We have recently shown that the C-type lectin-like receptor, CLEC-2, is expressed on platelets and that it mediates powerful platelet aggregation by the snake venom toxin rhodocytin. In addition, we have provided indirect evidence for an endogenous ligand for CLEC-2 in renal cells expressing HIV-1. This putative ligand facilitates transmission of HIV through its incorporation into the viral envelope and binding to CLEC-2 on platelets. The aim of the present study was to identify the ligand on these cells which binds to CLEC-2 on platelets. Recombinant CLEC-2 exhibits specific binding to HEK-293T (human embryonic kidney) cells in which the HIV can be grown. Furthermore, HEK-293T cells activate both platelets and CLEC-2-transfected DT-40 B-cells. The transmembrane protein podoplanin was identified on HEK-293T cells and was demonstrated to mediate both binding of HEK-293T cells to CLEC-2 and HEK-293T cell activation of CLEC-2-transfected DT-40 B-cells. Podoplanin is expressed on renal cells (podocytes). Furthermore, a direct interaction between CLEC-2 and podoplanin was confirmed using surface plasmon resonance and was shown to be independent of glycosylation of CLEC-2. The interaction has an affinity of 24.5±3.7 μM. The present study identifies podoplanin as a ligand for CLEC-2 on renal cells.
We have recently identified the C-type lectin-like receptor, CLEC-2, as a novel activating receptor that is expressed on the surface of platelets and megakaryocytes . CLEC-2 can be activated by the snake venom toxin rhodocytin or by a specific antibody, triggering powerful platelet aggregation . CLEC-2 signals through a single YXXL (Tyr-Xaa-Xaa-Leu) motif that is present in its cytoplasmic domain . Cross-linking of CLEC-2 induces Src kinase-dependent tyrosine phosphorylation of the YXXL sequence, inducing activation of the tyrosine kinase Syk and initiation of a signalling pathway that culminates in activation of PLCγ2 (phospholipase Cγ2) . The CLEC-2 signalling pathway is similar to, but distinct from, that used by receptors that signal through an ITAM (immunoreceptor tyrosine-based activation motif), such as the major signalling receptor for collagen on platelets, the GPVI (glycoprotein VI)–FcR (Fc receptor) γ-chain complex. These two signalling pathways differ in that activation of PLCγ2 by ITAMs has an absolute requirement for two YXXL motifs and the SLP-76 (Src homology 2 domain-containing leucocyte protein of 76 kDa)/Blnk family of adaptor proteins , whereas CLEC-2 has a single YXXL sequence and is only partially dependent on this family of adaptor proteins .
CLEC-2 is a member of the C-type lectin-like family of receptors, but its putative carbohydrate-recognition domain lacks the key structural features that confer binding to carbohydrates, suggesting that its endogenous ligand may be a protein. We have recently crystallized CLEC-2 and shown that it has a compact C-type lectin-like domain with a flexible loop on its surface . A role for this loop in ligand binding is suggested by mutagenesis and surface plasmon resonance binding studies with rhodocytin .
CLEC-2 has recently been reported to facilitate capture of HIV-1 by CLEC-2-transfected cells and platelets . Importantly, CLEC-2 does not bind to the HIV-1 envelope protein, suggesting that CLEC-2 binds to a membrane protein derived from the HIV-producing HEK-293T (human embryonic kidney) cells that is captured by the virus during viral budding. Thus HEK-293T cells must express an endogenous ligand for CLEC-2. Infection of renal cells and production of infective virus from these cells is well documented in HIV infection, and renal disease is an important complication of HIV infection [6,7]. Therefore this CLEC-2 interaction could be a significant mechanism in viral dissemination. The aim of the present study was to identify this ligand and to investigate its ability to bind and activate CLEC-2.
