Reassessment of Protein Tyrosine Phosphorylation in Thrombasthenic Platelets: Evidence That Phosphorylation of Cortactin and a 64-kD Protein Is Dependent on Thrombin Activation and Integrin α_IIbβ₃

PROTEIN TYROSINE phosphorylation is contemporary to platelet activation, aggregation, and spreading and is involved in platelet cytoskeletal reorganization.^1-5 Platelet tyrosine phosphorylation has been studied in detail during platelet aggregation, and phosphorylated polypeptides have been subdivided into three groups according to the three steps leading to platelet aggregation⁵: early activation (no aggregation), oligomerization of the fibrinogen receptor (α_IIbβ₃ integrin, also termed GPIIb-IIIa), and platelet-to-platelet adhesion, ie, aggregation. In early activation, the phosphorylated proteins identified include the tyrosine kinase Syk,6 the p21ras GTPase-activating protein (GAP), and two associated proteins of relative mass 59 kD and 68 kD,7 the cortical actin-associated protein cortactin,4,7 mitogen-activated protein kinases, as well as several unidentified proteins.5 The second step corresponds to tyrosine phosphorylated proteins associated with α_IIbβ₃ oligomerization following fibrinogen binding and include the tyrosine kinase Syk,6 and unidentified 140-kD and 50- to 68-kD polypeptides.3,8 Finally, aggregation correlates with tyrosine phosphorylation of an unidentified 100- to 105-kD doublet1-5 and the tyrosine kinase FAK or focal adhesion kinase, as well as with an increase in tyrosine phosphorylation of c-src.8 9 

The first experimental evidence correlating platelet aggregation with platelet protein tyrosine phosphorylation came from the study of platelets from Glanzmann thrombasthenia patients.2 Glanzmann thrombasthenia is an inherited bleeding disorder characterized by defective platelet aggregation consecutive to the failure of patients' platelets to bind fibrinogen after agonist stimulation.10 The defect is due to a quantitative or a qualitative defect in α_IIbβ₃ integrin.11 α_IIbβ₃, like other members of the integrin family, is a heterodimer composed of two subunits, α_IIb (GPIIb) (140 kD) and β₃ (GPIIIa) (95 kD), each encoded by a single gene.12 A fast growing number of mutations have recently been identified in Glanzmann thrombasthenia (for review see Bray13 ). They have been classified in three types^11,13,14: in type I, platelets have less than 5% the normal amount of α_IIbβ₃, are unable to retract fibrin clots and have no intracellular fibrinogen; in type II, platelets exhibit 5% to 20% the normal amount of α_IIbβ₃,14 15 bind fibrinogen, do retract fibrin clots, and possess intracellular fibrinogen; variant types have normal or near normal amounts of α_IIbβ₃ unable to bind fibrinogen, while intracellular fibrinogen and fibrin clot retraction vary considerably from case to case. Thus α_IIbβ₃-dependent platelet functions involving cytoskeletal reorganization are altered differently depending on Glanzmann type. We, therefore, decided to determine if differences in tyrosine phosphorylation patterns might correlate different Glanzmann types.

Here we report that in conditions of platelet aggregation, the same alterations in protein tyrosine phosphorylation patterns are found in platelets from three different types of Glanzmann thrombasthenia patients. This confirms the central role of functional α_IIbβ₃ in the initiation of protein tyrosine phosphorylation in platelet aggregation. This study not only confirms the absence of phosphorylation of a 100- to 105-kD doublet, but also demonstrates a lower overall phosphorylation rate of a 77- to 80-kD doublet and a 64-kD band. We identified the 77- to 80-kD doublet by immunoprecipitation as the cortical actin-binding protein named cortactin. A similar pattern of normal early and defective late phosphorylation of cortactin and the 64-kD band were observed when aggregation of control platelets was inhibited either by specific functional antagonists of platelet α_IIbβ₃ integrin, or in the absence of stirring. We conclude that tyrosine phosphorylation of cortactin and the 64-kD band are dependent both on thrombin activation of platelets and on α_IIbβ₃ engagement in aggregation. The relationship between these tyrosine phosphorylated substrates and involvement of specific kinase and/or phosphatase activities and the cytoskeletal rearrangements associated with thrombin activation and aggregation are discussed.

