Inhibition of the Catalytic Activity of Cell Adhesion Kinase b by Protein-tyrosine Phosphatase-PEST-mediated Dephosphorylation*

Protein-tyrosine phosphatase (PTP)-PEST is a cytoplasmic tyrosine phosphatase that can bind and dephosphorylate the focal adhesion-associated proteins p130 CAS and paxillin. Focal adhesion kinase (FAK) and cell adhesion kinase b (CAK b )/PYK2/CADTK/RAFTK are protein-tyrosine kinases that can colocalize with, bind to, and induce tyrosine phosphorylation of p130 CAS and paxillin. Thus, we considered the possibility that these kinases might be substrates for PTP-PEST. Using a combination of substrate-trapping assays and overexpression of PTP-PEST in mammalian cells, CAK b was found to be a substrate for PTP-PEST. Both the major autophosphorylation site of CAK b (Tyr 402 ) and activation loop tyrosine residues, Tyr 579 and Tyr 580 , were targeted for dephosphorylation by PTP-PEST. Dephosphorylation of CAK b by PTP-PEST dramatically inhibited CAK b kinase activity. In contrast, FAK was a poor substrate for PTP-PEST, and treatment with PTP-PEST had no effect on FAK kinase activity. Tyrosine phosphorylation of paxillin, which is greatly enhanced by CAK b overexpression, was

Tyrosine phosphorylation is a post-translational modification that is essential for many signal transduction cascades, including growth factor receptor and focal adhesion signaling. Cellular phosphotyrosyl levels are kept in homeostasis by the competing action of two classes of enzymes, protein-tyrosine kinases (PTKs) 1 and protein-tyrosine phosphatases (PTPs). There is a vast literature describing the role of PTKs in the regulation of signal transduction. As more PTPs are being discovered, it has become apparent that these proteins, like PTKs, play an integral role in regulating cellular signaling pathways.
PTP-PEST is one of a number of cytoplasmic phosphatases characterized by the presence of several PEST-rich regions (1)(2)(3)(4), motifs that have been found in many proteins that are rapidly degraded in the cell (5). PTP-PEST, however, appears to be quite stable (6). Structurally, it is composed of an Nterminal catalytic domain and a C-terminal tail containing five proline-rich regions, several of which constitute consensus Src homology 3 binding sites. Indeed, PTP-PEST binds the adaptor protein Grb2 and the tyrosine kinase Csk in an Src homology 3-dependent manner (7,8). PTP-PEST also binds the adaptor protein Shc via a phosphotyrosine binding domain-mediated interaction. However, this interaction occurs in a non-phosphotyrosine-dependent manner through an NPLH sequence in the C-terminal tail of PTP-PEST (9). Although PTP-PEST has been shown to associate with these signaling molecules, its role in regulation of these signaling pathways remains to be fully elucidated. Grb2 binding was reported to bring PTP-PEST into association with activated epidermal growth factor receptor; however, the epidermal growth factor receptor does not appear to be a substrate (7). An alternative hypothesis is that PTP-PEST may instead regulate epidermal growth factor receptor signaling by dephosphorylating receptor substrates. Csk was recently shown to cooperate with the tyrosine phosphatase PEP, a structural homologue of PTP-PEST, in inhibiting T-cell antigen receptor signaling (10). PEP could dephosphorylate and inactivate PTKs responsible for T-cell activation but only when associated with Csk. A role for PTP-PEST in Csk signaling events has yet to be established. The Shc proteins are adaptor molecules most closely linked with activation of the Ras pathway (11,12). Interestingly, Shc is phosphorylated on tyrosine itself (11). PTP-PEST, however, has not been implicated in regulation of the Ras signaling pathway, and Shc has not been identified as a substrate. In addition to binding these signaling molecules, the C-terminal domain of PTP-PEST has docking sites for two known substrates. p130 CAS binds PTP-PEST in an Src homology 3-dependent manner (13), and paxillin binds through its two C-terminal LIM domains to two nonoverlapping stretches of a 52-amino acid sequence in PTP-PEST (14,15). Both p130 CAS and paxillin are substrates for PTP-PEST, and their association with the C-terminal domain of PTP-PEST is required for dephosphorylation of these substrates in vivo (13,15,16).
Paxillin and p130 CAS localize to focal adhesions and undergo rapid tyrosine phosphorylation upon cell adhesion (17). Paxillin and p130 CAS also become tyrosine-phosphorylated in response to various physiological stimulants including bombesin, platelet-derived growth factor, nerve growth factor, and angiotensin II (18,19). Both proteins act as adaptor molecules in integrin signaling. Paxillin binds to several focal adhesion-associated proteins including vinculin, FAK, and c-Src (19). Integrin-mediated tyrosine phosphorylation of paxillin creates binding sites for the SH2 domains of Crk and Csk (20,19). Tyrosine phosphorylation of paxillin has been implicated in the control of biological events such as cell spreading (21), cell adhesion (22), and cell motility (23,24). Tyrosine phosphorylation of p130 CAS also creates docking sites for the SH2 domain of Crk (25). The p130 CAS -Crk complex has been proposed to regulate cell migration (26), and p130 CAS is thought to act as a mediator for FAK-promoted cell migration (27). PTP-PEST has been implicated in the regulation of cell motility (28,29) and may do so by targeting paxillin and/or p130 CAS for dephosphorylation.
