Integrin signaling: roles for the cytoplasmic tails of α_IIbβ₃ in the tyrosine phosphorylation of pp125^FAK

Integrins are cell adhesion receptors, each composed of an α and a β type I transmembrane subunit. The subunits have a large extracellular domain, a single transmembrane domain, and a short cytoplasmic tail generally consisting of 20-70 amino acids (Hynes, 1992; Sastry and Horwitz, 1993). The platelet-specific integrin, α_IIbβ₃, mediates platelet aggregation and spreading on extracellular matrices through interactions with fibrinogen and von Willebrand factor (Weiss et al., 1989; Haimovich et al., 1993). These interactions are regulated such that high affinity binding of soluble ligands requires platelet activation (Shattil, 1995). Similarly, resting platelets do not adhere to immobilized von Willebrand factor and they adhere only loosely to fibrinogen, whereas activated platelets adhere tightly to both ligands (Haimovich et al., 1993). This mode of regulation, referred to as inside-out signaling, may involve interactions of the α_IIb and/or β₃ cytoplasmic tails with as yet uncharacterized intracellular mediators (Shattil, 1995; O’Toole et al., 1994).

α_IIbβ₃ and other integrins also participate in outside-in signaling whereby extracellular biochemical and mechanical cues are transduced into the cell. One of the earliest detectable integrin signaling events in platelets and other cells is protein tyrosine phosphorylation (Clark and Brugge, 1995). Of partic-ular interest in this regard, the protein tyrosine kinase, pp125^FAK (FAK), is localized to integrin-rich focal adhesions in adherent cells and becomes activated and phosphorylated on tyrosine residues following ligand-induced integrin clustering (Hanks et al., 1992; Schaller and Parsons, 1994). Integrin ligation also triggers cytoskeletal reorganization and can influence programs of gene expression during cell growth, differentiation and programmed death (Juliano and Haskill, 1993; Roskelley et al., 1994; Meredith et al., 1993). FAK may function as a key mediator in these events by integrating signals from integrins with those from receptor tyrosine kinases and other plasma membrane receptors. Indeed, recent studies have identified regions or specific residues within FAK that target it to focal adhesions (Hildebrand et al., 1993; Schaller et al., 1995) or are responsible for its interactions with integrins (Schaller and Parsons, 1994), other kinases (Src, PI 3-kinase) (Cobb et al., 1994; Chen and Guan, 1994), and adaptor proteins (Grb2, paxillin) (Turner and Miller, 1994; Schlaepfer et al., 1994).

Fibrinogen-dependent platelet adhesion and aggregation are accompanied by the phosphorylation of specific proteins on tyrosine residues (Ferrell and Martin, 1989; Golden et al., 1990). Some of these substrates, including a protein tyrosine kinase, pp72^syk, are phosphorylated within seconds of fibrino-gen binding to α_IIbβ₃ (Huang et al., 1993; Clark et al., 1994). Others, including pp60^Src and FAK, are phosphorylated later during platelet aggregation or spreading (Huang et al., 1993; Haimovich et al., 1993), events that are associated with a major reorganization of the actin cytoskeleton (Hartwig, 1992; Fox et al., 1993). While sequences within the cytoplasmic tails of α_IIb and β₃ have been shown to influence certain aspects of outside-in signaling, such as integrin recruitment to focal adhesions (Ylanne et al., 1993, 1995), the function of the tails in initiating the tyrosine phosphorylation cascade in platelets has not been defined.

One approach to study the process of outside-in signaling through α_IIbβ₃ would be to delete portions of the cytoplasmic tails by mutagenesis. While this cannot be done in platelets, it can be done in Chinese hamster ovary (CHO) cells that have been stably transfected with α_IIbβ₃ (O’Toole et al., 1994). Therefore, to study the roles of the α_IIb and β₃ cytoplasmic tails in FAK phosphorylation, we have expressed α_IIbβ₃ mutants containing partial or complete deletions of the cytoplasmic tails in CHO cells. Tyrosine phosphorylation of FAK was examined in response to cell adhesion to immobilized α_IIbβ₃ ligands. The results show that both the α_IIb and β₃ cytoplasmic tails are involved in the regulation of FAK. Furthermore, under certain experimental conditions, FAK phosphorylation can occur in the absence of cell spreading, suggesting that full reorganiza-tion of the cytoskeleton is not required for initial activation of FAK.

