Platelet lipidomic profiling: novel insight into cPLA₂α activity and its role in human platelet activation

Abstract

With a newer, more selective and efficacious cytosolic phospholipase A₂α (cPLA₂α) inhibitor available we revisited the role of cPLA₂α activity in platelet activation and discovered an even larger component of platelet signaling relies on this enzyme than was previously appreciated. In a whole blood shear-based flow chamber assay, giripladib, a cPLA₂α inhibitor, reduced platelet adhesion and accumulation on collagen. Moreover, giripladib differentially affected P-selectin expression and GPIIbIIIa activation depending on the agonist employed. While protease activated receptor (PAR)1 mediated platelet activation was unaffected by giripladib, PAR4 and GPVI mediated platelet activation were significantly reduced. Meanwhile, the thromboxane A₂ receptor antagonist SQ29548 had no effect on PAR, GPVI, or puriniergic receptor mediated platelet activation, suggesting that another eicosanoid produced downstream of arachidonic acid liberation by cPLA₂α was responsible for this large component of PAR4 and GPVI mediated platelet activation. In parallel, we profiled PAR mediated changes in glycerophospholipid (GPL) mass with and without giripladib to better understand cPLA₂α mediated lipid metabolism. Phosphatidylcholine and phosphatidylethanolamine (PE) demonstrated the largest consumption of mass during thrombin stimulation. Additionally, we confirm phosphatidylinositol as a major substrate of cPLA₂α. A comparison of PAR1 and PAR4 induced metabolism revealed the consumption of more putative arachidonyl-PE species downstream of PAR1 activation. Instead of enhanced cPLA₂α activity and therefore more arachidonic acid liberation downstream of PAR4, these results suggest that a novel eicosanoid is produced in response to platelet activation that represents a large component of PAR4 and GPVI mediated responses.

Graphical Abstract

Platelet stimulation with thrombin results in the production of thromboxane A₂ (TXA₂) through both platelet-thrombin receptors, protease activated receptor (PAR)1 and PAR4^{1, 2}. Circulating TXA₂ activates the TXA₂ receptor (TP), a G_q-coupled receptor that enhances or amplifies other platelet stimuli^{3, 4}. As a result the pathway regulating its generation has been an area of pharmacological interest for the prevention of platelet-dependent vascular occlusion for many decades. TXA₂ is produced by the action of cyclooxygenase-1 (COX-1) and TXA₂ synthase on arachidonic acid liberated from glycerophospholipids (GPLs). Platelet COX-1 is the target of aspirin, which is widely used for the prevention of myocardial infarction, stoke, and occlusive vascular events^{5, 6}.

The liberation of arachidonic acid from GPL sources occurs through the phospholipase A₂ family of enzymes. Platelets contain the 85 kDa cytosolic phospholipase A2 α (cPLA₂α) which requires micromolar Ca²⁺ for activity ⁷ and the 14-kDa secretory PLA₂ (sPLA₂)⁸, that requires millimolar concentrations of Ca²⁺ for activity. However, there is no evidence for the activity of sPLA₂ within the platelet cytosol. Additionally, evidence for activity of a calcium independent PLA₂ (iPLA₂) in the platelet, has been suggested based on the effectiveness of the iPLA₂ inhibitor bromoenol lactone (BEL) against TXA₂ production^{9, 10}. Indeed, arachidonic acid release has been observed in platelets isolated from cPLA₂α^−/−/sPLA₂-IIA^−/− mice suggesting the involvement of another PLA₂ isoform¹¹. However, genetic evidence in human platelets demonstrate that the majority of platelet TXA₂ generation occurs as a result of cPLA₂α activity¹².

The lack of an efficacious cPLA₂α inhibitor has prevented a detailed investigation into arachidonic acid mobilization in the platelet or its influence on platelet activation. However, pharmaceutical companies have recently increased efforts to target PLA₂ enzymes due to their relationship to the progression of a number of inflammatory diseases. The cPLA₂α inhibitor Wyeth-2 also known as giripladib and PLA-695, inhibits cPLA₂α mediated release of arachidonic acid from GPL sources, and was originally developed and tested for efficacy against arthritis. Clinical trials were halted due to unforeseen gastrointestinal pain (clinical trials.gov), however the compound remains the most specific and potent inhibitor of cPLA₂α activity available. In 2008 Wyeth reported that oral treatment with 3 or 100 mg/kg twice daily significantly reduced clinical and microscopic disease scores in a murine, collagen induced arthritis model; Similarly oral treatment with 3 or 10 mg/kg of giripladib in a K/BxN model of arthritis significantly reduced ankle swelling and microscopic disease scores (Drug Data Report 30(7), 611 (2008)). Given its efficacy, it has been repurposed and is being studied as a possible radiosensitizing agent in the treatment of radioinsensitive tumors^{13, 14}. We utilized giripladib to better understand cPLA₂α activity during GPL remodeling after thrombin stimulation and the role of cPLA₂α in platelet activation.

