Divalent Cations Differentially Regulate Integrin αIIb Cytoplasmic Tail Binding to β3 and to Calcium- and Integrin-binding Protein*

We have used recombinant or synthetic αIIb and β3 integrin cytoplasmic peptides to study their in vitro complexation and ligand binding capacity by surface plasmon resonance. α·β heterodimerization occurred in a 1:1 stoichiometry with a weakK D in the micromolar range. Divalent cations were not required for this association but stabilized the α·β complex by decreasing the dissociation rate. α·β complexation was impaired by the R995A substitution or the KVGFFKR deletion in αIIb but not by the β3 S752P mutation. Recombinant calcium- and integrin-binding protein (CIB), an αIIb-specific ligand, bound to the αIIbcytoplasmic peptide in a Ca2+- or Mn2+-independent, one-to-one reaction with aK D value of 12 μm. In contrast,in vitro liquid phase binding of CIB to intact αIIbβ3 occurred preferentially with Mn2+-activated αIIbβ3conformers, as demonstrated by enhanced coimmunoprecipitation of CIB with PAC-1-captured Mn2+-activated αIIbβ3, suggesting that Mn2+activation of intact αIIbβ3 induces the exposure of a CIB-binding site, spontaneously exposed by the free αIIb peptide. Since CIB did not stimulate PAC-1 binding to inactive αIIbβ3 nor prevented activated αIIbβ3 occupancy by PAC-1, we conclude that CIB does not regulate αIIbβ3 inside-out signaling, but rather is involved in an αIIbβ3 post-receptor occupancy event.

Integrins are ␣␤ heterodimeric cell-surface receptors that promote not only adhesion to components present within the extracellular matrix or on the surface of opposite cells but also transfer information into and out of a cell (1). The adhesive functions of integrins can be regulated by intracellular processes referred to as "inside-out signaling." Conversely, ligand binding to the extracellular domain of integrins initiates a cascade of intracellular events termed "outside-in signaling" that generate a large spectrum of cellular responses, such as cell migration, proliferation, differentiation, and gene expression (2). Integrin cytoplasmic tails appear to be key elements in these bidirectional signaling pathways, despite their short size as compared with other signaling receptors and the absence of any demonstrable catalytic activity (3,4). Integrin ␣ and ␤ cytoplasmic domains are thought to mediate signaling events through modifications of their own structural and spatial organization and/or through interactions with specific cytoplasmic components. Various proteins have been identified that bind, at least in vitro, to the cytoplasmic tail of ␣ and ␤ subunits and are likely to play a role in regulating integrin signaling functions. These include cytoskeletal components such as talin and ␣-actinin, as well as several signaling or regulatory proteins such as integrin-linked kinase p59 ILK , focal adhesion kinase pp125 FAK , Grb2, ␤ 3 -endonexin, cytohesin-1, integrin cytoplasmic domain-associated protein ICAP-1, calreticulin and calcium-and integrin-binding protein CIB 1 (reviewed in Refs. 5 and 6).
Recently used methods for studying protein-protein interactions, such as the two-hybrid system, have allowed the identification of integrin-specific intracellular ligands (7)(8)(9)(10)(11)(12). These methods are based on the use of a unique linear amino acid sequence as a bait and consequently do not take into account the secondary and tertiary structural features of the interacting molecules. However, numerous studies tend to demonstrate that ␣ and ␤ cytoplasmic domains adopt a defined conformation and that the preservation of these structural constraints is crucial to maintain the functional properties of integrin receptors (13)(14)(15)(16)(17)(18)(19)(20).
One of the best studied integrins is the platelet fibrinogen receptor, integrin ␣ IIb ␤ 3 , that undergoes conformational changes necessary for receptor function. In order to elucidate further the structural relationship of the cytoplasmic tails of ␣ IIb and ␤ 3 , we have used surface plasmon resonance biosensor technology to monitor real time assembly of the integrin ␣ IIb and ␤ 3 cytoplasmic tails and to investigate their ligand binding capacity.
Platelets and Cell Lines-Outdated platelet concentrates were kindly provided by Dr. J.-C. C. Faber (Luxembourg Red Cross Blood Transfusion Center). The stable transfected CHO cell line A06, expressing high levels of human ␣ v ␤ 3 integrin (21), was grown in Iscove's modified Dulbecco's medium, and HEL-5J20 cells in RPMI medium (Life Technologies) (22). Culture medium was supplemented with glutamine, penicillin, and streptomycin, and 10% heat-inactivated fetal calf serum. The adherent CHO cells were routinely passaged with EDTA buffer, pH 7.4 (1 mM EDTA, 126 mM NaCl, 5 mM KCl, 50 mM Hepes).
Construction of pGEX-4T-2 Expression Plasmids-The cDNA encoding the wild type or S752P mutant human ␤ 3 integrin cytoplasmic tail (Lys 716 -Thr 762 ) was generated by the polymerase chain reaction (PCR) using full-length pBJ1-␤ 3 plasmids as templates (21). The upstream (sense) primer was a 28-mer with a BamHI site (G2GATCC) corresponding to the ␤ 3 nucleotide sequence 2245-2266, 5Ј-GGATC-CAAACTCCTCATCACCATCCACG-3Ј. The downstream (antisense) primer was a 30-mer corresponding to the ␤ 3 nucleotide sequence 2365-2388 and comprising an SmaI restriction site (CCC2GGG) followed by a stop codon 5Ј-CCCGGGTTAAGTGCCCCGGTACGTGATATT-3Ј. PCR amplification was performed using the Takara PCR kit (Shiga, Japan). The full-length cDNA encoding human CIB was obtained by reverse transcriptase-PCR (RT-PCR) of HEL-5J20 cell mRNA. Briefly, total RNA was isolated from 5 ϫ 10 6 cells according to the method of Chomczynski and Sacchi (23), and RT-PCR was performed using the RNA-PCR kit from Promega (Madison, WI). The sense primer was a 36-mer corresponding to the published CIB nucleotide sequence 1-30 (12) with an additional BamHI restriction site (G2GATCC), 5Ј-GGATCCAT-GGGGGGCTCGGGCAGTCGCCTGTCCAAG-3Ј. The downstream antisense primer was a 32-mer corresponding to the CIB nucleotide sequence 551-576 followed by a stop codon and an EcoRI restriction site (G2GATTC), 5Ј-GGATTCTCACAGGACAATCTTAAAGGAGCTGG-3Ј. All the primers used to generate cDNA fragments were obtained from Life Technologies, Inc. PCR and RT-PCR products were purified using the PCR Preps DNA Purification System from Promega. They were digested with the corresponding restriction enzymes and inserted into the glutathione S-transferase (GST) vector pGEX-4T-2 (Amersham Pharmacia Biotech, Uppsala, Sweden) containing a thrombin cleavage site. The expected nucleotide sequence was confirmed for each construct by direct sequencing using the T7 sequencing kit from Amersham Pharmacia Biotech.
