Disruption of the β₃ 663-687 disulfide bridge confers constitutive activity to β₃ integrins

The glycoprotein (GP) IIb-IIIa complex, integrin α_IIbβ₃, is a calcium-dependent, noncovalent heterodimer formed by GPIIb and GPIIIa. This complex is found in the plasma membrane of megakaryocytes, platelets, and some tumor tissues^1-3  and functions as a receptor for fibrinogen and other adhesive proteins like the von Willebrand factor, fibronectin, or vitronectin.⁴  The β₃ subunit may also complex the GP α_v to form the vitronectin receptor (integrin α_vβ₃) that shares with α_IIbβ₃ the binding of fibrinogen although with different affinity.⁵  The platelet α_IIbβ₃ complex is essential to maintain a normal hemostasis. Unlike other platelet receptors that are constitutively active, the α_IIbβ₃ is maintained in a low-affinity state for its ligands. Disruption of the vascular endothelium and exposure of platelets to the action of agonists and adhesive proteins from the subendothelial matrix induces a cellular activation. The activated cells interact with adhesive proteins from the extracellular matrix,^6,7  and the α_IIbβ₃ receptors are able to bind fibrinogen with high affinity (inside out signaling), resulting in platelet aggregation.⁸  Conversely, ligand-bound α_IIbβ₃ propagates signals to the interior of the cell (outside-in signaling) leading to enhanced interaction with the cytoskeleton, clustering of receptors (increased ligand avidity), and formation of focal contacts rich in signaling complexes.^9,10  The agonist-induced increase in ligand affinity of α_IIbβ₃ is thought to be the result of conformational changes of the heterodimer^11-13  initiated by the interaction of the cytoplasmic tails of α and β₃ subunits with cytosolic proteins. Despite the pathophysiologic importance of the platelet α_IIbβ₃ receptor, the knowledge of the mechanisms controlling its state of activation is rather limited.

Unlike previous reports,¹⁴  recent work from our laboratory¹⁵  showed that a truncated form of β₃ lacking the transmembrane and cytosolic domains failed to associate with α_IIb. The present work was aimed at further investigating the role played by the carboxy-terminal domain of β₃ in the surface expression and function of β₃ heterodimers. The results obtained in this study indicate that surface expression of α_IIbβ₃ could not occur in the absence of the transmembrane domain of β₃. The present study has also revealed that either deletion of the carboxy-terminal region of the β₃ ectodomain or disruption of the 663-687 disulfide bridge confers constitutive activity to the β₃ integrins, suggesting that this region is involved in maintaining the α_IIbβ₃ receptor in a resting, nonactivated state.

Restriction enzymes were obtained from Boehringer (Mannheim, Germany). The pcDNA3 expression vector was from Invitrogen (San Diego, CA). Oligonucleotides were synthesized by Invitrogen (Paisley, Scotland). Most other reagents were purchased from Sigma Chemical (St Louis, MO), Merck (Darmstadt, Germany), or Calbiochem-Novabiochem (San Diego, CA). Monoclonal antibodies (mAbs) anti-α_IIb (2bc1) and anti-β₃ (H1a) were obtained in our laboratory; the anti-α_vβ₃ MAB1976 was purchased from Chemicon (Temecula, CA), and fluorescein isothiocyanate (FITC)-PAC-1 was obtained from Becton Dickinson (San Jose, CA). WOW-1 F(ab′)₂ and PAC-1 F(ab′)₂ were obtained from Dr S. Shattil (The Scripps Research Institute, La Jolla, CA).

