Microsomal prostaglandin E2 synthase-1 (mPGES-1) constitutes an essential player in inflammation and is involved in the pathogenesis of rheumatoid arthritis. Platelets participate in the regulation of inflammatory processes by the release of proinflammatory mediators and platelet-derived microparticles (PMPs). However, the role of the inducible mPGES-1/PGE2 pathway in platelet functions has not been investigated. In the present study we report a significant impact of mPGES-1 on platelet functions during inflammation. Wild-type (WT) and mPGES-1−/− knockout (KO) mice were stimulated with lipopolysaccharide (LPS) for 24 h. Platelet counts and activation were assessed by flow cytometry analysing CD62P–CD154 expression, PMP numbers, platelet–leukocyte aggregates and platelet aggregation. The accumulation of platelets and fibrinogen in the liver was analysed by immunofluorescent staining. In native platelets from WT and mPGES-1 KO mice, there were no differences among the investigated functions. After LPS treatment, the number of platelets was significantly decreased in WT, but not in KO mice. Platelet activation, platelet–leukocyte aggregates and PMP numbers were all significantly lower in KO mice compared with WT mice after LPS treatment. In addition, KO mice displayed a significant reduction in platelet aggregation ex vivo. In the liver of LPS-stimulated WT and KO mice, there were no differences in platelet accumulation, although the percentage of total vessel area in the KO liver was significantly lower compared with the WT one. Our results demonstrate that systemic inhibition of mPGES-1 prevents platelet activation, which should have important implications with regard to the cardiovascular safety of mPGES-1 inhibitors.
Deletion of mPGES-1 in mice leads to reduced platelet activation and aggregation, platelet–leukocyte aggregates and microparticle formation after LPS stimulation.
In the liver of LPS-stimulated WT and KO mice, there were no differences in platelet accumulation, although the percentage of total vessel area in the KO liver was significantly lower compared with WT mice.
The data suggest a beneficial effect of mPGES-1 inhibition on platelet functions with regard to both inflammatory responses and cardiovascular complications. Also, systemic inhibition of mPGES-1 prevents platelet activation, which should have important implications for cardiovascular safety of mPGES-1 inhibitors.
Microsomal prostaglandin E2 synthase-1 (mPGES-1) catalyses the biosynthesis of PGE2 from cyclooxygenase-1/2 (COX-1/2)-derived prostaglandin H2 (PGH2). The mPGES-1 is increased several-fold by various proinflammatory stimuli such as tumour necrosis factor α (TNF-α) or interleukin (IL)-1β, and is commonly, but not exclusively, linked to COX-2; it may be regarded as the enzyme primarily responsible for the production of prostaglandin E2 (PGE2) at sites of inflammation [1,2].
PGE2 mediates several important physiological functions and is an essential player in the pathogenesis of inflammatory diseases and cancer [2–4]. Genetic deletion of mPGES-1, as well as its pharmacological inhibition, has been proven to be protective in several experimental models of inflammatory disorders, such as stroke, multiple sclerosis, atherosclerosis, osteoarthritis and rheumatoid arthritis (RA) [2,5]. In various experimental models of arthritis, mPGES-1 inhibition causes a reduction in disease incidence, inflammation, pain and joint damage [6,7]. The mechanisms behind delayed arthritis and symptom attenuation in mPGES-1-deficient mice have been explored and involve reduction of synovial hyperplasia, less infiltration of inflammatory cells, and reduction of osteoclast counts and the humoral immune response [6–8]. Based on these results, mPGES-1 seems a valuable target for suppressing multiple mechanisms involved in the pathogenesis of RA.
