Regulation of E-cadherin/Catenin Association by Tyrosine Phosphorylation*

Alteration of cadherin-mediated cell-cell adhesion is frequently associated to tyrosine phosphorylation of p120- and β-catenins. We have examined the role of this modification in these proteins in the control of β-catenin/E-cadherin binding usingin vitro assays with recombinant proteins. Recombinant pp60c-src efficiently phosphorylated both catenins in vitro, with stoichiometries of 1.5 and 2.0 mol of phosphate/mol of protein for β-catenin and p120-catenin, respectively. pp60c-src phosphorylation had opposing effects on the affinities of β-catenin and p120 for the cytosolic domain of E-cadherin; it decreased (in the case of β-catenin) or increased (for p120) catenin/E-cadherin binding. However, a role for p120-catenin in the modulation of β-catenin/E-cadherin binding was not observed, since addition of phosphorylated p120-catenin did not modify the affinity of phosphorylated (or unphosphorylated) β-catenin for E-cadherin. The phosphorylated Tyr residues were identified as Tyr-86 and Tyr-654. Experiments using point mutants in these two residues indicated that, although Tyr-86 was a better substrate for pp60c-src , only modification of Tyr-654 was relevant for the interaction with E-cadherin. Transient transfections of different mutants demonstrated that Tyr-654 is phosphorylated in conditions in which adherens junctions are disrupted and evidenced that binding of β-catenin to E-cadherin in vivo is controlled by phosphorylation of β-catenin Tyr-654.

E-cadherin is the predominant cadherin in epithelial tissue and, as a member of this family, is responsible for the correct establishment and maintenance of adherens junctions, through Ca 2ϩ -dependent homophilic interactions with adjacent cells (1). The adhesive function of E-cadherin requires its attachment to the actin cytoskeleton, association mediated by a set of proteins collectively named catenins. In adherens junctions, the intracellular domain of E-cadherin binds directly to ␤-catenin that, in turn, associates with ␣-catenin, which is thought to link cadherin complexes to the actin cytoskeleton (2)(3)(4)(5). In addition to plakoglobin, that plays a role similar to that of ␤-catenin but in desmosomes, other proteins have also been shown to be associated to cadherin complexes. The presence of EGF 1 receptor, phosphotyrosine phosphatases or tyrosine kinase substrates as p120, has suggested a role for tyrosine phosphorylation in the modulation of these complexes. This hypothesis has been supported by data showing that maintenance of the adhesion complex is modulated by Src family tyrosine kinases, as pp60 c-src itself; or receptors with tyrosine kinase activity of growth factors like EGF, hepatocyte growth factor, platelet-derived growth factor, and colony stimulating factor-1 (6 -10). ␤and p120-catenins are considered to be the tyrosine phosphorylation-sensitive components of the adhesion complexes, since both proteins are phosphorylated in response to stimuli that disrupt adherens junctions and dissociate Ecadherin from the cytoskeleton (11)(12)(13).
Several hypotheses have been proposed to explain the adhesive changes of cadherin complexes in response to catenin tyrosine phosphorylation. These hypotheses include 1) alterations in the affinity of p120-and ␤-catenin for the cytoplasmatic region of E-cadherin; 2) conformational changes in these components that result in the disruption of linkage within cadherin and the cytoskeleton; or 3) recruitment of unknown proteins, or perhaps p120, to the cadherin/catenin complexes that implicate the dissociation of their components (14). Since a direct relationship of ␤-catenin phosphorylation to E-cadherin loss of function has not been established yet, it is also possible that, as proposed by some authors, E-cadherin dissociation from the cytoskeleton after cell transformation is not directly related to tyrosine phosphorylation of catenins.
