Integrin-mediated Activation of Focal Adhesion Kinase Is Independent of Focal Adhesion Formation or Integrin Activation

Integrin αIIbβ3functions as the fibrinogen receptor on platelets and mediates platelet aggregation and clot retraction. Among the events that occur during either “inside-out” or “outside-in” signaling through αIIbβ3 is the phosphorylation of focal adhesion kinase (pp125FAK) and the association of pp125FAK with cytoskeletal components. To examine the role of pp125FAK in these integrin-mediated events, pp125FAK phosphorylation and association with the cytoskeleton was determined in cells expressing two mutant forms of αIIbβ3: αIIbβ3(D723A/E726A), a constitutively active integrin in which the putative binding site for pp125FAK is altered, and αIIbβ3(F727A/K729E/F730A), in which the putative binding site for α-actinin is altered. Both mutants were expressed on the cell surface and were able to bind ligand, either spontaneously or upon activation. Whereas cells expressing αIIbβ3(D723A/E726A) were able to form focal adhesions and stress fibers upon adherence to fibrinogen, cells expressing αIIbβ3(F727A/K729E/F730A) adhere to fibrinogen, but had reduced focal adhesions and stress fibers. pp125FAK is recruited to focal adhesions in adherent cells expressing αIIbβ3(D723A/E726A) and is phosphorylated in adherent cells or in cells in suspension in the presence of fibrinogen. In adherent cells expressing αIIbβ3(F727A/K729E/F730A), pp125FAK was phosphorylated despite reduced formation of focal adhesions and stress fibers. We conclude that activation of pp125FAK can be dissociated from two important events in integrin signaling, the assembly of focal adhesions in adherent cells and integrin activation following ligand occupation.

The cytoplasmic domains of integrins constitute an essential link between the extracellular, ligand-binding domain of the receptor and both signaling and structural mechanisms inside the cell. Ranging in length from 15 amino acids for the ␣ 1 subunit to over 1000 amino acids for the ␤ 4 subunit, the cytoplasmic domains extend inward from the plasma membrane toward various cytosolic components known to be involved in integrin-mediated signaling such as heterotrimeric G proteins, proteins involved in phospholipid metabolism, serine-threonine kinases, tyrosine kinases, calcium transport systems, and cy-toskeleton components (1)(2)(3)(4)(5)(6)(7). Although structurally distinct, the different ␣ and ␤ cytoplasmic domains show areas of regional homology, suggesting a conserved tertiary structure. The nine known ␤ subunits contain a highly conserved membrane-proximal polar region with the consensus sequence HDRREFAKFEKEK, which is separated by a stretch of nine amino acids from a central NPXY motif and a stretch of 16 -25 amino acids from a distal NPXY motif. The NPXY sequence prescribes a signal for clathrin-coated pit-mediated internalization of integral membrane proteins (8) and is a recognition site for Shc, an SH2-containing adaptor protein (9). The spacing of the tyrosine residues in the ␣ 1 , ␤ 3 , and ␤ 6 cytoplasmic tails conforms to the spacing of tyrosine residues in immunoglobulin family tyrosine-based activation motifs, which represent coupling sequences for protein tyrosine kinases and adapter molecules (10). Like the ␤ subunits, the ␣ subunits also contain a conserved membrane-proximal region with the consensus sequence, KXGFFKR, followed in many cases by a sequence predicting a ␤ turn. The structural diversity of the cytoplasmic domains is increased further by the expression of alternatively spliced forms of several subunits, including ␤ 1 (11)(12)(13), ␤ 3 (14), ␣ 3 (15), and ␣ 6 (16).
Several important integrin functions have been linked to sequences in the cytoplasmic domains. The recruitment of integrins into focal adhesions is mediated by the cytoplasmic domains. Solowska and co-workers (17) provided early evidence that deletion of the ␤ 1 subunit cytoplasmic domain blocked integrin association into focal adhesions. Subsequent deletion mapping and random mutagenesis studies have indicated that the sequences required for direction to focal adhesions are localized to several discrete regions in the ␤ cytoplasmic domains, including the dibasic sequence in the membrane-proximal region that is highly conserved in those ␤ cytoplasmic domains that localize to focal adhesions, the NPXY sequences, and the carboxyl terminus (18 -23). Some of the same general regions required for direction to focal adhesions have been implicated in the interaction of ␤ subunit cytoplasmic domains with ␣-actinin (24) and with talin (25)(26)(27), proteins that are components of the cellular cytoskeleton and are present in focal adhesions. Sequences in the ␣ subunit cytoplasmic domains also appear to play a role in normal ligand-dependent localization of integrins to focal adhesions (22,28,29). In addition, integrin cytoplasmic domains play an important role in regulating the affinity of the integrin for ligand. Both ␣ and ␤ subunit cytoplasmic tails appear to be important for affinity modulation. O'Toole et al. (30) found that the affinity of ␣ IIb ␤ 3 for ligand was increased by truncation of the cytoplasmic domain of ␣ IIb . Activation correlated with removal of the highly conserved membrane-proximal GFFKR motif, suggesting that this motif was required to maintain the receptor in a low affinity activation state. Deletion of the membrane-proximal LLITIHD region of the ␤ 3 cytoplasmic domain also produced a high affinity receptor (31). Further studies have identified Asp-723 in ␤ 3 and Arg-995 in ␣ IIb as important and suggested that these two residues form a molecular hinge that regulates receptor affinity (32). The NPXY motif in the ␤ subunit cytoplasmic domain also appeared to be important in the regulation of the affinity state of ␣ IIb ␤ 3 , since point mutations in this region inhibited activation through the ␣ subunit (33). A naturally occurring mutation in the ␤ cytoplasmic domain at serine 752 in a patient with Glanzmann's thrombasthenia inhibited activation of ␣ IIb ␤ 3 (34).
