GLYCOGEN SYNTHASE KINASE-3 NEGATIVELY REGULATES TISSUE FACTOR EXPRESSION IN MONOCYTES INTERACTING WITH ACTIVATED PLATELETS

Abstract

Background

At the site of vascular injury monocytes (MN) interacting with activated platelets (PLT) synthesize tissue factor (TF) and promote thrombus formation. Intracellular signals necessary for the expression of TF in MN, in the context of a developing thrombus, remain unknown.

Objective

The study was designed to investigate the role of glycogen synthase kinase 3 (GSK3), a serine-threonine kinase, down-stream insulin receptor pathway, on PLT-induced TF expression in MN.

Methods

To this purpose we used a well characterized in vitro model of human MN-PLT interactions that allows detailed analysis of TF activity, TF protein and gene expression.

Results

The results demonstrated that, in MN interacting with activated PLT: 1) TF activity, antigen and mRNA were low until 8–10 hours and dramatically increased thereafter, up to 24 hours. 2) According to the kinetics of TF expression in MN, GSK3β undergoes phosphorylation on serine 9, a process associated with down-regulation of enzyme activity. 3) Pharmacological blockade of GSK3 further increased TF expression and was accompanied by increased accumulation of NF-kB, in the nucleus. 4) Blockade of phosphoinositide-3 kinase (PI(3)K) by wortmannin inhibited PLT-induced TF expression. 5) According to the established role of GSK3 down-stream insulin receptor, insulin increased PLT-induced TF expression in a PI(3)K-dependent manner.

Conclusion

GSK3 acts as molecular brake of the signaling pathway leading to TF expression in MN interacting with activated PLT. PI(3)K, through Akt-dependent phosphorylation of GSK3, relieves this brake and allows TF gene expression. This study identifies a novel molecular link between thrombotic risk and metabolic disorders.

INTRODUCTION

Synthesis of TF, the initial activator of coagulation cascade leading to thrombin generation, is a key effector function of human monocytes (MN) [1] that may play an important role in fibrin accumulation in a growing thrombus, where circulating MN are actively recruited by P-selectin expressed by activated platelets (PLT) [2]. In fact, the interaction with activated PLT not only recruits MN to the site of vascular injury but also triggers the expression of TF procoagulant activity [2–4].

P-selectin, alone or in concert with other PLT-derived products, may trigger de novo synthesis of TF in MN through the regulation of gene expression at the transcriptional level, [5] and stimulate phosphatidylserine exposure and thrombin generation on the surface of monocytic cells [6]. Moreover, the capacity of soluble circulating P-selectin in inducing a hypercoagulable state has been demonstrated in mice expressing a modified form of P-selectin lacking the cytoplasmic domain, and high levels of the soluble form of the protein in the circulation [7], and in mice with hemophilia A injected with a soluble recombinant form of P-selectin [8].

Therefore, consistent observations suggested that engagement of P-selectin glycoprotein ligand-1 (PSGL-1) plays a pivotal role in inducing procoagulant responses in MN. However it is plausible to hypothesize that full expression of inflammatory and procoagulant phenotype requires a complex cross-talk between MN and PLT, which remains to be fully elucidated. In this context, the intracellular events that signal gene expression and TF synthesis remain completely unknown.

Clarifying the molecular steps, triggered by activated PLT, that mediate expression of procoagulant TF in MN, is fundamental for understanding hypercoagulable state in patients at high cardiovascular risk and for the identification of novel targets for anti-thrombotic interventions.

Glycogen synthase kinase-3 (GSK3), a serine-threonine kinase, originally identified as the enzyme that phosphorylates and inhibits glycogen synthase, down-stream insulin receptor, is now recognized to play a pivotal role in the regulation of many cellular functions including gene expression, cell movement, cell cycle and cell survival. For these reasons GSK3 is considered a promising target for pharmacological treatment of diabetes, neurological and inflammatory disorders and cancer [9].

Here we report the fundamental role of GSK3 in the regulation of TF expression in MN interacting with activated PLT, a cellular model that mimics the context of a developing thrombus. Overall the results of our study demonstrate that GSK3 negatively regulates TF gene expression that occurs in MN following prolonged interaction with activated PLT. Activation of phosphoinositide-3 kinase (PI(3)K), through Akt-dependent phosphorylation of GSK3 relieves this brake and allows TF gene expression. Our data identify a novel mechanism linking increased thrombotic risk and metabolic disorders in which GSK3 activity in circulating MN may be altered as consequence of the disease or of pharmacological treatment.

