SH2- and SH3-mediated Interactions between Focal Adhesion Kinase and Src*

Intramolecular SH2 and SH3 interactions mediate enzymatic repression of the Src kinases. One mechanism of activation is disruption of these interactions by the formation of higher affinity SH2 and SH3 interactions with specific ligands. We show that a consensus Src SH3-binding site residing upstream of the Src SH2-binding site in FAK can function as a ligand for the Src SH3 domain. Surface plasmon resonance experiments indicate that a FAK peptide containing both the Src SH2- and SH3-binding sites exhibits increased affinity for Src. Furthermore, the presence of both sites in vitro more potently activates c-Src. A FAK mutant (FAKPro-2) with substitutions destroying the SH3-binding site shows reduced binding to Src in vivo. This mutation also reduces Src-dependent tyrosine phosphorylation on the mutant itself and downstream substrates, such as paxillin. These observations suggest that an SH3-mediated interaction between Src-like kinases and FAK may be important for complex formation and downstream signaling in vivo.

The x-ray crystal structures of Src and Hck (a Src family member) in their inactive form have revealed intramolecular interactions that function to regulate these protein-tyrosine kinases (PTKs) 1 (1,2). As expected, the tyrosine phosphorylated, negative regulatory element binds to the SH2 domain (1,2). In addition, the SH3 domain binds to a polypeptide linking the SH2 and catalytic domains which assumes a polyproline type II helix that is structurally similar to SH3-binding sites (3)(4)(5). In this conformation, ␣-helix C in the small lobe of the catalytic domain is displaced altering the conformation of the ATP-binding site. It is noteworthy that the sequences that bind the SH2 and SH3 domains do not conform to high affinity binding sites (3, 6 -8).
Dephosphorylation of the negative regulatory element is one mechanism by which these PTKs can be activated. Consequently, this element fails to bind the SH2 domain and the inactive conformation cannot be maintained. A second mechanism by which the Src-like kinases could be activated is by the disruption of the weaker intramolecular SH2-SH3 interactions by the formation of stronger intermolecular SH2-SH3 domain interactions. It has been shown that disruption of the SH2-or SH3-mediated intramolecular interactions in vitro enhances the activity of the enzyme (9 -12). It is likely that similar mechanisms operate in vivo and that complex formation between Src and its binding partners, like FAK, results in its activation. FAK is a 125-kDa PTK that localizes to focal adhesions and functions in integrin signaling (13,14). Integrin-dependent cell adhesion or cross-linking of cell surface integrins induces the tyrosine phosphorylation of FAK and stimulates its activity (14 -18). The major site of FAK autophosphorylation is Tyr-397, whereas other sites of FAK phosphorylation, e.g. Tyr-576, -577, and -925, are phosphorylated by Src family PTKs (19 -21). The sequence flanking Tyr-397 conforms to a high affinity binding site for the Src SH2 domain and serves as a binding site for Src-like PTKs (6,19,22,23). Upstream of Tyr-397 is the sequence RALPSIPKL, which resembles the consensus Src SH3-domain binding motif, RXLPPLPR (3,7,8). The close proximity of this putative SH3-binding site (residues 368 -376) to the Src SH2-binding site in FAK (residues 397-400) suggests that both SH3 and SH2 domain interactions might mediate the association of Src with FAK.
We present evidence that residues 368 -378 of FAK can function as an SH3-binding site and that a FAK peptide containing both the SH2-and SH3-binding sites binds to Src with a higher affinity than a peptide containing the SH2-binding site alone. Furthermore, a peptide containing both binding sites activates c-Src in vitro more potently than peptides containing either the SH2-or SH3-binding site alone. Mutation of this SH3-binding site in FAK impairs association with Src (and Fyn) in vivo. This mutation also reduces the capacity of FAK to serve as a substrate for tyrosine phosphorylation in vivo and to induce tyrosine phosphorylation of paxillin when coexpressed with Src (or Fyn). These results support the hypothesis that both SH3 and SH2 interactions mediate Src/FAK binding and that interactions at both sites are critical for downstream signaling.