Expression and Purification of Recombinant Fusion Proteins-Native GST and in-frame GST fusion proteins were expressed in Escherichia coli JM105 (Amersham Pharmacia Biotech) according to the manufacturer's instructions. Bacterial pellets were suspended in PBS (140 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , pH 7.4) containing 10 mM EDTA and 50 M aminoethylbenzenesulfonyl fluoride and were incubated for 30 min at room temperature in the presence of 200 g/ml lysozyme. The bacterial suspension was submitted to short bursts of sonication and further treated with 1% Triton X-100 for 30 min at 4°C. The insoluble material was pelleted by a 20-min centrifugation at 10,400 ϫ g at 4°C. Supernatants were filtered on a 0.8-m membrane and submitted to affinity chromatography on a glutathione-Sepharose 4B (Amersham Pharmacia Biotech) column (100 ϫ 20-mm inner diameter) previously equilibrated with PBS. Bound fusion proteins were eluted with 10 mM reduced glutathione in 50 mM Tris-HCl, pH 8.0. GST-␤ 3 and GST-␤ 3 (S752P) fusion proteins were further purified by immunoaffinity chromatography using the anti-␤ 3 cytoplasmic tail mAb C3a.19.5 immobilized on CNBr-activated Sepharose CL4B (Amersham Pharmacia Biotech) in order to eliminate free GST.
Proteolytic Cleavage of GST Fusion Proteins and Purification of Recombinant Products-Cleavage of recombinant CIB or wild type ␤ 3 cytoplasmic tail peptide from GST was achieved by incubating 2 NIH units of thrombin protease (Amersham Pharmacia Biotech) per mg of purified material under gentle stirring for 5 h at room temperature. ␤ 3 peptide was purified by preparative reverse-phase high performance liquid chromatography (HPLC) using a C4 column (100 ϫ 20-mm inner diameter) with a 0 -40% linear gradient of acetonitrile in 0.05% trifluoroacetic acid. The amino acid sequence of the recombinant peptide was checked by microsequencing on an Applied Biosystems Procise sequencer. The purified peptide was stored dessicated at 4°C. For CIB purification, the thrombin hydrolysate was extensively dialyzed against PBS and then passed through a glutathione-Sepharose column in order to remove GST. The flow-through fraction containing CIB was kept frozen at Ϫ20°C until use.
␣ IIb ␤ 3 Binding to Immobilized Human Fibrinogen-␣ IIb ␤ 3 binding to fibrinogen was determined according to Kouns et al. (25) with some modifications. 96-well microtiter plates (Costar, Cambridge, MA) were coated overnight at 4°C with 100 l/well purified human fibrinogen (Sigma, Bornem, Belgium) at 5 g/ml in TBS. The plates were then saturated with TBS containing 3.5% bovine serum albumin and 0.05% NaN 3 (125 l/well) overnight at 4°C. The ConA-enriched platelet glycoprotein fraction was serially diluted in TBS/ELISA alone (TBS containing 1% bovine serum albumin, 0.035% Triton X-100) or in TBS/ ELISA containing either 2 g/ml D3GP3 mAb or 10 mM MnCl 2 . 100-l aliquots were added to the wells and incubated 4 h at room temperature. After three washes with TBS, 1 g/ml polyclonal rabbit anti-␣ IIb ␤ 3 antibodies in TBS/ELISA were added (100 l/well) for 2 h at room temperature. The wells were washed three times with TBS, followed by 90 min incubation at room temperature with 100 l/well donkey antirabbit Ig antibodies conjugated to horseradish peroxidase (Amersham Pharmacia Biotech). After three washes with TBS, 100 l of 0.1 mg/ml 3,3Ј,5,5Ј-tetramethylbenzidine, 0.01% H 2 O 2 in 140 mM sodium acetatecitrate buffer, pH 6.0, were dispensed into each well. The enzymatic reaction was stopped by addition of 25 l of 2 M H 2 SO 4 , and the absorbance was measured at 450 nm.
In Vitro Liquid Phase ␣ IIb ␤ 3 -CIB Binding Assays-ConA-purified platelet glycoproteins (250 g) or ␣ v ␤ 3 -enriched CHO-A06 cell glycoproteins (1 mg) were incubated with 50 g of purified GST-CIB or GST alone for 2 h at 4°C under gentle stirring. Experiments were carried out in ice-cold 0.1% Triton X-100, 150 mM NaCl, 50 mM Tris-HCl, pH 7.4, containing either 2 mM CaCl 2 , 2.5 mM EGTA, or 1 mM CaCl 2 , 1 mM MgCl 2 Ϯ 10 mM MnCl 2 . In Mn 2ϩ -related assays, platelet glycoproteins were first preincubated for 30 min at room temperature in 10 mM MnCl 2 or in Mn 2ϩ -free buffer before incubation with GST fusion proteins. For ␣ IIb ␤ 3 capture, the mixtures were incubated with 5 g of the 10E5 mAb for 2 h at 4°C or, alternatively, with 4 g of the mAb PAC-1 for 2 h at room temperature followed by 8.5 g of rabbit anti-mouse -chainspecific antibodies (Jackson Immunoresearch, West Grove, PA) for an additional 2 h at 4°C. Protein A-Sepharose CL 4B beads (80 l of a 50% slurry suspension) were added and incubated for 2 h at 4°C. For GST fusion protein capture, the mixtures were incubated for 2 h at 4°C with 80 l of 50% slurry glutathione-Sepharose 4B suspension. Control experiments were performed using non-substituted Sepharose CL4B. The adsorbents were washed three times with 500 l of the respective ice-cold incubation buffer, and the captured proteins were recovered by boiling the beads in 30 -50 l of 5% ␤-mercaptoethanol, 2% SDS, 10% glycerol, 25 g/ml bromphenol blue in 15.625 mM Tris-HCl, pH 6.8. Each sample was analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) followed by immunoblotting as described below.
