Eph kinases and ephrins support thrombus growth and stability by regulating integrin outside-in signaling in platelets

Abstract

The ability of activated platelets to adhere to each other at sites of vascular injury depends on the integrin α_IIbβ₃. However, as aggregation continues, other signaling and adhesion molecules can contribute as well. We have previously shown that human platelets express on their surface the Eph receptor kinases EphA4 and EphB1 and the Eph kinase ligand ephrinB1. We now show that EphA4 is physically associated with α_IIbβ₃ in resting platelets, increases its surface expression when platelets are activated, and colocalizes with α_IIbβ₃ at sites of contact between platelets. We also show that Eph/ephrin interactions can support the stable accumulation of platelets on collagen under flow and contribute to postengagement “outside-in” signaling through α_IIbβ₃ by stabilizing platelet aggregates and facilitating tyrosine phosphorylation of the β₃ cytoplasmic domain. β₃ phosphorylation allows myosin to bind to α_IIbβ₃ and clot retraction to occur. The data support a model in which the onset of aggregation permits Eph/ephrin interactions to occur, after which signaling downstream from ephrinB1 and its receptors favors continued growth and stability of the thrombus by several mechanisms, including positive effects on outside-in signaling through α_IIbβ₃.

Thrombus formation is a process by which freely circulating platelets are activated by collagen or thrombin, adhering first to the injured vessel wall and then to each other, concomitant with the establishment of a fibrin clot. The formation of a thrombus can be an appropriate response to vascular injury or a pathological response to disease. In either setting, the recruitment of platelets into an expanding thrombus is normally supported by soluble agonists, such as thromboxane A₂ and ADP, which, along with thrombin and collagen, initiate the signaling events that activate the integrin α_IIbβ₃ (1). The activated integrin then binds to fibrinogen or von Willebrand factor, bringing platelets into contact with each other. How such contacts are maintained once they are formed is not fully understood, but increasing evidence suggests that the boundaries between adjacent platelets, like those between epithelial cells, are populated by signaling and adhesion molecules that help to sustain platelet activation and cement contacts between platelets (2). Some of the contact-dependent signaling is mediated by the integrins themselves (3), but other candidates exist as well, including the Eph kinases and their ligands, known as ephrins.

Although Eph kinases and ephrins are best known for their role in neuronal organization (4, 5) and as markers distinguishing arteries from veins during vasculogenesis (6), we have shown that human platelets express two Eph kinases, EphA4 and EphB1, and at least one ligand, ephrinB1 (7, 8). The Eph kinases are a large family of receptor tyrosine kinases with an extracellular ligand binding domain and an intracellular tyrosine kinase domain. The Eph B subfamily is distinguished from the Eph A subfamily by an insertion within the extracellular domain that helps to define the ligand preferences for the receptor (9). Ephrins are membrane-attached molecules that fall into two groups based on their mode of attachment. Ephrin A family members have a glycophosphati-dylinositol anchor. Ephrin B family members have a transmembrane domain. Ephrin A family members typically bind to Eph A kinases, and ephrin B family members usually bind to Eph B kinases, but exceptions do exist, including EphA4, which can bind to B and A ephrins. The ephrin B cytoplasmic domain includes sites for tyrosine phosphorylation and a docking site for proteins with PDZ domains. These sites give the B ephrins at least two ways to engage in protein–protein interactions (4, 10). As a result, when cells expressing ephrins contact cells expressing Eph kinases, bidirectional signaling can occur (11).

