A Cytosolic Protein-tyrosine Phosphatase PTP1B Specifically Dephosphorylates and Deactivates Prolactin-activated STAT5a and STAT5b*

Prolactin (PRL) plays a central and crucial role in the regulation of milk protein gene expression in mammary epithelial cells. PRL binding to its cognate receptor leads to receptor dimerization and activation of the tyrosine kinase Janus kinase 2 (JAK2), associated with the membrane-proximal, intracellular domain of the receptor. In turn, JAK2 phosphorylates and activates STAT5, a member of the signal transducers and activators of transcription (STAT) family. We have recently reported that 16 different protein-tyrosine phosphatases (PTP) were expressed in lactating mouse mammary gland and mammary epithelial cells (Aoki, N., Kawamura, M., Yamaguchi-Aoki, Y., Ohira, S., and Matsuda, T. (1999) J. Biochem.(Tokyo) 125, 669–675). We investigated the involvement of each PTP in PRL signaling. Among the 12 phosphatases including SHP-2 examined, a cytosolic phosphatase PTP1B was found to specifically dephosphorylate STAT5a and STAT5b in transfected COS7 and in vitro. Nuclear translocation of STAT5a and STAT5b was largely inhibited upon overexpression of PTP1B. The PRL-dependent transcriptional activation of the β-casein gene promoter was also inhibited by PTP1B. Furthermore, retrovirus-mediated overexpression of PTP1B resulted in dephosphorylation of endogenous STAT5 and down-regulation of β-casein gene expression in mammary epithelial COMMA-1D cells when the cells were treated with lactogenic hormones. Endogenous tyrosine-phosphorylated STAT5 proteins in mammary epithelial COMMA-1D cells as well as tyrosine-phosphorylated STAT5a and STAT5b expressed in COS7 cells were co-precipitated by substrate-trapping mutants of recombinant PTP1B. These results strongly suggest that PTP1B dephosphorylates PRL-activated STAT5a and STAT5b, thereby negatively regulating PRL-mediated signaling pathway.

The polypeptide hormone prolactin is produced in the anterior pituitary, regulates the activity of milk protein gene promoters in mammary epithelial cells, and plays an important role in the growth and differentiation of lymphocytes (1). It exhibits its activity via its cognate receptor and the activation of intracellular signaling molecules such as the Janus kinase (JAK) 1 signal transducers and activators of transcription (STAT) pathway. The prolactin (PRL) receptor, belonging to the hematopoietin receptor superfamily (2), does not possess intrinsic tyrosine kinase activity but is associated with the cytoplasmic tyrosine kinase JAK2 (3)(4)(5). Ligand binding leads to dimerization of the receptor and activation of JAK2 (5). JAK2 phosphorylates not only the prolactin receptor but also the transcription factor STAT5. Upon phosphorylation, STAT5 forms homodimers, translocates to the nucleus, and specifically binds to the promoter regions of target genes, thus activating transcription (6,7).
Two closely related STAT5a and STAT5b have been identified and were shown to be encoded by similar but different genes (8 -11). STAT5a and STAT5b share 93% identity at the amino acid level with the primary differences found at their C termini (9 -11). STAT5a was originally identified as a critical mediator of PRL response in mammary epithelial cells (12,13). STAT5b was cloned from hematopoietic cells, mammary gland, and liver (8 -10, 14). It is now known that both STAT5a and STAT5b are ubiquitously expressed in most cell lines and tissues at comparable levels with a few exceptions (10).
