Redox control of β2-glycoprotein I–von Willebrand factor interaction by thioredoxin-1

Background:β2-Glycoprotein I (β2GPI) is an abundant plasma protein that is closely linked to blood clotting, as it interacts with various protein and cellular components of the coagulation system. However, the role of β2GPI in thrombus formation is unknown. We have recently shown that β2GPI is susceptible to reduction by the thiol oxidoreductases thioredoxin-1 and protein disulfide isomerase, and that reduction of β2GPI can take place on the platelet surface. Methods:β2GPI, reduced by thioredoxin-1, was labeled with the selective sulfhydryl probe Na-(3-maleimidylpropionyl)biocytin and subjected to mass spectrometry to identify the specific cysteines involved in the thiol exchange reaction. Binding assays were used to examine the affinity of reduced β2GPI for von Willebrand factor (VWF) and the effect of reduced β2GPI on glycoprotein (GP)Ibα binding to VWF. Platelet adhesion to ristocetin-activated VWF was studied in the presence of reduced β2GPI. Results: We demonstrate that the Cys288–Cys326 disulfide in domain V of β2GPI is the predominant disulfide reduced by thioredoxin-1. Reduced β2GPI in vitro displays increased binding to VWF that is dependent on disulfide bond formation. β2GPI reduced by thioredoxin-1, in comparison with non-reduced β2GPI, leads to increased binding of GPIbα to VWF and increased platelet adhesion to activated VWF. Conclusions: Given the importance of thiol oxidoreductases in thrombus formation, we provide preliminary evidence that the thiol-dependent interaction of β2GPI with VWF may contribute to the redox regulation of platelet adhesion.

Summary. Background: b 2 -Glycoprotein I (b 2 GPI) is an abundant plasma protein that is closely linked to blood clotting, as it interacts with various protein and cellular components of the coagulation system. However, the role of b 2 GPI in thrombus formation is unknown. We have recently shown that b 2 GPI is susceptible to reduction by the thiol oxidoreductases thioredoxin-1 and protein disulfide isomerase, and that reduction of b 2 GPI can take place on the platelet surface. Methods: b 2 GPI, reduced by thioredoxin-1, was labeled with the selective sulfhydryl probe N a -(3-maleimidylpropionyl)biocytin and subjected to mass spectrometry to identify the specific cysteines involved in the thiol exchange reaction. Binding assays were used to examine the affinity of reduced b 2 GPI for von Willebrand factor (VWF) and the effect of reduced b2GPI on glycoprotein (GP)Iba binding to VWF. Platelet adhesion to ristocetin-activated VWF was studied in the presence of reduced b 2 GPI. Results: We demonstrate that the Cys288-Cys326 disulfide in domain V of b 2 GPI is the predominant disulfide reduced by thioredoxin-1. Reduced b 2 GPI in vitro displays increased binding to VWF that is dependent on disulfide bond formation. b 2 GPI reduced by thioredoxin-1, in comparison with non-reduced b 2 GPI, leads to increased binding of GPIba to VWF and increased platelet adhesion to activated VWF.
Introduction b 2 -Glycoprotein I (b 2 GPI) is a circulating plasma protein that consists of five repeating amino acid domains [1,2]. b 2 GPI may play a role in platelet adhesion as, in vitro, it interacts with von Willebrand factor (VWF) [3] and the platelet receptors glycoprotein (GP)Iba and ApoER2 [4,5]. b 2 GPI contains 11 disulfide bonds and no unpaired cysteine. The first four domains include four cysteines each, with disulfide bridges joining the first to the third and the second to the fourth cysteine. The fifth domain has an extra 20 amino acid tail, with an unusual termination in a cysteine, that forms a loop-back disulfide link with an extra cysteine found midway between the standard second and third cysteine positions [1].