While the association of integrins with focal adhesions has been well established, the factors regulating this association and the mechanisms by which integrins signal cytoskeletal assembly remain uncertain. Focal adhesion kinase (pp125 FAK ), a 125-kDa cytoplasmic tyrosine kinase (35), is a component of focal adhesions and is phosphorylated following integrin-mediated adhesion or integrin clustering (36 -40). Tyrosine phosphorylation of pp125 FAK is an early event in adhesion and phosphorylation persists as long as cells remain attached to substrata (41). Inhibition of pp125 FAK phosphorylation by tyrosine kinase inhibitors is associated with reduced formation of focal adhesions and stress fibers (38,41). Conversely, inhibition of tyrosine phosphatases results in increased phosphorylation of pp125 FAK , which correlated with the assembly of focal adhesions and actin stress fibers (42,43). These observations led to a suggestion that tyrosine phosphorylation of pp125 FAK might be implicated in cytoskeletal assembly and formation of focal adhesions.
The present studies were undertaken to examine further the role of pp125 FAK in integrin function. We have used two unique mutants of the ␤ 3 cytoplasmic domain to examine the role of pp125 FAK in assembly of focal adhesions and in integrin activation. One mutation is in the region of the ␤ 3 cytoplasmic domain, which has been suggested to interact with ␣-actinin and interrupts the association of ␣ IIb ␤ 3 with focal adhesions. The second, which is similar to a recent mutation reported by Hughes et al. (32), is in the putative pp125 FAK binding site in the ␤ 3 cytoplasmic domain and results in a constitutively active integrin. To examine the role of pp125 FAK phosphorylation in its association with focal adhesions and in integrin activation, we have examined the state of pp125 FAK phosphorylation in cells expressing these mutant integrins. The results suggest that pp125 FAK phosphorylation is independent of integrin association with focal adhesions and that activation of ␣ IIb ␤ 3 can occur without inducing pp125 FAK phosphorylation.

MATERIALS AND METHODS
Antibodies and Reagents-AP2, an ␣ IIb ␤ 3 complex-specific monoclonal antibody (44), A2A9, a complex-specific monoclonal antibody (45), Tab, an ␣ IIb -specific monoclonal antibody (46), and PAC1, the activated ␣ IIb ␤ 3 complex-specific monoclonal antibody (47)  DNA Constructs-␤ 3 and ␣ IIb cDNA were isolated from a cDNA library provided by Lawrence Brass (University of Pennsylvania, Philadelphia, PA). The ␤ 3 cDNA was mutated to eliminate the EcoRI site at nucleotide 2270 (␤ 3 ⌬E). The cDNAs were subcloned into the EcoRI site in Bluescript II KS. The constructs were then subcloned into the Hin-dIII and XbaI sites in the multiple cloning region of pT7T3. A cytoplasmic domain cassette was created in ␤ 3 ⌬E using the molecular cloning strategy diagrammed in Fig. 1 with the oligonucleotides listed in Table  I. All restriction sites, which were created or modified, resulted in silent mutations. A unique NdeI restriction site (C to A at nucleotide 2247) flanking the 5Ј end of the cytoplasmic domain was introduced by PCR mutagenesis during an initial round of amplification using oligonucleotides 1 and 2 ( Fig. 1, Step A). An existing BsaBI restriction site at base 2418, 24 bases downstream from the stop codon, was altered by a C to G base change at position 2417, which was introduced during a second round of amplification ( Fig. 1, Step B). In this reaction, the initial PCR product from Step A was used as a megaprimer with oligonucleotide 3. The modification of the BsaBI site removed a dam methylation site, thereby making the construct susceptible to digestion with BsaBI when grown in standard host strains. Full-length ␤ 3 in pT7T3 was digested with ApaI and XbaI. The digestion fragment containing the vector and the 5Ј region of ␤ 3 up to the first ApaI site at nucleotide 110 was isolated and ligated to the similarly digested PCR product ( Fig. 1, Step C). Subsequently, the fragment from ApaI 110 to ApaI 2176 was cloned into the ApaI site in the correct orientation ( Fig. 1, Step D). This new construct lacked most of the 3Ј-untranslated region of ␤ 3 cDNA. The region of the construct from base 2176 to 2418, which was amplified by the polymerase chain reaction, was sequenced to confirm that there were no mutations. Three additional unique restriction sites, BssHII at base 2306 (A to G at 2304), AatII at base 2357 (C to G at 2355), and BglII, which is located at base 2405 in the 3Ј-untranslated region, were generated by PCR of the construct in Step D with oligonucleotides 4 and 5 ( Fig. 1, Step E). The PCR product was digested with NdeI and XbaI and subcloned into the similarly digested ␤ 3 /pT7T3 construct (Fig. 1, Step F). The construct from Step F was digested with BssHII and AatII, isolated from an agarose gel, and ligated to annealed oligonucleotide pair 6 and 7. This completed the construction of the ␤ 3 cDNA with four unique, silent restriction sites in the cytoplasmic domain.
Cytoplasmic domain mutations were generated by annealing appropriate oligonucleotide pairs at 85°C for 5 min and allowing to cool to 22°C over 1 h. Double-stranded fragments containing the base changes required for the generation of the desired amino acid substitutions were ligated to the appropriately digested construct (Fig. 1, Steps H, I, and J). These completed constructs were sequenced to verify that there were no synthesis errors and that the desired base changes were present. The constructs were digested with AflII and XbaI, and the fragment was isolated and subcloned into similarly digested ␤ 3 /pcDNAI/Amp. DNA for transfection was purified on a Qiagen-tip 100 (Qiagen, Chatsworth, CA).
