Identification of a Talin-binding Site in the Integrin β3 Subunit Distinct from the NPLY Regulatory Motif of Post-ligand Binding Functions

Following platelet aggregation, integrin αIIbβ3 becomes associated with the platelet cytoskeleton. The conserved NPLY sequence represents a potential β-turn motif in the β3 cytoplasmic tail and has been suggested to mediate the interaction of β3integrins with talin. In the present study, we performed a double mutation (N744Q/P745A) in the integrin β3 subunit to test the functional significance of this β-turn motif. Chinese hamster ovary cells were co-transfected with cDNA constructs encoding mutant β3 and wild type αIIb. Cells expressing either wild type (A5) or mutant (D4) αIIbβ3 adhered to fibrinogen; however, as opposed to control A5 cells, adherent D4 cells failed to spread, form focal adhesions, or initiate protein tyrosine phosphorylation. To investigate the role of the NPLY motif in talin binding, we examined the ability of the mutant αIIbβ3 to interact with talin in a solid phase binding assay. Both wild type and mutant αIIbβ3, purified by RGD affinity chromatography, bound to a similar extent to immobilized talin. Additionally, purified talin failed to interact with peptides containing the AKWDTANNPLYK sequence indicating that the talin binding domain in the integrin β3 subunit does not reside in the NPLY motif. In contrast, specific binding of talin to peptides containing the membrane-proximal HDRKEFAKFEEERARAK sequence of the β3 cytoplasmic tail was observed, and this interaction was blocked by a recombinant protein fragment corresponding to the 47-kDa N-terminal head domain of talin (rTalin-N). In addition, RGD affinity purified platelet αIIbβ3 bound dose-dependently to immobilized rTalin-N, indicating that an integrin-binding site is present in the talin N-terminal head domain. Collectively, these studies demonstrate that the NPLY β-turn motif regulates post-ligand binding functions of αIIbβ3 in a manner independent of talin interaction. Moreover, talin was shown to bind through its N-terminal head domain to the membrane-proximal sequence of the β3cytoplasmic tail.

Integrins are transmembrane ␣⅐␤ receptor complexes involved in numerous physiological processes such as embryogenesis, angiogenesis, immune response, and hemostasis (1). It is generally agreed that binding of adhesive proteins to integrins initiates a series of post-ligand occupancy events that are dependent on an intact cytoskeleton (2). On blood platelets, integrin ␣ IIb ␤ 3 is the most prominent adhesion receptor and plays an essential role in platelet aggregation and the retraction of platelet-rich fibrin clots (3,4). At sites of vascular injury, stimulation of platelets with physiological agonists is thought to result in a change of the native ␣ IIb ␤ 3 conformation from an inactive to an active state which is competent to bind soluble fibrinogen. Following fibrinogen binding to ␣ IIb ␤ 3 and platelet aggregation, this integrin becomes associated with the cytoskeleton (5)(6)(7). Increasing evidence suggests that the interaction between ␣ IIb ␤ 3 and the platelet cytoskeleton is crucial for ␣ IIb ␤ 3 -dependent post-ligand occupancy events such as clot retraction, protein tyrosine phosphorylation, and receptor clustering and redistribution (8 -12). In this regard, it has been shown that cytochalasins, which block actin polymerization, have profound inhibitory effect on these processes (10,12,13).
Biochemical and functional studies revealed that the ␤ 3 cytoplasmic tail modulates the ligand binding affinity state of the receptor (14 -16) and serves as an assembly site for cytoskeletal proteins and signaling molecules (17)(18)(19)(20). Thus, truncation of the cytoplasmic sequence of ␤ 3 , but not ␣ IIb , completely abolished the ability of ␣ IIb ␤ 3 to initiate cell spreading and focal adhesion formation upon ␣ IIb ␤ 3 -mediated cell adhesion to fibrinogen (21). Moreover, deletion of the ␤ 3 cytoplasmic tail also rendered the ␤ 3 integrins in transfected Chinese hamster ovary (CHO) 1 cells incapable of supporting clot retraction (21) and phosphorylation of focal adhesion kinase (22,23). In addition, site-directed mutagenesis of integrin ␤ 1 and ␤ 3 cytoplasmic domains defined the membrane-proximal region (cyto-1) and two highly homologous NXXY motifs (cyto-2 and cyto-3) as being important in the regulation of integrin affinity states, cell spreading, and receptor recruitment into adhesion plaques (24,25) (Fig. 1). Recently, it has been shown that the tyrosine residues in the NXXY sites of the ␤ 3 cytoplasmic tail are phosphorylated upon thrombin-induced platelet aggregation, and their phosphorylation state may regulate the binding of signaling molecules such as SHC and GRB2 (19), as well as cytoskeletal proteins including myosin (20).