We were particularly interested to evaluate the different possibilities that led to the disassembly of adherens junctions upon tyrosine phosphorylation of ␤-catenin. We have studied the ability of ␤-catenin and p120-catenin to bind E-cadherin in vitro, and the modulation of this binding by tyrosine phosphorylation. affinity chromatography on glutathione-Sepharose columns, the fusion protein was cleaved with PreScission protease (Amersham Pharmacia Biotech). A DNA fragment corresponding to amino acids 196 to end of murine p120-catenin was obtained from pcDNA3-Cas1 plasmid (provided by Dr. Albert Reynolds, Vanderbilt University, Nashville, TN) by digestion with BglII and NotI, and cloned into BamHI-NotI sites of pGEX-6P1 (Amersham Pharmacia Biotech). The expressed fusion GST-p120-catenin protein was purified and cleaved as above. A DNA fragment that corresponds to the cytosolic domain of murine E-cadherin was amplified by polymerase chain reaction using oligos corresponding to the nucleotide sequences 2268 -2282 and 2720 -2706, and containing BamHI and EcoRI sites at their 5Ј-ends, respectively. The 0.5-kilobase amplification fragment was digested with BamHI and EcoRI and cloned in the same sites of pGEX-6P3 plasmid. The absence of mutations in this fragment was verified by sequencing. GST-cytoEcad fusion proteins were prepared and isolated as above. Purified proteins were aliquoted and stored in 50% (v/v) glycerol at Ϫ40°C until use.
In Vitro Phosphorylation Assay-In vitro assays were performed in a final volume of 30 l in the following conditions: 25 mM Tris-HCl, pH 6.8, 25 mM MgCl 2 , 5 mM MnCl 2 , 0.5 mM EGTA, 0.25 mM VaO 4 Na 3 , 1 mM dithiothreitol, 0.1 mM [␥-32 P]ATP (1000 cpm/pmol), 1 g of either ␤-catenin or p120-catenin, and 0.5 units of recombinant pp60 c-src protein kinase (from Upstate Biotechnology, Inc.). Reactions were performed at 22°C for 0.5-4 h. Samples were analyzed by SDS-polyacrylamide gel electrophoresis in standard 10% polyacrylamide gels. After electrophoresis, gels were fixed in 25% methanol, 10% acetic acid, dried, and exposed to x-ray film at Ϫ80°C for 12 h. Quantitation of ␤-catenin and p120 phosphorylation was determined excising the band with a razor blade and counting in a liquid scintillation radioactivity detector.
In some cases, phosphorylation reactions were performed in the same conditions using GST fusion proteins as substrates and GST as control. The reactions were finished by diluting the samples in phosphatebuffered saline (PBS) (10 mM phosphate buffer, pH 7.4, 136 mM NaCl, 4 mM KCl), and 20 l of glutathione-Sepharose beads (50% suspension in PBS) were added and incubated with the samples for 30 min. The resin was pelleted down and washed three times with PBS, and the radioactivity present in the pellet was estimated as above.
Catenin-Cytosolic E-cadherin Binding Assay-When indicated 50 g of recombinant p120-or ␤-catenin were incubated with 9 units of pp60 v-src in the conditions indicated above, but using 0.1 mM ATP. As a control, reactions were performed in the same conditions but in the absence of ATP. Different concentrations of p120-or ␤-catenin ranging from 0.12 to 12 pmol (which correspond to 11 ng to 1.1 g for ␤-catenin and 9.4 ng to 0.94 g for p120-catenin) were incubated with a constant amount, 1.2 pmol (60 ng), of GST-cytoEcad fusion protein or with the same pmol of GST (30 ng) as control. Incubations were performed in the presence of 50 mM Tris-HCl, pH 7.3, 150 mM NaCl, 3 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, and 0.1% (w/v) Triton X-100, in a final volume of 200 l for 45 min at room temperature. GST-cytoEcad/ catenin complexes were isolated by incubation with 40 l of a 50% slurry of glutathione-Sepharose beads. After 30 min of gentle agitation, beads were collected by spinning in a microcentrifuge and washed three times with binding buffer. Bound protein complexes were solubilized in electrophoresis sample buffer and boiled for 5 min. Samples were separated by 10% SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose filters, and analyzed by Western blot with monoclonal antibodies (mAbs) against p120-or ␤-catenin (Transduction Laboratories, Lexington, KY). Immunoblots were developed with peroxidaseconjugated secondary antibody followed by an enhanced chemiluminescence detection system (ECL, Pierce). Phosphorylated catenin samples that were subjected to cytoEcad binding were analyzed in parallel by Western blot with an anti-phosphotyrosine (Tyr(P))-specific mAb (Transduction Laboratories), to check that catenin phosphorylation was completed.
