Inhibition of PYK2-induced actin cytoskeleton reorganization, PYK2 autophosphorylation and focal adhesion targeting by FAK

Focal adhesions are formed at regions where a cell is in close contact with extracellular matrix (ECM), and are essential for the growth of many adherent cells in culture and for cellular attachment in tissue formation during embryogenesis and wound healing (Hynes,1992EF20; Schwartz et al.,1995EF32; Parsons,1996EF26). Focal adhesions link the ECM with the actin cytoskeleton via aggregated integrins (Hynes,1992EF20; Schwartz et al.,1995EF32; Parsons,1996EF26). They serve as discrete plasma membrane signaling domains and mediate integrin signaling, which is important for cell growth, survival, migration and differentiation (Hynes,1992EF20; Schwartz et al.,1995EF32; Parsons,1996EF26). Although many structural and signaling components involved in focal adhesions have been identified, the signaling events that regulate focal adhesion assembly and disassembly, and cytoskeleton organization remain to be elucidated.

Tyrosine phosphorylation is essential in integrin signaling and actin cytoskeleton organization. Tyrosine phosphorylation of focal adhesion proteins including FAK and paxillin is associated with focal adhesion assembly(Burridge et al., 1988EF6;Kornberg et al., 1991EF22; Guan and Shalloway, 1992EF16; Schaller et al., 1992EF30). Inhibition of protein tyrosine kinases disrupts focal adhesion assembly (Burridge and Chrzanowska-Wodnicka, 1996EF5;Chrzanowska-Wodnicka and Burridge,1994EF9; Barry and Critchley,1994EF2). By contrast, over expression of protein tyrosine phosphatases inhibits focal adhesion assembly(Persson et al., 1997EF27). Interestingly, the reversal of the focal adhesion formation process —the disassembly of focal adhesions — is also associated with tyrosine phosphorylation. Focal adhesions are disrupted in cells transformed by the Rous sarcoma virus or by the Fujinami avian sarcoma virus, whose oncogenes encode tyrosine kinases (Tarone et al.,1985EF35; Gavazzi et al.,1989EF13). These cells form podosomes, dot-shaped spots where cells adhere transiently to their matrix,which contain actin and virtually all of the focal adhesion proteins (Tarone et al., 1985EF35; Gavazzi et al.,1989EF13). Concomitantly, tyrosine phosphorylation of focal adhesion components including integrin, talin,paxillin, Src and FAK is increased (Hirst et al.,1986EF19; Smart et al.,1981EF34; Schaller et al.,1992EF30). These observations indicate that regulated tyrosine phosphorylation of adhesion components seems to be important in both assembly and disassembly of focal adhesions.

FAK is a major protein tyrosine kinase localized in focal adhesions(Schaller et al., 1992;Richardson and Parsons, 1996;Ilic et al., 1995; Frisch et al., 1996). It contains a central catalytic domain and large N- and C-terminal non-catalytic regions that are devoid of SH2 or SH3 domains (Schaller et al.,1992; Hanks et al.,1992). Within the C-terminal domain is the FAT domain that is both necessary and sufficient for the targeting of FAK into focal adhesions (Hildebrand et al.,1993). Several lines of evidence indicate that FAK plays an important role in regulating cell migration and spreading. First, fibroblasts derived from FAK null mice(fak^-/-) exhibit reduced rates of cell migration and altered cellular morphology (Ilic et al.,1995). Second, overexpression of FAK in Chinese hamster ovary cells stimulates cell migration on fibronectin(Carey, 1996; Carey, 1998). Third, overexpression of FAK-related non-kinase(FRNK; a splice variant that contains only FAK's C-terminal domain and functions as a dominant negative protein) inhibits cell spreading on fibronectin (Richardson and Parsons,1996; Gilmore and Romer,1996).

