CalDAG-GEFI deficiency protects mice in a novel model of Fcγ RIIA-mediated thrombosis and thrombocytopenia.

Platelet activation via Fcγ receptor IIA (FcγRIIA) is a critical event in immune-mediated thrombocytopenia and thrombosis syndromes (ITT). We recently identified signaling by the guanine nucleotide exchange factor CalDAG-GEFI and the adenosine diphosphate receptor P2Y12 as independent pathways leading to Rap1 small GTPase activation and platelet aggregation. Here, we evaluated the contribution of CalDAG-GEFI and P2Y12 signaling to platelet activation in ITT. Mice transgenic for the human FcγRIIA (hFcR) and deficient in CalDAG-GEFI(-/-) (hFcR/CDGI(-/-)) were generated. Compared with controls, aggregation of hFcR/CDGI(-/-) platelets or P2Y12 inhibitor-treated hFcR platelets required more than 5-fold and approximately 2-fold higher concentrations of a FcγRIIA stimulating antibody against CD9, respectively. Aggregation and Rap1 activation were abolished in P2Y12 inhibitor-treated hFcR/CDGI(-/-) platelets. For in vivo studies, a novel model for antibody-induced thrombocytopenia and thrombosis was established. FcγRIIA-dependent platelet thrombosis was induced by infusion of Alexa750-labeled antibodies to glycoprotein IX (CD42a), and pulmonary thrombi were detected by near-infrared imaging technology. Anti-GPIX antibodies dose-dependently caused thrombocytopenia and pulmonary thrombosis in hFcR-transgenic but not wild-type mice. CalDAG-GEFI-deficient but not clopidogrel-treated hFcR-transgenic mice were completely protected from ITT. In summary, we established a novel mouse model for ITT, which was used to identify CalDAG-GEFI as a potential new target in the treatment of ITT.


Introduction
Platelets are essential components of the hemostatic response to vascular injury. However, platelets also play a role in pathologic conditions, such as atherothrombosis, and in immune-mediated thrombocytopenia and thrombosis (ITT). Several ITT syndromes, including heparin-induced thrombocytopenia and thrombosis (HIT), 1-3 bacterial sepsis-associated thrombocytopenia and disseminated intravascular coagulation, 4,5 and the thrombotic manifestations of antiphospholipid syndromes 6 are characterized by immunemediated platelet activation through the platelet Fc␥ receptor, Fc␥RIIA. In addition, thrombotic complications have been observed with the expanded use of therapeutic IgG antibodies, such as bevacizumab. [7][8][9] One of the barriers to successful treatment of these thrombotic syndromes is that therapeutic targeting of platelet activation pathways to prevent thrombosis is either not effective or comes with an inherent risk of bleeding complications.
In humans, Fc␥RIIA is expressed on platelets, neutrophils, monocytes, and macrophages and activates these cells following the binding of the Fc region of IgG-coated cells or IgG-containing immune complexes. 10 Mice lack the genetic equivalent of human Fc␥RIIA and, indeed, do not express a platelet Fc receptor. Consequently, most of the studies on the role of Fc␥RIIA in platelet activation were entirely dependent on the use of inhibitors, and they did not provide information on whether or not these inhibitors would reduce the risk of excessive platelet activation as observed in the clinical settings of ITT or HIT. To circumvent this limitation, we generated and characterized human Fc␥RIIA-transgenic (hFcR) mice in which Fc␥RIIA is expressed on mouse platelets and macrophages at levels equivalent to that in human cells. 11 For studies on HIT, we further crossed hFcR mice with mice deficient for mouse PF4 but transgenic for human PF4. 12 Using these mouse models, we demonstrated a critical role for Fc␥RIIA expression in antiplatelet antibody-induced and heparin-induced thrombocytopenia and thrombosis in vivo. [11][12][13][14] Fc␥RIIA is unique among the activating Fc␥ receptors in that its cytoplasmic tail contains an immunoreceptor tyrosine-based activation motif. 15 Residues in the immunoreceptor tyrosine-based activation motif domain become rapidly phosphorylated on receptor engagement and induce cell activation after binding by nonreceptor protein tyrosine kinases, such as spleen tyrosine kinase. 16,17 It is widely accepted that stimulation of Fc␥RIIA on platelets stimulates spleen tyrosine kinase, leading to PLC␥ activation and the generation of the second messengers Ca 2ϩ and DAG. In our recent work, we have described a central role for Ca 2ϩ and diacylglycerol regulated guanine nucleotide exchange factor I (CalDAG-GEFI) in Ca 2ϩ -dependent platelet activation. [18][19][20][21] CalDAG-GEFI catalyzes the activation of the small GTPases Ras-proximate (Rap)1 and Rap2. In platelets, Rap1B accounts for 90% of the total Rap protein, 22 and its importance in ␣IIb␤3 activation was demonstrated in Rap1B-deficient mice. 23 Our studies with CalDAG-GEFI Ϫ/Ϫ mice in combination with inhibitors to protein kinase C or the Gi-coupled adenosine diphosphate receptor, P2Y12, identified a 2-pathway model for integrin activation The online version of this article contains a data supplement.
The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked ''advertisement'' in accordance with 18 USC section 1734.
downstream of PLC activation. CalDAG-GEFI is a high-affinity sensor for Ca 2ϩ , which mediates the rapid but reversible activation of ␣IIb␤3. In the absence of CalDAG-GEFI, Rap1/integrin activation is delayed but sustained and depends on signaling by protein kinase C and P2Y12. Notably, our studies further suggested that CalDAG-GEFI is particularly important for platelet activation through GPVI, the immunoreceptor tyrosine-based activation motifcoupled platelet collagen receptor. 21 In this study, we evaluated the contribution of 2 critical platelet signaling pathways, Ca 2ϩ /CalDAG-GEFI/Rap1 and P2Y12/Rap1, to platelet activation in ITT. Deficiency in CalDAG-GEFI, and to a lesser extent inhibition of P2Y12, provided protection from Fc␥RIIA-mediated platelet aggregation both in vitro and in vivo.

