Factor XI Binding to the Platelet Glycoprotein Ib-IX-V Complex Promotes Factor XI Activation by Thrombin*

Factor XI binds to high affinity sites on the surface of stimulated platelets where it is efficiently activated by thrombin. Here, we provide evidence that the factor XI binding site on platelets is in the glycoprotein (GP) Ib (cid:1) subunit of the GP Ib-IX-V complex as follows. 1) Ber-nard-Soulier platelets, lacking the complex, are deficient in factor XI binding; 2) two GP Ib (cid:1) ligands, SZ-2 (a monoclonal antibody) and bovine von Willebrand factor, inhibit factor XI binding to platelets; 3) by surface plasmon resonance, factor XI bound specifically to glycocali- cin (the extracellular domain of GP Ib (cid:1) ) in Zn 2 (cid:2) -depend-ent fashion ( K d app (cid:1) 52 n M ). We then investigated whether glycocalicin could promote factor XI activation by thrombin, another GP Ib (cid:1) ligand. In the presence of high molecular weight kininogen (45 n M ), Zn 2 (cid:2) and Ca 2 (cid:2) ions, thrombin activated factor XI in the presence of glycocalicin at rates comparable with those seen in the presence of dextran sulfate (1 (cid:3) g/ml). With higher high Assay of FXI Activation— Activation of FXI (60 n M ) by thrombin (1.25 n M ) was measured by chromogenic assay. Incubations were carried out at 37 °C in 200 (cid:4) l of Tris-buffered saline (50 m M Tris, 150 m M NaCl, pH 7.3) with 1% bovine serum albumin and glycocalicin (80 n M ) or dextran sulfate (1 (cid:4) g/ml, average M r (cid:4) 500,000; Sigma), or gel-filtered platelets (activated by incubation at 37 °C for 1 min with thrombin receptor activation peptide, SFLLRN-amide, 25 (cid:4) M ). After dilution to a final volume of 1 ml with Tris-buffered saline with 1% bovine serum albumin containing 600 (cid:4) M S2366 (EPR para -nitroanilide, Chromogenix, Mo¨ln- dal, Sweden), the amount of free para -nitroanilide was determined by measuring the changes in absorbance at 405 nm ( A 405 ). The amount of FXIa generated was assayed by reference to a standard curve con-structed using purified FXIa. Surface Plasmon Resonance Studies of FXI Binding to Glycocalicin— Binding studies were performed on a Biacore 2000 Flow Biosensor (Biacore, Inc., Uppsala, Sweden). Glycocalicin contains seven cysteine residues, six of which are involved in intrachain disulfide linkages. One cysteine, Cys-65, remains free (24). Glycocalicin was therefore immobilized on a carboxymethyldextran sensor chip by thiol coupling. The carboxymethyl groups in the surface matrix were modified using 0.05 M N -hydroxysuccinimide, 0.2 M N -ethyl- N (cid:5) -(dimethylaminopropyl)carbo- diimide for 2 min. This was followed by a 4-min pulse of 80 m M 2-(2-pyridinyldithio)ethaneamine hydrochloride.

The primary structure of FXI reveals the presence of four repeated Apple domains (A1-A4) within the heavy chain followed by a light chain containing the serine protease catalytic domain (11). Our laboratory has demonstrated that the A3 domain of FXI is essential for the binding of FXI to platelets as both A3 domain peptides and a recombinant A3 domain bind specifically and saturably to the platelet surface (10,12). However, the platelet receptor(s) responsible for this interaction has not been identified.