Prediction of protein secondary structures by the Chou and Fasman algorithm (26) indicate that the N 744 PLY sequence forms a tight ␤-turn while the N 756 ITY sequence resides within a ␤-sheet structure. In addition, a N744Q/P745A double mutation would greatly diminish the probability of forming the ␤-turn within the ␤ 3 cytoplasmic tail. To examine the functional significance of the N 744 PLY ␤-turn motif, we generated this N744Q/P745A ␤ 3 mutant construct and co-transfected it with a wild type ␣ IIb construct into CHO cells. The abilities of cells expressing mutant ␤ 3 integrins to support cell spreading, focal adhesion formation, and protein tyrosine phosphorylation were determined. Since the corresponding region in the ␤ 1 cytoplasmic tail has been implicated in talin binding function (27,28), we also examined the ability of the purified mutant receptor to interact with talin in a solid phase binding assay. In these experiments, we found that the N744Q/P745A mutation in ␤ 3 blocked post-ligand binding functions but did not affect ␣ IIb ␤ 3 binding to talin. This led us to investigate further the talin-binding site within the ␤ 3 cytoplasmic tail as well as the integrin-binding site within talin. Results of our study show that the N-terminal head domain of talin interacts with the membrane-proximal region of the ␤ 3 cytoplasmic sequence.

MATERIALS AND METHODS
Antibodies, Peptides, and Reagents-The monoclonal antibodies mAb15 (29) and AP-2 (30) were generous gifts of Dr. M. H. Ginsberg and Dr. T. J. Kunicki, respectively, of the Scripps Research Institute, La Jolla, CA. The anti-talin monoclonal antibody 8d4 (31) and streptavidin were obtained from Sigma. The 6ϫ His monoclonal antibody was from CLONTECH Laboratories, Inc. Human fibrinogen (grade L) was purchased from KabiVitrum, Inc. For flow cytometry studies, AP-2, was conjugated with fluorescein isothiocyanate (FITC) using FITC-celite (Sigma) as described (32). For solid phase binding studies, mAb15, 8d4, and streptavidin were labeled with carrier-free Na 125 I (Amersham Pharmacia Biotech) using the IODO-BEADS iodination reagent (Pierce) to a specific activity of approximately 3 Ci/g for the antibodies and 0.5 Ci/g for streptavidin. Peptides were synthesized by solid phase synthesis using an Applied Biosystems model 431 peptide synthesizer or were obtained from Research Genetics, Inc. Synthetic peptides were represented by the single letter amino acid code corresponding to their sequences (33). Table I shows the amino acid sequences of the ␤ 3 cytoplasmic domain peptides used in the present study.
The N744Q/P745A mutant ␤ 3 construct was co-transfected with the wild type ␣ IIb construct (CD2b) into CHO cells by liposome-mediated transfection. Briefly, 2 g of each construct were incubated with 20 l of LipofectAMINE reagent (Life Technologies Inc.) in 180 l of unsupplemented DMEM at 22°C for 20 min, and 3.8 ml of unsupplemented DMEM was then added. The DNA-liposome complexes were overlaid onto CHO cells and incubated for 6 h at 37°C. The cells were washed with phosphate-buffered saline (PBS, 10 mM sodium phosphate, pH 7.4, 0.15 M NaCl) and incubated in complete medium at 37°C for 48 h with a change of medium at 24 h. The cells were analyzed for transient expression of mutated ␣ IIb ␤ 3 by flow cytometry using FITCconjugated AP-2, an anti-␣ IIb ␤ 3 complex specific monoclonal antibody (30). 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 that were high expressors of the mutant ␣ IIb ␤ 3 . The production of the control A5 cell line expressing wild type ␣ IIb ␤ 3 has been described previously (37).
Phase Contrast and Immunofluorescence Microscopy-Coverslips were coated with fibrinogen (100 g/ml in PBS) overnight at 4°C and then blocked with 1% BSA for 1 h at 22°C. CHO cells were harvested and seeded onto the fibrinogen-coated coverslips in a serum-free medium for 3 h. Adherent cells were fixed with 2% paraformaldehyde for 10 min at 4°C and subsequently neutralized with NH 4 Cl/Tris-buffered saline, pH 7.4. For phase contrast microscopy, the specimens were examined with a Leitz microscope using a 10ϫ dry objective and photographed with Eastman Kodak T max 400 film.
For indirect immunofluorescent staining of focal contacts, fixed adherent cells were permeabilized with 0.5% Triton X-100 in PBS for 10 min. Following incubation with the anti-␤ 3 monoclonal antibody mAb15 for 1 h, cells were stained with FITC-conjugated goat anti-mouse IgG (Sigma) for 1 h. After washing, the samples were mounted on a droplet of FITC-Guard TM (Testog Inc.) and viewed with a Jenaval phase/fluorescence microscope equipped with an HBO 50-watt mercury lamp, and an IVF1 epifluorescence condenser with BP 485 and 546 nm excitation filters and BP 520 -560 and LP 590 barrier filters. CHO cells were photographed with Eastman Kodak T max 400 film.