PYK2 is a second member of the FAK family (Lev et al.,1995EF23; Sasaki et al.,1995EF29; Avraham et al.,1995EF1; Li et al.,1996EF24); it is also known as cellular adhesion kinase β (CAKβ), related adhesion-focal tyrosine kinase (RAFTK) and calcium-dependent tyrosine kinase (CADTK). PYK2 catalytic activity is regulated by tyrosine phosphorylation in a manner similar to FAK. Phosphorylation at tyrosine 402 (Tyr402) creates a binding site for the SH2 domains of Src family kinases, resulting in Src activation, which increases the catalytic activity of PYK2 (Tokiwa et al.,1996EF36). However, PYK2 activation seems to be different from that of FAK. PYK2 is activated by various stimuli, including elevation of intracellular calcium levels,activation of protein kinase C and exposure to stress factors (e.g. UV light,tumor necrosis factor α) (Lev et al.,1995EF23; Tokiwa et al.,1996EF36; Li et al.,1996EF24), whereas FAK is mainly activated by integrin engagement (Guan and Shalloway,1992EF16; Schaller et al.,1992EF30). We have shown that PYK2 induces apoptosis when expressed in fibroblasts, whereas FAK seems to be important for cell survival (Xiong and Parsons,1997EF37; Frisch et al.,1996EF12). The mechanisms underlying the different effects of PYK2 and FAK, and the potential role of PYK2 in regulating cytoskeleton organization remain unclear.

In this paper, we show that PYK2 was expressed in fibroblastic cell lines and co-localized with FAK at focal adhesions. Overexpression of PYK2, but not FAK, in fibroblastic cells led to reorganization of actin-associated cytoskeleton structures and cell rounding. PYK2-mediated actin cytoskeleton reorganization required the N terminus, kinase activity and focal adhesion targeting. Interestingly, PYK2's effects could be suppressed by overexpression of FAK, probably via inhibition of PYK2 kinase activity and focal adhesion targeting. Taking together, our results suggest that PYK2 and FAK coordinately regulate actin-cytoskeleton organization and cellular morphology, which are important for cell migration and survival.

Monoclonal antibodies were purchased from Transduction Laboratories(anti-phosphotyrosine, anti-paxillin and anti-PYK2), Sigma (anti-vinculin) and Santa Cruz Biotechnology (anti-Myc). Polyclonal antibodies were purchased from UBI (anti-PYK2), Biosources International (anti-PY402, phosphotyrosine 402). The rabbit polyclonal antiserum against PYK2 was raised using a glutathione-S-transferase (GST) fusion protein containing the C-terminal 400 amino acids of rat PYK2 (amino acids 587-988). The specificity of this antiserum has been described previously (Xiong and Parsons,1997).

cDNAs of PYK2, FAK and FRNK were subcloned into expression vectors downstream of a Myc epitope tag (MEQKLISEEDL) under the control of the cytomegalovirus (CMV) promoter (pCMV-Myc) as described previously (Xiong and Parsons, 1997EF37). Chimeric PYK2/FAK-1 that contained PYK2 N terminus (amino acids 2-385) and FAK's kinase and C-terminal domains (amino acids 380-1052) has been described previously(Xiong and Parsons, 1997EF37). All mutants were generated using the polymerase chain reaction (PCR). The authenticity of constructs was verified by DNA sequencing. Green fluorescent protein (GFP) vector was purchased from Clontech.

