α2β1 Integrin, GPVI Receptor, and Common FcRγ Chain on Mouse Platelets Mediate Distinct Responses to Collagen in Models of Thrombosis

Objective Platelets express the α2β1 integrin and the glycoprotein VI (GPVI)/FcRγ complex, both collagen receptors. Understanding platelet-collagen receptor function has been enhanced through use of genetically modified mouse models. Previous studies of GPVI/FcRγ-mediated collagen-induced platelet activation were perfomed with mice in which the FcRγ subunit was genetically deleted (FcRγ−/−) or the complex was depleted. The development of α2β1−/− and GPVI−/− mice permits side-by-side comparison to address contributions of these collagen receptors in vivo and in vitro. Approach and Results To understand the different roles played by the α2β1 integrin, the GPVI receptor or FcRγ subunit in collagen-stimulated hemostasis and thrombosis, we compared α2β1−/−, FcRγ−/−, and GPVI−/− mice in models of endothelial injury and intravascular thrombosis in vivo and their platelets in collagen-stimulated activation in vitro. We demonstrate that both the α2β1 integrin and the GPVI receptor, but not the FcRγ subunit influence carotid artery occlusion in vivo. In contrast, the GPVI receptor and the FcRγ chain, but not the α2β1 integrin, play similar roles in intravascular thrombosis in response to soluble Type I collagen. FcRγ−/− platelets showed less attenuation of tyrosine phosphorylation of several proteins including RhoGDI when compared to GPVI−/− and wild type platelets. The difference between FcRγ−/− and GPVI−/− platelet phosphotyrosine levels correlated with the in vivo thrombosis findings. Conclusion Our data demonstrate that genetic deletion of GPVI receptor, FcRγ chain, or the α2β1 integrin changes the thrombotic potentials of these platelets to collagen dependent on the stimulus mechanism. The data suggest that the FcRγ chain may provide a dominant negative effect through modulating signaling pathways in platelets involving several tyrosine phosphorylated proteins such as RhoGDI. In addition, these findings suggest a more complex signaling network downstream of the platelet collagen receptors than previously appreciated.


Introduction
Hemostasis relies on the highly regulated balance of prothrombotic and antithrombotic components to prevent blood loss from the vasculature while at the same time maintaining blood fluidity. Platelets play a central role in this balance especially during arterial hemostasis and pathological thrombosis. Fibrillar collagens represent a potent prothrombotic stimulus for platelets at sites of vascular injury.
A role for a2b1 integrin-mediated adhesion in vascular disease was suggested by epidemiologic studies that linked a2b1 integrin density to pathologic thrombosis and enhanced bleeding. Kunicki et al. demonstrated that a2b1 integrin density on the platelet surface directly correlates with platelet adhesiveness to Type I collagen (7). We previously reported that the a2b1 integrin-deficient (a2b1 2/2 ) mice clearly exhibit impaired adhesion to collagen substrates under arterial flow conditions and observed a marked decrement in thrombus formation in vivo following arterial injury (8).
Similarly, the importance of the GPVI/FcRc complex in normal hemostasis was demonstrated in patients with a mild bleeding diathesis associated with either mutations in the Gp6 gene or presence of anti-GPVI antibodies (1,(9)(10)(11). Collagen or collagen-related peptides (CRPs) binding to the GPVI subunits stimulate clustering of the GPVI/FcRc complex, tyrosine phosphorylation of the ITAM motifs within FcRc chains, and activation of the Src family tyrosine kinases Fyn and Lyn that trigger platelet activation (12,13). Phosphorylation of the ITAM domain also results in activation of Syk and downstream effectors, including PLCc2, PI3K, and small GTPases that contribute to platelet activation and aggregation (13). Earlier studies of GPVI/FcRc-mediated collagen-induced platelet activation and thrombus formation were carried out using mice in which either the FcRc subunit was genetically deleted [FcRc-deficient (FcRc 2/2 ) mice] or the complex was depleted by antibody-mediated internalization. Of note, platelets derived from FcRc 2/2 mice fail to express GPVI. In these studies, lack of FcRc through genetic knockout or antibody-depletion resulted in attenuated collagen-stimulated platelet activation and thrombus formation under flow conditions in vitro, however the phenotype in vivo still remains unclear (14)(15)(16)(17)(18). Importantly, the FcRc 2/2 animals also lack FceRcI, FccRIII, and FccRI, and are immunodeficient with abnormalities in macrophage, NK cell, mast cell and B cell function. More recently, GPVI-deficient (GPVI 2/2 ) mice were developed. These mice were reported to be viable and fertile, and to exhibit normal bleeding times. However, GPVI 2/2 platelets did not aggregate in response to collagen or GPVIspecific collagen related peptide (CRP) (15). Although GPVI-null platelets did not form aggregates when perfused over a collagen surface, they did form an adherent monolayer. Kato et al. attributed the residual adhesion solely to von Willebrand factor (15). Conversely, other groups showed decreased platelet adhesion to collagen upon loss of GPVI (14,18,19). These discrepancies suggest that much remains to be understood about the mechanisms of collagen receptor signaling in platelets.