Protein Assay, Electrophoresis, and Western Blot Analysis-Protein concentration was determined using the Bio-Rad Protein assay reagent. SDS-PAGE was performed using the mini-Protean II electrophoresis system (Bio-Rad), and Tris-Tricine SDS-PAGE was carried out according to the method described by Schagger and von Jagow (26). Electrophoresed samples were transferred onto Hybond-C nitrocellulose membrane (Amersham Pharmacia Biotech) using a semi-dry transblot apparatus (Amersham Pharmacia Biotech). The membranes were blocked overnight in blotting buffer (5% dry milk, 0.1% Tween 20, 150 mM NaCl, 20 mM Tris-HCl, pH 7.4), and incubated for 2 h with the anti-␤ 3 4D10G3 mAb mixed with either the anti-␣ IIb S1.3 mAb or the anti-␣ v VNR 139 mAb diluted in blotting buffer. After three washes in blotting buffer, membranes were incubated for 1 h with diluted sheep anti-mouse Ig conjugated to horseradish peroxidase (Amersham Pharmacia Biotech). Membranes were again washed three times in blotting buffer and then in 137 mM NaCl, 20 mM Tris-HCl, pH 7.4 (TBS/WB), and developed using the chemiluminescence ECL kit (Pierce) according to the manufacturer's instructions. The membranes were then stripped by successive washes in TBS/WB containing 100 mM ␤-mercaptoethanol, 2% SDS, 52.5 mM Tris-HCl, pH 6.7, for 30 min at 50°C, and again in TBS/WB. After an overnight incubation in blotting buffer, the membranes were reprobed for 2 h with polyclonal goat anti-GST antibodies (Amersham Pharmacia Biotech), and antibody binding was detected as described above with horseradish peroxidase-conjugated rabbit antigoat IgG (Jackson Immunoresearch).
Surface Plasmon Resonance Binding Studies-Real time biomolecular interaction analysis was performed using the Bialite TM or Biacore X TM instruments (Biacore, Uppsala, Sweden). Purified proteins were covalently attached to carboxymethyl dextran (CM5) chips (Biacore) previously activated with a mixture of N-hydroxysuccinimide and Nethyl-NЈ-dimethylaminopropyl carbodiimide according to the manufacturer's instructions. Experiments were performed at 25°C using as running buffers either Biacore HBS (150 mM NaCl, 3.4 mM EDTA, 0.005% Tween 20, 10 mM Hepes, pH 7.4) or TBS/Bia (150 mM NaCl, 0.005% Tween 20, 50 mM Tris-HCl, pH 7.4) Ϯ 2 mM CaCl 2 . The sensorchips were regenerated with a short pulse of either 200 mM glycine HCl, pH 2.2 (anti-GST antibody-coated chip), or 10 mM HCl (GST fusion protein-derivatized chip). The amount of analyte bound to the immobilized ligand was monitored by measuring the variation of the surface plasmon resonance angle as a function of time. Results were expressed in resonance units (RU), an arbitrary unit specific for the Biacore instrument (1000 RU correspond to approximately 1 ng of bound protein/mm 2 and are recorded for a change of 0.1°in resonance angle) (27). The transformation of crude data, the preparation of overlay plots, and the determination of kinetic parameters of the binding reactions were performed using the Biaevaluation 3.0 software. The association rate constant (k on ) and the dissociation rate constant (k off ) were determined separately from individual association and dissociation phases, respectively, assuming a one-to-one interaction. The affinity constant K D was calculated as k off /k on . Experimental values from the first 20 s at the beginning of each phase were not considered in the fitting to avoid distortions due to injection and mixing.

Generation of Recombinant Wild Type or Mutant ␤ 3 Integrin
Cytoplasmic Peptides and of CIB-In order to perform in vitro studies of ␣ IIb and ␤ 3 cytoplasmic tail association, we generated recombinant GST fusion proteins containing a thrombin cleavage site and corresponding to the entire wild type or S752P mutant cytoplasmic domain of the ␤ 3 integrin (residues 716 -762) and to the ␣ IIb -binding protein CIB (residues 1-191). The fusion proteins were isolated from bacterial cell lysates by glutathione affinity chromatography, and the GST-␤ 3 proteins were further immunopurified using the anti-␤ 3 monoclonal antibody (mAb) C3a.19.5. Thrombin-released wild type ␤ 3 cytoplasmic tail peptide and CIB were purified free of GST using reverse-phase HPLC and glutathione affinity chromatography, respectively. The accurate amino acid sequence of the ␤ 3 peptides was confirmed by microsequencing. SDS-PAGE analysis of isolated proteins revealed a purity greater than 98%, as evaluated by densitometric scanning of the gel (Fig. 1). The apparent molecular masses of both GST-␤ 3 and GST-␤ 3 (S752P) (34 kDa), GST-CIB (48 kDa), CIB (25 kDa), and GST (29 kDa) were in good agreement with the predicted mass deduced from their amino acid composition. When analyzed with higher resolutive SDS-PAGE and on a C18 HPLC column, the 5.6-kDa ␤ 3 peptide appeared homogeneous with only slight impurities (Fig. 2).