Although the full range of responses mediated by Eph kinases and ephrins remains to be described, it has been shown that Eph/ephrin interactions can have biologically meaningful effects on cell migration and adhesion. In the developing central nervous system, interactions between ephrins and Eph kinases help to establish boundaries and prevent inappropriate interactions (12). Deletion of genes encoding ephrins and Eph kinase can affect neurological development in model organisms, and mutations in ephrins and Eph kinases have been associated with human neurological disorders (12–14). Loss of ephrinB2 or EphB2 affects development of the vasculature (15). In some cases, the effects of Eph/ephrin interactions on migration and adhesion have been linked to effects on integrin engagement which, in different cells and different settings, can be inhibited or promoted (16–21). In the case of platelets, Eph/ephrin interactions favor cohesion: Forced clustering of ephrinB1 or Eph kinases promotes adhesion and aggregation, whereas inhibition of Eph/ephrin interactions leads to premature disaggregation at low agonist concentrations (7, 8). These effects appear to be partially due to the ability of ephrinB1 to activate Rap1 (8), a Ras family member that supports integrin activation in platelets (22). In addition, in activated platelets, EphA4 becomes associated with the Src-related kinases Fyn and Lyn (7), an association that is relevant for the present studies.

Contacts between circulating platelets are normally transient, but the onset of integrin engagement permits Eph/ephrin interactions to occur as well. In the present studies, we have examined the interrelationships that occur between Eph kinases and integrins in platelets by examining the physical and functional associations that develop between α_IIbβ₃, EphA4, and ephrinB1 during platelet activation. The results show that EphA4 is constitutively associated with α_IIbβ₃ in platelets and, like α_IIbβ₃, shows an increase in surface expression when platelets are activated. The results also show that, in addition to promoting Rap1 activation, Eph/ephrin interactions support the stable accumulation of platelets on collagen under flow and promote postengagement “outside-in” signaling through α_IIbβ₃ by stabilizing platelet aggregates and facilitating tyrosine phosphorylation of the β₃ cytoplasmic domain. Phosphorylation of β₃ allows myosin to bind to α_IIbβ₃ and clot retraction to occur. The model that emerges is one in which the onset of aggregation permits Eph/ephrin interactions to occur, after which signaling downstream from ephrinB1 and its receptors favors continued growth and stability of the thrombus by several mechanisms, including positive effects on outside-in signaling through α_IIbβ₃.

Methods

Reagents. Anti-β₃ monoclonal antibody SSA6 has been described in ref. 23. Anti-phospho-β₃ pY773 (Y747 in mice) was from BioSource International (Camarillo, CA). The myosin antibody and Arg-Gly-Asp-Ser peptide were from Sigma. The anti-EphA4 murine monoclonal antibody 16C11 was raised against residues 125–149 of EphA4. Thrombin was a generous gift from John Fenton (New York State Department of Health, Albany). Monomeric His-6-tagged proteins corresponding to the exodomains of EphA4 (Thr-21–Val-551) and EphB1 (Ser-143–Asp-529), and GST-EphB1 were prepared as described in refs. 7, 8, and 23. Binding domain deleted His-6-tagged ΔEphA4 lacks Asp-130–F-142; His-6-tagged ΔEphB1 lacks Asp-118–Ile-134 (7).

Clot Retraction. Blood from medication-free volunteer donors was anticoagulated with 147 mM sodium citrate (1:9). Red cells were removed by centrifugation at 100 × g for 20 min. Clotting was initiated by adding 10 units/ml thrombin and studied at 37°C. The extent of clot retraction was quantified by measuring the volume of extruded serum.

Immunoprecipitation. Tyr phosphorylation of β₃ was detected by precipitating β₃ with antibody SSA6 followed by immunoblotting with anti-pY773-β₃. Lysates were prepared in Nonidet P-40 buffer (1% Nonidet P-40/50 mM Tris·HCl/75 mM NaCl/4 mM NaV0₄), precleared, and incubated overnight at 4°C with the indicated antibody. The protein complexes were then precipitated with protein G agarose. For the β₃-myosin coprecipitation studies, platelets were lysed and sonicated in modified Nonidet P-40 buffer (1% Nonidet P-40/50 mM Tris·HCl/150 mM NaCl/1 mM NaVO₄), then centrifuged at 16,000 × g before preclearing and immunoprecipitation. When indicated, the membranes were stripped in 62.5 mM Tris·HCl, pH 6.8, with 2% SDS and 100 mM 2-mercaptoethanol for 10 min at 63°C and reprobed.

