The Binding Site in β2-Glycoprotein I for ApoER2′ on Platelets Is Located in Domain V*

The antiphospholipid syndrome is caused by autoantibodies directed against β2-glycoprotein I (β2GPI). Dimerization of β2GPI results in an increased platelet deposition to collagen. We found that apolipoprotein E receptor 2′ (apoER2′), a member of the low density lipoprotein receptor family, is involved in activation of platelets by dimeric β2GPI. To identify which domain of dimeric β2GPI interacts with apoER2′, we have constructed domain deletion mutants of dimeric β2GPI, lacking domain I (ΔI), II (ΔII), or V (ΔV), and a mutant with a W316S substitution in the phospholipid (PL)-insertion loop of domain V. ΔI and ΔII prolonged the clotting time, as did full-length dimeric β2GPI; ΔV had no effect on the clotting time. Second, ΔI and ΔII bound to anionic PL, comparable with full-length dimeric β2GPI. ΔV and the W316S mutant bound with decreased affinity to anionic PL. Platelet adhesion to collagen increased significantly when full-length dimeric β2GPI, ΔI, or ΔII (mean increase 150%) were added to whole blood. No increase was found with plasma β2GPI, ΔV, or the W316S mutant. Immunoprecipitation indicated that full-length dimeric β2GPI, ΔI, ΔII, and the W316S mutant can interact with apoER2′ on platelets. ΔV did not associate with apoER2′. We conclude that domain V is involved in both binding β2GPI to anionic PL and in interaction with apoER2′ and subsequent activation of platelets. The binding site in β2GPI for interaction with apoER2′ does not overlap with the hydrophobic insertion loop in domain V.

anticoagulants and anticardiolipin antibodies) in the plasma of patients. These aPL are a heterogeneous group of antibodies directed to plasma proteins with affinity for anionic phospholipids (PL). We now know that the most important plasma protein, to which the aPL are directed, is ␤ 2 -glycoprotein I (␤ 2 GPI or apolipoprotein H) (5,6).
␤ 2 -Glycoprotein I is abundantly present in plasma (ϳ200 g/ml) and is mainly synthesized in the liver, although mRNA coding for ␤ 2 GPI has been found in a variety of cells such as trophoblasts, placental cells, endothelial cells, and neurons (7)(8)(9). The mature sequence of human ␤ 2 GPI consists of 326 (44 kDa) amino acids (aa) with four N-linked glycosylation sites. It is composed of five repeating units that belong to the complement control protein family. The first four domains have ϳ60 aa residues and 4 cysteines each, with potential disulfide bridges joining the first to third and the second to fourth cysteines to contribute to a "looped-back " structure, called Sushi domains. The fifth domain is aberrant, having 82 aa and three disulfide bridges. A positively charged (multiple lysine) region between Cys 281 -Cys 288 in domain V is highly conserved and a critical phospholipid-binding site (10 -13). The flexible loop Ser 311 -Lys 317 , containing Trp 316 , which is essential for phospholipid binding (14), is located in the middle of this charged region. Domain V has also been described to interact with anionic hydrophobic ligands (15). Domain I of ␤ 2 GPI harbors another cationic region. Involvement of this region in binding to PL has also been described (16). Apolipoprotein E receptor 2 (also known as apoER2 or LRP8) is a member of the low density lipoprotein (LDL) receptor family. It has been identified by Kim et al. in 1996 (17) and shares structural homology with the LDL and the very low density lipoprotein (VLDL) receptors (17). With respect to restricted tissue expression (brain, testis, and placenta) and structural homology, apoER2 is closer to the VLDL receptor (45-63% aa homology) than to the LDL receptor (18). The apoER2 cDNA encodes a cluster of eight complement-type repeat domains, not all of which are translated into the polypeptide chain. Because of alternative splicing, receptors with either four or five complement-type repeat domains are produced (19). Spatial and temporal differences in the expression pattern of these proteins suggest different physiological functions for individual receptor species. It seems that apoER2 has an alternative physiological function in vivo, as there is firm support that this protein is involved in signaling processes (20 -23).
Recently, a splice variant of apolipoprotein E receptor 2 (apoER2Ј or apoER2⌬5) was identified in platelets and megakaryocytic cell lines, as a member of the LDL receptor family (24). Platelet apoER2Ј mRNA encodes a 130-kDa protein including the LDL receptor class A repeats, epidermal growth factor homology repeats, O-linked sugar domain, a cytoplasmic domain that contains one internalization signal, and a single transmembrane region. In recent publications it has been shown that LDL and dimeric ␤ 2 GPI can interact with apoER2Ј on platelets (22,25). Until now, little is known about the interaction between (dimeric) ␤ 2 GPI and platelets. As for phospholipid binding, cationic patches might play an important role, because ligand binding to apoER2Ј is dependent on electrostatic interactions (26). ␤ 2 -Glycoprotein I contains two cationic regions, located in domain I (including the interface between domain I and II) and domain V. The largest cationic patch is in domain V. One may speculate that these domains play a role in binding of dimeric ␤ 2 GPI to apoER2Ј on platelets. To understand the mechanism of the interaction between apoER2Ј and dimeric ␤ 2 GPI it is essential to know: (i) which domain(s) and (ii) what structures in these domains are involved in interaction with apoER2Ј.

