Ask1 regulates murine platelet granule secretion, thromboxane A₂ generation, and thrombus formation

The mitogen-activated protein kinase (MAPK) signaling cascades are well conserved in eukaryotic cells.¹  There are 3 MAPK families: extracellular signal-regulated kinases (ERKs) that are mainly activated downstream of growth factors, c-Jun N-terminal kinases (JNKs), and p38 MAPKs that are activated by cellular stress.^2-4  Activation of each MAPK occurs through a sequential phosphorylation by 2 classes of serine/threonine kinases. MAPK is phosphorylated first by MAPK kinase (MAP2K/MEK/MKK), which is in turn phosphorylated by MAP2K kinase (MAP3K/MEKK/MKKK). In platelets all 3 MAPKs are expressed and are activated downstream of physiological agonists.⁵  Although ERK2 knockout mice are not viable,⁶  studies using pharmacological inhibitors showed prolonged arteriole occlusion time, uncovering an important role for these MAPK in regulating thrombosis.⁷  Genetic ablation of JNK1 results in defective thrombosis, primarily due to a defect in granule secretion.⁸  Although homozygous p38α knockout mice die in utero,⁹  heterozygous mice show defective thrombosis.¹⁰ 

Apoptosis signal-regulating kinase (ASK1) is a well-characterized member of the MAP3K family, which is activated by a number of cellular stressors, such as oxidative stress, endoplasmic reticulum (ER) stress, UV exposure, and inflammatory cytokines.^11,12  ASK1 regulates the JNK and p38 MAPK pathways by phosphorylating MKK4/MKK7 and MKK3/MKK6, respectively, in nucleated cells.¹³  Activation of ASK1 during cellular stress leads to apoptosis.¹³  In nucleated cells, ASK1 is held in an inactive state through the association with one of several endogenous inhibitors, such as thioredoxin (Trx), 14-3-3, and calcium and integrin binding protein 1 (CIB1).^14-16  Replacement of these inhibitors by activators, such as tumor necrosis factor-α receptor-activated factor (TRAF) 2 and TRAF6 results in the activation of ASK1 by autophosphorylation of a threonine residue (human T⁸³⁸; murine T⁸⁴⁵) in the activation loop.¹⁷  Several molecular mechanisms were reported to regulate ASK1 in the nucleated cells.¹⁸  Reactive oxygen species (ROS)-mediated dissociation of Trx and Ca²⁺-dependent dissociation of CIB1 activate ASK1,^14,15  and TRAF2 and TRAF6 binding facilitate ASK1 autophosphorylation and activation.^17,19,20  Conversely, 14-3-3ζ binding inhibits ASK1 activation.¹⁶ 

We have previously reported that CIB1, which plays an important role in platelet function,²¹  regulates ASK1 in cells.¹⁵  However, expression of ASK1 in platelets is not known. Here, we show for the first time that ASK1 is expressed in both human and murine platelets and is rapidly activated by physiological agonists. Disruption of Ask1 in mice results in impaired platelet functions in part due to failure of agonist-induced TxA₂ generation, a consequence of impaired p38 MAPK–dependent activation of cPLA₂, resulting in protection of mice from arterial thrombosis.

Adenosine 5′-diphosphate (ADP) and collagen were purchased from Chronolog. AYPGKF was purchased from AnaSpec. Human fibrinogen (Fg) and α-thrombin were purchased from Enzyme Research. Anti-ASK1, anti-HSC-70, anti-CD36, anti-GP1bα, anti-glycoprotein VI (GPVI), and anti-αIIb were obtained from Santa Cruz Biotechnology. Phosphospecific-p38, P-ERK1/2, P-ASK1 Thr845, P-MKK4, P-MKK7, P-MKK 3/6, P-JNK, anti-p38, anti-MKK3, anti-MKK4, anti-MKK7, and anti-JNK were from Cell Signaling. Fluorescein isothiocyanate (FITC)-Fg, FITC-anti-P-selectin, anti-mouse CD41, and anti-JAM-A were obtained from BD Pharmingen. Anti-mouse CLEC-2, clone 17D9, was obtained from Biolegend. Anti-β3 and anti-PAR4 were obtained from EMD Millipore. Convulxin was from Pentapharm. Phycoerythrin (PE)-conjugated JON/A, PE-conjugated GPIbα, and the control immunoglobulin G (IgG) were obtained from Emfret. U44619, 2MeSADP, and all other chemicals unless indicated were of analytical grade from Sigma. PE-conjugated anti-GPVI was from R&D Systems.