Immunofluorescence. Glass slides were coated overnight at 4°C with 100 μg/ml fibrinogen, washed in PBS, and blocked for 2 h in 5 mg/ml BSA. Gel-filtered platelets (1.25 × 10⁷ platelets in 0.5 ml) were added to each slide along with 100 nM phorbol myristate acetate (PMA). After 30 min at room temperature, attached platelets were fixed in 3.7% formaldehyde/PBS for 10 min and permeabilized in 0.2% Triton X-100 for 5 min, followed by a blocking step for 30 min in 10% goat serum. After blocking, platelets were incubated for 45 min at 37° with Alexa Fluor 488-labeled SSA6 anti-β₃ along with Alexa Fluor 594-labeled anti-EphA4. After being washed, the slides were mounted in PBS/glycerol (Citifluor, London).

Flow Cytometry and Platelet Aggregation. Platelets were isolated from plasma by gel filtration on Sepharose 2B and adjusted to a final concentration of 2 × 10⁸ per milliliter before being incubated with monoclonal anti-EphA4 (16C11), anti-β₃ (SSA6), or an isotype-matched control. After 25 min at room temperature with or without 100 nM PMA, the platelets were fixed with an equal volume of 2% formalin in Tyrode's buffer, washed, and then resuspended in PBS before being incubated with FITC-conjugated goat antimouse diluted in 10% goat serum in PBS for 45 min at 37°C and then resuspended in PBS before analysis by FACS. Platelet aggregation was observed in a ChronoLog (Havertown, PA) platelet aggregometer with gel-filtered platelets supplemented with 1 mM CaCl₂ and 300 μg/ml fibrinogen.

Cell fractionation was performed by a method adapted from Fauvel et al. (24). Washed platelets were prepared from one to two units of freshly obtained blood, resuspended in Buffer A (100 mM KCl/25 mM Tris·HCl, pH 7.4/3 mM ATP/3 mM MgCl₂), and disrupted by nitrogen cavitation in a Parr bomb for 20 min at 70 atm (1 atm = 101.3 kPa). The resulting lysate was centrifuged at 1,500 × g for 10 min at 4°C. The supernatant was applied to a Percoll gradient (6 ml of homogenate supernatant/12 ml of 2× Buffer A/1.68 ml of H₂O) and spun at 79,000 × g for 15 min at 4°C. Membrane-containing fractions were then collected and submitted to a second Percoll gradient (6 ml of homogenate supernatant/12 ml 2× Buffer A, pH 9.6/1.68 ml of H₂O). In this gradient, a combination of plasma membranes, surface connecting membranes, and membranes from the dense tubular system formed a band near the top of the gradient; granule membranes were present in fractions near the bottom. The two fractions were collected separately, diluted 3-fold in buffer A, sonicated twice for 30 sec with a MSE sonicator at maximum power before centrifugation at 200,000 × g for 60 min to sediment the membranes, which were resuspended in a small volume of buffer A.

Ex Vivo Perfusion Studies. Whole blood was anticoagulated with 93 μM d-phenyl alanyl-l-prolyl-l-arginine chloromethyl ketone dihydrochloride, a thrombin inhibitor. Platelets were fluorescently labeled with mepacrine and examined at a wall shear rate of 1,500 s^–1 in a flow chamber coated with 2.5 mg/ml type I fibrillar collagen and maintained at 37°C on the stage of a Zeiss Axiovert 135M-LSM410 inverted epifluorescence microscope and laser confocal microscope (25–27).