Preparation of Cell Lines-CHO-K1 cells were obtained from the University of North Carolina Tissue Culture Facility and cultured in Dulbecco's modified Eagle's medium (DMEM-H) supplemented with 10% fetal bovine serum (HyClone, Logan, UT), 100 units/ml penicillin, 100 g/ml streptomycin, and non-essential amino acids (Life Technologies, Inc.). Adherent CHO-K1 cells were detached in 1 ϫ trypsin-EDTA (Boehringer Mannheim), washed, and resuspended in PBS containing 1 g of linearized ␣ IIb cDNA in pRc/CMV-Neo and 10 g linearized ␤ 3 cDNA constructs in pcDNAI/Amp. The cells were electroporated at 25 microfarads and 0.75 kV for 5 s in a gene pulser apparatus (Bio-Rad). After 48 h, cells were cultured in the presence of 0.5 mg/ml active Geneticin (Life Technologies, Inc.), and neomycin-resistant cells were selected for 2 weeks. Cells were harvested and immunostained with AP2 as described below, and single positive cells were sorted into individual wells in 96-well plates on an Epics model 753 cell sorter (Coulter Electronics, Hialeah, FL).
Surface Labeling and Immunoprecipitation-Adherent cells grown to confluence were rinsed with PBS, incubated on ice with 5 mM NHS-LC-biotin (Pierce) in PBS for 30 min, and rinsed with cold 5 mM glycine in Tris-buffered saline. Cells were washed three times with cold PBS and solubilized in 75 mM NaCl, 1 mM MgCl 2 , 1% Nonidet P-40, 25 mM Tris-Cl, pH 8, containing 1 g/ml aprotinin, 10 g/ml leupeptin, 10 g/ml pepstatin, 0.5 mg/ml Pefabloc SC (Boehringer Mannheim), and 10 g/ml benzamidine for 30 min at 4°C. Lysates were clarified by centrifugation at 12,000 ϫ g for 10 min, then precleared with normal mouse serum and ␥-Bind Plus Sepharose (Pharmacia) for 1 h at 4°C. Precleared lysates were centrifuged and incubated with 10 g of purified Tab IgG for 1 h followed by ␥-Bind Plus Sepharose at 4°C overnight. Beads were washed five times with buffer containing 0.5 M NaCl, 0.05 M Tris-Cl, pH 8.0, and 0.5% Nonidet P-40, and immunoprecipitates were released from ␥-Bind Plus Sepharose in Laemmli buffer (50). Samples were electrophoresed on a 7.5% acrylamide gel with biotinyl-ated molecular weight markers (Sigma) and separated proteins were transferred to nitrocellulose. The blots were developed with streptavidin-horseradish peroxidase (Life Technologies, Inc.) and detected with BM chemiluminescence (Boehringer Mannheim).
Preparation of Cells-Clonal cell lines expressing wild-type ␣ IIb ␤ 3 or variants were harvested as follows. Briefly, cells grown to confluence in complete medium containing Geneticin were rinsed with PBS and incubated in 3.5 mM EDTA in PBS for 5 min at room temperature. Cells were trypsinized with the addition of 0.10 volume 0.1% L-1-tosylamido-2-phenylethyl chloromethyl ketone-trypsin (Sigma) and allowed to incubate for 5 min, before adding 0.50 volume 20% fetal bovine serum containing 2 mg/ml soybean trypsin inhibitor (Sigma) in PBS. The cells were washed twice in PBS and resuspended in the buffers indicated below for the respective studies.
Flow Cytometry-Cells were harvested as described above and resuspended in PBS containing 2% BSA and 0.1 mM each MgCl 2 and CaCl 2 (PBC). Cells were incubated with AP2 ascites (1/16,000 dilution) for 30 min on ice, washed twice with cold PBS, and incubated as above with FITC-conjugated goat anti-mouse IgG. Cells were washed once and resuspended in 400 l of PBS for fluorescence analysis on a FACScan (Becton Dickinson, San Jose, CA). For PAC1 staining, cells were incubated with purified IgM antibody at 1 g/ml in Tyrode's solution containing 1 mg/ml BSA, 1 mM MgCl 2 , 0.1 mM CaCl 2 , and 20 mM HEPES, pH 7.4 (PBB), for 30 min at room temperature, then stained with FITC-conjugated rabbit anti-mouse IgM. Staining with PAC1 antibody was also performed in the presence of 1 mM GRGDSP. To detect activation of expressed ␣ IIb ␤ 3 , cells were preincubated with LIBS6 ascites for 30 min at 37°C, washed once with cold PBS, and immunostained with PAC1 in the presence or absence of 1 mM GRGDSP.
