Identification of Src phosphorylation sites in the catenin p120ctn.

p120-catenin (p120(ctn)) interacts with the cytoplasmic tail of cadherins and is thought to regulate cadherin clustering during formation of adherens junctions. Several observations suggest that p120 can both positively and negatively regulate cadherin adhesiveness depending on signals that so far remain unidentified. Although p120 tyrosine phosphorylation is a leading candidate, the role of this modification in normal and Src-transformed cells remains unknown. Here, as a first step toward pinpointing this role, we have employed two-dimensional tryptic mapping to directly identify the major sites of Src-induced p120 phosphorylation. Eight sites were identified by direct mutation of candidate tyrosines to phenylalanine and elimination of the accompanying spots on the two-dimensional maps. Identical sites were observed in vitro and in vivo, strongly suggesting that the physiologically important sites have been correctly identified. Changing all of these sites to phenylalanine resulted in a p120 mutant, p120-8F, that could not be efficiently phosphorylated by Src and failed to interact with SHP-1, a tyrosine phosphatase shown previously to interact selectively with tyrosine-phosphorylated p120 in cells stimulated with epidermal growth factor. Using selected tyrosine to phenylalanine p120 mutants as dominant negative reagents, it may now be possible to selectively block events postulated to be dependent on p120 tyrosine phosphorylation.

Studies of endogenous Src family proteins suggest both positive and negative roles in adhesion. In one study, cell adhesion defects were observed in keratinocytes and epidermal cells from Fyn-deficient mice or from mice with double knockouts of Src and Fyn, suggesting that the activities of Src kinases are necessary for normal cell-cell adhesion (28). In contrast, another study showed that inhibition of the catalytic activity of the Src kinases promotes the stability of cadherin-dependent cell-cell contacts (29), suggesting that Src kinase activity is normally required for junction disassembly. The latter result is consistent with data from oncogene-transformed or mitogenstimulated cells, where increased tyrosine kinase activity is associated with weak junctions (for review see Ref. 30). The simplest explanation for these apparently contradictory findings is that controlled tyrosine kinase activity is necessary for both assembly and disassembly of adherens junctions, depending on regulatory influences that are poorly understood.
It has been suggested that constitutive tyrosine phosphorylation of p120 and ␤-catenin contributes to defects in cadherinmediated adhesion in Src-transformed cells (for review see Ref. 30). Although evidence suggests that these defects are independent of ␤-catenin tyrosine phosphorylation (31), a role for p120 tyrosine phosphorylation has not been excluded. Indeed, mutation of p120 tyrosine residue 217 can partially reverse the Src-induced defects in cadherin function (32) in L-cells. An unresolved problem, especially in Src-transformed cells, is how to separate the effects of p120 tyrosine phosphorylation from the simultaneous effects induced by tyrosine phosphorylation of other substrates. For example, most prominent Src substrates (e.g. tensin, focal adhesion kinase, Crk-associated substrate, actin filament-associated protein, cortactin, ezrin-radixin-moesin proteins, p120, ␤-catenin, etc.) are proteins that affect the organization of the actin cytoskeleton (33). Thus, it is possible that constitutive phosphorylation of these proteins affects cadherin function indirectly by negatively affecting the underlying actin framework that anchors the cadherin complex. Therefore, although a vast literature correlates cadherin malfunction to unscheduled tyrosine phosphorylation of the catenins, direct mechanistic evidence of a role for p120 phosphorylation is lacking.
To directly address the role of p120 tyrosine phosphorylation in cadherin function, we aim to generate dominant negative p120 Tyr 3 Phe mutant proteins that are selectively unable to be phosphorylated at key tyrosine residues. Here, as a first step toward this goal, we have employed two-dimensional tryptic mapping methods to identify all major sites of Src-induced p120 tyrosine phosphorylation. A method was developed to tyrosine phosphorylate p120 to high stoichiometry using a modification of the in vitro Src kinase assay. Deletion mapping revealed that virtually all p120 tyrosine phosphorylation occurs in a region amino-terminal to the first Arm repeat. Thus, we have named this region the p120 phosphorylation domain and suggest that it is involved in the regulation of p120 function. To confirm the identity of each site, individual candidate tyrosine residues were mutated to phenylalanine and assayed for elimination of the accompanying spot on the two-dimensional tryptic phosphopeptide map. Importantly, p120 tryptic maps from in vitro labeling experiments were nearly identical to p120 maps generated by [ 32 P]orthophosphate labeling in vivo, indicating that we have indeed mapped the sites that are likely to be relevant in vivo. Eight sites were identified, and changing these sites to phenylalanine resulted in a p120 mutant (p120 -8F) that could not be phosphorylated efficiently by Src and failed to interact with SHP-1, a phosphatase known to bind p120 in a tyrosine phosphorylation-dependent manner. Using mutants with individually altered tyrosine phosphorylation sites, it may now be possible to distinguish the specific effects of p120 phosphorylation from effects driven by phosphorylation of other substrates induced by Src and receptor tyrosine kinases.

