The Regulation of Platelet-activating Factor Production in Endothelial ...

Vol. 264. No. 11, Issue of April 15,pp. 6325-6333.1989 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY

The Regulationof Platelet-activatingFactor Production in Endothelial Cells THE ROLE OF CALCIUM AND PROTEIN KINASE C* (Received for publication, August 31,1988)

Ralph E. WhatleySO, Patricia Nelson$, GuyA. ZimmermanS, Dennis L. StevensT, Charles J. Parker$, Thomas M. McIntyreSIl, and StephenM. Prescott$II From the Departments of $Medicine and IIBiochemistry, University of Utah School of Medicine, Salt Lake City, Utah84112 and the TUniversity of Washington-Boise Veterans AdministrationMedical Center, Boise, Zdaho 83702

Endothelial cells (EC) synthesize platelet-activating factor (PAF) when stimulated with agonists that bind to cell-surface receptors. We examined events that link receptor bindingto synthesis of PAF by EC. Bovine EC stimulated with agonists that interact with specific cell-surface receptors accumulated PAF only in the presence of extracellular calcium. Hormonal stimulation of EC resulted inCa2+entry characteristicof that seen with receptor-operated calcium channels; Indo- 1 measurements demonstrated that this inward flux of Ca2+caused prolonged elevated levels of intracellular Ca2+.EC were exposed to melittin or theta toxin from Clostridium perfringens (pore-forming peptides that increase the permeabilityof the plasma membrane for small molecules) resulting in an inward flux of Ca2+ and accumulation of PAF. Ca” appears to be regulatory for PAF production at the level of phospholipase Az-medlated production of the PAF precursor 1-0alkyl-2-lyso-sn-glycero-3-phosphocholine, as Ca2+was required for the stimulated hydrolysis of l-O-alkyl-2acyl-sn-glycero-3-phosphocholine. PAF accumulation in EC is also regulated by protein kinase C. Pretreatment of EC with phorbol esters that activate protein kinase C or with dioctanoylglycerol, followed by stimulation, resulted in a 2-fold increase in stimulated PAF production. The regulatoryeffect of protein kinase C also appears tobe at a phospholipase A2-mediated hydrolysis of 1-0-alkyl-2-acyl-sn-glycero-3-phosphocholine.

produced by many cells involved in theinflammatory response upon appropriate stimulation (1, 2). It causes activation and aggregation of platelets and neutrophils (1, 2), and may play a role in inflammation (3, 4) and the regulation of cell-cell interactions in uiuo. We have demonstrated that plateletactivating factor is produced by endothelial cells from diverse vascular beds in response to stimulation with agonists that interact with specific cell-surface receptors (5-9). In other studies (5, 6) we have shown that PAF produced by EC mediates the binding of polymorphonuclear leukocytes, which suggests that EC-derived PAF may play a role in vascular disease. The mechanism by which ligand-receptor interaction is coupled to PAF production is not known; however, as in other cellular responses involving phospholipids, such as arachidonic acid release and phosphatidylinositol turnover, activation of phospholipase activity is a central event(10-12). As others have reported (10, l l ) , we have found that hormonal stimulation of humanand bovine endothelial cells results in turnover of phosphoinositides with production of two second messengers, diacylglycerol and inositol polyphosphates.’ Diacylglycerol activates proteinkinase C, an enzyme that regulates many intracellular processes (13-15). We have shown that PAF and leukotriene B4production in polymorphonuclear leukocytes is dependent on what appears to be a protein kinase C-dependentstep (16). Inositol phosphates raise intracellular calcium concentrations by mediating release of calcium from intracellular stores (1,4,5-IP3)and perhaps by entry of extracellular calcium (1,3,4,5-IP4) (17-19). Moreover, these effects may be synergistic in the intact cell, Platelet-activating factor (l-O-alkyl-2-acetyl-sn-glycero-3phosphocholine; PAF)’ is apotent lipid autacoid that is as diglyceride increases the affinity of protein kinase C for calcium (13-15). There is evidence that both calcium and * This work was supported by the Nora Eccles Treadwell Founda- protein kinase C regulate the production of prostaglandin Iz tion and Grants HL34127, HL35828, DK35830, and National Re- by endothelial cells (10, 11, 20, 21), which suggests an effect search Service Award HL07529 from the National Institutes of on the phospholipase(s) that mediates arachidonate release. Health, Grant-in-aid 84-975 from the American Heart Association, In neutrophils, calcium and diacylglycerol may directly actiEstablished Investigator Awards 87-225 and 85-204 from the Amerivate phospholipase Azby a protein kinase C-independent can Heart Association with funds contributed in part by the Utah Affiliate, and Veteran Administration Merit Reviews ( t o D. L. S. and mechanism (22). C. J. P.). The costs of publication of this article were defrayed in part In endothelial cells, the agonists that stimulate PAF proby the payment of page charges. This article must therefore be hereby duction also stimulate hydrolysis of phosphoinositides (7-11, marked “advertisement” in accordance with 18 U.S.C. Section 1734 20). Although there is no direct evidence that the activation solely to indicate this fact. 3 To whom correspondence and reprint requests should be ad- of these two processes sharesa common mechanism, the dressed CVRTI, Bldg. 100, University of Utah, Salt Lake City, UT production of second messengers (IP3 and diacylglycerol) as a result of phosphoinositide hydrolysis suggests that changes 84112. The abbreviations used are: PAF, platelet-activatingfactor; lysoPAF, l-O-alkyl-2-lyso-sn-glycero-3-phosphocholine; Hepes, 4-(2-hy- ester); PMA, phorbol 12-myristate 13-acetate; di-C8, 1,2-dioctanoyl droxyethy1)-1-piperazineethanesulfonicacid; EC, endothelial cells; glycerol; GPC, glycerophosphocholine;IP, inositol phosphate; EGTA, HU, hemolytic unit; [Ca2+]~,extracellular calcium concentration; [ethylenebis(oxyethylenenitrilo)]tetraacetic acid; 4a-PDD, 4a-phorHUVEC, human umbilical vein endothelial cell; Indo-1 AM, 1-[2- bo1 12,13-didecanoate;HBSS, Hanks’ balanced salt solution. amino-5-(6-carboxyindol-2-yl)-phenoxy]-2-(2’-amino-5’-methyl- R. Whatley, E. Stroud, G. A. Zimmerman, T. M. McIntyre, and phenoxy)-ethane-N,N,N’,N’-tetraacetic acid (pentaacetoxymethyl- S. M. Prescott, manuscript in preparation.