Adhesion-96-well tissue culture plates (Costar, Cambridge, MA) FIG. 1. Molecular cloning strategy for wild-type and mutant cDNA constructs. The polymerase chain reaction was used to introduce two unique restriction enzyme sites, NdeI and XbaI, flanking the cytoplasmic domain region of the ␤ 3 cDNA, and two unique restriction enzyme sites, BssHII and AatII, within the cytoplasmic domain cassette. Encircled numbers represent the oligonucleotides listed in Table I. A, PCR of wild-type ␤ 3 cDNA with oligonucleotide 2, which contains the NdeI site, and oligonucleotide 1. B, the PCR product from A was used as a megaprimer with oligonucleotide 3, which contains a modified BsaBI site and an XbaI site. C, this PCR product was digested with ApaI and XbaI and cloned into the ApaI 110 site of the vector. The ApaI 110 to ApaI 2176 fragment and the 3Ј-untranslated region were removed at this step. D, the ApaI 110 to ApaI 2176 fragment was cloned into the ApaI site. E, PCR of the new construct with oligonucleotide 4, which contains the sequence for the BssHII and AatII sites, and oligonucleotide 5. F, the product was digested with NdeI and XbaI and cloned into the construct. G, oligonucleotide pair 6 and 7 represents the wild-type sequence from 2304 to 2355. This pair was annealed and cloned into BssHII and AatII. H, oligonucleotide pair 8 and 9 and pair 10 and 11 were annealed, and the double-stranded fragments were ligated, purified on an agarose gel, and cloned into the NdeI and BssHII sites. I, the ligation product of oligonucleotide pair 12 and 13 to oligonucleotide pair 8 and 9 was subcloned into the NdeI and BssHII sites to generate mutant D723A/E726A. J, generation of mutant F727A/K729E/F730A with oligonucleotide pair 14 and 15.
were coated with either 2.5 g/ml fibronectin or 4 g/ml fibrinogen in PBS for 16 h at 4°C. Plates were then incubated with PBS containing 2% heat-inactivated BSA for 2 h at 37°C to block non-specific binding on the plates. Cells harvested as described above were resuspended at 1 ϫ 10 6 cells/ml in PBC and incubated in the presence or absence of 1 mM GRGDSP for 30 min at 22°C prior to plating 1 ϫ 10 5 cells/well. Cells were allowed to bind to the plates for 2 h at 37°C. Plates were washed three times with PBS to remove non-adherent cells. Adherent cells were stained with 0.5% crystal violet in PBS containing 20% methanol for 30 min at 22°C. Excess dye was removed by three washes with water, and cells were solubilized in 1% SDS for 16 h at 22°C. Cell binding was quantified by measuring the absorbance at 540 nm in a Bio Kinetics model EL 340 plate reader (Biotek Instruments). Specific binding was defined as the difference between the absorbance of cells binding to ligand-coated wells and the absorbance of cells binding to BSA-coated wells. All assays were performed in triplicate.
Immunofluorescence-Glass coverslips were coated with either 50 g/ml fibrinogen or 50 g/ml fibronectin in PBS at 4°C overnight. Matrix protein was removed and coverslips were rinsed twice with PBS. Cells were grown to confluence and incubated with medium containing 25 g/ml cycloheximide for 90 min at 37°C in 5% CO 2 . Cells were harvested as described above, resuspended in 12 ml of DMEM-H containing 25 g/ml cycloheximide, and maintained in suspension for 30 min at 37°C in 5% CO 2 with mixing at 15 min. Aliquots of the suspension were pipetted onto the coverslips and incubated for 2 h at 37°C. After gently swirling the plates, medium was removed and cells were fixed for 10 min at room temperature in 3.7% formaldehyde and washed twice with PBS. Following fixation, cells were permeabilized in 0.5% Triton X-100 in PBS for 5 min at room temperature and rinsed twice in PBS. Antibodies were used at dilutions ranging from 1:200 to 1:50 in PBS, 0.1% BSA to ensure optimal labeling. Coverslips were incubated with primary antibodies for 1 h at 37°C, extensively rinsed in PBS, and incubated with a mixture of either affinity-purified rhodamine-labeled donkey anti-mouse IgG or affinity-purified fluorescein-labeled donkey anti-rabbit IgG and fluorescein-conjugated phalloidin for 1 h at 37°C. Coverslips were then washed in PBS, rinsed in deionized water, and mounted in Fluorsave. Coverslips were viewed on a Zeiss Axiophot microscope, and photographs were taken on T-max 400 film (Eastman Kodak Co.) Adhesion-dependent pp125 FAK Phosphorylation-Plastic Petri dishes (10 cm) were precoated with 15 ml of either 50 g/ml fibrinogen in PBS or 2 g/ml poly-D-lysine in PBS for 18 h at 4°C. Before adherence, the plates were rinsed once with 5 ml of PBS. Cells were grown to confluence and then serum-starved in 0.5% fetal bovine serum in DMEM-H for 18 h. Cells from six 75-cm 2 flasks were harvested, washed twice in PBS, resuspended at 1.5 ϫ 10 6 cells/ml in 20 ml of DMEM-H, and maintained in suspension at 37°C for 30 min. Cells (7.5 ϫ 10 6 ) were pipetted onto plates precoated with either fibrinogen or poly-D-lysine or kept in suspension in 15-ml conical tubes and incubated for 25 min at 37°C. Cells maintained in suspension were centrifuged for 5 min and lysed in 0.5 ml of lysis buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 50 mM Tris, pH 7.5, containing 1 mM sodium orthovanadate, 1 mM p-nitrophenyl phosphate, 10 g/ml each leupeptin and pepstatin, 1 g/ml aprotinin, and 0.5 mM Pefabloc). For adherent cells, medium was removed from plates, 0.5 ml of lysis buffer was added, and the plates were kept on ice for 10 min. The plates were then scraped and the lysates transferred to a 1.5-ml tube. All lysates were centrifuged at 12,000 ϫ g for 10 min, and supernatants were transferred to a new tube. pp125 FAK was immunoprecipitated with 1.5 g of monoclonal antibody 2A7 for 1 h at 4°C on a rocker. The antigen-antibody complex was captured with 40 l of a 50% suspension of goat anti-mouse IgGconjugated agarose overnight on a rocker. Immunoprecipitates were washed four times with lysis buffer, and beads were boiled in 50 l of reducing buffer for 3 min. Samples were electrophoresed on a 7.5% acrylamide gel and blotted onto Immobilon-P. Blots were blocked for 1 h in TTBS (10 mM Tris, pH 7.4, 150 NaCl, 0.1% Tween 20) containing 3% BSA. Blots were incubated for 1 h with a 1/1000 dilution of polyclonal anti-FAK antibody (HuB3) in TTBS, washed extensively, and incubated for 1 h with anti-rabbit IgG (1:10,000) peroxidase conjugate. Phosphotyrosine was detected on a second blot with monoclonal antibody 4G10 (Upstate Biotechnology, Inc.) and 1:10,000 anti-mouse IgG peroxidase conjugate. Blots were washed, and bands were detected with BM chemiluminescence (Boehringer Mannheim).
