Receptor Protein-tyrosine Phosphatase RPTPm Binds to and Dephosphorylates the Catenin

RPTPm is a prototypic receptor-like protein-tyrosine phosphatase (RPTP) that mediates homotypic cell-cell interactions. Intracellularly, RPTPm consists of a relatively large juxtamembrane region and two phosphatase domains, but little is still known about its substrate(s). Here we show that RPTPm associates with the catenin p120, a tyrosine kinase substrate and an interacting partner of cadherins. No interaction is detectable between RPTPm and b-catenin. Furthermore, we show that tyrosine-phosphorylated p120 is dephosphorylated by RPTPm both in vitro and in intact cells. Complex formation between RPTPm and p120 does not require tyrosine phosphorylation of p120. Mutational analysis reveals that both the juxtamembrane region and the second phosphatase domain of RPTPm are involved in p120 binding. The RPTPm-interacting domain of p120 maps to its unique N terminus, a region distinct from the cadherin-interacting domain. A mutant form of p120 that fails to bind cadherins can still associate with RPTPm. Our findings indicate that RPTPm interacts with p120 independently of cadherins, and they suggest that this interaction may serve to control the tyrosine phosphorylation state of p120 at sites of cellcell contact.

Protein tyrosine phosphorylation is controlled by the opposing actions of protein-tyrosine kinases (PTKs) 1 and proteintyrosine phosphatases (PTPs). Like the PTKs, the PTPs occur as both receptor and non-receptor forms. The receptor-like PTPs (RPTPs) are characterized by one or two intracellular PTP domains and a variable, putative ligand-binding ectodomain; they are classified according to structural motifs in their ectodomains (for review, see Refs. [1][2][3]. While many RPTP family members have been identified to date, little is still known about their intracellular substrates and signaling properties. We have identified RPTP and shown that it functions as a homotypic cell-cell adhesion receptor and, furthermore, that its expression is up-regulated with increasing cell density (4 -6). Its ectodomain consists of an N-terminal "MAM" domain followed by an Ig-like domain and four fibronectin type III-like repeats (4,7). The MAM domain, which is also present in the related RPTP (8) and PCP-2 (also called RPTP or RPTP) (9 -11), is essential for mediating homophilic cell-cell interactions (12). Intracellularly, RPTP consists of a relatively large juxtamembrane region (about 150 amino acids) and two conserved PTP domains (4). As in most RPTPs, the first (membrane-proximal) PTP domain of RPTP is catalytically active, whereas its second (C-terminal) domain is inactive, at least in vitro (see Refs. 1-3, and references therein).
Several RPTPs have recently been implicated in the regulation of cadherin/catenin complexes. For example, RPTP was reported to associate with both ␤and ␥-catenin (plakoglobin), but not with ␣-catenin or cadherins (13). Another member of the MAM-containing subfamily, termed RPTP, was found in complex with ␤-catenin (10), while leukocyte common antigenrelated RPTP was reported to interact with both ␤-catenin and plakoglobin, but not with ␣-catenin or E-cadherin (14,15). As for RPTP, Brady-Kalnay et al. (16,17) reported that it associates with more than 80% of the cellular cadherin pool. However, our own studies did not confirm the proposed RPTPcadherin interaction (18).
In the present study, we have examined the possible interaction between RPTP and the catenin p120 ctn (formerly termed p120 cas ; Refs. 19 and 20), a protein originally identified as a substrate for the Src tyrosine kinase (21,22). Like ␤-catenin and plakoglobin, p120 ctn belongs to the Armadillo family of proteins as it contains 11 Armadillo repeats flanked by unique N-and C-terminal sequences (23). Through its Armadillo repeats, p120 ctn associates with cadherins, although on a different site and with lower affinity than the classical ␤-catenins and plakoglobin (19, 24 -26). p120 ctn differs from ␤-/␥-catenin in that it does not interact with ␤-catenin (24), nor does p120 bind the tumor suppressor adenomatous polyposis coli protein as do ␤-catenin and plakoglobin (27)(28)(29). Overexpression of p120 ctn in 3T3 cells induces striking morphological changes characterized by the presence of long dendrite-like processes (30), suggesting a role for p120 ctn in cytoskeletal organization.
