Src Family Protein-tyrosine Kinases Alter the Function of PTEN to Regulate Phosphatidylinositol 3-Kinase/AKT Cascades*

Src family protein-tyrosine kinases, which play an important role in signal integration, have been implicated in tumorigenesis in multiple lineages, including breast cancer. We demonstrate, herein, that Src kinases regulate the phosphatidylinositol 3-kinase (PI3K) signaling cascade via altering the function of the PTEN tumor suppressor. Overexpression of activated Src protein-tyrosine kinases in PTEN-deficient breast cancer cells does not alter AKT phosphorylation, an indicator of signal transduction through the PI3K pathway. However, in the presence of functional PTEN, Src reverses the activity of PTEN, resulting in an increase in AKT phosphorylation. Activated Src reduces the ability of PTEN to dephosphorylate phosphatidylinositols in micelles and promotes AKT translocation to cellular plasma membranes but does not alter PTEN activity toward water-soluble phosphatidylinositols. Thus, Src may alter the capacity of the PTEN C2 domain to bind cellular membranes rather than directly interfering with PTEN enzymatic activity. Tyrosine phosphorylation of PTEN is increased in breast cancer cells treated with pervanadate, suggesting that PTEN contains sites for tyrosine phosphorylation. Src kinase inhibitors markedly decreased pervanadate-mediated tyrosine phosphorylation of PTEN. Further, expression of activated Src results in marked tyrosine phosphorylation of PTEN. SHP-1, a SH2 domain-containing protein-tyrosine phosphatase, selectively binds and dephosphorylates PTEN in Src transfected cells. Both Src inhibitors and SHP-1 overexpression reverse Src-induced loss of PTEN function. Coexpression of PTEN with activated Src reduces the stability of PTEN. Taken together, the data indicate that activated Src inhibits PTEN function leading to alterations in signaling through the PI3K/AKT pathway.

Nonreceptor Src family protein-tyrosine kinases transduce signals controlling a variety of cellular processes including proliferation, differentiation, motility, and adhesion (1)(2)(3). Src kinases are activated following engagement of many different classes of cellular receptors and participate as a convergence point in different signaling pathways (1,3,4). In this regard, Src is a critical component of the signaling cascades initiated by tyrosine kinase-linked receptors, such as the epidermal growth factor receptor, G protein-coupled receptors including the lysophosphatidic acid receptors and steroid receptors including the estrogen receptor (1,(5)(6)(7)(8). Src has been found to be overexpressed or highly activated in a number of human neoplasmas, including ovary, breast, lung, colon, esophagus, skin, parotid, cervix, and gastric carcinomas, as well as neuroblastomas and myeloproliferative disorders (9 -14). An activated mutation of Src may be present in a number of neoplasms, contributing to their development or progression (15,16). Transgenic mice expressing the polyomavirus middle T oncogene in the mammary epithelium develop multifocal mammary tumors that metastasize with high frequency because of an ability to associate with and activate Src family kinases, phosphatidylinositol 3-kinase (PI3K) 1 and the Shc adapter protein (17)(18)(19)(20). Mice expressing polyomavirus middle T antigen in the absence of functional c-Src rarely develop mammary tumors, indicating that c-Src tyrosine kinase activity is required for polyomavirus middle T antigen-induced mammary tumorigenesis (21). Indeed, the c-Src protein level and activity are increased 30 -40fold in human breast cancers compared with normal breast tissue (14). The activation of c-Src activity in mammary tumors has been linked to its association and interaction with the HER2/Neu epidermal growth factor receptor family member (22,23). Activation of Src kinase is an indicator of poor clinical prognosis in colorectal carcinomas, implicating Src in tumor progression (12). Together, the Src protein-tyrosine kinase appears to be an important mediator of tumor initiation and progression.
Src family protein-tyrosine kinases consist of nine members including Src, Fyn, and Yes, which are expressed in most tissues, and Blk, Yrk, Fgr, Hck, Lck, and Lyn, which are selectively expressed in particular cell lineages (9). Each member of the Src family members shares a series of common structure features (1,2,9). An amino-terminal myristylation site is required for association with cytoplasmic membranes and is essential for oncogenic transformation. Autophosphorylation or transphosphorylation by activating kinases of Src at tyrosine 416 is required for optimal activity. Displacement of the interaction between phosphorylated carboxyl-proximal tyrosine-527 and the SH2 domain of Src or the binding of SH3 domain to the SH2-kinase linker region converts Src from a relatively inactive form to an active form able to phosphorylate its downstream effector molecules.