METHODS

Cells isolation and experimental procedure. PLT and MN were isolated from whole blood obtained from healthy donors who gave their informed consent to participate in the study and did not take drugs within ten days before blood donation. Approval was obtained from the Institutional Review Board for these studies. PLT were prepared following standard procedures [10]. MN were purified to more than 70% by two steps centrifugation on Lymphoprep (Axis-Shield, Oslo, Norway) followed by Percoll (GE Healthcare Bio-Science, Upsala, Sweden) gradient as previously described [11]. The platelet preparations were controlled for the presence of contaminating leukocytes by flow cytometry. Contaminating leukocytes (mainly lymphocytes) were usually, less than 2 cell per 20.000 platelets.

Washed PLT and MN were resuspended in serum-free RPMI-1640 medium. PLT, resting or activated for 30 minutes with 25 μMoles/L of PAR-1 activating peptide, SFLLRN, (TRAP-6) (Bachem, Bubendorf, Switzerland) in the presence of RGDS (Sigma-Aldrich, Milan, Italy) (400 μMoles/L), in order to reduce PLT homotypic aggregation, were coincubated with isolated MN at 37°C for different times. MN, unstimulated or challenged with endotoxin, 100 ng/ml(LPS) (Escherichia Coli serotype 055:B5, Sigma-Aldrich, Milan, Italy) and PLT alone were incubated in parallel in the same conditions. GSK-3 inhibitors: SB216763 and SB415286, Azakenpaullone and LiCl, (all from Sigma-Aldrich), insulin (Sigma-Aldrich), wortmannin (Calbiochem, La Jolla, CA), or NFkB activation inhibitor BAY117085 (Calbiochem), were added to MN for 15 minutes before starting coincubation with PLT.

Procoagulant activity of TF was measured in cell lysates by one stage clotting time as previously described [12]. Briefly, 100 μl of cell lysates was added to a prewarmed tube containing 100 μl of pooled normal human plasma. After addition of 100 μl (20 mMoles/L) CaCl₂, clotting time was determined by Amellung Coagulometer KC4A (Mascia Brunelli, Milan). Clotting times were converted to arbitrary units (aU) by interpolation with a standard curve generated with serial dilutions of human recombinant thromboplastin (Recombiplastin, Hemoliance, Instrumentation Laboratory). Since the concentration of recombinant TF (Recombiplastin, Hemoliance, Instrumentation Laboratory) used to produce the standard curve of activity is not reported by manufacturer, we have measured TF concentrations in the serial dilutions used to produce standard curve of activity, by ELISA. The correlations between TF activity and concentration of TF antigen in the recombinant TF preparations are reported in supplemental figure 1.

Factor VII-deficient plasma and anti-TF antibodies were from (American Diagnostic Inc, International Laboratory, Milan, Italy).

TF antigen was measured in cell lysates by ELISA-Imubind (International Laboratory) according to the manufacturer instruction.

Western blot analysis of GSK3 and Akt. The levels of GSK3 phosphorylation were analyzed by Western blotting using phosphospecific antibodies recognizing phosphorylated (p) serine-21 and serine-9 on GSK3α and βrespectively (anti pGSK3 antibodies were from Santa Cruz Biotechnology, Inc). In parallel total GSK3 was detected by an anti-whole protein antibody (Cell Signaling Technology). Similarly phosphorylated and total Akt were detected by anti-pAkt1/2/3(ser473) and anti-Akt total protein. Both antibodies were provided by Santa Cruz Biotechnology, Inc.

Real-time reverse transcriptase polymerase chain reaction (RT-PCR). Total RNA was extracted from cells by thiocyanate/cesium chloride method. One μg of total RNA was converted to cDNA using Moloney murine leukemia virus reverse transcriptase (Applied Biosystems). Real-time PCR measurements of TF were performed using 10 ng of cDNA, 50 nMoles/L of each primer and SYBR Green master mix (Applied Biosystems) in 20 μl reactions, by Applied Biosystems PRISM 7500 Fast Real-time PCR System. The reverse transcription was primed by random examers. The following TF gene sequences were used as validated primers: cag tga ttc cct ccc gaa ca, (forward), tgc ctt tct aca act gtg tag ag, (reverse) The sequences are located on exons 5 and 6, at the position 667 and 827 for forward and reverse sequence, respectively. Preliminary validation assays showed a band on agarose gel electrophoresis of the expected size for amplification of TF cDNA in samples of MN/PLT coincubated for 24 hours. TF cDNA band was present in samples treated with DNAaseI but not in samples in which M-MuLV was omitted indicating that the signal is not from contaminating genomic DNA (Supplemental figure 2).

Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as house keeping gene. Each sample was measured in triplicate and the generated data were analyzed with the SDS 2.0 software (Applied Biosystems) by the delta-delta method (2^−DDCT) for comparing relative expression results. Resting MN incubated alone was considered the reference sample.