The development of a2b1 2/2 , FcRc 2/2 , and GPVI 2/2 mice allowed us to compare the roles and contributions of these platelet collagen receptors in sideby-side comparisons in vivo and in vitro (15,20,21). Here we report data comparing GPVI 2/2 , FcRc 2/2 , and a2b1 2/2 mice using in vivo and in vitro assays of thrombosis. Unexpectedly, the GPVI 2/2 and FcRc 2/2 mice demonstrated different defects, suggesting distinct phenotypes of platelets lacking GPVI or FcRc. These data show that the platelet responses to collagen in FcRc 2/2 mice differ from GPVI 2/2 mice and raises caution to utilizing these two knockout mice as similar systems.
The a2b1 2/2 mice, backcrossed onto the C57BL/6 background were previously reported (20). FcRc-deficient mice on the C57BL/6 and C57BL/6 X 129/SvJ background were purchased from Jackson Labs. GPVI-deficient mice on a C57BL/ 6 X 129/SvJ background were developed by Kato et al. (15). GPVI-deficient mice were backcrossed 8 times to the C57BL/6 background using a microsatellite marker-assisted selection (''speed congenics''), as previously described (20). Animals were housed in pathogen-free conditions at Vanderbilt University Medical Center in compliance with IACUC regulations. The protocol was approved by the Internal Review Board at Vanderbilt University (protocol #M/ 05/324). All animals were appropriately age and sex matched, and efforts were made to minimize suffering.

Platelet isolation
Murine PRP or washed platelets were prepared from blood obtained on the day of the experiment according to protocols described previously (22).

Platelet adhesion assay
Adhesion assays were carried out using washed platelets (1610 8 platelets/mL) as done previously (22). Measurements for each data point were performed in triplicate.

Platelet aggregation
Aggregation assays using PRP were performed on a BIO/DATA Corporation PAP-4 aggregometer at 37˚C with stirring (1200 rpm) as described (20). Agonists were added at designated final concentrations.
In vivo photochemical injury of the carotid artery of mice Carotid artery thrombosis was induced as described previously (8). Briefly, male mice approximately 12 weeks of age were anesthetized with an intraperitoneal injection of sodium pentobarbital, secured in the supine position, and placed under a dissecting microscope. The right common carotid artery was isolated through a midline cervical incision, and an ultrasonic flow probe (Model 0.5 VB; Transonic Systems) was applied. A 1.5-mW, 540-nm laser beam (Melles Griot) was applied to the artery from a distance of 6 cm. Rose-Bengal dye (Fisher Scientific), 50 mg/kg body weight, was then injected into the tail vein, and flow in the vessel was monitored until complete occlusion occurred.

In vivo collagen-induced pulmonary thromboembolism in mice
Collagen-induced thrombosis was carried out as previously described (8). Briefly, female mice were anesthetized by intraperitoneal injection of 100 to 150 mL of a mixture of ketamine and xylazine. Blood was collected into EDTA-coated microtainer tubes for determination of the baseline platelet count and hematocrit. 25 mg of collagen (equine tendon Type I fibrillar collagen) along with 1 mg epinephrine (Sigma) in phosphate-buffered saline (PBS), or PBS alone, were injected into the right jugular vein; 1 minute after injection a second blood sample was taken and cell counts were measured. Mice were humanely sacrificed 3 minutes after injection and lungs were collected and placed in formalin. Pulmonary thrombi were quantitated using digital imaging of lung sections stained with hematoxylin and eosin using an Olympus Camedia C-3040 Zoom camera. Five random 20X fields were photographed for each specimen. Analysis of thrombus number for each mouse lung was made using Olympus Camedia Master 2.5 software, and then expressed as thrombi per square millimeter¡SEM.