Src regulates cell cycle progression and mediates cell survival through the activation of the PI3K/AKT pathway (2)(3)(4)9). The PI3K signaling cascade is implicated in a broad range of cancer-related cellular processes (24). PI3K catalyzes phosphorylation of phosphatidylinositols (PtdIns) at the D3 position of the inositol ring. The resulting PI3K products, PtdIns 3,4,5trisphosphate and PtdIns 3,4-bisphosphate, bind to pleckstrin homology (PH) and other domains of intracellular signaling molecules recruiting them to the cell membrane. Phosphorylation and activation of the PH domain containing c-Akt contributes to cell proliferation, survival, and other cellular processes (24, 25). Growth factor responsive Class IA PI3Ks consist of heterodimers of a p110-kDa catalytic subunit associated with a p85-kDa noncatalytic regulatory subunit. We have previously shown that phosphorylation of Tyr 688 on p85 by Src relieves the inhibitory effect of p85 on p110 catalytic activity (26). We demonstrate herein that Src kinases also regulate PI3K signaling cascades via altering the function of PTEN, a negative regulator of the PI3K pathway.
PTEN, a tumor suppressor, dephosphorylates membrane Pt-dIns and cytosolic polyphosphoinositols as well as highly acidic tyrosine, serine, and threonine phosphorylated peptides (27)(28)(29). The protein phosphatase activity of PTEN has been proposed to decrease cyclin D1 expression, thereby contributing to cell cycle arrest and also to dephosphorylate focal adhesion kinase, consequently reducing cell migration and focal adhesion formation (30,31). However, the tumor suppressor role of PTEN is thought to derive primarily from the capacity of PTEN to counteract the activity of PI3K by dephosphorylating the D3 position of membrane PtdIns 3,4,5-trisphosphate and PtdIns 3,4-bisphosphate (32,33).The lipid phosphatase activity of PTEN is associated with down-regulated activities of multiple downstream components of the PI3K pathway, including most notably AKT and its downstream targets, BAD, p70S6K, GSK3, p21, p27, MDM2, TSC1/2, and forkhead transcription factors and altered expression of p27 (31, 34 -36). Because PI3K/AKT activation has been implicated in cell proliferation, differentiation, survival, metastasis, angiogenesis, and cytoskeletal rearrangement, the inhibition of this pathway is consistent with a major role for PTEN in tumor suppression. In addition, the association of its lipid phosphatase activity with PTEN tumor suppressor activity has been corroborated by data revealing the ability of PTEN to regulate cell growth to be severely diminished in a PTEN protein carrying a mutation (G129E), which abrogates lipid but not protein phosphatase activity (33).
Although loss-of-function PTEN gene mutations have been detected in many types of tumors, reduced PTEN expression in at least some transformed cells appears to arise independently of gene mutation and in some instances has been linked to methylation of the promoter (37)(38)(39). Little is known about the regulation of PTEN function. PTEN is phosphorylated on serine and threonine residues in its tail by CK2 (40,41). This phosphorylation regulates its stability, activity, and potentially its association with other proteins through its PDZ-binding motif (42)(43)(44). PTEN contains 23 tyrosine residues, several of which are in consensus phosphorylation sites and protein-protein interaction motifs, such as SH2-binding sites. Thus, tyro-sine phosphorylation represents one potential mechanism whereby activity and/or expression of this phosphatase might be modified post-translationally. This possibility is supported by previous data from our group indicating that mutation of Tyr 240 or Tyr 315 abrogated the ability of PTEN to alter cell growth rate, saturation density, and colony formation in vitro. Mutation of Tyr 240 or Tyr 315 markedly decreased the ability of PTEN to limit tumor growth in nude mice (45). These data raise the possibility that tyrosine phosphorylation of PTEN could regulate the PI3K/AKT signaling pathway. However, either basal or induced tyrosine phosphorylation of PTEN has not been detected in previous studies (40 -42). To evaluate the potential interactions between Src tyrosine kinases and PTEN in breast cancer, we have now investigated the effects of Src tyrosine kinases on PTEN function and on PI3K-mediated AKT activation. The results of our studies demonstrate that Src protein-tyrosine kinases regulate the PI3K/AKT pathway by altering PTEN function, which is associated with increased tyrosine phosphorylation of PTEN, leading to a decreased ability of PTEN to hydrolyze PtdIns and reduced PTEN stability.

Expression Vectors
FLAG epitope-tagged full-length wild type PTEN and catalytic inactive PTEN (C124S) were from Drs. Dixon and Maehama (Ann Arbor, MI). Lipid-phosphatase inactive mutant of PTEN G129E was generated by site-directed mutagenesis (Stratagene, La Jolla, CA). The construct was sequenced in its entirety to confirm the mutagenesis. GFP fusion protein of full-length wild type PTEN and CK2 phosphorylation defective mutant PTEN (A4, alanine substitutes at Ser 380 , Thr 382/383 , and Ser 385 ) were from Dr. William R. Sellers (Boston, MA). HA epitopetagged wild type AKT was from Dr. Downward. The AKT-PH domain fused to GFP was provided by Dr. Tobias Myer (Stanford University) and Dr. S. Grinstein (Toronto, Canada). The cDNA plasmid for activated Lck Y505F was a generous gift of Dr. A. Veillette (Montreal, Canada). The cDNA plasmids for activated Src Y527F, wild type SHP-1, and inactive SHP-1 (C453S) were described previously (5). Adenovirus containing the transgene of wild type PTEN was described previously (46).