EXPERIMENTAL PROCEDURES
Cell Culture and Transient Transfection-Cells were cultured in Dulbecco's modified Eagle's medium containing L-glutamine (Life Technologies, Inc.), 10% fetal bovine serum (Hyclone), and 1% penicillinstreptomycin (Life Technologies, Inc.). Cos-7 (African green monkey kidney) were obtained from American Type Culture Collection and from Dr. Steven Hanks (Vanderbilt University); HEK 293 (human embryonic kidney) cells were obtained from the American Type Culture Collection.
Transient transfections were performed with SuperFect transfection reagent (Qiagen) according to the manufacturer's directions, except that DNA-SuperFect complexes were allowed to form for 15 min prior to the addition to cells. In co-transfection experiments, equal microgram amounts of each plasmid were transfected (5 g of total plasmid/60-mm dish). Transfections were carried out with 5 l of SuperFect/1 g of DNA for Cos-7 cells and 6 l of SuperFect/1 g of DNA for 293 cells. Experiments on transfected cells were performed 1 day posttransfection.
Cell Lysis, Immunoprecipitation, and Western Blotting-Cell lysis, immunoprecipitation, and Western blotting procedures have been described in detail previously (13). Briefly, cells were washed one time in phosphate-buffered saline (10 mM phosphate, pH 7.4, 150 mM NaCl) and then lysed in 1 ml of ice-cold RIPA buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholic acid, 0.1% SDS) containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 2 g/ml aprotinin, 1 mM EDTA) for 5 min at 0°C. Lysates were clarified by centrifugation for 5 min. Monoclonal antibody (mAb) was added to the supernatant and incubated for 1 h at 4°C with end-overend rotation. 20 l of protein A-Sepharose bead slurry (Amersham Pharmacia Biotech; 1:1 slurry of beads and phosphate-buffered saline) coupled to a rabbit anti-mouse bridge antibody (Jackson Immunoresearch, Inc.) were then added to lysates and incubated for an additional 1 h at 4°C. Immunoprecipitates were washed five times with lysis buffer, resuspended in 2ϫ Laemmli sample buffer, and boiled for 3 min. Denatured proteins were separated by SDS-polyacrylamide gel electro-phoresis on 7% gels. The proteins were then transferred to polyvinylidene difluoride membranes (Millipore) for phosphopeptide mapping procedures or to nitrocellulose for Western blotting. Western blotting was performed with p120 mAb pp120 (Transduction Laboratories), diluted to 0.1 g/ml in 3% nonfat milk in Tris-buffered saline (10 mM Tris, pH 7.4, 150 mM NaCl) and with phosphotyrosine mAb PY20 (Transduction Laboratories) diluted to 1 g/ml in 5% bovine serum albumin in Tris-buffered saline. The blots were developed with ECL (Amersham Pharmacia Biotech) with the exception of the use of alkaline phosphatase color reaction for Fig. 1C (lower panel).
In Vivo Labeling of p120 -The cells were washed twice with Trisbuffered saline before being starved for 15 min in phosphate-free Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 4% dialyzed fetal bovine serum (Life Technologies, Inc.) (4 ml/100-mm dish, 1.5 ml/60-mm dish). 32 P-Labeled H 3 PO 4 (0.5 mCi/l in H 2 O, carrierfree; ICN Biomedicals Inc.) was then added directly to cells (1 mCi/ 100-mm dish, 350 Ci/60-mm dish). The cells were then incubated for 4 h before treatment with pervanadate (described below), followed by lysis. The cells were lysed with RIPA containing protease inhibitors and phosphatase inhibitors (pervanadate (0.3 mM Na 3 VO 4 , 0.6 mM H 2 O 2 ) and 50 mM NaF). The immunoprecipitations were carried out as described above. Transfected murine p120 was specifically immunoprecipitated with 3 g of mAb 8D11, which does not react with human or monkey p120 epitopes.
Pervanadate Treatment-The cells were incubated with pervanadate (0.1 mM Na 3 VO 4 , 0.2 mM H 2 O 2 ) for 6 min before being washed with phosphate-buffered saline containing pervanadate (0.3 mM Na 3 VO 4 , 0.6 mM H 2 O 2 ). Pervanadate was prepared fresh before each experiment by incubating appropriate concentrations of Na 3 VO 4 and H 2 O 2 for 20 min at room temperature before storing on ice until ready to use.
In Vitro Kinase Assays-Cos-7 cells were co-transfected in 60-mm dishes with an equal number of g of RcRSV/Src527F (chicken) and RcRSV/p120 -1A (murine) plasmid DNA. Kinase-dead c-Src(K430R) in pcDNA3.1 vector was obtained from Dr. Sarah Parsons (University of Virginia). Lysis and immunoprecipitation procedures were performed as described above, except that 3 g each of p120 mAb 8D11 and Src mAb 327 were added to each lysate to precipitate immune complexes of Src and p120, and only 10 l of protein A-Sepharose slurry coupled to rabbit anti-mouse bridge antibody was added to each immunoprecipitation reaction to allow saturation of the beads with immune complexes. Immunoprecipitates were washed four times in ice cold RIPA, followed by two times in Tris-buffered saline containing 0.1 mM Na 3 VO 4 . Src-p120 immune complexes were then incubated in 20 l of kinase buffer (20 mM HEPES, pH 7.4, 10 mM MnCl 2 ) containing either 10 Ci of [␥-32 P]ATP (4500 Ci/mmol, ICN Biomedicals Inc.) or 20 M unlabeled ATP (Sigma) for 10 min at room temperature with constant agitation. Reactions were terminated by washing twice with RIPA containing 5 mM EDTA followed by boiling in 2ϫ Laemmli sample buffer for 3 min.
