Phosphorylation and Free Pool of β-Catenin Are Regulated by Tyrosine Kinases and Tyrosine Phosphatases during Epithelial Cell Migration*

Cell migration requires precise control, which is altered or lost when tumor cells become invasive and metastatic. Although the integrity of cell-cell contacts, such as adherens junctions, is essential for the maintenance of functional epithelia, they need to be rapidly disassembled during migration. The transmembrane cell adhesion protein E-cadherin and the cytoplasmic catenins are molecular elements of these structures. Here we demonstrate that epithelial cell migration is accompanied by tyrosine phosphorylation of β-catenin and an increase of its free cytoplasmic pool. We show further that the protein-tyrosine phosphatase LAR (leukocyte common antigen related) colocalizes with the cadherin-catenin complex in epithelial cells and associates with β-catenin and plakoglobin. Interestingly, ectopic expression of protein-tyrosine phosphatase (PTP) LAR inhibits epithelial cell migration by preventing phosphorylation and the increase in the free pool of β-catenin; moreover, it inhibits tumor formation in nude mice. These data support a function for PTP LAR in the regulation of epithelial cell-cell contacts at adherens junctions as well as in the control of β-catenin signaling functions. Thus PTP-LAR appears to play an important role in the maintenance of epithelial integrity, and a loss of its regulatory function may contribute to malignant progression and metastasis.

The cadherins represent a family of transmembrane receptors that mediate homophilic, Ca 2ϩ -dependent cell-cell adhesion. In epithelial cells, the members of this family, such as the classical E-, N-, and P-cadherins, are primarily found at the adherens junctions of adjacent cells (1). ␤-Catenin as well as plakoglobin (␥-catenin) associate directly with the highly conserved cytoplasmic domain of classical cadherins in a mutually exclusive manner (2,3). The cadherin-catenin complex is linked via ␣-catenin either directly (4) or indirectly to the actin filament network via the actin-binding proteins ␣-actinin or vinculin (5,6). The association of the cadherin-catenin complex with the cytoskeleton is essential for tight cell-cell interaction.
Nevertheless, cadherin/catenin-mediated cell-cell contacts have to be highly dynamic because, particularly during embryonic development or wound healing, adherens junctions have to be rapidly disassembled and reassembled (7). Down-regulation of cadherins results in the separation of neighboring cells, a phenomenon that is observed during embryonic development at the epithelial-mesenchymal transition (EMT) 1 of forming mesoderm (8) as well as in tumor cells, allowing their invasion and dissemination throughout the body (9). During epithelialmesenchymal transition, cells transiently lose their epithelial features and acquire a fibroblastoid morphology (10). The critical importance of an intact cadherin-catenin complex is underscored by the observation that down-regulation of any of its components resulting in the loss of the tumor-suppressive actions of adherens junctions correlates with tumor invasion and metastasis (11). Moreover, the integrity of adherens junctions appears to be dynamically regulated by tyrosine phosphorylation. Transfection of a v-src oncogene (12,13) or treatment with growth factors (14,15) causes unstable cell-cell adhesion and migration of cells, and inhibition of PTPs enhances this destabilizing effect (16). The model in which reversible tyrosine phosphorylation serves to regulate cadherin-mediated cell-cell adhesion is further supported by the demonstration of cadherin-catenin complex association with the receptor tyrosine kinases EGF receptor and human EGF receptor 2/Neu (17,18) as well as with the transmembrane PTPs , , and (19 -21).
␤-Catenin and plakoglobin are mammalian homologues of the Drosophila protein Armadillo, whose function is critical for normal segmental pattern formation during development (22). The presence of a repeating 42-amino acid sequence motif defines members of the "armadillo family" (23). Data obtained in the Drosophila and Xenopus systems suggest an additional function for ␤-catenin independent of cadherin-mediated cell adhesion. This involves translocation of ␤-catenin to the nucleus that is preceded by its accumulation in the cytoplasm. Thus, free ␤-catenin is involved in transcriptional regulation of specific genes that are essential for embryonic development (24). The signals resulting in a free pool of ␤-catenin include the binding of Wingless/Wnt to its transmembrane receptor Frizzled and the inhibition of the serine/threonine kinase Zeste-White 3 (Shaggy) or its vertebrate homologue glycogen synthase kinase 3 (25). Further functions for ␤-catenin and plakoglobin are indicated by their association with the adenomatous polyposis coli (APC) tumor suppressor protein, which is thought to serve as a cytoplasmic effector of ␤-catenin, negatively regulating the accumulation of cytosolic ␤-catenin in concert with glycogen synthase kinase 3 (26) and axin/conductin (27)(28)(29) by inducing ubiquitin-dependent degradation of ␤-catenin (30).