Cell Lines and Transient Transfection
Tumor cell lines, MDA-MB-468, BT-549, MDA-MB-231, SKBr3, MCF-7, and Cos-7 cells were cultured in complete medium (RPMI 1640 supplemented with 10% fetal bovine serum). The cells were transiently transfected with different cDNA plasmids by FuGENE 6 TM transfection reagent (Roche Applied Science) or by Nucleofector Reagent (Amaxa Biosystems, Koeln, Germany) as suggested by the manufacturer. In some experiments, the cells were infected with adenovirus containing wild type PTEN as previously described (46). After transfection or infection, the cells were cultured for 30 h in complete medium and then serum-starved for 18 h prior to cell stimulation or cell lysis.

PTEN Lipid Phosphatase Assay
Micelle PtdIns Assay-FLAG epitope-tagged wild type PTEN was transiently transfected with or without activated Lck (Y505F) into BT-549 cells (which lack endogenous PTEN). 48 h after transfection, the cells were lysed in lysis buffer without NaF and Na 3 VO 4 . PTEN was immunoprecipitated by anti-FLAG M2 monoclonal antibody as described above. The immunoprecipitates were washed and incubated with 32 P-labeled phosphatidylinositol 3,4-bisphosphate in 100 mM Tris-HCl, pH 8.0, 10 mM dithiothreitol at 37°C for 1 h. 32 P-Labeled phosphatidylinositol 3,4-bisphosphate was generated by phosphorylation of phosphatidylinositol 4-monophosphate in a PI3K reaction. The lipid phosphatase reaction was terminated by the addition of 0.1 N HCl, and the lipids were extracted by methanol:chloroform (1:1). The lipids were resolved by TLC and visualized by autoradiography.
Soluble PtdIns Assay-FLAG epitope-tagged wild type PTEN was transiently transfected with or without activated Lck (Y505F) or activated Src (Y527F) into Cos-7 or BT-549 cells. 48 h after transfection, the cells were lysed in 50 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA, 10% glycerol, 1% Triton X-100 with protease inhibitors (Roche Applied Science). PTEN was immunoprecipitated by anti-FLAG M2 monoclonal antibody or rabbit anti-PTEN antibody as described above. The immunoprecipitates were washed and incubated with water-soluble diC 8

Pulse-Chase Assay
Wild type PTEN was transiently transfected with or without activated Src kinase as described above. 24 h after transfection, the cells were washed with phosphate-buffered saline and incubated with methionine-free medium supplemented with 10% dialyzed fetal bovine serum for 15 min. The cells were then pulse-labeled with [ 35 S]methionine containing medium (50 Ci/ml) for 1 h. The cells were washed and incubated in chasing medium (regular RPMI 1640 with 25 mM Hepes but without fetal bovine serum) for the time period as indicated. The cells were collected and lysed. PTEN was immunoprecipitated by goat anti-PTEN antibody as described above. The proteins were resolved by SDS-PAGE and transferred to PVDF membranes. 35 S-Labeled PTEN was visualized by autoradiography. The density of each band was analyzed by NIH Image software. The data are presented as percentages of control using the 0-h point as the basal level.