We have recently shown that b 2 GPI can be reduced by thioredoxin-1 (TRX-1) and protein disulfide isomerase (PDI) in vitro [6,7]. Thiol oxidoreductases are becoming increasingly recognized as important mediators of platelet function [8,9]. The prototype member, PDI, is involved in regulation of activation of the fibrinogen receptor a IIb b 3 [10] and tissue factor [11]. Several novel members of the thiol isomerase family have been recently shown to translocate to the platelet surface following platelet activation [12]. TRX-1 is another member of the superfamily, and has been shown to protect endothelial cells from oxidative stress [13,14]. Extracellular TRX-1 mediates lymphocyte effector function [15] and may regulate platelet adhesion [16]. We have shown that reduction of b 2 GPI is achieved on the platelet surface and endothelial cells [6,7]. In the current study, we show that b 2 GPI is reduced by TRX-1 at Cys288-Cys326 in domain V of b 2 GPI, and that this reduction results in a significantly increased affinity of b 2 GPI for VWF.

Methods
Structural analysis of b 2 GPI The structural features of all disulfide bonds in two structures of b 2 GPI (Protein Data Bank ID: 1C1Z and 1QUB) were determined with the disulfide bond analysis tool available at http://www.cancerresearch.unsw.edu. au/CRCWeb.nsf/page/Disulfide+Bond+Analysis.
The secondary structures, in which the cysteines reside, and their solvent accessibility values are from DSSP (http://swift. cmbi.ru.nl/gv/dssp/). The dihedral strain energy of the disulfides was estimated from the magnitude of the five v angles that constitute the bond [19]. The analysis of the disulfide bonds in b 2 GPI is shown in Table 1.
Mass spectrometry To localize the specific cysteine(s) of b 2 GPI that are reduced by TRX-1, nb 2 GPI was treated with a TRX-1/TRX-R/NADPH mixture and labeled with the selective sulfhydryl reagent MPB, on the basis of a previously described method [20]. Specifically, TRX-1 (5 lM) was reduced by incubation for 1 h at 37°C with TRX-R (10 nM) and NADPH (200 lM) in a total volume of 300 lL. nb 2 GPI or rb 2 GPI was added at a concentration of 0.2 lM to TRX-1/ TRX-R/NADPH and incubated for 1 h at 37°C. To label free thiols, MPB at a concentration of 100 lM was added to b 2 GPI/ TRX-1/TRX-R/NADPH and incubated for 10 min at 37°C. The reaction was quenched by the addition of GSH at a concentration of 200 lM for 10 min at 37°C. All reactions were performed in 20 mM HEPES buffer containing 0.14 M NaCl (pH 7.4) (HBS).
Samples were separated by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) (4-12%) under nonreducing conditions, and stained with Coomassie Blue stain. The position of reduced (by TRX-1/TRX-R/NADPH) b 2 GPI ± MPB on Bis-Tris gel after separation by SDS-PAGE has been previously determined [6]. The corresponding bands of b 2 GPI, reduced b 2 GPI and reduced b 2 GPI + MPB were excised from the gel, and b 2 GPI peptides were prepared and analyzed by mass spectrometry and MASCOT and X!TAN-DEM searches (the preparation of b 2 GPI peptides and analysis of biotinylation are provided in detail in Data S1).
Assays for binding of b 2 GPI to immobilized VWF Binding of b 2 GPI reduced by TRX-1/TRX-R/NADPH to immobilized VWF Ninety-six well microtiter plates were coated with 100 lL of human VWF (10 lg mL )1 ) in 0.1 M NaHCO 3 (pH 8.3), as previously described [21]. Wells were washed with HBS, and non-specific binding sites were blocked by adding 200 lL of 2% BSA in HBS for 2 h at room temperature; this was followed by washing with HBS. Eight micromolar TRX-1 was incubated with 35 nM TRX-R and 180 lM NADPH for 1 h at 37°C. Reduced nb 2 GPI was prepared by incubating 187 lL of nb 2 GPI (1.8 lM) or HBS with 38 lL of TRX-1/TRX-R/NADPH for 1 h at 37°C. Reactions were diluted 1 : 1 in HBS, and VWF-coated wells were incubated with 100-lL reaction mixtures for 1 h at room temperature. Wells were washed four times with HBS containing 1 M NaCl, and 100 lL of anti-b 2 GPI mAb (20 lg mL )1 ) was added and incubated for 1 h at room temperature. After washing with HBS, 100 lL of a 1 : 1000 dilution of goat anti-mouse alkaline phosphatase (AP)conjugated antibody was added and incubated for 1 h at room temperature. Wells were washed with HBS, and 100 lL of pNPP (1 mg mL )1 ) in 1 M diethanolamine buffer and 0.5 mM MgCl 2 (pH 9.8) was added to each well. In some experiments, free thiols in b 2 GPI/TRX-1/TRX-R/NADPH were blocked by adding MPB (100 lM) for 10 min at 37°C before addion of the mixture to VWF-coated plates. In other experiments, TRX-R was inhibited, as previously described [22], by incubating 3.5 lM DNCB with TRX-R/NADPH before addition to TRX-1 (the Reduced b 2 GPI and von Willebrand factor 1755 final concentration of DNCB was 0.25 lM per well). The dose response of binding of (reduced and non-reduced) nb 2 GPI to immobilized VWF (5 lg mL )1 ) was measured by adding b 2 GPI at concentrations between 0.01 and 4 lM.