Phosphopeptide Mapping and Phosphoamino Acid Analysis-Phosphoproteins labeled in vivo or in vitro were separated by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride. Polyvinylidene difluoride membranes were rinsed well with deionized H 2 O twice before visualization of phosphoproteins by autoradiography. The phosphoproteins were then excised, and the membrane pieces were re-wet with methanol followed by water. The membranes were blocked with 50 mM NH 4 HCO 3 containing 0.1% Tween 20 (Sigma) for 30 min at room temperature and washed three times with 50 mM NH 4 HCO 3 before enzymatic cleavage of phosphoproteins from the polyvinylidene difluoride with L-(tosylamido-2-phenyl) ethyl chloromethyl ketonetreated bovine pancreas trypsin (Worthington). The peptides were then subjected to two-dimensional phosphopeptide mapping according to the procedures described in detail by Boyle et al. (34) and VanDerGeer et al. (35). Performic acid oxidation was not routinely performed. Partial hydrolysis fingerprints were generated by incubation of phosphopeptides for 30 min in 6 N HCl.
Mutagenesis-p120 deletion mutants have been described previously (13). Point mutations were generated by QuickChange Mutagenesis (Stratagene) according to the manufacturer's instructions except that Pfu polymerase (Stratagene) was used for all amplification reactions, for high fidelity. All mutations were verified by DNA sequencing.
GST Pull-down Experiments-GST-SHP-1 constructs and methods for generating the GST fusion proteins were described previously (36). Briefly, overnight cultures of BL21(DE3) Escherichia coli transformed with pGEX-5X (Amersham Pharmacia Biotech) or pGEX 5X-SH2 SHP-1 plasmids were diluted 10 -20-fold into a fresh 200-ml culture of Luria broth containing 100 g/ml ampicillin and incubated at 37°C until A 600 ϭ ϳ0.7. Protein expression was then induced with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside at ambient temperature (25-29°C) for 14 h. Bacteria were harvested by centrifugation and resuspended in detergent buffer (10 mM NaH 2 PO 4 pH 7.4/150 mM NaCl/0.5% Nonidet P-40) containing 1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 2 g/ml aprotinin, 10 mM ␤-mercaptoethanol, and 1 mM EDTA. Bacterial suspensions were sonicated on ice, and lysate was clarified by centrifugation (14,000 ϫ g, 10 min) at 4°C. GST proteins were immobilized by incubating the clarified lysate with 250 l of glutathione-Sepharose beads (Sigma) for 1 h at 4°C with end-over-end rotation. The beads were washed twice with detergent buffer and twice with phosphate-buffered saline and resuspended to a final volume of 500 l to generate a 50:50 bead slurry for pull-down experiments. GST proteins were gel-quantified relative to known protein loading standards.
GST Pull-down of Tyrosine-phosphorylated p120 -Cos-7 cells were transiently co-transfected with Src527F and murine p120, and in vitro kinase assays were performed with the immunoprecipitated protein, as described above. Kinase assays were done in the presence of 20 M ATP (Sigma) rather than with [␥-32 P]ATP. Following kinase assays, beads were washed twice with RIPA and then boiled for 5 min in denaturation buffer (50 mM HEPES, pH 7.5, 1% SDS, 1% ␤-mercaptoethanol). After samples cooled to ambient temperature, an aliquot of sample was reserved for Tyr(P) and p120 Western blotting to determine the relative amount of p120 present in each pull-down reaction. Then 600 l of 0.5% Triton X-100 buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.5% Triton X-100) containing 1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 2 g/ml aprotinin, 1 mM EDTA, and 0.1 mM Na 3 VO 4 was added to each sample to dilute the SDS to less than 0.05%. The beads were discarded from samples, and the proteins were allowed to partially renature in this solution for 30 min at 4°C. Glutathione-Sepharose beads loaded with GST or SHP-1 SH2 domain GST fusion protein were added, and samples were incubated at 4°C for 1.5 h with end-over-end rotation. Then beads were washed three times with 50 mM HEPES, pH 7.4, 150 mM NaCl, 10% glycerol, 0.1% Triton X-100 and boiled in 2ϫ Laemmli sample buffer for 3 min.