6325

6326

The Role of Calcium and Protein Kinase

in PAF Production

in cellular calcium and/or protein kinase C activation might The incubations were performed at 25 “C and were stopped at the times shown by the addition of0.5mlof 50 mM acetic acid in regulate PAF production. A regulatory role forcalciumis suggested by our observationthat calcium ionophore A23187 methanol. Control incubations were performed identically, but in the absence of an agonist. Incubations performed in calcium-free buffer causes PAF accumulation in EC (9, 23), but ionophores may used HBSS that was nominally calcium-free. Experiments using have effects other than translocation of calcium from the phorbol ester ordi-C8 pretreatment were performed identically except extracellular space into the cell. that phorbol esters or di-C8 (dissolved in dimethyl sulfoxide) were In the experiments reported here, we have examined the added to the complete culture media for the indicated times prior to role of calcium and protein kinase C in PAF production by initiating the PAF assay. In experiments using sangivamycin, the hormonally-stimulated endothelialcells. Endothelial cellsare sangivamycin (or control solution) was added directly to themedium for 5 min prior to the addition of the phorbol ester. In experiments particularly important cells in which to examine these issues using sphingosine, the monolayers were exposed to sphingosine (in 1 because: 1) EC-derivedPAF appears to provide a novel mech- ml of HBSS plus 1 mg/ml of fatty acid-free bovine serum albumin) and therefore it is of particular or control buffer (HBSS, 1 mg/ml of fatty acid-free bovine serum anism for cell-cell interaction interest to understand its regulation, and 2) EC synthesize albumin) for 15 min prior to the initiation of the PAF assay. After the addition of the acidified methanol, the cells were scraped PAF in response to a variety of agonists that act via well from the surface of the dish. The lipids were extracted by the method characterized receptors on the cell surface. This contrasts of Bligh and Dyer (27) and were then separated by thin layer chrowith most other cellular models (e.g. polymorphonuclear leu- matography on precoated plates of Silica Gel 60 (Merck Darmstadt, kocyte) for studying PAF synthesis, in which nonphysiological Federal Republic of Germany) in solvent system I1 of Mueller et al. (26) The incorporation of [3H]acetate into a lipid that comigrated agonists must be used to achieve maximal synthesis. We present evidence that PAF production in response to agonist with authentic PAF was determined by scintillation spectroscopy. stimulation requires a prolongedelevation in intracellular We previously demonstrated (9) that we routinely have greater than 90% recovery of PAF using this method, and that the radiolabeled calcium concentration, which is achieved by an inward flux product is identical to authentic PAF by comigration in the TLC ofextracellularcalcium.Wedemonstrate that calcium is system, identical elution volume on high performance liquid chronecessary for the phospholipase Az-mediated production of matography system (28), the effects of phospholipases AI, A2, and PAF acetylhydrolase (7,9), andby its bioactivity (7, 9). t h e PAF precursor, l-0-alkyl-2-lyso-sn-glycero-3-phosphoAssay of 45Ca2+Flux into Endothelial Cells-The medium was choline. In addition, we find that endothelial cell production removed from confluent monolayers of endothelial cells and they of PAF in response to a calcium stimulus is regulated by were washed twice with 1 mlof HBSS. These monolayers were protein kinase C; this effect also occurs at the level of t h e incubated at 25 “C for the indicated times with 1ml of HBSS ([Ca2+] phospholipase A2-mediatedproduction of t h e PAF precursor. = 1.3 mM, pH 7.4) containing 5 pCi/ml of “CaC12 and the indicated concentrations of agonist. Control dishes were incubated identically, EXPERIMENTAL PROCEDURES but without an agonist. At the indicated times, the incubation buffer Materials-Bradykinin, ATP, angiotensin 11, calcium ionophore was removed and the culture dish immersed in 3 sequential washes A23187, sphingosine, phorbol 12-myristate13-acetate(PMA), 4a- of ice-cold HBSS (withoutcalcium) containing 0.1% EDTA to remove phorbol12,13-didecanoate(4a-PDD), and1,2-dioctanoylglycerol (di- any remaining extracellular “Ca2+. The monolayers were solubilized C8) were purchased from Sigma. Mellitin, P,y-methylene ATP, and in 1 ml of 1 M NHdOH and placed in scintillation vials for determination of radioactivity. Experiments using phorbol ester pretreatment &y-imido ATP were purchased from Fluka Chemical Corp. The were performed identically except that phorbol esters (dissolved in melittin contained no phospholipase A2 activity as determined using dimethyl sulfoxide) were added to thecomplete culture media for the a modification (7) of the method of Blackwell et al. (24). [3H]Acetate indicated times prior to initiating the “Ca2’ flux assay. (3.4 Ci/mmol), [3H]arachidonicacid (94.5 Ci/mmol), and ‘5CaC12 were A trivial explanation for apparently increased calcium flux could purchased from Du Pont-New England Nuclear. PAF was purchased be that the assay reflected binding of calcium to theexternal surface from Avanti Polar Lipids, Inc. Tissue culture medium was purchased of the cell. This was unlikely in view of the washes in EDTA, but to from KC Biological (Lenexa, KS) or MA Bioproducts (Walkersville, exclude this possibility, we measured the stimulated loss of calcium MD). Fetal bovine serum was purchased from Hyclone Laboratories from cells that had been preloaded with “Ca”. Culture medium was (Logan, UT). Collagenase was purchased from Cooper Biomedical removed from endothelial cell monolayers and replaced with 1 ml of (Malvern, PA). Purified human thrombin was a gift from Dr. George medium M199 (no supplementation), containing5 pCi/ml of “CaC12. Broze, Washington University, St. Louis, MO, or from John W. After 1 h of incubation at 37 “C, the labeling medium was removed Fenton 11, New York Department of Health, Albany, NY. Indo-1 AM and the monolayers were washed twice with 1 mlof HBSS. The was purchased from Behring Diagnostics. Sangivamycin was supplied monolayers were stimulated with the indicated concentration of agby the NaturalProductsBranch, Division of Cancer Treatment, onist in 1 ml of calcium-free HBSS (pH 7.4) at 25 “C. These condiNational Cancer Institute, National Institutes of Health. tions result in a large outward gradient and, if increased permeability Cell Culture-Bovine aortic and pulmonary artery endothelial cells is present, efflux is seen. Control incubations were performed idenwere cultured as described (9). Cells were cultured in 35-mm plastic tically, but without an agonist. At the indicated times, the incubation culture dishes (Corning) in medium 199 with 20% fetal bovine serum buffer was removed and the monolayers were washed with 1 mlof containing 100 units/ml penicillin and 100 pg/ml streptomycin. Hu- HBSS. The monolayers were solubilized in 1 ml of 1 M NHIOH and man umbilical vein endothelial cells were cultured using a modifica- placed in scintillation vials for determination of radioactivity. Both tion of the method of Jaffe et al. (25) as previously described (5-8). ATP andbradykinin caused loss of calcium from preloaded cells (not The cells were cultured in 35-mm dishes in medium 199 containing shown), demonstrating that these agonists induce an increased cal25 mM Hepes buffer supplemented with L-glutamine (2 mM), penicil- cium permeability in the cell membrane. lin (100 units/ml), and streptomycin (100 pglml), and 20% pooled Measurements of Intracellular-free Calcium Concentrations-Meashuman serum. urements of the time course of changes in intracellular calcium All experiments were performed on endothelial cells in primary activity were performed using the calcium-sensitive dye Indo-1 as culture; at confluence there were approximately lo6 cells/dish. We described (29). Confluent cultures of endothelial cells grown on coverhave previously shown that these cells have typical endothelial mor- slips wereexposed to the acetomethoxyester of Indo-1 (1 pM in phology, express von Willebrand factor, take up acetylated low den- standard culture medium) for 1 h at 37 “C. The cells were then sity lipoprotein, and possess angiotensin-converting enzyme activity washed in dye-free medium for 20 min. The coverslip was placed in the perfusion chamber of a custom designed spectrofluorimeter (360 (9). Assay of Platelet-activating Factor Production-Platelet-activating nm excitation). Fluorescence was measured at 410 and 480 nm and factor accumulation was measured by incorporation of [3H]acetate the ratio of the signals reflects intracellular calcium concentrations into PAF by a modification of the method of Mueller et al. (26) as (29). previously described (8,9). Briefly, culture medium was removedfrom Measurements were made by perfusing the endothelial cell monoconfluent monolayers of endothelial cells and replaced with 1 ml of layers first with control buffer (HBSS, 10 mM Hepes, pH 7.4) to HBSS (1.3 mM Ca2+,10 mM Hepes, pH 7.4) that contained 25 pci of measure basal levels of [Ca”] followed by perfusion with the same carrier-free [3H]acetate,and theappropriate concentrationof agonist. buffer containing an agonist. Only one intervention was performed

