Steve P. Watson, Institute of Biomedical Research, Medical School, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. Tel.: +44 0 121 414 6514; fax: +44 0 121 415 8817; e-mail: email@example.com
Summary. Background: Glycoprotein VI (GPVI) is a physiologic receptor for collagen expressed at the surface of platelets and megakaryocytes. Constitutive dimerization of GPVI has been proposed as being necessary for the interaction with collagen, although direct evidence of dimerization has not been reported in cell lines or platelets. Objectives: To investigate oligomerization of GPVI in transfected cell lines and in platelets under non-stimulated conditions. Methods and results: By using a combination of molecular and biochemical techniques, we demonstrate that GPVI association occurs at the surface of transfected 293T cells under basal conditions, through an interaction at the extracellular domain of the receptor. Bioluminescence resonance energy transfer was used to confirm oligomerization of GPVI under these conditions. A chemical crosslinker was used to detect constitutive oligomeric forms of GPVI at the surface of platelets, which contain the Fc receptor (FcR) γ-chain. Conclusions: The present results directly demonstrate GPVI–FcR γ-chain oligomerization at the surface of the platelet, and thereby add to the growing evidence that oligomerization of GPVI may be a prerequisite for binding of the receptor to collagen, and therefore for proper functioning of platelets upon vascular damage.
The extracellular matrix protein collagen is the major and most thrombogenic component of the vessel wall. Circulating platelets adhere to exposed collagen and undergo activation, leading to thrombus formation. The interaction with collagen is mediated through two distinct receptor classes on the platelet surface, the integrin αIIβ1 and the glycoprotein VI (GPVI)–FC receptor (FcR) γ-chain receptor complex . Blocking of either receptor in vitro using specific antibodies inhibits or delays, respectively, collagen-induced platelet aggregation [2–4]. Similarly, platelets deficient in αIIβ1 or GPVI–FcR γ-chain show loss of reactivity towards collagen in vitro [5–8]. Furthermore, mice deficient in GPVI–FcR γ-chain are protected against lethal thromboembolism [8,9], illustrating the crucial role that the receptor plays in vivo under pathologic conditions.
Although the intracellular signaling events mediated by αIIβ1 in platelets have remained elusive [10,11], the mechanism of action of GPVI has been well documented and remains an area of intense research. GPVI mediates platelet activation in response to collagen through a pathway that shares many features with those used by immune receptors such as FcεRI, and T-cell and B-cell antigen receptors . As GPVI has no intrinsic signaling capacity, it is widely recognized that it must be coexpressed in association with the FcR γ-chain, which acts as the signaling partner. Furthermore, this association is a prerequisite for surface expression of GPVI on mouse platelets .
From a structure–function point of view, there are several lines of circumstantial evidence to suggest that GPVI functions as a dimer on the platelet surface. Moroi and coworkers have demonstrated, using recombinant protein, that collagen binds to the dimeric but not the monomeric form of GPVI, and that only the former is able to attenuate collagen-induced platelet aggregation . In contrast, both the monomeric and dimeric forms of GPVI bind to immobilized convulxin and inhibit platelet aggregation induced by the snake toxin with similar concentration dependencies . The possibility that GPVI functions as a dimer is strongly reinforced by studies analyzing the ability of a series of synthetic peptides with differentially spaced GPVI-recognition motifs to activate the collagen receptor in platelets . Finally, structural studies of the two immunoglobulin (Ig) domains of human GPVI have revealed the formation of a back-to-back dimer in the crystal structure, which is mediated through the more membrane-proximal of the two Ig domains .
As the FcR γ-chain is present as a disulfide-linked homodimer, it has been proposed that each chain associates independently with GPVI . In light of a recent report indicating that the two chains of the FcR γ-chain are necessary for binding a single GPVI molecule , this model needs to be reviewed. Furthermore, direct evidence that GPVI is expressed at the cell surface as a dimer or possibly as a larger complex is not available. We set out to investigate this in platelets and in transfected cell lines using distinct biochemical and molecular approaches. Our results confirm that GPVI is capable of undergoing oligomerization in transfected cells and forming oligomers in platelets, and indicate that a modified version of the current model for GPVI dimerization may be necessary.