We used a lipidomic approach developed by the Brown lab ¹⁵ to profile changes in the mass of GPLs following stimulation with thrombin. The application of electron spray ionization-mass spectrometry to the field of lipidomics has accelerated our understanding of lipid metabolism by allowing the detection of lipid molecular species quickly from a small amount of source material. With an initial phase of separation by liquid chromatography prior to direct injection, lipid species from different classes can be identified from a complex mixture. Furthermore, the injection of internal standards has facilitated reliable quantitation of these lipid species. We profiled changes in GPLs at the species level, and challenged these patterns with the cPLA₂α inhibitor giripladib.

Experimental Procedures

Blood collection and platelet isolation

Human platelets were obtained from healthy volunteers. The studies were approved by the Vanderbilt University Internal Review Board. Informed consent was obtained from all individuals prior to the blood draw. Blood was collected into sodium citrate anticoagulant (final concentration 0.32%) through a 19 gauge needle. Washed platelets were prepared as previously described ¹⁶ and suspended in Tyrodes buffer (10 mM HEPES, 11.9 mM sodium bicarbonate, 127.2 mM sodium chloride, 5 mM potassium chloride, 0.4 mM sodium phosphate monobasic, 1 mM magnesium chloride hexahydrate, and 5 mM D-glucose) to 3.0×10⁸ cells/ml.

Flow Cytometry

For detection of P-selectin or GPIIbIIIa activation, platelets at 1.5×10⁷ cells/ml were preincubated with Allophycocyanin (APC) conjugated CD62P and Phycoerythrin (PE) conjugated PAC1 (BD biosciences, San Jose, CA) before a 10 min preincubation with inhibitor or vehicle followed by stimulation with the appropriate agonist for 10 min. Giripladib was a gift provided by Dr. Craig Lindsley and synthesized at the Vanderbilt Center for Neuroscience and Drug Discovery (Nashville, TN). SQ29584 and U46619 were from Cayman chemical (Ann Arbor, MI). PAR1-activating peptide (AP) (SFLLRN) and PAR4-AP (AYPGKF) were purchased from GL Biochem (Shanghai, China). Adenosine di-phosphate (ADP) was purchased from Sigma-Aldrich. Convulxin (CVX) was purchased from Santa Cruz (Dallas, TX). Thrombin was purchased from Enzyme Research Laboratories (South Bend, IN). Samples were fixed with 1% paraformaldehyde in PBS for 20 min before dilution of the samples with Tyrode’s buffer. Data were analyzed using FACS DiVa acquisition software (BD biosciences) and Winlist software (Verity Software House) for analysis. Mean fluorescence intensity was determined by collecting 100,000 events within the platelet gate.

Aggregation and ATP release

Washed platelets were suspended in Tyrode’s buffer at a density of 3.0×10⁸ cells/ml and loaded into glass cuvettes with stir bars. Aggregation (optical density) and ATP release (luminescence) were assessed simultaneously in a Chrono-log (Havertown, PA) model 700 lumiaggregometer. Platelets were preincubated for 10 min with DMSO or giripladib and aggregation/ATP release was monitored for at least 10 min.

TXB2 ELISA

Platelets were suspended in Tyrode’s buffer at a density of 3.0×10⁸ cells/ml and aliquoted into 500 μL volumes. Each sample was treated with either DMSO or giripladib before being stimulated with the indicated agonist. Platelets were immediately pelleted at 13,000xg for 3 min. Supernatants were collected carefully avoiding platelet pellets and immediately frozen for future analysis. Samples were analyzed for TXB₂ content using the Enzo TXB2 ELISA kit according to the manufacturer’s instructions.