RA is a chronic autoimmune inflammatory disease characterized by synovial inflammation, cartilage and bone destruction. The joint inflammation in RA is a complex system involving many different cells such as synovial fibroblasts, chondrocytes, bone cells, leukocytes and multiple inflammatory mediators. Cardiovascular comorbidity is also common, probably because of the enhanced inflammation of the vessel walls . Manfredi et al.  demonstrated a link between platelet activation and vascular inflammation via an TNF-α-dependent pathway, which could render atherosclerotic plaques more prone to rupture in RA patients. Activated platelets have also been recognized as one of the major players in autoimmune diseases such as RA, by binding to neutrophils to provide a sticky surface to recruit leukocytes to the vessel walls, thus promoting inflammation further [11–13]. Furthermore, activated platelets expressing activation markers, such as P-selectin (CD62P) and CD40 ligand or CD40L (CD154) as well as platelet-derived microparticles (PMPs), are commonly found accumulated in the blood and synovial fluid of RA patients [14–16]. Beyond their role in haemostasis, platelets have been shown to play a part in the pathophysiology of RA by releasing cytokines, chemokines and lipid mediators [13,17,18]. It is also known that markers of platelet activation, CD40L and P-selectin, can attract and activate neutrophils . PMPs have been reported to be internalized by neutrophils , resulting in their activation. In the joints of RA patients, PMPs can be actively involved in the pathogenesis by exposing IL-1β and aggravating cytokine response from synovial fibroblasts . Boilard et al.  have shown that platelet-depleted mice exhibited a marked reduction of arthritis, confirming the important role of platelets in inflammatory arthritis.
A recent study demonstrated that human platelets do not express mPGES-1 and cannot produce inducible PGE2 . However, platelets express PGE2 receptors, so their functions might be modulated by the mPGES-1/PGE2 pathway, induced in inflammatory conditions, in other cells expressing mPGES-1.
Platelets, therefore, constitute a candidate to link active synovitis with systemic inflammation and cardiovascular manifestations in RA patients. In the present study, we aimed to understand the effect of genetic deletion of mPGES-1 on platelet number, activation, aggregation and release of PMPs in normal and inflamed mice.
MATERIALS AND METHODS
Mice with a deletion of the Ptges gene, which encodes mPGES-1 (mPGES-1−/−), and on a DBA/1lacJ genetic background, were generated by breeding heterozygous littermates as previously described . The mice's wellbeing and examinations to make sure that they were free from screened pathogens were supervised according to guidelines from the Swedish Veterinary Board (SVA). Mice were kept under climate-controlled conditions with a 12-h light–dark cycle, fed standard rodent chow and administered water freely. All mice experiments were executed according to the guidelines for the care and use of animals sanctioned by the Ethics Committee at the Karolinska Institutet, Stockholm, Sweden.
Animal experiments and collection of blood by cardiac puncture
The mPGES-1 wild-type (WT) and knockout (KO) mice (8–12 weeks old) were injected with 2 μg of lipopolysaccharide (LPS, Sigma-Aldrich) or saline (Sigma-Aldrich), intraperitoneally for 24 h. Mice were anaesthetized with isoflurane (Baxter) and blood was slowly drawn by cardiac puncture with a 1-ml syringe containing 100 μl of 3.8% sodium citrate; after that it was transferred to Eppendorf tubes. In addition, mouse liver and spleen tissues were collected. The Regional Ethics Committee in Stockholm granted their approval (numbers N86/13 and N364/11).
Platelet-rich plasma (PRP) was prepared by centrifugation. Briefly, citrated whole blood was immediately centrifuged at 170 g for 10 min at room temperature (RT). The PRP was then carefully isolated by pipetting the upper two-thirds of the plasma. All analyses were performed within 1 h.
Western blot analysis of mPGES-1 expression in murine platelets
Spleen tissue and PRP from LPS-stimulated WT and mPGES-1 KO mice were lysed on wet ice for 30 min in Tissue Protein Extraction Reagent (T-PER) (Thermo Scientific) which was supplemented with 1× complete protease inhibitor cocktail tablet (Roche Diagnostics GmbH). Protein concentration was determined using the NanoDrop technique. An equal amount of protein was separated by gel electrophoresis on a NuPage Novex Bis-Tris gel system (Invitrogen). A Trans-Blot SD semi-dry transfer cell (Bio-Rad Laboratories AB) was used to transfer proteins to PVDF. Transference of proteins was followed by a blocking step with 5% milk (Bio-Rad Laboratories AB) in PBS (0.1% Tween 20) for 30 min at RT on a rocker, followed by incubation with primary antibody (polyclonal mPGES-1–Cayman Chemical) at 4°C overnight. Primary antibody was rinsed by washing the membrane three times for 10 min in PBS (0.1% Tween 20). Then secondary antibody horseradish peroxidase-coupled anti-rabbit IgG from donkeys (GE Healthcare) was applied and incubated with the membrane for 1 h at RT. After incubation with the second antibody, the membrane was washed three times for 10 min in PBS (0.1% Tween 20). Visualization of the bands was carried out using an enhanced chemiluminescence (ECL) kit (GE Healthcare) on film (Amersham Hybond, GE Healthcare). β-Actin antibody (Sigma-Aldrich) was used as a protein-loading control. A549 cells overexpressing mPGES-1 were used as a positive control.