The Role of Calcium and Protein Kinase in PAF Production on each monolayer, except in the case of a lack of a response. In such cases a positive control (bradykinin) was subsequently tested to exclude technical artifacts as a cause of the lack of response. Assays Using Clostridium Perfringens Theta Toxin-Theta toxin was purified as described (30) and was homogeneous as evidenced by the finding of a single band following electrophoresis in a polyacrylamide gel that contained sodium dodecyl sulfate. Activation of the theta toxin requires the presence of sulfhydryl reagents (30, 31) and incubations performed with theta toxin contained 33 mM cysteine. Theta toxin is inactivated by exposure to oxygen (31) and this was performed by bubbling 100% oxygen through the thetatoxin-containing buffer for 30 min. In allexperiments using theta toxin, the endothelial cell monolayers were examined under phase-contrast microscopy and therewas no evidence of cell lysis. r H ~ A r a c ~ i d Q n aLabeling te and S t i m u ~ of~ E ~ ~n o t h e ~ iCelkal Confluent cultures were incubated for 4 h in 1 ml of medium (M199 + 1 mg/ml fatty acid-free bovine serum albumin) containing 0.2 pCi [3H]arachidonic acid/ml. The labeling medium was removed and the monolayers were washed twice with HBSS. The monolayers were then incubated for the indicated times in buffer (HBSS, pH 7.4) containing bradykinin M). Control incubations were performed identically but without bradykinin. Paratlel incubations were performed identically in nominally calcium-free buffer (HBSS, 0 Ca”, pH 7.4). The incubations were terminated by the addition of 0.5 ml of acidified methanol and thelipids extracted by the method of Bligh and Dyer (27). The organic phase was dried under N1and applied to precoated plates of Silica Gel 60 (Merck). Phospholipids were separated by TLC using the solvent system: chloroform/methanol/glacial acetic acid/H20 (25:15:4:2).They were identified by comparison with the migration of authentic standards run in parallel, were scraped from the plate, and theamount of radioactivity in each phospholipid was determined by scintillation spectroscopy. Experiments using phorbol ester pretreatment were performed identically except that afterremoval of the [3H]arachidonate-containing medium and the subsequent washes, complete culture medium containing the phorbol esters was placed on the monolayers. After exposure for the time indicated, the phorbol-con~ining medium was removed and themonolayers were washed with HBSS. The remainder of the experiment was performed exactly as described above. ~ H ~ A r ~ Labeling h ~ oand ~ Release ~ e from 1-0-Alkyl-2-arachia’omyl-sn-glycero-3-phosphocholine-Confluent cultures of bovine endothelial cell in 75-em2flasks were incubated for 2 h in labeling medium (M199 -k 1 mg/ml fatty acid-free bovine serum albumin) containing 0.5 pCi/ml of [3H]arachidonic acid. The labeling medium was removed and themonolayers were washed twice with HBSS. The monolayers were then incubated in M199 f 20% fetal bovine serum for 24 h. The medium was removed and themonolayers were washed twice with HBSS. The monolayers were then incubated for 15 min in stimulation buffer containing the indicated agonist. Stimulation buffers contained either 1.3 mM Ca” (HBSS, 10 mM Hepes, pH 7.4) or 0 Ca2+ (HBSS without calcium, 10 mM Hepes, pH 7.4). Control incubations were performed identically but in the absence of an agonist. Experiments using phorbol ester pretreatment were performed identically except that afterremoval of the [3H]arachidonatecontaining medium and the subsequent washes, complete culture medium containing the phorbol esters or vehicle (dimethyl sulfoxide) was placed on themonolayers. After exposure for the time indicated, the phorbol-containing medium was removed and the monolayers were washed with HBSS. The remainder of the experiment was performed exactly as described above. The incubation was terminated by removal of the stimutation buffer and placing 3 ml of ice-cold water/methanol ( 2 1 ) on the monolayers and scraping the cells from the flask. The lipids were extracted and 20% of the organic phase was removed and theamount of radioactivity determined by liquid scintillation spectroscopy. Phospholipids were separated by TLC as described above and the percentage of ra~oactivity each in phosphoiipid was determined using a TLC radioactivity scanner. The choline phosphoglyceride fraction was scraped from each TLC lane and extracted by the method of Bligh and Dyer (27). This extract was separated into subclasses using the method of Blank et al. (32) and the percentage of label in each subclass was determined. Briefly, the choline phosphoglycerides were hydrolyzed to diradylglycerols with phospholipase C,followedby benzoylation. The benzoyl derivatives of the diradylglycerols were separated into subclasses (diacyl, alkylacyl, and alk-1-enylacyl) by TLC using benzene/hexane/diethyl ether (5045:4). The amount of label in each subclass was determined by TLC radioactivity scanning.

6327

RESULTS

Extracellular Calcium Is Required for PAF Synthesis-As we have demonstrated previously (9), stimulation of bovine EC with ATP ( W 3 M) or bradykinin (lo-’ M) resulted in PAF accumulation in the presence of ea2+(1.3 mM) (Table I). However, in the absence of ea2+ (nominally calcium-free buffer), there was no accumulation ofPAF above control levels (Table I). This suggested that PAF synthesis by bovine endothelial cells requires extracellular calcium. However, it was possible that the response was only delayed. To exclude this possibility we examined the time course of the response to bradykinin, which demonstrated that there was no detectable PAF accumulation in the absence of calcium, even at 20 min (not shown). We next stimulated bovine EC with bradykinin in the presence of varying concentrations of calcium. PAF production demonstrated a clear dependence on external calcium with an EC50of approximately M (Ca2+]B(Fig. 1). Maximal production occurred at approximately M [Ca2+]E,and there was no detectable PAF accumulation at concentrations of calcium below M. We also found that PAF accumulation by human umbilical vein EC was similarly dependent upon extracellular ea2+.PAF production by HUVEC stimulated with thrombin ina calcium-free, 1 mM EGTA buffer was 22.5% of that seen in 1 mM calcium buffer (n = TABLE I The effect of external calcium on p ~ t e ~ t - a c ~ i u factor a~~ng production by endothelial cells Cultured bovine aortic endothelial cells (IO6 cells/dish) were incubated with the indicated agonist in buffer containing 1.3 mM calcium or the same buffer containing no calcium for 10 min a t room temperature. The incubation buffer contained 25 pCi/ml [3H]acetate. Incorporation of I3H]acetate into PAF was determined and is presented as disintegrations/min/dish (mean f S.D.). +Calcium (1.3 mMf No calcium d m

Bradykinin (lo-’ M) ATP (10” M) Control (no agonist)

0

6249 f 386 ( n = 4) 8333 f 281 (n = 3) 121 rt 35 ( n = 4)

5

149 f 53 ( n= 4) 377 +. 91 (n= 4)

I

I

I

4

3

2

-Log [Ca *+I, M

FIG.1. The effect of varying external calcium concentrations on PAF production by endothelial cells stimulated with bradykinin. Bovine aortic endothelial cells were exposed to lo-’ M bradykinin in buffer (HBSS, pH 7.4) containing different concentrations of calcium. Incubations were performed for 10 min at 25 ‘C. The reaction was stopped and the incorporation of [3H]acetate into PAF was determined as described under “Experimental Procedures.” Each point represents the mean of determinations in two separate dishes (lo6cells/dish). The results shown are representative of three experiments.

