Platelet Tissue Factor Synthesis in Type 2 Diabetic Patients Is Resistant to Inhibition by Insulin

We obtained human recombinant insulin (solubilized according to the recommendations of the manufacturer in 10 mmol/l acetic acid, 100 mmol/l NaCl, and 0.01% BSA to a stock concentration of 100 μmol/l); wortmannin; 1-(5-chloronaphthalene-1-sulfonyl)-1H-hexahydro-1,4-diazepine (ML-9); LPS (E. coli 0111:B4); indomethacin; puromycin; collagen types I and III; ADP from Sigma (St. Louis, MO); Akt inhibitor 1701-1 from Biovision (Mountain View, CA); iloprost from Schering (Berlin, Germany); forskolin, BAPTA(AM), ionophore A23187, Tg003, and cantharidin from Calbiochem (La Jolla, CA); LY294002 from Biomol (Plymouth Meeting, PA, USA); cycloheximide from MP Biochemicals (Santa Ana, CA, USA); collagen reagent Horm from Nycomed Pharma; a mixture of 95% type I and 5% type III collagen (Munich, Germany); thrombin and fibrinogen from Enzyme Research Laboratories (South Bend, IN); and U-44619 from Cayman Chemicals (Ann Arbor, MI). Plasma-derived VWF was purified from VWF/FVIII concentrate (Hemate P; Behring, Marburg, Germany) (18). Factor X (FX) was purified from fresh-frozen plasma (19). Recombinant FVIIa (rFVIIa) was from Novo Nordisk (Bagsværd, Denmark), and recombinant TF (Innovin) was from Dade Behring (Liederbach, Germany). Antibodies against the horseradish peroxidase–labeled anti-rabbit antibody were from Cell Signaling Technology (Danvers, MA), anti–phospho-Ser polyclonal antibody was from Upstate Biotechnology (Bucks, U.K.), peroxidase-linked goat anti-mouse antibody was from DAKO (Glostrup, Denmark), anti–splicing factor 2 (SF2)/–alternative splicing factor (ASF) antibody was from Abcam (Cambridge, U.K.), anti-TF antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA) for immunoprecipitation and Affinity Biologicals (Ancaster, ON, Canada) for Western blotting, and anti-TF pathway inhibitor (TFPI) antibody from American Diagnostica (Stamford, U.K.) An inhibitory antibody against TF was a generous gift from Dr. M. Kjalke (Hemostasis Biology, Novo Nordisk, Malov, Denmark). The ADP receptor P2Y12 antagonist, N6-(2-methylthioethyl)-2-(3,3,3-trifluoropropylthio)-β,γ-dichloromethylene ATP (AR-C69931MX), was a kind gift from Astra Zeneca (Loughborough, U.K.). All other chemicals used were of analytical grade.

Isolated platelet suspensions were made monocyte free and adhered to different surface-coated adhesive proteins. Splicing of TF premRNA and synthesis of TF protein and TF procoagulant activity were measured. Pharmacologic interference with extra- and intracellular signaling was investigated, focusing specifically on the role of insulin in healthy individuals and patients with type 2 diabetes.

The study had been approved by the medical ethical review board. Type 2 diabetic subjects were recruited from the outpatient clinics of the Utrecht Medical University Hospital (Utrecht, the Netherlands). All patients were anti-GAD negative and had plasma C-peptide levels >0.30 nmol/l. We recruited insulin-using patients in order to circumvent the possibility that oral hypoglycemic agents interfere with the study; only metformin use was acceptable (it was stopped the evening for the study). Six patients used four injections with insulin per day, one used continuous subcutaneous insulin infusion. Their further characteristics are given in Table 1. Patients and matched control subjects gave their informed consent prior to participation in the study.

Freshly drawn venous blood from healthy volunteers and type 2 diabetic patients was collected into 0.1 volume of 130 mmol/l Na₃ citrate. Platelets were isolated as described (11).