In this report we show that during epithelial migration, ␤-catenin accumulates in the cytosol in a free, uncomplexed and tyrosine-phosphorylated form. However, in confluent cells, ␤-catenin maintains epithelial cell integrity as an essential part of the cadherin-catenin tumor suppressor system. Our findings suggest that tyrosine phosphorylation regulates the function of ␤-catenin as a signaling molecule during epithelial cell migration. We further demonstrate PTP LAR to be a modulator of epithelial cell migration, which strongly supports a function of this protein tyrosine phosphatase in the regulation of cell-cell contacts and epithelial cell integrity. Moreover, we show that ectopic expression of PTP LAR inhibits tumor formation in nude mice. A dysfunction of PTP LAR may therefore lead to tumor invasion and metastasis.

MATERIALS AND METHODS
Cloning of ␣-Catenin, ␤-Catenin, and Plakoglobin (␥-Catenin)-Human ␣-catenin (accession number D14705), ␤-catenin (Z19054), and Plakoglobin (M23410) were amplified from cDNA generated from MCF7 cells by PCR. PCR products were cloned in a eukaryotic expression vector under the control of the cytomegalovirus promoter and confirmed by sequence analysis.
Cell Lines and Cell Culture-All cell lines were obtained from the American Tissue Culture Collection. NBT II cells (CRL-1655) were grown in minimal essential medium supplemented with 1% nonessential amino acids, Earle's balanced salt, 2 mM glutamine, 1 mM sodium pyruvate, and 10% fetal calf serum (Sigma). MCF7 cells (HTB22) were grown in RPMI medium supplemented with 2 mM glutamine and 10% fetal calf serum, human embryonic 293 kidney cells (CRL 1573) in Dulbecco's modified Eagle's medium supplemented with 1 mg/ml glucose, 2 mM glutamine, and 10% fetal calf serum. Fetal calf serum was routinely heat-inactivated. All media and supplements were obtained from Life Technologies, Inc.
Migration Assays-For in vitro wound assays, NBT II cells were plated at a density of 7 ϫ 10 4 cells/cm 2 . After 24 h, the confluent monolayer was scratched with a pipette tip to create a cell-free area, growth factors were added, and wound closure was documented by photography. For scatter assays, cells were plated at a cell density of 1 ϫ 10 4 cells/cm 2 . Growth factors were added after 24 h, and the morphology of cells and the dispersion of small colonies were documented by photography. Quantification of migration was performed by counting single cells with fibroblastoid migration morphology compared with cells in groups with epithelial morphology of different randomly chosen microscopic fields. 1000 cells/dish were counted. All migration assays were performed in triplicates. Tyrosine kinases (TK), mitogenactivated protein kinase, or phosphatidylinositol 3-kinase were inhibited by genistein (250 M; Sigma), PD98059 (50 M; Biolabs), or LY295002 (20 M; Biomol), respectively, added 1 h before the addition of the growth factors. As a control, the solvent Me 2 SO was used. DNA synthesis was inhibited by the specific inhibitor of DNA polymerase ␣ aphidicolin (1, 5 M; Serva) 1 h before adding the growth factors.
Tumor Formation in Nude Mice-NBT II cells were resuspended at a cell density of 2 ϫ 10 6 /50 l in minimal essential medium and injected subcutaneously into the flank region of Swiss nude mice (IGR Villejuif, Paris, France). Tumor formation was monitored by measuring the width (W) and length (L) of the tumors with W Ͻ L. The tumor volume was calculated according to the formula (W 2 ϫ L ϫ /6). Assays were performed at least in triplicates.
Generation of Recombinant Retroviruses and Retrovirus-mediated Gene Transfer-Full-length PTP LAR (H. Saito, Harvard Medical School, Boston, MA) was subcloned into pLXSN vector (31). Stable NBT II cell lines were generated by retroviral gene transfer as described (32). Polyclonal and clonal cell lines were selected in medium containing 0.5 mg/ml G418 (Life Technologies, Inc.). Ectopic expression was confirmed by immunoprecipitation and Western blot analysis.