Src Protein-tyrosine Kinases Regulate AKT Phosphorylation by Compromising the Function of PTEN-We have previously
shown that breast cancer cells lacking functional PTEN display high basal levels of AKT phosphorylation, an indicator of signal transduction through the PI3K pathway (34). Enforced expression of PTEN in PTEN null cells markedly reduces the basal phosphorylation of AKT. In contrast, enforced expression of PTEN decreases the duration but not the magnitude of ligandinduced PI3K/AKT signaling, suggesting that PTEN is constitutively active and that its activity is constrained following ligand perturbation (34). Further, mutation of PTEN tyrosine residues 240 or 315, located in consensus phosphorylation sites and protein-protein binding motifs, significantly alters the ability of PTEN to reduce tumor growth in vitro and in vivo, suggesting that these tyrosine residues are important for PTEN to fulfill its tumor suppressor function (45). To determine whether Src protein-tyrosine kinases alter the capacity of PTEN to modulate PI3K signaling, Src protein-tyrosine kinases were introduced together with HA-AKT into BT-549 cells with and without expression of functional PTEN (Fig. 1). An active form of a Src family protein-tyrosine kinase, Lck(Y505F), in which the negative regulatory carboxyl tyrosine was replaced with phenylalanine, was used in this assay because the kinase activity of this mutant enzyme is not affected by protein-tyrosine phosphatases (47). As expected, a lack of functional PTEN in the BT-549 breast cancer cell line is associated with high basal levels of AKT phosphorylation (lane 2), as detected by anti-phospho-AKT (Ser 473 ) immunoblotting. Coexpression of wild type PTEN with HA-AKT in these PTENdeficient cells engendered a significant down-regulation in the phosphorylation of AKT (lane 3). As a control, AKT phospho- Active Src Kinase Reduces the Ability of PTEN to Dephosphorylate PtdIns-The ability of PTEN to efficiently dephosphorylate PtdIns has been linked to both its catalytic activity and its positioning via the C2 domain in proximity to PtdIns substrates at the membrane (48). To determine whether PTEN-mediated dephosphorylation of membrane PtdIns is also influenced by Src protein-tyrosine kinases, PTEN was introduced into BT-549 cells in the presence and absence of activated Lck, and the capacity of immunoprecipitated PTEN to dephosphorylate PI3K-phosphorylated lipid micelles of 32 P-labeled PtdIns 3,4-bisphosphate was then evaluated. The results of this analysis revealed PtdIns 3,4-bisphosphate to be efficiently dephosphorylated by wild type PTEN but to be unaffected by lipidphosphatase incompetent G129E PTEN (Fig. 2, A and B). The lipid phosphatase activity of PTEN immunoprecipitates was found to be reduced by about 40% in the context of PTEN coexpression with activated Lck (Fig. 2, A and B), a result suggesting that Src kinases suppress the capacity of PTEN to dephosphorylate PtdIns (p ϭ 0.033 by Student's t test, PTEN (WT) versus PTEN (WT) ϩ Lck). Because phosphatase inhibitors could not be used in this assay, PTEN may have undergone some degree of autodephosphorylation, and accordingly, the effects of Src family kinases on PTEN activity may have been underestimated. However, in contrast to the effects of activated Lck on PTEN-directed dephosphorylation of micellar PtdIns, activated Lck or activated Src had no effect on PTEN-mediated dephosphorylation of water-soluble diC 8  These observations suggest that the inhibitory effects of Src family kinases on PTEN lipid phosphatase activity reflect interference with the ability of PTEN to associate productively with membrane lipids rather than a direct inhibition of its enzymatic activity.
Upon growth factor stimulation, AKT is recruited to the activated compartments of cellular plasma membrane through the binding of its PH domain to PI3K products, PtdIns 3,4,5trisphosphates and PtdIns 3,4-bisphosphates. Effect of activated Src kinases on the ability of PTEN to dephosphorylate PtdIns was further investigated in intact BT-549 cells expressing GFP fusion protein of AKT-PH domain. Cellular membrane localization of AKT-PH-GFP fusion protein in the presence or absence of activated Src and/or PTEN would represent the amount of PI3K products PtdIns in cytoplasma membrane. As shown in Fig. 3, the GFP-AKT-PH fusion protein generally locates in cytoplasm under resting conditions (Fig. 3A). Although activated Src does not significantly alter the localization of the fusion protein in the absence of functional PTEN (Fig. 3B), wild type PTEN itself effectively constrains AKT-PH in cytosol (Fig. 3C). In contrast, coexpression of activated Src with wild type PTEN markedly promotes the translocation of the AKT-PH fusion protein to the leading edge of cytoplasm membrane (data not shown), indicating the loss of PTEN function to dephosphorylate PtdIns. Introducing wild type SHP-1, a protein-tyrosine phosphatase, into the cells expressing both activated Src and wild type PTEN reconstitutes the localization of AKT-PH in cytosol (Fig. 3E). These data, consistent with in vitro PTEN phosphatase assays (Fig. 2, A and B), demonstrate that activated Src alters the ability of PTEN to dephosphorylate PtdIns.