Results

To begin to assess the role of Eph/ephrin interactions in postaggregation events, we asked whether interrupting Eph/ephrin interactions would affect clot retraction, an event that depends on outside-in signaling through α_IIbβ₃. Clot retraction occurs when actin–myosin filaments within aggregated platelets contract, pulling on α_IIbβ₃ by means of a direct interaction between myosin and the β₃ cytosolic domain (28). Clot retraction is thought to enhance thrombus stability and prevent premature dissolution of the clot by plasmin, and it may also facilitate contact-dependent signaling by compressing the spaces between platelets (29). In the studies shown in Fig. 1, clot retraction was observed in the presence and absence of soluble monomeric proteins corresponding to the extracellular domains of EphA4 and EphB1. These monomers act as antagonists by binding to ephrinB1, preventing engagement with endogenous Eph kinases and the subsequent oligomerization required for signaling (30). Clotting was initiated by adding thrombin, activating the platelets, and causing the rapid formation of a platelet-embedded fibrin clot that filled the cuvette. This process was followed by the gradual retraction of the clot that was complete within 60 min. Blockade of Eph/ephrin interactions greatly retarded the rate of clot retraction, which was still incomplete after 2 h. Monomers lacking residues required for binding to ephrinB1 (Fig. 1, ΔMonomers) had no effect.

Fig. 1.

Inhibition of Eph/ephrin interactions impairs clot retraction and the association of myosin with α_IIbβ₃. (A and B) Platelet-rich plasma was preincubated for 10 min with either PBS, 10 μg/ml His-EphA4 plus His-EphB1 (Monomers), or 10 μg/ml binding-domain-deleted His-ΔEphA4 plus His-ΔEphB1 (ΔMonomers), after which 5–10 units/ml thrombin was added to initiate thrombus formation. The clots were photographed after 120 min, and clot volume was determined at each time indicated. (C) Platelets were incubated for 10 min with thrombin. Proteins were immunoprecipitated (IP) with anti-β₃ or an isotype-matched control (M2) and then immunoblotted (IB) with anti-β₃ plus anti-myosin (My). The experiments shown are representative of three similar studies.

We next asked whether the defect in clot retraction could be due to a defect in the binding of myosin to the cytoplasmic domain of β₃. Lysates were prepared from resting and activated platelets, immunoprecipitated with anti-β₃ and then immunoblotted for myosin. Myosin coprecipitated with β₃ when the lysates were prepared from activated platelets but not from resting platelets or from platelets that had been activated in the presence of the ephrinB1-blocking monomers (Fig. 1C). Therefore, blockade of Eph/ephrin interactions inhibits the association of myosin with β₃, which accounts for the observed inhibition of clot retraction.

Myosin-binding requires the phosphorylation of two Tyr residues (Tyr-773 and Tyr-785) within the β₃ cytoplasmic domain (28). An elegant demonstration of the requirement was obtained in a mouse model in which substitution of Phe for Tyr-773 and Tyr-785 (Tyr-747 and Tyr-759 in mice) prevented myosin binding and impaired clot retraction (31), much as we observed in human platelets when Eph/ephrin interactions were blocked. To determine whether Eph/ephrin interactions play a role in β₃ phosphorylation, platelets were incubated with an agonist or with GST–EphB1, a fusion protein comprised of the extracellular domain of EphB1 joined to GST that can induce clustering and activation of ephrinB1 in platelets (7, 8). Afterward, the platelets were lysed, β₃ was immunoprecipitated, and the precipitate was probed with an antibody specific for β₃ phosphorylated on Tyr-773 (pTyr-773-β₃). Each of the agonists caused β₃ phosphorylation, as did GST–EphB1 (Fig. 2). Phosphorylation required fibrinogen and stirring and was inhibited by the Src family inhibitor PP2 or by blocking Eph/ephrin interactions (Fig. 2 B, C, and D). As reported in refs. 7 and 8, for platelet aggregation (and demonstrated again in Fig. 2E), the inhibition of phosphorylation produced by Eph/ephrin blockade was most evident at low agonist concentrations and less evident at high concentrations, suggesting that more than one mechanism is active in the phosphorylation of β₃ (see Discussion).

Fig. 2.