In Vitro Complex Formation of ␣ IIb and ␤ 3 Cytoplasmic Domains as Monitored by SPR Analysis-In vitro complex formation of wild type ␣ IIb and ␤ 3 integrin cytoplasmic tails was investigated by surface plasmon resonance (SPR) in order to monitor real time biomolecular interactions. We first examined whether the synthetic ␣ IIb peptide was able to bind to the ␤ 3 cytoplasmic tail fused to GST. Purified anti-GST polyclonal antibodies were immobilized on a sensor chip through amine coupling and were allowed to stably capture native GST or the GST-␤ 3 fusion protein before injection of the ␣ IIb peptide. As shown in Fig. 3A, a characteristic binding signal was monitored when the peptide was brought into contact with captured GST-␤ 3 but not with either an uncoated surface, immobilized antibodies alone, or GST-antibody complexes. In these latter control experiments, the rapid change in the resonance signal was due to a dilution buffer-induced, nonspecific change in the bulk refractive index. The maximum response monitored at the end of the peptide injection phase was about 60 -70 RU for 1100 RU of initially captured GST-␤ 3 protein. The corresponding molar ratio was estimated at ϳ0.8 mol/mol and was consistent with a 1:1 interaction. Further studies showed that ␣ IIb binding was dose-dependent and could be almost completely inhibited by soluble GST-␤ 3 protein but not by GST alone (Fig. 3, B and  C). Taken together, these data demonstrate that ␣ IIb specifi- cally interacts with the ␤ 3 cytoplasmic tail and that this interaction can be monitored using SPR technology.
The ␣ IIb ␤ 3 Cytoplasmic Domain Complex Is Stabilized by Divalent Cations-In order to determine the role of divalent cations in integrin ␣ IIb ␤ 3 cytoplasmic domain association, we investigated the binding of ␣ IIb to immobilized GST-␤ 3 in the presence of 2 mM CaCl 2 or MgCl 2 . The influence of Ca 2ϩ on the interaction was apparent from the slopes of the sensorgrams in Fig. 4. Interestingly, the rates of association of ␣ IIb with GST-␤ 3 were similar, independent of the presence or absence of Ca 2ϩ . In contrast, the dissociation of ␣ IIb was more rapid in the absence of Ca 2ϩ , as demonstrated by a greater slope in the curve. Similar results were obtained with a low cation concentration (50 M) or with a Mg 2ϩ -containing buffer (data not shown). To characterize further the dynamic parameters of the interaction, we determined binding isotherms in the presence or absence of Ca 2ϩ by injecting ␣ IIb peptide solutions ranging from 4.2 to 83.3 M over a GST-␤ 3 fusion protein-coated chip (Fig. 3B). From these curves, the association and dissociation rates and the apparent K D of the binding were determined as indicated under "Experimental Procedures." As shown in Table  I, the on rates with or without Ca 2ϩ were very similar. In contrast, the peptide dissociation was slower when Ca 2ϩ ions were present in the flow as compared with Ca 2ϩ -free buffer. This resulted in an increased affinity, although the K D value was still in the range of weak interactions. Taken together, these data suggest that divalent ions have a different effect on the association and dissociation of ␣ IIb and ␤ 3 cytoplasmic tails.
In Vitro ␣/␤ Heterodimerization Is Impaired by the R995A Substitution or the KVGFFKR Deletion within the ␣ IIb Cytoplasmic Tail but Not by the S752P Substitution in the ␤ 3 Cytoplasmic Domain-Several mutations or deletions within the ␣ IIb and ␤ 3 cytoplasmic tails have been shown to disturb ␣ IIb ␤ 3mediated signaling, such as the ␣ IIb (R995A) substitution, the ␣ IIb membrane-proximal GFFKR truncation, or the ␤ 3 (S752P) point mutation (15,28,29). To investigate the influence of these mutations on the in vitro ␣⅐␤ complexation, a chip coated with GST-␤ 3 was used. The binding curves obtained following the injection of equimolar solutions of ␣ IIb (R995A) or ␣ IIb (Asn 996 -Gln 1008 ) peptides were strongly reduced as compared with the curve monitored with wild type ␣ IIb (Fig. 5A). Residual binding calculated from these curves 20 s after the end of the sample injection, using wild type ␣ IIb peptide binding as a 100% reference, were only ϳ15% for the substituted mutant and ϳ20% for the deleted mutant. In contrast, sensorgrams obtained when wild type ␣ IIb peptide was injected over GST-␤ 3 or GST-␤ 3 (S752P) were superimposable (Fig. 5B). The presence of 2 mM Ca 2ϩ in the running buffer only slightly improved  In Ca 2ϩ -related experiments, the Ca 2ϩ -containing buffer was injected during the dissociation phase to replace the running buffer (arrow). Data are expressed as absolute responses. The sensor chip used was the same as in Fig. 3, B and C.
␣ IIb mutant binding to GST-␤ 3 (22% residual binding for both). Calcium stabilized the interaction of the wild type ␣ IIb peptide with GST-␤ 3 (S752P) in the same way as with GST-␤ 3 , and kinetic parameters determined for the ␣ IIb /GST-␤ 3 (S752P) interaction in the absence or presence of 2 mM CaCl 2 were essentially the same as those obtained with wild type ␤ 3 (data not shown). These results demonstrate that the ␣ IIb membrane-proximal KVGFFKR sequence and the ␣ IIb residue Arg 995 are crucial for the formation of ␣⅐␤ complexes, as well as their subsequent stabilization by divalent cations. In contrast, the S752P mutation in the ␤ 3 cytoplasmic tail does not affect ␣⅐␤ dimerization and stabilization, suggesting that the ␤ 3 Cterminal part is not involved in these processes.
Calcium-independent Interaction of CIB with the ␣ IIb Cytoplasmic Tail-We have used purified recombinant CIB as a control reporter protein to monitor ␣ IIb cytoplasmic tail ligand binding functions. As shown in Fig. 6A, a binding signal was recorded when the ␣ IIb peptide was passed over sensorchips with antibody-captured GST-CIB but not with anti-GST antibodies alone, GST, or ethanolamine-substituted dextran, indicating that the interaction occurred specifically between ␣ IIb and CIB. The molar ratio calculated from these curves was ϳ0.7 mol/mol and was consistent with a 1:1 stoichiometry. Binding experiments performed in the presence or absence of 2 mM CaCl 2 in the running buffer showed that CaCl 2 did not significantly modify either the association or the dissociation phase (Fig. 6B), demonstrating that the ␣ IIb interaction with CIB was Ca 2ϩ -independent, despite the fact that CIB is a Ca 2ϩ -binding protein (12). The on and off rates, and the apparent K D , calculated as described previously from overlaid sensorgrams, are summarized in Table II. As predicted from the sensorgrams in Fig. 6B, all the dynamic parameters were essentially the same, independent of the presence or absence of Ca 2ϩ , and revealed a weak affinity of 12 M.