Paxillin and p130 CAS are potential substrates for several candidate tyrosine kinases including FAK and CAK␤. Paxillin and p130 CAS can directly bind to FAK and CAK␤ and become tyrosine-phosphorylated in response to stimuli that activate these PTKs (17). FAK and CAK␤ are structurally related cytoplasmic tyrosine kinases composed of a central catalytic domain flanked by N-and C-terminal noncatalytic domains (30). Although structurally similar and able to associate with many of the same proteins, it is unclear whether these kinases share similar functions. While FAK has been most strongly implicated as a major component of integrin-mediated signaling pathways (17), there have been conflicting reports as to whether CAK␤ is regulated by integrin mediated cell adhesion (31)(32)(33). Unlike FAK, CAK␤ is regulated by stimuli that induce changes in intracellular Ca 2ϩ (30). Like other PTKs, tyrosine phosphorylation of FAK and CAK␤ plays an important role in regulating signaling. FAK and CAK␤ share four conserved sites of tyrosine phosphorylation. They are the Src-family SH2 domain binding site at Tyr 397 in FAK and Tyr 402 in CAK␤, the Grb2-SH2 domain binding site at Tyr 925 in FAK and Tyr 881 in CAK␤, and two regulatory sites in the activation loop of the kinase domains: Tyr 576/577 in FAK and Tyr 579/580 in CAK␤ (30,34).
FAK and CAK␤ are particularly good candidate PTP-PEST substrates, since these kinases both bind paxillin and p130 CAS (35), two previously identified substrates (15,16). Two lines of evidence suggest that FAK may be a substrate. First, PTP-PEST activity was coimmunoprecipitated with FAK (36), although this association is indirect and appears to be mediated by paxillin. Second, PTP-PEST Ϫ/Ϫ cells exhibited not only increased paxillin and p130 CAS phosphotyrosine but also increased tyrosine phosphorylation of FAK (29). These results suggest that FAK may be a substrate of PTP-PEST and that PTP-PEST may regulate signaling both at the level of the kinase and the substrate. On the other hand, PTP-PEST may not play a major role in FAK dephosphorylation, since the two proteins fail to colocalize intracellularly. 2 Alternatively, PTP-PEST may regulate CAK␤ signaling. In this report, we explore the hypothesis that FAK and CAK␤ may be PTP-PEST substrates. We identify CAK␤ as a specific substrate of PTP-PEST using a substrate-trapping approach and by demonstrating dephosphorylation of CAK␤ by PTP-PEST in vivo. In contrast, PTP-PEST bound very weakly to FAK in a substrate-trapping assay and was a weaker substrate for dephosphorylation than CAK␤. We determined that PTP-PEST could target the major autophosphorylation site of CAK␤ as well as one or both tyrosines in the activation loop of the catalytic domain. We demonstrate that PTP-PEST is able to inhibit CAK␤ kinase activity both in vitro and in vivo. Finally, endogenous PTP-PEST and endogenous CAK␤ were shown to reside in similar cellular compartments within GN4 and A7r5 cells. These results suggest that PTP-PEST can inhibit CAK␤ signaling by impairing the activity of the kinase and by dephosphorylating downstream substrates.

EXPERIMENTAL PROCEDURES
Cells-HEK 293 cells were maintained in Dulbecco's modified Eagle's medium/Ham's F-12 medium supplemented with 10% fetal bovine serum and A7r5 vascular smooth muscle cells in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. GN4 cells were obtained from Dr. H. Shelton Earp and were maintained in Richter's minimal essential medium supplemented with 10% fetal bovine serum. Cells were transfected using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's recommended protocol.
Molecular Biology-The epitope-tagged (KT3) wild type, dl367-400, dl297-493, and P337A PTP-PEST constructs have been described previously (15,36). To construct pcDNA 3.1-FAK, the FAK cDNA was excised from pBS-FAK (37) using XbaI and SalI, which cut in the multiple cloning site flanking the FAK insert. The FAK-containing fragment was inserted between the NheI and XhoI sites of pcDNA3.1. Wild type and CAK␤ YF mutants were kindly provided by Drs. Xiong Li and H. Shelton Earp (University of North Carolina) and have been described previously (34). For expression as a glutathione S-transferase (GST) fusion protein, a fragment of the PTP-PEST cDNA encoding the catalytic domain (amino acids 1-301) was amplified by polymerase chain reaction and subcloned into pGEX2TK (Amersham Pharmacia Biotech) in frame with the GST coding sequence. The C231S PTP-PEST substrate-trapping mutant (GST/PTP-PEST C231S ) was generously provided by Dr. André Veillette (McGill University). This mutant encodes amino acids 1-301 with cysteine 231 mutated to serine. GST-paxillin N-C3 was generated by amplifying a fragment of the paxillin cDNA (encoding amino acids 1-312) by polymerase chain reaction using primers that created a BamHI site at the 5Ј-end and an EcoRI site at the 3Ј-end. The fragment was subcloned into pGEX2TK in frame with the GST coding sequence. The fusion proteins were expressed by incubation with 0.3 mM isopropyl ␤-D-thiogalactopyranoside for 5 h at 30°C, the bacteria were lysed by sonication, and the fusion proteins were purified using glutathione-agarose beads (Sigma) (38).