In order to quantitate the amount of ␤-catenin bound to E-cadherin, the autoradiograms were scanned in a densitometer and the results were compared with those obtained with p120 or ␤-catenin standards analyzed by Western blot in parallel. Data obtained from the binding of five or six different concentrations of catenin (each one performed in duplicate) were analyzed by Scatchard plot. The ratio of bound to free catenin was plotted versus the concentration of catenin bound to GST-cytoEcad; the association constant (K a ) was determined from the slope of the regression line graph thus obtained. Determination of K a was repeated from four independent sets of measurements, and the mean Ϯ standard deviation of these four association constants was calculated.
Transfection to Caco-2 cells and Analysis of ␤-Catenin Phosphorylation and ␤-Catenin-E-cadherin Association-Wild-type or mutant ␤-catenin cDNAs were inserted in the BamHI site of pcDNA3.1 His (C) plasmid (Invitrogen, Carlsbad, CA). This plasmid labels the NH 2 end of expressing proteins with a polyhistidine tag, which facilitates their purification, and with an epitope recognized by an anti-Xpress TM antibody. 50% confluent Caco-2 cells, grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (Life Technologies, Inc.), were transfected with these plasmids Using LipofectAMINE (Life Technologies, Inc.) according the instructions of the manufacturer. After transfection, cells were incubated for 30 h in Dulbecco's modified Eagle's medium plus fetal calf serum; when indicated, cells were incubated the last 6 h with 0.25 mM sodium orthovanadate (Na 3 VaO 4 ). Cells extracts were prepared in RIPA buffer (20 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% sodium deoxycholate, 1% Nonidet P-40, 0.1% SDS, 1 mM EDTA) supplemented with 10 g/ml aprotinin, 20 g/ml leupeptin, 1 mM PMSF, and 0.5 mM Na 3 VaO 4 . Lysates were centrifuged at 13,000 rpm in a microcentrifuge for 5 min at 4°C. 250 g of extract were incubated in a final volume of 0.2 ml with 15 l of a 50% (w/v) suspension of nickel-NTA-agarose (Qiagen, Hilden, Germany) for 30 min at 4°C. Beads were washed with RIPA buffer, and bound proteins were eluted with electrophoresis sample buffer. Samples were analyzed by SDS-PAGE and Western blot using antibodies against ␤-catenin (from Transduction Laboratories). In order to reprobe the membranes, blots were stripped as described (15) and re-analyzed with mAbs against Tyr(P) or E-cadherin (Transduction Laboratories). The absence of signal after stripping was always checked incubating with the correspondent secondary antibody and ECL reagent.

RESULTS
Full-length recombinant ␤-catenin or a fragment of p120catenin comprising all the arm repeats were produced as GST fusion proteins and purified as indicated under "Experimental Procedures." The purity of these preparations is shown in Fig.  1A. Both catenins were phosphorylated in vitro by recombinant pp60 c-src (Fig. 1B); only Tyr residues were modified (data not shown). ␤-catenin or p120 phosphorylation (Fig. 1B) was totally blocked by addition of herbimycin, a pp60 c-src specific inhibitor (16). Similar results were obtained with a purified preparation of pp60 v-src (data not shown).
The stoichiometry of these phosphorylations was investigated. 1 unit of pp60 c-src was able to incorporate up to 2.4 nmol of P/1.6 nmol of ␤-catenin or 2.7 nmol of P/1.3 nmol of p120catenin, when incubated at 22°C (Fig. 1C). At 30°C, the reactions were faster but proceeded to a lower extent, probably because pp60 c-src was being inactivated (data not shown). These stoichiometries of phosphorylation, 1.5 and 2 nmol of P/nmol of ␤or p120-catenin, respectively, indicated that, in both cases, more than one site was being modified.