Wild β₃-cDNA cloned in EcoRI and HindIII sites of pBluescript II-KS (Stratagene, La Jolla, CA) was digested with XhoI and partially with BamHI, and subcloned in pcDNA3. pcDNA3-β₃Δ616 construct was prepared as described previously.¹⁵  To generate truncated β₃ at 638, 657, 675, and 693 amino acids, stop codons were introduced by polymerase chain reaction (PCR) using pcDNA3-β₃ as template, and oligonucleotide β₃-sense (1670-1689): 5′-ACTGCAACTGTACCACGCGT-3′ and the antisense primers β₃-AS-638-stop-XhoI: 5′-TATCTCGAGCTAGTCACGGCAG-3′, β3-AS-657-stop-XhoI: 5′-TATCTCGAGTTAGGTACAATTCAC-3′, β3-AS-675-stop-XhoI: 5′-TAACTCGAGTCAACTAGAATCTTC-3′, and β3-AS-693-stop-XhoI: 5′-TATCTCGAGTTAGTCAGGGCCC-3′, respectively. Mutations introduced to generate stop codons are underlined. PCR products were digested with NotI and XhoI to replace the normal fragment from pcDNA3-β₃.

cDNAs encoding [608Ala]β₃, [614Ala]β₃, [617Ala]β₃, [663Ala]β₃, and [687Ala]β₃ mutants were prepared by the splicing by overlap extension PCR procedure using the oligonucleotide primers β₃-sense (1670-1689) and β₃-AS-(2319-2296)-XhoI: 5′-TGATAATGACTCGAGGATGACTGC-3′, and the overlapped pairs of the following primers: [608Ala]β₃-S: 5′-CCAGATGCCGCCACCTTTAAG-3′ and [608Ala]β₃-AS: 5′-CTTAAAGGTGGCGGCATCTGG-3′; [614Ala]β₃-S: 5′-AAGAAAGAAGCTGTGGAGTGT-3′ and [614Ala]β₃-AS: 5′-ACACTCCACAGCTTCTTTCTT-3′; [617Ala]β₃-S: 5′-TGTGTGGAGGCTAAGAAGTTTGAC-3′ and [617Ala]β₃-AS: 5′-GTCAAACTTCTTAGCCTCCACACA-3′; [663Ala]β₃-S: 5′-GAGGATGACGCTGTCGTCAGA-3′ and [663Ala]β₃-AS: 5′-TCTGACGACAGCGTCATCCTC-3′; [687Ala]β₃-S: 5′-GAGCCAGAGGCTCCCAAGGGC-3′ and [687Ala]β₃-AS: 5′-GCCCTTGGGAGCCTCTGGCTC-3′. The [608Ala]β₃, [614Ala]β₃, [617Ala]β₃, and [663Ala]β₃ cDNAs were used as template to generate [608Ala/655Ala]β₃, [614Ala/635Ala]β₃, [617Ala/631Ala]β₃, and [663Ala/687Ala]β₃ mutant cDNAs, respectively, and the overlapped pairs of primers: [655Ala]β₃-S: 5′-GCAGTGAATGCTACCTATAAG-3′ and [655Ala]β₃-AS: 5′-CTTATAGGTAGCATTCACTGC-3′; [635Ala]β₃-S: 5′-AACCGTTACGCCCGTGACGAG-3′ and [635Ala]β₃-AS: 5′-CTCGTCACGGGCGTAACGGTT-3′; [631Ala]β₃-S: 5′-GACGAAAATACCGCCAACCGTTAC-3′ and [631Ala]β₃-AS: 5′-GTAACGGTTGGCGGTATTTTCGTCATG-3′; and [687Ala]β₃-S and [687Ala]β₃-AS, respectively. Final PCR products carrying the desired mutations were digested with XhoI and MluI and ligated into pcDNA3 carrying normal β₃ cDNA digested with XhoI and partially digested with MluI.

β₃(685-690del), β₃(654-690del), and β₃(616-690del) were obtained by PCR amplification from the plasmid pBJ1-β₃, with antisense primers located just before the region to delete: β₃-AS-(2055-2030): 5′-TGGCTCTTCTACCACATACGGATGG-3′, β₃-AS-(1959-1938): 5′-CACTGCATCCTTGCCAGTGTCC-3′ and β₃-AS-(1845-1819): 5′-CACACATTCTTTCTTAAAGGTGCAGGC-3′, respectively, and the sense primer β₃-S-(2071-2092): 5′CCTGACATCCTGGTGGTCCTGC-3′, located after the fragment to be removed. The PCR products were phosphorylated and ligated to obtain pBJ1-β₃(685-690del), pBJ1-β₃(654-690del), and pBJ1-β₃(616-690del) that were used as template for another PCR performed with oligonucleotides β₃-S-(1670-1689) and β₃-AS-(2319-2296)-XhoI. The resulting PCR products were digested with NotI and XhoI to replace the normal fragment from pcDNA3-β₃.