NF-kB DNA-binding assay. Nuclear extracts were incubated with P³² radiolabeled double-stranded oligonucleotide probes containing NF-kB binding site 5'-AgTTgAggggATTTCCCAggC-3' (Santa Cruz Biotechnology). Protein-DNA complexes were separated by 5% non denaturating acrylamide gels in 0.5× TBE. Autoradiography was performed on Kodak XAR films.

Statistical analysis

Data are expressed as mean±SEM. Statistical differences between two groups were determined by paired t test. To test for differences across different treatment groups, repeated-measures ANOVA and Dunnett test were used. Statistical significance was defined as P<0.05.

RESULTS

Activated PLT induce TF expression in MN

In initial experiments we assessed, procoagulant activity of lysates of MN incubated for different times with autologous PLT activated by thrombin receptor (PAR-1) activating esapeptide (TRAP-6). For comparison MN alone were stimulated by endotoxin (LPS). Procoagulant activity induced by LPS, reached a maximum at 5h and then slowly declined. In contrast procoagulant activity in mixed PLT/MN suspensions was consistently increased over basal levels only after 6–10 hours of incubation and steadily increased thereafter. When MN and PLT were incubated separated for 24 hours and mixed just before cell lysis, procoagulant activity of lysates was not significantly different from that expressed by mixed cell population at time 0 (Figure 1A). Therefore PLT-dependent increase of procoagulant activity in MN required prolonged cell-cell interaction. PLT-induced procoagulant activity in mixed cell suspension incubated for 24 hours could be consistently appreciated at a PLT/MN cell ratio of 50 and significantly increased with increasing PLT concentration (Figure 1B). Moreover the effect of TRAP-6-activated PLT was significantly higher than that of non-activated PLT (Figure 1C).

Figure 1. PLT induce TF in MN.

A) PLT were activated by TRAP-6 (25 μMoles/L) in the presence of RGDS (400 μMoles/L) for 30 minutes before mixing with MN (PLT/MN ratio 200/1). Mixed cell suspensions were incubated for different times before lysis. In parallel MN and TRAP-activated PLT were incubated separated for 24 hours and recombined just before lysis. For comparison MN were challenged with 100 ng/ml of endotoxin (LPS). Procoagulant activity was assessed by one stage clotting time. (see Materials and Methods).

Data are representative of 2 experiments performed with cells from different donors.

B). MN mixed with increasing concentrations of resting or TRAP-activated PLT were incubated for 24 hours before lysis. Graphs are representative of 2 experiments performed with cells from different donors.

C) Bars are means and SEM of 6 experiments performed with cells from different donors. Asterisk (*) indicates P<0.05 by paired t test

D) MN and resting or TRAP-activated PLT were incubated alone or in mixed cell suspensions (ratio 1/200) for 24 hours. At the end of incubation cells were lysed and TF antigen was assessed by ELISA. TF antigen in PLT alone was under the assay detection limit (25 pg/ml). In MN alone the antigen was barely detectable 35±6 and 30±2 pg/ml in the absence or in the presence of TRAP and RGDS. Data are Means±SEM of 3 experiments performed in triplicates with cells from different donors.

Asterisk (*) indicates P<0.05 by paired t test

E) MN were incubated with resting or TRAP-activated PLT at 200/1 ratio. At each time point RNA was extracted and converted to cDNA. Real-time PCR measurements of TF were performed using validated primers. GAPDH served as house keeping gene. Each sample was measured in triplicate and data were elaborated by the delta-delta method (2^−DDCT) for comparing relative expression results. MN incubated alone for different times was considered the reference sample.

Data are Means±SD of triplicates from one representative of 3 experiments performed with cells of different donors.

Addition of polymixinB (10 μg/ml) completely abolished the activity of LPS, 0.6±0.004 and 42.0±11.15 arbitrary units (means±SEM of n=10 experiments with cells from different donors) in the presence or in the absence of polymixinB, respectively. In contrast, PLT-induced procoagulant activity was similar in the presence or in the absence of polymixinB, 56.16±14.6 versus 65.0±22 arbitrary units (means±SEM of n=14 experiments with cells from different donors), respectively. Thus, excluding an important contribution of contaminating LPS.

The procoagulant activity expressed by human MN is mainly of TF type [13]. We used factor VII-deficient plasma and an inhibitory antibody against TF to confirm that procoagulant activity in mixed cell population was mediated by TF. When the coagulation assay was performed with FVII-deficient plasma, procoagulant activity was reduced by 95,2% respect to the activity measured in normal plasma. Treatment of cell lysates with an inhibitory anti-TF antibody reduced the procoagulant activity by 84,8% respect to the activity measured in samples treated with an irrelevant mouse monoclonal antibody (data are means of 2 different experiments).

The role of TF was further strengthened by measurement of TF protein by ELISA. As shown in Figure 1D, TF protein, barely detectable in lysates of PLT or MN incubated alone (lower than 25,0 and 40,0±14 pg/ml, respectively), increased to 360,0±85 and 1.120,0±475,0 pg/ml in MN coincubated for 24 hours, with resting or TRAP-6-activated PLT, respectively.