To determine the relevance of this modification, the abilities of Tyr-phosphorylated catenins to bind E-cadherin were analyzed and compared with those of the unmodified proteins. In order to perform this assay, the cytosolic domain of E-cadherin (cytoEcad) was prepared and purified as a GST fusion protein (Fig. 1A). It has been previously described that a limited fragment of this E-cadherin-domain is sufficient to bind ␤-catenin in vivo as well as in vitro (17). Different amounts of recombinant p120-and ␤-catenins, from 0.12 to 12 pmol, were incubated separately with 1.2 pmol of GST-cytoEcad. The complexes were bound to glutathione-Sepharose and the presence of the catenins associated to the beads was analyzed by Western blot with specific mAbs (Fig. 2). The results were compared with the signal obtained from known amounts of standards that were included in the Western blot analysis. Scatchard analysis were carried out and the association constant (K a ) was determined from the slope of the regression line obtained. K a for E-cadherin/␤-catenin or E-cadherin/p120-catenin were estimated to be 1.0 Ϯ 0.2 ϫ 10 8 and 3.6 Ϯ 0.5 ϫ 10 6 M Ϫ1 , respectively (Table I). The value of this constant for ␤-catenin/Ecadherin binding was similar to that obtained by Obama and Ozawa (18) for the association of this same catenin to ␣-catenin. p120-catenin presented a lower affinity, as has been suggested previously (14). Phosphorylation of ␤-catenin by pp60 c-src significantly decreased the affinity by E-cadherin ( Fig.  2A); the association constant was 5-fold lower (Table I). On the other hand, phosphorylation of p120-catenin promoted the opposite effect; association of this catenin to E-cadherin was increased by a factor of 4 ( Fig. 2B and Table I).
It has been reported previously that p120-catenin does not bind to the same region of the cytosolic domain of E-cadherin than ␤-catenin. However, a possibility to be considered consisted in that the association of p120-catenin could hamper ␤-catenin binding only when both proteins were tyrosine-phosphorylated. Therefore, ␤-catenin binding assays were performed in the presence of p120-catenin. The addition of increasing amounts of this last protein (in a molar excess up to 4), either phosphorylated or not, did not modify the binding of control or Tyr-phosphorylated ␤-catenin to cytoEcad (Fig. 2C). The association constants measured in these conditions were not substantially different from those calculated for ␤-catenin in the absence of p120 (Table I).
Since ␤-catenin phosphorylation presented a functional relevance for E-cadherin activity, the identification of the modified Tyr residues was pursued. Phosphorylation of several ␤-catenin deletion mutants demonstrated the existence of Tyr residues capable to be phosphorylated in two different parts of the molecule (Fig. 3). A short N-terminal mutant, comprising amino acids 1-106, was efficiently phosphorylated, incorporating approximately 1 pmol of phosphate pmol of protein. No significant differences were detected between the amount of phosphate incorporated by this form and a longer N-terminal mutant (amino acids 1-575). This result suggests that amino acids placed between 106 and 575 were not phosphorylated by pp60 c-src . The C-terminal tail of the molecule (575-783) was also substrate of this tyrosine kinase, although in a lower extent than the N terminus (approximately 0.5 pmol of phosphate/pmol of protein). Another shorter C-terminal fragment (575-693) was phosphorylated identically to the complete Ctail, indicating that the modified residue (or residues) was confined to this short sequence. In accordance, a fragment comprising residues 693-783 was not phosphorylated (Fig. 3).
The Tyr residues present in the 1-106 and 575-693 fragments were analyzed and compared with sequences of well known substrates of Src protein kinases. None of the Tyr residues present in the 575-693 fragment fit well to the optimal phosphorylation sequence of this kinase, but among the three Tyr, Tyr-654 showed the best match. Tyr-670 was discarded because it contained three basic amino acids at position ϩ1 to ϩ3, a feature that makes phosphorylation by pp60 c-src difficult (19 -21). In respect to Tyr-604, Tyr-654 presents an acidic residue on the N-terminal side (position Ϫ5), two Ala residues at position ϩ1 and ϩ2, and a Lys at ϩ7, all characteristics repeatedly found in Src substrates.
Three Tyr residues are also present in the 1-106 ␤-catenin fragment. Among these, Tyr-86 presents the best features: Arg at position ϩ7, Ala at position ϩ1 and, especially, two Asp residues upstream (Ϫ3 and Ϫ6). In addition, it contains a Gln at Ϫ1, another feature found in pp60 c-src substrates (19,20). None, or only one, of these characteristics was observed in the sequences surrounding other Tyr residues.