We report here that RPTP associates with p120 ctn , but not with ␤-catenin, and show that tyrosine-phosphorylated p120 ctn serves as a substrate for RPTP. We have mapped the RPTPbinding site on p120 ctn to its unique N terminus, a region distinct from the cadherin-interacting Armadillo repeats. Furthermore, we find that the interaction between RPTP and p120 ctn requires both the juxtamembrane region and the second PTP domain of RPTP. Our findings suggest that RPTP functions in a cell-cell signaling pathway that involves p120 ctn .

MATERIALS AND METHODS
Cell Culture, Transfection, and Antibodies-Mink lung cells (Mv1Lu), human lung carcinoma cell-line A549, and COS-7 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with antibiotics and 8% fetal calf serum. Culture and infection of Sf9 insect cells were done as described (12). Transfection of COS7 cells was performed using the DEAE-dextran procedure. Briefly, COS cells seeded 1 day before transfection were rinsed with phosphatebuffered saline (PBS) and overlaid with a mixture of plasmid DNA and DEAE-dextran (0.5 mg/ml) in PBS. Following 30 min of incubation at 37°C, cells were incubated in Dulbecco's modified Eagle's medium supplemented with 80 M chloroquine (Sigma), 8% fetal calf serum, and antibiotics. After 3 h of incubation at 37°C, cells were shocked in 10% Me 2 SO and cultured in fresh medium for 2 days before analysis. Monoclonal antibodies 3G4 and 3D7 directed against the extracellular domain and polyclonal antibody Ab37 against the intracellular domain of RPTP have been described previously (4,6). Anti-HA tag and anti-Myc tag antibodies were obtained from culture supernatant of hybridoma cell lines 12CA5 and 9E10, respectively. Monoclonal antibodies against ␤-catenin, p120 ctn , and phosphotyrosine were purchased from Transduction Laboratories (Lexington, KY). Polyclonal anti-p120 ctn antiserum F1 has been described previously (30).
DNA Constructs-The pMT2 expression vector containing the fulllength RPTP cDNA has been described previously (31). RPTP-CS represents a double point-mutated RPTP cDNA in which the cysteine residues in the active sites of both the catalytic domains are replaced by serine residues. RPTP-CS was obtained by standard cloning and polymerase chain reaction procedures using primers 5Ј-GGTGGTGCAC-TCGAGTGCTGGTGCAGGGAGG-3Ј and 5Ј-GCACCAGCACTCGAGT-GCACCACCAGTGGGC-3Ј to mutate the first catalytic domain, and primers 5Ј-GTTGTGCATAGCTTGAACG-3Ј and 5Ј-GTTCAAGCTATG-CACAACG-3Ј to mutate the second catalytic domain. The C-terminally truncated RPTP-⌬2 construct was generated by deletion of the second catalytic domain through EcoRI digestion. Generation of construct RPTP-ExJ, which lacks both phosphatase domains but still contains the complete juxtamembrane region, has been described (5). Finally, the RPTP-ExT construct encodes the extracellular domain and the transmembrane region plus 12 additional amino acids to serve as membrane anchor. This construct was obtained by introducing a membraneproximal stop codon using primers 5Ј-CTTTGCTGCAGAATTCCCCGC-AGACAGCCTC-3Ј and 5Ј-CGCCTCTAGACTAGGTCTCTTTCCGCTT-CTTG-3Ј. Immunofluorescence and biochemical analysis showed that the RPTP-ExT protein was expressed and retained on the cell surface (data not shown). All sequences amplified by polymerase chain reaction were verified by DNA sequencing. Myc-tagged ␤-catenin cDNA was generously provided by Dr. Jorg Hulsken (Max Delbrü ck Centrum, Berlin, Germany) and subcloned into the pMT2 expression vector. An expression vector containing an HA-tagged form of p120 ctn 1A has been described previously (19). Constructs encoding an N-terminally truncated mutant (p120 N2) and the p120⌬R3-11 mutant that lacks Armadillo repeats 3-11 are described by Reynolds et al. (30). A pMT2 expression plasmid encoding a dominant form of p60 src (Y527F substitution) was constructed and provided by Dr. Dick Schaap (Organon Tecnika, Oss, The Netherlands).