Characterization of p120 Phosphorylation in Vivo and in
Vitro-To determine the extent of total p120 phosphorylation under steady state conditions in vivo, murine p120 was transiently overexpressed and orthophosphate-labeled in rapidly growing 293 cells. Two-dimensional tryptic phosphopeptide mapping of p120 resolved two peptides that were robustly phosphorylated, as well as a complex array of minor phosphopeptides (Fig. 1A). Phosphoamino acid analysis of the total protein revealed that a majority of the phosphorylation was on serine, with a relatively small amount of phosphorylation on threonine and tyrosine. Phosphoamino acid analysis of the two major peptides (Fig. 1A, peptides C and H) indicated that the phosphate was on serine (data not shown). Because p120 was initially discovered as a major in vivo substrate for Src tyrosine kinase, we overexpressed p120 together with Src527F, a constitutively active Src mutant. Overexpression of Src527F lead to a sharp increase in the phosphotyrosine content of p120, as determined by phosphoamino acid analysis (Fig. 1B, compare Ϫ and ϩSrc panels), producing a unique array of p120 phosphopeptides (Fig. 1A, panel ii). Because of the high stoichiometry of tyrosine phosphorylation, phosphoserine-containing peptides were no longer detectable at optimal exposure times for phosphotyrosine detection. Five-minute pervanadate treatment prior to cell lysis was required for optimal detection of these phosphorylation events, suggesting that the sites are rapidly dephosphorylated under normal circumstances.
To determine whether Src could phosphorylate p120 directly, we devised a modified in vitro kinase assay (see "Experimental Procedures") based on coimmunoprecipitation of p120 and Src527F, followed by incubation of the purified complexes with [␥-32 P]ATP and manganese, the relevant co-factor. Phospholabeling p120 by this method induced significantly higher stoichiometry of phosphorylation than could be obtained by in vivo labeling of cells with [ 32 P]orthophosphate while reducing working levels of radioactivity. p120 was heavily tyrosine-phosphorylated in the presence of Src527F (Fig. 1C, lane 3), but not in its absence (Fig. 1C, lane 2), or in the presence of kinase-dead c-Src (K430R) (Fig. 1C, lane 4), indicating that the coprecipitated Src was directly responsible for the phosphorylation of p120. In addition, phosphoamino acid analysis confirmed that the in vitro phosphorylation event occurred on tyrosine only (data not shown). In vitro phosphorylation of p120 produced a peptide pattern that was nearly identical to that produced upon Src overexpression in vivo (Fig. 1A, compare panels ii and iii), generating phosphopeptides with similar electrophoretic and chromatographic mobility as well as similar stoichiometry of phosphorylation among peptides. The tryptic maps of in vitro phosphorylated p120 were highly reproducible, although occasional differences in the relative intensity between peptides 4 and 5 and among peptides 1-3 were detected. Mixing experiments confirmed that the in vitro and in vivo labeled peptide co-migrated on the tryptic peptide maps (Fig. 1A, panel iv). Together, these data suggest that Src phosphorylates p120 directly and that the in vivo and in vitro phosphorylation sites are similar, if not identical.
Identification of the Major Sites of p120 Tyrosine Phospho- To confirm that the peptides phosphorylated in vivo and in vitro were the same, the two samples (from panels ii and iii) were mixed and run together (panel iv) to show that they exactly co-migrate. The horizontal arrow indicates the direction of electrophoresis at pH 1.9 (toward the cathode). The vertical arrow indicates the direction of chromatography. Exposure times were 1000 cpm, 3 days (panel i); 3000 cpm, 1.5 days (panel ii); 20,000 cpm, 2.5 h (panel iii); in vitro 500 cpm and in vivo 1000 cpm, 1 day (panel iv). B, phosphoamino acid analysis in the presence and the absence of activated Src. Human 293 cells were transiently co-transfected with murine p120 and either vector alone (ϪSrc 527F ) or the same vector containing activated Src (ϩSrc 527F ) and labeled in vivo with [ 32 P]orthophosphate. Immunoprecipitated p120 was then subjected to phosphoamino acid analysis. The locations of phosphoserine, phosphothreonine, and phosphotyrosine are circled. C, direct p120 phosphorylation by Src in vitro. Immune complexes containing the indicated (ϩ/Ϫ) combinations of p120 or Src were subjected to in vitro kinase assay, separated by SDS-polyacrylamide electrophoresis, and visualized by autoradiography. kd represents kinase-dead Src. p120 phosphorylation occurs only in the presence of both p120 and kinase active Src. 32 P autoradiogram exposure time, 15 s. rylation-Initially, a panel of p120 deletion mutants was assayed to identify regions of p120 that could be phosphorylated by Src. Only deletions in the amino terminus of p120 affected the phosphorylation pattern on tryptic maps (Fig. 2), in agreement with previous data indicating that most of the major phosphorylation sites were contained in an 18-kDa aminoterminal fragment generated by cleavage with the protease Endolysine C. 2 Deletion of the first 157 amino acids eliminated the major phosphopeptides 6, 9, and 10 ( Fig. 2). Deletion of amino acids 29 -233 eliminated one additional peptide, peptide 7, suggesting that three of the major phosphopeptides originated from amino acids 29 -157, whereas another peptide lay between amino acids 158 and 233. Finally, deletion of amino acids 233-387 eliminated the remaining major phosphopeptides, 1-5 and 8, suggesting the location of the remaining sites. In addition, the latter deletion eliminated peptide 7, and a new peptide designated peptide A was observed. These experiments place all of the major sites within the amino-terminal 387 residues.
Partial tryptic digestion frequently results in overlapping peptides containing the same phosphorylated residue. To determine which of the phosphopeptides shared amino acid sequence similarity, individual peptides were isolated and treated with hydrochloric acid to detect similarities in the pattern of partial hydrolysis products that are characteristic of each peptide (Fig. 3). These "peptide fingerprints" showed a relationship between peptides 1-3 (Fig. 3A), peptides 4 and 5 (Fig. 3B), peptides 9 and 10 (Fig. 3C), and peptides 7 and A (not shown). Furthermore, each of these sets of peptides behaved similarly when oxidized with performic acid (not shown).
