Dual Role of Platelet Protein Kinase C in Thrombus Formation

Human platelets express at least six protein kinase C (PKC)⁴ isoforms, namely α, β_I/II θ, η′, and ζ, which phosphorylate multiple proteins during platelet activation (Refs. 1 and 2 and references therein). In particular, the classical Ca²⁺-dependent PKC isoforms α and β_I/II play important but often unclear signaling roles in platelet responses (3, 4), and still little is known of the other atypical isoforms. In 1983 it was proposed, and later confirmed, that PKC activity in synergy with Ca²⁺ regulates the secretion of dense and α-granules following platelet stimulation with phospholipase C-stimulating agonists like collagen and thrombin (5–7). Secretion of ADP, fibrinogen, and other stored compounds enhances the activation process (8, 9). Phosphorylation by PKC may also contribute to the promotion of conformational changes of integrin αIIbβ3, required for fibrinogen binding and platelet aggregation (10, 11). This regulation is complex, because αIIbβ3 can also be activated in the absence of PKC (12). Furthermore, PKC and Ca²⁺ can synergize with another G_i-mediated pathway of αIIbβ3 activation (2), putatively via the G-protein exchange factor CalDAG-GEFI (13). Conversely, activated integrins themselves may stimulate PKC via outside-in signaling, to result in filopodial formation and platelet spreading (14, 15). Thus, Ca²⁺-dependent PKC isoforms seem to contribute to platelet aggregation in two different ways, directly via integrin phosphorylation and indirectly via secretion.

On the other hand, PKC can also have inhibitory effects on platelets. In the 1980s, it was reported that PKC inhibition reduces Ca²⁺ extrusion from platelets (5, 16) and increases (rather than suppresses) phospholipase C activity (17). This contrasts to the later suggestion that PKC stimulates an unidentified Ca²⁺ influx pathway involved in platelet activation (18). Another platelet-inhibiting effect of PKC is its ability to desensitize G-protein-coupled receptors (19). How these platelet-inhibiting effects of PKC interfere with its activating action is not understood.

During hemostasis and thrombosis, platelets have two distinct but additive functions in the formation of thrombi, aggregation and procoagulant activity. Once adhered to vascular collagen, platelets become activated via the signaling immunoreceptor, glycoprotein VI (GPVI), in interplay with the collagen receptor α2β1 (20–23). At high shear flow conditions, this results in αIIbβ3 activation, in the secretion of ADP, and the formation of thromboxane, which activate nearby platelets to form aggregates, e.g. by stimulation of G_q-coupled receptors (24, 25). In addition, GPVI activation enhances the coagulation process (26), because it triggers the Ca²⁺-dependent exposure of negatively charged phosphatidylserine (PS) at the platelet outer surface, where coagulation factors assemble to form thrombin (21, 22). Thrombin, in turn, further activates platelets and mediates fibrin clot formation. Recent in vitro and in vivo investigations with mice have highlighted the dual importance of GPVI in the thrombus-forming process by inducing intravascular platelet aggregation (27, 28) as well as procoagulant platelet formation (29). Concerning the role of PKC, it is only known that under flow conditions PKC contributes to the stable adhesion of platelets to collagen but not to the initial attachment (30).

Here we investigate the stimulating and inhibitory properties of all PKC isoforms both in the thrombus-forming process as a whole and in distinct platelet responses, using high affinity PKC inhibitors that were proved to be effective in whole blood. The results provide the first evidence for an overall balancing function of PKC, in controlling the proportions of aggregated and procoagulant platelets.

EXPERIMENTAL PROCEDURES

Antibodies and Materials—H-Phe-Pro-Arg chloromethyl ketone (PPACK) was obtained from Calbiochem, as were GF109203X (GF10, bisindolylmaleimide I), RO318220 (bisindolylmaleimide IX), RO318425 (RO31, bisindolylmaleimide X), and the non-maleimide compound Gö6976. Annexin A5 labeled with fluorescein isothiocyanate (FITC) was from Nexins Research. Fura-2 and Fluo-3 acetoxymethyl esters, Oregon Green 488-conjugated fibrinogen, and annexin A5 labeled with Alexa Fluor (AF)647 came from Molecular Probes. Bovine serum albumin (BSA), phorbol myristate acetate (PMA), and human α-thrombin were from Sigma; recombinant tissue factor was from Dade; Z-Gly-Gly-Arg aminomethylcoumarin (Z-GGR-AMC) was from Bachem; fibrillar type I collagen (Horm) was from Nycomed; and collagen-related peptide was from Baylor College of Medicine. Convulxin was purified as described (31). FITC-labeled PAC1 mAb was from BD Biosciences, and labeled anti-CD62 (anti-P-selectin) mAb was from Sanquin. Anti-murine GPVI mAb JAQ1 and anti-murine αIIbβ3 mAb JON/A were produced and modified as described (32), and where indicated, Fab fragments were generated and used in part for fluorescein labeling (33). Other materials were obtained from sources described before (34).