Tyr phosphorylation of the β₃ cytoplasmic domain. Except where indicated, platelets were incubated with an agonist for 10 min under aggregating conditions (300 μg/ml fibrinogen plus stirring), then lysed, immunoprecipitated with anti-β₃, and immunoblotted with a phospho-specific antibody to β₃ Tyr-773 (pTyr-773-β₃). (A) ADP (10 μM), thrombin (1 units/ml), or GST-EphB1 (16 μg/ml). (B) GST-EphB1 (16 μg/ml) or collagen (100 μg/ml) with or without the Src family inhibitor PP2 (10 μM). (C) ADP (5 μM), His-EphA4 (20 μg/ml), or GST-EphB1 (16 μg/ml). (D) ADP dose–response curve for β₃ phosphorylation preincubated for 10 min with 10 μg/ml His-EphA4 plus His-EphB1 (denoted Monomers or M) or His-ΔEphA4 plus His-ΔEphB1 (ΔMonomers or ΔM). (E) The aggregation traces for the samples analyzed in D.

The Association of EphA4 with α_IIbβ₃. We next asked whether integrins and Eph kinases are physically and functionally associated in platelets. Flow cytometry with a monoclonal antibody directed at the EphA4 extracellular domain showed that EphA4 is detectable on the surface of resting platelets and, like α_IIbβ₃, increases in expression during platelet activation (Fig. 3A). The increase in EphA4 expression is not due to recruitment from α-granule membranes. Cell fractionation studies in which platelet lysates produced by nitrogen cavitation were resolved on Percoll gradients showed EphA4 in the membrane fraction in resting and activated platelets, as opposed to the α-granule protein P-selectin, which was in the granule fraction in resting platelets but moved to the membrane fraction when platelets were activated (Fig. 3B). Fluorescence microscopy showed that EphA4 and β₃ have a similar distribution on the surface of activated platelets (Fig. 3C). Immunoprecipitation studies showed that EphA4 coprecipitated with β₃ by using an anti-β₃ antibody or an anti-EphA4 antibody (Fig. 3 D and E). This association was observed in resting and activated platelets, and it persisted even when Eph/ephrin interactions were blocked, in contrast with the association between EphA4 and the Src family members Lyn and Fyn, which occurs only in activated platelets and only when Eph/ephrin interactions are permitted (7).

Fig. 3.

Association of EphA4 with α_IIbβ₃. (A) Flow cytometry of platelets performed with monoclonal anti-EphA4 (clone 16c11) and anti-β₃ (clone SSA6). The results shown are representative of three similar studies. (B) Platelet lysates were resolved by Percoll gradient centrifugation into granule and membrane fractions. The results shown are representative of two similar studies. (C) Platelets were activated with 100 nM PMA and stained with anti-EphA4 or anti-β₃. The experiment shown is representative of two similar studies. (D and E) Platelets were incubated for 5 min with 20 μM ADP or 2 μM SFLLRN under aggregating conditions in the presence of PBS, 10 μg/ml His-EphA4 plus His-EphB1 (Monomers), or 10 μg/ml binding-domain-deleted His-ΔEphA4 plus His-ΔEphB1 (ΔMonomers), then immunoprecipitated (IP) with anti-β₃, anti-EphA4, or (as a negative control) antibody M2 and immunoblotted (IB). The experiment shown is representative of three similar studies. (F) Lysates were prepared from platelets activated with 100 nM PMA for 30 min, immunoprecipitated with anti-EphA4 (IP#1), and probed for β₃ and EphA4. Afterward, the supernatant from the first immunoprecipitate was reprecipitated with anti-β₃ (IP#2) and probed for β₃, EphA4, and phospho-β₃. The results shown are representative of two similar studies.