Materials and methods
Reagents and antibodies
Convulxin was purchased from Latoxan (Valence, France). Anticonvulxin was a kind gift from M. Leduc (Institute Pasteur, Paris, France). Anti-CD2 antibody was kindly supplied by V. Horejsi (Institute of Molecular Genetics, Academy of Sciences, Prague, Czech Republic). Anti-FcR γ-chain was obtained from Upstate Biotechnology (Buckingham, UK). Anti-Flag (M2) was obtained from Sigma (Dorset, UK). Anti-Myc (9B11) was obtained from Cell Signalling Technology (Hertfordshire, UK). All other reagents were obtained from previously described sources  unless otherwise stated.
Human blood was taken from drug-free volunteers on the day of experiment using acidic citrate dextrose (120 mm sodium citrate, 110 mm glucose, 80 mm citric acid). Platelet-rich plasma was obtained by centrifugation at 200 × g for 20 min, and platelets were isolated by centrifugation at 1000 × g for 10 min in the presence of prostacyclin (0.1 μg mL−1). Platelets were resuspended in modified Tyrode’s/HEPES buffer (134 mm NaCl, 2.9 mm KCl, 0.34 mm Na2HPO4.12H2O, 12 mm NaHCO3, 20 mm HEPES, 1 mm MgCl2, 5 mm glucose, pH 7.3, at 37 °C) in the presence of prostacyclin (0.1 μg mL−1), recentrifuged at 1000 × g for 10 min, and resuspended in the above buffer to a density of 5 × 108 cells mL−1.
293T cells were grown in DMEM supplemented with 100 U mL−1 penicillin, 100 μg mL−1 streptomycin and 10% heat-inactivated fetal bovine serum under 5% CO2/95% air in a humidified incubator. Cells were kept at the exponential phase of growth.
Cells were transfected using the calcium phosphate method as previously described , and incubated in complete medium for 48 h prior to experimentation. For stable transfections, plasmid DNA was cut with an appropriate restriction enzyme, and after transfection following the above method, divided into 24 wells; the required antibiotic for selection was then added. Approximately 10 days later, individual clones of cells were selected and placed in 96-well plates for expansion.
GPVI–Flag and (Δ288)GPVI–Flag were subcloned into the pRc plasmid, whereas the FcR γ-chain was subcloned into the pMG plasmid, as previously described . GPVI–Myc was obtained by standard PCR using a vector primer (T7) and GPVI–Myc primer (5′-CCCTAAGCGGCCGCTCACAGATCCTCTTCTGAGATGAGTTTTTGTTCTGAACATAACCCGCGGC-3′). The final amplified product was digested with HindIII and NotI, and inserted into the similarly cut mammalian expression vector pcDNA 3.1. The CD2 extracellular domain was fused to the transmembrane region and cytoplasmic tail of GPVI using standard overlapping PCR techniques. The extracellular domain of CD2 was amplified using oligo 1 (5′-CCCTAAAAGCTTACCATGAGCTTTCCATGTAAATTT-3′) and oligo 2 (5′-GGTGTAGTAGTCCAGACCTTTCTCTGGACA-3′), and a fragment containing the transmembrane and intracellular domains of GPVI was amplified using oligo 3 (5′-GGTCTGGACTACTACACCAAGGGCAACCTG-3′) and a vector primer (sp6). The two fragments were subsequently mixed together, and oligo 1 and sp6 were added to perform a second overlap PCR. The final amplified product, encoding a chimeric protein containing the extracellular part of CD2 and the transmembrane and intracellular domains of GPVI, was digested with HindIII and XbaI and inserted into the similarly cut mammalian expression vector pcDNA 3.1. GPVI–green fluorescent protein (GFP) and GPVI–luciferase constructs used for bioluminescence resonance energy transfer (BRET) analysis were generated by excision of GPVI from pRc–GPVI–Flag, followed by cloning into pGFP2-N3 and pRluc-N3 (PerkinElmer, UK), respectively. CD2 and cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) BRET constructs were prepared as previously described . The integrity and authenticity of constructs was confirmed by nucleotide sequencing.