EXPERIMENTAL PROCEDURES
Construction of Dimeric Constructs of ␤ 2 GPI-The dimer apple 4-␤ 2 GPI and the apple 2-␤ 2 GPI, which is not able to form dimers, were constructed as described previously (27). To exclude the possibility that apple 4-␤ 2 GPI binds via the dimerization domain of factor XI (apple 4), dimer apple 4 was constructed. The sequence of dimeric apple 4 was amplified from the vector apple 4-tissue-type plasminogen activator-S478A with the primers apple 4-BglII (GCCAGATCTTTCTGCCAT-TCTTCA) and apple 4-XbaI (GGTCTAGACTCGAGTCCCTCCTT-TGATGCGTG). The PCR product was subcloned into the vector pCR®2.1-TOPO® (Invitrogen, Breda, The Netherlands), and cloned into the vector apple 4-tissue-type plasminogen activator-S478A with BglII and XbaI (underlined in apple 4-BglII and apple 4-XbaI, respectively). The starting point for the construction of the domain deletion mutants (DM) was the full-length cDNA of apple 4-C321S-␤ 2 GPI (in short apple 4-␤ 2 GPI) cloned into the vector apple 4-C321S-tissue-type plasminogen activator-S478A. The domain I deletion was constructed with the primers domain II ␤ 2 GPI-XhoI forward (CCCTCGAGATCCCAGA-GTATGTCCTTTTGCTG) and ␤ 2 GPI-XbaI reverse (GCTCTA-GAAAACAAGTGTGACATTTTATGTGGA). For the construction of the domain II deletion a set of two primers was used: for domain I amplification the primers ␤ 2 GPI-XhoI forward (CCCTCGAGGACG-GACCTGTCCCAAGCC) and domain I ␤ 2 GPI reverse (TGTACATT-TCAGAGTGTTGATG) and for domain III-V amplification the primers domain III ␤ 2 GPI forward (ACTCTGAAATGTACACC-CATCATCTGCCCTCCACCA) and ␤ 2 GPI-XbaI reverse. These two products served as a template in a second PCR to amplify the full-length domain I deletion using ␤ 2 GPI-XhoI forward and ␤ 2 GPI-XbaI reverse. The domain V deletion was constructed with the primers ␤ 2 GPI-XhoI forward and domain IV ␤ 2 GPI-XbaI reverse (TCTAGATCATTTA-CAACTTGGCATGGCAGACCA). To construct fusion proteins of apple 4 and the domain deletion mutants of ␤ 2 GPI, the PCR product was cloned with XhoI and XbaI into the vector apple 4-C321S-tissuetype plasminogen activator-S478A. In this way DM of apple 4-␤ 2 GPI were constructed. Sequence analyze were performed to confirm correct amplification of the cDNAs.
Construction of Soluble Human ApoER2Ј-Mature megakaryocytes were cultured from citrated umbilical cord blood as described by Den Dekker et al. (28). cDNA was synthesized from mRNA of mature megakaryocytes using Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Soluble human apoER2Ј (sh-apoER2Ј) was then cloned from this cDNA using Phusion DNA polymerase (Finnzymes). Primer design was such that the signal peptide was omitted and the stop codon was deleted, forward primer: sh-apoER2Ј BamHI, GGATCCGGGCCGGCCAAGGATTGCGAA-AAGG and reverse primer: sh-apoER2Ј NotI, GCGGCCGCCTTG-CAGTTCTTGGTCAGTAGGTCC. Sh-apoER2Ј was then cloned into PTT3-SR␣-GH-HISN-TEV. This expression vector is constructed from the pTT3 (29) and the pSGHV0 expression vectors (30).
