PTP1B Regulates Cortactin Tyrosine Phosphorylation by Targeting Tyr446*

The emergence of protein-tyrosine phosphatase 1B (PTP1B) as a potential drug target for treatment of diabetes, obesity, and cancer underlies the importance of understanding its full range of cellular functions. Here, we have identified cortactin, a central regulator of actin cytoskeletal dynamics, as a substrate of PTP1B. A trapping mutant of PTP1B binds cortactin at the phosphorylation site Tyr446, the regulation and function of which have not previously been characterized. We show that phosphorylation of cortactin Tyr446 is induced by hyperosmolarity and potentiates apoptotic signaling during prolonged hyperosmotic stress. This study advances the importance of Tyr446 in the regulation of cortactin and provides a potential mechanism to explain the effects of PTP1B on processes including cell adhesion, migration, and tumorigenesis.

The emergence of protein-tyrosine phosphatase 1B (PTP1B) as a potential drug target for treatment of diabetes, obesity, and cancer underlies the importance of understanding its full range of cellular functions. Here, we have identified cortactin, a central regulator of actin cytoskeletal dynamics, as a substrate of PTP1B. A trapping mutant of PTP1B binds cortactin at the phosphorylation site Tyr 446 , the regulation and function of which have not previously been characterized. We show that phosphorylation of cortactin Tyr 446 is induced by hyperosmolarity and potentiates apoptotic signaling during prolonged hyperosmotic stress. This study advances the importance of Tyr 446 in the regulation of cortactin and provides a potential mechanism to explain the effects of PTP1B on processes including cell adhesion, migration, and tumorigenesis.
Protein-tyrosine phosphatase (PTP) 4 1B is recognized as an important regulator of metabolic signaling in mice. Ablation of the gene encoding PTP1B, Ptpn1, causes tissue-specific hypersensitivity to insulin and leptin, resulting in resistance to diabetes and obesity (1,2). PTP1B is a ubiquitously expressed enzyme that is localized to the cytoplasmic face of the endoplasmic reticulum (3). Its catalytic domain can directly dephosphorylate and inactivate the insulin receptor and other receptor and nonreceptor protein-tyrosine kinases (4). Despite the proto-oncogenic functions of many of these protein-tyrosine kinases, Ptpn1-null mice are not prone to tumorigenesis. On the contrary, two recent studies found that mice lacking PTP1B are markedly resistant to mammary tumorigenesis induced by active mutants of the receptor tyrosine kinase ErbB2 (5,6).
The effect of PTP1B deficiency in these cancer models has highlighted its diverse cellular functions. One emerging role of PTP1B is in the regulation of the actin cytoskeleton. Early stud-ies in fibroblasts showed that PTP1B is required for proper cell adhesion and spreading on extracellular matrix proteins (7). Cell adhesion to the extracellular matrix is mediated largely by cell-surface integrins, which initiate signaling cascades and orchestrate changes in the actin cytoskeleton at adhesive contacts. Protein-tyrosine kinases involved in transmission of integrin signals, including c-Src (8), focal adhesion kinase (9), and Csk (10), have been proposed as PTP1B substrates; however, none has consistently been shown to be hyperphosphorylated in the absence of PTP1B. It is tempting to speculate that, rather than regulating upstream signaling, PTP1B may directly target one or more non-protein-tyrosine kinase actin regulatory proteins.
Cortactin was identified as a prominent tyrosine phosphoprotein in cells expressing the active protein-tyrosine kinase v-Src (11). It has subsequently been implicated in various processes requiring dynamic actin assembly, including cell adhesion and migration, vesicular transport, and microbial infection (12,13). Cortactin exerts its effects on the actin cytoskeleton by interacting directly with the Arp2/3 complex (via its N-terminal acidic domain) (see Fig. 3), F-actin (via a central repeat region), and other actin regulatory proteins such as N-WASP and MIM (via its C-terminal SH3 domain) (12). A variety of additional binding partners mediate the effects of cortactin in specific contexts, including receptor tyrosine kinase down-regulation and cell-cell adhesion.
