Bidirectional Transmembrane Modulation of Integrin αIIbβ3 Conformations*

Activation of blood platelets by physiological stimuli (e.g. thrombin, ADP) at sites of vascular injury induces inside-out signaling, resulting in a conformational change of the prototype integrin αIIbβ3 from an inactive to an active state competent to bind soluble fibrinogen. Furthermore, ligand occupancy of αIIbβ3initiates outside-in signaling and additional conformational changes of the receptor, leading to the exposure of extracellular neoepitopes termed ligand-induced binding sites (LIBS), which are recognized by anti-LIBS monoclonal antibodies. To date, the mechanism of bidirectional transmembrane signaling of αIIbβ3 has not been established. In this study, using our newly developed anti-LIBScyt1 monoclonal antibody, we showed that extracellular ligand binding to αIIbβ3 on blood platelets induces a transmembrane conformational change in αIIbβ3, thereby exposing the LIBScyt1 epitope in the αIIb cytoplasmic sequence between Lys994 and Asp1003. In addition, a point mutation at this site (P998A/P999A) renders αIIbβ3 constitutively active to bind extracellular ligands, resulting in fibrinogen-dependent cell-cell aggregation. Taken collectively, these results demonstrated that the extracellular ligand-binding site and a cytoplasmic LIBS epitope in integrin αIIbβ3 are conformationally and functionally coupled. Such bidirectional modulation of αIIbβ3 conformation across the cell membrane may play a key role in inside-out and outside-in signaling via this integrin.

Interaction of adhesive proteins with transmembrane integrin adhesion receptors is essential for diverse biological processes including embryogenesis, angiogenesis, immune response, and hemostasis (1,2). It is generally agreed that insideout signaling processes regulate the affinity state of integrins for binding extracellular ligands (2,3). Thus, upon activation of blood platelets by physiological stimuli (e.g. thrombin or ADP) at sites of vascular injury, the prototype integrin ␣ IIb ␤ 3 undergoes a conformational change from an inactive to an active state competent to bind soluble fibrinogen (1)(2)(3)(4)(5). The cytoplasmic domain of ␣ IIb ␤ 3 appears to play a regulatory role in ␣ IIb ␤ 3 activation, since truncations of the entire cytoplasmic sequence of either the ␣ IIb or ␤ 3 subunit, including their membrane-proximal regions, were found to increase the ligand binding affinity of the mutant receptor expressed on Chinese hamster ovary (CHO) 1 cells (6,7). More recently, a potential salt bridge hinge formed between the ␣ IIb and ␤ 3 cytoplasmic sequences has been suggested to maintain ␣ IIb ␤ 3 in a default low affinity state; disruption of this structure may result in receptor activation (8).
It is well established that binding of adhesive ligands to integrins initiates outside-in signaling processes that mediate post-ligand binding events including cytoskeleton reorganization, receptor clustering, and gene transcription (2,3,9). Although the mechanisms regulating outside-in signaling of integrins remain elusive, the binding of cytoskeletal proteins and signaling molecules to the receptor's cytoplasmic domain as well as the receptor's conformational state have been implicated to play an important role in this process. In this regard, it has been suggested that ligand occupancy of the ␣ 5 ␤ 1 integrin may induce a transmembrane conformational change of the receptor, thereby unmasking specific regions in the receptor cytoplasmic domain mediating cytoskeletal attachment, which ultimately leads to receptor localization to focal contacts (10). However, to date, ligand-induced transmembrane conformational changes of an integrin receptor have not been demonstrated.
It has previously been shown that ligand binding to ␣ IIb ␤ 3 induces further conformational changes of the receptor extracellular domain, resulting in the exposure of neoantigenic sites termed ligand-induced binding sites (LIBS), which are recognized by anti-LIBS monoclonal antibodies (mAbs) (11)(12)(13)(14)(15)(16). Furthermore, certain anti-LIBS mAbs were found to activate ␣ IIb ␤ 3 to bind soluble fibrinogen (14,15). In this study, we postulated that bidirectional conformational changes of ␣ IIb ␤ 3 transducing through the receptor's transmembrane segment occur as a result of cellular activation and ligand binding. To test this possibility, we examined whether extracellular ligand binding induces the exposure of LIBS epitope(s) in the cytoplasmic domain of ␣ IIb ␤ 3 . In addition, we evaluated the functional role of a putative cytoplasmic LIBS epitope in regulating ␣ IIb ␤ 3 ligand binding affinity.

