Collagen, convulxin and thrombin stimulate aggregation-independent tyrosine phosphorylation of CD31 in platelets: evidence for the involvement of Src-family kinases.

this study we investigated the role of this collagen in the of CD31 tyrosine phosphorylation. We demonstrate that activation of this receptor and also platelet thrombin receptors result in tyrosine phosphorylation of CD31 and that this is not dependent on integrin function or platelet aggregation. We show that platelet activation by GPVI and thrombin receptors lead to an increase in the platelet surface expression of CD31 and present data that suggests Src-family kinases are responsible for CD31 tyrosine phosphorylation in platelets.


Introduction
Platelet endothelial cell adhesion molecule-1 (CD31/PECAM-1 1 ) is a membrane spanning glycoprotein of 130 kDa that is expressed on the surface of platelets, endothelial cells, neutrophils, monocytes, and some T-lymphocyte subsets (1)(2)(3)(4). The functions that have been identified for CD31 are diverse and include transendothelial migration of leukocytes (5,6), angiogenesis (3,7) vasculogenesis (8) integrin regulation (9,10) and T-cell receptor function (11). The cloning of CD31 was as a consequence of its expression on platelets (4), however a clear role for CD31 in platelets remains obscure. CD31 was classified to the family of cellular adhesion molecules on basis of structural and amino acid sequence. Homophilic ligand binding interactions are believed to underlie the functions of CD31, although heterophilic interactions with αvβ3, CD38 and unidentified proteins have been suggested (12) (13).
In recent years, considerable attention has been directed at the ability of CD31 to participate in signal transduction. Numerous studies have shown the CD31 becomes tyrosine phosphorylated in response to a variety of conditions that include CD31 cross-linking (14), activation of the high affinity receptor for immunoglobulin E, FcεRI (15), platelet aggregation (16), shear (17) and oxidative stress (18). The sites of CD31 tyrosine phosphorylation have been mapped to two residues that fall within a conserved signalling motif known as an immunoreceptor tyrosinebased inhibitory motif (ITIM). ITIMs are found in an increasing list of receptors that regulate immune function and are defined by the consensus sequence L/I/V/S-x-Y-x-x-L/V. Many proteins, including CD31, contain tandem ITIMs separated by a 15 to 30 amino acid residue sequence (19). The phosphorylated ITIM provides a docking site for signalling proteins that bind via Src-homology 2 (SH2) domains. The protein tyrosine phosphatases SHP-2 and SHP-1 by guest on  http://www.jbc.org/ Downloaded from 4 have been shown to associate with tyrosine phosphorylated CD31 (16,20), and for SHP-2 this interaction has been shown in vitro to stimulate its activation (14). As a consequence, ITIMcontaining receptors may counter the effects of tyrosine kinases and thereby negatively regulate signalling pathways, although SHP-2 has been shown to positively regulate growth factor receptor signalling. It has been suggested recently that this molecule would be more correctly classified as a member of the Ig-ITIM family of proteins (21).
In contrast to the effects of ITIM-containing receptors are the functions of a number of receptors that possess a conserved signalling motif termed an immunoreceptor tyrosine-based activation motif (ITAM). The consensus sequence for this motif is Y-x-x-L-x (6)(7)(8) -Y-x-x-L (22) and is found on the cytoplasmic tail of several immune receptors including T and B cell antigen receptors (TCR, BCR) and a number of immunoglobulin receptors (23,24). It is believed that ITAM-and ITIM-bearing receptors may in some contexts act antagonistically, when expressed on the same cell. Indeed a number of closely related receptors, such as FcγRIIa (ITAM) and FcγRIIb (ITIM), are thought to function in this way (25).
We and others have recently identified the collagen receptor that underlies platelet activation.
Collagens are principal platelet agonists at sites of vascular damage and are therefore central to the process of haemostasis. This collagen receptor comprises the recently cloned glycoprotein VI (GPVI) (26-28) and the Fc receptor (FcR) γ-chain (29). The FcR γ-chain contains an ITAM within it's cytoplasmic domain that is tyrosine phosphorylated on collagen-stimulation. The phosphorylated ITAM is responsible for recruiting signalling molecules such as Syk and phosphatidylinositol 3-kinase to the receptors, and thus multiple signalling pathways are by guest on July 21, 2018 http://www.jbc.org/ Downloaded from 5 activated leading to platelet secretion and aggregation (29)(30)(31).