Transient Expression, Cell Lysis, and Immunoprecipitation-Transient transfection of human 293 embryonic kidney cells was performed as described (33). For biochemical analysis, NBT II cells were plated at a density of 1 ϫ 10 4 cells/cm 2 at the day before lysis. When indicated, cells were pretreated before lysis with sodium orthovanadate (1 mM) for the indicated period of time. After washing with ice-cold phosphatebuffered saline (PBS), cells were lysed in ice-cold lysis buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM EDTA, 10% glycerol, 20 mM pyrophosphate, 1% Triton X-100, 100 mM NaFl, 2 mM phenylmethylsulfonyl fluoride, 20 g/ml aprotinin, 20 g/ml leupeptin, 0.7 g/ml pepstatin, 0.2 mM ammonium molybdate, 2 mM sodium orthovanadate) and precleared by centrifugation at 12,500 ϫ g for 10 min at 4°C. The protein concentrations of the supernatants were adjusted to be equal. HNTG buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 10 mM pyrophosphate, 10 mM NaFl, 0.2 mM ammonium molybdate, 10% glycerol, 0.1% Triton X-100, 2 mM sodium orthovanadate) was added in a 1:1 ratio, and immunoprecipitations were performed for 2 h at 4°C. Protein A-or protein G-Sepharose was added for an additional 2 h. Precipitates were washed three times with HNTG buffer, and beads were resuspended in SDS sample buffer. For subsequent Western blot analysis, proteins separated by SDS-PAGE were transferred to nitrocellulose (Schleicher and Schuell) and incubated with the respective antibody. Proteins were visualized by the ECL system (Amersham Pharmacia Biotech). Before reprobing, blots were stripped by incubation for 1 h in 68 mM Tris-HCl, pH 6.8, 2% SDS, and 0.1% ␤-mercaptoethanol at 50°C.
In Vitro Binding Assays-The plasmid coding for the GST-hPTP LAR i fusion protein was constructed by PCR amplification of the cDNA sequence between amino acids 1259 and 1881 of human PTP LAR and cloned into the appropriate pGEX vector (Amersham Pharmacia Biotech). In vitro mutagenesis (34) of the original vector pSP65-LAR yielded PTP LAR with either inactivated PTP domain 1 PTP LAR D 1 C1522S or inactivated PTP domain 2 PTP LAR D 2 C1813S, which were also amplified by PCR between amino acids 1259 and 1881 of human PTP LAR and subcloned into pGEX vector. The integrity of the subcloned PCR products was confirmed by sequence analysis. GST fusion proteins were expressed in Escherichia coli and purified as described (35). 3 g of GST-hPTP LAR i fusion protein or a 3-fold molar excess of GST were incubated at 4°C with equal amounts of cell lysates, immobilized on glutathione-Sepharose (Sigma), and washed three times with HNTG. Bound proteins were separated by SDS-PAGE for Western blotting.
Affinity Precipitation-The plasmid coding for the GST-E-cadherin cytoplasmic fusion protein was constructed by amplification of the cDNA sequence between amino acids 734 and 884 of murine E-cadherin (36) by PCR and cloned into the appropriate pGEX vector. For affinity precipitation, equal concentrations of precleared NBT II cell lysates were incubated with 5 g of purified GST-E-cadherin cytoplasmic fusion protein or a 3-fold molar excess of GST and immobilized on glutathione-Sepharose. The resulting complexes were washed three times with HNTG, and bound proteins were separated by SDS-PAGE for Western blotting.
Antisera-Monoclonal antibody against phosphotyrosine (4G10) was obtained from UBI, and ␣-catenin, ␤-catenin, plakoglobin, and E-cadherin antibodies were obtained from Transduction Laboratories. A second antibody against E-cadherin was raised against a GST fusion protein containing amino acids 834 -913 of human E-cadherin. Monoclonal antibody 11.1A (M. Streuli, Dana Farber Cancer Institute, Boston, MA) recognizes the extracellular domain of human PTP LAR (37). Rabbit antiserum 320 (Y. Schlessinger, New York University Medical Center, New York) is directed against a peptide corresponding to the C terminus of PTP LAR (amino acids 1868 -1881).
In Vitro Dephosphorylation Assay-␤-Catenin was transiently overexpressed in human 293 embryonic kidney cells. Cells were serumstarved, treated with pervanadate (0.3 M H 2 O 2 , 0.1 mM sodium orthovanadate) for 10 min, and lysed. Immunoprecipitations were performed as described above, except that HNTG without pyrophosphate, ammonium-molybdate, and sodium orthovanadate was used. PTP activity toward tyrosine-phosphorylated ␤-catenin was assayed in 200-l reactions at 25°C containing 25 mM HEPES, pH 7.5, 5 mM EDTA, 10 mM dithiothreitol, and 1 mg/ml bovine serum albumin with 200 ng of GST-hPTP LAR cytoplasmic, GST-hPTP LAR D 1 C1522S, or GST-hPTP LAR D 2 C1813S added. Reactions were stopped by washing with HNTG and subsequently separated by SDS-PAGE. The tyrosine phosphorylation status was analyzed by Western blotting using antiphosphotyrosine antibody.
