Differential Signaling by the Focal Adhesion Kinase and Cell Adhesion Kinase β*

pp125FAK and CAKβ/Pyk2/CadTK/RAFTK are related protein-tyrosine kinases. It is therefore of interest whether CAKβ shares some of the properties of pp125FAK. Using recombinant glutathioneS-transferase fusion proteins, we show that the C-terminal domains of both proteins bind paxillin in vitro. The C-terminal domain of CAKβ was engineered to be autonomously expressed in chicken embryo cells and, like pp125FAK and p41/43FRNK (the C-terminal noncatalytic domain of pp125FAK), was found to localize to cellular focal adhesions. In contrast, full-length CAKβ was generally found diffusely distributed throughout the cell, although a fraction of the cells exhibited focal adhesion localization. Vanadate treatment of pp125FAK- and CAKβ-overexpressing CE cells induced a dramatic increase in the phosphotyrosine content of a common set of proteins including tensin, paxillin, and p130Cas, but some of these substrates, particularly p130Cas, appeared to be differentially phosphorylated by pp125FAK and CAKβ. Levels of tyrosine phosphorylation were higher in CAKβ-overexpressing cells, and additional phosphotyrosine-containing species were specifically immunoprecipitated. In addition, vanadate treatment of CE cells overexpressing CAKβ, but not pp125FAKoverexpressors, induced a profound morphological change, which could be a consequence of the observed differences in substrate phosphorylation.

pp125 FAK , 1 the focal adhesion kinase, is a 125-kDa proteintyrosine kinase (PTK) that is discretely localized to cellular focal adhesions (1,2). Thus, pp125 FAK colocalizes with the integrins, which are heterodimeric, transmembrane receptors that bind to proteins in the extracellular matrix, e.g. fibronectin and collagen (3,4). The integrins function in cell adhesion cytoskeleton anchorage and in the transduction of extracellular stimuli into cytoplasmic signals, including the phosphorylation of proteins on tyrosine (5,6). pp125 FAK is one of the major substrates for integrin-dependent tyrosine phosphorylation, and concomitant with tyrosine phosphorylation, it becomes enzymatically active (2,(7)(8)(9)(10). It is therefore anticipated that pp125 FAK will prove a fundamental element in an integrinregulated signaling pathway. In addition, a number of other stimuli, including neuropeptides and growth factors, have been reported to induce tyrosine phosphorylation of pp125 FAK (5).
The function of pp125 FAK has been explored by disrupting endogenous pp125 FAK signaling. Overexpression of the C-terminal noncatalytic domain of pp125 FAK , which contains the focal adhesion targeting sequence, blocks pp125 FAK -induced tyrosine phosphorylation of substrates and impairs the spreading of chicken embryo cells on fibronectin (11). Microinjection of a similar fragment of pp125 FAK into human umbilical vein endothelial cells had little effect upon the structure of focal adhesions but abolished focal adhesion-associated phosphotyrosine and inhibited cell migration (12). pp125 FAK expression has also been obliterated by gene knockout, which results in embryonic lethality (13). Cell lines derived from fak Ϫ/Ϫ embryos exhibit a more rounded morphology, a distribution of focal adhesions that differs from wild type cells, and a retarded rate of cell migration (13). However, normal phosphorylation of pp125 FAK substrates was reported. These results support the contention that pp125 FAK may regulate cell spreading and motility; however, the mechanism of regulation has yet to be established.