MATERIALS AND METHODS
Cells and Viruses-Chicken embryo (CE) cells were harvested from 9-day-old embryos and grown in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum and 1% chicken serum as described (24). FAK variants, c-Src, and Fyn were expressed in CE cells using RCAS A and RCAS B (25), which are replication competent, avian retroviral expression vectors. Cells were transfected with vector DNA as described (24) and lysed 7-9 days post-transfection. For coexpression studies, cells expressing RCAS A FAK, FAK Pro-2 , or FAK 397F were mixed with cells expressing RCAS B Src or Fyn 7 days post-transfection. Cells were lysed a week later. For some experiments cells were treated with 50 M vanadate for 16 h prior to lysis (26).
Site-directed Mutagenesis and Cloning-Site-directed mutagenesis was performed using the Altered Sites Kit (Promega, Madison, WI) to introduce alanine substitutions at Pro-371 and Pro-374 of FAK. Mutants were identified by nucleotide sequencing using Sequenase (Amersham). A 357-base pair BsmI/MscI fragment containing the mutations (nucleotide 821 to 1178 of the FAK cDNA) was then cloned into full-length FAK in pBluescript (Stratagene, La Jolla, CA). The fulllength mutant FAK cDNA was then subcloned into RCAS.
Protein Analysis-Cells were lysed in modified RIPA buffer as described and protein concentrations determined using the bicinchoninic acid (BCA) protein assay kit (Pierce) (27). For immunoprecipitations, 300 g of lysate were incubated on ice for 1 h with primary antibody. Polyclonal antisera BC3 (13) and 428 (gift from Dr. Andre Veillette, McGill University) (28) were used to recognize FAK and Fyn, respectively, and monoclonal antibody EC10 (gift from Dr. Sarah Parsons, University of Virginia) (29) was used to immunoprecipitate Src. A commercially available monoclonal antibody was used to immunoprecipitate paxillin (Transduction Labs, Lexington, KY). Immune complexes were collected using Protein A-Sepharose (Sigma) (for polyclonal antiserum) or goat anti-mouse IgG conjugated to agarose beads (Sigma) (for monoclonal antibodies) by rotating at 4°C for 1-2 h. Src immune complexes were collected using Protein A-Sepharose beads previously coated with rabbit anti-mouse antibody (Jackson Immunoresearch, West Grove, PA). Immune complexes were washed twice with modified RIPA and twice with Tris-buffered saline (TBS) (10 mM Tris pH 7.5, and 150 mM NaCl). The immune complexes were then boiled in Laemmli sample buffer (30) and analyzed by SDS-PAGE. Western blotting was performed using FAK (BC3), Src (EC10), and commercially available Fyn, paxillin, and phosphotyrosine antibodies (Transduction Laboratories, Lexington, KY) as described (27,31). Protein bands were visualized using the ECL detection system (Amersham) following application of horseradish peroxidase-conjugated secondary antibodies. For in vitro PTK assays, immune complexes were incubated in kinase reaction buffer (20 mM PIPES, pH 7.2, 3 mM MnCl 2 ) containing 10 Ci of [␥-32 P]ATP for various times at room temperature. Reactions were terminated by addition of Laemmli sample buffer, the samples boiled and analyzed by SDS-PAGE.
In Vitro Binding Assays-GST fusion proteins containing the SH2 domains from Src, Grb2, and phospholipase C␥ have been described (26). Protein expression was induced in Escherichia coli with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside and the GST fusion proteins isolated on glutathione-agarose beads following sonication (32). 200 g of CE cell lysates were precleared with 5 g of GST then incubated with 2, 5, or 10 g of the GST-SH2 fusion protein for 1 h at 4°C (26,33). The beads were washed twice with RIPA and twice with TBS, resuspended in Laemmli sample buffer, and analyzed by Western blotting as above.
