Abstract
The multidrug transporter, breast cancer resistance protein, ABCG2, is up-regulated in certain chemoresistant cancer cells and in the mammary gland during lactation. We investigated the role of the lactogenic hormone prolactin (PRL) in the regulation of ABCG2. PRL dose-dependently induced ABCG2 expression in T-47D human breast cancer cells. This induction was significantly reduced by short-interfering RNA–mediated knockdown of Janus kinase 2 (JAK2). Knockdown or pharmacologic inhibition of the down-stream signal transducer and activator of transcription-5 (STAT5) also blunted the induction of ABCG2 by PRL, suggesting a role for the JAK2/STAT5 pathway in PRL-induced ABCG2 expression. Corroborating these findings, we observed PRL-stimulated STAT5 recruitment to a region containing a putative γ-interferon activation sequence (GAS) element at −434 base pairs upstream of the ABCG2 transcription start site. Introduction of a single mutation to the −434 GAS element significantly attenuated PRL-stimulated activity of a luciferase reporter driven by the ABCG2 gene promoter and 5′-flanking region containing the −434 GAS motif. In addition, this GAS element showed strong copy number dependency in its response to PRL treatment. Interestingly, inhibitors against the mitogen-activated protein kinase and phosphoinositide-3-kinase signaling pathways significantly decreased the induction of ABCG2 by PRL without altering STAT5 recruitment to the GAS element. We conclude that the JAK2/STAT5 pathway is required but not sufficient for the induction of ABCG2 by PRL.
Introduction
The breast cancer resistance protein ABCG2 is the second member of the ABCG subfamily within ATP-binding cassette–transporter superfamily. Classically associated with multidrug resistance in certain cancer cells, ABCG2 is now known to be expressed in normal epithelial cells from such tissues as placenta, intestine, liver, and kidney (Maliepaard et al., 2001; Fetsch et al., 2006). Since ABCG2 is predominately localized to the apical side of biologic barriers (e.g., the intestine and the blood–brain barrier), it is considered a protective efflux transporter. For example, Abcg2/Bcrp1-null mice exhibit increased systemic exposure to dietary carcinogens, such as aflatoxin B1 (van Herwaarden et al., 2006); phototoxins, such as pheophorbide a (Jonker et al., 2002); drugs, such as topotecan and phytoestrogens, such as daidzein (Enokizono et al., 2007) compared with wild-type mice. These mice also demonstrate higher fetal-to-maternal plasma concentration ratio for genistein and topotecan (Enokizono et al., 2007).
ABCG2 is dramatically up-regulated in the mammary gland during lactation, where it contributes to not only the accumulation of riboflavin (vitamin B2) and other vitamins in milk but also, for reasons unknown, the excretion of drugs and toxins into breast milk (Jonker et al., 2005). However, the mechanism of the massive upward surge in mammary ABCG2 expression during lactation is poorly understood. Given the roles of ABCG2 in both therapeutic failure of cancer treatment and host protection from toxins, mechanistic understanding of this lactation-related phenomenon may inform a novel therapeutic strategy for ABCG2-mediated drug resistance in cancer and provide insight into an approach for facilitated detoxification.
A wide range of modulators of ABCG2 expression have been identified, which include but are not limited to hormones (Ee et al., 2004; Wang et al., 2008a), cytokines (Evseenko et al., 2007; Vee et al., 2009), xenobiotics (Jigorel et al., 2006), and epigenetic factors (To et al., 2006; Turner et al., 2006). Among these, several have been identified in the 5′ flanking region of the human ABCG2 gene: the DNA-binding elements of the estrogen receptor (Ee et al., 2004), hypoxia inducible factor 1-α (Krishnamurthy et al., 2004), progesterone receptor (Wang et al., 2008a), aryl hydrocarbon receptor (Tan et al., 2010), peroxisome proliferator-activated receptor γ (Szatmari et al., 2006), NF-E2-related factor 2 (Singh et al., 2010), and nuclear factor-κB (Pradhan et al., 2010). However, none of these transcription factors has been implicated in the up-regulation of mammary gland ABCG2 during lactation, implying an existence of a novel regulatory pathway.
Lactogenesis requires the interplay between many different hormones, but of primary importance is prolactin (PRL), which is essential for both mammary gland development and lactation (Neville et al., 2002). Binding of PRL to the cell-surface prolactin receptor (PRLR) triggers an intracellular signaling cascade that involves Janus kinase 2 (JAK2) autophosphorylation and activation (Chilton and Hewetson, 2005; Clevenger et al., 2009). Once activated, JAK2 phosphorylates tyrosine residues on the PRLR that serve as docking sites for other proteins, such as signal transducer and activator of transcription-5 (STAT5). Upon recruitment to the PRLR, STAT5 is phosphorylated by JAK2, after which it dimerizes and enters the nucleus where it binds DNA motifs called γ-interferon activated sequence (GAS) elements with the consensus sequence 5′-TTCNNNGAA-3′ (Chilton and Hewetson, 2005). Examples of genes regulated in this manner include the cytokine-inducible SH2–containing protein (CISH) (Verdier et al., 1998), and milk proteins β-casein (CSN2) (Gouilleux et al., 1994) and whey acidic protein (WAP) (Li and Rosen, 1995). In addition to JAK2/STAT5, PRL also activates mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K) signaling (Clevenger et al., 2003). However, how these pathways modulate PRL-induced gene expression is not well characterized.
Given the lactation-specific up-regulation of ABCG2 in the mammary gland, which corresponds to the PRL surge during lactation, we hypothesized that the PRL signaling pathways are involved in the regulation of ABCG2. In this study, we show that PRL induces ABCG2 expression by activating not only JAK2/STAT5 but also MAPK and PI3K pathways.
