|Home | About | Journals | Submit | Contact Us | Français|
In addition to hemostasis, platelets mediate inflammation and clearance of bacteria from the bloodstream. As with platelet-platelet interactions, platelet-bacteria interactions involve cytoskeletal rearrangements and release of granular content. Stimulation of the immune Toll-like receptor 2 (TLR2) on the platelet surface, activates phosphoinositide-3-kinase (PI3K) and causes platelet activation and platelet-dependent thrombosis. It remains unknown if platelet activation by immune vs. thrombotic pathways leads to the differential regulation of signal transduction, protein-protein interactions, and α-granule release, and the physiological relevance of these potential differences. We investigated these processes after immune vs. thrombotic platelet stimulation. We examined selected signaling pathways and found that phosphorylation kinetics of Akt, ERK1/2 and p38 differed dramatically between agonists. Next, we investigated platelet protein-protein interactions by mass spectrometry (MS)-based proteomics specifically targeting cytosolic factor XIIIa (FXIIIa) because of its function as a cytoskeleton-crosslinking protein whose binding partners have limited characterization. Four FXIIIa-binding proteins were identified, two of which are novel interactions: FXIIIa-focal adhesion kinase (FAK) and FXIIIagelsolin. The binding of FAK to FXIIIa was found to be altered differentially by immune vs. thrombotic stimulation. Lastly, we studied the effect of thrombin- vs. Pam3CSK4-stimulation on α-granule release and observed differential release patterns for selected granule proteins and decreased fibrin clot formation compared with thrombin. The inhibition of PI3K caused a decrease in protein release after Pam3CSK4- but not after thrombin-stimulation. In summary, stimulation of plates by either thrombotic or immune receptors leads to markedly different signaling responses and granular protein release consistent with differential contribution to coagulation and thrombosis.
Cells of the innate immune system distinguish between pathogen and self by utilizing signals from Toll-like receptors (TLRs). Stimulation of TLRs in vascular cells and monocytes by pathogen-associated molecules results in activation of nuclear factor κB (NF-κB) and production of proinflammatory cytokines (1, 2). TLRs are also expressed on the platelet surface (3, 4). Although platelets are present at sites of inflammation, promote bacterial clearance (5), and cross-talk with other immune cells, their specific role in innate immunity is not well understood.
TLR2, in particular, recognizes bacterial lipopeptides, lipoproteins, and peptidoglycans from gram-positive bacteria (6). The lipopeptide Pam3CSK4, a TLR2/1 agonist, has been shown to induce physiological platelet activation and serotonin release (7). Recent data from our laboratory demonstrate that Pam3CSK4-stimulation of TLR2 leads to P-selectin expression, activation of integrin αIIbβ3, and activation of PI3K/Akt signaling pathway (8).
Thrombin, on the other hand, has been studied extensively in platelets. As one of the major platelet agonists, it activates the protease activated receptor 1 and 4 (PAR1/4) on the platelet surface which leads to activation of PI3K and phospholipase Cβ (PLCβ) pathways. The activation of these pathways leads to an increase in intracellular calcium concentrations, platelet shape change, granule secretion, and platelet aggregation (9). However, many details of these signaling cascades, and the mechanisms of their downstream events remain to be elucidated.
The thrombotic vs. immune stimulation of platelets and their subsequent differential effects on the activation of signaling pathways, on specific protein-protein interactions, and on α-granule release have not been examined. To differentiate these forms of platelet stimulation and define their mechanism of action, we compared platelet stimulation with thrombin vs. Pam3CSK4 and investigated differential changes in (1) the phosphorylation kinetics of Akt, ERK1/2 and p38, (2) FXIIIa-interacting proteins, and (3) the release of several α-granule proteins, as well as the effect of PI3K-inhibition on FXIIIa-interactions and α-granule release.
Platelets were isolated from healthy volunteers and platelet aggregation was monitored using a PAP-4 platelet aggregometer (Bio/Data Corporation, Horsham, PA) as described previously (10). Briefly, human α-thrombin (0.5 U/mL; Enzyme Research Laboratories, South Bend, IN), Pam3CSK4 (10 μg/mL; Invivogen, San Diego, CA), or ADP (10 μM; Chrono-log Corporation, Havertown, PA) were added, while stirring, to 2×108/mL platelets supplemented with 1 mg/ml fibrinogen, 1 mM CaCl2, and 2 mM MgCl2 for 12 min, 37°C. The peptide Gly-Pro-Arg-Pro (GPRP; Sigma-Aldrich, St. Louis, MO) was added to thrombin-activated samples to inhibit fibrinogen polymerization (11).
