Tyrosine phosphorylation of type Iγ phosphatidylinositol phosphate kinase by Src regulates an integrin–talin switch

Integrin binding to the ECM stimulates the formation of cell–matrix adhesions and mediates the linkage between the ECM and actin cytoskeleton (for review see Geiger et al., 2001; Zamir and Geiger, 2001). This linkage is generated by a submembrane interconnecting complex that consists of structural proteins such as talin, vinculin, paxillin, α-actinin, and tensin, signaling molecules including tyrosine kinases such as FAK and Src, and serine/threonine kinases, as well as various adaptor proteins. The molecular complexity of cell–matrix adhesions enables them to modulate both cell anchorage and transmembrane signaling (Sastry and Burridge, 2000; Geiger and Bershadsky, 2001). The most common forms of integrin-mediated cell–matrix adhesions in cultured cells are termed focal adhesions (FAs), fibrillar adhesions, or focal complexes, based on their size, shape, and location within the cell. Signal transduction through FAs has been implicated in the regulation of a number of key cellular processes, including growth factor–induced mitogenic signals, cell survival, cell locomotion and morphogenesis, and tissue assembly (Burridge and Chrzanowska-Wodnicka, 1996).

The dynamic regulation of FA assembly and disassembly is the central process for cell motility in response to extracellular stimuli (for review see Parsons et al., 2000). In addition to regulation by Rho family small GTPases, another key mechanism for modulation of FA organization and stability is tyrosine phosphorylation. Nonreceptor tyrosine kinases, FAK, and Src family kinases are involved in the phosphorylation of several FA proteins including paxillin, p130Cas, tensin, and FAK itself (for review see Cary and Guan, 1999; Frame et al., 2002). Tyrosine phosphorylation provides docking sites for additional molecules that contain phosphotyrosine (pY)-binding (PTB) domains (for review see Zamir and Geiger, 2001). Inhibition of tyrosine kinases decreases the level of intracellular tyrosine phosphorylation and blocks the development of FAs (for review see Frame et al., 2002). Consistently, activation of protein tyrosine phosphatases disrupts FAs, whereas their inhibition stimulates FA assembly. On the other hand, overactivated Src results in high levels of tyrosine phosphorylation but disruption of FAs (Kellie et al., 1986), and Src^−/− cells form FAs that fail to turn over into fibrillar adhesions (Volberg et al., 2001). In addition, Src family kinases were implicated in regulation of integrin–cytoskeleton interactions (Felsenfeld et al., 1999). These data suggest that tyrosine phosphorylation plays an essential role in the formation and turnover of FAs. However, the exact mechanism underlying tyrosine phosphorylation–mediated modification of the integrin–cytoskeleton linkage and FA dynamics are still poorly defined.

Phosphatidylinositol 4,5-bisphosphate (PI4,5P₂), a lipid second messenger, modulates FA formation by binding to vinculin, α-actinin, and talin facilitating protein–protein interactions and by regulating Rho family small G proteins (for review see Sechi and Wehland, 2000). The concentration of PI4,5P₂ appears to be regulated locally, enabling spatial and temporal changes in its availability and signaling. Type Iγ661 phosphatidylinositol phosphate kinase (PIPKIγ661) targets to FAs and regulates FA assembly by generating PI4,5P₂ and targeting talin to FAs (Di Paolo et al., 2002; Ling et al., 2002). Our previous data showed that PIPKIγ661 was tyrosine phosphorylated by FAK signaling, and this enhances PIP kinase activity and correlated with increased talin association (Ling et al., 2002). Here, we demonstrate that PIPKIγ661 is tyrosine phosphorylated on Y644 by Src family kinases and is potentially regulated by FAK signaling. Tyrosine phosphorylation of PIPKIγ661 is required for its strong talin interaction and stable FA targeting. Interestingly, PIPKIγ661 tyrosine phosphorylation strengthens its interaction with talin, effectively competing with β1-integrin for binding to talin. The combined data support a novel mechanism regulating FA dynamics by tyrosine phosphorylation and PI4,5P₂ signaling.