Animals—Control C57BL/6 mice of either sex were obtained from Charles River Breeding Laboratories. G_qα-deficient and wild type C57BL/6 mice were kindly provided by Dr. S. Offermanns (Heidelberg, Germany). Murine blood was collected under anesthesia by orbital puncture in 40 μm PPACK and 5 units/ml heparin, as described (33). Animal studies were approved by the local animal care and use committees.

Preparation of Platelet-rich Plasma and Washed Platelets—PRP was prepared from acetate-citrate-glucose (ACD) anticoagulated blood obtained from aspirin-free healthy volunteers or mice by differential centrifugation steps; PRP was used to prepare washed platelets. Human platelets were suspended in buffer composed of 5 mm Hepes, 136 mm NaCl, 10 mm glucose, 2.7 mm KCl, 2 mm MgCl₂, and 1 mg/ml BSA (pH 7.45). Murine platelets were washed with buffer consisting of 5 mm Hepes, 136 mm NaCl, 10 mm glucose, 2.7 mm KCl, 2 mm MgCl₂, 0.42 mm NaH₂PO₄, 1 mg/ml BSA (pH 7.45). Cells were counted and adjusted to the appropriate concentration.

Platelet Aggregation and Flow Cytometry—Platelet aggregation was measured by turbidometry with Chronolog aggregometers. For each sample, 300–500 μlof2 × 10⁸ platelets/ml in Hepes buffer or PRP were used. Samples were preincubated at 37 °C with vehicle (1% Me₂SO), 0.1–20 μm RO31, 0.1–20 μm GF10, or 0.025–20 μm Gö6976 for 10 min prior to stimulation with indicated agonists. Flow cytometry of unstirred activated and control platelets in buffer or plasma (2 × 10⁸/ml) was performed using primary FITC-labeled antibodies, as described for human (34) and mouse (35) platelets. For flow cytometry, platelets were preincubated with various doses of RO31, GF10, or Gö6976, as described above.

Measurement of PKC Activity—PKC activity of platelets in buffer, plasma, or blood was determined by measuring Ser phosphorylation of the modified PKC/cAMP-dependent protein kinase pseudo-substrate, RFARKGSLRQKNV (36), using a biotinylated mAb recognizing this phosphorylated form (Calbiochem). Washed platelets, PRP, or whole blood (normalized to 2 × 10⁸ platelets/ml) was treated with the indicated concentrations of RO31, GF10, or Gö6976 for 10 min and then stimulated as appropriate (PMA 100 nm, convulxin 50 ng/ml, ADP 10 μm). Samples of 1 × 10⁸ cells were centrifuged in a small volume of ice-cold phosphate-buffered saline, immediately sonicated on ice, and then further processed as indicated by the manufacturer. When required, PRP was isolated from blood prior to assaying. In control experiments, platelets treated with 20 μm RO31, GF10, or Gö6976 were stimulated with 1 μm iloprost and examined for cAMP-dependent phosphorylation of VASP (37). No influence of this treatment on the phosphorylation was detected (data not shown).

Measurement of [Ca²⁺]_i—Changes in cytosolic [Ca²⁺]_i were measured in washed suspensions of Fura-2-loaded platelets (2 × 10⁸/ml) by ratio fluorometry, as described (38). Final suspensions did not contain aspirin; activations were carried out in the presence of 1 mm CaCl₂. Control calibrations were performed when colored substances were present.

