Filamin A (FlnA) cross-links actin filaments and connects the Von Willebrand factor receptor GPIb-IX-V to the underlying cytoskeleton in platelets. Because FlnA deficiency is embryonic lethal, mice lacking FlnA in platelets were generated by breeding FlnAloxP/loxP females with GATA1-Cre males. FlnAloxP/y GATA1-Cre males have a macrothrombocytopenia and increased tail bleeding times. FlnA-null platelets have decreased expression and altered surface distribution of GPIbα because they lack the normal cytoskeletal linkage of GPIbα to underlying actin filaments. This results in ∼70% less platelet coverage on collagen-coated surfaces at shear rates of 1,500/s, compared with wild-type platelets. Unexpectedly, however, immunoreceptor tyrosine-based activation motif (ITAM)- and ITAM-like–mediated signals are severely compromised in FlnA-null platelets. FlnA-null platelets fail to spread and have decreased α-granule secretion, integrin αIIbβ3 activation, and protein tyrosine phosphorylation, particularly that of the protein tyrosine kinase Syk and phospholipase C–γ2, in response to stimulation through the collagen receptor GPVI and the C-type lectin-like receptor 2. This signaling defect was traced to the loss of a novel FlnA–Syk interaction, as Syk binds to FlnA at immunoglobulin-like repeat 5. Our findings reveal that the interaction between FlnA and Syk regulates ITAM- and ITAM-like–containing receptor signaling and platelet function.
The filamin family consists of three large dimeric proteins (filamin A [FlnA], FlnB, and FlnC) that cross-link actin filaments, tether membrane glycoproteins, and serve as scaffolds for signaling intermediates (Stossel et al., 2001; Zhou et al., 2010). The most abundant isoform of the family, FlnA, is encoded by the X chromosome in humans and mice (Feng and Walsh, 2004; Robertson, 2005). FlnA is composed of an N-terminal actin-binding domain followed by 24 Ig repeats, the C-terminal of which mediates their dimerization (Pudas et al., 2005). Human melanoma cells that lack FlnA have poor motility and continuous membrane blebbing (Cunningham et al., 1992; Flanagan et al., 2001). FLNA mutations have been associated with periventricular heterotopia, Ehlers-Danlos Syndrome, or familial cardiac valvular dystrophy (Fox et al., 1998; Robertson et al., 2003; Sheen et al., 2005; Kyndt et al., 2007; Unger et al., 2007). Loss of FlnA in mice results in embryonic lethality caused by pericardiac and visceral hemorrhage, severe cardiac structural defects, and aberrant vascular patterning (Feng et al., 2006; Hart et al., 2006).
Platelets predominantly express FlnA (5 µM), although a small amount of FlnB is also expressed (<0.5 µM). Platelet FlnA has a critical structural role in attaching the Von Willebrand Factor (VWF) receptor complex GPIb-IX-V to the underlying actin cytoskeleton. aa 556–577 in the cytoplasmic tail of GPIbα constitutively interacts with FlnA Ig repeat 17 (Nakamura et al., 2006), and the loss of GPIbα in mice results in enlarged platelets, a phenotype which can be rescued by expression of a chimeric protein construct containing the cytoplasmic domain of human GPIbα (Ware et al., 2000; Kanaji et al., 2002). FlnA binding facilitates GPIbα surface expression in Chinese hamster ovary (CHO) cells (Feng et al., 2005), and the interaction between FlnA and GPIbα has been reported to influence VWF receptor function, although conflicting effects are found in the literature. CHO cells transfected with GPIbα mutants that lack the FlnA binding site have decreased VWF binding (Dong et al., 1997; Schade et al., 2003), VWF-induced cell aggregation (Mistry et al., 2000), and/or adhesion to a VWF matrix under high shear (Cranmer et al., 1999; Williamson et al., 2002; Cranmer et al., 2005). In contrast, another study has shown that FlnA binding to GPIbα negatively regulates VWF binding to CHO cells and CHO cell adhesion under both static and flow conditions (Englund et al., 2001). Platelets treated with cell-permeable peptide mimics of the FlnA binding site on GPIbα have decreased ability to activate their fibrinogen receptor, the integrin αIIbβ3, and change shape in response to VWF stimulation, suggesting that the interaction between FlnA and GPIbα positively modulates signaling events initiated by VWF in platelets (Feng et al., 2003; David et al., 2006).
In this study the role of FlnA was probed in platelets. Mouse platelets lacking FlnA were generated by breeding FlnAloxP/loxP females with GATA1-Cre males (Jasinski et al., 2001). Offspring FlnAloxP/y GATA1-Cre males have <15% of normal blood platelet count, and their platelets lack FlnA. As expected, FlnA-null platelets are large, lack normal actin-GPIbα membrane attachments, and have an altered distribution of GPIbα on their surface. However, FlnA-null platelets surprisingly also have severe functional impairment in signaling responses downstream of the immunoreceptor tyrosine-based activation motif (ITAM)– and ITAM-like–mediated signal receptors GPVI and C-type lectin-like receptor 2 (CLEC-2). This specific signaling defect results from the loss of a novel and direct FlnA–Syk interaction mediated through Ig repeat 5 of FlnA. As a consequence, Syk is mistargeted in FlnA-null platelets, and ITAM signaling that normally leads to actin assembly, α-granule secretion, integrin αIIbβ3 activation, and protein tyrosine phosphorylation is interrupted. FlnA is, therefore, required for normal ITAM- and ITAM-like–based signaling responses.