Stimulations were terminated by the addition of an equal volume of ice-cold lysis buffer (2% Nonidet P-40, 300 mm NaCl, 20 mm Tris, 10 mm EDTA, 2 mm Na3VO4, 1 mm phenylmethanesulfonyl fluoride, 10 μg mL−1 leupeptin, 10 μg mL−1 aprotinin, and 1 μg mL−1 pepstatin A, pH 7.4). Insoluble cell debris was removed by centrifugation for 15 min at 12 000 × g at 4 °C. Samples were immunoprecipitated using an appropriate antibody overnight, and 30 μL of 50% (v/v) protein A–sepharose or protein G–sepharose was then added for 2 h. Beads were washed with lysis buffer, and resuspended in 20 μL of sodium dodecylsulfate sample buffer.
Immunoblotting and ligand blotting
Samples were separated by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS–PAGE) and transferred to poly(vinylidene difluoride) membranes, and then blocked with TBS-T [0.5 m Tris, 1.5 m NaCl, 0.1% (v/v) Tween-20, pH 7.4] containing 10% (w/v) bovine serum albumin (BSA) for at least 1 h. Blots were incubated with appropriate antibodies and developed using an enhanced chemiluminescence (ECL) detection system. For GPVI ligand blotting with convulxin, membranes were incubated with 10 μg mL−1 of convulxin for 1 h at room temperature, and incubated with anticonvulxin antibody. Prestained molecular weight protein markers were obtained from Bio-Rad (Hemel Hempstead, UK) and New England Biolabs (Hertfordshire, UK).
Cells were resuspended in phosphate-buffered saline (PBS) containing 1 mg mL−1 BSA. Cells were incubated with 10 μg mL−1 of primary antibody for 15 min, washed, and incubated with fluorescein isothiocyanate-conjugated secondary antibody for a further 15 min. Stained cells were analyzed immediately using FACScalibur (Becton Dickinson, Oxford, UK). Data were recorded and analyzed using cellquest software.
Chemical crosslinking assay
The crosslinking reaction was performed on washed platelets resuspended in PBS at room temperature. Sulfo-EGS is a water-soluble analog of a homobifunctional N-hydroxysuccimide ester (NHS-ester), EGS. Primary amines are principal targets for NHS-esters. Platelets were incubated with freshly prepared crosslinker solution at concentrations of 1 mm, 1.5 mm and 2 mm sulfo-EGS in PBS for 30 or 60 min at room temperature. The reaction was quenched by adding Tris–HCl (pH 7.5) at a final concentration of 50 mm for 20 min. Samples were lyzed and separated by SDS–PAGE under non-reducing conditions.
The BRET analysis was carried out essentially as previously described . Briefly, FuGene (Roche, Hertfordshire, UK) was used to transfect 293T cells, with varying ratios of the GFP and luciferase constructs. Cells were harvested 24 h post-transfection, and, for each transfection, 10 μm DeepBlueC (final concentration) was added to 100 μL of cells in a 96-well plate, and light emission in the 410 ± 40-nm (LU-A) and 515 ± 15-nm (LU-B) wavelength ranges was measured immediately. GFP and luciferase expression was measured in a separate well, and converted to a ratio of concentrations. BRETeff values were calculated, after background subtraction, as LU-B/LU-A, corrected for luciferase expression alone (typically 7% of LU-A). To assess the ability of ligands to alter GPVI oligomerization, 10 μL of convulxin (10 μg mL−1), collagen (20 μg mL−1) or PBS (control) was incubated with GPVI-transfected cells for 15 min before assaying for BRET as above.
GPVI undergoes oligomerization in 293T cells
We sought to investigate the possible dimerization or formation of higher oligomers of GPVI in a transfected cell line via immunoprecipitation. Plasmid constructs were generated coding for GPVI coupled to either Myc or Flag tags at the cytosolic C-terminus (Fig. 1A), and transiently transfected into 293T cells, a human kidney cell line with a high efficiency of transfection. The expression of both forms of GPVI was detected by western blotting for Myc and Flag (not shown). The two forms of GPVI were expressed at similar levels at the surface of the cells in the absence of FcR γ-chain, as demonstrated by flow cytometry using an anti-GPVI antibody (Fig. 1B).