Based on deletion mapping, peptide fingerprinting, and performic acid oxidation data (not shown), the search was narrowed to specific candidate residues. To definitively identify each site, suspected tyrosines were individually mutated to phenylalanine and re-evaluated by in vitro kinase assay and two-dimensional tryptic mapping (Fig. 4). Mutation of tyrosines 112, 228, and 280 to phenylalanine resulted in elimi-nation of phosphopeptide spots 6, 7, and 8, respectively. Mutation of tyrosine 96 eliminated spots 9 and 10, whereas mutation of tyrosine 257 eliminated spots 4 and 5. The latter results confirm the partial hydrolysis fingerprinting data, which sug-2 A. B. Reynolds, unpublished observations.

FIG. 2. Approximate localization of p120 phosphorylation sites by deletion analysis.
Wild type p120 and p120 deletion mutants were coexpressed with Src527F in Cos-7 cells and analyzed by in vitro kinase assay and two-dimensional tryptic mapping. A, schematic showing the boundaries of the p120 deletions and the loss of specific phosphopeptides from the associated two-dimensional tryptic maps of each mutant. B, two-dimensional tryptic maps of p120 deletion mutants phosphorylated in vitro by Src527F. Peptides eliminated by each deletion are illustrated by a circle around their predicted position in wild type maps. A unique peptide (peptide A) was present in the map of the ⌬233-387 p120 mutant. Exposure times were 10,000 cpm each, all exposed for 4.5 h except ⌬233-387, which was exposed for 3.75 h.ϫ, sample origin. The horizontal arrow indicates the direction of electrophoresis at pH 1.9 (toward the cathode). The vertical arrow indicates the direction of chromatography.
FIG. 3. Identification of related peptides by partial hydrolysis fingerprinting. Following two-dimensional tryptic mapping of in vitro phosphorylated p120, peptides 1-5, 9, and 10 were recovered, partially hydrolyzed in HCl, and subjected again to two-dimensional mapping. The pattern of partial hydrolysis products for peptides 1, 2, and 3 revealed a close relationship between these peptides (A). Likewise, nearly identical patterns were observed for peptides 4 and 5 (B), and peptides 9 and 10 (C). ϫ, sample origin. An arrow indicates the direction of electrophoresis in each dimension (toward the cathode). The pH levels of the electrophoresis buffers are as indicated. Exposure times were: peptide 1, 200 cpm, 5.5 days; peptide 2, 475 cpm, 4 days; peptide 3, 155 cpm, 5.5 days; peptide 4, 295 cpm, 4 days; peptide 5, 450 cpm, 4 d days; peptide 9, 315 cpm, 6 days; peptide 10, 117 cpm, 2 weeks. gested that each of these pairs of spots arose from partial tryptic digestion of the same sequence. Identification of Tyr 228 in peptide 7 is in accordance with the deletion analysis, which indicated that peptide 7 resided between amino acids 158 and 233. The unique peptide A present in p120⌬233-387 apparently results from the new fusion peptide formed by the deletion.
In addition, each p120 point mutant was transiently cotransfected with Src527F into 293 cells and labeled in vivo by the addition of [ 32 P]orthophosphate to the cell culture medium. Note that the transfected p120 mutants were generated from murine p120 and can therefore be selectively immunoprecipitated from human 293 cells using p120 mAb 8D11, which recognizes murine but not human p120. Two-dimensional tryptic maps generated by directly immunoprecipitating the trans-fected p120 from these cells demonstrated that the major peptides that were lost corresponded exactly to those identified in vitro (Fig. 4, compare in vitro and in vivo columns). These maps of in vivo labeled p120 also revealed occasional unique phosphopeptides that were not readily reproducible.
Tryptic mapping of point mutants initially failed to lead to identification of peptides 1, 2, and 3, which had been shown to be related to each other by partial hydrolysis fingerprinting (Fig. 3). To identify the phosphorylated tyrosines in these peptides, we explored the possibility that these spots arose from a single peptide that was phosphorylated at multiple sites. Because increased phosphorylation (and therefore, net negative charge) of a peptide decreases both electrophoretic potential and hydrophobicity, single peptides phosphorylated at multiple sites are predicted to migrate on a diagonal with positive slope. This pattern characterized the migration of peptides 1, 2, and 3. Peptide 3 migrated toward the cathode, indicating a low charge-to-mass ratio, and was relatively hydrophilic. Mutating Tyr 291 , Tyr 296 , or Tyr 302 , which lie in a single predicted tryptic fragment, each resulted in elimination of peptide 3 while concomitantly increasing the hydrophobicity of peptides 1 and 2 (Fig. 5, 296F). Further mutation of both Tyr 296 and Tyr 302 resulted in elimination of both peptides 2 and 3, and a further increase in hydrophobicity of peptide 1 (Fig. 5, 296/302F). These results suggest that peptides 1, 2, and 3 represent singly, doubly, and triply phosphorylated variants, respectively, of the same peptide, containing Tyr 291 , Tyr 296 , and Tyr 302 . Indeed, triple mutation of each of these putative phosphorylation sites completely eliminated peptides 1, 2, and 3 without detectably affecting the migration of any other phosphopeptides (Fig. 5, 291/296/302F). This result was confirmed by in vivo orthophosphate labeling of ectopic murine p120 containing Y291F/Y296F/Y302F mutations (not shown). The results of the point mutation analyses and the final assignments of each site are summarized in Table I.