Cell Lysis, Immunoprecipitation, and Blotting-Cells were washed once with cold PBS and lysed at 4°C in either Nonidet P-40 lysis mix (20 mM Tris, pH 7.5, 1% Nonidet P-40, 100 mM NaCl, 5 mM EDTA) or RIPA (20 mM Tris, pH 7.5, 1% Triton X-100, 0.1% SDS, 1% sodium deoxycholate, 150 mM NaCl, 5 mM EDTA). Lysis buffers were supplemented with 5 g/ml leupeptin, 2.5 g/ml aprotinin, and with the phosphatase inhibitors sodium vanadate and molybdate (100 M each). After centrifugation, supernatants were incubated for 4 h with Sepharose-protein A beads (Amersham Pharmacia Biotech), precoupled to the appropriate antibodies. Immunoprecipitates were washed three times in lysis buffer, boiled in SDS sample buffer, and separated by SDSpolyacrylamide gel electrophoresis. After transfer of proteins to nitrocellulose, blots were probed with specific antibodies using standard procedures and visualized using enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech) Production of Glutathione S-Transferase (GST) Fusion Proteins-The N-terminal fusion of the catalytic phosphatase domains of RPTP to GST has been described (31). The bacterial expression construct B5 encodes a GST fusion protein that contains part of the juxtamembrane region and both RPTP phosphatase domains. The B5M construct is identical to the B5 construct except for a single point mutation in the active site of the first catalytic domain. The resulting cysteine to serine substitution in the B5M construct was previously shown to abolish phosphatase activity in vitro (31). Inserts from the bacterial B5 and B5M expression constructs were subcloned into the baculo-transfer vector pVL1392 (PharMingen) resulting in pVL-C1C2 and pVL-M1C2, respectively. Recombinant baculoviruses were obtained using the Bacu-loGold transfection kit (PharMingen) according to the manufacturer's instructions. Two days after infection, Sf9 cells were lysed in Nonidet P-40 lysis mix supplemented with 2 mM dithiothreitol and protease inhibitors. GST fusion proteins were precipitated with Sepharose-glu-tathione beads (Amersham Pharmacia Biotech), washed in lysis buffer, and eluted from the beads in 50 mM Tris (pH 8.0) plus 5 mM reduced glutathione. The amount of purified fusion-protein was estimated from SDS-polyacrylamide gel electrophoresis, and aliquots were stored in 40% glycerol, 0.1 mg/ml bovine serum albumin, and 2 mM dithiothreitol at Ϫ20°C.
In Vitro Dephosphorylation Assay-To obtain tyrosine-phosphorylated p120 ctn , COS cells were cotransfected with pRC/CMV-p120 ctn and pMT2-Src 527F . Two days after transfection, cells were lysed in RIPA supplemented with protease inhibitors, 100 M sodium vanadate, and 100 M sodium molybdate. Lysates were cleared by centrifugation and immunoprecipitated using anti-HA antibody coupled to Sepharose-protein A beads (Amersham Pharmacia Biotech). Immunoprecipitates were washed four times in lysis buffer and two times in PTP buffer (40 mM imidazole, pH 7.2, 0.1 g/ml bovine serum albumin, 2 mM dithiothreitol) without phosphatase inhibitors. Immunocomplexes were then incubated in PTP buffer at 30°C with approximately 200 ng of GST-C1C2 or GST-M1C2 fusion proteins per time point. At the times indicated, aliquots were collected, mixed with 2ϫ SDS sample buffer, and boiled. Samples were separated on SDS-polyacrylamide, transferred onto nitrocellulose, and analyzed using anti-phosphotyrosine antibody.