The ␣ IIb Cytoplasmic Peptide Can Bind Simultaneously to the ␤ 3 Cytoplasmic Tail and to CIB-We further examined whether the binding of the ␣ IIb peptide to antibody-captured GST-CIB was influenced by the ␤ 3 cytoplasmic tail. Samples of ] Experiments were performed on a Bialite instrument using a flow rate of 50 l/min and TBS/Bia buffer Ϯ 2 mM CaCl 2 as running and dilution buffer. The association (k on ) and dissociation (k off ) rate constants were generated using the Biaevaluation analysis software, from data recorded for a range of synthetic ␣ IIb peptide concentrations (4.2-83.3 M) injected over a GST-␤ 3 fusion protein-coated surface (ϳ12,500 RU). The affinity constants (K D ) were calculated as k off /k on from curvefitting analysis as described under "Experimental Procedures." The kinetic data presented are the mean values Ϯ S.D. of at least two separate experiments.
Buffer each of the two isolated peptides were injected sequentially or simultaneously after an in vitro incubation in the presence or absence of CaCl 2 , in order to allow ␣⅐␤ complex formation. Experiments were performed in a Ca 2ϩ -free running buffer. As expected from previous results, the ␣ IIb peptide always bound to captured GST-CIB when injected alone (Fig. 7A). In contrast, the purified ␤ 3 peptide did not. Injection of the ␣ IIb and ␤ 3 peptides preincubated in a Ca 2ϩ -free buffer led to a response almost similar to that monitored with the ␣ IIb peptide alone, whereas the ␣/␤ mixture preincubated in the presence of Ca 2ϩ supported an enhanced resonance signal (Fig. 7B). Comparable results were obtained when the experiments were carried out with running buffer containing 2 mM CaCl 2 (data not shown). Considering that the SPR response is directly related to the mass of the analyte adsorbed on the chip surface, we conclude from these results that binding of preformed ␣ IIb ⅐␤ 3 complexes was responsible for the difference observed with the reference signal recorded with the ␣ IIb peptide alone. These data indicate that ␣ IIb and ␤ 3 cytoplasmic tails can form a multimolecular complex together with CIB, suggesting that the regions in the ␣ IIb amino acid sequence implicated in the interaction with ␤ 3 and CIB are distinct. Since the ␣ IIb membrane-proximal region has been shown to contain the key contact sites for ␤ 3 binding, CIB interaction is likely to involve the ␣ IIb C terminus.
The Membrane-proximal KVGFFKR Domain and the Arg 995 Residue of ␣ IIb Cytoplasmic Tail Are Also Required for Optimal ␣ IIb Binding to CIB-To delineate further the contact site of CIB within the ␣ IIb cytoplasmic tail, we investigated the binding of ␣ IIb (R995A) and ␣ IIb (Asn 996 -Gln 1008 ) peptides to GST-CIB. Interestingly, the sensorgrams obtained with both mutants were markedly reduced, as compared with the binding curve monitored with the wild type ␣ IIb peptide (Fig. 8), although residual binding of ϳ30% persisted for the two mutants independent of the presence or absence of calcium. These results indicate that CIB binding to ␣ IIb does not rely exclusively on the C-terminal part of ␣ IIb and that the KVGFFKR sequence or the Arg 995 residue within ␣ IIb are required for optimal binding.
Calcium-independent Binding of CIB to Intact Platelet ␣ IIb ␤ 3 -To confirm further the calcium-independent interaction of CIB with the ␣ IIb cytoplasmic domain, we performed in vitro liquid phase binding assays using intact ␣ IIb ␤ 3 . For this purpose, an ␣ IIb ␤ 3 -rich glycoprotein fraction was prepared from platelets by ConA affinity chromatography. Similarly, an ␣ v ␤ 3enriched glycoprotein concentrate was prepared from CHO cells expressing human ␣ v ␤ 3 for control experiments. SDS-PAGE and Western blot analysis demonstrated that these samples contained appreciable levels of ␤ 3 integrins (Fig. 9A, lanes  2 and 3). Platelet or CHO-A06 cell glycoproteins were incubated with GST-CIB or GST alone in the presence of 2 mM CaCl 2 or 2.5 mM EGTA. GST fusion proteins were recovered using glutathione-Sepharose beads, and bound proteins were analyzed by SDS-PAGE and Western blotting using anti-␣ IIb , FIG. 7. Interaction of the ␣ IIb ⅐␤ 3 cytoplasmic tail complex with GST-CIB fusion protein. Experiments were carried out on a Biacore X apparatus at a flow rate of 20 l/min. TBS/Bia without CaCl 2 was used as running and dilution buffer. Data are expressed as relative responses. Both ␣ IIb and ␤ 3 peptides were injected at a concentration of 40 M. Sensorgrams were obtained using an antibody-captured GST-CIB chip surface (ϳ1700 -1800 RU) after sequential injection (A) of each of the peptides (curve 1, ␣ IIb followed by ␤ 3 ; curve 2, ␤ 3 followed by ␣ IIb ) or after a single injection (B) of the two peptides previously incubated together for at least 15 min at room temperature with (curve 3) or without 2 mM CaCl 2 (curve 4). Sample injections are indicated by arrows.   ) interacting with antibody-captured GST-CIB fusion protein (1200 -1400 RU). Measurements were carried out on a Biacore X instrument at a flow rate of 50 l/min using TBS/Bia Ϯ 2 mM CaCl 2 as running and dilution buffer. On and off rates (with K D ϭ k off /k on ) were evaluated using Biaevaluation software as described under "Experimental Procedures." Kinetic data presented are the mean values Ϯ S.D. of three separate experiments. Buffer anti-␣ v , anti-␤ 3 , and anti-GST antibodies. As shown in Fig. 9B (panel 1), GST-CIB was able to retain ␣ IIb ␤ 3 independent of the presence or absence of Ca 2ϩ , whereas GST alone was not. In contrast, no ␣ v ␤ 3 integrin was detected in either GST-CIB-or GST-containing samples incubated with the ␣ v ␤ 3 -enriched fraction (Fig. 9B, panel 2). In competitive inhibition experiments, an excess of free ␣ IIb cytoplasmic peptide abrogated ␣ IIb ␤ 3 /GST-CIB coprecipitation (Fig. 9B, panel 3). Taken together, these results demonstrate that CIB binds specifically to intact platelet ␣ IIb ␤ 3 in a Ca 2ϩ -independent manner. CIB Binds Preferentially to Mn 2ϩ -activated ␣ IIb ␤ 3 -Since both inactive and active ␣ IIb ␤ 3 conformers are isolated from platelet lysate by ConA affinity chromatography (30), we determined the activation state of the ␣ IIb ␤ 3 heterodimers present in our ConA-purified platelet extract by examining the binding capacity of ␣ IIb ␤ 3 to immobilized fibrinogen. Fig. 10A shows that specific ␣ IIb ␤ 3 binding was significantly weaker with crude sample than with fractions incubated either with the activating anti-␤ 3 mAb D3GP3 or with 10 mM Mn 2ϩ , demonstrating that the ConA-purified platelet extract contained essentially inactive ␣ IIb ␤ 3 and minor amounts of activated ␣ IIb ␤ 3 .