MATERIALS AND METHODS
Peptides and Antibodies-Peptides, represented as sequences of single letter amino acid codes (17), were synthesized by solid-phase synthesis using an ABI model 431 peptide synthesizer or were purchased from Research Genetics, Inc. (Huntsville, AL). The amino acid composition of each peptide was consistent with its desired sequence. The anti-␣ IIb ␤ 3 antibodies PMI-1 (18), anti-V41 (19) ␣ IIb C, an antipeptide polyclonal antibody raised in rabbits against the ␣ IIb cytoplasmic sequence Phe 992 -Glu 1008 , was a generous gift of Dr. X. Du of the University of Illinois (Chicago, IL).
Production of Anti-LIBS cyt 1-For the production of antipeptide mAbs against the ␣ IIb cytoplasmic sequence, the full-length P2b peptide (CK-VGFFKRNRPPLEEDDEEGE) was coupled to keyhole limpet hemocyanin using m-maleimidobenzoic acid N-hydroxysuccinimide ester and used as immunogen for BALB/c mice. Isolated splenocytes were fused with P3-X63Ag8.653 myeloma cells. Hybridomas were grown in selective media (hypoxanthine/aminopterin/thymidine), and their supernatants were tested in an ELISA for the presence of antibodies reactive with RGD affinity-purified ␣ IIb ␤ 3 (13,21). A positive hybridoma 3F5, which secreted anti-␣ IIb antibodies belonging to the IgG 1 subclass, was subcloned twice at limiting dilutions of 0.5 cell/well. The antibody was produced as ascites and purified by chromatography on protein A-Sepharose CL-4B (Amersham Pharmacia Biotech).
Immunoprecipitation-Gel-filtered platelets were surface-labeled with Na 125 I and solubilized in lysis buffer containing 50 mM octyl glucoside (21). Cell lysates were incubated with GRGDSP, GRGESP, or vehicle buffer for 30 min at 37°C. Antibodies were then added and incubated overnight at 4°C. The immunoprecipitated proteins were collected on protein G-Sepharose, electrophoresed on SDS-7% polyacrylamide gels under nonreducing conditions, and analyzed by autoradiography.
Indirect Immunofluorescent Microscopy-Washed human platelets resuspended in Tyrode's solution (2.5 ϫ 10 8 cells/ml) were incubated with the indicated reagents (see legend of Fig. 2) at 37°C for 30 min and subsequently fixed with 1% paraformaldehyde on ice for 1 h. After blocking unreacted aldehyde with Tris-buffered saline (30 mM Tris, 120 mM NaCl, pH 7.4) containing 0.5 M NH 4 Cl, cells were allowed to settle onto polylysine-coated glass coverslips and incubated with 0.2 mg/ml lysophosphatidylcholine (LPC) for 5 min to render them permeable. Permeabilized cells were rinsed with Tris-buffered saline containing 0.1% bovine serum albumin and incubated with anti-LIBS cyt 1 followed by rhodamine-conjugated goat anti-mouse IgG. Samples were mounted with a droplet of FITC guard, and platelets were viewed with a Jenaval phase/fluorescence microscope (Jenoptik Jena GmbH) and photographed with Eastman Kodak Tri-X panchromatic film (22).
Competitive ELISA-Microtiter wells were coated with the fulllength P2b peptide (5 g/well) and blocked with 3% bovine serum albumin. Anti-LIBS cyt 1 was incubated with 10 M inhibitory peptides at 37°C for 30 min and added to the P2b-coated wells. Antibody binding to the adsorbed P2b proceeded at 37°C for 1 h. The wells were washed, and bound antibody was detected with horseradish peroxidase-conjugated goat anti-mouse IgG using o-phenylenediamine as substrate (12). Absorbance at 490 nm (A 490 ) was measured, and percentage inhibition was calculated relative to control without inhibitor.
