Thrombin and Collagen Induce a Feedback Inhibitory Signaling Pathway in Platelets Involving Dissociation of the Catalytic Subunit of Protein Kinase A from an NFκB-IκB Complex*

Protein kinase A (PKA) activation by cAMP phosphorylates multiple target proteins in numerous platelet inhibitory pathways that have a very important role in maintaining circulating platelets in a resting state. Here we show that in thrombin- and collagen-stimulated platelets, PKA is activated by cAMP-independent mechanisms involving dissociation of the catalytic subunit of PKA (PKAc) from an NFκB-IκBα-PKAc complex. We demonstrate mRNA and protein expression for most of the NFκB family members in platelets. From resting platelets, PKAc was co-immunoprecipitated with IκBα, and conversely, IκBα was also co-immunoprecipitated with PKAc. This interaction was significantly reduced in thrombin- and collagen-stimulated platelets. Stimulation of platelets with thrombin- or collagen-activated IKK, at least partly by PI3 kinase-dependent pathways, leading to phosphorylation of IκBα, disruption of an IκBα-PKAc complex, and release of free, active PKAc, which phosphorylated VASP and other PKA substrates. IKK inhibitor inhibited thrombin-stimulated IkBα phosphorylation, PKA-IkBα dissociation, and VASP phosphorylation, and potentiated integrin αIIbβ3 activation and the early phase of platelet aggregation. We conclude that thrombin and collagen not only cause platelet activation but also appear to fine-tune this response by initiating downstream NFκB-dependent PKAc activation, as a novel feedback inhibitory signaling mechanism for preventing undesired platelet activation.

Protein kinase A (PKA) activation by cAMP phosphorylates multiple target proteins in numerous platelet inhibitory pathways that have a very important role in maintaining circulating platelets in a resting state. Here we show that in thrombin-and collagen-stimulated platelets, PKA is activated by cAMP-independent mechanisms involving dissociation of the catalytic subunit of PKA (PKAc) from an NFB-IB␣-PKAc complex. We demonstrate mRNA and protein expression for most of the NFB family members in platelets. From resting platelets, PKAc was co-immunoprecipitated with IB␣, and conversely, IB␣ was also co-immunoprecipitated with PKAc. This interaction was significantly reduced in thrombin-and collagen-stimulated platelets. Stimulation of platelets with thrombin-or collagen-activated IKK, at least partly by PI3 kinase-dependent pathways, leading to phosphorylation of IB␣, disruption of an IB␣-PKAc complex, and release of free, active PKAc, which phosphorylated VASP and other PKA substrates. IKK inhibitor inhibited thrombin-stimulated IkB␣ phosphorylation, PKA-IkB␣ dissociation, and VASP phosphorylation, and potentiated integrin ␣IIb␤3 activation and the early phase of platelet aggregation. We conclude that thrombin and collagen not only cause platelet activation but also appear to fine-tune this response by initiating downstream NFB-dependent PKAc activation, as a novel feedback inhibitory signaling mechanism for preventing undesired platelet activation.
Platelets are small anucleate cells derived from megakaryocytes in the bone marrow, in a process in which megakaryocyte cytoplasmic extensions into microvessels are sheared from their transendothelial stems by flowing blood (1)(2). Platelets play a key role in the normal homeostatic process through their ability to rapidly adhere to activated and/or injured endothe-lium and subendothelial matrix proteins (platelet adhesion), and to other activated platelets (platelet aggregation). Many factors bind to specific platelet receptors and regulate signaling pathways, which promote or inhibit platelet adhesion, aggregation, and secretion. In vivo, circulating platelets are continually exposed to a variety of activating factors including collagen, fibrinogen, ADP, von Willebrand Factor (vWF), thrombin, and thromboxane (3)(4)(5), as well as inhibitory factors such as endothelial-derived nitric oxide (NO), prostacyclin (PG-I 2 ), and ADPase (3,(5)(6). A strong equilibrium between the two opposing processes of platelet stimulation and inhibition is thought to be essential for normal platelet and vascular function. An impairment of this equilibrium will promote either thrombotic or bleeding disorders.
In the initial steps of platelet activation, the platelet receptor glycoproteins (GP) 3 1b and GPVI interact with extracellular matrix (ECM) proteins, causing platelets to tether and roll on the injured endothelium or subendothelial ECM (5). Stimulation of these receptors triggers intracellular signaling cascades that activate integrin ␣IIb␤3 and induce the release of secondary mediators like ADP and thromboxane A 2 (TXA 2 ), leading to full platelet activation and thrombus formation. However, most of the platelets that receive stimulatory signals and initially adhere to the ECM are later detached from the ECM by blood flow and returned back into the circulation.
In human platelets, established platelet inhibitors such as NO and PG-I 2 directly activate either the soluble guanylyl cyclase (sGC) or G s -protein-coupled prostanoid membrane receptors, respectively, and thereby increase the intracellular second messengers, cGMP and cAMP, both of which have been shown to play a crucial role in platelet inhibition (6 -9). The effects of the cyclic nucleotides are mediated via their respective cGMP-and cAMP-dependent protein kinases (PKG and PKA), which phosphorylate substrate proteins involved in platelet inhibitory pathways (6,9).
Recently we demonstrated cross-talk between platelet stimulatory and inhibitory pathways. Activation of human platelets by vWF caused NO-independent activation of soluble guanylyl cyclase and stimulation of cGMP production and PKG, thus initiating a feedback inhibitory pathway (10). We now demonstrate that thrombin and collagen stimulation of human platelets activate another distinct feedback inhibitory mechanism based on cAMP-independent activation of PKA.