To assess the effect of TRX-1/TRX-R/NADPH on the capacity of immobilized VWF to bind to b 2 GPI, VWF-coated wells were treated with or without 100 lL of TRX-1/TRX-R/ NADPH at concentrations equimolar to that used for reduction of b 2 GPI. Subsequently, b 2 GPI was reduced with TRX-1/TRX-R/NADPH, half of which was subsequently incubated with DNCB (as described above), to inactivate residual activity of TRX-R before it was added to VWF.
Binding of recombinant GPIba to immobilized VWF in the presence of reduced nb 2 GPI vs. non-reduced b 2 GPI nb 2 GPI or equimolar BSA was treated with TRX-1/TRX-R/NADPH as above. Fifty-microliter volumes of the reaction mixtures were added to 50 lL of recombinant GPIba (10 lg mL )1 ) and incubated for 1 h at 37°C. One-hundred-microliter volumes of the reaction mixture were applied to VWF-coated wells. The amount of GPIba bound was determined using 100 lL of 1 lg mL )1 anti-CD42b mAb and secondary anti-mouse AP-conjugated antibody (1 : 1000).
Assays for binding of VWF to immobilized b 2 GPI Binding of VWF in solution, in the presence or absence of ristocetin, to immobilized reduced vs. non-reduced b 2 GPI Non-reduced and reduced rb 2 GPI (reduced by TRX-1/TRX-R/NADPH) were coated on ELISA plates at a concentration of 10 lg mL )1 for the b 2 GPI component, according to Hulstein et al. [3]. One hundred microliters of VWF (10 lg mL )1 ) alone or after preincubation with 1 mg mL )1 ristocetin for 5 min at room temperature was added to the wells and incubated for 1 h at room temperature. For some experiments, 3.5 lM DNCB was incubated with VWF or VWF/ristocetin before addition to immobilized b 2 GPI.
To study the effect of TRX-1/TRX-R/NADPH on VWF in solution, VWF (10 lg mL )1 ) was incubated with or without TRX-1/TRX-R/NADPH as above; half of the VWF mixture was subsequently activated with ristocetin. These VWF mixtures were then added to wells coated with b 2 GPI. The amount of bound VWF was assessed with the use of 100 lL of anti-VWF mAb (5 lg mL )1 ) and secondary anti-mouse AP-conjugated antibody (1 : 1000).
All enzyme-linked immunosorbent assay (ELISA) incubations were performed under argon, as exposure of reduced b 2 GPI to air led to its gradual reoxidation [7].
The optical density (OD) for all binding assays was read at 405 nm with a Microplate Scanning Spectrophotometer (Bio-Tek Instruments, Winooski, VT, USA).