Cell Aggregation-Harvested cells were resuspended at 2 ϫ 10 7 /ml in Tyrode's solution (137 mM NaCl, 2.8 mM KCl, 12 mM NaHCO 3 , 5.5 mM glucose, 0.4 mM NaH 2 PO 4 , and 10 mM HEPES, pH 7.4) containing 3.5 mg/ml BSA (two 15-cc conical tubes for each cell line). Cells were treated with 10 mM DTT or buffer for 20 min at room temperature. Cells were washed once with Tyrode's solution and resuspended at the same concentration in Tyrode's solution without albumin. Aggregation was performed in a 48-well plate by mixing 150 l of cells, 20 l of 2 mg/ml fibrinogen, or Tyrode's solution and 10 l of 15 mM CaCl 2 . Cells were TABLE I Oligonucleotides used in the molecular construction of the ␤ 3 cytoplasmic domain cassette and mutant cDNAs Oligonucleotides 1-5 are PCR primers that were used to introduce unique restriction sites into the cDNA. Base changes within the coding region yielded silent mutations. Oligonucleotides 6 -15 are pairs that were annealed and ligated into the cDNA in order to introduce wild-type sequence (oligonucleotides 6 -11) or to code for a desired amino acid substitution (oligonucleotides 12-15). The asterisks indicate base changes from the wild-type sequence. 1.

3Ј-GTGCTGGCTTTTCTTCGTCGACTTCGTCTCCTTCTCGCGC-5Ј A E A BssHII
incubated on a gyrotory shaker (New Brunswick Scientific, Edison, NJ) at 80 rpm for 20 min, and cell suspensions from each well were mounted on slides with coverslips and analyzed by brightfield photomicroscopy. Clot Retraction-Harvested cells were resuspended in serum-free DMEM-H at 1 ϫ 10 7 /ml, and 200 l was transferred to a glass aggregometry cuvette (Chrono-Log, Havertown, PA). Purified fibrinogen was diluted to 2 mg/ml in DMEM, and 100 l was added to the cells with stirring. Clot formation was initiated by the addition of 100 l thrombin (10 units/ml) in DMEM containing 28 mM CaCl 2 and 25 mM HEPES, pH 7.4. The stir bar was removed after stirring for 7 s, and tubes were incubated in a 37°C water bath overnight and photographed.

RESULTS
The amino acid sequences of the wild-type and mutant cytoplasmic domains are shown in Fig. 2. Two mutant forms of ␤ 3 were constructed and expressed with wild type ␣ IIb in CHO cells, one termed D723A/E726A contained alanine substitutions in the membrane proximal region of the ␤ 3 cytoplasmic domain, which has been hypothesized to interact with pp125 FAK (51), and the other, termed F727A/K729E/F730A, contained substitutions in the region implicated in the binding of ␣-actinin (24).
Complex Formation and Surface Expression-To determine whether the ␣ IIb ␤ 3 mutants formed heterodimeric complexes that were expressed on the cell surface, transfectants cultured in selective medium for 2 weeks were surface labeled with biotin, lysed, and immunoprecipitated with the ␣ IIb -specific antibody, Tab. As shown in Fig. 3, biotin-labeled ␣ IIb and ␤ 3 were recovered in the Tab immunoprecipitates from each cell line, indicating 1) that heterodimer complex formation had occurred, 2) that the complex was expressed on the cell surface, and 3) that the Tab epitope in the extracellular domain was intact. The heterodimer complex formation and surface expression of the ␣ IIb ␤ 3 mutant constructs were confirmed by flow cytometry using AP2, a complex-specific monoclonal antibody. Like the epitope for Tab described above, the epitope for AP2 was preserved in the extracellular domains of the two mutants. All three constructs were expressed at similar levels on the cell surface as assessed semiquantitatively by fluorescence staining (see left column, Fig. 4).
Although agonist-stimulated mechanisms for activating ␣ IIb ␤ 3 in CHO cells have not been identified, there are several methods that can be used to activate the complex externally to examine competence of the receptor to bind ligand. Partial reduction of disulfide bonds in ␣ IIb and/or ␤ 3 induces an activated conformation that is able to bind ligand (52). LIBS6 is a murine monoclonal antibody that binds to ␣ IIb ␤ 3 and increases the affinity of the receptor for ligands such as fibrinogen or reporter antibodies like PAC1 (48). In response to LIBS6 (Fig.  4, third column from left), cells expressing the wild-type ␣ IIb ␤ 3 showed increased binding of PAC1, a monoclonal antibody that recognizes the activated form of ␣ IIb ␤ 3 . Mutant F727A/K729E/ F730A demonstrated a similar response to LIBS6 with increased binding of PAC1. In contrast, the mutant D723A/ E726A complex bound increased PAC1, even in the absence of LIBS6 (Fig. 4, second column from left), suggesting that D723A/E726A was constitutively active. The addition of LIBS6 to cells expressing the mutant D723A/E726A complex induced additional binding of PAC1, implying that while the complex appears to be constitutively active, it may be only partially activated. D723A/E726A was not further activated by DTT (Fig. 4, right column).