One potential receptor was suggested by our earlier work (13) with platelets from patients with the congenital bleeding disorder Bernard-Soulier syndrome (14,15). This disorder results from a deficiency or functional defects of the platelet glycoprotein (GP) Ib-IX-V complex, a large plasma membrane complex comprising four polypeptide chains: GP Ib␣, GP Ib␤, GP IX, and GP V, arranged in a stoichiometry of 2:2:2:1, respectively (16 -18). Approximately 25,000 copies of the first three peptides reside in the platelet surface along with half as many copies of GP V. The GP Ib-IX-V complex is the receptor responsible for adhering platelets to sites of injury, a function that it carries out by binding von Willebrand factor. The GP Ib-IX-V complex also binds thrombin with high affinity (19), thereby increasing the efficiency of platelet activation by low concentrations of this agonist. The binding sites within the complex for both von Willebrand factor and thrombin reside within a region encompassing the first 300 amino acids of GP Ib␣ and are partially overlapping (20). Platelet activation by low concentrations of thrombin requires intact GP Ib␣, and the rate of activation correlates with binding to a small number of high affinity sites (K d ϳ0.30 nM) that are missing on Bernard-Soulier platelets. Bernard-Soulier platelets have also been shown to be defective in prothrombin consumption, possibly as a result of their failure to bind FXI (13,19,21). Because our earlier study suggests that FXI might bind GP Ib, we investigated whether FXI may interact with GP Ib-IX complex using both activated gel-filtered platelets and glycocalicin (the soluble extracellular region of GP Ib␣) purified from human platelets. Furthermore, we directly examined Bernard-Soulier platelets to determine whether they bind FXI as normal platelets do. We report evidence that on activated platelets FXI binds GP Ib␣ within the GP Ib-IX-V complex, an interaction that may be important in promoting the activation of FXI by thrombin. (Fairhaven, MA). Carrier-free Na 125 I was from Amersham Biosciences, Inc. The chromogenic substrate for measurement of FXIa activity (S2366) was obtained from Chromogenix (Mölndal, Sweden). The thrombin receptor agonist peptide (TRAP), SFLLRN-amide, was synthesized at the Protein Chemistry Facility of the University of Penn-sylvania on the Applied Biosystems 430A synthesizer (Foster City, CA), and reverse-phase high performance liquid chromatography was used to purify it to Ͼ99% homogeneity. A monoclonal antibody (SZ-2), which recognizes the N-terminal extracellular globular domain of GP Ib␣, blocks thrombin binding to platelets at low concentration, and inhibits thrombin-induced platelet aggregation (22), was used in experiments examining FXI binding to platelets. An isotype-specific mouse IgG2A chain control antibody was purchased from Sigma.

Materials-Human
Bernard-Soulier Platelets-Bernard-Soulier platelets were isolated from the platelet-rich plasma of two well-studied patients (T. H. and A. H.) whose clinical histories have been previously reported (13). T. H. is a 42-year-old male of maternal African-American and West Indian descent who experienced prolonged hemorrhage after circumcision and was noted to have thrombocytopenia, giant platelets, a prolonged bleeding time, and megakaryocytes of normal number and appearance, with multiple subsequent hospitalizations for recurrent epistaxis (spontaneous and traumatic), intrapleural bleeding, and hematuria. A. H. (maternal first cousin of T. H.) is a 39-year-old female of African-American ancestry who has experienced recurrent severe epistaxis and multiple spontaneous bleeding episodes since the age of 2.
Radiolabeling of FXI-Purified FXI was radiolabeled with 125 I by a minor modification (8) of the iodogen method to a specific activity of 5 ϫ 10 6 cpm/g. The radiolabeled FXI retained Ͼ98% of its biological activity.
Preparation of Washed Platelets-Platelets were prepared from normal donors as described (9,10,12). Platelet-rich plasma obtained from citrated human blood was centrifuged, and the platelets were resuspended in calcium-free Hepes-Tyrodes buffer (126 mM NaCl, 2.7 mM KCl, 1 mM MgCl 2 , 0.38 mM NaH 2 PO 4 , 5.6 mM dextrose, 6.2 mM sodium Hepes, 8.8 mM Hepes-free acid, 0.1% bovine serum albumin), pH 6.5, and gel-filtered on a column of Sepharose 2B equilibrated in calciumfree Hepes-Tyrodes buffer, pH 7.2. Platelets were counted electronically (Coulter Electronics, Hialeah, FL). Blood was obtained from the Bernard-Soulier patients, anticoagulated with citrate/phosphate/dextrose/ adenine, and allowed to sediment at 1 ϫ g for 4 h. The supernatant platelet-rich plasma (ϳ10 8 platelets/ml) was centrifuged briefly (1,000 ϫ g, 3 min) to remove erythrocytes and then loaded on a Sepharose 2B column and further processed as described for normal platelets above.