The Role of Calcium andProtein Kinase in PAF Production

6328

lo), confirming an earlier observation under nominally calTABLEI1 cium-free conditions (7). The effect of divalent cations on the stimulated production of platelet-actiuating factor by endothelial cells We subsequently tested whetherthe lack of response in the Cultured bovine aortic endothelial cells (lo6cells/dish) were incuabsence of external calcium could be specifically reversed by calcium. In particular, we wished to know if an activated state bated for 10 min in stimulation buffer (HBSS, no added calcium or magnesium, pH 7.4) containing lo-’ M bradykinin, 25pCi [3H] could be attained by stimulation in the absence of Ca2+,and acetate/ml, and theindicated cation. Control incubations used Ca2+/ allow subsequent PAF production by the addition of just M$-free buffer (“None”) orthe same buffer plus 1m M EDTA. Each calcium. In this experiment bovine EC were incubated in point is the mean of measurements made in two separate dishes and normal buffer or calcium-free buffer (1 mM EGTA). In the represents incorporation of [‘Hlacetate into platelet-activating factor. presence of calcium, accumulation of PAF in response to Cation added PAF Calcium response (1 mM) accumulation stimulation with bradykinin (lo-’ M) was maximal within 510 min (Fig. 2). When endothelial cells were stimulated with % dpm bradykinin in the absence of calcium (+EGTA), as before, Ca2+ 4528 100 563 there was no accumulation of PAF. However, when calcium 15 Mi? Mn2+ 379 8 was added (final [Ca”], of 10 mM), there was an immediate cu2+ 635 14 and rapid accumulation of PAF. The amount of [3H]PAFthat Zn2+ 556 12 accumulated after calcium addition equaled that present in None 542 12 cells maintained in the continual presence of calcium and 121 None/l mM EDTA 3 subsequently followed normal kinetics (Fig. 2). This experiment also excluded the possibility that EGTA was having an TABLE I11 effect other than chelating calcium. The importance of the calcium flux wasfurther emphasized Accumulation of PAF in bouine aortic endothelial cells in response to stimulation with ATP analogs by the failure of other divalent cations to support PAF proCultured endothelial cells (lo6cells/dish) were stimulated with the duction (Table 11).Endothelial cells were stimulated inbuffer in which MgZ+, Mn2+,Cu2+,or Zn2+had been substituted for indicated agonist for 10 min at room temperature. PAF synthesiswas determined as the incorporation of [’Hlacetate into PAF (mean f calcium. In each case, PAF accumulation was less than 16% S.D., n = 4). of the response seen when cells were stimulated in the presaccumulation PAF Agonist ence of calcium and equivalent to that seen when the incudpm bations were performed in a nominally calcium-free buffer Buffer 60 f 5 (12%). Stimulation inzero calcium buffer (nominally calciumATP M) 7440 f 783 free buffer plus 1 mM EDTA) resulted in PAF production 4660 f 690 ADP M) that was only 3% of that seen in a 1.3 mM calcium buffer M) 705 f 95 P,y-Methylene ATP (Table 11). 6705 f 520 0,y-Imido ATP (lo-’ M) Agonists That Stimuhte Synthsis Also Cause Calcium Influx-These experiments demonstrated that endothelial cells require extracellular calcium for synthesis of PAF in response allow PAF synthesis. To testthis hypothesis directly, we to stimulation with a receptor-mediated agonist, and sug- determined whether agonists that cause PAF accumulation gested that an inward movement of calcium is necessary to also cause increased calcium entry. Cells were stimulated in buffer containing “Ca2+ and the inward flux of calcium was measured. When bovine cells were stimulated with bradykinin (lo-’ M), there was an inward movement that was equivalent to that produced with calcium ionophore A23187 (lo-‘ M) (Fig. 3). This was also seen with ATP, an agonist that offered an opportunity to further test the association of PAF synthesis with Ca2+ entry by the use of ATP analogs. Several important effects of ATP have been described that suggest that ATP functions as an extracellular, receptor-mediated agonist. These activities include modulation of vascular tone, hydrolysis of phosphoinositides, and release of arachidonic acid metabolites by endothelial cells (33-35). We stimulated cultured bovine endothelial cells with ATP analogs and related compounds. Accumulation of PAF occurred with ATP = &y-imido ATP > ADP >> &y-methylene ATP (Table 111). Adenosine did not cause accumulation of PAF significantly above buffer control (data notshown). These compounds then 0 10 20 30 were tested for their ability to stimulate the uptake of “Ca2+ and we found an identical rank order of potency. Time(minutes) We next examined whether the hormone-induced calcium FIG. 2. Time course of PAF accumulation in endothelial entry was characteristic of that seen with voltage-dependent cells stimulated with bradykinin, with addition of calcium at M) and nifedipine (5 X different times. Bovine aortic endothelial cells (lo6cells/dish) were calcium channels. Verapamil M), which block such channels, had no effect on either exposed to lo-’ M bradykinin in buffer (HBSS, pH 7.4) containing 10 mM [Ca”] (0)or 0 [Ca”] (0).At 9 min (indicated by the arrow), PAF production or “Ca2+ entry in bradykinin-stimulated EC sufficient calcium chloride was added to the plates containing cal- (not shown). Moreover, there was no PAF production in EC cium-free buffer to achieve a free calcium concentration of lo-’ M. that were depolarized by exposure to a high K+, low Na+ At the indicated times, the reactions were stopped and the incorporation of [‘Hlacetate into PAF determined. Each point representsthe buffer (10 mM Na+, 136 mM K+) (36). These results exclude mean of two separate dishes. The results shown are representative of a voltage-dependent Ca” channel, and we conclude that the hormone-induced influx of calcium that supports PAF syntwo experiments.

%

I

I

The Rote of C ~ ~ cand i uProtein ~ Kinase in PAF Production IS

6329

b.

a

IO

5 10

0

20

time(mhlne8)

0

I

I

5

10

Tirne(rninutes)

FIG.3. Bradykinin stimulates''Cas+ influx intoendothelial cells. Bovine aortic endothelial cells (10' cells/dish) were incubated at 25 "C inthe presence of '5Ca2+(5 pCi/ml) in HBSS (1.3 mM ca"). The incubations containedbuffer alone (O),bradykinin (lo-' M) (x), or calcium ionophore A23187 (lo-' M) (0).At the indicated times, the incubation buffer was removed and thedishes were rapidly washed in three sequential baths of ice-cold Ca2+-freeHBSS containing0.1% EDTA. The cells were solubilized with 1ha ammonium hydroxide and the cellular 'T!az+ content determined by scintillation spectroscopy. Each point representsa single determination in onedish. The results shown are typical of multiple experiments.

30

0

20

40

60

80

1CQ

RHt.Mxln (HUlml)

FIG.4. Clostridial theta toxin causes PAF accumulation in endothelial cells. A , time course. Bovine aortic endothelial cells were exposedto clostridial theta toxin (25 HU/ml, 9 pg/ml) in HBSS (1.3mM [Ca'+], pH 7.4)containing 25 pCi/ml [3H]acetateand 33 mM cysteine. Incubations were performed at 25 "C and terminated at the times indicated. The incorporation of [3H]acetate into PAF was determined. Each point (0) represents the mean of determinations in two separate dishes (IO6cells/dish). B, concentration response. Bovine aortic endothelial cells were exposed to varying concentrations of theta toxin (1 HU = 0.36 pg of protein) in buffer (HBSS, 1.3 mM [Ca*+],33 mM cysteine, pH 7.4) containing 25 pCi/ml [3H]acetate. The incubations were performed a t 25 "C and terminated after 10 min. Incorporation of [3H]acetate into PAF was determined for monolayers ( IO6 cells/dish) exposed to thetatoxin (0)and theta toxin exposed to 100% oxygen for 30 min (0). i.o

