Activated human platelets synthesize prostaglandin (PG) E2, although at lower rate than thromboxane A2. PGE2 acts through different receptors (EP1–4), but its role in human platelet function remains poorly characterized compared with thromboxane. We studied the effect of PGE2 and its analogs on in vitro human platelet function and platelet and megakaryocyte EP expression. Platelets preincubated with PGE2 or its analogs were stimulated with agonists and studied by optical aggregometry. Intraplatelet calcium mobilization was investigated by the stopped flow method; platelet vasodilator-stimulated phosphoprotein (VASP), P-selectin, and microaggregates were investigated by flow cytometry. PGE2 at nanomolar concentrations dose-dependently increased the slope (velocity) of the secondary phase of ADP-induced platelet aggregation (EC50, 25.6 ± 6 nM; Emax of 100 ± 19% increase versus vehicle-treated), without affecting final maximal aggregation. PGE2 stabilized reversible aggregation induced by low ADP concentrations (EC50, 37.7 ± 9 nM). The EP3 agonists, 11-deoxy-16,16-dimethyl PGE2 (11d-16dm PGE2) and sulprostone enhanced the secondary wave of ADP-induced aggregation, with EC50 of 48.6 ± 10 nM (Emax, 252 ± 51%) and 5 ± 2 nM (Emax, 300 ± 35%), respectively. The EP2 agonist butaprost inhibited ADP-induced secondary phase slopes (IC50, 40 ± 20 nM). EP4 stimulation had minor inhibitory effects. 11d-16dm PGE2 alone raised intraplatelet Ca2+ and enhanced ADP-induced Ca2+ increase. 11d-16dm PGE2 and 17-phenyltrinor PGE2 (EP3 > EP1 agonist) at nanomolar concentrations counteracted PGE1-induced VASP phosphorylation and induced platelet microaggregates and P-selectin expression. EP1, EP2, EP3, and EP4 were expressed on human platelets and megakaryocytes. PGE2 through different EPs finely modulates human platelet responsiveness. These findings should inform the rational selection of novel antithrombotic strategies based on EP modulation.
Activated human platelets synthesize and release prostaglandin (PG) E2 during whole-blood clotting, although at concentrations approximately 30-fold lower compared with thromboxane (TX) A2 (Patrignani et al., 1982). TXA2, by binding the PGH2/TXA2 receptors, promotes prohemostatic responses such as vasoconstriction and platelet aggregation (Patrono and Rocca, 2010). PGE2 can bind at least four structurally different receptor subtypes (EP1–4), resulting in diverse and often opposite final biological responses (Narumiya et al., 1999; Tsuboi et al., 2002). The role of PGE2 in inflammation, pain, fever, gastroprotection, and labor is also well established (Narumiya et al., 1999; Tsuboi et al., 2002) and can be pharmacologically modulated by selective and/or nonselective cyclooxygenase (COX) inhibitors (Patrono and Rocca, 2009) or by PGE2 synthetic analogs, as in the case of misoprostol for gastroprotection or early pregnancy termination (Rocca, 2006).
The role of PGE2 in hemostasis and, possibly, thrombosis has emerged relatively more recently, mainly from the study of EP-deleted mice. Indeed, EP3-deleted mice display lower platelet response to subthreshold concentrations of agonists and are less susceptible to experimental thrombosis than wild-type mice (Fabre et al., 2001; Ma et al., 2001; Gross et al., 2007). mRNAs for EP2, EP4, and EP3 subtypes have been isolated from human platelets (Paul et al., 1998). Early studies of PGE2 showed its capacity to potentiate platelet aggregation in response to subthreshold concentrations of agonists (Shio and Ramwell, 1972; Bruno et al., 1974; Vezza et al., 1993). Upon the pharmacological characterization of EPs, EP3 receptor agonists have been shown to potentiate human platelet aggregation in response to low concentrations of various agonists (Matthews and Jones, 1993; Heptinstall et al., 2008; Iyú et al., 2010), and clinical development has begun for an EP3 antagonist (Singh et al., 2010). However, the human EP3 receptor has many splice variants that have different signaling pathways (Kotani et al., 1995; Schmid et al., 1995), and mRNAs for at least four EP3 splicing variants have been isolated from human platelets (Paul et al., 1998). Moreover, a “dual” effect of PGE2 on platelet response (i.e., activatory at nanomolar concentrations and inhibitory at micromolar concentrations) was repeatedly observed in early studies (Salzman et al., 1972; Shio and Ramwell, 1972; Bruno et al., 1974; Andersen et al., 1980; Tynan et al., 1984; Vezza et al., 1993). The inhibitory effect has been attributed, at least in mice, to heterologous activation of the prostacyclin receptor (Fabre et al., 2001), but it remains unclear whether this heterologous activation occurs in humans and whether the inhibitory effect results from different EP activation (Tynan et al., 1984; Gray and Heptinstall, 1985).
In many biological systems, within the same tissue or cell, such as T cells, mast cells, human fibroblasts, and cells isolated from ovary interstitium or cortex PGE2 can act as positive and negative modulator, eliciting different final responses, depending on its concentration in the microenvironment, type of available EPs, different affinities for the EPs present in the system at a given concentration of the agonist (Tsuboi et al., 2002; Rocca, 2006; Hoshikawa et al., 2009; Markosyan and Duffy, 2009; Li et al., 2010). Likewise, platelets might be modulated by PGE2 in a more complex way, through different EP subtypes and variants.