Therefore, the best candidates, Tyr-654 and Tyr-86, were mutagenized to Phe residues and mutant ␤-catenins were expressed and purified (Fig. 4A). Phosphorylation experiments revealed that our predictions were correct; both Tyr-86 3 Phe and Tyr-654 3 Phe were significantly less phosphorylated by pp60 c-src than the wild-type ␤-catenin (Fig. 4). The phosphorylation of the double mutant was undetectable (Fig. 4). As expected from the experiments performed with the deletion mutants, that suggested a higher stoichiometry of phosphorylation for Tyr-86 than for Tyr-654 (Fig. 3), the presence of the muta- tion Tyr-86 3 Phe diminished more ␤-catenin tyrosine phosphorylation than the mutation Tyr-654 3 Phe (Fig. 4).
The capacity of both ␤-catenin mutants to bind E-cadherin was also assayed and compared with that of wild-type ␤-catenin. As shown in Fig. 5A, Tyr-86 3 Phe and Tyr-654 3 Phe ␤-catenin mutants associated to cytoEcad with a very similar affinity to the wild-type form: the association constants were not significantly different (Table I). pp60 c-src phosphorylation decreased binding of Tyr-86 3 Phe ␤-catenin mutant to cytoEcad in a similar fashion to binding of wild-type ␤-catenin, but it did not alter binding of Tyr-654 3 Phe mutant ( Fig. 5A and Table I). This result suggests that phosphorylation of Tyr-654 is relevant for the decrease in cytoEcad binding caused by pp60 c-src .
To verify this hypothesis, the two Tyr residues were mutagenized to Glu and mutant proteins expressed and purified (Fig. 5B). This mutation introduces a negative charge in the position of Tyr and mimics the effect of phosphorylation of this residue. The results of the binding experiments indicate that, whereas Tyr-86 3 Glu ␤-catenin associated to cytoE-cad in a manner identical to that for the wild-type form, Tyr-654 3 Glu did it much worse (Fig. 5C); the K a for this interaction was even lower than that calculated for ␤-catenin phosphorylated with pp60 c-src (Table I). This fact is probably due to the incomplete phosphorylation of Tyr-654 in our in vitro assays.
The functional relevance of Tyr-654 phosphorylation was also evidenced in vivo. Caco-2 cells were transiently transfected with wild-type ␤-catenin or the two Tyr-654 mutants (Tyr-654 3 Phe, Tyr-654 3 Glu) labeled with a polyhistidine tag, to facilitate their purification. After 24 h, cells were incubated with the Tyr(P) phosphatase inhibitor Na 3 VaO 4 , a compound that increases ␤-catenin phosphorylation in Tyr residues, at the same time that dissociates adherens junctions (15,22,23). Transfected forms of ␤-catenin were purified by Ni 2ϩ -agarose chromatography, and their phosphorylation status was analyzed. As shown in Fig. 6A, phosphorylation of wild-type ␤-catenin was remarkably increased by incubation with Na 3 VaO 4. On the other hand, the levels of Tyr(P) present in the two Tyr-654 mutants were not increased by this compound. Curiously, the

TABLE I Affinity constants for binding of cytoEcad to ␤-catenin or p120-catenin
Binding assays were performed as described under "Experimental Procedures." The affinity constants presented are the mean Ϯ standard deviation of the constants calculated in four different experiments, each one performed with five or six different concentrations of p120-or ␤-catenin. In the case of the ␤-catenin plus p120-catenin binding assay, only the K a for E-cadherin/␤-catenin association is shown. In these assays the ratio of ␤-catenin to p120-catenin was kept at 1:4 at all the concentrations tested. Representative binding assays corresponding to the data presented here are shown in Figs ␤-Catenin ϩ p120-catenin (4-fold molar excess) 8.2 Ϯ 0.5 ␤-Catenin (phosphorylated) ϩ p120-catenin (phosphorylated) (4-fold molar excess) 0.77 Ϯ 0.14 basal Tyr(P) content of the three forms was very similar. This experiment suggests that Na 3 VaO 4 induces phosphorylation of Tyr-654, but this residue is not phosphorylated in basal conditions. As previously mentioned, Na 3 VaO 4 effects on tyrosine phosphorylation are accompanied by a lower number of E-cadherin/ ␤-catenin cellular complexes. E-cadherin was retained by Ni 2ϩagarose only when ␤-catenin was bound to this resin (Fig. 6A). After addition of Na 3 VaO 4 , the amount of E-cadherin that copurified with ␤-catenin was substantially reduced. Similar levels of E-cadherin were observed in the beads when wild-type ␤-catenin or Tyr-654 3 Phe mutant were expressed (Fig. 6A). However, the amount of E-cadherin that copurified with ␤-catenin Tyr-654 3 Phe was not modified by Na 3 VaO 4 as this form is not phosphorylated.