To further determine if extracellular ligand binding domains of the mutant receptors were functionally intact, fibrinogen-dependent cell aggregation in response to DTT treatment was examined. As shown in Fig. 5, DTT-treated "activated" (A) wild-type cells formed large aggregates after 20 min, whereas the controls, which include DTT-treated cells in the absence of fibrinogen (not shown), and untreated cells in the presence (B) or absence (not shown) of fibrinogen, did not aggregate. Mutant D723A/E726A also formed large aggregates when activated, and, strikingly, also formed aggregates with no DTT pretreatment, a finding consistent with the demonstration of a constitutively active receptor by flow analysis with PAC1. DTTtreated mutant F727A/K729E/F730A also aggregated in response to DTT, which indicated that the extracellular region was functional.
Adhesion-Adherence of CHO cells expressing the various ␣ IIb ␤ 3 constructs to fibrinogen was measured using a solid phase binding assay (Fig. 6). Wild-type and mutant D723A/ E726A cells adhered to fibrinogen, and adhesion was inhibited by A2A9, a complex-specific antibody, indicating that adhesion was ␣ IIb ␤ 3 -specific. Whereas adhesion of wild-type cells was completely blocked by 1 mM GRGDSP, adhesion of mutant D723A/E726A cells was only partially blocked by 1 mM GRGDSP, but was completely blocked by 2 mM GRGDSP (not shown), suggesting that the receptor in these cells has a higher affinity for the ligand than wild-type. Mutant F727A/K729E/ F730A cells adhered to fibrinogen, and adhesion was blocked with 1 mM GRGDSP, indicating that the extracellular portion of the receptor functions normally.
Clot Retraction-One of the functions of the cytoplasmic tails of integrins is their interaction with cytoskeletal components linking extracellular events with the cytoskeleton. Among other things, this interaction mediates retraction of a fibrin clot (22). To look at the interaction of the mutant cytoplasmic domains with cytoskeletal components, the effect of the ␤ 3 cyto- plasmic domain amino acid substitutions on the integrin capacity to retract a fibrin clot was examined (Fig. 7). Harvested cells were mixed with purified human fibrinogen, and clot formation was initiated by the addition of thrombin. Wild type ␣ IIb ␤ 3 was able to efficiently retract a clot. In contrast, mutant F727A/K729E/F730A was unable to retract a clot. Mutant D723A/E726A retained the ability to retract a clot. Since transfection of CHO cells with ␤ 3 cDNA also results in complex formation between the ␤ 3 subunit and the endogenous ␣ v subunit to form ␣ v ␤ 3 heterodimers, and since ␣ v ␤ 3 can retract a fibrin clot in nucleated cells (53,54), studies were performed to assess the relative contributions of ␣ IIb ␤ 3 and ␣ v ␤ 3 in CHO cell-mediated retraction of the fibrin clot. Incubating cells with saturating amounts of AP2 inhibited clot retraction, whereas LM609 had no effect (not shown), suggesting that any vitronectin receptor expressed on the cell surface was not contributing significantly to clot retraction.
Formation of Focal Adhesions-During adhesion to extracellular matrix proteins, integrins become localized to focal adhesions at the ends of actin stress fibers. The interactions of the cytoplasmic domains of integrins with the the cytoskeleton in focal adhesions involve the association of a number of proteins. We examined the ability of the wild-type or mutant ␣ IIb ␤ 3 heterodimers to form focal adhesions and to organize the actin cytoskeleton. Each of the CHO cell lines were treated with cycloheximide for 90 min to prevent matrix formation and plated onto fibrinogen-coated glass coverslips in serum-free media. Cells were fixed after 2 h, and the organization of actin and localization of ␣ IIb ␤ 3 was examined by immunofluorescence microscopy using fluorescein-conjugated phalloidin (Fig.  8, top row) and the monoclonal antibody AP2 (Fig. 8, middle  row). Wild-type ␣ IIb ␤ 3 was found to localize to discrete structures at the termini of actin stress fibers (Fig. 8d), typical of focal adhesions, and ␣-actinin co-localized with stress fibers (Fig. 8g). The mutant D723A/E726A was also found to form focal adhesions (Fig. 8f) and promote the organization of stress fibers (Fig. 8c), which were stained with ␣-actinin (Fig. 8i). The mutant F727A/K729E/F730A did not spread as well as either wild-type or D723A/E726A and had fewer focal adhesions and stress fibers, as determined by AP2 (Fig. 8e) and phalloidin staining (Fig. 8b), respectively. The few focal adhesions present stained less strongly with AP2, were confined to the corners of the cell, and stress fibers tended to extend between these focal adhesions with few stress fibers terminating within the center of the cell. There was little staining for ␣-actinin (Fig. 8h). ␣ IIb ␤ 3 distribution appeared more diffuse throughout the cell. All three cell lines were able to spread equally on fibronectincoated coverslips and form focal adhesions and stress fibers as judged by immunofluorescence localization of vinculin and actin (data not shown).
pp125 FAK Analysis-pp125 FAK was normally distributed in focal adhesions in adherent cells expressing either wild-type (Fig. 9g) or mutant D723A/E726A (Fig. 9i). Mutant F727A/ K729E/F730A (Fig. 9h) showed a more diffuse pattern of staining for pp125 FAK . The pattern of tyrosine phosphorylation (Fig.  9, a-c) and paxillin (Fig. 9, d-f) staining was similar to that of pp125 FAK . When plated onto fibronectin-coated coverslips, all three cell lines formed focal adhesions that stained positively for pp125 FAK , paxillin, and phosphotyrosine, demonstrating that there was no defect in the capacity of these proteins to associate with focal adhesions (not shown).