Mice
Mice transgenic for hFc␥RIIA, 11 CalDAG-GEFI Ϫ/Ϫ , 18 and littermate control wild-type (WT) mice were bred on a C57Bl6/J strain background and housed in the mouse facility of Thomas Jefferson University. Experimental procedures were approved by the Animal Care and Use Committee of Thomas Jefferson University. Where indicated, mice were treated with clopidogrel 24 and 3 hours before the experiment at a dosage of 75 mg/kg body weight orally by gavage feeding.

Flow cytometry
Surface expression of Fc␥RIIA. Blood was obtained from the retroorbital plexus of anesthetized mice. Blood samples were stained with a phycoerythrin-labeled antibody against GPIb␣ and an Alexa488-labeled antibody against Fc␥RIIA (IV.3). Surface expression of Fc␥RIIA was determined as the mean fluorescence intensity in channel 1 (FL-1) for all events that stained positive for FL-2.
Platelet count. Blood samples were stained with a phycoerythrinlabeled antibody against ␣IIb␤3, and platelets were counted by flow cytometry gating for FL-2 (phycoerythrin) positive events. Platelet counts at t ϭ 0 were defined as 100%.

Aggregometry
Light transmission was measured in PRP or washed platelets activated under stirring conditions at 37°C. Inhibitors and agonists were added at the indicated concentrations, and light transmission was recorded for 10 to 15 minutes on a Chrono-log 4-channel optical aggregation system (Chronolog).

Western blotting
Proteins were separated by SDS-PAGE on 10%-20% gradient gels and transferred to polyvinylidene fluoride membranes. Standard Western blotting procedures were used. Rap1 was detected, and band intensity was quantified with the Odyssey Infrared Imaging System (LICOR Biosystems).