The role of the collagen receptor GPVI in CD31 signalling in platelets is particularly interesting since this receptor signals through an ITAM. In this study we have investigated the role of this collagen receptor in the stimulation of CD31 tyrosine phosphorylation. We demonstrate that activation of this receptor and also platelet thrombin receptors result in tyrosine phosphorylation of CD31 and that this is not dependent on integrin function or platelet aggregation. We show that platelet activation by GPVI and thrombin receptors lead to an increase in the platelet surface expression of CD31 and present data that suggests Src-family kinases are responsible for CD31 tyrosine phosphorylation in platelets.
Preparation and stimulation of Platelets -Human platelets from drug-free volunteers were prepared on the day of the experiment by differential centrifugation as described previously (31) and suspended in modified Tyrodes-Hepes buffer (134mM NaCl, 0.34mM Na 2 HPO 4 , 2.9mM KCl, 12mM NaHCO 3 , 20mM Hepes, 5mM glucose, 1mM MgCl 2 , pH7.3) to a density of 8 x 10 7 (delivered in 50µl) was performed at 37°C in an aggregometer with continuous stirring (1200rpm). Mouse platelets were prepared as described previously (30) and suspended in modified Tyrodes-Hepes buffer to a density of 1.3 x 10 8 cells/ml. 90µl of suspension was used for each assay with collagen, Cvx or thrombin, added in a volume of 10µl. For protein precipitation experiments, platelets were resuspended in buffer that contained 1mM EGTA to prevent aggregation. In certain experiments platelets were preincubated prior to stimulation with the tetrapeptide RGDS (0.5mM) (16) or the fibrinogen γ-peptide DVHHLGGAKQAG (125µM) (34). Where required, inhibitors and other reagents were added such that the final solvent concentration was 0.2% (v/v) and incubated at 37°C for 5 min before stimulation. Apyrase was prepared in modified Tyrodes-Hepes buffer and indomethacin, PP2 and PP3 were dissolved in dimethylsulphoxide (DMSO).
Immunoprecipitation studies -Platelet stimulation was terminated by the addition of an equal volume of ice cold lysis buffer (2% (v/v) Nonidet P40 or 2% (v/v) Brij96, 20mM Tris, 300mM NaCl, 10mM EDTA, 1mM phenylmethylsulfonyl fluoride, 2mM Na 3 VO 4 , 10µg/ml leupeptin, 10µg/ml aprotinin, 1µg/ml pepstatin A, pH7.3). Detergent-insoluble debris was removed and the lysates were pre-cleared by mixing with protein A-Sepharose for 1 hour at 4°C (20µl of a 50% (w/v) suspension of protein A-Sepharose in tris-buffered saline-Tween (TBS-T; 20mM Tris, 137mM NaCl, 0.1% (v/v) Tween 20, pH7.6)). Protein A-Sepharose was removed from the lysates before the addition of relevant antibodies (1 or 2.5µg anti-CD31, 1µg anti-Lyn, Fyn, Src, Yes and Hck). Following rotation at 4°C for 1 hour, 0.5µl secondary antiserum was added where required (CD31 (human platelets), Lyn and Fyn: rabbit anti-mouse IgG; CD31 (mouse platelets) and Src: sheep anti-goat IgG) and mixed for a further 30 mins. 25µl protein A-8 Sepharose suspension was added to each sample and mixing continued for 1 hour before washing the Sepharose pellet in lysis buffer followed by a second wash with TBS-T, and the addition of Laemmli sample-treatment buffer. Proteins were separated by SDS-PAGE using 10% gels and transferred to polyvinylidene difluoride membranes by semi-dry western blotting.