Washed platelets were incubated with 10 μM calcein-AM (Invitrogen, Carlsbad, CA) (30 min), then resuspended in HBSS. Subsequently, platelets were stimulated with 0.5 U/mL thrombin, 10 μg/mL Pam3CSK4, 10 μM ADP, or left untreated (10 min). Using a vacuum-based cell adhesion flow chamber (Immunetics, Inc., Boston, MA), platelets were passed at a constant rate over collagen-coated cover slips for 20 min. Cover slips were mounted on glass slides for analysis with a fluorescence microscope (Nikon, Melville, NY; 20X magnification, 505 nm emission).
Platelets (2x108/mL) in Tyrodes-buffer were treated with 0.5 U/mL thrombin, 10 μg/mL Pam3CSK4, or 10 μM ADP for 15 min. To resting platelets, prostaglandin E1 (PGE1) (1 μM; Cayman Chemical, Ann Arbor, MI) and apyrase (0.2 U/mL; Sigma-Aldrich) were added. Activation was stopped by adding PGE1/apyrase and pelleting the platelets. Samples were washed with Tyrodes-buffer containing PGE1/ apyrase, and lysed in equal volumes of Tyrodes-buffer and 2x lysis-buffer (2% NP-40, 30 mM Hepes, 150 mM NaCl, 2 mM EDTA, pH 7.4). To remove platelet-derived microvesicles, releasates were centrifuged at 150,000 × g, 90 min, 4°C (12), and tested for the absence of microvesicles by immunoblotting for integrin αIIb (13). Protein concentrations were determined using the DC-protein assay (BioRad, Hercules, CA).
For kinetics studies, resting platelets and platelets activated for indicated times with thrombin (0.5 U/mL) or Pam3CSK4 (10 μg/mL), were pelleted, washed and lysed. For PI3K-inhibition, platelets were incubated with 25 μM or 50 μM LY294002 (Invivogen) (30 min) prior to platelet activation for 15 min. Lysate and releasate samples were obtained as described above.
Sheep polyclonal anti-FXIIIa (Cedarlane Laboratories, Hornby, ON, Canada) and IgG control antibodies (Santa Cruz Biotechnology, Santa Cruz, CA), or mouse monoclonal anti-thrombospondin and IgG1 control antibodies (LabVision Corporation, Fremont, CA), respectively, were used. After preclearing with protein G+ beads (Santa Cruz Biotechnology), lysates were incubated with either the target-protein-specific or control antibodies. Subsequently, antibodies were precipitated with protein G+ beads and washed with lysis buffer. For 2D gel electrophoresis (2D-GE), precipitates were washed with lysis buffer, followed by 1 mM Hepes/NaCl (pH 7.6), and 1 mM Hepes (pH 7.6) to remove residual salts/detergents.
For 1D-GE, samples were loaded onto 10% Criterion-XT Bis-Tris gels (BioRad). After electrophoresis, gels were either stained with Coomassie Blue Imperial Stain (Pierce, Rockford, IL) or prepared for immunoblotting. For 2D-GE of releasates, samples were prepared using a 2D clean-up kit (GE Healthcare, Piscataway, NJ). Samples (75-125 μg protein) were solubilized in 2D Rehydration/Sample Buffer (BioRad) containing 50 mM dithiothreitol (DTT). For 2D-GE of immunoprecipitates, bead pellets were directly solubilized in the same manner as releasate samples and pelleted. Bio-Lyte 3/10 ampholytes (BioRad) were added to all samples. Immobilized pH gradient (IPG) strips (11 cm; pH 4-7 (for releasates), pH 3-10 (for immunoprecipitates; BioRad) were actively rehydrated with samples overnight in a PROTEAN™ IEF cell (BioRad) followed by isoelectric focusing in the first dimension. Afterwards, proteins in the IPG strips were reduced with DTT and alkylated with iodoacetamide using Equilibrium Buffer I/II (BioRad). IPG strips were placed onto 10% Criterion-XT gels and subjected to separation in the second dimension. Gels were either stained with Coomassie Blue or silver stain (GE Healthcare), or prepared for immunoblotting.
Proteins transferred onto PVDF membranes (Millipore, Billerica, MA) were probed for the following: anti-FXIIIa, anti-thrombospondin (LabVision); anti-HSP27, anti-fibrinogen β, anti-gelsolin, anti-FAK, anti-platelet basic protein, anti-platelet factor 4, anti-integrin αIIb (Santa Cruz Biotechnology), anti-myosin IIA (Covance Research Products, Berkeley, CA), anti-phospho-Akt(Ser473), anti-Akt, anti-phospho-ERK1/2(Thr202/Tyr204), anti-ERK1/2, anti-phospho-p38(Thr180/Tyr182); and anti-p38 (Cell Signaling Technology, Danvers, MA). Protein bands were quantified by densitometry (VersaDoc Model 3000 Imaging System; Quantity One software version 4.4.1; BioRad).