Site-directed Mutagenesis-The expression constructs encoding wild type ␣ IIb (CD2b) and ␤ 3 (pc3A) have been previously described (23,24). To generate the pc2b construct encoding wild type ␣ IIb , a 3.3-kilobase fragment of ␣ IIb containing the entire coding sequence and the 3Јuntranslated region was excised from CD2b by digestion with XbaI and ligated into the expression vector pcDNA3. The resultant construct was designated as pc2b. Both pc2b and pc3A were kindly provided by Dr. J. C. Loftus at the Mayo Clinic (Scottsdale, AZ). The P998A/P999A mutation in ␣ IIb was generated by splice overlap extension mutagenesis (25). Overlapping fragments containing this mutation were first made by polymerase chain reaction amplifications on pc2b using the following oligonucleotide pairs: (a) 5Ј-CACAAGCGGGATCGCAGACAGATCTTC-CTGCCAGA-3Ј and (b) 5Ј-CTTCTTCCAGGGCTGCCCGGTTCCGCTT-G-3Ј; (c) 5Ј-CAAGCGGAACCGGGCAGCCCTGGAAGAAG-3Ј and (d) 5Ј-GGACAGTGGGAGTGGCACCTTCCAGGGTCAAGGAA-3Ј as primers. The overlapping fragments were combined, denatured by heating at 94°C for 5 min, and reannealed by cooling to 55°C. The ends were filled in with Pfu, and the double-stranded fragments were then amplified by polymerase chain reaction using the oligonucleotide pair a and d. The amplified product was digested with BamHI and XbaI and reinserted into a BamHI-XbaI-digested pc2b vector fragment. The mutant construct was identified by automated DNA sequencing, purified by chromatography on QIAGEN Tip-100, and co-transfected with the wild type ␤ 3 construct (pc3A) into CHO-K1 cells (ATCC, Rockville, MD) by liposome-mediated transfection as described (7). Surface expression of mutated ␣ IIb ␤ 3 was analyzed by flow cytometry using FITC-conjugated AP-2. Stable cell lines were selected in medium containing 0.75 mg/ml G418 (Sigma), and single cell sorting was performed to obtain stable clonal lines, which were high expressors of the mutant ␣ IIb ␤ 3 .

Ligand Binding Induces A Transmembrane Conformational
Change of ␣ IIb ␤ 3 -To examine the possibility that ligand binding induces the exposure of LIBS epitopes in the cytoplasmic domain of ␣ IIb ␤ 3 , we developed anti-peptide mAbs against the receptor's cytoplasmic sequences and screened for antibodies that preferentially bind to the ligand-occupied conformer of ␣ IIb ␤ 3 . In the present study, we focused on the ␣ IIb cytoplasmic tail. Initially, mAbs reactive with RGD affinity-purified ␣ IIb ␤ 3 in an ELISA system were further characterized by immunoblotting and immunoprecipitation studies. The mAb obtained from clone 3F5 recognizes a ligand-induced binding site in the cytoplasmic domain of ␣ IIb ␤ 3 (LIBS cyt ), and therefore this mAb is designated as anti-LIBS cyt 1. Fig. 1A shows that anti-LIB-S cyt 1 specifically immunoblotted the 140-kDa nonreduced ␣ IIb subunit in RGD affinity-purified ␣ IIb ␤ 3 (lane 1) and in a detergent extract of platelet proteins (lane 2). Upon reduction of purified ␣ IIb ␤ 3 and proteins in the platelet lysate, anti-LIBS cyt 1 immunoblotted the 27-kDa light chain of ␣ IIb , which contains its cytoplasmic sequence (lanes 3 and 4). To determine whether the interaction of anti-LIBS cyt 1 with nondenatured ␣ IIb ␤ 3 is dependent on ligand occupancy, we performed immunoprecipitation experiments using lyates of surface-radioiodinated platelets in the presence and absence of an RGD peptide. As judged by densitometric scanning of the immunoprecipitated 125 I-labeled protein bands, incubation of platelet lysates with GRGDSP caused a 7.2-fold increase in the amount of ␣ IIb ␤ 3 immunoprecipitated by anti-LIBS cyt 1 (Fig. 1B). In contrast, the variant GRGESP peptide was much less effective (1.5-fold). As controls, we used the well characterized anti-LIBS1 mAb (13), which demonstrated a similar effect in RGD-dependent immu- noprecipitation of ␣ IIb ␤ 3 . However, using the control mAb 15, whose binding to ␣ IIb ␤ 3 is not markedly influenced by ligand occupancy (13), we observed that GRGDSP incubation induced only a slight (1.5-fold) increase in the immunoprecipitation of ␣ IIb ␤ 3 . Thus, these results suggest that interaction of GRGDSP with the extracellular ligand binding site of ␣ IIb ␤ 3 induces a conformational change in the receptor's cytoplasmic domain. In support of our finding that the cytoplasmic domain of ␣ IIb ␤ 3 can exist in different conformational states, the cytoplasmic sequences of ␣ IIb and ␤ 3 have been shown to interact with each other, and at least two docking models with different tertiary structures of the ␣ IIb ␤ 3 cytoplasmic domain have been proposed (26 -28).