Previously, we observed that tyrosine phosphorylation of PIPKIγ661 was dependent on adhesion and FAK signaling (Ling et al., 2002). Here, we show that PIPKIγ661 tyrosine phosphorylation is dependent on both FAK phosphorylation on Y397 and FAK activity because neither coexpressed Y397F nor kinase-dead mutants of FAK induced phosphorylation of PIPKIγ661 (Fig. 1 A). To map the phosphorylation sites, we used the following PIPKIα/γ chimeras: PIPKIα/γ635, PIPKIα/γ661, and PIPKIα/γ636–661 as defined in the Materials and methods. We coexpressed the chimeras with wild-type FAK, and then analyzed their tyrosine phosphorylation by immunoprecipitation and immunoblotting. Both PIPKIα/γ661 and PIPKIα/γ636–661 (but not PIPKIα/γ635) were phosphorylated after FAK coexpression (Fig. 1 B). The COOH-terminal truncation mutants of PIPKIγ661 showed consistent results; PIPKIγ661Δ20 (1–641) was not phosphorylated, PIPKIγ661Δ13 (1–648) was phosphorylated but to a lesser extent, and PIPKIγ661Δ3 (1–657) was phosphorylated similar to the wild-type kinase (unpublished data). These data suggest that the major FAK-mediated phosphorylation sites are in the last 26 amino acids of PIPKIγ661. To define the phosphorylated residues, two Y to F point mutants were made, Y644F and Y649F (Fig. 1 C). The wild-type or Y to F mutant PIPKIγ661 was coexpressed in HEK293T cells with wild-type FAK or vector control, and tyrosine phosphorylation of PIPKIγ661 was determined after immunoprecipitation using anti-PIPKIγ antibody. As shown in Fig. 1 C, when FAK was overexpressed, PIPKIγ661Y649F was phosphorylated similar to the wild type. However, PIPKIγ661Y644F and PIPKIγ661Y644/649F showed dramatically reduced phosphorylation (Fig. 1 C), suggesting that Y644 is the phosphorylation site downstream of FAK signaling.

Src family tyrosine kinases are key factors that associate with FAK, thereby playing a fundamental role in downstream signaling and FAK-regulated processes (Parsons et al., 2000). Y397 is the autophosphorylation site in FAK and serves as a Src-binding site. Previous results have shown that either the Src-binding site (Y397) mutant or a kinase-dead mutant abrogated FAK's ability to promote PIPKIγ661 phosphorylation (Fig. 1 A). This suggests that phosphorylation of PIPKIγ661 is mediated by Src or FAK. To address the role of Src family kinases in PIPKIγ661 phosphorylation, Src activity was examined. The Src-specific inhibitor PP2 (but not the inactive analogue PP3) inhibited both basal and FAK-induced tyrosine phosphorylation of PIPKIγ661 (Fig. 2 A), showing that Src family kinases play an important role in tyrosine phosphorylation of PIPKIγ661. To further explore this possibility, we used the c-Src, Fyn, and Yes triple knockout cells (SFY^−/−) and c-Src–reintroduced SFY^−/− cells (Src^+/+) to analyze the tyrosine phosphorylation of endogenous PIPKIγ661. As shown in Fig. 2 B, endogenous PIPKIγ661 showed very low phosphorylation in SFY^−/− cells, whereas reintroduced c-Src restored PIPKIγ661 phosphorylation. Our results indicate that c-Src is directly involved in the tyrosine phosphorylation of PIPKIγ661, whereas the role of other Src family kinases members Fyn and Yes need further clarification. When the FAK knockout (FAK^−/−) or their parent (FAK^+/+) cells were examined, the endogenous PIPKIγ661 showed similar tyrosine phosphorylation in both cell lines (Fig. 2 B), indicating that FAK did not phosphorylate PIPKIγ661 directly, but enhanced its phosphorylation via regulation of Src.

These whole-cell assays were confirmed by in vitro kinase assays, which demonstrated that c-Src directly phosphorylated the His-tagged recombinant PIPKIγ661 tail (439–661; Fig. 2 C), whereas FAK only phosphorylated it very weakly (unpublished data). Both His-tagged recombinant PIPKIγ661Y644F and PIPKIγ661Y644/649F tails were not efficiently phosphorylated by c-Src in vitro compared with the wild-type PIPKIγ661, whereas the PIPKIγY649F tail was efficiently phosphorylated (Fig. 2 C). Furthermore, the phosphopeptide-mapping results confirmed that Y644 is the major substrate of c-Src because the Y644F mutant failed to yield two major phosphorylation spots compared with the wild-type PIPKIγ661 map (Fig. 2 D). These results demonstrate that Y644 is the major tyrosine phosphorylation site of PIPKIγ661 and that Src is the enzyme responsible for PIPKIγ661 phosphorylation.

To decipher the mechanism of PIPKIγ661 phosphorylation by Src, we examined whether there was an interaction between Src and PIPKIγ661. HEK293T cells were cotransfected with c-Src and PIPKIγ661 for 48 h, harvested, and PIPKIγ661 was immunoprecipitated with purified anti-PIPKIγ antibody. As shown in Fig. 3 A, c-Src was coimmunoprecipitated with PIPKIγ661. To further explore FAK's role in PIPKIγ661 phosphorylation, we coexpressed FAK with c-Src and PIPKIγ661. Coexpression of the wild-type FAK enhanced both tyrosine phosphorylation and Src association with PIPKIγ661 (Fig. 3 B). FAKY397F, which lacks the docking site for Src, inhibited the PIPKIγ661–c-Src association and failed to increase PIPKIγ661 phosphorylation to the same extant as FAK (Fig. 3 B). This demonstrates that the interaction of Src with FAK is required for FAK signaling to enhance Src binding and phosphorylation of PIPKIγ661.