In Vivo Dephosphorylation Assay-COS cells were cotransfected with expression plasmids encoding HA-tagged p120 ctn and either wild-type RPTP or the catalytically inactive point mutant RPTP-CS. Two days after transfection, cells were serum-starved for 6 h and stimulated with 10 ng/ml EGF for the times indicated. Cells were lysed in RIPA containing phosphatase inhibitors, and p120 ctn was immunoprecipitated using anti-HA tag antibody 12CA5 as described above. Total cell lysates were probed with anti-phosphotyrosine antibody PY20 or with antibody Ab37 directed against RPTP. Blots containing anti-p120 ctn immunocomplexes were first probed with PY20, stripped, and then reprobed with anti-HA tag antibody to control for equal expression and loading.
Immunofluorescence-Cells, grown overnight on coverslips, were washed once in PBS and then fixed in methanol for 5 min at Ϫ20°C. Coverslips were blocked in 10% normal goat serum and incubated with primary antibodies at room temperature for 1 h. Coverslips were then washed in PBS and incubated with Texas Red-conjugated goat antirabbit IgG (Molecular Probes, Eugene, OR) or fluorescein-conjugated goat anti-mouse IgG secondary antibodies (Zymed Laboratories Inc.). After washing in PBS, coverslips were mounted with VectaShield (Vector Laboratories Inc., Burlingame, CA) and analyzed by confocal laser scanning microscopy (model MRC-600, Bio-Rad).

RESULTS
RPTP Colocalizes with p120 ctn to Cell-Cell Contacts-Both RPTP and p120 ctn are known to localize to sites of cell-cell contact (6,19). We analyzed colocalization of RPTP and p120 ctn in A549 human lung carcinoma cells and mink lung cells, in which both proteins are endogenously expressed. Immunofluorescence analysis reveals that the localization of RPTP and p120 ctn exactly overlaps in regions of cell-cell contact (Fig. 1). Mink lung cells, at low confluence, form filopodiallike extensions that contact neighboring cells. Also in these specialized structures, RPTP and p120 ctn are seen to colocalize ( Fig. 1, C and D). In addition, there is a correlation between RPTP and p120 ctn protein levels, as indicated by the open and closed arrows (Fig. 1, D and E), which point to regions of relatively low and high expression of RPTP and p120 ctn , respectively. These findings raise the possibility of an interaction between RPTP and p120 ctn and/or related catenins.
RPTP Interacts with p120 ctn , but Not with ␤-Catenin-We set out to examine the possible interaction between RPTP and p120 ctn and/or the related Armadillo member ␤-catenin through coprecipitation studies. Small amounts of endogenous p120 ctn , but not ␤-catenin, were detected in anti-RPTP immunoprecipitates from human A549 as well as mink lung cells (data not shown).To investigate RPTP-catenin interactions in better detail, we used the COS cell expression system. Expression vectors encoding full-length RPTP and an HA-tagged p120 ctn construct were cotransfected, and anti-RPTP precipitates were probed for the presence of p120 ctn . In parallel experiments, we cotransfected ␤-catenin. Fig. 2 shows that RPTP associates with p120 ctn and that coprecipitation of p120 ctn is strictly dependent on the presence of RPTP (lanes 5 and 6). Similar association was found between p12 ctn and the RPTP-related RPTP (data not shown). We note that RPTP interacts with a relatively small amount of p120 ctn , similar to what has been reported for the interaction between cadherins and p120 ctn (32). This suggests that the RPTP-p120 ctn interaction occurs with relatively low affinity or, alternatively, with low stoichiometry. Probably because of this, reciprocal precipitations, in which we analyzed anti-p120 ctn immunoprecipitates for the presence of RPTP, were not successful.