Recently, a pp125 FAK -related PTK, called CAK␤/Pyk2/ RAFTK/CadTK, was isolated (14 -17). It exhibits approximately 45% amino acid identity with pp125 FAK . CAK␤ is absent from focal adhesions when expressed exogenously in COS cells, and its phosphotyrosine content is independent of cell adhesion (14). However, in a megakaryocytic cell line, CAK␤ colocalizes with vinculin and exhibits cell adhesion-dependent tyrosine phosphorylation (18). Furthermore, cross-linking ␤ 1 integrins on the surface of B cells using mAbs induces tyrosine phosphorylation of CAK␤ (19). Tyrosine phosphorylation of CAK␤ in PC12 cells occurs in response to a number of stimuli including bradykinin, lysophosphatidic acid, phorbol 12-myristate 13-acetate, carbachol, and membrane depolarization (15,20). Since calcium ionophores stimulate CAK␤ tyrosine phosphorylation and chelation of extracellular Ca 2ϩ abrogates tyrosine phosphorylation of CAK␤ in response to various stimuli, this kinase is believed to be a Ca 2ϩ -regulated PTK (15,20). pp125 FAK and CAK␤ exhibit the same overall topology with large N and C termini approximately 400 residues in length flanking a central catalytic domain. The C-terminal 150 residues of pp125 FAK represent the focal adhesion targeting (FAT) sequence, which directs pp125 FAK to its location in the cell and contains binding sites for two focal adhesion-associated proteins, paxillin and talin (21)(22)(23). The most highly conserved region between these two kinases is the C-terminal FAT sequence, which exhibits 61% identity between CAK␤ and pp125 FAK (the catalytic domains are only 60% identical). There are four sites of tyrosine phosphorylation within pp125 FAK (24,25) that are conserved in CAK␤. Autophosphorylation of pp125 FAK (at tyrosine 397) and CAK␤ (at tyrosine 402) create binding sites for the SH2 domain of the Src-like PTKs (20,24). Phosphorylation of an additional site in pp125 FAK , tyrosine 925, creates a binding site for the SH2 domain of the adaptor protein GRB2 (26), suggesting a function for pp125 FAK in regulating the Ras signaling pathway and ultimately mitogenactivated protein kinase. There is evidence, however, that mitogen-activated protein kinase activation in response to integrin stimulation and other stimuli, like bombesin, is pp125 FAK -independent (27)(28)(29). There is convincing evidence that CAK␤ regulates the activity of mitogen-activated protein kinase (15,20) and some evidence that Jun kinase may be regulated by CAK␤ (17,30). There are also several proline-rich regions in the C-terminal domain that are conserved between these PTKs. In pp125 FAK , these serve as binding sites for the Src homology 3 domains of other signaling molecules including p130 Cas (31,32), phosphatidyl-inositol 3Ј-kinase (33), and a GTPase-activating protein called GRAF (34). Very recently, CAK␤ was reported to associate at low stoichiometry with p130 Cas (19,35).
Given the relatedness of these sequences, it seemed likely that the C-terminal domains of CAK␤ and pp125 FAK might serve similar functions. We report here a comparative analysis of pp125 FAK and CAK␤ and demonstrate that the C-terminal domain of CAK␤ can bind the focal adhesion-associated protein paxillin in vitro and targets to focal adhesions when expressed in chicken embryo (CE) cells. We show that full-length CAK␤ is predominantly found diffusely distributed when exogenously expressed in CE cells, although a fraction of the cells display prominent focal adhesion localization. Comparison of tyrosine phosphorylation of cellular substrates suggests that both pp125 FAK and CAK␤ can phosphorylate similar proteins, although the regulation of phosphorylation and sites targeted are likely to differ. Finally we report that unique morphological changes are induced in CAK␤-expressing CE cells upon vanadate treatment, indicating differential consequences of CAK␤and pp125 FAK -induced tyrosine phosphorylation. These observations suggest that pp125 FAK and CAK␤ may have partially redundant functions but also exhibit very distinct functions, particularly in substrate recognition, when expressed in CE cells.

MATERIALS AND METHODS
Cells-CE cells were prepared as described (36). Exogenous proteins were expressed using the RCAS A replication-competent, avian retroviral vector (37). Plasmid DNA was transfected into CE cells as described (36), and the cells were passaged for 7-10 days, allowing spread of the virus throughout the culture. In experiments to study adhesiondependent tyrosine phosphorylation, cells were trypsinized, and the trypsin was neutralized with soybean trypsin inhibitor (Sigma) and then plated onto Petri dishes coated either with a solution of 0.1 mg/ml of poly-L-lysine or fibronectin (2.5 g/cm 2 ) (10). In some experiments, cells were treated with 50 M sodium orthovanadate for 16 h at 37°C (38). Cells were visualized using a Nikon TMS inverted microscope and photographed using a polaroid camera (NPC Photo Division; Newton, MA) and Type 107 film.