Nevertheless, the PIPKIγ661 phosphorylation by Src is not absolutely dependent on FAK because PIPKIγ661 can be phosphorylated in FAK-null cells (Fig. 2 B). In addition, PIPKIγ635 and the PIPKIγ661 Y to F mutants coimmunoprecipitated with c-Src (unpublished data), suggesting that c-Src association with PIPKIγ requires a site distinct from the last 26 amino acids of PIPKIγ661 and independent of phosphorylation at Y644 or Y649. In A431 cells, endogenous PIPKIγ661 was tyrosine phosphorylated and associated with endogenous Src upon cell adhesion to type I collagen (Fig. 3 C). These results indicate that signals initiated from integrin binding to ligand regulate the Src–PIPKIγ661 association, and consequently, tyrosine phosphorylation of PIPKIγ661. The requirement of integrin-mediated cell adhesion suggests that FAK regulation is one potential signaling pathway that modulates the Src–PIPKIγ interaction. A431 cells highly express EGF receptors and, as a result, Src activity could be greatly enhanced. In other cell lines assayed, including HEK293T, NRK, and NIH3T3, the interaction between endogenous PIPKIγ661 and Src was more difficult to detect, although endogenous PIPKIγ661 is tyrosine phosphorylated in these cell lines and is dependent on adhesion on type I collagen (unpublished data; Ling et al., 2002).

Tyrosine phosphorylation of FA proteins regulates cell spreading and migration by generating protein–protein interaction sites for SH2 and PTB domains (for review see Panetti, 2002). To further explore the physiological significance of PIPKIγ661 tyrosine phosphorylation, we determined if the endogenous PIPKIγ661 bound to talin is tyrosine phosphorylated. Compared with endogenous PIPKIγ661 immunoprecipitated by anti-PIPKIγ antibody, endogenous PIPKIγ661 coimmunoprecipitated by anti-talin antibody had equivalent tyrosine phosphorylation but less PIPKIγ661 (Fig. 4 A), suggesting that PIPKIγ661 bound to talin is tyrosine phosphorylated to a greater extent than free PIPKIγ661 in vivo.

Cotransfection experiments also indicated that increased PIPKIγ661 tyrosine phosphorylation, induced by FAK, correlates with an increase in PIPKIγ661 bound to talin (Fig. 4 B; Ling et al., 2002). To characterize the role of PIPKIγ661 tyrosine phosphorylation, we examined the talin interaction with PIPKIγ661 phosphorylation-defective mutants by coimmunoprecipitation using HEK293T cells. Compared with the wild-type PIPKIγ661, PIPKIγ661Δ20 did not interact with talin, PIPKIγ661Δ13 had decreased talin association, and PIPKIγ661Δ3 maintained identical talin association (Fig. 4 C). The tyrosine point mutants PIPKIγ661Y644F and PIPKIγ661Y644/649F both showed substantially diminished talin association, whereas PIPKIγY649F retained talin association comparable with wild-type PIPKIγ661 (Fig. 4 C). These data demonstrate that tyrosine phosphorylation of the PIPKIγ661 COOH terminus regulates its interaction with talin.

The PIPKIγ661–talin association is important for targeting PIPKIγ661 to FAs (Di Paolo et al., 2002; Ling et al., 2002). To further explore targeting requirements, FA localization of PIPKIγ661 mutants were examined in NRK cells. Consistent with the ability to interact with talin in coimmunoprecipitation experiments, PIPKIγ661Δ20 lost FA targeting, PIPKIγ661Δ3 targeted to FAs and colocalized with talin similar to wild-type PIPKIγ661, and PIPKIγ661Δ13 partially retained FA targeting (Fig. 5 A). These data indicate that in addition to the minimum region ⁶⁴²WVYSPLH⁶⁴⁸ for in vitro talin binding (Di Paolo et al., 2002), in vivo talin association, FA targeting, and phosphorylation of PIPKIγ661 also require other residues in the COOH-terminal 26 amino acids. Consistently, the Y to F mutants lacking phosphorylation showed reduced talin association (Fig. 4 C) and lost FA targeting, whereas PIPKIγY649F localized to FAs similar to its wild-type counterpart (Fig. 5 A). Analysis of 600 cells under each condition in three independent experiments revealed a correlation with talin association (Fig. 4 C), FA targeting (Fig. 5 B), and tyrosine phosphorylation of PIPKIγ661 (Fig. 1 B). Additionally, PIPKIγ661 overexpressed in Src^+/+ cells showed significantly greater FA targeting and colocalization with the pY staining compared with those overexpressed in SFY^−/− cells (Fig. 5 C). In the FAK^−/− cells, overexpressed PIPKIγ661 showed comparable FA targeting as in the parent cells (unpublished data), consistent with results demonstrating that FAK does not directly phosphorylate PIPKIγ661 in vivo. Together, these data show that c-Src–induced tyrosine phosphorylation of PIPKIγ661 enhances its FA targeting.