Materials and Methods
Reagents.
Human ductal breast epithelial tumor cells (T-47D), nontumorigenic human mammary epithelial cells (MCF-10A), and human breast adenocarcinoma cells (MCF-7) were purchased from the American Type Culture Collection (Manassas, VA). Clonetics primary human mammary epithelial cells (HMEC) were purchased from Lonza (Walkersville, MD). Fetal bovine serum (FBS), recombinant human insulin, RPMI-1640, Dulbecco’s modified Eagle’s medium (DMEM), and phenol-red free DMEM/F12 were from Wiscent Inc. (Montreal, QC). DMEM/F12, phenol red–free high-glucose DMEM, sodium pyruvate, recombinant human epidermal growth factor (EGF), horse serum, and GlutaMAX were from Gibco (Invitrogen; Burlington, ON). Actinomycin D and cholera toxin were purchased from Sigma-Aldrich Canada, Ltd. (Oakville, ON). STAT5 inhibitor [N′-([4-oxo-4H-chromen-3-yl]methylene)nicotinohydrazide], PD98059, U0126, wortmannin, and LY294002 were from EMD Bioscience (San Diego, CA) and were dissolved to working concentrations in dimethylsulfoxide (DMSO). FugeneHD and pGL4.23(luc2/minP) were purchased from Promega (Madision, WI). Short-interfering RNAs (siRNAs) were purchased from Dharmacon (ThermoScientific; Lafayette, CO). Recombinant human PRL and recombinant growth hormone (GH) were obtained from Dr. A. F. Parlow at National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases’s National Hormone and Peptide Program, Harbor-UCLA Medical Center (Torrance, CA), stored as lyophilized powder, and dissolved in phosphate buffered solution (PBS) to working concentrations. All other chemicals were purchased from Sigma-Aldrich.
Cell Culture and Serum Starvation.
T-47D cells were maintained in RPMI-1640 supplemented with 10% FBS. For serum-starvation, T-47D cells were washed twice with phenol red–free RPMI-1640, and then incubated with serum-free and phenol red–free RPMI-1640 supplemented with 0.05% fatty acid–free bovine serum albumin and 0.01 mg/ml holo-transferrin. MCF-7 cells were grown in high-glucose DMEM containing sodium pyruvate supplemented with 10% FBS. MCF-7 cells were starved in phenol red–free DMEM supplemented with 1 mM sodium pyruvate, 4 mM GlutaMAX, 0.05% fatty acid–free bovine serum albumin, and 0.01 mg/ml holo-transferrin. MCF-10A cells were maintained in DMEM/F12 supplemented with 20 ng/ml EGF, 0.5 µg/ml hydrocortisone, 100 ng/ml cholera toxin, 10 µg/ml insulin, and 5% horse serum. For serum starvation, MCF-10A cells were washed twice and incubated with phenol red–free DMEM/F12. Prior to treatment, all cells were serum-starved overnight for 18 to 20 hours. Clonetics HMEC were grown as per instructions provided by the supplier (Lonza). In brief, HMEC were grown in mammary epithelial basal medium (MEBM) supplemented with 0.4% bovine pituitary extract, 10 ng/ml EGF, 5 µg/ml insulin, 0.5 µg/ml hydrocortisone, 15 ng/ml amphotericin B, and 30 µg/ml gentamicin and used in the second passage after thawing (approximately sixth division). HMEC were serum-starved in phenol red–free MEBM or phenol red–free MEBM supplemented with 5 µg/ml insulin and 0.5 µg/ml hydrocortisone. All cells were maintained in a humidified incubator at 37°C under a 5% CO2 atmosphere.
RNA Isolation and Real-Time Polymerase Chain Reaction.
RNA was isolated using RNeasy Kit (Qiagen; Toronto, ON), and quantified using a Nanodrop 2000 spectrophotometer (Wilmington, DE). RNA integrity was assessed by agarose gel electrophoresis. RNA was reverse transcribed to cDNA using Moloney murine leukemia virus-reverse transcriptase and random primers (Invitrogen). Real-time reverse transcriptase polymerase chain reaction (RT-PCR) was performed using inventoried Taqman primers from Applied Biosystems (Invitrogen) and the Applied Biosystems 7500 Real-Time PCR system. Relative mRNA expression was quantified using the 2-∆∆CT method, normalizing to glyceraldehyde-3-phosphate dehydrogenase mRNA expression.
Preparation of Crude Membrane Fraction and Whole Cell Lysate.
Crude membrane fractions were prepared as described previously (Wang et al., 2003) with modifications. Each T225 flask of T-47D cells was washed twice with ice-cold PBS and cells detached with a rubber policeman in ice-cold PBS containing 1 mM phenylmethylsulfonyl fluoride (PMSF). The cell suspension was pelleted by centrifugation at 150 g for 5 minutes at 4°C, the cell pellet was resuspended in hypotonic lysis buffer (10 mM KCl, 1.5 mM MgCl2, 10 mM Tris-HCl, 1 mM EDTA, 1 mM PMSF, 1× protease inhibitor cocktail, pH 7.4) and incubated on ice for 10 minutes. Cells were homogenized with a Dounce homogenizer, and removal of plasma membrane was confirmed by visual inspection under a microscope. The homogenate was centrifuged at 1500 g for 10 minutes at 4°C, and the resulting supernatant was collected and centrifuged at 100,000 g for 60 minutes at 4°C. The crude membrane pellet was resuspended in 200 µL resuspension buffer (10 mM Tris-HCl, 0.25 M sucrose, 150 mM NaCl, 1 mM PMSF, 1× protease inhibitor cocktail, pH 7.4), passed through a 26-gauge needle 10 times, and aliquoted. Aliquots were stored at −80°C. For preparation of whole cell lysate from T-47D, MCF-7, MCF-10A, and HMEC, cells were washed twice with ice-cold PBS (containing 1 mM Na3VO4 and 2.5 mM NaF when lysate was to be used for detection of phosphorylated proteins) and lysed with radioimmunoprecipitation assay buffer (50 mM Tris-HCl, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) containing 2 mM Na3VO4, 5 mM NaF, 2 mM EDTA, 1 mM PMSF, and 1× protease inhibitor cocktail. Cells were scraped with a rubber policeman, transferred to a microcentrifuge tube, and rotated end-over-end at 4°C for 20 minutes. Cellular debris was pelleted by centrifugation at 10,000 rpm for 10 minutes at 4°C. The supernatant was collected, aliquoted, and stored at −80°C. Protein concentration was measured using the Bradford assay.