Preparation of Glycocalicin from Human Platelets-Glycocalicin was extracted from human platelets and purified as previously described (23).
Platelet Binding Experiments-Platelets were pre-warmed to 37°C and incubated at a concentration of 1 ϫ 10 8 /ml in calcium-free Hepes-Tyrodes buffer, pH 7.3, in a 1.5-ml Eppendorf plastic centrifuge tube with a mixture of radiolabeled FXI, divalent cations, a thrombin receptor activation peptide (SFLLRN-amide) as a platelet agonist (8,10), and HK or antibodies or other proteins as designated in the figure legends. All incubations were performed at 37°C without stirring after an initial mixing of the reaction mixture. At various added FXI concentrations, aliquots were removed (100 l) and centrifuged through a mixture of silicone oils as described (8,10). Saturable binding was corrected for any nonspecific component by subtracting the amount of radioactivity bound in the presence of a 100-fold excess of unlabeled FXI.
Assay of FXI Activation-Activation of FXI (60 nM) by thrombin (1.25 nM) was measured by chromogenic assay. Incubations were carried out at 37°C in 200 l of Tris-buffered saline (50 mM Tris, 150 mM NaCl, pH 7.3) with 1% bovine serum albumin and glycocalicin (80 nM) or dextran sulfate (1 g/ml, average M r ϭ 500,000; Sigma), or gel-filtered platelets (activated by incubation at 37°C for 1 min with thrombin receptor activation peptide, SFLLRN-amide, 25 M). After dilution to a final volume of 1 ml with Tris-buffered saline with 1% bovine serum albumin containing 600 M S2366 (EPR para-nitroanilide, Chromogenix, Mölndal, Sweden), the amount of free para-nitroanilide was determined by measuring the changes in absorbance at 405 nm (A 405 ). The amount of FXIa generated was assayed by reference to a standard curve constructed using purified FXIa.
Surface Plasmon Resonance Studies of FXI Binding to Glycocalicin-Binding studies were performed on a Biacore 2000 Flow Biosensor (Biacore, Inc., Uppsala, Sweden). Glycocalicin contains seven cysteine residues, six of which are involved in intrachain disulfide linkages. One cysteine, Cys-65, remains free (24). Glycocalicin was therefore immobilized on a carboxymethyldextran sensor chip by thiol coupling. The carboxymethyl groups in the surface matrix were modified using 0.05 M N-hydroxysuccinimide, 0.2 M N-ethyl-NЈ-(dimethylaminopropyl)carbodiimide for 2 min. This was followed by a 4-min pulse of 80 mM 2-(2pyridinyldithio)ethaneamine hydrochloride. Glycocalicin, 20 g/ml in sodium acetate buffer, pH 4.5, was injected to a response level of 500 response units. The remaining active sites were blocked with a 4-min infusion of 50 mM L-cysteine, 1 M NaCl. The base line of immobilized glycocalicin was stabilized by repeated injections of 1 M NaCl to a resulting response level of ϳ100 response units. Nonspecific binding was determined by protein binding to a derivatized and blocked flow cell. Serial dilutions of FXI in Hepes-buffered saline, 0.005% surfactant P-20 (Biacore, Inc.) with or without 10 M ZnCl 2 were injected with a 10-min association time and a 7-min dissociation time. After subtraction of nonspecific binding curves, the association and dissociation rate constants and the resulting equilibrium dissociation constant (K d ) were determined using a global fit to a one-to-one Langmuir model on Biaevaluation 3.0 software (Biacore, Inc.). The best fit was determined by a 2 value of less than 10 or less than 5% of the equilibrium response units for the highest concentration. The 2 value is the square of the differences between the theoretical ideal curve and the actual curve, calculated according to the equation, where r f indicates the fitted value at a given point, r x indicates the experimental value at that point, n indicates the number of data points, and p indicates the number of fitted parameters. For comparison, data were also analyzed using the steady-state affinity approach (Biaevaluation 3.1), where the response units achieved at equilibrium (R eq ) are plotted versus the analyte concentration and fitted to a hyperbolic binding curve.

