Requirement of Phosphatidylinositol 3-Kinase in Focal Adhesion Kinase-promoted Cell Migration*

We have previously shown that overexpression of focal adhesion kinase (FAK) in Chinese hamster ovary (CHO) cells promoted their migration on fibronectin. This effect was dependent on the phosphorylation of FAK at Tyr-397. This residue was known to serve as a binding site for both Src and phosphatidylinositol 3-kinase (PI3K), implying that either one or both are required for FAK to promote cell migration. In this study, we have examined the role of PI3K in FAK-promoted cell migration. We have demonstrated that the PI3K inhibitors, wortmannin and LY294002, were able to inhibit FAK-promoted migration in a dose-dependent manner. Furthermore, a FAK mutant capable of binding Src but not PI3K was generated by a substitution of Asp residue 395 with Ala. When overexpressed in CHO cells, this differential binding mutant failed to promote cell migration although its association with Src was retained. Together, these results strongly suggest that PI3K binding is required for FAK to promote cell migration and that the binding of Src and p130Cas to FAK may not be sufficient for this event.


We have previously shown that overexpression of focal adhesion kinase (FAK) in Chinese hamster ovary (CHO) cells promoted their migration on fibronectin.
This effect was dependent on the phosphorylation of FAK at Tyr-397. This residue was known to serve as a binding site for both Src and phosphatidylinositol 3-kinase (PI3K), implying that either one or both are required for FAK to promote cell migration. In this study, we have examined the role of PI3K in FAK-promoted cell migration. We have demonstrated that the PI3K inhibitors, wortmannin and LY294002, were able to inhibit FAK-promoted migration in a dose-dependent manner. Furthermore, a FAK mutant capable of binding Src but not PI3K was generated by a substitution of Asp residue 395 with Ala. When overexpressed in CHO cells, this differential binding mutant failed to promote cell migration although its association with Src was retained. Together, these results strongly suggest that PI3K binding is required for FAK to promote cell migration and that the binding of Src and p130 Cas to FAK may not be sufficient for this event.
Focal adhesion kinase (FAK) 1 is a key component of integrinmediated signal transduction (1)(2)(3). This 125-kDa nonreceptor protein-tyrosine kinase (PTK) is localized to focal contacts in fibroblasts and represents the prototype of a distinct family of nonreceptor PTKs (4,5). It rapidly becomes phosphorylated following cell adhesion to extracellular matrix (ECM) proteins or integrin clustering by antibodies (6 -9). In addition to its role in integrin signaling, FAK has also been suggested to be a point of convergence of signaling by other extracellular stimuli (10). The ability of FAK to transmit signals to downstream targets is dependent on its ability to interact with several intracellular signaling molecules including Src family kinases (11,12), phosphatidylinositol 3-kinase (PI3K; Ref. 13), Grb2 (14), and p130 Cas (15). Tyr-397 has been identified as the major site of FAK autophosphorylation (16) and the binding site for the Src-homology (SH)2 domains of Src (11,12) and PI3K (17). The binding site for the SH2 domain of Grb2 has been mapped to Tyr-925 (18). The proline-rich region of FAK (residues 712-718) has been identified as the major binding site for the SH3 domain of p130 Cas (15,19).
Mounting evidence suggests that FAK plays an important role in regulating cell migration in response to cell adhesion to ECM proteins (20,21). We have shown previously that the overexpression of FAK in Chinese hamster ovary (CHO) cells promoted migration on fibronectin (21). We have also demonstrated that p130 Cas association with FAK was required for FAK-promoted cell migration (22). Furthermore, a substitution of Tyr-397 with Phe abolished FAK-promoted cell migration (21), implicating a role for Src and/or PI3K in this phenomenon.
PI3K is a heterodimer consisting of an 85-kDa regulatory subunit (p85), and a 110-kDa catalytic subunit (p110) (23,24). p85 contains an SH3 domain and two SH2 domains. This kinase phosphorylates the D-3 position of the inositide ring of phosphatidylinositol and its derivatives, phosphatidylinositol 4-phosphate and phosphatidylinositol 4,5-bisphosphate (25). These lipid products are believed to act as second messengers in a variety of signaling processes including cell survival (26,27) and migration (28 -30). We have shown that PI3K associates with FAK in response to cell adhesion (13) or plateletderived growth factor (PDGF) stimulation (31) in NIH 3T3 cells. The significance of this interaction in cellular processes is unknown.