The present study was aimed at characterizing 1) the effect of PGE2 on human platelet function in response to different agonists used over a range of concentrations, 2) the effects of selective EP activation, 3) the EP protein expression pattern both in platelets and megakaryocytes, and 4) the effects of EP agonists on intraplatelet calcium, vasodilator-stimulated phosphoprotein (VASP) phosphorylation, microaggregate formation, and P-selectin expression. The results of these studies suggest that PGE2 acts as a fine tuner of platelet function primarily through its interaction with EP2 and EP3, allowing diverse pharmacological strategies potentially useful in cardiovascular treatment and prevention.
Materials and Methods
Collagen and ADP were purchased from Mascia Brunelli (Milan, Italy). Phycoerythrin-conjugated anti-CD61 and isotype- and fluorochrome-matched irrelevant mouse IgGs were purchased from eBioscience (San Diego, CA). FITC-conjugated anti-CD62P (P-selectin) was purchased from Caltag Laboratories (Burlingame, CA). Anti-CD61 monoclonal antibody was obtained from Dako Denmark A/S (Glostrup, Denmark). Affinity-purified rabbit polyclonal antibodies against EP2, EP3, and EP4 receptors, 11-deoxy-16,16-dimethyl PGE2, 17-phenyl-trinor PGE2, 6-isopropoxy-9-oxoxanthene-2-carboxylic acid (AH6809), butaprost, sulprostone, 2-(3-hydroxyoctyl)-5-oxo-1-pyrrolidineheptanoic acid (CAY10580), aspirin, indomethacin, and 5-(4-chlorophenyl)-1-(4-methoxyphenyl)-3-(trifluoromethyl)-1H-pyrazole (SC-560) were purchased from Cayman Chemical (Ann Arbor, MI). Rhodamine (TRITC)-conjugated goat anti-mouse IgG was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). FITC-conjugated goat anti-rabbit IgG was obtained from Vector Laboratories (Burlingame, CA). Rapide Decalcificant Osseuse was purchased from Eurobio (Les Ulis, France). Biotinylated polyvalent antibody-streptavidin peroxidase kit and DAB substrate kit were purchased from Scy-Tek Laboratoires (Logan, UT). Platelet VASP-P2Y12 kit was obtained from Biocytex (Marseille, France). Fura-2 acetoxymethyl ester was purchased from Sigma-Aldrich (St. Louis, MO). PGE2, its stable analogs, and COX inhibitors (aspirin, indomethacin, SC-560) were dissolved in ethanol at a concentration at least 500 to 1000× greater than the final working concentration.
Blood and Bone Marrow Collection.
Venous blood was obtained after informed consent by healthy volunteers from among the laboratory personnel who denied taking any medication in the preceding 14 days. Blood was taken by forearm venipuncture and dispensed into polystyrene tubes containing trisodium citrate [final concentration, 0.38% (w/v)]. Bone marrow slides were obtained from the archives of the Pathology department from post mortem tissues of subjects who died from accidental causes.
For platelet-rich plasma (PRP) preparation, citrated whole blood was centrifuged at 800 rpm for 15 min, and the supernatant PRP was removed. Platelet aggregation was measured in PRP by standard light-transmittance aggregometry (LTA) using a PACKS-4 aggregometer (Helena Laboratories, Beaumont, TX) in response to ADP, collagen, or arachidonic acid (AA), as indicated under Results. In a typical experiment, 1000× stock solutions of PGE2 or its analogs in ethanol were added to PRP at 1:1000 dilution to the indicated final concentration, in control samples an equivalent volume of ethanol was added; thus, in each sample, the final concentration of ethanol was always 0.001%. In preliminary experiments versus saline alone, this ethanol concentration did not modify platelet response patterns. After 90 s of incubation of PRP with PGE2 or its analogs under stirring 150g at 37°C, the platelet agonist was added. The following parameters of the aggregation tracings were considered for the analyses: the maximal extent of aggregation (Tmax) expressed in percentage of light transmittance, the slope of the aggregation tracings as an index of aggregation velocity (expressed in millimeters per minute) and the lag-time between agonist addition and onset of aggregation in the presence of collagen (expressed in seconds). In experiments with lower ADP concentrations giving a single-phase, fully reversible aggregation, the percentage of light transmittance 240 s after agonist addition was used to analyze experimental data. In experiments using cyclooxygenase inhibitors, PRPs were preincubated with each inhibitor for 20 min at 37°C, and then LTA experiments were performed as described above.
Flow Cytometry Analyses.
In a typical experiment investigating P-selectin expression on platelet surface, aliquots of citrated whole blood were incubated with different concentrations of PGE2 analogs, as indicated under Results, diluted from 1000× stock solutions, or with a similar volume of ethanol as control. After mixing and 5-min incubation without stirring, samples were stained with the primary fluorescent antibody [anti-P-selectin (final dilution, 1:20) or anti-CD61 (final dilution, 1:20)] or isotype- and fluorochrome-matched irrelevant mouse IgGs as negative control for 1 h at 4°C. The platelet population was identified on a Log scale on the basis of forward and side-scatter distribution and CD61 positivity, and 3 ×104 CD61-positive platelets were acquired and analyzed using a FACSCanto flow cytometer (BD Biosciences, San Jose, CA). Data were analyzed as mean fluorescence intensity using the FACSDiva software package (BD Biosciences).