Flow cytometry analysis of platelets and PMPs
Platelets from whole blood and PMPs from PRP were analysed using flow cytometry to investigate platelet function. Briefly, the samples (either whole blood or PRP) were labelled with anti-CD61-Alexa 488 (AbD Serotec) combined with anti-CD62P-PE (eBioscience) and anti-CD154-APC (eBioscience) antibodies. Platelets and PMPs were defined by their size characteristics and CD61 expression. Furthermore, co-expression of CD62P and CD154 was investigated. In addition, we also measured platelet–leukocyte aggregates in whole blood by gating for CD62P expression in the leukocyte gate (leukocytes were defined by size).
Platelet aggregation was measured by a flow cytometry-based platelet aggregation assay (FCA) using a modified protocol . Briefly, whole blood was divided into two equal portions, each portion then being labelled with CD61-Alexa 488 or CD61-PE (AbD Serotec). After incubation for 15 min at RT, in the dark, the portions were combined (1:1, v/v). After 15 min of incubation (RT, in the dark) the platelets were activated with 6.5 μM ADP (Roche) and again incubated for 10 min, with shaking (1000 rev./min) at RT. Cell fix (1:20, v/v; BD Biosciences) was added to the samples before analysis by flow cytometry. The platelets were defined by their size. Moreover, the platelets that expressed both markers simultaneously (CD61-Alexa 488 and CD61-PE) were considered to be aggregated. The results are presented as percentage aggregated platelets.
Immunofluorescent staining of liver from LPS-stimulated WT and mPGES-1 KO mice
Livers from the WT and mPGES-1 KO mice were analysed for total expression of CD41 and fibrinogen presence/accumulation in the vessels. Briefly, frozen liver tissue was sectioned into 7-μm-thin sections using a cryostat and fixed with 2% formaldehyde for 20 min. Liver sections were washed with PBS containing 0.1% saponin, pH 7.4 (PBS-S) for 10 min and blocked in 20% normal mouse serum for 45 min at RT. Sections were then incubated with primary antibodies (anti-CD41, 1:1000, v/v, AbD Serotec; anti-fibrinogen, 1:5000, v/v, Dako) containing 3% human serum, overnight at RT. Slides were washed with PBS-S and incubated with secondary antibodies conjugated with the fluorescent dye (anti-rabbit Alexa Fluor 594, anti-rat Alexa Fluor 488, 1:1000, v/v, Molecular Probes) at RT for 30 min followed by additional PBS-S washes. After the final wash with PBS, sections were mounted in PBS–glycerol (1:10, v/v). PBS-S was used for all antibody dilutions. Isotype-matched irrelevant antibodies were used as a negative control. After immunofluorescent labelling, the CD41- and fibrinogen-positive staining was quantified using computer-assisted image analysis. One section per mouse (n=5 mice per group) was used for the analysis; the entire section was examined with an average of 10 images per section, depending on the size and the quality of the tissue area. The images were analysed using the Qwin IM500 software (Leica). For quantification of platelets, the total area of CD41 staining was measured and normalized to the total area of the tissue. For quantification of fibrinogen in the vessels, the area of positive staining was measured in vessels and normalized to the total area of the vessels. For comparison of analysis performed on different occasions, the values were also normalized to the maximal value obtained at each staining.
Before statistical analysis, data were log transformed, if necessary, to obtain a normal distribution. WT and KO mice were compared using the unpaired Student's t-test. P<0.05 was considered to indicate a statistically significant difference. Statistical analysis was performed using GraphPad Prism (6.0h, GraphPad Software Inc.) software. Data are expressed as means±S.E.M.s.