Scanning electron microscopy
Platelet adhesion assays similar to those described above were performed with minor changes. Platelets at a concentration of 2610 7 platelets/mL were allowed to adhere to substrates (30 mg/mL) bound to round glass coverslips (Electron Microscopy Sciences; 22 mm diameter) for 1 hour at 37˚C. Coverslips were washed 3 times with adhesion buffer. Adherent platelets were fixed using 2% glutaraldehyde for 30 minutes at 21˚C, washed 3 times with 0.1 M sodium cacodylate buffer and processed (fixed, dried, and sputter coated) in the VUMC Cell Imaging Shared Resource and the EM Core. Imaging was done using a Hitachi S-4200 Scanning Electron Microscope.

Mouse platelet phosphotyrosine analysis
Mouse platelets were resuspended at a concentration of 5610 8 platelets/mL in HBSS-containing 2 mM MgCl 2 . Mouse platelets (wild type, GPVI 2/2 , or FcRc 2/2 ) were either untreated or stimulated with 10 mg/mL collagen Type I for 1 minute at 21˚C after which an excess of ice-cold HBSS-was added to stop the interaction. Platelets were pelleted at 4,000 rpm for 4 minutes at 4˚C. Platelets were lysed using SDS-PAGE sample buffer containing protease and phosphatase inhibitors and run on a reducing 10% SDS-PAGE gel. The proteins were transferred to a nitrocellulose membrane followed by immunoblot analysis with anti-phosphotyrosine (12000) or anti-actin (12000) antibodies. Appropriate secondary antibodies linked with horseradish peroxidase were used with a chemiluminescence substrate to image the labeled protein bands using a BioRad ChemiDoc with Quantity One software. Densitometry on protein bands of interest from image files was done using ImageJ software.

Statistical analyses
The data from multiple different animals were analyzed by Fisher's least significant difference approach. Means, standard deviations (SD), standard error of the means (SEM), t-test, one-way, and two-way ANOVA for column statistics, and nonlinear curve fits were calculated using GraphPad Prism 4 software.

Deletion of the GPVI and FcRc receptors, but not the a2b1 integrin alters collagen-induced intravascular thrombosis and pulmonary embolism
Distinct biological roles for the a2b1 integrin and the GPVI-FcRc receptor in platelet activation and thrombus formation in vitro have been well described. To further understand the complex roles played in vivo by the a2b1 integrin and the GPVI/FcRc complex, mice with targeted deletion of the distinct receptors were evaluated using models of collagen-induced thrombosis. Intravenous injection of collagen Type I into wild type animals resulted in rapid onset of intravascular thrombosis, a profound decrease in the number of circulating platelets, massive pulmonary emboli, and death (as previously reported and Figure 1A and B). Pulmonary thrombosis occurred rapidly, therefore the experiment was completed in 3 minutes. It was difficult to evaluate time-dependent differences in this model in which collagen-induced platelet activation is independent of shear stress.
In the absence of FcRc receptor, the a2b1 integrin contributes to intravascular thrombus formation To define the overlapping and/or synergistic roles of the platelet collagen receptors in vivo, we compared platelet decrement and pulmonary thrombi in mice with combined deficiency of the GPVI or FcRc receptor and the a2b1 integrin on a mixed SvJ129/C57BL/6 background. The platelet count decrement in the a2b1 2/2 /FcRc 2/2 animals was 20%¡3% (n58), significantly less than the platelet decrement in wild type or a2b1 2/2 mice (P,0.0001 for both analyses), but surprisingly also significantly less than the decrement in the FcRc 2/2 mice (P50.02) ( Figure 1A). In the a2b1 2/2 /GPVI 2/2 mice (n510), the platelet count decreased by 12%¡2% (n510) ( Figure 1A), therefore, significantly less than wild type mice or a2b1 2/2 mice (P,0.0001 for both) and significantly less than a2b1 2/2 /FcRc 2/2 (P50.02) but not different from mice lacking the GPVI receptor alone (P50.99). The lack of a difference in mice lacking both GPVI and a2b1 integrin may be due to the very low level of platelet aggregation and further changes could not be detected. However, loss of a2b1 did attenuate thrombosis when paired with loss of FcRc chain.