Immunoblotting -Following western blotting, polyvinylidene difluoride membranes were blocked by incubation in 10% (w/v) bovine serum albumin (BSA) dissolved in TBS-T. Primary and secondary antibodies were diluted in TBS-T containing 2% (w/v) BSA, and incubated with polyvinylidene difluoride membranes for 1 hour at room temperature. Blots were washed for 2 hours in TBS-T following each incubation with antibodies for 1 hour at room temperature, and then developed using an enhanced chemiluminescence detection system. Primary antibodies were used at a concentration of 1µg/ml (anti: phosphotyrosine, CD31, Lyn, Fyn, Src, Yes and Hck). Horseradish peroxidase-conjugated secondary antibodies were diluted 1:10000.  Flow cytometry -Human platelets were stimulated at a density of 8x10 8 cells/mL with Cvx (62.5ng/mL, 125ng/mL) or thrombin (0.1U/mL, 1U/mL) in the presence of EGTA (1mM), indomethacin (10µM) and apyrase (2U/mL). Stimulation was terminated by dilution to 1x10 8 cells/mL by addition of ice-cold buffer (modified Tyrodes-Hepes buffer pH7.3 containing BSA (1% w/v), EGTA (1mM) and sodium azide (0.04% w/v)). Platelet suspensions were then incubated in the dark at room temperature for 1 hour with a FITC-conjugated anti-CD31 monoclonal antibody (AB468F) diluted to 1:200. Non-specific antibody binding was determined using a FITC-conjugated isotype-matched IgG (AB600F). Data was collected using a FACScan™ flow cytometer (Becton Dickinson), and analysed using CELLQuest™ software (Becton Dickinson).
In Vitro Kinase Assay -CD31 was immunoprecipitated as described above and assayed for kinase activity whilst immobilised on protein-A Sepharose following the final washing step. The supernatant was removed and replaced with assay buffer (105mM NaCl, 20mM Hepes, 5mM MnCl 2 , 5mM MgCl 2 , 10µM ATP, 5µCi of [γ-32 P]ATP, pH7.4). Samples were incubated at room temperature for 20 min whereupon reactions were terminated by the addition of reducing Laemmli SDS-PAGE sample-treatment buffer. Samples were boiled and proteins separated by SDS-PAGE on 10% slab gels and western blotted onto polyvinylidine difluoride membranes. Blots were exposed to storage phosphor screens and developed on a PhosphoImager SI™ (Amersham Pharmacia Bioctech) to detect the incorporation of 32 P into proteins present in the immunoprecipitate. Densitometry was performed using ImmageQuant software (Amersham Pharmacia Bioctech). The identity of 32 P labelled proteins was confirmed by immunoblotting.

Results
Collagen, convulxin and thrombin stimulate CD31 tyrosine phosphorylation in platelets -At sites of vascular damage platelets may be activated in response to a number of factors, amongst which the extracellular matrix protein collagen, and the enzyme thrombin are considered to be most thrombogenic. Collagen-stimulated activation is mediated through a tyrosine kinasedependent pathway via the receptor GPVI, while thrombin stimulates an alternative signalling pathway via activation of G protein-coupled thrombin receptors.
Since the physiological significance of CD31 in platelet function has not been determined, we sought to characterise the effect of collagen and thrombin receptor mediated platelet activation on the tyrosine phosphorylation state of CD31. In addition to use of collagen and thrombin, platelets were also stimulated with convulxin (Cvx), a protein purified from the venom of the rattlesnake (Crotalus durissus terrificus) that is a selective GPVI agonist (31,32).
Platelets were stimulated with collagen (10 -100µg/ml), Cvx (62.5 -500ng/ml) or thrombin (0.01 -0.5U/ml) for 90s and CD31 was immunoprecipitated from cell lysates. Proteins were separated by SDS-PAGE and immunoblotted to detect tyrosine phosphorylation. Each of the platelet agonists induced robust tyrosine phosphorylation of CD31 in a dose-dependent manner ( Figure 1A.). With collagen, tyrosine phosphorylation was detectable at a concentration of 10µg/ml and reached maximal phosphorylation at approximately 100µg/ml. Cvx is a very potent GPVI agonist (32) and stimulated maximal tyrosine phosphorylation at 62.5ng/ml. The level of CD31 tyrosine phosphorylation on thrombin stimulation peaked at 0.1U/ml, however at higher concentrations additional tyrosine phosphorylated proteins were co-immunoprecipitated. These unidentified proteins were also detected on stimulation of platelets with supramaximal concentrations of collagen and convulxin (not shown). Src-family kinases have been implicated in the phosphorylation of CD31. Therefore, the possibility that the smaller tyrosine phosphorylated protein that co-immunoprecipitated with CD31 ( Figure 1A) was a Src-family kinase, was investigated. The protein did not co-migrate precisely with any of five Src-family kinase members that are expressed in platelets. We report later in this paper that such kinases do associate with CD31 in platelets, however this is not dependent on platelet activation. The larger tyrosine phosphorylated co-immunoprecipitated protein has an apparent molecular mass of 68kDa (determined using QuantityOne ® software, BioRad) is likely to be SHP-2, whose association with tyrosine phosphorylated CD31 is well documented.