Detection of TRX-1 and TRX-R in platelet lysates Washed platelets from healthy donors were prepared as previously described [20,23]. A platelet suspension, 4 · 10 11 L )1 in 20 mM HEPES, 137 mM NaCl, 4 mM KCl, 0.5 mM Na 2 HPO 4 and 0.1 mM CaCl 2 (pH 7.4) was activated with thrombin (100 nM) Table 1 Properties of the b 2 -glycoprotein I (b 2 GPI) disulfide bonds in two X-ray structures of the protein [1,2]; analysis of the properties of the disulfide bonds in b 2 GPI was performed as previously described [36,37] PDB ID   Platelet adhesion to reduced b 2 GPI coated on microtiter wells in the presence of VWF The method is based on that reported by Lahav et al. [8]. Washed platelets in Jamieson buffer were pelleted and resuspended in platelet adhesion buffer (Trisbuffered saline containing 0.05 M Tris-HCl, 140 mM NaCl, 2 mM Mg 2+ and 0.1% BSA, pH 7.4) at 12 · 10 11 L )1 . Ninetysix-well plates were coated with non-reduced or reduced rb 2 GPI (reduced by TRX-1/TRX-R/NADPH), washed three times with HBS, and incubated with VWF in the presence of ristocetin (1 mg mL )1 ), as described above. Platelets in adhesion buffer with or without TRX-1/TRX-R/NADPH or DNCB were aliquoted into the corresponding (± TRX-1/TRX-R/ NADPH) wells and incubated at room temperature for 1 h. After washing with phosphate-buffered saline (five times), platelets that had adhered to the wells were visualized with a Leica DM IRB (Leica, Wetzlar, Germany) microscope. Images were obtained with a Leica DC200 (Leica) camera (· 10 magnification). The number of adherent platelets was quantified by adding lysis buffer [0.07 M trisodium citrate, 0.3 M citric acid, 5 mM pNPP and 0.1% (v/v) Triton X-100, pH 5.4] for 1 h. The reaction was terminated by addition of 2 M NaOH, and absorbance was read at 405 nm.

Statistical analysis
The SPSS 17.0 statistics software (SPSS Inc., Chicago, IL, USA) was used for the analysis of data. Data were analyzed by ANOVA followed by DunnettÕs correction for multiple comparisons. A P-value of < 0.05 was considered to be statistically significant.

Results
Identification of a free thiol in Cys326 of b 2 GPI after reaction with TRX-1 Mass spectrometry showed biotinylation of nb 2 GPI treated with TRX-1/TRX-R/NADPH/MPB (see Tables S1 and S2 for a full list of peptides for each sample). As a number of cysteines were found to be biotinylated in b 2 GPI, the MBP/iodoacetamide-labeled peptide ratio was used to determine the cysteine target of reduced TRX-1. As shown in Table S3, Cys326 is by far the most heavily modified cysteine in the protein.
Structural analysis of disulfide bond Cys288-Cys326 of the b 2 GPI molecule Nine of the 11 disulfides in b2GPI have a spiral configuration (RHS, right-handed spiral or LHS, left-handed spiral), the most common configuration and one associated with structural disulfides ( Table 1). The Cys288-Cys326 disulfide has a )/+ right-handed hook ()/+RHHook) configuration in both crystal structures of the protein [1,2]. None of the bonds are particularly strained based on the measure of the dihedral strain energy. Notably, Cys326 is particularly exposed to solvent. This exposure is likely a factor in the reduction of the Cys288-Cys326 disulfide bond by protein oxidoreductases.
Redox control of the interaction of b 2 GPI with VWF Assays for binding of b 2 GPI to immobilized VWF. b 2 GPI reduced by TRX-1 shows increased binding to immobilized VWF. Given the importance of thiol linkage in VWF multimerization [24] and its previously described ability to bind b 2 GPI [3], we proceeded to examine whether free thiols generated in b 2 GPI are involved in its interaction with VWF. We applied b 2 GPI reduced by TRX-1/TRX-R/NADPH to VWF-coated ELISA plates, and detected the amount of b 2 GPI bound to VWF with the mAb 4B2E7. The binding of b 2 GPI treated with TRX-1/TRX-R/NADPH to immobilized VWF was increased 3.5-fold when compared with untreated b 2 GPI (Fig. 1A). Non-reduced b 2 GPI displayed low binding to VWF, regardless of the concentrations used, whereas binding of reduced b 2 GPI to VWF showed a Ôbell-shapedÕ curve with maximal binding at 0.8 lM (Fig. 1B) and decreased binding at higher concentrations. This can be attributed to the Ôhigh-dose hook effectÕ described for high concentration of some ligands in ELISA systems [25]. For a monotonic increase in signal with concentration, there must be an excess of antibodies, both capture and enzyme-conjugated, relative to the analyte being detected. As the concentration of analyte (in this case, b 2 GPI) begins to exceed the amount of antibody, the dose-response curve will plateau and, with further increase, may, paradoxically, become negatively sloped. In some cases, this phenomenon has also been attributed to the coating density of the capture antibody (the capture protein in this case being VWF). Thiol dependence of reduced b 2 GPI binding to VWF. The increased binding of reduced b 2 GPI to immobilized VWF was abrogated when the thiol-reactive molecule MPB was added to reduced b 2 GPI (reduced by TRX-1/TRX-R/ NADPH) before addition to VWF-coated wells. This indicated Reduced b 2 GPI and von Willebrand factor 1757 that the binding of reduced b 2 GPI to VWF was dependent on disulfide bond formation between the two molecules, which was prevented in the presence of MPB (Fig. 1A).