Congenic Ask1^−/− mice in C57BL/6 genetic background generated by backcrossing for more than 10 generations was described previously.²²  Age (8-16 weeks) and gender-matched wild-type (WT) C57BL/6 mice were used as controls for all in vivo and in vitro experiments. Integrin β₃^−/− mice were obtained from Jackson Laboratories. Approval for animal experimental studies was received from the institutional animal care and use committees of the University of Delaware and Thomas Jefferson University.

Whole blood was drawn by venipuncture from healthy adult volunteers (of both gender and race) with informed consent. Approval was obtained from the institutional review boards of the University of Delaware and Thomas Jefferson University, according to the Declaration of Helsinki. Blood was collected in acidified citrate dextrose as an anticoagulant. Human and murine platelet-rich plasma (PRP) and washed platelets were prepared, as has been previously described.^23,24  Platelet aggregation was performed using PRP containing 2 × 10⁸/mL platelets using a Chrono-Log Lumi-Aggregometer (Chrono-Log), as has been described.²⁵  Aggregation traces were recorded using Aggrolink software (Chrono-Log).

An aliquot of 100 μL of washed platelet suspension (4 × 10⁸ platelets/mL) in Tyrode’s buffer was resting or stimulated with thrombin (0.1 U/mL), ADP (20 μM), U46619 (5 μM), epinephrine (2 μM), or convulxin (10 ng/mL) at various time points, and the reaction was stopped by adding 5× reducing sample buffer. Immunoblotting was performed, as has been described.²⁶ 

P-selectin surface expression, FITC-labeled Fg, and PE-labeled JON/A binding to washed murine platelets (0.6 × 10⁸/mL) resting or activated with agonist were measured by flow cytometry (Accuri C6, BD), as has been described.²⁵ 

Tail bleeding of age- and gender-matched WT and Ask1^−/− mice (4-6 weeks) was performed under anesthesia prior to genotyping, as has been described.²⁵  Time (in seconds) required for cessation of blood flow was recorded.

Platelet transfusion experiments were performed, as has been described.²⁷  In recipient WT and Ask1^−/− mice, platelets were depleted by injecting anti-CD41 antibody, as has been described.²⁸  After 24 h the reduction in platelet counts was assessed by Hemavet. Donor platelets isolated from WT and Ask1^−/− mice were transfused into the recipient mice, and the FeCl₃-induced carotid artery thrombosis assay was performed as has been described.²⁵ 

FeCl₃-induced carotid artery thrombosis assay was performed, as has been described.²⁵  Briefly, the carotid arteries of anesthetized WT and Ask1^−/− mice were exposed, and an injury was inflicted upon the arteries by placing a piece of Whatman No. 1 filter paper (1 mm × 1 mm) saturated with freshly prepared 10% FeCl₃ (anhydrous) for 2 min. The time for complete occlusion (lack of detectable blood flow) was recorded.

Pulmonary thromboembolism assay was performed, as has been described.²⁴  Briefly, a mixture of collagen (0.4 mg/kg; Chronolog) and epinephrine (60 mg/kg; Sigma) in 100 μL of phosphate-buffered saline (PBS) were administered through the tail-vein injection into anesthetized WT and Ask1^−/− mice. Two minutes after the onset of respiratory arrest, but while the heart was still beating or at the completion of the 10-minute observation period, 0.5 mL of Evans blue solution (1% in saline) was injected into the heart. Lungs were excised, photographed with a Nikon Coolpix camera, formalin fixed, and embedded into paraffin. Lung sections were stained with hematoxylin and eosin (H&E).

