INTRODUCTION

Integrins are a widely distributed family of heterodimeric transmembrane proteins that mediate cell adhesion by binding to extracellular adhesive proteins (1). In addition, integrins are thought to mediate the attachment of actin filaments to the cell membrane presumably by binding to intracellular cytoskeletal proteins (2). The most prominent platelet membrane glycoprotein GPIIb-IIIa¹ (integrin _IIb3) plays a central role in hemostasis and thrombosis by acting as a receptor for several adhesive proteins including fibrinogen, fibronectin, von Willebrand factor, and vitronectin (3, 4). Furthermore, it has been demonstrated that following the binding of adhesive proteins to GPIIb-IIIa on activated platelets, this receptor complex becomes associated with the platelet cytoskeleton (5, 6, 7, 8, 9). This process has been implicated in several post-occupancy events including clot retraction (10), protein tyrosine phosphorylation (11, 12), as well as receptor clustering (13) and redistribution (14, 15). Although extracellular interactions between adhesive proteins and GPIIb-IIIa have been well characterized (16), the mechanisms mediating interactions between the cytoplasmic domains of GPIIb-IIIa and cytoskeletal proteins have not been clearly defined. Nevertheless, in other cell types such as endothelial cells and fibroblasts, several cytoskeletal proteins including talin, vinculin, and -actinin have been suggested to provide linkages between integrins and actin filaments (2). Morphological studies have demonstrated that these cytoskeletal proteins are concentrated in focal adhesions where integrins (e.g. ₅₁) cluster and interact with proteins in both the extracellular matrix and cytoskeleton (2).

Several biochemical studies have suggested an interaction between talin and the cytoplasmic domains of the  subunits of integrins (17, 18). Furthermore, interactions between talin and vinculin (19, 20, 21), as well as between vinculin and -actinin (22), have also been reported. Since -actinin is an actin cross-linking protein, this leads to the speculation that actin filaments are anchored to membrane bound integrins via a chain of protein-protein interactions involving talin, vinculin, and -actinin. However, recent studies have suggested alternative mechanisms of cytoskeletal attachment of integrins in which talin (23) or -actinin (24, 25) directly links integrins to actin filaments. In addition, the recent findings that vinculin contains a cryptic actin-binding site raise the possibility that talin and vinculin mediate the binding of integrins to actin filaments in a manner independent of -actinin (26).

Regarding the attachment of GPIIb-IIIa to the platelet cytoskeleton, it has been shown that in resting platelets, GPIIb-IIIa associates with a membrane skeleton containing talin and vinculin. As a result of platelet aggregation, there is a concomitant redistribution of GPIIb-IIIa and cytoskeletal proteins to the cross-linked actin filaments in the cytoplasmic cytoskeleton (9). In a preliminary report utilizing ligand blotting techniques, it has been suggested that talin interacts with the cytoplasmic domain of GPIIIa (27). More recently, specific binding of -actinin to purified GPIIb-IIIa adsorbed onto microtiter wells or incorporated into phospholipid vesicles has been reported (24). Using a solid-phase binding assay, we further examine the interaction of GPIIb-IIIa with talin, as well as the role of the cytoplasmic domain of GPIIb-IIIa in mediating interaction with this cytoskeletal protein.