Attempts to remove contaminating monocytes included 1) incubation of a platelet suspension (2 × 10⁸ platelets/ml) with either anti-CD14 antibody– or anti-CD45 antibody–coupled beads (15 min, 22°C, 35 μl anti-CD14 or anti-CD45 beads per 40 × 10⁸ platelets) and subsequent magnetic separation (Miltenyi Biotec, Bergisch Gladbach, Germany) and 2) a double incubation of a concentrated platelet suspension (∼15 × 10⁸ platelets/ml) with a cocktail of anti-CD14 antibody– and anti-CD45 antibody–coupled beads (15 min, 22°C, 35 μl anti-CD14 and 35 μl anti-CD45 beads per 40 × 10⁸ platelets) and magnetic separation. Contamination with monocytes was measured by RT-PCR of CD14 mRNA. To this end, total RNA was isolated using the kit from RNA-Bee (Tell Tech, Friendswood, TX), and RT-PCR was performed using the forward primer 5′-AAAGCACTTCCAGAGCCTGC-3′ and the reverse primer 5′-TC GAGCGTCAGTTCCTTGAGG-3′ on a PTC-200 apparatus (MJ Research, Waltham, MA).

To investigate TF synthesis in platelets, an aliquot of suspension (2 × 10⁸ platelets) was incubated for static adhesion experiments in six-well plates (Corning ultra-low attachment surface; Corning, Acton, MA) or kept in suspension for the indicated times at 37°C. Wells were coated with 50 μg/ml horm collagen, 50 μg/ml collagen type I, 50 μg/ml collagen type III, 1 unit/ml thrombin, 100 μg/ml fibrinogen, and 10 μg/ml VWF for 2 h at 37°C. For analysis of TF procoagulant activity, adhered platelets, collected in 250 μl HEPES/Tyrode buffer, or suspensions were lysed by freeze/thawing three times and centrifuged (16200g, 2 min, 22°C). Pellet fractions were dissolved in 50 μl HEPES/Tyrode buffer. Samples were incubated with 2 μg/ml anti-TFPI antibody (15 min, 22°C) and thereafter with 10 μg/ml FX, 5 units/ml rFVIIa, and 5 mmol/l CaCl₂ (45 min, 37°C). FXa generation was measured in a fluorescence reader at 405 nm after addition of FXa substrate (Pentapharm, Basel, Switzerland). Procoagulant activity was determined based on a standard curve derived from serial dilutions of recombinant TF.

For analysis of splicing of TF premRNA, 12 × 10⁸ platelets were adhered to a coated surface or kept in suspension, as indicated. Total RNA was isolated (RNA-Bee; Tell Tech), and RNA levels adjusted for variations per sample (by measuring the optical density OD260 on a NanoDrop spectrophotometer), and RT-PCR was performed on a PCR apparatus (PTC-200; MJ Research) using primers that overlapped intron 4, with forward primer 5′-CTCGGACAGCCAACAATTCAG-3′ and reverse primer 5′-CGGGCTGTCTGTACTCTTCC-3′.

The number of adhered platelets was inferred from the content of alkaline phosphatase in collected samples, as published (20). Alkaline phosphatase activity was related to platelet number using a serial dilution of platelets in suspension.

To determine levels of TF protein and the Ser phosphorylation of SF2, aliquots of 2 × 10⁸ platelets were adhered to a coated surface in six-well plates or kept in suspension (1 h, 37°C). Adhered platelets and suspensions were collected in lysis buffer (1% NP40, 0.5% octylglucoside, 0.1% SDS, and 5 mmol/l EDTA in PBS) supplemented with 10% protease inhibitor cocktail, 1 mmol/l NaVO₃, and 10 μmol/l cantharidin. Proteins were precipitated with protein G-Sepharose in combination with 1 μg/ml anti-TF or anti-SF2 antibody overnight at 4°C. Precipitates were washed three times and dissolved in reducing Laemmli sample buffer.