STAT5a and STAT5b are also activated by other cytokines, including growth hormone, erythropoietin, granulocyte macrophage-colony stimulating factor (14,15), thrombopoietin (16), interleukin (IL)-2 (17,18), IL-3 (8,14), , IL-7, IL-15 (18), as well as epidermal growth factor through its respective receptor tyrosine kinase and by non-receptor tyrosine kinases Src and Bcr-Abl (19 -21). These studies indicate that STAT5a and STAT5b are involved in many different signaling pathways. Gene disruption of individual genes in mice revealed that both STAT5a and STAT5b play essential but often redundant roles in the physiological responses associated with PRL (22). Although they share high homology, it is also reported that STAT5a and STAT5b may be differentially activated (23) and bind DNA sequences with distinct specificities (24) STAT5 undergoes a rapid and transient activation and deactivation cycle through tyrosine phosphorylation upon cytokine stimulation (25). It is generally thought that phosphatases attenuate or block tyrosine phosphorylation-mediated signals and play a negative role. However, since the finding that dephosphorylation of c-Src leads to the activation (26), it is conceivable that the phosphatases could also play a positive role in some signaling cascades. Actually, one of the SH2-containing protein-tyrosine phosphatases (PTPs), SHP-2, was shown to be essential for interferon ␣/␤-induced gene transcription (27), and recent publications have shown that SHP-2 contributes to ␤-casein promoter activation in a positive manner (28,29). However, dephosphorylation of the activated JAK2 and STAT5 through the PRL receptor and the involvement of the PTPs in a negative regulation have poorly been understood.
Recently, we found that 16 different PTPs including SHP-1 and SHP-2 were expressed in mammary glands and mammary epithelial cells and that most of them were down-regulated in lactating mammary glands (30). To extend these findings, in this study we investigated the involvement of each PTP in PRL receptor-mediated signaling pathway by using expression constructs for prolactin receptor, STAT5a/b, and each PTP, and we found that both of the prolactin-induced tyrosine phosphorylations of STAT5a and STAT5b and promoter activation of ␤-casein gene were abolished in COS7 cells when cytosolic PTP1B was overexpressed. Nuclear translocation of STAT5 was also inhibited by PTP1B. Overexpression of PTP1B in mammary epithelial COMMA-1D cells also resulted in dephosphorylation of PRL-activated STAT5 and down-regulation of ␤-casein gene expression upon lactogenic hormone treatment. STAT5a and STAT5b were dephosphorylated by recombinant PTP1B in vitro and were co-precipitated by substrate-trapping mutants of PTP1B. These results strongly suggest that STAT5a and STAT5b are specific substrates of PTP1B and are deactivated by the phosphatase in PRL-mediated signaling pathway.
Retrovirus-mediated Gene Delivery-HA-tagged PTP1B was ligated into pLXSN retroviral vector (CLONTECH) via EcoRI site and introduced into Phoenix ecotropic packaging cells, which were obtained from Dr. Garry P. Nolan (Stanford University), by the modified calcium phosphate precipitation method (33). COMMA-1D cells were infected with the retrovirus-containing culture medium and then selected in the presence of G418 (1 mg/ml). To eliminate clonal deviation, G418-resistant polyclonal cells were used for experiments.
In Vitro Dephosphorylation Assay-GST fusion proteins containing full-length PTP1B were purified on glutathione-Sepharose beads and eluted with neutralized glutathione. Enzymatic activities of the GST fusion proteins were determined using para-nitrophenyl phosphate, as described previously (35). STAT5a and STAT5b immune complexes prepared from PRL-treated COS7 cells that had been co-transfected with PRL receptor and STAT5a were washed twice with dephosphorylation assay buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.4, 5 mM dithiothreitol) and incubated with the indicated amounts of GST-PTP1B fusion proteins in dephosphorylation assay buffer for 30 min at 37°C. Reactions were terminated by adding SDS sample buffer.
Nuclear Translocation-Procedures for obtaining nuclear extracts were carried out as described previously (36) with some modifications. Briefly, transiently co-transfected COS7 cells were collected by centrifugation, washed with phosphate-buffered saline, and then lysed in the hypotonic buffer (10 mM HEPES-KOH, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 1 mM Na 3 VO 4 , 20 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin). Cells were incubated for 15 min and then vortexed vigorously and centrifuged at 12,000 ϫ g at 4°C for 20 s. The pellet was washed once with cold phosphate-buffered saline, and then the nuclear extracts were obtained by adding a high salt buffer (25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 1 mM Na 3 VO 4 , 20 mM NaF, 10 g/ml leupeptin), shaking for 30 min at 4°C, and then centrifuging at 12,000 ϫ g for 5 min.

