Expression of the Receptor Protein-tyrosine Phosphatase, PTPμ, Restores E-cadherin-dependent Adhesion in Human Prostate Carcinoma Cells*

Normal prostate expresses the receptor protein-tyrosine phosphatase, PTPμ, whereas LNCaP prostate carcinoma cells do not. PTPμ has been shown previously to interact with the E-cadherin complex. LNCaP cells express normal levels of E-cadherin and catenins but do not mediate either PTPμ- or E-cadherin-dependent adhesion. Re-expression of PTPμ restored cell adhesion to PTPμ and to E-cadherin. A mutant form of PTPμ that is catalytically inactive was re-expressed, and it also restored adhesion to PTPμ and to E-cadherin. Expression of PTPμ-extra (which lacks most of the cytoplasmic domain) induced adhesion to PTPμ but not to E-cadherin, demonstrating a requirement for the presence of the intracellular domains of PTPμ to restore E-cadherin-mediated adhesion. We previously observed a direct interaction between the intracellular domain of PTPμ and RACK1, a receptor for activated protein kinase C (PKC). We demonstrate that RACK1 binds to both the catalytically active and inactive mutant form of PTPμ. In addition, we determined that RACK1 binds to the PKCδ isoform in LNCaP cells. We tested whether PKC could be playing a role in the ability of PTPμ to restore E-cadherin-dependent adhesion. Activation of PKC reversed the adhesion of PTPμWT-expressing cells to E-cadherin, whereas treatment of parental LNCaP cells with a PKCδ-specific inhibitor induced adhesion to E-cadherin. Together, these studies suggest that PTPμ regulates the PKC pathway to restore E-cadherin-dependent adhesion via its interaction with RACK1.

A diverse set of cellular behaviors including growth, differentiation, adhesion, and migration are regulated by protein tyrosine phosphorylation. Protein tyrosine kinases and proteintyrosine phosphatases (PTPs) 1 regulate intracellular phospho-tyrosine levels. A subfamily of receptor-like PTPs (RPTPs) has extracellular segments containing adhesion molecule-like domains and intracellular segments that possess tyrosine phosphatase activity (1,2). This structural arrangement suggests that RPTPs directly send signals in response to cell adhesion.
The receptor protein-tyrosine phosphatase PTP is a member of the Ig superfamily of adhesion molecules. The extracellular segment of PTP contains a MAM ( (Meprin/A5/PTP ) domain, an Ig domain, and four fibronectin type III repeats (3). Expression of PTP induces aggregation of non-adherent cells (4,5) through a homophilic binding site that resides within the Ig domain (6). The MAM domain plays a role in cell-cell aggregation by determining the specificity of the adhesive interaction (7). PTP contains a single membrane-spanning region with two cytoplasmic PTP domains. Only the membrane proximal PTP domain has catalytic activity (8). The role of the membrane distal PTP domain is not known, but this domain has been implicated in directing protein-protein interactions in other RPTPs (reviewed in Ref. 2). The intracellular juxtamembrane domain of PTP contains a region that is homologous to the conserved intracellular domain of the cadherins (9).
Cadherins are a family of calcium-dependent adhesion molecules that play an essential role in the formation of the cellcell contacts termed adherens junctions (10). Cadherin-dependent adhesion is important for many physiological processes including establishment of cell polarity, morphogenetic movements such as epithelial/mesenchymal transitions, and cell type sorting during development (10,11). Cadherins interact with the actin cytoskeleton via binding of the cytoplasmic domain to catenins (12). The catenins include ␣, ␤, ␥/plakoglobin, and p120. ␤and ␥-catenin bind directly to the cytoplasmic segment of cadherin, whereas ␣-catenin binds to ␤or ␥-catenin thereby linking the cadherin-catenin complex to the cytoskeleton. Deletions in the catenin-binding region of cadherins disrupt cadherin-mediated adhesion despite the presence of an intact extracellular segment (12).
Despite the importance of cadherin-mediated cell-cell adhesion, the underlying mechanisms that regulate adhesion are still poorly understood. PTP has been shown to associate with the cadherin-catenin complex (13)(14)(15). Specifically, PTP interacts with a number of classical cadherins including E-cadherin, N-cadherin, and cadherin 4 (also called R-cadherin) (14). The classical cadherins have a highly conserved cytoplasmic domain, and PTP has been shown to bind directly to the C-terminal 38 amino acids of the intracellular domain of Ecadherin, which is the likely binding site in the other classical cadherins as well (14). We have shown that PTP regulates N-cadherin-mediated neurite outgrowth (16). In fact, expression of a catalytically inactive form of PTP perturbed Ncadherin-mediated neurite outgrowth. This demonstrates that the phosphatase activity of PTP is required for N-cadherinmediated signal transduction and/or regulation of the cytoskeleton in neurons.
In this study, we employed the LNCaP prostate carcinoma cell line (17) to investigate the role of PTP in E-cadherinmediated adhesion. Unlike normal prostate epithelial cells, LNCaP cells do not express endogenous PTP. Although these cells express the proteins in the cadherin-catenin complex, we found that they did not mediate E-cadherin-dependent adhesion in aggregation assays and in an in vitro adhesion assay. By using a retroviral/tetracycline-repressible system, we re-expressed wild type and mutant forms of PTP in LNCaP cells and tested their effect on cell adhesion. Our data indicate that the cytoplasmic domain is important for restoring E-cadherindependent adhesion regardless of catalytic activity. In a recent paper (18), we isolated RACK1 as a PTP-interacting protein using a two-hybrid screen. RACK1 is a scaffolding protein that was originally identified as a receptor for activated protein kinase C (19). Because RACK1 binds activated protein kinase C, we tested whether PKC may play a role in the ability of PTP to regulate E-cadherin-mediated adhesion. Data presented here suggest that the cytoplasmic domain of PTP regulates E-cadherin-mediated adhesion through modulating PKC via the PTP/RACK1 interaction.