CHO cells were stably-transfected with various α_IIbβ₃ constructs and grown in the presence of fetal calf serum as described (O’Toole et al., 1990, 1991, 1994; Ylanne et al., 1993). Cell surface expression of α_IIbβ₃ was quantitated by flow cytometry, using a murine monoclonal antibody (Ab D57) specific for the α_IIbβ₃ complex (O’Toole et al., 1994).

Tissue culture plates (100 mm; Falcon) were coated overnight at 4°C with one of the following proteins in coating buffer (150 mM NaCl, 50 mM NaH₂PO₄, 50 mM Na₂HPO₄, pH 8): 5 mg/ml bovine serum albumin (BSA; fraction V, Sigma), 100 μg/ml purified human fib-rinogen (Ugarova et al., 1993), 10 μg/ml murine myeloma protein IgG₁ (MOPC-21, Sigma), 10 μg/ml Ab PM1.1 (an anti-α_IIb mono-clonal antibody; Ginsberg et al., 1986), 10 μg/ml Ab D57 (an anti-α_IIb monoclonal antibody), 10 μg/ml 7G7B6 (an anti-IL2 receptor monoclonal antibody; Chen et al., 1994), or 5 μg/ml poly-L-lysine (Sigma). The plates were then washed twice with phosphate-buffered saline (PBS), blocked for 2 hours at 37°C with 5 mg/ml BSA and then washed twice more with PBS.

CHO cells were trypsinized, washed once and resuspended to 1×10⁷/ml in an incubation buffer containing 137 mM NaCl, 2.7 mM MgCl₂, 5.6 mM glucose, 3.3 mM NaH₂PO₄ and 20 mM Hepes, pH 7.4. Then 1 ml of cells was added to each protein-coated plate for 90 minutes at 37°C. Non-adherent cells were diluted 1:1 with PBS, sed-imented at 100 g for 5 minutes, rinsed twice more with PBS, and lysed with RIPA buffer (158 mM NaCl, 1 mM Na₂EGTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS and 10 mM Tris-HCl, pH 7.2) containing 1 mM phenylmethylsulfonyl fluoride, 1 mM Na₃VO₄ (Fisher), and 100 KIU/ml aprotinin (Boehringer-Mannheim). Adherent cells were rinsed three times with PBS and either photographed by phase microscopy or scraped into RIPA buffer contain-ing inhibitors. Lysates were clarified by centrifugation in a microfuge at 14,000 rpm at 4°C for 30 minutes and protein content was determined with the BCA reagent (Pierce).

Equal amounts of protein from each lysate (250-500 μg, depending on the experiment) were incubated overnight at 4°C with either an affinity-purified rabbit polyclonal anti-phosphotyrosine antibody (UP28) (a gift from Joan Brugge, Ariad Pharmaceuticals, Inc., Cambridge, MA) (Huang et al., 1993), a rabbit antiserum specific for pp125^FAK (BC3, a gift from J. Thomas Parsons, University of Virginia, Charlottesville, VA) (Lipfert et al., 1992), or an appropriate control antibody. Immune complexes were precipitated at 4°C with Protein A-Sepharose that had been previously blocked with 10 mg/ml BSA for 10 minutes, followed by 3 washes in ice-cold RIPA buffer containing 1 mM Na₃VO₄. Immune complexes were extracted into boiling Laemmli sample buffer containing 10% β-mercaptoethanol and subjected to SDS-PAGE on 7.5% polyacrylamide gels and elec-trotransferred to nitrocellulose. Western blots were prepared and analyzed for phosphotyrosine-containing proteins as described previ-ously (Huang et al., 1993), using the anti-phosphotyrosine antibody PY72 (a gift from Bart Sefton, Salk Institute, La Jolla, CA).

Adhesion of platelets to immobilized fibrinogen requires surface expression of α_IIbβ₃. Over the course of 15-90 minutes, the adherent platelets spread and exhibit tyrosine phosphory-lation of numerous proteins, including pp125^FAK (Haimovich et al., 1993). As with platelets, a stable CHO cell line (A5) that expresses human α_IIb and β₃ adheres to and spreads on fib-rinogen, whereas untransfected cells do not (Ylanne et al., 1993). Therefore, to study the role of the cytoplasmic tails of

α_IIb or β₃ in the tyrosine phosphorylation of FAK, human α_IIb and β₃ containing specific truncations of the tails were stably expressed in CHO cells.