Identification of PTPs That Dephosphorylate PRL-induced
Phosphorylated STAT5a-We have experienced that mammary epithelial cell lines COMMA-1D, HC11, and primary mammary epithelial cells are quite resistant to gene transfection. Therefore, to study the involvement of each PTP identified in PRL receptor-mediated signaling, co-transfection strategy was employed using COS7 cells. With this strategy, SHP-2 was shown to be a positive regulator of PRL receptor-mediated signal transduction pathway (29). Among the 16 PTPs identified (30), the following 12 PTPs were examined: PTP1B, SHP-1, SHP-2, PTP36, HCSF, PTP-BAS, PTP␣, PTP⑀, LAR, PTP, PTP, and PTP. COS7 cells were triple co-transfected with the expression plasmids for PRL receptor, STAT5a, and the respective PTPs and were stimulated with PRL following serum starvation. STAT5a was immunoprecipitated, and its tyrosine phosphorylation level was assessed by immunoblotting with the anti-phosphotyrosine antibody. As shown in Fig. 1

W I T H D R A W N W I T H D R A W N
F e b r u a r y 1 5 , 2 0 1 1 F e b r u a r y 1 5 , 2 0 1 1 protein tyrosine phosphatase PTP1B was co-transfected, tyrosine phosphorylation of STAT5a was strongly reduced and nearly undetectable. Although the expression level of other PTPs was obvious ( Fig. 1C), the other PTPs including SHP-1 and SHP-2, on the other hand, exhibited apparently no effect on the tyrosine phosphorylation of STAT5a, which was consistent with previous reports (29). PTP1B Dephosphorylates PRL-activated STAT5a and STAT5b-To confirm the possible dephosphorylation of STAT5 by PTP1B, catalytically inactive mutants of PTP1B were constructed and co-transfected into COS7 cells with PRL receptor and STAT5a or STAT5b. Upon overexpression of PTP1B wildtype, PRL-induced tyrosine phosphorylation of both STAT5a and STAT5b was largely abolished ( Fig. 2A). Dephosphorylation activity was not observed when the cells were co-transfected with catalytically inactive Cys/Ser and Asp/Ala mutants of PTP1B ( Fig. 2A), suggesting that phosphatase activity is essential for the dephosphorylation of STAT5 proteins.
To relate the expression level of PTP1B with the dephosphorylation of STAT5a and STAT5b, various amounts of expression plasmids for PTP1B were co-transfected into COS7 cells, and tyrosine phosphorylation of STAT5a and STAT5b was assessed. Approximately 80% of STAT5a was dephosphorylated when 0.1 g of PTP1B was co-transfected (Fig. 2B, lane 3), and transfection with higher amounts of plasmid (1 and 2 g) abolished tyrosine phosphorylation of STAT5a (Fig. 2B, lanes 2 and 1, respectively). Nearly the same dephosphorylation of STAT5b was caused by expression of PTP1B wild-type in a manner dependent of its expression level (Fig. 2B, lanes 5-8).
Prolactin binding to its cognate receptor results in the autophosphorylation and activation of JAK2, and in turn JAK2 tyrosine phosphorylates both STAT5a and STAT5b. Tyrosine phosphorylation of JAK2 is essential for its full activation (5). If PTP1B dephosphorylates JAK2, tyrosine phosphorylation of STAT5a and STAT5b should be reduced accordingly. To know direct or indirect dephosphorylation activity of PTP1B against STAT5, COS7 cells were co-transfected with PRL receptor and PTP1B. Cells were serum-starved, stimulated with PRL for 30 min, lysed, and then subjected to immunoprecipitation with anti-JAK2 antibody followed by immunoblotting with antiphosphotyrosine antibody. As clearly shown in Fig. 2C, no dephosphorylation of JAK2 was observed when PTP1B was co-transfected.