Reagents and suppliers were as follows: protein A-Sepharose CL-4B, phenylmethanesulphonyl difluoride (PMSF), sodium orthovanadate (Na₃VO₄), leupeptin, NP40, EGTA, Arg-Gly-Asp-Ser (RGDS) peptide, bovine thrombin, and cytochalasin D (Sigma Chemical Co, St Louis, MO); 10E5 anti-α_IIbβ₃ monoclonal antibody (Centocor, Inc, Malvern, PA); sheep antimouse horseradish peroxidase-conjugated IgG, enhanced chemiluminescence (ECL) reagents, hyperfilm-ECL (Amersham International, Buckinghamshire, UK); metrizamide (Nycomed Pharma AS, Oslö, Norway); electrophoresis reagents (Euromedex, Souffelweyersheim, France); antiphosphotyrosine mouse monoclonal antibody 4G10, anticortactin (anti-p80/85) mouse monoclonal antibody (Upstate Biotechnology Inc, Lake Placid, NY); affinipure rabbit antimouse IgG (H+L) polyclonal antibody (Jackson Immunoresearch Laboratories, Inc, West Grove, PA); sodium fluoride (NaF) (Aldrich Chemical Company, Inc, Milwaukee, WI); EDTA (Prolabo, Paris, France); Na-pyrophosphate (Merck, Darmstadt, Germany) nitrocellulose membranes (Schleicher and Schuell, Cera-Labo, Dassel, Germany).

Glanzmann thrombasthenia patients.The three patients studied have been characterized previously. Glanzmann type I platelets (patient CB) had no α_IIb (GPIIb) nor β₃ (GPIIIa) detectable by Western blotting due to an homozygous mutation carried by the GPIIIa gene.16 Platelets from Glanzmann type II patient (patient MH) exhibited approximately 10% α_IIbβ₃,15 and were able to retract fibrin clot.14 The mutation(s) involved has not been characterized at the molecular level. Platelets of Glanzmann variant type (patient RP) exhibited approximately 50% the normal amount of α_IIbβ₃, which was unable to expose fibrinogen binding sites (also often referred to as α_IIbβ₃ “activation”) during platelet activation, due to a Ser752→Pro substitution in the cytoplasmic tail of β₃.17 

Platelet preparation.Platelets were prepared as previously described.18 Briefly, blood was obtained with informed consent from healthy donors or patients with Glanzmann's thrombasthenia. Blood was anticoagulated with ACD-C (124 mmol/L Na₃-citrate, 130 mmol/L citric acid, and 110 mmol/L glucose: 1 volume for 9 volumes of blood) and then centrifuged for 10 minutes at 100g to obtain platelet-rich plasma (PRP). Platelets were isolated from plasma on a metrizamide gradient.18 The “metrizamide gradient platelets” thus obtained were resuspended in an aggregation buffer containing 10 mmol/L HEPES (pH 7.4), 140 mmol/L NaCl, 3 mmol/L KCl, 0.5 mmol/L MgCl₂, 5 mmol/L NaHCO₃, 10 mmol/L glucose and adjusted to a density of 4 × 10⁸ cells/mL.

Aggregation assays.Platelets (0.4 mL) were preincubated without stirring at 37°C for 5 minutes. Platelet aggregation was then initiated by the addition of 0.05 U/mL bovine thrombin with constant stirring (1,200 rpm) in an aggregometer cuvette (Chronolog dual beam aggregometer). Aggregation was measured and expressed as the percent change in light transmission, 100% referring to the blank sample (buffer without platelets). In some experiments platelets were preincubated for 10 minutes at 37°C with 100 μmol/L cytochalasin D (4 μL 10 mmol/L stock solution) or an equivalent volume of DMSO.

Immunoblotting.Samples were processed and subjected to immunoblotting as described previously.19 Briefly, after incubation, platelet suspensions were adjusted to 2% (wt/vol) sodium dodecyl sulfate (SDS) and 1 mmol/L EDTA and heated at 60°C for 30 minutes in the presence of 5% (vol/vol) 2-mercaptoethanol. Proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE), on a 7% polyacrylamide gel. Proteins were transferred onto a nitrocellulose filter membrane by semidry transfer (Semi-Phor; Hoefer Scientific Instruments, San Francisco, CA). Nonspecific binding was blocked in 5% (wt/vol) low fat powder milk in buffer (20 mmol/L Tris, 137 mmol/L NaCl, 0.1% (vol/vol) Tween-20, pH 7.6) before exposure of the membrane to primary antibodies for 60 minutes at room temperature. After washing steps, nitrocellulose strips were incubated with secondary sheep antimouse immunoglobulins bound to peroxidase. Immunoreactive bands were visualized by chemiluminescence followed by autoluminography using the ECL kit (Amersham International, Buckinghamshire, UK), according to the manufacturer's instructions.