Protein Analysis-Cells were lysed in modified radioimmune precipitation assay buffer (36), and protein concentrations were determined using the BCA assay (Pierce). Immunoprecipitations were performed using ϳ500 g of cell lysate and 10 l of polyclonal antisera or 2 g of purified antibody. For CAK␤ immunoprecipitations, a polyclonal antiserum was used (described below). Paxillin was immunoprecipitated using a polyclonal antiserum described previously (39), and FAK was immunoprecipitated using the BC4 polyclonal antiserum (37), a gift of Dr. Thomas Parsons (University of Virginia). The immune complexes were recovered by incubation with protein A-Sepharose beads (Amersham Pharmacia Biotech) for 1-2 h at 4°C. Immune complexes were washed twice with lysis buffer and twice with Tris-buffered saline (10 mM Tris-HCl (pH 7.5), 150 mM NaCl). The samples were boiled in Laemmli sample buffer (40) and analyzed by SDS-PAGE and Western blotting (41). For Western blotting, monoclonal antibodies recognizing CAK␤, paxillin, and the RC20 phosphotyrosine antibody were purchased from BD Transduction Laboratories (Lexington, KY). The FAK 2A7 monoclonal antibody was kindly provided by Dr. Thomas Parsons (University of Virginia). A rabbit polyclonal antiserum recognizing the noncatalytic domain of PTP-PEST has been described previously (15). CAK␤ phosphorylation site-specific antibodies (PY402, PY579, and PY579/580) were obtained from BIOSOURCE International (Camarillo, CA). These affinity-purified (using both negative and positive affinity purification methods) rabbit polyclonal antibodies have been shown to be highly selective for the targeted phosphorylation site by analyzing site-directed (YF) mutants at the phosphorylation site of interest (see Fig. 5). For RC20 phosphotyrosine immunoblots, membranes were blocked in Tris-buffered saline (10 mM Tris-HCl (pH 7.5), 150 mM NaCl), supplemented with 0.1% Tween 20 (Sigma). For CAK␤ phosphospecific antibodies, immunoblots were blocked in Tris-buffered saline, supplemented with 0.1% Tween 20 (Sigma) and 2% fish gelatin (Sigma). Immunoblots were incubated with 0.5 g/ml phosphospecific antibody for 1 h at room temperature, washed, and incubated with horseradish peroxidase-conjugated protein A (Amersham Pharmacia Biotech). Enhanced chemiluminescence (Amersham Pharmacia Biotech) was used for detection of proteins.
Generation of CAK␤ Polyclonal Antiserum-The rat CAK␤ cDNA was used to generate a GST fusion protein containing amino acids 226 -306. These residues were amplified by polymerase chain reaction and subcloned into pGEX2TK in-frame with the GST coding sequence. This fusion protein was gel-purified and used as an antigen to prepare polyvalent rabbit antiserum as described previously (37). For primary immunizations, 1 mg of fusion protein in phosphate-buffered saline was emulsified with complete Freund's adjuvant and injected subcutane-ously into female New Zealand White rabbits. For all boosts, 500 g of antigen in phosphate-buffered saline was emulsified with Freund's incomplete adjuvant and injected subcutaneously. Serum was prepared by centrifugation of cellular and clotted material from whole blood.
Substrate-trapping Assay-Cells were treated with pervanadate and lysed as described previously (16). Briefly, confluent cells were treated with 50 mM pervanadate for 30 min and lysed in radioimmune precipitation assay buffer supplemented with 5 mM iodoacetic acid for 30 min at 4°C. After lysis, 10 mM dithiothreitol was added to inactivate any unreacted iodoacetic acid. Pervanadate-treated cell lysates (500 g) were precleared with 10 g of GST bound to glutathione-agarose beads at 4°C for 1 h. The cleared supernatants were then incubated with 5 g of either GST alone, GST/PTP-PEST WT , or GST/PTP-PEST C231S fusion protein immobilized on glutathione-agarose beads at 4°C for 1 h. The beads were washed as described for immune complexes above, and bound protein was detected by Western blotting.
In Vitro Kinase Assay-CAK␤ and FAK were immunoprecipitated using polyclonal antiserum as described above. Immune complexes were washed twice in lysis buffer and twice in kinase reaction buffer (50 mM Tris-HCl (pH 7.4), 5 mM MnCl 2 , 5 mM MgCl 2 ) and then resuspended in 20 l of kinase reaction buffer supplemented with 20 M cold ATP, 10 Ci of [␥-32 P]ATP (6000 Ci/mol; PerkinElmer Life Sciences), and 5 g of GST-paxillin N-C3 as an exogenous substrate. The reaction was stopped by the addition of Laemmli sample buffer after incubation at room temperature for 10 min. Samples were resolved by SDS-PAGE and analyzed by autoradiography. In some experiments, pervanadatetreated lysates (500 -1000 g) were incubated with GST/PTP-PEST WT (1-4 g) immobilized on glutathione beads for up to 20 min at room temperature or 1 h on ice. The phosphatase was removed by centrifugation, vanadate was added to the sample, and CAK␤ or FAK was immunoprecipitated. The immune complexes were analyzed in the in vitro kinase assay or by Western blotting.
Immunofluorescence-Cells were fixed in 3.7% formaldehyde and permeabilized with 0.5% Triton X-100 as previously described (42). Commercially available monoclonal antibodies directed against PTP-PEST (Exalpha Biologicals Inc., Boston MA) and CAK␤ (BD Transduction Laboratories) were used. A secondary rabbit anti-mouse bridging antibody and tertiary donkey anti-rabbit antibody conjugated to fluorescein (Jackson Immunoresearch Laboratories Inc., West Grove PA) were used to detect the primary antibody. The cells were visualized using a Leitz Orthoplan microscope, and images were collected with a Hamamatsu CCD camera and MetaMorph imaging software (Universal Imaging Corp., West Chester, PA).