PKA is a tetrameric holoenzyme consisting of a regulatory (PKAr) subunit dimer and two catalytic (PKAc) subunits. Elevation of cAMP levels and binding of cAMP to PKAr causes dissociation of the kinase complex and release of free active catalytic subunits (11)(12)(13)(14). However, in addition to this "classical" cAMP-dependent regulation of PKA activity, cAMP-independent activation of PKA has been demonstrated in different cell types (15)(16)(17). Some portion of PKAc molecules (independently from PKAr) is bound to IB in an NFB-IB complex. Stimulation of cells with inducers of NFB activity dissociates NFB from IB, leading to IB degradation and release, and cAMP-independent activation of PKAc (15). The NFB complex plays a significant role in megakaryocyte differentiation and maturation (18 -19) and is also expressed in platelets (20), in which, however, no functional role has yet been identified.
Here we show that, in platelets, PKAc is associated with an NFB-IB complex, and that during platelet activation by thrombin or collagen, active PKAc is released and phosphorylates VASP Ser157 as well as other PKA substrates. This particular pathway for thrombin/collagen activation of PKA is described for the first time in platelets, and has characteristics of a novel feedback inhibitory mechanism, which would reduce the likelihood of platelet activation, particularly under weak stimulus conditions.
Preparation of Washed Human Platelets-Human platelets were prepared and used experimentally as indicated in detail below, as reported previously (10,22) with small modifications. Blood was obtained from healthy volunteers according to our institutional guidelines and the Declaration of Helsinki, and our studies with human platelets were approved and recently (Sept. 24, 2008) reconfirmed by our local ethics committee of the University of Würzburg (Studies No. 67/92 and 114/04).
Blood was collected into ACD solution (12 mM citric acid, 15 mM sodium citrate, 25 mM D-glucose, final concentrations). Platelet-rich plasma (PRP) was obtained by 5 min of centrifugation at 330 ϫ g, and then apyrase (0.01 unit/ml, final concentration) was added. To reduce leukocyte contamination, PRP was diluted 1:1 with phosphate-buffered saline and centrifuged at 240 ϫ g for 10 min. Subsequently, the supernatant was centrifuged for 10 min at 430 ϫ g, then the pelleted platelets were washed once in CGS buffer (120 mM sodium chloride, 12.9 mM trisodium citrate, 30 mM D-glucose, pH 6.5), and resuspended in HEPES buffer (150 mM sodium chloride, 5 mM potassium chloride, 1 mM magnesium chloride, 10 mM D-glucose, 10 mM HEPES, pH 7.4) to a final concentration of 3 ϫ 10 8 platelets/ml. After 15 min rest in a 37°C water bath, washed platelets were used for experiments. Leukocyte contamination, counted using a leucocount kit (BD Biosciences), was less than 1 leukocyte per 10 6 platelets.
Aggregation Experiments-Platelet aggregation was measured using an Apact (LabiTec) aggregometer. Washed human platelets (3 ϫ 10 8 /ml) were preincubated with vehicle (0.01% DMSO), or 2 m IKK inhibitor VII for 5 min. Platelet aggregation was induced by addition of different concentrations of thrombin or collagen. Aggregation in response to thrombin or collagen in the absence or presence of IKK inhibitor was registered as change in light transmission and was calculated as the area under the aggregation curve (AUC, % aggregation ϫ s) and expressed as arbitrary units; aggregation caused by thrombin or collagen alone was designated as 100%.
Calcium Measurement-Calcium transients were determined with the fluorescence indicator Fura-2/AM. Platelets in PRP were loaded with Fura-2/AM for 45 min at 37°C. Excessive dye and plasma were removed by centrifugation. The pelleted platelets were then resuspended in HEPES buffer and diluted to a cell density of 2 ϫ 10 8 platelets/ml. Fura-2 fluorescence was measured at 340 nm with a Perkin-Elmer LS50 luminometer. Ca 2ϩ (1 mM) was added immediately before the experiment.
Immunoprecipitation and Western Blot Analysis-For immunoprecipitation (IP) platelets were pelleted and resuspended in IP buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% IGEPAL, and protease inhibitors (Complete Mini, Roche, Mannheim, Germany). Lysates were centrifuged for 20 min at 14,000 ϫ g and 4°C, and supernatants were incubated with either IB␣ or PKAc antibodies, or corresponding mouse or rabbit IgG as controls, overnight at 4°C. Then protein G-or cAMP-independent PKA Activation protein A-Sepharose 4B fast flow beads (GE Healthcare, Freiburg, Germany) were added for 1 h. Immunocomplexes were washed three times with IP buffer and twice with phosphate-buffered saline, then precipitated proteins were detected by immunoblotting.
For Western blotting, total platelet and HL-60 cell lysates were separated by SDS-PAGE, transferred to nitrocellulose membranes, and the membranes incubated with appropriate primary antibodies overnight at 4°C. For visualization of the signal, goat anti-rabbit or anti-mouse IgG conjugated with horseradish peroxidase was used as secondary antibodies, followed by ECL detection (GE Healthcare). Blots were analyzed densitometrically using NIH Image J software for uncalibrated optical density.
RNA Isolation from Platelets, and RT-PCR Analysis-Total RNA was extracted from washed platelets (1 ϫ 10 9 ) and HL-60 cells (1 ϫ 10 7 cells) by TRIzol Reagent (Invitrogen, Karlsruhe, Germany). Isolated RNA was reverse transcribed using the first strand cDNA synthesis kit (Stratagene, Waldbronn, Germany) according to the manufacturer's instructions. 1 l of cDNA synthesis reaction was subjected to PCR and amplified for 35 cycles. Primers sequences and expected sizes of transcribed gene products are given in Table 1. The purity of platelet cDNA was controlled with primers specific for genomic DNA; only those samples negative for genomic DNA contamination were selected for PCR amplification. The specificity of all PCR products was confirmed by sequencing using product-specific PCR primers.