The kinetics of CD31 tyrosine phosphorylation stimulated with these agonists were examined and are shown in Figure 1B). Collagen-(100µg/ml) stimulated tyrosine phosphorylation was detectable at 20s of stimulation and the level peaked at 90s. This kinetic profile is very similar to that previously reported for the tyrosine phosphorylation of the FcR γ-chain on stimilation of platelets with collagen (29), and is consistent with this being an early signalling event. Cvx stimulated tyrosine phosphorylation was rapid and reached maximal levels after 20s of stimulation. Stimulation with thrombin also resulted in rapid tyrosine phosphorylation of CD31 that was detectable at 20s, however levels continued to rise over 300s.

CD31 tyrosine phosphorylation is independent of aggregation and secretion -The stimulation
of CD31 tyrosine phosphorylation in platelets following incubation with thrombin-related activatory peptide (TRAP) has been reported to be dependent on aggregation and integrin 12 function. The results discussed above were therefore surprising since the stimulations were performed in the presence of EGTA (1mM) which prevented platelet aggregation by chelating extracellular calcium that is required for stabilisation and function of the fibrinogen receptor, integrin α IIb β 3 . To confirm this observation, platelets were stimulated with Cvx (62.5ng/ml) or thrombin (0.1U/ml) under alternative conditions that prevent fibrinogen binding to integrin α IIb β 3 and thereby block aggregation (Figure 2A). Tyrosine phosphorylation was induced by both agonists in the presence of EGTA, the tetrapeptide RGDS (0.5mM) or the fibrinogen γ-peptide (125µM). The treatments alone did not affect basal levels of CD31 tyrosine phosphorylation (not shown) and the differences in the levels of CD31 tyrosine phosphorylation in resting platelets is a consequence of variability between donors and platelet preparations. Aggregation assays were performed on the samples and confirmed complete inhibition of aggregation (results not shown).
Platelet activation leads to degranulation and the release or secretion of various factors that further enhance platelet activation. These include thromboxane (Tx) A 2 formed from liberated arachidonic acid via the cyclooxygenase pathway, and ADP, which is secreted from dense granules. In order to determine whether collagen-, Cvx-or thrombin-stimulated CD31 tyrosine phosphorylation was dependent on the release of TxA 2 , platelets were stimulated in the presence or absence of indomethacin (10µM), an inhibitor of cyclooxygenase. Figure 2B shows that the level of CD31 tyrosine phosphorylation is not altered in the presence of indomethacin.
Furthermore the inclusion of apyrase (2U/ml), that catalyses the conversion of ADP to AMP, had no effect on the level of CD31 tyrosine phosphorylation stimulated by these agonists (Figure  13 2B). Slightly different levels of CD31 tyrosine phosphorylation in unstimulated cells in the two experiments shown in this figure are due to donor and platelet preparation variability.
Platelet degranulation following stimulation may be assessed through measurement of 5HT secretion. Figure 2C (i) shows the results of 5HT secretion assays performed on cells stimulated in the absence or presence together of EGTA (1mM) indomethacin (10µM) and apyrase (2U/ml).
