Tight Binding Inhibitors of 85-kDa Phospholipase A, but Not 14-kDa Phospholipase A, Inhibit Release of Free Arachidonate in Thrombin-stimulated Human Platelets*

An analogue of arachidonic acid in which the COOH group is replaced by a trifluoromethyl ketone group (COCF,) has recently been shown to be a tight binding inhibitor of the 85-kDa cytosolic phospholipase A, that is found in platelets and other cells (Street, I. P., Lin, H.-K., Lalibertb, F., Ghomashchi, F. G., Wang, Z., Perrier, H., Tremblay, N. M., Huang, Z., Weech, P. K., and Gelb, M. H. (1993) Biochemistry 32,5935-5940). This trifluoromethyl ketone inhibits most of the arachidonate release from the phospholipid pool in thrombin-stimulated human platelets at concentrations of 0-40 with 4 x 10' plateletdml. A structure-function analysis of related compounds reveals a good correlation between the inhi- bition of the purified phospholipase A, and the blockage

Essentially all cellular arachidonate is esterified at the sn-2 position of phospholipids (18). Phospholipase 4 ( P a ) , ' which selectively cleaves this ester linkage, has been considered as a prime candidate to catalyze the release of this polyunsaturated fatty acid (19)(20)(21)(22)(23)(24)(25)(26)(27)(28), although the role of lysophospholipase (29) or phospholipase C together with lipases (3) has also been proposed. Ambiguities in the definition of the pathway for arachidonate liberation also persist because of the lack of a suitable means of dissecting functional roles of different enzymes in intact cells. In addition, many cell types contain different classes of P w s . Platelets contain a secretable 14-kDa phospholipase A, ( s P~) with a high turnover number and broad specificity toward phospholipids with different fatty acyl chains and polar head groups (30)(31)(32)(33) and also an 85-kDa enzyme (cP-1 with a low turnover number but high specificity for phospholipids that contain sn-2 arachidonate (34-36). The cP& shows very little head group specificity; however, diacylglycerol is not a substrate (37-39). I n this report we show that in intact human platelets, most of the thrombin-induced arachidonate release can be blocked by specific competitive inhibitors of cPL4, whereas specific inhibitors of s P L 4 are without effect.

EXPERIMENTAL PROCEDURES
Materials-All cPLA, and SP-inhibitors (Fig. 1) used in this study were synthesized as described previously (13,17,33,40,41). Since the unsaturated fatty acid analogues can oxidize over time, stock solutions of these compounds were prepared in ethanol and stored at -80 "C under argon in vials with Teflon septa-lined screw caps. The punties of the unsaturated fatty acid analogues were routinely checked by thin layer chromatography on silica plates (40). Pure recombinant sPLA, and cPLA, were obtained as described (33,40). Phenidone, indomethacin, and bovine thrombin are from Sigma.
Preparation of Platelets and Analysis of Arachidonate Release-Blood was obtained by clean venipuncture from healthy human donors who did not take any drugs during the previous 10 days, and it was collected onto ACD (130 m M citric acid, 152 m M sodium citrate, 111 m M glucose; 3.5 ml of blood + 1.5 ml of ACD). Platelet-rich plasma was  room temperature. All platelet-rich plasma samples were pooled and centrifuged at 1600 x g for 5 min at room temperature.
The pellet obtained from the blood of 6-8 donors (13.5 ml of blood from each donor) was gently resuspended in 1. mg of bovine serum albumin in 0.15 ml of buffer was added. After 5 min the suspension was filtered through Sepharose 4B (Pharmacia Biotech Inc.) preequilibrated with modified Tyrode buffer containing 0.2% bovine serum albumin. The platelet count in the gel-filtered platelets was adjusted to 4 x 108/ml (determined with a Baker MK4EIC platelet counter) with modified Tyrode buffer. The final calcium concentration was adjusted to 1 mM with CaCl,. The suspension was kept a t 22 "C for 30 min and was used in experiments within 2 h. Platelets prepared in this way are not activated since they produce a full burst of free arachidonate after treatment with thrombin (presented below), and analysis in the aggregometer shows that they respond normally to commonly used agonists (42). For arachidonate release studies, 10 PM indomethacin and 0.25 mM phenidone (final concentrations) were added 30 min prior to the addition of arachidonic acid analogues.
All assays of arachidonate release were carried out with 0. Platelet aggregation in the samples was simultaneously monitored in a dual channel aggregometer (Chronolog 500, Chronolog Corp., Haverston, PA) for up to 5 min as described elsewhere (42).