Surface Plasmon Resonance (SPR)-The affinity of the interactions between FAK sequences (Fig. 1) and the SH2 and SH2/SH3 domains of Src were measured by SPR. Peptides mimicking the putative SH3binding site, called P3 (AAAARALPSIPKLAN), and the Src SH2-binding site of FAK, called P2 (SVSETDDYAEIID), were synthesized by the Protein Chemistry Lab of University of North Carolina, Chapel Hill/ National Institute of Environmental Health Sciences. Two versions of the latter peptide were made, one containing tyrosine and the other containing phosphotyrosine. The tyrosine-phosphorylated peptide AAAARALPSIPKLANNEKQGVRSHTVSVSETDDYAEIID was also synthesized and is referred to as P3/2. A control tyrosine-phosphorylated peptide (P3/2m) with alanines substituted for the proline residues that are critical for SH3 binding was also synthesized (AAAAR-ALASIAKLANNEKQGVRSHTVSVSETDDYAEIID). Peptides were immobilized on a sensor chip by covalent coupling to a carboxylated dextran matrix on its surface as described (34). Purified recombinant Src SH2 and SH3/SH2 domains were expressed and purified as described (35,36) and injected over the sensor chips in 10 mM HEPES pH 7.4, 150 mM NaCl, and 0.05% (v/v) of a 10% P-20 surfactant solution. As described (34) data points were collected when the responses reached steady state and the equilibrium dissociation constants calculated using the Hill equation, where R is the steady state response, C is the concentration of the SH2 or SH2/SH3 domain, R max is the calculated maximum response, n is the Hill coefficient, and K d is the equilibrium dissociation constant. Differences in K d were analyzed by two-tailed t test to assess the statistical significance.
Tyrosine Kinase Assays-Src TK was expressed in insect cells using Baculovirus and purified as described (37). Using recently established autophosphorylation conditions, the enzyme was stoichiometrically phosphorylated on tyrosine 530, the negative regulatory site of human c-Src, resulting in enzymatic repression (34). The enzyme was then used to phosphorylate the synthetic substrate RRLIEDAEYAARG. The production of ADP was coupled to the oxidation of NADH using phosphoenolpyruvate, pyruvate kinase, and lactate dehydrogenase and the decrease in absorbance at 340 nm was measured as described (34,38). Initial reaction rates were measured and kinetic parameters were determined by nonlinear regression analysis of the rates using the equation, where ϭ the measured velocity, V max is the maximum velocity, [S] is the substrate concentration and K m is the Michaelis constant for the substrate. The activation constant, K act , was determined by nonlinear regression analysis of the rates as a function of peptide concentration using the equation, where a ϭ velocity measured in the presence of peptide minus the velocity measured in its absence, V act ϭ the maximal activated velocity minus the velocity measured in the absence of peptide, and [L] is the peptide concentration (34).

RESULTS
To determine whether the putative SH3-binding site of FAK binds the Src SH3 domain, a synthetic peptide mimicking this sequence was tested for binding by SPR (Fig. 1). The peptide (P3) failed to bind the SH2 domain of Src but bound to a recombinant protein containing both the SH3 and SH2 domains of Src with a K d of 31 M (Table I). A phosphopeptide corresponding to the FAK autophosphorylation/SH2-binding site (P2) bound to the both the SH2 domain and the protein containing both SH3 and SH2 domains (Table I). Thus peptides mimicking both the Src SH3-binding site of FAK (residues 368 -378) and the autophosphorylation/Src SH2-binding site (at tyrosine 397) can interact with the SH3/SH2 domain of Src, although the SH2 interaction was stronger. To compare binding of a peptide with both SH2 and SH3-binding sites to a peptide with an SH2-binding site alone, peptides P3/2 and P2 were immobilized on a sensor chip and their affinities for the Src SH2 domain and SH3/SH2 domain measured. Peptide P3/2 bound to the SH3/SH2 domains with a higher affinity (K d of 0.016 M) than peptide P2 (K d of 0.046 M) (p Ͻ 0.0001) ( Table  I; experimental series 1). To determine if the SH3-binding site of peptide P3/2 was responsible for the increased binding affinity of this peptide, a control peptide, with substitutions in the SH3-binding site and an intact SH2-binding site (P3/2m), was synthesized ( Fig. 1). Peptides P3/2 and P3/2m were immobilized on a sensor chip and their affinities for the SH2 domain of Src and the Src SH3/SH2 recombinant protein measured. Both peptides bound to the SH2 domain with similar affinity (Table  I; experimental series 2; p ϭ 0.21). However, peptide P3/2 bound with a significantly higher affinity to the SH3/SH2 domain than peptide P3/2m (p Ͻ 0.0001). Thus interactions with the SH3-binding site are responsible for the increased affinity of the full-length peptide. The preliminary results of competition experiments also support the conclusion that the peptide with both SH3-and SH2-binding sites associates with the SH3/SH2 domain of Src with higher affinity than the SH2binding phosphopeptide alone (data not shown).
