1 INTRODUCTION

The C-type lectin proteins belong to a large family containing C-type lectin-like domains.¹ These proteins activate receptors that are both cytoplasmic and transmembrane in a variety of cells.² C-type lectin receptors (CLRs) are functionally diverse and play an essential role in several processes including phagocytosis, endocytosis, pathogen recognition, and complement activation.^{1, 3} These CLRs are mainly expressed in immune cells but are also expressed on platelets.⁴ The immunoreceptor tyrosine-based activation motif (ITAM) contains two canonical YXXL motifs separated by a defined distance. Other C-type lectin receptors contain a single ITAM-like motif (hemITAM) with either YXXL or YXXXL in the intracellular domain through which they signal.⁵

CLEC-2 is highly expressed in platelets and is known to play a crucial role in the separation of lymphatic and blood vessels during development.^{6, 7} CLEC-2 is activated by podoplanin expressed on the endothelial cells at the junction of lymphatic and blood vessels causing platelet activation and plug formation.^{6, 7} The lack of CLEC-2 receptors results in the blood flowing into the lymphatic vessels.^{6, 7} CLEC-2 receptors can also be activated by the snake venom protein rhodocytin or by crosslinking the receptors with an antibody or by double crosslinking the antibody bound to CLEC-2 receptors with a secondary antibody.⁵ Such double crosslinking of the primary antibody by secondary antibodies has been used for clustering FcRIIA receptors in platelets.⁸ CLEC-2 receptors contain a single YXXL motif in the intracellular domain (hemITAM) that is phosphorylated upon agonist interaction by Src family kinases.⁵ Such phosphorylated hemITAM was proposed to bind to Syk by bridging two receptor molecules to one Syk molecule and activate Syk.^{9, 10} Once Syk is activated, it triggers a signaling cascade similar to ITAM signaling involving the linker for activation of T cells (LAT) signalosome, leading to phospholipase C gamma-2 (PLCγ2) activation.⁵ In this model, there is activation of PI3 kinase downstream of Syk that activates Tec family kinases that phosphorylate PLCγ2 in a redundant pathway.¹¹ Thus, inhibition of PI3 kinases or Tec kinases only leads to partial inhibition of PLCγ2 activation. While there was no direct evidence for such interaction, this model was based on the peptide interaction studies.^{9, 10}

We studied CLEC-2 signaling in platelets and proposed a different model for Syk activation by phosphorylated hemITAMs.¹² In this model, PI3 kinase is activated by binding to the phosphorylated hemITAM and generates phosphatidylinositol triphosphate (PIP3). Tec family kinases translocate to the membrane by binding to PIP3 and get activated, possibly involving phosphorylation by Src family kinases (SFKs). In support of our model, activated Tec family kinases (Btk and Tec) were shown to phosphorylate and activate Syk that is bound to the phosphorylated hemITAM.¹⁰ Hence, inhibition of either PI3 kinase or Tec family kinases leads to complete inhibition of hemITAM signaling. Recently, a study was published that showed no inhibition of rhodocytin-induced CLEC-2 signaling by the inhibition of Tec family kinases.¹³ Hence, we initiated our study to resolve the differences between these studies.

Using multiple pharmacological and molecular genetic approaches, we show that low-level activation of CLEC-2 receptors is influenced by the PI3 kinase/Tec family kinase/Syk activation pathway while increased crosslinking of the CLEC-2 receptors switches from this signaling to a mechanism that is primarily independent of Tec family kinases in the activation of Syk.