Mild thrombocytopenia in female mice carriers for FlnA deficiency
FlnA deficiency in mice results in lethality in utero as a result of pericardiac and visceral hemorrhage, severe cardiac structural defects, and aberrant vascular patterning (Feng et al., 2006; Hart et al., 2006). Because the FLNA gene is located on the X chromosome, we determined whether it conferred survival value to platelets in 6–8-wk-old heterozygous female mice carrying both one WT and one deleted FLNA gene. FlnA+/− females had a mild thrombocytopenia (Table I). Their blood platelet count was 678 ± 278 × 103/µl (mean ± SD; n = 14), compared with 994 ± 294 × 103/µl (n = 17; P < 0.01) in WT female littermates, revealing a reduction of 32%. The mean platelet volume was increased by 8%, compared with WT female littermates (P < 0.05). FlnA+/− females also had increased leukocyte count, notably lymphocytes and granulocytes. Erythrocyte counts were normal.
Platelets were isolated from FlnA+/− and WT mouse blood, fixed and permeabilized, and FlnA expression was determined by intracellular flow cytometry using a rabbit antibody directed against FlnA hinge 1 (Fig. 1, A and B). Two populations, one positive and one negative, were identified in FlnA+/− females, compared with only one single FlnA-positive population in WT female littermates. 95 ± 2% (mean ± SD; n = 14) of platelets isolated from FlnA+/− females contained FlnA, which was significantly greater than the expected 50% (P < 0.0001). FlnA-null platelets were large, as indicated by forward and side scatter dot plot analysis (unpublished data).
Generation of mice carrying platelets that selectively lack FlnA
GATA1-Cre transgenic mice were used to excise the FlnAloxP gene in the erythroid/megakaryocytic lineage (Jasinski et al., 2001). FlnAloxP/loxP females were bred with GATA1-Cre males. In this cross, all offspring expressed GATA1-Cre, and females were heterozygous and males hemizygous for FlnAloxP. Only 79 (14.2%) out of the 555 offspring obtained were males. Analysis of embryos collected at different stages of development showed a high embryonic mortality associated with complex developmental abnormalities (unpublished data), which is consistent with leaky GATA1-Cre expression in early embryogenesis leading to FlnAloxP gene recombination in multiple nonhematopoietic tissues (Mao et al., 1999; Jasinski et al., 2001).
Macrothrombocytopenia in male mice with platelets lacking FlnA
Blood counts revealed a macrothrombocytopenia in 6–8-wk-old FlnAloxP/y GATA1-Cre males (Table II). The mean blood platelet count was 284 ± 70 × 103/µl (n = 16), compared with 1,773 ± 403 × 103/µl (n = 19; P < 0.001) in control FlnA+/y GATA1-Cre males, an 84% reduction. The mean volume of FlnA-null platelets was 13.2 ± 2.1 fl, compared with 6.3 ± 0.5 fl in control males (P < 0.001). FlnAloxP/y GATA1-Cre males had increased lymphocyte count compared with FlnA+/y GATA1-Cre males. Other blood cell parameters were in the normal range. The ∼2.1-fold increased platelet volume correlated with a ∼2.7-fold increase in total protein content, from 927 ± 95 fg per FlnA-expressing platelet (n = 6) to 2,464 ± 263 fg per FlnA-null platelet (n = 9; P < 0.0001), as determined by Bradford assay using BSA as standard.
FlnA expression was evaluated in platelets isolated from FlnA+/y GATA1-Cre and FlnAloxP/y GATA1-Cre males by intracellular flow cytometry (Fig. 1 C) and by SDS-PAGE and immunoblotting (Fig. 1 D). FlnA expression was absent in most platelets isolated from FlnAloxP/y GATA1-Cre males. A small population (<5%) of FlnA-containing platelets could be detected in some animals, suggesting mosaic GATA1-Cre expression, rendering incomplete FlnAloxP excision. In contrast, FlnB expression levels appeared to be increased by approximately twofold in FlnA-null platelets, compared with WT (Fig. 1, E and F).
Increased tail bleeding time in mice with platelets lacking FlnA
FlnAloxP/y GATA1-Cre males had prolonged tail bleeding times (Fig. 1 G). 13 (81%) out of 16 FlnAloxP/y GATA1-Cre males had tail bleeding >10 min, which was our selected end measurement point. In contrast, the median tail bleeding time for FlnA+/y GATA1-Cre males was 165 s (n = 16; log-rank P = 1.27 × 10−7).
Structure of platelets lacking FlnA
Loss of the GPIbα–FlnA interaction alters the morphology of the resting platelet and its underlying cytoskeleton. Platelets isolated from FlnAloxp/y GATA1-Cre male mice are large but discoid, ∼4.0 µm in diameter at their longest axis, compared with ∼2.5 µm in controls. Fig. 2 (A and B) shows representative electron micrographs of the cytoskeletons of FlnA-null platelets prepared by removing the plasma membrane in the absence of glutaraldehyde with Triton X-100 detergent. In the absence of fixative, the bulk of the filaments composing the cytoskeleton are lost. In other words, cytoskeletal integrity is diminished as the bulk of the F-actin, but not the microtubule ring, rapidly dissociates from the cytoskeleton when FlnA-null platelets are permeabilized in buffers that normally stabilize an intact F-actin cytoskeleton. The percentage of actin assembled into filaments was, however, comparable in the FlnA-null and WT platelets, representing 40% of the total actin levels. Because FlnA-null platelets are larger, they contain more total actin filaments than the control cells.