To confirm dephosphorylation action of PTP1B on STAT5a and STAT5b, recombinant GST fusion proteins containing fulllength PTP1B were expressed in Escherichia coli and purified. GST-PTP1B wild-type exhibited catalytic activity against artificial substrate para-nitrophenyl phosphate, whereas Cys/Ser and Asp/Ala mutants showed no activity (data not shown). COS7 cells that had been co-transfected with PRL receptor and STAT5a or STAT5b were stimulated with PRL and phosphorylated, STAT5a or STAT5b was immunoprecipitated. The in-

W I T H D R A W N W I T H D R A W N
F e b r u a r y 1 5 , 2 0 1 1 F e b r u a r y 1 5 , 2 0 1 1 dicated amounts of the recombinant GST-PTP1B fusion proteins were added to the immune complexes and incubated at 37°C for 30 min. As clearly illustrated in Fig. 3A, tyrosine phosphorylation level of STAT5a was reduced to approximately 50% by 1 g of GST-PTP1B wild-type and incubation with 10 g of the fusion protein resulted in complete dephosphorylation of STAT5a (upper panels). In a similar manner, STAT5b was dephosphorylated by GST-PTP1B wild-type (lower panels). Incubation of the immune complexes with empty GST and GST fused to catalytically inactive mutants of PTP1B resulted in no reduction in tyrosine phosphorylation level of STAT5a and STAT5b. In addition, phosphorylated JAK2 was immunoprecipitated from PRL-stimulated cells with anti-JAK2 antibody and incubated with 10 g of GST, GST-PTP1B wild-type, or its mutants as above. No tyrosine dephosphorylation was observed, which is consistent with the in vivo data shown in Fig. 2C.
We then examined the kinetics of STAT5a and STAT5b dephosphorylation by PTP1B. Transfected COS7 cells were serum-starved and stimulated with PRL for 2-60 min. STAT5a and STAT5b were immunoprecipitated and assessed by phosphotyrosine immunoblotting. As shown in Fig. 4A, PRL induced rapid tyrosine phosphorylation of STAT5a and STAT5b within 5 min in mock-transfected COS7 cells, and the levels of tyrosine phosphorylation were kept high even 60 min after PRL stimulation. This time course of tyrosine phosphorylation of STAT5a was quite similar to endogenous STAT5a/b of mammary epithelial COMMA-1D cells (Fig. 4B). By overexpressing PTP1B wild-type, dephosphorylation of STAT5a and STAT5b already occurred 5 min after PRL stimulation, and only faint signals were detected 30 and 60 min after PRL stimulation.
PTP1B Inhibited Nuclear Translocation of STAT5a and STAT5b-It has been reported that PTP1B is localized in endoplasmic reticulum through its C-terminal hydrophobic amino acid residues (37). To address the possibility that STAT5 might be dephosphorylated by PTP1B in cytosol, subcellular localization of STAT5a and STAT5b following PRL stimulation was FIG. 2. Dephosphorylation of STAT5a and STAT5b by PTP1B in transfected COS7 cells. A, COS7 cells were co-transfected with expression plasmids for PRL receptor (1 g), STAT5a or STAT5b (1 g), and empty vector (mock) or each HA-PTP1B wild type (WT), catalytically inactive Cys/Ser (C/S) and Asp/Ala (D/A) mutants (2 g for each). Following serum starvation, cells were left untreated (Ϫ) or stimulated (ϩ) with PRL (5 g/ml) for 30 min and lysed. STAT5a and STAT5b were immunoprecipitated (IP), separated on SDS-PAGE, transferred to a nitrocellulose membrane, and probed with anti-phosphotyrosine antibody (Ab). The membrane was stripped and reprobed with anti-STAT5 antibody. B, COS7 cells were co-transfected with expression plasmids for PRL receptor (1 g), STAT5a or STAT5b (1 g), and varying amounts of HA-PTP1B wild type as indicated and stimulated with PRL (5 g/ml) following serum starvation. STAT5a and STAT5b were immunoprecipitated and processed as mentioned above. An aliquot of total cell lysate was immunoblotted with anti-HA antibody for the expression of HA-PTP1B. C, COS7 cells were co-transfected with expression plasmids for PRL receptor (2 g) and empty vector (mock) or each of HA-PTP1B wild type, catalytically inactive Cys/Ser and Asp/Ala mutants (2 g for each), serum-starved, and left untreated (Ϫ) or stimulated with PRL (5 g/ml) as above. JAK2 was immunoprecipitated, separated by SDS-PAGE, and immunoblotted with anti-phosphotyrosine antibody. The membrane was stripped and reprobed with anti-JAK2 antibody.

FIG. 3. In vitro dephosphorylation of STAT5 by recombinant PTP1B.