Cortactin is tyrosine-phosphorylated in response to a wide range of stimuli that induce cytoskeletal rearrangement, including growth factor stimulation, cell adhesion, and hyperosmotic stress (14). Src phosphorylates murine cortactin predominantly at three key sites in vitro, Tyr 421 , Tyr 466 , and Tyr 482 (corresponding to Tyr 421 , Tyr 470 , and Try 486 in human cortactin), resulting in decreased actin cross-linking activity (15). The combined mutation of these three residues abolishes tyrosine phosphorylation of cortactin in cells under various conditions (15)(16)(17). Thus, these Src sites have been the focus for functional characterization of cortactin tyrosine phosphorylation. Nonetheless, several mass spectrometry-phosphoproteomic studies have identified additional phosphorylated tyrosine residues (18 -24). A number of individual phosphotyrosine sites have been reported independently in different cell types and in response to diverse stimuli, but their regulation and function remain to be investigated.
In this study, we have identified cortactin as a substrate of PTP1B. A trapping mutant of PTP1B binds cortactin at a previously uncharacterized tyrosine residue, Tyr 446 . We show that PTP1B regulates cortactin phosphorylation induced by hyperosmolarity and that Tyr 446 is required for protection from apo-ptosis induced by hyperosmotic stress. Our results are the first to implicate a specific PTP in the regulation of cortactin, and they reveal a novel mechanism by which PTP1B may influence the cytoskeleton.
EGF and Sucrose Stimulation-For overexpression experiments, HeLa cells (total of 10 6 cells/12-well plate) were transfected with 0.5 g of pcDNA3-cortactin (WT or Y446F), pcDNA3.1-PTP1B (WT or C215S), or a control empty vector (pcDNA3.1) and grown for 36 h in normal growth medium. Cells were then serum-starved in DMEM for 4 h prior to treatment with hyperosmotic sucrose (300 mM sucrose in DMEM) or human EGF (100 nM; R&D Systems). For inhibitor experiments, untransfected HeLa cells were serum-starved for 2 h in DMEM followed by 2 h in DMEM containing a cell-permeable PTP1B inhibitor (5 M final concentration) as described previously (28). The hyperosmotic sucrose medium used for cell stimulation contained the same concentration of inhibitor. For all experiments, cells were lysed, and protein extracts were analyzed by immunoblotting as described previously (26).
Apoptosis Assays-HeLa cells plated in 6-cm dishes were transfected with pcDNA3-cortactin (WT or Y446F). At 24 h post-transfection, cells were trypsinized and seeded at 10 4 cells/ well in a 96-well plate. Cells were allowed to adhere for 10 h and subsequently serum-starved in 0.1% fetal bovine serum-containing DMEM for 16 h prior to treatment with hyperosmotic sucrose (300 mM in DMEM). At the indicated time points, caspase activity was measured using the luminescent Caspase-Glo 3/7 assay (Promega) following the manufacturer's protocol.

RESULTS
Identification of Cortactin as a PTP1B Substrate-To identify novel PTP1B substrates, a cell-based pulldown assay was employed in which GST-tagged WT PTP1B or mutant D181A was coexpressed in COS-7 cells with v-Src. The interaction of WT PTP1B with its substrates is normally transient. Asp 181 is a critical catalytic residue, and the D181A trapping mutant is able to form a stable, covalently linked complex with phosphorylated target proteins. The proteins bound to these two forms of PTP1B were purified using glutathione beads. As expected, the total cell lysate of COS-7 cells expressing v-Src contained numerous tyrosine phosphoproteins, as demonstrated by immunoblotting with an anti-phosphotyrosine antibody (Fig.  1A, third lane). Although WT PTP1B did not associate significantly with any of these proteins (first lane), the D181A mutant bound preferentially to two phosphoproteins of ϳ80 and ϳ65 kDa (second lane).
By immunoblotting for candidate substrates, we found that cortactin bound PTP1B D181A and that its molecular mass precisely matches the ϳ80-kDa protein we had detected by anti-phosphotyrosine blotting. The trapping mutant of PTP1B was able to bind endogenous cortactin in both HeLa and COS-7 cells (Fig. 1B). The form of cortactin expressed in COS-7 (monkey) cells migrated somewhat slower than that in HeLa (human) cells. An alternative antibody that selectively detects human cortactin was used to confirm binding of PTP1B D181A to cortactin in HeLa cells (supplemental Fig. 1). In both cell types, expression of v-Src enhanced, but was not required, for the cortactin-PTP1B interaction, suggesting that the basal phosphorylation of cortactin is sufficient for binding. An association of WT PTP1B with cortactin was never detected, even after long exposures, indicating that like many of its substrates, cortactin and WT PTP1B do not form a stable complex.