PTP-PEST Substrate Traps CAK␤-
The hypothesis that FAK and CAK␤ are PTP-PEST substrates was initially tested using an in vitro substrate-trapping assay. FAK and CAK␤ were overexpressed in HEK 293 cells, and the cells were treated with pervanadate to increase the cellular pool of tyrosine-phosphorylated proteins. Cell lysates were prepared and incubated with GST fusion proteins containing the catalytic domains of either wild type PTP-PEST (GST/PTP-PEST WT ) or the catalytically inactive, substrate-trapping mutant, GST/ PTP-PEST C231S . As shown in Fig. 1A, CAK␤ bound to GST/ PTP-PEST C231S (lane 3) but was unable to bind GST/PTP-PEST WT or GST alone (lanes 1 and 2). Upon overexposure of the blot, however, a very small amount of CAK␤ was observed to bind to GST/PTP-PEST WT (data not shown), implying that there may be a weak interaction between CAK␤ and an unidentified site in the PTP-PEST catalytic domain. CAK␤ from lysates of cells not stimulated with pervanadate failed to bind to GST/PTP-PEST C231S (data not shown; see Fig. 4A). These two results suggest that the observed phosphotyrosine dependent interaction of CAK␤ with GST/PTP-PEST C231S was due to substrate binding to the active site of the phosphatase. Whereas CAK␤ was readily detectable in complex with GST/ PTP-PEST C231S (Fig. 1A, lane 3), no detectable FAK was bound (Fig. 1B, lane 3). FAK could very weakly bind to the GST/PTP-PEST C231S fusion protein as seen upon overexposure of the Western blot (data not shown). Note that the signal intensities of the loading controls for the FAK blot (Fig. 1B, lane 4) and CAK␤ blot (Fig. 1A, lane 4) are similar. These data suggest that CAK␤ is a preferred substrate for PTP-PEST in comparison with FAK. In order to examine the ability of PTP-PEST to substrate trap endogenous CAK␤, the A7r5 vascular smooth muscle cell line was employed. Cells were treated with pervanadate, and lysates were prepared. CAK␤ from these lysates bound to GST/PTP-PEST C231S but was unable to bind GST/ PTP-PEST WT or GST alone (Fig. 1C, lanes 1-3). CAK␤ from cells not treated with pervanadate also failed to bind to any of the fusion proteins (data not shown). These results suggest that CAK␤ was being trapped by the catalytically inactive mutant, GST/PTP-PEST C231S , and is likely a substrate for PTP-PEST.
PTP-PEST Dephosphorylates CAK␤ in Vivo-Next, the ability of PTP-PEST to dephosphorylate CAK␤ in vivo was examined. CAK␤ was expressed either alone or with PTP-PEST in HEK 293 cells. The cells were then lysed, and CAK␤ was immunoprecipitated and Western blotted for phosphotyrosine. As shown in Fig. 2A, CAK␤ was phosphorylated on tyrosine when expressed alone (top panel, lane 1) but was dramatically dephosphorylated upon PTP-PEST coexpression (top panel, lane 2). The differences in phosphotyrosine were not due to differential recovery of CAK␤ as shown by stripping and reprobing the phosphotyrosine Western blot with a CAK␤ monoclonal antibody ( Fig. 2A, middle panel). A Western blot of whole cell lysates revealed expression of PTP-PEST in the transfected cells ( Fig. 2A, bottom panel, lane 2). To determine if CAK␤ dephosphorylation was specific, whole cell lysates were Western blotted for phosphotyrosine. Except for a band corresponding to CAK␤, the phosphotyrosine content of cellular proteins in CAK␤ expressers and cells expressing both CAK␤ and PTP- PEST was similar (Fig. 2B, lanes 1 and 2). A reduction in the phosphotyrosine content of a few cellular proteins was observed, including a 68-kDa protein that is probably paxillin. Another CAK␤ and PTP-PEST substrate is p130 CAS , which also becomes dephosphorylated upon expression of PTP-PEST in HEK 293 cells (15). Two other CAK␤-associated substrates are the Nir proteins and PAP/KIAA0400 (43,44). Nir2, the most broadly expressed Nir family member, is ϳ140 kDa, and PAP is 112 kDa. Although there are reduced phosphotyrosine levels in this range of molecular masses in cells coexpressing CAK␤ and PTP-PEST, tyrosine phosphorylation of endogenous proteins is difficult to discern, since bands in this region of the gel are obscured by the exogenously expressed CAK␤. Further experiments will be required to verify that additional CAK␤ substrates become dephosphorylated in PTP-PEST-expressing cells. These results demonstrate that PTP-PEST selectively dephosphorylates cellular proteins when expressed in HEK 293 cells, suggesting that CAK␤ is a substrate of PTP-PEST in vivo.
To further explore the specificity of CAK␤ dephosphorylation, the ability of PTP-PEST to dephosphorylate FAK in vivo was also examined. FAK or CAK␤ were expressed in HEK 293 cells either alone or with different amounts of PTP-PEST, and their phosphotyrosine content was examined by immunoprecipitation and Western blotting. As seen in Fig. 3A, coexpression of PTP-PEST with FAK resulted in a reduction in tyrosine phosphorylation of FAK. However, even in the presence of relatively high levels of PTP-PEST, very substantial levels of tyrosine-phosphorylated FAK remain (Fig. 3A, lane 4). In contrast, coexpression of even low levels of PTP-PEST with CAK␤ virtually abolished tyrosine phosphorylation of CAK␤ (Fig. 3B,  lane 2). To further explore the regulation of tyrosine phosphorylation of FAK by PTP-PEST, the level of phosphotyrosine present on endogenous FAK was examined. Expression of PTP-PEST in HEK 293 induced a very modest decrease in the phosphotyrosine content of endogenous FAK (Fig. 3C). The expression of endogenous CAK␤ in HEK 293 cells was virtually undetectable, precluding a similar analysis for CAK␤ (data not shown). These results are consistent with the hypothesis that CAK␤ is a specific substrate for PTP-PEST, whereas FAK is a much weaker substrate.
PTP-PEST Targets the Major Autophosphorylation Site and Activation Loop of CAK␤-In order to determine what site(s) of tyrosine phosphorylation of CAK␤ is targeted by PTP-PEST, site-specific CAK␤ mutants in which tyrosines were replaced with phenylalanine were used (34). These mutants (Y402F, Y579F/Y580F, Y881F) and wild type CAK␤ were expressed in HEK 293 cells. The cells were then either left unstimulated or treated with pervanadate, lysed, and subjected to the substrate-trapping assay. As expected, GST/PTP-PEST C231S was able to trap wild type CAK␤ from pervanadate-treated cells, while GST/PTP-PEST WT could not (Fig. 4A, compare lanes 2  and 4). Further, neither GST/PTP-PEST WT nor GST/PTP-PEST C231S bound to CAK␤ from cells that were not treated with pervanadate. GST/PTP-PEST C231S also trapped Y881F to a similar extent as wild type CAK␤ (Fig. 4A, compare lanes 4  and 16). The Y402F mutant was also trapped by GST/PTP-PEST C231S , although binding was reduced relative to wild type or Y881F CAK␤ (Fig. 4A, compare lane 8 with lanes 4 and 16). However, the Y579F/Y580F mutant was only very weakly trapped by GST/PTP-PEST C231S (Fig. 4A, lane 12), and this was only apparent upon overexposure of the Western blot (data not shown). Similar levels of each CAK␤ variant were detected in cell lysates (Fig. 4B); thus, these results are not due to differential expression. These results suggest that the activation loop tyrosines are specifically targeted for dephosphorylation by CAK␤.