To determine whether phenylalanine substitution or loss of the phosphorylated tyrosine is responsible for the reduced talin binding of PIPKIγ661 mutants, talin interactions were assessed by GST pull-down using recombinant nonphosphorylated His-PIPKIγ661 wild-type or Y to F mutant COOH terminus and GST-talin head domain. His-PIPKIγ661 Y to F mutants showed identical binding to talin compared with the wild-type protein (Fig. 6 A), suggesting that the decreased in vivo talin association resulted from the lack of tyrosine phosphorylation and not the phenylalanine substitution.

To address the role of tyrosine phosphorylation, we designed PIPKIγ661 peptides containing the talin-binding sequence ⁶⁴²WVYSPLH⁶⁴⁸ and flanking sequences, including normal peptide (PN), Y644-phosphorylated (pY644), Y649-phosphorylated (pY649), or Y644/Y649 dual-phosphorylated (pY644/649) peptides (Fig. 6 B). HEK293T cell lysates were supplemented with these peptides and subsequently used for immunoprecipitation. The phosphorylated peptides inhibited PIPKIγ661–talin association much more efficiently than the nonphosphorylated peptide (Fig. 6 B). To further explore this, we labeled the PIPKIγ661 peptides with fluorescein at the NH₂ terminus and examined their direct binding with purified GST-talin head using fluorescence polarization anisotropy. As shown in Fig. 6 C, all of the phosphorylated peptides showed higher binding affinity for GST-talin head compared with the nonphosphorylated peptide. These results demonstrate that PIPKIγ661 tyrosine phosphorylation enhances talin binding.

Talin plays a key role in integrin-mediated signaling by linking integrins to the actin cytoskeleton and regulating integrin activation (Liddington and Ginsberg, 2002). The F3 lobe of the band 4.1, ezrin, radixin, moesin homology (FERM) domain of talin, structurally homologous to a PTB domain, binds to the β-integrin cytoplasmic tail (Calderwood et al., 2002; Garcia-Alvarez et al., 2003) and the last 26 amino acids of PIPKIγ661 (Di Paolo et al., 2002; Ling et al., 2002). Because tyrosine phosphorylation of PIPKIγ661 enhanced its interaction with talin, it is important to explore whether PIPKIγ661 would displace β-integrin from talin in a phosphorylation-dependent manner. We examined the ability of recombinant PIPKIγ proteins or peptides to compete with His-β1-integrin tail for binding to talin in vitro. His-PIPKIγ661 tail, but not His-PIPKIγ635 tail, competed with His-β1-integrin tail in a dose-dependent manner (Fig. 7 A), consistent with the recent report by Barsukov et al. (2003). In addition, we found that the phosphorylated PIPKIγ661 peptides were ∼20-fold more effective in displacement of His-β1-integrin tail from GST-talin head than the nonphosphorylated peptide (Fig. 7 B). The analysis of the integrin competition data revealed that Y644 or Y649 single- or double-phosphorylated PIPKIγ661 peptides shared similar efficiencies for displacing His-β1-integrin tail from GST-talin head that is ∼20-fold higher than the nonphosphorylated PIPKIγ661 peptide (Fig. 7 C). Our current data suggest that tyrosine-phosphorylated PIPKIγ661 could be a key element in Src-mediated regulation of integrin–cytoskeleton linkage by competing with β-integrin binding to the talin FERM domain.

As a key residue for binding to talin (Di Paolo et al., 2002), W642 of PIPKIγ661 aligns with the conserved tryptophan in the β-integrin tail (Garcia-Alvarez et al., 2003). PIPKIγ661 binding to talin through an interaction with W642 would position Y644 of PIPKIγ661 adjacent to K357 and R358 of talin. To determine if the increased binding affinity of the phosphorylated PIPKIγ661 sequences was due to these two basic residues, each residue was respectively mutated to glutamine, which is the most conservative mutation that eliminates charge but maintains side chain size. Although both K357Q and R358Q mutants abolished His-β1-integrin tail binding, neither mutation had a detectable effect on His-PIPKIγ661 tail binding (Fig. 8 A). The binding curves of His-PIPKIγ661 tail to the wild-type or mutant GST-talin head showed comparable properties (Fig. 8 B), showing that both K357Q and R358Q mutants have no effect on talin binding to PIPKIγ661. These data indicate that PIPKIγ661 binds to talin on an overlapping yet distinct site compared with β1-integrin, as concluded by Barsukov et al. (2003).