Mutational Analysis of SHP-1 Binding to Tyrosine-phosphorylated p120 -SHP-1 is an SH2 domain-containing nonreceptor tyrosine phosphatase shown previously to interact with tyrosine-phosphorylated p120 (37). In theory, our mapping data should allow the identification of specific p120 tyrosine residues that mediate the interaction. To test this hypothesis and confirm the relevance of the identified sites, we performed a series of GST-SHP-1 pull-down experiments. GST was fused to the tandem SH2 domains of SHP-1 (generating GST-SHP-SH2) and used to pull down p120 mutant proteins containing specific tyrosine to phenylalanine mutations.
Consistent with previous data, p120 that was phosphorylated in vitro by Src associated tightly with the GST-SHP-SH2 probe but not with GST alone (Fig. 6). To test whether this association was dependent on p120 tyrosine phosphorylation, we repeated the pull-down assays with a p120 mutant where each of the eight identified tyrosine phosphorylation sites was changed to phenylalanine (i.e. p120 -8F). The p120 -8F mutant failed to interact with GST-SHP-SH2 (Fig. 6A, compare lanes  1-3). As expected, tyrosine phosphorylation of p120 -8F was nearly undetectable on anti-phosphotyrosine Western blots (Fig. 6C, lane 3). These data further confirm that the p120-SHP-1 interaction is tyrosine phosphorylation-dependent and suggest that the identified sites of p120 phosphorylation are sufficient to mediate the interaction with the SHP-1 SH2 domain.
On the other hand, no single p120 Tyr 3 Phe mutation was sufficient to significantly decrease the p120 SHP-1 SH2 interaction (data not shown). One possibility is that each of the two SH2 domains of SHP-1 interacts simultaneously with unique sites in p120. Alternatively, individual SH2 domains could FIG. 4

. Identification of major phosphorylation sites by mutational analysis in vitro (left panels) and in vivo (right panels).
Left panels, p120 mutants were labeled by in vitro Src kinase assay and subjected to two-dimensional tryptic mapping Exposure times were: 20,000 cpm each; all exposed for 1 h, except for 112F and 228F, which were exposed for 2.5 and 1.5 h, respectively. Right panels, p120 mutants were transiently co-expressed in 293 cells with Src527F, orthophosphate-labeled in vivo, and subjected to two-dimensional mapping. Exposure times were: Wild type, 112F, and 228F, 820 cpm, 2.3 days; 96F, 257F, and 280F, 700 cpm, 4.5 days. The data were compiled from independent, representative experiments. ϫ, sample origin. The horizontal arrow indicates the direction of electrophoresis at pH 1.9 (toward the cathode). The vertical arrow indicates the direction of chromatography.
interact with multiple p120 phosphotyrosines. Previous data suggest that the amino-terminal SH2 domain of SHP-1 was largely responsible for the interaction, because SHP-1 with a mutated amino-terminal SH2 domain could not interact with p120 (37). To further examine the apparent lack of specificity for individual phosphotyrosine motifs, we assayed additional p120 mutants containing combinations of Tyr 3 Phe mutated sites (Fig. 6). Some of these mutants did indeed uncouple the interaction significantly but not entirely. However, the extent of protein-protein interaction best correlated with the total number of phosphorylation sites left intact (Fig. 6A, compare  lanes 4 -7) rather than the presence or the absence of any specific site or combination of sites. Similar data were also obtained with full-length SHP-1 containing in its catalytic do-main a "trapping mutation," C455S, which renders the phosphatase unable to release its target (not shown). Therefore, although we were able to demonstrate dependence upon the tyrosine phosphorylation sites we identified, we were unable to define specific phosphorylation events that mediate the p120-SHP-1 interaction.