¹⁴C-serotonin release was performed, as has been described.²⁹  PRP from WT and Ask1^−/− mice was incubated with 0.1 μCi of ¹⁴C-serotonin for 1 h at 37°C. Incorporation was stopped by adding imipramine (1 mM). An aliquot of 100 μL (2.5 × 10⁸ platelets/mL) was stimulated with or without various concentrations of thrombin for 5 minutes, and the reaction was stopped by adding equal volume of stopping solution (2 mL of 1× PBS, 0.25 mL 37% formaldehyde, 2.25 mL of 77-mM EDTA pH 7.5). Another set of unstimulated aliquots of 100 μL were lysed with an equal volume of 2% Triton-X (control for serotonin uptake in platelets). An aliquot (150 μL) of individual sample supernatants was counted in a liquid scintillation counter. The percentage of ¹⁴C-serotonin release was calculated using the following formula:

Thromboxane generation assay was performed by using an enzyme immunoassay kit (Enzo Life Sciences).²⁶  Briefly, 100 μL of washed WT and Ask1^−/− platelets (2.0 × 10⁸/mL) were untreated or pretreated with pyrrophenone (1 μM) and stimulated with various agonists for 10 minutes. The level of thromboxane B₂, the stable metabolites of TxA₂, was determined from supernatant (1:10), according to the manufacturer’s instructions.

Microfluidic flow assays were performed in a 4-channel device with channel dimensions of 100 μm in height and 500 μm in width over collagen (200 μg/mL) that was absorbed into the clean glass surface, as has been described.³⁰  Briefly, blood from WT and Ask1^−/− mice (9-16 weeks) was collected using heparin (2 U/mL) and 40 μM of d-phenylalanyl-l-prolyl-l-arginine chloromethyl ketone (PPACK; Calbiochem). Whole blood was diluted 1:2 in Tyrode’s buffer and incubated with DiOC6 (1 mg/mL; Sigma) for 10 minutes at 37°C. Blood was perfused through the flow channel for 3 minutes at a shear rate of 800 s⁻¹. The chamber was rinsed with Tyrode’s buffer for 3 minutes and fixed with 5% paraformaldehyde for 30 minutes. Images were taken with a ×20 objective on an EVOS microscope (Thermo Scientific) and analyzed with Image J (National Institutes of Health); fluorescence intensity was calculated by using a NIES-Elements AR 3.2 (Nikon).

Intracellular calcium levels were measured, as has been described.³¹  Washed platelets from WT and Ask1^−/− mice (1.0 × 10⁷/mL) were incubated with 2.5 μM Fluo-4-AM (Life Technologies) for 1 minute in calcium-free Tyrode’s buffer at 37°C. Platelets were stimulated in the presence of 1 mM of extracellular Ca²⁺ with various agonists, following 60-s baseline, and fluorescence intensity was measured by Accuri C6 flow cytometer (BD).

Murine whole blood (50-75 μL) was drawn by retro-orbital bleed into a tube containing 1 μL (10 units) of heparin to prevent coagulation. The blood cell analysis was performed using a Hemavet 950 (Drew Scientific).

Occult fecal blood was detected using Hemoccult kit (Beckman Coulter), following the manufacturer’s instructions.

Transmission electron microscopy (TEM) was performed, as has been described.³²  Ultrathin counterstained sections were imaged by a Zeiss Libra 120 TEM, equipped with a CCD camera (Zeiss).

Statistical analysis of the data was performed using Prism 6 (GraphPad). Student’s t test (mean ± standard error of the mean) or Fisher’s exact test were used to determine statistical significance. P ≤ .05 was regarded as statistically significant. Each experiment was repeated independently at least 3 times.