Thrombus Formation on Collagen under Flow—PPACK-anticoagulated human and mouse blood was used for perfusion experiments over collagen-coated coverslips, as described before (23). Blood was incubated for 15 min with the indicated inhibitors and/or fluorescent probes, placed in a syringe, and perfused over collagen at a shear rate of 150–1000 s^–1 for 4 min. After rinsing with Hepes buffer (pH 7.45) containing 2 mm CaCl₂ and 1 unit/ml heparin and fluorescently labeled annexin A5 (0.5 μg/ml), bright field phase-contrast images and nonconfocal fluorescence images of adherent platelets were recorded using a two camera system (22). A video recorder was connected to one of the cameras for recording dynamic information (stable platelet adhesion). Platelet surface coverage was analyzed with ImagePro software (Media Cybernetics) for phase-contrast images and Quanticell software (Visitech) for fluorescence images. At least 10 different microscopic fields were averaged per experiment (no image processing). To provide a measure of the proportion of procoagulant platelets independent of platelet deposition, the ratio of annexin A5-binding surface coverage to phase-contrast surface coverage was calculated and was termed procoagulant index (Pi) (22). Although the procoagulant area was slightly overestimated through fluorescent glare in the optics, Pi provided a means of distinguishing the treatments of procoagulant expression from those of platelet deposition and aggregation.

For single cell Ca²⁺ measurements under flow, blood was supplemented with 5% autologous Fluo-3-loaded platelets (23). Changes in fluorescence in collagen-adhered Fluo-3-loaded-platelets were obtained by high speed (5 Hz) recording of fluorescence images and off-line analysis of regions-of-interest representing single platelets. Calibration to [Ca²⁺]_i was by pseudo-ratioing to give F/F_o values and fixed calibration values (39). For quantitative data, traces from individual cells were superimposed, so that [Ca²⁺]_i initially increased after 3.0 s.

Two-photon Laser Scanning Microscopy—For two-photon laser scanning microscopy, coverslips with thrombi were observed with a Bio-Rad 2100 multiphoton system (40). Excitation was by a Spectra-Physics Tsunami Ti:Sapphire laser, tuned, and mode-locked at 800 nm, producing pulses of 100 fs wide (repetition rate 82 MHz). Excitation at 647 nm was by a parallel-placed red diode laser. Fluorescence was detected using appropriate wave-length filters (29). Thrombi in flow chambers, double-labeled with Oregon Green 488 fibrinogen and Alexa Fluor 647-annexin A5, were scanned at the end of perfusions. Optical sections were recorded in Kalman filtering mode; no further image processing was performed.

Thrombin Generation Measurement—PRP and platelet-free plasma collected on citrate were used to measure thrombin generation using the thrombogram method (34). Briefly, normalized PRP (1.5 × 10⁸ platelets/ml) or platelet-free plasma was preincubated with inhibitor for 15 min and then with collagen for 10 min. Samples of PRP or platelet-free plasma (4 volumes) were added to wells of a 96-well plate (Immulon 2HB; Dynex Technologies), containing 1 volume of 20 mm Hepes, 140 mm NaCl, 5 mg/ml BSA, and tissue factor (1 pm, final concentration). Coagulation was started by addition of 1 volume thrombin substrate, Z-GGR-AMC. Pre-warmed plates were inserted into a fluorescence well plate reader (Thermolab Systems) and processed at 37 °C. Fluorescence accumulation from cleaved AMC was measured, and first derivative curves of accumulation of fluorescence were generated; calibrations were performed with human thrombin (41).

Statistical Analysis—Differences between experimental groups were tested for significance with a nonparametric Mann-Whitney U test using the statistical package for social sciences (SPSS 11.0). p < 0.05 was considered to be statistically significant.

RESULTS

Partial Blockage of Platelet Aggregation and Integrin αIIbβ3 Activation by PKC Inhibition—To investigate the regulation of thrombus formation by PKC, we first tested several compounds with known selective PKC-inhibiting effects on protein phosphorylation and platelet responses. These compounds were added to platelets in the presence of plasma or in whole blood to establish their efficacy at physiological conditions. The bisindolylmaleimide derivatives, RO318425 (RO31) and GF10, interacting with the ATP-binding site of PKC, were effective in PRP at concentrations ≤20 μm (see below), but the related compound RO318220 had less effect in plasma (data not shown). Using soluble enzymes, it has been reported that RO31 and GF10 inhibit phosphorylation reactions of Ca²⁺-dependent and -independent PKC isoforms at concentrations of 10–40 nm (42, 43). We tested the kinase activity in lysates of washed platelets (2 × 10⁸ platelets/ml, ∼80 μm phospholipids), by measuring the Ser phosphorylation of an Ala → Ser-modified PKC pseudo-substrate, RFARKGSLRQKNV (36). With RO31 or GF10 added to platelet lysate, IC₅₀ values of pseudo-substrate phosphorylation were 0.1 and <0.2 μm, respectively. However, when added to platelets in plasma, these IC₅₀ values increased to 3.0 and 10 μm. In PRP, 20 μm RO31 or GF10 blocked phosphorylation with 85 ± 6 and 79 ± 5%, respectively (mean ± S.E., n = 4–5).