Src Kinases Induce PTEN Tyrosine Phosphorylation and SHP-1 Phosphatase Down-regulates the Tyrosine Phosphorylation-
The PTEN carboxyl-terminal C2 domain, a region required for PTEN binding to membrane PtdIns, contains three tyrosine residues located in phosphorylation and protein-protein binding consensus motifs. Tyr 240 and Tyr 336 map within consensus SH2 domain-binding motifs for SHP-1. Tyr 315 locates within a consensus Shc-binding site. Tyr 336 is also located within a consensus phosphorylation site for the Abelson (Abl), epidermal growth factor receptor, and platelet-derived growth factor receptor kinases. Accordingly, the capacity of PTEN to undergo inducible tyrosine phosphorylation and its correlation with the ability to reduce AKT phosphorylation were examined. Compatible with previous reports (40 -42), PTEN is not tyrosine-phosphorylated in resting cells, and ligand stimulation did not cause significant tyrosine phosphorylation of PTEN in intact cells (data not shown). We assessed the possibility that PTEN could be tyrosine-phosphorylated in vivo but susceptible to rapid dephosphorylation. Pervanadate inhibits tyrosine phosphatases in intact cells, resulting in an increase in the level of tyrosine phosphorylation of many signaling proteins triggering pathways normally regulated by growth factors, such as those stimulated by insulin (49,50). As indicated by the anti-phosphotyrosine immunoblotting shown in Figs. 4 and 5, treatment of breast cancer cells with pervanadate causes a marked increase in the tyrosine phosphorylation of endogenous (Fig. 4) or ectopically expressed PTEN (Fig. 5). Incubation of the cells with Src kinase inhibitors, PP1, or herbimycin prior to pervanadate treatment significantly reduces the tyrosine phosphorylation of PTEN (Fig. 5A). In contrast, the Abl inhibitor STI571 does not affect pervanadate-induced phosphorylation of PTEN (Fig. 5A), demonstrating that Src protein-tyrosine kinases are involved in pervanadate-induced tyrosine phosphorylation of PTEN. Consistent with this effect, pervanadate, similar to activated Src kinases, abrogates the ability of PTEN to reduce AKT phosphorylation (Fig. 5B). Inhibition of PTEN tyrosine phosphorylation by PP1 or herbimycin reconstitutes the function of PTEN as assessed by Akt phosphorylation (Fig. 5B).
In light of our previous data showing that the association of the SHP-1 tyrosine phosphatase with the PI3K p85 subunit depends upon Src family PTK-mediated phosphorylation of p85 (47), the possibility that Src family kinases induce tyrosine phosphorylation of PTEN and its association with SHP-1 was further investigated. The ability of activated Src family kinases to induce association of SHP-1 with PTEN was assessed by coexpression of activated forms of Src (Y527F) or Lck (Y505F), catalytically inert PTEN (C124S), and either wild type or catalytically inactive SHP-1 (C453S). An enzymatically inactive PTEN was used here to prevent autodephosphorylation and/or dephosphorylation of PTEN-binding proteins. As shown in Fig.  6A, coexpression of either activated Src or Lck with PTEN induces a marked increase in the tyrosine phosphorylation of PTEN (lane 3 and 7). In the presence but not the absence of activated Src, PTEN associated with the catalytically inert form of SHP-1 (lane 5). By contrast, expression of wild type rather than inactive SHP-1 abrogated both the Src-induced tyrosine phosphorylation of PTEN and the association of PTEN with SHP-1 (lane 4). Thus, Src family kinases induce association of PTEN with SHP-1, likely through association of its SH2 domain with phosphorylated PTEN, which subsequently dephosphorylates PTEN. The capacity of SHP-1, and potentially other tyrosine phosphatases, to efficiently dephosphorylate PTEN may contribute to the difficulty in detecting PTEN phosphorylation in cells replete in SHP-1 and may contribute to the ability of pervanadate to increase PTEN phosphorylation in a Src-dependent manner.
To determine whether Src-induced PTEN tyrosine phosphorylation and the ability of SHP-1 to reverse the effect of Src could translate into alterations in signaling through PI3K cascade, HA-AKT was coexpressed with PTEN in the presence or absence of activated Src and/or wild type SHP1 and AKT phosphorylation evaluated. As shown in Fig. 6B, Src induces tyrosine phosphorylation of PTEN and alters the ability of PTEN to reduce AKT phosphorylation (lane 3). SHP-1 decreases Srcinduced phosphorylation of PTEN reconstituting its ability to decrease AKT phosphorylation (lane 4). In contrast, SHP-1 itself does not influence AKT phosphorylation (lane 6). These studies further support the concept that Src kinases alter the FIG. 2. Active Src kinase reduces the ability of PTEN to dephosphorylate membrane PtdIns. A, Src protein-tyrosine kinases reduce the ability of PTEN to dephosphorylate PtdIns in micelles. Wild type PTEN or PTEN (G129E) was transfected into BT-549 cells, with or without activated Lck (Y505F) (as described in the legend to Fig. 1). 