To investigate further whether Mn 2ϩ -induced activation of platelet ␣ IIb ␤ 3 was able to influence ␣ IIb ␤ 3 /CIB interaction, we performed in vitro binding studies in the presence or absence of MnCl 2 . Interestingly, the binding of ␣ IIb ␤ 3 to GST-CIB was more pronounced in the presence of 10 mM MnCl 2 than in Mn 2ϩ -free buffer (Fig. 10B). Also, ␣ IIb ␤ 3 heterodimers were undetectable in competitive assays using an excess of ␣ IIb cytoplasmic peptide or in control experiments performed with purified GST alone, demonstrating that the specificity of the interaction was not modified by Mn 2ϩ . To confirm further these results, we studied the binding of CIB to platelet ␣ IIb ␤ 3 as a function of Mn 2ϩ by immunoprecipitation experiments. Crude or Mn 2ϩ -treated platelet glycoproteins were first incubated with purified GST-CIB and then with the anti-␣ IIb ␤ 3 complexspecific mAb 10E5 to retain total ␣ IIb ␤ 3 , or with the anti-␤ 3 fibrinogen-mimetic mAb PAC-1 to capture activated ␣ IIb ␤ 3 . The immunoprecipitates were analyzed by SDS-PAGE and Western blotting using anti-␣ IIb , anti-␤ 3 , and anti-GST antibodies. As shown in Fig. 11, ␣ IIb ␤ 3 binding to PAC-1 was weak in the absence of Mn 2ϩ (B, lane 1) and was significantly enhanced in 10 mM MnCl 2 buffer (B, lane 2), demonstrating that the activated ␣ IIb ␤ 3 conformers were increased in the Mn 2ϩ -treated platelet glycoprotein concentrate. In Mn 2ϩ -free buffer, few GST-CIB coimmunoprecipitated with 10E5-captured ␣ IIb ␤ 3 (A, lane 3), whereas GST-CIB was not detectable with PAC-1bound ␣ IIb ␤ 3 , probably on account of the small recovery in ␣ IIb ␤ 3 obtained with this sample, containing a weak proportion of active heterodimers (B, lane 3). Conversely, Mn 2ϩ treatment of the platelet glycoproteins led to a marked increase in GST-CIB coprecipitation in experiments performed with either 10E5 or PAC-1 mAb (A and B, lane 4). This effect was particularly apparent from the level of GST-CIB coprecipitated in 10E5 assays, which was significantly greater in the presence of 10 mM Mn 2ϩ than in Mn 2ϩ -free buffer, although the amount of captured ␣ IIb ␤ 3 remained unchanged (A, lanes 3 and 4). As expected from our previous assays, addition of ␣ IIb cytoplasmic peptides to the incubation mixture almost completely inhibited GST-CIB binding to platelet ␣ IIb ␤ 3 (A and B, lane 5), and no GST was coprecipitated with ␣ IIb ␤ 3 (A and B, lane 6). Taken together, these results demonstrate that CIB binds preferentially to the Mn 2ϩ -treated, activated form of ␣ IIb ␤ 3 integrin. These data further suggest that CIB is unlikely to have a regulatory effect on ␣ IIb ␤ 3 ligand binding function, since its interaction with ␣ IIb ␤ 3 does not stimulate PAC-1 binding to inactive ␣ IIb ␤ 3 nor inhibit Mn 2ϩ -activated ␣ IIb ␤ 3 occupancy by PAC-1. DISCUSSION SPR technology has been successfully used by several authors to analyze the interaction of structural domains of receptor cytoplasmic tails with their intracellular targets, such as the interaction of cytoplasmic tail sorting sequences with the adaptor proteins AP-1, AP-2, and AP-3 (31)(32)(33) or of growth factor receptors with the Src homology 2 (SH2) or the phosphotyrosine-binding (PTB) domains of tyrosine kinases (34,35). SPR has also been used to characterize the interaction of various integrin receptors with extracellular matrix proteins (36 -38), ligand-mimetic antibodies (39,40), or with integrin counter-receptors (41). In this report, we have used SPR technology to study in vitro complexation of the ␣and ␤-cytoplasmic tails of integrin ␣ IIb ␤ 3 , as data from mutagenesis, spectroscopic, and computer modeling studies indicate this heterodimerization FIG. 9. In vitro binding of GST-CIB to platelet ␣ IIb ␤ 3 . A, ConApurified glycoproteins from either nontransfected CHO cell lysate (100 g, lane 1), human ␣ v ␤ 3 -expressing CHO-A06 cell lysate (100 g, lane 2), or ␣ IIb ␤ 3 -rich platelet extract (25 g, lane 3) were resolved by 7.8% reducing SDS-PAGE (upper panel) and subjected to Western blot analysis using specific anti-␤ 3 (4D10G3), anti-␣ IIb (S1.3), or anti-␣ v (VNR 139) mAbs (lower panel). Considering the chemiluminescence signal obtained using an equivalent dilution of anti-␤ 3 mAb and the same film exposure time, the level of ␣ IIb ␤ 3 in the platelet glycoprotein extract was estimated to be ϳ4 times higher than that of ␣ v ␤ 3 in the CHO-A06 glycoprotein fraction. The slight difference in the electrophoretic mobility of ␣ IIb and ␣ v is not apparent in this mini-gel. B, ConA-purified glycoproteins from platelet lysate (250 g, panel 1) or ␣ v ␤ 3 -expressing CHO-A06 cell lysate (1 mg, panel 2), both containing approximately equal amounts of ␤ 3 integrins, were incubated with 50 g of purified GST or GST-CIB in the presence of 2 mM CaCl 2 (ϩ) or 2.5 mM EGTA (Ϫ). Binding inhibition experiments were performed by adding 25 nmol of ␣ IIb cytoplasmic peptide to the platelet glycoproteins/GST-CIB or GST mixture (panel 3). Proteins were captured onto glutathione-Sepharose beads and were analyzed by 7.8% reducing SDS-PAGE and Western blot. The membrane was first probed with anti-␤ 3 (4D10G3) mAbs combined with either (panels 1 and 3) anti-␣ IIb (S1.3) or (panel 2) anti-␣ v (VNR 139) mAbs (upper