Thromboxane generation assay
Platelets were stimulated in standard aggregometry. At different time points, 50 L of sample was withdrawn and 5mM EDTA /1mM acetylsalicylic acid was added. The cells were removed by centrifugation, and TxB 2 levels were determined in the supernatant by enzyme immunoassay. TxB 2 is the stable product of the nonenzymatic hydration of thromboxane (TxA 2 ), which itself has a half-life of only 37 seconds under physiologic conditions.

In vivo immunoglobulin-mediated thrombosis model
Animals were matched according to their platelet Fc-receptor expression before the experiment. Mice were anesthetized with isoflurane, and anti-GPIX-750 was administered by retro-orbital injection at a dose of 0.5 g/g or 1 g/g body weight. After 4 hours, the mice were anesthetized and killed by cervical dislocation. Laporotomy and thoracotomy were performed to gain access to the internal organs. The organs were carefully flushed with 2 mL PBS by left ventricular and right ventricular puncture to perfuse the lung and systemic vessels. After flushing, the organs were harvested, cleared from fatty and connective tissue, and incubated in 4% paraformaldehyde for 24 hours before they were read on a conventional LICOR Odyssey scanner (LICOR).
Pearl imaging method. WT animals and hFcR animals were anesthetized with 100 mg/kg ketamine (Fort Dodge Animal Health) and 10 mg/kg xylazine (Lloyd Lab), the abdominal and thoracic region was shaved with an electric clipper, and the remaining hair was removed with a depilatory cream. Subsequently, the mice were injected with an anti-GPIX-750 antibody. After 4 hours, bio imaging was performed using a Pearl Impulse in vivo imaging chamber (LICOR Biosystems) at 800 nm. H&E staining. Extracted lungs were cryosectioned (8-m sections) and stained with H&E at the histology core within the Pathology Department at Thomas Jefferson University.

Statistics
Results are reported as mean Ϯ SEM. Statistical significance was assessed by unpaired 2-tailed Student t test. A P value Ͻ .05 was considered significant.