Deletion of either the a2b1 integrin or the GPVI subunit, but not the FcRc receptor delays carotid artery thrombosis We analyzed platelet activation in a second in vivo model of thrombosis that involves photochemical damage of the artery to produce endothelial cell denudation and subendothelial extracellular matrix exposure under shear stress conditions (23). This assay uses laser-activated Rose-Bengal dye to produce the photochemical injury of the mouse carotid artery in order to measure the time to complete vessel occlusion ( Figure 2). We compared the time required for complete arterial occlusion in a2b1 2/2 , FcRc 2/2 , GPVI 2/2 and wild type mice. As previously reported, the time to complete occlusion was significantly prolonged in the a2b1 2/2 animals (74.5¡19.8 minutes) compared to wild type littermates (44.4¡7.8 minutes) (P50.0002) and to animals lacking FcRc (P50.008). Occlusion times for FcRc 2/2 mice (39.3¡15.9 minutes) were not different from wild type mice (P50.99). In contrast, occlusion times for the GPVI 2/2 mice (74.6¡28.3 minutes) were statistically increased compared to those in wild type (P50.0002) and FcRc 2/2 mice (P50.004), but similar to a2b1 2/2 mice (P50.3). In this second in vivo model of platelet function, the time to complete arterial occlusion was different in the FcRc 2/2 and GPVI 2/2 mice.
The in vivo thrombotic differences are independent of genetic background Since the differences observed between the GPVI 2/2 and FcRc 2/2 mice were unexpected and platelet responses are known to be dependent in some circumstances on genetic background, we acquired or generated animals on a pure C57BL/6 background, as described in Methods. Studies of platelet response to collagen were repeated and platelet count decrements in mice on a pure C57BL/6 background were determined. The results obtained for platelet count decrement and number of pulmonary thrombi following intravenous injection of collagen into wild type, GPVI 2/2 , FcRc 2/2 or a2b1 2/2 pure C57BL/6 animals were similar to that observed on the mixed background ( Figure 3A and 3B). As observed in mice on a mixed background, there was no difference in platelet count decrement or in the number of thrombi between wild type (n57) and a2b1 2/2 animals (n511) (P50.4), but a significant difference was seen between wild type animals and FcRc 2/2 (n59) (P50.0001) or GPVI 2/2 mice (n57) (P50.0001). In addition, as observed on the mixed background, there was a significant difference in platelet decrement (P50.0001) and the number of thrombi (P,0.0001) between the FcRc 2/2 and GPVI 2/2 mice on the pure background The impact of genetic background on injury induced carotid artery occlusion was also evaluated in a2b1 2/2 , FcRc 2/2 , GPVI 2/2 and wild type mice on a pure C57BL/6 genetic background (Figure 4). Time to complete occlusion for the a2b1 2/2 mice (70¡7 minutes) was significantly prolonged compared to wild type animals (52.7¡2.2 minutes), as expected (P50.02). The FcRc 2/2 mice Figure 2. Deletion of either the a2b1 integrin or the GPVI subunit, but not the FcRc receptor delays carotid artery thrombosis. The length of time to complete arterial occlusion following photochemical injury of the carotid artery was recorded in wild type (WT), a2b1 2/2 , FcRc 2/2 , and GPVI 2/2 mice on a mixed genetic background (129/SvJ 6 C57BL/6). The values represent the mean ¡ SD for WT (n512), a2b1 2/2 (n515), GPVI 2/2 (n57), or FcRc 2/2 (n57) animals.  The platelet count decrement (% change from baseline) following intravenous injection of Type I collagen (25ug) and epinephrine (1ug) into wild type (WT) (n57), GPVI 2/2 (n57), FcRc 2/2 (n59) or a2b1 2/2 (n511) mice on a pure C57BL/6 background was determined. (B) The number of thrombi observed in the lungs at 3 minutes after injection of Type I collagen (25ug) and epinephrine (1ug) was determined for wild type (WT), a2b1 2/2 , GPVI 2/2 , FcRc 2/2 mice on the C57BL/6 background. Thrombi were recorded per mm 2 in 6 random 20X fields. demonstrated an occlusion time of 54.8¡4.8 minutes, similar to wild type (P50.7). Time to occlusion for the pure GPVI 2/2 mice (99¡20 minutes) was significantly prolonged compared either wild type (P50.006) or FcRc 2/2 mice (P50.02). These data show that mice deficient in the GPVI receptor, but not the FcRc chain manifest a major defect in carotid artery thrombosis induced by photochemical injury. Overall, the genetic background did not contribute to the in vivo variances observed between genotypes in thrombosis.