MATERIALS AND METHODS

Peptide sequences were represented by the single-letter amino acid code (28). A 41-amino acid peptide (P3a, HDRKEFAKFEEERARAKWDTANNPLYKEATSTFTNITYRGT) corresponding to residues 722-762 of GPIIIa and its scrambled variant (scr-P3a, FEEWETIKNTPDPHLRLAEARRTNISSKGNYYTKTKIKEFLTAARDA), as well as a scrambled peptide of the cytoplasmic sequence of GPIIb (scr-P2b, DKFGRPPKVENEELEDRGEF), were generous gifts of Dr. Leslie V. Parise of the University of North Carolina, Chapel Hill. The following peptides were synthesized by solid-phase synthesis using an ABI model 431 peptide synthesizer or were purchased from the Peninsula Laboratories, Inc. A 21-amino acid peptide (CKVGFFKRNRPPLEEDDEEGE) corresponding to the GPIIb cytoplasmic sequence with a cysteine residue added to the N terminus to facilitate coupling to carrier proteins was designated as P2b. The dodecapeptide corresponding to residues 400-411 of the  chain of fibrinogen with the sequence HHLGGAKQAGDV was designated as H₁₂. Amino acid analysis of each peptide was consistent with the desired sequence. Monoclonal antibodies against talin (8d4) (29) and vinculin (hVIN-1) (30), as well as normal mouse serum, were obtained from Sigma. Monoclonal antibodies against GPIIb (PMI-1) (31) or GPIIIa (mAb15 and PMI-2) (32, 33), as well as the polyclonal anti-peptide antibody against the GPIIb light chain (anti-V41) (34), were kindly provided by Dr. Mark H. Ginsberg of the Scripps Research Institute, La Jolla, CA. Antibodies were purified from ascites fluid or from normal mouse serum by chromatography on protein A-Sepharose CL-4B (Pharmacia Biotech, Inc.). In binding studies, mAb15 and 8d4 were labeled with carrier-free Na¹²⁵I (Amersham Corp.) using the IODO-BEADS iodination reagent (Pierce) to a specific activity of approximately 2 µCi/µg.

P2b was coupled to keyhole limpet hemocyanin using m-maleimidobenzoic acid N-hydroxysuccinimide ester (35) and used as immunogen for BALB/c mice. Immunization and fusion of splenocytes with P3-X63Ag8.653 myeloma cells were performed as described previously (32). Hybridomas were grown in selective media (hypoxanthine/aminopterin/thymidine), and their supernatants were tested in an enzyme-linked immunosorbent assay for the presence of antibodies reactive with RGD affinity purified GPIIb-IIIa. A positive hybridoma 10B2, which secreted anti-GPIIb antibodies belonging to the IgG_2a subclass, was subcloned twice at limiting dilutions of 0.1-0.5 cell/well. As shown in Fig. 1, immunoblot analyses using nonreduced platelet lysates or RGD affinity purified GPIIb-IIIa showed that mAb10B2 reacted specifically with a ~140-kDa protein corresponding to GPIIb. Moreover, upon reduction of proteins in the platelet lysate, mAb10B2 recognizes a ~25-kDa polypeptide identified by its reactivity with anti-V41 on Western blots (34) as the light chain of GPIIb. The antibody was produced as ascites and purified by chromatography on protein A-Sepharose CL-4B. Half-maximal binding of ¹²⁵I-labeled mAb10B2 to P2b- or GPIIb-IIIa-coated microtiter wells occurred at an antibody concentration of 0.5 µM.

Purification of GPIIb-IIIa from platelet lysates by RGD affinity chromatography was performed as described previously (36). Washed outdated platelets were solubilized in lysis buffer containing 10 mM HEPES, pH 7.5, 0.15 M NaCl, 1 mM CaCl₂, 1 mM MgCl₂, 0.1 mM leupeptin, 10 mM N-ethylmaleimide, 1 mM phenylmethanesulfonyl fluoride, and 50 mM octyl glucoside. The platelet lysate was applied to GRGDSPK-coupled Sepharose 4B and incubated at 4 °C overnight. After washing unbound proteins with column buffer (same as the lysis buffer except it contained 25 mM octyl glucoside), GPIIb-IIIa was eluted with 1.7 mM H₁₂ (Fig. 2A, lane 1). Since we and other investigators (37, 38) found that a subpopulation of affinity purified GPIIb-IIIa existed as a high molecular weight complex with actin filaments, we employed gel filtration to purify further GPIIb-IIIa for binding studies. Thus, the H₁₂ eluate was applied to a Sephacryl S-300 High Resolution (HR) column (1.5 × 95 cm) and eluted with 10 mM HEPES, pH 7.5, 0.15 M NaCl, 1 mM CaCl₂, 1 mM MgCl₂, and 25 mM octyl glucoside. As shown in Fig. 2A, lane 2, a small fraction (<10%) of GPIIb-IIIa was eluted with actin in the void volume, and further elution yielded monomeric GPIIb-IIIa (lane 3) with a Stokes radius of 61 Å as previously reported (37, 39). To determine the concentration of purified GPIIb-IIIa, proteins were resolved by SDS-PAGE, and the Coomassie Blue-stained bands were quantitated by densitometric scanning using the Discovery Series^TM densitometer and Quantity One software package (pdi Inc., Huntington Station, NY) with bovine serum albumin (BSA) as standards. Purified GPIIb-IIIa was stored at 4 °C and used within a week of isolation.