In order to test whether procoagulant activity measured in cell lysates reflects the expression of active TF on the cell surface, MN alone or MN/PLT mixed cell populations were incubated for different times. At each time point samples were divided in 2 and procoagulant activity was measured in parallel, in intact cells and in lysates.

The results showed that, at 24 hours, procoagulant activity increased from basal values of 0.08±0.001 and 0.03±0.01 to 2.44±1.01 and 32.04±10.67 arbitrary units, in intact cells or in lysates, respectively. At 48 hours, procoagulant activity in whole cells further increased to 13.35±5.75 while remaining stable in cell lysates (37.18±16.97 arbitrary units) (Supplemental Table 1).

Thus, a large part of TF expressed at 24 hours is not available (or inactive) for procoagulant activity but it becomes available at later time points. The mechanism that regulates these processes are currently under investigation.

Then we analyzed by real time RT-PCR the kinetics of TF mRNA expression in MN incubated alone or in the presence of resting or TRAP-activated PLT. Respect to MN alone, coincubation with resting or TRAP-activated PLT triggered an early, transient response at 1.5 h, followed by dramatic increase of TF mRNA at 18 hours of incubation. (Figure 1E). At this late time point we were unable detect TF mRNA in untreated or TRAP-activated PLT alone, identifying MN as the sole source of TF mRNA under these experimental conditions (data not shown). According to the procoagulant activity, TF mRNA expression induced by TRAP-activated PLT was consistently higher than that induced by resting PLT.

Procoagulant activity in mixed cell suspension was dramatically reduced to 25±5 and 15±3% of the control by actinomycin D and by puromicine that inhibit mRNA and protein de novo synthesis, respectively (data not shown).

Altogether these initial experiments confirmed previous data and definitely demonstrate that prolonged interaction with activated PLT induces TF expression in MN acting at the transcriptional level. These results prompted us to explore the role of GSK-3 as potential regulator of this process.

Role of PI(3)K-Akt pathway on the regulation of GSK3 activity in MN coincubated with activated PLT

Unlike many protein kinases, GSK3 is active in cells under resting conditions but its activity is regulated by phosphorylation-dependent inactivation. Thus in resting cells GSK3 activity is high, whereas upon specific agonist stimulation of the cells, GSK3 activity is reduced. Two isoforms of GSK3 are present in different cell types, GSK3-α and β. Both isoforms are mainly regulated by phosphorylation and dephosphorylation within the amino-terminal domain. GSK3-α is phosphorylated at the serine 21, and GSK3-β is phosphorylated at the equivalent site, serine 9. Phosphorylation at this site inhibits the enzyme activity [14]. We analyzed whether GSK3-β undergoes serine-9 phosphorylation, in our experimental conditions. Whole cell lysates from MN incubated alone or in the presence of TRAP-activated PLT were prepared at different times and GSK3-β phosphorylation was analyzed by western blotting with phospho-serine-9 specific antibody. The results of this experiment showed a time-dependent increase of GSK3-β phosphorylation, in mixed cell suspensions, that was consistently detectable in different experiments at the time points between 6 and 12 hours. In contrast GSK3-β phosphorylation did not occur in MN incubated in the absence of PLT (Figure 2A). A similar behavior was observed in some, but not all experiments for GSK3-α on serine 21 (data not shown).

Figure 2.

Time-dependent Ser-9 phosphorylation of GSK3β. R•••••••••(3)K.

A) PLT were activated with TRAP in the presence of RGDS for 30 minutes before addition of MN.

MN alone or mixed with activated PLT (ratio 1:100).

At each time point cells were lysed, and phosphorylated and GSK-3β was revealed by a phosho-serine-9 specific antibody or with an anti-GSK-3βwhole protein. All experiments were performed running samples in parallel gels. For this reason the figure shows densitometry analysis of both phosphorylated and total GSK3β. The results are representative of 3 experiments performed with cells from different donors.

Asterisks (*) indicate P<0.05 versus 0.25, 1 and 20 hours, by ANOVA.

B) MN were preincubated with wortmannin (100 nMoles/L) or DMSO for 15 minutes and then cultured in the absence or presence of activated PLT (ratio 1:100). At each time point cells were lysed, and phosphorylated GSK-3β was revealed by phosho-serine-9 specific antibody. The results are representative of 5 experiments performed with cells from different donors.

Asterisks (*) indicate P<0.05 versus control samples, by ANOVA.