EXPERIMENTAL PROCEDURES
Antibodies and Reagents-Monoclonal antibodies against the intracellular (SK7) and extracellular (BK2) domains of PTP have been described (4,6). A monoclonal antibody against ␥-catenin (5172) was kindly provided by Dr. Pamela Cowin (New York University). Monoclonal antibodies against chick L1 (8D9) were generated in our lab using hybridoma cells generously provided by Dr. Vance Lemmon (Case Western Reserve University, Cleveland, OH). Monoclonal antibodies against E-cadherin, p120, ␣and ␤-catenin, and RACK1 were purchased from Transduction Laboratories (Lexington, KY). A monoclonal antibody against vinculin and a monoclonal antibody against the extracellular domain of E-cadherin (DECMA) were purchased from Sigma. Goat anti-mouse IgG and IgM immunobeads were obtained from Zymed Laboratories Inc. Laboratories (San Francisco) or goat anti-mouse IgG immunobeads were alternatively obtained from Cappel (Costa Mesa, CA). Normal prostate epithelial cells were purchased from Clonetics (San Diego, CA). LY294002, PMA, rottlerin, chelerythrine chloride, GF109203X, and Gö6976 were purchased from Calbiochem. RPMI 1640 medium, SMEM medium, and laminin were obtained from Invitrogen. Fetal bovine serum was obtained from HyClone (Logan, UT). Tween 20 was obtained from Fisher. All other reagents were obtained from Sigma.
Construction and Expression of the PTP Retroviruses-The retroviral system used is a tetracycline-repressible ("tet-off") promoter-based system (20). By using the pBPSTR1 vector generously provided by Dr. Steven Reeves (Harvard Medical School, Charlestown, MA), the following constructs were generated: wild type PTP, the C-S mutant form of PTP, and PTP-extra. The wild type PTP plasmid (PTPWT) and the PTPC1095S (C-S) catalytically inactive mutant have been described previously (16). Briefly, the wild type PTP plasmid contained almost the entire coding sequence of PTP (base pairs 1-4350, i.e. it only lacked the last two amino acids and the stop codon). This was done to create an in-frame fusion with the green fluorescence protein at the C terminus (PTP-GFP). The mutant form of PTP is also GFPtagged and contains a cysteine to serine mutation at residue 1095. A construct containing the extracellular, the transmembrane, and 55 amino acids of the intracellular domains has been previously described (4) (Note: PTP-extra is not GFP-tagged.) This construct was subcloned into the tetracycline-regulatable retroviral vector, pBPSTR1. Replication-defective amphotrophic retroviruses were made by transfecting the PA317 helper cell line (ATCC CRL-9078) with the respective PTPcontaining plasmids. Control virus was generated by transfecting PA317 helper cells with the pBPSTR1 plasmid.
Expression and Purification of GST Fusion Proteins-An E-cadherin GST fusion protein construct containing amino acids 9 -139 of mouse E-cadherin was obtained from Dr. Robert Brackenbury (University of Cincinnati, Cincinnati, OH). The E-cadherin GST fusion protein was constructed by restriction digest of pBATEM2 with PvuII and HincII. The fragment was ligated into the SmaI site of pGEX-KG, which results in a fusion protein containing amino acids 9 -139 of E-cadherin with GST at the N terminus. The GST fusion protein construct for expression of the entire extracellular domain of PTP has been described previously (4). Expression of GST-tagged proteins in E. coli was induced by isopropyl-1-thio-␤-D-galactopyranoside. The bacteria were collected by centrifugation at 3000 ϫ g for 10 min and lysed in PBS containing 1% Triton X-100, 5 g/ml leupeptin, 5 g/ml aprotinin, and 1 mM benzamidine, sonicated, and centrifuged again at 3000 ϫ g for 10 min to remove debris. The supernatant was passed over glutathione-Sepharose beads (Amersham Biosciences) and washed, and the bound protein was eluted with 10 mM glutathione as described previously (4).
Tissue Culture and Retroviral Infection of LNCaP Cells-LNCaP cells (17) were grown in RPMI 1640 supplemented with 10% fetal bovine serum and 1 g/ml gentamicin at 37°C and 5% CO 2 . Cells were infected with retrovirus by the addition of Polybrene (5 g/ml) and virus-containing medium. The cells were incubated overnight at 37°C, and the medium was exchanged with normal culture medium. Five days after infection, the cells were checked for expression of the PTP proteins, which were tagged with the green fluorescent protein (GFP) by fluorescence microscopy.
Protein Extraction and Immunoblotting-LNCaP cells were rinsed once with PBS, and the cells were lysed in Triton-containing buffer (20 mM Tris, pH 7.6, 1% Triton X-100, 2 mM CaCl 2 , 1 mM benzamidine, 200 M phenylarsine oxide, 1 mM vanadate, 0.1 mM ammonium molybdate, and 2 l/ml protease inhibitor mixture) and scraped off the dish. In all experiments where RACK1 was co-immunoprecipitated, the cells were lysed in a buffer containing 20 mM Tris, pH 7.6, 1% Triton X-100, 50 mM NaCl, 1 mM benzamidine, 1 mM vanadate, and 2 l/ml protease inhibitor mixture. After incubation on ice for 30 min, the lysate was centrifuged at 14,000 rpm for 3 min, and the Triton-soluble material was recovered in the supernatant. The amount of protein was determined by the Bradford method using BSA as a standard. Lysates were boiled in equal volume of 2ϫ sample buffer, and the proteins were separated by 6 or 10% SDS-PAGE and transferred to nitrocellulose for immunoblotting as described previously (4).