thesis is due to a receptor-operated calcium channel (37). A Sustained Elevation in Cytophmic-free Calcium Is Sufficient to Initiate PAF Synthesis-Because the evidence indicated that an inward flux of calcium is required for PAF synthesis inresponse to hormonal stimulation, we next asked whether an inward flux of calcium is sufficient to initiate PAF production in endothelial cells. This was suggested by the ability of A23187 to induce PAF synthesis (9,23), butA23187 is known to affect intracellular membranes as well as the plasma membrane (38). Therefore, we examined the effect of substances that areknown to selectively increase the permeability of the plasma membrane. One such substance is theta toxin, a peptide produced by C ~ s t r ~ ~ ~ m p ethat ~ r under reducing conditions, creates a pore in cholesterol-containing (i.e. plasma) membranes (31, 39-41). eli it tin, a peptide in bee venom, alters plasma membranes in an analogous fashion (42-44). In control experiments we showed that theta toxin and melittin each caused an inward flux of calcium equivalent to that induced by A23187. Furthermore, bovine M) for 10 endothelial cells that were exposed to melittin ( min in buffer containing 1.3mM Ca2+accumulated PAF (5942 +. 286 dpm; n = 2), a response equivalent to that observed with receptor-mediated agonists such as bradykinin or ATP. Clostridial theta toxin also induced PAF accumulation by EC. This occurred in a time- and concentration-dependent fashion with maximal accumulation at 10 min and with an ECm of approximately 15 hemolytic units of theta toxin (Fig. 4, a and b). PAP synthesisdid not occur in medium that lacked calcium (not shown). We confirmed that thetatoxin was the substance causing these effects in two controls: first, theta toxin that had been inactivated by exposure to oxygen (31), demonstrated a dramatic loss of effect (Fig. 4b). Second, activation of the toxin requires the presence of reducing sulfhydryl reagents (31). In the absence of sulfhydryl reagents, PAF accumulation was markedly decreased (>95% reduction, data not shown). Theta toxin also caused PAF accumulation in human EC. Calcium was required, and the toxin was as effective as thrombin which gives a maximal response (data not shown) (6-8).

1

J

l ~ u t e

i n g e ~ FIG.5. Measurements o f changes in intracellular calcium activity using Indo- 1. Bovine endothelial cells grown on coverslips were lahefedwith Indo-1 AM (10 p~ in standardculture medium) for 1 h at 37 "C. The labeling medium was then removed and replaced with dye-free medium for 20 min. The coverslips were placed in the perfusion chamber of a custom spectrofluorimeter (excitation 360 nm) and fluorescence was measured at 410 and 480 nm. The ratio 410/480 nm, expressed here as arbitrary units, is proportional to the intracellular calcium concentration. The cells were initially perfused with control buffer (HBSS, X0 mM Hepes, pH 7.4)and thenwith the indicated agonist in the same buffer. A; cells exposed to lo-' M bradykinin. The dashed line indicates the effect of changing the perfusate to control buffer. B, cells exposed to bradykinin in a low calcium (200nM) buffer. C, cells exposed to lo4 M angiotensin 11 in standard calcium buffer (1.3 mM [CaZ+]).D, cells perfused with thrombin (2u n i ~ / m l )in standard calcium buffer (1.3 mM [Ca"']). E, cells perfused with clostridial theta toxin (12.5HU/ml). (Bothcontrol and stimulation buffers contain 1.3 mM [Ca"] and 33 mM cysteine.)

We next examined whether the inward movement of calcium elevated the intracellular-free calcium concentration by measuring the cellular calcium activity with the calciumsensitive fluorescent dye, Indo-1. When bovine endothelial cells were exposedto substances that cause PAF accumulation and 45Ca2+influx, prolonged elevations in intracellular calcium were observed (Fig. 5). Stimulation with the receptormediated agonist bradykinin (lo-' M) in 1.3 mM calcium buffer resulted in an initial spike followed by a plateau of elevated calcium (Fig. 5A). The plateau phase was as long as

6330

The Role of Calcium and Protein Kinase in PAF Production

15 min and could be terminated either by the removal of the by bradykinin in the presence of calcium (Table IV). In the bradykinin or the calcium from the perfusion buffer. When absence of calcium there was no decrease in the amount of cells were stimulated with bradykinin in a low calcium (200 [3H]arachidonate in alkylacyl-GPC (Table IV) compared to nM) buffer, the initial spike occurred but therewas no plateau unstimulated controls. Similar results were seen when the (Fig. 5B). We conclude that theinitial sharp rise represented endothelial cells were exposed to melittin (Table IV). This release of calcium from intracellular stores and the plateau demonstrated that calcium is necessary for the hydrolysis of M) l-O-alkyl-2-arachidonoyl-sn-glycero-3-phosphocholine was due to influx of extracellular calcium. Verapamil( in had no effect on intracellular calcium levels in bradykinin- stimulated endothelial cells. Protein Kinase C Activity Appears to be Essential for PAF stimulated cells, providing additional evidence that the stimulated influx is not via a voltage-dependent calcium channel Synthesis-Our initial experiments to examine the role of protein kinase C in EC PAF production utilized an activator (not shown). We tested other agonists to determine whether the associ- of protein kinase C, PMA.Treatment of endothelial cells with ation of PAF synthesis with sustained intracellular [Ca2+] PMA alone did not cause production of platelet-activating was a generalized phenomenon. Angiotensin 11, which is a factor (23), release of arachidonate from phospholipids, or weak agonist for PAF production in these cells (9), caused calcium entry, and had no effect on agonist-induced calcium increased intracellular calcium, but of a smaller magnitude influx (data not shown). However, a short pre-exposure of than bradykinin (Fig. 5C). Thrombin, which, in contrast to endothelial cells to PMA, followed by a stimulus that raised its effect on human umbilical vein EC, is not an agonist for intracellular calcium, resulted in an increase in PAF producPAF production in cultured bovine EC (9), caused an initial tion over that seen in controls (Fig. 6). This was observed transient spike, but no prolonged elevation of intracellular whether the calcium rise was caused by a hormone (bradykicalcium (Fig. 5D).Cellsexposed to clostridial theta toxin nin) or a calcium ionophore (A23187). An isomer of PMA, demonstrated a sustained elevation of intracellular calcium 4a-phorbol 12,13-&decanoate (4a-PDD), which does not acbut without an initial spike (Fig. 5E). This is consistent with tivate protein kinase C, had no effect on PAF production. the known actions of this substance in causing an increased The effect of PMA was time-dependent, with maximal actipermeability of the plasma membrane and allowing an influx TABLEIV of extracellular calcium. Thus,this series of experiments The effectof calcium on the stimulated loss of pH]arachidonic acid demonstrated a direct correlation between the ability of an from alkylacyl-GPC agonist to cause prolonged elevations of free cytoplasmic Bovine pulmonary artery endothelial cells (10' cells/point) were calcium and the initiation of PAF synthesis. Calcium Regulates the Phospholipase A2 in the P A F Syn- prelabeled with [3H]arachidonicacid for 2 h. 24 h later the cells were exposed to the indicated agonist for 15 min in buffer (HBSS) conthetic Pathway-Our next experiments were designed to iden- taining 1.3 mM Ca2+or 0 calcium (HBSS-no calcium). Phospholipids tify the calcium-regulated step in the PAF synthetic pathway. were separated by TLC and theamount of label in subclasses (diacyl, The initial step in the synthesis of PAF by endothelial cells alkylacyl, and alk-1-enyl acyl) of choline phosphoglycerides was deis a phospholipase A*-mediatedhydrolysis of 1-0-alkyl-2-acyl- termined as described under "Experimental Procedures" (mean f sn-glycero-3-phosphocholine,with the production of lyso- S.D., n = 2; percentage shown is the percent reduction from control). +Calcium (1.3 mM) No calcium PAF. Phosphlipase Az also catalyzes the release of arachidonate from the sn-2 position of membrane phospholipids in EC dPm (45) in a calcium-dependent manner (21). Other workers (46, 20,683 f 1,533 Control 16,262f 340 20,621 f 2,017 47) have shown that l-O-alkyl-2-arachidonoyl-sn-glycero-3- Bradykinin (lo" M) (0.3%) (21.4%) phosphocholine is a precursor for PAF synthesis, and we have 13,602f 2618,307 f 1,040 Melittin M) bovine found that this molecule is present in human and (34.2%) (11.5%) endothelial cells (48). We hypothesized that the effect of calcium in regulating PAF production is on the phospholipase 25 I I Az-mediated hydrolysis of l-O-alkyl-2-acyl-sn-glycero-3phosphocholine. To test this, we examined the calcium dependence of bradykinin-induced release of arachidonate from all phospholipids and specifically from alkylacyl-GPC. Bovine Stimulus: ECwere prelabeled with [3H]arachidonateandstimulated 0 Bradykinin with bradykinin in buffers containing 1.3 mM calcium or 0 calcium. The release of arachidonate for phospholipids was \23187 dependent upon extracellular calcium: 6.2% of the [3H]arachidonate was released from phospholipids in the presence of extracellular calcium, whereas in a calcium-free buffer it was reduced to 1.5% (means of four determinations). A similar dependence on calcium was seen for release of arachidonate PPr eMl B rAeuaftfme re n l : PDD from EC exposed to melittin (data not shown). This demonFIG. 6. Effect of PMA pretreatment on agonist-induced strated that the entryof extracellular calcium is required for PAF production by endothelial cells. Bovine pulmonary artery release of arachidonate from cellular phospholipids. We next endothelial cells (lo6 cells/dish) were exposed to PMA (lo-' M), 4aexamined directly the calcium dependence of the stimulated PDD M), orcontrol (dimethyl sulfoxide) (which were added to The cells were standard culture medium from stock solutions), for 10 min at 37 "C. hydrolysis of 1-0-alkyl-2-arachidonoyl-GPC. The medium was removed and the monolayers were washed twice labeled with [3H]arachidonate and then exposed to a hormonal agonist (bradykinin), an agonist that directly causes with HBSS. The monolayers were then exposed to buffer (HBSS, 10 mM Hepes, pH 7.4,25pCi [3H]acetate/ml) that contained bradykinin calcium entry (melittin), or buffer only. Each condition was M) or calcium ionophore A23187 (10- M) for 10 min at room examined in buffer that contained 1.3 mM calcium or was temperature. The incubations were terminated and theincorporation nominally calcium-free. The amount of [3H]arachidonate in of [3H]acetate into PAF was determined. Each point is the mean of alkylacyl-GPC decreased by 21% when cells were stimulated determinations in two dishes.