Several protein kinases may phosphorylate and inactivate GSK3. Most notably, protein kinase B/Akt targets GSK3-α and-β at the sites Ser-21 or 9 and is the predominant enzyme that performs this task downstream PI(3)K. In fact, treatment of MN with wortmannin, at concentrations (100 nMoles/L) that selectively inhibits PI(3)K activity significantly reduced GSK3 ser-9 phosphorylation in MN incubated with activated platelets confirming that this process involved PI(3)K Figure 2B.

In order to further investigate whether PI(3)K-dependent phosphorylation of Akt takes place in MN after prolonged interaction with activated platelets, we performed time course experiments in the absence and in the presence of wortmannin. The results demonstrate time-dependent phosphorylation of Akt in MN coincubated with activated PLT. Akt phosphorylation increased, over basal levels, at 3 hours, declines at 6 hours and increased again at later time points. Treatment with wortmannin abolished Akt phosphorylation (Figure 3).

Figure 3.

PI(3)K mediates Ser-473 phosphorylation of Akt and Ser-9 phosphorylation of GSK-3β.

MN were preincubated with wortmannin (100 nMoles/L) or DMSO for 15 minutes.

PLT were activated with TRAP in the presence of RGDS for 30 minutes before addition of MN.

MN were cultured alone or mixed with activated PLT (ratio 1:100). At each time point cells were lysed, and phosphorylated and total Akt was revealed by a phosho-serine-473 specific antibody or with an anti-Akt whole protein. All experiments were performed running samples in parallel gels. For this reason the figure shows densitometry analysis of both phosphorylated and total Akt or GSK3beta. The results are representative of 2–3 experiments performed with cells from different donors for Akt and GSK3 analysis, respectively.

Asterisks (*) indicate P<0.05 versus basal level, by ANOVA.

Therefore, negative functional modulation of GSK3 occurred in MN interacting with PLT. The coincidence between the kinetics of GSK3 phosphorylation and the kinetics of TF expression suggested a relationship between these two events.

GSK3 limits PLT-dependent TF expression in MN

In order to analyze the functional relevance of GSK3 blockade, in our model, we tested the effect of four structurally different inhibitors of GSK3 activity, SB216763, SB415286, azakenpaullone and LiCl [9], on PLT-dependent TF expression in MN. All these compounds dose-dependently increased PLT-induced TF activity in MN after 24 hours of incubation (Figure 4A). The enhancing effect of GSK3 blockade was evident at all time points within 10 and 24 hours in mixed cell populations but not in MN alone (Figure 4B). TF activity in MN incubated alone for 24 hours with azakenpaullone remained unchanged (0.02±0.004 and 0.03±0.013 arbitrary units in the absence or in the presence of 5 μMoles/L azakempaullone, respectively. Data are means±SEM of 6 different experiments). Similarly, GSK3 blockade did not modify procoagulant activity in PLT incubated alone for 24 hours (0.147±0.029, 0,152±0,022, 0,112±0,025, 0,128±0,023 arbitrary units in the absence or in the presence of DMSO, 5 μMoles/L azakempaullone, 10μMoles/L SB216773 and 20 mMoles/L of LiCl, respectively. Data are means±SEM, n=3). Since previous studies have shown that thrombin receptor activation causes Ser9 phosphorylation of GSK-3beta in platelets, and that GSK-3β blockade amplifies-induced platelet activation (15), we reasoned that inhibiting this pathway in platelets could indirectly exaggerate the platelet-dependent stimulation of monocyte tissue factor expression. In order to further clarify this point we tested the effect of GSK3 inhibitors on CD40L release by platelets activated for different times (from 15 min, 2, and 24 hours) by TRAP-6 (25 μMoles/L) in our experimental conditions. The results of these experiments, that are shown in Supplemental Table 2, indicated that GSK3 inhibition by SB or lithium did not modify CD40L release during a prolonged time course. Thus it can be reasonably conclude that in our model, the effect of GSK3 inhibitors on PLT is negligible.

Figure 4.

GSK-3 blockade upregulates TF-activity in MN-PLT suspensions.

A) MN, pretreated with different GSK-3 inhibitors or the appropriate vehicle, were incubated with TRAP-activated PLT in mixed cell suspensions (ratio 1/200).

Data (% of control) are means±SEM of 6, 3, 6, 9 experiments performed with cells from different donors: for SB216763, SB415286, Azakenpaullone and LiCl, respectively. Control values corresponded to 71.4±24.8, 101.7±29.4, 56.3±31.3 and 231.18±102.3 arbitrary units, for SB216763, SB415286, azakenpaullone and LiCl, respectively. Data are means±SEM.

Asterisks (*) indicate P<0.05 versus untreated sample by ANOVA.

B) MN, pretreated with azakenpaullone (5μMoles/L) or the appropriate vehicle (DMSO) were incubated with or without PLT in mixed cell suspensions (ratio 1/200). At different times, cells were disrupted by 3 freeze-thaw cycles and the procoagulant activity was assessed by one stage clotting time. Data are from a representative of 5 experiments performed with cells from different donors.