Gel bands/spots were excised, destained, washed, and dried. Proteins in 1D gel pieces were reduced with DTT, then alkylated with iodoacetamide. Proteins from 2D gels were reduced/alkylated prior to loading of the IPG strip onto the 2D gel. In-gel digestion was performed with Trypsin Gold (Promega, Madison, WI), overnight, 37°C in 50 mM NH4HCO3, 5% acetonitrile (ACN). Tryptic peptides were extracted with 1% trifluoroacetic acid in 50% ACN, desalted with ZipTipC18 pipette tips, and eluted into 0.1% TFA, 50% ACN.
After co-crystallization of purified proteolytic peptides with matrix (2,5-dihydroxybenzoic acid (Bruker Daltonics, Billerica, MA; 2 mg/ml in 50% ACN, 0.1% TFA)) onto an AnchorChip™ target (Bruker Daltonics), mass spectra were acquired on a Reflex IV MALDI-TOF instrument (Bruker Daltonics) operated in positive-ion reflectron mode, calibrated with Bruker peptide standards. MALDI-TOF mass spectra were analyzed with MoverZ™ software (Proteometrics, LLC, New York, NY). The resulting peak lists were submitted for peptide mass fingerprint (PMF) analysis to Mascot™ (Matrix Science Inc., Boston, MA) against the SwissProt or NCBI non-redundant protein databases, using the following restrictions: human taxonomy; up to 2 missed cleavages allowed with trypsin as the digestion enzyme; cysteine carbamidomethylation as fixed and methionine oxidation as variable modifications; peptide mass error tolerance of ± 0.1-0.3 Da. Under these conditions, a probability-based Mascot™ Mowse score greater than 60 indicated a p-value less than 0.05 and was considered to represent a significant identification.
Activated platelets were placed onto 0.01% poly-L-lysine-coated slides and fixed in methanol:acetone (70:30). Slides were rehydrated, treated with 0.3% H2O2 in methanol, then permeabilized (0.2% Triton X-100). After blocking in Protein block serum-free solution (DakoCytomation, Carpinteria, CA), slides were incubated with antibodies to FXIIIa and FXIIIa-interacting protein of interest (overnight, 4°C). Primary antibodies were the same as those used for immunoblotting, except anti-thrombospondin (Ab-8, LabVision). Slides were incubated with secondary FITC- and TexasRed-conjugated antibodies. Control slides included labeling with non-specific IgGs. Slides were examined by confocal microscopy (Leica TCS SP2 instrument; Leica Microsystems, Exton, PA) using a compensation system to remove spectral overlap (FITC: 488 nm; Texas Red: 595 nm) and Leica Confocal Software (Leica Microsystems). Images were taken at 100X magnification.
Isolated platelets were resuspended in Tyrodes-buffer at a final concentration of 3×108/mL, with 1 mM CaCl2. Platelets were either stimulated with 0.5 U/mL thrombin or 10 μg/mL Pam3CSK4 ± 1 mg/mL fibrinogen, for 1h. Clots that formed were removed and the volume of buffer remaining was determined. Data is presented as the percent of buffer remaining.
Whole blood (600 μL) was added to 200 μL of 1X PBS and stirred continuously for 10 minutes at 37°C. Samples were left untreated or treated with either 0.5 U/mL thrombin or 10 μL Pam3CSK4. Samples were then incubated with anti-CD41 FITC conjugated antibodies (platelet marker; eBioscience, San Diego, CA) and anti-CD14 PE-Cy7 conjugated antibodies (monocyte marker; eBioscience). Red blood cells (RBC) were then lysed using 1X BD FACS™ Lysing Solution (BD Biosciences, San Jose, CA). Samples were then processed using FACScan Flow Cytometer with CellQuest Pro Software (BD Biosciences). Monocytes were gated for using a combination of light scattering and CD14 PE-Cy7 fluorescence. A large gate was placed around this population to measure an increase in forward light scattering, indicative of heterotypic aggregate formation. 2000 monocytes were acquired. Further discrimination was done to quantitate the percent of platelet-positive monocytes (CD14-positive and CD41-positive).