A β₃-β₇ chimera was obtained as follows. The 2078-2323 fragment of β₇ was amplified by PCR with the oligonucleotides β₇-S-(2078-2095): 5′-ACACCATGGCTAGCACACCGGGAC-3′ and β₇-AS-(2323-2298): 5′-CGTGTGGTATCGATCCTTTTCTTGG-3′, containing a NheI and a ClaI site, respectively. The PCR product was digested with NheI and ClaI and ligated to pcDNA3-β₃ where the homologous fragment had been deleted. For this, NheI and ClaI were previously introduced in the β₃ sequence using the overlapped oligonucleotide primers β₃-S (2137-2160)-ClaI: 5′-TGTCCCAAGGATCGATACATCCTG-3′ and β₃-AS-(2160-2137)-ClaI: 5′-CAGGATGTATCGATCCTTGGGACA-3′. The cDNA so obtained was used as a template to introduce the NheI site with the following overlapped oligonucleotides: β₃-S (1918-1939)-NheI: 5′-TGTGTGGAGCTAGCGAAGTTTGAC-3′ and β₃-AS (1939-1918)-NheI: 5′-GTCAAACTTCGCCTAGCTCCACACA-3′. The final PCR product carrying the NheI and ClaI sites was digested with XhoI and MluI and ligated into pCDNA3-β₃ cDNA digested with XhoI and partially digested with MluI. The DNA sequence of each construct was verified. The α_IIb cDNA was cloned into the HindIII site of pcDNA3, as previously described.¹⁶ 

CHO cells were grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS). Cells were transiently cotransfected by the diethylaminoethyl (DEAE)-dextran method¹⁷  with normal or mutated pcDNA3-β₃ constructs and, when indicated, with pcDNA3-α_IIb. Forty-eight hours after transfection the cells were harvested and the surface expression of α_IIbβ₃ complexes was assessed by flow cytometry analysis or immunoprecipitation.

CHO cell lines stably expressing β₃ or β₃(616-690del) were established by transfection with 15 μg pcDNA3-β₃ or pcDNA3-β₃(616-690del), using the calcium phosphate precipitation procedure. The transfected cells were fed with medium containing 800 μg/mL G-418 every 3 to 4 days, and clones of cells stably expressing β₃ were selected by flow cytometry analysis. Cells stably expressing β₃ or β₃(616-690del) were transfected with pcDNA3-α_IIb, and cells expressing α_IIb were cloned by cell sorting with fluorescence-activated cells sorting (FACS Vantage; Becton Dickinson) with argon laser tuned to 488 nm of excitation and specific fluorescence signals collected by a 530-bp filter.

The 96-well flat-bottomed plates were coated with 10 μg/mL fibrinogen in phosphate-buffered saline (PBS) pH 7.2 (100 μL/well) for 2 hours at 37°C and then washed with PBS, blocked with 200 μL 1% bovine serum albumin (BSA) for 1 hour at 37°C, and washed with PBS. CHO cell lines were labeled with 10 μM calcein-am (Molecular Probes, Leiden, The Netherlands) for 10 minutes, washed by centrifugation, and resuspended in serum-free DMEM, plated onto dishes in duplicate at a density of 4 × 10⁴ cells/well and incubated for different periods of time at 37°C. Nonadherent cells were removed by carefully washing twice with 200 μL PBS. Cell adhesion was quantitated by cytofluorometry. Basal adherence to BSA (cell binding to BSA-coated wells was always < 1%) was subtracted from attachment values. To study adhesion of cell lines to vitronectin, cytomatrix cell adhesion strips with human vitronectin (Chemicon) were used following the manufacturer's recommendations. At the end of the assays, adherent cells were examined with a phase-contrast Nikon TMS microscope (Barcelona, Spain) and micrographs were taken with a digital camera.