In this report, we have found that a FAK mutant capable of binding to Src, but not to PI3K, failed to stimulate cell migration and that inhibition of PI3K decreases cell migration. Taken together, these results strongly implicate a role for PI3K in the facilitation of FAK-promoted cell migration and suggest that the binding of Src and p130 Cas to FAK may not be sufficient for this process.

EXPERIMENTAL PROCEDURES
Antibodies-The mouse mAb KT3 (21) and the mouse mAb 12CA5 (anti-HA) (17), which recognize an epitope (KPPTPPPEPET) of the SV40 large T antigen and an epitope (YPYDVPDYA) of the hemagglutinin (HA) protein of the influenza virus, respectively, have been described previously. The rabbit polyclonal anti-FAK (11) and anti-p85 (17) sera have also been described previously. The anti-Src mAb 327 was purchased from Calbiochem (San Diego, CA). The anti-Src mAb 2-17 was purchased from Quality Biotech (Camden, NJ). The rabbit polyclonal anti-p130 Cas (C-20) was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The anti-phosphotyrosine (anti-PY) mAb PY20 was purchased from Transduction Laboratories (Lexington, KY).
Construction of Expression Plasmids-The pKH3 plasmids encoding HA epitope-tagged FAK and Y397F mutants were described previously (17). The cDNA encoding other FAK mutants (D395A, D396A, D395/ 396A, P712/715A) used in this study were generated using site-directed mutagenesis by overlap extension using the polymerase chain reaction. The polymerase chain reaction products containing the desired mutations were treated with restriction enzymes KpnI and NdeI and then used to replace the corresponding fragments in the plasmid pGEX-FAK. The desired mutations were confirmed by dideoxy DNA sequencing. The cDNAs encoding FAK mutants in pGEX2T vectors were in-frame and transferred to pKH3 vectors using the BamHI and EcoRI sites. The pKH3 plasmids encoding FAK and its mutants were used to transiently express FAK proteins in human embryonic kidney (HEK) 293 cells.
The expression plasmid pCDM8-FAK and pCDM8-FAK P712/715A have been described previously (21,22). pKH3-FAK D395A was digested with MscI and NheI to generate a 1.8-kilobase fragment containing the D395A substitution. This was then subcloned into pCDM8-FAK from which the corresponding MscI-NheI fragment had been excised. The resulting plasmid was designated pCDM8-FAK D395A and used to transfect CHO cells.
The cDNA encoding the p85 subunit of PI3K that was kindly provided by Dr. L. C. Cantley (Harvard University) was inserted into pKH3 at the BamHI and EcoRI sites. The pEVX plasmid encoding c-Src was kindly provided by Dr. D. Shalloway (Cornell University).
Cell Culture and Transfections-HEK 293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Life Technologies, Inc.). One day after plating 5 ϫ 10 5 cells on 60-mm dishes, HEK 293 cells were transiently transfected with 1 g of pKH3 plasmids encoding FAK or its mutants using 10 l of Lipo-fectAMINE (Life Technologies, Inc.). In some cases, HEK 293 cells were co-transfected with pKH3-FAK or its mutants, pKH3-p85, and pEVX-Src. Two days after transfection, the cells were lysed in 1% Nonidet P-40 lysis buffer containing protease inhibitors as described previously (8).
CHO cells expressing wild type (WT) FAK, Y397F, and P712/715A have been described previously (21,22) and were maintained in F-12 medium with 10% fetal bovine serum and 0.5 mg/ml G418 (Life Technologies, Inc.). To generate cells stably expressing the D395A FAK mutant, CHO cells were grown on 10-cm dishes and transfected essentially as described (21) using LipofectAMINE following the manufacturer instructions. Clones were selected in G418-containing medium and screened by immunoblotting using anti-FAK and KT3.