Immunoprecipitations-Antibodies (5 g of IgG/IP or 1.25 g of IgM/IP) were incubated with goat anti-mouse IgG or IgM immunobeads, respectively, for 2 h at room temperature and then washed 3 times with PBS (9.5 mM phosphate, 137 mM NaCl, pH 7.5). Immunoprecipitates were prepared by incubating lysates containing either 250 g of protein (Fig. 5) or 400 g of protein ( Fig. 6) with antibody-coupled beads overnight at 4°C. The beads were washed extensively with lysis buffer, then boiled in sample buffer, and separated by 6 or 10% SDS-PAGE. One-fifth of the immunoprecipitate was loaded per lane. Proteins were transferred to a nitrocellulose membrane and immunoblotted as described (4).
Calcium-dependent Aggregation Assay-Aggregation assays were performed as described previously (21). Briefly, cells were trypsinized in the presence of calcium, which selectively preserved cadherins (22). Uninfected LNCaP cells or cells infected with either PTPWT or C-S mutant were rinsed twice in HCMF buffer (10 mM HEPES, pH 7.4, 137 mM NaCl, 5.4 mM KCl, 0.3 mM Na 2 HPO 4 ⅐7H 2 O, 5.5 mM glucose) containing 2 mM CaCl 2 and trypsinized into single cells by incubation with 0.04% trypsin in HMCF buffer supplemented with 2 mM CaCl 2 . The trypsin was inactivated by the addition of RPMI containing 10% serum. The cells were pelleted and resuspended in 2 ml of SMEM, followed by a 20-min incubation with 20 units/ml DNase on ice. Some of the cells re-expressing PTPWT were treated with 200 g/ml of an E-cadherin function-blocking antibody (DECMA) for an additional 20 min on ice. 2 ϫ 10 6 cells were added to scintillation vials containing HMCF buffer with a final concentration of 2 mM CaCl 2 . Where indicated, the CaCl 2 was substituted with 5 mM EDTA (final concentration). Aggregation was initiated by placing the vials at 37°C at 90 rpm in a gyratory shaker. Aliquots of the samples were diluted 50-fold in PBS, and the number of particles were determined using a Coulter Counter. The Coulter Counter was set at a lower threshold of 10%, 1/aperture current of 16, 1/amplification of 2. Percent aggregation was calculated by subtracting the number of particles after 1 h (N t ) from the initial particle number and dividing by the initial number {((N 0 Ϫ N t )/N 0 ) ϫ 100}.
LNCaP Adhesion to Purified Proteins-Sterile coverslips were coated overnight with 100 g/ml poly-L-lysine (Sigma), washed twice in sterile water, and allowed to dry. Subsequently, the coverslips were coated with nitrocellulose in methanol (23) and allowed to dry. Purified recombinant proteins were diluted in PBS containing 2 mM CaCl 2 to a con-centration of 75 g/ml for PTP and E-cadherin, respectively, and 40 g/ml for laminin. To identify the individual protein spots on the coverslips, the protein solutions were supplemented with 20 g/ml Texas Red BSA (Sigma). Three distinct spots, each containing a single adhesion molecule (laminin, E-cadherin, and PTP), were generated by spotting 20 l of each protein solution on one coverslip. After a 20-min incubation at room temperature, the protein solutions were aspirated, and this procedure was repeated once. Remaining binding sites on the nitrocellulose were blocked with 2% BSA in PBS, and the dishes were rinsed with RPMI 1640 medium. LNCaP cells infected with the indicated retrovirus were fully trypsinized with 0.05% trypsin, 0.53 mM EDTA (Invitrogen), and 3 ϫ 10 5 cells were added to coverslips, and the cells were allowed to adhere overnight to regenerate cell surface proteins. In the control experiments, function-blocking antibodies to either PTP (BK2, 10 l/ml ascites) or E-cadherin (DECMA, 1:1 dilution of culture supernatant), or 5 mM EDTA (final concentration) was added to the dishes just prior to the addition of cells. In some experiments, the overnight incubation was followed by a 15-min incubation with 20 nM PMA or an equal volume of Me 2 SO. Alternatively, uninfected LNCaP cells were added to coverslips and incubated overnight followed by a 45-min incubation with either 5 M rottlerin, 10 M chelerythrine chloride, 0.5 M GF109203X, 15 nM Gö6976, 10 M LY294002, or Me 2 SO alone. At the concentrations used, chelerythrine chloride (IC 50 ϭ 0.66 M) and GF109203X (IC 50 ranges between 8 nM and 5.8 M for different isoforms of PKC) are specific for PKC, whereas rottlerin is specific for PKC␦ (IC 50 ϭ 6 M), and Gö6976 is specific for PKC␣ and -␤ (IC 50 ϭ 2.3 and 6 nM, respectively). LY294002 inhibits the phosphatidylinositol 3-kinase (IC 50 ϭ 1.4 M). The medium was then removed, and the coverslips were rinsed once in PBS to remove unattached cells. The cells were subsequently fixed with 4% paraformaldehyde, 0.01% glutaraldehyde in PEM buffer (80 mM Pipes, 5 mM EGTA, 1 mM MgCl 2 , 3% sucrose), pH 7.4, for 30 min at room temperature. The coverslips were washed twice in PBS and mounted in IFF mounting medium (0.5 M Tris-HCl, pH 8.0, containing 20% glycerol, and 0.1% p-phenylenediamine). Adherent cells were detected by dark field microscopy, using a 5ϫ objective, and photographed. To quantify the number of adherent cells, the 35-mm negatives were scanned, and the digitized images were analyzed using the Metamorph image analysis program (Universal Imaging Corp., West Chester, PA). The number of adherent cells per image was approximated by highlighting the cells using the threshold function, and the total number of highlighted cells per image was calculated. The data obtained in 4 -6 separate experiments were analyzed by Student's t test (Statview 4.51, Abacus Concepts, Inc.).