Transfection, Expression, Cell Culture, and Purification of Dimeric Constructs of ␤ 2 GPI and Sh-ApoER2Ј-Transfection of baby hamster kidney cells with the calcium phosphate method was performed as described previously (31). Expression of all fusion constructs was performed in conditioned serum-free medium (Dulbecco's modified Eagle's medium/F-12 medium supplemented with 0.5% UltroserG; Invitrogen). Protein expression was measured using a ␤ 2 GPI-enzyme linked immunosorbent assay (ELISA). Domain deletion mutants of apple 4-␤ 2 GPI fusion proteins were purified from cell culture medium with a monoclonal antibody against ␤ 2 GPI (21B2) coupled to a CNBractivated Sepharose column (Amersham Biosciences). Bound DM was eluted with 0.1 M glycine (pH 2.7). The purified proteins were immediately neutralized with 1 M Tris (pH 9). The purified proteins were further subjected to purification on a Mono S column using fast protein liquid chromatography (Amersham Biosciences). Fusion proteins were eluted with a linear salt gradient from 50 mM to 1 M NaCl. After determination of the purity of the protein fractions on a SDS 4 -15% PAGE, fractions with DM of apple 4-␤ 2 GPI were pooled, concentrated with polyethylene glycol, and dialyzed against TBS (50 mM Tris, 150 mM NaCl, pH 7.4). Apple 4 was purified using monoclonal antibody XI-1 (generous gift of Dr. J. C. M. Meijers, Academic Medical Hospital, Amsterdam), which recognizes the apple 4 domain, coupled to CNBractivated Sepharose. Bound proteins were eluted with 0.1 M glycine (pH 2.7). The purified proteins were immediately neutralized with 1 M Tris (pH 9). For sh-apoER2Ј production, HEK293-EBNA cells were transfected by the DNA-polyethyleneimine method according to Durocher et al. (29). Sh-apoER2Ј production was done in a 1-liter suspension culture (in medium containing 90% freestyle, 10% calcium-free Dulbecco's modified Eagle's medium, 0.5% fetal calf serum, Invitrogen) for 4 days. Sh-apoER2Ј was affinity purified using receptor-associated protein-Sepharose from expression medium. Protein concentrations were determined using a bicinchoninic acid (BCA) protein assay (Pierce) according to the instructions of the manufacturer, and with bovine serum albumin as a standard. Purified constructs were analyzed by SDS-PAGE.
Purification of Plasma ␤ 2 GPI-Plasma ␤ 2 GPI was isolated from fresh citrated human plasma as described previously (32). In short, dialyzed human plasma was subsequently applied to the following columns: DEAE-Sephadex A50, protein G-Sepharose, S-Sepharose, and finally heparin-Sepharose (all Sepharoses were obtained from Amersham Biosciences). Bound proteins were eluted with a linear salt gradient. Afterward, ␤ 2 GPI was dialyzed against TBS. Purity of the protein was checked with SDS-PAGE analysis. Concentration of the protein was determined using the BCA protein assay.
Preparation of Phospholipid Vesicles-Phospholipid vesicles containing 20% phosphatidylserine (PS) and 80% phosphatidylcholine (PC) were prepared according to Brunner et al. (33), with some modifications as described by Van Wijnen et al. (34). The phospholipid concentration was determined by phosphate analysis (35).
Binding of Domain Deletion Mutants to Phospholipid Vesicles-Binding of DM of apple 4-␤ 2 GPI to PS/PC vesicles was tested in a solid phase binding assay. High binding 96-well ELISA plates (Costar, Corning Inc., 9102) were coated with 20% PS, 80% PC (25 M in TBS; 50 l/well) overnight at 4°C. Wells were blocked with TBS, 0.5% gelatin (150 l/well) for 2 h at 37°C. Subsequently, wells were incubated with different concentrations of DM (0.25-32 g/ml) for 1.5 h at 37°C, followed by incubation with monoclonal antibody 2B2 (3 g/ml; 50 l/well; 1.5 h at 37°C), a generous gift of Dr. J. Arnout, Leuven, Belgium. Apple 4-␤ 2 GPI was used as a positive control and plasma ␤ 2 GPI as a negative control. Afterward the wells were incubated with peroxidaseconjugated rabbit anti-mouse antibody (1:1000, 50 l/well, 1.5 h at 37°C), followed by a staining procedure using orthophenylenediamine. Samples were diluted in TBS, 0.5% gelatin. Nonspecific binding was determined using non-coated wells. Results are expressed as mean Ϯ S.D. (n ϭ 3).
Determination of the Effect of Domain Deletion Mutants on Clotting Time-Coagulation assays were performed in a KC-10 coagulometer (Amelung, Lemgo, Germany). To detect the effect of the DM on clotting time, the prothrombin time (PT) was performed as follows: 25 l of normal pooled plasma and 25 l of DM, apple 4-␤ 2 GPI, plasma ␤ 2 GPI (final concentration 100 g/ml), or buffer were incubated for 30 min at 4°C, followed by an incubation of 90 s at 37°C. Clotting was initiated by the addition of 50 l of Innovin (Dade Behring, Marburg, Germany).