In order to demonstrate association, 293T cells were transiently cotransfected with both GPVI constructs, either with or without FcR γ-chain. After 2 days in culture, the cells were lyzed, and protein extracts were immunoprecipitated using an anti-Flag antibody. Subsequent western blot analysis using an anti-Myc antibody demonstrated association between the two tagged forms of GPVI (Fig. 1C), independent of FcR γ-chain. Cotransfection of the latter did not increase the yield of association (Fig. 1D). Significantly, the experiments shown in Fig. 1C,D were performed on the same day, and the ECL exposures are time-matched. Ligand blots using convulxin revealed similar levels of expression of GPVI in the cotransfected cells. In order to rule out the possibility that association was an artefact of this experimental approach, lysates from cells with either GPVI–Flag or GPVI–Myc were combined and subjected to immunoprecipitation and western blotting as above. Under these conditions, no association was detected (data not shown). These observations confirm that GPVI is able to undergo oligomerization in transfected cells, and that this is not dependent on the FcR γ-chain.
Oligomerization takes place through the extracellular domain of GPVI
Further studies were performed to establish whether the intracellular or extracellular regions of GPVI are required for the homophilic interaction. For these studies, a chimeric and a mutant protein were generated (Fig. 2A). The chimeric protein consisted of the extracellular domain of the Ig surface protein CD2 coupled to the transmembrane and intracellular domains of GPVI, and tagged with Flag (CD2–GPVI–Flag) (Fig. 2A). For the mutant protein, GPVI was truncated at the interface of the transmembrane and cytoplasmic domains, and tagged with Flag at the C-terminus (Δ288)GPVI–Flag) (Fig. 2A). Similar levels of expression of the two constructs was demonstrated by transient transfection into 293T cells followed by western or ligand blotting (not shown), and by flow cytometry using specific antibodies for CD2 or GPVI (Fig. 2B).
The two modified forms of GPVI were cotransfected with Myc-tagged wild-type GPVI (GPVI–Myc) in 293T cells. Protein association was detected by immunoprecipitation with an antibody to Flag and western blotting for Myc (Fig. 2C). (Δ288)GPVI–Flag was found to coprecipitate with wild-type GPVI–Myc, demonstrating that the GPVI cytosolic tail is not required for the GPVI–GPVI interaction. In contrast, there was no association between CD2–GPVI and wild-type GPVI–Myc, confirming that the extracellular portion of GPVI is essential for the homophilic interaction.
GPVI oligomerization by BRET analysis
BRET relies on the transfer of energy from a donor molecule (luciferase) to an acceptor (GFP) that is only effective at distances of < 10 nm. Proteins of interest can be genetically fused to these fluorophores, permitting the analysis of protein interactions in live cells, with the level of energy transfer (BRETeff) and its dependence on the acceptor/donor ratio allowing the assignment of stoichiometry . Human 293T cells were transiently cotransfected with GPVI–GFP (GPVIGFP)-expressing and GPVI–luciferase (GPVILuc)-expressing plasmids as a ‘BRET pair’ (GPVIBP), along with FcR γ-chain. Subsequent analysis demonstrated that GPVI was capable of oligomerization, with the dependence of BRETeff values on the acceptor/donor ratio fitting best to a dimer model (Fig. 3A,B). The level of BRET observed was intermediate between that of the known monomer, CD2, and the disulfide-linked homodimer, CTLA-4, implying that GPVI dimers are likely to be in equilibrium with the monomeric moiety at the cell surface (Fig. 3A,B), as observed previously for CD80 . Expression of GPVI and CD2 as a BRET pair gave BRETeff values that were lower than those of CD2 and exhibited an independence from the acceptor/donor ratio, which is a characteristic of random interactions at the cell surface . GPVI dimerization was observed regardless of FcR γ-chain coexpression, although the level of dimerization was slightly enhanced in its presence (Fig. 3A). Incubation of GPVI-transfected cells with either collagen or convulxin did not increase the level of BRETeff observed for GPVI in the absence of ligand (Fig. 3C), with no effect being observed over 30 min (not shown). This suggests that GPVI binding to these two ligands does not significantly alter the distance between luciferase- and GFP-coupled GPVI molecules measured in BRET.