Asterisk (*) indicate P<0.05 versus control (DMSO) sample by ANOVA

C) MN, pretreated with SB216763 (20 μMoles/L) or DMSO, were incubated TRAP-activated PLT (ratio 1/200) for 24 hours. At the end of incubation, cells were lysed and TF antigen was measured by ELISA. Data are expressed as percent of the control (DMSO). Control values corresponded to 369.6±94.2 pg/ml. Means±SEM of 3 experiments performed with cells from different donors.

Asterisk (*) indicates P<0.05 by paired t test.

D) MN, pretreated with SB216763 (20 μMoles/L) or DMSO, were incubated with resting or TRAP-activated PLT for 24 hours (ratio 1/200). At the end of incubation, total RNA was extracted from cells by thiocyanate/cesium chloride method and processed for TF mRNA analysis by real time PCR. Each sample was measured in triplicate and the generated data were analyzed with the SDS 2.0 software (Applied Biosystems) by the delta-delta method (2^−DDCT) for comparing relative expression results. The comparator for the delta-delta Ct analysis was the sample of MN alone. Data are reported as Means±SD of 3 different experiments with cells from 3 different donors.

Asterisk (*) indicates P<0.05 by paired t test.

In agreement with procoagulant activity, GSK3 blockade by SB217663 increased TF antigen (Figure 4C) as well as TF mRNA (Figure 4D) at 24 and 18 hours, respectively.

Role of PI(3)K-Akt pathway on PLT-mediated TF expression

These results prompted us to explore the functional role of PI(3)K-Akt cascade on TF expression in MNPLT suspensions.

As shown in figure 5, treatment with wortmannin or LY294002 consistently reduced TF activity expressed at 24 hours. Important to note, the inhibitory effect of both wortmannin or LY294002 was reversed by concomitant treatment of the cells with the GSK3 inhibitors (SB216773 or azakempaullone). This furthermore supports that, in our model, GSK3 acts downstream PI(3)K.

Figure 5.

PI3 kinase blockade inhibits TF-activity in MN-PLT suspensions in a GSK3-dependent manner.

MN, pretreated with DMSO, wortmannin (100 nMoles/L), SB216773 (20μMoles/L), wortmannin and SB216773, Ly294002 (5μMoles/L), azakempaullone (5μMoles/L) or Ly294002 and azakempaullone, were incubated with TRAP-activated PLT in mixed cell suspensions (ratio 1/200) for 24 hours. After incubation cells were disrupted by 3 freeze-thaw cycles, and procoagulant activity measured by one stage clotting time. Data are means of 2–4 experiments. Control values corresponded to 148.8±59.6 arbitrary units.

Since phosphorylation on Ser-9 and inhibition of GSK3 is a key step of insulin receptor signaling mediated by PI(3)K-Akt cascade, we have further explored our hypothesis testing the effect of insulin on PLT-induced TF expression in our model. In agreement with previous results, insulin induced a dose-dependent increase of TF activity, that was almost abolished by PI(3)K blockade (Figure 6).. In contrast, treatment with 1μMoles/L insulin did not affect TF expression in MN incubated alone for 24 hours (0.078±0.082 and 0.074±0.067 arbitrary units in the absence or in the presence of insulin, means±SEM, n=6). PI3-kinase blockade increased TF activity in MN alone activated by LPS (Figure 6).

Figure 6.

Insulin upregulates TF-activity in MN-PLT suspensions in a PI(3)K-dependent manner.

MN, pretreated with DMSO, wortmannin (100 nMoles/L), were incubated with TRAP-activated PLT in mixed cell suspensions (ratio 1/200), in the absence or in the presence of different concentrations of insulin, for 24 hours. For comparison control or wortmannin-treated MN were activated with LPS for 5 hours. After incubation cells were disrupted by 3 freeze-thaw cycles, and procoagulant activity measured by one stage clotting time. Data are means of 4 experiments performed with cells from different donors.

Asterisks (*) indicate P<0.05 versus no insulin and insulin 10 nMoles/L, by ANOVA.