To assess the impact of PIPKIγ661 phosphorylation, the phosphorylated peptides were used to compete with His-PIPKIγ661 binding to the wild-type or mutant GST-talin head. As shown in Fig. 8 C, the K357Q mutant retained high affinity binding for the Y644-phosphorylated peptide, but lower affinity for the Y649-phosphorylated peptide. R358Q showed a loss of increased affinity for both Y644- and Y649-phosphorylated peptides (Fig. 8 C). These results indicate that both phosphorylated Y644 and Y649 require R358 for high affinity binding, but phosphorylated Y649 appeared to also require K357. To further examine this hypothesis, we examined the direct binding of fluorescein-labeled PIPKIγ661 peptides with the mutant GST-talin head. To the K357Q mutated talin head, the binding affinity of Y649-phosphorylated peptide dropped to the comparable level with the nonphosphorylated peptide, whereas the Y644- and double-phosphorylated peptides maintained their higher binding affinity (Fig. 8 D). All of the phosphorylated peptides showed decreased binding affinity to the R358Q mutant compared with the wild-type talin head (Fig. 8 D), suggesting that R358 is critical for phosphorylated PIPKIγ661 peptide binding. The anisotropy data revealed that phosphorylation at Y644 or Y649 enhanced the binding affinity to talin head ∼15-fold. These results support the GST pull-down displacement data that K357 is responsible for tighter binding induced by phosphorylation of Y649, whereas R358 is critical for tighter binding to talin induced by phosphorylation of both Y644 and Y649. Both nonphosphorylated and phosphorylated PIPKIγ661 peptides showed 3–4 times lower maximum change of anisotropy with the R358Q talin head mutant, consistent with the recent work by Barsukov et al. (2003). Previous data demonstrate that there is no detectable effect on full-length PIPKIγ661 tail binding to R358Q talin head (Fig. 8, A and B). This gives rise to the possibility that residues in the last 26 amino acids other than the 15 amino acids in the synthesized PIPKIγ661 peptide are involved in the interaction with talin head domain. In addition, the K_d value of nonphosphorylated PIPKIγ661 peptide for the full talin head is eightfold lower than the reported value for F2F3 domain of talin (Barsukov et al., 2003), suggesting additional contribution to the binding energy.

Because dimerization of GST may introduce artifactual results, we also performed experiments using His-tagged PIPKIγ661 tail and S-tagged recombinant talin head proteins to repeat the representative results. These experiment showed consistent results with those obtained from the GST pull-downs (unpublished data).

The enhanced association of the phosphorylated PIPKIγ661 peptides with talin raised the question of whether the β1-integrin tail may bind to talin with increased affinity when phosphorylated at the ⁷⁸⁰WDT⁷⁸² and ⁷⁸⁵NPIY⁷⁸⁸ motif because the WDT motif aligns with ⁶⁴²WVY⁶⁴⁴ of PIPKIγ661. To address this question, β1-integrin peptides phosphorylated at T783 and/or Y788 were synthesized and used to compete integrin tail binding to talin. The T783-phosphorylated peptide competed with the integrin tail for talin to a similar extent as the nonphosphorylated peptide, whereas the Y788-phosphorylated peptide showed lower binding affinity (Fig. 9). These data indicate that phosphorylation of the Y788 residue diminished the interaction with GST-talin, consistent with previous reports (Tapley et al., 1989; Garcia-Alvarez et al., 2003). These results identify a region on the F3 lobe of talin as a PTB domain that can bind both phosphorylated PIPKIγ661 or unphosphorylated integrin. The loss of talin binding to phosphorylated integrin and the enhanced binding to phosphorylated PIPKIγ661 defines a mechanism by which tyrosine phosphorylation of PIPKIγ661 tightly regulates FA dynamics via an integrin–talin switch.

PIPKIγ661 is a newly recognized FA-targeting phosphatidylinositol phosphate kinase isoform, and was identified as an important regulator for FA assembly through recruiting talin to and generating PI4,5P₂ at FAs (Di Paolo et al., 2002; Ling et al., 2002). Previously, the tyrosine phosphorylation of PIPKIγ661 was shown to regulate its PIP kinase activity (Ling et al., 2002). The present work demonstrates that the PIPKIγ661–talin interaction and the PIPKIγ661 FA targeting require Src-mediated tyrosine phosphorylation of PIPKIγ661 at Y644. This enables PIPKIγ661 to produce PI4,5P₂ at FAs, thereby regulating FA assembly. Tyrosine phosphorylation has been shown to regulate FA turnover; however, the underlying mechanism is poorly defined. We speculate that tyrosine phosphorylation of PIPKIγ661 provides a plausible mechanism for the dynamic regulation of FA organization.

PIPKIγ661 binds talin via the talin FERM domain. The binding region has been narrowed to the F3 lobe of the FERM domain, and this lobe has structural homology to PTB domains (Calderwood et al., 2002; Garcia-Alvarez et al., 2003). Here, we have shown that phosphorylated PIPKIγ661 peptides bind with high affinity to the F3 lobe, and K357 and R358 are key residues that bestow increased affinity to tyrosine-phosphorylated PIPKIγ661. To understand the interaction between talin and phosphorylated PIPKIγ661, we structurally modeled the talin F3 lobe with the dual phosphorylated PIPKIγ661 peptide (Fig. 10 A). This model shows that Y644 is located close to K357 and R358 of talin and forms a charge–charge interaction with these residues when phosphorylated. This model is supported by the observation that mutation of K357 and R358 to glutamine ablated the enhanced binding of the Y644-phosphorylated sequences. Unlike the integrin sequence, P646 of PIPKIγ661 creates a turn that also positions Y649 directly adjacent to K357 of talin. This model is consistent with the result that the K357Q mutant lost the enhanced binding to the Y649-phosphorylated PIPKIγ661 peptide. Y649 of PIPKIγ661 is in a similar position to Y788 of β1A-integrin, which was reported to be phosphorylated in Src-transformed cells (Sakai et al., 2001). However, phosphorylation of Y788 resulted in the loss of integrin binding to the talin head (Fig. 9; Tapley et al., 1989; Garcia-Alvarez et al., 2003). This indicates that P786 of integrin locates integrin Y788 to a position where the tyrosine-phosphorylated residue may be sterically hindered from interacting with K357 or R358 of talin.