Plasmids-RCAS A constructs containing epitope-tagged derivatives of full-length FAK and the autonomously expressed C-terminal noncatalytic domain, called FRNK, have been described (10). An epitopetagged derivative of the C-terminal noncatalytic domain of CAK␤ was engineered using polymerase chain reaction. Two primers (5Ј-TGACG-GATCCAAGATGGAGAGGGACATTGC and 3Ј-CCCTAAGCTTCTCTG-CAGGAGGGTGGGCCA) were used on the rat CAK␤ cDNA as a template to amplify the sequences between nucleotides 2050 and 3027. The 5Ј-primer created a BamHI site and a point mutation creating a more optimal consensus translation initiation site at methionine 685 (nucleotides 2053-2055) (39). The 3Ј-primer removed the CAK␤ termination codon and created a HindIII site. Approximately 10 ng of template was amplified using 1 M primers, a 0.2 mM concentration of each dNTP, and 2 units of Vent polymerase (New England Biolabs; Beverly, MA) in the manufacturer's reaction buffer for 30 cycles. The product was inserted between the BamHI and HindIII sites of pctag, an epitope tagging vector (10). The resulting construct contains the CAK␤ sequences from methionine 685 to the penultimate codon (glutamic acid 1009) fused in frame with sequences encoding the KT3 epitope tag (40). Nucleotide sequencing of the entire amplified fragment verified that no mutations were introduced during the procedure. An epitope-tagged full-length CAK␤ construct was created by cleaving the original CAK␤ cDNA with EcoRI and XbaI to liberate the fragment from the 5Ј-end of the cDNA to the XbaI site at nucleotide 2803. This fragment was used to replace nucleotides 2050 -2803 of the epitope-tagged C-terminal domain construct in pctag. Both the tagged C-terminal and full-length constructs were then rescued into RCAS A.
Protein Analysis-CE cells were lysed in modified RIPA buffer (41), and protein concentrations were determined using the bicinchoninic acid protein assay (Pierce). Proteins of interest were immunoprecipitated, or lysates were analyzed directly by SDS-PAGE (42). Briefly, 0.5 mg of cell lysate was incubated on ice with 2-5 g of mAb for 1-2 h, and then the immune complexes were captured with goat anti-mouse IgG conjugated to agarose beads (Sigma) by incubation for 1-2 h at 4°C. For denaturation experiments, SDS (1% final concentration) was added to the lysates, and the samples were boiled for 5 min. The SDS was diluted to 0.1% final concentration with RIPA, and immunoprecipitations were performed. Immune complexes were boiled in sample buffer (42) and analyzed by SDS-PAGE and Western blotting. For immunoprecipitations and blotting, commercially available mAbs recognizing paxillin and p130 Cas were used (Transduction Laboratories; Lexington, KY). Monoclonal antibodies recognizing the tag (KT3) and tensin (5B9) were a gift of Dr. J. T. Parsons. Primary antibodies were detected with horseradish peroxidase-conjugated secondary antibodies and then developed by enhanced chemiluminescence (Amersham Corp.). Phosphotyrosine was detected using the recombinant antibody RC20 (Transduction Laboratories) and ECL.

Recombinant Fusion Proteins and in Vitro
Binding-The C-terminal noncatalytic domains of pp125 FAK (residues 687-1054) (22) and CAK␤ (residues 670 -1009) were expressed as glutathione S-transferase fusion proteins in Escherichia coli DH5. Expression was induced at 37°C with 100 M isopropyl ␤-D-thiogalactopyranoside. The bacteria were lysed by sonication in NETN buffer (44), the lysates were clarified, and GST proteins were purified by binding to glutathione-agarose (Sigma) (45). The concentration of GST protein bound to beads was estimated by comparison with a bovine serum albumin standard following SDS-PAGE and staining with Coomassie Blue. Approximately 200 g of total CE cell lysate was precleared by incubation with 24 g of GST immobilized to glutathione-agarose for 1 h at 4°C with constant rocking. The cleared supernatant was incubated with glutathione-agarose beads bound to 2 g of GST-Cterm (pp125 FAK fusion protein), GST-cDom (CAK␤ fusion protein), or GST as a negative binding control. After a 1-h incubation at 4°C with constant rocking, the beads were collected, washed twice with lysis buffer, washed twice with Tris-buffered saline, and boiled in sample buffer (42). The samples were then analyzed by Western blotting.
Immunofluorescence-Cells were plated onto glass coverslips and 24 h later fixed with paraformaldehyde and permeabilized with Triton X-100 as described (36,46). Permeabilized cells were incubated with mAb KT3 (5 g/ml in phosphate-buffered saline), washed with phosphate-buffered saline, and then incubated with rhodamine-conjugated goat anti-mouse secondary antibodies (1 g/ml in phosphate-buffered saline) (Affinipure, min X; Jackson Immunoresearch Laboratories; West Grove, PA). The cells were visualized using a Leitz Orthoplan microscope with a Leitz ϫ 63 (1.4 N.A.) objective, and images were collected with a Hamamatsu CCD camera and MetaMorph imaging software (Universal Imaging Corp.; West Chester, PA).