One trivial explanation why the Y579F/Y580F mutant was poorly trapped is that its phosphotyrosine content is lower than  that of the wild type protein or the other CAK␤ mutants from pervanadate-treated cells. To examine this possibility, CAK␤ was immunoprecipitated from pervanadate-treated lysates and Western blotted for phosphotyrosine. Whereas similar levels of phosphotyrosine were seen in wild type and Y881F (data not shown), Y402F exhibited a slight decrease in its phosphotyrosine content relative to wild type, and Y579F/Y580F had dramatically reduced levels of phosphotyrosine (Fig. 5A). A CAK␤ Western blot shows that equal amounts of each variant were immunoprecipitated (Fig. 5D); thus, the reduced signal on Y579F/Y580F is really due to lower levels of phosphotyrosine. This result could simply reflect the fact that Y579F/Y580F is missing two sites of phosphorylation and thus is expected to be hypophosphorylated. Alternatively, since Tyr 579 and Tyr 580 are important for CAK␤ kinase activity, it was possible that other sites of tyrosine phosphorylation, especially the major autophosphorylation site at Tyr 402 , were hypophosphorylated in the Y579F/Y580F mutant. In order to test this hypothesis, phosphospecific antibodies recognizing distinct CAK␤ phosphotyrosine residues were employed. CAK␤ immune complexes prepared from lysates of pervanadate-treated HEK 293 cells were analyzed by Western blotting. An antibody specific for phosphorylated Tyr 402 (PY402) was used to examine phosphorylation of this site. Western blotting with PY402 revealed strong reactivity with wild type CAK␤ (Fig. 5B, lane 2). This antibody was specific, since it did not react with the Y402F mutant of CAK␤ (Fig. 5B, lane 3). The Y579F/Y580F mutant did contain dramatically reduced levels of phosphorylation on Tyr 402 as compared with wild type CAK␤ (Fig. 5B, compare lane 4 with  lane 2). An antibody specific for CAK␤ that is dually phosphorylated on Tyr 579 and Tyr 580 (PY579/580) recognized wild type CAK␤ but very poorly recognized the Y579F/Y580F mutant (Fig. 5C, lanes 2 and 4); thus, this antibody is also specific. The PY579/580 antibody recognized the Y402F mutant to a lesser extent than wild type CAK␤ (Fig. 5C, lanes 2 and 3), suggesting that Y402F exhibits reduced phosphorylation of the activation loop tyrosine residues. The results of these experiments underscore the complex regulation of CAK␤ by tyrosine phosphorylation. The activation loop tyrosine residues must be phosphorylated for maximal catalytic activity of CAK␤, and thus the Y579F/Y580F mutant exhibits reduced tyrosine phosphorylation of the autophosphorylation site (Tyr 402 ). Conversely, the association of Src family PTKs with the autophosphorylation site of CAK␤ is required for maximal phosphorylation of the activation loop tyrosine residues, and thus the Y402F mutant exhibits reduced tyrosine phosphorylation at these sites.
Dephosphorylation of Tyr 402 and Tyr 579/580 by PTP-PEST in Vitro-In order to confirm the dephosphorylation of specific tyrosine residues on CAK␤ by PTP-PEST, an in vitro phosphatase assay was employed. Lysates of pervanadate-treated HEK 293 cells expressing wild type CAK␤ were incubated with GST/ PTP-PEST WT for various times at room temperature or on ice for 1 h. CAK␤ was immunoprecipitated from the lysates and analyzed for phosphotyrosine by Western blotting. Incubation with the phosphatase for 5 min at room temperature resulted in a reduction in the phosphotyrosine content of CAK␤, although significant levels of phosphotyrosine remained (Fig. 6A,  lane 2). Incubation for up to 20 min at room temperature did not result in further dephosphorylation of CAK␤ (Fig. 6A, lane  4), and CAK␤ was not dephosphorylated when incubated for 20 min with GST alone (Fig. 6A, lane 5). In contrast to the result using a phosphotyrosine antibody (Fig. 6A), the results using phosphospecific antibodies demonstrated very dramatic reductions in the levels of phosphorylation of Tyr 402 and Tyr 579/580 (Fig. 6, B and C, lanes 2 and 3). A CAK␤ Western blot verified that similar amounts of protein were recovered in the immune complexes (Fig. 6D). Since little tyrosine phosphorylation of CAK␤ was detected with the phosphospecific antibodies, the phosphotyrosine seen in Fig. 6A is probably due to phosphorylation of other tyrosine residues. FAK can be phosphorylated on at least six tyrosine residues (45,46), and similarly CAK␤ is anticipated to have other sites of tyrosine phosphorylation. These results demonstrate that the PTP-PEST can target Tyr 402 and Tyr 579/580 for dephosphorylation in vitro yet spares other tyrosine residues within CAK␤ from dephosphorylation.