Our data demonstrate that the FERM domain of talin acts as a PTB domain by coordinating binding of two different tyrosine-phosphorylated sequences in PIPKIγ661. This may be a general function of FERM domains because the K357 and R358 positions are conserved in other FERM domains including moesin, radixin, ezrin, band 4.1. This suggests that there is structurally conserved binding between FERM domains and phosphorylated tyrosine residues in sequences similar to the PIPKIγ661 tail. In addition, the F3 lobe of talin does bind tyrosine-phosphorylated residues in the context of Y649 in the sequence PLHY, which is similar to the NPXY sequences that PTB domains normally bind. This demonstrates that the F3 lobe of talin is a remarkably flexible PTB domain that can bind two distinct phosphorylated tyrosine residues in different sequence contexts.

Current results suggest a mechanism as illustrated in Fig. 10 B, where talin switches between an association with integrin or PIPKIγ661 in a dynamic and highly regulated fashion. It is demonstrated here that both FAK and Src are involved in PIPKIγ661 tyrosine phosphorylation. Endogenous PIPKIγ661 tyrosine phosphorylation was dependent on adhesion to type I collagen, an α₂β₁ integrin-dependent signaling process. As a component of this pathway, FAK activates Src, which then phosphorylates PIPKIγ661. In addition, FAK stimulates an association between Src and PIPKIγ661. This association was also stimulated by adhesion to type I collagen, indicating that the Src association is physiologically relevant. The Src–PIPKIγ661 association is likely important for efficient phosphorylation of PIPKIγ661. These combined data suggest that α₂β₁ integrin signaling is one mechanism that leads to phosphorylation of PIPKIγ661.

The PIPKIγ661-binding site on talin overlaps with the integrin-binding site, and this competitive interaction is likely the key for modulation of PIPKIγ661 function at FAs. Current results demonstrate that tyrosine phosphorylation of PIPKIγ661 dramatically enhances talin interaction and FA targeting, which enable the generation of PI4,5P₂ at FAs to improve the integrin–talin interaction. The enhancement of the integrin–talin association by PI4,5P₂ may ultimately displace PIPKIγ661 from talin, resulting in a reduction of PI4,5P₂. Via this mechanism, PI4,5P₂ production is self limiting, and in the context of this model would be highly dynamic. The tyrosine phosphorylation of PIPKIγ661 may also be regulated by PI4,5P₂ production because the kinase-dead enzyme is 20-fold less tyrosine phosphorylated (Ling et al., 2002). This results in a spatial and temporal generation of PI4,5P₂ that is regulated by integrin binding to ECM, and by Src, FAK, or other signals. Although we have not yet identified a kinase that phosphorylates Y649, this potential phosphorylation site may be a link to other signaling pathways for the modulation of PIPKIγ661–talin association and FA dynamics.

FAK and Src family members are important regulators of FA dynamics (for review see Cary and Guan, 1999; Frame et al., 2002). Src is required in directed migration stimulated by growth factors. There is also evidence that adhesion to the ECM is required for cells to become responsive to growth factors, and this responsiveness correlates with stimulation of PI4,5P₂ production by ECM adhesion (McNamee et al., 1993). As shown in Fig. 10 B, Src activation by integrin leads to phosphorylation of PIPKIγ661, enhancing both talin binding and PIP kinase activity (Ling et al., 2002). Because growth factor receptor and integrin signaling regulate Src activity, both pathways may lead to phosphorylation of PIPKIγ661 and regulation of migration.

Recent work has revealed that talin provides the initial connections between the ECM, integrin, and the cytoskeleton (Jiang et al., 2003). Moreover, talin binding to integrin has been identified as the final step leading to integrin activation (Tadokoro et al., 2003). This “inside-out” integrin signaling modulates both integrin-binding affinity to the ECM and internal connection to the cytoskeleton via the regulated association with talin (Calderwood and Ginsberg, 2003; Calderwood et al., 1999, 2002; Ulmer et al., 2003). Currently, the only known enhancer of the integrin–talin interaction is PI4,5P₂, which induces an ∼10-fold increase of the binding affinity (Martel et al., 2001). Regulated production of PI4,5P₂ at FAs, via Src-mediated phosphorylation of PIPKIγ661, may be key for modulation of the integrin–talin interaction, and subsequently, integrin activity and binding affinity for the ECM. This would provide a mechanism by which migrating cells can spatially and temporally control adhesion to the ECM differently at the leading or trailing edge.