CHO cells were incubated with 1 mg/mL fibrinogen for 15 minutes at room temperature at a density of 2.5 × 10⁶ cells/mL. Then, 250 μL cell suspension was plated onto wells of a 24-well culture dish precoated with 1 mg/mL BSA. Ten to 15 minutes after plating, cell aggregates were examined by phase-contrast microscopy as described. In some cases, cells were preincubated with 10 μg anti-β₃ antibody H1a or 10 mM EDTA (ethylenediaminetetraacetic acid).

Transfected cells were harvested using 0.5 mM EDTA in PBS, washed with PBS, resuspended at a density of 10⁶ cells/100 μL, and incubated for 20 minutes at 4°C with mAbs specific to α_IIb (2bc1), β₃ (H1a), or α_vβ₃ (MAB1976). Next, cells were washed and exposed to FITC-F(ab′)₂ fragment of rabbit antimouse Ig (Dako, Glostrup, Denmark) at 4°C for 20 minutes, and the surface fluorescence was analyzed in a Coulter flow cytometer, model EPICS XL (Miami, FL).

Human fibrinogen was labeled with FITC as described previously.¹⁶  CHO cells (5 × 10⁵) coexpressing normal or mutated β₃, and either human α_IIb or endogenous α_v were resuspended in Tyrode buffer (5 mM HEPES [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid], 2 mM MgCl₂, 0.3 mM NaH₂PO₄, 3 mM KCl, 134 mM NaCl, 12 mM NaHCO₃, 0.1% glucose, 0.1% BSA, and 1 mM CaCl₂, pH 7.0) with 10 μg FITC-fibrinogen and incubated for 25 minutes at room temperature. After washing, cells were resuspended and analyzed by flow cytometry. When indicated, cells were preincubated for 30 minutes at 4°C with mAbs directed against activated α_vβ₃ (WOW-1) or activated α_IIbβ₃ (PAC-1), 2 mM MnCl₂, or 10 mM EDTA.

To assess the binding of PAC-1, cells were resuspended in Tyrode buffer, incubated at room temperature for 15 minutes with FITC-PAC-1, and then cell fluorescence was determined by flow cytometry.

Metabolic labeling of transfected CHO cells was performed 48 hours after transfection. First, cells were incubated in medium without methionine and cysteine for 30 minutes and, then, with [³⁵S]-methionine/cysteine (400 μCi/mL; 14.8 MBq) for 3 hours. Cells were washed 3 times with PBS and extracted in 0.5 mL lysis buffer (50 mM Tris [tris(hydroxymethyl)aminomethane], pH 7.4, 150 mM NaCl, 2 mM phenylmethylsulfonyl fluoride [PMSF], 1% Triton X-100, 0.05% Tween 20, and 0.03% sodium azide). Precleared lysates were immunoprecipitated with either anti-β₃ or anti-α_IIb. The immunoprecipitates were then bound to protein A-Sepharose CL-4B beads (Amersham Pharmacia Biotech, Uppsala, Sweden), washed, and eluted by incubating 10 minutes at 100°C in 50 μL reducing loading buffer (50 mM Tris-HCl, pH 6.8, 10% glycerol, 5% β-mercaptoethanol, 4% sodium dodecyl sulfate [SDS], and 0.002% bromophenol blue) and, finally, resolved by electrophoresis. The gels were vacuum dried and exposed to hypersensitive x-ray film for 48 hours.

