INTRODUCTION

A variety of biological signals that regulate cell growth, migration, differentiation, and survival are mediated by protein tyrosine phosphorylation (1, 2). These processes are thought to be counteracted by a family of proteins termed protein-tyrosine phosphatases (PTP)¹ (3). Dephosphorylation by PTPs can either inactivate a signal generated by protein phosphorylation (4, 5) or, alternatively, promote signal transduction by dephosphorylation of phosphotyrosine residues with negative regulatory functions such as, for example, in the case of Src-like kinases (6). PTPs can be subdivided into receptor-like plasma membrane-spanning and soluble cytosolic forms. Their biological function appears to be determined not only by specific target substrates but also by their intracellular localization to distinct compartments (7, 8). The subfamily of receptor-like PTPs (RPTP) is composed of five different classes based on structural motifs in their extracellular domain (9). Identification of amino acid motifs that resemble those of cell adhesion molecules in the extracellular domain of class II and III RPTPs has led to the suggestion that these PTPs may be involved in the regulation of cell contact formation. The observation that overexpression of PTPµ and PTP can induce cell aggregation by homophilic interaction of their respective extracellular regions supports this hypothesis (10, 11, 12). In this process the so called ``MAM'' domain (13) appears to determine the specificity of interaction (14).

The key molecules involved in the formation of cell-cell adhesions are members of the cadherin-catenin family (15), which connect adjacent cells via cadherin extracellular domain-mediated homophilic, Ca²⁺-dependent interactions (16, 17). The cytoplasmic domain of E-cadherin on the other hand is responsible for the complex formation with the intracellular catenins (18, 19, 20, 21), which, in turn, link cadherins to the actin filament network (22, 23, 24, 25, 26). -Catenin and -catenin/plakoglobin form two distinct and mutually exclusive complexes with E-cadherin and -catenin but are also found in E-cadherin-independent pools (27, 28) and have been shown to associate with the tumor suppressor gene product adenomatous polyposis coli (29, 30, 31). The biological function of cell-cell adhesions, however, extends beyond a mere maintenance of the cellular architecture, and a direct involvement of these specialized molecules in signal transduction events has been postulated.

-Catenin and -catenin/plakoglobin are homologous to the Drosophila armadillo (arm) protein (32, 33, 34, 35), and this gene product has been shown to be essential for a signal transduction pathway that involves the segmental pattern formation during Drosophila development (36). The characteristic feature of these proteins is the presence of a variable number of 42 amino acid repeats (arm motif; Ref. 32), which is the basic motif of the armadillo family, including -catenin and -catenin/plakoglobin (35). The observation of tyrosine phosphorylation of the cadherin-catenin complex after growth factor stimulation or v-src transformation (37, 38, 39, 40) as well as the association of -catenin and -catenin/plakoglobin with the epidermal growth factor receptor and HER2/c-erbB-2 (41, 42, 43) further support a direct involvement of cell adherens junctions in cell signaling. The fact that a disruption of the cadherin-catenin complex promotes the invasiveness of malignant tumors, that E-cadherin has been demonstrated to be a tumor suppressor gene (44, 45), and the notion that tyrosine phosphorylation seems to interfere with cadherin function (38, 39) makes the question as to how PTPs are involved in cell contact formation all the more compelling.

In this report, we describe the identification of a human PTP that represents the homolog of murine PTP (46). This class II transmembrane phosphatase is expressed in a cell density-dependent fashion and is recruited to areas of cell-cell contact. hPTP colocalizes with cell adhesion molecule-associated proteins at adherens junctions and associates in vitro and in intact cells with two members the armadillo family of proteins. Our observations strongly suggest a biological role of hPTP in the regulation of cell contact and adhesion.

MATERIALS AND METHODS

PCR and cDNA Cloning of hPTP

Poly(A)⁺ RNA was isolated from HTB30 cells and cDNA synthesized using avian myeloblastosis virus reverse transcriptase as described (47). PTP sequence fragments were amplified using the PCR with a pool of degenerated oligonucleotide primers based on conserved amino acid sequences in the PTP catalytic domain (48) under standard conditions. PCR products were cloned in Bluescript KS⁺ vector (Stratagene) and sequenced by the dideoxynucleotide chain termination method (49). A  ZAP II library (Stratagene) generated from HTB30 poly(A)⁺ RNA was screened with a PCR fragment probe under high stringency conditions (50). The full-length cDNA of hPTP was cloned into a cytomegalovirus promoter-based eukaryotic expression plasmid (pCMV; Ref. 51).