To determine if the pp125 FAK in focal adhesions was phosphorylated, cell lysates were immunoprecipitated with anti-pp125 FAK . Samples were divided in half, electrophoresed, and transferred to nitrocellulose. Blots were incubated with either anti-pp125 FAK or anti-phosphotyrosine (Fig. 10). pp125 FAK immunoprecipitated from wild-type cells maintained in suspension or incubated in poly-D-lysine-coated flasks was minimally phosphorylated. As expected, integrin engagement on a fibrinogen matrix resulted in increased tyrosine phosphorylation of pp125 FAK in wild-type cells. Cells expressing mutant D723A/ E726A also showed no evidence for constitutive phosphorylation in cells maintained in suspension but displayed phosphorylation in cells plated on fibrinogen. The addition of fibrinogen to mutant D723A/E726A cells in suspension without an activating agent resulted in phosphorylation of pp125 FAK , and this was blocked by 1 mM GRGDSP. Interestingly, adhesion of mutant F727A/K729E/F730A cells to fibrinogen also resulted in phosphorylation of pp125 FAK , despite the failure of these cells to form significant focal adhesions or stress fibers. Semiquantitation of pp125 FAK phosphorylation, expressed as the ratio of phosphorylated pp125 FAK to pp125 FAK antigen, indicated that pp125 FAK phosphorylation in mutant F727A/K729E/F730A was approximately 85% that in wild-type cells (Table II). DISCUSSION These results provide evidence that activation of pp125 FAK can be dissociated from two important events in integrin signaling. First, activation of pp125 FAK can occur despite inhibition of integrin association with and assembly of focal adhesions and stress fibers. Second, pp125 FAK was not activated in cells in which the ␣ IIb ␤ 3 receptor was constitutively active but not occupied by ligand. These observations suggest that events other than pp125 FAK phosphorylation are needed for assembly of focal adhesions and that events other than integrin activation are needed for phosphorylation of pp125 FAK .
The finding that phosphorylation of pp125 FAK occurred only after ligand binding and was not observed in cells expressing the activated form of the receptor without ligand binding fits well with observations by Shattil and co-workers (40) that  F730A (b, e, h), or mutant D723A/E726A (c, f, i) were preincubated with media containing 25 g/ml cycloheximide, harvested, washed, and allowed to adhere to the coverslips for 2 h. Cells were fixed in 3.7% formaldehyde and permeabilized in 0.5% Triton X-100. Coverslips were immunostained for 1 h at 37°C, rinsed, and stained with secondary antibody for 1 h at 37°C. Coverslips were mounted in Fluorsave. a-c, phalloidin; d-f, anti-␣ IIb ␤ 3 ; g-i, anti-␣-actinin. phosphorylation of pp125 FAK during "outside-in" signaling did not occur with simple activation of ␣ IIb ␤ 3 but required a costimulatory signal. Based on the data of Shattil and others, it has been suggested that ␣ IIb ␤ 3 exists in several states of activation: a resting state, which is unable to bind fibrinogen or other ligands, a state of partial activation in which the receptor has undergone a conformational change which makes it able to bind ligand, and a state of activation in which ligand is bound and additional ligand-induced conformational changes have occurred. The results presented here are consistent with a multistate receptor and imply that the mutant receptor is only partially activated. Full activation of the receptor and generation of inwardly directed signals to pp125 FAK occur after binding of ligand.
This behavior of the partially activated mutant form of ␣ IIb ␤ 3 appears to be similar to that reported by Hughes et al. (32) using a ␤ 3 mutation in the same region. They suggested that the membrane-proximal regions of both ␣ and ␤ subunits act as a molecular hinge. In their model, the charged residue, Asp-723, in ␤ 3 cytoplasmic domain interacted with Arg-995 of the ␣ IIb cytoplasmic domain to maintain the receptor in an inactivated state. Mutation of either charged residue resulted in activation of the receptor, perhaps by breaking the molecular hinge. However, in contrast to our findings with mutant D723A/E726A in which phosphorylation of pp125 FAK occurred only after ligand binding, Hughes et al. (32) found that constitutively active ␣ IIb (F992A)␤ 3 or ␣ IIb ␤ 3 (D723A) mediated both ligand-dependent and ligand-independent phosphorylation of pp125 FAK . Whether these represent different effects of different mutations remains unclear.
One can speculate on the mechanism of activation of ␣ IIb ␤ 3 in the mutant D723A/E726A cell line. Perhaps the simplest explanation is that the mutations in the cytoplasmic tail induce local conformational changes that are transmitted to the extracellular domains, increasing affinity for ligand. In this case, the change in the extracellular domain should be a direct consequence of the cytoplasmic domain structural change and inde- FIG. 9. Tyrosine kinase substrates in focal adhesions. Experimental procedure was the same as in Fig. 8. a, d, 1, 6, 11), poly-D-lysine (lanes 2, 7, 12), maintained in suspension (lanes 3, 8,13), maintained in suspension with the addition of fibrinogen (lanes 4,9,14), and maintained in suspension with the addition of fibrinogen and a molar excess of GRGDSP (lanes 5, 10,15). The cells were lysed and pp125 FAK immunoprecipitated. Upper panel, pp125 FAK antigen. Lower panel, anti-Tyr(P) (anti-pTyr). pendent of cellular energy requirements. However, we found that PAC1 did not bind to these cells in the presence of inhibitors of oxidative phosphorylation, 2 raising the likelihood that some intracellular intermediate may be involved in receptor activation. Another possibility is that the mutation in the cytoplasmic tail affects the interaction of the ␤ 3 tail with cytosolic signaling molecules. In addition to ␣-actinin (24) and pp125 FAK (51), several proteins have been identified that interact in vitro with either the ␤ 1 or the ␤ 3 cytoplasmic domains, including paxillin (51), talin (25)(26)(27), integrin-associated protein (55), ␤ 3 -endonexin (56), and an integrin-linked protein kinase (57).