Inhibition of TRX-R activity by DNCB (before the inclusion of TRX-R in the TRX-1/TRX-R/NADPH mixture used to reduce b 2 GPI) decreased the binding of b 2 GPI treated with TRX-1/TRX-R/NADPH to immobilized VWF, and the binding was comparable to that of untreated b 2 GPI (Fig. 1C).
Reduction of VWF by TRX-1 is not necessary for interaction with b 2 GPI. There is evidence from the literature that exposure of VWF to TRX-1/DTT reduces disulfide bonds in the molecule, decreasing the binding affinity of collagen [21]. To exclude the effect of TRX-1/TRX-R/ . Plates were coated with VWF, and b 2 GPI mixtures were applied to the wells and incubated for 1 h at room temperature. The amount of bound b 2 GPI was assessed with an anti-b 2 GPI monoclonal antibody (mAb). b 2 GPI reduced by thioredoxin-1 (TRX-1)/thioredoxin reductase (TRX-R)/NADPH showed a significantly higher level of binding to VWF than non-reduced b 2 GPI. The increased binding of reduced b 2 GPI to VWF was inhibited when free thiols introduced into b 2 GPI by TRX-1/TRX-R/NADPH were previously blocked with MPB. (B) Dose-response curves of nb 2 GPI (reduced and non-reduced) binding to immobilized VWF. The amount of bound b 2 GPI was assessed with an anti-b 2 GPI mAb. Maximal binding for b 2 GPI reduced by TRX-1/TRX-R/NADPH is at 0.8 lM (n = 2). (C) Inhibition of reduced nb 2 GPI binding to VWF by the TRX-R inhibitor 1-chloro-2,4-dinitrobenzene (DNCB). Plates were coated with VWF and blocked. TRX-R activity was inhibited by DNCB before incubation with TRX-1. b 2 GPI was treated with the inactive TRX-1 preparations and applied to the wells. The amount of b 2 GPI bound to VWF was significantly reduced when TRX-R had been inactivated with DNCB. (D) The increased binding of reduced nb 2 GPI to VWF is mediated through the effect of the TRX-1 reducing mixture on b 2 GPI and not on VWF. Plates were coated with VWF and blocked. VWFcoated wells were then treated with TRX-1/TRX-R/NADPH or buffer alone. Separately, nb 2 GPI was incubated with TRX-1/TRX-R/NADPH, and half of the mixture was incubated with DNCB to inactivate residual TRX-R activity before addition of the b 2 GPI preparations to the wells. Binding of b 2 GPI was determined with an anti-b 2 GPI mAb. The binding of reduced b 2 GPI to VWF was not affected by pretreatment of the coated VWF with TRX-1/TRX-R/NADPH before addition of the b 2 GPI mixtures. Also, inhibition of the residual TRX-R activity contained in the b 2 GPI/TRX-1/TRX-R/NADPH mixture before addition to coated VWF did not alter the binding of already reduced b 2 GPI to VWF (whether or not VWF was exposed to active TRX-R or DNCB-inactivated TRX-R). (E) Binding of recombinant GPIba to immobilized VWF in the presence of reduced nb 2 GPI. Plates were coated with VWF and blocked. b 2 GPI with or without TRX-1/TRX-R/NADPH treatment and bovine serum albumin treated with TRX-1/TRX-R/NADPH as control were incubated with GPIba. The reaction mixtures were then applied to the wells and incubated. The amount of GPIba bound was determined with a specific anti-GPIba mAb, and was increased in the presence of reduced b 2 GPI. OD, optical density; b 2 , b 2 GPI; M, MPB; TRN, TRX-1/TRX-R/NADPH; B, bovine serum albumin; NS, not significant. For all enzyme-linked immunosorbent assays, values are the mean ± standard error of the mean (n = 3). ***P < 0.001. NADPH on vWF in our experiments, we used two different approaches. First, we treated VWF coated on the wells with TRX-1/TRX-R/NADPH or buffer alone, and compared binding affinities for b 2 GPI. Treatment of VWF with the reducing mixture before addition of b 2 GPI did not affect b 2 GPI binding: non-reduced b 2 GPI showed low binding to VWF, whether or not VWF had been pretreated with TRX-1/ TRX-R/NADPH. Also, reduced b 2 GPI showed a high level of binding to VWF, whether or not VWF had been pretreated with TRX-1/TRX-R/NADPH (Fig. 1D).