In Vitro Binding Assays-The plasmids pGEX-Src.SH2 and pGEX-p85.NSH2 were described previously (17). The cDNA encoding the SH3 domain of p130 Cas (22) was cloned into pGEX2T at the BamHI and EcoRI sites to generate pGEX-Cas.SH3. GST fusion proteins were immobilized on glutathione-agarose beads and then incubated with HEK 293 cell lysates (ϳ10 g) containing various HA epitope-tagged FAK proteins in 1% Nonidet P-40 lysis buffer for 1 h at 4°C. The complexes were washed four times with 1% Nonidet P-40 lysis buffer, resolved by SDS-polyacrylamide gel electrophoresis, and analyzed by immunoblotting with anti-HA.
To measure the tyrosine phosphorylation and in vitro autophosphorylation of ectopically expressed FAK proteins, HA epitope-tagged FAK proteins were immunoprecipitated with anti-HA from HEK 293 cell lysates (50 g) and subjected to immunoblotting with anti-PY or to an in vitro kinase assay as described earlier (8).
To measure the PI3K activity associated with ectopically expressed FAK proteins in CHO cells, epitope-tagged FAK proteins were immunoprecipitated with KT3 from CHO cell lysates (750 g) and subjected to an in vitro PI3K assay as described previously (13).
Cell Migration Assays-CHO cell migration assays in 48-well chemotaxis chambers were carried out as described previously (21). For experiments with the PI3K inhibitors wortmannin and LY294002 (Sigma) (reconstituted in Me 2 SO and stored at Ϫ20°C), cells were harvested and washed as described earlier and then resuspended in F-12 medium containing the indicated concentrations of inhibitor or Me 2 SO such that the final concentration of Me 2 SO was constant. The cells were pretreated with the inhibitor at 37°C and 5% CO 2 for 30 min before loading them onto the chemotaxis chamber. They were allowed to migrate on 12 g/ml fibronectin for 6 h in the presence of the inhibitor and were then fixed and stained as described previously. For all cell migration assays, migrated cells were enumerated under a light microscope at ϫ200 magnification using the Image-Pro Plus software, Version 3.0.

RESULTS
It is known that PI3K plays an important role in growth factor-induced cell migration (28 -30). To examine the putative role of PI3K in FAK-promoted cell migration, CHO cells overexpressing WT FAK and control cells (Neo) were subjected to cell migration assays in the presence of specific PI3K inhibitors (Fig. 1). Treatment of WT cells with either PI3K inhibitor resulted in an inhibition of cell migration in a dose-dependent manner. Furthermore, in the presence of either 100 nM wortmannin or 75 M LY294002, the majority (ϳ80%) of cell migration promoted by FAK overexpression in WT cells was inhibited, with little effect on control cell migration. These results demonstrate that PI3K plays a role downstream of FAK in promoting CHO cell migration on fibronectin. However, from these results, it was not clear whether FAK-promoted cell migration is dependent on PI3K, either through direct binding at Tyr-397 of FAK or through another FAK-promoted signaling pathway.
To determine whether the direct binding of PI3K to FAK is a prerequisite for its function in FAK-promoted cell migration, a FAK mutant deficient only in PI3K binding was generated by a substitution of Asp-395 with Ala (see below). Songyang et al. (32) suggested that residues both N-and C-terminal to the phosphotyrosine may contribute to the specific binding of the SH2 domain to its cognate phosphotyrosine and showed that the optimal binding sequence for the p85 N-terminal SH2 domain is EEDpYVEM. An examination of FAK sequences revealed that sequences flanking Tyr-397 (TDDpYAEI) conformed well with the binding motif for the p85 N-terminal SH2 domain. Based on these observations, two Asp residues upstream of Tyr-397 were mutated either singly or in combination to generate a PI3K binding-defective FAK mutant. These FAK mutants were transiently expressed in HEK 293 cells, and their ability to bind the SH2 domains of Src and p85 was examined in vitro ( Fig. 2A). Similar to the mutation in Tyr-397, the mutation converting Asp-396 (or in combination with Asp-395) to Ala abolished FAK binding to the SH2 domains of Src and p85. Interestingly, a mutation in Asp-395 inhibited FAK binding to the SH2 domain of p85 but not Src. To examine if the FAK D395A mutant also selectively binds to Src but not p85 in intact cells, HEK 293 cells were transiently co-transfected with expression plasmids encoding HA epitope-tagged FAK or its mutants, c-Src, and p85. Two days after transfection, cells were lysed, and the association of FAK with Src or p85 was detected by co-immunoprecipitation (Fig. 2B). Consistent with the results from in vitro binding experiments, the mutation in Asp-396 abolished FAK association with both Src and p85, and importantly, the mutation in Asp-395 only prevented FAK association with p85 but not Src.