Flow chamber assay

Experiments were conducted as described by Gailani et al¹⁷. Glass capillary tubes (1 mm x 0.1 mm) were coated overnight with 100 μg/ml collagen I (Chrono-log, Havertown, PA) at pH 4 in acetate buffer. The day of the experiment, tubes were rinsed once with Tyrode’s buffer then blocked with 0.5% fatty acid free BSA (Sigma-Aldrich, St. Louis, MO) in Tyrode’s for 1 hour at room temperature. Whole blood was collected into syringes filled with 1/10^th volume of 3.2% Na Citrate (final 0.32%). Blood was treated with DiOC₆ (Sigma-Aldrich) for at least 20 min, and inhibitor or vehicle for at least 10 min before perfusion through the capillary tube with a syringe pump. Both control and treated conditions are run in parallel to avoid unwanted effects of longer incubation periods in the vehicle or any time dependent changes in the reactivity of the blood. Prior to entering the capillary tube blood was mixed with 1/5^th volume of CaCl₂/MgCl₂ in HEPES-saline buffer (final concentrations: 2.5 mM CaCl₂, 1.25 mM MgCl₂). After 15 min of perfusion, blood is replaced with Tyrode’s buffer, and then 4% paraformaldehyde each for 1 min at a consistent flow rate. Images were capture with a 4X objective on an LSM710 META inverted microscope (Zeiss, Oberkochen, Germany). Analysis was performed using ImageJ. Volume was assessed by calculating total area for each stack then multiplying this value by the height of the stack before adding them together. % coverage was calculated from a z-project (sum of slices) picture as depicted in Fig 3A.

Figure 3. An iPLA₂β inhibitor reduces platelet activation.

A. The effect of 10 μM Varespladib, 5 μM R-BEL, and 5 μM S-BEL on P-selectin expression (A.) and GPIIbIIIa activation (B.) by 10 nM thrombin (Thr), 20 μM Protease activated receptor (PAR)1-AP, 200 μΜ PAR4-AP, 20 μM ADP, and 500 ng/ml Convulxin (CVX). SEM, n=4.

Preparation of samples, mass spectrometry, and data analysis

Platelets at a density of 3.0×10⁸ cells/ml were preincubated with vehicle control or giripladib for 10 min followed by stimulation with PAR agonists. Reactions were stopped with an equal volume of acidified methanol (0.1N HCl:MeOH (1:1)). Lipids were extracted using a modified Bligh and Dyer procedure ¹⁵. An equal volume of chloroform is added and after vortexing, samples were allowed to separate into two phases overnight at 4°C. The methanol layer is collected and dried under vacuum (Labconoco Centrivap Concentrator, Kansas City, MO). Samples were reconstituted in methanol:chloroform and spiked with known concentrations of standards (Avanti, Alabaster, AL) prior to analysis by electron spray ionization assisted mass spectrometry as previously described ^{18, 19}.

Results

cPLA₂α inhibition reduces PAR4 and GPVI mediated platelet activation

To determine the contribution of cPLA₂α activity to platelet activation we tested the effect of the cPLA₂α inhibitor giripladib on platelet activation induced by PAR, collagen receptor GPVI and purinergic receptor stimulation. P-selectin expression and GPIIbIIIa inside-out activation (detected with PAC1) were used as readouts of platelet activation. Preincubation with giripladib selectively inhibited PAR4-AP mediated platelet P-selectin expression by 46% (Fig 1A) with minimal effects on other agonists. Although technically significant, the effect of giripladib on P-selectin expression by 10 nM thrombin (7% reduction) and convulxin (10% reduction) was meager. No significant effects were noted on PAR1-AP, and ADP induced P-selectin expression was enhanced (Fig 1A). In contrast to P-selectin expression, GPIIbIIIa activation by all agonists was affected, however the largest reductions in activity occurred with PAR4-AP (62% reduction), followed by convulxin (45% reduction), ADP (33% reduction), PAR1-AP (30% reduction) and thrombin (29% reduction) stimulated platelets (Fig 1B). No inhibitory effects of giripladib were observed when platelet activation was monitored by aggregation regardless of the agonist employed (data not shown). However, significant reductions in ATP release from giripladib-treated platelets were noted for all agonists (Fig 1C). Free arachidonic acid is quickly utilized or reincorporated back into GPL sources and therefore transient and difficult to measure. In order to confirm efficacy we measured the effect of pre-incubation with giripladib on TXB₂ production. TXB₂ is the stable product of TXA₂ which is rapidly hydrolyzed in aqueous solutions. TXA₂ production relies on the liberation of arachidonic acid, therefore an effective cPLA₂α inhibitor should blunt the amount of TXB₂ detected in stimulated platelet supernatants. Preincubation with giripladib abolished TXB₂ production in response to platelet stimulation with thrombin or convulxin (Fig 1D). In parallel, we tested the effects of a TP antagonist, SQ29584, on platelet activation (Fig 1E and 1F). Despite complete inhibition of the thromboxane receptor agonist U46619-induced platelet activation (Fig 1E and 1F), SQ29584 had no effect on PAR, GPVI, or purinergic receptor mediated platelet activation. Therefore, an eicosanoid other than TXA₂ must be responsible for modulating cPLA₂α dependent components of PAR4 and GPVI mediated platelet activation.

Figure 1. cPLA₂α inhibition reduces PAR4 and GPVI mediated platelet activation.