Blood Collection for Perfusion Experiments-Freshly drawn venous blood was collected from healthy donors (with informed consent) into 1/10 of volume of 3.2% trisodium citrate (w/v). These donors denied taking aspirin or other platelet function inhibitors during the previous 10 days.
Perfusion Experiments-Perfusions were performed in a single-pass perfusion chamber under nonpulsatile flow conditions using a modified parallel plate perfusion with a slit width of 2 mm and a slit height of 0.1 mm. Experiments with collagen type III as a surface were performed with a perfusion time of 90 s, at a shear rate of 800 s Ϫ1 , which represents the flow rate in small arteries. Plasma-derived ␤ 2 GPI, apple 4, apple 2-␤ 2 GPI, apple 4-␤ 2 GPI, DM, or buffer were added to whole blood 5 min before the start of the perfusion at a concentration of 100 g/ml and incubated at 37°C. The prewarmed blood was drawn through the perfusion chamber by an infusion pump (pump 22, model 2400 -004; Harvard, Natrick, MA). Afterward, the coverslips were taken from the perfusion chamber and rinsed with Hepes-buffered saline (10 mM Hepes, 150 mM NaCl, pH 7.4), fixed in 0.5% glutaraldehyde in PBS, dehydrated with methanol, and stained with May-Grünwald/Giemsa as described previously. Platelet deposition was evaluated with a light microscope equipped with a JAI-CCD camera (Copenhagen, Denmark) coupled to a Matrox frame grabber (Matrox Electronic Systems Ltd., Quebec, Canada) using Optimas 6.2 software (Optimas Inc., Seattle, WA) for image analysis. Evaluation for platelet adhesion was performed on 20 fields, perpendicular to the flow direction. Platelet adhesion was expressed as the percentage of the surface covered with platelets relative to adhesion in the presence of buffer. Results are expressed as mean relative coverage (mean Ϯ S.D., n ϭ 3). Statistical analysis was performed with a Student's t test.
Immunoprecipitations-500-l aliquots of washed platelets (300,000/l) resuspended in Hepes/Tyrode buffer were incubated for 5 min at 37°C with buffer or with plasma ␤ 2 GPI, apple 4-␤ 2 GPI, apple 4-␤ 2 GPI-W316S, or DM of apple 4-␤ 2 GPI (final concentration 100 g/ml). Incubations were performed in the presence of 3 mM CaCl 2 . For competition experiments, proteins (final concentration 100 g/ml) were incubated with the inhibiting peptides (final concentration 500 g/ml) for 5 min at 37°C. As control, platelets were incubated with buffer and the inhibiting peptides. Afterward, platelets were lysed on ice with 1% CHAPS, containing 50 mM MES, and 150 mM NaCl (pH 7.4). Proteins were precipitated with 1 g/ml of a polyclonal anti-apoER2Ј antibody (sc-10112, Santa Cruz Biotechnology, Santa Cruz, CA) and protein G-Sepharose (Amersham Biosciences). The immunoprecipitations were incubated for 18 h at 4°C in a top over top rotor, washed three times with lysis buffer, resuspended in non-reducing Laemmli sample buffer (0.001% (w/v) bromphenol blue, 2% (w/v) SDS, 10% (v/v) glycerol in 62.5 mM Tris, pH 6.8), and boiled for 5 min. The supernatants were subjected to 10% SDS-PAGE and electroblotted onto an Immobilon-P polyvinylidene difluoride membrane. Blots were blocked with TBS with 0.1% (v/v) Tween 20 (TBST) containing 2% (w/v) nonfat dry milk for 1 h at room temperature. Incubation with anti-␤ 2 GPI antibody 2B2 (3 g/ml) was performed overnight in TBST supplemented with 1% nonfat dry milk. The membranes were washed three times and incubated with peroxidase-conjugated rabbit anti-mouse antibody (1:2500; Dako, Glostrup, Denmark) in the same buffer. Bands on blots were visualized with enhanced chemiluminescence. This experiment was performed with washed platelets from three different donors.