Talin was purified from Triton X-100 extracts of outdated human platelets by a combination of DEAE-cellulose anion exchange chromatography, gel filtration on Sepharose CL-6B and hydroxyapatite column chromatography as described (40). Purified talin was dialyzed against 20 mM Tris acetate, pH 7.6, 20 mM NaCl, 0.1 mM EDTA, 1 mM EGTA, and 0.1% -mercaptoethanol. In some experiments as indicated, -mercaptoethanol was omitted in the dialysis buffer. SDS-PAGE analysis of the purified protein under reducing conditions showed a single Coomassie Blue-stained band of ~235 kDa which reacted specifically with the anti-talin monoclonal antibody 8d4 on Western blot (Fig. 2B). For measuring the concentration of purified talin, the protein determination kit based on Peterson's modification of the micro-Lowry method (Sigma) was used.

Proteins were resolved by SDS-PAGE and transferred onto nitrocellulose filters by standard procedures (41). After blocking with 5% non-fat dry milk (Bio-Rad), the filters were probed with the indicated primary antibodies. Bound antibodies were detected by incubation with ¹²⁵I-labeled species-specific secondary antibodies followed by autoradiography.

Microtiter wells (Immulon 2 removawell strips) were coated with talin (45 µg/ml) for 48 h at 4 °C and then blocked with 3% BSA overnight at 4 °C. In some studies, talin was captured by immobilized 8d4 by incubating talin (150 µg/ml) for 2 h at 37 °C with 8d4-coated wells. After two washes with a buffer containing 10 mM HEPES, 0.15 M NaCl, 1 mM CaCl₂, 1 mM MgCl₂, 25 mM octyl glucoside, pH 7.4, followed by two additional washes with the same buffer without octyl glucoside, purified GPIIb-IIIa was added and incubation proceeded at 37 °C for 4-5 h. Unbound GPIIb-IIIa was then removed, and the washing procedure described above was repeated. Bound receptor was detected by incubation with ¹²⁵I-labeled mAb15 (50 nM) for 1 h at 22 °C. The wells were then washed extensively with Tris-buffered saline (TBS, 10 mM Tris, 0.15 M NaCl, pH 7.4) containing 0.05% Tween 20, and bound radioactivity was detected by -counting. Background absorption of ¹²⁵I-labeled mAb15 to talin-coated wells, which had not been incubated with GPIIb-IIIa, was measured in parallel and was subtracted from the total amount of bound labeled antibody. To quantitate the amounts of adsorbed talin, the binding of ¹²⁵I-labeled 8d4 (67 nM) to control wells was determined. In inhibition studies, GPIIb-IIIa was preincubated with the indicated antibodies for 1 h at 37 °C before the receptor was added to the wells.

RESULTS

As an initial approach to examine the interaction of GPIIb-IIIa with talin, we purified both proteins (see Fig. 2 in ``Materials and Methods'') and developed a solid-phase binding assay in which soluble GPIIb-IIIa was allowed to bind to talin immobilized onto microtiter wells. In order to minimize modification of GPIIb-IIIa which might affect binding, we detected unlabeled bound receptor with a saturating concentration of ¹²⁵I-labeled mAb15. Assuming that mAb15 binds to GPIIb-IIIa with a 1:1 stoichiometry, this method allows us to quantitate directly the amounts of bound receptor. As shown in Fig. 3, preincubation of talin-coated wells with GPIIb-IIIa resulted in an increased binding of ¹²⁵I-labeled mAb15 as compared with control wells incubated with the vehicle buffer of GPIIb-IIIa. In contrast, no detectable difference was observed with BSA-coated wells incubated with or without GPIIb-IIIa. These results suggest that GPIIb-IIIa bound specifically to talin immobilized onto microtiter wells. In subsequent experiments, we routinely measured the amounts of nonspecific absorption of mAb15 to talin- or BSA-coated wells, and these values were subtracted from the total amounts of bound antibody. Using this assay system, we then examined the time course of GPIIb-IIIa binding to immobilized talin (Fig. 4). At an input GPIIb-IIIa concentration of 10 nM, the binding of GPIIb-IIIa to immobilized talin was time-dependent and approached steady-state after 4-5 h incubation at 37 °C. Again, no significant binding of GPIIb-IIIa to control BSA-coated wells was observed over this period. To characterize further the interaction of GPIIb-IIIa and talin, we performed binding isotherms with varying concentrations of GPIIb-IIIa. Fig. 5 shows that the dose-dependent binding of GPIIb-IIIa to talin was saturable, and half-saturation occurred at approximately 15 nM GPIIb-IIIa. Furthermore, as detected by ¹²⁵I-labeled mAb15 binding, there was 211 ± 8 fmol of GPIIb-IIIa bound per well at saturating concentrations of GPIIb-IIIa. To estimate the amounts of talin adsorbed onto the wells, we performed direct binding of ¹²⁵I-labeled 8d4 to control wells and found that there was 117 ± 10 fmol of talin/well.