2 MATERIALS AND METHODS

3 RESULTS

3.1 Effect of ibrutinib or acalabrutinib on CLEC-2–mediated platelet activation

The CLEC-2 receptor can be activated by a number of agonists that cluster the receptor, including a monoclonal antibody against CLEC-2, the snake venom protein rhodocytin, and podoplanin, CLEC-2’s endogenous ligand expressed on cells.^{5, 16-18} Of these agonists, the monoclonal antibody is highly selective, can be added to platelets in a soluble form, and clusters two CLEC-2 molecules to initiate signaling. Hence, we used a CLEC-2 monoclonal antibody (mAb) to activate platelets and evaluated the effect of the Tec family kinase inhibitor ibrutinib. As shown in Figure 1A, the CLEC-2 mAb activated mouse platelets in a concentration-dependent manner, and ibrutinib (20 nM) abolished platelet aggregation and secretion induced by low concentrations of the CLEC-2 mAb, inhibited at moderate concentrations of CLEC-2 mAb but was without effect at high concentrations. These experiments were slightly more difficult to perform in human platelets, as human platelets express FcγRIIA receptors that can be activated by the CLEC-2 mAb. We used an antibody to CD32 (FcγRIIA), clone IV.3, to block these receptors and then performed the experiment with an antibody to human CLEC-2. Unlike mouse platelets, the human platelets required higher concentrations of a CLEC-2 mAb to elicit a response, possibly because human platelets express far less CLEC-2 receptors (≈2000 copies)¹⁹ compared to mouse platelets (about 40 000 copies).²⁰ As shown in Figure 1D, ibrutinib abolished human platelet aggregation and secretion when lower concentrations of the anti-human CLEC-2 mAb were used, but not when higher mAb was used. These data indicate that at low clustering signaling conditions, CLEC-2 depends on Tec family kinases but at higher clustering, CLEC-2 signals independently of Tec family kinases.

To evaluate the effects of ibrutinib on downstream signaling events, we measured phosphorylation of LAT and PLCγ2 under the conditions of low and high clustering of CLEC-2 in both mouse and human platelets. As shown in Figure 2, phosphorylation of Syk at 519/520, 346, and 342 were dramatically inhibited by ibrutinib at low CLEC-2 clustering conditions, but this inhibition was not significant at high clustering conditions. LAT tyrosine residue 191 was dramatically inhibited by ibrutinib in the low clustering conditions of CLEC-2 in mouse platelets. However, at higher clustering conditions, these phosphorylations were unaffected by ibrutinib (Figure 2E). Similar effects were observed on PLCγ2 tyrosine 1217 phosphorylation (Figure 2E), except that even at the high clustering conditions, PLCγ2 phosphorylation was significantly inhibited. This could be because of an additional role for PI3K/Btk downstream of Syk in the activation of PLCγ2.²¹ As shown in Figure 3, identical results of Syk, LAT, and PLCγ2 phosphorylations were obtained when human platelets were used with varied CLEC-2 clustering. Similar results were obtained when the second-generation Btk inhibitor acalabrutinib was used (Figures 4,5).

3.2 CLEC-2–induced platelet activation in Xid mouse platelets

A genetically altered mouse with an inactivating mutation in the Tec kinase Btk is available and has been used in our previous study with rhodocytin. To evaluate the role of Btk in CLEC-2 clustering-dependent platelet activation, we stimulated Xid mouse platelets with different concentrations of the CLEC-2 mAb. These data were compared to control mouse platelets as recommended by Jackson Laboratory. As can be seen in Figure 6, when low concentrations of CLEC-2 mAb were used, platelet activation was dramatically inhibited in Xid mouse platelets compared to control mouse platelets. However, as the concentration of mAb was increased, this inhibition of platelet activation was less apparent in Xid mouse platelets. Consistently, LAT and PLCγ2 phosphorylations were dramatically inhibited in Xid mouse platelets with low CLEC-2 clustering conditions, while LAT was unaffected with high clustering of CLEC-2 (Figure 6C-E). Phosphorylation of PLCγ2 on tyrosine 1217 was still inhibited at high clustering conditions, indicating that Btk may contribute to its phosphorylation (Figure 6E). These data suggest an essential role for Btk in low clustering conditions of CLEC-2, while Btk appears to be dispensable when CLEC-2 receptor clustering is very high. The reason for the lack of abolishment of platelet responses in low clustering conditions in Xid mice is that, although they lack a functional Btk, they still express Tec kinase, which might have a redundant function as Btk, similar to glycoprotein VI (GPVI) activation.²²