Although β₃ and EphA4 coprecipitated, the flow cytometry data suggest that the integrin is more abundant on the platelet surface than the kinase (Fig. 3A). The precise stoichiometry has not been reported, but it is likely that only some of the β₃ in platelets becomes phosphorylated during platelet aggregation. To determine whether the fraction of β₃ that becomes phosphorylated is restricted to the fraction of α_IIbβ₃ that remains associated with EphA4, the supernatant left behind after an initial immunoprecipitation with antiEphA4 (IP #1 in Fig. 3F) was reprecipitated with anti-β₃ (IP #2), and the resulting precipitate was probed for β₃, phospho-β₃, and EphA4. The results show that there is residual β₃ in the anti-EphA4 supernatant that is not associated with EphA4 and that some of this “free” β₃ is phosphorylated. Assuming that this result is not simply due to a partial dissociation of α_IIbβ₃/EphA4 complexes when the platelets were lysed, the presence of phospho-β₃ that does not coprecipitate with EphA4 could reflect movement of the integrin in and out of complexes with EphA4 during platelet activation. Alternatively, free phospho-β₃ could be because more than one mechanism exists for causing β₃ phosphorylation during platelet activation, with Eph/ephrin signaling being most relevant at sites where contact between platelet allows Eph/ephrin interactions to occur and less relevant at sites where fibrinogen binding has occurred but platelet–platelet contact has not.

Thrombus Growth Under Arterial Flow Conditions. In addition to facilitating the binding of myosin to α_IIbβ₃, Tyr phosphorylation of the cytoplasmic domain of β₃ makes possible the binding of signaling and adaptor molecules whose subsequent contribution to intracellular events promotes thrombus growth and stability (3). Therefore, by modulating outside-in signaling, Eph/ephrin interactions might impact these events as well. To determine the effects of Eph/ephrin interactions on developing thrombi, fluorescently labeled human platelets were perfused over a collagen-coated surface at arterial flow rates. The growth of platelet thrombi was observed, and mean thrombus volume was determined at several points in the flow chamber by addition of the corresponding cross sectional areas in successive confocal planes. The results show that mean thrombus volume decreased by 40% when Eph/ephrin interactions were blocked (Fig. 4).

Fig. 4.

Thrombus formation under arterial flow conditions. Mepacrine-labeled platelets in whole blood were perfused across a collagen-coated slide at a shear rate of 1,500 s^–1 in the presence of 10 μg/ml His-EphA4 plus His-EphB1 (Monomers) or 10 μg/ml binding-domain-deleted His-ΔEphA4 plus His-ΔEphB1 (ΔMonomers). After 5 min, the cells were fixed, and the slide was examined by confocal microscopy to determine thrombus volume. (A) Representative cross sections of the accumulated platelet thrombi in a series of parallel planes 10, 13, and 16 μm above the slide surface. (B) Average thrombus volume measured at four surface locations in each of four different experiments (mean ± SEM).

Discussion

Platelets are freely circulating blood cells that normally adhere to each other only at sites of vascular injury. Although injury can occur anywhere, the contribution of platelets to hemostasis is most critical in the arterial circulation, where rapid blood flow and ever-varying shear forces tend to oppose or destabilize hemostatic plug formation by pulling platelets away from each other and preventing the local accumulation of soluble agonists. Therefore, the ability of activated platelets to remain in contact with each other depends on the strength of the contacts between platelets, including the ability of integrins to remain engaged with their ligands and new adhesive and signaling interactions to occur once platelets come into contact. The present studies examine the contribution of ephrins and Eph kinases to this process. Previous work in cells other than platelets has shown that Eph/ephrin interactions can affect cell migration and adhesion by affecting integrin engagement. Depending on the cell and the context, Eph/ephrin engagement can promote or inhibit integrin activation (16–21). As a result, cells can move closer together when ephrins engage Eph kinases, or they can move apart. This process helps to establish boundaries within the developing central nervous system, as well as promoting coordinate development of the vasculature.

In the case of platelets, the evidence shows that Eph/ephrin interactions favor cohesion: Forced clustering of ephrinB1 or Eph kinases promotes adhesion and aggregation, whereas inhibition of Eph/ephrin interactions leads to premature disaggregation at low agonist concentrations (7, 8). The present studies suggest that, in addition to facilitating the inside-out signaling events that lead to irreversible integrin engagement, Eph/ephrin interactions play a previously unappreciated role in integrin outside-in signaling by contributing to the phosphorylation of Tyr residues within the β₃ cytoplasmic domain. In turn, β₃ phosphorylation affects the binding of cytoplasmic proteins to α_IIbβ₃ and supports postaggregation events, such as clot retraction. The biological relevance of these events is reflected by studies showing that mutation of the Tyr residues in the β₃ cytoplasmic domain produces a phenotype in mice in which platelet aggregation becomes reversible at lower agonist concentrations, clot retraction is inhibited, and thrombus stability is impaired (31), phenotypes similar to those we have observed with human platelets when Eph/ephrin interactions are inhibited (ref. 7 and this study).