First, it was necessary to prove that tyrosine phosphoryla-tion of hamster FAK occurs in A5 cells in response to cell adhesion to an α_IIbβ₃ ligand. As expected, A5 cells attached to and spread on a fibrinogen-coated surface, but they did not adhere to a BSA-coated surface. Untransfected CHO cells did not bind to either surface. To study protein tyrosine phospho-rylation, A5 cell lysates were immunoprecipitated with an anti-phosphotyrosine polyclonal antibody to, in effect, concentrate phosphotyrosine-containing proteins. The immunoprecipitates were then probed on western blots with a monoclonal anti-phosphotyrosine antibody. A5 cells suspended over BSA for 90 minutes at 37°C exhibited tyrosine phosphorylation of three prominent bands at ∼40-50 kDa, 70-80 kDa and 120-130 kDa. In contrast, when the cells were allowed to attach to and spread on fibrinogen for 90 minutes, there was a marked decrease in intensity of the 40-50 kDa and 70-80 kDa bands and an increase in intensity of the 120-130 kDa band (Fig. 1A). Thus, adhesion of CHO cells via α_IIbβ₃ can exert differential effects on tyrosine kinase substrates, increasing the state of phospho-rylation of some and decreasing others.

To establish if one of the tyrosine-phosphorylated proteins in the 120-130 kDa region of the blot shown in Fig. 1 repre-sented FAK, parallel cell lysates were immunoprecipitated with a polyclonal anti-FAK antibody and western blots were probed with a monoclonal anti-phosphotyrosine antibody. No FAK phosphorylation was observed in A5 cells maintained in suspension for 90 minutes over a BSA-coated surface. In contrast, FAK became phosphorylated during cell adhesion and spreading on fibrinogen (Fig. 1B). Tyrosine phosphoryla-tion was detectable at the earliest time point studied (15 minutes), was maximal by 60 minutes and remained steady for up to 90 minutes. Re-probing the blots with a monoclonal anti-FAK antibody showed that equal amounts of this protein had been immunoprecipitated from each sample, indicating that cell adhesion had increased the extent of FAK phosphorylation (not shown).

FAK phosphorylation in A5 cells required the specific interaction of α_IIbβ₃ with fibrinogen because: (1) A5 cells adherent to a non-specific substrate, poly-L-lysine, exhibited no tyrosine phosphorylation of FAK; (2) cells expressing a point mutation in the extracellular portion of the β₃ subunit (D119A) that abolishes ligand binding (Loftus et al., 1990) failed to adhere to fibrinogen or exhibit FAK phosphorylation; and (3) FAK became phosphorylated in cells expressing either α_IIbβ₃ or α_IIbβ₃(D119A) when they were allowed to attach to Ab PM1.1, a non-function-blocking antibody against α_IIb (Fig. 1B). Thus, the adhesion of CHO cells to fibrinogen is dependent on α_IIbβ₃ and leads to tyrosine phosphorylation of FAK.

Clustering of β₁, β₃ or β₅ integrin cytoplasmic tail chimeras in fibroblasts is sufficient to cause tyrosine phosphorylation of FAK (Akiyama et al., 1994; Lukashev et al., 1994). To examine the requirement for the β₃ tail in the context of an intact integrin, wild-type α_IIb was stably expressed in CHO cells along with one of three truncated forms of β₃: β₃Δ728, β₃Δ724 or β₃Δ717. β₃Δ728 lacks the COOH-terminal 35 residues of the 47 residue β₃ tail, β₃Δ724 lacks the COOH-terminal 39 residues and β₃Δ717 lacks the COOH-terminal 46 residues (Table 1). All variants of α_IIbβ₃ were expressed on the surface of the CHO cells, although expression of α_IIbβ₃Δ728 was typically somewhat greater and α_IIbβ₃Δ717 typically less than that of α_IIbβ₃ (Fig. 2). Unlike A5 cells, α_IIbβ₃Δ728 and α_IIbβ₃Δ724 cells failed to spread (Fig. 3) or to form focal adhesions (Ylanne et al., 1993) during cell adhesion to immo-bilized fibrinogen or to Ab PM1.1 and D57, antibodies specific for α_IIb and α_IIbβ₃, respectively. Similarly, α_IIbβ₃Δ717 cells did not spread on these antibodies, but some cells did assume a partially-spread orientation on fibrinogen (Fig. 3).