Protein Tyrosine Phosphorylation-Tissue culture plates (100 mm, Falcon) were coated with 100 g/ml fibrinogen in PBS overnight at 4°C and then blocked with 0.5% BSA for 2 h at 37°C. Harvested cells (1 ϫ 10 7 cells), suspended in 1 ml of 20 mM HEPES, pH 7.4, containing 137 mM NaCl, 2.7 mM MgCl 2 , 5.6 mM glucose, and 3.3 mM NaH 2 PO 4 , were added to the fibrinogen-coated plates and allowed to adhere for 90 min at 37°C. Non-adherent cells on control BSA-coated plates and adherent cells on fibrinogen-coated plates were lysed with RIPA buffer (10 mM Tris-HCl, 158 mM NaCl, 1% sodium deoxycholate, 1% Triton X-100, 0.1% SDS, 1 mM Na 2 EGTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 , 100 KIU/ml aprotinin, pH 7.2). The lysates were clarified by centrifugation at 12,500 ϫ g at 4°C for 30 min, and their protein concentrations were determined by the BCA protein assay (Pierce). Proteins were separated by electrophoresis on 7% polyacrylamide gels under reducing conditions and transferred onto nitrocellulose membranes. In some experiments, the cell lysates were subjected to immunoprecipitation using an anti-FAK monoclonal antibody (Transduction Laboratories) coupled to protein G-Sepharose (GammaBind Plus Sepharose, Amersham Pharmacia Biotech). Protein tyrosine phosphorylation was analyzed by immunoblotting using the anti-phosphotyrosine monoclonal antibody PY20 (ICN) and detected by enhanced chemiluminescence (Pierce) as described (10,23).
The concentrations of ␣ IIb ␤ 3 and talin were determined using the BCA protein assay (Pierce) with BSA as the standard. In the binding studies, purified talin (100 g/ml, 50 l/well) was coated onto microtiter wells (Immulon 2 Removawell strips, Dynex Technologies, Inc.) for 48 h at 4°C. After blocking with 3% BSA, purified ␣ IIb ␤ 3 was added and incubated for 4 h at 37°C. Following extensive washing, bound receptor was detected with 125 I-labeled mAb15 as described (18).
For the binding of talin to synthetic peptides encompassing partial sequences of the ␤ 3 cytoplasmic tail, the peptides (0.5 mM, 50 l/well) were coated onto microtiter wells (Immulon 2, Removawell strips) overnight at 22°C. After blocking the peptide-coated wells with 3% BSA, purified talin was added and incubated for 4 h at 37°C. Bound talin was detected with 125 I-labeled 8d4, a monoclonal antibody directed against the 190-kDa C-terminal tail domain of talin (31). To determine the amounts of immobilized peptides, the peptide-coated wells were blocked with 1% gelatin and incubated overnight at 22°C with 10 mM EZ-link TM PEO-maleimide activated biotin (Pierce), which reacts with the free sulfhydryl group of the N-terminal cysteine residue of the peptides. After extensive washing, the amounts of coupled biotin were quantitated with 0.1 M 125 I-labeled streptavidin.
In some experiments, talin binding to peptides with a biotinylated N-terminal lysine residue (Research Genetics, Inc., Table I) was examined. The biotinylated peptides (0.5 mM, 50 l/well) were coupled onto Reacti-Bind TM NeutrAvidin-coated polystyrene microtiter wells (Pierce) at 37°C for 2 h, and the binding of soluble talin to the immobilized peptides was performed as described above. Alternatively, talin (50 g/ml, 50 l/well) was immobilized onto Immulon 2 microtiter wells overnight at 22°C. After blocking with 3% BSA, biotinylated peptides were allowed to bind to the immobilized talin for 3 h at 37°C. Bound peptides were detected with 0.5 M 125 I-labeled streptavidin.

Generation of a Recombinant Protein Fragment Corresponding to the 47-kDa N-terminal Head Domain of Human Talin (rTalin-N)-At the time of our experiments, the human talin sequence was not available.
Thus, based on the mouse sequence (GenBank TM accession number L46861), we performed 5Ј-RACE reaction to determine the nucleotide sequence encoding the N terminus of human talin (39). Briefly, total cellular RNA was extracted from human skin fibroblasts (CCD 1064Sk, ATCC, Rockville, MD) using the RNeasy TM mini kit (Qiagen) and reversed-transcribed using primer a, 5Ј-CACCTGGAAATTCTCAGGAC-CAGAGGC-3Ј, derived from the mouse sequence. 5Ј-RACE was performed using the 5Ј/3Ј-RACE Kit (Mannheim Boehringer) according to the manufacturer's instructions. The coding sequence of the human talin N terminus was determined to be 5Ј-ATGGTTGCACTTTCACT-GAAGATCAGCATTGGG-3Ј, and a forward primer b corresponding to this sequence was synthesized. To create the rTalin-N fragment, PCR was performed using Pfu and the oligonucleotide pair b and a with the NdeI and XhoI restriction sites added to the 5Ј-ends of these primers, respectively. The rTalin-N fragment was cloned into a pET-30-a(ϩ) vector (Novagen) as a fusion protein with a hexahistidine tag at its C terminus and transformed into BL21(DE3) competent cells (Novagen). DNA sequencing revealed that the human talin N-terminal domain nucleotide sequence was 91.6% identical to the mouse sequence. As compared with the mouse talin protein (40), there are 6 changes out of 465 amino acid residues in this region: M31I, L40P, N45S, D139E, G141I, and S352N (mouse to human). The expression of rTalin-N his-tidine-tagged fusion protein was induced in transformed bacteria with isopropylthio-␤-D-galactoside, and the histidine-tagged fusion protein was purified by chromatography on Ni 2ϩ resin (Novagen). The molecular mass of the purified fusion protein was ϳ52 kDa, and it contained 32 amino acid residues downstream of the calpain cleavage site plus the histidine tag (Fig. 2, lane 2). To produce the 47-kDa rTalin-N fragment, the isolated fusion protein was digested with m-calpain (Sigma), and the released ϳ5-kDa fragment was removed by extensive dialysis against 20 mM Tris acetate, pH 7.6, 20 mM NaCl, 0.1 mM EDTA, 1 mM EGTA, 0.1% ␤-mercaptoethanol (Fig. 2, lane 3).