Expression and Purification of Domain Deletion Mutants-
To study the effect of domain involvement of apple 4-␤ 2 GPI on PL binding, clotting time, and platelet adhesion, dimeric constructs of ␤ 2 GPI fused to the C terminus of the dimerization domain (apple 4) of factor XI were made. Baby hamster kidney cells were transfected with expression vectors containing DM of apple 4-␤ 2 GPI. Protein expression was confirmed by Western blotting using an anti-␤ 2 GPI monoclonal antibody. Cell lines with the highest expression were selected using a ␤ 2 GPI ELISA. The proteins were purified using a monoclonal ␣-␤ 2 GPI antibody (monoclonal antibody 2B2) column followed by further purification on a Mono S column using fast protein liquid chromatography. After purification, DM were applied on a 7.5% SDS-PAGE under nonreducing (Fig. 1, panel A) and reducing (panel B) conditions and stained with Coomassie Brilliant Blue. In the presence of SDS full-length apple 4-␤ 2 GPI (D), apple 4-␤ 2 GPI-W316S (W), apple 4-⌬1␤ 2 GPI (⌬I), apple 4-⌬2␤ 2 GPI (⌬II), and apple 4-⌬5␤ 2 GPI (⌬V) migrated as monomers with an apparent molecular mass of 50 kDa. Plasma ␤ 2 GPI (M) migrated with a molecular mass of 45 kDa under non-reducing conditions. Under reducing conditions, full-length apple 4-␤ 2 GPI and apple 4-␤ 2 GPI-W316S migrated with a molecular mass of ϳ62 kDa. Apple 4-⌬1␤ 2 GPI, apple 4-⌬2␤ 2 GPI, and apple 4-⌬5␤ 2 GPI migrated slightly slower with a molecular mass of ϳ56 kDa. Plasma ␤ 2 GPI migrated with a molecular mass of ϳ52 kDa.
Effect of Apple 4-␤ 2 GPI Domain Deletion Mutants on Clotting Time-To study the role of individual domains of apple 4-␤ 2 GPI for competing with coagulation factors, we studied the effect of the DM on clotting time. For this purpose we performed the PT. Concentrations of 200 g/ml plasma-derived ␤ 2 GPI, apple 4-␤ 2 GPI, or DM, diluted in TBS, were mixed 1:1 with normal pooled plasma and incubated for 30 min at 4°C (final concentration 100 g/ml). This was followed by measurement of the PT. The clotting time in the presence of buffer was set at 100%. The observed effect for ⌬I and ⌬II was comparable with fulllength apple 4-␤ 2 GPI; apple 4-␤ 2 GPI showed a relative prolongation of the clotting time to 171.3 Ϯ 3.7%, ⌬I to 163.6 Ϯ 7.9%, and ⌬II to 180.7 Ϯ 23.8%. Results are presented in Fig. 2. The addition of plasma ␤ 2 GPI, ⌬V, or apple 4-␤ 2 GPI-W316S to normal pooled plasma did not influence the clotting time. Furthermore, the control proteins apple 4 and apple 2-␤ 2 GPI did not influence the clotting time.
Binding of Domain Deletion Mutants to Immobilized Phospholipids-The phospholipid binding features of apple 4-␤ 2 GPI fusion proteins were tested in a solid phase binding assay. Phospholipid vesicles (25 M, 20% PS, 80% PC) were immobilized on 96-well ELISA plates, and binding of plasma-derived ␤ 2 GPI, and DM of apple 4-␤ 2 GPI was measured. As shown in Fig. 3, half-maximal binding of apple 4-␤ 2 GPI to phospholipid vesicles occurred at a concentration of 2.1 g/ml (TABLE  ONE). For the domain deletion mutants a similar interaction with phospholipid vesicles was observed: with ⌬I having half-maximal binding to phospholipids at a concentration of 2.9 g/ml and ⌬II at a concentration of 4.1 g/ml. Half-maximal binding to immobilized phospholipids of ⌬V occurred at a concentration of 29.2 g/ml. For apple 4-␤ 2 GPI-W316S, half-maximal binding was observed at a concentration of 26.0 g/ml. The presence of an aa substitution in the phospholipid-insertion loop explains why the W316S mutant hardly binds to anionic phospholipids. Plasma-derived ␤ 2 GPI showed little binding at a concentration of 16 g/ml.

Effect of Domain Deletion Mutants on Platelet Deposition to Collagen Type III under Conditions of Flow-
To determine which domain of apple 4-␤ 2 GPI is involved in platelet sensitization, we performed perfusion experiments with citrated whole blood preincubated with plasma ␤ 2 GPI, apple 4-␤ 2 GPI, or DM (final concentration 100 g/ml). To determine basal platelet adhesion to collagen type III, whole blood was preincubated with buffer. The basal platelet coverage after 90 s was 9.4 Ϯ 2.1%, which was set at 100% (baseline). As shown in Fig. 4A, no increase in platelet adhesion was found when plasma ␤ 2 GPI, apple 4, apple 2-␤ 2 GPI, or ⌬V were added to whole blood (105.5 Ϯ 11.4, 103.8 Ϯ 15.6, 97.3 Ϯ 16.1, and 99.5 Ϯ 5.5%, respectively). As has also been shown by Lutters et al. (27), apple 4-␤ 2 GPI-W316S did not induce increased platelet adhesion to collagen. In contrast, platelet adhesion increased significantly when full-length apple 4-␤ 2 GPI, ⌬I, or ⌬II was added to whole blood (155.4 Ϯ 11.0, 148.3 Ϯ 8.6, and 157.5 Ϯ 7.9%, respectively). Morphology of platelets and number of platelet aggregates were similar   in conditions with full-length apple 4-␤ 2 GPI (Fig. 4B, panel a), ⌬I (not shown), and ⌬II (not shown). Plasma ␤ 2 GPI, ⌬V (panels b and d, respectively), and apple 4-␤ 2 GPI-W316S (not shown) displayed comparable effects on morphology and number of platelet aggregates as incubation with buffer (panel a).