The ability of Eph/ephrin interactions to affect integrin engagement and integrin signaling shows that there is a functional relationship between α_IIbβ₃ and Eph kinases expressed in platelets. The present studies show that there is a physical relationship as well. Like α_IIbβ₃, EphA4 is broadly distributed on the platelet surface and can be found at the sites of contact between activated platelets. The surface expression of both molecules increases when platelets are activated, presumably because of recruitment from the surface connecting membrane system, and the two molecules coprecipitate from lysates of resting or activated platelets. These findings suggest that there is a constitutive association between the integrin and the Eph kinase. The data in the present studies do not establish whether EphA4 interacts directly or indirectly with α_IIbβ₃, and it is likely that the relationship between them is complex. The exact number of copies of EphA4 per platelets has not been determined, but the FACS data suggest that it is likely to be less than α_IIbβ₃, which is expressed at very high density. The reprecipitation experiments in Fig. 3 show that, at any given time, some of the α_IIbβ₃ is not associated with EphA4 and that unassociated α_IIbβ₃ includes some of the phospho-β₃. Whether the presence of phosphorylated β₃ that is not associated with EphA4 reflects the existence of more than one mechanism for producing phosphorylated α_IIbβ₃ or whether it is because α_IIbβ₃ moves in and out of multimolecular complexes that include EphA4 remains to be determined.

Taken together, these observations suggest the following model. When platelets are activated by agonists, such as collagen, thrombin, or ADP, rapid signaling through their respective receptors initiates platelet aggregation by activating α_IIbβ₃, which binds to fibrinogen or von Willebrand factor, forming bridges between platelets. These bridges bring platelets into sustained contact with each other, allowing ephrinB1 on the surface of one platelet to engage its receptors on an adjacent platelet. Eph/ephrin interactions contribute another high-affinity interaction between the platelets and cause signaling complex formation that appears to have a number of consequences. One known consequence is helping to sustain Rap1B activation and, therefore, integrin engagement. Another consequence is the observed effect on α_IIbβ₃ phosphorylation, which supports integrin-based signaling and clot retraction.

How might the binding of ephrinB1 to its receptors support β₃ phosphorylation? It is likely that more than one mechanism is involved. The first mechanism reflects the ability of Eph/ephrin interactions to support the integrin engagement required for agonist-induced β₃ phosphorylation to occur. By contributing to aggregation, Eph/ephrin interactions increase the level of β₃ phosphorylation. In keeping with this first mechanism, we found that Eph/ephrin blockade inhibited ADP-induced platelet aggregation and ADP-induced β₃ phosphorylation. However, we also found that blockade of Eph/ephrin interactions had a greater effect on β₃ phosphorylation than on aggregation at low concentrations of ADP (Fig. 2) and that clustering ephrinB1 with GST–EphB1 was sufficient to cause β₃ phosphorylation. Therefore, we propose that there is a second mechanism by which Eph/ephrin interactions support β₃ phosphorylation, one that takes into account downstream signaling events in addition to Rap1 activation. Ephrin B family members are known to associate with Src family members and to signal via cytoplasmic proteins bound to the ephrin PDZ target domain (4, 10). EphA4 becomes associated with two Src family members, Fyn and Lyn, during platelet activation (7). Therefore, the ephrin and its receptors can gather together kinases that can potentially phosphorylate β₃, bringing those kinases into complexes that include the integrin. In keeping with such a mechanism, we found that β₃ phosphorylation in response to ephrinB1 clustering by GST–EphB1 is prevented by an inhibitor of Src family kinases. In contrast, Eph/ephrin interactions are clearly not the sole means by which Src family members are brought into contact with α_IIbβ₃. Arias-Salgado et al. (32) have recently reported that Src constitutively associates with β₃ integrins. Other Src family members (Fyn, Lyn, Hck, and Yes) were shown to associate with the cytoplasmic domain of β₃ when a fusion protein including that domain was incubated with platelet lysates. Might the association we observed previously between EphA4, Fyn, and Lyn be more simply attributed to the constitutive association of these kinases with α_IIbβ₃? Possibly, but this association seems unlikely to be the entire explanation. Although EphA4 is associated with α_IIbβ₃ in resting and activated platelets, we found previously that Fyn and Lyn associate with EphA4 only after the platelets are activated and that Src does not associate with EphA4 in either resting or activated platelets (7). Therefore, we propose that in addition to the constitutive association between α_IIbβ₃ and Src, there is an activation-dependent recruitment of other Src family members as a consequence of Eph/ephrin interactions, which, in turn, promotes the phosphorylation of β₃. This contribution is particularly helpful at suboptimal agonist concentrations, but is needed less at high agonist concentrations.