Flow Cytometry-The level of integrin ␣ IIb ␤ 3 activation was determined using PAC-1 FITC-conjugated antibody against activated integrin ␣ IIb ␤ 3 . PAC-1 FITC-conjugated antibody (2 l) was directly added to 20 l of platelets (3 ϫ 10 8 /ml), diluted with 400 l of phosphate-buffered saline/0.3% bovine serum albumin, and subsequently analyzed in a Becton Dickinson FACS Calibur using CELLQuest software, version 3.1f. The platelet population was identified by its forward and side scatter distribution, and 10,000 events were analyzed for mean fluorescence.
cAMP and cGMP Measurements-Levels of cAMP and cGMP were evaluated using a cAMP EIA kit and cGMP EIA kit, respectively, following the manufacturer's instructions (Cayman Chemical, Hamburg, Germany).
NFB Activity Assay-NFB activity from platelets and HL-60 cells was assessed by the binding of activated p65 to a consensus binding site on biotinylated oligonucleotides, measured by an ELISA kit from Panomics (Fremont, CA) according to the manufacturer's instructions and was expressed as arbitrary units of DNA binding.
Data Analysis-All experiments were performed at least in triplicate, and data shown are means Ϯ S.E. The n-values refer to the number of experiments, each made with different blood donors. Differences between groups were analyzed by Student's t test. ANOVA was used for time and dose-response experiments. p Ͻ 0.05 was considered statistically significant.

Cyclic Nucleotide-independent Phosphorylation of VASP Ser157 and Other PKA Substrates in Thrombin-and Collagen-stimulated
Platelets-VASP is one of the major PKA and PKG substrates in platelets and has been used to monitor the activation state of both kinases in numerous studies (23). Treatment of platelets with thrombin resulted in phosphorylation of VASP at Ser 157 , the site preferred by PKA (Fig. 1A). VASP Ser239 , the site more preferred by PKG, was not phosphorylated after thrombin treatment (Fig. 1A). To investigate whether thrombin-induced VASP Ser157 phosphorylation was related to elevated cyclic nucleotide levels, we determined cAMP and cGMP concentrations. Increase of cGMP in thrombin-and collagen-stimulated platelets has been described in several publications (reviewed in Ref. 24), whereas cAMP has been shown to decrease in thrombin-stimulated platelets (25)(26). In our experiments, we could not detect any increase of cGMP in thrombin-or collagen-stimulated platelets (Fig. 1B), which directly correlated with the unchanged, or even slightly reduced, VASP Ser239 phosphorylation observed in response to thrombin in Fig. 1A. We detected a significant decrease in cAMP only in thrombin-stimulated, not collagen-stimulated platelets. cAMP concentration in control was 17.1 Ϯ 0.6 pmol/ 10 9 platelets, in collagen-stimulated platelets 16.5 Ϯ 0.7 pmol/ 10 9 platelets, and in thrombin-stimulated platelets 11.4 Ϯ 0.5 pmol/10 9 platelets (n ϭ 4, p Ͻ 0.05). cGMP content in control was 0.11 Ϯ 0.02 pmol/10 9 platelets and was not significantly changed in thrombin-(0.12 Ϯ 0.03), or collagen-(0.13 Ϯ 0.03) stimulated platelets (Fig. 1B). In contrast, forskolin and sodium nitroprusside (SNP) (5 M each) elicited expected cAMP and cGMP increases, respectively.
To investigate whether thrombin-induced VASP phosphorylation at Ser 157 was mediated by PKA, we used two different PKA inhibitors, H-89 that inhibits ATP binding to PKAc and Rp-8-Br-cAMPS that is a competitive inhibitor of cAMP-binding sites in PKAr. Both inhibitors strongly decreased the VASP phosphorylation caused by forskolin, an agent which activates cAMP-independent PKA Activation adenylyl cyclase, and as a consequence, cAMP-dependent PKA stimulation. In contrast, thrombin ( Fig. 1A) (and collagen, data not shown) stimulated VASP phosphorylation was only inhibited by H-89, not by Rp-8-Br-cAMPS, suggesting that thrombin action was cAMP-independent. Thrombin or collagen stimulation of VASP Ser157 phosphorylation was enhanced by a very low forskolin concentration (100 nM), suggesting that the cAMPdependent (forskolin) and cAMP-independent effects were additive (see supplemental Fig. S1). We determined that VASP is not the only PKA substrate phosphorylated in response to thrombin and collagen stimulation of platelets by examining the phosphorylation of Rap1GAP2, which has been recently identified as a substrate of PKA and PKG in platelets (21,27). In our experiments, Rap1GAP2 was indeed phosphorylated in thrombin-and collagen-stimulated platelets (Fig. 1C). To gain a broader view of PKA-mediated phosphorylation events, we used a PKA substrate antibody that recognizes the RRX(S/T) consensus phosphorylation site when S or T is phosphorylated. Numerous phosphoproteins are recognized by this antibody after forskolin, as well as thrombin or collagen treatment. Several of them (arrows in supplemental Fig. S2) have the same molecular weight, suggesting that thrombin and collagen activate pathways that lead to phosphorylation of PKA substrates, which are also targeted by forskolin. Taken together, our data ( Fig. 1, A-C) suggest that, unlike forskolin, thrombin and collagen induce phosphorylation of VASP and other PKA substrates independently of cyclic nucleotides. Next, we evaluated the concentration and time dependence of VASP ser157 phosphorylation in thrombin-, collagen-, and forskolin-stimulated platelets (Fig. 2). Platelets were stimulated with thrombin (0.01, 0.05, 0.1 unit/ml), collagen (1, 5, 10 g/ml), or forskolin (0.1, 0.5, 1 M) for 2 min ( Fig. 2A), or with 0.01 unit/ml of thrombin, 1 g/ml of collagen, and 0.1 M forskolin from 1 to 10 min (Fig.  2B). VASP Ser157 phosphorylation was independent of collagen and thrombin concentration, but dependent on forskolin concentration ( Fig. 2A). VASP Ser157 phosphorylation induced by all three compounds was not significantly changed from 1 to 10 min of stimulation (Fig. 2B).