Human talin1 head region (1–435) was cloned from HEK293T cells by RT-PCR using the OneStep RT-PCR kit (QIAGEN) and was subcloned into pET42. The COOH-terminal truncated or site-directed mutagenesis for the PIPKIγ661 mutants or talin head was performed using PCR-primer overlap extension with mutagenic primers. PIPKIγ661 mutants were then subcloned into pcDNA3.1, and talin head mutants were subcloned into pET42. His-tagged mouse β1-integrin tail was made following the published method (Pfaff et al., 1998), and was subcloned into pET28. The COOH terminus of PIPKIγ635 (439–635), PIPKIγ661 (439–661), or the Y to F mutants of PIPKIγ661 were PCR amplified from the corresponding pcDNA3.1 constructs and subcloned into pET28. PIPKIα/γ439–635 or PIPKIα/γ439–661 were constructed by fusing residues 1–444 of PIPKIα and residues 439–635 of PIPKIγ635 (PIPKIα/γ635) or 439–661 of PIPKIγ661 (PIPKIα/γ661) as described in Ling et al. (2002). PIPKIα/γ636–661 was constructed by fusion of the COOH-terminal 26 residues of PIPKIγ661 (PIPKIα/γ636–661) to the COOH terminus of PIPKIα as described by Ling et al. (2002). All mutants were confirmed by DNA sequence analysis. FAK, FAKY397F, and kinase-dead FAK constructs were gifts from Dr. Patricia J. Keely (University of Wisconsin, Madison, Madison, WI). The c-Src construct was provided by Dr. Alan C. Rapraeger (University of Wisconsin, Madison, Madison, WI).

HEK293T cells, NRK cells, and A431 cells were cultured using DME supplemented with 10% FBS. FAK^−/−, FAK^+/+, SFY^−/−, and Src^+/+ cells, provided by Dr. Patricia J. Keely, were maintained in DME supplemented with 10% FBS. NRK cells were plated on glass coverslips and transfected using FuGENE™ 6 following the manufacturer's instructions. 0.5 × 10⁶ HEK293T cells were transfected by calcium phosphate using 2 μg PIPKIγ expression vectors together with 4 μg empty pcDNA3 vector, 2 μg empty pcDNA3 vector plus 2 μg wild-type or mutant FAK, 2 μg empty pcDNA3 vector plus 2 μg c-Src, or 2 μg wild-type or mutant FAK plus 2 μg c-Src. SFY^−/− and Src^+/+ cells were plated on glass coverslips for 16 h and then used for immunofluorescence.

Anti-talin and anti-Flag (M2) antibodies were purchased from Sigma-Aldrich. Monoclonal mouse anti-PY (4G10), anti-Src (EC10), and anti-FAK (4.47) were obtained from Upstate Biotechnology. Anti-HA and anti-c-Myc antibodies were purchased from Covance. HRP-conjugated anti-GST antibody was purchased from Amersham Biosciences. Anti-His and HRP-conjugated anti-T7 antibodies were purchased from Invitrogen. Polyclonal PIPKIγ anti-serum was generated and purified as described previously (Ling et al., 2002). Secondary antibodies were obtained from Jackson ImmunoResearch Laboratories.

pET28 or pET42 constructs were transformed in BL21(DE3) (Novagen). Proteins were expressed and purified using His resin following the manufacturer's instructions (Novagen).

Primary chicken embryo cells were prepared as described previously (Reynolds et al., 1989). Exogenous c-SrcY527F was expressed using the RCAS B replication competent avian retroviral vector as described previously (Schaller et al., 1999). Cells were transfected with LipofectAMINE™ (Life Technologies) and LipofectAMINE™ Plus (Life Technologies) according to the manufacturer's instructions. Approximately 7–9 d after transfection, cells were lysed in RIPA buffer.

Overexpressed c-Src was immunoprecipitated from ∼1 mg chicken embryo cell lysates using EC10 and protein A–Sepharose (Amersham Biosciences) for 1 h at 4°C. Immune complexes were washed twice with RIPA buffer, twice with TBS, and twice with kinase reaction buffer (20 mM Pipes, pH 7.2, 5 mM MnCl₂, and 5 mM MgCl₂). The immune complexes were incubated with 4 μg recombinant substrates for various times at RT in kinase reaction buffer containing 35 μM ATP including 10 μCi γ[³²P]ATP (6,000 Ci mmol⁻¹; PerkinElmer), and were then terminated by adding sample buffer. Samples were resolved by SDS-PAGE, stained with SYPRO^® Orange (Molecular Probes, Inc.), and analyzed after drying for fluorescence and by phosphorimaging, using a PhosphorImager^® (Storm^®; Molecular Dynamics).