Immunodetection of Protein by Gel Electrophoresis/Western Blot.
Crude membrane preparations (30 µg) and whole cell lysates (10 µg) were resolved in 4–12% 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol gels and transferred to nitrocellulose membrane using the Novex NuPAGE SDS-PAGE system (Invitrogen). Membranes were blocked overnight at 4°C prior to incubation with primary antibody. In general, blots were incubated with primary antibody overnight at 4°C and then with horseradish peroxidase conjugated secondary antibody for 1 hour at room temperature. Bands were visualized by enhanced chemiluminescence (GE Healthcare Life Sciences; Piscataway, NJ) followed by film exposure. Blots were stripped with Restore Western blot stripping buffer (ThermoScientific) at room temperature for 20 minutes and re-blocked overnight before incubating with a different primary antibody. Antibodies used were as follows: breast cancer resistance protein (1:200 BXP-21, Millipore; Billerica, MA), β-actin (1:20,000, Sigma-Aldrich), PRLR (1:500, 35-9200, clone 1A2B1, Sigma-Aldrich), JAK2 (1:1000, D2E12, Cell Signaling Technology; Beverly, MA), phospho-STAT5 (1:4000, 71-6900, Invitrogen) STAT5A (1:2000, sc-1081, Santa Cruz Biotech; Santa Cruz, CA), STAT5B (1:1000, sc-1656, Santa Cruz Biotech), phospho-ERK1/2 (1:1000, 197G2, Cell Signaling Technology), total ERK1/2 (1:2000, L34F12, Cell Signaling Technology), phospho-AKT (1:4000, C31E5E, Cell Signaling Technology), and total (pan) AKT (1:10,000, C67E7, Cell Signaling Technology).
Short-Interfering RNA.
T-47D cells were seeded at 500,000 cells per well on 6 well plates, grown for 24 hours, and transfected with siRNA using Dharmafect 1 as per manufacturer protocol (Dharmacon). For simultaneous knockdown of both isoforms STAT5A and STAT5B, cells were transfected with 25 nM custom siRNA duplex sense 5′-CUACAGUCCUGGUGUGAGAUU-3′ and antisense 5′-UCUCACACCAGGACUGUAGUU-3′ used previously by others (Gutzman et al., 2007). A nontargeting siRNA was used as a control (siGENOME nontargeting RNA #3, Dharmacon). For knockdown of JAK2, cells were transfected with 20 nM each of two predesigned siRNA (J-003146-12 and J-003146-13) against JAK2, and for control, 20 nM each of nontarget siRNA 1 and siRNA 2 (ON-TARGETplus siRNA, Dharmacon). Twenty-four hours after transfection, cells were incubated with fresh growth medium, grown for an additional 24 hours, then serum- starved overnight. Cells were either lysed to obtain whole cell lysate for assessing knockdown of protein expression by Western blot, or treated with PBS (0.1% v/v) or PRL for 6 hours to assess mRNA expression.
Small Molecule Inhibitors of STAT5, MAPK, and PI3K Signaling.
T-47D cells were serum-starved overnight and pretreated with STAT5 inhibitor (200 µM), MEK inhibitors PD98059 (20 µM) and U0126 (10 µM), or PI3K inhibitors LY294002 (10 µM) and Wortmannin (25 nM), and appropriate concentration of DMSO vehicle control for 1 hour. After 1 hour, cells were treated with PBS (0.1% v/v) or 100 ng/ml PRL for 15 minutes to 6 hours.
Plasmid Constructs.