FlnA-null platelets adhere to the collagen-related peptide (CRP)–coated surfaces, which are specific for the collagen receptor GPVI but spread poorly, and have one or more unusual F-actin enriched foci within their cytoplasm (Fig. 2, C and D). In the electron microscope, these foci are aggregates of membrane skeleton and F-actin. In contrast, WT platelets activate on CRP-coated surfaces, spread, and generate blunt filopodia that are enriched in newly assembled F-actin, as indicated by both Alexa-phalloidin staining of F-actin and electron microscopy.
Fig. 2 (E and F) also shows that the limited shape change reaction of FlnA-null platelets to CRP results from a blunted actin assembly process. Although the F-actin assembly response mediated by the G protein–coupled receptor (GPCR) agonist thrombin is normal, FlnA-null platelets mount only a modest shape change and actin assembly reaction after CRP ligation of GPVI.
GPIbα expression and distribution in FlnA-null platelets
Expression of major platelet surface glycoproteins on FlnA-null platelets was examined by flow cytometry (Table III). FlnA deficiency resulted in a diminution of the expression of the VWF receptor, the GPIb-IX-V complex. Expression of the GPIbα (CD42b) subunit was reduced by 45% and that of GPIbβ (CD42c), GPV (CD42d), and GPIX (CD42a) by 25–30%. In contrast, expression of the fibrinogen receptor subunit, the integrin β3 (GPIIIa and CD41), and the collagen receptor GPVI was increased by 230 and 40%, respectively.
To demonstrate that the loss of FlnA results in untethered VWF complexes, the attachment of GPIbα to the actin cytoskeleton of FlnA-null platelets was evaluated (Fig. 3 A). FlnA+/y GATA1-Cre and FlnAloxP/y GATA1-Cre platelets were lysed in PHEM buffer containing Triton X-100, and F-actin was collected by centrifugation at 100,000 g for 30 min (Falet et al., 2002). We found that 41 ± 4% of total GPIbα was recovered in the Triton X-100–insoluble fraction of resting FlnA-positive platelets, which is consistent with earlier accounts of ∼50% of GPIbα associated with the resting cytoskeleton (Fox, 1985a,b). In contrast, all GPIbα remained Triton X-100 soluble in FlnAloxP/y GATA1-Cre platelets. The lack of a cytoskeletal linkage altered the VWF receptor topology such that it was dispersed randomly across the platelet surface (Fig. 3 B). Particle counts revealed that only 10.6 ± 3.4% of GPIbα was arrayed into linear rows in FlnA-null platelets, compared with 72.0 ± 6.2% in WT controls (5 µm2 area; n = 5), as described previously (Hoffmeister et al., 2003).
Platelet adhesion under arterial shear condition
The functionality of FlnAloxP/y GATA1-Cre platelets was tested in flow chamber experiments. Binding to a collagen-coated surface was measured in whole blood after labeling of platelets and normalization of platelet counts. Perfusion was performed under arterial shear condition (shear rate 1,500/s), which mediates binding of plasma VWF to surface-bound collagen (Ruggeri and Mendolicchio, 2007; Diener et al., 2009). Fluorescent FlnA-null platelets covered ∼70% less of the collagen-coated surface than did WT platelets (Fig. 4, A and B). Considering that FlnA-null platelets are larger in size than WT platelets, this reveals a highly significant defect in binding to collagen-coated surface under arterial shear.
To test the hypothesis that the GPIbα defect might result from a decreased stability of ligand binding rather than a decreased initial binding efficiency, we analyzed individual platelet binding to the coated surface with high temporal resolution under the same experimental conditions (Fig. 4 C). After initial tethering, FlnA-null platelets showed significantly less firm adhesion than WT platelets. Although all WT platelets dwelled for at least 500 ms on the surface and 62% adhered firmly during the observation period, 23% of the FlnA-null platelets detached after <500 ms.
ITAM and ITAM-like signaling defects in FlnA-null platelets
The significance of FlnA expression in platelet responses to soluble agonists was tested (Fig. 5). Platelets isolated from FlnA+/y GATA1-Cre or FlnAloxP/y GATA1-Cre males were treated with GPCR agonists (i.e., ADP, the thromboxane analogue U46619, and/or epinephrine) and analyzed by flow cytometry using the antibody JON/A, which is specific for active αIIbβ3 (Fig. 5 A). ADP induced comparable αIIbβ3 activation in WT and FlnA-null platelets alone or in combination with U46619, epinephrine, or both. Furthermore, U46619 induced similar actin assembly responses in WT and FlnA-null platelets (unpublished data).
The secretory and aggregatory capacities of FlnA-null platelets were further tested. WT and FlnA-null platelets were treated with PMA, which activates protein kinase C directly, and analyzed by flow cytometry for P-selectin (CD62P) expression, as a marker for α-granule secretion, or fibrinogen binding, as a marker for integrin αIIbβ3 activation (Falet et al., 2009). FlnA-null platelets expressed P-selectin and bound fibrinogen normally in response to PMA (Fig. 5, B and G).
Platelets isolated from FlnA+/y GATA1-Cre or FlnAloxP/y GATA1-Cre males were also treated with the GPCR agonist thrombin. Thrombin induced a concentration-dependent increase of P-selectin expression and fibrinogen binding in FlnA+/y GATA1-Cre platelets, reaching 93.1 ± 0.5% of P-selectin–expressing and 82.8 ± 6.9% of fibrinogen-binding platelets with 0.5 U/ml thrombin (Fig. 5, C and H). FlnA deficiency resulted in a significant decrease of platelet responses to thrombin, as only 55.9 ± 2.2% of P-selectin–expressing and 65.6 ± 5.9% of fibrinogen-binding platelets were obtained with 0.5 U/ml of thrombin.