RESULTS
The C Terminus of CAK␤ Binds Paxillin in Vitro-Given the degree of sequence identity between the C termini of pp125 FAK and CAK␤, it seemed likely that the two might have conserved C-terminal functions. To determine if the C-terminal sequences within CAK␤ were capable of binding paxillin, an in vitro binding assay was utilized. The C-terminal domains of pp125 FAK and CAK␤ were expressed in E. coli as GST fusion proteins and purified (45). The fusion proteins, immobilized on glutathione beads, were incubated with RIPA lysates prepared from CE cells for 1 h at 4°C. After washing, the paxillin bound to the beads was visualized by Western blotting. As previously reported, paxillin bound to the GST fusion protein containing the C-terminal domain of pp125 FAK (22), and a similar amount of paxillin bound to beads containing the C terminus of CAK␤ ( Fig. 1). There was no detectable paxillin bound to GST alone immobilized on glutathione beads, indicating that the pp125 FAK and CAK␤ sequences mediated binding to paxillin. This result suggests that sequences common to pp125 FAK and CAK␤ compose the paxillin binding site.
The C-terminal, Noncatalytic Domain of CAK␤ Targets to Focal Adhesions-Two previous analyses using different cell types report differences in the subcellular localization of CAK␤ (14,18). To further address this issue, we sought to directly compare the focal adhesion targeting capacity of the C-terminal domains of pp125 FAK and CAK␤ in CE cells. The C-terminal domain of CAK␤ was engineered to be expressed autonomously using an internal methionine codon (685 in full-length CAK␤) for translation initiation. This is the CAK␤ equivalent of p41/ 43 FRNK , the C-terminal domain of pp125 FAK , which is expressed autonomously in CE cells and targets to focal adhesions (10). This construct, called CRNK, was epitope-tagged and expressed in CE cells using RCAS A.
Approximately 10 days post-transfection, cells were lysed in RIPA buffer, and 25 g of protein was analyzed by Western blotting using KT3, the mAb directed against the tag. CE cell lysates contained no KT3-immunoreactive bands. Lysate from cells expressing tagged p41/43 FRNK exhibited a single KT3reactive band of M r 43,000 ( Fig. 2A). A 42-kDa band in lysates from cells expressing CRNK was recognized by KT3 ( Fig. 2A). This band could sometimes be resolved into a doublet ( Fig. 2A).
Although not obvious from this experiment, p41/43 FRNK is frequently resolved into a doublet, the slower form representing a phosphorylated isoform of the faster species (10,47). Further analysis indicated that KT3 could immunoprecipitate both p41/ 43 FRNK and CRNK and that polyclonal antiserum and a mAb directed against the C-terminal domain of pp125 FAK recognized p41/43 FRNK but not CRNK (data not shown). These results demonstrate the successful expression of the C-terminal CAK␤ construct and show that KT3 recognizes the tag at its C terminus in both immunoprecipitation and Western blotting assays.
The subcellular localization of tagged CRNK was determined by indirect immunofluorescence using mAb KT3. Uninfected CE cells exhibited a faint, diffuse background staining with KT3 ( Fig. 3A), whereas cells expressing epitope-tagged FRNK exhibited a characteristic focal adhesion staining as reported previously (Fig. 3B) (10). Virtually every CE cell expressing the tagged CRNK construct exhibited a focal adhesion staining pattern that was indistinguishable from the pattern seen with tagged FRNK (Fig. 3C). Therefore, at least when autonomously expressed in CE cells, the C-terminal domain of CAK␤ is effectively localized to cellular focal adhesions.
Subcellular Localization of Full-length CAK␤-Having established that the C terminus of CAK␤ contains sequences that target it to focal adhesions in CE cells, the subcellular localization of the full-length protein was reevaluated. An epitopetagged full-length CAK␤ construct was engineered and expressed in CE cells using RCAS A. Cells were lysed in RIPA buffer, and 25 g of total cellular protein was analyzed by Western blotting with KT3. In lysate from CE cells expressing epitope-tagged pp125 FAK (10), a 125-kDa protein was recognized ( Fig. 2B). KT3 recognized a slightly smaller protein in cells expressing-tagged CAK␤ (Fig. 2B). Both the epitopetagged pp125 FAK and tagged CAK␤ could be immunoprecipitated with KT3 and subsequently detected with KT3 on a Western blot (Fig. 2C). Furthermore, immunoprecipitated tagged pp125 FAK was recognized by mAb 2A7 on a Western blot, whereas tagged CAK␤ was not (data not shown). Conversely, immunoprecipitated, tagged CAK␤ was recognized with a specific polyclonal antiserum raised against CAK␤/ CadTK (a gift of X. Li and Dr. S. Earp) (17), whereas tagged pp125 FAK was not (data not shown). These results verify the expression of the tagged CAK␤ protein and document its recognition with KT3 in both immunoprecipitation and Western blotting.