Immunoprecipitation.After stimulation with thrombin, 0.4 mL, platelet suspension was lyzed with 0.1 mL of lysis buffer (sample buffer, pH 7.4, containing 1% (vol/vol) NP40, 1 mmol/L EDTA, 1 mmol/L EGTA, 3 mmol/L Na₃VO₄, 50 mmol/L NaF, 5 mmol/L Na-pyrophosphate, 20 μmol/L leupeptin). After 20 minutes at 4°C and centrifugation (15 minutes at 10,000g at 4°C), soluble fractions were collected and then incubated for a minimum of 2 hours at 4°C with 20 mg of protein A-Sepharose, 0.05 mg/mL rabbit antimouse IgG, 5 μg/mL mouse anticortactin antibody. After four washing steps in lysis buffer, immunoprecipitates were eluted by boiling in SDS-PAGE sample buffer containing 2% (wt/vol) SDS and 1 mmol/L EDTA. Tyrosine phosphorylated proteins were immunoprecipitated as described above, except that soluble fractions were incubated for a minimum of 2 hours at 4°C with 20 mg of protein A-Sepharose and 5 μg/mL mouse antiphosphotyrosine antibody.

Kinetics of thrombin-induced tyrosine phosphorylation of protein substrates in platelets from controls and from three patients of different Glanzmann thrombasthenia types.The time-course pattern of tyrosine phosphorylation of platelet proteins induced during thrombin aggregation as analyzed by phosphotyrosine-specific Western blotting was compared between three controls and three patients exhibiting Glanzmann thrombasthenia type I (Fig 1A), type II (Fig 1B), and variant type (Fig 1C). The earliest bands were a strong 60-kD band, already detected in resting platelets and corresponding to src,1 and a 64-kD, a 77-80–kD, and a 130-kD band detectable within 30 seconds; the latest was a 100- 105–kD doublet, which, as documented by others,1-3 correlated with significant rates of aggregation (over 24%, see Fig 1). The tyrosine phosphorylation patterns of platelets from all three thrombasthenic types differed significantly from the controls and showed identical alterations. The most affected bands were the 100-105–kD doublet, which was completeley absent, confirming previous reports.1-3 Second the 64-kD band and the 77-80–kD doublet were strongly reduced (a finding similar to previous reports7,8 ), but at essentially late time points and with different kinetics. The phosphorylation of the 77-80–kD doublet was normal until 30 seconds of thrombin aggregation (threefold to fivefold enhancement over the level of resting platelets, Table 1), but remained steady (or even tended to slightly decrease in some experiments) at later aggregation time points, as opposed to controls (fourfold to fivefold relative enhancement at 240 seconds v nearly 20-fold in controls, Table 1). Phosphorylation of the 64-kD band was normal up to 60 seconds, then enhanced (as opposed to the 77-80–kD doublet), but to a lesser extent than the control: 10-fold increase at 240 seconds aggregation, compared with nearly 30-fold for controls (Table 1). The 130-kD band appeared decreased in some experiments. There was no major change in phosphorylation for src in the three thrombasthenics studied as previously reported.1-5 

Our results correlate defective α_IIbβ₃ and aggregation in thrombasthenic platelets and alteration in extent and/or kinetics of tyrosine phosphorylation of several polypeptides. In addition to confirming previous observations, our results obtained with three thrombasthenic patients exhibiting three different α_IIbβ₃ mutations further confirm the key role of α_IIbβ₃ in protein phosphorylation of platelets. Moreover, our results are consistent with tyrosine phosphorylation of the 77-80–kD doublet and the 64-kD band being dependent on both thrombin activation and functional α_IIbβ₃. To extend these data further, we assessed the phosphorylation of these substrates in control platelets in conditions known to inhibit fibrinogen binding and/or aggregation.

Tyrosine phosphorylation of the 77-80–kD inhibited by the RGDS peptide and the monoclonal antibody, 10E5, two competitive inhibitors of α_IIbβ₃. To confirm that altered tyrosine phosphorylation of the 64 kD and the 77-80–kD doublet in thrombasthenic platelets was due to the defect in α_IIbβ₃, we blocked the binding of adhesive ligands such as fibrinogen to α_IIbβ₃ in control platelets by two competitive inhibitors of α_IIbβ₃ ligand binding, the tetrapeptide RGDS,20 and 10E5, a monoclonal antibody specific for α_IIbβ₃.21 As shown in Fig 2A and B, inhibition of aggregation by 1 mmol/L RGDS or 10 μg/mL 10E5 generated a pattern of tyrosine-phosphorylated polypeptides similar to thrombasthenic platelets: the 77-80–kD doublet reached a plateau as early as 30 seconds after thrombin activation, and the 64-kD kinetics of phosphorylation appeared decreased (though to a comparatively lower extent) compared with control. This confirmed that tyrosine phosphorylation of these substrates is partially dependent on engagement of α_IIbβ₃ in aggregation. Taken together both these experiments and analyses of thrombasthenic platelets suggest that the 77-80–kD doublet is subject to two waves of tyrosine phosphorylation: A first wave is contemporary to the early steps of platelet activation and a second wave is dependent on α_IIbβ₃ engagement in aggregation. The earlier onset of phosphorylation of the 77-80–kD doublet suggests it is involved in the first steps of platelet activation, such as shape change, while its strong dependence on aggregation is consistent with its involvement in “post-ligand occupancy” α_IIbβ₃-dependent events. Because both are contemporary of actin cytoskeletal rearrangements, which are essential to platelet aggregation,22 we focused our efforts on the 77-80–kD doublet.