To assess whether the SH3-binding site of FAK could contribute to the enzymatic activation of Src in vitro, a synthetic peptide mimicking the SH3-binding site or phosphotyrosine containing peptides mimicking the SH2-binding site alone or containing both the SH2-and SH3-binding sites were used in a Src activation assay. Peptides containing either the SH3-binding site or the SH2-binding site activated Src in vitro, although the latter peptide was more effective (Table II). The presence of both binding sites resulted in more potent activation in vitro when compared with peptides containing either binding site alone (Table II). Therefore, sequences from FAK, a naturally occurring binding partner of Src, can activate Src by disruption of either the SH3 or SH2 intramolecular interaction, but most effectively by disruption of both interactions.
The role of the Src SH3-binding site of FAK in vivo was investigated by creating a mutant (FAK Pro-2 ) with alanines substituted for the prolines at positions 371 and 374, i.e. in the SH3-binding site. This mutant was expressed in CE cells using the RCAS retroviral vector. Immunoprecipitation and Western blot analysis showed that FAK Pro-2 and wild type FAK could be equivalently overexpressed but that the phosphotyrosine level on FAK Pro-2 was diminished when compared with wild type FAK ( Fig. 2A). One possible explanation is that disruption of the tertiary structure in the mutant could impair its ability to autophosphorylate on Tyr-397 and create the Src SH2-binding site. Two experiments were performed to address this possibility. First, the abilities of FAK and FAK Pro-2 to autophosphorylate in vitro were compared. Both enzymes exhibited autophosphorylating activity, although the rate of FAK Pro-2 autophosphorylation was slightly reduced (Fig. 2B). Densitometric analysis of four experiments suggests that FAK Pro-2 autophosphorylates at approximately 75% of the rate of FAK. Second, the integrity of the Src SH2-binding site in vivo was probed using a GST fusion protein containing the Src SH2 domain. Lysates from cells overexpressing wild type FAK or FAK Pro-2 were incubated with the GST-Src SH2 fusion protein, and the complexes were then analyzed by FAK Western blotting. The GST-Src SH2 fusion protein associated with equal amounts of wild type FAK and FAK Pro-2 (Fig. 2C). This interaction was specific since FAK failed to bind to GST (Fig. 2C, lane 9) or to a GST-phospholipase C␥ SH2 fusion protein (Fig.  2C, lane 10). In addition a FAK mutant (FAK 397F ) lacking the Src SH2-binding site failed to bind to the GST-Src SH2 fusion protein (Fig. 2C, lane 8). Thus despite slightly reduced autophosphorylating activity in vitro, these data suggest that the integrity of the Src SH2-binding site of FAK at Tyr-397 is intact in vivo.
The ability of Src and Fyn to bind FAK Pro-2 in vivo was assessed by coexpressing FAK variants with Src and Fyn in CE cells. Src immune complexes were isolated from RIPA lysates and probed by Western blotting for FAK. Wild type FAK was coimmunoprecipitated with Src and Fyn (Fig. 3A, lane 5, and data not shown). FAK 397F , which contains a mutated Src SH2binding site, failed to coimmunoprecipitate with Src or Fyn (Fig. 3A, lane 8, and data not shown). FAK Pro-2 could also be coimmunoprecipitated with both Src and Fyn, however, the amount of mutant protein coprecipitated was less than wild type FAK (Fig. 3A, lanes 5 and 6, and data not shown). Western blotting demonstrated that equivalent amounts of exogenous FAK proteins were expressed (Fig. 3A, lanes 1-4) and that Src was equally coexpressed with each (Fig. 3B). Despite equal levels of protein, FAK and FAK Pro-2 dramatically differ in their phosphotyrosine content when coexpressed with Src or Fyn (see Fig. 4). As above, a GST-Src SH2 domain was used as a probe to indirectly compare the phosphotyrosine status of tyrosine 397. The amount of FAK Pro-2 recovered was diminished relative to the amount of FAK recovered despite equal levels of expression (Fig. 3C). Identical results were obtained upon coexpression of FAK Pro-2 and Fyn (data not shown). These results suggest that the Src SH3-binding site of FAK is important for phosphorylation at tyrosine 397 upon coexpression with Src and for the efficient formation of the Src⅐FAK complex in vivo.