A search of platelet proteomic databases showed that ASK1 (MAP3K5)—but not ASK2 (MAP3K6) and ASK3 (MAP3K15), two other members of the ASK1 family—is expressed in platelets.^33,34  We found expression of ASK1 in both human and murine platelets (Figure 1A).³⁵  To determine whether ASK1 is activated during human platelet stimulation by physiological agonists, we immunoblotted resting and activated washed human platelet lysates using antiphospho-T⁸⁴⁵ (P-T⁸⁴⁵) ASK1. P-T⁸⁴⁵ASK1 was undetectable in resting platelet lysates. ASK1 was rapidly phosphorylated by thrombin within 1 minute, which peaked at 3 minutes and then decreased (Figure 1B). ADP also activated ASK1 with kinetics similar to that of thrombin; however, the extent of activation was slightly lower (Figure 1C). Convulxin, a GPVI-specific agonist, robustly increased ASK1 phosphorylation with similar kinetics to that of thrombin and ADP (Figure 1D). U46619, a TxA₂ mimetic, also strongly induced ASK1 phosphorylation. However, unlike ADP and thrombin, U46619-induced ASK1 phosphorylation persisted beyond 5 minutes (Figure 1E). Epinephrine on the other hand did not activate ASK1 (Figure 1F). These results strongly suggest that most platelet agonists are capable of activating ASK1 within a rapid time scale that is relevant to platelet activation.

Because ASK1 is robustly and rapidly activated by platelet agonists, we evaluated platelet functions in Ask1^−/− mice. As was expected, Ask1^−/− platelets do not express Ask1 protein. However, we found no difference in expression levels of major platelet membrane proteins, nor in signaling proteins associated with MAPK signaling pathway between WT and Ask1^−/− platelets (Figure 2A-B). The surface expression of CD41, GP1bα, CLEC-2, and GPVI was not affected because of the lack of Ask1 in platelets (Figure 2C). Platelet aggregations induced by a low dose of thrombin, U46619, collagen, and ADP were severely affected in Ask1^−/− mice in comparison with WT mice (Figure 2D). However, the aggregation defect was overcome by a higher dose of agonists (Figure 2D). Furthermore, we found a significant attenuation of both JON/A- and Fg-binding induced by AYPGKF in Ask1^−/− platelets in comparison with WT platelets, indicating a defect in integrin α_IIbβ₃ activation (Figure 2E-F). These results suggested that Ask1 is required for agonist-induced platelet activation.

It is known that intracellular Ca²⁺ rise is one of the common and early steps of platelet activation. Platelet agonist-induced intracellular Ca²⁺ rise was unaffected in Ask1^−/− platelets in comparison with WT platelets, suggesting that Ask1 functions downstream of Ca²⁺ release (Figure 3A-B). Thrombin induced a dose-dependent Ca²⁺ rise, which was similar in both WT and Ask1^−/− platelets (Figure 3Bii). On the other hand, thrombin-induced dense granule secretion was significantly attenuated in Ask1^−/− platelets in comparison with WT platelets (Figure 3C). Total ¹⁴C-serotonin incorporation was comparable in platelets of both genotype, suggesting that the defect is not in the granular content of Ask1^−/− platelets. We also found that agonist-induced ATP release was also significantly lower in Ask1^−/− platelets than in WT platelets (Figure 4H). Similar to dense granules, agonist-induced release of α-granules was also significantly attenuated in Ask1^−/− platelets in comparison with WT platelets (Figure 3D). The platelet morphology and granule number were similar in Ask1^−/− and WT platelets, suggesting that the attenuation of secretion was not caused by defective granule biogenesis (Figure 3E).