To examine the effects of FAK mutations in Asp-395 and Asp-396 on its tyrosine phosphorylation and in vitro autophosphorylation, HA epitope-tagged WT FAK or its mutants were transiently expressed in HEK 293 cells, immunoprecipitated with anti-HA, and subjected to immunoblotting with anti-PY or to an in vitro kinase assay (Fig. 3A). The tyrosine phosphorylation and in vitro autophosphorylation of FAK mutants containing a substitution of Asp-396 with Ala were as low as those of the Y397F mutant, indicating that Asp-396 is essential for the phosphorylation event upon Tyr-397. This also reflects the inability of the FAK D396A mutant to bind the SH2 domains of Src and p85. In contrast, the tyrosine phosphorylation and in vitro autophosphorylation of FAK D395A mutant remained comparable with the WT. These results indicated that the inability of the D395A mutant to bind PI3K was most likely not because of an overall conformational change. To exclude the possibility that the mutation in Asp-395 causes a subtle change in protein conformation and thereby affects the binding of FAK to other proteins, the capacity of FAK D395A to bind to the SH3 domain of p130 Cas was examined in vitro (Fig. 3B). The result clearly showed that the ability of FAK to bind the SH3 domain of p130 Cas was not affected by mutating Asp-395. Together, these data suggest that the inability of FAK D395A to bind p85 is indeed because of the preference of the p85 SH2 domain for an aspartic acid at residue 395 in FAK.
To determine whether PI3K binding is required for FAK-

FIG. 2. A mutation in Asp-395 of FAK abolishes its binding to p85 but not Src.
A, the lysates were prepared from HEK 293 cells that had been transfected with pKH3 plasmids encoding HA epitope-tagged FAK or its mutants as indicated. An aliquot of cell lysates was analyzed by immunoblotting with anti-HA to verify a similar expression of epitope-tagged FAK proteins in all samples (input). The lysates were incubated with immobilized GST-Src.SH2 or GST-p85.NSH2. After washing, the bound proteins were resolved by SDS-polyacrylamide gel electrophoresis and analyzed by immunoblotting with anti-HA. B, an equal amount of lysates from HEK 293 cells that had been co-transfected with expression plasmids encoding HA epitope-tagged FAK or its mutants, c-Src, and p85 was incubated with anti-Src (327) or anti-p85. The immunocomplexes were washed and analyzed by immunoblotting with anti-HA.

FIG. 3. A mutation in Asp-395 of FAK does not affect its tyrosine phosphorylation, in vitro autophosphorylation, and association with the SH3 domain of p130 Cas .
A, the lysates prepared as described in Fig. 2A were incubated with anti-HA. The immunocomplexes were subjected to immunoblotting with anti-FAK (top) or anti-PY (middle) or to an in vitro kinase assay (bottom) as described under "Experimental Procedures." B, the lysates from HEK 293 cells that had been transiently transfected with pKH3 plasmids encoding FAK (WT), D395A, or P712/715A mutants were incubated with immobilized GST-Cas.SH3 or GST-p85.NSH2. The bound proteins and an aliquot of lysates (input) were analyzed by immunoblotting with anti-HA. promoted cell migration, stable CHO cell lines overexpressing FAK D395A were established. The expression of FAK D395A was examined by immunoblotting with anti-FAK or KT3 (Fig.  4A). The expression levels of ectopic WT FAK and D395A mutant in CHO cells were comparable and severalfold higher than that of endogenous FAK. To examine the association of endogenous PI3K with WT and mutant FAK in CHO cells, ectopically expressed FAK proteins were immunoprecipitated by mAb KT3, and the immunocomplexes were subjected to an in vitro PI3K activity assay (Fig. 4B). Approximately 10% of total cellular PI3K activity were found to associate with ectopic WT FAK (data not shown). As expected, the levels of PI3K activity associated with WT FAK and P712/715A mutant were similar. Surprisingly, approximately 15% of PI3K activity that WT FAKs had were co-precipitated with Y397F and D395A mutants. The cell migration assays indicated a marked increase (ϳ2.5-fold) in cell migration for WT cells compared with both Y397F and control (Neo) cells, which were similar in their rates of migration. Interestingly, both D395A clones failed to promote cell migration (Fig. 4C). These results together suggest that the residual binding of PI3K to FAK Y397F or D395A mutant in CHO cells is not sufficient for promoting migration. Moreover, adhesion assays indicated that all clones had similar adhesive strength to fibronectin (data not shown), indicating that the decrease in migration observed in D395A clones was not because of an alteration in cell adhesion.