The association of E-cadherin to ␤-catenin mutants Tyr-654 3 Glu and Tyr-86 3 Glu was also analyzed as these forms behave as constitutively phosphorylated. The amount of Ecadherin associated to Tyr-654 3 Glu mutant was substantially lower than to the wild-type form (Fig. 6, A and B) and was not modified by addition of Na 3 VaO 4 . Unlike Tyr-654 3 Glu, Tyr-86 3 Glu interacted with E-cadherin identically than wildtype ␤-catenin (Fig. 6B).
All these results indicate that Tyr-654 phosphorylation is relevant for the modulation in vivo of the interaction ␤-catenin-E-cadherin.

DISCUSSION
Several authors have determined that de-assembly of adhesion junctions was frequently associated to augmented tyrosine phosphorylation of proteins present in the complex (14,(22)(23)(24). Among these proteins, the most relevant changes have been detected in ␤-catenin; it has been assumed that increased tyrosine phosphorylation of this protein was the cause of Ecadherin-loss-of-function. However, a conclusive relationship had not been established so far, and some authors have proposed alternative explanations (25). In this article we show that tyrosine phosphorylation of ␤-catenin decreases its binding to E-cadherin in assays using purified proteins, evidencing a direct cause-effect relationship. We have identified the Tyr involved in regulation of this binding as Tyr-654 and have demonstrated that modification of this residue alters the modulation of E-cadherin/␤-catenin binding in vivo.
In our in vitro assays, ␤-catenin phosphorylation was performed by pp60 c-src tyrosine kinase. This protein kinase, or its viral homologue pp60 v-src , has been reported to promote deassembly of adherens junction, loss of function of E-cadherin, and phosphorylation of ␤-catenin in Madin-Darby canine kidney cells. Our results suggest that pp60 c-src might be the tyrosine kinase directly responsible for the phosphorylation and, thus, the inactivation of ␤-catenin. However, it cannot be discounted that another protein-tyrosine kinase, activated by pp60 c-src , is the direct ␤-catenin kinase, or contributes to the complete tyrosine phosphorylation of this protein. At this respect, it should be remembered that the two Tyr residues modified are not phosphorylated with the same efficiency. It is possible that, in vivo, pp60 c-src is catalyzing the direct phosphorylation of Tyr-86 but not of Tyr-654. In addition to pp60 c-src , other plausible candidates to be responsible for Tyr-654 phos- were performed in the indicated conditions but using as substrate the four GST-␤-catenin deletion mutants indicated. Reactions were stopped by diluting the reaction in glutathione-Sepharose binding buffer, and binding to this resin was performed as described. After washing three times, the amount of radioactivity associated to glutathione-Sepharose was determined. No radioactivity was detected associated to the resin when GST was used as substrate.
phorylation are the Fer kinase and the EGF receptor, two tyrosine kinases that are associated to ␤-catenin (26,27). In this respect, both EGF receptor and its homologue the proto-oncogene c-erb-2 have been shown to associate to the last three armadillo repeats of ␤-catenin (10), precisely the region of this protein where Tyr-654 is located.
Altered cell adhesion has been reported to be also associated Coomassie staining is shown. Panel C, a binding assay similar to that described in panel A was also performed with equivalent amounts of Tyr-86 3 Glu (Y86E) or Tyr-654 3 Glu (Y654E) mutants. No signal was detected when GST was used as bait. Although not shown, the anti-␤catenin mAb used in our assays reacted identically with all the ␤-catenin forms analyzed.