After treatment with either 0.5 U/mL thrombin, 10 μg/mL Pam3CSK4, or resting for 10 min, a mixture of platelets, red blood cells, and white cells (isolated through 1-Step ® Polymorphs Method) was fixed with 2.5% glutaraldehyde in 0.1 M PBS (pH 7.4) for 1 h. Fixed cells were washed in buffer solution and post-fixed with 1.0% osmium tetroxide (OsO4) in 0.1 M PBS (pH 7.4) for 2 h. Samples were rinsed with distilled water (3×5min), dehydrated serially in 30%, 50%, 70%, 90%, 100% ethanol, then agitated in acetone. After critical point drying, cells were mounted on specimen stub, coated with gold alloy, and examined with a JEOL (JSM-6100) scanning electron microscope (magnification 15,000X).
Data are expressed as mean ± standard error. Differences between two groups were assessed by Student's t-test with p < 0.05 being statistically significant.
In this study, we compared platelet responses to thrombin and Pam3CSK4 to examine the functional effects of thrombotic vs. immune stimulation. To test the functionality of both agonists, we performed aggregation studies with washed platelets. After 12 minutes, platelet aggregation had progressed to 98 ± 3% for thrombin, 69 ± 5% for Pam3CSK4, and 5 ± 2% for ADP. In Figure 1A, representative platelet aggregation curves for each agonist are shown. The amount of thrombin used (0.5 U/mL/4.43 nM) was below physiological levels (50 nM) (14). ADP (10 μM) was also below its physiologic range (0.05 to > 1 mM) (15). The concentration of Pam3CSK4 used was based on previous work (8).
To further investigate the functionality of thrombin vs. Pam3CSK4, we measured platelet adhesion to collagen-coated slides under flow conditions (Figure 1B). Stimulation with thrombin (Figure 1Bi) or Pam3CSK4 (Figure 1Bii) resulted in significant platelet adhesion, whereas stimulation with ADP (Figure 1Biii) induced less platelet adhesion. Unactivated platelets did not adhere to the collagen surface (Figure 1Biv). Taken together, Pam3CSK4 appears to activate platelets significantly more than the weak agonist, ADP.
To elucidate whether thrombin and Pam3CSK4 activate signaling pathways differentially downstream of their respective receptors, we examined the phosphorylation kinetics of three major signaling proteins. Previous data from our group showed that Pam3CSK4-stimulation of TLR2 activates the PI3K/Akt signaling pathway (8). Here, we measured the phosphorylation kinetics of Akt(Ser473) as a downstream target of PI3K after platelet activation with thrombin or Pam3CSK4 for up to 15 minutes (Figure 2A). Interestingly, at early time points of activation, each agonist led to distinct phosphorylation kinetics of Akt, whereas at 15 minutes the extent of Akt-phosphorylation became comparable. We also tested ERK1/2- and p38-phosphorylation as they have been shown to be involved in platelet signaling after thrombin stimulation (16, 17). ERK1/2-phosphorylation peaked around 3 minutes after Pam3CSK4-activation and then abated, while thrombin-activation caused a quick increase, which remained constant after 10 minutes (Figure 2B). For p38, thrombin caused an almost immediate increase in phosphorylation, reaching a plateau after 3 minutes, while Pam3CSK4 led to only 60-70% of phosphorylation relative to thrombin (Figure 2C).
Next, we investigated whether protein-protein interactions are also differentially affected by thrombin- vs. Pam3CSK4-stimulation of platelets. We selected FXIIIa to study such interactions because its intracellular function lies in regulating the cytoskeletal remodeling during platelet activation and aggregation, and in cross-linking cytoskeletal proteins such as actin, myosin and vinculin (18). In addition, FXIIIa is found abundantly in platelets and, therefore, is well-suited for a MS-based proteomics approach.