Platelet adhesion, spreading and aggregation on Type I collagen are dependent on the a2b1 integrin, the GPVI receptor and the FcRc receptors The data presented above describe an unexpected difference in vivo in thrombosis between mice lacking the GPVI and the FcRc subunits. To better define the mechanisms for this difference, platelet adhesion using platelets from pure C57BL/6 animals was evaluated in vitro at 60 minutes ( Figure 5A). Wild type platelets adhered to collagen substrates in a time dependent manner, but not to BSA ( Figure 5A). As previously reported, a2b1 2/2 platelets failed to adhere. Although FcRc 2/2 and GPVI 2/2 platelets adhered to collagen substrates, adhesion was significantly reduced at 60 minutes compared to adhesion of wild type platelets. Although reduced compared to wild type, adhesion of FcRc 2/2 platelets was significantly greater than adhesion of GPVI 2/2 platelets at each time point ( Figure 5B). Expression of the a2b1 integrin was similar on GPVI 2/2 and FcRc 2/2 platelets and therefore did not explain the difference in adhesion to collagen I (data not shown).
Next we analyzed platelet aggregation. Wild type platelets formed small and large platelet aggregates during adhesion to collagen I after 60 minutes, but FcRc 2/2 and GPVI 2/2 platelets did not (data not shown). There was no difference in platelet aggregation between the FcRc 2/2 and GPVI 2/2 platelets. Correspondingly, wild type and a2b1 2/2 platelets aggregated in response to soluble collagen, as measured by turbidometric aggregometry, but neither the FcRc 2/2 nor GPVI 2/2 platelets aggregated (data not shown).
We also examined wild type, a2b1 2/2 , GPVI 2/2 and FcRc 2/2 platelets using scanning electron microscopy to determine if there were variances in platelet morphology during adhesion to collagen I ( Figure 5D). Individual wild type, GPVI 2/2 , and FcRc 2/2 formed similar filopodial extensions on collagen I. No a2b1 2/2 platelets attached to collagen I and therefore the image was similar to the BSA negative controls (data not shown).
Protein phospho-tyrosine analyses of wild type, GPVI 2/2 , and FcRc 2/2 mouse platelets To determine whether the different phenotypes observed in vivo between wild type, GPVI 2/2 , and FcRc 2/2 mice were a consequence of alterations in a2b1 integrin-dependent platelet activation signals in response to collagen, we determined the extent of protein tyrosine phosphorylation following collageninduced platelet activation in vitro ( Figure 6A). Mouse platelets from wild type, GPVI 2/2 , or FcRc 2/2 mice were either untreated or stimulated with 10 mg/mL collagen I for 1 minute followed by immunoblot analysis of proteins containing phosphorylated tyrosines. Interestingly, control and collagen-stimulated platelets from wild type, GPVI 2/2 , and FcRc 2/2 animals showed slightly or moderately different levels of tyrosine phosphorylation of 72 and 25 kDa proteins, respectively. These variations in protein tyrosine phosphorylation between genotypes was quantified for the 72 and 25 kDa proteins in Figure 6A, and the amounts of phosphorylation correlates with the levels of thrombosis observed in vivo and with adhesion to collagen substrates in vitro. Wild type platelets demonstrated a low basal level of phosphorylation of a 72 kDa protein that was augmented by collagen I stimulation.  (A) Purified platelets from WT, GPVI 2/2 , or FcRc 2/2 mice were either untreated or stimulated with 10 mg/mL collagen I In the presence of 2 mM MgCl 2 for 1 minute followed by immunoblot analysis using antibodies against phospho-tyrosine (pTyr) or actin. Protein bands of interest are indicated with arrows at molecular weights of about 25 and 72 kDa. Quantification of the 25 and 72 kDa pTyr bands were performed by densitometry and normalized to actin. (B) WT, GPVI 2/2 , or FcRc 2/2 mouse platelets stimulated with CNI were analyzed for phosphorylated Syk (pSyk) and total Syk by immunoblot. In the presence of 2 mM MgCl 2 , control platelets or platelets treated with 10 mg/mL CNI for 1 minute, were lysed, and analyzed by Western blot using antibodies for phospho-Syk (Tyr525/526), total Syk protein, or total phospho-tyrosine. Quantification of the pSyk and total Syk was performed by densitometry and the value of pSyk normalized to total Syk. (C) Western blot analysis of WT, GPVI 2/2 , or FcRc 2/2 mouse platelets for phosphorylated RhoGDI and total RhoGDI following CNI stimulation as described. Quantification of the pRhoGDI and total RhoGDI was performed and the value of pRhoGDI normalized to total RhoGDI. (D) Western blot analysis of unstimulated WT, GPVI 2/2 , or FcRc 2/2 mouse platelets for total FcRc protein or actin expression. phosphorylation of the 25 kDa protein that were much higher than the levels observed in GPVI 2/2 platelets. Both wild type and FcRc 2/2 platelets demonstrated enhanced phosphorylation of the 25 kDa protein with collagen I stimulation. These data further support an underlying difference between platelet activities in FcRc 2/2 and GPVI 2/2 mice.