Total RNA was isolated by the guanidinium isothiocyanate method (52) from cultured cells grown to 30, 70, and 100% confluence. Poly(A)⁺ RNA was prepared by oligo(dT)-cellulose chromatography (53) and Northern blot analysis was performed as described (47).

Cloning of -Catenin and -Catenin/Plakoglobin

Human -catenin () and -catenin/plakoglobin () were amplified from cDNA generated from MCF7 cells using the PCR method. PCR products were cloned in an eukaryotic expression vector under the control of the CMV promoter and confirmed by sequence analysis.

All cell lines were obtained from the American Tissue Culture Collection (ATCC). SK-BR-3 cells (HTB30) and HT29 cells (HTB38) were grown in McCoy's medium supplemented with 1% glutamine and 15% or 10% FCS (Life Technologies, Inc.), respectively. MCF7 cells (HTB22) were grown in RPMI supplemented with 1% glutamine and 10% FCS. NBT-II cells (CRL-1655) were grown in Eagle's minimal essential medium supplemented with 1% glutamine, nonessential amino acids, 1 mM sodium pyruvate, Earle's balanced salt solution, and 10% FCS. Human embryonic 293 kidney cells (CRL 1573) were grown in Dulbecco's modified Eagle's medium, supplemented with 1% glutamine and 10% FCS.

For immunofluorescence studies, NBT-II cells were grown on uncoated glass coverslips, fixed for 30 min with 2% paraformaldehyde in 0.1 M cacodylate buffer (pH 7.4) on ice, washed repeatedly with phosphate-buffered saline (PBS) containing glycine, and permeabilized for 10 min with 0.1% Tween in PBS. Nonspecific antibody binding was blocked for 1 h with 5% normal goat serum in phosphate-buffered gelatin (PBG: PBS, 0.5% bovine serum albumin, 0.045% fish gelatin). Incubation with primary antibodies was done at room temperature for 2 h after dilution in PBG, 1:100 for polyclonal anti-hPTP-JM-1, 1:200 for monoclonal anti-plakoglobin/-catenin and monoclonal anti--catenin. After three washes in PBG, primary antibody binding was detected with isotype-specific secondary antibody, dichlorotriazinyl-amino-fluorescein-conjugated donkey-anti-rabbit IgG (1:200), or Cy3-conjugated donkey-anti-mouse IgG (1:300, Jackson Laboratories, West Grove, PA). Nuclei were counterstained with 4,6-diamidino-2-phenylindole and coverslips mounted under glycerol-2.5% 1,4-diazo-bicyclo(2.2.2)octan.

Transient transfection of human 293 embryonic kidney cells was performed as described (54). After washing with PBS, cells were lysed in 1 ml of lysis buffer/10-cm plate (50 mM HEPES (pH 7.5), 150 mM NaCl, 1.5 mM MgCl₂, 5 mM EDTA, 10% glycerol, 1% Triton X-100, 10 mM NaFl, 2 mM phenylmethylsulfonyl fluoride, 10 mg/ml aprotinin, 1 mM sodium orthovanadate), and lysates were precleared by centrifugation at 12,500 × g for 10 min at 4 °C. Prior to immunoprecipitations, lysates were adjusted for equal protein concentrations, and the appropriate antibody and protein A-Sepharose were added to the lysate and incubated for 3 h at 4 °C. Precipitates were washed three times with HNTG buffer (20 mM HEPES (pH 7.5), 150 mM NaCl, 10% glycerol, 0.1% Triton X-100) and SDS-sample buffer was added. For subsequent Western blot analysis, the separated proteins were transferred to nitrocellulose (Schleicher and Schuell) and incubated with the respective antibody. The ECL system (Amersham Corp.) in conjunction with horseradish peroxidase-conjugated goat anti-mouse or goat anti-rabbit antibody (Bio-Rad) was used to visualize proteins recognized by the antibodies. Before reprobing blots were incubated for 1 h in 67 mM Tris-HCl (pH 6.8), 2% SDS, and 0.1% -mercaptoethanol at 50 °C.

