Proinflammatory Cytokines Enhance Estrogen-dependent Expression of the Multidrug Transporter Gene ABCG2 through Estrogen Receptor and NFκB Cooperativity at Adjacent Response Elements*♦

Constitutive activation of NFκB in estrogen receptor (ER)-positive breast cancer is associated with tumor recurrence and development of anti-estrogen resistance. Furthermore, a gene expression signature containing common targets for ER and NFκB has been identified and found to be associated with the more aggressive luminal B intrinsic subtype of ER-positive breast tumors. Here, we describe a novel mechanism by which ER and NFκB cooperate to up-regulate expression of one important gene from this signature, ABCG2, which encodes a transporter protein associated with the development of drug-resistant breast cancer. We and others have confirmed that this gene is regulated primarily by estrogen in an ER- and estrogen response element (ERE)-dependent manner. We found that whereas proinflammatory cytokines have little effect on this gene in the absence of 17β-estradiol, they can potentiate ER activity in an NFκB-dependent manner. ER allows the NFκB family member p65 to access a latent NFκB response element located near the ERE in the gene promoter. NFκB recruitment to the gene is, in turn, required to stabilize ER occupancy at the functional ERE. The result of this cooperative binding of ER and p65 at adjacent response elements leads to a major increase in both ABCG2 mRNA and protein expression. These findings indicate that estrogen and inflammatory factors can modify each other's activity through modulation of transcription factor accessibility and/or occupancy at adjacent response elements. This novel transcriptional mechanism could have important implications in breast cancer, where both inflammation and estrogen can promote cancer progression.

The estrogen 17␤-estradiol (E 2 ) 2 is a steroid that plays an important role in reproductive tissues, as well as in the skeletal, cardiovascular, immune, and central nervous systems, by regu-lating a number of cellular processes, such as proliferation, differentiation, and survival. In the classical mechanism of estrogen action, E 2 binds to the estrogen receptor (ER) to promote receptor homodimerization. The ligand-bound receptor binds to cognate DNA sequences, called estrogen response elements (EREs), which leads to coregulator recruitment and target gene transcription. In addition to direct DNA binding, ER can also regulate gene transcription via protein-protein interaction with other DNA-binding transcription factors, such as the proteins of the AP-1 complex and Sp1 (1,2).
Estrogen can also modulate gene transcription by the proinflammatory transcription factor NFB, which, like ER, influences numerous cellular processes. In the classical NFB pathway, binding of proinflammatory cytokines to their receptors activates the IB kinase (IKK) complex, which phosphorylates the inhibitory protein IB, leading to its subsequent ubiquitination and proteasomal degradation. NFB family members p65 and p50 can then translocate to the nucleus and regulate transcription of a cohort of genes by binding to specific DNA elements called NFB response elements (NFBREs). Numerous studies in a variety of physiological systems and models have demonstrated that mutual transrepression occurs between ER and NFB, with ER repressing transcription by NFB (3) and NFB repressing transcription by ER (4,5).
However, in breast cancer cells, positive cross-talk between ER and NFB may also be particularly important. Inhibition of NFB has been shown to restore responsiveness to anti-estrogens in cell lines that were originally resistant to endocrine therapy (6,7). Recent studies have also found that hormone-dependent tumors with a high risk for recurrence have constitutive activation of NFB (8). Moreover, recent work from our laboratory highlights a high degree of positive cross-talk between ER and NFB in the synergistic up-regulation of a number of common target genes. This synergistic gene signature is associated with the more aggressive luminal B subtype of ER-positive breast tumors and delineates the responsiveness of ER-positive breast tumors to tamoxifen therapy (9).
One gene from this signature that is highly important in breast cancer response to therapeutic drugs is ABCG2 (ATPbinding cassette transporter G2) (9). The ABCG2 gene encodes an ABC transporter that is capable of pumping a number of endogenous and exogenous agents out of cells (10). ABCG2, also known as BCRP (breast cancer resistance protein), causes the efflux of a spectrum of anticancer drugs out of breast cancer cells, including mitoxantrone, camptothecin-derived and indolocarbazole topoisomerase I inhibitors, methotrexate, flavopiridol, and quinazoline ErbB1 inhibitors, and can thereby contribute to drug-resistant breast cancers (10). Furthermore, overexpression of ABCG2 has been shown to confer anticancer drug resistance in the breast cancer cell line MCF-7 (11).