Specific association of RPTP with p120 ctn is indicated by the findings that (i) the precipitating antibody does not crossreact with p120 ctn , and (ii) no interaction is detected with ␤-catenin (Fig. 2, lane 8) even after prolonged exposure. We note that epitope-tagged p120 ctn protein appears as multiple bands in total cell lysates (Fig. 2, lanes 1 and 2). The upper band migrates at 120 kDa, while the lower triplet runs around 100 kDa. Since p120 ctn is C-terminally tagged, the 100-kDa bands must represent breakdown products that lack part of the N terminus. These 100-kDa products do not interact with RPTP, suggesting that the N terminus of p120 ctn is required for binding to RPTP (as will be confirmed below). Several studies have shown that p120 ctn interacts with cadherins (19, 25, 26, 32), while conflicting data exist about an association between RPTP and cadherins (16 -18). Reprobing of the RPTP-p120 ctn immunocomplexes with anti-cadherin antibodies did not reveal a detectable cadherin signal in the complexes (data not shown). This suggests that the p120 ctn -RPTP complexes are distinct from p120 ctn -cadherin complexes.
Association Is Independent of the Tyrosine Phosphorylation State of p120 ctn -Since p120 ctn is a substrate for protein-tyrosine kinases (21,22,33), we tested how the RPTP-p120 ctn interaction depends on the phosphorylation state of p120 ctn . Cells cotransfected with RPTP and p120 ctn were treated with EGF or pervanadate, or cotransfected with activated p60 src (Src 527F ). Although p120 ctn was heavily tyrosine-phosphorylated in all three cases (see below), its binding to RPTP was not affected (data not shown). We therefore conclude that complex formation between p120 ctn and RPTP is phosphotyrosine-independent.
The Juxtamembrane and Second Phosphatase Domain of RPTP Are Involved in p120 ctn Binding-To map the regions of RPTP that interact with p120 ctn , we generated several mutant RPTP constructs and tested them for p120 ctn binding (Fig. 3A). Protein-protein interaction was analyzed by cotransfection and subsequent immunoprecipitation using an antibody against the RPTP ectodomain. Construct RPTP-CS is a catalytically inactive form of RPTP, in which critical cysteine residues in both PTP domains were mutated. Only the first, membrane-proximal PTP domain shows catalytic activity, with residue Cys 1095 being essential for activity (31). As shown in Fig. 3B (upper panel), p120 ctn coprecipitates equally well with wild-type RPTP and RPTP-CS, indicating that the catalytic activity of RPTP is dispensable for association with p120 ctn .
In mutants RPTP-⌬2 and RPTP-ExJ, the second or both PTP domains were deleted, respectively. Deletion of the second PTP domain significantly, though not completely, reduces the interaction with p120 ctn (Fig. 3B, lane 4, construct RPTP-⌬2). This suggests that the second PTP domain interacts with p120 ctn or, alternatively, that deletion of the second domain destabilizes p120 ctn binding to another binding site on RPTP. When both PTP domains are deleted (construct RPTP-ExJ), p120 ctn is still capable of interacting with similar efficacy to the RPTP-⌬2 mutant. This indicated that the first catalytic domain does not contribute to p120 ctn binding. However, the interaction with p120 ctn is completely abolished after deletion of the juxtamembrane region of RPTP (using construct RPTP-ExT, containing the ectodomain, the transmembrane domain and only 12 intracellular residues). Taken together, these results indicate that the juxtamembrane region of RPTP confers binding to p120 ctn , and that the second PTP domain may contain additional binding sites or else may serve to stabilize binding of p120 ctn to the juxtamembrane region.