The −1285/+362 ABCG2 construct reportedly previous (Bailey-Dell et al., 2001), which contains the ABCG2 proximal GAS element, surrounding 5′-flanking region and promoter, inserted into the pGL3-basic construct (Promega) is referred to here as pGL3-ABCG2. Fragments of DNA containing the ABCG2 distal GAS element were prepared by either annealing chemically synthesized oligonucleotides containing MluI sticky ends and corresponding to −4476/−4442ABCG2 (forward: 5′-CGCGTCCAAATGAATATTTTCTCTGAACCCTAAGATAGCCA-3′ and reverse: 5′-CGCGTGGCTATCTTAGGGTTCAGAGAAAATATTCATTTGGA-3′, GAS element underlined), or by PCR amplification of the −4565/−4414ABCG2 region using human genomic DNA and primers that contain MluI restriction sites (forward primer 5′-TGATACTAACGCGTTAAGGGACCTGACACTACCAATAAC-3′; reverse primer 5′-TGATACTAACGCGTGGAAGGTGAGAGAAAGGAATATAGC-3′). Annealed oligonucleotides and MluI digested PCR product was subsequently cloned into the pGL3-ABCG2 construct. To introduce a single mutation into the proximal GAS element of the pGL3-ABCG2 construct, site directed mutagenesis was performed using Pfu Turbo as per manufacturer instructions (Stratagene/Agilent Technologies; Santa Clara, CA). Primers used to generate the ABCG2/GASmut construct were as follows: forward 5′-CATCCACTTTCTCAGCATCCCATTCACCAG-3′; reverse 5′-CTGGTGAATGGGATGCTGAGAAAGTGGATG-3′; with site of mutation underlined. Successful mutagenesis was confirmed by sequencing, and the mutated insert was released by digestion with MluI/BglII and re-cloned back into the pGL3-basic vector. The pGL4-CISH construct contains the −1304 to +1 region of the human CISH gene cloned into the pGL4.10 vector (Fang et al., 2008). Tandem GAS constructs were cloned using chemically synthesized oligonucleotides that contain variable numbers of copies of the putative proximal GAS element (−448 to −422 regions) of the human ABCG2 gene. For GAS1v2.1, annealed oligonucleotides with sequence 5′-CAGCATCCACTTTCTCAGAATCCCATTCA-3′ and 5′-GATCTGAATGGGATTCTGAGAAAGTGGATGCTGGTAC-3′ were cloned into KpnI/BglII digested pGL4.23[Luc2/minP] construct (Promega). GAS3v2.5 was constructed by inserting HindIII digested annealed oligonucleotides with sequence 5′-CAGACAAGCTTAGCATCCACTTTCTCAGAATCCCATTCAGACAGCATCCACTTTCTCAGAATCCCATTCAGACAGCATCCACTTTCTCAGAATCCCATTCAAGCTTAGACA-3′ and 5′-GATCTGTCTAAGCTTGAATGGGATTCTGAGAAAGTGGATGCTGTCTGAATGGGATTCTGAGAAAGTGGATGCTGTCTGAATGGGATTCTGAGAAAGTGGATGCTAAGCTTGTCTGGTAC-3′ into pGL4.23. GAS6v2.4 was constructed by ligating undigested annealed oligonucleotides used to construct GAS3v2.5, which contain KpnI/BglII sticky ends, to GAS3v2.5. The same approach was used to construct GAS6v2mut.8 using annealed oligonucleotides 5′-CAGACAAGCTTAGCATCCACTGGCTCAGAATCCCATTCAGACAGCATCCACTGGCTCAGAATCCCATTCAGACAGCATCCACTGGCTCAGAATCCCATTCAAGCTTAGACA-3′and 5′-GATCTGTCTAAGCTTGAATGGGATTCTGAGCCAGTGGATGCTGTCTGAATGGGATTCTGAGCCAGTGGATGCTGTCTGAATGGGATTCTGAGCCAGTGGATGCTAAGCTTGTCTGGTAC-3′ (GAS sequence underlined). The pCMV-wtStat5a construct contains the rat STAT5A cloned into the pCMV-Tag 3B expression vector (Fang et al., 2009). Mouse STAT5A and STAT5B expression plasmids (pcDNA3-Stat5a and pcDNA3-Stat5b) were generous gifts from Dr. Dwayne L. Barber (Ontario Cancer Institute, Princess Margaret Hospital, Toronto, ON).
Plasmid constructs were sequenced (The Centre for Applied Genomics, SickKids, Toronto, ON) and prepared using HiSpeed Plasmid Midi and Maxi Kit (Qiagen) for transfection.
Transient Transfection and Luciferase Assay.
For experiments using firefly luciferase reporter constructs driven by the ABCG2 and CISH promoter, T-47D cells were seeded at 3 × 105 cells/well on 12-well plates and grown for 24 hours. Cells were then transfected in growth medium with 500 ng/well pGL3-basic, variations of the pGL3-ABCG2 constructs, or 75 ng/well pGL4.10 or pGL4-CISH using a 3:1 ratio Fugene HD:DNA (Promega). Twenty-four hours after transfection, cells were serum-starved overnight and treated with 100 ng to 500 ng/ml recombinant human PRL for 24 hours. For experiments using the tandem GAS constructs, T-47D cells were transfected as described above with slight modifications. T-47D cells were seeded at 1.25 × 105 cells on 24-well plates and transfected with 300 ng/well pGL4.23 or variations of this construct containing different copies of the proximal ABCG2 GAS element. To overexpress rat STAT5A, cells were co-transfected with 300 ng/well pCMV-wt Stat5a or its empty vector for control. At the end of treatment, cells were lysed with 100 µl (24-well plates) or 250 µl (12-well plates) phosphate lysis buffer, and 10 µl of lysate was used to measure luciferase activity using the Dual Luciferase Reporter Assay Kit from Promega on a Sirius luminometer (Berthold Detection Systems; Pforzheim, Germany) . Firefly luciferase reporter activity was normalized to renilla luciferase activity from a co-transfected pRL-TK plasmid (5 ng/well and 10 ng/well for 24- and 12-well plates, respectively) and presented as fold change over vehicle and empty vector control. All experiments were conducted in triplicate wells at least three times.
Chromatin Immunoprecipitation.