Analysis of Intracellular Mechanisms of Thrombin-and Collagen-induced VASP Ser157 Phosphorylation in Platelets-To
exclude that other S/T kinases, which are also activated by thrombin or collagen, might be involved in VASP Ser157 phosphorylation, we analyzed effects of different kinase inhibitors using established kinase substrates as controls. VASP Ser157 has been shown to be directly phosphorylated in vitro by protein kinase C (PKC) (28). In thrombin-stimulated platelets VASP Ser157 has been described to be phosphorylated by PKCdependent and -independent mechanisms (29 -30). To investigate whether PKC phosphorylates VASP Ser157 in isolated plate-FIGURE 1. cAMP-independent PKA activation in thrombin-and collagenstimulated platelets. A-C, washed human platelets (3 ϫ 10 8 /ml) were stimulated for 2 min with thrombin (0.01 unit/ml), collagen (10 g/ml), or forskolin (0.1 M), or prior to stimulation preincubated with the indicated inhibitors for 10 min, and then processed for Western blot analysis (A, C) or cAMP/cGMP measurement (B). A, platelets were untreated or preincubated with PKA inhibitors (H-89, 10 M; Rp-8-pCPT-cAMPS, 200 M), prior to stimulation with thrombin or forskolin. Platelet activation was monitored by p38 phosphorylation. Immunoblots were scanned and quantified by the Image J program, the intensity of the VASP Ser157 signal normalized to the total p38 signal, and then this ratio for each sample was expressed relative to the ratio for thrombin stimulation which was designated as 1. H-89 inhibited both thrombinand forskolin-stimulated VASP Ser157 phosphorylation, whereas Rp-8-pCPT-cAMPS inhibited only forskolin-stimulated VASP Ser157 phosphorylation. Results are means Ϯ S.E., n ϭ 4, ϩ p Ͻ 0.05 compared with the control; *, p Ͻ 0.05 compared with the corresponding stimulus. B, cAMP and cGMP levels in thrombin-and collagen-stimulated platelets. Forskolin-and sodium nitroprusside (SNP) (both 5 M, 2 min)-stimulated platelets were used as positive controls. cAMP/cGMP concentrations were measured as described under "Experimental Procedures." cAMP concentration significantly decreased only in thrombin-stimulated platelets. Results are means Ϯ S.E., n ϭ 4, ϩ p Ͻ 0.05 compared with control. C, platelets were incubated with thrombin (0.01 unit/ml) or collagen (10 g/ml), or each of these after preincubation with H89 (10 M, 10 min) as indicated, and analyzed for VASP Ser157 and Rap1GAP2 Ser7 phosphorylation, or total Rap1GAP2 expression. Results are representative of three independent experiments.

cAMP-independent PKA Activation
lets, we first tested several PKC inhibitors (Bis IX and I, Gö 6976, Gö 6983, hispidin, and chelerythrine), demonstrating that all of these compounds inhibited thrombin-and collageninduced platelet integrin ␣IIb␤3 activation (PAC-1 binding assay), and the phosphorylation of two independent PKC substrates (MARCKS and pleckstrin) (Bis IX and I Fig. 3A, and other data not shown). However, thrombin-and collagen-induced VASP Ser157 phosphorylation was not inhibited by any of these compounds, and some of them (Bis IX and I) even further increased VASP Ser157 phosphorylation in some experiments ( Fig. 3A and data not shown).
In endothelial cells, thrombin-induced VASP Ser157 phosphorylation has also been shown to be dependent on RhoA kinase and MEKK/ERK pathways (16). Here we tested whether any of these kinases might be involved in thrombin-induced VASP Ser157 phosphorylation in platelets (Fig. 3, E and F). In thrombin-and collagen-stimulated platelets, the Rho kinase inhibitor Y27623 specifically inhibited only phosphorylation of its established substrate myosin light chain 2 (MLC) but had no effect on VASP Ser157 and p38 MAP kinase phosphorylation (Fig. 3E). Even high concentrations (up to 50 M) of U0126, an inhibitor of the MEKK/ERK pathway, specifically inhibited only ERK phosphorylation, not VASP Ser157 phosphorylation, in collagen (data not shown)-or thrombin (Fig. 3F)-stimulated platelets. We conclude that thrombin-and collagen-induced phosphorylation of VASP Ser157 most likely involves PI3K and PKA, but none of the other kinases are suggested to phosphorylate VASP at this site.