I-FXI Binding to Bernard-Soulier Platelets-Because previous studies suggested a deficient interaction of FXI with
Bernard-Soulier platelets (13), we studied the binding of 125 I-FXI to platelets derived from two well-studied cases of Bernard-Soulier syndrome (A. H. and T. H. (25)). These patients are first cousins in whom the syndrome is caused by a missense mutation of the codon for Leu-129 of GP Ib␣, causing deficiency and dysfunction of the polypeptide. Platelets were incubated unstirred at 37°C with a mixture of unlabeled and 125 I-labeled FXI in the presence of either 45 nM HK, 25 M ZnCl 2 , 2 mM CaCl 2 , or only 25 M ZnCl 2 . Aliquots of incubation mixtures were centrifuged through silicone oil barriers to separate platelets from unbound proteins. Binding of FXI to stimulated Bernard-Soulier platelets in the presence of HK, Zn 2ϩ , and Ca 2ϩ demonstrate 675 Ϯ 150 sites per platelet (K d app ϭ 12.1 nM Ϯ 1.3) for T. H. and 1,095 Ϯ 220 sites per platelet (K d app ϭ 11.75 nM Ϯ 1.1) for A. H. compared with 1,550 nM Ϯ 350 FXI sites per platelet (K d app ϳ12.5 nM Ϯ 1.8) for a normal donor (Fig. 1A). In the presence of Zn 2ϩ without added HK, Bernard-Soulier platelets from patient T. H. demonstrated the total absence of FXI binding, whereas platelets from patient A. H. had 347 Ϯ 80 FXI binding sites per platelet (K d app of 10.50 nM Ϯ 0.95) compared with 771 Ϯ 142 sites per platelet (K d app of 13.75 nM Ϯ 1.5) for normal platelets (Fig. 1B). Thus, the number of FXI-binding sites is reduced on Bernard-Soulier platelets, suggesting the possibility that FXI may interact with GP Ib on the platelet surface in a Zn 2ϩ -dependent fashion.
Effects of a GP Ib␣ Monoclonal Antibody and of Bovine von Willebrand Factor on 125 I-FXI Binding to Platelets-To test the hypothesis that FXI binds to the GP Ib-IX-V complex, we determined the effects on platelet FXI binding of a monoclonal antibody (SZ-2) directed against the N-terminal extracellular globular domain of GP Ib␣. SZ-2 effectively inhibited FXI binding to activated platelets in the presence of HK and Zn 2ϩ ions (IC 50 ϭ 5 ϫ 10 Ϫ7 M, Fig. 2A). An IgG2A isotype control showed no effect on FXI binding to activated platelets. Bovine von Willebrand factor, which binds spontaneously to the GP Ib␣ N terminus in a region close to the thrombin binding site, blocks FXI binding Ͼ90% at a concentration of 0.4 g/ml (IC 50 ϭ 0.1 g/ml Fig. 2B). These experiments provide additional evidence that FXI binds directly to the GP Ib-IX-V complex on the platelet surface.
FXI Binding to Glycocalicin Using Surface Plasmon Resonance-FXI, at concentrations between 2.5 and 100 nM in the presence and absence of 10 M ZnCl 2 , was infused across a Biosensor surface-containing immobilized glycocalicin. As seen in Fig. 3A, there was virtually no binding of 100 nM FXI in the absence of zinc (curve 1). In the presence of zinc, there was specific and saturable binding of FXI to glycocalicin (curves 2-6). The association rate constant (k on ϭ 7.4 Ϯ 2.2 ϫ 10 4 M Ϫ1 s Ϫ1 ), dissociation rate constant (k off ϭ 3.9 Ϯ 0.64 ϫ 10 Ϫ3 s Ϫ1 ), and K d app (52 Ϯ 6.5 nM) were obtained by global fit to a Langmuir one-to-one binding model. The K d app , obtained by steadystate affinity analysis, was 60 Ϯ 19.2 nM (Fig. 3B).