30 h after transfection, the cells were serum-starved overnight prior to cell lysis. FLAG epitope-tagged PTEN was immunoprecipitated by anti-FLAG M2 monoclonal antibody. The immunoprecipitates were subjected to in vitro lipid phosphatase activity assay using 32 P-labeled phosphatidylinositol 3,4-bisphosphate in micelles as substrate. After the reaction was terminated, the lipids were extracted, resolved by TLC, and visualized by autoradiograph. B, the results from A were quantified by phosphorimaging. PTEN phosphatase activity was calculated (from arbitrary units) and presented as percentages of wild type ϭ (Mock Ϫ wild type PTEN expressed with Lck or PTEN G129E)/(Mock Ϫ wild type PTEN expressed by itself) ϫ 100%. The data were shown as the means Ϯ S.D. from two similar experiments. Statistical significance was determined by Student's t test (p ϭ 0.033, PTEN (WT) versus PTEN (WT) ϩ Lck). C and D, Src protein-tyrosine kinases do not affect the ability of PTEN to dephosphorylate water-soluble diC 8 PtdIns. Wild type PTEN or PTEN (C124S) was transfected into Cos-7 (C) or BT-549 cells (D) with or without activated Lck (Y505F) (C) or activated Src (Y527F) (D). The cells were serum-starved overnight prior to cell lysis. FLAG epitope-tagged PTEN was immunoprecipitated by anti-FLAG M2 monoclonal antibody. The immunoprecipitates were subjected to in vitro lipid phosphatase activity assay using water-soluble diC 8 PtdIns 3,4,5-trisphosphate as substrate. PTEN activity was measured by colorimetric detection of phosphate release and quantitated by A 650 nm and calculated for relative activity as described under "Experimental Procedures." The data are presented as the means Ϯ S.D. from three similar experiments. Anti-PTEN immunoblot confirmed that equal amount of PTEN was immunoprecipitated from each sample (data not shown). Statistical significance was determined by Student's t test. PI3K signaling pathway by inducing the phosphorylation of specific substrates including PTEN, a process that is reversed by tyrosine phosphatases such as SHP-1.
Cotransfection of Src PTKs with PTEN resulted in PTEN tyrosine phosphorylation in intact cells. Furthermore, pervanadate increases PTEN tyrosine phosphorylation in a Src-dependent manner. However, recombinant Src family kinases were unable to phosphorylate purified PTEN in in vitro kinase assays (data not shown). This observation suggests the involvement of an intermediary kinase coupling the Src PTK to PTEN tyrosine phosphorylation, a hypothesis supported by a failure in the detection of PTEN tyrosine phosphorylation in in vitro kinase assays incorporating either immunoprecipitated PTEN or immunoprecipitated Src PTK (data not shown).
Src Kinases Decrease PTEN Protein Stability-Phosphorylation has been implicated in the regulation of protein stability (51). We thus assessed whether Src kinase-mediated phospho-rylation of PTEN alters PTEN stability. FLAG epitope-tagged PTEN was coexpressed in Cos-7 cells with or without activated Lck (Y505F) and pulse-labeled with [ 35 S]methionine. FLAG-PTEN was recovered by immunoprecipitation and detected by autoradiography after different "chasing" time periods. As shown in Fig. 7 CK2 (40, 41). Phosphorylation of these sites in PTEN by CK2 regulates both PTEN function and stability (40 -42). Thus, the effects of Src family kinases on PTEN could be through changes in CK2mediated phosphorylation of the PTEN tail. As indicated in To further assess whether alterations in CK2-mediated phosphorylation of PTEN could contribute to the ability of Src to alter signal transduction through the PI3K pathway, wild type PTEN and A4 PTEN were introduced into BT-549 cells, and AKT phosphorylation was assessed in the presence or absence of activated Src. The levels of A4-PTEN in intact cells were decreased compared with wild type PTEN compatible with the concept that CK2 phosphorylates and stabilizes PTEN (data not shown). When A4-PTEN levels were adjusted by introduction of higher amounts of the expression vector, A4-PTEN reduces AKT phosphorylation in BT-549 cells in a similar manner to wild type PTEN (Fig. 8). Furthermore, the mutation of the CK2 phosphorylation sites did not limit the ability of activated Src to alter PTEN function as indicated by AKT phosphorylation (Fig. 9). Thus, Src regulates PTEN function independently of CK2-induced phosphorylation of PTEN. Together these observations demonstrate that Src kinases regulate signaling through the PI3K pathway by altering PTEN function, which includes increased tyrosine phosphorylation of PTEN associated with its decreased ability to hydrolyze PtdIns and reduced PTEN stability. DISCUSSION In the current study, we have investigated the possibility that Src protein-tyrosine kinases regulate the PI3K/AKT prosurvival pathway via altering PTEN function and stability. Src participates in the regulation of many signal transduction pathways. Src regulates the PI3K/AKT cascade through multiple mechanisms. Tyrosine phosphorylation of cellular proteins induced by Src family kinases recruits PI3K to activated membrane compartments wherein PI3K phosphorylates membrane PtdIns and recruits its downstream effector molecules, thus fulfilling its biological activities (24, 25). Src also activates PI3K via the binding of the Src-SH3 domain to the proline-rich region in the PI3K p85 subunit (52). The regulatory p85 subunit of PI3K stabilizes the catalytic p110 subunit and inhibits the PI3K enzymatic activity (53). Tyrosine phosphorylation of p85, more specifically of Tyr 688 , by Src family kinases induces the formation of an intramolecular complex in p85, thereby relieving the inhibitory effect of p85 on p110 leading to the activation of the PI3K/AKT cascade (26). The ability of Src to inactivate PTEN function provides an additional interaction site between Src and the PI3K pathway.