Responses to CRP and convulxin, which are both specific for the collagen receptor GPVI (Watson et al., 2005), were severely affected by FlnA deficiency (Fig. 5, D, E, I, and J). Only 32.7 ± 6.7% of CD62P-expressing and 32.5 ± 6.6% of fibrinogen-binding FlnAloxP/y GATA1-Cre platelets were obtained with 10 µg/ml CRP, compared with 80.0 ± 6.2 and 72.5 ± 6.9% of FlnA+/y GATA1-Cre platelets, respectively. Similarly, only 42.6 ± 1.1% of P-selectin–expressing and 18.0 ± 4.8% of fibrinogen-binding FlnAloxP/y GATA1-Cre platelets were obtained with 200 ng/ml convulxin, compared with 82.8 ± 1.5% of P-selectin–expressing and 47.2 ± 2.3% of fibrinogen-binding FlnA+/y GATA1-Cre platelets, respectively.
To determine if the poor response to GPVI is extended to ITAM-like signaling, we assessed the ability of rhodocytin to activate WT and FlnA-null platelets through CLEC-2 (Suzuki-Inoue et al., 2006; Fig. 5, F and K). The data reveal that signals coming from CLEC-2 that lead to P-selectin expression and integrin αIIbβ3 activation were also highly diminished in FlnA-null platelets, in a similar fashion to those coming from GPVI.
Impaired tyrosine phosphorylation in FlnA-null platelets
The data described in the previous section suggest that FlnA modulates platelet signaling downstream of ITAM- and ITAM-like–containing receptors (i.e., GPVI and CLEC-2), cascades initiated by tyrosine phosphorylation, and activation of signaling proteins that include the protein tyrosine kinase Syk and phospholipase C-γ2 (PLC-γ2; Watson et al., 2005; Suzuki-Inoue et al., 2006). Hence, the ability of FlnA-null platelets to phosphorylate proteins on tyrosine in response to CRP, convulxin, or rhodocytin was investigated (Fig. 6). In FlnA+/y GATA1-Cre platelets, CRP, convulxin, and rhodocytin induced tyrosine phosphorylation of numerous platelet proteins, including a 130-kD protein that comigrates with PLC-γ2. The profile of tyrosine phosphorylation was markedly reduced in FlnAloxP/y GATA1-Cre platelets stimulated in the same conditions. Immunoprecipitation confirmed that Syk and PLC-γ2 phosphorylation was severely compromised in these platelets, thus revealing an important early role for FlnA in the ITAM- and ITAM-like–mediated signaling cascades (Fig. 6).
FlnA associates with Syk
Because the block in ITAM-based signaling in platelets occurs early, as judged by the decreased phosphorylation of Syk and PLC-γ2, we investigated whether FlnA interacts with any signaling molecules involved in the GPVI and CLEC-2 signaling pathways and found that FlnA directly binds to Syk. Syk immunoprecipitates were prepared from human platelet lysates and were probed against FlnA (Fig. 7 A). FlnA associated with immunoprecipitated Syk in both resting and CRP-activated platelets.
The localization of Syk in WT and FlnA-null platelets was analyzed after adherence to CRP-coated surfaces. Fig. 7 B shows WT and FlnA-null platelets on CRP-coated surfaces after staining with rabbit anti-Syk antibody BR15 and Alexa Fluor 488–conjugated phalloidin and TRITC-labeled goat anti–rabbit IgG. Syk appears to be associated with the membrane in WT platelets and more cytoplasmic in FlnA-null platelets, indicating that FlnA contributes to Syk spatial distribution to the cytoplasmic surface of the platelet plasma membrane.
The FlnA–Syk interaction was further investigated in vitro using recombinant proteins. His-tagged FlnA constructs containing repeats 1–8+24, 8–15+24, or 16–24 were incubated with a fusion protein containing GST-Syk. The construct containing FlnA repeats 1–8+24 was pulled down by GST-Syk, but not GST alone, as indicated by anti-His immunoblotting (Fig. 7 C). Binding to repeats 8–15+24 and 16–24 was not detected, indicating that Syk binds to the FlnA N-terminal region containing repeats 1–8. We further determined that Syk binding occurs to FlnA repeat 5 (Fig. 7 D), as GST-His-Syk predominantly pulled down eGFP-tagged FlnA repeat 5. Weak binding to repeats 1–3, 4, and 6–7, but not to FlnA actin-binding domain and repeat 8, was also observed.
Our results show that FlnAloxP/y GATA1-Cre males have <15% of normal blood platelet count and increased tail bleeding time. These mice have enlarged platelets that lack FlnA. FlnA-null platelets lack normal actin membrane attachments and have decreased and altered GPIbα expression on their surface and impaired adhesion to collagen under flow conditions. Unexpectedly, FlnA-null platelets also have severe functional and signaling impairment in response to stimulation through ITAM- and ITAM-like–mediated signal receptors GPVI and CLEC-2, signaling which requires activation of the protein tyrosine kinase Syk. Interaction between platelet FlnA and Syk was detected by immunoprecipitation and reconstituted using purified proteins.