Expression of NFB Family Proteins and mRNA in Human Platelets-The NFB transcription factors are members of three families (IKK, NFB, and IB) of proteins (33). Expression of only some of the NFB family members has been demonstrated in platelets (20). Therefore we performed a more comprehensive examination of the protein and mRNA expression of the most common members of the NFB, IB, and IKK family in platelets, using HL-60 cells as a nucleated positive control (Fig. 4). The purity of platelet cDNAs was controlled using primers specific for genomic DNA (data not shown). All three IKK (␣, ␤, and ␥) family members are expressed in platelets and HL-60 cells, with IKK␤ protein being even more strongly expressed in platelets than in HL-60 cells. Of the NFB/Rel family, Rel B protein was found to be most strongly expressed, and C-Rel undetectable, in platelets (Fig. 4). Of the IB family, Bcl-3 has been described as expressed in platelets (34), and we found that IB␣ and IB␤ proteins are also expressed in platelets, however at lower levels than in HL-60 cells (Fig. 4). Washed human platelets (3 ϫ 10 8 /ml) were incubated (A) for 1 min with indicated concentrations of thrombin, collagen, or forskolin, or (B) with thrombin (0.01 unit/ml), collagen (1 g/ml), or forskolin (0.1 M) for the indicated times, and then processed for Western blot analysis (VASP Ser157 , total VASP, and GAPDH). Immunoblots were scanned and quantified as described in Fig.  1A. Results are means Ϯ S.E., n ϭ 6, ϩ p Ͻ 0.05 compared with no stimulus. Only forskolin-stimulated VASP Ser157 phosphorylation was time-and concentration-dependent.  C) represents integrin ␣IIb␤3 activation and is expressed as % of the thrombin effect, which was designated as 100%. Immunoblots (D) were scanned and the intensity of bands quantified by the Image J program. VASP (Ⅺ) and PKB (f) phosphorylation were normalized to the total PKB signal. Results are means Ϯ S.E., n ϭ 4. A and C, ϩsignificant difference from control; *, significant difference from thrombin-or convulxin-stimulated platelets, respectively. D, ϩsignificant difference from thrombin-or forskolinstimulated platelets, respectively.

cAMP-independent PKA Activation in Thrombin-and Collagen-stimulated Human Platelets Is Mediated by Dissociation of
PKAc from an NFB-IB Complex-After the first description that PKA activity may be regulated independently of cAMP by dissociation of PKAc from an NFB-IB complex (15), this mechanism was described in several cell types (16 -17). In nucleated cells, activation of PKA is associated with increased transcriptional activity of NFB. However, the function of NFB complexes in anucleate platelets has not been described. We therefore examined whether (i) PKAc is also associated with the NFB-IB complex in platelets, (ii) PKAc is released from this complex in agonist-stimulated platelets, (iii) NFB is activated in agonist-stimulated platelets, and (iv) whether cAMP-independent PKA activation in thrombin-stimulated platelets has functional relevance.
In resting platelets, PKAc was co-immunoprecipitated with IB␣ (Fig. 5, A and B), and IB␣ was co-immunoprecipitated with PKAc (Fig. 5, C and D). In thrombin-and collagen-stimulated platelets, both the amount of PKAc co-precipitated with IB␣, and the amount of IB␣ co-precipitated with PKAc was significantly reduced (Fig. 5, E and F). This reduced binding of PKAc to IB␣ indicates that in platelets, as in other cell types, activation disrupts the IB␣-PKAc complex, resulting in the release of free active PKAc. Disruption of the IB␣-PKAc complex and phosphorylation of VASP Ser157 in thrombin-stimulated platelets was inhibited by preincubation with IKK inhibi-  Washed human platelets (1 ϫ 10 9 /ml) were incubated for 2 min with thrombin (0.01 unit/ml) or collagen (10 g/ml), pelleted, and resuspended in IP buffer. Specific antibodies were used to precipitate IB␣ (A, B, G, H) or PKAc (C, D, G) overnight at 4°C, followed by protein A or G-Sepharose beads for 1 h (IP). Precipitated proteins were solubilized, separated by SDS-PAGE, and immunoblotted (WB), and blots were treated with IB␣ (A, D, G) or PKAc (B, C, G) antibodies. As control, nonspecific antibodies (IgG) were used for IP. In G, platelets were preincubated with or without IKK inhibitor VII (5 M, 10 min), then stimulated with thrombin (0.01 unit/ ml) for 2 min and analyzed by Western blotting for VASP phosphorylation (WB, without prior IP), or analyzed by IP with IB␣ or PKAc antibodies and subsequent WB. To estimate the total concentration of PKAc expressed in platelets, in comparison to the amount of PKAc that forms a complex with IB␣, different amount of platelet proteins and purified PKAc were loaded on the same gel as the samples from IP of IB from platelets (H). Immunoblots were scanned, and the intensity of bands quantified by the Image J program; in E and F, intensities were expressed as fold change with respect to control samples which were designated as 1. Results are means Ϯ S.E., n ϭ 4, ϩ significant difference from respective controls. cAMP-independent PKA Activation tor (Fig. 5G). Next, we calculated the amount of PKAc that forms a complex with IB␣. For this, we immunoprecipitated IB␣ and performed a Western blot analysis of PKAc in the precipitate, as well as in defined amounts of platelet lysate and purified PKAc (Fig. 5G). Immunoblots were scanned and the intensity of the PKAc bands was quantified using the Image J software. Regression analysis of PKAc band intensity showed a linear correlation with g of loaded platelet proteins and ng of purified PKAc. From these data we observed that 1 g of platelet protein contains 0.58 Ϯ 0.03 ng of PKAc, in agreement with our previous calculations (35). Collectively, the data in Fig. 5G can be used to deduce that 1.63% of total platelet PKAc forms a complex with and is co-precipitated with IB␣. However, this amount of PKAc may be underestimated because it depends on the efficiency with which the IB␣ antibody immunoprecipitates the IB␣-PKAc complex.