Murine platelets do not express mPGES-1
To address whether murine platelets express mPGES-1, we performed Western blot analysis on spleen and platelet cell lysate from WT and mPGES-1 KO mice. As expected the spleen tissue from LPS-induced WT mice expressed mPGES-1 whereas mPGES-1 was not expressed in platelets. There was no expression of mPGES-1 in either spleen tissue or platelets from LPS-induced KO mice (Figure 1).
Protein expression of mPGES-1 in murine platelets
Effect of mPGES-1 deletion on platelet activation and platelet–leukocyte aggregates in mice
The platelet whole blood counts were reduced in WT mice treated with LPS compared with control mice (Figure 2). In contrast, the platelet counts in KO mice were similar in LPS-stimulated and untreated mice, although significantly higher than in LPS-stimulated WT mice (8.8±3.0 vs 5.3±1.1×105 platelets/μl, P<0.001) (Figure 2). Platelet activation assessed by CD62P and CD154 expression was found to be increased in both WT and KO mice treated with LPS (Figures 3A and 3B). However, treated WT mice displayed a significantly higher degree of platelet activation compared with KO mice (CD62P: 12.0±3.9 vs 4.3±2.8%, P<0.001; CD154: 11.4±4.3 vs 6.1±2.1%, P<0.01) (Figures 3A and 3B). The level of platelet–leukocyte aggregates was higher in LPS-stimulated WT mice than in KO mice (10.1±4.8 vs 6.4±2.9%, P<0.05) (Figure 3C). In addition to measuring surface markers on platelets, we also measured PMP levels in PRP. Our results demonstrate that WT mice, after stimulation with LPS, showed higher levels of PMPs compared with KO mice (6.1±1.7 vs 2.5±1.1×103 PMP events, P<0.01) (Figure 3D).
Platelet counts in whole blood
Platelet activation in WT and mPGES-1 KO mice
Effect of mPGES-1 deletion on platelet aggregation in mice
Platelet aggregation measured by flow cytometry
Platelet and fibrinogen accumulation in the liver from LPS-stimulated WT and mPGES-1 KO mice
Previous studies have provided evidence that stimulation with LPS might induce a translocation of platelets to tissues such as the lungs and liver . We analysed platelet accumulation and aggregation in the mouse liver using immunofluorescent staining. The expression of CD41 and fibrinogen was examined in livers from WT and KO mice stimulated with LPS for 24 h. No differences were seen in CD41 expression in WT compared with KO mice (Figure 5A). Fibrinogen staining was detected in vessels only (Figures 5B and 5D). We could observe a trend towards less fibrinogen staining in the vessels of KO mice stimulated with LPS compared with WT mice (P=0.080) (Figure 5B). In addition, the percentage of total vessel area was significantly lower in KO compared with WT mice (P=0.042) (Figure 5C).
Immunofluorescent staining of platelet accumulation and aggregation in mouse liver
It is well established that platelets have not only a function in haemostasis but also a role in inflammation . Platelets cross-talk with immune cells and recruit leukocytes to sites of inflammation . P-selectin on platelets binds to P-selectin glycoprotein ligand-1 (PSGL-1) expressed on neutrophils, and then transports them to the scene of inflammation . Soluble CD40L released by activated platelets, on the other hand, binds to its cognate receptor CD40, expressed on neutrophils, and modulates inflammatory immune responses . These abilities of platelets make them major players in inflammation; thus reduction of platelet activation may subsequently lead to impaired neutrophil activation and consequently reduced inflammation. Moreover, it has been shown by Gudbrandsdottir et al.  that platelets bind to lymphocytes and promote their activation, proliferation and secretion of IL-17 and interferon-γ via glycoprotein VI expressed on CD4+ T cells in RA patients. Ertenli et al.  showed that synovial fluid from RA patients contains high levels of activated platelets. They have also compared plasma from RA patients and healthy controls and demonstrated higher levels of P-selectin in RA patients compared with healthy controls. Moreover, there is a correlation between the disease activity and the amount of soluble P-selectin and CD40L in plasma of RA patients . Several publications have revealed that deletion or pharmacological inhibition of mPGES-1 markedly reduces inflammatory responses in experimental models of arthritis via different mechanisms [6,7]. We observed significantly higher levels of P-selectin and CD40L expression in WT mice treated with LPS compared with KO mice, reflecting increased platelet activation by PGE2.