Recombinant proteins: cloning, production, purification and enzymatic biotinylation
For eukaryotic expression of glycosylated CLEC-2 protein, the extracellular domain of CLEC-2 was cloned into the plasmid pQCXIX (Clontech) to encode a protein with an N-terminal hexahistidine tag followed by a BirA recognition sequence. The packaging cell line HEK-293-GP2 was transfected with the CLEC-2 expression construct, and the retrovirus was harvested and used to infect HEK-293T cells. The cell supernatant was centrifuged at 2100 g for 15 min and filtered through a 0.45 μm filter, and protein was purified by nickel-affinity chromatography and size-exclusion gel filtration using a Superdex 75 HR 26/60 column on an AKTA Purifier (GE Healthcare).
For bacterial expression of unglycosylated protein, the extracellular domain of CLEC-2 was cloned into the plasmid pGMT7 to encode a protein with an N-terminal BirA recognition sequence. Recombinant CLEC-2 was expressed in Escherichia coli strain BL21(DE3)pLysS cells as inclusion bodies as described previously [4,8]. Briefly, inclusion bodies were isolated, washed in detergent, solubilized in guanidine and refolded in the presence of a redox couple, and refolded protein was purified by size-exclusion chromatography.
Podoplanin was made as described previously . Briefly, podoplanin was cloned into the plasmid pCDM7Ig, which was transiently transfected into HEK-293T cells to express podoplanin fused at its C-terminus to the human IgG1 Fc region. The fusion protein was purified by Protein A-affinity chromatography and size-exclusion chromatography.
BirA enzyme was produced and used to biotinylate proteins with BirA recognition tags as described previously . A 4-fold molar excess of biotinylated protein was conjugated to phycoerythrin-labelled extravidin (Sigma).
Flow cytometry and competition experiments
Cells were washed in cold PBS containing 0.01% sodium azide, blocked with 1% BSA for 10 min on ice, washed again and incubated with fluorochrome-labelled multimeric CLEC-2 protein (3.5 μM) for 1 h. Cells were then washed three times, fixed in PBS/1% (v/v) formaldehyde and analysed on a FACS Canto machine (Becton-Dickinson) and subsequently with FlowJo software (Tristar). For competition studies, cells were pre-incubated with unlabelled CLEC-2 protein, NKG2D, rat monoclonal anti-(human podoplanin) antibody NZ-1 (AngioBio) or isotype control for 1 h before the addition of multimeric fluorochrome-labelled CLEC-2. The monoclonal anti-(human podoplanin) antibody 18H5 (Santa Cruz Biotechnology) was also used for flow cytometry. For platelet analysis, whole blood was isolated in trisodium citrate, labelled directly, then diluted before being subjected to flow cytometry.
Functional studies in platelets and transfected DT-40 cell lines
Platelet-rich plasma was isolated as described previously  and placed in a Born lumiaggregometer with stirring at 1200 rev./min at 37 °C for 5 min before addition of HEK-293T cells suspended in PBS. In some experiments, platelets were pre-incubated with the GPIIbIIIa antagonist, integrilin (9 μM), or the Src kinase inhibitor, PD173952 (20 μM) (Pfizer), for 2 min before experimentation. Activation of CLEC-2 was monitored in DT-40 cells using an NFAT (nuclear factor of activated T-cells)–luciferase reporter assay following transient transfection of CLEC-2 or the Y7F mutant of CLEC-2, as described previously . The transfected DT-40 cells were incubated with HEK-293T cells for 6 h before measurement of luciferase activity. In competition experiments, HEK-293T cells were pre-incubated with 20 μg/ml rat IgG or rat anti-(human podoplanin) antibody NZ-1 on ice for 20 min before experimentation.