Binding of Apple 4-␤ 2 GPI and the W316S Mutant to Immobilized
Soluble Human ApoER2Ј-Binding of apple 4-␤ 2 GPI and the W316S mutant to immobilized sh-apoER2Ј was measured using an ELISA setup. As shown in Fig. 5, half-maximal binding of both apple 4-␤ 2 GPI and apple 4-␤ 2 GPI-W316S to immobilized sh-apoER2Ј occurred at concentrations as low as 2.9 Ϯ 0.7 and 6.0 Ϯ 0.8 g/ml, respectively (corresponds to 25 and 52 nM, respectively). Also, plasma ␤ 2 GPI in the presence of a monoclonal ␣-␤ 2 GPI antibody (19H9) displayed binding to sh-apoER2Ј. Half-maximal binding occurred at a concentration of 2.4 Ϯ 0.7 g/ml (corresponds to 21 nM). No binding was found with plasma ␤ 2 GPI, apple 4, or ⌬V. Apple 2-␤ 2 GPI displayed only a slight interaction with sh-apoER2Ј.

Association of Domain Deletion Mutants and the W316S Mutant of Apple 4-␤ 2 GPI with ApoER2Ј on Platelets-
The only member of the LDL receptor family known to be present on platelets is apoER2Ј. To demonstrate that a cationic patch in domain V of ␤ 2 GPI and not the hydrophobic PL-insertion loop in domain V (Ser 311 -Lys 317 ) is responsible for interaction with apoER2Ј, immunoprecipitations were performed. Platelets were incubated with buffer, plasma-derived ␤ 2 GPI, full-length apple 4-␤ 2 GPI, DM, or apple 4-␤ 2 GPI-W316S, lysed, and subjected to immunoprecipitation with an anti-apoER2Ј antibody. Afterward, Western blots were incubated with a monoclonal anti-␤ 2 GPI antibody to detect interaction between ␤ 2 GPI and apoER2Ј. Association with apoER2Ј was observed with apple 4-␤ 2 GPI (Fig. 6). Hardly any association was observed when platelets were incubated with plasma ␤ 2 GPI or ⌬V. To demonstrate that this finding was not the result of the inability of ⌬V to bind to anionic phospholipids, immunoprecipitations were performed with apple 4-␤ 2 GPI-W316S. The interaction of the W316S mutant with apoER2Ј on the surface of platelets was similar to that of full-length apple 4-␤ 2 GPI. Interaction between apple 4-␤ 2 GPI and sh-apoER2Ј was measured in a solid phase binding assay. Soluble human apoER2Ј was coated on a hydrophobic 96-well ELISA plate in a concentration dependent manner (0 -10 g/ml). After blocking, plasma ␤ 2 GPI (E), apple 4 (224), apple 2-␤ 2 GPI (ϩ), apple 4-␤ 2 GPI (OE), ⌬V (‚), apple 4-␤ 2 GPI-W316S (ƒ) mutant and plasma ␤ 2 GPI ϩ 19H9 (•) were incubated (3 g/ml). Afterward bound protein was detected with rabbit polyclonal anti-␤ 2 GPI. Bound antibody was detected using peroxidase-conjugated swine anti-rabbit antibody. Results are expressed as mean Ϯ S.D. (n ϭ 3).

Apparent dissociation constants of domain deletion mutants for PS/PC vesicles
Curves of plasma ␤ 2 GPI and apple 4-␤ 2 GPI fusion proteins were fitted according to a one-site binding model in GraphPad. Half-maximal binding is given as apparent K d (K d (app)) both in g/ml as in nanomole/liters (nM). Results are expressed as mean Ϯ S.D. (n ϭ 3).  Association of Apple 4-␤ 2 GPI with ApoER2Ј in the Presence of CKNKEKKC, EKCKNKCK, (scrambled) or VSRGGMRK Peptides-Association of ligands with members of the LDL receptor family is sup-ported by electrostatic interactions. To investigate the possibility that association of apple 4-␤ 2 GPI with apoER2Ј is supported by a cationic patch in domain V of apple 4-␤ 2 GPI, competition studies were performed with positively charged peptides. Incubation of washed platelets with apple 4-␤ 2 GPI and CKNKEKKC (represents a cationic patch at aa position 282-287 in domain V of ␤ 2 GPI) peptide resulted in decreased binding of apple 4-␤ 2 GPI to apoER2Ј (Fig. 7A). This effect was not seen with the VSRGGMRK peptide (represents a cationic patch at aa position 37-44 in domain I of ␤ 2 GPI). The decreased binding of apple 4-␤ 2 GPI to apoER2Ј in the presence of CKNKEKKC was also seen for the W316S mutant. Incubation of washed platelets with apple 4-␤ 2 GPI and EKCKNKCK (scrambled peptide) resulted in a minor reduction of the association between apple 4-␤ 2 GPI and apoER2Ј (Fig. 7B).