It is generally agreed that coating of proteins onto plastic microtiter wells would induce conformational changes of the adsorbed proteins. In this regard, cryptic GPIIb-IIIa binding sites might have been exposed on immobilized talin. Therefore, we examined whether GPIIb-IIIa also bound to talin captured by immobilized 8d4, an anti-talin monoclonal antibody. To prevent the reduction of immobilized antibodies in these experiments, -mercaptoethanol was removed from purified talin by dialysis prior to its addition to antibody-coated wells. After incubation at 37 °C for 2 h, unbound talin was removed by washing, and the binding of GPIIb-IIIa to captured talin proceeded as described above. As shown in Fig. 6, without the preincubation of talin with 8d4-coated wells, minimal binding of GPIIb-IIIa was detected. However, the binding of talin to immobilized 8d4 caused a dramatic increase in GPIIb-IIIa binding. Parallel binding was also performed with control wells coated with purified normal mouse IgG, or with hVIN-1, an isotype-matched monoclonal antibody against vinculin. As expected, preincubation of talin to these wells coated with irrelevant antibodies did not cause any increase in GPIIb-IIIa binding. In three separate experiments, we consistently observed an increased binding of GPIIb-IIIa to talin captured by immobilized 8d4, although the amounts of bound GPIIb-IIIa varied among experiments (ranged from 13 to 47 fmol/well). Nonetheless, these results demonstrated that in addition to immobilized talin, GPIIb-IIIa also bound to antibody-captured talin which is more likely to assume the conformation of soluble talin.

In an attempt to examine whether the observed binding of GPIIb-IIIa to talin was mediated by the cytoplasmic domain of the receptor, we tested the effect of our newly developed anti-peptide monoclonal antibody (mAb10B2) directed against the cytoplasmic sequence of GPIIb (see Fig. 1 in ``Materials and Methods''). In these experiments, purified mAb10B2 or normal mouse IgG was incubated with GPIIb-IIIa at 37 °C for 30 min before the receptor was allowed to bind to talin captured by immobilized 8d4. At an antibody concentration of 1 µM, mAb10B2 inhibited GPIIb-IIIa binding to 8d4-captured talin by 48%, whereas normal mouse IgG did not cause a significant inhibition (Table I). Similar results were obtained when these antibodies were tested for inhibition of GPIIb-IIIa binding to immobilized talin (not shown). The observed partial inhibition of GPIIb-IIIa binding to talin caused by mAb10B2 may be due to the low affinity of this monoclonal antibody (K_d ~0.5 µM). Alternatively, there may be multiple talin-binding sites on GPIIb-IIIa, some of which may not be inhibitable by mAb10B2. In order to characterize further the interaction between talin and the cytoplasmic domain of GPIIb-IIIa, we examined whether talin binds directly to synthetic peptides corresponding to the cytoplasmic sequences of GPIIb (P2b) and GPIIIa (P3a). In this assay, purified talin was incubated with microtiter wells coated with these peptides, and bound talin was detected with ¹²⁵I-labeled 8d4. As controls, we also examined talin binding to wells coated with scrambled variants of these peptides (scr-P2b and scr-P3a) with the same amino acid compositions but arranged in random sequences. As shown in Fig. 7A, talin bound more effectively to wells coated with P3a (bar 2) as compared to control wells coated without peptide (bar 1) or with scr-P3a (bar 3). In two separate experiments, we observed that the binding of talin to the native P3a sequence was ~50-60% more effective than to its scrambled sequence. Likewise, we observed that talin also bound to wells coated with P2b (Fig. 7B, bar 2). Again, much less binding was detected to control wells coated with the scr-P2b peptide (bar 1). To substantiate further the specificity of talin binding to P2b, we examined the inhibitory effect of mAb10B2. Thus, P2b-coated wells were preincubated with mAb10B2 (Fig. 7B, bar 3) or normal mouse IgG (bar 4) at 37 °C for 1 h prior to the addition of talin. As expected, mAb10B2 blocked talin binding to P2b by 95%, whereas normal mouse IgG was completely ineffective.