PLT-induced NF-kB activity in MN is upregulated by GSK-3 blockade

TF gene expression in MN and in monocytic cell lines is mainly controlled by NF-kB/rel transcription factors [16]. To explore the role of NF-kB on PLT-induced TF activity in MN we tested the effect of BAY-117085, a reported inhibitor of NF-kB activation, in our model. BAY-117085 dose-dependently reduced procoagulant activity of mixed cell suspension incubated for 24 hours, reaching a maximum, 23.3±3.0% of the control, at concentration of 10 μMoles/L (Figure 7A). At this concentration the compound also abolished procoagulant activity of MN challenged with LPS for 4 hours (0.03 versus 32.7 arbitrary units in the presence or in the absence of BAY-117085, means of 2 experiments with cells from different donors). Activity of NF-kB/Rel proteins is regulated at different levels [17–19]. We hypothesized that GSK3 may regulate TF gene expression in MN interacting with activated PLT through NF-kB. In order to test this hypothesis we analyzed, by gel shift assay, accumulation of NF-kB in nuclear extracts obtained at different time points from MN incubated alone or in the presence of TRAP-6 activated PLT. As shown in Figure 7B, NF-kB increased in nuclear extracts of MN interacting with TRAP-6-activated platelets for 18 hours, respect to MN alone. At this time point GSK3 blockade by SB216763 further increased NF-kB accumulation.

Figure 7.

Role of NF-kB in TF expression in MN incubated with activated PLT. Upregulation by GSK-3 blockade.

A) MN, pretreated with DMSO or increasing concentrations of BAY-117085 were incubated with TRAP-activated PLT in mixed cell suspensions (ratio 1/200). After incubation cells were disrupted by 3 freeze-thaw cycles, and procoagulant activity measured by one stage clotting time.

B) MN pretreated with DMSO or SB216763 (20μM) were incubated alone with or without LPS, or combined with TRAP-activated PLT for 1 or 18 hours. At the end of incubation cells were lysed and nuclear extracts were incubated with radiolabeled oligonucleotide probes containing NF-kB binding site. Protein-DNA complexes were separated by 6% non denaturating acrylamide gels and visualized by autoradiography.

DISCUSSION

Using an in vitro cellular model that mimics MN-PLT interactions as these occur at the site of vascular injury, we demonstrate that GSK3 represents a crucial, gating element of the signaling pathway leading to TF gene expression in MN.

PLT-induced TF appeared after a prolonged time (6–10 hours) of interaction and steadily increased thereafter. The procoagulant activity was accompanied by accumulation of TF protein and of TF mRNA, and was reduced by actinomicin D, an inhibitor of transcription, demonstrating that MN are the major source of TF and that its expression was regulated at the transcriptional levels. This also indicates that in our model the role played by PLT-derived TF [20–22] is negligible.

In this model pharmacological blockade of GSK3 activity resulted in a significant up-regulation of TF gene and protein expression.

Like in other cell types [23] GSK3α and β isoforms are constitutively active in human MN in resting conditions and undergo phosphorylation on serine-21 and 9 respectively, in a PI-(3)K-dependent manner following cell stimulation. This results in functional inactivation of the enzyme [23]. In our experimental conditions western blot analysis, with GSK3-β-serine-9 phospho-specific antibodies revealed phosphorylation of the enzyme with a kinetics consistent with its role in the regulation of TF expression in MN. In fact, like TF activity, GSK3-β-serine-9 phosphorylation was detectable at time points between 6 and 10 hours in MN incubated in the presence but not in the absence of PLT.

Several protein kinases may phosphorylate and inactivate GSK3. Most notably, protein kinase B/Akt targets GSK3-α and β at the sites Ser-21 or 9 and it appears to be the predominant enzyme that performs this task in response to insulin [9]

Also in our model PI(3)K-Akt cascade appears to be involved in GSK3 phosphorylation and subsequent TF expression.

Western blot analysis showed that a PI(3)K-dependent Akt and GSK3 phosphorylation takes place during prolonged MN/PLT interaction. Treatment with wortmannin prevented TF expression at 24 hours and this effect was reversed by GSK3 blockade, thus supporting the notion that PI(3)K-Akt mediated phosphorylation, and down-regulation of GSK3 activity, plays a pivotal role in inducing TF expression. The role of PI(3)K-Akt cascade, was further strengthened by the effect of insulin on PLT-induced TF expression in MN. Insulin induced a dose-dependent increase of TF activity, that was almost abolished by PI(3)K blockade. In contrast PI(3)K blockade increased TF activity in MN alone activated by LPS. This observation, in agreement with previous studies [25,26], suggests a divergent role of PI(3)K in the regulation of TF expression in MN depending on whether the initial trigger is induced by activated PLT or by bacterial-derived LPS.

Recent studies in MN demonstrated that GSK3 negatively regulates anti-inflammatory cytokines and promote pro-inflammatory cytokine gene expression triggered by toll like receptors. In this context, GSK3 differentially modulates NF-kB subunit p65 and CREB interaction with the co-activator CBP [23]. Moreover, in monocyte-macrophages, GSK3 plays a key role in the regulation of immune and inflammatory response triggered by IFN-γ[27]. In human endothelial cells, pharmacological inhibition of GSK3 prevented TNF-α-induced NF-kB DNA binding activity and TF and VCAM gene expression [28]. In contrast, a more recent study, showed that over-expression of GSK3β or inhibition of GSK3β expression or function, reduced or increased, respectively, TNF-induced expression of IL-6, MCP-1 and VCAM-1 in microvascular endothelial cells [29]. Therefore, GSK3 may differentially regulate pro-inflammatory and pro-thrombotic gene expression depending on the cellular context and the nature of the initial stimulation.