High levels of 5HT secretion are stimulated with Cvx (62.5ng/mL) and thrombin (0.1U/mL) and this is almost abolished in the presence of EGTA, indomethacin and apyrase (Cvx: reduced from 79 ± 1.7% to 6.2 ± 3.0%; thrombin reduced from 61.8 ± 6.5% to 5.6 ± 4.7%, (mean ± S.E., n=3)). However, Cvx and thrombin-stimulated tyrosine phosphorylation of CD31 is maintained under these conditions Figure 2C(ii). These data support the notion that tyrosine phosphorylation of CD31 following stimulation of the GPVI collagen receptor or the thrombin receptor is an early signalling event that is not dependent on the activity of factors released by the platelets. All subsequent experiments were performed in the presence of EGTA (1mM), indomethacin (10µM) and apyrase (2U/mL).
Platelet-platelet contact may contribute to CD31 tyrosine phosphoryation -CD31 has been reported to participate in both homophilic and heterophilic binding and CD31 dimerisation or clustering has been shown to stimulate its tyrosine phosphorylation. However, the binding characteristics of this protein on platelets has not been studied. It was not clear from our experiments whether CD31 tyrosine phosphorylation was a direct consequence of intracellular increases the likelihood of inter-cellular ligand binding of CD31, which may lead to increased tyrosine phosphorylation of the protein. In accordance with this, we observed that stirring platelets induces low level and time-dependent tyrosine phosphorylation of CD31. This effect is shown in CD31 immunoprecipitates from stirred platelets in Figure 3A. The level of tyrosine phosphorylation is considerably lower than when stimulated with collagen, Cvx or thrombin, and continues to rise over a period of 5 minutes.
An increase in CD31 tyrosine phosphorylation on stimulation with collagen, Cvx and thrombin may be a direct consequence of activated intracellular signalling mechanisms and/or CD31ligand binding. If CD31-ligand binding is important for this process, one may expect to observe an increased capacity for such binding in platelets stimulated with Cvx or thrombin. To examine this, the level of cell surface expression of CD31 in resting platelets and platelets stimulated with Cvx or thrombin was examined by flow cytometry using a FITC-conjugated anti-CD31 antibody.
The results shown in Figure 3B demonstrate that the cell-surface expression of CD31 is increased following stimulation of platelets with Cvx or thrombin in a dose-dependent manner. The mean fluorescence levels for Cvx-stimulated platelets (125ng/ml) are increased by approximately 30% above resting levels, and for thrombin the increase is approximately 190%. These results are representative of four separate experiments. Comparison of Figure 3B with the CD31 tyrosine phosphorylation dose-responses shown in Figure 1A indicate that surface expression is increased at high concentrations where further increases in tyrosine phosphorylation are not observed.
This indicates that increased surface expression of CD31 and the increased ligand binding that this would facilitate, may contribute to CD31 tyrosine phosphorylation, but additional contributions by direct GPVI and thrombin receptor signalling are also likely to be involved. Reconstitution studies have partially addressed which kinases can phosphorylate CD31 and have implicated Src, Lyn and the Csk kinase (not a Src-family kinase), although an exhaustive investigation has not been made. The identity of the kinase(s) that phosphorylate CD31 at normal cellular levels in vivo is, however, not clear, and this question has not been addressed in platelets. (37), FcγRI (38), TCR (39), BCR (40) and the platelet-derived growth factor receptor (41) that are activated through phosphorylation by Src-family kinases have been shown to physically associate with the appropriate kinase. We therefore conducted in vitro kinase assays on CD31 immunoprecipitates obtained from resting and stimulated platelets. Such experiments revealed that a kinase is co-immunoprecipitated with CD31 and is able to phosphorylate CD31 in vitro ( Figure 4A).