In addition to Ca²⁺, TxA₂ generation is also important for the regulation of granule secretion.³⁶  We found that thrombin, collagen, and convulxin caused robust generation of TxA₂ in WT platelets, as was expected (Figure 4A). However, TxA₂ generation induced by these agonists was severely attenuated in Ask1^−/− platelets (Figure 4A). To determine whether the observed defect in granule secretion was in fact due to reduced TxA₂ generation, we determined the thrombin-induced ¹⁴C-serotonin secretion in the presence of aspirin, which inhibits Cox1, a key enzyme required for TxA₂ generation.³⁷  As was expected, we found that aspirin significantly attenuated ¹⁴C-serotonin secretion in WT platelets; however, it had no effect on thrombin-induced ¹⁴C-serotonin secretion in Ask1^−/− platelets (Figure 4B). Interestingly, the amount of ¹⁴C-serotonin secretion in the WT platelets in the presence of aspirin was the same as that in Ask1^−/− platelets in the presence or absence of aspirin (Figure 4B). These data suggest that the granule secretion defect seen in Ask1^−/− platelets was mainly dependent on TxA₂ generation. Because cPLA₂ is a key enzyme in the biogenesis of TxA₂, we used pyrrophenone, a specific cPLA₂ inhibitor, to determine the contribution of cPLA₂ in agonist-induced TxA₂ generation. We found that pretreatment of both WT and Ask1^−/− platelets with 1 μM of pyrrophenone strongly inhibited thrombin-, collagen-, and convulxin-induced TxA₂ generation, suggesting that cPLA₂ was primarily responsible for the observed agonist-induced TxA₂ generation in murine platelets (Figure 4C). It is known that cPLA₂ activity is regulated by phosphorylation of its S⁵⁰⁵ residue.³⁸  Although treatment of WT platelets with thrombin showed an increase in S⁵⁰⁵ phosphorylation of cPLA₂ in a time-dependent manner, thrombin failed to induce any S⁵⁰⁵ phosphorylation of cPLA₂ in Ask1^−/− platelets (Figure 4D). ADP and collagen also failed to induce S⁵⁰⁵ phosphorylation of cPLA₂ in Ask1^−/− platelets (Figure 4E). These results suggest that agonist-induced cPLA₂ S⁵⁰⁵ phosphorylation is entirely dependent on Ask1. If the sole function of Ask1 is activation of cPLA₂, which is required for arachidonic acid (AA) release, then exogenous addition of AA should rescue functional defects seen in Ask1^−/− platelets. We found that AA-induced integrin activation was significantly attenuated in Ask1^−/− platelets (Figure 4F). Because TxA₂ stimulates signaling by activating TPα in platelets, we also tested whether TxA₂-induced granule secretion is rescued in Ask1^−/− platelets. We found significant attenuation of α-granule as well as dense granule secretion induced by U46619, a TxA₂ mimetic (Figure 4G-H). These results suggested that in addition to TxA₂ generation, Ask1 also functions in regulating other pathways downstream of platelet agonists, including TxA₂.

The kinase responsible for the phosphorylation and activation of cPLA₂ is controversial.⁵  It was reported that both p38 and ERK MAPKs regulate cPLA₂ activity by phosphorylation of S⁵⁰⁵ downstream of agonist activation.^39,40  Because we found a complete lack of phosphorylation of cPLA₂ in Ask1^−/− platelets, we investigated which MAPK is actually responsible for cPLA₂ phosphorylation in murine platelets. We found that thrombin, ADP, and collagen robustly induced ERK1/2 as well as JNK phosphorylation in both WT and Ask1^−/− platelets, suggesting that the activation of JNK and ERK1/2 is not dependent on Ask1 and that they are not responsible for cPLA₂ phosphorylation (Figure 5A-B). On the contrary, agonists failed to induce p38 MAPK phosphorylation in Ask1^−/− platelets (Figure 5C-D), suggesting that Ask1 is required for p38 MAPK phosphorylation and p38 MAPK is responsible for cPLA₂ activation in murine platelets. To determine which MKK is activated by Ask1, we evaluated thrombin-induced phosphorylation of MKK3, MKK4, MKK6, and MKK7. We found robust time-dependent activation of all of these MKKs in WT platelets (Figure 5E-G). However, thrombin failed to phosphorylate MKK3, MKK4, and MKK6 at any time point tested (Figure 5E-F). On the other hand, phosphorylation of MKK7 occurred at the 1-minute time point but was rapidly attenuated in Ask1^−/− platelets in comparison with WT platelets (Figure 5G). Interestingly, ADP- and collagen-induced phosphorylation was significantly attenuated in Ask1^−/− platelets, suggesting that MKK7, a kinase responsible for activation of JNKs, is not under control of Ask1 (Figure 5H).