Cells that overexpress FAK and Src or Fyn exhibit an increase in tyrosine phosphorylation on the exogenously expressed FAK and on endogenous paxillin. 2 This observation was exploited to examine the functional ability of FAK Pro-2 to send downstream signals. Coexpression of FAK and c-Src or Fyn resulted in enhanced tyrosine phosphorylation of wild type FAK (Fig. 4A, lanes 1 and 4, and data not shown). Despite expression of approximately the same amount of FAK Pro-2 as  wild type FAK, tyrosine phosphorylation of FAK Pro-2 was not enhanced when coexpressed with Src (Fig. 4A, lanes 2 and 5). Paxillin immune complexes were isolated from lysates of cells coexpressing Src and either FAK Pro-2 or wild type FAK and probed by Western blot using an anti-phosphotyrosine antibody. Coexpression of wild type FAK with Src increased the phosphotyrosine content of paxillin (Fig. 4B). Also note a shift in the electrophoretic mobility of paxillin in these cell lysates (Fig. 4B, lane 4, bottom panel). Coexpression of FAK Pro-2 with Src did not induce paxillin tyrosine phosphorylation nor did it alter its electrophoretic mobility (Fig. 4B, lane 5). A paxillin blot shows that approximately equal amounts of paxillin are present (Fig. 4B, bottom panel). Identical results were obtained upon coexpression of Fyn with the FAK variants (data not shown). These findings indicate that the SH3-binding site of FAK is not only required for optimal binding to Src family  Similar experiments were performed on FAK and FAK Pro-2 overexpressing CE cells treated with vanadate. Under these conditions there is a FAK-dependent increase in phosphotyrosine levels on paxillin and on exogenously expressed FAK itself (26). In this assay both FAK and FAK Pro-2 induced tyrosine phosphorylation and became phosphorylated on tyrosine themselves (Fig. 5A). Presumably, FAK Pro-2 can signal in this assay because it is leaky. The inability of cellular phosphatases inhibited by vanadate to reverse phosphorylation events may lead to an accumulation of phosphotyrosine over time. Despite the ability of FAK Pro-2 to support vanadate-induced tyrosine phosphorylation of proteins, it does not completely function like wild type. Tyrosine phosphorylation of FAK at residue 925 produces a binding site for the SH2 domain of Grb2. In vanadate-treated FAK overexpressing cells, this site is apparently phosphorylated since a GST-Grb2 SH2 domain fusion protein can bind FAK (Fig. 5B, lane 7). FAK Pro-2 , however, was not bound by the GST-Grb2 SH2 fusion protein (Fig. 5B, lane 8), despite the fact that both proteins were equally expressed and each exhibited similar elevations in phosphotyrosine content in vanadate-treated cells (Fig. 5A). This observation implies that the Src SH3-binding site of FAK is required to direct phosphorylation of selected tyrosine residues in FAK.

DISCUSSION
The data presented herein supports the conclusion that an SH3-binding site in FAK functions as a Src-binding site and may play a role in regulating the activation of Src. Surface plasmon resonance experiments demonstrated that this site could function to bind the Src SH3 domain in vitro and a FAK peptide containing both SH3-and SH2-binding sites was a higher affinity ligand for the Src SH3/SH2 domain polypeptide than peptides with single binding sites. Furthermore, in an in vitro assay, the peptide with both binding sites more potently activated c-Src. The FAK Pro-2 mutant with substitutions in the Src SH3-binding site of FAK bound to Src less efficiently than wild type FAK in vivo. Moreover, FAK Pro-2 was impaired in the induction of tyrosine phosphorylation of paxillin and served less effectively itself as a substrate for tyrosine phosphorylation. Collectively, these data support the conclusion that assembly of the Src⅐FAK complex and subsequent phosphorylation of downstream substrates is dependent upon both SH3and SH2-mediated interactions.