Table I. Antibody inhibition of GPIIb-IIIa binding to talin Talin was allowed to bind to 8d4-coated wells as described in the legend of Fig. 6. Purified GPIIb-IIIa was incubated with 1 µM of the indicated antibody for 1 h at 37 °C prior to addition to wells with captured talin. Binding proceeded as described, and percent inhibition of binding was calculated relative to controls without antibody. mAb15 bound Inhibition fmol % Control 12.9  ± 0.9 mAb10B2 6.7  ± 0.7 48 Normal mouse IgG 12.0  ± 0.7 7

DISCUSSION

Based on the co-localization of talin and integrins in focal adhesions, it is generally believed that talin plays a crucial role in the attachment of integrins to cytoskeleton (2). However, to date, the only biochemical evidence for a direct integrin-talin interaction was the reported elution profile shift toward a higher molecular weight complex during equilibrium gel filtration of the avian integrin complex (CSAT antigen) in the presence of talin (17, 18). Using highly purified GPIIb-IIIa and talin in the present study, we have developed a solid-phase binding assay and demonstrated direct interaction between GPIIb-IIIa and talin. In this assay, we observed that GPIIb-IIIa bound to immobilized talin as well as to talin captured by the anti-talin monoclonal antibody 8d4. Furthermore, the observed GPIIb-IIIa/talin interaction was mediated by the cytoplasmic domain of the receptor since it was inhibited by the anti-GPIIb cytoplasmic domain monoclonal antibody mAb10B2. Finally, the findings that talin bound directly to synthetic peptides corresponding to the cytoplasmic sequences of both GPIIb (P2b) and GPIIIa (P3a) suggest that a potential talin-binding site(s) is located in the cytoplasmic domain of both subunits of this integrin.

Talin is abundant in human platelets accounting for >3% of total platelet proteins (42). Morphological studies have demonstrated that talin is uniformly distributed in the cytoplasm of resting platelets but is translocated to the cytoplasmic face of the plasma membrane following cellular stimulation with thrombin (43). By differential sedimentation of Triton X-100 extracts of platelet proteins, it has been shown that in unstimulated platelets a subpopulation of GPIIb-IIIa is associated with a membrane skeleton fraction containing talin, vinculin, spectrin, and other signaling molecules. Following thrombin-induced platelet aggregation, GPIIb-IIIa and talin, among other proteins in the membrane skeleton, become associated with the actin filaments in the cytoplasmic cytoskeleton (9). Our findings that talin bound directly to the cytoplasmic domain of GPIIb-IIIa provide evidence that it acts as a linkage protein between GPIIb-IIIa and short actin filaments in the membrane skeleton. Furthermore, it is tempting to speculate that thrombin-induced redistribution of talin in platelets would facilitate the interaction of this cytoskeletal protein with GPIIb-IIIa. Nonetheless, it should be noted that our results do not exclude the involvement of other cytoskeletal proteins in the membrane skeleton fraction, and further studies are required to define their role in mediating GPIIb-IIIa attachment to the cytoskeleton.

To demonstrate GPIIb-IIIa/talin interaction, we initially measured the binding of soluble GPIIb-IIIa to immobilized talin. However, coating of talin onto microtiter wells might change the protein conformation and expose artifactual binding sites for GPIIb-IIIa. To eliminate this possibility, we examined the ability of GPIIb-IIIa to bind to talin captured by the anti-talin monoclonal antibody 8d4. The findings that GPIIb-IIIa interacts with 8d4-captured talin indicate that the GPIIb-IIIa binding domain(s) are exposed on the surface of the native talin molecule. Furthermore, these results also rule out the possibility that GPIIb-IIIa bound to a contaminant in the talin preparations.