Activation of FXI by Thrombin in the Presence of Glycocalicin-Because activated platelets bind FXI (K d ϳ10 nM) and promote its activation by thrombin in the presence of HK (8), and FXI can bind with similar affinity (K d app ϳ52 nM) to GP Ib␣ (glycocalicin) (Fig. 3) and because thrombin can also bind to GP Ib␣ (26) with high affinity (K d app ϳ 0.25 nM), we reasoned that binding of both FXI and thrombin to GP Ib␣ might promote the activation of FXI by thrombin. When FXI (60 nM) was incubated with thrombin alone (1.25 nM) or with glycocalicin alone (80 nM), very low rates of FXIa generation were observed, whereas in the presence of both thrombin and glycocalicin the rate of FXI activation was significantly enhanced, even in the absence of added ZnCl 2 . In the presence of ZnCl 2 (25 M), glycocalicin, and thrombin, FXI activation rates of ϳ7.2 Ϯ 0.8 nM/min were achieved ( Fig. 4A and Table II). In contrast, when purified GP IIb-IIIa (80 nM) replaced glycocalicin, no rate enhancement was observed. When FXI was incubated with thrombin and ZnCl 2 in the presence of HK (45 nM), virtually no FXI activation was observed, whereas when either activated platelets (10 8 /ml activated with 25 M SFLLRN-amide) or glycocalicin (80 nM) were present, major rate enhancements were observed, the initial rates being ϳ21 Ϯ 4 nM/min with activated platelets and ϳ8.8 Ϯ 1.0 nM/min with glycocalicin ( Fig. 4B and Table II). By comparison, FXI activation rates of ϳ9.6 Ϯ 1.0 nM/min were observed with dextran sulfate (1 g/ml) in the absence of HK ( Fig. 4B and Table II), which inhibited FXI activation by thrombin in the presence of dextran sulfate (data not shown). When concentrations of HK were increased to that present in normal human plasma (ϳ640 nM), the initial rates of FXI activation by thrombin progressively increased to 25.2 Ϯ 2.0 nM/min in the presence of glycocalicin (Table II), whereas in the presence of activated platelets HK decreased initial FXI activation rates in a concentration-dependent manner (data not shown) unless prothrombin (1.2 M) was also present at its normal plasma concentration (see also Ref. 8). These data support the conclusion that in the presence of HK and Zn 2ϩ ions, initial rates of FXI activation by thrombin in the presence of glycocalicin are similar to or greater than those observed in the presence of activated platelets and 2.5-fold greater than those observed with dextran sulfate.
Because prothrombin and Ca 2ϩ ions can substitute for HK and Zn 2ϩ ions as cofactors for FXI binding to activated platelets and for thrombin-catalyzed FXI activation on the platelet surface (8), we also examined FXI activation by thrombin in the presence of prothrombin (1.2 M) and CaCl 2 (5 nM). In the presence of activated platelets, initial FXI activation rates of ϳ20 Ϯ 2.5 nM/min were achieved, whereas in the presence of glycocalicin the maximum initial rates observed were ϳ3.2 Ϯ 0.5 nM/min ( Fig. 4C and Table II).