T-47D cells were grown to 80–90% confluence in 10-cm Petri dishes, serum-starved overnight for 18–20 hours, and treated with PBS (0.1% v/v), indicated concentration of PRL, or 200 ng/ml GH for up to 6 hours. For study using T-47D cells transfected with siSTAT5A/B or siRNA control, cells were transfected as per protocol for 6-well plates but scaled up to account for the larger surface area. For STAT5 inhibitor study, cells were pretreated with 200 µM STAT5 inhibitor or vehicle control (0.2% DMSO v/v) for 1 hour prior to PRL treatment. Cells were fixed with 1% formaldehyde at room temperature for 10 minutes. The cross-linking reaction was quenched with 125 mM glycine and rotated at room temperature for 5 minutes. Fixed cells were scraped in PBS containing 1 mM PMSF and 1× protease inhibitor cocktail, and pelleted at 10,000 rpm for 3 minutes at 4°C. The cell pellet was frozen in an ethanol-dry ice bath and stored at −80°C until use. To immunoprecipitate STAT5, cell pellets were resuspended in 400 µL TSEI (20 mM Tris, 150 mM NaCl, 2 mM EDTA, 1% TritonX-100, 0.1% SDS, pH 8.0) and sonicated twice at 30% amplitude, 10 pulses for 10 seconds, and cycled 5 times to a combined sonication of 100 seconds using a Branson 450 sonifier (Heinemann; Schwäbisch Gmünd, Germany). Insoluble debris was pelleted at 13,200 rpm, at 4°C for 10 minutes, and the supernatant was pre-cleared with 10 µl 50% slurry protein-A agarose beads (Sigma-Aldrich). A small volume (4.25 µl) was saved to quantify total input. The pre-cleared lysate (85 µl) was then incubated with 2 µg anti-STAT5 antibody (N-20, sc-836×, Santa Cruz Biotech) or nonspecific rabbit IgG (Sigma-Aldrich) rotating at 4°C overnight. Immunocomplexes were precipitated with 20 µl pre-blocked 50% protein-A agarose beads/salmon sperm DNA for 1.5 hours at 4°C, and subsequently washed for 5 minutes three times with TSEI, once each with TSEII (20 mM Tris, 500 mM NaCl, 2 mM EDTA, 1% TritonX-100, 0.1% SDS, pH 8.0) and LiCl buffer (20 mM Tris, 250 mM LiCl, 1 mM EDTA, 1% NP-40, 1% sodium deoxycholate, pH 8.0), and twice with TE buffer as previously described (Ahmed et al., 2009). Immunocomplexes were eluted with 1% SDS in TE buffer at room temperature for 30 minutes, and cross-links were reversed overnight at 65°C. DNA was purified using EZ-10 Spin Column (BioBasic Inc.; Markham, ON) and eluted with 50 µl ddH2O. For quantification of STAT5 binding, real-time PCR was performed using 1 µl eluted DNA in a 10-µl reaction containing SsoFAST EvaGreen Supermix (Bio-Rad; Hercules, CA) and 100 nM each of forward and reverse primer. Primers used were as follows. CISH GAS elements: forward 5′-CCCCTCTGGGTAGCTTCAG-3′; reverse 5′-CCCTGAGCAGTGAAAGGAAA-3′. ABCG2 proximal GAS element: forward 5′-AAGTTTCTCCCCTTTCCTTCC-3′; reverse 5′-ACAGGTTGCCCAGTCACAAG-3′. ABCG2 distal GAS element: forward 5′-GCTTCCTAAGGGACCTGACAC-3′; reverse 5′-GGAAGGTGAGAGAAAGGAATATAGC-3′. Recruitment was calculated as % of total input.
Statistical Analysis.
All experiments, except for HMEC experiments, were conducted at least three times. Data presented are means ± S.E. Test of significance was performed using the Student’s t test or analysis of variance, followed by post-hoc pairwise multiple comparisons using the Tukey’s or Dunnett’s test. A P value < 0.05 was considered significant and shown as a nominal value without correcting for multiple testing. All statistical analyses were performed using SPSS Statistics Version 19 (IBM; Armonk, NY).
Results
PRL Induces ABCG2 in T-47D Cells.
To examine if PRL regulates ABCG2 expression, we first treated serum-starved PRLR-expressing T-47D human breast cancer epithelial cells with varying concentrations of PRL for 24 hours. PRL dose-dependently increased ABCG2 mRNA and the mRNA of cytokine-inducible SH2–containing protein (CISH), expressed from a known PRL-responsive gene (Fig. 1A). ABCG2 mRNA was maximally induced at 6 hours, whereas CISH mRNA was induced as early as 2 hours after PRL treatment (Fig. 1B). This induction was blocked by co-treatment with the transcription inhibitor actinomycin D (Supplemental Fig. 1). In addition, immunoblotting for ABCG2 protein in crude membrane preparations from T-47D cells treated with PRL for 24 hours showed that ABCG2 protein is also induced in a dose-dependent manner (Fig. 1C). In contrast, other human breast epithelial cells (MCF-7, MCF-10A, and primary HMEC, Supplemental Figs. 2 and 3), which express significantly less PRLR (Supplemental Fig. 4), did not show PRL-induced expression of ABCG2 and CISH mRNA despite responsiveness to the aryl hydrocarbon receptor agonist TCDD, a known inducer of ABCG2, CYP1A1, and TiPARP (Tan et al., 2010; Kress and Greenlee, 1997; Ma et al., 2001).
JAK2- and STAT5-Dependency in the Induction of ABCG2 by PRL.
JAK2, one of the most proximal proteins in the PRL signaling pathway, plays an important role in phosphorylating and hence modulating downstream signaling cascades. To determine the role of JAK2 in the induction of ABCG2 by PRL, we used a combination of two siRNAs that target JAK2 to knockdown JAK2 expression in T-47D cells (Fig. 2A). JAK2-targeting siRNAs significantly attenuated the response to PRL treatment (Fig. 2B). STAT5 represents the major effector of activated JAK2. Using a similar approach to determine if STAT5 mediates PRL-induced expression of ABCG2, T-47D cells were transfected with a siRNA (siSTAT5A/B) that targets both isoforms STAT5A and STAT5B (Fig. 3A). The PRL induction of ABCG2 and CISH mRNA was significantly reduced in siSTAT5A/B-transfected T-47D (Fig. 3B). Consistent with these results, co-treatment with a novel STAT5 inhibitor (Müller et al., 2008) that partially reduced PRL-stimulated STAT5 phosphorylation/activation (Supplemental Fig. 5) significantly blunted the induction of ABCG2 and CISH mRNA by PRL (Fig. 3C). These experiments demonstrated that the JAK2/STAT5 pathway was important in the induction of ABCG2 by PRL.