VASP phosphorylation was assessed in citrated whole blood by flow cytometry using the Platelet VASP-P2Y12 kit according to the manufacturer's instructions. In a typical experiment using PGE2 analogs, citrated aliquots of whole blood were preincubated with different concentrations of each compound or vehicle for 5 min without stirring and then processed according to the manufacturer's instructions, with or without ADP stimulation, as indicated under Results.
Immunohistochemistry was performed on washed platelets as described previously (Rocca et al., 2002), using anti-CD61 (final dilution, 1:50), anti-EP2 (1:100), anti-EP3 (1:150), and anti-EP4 (1:300) antibodies. After incubation with the anti-CD61 antibody, immunodetection was performed using a goat anti-mouse TRITC conjugate. Primary anti-EP2, anti-EP3, and anti-EP4 antibodies were detected with a goat anti-rabbit FITC conjugate. Bone marrow biopsies were fixed 12 h in 4% formalin, decalcified in rapid decalcificant for 4 h, embedded in paraffin, cut at 3 μm, mounted on polarized slides, and dewaxed. Slides were rehydrated, washed in phosphate-buffered saline containing 0.1% Triton X for 2 min, and treated with 3% H2O2 for 5 min to block endogenous peroxidase. The antigen retrieval was performed with 0.01 M citrate buffer, pH 6, in a microwave oven set at 750 W for 10 min. Bone marrow slides were incubated overnight at 4°C with one of the primary antibodies [anti-EP2 (1:100), anti-EP3 (1:150), and anti-EP4 (1:300)] with or normal goat serum as negative control. For immunoperoxidase, biotinylated polyvalent antibody-streptavidin peroxidase kit was used and developed with the DAB substrate kit. Slides were lightly counterstained with hematoxylin. Specimens were observed and digitalized by a Zeiss Axioskop (Zeiss, Jena, Germany) equipped with an intensified charge-coupled device camera system (Photometrics, Tucson, AZ).
Intracellular Ca2+ was measured in platelets labeled with the fluorescent dye Fura-2 using a Cary Eclipse fluorescence spectrophotometer (Varian, Palo Alto, CA), equipped with RX2000 Rapid Mixing Stopped-Flow Unit (Applied Photophysics Ltd, Leatherhead, UK) as described previously (Sage et al., 1990). Platelet suspensions were excited at 360 nm, and the emitted light was measured at 510 nm, using slits of 10 nm. PRP samples were first incubated with Fura-2 at 37°C for 30 min, then washed by gel-filtration chromatography, resuspended in Tyrode buffer (10 mM HEPES, 135 mM NaCl, 3.5 mM KCl, 5.5 mM glucose, and 0.1% BSA, pH 7.4) and adjusted at a final concentration of 13 × 104 platelets/μl. Three milliliters of Fura-2 loaded platelets were injected into a cell holder, and 3-ml solutions of agonists in Tyrode buffer were simultaneously injected through another syringe into the same cell holder. The tubes carrying the solutions into the mixing chamber passed through a flexible jacket, the temperature of which was controlled by a Haake water circulator (Thermo Fisher Scientific, Waltham, MA) and a Varian Peltier apparatus. The syringe pistons were moved by compressed air set at 3 bar, and 150 μl of each solution entered the chamber before flow was stopped. For each measurement, ADP with or without different PGE2 analogs or vehicle was used as indicated. Based on preliminary experiments to identify the ADP concentration (1.25–10 μM), which better enabled the study of different parameters of the kinetics of intraplatelet Ca2+, 2.5 μM ADP (final concentration) was selected. Fluorescence was recorded every 50 ms and analyzed by Cary Eclipse kinetics application (Varian). The [Ca2+] was calculated using the general formula Ca2+ = Kd (Fobs − Fmin/(Fmax − Fobs), where Kd is the dissociation constant for Ca2+ binding to the indicator, Fobs is the fluorescent unit, and the Kd calculated for Fura 2 is 224 nM (Grynkiewicz et al., 1985), whereas Fmax and Fmin were determined for any data set as detailed previously (Grynkiewicz et al., 1985). In the kinetics of intraplatelet Ca2+ changes, three sequential phases were identifiable on the tracings: 1) a lag time (LT), 2) the increase in cytoplasmic [Ca2+], and 3) a slower decrease in return of intraplatelet Ca2+ to baseline. In our experiments, we analyzed the first two phases, (e.g., LT and [Ca2+] increase) as follows: LT was defined as the extrapolation of the maximum velocity (Vmax) of thrombin-induced Ca2+ increase to its basal value at t = 0 (Ca02+). The [Ca2+] at t = 0 was obtained by linear regression of the first 20 points of the kinetic curve. The maximum velocity of Ca2+ increase was computed by linear regression of a 20 experimental points centered around a given Ca2+ concentration, as detailed previously (Grynkiewicz et al., 1985). In the fitting procedure, only the ascending part along with the plateau value of Ca2+ concentration was analyzed to compute Vmax. The time required to reach Vmax was defined as tmax. When Vmax, Ca02+, and tmax were calculated on the basis of simple trigonometric relations, the LT value was calculated as follows: LT = tmax − (Catmax2+ − Ca02+) Vmax, where Catmax2+ is the value of intraplatelet Ca2+ concentration at Vmax. Only the ascending part along with the plateau value of Ca2+ concentration was analyzed to compute Vmax.