FXI Activation by Thrombin in the Presence of Bernard-Soulier Platelets-To further examine the hypothesis that the GP Ib-IX-V complex on activated normal platelets comprises a FXI binding site essential for platelet-mediated FXI activation by thrombin, we compared the contributions of Bernard-Soulier platelets to those of normal platelets in thrombin-catalyzed FXI activation in the presence of HK (45 nM) and ZnCl 2 (25 M). When activated Bernard-Soulier platelets were used as a surface, the initial rates of FXI activation by thrombin were significantly decreased either in the presence or absence of HK ( Fig. 5 and Table II). Interestingly, not only is the initial rate of FXI activation by thrombin significantly decreased, but the final yield of FXIa generated on Bernard-Soulier platelets was only 20 -40% of that observed for normal platelets, a pattern we have previously observed when HK is excluded as a cofactor for FXI binding to normal platelets (8). These studies further confirm a role for GP Ib-IX-V as a platelet surface receptor important for FXI activation by thrombin. DISCUSSION The platelet GP Ib-IX-V complex is involved in several activities crucial to normal platelet function, including initial adhe- sion of platelets to exposed subendothelium, regulating certain cytoskeletal properties such as actin polymerization, and mediating the response of platelets to low concentrations of thrombin (26). It has been postulated that platelet membrane GP Ib is required for responses to low concentrations of thrombin, as evidenced from the observation that GP Ib-deficient platelets (from Bernard-Soulier patients) show a decreased sensitivity to thrombin and a low rate of activation (21). It has been determined that the 45-kDa N-terminal domain of GP Ib␣ is involved in thrombin binding primarily through a stretch of negatively charged residues spanning amino acids 268 -282 (27). This region includes three sulfated tyrosine residues (28) and has been implicated in botrocetin-induced von Willebrand factor binding (29), although the physiological relevance of botrocetin-induced binding has been recently called into question (30). It has also recently been reported that HK and FXII interact with the GP Ib-IX complex in a Zn 2ϩ -dependent manner and might regulate thrombin-induced platelet aggregation (31,32). Moreover, studies with platelets from Bernard-Soulier patients showing them to be deficient in FXI binding gave rise to the possibility that the defective prothrombin consumption observed in these platelets might have resulted from their failure to bind FXI (13). These studies prompted us to investigate whether FXI may interact with the GP Ib complex on the platelet surface.
In the presence of HK (and Zn 2ϩ ions) or prothrombin (and Ca 2ϩ ions) FXI binds reversibly and specifically to high affinity sites on the surface of stimulated platelets (8,10). The functional consequence of FXI binding to activated platelets is a 5,000 -10,000-fold acceleration of the rate of FXI activation by thrombin (8). The binding of FXI to platelets is mediated by a sequence (Ser-248 to Val-271) in the A3 domain of FXI that binds to activated platelets in a specific, reversible, and saturable manner (10,12). Two different conditions were utilized in the present study to characterize FXI interactions with activated platelets: 1) in the presence of HK and Zn 2ϩ ions or prothrombin and Ca 2ϩ ions and 2) in the presence of Zn 2ϩ alone. The interaction of FXI with normal platelets is optimal in the presence of HK, Ca ϩ2 , and Zn ϩ2 with ϳ1,500 sites/ platelet and a K d app ϳ 10 -15 nM (9). However, in the presence of Zn 2ϩ ions without HK, FXI interacts with approximately half the number of sites on normal platelets (800 sites/platelet), with a K d app ϳ 12 nM (9). 2 Although the mechanism and physiological relevance of FXI binding to activated platelets in the absence of cofactor molecules is not entirely clear, a possible mechanism to account for this binding in the absence of added HK is the release of HK from platelet ␣-granules after platelet activation (33). Furthermore, the binding of FXI in the absence of added HK to ϳ40% of the number of platelet sites generated in the presence of added HK correlates well with the observation that both initial rates and total yield of FXIa generated by thrombin in the absence of added HK are ϳ40% of those obtained in the presence of added HK or prothrombin (8,34).
The interaction of FXI with activated platelets obtained from two patients with the hereditary giant platelet (Bernard-Soulier) syndrome was deficient in the presence of HK, Zn 2ϩ , and Ca 2ϩ , with significantly decreased stoichiometry and normal affinity, and similar defects were observed when HK was excluded from the incubation mixture ( Fig. 1 and Table I). The two subjects of these studies (T. H. and A. H.) are first cousins (both homozygotes) whose platelets have been studied by various laboratories (13,21,25,(35)(36)(37)(38). Li et al. (25) show decreased levels of GP Ib-IX with normal amounts of GP V. A point mutation was found in codon 129 of the GP Ib␣ gene that results in the substitution of proline for leucine in the first position of the fifth leucine-rich repeat of the mature gene product and does not affect transcription of the GP Ib-IX genes but does inhibit surface expression of the receptor. Platelets from both patients exhibited ϳ40% of normal von Willebrand binding and ϳ40% of normal GP Ib-IX surface antigens. This suggests (25) that the mutation (Leu 129 3 Pro) affects the conformation of the GP Ib-IX complex, altering its availability on platelets. These findings are particularly intriguing in the context of our results (Table I), which demonstrate that the number of FXI binding sites on the same Bernard-Soulier platelets is reduced to 0 -70% (mean 43%) of normal. These data strongly suggest that the GP Ib-IX-V complex is the receptor localizing FXI to the surface of activated platelets. An alternative explanation that could account for the deficiency of FXI binding sites on Bernard-Soulier platelets is a defect in platelet activation with a failure to expose a FXI-binding site. However, in our studies the platelets were activated with the PAR-1 peptide, which unlike thrombin, activates Bernard-Soulier platelets normally (39).