The mechanisms and mediators by which activated PLT induce phosphorylation of GSK3 and trigger TF expression, in our model, were out of the focus of the present study. However, several studies in the literature already established that a complex array of molecular mechanisms operate in PLT-MN interaction and involve signaling via adhesive receptors acting in concert with cytokines and other factors [30]. In our model, the kinetics of GSK3 serine-9 phosphorylation and that of TF expression did not coincide with the process of cell-cell adhesion; in fact MN-PLT conjugates formed within few minutes of interaction and remained stable for up to 24 hours. Moreover cell free supernatants from PLT activated with TRAP-6 for 5 or 24 hours did not increase TF activity or TF protein, suggesting that the process leading to TF expression in MN adherent to activated PLT requires a complex interaction that cannot be reproduced by PLT released mediators or by signals directly triggered by engagement of adhesive receptors. This interpretation is in agreement with recent observation that de novo expression of COX-2 in MN interacting with activated PLT occurs via combinatorial regulation involving P-selectin-PSGL-1 interaction and signaling by newly synthesized interleukin (IL)-1β [31]. Detailed analysis of the kinetics of COX-2 mRNA accumulation demonstrated that P-selectin-PSGL-1 interaction rapidly induces NF-kB activation and COX-2 gene expression, however COX-2 mRNA accumulation and COX-2 protein synthesis could be achieved at later time points, only after the generation of IL-1β□□□□ promoted mRNA stabilization.

Interestingly the kinetics of COX-2 mRNA, protein and activity coincides with that of TF expression in our study, suggesting that a similar combined signaling may regulate PLT-induced TF expression in MN.

The novelty and importance of our data is the demonstration of a key role of GSK3 on TF expression in MN interacting with activated PLT, a potential mechanism linking increased thrombotic risk and metabolic disorders, particularly diabetes, in which GSK3 activity in circulating monocytes and in plaque resident macrophages may be reduced as consequence of the primary disease or of pharmacological treatments.

GSK3 has attracted much attention as an important mechanism of regulation of a wide variety of cellular functions relevant in metabolic and inflammatory disorders and in cancer [23]. The classical model of GSK-3-mediated regulation of down-stream targets foresees that GSK3, whose activity is high in resting conditions, phosphorylates and maintains in an inactive state a plethora of signaling proteins. Cell activation inhibits GSK-3 activity, thus releasing downstream mediators and signaling pathways [23]. One of the most important pathways involving GSK3 is that induced by insulin: via PI(3)K-Akt, insulin induces phosphorylation and inhibits GSK3, thus releasing the activity of glycogen synthase. In type-2 diabetes as well as in pre-diabetic metabolic disorders “insulin resistance” is a central feature characterized by reduced insulin ability to regulate glucose homeostasis, increased insulin secretion and increased insulin levels in circulating blood to compensate reduced activity on peripheral tissues. Metabolic disorders and “insulin resistance” are associated with high risk of cardiovascular disease [32–34]. Important for the discussion of our results, Vaidyula et al recently reported that exposure of healthy volunteers to 24 hours of combined hyperinsulinemia and hyperglycemia resulted in a significant increase of TF mRNA, protein and procoagulant activity in isolated MN [35]. Moreover increased levels of circulating TF protein and procoagulant activity were found in diabetic patients after 24 hours of hyperinsulinemia or of combined hyperinsulinemia and hyperglycemia [36] suggesting that altered metabolic homeostasis contributes to the procoagulant phenotype of MN that predisposes these patients to acute cardiovascular events. Our in vitro data further support this concept providing the cellular context and the molecular link that mediate TF expression in MN.

Insulin concentrations in peripheral blood of type-2 diabetic patients are extremely variable. However many clinical studies report that insulin concentration in fasting condition ranges around 100 pMoles/L, that may increase 2–5 times after oral glucose load. Thus, insulin concentration showing efficacy in our in vitro assay, are at least an order of magnitude higher than those found in the systemic circulation.

However it must be considered that MN might encounter higher insulin concentration in the regional pancreatic and portal circulation.

Prolonged exposure of circulating MN or resident macrophages to high insulin and glucose levels would result in a persistent down-regulation of GSK3 activity and consequent removal of a key, gating element in the regulation of TF expression. These “primed” MN exposed to activated PLT (another common feature of dismetabolic conditions) in the circulation or at the site of vascular injury, would express high amount of TF and increase the risk of thrombus formation. Therefore GSK-3 may represent the molecular link between metabolic disorders and increased risk of atherothrombosis.