RESULTS
The NPLY Motif in the ␤ 3 Cytoplasmic Tail Is Essential for ␣ IIb ␤ 3 -mediated Post-ligand Binding Functions-By using the Chou and Fasman and Gor II methods to predict protein secondary structures, it has previously been reported that the N 744 PLY sequence represents a potential ␤-turn motif in the cytoplasmic tail of the integrin ␤ 3 subunit (41). To examine the functional significance of this ␤-turn motif, we performed sitedirected mutagenesis and replaced the NPLY sequence with QALY which was predicted to alter the secondary structure of this region. The N744Q/P745A ␤ 3 mutant construct was cotransfected with a wild type ␣ IIb construct into CHO-K1 cells, and stable clonal lines expressing the mutant receptor were developed for functional studies. All experiments described below were performed with at least three clonal cell lines yielding similar results. For simplicity, only those obtained with the D4 cell line are presented. Additionally, the control A5 cell line bearing wild type ␣ IIb ␤ 3 in CHO-K1 cells (37) was used for comparison. In preliminary experiments, we determined the levels of surface expression of the ␤ 3 integrins on A5 and D4 cells by flow cytometry. Both A5 and D4 cells bound similar amounts of FITC-conjugated AP-2, a complex-specific anti-␣ IIb ␤ 3 monoclonal antibody (30), indicating similar expression of ␣ IIb ␤ 3 on these cell lines (mean fluorescence intensity: A5, 78.0 Ϯ 4.9; D4, 76.0 Ϯ 5.0, mean Ϯ S.D., n ϭ 3). By using these two cell lines, we examined the effect of mutation on ␣ IIb ␤ 3mediated cell adhesion and spreading, as well as focal adhesion formation. As shown in Fig. 3, both A5 and D4 cells bound to fibrinogen-coated coverslips. In the presence of 1 mM GRGDSP, adhesion of both cell types was inhibited by Ͼ80% (data not shown). Examination of the morphology of the adherent cells revealed that most of the adherent A5 cells became fully spread (Fig. 3A, left) with ␣ IIb ␤ 3 clustered into punctate focal adhesion plaques (Fig. 3B, left). In contrast, D4 cells remained round (Fig. 3A, right) with a diffused distribution of ␣ IIb ␤ 3 (Fig. 3B,  right).
Cell adhesion through integrins initiates signaling events such as protein tyrosine phosphorylation, and this process is thought to be dependent on the interaction between integrins and the cytoskeleton (42). Therefore, we examined whether the NPLY sequence in ␤ 3 integrins modulates protein tyrosine phosphorylation. In these experiments, A5 or D4 cells were allowed to adhere to fibrinogen-or BSA-coated tissue culture plates. Non-adherent and adherent cells were recovered, lysed, and subjected to immunoblotting with PY20, an anti-phosphotyrosine antibody. As shown in  shows that similar amounts of ␤ 3 integrins were present in all samples.
In these experiments, FAK in cell lysates was isolated by immunoprecipitation and subjected to immunoblotting with PY20. As shown in Fig. 4C, FAK in adherent A5 cells (lane 2), but not D4 cells (lane 4), was phosphorylated on tyrosine residues. As controls, parallel samples were subjected to immunoblotting with an anti-FAK monoclonal antibody, confirming that similar amounts of FAK were being immunoprecipitated from adherent A5 and D4 cells (Fig. 4D). Collectively, these results indicate that the structural integrity of the NPLY motif in the ␤ 3 cytoplasmic tail is essential for mediating cell spreading, focal adhesion formation, and protein tyrosine phosphorylation.