A, COS7 cells were co-transfected with expression plasmids for PRL receptor (2 g) and STAT5a or STAT5b (2 g) and stimulated with PRL (5 g/ml) for 30 min following serum starvation. STAT5a and STAT5b were immunoprecipitated, washed with lysis buffer, and then subjected to in vitro dephosphorylation assay as described under "Experimental Procedures." Following termination of the incubation, proteins were separated by SDS-PAGE and analyzed with anti-phosphotyrosine antibody (Ab) (4G10). The same blot was reprobed with anti-STAT5 antibody following stripping. B, COS7 cells were co-tranfected with expression plasmids for PRL receptor (2 g) and JAK2 (2 g) and stimulated with PRL (5 g/ml) for 30 min following serum starvation. JAK2 was immunoprecipitated and processed as above.

W I T H D R A W N W I T H D R A W N
F e b r u a r y 1 5 , 2 0 1 1 F e b r u a r y 1 5 , 2 0 1 1 examined. Cytoplasmic and nuclear fractions were prepared after PRL stimulation for the times indicated, and STAT5 proteins were immunoprecipitated and immunoblotted with anti-STAT5 antibody. As shown in Fig. 5A, the amounts of STAT5a as well as STAT5b in the nucleus increased within 5 min following PRL stimulation. Conversely, the amounts of cytoplasmic STAT5a and STAT5b decreased. Such nuclear translocation of STAT5 proteins was observed in COMMA-1D cells in a similar kinetics (Fig. 5C). Upon overexpression of PTP1B wild-type, nuclear translocation of STAT5a and STAT5b was disrupted, and most of them were retained in cytosol, whereas the catalytically inactive mutants exhibited no effect (data not shown). Subcellular localization of PTP1B was also examined. Upon PRL stimulation for the times indicated, PTP1B was immunoprecipitated and immunoblotted with anti-HA antibody. Throughout the times following PRL stimulation, most of PTP1B was present in cytosol and detected as an expected 50-kDa protein, whereas only marginal amounts of PTP1B were detected in the nuclear fraction (Fig. 5B).
Transcriptional Induction of the ␤-Casein Gene Promoter Was Inhibited by PTP1B-We then studied the effect of PTP1B on PRL-induced transcriptional activation of the ␤-casein gene promoter. PTP1B was transfected into COS7 cells together with PRL receptor, STAT5a or STAT5b, and the ␤-casein gene promoter-luciferase construct. A ␤-galactosidase gene was also included to normalize for transfection efficiency. Luciferase activity was determined in extracts from cells left untreated or stimulated with PRL. As shown in Fig. 6A, both STAT5a and STAT5b together with the PRL receptor could strongly activate the ␤-casein gene promoter upon PRL stimulation of the cells that had been mock-tranfected with the empty plasmid without PTP (lanes 1, 2, 9, and 10). When PTP1B wild-type was coexpressed, such transcriptional induction was completely suppressed (lanes 4 and 12), whereas PTP1B Cys/Ser and Asp/Ala mutants exhibited slightly higher ␤-casein promoter activation as compared with mock transfectants (lanes 6, 8, 14, and 16). Furthermore, expression of PTP1B wild type caused a dose-dependent suppression of transactivation of STAT5a and STAT5b by PRL (Fig. 6B). These results indicate that PTP1B is a negative regulator of PRL-mediated ␤-casein promoter activation, by dephosphorylating STAT5a and STAT5b.
PTP1B Is a Negative Regulator in PRL Receptor-mediated Signaling in Mammary Epithelial Cells-As mentioned, mammary epithelial cells are quite resistant to gene transfection. Therefore, it was required to use virus-mediated gene infection strategy. PTP1B cDNA was ligated into a retroviral vector and introduced into mammary epithelial COMMA-1D cells. Cells were selected in cell culture medium containing G418 and then directly used for experiments. As shown in Fig. 7A, nearly the same amounts of HA-tagged PTP1B wild type, Cys/Ser, and Asp/Ala were expressed in the cells. Cells were serum-starved and stimulated with PRL, and then endogenous STAT5 was immunoprecipitated followed by immunoblotting with antiphosphotyrosine antibody. As was seen in reconstituted COS7 cells, phosphorylation level of STAT5 was significantly low when PTP1B wild-type was overexpressed, whereas overexpression of PTP1B mutants caused no dephosphorylation of STAT5 in mammary epithelial cells (Fig. 7B). On the other hand, overexpression of PTP1B did not affect the tyrosine phosphorylation level of JAK2 (Fig. 7C). Moreover, ␤-casein gene expression was suppressed in PTP1B wild type-expressing COMMA-1D cells, whereas that in PTP1B mutant-expressing clones was comparable to that in mock-infected cells (Fig. 7D). These clearly indicate that PTP1B is a negative regulator in PRL receptor-mediated signaling pathway in mammary epithelial cells.