Because of its known role in promoting drug-resistant breast tumors, we sought to understand the mechanism by which the combination of E 2 and proinflammatory cytokines leads to highly elevated levels of ABCG2 expression in breast cancer cells. Although previous studies have shown that ABCG2 can be either up-or down-regulated by E 2 or proinflammatory cytokines acting independently (12)(13)(14)(15), here, we establish that cooperativity between ER and p65 occurs at two adjacent response elements, leading to a synergistic increase in both ABCG2 mRNA and protein expression in breast cancer cells.
mRNA Analysis by Quantitative PCR (qPCR)-RNA isolation, reverse transcription, and qPCR were carried out as described previously (16). 36B4 was used as an internal control. Primer sequences for qPCR are listed in supplemental Table 1.
Plasmids and Mutagenesis-Fragments of the ABCG2 promoter (Ϫ243 to ϩ362 and Ϫ115 to ϩ362) subcloned into the pGL3-basic reporter plasmid were kindly provided by Drs. D. Ross (University of Maryland) and W. Beck (University of Illinois at Chicago). The ABCG2 (Ϫ243 to ϩ362) plasmid was used to mutate the ERE and either one or both NFBRE sequences using the QuikChange Lightning site-directed mutagenesis kit (Stratagene, La Jolla, CA). Wild-type and mutant sequences are given in supplemental Table 1. Mutagenesis was confirmed by sequencing using a 3730 DNA analyzer (Applied Biosystems, Foster City, CA). MCF-7 cells were transiently transfected with reporter constructs, and Dual-Luciferase assays were carried out as described previously (16).
ChIP Assay-ChIP assays were carried out as described previously with certain modifications (16,17). MCF-7 cells were seeded in 10-cm plates and grown in phenol red-free medium containing 5% charcoal/dextran-stripped calf serum. On reaching 90% confluency, cells were treated as described in the figure legends. Cross-linking was carried out with 1.5% formaldehyde for 15 min at room temperature. Cells were lysed with SDS lysis buffer (1% SDS, 10 mM EDTA, pH 8.0, and 50 mM Tris-HCl, pH 8.1), and chromatin was sonicated three times for 10 s using a Fisher Model 100 Sonic Dismembrator. Protein A-coated magnetic beads (100-02D, Dynabeads, Invitrogen) washed three times with 5% BSA in PBS were incubated overnight at 4°C with 1 g of ER␣ or 4 g of p65 antibody. Antibody-bound beads were washed two times with 5% BSA in PBS and incubated overnight at 4°C with sonicated chromatin diluted in dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, pH 8.0, 16.7 mM Tris-HCl, pH 8.1, and 167 mM NaCl). Beads were serially washed with low salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, pH 8.0, 20 mM Tris-HCl, pH 8.1, and 150 mM NaCl), high salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, pH 8.0, 20 mM Tris-HCl, pH 8.1, and 500 mM NaCl), and LiCl immune complex wash buffer (0.25 M LiCl, 1% IGEPAL CA-630, 1% deoxycholic acid (sodium salt), 1 mM EDTA, pH 8.0, and 10 mM Tris-HCl, pH 8.1) and two times with Tris/ EDTA (10 mM Tris-HCl, pH 8.0, and 1 mM EDTA). DNA was eluted from the beads in elution buffer containing 0.1 M sodium bicarbonate and 1% SDS. Decross-linking was carried out for 16 h at 65°C. DNA was purified using QIAquick columns (Qiagen, Valencia, CA) and eluted in prewarmed water. Approximately 5% of the cell lysate was removed as input prior to immunoprecipitation. Inputs were diluted at 1:20 or 1:200. The percent maximum occupancy was calculated as follows: the -fold change for each treatment was determined relative to the input for each sample and the untreated control for each experiment using the ⌬⌬C t method. The percent maximum occupancy for each assay was then calculated relative to the treatment with the highest -fold change, which was set to 100%. Data were plotted as the mean percent maximum occupancy Ϯ S.E. for at least three independent assays. The primers for ChIP-qPCR are listed in supplemental Table 1.