Re-expression of PTP-
The receptor tyrosine phosphatase PTP has been shown previously to interact with E-cadherin in a variety of tissues by immunoprecipitation (13)(14)(15). To investigate whether PTP plays a functional role in E-cadherinmediated adhesion, we employed the LNCaP prostate carcinoma cell line (17). Unlike normal prostate epithelial cells (NPr), these cells do not express PTP (Fig. 1a, VEC). To re-express PTP in LNCaP cells, we generated a tetracyclineregulatable retrovirus encoding the PTP cDNA sequence tagged with the green fluorescence protein (PTP-GFP) (16). By using this retrovirus, we re-expressed wild type PTP (PTPWT) in the LNCaP cells. Five days after retroviral infection, the cells were analyzed for expression of PTPWT-GFP by immunoblot and by fluorescence microscopy. Immunoblot analysis showed that LNCaP cells infected with retrovirus containing an empty vector do not express PTP (Fig. 1a, VEC). Cells infected with retrovirus containing PTPWT (Fig. 1a, WT) expressed both the full-length protein (200 kDa) as well as the proteolytically processed forms (ϳ100 kDa) (6). Due to the GFP tag, both the full-length and the proteolytically processed forms of the re-expressed PTPWT migrated at a higher molecular weight than the PTP expressed in normal prostate cells (Fig.  1a, NPr). The retroviral system we used is a tet-off system, and in the presence of tetracycline the gene is not expressed. The re-expression of PTPWT was inhibited by treating the cells with tetracycline (Fig. 1a, WTϩT). Fluorescence microscopy revealed that between 70 and 90% of the LNCaP cells expressed PTPWT and that PTP was primarily localized to the plasma membrane as expected (Fig. 2, C and D). This expression was repressed when the cells were grown in the presence of 4 g/ml tetracycline (Fig. 2, E and F). Control cells infected with a virus containing an empty vector did not show any fluorescence (Fig. 2, A and B).
To assess the functional role of PTP catalytic activity in the regulation of E-cadherin-mediated adhesion, we have generated tetracycline-repressible retrovirus encoding a mutant form of PTP-GFP containing a single amino acid mutation in the catalytic site (16). Mutation of the conserved cysteine residue PTPC1095S (C-S) results in a catalytically inactive enzyme. Immunoblot analysis showed that the C-S mutant was expressed at a similar level to PTPWT in LNCaP cells (Fig.  1a, C-S). Fluorescence microscopy confirmed that infection with the C-S mutant (Fig. 2, G and H) resulted in expression at the plasma membrane at a similar level as PTPWT, demonstrating that the expression and subcellular localization are not affected by the loss of catalytic activity.
Re-expression of PTP Enhanced Calcium-dependent Aggregation of LNCaP Cells-To investigate whether PTP plays a role in E-cadherin-mediated adhesion in LNCaP cells, we trypsinized the cells in the presence of CaCl 2 to selectively preserve the cadherins (22). This assay only measures calciumdependent aggregation predominantly mediated by the cadherins (22). The cells were then allowed to aggregate for 1 h. LNCaP cells infected with an empty vector weakly aggregated (27.4%). Re-expression of PTPWT increased the aggregation 3-fold (72.4%), as did expression of the C-S mutant (72.5%). The increased aggregation was only partly dependent on E-cadherin function, because the presence of an E-cadherin functionblocking antibody did not completely reduce the aggregation induced by re-expression of PTP (49.7%). However, the increased aggregation was Ca 2ϩ -dependent, because the presence of EDTA reduced the aggregation to a level below that seen in cells infected with an empty vector (10.8%). The residual Ca 2ϩ -dependent adhesion is at least due in part to the fact that LNCaP cells express N-cadherin (data not shown). Taken together, these findings demonstrate that re-expression of PTP in LNCaP cells induced Ca 2ϩ -dependent aggregation that is partly because of E-cadherin-dependent cell-cell adhe- To verify that protein expression is under tetracycline control, 4 g/ml tetracycline was added daily to cells infected with retrovirus containing PTPWT (WTϩT). Five days after infection, the cells were lysed, and 30 g of lysate from normal prostate epithelial cells (NPr) and LNCaP cells were separated by 6% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with monoclonal antibodies to PTP. Note that both PTPWT and the C-S mutant have a GFP tag and therefore migrate at a higher molecular weight than PTP expressed in the normal prostate epithelial cells. a, Western blot using the monoclonal antibody SK7 against the intracellular domain of PTP. b, Western blot using the monoclonal antibody BK2 against the extracellular domain of PTP.
sion. Thus, aggregation assays were not ideal to specifically study E-cadherin-dependent adhesion in LNCaP cells. Therefore, we utilized an in vitro adhesion assay that measures specific binding to a given adhesion molecule which is similar to our previously published assay (4).