PTP1B belongs to the family of 38 classical, phosphotyrosine-specific PTPs (29). We examined the specificity of the cortactin-PTP1B interaction by testing the binding of cortactin to two other classical PTPs, TCPTP and PTP-PEST. TCPTP is the most similar enzyme to PTP1B, and although it is predominantly nuclearly localized, many of its known substrates, including receptor tyrosine kinases and STAT (signal transducer and activator of transcription) transcription factors, overlap with PTP1B (4). PTP-PEST is a cytoplasmic PTP that is a well established regulator of the actin cytoskeleton. The WT and trapping mutant forms of PTP1B, TCPTP, and PTP-PEST were coexpressed as GST fusion proteins in COS-7 cells with WT cortactin (Fig. 2). After precipitation using glutathione beads, bound cortactin was detected only in cells expressing PTP1B D181A. Thus, among the candidate enzymes tested, the ability to interact with cortactin is specific to PTP1B.
Mapping of the Cortactin-PTP1B Interaction-To localize the PTP1B-binding site on cortactin, we tested whether cortactin deletion mutants lacking the SH3 domain or the entire C terminus (Fig. 3A) could be trapped by PTP1B D181A. Despite potential interactions between the cortactin SH3 domain and the PTP1B proline-rich motifs, the ⌬SH3 mutant retained binding to PTP1B (Fig. 3B). Similarly, deletion of either proline-rich motif of PTP1B D181A did not impair its ability to pull down cortactin (data not shown). In contrast, deletion of the cortactin C terminus, including the SH3 and ␣-helical/prolinerich regions, prevented the binding of PTP1B (Fig. 3B). The ␣-helical/proline-rich region contains a cluster of tyrosine residues including the canonical Src sites (Fig. 3A). Tyrosine-tophenylalanine mutants were generated for each of these residues, and binding to PTP1B D181A was assessed after cotransfection. Strikingly, mutation of a single tyrosine residue, Tyr 446 , nearly abolished the interaction (Fig. 3C). The results of independent binding assays were quantified by densitometry (Fig. 3D) (18 -22, 24), and multiple species alignment of cortactin protein sequences indicates that Tyr 446 has been evolutionarily conserved (Fig. 4A). These results suggest that Tyr 446 is a functionally important site for  the regulation of cortactin. To evaluate the role of PTP1B in modulating cortactin phosphorylation, a polyclonal antibody for human cortactin phospho-Tyr 446 was prepared by immunization of rabbits with a corresponding phosphopeptide (Fig. 4A). The specificity of the anti-phospho-Tyr 446 antibody was tested using lysates from transfected HeLa cells treated with EGF. The overexpressed cortactin, which corresponds to the long isoform of the human protein (30), migrated slightly slower on SDS-polyacrylamide gel than the endogenous form expressed in these cells. Immunoblot analysis showed that the antibody bound endogenous and overexpressed WT cortactin, but not the overexpressed Y446F mutant (Fig. 4B). The reactivity of the antibody with cortactin was increased with EGF treatment, reaching a maximum at a stimulation time of 15 min. These results confirm that Tyr 446 is a target of EGF signaling and that this novel phospho-specific antibody is specific for this site.

PTP1B Regulates Cortactin Tyrosine
Phosphorylation-In initial experiments, we found that PTP1B overexpression dramatically decreased cortactin phosphorylation in response to EGF (data not shown). However, the EGF receptor and other receptor tyrosine kinases that promote cortactin phosphorylation are themselves substrates of PTP1B. Thus, the effect of PTP1B overexpression in this context is likely due to its cumulative activity on upstream receptor tyrosine kinases as well as on cortactin itself.
Hyperosmotic media can potently induce tyrosine phosphorylation of cortactin independently of receptor tyrosine kinases. Cell shrinkage caused by hyperosmolarity leads to successive activation of the Fyn and Fer cytoplasmic protein-tyrosine kinases (16). Activated Fer was believed to phosphorylate cortactin at the previously identified Src sites, Tyr 421 , Tyr 470 , and Tyr 486 , because mutation of these residues abrogates hyperosmolarity-induced cortactin tyrosine phosphorylation (16). However, our results show that Tyr 446 is also targeted under these conditions: treatment with hyperosmotic sucrose induced Tyr 446 phosphorylation in HeLa cells expressing WT cortactin (endogenous or overexpressed) (Fig.  5A). Immunoblotting with an antibody specific for Tyr 421 showed that, as expected, this site was also phosphorylated in response to hyperosmolarity. Notably, we found that the Y446F mutation resulted in decreased Tyr 421 phosphorylation (Fig. 5A), suggesting that the phosphorylation state of these residues is interdependent.