We have recently reported a nonsense mutation producing a truncated form of β₃ (β₃Δ616) associated with thrombasthenic phenotype (type I Glanzmann thrombasthenia).¹⁵  The underlying molecular mechanism for the thrombasthenic phenotype was the failure of β₃Δ616 to complex α_IIb. This observation appeared to be in conflict with previous reports in which β₃-truncated proteins complex α subunits^14,18,19  and the mutated complexes reached the cell surface.¹⁴  Because the apparent discrepancy could be due to differences in the lengths of the deleted fragments, we performed a transient transfection analysis of progressive carboxy-terminal or internally deleted forms of β₃. Figure 1 depicts the deletion mutants used for transfection analysis (Figure 1A) as well as some structural features of the carboxy-terminal ectodomain of β₃ (Figure 1B) based on recent studies on the crystal structure of heterodimeric α_v and β₃ ectodomains.²⁰ 

Whether the carboxy-terminal-deleted constructs were transfected alone or with α_IIb, no detectable surface expression of β₃ or β₃ heterodimers was appreciated by flow cytometry (Figure 2). In contrast, deletion of the 616-690 carboxy-terminal region of the β₃ ectodomain (616-690del) did not prevent the surface expression of α_vβ₃ or α_IIbβ₃ heterodimers. According to these observations, a normal level of surface exposure of these receptors demands the presence of the transmembrane and cytosolic regions of β₃.

Immunoprecipitation analysis of metabolically labeled cells cotransfected with α_IIbβ₃ were performed (Figure 3). When anti-β₃ was used, immunoprecipitates of the expected size for the normal and mutated forms of β₃ were obtained. All the anti-β₃ immunoprecipitates contain products with apparent molecular size of pro-α_IIb, whereas mature heavy chains of α_IIb were detected only in cells transfected with either normal β₃ or β₃(616-690del). When anti-α_IIb was used, pro-α_IIb was precipitated in all cases, and α_IIb and β₃ were only coprecipitated in cells transfected with either normal β₃ or β₃(616-690del). Because the proteolytic cleavage of pro-α_IIb occurs only when the subunit is complexed to β₃, the absence of the mature heavy chain of GPIIb suggests absent or very low rate of pro-α_IIb-β₃ complex formation. Immunoprecipitation of concentrated 48-hour-conditioned medium with either anti-β₃ or anti-α_IIb did not yield detectable amounts of truncated β₃. Thus, Δβ₃ forms are not secreted to the extracellular medium, either alone or associated with α heavy chain, at a significant rate; however, it cannot be excluded that the Δβ₃ or the α_IIbΔβ₃ complexes could have been rapidly degraded.

CHO cells expressing recombinant α_IIbβ₃ receptors adhere spontaneously to solid-phase fibrinogen. Thus, we found it of interest to examine whether the mutant α_IIbβ₃(616-690del) complexes exhibited a distinct capacity of adhesion onto ligand-coated plates. The CHO-α_IIbβ₃(616-690del) cells were larger than cells expressing normal complexes and showed enhanced fibrinogen concentration-dependent adherence onto ligand-coated plates (Figure 4A). Because CHO cells express endogenous α_v, we stably transfected normal β₃ or β₃(616-690del) and studied the adherence of normal CHO-α_vβ₃ and CHO-α_vβ₃(616-690del) cells to plates coated with either vitronectin (Figure 4B) or fibrinogen (Figure 4C). In both conditions the cells expressing the mutant complexes showed enhanced adherence to coated plates.