Swiss 3T3 and HEK 293 cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) containing 10% fetal calf serum, 100 μg ml^-1penicillin G and 100 μg ml^-1 streptomycin (GIBCO). FAK null(fak^-/-) cells, kindly provided by D. Ilic, were maintained in DMEM containing 10% fetal calf serum, 100 μg ml^-1penicillin G, 100 μg ml^-1 streptomycin (GIBCO), 4.5 mg ml^-1 D-glucose, 584 μg ml^-1 glutamine and 49 μMβ-mecaptoethonal (Ilic et al.,1995). For microinjection,Swiss 3T3 cells were plated at a density of 10³ cells per 22 mm cover slip 12 hours before injection. Plasmid DNAs (50 ng μl^-1)were microinjected into nuclei using the Eppendorf micromanipulation system. Two hours after injection, cells were fixed and immunostained with specific antibodies. For transfection, cells were plated at a density of 10⁶cells per 100 mm culture dish, allowed to grow for 12 hours and transfected using the calcium phosphate precipitation method for HEK 293 cells and the SuperFect method (Qiagen) for fak^-/- cells. Cells were lysed with modified RIPA buffer (50 mM Tris-HCL, pH 7.4, 150 mM sodium chloride, 1% NP40, 0.25% sodium deoxycholate and proteinase inhibitors) 24 hours after transfection. Cell lysates were subjected to analyses by SDS-PAGE and western blotting.

Focal adhesions were characterized by immunostaining using antibodies against focal adhesion proteins as previously described (Xiong and Parsons,1997). Briefly, cells were plated on fibronectin-coated coverslips (40 μg ml^-1) for 2-4 hours, fixed with 4% paraformaldehyde, permeabilized with 1% Triton X-100 and incubated with a primary antibody at room temperature for 1 hour and subsequently incubated with a fluorescein-conjugated anti-rabbit or anti-mouse secondary antibody (1:300 dilution) for 1 hour at room temperature. Immunoreactivity was visualized using a Nikon fluorescence microscope. Cell morphology was determined by examining the cell length : cell width ratio of fibroblasts. A ratio >2 was considered to be a normal elongated cell shape. Abnormal or rounded cells usually had a ratio <2. More than 100 microinjected cells were examined for each protein. All the results are presented as percentages of the mean + s.d. Statistical analyses were carried out using Mann-Whitney u test.

Immunoprecipitation of individual proteins was carried out as previously described (Xiong and Parsons,1997EF37). Briefly, 1 mg of cell lysates was incubated with anti-PYK2 antibodies (1-10 μg) in 1 ml of RIPA buffer at 4°C for 1 hour with agitation. After the addition of protein-A/agarose beads, the reaction was incubated at 4°C for another hour. Immune complexes were collected by centrifugation and washed. Bound proteins were resolved by SDS-PAGE and subjected to western blotting using the indicated antibodies.

Expression of endogenous PYK2 in several fibroblast cell lines was examined by western blot analysis using a PYK2 antibody (polyclonal, from UBI). This antibody was PYK2 specific and did not cross-react with overexpressed FAK in HEK 293 cells (Fig. 1A). Endogenous FAK protein was high in 10T1/2 and Swiss 3T3 cells, low in 3Y1 and osteoclasts, and undetectable in fak^-/- fibroblasts. Despite this, PYK2 protein was abundantly expressed in osteoclasts in agreement with a previous report (Duong et al.,1998)(Fig. 1B). Remarkably, PYK2 protein was also detected in 3Y1, 10T1/2 and Swiss3T3 fibroblasts(Fig. 1B). To characterize the subcellular locations of PYK2, 10T1/2 fibroblasts were fixed and immunostained with the polyclonal anti-PYK2 antibody and a monoclonal antibody against FAK or paxillin, a marker of focal adhesions. As shown in Fig. 1C, PYK2 co-localized with paxillin or FAK at the focal adhesions in 10T1/2 cells. PYK2 focal adhesion targeting seemed to be transient and dependent on fibronectin stimulation. There was no apparent focal adhesion localization of PYK2 in 10T1/2 cells within the first hour after being plated on fibronectin-coated coverslips(Fig. 1D). However, PYK2 focal adhesion localization became obvious 2-4 hours after plating on fibronectin-coated coverslips (Fig. 1D). In addition to focal adhesions, PYK2 was also distributed in perinuclear regions that agree with the previous reports(Fig. 1C,D) (Schaller and Sasaki, 1997; Zheng et al.,1997; Sieg et al., 1998). To exclude possible cross-reaction of the polyclonal anti-PYK2 antibody with FAK protein at focal adhesions in 10T1/2 cells, we examined PYK2 subcellular localization in fak^-/- fibroblasts, in which the fak gene is knocked out and PYK2 expression is upregulated(Fig. 1B) (Sieg et al.,1998). Consistent with the results obtained in 10T1/2 cells, PYK2 co-localized with paxillin at focal adhesions in fak^-/- fibroblastic cells(Fig. 1C). These results indicated that PYK2 was expressed in fibroblasts and localized at focal adhesions.