To examine whether RGD occupancy also induces transmembrane conformational changes of ␣ IIb ␤ 3 in situ, we performed indirect immunofluorescence microscopy using whole platelets preincubated with or without GRGDSP followed by paraformaldehyde fixation and LPC permeabilization to allow antibody access. As shown in Fig. 2A, incubation of platelets with GRGDSP (panel a) resulted in significant intracellular staining of anti-LIBS cyt 1 as opposed to control platelets incubated with vehicle buffer (panel b) or GRGESP (panel c). Furthermore, the observed rim staining pattern with GRGDSP-treated permeabilized platelets is suggestive of anti-LIBS cyt 1 localization to the inner face of the plasma membrane, since minimal staining was observed with nonpermeabilized cells (not shown). To investigate whether binding of the physiological ligand fibrinogen to ␣ IIb ␤ 3 on activated platelets also induces the exposure of LIBS cyt 1, we performed indirect immunofluorescence studies with ADP-stimulated platelets in the presence and absence of exogenous fibrinogen. Again, anti-LIBS cyt 1 staining was performed following fixation and cell permeabilization. The addition of fibrinogen to ADP-stimulated platelets dramatically increased anti-LIBS cyt 1 staining as compared with activation of platelets with ADP alone (Fig. 2B, panels a and b). In control samples, resting platelets failed to stain for anti-LIBS cyt 1 in the presence and absence of fibrinogen (Fig. 2B, panels c and  d). Therefore, these results demonstrate that anti-LIBS cyt 1 recognizes the ligand-occupied but not the activated unoccupied conformer of ␣ IIb ␤ 3 . By immunogold staining with AP6, an anti-LIBS mAb directed against the ␤ 3 extracellular domain, Nurden et al. (16) previously reported that a pool of ␣ IIb ␤ 3 in the ␣-granules of unactivated platelets exists in the ligandoccupied state. However, using anti-LIBS cyt 1, we failed to detect immunofluorescent staining of ␣ IIb ␤ 3 in the ␣-granules of resting platelets. This may be due to the association of ␣ IIb ␤ 3 with cytoskeletal components that mediate internalization and transport of the fibrinogen-␣ IIb ␤ 3 complex to the platelet ␣-granules (22,29,30), thus blocking interaction of anti-LIB-S cyt 1 with the cytoplasmic domain of ligand-occupied ␣ IIb ␤ 3 in the ␣-granule membranes. Nonetheless, the observation that fibrinogen binding to ␣ IIb ␤ 3 on the platelet surface induces a transmembrane conformational change of the receptor provides a possible mechanism by which ligand occupancy of ␣ IIb ␤ 3 mediates a variety of post-ligand binding function of blood platelets including clot retraction, receptor internalization, and cytoskeletal attachment.