We had to first identify potential FXIIIa-binding partners by immunoprecipitating FXIIIa from platelets. We confirmed that the immunoprecipitation of FXIIIa was specific through immunoblotting (Figure 3A). Separation of protein mixtures by 1D- and 2D-GE often result in identification of complementary sets of proteins due to the differing nature of sample preparation and separation characteristics (19), therefore, the FXIIIa-immunoprecipitates were separated by both methods. In 1D-GE, we used only resting platelets and excised all distinguishable protein bands, then subjected them to tryptic in-gel digestion, MALDI-TOF MS and PMF analysis (Figures 3B, 3C). Details of the identification of all marked proteins are summarized in Table 1. In Figure 3C, we present examples of acquired MALDI-TOF mass spectra for three identified proteins. Major peptide ions of the proteins are labeled with their detected m/z values and corresponding amino acid intervals. In the 2D-GE experiment we also included samples of thrombin-and Pam3CSK4-activated platelets in order to establish whether differences in the FXIIIa-associated protein profiles exist. Gels were stained with either silver stain (spot detection) or Coomassie Blue (spot excision), and all excised protein spots were digested with trypsin and again analyzed by MALDI-TOF MS and PMF. In Figure 4, we show a representative set of silver stained 2D-gels. Control antibody and protein G+ beads bound many proteins non-specifically in resting platelets (Figure 4A). The FXIIIa-specific immunoprecipitate from resting platelets is shown in Figure 4B, where the positions of all excised and identified protein spots are indicated. The spots at position 1 and 2 correspond to FXIIIa and suggest the presence of various FXIIIa isoforms. The identity of the remaining spots are listed in Table 1. Figures 4C and and4D4D represent FXIIIa-immunoprecipitates from thrombin- and Pam3CSK4-activated platelets, respectively. All subunits of fibrinogen - fibrinogen α (Figure 4B, spots 5a-d), fibrinogen β (spots 6a,b), fibrinogen γ (spots 9a,b) - were absent in the immunoprecipitate from thrombin-activated platelets (Figure 4C) but remained present in the immunoprecipitate from Pam3CSK4-activated platelets (Figure 4D). The 2D patterns of immunoprecipitated proteins were reproducible, but spot intensities varied between experiments and, thus, necessitated thorough validation of the FXIIIa-protein interactions by auxiliary techniques. Several spots visible in silver stained gels were not detected in Coomassie Blue stained gels and, therefore, eluded identification.
To confirm whether the identified proteins (Figures 3B, ,4B,4B, Table 1) were interacting specifically with FXIIIa, we performed immunoblot analysis for all 18 proteins and found, in resting platelets, FXIIIa binds specifically to four proteins: FAK, gelsolin, myosin IIA, and thrombospondin (Figure 5A, lanes “Resting”). All other identified proteins could not be confirmed to bind to FXIIIa in a specific manner as summarized in Table 1. While the identified FXIIIa-binding proteins myosin and thrombospondin are known substrates of FXIIIa (20), FAK and gelsolin have not previously been recognized. Since heat shock protein 27 (HSP27) had been reported to bind to FXIIIa in resting platelets (21), it was used as a positive control. However, it was not identified in our FXIIIa-immunoprecipitation experiments due to the low abundance of signaling molecules in platelets and the detection limit of the applied gel staining methods.
For further verification of the interaction between thrombospondin and FXIIIa, we performed co-immunoprecipitation experiments in resting platelets. Thrombospondin was immunoprecipitated and immunoblotted for FXIIIa (Figure 5B, bottom image) resulting in a much stronger signal for thrombospondin compared to the control, which demonstrates the specificity of the FXIIIa-thrombospondin interaction. Furthermore, we positively confirmed by confocal microscopy that the four FXIIIa-binding proteins co-localize with FXIIIa in resting as well as in activated platelets (Figure 6, last column of images in panels A-D).
We determined whether the interaction between FXIIIa and the four identified binding proteins changed upon platelet activation with different agonists. We immunoprecipitated FXIIIa and then immunoblotted for FAK, gelsolin, myosin and thrombospondin (Figure 5A, lanes “anti-FXIIIa”). The signal of immunoprecipitated FXIIIa did not change upon platelet activation (Figure 5C). For HSP27, platelet activation with thrombin led to reduced FXIIIa-binding, compared to resting platelets. Activation with Pam3CSK4, however, showed an even stronger reduction in FXIIIa-association. FAK-binding to FXIIIa showed no significant difference for resting vs. thrombin-activated platelets, however, its binding to FXIIIa decreased significantly in Pam3CSK4-vs. thrombin-activated platelets. For gelsolin and myosin, band intensities varied greatly and the averaged densitometry values did not show significant differences before and after activation or between thrombin- and Pam3CSK4-activated platelets. Thrombospondin, however, showed a significant signal reduction after platelet activation which was independent of the agonist.
Since thrombin and Pam3CSK4 activate PI3K, we investigated the effect of PI3K-inhibition by LY294002 on the FXIIIa-protein interactions. Although we showed in Figure 2A that the extent of Akt-phosphorylation for either agonist was comparable after 15 minutes of activation, the inhibition of PI3K had no effect on the interaction of FXIIIa and its binding proteins (data not shown) suggesting that the PI3K/Akt-signaling pathway is not involved in these protein-protein interactions.