To study the role of PYK2 and FAK in regulating cytoskeleton organization,we microinjected PYK2 cDNA into Swiss 3T3 cells, a typical mouse fibroblastic cell line with well-characterized cytoskeleton structures and examined focal adhesions and cell morphology in microinjected cells. Swiss 3T3 cells expressing FAK showed normal focal adhesions(Fig. 2D-F). By contrast, cells expressing PYK2 exhibited altered cytoskeleton structures with either no focal adhesions or abnormal `podosome-like' focal adhesions with peculiar ring- or crescent-like adhesion sites at the periphery of cells(Fig. 2A). The `podosome-like'structures in PYK2 expressing cells could be labeled with antibodies against various focal adhesion elements including paxillin or vinculin(Fig. 2B). We next examined whether the organization of stress fibers and microtubules were altered by PYK2. In PYK2-expressing cells, stress fibers were disrupted in the central region and enriched at the peripheral region of cells(Fig. 2G-I). However, cells expressing FAK exhibited normal stress fibers(Fig. 2J-L). Microtubules labeled by anti-β-tubulin antibody seemed normal in PYK2-expressing cells (data not shown).

These data demonstrate that actin-associated cytoskeleton structures, but not microtubules, were reorganized by PYK2 in Swiss 3T3 cells. In addition to altered focal adhesions and reorganized stress fibers, Swiss 3T3 cells expressing PYK2 exhibited rounded morphology (ratio of length to width is<2) with condensed cytoplasm (Fig. 2A-C,G-I), which was consistent with our previous report in Rat-1 cells (Xiong and Parsons,1997), whereas cells expressing FAK showed the normal elongated cell shape (ratio of length over width is >2) (Figs 2D-F,J-K). Similar results were also observed in 3Y1 and 10T1/2 fibroblasts, and MDCK cells expressing PYK2(Fig. 3). Moreover,PYK2-induced cytoskeleton reorganization seemed to correlate with the level of PYK2 expression. Cells overexpressing PYK2 at fivefold or more that of endogenous PYK2, estimated by quantitative fluorescence analysis, showed abnormal focal adhesions and morphology. These results indicated that overexpression of PYK2, but not FAK, in Swiss 3T3 cells resulted in reorganization of actin cytoskeleton and changes of cell morphology.

To understand mechanisms of PYK2-mediated cytoskeleton reorganization, we attempted to map the domains required for this event. PYK2 contains a central catalytic domain (amino acids 419-679), flanked by non-catalytic N-terminal(amino acids 1-418) and C-terminal (amino acids 680-1009) domains(Fig. 4b). The C-terminal region has two proline-rich sequences and the FAT domain(Fig. 4b). We generated a series of PYK2 deletion mutants and studied their cytoskeleton-reorganizing activity and focal adhesion targeting in Swiss 3T3 cells(Fig. 4). Whereas Swiss 3T3 cells expressing PYK2 demonstrated abnormal focal adhesions, Swiss 3T3 cells expressing the FAT-domain deletion mutant (PYK2Δ936-1009),which failed to target to focal adhesions, exhibited normal focal adhesion structures (Fig. 4a). This result suggests that the FAT domain of PYK2, or PYK2 focal adhesion targeting,was required for PYK2-mediated reorganization of focal adhesions. However,PYK2 focal adhesion targeting was not sufficient to induce reorganization of focal adhesions. Swiss 3T3 cells producing PYK2 FAT domain or C-terminal regions, which localized to focal adhesions, failed to induce reorganization of focal adhesions (Fig. 4a). This result suggests that, in addition to the FAT domain, PYK2 N terminus and kinase domain were also required for focal adhesion reorganization. The requirement for PYK2 N terminus by this event was further demonstrated by the finding that cells expressing PYK2 N-terminal deletion mutants(PYK2Δ1-88 or PYK2Δ1-416) showed normal focal adhesions(Fig. 4a). Interestingly,although the C-terminal domain of PYK2 can localize to focal adhesions, the partial N-terminal deletion mutant failed to localize to the focal adhesions(Fig. 4a), even though they contain the intact FAT domain, implying that the PYK2 FAT-domain-mediated focal adhesion targeting might be regulated by PYK2 kinase domain and /or N terminus. Taken together, these results indicated that both PYK2 N terminus and FAT domain were required for PYK2-mediated reorganization of focal adhesions.