3.3 CLEC-2 activation of Src family kinase–deficient mouse platelets

Severin et al²³ have shown that aggregation was dramatically delayed in Lyn null mouse platelets when the platelets were activated with CLEC-2 mAb but not in Fyn or Src null platelets. However, they used a single concentration of CLEC-2 mAb. We investigated the effect of different concentrations of CLEC-2 mAb on the functional responses in Lyn null mouse platelets, to evaluate the role of Lyn under different clustering conditions of CLEC-2. As shown in Figure 7, platelet aggregation and secretion were abolished when a low concentration of the CLEC-2 mAb was used in Lyn null mice, but the functional responses began to recover as the concentration of CLEC-2 mAb was increased. Consistently, PLCγ2 and LAT phosphorylations were also abolished when low CLEC-2 mAb was used. When high mAb was used, PLCγ2 phosphorylation completely recovered (Figure 7E), although LAT phosphorylation recovered partially (Figure 7D).

4 DISCUSSION

CLRs, such as Dectin-1 and CLEC-2, have been known to signal through phosphorylation of a hemITAM motif in the cytoplasmic tail. The collagen receptor, which is not a CLR, also associates with the FcRγ chain and signals through ITAM motifs. In both cases, these ITAM and hemITAM motifs are phosphorylated by SFKs upon receptor clustering. The phosphorylated ITAM, containing two YXXL motifs separated by a specific distance, can bind to the two Src homology 2 (SH2) domains of Syk and activate Syk possibly by inducing a transition from autoinhibited to the active conformation and subsequent phosphorylation events. However, in the case of the hemITAM, there is only one YXXl/L motif that can be phosphorylated and only one SH2 of Syk that can bind to it. Hence, extending the model of ITAM binding, a model was proposed for hemITAM signaling in which each SH2 domain of Syk binds to one phosphorylated hemITAM on each receptor. This was a widely accepted model, although there is no direct evidence to support this model. Indirect evidence comes from the use of peptides from hemITAM that can bind to Syk and vice versa.

We have published that CLEC-2/hemITAM signaling involves PI3 kinase and Tec kinase activation that is essential for the stimulation of Syk.¹² In this model, we provided evidence that inhibition of either PI3 kinase or Tec kinases leads to abolished CLEC-2–mediated phosphorylation of Syk as well as its in vitro kinase activity.¹² We demonstrated that this was not the case for the ITAM-linked GPVI-mediated Syk activation.¹² However, a recent study indicated that ibrutinib, a Tec family kinase inhibitor, failed to inhibit 300 nM rhodocytin-induced platelet activation.¹³ Hence, we investigated the discrepancies between our studies.

While there are several agonists that can activate CLEC-2, the mAb against CLEC-2 provides the lowest activation as it can cluster two CLEC-2 molecules. The strength of CLEC-2 clustering can be enhanced by crosslinking the mAb with anti-IgG secondary antibodies in which four CLEC-2 receptors could be activated, as has been shown for the FcRIIA receptor on platelets.⁸ Rhodocytin is a snake venom protein and can crosslink 8 CLEC-2 molecules,²⁴ but its effects on other platelet receptors cannot be ruled out.²⁵ Finally, podoplanin, a physiological agonist, can only be used when it is expressed on cells, and its avidity depends on the expression levels and cannot be regulated accurately. CLEC-2 receptors have been shown to exist as monomers, oligomers, and oligomers of higher order on the platelet surface.⁹ These different structures of CLEC-2 might have varied affinity for the ligands and might proceed with activation of monomers to higher-order oligomers as the ligand concentrations increase.²⁶ This aspect is further complicated by the fact that human platelets express far fewer CLEC-2 receptors (≈2000 copies)¹⁹ compared to mouse platelets (≈40 000 copies).²⁰ Thus, work with mouse platelets, as they are more sensitive, requires much more careful design of the ligand concentrations.