To establish the role of PTP1B in this process, we tested whether altering its expression or activity affects hyperosmolarity-induced cortactin phosphorylation. Transient overexpression of WT PTP1B reduced phosphorylation at both Tyr 446 and Tyr 421 (Fig. 5B), whereas an inactive mutant of PTP1B (C215S) had no effect. Correspondingly, preincubation of cells with a membrane-permeable PTP1B inhibitor enhanced hyperosmolarity-induced phosphorylation of these sites (Fig. 5C). Quantitation of data from multiple experiments showed that inhibitor treatment resulted in an ϳ50% increase in Tyr 446 phosphorylation at the basal level and after stimulation with hyperosmotic sucrose (Fig. 5D). Thus, the cellular effects of PTP1B overexpression and inhibition are consistent with its characterization as a cortactin phosphatase. A, the structure of full-length (FL) human cortactin is shown, with the positions of the 6.5 tandem repeats (TR), the ␣-helical/proline-rich (AHPR) region, and the C-terminal (C-term) SH3 domain indicated. The ␣-helical/proline-rich region contains nine tyrosine residues, including the three Src sites (denoted by asterisks). B, deletion of the entire C terminus of cortactin, but not the SH3 domain alone, disrupts binding to PTP1B. WT cortactin and mutants were coexpressed in COS-7 cells with GST-PTP1B (D181A or WT), and interacting proteins were detected by immunoblotting (IB) after GST pulldown (PD). C, mutation of cortactin at Tyr 446 blocks binding of PTP1B. GST-PTP1B D181A-cortactin binding was tested as described for B. For the control (Cont) assay, GST-WT PTP1B was coexpressed with WT cortactin. D, the results of three independent binding experiments are quantified. The band density of cortactin isolated by pulldown and in the total cell lysate (TCL) was quantified by densitometry. Error bars represent S.E.

Mutation of Cortactin Tyr 446 Increases Hyperosmolarity-induced
Apoptosis-Remodeling of the actin cytoskeleton is a protective mechanism in cells exposed to a hyperosmotic envi-ronment. Nonetheless, prolonged exposure leads to apoptotic cell death in a variety of cell types (31). A recent study showed that a single tyrosine residue on the protein-tyrosine kinase focal adhesion kinase, phosphorylated during hyperosmotic stress, is important for protection from apoptosis (32). Because cortactin is tyrosine-phosphorylated in response to hyperosmolarity (33), we investigated whether cortactin Tyr 446 could also play a cytoprotective role. Hyperosmotic sucrose induced apoptosis in HeLa cells as indicated by robust caspase-3/7 activation (Fig. 6A). After a 6-h treatment, transient overexpression of cortactin Y446F resulted in a significant ϳ30% increase in caspase activation compared with cells overexpressing the WT protein (Fig. 6B). This result indicates that mutation of Tyr 446 potentiates apoptotic signaling in HeLa cells under hyperosmotic stress.