In both washed platelets (Fig. 1A) and platelets in plasma (not shown), RO31 and GF10 completely suppressed PMA-induced aggregation, as expected. We also tested the effects of these compounds on the aggregation induced by key agonists involved in thrombus formation. Either greatly suppressed platelet aggregation in response to the GPVI-dependent agonist, collagen, but only when added to washed platelets and not in PRP (Fig. 1A). Similar results were obtained for platelets stimulated with ADP, an agonist that acts via the G protein-coupled receptors P2Y₁ and P2Y₁₂ (data not shown). Dose-response curves of collagen-induced aggregation in buffer medium showed maximal inhibition from about 10 μm RO31 or GF10 (Fig. 1B).

Supporting experiments were performed with the non-maleimide compound, Gö6976, which has a high affinity for the classical PKC isoforms α and β_I (42). In vitro kinase measurements with lysed platelets showed that Gö6976 up to 20 μm inhibited only 40% of pseudo-substrate phosphorylation, which is compatible with its selectivity for PKC α/β_I. Gö6976 nearly completely inhibited PMA- and collagen-induced aggregation of washed platelets, the latter already at 1–2.5 μm (Fig. 1, A and B).

To understand the difference in effects of PKC inhibitors on platelet aggregation in buffer and plasma, flow cytometry was used to measure integrin αIIbβ3 activation. Inhibitor-treated platelets were stimulated with PMA, convulxin,⁵ or ADP. The appearance of activated epitopes on αIIbβ3 was detected with fluorescently labeled PAC1 mAb. In the presence of plasma at 20 μm (Fig. 1C), or in washed platelets at 10 μm (not shown), both RO31 and GF10 fully antagonized PMA-induced PAC1 binding, confirming that PKC-induced integrin activation was completely inhibited. In contrast, these compounds only partly reverted PAC1 binding in response to convulxin and did not affect PAC1 binding with ADP (Fig. 1C). When washed platelets were stimulated with PMA, convulxin, or ADP, RO31 reduced the PAC1 binding with 95 ± 3, 63 ± 6, or 5 ± 4%, respectively (mean ± S.E., n = 3). These results confirm the presence of two pathways of integrin activation (2, 44), only one of which appears to be PKC-dependent.

Platelet PKC Activity Is Required for Secretion—Expression of P-selectin at the platelet surface is a measure of secretion (exocytosis). Experiments with PRP showed that RO31 and GF10 greatly suppressed the high P-selectin expression induced by PMA or convulxin, and also the lesser P-selectin expression with ADP (Fig. 2A). Dose-response curves were generated to compare effects of the PKC inhibitors in the absence and presence of plasma. In washed platelets, the potency of these compounds to block P-selectin expression was quite similar to their anti-aggregatory activity, in the order of Gö6976 > RO31 ≥ GF10 (Fig. 2B, compare with Fig. 1B). In the presence of plasma, substantially higher concentrations were needed, and the potency of inhibition changed to RO31 > Gö6976 ≥ GF10 (Fig. 2B). With plasma present, 20 μm of all compounds gave complete or substantial inhibition. The relatively high concentrations needed for inhibition in plasma can be explained by binding of the hydrophobic compounds to plasma lipids, which impedes their incorporation into platelets. Typically, RO31 carries a charged primary amino group, which may decrease its binding to plasma proteins and hence favor uptake by platelets.

These inhibitor studies suggested that the contribution of PKC to collagen-induced aggregation in washed platelets is mediated by (fibrinogen and/or ADP) secretion. This was confirmed by the observation that addition of 2 mg/ml fibrinogen to washed RO31-treated platelets restored the collagen-induced aggregation response to >95% of control. Accordingly, the contribution of PKC isoforms to aggregation is partly via secretion of autocrine agents.