To confirm the hypothesis that FXI binds to GP Ib on the platelet surface, we employed two GP Ib ligands in competition studies and demonstrated that both the monoclonal antibody SZ-2 and bovine von Willebrand factor, both of which bind to GP Ib␣, potently and completely inhibited FXI binding to activated platelets (Figs. 2, A and B). To obtain direct evidence of a specific interaction between GP Ib␣ and FXI, we utilized surface plasmon resonance to demonstrate that FXI binds to glycocalicin with high affinity (K d app ϳ52 nM) in the presence of Zn 2ϩ ions at the same concentration as that required for FXI binding to platelets. Interestingly, the requirements for FXI binding to activated platelets and to glycocalicin are somewhat different since optimal binding to platelets requires the presence of HK and Zn 2ϩ ions or prothrombin and Ca 2ϩ ions, whereas the direct binding of FXI to glycocalicin requires only Zn 2ϩ ions. It is possible that the low level (ϳ40%) of FXI binding to platelets in the absence of added HK (Fig. 1)  reflected by the FXI binding to glycocalicin in the surface plasmon resonance studies. Alternatively, it is possible that FXI, subjected to shear stress during the flow conditions employed for surface plasmon resonance, undergoes a conformational alteration that promotes its capacity to bind to glycocalicin. Thus, our previous studies demonstrate that HK (in the presence of Zn 2ϩ ions) can bind to the A1 domain of FXI and promote FXI binding to platelets via amino acids exposed on the surface of the A3 domain (10, 12, 40 -42). The putative allosteric alteration arising from ligation of the A1 domain by HK could be similar to that which may occur when FXI is exposed to shear stress during the surface plasmon resonance procedure.
Because both FXI and thrombin bind to GP Ib␣, we asked whether glycocalicin can promote FXI activation by thrombin.
In the presence of Zn 2ϩ , rates of FXI activation by thrombin were significantly increased by glycocalicin (Fig. 4A), and HK further increased these rates to those observed in the presence of activated platelets (Fig. 4B, Table II). The notion that GP Ib on the platelet surface binds FXI and promotes its activation by thrombin is confirmed by the observation that Bernard-Soulier platelets were defective in promoting the activation of FXI by thrombin when compared with normal platelets (Fig. 5 and Table II). These experiments clearly define the role of GP Ib on the platelet surface for 1) binding FXI in the presence of Zn 2ϩ ions and 2) promoting the activation of FXI by colocalizing FXI and thrombin on the activated platelet surface.
Two possibly related questions remain to be addressed, one of which relates to the stoichiometry of FXI binding to the GP Ib-IX-V complex and the other of which relates to the effects of platelet activation on GP Ib-IX-V complex exposure in platelets and on FXI binding to platelets. First, if the GP Ib-IX-V complex comprises the platelet receptor for FXI, how do we account for the fact that there are ϳ25,000 copies of GP Ib-IX-V per platelet but only ϳ1,500 FXI binding sites per platelet? Second, because it has been reported that platelet activation results in a ϳ65% decrease of GP Ib from the platelet surface (43) and a redistribution to the surface canalicular system (44), how do we account for the fact that platelet activation is required for the binding of FXI to the platelet GP Ib-IX-V complex and its activation by thrombin? We do not have definitive answers to these important questions, both of which are currently being investigated in our laboratory.   Table 1 have been retracted by the authors for the following reasons.