Tyrosine phosphorylation of FAK was studied both in the presence and absence of 20 μM cycloheximide. This compound was used in an attempt to prevent the de novo synthesis of fibronectin or other adhesive ligands by the CHO cells and thus to minimize induction of FAK phosphorylation by ligation of hamster integrins. Cycloheximide was not generally toxic to CHO cells, since all of these cell lines exhibited tyrosine phosphorylation of FAK upon adhesion to purified fibronectin, presumably via hamster β₁ integrins (Fig. 4). Concentrations of cycloheximide in this range have been used for similar purposes by others (Schultz and Arman, 1995). When each of the three β₃ truncation mutants were plated over BSA in the presence of cycloheximide, they did not adhere and no FAK phosphorylation was observed. However, as shown for a single experiment in Fig. 4, FAK phosphoryla-tion was observed in A5 cells after adhesion to the α_IIbβ₃ antibody, Ab D57. FAK phosphorylation was also detectable in α_IIbβ₃Δ728 and in α_IIbβ₃Δ724 cells adherent to Ab D57, but less so in α_IIbβ₃Δ717 cells (Fig. 4). Densitometric analyses of western blots from 3 such experiments were carried out to quantitate the extent of FAK phosphorylation triggered by cell adhesion to the α_IIbβ₃ antibody. Compared to A5 cells, FAK phosphorylation in α_IIbβ₃Δ728 cells was reduced by an average of only 4.4%, while phosphorylation of FAK in α_IIbβ₃Δ724 and α_IIbβ₃Δ717 cells was reduced by 46.7% and 71.1%, respectively (Fig. 5). Parallel immunoblots probed with an anti-FAK antibody demonstrated that equal amounts of FAK had been immunoprecipitated from each of these cell lines (not shown). The residual FAK phosphorylation in the α_IIbβ₃Δ717 cells observed in some experiments could not be explained by expression of the β₃ cytoplasmic tail because α_IIbβ₃ immuno-precipitated from α_IIbβ₃Δ717 cell lysates failed to react on western blots with a polyclonal antibody specific for the β₃ tail. Two conclusions can be drawn from these experiments. First, FAK phosphorylation triggered through α_IIbβ₃ occurs in the absence of distal sequences of the β₃ cytoplasmic tail. Second, since the α_IIbβ₃Δ728 and α_IIbβ₃Δ724 cells failed to spread on fibrinogen yet exhibited FAK phosphorylation, initiation of FAK phosphorylation in these CHO cell lines is not dependent on cell spreading.

To examine the role of the α_IIb cytoplasmic tail in integrin signaling, wild-type β₃ was expressed in CHO cells along with one of two truncated forms of α_IIb. α_IIbΔ996 lacks the carboxy-terminal 13 residues of the 20 residue α_IIb cytoplasmic tail, but it preserves the highly conserved membrane-proximal KVGFFKR segment. α_IIbΔ991 lacks the carboxy-terminal 18 residues of the cytoplasmic tail (Table 1). Previous studies have shown that, like wild-type α_IIbβ₃, α_IIbΔ996β₃ exists in a low-affinity state in CHO cells with respect to soluble ligands, while α_IIbΔ991β₃ exists in a high affinity state (O’Toole et al., 1994). Both mutant integrins differ from α_IIbβ₃ in that they are recruited to focal adhesions in a ligand-independent manner (Ylanne et al., 1993). As shown in Fig. 6 for α_IIbΔ996β₃, these integrins were expressed on the cell surface to levels of about 50% of that observed for α_IIbβ₃ in A5 cells. Nonetheless, when the mutant cells were incubated for 90 minutes over a fibrino-gen matrix, they became adherent and fully spread (not shown) and they exhibited tyrosine phosphorylation of FAK (Fig. 7). Thus, the α_IIb cytoplasmic tail is not necessary for FAK phos-phorylation in response to cell adhesion via α_IIbβ₃. This is con-sistent with a previous study of α₅β₁ in fibronectin-adherent CHO cells that concluded that the α₅ cytoplasmic domain is not essential for integrin-mediated tyrosine phosphorylation (Bauer et al., 1993).