We have previously demonstrated that phosphorylation of Tyr-397 in FAK is required for binding the SH2 domains of Src and PI3K (17). Thus, it is unclear whether the inability of FAK Y397F to promote cell migration is because of its inability to bind Src and/or PI3K. To clarify this, the association of FAK D395A with c-Src in CHO cells was examined. Both WT FAK and FAK D395A but not FAK Y397F co-precipitated with c-Src in CHO cells (Fig. 5A). We have previously demonstrated that the association of p130 Cas with FAK and FAK-promoted phosphorylation of p130 Cas by Src are critical for FAK-promoted migration (22). We examined the phosphorylation state of p130 Cas in FAK D395A cells as an indicator of a FAK/Srcdepending cell migration pathway as well as an indirect indicator of FAK/p130 Cas association. Consistent with prior experiments (22), WT FAK exhibited a promotion of p130 Cas phosphorylation compared with the FAK P712/715A (Fig. 5B), which is unable to bind p130 Cas or promote p130 Cas phosphorylation (22). FAK D395A was able to promote p130 Cas phosphorylation at an estimated 2-fold level above WT FAK, indicating that the FAK/Src specific pathway is still functioning in the D395A mutant. Taken together, these results strongly suggest that PI3K binding to FAK is required for FAK to promote cell migration and that FAK:Src association alone may be insufficient for this cellular function. DISCUSSION In this report, we examined the role of PI3K in FAK-promoted cell migration. First we showed that two PI3K inhibitors, wortmannin and LY294002, were able to inhibit FAKpromoted migration in a dose-dependent manner. Our results are consistent with other work showing the importance of PI3K in increased cell motility stimulated by growth factors such as PDGF and hepatocyte growth factor (28 -30). However, these data do not indicate whether PI3K association with FAK is required for increased cell migration and cannot rule out the involvement of PI3K in a FAK-independent migration pathway. To address this, we next established stable cell lines overexpressing FAK and a mutant defective in PI3K binding. Using this system, we showed that the PI3K binding-deficient mutant failed to promote cell migration toward fibronectin. These results strongly suggest that FAK:PI3K association is required for FAK-promoted cell migration.
We have previously identified that the autophosphorylation site Tyr-397 is the primary site for PI3K binding (17) and is FIG. 4. FAK D395A mutant fails to promote migration in CHO cells. A, an equal amount of lysates prepared from a control clone (Neo), a CHO cell clone expressing WT FAK, or two independent clones expressing FAK D395A (designated D395A-1 and D395A-2) was analyzed by immunoblotting with anti-FAK or KT3. B, ectopically expressed FAK and mutants were immunoprecipitated from CHO cell lysates using mAb KT3, and the immunocomplexes were subjected to an in vitro assay for PI3K activity, as described under "Experimental Procedures." The location of the origin (Ori) and phosphatidylinositol 3-phosphate (PIP) are indicated at the right. C, cells from a control clone or cells overexpressing WT FAK, Y397F, or D395A-1, and D395A-2 were subjected to a cell migration assay as described under "Experimental Procedures." Cells in the upper half of the chamber were allowed to migrate through a porous membrane toward the lower chamber containing fibronectin (FN) as a chemoattractant at the indicated concentrations. Migrated cells were fixed, stained, and counted using a light microscope. Mean cell counts are shown from at least twelve fields and three independent experiments. Relative cell migration was calculated based on the level of WT cell migration at 6 g/ml fibronectin. Error bars represent the standard errors. required for the function of FAK in promoting cell migration (21). We show here that the introduction of a mutation in Asp-395 completely abolishes FAK association with PI3K SH2 domain but not Src SH2 domain in vitro. The D395A mutant transiently expressed in HEK 293 cells also loses its ability to bind to intact p85 but not Src. However, in CHO cells some residual binding of PI3K to Y397F and D395A mutants can be detected, indicating that other mechanisms such as SH3 interaction may also be involved. This result is consistent with the previous report that the SH3 domain of p85 could bind to a proline-rich region (amino acids 875-884) of FAK in vitro (33) and suggest that an SH3 interaction besides SH2 interaction might also participate in FAK:PI3K interactions in response to certain stimuli in different cell types. Furthermore, together with our observation that FAK D395A mutant retained its tyrosine phosphorylation and in vitro phosphorylation to the same level as WT FAK, it appears that Asp-395 makes a major contribution for specific binding of the SH2 domains of PI3K to FAK. This finding also supports the notion that residues Nterminal of phosphotyrosine are critical for specific binding to certain SH2 domain-containing proteins.