A potential candidate for the tyrosine-phosphorylated 72 kDa protein is Syk, a protein tyrosine kinase important in GPVI/FcRc signal transduction and platelet activation. Therefore, we analyzed Syk phosphorylation at tyrosines 525 and 526, which are required for Syk activation. Mouse platelets (wild type, GPVI 2/2 , or FcRc 2/2 ) were left untreated or stimulated with 10 mg/mL collagen I for 1 minute followed by immunoblot analysis of phospho-Syk (Tyr525/526) in correlation with total Syk protein and total protein tyrosine phosphorylation of the 72 kDa band ( Figure 6B). Collagen-treated platelets showed a slightly elevated levels of phospho-Syk (Tyr525/527) that correlated with total phospho-tyrosine of the 72 kDa protein band when comparing wild type, GPVI 2/2 , and FcRc 2/2 platelets. Wild type had a low basal level of Syk phosphorylation that increased with collagen I treatment; GPVI 2/2 had the lowest level of basal and collagenstimulated Syk phosphorylation; and FcRc 2/2 platelets also had low to intermediate levels of basal and collagen-stimulated Syk phosphorylation as shown in the quantitation of phospho-Syk ( Figure 6B). The difference in phospho-Syk levels between resting and activated platelets was not significant and therefore did not contribute to observed phenotype.
Identification of a candidate for the 25 kDa phosphotyrosine protein band was undertaken, and a potential candidate of this molecular weight was the family of RhoGDI proteins. RhoGDI has been shown to modulate cellular activities through regulation of Rho GTPases by inhibiting Rho GTPase activation through a sequestration mechanism, which is diminished when RhoGDI in tyrosine phosphorylated (24). We compared baseline and collagen I-induced RhoGDI2 tyrosine phosphorylation in wild type, FcRc 2/2 , and GPVI 2/2 platelets. As shown in Figures 6C, the 25 kDa phospho-protein band overlays with RhoGDI2. Significantly enhanced tyrosine phosphorylation of RhoGDI2 in wild type and FcRc 2/2 , but not GPVI 2/2 platelets occurred in response to Type I collagen.
Since a2b1 integrin-dependent collagen adhesion and RhoGDI2 phosphorylation occurred in FcRc 2/2 platelets, but not in GPVI 2/2 platelets, the continued expression of FcRc in GPVI 2/2 platelets could be causing a dominant negative phenotype in the GPVI 2/2 mice by interfering with signaling pathways in these platelets. We therefore determined the expression of FcRc receptor in platelets from GPVI 2/2 mice in comparison to wild type and FcRc 2/2 ( Figure 6D). Surprisingly, GPVI 2/2 platelets have a similar expression level of FcRc compared to wild type platelets and not a partial reduction. As expected, no expression was seen in the FcRc 2/2 platelets. The wild type expression levels of FcRc receptor in GPVI 2/2 platelets may cause the varied thrombotic phenotypes observed between GPVI 2/2 and FcRc 2/2 mice through increased RhoGDI sequestration of Rho GTPases and attenuation on actin cytoskeletal dynamics, which are important in platelet adhesion and thrombosis.