The N Terminus of p120 ctn Mediates Binding to RPTP-p120 ctn consists of an N-terminal region of about 280 amino acids, a central core of 11 Armadillo repeats, and a unique C terminus of about 100 amino acids (Fig. 4A). To map the RPTP binding region, we co-expressed RPTP with an Nterminally deleted form of p120 ctn (p120⌬N2). In addition, we used a deletion mutant that lacks the Armadillo repeats 3-11 but has intact N and C termini (p120⌬R3-11). As shown in Fig.  4, both full-length p120 ctn and the p120⌬R3-11 mutant are detected in anti-RPTP immunoprecipitates. This indicates that the Armadillo repeats 3-11 are dispensable for binding to RPTP.
In contrast, when part of the N terminus of p120 ctn is deleted (amino acids 28 -233), RPTP binding is abolished. Thus, the binding of p120 ctn to RPTP is mediated by its unique N terminus without an apparent role for the Armadillo repeats. Identical results were obtained using RPTP (data not shown), indicating that p120 ctn can bind to both RPTP and RPTP via its N-terminal part. Of note, the association of p120 ctn with   FIG. 2. Association between p120 ctn and RPTP. cDNAs encoding RPTP, HA-tagged p120 ctn , and Myc-tagged ␤-catenin were cotransfected into COS cells as indicated. Total cell lysates (lanes 1-4) and anti-RPTP immunoprecipitations (lanes 5-8) were analyzed on Western blots. p120 ctn was detected by anti-HA antibody (upper lanes 1,  2, 5, and 6), ␤-catenin by anti-Myc antibody (upper lanes 3, 4, 7, and 8), and RPTP by antibody 3G4 (lower panels, lanes 1-8). Lanes 1-4 (total cell lysates) show expression controls. Note that p120 ctn is expressed as multiple bands (lanes 1 and 2) and that only the upper band associates with RPTP. cadherins is mediated by its Armadillo repeats and is not detected with the p120⌬R3-11 mutant (30). The finding that p120⌬R3-11 fails to bind cadherins but still associates with RPTP supports the notion that complex formation between p120 ctn and RPTP occurs independently of cadherins. p120 ctn Is Dephosphorylated by the First Catalytic Domain of RPTP-We tested whether tyrosine-phosphorylated p120 ctn can serve as a substrate for RPTP. Tyrosine-phosphorylated p120 ctn was purified from COS cells cotransfected with p120 ctn and Src 527F . The immune complexes were incubated with GST fusion proteins containing the PTP domains of RPTP (Fig. 5). Next, the tyrosine phosphorylation state of p120 ctn was monitored by anti-phosphotyrosine blotting. Fig. 5A shows that p120 ctn is tyrosine-phosphorylated following EGF stimulation or Src expression. Incubation of tyrosine-phosphorylated pp120 ctn with the PTP domains of RPTP (GST-C1C2) results in a reduced phosphotyrosine signal over time, indicating that RPTP can dephosphorylate p120 ctn (Fig. 5B). As anticipated, inactivation of the first PTP domain abolishes the phosphatase activity toward p120 ctn (GST-M1C2, Fig. 5C). The role of the second PTP domain, which lacks activity in vitro (31), has remained elusive to date. Our results suggest that the second domain may serve a role in presenting substrates to the first catalytic domain.
Dephosphorylation of p120 ctn by RPTP in Intact Cells-We next tested RPTP-induced dephosphorylation of p120 ctn in intact cells. To this end, p120 ctn and RPTP were co-expressed in COS cells. Phosphorylation of p120 ctn was induced by EGF and its dephosphorylation was followed over time by immunoprecipitation and anti-phosphotyrosine blotting. As shown in Fig. 6, EGF stimulation of COS cells results in massive tyrosine phosphorylation of total cellular protein within minutes after stimulation. The high phosphorylation state is maintained for at least 3 h, without any significant reduction observed in cells expressing either active or inactive RPTP. However, differences became apparent when p120 ctn was precipitated from total cell lysates and analyzed for phosphotyrosine content. First, it is seen that cotransfection of catalytically inactive RPTP results in phosphorylation of p120 ctn even in the absence of EGF (Fig. 6, compare lanes 1 and 9). Moreover, EGF-induced phosphorylation of p120 ctn is higher in cells transfected with inactive RPTP than in cells transfected with wild-type RPTP. This suggests that expression of inactive RPTP protects p120 ctn from dephosphorylation by endogenous PTPs. In addition, phosphorylation of p120 ctn decreases over time when active RPTP is co-expressed, a response not observed with inactive RPTP. Thus, tyrosine-phosphorylated p120 ctn serves as a substrate for RPTP both in vitro and in intact cells.