Proteins were analyzed by SDS-PAGE and Western blotting. After blocking with Odyssey buffer (1 h, 22°C), membranes were incubated (16 h, 4°C) with 1 μg/ml anti-TF, anti–phospho-Ser, or anti-SF2 antibody. Immunoblots were visualized using the Odyssey infrared imaging system (LI-COR Biosciences, Lincoln, NE) and Alexa-labeled antibodies according to the manufacturer's instructions. The intensity of the bands was quantified with ImageJ software.

Data are expressed as means ± SE, with n observations, and was analyzed with Student t test for unpaired observations or Mann-Whitney U test, as indicated. Differences were considered significant at P < 0.05.

Since monocytes have a much higher capacity to synthesize TF than platelets, it was essential to prepare pure platelet suspensions. Platelet preparations prepared by conventional centrifugation/resuspension (7,11) without and with treatment with anti-CD45 magnetic beads were positive for the monocyte marker CD14 detected by RT-PCR (Fig. 1,A, top panel). Pure preparations were obtained by a double incubation of concentrated suspensions with anti-CD14/CD45 magnetic beads and monocyte removal by magnetic sorting. After 35 cycli, RT-PCR of CD14 was negative in the platelet preparation and positive in the eluate (Fig. 1,A, bottom panel). Upon stimulation with LPS, CD14-mRNA–containing suspensions adhering to fibrinogen synthesized TF activity detected in the FXa assay, but CD14-mRNA–negative preparations did not (Fig. 1 B). To ensure purity of platelet suspensions, a sample of adhered platelets with LPS was included in all experiments, and series with LPS-induced TF synthesis were discarded. Addition of anti-TF antibody (0.5 mg/ml, 15 min, 22°C) completely blocked TF activity in activated platelets confirming specificity of the FXa assay.

TF synthesis and TF premRNA splicing were studied on fibrinogen- and collagen-coated surfaces. Platelets adhering to a fibrinogen-coated surface (4 h, 37°C) synthesized 0.24 ± 0.04 ng TF/2 × 10⁸ platelets (n = 10). TF synthesis in adhered platelets was inhibited by the translation blockers cycloheximide (58 ± 2%, n = 3) and puromycin (76 ± 6%, n = 12) and the splicing blocker Tg003 (51 ± 7%, n = 6) (Fig. 1,C), confirming earlier work (17,21). To demonstrate splicing of TF premRNA, an RT-PCR with primers overlapping intron 4 was performed on platelets adhered to collagen type I (2 h, 37°C). The result showed the expected band for spliced mRNA at 297 bp and for unspliced mRNA at 904 bp in the presence of Tg003 (Fig. 1,D). Regulation of TF premRNA splicing was examined by measuring the Ser phosphorylation of SF2, an effector of Cdc2-like kinase (Clk-1) (Fig. 1 E). The low levels of SF2-Ser phosphorylation in platelets in suspension were unchanged after stimulation with thrombin and increased when platelets had adhered to fibrinogen (4 h, 37°C). Tg003 reduced the phosphorylation, confirming its dependence on Clk-1. These results confirm that adhering platelets synthesize TF through splicing of premRNA.

The question was addressed regarding which adhesive surface induced the highest TF premRNA splicing. Puromycin was used to separate translation-dependent TF activity from background TF. TF synthesis on horm collagen, collagen I, and thrombin was 0.34 ± 0.10, 0.32 ± 0.07, and 0.30 ± 0.05%, respectively. A lower TF activity was found on fibrinogen (0.24 ± 0.04), collagen III (0.17 ± 0.02%), and VWF (0.09 ± 0.03%) (n = 4) (Fig. 2,A). Since different adhesive surfaces induced different extents of adhesion/aggregation, bound platelets were quantified on the basis of alkaline phosphatase content. The highest adhesion/aggregation was found with collagen III, collagen I, and horm collagen (∼125 × 10⁶ platelets), whereas adhesion to thrombin, fibrinogen, and VWF was ∼30% lower (Fig. 2,B). Expression per platelet and correction for background TF (defined as ΔTF activity), indicated that horm collagen, thrombin, and collagen I induced the highest TF synthesis followed by fibrinogen, collagen III, and VWF (Fig. 2,C). Further experiments were based on collagen III, because it induced the highest platelet adhesion and is present in the vasculature. We examined whether adhered platelets produce more TF than platelets in suspension. Collagen III–adhered platelets (4 h, 37°C) produced 600-fold more TF than platelets in suspension stimulated with collagen III (0.17 ± 0.09 and 0.0003 ± 0.0001 ng TF/2.0 × 10⁸ platelets, respectively, n = 3), indicating that splicing of TF premRNA is strongly facilitated by adherence to a surface (Fig. 2 D), in contrast with an earlier finding (17,21).