DISCUSSION
Although Src-and receptor-induced tyrosine phosphorylation of p120 has been demonstrated in many systems, it is unclear how this modification affects p120 or cadherin function. A major obstacle is the fact that these kinases regulate multiple substrates simultaneously, making it difficult to distinguish the effects of one modified substrate from another. Moreover, p120 is phosphorylated at multiple sites, each of which might have different functions. To pinpoint the specific effects of p120 in these processes, we aim to create specific dominant negative p120 mutants that lack the ability to be phosphorylated at key sites. As a first step toward this goal, we have identified the major Src-induced tyrosine phosphorylation sites on p120 and generated a panel of tyrosine to phenylalanine p120 mutants, including one in which all eight sites are mutated simultaneously (p120 -8F). The latter mutant may be particularly useful in screens designed to initially identify activities dependent on p120 tyrosine phosphorylation. Indeed, p120 -8F was not significantly tyrosine-phosphorylated in in vitro Src kinase assays, and it failed to interact with SHP-1, a nonreceptor tyrosine phosphatase shown previously to bind tyrosine-phosphorylated p120 via the SHP-1 amino-terminal SH2 domain (37). Fig. 7 shows the location of all phosphorylated and unphosphorylated p120 tyrosine residues. Of the 30 tyrosines, 11 are located carboxyl-terminal to the first Arm repeat, and none of these are phosphorylated by Src. Thus, the Arm domain does not appear to be an important target for phosphorylation. Of the remaining 19 tyrosines located amino-terminal to the first Arm repeat, eight were phosphorylated. Six of these occur in a relatively short segment between residues 228 and 302, an apparent phosphorylation domain. This region also contains the majority of p120 serine phosphorylation sites, 3 suggesting that most regulatory modification of p120 function is likely to occur through serine and tyrosine phosphorylation within this domain. Only two sites, tyrosines 96 and 112 occurred outside this region; residue 96 is present in isoform 1 (the most common form in mesenchymal cells) but absent from isoform 3 (the most common form in epithelial cells), suggesting a potential role in cell motility. Of the eight identified sites, only tyrosine 291 is not conserved among human, mouse, and chicken (phenylalanine in chicken p120).
Although the clustering of sites within the p120 phosphorylation domain was striking, it was apparently not the only factor influencing the choice of a particular residue by Src because many of the tyrosine residues in this region clearly were not phosphorylated. Consensus motifs for preferential phosphorylation by Src and receptor tyrosine kinases have been identified in vitro by screening peptide libraries (38). The predicted motifs based on such studies include 1) acidic residues favored but not necessary in the Ϫ3 and Ϫ2 positions upstream of the phosphorylation site, 2) a ␤-branched residue (isoleucine, valine) in the Ϫ1 position, 3) a small or acidic amino acid in the ϩ1 position, and 4) a large hydrophobic amino acid in the ϩ3 position (38). Indeed, aside from the single exception of tyrosine 662, all of the p120 tyrosine motifs that contained at least two of the four parameters described above were in fact identified as phosphorylation sites in our study. Of these, the most con-3 D. J. Mariner, unpublished data.
FIG. 5. Identification of peptides 1-3 as a single triply phosphorylated peptide. Wild type or mutated p120 (panels 296F, 296/ 302F, and 291/296/302F) was phosphorylated in vitro by Src527F and subjected to two-dimensional tryptic mapping. Mutation of Tyr 296 to Phe (panel 296F) resulted in increased hydrophobicity of peptides 1 and 2, resulting in an upward shift to positions 1Ј and 2Ј, respectively. Additionally, peptide 3 was no longer phosphorylated. The circles show the normal positions of these peptides on a wild type p120 map. Mutation of both Tyr 296 and Tyr 302 to Phe (panel 296/302F) further increased the hydrophobicity to peptide 1, resulting in an additional upward shift to position 1Љ, and peptide 2 was no longer phosphorylated. Mutation of Tyr 291 , Tyr 296 , and Tyr 302 to Phe (panel 291/296/302F), which are predicted to lie in a single tryptic peptide, completely abolished phosphorylation of peptides 1, 2, and 3, indicating that these peptides represent a single triply phosphorylated peptide. ϫ, sample origin. The horizontal arrow indicates the direction of electrophoresis at pH 1.9 (toward the cathode). The vertical arrow indicates the direction of chromatography. Exposure time was: 10,000 cpm, 3 h. sistent predictive features were the first criterion (acidic residues at positions Ϫ2 or Ϫ3) and third criterion (a small side chain residue, usually glycine, at position ϩ1). Interestingly, the tyrosine 662 motif lacked these features. Although particular motifs were apparently favored by Src, the clustering of these motifs in a focused region of p120 is unlikely to be accidental, and both location and sequence context may contribute to the choice of phosphorylation sites by Src. Importantly, when Src and p120 were coexpressed in vivo, the same sites were phosphorylated, strongly suggesting that the physiologically important sites have been correctly identified.
Two lines of evidence suggest that deletion of the aminoterminal end of p120 can eliminate a negative regulatory function (6,39), thereby restoring positive effects of p120 on adhesion. Indirect evidence implicates the loss of key serine phosphorylation events, which indeed, are probably located within the amino-terminal region that was deleted. 3 A second possibility is that the loss of tyrosine phosphorylation sites may account for these effects. Ozawa and Ohkubo (32) show that v-Src-induced effects on cadherin function are partially reversed by mutation of p120 tyrosine 217 (but not tyrosine 221) to phenylalanine. Our failure to detect phosphorylation of Tyr 217 raises the possibility that we have missed important residues. However, tyrosine residues 217, 221, and 228 are all contained within a single tryptic peptide, which is easily detected as spot 7 on our maps. In our hands, phenylalanine mutation of tyrosine residues 217 or 221 resulted in increased hydrophobicity of the peptide, as expected (data not shown), but only Tyr 228 mutation eliminated its phosphorylation (Fig.  4, 228F panels). Thus, it seems unlikely that Tyr 217 is directly phosphorylated by Src. A possible explanation for the apparent discrepancy with the data of Ozawa and Ohkubo is that mutation of tyrosine 217 may affect p120 function by mechanisms unrelated to tyrosine phosphorylation of that residue. In future studies, it will be important to discriminate between the several alternative mechanisms that might account for functional changes associated with deletion or mutation of the p120 amino-terminal end. Our work adds several new tyrosine phosphorylation sites to the list of possible modifications that might regulate the role of this domain.