STAT5 Is a Specific Substrate of PTP1B-To confirm further that STAT5 is a specific substrate of PTP1B, a co-precipitation study was carried out using recombinant GST-PTP1B fusion proteins. COMMA-1D cells were stimulated with PRL for 30 min following serum starvation and lysed. The cell lysates were mixed with 10 g of empty GST or GST-PTP1B fusion proteins and were being rocked together with GSH-Sepharose beads at 4°C for 3 h. The GSH-Sepharose beads were washed with the lysis buffer and dissolved in SDS sample buffer. Proteins were separated by SDS-PAGE (10% gel) and blotted onto nitrocellulose membranes. The membranes were probed with anti-phosphotyrosine antibody. As shown in Fig. 8A, many tyrosinephosphorylated proteins including a 97-kDa protein were coprecipitated by GST-PTP1B Asp/Ala mutant, whereas much less proteins were co-precipitated by the wild-type and the catalytically inactive Cys/Ser mutant of PTP1B recombinant proteins. The 97-kDa band was marginally detected in the Cys/Ser precipitates. The membranes were stripped and reprobed with anti-STAT5 antibody. The tyrosine-phosphorylated 97-kDa band co-precipitated with the PTP1B Asp/Ala mutant was demonstrated to be STAT5 (Fig. 8A, lower panel). STAT5 was also present in the precipitates of PTP1B Cys/Ser mutant. Interestingly, STAT5 was detected in the precipitates of PTP1B wild-type, although no tyrosine-phosphorylated band corresponding to 97 kDa was detected by anti-phosphotyrosine antibody. Other tyrosine-phosphorylated bands co-precipitated FIG. 4. Time course of PRL-stimulated STAT5 tyrosine phosphorylation. A, COS7 cells were co-transfected with expression plasmids for PRL receptor (1 g), STAT5a or STAT5b (1 g), and empty vector (mock) or HA-PTP1B wild type (2 g for each). Following serum starvation, cells were left untreated (0 min) or stimulated with PRL (5 g/ml) for the times indicated. Cells were lysed, and STAT5a and STAT5b were immunoprecipitated (IP), separated by SDS-PAGE, transferred to a nitrocellulose membrane, and probed with anti-phosphotyrosine antibody (Ab). The membrane was stripped and reprobed with anti-STAT5 antibody. B, COMMA-1D cells were left untreated (0 min) or stimulated with PRL (5 g/ml) for the indicated times following serum starvation and then processed as above.  F e b r u a r y 1 5 , 2 0 1 1  F e b r u a r y 1 5 , 2 0 1 1 with GST fusion proteins have not been well characterized so far. To confirm that, COS7 cells, which had been co-transfected with PRL receptor and STAT5a or STAT5b, were stimulated with PRL and the lysates were precipitated with the GST-PTP1B fusion proteins as above. As clearly shown in Fig. 8B, phosphorylated STAT5a or STAT5b was detected in the precipitates of the Asp/Ala and to a lesser extent in Cys/Ser mutants, whereas no signal was observed in the precipitates of the wild type. The immunoblotting with anti-STAT5 also showed that the Asp/Ala mutant precipitated the STAT5 most strongly. However, the anti-STAT5 immunoblotting also revealed that STAT5a and STAT5b were co-precipitated by not only substrate-trapping mutants of PTP1B but also the wild type, suggesting some contribution of phosphotyrosine-independent interaction between STAT5 and PTP1B. DISCUSSION Although the mechanisms how STAT proteins become activated have been well characterized, much less is known about their subsequent deactivation process. Recent publications have focused on the negative regulation of STAT proteins, and several mechanisms have been documented for deactivation of STAT5. A family of JAK kinase-binding protein or cytokineinducible SH2-containing protein has been shown to downregulate STAT5 by inhibiting upstream JAK kinase activity or preventing recruitment of STAT5 to cytokine receptors (38 -43). It has also been demonstrated that the function of STAT5 can be modulated by the ubiquitin-proteasome pathway (25, 44 -46). Although tyrosine phosphorylation of STAT5 is the essential step for its full biological functions (36), dephosphorylation of STAT5 has been largely unknown.