The effect of giripladib P-selectin expression (A.) and GPIIbIIIa activation (B.) on platelet activation by 10 nM thrombin (Thr), 20 μM protease activated receptor (PAR)1-AP, 200 μΜ PAR4-AP, 20 μM ADP, and 500 ng/ml convulxin (CVX), SEM, n=8, except CVX n=4. C. The effect of giripladib on ATP release. SEM n=4. D. The effect of giripladib on TXB₂ release. SEM, n=3. The effect of SQ29584 on P-selectin expression (E.) and GPIIbIIIa activation (F.) by the aforementioned agonist and additionally 10 μM U46619 (U466) a thromboxane A2 receptor agonist. SEM, n=4, except U46619 n=3. Both giripladib and SQ29584 were used at 10 μM.

cPLA₂α inhibition reduces platelet accumulation on collagen under shear stress

To further demonstrate the importance of arachidonic acid in hemostasis and thrombosis, we tested the effect of giripladib on platelet accumulation on collagen under flow. Whole blood treated with giripladib or vehicle control was pushed through capillary tubes at volumetric flow rates that yielded a shear rate of 1500 s⁻¹. Citrated blood was recalcified prior to entering the capillary tube to allow thrombin generation to occur so that fibrin would be generated and PARs would also be engaged. Preincubation with giripladib significantly reduced platelet adhesion (as measured by % coverage) and accumulation (as measured by volume) (Fig 3A and 3B). These data demonstrate the significant role that cPLA₂α activity plays in the hemostatic response of blood.

An iPLA₂β inhibitor reduces platelet activation

We also tested the effect of sPLA₂ and iPLA₂ inhibitors on platelet activation (Fig 3A and 3B). The sPLA₂ inhibitor Varespladib had no effect on platelet activation. However, the iPLA2 inhibitors R-BEL and S-BEL had significant effects on PAR, collagen receptor, and purinergic receptor mediated platelet activation. R-BEL, which is specific to iPLA₂γ, significantly reduced P-selectin expression in response to PAR1-AP, however the magnitude of the reduction was negligible. S-BEL, which is 10-fold more specific to iPLA₂β significantly reduced platelet activation in response to each agonist.

Changes in glycerophospholipid mass in response to stimulation with thrombin

Given the dramatic effect of giripladib on PAR and Collagen mediated platelet activation, we sought a biochemical understanding of cPLA₂α activity on different GPL pools. We did not utilize arachidonic acid labeling as this technique relies on arachidonic acid reacylation and reincorporation into the GPL pool through the Land’s pathway, which will occur at different rates for each GPL class resulting in the preferential labeling of GPL pools with high turnover under resting conditions. Free arachidonic acid is rapidly oxidized by cycloxygenase and lipoxygenase enzymes to make a diverse array of eicosanoids. Due to its transient nature measurement of arachidonic acid is not a reliable assessment of cPLA₂α activity. We observed changes in GPL mass from the perspective of arachidonate. The current analysis prevents the determination of the mass and degree of saturation for individual acyl chains. However, determination of the combined mass of both acyl chains enables the inference of a GPL containing one 20:4 acyl chain. Therefore we present here only those lipids detected that putatively contain arachidonyl according to the appearance of their “−20:4” lysolipid counterpart during stimulation. Changes in all GPLs detected are presented in the supplement for the reader to peruse and summarized in Figure 4 C, D, and E. Figure 4A presents a time course of changes in putative arachidonyl-GPL species mass in human platelets challenged with 2 nM thrombin for up to 15 min. Changes in putative arachidonyl-GPL mass reach maximal levels by 5 min. The greatest changes in mass occur in the phosphatidylethanolamine (PE) and phosphatidylcholine (PC) pools, which experience a loss of roughly 1000 pmols. There were also significant losses in mass in the phosphatidylinositol (PI) pool as well. In contrast to the PE, PC, and PI pools the phosphatidic acid (PA) pool displayed a significant and persistent increase in mass of roughly 400 pmol in response to thrombin stimulation. Although we attempted to measure changes in the phosphatidylglycerol (PG), lipid levels were barely above the detection limit and were therefore excluded from the manuscript. Changes in phosphatidylserine (PS) mass were inconsistent between donors and therefore is presented in the supplement for the reader to peruse.

Figure 4. Thrombin induces changes in glycerophospholipid mass.

Human platelets were stimulated with 2 nM thrombin for the indicated timeframes. Glycerophospholipids (GPLs) were isolated and prepared for mass spectrometry as detailed in materials and methods. GPLs were quantitated according to known amounts of injected internal standards. Only species with putative arachidonyl are represented. The data is presented as the change in GPL mass relative to the unstimulated control for each experiment (n=3). A. Changes in the mass of each of 5 classes of arachidonyl-glycerophospholipid species including phosphatidic acid (PA), phosphatidylcholine (PC), phosphatidylethanolamine (PE), and phosphatidylinositol (PI). B. Changes in mass of PA separated into individual species. C. Changes in mass of PE separated into individual species. D. Changes in mass of PI separated into individual species. E. Changes in mass of PC separated into individual species. Ether linked species are denoted as XX:Xe and plasmenyl linked species are denoted as XX:Xp.