In the second approach, we utilized the TRX-R inhibitor DNCB to inactivate the residual reducing activity of TRX-1/ TRX-R/NADPH after it had already reduced b 2 GPI. When DNCB had been incubated with TRX-1/TRX-R/NADPH before it was used to reduce b 2 GPI, b 2 GPI could not be reduced, and therefore the mixture of b 2 GPI and inactive TRX-1/TRX-R/NADPH showed the same low level of binding to VWF as b 2 GPI alone (Fig. 1C). However, when DNCB was added to TRX-1/TRX-R/NADPH after it had already reduced b 2 GPI, b 2 GPI/inactivated TRX-1/TRX-R/NADPH showed the same high level of binding as b 2 GPI/active TRX-1/TRX-R/ NADPH, thus showing that the critical event in VWF-b 2 GPI binding is the reduction of b 2 GPI (Fig. 1D). b 2 GPI reduced by TRX-1 promotes binding of GPIba to immobilized VWF. The presence of reduced b 2 GPI (reduced by TRX-1/TRX-R/NADPH) increased the binding of GPIba to immobilized VWF in comparison with untreated b 2 GPI or BSA treated with TRX-1/TRX-R/NADPH as control protein (Fig. 1E).
Assays for binding of VWF to immobilized b 2 GPI. Dependence of VWF binding to reduced b 2 GPI on ristocetin activation of VWF. Ristocetin-activated VWF bound more than non-activated VWF to coated, non-reduced b 2 GPI. Furthermore, there was a significant increase in the binding of ristocetin-activated VWF to coated, reduced b 2 GPI as compared with non-activated VWF ( Fig. 2A). This shows that the affinity of reduced b 2 GPI for VWF is high when VWF is activated either by coating on microtiter wells (Fig. 1A) or being treated in solution with ristocetin ( Fig. 2A). Incubating TRX-1/TRX-R/NADPH with VWF in solution did not affect its capacity to bind to b 2 GPI (Fig. 2A).
The binding of VWF (activated or non-activated by ristocetin) to immobilized (reduced or non-reduced) b 2 GPI was not affected by the presence or not of DNCB in the VWF solution (Fig. 2B).