Tyrosine phosphorylation of the cytoskeletal protein cortactin is altered in thrombasthenic platelets.Cortactin is a cytoskeletal tyrosine phosphorylated protein that migrates as a doublet within the 80-kD range and was reported previously to be tyrosine-phosphorylated in activated platelets.5,7 23 This prompted us to examine the possibility that cortactin corresponded to or was part of the 77-80–kD doublet seen in our system. We, thus, immunoprecipitated normal control as well as thrombasthenic platelet lysates with a monoclonal antibody specific for cortactin followed by antiphosphotyrosine Western blotting (Fig 3), and conversely, immunoprecipitated phosphotyrosine proteins followed by Western blotting using the anticortactin antibody. Using both approaches, a 77-80–kD doublet was obtained in control platelets, which reached a premature phosphorylation plateau in thrombasthenic platelets after 30 seconds aggregation. The same pattern was obtained when RGDS-inhibited control platelets were used instead of thrombasthenic platelets. This experiment thus confirms that early and late tyrosine phosphorylation of cortactin are independent from and dependent on functional α_IIbβ₃, respectively.

Tyrosine phosphorylation of platelet cortactin was examined in a previous study7 and was reported comparable between control and thrombasthenic platelets, a significant difference with the present study where cortactin phosphorylation is weak in thrombasthenic platelets, presumably due to its occurence only during the early steps of activation. The reason for this difference is unclear. This may be due to technical differences, in particular, as Fox et al7 analyzed subcellular fractions, from which quantitation of cortactin phosphorylation may be difficult to assess reproducibly.

The thrombin-induced tyrosine phosphorylation of cortactin and the 64 kD are differentially dependent on aggregation and actin polymerization.Alteration of the kinetics of tyrosine phosphorylation of cortactin (and the 64-kD protein) in thrombasthenic platelets, as well as in control platelets, in the presence of α_IIbβ₃ inhibitors is secondary to the absence of binding of fibrinogen to α_IIbβ₃ or to the absence of aggregation per se. To distinguish between these two possibilities, we examined the tyrosine phosphorylation pattern of thrombin-activated control and type I thrombasthenic platelets in the absence of stirring. These are conditions known to prevent platelet aggregation without interfering with fibrinogen binding to α_IIbβ₃,4 as opposed to calcium chelators. In unstirred conditions (Fig 4A), cortactin tyrosine phosphorylation remained steady from 30 to 120 seconds and was comparable in normal and thrombasthenic platelets. This suggests that the α_IIbβ₃-dependent phase of phosphorylation of cortactin is mainly due to aggregation itself rather than fibrinogen binding to α_IIbβ₃ and/or clustering of fibrinogen-bound α_IIbβ₃. In addition, 64 kD phosphorylation increased after stirring of control platelets, but not of thrombasthenic platelets (data not shown) confirming α_IIbβ₃-dependent phosphorylation of 64 kD triggered by aggregation.

Fibrinogen binding to α_IIbβ₃ as triggered by an α_IIbβ₃-activating Fab antibody (LIBS 6) in the absence of platelet activation led to the tyrosine phosphorylation of a 140-kD substrate along with a group of proteins between 68 and 50 kD.8 Likewise, several experiments demonstrated the presence of a band within the 70- to 80-kD range in these conditions, ie, only when fibrinogen binding and α_IIbβ₃ oligomerization (no activation and no aggregation) were allowed. It is possible that our 64-kD band and cortactin correspond to the 68-kD and the 70- to 80-kD bands seen in this system. Studies using LIBS6 will be required to test this possibility.