Apparent K d
Binding of Apple 4-␤ 2 GPI to Sh-ApoER2Ј in the Solid Phase Binding Assay in the Presence of CKNKEKKC, EKCKNKCK (Scrambled), or VSRGGMRK Peptides-To show direct competition between apple 4-␤ 2 GPI and the different peptides, apple 4-␤ 2 GPI was able to interact with sh-apoER2Ј in the presence of increasing concentrations of peptides VSRGGMRK, CKNKEKKC, or EKCKNKCK (scrambled). As shown in Fig. 7C, inhibition was observed for both the CKNKEKKC and EKCKNKCK peptides. Binding of apple 4-␤ 2 GPI to sh-apoER2Ј in the presence of the CKNKEKKC peptide was reduced to 12.5 Ϯ 3.6%. In the presence of the scrambled peptide EKCKNKCK binding was reduced to 52.8 Ϯ 7.4%. In the presence of the VSRGGMRK peptide no inhibition was observed (data not shown).
Effect of the CKNKEKKC and EKCKNKCK Peptides on the Prothrombin Time-The inhibiting effect of the peptides on association between apple 4-␤ 2 GPI and apoER2Ј might be because of interference with binding of the apple 4-␤ 2 GPI constructs to phospholipids. Therefore, we investigated the influence of the peptides in the clotting time (PT), which is a phospholipid-dependent clotting assay. The clotting time in the presence of buffer was set at 100%. Apple 4-␤ 2 GPI displayed a relative prolongation of the PT to 194.0 Ϯ 2.8% (Fig. 8). In the presence of the CKNKEKKC or EKCKNKCK peptides (500 g/ml) prolongation of the clotting time was observed to 191.3 Ϯ 7.8 and 216.5 Ϯ 23.3%, respectively. The CKNKEKKC or EKCKNKCK peptides (500 g/ml) did not influence the clotting time in the absence of dimeric ␤ 2 GPI up to concentrations of 1 mg/ml (data not shown).

DISCUSSION
It is difficult to envision that mere binding of ␤ 2 GPI to anionic PL on the cell surface can activate these cells. Therefore, a search for a cellular FIGURE 7. Immunoprecipitations with apple 4-␤ 2 GPI and apoER2 on platelets in the presence of CKNKEKKC, EKCKNKCK, or VSRGGMRK peptides. Competition experiments were performed to investigate electrostatic involvement between apple 4-␤ 2 GPI and apoER2Ј. For this purpose washed platelets were incubated with buffer (T), plasma ␤ 2 GPI (M), apple 4-␤ 2 GPI (D), and apple 4-␤ 2 GPI-W316S (W) in the presence of cationic peptides; incubations were in the presence of CKNKEKKC or VSRGGMRK peptides (panel A) or in the presence of the EKCKNKCK peptide (panel B) (final peptide concentration; 500 g/ml) for 5 min at 37°C. Afterward immunoprecipitations were performed as described previously. Blots represent three different experiments. To show direct competition between apple 4-␤ 2 GPI and the CKNKEKKC (OE) and EKCKNKCK (⌬) peptides, binding of apple 4-␤ 2 GPI in the presence of the peptides was investigated in the solid phase binding assay (panel C). FIGURE 6. Immunoprecipitations with domain deletion mutants of apple 4-␤ 2 GPI and apoER2 on platelets. Washed platelets (300,000/l), resuspended in Hepes/Tyrode buffer containing 3 mM CaCl 2 were incubated with buffer, apple 4-␤ 2 GPI (D), ⌬I(⌬I), ⌬II (⌬II), ⌬V(⌬V), or apple 4-␤ 2 GPI-W316S (W) (final concentration 100 g/ml; 10% v/v) for 5 min at 37°C followed by lysis with CHAPS on ice. Lysed platelets were subjected to immunoprecipitation with an anti-apoER2Ј antibody in the presence of protein G-Sepharose. Afterward Western blots were incubated with an anti-␤ 2 GPI antibody followed by visualization using chemiluminescence. The blot represents three different experiments. receptor for ␤ 2 GPI on platelets was initiated. Lutters et al. (27) have demonstrated that dimerization of ␤ 2 GPI (either artificially by fusing ␤ 2 GPI with the apple 4 domain of factor XI, or physiologically by binding aPL to ␤ 2 GPI) results in increased affinity of ␤ 2 GPI for platelets, which results in an increased platelet deposition to collagen under conditions of flow. The increase in platelet adhesion could be blocked by the addition of receptor-associated protein, suggesting that a member of the LDL receptor family as the platelet receptor for dimeric ␤ 2 GPI was involved. This receptor has later been identified as apoER2Ј (25). In the present study, the domain of ␤ 2 GPI responsible for interaction with apoER2Ј has been determined by using constructs of dimeric ␤ 2 GPI lacking domains I, II, or V and a construct with an aa substitution (W316S) in the hydrophobic loop in domain V. Substitution of Trp 316 by a serine completely abolished binding of ␤ 2 GPI to anionic phospholipids (14). In this study, we showed that a cationic patch in domain V of dimeric ␤ 2 GPI is involved in interaction with apoER2Ј on platelets and that, by using the W316S mutant, the binding site for apoER2Ј does not coincide with the phospholipid-binding site within domain V.