Construction of hPTP Mutants

In vitro mutagenesis of PTP was performed according to the method described by Kunkel (55). The following oligonucleotides were used: 5-pTCGTCCAGCACCAGCATCGGCATGTACAACGATGGGGCC-3 for Cys-Ala mutation of the NH₂-terminal PTPase domain (aa 1082) and 5-pTCGCCCGCCACCATTTAGAGCGTGGATAATCGTCCGGCC-3 for Cys-Ala mutation in the COOH-terminal PTPase domain (aa 1376). The double Cys mutant was designated hPTP-C/A.

Plasmids coding for hPTP/Glutathione S-transferase (GST)-fusion proteins were constructed by amplification of the sequences between amino acids 783-1439 of hPTP; amino acids 783-1439 of hPTP-C/A and the juxtamembrane region of hPTP, amino acids 783-904, using the PCR method (named GST-hPTP_i, GST-hPTP_i-C/A, and GST-hPTP-JM, respectively). PCR products were subcloned in the appropriate pGEX vector (Pharmacia Biotech Inc.) and sequenced to confirm the integrity of the sequence. GST fusion proteins were prepared and purified as described (54).

For the in vitro binding assay -catenin/plakoglobin and -catenin were transiently expressed as described above. Lysates were incubated with 5 µg of GST-hPTP_i, GST-hPTP-JM, or 8 µg of GST alone, immobilized on Glutathione Sepharose, rotated 3 h at 4 °C, and washed three times with HNTG. Bound proteins were separated by SDS-PAGE for Western blotting.

-Catenin was transiently overexpressed in human embryonic kidney cells. Cells were serum-starved, treated with pervanadate (0.3 mM H₂O₂, 0.1 mM sodium orthovanadate) for 10 min and lysed. Immunoprecipitation and washing were performed as described above. PTPase activity toward tyrosine phosphorylated -catenin was assayed in 100-µl reactions containing 50 mM acetate (pH 5.5), 10 mM DTT, 1 mg/ml BSA, and 0.1% CHAPS (Sigma) with 50 ng of GST-hPTP_i or GST-hPTP_i-C/A added. Enzymatic activity was stopped by addition of 50 µl of SDS sample buffer and immediate boiling. The reaction mix was separated by SDS-PAGE and the tyrosine phosphorylation status analyzed by Western blotting using anti-phosphotyrosine antibody.

Ab 116 represents a polyclonal antiserum raised against a peptide sequence (residues 60-76) within the extracellular domain of hPTP. Additional hPTP-specific antisera were generated by subcloning a fragment encoding the 93 carboxyl-terminal amino acids (CT-1), the nucleotides corresponding to the amino acids 783-904 (JM-1) or amino acids 1181-1336 (D2-1), respectively, in the fusion protein expression vector pGEX (Pharmacia). The fusion proteins were purified as described (56) and used for immunizing rabbits. Monoclonal antibody against phosphotyrosine (4G10) was obtained from Upstate Biotechnology, Inc., -catenin, and -catenin/plakoglobin antibodies were obtained from Transduction Laboratories. The monoclonal anti-GAPDH antibody was described before (57).

RESULTS

Isolation and Analysis of hPTP cDNA Clones

To investigate PTP expression and function in cancer cells, we performed PCR experiments employing mRNA preparations from the human mammary carcinoma cell line SK-BR-3 and degenerated oligonucleotide primer pools corresponding to conserved sequences within the PTP catalytic domains (48). Sequence analysis of the cloned PCR fragments revealed the expression of several previously characterized as well as novel PTPs. One of the sequences was highly represented (18%) in the 121 clones examined and was used to screen a  ZAP II SK-BR-3 cDNA library at high stringency. Eleven overlapping clones were assembled to a full-length clone of 6.1 kb in size (not shown). Its open reading frame, coding for a 1439-amino acid sequence, displayed a high degree of homology (95.4%) with the murine type II transmembrane PTP (46). We concluded that our clone represented the human homolog of this previously identified mouse PTP and accordingly termed it hPTP. The deduced amino acid sequence of hPTP is shown in comparison with mPTP in Fig. 1A. Using Northern blot analysis hPTP mRNA transcripts were found at high levels in human lung, brain, and colon, and to a lesser extent in liver, pancreas, stomach, kidney, and placenta as well as in mammary carcinoma cell lines MDA-MB-453, HBL-100, T-47D, MDA-MB-435, BT-20, BT-474, SKBR-3, MDA-MB-231 and MCF-7 (data not shown).