To identify specific residues within the ␣ IIb cytoplasmic sequence mediating interaction with anti-LIBS cyt 1, we performed competitive ELISA analyses using peptides corresponding to the full-length or partial sequences of the ␣ IIb cytoplasmic tail. As shown in Table I, the full-length P2b peptide, as well as the truncated 15-mer (KVGFFKRNRPPLEED) effectively blocked anti-LIBS cyt 1 binding to immobilized P2b peptide. Moreover, using two overlapping peptides, we further localized the anti-LIBS cyt 1 epitope to the KRNRPPLEED sequence. Molecular modeling suggests that this region in both ␣ IIb and ␣ v subunits would form a tight ␤-turn (28,31). Since Pro 998 -Pro 999 may facilitate this ␤-turn formation, we tested the ability of KRN-RAALEED to inhibit anti-LIBS cyt 1 binding. The inhibitory effect of the peptide was found to be significantly diminished by substitution of the two proline residues with alanine, indicating that anti-LIBS cyt 1 recognizes a structural motif dependent on these two proline residues.
A Site-directed Mutation of the Anti-LIBS cyt 1 Epitope Acti-   (14,15) suggests that these LIBS epitopes may regulate the ligand binding affinity state of the receptor. Therefore, we evaluated the functional significance of Pro 998 -Pro 999 in the regulation of extracellular ligand binding to ␣ IIb ␤ 3 . In these studies, a double P998A/ P999A mutation in ␣ IIb was generated by splice overlap extension mutagenesis (25), and the mutant ␣ IIb construct was cotransfected with a wild type ␤ 3 construct into CHO cells. A stable clonal cell line (G4) expressing the mutant ␣ IIb ␤ 3 was established, and comparative analyses were performed with the control A5 cell line expressing wild type ␣ IIb ␤ 3 (32). As determined by flow cytometry using FITC-conjugated AP-2, a complex-specific anti-␣ IIb ␤ 3 mAb (20), both G4 and A5 cells expressed similar amounts of ␣ IIb ␤ 3 (mean fluorescence intensity: G4, 73.9; A5, 74.9). It has previously been shown that truncation of the entire ␣ IIb cytoplasmic sequence at residue 991 resulted in an increase of the ligand binding affinity of ␣ IIb ␤ 3 (6); therefore, we examined whether the P998A/P999A mutation might result in proteolytic cleavage of the ␣ IIb cytoplasmic domain, which would lead to receptor activation. Initially, we compared the molecular mass of the P998A/P999A mutant ␣ IIb light chain with those of wild type and truncated ␣ IIb by immunoblotting with anti-V41, an antipeptide antibody directed against the amino terminus of the ␣ IIb light chain (19). As shown in Fig. 3A, the light chain of the P998A/P999A mutant migrated with a similar molecular mass as the inactive wild type ␣ IIb . In contrast, the constitutively active ⌬991 truncation mutant of ␣ IIb (6) migrated with an increased mobility on SDS-polyacrylamide gel electrophoresis. Furthermore, Fig.  3B shows that both wild type and the P998A/P999A mutant ␣ IIb reacted with PMI-1, a mAb directed against an extracellular epitope in the ␣ IIb heavy chain (18,33), and with anti-␣ IIb C, an antipeptide polyclonal antibody raised against the ␣ IIb cytoplasmic sequence (Phe 992 -Glu 1008 ). Collectively, these results indicated that the P998A/P999A mutation did not result in proteolytic cleavages of the ␣ IIb cytoplasmic tail. As expected, anti-LIBS cyt 1 immunoblotted the wild type but not the mutant ␣ IIb . The affinity state of the mutant ␣ IIb ␤ 3 was then examined by the binding of FITC-conjugated PAC-1, an activation-specific mAb that, like fibrinogen, preferentially binds to activated ␣ IIb ␤ 3 (34). As reported previously (32), A5 cells expressing wild type ␣ IIb ␤ 3 bound minimal amounts of PAC-1 in the absence of receptor activation (Fig. 4A). In contrast, we observed constitutive binding of PAC-1 to the mutant ␣ IIb ␤ 3 on G4 cells, and this process was specifically blocked with 1 mM GRGDSP (Fig.  4A). Since the binding of fibrinogen to activated ␣ IIb ␤ 3 on platelets and transfected cells resulted in cell aggregation, we examined the ability of G4 cells to aggregate in the presence of fibrinogen. Fig. 4B shows that G4 but not A5 cells aggregated upon fibrinogen addition. Again, aggregation of G4 cells was specifically blocked with 1 mM GRGDSP. As a specificity control, we mutated the putative N 744 PLY ␤-turn motif in the ␤ 3 cytoplasmic sequence to QALY. Cells expressing wild type ␣ IIb and mutated ␤ 3 heterodimers failed to bind soluble fibrinogen and undergo aggregation (not shown). These results indicate that a structural change in the anti-LIBS cyt 1 binding site in the ␣ IIb cytoplasmic tail induces a transmembrane conformational change of ␣ IIb ␤ 3 , mimicking receptor activation due to insideout signaling.