Figure 4B–F represents the same results as in Figure 4A segregated into GPL species at the 5 min time point. Across all classes of GPLs the largest changes in mass occurred in the 38:4 species. Multiple plasmenyl and ether linked arachidonyl-GPL species were detected in the PC and PE pools (Fig 4C and 4E) however there did not appear to be a preference for the breakdown of these particular species. Intriguingly, the PA mass created in response to thrombin stimulation was composed exclusively of a 38:4 PA species (Fig 4B).

We also tracked lysolipid species of each of the 5 major GPL classes. Most notably, we observed a persistent increase in lysophosphatidyl inositol (LPI). When separated into specific molecular species, all of the LPI mass was accounted for by an 18:0 LPI (Fig 4D), a strong indication that it is a product of the breakdown of 38:4 PI for the release of arachidonic acid (20:4). Various species of lysophosphatidylethanolamine (LPE) and lysophosphatidylcholine (LPC) were also detected (Fig 4C and 4E).

The effect of cPLA₂α inhibition on the loss of arachidonyl GPL and the production of lysolipids in response to thrombin stimulation

cPLA2α is considered the primary PLA₂ enzyme responsible for the liberation of arachidonic acid from GPL sources in human platelets. In order to probe the contribution of cPLA₂α activity to the observed changes in mass of arachidonyl-GPL, we preincubated platelets with the specific cPLA₂α inhibitor giripladib. Giripladib inhibited the loss of mass in the PC and PE pools, although this difference did not reach statistical significance in the PE pool. At early time points in the presence of giripladib, arachidonyl- PC and arachidonyl-PE displayed increases in mass suggesting a flux of arachidonyl-PC/PE, being created as it is consumed (Fig 5C and 5D). At later time points arachidonyl-PC and arachidonyl-PE mass continues to decrease leaving open the possibility of an additional lipase working on these pools. Loss of arachidonyl-PI was partially but significantly inhibited by giripladib (Fig 5B) suggesting that the activity of another enzyme, most likely phospholipase C (PLC) is responsible for the changes observed with thrombin.

Figure 5. cPLA2 inhibition blunts changes in putative arachidonly-glycerophospholipid mass induced by thrombin.

Platelets were preincubated with 10 mM Giripladib or vehicle control for 10 min prior to stimulation with 2 nM thrombin. The data is presented as the change in glycerophospholipid (GPL) mass relative to the unstimulated control for each experiment (n=3). A. Changes in the 38:4 phosphatidic acid (PA) mass. B. Changes in 38:4 phosphatidylinositol (PI) mass. C. Changes in the sum of putative arachidonly-phosphatidylcholine (PC) mass. D. Changes in the sum of putative arachidonly-phosphatidylethanolamine (PE) mass. E. Changes in the 18:0 lysophosphatidylethanolamine (LPE) mass. F. Changes in the 18:0 lysophosphatidylinositol (LPI) mass.

Preincubation with giripladib completely abolished the production of 18:0 LPI and 18:0 LPE (Fig 5E and 5F). Complete inhibition of the production of 18:0 LPI and 18:0 LPE by giripladib confirm the efficacy of this cPLA₂α inhibitor and cPLA₂α activity on the PE and PI pools. These results also confirm cPLA₂α activity as the source of the large 18:0 LPI signal. Finally, the gain in mass observed for 38:4 PA was blunted by giripladib suggesting that cPLA₂α activity is partially required for the production of this GPL (Fig 5A).

Comparing PAR1-AP and PAR4-AP mediated changes in GPL mass

To potentially gain some insight into the difference in the effect of giripladib on PAR1 and PAR4 stimulated platelets we conducted a similar experiment using PAR1 and PAR4 specific peptides (PAR1-AP, PAR4-AP) tracking changes in GPL mass in response to agonist over the course of 15 min paying attention to GPL pools that were affected by preincubation with giripladib. Figure 6 compares changes in GPL mass between PAR1-AP and PAR4-AP stimulated platelets. Looking across GPL classes major differences in metabolism were noted between PAR1 and PAR4 in the PE and PA pools. Significantly more 38:4 PA was produced downstream of PAR4. We also observed a preference for the consumption of arachidonyl-PE downstream of PAR1 activation, in direct contrast to the effect of giripladib on PAR1 and PAR4 responses. In conclusion, cPLA₂α activity as assessed by loss of putative arachidonyl-GPLs could not account for the difference in giripladib’s effects on PAR1 and PAR4.