Several structural features suggested that hPTP may be involved in cell adhesion, including a MAM domain (13), one possible Ig-like domain, and four fibronectin type III-related domains that match the FNIII-like repeats of LAR (58), PTP (59), PTP (60), and RPTPµ (61).

To investigate the properties of the hPTP gene product, the complete cDNA was cloned into a cytomegalovirus early promoter-based expression vector and transfected into 293 human embryonic kidney cells. Western blot analysis with antibodies directed against a fragment within the extracellular domain (Fig. 1B, left panel) or against the carboxyl terminus (Fig. 1B, right panel) of hPTP resulted in the detection of three bands of 185 kDa, approximately 115 kDa, and 97 kDa, respectively. Since the calculated mass of hPTP was determined to be 163 kDa, it was likely that the 185-kDa band represented the glycosylated form of the entire hPTP polypeptide chain, while the 115- and 97-kDa bands could be defined subunits of hPTP. Among other possibilities these fragments may be the result of endoprotease furin cleavage in the extracellular domain of hPTP (Fig. 1A; residues 640-643) (62, 63), since similar processing events have been described for several other type II PTPases (46, 60, 64). The 115-kDa band, which was identified to contain extracellular sequences because it could be detected with an antiserum raised against amino acid residues 60-76 in the MAM domain, was termed -fragment (Fig. 1B, left panel). The 97-kDa fragment containing the transmembrane and intracellular domain was termed -fragment (Fig. 1B, right panel). That the RTKR sequence is indeed the cleavage site was confirmed by site-directed mutagenesis, which demonstrated that alteration of RTKR to LTNR resulted in a noncleaved hPTP protein (data not shown).

Cell Density Regulation of hPTP Expression

The motifs within the extracellular domain of hPTP that resemble those of proteins involved in cell-cell and cell-extracellular matrix interactions led us to investigate whether the level of hPTP expression was affected by density of cells growing in culture. SK-BR-3 mammary carcinoma cells were harvested at 30, 70, and 100% density, and poly(A)⁺ RNA was prepared. Northern blot analysis with a probe corresponding to the extracellular domain of hPTP revealed an elevation in hPTP transcript levels with increasing cell density (Fig. 2A, top). The intensity of the signal obtained with a GAPDH probe was not affected (Fig. 2A, bottom). Thus, expression of the hPTP gene appears to be up-regulated by signals triggered upon cell contact. To confirm that the increase in hPTP mRNA transcripts is reflected on the protein level, cells were grown as described above, lysed, adjusted for protein concentration, and analyzed by SDS-PAGE. Western blot analysis with the anti-hPTP specific antibody D2-1 revealed a severalfold increase in hPTP protein level with increasing cell density (Fig. 2B, top). Reprobing with an anti-GAPDH specific monoclonal antibody revealed comparable amounts of protein (Fig. 2B, bottom).

Cell density-dependent induction of hPTP.

Colocalization of PTP with Catenins

We investigated the localization of PTP by immunocytochemical methods in different cell lines such as NBT-II and HT29 cells. Conventional fluorescent microscopy revealed that PTP was localized predominantly along the cell-cell contacts of adjacent cells, while only weak fluorescence was detectable at the contact-free cell membranes. A representative staining of NBT-II cells is shown in Fig. 3A. To confirm the specificity of the fluorescent signal, cells were incubated with nonimmuneserum, with anti-hPTP antiserum preincubated with GST-hPTP fusion protein or without primary antibody (Fig. 3, B, C, and D, respectively). Since intercellular junctions are the predominant localization of the cadherin-associated proteins -catenin and -catenin/plakoglobin, we performed double labeling experiments for PTP (Fig. 3, E and G) and -catenin or -catenin/plakoglobin (Fig. 3, F and H, respectively). Using laser confocal microscopy with monoclonal anti--catenin or anti--catenin/plakoglobin antibodies and the anti-hPTP antibody JM-1, we found colocalization of these proteins at cell-cell contacts as indicated by yellow fluorescence (Fig. 3, I and K, respectively), which arises from the superimposition of the red -catenin/plakoglobin and the green hPTP label.