DISCUSSION
While many new members of the RPTP family have been identified in recent years, very little is still known about their intracellular binding partners and substrates. In the present study we demonstrate that RPTP, a prototypic RPTP that functions in homotypic cell-cell interactions (5,34), associates with the catenin p120 ctn . Since no cadherins can be detected in p120 ctn -RPTP complexes, we conclude that complex formation occurs independently of cadherins. Similar results were obtained with the related RPTP (data not shown). That the interaction is independent of cadherins is supported by the finding that a mutant form of p120 ctn , which lacks most of the Armadillo repeats (p120⌬R3-11), still interacts with RPTP but fails to bind cadherins (30). Since p120 ctn does not bind to ␣-catenin, ␤-catenin, ␥-catenin, or APC (24), it is unlikely that any of these proteins mediates the interaction between p120 ctn to RPTP. The tyrosine dephosphorylation of p120 ctn induced by RPTP, observed not only in vitro but also in intact cells, suggests that both proteins are in close proximity. However, we cannot exclude the formal possibility that p120 ctn binding to RPTP (or the closely related RPTP) involves other, as-yetunidentified proteins. Whether the relatively low amounts of FIG. 3. Mapping of the p120 ctn -interacting regions on RPTP. A, wild-type and mutant RPTP constructs used to analyze the p120 ctn -interacting regions. See "DNA Constructs" and "Materials and Methods" for details. B, empty vector (lane 1) or RPTP constructs were cotransfected with p120 ctn as indicated. Wild-type and mutant RPTP proteins were immunoprecipitated and analyzed on Western blot for the presence of p120 ctn (upper panel) and RPTP (middle panel). Expression controls for p120 ctn are shown in the lower panel.
FIG. 4. The N terminus of p120 ctn mediates binding to RPTP. A, constructs used to map the RPTP-interacting site of p120 ctn . See "DNA Constructs" for details. All constructs are HA epitope-tagged at their C termini. B, wild-type and mutant p120 ctn constructs were cotransfected with RPTP as indicated. To control for specificity, full-length p120 ctn was transfected without RPTP (lane 1). Cell lysates were immunoprecipitated with antibody 3D7, directed against the ectodomain of RPTP. Immunoprecipitates and total lysates were analyzed on Western blots using anti-HA antibody to detect p120 ctn proteins and polyclonal antibody 37 to detect RPTP. p120 ctn found in RPTP precipitates are indicative of a low affinity or a low stoichiometry interaction cannot be deduced from the present finding.
The p120 ctn protein is known to exist in at least four different isoforms (through alternative splicing; Refs. 35 and 36), two of which lack significant part of the N terminus. Through mutational analysis, we found that the N terminus of p120 ctn is required for RPTP binding. This raises the possibility that p120 ctn isoforms selectively associate with RPTP, depending on the presence of an intact N terminus. Further experiments are required to test this possibility. Mutational analysis further indicates that binding of p120 ctn to RPTP requires both the juxtamembrane region and the second PTP domain of RPTP, but is independent of RPTP phosphatase activity and on the tyrosine phosphorylation state of p120 ctn . We also showed that, in vitro, p120 ctn is dephosphorylated by a GST fusion protein containing both PTP domains of RPTP, and that only the first PTP domain is active toward p120 ctn . This suggests a model in which the second PTP domain may help to recruit p120 ctn , thereby bringing p120 in the correct orientation toward the first PTP domain. The second PTP domain may thus play a role in presenting substrates to the first PTP domain of RPTP and possibly other RPTPs. In this respect, it is of note that two proteins interacting with leukocyte common antigen-related RPTP were also shown to bind to the second PTP domain (37,38). Although these proteins themselves are unlikely to be direct substrates for leukocyte common antigen-related RPTP, they could be involved in binding and presenting substrates to the first catalytic domain.