PTEN has been demonstrated to undergo phosphorylation at multiple serine and threonine sites in its carboxyl-terminal tail (40 -43). Phosphorylation at these sites, which is at least in part mediated by the CK2 kinase, has been shown to alter both stability and catalytic activity of the PTEN protein (40,42,43). The identity of the CK2-mediated phosphorylated residues in the PTEN tail with mass spectrometry results in contradictory conclusions with mutagenesis studies (40,41). It appears likely that the dominant phosphorylation sites in PTEN are Ser 370 and Ser 385 . However, it is possible that additional Ser and Thr residues contribute to PTEN phosphorylation, particularly when the conformational structure of PTEN is altered by mutagenesis (41). Phosphorylation of serine/threonine residues within the PTEN tail acts as an inhibitory switch by preventing its recruitment into a protein complex (43,44). Consequently, it interferes with PTEN binding to MAGI family proteins and may thereby impede PTEN association with cell membranes, particularly at points of focal contact (42,54).
Src and CK2 both decrease the ability of PTEN to dephosphorylate PtdIns. In contrast Src decreases PTEN stability and CK2 increases PTEN stability (40,42). Thus, it is unlikely that the effects of Src on PTEN are mediated by alterations in CK2-mediated PTEN phosphorylation. Furthermore, Src does not alter CK2 kinase-induced serine/threonine phosphorylation of PTEN. Mutation of putative serine and threonine CK2 phosphorylation sites in the tail of PTEN does not alter the ability of Src to inhibit PTEN function as indicated by changes in AKT phosphorylation. Although the identity of the CK2 phosphorylation sites of PTEN is controversial, the PTEN mutants utilized in mutagenesis studies may not present as a physiological substrate for CK2 either as a consequence of structural changes in the PTEN tail or serving as a pseudosubstrate with highly negative charged residues (40,41).
Tyrosine phosphorylation of PTEN has not been detected in previous studies (40 -42). Indeed, other than cotransfection of activated Src family kinases or inhibition of tyrosine phosphatases, we are unable to reveal consistent high levels of PTEN tyrosine phosphorylation. However, it remains possible that specific phosphatase in particular cellular lineages may limit the detection of PTEN tyrosine phosphorylation. The ability of SHP-1 to bind and dephosphorylate PTEN suggests that phosphatase-mediated dephosphorylation of PTEN may mask PTEN tyrosine phosphorylation under certain physiological conditions and especially in cells stimulated with ligands such as epidermal growth factor, platelet-derived growth factor, or insulin-like growth factor (40 -42). Thus, the difficulty in detecting tyrosine-phosphorylated PTEN following growth factor stimulation may be due to rapid attenuation by SHP-1 or other tyrosine phosphatase-mediated dephosphorylation of PTEN and/or by tyrosine phosphorylation-induced degradation of PTEN. Nevertheless, because Src family kinases are important mediators in tumorigenesis in breast and other tissues, the ability of Src family kinases to decrease PTEN activity and stability may contribute to the effects of Src on tumorigenesis.
Despite the presence of functional PTEN in most cells, extracellular ligands are normally able to induce transient increases in intracellular D3 phosphorylated PtdIns and signaling through the PI3K cascade. Similarly, adenoviral-induced high level expression of PTEN does not abrogate PI3K signaling and the consequent induction of AKT, BAD, and p70S6K phosphorylation in PTEN replete cells (34,46). These observations indicate that PTEN is either catalytically inefficient, recruited relatively slowly to the cell membrane, and/or posttranslationally regulated. The data presented here favor the last possibility, suggesting that Src-mediated phosphorylation of PTEN may contribute to the inactivation of this phosphatase during ligand perturbation and thereby provide a "permissive" role in promoting the activation of the PI3K signaling pathway.