Metabolic labelling and Western blotting
Cells were resuspended in methionine-free medium and incubated for 1 h. Cells were then incubated overnight in 35S-labelled methionine (GE Healthcare), then washed and lysed on ice in 0.5% Nonidet P40, 0.25M NaCl and 10 mM Tris/HCl (pH 7.5) with protease inhibitors. Lysates were pre-cleared with Protein G–agarose beads (Sigma) and incubated with or without 100 μg of recombinant biotinylated CLEC-2 for 1 h on ice with gentle agitation. Anti-polyhistidine antibody (Sigma) was added for 1 h, then Protein G–agarose was added, and the mixture was rotated at 4 °C for 3 h. Beads were washed five times, and proteins were eluted in Laemmli reducing sample buffer and boiled for 5 min. Proteins were separated on SDS/12% PAGE gels, which were dried and exposed to photographic film. Western blots were performed in an equivalent manner without metabolic labelling. Gels were blotted on to nitrocellulose membranes, which were blocked in 5% (w/v) non-fat dried skimmed milk powder, washed and incubated with 1 μg/ml NZ-1, washed and incubated with HRP (horseradish peroxidase)-conjugated anti-(rat immunoglobulin) antibody (Dako).
Surface plasmon resonance binding studies
Surface plasmon resonance binding studies were conducted using a Biacore T100 machine, as described previously . The biological activities of recombinant CLEC-2 and podoplanin were established by demonstrating binding of each fluorescently labelled protein to cells transfected with the other protein. Proteins were attached to the carboxymethylated dextran-coated surface of CM5 biosensor chips, using amine-coupling chemistry, and experiments were performed in 10 mM Hepes (pH 7.4), 150 mM NaCl, 3 mM EDTA and 0.005% polysorbate 20 surfactant. Non-specific interactions were controlled for by subtraction of the signal from a reference flow cell coated with an irrelevant protein. All CLEC-2 protein used in binding studies was monomeric. To avoid avidity effects with podoplanin, which was expressed as a dimeric Fc-fusion protein, experiments were only performed with podoplanin immobilized on the chip surface. Kd values were obtained by non-linear curve fitting of the Langmuir isotherm to the data using the Levenberg–Marquardt algorithm as implemented in the program Origin (Microcal Software).
CLEC-2 protein binds to a cell-surface ligand on HEK-293T cells
To confirm that HEK-293T cells express a ligand for CLEC-2 and to provide a route to its identification, we expressed the extracellular domain of CLEC-2 in eukaryotic cells as a recombinant protein fused at its N-terminus to a BirA recognition sequence . This allowed site-specific enzymatic biotinylation at the membrane insertion point of the protein. Biotinylated protein was tethered to a fluorochrome-labelled avidin derivative with four biotin-binding sites to create a species that is tetrameric with respect to CLEC-2 and in which CLEC-2 retains the orientation that it would have on the cell surface, with the C-terminus of the protein available for interaction with a ligand. Flow-cytometric studies using this fluorochrome-labelled CLEC-2 demonstrated binding of CLEC-2 to HEK-293T cells, consistent with expression of an endogenous ligand for CLEC-2 on these cells (Figure 1A). No binding was seen with the fluorochrome-labelled extravidin alone. To confirm that the interaction of multimeric CLEC-2 protein was specific, binding was carried out in the presence of free CLEC-2. Increasing concentrations of free unlabelled CLEC-2 inhibited binding of the fluorescently labelled protein in a concentration-dependent manner, consistent with CLEC-2 exhibiting a specific mode of binding (Figures 1A and 1B). In contrast, incubation with a range of concentrations of a free irrelevant protein (NKG2D) did not inhibit binding of fluorescent CLEC-2 (Figure 1B).
CLEC-2 binds specifically to HEK-293T cells
To confirm that the binding seen was not artefactual due to the multimeric nature of the CLEC-2 used, the experiment was repeated using monomeric CLEC-2. When monomeric biotinylated CLEC-2 was bound to cells, it could be detected by the addition of fluorochrome-labelled extravidin (Figure 1C). However, the binding was slightly less, as predicted by the loss of avidity effects in the primary binding of monomeric CLEC-2 to the cells.