In some experiments, arachidonate release was measured by mass spectrometry. Platelets were prepared, labeled with r3H1arachidonate, and stimulated with thrombin as described above except that 1 ml of platelet suspension (4 x 1 0 ' platelets) was used for each assay reaction. after the layers separated, 1.8 ml of organic phase was transferred to vials equipped with Teflon septa-lined screw caps. The samples were purged with argon or nitrogen and stored at -80 "C until analyzed. For analysis, the solvent was removed in a vacuum centrifuge (Speed-Vac). 100 p1 of freshly prepared 10% diisopropylethylamine (Aldrich) and 0.1% pentafluorobenzyl bromide (Pierce) in acetonitrile (Baker high performance liquid chromatography grade) were added to the residues. The capped vials were purged briefly with argon and placed in an oven at 60 "C for 15 min. After solvent removal in a Speed-Vac, 50 pl of toluene was added, and the vials were purged with argon and stored at -80 "C until analyzed. For analysis by mass spectrometry, portions of the samples were diluted 200-and 500-fold with toluene, and 1-1.11 aliquots were injected onto the gas chromatography column (DB-5, 0.25 pm, 30 m, J & W Scientific) installed on a Hewlett Packard 5890 gas chromatograph. The thermal program was 1 min at 200 "C, then 20 'C/min to 280 "C, and 3 min at 280 "C. The gas chromatograph is interfaced to a mass spectrometer operating in the detection mode of electron capture chemical ionization (Fissons Trio 100). The arachidonate esters elute at 5.85 min, and the ion intensities at 303.2 -c 0.5 and 311.2 0.5 atomic mass units, corresponding to the nondeuterated and deuterated arachidonate anionic fragments arising from the loss of the pentafluorobenzyl radical, respectively, were monitored. The amount of platelet arachidonate released was obtained as the moles of internal standard added to the samples multiplied by the ratio of areas of the ion chromatogram peaks.
Experiments with recombinant sPLA, were carried out by adding this enzyme (final concentration, 50 pg/ml) to 0.4-ml samples of platelet suspension described above for the experiments with inhibitors. After 5 min, the reaction was quenched and analyzed for [3Hlarachidonate release as described above. In some experiments, thrombin was added followed by the addition of sPLA, 5 min later, and the reaction was quenched after 1 min.
Analysis of sPLA, a n d cPLA, Znhibitors in Vitro-Inhibitors of cPLA, were analyzed in vitro using substrate composed of l-stearoyl-2-[l-14Clarachidonyl-sn-glycero-3-phosphocholine dispersed in micelles of Triton X-100 (40). AACOCF, and AACOCF,Cl are slow binding inhibitors of cPLA, (40). The potencies in Table I  Analysis of Thromboxane B, Production-Thromboxane B, production was measured by enzyme-linked immunosorbent assay using a n immunoassay kit (Cayman Chemicals, catalog no. 419031). The platelets were prepared and stimulated with thrombin as described above except that indomethacin was omitted. Each assay contained 4 x lo8 platelets/ml, and inhibitors were added as described above. The reaction mixture was diluted 5000-fold into the microtiter plates containing the enzyme-linked immunosorbent assay components. Each platelet incubation reaction was done in triplicate, and each enzyme-linked immunosorbent assay measurement was done in duplicate.

RESULTS
Effect of PLA, Inhibitors on Arachidonate Release in Thrombin-stimulated Human Platelets-Thrombin-induced arachidonate release from gel-filtered platelets in the presence of inhibitors of cyclooxygenase and lipoxygenase amounts to about 20% of the total arachidonate present in platelets, and the release occurs with a half-time of about 20 s (28,45). As shown in Fig. 2, arachidonate release decreases in the presence of increasing amounts of AACOCF, (an analogue of arachidonic acid in which the COOH group is replaced with a trifluoromethyl ketone, Fig. 1). The inhibition is described by a single inhibition constant and an IC,, of 8-10 V M measured in the presence of 4 x lo8 plateletdml. This bulk concentration of inhibitor corresponds to a mole fraction of inhibitor in platelet membranes of 0.04 under the reasonable assumption that AA-COCF, partitions mainly into the membrane phase. This calculation of mole fraction also assumes that AACOCF, is present in all of the various platelet membranes (4 x 1 0 ' plateletdml contain 200 PM total membrane lipid (46)), which is reasonable since AACOCF, can rapidly exchange between the membrane and aqueous phases. When reporting IC,, values for the action of a membrane-soluble inhibitor in cells, it is also important to specify the number of cells used (or more specifically the mole fraction of inhibitor in the membrane); the latter is related to the surface concentration of inhibitor in the membrane phase that controls the degree of inhibition. Most, if not all, of the thrombin-induced release of arachidonate is blocked at concentrations of AACOCF, above 30 VM (Fig. 2).