Chemical crosslinking of GPVI reveals the presence of high molecular mass complexes in platelets
To determine the molecular organization of GPVI on the platelet surface, platelets were treated with a chemical crosslinker that stabilizes pre-existing structures, including the presence of homodimers and associations with other membrane proteins. The membrane-impermeable crosslinker sulfo-EGS was chosen for these studies, as it has been previously used in similar studies [21,22]. Three concentrations (1, 1.5 and 2 mm) of sulfo-EGS were assessed for incubation times of 30 and 60 min as recommended by the manufacturer. Similar data were obtained using all combinations, as illustrated following a 30-min incubation with 1.5 mm sulfo-EGS (Fig. 4), with no major differences being seen between the 30-min and 60-min time points. The subsequent separation of whole cell lysates by SDS–PAGE under non-reducing conditions, combined with ligand blotting using the snake venom toxin, convulxin, identified three major molecular mass bands of approximately 55, 70 and 220 kDa in sulfo-EGS-treated cells, in comparison to a single major band of around 55 kDa in control samples. In addition, several uncharacterized, minor bands could also be seen that could be due to additional GPVI complexes, non-specific binding of convulxin, or the presence of contaminants in the snake toxin preparation. The major band of 55 kDa corresponds to the GPVI monomer, as confirmed by western blotting with an antibody to GPVI (not shown). The possibility that the 70-kDa band corresponds to a complex of GPVI and a dimer of FcR γ-chain, which would be preserved under the non-reducing conditions of the experiment, was supported by the appearance of a specific comigrating band in sulfo-EGS-treated cells upon western blotting for FcR γ-chain (Fig. 4B). Similarly, the 220-kDa band was detected by western blotting for FcR γ-chain, suggesting that it may represent a trimeric form of the GPVI–FcR γ-chain dimer association [i.e. (GPVI–FcR γ-chain)3]. Alternatively, it could represent the formation of a complex between GPVI–FcR γ-chain and one or more other proteins on the platelet surface, or an alternative combination of GPVI and FcR γ-chain. The efficiency of the crosslinker was investigated by heterodimerization of the platelet integrin subunits αIIb and β3 (Fig. 4C). We show that under the same conditions, treatment with sulfo-EGS induces formation of a complex with a molecular mass greater than 200 kDa, which appears to contain two prominent and other more minor bands. The size of this complex corresponds to that of the αIIbβ3 heterodimer, with the multiple bands possibly reflecting differential glycosylation. Significantly, however, as with GPVI and other platelet receptors , the monomeric form of αIIb was the predominant band in the presence of the crosslinking reagent, which might reflect the inefficiency of the crosslinking process.
The organization of the GPVI–FcR γ-chain receptor complex at the surface of platelets is of considerable interest. GPVI has been proposed as a possible target in the fight against thrombus-related diseases. GPVI plays a recognized role in platelet activation upon vascular injury under normal physiologic conditions, and its absence has been linked to mild bleeding disorders in humans and to protection against thrombus formation [24,25]. Furthermore, it has been proposed that GPVI antibodies and soluble GPVI dimers can protect against thrombus formation in animal models, although results for the latter have varied between different research groups [26,27].
It is widely accepted that GPVI must dimerize under basal conditions in order to bind collagen, in view of the low affinity of the monomeric form for the matrix protein. Furthermore, the concept of dimerization of GPVI is supported by a recent report showing that the crystal structure of GPVI contains back-to-back dimers of the receptor through an interaction in the extracellular domain . The accepted molecular model for GPVI dimerization proposes that an FcR γ-chain homodimer binds a molecule of GPVI on either side , with the site for interaction being in the transmembrane region, between the aspartic acid of the FcR γ-chain and the arginine residue of GPVI . However, this model has recently been challenged by a study showing that the two aspartic residues in the FcR γ-chain homodimer are necessary for association with the single arginine residue in the transmembrane region of GPVI . This model is energetically stable, as prior protonization of at least one of the carboxyl groups in the aspartic acid may occur, so that there is no charge imbalance in the assembled structure [28,29]. Furthermore, this seems to be a distinctive but common assembly mechanism for a number of activating immune receptors that are structurally similar to GPVI .