However, studies of these α_IIb mutant cells in suspension indicated that the α_IIb cytoplasmic tail can modulate the process of FAK tyrosine phosphorylation. When A5 cells were removed from tissue culture plastic and incubated in suspen-sion, FAK phosphorylation became undetectable within five minutes and remained so for up to 90 minutes. In contrast, when α_IIbΔ996β₃ cells were maintained in suspension, FAK remained phosphorylated on tyrosine residues for up to 90 minutes (Fig. 7). Identical results were obtained with the α_IIbΔ991 cell line (not shown). Although the extent of residual FAK phosphorylation varied from experiment to experiment, this basic observation held true for multiple clonal isolates of each cell line. The presence of the β₃ tail was necessary for persistent FAK phosphorylation in the suspended α_IIb mutant cells because it was not observed in an α_IIb mutant cell line that also lacks the β₃ tail (α_IIbΔ996β₃Δ717) (not shown).

Four separate observations indicated that the persistent phosphorylation of FAK in the suspended cells did not require occupancy of the α_IIb mutant integrins by adhesive ligands. First, no soluble ligand was added to the suspended cells. Second, the α_IIbΔ996β₃ mutant is in a low affinity state and would not be expected to bind a soluble ligand. Third, addition of 20 μM cycloheximide to minimize endogenous synthesis of a ligand had no effect on FAK phosphorylation in the α_IIb mutants. Fourth, persistent tyrosine phosphorylation of FAK was still observed in suspensions of a stable cell line that expressed a double mutation, α_IIbΔ996β₃(D119A) (Fig. 7). The β₃(D119A) mutation is in the extracellular domain of β₃ and prevents ligand binding (Loftus et al., 1990). Taken together, these experiments show that removal of at least two-thirds of the residues from the COOH terminus of the α_IIb cytoplasmic tail leads to anomalous tyrosine phosphorylation of FAK in suspended CHO cells bearing α_IIbβ₃. This suggests that the α_IIb tail may normally function to dampen integrin signaling when cells are not adherent to a surface.

In several cell types, ligand binding to and clustering of integrins is associated with tyrosine phosphorylation and acti-vation of FAK and localization of the enzyme to focal adhesions (Guan et al., 1991; Kornberg et al., 1991; Hanks et al., 1992; Lipfert et al., 1992; Hildebrand et al., 1993). Membrane-based focal adhesions or closely-related structures form in many types of adherent cells, including platelets (Nachmias and Golla, 1991), and they couple extracellular matrix ligands to actin stress fibres through integrins (Turner and Burridge, 1991). Since focal adhesions contain numerous enzymes (e.g. pp60^Src, PI 3-kinase, protein kinase C) as well as cytoskeletal proteins (e.g. talin, α-actinin, paxillin), they may function as a type of ‘signaling organelle’ to facilitate information flow from the plasma membrane to the nucleus (Turner and Burridge, 1991; Cobb et al., 1994; Chen and Guan, 1994; Woods and Couchman, 1992; Sastry and Horwitz, 1993). Integrins are presumed to interact with other focal adhesion components through their cytoplasmic tails. Consis-tent with a role for these tails in integrin signaling, clustering of integrin β₁, β₃ or β₅ cytoplasmic tail chimeras in fibroblasts is sufficient to effect tyrosine phosphorylation of FAK (Akiyama et al., 1994; Lukashev et al., 1994).

To study the signaling function of integrin tails in more detail, we have expressed α_IIbβ₃ in CHO cells. The effect of deletions of cytoplasmic tail sequences on tyrosine phospho-rylation of hamster FAK was studied by plating the cells on matrices selective or specific for α_IIbβ₃. This experimental system was validated by showing that transfectants expressing wild-type α_IIbβ₃ failed to adhere to a BSA-coated matrix or to exhibit tyrosine phosphorylation of FAK, and no phosphory-lation occurred during ‘non-specific’ cell attachment to poly-L-lysine. However, when cells were incubated either over a fib-rinogen matrix or anti-α_IIbβ₃ monoclonal antibodies, they attached and spread and tyrosine phosphorylation of FAK was observed. It is interesting to note that certain phosphotyrosine-containing proteins became dephosphorylated during CHO cell adhesion to fibrinogen, consistent with observations in platelets that adhesion via α_IIbβ₃ stimulates both protein tyrosine kinases and phosphatases (Frangione et al., 1993; Luber and Siess, 1994; Ezumi et al., 1995). Thus, although the current studies were focused on FAK, the CHO cell system may also prove useful in studying the functions of tyrosine phosphatases during integrin signaling.