The N744Q/P745A Mutation Does Not Affect the Binding of ␣ IIb ␤ 3 to Immobilized Talin-By using a solid phase binding assay, we previously demonstrated direct binding of RGD affinity purified ␣ IIb ␤ 3 to immobilized talin (18). Since the WDT-GENPIYK peptide derived from the ␤ 1 cytoplasmic tail containing the NPXY motif has been shown to block talin binding to the avian integrin complex (27), we examined the effect of the N744Q/P745A mutation on the interaction of ␣ IIb ␤ 3 with talin. In these studies, ␣ IIb ␤ 3 was purified from lysates of A5 and D4 cells by RGD affinity chromatography and allowed to bind to immobilized talin for 4 h at 37°C. Following washing, bound receptor was detected with 125 I-labeled mAb15. Fig. 5A (top) shows that similar amounts of wild type and mutant ␣ IIb ␤ 3 bound to talin-coated wells but not to control wells coated with BSA. As a control, we examined the binding of ␣ IIb ␤ 3 with truncations of both ␣ IIb (⌬991) and ␤ 3 (⌬728) cytoplasmic sequences (43). As expected, the ␣ IIb ⌬991/␤ 3 ⌬728 truncation mutant failed to bind to immobilized talin. To ascertain that equal concentrations of ␣ IIb ␤ 3 were added to the microtiter wells, the samples were subjected to immunoblotting with 125 Ilabeled mAb15 and found to contain similar amounts of integrin ␤ 3 subunit (Fig. 5A, bottom). To examine further whether the N744Q/P745A mutation affects the affinity of ␣ IIb ␤ 3 -talin interaction, we performed binding isotherms with varying concentrations of ␣ IIb ␤ 3 . As shown in Fig. 5B, at all input concentrations of the receptor, similar extents of binding of wild type and mutant ␣ IIb ␤ 3 to immobilized talin were observed. Thus, these results demonstrated that the cytoplasmic domain of recombinant ␣ IIb ␤ 3 mediates its interaction with talin; however, the N744Q/P745A mutation of the ␤ 3 cytoplasmic tail has no effect on this process.
Talin Binds to the Membrane-proximal Region of the ␤ 3 Cytoplasmic Tails-To further investigate the role of the NPLY sequence in talin binding, we examined whether talin binds to synthetic peptides encompassing partial sequences of the ␤ 3 cytoplasmic tail. In the integrin ␤ 1 cytoplasmic sequence, three distinct regions, namely cyto-1, cyto-2, and cyto-3, have been implicated in mediating the localization of ␤ 1 integrins in focal adhesions (24). In the present study, Cys-␤ 3 -(722-738) and  2 and 4) were subjected to SDS-PAGE on 7% polyacrylamide gels under reducing conditions. Protein tyrosine phosphorylation was analyzed by immunoblotting with PY20 and detected by enhanced chemiluminescence. B, to determine the amounts of ␤ 3 integrins in the samples of A, identical samples were subjected to SDS-PAGE under non-reducing conditions and immunoblotted with 125 I-labeled mAb15. C, FAK in the cell lysates described in A was isolated by immunoprecipitation with an anti-FAK mAb, subjected to SDS-PAGE under reducing conditions, and immunoblotted with PY20. D, the amounts of immunoprecipitated FAK were analyzed by immunoblotting with an anti-FAK monoclonal antibody. Cys-␤ 3 -(737-748) peptides with partial ␤ 3 cytoplasmic sequences containing the corresponding cyto-1 and cyto-2 regions, respectively, were synthesized and coated onto microtiter wells (Fig. 1). Since these peptides contain a free sulfhydryl group in their N-terminal cysteine residue, their coating efficiencies were estimated by the binding of PEO-maleimideactivated biotin, followed by detection with 125 I-labeled streptavidin. As shown in Table I, similar amounts of both peptides were coated onto microtiter wells. To examine the binding of talin to the immobilized peptides, purified talin was incubated with peptide-coated wells at 37°C for 4 h, followed by detection with 125 I-labeled 8d4, an anti-talin monoclonal antibody. Fig. 6 shows that talin bound in a dose-dependent manner to Cys-␤ 3 -(722-738)-coated wells but not to wells coated with Cys-␤ 3 -(737-748) or BSA. Since Cys-␤ 3 -(737-748) contains the N 744 PLY sequence, these results indicate that this motif is by itself not sufficient in mediating talin binding to ␤ 3 integrins. To demonstrate the specificity of talin binding to Cys-␤ 3 -(722-738), we used a scrambled Cys-␤ 3 -(722-738) peptide that was found to be much less effective in supporting talin binding at the highest input talin concentration of 200 g/ml (Fig. 6, open square).