hPTP Associates with -Catenin and -Catenin/Plakoglobin

The colocalization of hPTP with -catenin and -catenin/plakoglobin prompted us to investigate whether these proteins associate with hPTP in vitro. We constructed GST fusion proteins containing either the whole cytoplasmic part of hPTP (GST-hPTP_i) or only the juxtamembrane region (GST-hPTP-JM), which displays limited homology to the intracellular domain of E-cadherin (65). -Catenin and -catenin/plakoglobin were transiently overexpressed in human 293 embryonic kidney fibroblasts and lysates of these cells incubated with GST, GST-hPTP_i, or GST-hPTP-JM immobilized on glutathione-Sepharose. Matrix bound protein was subjected to Western blot analysis with either anti--catenin or anti--catenin/plakoglobin antibody and led to the detection of both proteins bound to GST-hPTP_i (Fig. 4A, lane 4-6) as well as to the GST-hPTP-JM fusion protein (Fig. 4A, lane 7-9). No binding was detected to GST protein alone (Fig. 4A, lane 1-3). This interaction appeared to be specific since -catenin, another member of the catenin family that does not contain an arm motif, and E-cadherin did not associate with hPTP under the same conditions (data not shown). We therefore concluded that hPTP binds specifically to the arm motif containing adhesion protein-associated -catenin and -catenin/plakoglobin. The presence of the juxtamembrane region of hPTP appeared to be sufficient for this association (Fig. 4A).

association of hPTP with -catenin and plakoglobin.

Next, we determined whether hPTP and ``arm proteins'' could be shown to associate in intact cells. To this end, human HT29 cells were serum-starved or stimulated with the tyrosine phosphatase inhibitor pervanadate, lysed, and immunoprecipitated with an anti-hPTP-specific antibody. Western blot analysis using a monoclonal anti--catenin or anti--catenin/plakoglobin antibody detected both proteins in anti-hPTP immunoprecipitates (Fig. 4B, lane 3 and 4; top and middle panel, respectively) and therefore confirmed the specificity of this association. No co-immunoprecipitation was detected with the antibody JM-1 after preincubation with the specific antigen (Fig. 4B, lanes 5 and 6) or nonimmuneserum (Fig. 4B, lane 1 and 2). The presence of hPTP was confirmed by reblotting with anti-hPTP antibody D2-1. We found the association to be independent of the phosphorylation state of -catenin or -catenin/plakoglobin, which are both phosphorylated after treatment with pervanadate (data not shown). The precipitated proteins, however, were phosphorylated after pervanadate treatment, as indicated by the shift in the apparent protein size (Fig. 4B, lane 4).

The in vivo and in vitro association of -catenin and -catenin/plakoglobin with hPTP led us to investigate whether these proteins could serve as potential substrates for hPTP. To examine this possibility we performed an in vitro assay using GST fusion proteins of hPTP (GST-hPTP_i) as well as a catalytically inactive mutant of hPTP (GST-hPTP_i-C/A). Prior to the experiment, we determined the enzymatic activity of GST-hPTP_i or GST-hPTP_i-C/A, respectively, using pNPP as a substrate (data not shown). Next, -catenin was transiently overexpressed in human 293 embryonic kidney fibroblasts, which were treated with pervanadate to induce tyrosine phosphorylation of substrates. Immunoprecipitated phosphotyrosine containing -catenin was then incubated with GST-hPTP_i or GST-hPTP_i-C/A. The reactions were terminated after different time intervals and samples were separated by SDS-PAGE. Western blot analysis with an anti-phosphotyrosine-specific antibody revealed a strong reduction in the tyrosine phosphorylation signal within the first 15 min when phosphorylated -catenin was incubated with GST-hPTP_i (Fig. 5A, lanes 1-4). No detectable change in phosphorylation levels, however, was observed after treatment with GST-hPTP_i-C/A (Fig. 5A, lanes 5 and 8). Blots were reprobed with an anti--catenin-specific antibody to confirm that identical amounts of protein were loaded (Fig. 5B). We therefore concluded that -catenin may represent a substrate of hPTP.

hPTP activity against tyrosine-phosphorylated -catenin.

DISCUSSION

Little is known about interacting proteins or in vivo substrates of type II and type III phosphatases. Type II PTP LAR was recently described to interact and colocalize with a novel protein termed LIP-1 (LAR interacting protein-1) at focal adhesions. However, LIP-1 does not appear to be a substrate for LAR PTP activity (66). Interestingly, a novel adhesion molecule-like protein was found to interact with and serve as a substrate of the Drosophila phosphatase DPTP10D (67). Recent findings by Brady-Kalnay et al. (68) and our observations now provide evidence for the association between proteins of the cadherin/catenin complex with the receptor tyrosine phosphatases PTPµ and hPTP, respectively.