The best known and by far the most prevalent role of protein tyrosine phosphorylation is to generate binding sites for proteins that contain phosphotyrosine-specific binding domains (e.g. SH2 and PTB) (38). By analogy, specific phosphopeptides in p120 may serve as docking sites for the recruitment of molecules that regulate cadherin function. It should be possible to identify such proteins by affinity chromatography using the p120 phosphopeptide motifs as ligand.
Recently, p120 was identified as a binding partner of the SHP-1 tyrosine phosphatase following activation of the epidermal growth factor receptor (37). SHP-1 belongs to a small family of nonreceptor tyrosine phosphatases, each containing two tandem SH2 domains in the amino-terminal end and a carboxyl-terminal catalytic domain (for review see Ref. 40). Surprisingly, none of the single Tyr 3 Phe mutations had any significant effect on p120 binding. Several of the p120 phosphomotifs partially overlap with the consensus binding site for the amino-terminal SH2 domain of SHP-1 (I/V/L/S)XpYXX(L/V) (41) (compare with peptide sequences in Table I). Thus, SHP-1 may be capable of binding to multiple sites in p120. Although previous work demonstrated that the interaction is largely mediated by the amino-terminal SH2 domain (37), we cannot TABLE I Sites of p120 tyrosine phosphorylation by Src Trypsin cleaves peptides after lysine or arginine but cuts inefficiently between adjacent arginines or lysines. It does not cut when a proline directly follows an arginine or lysine. The listed tryptic fragments are those that would be generated by complete tryptic digestion. Amino acid numbers are based on the murine p120 isoform 1A sequence. FIG. 6. The SHP-1 SH2 domains interact with p120 in a phosphotyrosine-dependent manner. Wild type p120 (isoform 1A) or p120 containing Tyr 3 Phe point mutations (see below) was phosphorylated in vitro by Src and removed from the beads by exposure to SDS and 1% ␤-mercaptoethanol (see "Experimental Procedures"). An aliquot of the sample was directly analyzed by Western blotting with p120 mAb pp120 (B) or with antibodies to phosphotyrosine (C). The remainder was partially renatured and coprecipitated in pull-down assays with GST alone (A, lane 1) or GST linked to the SH2 domains of SHP-1 (lanes 2-7). Lanes contain wild type p120 -1A (lanes 1 and 2)   and phosphorylated (open circles on stems) tyrosines are indicated relative to the overall domain structure of p120. All of the major Src phosphorylation sites reside in the amino terminus. Six of the these cluster in a short region amino-terminal to the first Armadillo repeat (shaded), which we refer to as the phosphorylation domain (hatched). RNA splicing leads to alternative usage of four different ATG start sites (designated by arrows 1-4). Isoforms 1 and 3 are the most abundant forms, whereas isoforms 2 and 4 are rarely observed. rule out some contribution from the carboxyl-terminal SH2 domain. However, assaying several combinations of Tyr 3 Phe p120 mutants suggested that the strength of the interaction best correlated with the number of p120 phosphotyrosine residues left intact. Thus, although p120 tyrosine phosphorylation was clearly prerequisite for SHP-1 binding, we were unable to identify particular sites that specifically mediated the SHP-1 interaction. The epidermal growth factor receptor phosphorylation domain behaves similarly with regard to Shc, which binds to each of the epidermal growth factor receptor autophosphorylation sites with some degree of preference but without absolute specificity for a single site (42).
Because tyrosine phosphorylation of p120 was originally reported to correlate with Src-induced cell transformation, we overexpressed our p120 -8F mutant in Src transformed NIH3T3 or MDCK cells to screen for any obvious dominant negative effects. Despite the inability of the p120 -8F protein to be phosphorylated by Src, expression of this mutant did not appear to affect the morphology or actin cytoskeleton in these cells. 3 These data do not rule out an important role for p120 in the transformation process. Rather, they highlight the difficulty in discriminating among the pleiotropic effects of activated Src kinases. Thus, although activated Src has been a useful tool for identifying Src substrates and potentially important sites of tyrosine phosphorylation, it is evident that the use of activated Src as a model system may not be a good choice for studying the normal role of p120 phosphorylation in cadherin complexes.
The activity of c-Src is tightly regulated in cells. In the context of cadherin function, p120 in nascent cell-cell contacts is believed to be transiently tyrosine-phosphorylated (43), which reportedly increases its affinity for cadherins (28, 44 -47). Technically, it is difficult to detect such transient and low level events, especially in the context of cell-cell adhesion. p120 mutants containing appropriate tyrosine to phenylalanine mutations should act in dominant negative fashion to block such events, thereby directly implicating the site in p120 function. Moreover, it should now be possible to efficiently screen for putative p120 phosphotyrosine-dependent binding partners and to generate p120 tyrosine phospho-specific monoclonal antibodies, developments that will greatly facilitate the identification of upstream and downstream events associated with p120 tyrosine phosphorylation.