Normal resting CHO-α_IIbβ₃ cells in suspension do not bind to soluble fibrinogen at significant rates and, therefore, as expected, no spontaneous aggregation was observed (Figure 5). In contrast, CHO-α_IIbβ₃(616-690del) cells in suspension underwent receptor and soluble fibrinogen-dependent spontaneous aggregation and this effect was prevented by pretreatment with either EDTA or an anti-β₃ function blocking mAb (Figure 5). The fibrinogen-dependent aggregation of CHO-α_IIbβ₃(616-690del) cells suggested that the mutant receptor was constitutively active. To further investigate this point, we studied the spontaneous binding to soluble FITC-labeled fibrinogen of cells expressing either normal or mutated complexes. As shown in Figure 6, CHO-α_IIbβ₃(616-690del) but not CHO-α_IIbβ₃ cells bound to soluble fibrinogen. Fibrinogen (Fg) is a ligand for both α_IIbβ₃ and α_vβ₃. Thus, to assess the binding of FITC-Fg to each integrin, the cells were incubated with function-blocking, activation-dependent mAbs directed against either α_IIbβ₃ (PAC-1) or α_vβ₃ (WOW-1). As observed in Figure 6, the binding of FITC-Fg was partially prevented by both WOW-1 and PAC-1. In agreement with their different sites of action, the combined administration of WOW-1 and PAC-1 showed additive effects, so that the binding of cells to FITC-Fg was virtually abolished (data not shown). The presence of Mn²⁺ enabled CHO-α_vβ₃ or CHO-α_IIbβ₃ cells to bind to soluble fibrinogen. Cells expressing the β₃(616-690del) subunit showed spontaneous binding to FITC-Fg that was further stimulated by Mn²⁺ (Figure 6 lower panels). In all conditions, the binding of cells to FITC-Fg was completely abolished by preincubation with EDTA. We next tried to determine whether β₃ subunits containing deletions shorter than 616-690del, such as 654-690del or 685-690del, would confer receptor activity. All the constructs yielded normal rates of cell surface expression of α_IIbβ₃ complexes (Figure 7A). The cells carrying α_IIbβ₃ receptors with internally deleted β₃ subunit, even one as small as 5 residues, β₃(685-690del), exhibited enhanced binding to PAC-1 and FITC-Fg (Figure 7B-C). EDTA prevented the binding of cells to FITC-Fg (Figure 7C).

A recent crystallographic study on the α_vβ₃ integrin described a new fold in the carboxy-terminal end 606-690 of the β₃ ectodomain, named β tail domain (βTD), that shows a weak homology to the cytostatin fold and contains 4 disulfide bridges (608Cys-655Cys, 614Cys-635Cys, 617Cys-631Cys, and 663Cys-687Cys).²⁰  Since the smaller deletion conferring receptor activity, β₃(685-690del), disrupts the 663-687 disulfide bridge, we next investigated the functional effect of disrupting each of the disulfide bridges of the βTD, by replacing the cysteine residues by alanine. None of these mutations altered the normal levels of surface expression of α_IIbβ₃ complexes (Figure 8A); however, disruption of the 663-687 disulfide bridge enabled the cells to bind spontaneously to soluble fibrinogen and PAC-1 (Figure 8B-C). To further verify the role of this disulfide bridge, we mutated only the 687C>A. The β₃(685-690del) construct as well as [663Ala-687Ala]β₃ and [687Ala]β₃ mutants were less efficient than β₃(616-690del) in enhancing the receptor activity (approximately 50%). The latter observation suggested that both shortening of the carboxy-terminal ectodomain of β₃ and disulfide bridge disruption could have contributed to confer constitutive activation to β₃(616-690del). To further investigate this point, we prepared a chimeric β₃ subunit in which the 616-690 fragment was replaced with the homologous domain of β₇ that lacks the cysteines found in β₃ at positions 663 and 687. Figure 8B-C shows that heterodimers of the β₃-β₇ chimera bound spontaneously to PAC-1 and fibrinogen to a similar extent as the 663Ala-687Ala and 687Ala β₃ mutants, that is, less than β₃(616-690del). This observation supports the idea that both disruption of the 663Cys-687Cys disulfide bridge and shortening of the carboxy-terminal ectodomain enhance the activity of β₃ heterodimers by different yet additive mechanisms. Although the importance of the βTD in maintaining the basal, nonactive state of the β₃ receptors is evident under our experimental conditions, the physiologic significance of these findings is difficult to ascertain. The possibility exists that a normal mechanism of receptor activation could involve the sliding of the subunits with the result of a relative shortening of the β₃. On the other hand, molecular genetic defects in the βTD of β₃ might be associated with constitutive active β₃ receptors and subsequent pathologic disorders.