Immunofluorescence-NBT II cells were plated at a density of 1 ϫ 10 4 cells/cm 2 or 7 ϫ 10 4 cells/cm 2 . 24 h later, cells were fixed with 2% paraformaldehyde in PBS (pH 7.4, 125 mM sucrose). Autofluorescence was quenched with PBS glycine (100 mM glycine, 0.1% borohydrate in PBS), and the cells were permeabilized with 0.5% saponin in PBS (5 min). Nonspecific antibody binding was blocked for 1 h with phosphatebuffered gelatin (PBS, 0.5% bovine serum albumin, 0.045% cold water fish gelatin, 5% donkey serum). Primary antibody incubation was performed at room temperature for 2 h after dilution in phosphate-buffered gelatin. After three washes in phosphate-buffered gelatin, primary antibody binding was detected with isotype specific secondary antibody, either fluorescein isothiocyanate(DTAF)-or Cy3-conjugated (The Jackson Laboratories). For double-labeling experiments, antibody decoration was performed consecutively. Coverslips were mounted under Permafluor mounting medium (Immunotech, France) and viewed either with a conventional fluorescence microscope (Leica, FRG) or with a CLSM laser confocal microscope (Leica, FRG). Controls were recorded at identical settings.

Relocalization of the Cadherin-Catenin Complex upon Induction of Epithelial Cell Migration-Migration of epithelial cells
in tissue culture represents in many aspects a model system of the epithelial-mesenchymal transition (38). We chose for our studies the rat bladder carcinoma cell line NBT II (39) because certain growth factors and components of the extracellular matrix induce migration of these cells (40 -42). Results of an in vitro wound assay are shown in Fig. 1A for EGF and the commercial serum substitute Ultroser G. To rule out proliferation as the cause of wound closure, we used aphidicolin to inhibit DNA synthesis, which did not interfere with migration (data not shown). In contrast, inhibition of phosphatidylinositol 3-kinase by LY295002, TK by genistein, or mitogen-activated protein kinases by PD98059 completely abolished cell migration (data not shown). The protein levels of the components of the cadherin-catenin complex remained unchanged during migration (data not shown), and even overexpression of E-cad-herin could not prevent disruption of intercellular contacts (43). However, the localization of members of the cadherincatenin complex during EMT was altered. E-cadherin (green fluorescence) and ␤-catenin (red fluorescence) were located along the entire cell-cell contact region of adjacent, subconfluently plated cells (Fig. 1B, upper panel). Fluorescence on contact-free membrane portions was only weak or absent. Some weak ␤-catenin staining was detected in the nucleus of control cells, most probably because of the fact that these subconfluent cells did not establish their adherens junctions completely. However, upon induction of migration by EGF, E-cadherin and ␤-catenin redistributed over the entire cell surface and into the cytoplasm. Interestingly, increased nuclear localization of ␤-catenin was also detectable during epithelial cell migration.
Tyrosine Phosphorylation of the Cadherin-Catenin Complex-Because tyrosine phosphorylation seems to be involved in the regulation of adherens junctions (12,13), we investigated the specificity of tyrosine kinases on the cadherin-catenin complex. Transient cotransfection of TKs with members of the cadherin-catenin complex demonstrated that only TKs that play a role in epithelial cell migration, such as EGF receptor, c-Src, or fibroblast growth factor receptor 2, are capable of phosphorylating ␤-catenin and plakoglobin. However, ␣-catenin and E-cadherin were not substrates of these TKs. These observations could be confirmed in nontransfected NBT II cells stimulated to migrate. After cell lysis, all members of the cadherin-catenin complex were immunoprecipitated with specific antibodies. This technique allows the precipitation of the members of the cadherin-catenin complex regardless of their localization and binding partners and, therefore, also allows precipition of E-cadherin-bound ␤-catenin. Tyrosine phosphorylation levels were analyzed by Western blotting with an anti-phosphotyrosine antibody (Fig. 2, top) followed by the detection of E-cadherin (Fig. 2, middle) or the catenins (Fig. 2, bottom) by specific antibodies. Tyrosine phosphorylation of ␤-catenin and plakoglobin, but not of ␣-catenin and E-cadherin, was detectable already 30 min after induction of migration. The phosphorylation was transient, lasted for about 9 h, and was no longer detectable after 24 h (data not shown). For immunoprecipitation, pretreatment of the cells with 1 mM sodium orthovanadate, a specific inhibitor of PTPs, was necessary to detect tyrosine phosphorylation. This indicated that the phosphorylation state of ␤-catenin and plakoglobin was tightly regulated by TKs and PTPs. The correlation of epithelial cell colony dispersion with tyrosine phosphorylation of ␤-catenin and plakoglobin could also be demonstrated in HT29 and Ha-CaT cells (data not shown), suggesting a common regulatory mechanism during the induction of cell migration.