To examine the mechanisms that couple surface activation to TF synthesis, platelets were adhered to collagen III without and with coincubation with AR-C69931 MX (which mimics blockade of the P2Y12 receptor by the active metabolite of clopidogrel), with indomethacin, an easily soluble COX-1 inhibitor which mimics the action of aspirin, or a combination. The inhibitors did not change platelet adhesion during prolonged contact with collagen III (Fig. 3,A, insert). AR-C69931 induced a 56 ± 16% decrease in TF synthesis, indicating that in its absence TF synthesis is enhanced by ADP secretion and P2Y12 receptor–mediated signaling. Indomethacin induced an even steeper reduction and inhibited TF synthesis by 82 ± 9%. The combination of the two inhibitors reduced TF synthesis by 96 ± 2% (n = 4, Fig. 3 A). Thus, platelet TF synthesis strongly depends on the extracellular feedback loops mediated by secretion of granule ADP and release of thromboxane A₂ (TxA₂).

There was little effect of metabolic inhibitors on adhesion (not shown). The adenylyl cyclase activator forskolin and the stable prostacyclin mimetic iloprost raise cAMP and inhibited TF synthesis on collagen III by 51 ± 14 and 82 ± 6%, respectively. BAPTA/AM, which lowers the cytosolic Ca²⁺ concentration and quenches Ca²⁺ increases in activated platelets, inhibited TF synthesis almost completely. Incubation with ionophore A23187 alone did not change collagen III–induced TF synthesis, but coincubation with 1 mmol/l Ca²⁺ increased TF synthesis to 443 ± 57%. Inhibitors of PI3-K (wortmannin, LY294002) and PKB (Akt inhibitor 1701-1, ML-9) decreased TF synthesis by ≥60%, indicating that in the absence of inhibitors, the PI3-K/PKB pathway signals to TF premRNA splicing (Fig. 3,B). To investigate whether the inhibitors interfere with release of ADP/TxA₂ and/or with signaling from ADP/TxA₂ receptors to the spliceosome, experiments were repeated in the presence of excess of ADP (50 μmol/l) and of the stable TxA₂ analog U-46619 (2 μmol/l). Inhibition of TF synthesis by the metabolic inhibitors was preserved, indicating that in addition to the known suppression of ADP/TxA₂ release, these blockers interfere with the control of TF premRNA splicing (Fig. 3,C). Western blot analysis confirmed that iloprost inhibited synthesis of TF protein (Fig. 3 D). Together, these data reveal that surface-induced TF synthesis is inhibited by a rise in cAMP strongly depends on cytosolic Ca²⁺ and is induced by signaling through the PI3-K/PKB pathway.