We have recently reported a case of Glanzmann thrombasthenia associated with a point mutation that changed Glu616 to a termination codon resulting in a truncated β₃ protein that lacks the transmembrane and cytoplasmic regions.¹⁵  The lack of surface exposure of α_IIbβ₃Δ616 was caused by a failure of the truncated protein to complex α_IIb. This observation appeared to be in conflict with a previous report indicating that a truncated β₃ lacking the transmembrane and cytoplasmic domains (β₃Δ693) allowed assembly and surface exposure of α_IIbΔβ₃ complexes.¹⁴  Moreover, truncated forms of recombinant α_IIb and β₃ seem to form soluble Δα_IIb-Δβ₃ heterodimers capable of binding to ligands.²¹  To determine whether the apparent discrepancy could be due to the extension of the β₃ truncation in each experimental condition, we analyzed the ability of β₃-truncated (Δβ₃) forms of different lengths to complex α_IIb. We failed to demonstrate association of any of the Δβ₃ forms tested with α_IIb. The proteolytic cleavage of pro-α_IIb to yield α_IIb heavy (α_IIbH) and light (α_IIbL) chains requires a previous association of α and β subunits; therefore, the lack of coprecipitation of α_IIbH with Δβ₃ indicates that pro-α_IIb failed to form stable complexes with any of the Δβ₃ forms analyzed. This assertion agrees with the observation that the Iraqi-Jewish thrombasthenic phenotype associated with the β₃Δ650 mutation shows detectable platelet pro-α_IIb²²  and absence of platelet α_IIbβ₃ and α_vβ₃ receptors^23,24 ; moreover, the recombinant β₃Δ657 does not complex α_IIb.²⁵  However, truncated β₃ preserving the transmembrane domain and 1 or 8 of the 47 cytoplasmic amino acids have been reported to reach the cell surface.^26,27  On the other hand, the α_IIb and β₃ ectodomains can form heterodimers that bind ligands and exhibit immunochemical properties similar to the full-length receptor inserted in the plasma membrane.^21,28  On these grounds, it seems plausible to conclude that the cell surface exposure of detectable levels of α_IIbβ₃ demands the presence of the transmembrane region of β₃.