STAT5 Recruitment to a Putative Proximal GAS Element in the Human ABCG2 Gene.
In silico examination of the 10-kilobase pair 5′-flanking region of the human ABCG2 gene for GAS elements with sequence 5′-TTCNNNGAA-3′ revealed a distal site at −4459 and a proximal site at −434 (Fig. 4A). Chromatin immunoprecipitation (ChIP) analyses further showed that STAT5 was recruited to the proximal GAS element in a time-dependent manner, with peak recruitment achieved at 1 hour after PRL treatment, but the distal GAS motif did not show significant STAT5 binding (Fig. 4B). This time-course of STAT5 recruitment was consistent with that observed for the CISH gene, suggesting a similar mechanism of regulation. Further, STAT5 recruitment to the proximal ABCG2 GAS element and CISH GAS elements was attenuated by knockdown of STAT5 expression (Fig. 4C) or by inhibition of STAT5 activation (Fig. 4D).
Evidence for a Functional Proximal GAS Element in the ABCG2 Gene.
To test if the putative GAS elements were functional, we transfected T-47D with the luciferase reporter construct pGL3-ABCG2 (Bailey-Dell et al., 2001). This construct, which contained the proximal GAS element and surrounding region, was induced approximately twofold by PRL (Fig. 5A). We confirmed PRL responsiveness of our transfection protocol by using the pGL4-CISH reporter construct (Fang et al., 2008). The addition of the distal GAS element and the neighboring region to pGL3-ABCG2 did not further enhance reporter activity (Fig. 5B), which was consistent with the ChIP data (Fig. 4B). In contrast, by introducing a single mutation into the proximal GAS element, reporter activity was significantly reduced (Fig. 5C). Similar results were observed when a different single mutation was introduced into the proximal GAS element (data not shown). To investigate whether this proximal GAS element was functional in isolation, various numbers of copies of the GAS element were cloned into the minimum promoter-driven pGL4.23 luciferase reporter construct (Fig. 6A). The reporter activity of these constructs was induced by PRL in a dose- and copy number–dependent manner (Fig. 6B). In particular, GAS6v2.4 containing six tandem repeats of the proximal GAS element was very sensitive to PRL treatment. PRL-induced reporter activity was synergistically increased by co-transfection with rat STAT5A (Fig. 6C) or with mouse STAT5A and STAT5B (unpublished data) expression vector. As evidence that this reporter activity was dependent on an intact GAS element, double mutation to each of the GAS elements, as shown using the GAS6v2mut.8 construct, completely abolished reporter activity (Fig. 6C).
PRL-Induced ABCG2 Expression Is Attenuated by MAPK and PI3K Pathway Inhibitors.
In addition to JAK2/STAT5, the MAPK and PI3K pathways are also activated by PRL signaling. The effect of these pathways on the induction of ABCG2 by PRL was examined using the MAPK pathway inhibitors U0126 and PD98059, and the PI3K pathway inhibitors LY294002 and wortmannin. These inhibitors were used at concentrations that consistently blocked activation of the MAPK or PI3K pathway as assessed by reduction of PRL-stimulated phosphorylation of ERK1/2 or AKT as shown by Western blot (Fig. 7, A–D). Inhibition of MAPK and PI3K signaling attenuated the inductive effect of PRL on ABCG2 but not CISH mRNA at 6 hours after treatment (Fig. 7, E–H).
STAT5 Recruitment to the Proximal GAS Element Is Not affected by MAPK and PI3K Inhibitors.
One potential mechanism by which small molecule-–mediated inhibition of MAPK and PI3K pathways can attenuate the induction of ABCG2 by PRL is by repressing STAT5 recruitment. To test this, we performed ChIP to assess the effect of U0126, PD98059, LY294002 and wortmannin on PRL stimulated STAT5 recruitment. Co-treatment with these inhibitors did not affect STAT5 recruitment to the ABCG2 proximal GAS element and the CISH GAS elements (Fig. 8, A and B). This suggests that MAPK and PI3K pathways modulate PRL-induced ABCG2 expression by a mechanism distinct from STAT5 recruitment.
GH Induces ABCG2 mRNA and STAT5 Recruitment to the ABCG2 Proximal GAS Element.
GH is a member of the peptide hormone family to which PRL belongs and can activate similar signaling networks, including JAK2/STAT5. For this reason, to examine the generalization that JAK2/STAT5 activators are potential inducers of ABCG2, T-47D cells were treated with different concentrations of recombinant human GH for 6 hours. The mRNA expression of ABCG2 and CISH was up-regulated by GH in a dose-dependent manner (Fig. 9A). To determine if GH, like PRL, stimulates STAT5 recruitment to the ABCG2 proximal GAS element, we performed ChIP at 1 hour after GH treatment (200 ng/ml). GH treatment significantly increased STAT5 occupancy at the ABCG2 proximal GAS element (Fig. 9B), demonstrating a potential role for STAT5 in GH-induced ABCG2 expression.
Discussion
The multidrug resistance transporter ABCG2 is up-regulated in the mammary gland during lactation (Jonker et al., 2005). To date, few studies have addressed the potential regulatory mechanisms that may explain this phenomenon. In the present study, we show that PRL up-regulates ABCG2 in T-47D human breast cancer epithelial cells via activation of JAK2/STAT5, MAPK, and PI3K signaling. We further identify a functional STAT5 binding sequence (GAS element) at position −434bp upstream of the human ABCG2 transcription start site. To our knowledge, this is the first evidence of JAK2/STAT5–mediated regulation of ABCG2.
Our results extend the findings of Wang et al. (2008b), to provide mechanistic insight into the regulation of ABCG2 by PRL at the transcriptional level. We used T-47D cells as a PRLR-expressing model system since the PRL response monitored by known target genes was not observed in other mammary cell lines, such as MCF-7, MCF-10A, and primary HMEC, probably due to low PRLR expression. The expression of the functional long form of the PRLR in the mammary gland of rat and mouse is temporally regulated, showing lactation-associated up-regulation (Camarillo et al., 2001; Mizoguchi et al., 1996). Although experimental data to confirm this in humans is lacking, T-47D cells serve as a model system to study PRL signaling in the human mammary gland.
Our in silico search for GAS elements with sequences 5′-TTCNNNGAA-3′ in the proximal 10-kilobase pair 5′flanking region of the human ABCG2 gene revealed two putative sites (Fig. 4A). We then identified the proximal GAS element (position −434) to be functional and able to recruit STAT5 upon PRL stimulation (Figs. 4–6). This GAS element may serve as an important site for crosstalk between STAT5 and other nuclear receptors that bind the ABCG2 gene. The inability for the distal GAS site to bind STAT5 agrees with the more stringent consensus STAT5 (5′-TTCC/TNG/AGAA-3′)-binding motif that was identified by in vitro DNA binding site selection using pools of double-stranded oligonucleotides (Soldaini et al., 2000).
The JAK2/STAT5 pathway is the best characterized system activated in PRL signaling. It mediates many of the effects of PRL, such as induction of milk protein expression during lactation. Here we show that knockdown or pharmacologic inhibition of JAK2 (Fig. 2) or STAT5 (Fig. 3) attenuates PRL-induced ABCG2 mRNA expression. This demonstrates that the induction of ABCG2 by PRL depends on the JAK2/STAT5 pathway. This has broad implications because this pathway is activated not only by PRL but also by a variety of hormones and cytokines, including erythropoietin; interleukins 2, 3, 5, and 7; granulocyte-macrophage-colony stimulating factor; and GH (Tan and Nevalainen, 2008). In this study, we show that growth hormone also induces ABCG2 and stimulates the recruitment of STAT5 to the proximal GAS element (Fig. 9). Since the action of GH can be mediated by both the GH receptor and the PRL receptor (Goffin et al., 1996; Xu et al., 2011), these results must be interpreted with caution. Nonetheless, our results suggest that JAK2/STAT5 activators are potential inducers of ABCG2.
Meyer zu Schwabedissen et al. (2006) showed that the tyrosine kinase inhibitor AG1478 and the MEK inhibitor PD98059 abolished EGF-induced ABCG2 expression. This led the authors to conclude that EGF induced ABCG2 by activation of the MAPK cascade. In the present study, we show that co-treatment with two different MEK inhibitors U0126 and PD98059 attenuates the induction of ABCG2 by PRL. This further supports the involvement of the MAPK pathway in ABCG2 gene regulation.
Unlike the JAK2/STAT5 and MAPK pathway, regulation of ABCG2 by PI3K/AKT signaling is comparatively well studied; however, its exact role is a matter of debate. First reported by Mogi et al. (2003), bone marrow from Akt1 knockout mice shows significantly fewer side population (SP) cells than are seen in the marrow of wild-type mice, reflecting decreased expression of Bcrp1. Furthermore, the SP fraction in bone marrow cells can be reduced by inhibiting PI3K or conversely, increased by overexpressing constitutively active AKT. Using immunofluorescence microscopy, the authors concluded that this is a result of PI3K/AKT-induced alteration of ABCG2 membrane localization but not gene expression. Similarly, using the porcine kidney epithelial cell line LLC-PK1 that stably expresses human ABCG2, Takada et al. (2005) showed that ABCG2 expressed in the plasma membrane was internalized by treatment with PI3K inhibitors or by overexpression of dominant-negative AKT. More recently, it was reported that neurospheres deficient in PTEN, a protein that antagonizes the action of PI3K, contained more SP cells compared with controls (Bleau et al., 2009). Furthermore, inhibition of PI3K effectively reduced the SP fraction, with a shift in ABCG2 localization from membrane to cytoplasm, with no changes in protein and mRNA expression. Wang et al. (2010) using the human hepatocarcinoma PLC cell line, demonstrated that overexpression of OCT4 induces phosphorylation of AKT and the expression of ABCG2, the effect of which was blocked by inhibiting PI3K. Our findings that PI3K inhibitors reduce PRL-induced ABCG2 expression provide further evidence of the involvement of the PI3K pathway in the transcriptional regulation of ABCG2. Together, these data support not only a major role for PI3K activation in ABCG2 membrane localization but also a modulatory role in ABCG2 expression in the presence of other pathways, such as those activated by OCT4 or PRL.
In our study, PRL-stimulated STAT5 recruitment to the ABCG2 proximal GAS element was not affected by MAPK and PI3K pathway inhibitors (Fig. 8), suggesting that STAT5 recruitment to the GAS element is independent of MAPK and PI3K activation. Given that MAPK or PI3K pathway inhibitors attenuate ABCG2 induction by PRL, we further postulate that PRL induction of ABCG2 depends on activation of MAPK and PI3K pathways. One hypothesis is that STAT5 recruits co-regulators that are activated by MAPK and PI3K pathways, which is necessary for optimal ABCG2 gene expression (Fig. 10). The identity of these putative effector proteins remains to be determined. Intriguingly, in addition to JAK2/STAT5 mediating many of the effects of PRL, the MAPK and PI3K pathway components are over-represented in the transcriptome of the lactating mammary gland (Lemay et al., 2007; Maningat et al., 2009). All three pathways may therefore contribute to ABCG2 expression in the mammary epithelium during lactation.
Of note, CISH expression was used as an indicator of PRL responsiveness, but our results suggest that there are differences in transcriptional regulation between CISH and ABCG2. JAK2 and STAT5 knockdown caused a significant induction of basal ABCG2 but not CISH mRNA (Figs. 2B and 3B), the effect of which was not observed with the STAT5 inhibitor (Fig. 3C). We speculate that prolonged suppression of JAK2/STAT5 signaling achieved only with siRNA may have disinhibited pathways that are negatively regulated by JAK2/STAT5 and also those that are positive regulators of ABCG2 but not CISH. One notable example is the AHR pathway, which is a positive regulator of ABCG2 (Tan et al., 2010). For reasons unknown, the basal expression of AHR-target gene CYP1A1 was induced by 50% after JAK2 knockdown (unpublished observations). The lack of effect of MAPK and PI3K pathway inhibitors on PRL-induced CISH expression further highlights differences in transcriptional control between the two genes. The CISH gene appears to be regulated mainly by JAK2/STAT5, whereas regulation of ABCG2 is more multifactorial.
The significance of our findings extends beyond understanding how ABCG2 is regulated by PRL during lactation. The PRL receptor is overexpressed in 60%–95% of human breast cancers (Reynolds et al., 1997; Ormandy et al., 1997; Touraine et al., 1998; Gill et al., 2001). This is attributed to reduced phosphorylation of Ser349 within the phospho-degron of the PRLR, resulting in impaired degradation of the PRLR in human breast cancers (Li et al., 2006). PRL is also produced by the mammary epithelium (Reynolds et al., 1997, Clevenger et al., 1995) and has been shown to be up-regulated in breast cancers compared with normal/hyperplastic epithelium (McHale et al., 2008). There is strong evidence that PRL plays an important role in breast cancer progression, and for this reason, there have been efforts to find chemotherapies that target the PRL pathway (Goffin et al., 2005; Clevenger et al., 2008). In addition to its role in the biology of breast cancer progression, there is evidence that PRL can modulate the cytotoxicity of chemotherapeutic agents. PRL antagonists have been shown to enhance the cytotoxic effect of cisplatin, paclitaxel, and doxorubicin in breast cancer cell lines (Ramamoorthy et al., 2001; Howell et al., 2008), whereas pretreatment with PRL can attenuate the cytotoxicity of chemotherapeutic agents taxol, vinblastine, doxorubicin, and cisplatin (LaPensee et al., 2009). Our results suggest that PRL may confer resistance to such chemotherapeutics as doxorubicin by induction of ABCG2. The significance of this induction will require further study.
In summary, we show that PRL induces ABCG2 expression in T-47D human breast cancer cells. Our results are the first to demonstrate regulation of ABCG2 by JAK2/STAT5 and further support the involvement of the MAPK and PI3K pathways in the transcriptional regulation of ABCG2. These findings suggest a potential role for PRL in the regulation of ABCG2 during lactation and offers additional mechanistic insight into the regulation of ABCG2 expression in certain breast cancer cells.
Acknowledgments
The authors thank Dr. Dwayne L. Barber (Ontario Cancer Institute, Princess Margaret Hospital, Toronto, ON) for generously providing us the mouse Stat5a and Stat5b expression plasmids. Special thanks to Raymond Lo (Department of Pharmacology and Toxicology, University of Toronto) for technical guidance on the ChIP protocol.
Authorship Contributions
Participated in research design: Wu, Riddick, Matthews, Harper, Ito.
Conducted experiments: Wu, Dalvi, Lu, Yang.
Contributed new reagents or analytic tools: Matthews, Ross, Clevenger.
Performed data analysis: Wu, Harper, Ito.
Wrote or contributed to the writing of the manuscript: Wu, Riddick, Matthews, Ross, Clevenger, Harper, Ito.
Footnotes
This work was mainly supported by the Canadian Institutes of Health Research [Grants MT13747 and MOP-93759] and Frederick Banting and Charles Best Canada Graduate Scholarship, with additional support provided in part by University of Toronto Fellowship, Avon Foundation, and by a Merit Review grant from the US Department of Veterans Affairs.
This article has supplemental material available at molpharm.aspetjournals.org.
Abbreviations
- AKT
- protein kinase B
- bp
- base pair
- ChIP
- chromatin immunoprecipitation
- EGF
- epidermal growth factor
- FBS
- fetal bovine serum
- GAS
- γ-interferon activation sequence
- GH
- growth hormone
- CISH
- cytokine-inducible SH2-containing protein
- DMEM
- Dulbecco’s modified Eagle’s medium
- DMSO
- dimethylsulfoxide
- ERK1/2
- extracellular signal–regulated kinases 1 and 2
- FBS
- fetal bovine serum
- HMEC
- human mammary epithelial cells
- JAK2
- Janus kinase 2
- LY294002
- 2-(4-morpholinyl)-8-phenyl-1(4H)-benzopyran-4-one
- MAPK
- mitogen-activated protein kinase
- MEBM
- mammary epithelial basal medium
- PBS
- phosphate buffered saline
- PCR
- polymerase chain reaction
- PD98059
- 2-(2-amino-3-methoxyphenyl)-4H-1-benzopyran-4-one
- PI3K
- phosphoinositide-3-kinase
- PMSF
- phenylmethylsulfonyl fluoride
- PRL
- prolactin
- PRLR
- prolactin receptor
- RT-PCR
- reverse transcriptase polymerase chain reaction
- siRNA
- short-interfering RNA
- SP
- side population
- STAT5
- signal transducer and activator of transcription-5
- U0126
- 1,4-diamino-2,3-dicyano-1,4-bis(2-aminophenylthio) butadiene
- Received September 12, 2012.
- Accepted November 13, 2012.
- U.S. Government work not protected by U.S. copyright