Results
To evaluate the role of CalDAG-GEFI in Fc␥RIIA-mediated platelet activation, we generated mice transgenic for hFcR and deficient in CalDAG-GEFI (hFcR/CDGI Ϫ/Ϫ ). Staining of platelets for surface-expressed Fc␥RIIA demonstrated variation in its expression level from mouse to mouse. Thus, for further experiments, we selected mice with similar platelet Fc␥RIIA expression ( Figure 1). Importantly, the surface expression of GPIb␣, GPIX, GPVI, ␣IIb␤3 integrin, and ␤1 integrins on platelets from hFcR/CDGI Ϫ/Ϫ mice did not differ significantly compared with platelets from WT or hFcR mice (supplemental Table 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article).
More specifically, hFcR/CDGI Ϫ/Ϫ platelets lacked the early phase of Rap1 activation, whereas 2-MeSAMP-treated hFcR platelets were defective in sustained activation of the small GTPase. Rap1 activation was completely abolished in 2-MeSAMP-treated hFcR/ CDGI Ϫ/Ϫ platelets. An important platelet response controlled by Rap1 is the generation of the second wave mediator thromboxane (TxA 2 ). 21 As expected, the generation TxB 2 , a stable analog of TxA 2 , was almost completely abolished in 2-MeSAMP-treated hFcR/CDGI Ϫ/Ϫ platelets ( Figure 3C). Deficiency in CalDAG-GEFI led to a significant reduction in TxB 2 formation, whereas inhibition of P2Y12 alone did not affect TxB 2 release under these experimental conditions. Importantly, addition of an exogenous thromboxane analog (U46619) restored aggregation in hFcR/CDGI Ϫ/Ϫ platelets activated with a threshold dose of anti-CD9 antibodies, whereas either agonist alone failed to induce aggregation ( Figure 3D). Inhibition of TxA 2 generation by the COX-1 inhibitor acetylsalicylic acid (aspirin), however, had only a mild inhibitory effect on the aggregation of both hFcR and hFcR/CDGI Ϫ/Ϫ platelets (supplemental Figure 2). This impairment was much weaker than that observed in hFcR/CDGI Ϫ/Ϫ platelets in response to anti-CD9 (Figure 2), indicating that (1) CalDAG-GEFI predominantly affects Fc␥RIIA-dependent platelet aggregation because of its critical role in integrin ␣IIb␤3 activation, and (2) the defect in TxA 2 feedback is critical only in the absence of CalDAG-GEFI-mediated integrin activation.
Next, we sought to determine the contribution of CalDAG-GEFI and P2Y12 signaling to Fc␥RIIA-mediated platelet activation in ITT. In our previous work, we showed that infusion of antibodies against CD9 (rat IgG2a isotype) causes thrombocytopenia in mice independent of Fc␥RIIA and that Fc␥RIIA-independent platelet consumption protects mice from Fc␥RIIA-mediated thrombotic complications. 14 We also showed that antibodies to membraneproximal domains within the GPIb-IX subunit of the von Willebrand factor receptor complex have minimal effects on the survival of circulating platelets in regular WT mice, irrespective of their isotype. 24,25 Thus, we speculated that antibodies to GPIX could be well suited to induce Fc␥RIIA-dependent thrombocytopenia and thrombosis in mice. In vitro, antibodies to GPIX caused aggregation of U46619-primed hFcR but not hFcR/CDGI Ϫ/Ϫ platelets ( Figure 4A). Inhibition of P2Y12 signaling partially inhibited anti-GPIX-induced aggregation of hFcR platelets. Consistent with these findings, we observed that anti-GPIX antibodies induced Rap1 activation in hFcR platelets only in the presence of U46619. Whereas Rap1 activation induced by anti-GPIX/U46619 was abolished in hFcR/CDGI Ϫ/Ϫ platelets, it was partially reduced in 2-MeSAMP-treated hFcR platelets ( Figure 4B).
Infusion of antibodies to GPIX dose-dependently caused severe thrombocytopenia in hFcR but not regular WT mice ( Figure 5A), and thrombocytopenia was associated with the formation of pulmonary microthrombi ( Figure 5B). To quantify thrombus formation in the lungs, we injected mice with Alexa750 fluorophorecoupled antibodies to GPIX followed by near-infrared scanning of the whole animal ( Figure 5B) or isolated lungs ( Figure 6).
When scanned on a Pearl in vivo Imaging Chamber (LICOR) 4 hours after infusion of the antibody, WT mice did show accumulation of the fluorescence dye in the liver and spleen, but not in the thorax. However, when injecting hFcR mice with the labeled GPIX antibody, there was a distinct accumulation of fluorescence in the thorax (Figure 5B), suggesting the formation of platelet-rich thrombi in the lungs. H&E-stained histology sections of extracted lungs confirmed the formation of microthrombi rich in platelets and fibrin ( Figure 5B inset). For studies on the role of CalDAG-GEFI and P2Y12 in experimental ITT, we decided to quantify ITT in lungs extracted 4 hours after infusion of the antibody. At a dosage of 0.5 g/g body weight of the antibody, both hFcR/CDGI Ϫ/Ϫ mice and clopidogrel-treated hFcR mice were completely protected from thrombocytopenia ( Figure 6A) and pulmonary thrombosis ( Figure 6C,E) compared with hFcR mice. At 1 g/g body weight, however, hFcR/CDGI Ϫ/Ϫ mice were still protected from ITT, whereas clopidogrel treatment did not significantly affect pulmonary thrombosis in hFcR mice. hFcR mice injected with 1 g/g anti-GPIX antibodies also displayed a shock phenotype with marked tactile hypothermia (not shown), severely reduced motor activity, and shallow, rapid, and irregular breathing (supplemental Video). Clopidogrel treatment did not prevent these anti-GPIX-induced shock-like symptoms in hFcR mice, whereas no symptoms were observed in hFcR/CDGI Ϫ/Ϫ mice.