To eliminate the possibility that talin failed to bind to the Cys-␤ 3 -(737-748) peptide was due to obstruction of the NPLY motif resulting from peptide immobilization, we examined talin binding to biotinylated peptides that have been coupled to NeutrAvidin-coated microtiter wells (Table I). In control experiments, successful coupling of the biotin-Lys-␤ 3 -(737-748) peptide to NeutrAvidin-coated wells was monitored by an enzymelinked immunosorbent assay (ELISA) using an anti-peptide polyclonal antibody raised against Cys-␤ 3 -(737-748) (data not shown). Also, since the biotin group was conjugated onto the N-terminal lysine residue of the peptide, this would allow the C-terminal NPLY motif to be protruded from the NeutrAvidincoated wells. Consistent with the above observations, talin bound efficiently and dose-dependently to wells coated with biotin-Lys-␤ 3 -(716 -738) but poorly to wells coated with biotin-Lys-␤ 3 -(737-748) (Fig. 7A). Conversely, using 125 I-labeled streptavidin to detect the binding of biotinylated peptides to immobilized talin, we found that biotin-Lys-␤ 3 -(716 -738), but not biotin-Lys-␤ 3 -(737-748), bound in a dose-dependent manner to talin-coated wells (Fig. 7B). As expected, neither peptide bound to control wells coated with BSA. Together, these findings indicate that talin interacts with the ␤ 3 cytoplasmic tail at its membrane-proximal region which does not contain the NPLY regulatory motif of post-ligand binding functions.  (40,44). Ezrin/Radixin/Moesin (ERM) in this protein family have been shown to bind to basic amino acid clusters in the juxtamembrane cytoplasmic domains of membrane proteins (45). Thus, we postulated that talin may bind through its 47-kDa head domain to the ␤ 3 -(722-738) sequence. To test this possibility, we examined the inhibitory effect of a recombinant 47-kDa fragment of human talin Nterminal domain (rTalin-N) on the binding of intact talin to immobilized Cys-␤ 3 -(722-738) peptide. Fig. 8A shows that preincubation of Cys-␤ 3 -(722-738)-coated wells with 15 g/ml (0.3 M) rTalin-N completely blocked talin binding to the peptide, suggesting that the interaction is mediated through the head domain of talin. To ascertain that the rTalin-N fragment binds directly to the Cys-␤ 3 -(722-738) peptide, we performed an ELISA using the His-tagged rTalin-N fusion protein and a detecting monoclonal antibody directed against the hexahistidine tag. As shown in Fig. 8B, the recombinant talin head fusion protein bound directly to wells coated with Cys-␤ 3 -(722-738), whereas minimal binding to control wells coated with BSA was observed.
To examine further the role of the talin N-terminal head domain in binding to integrin, we performed direct binding of purified platelet ␣ IIb ␤ 3 to immobilized rTalin-N as described (18). Fig. 9 shows that ␣ IIb ␤ 3 bound saturably to wells coated with rTalin-N but not to control wells coated with BSA. Moreover, half-saturation binding was observed at approximately 12 nM ␣ IIb ␤ 3 , which is similar to that observed for ␣ IIb ␤ 3 binding to intact talin (18). Thus, these results indicated that the talin head domain contains a binding site for integrin ␣ IIb ␤ 3 and interacts with the membrane-proximal region of the ␤ 3 cytoplasmic tail. DISCUSSION Fibrinogen binding to integrin ␣ IIb ␤ 3 on activated platelets results in platelet aggregation which is accompanied by a series of post-ligand binding events dependent on the interaction between ␣ IIb ␤ 3 and the platelet cytoskeleton. In the present study, we investigated the functional role of the putative NPLY ␤-turn motif in the ␤ 3 cytoplasmic domain by site-directed mutagenesis. The major findings of these studies are as follows: 1) the N744Q/P745A double mutation in ␤ 3 does not affect cell attachment to fibrinogen but blocks post-ligand binding functions of ␣ IIb ␤ 3 including cell spreading, focal contact formation, FIG. 5. Binding of purified recombinant ␣ IIb ␤ 3 to immobilized talin. A, integrin ␣ IIb ␤ 3 was purified from A5, D4, and ␣ IIb ⌬991/␤ 3 ⌬728 cells by RGD affinity chromatography. In the top panel, purified ␣ IIb ␤ 3 was added to microtiter wells coated with talin (200 g/ml, 50 l/well) or BSA, and binding proceeded for 4 h at 37°C. Bound receptor was detected with 125 I-mAb15 (50 nM). Data shown represent means of triplicate determinations from two experiments, and error bars represent standard deviations. In the bottom panel, the amounts of integrin ␤ 3 subunit in the samples were analyzed by immunoblotting with 125 I-labeled mAb15. B, varying concentrations of wild type and mutant ␣ IIb ␤ 3 , isolated from A5 and D4 cells, respectively, were added to microtiter wells coated with talin or BSA. Binding proceeded as described above. Results are means of triplicate determinations of one experiment, and error bars represent standard deviations. and protein tyrosine phosphorylation; 2) the NPLY sequence in the integrin ␤ 3 cytoplasmic tail is not by itself sufficient to mediate the interaction between ␣ IIb ␤ 3 and talin; and 3) talin binds through its N-terminal head domain to the membraneproximal region of the ␤ 3 cytoplasmic sequence. Collectively, these findings indicate that the NPLY motif mediates postligand binding functions of ␣ IIb ␤ 3 in a manner independent of ␣ IIb ␤ 3 -talin interaction.