The Increased Free Pool of ␤-Catenin during Epithelial Cell Migration-We next examined the consequences of the ␤-catenin phosphorylation. To this end, ␤-catenin from EGF-treated cells was affinity-precipitated with a GST fusion protein comprising the entire cytoplasmic domain of E-cadherin, and the levels of free, uncomplexed ␤-catenin (Fig. 3, top) as well as its phosphotyrosine content (Fig. 3, bottom) were analyzed by immunoblotting. As previously demonstrated, this strategy allows, in contrast to immunoprecipitation, specifically and selectively the precipitation of the free, non-E-cadherin-bound pool of ␤-catenin (44). The induction of migration by EGF correlated with an increase of the free pool of ␤-catenin, which had a significantly higher phosphotyrosine content than ␤-catenin obtained by immunoprecipitation (compare Fig. 2 with Fig. 3). The free pool of ␤-catenin in nonstimulated control cells was unaffected. Although we could detect increases in free, tyrosine-phosphorylated ␤-catenin even without adding sodium orthovanadate (see Fig. 8), we added it in this set of experiments to improve detection. These findings indicate that phosphorylation by TKs leads to an increase of the free, uncomplexed pool of ␤-catenin during epithelial cell migration.
Colocalization of the Cadherin-Catenin Complex with PTP LAR-Pretreatment with sodium orthovanadate led to an increased tyrosine phosphorylation of ␤-catenin and plakoglobin in migrating NBT II cells. PTPs may therefore act as steady state equilibrium antagonists of TKs for regulatory events at adherens junctions. Because the transmembrane PTPs and were shown to associate with the cadherin-catenin complex (19,20), we extended the search for other PTPs at adherens junctions. As shown in Fig. 4, PTP LAR (red fluorescence) colocalized with the cadherin-catenin complex (green fluorescence) as indicated by the yellow fluorescence signal at adherens junctions of epithelial cells. PTP LAR was shown to be localized at focal adhesions of epithelial MCF7 cells (45). In NBT II cells, we detected PTP LAR predominantly at adherens junctions but also at focal adhesions; however, with a lower signal intensity. Because motile cells have to rapidly disassemble and reassemble adherens junctions and focal adhesions, the localization of PTP LAR supports its potential regulatory function during epithelial cell migration.
Association of ␤-Catenin and Plakoglobin with PTP LAR-To investigate the potential association of PTP LAR with the cadherin-catenin complex in intact cells, subconfluent human MCF7 cells were stimulated with EGF, lysed, and immunoprecipitated with antibodies either against PTP LAR or the cadherin-catenin complex. Western blotting analysis with specific antibodies to ␤-catenin and E-cadherin (Fig. 5A) or plakoglobin and E-cadherin (Fig. 5B) detected these proteins in anti-human PTP LAR immunoprecipitates with the monoclonal antibody 11.1A against the extracellular domain of human PTP LAR and vice versa. Interestingly, the association was constitutive and independent of EGF stimulation. No specific signal of members of the cadherin-catenin complex was obtained with the rabbit polyclonal antiserum 320 to PTP LAR, suggesting that the antigenic epitope was inaccessible in the complex. To investigate which component of the cadherin-catenin complex mediated the interaction with PTP LAR, we used a GST fusion protein comprising the entire cytoplasmic domain of PTP LAR (GST-PTP LAR i ) to affinity purify the individual components of the cadherin-catenin complex. The members of the cadherincatenin complex were individually and transiently overexpressed in human 293 embryonic kidney fibroblasts that had been treated with the tyrosine phosphatase inhibitor pervanadate before lysis. ␤-Catenin (Fig. 5C, left) and plakoglobin (Fig.   FIG. 2. Tyrosine phosphorylation of ␤-catenin and plakoglobin during epithelial cell migration. NBT II cells were plated at a density of 1 ϫ 10 4 cells/cm 2 . 24 h later, growth factors were added to induce migration (EGF, 100 ng/ml; acidic fibroblast growth factor (aFGF), 30 ng/ml including 50 g/ml heparin; Ultroser G: 2%). 90 min before lysis sodium orthovanadate (1 mM) was added (right) or not (left). As a positive control, cells were treated for 10 min with pervanadate. After lysis the cadherin/catenin complex was immunoprecipitated and precipitates were separated by SDS-PAGE. Tyrosine phosphorylation levels were analyzed by Western blotting with an anti-phosphotyrosine antibody (top). The membranes were reprobed with specific antibodies (Ab) to E-cadherin (middle) or the catenins (bottom) to assure equal amounts of precipitated proteins. Reproducibly it was observed that highly tyrosine-phosphorylated ␤-catenin (as here after pervanadate treatment) was not efficiently immunodecorated in the reblot. Arrows indicate the proteins of interest, and molecular mass standards are shown in kilodaltons on the left. IP, immunoprecipitation. 5C, right) were found to associate specifically with the cytoplasmic domain of PTP LAR under these conditions. E-cadherin and ␣-catenin did not interact with PTP LAR directly under the same experimental conditions (data not shown). The phosphorylation state of ␤-catenin or plakoglobin, which were both tyrosine-phosphorylated after treatment with pervanadate did not affect their association with the GST-PTP LAR cytoplasmic fusion protein. This interaction required the complete cytoplasmic domain of PTP LAR, because deletion mutants of PTP LAR did not bind ␤-catenin or plakoglobin efficiently (data not shown).