Although Src family kinases efficiently induced PTEN tyrosine phosphorylation when these proteins were overexpressed in cell lines, purified or recombinant Src or Lck did not phosphorylate purified PTEN in in vitro assays. These data suggest the absence in the in vitro assay of required cofactors or, alternatively, an intermediary kinase whose activation depending upon Src family kinases. Because the PTEN Tyr 336 residue is located within a consensus site for phosphorylation by the Abl tyrosine kinase, the potential for this protein-tyrosine kinase to phosphorylate PTEN was examined by overexpression of PTEN and Abl in Cos-7 cells. However, no induction of PTEN tyrosine phosphorylation was observed in this assay (data not shown). Further, the Abl inhibitor STI571 does not block pervanadate-induced tyrosine phosphorylation of  PTEN, indicating that Abl is not required for pervanadate-and Src-induced increases in tyrosine phosphorylation of PTEN. The identity of a protein subserving an intermediary role in tyrosine phosphorylation of PTEN remains to be determined. SHP-1 has been shown to limit signaling through multiple cell surface receptors and thereby to inhibit ligand-induced cell growth (55). Further, we have demonstrated that SHP-1 regulates Src family protein-tyrosine kinases (56) and dephosphorylates multiple different Src phosphorylated targets (47). Signaling through the PI3K/AKT pathway, which plays an important role in tumorigenesis in multiple lineages, is increased in homozygous SHP-1 negative me mice (26), potentially contributing to the development of tumors in these mice. We have also demonstrated that SHP-1 modulates signaling through the PI3K pathway via its association with and dephosphorylation of the PI3K p85 regulatory subunit (47). The studies described herein reveal the additional possibility that SHP-1 could exhibit a tumor suppressor role by restoring the function of PTEN, potentially through dephosphorylating PTEN and allowing PTEN to productively associate with its substrates in the cell membranes.
Although the tyrosine phosphorylation of PTEN was associated with a reduction in PTEN-mediated dephosphorylation of membrane PtdIns in vitro and in PTEN-mediated down-regulation of AKT phosphorylation and membrane localization in intact cells, tyrosine phosphorylation did not affect PTENmediated dephosphorylation of water-soluble PtdIns. These observations suggest that phosphorylation of PTEN may alter the capacity of PTEN to productively associate with membrane PtdIns but does not result in direct inactivation of the PTEN catalytic domain. PTEN possesses a C2 domain, which has been previously demonstrated to be required for PTEN to bind membrane PtdIns, productively positioning the catalytic domain where it can fulfill its 3-phosphatidylinositol phosphatase enzymatic activity (48). Mutations of PTEN in the C2 domain found in human cancers suggest that an intact C2 domain may not only participate in membrane recruitment of PTEN but may also regulate PTEN protein stability (48,57). Tyrosine 336 is located on the surface of the C2 domain, where it would be FIG. 8. Src kinases do not interfere with protein kinase CK2-induced phosphorylation of PTEN. Wild type PTEN or PTEN with defective CK2 phosphorylation sites were cotransfected into Cos-7 cells with or without activated Src (Y527F). The cells were serum-starved overnight prior to cell lysis. PTEN was immunoprecipitated with goat anti-PTEN antibody. The immunoprecipitates were separated by 8% SDS-PAGE and immunoblotted by anti-phospho-PTEN (Ser 380 , Thr 382/383 ) antibody. Efficacy of PTEN immunoprecipitation was detected by reprobing the membrane with anti-PTEN antibody (upper panel). The expression of Src (Y527F) was confirmed by immunoblotting of the total cell lysates with anti-Src antibody (lower panel). Anti-PI3K p85 immunoblot was included to verify that equal amounts of cellular proteins were applied. Similar results were obtained in two independent experiments.
FIG. 9. Activated Src kinase alters the ability of PTEN to reduce AKT phosphorylation independent of protein kinase CK2induced phosphorylation of PTEN. HA-AKT was cotransfected into BT-549 cells with GFP fusion protein, wild type PTEN, or Ser/Thr phosphorylation-defective mutant A4-PTEN in the presence or absence of the activated Src family protein-tyrosine kinase, Src (Y527F). The cells were serum-starved overnight prior to cell lysis. HA-AKT was immunoprecipitated by anti-HA monoclonal antibody. The immunoprecipitates were subjected to SDS-PAGE and immunoblotted with antiphospho-AKT (Ser 473 ) antibody as described in the legend to Fig. 1. The membrane was stripped and reprobed with anti-AKT antibody to verify that equal amounts of HA-AKT protein were immunoprecipitated. The results from the immunoblotting were quantified by densitometry, adjusted by protein expression level, and presented as AKT phosphorylation (arbitrary units). Similar results were obtained in three independent experiments and presented here as the means Ϯ S.D. highly accessible for tyrosine phosphorylation. Alternatively, phosphorylation of PTEN may contribute in some manner of molecular switch to the decreased ability of PTEN to dephosphorylate membrane PtdIns and thereby regulate PI3K signaling. Although these possibilities require further investigation, the current data strongly suggest that PTEN lipid phosphatase activity is regulated by Src-induced tyrosine phosphorylation in intact cells.
Together, these data reveal that not only mutational inactivation observed in multiple lineages of tumor but also phosphorylation and potentially other post-translational modifications regulate PTEN catalytic activity, protein-protein interaction, localization, and substrate contact. Further elucidation of the mechanisms regulating PTEN may provide insight the role of PTEN and Src-family kinases on PI3K/AKT signaling pathways and thus modulation of cell growth and tumorigenesis.