DISCUSSION
The effect of PTP1B on the onset of diabetes, obesity, and breast tumorigenesis in mice has focused attention on its potential as a therapeutic target. However, the mechanisms by which it accomplishes some of its known cellular functions, including the modulation of cell-extracellular matrix adhesion, remain unclear. Here, we have shown that the actin regulatory protein cortactin is a target of PTP1B. In COS-7 cells expressing v-Src, the D181A trapping mutant of PTP1B binds selectively to cortactin, at ϳ80 kDa, and a second protein of ϳ65 kDa. This degree of specificity is remarkable considering the multitude of tyrosine phosphoproteins present in these cells and the numerous known substrates of PTP1B. Previously, our laboratory identified p62Dok as a PTP1B substrate in mouse fibroblasts using a similar approach (34). However, on the basis of immunoblotting data (not shown), we suspect that the ϳ65-kDa protein is not p62Dok, and we are currently investigating its identity. The ability of PTP1B to bind cortactin is not shared by the related enzymes TCPTP and PTP-PEST. The specificity of the cortactin-PTP1B interaction is further demonstrated by the fact that their binding depends largely on a single tyrosine residue, Tyr 446 . Published phosphoproteomic data provide convincing evidence that cortactin Tyr 446 is a bona fide phosphorylation site: phospho-Tyr 446 peptides have been detected following activation or overexpression of receptor tyrosine kinases (18,20,21,23) as well as in cells treated with pervanadate, a general PTP inhibitor (19,22,24). Here, using immunological A polyclonal antibody was produced by immunizing rabbits with a peptide corresponding to the sequence surrounding Tyr 446 (boxed). B, Tyr 446 is phosphorylated in response to EGF. HeLa cells overexpressing WT cortactin or Y446F were treated with EGF for the indicated times, and lysates were analyzed by immunoblotting (IB). FIGURE 5. PTP1B regulates sucrose-induced phosphorylation of cortactin. A, cortactin (cort) Tyr 446 is phosphorylated in response to hyperosmolarity. HeLa cells were transiently transfected with expression constructs encoding WT cortactin and Y446F or an empty vector and subsequently stimulated with 300 mM sucrose (Suc) for the indicated times. Lysates were analyzed by immunoblotting (IB) with the indicated antibodies B, hyperosmolarity-induced cortactin phosphorylation is reduced by WT but not inactive PTP1B. HeLa cells were transfected with expression constructs encoding WT PTP1B and C215S (CS) or an empty vector and stimulated as described for B. C, PTP1B inhibition increases hyperosmolarity-induced cortactin phosphorylation. HeLa cells were treated with a small molecule PTP1B inhibitor prior to stimulation with sucrose as described for A. D, quantitation of four independent experiments described in C. Error bars represent S.E. Asterisks denote data points that are significantly greater than the control points (p Ͻ 0.05) calculated using an unpaired, two-tailed, Student's t test.
methods, we have confirmed that Tyr 446 is a target of EGF receptor signaling and shown that it is also phosphorylated in response to hyperosmotic stress.
There is evidence that cortactin is phosphorylated in a progressive manner, with phospho-Tyr 421 potentially acting as a docking site for Src, allowing it to phosphorylate Tyr 470 (35). However, it seems unlikely that Tyr 446 represents an additional secondary site. First, mass spectrometry studies have identified, in several cases, Tyr 446 as the sole tyrosine-phosphorylated cortactin residue (20,21,24). Second, our results show that in response to hyperosmolarity, the phosphorylation state of Tyr 421 is dependent on the presence of Tyr 446 . Correspondingly, although the trapping mutant of PTP1B binds to Tyr 446 and not Tyr 421 , inhibition of PTP1B or overexpression of the WT enzyme affects both residues. Finally, we have shown that Tyr 446 is required for protection of cells from hyperosmolarityinduced apoptosis, pointing to the potential importance of this single residue in the regulation of actin remodeling. In future studies, we plan to investigate further the interrelationship between Tyr 446 and the canonical Src sites as well as whether particular effectors of cortactin function depend on Tyr 446 phosphorylation.
An exciting possibility is that the regulation of cortactin could contribute to the effect of PTP1B on breast tumorigenesis in mice (5,6). Amplification of a genomic region containing EMS1, encoding human cortactin, is associated with human cancer, and cortactin has been implicated in the promotion of tumor cell migration and invasion (12). It will be interesting to determine whether expression of cortactin Y446F affects the tumorigenicity of cancer cells. Nonetheless, deciphering the specific importance of PTP1B-mediated cortactin regulation in this process will be a challenge, given the pleiotropic effects of this enzyme on cancer cell signaling. Potentially, the phosphorylation state of Tyr 446 could also be used as an in vivo reporter to monitor the effectiveness of PTP1B-targeted therapies.
In conclusion, we have shown that PTP1B regulates cortactin tyrosine phosphorylation, likely by directly dephosphorylating Tyr 446 . This is the first report implicating a specific enzyme in the modification of this site, and it also establishes its functional significance in the protection of cells during hyperosmotic stress. On the basis of these results and published phosphoproteomic data, we propose that the current model of cortactin regulation by tyrosine phosphorylation, which holds that Tyr 421 , Tyr 470 , and Tyr 486 are of primary importance, should be revised to include Tyr 446 . Furthermore, this study suggests a novel mechanism by which PTP1B may affect a variety of cellular processes, including tumorigenesis, through regulation of the actin cytoskeleton.