The observed differential changes in selected signaling pathways and protein-protein interactions that were induced by the thrombotic vs. immune stimulation of platelets, were further examined by studying the release of α-granule proteins after activation. Platelets were activated for 15 minutes with either agonist. Releasates were then separated by 2D-GE and immunoblotted for the following proteins: thrombospondin, fibrinogen β, FXIIIa and gelsolin (known α-granule proteins), as well as platelet basic protein (PBP) and platelet factor 4 (PF4) (platelet-specific chemokines) (Figure 7). Releasates tested negative for microvesicle contamination (data not shown). As control, immunoblots of releasates from unstimulated platelets are shown (Figure 7Ai-vi). Thrombospondin was present in the releasate of thrombin- and Pam3CSK4-activated platelets with an agonist-specific pattern, whereas ADP-stimulated platelets did not contain this protein in a detectable amount (Figure 7Ai). The presence of fibrinogen β in the releasate was similar for Pam3CSK4- and ADP-activated platelets, with small differences in isoform patterns and intensities (Figure 7Aii). The very weak signal after thrombin-stimulation might be attributed to the its conversion of fibrinogen into fibrin (22). FXIIIa displayed a similar pattern of isoforms in each releasate (Figure 7Aiii). Interestingly, the spot intensity of each isoform varied and appeared to be agonist-dependent. Particularly after Pam3CSK4-activation, an additional spot at pH 6.5 was observed. Gelsolin, known to be released from platelets after thrombin-stimulation (13), was also released after activation with Pam3CSK4 and ADP (Figures 7Aiv) with releasate patterns similar to each other. However, the relative abundance of several gelsolin isoforms varied between agonists. The chemokine PBP was released in a similar manner after platelet activation with thrombin and Pam3CSK4 (Figure 7Av), while ADP generated a smaller amount. The chemokine PF4 (Figure 7Avi) was highest in the thrombin stimulated platelets, followed by Pam3CSK4-activation. ADP-stimulation yielded the weakest signal for PF4.
In addition, the effect of PI3K-inhibition by LY294002 on platelet α-granule release was examined. The concentration of LY294002 used was based on previous work done in platelets (8). As shown in Figure 7B, the extent of Akt-phosphorylation for either agonist was comparable at 15 minutes of activation. Increasing amounts of LY294002 caused a decrease in the release of PBP and thrombospondin in Pam3CSK4 stimulated platelets, while, the release of these two proteins was not affected by LY294002 after thrombin-stimulation. This suggests that the PI3K/Akt-pathway is differentially involved in the thrombin- vs. Pam3CSK4-induced α-granule release.
With the differences seen between thrombin and Pam3CSK4 in signaling, protein-protein interactions, αgranule release, and, specifically, FXIIIa, the question remains as to how these agonists affect platelet-dependent coagulation. Using the fibrin clot retraction assay, washed platelets were treated with fibrinogen and either 0.5 U/mL thrombin or 10 μg/mL Pam3CSK4, and incubated for 1 h. As seen in Figure 8A, the addition of thrombin resulted in a stable platelet aggregate, while Pam3CSK4 treatment did not cause any significant clot formation. Figure 8B shows quantitatively the amount of buffer remaining. Thrombin significantly reduces the volume to 82.6% ±5.4, while after Pam3CSK4 treatment the buffer volume remained the same (99.6% ±0.4), suggesting that FXIIIa released from platelets can crosslink fibrin monomers that were converted from fibrinogen by thrombin.
The releasate of thrombin-stimulated platelets contains a small amount of fibrinogen β, unlike Pam3CSK4-stimulation (Figure 7Aii). It is hypothesized that this discrepancy is due to the presence of thrombin converting fibrinogen into fibrin. If this is true, then any fibrinogen released from the platelets would be involved in aggregate formation, albeit small, in the presence of thrombin. Using the fibrin clot retraction assay (Figure 8A,B) without the addition of fibrinogen, thrombin stimulation of platelets resulted in a small, stable clot, which was absent after Pam3CSK4-stimulation. Therefore, fibrinogen released from the platelets was converted into fibrin monomers by thrombin, which binds αIIbβ3. FXIIIa released by the platelet would crosslink the fibrin to form a network onto which platelets adhere and form a stable aggregate (23).