Data were first checked for normality of distribution. Differences versus control samples were analyzed with paired or nonpaired t test as appropriate. Data are reported as mean ± S.D. or S.E. as indicated. Significance was defined as p < 0.05. Analyses were performed using SPSS (version 13.0; SSPS Inc., Chicago, IL) and SigmaStat 3.1 (Systat Software Inc., Hounslow, UK). Grafit software (Erithacus Software, Staines, UK) was used for fittings of the dose-response curves and for calculating the IC50, EC50, or Emax values.
Platelet Aggregation and EP Expression.
We first tested the effects of preincubation with PGE2 (0.2 nM–150 μM) on LTA induced by increasing concentrations of three different agonists (i.e., ADP, collagen, and AA). PGE2 was added to PRPs 90 s before each agonist. LTA continuously recorded during this preincubation under stirring detected no optical signal variation.
Based on preliminary experiments, we selected three different ranges of ADP concentrations to generate three distinct patterns of aggregation: lower concentrations (2–4 μM) generated a small monophasic fully reversible aggregation; intermediate concentrations (6–8 μM) generated two distinct primary and secondary phases of aggregation; and a high concentration (20 μM) giving a monophasic, complete, and irreversible aggregation. This was designed to selectively explore the effects of PGE2 on primary and secondary ADP-induced phases of aggregation (Andersen et al., 1980; Hechler et al., 2005) despite the individual variability in the sensitivity to ADP and despite different commercial ADP preparations. In fact, the two sequential phases of ADP-induced aggregation are known to reflect different components of response to ADP: a P2Y1-dependent, reversible phase and a P2Y12- and release reaction-dependent irreversible phase (Hechler et al., 2005). Furthermore, in some experiments, we also used 5 μM ADP, which was consistently giving a stable primary wave of aggregation.
At the highest (20 μM) ADP concentrations, we did not observe any effect of PGE2 preincubation up to 400 nM on both Tmax and slope of aggregometric traces (data not shown). When the intermediate (mainly 8 μM) ADP concentration was used, resulting in a two-wave aggregation, PGE2 dose-dependently and selectively increased the slope of the secondary phase of aggregation with an EC50 of 25.6 ± 6 nM (Emax, 100 ± 19% slope increase versus vehicle-treated samples) (Fig. 1, A and B) without affecting either Tmax or the primary phase of aggregation. At the lowest ADP concentrations generating a small, reversible aggregation, PGE2 dose-dependently stabilized aggregation by reverting disaggregation and originated a small secondary wave of aggregation at the highest concentrations (approximately 200 nM; Fig. 1C) without modifying the slope and Tmax of the primary wave. The EC50 of PGE2 in blunting reversible aggregation was 37.7 ± 9 nM (Emax, 137 ± 26% increase of aggregation measured 4 min after agonist addition versus vehicle-treated samples) (Fig. 1, C and D), being close to the one accelerating secondary aggregation.
Aspirin blocks the platelet synthesis of TXA2, which enhances platelet release reaction and stabilizes aggregates. Thus, aspirin is known to blunt the secondary phase of ADP-induced aggregation, resulting in reversible aggregation (Shio and Ramwell, 1972). Given the effect of PGE2 on enhancing the secondary wave of ADP-induced aggregation and on stabilizing reversible aggregation (Figs. 1 and 2, A and B), we explored whether exogenous PGE2 was able to partially revert the effect of 50 μM aspirin added in vitro. At this concentration, aspirin suppresses by >99% TXA2 generation from platelets in vitro (Dragani et al., 2010). As expected, 50 μM aspirin inhibited the secondary aggregation wave induced by 8 μM ADP (Fig. 2A) and reverted stable primary aggregation induced by 5 μM ADP (Fig. 2B). When 200 nM PGE2 was added to aspirin-pretreated samples, the effect of aspirin on ADP-induced aggregation was partially reverted (percentage of light transmittance measured at 4 min after agonist addition: 16 ± 17% with aspirin alone versus 24.3 ± 17.3% with aspirin plus 200 nM PGE2, p = 0.003 for paired samples, n = 6; and Fig. 2, A and C). Similar effects were observed by pretreating PRP with 20 μM indomethacin or 20 μM SC-560, a selective COX-1 inhibitor (data not shown). PGE2 in the high nano- and micromolar range (starting from 500 nM) dose dependently inhibited both Tmax and slopes of ADP-induced aggregation independently of ADP concentrations, with IC50 of 5 ± 2 μM for slope reduction and of 15 ± 8 μM for Tmax reduction (Fig. 3, A–D).