␤-Catenin Is a Substrate of PTP LAR in Vitro-The association of ␤-catenin and plakoglobin with PTP LAR prompted us to investigate whether these proteins could represent actual sub-  transiently overexpressing, pervanadate-treated human 293 cells. After different time intervals, the reactions were terminated and analyzed with an anti-phosphotyrosine-specific antibody. A significant reduction in the tyrosine phosphorylation signal within the first 5 min after incubation of ␤-catenin with GST-hPTP LAR cytopl. or GST-hPTP LAR PTP D 2 C1813S (Fig. 6, top, left, and right) was revealed. No change in phosphorylation levels was observed after incubation with GST-hPTP LAR PTP D 1 C1522S (Fig. 6, top and middle), confirming results of enzymatic activity measurements of the GST fusion proteins using p-nitrophenyl phosphate as a substrate (data not shown). These data are consistent with previous findings that the first PTP domain of PTP LAR is essential for catalytic activity, whereas the second is catalytically inactive (46). Blots were reprobed with an anti-␤-catenin-specific antibody to confirm that equal amounts of protein were used in the assay (Fig.  6, bottom). We therefore conclude that ␤-catenin is a substrate for PTP LAR.
PTP LAR Inhibits Epithelial Cell Migration and Tumor Formation in Nude Mice-We next examined whether PTP LAR has a direct regulatory function during epithelial cell migration. We therefore ectopically expressed human PTP LAR in NBT II cells at levels comparable with the endogenous protein, thereby yielding polyclonal as well as clonal cell lines with about twice the PTP LAR expression relative to the parental cells (data not shown). With these cell lines we performed scatter assays to quantify migration after EGF treatment. This modest ectopic expression of hPTP LAR in NBT II cells significantly inhibited EMT and motility (Fig. 7A) to about 40% that of the vector control-infected cell lines (NBT II pLXSN, Fig. 7B) without affecting the kinetics of the onset of migration. No differences in activation and autophosphorylation of the EGF receptor and its association with the adapter protein Shc were detected (data not shown), suggesting that ectopic hPTP LAR does not function by inactivating the EGF receptor. Furthermore, downstream events of EGF signaling like DNA synthesis and proliferation rate of these NBT II-hPTP LAR cell lines were not affected by ectopic expression of hPTP LAR (data not shown).
The epithelial cell migration in vitro resembles a simplified model system of tumor formation and metastasis in vivo. Therefore we tested whether ectopic expression of hPTP LAR correlated also with an decreased capability of these cells to form tumors in vivo in nude mice. To this end, we injected NBT II hPTP LAR cells subcutaneously into the flank region of Swiss nude mice. Interestingly, these cells displayed significantly reduced tumor growth in comparison to NBT II pLXSN cells (Fig. 7C). The capability to form tumors of polyclonal and clonal cell lines did not differ. The parental NBT II cells form tumors to the same extent as the vector control-infected cell lines, which rules out a clonal artifact (data not shown). These data strongly suggest that PTP LAR serves as a negative regulator of epithelial cell migration and tumor formation.
PTP LAR Inhibits Tyrosine Phosphorylation and the Increase of the Free Pool of ␤-Catenin-These data prompted us to investigate biochemical parameters, which correlated with the inhibition of EGF-induced migration in hPTP LAR-expressing NBT II cells. We therefore affinity-precipitated ␤-catenin in control and hPTP LAR-expressing NBT II cells using a GST-E-cadherin cytoplasmic fusion protein. Because we were interested in the function of PTP LAR, we did not treat the cells with the inhibitor of PTPs, sodium orthovanadate. Nevertheless, even under these conditions we were able to detect an increase in the free, uncomplexed and tyrosine-phosphorylated pool of ␤-catenin in vector-infected control cells, demonstrating that an increase in free, tyrosine-phosphorylated ␤-catenin occurred also in cells without sodium orthovanadate pretreatment (Fig.  8, left). However, in NBT II-hPTP LAR-expressing cells, neither tyrosine phosphorylation nor an increase in the free pool of ␤-catenin were detectable (Fig. 8, right). Because ␤-catenin is a substrate of PTP LAR in vitro and ectopic expression of human PTP LAR in NBT II cells does not interfere with EGF receptormediated signaling, we suggest that PTP LAR represents a specific negative regulator of ␤-catenin tyrosine phosphorylation, which prevents an increase in free ␤-catenin, thereby inhibiting epithelial cell migration. DISCUSSION Cadherin-catenin complex-mediated cell-cell adhesion as well as adhesion-independent functions of catenins have been implicated in the modulation of multicellular differentiation, proliferation, and malignant transformation of epithelial cells (7,8,11). Cell migration is an essential process during embryonic development and in epithelial regeneration of adult organ- isms, for example in wound healing, and requires precise control, which is altered or lost when tumor cells become invasive and metastatic. In this study we have defined at the cellular and biochemical level distinct mechanisms regulating epithelial cell migration. Furthermore, we have characterized in motile cells the impact of tyrosine phosphorylation of ␤-catenin and its regulation by tyrosine kinases and the protein-tyrosine phosphatase LAR.