panel), stripped and reprobed with anti-GST antibodies (lower panel). The weak band observed in some GST immunoblot analysis corresponds to an unidentified protein distinct from GST-CIB, as demonstrated by slightly different electrophoretic mobilities. (14 -16, 19). SPR provided a direct and more informative approach to investigate ␣⅐␤ complexation, since continuous real time monitoring of the interaction allowed a detailed characterization of specificity, affinity, and kinetics. Our experimental data are consistent with a 1:1 interaction that proceeds transiently according to affinity constants (K D ) in the micromolar range. These affinities are low when compared with those obtained by SPR for receptor cytoplasmic tails (32,33,35) or signaling molecules (42)(43)(44)(45) which usually range from 10 Ϫ7 to 10 Ϫ9 M. As our kinetic data are close to the limits specified for the Biacore TM instruments (46), they should be considered as an indication of an order of magnitude rather than definite numerical data. Nevertheless, the affinities found here are in agreement with transient and potentially easily modulatory processes, such as those expected intuitively in signal transduction pathways, and predict that the interactions actually proceed in vivo if the binding partners are locally concentrated. This is easily conceivable for integrin ␣ and ␤ cytoplasmic tails which are maintained in close proximity at the cytoplasmic face of the plasma membrane through tight interactions of the integrin extracellular domains.
In contrast to previous studies, in which ␣ or ␤ cytoplasmic tail peptides were linked to an helical coiled-coil motif mimicking the transmembrane domain (14,20), the ␣⅐␤ intersubunit binding shown here occurs in the absence of any additional structural motifs. Furthermore, no homodimerization of ␤ 3 peptides was observed in our study, although this process has been recently described in cells transfected with a ␤ 3 cytoplasmic tail connected to an ␣-helical domain (47), suggesting that transmembrane domains are necessary for integrin oligomerization but not for ␣⅐␤ cytoplasmic tail complexation. Interestingly, our data demonstrate that the S752P substitution in ␤ 3 did not affect ␣⅐␤ heterodimerization, indicating that the Cterminal part of the ␤ 3 cytoplasmic tail is not involved in this process. As the ␤ 3 (S752P) mutation has previously been shown to disrupt bidirectional signaling in platelets and transfected CHO cells (28,29), our results suggest that this defect cannot be attributed to an alteration in ␣⅐␤ cytoplasmic tail interaction but rather to a disruption in a specific ␤ 3 interaction with intracellular regulatory molecules. Actually, the ␤ 3 (S752P) mutation has been shown to markedly reduce ␤ 3 -endonexinspecific binding to ␤ 3 and to impair its modulating effect on the ␣ IIb ␤ 3 affinity state (8,48). Our data further indicate that the ␣ IIb membrane-proximal region is critical for ␣⅐␤ heterodimer assembly, as both ␣ IIb (R995A) and ␣ IIb (Asn 996 -Gln 1008 ) peptides failed to interact efficiently with the ␤ 3 cytoplasmic tail. Our data apparently contradict previous biophysical data which demonstrated that the interactive sites involved in ␣⅐␤ dimerization are located within the ␤ 3 Ile 721 -Asp 740 and ␣ IIb Pro 999 -Gln 1008 sequences (16). However, it should be noted that their data were primarily derived from terbium luminescence spectroscopy experiments which did not allow investigation of an interaction between the N-terminal regions of the ␣ IIb and ␤ 3 cytoplasmic tails, since neither ␣ IIb Leu 985 -Pro 998 nor ␤ 3 were able to bind Tb 3ϩ ions necessary for the generation of a specific fluorescence energy transfer signal. In contrast, our data are in good agreement with previous mutagenesis studies (15) and also indicate that the ␣IIb KVGFFKR motif is necessary but not sufficient to support ␣⅐␤ heterodimerization. Indeed, alterations within this sequence did not totally abolish the ␣ IIb ⅐␤ 3 interaction, suggesting that binding sites distinct from the ␣ IIb membrane-proximal domain are involved in ␤ 3 engagement. These additional contact sites are probably located within the ␣ IIb C-terminal acidic tail. We have also provided direct evidence that Ca 2ϩ and Mg 2ϩ FIG. 10. In vitro binding of ConA-purified ␣ IIb ␤ 3 to immobilized fibrinogen and GST-CIB as a function of MnCl 2 . A, various concentrations of crude ConA-purified platelet glycoproteins were incubated for 4 h at room temperature with 2 g/ml D3GP3 mAb (OE), 10 mM MnCl 2 (f), or buffer alone (q) in 96-well microtiter plates coated with 5 g/ml human fibrinogen. After washing, bound ␣ IIb ␤ 3 was detected by ELISA using rabbit anti-␣ IIb ␤ 3 antibodies as stated under "Experimental Procedures." The figure is representative of two independent experiments, each data point corresponding to the mean of duplicates. B, ConA-purified platelet glycoproteins (250 g) were incubated with 50 g of purified GST-CIB or GST in the presence (ϩ) or absence (Ϫ) of 10 mM MnCl 2 . Binding inhibition experiment was performed using 25 nmol of ␣ IIb cytoplasmic peptide. Proteins were recovered using glutathione-Sepharose beads and were analyzed by 7.8% reducing SDS-PAGE and by Western blot. The membrane was first probed