We also studied collagen-induced platelet aggregation. As shown in Fig. 3E, at the lowest collagen concentration (1.25 μg/ml), PGE2 significantly increased Tmax (37 ± 18.4% in vehicle-treated samples versus 49.7 ± 18.5% in 200 nM PGE2-pretreated samples, n = 7, p = 0.031) without affecting slope (25.6 ± 5 versus 27.6 ± 12 mm/min in vehicle-treated versus 200 nM PGE2, n = 7, p = 0.68), and there was a nonsignificant trend toward a reduction of the lag interval (123 ± 12 s versus 111 ± 25 s in vehicle-treated versus 200 nM PGE2, respectively; n = 7, p = 0.10). At higher collagen concentrations (2.5–5 μg/ml), PGE2 preincubation in the nanomolar range did not significantly affect Tmax or slopes (data not shown) and was associated with a nonsignificant shortening of the lag intervals (73.13 ± 13 s and 68.3 ± 13.6 s in vehicle- and PGE2-treated samples, respectively, n = 9, p = 0.25). Micromolar concentrations of PGE2 dose-dependently inhibited collagen-induced aggregation independently of agonist concentrations, reducing both Tmax and slope. Finally, preincubation up to 200 nM concentrations of PGE2 did not affect AA-induced aggregation to any statistically significant extent (data not shown).
We next used different PGE2 analogs to study the contribution of each EP to the net effect on platelets observed with PGE2. The 11-deoxy-16,16-dimethyl PGE2 (11d-16dm PGE2) dose-dependently and selectively accelerated the secondary phase of 8 μM ADP-induced aggregation with an EC50 of 48.6 ± 10 nM (Fig. 4, A and C), and the enhancement of the secondary wave of aggregation was greater than the one observed with PGE2, as indicated by the corresponding Emax values (252 ± 51 versus 100 ± 19% increase, respectively; p < 0.01). Moreover, 11d-16dm PGE2 was also able to trigger a secondary aggregation, dose-dependently, when lower concentrations of ADP were used, with an EC50 of 43.5 ± 7.1 nM (Fig. 4B). Similar effects were observed with sulprostone, which enhanced the secondary ADP-induced aggregation phase with an EC50 of 5 ± 2 nM and an Emax of 300 ± 35%. In addition, sulprostone partially reverted the effect of aspirin on ADP-induced aggregation (Fig. 2D). 11d-16dm PGE2 did not modify aggregation induced by the highest concentrations of ADP (20 μM). In response to 2.5 and 5 μg/ml collagen, increasing concentrations of 11d-16dm PGE2 did not affect Tmax or slopes, but the lag time was dose-dependently shortened, with an IC50 of 36 ± 23 nM, and the shape change was abolished between 50 and 70 nM (Fig. 4D). At variance with micromolar concentrations of PGE2, both sulprostone and 11d-16dm PGE2 at concentrations ≥500 nM induced a full irreversible platelet aggregation in the absence of other agonists (data not shown). Based on Kd and Ki values, sulprostone and 11d-16dm PGE2 act preferentially as EP3 agonists (Kiriyama et al., 1997; Abramovitz et al., 2000).
Butaprost, a preferential EP2 agonist (Kiriyama et al., 1997; Abramovitz et al., 2000), dose-dependently and selectively inhibited the secondary wave of 8 μM ADP-induced aggregation (Fig. 5, A and B) with an IC50 of 40 ± 20 nM. Moreover, butaprost reduced the Tmax of 2 to 4 μM ADP-induced reversible aggregation with a similar IC50 of 30.4 ± 17 (Fig. 5C). When used before 2.5 μg/ml collagen, butaprost significantly reduced the slope, with a significant effect at 1 μM: 68.5 ± 12 versus 42.3 ± 15.5 mm/min in vehicle- and 1 μM butaprost-treated samples, respectively, n = 4, p = 0.01. We also observed a nonsignificant prolongation of lag times and reduction of Tmax (Fig. 5D). The CAY10580, an EP4 agonist (Billot et al., 2003), had only modest effects on platelet response. A statistically significant reduction in slope and Tmax of ADP-induced aggregation was detectable only at the highest concentrations of 1 μM: 23.9 ± 9% reduction in Tmax and 23.8 ± 8% reduction in slope, p = 0.036, n = 5.
In light of these functional data, we investigated EP protein expression by immunohistochemistry in peripheral platelets and bone marrow megakaryocytes. These studies revealed an immunoreactivity for EP1 (not shown), EP2, EP3, and EP4 in platelets (Fig. 6) and megakaryocytes (Fig. 7). It is noteworthy that the pattern of positivity of the EPs in megakaryocytes was different, EP3 positivity showing reinforcement at the periphery of the cytoplasm, where proplatelets are formed and released, whereas the EP2 and EP4 showed a more diffuse cytoplasmic positivity.
As a result of the activating effects of 11d-16dm PGE2, we investigated intraplatelet Ca2+ movement, a known proaggregator signal. The kinetics of rapid Ca2+ movements was continuously recorded by the stopped-flow method in washed platelets treated with vehicle, 2.5 μM ADP alone, 11d-16dm PGE2 alone (10 nM–1 μM), ADP plus 100 nM 11d-16dm PGE2, 500 nM butaprost or CAY10580 with or without ADP, and 100 nM 11d-16dm PGE2 plus the EP3-III antagonist AH6809 (50 μM) (Abramovitz et al., 2000).