Having established conditions for measuring in vitro GPIIb-IIIa interaction with talin, we proceeded to identify region(s) within GPIIb-IIIa mediating this interaction. Previous data suggest that a putative talin-binding site is present within a highly conserved sequence (WDTGENPIYK) in the cytoplasmic domain of the ₁ integrin subunit since a synthetic peptide from this region inhibits integrin-talin interaction (18). Additionally, in a preliminary report, Simon and Burridge (44) observed that talin bound to a synthetic peptide corresponding to the entire cytoplasmic domain of the ₁ integrin subunit. Consistent with these findings, we found that talin bound to the P3a peptide corresponding to the entire GPIIIa cytoplasmic sequence. In control samples, we found that talin also bound to a lesser extent to wells coated with a scrambled variant of P3a. Thus, as suggested for the binding of -actinin to the C-terminal region of the ₁ integrin cytoplasmic domain (25), the overall composition of the peptide rather than the order of amino acids may be important for its interaction with talin. Alternatively, the scr-P3a peptide used in the present study may retain some sequence similarity to the native sequence, thereby exhibiting a lower affinity interaction with talin. Nonetheless, our results confirm earlier observations that a talin binding site(s) is present in the cytoplasmic domains of integrin  subunits (18, 44) and further substantiate functional studies in which truncation and certain point mutations of the GPIIIa cytoplasmic sequence were shown to abolish cell spreading as well as the recruitment of GPIIb-IIIa and talin to focal adhesions (45, 46).

Besides interacting with the GPIIIa cytoplasmic sequence, we observed that talin also binds to immobilized P2b peptide corresponding to the entire GPIIb cytoplasmic sequence. The specificity of this interaction was demonstrated by the failure of a scrambled version of P2b to support talin binding and by specific inhibition of talin binding to P2b caused by the anti-GPIIb cytoplasmic domain monoclonal antibody mAb10B2. Therefore, these results indicate that a talin-binding site is also present within the  subunit of this integrin. In support of this finding, Pavalko and Otey (47) noted in a review article that talin binds to synthetic peptides corresponding to the cytoplasmic tails of ₄ and ₅ but not ₃. However, it is noteworthy that besides the highly conserved GFFKR sequence near the membrane, there is no apparent homology among the cytoplasmic domains of these integrin  subunits. Thus, it is not clear whether the diverse cytoplasmic sequences of _IIb, ₄, and ₅ interact with the same or different regions on talin. Recently, it has been suggested that the cytoplasmic sequences of GPIIb and GPIIIa associate with each other to form a defined tertiary structure (48). Since talin interacts with the cytoplasmic domains of both  and  subunits of certain integrins, it is conceivable that interaction of the two cytoplasmic tails in the native receptor would constitute a high affinity binding site for talin.

Platelet stimulation activates GPIIb-IIIa resulting in high affinity binding of macromolecular ligands such as fibrinogen (49, 50). Moreover, it has been shown that small ligand-mimetic peptides are also capable of activating GPIIb-IIIa (51). In this regard, RGD affinity purified GPIIb-IIIa was found to exist in the active conformer with high affinity fibrinogen-binding function (51, 52). Inasmuch as RGD affinity purified GPIIb-IIIa was utilized in our binding assay, we cannot exclude the possibility that the observed talin binding function of GPIIb-IIIa is confined to the active population of receptor isolated by this procedure. Therefore, whether conformational changes in GPIIb-IIIa also modulate its binding affinity to talin remains to be established. Similarly, thrombin activation of platelets leads to an increased phosphorylation of talin (53), and it would be interesting to determine whether phosphorylation would affect the affinity of talin for interaction with GPIIb-IIIa.

In summary, we have developed a rapid and sensitive solid- phase binding assay to detect direct interaction between GPIIb-IIIa and talin. Moreover, we demonstrated that purified talin binds to the cytoplasmic sequences of both  and  subunits of this integrin. This assay system would be useful for further studies to examine the interaction between talin and other integrins, as well as for the identification of specific determinants within the cytoplasmic domains of integrins mediating interaction with this cytoskeletal protein.