The Role of Calcium and Protein Kinase in PAF Production vation within 5-10 min (Fig. 7). With longer pre-exposure times (60 min), PMA inhibited subsequent PAF production (Fig. 7), which may be due to inactivation of protein kinase C that occurs with longer exposures to phorbols (49,50). This effect was not observed when 4a-PDD was used for the pretreatment (not shown). The effect of PMA was concentration-dependent with a maximal effect at approximately low7 M PMA; no effect was seen below concentrations of lo-* M (not shown). To further examine whether the effect of PMA was due to protein kinase C we utilized 1,2-dioctanoylglycerol, a water-soluble diglyceride. Like PMA, short preexposure of endothelial cells to 1,2-dioctanoyl glycerol followed by bradykinin stimulation, resulted in a 2-fold increase in PAF production over that seen with control pretreatment (not shown). We next examined the regulation of EC PAF production using inhibitors of protein kinase C. Recently, sphingosine and other long chain amineshave been shown to be inhibitors of protein kinase C in vitro (51) and in whole cells (16, 52). Pre-exposure of endothelial cells to sphingosine for 15 min inhibited PAF production in response to a subsequent stimulus (IC50 of sphingosine = 10 PM), although complete inhibition required high concentrations (100 p ~ (Fig. ) 8). Because sphingosine is poorly soluble in aqueous media, these experiments used albumin as a carrier; consequently, the effective concentration of sphingosine may be much lower than indicated. Finally, we utilized sangivamycin, an unrelated inhibitor of protein kinase C that acts at theATP binding site and does not compete for the diacylglycerol (phorbol) binding site (53). Bradykinin-stimulated PAF production following sangivamycin pretreatment (10 p ~ 5,min) was only 61% of control. Moreover, the ability of PMA to augment PAF production in stimulated EC (241% of control) was completely blocked by pretreatment (5 min) with 10 PM sangivamycin (109% control). The Regulatory Effect of Protein Kinase C Is at the Level of

'

4

2

0 0

20

40

60

Length of PMA Pretreatment (minutes)

FIG.7 . Effect of increasing time of PMA pretreatment on bradykinin-induced PAF production by bovine endothelial cella. Bovine pulmonary artery endothelial cells (lo6cells/dish) were exposed to P M A M) (added to standard culture medium from a stock solution) for the indicated times at 37 "C.The medium was removed and the monolayers were washed twice with HBSS. The monolayers were then exposed to buffer (HBSS, 10 mM Hepes, pH 7.4,25 pCi [3H]acetate/ml) that contained bradykinin (lo" M ) for 10 min a t room temperature. The incubations were terminated and the incorporation of [3H]acetateinto PAF was determined. Each point is the mean of determinations in two dishes. The results shown are representative of multiple determinations performed in cells cultured from multiple isolates.

0

6331

I

I

7

6

I 5

4

-Log [Sphingosine] (M)

FIG.8. Inhibition of bradykinin-induced PAF production by sphingosine. Bovine endothelial cells (lo6 cells/dish) were exposed to increasing concentrations of sphingosine (in HBSS 1mg/ ml of fatty acid-free bovine serum albumin) for 15 min at 37 'C. This buffer was removed and the monolayers were washed twice with HBSS. The cells were then exposed to stimulation buffer (HBSS, 10 mM Hepes, pH 7.4,25 pCi [3]acetate/mi) containing bradykinin M)for 10 min. The incubations were terminated and theincorporation of [3H]acetate into PAF was determined. Each point is the mean of determinations in two dishes.

+

P ~ s p ~ ~Az-Recently, i p ~ e others have found that activation of protein kinase C causes an increase in the stimulusinduced production of prostaglandin IZ by endothelial cells (20) and suggested that thiseffect wasdue to aneffect on the phospholipase Az-mediated release of arachidonate from phospholipids. We hypothesized that theincrease in the stimulated production of PAF that occurs in response to protein kinase C activation may arise from an effect on the phospholipase A2-mediatedhydrolysis of 1-0-alkyl-2-acyl-sn-glycero3-phosphocholine. To test this, we examined the effect of protein kinase C activation on the stimulated release of [3H] arachidonate from phospholipids, including the precursor of PAF, 1-0-alkyl-2-arachidonoyl-GPC. [3H]Arachidonate release fromendothelial cell phosphatidylcholine in response to stimulation with bradykinin (lo-' M for 15 min) was 3.1 k 0.5% ( n = 4); 10-min pretreatment with M PMA, followed by bradykinin stimulation, increased the release to 5.6 +. 1.4% ( n = 4) ( p < 0.01). Control experiments demonstrated that PMA alone did not cause release of arachidonate and the inactive phorbol 4a-PDD had no effect on agonist-induced arachidonate release (not shown). PMA pretreatment similarly augmented total release of [3H]arachidonate from phosphatidylcholine in response to the calcium ionophore A23187 (10 p M for 15 min)from 4.3 k 1.2% ( a = 4) to 18.2 +- 1.7% ( n = 4) ( p < 0.001). More importantly, PMA pretreatment increased the release of 13H]arachidonate from l-O-alkyl-2arachidonoyl-GPC. Pretreatment with PMA, followedby stimulation with calcium ionophore resulted in 17.6 f 2.3% ( n= 4) release of label compared to 3.8 f 1.8%( n = 4) release fromcells not pretreated ( p < 0.001). PMA pretreatment followed by stimulation with bradykinin resulted in a 6.4 & 1.1%( n = 4) compared to 2.8 k 1.1%( n = 4) ( p < 0.01) release from cells not pretreated. These observations support the hypothesis that protein kinase C regulates the phospholipase Az-mediated hydrolysisof 1-0-alkyl-2-acyl-GPC. DISCUSSION

The receptor-mediated agonists that stimulate the production of PAF by endothelial cells also induce phosphoinositide