Potential role of reduced b 2 GPI and TRX-1 in platelet adhesion TRX-1 and TRX-R are detected by western blot in platelet lysates As VWF tethers platelets, we were interested to determine whether the source of b 2 GPIÕs reducing agent, TRX-1, can be found in platelets. Both TRX-1 and TRX-R were detected in platelet lysates of resting and thrombin-activated platelets. TRX-1 was also detected in the releasates from resting and thrombin-activated platelets (Fig. 3A). b 2 GPI reduced by TRX-1 increases platelet adhesion to activated VWF The VWF-platelet GPIba receptor interaction is important for the initial step of platelet adhesion. Ristocetin promotes VWF binding to GPIba in Non-reduced and reduced rb 2 GPI [reduced by thioredoxin-1 (TRX-1)/ thioredoxin reductase (TRX-R)/NADPH] were coated on enzyme-linked immunosorbent assay (ELISA) plates (under argon) at a concentration of 10 lg mL )1 for the b 2 GPI component. After washing, VWF (10 lg mL )1 ) was added alone or after incubation with TRX-1/TRX-R/NADPH (at the same concentration and with the same method as used for the reduction of b 2 GPI). Half of the VWF solution was further incubated with ristocetin before addition to the wells. The amount of bound VWF was assessed with an anti-VWF monoclonal antibody. Ristocetin-activated VWF showed a marked increase in binding to reduced b 2 GPI as compared with nonactivated VWF.Treatment of VWF in solution with the reducing mixture TRX-1/TRX-R/NADPH did not affect its binding to b 2 GPI in the presence or absence of ristocetin. (B) Binding of VWF to coated b 2 GPI in the presence of 1-chloro-2,4-dinitrobenzene (DNCB). Reduced and non-reduced b 2 GPI (reduced by TRX-1/TRX-R/NADPH) was coated onto 96well plates under argon. Preparations of VWF or VWF activated by ristocetin were prepared, and half of these were incubated with DNCB. The VWF mixtures were added to the wells and incubated for 1 h at room temperature. Detection with an anti-VWF antibody showed that the presence of DNCB did not affect the binding of VWF to already reduced b 2 GPI on the plate. OD, optical density; b 2 , b 2 GPI; TRN, TRX-1/TRX-R/NADPH; risto, ristocetin; NS, not significant. For all ELISAs, values are the mean ± standard error of the mean (n = 3). ***P < 0.001.
solution. We demonstrated that ristocetin-activated VWF had a greater affinity for reduced b 2 GPI ( Fig. 2A), so we tested whether reduced b 2 GPI could support platelet adhesion. The method was based on one previously published, which uses an ELISA to measure acid phosphatase in adherent platelets [8].
In this system, reduced b 2 GPI caused a significant increase, by 41%, in platelet adhesion as compared with non-reduced b 2 GPI when VWF was activated with ristocetin (OD 0.68 ± 0.04 vs. 0.40 ± 0.02, respectively, mean ± standard deviation, n = 3 in triplicate, P < 0.0001). Platelet adhesion was partially inhibited by addition of the TRX-R inhibitor DNCB, supporting the notion that reduction of b 2 GPI by TRX-1 was partially responsible for the adhesion of the platelets (Fig. 3B). Representative images of the platelet adhesion assay are shown in Fig. 3C. Platelets displayed low adhesion to VWF not treated with ristocetin, with no differences being seen between coated b 2 GPI, HSA, TRX-1treated b 2 GPI or TRX-1-treated HSA (data not shown).

Discussion
In the current study, we have shown that b 2 GPI reduced by TRX-1 demonstrates thiol-dependent increased binding to VWF and platelet adhesion to activated VWF. TRX-1 is ubiquitously expressed and is secreted to the cell surface [26]. TRX-1 is also present in platelet lysates [27] and releasates. We have previously shown that b 2 GPI is partially reduced on the platelet surface by TRX-1/TRX-R/NADPH [6]. We now demonstrate that the predominant cysteine reduced by TRX-1 is Cys326 in domain V. This finding is in agreement with the majority of biological functions of b 2 GPI being attributed to domain V including phospholipid [1], thrombin [28] and GPIba binding [4,5], and FXIa cleavage [29]. The fifth domain is predicted to be anchored to the plasma membrane, providing the appropriate interface to react with cell surface proteins such as platelet oxidoreductases. (rb 2 GPI) was treated or not treated with TRX-1/TRX-R/NADPH and coated onto 96-well plates. After blocking, wells were incubated with VWF activated with ristocetin. Platelets in adhesion buffer with or without TRX-1/TRX-R/NADPH or 1-chloro-2,4-dinitrobenzene (DNCB) were aliquoted into the corresponding (± TRX-1/TRX-R/NADPH) wells and incubated at room temperature for 1 h. After extensive washing with phosphate-buffered saline, platelets that had adhered were quantified with lysis buffer for 1 h. The reaction was terminated by the addition of 2 M NaOH, and absorbance was read at 405 nm. OD, optical density; b 2 , b 2 GPI; TRN, TRX-1/TRX-R/NADPH; HSA, human serum albumin.