Platelet PKC Down-regulates Ca²⁺ and Procoagulant Responses—In thrombus formation, calcium signal generation via GPVI and G_qα-coupled receptors is essential for a number of platelet responses such as secretion, thromboxane formation, and procoagulant activity (26). We thus monitored the effects of PKC antagonism on Ca²⁺ responses in Fura-2-loaded platelets. Strikingly, platelet treatment with RO31 or GF10 substantially increased the Ca²⁺ signals by all tested agonists: ADP, convulxin, or thrombin (Fig. 3A). Dose-dependent measurements indicated that the increase with convulxin reached a maximum at around 10 μm RO31 or GF10 (not shown, but see below). Conversely, platelet treatment with PKC-activating PMA nearly completely (with ADP or thrombin) or substantially (with convulxin) blocked the Ca²⁺ responses, indicating that the overall effect of PKC activation is suppression of Ca²⁺ signaling. Time integrals of agonist-induced [Ca²⁺]_i increases were calculated as a measure of the extent of the Ca²⁺ signal and the Ca²⁺-dependent procoagulant response (38). Treatment with RO31 or GF10 resulted in a 3–6-fold increase in the [Ca²⁺]_i time integrals with ADP or thrombin, but in a smaller increase with convulxin (Fig. 3B). Surprisingly, treatment with Gö6976 (which has increased affinity for PKCα/β isoforms) was inactive on the Ca²⁺ responses with ADP or thrombin but nearly completely blocked the Ca²⁺ response with convulxin (Fig. 3B), under the same conditions where it completely inhibited secretion (Fig. 2B). The inhibitory effect of Gö6976 on P-selectin expression was verified in Fura-2-loaded platelets (not shown). Taken together, this suggested that the net Ca²⁺-increasing effect of PKC inhibition is not mediated by PKCα/β isoforms and furthermore that activation of these isoforms is required for convulxin-induced Ca²⁺ signaling.

The function of PKC in platelet procoagulant activity was established by measuring thrombin generation in PRP, which was triggered with tissue factor and CaCl₂. Under these plasma conditions, thrombin is formed at the surface of activated platelets with prolonged, high [Ca²⁺]_i, exposing procoagulant PS (34). Both GF10 and RO31, but not Gö6976, greatly enhanced the thrombin generation process in PRP and even enhanced the collagen-evoked (GPVI-mediated) increase in thrombin generation (Fig. 4). Control experiments indicated that neither RO31 nor GF10 influenced thrombin generation, if platelet activation was inhibited with cAMP-elevating iloprost (Fig. 4B). Controls without platelets indicated that the compounds did not interfere with the coagulation process itself (not shown). Together, these results point to a clear suppressive effect of PKC on platelet Ca²⁺ signaling and ensuing procoagulant response.

Dual Effects of PKC Inhibition on Shear-induced Thrombus Formation—The pro-aggregatory and procoagulant functions of platelets were simultaneously monitored in flow experiments, where whole blood was perfused over a collagen surface. At a moderately high shear rate of 1000 s^–1 (representative arterial shear rate) flow over collagen rapidly results in GPVI-dependent formation of platelet aggregates and exposure of procoagulant PS (detected with FITC-annexin A5) (33). Under this condition, pretreatment of blood with 20 μm RO31 or GF10, i.e. concentrations that substantially blocked PKC pseudo-substrate phosphorylation, reduced platelet adhesion with 30–40% and nearly abolished aggregate formation (Fig. 5A). Surface area coverage of the platelets was more than halved, whereas coverage with PS-expressing platelets was significantly increased with RO31 but not with GF10 (Fig. 5B). The ratio of PS-exposing surface coverage to total surface coverage with platelets was calculated as the Pi (22). The Pi was used as a means of determining the relative procoagulant expression of the deposited platelets (see “Experimental Procedures”). The Pi increased significantly with 20 μm RO31 or GF10 from 0.19 to about 1 (Fig. 5B, lower panel), thus pointing to increased procoagulant activity. This parameter increased with the inhibitor dose, e.g. with 5 μm RO31, GF10, and also Gö6976, and the Pi was raised from 0.15 to 0.31, 0.32, or 0.42, respectively, particularly because of a reduction in platelet deposition (n = 2).