Discussion
We here describe the generation of a novel mouse model for ITT, in which Alexa750 fluorophore-labeled antibodies against the platelet surface receptor GPIX are used (1) to selectively induce thrombocytopenia and thrombosis in Fc␥RIIA-positive but not WT Fc␥RIIAnegative mice and (2) to facilitate the rapid and quantitative visualization of pulmonary thrombi by near-infrared imaging technology. Using this cutting-edge technology, we were able to demonstrate that Fc␥RIIA-dependent platelet aggregation occurs in the systemic venous system and/or the lungs of mice and that pulmonary platelet accumulation was completely abolished in mice transgenic for hFc␥RIIA but deficient in CalDAG-GEFI. In contrast, clopidogrel bisulfate (Plavix) treatment only partially inhibited ITT in mice.
Studies in mice on the mechanisms regulating immunemediated thrombosis have been difficult for several reasons. First, mice do not express the activating Fc receptor, Fc␥RIIA, on circulating platelets and thus do not develop HIT/ITT. To circumvent this limitation, we previously generated and characterized mice transgenic for the human Fc␥RIIA receptor. 11 Another important limitation for studies on ITT in mice is based on the ability of many antibodies directed against mouse platelet surface receptors to induce thrombocytopenia independent of Fc␥RIIA expression. Platelet elimination by the reticuloendothelial phagocytic system, however, complicates studies on ITT, as Fc␥RIIAindependent platelet removal protects mice from Fc␥RIIAdependent thrombosis. 14 Clinically relevant antibodies/immune complexes, such as the heparin/PF4/anti-PF4 complex 12 or the heparin/vascular endothelial growth factor (VEGF)/anti-VEGF complex, 7 however, do not cause thrombocytopenia in the absence of platelet Fc␥RIIA. Finally, there is a lack of technologies that allow a simple but quantitative visualization of microvascular thrombi in the vital organs of ITT mice. So far, most studies relied on the peripheral platelet count in combination with hematoxylin and eosin tissue sections as a read-out for thrombotic complications. We here describe a novel mouse model for ITT in Fc␥RIIAtransgenic mice, which showed good correlation to HIT and antibody-induced thrombotic complications, both in vitro and in vivo. Our model also markedly simplifies the detection of tissue microthrombi, as it facilitates the quantitative visualization of thrombosis in antibody-treated mice. We selected antibodies to the GPIX subunit of the von Willebrand factor receptor complex to induce thrombocytopenia and thrombosis in an Fc␥RIIAdependent manner, as such antibodies show marginal effects on circulating platelet counts in control mice compared with antibodies against other platelet surface receptors, such as ␣IIb␤3, GPIb␣, or CD9. 14,24,25 In mice transgenic for hFcR, however, anti-GPIX antibodies caused severe thrombocytopenia within hours ( Figure  5). Thrombocytopenia was accompanied by the formation of fibrin-rich thrombi, particularly in the lungs ( Figures 5-6), marked tactile hypothermia, severely reduced motor activity, and shallow, rapid, and irregular breathing. Very similar findings have been made for hFcR/hPF4 double-transgenic mice after injection of heparin and anti-PF4 antibodies (S.E.M., unpublished observation) and for Fc␥RIIA-transgenic mice treated with the antiangiogenic drug bevacizumab in combination with VEGF and heparin. 7 Importantly, patients with acute HIT can develop a plethora of complications of arterial and venous thrombosis, including pulmonary thromboembolism, stroke, and myocardial infarction. Consistently, they can present with a variety of symptoms, including dyspnea, neurologic deficiencies, chest pain, and shock. 26,27 Interestingly, antibodies to GPIX are weak agonists for Fc␥RIIAdependent mouse platelet activation/aggregation in vitro ( Figure  4), similar to what has been described for immune complexmediated activation of these cells. 7 Consistent with this observation, it is also widely known that preactivation of low responding human platelets can produce positive Fc␥RIIA-dependent aggregation responses to HIT complexes. 28,29 Thus, our findings that Fc␥RIIA-dependent mouse platelet activation by anti-GPIX antibodies benefits from priming of these cells are consistent with observations on immune complex-mediated activation in human and murine platelets.