Western Blotting-Whole-cell and nuclear extracts were prepared from MCF-7 cells using M-PER and NE-PER reagents, respectively (Pierce), containing protease and phosphatase inhibitors (Roche Applied Science). SDS-PAGE was carried out, and proteins were detected by Western blotting using the antibodies indicated, with ␤-actin or H3 serving as an internal loading control.
DNA Affinity Pulldown Assay (DAPA)-DNA probes were prepared from ABCG2 reporter constructs using a biotinylated forward primer. The resulting 434-bp probes with either an intact or mutant ERE or NFBRE were hybridized to streptavidin-coated magnetic beads (Invitrogen). Nuclear extracts were prepared from MCF-7 cells using a nuclear extraction kit (AY2002, Panomics, Fremont, CA) following the manufac-turer's instructions. The hybridized probes were incubated with nuclear extracts for 30 min at room temperature in a 60-l binding reaction as described previously (18). The probes were then immobilized on a magnetic stand, and unbound proteins were removed by washing three times with wash buffer (18). Proteins retained by the probes were reduced with 1ϫ electrophoresis buffer, and SDS-PAGE was carried out. ER␣ and p65 bound to the probes were detected by Western blotting.
Statistical Analysis-qPCR and reporter data were analyzed by two-way analysis of variance, followed by a post hoc Bonferroni test. Significance for all statistical tests was set at p Ͻ 0.05. The data shown are the mean Ϯ S.E. from at least three independent determinations.

TNF␣ Potentiates E 2 -mediated
Regulation of ABCG2 in an NFBdependent Manner-To confirm the finding from our microarray study that showed synergistic up-regulation of ABCG2 by the combination of E 2 and TNF␣ compared with either E 2 or TNF␣ alone (9), we examined regulation of ABCG2 mRNA in ER-positive MCF-7 breast cancer cells over a period of 4 h. We observed that a rapid and robust up-regulation of ABCG2 expression occurred in response to E 2 ϩ TNF␣ compared with a modest increase in response to E 2 and minimum regulation in response to TNF␣ (Fig. 1A). Doseresponse studies showed that ABCG2 mRNA was regulated by E 2 in a typical dose-dependent manner, an effect that was significantly increased by the presence of TNF␣ (Fig. 1B). In contrast, TNF␣ had a minimum effect on ABCG2 expression at all doses tested in the absence of E 2 but showed a typical dose response in the presence of E 2 (Fig. 1C). Examination of ABCG2 protein levels showed a pattern of regulation similar to mRNA levels, with E 2 stimulation of ABCG2 expression greatly enhanced by E 2 in combination with TNF␣ (Fig.  1D). These findings indicate that TNF␣ can significantly potentiate the effect of E 2 on ABCG2 expression at both the mRNA and protein levels but has little effect on the gene in the absence of E 2 .
To understand the mechanism by which TNF␣ enhances E 2 -mediated expression of ABCG2, we initially examined the role of ER. We found that the pure ER antagonist ICI 182,780 completely blocked ABCG2 regulation both by E 2 alone and by E 2 ϩ TNF␣ ( Fig. 2A). Furthermore, use of a small molecule inhibitor that prevents ER binding to DNA, 8-[(benzylthio)methyl]theophylline (19), had an effect similar to ICI FIGURE 1. TNF␣ potentiates the up-regulation of ABCG2 mRNA and protein by E 2 . A, MCF-7 cells were treated with E 2 (10 nM), TNF␣ (10 ng/ml), or both for up to 4 h. RNA was isolated, and ABCG2 mRNA levels were examined by qPCR using the ⌬⌬C t method. 