Receptors such as FcεRI
Furthermore, the ability of this enzyme to phosphorylate CD31 is increased following stimulation of platelets with convulxin or thrombin ( Figure 4A upper panel). Quantification of increased 32 P incorporation into CD31 is shown in Figure 4B. Blots were probed for tyrosine phosphorylation, to reveal a similar profile to 32 P incorporation in the kinase assays ( Figure 4A middle panel). The identity of the protein on the phosphorimager radiograph was confirmed by reprobing the blot for CD31. Since a kinase co-immunoprecipitates with CD31 from platelets, that is able to phosphorylate CD31 in vitro and whose activity is increased on stimulation of platelets with Cvx or thrombin, the possibility that the associated enzyme may be a Src-family kinase was explored. Platelets  (Figure 6). These results were reproducible, however, the level of reduction in tyrosine phosphorylation was variable. This is likely to be a consequence of redundancy in function of the multiple Src-family kinases that are expressed in platelets. A high concentration of thrombin was used in these experiments to ensure detectable levels of tyrosine phosphorylation because, unlike human platelets, the dose response of mouse platelets to thrombin is very steep between 1 and 10U/ml. In addition, mouse platelets are less sensitive than human platelets to Cvx, and therefore 125ng/ml was used to ensure detectable phosphorylation of CD31. Similar results were obtained with Lyn-deficient platelets (result not shown).
However, these mice are thrombocytopenic and this has prevented more extensive study. Much of the characterisation of CD31 phosphorylation and signalling has been performed in platelets, however, the role of CD31 in these cells is not clear. Clustering of CD31 using antibodies has been shown to up-regulate integrin function on the platelet surface, and to enhance ADP-mediated platelet activation (44). However, recent studies with CD31 knockout mice concluded that CD31 is not critically involved in ADP stimulated platelet-platelet adhesion during aggregation (45). More extensive functional assays using CD31 -/-platelets and the use of alternative agonists with these cells have not been reported. Platelet activation using TRAP has also been shown to result in the tyrosine phosphorylation of CD31 at Tyr 663 and Tyr 686 , and this is reported to be dependent on integrin α IIb β 3 engagement (16).
In this study we aimed to examine the regulation of CD31 tyrosine phosphorylation in platelets in response to activation of the GPVI collagen receptor. This was done was using collagen and the GPVI-selective agonist convulxin (Cvx) that allows examination of GPVI signalling in the absence of integrin α 2 β 1 binding. This protein has been shown to bind GPVI with high affinity and we have shown using FcR γ-chain deficient mice that Cvx-stimulated platelet activation is mediated through the GPVI receptor (31). Each of collagen, Cvx and thrombin stimulated robust concentration-dependent tyrosine phosphorylation of CD31. The kinetics of phosphorylation were rapid (detectable within 20s), and in the case of collagen was very similar to the kinetics of phosphorylation of the FcR γ-chain, suggesting that this is an early signalling event (29). Of particular note was that tyrosine phosphorylation was not dependent on platelet aggregation, as has been previously reported following activation with TRAP (16). Our observations were reproduced in the presence of three treatments that prevent platelet aggregation through inhibition of fibrinogen binding to integrin αIIbβ3.
A critical difference between our experiments and those of Jackson and colleagues, lie behind the definition of non-aggregating conditions. In the above report, negligible increases in tyrosine phosphorylation were detected under non-aggregating conditions, but these experiments were conducted by treating platelets under static conditions. Aggregation was induced by agonists and stirring, and prevented in stirring conditions through the addition of the tetrapeptide RGDW. It is possible that their observations may be explained by the fact that stimulation without stirring limits cell-cell contact, and this may in itself reduce CD31 tyrosine phosphorylation levels. As outlined above, CD31 is recognised for its propensity for homophilic ligand binding interactions, that may occur in an inter-cellular manner and give rise to activation of CD31. This is supported by our observation that stirring platelets stimulates low-level time-dependent tyrosine phosphorylation of CD31 ( Figure 3A). Furthermore we report here that stimulation of platelets with Cvx or thrombin leads to a substantial increase in the surface expression of CD31. This may lie behind the mechanism that leads to increased tyrosine phosphorylation in agoniststimulated platelets. We have examined the membrane localisation of CD31 by immunoblotting of platelet external and internal membrane preparations (prepared by high-voltage free flow electrophoresis and kindly provided by Dr K Authi, Thrombosis Research Institute, London, UK). CD31 was detected on internal as well as external platelet membranes (unpublished results), which would be expected given our observation that surface expression may be upregulated on stimulation. This is in agreement with Cramer et.al., who show using immunogold electron microscopy the presence of CD31 in α-granules (46). It is interesting to note that a greater increase in surface expression is obtained with thrombin in comparison with Cvx. This is consistent with higher levels of CD31 tyrosine phosphorylation on thrombin stimulation ( Figure   1B) at the concentrations of agonists that were used in flow cytometry experiments. In addition, the association of CD31 with platelet cytoskeleton following stimulation with thrombin has been reported (47). This interaction may be important in the mechanism for increasing surface expression of this molecule.