Because FlnA deficiency is embryonic lethal in mice (Feng et al., 2006; Hart et al., 2006), mice lacking FlnA in platelets were generated using the Cre–loxP system. FlnAloxP/loxP female mice were bred with erythroid/megakaryocytic-specific GATA1-Cre transgenic male mice. In agreement with the inactivation of the PIGA gene (Jasinski et al., 2001), we found that GATA1-Cre caused high efficiency FLNA gene inactivation in mouse megakaryocytes, as FlnA expression was absent in most platelets isolated from FlnAloxP/y GATA1-Cre male mice. A small population (<5%) of FlnA-containing platelets could be detected in some animals, suggesting mosaic GATA1-Cre expression, rendering incomplete FlnAloxP excision. Whether these platelets contain full-length FlnA or a truncated form, as shown for talin (Priddle et al., 1998), remains to be investigated. The high efficiency of gene inactivation is similar to that described for the widely used megakaryocyte-specific Pf4-Cre transgenic mouse and posits the GATA1-Cre mouse as a strong alternative for studies in megakaryocytes and platelets (Léon et al., 2007; Petrich et al., 2007; Tiedt et al., 2007; Hitchcock et al., 2008; Wen et al., 2009).
FlnA deficiency in FlnAloxP GATA1-Cre mice was associated with a high embryonic mortality, as our breeding strategy resulted in only 14.2% of pups being hemizygous FlnA-deficient males, which was lower than the expected 50%. GATA1-Cre is turned on early in embryogenesis and is also expressed outside of the hematopoietic system (Mao et al., 1999; Jasinski et al., 2001). Analysis of embryos collected at different stages of development showed complex developmental abnormalities, which is consistent with leaky Cre expression in early embryogenesis leading to FlnAloxP gene recombination in multiple nonhematopoietic tissues. The FlnAloxP/y GATA1-Cre males that survived had a macrothrombocytopenia, with <15% of normal blood platelet count and ∼2.5-fold enlarged platelets, but displayed no gross anatomical anomalies.
FlnA-null platelets had deleterious cytoskeletal alterations, resulting in inadequately cross-linked cytoplasmic actin filaments and poor attachment of the filaments to the membrane. FlnB expression levels appeared to be increased in FlnA-null platelets, compared with WT, which is consistent with a recently described compensatory FlnB expression increase in FlnA knockdown endothelial cells (Del Valle-Pérez et al., 2010). FlnB concentration levels, however, may be comparable in WT and FlnA-null platelets, relative to the increased volume and total protein content of FlnA-null platelets, compared with WT controls. Nevertheless, although FlnB has the same binding site for the cytoplasmic tail of GPIbα on repeat 17 (Nakamura et al., 2006), it does not compensate for FlnA absence, and no GPIbα was detected linked to F-actin in FlnA-null platelets. The F-actin/G-actin ratio at rest and the actin assembly reaction of FlnA-null platelets in response to thrombin were normal, as assessed by centrifugation of Triton X-100 platelet lysates and/or intracellular flow cytometry. The data show that the F-actin disassembly/assembly steps in the activation reaction occur normally in FlnA-null platelets, as described for FlnA-null fibroblasts (Feng et al., 2006; Hart et al., 2006).
As predicted, biochemical experiments revealed that all of the VWF receptor GPIbα was Triton X-100 soluble and, therefore, untethered to the actin cytoskeleton in FlnA-null platelets, resulting in a disorganized GPIbα topology on the platelet surface. Together, the data show that FlnA is required for normal platelet morphology and circulation and confirm that the macrothrombocytopenia associated with Bernard-Soulier syndrome results from the lack of the FlnA-GPIbα linkage to the actin cytoskeleton (Ware et al., 2000; Kanaji et al., 2002). Low platelet counts and large platelets have been described in periventricular heterotopia female patients (Parrini et al., 2006). These patients were carriers with only one mutant allele and are expected to have two platelet populations: one that is normal and one expressing mutant FlnA. FlnA+/− female mice have a mild thrombocytopenia with only 5% of their circulating platelets being FlnA null and enlarged. Further studies are required to determine the distribution of circulating FlnA-deficient platelets in periventricular heterotopia patients.
FlnAloxP/y GATA1-Cre males had a more severe bleeding phenotype than their macrothrombocytopenia would predict, indicating that FlnA-null platelets have functional defects. Indeed, there was loss of GPVI and GPIbα function in the absence of FlnA, as adhesion to collagen and collagen-associated VWF was diminished in FlnA-null platelets under flow conditions. Whether FlnA influences the ligand binding properties of GPIbα has been controversial in the literature. Some groups have reported that it enhances (Dong et al., 1997; Cranmer et al., 1999, 2005; Mistry et al., 2000; Williamson et al., 2002; Schade et al., 2003) and others that it diminishes (Englund et al., 2001). Our studies directly confirm the former, pointing out the importance of this interaction to platelet tethering under arterial shear conditions. Diminished binding to collagen and collagen-associated VWF and poor adherence under flow could result from the decreased density of GPIbα on the surface of FlnA-null platelets or from its changed topology.
Unexpectedly, FlnA-deficient platelets failed to respond normally to agonists for the ITAM- and ITAM-like–containing receptors GPVI or CLEC-2. In particular, the actin assembly reaction was blunted in response to rhodocytin (unpublished data) or CRP, but not thrombin. FlnA-null platelets also had decreased α-granule secretion and integrin αIIbβ3 activation. This diminished response is not a result of the lack of GPVI or CLEC-2, as FlnA-null platelets express normal or higher numbers of these receptors per cell (unpublished data). The possibility that FlnA-null platelets have general secretory and aggregatory defects was also eliminated because of their normal response to PMA, a direct activator of protein kinase C which bypasses ligation of platelet membrane receptors. Furthermore, αIIbβ3 activation in response to the GPCR agonist ADP alone or in combination with the thromboxane analogue U46619 and/or epinephrine was comparable between FlnA-null and WT platelets.