Because cortactin and the 64-kD protein are tyrosine-phosphorylated in a manner dependent, at least in part, on aggregation, we tested their dependence on aggregation-dependent cytoskeletal assembly. We inhibited platelet cytoskeleton assembly during aggregation by preincubation with 100 μmol/L cytochalasin D²⁴ (Fig 4B). As previously reported,8 9 phosphorylation was markedly reduced for the 130-kD band or absent for the 77-80–kD and the 100-105–kD doublets. Both src and the 64-kD band were only partially affected. This pattern is in fact indistinguishable from thrombasthenic platelets after 2 minutes aggregation (see Fig 1). This confirms that tyrosine phosphorylation of cortactin and (to a lesser extent) of the 64-kD band is dependent on actin cytokeleton. It is likely that the cytochalasin-insensitive fraction of phosphorylated 64 kD corresponds to the 64 kD undergoing phosphorylation in thrombasthenic platelets (and in unstirred control platelets). It is, therefore, tempting to speculate that α_IIbβ₃-dependent and -independent phosphorylated 64 kD correspond to actin cytoskeleton-dependent and -independent 64 kD, respectively. Whether these are indeed two pools of the same molecule or two (at least) different molecules remains to be elucidated.

What is the significance of the tyrosine-phosphorylation of cortactin? Among substrates undergoing α_IIbβ₃-dependent tyrosine phosphorylation, cortactin is one of the earliest, occurring in parallel with platelet shape change and the formation of filopods, which are dependent on cytoskeletal rearrangements involving mainly depolymerization of the submembranous cytoskeleton and polymerization of cytoplasmic actin.7,24 The early tyrosine phosphorylated cortactin (which is phosphorylated in thrombasthenic platelets) may participate in this actin cytoskeleton, the remainder (nonphosphorylated in thrombasthenic platelets) being incorporated in the later α_IIbβ₃-dependent actin bundles.24 Our view is consistent with the two pools of cortactin suggested by Ozawa et al,25 based on experiments of cosedimentation of cortactin with actin in thrombin-stimulated platelets. However, because these investigators did not evaulate the tyrosine phosphorylation status of their cosedimented fractions, future studies are required to test this hypothesis.

The kinase and/or phosphatase activities responsible for cortactin and the 64-kD phosphorylation are unknown. With regard to cortactin, Src has recently been demonstrated to be associated with tyrosine phosphorylated cortactin.26 In addition src exhibits an early and a late activity during platelet aggregation⁵: for these reasons src appears as a potential candidate for the kinase activities involved in the two waves of phosphorylation of cortactin. Another candidate tyrosine kinase potentially involved in cortactin phosphorylation, is the pp72^syk tyrosine kinase, which follows similar kinetics and subcellular redistribution: syk is activated (and undergoes autophosphorylation) in part by thrombin activation and in part by platelet aggregation,6 and undergoes a comparable two-step cytoskeletal translocation.22 Moreover, a direct association between syk and cortactin has been demonstrated in another cell type (the erythroleukemia cell line K562).27 Finally, a recent report has shown α_IIbβ₃-dependent tyrosine phosphorylation in the absence of platelet activation, of several substrates, one of them identified as syk in its activated state²⁸. This observation suggests a close relationship between a_IIbb₃ and syk (presumably activated directly by α_IIbβ₃ after subunit dissociation by external calcium chelation). Syk is thus a reasonable candidate kinase for coupling α_IIbβ₃ to phosphorylation of cortactin and possibly 64 kD (likely to correspond to the p62 and p68 substrates described in Negrescu and Siess28 ). A third candidate kinase is FAK or the focal adhesion kinase, the activity of which is completely dependent on α_IIbβ₃ engagement in aggregation (for review, see Shattil4 ). Thus, FAK would be involved in the late phosphorylation events of cortactin and possibly of the 64 kD.

Low phosphorylation rates of the 64 kD and cortactin in thrombasthenic platelets or in nonaggregated platelets may be due to nonactivation of specific tyrosine kinases, but a role for activated tyrosine phosphatases cannot be ruled out. Consistent with this hypothesis is the fact that in several experiments with thrombasthenic platelets, a decreased phosphorylation rate of cortactin was observed at late aggregation times (Fig 1A, for example). This model thus implies that in platelets, a tyrosine phosphatase activity exists, which is under the positive control of thrombin and is negatively regulated by α_IIbβ₃. This activity would be distinct from tyrosine phosphatase activities such as PTP1B, which is positively regulated by α_IIbβ₃.29 However, direct evidence for such an α_IIbβ₃-negatively regulated tyrosine phosphatase activity awaits future investigations.

We thank Dr B.S. Coller for providing the anti-α_IIbβ₃ mouse monoclonal antibody 10E5.