W I T H D R A W N W I T H D R A W N
SHP-2 and structurally related SHP-1 have been shown to be widely implicated in distinct JAK-STAT pathways. Positive involvement of SHP-2 in PRL receptor-mediated signaling pathway has been shown by two groups (28,29). Upon PRL stimulation, SHP-2 forms a trimeric complex with PRL receptor and JAK2, and SHP-2 itself becomes tyrosine-phosphorylated. Catalytically inactive mutants of SHP-2 inhibited the signaling pathway in a dominant negative fashion, suggesting that the phosphatase activity of SHP-2 is essential, possibly through dephosphorylating essential tyrosine residues on JAK2 for its full activation (29). SHP-2 has also been shown to be involved positively in IL-2-mediated signaling in NK cells (47), whereas it plays a negative role in gp130 signaling (48). It is also reported that SHP-1 is activated upon growth hormone stimulation and might dephosphorylate STAT5b in rat liver cells (49). Thus, SHP-2 and SHP-1 might play a positive or negative role depending on the cells and tissue where they express.
In the present study, we showed that a cytosolic phosphatase PTP1B dephosphorylated PRL-activated STAT5a and STAT5b In contrast, other cytoplasmic PTPs, especially SHP-1 and SHP-2, and receptor type PTPs had no effect on the tyrosine phosphorylation level of STAT5a (Fig. 1) and STAT5b (data not shown). Further detailed analyses revealed that catalytic activity of PTP1B is essential for the dephosphorylation of STAT5a and STAT5b and that PTP1B directly dephosphorylated STAT5a and STAT5b but not JAK2, upstream regulator of STAT5 (Figs. 2 and 3). Additionally, tyrosine-phosphorylated STAT5a and STAT5b were co-precipitated with the substrate-trapping mutants of PTP1B (Fig. 8), and PRL-induced transcriptional activation of STAT5a and STAT5b was disrupted when PTP1B was overexpressed (Fig. 6). Moreover, PTP1B was shown to be a negative regulator in PRL receptor-mediated signaling pathway leading to up-regulation of ␤-casein gene expression in mammary epithelial COMMA-1D cells (Fig. 7). These results strongly and clearly suggest that PTP1B is a principal PTP specifically dephosphorylating and deactivating PRL-activated STAT5a and STAT5b. Deactivation of tyrosine-phosphorylated nuclear STAT proteins involves both tyrosine dephosphorylation and export from nuclear back to the cytosol for a subsequent cycle of activation and inactivation (50). Our data showed that nuclear translocation of STAT5a and STAT5b was largely inhibited when PTP1B wild-type was overexpressed. Since tyrosine phosphorylation of STAT5 is essential for its dimerization, nuclear translocation, and binding to its target genes, our data might simply imply that PTP1B present in cytosol blocks the nuclear translocation by trapping and dephosphorylating tyrosine-phosphorylated STAT5.