PTP-PEST Inhibits CAK␤ Kinase Activity-Phosphorylation of Tyr 579 and Tyr 580 in the activation loop of CAK␤ enhances its kinase activity (34). To test the possibility that PTP-PEST could function to negatively regulate CAK␤ kinase activity, HEK 293 cells overexpressing either wild type CAK␤ or the Y579F/Y580F mutant were treated with pervanadate. Lysates were then incubated with either GST alone or GST/PTP-PEST WT to catalyze dephosphorylation of CAK␤ in vitro. CAK␤ was then immunoprecipitated and subjected to an in vitro kinase assay using GST-paxillin N-C3 as an exogenous substrate. CAK␤ very effectively phosphorylates this substrate in vitro (Fig. 7A, lane 1). As shown in Fig. 7A, CAK␤ kinase activity was dramatically reduced in lysates incubated with GST/PTP-PEST WT compared with lysates incubated with GST alone (compare lanes 1 and 2). Further, phosphorylation of the activation loop tyrosines was important for optimal catalytic activity of CAK␤, since Y579F/Y580F exhibited dramatically reduced catalytic activity relative to wild type CAK␤ (Fig. 7A,  lane 3). Western blotting for CAK␤ revealed that equal amounts of wild type and Y579F/Y580F were present in the immune complexes (Fig. 7B, lanes 1 and 2). If PTP-PEST is negatively regulating CAK␤ kinase activity by dephosphorylation of activation loop tyrosines, then the phosphatase should have no effect on the Y579F/Y580F CAK␤ mutant. This was indeed the case, since Y579F/Y580F had the same level of kinase activity regardless of whether the lysate was incubated with GST or GST/PTP-PEST WT (Fig. 7A, compare lanes 3 and  4). Interestingly, when wild type CAK␤ was incubated in the presence of GST/PTP-PEST WT , its kinase activity was reduced to almost the level of activity exhibited by the Y579F/Y580F mutant (Fig. 7A, compare lanes 2 and 3). This result supports the hypothesis that PTP-PEST is targeting activation loop tyrosines and that dephosphorylation of CAK␤ by PTP-PEST impairs catalytic activity of the kinase. Unlike CAK␤, PTP-PEST had little or no effect on the kinase activity of FAK. Similar catalytic activities were observed following incubation with GST alone and with GST/PTP-PEST WT (Fig. 7C, compare  lanes 1 and 2). This finding demonstrates the substrate specificity of PTP-PEST, since it inhibited the catalytic activity of CAK␤ in vitro but had no effect on the highly related tyrosine kinase, FAK.  1 and 2) and FAK (lanes 3 and 4) were immunoprecipitated and subjected to kinase assays as in A and C. PTP-PEST was also able to inhibit the kinase activity of CAK␤ in vivo when the two were coexpressed in HEK 293 cells. The catalytic activity of CAK␤ was measured by immunoprecipitating the kinase from cell lysates and incubating the immune complex in an in vitro kinase assay. The catalytic activity of CAK␤ immunoprecipitated from cells coexpressing PTP-PEST was dramatically reduced relative to the catalytic activity of CAK␤ when expressed alone (Fig. 7D, compare lanes 1  and 2). In contrast, the catalytic activity of FAK immunoprecipitated from cells coexpressing PTP-PEST was the same as the activity of FAK expressed alone (Fig. 7D, compare lanes 3  and 4). Thus, PTP-PEST was unable to inhibit FAK kinase activity in vivo. These data support the substrate trapping data and suggest that PTP-PEST prefers CAK␤ to FAK as a substrate. Furthermore, they suggest that PTP-PEST can directly regulate the catalytic activity of CAK␤ by dephosphorylation of activation loop tyrosine residues.
PTP-PEST Negatively Regulates CAK␤-dependent Tyrosine Phosphorylation in Vivo-PTP-PEST is able to bind and dephosphorylate paxillin and p130 CAS (13,15). Since paxillin and p130 CAS are also binding partners of CAK␤, it seemed plausible that either of these proteins might serve as a scaffold to bring PTP-PEST into the proximity of CAK␤ to allow dephosphorylation to occur. In order to address this, CAK␤ was expressed in HEK 293 cells either alone or in combination with wild type PTP-PEST or deletion mutants of PTP-PEST (⌬367-400 and ⌬297-493), shown previously to be unable to bind paxillin (15), or a PTP-PEST mutant (P337A PTP-PEST) that is unable to bind p130 CAS (13). CAK␤ was immunoprecipitated from cell lysates and Western blotted for phosphotyrosine. As already shown in Figs. 2A and 3B, coexpression with wild type PTP-PEST induced a complete loss of phosphotyrosine on CAK␤ (Fig. 8A, lane 2, top panel). Both deletion mutants and the P337A mutant were able to induce CAK␤ dephosphorylation to the same extent as wild type PTP-PEST (Fig. 8A, lanes 3-5, top  panel). The blot was stripped and reprobed with a CAK␤ monoclonal antibody to show that equal levels of protein were immunoprecipitated (Fig. 8A, bottom panel). This result suggests that the association of PTP-PEST with neither paxillin nor p130 CAS is necessary for dephosphorylation of CAK␤ by PTP-PEST.