A range of other cell types were also screened for binding to the CLEC-2 tetramer. High levels of specific binding (i.e. binding that could be competed out with unlabelled CLEC-2), comparable with that seen on HEK-293T cells, were observed on SuSa cells, which are derived from testis. Intermediate levels of specific binding were observed on HT1080 and U2OS cells, which are derived from connective tissue and osteocytes respectively, and low-level binding was observed on the erythroid/megakaryocyte progenitor cell line, K562, and on ovarian SKOV3 cells. In contrast, there was no specific binding on T-cells (Jurkat), B-cells (Raji, Daudi, CIR), natural killer cells (NK92), mammary epithelial cells (T47D), monocytes (U937 and HL60), platelets or primary endothelial cells [HUVECs (human umbilical vein endothelial cells)]. These data are summarized in Table 1, and representative binding as measured by flow cytometry is shown in Figure 1(D).
|Cell line||Glycosylated fluorescent CLEC-2||Non-glycosylated fluorescent CLEC-2||Anti-podoplanin antibody|
|Cell line||Glycosylated fluorescent CLEC-2||Non-glycosylated fluorescent CLEC-2||Anti-podoplanin antibody|
Glycosylation of CLEC-2 is not required for ligand binding
CLEC-2 is able to interact with a range of different cell types of diverse origin, suggesting that the ligand is widely expressed. Native CLEC-2 on platelets is glycosylated , and recombinant CLEC-2 has a molecular mass of ∼33 kDa, compared with the predicted molecular mass of ∼26 kDa (Figure 2A). Treatment of this protein with PNGaseF (peptide N-glycosidase F), which removes N-linked glycans, decreased the molecular mass to its predicted value (Figure 2A), confirming that recombinant CLEC-2 is glycosylated. To establish whether the binding of CLEC-2 to its ligand was dependent on carbohydrate groups on CLEC-2, unglycosylated CLEC-2 was produced in E. coli. This protein was biotinylated and tethered to phycoerythrin-labelled extravidin and used for flow cytometry in the same manner as the CLEC-2 expressed from eukaryotic cells. Unglycosylated CLEC-2 binds to primary and cultured cells in a similar manner to glycosylated CLEC-2, as illustrated in Figure 2(B) and Table 1. Two other irrelevant C-type lectin-like molecules, made in a similar way in E. coli and refolded from inclusion bodies, did not show any binding to the cells (results not shown). In addition, PNGaseF-treated CLEC-2, which was made in eukaryotic cells, also bound to cells specifically, as demonstrated in Figure 2(C). These observations demonstrate that sugar moieties on CLEC-2 are not necessary for CLEC-2 to bind to its ligand. Thus recombinant CLEC-2 binds specifically to the surface of a number of cell types, including HEK-293T cells, and this binding is independent of its glycosylation. These data strongly suggest that one, or possibly more than one, transmembrane protein functions as an endogenous ligand for CLEC-2.
Glycosylation of CLEC-2 is not required for ligand binding
HEK-293T cells induce platelet aggregation and activate CLEC-2-transfected DT-40 cells
The above results demonstrate the presence of a ligand for CLEC-2 on HEK-293T cells and confirm our previous indirect evidence for this . Studies were therefore undertaken to investigate whether this putative ligand is able to activate platelets. Figure 3(A) demonstrates that HEK-293T cells stimulate powerful aggregation of platelets, with the magnitude of response being determined by the number of HEK-293T cells used. Importantly, this response was inhibited by the GPIIbIIIa integrin antagonist, integrilin, demonstrating that it is true integrin-dependent aggregation rather than agglutination (Figure 3B). Furthermore, activation by HEK-293T cells was inhibited in the presence of the Src kinase inhibitor, PD173952 (Figure 3B). This is consistent with the observation that CLEC-2 signals through a Src kinase-dependent pathway . Moreover, HEK-293T cells also stimulated activation of FcR γ-chain-deficient mouse platelets (results not shown), demonstrating that activation is independent of the platelet collagen receptor GPVI, the other major platelet activation receptor that induces aggregation via Src kinases.