The NPXY motif is highly conserved among the cytoplasmic sequences of integrin ␤ subunits and is present in several non-integrin receptors (46). Previous mutational studies on the N 744 PLY sequence in ␤ 3 integrins focused on the tyrosine residue since its phosphorylation state may regulate receptor interaction with cytoskeletal proteins and/or signaling molecules (19,20). However, the secondary structure of this sequence may also be important in mediating protein-protein interaction(s). In this regard, NMR analysis of synthetic peptides encompassing the NPVY internalization sequence of the LDL receptor confirmed its predicted secondary structure as a type I ␤-turn (47). Moreover, asparagine and proline are strong ␤-turn promoters and are frequently found in the first two positions of type I ␤-turns in proteins (48). Similarly, the aromatic side chain of the tyrosine residue has been suggested to contribute to the overall structural stability of the ␤-turn motif (47). To address the structural importance of the N 744 PLY sequence in the integrin ␤ 3 subunit without modifying the tyrosine residue, we mutated asparagine to glutamine and proline to alanine. These amino acid substitutions were predicted to alter the type I ␤-turn into an ␣-helical structure.
It has been shown that replacement of Asn 744 or Tyr 747 , but not Pro 745 , with alanine in the NPLY sequence of the integrin ␤ 3 subunit resulted in a complete loss of ␣ v ␤ 3 -mediated cell adhesion to vitronectin (14). However, comparable adhesion of A5 and D4 cells bearing wild type and mutant ␣ IIb ␤ 3 to fibrin-ogen was observed, indicating that the N744Q/P745A mutation does not affect extracellular ligand binding activities of the receptor. One possible explanation of the difference in cell adhesion between our N744Q/P745A mutant and the N744A mutant in the earlier study is that glutamine for asparagine is a much more conservative substitution than alanine for asparagine. Thus, N744Q, in conjunction with the permissible P745A substitution, has no effect on cell attachment to fibrinogen mediated by the mutant ␣ IIb ␤ 3 . Nonetheless, in agreement with earlier studies that demonstrated that Y747A and N744A single mutations in ␤ 3 abolished cell spreading (14,25), we found that adherent D4 cells remained round and were not able to induce focal contact formation. Interestingly, point mu- Biotin-KAKWDTANNPLYK a The amounts of cysteine-containing peptides coated onto microtiter wells were determined by the binding of PEO-maleimide-activated biotin followed by detection with 125 I-labeled streptavidin as described under "Materials and Methods." Data shown are means Ϯ S.D. of triplicate determinations from two experiments.
FIG. 6. Binding of talin to immobilized peptides containing partial sequences of the integrin ␤ 3 cytoplasmic tail. Microtiter wells were coated with the indicated peptides (0.5 mM, 50 l/well), and blocked with 3% BSA; the amounts of peptides coated onto the wells were shown in Table I tations of the corresponding ␤-turn motif in the integrin ␤ 1 cytoplasmic tail had similar inhibitory effects on cell spreading and focal adhesion (24). These results suggest a role for the conserved NPXY ␤-turn motif in mediating interaction(s) with cytoskeletal proteins and/or signaling molecules necessary for these cellular processes.
A major outside-in signaling event following cell adhesion through integrins is protein tyrosine phosphorylation. In this regard, it has been shown that a number of proteins including FAK becomes tyrosine-phosphorylated in aggregated platelets, and this process is blocked by cytochalasins indicating an involvement of integrin-cytoskeleton interaction (10). By using single subunit chimeric receptors containing the interleukin-2 receptor extracellular and transmembrane domains connected to mutated ␤ 3 cytoplasmic sequences, Tahiliani et al. (22) reported that the NPLY motif is essential for the chimeric receptor to induce FAK phosphorylation. Our present finding that N744Q/P745A mutation in the native integrin ␤ 3 subunit abolished ␣ IIb ␤ 3 -dependent FAK phosphorylation in D4 cells adherent to fibrinogen is in agreement with these results. Furthermore, tyrosine phosphorylation of other proteins with molecular masses of 70 -90 kDa was similarly affected, suggesting that the NPLY motif in ␤ 3 is crucial for this integrinmediated signaling event.
Talin and ␣-actinin have been suggested to serve as linkage proteins between integrins and actin filaments (17,49). Mutational studies have identified three amino acid clusters (i.e. cyto-1, cyto-2, and cyto-3, see Fig. 1) in the ␤ 1 integrin cytoplasmic sequence that are important for integrin localization to focal adhesions (24). An ␣-actinin-binding site has been mapped to the cyto-1 region (50). The NPIY motif at cyto-2 is generally believed to mediate talin binding since a synthetic peptide encompassing this sequence was found to inhibit talin interaction with the avian integrin complex (27); however, it is unclear whether this inhibitory effect was due to a direct interaction between talin and the ␤ 1 peptide. In a more recent study, talin has been shown to bind to affinity matrices containing ␤ 1A or ␤ 1D cytoplasmic sequences, and the binding was abolished by a Y788A substitution at the NPIY 788 motif of the ␤ 1A sequence (28). Although these findings suggest that the NPXY motif plays a regulatory role in talin binding to the ␤ 1 peptides, whether talin interacts directly with this sequence in integrin receptors has not been conclusively demonstrated. In fact, it has been reported that talin bound to antibody-captured integrin ␣ 5 ␤ 1 with either Y788S or Y800S point mutation in the ␤ 1A cytoplasmic tail, suggesting that these motifs do not constitute the talin-binding site of the ␣ 5 ␤ 1 integrin (51). Consistent with the latter mutational study, we found that both wild type and N744Q/P745A-mutated ␣ IIb ␤ 3 bound equally well to immobilized talin in a solid phase binding assay. Thus, in intact integrin receptors, the NPXY structural motif is not essential in mediating talin binding.