LPS stimulation of mice is known to induce platelet activation and aggregation while also reducing the platelet counts due to consumption of platelets, presumably by attachment to cells/endothelium or by accumulating in the liver or lung [22,27]. Ståhl et al.  demonstrated that human platelets bind LPS derived from E. coli and become activated via the complex of toll-like receptor 4 (TLR4) and CD62. As expected, the platelet counts in whole blood diminished in WT mice treated with LPS compared with WT control mice. In addition, we showed that WT mice stimulated with LPS exhibited a significantly higher degree of platelet activation compared with KO mice. Our data suggest an additional mechanism behind the anti-inflammatory effect of mPGES-1 deletion via modulation of platelet activation.
Non-steroidal anti-inflammatory drugs (NSAIDs) are frequently used worldwide to dampen pain in patients suffering from RA. Long-term consumption of NSAIDs sometimes brings side effects and is associated with increased cardiovascular hazard due to suppression of COX-2-derived prostacyclin (PGI2) in patients and animal models of inflammation . In contrast to NSAIDs, which act as COX inhibitors, the inhibition or deletion of mPGES-1 does not alter blood pressure or predispose to thrombosis in mice models . One explanation for this observation could be the reduced PGE2 production leading to a re-direction of accumulated PGH2 to increased production of PGI2, which is cardioprotective [30–32]. These observations are in line with our results about platelet accumulation and aggregation in the mouse liver. Although no significant differences were seen for CD41 expression in WT compared with KO mice, we observed a trend towards less fibrinogen accumulation, suggesting reduced platelet activation. Furthermore, we noted that the percentage of total vessel area was significantly lower in KO compared with WT mice; one plausible explanation for what causes this effect could indeed be the lack of inducible PGE2, which is known to contribute to vasodilatation, increased blood flow and hyperpermeability . Indeed, strongly increased vascular diameter was detected in response to treatment with PGE2in vivo . Also worth mentioning, there are several mechanisms involved in the vasodilatation, increased blood flow and hyperpermeability during inflammation. Thus, co-induction of COX-2 and expression of the inducible isoform of nitric oxide synthase was demonstrated in rat aorta and brain in response to LPS, consequently leading to the production of vasodilating PGI2, PGE2 and nitric oxide (NO) [34,35]. In our mouse model, all these mediators could contribute to vasodilatation in the liver. However, the lack of the inducible mPGES-1 led to reduced vessel area, suggesting that LPS-induced PGE2 is involved in vasodilatation in the liver via direct or indirect mechanisms.
Previous reports have described exogenous effects of PGE2 on human platelets, but the endogenous role of the COX/mPGES-1 pathway on platelet functions remains poorly understood. Several studies have shown that PGE2 acts as a positive or negative modulator of platelet functions depending on its concentration, as well as the types of expressed EP receptors [36–38]. There are several EP receptors expressed on murine platelet surface, resulting in diverse and often opposite biological responses [39,40]. PGE2 binding to the EP3 receptor expressed on platelets triggers platelet activation and promotes thrombosis [41,42], whereas EP4 stimulation leads to inhibition of platelet aggregation . It is important to note that our results demonstrate that murine platelets do not express mPGES-1 in an inflammatory setting, which is in agreement with data showing that mPGES-1 is not expressed in human platelets . Consequently, the observed effects described herein depend on the mPGES-1/PGE2 pathway induced in other cells, rather than the platelets themselves, on LPS treatment in vivo. It is well established that mPGES-1 is essential for LPS-induced synthesis of PGE2 [43,44] and mPGES-1 deletion results in no augmentation of PGE2 production in response to LPS [44,45]. Exposure to circulating LPS causes induction of mPGES-1 in murine brain, lung, heart, liver and colon . Furthermore, marked up-regulation of mPGES-1 was detected in LPS-induced monocytes  and peritoneal macrophages [7,43], as well as in macrophages accumulated in the liver after acute endotoxaemia . The inducible mPGES-1 is also the essential enzyme involved in PGE2 production by vascular cells in inflammatory conditions [49,50]. Thus, platelets could be activated in response to PGE2 produced by the inducible mPGES-1 in multiple cells in both the circulation and different tissues.