with anti-␤ 3 (4D10G3) and anti-␣ IIb (S1.3) mAbs (upper panel), stripped and reprobed with anti-GST antibodies (lower panel). Two nonspecific bands of high electrophoretic mobility were observed with the GST-CIB-containing samples. are not required for ␣ IIb ⅐␤ 3 complexation but rather stabilize the heterodimeric structure by reducing the dissociation rate. Since ␣⅐␤ association proceeded with the same extent, independent of the presence of cations, the cation coordination sites within ␣ IIb and ␤ 3 cytoplasmic tails are likely to be distinct from those involved in intersubunit binding, namely the membrane-proximal region of each integrin subunit. Interestingly, our study indicates that the ␣ IIb membrane-proximal region is involved in divalent cation-induced ␣⅐␤ complex stabilization, since an alteration in the KVGFFKR sequence impaired the stability of the heterodimers. In support of these results, a functional cation binding domain has previously been mapped to the negatively charged acidic C-terminal region of ␣ IIb (residues 999 -1008) and was found to bind divalent cations in coordination with sites located in the ␣ IIb 985-998 sequence (16,19). Based on the structural model of ␣ IIb ␤ 3 , Haas and Plow (19) speculated that a cation coordination site rearrangement could occur upon ␣ IIb ⅐␤ 3 cytoplasmic tail complexation. The faster complex dissociation observed here in the absence of cations actually suggests that additional intra-and/or intersubunit cation coordination site(s) probably appear(s) during the complexation resulting in a more stable structure.
Because CIB is the only intracellular protein identified so far that specifically interacts with the ␣ IIb integrin subunit cytoplasmic tail, we used recombinant CIB as a reporter protein to monitor the ligand binding capacity of ␣ IIb either as a monomer or as a heterodimer in association with ␤ 3 . Our data clearly demonstrate that CIB interacts with the ␣ IIb peptide in a one-to-one, weak affinity reaction (K D ϭ 12 M) and also binds to the preformed ␣ IIb ⅐␤ 3 cytoplasmic complex, suggesting that the contact sites within the ␣ IIb amino acid sequence involved in the interaction with CIB are distinct from those engaged in ␤ 3 binding. As CIB has been shown to interact only with ␣ IIb and not with ␣ v , ␣ 2 , or ␣ 5 in the yeast two-hybrid system, Naik et al. (12) concluded that CIB was unlikely to bind to the highly conserved GFFKR motif common to all ␣ subunits but rather to the highly acidic C-terminal part of the ␣ IIb cytoplasmic tail. Our data, however, provide evidence that the KVGFFKR sequence is necessary for optimal CIB⅐␣ IIb interaction, since both ␣ IIb (R995A) and ␣ IIb (Asn 996 -Gln 1008 ) peptides failed to interact efficiently with CIB. In the three-dimensional model of ␣ IIb ␤ 3 , the negatively charged C terminus of the ␣ IIb cytoplasmic tail is predicted to fold back onto itself and to interact with the positively charged N terminus through several side chain and backbone contacts (19). Thus, it is conceivable that the deletion of the KVGFFKR sequence or the R995A substitution disrupts some of the intrasubunit contacts that stabilize the ␣ IIb conformation, leading to an alteration in the capacity of the cytoplasmic tail peptide to bind to CIB. Interestingly, although CIB has two conserved EF-hand motifs within its protein structure, and has been found to bind Ca 2ϩ in vitro (12), our data demonstrate a Ca 2ϩ -independent binding of CIB to the ␣ IIb cytoplasmic peptide. This finding indicates that the potential Ca 2ϩ -dependent regulatory function of CIB, based on its sequence homology with calmodulin and calcineurin B (12), is not involved in the association with ␣ IIb but rather in the activity or in the binding to an as yet unidentified CIB-associated protein.
Most strikingly, when binding studies were performed with intact ␣ IIb ␤ 3 , CIB bound preferentially to Mn 2ϩ -activated ␣ IIb ␤ 3 , suggesting that the accessibility of the CIB-binding site within ␣ IIb is increased in active ␣ IIb ␤ 3 conformers. Since Mn 2ϩ did not affect the binding characteristics of the GST-CIB⅐␣ IIb cytoplasmic peptide complex in SPR studies (data not shown), our results suggest that Mn 2ϩ activation of intact ␣ IIb ␤ 3 induces a conformational change that is transmitted from the ␣ IIb ␤ 3 extracellular domains to the cytoplasmic tails. Such long range transmembrane structural alterations have been anticipated from studies showing that specific epitopes are exposed or disappear within the cytoplasmic tails of ␣ IIb and ␤ 3 subunits following ␣ IIb ␤ 3 activation or ligand occupancy (49,50). Alternatively, an increase in ␣ IIb ␤ 3 avidity for CIB cannot be excluded, since the fibrinogen-mimetic mAb PAC-1 used to capture active ␣ IIb ␤ 3 is a multimeric IgM antibody and is thus likely to trigger oligomerization of ␣ IIb ␤ 3 complexes that mimic integrin clustering (51). Finally, our data provide evidence that CIB is unlikely to have a regulatory effect on ␣ IIb ␤ 3 ligand binding function, since its interaction with ␣ IIb does not trigger ligand binding to inactive ␣ IIb ␤ 3 nor inhibit activated ␣ IIb ␤ 3 occupancy by a ligand, suggesting that CIB is most likely involved in ␣ IIb ␤ 3 post-receptor occupancy events.