Platelets express at least three ITAM or ITAM-like–containing receptors: the low affinity IgG Fc receptor FcγRIIA, the collagen receptor GPVI–FcRγ chain complex, and CLEC-2. FcγRIIA is expressed on human but not mouse platelets. After ligation of GPVI with collagen, CRP, or convulxin, the two YXXL motifs of the associated FcRγ chain ITAM become phosphorylated by Src family kinases, i.e., Fyn and Lyn (Ezumi et al., 1998). Syk recognizes and binds the phosphorylated ITAM, leading to Syk autophosphorylation and activation (Poole et al., 1997). Downstream signals through adapter proteins, such as LAT and SLP-76, lead to PLC-γ2 phosphorylation and activation (Watson et al., 2005). Active PLC-γ2 hydrolyzes membrane bound polyphosphoinositides, notably phosphatidylinositol 4,5-bisphosphate, to form soluble inositol 1,4,5-trisphosphate, which releases cytosolic calcium from the internal stores, and membrane-bound 1,2-diacylglycerol, which activates protein kinase C. Signals are also sent to phosphoinositide 3-kinase, generating D3-containing polyphosphoinositides (Pasquet et al., 1999). CLEC-2 only contains a single YXXL motif but appears to be active as a dimer (Watson et al., 2009; Hughes et al., 2010) and signals through many of the same components as GPVI because platelets lacking LAT, SLP-76, or PLC-γ2 are not receptive to ligation of this receptor with rhodocytin (Suzuki-Inoue et al., 2006; Fuller et al., 2007).
We traced the failure of FlnA-null platelets to respond normally to agonists for GPVI or CLEC-2 to decreased Syk phosphorylation, which then poorly activates downstream signaling effectors such as PLC-γ2. Furthermore, Syk appeared to be cytoplasmic in FlnA-null platelets stimulated through GPVI rather than membrane associated as in WT controls. Together, the data indicate that FlnA contributes to Syk spatial distribution to the cytoplasmic surface of the platelet plasma membrane, a prerequisite for the docking to the ITAM or ITAM-like sequences of GPVI and CLEC-2 and phosphorylation of Syk. This is a surprising finding and one that posits FlnA as essential for the very early Syk dependent steps in ITAM and ITAM-like based signaling.
Previous observations have suggested that GPVI and CLEC-2 must be coupled to the platelet cytoskeleton to function. CLEC-2 signaling is inhibited by cytochalasin d (Pollitt et al., 2010) and signaling of GPVI is greatly diminished by treating platelets with latrunculin A (unpublished data), an agent which depolymerizes actin filaments. The actin filament cross-linking activity of FlnA may be important for Syk recruitment and activation by the two receptors. FlnA and Syk were associated in both resting and activated platelets. Direct binding was reconstituted using purified Syk and FlnA fusion proteins and was restricted to Ig repeat 5 of FlnA using a series of FlnA truncates. Binding to FlnA Ig repeat 5 is novel, as most FlnA interacting partners bind to the C-terminal region containing Ig repeats 16–24 (Stossel et al., 2001; Feng and Walsh, 2004; Zhou et al., 2010). Mutations in FlnA Ig repeat 5 have been associated with familial cardiac valvular dystrophy (Kyndt et al., 2007). However, the consequence of these mutations on platelet activation remains to be determined, as does the location of the binding interface on Syk.
Stable platelet adhesion and aggregate formation under arterial shear condition, events which normally require activation of the αIIbβ3 integrin downstream of GPVI ligation by collagen and that of GPIbα by VWF, were severely impaired in the absence of FlnA (Ruggeri and Mendolicchio, 2007; Diener et al., 2009). Syk participates in the signaling cascade induced by GPIbα ligation, particularly in the calcium rise and activation of PI-3 kinase which follow ligation, and is intimately involved in downstream signaling to αIIbβ3 (Asazuma et al., 1997; Yanabu et al., 1997; Falati et al., 1999; Ozaki et al., 2005). It is likely that FlnA modulates Syk phosphorylation and activation downstream of GPIbα ligation as well. Further experimental evidence is required to test this hypothesis.
Our data further show that P-selectin expression and fibrinogen binding, but not actin assembly, were attenuated in FlnA-null platelets stimulated with the GPCR agonist thrombin. This could be a result of the altered expression and location of GPIbα, such that it is less able to present thrombin to its GPCR PAR4. Another explanation is that signals originated from GPVI or αIIbβ3 contribute to the thrombin signaling cascade. Boylan et al. (2006) have reported that platelets that lost GPVI after in vivo injection of an anti-GPVI antibody had diminished functional responses to thrombin and that outside-in signaling downstream of αIIbβ3 engagement also uses an ITAM-based pathway (Boylan et al., 2008). Older studies have shown that Syk is phosphorylated and activated after platelet stimulation with thrombin (Taniguchi et al., 1993; Clark et al., 1994). Thus, it is also possible that FlnA plays a yet undefined scaffolding role in thrombin receptor signaling.
In conclusion, our results show that FlnA is not only required for platelet morphology and circulation but also for signaling through platelet ITAM and ITAM-like containing receptors (i.e., GPVI and CLEC-2), which requires activation of the protein tyrosine kinase Syk. FlnA interacts with Syk through its Ig repeat 5 and, thus, docks Syk near the plasma membrane where it can bind GPVI, CLEC-2, and possibly GPIbα, initiating or reinforcing/amplifying their signaling cascades.