Three major conclusions can be drawn from the present studies: (1) The distal β₃ tail is not necessary for integrin-triggered tyrosine phosphorylation of FAK. (2) Although the

α_IIb tail is not necessary for FAK phosphorylation, sequences within the membrane-distal portion of the α_IIb tail can modulate this process. (3) Under certain experimental con-ditions, tyrosine phosphorylation of FAK does not require spreading of CHO cells on the adhesive matrix. As discussed later, this finding has implications for the relationship of FAK function to cytoskeletal reorganization.

Tyrosine phosphorylation of FAK was reduced in CHO cells adherent through a form of α_IIbβ₃ that lacks the COOH-terminal 46 residues from the β₃ tail. This is consistent with previous studies using different experimental approaches indicating that a β tail is involved in integrin-mediated FAK phos-phorylation (Guan et al., 1991; Akiyama et al., 1994; Lukashev et al., 1994). It should be noted, however, that reduced surface expression of α_IIbβ₃(Δ717) compared to α_IIbβ₃ (Fig. 2) might have played some role in the reduced FAK phosphorylation (Figs 4 and 5). On the other hand, FAK phosphorylation was compared in cells adherent to an α_IIbβ₃-specific subtrate. Con-sequently, differences in α_IIbβ₃ surface expression may not have been as marked in these adherent cells. FAK phosphory-lation in adherent cells appeared to be initiated through the transfected and truncated integrin since an antibody specific for α_IIbβ₃ was used as the ligand, and this antibody did not support the adhesion of untransfected CHO cells. The residual FAK phosphorylation exhibited by adherent β₃ truncation mutants may have been due to a component of adhesion through endogenous hamster integrins, despite the presence of cyclo-heximide to inhibit matrix synthesis. These data suggest that sequences within the β₃ tail are required for maximal FAK phosphorylation. The findings in this study seem to differ from a previous report showing that deletion of only 4 residues from the COOH terminus of the β₁ tail abolished tyrosine phospho-rylation of a ∼120 kDa band in fibroblasts adherent to fibronectin (Guan et al., 1991). Although the reasons for these differences are not apparent, the two studies focused on different integrins and different cell types, and the previous study performed anti-phosphotyrosine immunoblots on total cell lysates rather than on FAK immunoprecipitates.

In theory, the β₃ tail could function as a direct binding site for FAK. Upon fibrinogen binding to α_IIbβ₃, integrin cluster-ing might then lead to FAK clustering, a process that could lead to FAK auto-phosphorylation. Alternatively, the β₃ tail might serve as a nucleation site for other protein tyrosine kinases that in turn could phosphorylate and activate FAK. Both of these possibilities are plausible. In vitro, certain integrin β tails have been shown to bind to α-actinin and talin (Otey et al., 1993; Horwitz et al., 1986). In addition, synthetic peptides derived from the membrane-proximal 13 residues of β₁ and β₃ can bind to the unique NH₂-terminal domain of FAK (Schaller et al., 1995). However, it is not yet proven that these interactions actually take place within cells. Moreover, integrin clustering by fibrinogen or antibodies is not sufficient to initiate the phosphorylation of FAK in platelets. Rather, co-stimulation of platelets with excitatory agonists as well as platelet aggregation (or spreading) are required (Shattil et al., 1994). After fibrinogen binding to platelet α_IIbβ₃, activation of pp72^Syk and pp60^Src precedes activation of FAK (Clark et al., 1994). Thus, in platelets at least, events in addition to integrin clustering are necessary for FAK phosphorylation. Of potential importance in this regard, activated pp60^c-Src localizes to focal adhesions through its SH2 and SH3 domains (Kaplan et al., 1994) where it may participate in the recruitment and phos-phorylation of FAK (Cobb et al., 1994).

The reduced FAK phosphorylation observed with β₃ cyto-plasmic tail truncation could be due to fewer numbers of FAK molecules phosphorylated per cell or to less phosphorylation of each FAK molecule. FAK contains a number of potential sites of tyrosine phosphorylation that might be differentially regulated during cell adhesion, leading to different routes of integrin signaling. For example, Tyr³⁹⁷ in the NH₂-terminal domain of FAK is the major site of autophosphorylation and serves as a docking site for the SH2 domain of pp60^Src (Schaller and Parsons, 1994; Calalb et al., 1995). On the other hand, Tyr⁹²⁵ in the COOH-terminal domain may function as a docking site for the SH2 domain of Grb2, thereby linking integrin signaling to Ras pathways (Schlaepfer et al., 1994).