Intraplatelet Ca2+ increase displays a complex kinetics after a reaction of A→B→C type (Sage et al., 1990), and the complex Ca2+ rise induced by ADP is associated with a very rapid influx by receptor-operated channels in the plasma membrane. The delayed phase of the ADP-evoked intraplatelet Ca2+ rise is likely to result from the release of Ca2+ from intracellular stores (Sage et al., 1990). It is noteworthy that the timing of the delayed ADP-evoked event is modulated by intracytosolic Ca2+ concentration, being more rapid in onset when internal [Ca2+]i is high (Sage et al., 1990). Figure 8A shows indeed that both the Vmax and the maximum calcium concentration values are progressively higher as an inverse function of the LT values, or, conversely, as a direct function of the rapid increase in the intraplatelet Ca2+ (inset Fig. 8, A and B, and Table 1).
The 11d-16dm PGE2 dose-dependently increased cytoplasmic Ca2+ (Fig. 8A) with an EC50 of 39 ± 3 nM, and it strongly synergized with 2.5 μM ADP in further augmenting intraplatelet Ca2+ to values of approximately 1 μM (Fig. 8B). The AH6809 blunted the effect of 11d-16dm PGE2, whereas neither butaprost nor CAY10580, alone at different concentrations or in combination with ADP, had any effect on Ca2+ movements (data not shown).
VASP-P Phosphorylation, Microaggregates, and P-Selectin Expression.
We next investigated the effects of different PGE2 analogs on the phosphorylation of platelet VASP (VASP-P), surface P-selectin expression and platelet microaggregates identified on the basis of forward and side scattering (Fox et al., 2004) together with CD61 positivity.
Preincubation of whole-blood samples with 200 nM 11d-16dm PGE2 or PGE2 alone decreased platelet VASP-P induced by PGE1 (Fig. 9A). We also tested another compound, the 17-phenyl-trinor PGE2 (EP3 > EP1 agonist; Kiriyama et al., 1997), which gave results similar to those of 11d-16dm PGE2. At variance with 11d-16dm PGE2, PGE2, or 17-phenyl-trinor PGE2, butaprost added alone to whole blood dose-dependently increased platelet VASP-P compared with controls (Fig. 9B) and counteracted the dephosphorylating effect of ADP on VASP when added to samples treated with PGE1 and ADP (Fig. 9C). The CAY10580 up to 2 μM did not cause any variation of VASP-P or modified the effect of ADP in ADP/PGE1-treated samples (data not shown).
Based on forward scatter (FSC) and side scatter (SSC) parameters of flow cytometry plots together with CD61 positivity, both 11d-16dm PGE2 and 17-phenyl trinor PGE2 alone caused a shift in the CD61-positive platelet population gate compared with controls, which is likely a result of platelet microaggregate formation (Fox et al., 2004) (Fig. 10, A–D). This effect was dose-dependent and visible on flow cytometry FSC/SSC plots starting from 20 nM 11d-16dm PGE2 and 40 nM 17-phenyl trinor PGE2. No shift in the platelet population gate was observed when butaprost of CAY10580 were used up to 2 μM (data not shown).
Given the formation of platelet microaggregates, we checked whether 11d-16dm PGE2 or 17-phenyl trinor PGE2 alone were able to induce markers of activation on platelet surface and investigated surface P-selectin (CD62p) expression on nonpermeabilized platelets. Both compounds added alone dose-dependently increased P-selectin expression on platelets compared with vehicle-treated samples (Fig. 10E). Butaprost and the CAY10580 at concentrations up to 2 μM did not induce P-selectin exposure on platelet membranes (data not shown). Finally, AH6809 up to 50 μM had no effect on VASP-P or P-selectin induction (data not shown).
The present work explored the effects of PGE2 in modulating human platelet responsiveness. Different PGE2 analogs were used to dissect the relative contribution of individual receptor subtypes. Moreover, we provided the first evidence for the presence of several EP proteins in both platelets and megakaryocytes.
We have shown a selective role for nanomolar concentrations of PGE2 in amplifying the secondary phase of ADP-induced aggregation and stabilizing reversible aggregation, by using increasing concentrations of ADP. Although a role for PGE2 as positive modulator of platelet aggregation has been reported in mice (Fabre et al., 2001; Gross et al., 2007) and humans (Matthews and Jones 1993; Bruno et al., 1974; Vezza et al., 1993), this is the first report identifying the specific phase of platelet response, which is modulated by PGE2 as a function of a range of agonist concentrations. Previous studies on mice used subthreshold, nonaggregating concentrations of ADP. More recent studies using a whole blood method based on platelet counting (Heptinstall et al., 2008) investigated a single, low ADP concentration and could not detect any direct effect of PGE2 (0.01–10 μM) on platelets, thus the role of PGE2 was indirectly extrapolated by studying the effect of EP antagonists in PGE2-stimulated samples. We also showed for the first time a positive effect of PGE2 on collagen-induced aggregation at low and intermediate collagen concentrations. Our data suggest that nanomolar PGE2 acts as positive platelet modulator, stabilizing transient aggregation (low ADP or collagen concentrations) and accelerating the completion of aggregation (e.g., slope increase of ADP-induced secondary aggregation, shortening lag time of collagen-induced aggregation). PGE2 consistently counteracted, at least in part, the inhibition exerted by aspirin, again stabilizing and amplifying aggregation.