It has previously been shown that ligand binding to integrin ␣ IIb ␤ 3 induces long range conformational changes in the extracellular domains of both ␣ IIb and ␤ 3 subunits (12,35). Our results demonstrated that such conformational changes transduce through the cell membrane to the cytoplasmic domain of the receptor. Besides ␣ IIb ␤ 3 , several other integrins such as ␣ v ␤ 3 and ␣ 5 ␤ 1 have been shown to undergo extracellular conformational changes upon ligand occupancy (13,36); therefore, it is tempting to speculate that ligand-induced conformational changes also occur in the cytoplasmic domains of other integrins. Since ligand binding to integrins results in cytoskeletal rearrangement and the generation of intracellular signals (2,3,9), the conformational state of integrin cytoplasmic domains may play a regulatory role in the assembly of cytoskeletal proteins and/or signaling molecules. In this regard, it has been shown that antibody-induced clustering of the ␣ 5 ␤ 1 integrin in the absence of ligand occupancy is sufficient for the intracellular accumulations of tensin and at least 20 signal transduction molecules (e.g. RhoA, Rac1, Ras, Raf, MEK, extracellular signal-regulated kinase, c-Jun N-terminal kinase, and focal adhesion kinase) (37,38). In contrast, both ligand occupancy and clustering of ␣ 5 ␤ 1 are required for transmembrane accumulations of several cytoskeletal proteins (e.g. talin, vinculin, and ␣-actinin). In light of these findings, our present data suggest that ligand-induced conformational changes of integrin cytoplasmic domains may play an essential role in the intracellular assembly of cytoskeletal proteins found in focal adhesions.
Emerging evidence has implicated ligand-induced oligomerization and/or conformational changes of transmembrane receptor complexes as potential mechanisms for receptor-mediated signal transduction. Specifically, it has been demonstrated that following ligand binding and dimerization of the platelet-derived growth factor receptor, there is a phosphorylation-dependent conformational change in the receptor cytoplasmic domain (39 -41). Although integrin ␣ IIb ␤ 3 on platelets becomes tyrosine-phosphorylated as a result of ligand binding and cell aggregation, the monovalent RGD peptide has been shown to block receptor phosphorylation (42). Inasmuch as GRGDSP binding to ␣ IIb ␤ 3 is capable of inducing the exposure of the anti-LIBS cyt 1 epitope, receptor phosphorylation is apparently not required for the observed effect. Thus, our results provide the first evidence of a direct effect of ligand occupancy FIG. 3. Immunoblot characterization of the P998A/P999A mutant ␣ IIb . Lysates of CHO cells expressing ␤ 3 integrins complexed with wild type, P998A/P999A mutated, or ⌬991 truncated ␣ IIb were subjected to immunoblotting with the indicated antibodies. A, proteins were resolved on SDS-18% polyacrylamide gel under reducing conditions and immunoblotted with anti-V41, an antipeptide antibody directed against the amino terminus of the ␣ IIb light chain. B, proteins were resolved on SDS-7% polyacrylamide gels under nonreducing conditions and immunoblotted with PMI-1 directed against an extracellular epitope of ␣ IIb , anti-␣ IIb C directed against the cytoplasmic sequence of ␣ IIb (Phe 992 -Glu 1008 ), or anti-LIBS cyt 1. on the conformation of the cytoplasmic domain of an intact integrin receptor. Additionally, site-directed mutation of the identified LIBS cyt 1 epitope resulted in an increase of ligand binding affinity of ␣ IIb ␤ 3 , indicating that the extracellular ligand-binding site and the cytoplasmic LIBS epitope of the receptor are functionally coupled. In sum, these findings suggest a bidirectional modulation of ␣ IIb ␤ 3 conformations across the cell membrane. Such conformational regulation may provide a novel mechanism for transmembrane receptor-mediated signal transduction.