FAK was rapidly dephosphorylated on tyrosine residues when CHO cells expressing wild-type α_IIbβ₃ were suspended in physiological buffer. In contrast, the α_IIbΔ996β₃ cell line containing a deletion of all but the membrane-proximal 7 residues of the 20 residue α_IIb cytoplasmic tail showed per-sistent FAK phosphorylation when the cells were maintained in suspension for up to 90 minutes. Similar results were obtained with α_IIbΔ991β₃ cells lacking the COOH-terminal 18 residues from the α_IIb tail. These results could not be explained by the binding of soluble ligands to these integrins because α_IIbΔ996β₃ is in a low affinity state in CHO cells (O’Toole et al., 1994). Moreover, persistent FAK phospho-rylation was observed in a double-mutant cell line,

α_IIbΔ996β₃(D119A), that is incapable of binding soluble

α_IIbβ₃ ligands. It is not known if the persistent phosphoryla-tion in the α_IIb mutants is due to continued activity of a tyrosine kinase and/or to reduced activity of a tyrosine phos-phatase. While further studies will be required to assess the functional significance of this anomalous signaling, these findings suggest that the distal portion of the α_IIb tail may normally function to dampen integrin signaling when cells are not adherent to a matrix.

Circular dichroism measurements of synthetic peptides and fluorescence quenching studies of neoproteins chemically syn-thesized to simulate the α_IIbβ₃ tails suggest that the tails can interact with each other (Haas et al., 1993; Muir et al., 1994). If this is true inside of cells, the α_IIb tail may function in non-anchored cells to occlude binding sites on the β₃ tail for proteins that are involved in integrin signaling. In the absence of the α_IIb tail, signaling might therefore proceed in an anomalous fashion, e.g. when cells are in suspension. Similarly, when normal cells become adherent to an integrin ligand, these physical relationships between the α and β tails might be modified so as to promote association of the β₃ tail with FAK and other focal adhesion proteins. Consistent with this hypothesis, both α_IIbΔ996β₃ and α_IIbΔ991β₃ become recruited to focal adhesions in a ligand-independent fashion during CHO cell adhesion to fibronectin via hamster β₁ integrins. This recruitment does not take place in the absence of the β₃ tail (Ylanne et al., 1993).

In platelets and other cells, FAK phosphorylation is usually associated with cell spreading or aggregation, responses that require actin polymerization and rearrangements of the cytoskeleton, including formation of F-actin stress fibres and assembly of focal adhesions. In some cell types in which careful time course studies have been performed, FAK phos-phorylation precedes cell spreading, implying that FAK may play a role in the spreading process (Guan et al., 1991). The finding that a tyrosine kinase inhibitor prevents FAK phos-phorylation and cell spreading in fibroblasts is consistent with this interpretation (Burridge et al., 1992). On the other hand, in mouse aortic smooth muscle cells adherent to fibronectin, FAK activation is not required for the assembly of F-actin stress fibres or focal adhesions (Wilson et al., 1995). It is also possible that cytoskeletal rearrangements may promote FAK phosphorylation and activation by recruiting this enzyme to the correct subcellular location. Indeed, cytochalasin B or D, compounds that disrupt actin filaments, also inhibit integrin-triggered FAK phosphorylation in platelets and fibroblasts (Lipfert et al., 1992; Sinnett-Smith et al., 1993; Zachary et al., 1993). Yet, in the present study, cells bearing β₃ truncation mutants underwent some degree of FAK phosphorylation despite the fact that they did not spread on an anti-α_IIb antibody. This indicates that complete cytoskeletal reorgani-zation is not required for the initiation of FAK phosphoryla-tion in these cells. These diverse observations indicate that the exact relationship between FAK activation and cytoskeletal reorganization during cell adhesion may depend on the partic-ular matrix, integrin and cell type studied.

These studies were supported by grants from the NIH (HL 40387, HL 48728) and by a fellowship to R.E.H. from the American Heart Association. This work was presented in part at the 1994 Annual Meeting of the American Heart Association, Dallas, TX and published in abstract form (Circulation 90:I-86, 1994).