The subcellular localization of tagged CAK␤ was compared with that of tagged pp125 FAK . Immunofluorescent staining of CE cells expressing pp125 FAK with mAb KT3 produced a characteristic focal adhesion staining pattern in virtually every cell (Fig. 3D). CE cells expressing tagged CAK␤ exhibited two distinct staining patterns. The majority of cells showed a diffuse, nondescript staining pattern (Fig. 3E). A small fraction (ϳ10%) of the CAK␤ overexpressors exhibited an intense focal adhesion staining pattern (Fig. 3F). Thus, regardless of the focal adhesion targeting exhibited by the autonomously expressed Cterminal domain, full-length CAK␤ is generally distributed more diffusely throughout the cell, although under some circumstances it can reside in focal adhesions. Tyrosine Phosphorylation of CAK␤-Since pp125 FAK is one of the major phosphotyrosine-containing proteins in CE cells growing in culture, the phosphotyrosine content of exogenously expressed pp125 FAK and CAK␤ in subconfluent CE cells was compared. pp125 FAK and CAK␤ were immunoprecipitated with mAb KT3 and blotted for phosphotyrosine. The results show substantially less phosphotyrosine on CAK␤ than on pp125 FAK (Fig. 4A, top). Approximately equal amounts of pp125 FAK and CAK␤ were detected in the immune complexes by Western blotting with KT3 (Fig. 4A, bottom). Thus, CAK␤ is underphosphorylated on tyrosine relative to pp125 FAK under standard conditions of growth.
Tyrosine phosphorylation of CAK␤ was originally reported to be independent of cell adhesion (14), but recent evidence has suggested that the phosphotyrosine content of CAK␤ could be regulated by cell adhesion and integrin cross-linking (18,19). The adhesion dependence of tyrosine phosphorylation of CAK␤ was reassessed in CE cells. Cells were trypsinized, held in suspension, or replated onto Petri dishes coated with either poly-L-lysine, to which cells adhere in an integrin-independent manner, or fibronectin, which supports integrin-dependent cell adhesion and spreading. Tagged CAK␤ was immunoprecipitated from cell lysates using KT3, and the immune complexes were analyzed by Western blotting for phosphotyrosine. Invariably, tyrosine phosphorylation of CAK␤ vanished when cells were taken into suspension (Fig. 4B, top). Upon plating onto fibronectin, there was a small increase in the phosphotyrosine content of CAK␤ that was not observed on poly-L-lysine (Fig.  4B, top). Western blotting with KT3 revealed equal amounts of protein in each lane (Fig. 4B, bottom). Therefore, the small amount of phosphotyrosine detected on CAK␤ in cells growing in culture seems to be adhesion-dependent.
CAK␤-dependent Tyrosine Phosphorylation of Cellular Proteins-Initially, Western blots of whole cell lysates were probed with a phosphotyrosine antibody to compare pp125 FAK -and CAK␤-induced tyrosine phosphorylation of cellular proteins. As reported previously, overexpression of pp125 FAK had little effect upon tyrosine phosphorylation, with the exception that the exogenous pp125 FAK was phosphorylated on tyrosine (38) (Fig. 5). Overexpression of CAK␤ induced an increase in the phosphotyrosine content of proteins of approximately 200, 100 -120, and 65-80 kDa (Fig. 5), although the changes at 100 -120 kDa are not particularly obvious in this experiment. Treatment of pp125 FAK -or CAK␤-overexpressing CE cells with 50 M vanadate overnight induced a profound increase in the phosphotyrosine content of many proteins (Fig. 5). These proteins are probably direct or indirect substrates for these two PTKs, since their phosphorylation was dependent upon expression of the exogenous PTKs. Proteins from control CE cells exhibited little change in phosphotyrosine upon vanadate treatment; in fact, the phosphotyrosine content at some positions appears to decline. The profiles of tyrosine-phosphorylated proteins in vanadate-treated pp125 FAK and CAK␤ overexpressors are very similar (Fig. 5B). However, there are several differences in these profiles. For example, the enhanced tyrosine phosphorylation at 120 -130 kDa is much more prominent in pp125 FAK than in CAK␤ overexpressors (Fig. 5B), and a ϳ50-kDa phosphotyrosine-containing protein is evident in CAK␤ overexpressors but not in pp125 FAK overexpressors (Fig.  5A).