We initiated our studies using a mAb against CLEC-2 to cluster under the lowest activation conditions using the lowest concentration of the mAb. Under these conditions, the signal is low but sufficient to activate platelets. Our data show that, under these conditions, ibrutinib abolished platelet activation induced by the CLEC-2 mAb. When the concentrations of mAb were increased, the inhibition by ibrutinib became less apparent. Thus, as the mAb concentration is increased, more CLEC-2 receptors are engaged in signaling, and this results in the decreased effect of ibrutinib. This effect was demonstrated in both human and mouse platelets (Figures 1-3). Consistently, ibrutinib abolished phosphorylations of LAT and PLCγ2 when a low concentration of the CLEC-2 mAb was used but not when higher mAb concentrations were used.

In our experience, platelet preparations play a key role in the responsiveness to the CLEC2 mAb minimal concentrations to activate platelets. We caution that blindly using one concentration of agonist to evaluate the effect of ibrutinib will not accurately represent the state of signaling, and one must do a titration of agonist to determine the lowest concentration to activate platelets. We could correlate the differences in this platelet reactivity to the desensitization of ADP receptors; the less responsive the ADP receptors are, the more mAb needed. The second point we want to emphasize is that the concentration of rhodocytin does not accurately represent the active amount of rhodocytin in the preparation.¹³ We have noticed that there is a great batch-to-batch variation of rhodocytin. While some batches are weak and require higher concentrations (up to 40 nM), other batches needed only 3 nM to activate platelets. Hence, one cannot blindly follow the concentration of rhodocytin in the publications and should titer each batch in the laboratory to determine the low and high concentrations required for platelet activation. In this context, we feel that 300 nM of rhodocytin used by Nicolson et al¹³ could be high, and that could be the reason they did not see any inhibition by ibrutinib.

The results obtained with ibrutinib were confirmed using Xid mice that lack a functional Btk (Figure 5). In the Xid mice, platelet aggregation and secretion were dramatically inhibited by low-avidity conditions of CLEC-2 clustering compared to wild-type mice. However, at high clustering conditions of CLEC-2 engagement, there was no difference between Xid and wild-type mouse platelet aggregation and secretion. The lack of complete inhibition of platelet aggregation in Xid mice at low clustering conditions is due to Tec kinase, which might be taking over the function of Btk to some extent. Consistently, the phosphorylations of LAT and PLCγ2 were also dramatically inhibited under low CLEC-2 clustering conditions but unaffected at high clustering conditions, compared to wild-type platelets. Consistent with the role of Tec kinases, we have previously shown that separation of lymphatic and blood vessels is disrupted only in the Btk/Tec double-knockout mice but not in single-knockout mice. These data indicate that Btk and Tec can serve redundant functions downstream of CLEC-2 receptors during development.

While the platelet functional studies are carried out with ex vivo platelets and low CLEC-2 clustering conditions require Btk for the functional activation of Syk, one wonders what the relevant physiological condition for CLEC-2 clustering may be. Under physiological conditions, the primary function of platelet CLEC-2 is to interact with endothelial podoplanin and form a microthrombus at the junctions of lymphatic and blood vessels, thereby aiding the lymphatic and blood vessel separation during development. Our data showing the disruption of lymphatic and blood vessel separation during development in Btk/Tec double-knockout mice demonstrates that the Tec kinases play a role in the CLEC-2 signaling under physiological conditions. If the physiological podoplanin engagement of platelet CLEC-2 were with high clustering then, based on the results where we show that Tec kinases are not important under high clustering conditions, we should not have seen dysfunctional lymphatic and blood vessel separation in Btk/Tec double-knockout mice.