Discussion
The findings reported in this study demonstrate novel requirements in vivo and in vitro for the platelet collagen receptors, the a2b1 integrin and the GPVI/FcRc receptor complex. First, the in vivo data demonstrate by using multiple genetically modified animals that deletion of the GPVI receptor results in a phenotype that is distinctly different from that resulting from deletion of the common FcRc chain. Second, both the a2b1 integrin and the GPVI receptor, but not the FcRc subunit, influence carotid artery occlusion in vivo. In contrast, the loss of GPVI receptor or the FcRc chain decreases intravascular thrombosis in response to soluble Type I collagen, but surprisingly the FcRc 2/2 animals retain some thrombotic potential to collagen. These findings are the first to show disparate effects in thrombosis between GPVI receptor and FcRc chain knockout mice. Much of our earlier understanding of the role of the GPVI receptor in collagen-induced platelet activation and thrombus formation was based on mice lacking the FcRc chain. The development of the a2b1 integrin-deficient mice (20,21), FcRc-deficient mice, and GPVI-deficient mice (15), affords an unambiguous opportunity to address definitively the roles and contributions of these receptors to platelet activation by collagen in vivo and in vitro. Third, earlier data from a number of laboratories, including our own, failed to identify a role for the a2b1 integrin in collagen-induced intravascular thrombosis. We now show that in the absence of the common FcRc chain the a2b1 integrin promotes a low level of collageninduced platelet activation and thrombosis. Finally, the molecular basis for the difference in activation status of platelets from FcRc chain-deficient mice and GPVI-deficient mice correlates to differences in collagen-induced tyrosine phosphorylation of RhoGDI2.
The identification of different thrombotic responses to collagen observed in mice with deletion of GPVI or FcRc genes was initially surprising, and it was logical to assume that this was attributable to the genetic background of mice on a mixed genetic background. A study by Cheli et al. (25) using the same mixed GPVI 2/2 mice on the 129/SvJ x C57BL/6 background identified a Modifier of hemostasis (Mh) locus on chromosome 4 that correlated with the extreme but transient dichotomy in tail bleeding time (tBT) seen in the earlier generations of these mice. A modest correlation was also observed in the in vivo ferric chlorideinduced carotid artery injury model. With progressive backcrosses to C57BL/6, this phenotypic difference gradually diminished until later generations of GPVI 2/ 2 mice that were congenic on the C57BL/6 background exhibit a normal tBT. Nonetheless, a follow-up study of Mh has identified at least one candidate gene of interest that is currently under further investigation. Our studies in this report rule out the possibility that genetic differences arising from the original mixed background (129/SvJ 6 C57BL/6) are responsible for this unexpected result since animals with genotypes on pure inbred background (C57BL/6) showed similar experimental outcomes suggesting that a genetic modifier (such as Mh) was not responsible for the differences. Thus, we conclude that there is yet another a2b1 Integrin and GPVI/FcRc Receptor in Mouse Models of Thrombosis mechanism responsible for thrombotic differences between GPVI 2/2 or FcRc 2/2 mice.
To determine whether differences observed in vivo were demonstrable in isolated platelets, the ability of FcRc 2/2 and GPVI 2/2 platelets to respond to collagen was determined. GPVI 2/2 and FcRc 2/2 platelets adhered to Type I collagen substrates, however FcRc 2/2 platelets were significantly more adherent to collagen in an a2b1 integrin-dependent manner than GPVI 2/2 platelets. No differences between GPVI 2/2 and FcRc 2/2 platelets were seen in in vitro analyses of aggregation or morphology. Taken together these results suggested that FcRc 2/ 2 platelets, but not GPVI 2/2 platelets, adhered to collagen in an a2b1 integrindependent fashion. During the platelet response to collagen, distinct phosphotyrosine protein profiles were observed between the GPVI 2/2 and FcRc 2/2 platelets, especially proteins of 72 and 25 kDa. The 25 kDa phosphoprotein was identified as RhoGDI2, an important regulator of RhoGTPase signaling.