4 μg recombinant substrates were tyrosine phosphorylated using c-SrcY527F in vitro as described above. The samples were resolved by SDS-PAGE, transferred to nitrocellulose membranes, and visualized by autoradiography. The ³²P-labeled substrate bands were excised and digested twice with 10 μg trypsin in 200 μl 0.05 M ammonium bicarbonate for 2 h each time at 37°C. The trypsinized membrane bands were oxidized on ice using 75 μl performic acid for 1 h and lyophilized as described by Boyle et al. (1991). Analysis by two-dimensional phosphopeptide mapping on thin-layer cellulose plates was performed using the Hunter thin-layer peptide mapping system (CBS Scientific) with methods described by Boyle et al. (1991). Electrophoretic resolution was performed in pH 1.9 buffer (2.2% formic acid, 7.8% acetic acid) for 30 min at 1,000 V. Then, ascending chromatography was performed for 3–4 h in phosphochromo buffer (37.5% n-butanol, 25% pyridine, and 7.5% acetic acid in water). The plates were dried and visualized by autoradiography at −70°C for 7 d with intensifying screens.

Cells were harvested and lysed in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40, 5.0 mM NaF, 2 mM Na₃VO₄, 4 mM Na₂P₂O₇, 1 mM EDTA, and 1 mM PMSF. 1–2-mg cell lysates were incubated with protein A–Sepharose and 5 μg anti-HA, anti-flag, anti-talin, or 10 μg purified anti-PIPKIγ antibody as indicated at 4°C overnight. The immunocomplexes were separated by SDS-PAGE, and analyzed as indicated.

Immunoblotting data were scanned, and the gray scale of each band was determined by NIH image 1.62 and then plotted using SigmaPlot 4.0 as described before (Ling et al., 2002).

Immunofluorescence was performed as described previously (Ling et al., 2002). In brief, NRK cells 16 h after transfection, SFY^−/−, or Src^+/+ cells were washed with PBS at 37°C, fixed by 4% PFA at RT for 10 min, and permeabilized by 0.2% Triton X-100 in PBS at RT for 10 min. Then, cells were blocked by 3% BSA in PBS at RT for 30 min, incubated with the primary antibody for 1 h at 37°C, washed with 0.1% Triton X-100 in PBS, incubated with fluorescence-labeled secondary antibody at RT for 30 min, and then washed with 0.1% Triton X-100 in PBS. Cells were maintained and examined using a 60× Plan oil immersion lens on a confocal laser-scanning microscope (model MR-1000; Bio-Rad Laboratories) mounted transversely to an inverted microscope (Diaphot 200, Nikon; W.M. Keck Laboratory for Biological Imaging, Madison, WI). Images were processed as described previously (Ling et al., 2002) using Photoshop^® 7.0.

Phosphorylated or nonphosphorylated peptides relevant to PIPKIγ661, named as PN (CDERSWVYSPLHYSAR), pY644 (CDERSWVpYSPLHYSAR), pY649 (CDERSWVYSPLHpYSAR), and pY644/649 (CDERSWVpYSPLHpYSAR), were synthesized and purified by HPLC at >95% purity (Invitrogen). Fluorescein-labeled tyrosine-phosphorylated or nonphosphorylated PIPKIγ661 peptides (labeled at the NH₂ terminus) and peptides corresponding to β1-integrin tail, named as β1InPN (CMNAKWDTGENPIYKSA), β1InpT (CMNAKWDpTGENPIYKSA), β1InpY (CMNAKWDTGENPIpYKSA), and β1InpTpY (CMNAKWDpTGENPIpYKSA), were synthesized and purified by HPLC at >97% purity in the Peptide Synthesis Facility at the University of Wisconsin Biotechnology Center (Madison, WI). Purities, sequences, and correct phosphorylation of all synthesized peptides were confirmed by the Mass Spectrometry/Bioanalytical Facility at the University of Wisconsin (Madison, WI).

Purified GST-talin head proteins were incubated with PIPKIγ661 tails and/or β1-integrin tail, with or without synthesized PIPKIγ661 or β1-integrin peptides, together with glutathione Sepharose 4 Fast Flow (Amersham Biosciences) in 500 μl buffer B (PBS, 0.2% NP-40, and 2 mM DTT) for 4 h at 4°C. The beads were washed with 1 ml buffer B five times and analyzed by Western blot.

20 nM of each fluorescein-labeled PIPKIγ661 peptide was incubated without or with different titration of wild-type, K357Q, or R358Q mutated GST-talin head in PBS containing 0.1% BSA for 30 min at RT. Then the anisotropy value of each sample was determined using a photo counting spectroflurometer (SLM 8000C^®; SLM Instruments) operating with an SLM 8100 hardware/software upgrade (Spectronic Instruments) at 25°C. Anisotropy value of peptide alone was abstracted from each of the relevant values of peptide bound to talin head. Data were analyzed and binding curves were plotted using SigmaPlot 4.0.

We thank Dr. Anna Huttenlocher and Dr. Patricia J. Keely at the University of Wisconsin, Madison (Madison, WI) for excellent discussions and reagents. We thank Martin G. Ensenberger for his technical help. We are grateful to Chateen Carbonara for her comments on the manuscript.

This work was supported by American Heart Association grant 133-EY51 to Kun Ling and by National Institutes of Health grant GM57549 to Richard A. Anderson and grant HL21644 to Deane F. Mosher. The authors declare that they have no competing financial interests.