Our studies extend the observations made by Severin et al²³ who reported that CLEC-2 -induced platelet aggregation was delayed in Lyn null mice with CLEC-2 mAb but not with rhodocytin. In our studies, we also show that under low clustering conditions, Lyn is absolutely necessary for platelet activation by CLEC-2 but becomes dispensable as more CLEC-2 receptors are engaged. Furthermore, specific phosphorylations on LAT and PLCγ2 are also abolished when low CLEC-2 mAb is used. Our studies are also consistent with the observations of Severin et al²³ that Lyn plays an important role when low-valency agonists of CLEC-2 are used and multivalent ligand engagement with podoplanin expressing Chinese hamster ovary cells or CLEC-2 IgM antibodies can bypass the Lyn-dependent pathways.

We connect the dependency of CLEC-2 signaling on Lyn and Tec family kinases at low clustering conditions and propose that Lyn initially phosphorylates the hemITAM motif on CLEC-2 under low clustering conditions (model outlined in Figure 8). Subsequently, PI3 kinase binds to the phosphorylated hemITAM and is activated, converting phosphatidylinositol bisphosphate (PIP2) to PIP3. Tec family kinases translocate to the membrane by binding to PIP3 and then phosphorylate and activate Syk, which might be bound to the phosphorylated hemITAM or in the cytosol. Lyn might also be involved in the phosphorylation of Y551 on PIP3-bound Btk, and this phosphorylation might be necessary for the Tec family kinase activation.²⁷ Hence, under low clustering CLEC-2 conditions, Lyn deficiency or inhibition of Tec family kinases with ibrutinib, Syk activation is abolished resulting in dramatically inhibited platelet functional responses. Interestingly, Fyn and Src, still present in the Lyn null platelets, are not able to phosphorylate CLEC-2 under these conditions, most likely due to the conformational requirements of CLEC-2 for these kinases. As the multivalent ligands are used, under high clustering engagement of CLEC-2, Lyn is dispensable and other SFKs can phosphorylate the hemITAM, and Syk is activated by alternate pathways. Interestingly, Severin et al²³ observed that the CLEC-2 cytoplasmic tail was not phosphorylated in Syk-deficient platelets, suggesting an essential role for Syk in this event. They also observed that pyrazolopyrimidine compound PP2, an inhibitor of SFKs, failed to inhibit CLEC-2 tyrosine phosphorylation, although it blocked CLEC-2–mediated platelet activation. Hence, they concluded that SFKs play an important role in the downstream signaling events. However, it is not clear how Syk is activated downstream of CLEC-2 if it must phosphorylate and bind to phosphorylated hemITAM. It is possible that if the hemITAM is phosphorylated, then the classical model of one Syk molecule binding to two CLEC-2 receptors through phosphorylated hemITAM occurs as the clustering increases and the CLEC-2 intracellular domains come closer. An alternative mechanism was proposed in which Syk could be activated by actin-cytoskeleton/ezrin, radixin, moesin proteins, by translocating Syk to the plasma membrane.²⁸ This has the potential to represent an alternate mode of Syk recruitment to the membrane. Thus, while we have supporting evidence for the involvement of Lyn and Tec family kinases under the conditions of low CLEC-2 clustering, we only have possible models of alternate pathways of Syk engagement and activation under a high CLEC-2 clustering condition, which certainly do not depend on Lyn or Tec kinases.

In conclusion, we provide evidence that CLEC-2 signals differently under low and high clustering conditions. Under low clustering conditions, CLEC-2 signaling depends on Lyn and Tec kinases, while under high clustering conditions, alternate pathways, without the involvement of Lyn and Tec family kinases, take over.

RELATIONSHIP DISCLOSURE

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

CD performed experiments, analyzed the data, and wrote the manuscript; HRV performed experiments; MW performed experiments; JCK performed experiments and analyzed the data including statistics; SPK was involved in conceptualization and overall supervision of the project.