Re-expression of PTP-induced Adhesion to Purified PTP-To study the specific interactions between cell-cell adhesion molecules in LNCaP cells, we developed an in vitro adhesion assay where purified, recombinant proteins were immobilized on nitrocellulose-coated coverslips. Basically, three spots of protein (laminin, E-cadherin, and PTP) were added to each nitrocellulose-coated coverslip. The field shown in each panel represents virtually the entire spot for a given adhesion molecule. PTP has been shown to mediate cell-cell adhesion via homophilic binding (4,5). To verify that the re-expressed forms of PTP were able to mediate homophilic binding in LNCaP cells, we investigated the adhesion of LNCaP cells to purified recombinant PTP that was immobilized on nitrocellulose-coated coverslips. As expected, cells infected with an empty vector did not adhere to PTP (Fig. 3A) because these cells do not express PTP. Re-expression of PTPWT induced LNCaP adhesion to purified PTP (Fig. 3D), as did re-expression of the C-S mutant (Fig. 3G). Quantitation of the adhesion assays (n ϭ 6) showed that the number of cells that adhered to purified PTP was significantly higher for cells infected with both the WT and the C-S mutant form of PTP as compared with cells infected with vector only (Table I). However, there was no difference between cells expressing PTPWT compared with the C-S mutant in their ability to adhere to purified PTP (Table I). To ensure the specificity of the adhesion assay, we repeated the experiments in the presence of function-blocking antibodies to either PTP or E-cadherin. The presence of an antibody to the extracellular domains of PTP specifically inhibited the adhesion to recombinant PTP of LNCaP cells re-expressing PTPWT (Fig. 4a, WTϩPTP Ab) or cells expressing the C-S mutant (Fig. 4a, C-SϩPTP Ab). As expected, the presence of the E-cadherin antibody had no effect on adhesion to PTP (Fig. 4a, WTϩE-ca. Ab, C-SϩE-cad Ab). Taken together, these data confirm that the re-expressed PTP is present at the cell surface and capable of mediating homophilic binding. In addition, PTP phosphatase activity is not necessary for PTP-dependent adhesion to occur as demonstrated previously (4).
As an internal control in each experiment, cells were allowed to adhere to laminin. Adhesion to extracellular matrix proteins such as laminin is mediated through integrin receptors. Because there is no evidence indicating that PTP regulates integrin function, LNCaP adhesion to laminin should not be affected by the re-expression of PTP. As expected, LNCaP cells infected with an empty vector adhered to laminin (Fig.  3B), and this adhesion was not significantly affected by reexpression of either WT (Fig. 3E) or C-S mutant forms of PTP ( Fig. 3H and Table I). None of the retrovirally infected cells adhered to nitrocellulose coated with BSA only (data not shown).
Re-expression of PTP Restores E-cadherin-mediated Adhesion-To study the role of PTP in the regulation of E-cadherin-mediated adhesion in LNCaP cells, we immobilized purified recombinant E-cadherin on the nitrocellulose-coated coverslips. Despite the fact that these cells express E-cadherin as well as ␣-, ␤-, and ␥catenin and p120 (Fig. 5), LNCaP cells infected with an empty vector did not adhere to E-cadherin (Fig. 3C). Re-expression of PTPWT restored the ability of LNCaP cells to adhere to E-cadherin (Fig. 3F). Quantitation of the adhesion assays show that the number of cells infected with PTPWT that adhered to E-cadherin was significantly higher than the number of cells infected with vector only (Table I). These data show that expression of PTP is necessary for E-cadherin-mediated adhesion in LNCaP cells. Because tyrosine phosphorylation has been reported to negatively regulate cadherin-mediated adhesion, we investigated whether PTP restored E-cadherin-mediated adhesion by dephosphorylating key components of the cadherin-catenin complex. To do this, we repeated the adhesion assays with cells expressing the C-S mutant form of PTP. Expression of the C-S mutant restored E-cadherin-mediated adhesion (Fig. 3I). As shown in Table I, re-expression of WT or the C-S mutant form of PTP induced a significant increase in adhesion to E-cadherin as compared with LNCaP cells infected with an empty vector. In contrast, there was no difference in adhesion between cells infected with PTPWT compared with cells infected with the C-S mutant. The statistical analysis for LNCaP cell adhesion to E-cadherin is summarized in Table I. In control experiments, the adhesion to E-cadherin was totally blocked by a function-blocking antibody to E-cadherin (Fig. 4b, WTϩE-cad Ab and C-SϩE-cad Ab, respectively). In contrast, the PTP antibody did not affect the adhesion to E-cadherin induced by the re-expression of PTPWT (Fig. 4b, WTϩPTP Ab) or by the expression of the C-S mutant (Fig. 4b, C-SϩPTP Ab) as expected. E-cadherin-mediated adhesion in this assay is Ca 2ϩ -dependent, and addition of 5 mM EDTA abolished adhesion to E-cadherin (Fig. 4b, WTϩEDTA and C-SϩEDTA, respectively). However, the presence of EDTA did not affect the adhesion to PTP, which is calcium-independent (4), of cells either reexpressing PTP-WT (Fig. 4a, WTϩEDTA) or expressing the C-S mutant (Fig. 4a, C-SϩEDTA). Taken together, these data  . Five days after infection, the cells were lysed, and 250 g of the lysates were subjected to immunoprecipitation (IP) using antibodies to E-cadherin (E-cad) or L1 (8D9). The immunoprecipitates were separated by 6% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with antibodies to the indicated proteins. As a positive control, 30 g of lysate from vector-infected cells (lysate) is shown in each panel.
indicate that although the presence of the PTP protein is required for E-cadherin-mediated adhesion in LNCaP cells, it does not require PTP catalytic activity.