HEK-293T cells induce platelet aggregation and activate CLEC-2-transfected DT-40 cells
To confirm that HEK-293T cells are able to activate platelets via CLEC-2, we performed a series of studies on CLEC-2-transfected DT-40 cells, using an NFAT reporter assay . Activation of CLEC-2 in this model cell line leads to induction of NFAT activity and expression of luciferase, which can be measured by luminometry. Strikingly, exposure to HEK-293T cells increases NFAT activity in CLEC-2-transfected, but not mock-transfected, DT-40 cells (Figure 3C), confirming that they express a ligand that is able to activate the C-type lectin-like receptor. Moreover, DT-40 cells transfected with a mutant of CLEC-2 in which the cytoplasmic motif YXXL was changed to FXXL (Phe-Xaa-Xaa-Leu) (Figure 3C) were not activated by HEK-293T cells. A similar result has been reported previously for activation of CLEC-2 by rhodocytin in DT-40 cells . Thus these results confirm that HEK-293T cells express a ligand for CLEC-2 that induces activation of the C-type lectin-like receptor through the same pathway as that used by the snake toxin rhodocytin.
HEK-293T cells express the CLEC-2 ligand podoplanin
To characterize the surface proteins in HEK-293T cells that bind to CLEC-2, pull-down experiments were performed with recombinant CLEC-2 on cells that had been metabolically labelled with [35S]methionine. This approach identified several minor bands and a major band of 36 kDa that were not present in the control lane (Figure 4A). During the course of these studies, podoplanin, a 36 kDa protein that is expressed on certain tumours, was shown to interact with CLEC-2 . Podoplanin derives its name from podocytes which express it [14,15], and which are kidney epithelial cells and therefore closely related to the human embryonic kidney HEK-293T cells.
Co-precipitation demonstrates a podoplanin–CLEC-2 interaction
Using a specific antibody against podoplanin, we have demonstrated that recombinant CLEC-2 precipitates podoplanin from HEK-293T cells and that this migrates in the same region as the 36 kDa protein identified in the metabolic labelling studies (Figure 4B). Furthermore, we have confirmed expression of podoplanin on HEK-293T cells, with strong staining using an anti-podoplanin antibody (Figure 5A), comparable with the staining seen with recombinant CLEC-2 (Figure 1). Moreover, binding of the podoplanin antibody paralleled that of binding of recombinant CLEC-2 on a range of cell lines as shown in Figure 5(B) and Table 1. Thus these observations strongly suggest that podoplanin is the ligand that confers binding of HEK-293T and a number of other cell lines on CLEC-2. Direct confirmation that this was the case was obtained using a specific antibody against podoplanin, NZ-1, that has been shown to block binding of podoplanin-expressing cells to platelets . As shown in Figure 5(C), NZ-1 inhibited binding of recombinant CLEC-2 to HEK-293T cells in a similar way to that observed with unlabelled CLEC-2 (Figure 1), confirming that binding was mediated by podoplanin. Moreover, NZ-1 also blocked activation of CLEC-2-transfected DT-40 cells by HEK-293T cells (Figure 3D), thereby confirming that HEK-293T cells bind and activate CLEC-2 through podoplanin. In addition, recombinant podoplanin was able to activate CLEC-2-transfected DT-40 cells directly (Figure 5D), confirming that the interaction between podoplanin and CLEC-2 triggers signalling. Furthermore, the signalling was dependent on the YXXL sequence of CLEC-2, confirming that it was mediated through the same mechanism as that used by rhodocytin (Figure 5D).
Podoplanin is a ligand for CLEC-2 on HEK-293T cells
CLEC-2 and podoplanin interact directly with an affinity of 24.5±3.7 μM
Thus this work confirms not only that podoplanin is a specific ligand for CLEC-2, but also the biological activity of both the recombinant CLEC-2 and the recombinant podoplanin that was used in these studies. In order to confirm a direct interaction between CLEC-2 and podoplanin and determine the affinity of the interaction, we have used the two recombinant proteins in surface plasmon resonance binding studies (Figure 6). For these studies, podoplanin was tethered to a biosensor surface and interacted with different concentrations of recombinant CLEC-2. A progressively higher signal was seen with increasing concentrations of CLEC-2 protein. No interaction was seen with control proteins. The data were consistent with a single-site-binding model and demonstrate that the interaction between the two proteins is direct, with an affinity of 24.5±3.7 μM.