We used a series of PGE2 analogs to try to identify the receptor(s) mediating PGE2-evoked responses. Both 11d-16dm PGE2 and sulprostone reproduced the positive pattern of PGE2 on the amplification/stabilization of platelet aggregation, but more effectively, as shown by the Emax for accelerating the secondary ADP-induced phase, the shortening of the lag interval in collagen-induced aggregation and their capacity of triggering a complete, irreversible aggregation in the high nanomolar/low micromolar range. Based on Ki values, sulprostone is approximately 30- to 300-fold more selective for EP3 versus EP1, whereas 11d-16dm is approximately 20-fold more selective for EP3 versus EP4 (Kiriyama et al., 1997; Abramovitz et al., 2000). Considering the rank order of potency for accelerating the secondary phase of ADP-induced aggregation relatively to PGE2 EC50, sulprostone had a value of 0.19, whereas 11d-16dm of 1.8. This hierarchy (sulprostone > PGE2 > 11d-16dm PGE2) and the characteristics of each compound (Kiriyama et al., 1997; Abramovitz et al., 2000), support EP3 as a mediator of the activating effects of PGE2. The data of 17-phenyl-trinor PGE2 (EP3 > EP1 agonist based on Ki values; Kiriyama et al., 1997), mimicking 11d-16dm PGE2 on VASP-P, microaggregates, and P-selectin expression, further support this hypothesis. Our results are consistent with and extend previous data (Heptinstall et al., 2008; Iyú et al., 2010), but we used diverse PGE2 analogs, calculated the EC50 values, ranked potency profiles on specific effects (enhancement of the secondary aggregation, stabilization of aggregates), and explored different platelet agonists at various concentrations. Furthermore, this is the first description of 11d-16dm PGE2-responsive receptors on platelets. This compound is likely to identify a specific, pharmacologically defined subtype of the EP3, which was first identified in erythroleukemia, a megakaryocyte-related cell line (Feoktistov et al., 1997). Consistent with data on erythroleukemia, 11d-16dm PGE2 increased Ca2+ in platelets, and EC50 values were similar in the two cell types (28 nM in erythroleukemia cells, 39 nM in platelets). In our experimental setting, 11d-16dm PGE2 not only enhanced the ADP-induced increase in intraplatelet Ca2+ but also dose-dependently increased intraplatelet Ca2+ when used alone. This is the first report of an EP3 agonist alone raising intraplatelet Ca2+, whereas previous studies described synergizing effects with other agonists (Heptinstall et al., 2008). It is possible that continuous recording in stopped flow is a more sensitive technique compared with flow cytometry-based single measurements. The kinetics of the Ca2+-induced increase in platelets was similar to that in cells transfected with human EP3 splice variants (An et al., 1994; Schmid et al., 1995). Moreover, the kinetics of Ca2+ increase induced by 11d-16dm PGE2 alone in platelets resembled that of ADP, possibly reflecting similar signaling.
EP3 splice variants have been described to signal not only through Ca2+ increase but also through cAMP reduction (An et al., 1994; Kotani et al., 1995; Schmid et al., 1995). 11d-16dm PGE2 and 17-phenyl-trinor PGE2 were consistently able to counteract PGE1-induced increase in platelet VASP-P. Similar findings have been reported using sulprostone (Heptinstall et al., 2008). Platelet VASP phosphorylation is due to an increase in cAMP. Thus, EP3 stimulation seems also to counteract cAMP increase, thus increasing platelet reactivity. Based on our data, different EP3 variants might operate in platelets increasing intraplatelet Ca2+ and/or reducing cAMP. It is noteworthy that only some human EP3 splice variants increase Ca2+, whereas others signal through cAMP (An et al., 1994; Schmid et al., 1995). In our experimental setting, AH6809 (EP3-III antagonist; Kiriyama et al., 1997) antagonized 11d-16dm PGE2-induced Ca2+ increase but not VASP-P or P-selectin expression. mRNAs for four EP3 splice variants (Ib, III, III, and IV) have consistently been isolated in platelets. All together these data might indicate that different EP3 variants potentiate platelet function.
Both 11d-16 dm PGE2 and 17-phenyl-trinor PGE2 shifted the FSC/SSC of the CD61-positive gate at flow cytometry, which can be accounted for by microaggregate formation (Fox et al., 2004). Flow cytometry, being more sensitive than LTA, possibly detected early microaggregates at nanomolar concentrations of EP3 agonists, whereas LTA could detect macroaggregates only at higher (high nanomolar) concentrations. In agreement with an activatory pattern of PGE2/EP3, platelet membrane P-selectin was induced by 11d-16dm PGE2 and 17-phenyl-trinor PGE2. This can be a relevant mechanism bridging platelets between coagulation and inflammation. PGE2 is synthesized at sites of inflammation and platelet P-selectin mediates platelet-leukocyte interaction (Dixon et al., 2006). Indeed, we observed enrichment in CD61 positivity in the FSC/SSC gate of leukocytes when the EP3 agonists were used (data not shown).