The K d of the interaction between a recombinant Src SH3/ SH2 domain and a peptide mimicking the Src SH3-binding site within FAK was calculated to be approximately 30 M by SPR. Affinity measurements of SH3 domains of a number of proteins for their peptide ligands have been made using a number of techniques including SPR, fluorescence titration, and titration calorimetry. From the literature, there is a wide range of affinities of SH3 domains for peptide ligands. The Grb2 SH3 domain can bind to a proline-rich peptide from SOS with an affinity of 1.48 nm (39). Others have determined the affinity of this interaction to be approximately 3.5 M (40). The affinity of the Sem5 SH3 domain (a Grb2 homolog) for its binding site in SOS was determined to be approximately 30 M (41). The affinity of the crk SH3 domain for a variety of ligands ranges from 350 nM to 3 M (40, 42). The Abl SH3 domain binds to a peptide corresponding to its binding site in 3BP1 with an affinity of 34 M (43). Hck interacts via its SH3 domain with an HIV protein, Nef. The affinity of the Hck SH3 domain for the SH3-binding site of Nef was determined to be 91 M (44). The context of this binding site proved important for its interaction, since full-length Nef bound to the Hck SH3 with a dissociation constant ranging from 188 to 250 nM (44). The interaction of the FAK SH3 binding peptide with the Src SH3/SH2 domain falls within the range of affinities of SH3 interactions reported in the literature, although it exhibits a moderate affinity.
Similarly, we have used SPR to measure the affinity of a peptide mimicking the autophosphorylation site of FAK for the SH2 domain of Src. The K d was determined as 30 -50 nM. A number of measurements of affinities of Src family PTK SH2 domains for peptide ligands have been determined using SPR, fluorescence titration, and titration calorimetry. A naturally occurring high affinity ligand for Src family SH2 domains is a sequence within polyomavirus middle T antigen which matches the consensus sequence pYEEI. The affinity of this sequence ranges from 65 to 650 nM, depending upon the length and composition of amino acids that flank this binding site (45)(46)(47). Using the GST-SH2 domain of Lck, a very high binding affinity for pYEEI has been determined (3.7 nM) (48). Our own measurements of the affinity of a consensus binding site to the Src SH2 domain is approximately 50 nM. 3 In contrast to the favored pYEEI binding sequence, the Lck regulatory domain with the sequence pYQPQ has a relatively weaker affinity for SH2 do-3 B. Ellis, unpublished data.  1 and 4), or FAK Pro-2 (lanes 2 and 5). Some cells were treated with 50 M vanadate for 16 h prior to lysis (lanes 4 -6). B, 5 g of GST-Grb2 SH2 fusion protein was incubated with lysates of CE cells (lanes 11 and 12) or CE cells expressing FAK (lanes 7 and 9), or FAK Pro-2 (lanes 8 and 10) and bound FAK detected by Western blotting with BC3. Some cells were treated with 50 M vanadate for 16 h prior to lysis (lanes 7, 8, and 12). As negative controls lysates of vanadate-treated CE cells overexpressing FAK were incubated with 10 g of GST (lane 13) or 10 g of GST-phospholipase C␥ SH2 (lane 14) fusion protein. Twenty-five g of total protein from lysates of CE cells (lanes 3 and 6), CE cells expressing FAK (lanes 1 and 4), and CE cells expressing FAK Pro-2 (lanes 2 and 5) was blotted with BC3 to assure equal expression of FAK and FAK Pro-2 . Some cells were treated with vanadate prior to lysis (lanes 4 -6). mains, reportedly 1.4 to 4.2 M (45, 46, 49). Similarly, we find that the Src C-terminal peptide binds to the SH2 domain of Src with an affinity of 3.5 M. 3 Our measurements of the affinity of the FAK autophosphorylation site for the Src SH2 domain fall within the range of the high affinity binding sites reported in the literature. In fact the measured affinity is at the lower end of the range.
The presence of both the Src SH3-and SH2-binding sites of FAK within a single peptide potentiates binding to the SH3/ SH2 domain, but not to the SH2 domain alone. It is intriguing that the presence of the lower affinity SH3-binding site has such a profound effect upon binding of the peptide. The integrity of the SH3-binding site is required for this effect, since substitution of alanine for key proline residues abolishes the increase in affinity of the peptide. It is possible that binding of both sites simultaneously stabilizes the interaction of the peptide with the SH3/SH2 domain leading to an decreased K d . Alternatively, the mechanism of enhanced binding may be more complex. It has been reported that occupancy of either the SH3 or SH2 domain of Fyn enhanced the affinity of the other domain for its ligands (50). It has also been shown that the affinity of Grb2 for its SH2 ligands is enhanced when Grb2 is in complex with mSos1, an interaction mediated by the SH3 domains of Grb2 (51). The emerging theme from these studies is that the SH3 and SH2 domains of proteins like Grb2 and the Src family PTKs may be interdependent. Thus, presentation of SH3 and SH2 ligands together, as is the case for the naturally occurring Src ligand, FAK, may be an extremely potent mechanism for engaging the SH3/SH2 domains.