CLEC-2 interacts directly with podoplanin
CLEC-2 is a newly characterized C-type lectin-like receptor, which has been shown to mediate platelet activation and aggregation upon binding to rhodocytin, a snake venom protein . CLEC-2 has also been shown to enhance infectivity of HIV-1 produced in HEK-293T cells . This effect was not mediated by the viral envelope protein, implying that a protein from the HEK-293T cells was captured during viral budding and was responsible for the interaction of the virus with CLEC-2. These data suggested the presence of an endogenous surface-bound ligand for CLEC-2 on HEK-293T cells. We used recombinant CLEC-2 protein to confirm that HEK-293T cells express a ligand for the receptor. This recombinant CLEC-2 was used to pull down potential ligand molecules from cell lysates and identified a candidate ligand of 36 kDa that corresponds in size to podoplanin. We have shown that HEK-293T and other CLEC-2-binding cells used in this study express podoplanin and that anti-podoplanin antibody inhibits both binding to and activation of CLEC-2 by HEK-293T cells. A direct interaction between podoplanin and CLEC-2 was demonstrated formally by co-immunoprecipitation and surface plasmon resonance experiments, which determined the affinity to be 24.5 μM.
These results are consistent with a report that certain cancer cells express podoplanin on their surface, which allows them to interact with CLEC-2 . The study by Suzuki-Inoue et al.  sought to identify a receptor for podoplanin and demonstrated that Src inhibitors blocked podoplanin-induced signalling through CLEC-2. The present study defines the signalling in CLEC-2 that is triggered by podoplanin and demonstrates that it is mediated through the YXXL motif in the cytoplasmic domain of CLEC-2. We have demonstrated further that glycosylation of CLEC-2 is not required for this interaction. There is good evidence that HIV infects renal cells, that these cells can produce infective HIV virions and that renal cells express podoplanin which may be incorporated into the budding virion [6,14]. The present study has demonstrated that podoplanin binds directly to CLEC-2 with an affinity in the micromolar range, so inhibitors of this interaction may help to prevent dissemination of HIV-1 by platelets.
Podoplanin was originally identified on a range of cells, including airway epithelia, fibroblasts, keratinocytes, osteoblasts and renal tubular epithelial cells [17–20]. Podoplanin is also expressed at high levels on lymphatic endothelial cells  and on certain tumour cells, where it has been shown to activate platelets . In the renal glomerulus, podoplanin helps to maintain the structure of podocytes, which are required for efficient glomerular filtration . Podocytes are not normally in contact with blood, but acute renal damage could expose them to blood, allowing interaction between CLEC-2 on platelets and podoplanin on podocytes. Physiologically, this would trigger platelet activation and formation of a haemostatic plug, thereby preventing further leakage into the collecting ducts, and possibly helping in renal repair through release of growth factors. However, this could also trigger intrarenal and especially intraglomerular thrombosis, which can be observed in renal diseases when there is renal epithelial damage, including during renal transplant rejection.
Inhibitors of the podoplanin–CLEC-2 interaction have potential use as novel anti-platelet agents. The significance of the interaction of podoplanin and CLEC-2 is unclear, as the majority of cells that express podoplanin, with the exception of tumour cells undergoing metastasis, do not normally come into contact with platelets. As with certain other C-type lectin-like receptors, it is possible that the function of CLEC-2 is primarily in defence, and specifically to prevent excessive blood loss and to respond to infection following contact with podoplanin. Further studies will provide information on the physiological and pathophysiological significance of the interaction of podoplanin and CLEC-2.
This work was supported by the Medical Research Council, the British Heart Foundation and the Wellcome Trust. We are grateful to Paul Harrison, Norma Masson, Neale Banerji, Alain Townsend, Hal Drakesmith, Lisa Schimanski, Emma Sweetland and Da Lin for helpful discussions.
These authors contributed equally to this work.