As a general rule, using cells labeled with [,H]arachidonate
to monitor agonist-induced release of this fatty acid can be misleading. This is because not all pools of cellular arachidonate may be radiolabeled to the same extent, and, thus, the release of L3H1arachidonate may not reflect the total mass of liberated arachidonate (6). However, under the conditions employed in this study, all major phospholipid classes are substantially radiolabeled with exogenously supplied arachidonate, and the specific radioactivity of the arachidonate in phosphatidylinositides is essentially the same as that of the cellular arachidonate pool (27). More importantly, as shown in Fig. 3 the amounts of [,H]arachidonate and the total mass of arachidonate (measured by mass spectrometry) released in thrombinstimulated platelets are correlated. The total amount of arachidonate released after stimulation with thrombin in the absence of AACOCF, is 4 nmol from 2 x lo8 platelets; this number agrees well with the previously reported yield of 3-4 nmol (27,28). AACOCF, has been recently shown t o be a tight binding inhibitor of the human cPLA, (401, and in vitro inhibition data  with the purified cPLA, is given in Table I. AACOCF, is more than 1000-fold less potent as an inhibitor of sPLA, (40). The extent of inhibition of arachidonate release is the same when the platelets were incubated with 7.5 ~MAACOCF, for 1,15, or 30 min prior to the addition of thrombin, suggesting that the trifluoromethyl ketone is not appreciably metabolized in these cells during 30 min.
Minor alterations to the structure of AACOCF, have a dramatic effect on the inhibition of arachidonate release, and the data in intact cells correlate well with the in vitro results ( Table  I). The compounds in which the trifluoromethyl ketone is reduced to the corresponding alcohol (AACH(0H)CFJ or replaced by a methyl ketone (AACOCH,) cause neither detectable inhibition of arachidonate release in platelets nor inhibition of purified cPLA,. The chlorodifluoromethyl ketone (AACOCF,Cl) is 8-fold less potent compared with AACOCF, in blocking arachidonate release in platelets and is 3-fold less potent as an inhibitor of purified c P L 4 . The fact that 18:3-COCF3, the trifluoromethyl ketone analogue of y-linolenic acid, is also a poorer inhibitor of arachidonate release is consistent with the finding that the cPLA, prefers sn-2 arachidonate-containing phospholipids over those with sn-2 trienoic fatty acids by about 2-4-fold (39). These structure-activity data together with the fact that AACOCF, blocks arachidonate release when present in platelets at low mole fractions strongly argue that the compound is acting as a specific inhibitor of cP&. None of the compounds listed in Table I induce lysis of platelets as shown by the lack of appearance of lactate dehydrogenase in the extracellular fluid (data not shown).
The fatty acid amides (Table I) Table I) have previously been shown to inhibit the action of the sPLA, in vitro with dissociation constants in the range of 0.002-0.2 mole fraction (17,33). None of these compounds produce detectable inhibition of purified human recombinant cPLA,.' As summarized in Table I, no inhibition of arachidonate release is observed when these sPLA, inhibitors were added to platelets. Platelet sPLA, is present in granules and is secreted extracellularly following activation with thrombin (30)(31)(32). Purified human recombinant s P L 4 (33) (50 pg/ml), added t o a suspension of platelets under the conditions of the experiments shown in Fig. 2, causes no noticeable release of arachidonate. In these experiments the concentration of extracellular calcium is 1 mM and is sufficient to serve as a catalytic cofactor of sPL& (33). Furthermore, no additional arachidonate formation is detected when s P L 4 is added to platelets just following the addition of thrombin compared with levels of arachidonate formed when thrombin is added alone.
The inhibitors listed in Table I did not produce any effect on thrombin-induced aggregation of gel-filtered platelets (data not M. H. Gelb and I. P. Street, unpublished data. shown). The aggregation response in the presence of sufficient AACOCF, to cause r90% inhibition of thrombin-induced arachidonate release is the same as the control samples. Similar results were obtained when aggregation was initiated with 5 pg/ml collagen.
Inhibition of Thromboxane B2 Production-The sPLA, and cPLA, inhibitors were tested for their abilities to block thromboxane B, formation in thrombin-stimulated platelets. Stimulation of platelets with thrombin leads to the production of 118 ng of thromboxane B&O* platelets, whereas only 1 ng is measured in the absence of thrombin ( Table 11). As summarized in Table 11, none of the s P L 4 inhibitors significantly reduce the amount of thromboxane B, formation when tested at concentrations up to 40 p~. On the other hand, AACOCF, and 18:3-COCF, did inhibit thromboxane B, formation in a dose-dependent manner (Table 11), and the dose responses are similar to those measured for the inhibition of arachidonate release (Fig.  2). However, as also shown in Table 11, AACOCF,Cl inhibited thromboxane B, formation at concentrations less than those required to inhibit arachidonate release, and AACH(OH)CF, inhibits the formation of this eicosanoid when added to platelets at concentrations of 10 and 30 p~, even though this compound does not inhibit arachidonate release.

DISCUSSION
The results obtained in this study provide pharmacological evidence for the role of cPLA, and not sPLA, in producing arachidonate that is released following the stimulation of human platelets with small amounts of thrombin. Inhibition studies alone do not provide unequivocal proof for the role of cPLA, in arachidonate liberation, but the good agreement between the structure-function data obtained with the arachidonate analogues for the inhibition of cPLA, in vitro and arachidonate release in platelets ( Table I) further strengthens the role of this enzyme in initiating the eicosanoid cascade.
The fact that most, if not all, of the thrombin-induced arachidonate release is blocked by addition of cPLA, inhibitors suggests that the contribution of other pathways, if any, is minor. Based on this direct evidence, the involvement of a pathway for arachidonate release involving phosphatidylinositide-specific phospholipase C and diglyceride lipase in thrombin-stimulated platelets seems highly unlikely. Other studies also support this conclusion (7, 8, 27, 28, 47, 48). There is a general consensus

Phospholipase A,-catalyzed Arachidonate
Release 15629 that thrombin induces release of about 20% of the total arachidonate present in platelet phospholipids over a period of 1 min and that about one-third of this (27 nmol from a total of about 85 nmol in 5 x lo9 platelets) is from phosphatidylinositides (27,28). Thus, the pathway involving a phosphatidylinositide-specific phospholipase C and a diglyceride lipase could possibly account for a maximum of 30% of the released arachidonate. Even this does not appear to be the case on the basis of the intrinsic rates and specificities of the enzymes involved in the diglyceride pathway (3)(4)(5)(6)(7)(8) and from the observation that inhibition of diacylglyceride lipase has little effect on the thrombininduced release of arachidonate in intact platelets (7,8i. On the other hand, as shown here, specific inhibition of c P L 4 accounts for most, if not all, of the arachidonate release in platelets, which stresses the capital importance of this enzyme over other proposed pathways.
Additional recent studies suggest a role of cPLA, in agonistinduced release of arachidonate. Overproduction of cPLA, in Chinese hamster ovary cells transfected with the gene for this enzyme leads to an enhancement in the release of arachidonate triggered by thrombin, and such treatment results in the phosphorylation of cPLA, most likely by mitogen-activated protein kinase (49,50). Phosphorylation of cPLA, also occurs in thrombin-stimulated platelets (361, agonist-treated HL60 cells (511, and rat macrophages (521, and this post-translational modification may act as a switch to initiate phospholipid hydrolysis. All of the sPLA, inhibitors were without effect on thrombinstimulated arachidonate release. The phospholipid analogues HK42 and MJ45 are potent inhibitors of sPLA, in vitro, producing 50% inhibition when present in substrate vesicles at concentrations of 1 inhibitod250 or 50 substrate phospholipids, respectively (33). The fatty acid amides inhibit sPLA, when present in vesicles at concentrations of 1 inhibitodseveral hundred substrates (17). Based on their chemical structures, it is highly unlikely that these compounds can be metabolically altered. Although the fatty acid amides almost certainly can flip to the cytosolic face of the plasma membrane as well as migrate to internal membranes, the phospholipid analogues may remain only in the extracellular face of the plasma membrane. However, the platelet sPLA, seems t o function extracellulary since it is secreted from platelets after stimulation with thrombin (30)(31)(32).
It is surprising that the addition of an enormous amount (50 pg/ml) of sPLA, did not liberate arachidonate from platelets.