NFB Activation in Thrombin-stimulated Platelets-In nucleated cells, activation of NFB leads to phosphorylation, ubiquitination, and degradation of IB proteins. In thrombinstimulated platelets, IB␣ was phosphorylated at Ser 32/36 , and the phosphorylation was blocked by IKK inhibitor (panel ϩ IKK inh in Fig. 6A). Also, the level of total IB␣ was significantly decreased after 10 min of thrombin stimulation (bar graph). This degradation of IB␣ was prevented by preincubation with the proteasome inhibitor MG-132 (panel ϩ MG-132 in Fig.  6A). Activated NFB binds to its DNA consensus sequence, and NFB activity can be measured by ELISA with biotinylated oligonucleotides containing this consensus sequence. In both platelets and HL-60 cells (included as a positive control), thrombin activated NFB and induced binding of activated NFB to target oligonucleotides (Fig. 6B). NFB DNA binding increased in thrombin-stimulated platelet lysate continually over the 1-h incubation period, whereas it peaked after 10 min in HL60 cells.

Inhibition of IKK Potentiates Thrombin-and Collagen-stimulated Platelet Activation without Significant Changes in Cal-
cium Mobilization-To demonstrate the physiological relevance of thrombin-and collagen-induced VASP Ser157 phosphorylation, we investigated effects of IKK inhibitor on integrin ␣IIb␤3 activation, calcium mobilization, and platelet aggregation. Preincubation of platelets with different concentrations of IKK inhibitor strongly reduced thrombin-induced VASP Ser157 phosphorylation, but had no effect on forskolininduced VASP Ser157 phosphorylation (Fig. 6C). IKK inhibitor, at low concentrations up to 5 M, not only inhibited VASP phosphorylation, but also significantly increased thrombinstimulated integrin ␣IIb␤3 activation (PAC-1 binding). In collagen-stimulated platelets IKK inhibitor also reduced VASP Ser157 phosphorylation, however the potentiating effect on PAC-1 binding was very low and not significant (data not shown). Preincubation of platelets with IKK inhibitor caused only slight (not significant) potentiation of thrombin-, but not collagen-induced calcium mobilization (supplemental Fig. S3), indicating that changes other than in intracellular calcium are involved in NFB effects on platelets. The effects of all concentrations of thrombin (0.001, 0.01, 0.02, and 0.05 unit/ml) and collagen (1, 5, and 10 g/ml) on the early phase of platelet aggregation were significantly potentiated by IKK inhibitor VII FIGURE 6. NFB activation in thrombin-stimulated platelets. A, IB␣ phosphorylation and degradation in thrombin-stimulated platelets. Washed human platelets (3 ϫ 10 8 /ml) were incubated with thrombin (0.01 unit/ml, all 6 Western blot panels) for the indicated times, or were first preincubated for 10 min with either 5 M of IKK inhibitor VII (ϩIKK inh., third panel from top), or with 10 M proteasome inhibitor (ϩMG132, fifth panel from top). Samples were analyzed by Western blotting for VASP Ser157 and IB␣ Ser32/36 phosphorylation, as well as for NFB and IB␣ total protein expression. Immunoblots were scanned, and the intensity of bands was quantified by the Image J program. IB␣ was normalized to the NFB signal, and the ratio was expressed as fold change with respect to the control sample (0 min thrombin), which was designated as 1. B, time-dependent thrombin-induced platelet and HL-60 cell NFB activation. Washed human platelets (3 ϫ 10 8 /ml) and HL-60 cells (1 ϫ 10 7 cells/ml) were incubated with thrombin (0.01 unit/ml for platelets, 0.1 unit/ml for HL-60 cells) for the indicated times, then analyzed for NFB activation using a DNA binding assay (ELISA), as described under "Experimental Procedures." C, IKK inhibitor inhibits thrombin, but not forskolin-induced VASP Ser157 phosphorylation, and potentiates thrombin-stimulated PAC-1 binding. Washed human platelets (3 ϫ 10 8 /ml) were stimulated with thrombin (0.001 unit/ml) or forskolin (1 M), or preincubated with the indicated concentrations of IKK inhibitor VII for 10 min prior to stimulation, then analyzed by FACS for integrin ␣IIb␤3 activation (PAC-1 binding), and by Western blotting for VASP and p38 phosphorylation. Results are means Ϯ S.E., n ϭ 3, ϩ, significantly different from control in A and B, and from thrombin in C.

DISCUSSION
Platelet inhibitory mechanisms play an important role in preventing circulating platelets from undesired activation. It is widely accepted that cyclic nucleotides (cAMP and cGMP) and their corresponding protein kinases (PKA and PKG) are key players in regulation of platelet inhibition. However, recently, several other mechanisms involved in platelet inhibition have been described. For example, constitutive activity of a novel platelet receptor G6b-B is involved in inhibition of GPVI and CLEC-2 signaling, and speculated to have an important physiological role in helping to prevent platelet activation in vivo (36). The canonical Wnt signaling was recently described in platelets and shown to be involved in inhibition of platelet adhesion, shape change, dense granule secretion, RhoA activation, and aggregation (37). Activation of peroxisome proliferator-activated receptor ␥ (PPAR␥) inhibited collagen-stimulated platelet aggregation, intracellular calcium mobilization, P-selectin exposure, and thrombus formation (38). Protease nexin-1 (PN-1), stored in ␣-granules and released during platelet activation, is involved in inhibition of tissue factor-induced thrombin generation, low-dose thrombin-induced P-selectin surface expression, and platelet aggregation (39) An adapter protein disabled-2 (DAB2), which is also associated with ␣-granules and secreted from agonist-stimulated platelets, inhibited platelet integrin ␣IIb␤3 activation and platelet aggregation by binding to the extracellular region of ␣IIb integrin (40). Diadenosine 5Ј, 5ЈЉ-P 1 ,P 4 -tetraphosphate (Ap 4 A), a component of platelet dense granules, is involved in inhibition of ADP-induced platelet activation (41). Interestingly, the last three inhibitors of platelets (PN-1, DAB2, and Ap 4 A) act as feedback inhibitors that are activated only in agonist-stimulated platelets.