Figure 6. Differences in PAR1 and PAR4 induced changes in glycerophospholipid mass.

Platelets were stimulated with 20 μM Protease activated receptor (PAR)1-activating peptide (AP) or 200 μM PAR4-AP for the indicated time frames and samples were prepared for mass spectrometry as detailed in materials and methods. A. Changes in 38:4 phosphatidic acid (PA) mass. B. Changes in 38:4 phosphatidylinositol (PI) mass. C. Changes in 38:4 phosphatidylcholine (PC) mass. D. Changes in putative arachidonyl- PC mass. E. Changes in 38:4 phosphatidylethanolamine (PE) mass. F. Changes in putative arachidonly-PE mass.

Discussion

The liberation of arachidonic acid from GPL sources for the production of TXA₂ has long been a subject of interest in the cardiovascular field due to the importance of TXA₂ to normal hemostasis and the effectiveness of Aspirin in the prevention of recurrent thrombotic events. However, the lack of a specific cPLA₂α inhibitor has prevented an assessment of the effect of blocking arachidonic acid release on platelet function as well as a detailed understanding of cPLA₂α substrates and therefore GPL sources of arachidonic acid. Giripladib was developed by Wyeth as part of a program to target various PLA₂ isoforms for the potential treatment of a spectrum inflammatory diseases with an eicosanoid component. In addition to the reported efficacy of giripladib in two different models of arthritis (Drug Data Report 30(7), 611 (2008)) the patent which first claimed giripladib reported an IC50 of 20 nM in a rat whole blood TXB₂ release assay and an IC50 of 30 nM in a GLU micelle assay containing only lipid, detergent and the enzyme. Efficacy was also demonstrated in the Platelet Function Analyzer which utilizes flow based shear and collagen to activate platelets, a FeCl₃-induced model of arterial thrombosis, and a mouse experimental autoimmune encephalomyelitis model of multiple sclerosis (WO/2006/128142, example 14). We used giripladib to probe cPLA₂α function in human platelets.

Aspirin, which inhibits the conversion of free arachidonic acid into TXA₂ has previously been demonstrated to affect platelet accumulation but not adhesion in ex vivo and in vivo models of thrombosis^20–25. The effects of TP antagonism on thrombosis have also been explored demonstrating good efficacy but similar phenotypes to aspirin treated blood^26–28. However, the effects of blocking cPLA₂α activity and therefore arachidonic acid release on platelet function under flow have yet to be determined. The effect of giripladib on platelet adhesion to collagen under flow highlights the importance of this pathway in platelet function during hemostasis and thrombosis. The inhibition of cPLA₂α has a much greater effect on platelet adhesion than administration of aspirin or the TP receptor antagonist SQ29584, which have been reported to affect only platelet accumulation in such models. These data indicate there is an additional unappreciated component of platelet signaling downstream of arachidonic acid release that can’t be accounted for by TXA₂ production.

Consistent with observations in the flow chamber model, preincubation with giripladib revealed that a large component of the PAR4 and GPVI mediated GPIIbIIIa activation was dependent upon arachidonic acid release. The partial reduction in response is likely due to the preclusion of the production of an eicosanoid otherwise produced downstream of arachidonic acid release that synergizes with PAR and GPVI mediated responses. The greater impact of giripladib on PAR4 versus PAR1 mediated platelet activation likely reflects the ability of this unidentified eicosanoid to synergize with PAR4 signaling better than PAR1. Perhaps PAR1 signaling does not rely on the unidentified eicosanoid production as much as PAR4. The residual PAR or GPVI mediated GPIIbIIIIa activation, P-selectin expression and ADP secretion is attributable to canonical receptor signaling which is independent of this putative eicosanoid-mediated augmentation of responses. Residual GPIIbIIIa activation, P-selectin expression, and ADP secretion could also be sufficient to induce full aggregation in response to maximal doses of the agonist and therefore explain the lack of effect of giripladib treatment on aggregation. Aggregation is conducted in glass cuvettes with a stir bar, essentially a closed system where the same partially activated platelets are continually exposed to one another. Platelets are also stimulated with maximal doses of agonist to induce full aggregation. In the flow chamber, however, naïve platelets are perfused past exposed collagen and previously activated/adherent platelets. Subtle gradients of agonists are generated in a flow system leading to partial platelet activation; activated platelets have fewer chances to adhere to and aggregate with other platelets before being pushed out of the capillary tube. Therefore, it is not unexpected to see a much greater response from an inhibitor in the flow chamber model versus aggregation.