Since surface-induced TF synthesis was sensitive to agents that raise cAMP or inhibit the PI3-K/PKB pathway, we addressed the question whether insulin interferes with adhesion-induced TF synthesis. Platelets were preincubated with 100 nmol/l insulin (5 min, 37°C) and adhered to different adhesive surfaces. The degree of inhibition differed between adhesive proteins and reached 94 ± 6, 64 ± 27, 76 ± 27, 67 ± 14, and 41 ± 7% (n = 5) for horm collagen, thrombin, collagen I, fibrinogen, and collagen III, respectively. On VWF, no inhibition by insulin was found (Fig. 4,A). To elucidate whether insulin interferes with splicing before or after release of ADP/TxA₂, experiments were repeated in the presence of excess of ADP and TxA₂ mimetic. Inhibition of collagen III–induced TF synthesis was preserved, indicating that insulin inhibits TF synthesis by interfering with the control of premRNA splicing induced by ADP and TxA₂ receptors (Fig. 4,B). Collagen III–adhered platelets were used to further evaluate the inhibition by insulin. Adhered platelets showed a dose-dependent inhibition between 1 and 100 nmol/l insulin (Fig. 4,C). Inhibition of premRNA splicing by insulin was clearly detectable, showing decreased levels of mature mRNA compared with platelets in the absence of insulin (Fig. 4,D). In type 2 diabetic patients, collagen III–induced TF activity inhibition by insulin was lost, resulting in higher levels of collagen III–induced TF activity (96 ± 12%, n = 8) compared with matched control subjects (59 ± 12%, n = 8; P < 0.05 with Mann-Whitney U test; Fig. 4 E).

Novel findings in the present study are 1) the identification of the extra- and intracellular mechanisms that couple surface activation to initiation of TF synthesis in platelets, 2) the inhibition of TF synthesis by insulin in normal platelets, and 3) the loss of insulin inhibition in platelets from patients with type 2 diabetes resulting in a 1.6-fold–higher TF synthesis. The increased TF synthesis might be a cause of the hypercoagulant state observed in type 2 diabetic patients and contribute to the increase in thrombotic risk.

The presence of TF synthesis in platelets is controversial. Some investigators attributed a positive outcome to contamination with monocytes, which have a many-fold–higher translation capacity (22,–24). Therefore, we included a purity check in every experiment to demonstrate unequivocally that platelets synthesize TF. The CD14 mRNA–free platelet suspensions showed no TF premRNA splicing upon stimulation with LPS, which is a potent inducer of TF synthesis in monocytes (16). Apparently, splicing of TF premRNA is not under control of the platelet LPS receptor, which contrasts with other studies that show LPS-induced splicing of interleukin (IL)-1β and cyclooxygenase (COX)-2 premRNA in platelets (25).

Platelets adhering to surface-coated adhesive proteins start splicing of TF premRNA and synthesis of TF protein and induce the capacity to initiate coagulation, confirming earlier observations (17,21). The anucleate platelet has long been considered incapable of synthesizing proteins. Intron removal was thought to be confined to the nucleated megakaryocyte and platelet proteins to be the exclusive result of transcription and translation in this progenitor cell (26). Unexpectedly, the platelet cytoplasm SF2/ASF, a regulator of constitutive and alternative splicing (27) together with a number of premRNAs with message for IL-1β, B-cell lymphoma 3, plasminogen activator inhibitor-1, COX-2, and TF (28,–30). Platelets stimulated with soluble agonists or adhering to fibrinogen start formation of mature RNAs and translation into functional proteins. Splicing of TF premRNA results in formation of TF protein and the capacity to start coagulation and is under control of dual specificity Clk-1, which contains an NH₂-terminal region enriched in serine/arginine dipeptides (SR) that interacts with SF2/ASF (31). Inhibition of splicing by Tg003 and of protein synthesis by puromycin confirm that platelets are capable of transcription/translation-dependent TF synthesis (17). However, the inhibition was incomplete, suggesting that platelets contain a small store of encrypted TF that becomes active when the cells adhere to an adhesive surface.