To examine whether the catalytic activity of PYK2 was required for PYK2-induced cytoskeleton reorganization, a catalytically inactive mutant(PYK2-KD) was generated that contained a lysine (K) to alanine (A) mutation in the ATP binding site (Fig. 5). Being unable to bind to ATP, this mutant showed little if any catalytic activity when expressed in mammalian cells by in vitro kinase assays or anti-phosphotyrosine blotting (Fig. 5B). In addition, tyrosine phosphorylation of several proteins was increased in cells expressing wild-type PYK2 but not PYK2-KD(Fig. 5C). One of these proteins was identified as p130^Cas, a constituent in the focal adhesions that is known to be a potential substrate of PYK2(Fig. 5C). Swiss 3T3 cells microinjected with the plasmid DNA encoding PYK2-KD were analyzed in parallel with the cells expressing wild-type PYK2. As shown in Fig. 5A, unlike those expressing PYK2-WT, which had abnormal or no focal adhesions, a fraction of PYK2-KD-expressing cells (∼33%) exhibited normal focal adhesions. In addition, a fraction of PYK2-KD-expressing cells(∼45%) seemed elongated in shape, with a ratio of length to width of>2. Fig. 5D summarizes results from three independent experiments. Significantly more cells showed normal focal adhesions or cell shape (a ratio of length to width of >2) in PYK2-KD-expressing cells than in PYK2-WT-expressing cells. These results indicated that the catalytic activity of PYK2 contributed to,but was not absolutely required for, the induction of cytoskeleton reorganization. In addition, the PYK2-Y402F mutant, which contains a tyrosine(Y402F) mutation, also showed decreased cytoskeleton reorganization activity(Fig. 5D), suggesting that autophosphorylation of PYK2 might also contribute to this event.

Considering that FAK is important for regulating focal adhesion turnover(Ilic et al., 1995),cytoskeleton reorganization by PYK2 but not FAK led us to determine whether FAK could regulate PYK2's effect. Swiss 3T3 cells microinjected with cDNAs encoding PYK2 and FAK or GFP were examined for focal adhesions and cell morphology. Unlike cells expressing PYK2 alone, cells expressing both FAK and PYK2 exhibited normal focal adhesions and cell morphology, suggesting that FAK could inhibit PYK2-induced cytoskeleton reorganization and cell rounding. This inhibitory effect was specific, in that co-injection of GFP failed to rescue the phenotype(Fig. 6A). Moreover, the rescue of the PYK2 phenotypes by FAK seemed to be dose dependent. When the ratio of injected cDNAs of FAK to PYK2 was increased, FAK inhibitory activity was enhanced from ∼20% (for 1:0.5 ratio) to 70% (1:2 ratio)(Fig. 6B). Finally, FAK's inhibitory effect on PYK2 did not require FAK autophosphorylation, because the mutant of the autophosphorylation site (Tyr397) (FAK-Y397F) could suppress PYK2-induced cytoskeleton reorganization(Fig. 6B). Interestingly,co-expression of FRNK seemed to be able to rescue PYK2-induced alterations only in focal adhesions, not in cell morphology(Fig. 6), suggesting that different mechanisms might be involved in PYK2-induced alternations in focal adhesions and cell morphology. These results suggest that PYK2 and FAK,although highly homologous in structure, might mediate different (and possibly opposing) roles in cytoskeleton organization.