Based upon the M r of the tyrosine phosphorylated proteins and previous analysis of pp125 FAK -induced tyrosine phosphorylation (38), it seemed likely that tensin, p130 Cas , and paxillin were substrates for phosphorylation. This was tested by immunoprecipitation and Western blotting. The results of this analysis demonstrate that these three proteins are among the tyrosine-phosphorylated proteins in both pp125 FAK -and CAK␤overexpressing cells (Fig. 6) but that the two PTKs phosphorylate these substrates differently. Tyrosine phosphorylation of tensin is elevated in CAK␤ overexpressors in the absence of vanadate but not in pp125 FAK -overexpressing cells (Fig. 6A). Vanadate treatment increases the phosphorylation of tensin on tyrosine in both pp125 FAK -and CAK␤-overexpressing CE cells. The amount of tensin immunoprecipitated from each lysate is equivalent as revealed by Western blotting. Additional tyrosine-phosphorylated proteins present in the tensin immune complex (125 and 88 kDa) are nonspecifically precipi- tated, since they are present in controls lacking the tensin antibody (data not shown).
Like tensin, the phosphotyrosine content of paxillin is elevated in CAK␤-expressing CE cells but not in pp125 FAK overexpressors (Fig. 6B). Vanadate treatment of both CAK␤and pp125 FAK -overexpressing CE cells induces a profound increase in the phosphotyrosine content of paxillin (Fig. 6B). Equivalent amounts of paxillin were detected in each immune complex (Fig. 6B, bottom). Note that there is a shift in the M r of paxillin upon vanadate treatment. In addition to the broad phosphotyrosine band representing paxillin, the paxillin immune complexes contained additional phosphotyrosine-containing proteins of 195, 135, and 120 kDa (Fig. 6B, top). The 120-kDa protein is nonspecifically precipitated in the absence of the primary paxillin antibody, but the 195-and 135-kDa proteins are specifically coimmunoprecipitated with paxillin (data not shown). In addition, paxillin immune complexes from vanadate-treated CAK␤-expressing cells contain a unique phosphotyrosine-containing protein of 50 kDa. This phosphoprotein is not present in control immunoprecipitations and can be immunoprecipitated with the paxillin mAb from cell lysates that have been denatured (Fig. 7A).
A small amount of phosphotyrosine was detected in p130 Cas immune complexes from CE cells or pp125 FAK overexpressors (not detected in the exposure in Fig. 6). In vanadate-treated pp125 FAK -expressing cells, there was a increase in the phosphotyrosine content of a 125/130-kDa doublet found in p130 Cas immune complexes (Fig. 6C). The 125-kDa species was nonspecifically trapped in the immune complex, since it was also precipitated in the absence of the p130 Cas mAb (Fig. 7B). CAK␤ overexpression induced tyrosine phosphorylation of a 100-kDa protein that was immunoprecipitated with the p130 Cas antibody (Fig. 6C). Vanadate treatment of these cells induced a shift in the mobility of this species. Most strikingly, vanadate treatment of CAK␤ overexpressors induced the appearance of a 145-kDa phosphotyrosine-containing protein in the p130 Cas immune complex (Fig. 6C). These major tyrosine-phosphorylated proteins were not detected in control immune complexes that did not contain the p130 Cas antibody (Fig. 7B). p130 Cas immune complexes from vanadate-treated CAK␤ overexpressors did contain a 120-kDa phosphotyrosine-containing species that was nonspecifically immunoprecipitated and also found in control immunoprecipitations (Fig. 7B). Western blotting revealed a single major p130 Cas -reactive band in these immune complexes and indicated that approximately equal amounts of protein were in each immune complex (Fig. 6C, bottom). However, the p130 Cas from vanadate-treated CAK␤-expressing CE cells migrated more heterogeneously, exhibiting a partial shift to a slower M r (Fig. 6C, bottom, lane 6). To determine if these tyrosine-phosphorylated species, particularly the 100-kDa protein from CAK␤ overexpressors, were in fact p130 Cas -associated proteins, cell lysates were denatured prior to immunoprecipitation. The immunoprecipitation of the 100-kDa tyrosinephosporylated protein from CAK␤ overexpressors with the p130 Cas mAb was not impaired, suggesting that it might not be  FIG. 7. Immunoprecipitation of tyrosine-phosphorylated proteins after denaturation. A, lysates from vanadate-treated CAK␤overexpressing cells were immunoprecipitated with the paxillin mAb (lanes 1 and 3) or with goat anti-mouse agarose in the absence of primary mAb (lane 2). The sample in lane 3 was denatured by boiling in SDS prior to immunoprecipitation. The immune complexes were Western blotted for phosphotyrosine. B, vanadate-treated pp125 FAK -overexpressing CE cell lysates (lanes 1-3) and lysates from CAK␤-overexpressing cells treated with (lanes 7-9) or without vanadate (lanes 4 -6) were subjected to immunoprecipitation with a p130 Cas mAb (lanes 1, 3, 4, 6, 7, and 9) or to control immunoprecipitations in the absence of primary mAb (lanes 2, 5, and 8), and the immune complexes were blotted for phosphotyrosine. The samples in lanes 3, 6, and 9 were denatured by boiling in SDS prior to immunoprecipitation. The arrowheads indicate proteins nonspecifically precipitated.