MATERIALS AND METHODS
FlnA+/− and FlnAloxP/loxP mice on a 129/Sv × C57BL/6 background were provided by Y. Feng and C. Walsh (Howard Hughes Medical Institute, Boston, MA; Feng et al., 2006). GATA1-Cre transgenic mice on a BALB/c background were provided by Y. Fujiwara and S.H. Orkin (Dana Farber Cancer Institute, Howard Hughes Medical Institute, Boston, MA; Jasinski et al., 2001). FlnAloxP/loxP females were bred with homozygous GATA1-Cre transgenic males to excise the floxed FLNA allele in the erythroid/megakaryocytic lineage. Mice were treated as approved by the Children’s Hospital Animal Care and Use Committee.
Antibodies and reagents.
Rabbit antibody directed against aa 1758–1767 of mouse FlnA (APQYNYPQGS) was produced by Invitrogen. Rabbit anti-FlnA antibody 4762 was purchased from Cell Signaling Technology. Rabbit anti-FlnB antibody and mouse anti-phosphotyrosine antibody 4G10 were purchased from Millipore. Mouse anti-phosphotyrosine antibody PY20 was obtained from BD. Rabbit anti-Syk antibody BR15 was supplied by M. Tomlinson (DNAX Research Institute, Palo Alto, CA). Control rabbit antibody, rabbit anti-Syk antibody N19, and rabbit anti–PLC-γ2 Q20 were obtained from Santa Cruz Biotechnology, Inc. Antibodies directed against mouse GPIbα (CD42b), GPIbβ (CD42c), GPIX (CD42a), GPV (CD42d), GPVI, and active αIIbβ3 were obtained from Emfret Analytics. Antibodies directed against mouse P-selectin (CD62P) and GPIIIa (CD61 and integrin β3) were obtained from BD. CM-Orange, Oregon green 488–labeled fibrinogen, and Alexa Fluor 488–labeled donkey anti–mouse IgG and anti–rabbit IgG antibodies were obtained from Invitrogen. Protein G Sepharose and glutathione Sepharose beads were obtained from GE Healthcare. The CRP (amino acids GCO(GPP)10GCOG) was synthesized by the Tufts University Core Facility and cross-linked as described previously (Morton et al., 1995). Rhodocytin was purified from the venom of Calloselasma rhodostoma (Suzuki-Inoue et al., 2006). ADP was obtained from Chrono-log and U46619 from Enzo Life Sciences. All other reagents were of the highest purity available (Sigma-Aldrich).
Complete blood count.
Blood was collected from mice by retroorbital plexus bleeding in EDTA. Peripheral blood platelet count was performed using an ADVIA 120/2120 automated hematology analyzer (Bayer HealthCare; Falet et al., 2009).
Bleeding time assay.
Mouse tail bleeding times were determined by snipping 5 mm of distal mouse tail and immediately immersing the tail in 37°C isotonic saline (Ware et al., 2000). A complete cessation of bleeding was defined as the bleeding time. Measurements exceeding 10 min were stopped by cauterization of the tail. Data were statistically analyzed by the Kaplan-Meier method (Kaplan and Meier, 1958).
Blood was collected from mice by retroorbital plexus bleeding and anticoagulated in ACD for experiments with washed platelets or in citrate for experiments with PRP. Mouse PRP was obtained by centrifugation of the blood at 100 g for 8 min, followed by centrifugation of the supernatant and the buffy coat at 100 g for 6 min. Mouse platelets were isolated by two sequential centrifugations of the PRP at 1,200 g for 5 min in 140 mM NaCl, 5 mM KCl, 12 mM trisodium citrate, 10 mM glucose, and 12.5 mM sucrose, pH 6.0 (washing buffer), and resuspended in 10 mM Hepes, 140 mM NaCl, 3 mM KCl, 0.5 mM MgCl2, 5 mM NaHCO3, and 10 mM glucose, pH 7.4 (resuspension buffer), as previously described (Hoffmeister et al., 2003). Platelet concentration was adjusted to 2–5 × 108/ml depending on the assay performed, and platelets were allowed to rest for 30 min before use.
Washed platelets were fixed and permeabilized in Cytofix/Cytoperm solution (BD), washed in Perm/Wash solution (BD), and incubated with a rabbit anti-FlnA antibody or a rabbit anti-FlnB antibody, followed by Alexa Fluor 488–labeled donkey anti–rabbit IgG antibody. For the actin assembly assay, resting or activated, fixed, and permeabilized platelets were incubated with 2 µM TRITC-labeled phalloidin (Falet et al., 2009).
For the functional assays, resting or activated washed platelets were incubated with a FITC-labeled anti–P-selectin antibody or Oregon green 488–labeled fibrinogen for 30 min at room temperature (Falet et al., 2009). Platelets were activated in PRP in the presence of PE-labeled JON/A. Fluorescence was quantified using a FACSCalibur flow cytometer (BD). A total of 20,000 events were analyzed for each sample.
Platelet cytoskeletal rearrangements.