To directly compare αIIbβ3 activation and PS exposure under flow conditions, perfusions were performed in the presence of Oregon Green 488-labeled fibrinogen and Alexa Fluor 647-labeled annexin A5. Two-colored images were recorded using the high sensitivity of two-photon fluorescence microscopy. Treatment with RO31 considerably reduced the fibrinogen binding to collagen-adhered platelets, and again annexin A5 binding was less affected (Fig. 5C). By addition of autologous, Fluo-3-loaded platelets to the blood (22), changes in [Ca²⁺]_i of platelets upon adhesion under flow were evaluated. Control platelets had peak-shaped rises in [Ca²⁺]_i followed by a prolonged continuous rise; treatment with RO31 resulted in Ca²⁺ responses of similar shape but increased amplitude (Fig. 6). At 30 s after the start of the Ca²⁺ signal, vehicle- and RO31-treated platelets showed averaged [Ca²⁺]_i of 470 ± 33 and 675 ± 98 nm, respectively (p = 0.011, n = 31–38). Because integrin activation, aggregate formation (via ADP secretion), Ca²⁺ signaling, and PS exposure are all GPVI-mediated events in this flow model (21, 40), these data together point to a dual effect of PKC inhibition as follows: on the one hand, reduction in secretion and integrin activation, leading to suppressed aggregation, and on the other hand, enhancement of Ca²⁺ mobilization and PS exposure.

Partial Role of PKC in Murine Platelet Aggregation and Integrin Activation—Effects of PKC inhibition were evaluated in wild type mouse blood. With mouse PRP, it was confirmed that 20 μm RO31 treatment inhibited PMA-induced pseudo-substrate phosphorylation by 89%. Both RO31 and GF10 inhibited the aggregation of washed mouse platelets induced by PMA or collagen peptide (Fig. 7A). In the presence of plasma, only PMA-induced aggregation was abolished, whereas aggregation with other agonists was no more than delayed, i.e. similar to the human situation. Flow cytometric analysis was performed using platelets stained with fluorescent labeled JON/A mAb, which displays an increased binding to activated mouse αIIbβ3. Similar to RO31, GF10 antagonized PMA- and collagen peptide-induced αIIbβ3 activation but had only small effects on ADP- and thrombin-induced αIIbβ3 activation (Fig. 7B). Again, these compounds increased the procoagulant activity of mouse platelets, as apparent from a left-ward shift in thrombin generation experiments (data not shown).

Effect of PKC Inhibition on Murine Thrombus Formation in the Absence of Autacoid G_q Signaling—Flow experiments over collagen were performed with blood from wild type control mice and G_qα-deficient mice, to distinguish between direct and indirect (secretion-dependent) effects of PKC, because these knock-out mice have abolished responses toward the autacoids ADP and thromboxane via their P2Y₁ and TP receptors, respectively (21, 25). Flow of G_qα-deficient blood did not reduce platelet adhesion to collagen but resulted in greatly diminished thrombus formation with only small, two-layered aggregates remaining (Fig. 8A). The adhered knock-out platelets still displayed a procoagulant response by exposing PS. Nevertheless, PKC inhibition with GF10 further reduced the aggregation of both wild type and G_qα-deficient platelets and, in comparison, increased the PS exposure and as result a significantly increased procoagulant index Pi (Fig. 8B). Thus, even in the absence of autacoid G_qα signaling, where aggregation was suppressed, PKC still positively regulated aggregation and negatively regulated the procoagulant response.

DISCUSSION

The present results provide a first unifying concept to understand the seemingly contradictory functions of PKC isoforms in platelets. They indicate that in platelets PKC phosphorylation has a dual role in controlling collagen-induced thrombus formation. On the one hand, inhibition of PKC with RO31 or GF10 inhibits secretion and incompletely down-regulates αIIbβ3 activation, which results in decreased aggregate formation, under conditions when secretion is a limiting factor, i.e. in washed platelets. Importantly, inhibition of PKC with RO31 or GF10 in whole blood also abrogated thrombus formation under conditions of flow. On the other hand, PKC inhibition increases agonist (GPVI)-induced Ca²⁺ responses and Ca²⁺-dependent PS exposure with (and as a result of) increased coagulation activity, as was established in thrombin generation experiments. Accordingly, in human and murine thrombus formation, PKC activation is the first signaling pathway identified that can balance the two principal roles of platelets, i.e. aggregate formation and coagulation stimulation. Thus, PKC simultaneously increases the pro-aggregatory activity of platelets and suppresses the procoagulant properties of platelets (Fig. 9). This dual role of PKC was still apparent when most of the autocrine ADP- and thromboxane-induced thrombus formation, via P2Y₁ and thromboxane A₂ receptor and protease-activated receptor signaling, was suppressed in platelets lacking the G_qα subunit.