The contribution of Eph kinases and ephrins to platelet function raises the larger issue of the role of contact-facilitated signaling in the later stages of platelet activation. There is increasing evidence that the boundaries between platelets within a growing thrombus are populated by adhesion and signaling molecules that help platelets remain in sustained contact with each other. Some of these molecules activate Tyr kinases. In addition to the Eph/ephrin interactions that are the subject of the present study, two laboratories have shown that platelets express receptors for Gas6, a vitamin K-dependent protein secreted from platelet α-granules (33, 34). Gas6 is a ligand for the Axl/Sky/Mer family of receptor Tyr kinases found on the surface of platelets. Like the Tyr→Phe mutants of the β₃ tail, deletion of the genes encoding Gas6 or one of its receptors produces a phenotype of reversible platelet aggregation in mice. Deletion of these genes also causes resistance to thrombosis (33, 35). It is reasonable to speculate that the signaling events downstream of the Axl receptor Tyr kinases will prove to have strong parallels with those downstream of Eph kinases in human platelets.

Once thrombus formation begins, contacts between platelets become increasingly stable and the spaces between platelets become increasingly constrained. Bioactive molecules that are secreted from platelet storage granules (such as ADP and Gas6), released from within the platelet (thromboxane A₂) or shed from the platelet surface (CD40L) can reach higher local concentrations at their receptors by being confined to a small space and protected from being washed out by blood flow. This contribution of clot retraction and, by extension, Eph/ephrin interactions may turn out to be one of the most relevant. Clot retraction would be expected to reduce the space between adjacent activated platelets, which increases the effective local concentration of bioactive molecules, a paradigm recently established for epithelial cells subjected to compressive stress (29). By promoting clot retraction, Eph/ephrin interactions may also limit the access of fibrinolytic enzymes to the clot interior and promote signaling through other cell surface receptors, including those that contribute to wound healing.

In conclusion, we have shown previously that in activated platelets Eph kinases and ephrins become part of a signaling complex that is not present in resting platelets. We now show that EphA4 is constitutively associated with α_IIbβ₃ and that Eph/ephrin interactions contribute to integrin outside-in signaling, helping to regulate clot retraction and improving aggregate stability. Although explored here in the context of platelets, we anticipate that the contribution of Eph kinases and ephrins to integrin outside-in signaling is not unique to platelets but will be found in other types of cells as well. Because they support thrombus growth and stability, Eph/ephrin interactions provide a novel target for the development of therapeutic antiplatelet agents that can be used in the prevention of strokes and recurrent coronary artery occlusions, adverse events that are ultimately due to the accumulation of inappropriate platelet aggregates. Although small-molecule inhibitors of Eph/ephrin interactions have not been described, a 12-residue peptide that selectively blocks ephrin binding to EphA4 was recently reported and may provide the basis for developing future therapeutic antagonists (36).