Previous results from our laboratory showed that a FAK mutant (P712/715A) deficient in p130 Cas binding failed to promote cell migration (22). In this report, we showed that FAK D395A deficient in PI3K binding also failed to increase cell migration. Although a residual binding of PI3K to D395A mutant was detected in CHO cells, it was apparently not sufficient for promoting cell migration. In fact, an analogous situation has also been described that a residual binding of p130 Cas to FAK P712/715A mutant possibly mediated by a second prolinerich region (amino acids 875-884) on FAK does not allow for FAK-promoted migration (22). These results together suggest that thresholds of signals transmitted independently through the associations of p130 Cas and PI3K with FAK have to be reached for enhanced cell migration. Lack of either one will not turn on an "intracellular machinery" for the induction of cell migration. Another possible interpretation is that the inability of FAK D395A to promote cell migration is because of a defect in p130 Cas binding. This defect may result from a subtle conformation change by the D395A substitution, or a mutually dependent binding of PI3K and p130 Cas to FAK. The latter possibility was excluded by our observation that the D395A mutant retained its ability to bind the SH3 domain of p130 Cas in vitro and promote p130 Cas phosphorylation in vivo.
The tyrosine phosphorylation of p130 Cas and its subsequent association with the adapter protein Crk have been shown to play an important role in promoting cell migration (34). In addition, the tyrosine phosphorylation of p130 Cas has also been shown to be a result of binding to FAK (22). Here, we have shown that expression of FAK D395A increased p130 Cas tyrosine phosphorylation at a roughly 2-fold level above WT FAK in CHO cells. Although the reason for this elevation in p130 Cas phosphorylation is unclear, this result supports our conclusion that p130 Cas binding and phosphorylation is necessary but not sufficient for FAK-promoted cell migration. Furthermore, several studies have shown that Src is important for phosphorylation of p130 Cas (35,36). Because FAK D395A still binds to Src, our results also suggest that FAK:Src association alone is not sufficient for the enhancement of cell migration. However, we cannot exclude the necessity of Src in FAK-promoted cell migration because of its function, at least, in phosphorylating p130 Cas . Therefore, it is likely that FAK-promoted cell migration depends on the coordinate regulation of a number of these signaling events involving PI3K, p130 Cas , and Src.
We have previously shown that FAK:PI3K association is enhanced by cell adhesion to ECM proteins (13) or PDGF stimulation (31). In this report, we have shown that this association is likely to be involved in cell migration upon cell adhesion to ECM proteins. However, it is not clear if this interaction is also involved in PDGF-stimulated cell migration. Furthermore, we do not know if FAK generally participates in growth factor-induced cell motility using a mechanism similar to that in ECM protein-induced cell migration. It is known that the tyrosine phosphorylation of FAK is stimulated by a number of growth factors (31,37). Although the mechanisms by which growth factor receptors and integrins regulate FAK phosphorylation may be different (38), it is possible that phosphorylated FAK recruits the same set of intracellular signaling molecules to facilitate cell migration. Answers to these questions will be interesting, and experiments are now in progress to test these possibilities.