PYK2's catalytic activity was required for cytoskeletal reorganization and FAK could inhibit PYK2's effect. We thus hypothesized that FAK might regulate PYK2's catalytic activity. Phosphorylation of PYK2 at Tyr402 is crucial for PYK2 kinase activity in vivo (Lev et al.,1995EF23; Dikic et al.,1996EF10). We tested whether FAK affected phosphorylation of PYK2 Tyr402. As shown in Fig. 7A, Tyr-402 phosphorylation detected by the anti-PY402 antibody was significantly reduced when PYK2 was co-expressed with FAK in HEK 293 cells. Interestingly, FRNK, the splice variant containing the C-terminal domain of FAK, did not seem to have an effect on PYK2 autophosphorylation (Fig. 7A), suggesting that the FAK N-terminal region and/or kinase domains are required for the inhibitory effect. We next examined whether FAK N-terminal region was required for the inhibition of PYK2 autophosphorylation. Whereas PYK2 autophosphorylation is significantly decreased in cells co-expressing FAK and FAKΔC20 (a C-terminal FAT domain deletion mutant),FAKΔN-term (a mutant with a deletion of FAK's N-terminal 1-412 amino acids) was unable to mediate the inhibitory effect, suggesting again that the N-terminal domain (amino acids 1-412) of FAK was required(Fig. 7A). To further confirm this finding, we generated a chimeric construct, PYK2/FAK-1, in which the N-terminal domain of FAK (amino acids 2-380) was replaced with the N-terminal domain of PYK2 (amino acids 2-385). Expression of this chimeric protein did not seem to affect PYK2 autophosphorylation(Fig. 7A). These data suggested to us that the FAK N-terminal domain (2-380) was essential for the inhibition of PYK2 autophosphorylation in HEK 293 cells.

PYK2 activity can be regulated by tyrosine phosphorylation. Phosphorylation at Tyr402 of PYK2, like Tyr397 in FAK, creates a binding site for the SH2 domain of Src. Src binding to PY402 leads to activation of Src, which in turn phosphorylates and activates PYK2 (Dikic et al.,1996EF10). Overexpression of FAK,which has similar binding property, might absorb Src or a Src-like kinase that is essential for PYK2 activation and thus decreases PYK2 autophosphorylation. To test this hypothesis, we determined whether FAK tyrosine phosphorylation,including PY397, was required for the inhibition of PYK2 autophosphorylation. As shown in Fig. 7B, mutations of key tyrosine residues in FAK did not seem to alter its inhibitory effect on PYK2. These results demonstrated that tyrosine phosphorylation of FAK is not required for FAK-mediated inhibition of PYK2 autophosphorylation and ruled out the possibility that inhibition of PYK2 by FAK resulted from a decrease in Src that effectively interacts with PYK2.

PYK2 was localized with FAK to focal adhesions in fibroblasts(Fig. 1) and PYK2 focal adhesion targeting was required for PYK2-mediated cytoskeletal reorganization(Fig. 4). We thus hypothesized that FAK might rescue PYK2's phenotypes by regulating PYK2 focal adhesion targeting. To test this hypothesis, we examined the PYK2 focal adhesion localization in fak^-/- fibroblasts transiently transfected with FAK or FRNK. Whereas PYK2 localizes with paxillin to focal adhesions in fak^-/- fibroblasts(Fig. 8aA-aC,b), PYK2 focal adhesion targeting and the number of focal adhesions seemed to be decreased in cells expressing FAK compared with cells that are not expressing FAK(Fig. 8aD-aF,b). Similar results were also observed in fak^-/- fibroblasts expressing FRNK (Fig. 8aG-aI,b) or FAK-Y397F (data not shown), suggesting that FAK C-terminal domain is sufficient for the reduction of PYK2 focal adhesion targeting. These results suggested that overexpressed FAK might displace endogenous PYK2 from focal adhesions, probably via the C-terminal FAT domain.