In addition, a recent study of activated ␣ IIb ␤ 3 receptor provides evidence for association with pp60 c-src and pp54/58 c-fyn (58). As a result of an altered interaction with such molecules, changes may occur in the extracellular domain to increase affinity for ligand. Since the mutation is in a putative binding site for pp125 FAK , it is tempting to speculate that activation may be caused by an altered interaction with pp125 FAK . Finally, because CHO cells do not possess the same mechanisms for activating ␣ IIb ␤ 3 that are present in platelets, it is possible that the mutant ␤ 3 interacts with a CHO signaling protein that does not normally appear to be available in platelets. Based on early results from several laboratories suggesting that tyrosine phosphorylation of pp125 FAK occurs coincident with focal adhesion and stress fiber assembly and that inhibition of phosphorylation of pp125 FAK correlated with decreased focal adhesion and stress fiber assembly (38,41), it has been suggested that phosphorylation of pp125 FAK precedes and is necessary for focal adhesion and stress fiber formation. Since pp125 FAK was phosphorylated in adherent cells expressing mutant F727A/K729E/F730A, which had reduced formation of focal adhesions, our studies suggest that it is a change related to the integrin receptor itself and not subsequent cytoskeletal assembly into focal adhesions, which leads to phosphorylation of pp125 FAK . This apparent dissociation of p125 FAK phosphorylation from focal adhesion formation mirrors recent findings showing that pp125 FAK can be displaced from focal adhesions and the tyrosine phosphorylation in focal adhesions reduced to an undetectable level without affecting the assembly or stability of these structures (59). Interestingly, although our results indicate that fibrinogen binding was associated with phosphorylation of pp125 FAK , occupation by small peptide ligands was not sufficient for phosphorylation. This difference between fibrinogen and peptide ligands suggests that receptor clustering may be crucial for phosphorylation of pp125 FAK .
The disruption of adhesion-dependent formation of focal adhesions and stress fibers with mutation of membrane proximal sequences may have several explanations. First, the mutated sequences are part of a region in ␤ 3 identified in peptide studies as an ␣-actinin binding site (24). Mutations in this region might inhibit the interaction of ␣-actinin with ␤ 3 and therefore prevent the clustering of ␣ IIb ␤ 3 that generates focal adhesions. Second, the mutated sequences are also part of a region in ␤ 3 identified as a talin binding site (27). Work showing that talin can associate with actin directly (60) or indirectly through an interaction with vinculin (61-63) suggests another possible mechanism by inhibition of the interaction of ␤ 3 with talin. Further studies using this mutant to examine the interaction with ␣-actinin and talin may help clarify the relative roles of these two proteins in the association of ␣ IIb ␤ 3 with focal adhesions. A third possibility is that the mutated region interacts with other as yet unidentified cytoskeletal components.
Interestingly, mutant D723A/E726A cells adherent to immobilized fibrinogen formed focal adhesions that stained posi-tively by immunofluorescence for ␣ IIb ␤ 3 , phosphotyrosine, and pp125 FAK , indicating that pp125 FAK is recruited into focal adhesions despite mutation of a putative pp125 FAK binding site within the ␤ 3 cytoplasmic domain. On the surface, this appears inconsistent with the findings of Schaller et al. (51) who reported that pp125 FAK binds to a synthetic peptide corresponding to the membrane-proximal region of ␤ 1 . When the aspartic acid and glutamic acid were substituted with alanines, pp125 FAK no longer bound to the peptide. However, there are several potential explanations for our results. First, Hildebrand et al. (64) identified a focal adhesion-targeting sequence in the COOH-terminal region of pp125 FAK , but it is the NH 2terminal region of the molecule that interacts with the integrin. As a consequence, it should be possible to block pp125 FAK binding to integrin and still observe incorporation of pp125 FAK into focal adhesions. Consistent with this, introduction of the carboxyl-terminal focal adhesion-targeting sequence of pp125 FAK into cells displaces pp125 FAK from focal adhesions (59,65). Second, the previously demonstrated interaction between purified pp125 FAK and the cytoplasmic domain peptide may not actually represent the physiological interaction. Third, the mutations in the ␤ 3 cytoplasmic tail, although within the region implicated in pp125 FAK binding may not prevent pp125 FAK binding. It is possible that a second region in the ␤ 3 cytoplasmic domain may contribute to pp125 FAK binding. The results of truncation studies indicating that the COOH-terminal amino acids are required for pp125 FAK association with integrin (66) are consistent with this explanation. Finally, recent work has shown an interaction of talin and paxillin with pp125 FAK (67)(68)(69)(70). Thus, pp125 FAK may be recruited to focal adhesions in association with talin or paxillin and not with the ␤ 3 cytoplasmic domain.
In summary, these results document the importance of membrane proximal cytoplasmic domain sequences in integrin function. Studies to further define the sites of interaction, the proteins with which these sites interact, and the molecular events that attend these interactions will significantly enhance our understanding of integrins. The constructs we have presented here may be useful in this regard.