Growth factors such as EGF, acidic fibroblast growth factor, or hepatocyte growth factor/scatter factor have been shown to induce migration of epithelial cells (38). A correlation has been suggested between migration and tyrosine phosphorylation of ␤-catenin (12,14), but its significance remained unclear. Inhibition of tyrosine kinase signaling inhibits epithelial cell migration of NBT II cells. Receptor tyrosine kinases like the EGF receptor (17) or cytoplasmic tyrosine kinases like Src are able to mediate tyrosine phosphorylation of ␤-catenin (13). In NBT II cells, a dominant-negative c-Src mutant was able to inhibit migration (47). We demonstrate here that after induction of migration, the pool of free, uncomplexed ␤-catenin is increased and that this increase correlates with enhanced ␤-catenin tyrosine phosphorylation. Therefore, tyrosine phosphorylation may result in a reduced interaction of ␤-catenin with both E-cadherin and the actin-cytoskeleton. Because the integrity of the cadherin-catenin complex is essential for strong cell-cell adhesion, this reduced interaction may lead to an overall decrease in intercellular contacts. Interestingly, a fusion protein of E-cadherin and ␣-catenin was reported to be able of mediating the interaction of E-cadherin with the cytoskeleton independent of ␤-catenin. However, these cells were no longer capable of migration (48). In light of these data, tyrosine phosphorylation of ␤-catenin may lead to disruption of the contact between E-cadherin and the cytoskeleton and to an increased pool of free ␤-catenin. This suggests a function for ␤-catenin independent of cadherin-mediated cell adhesion.
Besides their role at adherens junctions, a signaling function of ␤-catenin or its Drosophila homologue Armadillo was shown to be essential for normal embryonic development in Drosophila and Xenopus. Interference with the signaling function of free, uncomplexed ␤-catenin abolished proper vertebrate development (24). Subsequent studies in Drosophila and Xenopus led to the discovery of a signaling cascade that regulates the cytoplasmic pool of ␤-catenin. Without a signal, ␤-catenin is localized mainly at adherens junctions, and any free ␤-catenin is down-regulated in a ubiquitin-dependent manner (30). An extracellular signal such as Wnt or Wingless (49,50) leads via the receptor Frizzled (51) to an inhibition of glycogen synthase kinase 3/Zeste White 3 activity, thus stabilizing ␤-catenin/ Armadillo in its free form (44,52) by inhibiting APC-and ubiquitin-dependent degradation (30,53). We could show an increased pool of free ␤-catenin in the cytoplasm and the nucleus in migrating cells after EGF treatment, suggesting another way of regulating the free pool of ␤-catenin during epithelial cell migration beside Wnt/Wingless signaling. Tyrosine phosphorylation of ␤-catenin may also be able to stabilize free ␤-catenin, because it was shown for the homologous plakoglobin that the tyrosine-phosphorylated form did no longer associate with APC (54), thereby possibly preventing APC-mediated degradation. Indeed, we were able to demonstrate that the free pool of ␤-catenin has an increased phosphotyrosine content during EGF-dependent epithelial cell migration. Additionally, EGF is also able to inhibit glycogen synthase kinase 3 activity (55), and tyrosine phosphorylation of ␤-catenin was shown to correlate with carcinoma formation and tumor invasion (56). Only free ␤-catenin or plakoglobin is able to associate with members of the high mobility group (57) family of transcription factors, namely LEF1, T-cell factor 3, and T-cell factor 4 (58 -62). However, recently it was demonstrated that ␤-catenin and plakoglobin differ in their nuclear translocation and complexing with LEF1 and that LEF1-dependent transactivation is preferentially driven by ␤-catenin (63). The transcription factors LEF1 and T-cell factor are able to induce dorsal mesoderm when expressed together with ␤-catenin in Xenopus embryos (58 -60). Data from LEF1-deficient and LEF1-transgenic mice demonstrate an important role of this transcription factor during EMT, because it is normally up-regulated during EMT and inductive processes between mesenchyme and epithelium (64 -66). It is tempting to speculate that a common regulatory system exists that coordinates the expression of genes for mesenchymal and epithelial phenotypes during embryonic development as well as in epithelial cell migration. During EMT or cell migration this common regulator would switch off expression of epithelial genes while switching on genes for the mesenchymal phenotype. The ␤-catenin⅐LEF1 complex may represent this common regulator, and molecules that control the free pool of ␤-catenin may be essential for proper development or wound healing.