6332

The Role of Calcium and Protein Kinase in PAF Production

turnover, resulting in the production of second messengers that cause intracellular calcium to rise and protein kinase C to be activated. We have investigated the role ofthese signals in coupling agonist-receptor binding to thesynthesis of platelet-activating factor. We find that the production of PAF in response to these agonists is dependent upon the presence of extracellular calcium. Moreover, these agonists increase the permeability of the membrane for calcium, resulting in an inward flux of calcium and prolonged elevation in free intracellular calcium concentrations. In addition to demonstrating a requirement for calcium for PAF synthesis, we have shown that a prolonged increase in intracellularcalcium is sufficient to initiate PAF synthesis. Treatment of endothelial cells with theta toxin or melittin, which selectively permeabilize the plasma membrane, resulted in an inward flux of calcium and production of platelet-activating factor. Other cellular responses may have a different calcium threshold as recently demonstrated by Jaffe et al. (IO),who found that prostacyclin production by stimulated HUVEC waslargely dependent upon release of calcium from internal stores. The mechanism(s) that couple receptor-ligand binding in endothelial cells to elevations in cellular calcium arenot completely known but include opening of plasma membrane channels (either voltage-dependent or non-voltage dependent), production of inositol triphosphate (IP3) with consequent release of calcium from intracellular stores,production of inositol tetrakisphosphate (IP4)which may effect a transmembrane influx of calcium (18),or a combination of these. Our studies demonstrate a stimulated increase in calcium that is due to both release from intracellular stores as well as an influx of extracellular calcium. Our experiments also indicate that EC, like certain other excitable cells (37), have receptoroperated Ca2+channels, since maneuvers that alter voltageactivated Ca2+ channels had no effect on basal or ligandstimulated PAFsynthesis. Jaffe et al. (10) have demonstrated production of IPS in thrombin-stimulated HUVEC that coincides with an elevation in intracellular calcium, suggesting that inositol phosphates may function as second messengers to raise intracellular calcium in stimulated endothelial cells. Our results show that the mechanism by which elevated intracellular calcium causes PAF accumulation is activation of a phospholipase A2, whichresults in the hydrolysis of 1-0alkyl-2-acyl-sn-glycero-3-phosphocholine to 1-0-alkyl-2-lysosn-glycero-3-phosphocholine, which is subsequently acetylated to PAF. The activation of a phospholipase Az by increases in intracellular calcium concentrations is supported by the observations of Whorton et al. (21) who found a similar calcium dependence on stimulated arachidonate release from porcine endothelial cells. Activation of a cellular phospholipase A2by a stimulatedincrease in intracellular calcium would result in a concomitant production of both the precursor for platelet-activating factor and release of free arachidonate as proposed by Chilton et al. (46) and Albert and Snyder (47). Calcium may also regulate subsequent steps in the PAF synthetic pathway. Conversion of 1-0-alkyl-2-lyso-GPC to PAF is catalyzed by acetyl-CoA:alkyllyso-GPC acetyltransferase (54), which is a calcium-dependent enzyme (55-57). Moreover, acetyltransferase activity measured in homogenates of neutrophils and eosinophils is increased if the cells havebeen stimulated with calcium ionophore (58, 59). Although calcium is required at subsequent steps in the PAF synthetic pathway, we have shown in intactcells that calcium is absolutely required for the first step, which suggests that this is the primary regulatory point. A requirement for extracellular calcium in PAF synthesis has been demonstrated previously in endothelial (7) and other

cells (60, 61). Our results have extended those observations as we have unequivocally identified calcium as the relevant ion (Fig. 2), andshown that other divalent cations are without effect (Table 11). We have also demonstrated a strict correlation of the potency of agonists for PAFsynthesisanda calcium signal, and examined the effects of each of these manipulations on free cytosolic calcium (Fig. 5). This last point is particularly relevant since incubation of cells in calcium-free buffer can alter basal cytoplasmic calcium and thereby prevent a subsequent response. This wouldgive a false indication of a dependence on extracellular calcium. We have excluded this potential artifact by direct measurements using a calcium-sensitive dye. Direct demonstration of the essential role of calcium was obtained by stimulating EC in the absence of calcium. This induced an activated state where PAF synthesis could be induced if, and only if, calcium was restored (Fig. 2). In addition to the regulatory role of calcium, stimulusinduced endothelial cell PAF production appears to be regulated by protein kinase C, a ubiquitous enzyme that regulates many cellular processes (13, 15). Our observations show that active phorbol esters increase endothelial cell PAF production, but only in the presence of a stimulus that increases intracellular calcium. Conversely, the inhibition of stimulated PAF production by protein kinase C inhibitors further suggests that protein kinase C activation is necessary for PAF synthesis. Importantly, the regulatory role of protein kinase C does not appear to be via alterations in calcium concentrations since we find that activation of protein kinase C by phorbol estershas no effect on 45Ca2+entry under basal conditions or in response to hormonal stimulation. Although protein kinase C has diverse intracellular effects, the regulatory effect of protein kinase C on stimulated PAFproduction appears to be at the level of phospholipase Az. This would explain the observation that protein kinase C activation causes an increase in stimulus-induced production of prostaglandin I2 from endothelial cells (20) and our observations that activation of protein kinase C causes an increase in stimulated release of [3H]arachidonate from endothelial cell phospholipids, including the specific PAF precursor, 1-0alkyl-2-arachidonoyl-GPC. This evidence strongly suggests that protein kinase C is regulatory at thelevel of the production of the immediate precursor of PAF, 1-0-alkyl-2-lysoGPC. This is also consistent with the recent observations of Slivka and Insel (62) that protein kinase C regulates phospholipase Az activity in cultured canine kidney cells. Conversely, other groups have found that phosphorylation increases the activity of acetyl-CoAlyso-PAF acetyltransferase (55, 57, 59). One report (55) suggests that this is due to the action of protein kinase Cbutothers find otherprotein kinases to be responsible (57, 59). Nonetheless, our studies have revealed an interesting circumstance in which the stimulus-induced production of two potent lipid autacoids, PAF and prostaglandin 12, is modulated by endogenous concentrations of another lipid, diacylglycerol. In summary, we have demonstrated that PAF production by endothelial cells is dependent upon increases in intracellular calcium that are mediated by hormone-induced activation of a calcium channel or by substances that create transplasma membrane channels (melittin, theta toxin). The regulatory effect of calcium is on thephospholipase A2-mediated production of the precursor of PAF, 1-0-alkyl-2-lyso-sn-glycero-3-phosphocholine. In addition, endothelial cell PAF production in response to a calcium signal is increased %fold when protein kinase Cisactivated by phorbol estersor diacylglycerol, an effect that also arises at a phospholipase AZ