Expression of CAK␤ in chick embryo fibroblasts results in enhanced phosphorylation of paxillin (47). Similarly, paxillin becomes hyperphosphorylated when CAK␤ is overexpressed in HEK 293 cells (Fig. 8B, compare lanes 1 and 3). Coexpression of PTP-PEST with CAK␤ results in a reduction in paxillin phosphotyrosine levels (Fig. 8B, compare lanes 3 and 4). This could be the result of PTP-PEST inhibiting phosphorylation of paxillin by negatively regulating CAK␤ kinase activity. Alternatively, PTP-PEST could also be dephosphorylating paxillin itself, since paxillin was shown to be a substrate for PTP-PEST (15). The PTP-PEST deletion mutants deficient in paxillin binding were used to address this question. Paxillin binding is essential for the direct dephosphorylation of paxillin by PTP-PEST (15). As shown in Fig. 8A, paxillin binding is completely dispensable for dephosphorylation of CAK␤ by PTP-PEST. Either ⌬367-400 or ⌬297-493 PTP-PEST mutants were coexpressed with CAK␤ in HEK 293 cells. Paxillin was then immunoprecipitated, and its phosphotyrosine content was examined by Western blotting. Both ⌬367-400 and ⌬297-493 mutants were able to reduce paxillin phosphotyrosine levels compared with cells expressing CAK␤ alone (Fig. 8B, compare lanes 5 and 6 with lane 3). Since both mutants fail to directly dephosphorylate paxillin (15), this result implies that PTP-PEST is indirectly reducing tyrosine phosphorylation of paxillin by inhibiting CAK␤. However, both deletion mutants were unable to reduce paxillin phosphotyrosine levels to the same extent as wild type PTP-PEST (Fig. 8B, compare lanes 5 and 6 with lane  4). Whole cell lysates were Western blotted with the polyclonal PTP-PEST antiserum to demonstrate that differences in levels of tyrosine phosphorylation of paxillin were not due to differences in expression levels of wild type PTP-PEST and the various mutants (Fig. 8C). These results suggest that PTP-PEST also acts to directly dephosphorylate paxillin. Therefore, PTP-PEST is able to negatively regulate a signal sent by CAK␤ in vivo by targeting both the kinase and its substrate for dephosphorylation. Subcellular Localization of Endogenous PTP-PEST and CAK␤-FAK is unlikely to be a major target for PTP-PEST in vivo, since FAK is localized in focal adhesions and PTP-PEST is found in the cytoplasm (37,48). To determine if PTP-PEST and CAK␤ reside in the same cellular compartment, their subcellular localization was examined by immunofluorescence in the GN4 and A7r5 cell lines, which express both of these proteins endogenously. In both cell types, immunofluorescent staining for both PTP-PEST and CAK␤ was relatively nondescript. Each protein was located within the cytoplasm of the cells, and the staining appeared particulate (Fig. 9). As a control, the primary antibodies were omitted, resulting in very dim background fluorescence (Fig. 9, A and B). This result confirms previous findings that PTP-PEST and CAK␤ are localized in the cytoplasm of cells (47)(48)(49) and supports the contention that CAK␤ may be a physiological substrate for PTP-PEST, since both are found in a similar cellular compartment.

DISCUSSION
In this study, we have demonstrated the direct regulation of the PTK CAK␤ by PTP-PEST. CAK␤ is substrate-trapped by a catalytically inactive PTP-PEST mutant and is dephosphorylated in vivo when coexpressed with PTP-PEST. PTP-PEST targets tyrosine residues 579 and 580 in the activation loop of CAK␤ for dephosphorylation. As a result, PTP-PEST decreases CAK␤ kinase activity both in vitro and in vivo. This inhibition of CAK␤ kinase activity results in the reduced ability of CAK␤ to phosphorylate paxillin in vivo. In addition, PTP-PEST targets tyrosine residue 402 for dephosphorylation. Loss of phosphorylation at this site will prevent Src family kinases from binding to CAK␤, which in turn will result in reduced tyrosine phosphorylation of CAK␤ and other downstream substrates in the pathway. Perhaps as interesting, the results also demonstrate that PTPs can exhibit very strong specificity for substrates, since PTP-PEST appears to strongly prefer CAK␤ as a substrate to its highly homologous relative, FAK.
There are several observations regarding tyrosine phosphorylation of CAK␤ that merit further discussion. First, when PTP-PEST is coexpressed with CAK␤, a complete loss of phosphotyrosine on CAK␤ is observed ( Fig. 2A, top panel). Since PTP-PEST appears to target Tyr 402 , Tyr 579 , Tyr 580 , the complete loss of phosphotyrosine on CAK␤ is intriguing. Perhaps these three sites are the only residues phosphorylated when CAK␤ is exogenously expressed in cells. If other tyrosine residues are phosphorylated upon CAK␤ expression, how are these sites dephosphorylated upon PTP-PEST overexpression? One possible explanation is that by targeting the activation loop, PTP-PEST turns off CAK␤ kinase activity, and by directly dephosphorylating Tyr 402 the Src family kinase binding site is removed. As a result, Src can no longer bind CAK␤ and transphosphorylate other tyrosine residues, such as Tyr 881 , on CAK␤. Second, the observed pattern of tyrosine phosphoryla-tion of CAK␤ mutants following pervanadate treatment differs from the published pattern of phosphorylation from untreated cells (34). When expressed in HEK 293(T) cells, the Y402F mutant has been reported to be virtually devoid of phosphotyrosine, and the Y579F/Y580F mutant exhibits levels of phosphotyrosine dramatically lower than wild type CAK␤. In fact, we find similar results in experiments in which the cells are not pervanadate-treated (data not shown). These findings suggest that tyrosine 402 is the major site of tyrosine phosphorylation and/or a prerequisite site of phosphorylation for the phosphorylation of additional tyrosine residues. The observation that pervanadate treatment induces tyrosine phosphorylation of the Y402F mutant suggests that the SH2 domain-mediated association of Src with CAK␤ may not be absolutely required for phosphorylation of additional tyrosine residues in CAK␤. This result suggests that there is a second mechanism for tyrosine phosphorylation of CAK␤, although this is probably a minor mechanism of phosphorylation, since it only becomes evident following inhibition of PTPs with pervanadate. However, the SH2-mediated interaction between Src and CAK␤ may be necessary for optimal phosphorylation of the activation loop, since Y402F exhibits lower levels of phosphotyrosine on residues 579/580 than wild type CAK␤, even from pervanadate-treated cells (34) (Fig. 5B).