PTPs were proposed to play an important role in the regulation of cell-cell contacts because treatment with sodium orthovanadate, a potent inhibitor of phosphatase activity, diminished normal cell contact inhibition in epithelial cells and led to increased tyrosine phosphorylation at adherens junctions (16). PTPs may therefore act as steady state equilibrium antagonists of TKs for regulatory events at adherens junctions. Consistent with these findings we show that tyrosine phosphorylation of ␤-catenin and plakoglobin in migrating epithelial cells was significantly increased after pretreatment with sodium orthovanadate. Little is known about interacting proteins or in vivo substrates of transmembrane PTPs. PTP LAR was recently shown to interact and colocalize at focal adhesions with LAR interacting protein 1 (LIP-1) and the multidomain protein Trio. However, neither of these proteins appears to be a sub- FIG. 8. Ectopic expression of human PTP LAR in NBT II cells inhibits the phosphorylation in tyrosine residues of ␤-catenin as well as the increase of the free pool of ␤-catenin. NBT II cells infected with the empty vector pLXSN (left) or human PTP LAR (right) were plated at 1 ϫ 10 4 cells/cm 2 and stimulated after 24 h with EGF (100 ng/ml) for the indicated time intervals. Sodium orthovanadate pretreatment was omitted to avoid irreversible inhibition of protein-tyrosine phosphatase activity. Cells were lysed, and in vitro associations were performed with 5 g of the GST-E-cadherin cytoplasmic (cytopl.) protein or GST in a 3-fold molar excess. Proteins were separated by SDS-PAGE, and the levels of free, uncomplexed ␤-catenin were analyzed by Western blotting with an anti-␤-catenin antibody (top). Western blotting with an antiphosphotyrosine antibody shows the increased tyrosine phosphorylation in free, uncomplexed ␤-catenin only in the control-infected NBT II cell line (bottom). Arrows indicate the proteins of interest, and molecular mass standards are shown in kilodaltons on the left. strate for PTP LAR (45,67). Moreover, a PTP LAR-like PTP was reported to interact with the cadherin-catenin complex at adherens junctions of neurosecretory PC12 cells (68). We demonstrate here the colocalization and interaction of PTP LAR with the cadherin-catenin complex in epithelial cells. Furthermore, we show that only ␤-catenin and plakoglobin are able to associate directly with PTP LAR, and we present evidence that ␤-catenin is a substrate of PTP LAR. PTP LAR is of special interest because it is localized at adherens junctions and at focal adhesions (this report and Refs. 45,67,68); thus PTP LAR could represent an essential regulator of the disassembly and reassembly of cell-cell as well as cell-extracellular matrix adhesions during epithelial cell migration. We demonstrate that modest ectopic overexpression of PTP LAR significantly inhibited epithelial cell migration. A similar situation is found in Drosophila, where PTP LAR, contrary to vertebrates, is almost exclusively expressed in developing neurons. In flies lacking PTP LAR, motor axons bypass their normal target region and instead continue to extend without stopping (69). PTP LAR, PTP , PTP as well as PTP DEP-1 were found to be expressed at elevated protein levels with increased cell confluence (20, 70 -72), thereby contributing to the observed increased tyrosine phosphatase activity in confluent cells (73). The increase in phosphatase activity in confluent cells suggests a role of PTPs in the regulation and stabilization of cell-cell contacts and epithelial cell integrity. In contrast, during wound healing, cells at the wound edge are in a subconfluent situation, where PTP action is reduced. This could favor tyrosine kinase signaling through stimulation by growth factors such as EGF, transforming growth factor ␣, and keratinocyte growth factor, which results in an increase in cell migration and proliferation and, thus, to accelerated wound healing, because the expression of these growth factors was increased during a wound situation (74,75,76).
The importance of free and tyrosine-phosphorylated ␤-catenin during epithelial cell migration is underscored by our observations that interfering with this parameter by overexpressing hPTP LAR inhibits epithelial cell motility. Furthermore, expression of hPTP LAR in NBT II cells inhibited tumor formation in nude mice, although the growth characteristics of these cells in vitro were not altered. Ectopic expression of hPTP LAR to about twice the endogenous level was sufficient to result in the biological effects observed. Such modest ectopic expression of hPTP LAR appears to be critical, because high overexpression results in completely altered growth characteristics and apoptosis of the cells (77). The data presented in this report strongly support an important role for PTP LAR in the regulation of cell-cell adhesion and epithelial cell migration as well as tumorigenesis by controlling the free pool of signaling ␤-catenin. Because mutations or deletions in either APC or ␤-catenin leading to a stabilized pool of free ␤-catenin have been correlated with tumor formation (61,62,78), loss of PTP LAR function may also contribute to tumor formation and metastasis. As a potential tumor suppressor gene product, PTP LAR could therefore serve as a prognostic marker for human cancer.