In accord with the results of the mutational studies, we found that talin failed to bind to synthetic peptides encompassing the ␤ 3 -(737-748) sequence that contains the NPLY ␤-turn motif. In contrast, specific interaction of talin with peptides containing the ␤ 3 -(722-738) membrane-proximal sequence of the ␤ 3 cytoplasmic tail was observed. These findings indicate that a talin-binding site resides in the cyto-1 rather than the cyto-2 region of the ␤ 3 cytoplasmic domain. Similarly, we found that talin also binds to a synthetic peptide containing the cyto-1 region of the ␤ 1 cytoplasmic domain. 2 Thus, these results suggest that the conserved cyto-1 sequence in different integrin ␤ subunits contains a common recognition site for this cytoskeletal protein.
Talin is a single chain polypeptide consisting of a 47-kDa N-terminal globular head domain and a 190-kDa C-terminal tail domain that can be separated be proteolytic cleavage with calpain (38). Whereas vinculin-and actin-binding sites have been localized to the talin C-terminal tail domain (52-54), the

FIG. 8. Interaction of rTalin-N with immobilized ␤ 3 -(722-738).
A, inhibition of talin binding to Cys-␤ 3 -(722-738) by rTalin-N. Microtiter wells coated with Cys-␤ 3 -(722-738) or BSA were preincubated with or without rTalin-N (15 g/ml) for 2 h at 37°C as indicated. After removing rTalin-N, purified intact talin (200 g/ml) or its carrier buffer was added and binding proceeded as described in the legend of Fig. 6. B, binding of His-tagged rTalin-N to Cys-␤ 3 -(722-738). The fusion protein (5 g/ml) was added to the peptide-coated wells and incubated for 2 h at 37°C. After washing, bound protein was detected by an ELISA using the 6ϫ His monoclonal antibody, followed by a goat anti-mouse IgG conjugated to horseradish peroxidase. Data shown are means of triplicate determinations, and error bars represent standard deviations. integrin recognition site(s) in talin has not been identified. Sequence analyses of the primary structures of a number of membrane-cytoskeleton linkers including talin, band 4.1, Ezrin, Radixin, and Moesin reveal the presence of a homologous FERM domain near the N termini of these proteins (40,44). Our findings that a recombinant fragment of the talin head domain bound to the Cys-␤ 3 -(722-738) peptide and blocked talin binding to this sequence are consistent with the suggestion that the FERM domain provides an attachment site to transmembrane proteins. In addition, we found that purified ␣ IIb ␤ 3 bound saturably to the recombinant talin head fragment, reinforcing the notion that an integrin-binding site in talin resides within its N-terminal head domain.
Recently, Ezrin, Radixin, and Moesin have been shown to bind to clusters of three basic amino acid residues in the juxtamembrane cytoplasmic domains of CD44, CD43, and ICAM-2 (45). It is interesting to note that basic amino acid residues are also found in the juxtamembrane regions of most integrin ␣ and ␤ cytoplasmic tails. Furthermore, it has been shown that the cytoplasmic sequences of ␣ IIb and ␤ 3 interact with each other to form defined tertiary structures (41,55,56). Thus, it is conceivable that such interaction results in the formation of basic amino acid clusters to which the FERM domain of talin interacts. It has previously been reported that both in vitro and in vivo talin interaction with integrin ␣ 5 ␤ 1 is dependent on ligand occupancy of the receptor (51,57), suggesting that the talinbinding site in the cytoplasmic domains of integrins is dynamically regulated. In this regard, we recently showed that extracellular ligand binding induces a transmembrane conformational change in integrin ␣ IIb ␤ 3 (58). It is therefore an intriguing possibility that such conformational change in the receptor cytoplasmic domain may expose cryptic basic amino acid clusters in its membraneproximal region, thereby modulating the affinity of integrin-talin interaction. Thus, whether the juxtamembrane basic amino acid residues in integrin cytoplasmic domains play a role in talin binding merits further investigation.
The observation that mutations of the NPLY motif in the integrin ␤ 3 subunit block post-ligand binding functions but not talin binding suggests that ␣ IIb ␤ 3 -talin interaction is not sufficient for inducing outside-in signaling of ␣ IIb ␤ 3 . At present, the mechanisms by which the NPLY motif regulates post-ligand binding functions of ␣ IIb ␤ 3 remain elusive. Recently, it has been shown that the ␤ 3 cytoplasmic tail becomes phosphorylated on tyrosine residues upon ligand binding and cell aggregation, and several cytoskeletal proteins (e.g. myosin) and signaling molecules (e.g. GRB2 and SHC) interact with the NXXY motifs in a phosphorylation-dependent manner (19,20). Thus, the interaction of these proteins, rather than talin, with the phosphorylated NPLY motif in the ␤ 3 cytoplasmic tail may be important for inducing ␣ IIb ␤ 3 -dependent post-ligand binding events.