In our experiments, the EP2 agonists butaprost exerted a negative modulation of platelet function on the same phase of ADP- and collagen-induced platelet aggregation enhanced by EP3 activation. A negative effect of butaprost on PAF-induced platelet aggregation has also been reported (Matthews and Jones, 1993; Iyú et al., 2010). Butaprost did not modify intraplatelet Ca2+, with or without ADP, whereas it promoted VASP-P, indicating a cAMP-mediated effect, which is consistent with previous data on different cells (Regan et al., 1994). The Emax values indicate that sulprostone and 11d-16dm PGE2 were more effective than PGE2 in accelerating the secondary phase of ADP-induced aggregation, and, at variance with the inhibitory effect of PGE2, at the highest concentrations (>500 nM) triggered aggregation in the absence of other agonists. Thus, together with the data of butaprost, our results might indicate that the net response to nanomolar PGE2 concentrations might result from the balance between EP2 and EP3 (variants) activation. The Kd values indicate that PGE2 consistently has higher affinity for EP3 compared with EP2 (approximately 5- to 10-fold difference; Kiriyama et al., 1997; Abramovitz et al., 2000). Thus, an EP3-driven response might be triggered by lower PGE2 levels, whereas EP2 might prevail at higher concentrations. On the basis of previous data (Gray and Heptinstall, 1991) and our own, and on the lower affinity of PGE2 for the EP2 compared with the EP3 (Kiriyama et al., 1997; Abramovitz et al., 2000), it is also conceivable that the platelet-inhibiting effect of micromolar concentrations of PGE2 can be mediated by the EP2 rather than by an heterologous prostacyclin receptor stimulation, at least in humans. At variance with recent reports (Kuriyama et al., 2010), we could not detect any major effect of EP4 stimulation, even though EP4 was present in platelets and megakaryocytes.
PGE2 synthesized during in vitro whole-blood clotting, which represents the maximal biosynthetic capacity of platelet COX-1 activity, reaches levels of approximately 10 ng/ml corresponding to 28 nM (Patrignani et al., 1982). This concentration is surprisingly close to the EC50 values we observed for the activatory effects of PGE2 on platelets. The platelet origin of PGE2 synthesized in this experimental setting, at least under physiological conditions, is suggested by its virtually complete suppression by low-dose aspirin administered in vivo (7 ± 4.3 ng/ml, n = 18 healthy subjects versus 0.2 ± 0.1 ng/ml, n = 21 healthy subjects on aspirin; B. Rocca and G. Petrucci, unpublished data). Thus, during platelet activation, the amount of released PGE2 is likely to fall within the range required for amplifying platelet function. In addition, PGE2 could have a different cellular origin at sites of inflammation, and thus platelet function could be also modulated by PGE2 released from inflammatory cells. Based on our data, it is conceivable that pharmacological strategies based on selective TXA2 receptor blockade or inhibition of TX-synthase, might be less effective than expected because of unopposed PGE2/EP3 pathway. Indeed, a PGH2/TXA2 receptor antagonist (terutroban) failed to demonstrate superiority versus low-dose aspirin in a large phase 3 trial (Patrono and Rocca, 2010).
The presence of several EPs on the same cell, exerting different effects, is common for PGE2 (Rocca 2006; Hoshikawa et al., 2009; Markosyan and Duffy, 2009). The variable PGE2 concentration within the microenvironment possibly elicits an activatory or inhibitory modulation, finely tuning platelet responsiveness. Thus, the EP3 blockers in clinical development (Singh et al., 2010) might have the advantage of leaving the inhibitory PGE2/EP2 pathway unopposed. Selective agonism of platelet EP2 deserves further investigation as a potential adjuvant strategy for the prevention of atherothrombosis. Finally, the strong positivity of EPs in megakaryocytes might also suggest a role in megakaryopoiesis. Indeed, PGE2 is the prevalent eicosanoid during megakaryocyte maturation (Rocca et al., 2002). Although we detected EP1 protein in platelets and megakaryocytes, the lack of commercially available specific agonists precluded further characterization of its role in platelet function. In conclusion, we have shown that the net effect of PGE2 at low concentrations consists of enhancing stabilization and amplification of platelet response. This seems to result from a complex balance between EP3 variants- and EP2-mediated responses that positively and negatively modulate platelet aggregation, respectively. These findings should help in guiding the selection of novel antithrombotic strategies based on platelet EP modulation.
Participated in research design: Petrucci, De Cristofaro, Ranelletti, Patrono, and Rocca.
Conducted experiments: Petrucci, De Cristofaro, Rutella, Ranelletti, Pocaterra, Lancellotti, Habib, and Rocca.
Performed data analysis: De Cristofaro, Rutella, Habib, and Rocca.
Wrote or contributed to the writing of the manuscript: Patrono and Rocca.
We gratefully acknowledge the contribution to this article of the late Dr. Nicola Maggiano, who will be always remembered among his colleagues and friends.
This work was supported by the European Commission [FP6 funding LSHMCT-200-005033]; and the Catholic University School of Medicine of Rome [funding D1 70200368].
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- prostanoid E receptor
- vasodilator-stimulated phosphoprotein
- fluorescein isothiocyanate
- tetramethyl rhodamine isothiocyanate
- 6-isopropoxy-9-oxoxanthene-2-carboxylic acid
- 2-(3-hydroxyoctyl)-5-oxo-1-pyrrolidineheptanoic acid
- platelet-rich plasma
- light-transmittance aggregation
- arachidonic acid
- lag time
- platelet VASP
- forward scatter
- side scatter.
- Received September 6, 2010.
- Accepted November 2, 2010.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics