Convergence of Progesterone with Growth Factor and Cytokine Signaling in Breast Cancer

STATS (signal transducers and activators of transcription) are latent transcription factors activated in the cytoplasm by diverse cell surface signaling molecules. Like progesterone receptors (PR), Stat5a and 5b are required for normal mammary gland growth and differentiation. These two proteins are up-regulated during pregnancy, a period dominated by high levels of progesterone. We now show that progestin treatment of breast cancer cells regulates Stat5a and 5b, Stat3, and Stat1 protein levels in a PR-dependent manner. In addition, progestin treatment induces translocation of Stat5 into the nucleus, possibly mediated by the association of PR and Stat5. Last, progesterone pretreatment enhances the phosphorylation of Stat5 on tyrosine 694 induced by epidermal growth factor. Functional data show that progestin pretreatment of breast cancer cells enhances the ability of prolactin to stimulate the transcriptional activity of Stat5 on a β-casein promoter. Progesterone and epidermal growth factor synergize to control transcription from p21WAF1 and c-fospromoters. These data demonstrate the convergence of progesterone and growth factor/cytokine signaling pathways at multiple levels, and suggest a mechanism for coordination of PR and Stat5-mediated proliferative and differentiative events in the mammary gland.

Progesterone, acting through progesterone receptors (PR), 1 is important in the control of breast cell proliferation and differentiation (1). Mice lacking PR exhibit incomplete mammary gland ductal branching and failure of lobulo-alveolar development (2). Interestingly, similar disruptions of mammary gland development and lactation are observed in mice upon deletion of several other genes including cyclin D1 (3), prolactin receptors (4), the activin/inhibin ␤B gene (5) and most recently, two members of the STAT (signal transducers and activators of transcription) family, Stat5a (6) and Stat5b (7). It is evident that the products of these genes are involved in pathways that influence the mammary gland; however, it remains to be determined whether they function in distinct or convergent pathways.
STAT family members are latent cytoplasmic transcription factors activated via diverse cell surface signaling molecules (8). Although Stat5 is functional in many cell types, Stat5adeficient mice are normal except for lack of proliferative mammary lobulo-alveolar outgrowth, as well as inability of females to lactate, thus demonstrating a mandatory and specific role for this STAT in both processes (6). The role of Stat5a (originally identified as mammary gland factor in extracts from lactating mice) as a mediator of prolactin-induced transcription of milk protein genes during lactation is well documented (9). Stat5bdeficient mice show a phenotype mainly in sexual dimorphism of growth rates and liver gene expression; however, females also have impaired mammary gland development. Stat5bϪ/Ϫ females consistently abort their pups, a problem that can be overcome by progesterone administration during pregnancy (7). In relation to progesterone action, both Stat5a and 5b are particularly interesting since mRNA and protein levels rise during pregnancy (10), a period dominated by high levels of progesterone.
STATs are known to be activated by the ligand-induced intrinsic tyrosine kinase activity of members of the growth factor receptor family or by cytokine receptors that lack intrinsic tyrosine kinase activity, but associate with soluble tyrosine kinases known as Janus kinases (JAKs), as reviewed in Refs. 11 and 12. Phosphorylation of a tyrosine residue (tyrosine 694 in Stat5) conserved in all STAT family members induces their dimerization, which is followed by translocation into the nucleus, DNA binding, and regulation of numerous genes involved in growth and differentiation (8). In the lactating mammary gland, the cytokine prolactin, acting via prolactin receptors and JAK2, induces phosphorylation of Stat5a and 5b, which then bind sites on the promoters of mammary-specific genes such as ␤-casein. In addition, growth factors such as epidermal growth factor (EGF) can also stimulate Stat5, as well as Stat1 and Stat3, which then bind to STAT sites on the promoters of growth regulatory genes such as p21 WAF1 (13)(14)(15) and c-fos (16).
Despite conflicting in vivo and in vitro data regarding the proliferative versus growth inhibitory role of progesterone in the breast, the lack of mammary gland ductal branching and failure of lobulo-alveolar development observed in the PRknockout mouse (2) demonstrates that PR must play a proliferative role during development. Interestingly, in vitro, progestins stimulate breast cancer cells to progress through one round of cell division accompanied by the induction of cyclin D1, p21 WAF1 , EGF, EGFR, c-myc, and c-fos. This is followed by growth arrest at the G 1 /S phase of the second cycle (17)(18)(19). We have proposed that progesterone-arrested cells are poised to respond to secondary proliferative or differentiative signals ((18) and the accompanying paper (20)).
There is considerable evidence that progesterone and EGF have complementary effects on the mammary gland. Like progestins and PR, EGF and EGFR are required in the proliferative phase of mammary gland development (21,22). Mice carrying a spontaneous mutation resulting in a critical amino acid substitution in the kinase domain of the EGFR have underdeveloped ductal trees and impaired lactation (23). Furthermore, progestin treatment: 1) up-regulates EGFR (17,24,25) and other type I growth factor receptors (20) in breast cancer cell lines; 2) enhances the ability of EGF to induce proliferation of breast cancer cells (18); and 3) potentiates EGF mediated signaling pathways (20).
Since STAT proteins are downstream effectors of EGFR (13-16) we postulated that cross-talk between progesterone and EGF occurs at the level of STATs. That steroid hormone and growth factor signaling pathways converge at STAT proteins is suggested by the recent report of a functional interaction between glucocorticoid receptors (GR) and Stat5 in which GR act as coactivators for Stat5-mediated induction of the ␤-casein promoter by glucocorticoids and prolactin (26). Conversely, Stat3 can act as a coactivator of GR-mediated transcription on a mouse mammary tumor virus promoter in the presence of interleukin-6 and dexamethasone (27).
We now report that progestin treatment of breast cancer cells up-regulates Stat5 and Stat3 protein levels in a PR-dependent manner. In addition, progestin treatment induces translocation of Stat5 into the nucleus, possibly mediated by the physical association of PR and Stat5 proteins. Functional data show that progestin treatment of breast cancer cells enhances the ability of prolactin to stimulate the transcriptional activity of Stat5 on a ␤-casein promoter, and synergizes with EGF to control transcription of the p21 WAF1 and c-fos promoters. These data demonstrate the regulation of a key growth factor signaling molecule by progesterone and suggest a mechanism for coordination of PR and Stat5-mediated proliferative and/or differentiative events in the mammary gland.

MATERIALS AND METHODS
Cell Lines and Reagents-The wild type PR-positive T47Dco breast cancer cell line and its clonal derivatives T47D-Y, T47D-YA, and T47D-YB, have been described (28). Cells are routinely cultured in 75-cm 2 plastic flasks and incubated in 5% CO 2 at 37°C in a humidified environment. The stock medium consists of Eagle's minimum essential medium with Earle's salts (MEM), containing L-glutamine (292 g/liter) buffered with sodium bicarbonate (2.2 g/liter), insulin (6 ng/ml), and 5% fetal bovine serum (Hyclone, Logan, UT) without antibiotics. For routine subculturing, cells are diluted 1:20 into new flasks once per week, and medium replaced every 2-3 days. Cells are harvested by incubation in Hank's-EDTA for 10 min at 37°C. For time course experiments, cells are plated at 1 million cells per plate in MEM with supplements described above and were treated with 10 nM progesterone (Sigma) or R5020 (NEN Life Science Products Inc., Boston, MA). Cells were harvested at specified time points in RIPA buffer (10 mM sodium phosphate, pH 7.0, 150 mM NaCl, 2 mM EDTA, 1% deoxycholic acid, 1% Nonidet P-40, 0.1% SDS, 0.1% ␤-mercaptoethanol, 1 mM PMSF, 50 mM sodium fluoride, 200 M Na 3 VO 4 , and one Complete Protease Inhibitor Mixture tablet (Boehringer Mannheim, GmbH Germany) per 50 ml). EGF (Collaborative Biomedical Products, Bedford, MA) was used at 10 nM and prolactin (Sigma) at 5 g/ml.
Immunoblotting-Protein extracts were equalized to 100 g by the Bradford assay (Bio-Rad), resolved by SDS-PAGE, and transferred to nitrocellulose. Equivalent protein loading was confirmed by Ponceau S staining. Following incubation with the appropriate antibodies, protein bands were detected by enhanced chemiluminescence (Amersham, Arlington Heights, IL) and quantitated using a Molecular Dynamics Se-ries 300 Computing Densitometer and Molecular Dynamics Image-Quant Program. The following antibodies were obtained from Santa Cruz Biotechnology, Santa Cruz, CA: Stat5 C-17 (recognizes both Stat5a and 5b isoforms); Stat3 (C-20); JAK2 (HR-758); and p21 (C-18). Anti-phosphotyrosine (monoclonal 4G10), anti-human Stat5A, and anti-human Stat5B (specific for the Stat5a or 5b isoform), anti-phospho-Stat5 (specific for Tyr-694) and anti-human cdc2 kinase (PSTAIR) were purchased from Upstate Biotechnology, Lake Placid, NY. The anti-PR monoclonal antibody AB-52 was produced in our laboratories.
Northern Blotting-Breast cancer cells stably expressing either the PR-A (T47D-YA) or PR-B (T47D-YB) isoform were treated for 12 or 24 h with 10 nM progesterone or ethanol vehicle. Total RNA was isolated by ultracentrifugation of guanidinium isothiocyanate lysates through a CsCl cushion (29). RNA (30 g) from each treatment group was transferred to a Hybond nylon membrane (Amersham) and hybridized sequentially with the cDNA inserts described below labeled by random priming with [ 32 P]dCTP using the Mega-Prime DNA Labeling Kit (Amersham). The membrane was stripped in boiling 0.1 ϫ SSC, 0.1% SDS between hybridizations. A 2-kilobase insert was removed from a cDNA clone encoding fatty acid synthetase by restriction digest with EcoRI and HindIII followed by gel isolation and purification. The cyclin D1 cDNA was a 1.1-kilobase fragment removed from the vector by restriction digest with XbaI and HindIII. The Stat5a cDNA probe consisted of a 2.3-kilobase fragment cut from the vector by EcoRI digest. Last, a glyceraldehyde-3-phosphate dehydrogenase cDNA probe representing a non-regulated gene served as a control for RNA loading. Fatty acid synthetase clone pG8 (30)  Anti-phosphotyrosine Co-immunoprecipitations-Cells in 10-cm dishes were washed twice with ice-cold phosphate-buffered saline and lysed by scraping in extraction buffer (EB: 1% Triton X-100, 10 mM Tris-HCl (pH 7.4), 5 mM EDTA, 50 mM NaCl, 50 mM sodium fluoride, 2 mM Na 3 VO 4 , and 1 mM PMSF). Lysates were clarified by centrifugation for 10 min at maximum speed in a Savant (Speedfuge SFR13K) bench-top centrifuge and equal amounts of protein (1 mg/ml) were immunoprecipitated with the anti-phosphotyrosine monoclonal antibody 4G10 (4 g/mg) by rotation at 4°C for 2 h to overnight. Immuno complexes were captured by adding 30 l of washed protein A ((insoluble formalin-fixed StaphA-derived Sorbin (Sigma)) that had been preincubated with rabbit anti-mouse antibody, incubated by constant rotation at 4°C for an additional 2 h, then collected by centrifugation at 10,000 rpm for 3 min in a Savant bench-top centrifuge. Immunoprecipitates were washed twice in EB (1 ml), twice in PAN (10 mM PIPES (pH 7.0), 100 mM NaCl) containing 0.25% Nonidet P-40, and twice in PAN without Nonidet P-40. Washed pellets were resuspended in Laemmli sample buffer, boiled for 3 min, and analyzed by SDS-PAGE and immunoblotting.
Stable Cell Lines Expressing Flag-tagged PR and Co-immunoprecipitation of PR and Stat5-To facilitate immunoprecipitation studies, hPR1, the full-length PR-B cDNA cloned into the mammalian expression vector pSG5 from P. Chambon (Strasbourg, France) (31), was modified to disrupt the stop codon and add a carboxyl-terminal flag epitope consisting of amino acids DYKDDDDK. To verify that the FLAG epitope-tagged PR B (PR B :f) was expressed and functional in vivo, HeLa cells were co-transfected with PR B :f and the PRE 2 -TATA tk -CAT reporter using calcium phosphate precipitation as described previously (32), and treated with the synthetic progestin R5020. The chloramphenicol acetyltransferase activity produced by the PR B :f was equal to that of wild-type hPR1 (data not shown).
A stable cell line expressing PR B :f was established by co-transfecting HeLa cells with 4.5 g of PR B :f plasmid and 0.5 g of a plasmid encoding the neomycin resistance gene, pSV2neo. DNA precipitate was removed 18 h later, and cells were grown in MEM ϩ 5% fetal bovine serum containing 700 g/ml of the neomycin analog G418 (Life Technologies, Inc., Gaithersburg, MD) to kill non-transfected cells. Surviving neomycin-resistant colonies were expanded and cells were analyzed by immunoblotting and chloramphenicol acetyltransferase assay for clones expressing high levels of functional PR B :f.
For co-immunoprecipitations, wild type HeLa and HeLa PR B :f were cultured in MEM supplemented with 5% twice charcoal-stripped heatinactivated fetal bovine serum. Cells were treated with progesterone or R5020 (10 nM) for 1 h. Nuclear extracts were prepared according to the methods described in Ref. 33. After harvesting in Hanks'; EDTA, cells were washed in phosphate-buffered saline then resuspended in 5 packed cell volumes of buffer A (10 mM (pH 7.9) at 4°C, 1.5 mM MgCl 2 , 10 mM KCl, and 0.5 mM dithiothreitol, 0.5 mM PMSF, one complete protease inhibitor mixture tablet (Boehringer Mannheim, GmbH Germany) per 50 ml of buffer) and allowed to stand at 4°C for 10 min. The cells were collected by centrifugation at 2,000 rpm for 10 min, then resuspended in two packed cell volumes of buffer A and lysed by 10 strokes of a Kontes all glass Dounce homogenizer (B type pestle). The homogenate was centrifuged for 10 min at 2,000 rpm to pellet the nuclei. The supernatant was removed and the pellet subjected to a second centrifugation at 25,000 ϫ g for 20 min (Beckman Optima Le-80K Ultracentrifuge, 70.1 Ti rotor) to remove residual cytoplasmic material. The pellet was resuspended in 3 ml of buffer C per 5 ϫ 10 8 cells and re-homogenized as above. Buffer C consisted of 20 mM Hepes, 25% (v/v) glycerol, 0.42 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM PMSF, 0.5 mM dithiothreitol, and protease inhibitors. The homogenate was mixed by rotation for 30 min at 4°C then centrifuged for 30 min at 25,000 ϫ g. The resulting clear supernatant was dialyzed at 4°C for 5 h to overnight in 20 mM Hepes (pH 7.9), 20% glycerol, 0.1 M KCl, 0.2 M EDTA, 0.5 mM dithiothreitol, 0.5 mM PMSF, and protease inhibitors. One mg of nuclear extract was incubated with 100 l of a 50% slurry of Anti-Flag M2 Affinity Gel (Eastman Kodak, New Haven, CT) at 4°C for 4 h. The anti-Flag resin was then washed twice with TEDG containing 0.1 M NaCl, twice with TEDG containing 0.3 M KCl, and twice with TEDG containing 0.1 M NaCl and 0.1% Nonidet P-40, using 1 ml of buffer for each wash. The PR B :f protein was then eluted by competition with 0.2 mg/ml Flag peptide (N-DYKDDDDK-C) (Eastman Kodak) in 200 l of TEDG containing 0.3 M KCl and 0.1% Nonidet P-40, for 30 min at 4°C. Laemmli sample buffer was added to the eluate, and proteins were resolved by 7.5% SDS-PAGE, transferred to nitrocellulose, and identified by immunoblotting.
Transient Transfections-PR positive breast cancer cells plated at 1 million cells per 10-cm dish in MEM supplemented with 5% fetal bovine serum were treated with 10 nM R5020 or ethanol vehicle for 48 h prior to transfection. Cells were then transiently transfected with 3 g of ␤-casein Ϫ2300/ϩ490 promoter in the luciferase reporter plasmid pGL2-E (provided by Paul A. Kelly, Molecular Endocrinology, INSERM, Paris, France), 3 g of the ␤-galactosidase expression plasmid pCH110 (Amersham Pharmacia Biotech) to check transfection efficiency, and Bluescript carrier plasmid (Stratagene, La Jolla, CA) for a total of 20 g of DNA using calcium phosphate precipitation as described previously (32). Three hours after transfection, the medium was aspirated and the cells were shocked with 2 ml of phosphate-buffered saline containing 20% glycerol. Cells were then washed twice with serum-free MEM to remove the glycerol and 10 ml of MEM containing 5% charcoal-stripped fetal bovine serum was added for 18 h. Cells were then treated with either 10 nM R5020 or 5 g/ml prolactin in triplicate dishes, and harvested 24 h after hormone addition in 300 l of lysis solution (Analytical Luminescence Laboratories, Ann Arbor, MI), and 100 l of lysate was analyzed for luciferase activity using the Enhanced Luciferase Assay Kit and a Monolight 2010 Luminometer (Analytical Luminescence Laboratories, Ann Arbor, MI) as described by the manufacturer.
In experiments using the T47D-YB breast cancer cells to study the effects of progesterone plus EGF in combination on STAT site containing promoters p21 WAF1 and c-fos, the transfection protocol was as described above except cells were transfected 24 h after plating and treated with ethanol vehicle, progesterone, EGF, or both hormones together 18 h after glycerol shocking. Cells were harvested 24 h after hormone treatment. The Ϫ2320-base pair p21 promoter construct, a gift of A. Kraft and J. Biggs, Division of Oncology, University of Colorado Health Sciences Center, Denver, CO, was cloned into a pA3-LUC vector as described previously (34). The c-fos promoter, c-fos-81TK-luc, containing two copies of the Ϫ357/Ϫ276 promoter linked to a minimal (81 base pairs) thymidine kinase promoter was obtained from A. Gutierrez-Hartmann, Division of Endocrinology, University of Colorado Health Sciences Center, Denver, CO (35), and was modified from c-fos-TK-luc (36).

Regulation of STAT Protein Levels by Progesterone
Treatment-To examine the effects of progestins on STAT protein levels, PR-positive T47Dco breast cancer cells were treated with progesterone (not shown), 10 nM R5020, or ethanol vehicle (Fig. 1A) and cells were harvested 8 -72 h later. Total Stat5 protein present in whole cell lysates was detected using a polyclonal antibody that cross-reacts with both Stat5a and 5b.
Stat5 protein levels were elevated by 8 h and remained elevated until 72 h after R5020 treatment with increases of 4 -7fold (Fig. 1A). To test the PR dependence of this effect, the experiment was repeated using PR-negative T47D-Y cells (28) (Fig. 1B). In the absence of PR, R5020 did not up-regulate Stat5.
The two isoforms of Stat5, Stat5a and Stat5b, share 96% similarity at the protein level (Fig. 2). Aside from the nonconserved 5Ј-and 3Ј-untranslated regions, the main difference between the two isoforms is in the COOH terminus. The last 8 amino acids of the two isoforms are completely divergent, and Stat5a is 7 amino acids longer than Stat5b (37). To determine which of the two isoforms are regulated by PR, T47Dco cells were treated with or without R5020 for 8 -60 h. Protein blots of whole cell lysates were probed with antibodies that recognize epitopes unique to Stat5a or Stat5b, as well as with an antibody that recognizes both isoforms ( Fig. 2A). This study shows that both the longer (95 kDa) Stat5a and the shorter (92 kDa) Stat5b isoform are up-regulated by progestin treatment. Stat5a was below detectable levels in the absence of R5020 ( Fig. 2A). Note that a nonspecific (NS) protein migrating just above the 95-kDa Stat5a is recognized by Stat5a antibody, but is not regulated by R5020 and serves as a loading control. Similar studies using anti-Stat3 (Fig. 2B) and anti-Stat1 (Fig. 2C) antibodies show that these STAT family members are also progestin regulated. While Stat3 is clearly up-regulated by progestin treatment, Stat1 is slightly down-regulated. The same blot, after probing with anti-Stat1 antibody, was stripped and reprobed with antibody recognizing total Stat5, which showed that Stat5 was again strongly up-regulated as in Figs. 1 and 2A (data not shown).
PR Up-regulates Stat5a mRNA Levels-To determine whether progesterone, via PR, up-regulates Stat5a mRNA levels, total RNA was isolated from two T47Dco breast cancer cell lines stably expressing either the PR-A or PR-B isoform (28) that had been treated with progesterone or vehicle for 12 or 24 h (Fig. 3). Northern blot analysis demonstrates hormonedependent induction of Stat5a message at 12 and 24 h following progestin treatment (Fig. 3). Interestingly, Stat5a appears to be more strongly induced by the PR-B than the PR-A isoform (compare lanes 2 and 4; 6 and 8). In contrast, the same Northern blot hybridized with cDNA for two genes known to be PR regulated, fatty acid synthetase (30,38) or cyclin D1 (18, 39), shows equal induction of these genes by the two PR isoforms. The same Northern blot hybridized with cDNA for a nonregulated gene, glyceraldehyde-3-phosphate dehydrogenase, serves as an RNA loading control.
Progestin Treatment Promotes Translocation of Stat5 into the Nucleus-Upon activation by growth factor receptors via their intrinsic tyrosine kinase activity or by cytokines via members of the JAK tyrosine kinase family, STATs are phosphorylated and translocate into the nucleus. In addition to the well characterized activation of Stat5 by prolactin (9), EGF has been shown to activate STATs in breast cancer cells (13). Since progestins sensitize breast cancer cells to the effects of EGF (Ref. 18; Lange et al. (20)) in the accompanying article and Fig.  8 herein we asked whether they influence the ability of EGF to activate and translocate Stat5 to the nucleus (Fig. 4A). T47Dco cells were treated for 24 h with 10 nM R5020 or vehicle, then with 10 nM EGF or vehicle for 5 min. Nuclear and cytosolic extracts were prepared and equal amounts of protein were probed with antibody recognizing total Stat5 (Fig. 4A). In untreated cells, Stat5 was exclusively cytoplasmic (lane 1); no Stat5 was found in the nucleus (lane 2). As expected, EGF treatment promoted nuclear translocation of a portion of Stat5 (lanes 5 and 6). Surprisingly, R5020 not only increased total Stat5 levels (lanes 3 and 4; 7 and 8), but also promoted extensive nuclear translocation of Stat5 (compare lanes 2 and 4) independent of EGF treatment. Depending on the stringency with which nuclei are prepared, varying amounts of Stat5 are detected in the nuclear fraction in the untreated cells; however, the amount of Stat5 is consistently an average of 1.8- fold   FIG. 2. Stat5a and 5b, as well as Stat3 and Stat1, are regulated by progestin treatment of breast cancer cells. Stat5a and 5b isoforms differ primarily in the COOH terminus, where the last 8 amino acids of the two isoforms diverge. Stat5a is 7 amino acids longer than Stat5b and contains an additional activation domain (AD). DNA-binding domain (DBD); Src homology domains, SH2 and SH3; consensus tyrosine phosphorylation site (Y). A, PR-positive, T47Dco breast cancer cells were treated with 10 nM R5020 (ϩ) or ethanol vehicle (Ϫ) for 8 -60 h. Cells were harvested at the time points indicated and equal amounts of whole cell lysates (100 g) were resolved by SDS-PAGE and immunoblotted with antibodies specific to the 95-kDa Stat5a or the 90 -92 kDa Stat5b, or with an antibody that detects both isoforms (total Stat5). B, the same extracts as in A were analyzed with an antibody specific to Stat3 (89 kDa). C, a similar time course study was repeated and the immunoblot was probed with an antibody specific to Stat1 (91 kDa). was hybridized with 32 P-labeled cDNA encoding fatty acid synthetase (FAS) and cyclin D1. The same Northern blot was stripped and hybridized with a 32 P-labeled Stat5a cDNA probe. Last, a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probe representing a non-regulated gene served as a control for RNA loading.
higher in the nuclear fraction of progesterone-treated cells. These results suggest that Stat5 signaling molecules, previously thought to be activated only by association with ligandactivated cell surface receptors, can be translocated to the nucleus by steroid hormone treatment.
Interaction of Stat5 and PR-To explain the surprising effects of progestins on Stat5 nuclear translocation, we speculated that Stat5 might interact with PR. This would not be unreasonable since an interaction between GR and Stat5 has recently been described (26). To determine whether PR and Stat5 can interact, co-immunoprecipitation experiments were performed using wild type HeLa cells or HeLa cells stably transfected with the Flag-tagged B-isoform of PR (PR B :f HeLa) (Fig. 5A). Both PR B :f HeLa (lane 1) and wild type HeLa (lane 2) were treated with 10 nM R5020 for 1 h and nuclear extracts were immunoprecipitated with either anti-total Stat5 antibody followed by protein A-Sepharose (Fig. 5A, top) or Anti-Flag M2 Affinity Gel (Fig. 5A, bottom), then eluted and immunoblotted. The anti-Stat5 immunoprecipitates were first probed with the anti-PR antibody AB-52, followed by the anti-Stat5 antibody. The eluates obtained by competing PR and associated proteins off the Anti-Flag M2 Affinity Gel with the Flag peptide, were first probed with the anti-Stat5 antibody, followed by the anti-PR antibody.  lanes 1 and 2). These reciprocal studies show that PR and Stat5 are associated, either directly or in a multiprotein complex. In Fig. 5B, Stat5 was co-immunoprecipitated with PR from whole cell extracts of PR B :f HeLa cells treated with either ethanol (Ϫ) or R5020 (ϩ) for 48 h. Untreated cells contain much higher levels of PR than R5020-treated cells, which exhibit ligand-dependent PR down-regulation (40,41). When probed with anti-Stat5 antibody, the amount of Stat5 protein corresponds to the amount of PR, suggesting that stoichiometric amounts of Stat5 co-immunoprecipitate with PR.
Progestins Promote Co-immunoprecipitation of Stat5 and JAK2 with Anti-phosphotyrosine Antibody-Stat5 can be activated directly by the tyrosine kinase activity of growth factor receptors or indirectly by JAK2 (8). In both cases, activation involves tyrosine phosphorylation of Stat5. To determine the tyrosine phosphorylation state of Stat5 and JAK2 following treatment with R5020, T47Dco cells were transiently (1 h) and chronically (48 h) treated with R5020. Whole cell lysates were immunoprecipitated with an anti-phosphotyrosine antibody (4G10) then immunoblotted with specific antisera to Stat5 or JAK2 (Fig. 6). In the absence of R5020 pretreatment, JAK2 is immunoprecipitated with the anti-phosphotyrosine antibody only following EGF treatment (lane 2). Note that JAK2 is not present in phosphotyrosine immunoprecipitates after 1 h of R5020 treatment (lane 1). However, JAK2 is immunoprecipitated with the anti-phosphotyrosine antibody as efficiently after 48 h of R5020 treatment (lane 4) as with EGF treatment (lane 5). Although less dramatic, the same results described for JAK2 were also observed with Stat5. Co-immunoprecipitation of Stat5 and JAK2 with anti-phosphotyrosine antibody following long-term R5020 treatment was also observed in independent experiments in Fig. 3 of the accompanying article (20). These results indicate that progestin treatment can influence the JAK2/Stat5 pathway and that transient progestin treatment (1 h) has different biological effects than chronic progestin treatment (48 h). Although total Stat5 levels are up-regulated by progestin treatment, JAK2 levels are unaffected (data not shown).
To confirm that progestin influences the immunoprecipitation of Stat5 with anti-phosphotyrosine 4G10 antibody, duplicate plates of T47Dco cells were treated with ethanol or R5020 for 48 h followed by 5 min of EGF treatment. Whole cell lysates were immunoprecipitated with anti-phosphotyrosine antibody (4G10) then immunoblotted with antibody to total Stat5. Fig.  6B demonstrates: 1) that more Stat5 is immunoprecipitated by the anti-phosphotyrosine antibody after progestin pretreatment, and 2) that EGF stimulates a Stat5 upshift only following progestin treatment. We therefore asked whether cytokineor growth factor-mediated activation of Stat5 could be influenced by progestins.
Prolactin Induced Activation of Stat5 Is Progestin Dependent in T47Dco Breast Cancer Cells-Prolactin is a key cytokine influencing breast physiology that signals through the JAK2/ Stat5 pathway. To determine whether progestin treatment affects the ability of prolactin to activate Stat5, T47Dco cells were first treated with R5020 or vehicle for 48 h, then treated transiently with R5020 (1 h), ethanol vehicle (1 h), or prolactin (10 min). Whole cell lysates were immunoblotted first with a phospho-Stat5 specific antibody, and then with antibody to total Stat5 (Fig. 7A). Stat5 was present in the cell lysates under all treatment conditions (total Stat5, lanes 1-6) and was up-regulated by the progestin treatment (total Stat5, lanes 4 -6). However, Stat5 remained unphosphorylated (P-Stat5, lanes [1][2][3][4][5], except in the cells that were pretreated with R5020 for 48 h, prior to the 10-min prolactin treatment (P-Stat5, lane 6). Prolactin alone had no effect (lane 3) nor did transient (1 h) or chronic (48 h) R5020 treatment (lane 2). Thus, in T47Dco cells, Stat5 phosphorylation by prolactin requires progesterone pretreatment. FIG. 4. Both EGF and R5020 induce nuclear translocation of Stat5. PRpositive T47Dco cells were treated for 24 h with either 10 nM R5020 (ϩ) or ethanol vehicle (Ϫ) followed by 10 nM EGF (ϩ) or sterile water (Ϫ) for 5 min. Nuclear and cytosolic fractions were prepared and samples (100 g of each) were resolved by SDS-PAGE then immunoblotted with antibody to total Stat5 as indicated. Protein levels were quantitated by Bradford assay and Ponceau S staining (not shown), and by immunoblotting with an antibody against cdc2 kinase/PSTAIR.
The ␤-casein promoter, known to be induced by Stat5 when activated by prolactin via prolactin receptors and JAK2, was next used to test the function of the phosphorylated Stat5. T47D-YB cells were pretreated with either R5020 or ethanol vehicle for 48 h. Cells were then washed, transfected with a ␤-casein promoter-luciferase reporter, treated with either R5020 or prolactin, and harvested for luciferase assay 24 h after hormone treatment. We found that in T47D-YB breast cancer cells, prolactin alone has minimal to no significant effect on the ␤-casein luciferase promoter. However, after R5020 pretreatment, a 6 -7-fold induction of the ␤-casein luciferase promoter was observed (Fig. 7B). Thus R5020 sensitized these breast cancer cells to the transcriptional effects of prolactin mediated through Stat5.
Combined Progestins and EGF Have a Synergistic Effect on Induction of p21 WAF1 and c-fos Promoters-To determine whether growth factor signaling is also influenced by progesterone, we examined the transcriptional activity from the p21 WAF1 and c-fos promoters. Both genes are induced by progestins and EGF and both promoters contain STAT-binding sites termed sis-inducible elements. The cell cycle regulatory gene, p21 WAF1 is induced by progestins (18,34) and by EGF/STAT pathway (13,15). The mRNA of the immediate early gene c-fos is rapidly induced by progestins (17) and by EGF/STAT (16). PR-positive T47D-YB cells were transfected with either the Ϫ2320 p21-luc construct or the Ϫ357/Ϫ276 c-fos-81-TK-luc construct. The Ϫ2320 p21-luc construct contains one of three sis-inducible elements at position Ϫ640. The Ϫ357/Ϫ276 c-fos-81-TK-luc construct contains a sis-inducible element at Ϫ342. Following transfection, cells were treated with EGF or R5020 alone, or simultaneously with both hormones. Fig. 8A shows that the p21 WAF1 promoter is induced 4.5-fold by progesterone, 2.4-fold by EGF, and 13.2-fold by both hormones together. The average increase in five experiments was progesterone, 3.5-fold; EGF, 2.2-fold; both hormones, 11-fold. Fig. 8B demonstrates the corresponding increases in endogenous p21 levels when T47D-YB cells are treated for 24 h with ethanol, progesterone, EGF, or both hormones. Whole cell lysates were probed by immunoblotting with antibody to p21. Fig. 8C shows that the c-fos promoter is induced 19.3-fold by R5020 and 6.2-fold by EGF. Remarkably, together the two were treated with 10 nM R5020 for 1 h and nuclear extracts were prepared. Stat5 or PR-B receptors were immunoprecipitated from 1 mg of nuclear extract protein with either anti-total Stat5 antibody followed by protein A-Sepharose (top), or with Anti-Flag M2 Affinity Gel (bottom). Immunoprecipitates were either released from the protein A-Sepharose by boiling in Laemmli sample buffer, or eluted from the Anti-Flag M2 Affinity Gel with Flag peptide followed by boiling in Laemmli sample buffer, then resolved by SDS-PAGE and transferred to nitrocellulose. Anti-Stat5 immunoprecipitates were probed first with anti-PR antibody AB52, followed by anti-total Stat5 antibody. Eluates released from the Anti-Flag M2 Affinity Gel with the Flag peptide were probed first with anti-Stat5 antibody, then with anti-PR antibody, AB-52. White arrows indicate the position of the 97.4-kDa marker. 1, PR B :f HeLa; 2, wild type HeLa cells. B, PR B :f HeLa cells were treated with either ethanol (Ϫ) or R5020 (ϩ) for 48 h, PR were immunoprecipitated from whole cell lysates using the Anti-Flag M2 Affinity Gel, released with the Flag peptide, and coprecipitated proteins were resolved by SDS-PAGE and immunoblotted with AB-52 and Stat5 antibodies.
FIG. 6. Co-immunoprecipitation of JAK 2 and Stat5 with antiphosphotyrosine antibody. A, PR-positive T47Dco cells were pretreated for 48 h with 10 nM R5020 or ethanol vehicle, then subjected to short-term R5020 (1 h) or 10 nM EGF (5 min) treatments. Equal amounts (1 mg) of whole cell lysates were immunoprecipitated using anti-phosphotyrosine antisera (4G10). Immunoprecipitates were resolved by SDS-PAGE and immunoblotted with antibodies specific to JAK2 and total Stat5. B, duplicate plates of T47Dco cells were pretreated for 48 h with 10 nM R5020 or ethanol vehicle, then stimulated with vehicle or 10 nM EGF for 5 min. Equal amounts (1 mg) of whole cell lysates were immunoprecipitated using anti-phosphotyrosine antisera (4G10). Immunoprecipitates were resolved by SDS-PAGE and immunoblotted with antibody recognizing total Stat5. A nonspecific band is shown demonstrating equal protein loading.
hormones produced a 145-fold increase in luciferase activity. The control minimal thymidine kinase promoter linked to the luciferase reporter had 1% the activity of the c-fos containing construct, and no synergy was observed with both hormones. The experiment in Fig. 8C is also representative of five repeats in which the average fold increases were as follows: progesterone, 18-fold; EGF, 5.3-fold; and both hormones, 97-fold. The same pattern of induction of both the p21 and c-fos promoters is observed using T47Dco cells with endogenous PR, rather than stably transfected PR. Since the Ϫ2320 p21 promoter and the Ϫ357/Ϫ276 c-fos promoter lack a concensus progesterone response element (PRE), we speculate that the effect of progesterone and the synergistic effect of progesterone plus EGF are mediated through the sis-inducible element, possibly via PR interactions with STATs.
It is also possible that the synergistic effects of progesterone plus EGF are due to enhanced STAT activation via tyrosine phosphorylation. To test this, T47D-YB cells were treated for 24 h with 10 nM progesterone or vehicle, followed by 10 nM EGF for 5 min. Equal amounts of whole cell lysates were resolved by SDS-PAGE, transferred to nitrocellulose, and probed with an antibody specific for Stat5 phosphorylated on tyrosine 694 (P-Stat5), followed by an antibody that recognizes total Stat5 (Fig.  9A). As with prolactin (Fig. 7A), EGF does not induce phosphorylation of Stat5 unless cells are pretreated with progestin. The levels of total Stat5 are up-regulated by long-term progestin treatment. Since PR B :f HeLa cells respond to EGF without progesterone pretreatment, we asked what effect short-term progesterone treatment would have on activation of Stat5 by EGF. PR B :f HeLa were treated with progesterone for 1 h, EGF for 5 min, or 1 h progesterone followed by 5 min of EGF. Nuclear and cytosolic extracts were resolved by SDS-PAGE, transferred to nitrocellulose, and probed first with an antibody specific for Stat5 phosphorylated on tyrosine 694 followed by an antibody that recognizes total Stat5 (Fig. 9B). While there is no remarkable difference in total Stat5 levels with short-term progesterone alone, EGF alone, or progesterone plus EGF, there is a 3.3-fold increase in the amount of phospho-Stat5 in the nucleus if EGF is preceded by progesterone. This may explain the resulting synergism of the two hormones on STATregulated genes. DISCUSSION Since both progesterone and STATs are key molecules in the growth and differentiation of the normal breast, we have studied their relationship in breast cancer. We show here, using breast cancer cells, that progestins regulate the levels of STAT proteins in a PR-dependent manner. Additionally, we document that Stat5 can interact with PR and that progestin treatment induces Stat5 translocation to the nucleus. Stat5 and JAK2 are present in phosphotyrosine immunoprecipitates following chronic progestin treatment. This surprising result implies that progestins may activate a kinase that mediates the tyrosine phosphorylation of JAK2, or that JAK2 and Stat5 associate with, and coimmunoprecipitate with, another phosphotyrosine-containing protein following long-term progestin treatment. We have not addressed the role of a potential progestin initiated autocrine loop; however, we do not observe constitutive activation of EGFR or mitogen-activated protein kinases by progestins alone, as shown in Fig. 3-5 of the accompanying article (20). These data led to the hypothesis that progestins sensitize breast cancer cells to signaling by cytokines and growth factors. Indeed, we show that progestin pretreatment is necessary for prolactin signaling to the ␤-casein promoter via Stat5. Additionally we observe transcriptional synergy between progesterone and EGF on the promoters of two growth regulatory genes containing STAT sites. Finally, we show that progesterone enhances the ability of EGF to activate Stat5 through phosphorylation of tyrosine 694. Together, these results suggest novel pathways for the integration of progesterone and cytokine/growth factor signaling.
PR-mediated Regulation of Stat5-Knock-out studies have confirmed that Stat5 is essential for normal mammary gland development (6). Additionally, Stat5a-deficient mice are unable to lactate due to failure of the gland to differentiate appropriately during pregnancy (6). In fact, in wild-type animals, Stat5 mRNA and protein levels increase during pregnancy, peaking just prior to parturition, then declining during lactation (10,37). This rise and fall parallels changes in progesterone levels, suggesting a STAT-regulatory role for this pregnancy hormone. We now demonstrate the up-regulation of STAT mRNA and proteins by progestins in a PR-dependent manner in breast FIG. 7. Activation of Stat5 by prolactin and induction of the ␤-casein promoter are progestin dependent in T47Dco breast cancer cells. A, PR-positive T47Dco breast cancer cells were pretreated for 48 h with R5020 or ethanol vehicle, then subjected to short-term 10 nM R5020 (1 h) (ϩ), prolactin (10 min) (ϩ), or ethanol (Ϫ) treatments. Phosphorylated (P) Stat5 and total Stat5 were detected in whole cell extracts of T47Dco cells by immunoblotting first with an antibody specific for Stat5 phosphorylated on tyrosine 694, followed by an antibody that recognizes total Stat5. B, triplicate dishes of T47D-YB cells stably expressing the PR B-isoform, were pretreated for 48 h with 10 nM R5020 or ethanol vehicle, then transfected with the Ϫ2300/ϩ490 ␤-casein promoter fused to a luciferase reporter and treated with either R5020 or prolactin for an additional 24 h. Luciferase units are shown on the y axis and standard errors are indicated. f, R5020; u, prolactin. cancer cells (Figs. 1-3). The PR-B isoform appears to be more potent in this regard (Fig. 3). In contrast, two other PR-regulated genes, cyclin D1 and fatty acid synthetase, are equally up-regulated by both PR isoforms. The PR-B isoform is a stronger transcriptional activator than PR-A on some transiently expressed promoters in in vitro models and of the flavin-containing monooxygenase 5 gene in breast cancer cells (42). Whether STAT promoters contain a progesterone response element remains to be determined. It is also not known whether STATs play a role in the progression of breast cancers from steroid hormone responsiveness to steroid hormone resistance and growth factor dependence. High levels of STATs and increased STAT activity have been demonstrated in invasive breast cancers (43). The significance of this finding is unclear, and it will be important to determine whether STAT expression correlates with PR positivity in such tumors.
Progestin Regulation of Stat5 Localization through PR Binding?-Cytoplasmic Stat5a and 5b are substrates for tyrosine phosphorylation by growth factor tyrosine kinases such as EGFR, or by cytoplasmic protein tyrosine kinases of the JAK family activated by cytokines such as PRL. Phosphorylation of the conserved COOH-terminal tyrosine residue on Stat5 (tyrosine 694) leads to homo-or heterodimerization followed by nuclear translocation through unknown mechanisms (8,12). Recently, however, Ali and Ali (44), showed that Stat5 tyrosine phosphorylation can be dissociated from nuclear translocation. In addition, Stat1, mutated at the critical tyrosine phosphorylation site, retains activity and may act in its monomeric state as a non-DNA binding transcriptional coactivator (45). Since STATs lack concensus nuclear localization signals, speculation regarding translocation mechanisms has recently shifted to the involvement of cellular partners. For example, a functional NLS, located at the COOH terminus of interferon-␥, is postulated to complex with, and chaperone STATs to the nucleus (46).
The ␤-casein promoter is synergistically induced by glucocorticoids and prolactin. Stat5-dependent transcription from the ␤-casein gene promoter is enhanced by protein complexes formed between Stat5, bound to its consensus DNA-binding site, and GR (26). Since the ␤-casein promoter lacks a classical palindromic glucocorticoid response element, it has been suggested that glucocorticoid-dependent transcriptional synergy with prolactin occurs through GR binding to glucocorticoid response element half-sites, facilitated by the binding of Stat5 to GR (47). Glucocorticoid-dependent transcription from the mouse mammary tumor virus promoter may be similarly enhanced by Stat3 interactions with GR at a palindromic glucocorticoid response element (27). Based on the observation that Stat5 and GR can physically associate (26), and that nuclear translocation of Stat5 is enhanced by glucocorticoid treatment, Cella et al. (48) raise the possibility that Stat5 can be recruited to the nucleus by liganded GR. Similarly, we show here that independent of cytokine signaling, progestins alone induce the nuclear translocation of Stat5 (Fig. 4). This, coupled with the demonstration of an interaction between Stat5 and PR in nuclear fractions following 1 h of progestin treatment (Fig.  5), suggests that unactivated Stat5 can undergo nuclear translocation, co-transported by PR molecules shuttling between the cytoplasm and nucleus (49,50). This raises the interesting possibility that STATs, perhaps as monomers, can enter the nucleus when tethered to other liganded transcription factors that contain an endogenous nuclear localization signal, where they can serve as coregulators of the same DNA-bound factors.
Cross-talk between Cytoplasmic Receptor Tyrosine Kinases and Nuclear Steroid Receptors-Despite the fact that growth factors and cytokines function at the cell surface and steroid hormones function in the nucleus, triggering completely different signal transduction pathways, the resulting signals often converge on the same genes. Many examples of genes transcriptionally regulated by a complex set of signaling pathways involving steroid hormones, growth factors, and cytokines have been described, including the genes for ␤-casein, c-fos, and p21 WAF1 . Similar to the transcriptional synergy observed on the ␤-casein promoter in the presence of prolactin and glucocorticoids (26,47,48), we now describe synergy between EGF and progesterone in regulating transcription of c-fos, and p21 WAF1 promoters (Fig. 8). How do the growth factor and steroid hormone signaling pathways converge?
One possible site is in the cytoplasm or cell membrane. We describe here, and in the accompanying article (20), that progesterone, in a PR-dependent manner, can selectively increase the sensitivity of key tyrosine kinase signaling pathways to the actions of growth factors, by up-regulating receptor tyrosine kinases in the EGFR family, and by enhancing EGF-mediated activation of signaling intermediates.
However, another site appears to be in the nucleus, at the promoter in question. While the promoters for the ␤-casein, p21 and c-fos genes contain one or more STAT-binding sites, as well as binding sites for other transcription factors, none contains a concensus palindromic PRE. It is possible that regulatory effects of progestins on these promoters are mediated by degenerate half-sites, such as those found on the ␤-casein promoter (47). It is intriguing in this regard, that the mouse mammary tumor virus, the only promoter on which progesterone plus EGF synergy has been previously reported (51,52), also has multiple PRE half-sites as well as concensus palindromic PREs and Stat5-binding sites. Alternatively, transcriptional cooperativity between STATs and PR may occur without direct PR binding to DNA. We recently reported that transcriptional regulation of the p21 WAF1 promoter by progesterone involves recruitment of PR to the proximal promoter region, where the receptors bind to Sp1 and CBP/p300, rather than directly to DNA (34). Interestingly, maximal activation of the ICAM promoter requires a physical interaction between Sp1 and Stat1 in which Sp1 is thought to recruit Stat1 to the promoter and/or serves to link Stat1 to the basal transcription complex (53). Taken together these data suggest mechanisms for cross-talk between STATs and PR in which one or the other transcription factor is tethered to DNA indirectly.
Finally, we show here that progesterone pretreatment enhances the ability of EGF to activate Stat5 through phosphorylation of tyrosine 694. This raises the possibility that progestin-induced synergism does not require the binding of PR to the promoter. Instead, progestin treatment may enhance the amount of activated Stat5 capable of binding to the promoter.
In summary, our findings that progestins regulate STATs, foster protein-protein interactions between PR and Stat5, promote translocation of Stat5 to the nucleus, enhance the ability of prolactin and EGF to induce phosphorylation of Stat5, and synergize with EGF to activate transcription of growth regulatory genes, demonstrate that steroid and growth factor/cytokine signaling pathways can converge at multiple levels.