Our previous data showed that expression of PTP1B was largely suppressed in lactating mammary gland producing a huge amount of milk proteins under the control of PRL and that the expression level returned to that in virgin mammary gland immediately after suckling cessation and weaning of FIG. 6. Transcriptional induction of the ␤-casein gene promoter was inhibited by PTP1B. A, COS7 cells were co-transfected with expression plasmids for PRL receptor (1 g), STAT5a or STAT5b (1 g), ␤-casein gene promoter-luciferase (1 g), and empty vector (mock) or each of the HA-PTP1B wild types (WT), catalytically inactive Cys/Ser (C/S) and Asp/Ala (D/A) mutants (2 g). A ␤-galactosidase gene (0.4 g) was also included to normalize for transfection efficiency. Cells were induced with PRL for 15 h (ϩ, odd lanes) or left untreated (Ϫ, even lanes) and then lysed for enzymatic assay. B, COS7 cells were co-transfected with expression plasmids for PRL receptor (1 g), STAT5a or STAT5b (1 g), ␤-casein gene promoter-luciferase (1 g), and empty vector (mock) or various amounts of HA-PTP1B wild-type (2, 1, 0.1, and 0 g in lanes 3-6 and lanes 9 -12, respectively) and processed as above. Luciferase activity was represented as fold induction to that of mock transfectant without PRL induction. Data are shown as mean values Ϯ S.D. of three independent experiments. FIG. 7. PTP1B is a negative regulator in PRL receptor-mediated signaling pathway in mammary epithelial COMMA-1D cells. A, COMMA-1D cells were retrovirally infected with HA-PTP1B wild type (WT), Cys/Ser (C/S), or Asp/Ala (D/A) mutant and selected in the cell culture medium supplemented with G418 (1 mg/ml). Polyclonal clones for each were lysed, and aliquots were immunoblotted with anti-HA antibody (Ab). B and C, COMMA-1D cells expressing PTP1B were stimulated with PRL (5 g/ml) for 30 min following serum starvation. STAT5 (B) or JAK2 (C) was immunoprecipitated (IP) and processed as mentioned in the legend to Fig. 2. D, COMMA-1D cells expressing PTP1B were treated with PRL (5 g/ml) and hydrocortisone (0.1 M) for 48 h. RNA was prepared and reverse transcriptase-PCR amplification for ␤-casein and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was carried out as described (30).

W I T H D R A W N W I T H D R A W N
F e b r u a r y 1 5 , 2 0 1 1 F e b r u a r y 1 5 , 2 0 1 1 pups (30). This might partially support a possible role of PTP1B in negative regulation of the JAK2-STAT5 pathway leading to ␤-casein promoter activation, although most of the PTPs identified in mammary gland and mammary epithelial cells were also down-regulated in lactating mammary gland. Therefore, we cannot exclude the possibility that other cytoplasmic as well as nuclear phosphatases other than those examined in the present study dephosphorylate and deactivate the PRL-activated STAT5 proteins.
PTP1B was originally purified from the cytosolic fraction of human placenta as a 37-kDa protein (51) and is now known to be ubiquitously and abundantly expressed in various eukaryotic cells and be associated with endoplasmic reticulum through its C-terminal hydrophobic 35-amino acid region (37), although there is also evidence indicating its association with the plasma membrane (52). We showed that most of exogenously expressed PTP1B was recovered in cytosol, and its location, amount, and size were unchanged following PRL stimulation. Constitutive existence of endogenous PTP1B in cytosol might grant concomitant tyrosine dephosphorylation of STAT5. It has been reported that PTP1B is serine-phosphorylated upon 12-O-tetradecanoylphorbol-13-acetate treatment through the action of protein kinase C (53) and CLK1 and CLK2 (54) and that PTP1B undergoes tyrosine phosphorylation upon epidermal growth factor stimulation (55). In our examination, no phosphorylation of tyrosine and serine/threonine residues was detected by immunoblotting analysis using anti-phosphotyrosine and anti-phosphoserine/threonine antibodies (data not shown). Whether modulation of PTP1B occurs after PRL stimulation remains to be determined.
Recently, Ali and Ali (56) have reported that STAT5 tyrosine phosphorylation and nuclear translocation are regulated by two separate pathways by using a variety of mutants of the PRL receptor Nb2 form (56). Substitution of tyrosine 382 with alanine on the receptor did not affect tyrosine phosphorylation of STAT5, but both nuclear translocation and binding to the target DNA sequence were inhibited. This suggests potential involvement of PTPs, which dephosphorylate the tyrosine residues on the PRL receptor, in PRL-mediated target gene activation. Indeed, we have observed a receptor-type phosphatase PTP dramatically inhibited PRL-mediated ␤-casein promoter activation without affecting the phosphorylation level of STAT5. 2 In conclusion, we demonstrated in this study for the first time the specific tyrosine dephosphorylation of PRL-induced STAT5 by cytoplasmic PTP1B. We further reported that nuclear translocation and transcriptional activities of STAT5 were largely inhibited. In addition to STAT5, it has been shown that the PRL receptor-mediated signaling cascade also results in tyrosine phosphorylation of STAT1 and STAT3 (57). Whether PTP1B is also involved in negative regulation in other JAK-STAT pathways is currently in progress in our laboratory.