To date, the physiological function of p120 ctn remains unknown. This precludes meaningful assessment of the biological significance of the p120 ctn -RPTP interaction and RPTP-mediated dephosphorylation of p120 ctn . Currently, the effect of tyrosine phosphorylation of p120 ctn on cadherin-binding is controversial; increased affinity toward cadherins has been observed by some (39,40), while others have reported little or no change (19,32). Interestingly, tyrosine phosphorylation and molecular organization of cadherin/catenin complexes in endothelial cells is regulated by cell density. At low cell density, p120 ctn is phosphorylated, whereas its phosphorylation and binding to cadherins is reduced at confluence (40). Given the up-regulation of RPTP expression at high cell density (6), it will be interesting to investigate whether RPTP (and RPTP) regulates the organization of cadherin/catenin complexes in a cell density-dependent manner. Overexpression of p120 ctn in NIH3T3 fibroblasts induces profound morphological changes characterized by the formation of dendrite-like extensions (30). This phenotype is specific for p120 ctn in that overexpressed ␤-catenin causes only minor shape changes. Furthermore, our preliminary studies reveal that overexpression of p120 ctn in fibroblasts induces the disassembly of actin stress fibers. 2 It thus seems likely that p120 ctn plays a role in the regulation of the actin cytoskeleton. Recently, Yap et al. (41) suggested that a function for p120 ctn in lateral clustering of cadherins, thereby regulating cell-cell adhesion. While the precise biological function of p120 ctn remains to be elucidated, our findings suggest that RPTP regulates the tyrosine phosphorylation state of p120 ctn , and thereby may signal and help to maintain p120 ctnmediated reorganization of the cortical actin cytoskeleton and/or clustering of membrane proteins at sites of cell-cell contact.
FIG. 5. p120 ctn is a substrate for RPTP in vitro. A, EGF or Src induce tyrosine phosphorylation of the p120 ctn protein. COS cells were transfected with p120 ctn alone (lanes 1 and 2) or together with an activated form of p60 src (lane 3). Following serum starvation, cells were stimulated with epidermal growth factor for 5 min (EGF, lane 2). Anti-p120 ctn immunocomplexes were split and probed on Western blot with antibodies against phosphotyrosine (anti-PY) and anti-HA tag. B and C, tyrosine-phosphorylated p120 ctn (pp120) was incubated with GST fusion proteins containing both wild-type catalytic domains of RPTP (GST-C1C2, panel B) or with a mutant form in which the first catalytic domain was inactivated (GST-M1C2, panel C). At the indicated time points, aliquots were removed from the reaction mixture and boiled in SDS-sample buffer. Samples were split and analyzed on Western blot for phosphotyrosine and p120 ctn content. The additional bands observed after addition of fusion protein represent background reactivity of the anti-phosphotyrosine antibody to the GST fusion proteins.
FIG. 6. In vivo dephosphorylation of p120 ctn by RPTP. COS cells cotransfected with p120 ctn and wild-type (left panels) or with catalytically inactive RPTP (RPTP-CS, right panels), were stimulated with EGF for the indicated times (in minutes and hours). Total lysates and p120 ctn immunoprecipitates were analyzed on Western blot as indicated. Note that overexpression of mutant RPTP-CS enhances the basal phosphorylation of p120 ctn in the absence of EGF (lane 9).