In this paper, we have shown that PYK2 localized with FAK to focal adhesions in fibroblasts. The expression of PYK2, but not of FAK, resulted in reorganization of actin-associated cytoskeleton structures and cell rounding in Swiss 3T3 fibroblasts in a manner dependent on PYK2 catalytic activity and focal adhesion targeting. Remarkably,co-expression of FAK inhibited PYK2-induced cytoskeletal phenotypes. PYK2 autophosphorylation and focal adhesion targeting were also diminished by FAK, which might be mechanisms for the rescue of PYK2-induced actin cytoskeleton reorganization by FAK. These results suggested that PYK2 and FAK play distinct roles in regulating actin-based cytoskeleton organization and that a balance between PYK2 and FAK tyrosine kinases was important for the determination of focal adhesion turnover and cellular morphology.

Although PYK2 has been found to be located at cytoplasm or perinuclear regions (Schaller and Sasaki,1997; Zheng et al., 1997; Sieg et al., 1998), its presence at focal adhesions has not been documented. Here, we report that, in addition to cytoplasmic and perinuclear regions, PYK2 localizes with FAK to focal adhesions in fibroblastic cells. The anti-PYK2 antibody used in our studies does not cross-react with FAK. Western blot analysis using this antibody recognizes a single band that is different from that of FAK and does not cross-react with overexpressed FAK in HEK293 cells. In immunohistochemical studies, this antibody could label focal adhesion structures in fak^-/- cells, in which FAK was not expressed. Moreover, in fak^-/- cells overexpressing FAK, it did not stain overexpressed FAK. These results convincingly demonstrated that the anti-PYK2 antibody is specific and does not cross-react with FAK. The reasons why PYK2 has not been reported in focal adhesions are complex. The antibodies available for previous studies might not be as sensitive for immunostaining of endogenous PYK2. This might explain the failure to observe focal adhesion localization of endogenous PYK2 using the polyclonal anti-PYK2 antibody (amino acids 587-988) (Xiong and Parsons,1997). In studies in which PYK2 was overexpressed (Sasaki et al.,1995; Schaller and Sasaki,1997; Zheng et al.,1998), it could be difficult to visualize its subcellular localization. We found that only 10-20% of 10T1/2 cells overexpressing PYK2 exhibited focal adhesion targeting, in agreement with a previous report (Schaller and Sasaki,1997). Furthermore, we found that PYK2 focal adhesion targeting could be regulated by fibronectin stimulation or expression of FAK or FRNK. PYK2 focal adhesion localization became obvious when cells were plated on fibronectin-coated coverslips for 2-4 hours. In cells with high level of FAK or FRNK, PYK2 might not be localized at focal adhesions.

Our results suggest that PYK2 focal adhesion targeting has the following properties. First, PYK2 FAT domain was required and sufficient for its focal adhesion targeting. Second, PYK2 focal adhesion targeting seemed to be negatively regulated by its N terminus and catalytic activity. Third, PYK2 focal adhesion targeting was inhibited by FAK. The mechanism by which FAK or FRNK regulated PYK2 subcellular localization remains unclear. It is intriguing to speculate that the subcellular location of PYK2 might be regulated by FAK or FRNK, perhaps via competition for a common binding protein(s) whose association with PYK2 is necessary for PYK2 focal adhesion targeting, or by its catalytic activity, perhaps by modification of targeting sequences like the C-terminal FAT domain.