an associated protein but rather might be directly recognized by the mAb (Fig. 7B). This protein comigrated with a faster migrating, minor p130 Cas -reactive band (Fig. 6C, bottom, lane  5). The 145-kDa phosphotyrosine-containing protein in immune complexes from vanadate-treated CAK␤ overexpressors comigrated with the species of p130 Cas exhibiting the shifted M r .
Induction of Morphological Changes by CAK␤-During the course of this investigation, it was discovered that CAK␤ can induce morphological changes in CE cells. Overexpression of epitope-tagged pp125 FAK did not alter the normal, elongated, fibroblastic morphology of CE cells. Expression of CAK␤ in CE cells induced a subtle morphological change in that a number of rounded cells could be detected among the normal fibroblastic cells in the culture (Fig. 8E). To examine the effects of vanadate, sister plates of nearly confluent cells were incubated overnight with 50 M vanadate. This caused a very dramatic alteration in the shape of cells expressing CAK␤. Many cells became rounded, other cells became elongated and spindleshaped, and the overall appearance of the monolayer became disorganized (Fig. 8F). Treatment of either CE cells or pp125 FAK overexpressors with vanadate did not alter cellular morphology (Fig. 8). DISCUSSION Our comparative analysis of pp125 FAK and CAK␤ has revealed a number of shared properties and a number of properties that are distinct. Both can bind paxillin in vitro, and their C termini contain functional focal adhesion targeting sequences. In cells in culture, the phosphotyrosine content of pp125 FAK is higher than that of CAK␤, but, like pp125 FAK , the basal phosphotyrosine content of CAK␤ appears to be adhesiondependent. The subcellular localization of full-length pp125 FAK and CAK␤ differs, but CAK␤ appears to be capable of localizing to focal adhesions under certain circumstances. Although both PTKs can induce tyrosine phosphorylation of a similar set of proteins, our results indicate that these common substrates might be phosphorylated differently by CAK␤ and pp125 FAK . Furthermore, under some circumstances, CAK␤, but not pp125 FAK , can induce a morphological alteration of CE cells that may reflect differences in substrate recognition.
In the original report describing CAK␤, its subcellular localization was distinct from that of pp125 FAK (14), a finding that was recently challenged (18). Here we report that the exog-enously expressed C-terminal domain of CAK␤ can target to focal adhesions and thus contains a functional FAT sequence. In general, exogenously expressed full-length CAK␤ is diffusely distributed in CE cells, although some cells exhibit CAK␤ in focal adhesions. It is not clear why full-length CAK␤ and the autonomously expressed C-terminal domain localize differently in CE cells. The additional sequences could mask the focal adhesion targeting sequence in the C terminus of CAK␤, or there may be other dominant targeting sequences in CAK␤ that localize the protein to other structures. Under some circumstances, CAK␤ can localize to focal adhesions, since a fraction of the expressing CE cells display a prominent focal adhesion localization of the protein. It is intriguing to speculate that the subcellular localization of CAK␤ might be regulated, perhaps by modification of targeting sequences like the C-terminal FAT sequence.
CAK␤, like pp125 FAK , can bind through its C terminus to the focal adhesion-associated protein, paxillin, in vitro. We have not assessed the association of paxillin and CAK␤ in vivo, since our constructs have been epitope-tagged at their C termini to facilitate immunological detection, and fusion of this tag with the C terminus of pp125 FAK ablates paxillin binding in vivo and in vitro (22). However, two groups have described the coimmunoprecipitation of paxillin and CAK␤ (35,48). It is intriguing that the C terminus of CAK␤ exhibits paxillin binding activity and focal adhesion targeting activity, since it has been proposed that pp125 FAK is targeted to focal adhesions through its interaction with paxillin (49). However, despite the strong correlation reported between paxillin binding and focal adhesion targeting (49), these two functions have been dissociated in a single mutant of pp125 FAK (10,22). Therefore, paxillin binding and focal adhesion targeting probably reflect two distinct functions of the C-terminal domains of these PTKs.