Ask1^−/− mice are fertile and healthy with no visible abnormalities. While no statistically significant differences were observed in the hematological parameters between Ask1^−/− and WT mice, the reduced white blood cell (WBC) count was very close to significance (Table 1). Interestingly, however, we found blood in the feces of 70% of adult Ask1^−/− mice in comparison with 20% of WT mice. Integrin β₃^−/− mice, which are known to have internal bleeding, were 100% positive for fecal blood (Figure 6A).⁴¹  These results suggested that Ask1^−/− mice have bleeding diathesis. To determine whether the ablation of Ask1 affected hemostasis, we compared the duration of tail bleeding in WT and Ask1^−/− mice. WT mice had a mean tail-bleeding time of 100 ± 9.6 s, whereas Ask1^−/− mice had a mean tail-bleeding time of 430 ± 35.3 s (Figure 6B). These results suggest that Ask1 regulates hemostatic function in mice. To determine the role of ASK1 in thrombosis, we performed FeCl₃-induced carotid artery thrombosis assay. We found that carotid occlusion was significantly delayed in Ask1^−/− mice in comparison with WT mice (Figure 6C). In WT mice the average time to occlude was 7.16 ± 0.48 minutes; it was 10.39 ± 0.57 minutes in Ask1^−/− mice (Figure 6D). The majority of Ask1^−/− mice (57%) either did not occlude at all or had an unstable occlusion during the 30 minutes of the assay, in comparison with 22% of WT mice (Figure 6E). These results strongly suggest that Ask1 plays an important role in regulating thrombosis.

Because the Ask1^−/− mice used in this study were created through a global deletion,²²  it is possible that in addition to the lack of platelet Ask1, an absence of endothelial Ask1 also contributes to the observed thrombosis defect. To achieve thrombosis without major involvement of endothelial cells, we performed a collagen/epinephrine-induced thromboembolism assay.²⁴  We found that the majority of the WT mice died from asphyxia, and only <20% survived beyond 3 minutes after the administration of the agonist mixture (Figure 7A). In contrast, >80% of Ask1^−/− mice survived beyond 3 minutes, suggesting that they were protected from thromboembolism (Figure 7A). The inability of the Evan’s blue dye to penetrate the lungs of WT mice was visible by their pink color; on the contrary, Ask1^−/− mouse lungs were blue because of free passage of the dye, indicating lack of obstruction (Figure 7B). Furthermore, large thrombi obstructing the pulmonary vasculature were visible in the histological sections of the WT lungs, whereas no such thrombi obstructing the pulmonary vasculature were detected in Ask1^−/− lungs (Figure 7C). These results suggest that defective platelet thrombus formation protects the Ask1^−/− mice from pulmonary thromboembolism. To further evaluate the thrombus growth in the absence of the influence of the endothelium, we assayed in vitro thrombus formation by perfusing whole blood through a microfluidic device coated with fibrillar collagen. We found that 47% fewer Ask1^−/− platelets adhered to collagen in comparison with WT platelets (Figure 7Di,ii). Interestingly, the number of platelets in a thrombus (a measure of thrombus size) as indicated by fluorescence intensity of labeled platelets was significantly lower (4%) in Ask1^−/− mouse blood than in WT mouse blood (Figure 7Di,iii). These data suggest that platelet Ask1 contributes to thrombus growth. In order to conclusively rule out the contribution of endothelial and leukocyte Ask1 in this process, we performed FeCl₃-induced thrombosis in mice transfused with platelets after immunodepletion (Figure 7E). WT mice receiving WT platelets showed a mean occlusion time of 9 minutes, and Ask1^−/− mice receiving Ask1^−/− platelets showed a mean occlusion time of 21 minutes. WT mice receiving Ask1^−/− platelets also showed an increased mean occlusion time (24 minutes). Interestingly, Ask1^−/− mice receiving WT platelets showed a mean occlusion time of 7 minutes, which is similar to that of the WT recipient mice (Figure 7E). These results suggest that the thrombotic defect observed in Ask1^−/− mice is due to impaired platelet functions. This was further supported by studies using platelet-specific Ask1 deletion.⁴² 