Platelet α_IIbβ₃ receptors, about 80 000/cell,⁸  as well as their main ligands, fibrinogen and von Willebrand factor, are present in blood at high concentrations. Thus, to prevent intravascular aggregation, the receptors must be maintained in a low-affinity state for their ligands. The activation of platelets by physiologic agonists leads to increased fibrinogen binding to α_IIbβ₃ and the ligand-bound receptors propagate signals into the cell resulting in cytoskeletal organization and receptor clustering.^8-10  Domains within the cytoplasmic tails of α and β subunits seem to be involved in the bidirectional signaling through α_IIbβ₃.^29,30  The precise molecular mechanisms of receptor activation have not been yet elucidated, although early studies on this matter postulated that ligand affinity changes could be the result of conformational changes.¹¹  In agreement with this postulate, recent crystallographic studies of the ectodomains of β₃ revealed that the binding of the ligand mimetic RGD peptide is associated with precise conformational changes^31-34  and that clasping of the α and β tails prevents the integrin activation.³⁵  Moreover, analysis of peptide fragments from resting and activated α_IIbβ₃¹²  and 3-dimensional modeling of α_IIbβ₃ based on electron microscopy and x-ray crystallographic studies¹³  reported the existence of domain-specific conformational changes in receptors from activated platelets. Studies of binding of platelets to monoclonal antibodies also suggest specific agonist-induced conformational changes in α_IIbβ₃.³⁶  Furthermore, point mutations in β₃ leading to conformational changes are associated with changes in the state of activation of α_IIbβ₃.^37-42  Thus, a variety of experimental approaches support the contention that integrin activation is accompanied by conformational changes; however, the mechanisms by which the putative conformational changes increase the ligand binding affinity remain unclear.³⁵  The same comment applies to the mechanisms involved in locking the fibrinogen receptor in a low-affinity state. Our observation that deletion of residues 616 to 690 in β₃ endows constitutive activity to the α_IIbβ₃(616-690del) receptor suggested the presence within this region of a not-previously-identified domain essential to maintain a resting, nonactivated state. Further mutation analysis revealed that disruption of the 663Cys-687Cys β₃ disulfide bridge is sufficient to confer constitutive activity. This finding may provide a molecular basis for the postulate that disulfide swapping could play a role in integrin activation.⁴³  A chimeric receptor in which the region 616-690 of β₃ was replaced by its homologous region of β₇ that lacks cysteines 663 and 687 showed a constitutive activity similar to that observed when the 663Cys-687Cys disulfide bridge was disrupted, that is, less than 50% of the binding capacity of the β₃(616-690del) complexes. Moreover, the degree of α_IIbβ₃ activation was a function of the length of the carboxy-terminal deletion of β₃. These observations suggest that shortening of the β₃ carboxy-terminal ectodomain or disruption of the 663-687 disulfide bridge acted by different mechanisms in activating α_IIbβ₃. The role played by the 663-687 disulfide bridge of β₃ in activating α_IIbβ₃ adds to the list of natural^38-41  or experimental^41,44  mutations of β₃ cysteines associated with receptor activation. According to these observations, receptor activation may be achieved by different patterns of protein folding due to loss or rearrangement of disulfide links. Regardless of the mechanism by which the mutated β₃ subunits confer activity to β₃ receptors, signals must be propagated to elicit these changes in the ligand binding domain. In principle, 2 possibilities could be considered: first, conformational changes in the βTD are transmitted directly to the ligand-binding domain; second, structural changes in the βTD could force the α and β cytoplasmic tails to adopt a position unable to maintain the proper association with regulatory or cytoskeletal proteins. The first possibility finds support in the observation that a correlation exists between the length of the deleted β₃ fragment and the state of activation, suggesting that in the native complex shortening the β₃ stalk could make more accessible the ligand to the binding site in the amino-terminal globular domain. On the other hand, the observed changes in size and shape of CHO-α_IIbβ₃(616-690del) cells support the second possibility. In this regard, impairment of actin polymerization induces fibrinogen binding to platelets, suggesting that cytoskeletal interaction may contribute to maintain the α_IIbβ₃ receptor in a resting state.⁴⁵ 

The divalent manganese ion increases the ligand-binding affinity to integrins and promotes cell adherence to immobilized ligands.^46,47  It is worth noting that Mn²⁺ was able to further enhance the ligand binding in cells expressing mutated α_vβ₃ or α_IIbβ₃ receptors, suggesting that Mn²⁺ and mutations in the βTD of β₃ may act in activating β₃ integrins by different mechanisms. This and other observations³⁶  indicating alternative mechanisms of activation suggest that, under physiologic conditions, the state of activation of platelets may be the result of the combinatorial action of several positive and negative modulators.

The vitronectin and fibrinogen receptors share the β₃ subunit as well as some ligands. Nevertheless, their patterns of tissue expression as well as their pathophysiologic roles are markedly different. In most cell lines and in all adherent cells the vitronectin receptor seems to be in a permanent state of activation^28,48 ; however, in platelets and some lymphoid cells, the ligand affinity is subjected to regulation by physiologic agonists^49-51  and, most likely, the receptor occupancy generates cellulipetal (outside-in) signals of regulatory importance.^52,53  In principle, the ligand specificity and affinity of α_vβ₃ and α_IIbβ₃ receptors should be determined by their distinct α subunits; but the knowledge about the mechanisms controlling this process is rather limited. Our data indicate that the βTD of β₃ plays a previously unnoticed role in maintaining both receptors in a resting, nonactive state.

To conclude, the 616-690 region located in the carboxy-terminal tail of the β₃ ectodomain, so-called βTD, is not essential for cell surface expression of β₃ receptors. However, either deletions within this region or disruption of the C663-C687 disulfide bridge, confers constitutive activity to β₃ integrins. Thus, the normal conformation of this β₃ domain contributes to restrain β₃ integrins in a resting, low ligand-affinity state.

We thank Dr S. Shattil (The Scripps Research Institute, La Jolla, CA) for the gift of PAC-1 and WOW-1 Fabs.