Values are the mean ± standard error of the mean (n = 3, in triplicate). ***P < 0.001. (C) Representative images of platelet adhesion obtained with a Leica DC200 camera at · 10 magnification. b 2 GPI reduced by TRX-1/TRX-R/NADPH significantly enhanced platelet adhesion to vWF activated by ristocetin, whereas non-reduced b 2 GPI or TRX-1/TRX-R/ NADPH-treated HSA did not. Platelet adhesion was partially inhibited when TRX-1/TRX-R/NADPH was treated with DNCB.
It is interesting that b 2 GPI has been shown to bind both with GPIba and VWF [3][4][5]. Hulstein et al. [3] showed that b 2 GPI bound to the A1 domain of VWF with low affinity and inhibited platelet adhesion to immobilized VWF. However, when free thiols are introduced into b 2 GPI by TRX-1, the binding affinity for VWF increases significantly and promotes adhesion of GPIba and platelets to activated VWF. The GPIba-VWF interaction is crucial for hemostasis. Disulfide exchange may be an important feature of platelet tethering to exposed VWF, as shear has been shown to promote disulfide formation between VWF subunits and VWF binding to platelets [24]. Furthermore, GPIba shows physical proximity to PDI on the platelet surface [20]. Platelets may also provide the surface for reduction of b 2 GPI, which could be a modifier of disulfide reactions of VWF and/or GPIba.
Multiple interactions of b 2 GPI have been shown with various components of the coagulation and fibrinolysis system in vitro, often with conflicting results and interpretations. Although b 2 GPI deficiency does not lead to gross hemostatic abnormalities, we believe that reduction of circulating b 2 GPI can promote thrombus formation under specific conditions. This idea is supported by the facts that b 2 GPI )/) mice have impaired thrombin generation [30] and that b 2 GPI inhibits thrombin inactivation by heparin cofactor II [31]. Our laboratory has shown that b 2 GPI is predominantly in the reduced form in vivo [7]. Hence, it is possible that in vivo reduced b 2 GPI supports platelet adhesion, a function not apparent in in vitro studies, where the purified protein has been significantly oxidized. The design of in vivo studies would delineate the role of reduced vs. non-reduced b 2 GPI in platelet adhesion.
The presence of the reduced form of b 2 GPI in the circulation may be relevant to disease development. There are a number of publications addressing the relationship of b 2 GPI levels to thrombotic and atherothrombotic disease, some in the context of antiphospholipid syndrome. In a review by Inanc et al. [32], the levels of circulating b 2 GPI was not associated with thrombotic risk. High levels of circulating b 2 GPI have been reported to decrease the risk of myocardial infarction in elderly men [33]. With regard to the risk of thrombosis, the amount of reduced b 2 GPI in the circulation may be more important than the total levels of circulating b 2 GPI.
The finding for the first time that b 2 GPI can be involved in thiol exchange reactions with TRX-1 is of considerable importance, given b 2 GPIÕs high concentration in plasma makes it easily available for reactions where thiols are needed. There is increased scientific interest in the role of thiol-acting enzymes such as PDI in thrombus formation. PDI has been shown in vivo to be required for thrombus formation [34] and tissue factor activation [35]. Inhibition of PDI action by an inhibitory antibody or the thiol blocker bacitracin abrogated thrombus formation. Although the substrate of PDI has not been definitely established, existing data support integrin a IIb b 3 or GPIba. The demonstration that b 2 GPI is a substrate of TRX-1 and PDI has implications for a role of b 2 GPI in thrombus formation. Our results also provide considerable insight into the participation of extracellular TRX-1 in the regulation of platelet adhesion.

Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Data S1. Reduction of b 2 GPI by TRX-1 for mass spectrometry. Table S1. Results from X!Tandem search for tryptic peptides produced from the three samples: nB2GPI, nB2GPI/TRX-1/ TRX-R/NADPH and nB2GPI/TRX-1/TRX-R/NADPH + MPB. Table S2. Results from Mascot search for tryptic peptides produced from the three samples: nB2GPI (2A), nB2GPI/ TRX-1/TRX-R/NADPH+MPB (2B) and nB2GPI/TRX-1/ TRX-R/NADPH (2C). Table S3. MPB/IA labeled abundance ratios of cysteine containing peptides in nb2GPI treated with TRX-1/TRX-R/ NADPH/MPB. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.