Platelet activation during vascular injury could be viewed as a stress response. It is therefore not surprising to find that a stress response pathway is triggered during platelet activation, a terminal event analogous to apoptosis. MAPKs were shown to be activated at various steps of platelet activation and to play an important part in this process.⁵  However, how agonist-induced signaling events result in the activation of these kinases is not well understood. We found that ASK1, a MAP3K that relays signals to p38 and JNK MAPKs, is expressed in platelets and is robustly activated during platelet activation by physiological agonists. Furthermore, we show that in murine platelets, Ask1 is indispensable for p38 activation but is not required for activation of the ERK1/2 and JNK MAPKs. The upstream MAP3K that activates ERK1/2 in platelets was shown to be protein kinase C (PKC) and not Raf.⁴³  The MAP3K responsible for platelet JNK activation remains to be determined.

It is not clear how ASK1 is activated during platelet activation. In nucleated cells, the activation mechanisms of ASK1 have been well characterized.⁴⁴  ASK1 is maintained in a resting state by association with a number of proteins such as Trx and CIB1.^14,15  During oxidative stress, ROS oxidizes Trx, causing its dissociation from ASK1 and permitting ASK1 activation by autophosphorylation.¹⁴  In certain nucleated cells, CIB1, a Ca²⁺-binding protein,²¹  binds to ASK1 at low Ca²⁺ levels, keeping it inactive. When Ca²⁺ levels rise, CIB1 dissociates from ASK1, allowing its activation.¹⁵  CAMKII, a Ca²⁺-dependent kinase, has also been reported to activate ASK1 in nucleated cells and could be responsible for inducing the activation of ASK1 in platelets.^45,46  We find that ASK1 is activated downstream of most GPCRs in platelets (Figure 1). It is therefore possible that Gα_q-dependent intracellular Ca²⁺ rise induced by G-protein-coupled receptors (GPCRs) could be important for ASK1 activation. However, other pathways could not be ruled out at this time, because some of the platelet GPCRs are also coupled to Gα_12/13, and Gα_12/13 has been reported to be involved in ASK1 activation in nucleated cells.⁴⁷ 

ASK1 has been implicated to directly activate MKK3/6 as well as MKK4/7. MKK3/6 specifically activates p38, whereas MKK7 is specific to JNK1/2, and MKK4 can activate both p38 and JNK. The complete failure of activation of MKK3, MKK4, and MKK6 in Ask1^−/− platelets suggests that Ask1 is only responsible for activation of these MKKs and that it is not needed for MKK7 phosphorylation, although absence of Ask1 results in rapid deactivation of MKK7, probably through activation of a phosphatase. The lack of a significant defect on JNK activation in Ask1^−/− mice could be because of the fact that MKK7 is not activated by Ask1 in platelets.