Deletion of domain I or domain II does not have an effect on the function of dimeric ␤ 2 GPI. Both domain deletion mutants (⌬I and ⌬II) prolong the clotting time in the PT comparable with full-length dimeric ␤ 2 GPI (Fig. 2) and show comparable affinity for anionic phospholipids (Fig. 3). Furthermore, ⌬I and ⌬II cause an increased adhesion of platelets to collagen under conditions of flow comparable with full-length dimeric ␤ 2 GPI (Fig. 4A). Deletion of domain V or the W316S substitution abolished the prolongation of the clotting time induced by dimeric ␤ 2 GPI, which was to be expected, as domain V harbors the phospholipid-binding site. Indeed, both mutants show an approximate 15-fold decreased affinity for anionic phospholipids. Furthermore, in the perfusion model both mutants were not able to increase platelet adhesion to collagen, stressing the necessity of domain V in the activation of platelets.
Based on the observations in this article, we propose the following model for the activation of platelets by ␤ 2 -glycoprotein I (Fig. 9). ␤ 2 -Glycoprotein I binds to platelets with a low affinity. However, when bound to the platelet membrane a conformational change is induced in domain I of ␤ 2 GPI (36 -38), resulting in the exposure of a cryptic epitope in this domain. Binding of anti-␤ 2 GPI antibodies to this newly exposed epitope in domain I takes place. When one antibody interacts with two molecules of ␤ 2 GPI, the protein dimerizes resulting in an increased affinity for phospholipids on the outer surface of the platelet membrane. This increased affinity of the ␤ 2 GPI-anti-␤ 2 GPI complexes is mimicked by our recombinant dimeric ␤ 2 GPI construct. The binding to the platelet membrane also results in concentration of ␤ 2 GPI on the cellular surface, which allows interaction with apoER2Ј because of mass action effects. Stabilization of the binding of dimeric ␤ 2 GPI to phospholipids is crucial before it can bind to apoER2Ј, as the W316S mutant is not able to activate platelets under conditions of flow (Fig. 4A), despite the fact that the protein can bind to immobilized apoER2Ј (Fig. 5).
We cannot exclude that also a conformational change in domain V is necessary for interaction with apoER2Ј, as plasma ␤ 2 GPI is not able to associate with the receptor (Fig. 6). Conformational changes in domain V of ␤ 2 GPI after binding to phospholipids have been described before (39). The interaction between ␤ 2 GPI and apoER2Ј takes place via a cationic region in domain V, as peptides covering cationic amino acids present in domain V are able to inhibit the binding of dimeric ␤ 2 GPI to apoER2Ј (Fig. 7, A-C). The apoER2Ј interaction site does not completely overlap the phospholipid-binding domain, as the hydrophobic loop in domain V is not involved in this interaction.
After interaction with dimeric ␤ 2 GPI, dimerization of the receptor FIGURE 9. Proposed mechanism of platelet activation via dimerization of ␤ 2 GPI by aPL. ␤ 2 -Glycoprotein I circulates free in plasma with low affinity for phospholipids. A conformational change in domain I ␤ 2 GPI is induced after low-affinity binding to platelets. Dimerization of ␤ 2 GPI via binding of aPL occurs followed by stabilization of the complex on the platelet surfaced that subsequently results in firm adhesion of the ␤ 2 GPI-aPL complex to PL and association with apoER2Ј on the surface of platelets. We cannot exclude the possibility that dimeric ␤ 2 GPI dissociates from the cellular surface after interaction with apoER2Ј. The interaction results in downstream signaling, thromboxane A2 (TxA2) synthesis, and further platelet activation.