The two bisindolylmaleimide inhibitors used in this paper, RO31 and GF10, were selected because of their activity on platelets in the presence of plasma or whole blood. In lysates they bind with nanomolar affinity to the classical Ca²⁺-dependent isoforms PKCα, and β_I/II is absent in platelets but also with somewhat lower affinity to other isoforms (43). Dose-response curves of kinase activity, secretion, and platelet aggregation indicated that higher, 10–20 μm, concentrations of either compound were needed to block these responses, with a shift to the right in the presence of plasma. This compares well with the literature, where the same compounds have also been used at 10^–5 m concentrations to block PKC-dependent processes in suspensions of (washed) platelets (4, 15, 30, 45).

Various authors provide evidence that especially Ca²⁺-dependent PKC activity mediates αIIbβ3 activation and secretion in platelets (3, 44, 46), which implicates that the other nonclassical isoforms δ, θ, η′, and ζ (lacking the Ca²⁺-binding C2 region) may be less markedly involved in these processes. Although in this paper we did not aim to resolve the functions of specific isoforms, some remarkable results were obtained with the high affinity PKC α/β inhibitor Gö6976. It inhibited secretion and aggregation in a similar way to RO31 and GF10 and, in agreement with this, reduced thrombus formation under flow. On the other hand, Gö6976 was without effect on Ca²⁺ signaling or procoagulant activity in response to G-protein-coupled receptor agonists (ADP and thrombin), and it blocked GPVI-induced Ca²⁺ signals (convulxin). These data indicate that in platelets the stimulatory effects of PKC are regulated in a more distinct way than the suppressive effects of PKC, and they suggest that the latter are mediated by one or more nonclassical PKC isoforms. Mechanistically, these effects of Gö6976 are well explained by the recent observation that PKCα has a negative regulatory effect on Src kinases (45), the activity of which is of key importance in GPVI-induced signal transduction (23), although other authors propose that PKCδ is required for collagen-induced phospholipase C activation (47). In HEK293 and COS1 cells, PKCα is recognized as a quickly responding element to local elevation in [Ca²⁺]_i (48). Whether this is still the case in platelets remains to be investigated. At this moment, still very little is clear on the functions of the various PKC isoforms in platelets, and we consider that preferentially experiments with knock-out animals are needed to clarify this.

Both in human and mouse flow experiments, we found a reduced deposition on collagen with PKC-inhibited platelets, which was largely a consequence of greatly diminished aggregate formation. Close examination of the recorded video movies indicated that also the frequency of stable platelet adhesion was diminished under this condition (data not shown). This is in agreement with the earlier reported stabilizing effect of PKC on platelet adhesion to collagen, which is mechanistically still unexplained, but was found to be independent of Syk and focal adhesion kinase (30). We consider that this may be mediated by a decreased GPVI-induced αIIbβ3 activation in the absence of PKC, thus reducing irreversible platelet binding to von Willebrand factor that is bound to collagen. However, others have observed that activated αIIbβ3(e.g. by von Willebrand factor-induced GPIb signaling) in turn stimulates PKC, which points to the importance of positive feedback loops between αIIbβ3 and PKC (49).

Both in the absence and presence of flow, we observed that not all platelet aggregation was abolished in the absence of PKC activity. This is in agreement with optical aggregation measurements from others and clearly points to PKC-independent signaling mechanisms in the regulation of αIIbβ3 activation (44). Recently, evidence has been obtained for the existence of another diacylglycerol-dependent (PKC-independent) pathway, involving the Ras family-regulating protein, CalDAG-GEFI, as a modulator of integrin activation and platelet aggregation (13). In platelets, it likely acts as a regulator of Rap1b, which has also been implicated in integrin activation (50). In the absence of PKC activity, receptor signaling via both G_q and GPVI still ensures phospholipase C activity, diacylglycerol formation, and Ca²⁺ release, which allows CalDAG-GEFI activation. This can provide an alternative pathway to integrin activation, not indicated in Fig. 9. The present data suggest that in GPVI-induced thrombus formation, this alternative pathway is of only limited importance. In summary, we conclude that PKC isoforms have a dual controlling role in thrombus formation by permitting secretion and integrin activation required for platelet aggregation, while suppressing Ca²⁺-dependent PS exposure and the coagulation process. The PKC pathway thereby balances the pro-aggregatory and procoagulant properties of thrombi.