A role of PTPs in the regulation of cell-cell contacts was originally proposed by Klarlund (69), who found that orthovanadate, a potent inhibitor of phosphatase activity, diminished normal contact inhibition between NRK-1 cells. Recently, overexpression of PTP and PTPµ was shown to induce cell aggregation due to homophilic interactions of their extracellular domains, assigning to these PTPases a physiological role at cell adherens sites (10, 11, 12). Moreover, the receptor-like type III phosphatase DEP-1 was found to be expressed at elevated protein levels with increased cell confluence (70). Recently, Gebbink et al. (74) reported an increase in the expression of PTPµ protein when cells were grown to high density. Similarly, we found an approximately three fold elevation of hPTP in the mRNA transcript and a more drastic elevation in PTP protein level with increased cell density. Because the increase in protein PTP level seems to be more pronounced than the up-regulation of PTP mRNA transcripts we conclude that PTP expression is transcriptionally and in addition similar to PTPµ (74) also posttranscriptionally regulated in a cell density dependent manner.

Our observations are supported by earlier evidence suggesting an involvement of protein tyrosine phosphorylation in processes that regulate cadherin/catenin-mediated cell adhesion phenomena. Initially, a correlation was observed between v-src-transformation of cells leading to a loss of adhesive properties and to in vitro invasiveness of cells and tyrosine phosphorylation of the cadherin-catenin complex (38, 39, 71). Furthermore, epidermal growth factor can stimulate tyrosine phosphorylation of -catenin and -catenin/plakoglobin, and both proteins were shown to associate with HER2/c-ERB-2 and the epidermal growth factor receptor (41, 42, 43). In addition, growth factors such as epidermal growth factor, acidic fibroblast growth factor, and hepatocyte growth/scatter factor have been shown to induce migration of epithelial cells, and a correlation has been established between migration, phosphorylation of -catenin, and redistribution of the proteins of the cadherin-catenin complex (40, 72, 73).

We demonstrate here that hPTP colocalizes with and associates with -catenin and -catenin/plakoglobin in vivo and in vitro. Taken together with the in vitro phosphatase activity of hPTP toward -catenin as a substrate these data further suggest that -catenin is possibly a substrate of hPTP.

Interestingly, Brady-Kalnay et al. (65) recently noted that the juxtamembrane region of PTP and PTPµ shares sequence homology to the intracellular domain of cadherins, which has been shown to be involved in direct binding to the catenins (18, 19, 20, 21). This observation argues for a direct interaction between PTP and the -catenin or -catenin/plakoglobin, respectively. As we could not detect hPTP interacting with cadherin in vitro as was reported previously for the hPTP-related PTPµ (68), we suggest that these differences arise either from a different experimental approach, inherent differences in the protein binding capacities of the related phosphatases, or from limitations in the antibody sensitivity. Moreover, we cannot rule out the involvement of a third unidentified molecule, which could link PTP to the catenins. Considering the implications of the sequence homology of PTP and cadherins we believe, however, the interaction between hPTP and the arm motif containing proteins -catenin and -catenin/plakoglobin to be most likely direct.

It is tempting to speculate that PTPases like hPTP and PTPµ might serve to negatively regulate the action of tyrosine kinase-induced signal events at intercellular junctions by dephosphorylating -catenin and -catenin/plakoglobin or cadherins, respectively. hPTP might therefore contribute to the formation and maintenance of intact adherens junctions. The homophilic interaction between the extracellular domains of hPTP on neighboring cells could either lead to changes in activity toward specific substrates involved in the architecture of cell-cell junctions such as catenins or, alternatively, could simply serve to allow a correct localization of hPTP to its specific site of action at cell-cell contacts. However, no influence of the homophilic interaction on the catalytic activity of PTP and PTPµ toward an artificial substrate has been determined so far.

In summary, we conclude that the cell density-dependent expression, the localization to adherens junctions, and the association with and dephosphorylation of cell adhesion-associated -catenin and -catenin/plakoglobin strongly suggest that hPTP plays a biological role in the regulation of cell contact and adhesion controlled events such as cell proliferation, tumor invasiveness, and metastatic spread.