FIG. 6. In vivo phosphorylation and binding to E-cadherin of ␤-catenin point mutants. Wild-type or mutant ␤-catenin cDNAs were inserted into pcDNA3 plasmid, labeled with a polyhistidine tag, and transfected to Caco-2 cells as described under "Experimental Procedures." After 30 h, cell extracts were prepared from mock-transfected cells or from cells where the indicated mutants have been introduced. When mentioned, cells were incubated with Na 3 VaO 4 (250 mM) for the last 6 h. Expressed ␤-catenin forms were purified by chromatography on nickel-agarose, and proteins bound to the column were subjected to SDS-PAGE and Western blot anti-␤-catenin mAb to check that similar levels of expression were obtained in all the cases. Membranes were stripped and re-analyzed with mAbs against Tyr(P) or E-cadherin.
to disruption of ␣-catenin/␤-catenin binding (28). Tyrosine phosphorylation of unknown ␤-catenin residues seems to be responsible of this effect. Dissociation of these proteins is probably required for ␤-catenin to be transported to the nucleus and work as a transcriptional cofactor (reviewed in 29). We have tested the ability of our mutants to bind recombinant ␣-catenin. No changes in the in vitro association of these two proteins were observed after phosphorylation of ␤-catenin by pp60 c-src or when binding to ␣-catenin of Tyr-86 3 Glu and Tyr-654 3 Glu mutants were compared with the wild-type form. 2 These results suggest that pp60 c-src is not directly responsible for the modification of the Tyr residue that modulates ␣-catenin/␤catenin association and that neither Tyr-654 nor Tyr-86 are involved in this process. A more plausible candidate for this role is Tyr-143, the only Tyr residue present in the ␤-catenin domain necessary for ␣-catenin binding (30). In any case, the physiological relevance of tyrosine phosphorylation in the control of ␤-catenin/␣-catenin binding is likely to be dependent on the cell system, since in Caco-2 cells treatment with Na 3 VaO 4 did not modify the amount of ␣-catenin/␤-catenin binding. No differences were observed in the amount of ␣-catenin that copurified with transfected ␤-catenin in the experiment described in Fig. 6 after treatment with Na 3 VaO 4 (data not shown).
p120-catenin was initially characterized as a substrate of the activated pp60 v-src kinase and only recently has been described its binding to E-cadherin (14). Our group (15) and others (24) have reported that this protein is tyrosine-phosphorylated in conditions of E-cadherin-loss-of-function and that this modification is accompanied by a greater binding of p120-catenin to E-cadherin. These results have suggested that p120-catenin could act as a negative effector of adherens junctions, decreasing E-cadherin/␤-catenin binding. The data described in this paper indicated that this is not the case. Tyrosine-phosphorylated p120 associates better to E-cadherin than the unmodified catenin but does not compete with the binding of ␤-catenin. The exact role of p120-catenin phosphorylation, and its subsequent binding to E-cadherin, in the loss of the adhesive properties of this protein still has to be demonstrated. The description by Gumbiner and co-workers (31) that p120-catenin binds to Ecadherin in a cytoplasmic region necessary for E-cadherin clustering suggests that this protein might be involved in the regulation of this process.
Tyr-654, the tyrosine residue that controls ␤-catenin/E-cadherin binding, maps in the twelfth and last armadillo repeat of ␤-catenin (32). The structure of the armadillo repeats of ␤-catenin has been elucidated; they form a highly elongated domain composed entirely of ␣-helices with a long, positively charged groove. This structure is thought to accommodate the 30-amino acid sequence present in cytoEcad shown to be sufficient for binding to ␤-catenin (33). This E-cadherin sequence is quite acidic (pI of 3.3), like the sequences of other proteins known also to interact with ␤-catenin armadillo repeats. Therefore, it has been proposed that E-cadherin/␤-catenin interaction is dominated by ion pairs over an extended region, more than a close steric complementarity on a more limited area. It is easy to speculate that the introduction of a phosphate group, and thus, a negative charge, in the armadillo domain might affect this interaction, either diminishing the number of ion pairs or partially closing the long groove. The fact that the modified residue is not in the central part of this structure may help to explain the armadillo domain is still able to associate to other proteins after phosphorylation, as adenomatous polyposis coli (30, 34 -35).