Coimmunoprecipitation experiments demonstrate that the SH3-binding site of FAK is also important for association with Src and Fyn in vivo. The reduced recovery of FAK Pro-2 in Src immune complexes may be directly due to reduced affinity due to the SH3 site mutation. However, the phosphotyrosine content of FAK Pro-2 is dramatically less than that of wild type FAK when each is coexpressed with Src. Furthermore, under these conditions, FAK Pro-2 exhibits reduced binding to the Src SH2 domain in vitro relative to FAK binding, suggesting that phosphorylation of tyrosine 397 is reduced. Thus reduced phosphotyrosine could also contribute to the reduction in association with Src in vivo. In part, the difference in phosphorylation between FAK and FAK Pro-2 could reflect a tighter association between Src and wild type FAK which may block dephosphorylation by cellular phosphatases. There is precedent for such an effect since expression of a fragment of Src containing the SH3 and SH2 domains results in its constitutive association with FAK and constitutive tyrosine phosphorylation of FAK (52).
The enzymatic activity of the Src family kinases is tightly regulated (53). Genetic and biochemical evidence has implicated phosphorylation of a C-terminal regulatory tyrosine and its subsequent intramolecular interaction with the SH2 domain as critical events in enzymatic repression. Genetic evidence has also demonstrated that the SH3 domain is crucial for enzymatic repression (54 -56). The structure of the Src family PTKs has revealed that both SH3 and SH2 mediated intramolecular interactions function to inhibit the enzyme (1, 2). Disruption of these interactions with high affinity ligands for either the Src SH2 domain or the SH3 domain activates the enzyme in vitro (9,10,12,57). Maximal enzymatic activation occurs when both the SH3 and SH2 domain intramolecular interactions are disrupted simultaneously (9,10). We have demonstrated that high affinity binding sites for both the SH3 and SH2 domain of Src reside within FAK. Using synthetic peptides mimicking the Src-binding sites within FAK, we have shown that both the SH3-binding site and SH2-binding site can activate Src in vitro, but do so more potently when both binding sites are presented together.
This model of Src activation has been proposed to operate in vivo in a number of scenarios. Stimulation of cells with plateletderived growth factor induces tyrosine phosphorylation of the platelet-derived growth factor receptor, creation of a Src SH2binding site, and association with Src (12,(57)(58)(59)(60). This interaction may cause the enzymatic activation of Src. However, this interaction only targets the intramolecular SH2 domain interaction of Src for disruption. Other binding partners of Src, e.g. p130 cas , AFAP-110, and Sin, contain binding sites for both the SH3 and SH2 domains of Src (10,61,62). In the case of Sin, the combination of SH3-and SH2-mediated interactions can activate Src both in vitro and in vivo. The SH3 domain interaction is envisioned to recruit and activate Src which then phosphorylates and binds to the SH2-binding site. The mechanism of regulation of the initial interaction is not clear. We have described functional binding sites for both the SH3 and SH2 domains of Src within FAK. Assembly of a FAK⅐Src complex will cause the simultaneous disruption of both intramolecular SH3 and SH2 interactions within Src and hence enzymatic activation. Indeed, coexpression of FAK and Src leads to their association in vivo and the tyrosine phosphorylation of cellular substrates, including FAK itself and paxillin. Both the Src SH3-and SH2-binding sites of FAK are required for maximal induction of tyrosine phosphorylation in this system. Since the Src SH2-binding site of FAK is the higher affinity binding site and is absolutely essential for association with Src in vivo, it is likely that complex assembly is regulated by phosphorylation of this site. Thus autophosphorylation of FAK at tyrosine 397 may trigger Src binding via its SH2 domain and partial activation. Subsequent engagement of the Src SH3 domain may further stabilize the FAK⅐Src complex and enhance the activity of Src driving the phosphorylation of its substrates. Dual engagement of the SH3 and SH2 domains of Src may be functionally important in the transmission of signals downstream of FAK that control the biological processes of cell spreading, cell migration, and anoikis (63-65).