36B4 was used as an internal control. *, p Ͻ 0.05 compared with E 2 alone at the same time point. B, MCF-7 cells were treated for 2 h with or without TNF␣ (10 ng/ml) in the presence of increasing doses of E 2 . *, p Ͻ 0.05 compared with E 2 alone at the same dose. C, MCF-7 cells were treated for 2 h with or without E 2 (10 nM) in the presence of increasing doses of TNF␣. #, p Ͻ 0.05 compared with TNF␣ alone at the same dose. D, MCF-7 cells were treated with E 2 (E; 10 nM), TNF␣ (T; 10 ng/ml), or both for 6 h. ABCG2 and ␤-actin protein levels were examined by Western blotting as described under "Experimental Procedures." Treatment with vehicle (V) served as a control. FIGURE 2. TNF␣ potentiates ABCG2 expression by enhancing ER occupancy at an essential ERE. A and B, ABCG2 mRNA levels were examined by qPCR in MCF-7 cells pretreated for 2 h with or without the ER antagonist ICI 182,780 (ICI; 1 M) or a small molecule inhibitor of ER␣ binding to ERE, 8-[(benzylthio)methyl]theophylline (TPBM; 20 M), followed by treatment with E 2 (10 nM), TNF␣ (10 ng/ml), or both for an additional 2 h. *, p Ͻ 0.05 compared with E 2 alone. DMSO, dimethyl sulfoxide. C, MCF-7 cells were transfected with 1 g of an ABCG2luciferase reporter plasmid in which the ERE located at Ϫ180 was intact (Ϫ243), mutated (mutERE), or deleted (Ϫ115), along with 200 ng of the Renilla luciferase control plasmid pGL4.70. Dual-Luciferase assays were carried out following treatment with E 2 , TNF␣, or both for 4 h as described under "Experimental Procedures." *, p Ͻ 0.05 compared with E 2 alone. D, MCF-7 cells were treated with E 2 , TNF␣, or both for up to 60 min. ChIP assay was performed, and ER recruitment to the ABCG2 regulatory region containing the ERE was examined by qPCR. *, p Ͻ 0.05 compared with E 2 alone. ER occupancy in response to different treatments is expressed as the percentage of maximum ER occupancy, which was observed at 45 min of E 2 ϩ TNF␣ treatment. E, ChIP assay was performed with an antibody specific for ER␣ or a nonspecific IgG antibody as a negative control. qPCR was performed for the ABCG2 regulatory region containing the ERE (Enhancer) or for a far upstream region (Upstream) as a negative control. *, p Ͻ 0.05 compared with E 2 alone; nd, non-detectable. F, nuclear extracts were prepared from MCF-7 cells treated with E 2 and TNF␣ for 45 min and incubated with a DNA probe from the ABCG2 gene in which the ERE was either intact (ERE-WT) or mutated (ERE-mut). DAPA was performed as described under "Experimental Procedures," and ER␣ binding to the probes was examined by Western blotting.
182,780 (Fig. 2B), indicating that ER is essential for the ability of TNF␣ to enhance expression of ABCG2 by E 2 .
Previous studies have suggested that ABCG2 can be up-regulated by ER acting through an ERE located at Ϫ180 of the proximal promoter region of the gene (12); hence, we examined whether ER action at this response element is affected by TNF␣. An artificial reporter construct encompassing the ABCG2 ERE and promoter (Ϫ243/ϩ362) was transiently transfected into MCF-7 cells and found to be regulated similarly to the endogenous gene, with regulation by E 2 but not TNF␣ alone and enhanced regulation by the combination of the two (Fig.  2C). Mutation of the ERE or deletion of a region containing the ERE (Ϫ243 to Ϫ115) completely abolished the transcriptional activity of the promoter under all of the treatment conditions.
To examine ER occupancy at the ABCG2 ERE, ChIP assays were carried out over a 60-min time course. ER occupancy was rapidly and robustly increased by the combination of E 2 ϩ TNF␣ over E 2 alone, with maximum ER occupancy observed after 45 min of treatment with E 2 ϩ TNF␣ (Fig. 2D). qPCR for a far upstream region of the ABCG2 gene showed no increase in ER occupancy (Fig. 2E). In addition, ChIP assays performed using a nonspecific IgG antibody showed no enrichment at either the ERE or the upstream site (Fig. 2E), indicating the specificity of the assay for ER occupancy at the ABCG2 ERE.
To examine whether ER binds to the ERE of the ABCG2 promoter, DAPA was carried out. Nuclear extracts from MCF-7 cells treated with E 2 ϩ TNF␣ were incubated with DNA probes containing the regulatory region of ABCG2, with the ERE intact or mutated. As determined by Western blotting, ER binding to the probes was greatly reduced when the ERE was mutated (Fig. 2F), indicating specific ER binding to this site in the ABCG2 promoter. These studies indicate that the ability of TNF␣ to increase the transcriptional regulation of ABCG2 by E 2 is mediated by enhanced ER occupancy and activity specifically at a functional ERE in the ABCG2 promoter.