Our results suggest that GPVI dimerizes at the surface of GPVI-transfected 293T cells through an interaction that takes place in the extracellular domain. FcR γ-chain is not necessary for the dimerization, as it is not expressed in these cells. This would agree with the proposed model for GPVI dimerization from the recent crystallization study, which is independent of FcR γ-chain. Interestingly, in platelets, GPVI requires association with FcR γ-chain for expression on the platelet surface. However, this is not the case in the majority of transfected cell lines that have been studied, as demonstrated previously, not only for GPVI, but also for other immune receptors with similar structural organization that require FcR γ-chain for expression in the host cell [19,30,31]. Although FcR γ-chain is not essential for surface expression and dimerization of GPVI in the 293T cell line, it may help to stabilize expression, which may explain the difference in dimerization in cells that express FcR γ-chain, as demonstrated by BRET.
Interestingly, experiments using platelets pretreated with a chemical crosslinker that preserves surface complexes, followed by western blotting for either GPVI or FcR γ-chain, demonstrate the presence of two oligomeric structures that contain both proteins. The two bands of approximately 70 and 220 kDa containing GPVI and FcR γ-chain could correspond to a single GPVI molecule with two FcR γ-chains and a trimer of this complex, respectively. Alternatively, the latter could represent a protein multicomplex containing GPVI, FcR γ-chain and possibly one or more other protein(s). The possibility that a GPVI trimer may be present in this oligomeric structure is important, in that dimers show high affinity for fibrous collagen and inhibit platelet adhesion and aggregation to the injured vessel wall in vivo [14,17,26].
An important question is the extent to which activation of GPVI by multivalent ligands is mediated by oligomerization. The BRET results obtained indicate that neither convulxin nor collagen alter the average separation of GPVI molecules at the cell surface. The recent crystal structures of convulxin and GPVI may provide an explanation for this [16,32]. The structures indicate that only a monomer of convulxin can fit into the binding site of GPVI. Furthermore, the tetrameric shape of convulxin would not allow a GPVI dimer to form across the binding sites, because they are in the wrong orientation, suggesting convulxin may bind monomeric GPVI. On the other hand, the presence of multiple binding sites for GPVI in convulxin would enable the snake toxin to crosslink GPVI molecules and thereby induce a signal into the cell. This underlines the fundamental difference between basal receptor dimerization/oligomerization, which appears to be required for collagen binding, and agonist-induced receptor crosslinking/clustering, which is sufficient to induce signaling.
In the light of our own results as shown here, and those recently published elsewhere [16,18], we propose a modified version of the current model for GPVI dimerization at the platelet surface, whereby two GPVI molecules dimerize through their respective membrane-proximal Ig domains in the extracellular domain, and each GPVI molecule binds an FcR γ-chain homodimer in the transmembrane region. The two FcR γ-chain homodimers must be kept apart under basal conditions, in order to avoid activation, but the latter can be achieved upon collagen binding to initiate a signaling response.
Overall, we have demonstrated that surface oligomerization of GPVI in living cells occurs, and have provided evidence that this may also take place in platelets, where a protein multicomplex is formed that contains GPVI, FcR γ-chain, and possibly one or more other component(s). The correct expression and assembly of this multimolecular structure may be a prerequisite for formation of a receptor complex that retains the capacity to bind collagen with high affinity and trigger a signal cascade inside the cell. Establishing the organization of this complex has important consequences for our understanding of platelet regulation by collagen and for possible therapeutic strategies to fight cardiovascular disease.
O. Berlanga was supported by the BHF. T. Bori-Sanz was supported by the University of Birmingham (UK). J. R. James and S. J. Davis are funded by the Wellcome Trust. M. G. Tomlinson is an MRC research fellow. S. P. Watson holds a BHF chair.
Disclosure of Conflict of Interests
The authors states that they have no conflict of interest.