The basal levels of tyrosine phosphorylation of exogenously expressed pp125 FAK and CAK␤ in subconfluent CE cells are different, pp125 FAK exhibiting relatively more phosphotyrosine. This observation could be interpreted as evidence that the two PTKs are regulated differently or that the two are regulated by the same mechanism with different efficiencies. Other analyses indicate that pp125 FAK and CAK␤ can be regulated differently, since a number of stimuli induce a dramatic change in the phosphotyrosine content of CAK␤ in smooth muscle cells or liver epithelial cells without altering the phosphotyrosine content of pp125 FAK (35). 2 On the other hand, reports of tyrosine phosphorylation of either pp125 FAK or CAK␤ in response to various stimuli suggest that some agents, including lysophosphatidic acid and bradykinin, might regulate both PTKs (15,20,50,51). Our analysis demonstrates that the basal level of tyrosine phosphorylation of CAK␤, although relatively low compared with pp125 FAK , is cell adhesion-dependent, like pp125 FAK . This result is consistent with recent publications (18,19). These observations indicate that both pp125 FAK and CAK␤ are regulated by multiple stimuli and that the repertoires of stimuli that induce tyrosine phosphorylation of each of these kinases are partially overlapping.
Given the high degree of sequence conservation between pp125 FAK and CAK␤, it is not surprising that they appear to engage similar downstream pathways to mediate signaling. We show that similar profiles of proteins become tyrosine-phosphorylated in pp125 FAK -and CAK␤-overexpressing CE cells, specifically that the tyrosine phosphorylation of tensin, p130 Cas , and paxillin is induced by both PTKs. Another study has closely correlated tyrosine phosphorylation of paxillin with tyrosine phosphorylation of CAK␤ in the response of rat liver cells to treatment with a number of stimuli (35). Although similar downstream components are implicated as substrates for pp125 FAK and CAK␤, our studies suggest that they are phosphorylated differently by these kinases. Overexpression of CAK␤ is sufficient to induce some increase in the phosphotyrosine content of each of these proteins, whereas pp125 FAK overexpressors must be treated with vanadate (which presumably inhibits an antagonistic PTPase) to induce significant elevations in the phosphotyrosine content of these substrates. Vanadate treatment of CAK␤ overexpressors resulted in a much more dramatic elevation in p130 Cas phosphorylation than treatment of pp125 FAK overexpressors. Furthermore, distinct tyrosine-phosphorylated species are specifically detected in paxillin and p130 Cas immune complexes prepared from CAK␤-expressing CE cells. The 50-kDa species in paxillin immune complexes may represent an alternate species, since it is immunoprecipitated with the paxillin mAb following denaturation and migrates with a M r similar to that of a cellular protein directly recognized by the paxillin mAb in a Western blot of cell lysates (see Fig. 1). This would be the first report of tyrosine phosphorylation of this species, and its ability to serve as a substrate for CAK␤ and pp125 FAK is substantially different. The identity of the 100-kDa tyrosine-phosphorylated protein immunoprecipitated with the p130 Cas mAb from untreated CAK␤-overexpressing cells is unclear. It does comigrate with a minor p130 Cas species and can be immunoprecipitated following denaturation, suggesting that it may be an authentic p130 Cas species. Comparison of the intensity of the signals of the minor and major p130 Cas species in the p130 Cas blot and phosphotyrosine blot suggests that this minor species is relatively hyperphosphorylated. Alternatively, this may represent a p130 Cas -related protein or an associated protein that has the capacity to rapidly reassociate following disruption by denaturation.
In addition to these biochemical differences in substrate phosphorylation by pp125 FAK and CAK␤, the two PTKs elicit distinct morphological changes in CE cells. Treatment of CAK␤-expressing cells with vanadate induces a profound change in the morphology of the cells. The cells become more spindly and refractile, and many cells become rounded. Vanadate-treated pp125 FAK -overexpressing CE cells may become slightly more spindle shaped but generally are very similar in morphology to normal CE cells. We speculate that differential phosphorylation of cellular substrates may be responsible for the difference in morphology of pp125 FAK -and CAK␤-expressing cells. The mechanism by which the morphological alteration is brought about is unknown, but there are presumably corresponding changes in the cytoskeleton. Likewise, the significance of the morphological alteration remains to be elucidated. The morphology of the CAK␤ overexpressors treated with vanadate somewhat resembles that of oncogenically transformed CE cells; therefore, overexpression of CAK␤ cells may produce aspects of the transformed phenotype. Alternatively, since CAK␤ has been linked to activation of Jun kinase (17,30), the changes induced in the cellular morphology could be a consequence of sustained Jun kinase signaling. In any event, the effect of CAK␤ upon the morphology of CE cells is a provocative observation. Experiments to further characterize the phenotype and to begin to unravel the mechanism(s) controlling these changes are currently in progress in the laboratory.