All of the authors with the exception of F. A. Baglia retract the specific data listed above because recent experiments conducted by Dipali Sinha, Sergei Shikov, Wenman Wu, and Syed Ahmad in the laboratory of Peter N. Walsh failed to confirm the conclusion that activated platelets promote the activation of factor XI by thrombin. All of the other results reported in the paper are valid. A detailed explanation of the chronology of events leading to this retraction and the retraction of a paper from Biochemistry (Baglia, F. A., and Walsh, P.N. (1998) Prothrombin is a cofactor for the binding of factor XI to the platelet surface and for platelet-mediated factor XI activation by thrombin. Biochemistry 37, 2271-2281) has been published in the journal Biochemistry (manuscript bi-2007-01501k, accepted July 27, 2007). We apologize to the readers, reviewers, and editors of the Journal of Biological Chemistry for publishing these erroneous data.

VOLUME 275 (2000) PAGES 20514 -20519
Thrombin-mediated feedback activation of factor XI on the activated platelet surface is preferred over contact activation by factor XIIa or factor Xia.

PAGES 20514 -20519:
The authors are retracting the entire article for the following reasons. All of the authors with the exception of F. A. Baglia retract the paper listed above because recent experiments conducted by Dipali Sinha, Sergei Shikov, Wenman Wu, and Syed Ahmad in the laboratory of Peter N. Walsh failed to confirm the conclusion that activated platelets promote the activation of factor XI by thrombin. A detailed explanation of the chronology of events leading to this retraction and to the retraction of a paper from Biochemistry (Baglia, F. A., and Walsh, P. N. (1998) Prothrombin is a cofactor for the binding of factor XI to the platelet surface and for platelet-mediated factor XI activation by thrombin. Biochemistry 37, 2271-2281) has been published in the journal Biochemistry (manuscript bi-2007-01501k, accepted July 27, 2007). We apologize to the readers, reviewers, and editors of the Journal of Biological Chemistry for publishing these erroneous data.

VOLUME 278 (2003) PAGES 21744 -21750
The glycoprotein Ib-IX-V complex mediates localization of factor XI to lipid rafts on the platelet membrane. All of the authors with the exception of F. A. Baglia retract the specific data listed above because recent experiments conducted by Dipali Sinha, Sergei Shikov, Wenman Wu, and Syed Ahmad in the laboratory of Peter N. Walsh failed to confirm the conclusion that activated platelets promote the activation of factor XI by thrombin. All other results reported in this paper are valid. A detailed explanation of the chronology of events leading to the retraction and the retraction of a paper from Biochemistry (Baglia, F. A., and Walsh, P. N. (1998) Prothrombin is a cofactor for the binding of factor XI to the platelet surface and for platelet-mediated factor XI activation by thrombin. Biochemistry 37, 2271-2281) has been published in the journal Biochemistry (manuscript bi-2007-01501k, accepted July 27, 2007). We apologize to the readers, reviewers, and editors of the Journal of Biological Chemistry for publishing these erroneous data.

VOLUME 278 (2003) PAGES 48112-48119
Thrombin activation of factor XI on activated platelets requires the interaction of factor XI and platelet glycoprotein Ib␣ with thrombin anion-binding exosites I and II, respectively.  Table 1 has been retracted by the authors for the following reasons. All of the authors with the exception of F. A. Baglia retract the specific data listed above because recent experiments conducted by Dipali Sinha, Sergei Shikov, Wenman Wu, and Syed Ahmad in the laboratory of Peter N. Walsh failrd to confirm the conclusion that activated platelets promote the activation of factor XI by thrombin. All of the other results reported in this paper are valid. A detailed explanation of the chronology of events leading to this retraction and the retraction of a paper from Biochemistry (Baglia, F. A., and Walsh, P. N. (1998) Prothrombin is a cofactor for the binding of factor XI to the platelet surface and for platelet-mediated factor XI activation by thrombin. Biochemistry 37, 2271-2281) has been published in the journal Biochemistry (manuscript bi-2007-01501k, accepted July 27, 2007). We apologize to the readers, reviewers, and editors of the Journal of Biological Chemistry for publishing these erroneous data.