Here we describe a novel platelet feedback inhibitory mechanism involving PKA, which is activated in thrombin-and collagen-stimulated platelets. In our studies, thrombin and collagen activation of platelets increased phosphorylation of the VASP protein, which is a major substrate of PKA, a well-known inhibitor of platelet signaling. PKA is normally activated by cAMP-dependent dissociation of the regulatory and catalytic holoenzyme PKAr-PKAc. However, platelet activation by soluble agonists commonly involves a reduction in intracellular cAMP levels. For example, ADP and epinephrine induce G i -protein-dependent inhibition of adenylyl cyclases. Thrombin stimulation of platelets has been reported to decrease cAMP by PKB-and PKC-dependent phosphorylation and activation of PDE3A (25)(26). Our experiments also confirmed that thrombin decreases cAMP levels in platelets (Fig. 1B). Furthermore, we could show that thrombin activation of PKAc in platelets is cAMP independent, instead involves dissociation of an NFB-IB␣-PKAc complex to release free PKAc (Fig. 5), which in turn opposes the platelet activation initiated by thrombin.
Although PKA is mainly activated by cAMP, a fraction of total cellular PKA forms a complex with NFB-IB proteins and may be released upon NFB activation by different stimuli (15)(16), including thrombin in endothelial cells (15)(16). Expression of NFB (p65) and IB␣ proteins, and their activation by thrombin and other agonists have been shown in platelets (20). During preparation of this report for publication, three reports concerning NFB expression and function in platelets were published (42)(43)(44). In contrast to our data, authors of these works report that NFB plays a significant role in platelet activation. In the report of Malaver et al. (42), two NFB inhibitors (BAY 11-783 and Ro 106-9920) were used at high concentrations (up to 50 M) without any appropriate controls for their specificity. In our study, we also used both of these inhibitors and found that even much lower concentrations (less than 5 M), independently of NFB, significantly inhibited several platelet activation pathways including PI3K/ PKB, as well as inhibiting integrin ␣IIb␤3 activation, aggregation, and P-selectin expression (data not shown). The conclusions of this report (42) were also questioned by an editorial (45) published in the same issue of the Journal of Thrombosis and Hemostasis. An article by Lee et al. (43) reported that 5-20 M of the same NFB inhibitor (BAY 11-783) had significant inhibitory effects, in rat platelets. Species-dependent differences of NFB expression and function may exist in platelets, because our studies found that, in contrast to human platelets, thrombin-and collagen-stimulated mouse platelets did not show any significant VASP phosphorylation (data not shown), indicating that at least this NFB function is undetectable in mouse platelets. In agreement with our data (Fig. 4), Spinelli et al. (44) demonstrated expression of most NFB family members in platelets; however, in contrast to the two previous works (42)(43), reported significant platelet inhibitory effects already at low (0.5-5 M) concentrations of BAY 11-783 (44). However,

cAMP-independent PKA Activation
to prove the specificity of BAY 11-783, the authors used a rather questionable method of introducing recombinant full-length proteins into platelets.
The classical role of NFB is associated with the regulation of gene expression, a function which can be excluded in anucleate platelets. Here we describe a new function of NFB proteins involving cAMP-independent activation of PKA in thrombinand collagen-stimulated platelets, which we propose represents a novel platelet inhibitory feedback mechanism.
The first indication of PKA activation in thrombin-and collagen-stimulated platelets was phosphorylation of the PKA substrate VASP Ser157 . Furthermore, inhibition of PKAr by Rp-8-pCPT-cAMPS had no effect on thrombin-induced VASP Ser157 phosphorylation (Fig. 1A), indicating that under these conditions PKA was activated by a cAMP-independent pathway. The alternative, that PKA substrates could be phosphorylated by other kinases activated by thrombin and collagen, was examined.
Because several studies described PKC as a kinase which phosphorylates VASP Ser157 (28 -30), we analyzed this possibility in detail, using more than 10 different PKC inhibitors. We verified the functional activity of each inhibitor by measuring its effect on the phosphorylation of established PKC substrates (MARCKS, pleckstrin) and on the inhibition of integrin ␣IIb␤3 activation. Most PKC inhibitors inhibited PKC activity (assessed by MARCKS and pleckstrin phosphorylation and integrin ␣IIb␤3 activation) with different efficacy; however none of them inhibited thrombin-or collagen-induced VASP Ser157 phosphorylation in platelets ( Fig. 3A and data not shown). However, direct activation of PKC using PMA did induce VASP Ser157 phosphorylation, which was inhibited by H-89 (data not shown). PMA is a well known inducer of NFB activation in platelets, in a fashion similar to that of thrombin or collagen (20). 4 Based on these data, we conclude that PMA-induced VASP Ser157 phosphorylation in platelets is not mediated by PKC itself, but rather by PKC-dependent NFB activation followed by the release of active PKAc from an IB-PKAc complex.