Platelets were activated by attachment to CRP-coated coverslips for 10 min at 37°C. For F-actin staining in the light microscope, they were incubated with 0.1 µM Alexa Fluor 568–conjugated phalloidin after fixation with 1% formaldehyde, permeabilization with 0.1% Triton X-100, and appropriate washing and blocking. For electron microscopy, the CRP-attached and activated platelets were permeabilized with 0.75% Triton in PHEM buffer for 2 min at 37°C, washed in PHEM without Triton X-100, and then fixed with 1% glutaraldehyde in PHEM for 10 min. The samples were washed into water and rapidly frozen by slamming them into a liquid helium-cooled copper block. Frozen samples were transferred to a liquid nitrogen–cooled stage on a CFE-60 apparatus (Cressington), warmed to −80°C for 60 min, and metal cast with 1.4 nm tantalum-tungsten at 45° with rotation and 5 nm of carbon at 90° without rotation. Metal replicas were floated from the coverslips in 25% hydrofluoric acid, water washed, and examined by transmission electron microscopy (JEOL 1200 EX).
Immunogold anti-GPIbα glycoprotein labeling.
Resting platelets were attached to polylysine-coated coverslips by centrifugation at 300 g for 5 min in PBS containing 0.1% glutaraldehyde. They were fixed for 5 min in 0.5% glutaraldehyde in PBS, unreacted aldehydes blocked using a 2 min incubation with 0.1% sodium borohydride, and washed into PBS/BSA. The platelets were surface labeled with a rat anti-GPIbα monoclonal antibody. After washing, platelets were incubated with 10 nm of gold coated with goat anti–rat IgG. They were washed, postfixed with 1% gluteraldehyde for 10 min, washed into water, and processed by rapid freezing, freeze drying, and metal casting.
SDS-PAGE and immunoblot analysis.
Platelets were lysed at 4°C in 1% Nonidet P-40, 150 mM NaCl, and 50 mM Tris, pH 7.4, containing 1 mM EGTA, 1 mM Na3VO4, and Complete protease inhibitor cocktail (Roche). SDS-PAGE sample buffer containing 5% β-mercaptoethanol was added. After boiling for 5 min, platelet proteins were separated on 8% polyacrylamide gels and transferred onto an Immobilon-P membrane (Millipore). Membranes were incubated overnight in 0.2% Tween-20, 100 mM NaCl, and 20 mM Tris, pH 7.4, containing 1% BSA, and then probed with antibodies directed against proteins of interest. Detection was performed with an enhanced chemiluminescence system (Thermo Fisher Scientific).
For the immunoprecipitation studies, Nonidet P-40 lysates were centrifuged at 14,000 g for 10 min at 4°C. Soluble fractions were incubated with antibodies directed against proteins of interest for 2 h at 4°C, followed by incubation with protein G–conjugated Sepharose beads for 1 h at 4°C. After washing, immune complexes were resolved by SDS-PAGE as described in this section.
For the purified recombinant protein binding assays, His-tagged and eGFP-tagged FlnA fragments were expressed in Sf9 cells and purified as described previously (Nakamura et al., 2007). GST-Syk was purchased from Active Motif. FlnA fragments and GST-Syk were incubated for 2 h at room temperature in 1% Nonidet P-40, 150 mM NaCl, and 50 mM Tris, pH 7.4, containing 1 mM EGTA, 1 mM Na3VO4, and Complete protease inhibitor cocktail, followed by incubation with glutathione conjugated Sepharose beads for 1 h. After washing, immune complexes were resolved by SDS-PAGE as described in this section.
Flow chamber studies.
Blood was collected from mice by retroorbital plexus bleeding and anticoagulated in 40 µM PPACK and 20 µg/ml enoxaparin. PRP was separated by centrifugation at 80 g for 6 min and purified at 180 g for 5 min for WT and at 80 g for 5 min for FlnA-null platelets, respectively. Platelets were collected by centrifugation at 640 g for 6 min in the presence of 2 µg/ml prostacyclin and resuspended in modified Tyrode’s buffer (140 mM NaCl, 0.36 mM Na2HPO4, 3 mM KCl, 12 mM NaHCO3, 5 mM Hepes, and 10 mM glucose, pH 7.3) containing 0.2% BSA and labeled with 2.5 mg/ml calcein orange. Platelet-poor blood was reconstituted with labeled platelets and remaining plasma at 108/ml platelets. Using a 0.0127-cm silicon rubber gasket, a parallel plate flow chamber (GlycoTech) was assembled onto 35-mm diameter round glass coverslips, which had been coated with 100 µg/ml collagen type I (Nycomed). Perfusion was performed at a shear rate of 1,500/s for 3 min. Platelet adhesion to collagen-associated plasma VWF was monitored with an Axiovert 135 inverted microscope (Carl Zeiss, Inc.) at 32× and a silicon-intensified tube camera C 2400 (Hamamatsu Photonics) and analyzed with Image SXM 1.62 (National Institutes of Health).
Statistical analysis was performed using the unpaired Student’s t test. A p-value <0.05 was considered statistically significant.
We thank Drs. Yuanyi Feng and Christopher Walsh for providing the FlnA+/− and FlnAloxP/loxP mice, Drs. Yuko Fujiwara and Stuart Orkin for providing the GATA1-Cre mice, Dr. Mike Tomlinson for the anti-Syk antibody, and Teresa Collins and Dr. Fumihiko Nakamura for the recombinant FlnA constructs. We thank Drs. Karin Hoffmeister and Larry Frelinger for advice and reagents, Dr. François Maignen for statistical expertise, and Mike Marchetti for technical help.
This work was supported by National Institutes of Health grants HL-056252 (J.H. Hartwig), HL-056949 (J.H. Hartwig and D.D. Wagner), and HL-059561 (J.H. Hartwig and H. Falet). A.Y. Pollitt was supported by the British Heart Foundation (BHF) project grant (PG/07/116). S.P. Watson holds a BHF Chair.
The authors have no conflicting financial interests.