PTP-PEST is not the only phosphatase reported to associate with CAK␤. It was recently reported that SHP-1 binds constitutively to CAK␤ (50). The data suggest that SHP-1 is involved in dephosphorylation of Tyr 402 , the major autophosphorylation and Src-SH2 binding site. Unlike PTP-PEST, SHP-1 had no effect on CAK␤ kinase activity. SHP-2, another SH2 domaincontaining protein-tyrosine phosphatase, was also recently reported to constitutively associate with CAK␤ (51). SHP-2 binding to CAK␤ was shown to be independent of SHP-2 SH2 domains. Further, CAK␤ was dephosphorylated in vivo upon expression of a SHP-2 mutant with enhanced phosphatase activity that lacks its two N-terminal SH2 domains. Additional studies are required to establish which site(s) are targeted by SHP-2 and whether SHP-2 can regulate CAK␤ kinase activity. One intriguing possibility is that SHP-1 and SHP-2 function cooperatively with PTP-PEST in down-regulating CAK␤ signaling. According to this model, PTP-PEST may be responsible for turning off CAK␤ kinase activity by dephosphorylation of activation loop residues, and SHP-1 or SHP-2 may be responsible for inhibiting the assembly of the Src-CAK␤ complex by dephosphorylating the Src SH2 domain binding site.
Surprisingly, PTP-PEST substrate trapped CAK␤ much better than FAK. Small amounts of FAK could sometimes be found in association with the PTP-PEST substrate-trapping mutant, but it was always much less in comparison with CAK␤. 2 This does not rule out the possibility that FAK may be a substrate for PTP-PEST. For example, paxillin was unable to be substrate-trapped but nevertheless appears to be a substrate for PTP-PEST, since it directly binds PTP-PEST and becomes dephosphorylated in vivo upon overexpression of wild type PTP-PEST (15). Despite the failure of PTP-PEST to efficiently substrate-trap FAK, there is some evidence suggesting that PTP-PEST could regulate FAK dephosphorylation. First, PTP-PEST activity was coimmunoprecipitated with FAK from lysates of chicken embryo fibroblasts (36). Second, FAK was hyperphosphorylated in fibroblasts where PTP-PEST was removed by gene targeting (29). However, PTP-PEST and FAK have never been colocalized in cells. PTP-PEST is localized diffusely throughout the cytoplasm or at the membrane periphery upon cell adhesion to fibronectin (29), whereas FAK is predominately localized in focal adhesions (37). Clearly, further experimentation is required to define the role of PTP- The primary antibodies were detected with secondary rabbit antimouse and tertiary donkey anti-rabbit conjugated to fluorescein. The primary antibody was omitted in control slides (A and B). The GN4 cells were observed with a ϫ 63 objective, and the A7r5 cells were observed with a ϫ 40 objective.
The observation that CAK␤ can be substrate-trapped by PTP-PEST better than FAK raises the interesting question of substrate specificity, since CAK␤ and FAK are highly homologous, sharing 45% overall sequence identity and 60% identity in the catalytic domain. As stated previously, they share several conserved tyrosine residues, including two tyrosines in the catalytic domain. Why then does PTP-PEST so strongly prefer CAK␤? One reason may be a slight difference in amino acid sequence surrounding the adjacent tyrosine residues. In the activation loop, the CAK␤ sequence is 575 EDEDpYpYKAS 583 (where pY represents phosphotyrosine), while the equivalent FAK sequence is 572 EDSTpYpYKAS 580 . It may be that the difference in the sequences immediately upstream of the phosphorylated tyrosine residues is the reason that PTP-PEST prefers CAK␤ as a substrate. This is probably the case, since several reports have suggested that aspartic acid and glutamic acid within the first 5 residues N-terminal to the phosphorylated tyrosine residues increases the binding affinity of several phosphatases for their substrates and increase catalysis of the substrate (59 -61). In addition, the crystal structure of another cytoplasmic tyrosine phosphatase, PTP1B, complexed with a high affinity substrate has been determined (62). The crystal structure suggests that substrate specificity is conferred by the interaction of basic residues on the surface of the phosphatase with acidic residues of the peptide that lie to the N-terminal side of the phosphorylated tyrosine residue.
Paxillin is phosphorylated upon overexpression of CAK␤ in HEK 293 cells (Fig. 8) and is presumably a downstream effector of CAK␤ signaling. Both paxillin and CAK␤ are substrates for PTP-PEST, suggesting a model whereby PTP-PEST regulates this pathway at two distinct points (Fig. 10). First, PTP-PEST is able to directly bind and dephosphorylate paxillin. Second, PTP-PEST dephosphorylates activation loop tyrosine residues on CAK␤, rendering the kinase catalytically inactive, and further targets the major autophosphorylation and Src kinase binding site of CAK␤. Inhibition of the kinase activity of CAK␤ and blocking recruitment of Src kinases, in turn, prevents the CAK␤-Src kinase complex from phosphorylating its downstream substrate, paxillin. Perhaps regulation at these two distinct points is required for complete inactivation of this signal. Alternatively, there may be additional, independent mechanisms regulating targeting of these two PTP-PEST substrates for dephosphorylation. In this case, CAK␤ signaling could be inhibited by dephosphorylation of the kinase. Alternatively, one branch of the CAK␤ signaling cascade could be blocked by dephosphorylation of paxillin, allowing other CAK␤dependent signals to be propagated. This targeting of distinct members of this signaling pathway under specific conditions might increase the versatility of signals transmitted.
The discovery that PTP-PEST can regulate the catalytic activity of CAK␤ and phosphorylation of CAK␤ substrates provides important new insight into signaling via this pathway. The efficient transmission of downstream signals might require the regulation of both the PTK and the PTP, which could involve changes in the catalytic activity of these enzymes and/or alterations in protein-protein interactions. There is clearly considerable work to be done to unravel the complex interplay between these two enzymes and determine how their activities might be coordinated to allow the propagation of signals. In addition, PTP-PEST can dephosphorylate activation loop tyrosine residues on CAK␤, inhibiting kinase activity, and the major autophosphorylation site of CAK␤, inhibiting association with Src family kinases. Therefore, PTP-PEST can target multiple components of this signaling pathway for dephosphorylation.