A likely candidate for the elusive eicosanoid that synergizes with PAR4 and GPVI mediated responses was TXA_2. TXA₂ generation has previously been demonstrated to be significantly greater downstream of PAR4-AP versus PAR1-AP²⁹, suggesting that the disparity in the effect of giripladib could be the result of more TXA₂ generation, presumably as a result of more arachidonic acid liberation. Comparing changes in GPL mass in response to PAR1-AP and PAR4-AP, we did not observe the consumption of more putative arachidonyl-GPLs downstream of PAR4-AP, indicating that augmented cPLA₂α activity does not account for the large cPLA₂α -dependent component of the PAR4-AP response. Instead, this disparity is more likely due to the production of a unique eicosanoid downstream of PAR4 and not PAR1 or the activation of an unidentified receptor downstream of a commonly produced eicosanoid that synergizes with PAR4 and GPVI but not PAR1. In support of this, Holinstat et al. have recently published observations with a LOX inhibitor (ML355) that indicated a large component of the response downstream of PAR4 and GPVI, but not PAR1 is dependent upon LOX activity³⁰. Our data are in accord with this observation as we have blocked the production of this LOX-dependent eicosanoid by preventing the liberation of arachidonic acid, one step above LOX in the pathway. Moreover, they observe no effect of aspirin on platelet activation in agreement with our observations with the TP antagonist SQ29548 which had no effect on PAR or GPVI mediated platelet activation. The implication is another eicosanoid that is an even stronger feed-forward signal than TXA₂ is also produced by platelets. This eicosanoid appears to represent a previously unappreciated component of PAR4 and GPVI stimulated platelet activity. The molecular identity of this species remains undetermined. Identification of this eicosanoid represents a massive undertaking that is beyond the scope of this study.

In addition to profiling the effect of the cPLA₂α inhibitor giripladib on platelet activation we also looked at the potential involvement of sPLA₂ and iPLA₂ in platelet activation. sPLA₂ requires millimolar concentrations of Ca²⁺, is stored in granules and typically active against GPLs found in plasma. Therefore, we anticipated no response with the sPLA₂ inhibitor Varespladib in washed platelets. We did not anticipate any effect of the iPLA₂ inhibitors R-BEL or S-BEL. S-BEL, which significantly reduced platelet activation regardless of the agonist employed, is an inhibitor of iPLA₂β when used at 5 μM³¹. This data is in accord with other circumstantial evidence and BEL-based studies suggesting a role for iPLA₂ in arachidonic acid release and platelet activation. However, the BEL series, is known to have multiple cellular off-targets³² despite its sparing of cPLA₂α activity, therefore any conclusion with this inhibitor should not be taken as proof of this enzyme’s role in platelet activation. Indeed, this conclusion is difficult to resolve with the fact that giripladib abolishes TXB₂ production in platelets.

Seminal work in the field describes a hierarchical rank in the loss of mass from PE>PC>PI>PS during platelet stimulation with thrombin³³. Our results do not disagree with this ranking. We observed the most loss of mass in the PE and PC pools with minor losses from the PI pool. The preference of cPLA₂α for the PC and PE pools is consistent with the current understanding of this enzyme. Both PE and PC losses in mass were inhibited but not abolished by giripladib. However, at later time points, reductions in mass relative to the baseline continued, which leaves room for the interpretation that another lipase could be acting on these GPL pools. Consistent with this hypothesis arachidonic acid release is still observed in cPLA₂α^−/−/sPLA₂-IIA^−/− mice¹¹. Moreover, a recent study demonstrated the effectiveness of the iPLA₂ inhibitor BEL in blocking PLA₂ hydrolysis of plasmenyl-GPLs in the absence of Ca^{2+ 10}. Yet another study recently reported that iPLA₂γ deficient platelets demonstrated reduced ADP dependent aggregation and ADP or collagen dependent TXA₂ production³⁴. Indeed, we observed a reduction in platelet activation with the iPLA2β inhibitor S-BEL, suggesting that iPLA₂ activity may contribute to human platelet activation. The relative roles of these two PLA isoforms and the mechanism by which they contribute to platelet activation warrants further study.

Production of PC and PE species with cPLA₂α inhibition suggests a flux of arachidonyl GPLs through these pools and raise the possibility that arachidonyl-PC is produced during thrombin stimulation, but is not observed due to its rapid consumption by cPLA₂α. Although difficult to demonstrate experimentally, it is possible that the arachidonate is shuffled from a non-PC/PE pool to the PC/PE pool prior to its liberation from GPL sources by cPLA₂α. However, more specific inhibitors of transacylation enzymes are needed to definitively answer these hypotheses.