Importantly, tyrosine phosphorylation of RhoGDI inhibits binding to RhoA, Rac1, and Cdc42 freeing them for activation (26,27). Not much is known about RhoGDI functions in platelets even though its activity on Cdc42 was revealed in platelets (28)(29)(30)(31). We report that RhoGDI2 phosphorylation is enhanced in FcRc 2/2 platelets when compared to GPVI 2/2 platelets upon collagen stimulation. This suggests that FcRc 2/2 platelets may contain a signaling environment more primed for RhoGTPase activation than GPVI 2/2 platelets. In agreement with our result, Poole et al. observed a slight increase in Syk phosphorylation in FcRc 2/2 platelets with collagen I treatment (32). Mazzucato et al. showed that platelets lacking GPVI under flow conditions could still adhere to collagen I and elevate intracellular Ca 2+ through a2b1 integrin (33). Awareness of this difference between GPVI/FcRc knockouts may be pertinent to future research since a recent study by Boulaftali et al. showed that ITAM containing receptors (GPVI/FcRc and CLEC2) were critical for a novel form of hemostasis at sites of inflammation where GPCR signaling was not required (34,35). On the basis of these results, we propose that GPVI 2/2 and FcRc 2/2 platelets are poised at different resting states and GPVI 2/2 platelets have a larger activation barrier to overcome than wild type or FcRc 2/2 , which explains the differences observed with the in vivo thrombotic analyses.
Animals deficient in the common FcRc chain, an important signaling subunit of multiple cell surface receptors, have been extensively evaluated as a model of GPVI deficiency (32,36). These mice, which lack the GPVI/FcRc receptor, also lack FceRcI, FccRIII, and FccRI, and are immunodeficient with abnormalities in macrophage, NK cell, mast cell and B cell function, but fail to manifest a bleeding diathesis (37). Abnormalities observed with FcRc-deficient platelets in vitro include defective secretion, platelet activation and aggregation in response to collagen or GPVI-mimetics. Interestingly, bleeding and platelet function abnormalities are much less severe in FcRc-null mice than in mice lacking several of the downstream adaptor/signaling molecules such as SLP76 or PLCc2 (38,39). These studies initially suggested that other collagen receptors in addition to GPVI/ FcRc also contribute to collagen-induced signals. Work by Konstantinides et al. showed that during arterial thrombosis loss of GPVI had variable effects and was dependent upon wound severity and the presence of activating factors (40). We now raise the possibility that the platelet response to collagen in mice lacking the FcRc chain differ in a number of ways from mice lacking only the GPVI subunit. To exclude the possibility that expression of the FcRc receptor on other cell types contributes to the phenotype will require additional experiments with plateletselective, FcRc-null animals.
Enhanced adhesion to collagen I and the higher level of protein tyrosine phosphorylation in collagen-stimulated FcRc 2/2 platelets compared to GPVI 2/2 platelets suggests FcRc expression in GPVI 2/2 platelets mediates a dominantnegative phenotype. The level of FcRc expression remains at wild type levels even though there is loss of the GPVI/FcRc receptor complex, which suggests that FcRc could be associating with other interacting proteins in GPVI 2/2 platelets and negatively affecting platelet ITAM signaling pathways through sequestration or disruption mechanisms. Importantly, in mouse platelets there are no other known membrane proteins that directly bind FcRc. Dominant negative mutations have been shown to affect surface receptor function previously (e.g. PDGF and Her2 receptors) (41,42). This inhibitory activity is not present upon loss of FcRc expression or correct partnering of FcRc proteins. We speculate that the decreased levels of tyrosine phosphorylated proteins (RhoGDI) in collagen-stimulated GPVI 2/2 platelets attenuates platelet activities like cytoskeletal rearrangement, priming of a2b1 integrin, or synergy with GPCR signaling, which are important processes of inactivation of wild type platelets (22,43,44). Interestingly, the loss of a2b1 integrin on FcRc 2/2 platelets decreased collagen-stimulated aggregation in vivo and suggests a2b1 signaling is negatively regulated by inappropriate expression of FcRc in GPVI 2/2 platelets.
In summary, we identify in vivo and in vitro differences in thrombosis between a2b1 2/2 , GPVI 2/2 and FcRc 2/2 platelets, which surprisingly revealed variances between GPVI 2/2 and FcRc 2/2 platelets. This variation in phenotype seems to be attributable to normal expression levels of FcRc in GPVI 2/2 platelets and produces a dominant negative effect through a decrease in protein tyrosine phosphorylation of RhoGDI leading to increased inhibition of RhoGTPases. In addition, these findings suggest a complex signaling network downstream of the platelet collagen receptors. Shida and colleagues recently published a comprehensive analysis of von Willebrand factor and its interactions with collagen in conjunction with GPVI and the a2b1 integrin during thrombosis and showed major but overlapping functions (45). Further understanding of the functions of GPVI/FcRc on platelets and the involvement of RhoGDI and other molecules, including von Willebrand factor in this complex pathway is necessary and important as these receptors are potential targets for antithrombotic therapy.