Our previous studies on other common E 2 and TNF␣ target genes suggested that much of the cross-talk between these factors requires the NFB pathway (9,16). NFB was further implicated in the regulation of ABCG2 by our finding that IL-1␤, another proinflammatory cytokine that activates NFB, could enhance E 2 -regulated expression of ABCG2. On the other hand, IL-6, which does not activate NFB, could not potentiate E 2 action (Fig. 3A).
To further examine the role of NFB, IB␣-DN, which blocks NFB activation because it cannot be phosphorylated or subsequently degraded in response to cytokine treatment, was used. The enhanced regulation of ABCG2 by E 2 ϩ TNF␣ over that seen with E 2 alone was completely abrogated by IB␣-DN (Fig. 3B). Furthermore, the enhanced occupancy of ER at the ABCG2 ERE with E 2 ϩ TNF␣ was substantially reduced, nearly to the level of E 2 alone, by IB␣-DN (Fig. 3C) or by a small molecule inhibitor of IKK (Fig. 3D). These findings indicate a clear role for NFB in the ability of TNF␣ to potentiate E 2 action through increased ER occupancy at the ABCG2 gene. Taken together, our results indicate that ER is the primary factor required to drive transcription of ABCG2, whereas TNF␣, acting through the NFB pathway, mainly operates to augment ER action on this gene. E 2 Allows NFB Action through a Latent NFBRE-To determine the mechanism by which NFB can potentiate ER action, we examined the regulatory region of the ABCG2 promoter for possible response elements through which NFB may exert its effect. Interestingly, we found two putative NFBREs (N1 and N2) located downstream of the ABCG2 ERE (Fig. 4A). Because TNF␣ cannot regulate expression of ABCG2 on its own, these NFBREs appear not to be functional. However, mutational analysis revealed that the NFBRE closer to the ERE, N1, but not the one farther downstream, N2, was required for TNF␣ to potentiate E 2 action (Fig. 4B). The effect of mutating both NFBREs was similar to that of mutating N1 alone, indicating that this response element is sufficient to mediate the effects of NFB on this gene. However, these findings also indicate that N1 is a latent response element that requires the presence of E 2 along with TNF␣ to become functional.  A and B, ABCG2 mRNA levels were determined by qPCR in MCF-7 cells treated for 2 h with 10 ng/ml IL-1␤ or IL-6 with or without 10 nM E 2 or following infection with adenoviral constructs for either GFP (control) or IB␣-DN for 24 h prior to treatment with E 2 and/or TNF␣ for 2 h. *, p Ͻ 0.05 compared with E 2 or IL-1␤ alone. C and D, ChIP assays for ER were carried out 24 h after infection of cells with GFP or IB␣-DN adenoviral constructs or 2 h after pretreatment with an IKK inhibitor (1 M), followed by treatment with E 2 , TNF␣, or both for 45 min. *, p Ͻ 0.05 compared with E 2 alone in the same group.  OCTOBER 8, 2010 • VOLUME 285 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 31103
To examine whether the NFBRE N1 is capable of recruiting the NFB family member p65, a ChIP assay was carried out. Because N1 requires the presence of E 2 to become functional, recruitment of p65 to the ABCG2 regulatory region was determined at 45 min, which is the time point of maximum ER occupancy. Interestingly, p65 recruitment to ABCG2 was observed only in the presence of E 2 ϩ TNF␣ but not TNF␣ alone. Furthermore, this recruitment was totally prevented by ICI 182,780 (Fig. 5A), indicating that ER is required for p65 recruitment to this gene. This was in contrast to a control gene that showed typical p65 recruitment in response to TNF␣ alone (data not shown).
To examine if p65 is recruited to N1, DAPA was carried out using nuclear extracts from MCF-7 cells treated with E 2 ϩ TNF␣ and the regulatory region of the ABCG2 gene with intact or mutant N1 as DNA probes. Our findings revealed that p65 bound to the ABCG2 probe with an intact N1 site, but binding was significantly impaired when the N1 site was mutated (Fig.   5B), indicating that p65 recruitment to ABCG2 in the presence of E 2 ϩ TNF␣ requires the latent N1 response element.