Another interesting finding was the reduction of PMPs in mPGES-1-deleted mice. Our results indicate that WT mice, after stimulation with LPS, showed higher levels of PMPs compared with KO mice. Activated platelets are known to release PMPs which play a major role in communication between cells . PMPs are produced by budding and fission of the plasma membrane and therefore exhibit antigens from the cell of origin, such as receptors and cytokines . Several studies have shown that PMPs play an important role in the pathogenesis of RA by promoting inflammation [14–16]. Boilard et al.  proposed that PMPs can interact with fibroblast-like synovial cells and participate in the pathophysiology of arthritis. The same authors revealed that platelet-depleted mice exhibited a reduction in arthritis assessed by clinical scoring and histological analysis . Furthermore, Jüngel et al.  demonstrated that PMPs strongly induce the COX-2/mPGES-1/PGE2 axis in human synovial fibroblasts. These data provide further evidence for the role of PMPs in inflammation and pinpoint the importance of the reduction of PMPs detected in mPGES-1-deleted mice. Our findings that mPGES-1 deletion in mice leads to a decreased number of PMPs henceforth support the pathogenic role of mPGES-1 in PMP release.
In addition to our standardized platelet function assay, we used an assay for detecting platelet aggregation (FCA) by flow cytometry, investigating functional properties as well. This method allows detection of platelet aggregation in small volumes and low platelet numbers, and is particularly useful for experiments in mice. In the present study we demonstrate that mPGES-1-deleted mice present with less platelet aggregation, supporting our findings of low expression of P-selectin and CD40L, and low platelet–leukocyte aggregates. In line with this, Fabre et al.  demonstrated that a low concentration of PGE2 enhances platelet aggregation in mice, and activation of EP3 is sufficient to mediate the proaggregatory actions of a low PGE2 concentration. Ma et al.  showed that PGE2 potentiate platelet aggregation via EP3 by increasing calcium (Ca2+) and decreasing cAMP in mice. Furthermore, PGE2 facilitates the initiation of arterial thrombosis, leading to atherothrombosis in vivo and, by antagonizing the platelet EP3 receptor, this effect is abolished . The impact of systemic mPGES-1 deletion on platelet function has not been investigated previously. Our results indicate that mPGES-1-deficient mice have significantly less interaction between platelets and leukocytes as well as a lower expression of activation markers on platelets compared with WT mice.
In conclusion, mPGES-1 deletion in mice leads to less platelet activation, PMP formation, platelet–leukocyte aggregates and reduced platelet aggregation after LPS stimulation compared with WT mice. The results suggest a beneficial effect of mPGES-1 inhibition on platelet functions with regard to both inflammatory responses and cardiovascular safety.
Joan Raouf carried out the animal experiments, conducted the immunoblot analysis, participated in the design of the study and interpretation of the data, and contributed to drafting, revising and finalizing the manuscript. Fariborz Mobarrez carried out the flow cytometry analysis, participated in the design of the study and interpretation of the data, and contributed to revision of the manuscript. Karin Larsson conducted the immunofluorescent staining and participated in revising the manuscript. Per-Johan Jakobsson participated in the design of the study and revision of the manuscript. Marina Korotkova participated in the design of the study, interpretation of the data and revision of the manuscript. All authors read and approved the final manuscript.
These studies were supported by the Swedish Rheumatism Association, King Gustaf V 80 years Foundation, the Swedish Society of Medicine, Karolinska Institutet Foundation, and Medical Training and Research Agreement (ALF) funding [grant number SLL20140423].
Per-Johan Jakobsson is a member of the board of directors of NovaSAID.
CD40 ligand marker
flow cytometry-based platelet aggregation assay
microsomal prostaglandin E2 synthase-1
non-steroidal anti-inflammatory drug
P-selectin glycoprotein ligand-1
tumour necrosis factor-α