The formation of LPI has previously been observed in the field ³⁵. We have extended this observation and identified this species as 18:0 LPI. The cPLA₂α inhibitor significantly but only partially inhibited the loss of arachidonyl-PI, while abolishing the production of 18:0 LPI. Consistent with early observations in the field our results indicate that two different enzymes are consuming 38:4 PI during thrombin stimulation. A part of the 38:4 PI mass lost during thrombin stimulation can be accounted for by cPLA₂α activity. This is indicated by the effectiveness of giripladib in reducing this loss in mass and the arithmetic relationship between the mass of 38:4 PI preserved during cPLA₂α inhibition and the mass of 18:0 LPI lost by cPLA₂α inhibition, roughly 200 pmol. The remaining 100 pmol of 38:4 PI lost during thrombin stimulation that was not affected by the cPLA₂α inhibitor is most likely caused by PLCβ activity, a known effector of PAR activation in the platelet, responsible for subsequent Ca²⁺ mobilization and PKC activation. However, due to the lack of a specific PLC inhibitor, it is not possible to test this. The effective PLC inhibitors currently available have an inconvenient off-target effect on PLA type lipases. Moreover, Ca²⁺ mobilization mediated by PLC activity likely contributes to cPLA₂α activity rendering any experiment with a PLC inhibitor difficult to interpret when studying arachidonic acid release. However, these data definitively demonstrate that PI is a substrate of cPLA₂α in the human platelet and that the majority of PI lost during thrombin stimulation is a result of cPLA₂α activity.

Finally, it has recently been reported that thrombin stimulation of human platelets results in the selective hydrolysis of arachidonyl-plasmenylcholine and plasmenylethanolamine, with little activity towards diacyl phospholipids¹⁰. In accord with this study and others^{33, 36} we detected multiple ether-linked PC species and plasmenyl-linked PE species which demonstrated a loss in mass in response to stimulation with thrombin. However, in direct contrast to this study we did not observe a preference for these substrates over their diacyl counterparts.

This study demonstrates for the first time the effect of cPLA₂α inhibition on platelet activation, including defects in GPIIbIIIa activation, secretion, platelet adhesion and aggregation on collagen under flow. Importantly, these data indicate that arachidonic acid release and subsequent eicosanoid production represent a much larger component of platelet activation than was ever suggested by aspirin, TP receptor antagonists or previous cPLA₂α inhibitors. These effects appear to be unique to PAR4 and collagen receptor GPVI activation. Lipidomics suggest that there is no more loss in putative arachidonyl-GPL mass downstream of PAR4 versus PAR1, leading to the conclusion that perhaps a unique eicosanoid is produced downstream of PAR4 or GPVI activation or that an as yet unidentified eicosanoid receptor synergizes with PAR4 or GPVI, and not PAR1. These data suggest that there remains great potential in the pharmacological targeting of this signaling cascade to inhibit platelet activation and affect thrombosis. Moreover, they suggest caution be taken in a pharmacological strategy that would inhibit eicosanoid production and the platelet receptors with which they synergize.

Figure 2. cPLA₂α inhibition reduces platelet adhesion and accumulation under shear stress.

Citrated whole blood was preincubated with DiOC₆ for 20 min and giripladib or DMSO vehicle for 10 min prior to flow through a capillary tube coated with 100 μg/ml collagen and blocked with 0.5% fatty acid free BSA. Blood was flowed through the capillary tubes for 10 min at volumetric flow rates that yielded a shear rate of 1500s⁻¹. Immediately prior to entry into the tube blood was mixed with CaCl₂ and MgCl₂ to allow thrombin generation to occur. A. Shown are representative images from three independent experiments. Images were captured at 4x. Scale bar 500 μm. B. Quantification of three independent experiments. Top panel: Volume, Bottom panel: % coverage. SEM, n=3. C. Images of control samples without recalcification and collagen deposition.

Abbreviations:

TXA₂

thromboxane A₂

PAR

protease activated receptor

TP

thromboxane A₂ receptor

cPLA₂

cytosolic phospholipase A₂

GPVI

glycoprotein VI

COX-1

cyclooxygenase 1

PLA₂

phospholipase A2

sPLA₂

secretory phospholipase 2

iPLA2

inducible phospholipase A2

GPL

glycerophospholipid

APC

allophycocyanin

PE

phycoerythrin

AP

activating peptide

ADP

adenosine di-phosphate

CVX

Convulxin

GPIIbIIIa

glycoprotein IIbIIIa

PE

phosphatidylethanolamine

PC

phosphatidylcholine

PI

phosphtidylinositol

PA

phosphatidic acid

PG

phosphatidylglycerol

PS

phosphatidylserine

LPI

lysophosphatidylinositol

LPE

lysophosphatidylethanolamine

LPC

lysophosphatidylcholine