In platelets, cPLA₂ is primarily responsible for the release of arachidonic acid during activation. cPLA₂ translocation to the plasma membrane is dependent on intracellular Ca²⁺.⁴⁸  However, its activity is regulated by the phosphorylation of S⁵⁰⁵ by MAPKs.³⁸  Both p38 and ERK1/2 were reported to phosphorylate S⁵⁰⁵ of cPLA₂.^39,40  JNK1 does not appear to affect cPLA₂ activation because Jnk1^−/− platelets have no defect in TxA₂ generation.⁸  Our finding that cPLA₂ phosphorylation is abolished and that p38, but not ERK1/2, activation is eliminated in Ask1^−/− platelets indicates that in murine platelets, p38 is solely responsible for cPLA₂ phosphorylation and hence its activation. ERK1/2 has been shown to be negatively regulated by p38 in nucleated cells through protein phosphatase PP2A.⁴⁹  Our observation of enhanced activation of ERK1/2 in Ask1^−/− platelets could be because of the absence of p38 activation in these platelets.

Our in vivo experiments suggest that ASK1 plays a central role in regulating thrombosis by activating the downstream effectors of the MAPK cascade. This is supported by the finding that thrombosis is defective in Jnk1^−/− mice as well as in p38^+/− mice.^10,50  We attribute the defective thrombosis in Ask1^−/− mice to the absence of platelet Ask1, although the contribution of endothelial and leukocyte Ask1 to the severity of the phenotype cannot be ruled out at this time. In part, the defective thrombosis in Ask1^−/− mice could result from the lower TxA₂-dependent granule secretion. Our finding that U46619-induced platelet aggregation and granule secretion was strongly inhibited in Ask1^−/− platelets suggests that Ask1 may also be required for integrin α_IIbβ₃ activation and for aggregation in addition to dense granule secretion. As is depicted in Figure 7F, our results suggest that the increased intracellular Ca²⁺ levels downstream of the agonist receptors causes activation of ASK1 through an as yet unknown mechanism. ASK1 activates MKK3/4/6, which in turn activates p38 MAPK. cPLA₂ is then actived by p38 MAPK by phosphorylating S⁵⁰⁵ residue, resulting in AA release leading to TxA₂ production. Newly produced TxA₂ induces release of a population of dense granules. TxA₂ also activates TPα, inducing platelet activation. Activation of integrin α_IIbβ₃ may be regulated by p38 MAPK, leading to platelet aggregation. Increased Ca²⁺ can independently induce the major population of granule secretion and integrin α_IIbβ₃ activation, probably through PKC and Ca²⁺ and by DAG-regulated guanine nucleotide exchange factor I, respectively.^32,51 

In summary, we show here for the first time that ASK1, an upstream kinase of the stress-induced MAPK pathway, is activated in platelets. Most platelet agonists activate the ASK1 signaling pathway. This pathway appears to play a central role in TxA₂ generation and plays a significant role in granule secretion and integrin inside-out signaling. Together, these distinct functions of ASK1 strongly influence the thrombotic outcome.

The authors thank Brendan Bachman, Jeffrey Caplan, Natalie Barns, and Sharmila Chatterjee for technical assistance and Paul Bray and Arie Horowitz for helpful suggestions on the manuscript.

This work was supported by the National Institutes of Health, National Heart, Lung, and Blood Institute grants R01 HL113188 (U.P.N.) and R01 HL 120728 (K.B.N.).

Contribution: M.U.N. designed the experiments, performed in vivo thrombosis assays and activation of MAPK cascade, interpreted data, and wrote the paper; P.P. performed activation of Ask1 effectors and TxA₂ generation in murine platelets; R.D. performed ASK1 activation in human platelets; R.T. performed platelet aggregation, granular secretion, and flow cytometry; X.C. performed microfluidic flow experiments; K.G. performed calcium experiments; K.B.N. designed and constructed the microfluidic flow chamber; H.I. developed the Ask1 KO mouse; U.P.N. conceived the research, designed the experiments, interpreted data, and wrote and edited the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

The current affiliation for R.T. is the College of Doctoral Studies, Grand Canyon University, Phoenix, AZ.

Correspondence: Ulhas P. Naik, 1020 Locust St, Jefferson Alumni Hall, Suite 394, Philadelphia PA 19107; e-mail: ulhas.naik@jefferson.edu.