In our experiments, thrombin-induced VASP Ser157 phosphorylation was partially blocked by the PI3 kinase inhibitor wortmannin (Fig. 3B), which can also inhibit PKB activity downstream of PI3K. However, the wortmannin effect was not mediated by PKB, because the PKB inhibitor PKI-AKT (which inhibits phosphorylation of the PKB substrate GSK3) had no effect on VASP Ser157 phosphorylation (Fig. 3C). A second important observation from these experiments was the lack of specificity of the commonly used PKA inhibitor H-89. We expected that inhibition of PKA by H-89 would potentiate platelet activation, but in all of our experiments, H-89 dose-dependently and significantly inhibited thrombin-or collageninduced integrin ␣IIb␤3 activation, which can be explained by the inhibition of PI3 kinase/PKB pathways (inhibition of P-GSK3, Fig. 3C, and P-PKB, Fig. 3D). Unfortunately, in platelets only relatively high concentrations (greater than 5 M) of H-89 can be used, because lower concentrations had no effect on PKA activity assessed by VASP phosphorylation (Fig. 3D and data not shown). We also tested other commercially available PKA inhibitors (KT5720 and PKI) at different concentrations. KT5720 required even higher (greater than 20 M) concentrations to inhibit VASP phosphorylation and at these concentrations KT5720 started to inhibit other kinases (PI3K, PKB, PKC) and integrin ␣IIb␤3 activation (data not shown). PKI (5 M), was previously shown by us to have no effect on platelet PKA activity and higher concentrations even strongly stimulated platelets without inhibition of PKA (22).
In endothelial cells, thrombin-induced VASP Ser157 phosphorylation is mediated by RhoA/Rho kinase and MEKK1 that are downstream of the G␣ 13 pathway (16). In platelets, thrombin also stimulates the G␣ 13 pathway (46) and its downstream effectors MEKK1 and Rho kinase, however, in contrast to endothelial cells, neither of these kinases are involved in thrombininduced VASP Ser157 phosphorylation (Fig. 3, E and F).
Thrombin-induced NFB activation in both platelets and HL-60 cells (used as a positive control) was demonstrated by IB␣ phosphorylation and degradation, and by increased DNA-binding of activated NFB. We could show that in platelets, like in other cell types (15,17,47), a portion of PKAc binds to IB␣ and co-immunoprecipitates with IB␣. This interaction was significantly reduced in thrombin-and collagen-stimulated platelets, indicating release of catalytically active PKAc from IB␣ (Fig. 5). In addition, low dose IKK inhibitor (up to 5 M) inhibited thrombin-induced disruption of the IB␣-PKAc complex (Fig. 5G), inhibited thrombin-induced VASP Ser157 phosphorylation, and potentiated integrin ␣IIb␤3 activation, without having any effect on forskolin-and thus cAMP-dependent VASP Ser157 phosphorylation (Fig. 6C). Importantly, inhibition of NFB function by IKK inhibitor VII significantly potentiated thrombin-and collagen-stimulated platelet aggregation, especially the early phase (Fig. 7).
Platelet integrin activation and aggregation are consequences of several, partially independent, mechanisms, which include protein phosphorylation/dephosphorylation, calcium mobilization, changes in protein/protein interactions, etc. The potentiating effect of IKK inhibitor on the final steps of platelet activation (integrin activation, aggregation) is small, not more than 20%, and is potentially a composite of effects on one or more of the above-mentioned mechanisms, although not calcium mobilization (which was not significantly influenced by IKK inhibitor, see supplemental Fig. S3). A thorough analysis of intracellular mechanisms involved in mediation of NFB functions in platelets merits future investigation.
Based on our data, we propose the following mechanisms and effects of activation of an NFB-IB-PKAc complex in platelets (Fig. 8). Activation of thrombin and collagen receptors stimulates PI3 kinase/PKB/PKC pathways (for the sake of simplicity, other well-established signaling cascades are omitted from the scheme), ultimately leading to platelet activation. Both receptors partly utilize the PI3 kinase (and possibly other yet to be established) pathways to activate the NFB complex, which leads to degradation of IB, release of active PKAc, and consequently to the phosphorylation of VASP and other PKA substrates involved in platelet inhibitory pathways. However, our experiments do not completely rule out the possibility that other, yet unidentified kinases besides PKA can phosphorylate VASP and other PKA substrates and initiate feedback inhibition in platelets. Nor do our data exclude other potential, yet unidentified functions of NFB family members in platelets (Fig. 8).
In nucleated cells, NFB acts as a transcription factor and regulates gene expression under various conditions (33). NFB now appears, like several other transcription factors including peroxisome proliferator-activated receptor (48), steroid (49), and glucocorticoid receptors (50), and others, to be expressed in platelets and involved in diverse platelet functions. Most of the NFB transcription factor family members interact with and may regulate functions of other proteins. In addition to PKAc, some other interacting partners of NFB that are relevant to platelet functions include c-Src, which interacts with p65 (51) and the IKK complex (52), 14-3-3 proteins (53), PKC (54), and protein phosphatase A2 (55). The relevance of NFB family members in platelets is further supported by the abundance of their transcripts. Quantitative profiling of the platelet transcriptome revealed that IKK␣ and IKK␤ are represented by 88 and 24 tags, respectively, per 200,000 tags in the SAGE library derived from platelet cDNA (56). Tag counts in a SAGE library correlate with the abundance of their specific mRNAs (for review see Ref. 41), and the tag counts for IKK␣ in platelets exceed those found in any other library in SAGE Anatomic Viewer (see SAV). These data collectively indicate that NFB transcription factors could have functions other than regulation of gene expression and that anucleate platelets are a relevant model for investigating these functions.
In summary, we show here that most of the NFB family members are expressed in platelets at the mRNA and protein level. Stimulation of platelets with thrombin or collagen disrupts an NFB-IB␣-PKAc complex, which leads to PKA activation and phosphorylation of VASP and other PKA substrates involved in platelet inhibitory pathways. An IKK inhibitor inhibited this chain of events and potentiated ␣IIb␤3 integrin activation and aggregation of platelets, suggesting that NFBdependent PKAc activation represents a novel feedback inhibitory mechanism to modulate platelet functions.