It is possible that the PTP intracellular domain may recruit other proteins that aid in restoring E-cadherin-mediated adhesion. To determine whether the intracellular PTP domains of PTP were required to affect E-cadherin-dependent adhesion, we constructed a retrovirus encoding the extracellular, transmembrane, and 55 amino acids of the intracellular domains of PTP (PTP-extra) (4). Western blot analysis confirmed that this construct was expressed in LNCaP cells (Fig. 1b, Extra). The cytoplasmic domain of PTP is known to bind to E-cadherin (13). Immunoprecipitation experiments confirmed that PTP-extra does not associate with E-cadherin (data not shown). Expression of PTP-extra induced LNCaP adhesion to purified recombinant PTP ( Fig. 3J; Table I), confirming that the intracellular domains are not required for PTP to mediate homophilic binding (4). In addition, adhesion to PTP was blocked by an antibody to PTP (Fig. 4a, ExtraϩPTP Ab). The antibody to E-cadherin and 5 mM EDTA had no major effect on the adhesion to PTP (Fig. 4a, ExtraϩE-cad Ab and ExtraϩEDTA, respectively) as expected. However, PTP-extra did not restore LNCaP adhesion to recombinant E-cadherin (Fig. 3L), demonstrating that the intracellular domains of PTP are necessary for restoring E-cadherin-mediated adhesion. Because LNCaP cells expressing PTP-extra did not adhere to E-cadherin ( Fig. 3L; Fig. 4b, Extra), the presence of either the PTP antibody, the E-cadherin antibody, or 5 mM EDTA had no effect on adhesion to E-cadherin (Fig. 4b, Extra ϩ PTP Ab, Extra ϩ E-cad Ab, and Extra ϩ EDTA, respectively). As expected, the expression of PTP extra did not affect LN-CaP adhesion to laminin (Fig. 3K). Together, these results suggest that PTP-extra is expressed at the cell surface and capable of inducing adhesion to PTP but not restoring E-cadherin-dependent adhesion.
Similar to the results shown in Fig. 3, LNCaP cells expressing an empty vector did not adhere to either PTP or Ecadherin (Fig. 4a, VEC, and Fig. 4b, VEC, respectively), and this was not altered by the presence of either the PTP antibody, the E-cadherin antibody, or 5 mM EDTA (data not shown). Taken together, these experiments demonstrate that this in vitro adhesion assay can be used to study specific binding to cell-cell adhesion molecules.
Expression of Cadherins and Catenins-Cadherin-mediated cell-cell adhesion is dependent on the expression of both cadherins and catenins. Immunoblot analysis demonstrated that LNCaP cells expressed E-cadherin as well as ␣-, ␤-, and ␥-catenin and p120 (Fig. 5, lysate). This is in accordance with normal prostate epithelial cells, which were found to express similar amounts of E-cadherin as well as ␣-, ␤-, and ␥-catenin (data not shown). Infection of LNCaP cells with an empty vector, PTPWT, or the C-S mutant form of PTP did not alter the expression of any of the proteins in the cadherin-catenin complex (Fig. 5). It is possible that re-expression of wild type or mutant forms of PTP may alter the subcellular localization of the proteins in the cadherin-catenin complex, thereby altering the function of the complex. To address this question, we performed immunocytochemical analysis on LNCaP cells using antibodies to E-cadherin as well as ␣-, ␤-, and ␥-catenin and p120. However, re-expression of either PTPWT or the C-S mutant did not significantly alter the subcellular localization of any of the proteins examined (data not shown).
PTP Does Not Alter the Association of ␣-, ␤-, ␥-Catenin or p120 to E-cadherin-To examine the possibility that the presence of PTP affects the binding of ␣-, ␤-, and ␥-catenin or p120 to E-cadherin, we immunoprecipitated E-cadherin from cells infected with an empty vector (VEC), PTPWT (WT), C-S, or PTP-extra (Extra). As shown in Fig. 5, the immunoprecipitates from cells infected with an empty vector, PTPWT, as well as the C-S and PTP-extra contained equal amounts of E-cadherin. The immunoblot was stripped and reprobed with antibodies to the catenins. Immunoprecipitates from cells infected with an empty vector contained ␣-, ␤-, and ␥-catenin as well as p120. Infection of cells with various forms of PTP did not significantly alter the amounts of the catenins that coimmunoprecipitated with E-cadherin. As a control, a monoclonal antibody to chick L1 (8D9) was used. This antibody did not immunoprecipitate either E-cadherin or any of the catenins.
The PTP Cytoplasmic Domain, Regardless of Catalytic Activity, Is Required for the Interaction with RACK1-Even though the presence of PTP does not alter the composition of the E-cadherin-catenin complex, it is possible that full-length PTP regulates E-cadherin-dependent adhesion by recruiting other signaling molecules to the cadherin-catenin complex. In a recent paper (18), we demonstrated an interaction between the membrane-proximal phosphatase domain of PTP and RACK1, a receptor for activated PKC (19). Because RACK1 binds to the catalytic domain of PTP, we tested whether catalytic activity of PTP was required to interact with RACK1. We performed immunoprecipitation with antibodies directed against PTP or RACK1 and subjected the immunoprecipitates to SDS-PAGE and immunoblotted the gels with anti-RACK1 antibodies. Immunoprecipitation of RACK1 showed that LNCaP cells infected with an empty vector expressed RACK1 (Fig. 6a, VEC) and that infection of cells with various forms of PTP did not alter the expression of RACK1 (Fig. 6a, WT, C-S, and E, respectively). To investigate whether PTPWT and the C-S mutant associate with RACK1 in LNCaP cells, we immunoprecipitated PTP using an antibody to the extracellular domain of PTP (BK2). RACK1 was found to associate with both PTPWT (Fig. 6b, WT) and the C-S mutant (Fig. 6b, C-S) but not with PTP-extra (Fig. 6, b, E). As expected, PTP antibody did not immunoprecipitate RACK1 from cells infected with an empty vector (Fig. 6b, VEC). This experiment was repeated with an antibody to the intracellular do- . Five days after infection, the cells were lysed, and 400 g of the lysates were subjected to immunoprecipitation using a monoclonal antibody generated against RACK1 (a), a monoclonal antibody generated against the extracellular domain of PTP (BK2) (b), or a monoclonal antibody generated against the intracellular domain of PTP (SK7) (c). The immunoprecipitates were separated by 10% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with antibodies generated against RACK1 (a-c). d, the immunoblot of RACK1 immunoprecipitates shown in a was stripped and reprobed with a polyclonal antibody generated against PKC␦. main of PTP (SK7). As seen in Fig. 6c, the SK7 antibody also co-immunoprecipitated RACK1 from cells infected with PTPWT or C-S but not from cells infected with PTP-extra or an empty vector. Taken together, these data demonstrate that full-length PTP regardless of its catalytic activity associates with RACK1.