Since Pam3CSK4-stimulation does not affect platelet-platelet interactions (homotypic aggreagates), it is possible that Pam3CSK4 could affect the interaction of platelets with other cells, such as monocytes (heterotypic aggregates). Whole blood left untreated or stimulated with 0.5 U/mL thrombin or 10 μg/mL Pam3CSK4 for 10 minutes was analyzed by flow cytometry to detect and quantitate the percent of platelet-positive monocytes. As seen in Figure 9A, thrombin stimulation did not significantly increase the number of platelet-positive monocytes over resting (15.7±1.4 vs. 15.8±1.1, respectively). Interestingly, Pam3CSK4-stimulation did significantly increase the percent of platelet-positive monocytes (23.2±1.9) over resting or thrombin. Using SEM, we were able to visualize the differences between thrombin- and Pam3CSK4-stimulation on platelet interactions with other blood cells. In Figure 9B, untreated samples contain platelets with a round, spherical appearance. Upon activation with thrombin, platelets underwent a shape change as seen with the formation of pseudopodia, which interact with one another and, to a lesser extent, with circulating white cells. Pam3CSK4-stimulation also resulted in a change of platelet shape, however, with a larger number of platelets binding to white cells. More than one platelet appeared to bind onto each white cell. Pam3CSK4-activated platelets bound to white cells were also binding one another. This interaction is causing white cells to localize and interact with themselves. Therefore, Pam3CSK4 bound to TLR2/1 to activates platelets during inflammation in order to localize and form interactions between white cells. Platelets stimulated with Pam3CSK4 utilize white cells to form a stable clot, unlike thrombin-stimulated platelets, which form a clot via crosslinked fibrin monomers converted by the actions of thrombin and FXIIIa.
We activated human platelets with thrombin or Pam3CSK4, and examined agonist-dependent effects on the phosphorylation kinetics of signaling proteins, on FXIIIa-protein interactions, and on α-granule release. Three major signaling proteins - Akt, ERK1/2, and p38 - were examined as to whether they are differentially affected by thrombin- vs. Pam3CSK4-stimulation. Thrombin-induced signaling cascades in platelets have been intensively studied (9), and PI3K is a known key player in these events (24). However, the details of signaling pathways involving ERK1/2 or p38 remain controversial (16, 17). TLR2 signaling, on the other hand, culminates in NF-κB activation in various cell types (2, 25), and PI3K has been implicated as a regulator of TLR signaling (26). Although anuclear platelets express NF-κB (27), no reports describing mechanistic details of the TLR2 signaling pathway within platelets have been published. Recent data from our laboratory have shown that PI3K/Akt play a major role in Pam3CSK4-induced platelet responses (8). Here, we further demonstrate that Pam3CSK4 also induces phosphorylation of ERK1/2 and p38 in a time-dependent and agonist-dependent manner. This clearly suggests that other pathways beside PI3K/Akt are involved in TLR2 signaling and Pam3CSK4-induced platelet responses.
The increased phosphorylation of Akt, ERK1/2, and p38 after activation with thrombin results in the formation of a more stable clot, as seen in the platelet aggregation and fibrin clot retraction assays, compared to Pam3CSK4-stimulation. Furthermore, the phosphorylation kinetics of Akt is faster with thrombin treatment compared to Pam3CSK4. PI3K/Akt pathways have been implicated in the outside-in signaling that occurs when fibrinogen binds to αIIbβ3 and activates signaling pathways that stabilize platelet aggregates (28). Important to clot stabilization is the greater extent of phosphorylation of both ERK1/2 and p38 with thrombin compared to Pam3CSK4. Both kinases are phosphorylated in more rapid manner, suggestive of their role in inside-out signaling, when signaling events activated by an agonist lead to conformational changes in αIIbβ3 (29). Therefore, the more stable clot formation occurring with thrombin is a result of the rapid phosphorylation of ERK1/2 and p38 causing conformational change in αIIbβ3 to increase fibrinogen binding, which results in an increased phosphorylation of Akt to further stabilize the aggregate.
The second part of our study focused on protein-protein interactions with particular emphasis on FXIIIa. For anuclear platelets, MS-based proteomic methods have proven to be an important tool for characterizing protein expression, post-translational modifications, and protein release (13, 30-35). Eighteen FXIIIa-binding proteins were identified and validated in resting and activated platelets. Only four of these proteins were found to bind to FXIIIa specifically. Known FXIIIa-substrates have been classified into three groups: coagulatory/fibrinolytic proteins, adhesive proteins, and cytoskeletal proteins (20). The four FXIIIa-binding proteins we identified in this study were: FAK, gelsolin, myosin IIA and thrombospondin. FAK and gelsolin are novel FXIIIa-interacting partners, while myosin IIA and thrombospondin had been reported to function as substrates for FXIIIa. FAK is a tyrosine kinase that mediates signaling through integrins and translocates to the cytoskeleton in platelets in the second wave of cytoskeletal rearrangements (36-38). The FAK-FXIIIa interaction could lead to crosslinking of FAK to other proteins, or phosphorylation of FXIIIa by FAK. Gelsolin, a cytoskeletal protein that regulates actin assembly (39), has also been found in α-granules (35), where its interaction with FXIIIa might.