In accordance with the observations of Jackson et al. we have also observed higher levels of CD31 tyrosine phosphorylation when platelets are allowed to aggregate (e.g. in the absence of EGTA). We suggest that this is most likely to be more a consequence of greater platelet-platelet interactions in the aggregate fostering elevated CD31-ligand binding, rather than as a direct consequence of integrin α IIb β 3 engagement. However, the possibility that fibrinogen binding and other receptor interactions may contribute, cannot be excluded. The respective contributions of CD31 ligand-binding and signal transduction as a direct consequence of GPVI and thrombin receptor activation remain to be determined, although the rapid kinetics strongly suggest that direct signalling in addition to CD31-ligand binding is likely to be involved.
In this study we have begun to address the identity of the kinase(s) that is responsible for tyrosine phosphorylation of CD31 following stimulation of GPVI and the thrombin receptor in platelets. The Src-family of tyrosine kinases, more specifically Src, Lyn and Lck, have been widely implicated in the phosphorylation of CD31, and more recently Csk and the related kinase Chk have been added to this list. Other proteins such as Syk, which is activated on binding to the activated collagen receptor, have been shown to be unable to phosphorylate CD31 (36).
Most of this work has been performed in vitro and with recombinant proteins, or through overexpression reconstitution studies where elevated kinase levels may promote phosphorylation events that may not occur in vivo. It is however important to note the kinases that are capable of phosphorylating CD31 on tyrosine.
We have demonstrated that a tyrosine kinase is associated with CD31 in platelets and that it is able to phosphorylate CD31 in vitro. Furthermore, the ability of this enzyme to phosphorylate CD31 is increased on stimulation with Cvx or thrombin. We therefore sought to identify which kinase(s) may be present in CD31 immunoprecipitates and responsible for phosphorylating CD31 in platelets. A Src-family selective inhibitor, PP2, was used to determine whether Srcfamily kinases are likely to be involved in the phosphorylation of CD31 in platelets (42). This compound was able to inhibit or substantially reduce the level of CD31 tyrosine phosphorylation on stimulation with Cvx or thrombin in dose-dependent manner. The activation of the GPVI collagen receptor has been shown to involve the kinases Fyn and Lyn, (48,49) and therefore the effect on CD31 phosphorylation on Cvx stimulation may be a consequence of inhibition of GPVI activation. However, stirring-induced CD31 phosphorylation was abolished using PP2 and thrombin-stimulated platelet activation has not reported to be dependent on Src-family kinases.
This strongly implicates Src-family kinase involvement in agonist-and stirring-mediated CD31 tyrosine phosphorylation. where isolation of considerably greater levels of CD31 is possible (in comparison to coassociation in Src-family kinase immunoprecipitates) the levels of CD31 tyrosine phosphorylation is increased on stimulation with Cvx or thrombin. It is therefore likely that increases in CD31 tyrosine phosphorylation were not observed in Src-family kinase immunprecipitates due to insufficient assay sensitivity.
To further investigate a role for Src-family kinases in this process, we studied the phosphorylation of CD31 in mouse platelets that lack Fyn. Stirring induced phosphorylation was detectable in Fyn +/+ and Fyn -/-platelets and was indistinguishable. Cvx and thrombin induced CD31 tyrosine phosphorylation was reduced in Fyn -/-platelets. The reduction was variable   was immunoprecipitated from resting platelets and following stimulation with collagen (100µg/ml), cvx (125ng/ml) and thrombin (0.1U/ml) for the time points indicated. Proteins were separated by SDS-PAGE and immunoblotted to detect phosphotyrosine residues. Blots were reprobed for CD31 to ensure equal loading. I.P.,immunoprecipitates; Tyr(P), phosphotyrosine.
Immunoprecipitated proteins were separated by SDS-PAGE and immunoblotted for phosphotyrosine residues. Blots was reprobed for CD31 to verify equal protein loading. (C)