The association between PTP and RACK1 suggests that the presence of the PTP protein in LNCaP cells may regulate the PKC pathway, which could be involved in the restoration of E-cadherin-mediated adhesion by PTP. Several studies (24 -26) have shown that activation of the PKC pathway can either up-regulate or down-regulate E-cadherin-mediated adhesion depending on cell type. Therefore, we investigated whether activation of PKC by PMA affects the ability of PTP to restore E-cadherin-mediated adhesion. LNCaP cells infected with retrovirus containing PTPWT were allowed to adhere to PTP, laminin, or E-cadherin as described above. The cells were then treated with PMA for 15 min to activate PKC, which did not affect adhesion to either PTP or laminin (Fig. 7a). However, activation of PKC detached the PTP-expressing cells from E-cadherin (Fig. 7a). The statistical analyses for the effects of PMA on LNCaP adhesion are shown in Table I.
Inhibition of PKC␦ Induced LNCaP Adhesion to E-cadherin-To examine further the signal transduction pathways involved in restoring E-cadherin-mediated adhesion, we studied the effect of various kinase inhibitors on the ability of LNCaP cells to adhere to E-cadherin. Uninfected LNCaP cells were added to coverslips with immobilized E-cadherin, and the cells were subsequently treated with the inhibitors. Chelerythrine chloride and GF109203X are broad specificity compounds that inhibit most PKC isoforms. Rottlerin only inhibits the PKC␦ isoform of PKCs. We found that treatment of the cells with three different PKC inhibitors, chelerythrine chloride, GF109203X, and a PKC␦-specific inhibitor (rottlerin) induced LNCaP adhesion to E-cadherin (Fig. 7b, CHE, GF, and R, respectively). The induction of E-cadherin-dependent adhesion was due to inhibition of PKCs in general but we believe specific for inhibition of PKC␦, because treatment of LNCaP cells with the PKC␣-and PKC␤-specific inhibitor Gö6976 did not induce adhesion (Fig. 7b, Go). In addition, neither Me 2 SO nor the phosphatidylinositol 3-kinase inhibitor LY294002 induced LN-CaP adhesion to E-cadherin (Fig. 7b, Me 2 SO and LY, respectively). Also, the effect of the PKC␦ inhibition was specific in that it only affected the ability of LNCaP cells to adhere to E-cadherin but did not alter adhesion to laminin (data not shown). Interestingly, the PMA-induced detachment of the PTP-expressing cells from E-cadherin was blocked by preincubating the cells with rottlerin (data not shown). Furthermore, PKC␦ was found to associate with RACK1 in LNCaP cells, although this association was not affected by the presence of either PTPWT or the C-S mutant (Fig. 6d) (19). This result is not surprising because our experiments were done in the presence of serum, which activates PKC. Our data suggest that PTP negatively regulates PKC␦ activity to restore E-cadherindependent adhesion. However, the precise mechanism of PKC␦ regulation by PTP is not clear but is likely to involve RACK1. Taken together, these data indicate that PTP may restore E-cadherin-mediated adhesion in LNCaP cells by regulating the PKC pathway through the recruitment of RACK1 to the PTP complex. DISCUSSION Alterations in the function of the E-cadherin/catenin adhesion system occur frequently in a wide variety of human carcinomas (27). The molecular mechanisms underlying the loss of expression or functionality of individual components of the cadherin-catenin complex is still only partly understood. Previous studies (13)(14)(15) have shown that PTP associates with classical cadherins. The functional importance of this interaction was illustrated in a study from our lab where we demonstrated that PTP regulates N-cadherin-mediated neurite outgrowth of retinal ganglion cells (16). Therefore, it is possible that a loss of expression or function of PTP may result in a defect in cadherin-mediated adhesion. To investigate the role of PTP in E-cadherin-mediated adhesion, we employed the LN-CaP prostate carcinoma cell line (17). These cells provide a good model system in that they, unlike normal prostate epithelial cells, do not express endogenous PTP. This allowed us to re-express PTP and study the effects of wild type (WT) as well as mutant forms of PTP without the interference of endogenous PTP. Retroviral re-expression of both WT and the catalytically inactive mutant form of PTP induced LNCaP cell adhesion to purified recombinant PTP, demonstrating that PTP was indeed expressed at the cell surface at a level that could mediate homophilic binding. These results also show that perturbation of the phosphatase activity did not alter the subcellular localization or the ability of PTP to mediate homophilic binding as expected (4).
Although LNCaP cells express E-cadherin, as well as ␣-, ␤-, and ␥-catenin and p120, they were unable to mediate E-cadherindependent adhesion. Re-expression of PTPWT restored this adhesion, demonstrating a functional role for PTP in E-cadherin-mediated adhesion. The fact that the re-expression of the catalytically inactive mutant also restored E-cadherin-medi- ated adhesion indicates that PTP exerts an effect on E-cadherindependent adhesion that is independent of its catalytic activity. This process requires the presence of the intracellular domain, because expression of PTP-extra failed to restore E-cadherinmediated adhesion. It is possible that PTP alters the cadherin function by recruiting some signaling protein(s) to the cadherin complex through protein-protein interactions involving the PTP intracellular domain.