The Role of Calcium and Protein Kinase in PAF Production

6333

(1987) Am. J. Physiol. 2 5 3 , 1400-1408 30. Stevens, D. L., Mitten, J. E., and Henry, C. (1987) J. Infect. Dis. 156,324-333 31. Bernheimer, A. W . (1977) in Mechanisms in Bacterial Toxinology (Bernheimer, A. W . , ed) pp. 85-97, John Wiley and Sons, Ltd., Acknowledgments-We thank Drs. William Barry, George Peters, New York and Osami Kohmoto for assistance in making the intracellular cal- 32. Blank, M. L., Robinson, M., Fitzgerald, V., and Snyder, F. (1984) cium measurements. Work in their lab is supported by grants from J . Chromatogr. 298,473-482 the Nora Eccles Treadwell Foundation and the National Institutes 33. Pearson. J. D.. Slakev, L. L., and Gordon, J. L. (1983) B ~ h e m . of Health (HL30478 and HL07576). We thank Donelle Benson for J. 214,2731276 technical assistance and Linda Jara for help in preparing the manu- 34. Kennedy, C., and Burnstock, G. (1985) Blood Vessels 22, 145script. Drs. Mark Elstad andDiana Stafforini provided helpful com155 ments throug~out the project. 35. Pirotton, S., Raspe, E., Demolle, D., Erneux, C., and Boeynaems, J-M. (1987) J. Biol. Chem. 262,17461-17466 36. Northover, B. J. (1980) Adv. Microcirc. 9,135-160 REFERENCES 37. Neher, E.(1987) Nature 326,242 1. Hanahan, D. J. (1986) Annu. Rev. Biochem. 55,483-509 38. Pozzan, T., Lew, D. P., Wollheim, C . B., and Tsien, R. Y. (1983) 2. Snyder, F. (1985) Med. Res. Rev. 5,107-140 Science 221,1413-1415 3. Malonen, M., Palmer, J. D., Lohman, I. C., McManus, L. M., and 39. Johnson, M.K., and Aultman, K. S. (1977) J. Gen. Microbial. Pinckard, R. N. (1980) Am. Rev. Resp. Dis. 122,915-924 101,237-241 4. Pinckard, R. N., McManus, L. M., and Hanahan, D. J. (1982) in 40. Hase, J., Mitsui, K., and Shonaka, E. (1975) Jpn. J. Exp. Med. Advances in I n ~ m m a t i o nResearch (Weissman, G., ed) Vol. 4, 45,433-438 pp. 142-180, Raven Press, New York 41. Seeger, W., and Suttorp, N. (1987) Am. Rev. Resp. Dis. 136,4625. Zimmerman, G. A., McIntyre, T. M., and Prescott, S. M. (1985) 466 J. Clin. Invest. 7 6 , 2235-2246 42. Dufourc, E. J., Smith, I. C . P., and Dufourcq, J. (1986) BiochemMcIntyre, T. M., Zimmerman, G. A., and Prescott, S. M. (1986) 6. istry 25,6448-6455 Proc. Natl. Acad. Sci. U. S. A. 8 3 , 2204-2208 43. Coddington, J. M., Johns, S. R., Willing, R. I., Kenrick, J. R., 7. Prescott, S. M., Zimmerman, G. A., and McIntyre, T.M.(1984) and Bishop, D. G. (1983) Biochim. Biophys. Res. Commun. Proc. Natl. A c d . Sci. U. S. A. 81, 3534-3538 727, 1-6 8. McIntyre, T. M., Zimmerman, G. A., Satoh, K., and Prescott, S. 44. Mix, L. L., Dinerstein, R. J., and Villareal, M. L. (1984) Biochim. M. (1985) J. Clin. Invest. 76, 271-280 Biophys. Res. Commun. 119,69-75 9. Whatley, R. E., Zimmerman, G. A., McIntyre, T. M., and Pres- 45. Hong, S. L., and Deykin, D. (1982) J. Bioi. Chem. 2 5 7 , 7151cott, S. M. (1988) Arteriosclerosis 8, 321-331 7154 10. Jaffe, E. A., Grulich, J., Weksler, B. B., Hampel, G., and Watan- 46. Chilton, F. H.,O'Flaherty, J, T., Ellis, J. M., Swendson, C. L., abe, K. (1987) J. Biol. Chem. 262,8557-8565 and Wykle, R. L. (1983) J. Biol. Chem. 2 6 8 , 7268-7271 11. Lambert, T. L., Kent, R. S., and Whorton, A. R. (1986) J. BioZ. 47. Albert, D. H., and Snyder, F. (1983) J. Biol. Chem. 268,97-102 Chern. 261,15288-15293 48. Whatley, R. E., Zimmerman, G . A., McIntyre, T. M., Taylor, R., 12. Higashida, H., Streaty, R. A., Klee, W., and Nirenberg, M. (1986) and Prescott, S. M. (1987) Semin. T h r o m ~ . H e m o s t ~ 13, is Proc. Natl. Acad. Sci. U.S. A. 8 3 , 942-946 445-453 13. Nishizuka, Y. (1986) Science 233,305-312 49. Ballester, R., and Rosen, 0.M. (1985) J. Bioi. Chem. 260,1519414. Rando, R. R. (1988) FASEB J. 2 , 2348-2355 15199 15. Bell, R. M. (1986) Cell 45,631-632 50. Fournier, A., and Murray, A. W. (1987) Nature 330,767-769 16. McIntyre, T. M., Reinhold, S. L., Prescott, S. M., and Zimmer- 51. Hannun, Y. A., Loomis, C . R., Merrill, A. H., Jr., and Bell, R. M. man, G. A. (1987) J. Biol. Chem. 262, 15370-15376 (1986) J. Biol. Chem. 261,12604-12609 17. Majerus, P. W . , Neufeld, E. J., and Wilson, D. B. (1984) Cell 3 7 , 52. Wilson, E., Olcott, M.C., Bell, R. M., Merrill, A. H., Jr., and 701-703 Lambeth, J. D. (1986) J. Biol. Chem. 2 6 1 , 12616-12623 18. Imine, R. F., and Moor, R. M. (1986) Biochem. J. 240,917-920 53. Loomis, C. R., and Bell, R. M. (1986) J. Biol. Chem. 2 6 3 , 168219. Michell, B. (1986) Nature 324,613 1692 20. Halldorson, H., Kjeld, M., and Thorgeirsson, G. (1988) Arterio- 54. Wykle, R. L., Malone, B., and Snyder, F. (1980) J. Biol. Chern. sclerosis 8, 147-154 255,10256-10260 21. Whorton, A. R., Willis, C . E., Kent, R. S., and Young, S. L. 55. Lenihan, D. J., and Lee, T-C. (1984) Biochim. Biophys. Res. (1984) Lipids 1 9 , 17-24 Commun. 120,834-839 22. Billah, M. M., and Siegel, M. I. (1987) Biochem. Biophys. Res. 56. Gomez-Cambronero, J., Nieto, M. L., Mato, J. M., and SanchezCommun. 144,683-691 Crespo, M. (1985) Biochim Biophys. Acta 845, 511-515 23. Zimmerman, G. A., McIntyre, T. M., and Prescott, S. M. (1985) 57. Domenech, C., Machado-De Domenech, E., and Soling, H-D. Circulation 72,718-727 (1987) J. Biol. Chem. 262,5671-5676 24. Blackwell, G. J., Duncombe, W.G., Flower, R. J., Parsons, M. F., 58. Lee, T-C., Lenihan, D. J., Malone, B., Roddy, L. L., and Wasserand Vane, J. R. (1977) Br. J. Pharmocol. 59,353-366 man, S. I. (1984) J. Biol. Chem. 259,5526-5530 25. Jaffe, E. A., Nachman, R. L., Becker, C. G., and Minick, C. R. 59. Nieto, M. L., Velasco, S., and Sanchez Crespo, M. (1988) J. Biol. (1973) J. Clin. Invest. 5 2 , 2745-2756 Chem. 263,4607-4611 26. Mueller, H. W . , O'Flaherty, J. T., and Wykie, R. L. (1983) J. 60. Betz, S. J., and Henson, P. M. {1980) J. Immunol. 125, 2756Biol. Chem. 258,6213-6218 2763 27. Bligh, E. G., and Dyer, W. J. (1959) Can. J. Biochem. Physioi. 61. Ludwig, J. C., McManus, L. M., Clark, P. O., Hanahan, D. J., 37,911-917 and Pinckard, R. N. (1984) Arch. Biochem. Biophys. 232,10228. Blank, M. L., and Snyder, F. (1983) J. Chromatogr. 273, 415110 420 62. Slivka, S. R., and Insel, P. A. (1988) J . Biol. Chem. 2 6 3 , 1464029. Peeters, G.A., Hlady, V., Bridge, J. H. B., and Barry, W. H. 14647

step. This evidence demonstrates an important role for calcium and protein kinaseC as intracellular regulatorsof endothelial cellPAF production.

"