The finding of PYK2 at focal adhesions suggests that PYK2 might play an important role in focal adhesion signaling and/or regulate the formation and turnover of this subcellular structure. Indeed, we found that endogenous PYK2 autophosphorylation (assayed by PY402 antibody) is increased when 10T1/2 cells are plated on fibronectin-coated plates. The increase in PY402 is time dependent, and peaked around 1 hour grown on fibronectin-coated dishes (data not shown). Moreover, overexpression of PYK2 in fibroblasts changed cell shape and focal adhesion structures. PYK2-mediated cytoskeletal reorganization required its catalytic activity and focal adhesion targeting. These results support a role for PYK2 as a regulator of focal adhesion assembly. Altered focal adhesions mediated by PYK2 are characterized by podosome-like morphological structures. Podosome-like structures have been identified in v-Src-transformed cells (Tarone et al.,1985; Gavazzi et al.,1989) and non-transformed cells including peripheral mononuclear lymphocytes, macrophages and osteoclasts, in which PYK2 is highly expressed (Avraham et al.,1995; Duong et al.,1998). The abundant expression of PYK2 in these cells, the localization of PYK2 to podosomes (Duong et al., 1998) and our observation that expression of PYK2 results in cytoskeleton reorganization in Swiss 3T3 fibroblasts suggest that PYK2 might play an important regulatory role in the formation of podosome-like structures in vivo (Gavazzi et al.,1989; Boyles and Bainton,1979; Boyles and Bainton,1981).

Interestingly, FAK, a tyrosine kinase related to PYK2, could rescue PYK2-induced actin cytoskeleton reorganization, suggesting that PYK2 and FAK do not only mediate different and possibly opposing roles in regulating focal adhesion organization and cell morphology. What are the mechanisms for FAK-mediated suppression of PYK2's effects? Because FAK inhibits PYK2 autophosphorylation and focal adhesion targeting, which are required for PYK2-mediated cytoskeleton reorganization, we propose that FAK-mediated suppression of PYK2-induced cytoskeleton reorganization might be due to its inhibition of PYK2 autophosphorylation and/or focal adhesion localization. FAK inhibition of PYK2 autophosphorylation is consistent with previous reports that FAK and the FAK autophosphorylation mutant FAK-Y397F inhibit integrin-mediated PYK2 activation in fak^-/- cells (Owen et al., 1999). The mechanism of FAK-mediated inhibition of PYK2 activity remained unclear. FAK might inhibit PYK2 tyrosine phosphorylation by displacing PYK2 focal adhesion targeting. This might be a mechanism underlying FAK inhibition of integrin-mediated PYK2 tyrosine phosphorylation (Owen et al.,1999). However, FAK inhibition of PYK2 autophosphorylation in HEK 293 cells expressing PYK2 seemed to be caused by a different mechanism. The FAT domain deletion mutant,FAKΔC20 (which failed to localize to focal adhesions), but not FRNK(which decreases PYK2 focal adhesion targeting) could inhibit PYK2 autophosphorylation in HEK 293 cells. These results indicate that FAK inhibition of PYK2 autophosphorylation does not correlate with FAK focal adhesion targeting. Furthermore, our results suggest that FAK N-terminal region, which contains a Band 4.1 domain (Giranlt, J. A. et al.,1998), is essential for the inhibition of overexpression-induced PYK2 autophosphorylation.

PYK2 and FAK play important roles in regulating actin-associated cytoskeletal organization, and a balance between PYK2 and FAK tyrosine kinases is important for the determination of cytoskeletal organization and cell morphology. These data, together with our previous results (Xiong and Parsons,1997EF37), suggest that FAK and PYK2 could play an opposing roles in the overall regulation of the actin cytoskeleton organization and cell morphology, death and survival, which is reminiscent of the Yin-Yang principle.

We are grateful to J. T. Parsons for his support to the initiation of this project. We also thank D. Ilic for fak^-/- cells and M. Macklem for technical assistance on early microinjection experiments. This study was supported in part by a start-up fund from the School of Medicine,University of Alabama at Birmingham, and grants from American Heart Association Southeastern Affiliate grant in aid 051566B (to W.-C.X.) and NINDS(NS40480 and NS34062) (to L.M.).