We have previously demonstrated that CalDAG-GEFI is critical for Ca 2ϩ -dependent activation of Rap1 and ␣IIb␤3 integrin in platelets activated by thrombin or collagen, especially under physiologic flow conditions. 18,20,21 We here demonstrate that CalDAG-GEFI is critical for Fc␥RIIA-mediated activation of Rap1 and ␣IIb␤3 in platelets stimulated by antibodies directed against the surface receptors CD9, GPIX, or Fc␥RIIA itself (Figures 2 and  4; supplemental Figure 1). Compared with hFcR control platelets, aggregation of hFcR/CDGI Ϫ/Ϫ platelets required only a 2-fold increase in the concentration of PAR4 receptor-activating peptide, whereas more than 5-fold higher concentrations of stimulating antibodies (anti-CD9) were needed to aggregate hFcR/CDGI Ϫ/Ϫ cells. Feedback activation via P2Y12, however, was equally important for PAR4-and Fc␥RIIA-dependent platelet activation, as preincubation of hFcR platelets with 2-MeSAMP caused an approximately 2-fold shift in the dose-response to PAR4p or anti-CD9 antibodies. Consistent with previous findings on the role of Ca 2ϩ /CalDAG-GEFI signaling for platelet activation via the collagen receptor, GPVI, 21 TxA 2 release from antibody-stimulated hFcR/CDGI Ϫ/Ϫ platelets was markedly reduced and addition of exogenous TxA 2 restored aggregation in hFcR/CDGI Ϫ/Ϫ platelets stimulated via Fc␥RIIA. Importantly, inhibition of TxA 2 synthesis had only a minor effect on the aggregation response of hFcR platelets, demonstrating that CalDAG-GEFI is critical for Fc␥RIIAdependent platelet activation because of its critical role in both integrin ␣IIb␤3 activation and TxA 2 biosynthesis, a finding that is well in line with the documented role of TxA 2 in Fc␥RIIAmediated platelet activation. 30,31 Consistent with these in vitro findings, we observed complete protection from ITT in hFcR/CDGI Ϫ/Ϫ mice, whereas clopidogrel treatment was not effective when mice were challenged with a higher dose of anti-GPIX antibodies. This limited protective effect of clopidogrel in the setting of experimental ITT is in accordance with clinical studies demonstrating incomplete protection from HIT in patients receiving both aspirin and clopidogrel. 32 Our studies may explain the limited effect of P2Y12 inhibitors in the setting of HIT/ITT. P2Y12 inhibitors affect the sustained activation of Rap1, important for stable platelet adhesion at sites of vascular injury. However, they do not affect the rapid, CalDAG-GEFImediated activation of Rap1, which is sufficient for the formation of reversible platelet aggregates under low and high shear stress conditions. 33 In HIT/ITT, these aggregates occur systemically and lead to a permanent obstruction of the pulmonary microcirculation. In contrast, signaling via the Ca 2ϩ /CalDAG-GEFI/Rap1 signaling module is critical in the initiation of Fc␥RIIA-dependent platelet activation. Deficiency in CalDAG-GEFI results in (1) a marked right shift in the dose-response to Fc␥RIIA agonists and (2) a significant delay in Fc␥RIIA-mediated platelet activation/aggregation. Consequently, hFcR mice deficient for CalDAG-GEFI were completely protected from ITT.
In conclusion, our novel mouse model for ITT closely mimics experimental HIT and experimental thrombocytopenia and thrombosis induced by heparin/VEGF/anti-VEGF complexes. Our model represents an important step toward more standardized studies on the mechanisms underlying platelet activation in ITT, as it (1) eliminates the need for complicated mouse crosses, (2) depends on a single antibody instead of preformed immune complexes, and (3) provides a simple readout for the quantitative visualization of Fc␥RIIA-dependent thrombosis in vivo. Our studies also provide a rationale for the documented limited protective effect of P2Y12 inhibitors in HIT patients, and it establishes the Ca 2ϩ -sensor CalDAG-GEFI as a promising new target for the intervention with HIT/ITT.