The interaction between RPTPs and various proteins may serve to regulate either the subcellular localization of RPTPs or to recruit other signaling molecules to form a larger signaling complex. The fact that PTP, regardless of its catalytic activity, could restore E-cadherin-mediated adhesion suggests that part of its role in the cadherin complex is to recruit other signaling molecules that may be needed for functional E-cadherin-dependent adhesion. The importance of the intracellular domain of PTP is clearly demonstrated by the finding that LNCaP adhesion to E-cadherin was not restored by the expression of a construct where the majority of the intracellular domain of PTP had been deleted (PTP-extra). In this regard, we isolated RACK1 as a protein that binds to the first phosphatase domain of PTP in a two-hybrid screen (18). RACK1 is a homologue of the G␤ subunit of heterotrimeric G-proteins (19) and consists of seven WD repeats that are believed to form a propeller-like structure (28). RACK1 is thought to be a scaffolding molecule because each of the seven WD repeats could potentially mediate protein-protein interactions. RACK-1 was originally described as a receptor for activated PKC (19), but more recent studies have described its interaction with a variety of signaling proteins, such as Src (29), and with select pleckstrin homology domains in vitro (30). In this study, we found that full-length PTP interacts with RACK1 and that this interaction is not dependent upon the catalytic activity of PTP. The interaction between PTP and RACK1 suggests that PTP may regulate E-cadherin-mediated adhesion by recruiting RACK1 and other signaling molecules to the PTP adhesion complex.
Despite numerous attempts to clarify the regulation of cadherin function by tyrosine phosphorylation, it is not fully understood. Tyrosine phosphorylation has been correlated with loss of cadherin-mediated adhesion and destabilization of adherens junctions (reviewed in Ref. 2). Therefore, adhesive function may be controlled by reversible tyrosine phosphorylation. Components of the cadherin-catenin complex are phosphorylated by a number of cytoplasmic and receptor protein tyrosine kinases including Src, EGF receptor, and Met (the scatter factor receptor) (2). In addition, PTP and a few other PTPs interact with cadherins and catenins (2). The association of the cadherins with both kinases and phosphatases indicates a critical role for dynamic tyrosine phosphorylation in cadherin function.
We performed studies on the role of tyrosine phosphorylation in regulating the association between PTP and E-cadherin in cells transformed with a temperature-sensitive form of the Rous sarcoma virus (14). The mutant Rous sarcoma virus is temperature-sensitive for pp60 src tyrosine kinase activity. When grown at the permissive temperature, increased tyrosine phosphorylation induced by Src resulted in an increased tyrosine phosphorylation of E-cadherin, which correlated with a decreased association between PTP and E-cadherin. However, in this study we show that PTP regulates the cadherin function independently of its phosphatase activity, indicating that the cadherin-catenin complex may not be the primary substrates for PTP. We have shown previously that PTP catalytic activity is required for N-cadherin-mediated neurite outgrowth (16). These data indicate that PTP catalytic activ-ity may be required for signaling events that regulate the cytoskeleton and thus other cadherin-dependent functions downstream of adhesion.
An alternative hypothesis is that the C-S mutant form of PTP may indirectly alter the tyrosine phosphorylation of the cadherin-catenin complex. In a recent paper (18), we found that PTP and Src compete for binding to RACK1. RACK1 binds to the SH2 domain of Src, an interaction that inhibits Src kinase activity (29). The interaction between RACK1 and PTP may regulate the presence of the Src protein tyrosine kinase in the cadherin-catenin complex. The presence of the PTP protein could recruit RACK1 to the plasma membrane where it could dissociate from PTP and possibly bind to and inactivate Src. Inactivation of Src could indirectly regulate the tyrosine phosphorylation of either E-cadherin or the catenins, thereby restoring E-cadherin-mediated adhesion.
Several studies have indicated that PKC is involved in the regulation of E-cadherin-mediated adhesion and the formation of adherens junctions. The molecular mechanisms whereby PKC regulates E-cadherin function are unknown. Additionally, the activation of PKC has been reported to have the opposite effects on E-cadherin function in different cell types. For example, the calcium-induced formation of adherens junctions in keratinocytes is dependent on the activation of PKC (24). On the other hand, activation of PKC has been shown to induce the dissociation of E-cadherin from the cytoskeleton (25), followed by cell scattering in the HT29 intestinal cell line (26). In this study, we show that the inhibition of PKC␦ restored E-cadherin function in LNCaP cells. In addition, PTP restored E-cadherindependent adhesion, which could be reversed by PMA stimulation of PKCs. Together, these data suggest that PTP may negatively regulate PKC activity in LNCaP prostate carcinoma cells. Although the precise mechanism is unclear, it is likely to involve the PTP-RACK1 complex.
Others have shown (31,32) that serine/threonine phosphorylation of p120 negatively regulates E-cadherin-mediated adhesion. It is therefore possible that the inactivation of PKC␦ leads to decreased phosphorylation of p120 and thereby increased E-cadherin-mediated adhesion. However, we could not detect any alteration in the phosphorylation of p120 either after re-expression of PTP or after treatment of uninfected LNCaP cells with the PKC␦ inhibitor rottlerin (data not shown). In addition, activation of PKC caused cells expressing PTPWT to dissociate from an E-cadherin substrate. The fact that this dissociation occurred within 15 min after the addition of PMA argues that PKC directly affects the E-cadherin complex, rather than down-regulating the expression of either Ecadherin or the catenins. Therefore, the role of PTP in regulating E-cadherin-mediated adhesion could be to recruit RACK1 to the plasma membrane, thereby regulating the PKC pathway.