It was examined whether these FXIIIa-protein interactions are differentially altered upon thrombotic vs. immune stimulation in comparison to resting platelets. We found differences in FXIIIa-binding to FAK, thrombospondin, and HSP27. The FXIIIa-FAK interaction changed in a statistically significant, agonist-dependent manner. The intensity of the interaction was not altered significantly after thrombin-stimulation, indicative of their translocation to the cytoskeleton where both proteins are thought to function after activation. In Pam3CSK4-stimulated platelets, the FAK-FXIIIa interaction decreased significantly compared to thrombin-activated platelets, suggesting a decrease in binding due to differences in signaling. The FXIIIathrombospondin interaction under resting conditions stems from α-granules, since thrombospondin is located only in these granules. After thrombin- or Pam3CSK4-activation, the intensity of the FXIIIa-thrombospondin interaction decreased significantly indicating that less thrombospondin was bound to FXIIIa after activation with either agonist. The FXIIIa-HSP27 binding for thrombin-activated platelets decreased compared to resting but not to the extent as that seen with Pam3CSK4. HSP27, like FAK, is thought to translocate to the cytoskeleton upon platelet stimulation (40). The observed differences in signaling might have led to the differences in FXIIIa binding. Additional experiments showed the inhibition of PI3K did not affect the observed differences in these FXIIIa-interactions, suggesting that the PI3K/Akt-pathway is not involved in regulating FXIIIa-protein interactions.
Platelet activation with thrombin, Pam3CSK4 or ADP was examined whether it causes a differential release of various α-granule proteins. Platelet α-granules are the major storage compartment for adhesive proteins, chemokines, receptors, coagulatory and fibrinolytic proteins (35, 41). Granules are distributed randomly within platelets, but, upon stimulation, fuse with the plasma membrane or open canalicular systems to release their contents. Our data suggest a differential, agonist-dependent release for selected proteins upon platelet activation with thrombin, Pam3CSK4, or ADP. The releasate of ADP-stimulated platelets lacked thrombospondin, while the releasate of thrombin-activated platelets lacked fibrinogen β, likely because it had been converted into fibrin (22). For FXIIIa and gelsolin, differences occurred in protein spots representing multiple isoforms and in spot intensity. For PBP and PF4, the most abundant chemokines released from platelets (44) were differently abundant in the respective releasates. One might expect that Pam3CSK4-stimulation could lead to an excessive release of these chemokines, however, we did not observe this. Pam3CSK4-stimulation of platelets that had been treated with increasing amounts of PI3K-inhibitor, caused a decrease in the release of PBP and thrombospondin. The thrombin-stimulated release of these two proteins was not affected by the inhibitor, suggesting that the PI3K/Akt signaling pathway is significantly involved in Pam3CSK4-induced, but not in thrombin-induced, α-granule release.
Finally, these data demonstrate the functional differences that occur when platelets are stimulated in the context of inflammation and thrombosis. Using thrombin as the agonist, platelets release fibrinogen, which is converted into fibrin by thrombin. FXIIIa crosslinks fibrin to form a stable, platelet clot. In normal hemostasis, this stable clot prevents hemorrhage and allows for proper wound healing. However, in a pathologic setting, the same clot can also prevent blood flow and cause an ischemic event to occur, such as in myocardial infarction. Stimulation with Pam3CSK4, which binds to TLR2/1, leads to the release of fibrinogen, which not only binds to platelets, but also to immune cells such as neutrophils and monocytes. Platelets will bind these cells and each other to localize the immune response to the site of injury.
In conclusion, our data demonstrate for the first time that platelet stimulation with thrombin or Pam3CSK4 leads to differential activation of signaling proteins, changes in protein-protein interactions, as well as α-granule release. These results highlight the differences in the platelet's immune vs. thrombotic responses and form the basis for further functional and mechanistic studies to better understand the platelet's role in innate immunity and inflammation.
We thank Dr. Kahraman Tanriverdi for his assistance with gel imaging.
This work was supported by NIH-NHLBI grant P01 HL083801 (J.E.F.), NIH-NHLBI contract N01 HV28178 (C.E.C.), NIH-NCRR grants P41 RR10888 and S10 RR15942 (C.E.C.), NIH NHLBI grant T32 HL07224 (L.M.B.).