The sPLA, in vitro shows a strong requirement for anionic phospholipids for binding to membranes (33). The outer leaflet of the platelet plasma membrane takes on a significant negative charge, due to a redistribution of phosphatidylserine from the inner to outer membrane leaflets after the cells are activated with thrombin (53). However, the presence of extracellular sPLA, did not alter the level of thrombin-induced arachidonate release. The s P L 4 probably acts on cells other than platelets. Recent studies have shown that platelets that have secreted 60% of their total sPLA, still produce normal amounts of thromboxane B, when stimulated with thrombin (54). In addition, inhibition of the release of the intragranular content of platelets by the antithrombotic agent ajoene has no effect on thrombin-induced arachidonate release (55). All of these results suggest that sPLA, does not catalyze the release of arachidonate in thrombin-stimulated platelets. The possible importance of cPLA, and not sPLA, in increasing the production of prostaglandin E, in interleukin-1P-stimulated rheumatoid synovial fibroblasts has been proposed (56). cPLA, is almost certainly not the only enzyme responsible for arachidonate liberation in cells. Gross and co-workers (14) have found that an inhibitor of the calcium-independent phospholipase 4 found in muscle is able to decrease arachidonate re-lease in [Ar$]vasopressin-induced aortic smooth muscle cells. Platelet homogenates do not contain calcium-independent phospholipase 4 (36). Expression of antisense RNA against sPLA, inhibits arachidonate release and prostaglandin E, formation in the macrophage-like cell line P388D, induced by platelet-activating factor or lipopolysaccharide (57). The compound MJ33 has been recently shown to inhibit the release of oxidized fatty acid from the sn-2 position of lung surfactantderived phospholipid and to protect against ischemia reperfusion injury without affecting the production of thromboxane B, (58,59). These results underscore the fact that multiple PLA,s exist that have distinct biological functions. The role of different enzymes cannot by delineated simply on the basis of enzymatic activity alone but rather on the effects of enzyme-selective inhibitors.
All of the arachidonic acid analogues tested in this study inhibited the formation of thromboxane B, in thrombin-stimulated platelets. Additional studies show that these compounds inhibit the conversion of exogenously supplied arachidonate into thromboxane B, in intact platelets which suggests that they inhibit cyclooxygenase (60). It appears that the hydrocarbon chain, rather than the COOH group, of arachidonic acid is being recognized by cyclooxygenase. These results point to the importance of using inhibitors of the enzymes that utilize arachidonic acid as substrate when testing the effect of AA-COCF, on arachidonate release especially with those cell types that oxygenate a large fraction of the liberated arachidonate. In this way the total amount of liberated arachidonate rather than the net amount (liberated minus oxygenated) is being measured, and this is due solely to the action of cPLA,.
On the other hand, none of the arachidonate analogues inhibited the conversion of exogeneous arachidonate into 12-hydroxyeicosatetraenoic acid in platelets which suggest that these agents do not inhibit 12-lipoxygenase (60). Among the arachidonate analogues, only AACOCF, prevented the formation of 12-hydroxyeicosatetraenoic acid in platelets treated with calcium ionophore which suggests that the arachidonate production that is blocked by c P L 4 inhibition is converted, in part, to 12-hydroxyeicosatetraenoic acid (60).
The lack of effect of the c P L 4 inhibitors on platelet aggregation supports the idea that thromboxane 4, which is formed intracellularly and released into the medium, acts mainly as a secondary agonist intended to reinforce the action of weaker platelet agonists (61). It has been proposed that platelet activating factor is produced by the combined action of a phospholipase 4 (hydrolysis of l-O-alkyl-2-arachidonyl-sn-glycero-3-phosphocholine) and an acetyltransferase (acetylation of the lysophospholipid) (62). More recent work has generated evidence that a phospholipase 4 initiates the synthesis of platelet activating factor by hydrolysis of plasmalogens (mainly plasmenyl-phosphatidylethanolamine) to form lyso-plasmalogens. The latter is a substrate for a coenzyme A-independent transacylase that transfers the arachidonyl chain from 1-0-alkyl-2-arachidonyl-phosphatidylcholine to lysoplasmalogen to form 1-0-alkyl-lysophosphatidylcholine, which is acetylated to produce platelet activation factor (62). Since AACOCF, does not inhibit this transacylase, it should be useful as a tool t o probe the role of the cPLA, in the production of platelet activating factor. The active site-directed nature of the inhibitors reported in this study has been established by independent methods (17,33, 401, and the results described here extend the correlation between the in vitro inhibition data to corresponding effects in intact cells. This allays concerns about the distribution and local concentration of inhibitors in the various lipidic compartments, which ultimately control the concentration of the inhibitor that the enzyme encounters. Thus, a set of highly specific inhibitors such as those reported here should be useful in delineating the role of other phospholipases in cellular processes.