Splicing of TF premRNA was 600-fold more efficient in adhering platelets than in platelet suspensions, suggesting that close contact between platelets accelerates this process. In adhered platelets, the P2Y12 receptor blocker AR-C69931 MX inhibited TF synthesis by 56% and the COX-1 blocker indomethacin by 82%, illustrating that without inhibitors, TF synthesis is strongly enhanced by ADP secretion and TxA₂ formation and subsequent activation of the P2Y12 receptor (ADP) and TPα/β receptors (TxA₂). The combination of the blockers reduced TF synthesis almost by 100%. This observation might indicate that in vivo signaling to TF premRNA splicing is inhibited by the active clopidogrel metabolite as well as aspirin and that the combination abolishes the major part of platelet TF production. Apparently, there is little TF synthesis in the absence of feedback activation by released ADP and TxA₂, indicating that surface receptors that trigger TF premRNA splicing predominantly do so by inducing these positive feedback pathways. Collagen I, horm collagen, and thrombin had the strongest capacity to induce splicing of TF premRNA. Fibrinogen and collagen III induced less splicing, and VWF was the weakest activator. These differences might reflect the contribution of ADP/TxA₂ release in platelet activation by these agonists, which is strong in horm collagen (32) and thrombin (33) activation, weak in activation by surface-coated fibrinogen (34), and virtually absent in platelet activation by coated VWF (35).

Iloprost, BAPTA-AM, and inhibitors of PI3-K/PKB inhibited surface-induced TF synthesis, indicating that negative control through cAMP-dependent mechanisms and positive control through the PI3-K/PKB route determine the rate of TF synthesis. In addition to a direct effect of these inhibitors on the release of ADP/TxA₂, inhibition of TF activity also takes place after the released products start further signaling through the P2Y12 and TxA₂ receptors. Protein kinase A (PKA) α and PKAβ are activated by an increase in cAMP concentration induced by forskolin or 8-CPT-cAMP and phosphorylate SR proteins in vitro and change splice site selection in vivo (36). PKB contributes to phosphorylation of SR proteins, specifically SF2 (37), which potentiate their binding to recently transcribed mRNA and/or directly stimulate SR protein activities. Interestingly, in monocytes, TF expression is under negative control by the PKB pathway (16,38). Here, PKB inhibits nuclear factor-κB activity (38), such in contrast to anucleated platelets where PKB regulates the splicing mechanism.

The induction of ADP/TxA₂ release followed by signaling from ADP/TxA₂ receptors to the spliceosome is under control of insulin. In normal platelets, insulin interferes with the P2Y12 pathway by inducing the association of IRS-1 with G_iα₂, thereby blocking G_i-mediated cAMP suppression, which attenuates platelet functions. In addition, insulin stimulates the PI3-K/PKB pathway, initiating glucose uptake through upregulation of GLUT3 transporters (7,13). Interference with cAMP suppression by insulin impairs agonist-induced Ca²⁺ rises, aggregation, secretion, and the generation of a procoagulant surface, and the interference with TF synthesis presented here adds to the list of platelet functions under negative control by insulin.

Platelet TF does not seem to play a role in acute injury, since it takes long to express active TF. An animal model of laser-induced vessel-wall injury (14) shows the first appearance of TF at ∼100 s after induction of vessel wall damage, which is well before the first sign of platelet TF synthesis. Diabetes increases the risk of developing lower-extremity arterial disease, which is one of the first signs of a generalized atherosclerotic disease (39). A key event in the pathogenesis of type 2 diabetes is loss of atherosclerotic plaque stability. When the fibrous cap is disrupted, a platelet plug forms, which continuously increases in size until the lumen is completely obstructed. In the chronic process of arterial occlusive disease, the slow but consistent production of TF by platelets may well contribute to the propagation and stabilization of a thrombus. Indeed, previous studies have detected TF antigen and activity in the atherosclerotic plaque (40). The elevated TF synthesis in platelets form type 2 diabetic patients might result in a more stable clot inducing occlusion of an artery at sites where atherosclerotic lesions have developed.

This work was supported by the Dutch Diabetes Research Foundation (no. 2004.00.029) and the Netherlands Thrombosis Foundation (2006-1).

No potential conflicts of interest relevant to this article were reported.

The authors acknowledge the cooperation of the patients and control subjects.