To determine whether the effect of E 2 on p65 recruitment to DNA is due to a global change in the NFB pathway, whole-cell and nuclear extracts were prepared from MCF-7 cells treated with E 2 , TNF␣, or both for 5 or 15 min. TNF␣ caused a marked increase in IB␣ phosphorylation (phospho-IB␣) by 5 min and accompanying protein degradation at both time points. Neither of these events was affected by E 2 alone or in combination with TNF␣ (Fig. 5C). Similarly, TNF␣ increased the nuclear levels of phospho-p65 and total p65 at both time points, and neither was affected by E 2 . These results indicate that the effect of E 2 on p65 occupancy at N1 of the ABCG2 gene is a gene-specific, ER-dependent mechanism rather than the result of a global change in p65 activation or nuclear translocation caused by E 2 .
To understand why the N1 site requires ER for its functionality, we examined the sequence of the response element and found that it diverges considerably from a consensus p65 response element (Fig. 6A) (20). When we mutated the N1 sequence to a consensus p65 sequence (N1-Cons), we found that TNF␣ alone, in the absence of E 2 , could now significantly enhance ABCG2 promoter activity (Fig. 6B). Mutation of the ERE within the context of a consensus N1 site had no effect on the ability of TNF␣ to stimulate the promoter (data not shown), indicating that a consensus site no longer requires ER for its functionality. Furthermore, the combination of E 2 and TNF␣ led to a far greater stimulation of ABCG2 transcription (Fig.  6B), suggesting that cooperativity can still occur between ER and NFB even when a consensus NFBRE is present.

DISCUSSION
ABCG2 is a transmembrane transporter that carries out the important biological function of effluxing multiple endogenous and exogenous substances out of cells (10). On account of its ability to efflux multiple drugs, overexpression of this protein has been associated with drug-resistant cancers, including breast cancer (10,11). Also, a recent microarray study from our laboratory identified ABCG2 to be part of a gene signature that is associated with the more aggressive luminal B subtype of ER-positive breast tumors, which emphasizes the clinical relevance of this gene (9). Because of this noteworthy biological and clinical significance of ABCG2, the regulation of this gene has been extensively studied. Others have found that E 2 can up-or down-regulate ABCG2 depending on the cellular background, with differences potentially due to altered ratios of ER␣ to ER␤ expression in the cell (10, 12-14, 21, 22). Similarly, ABCG2 expression has also been shown to be either up-or down-regulated by numerous cytokines, including TNF␣ and IL-1␤, in a cell-specific manner (14,15). In addition, a number of other FIGURE 5. NFB family member p65 is recruited to the N1 site of the ABCG2 promoter in an ER-dependent manner. A, MCF-7 cells were pretreated for 2 h with or without the ER antagonist ICI 182,780 (ICI; 1 M), followed by treatment with E 2 , TNF␣, or both for 45 min. ChIP assay was carried out, and p65 recruitment to the ABCG2 promoter was examined by qPCR. *, p Ͻ 0.05 compared with TNF␣. B, nuclear extracts were prepared from MCF-7 cells treated with E 2 and TNF␣ for 45 min and incubated with a DNA probe from the ABCG2 gene in which N1 was either intact (N1-WT) or mutated (N1-mut). DAPA was performed as described under "Experimental Procedures," and p65 binding to the probes was examined by Western blotting. C, MCF-7 cells were treated with E 2 (E), TNF␣ (T), or both for 5 or 15 min, and whole-cell (WCE) and nuclear (NE) extracts were prepared. Phosphorylated and total IB␣ and ␤-actin protein levels were examined in whole-cell extracts, and phosphorylated and total p65 and histone H3 protein levels were examined in nuclear extracts by Western blotting. FIGURE 6. Conversion of the NFBRE of ABCG2 to a consensus p65 sequence renders the promoter responsive to TNF␣. A, the sequence of the N1 site of ABCG2 is compared with a consensus p65-binding motif (20). B, reporter activity was measured by a Dual-Luciferase assay in MCF-7 cells transfected with 1 g of ABCG2 reporter plasmid in which the N1 site was intact (Ϫ243) or mutated to a consensus p65-binding site (N1-Cons) and with 200 ng of Renilla luciferase control plasmid pGL4.70. Transfection was carried out, and luciferase activity was measured after 4 h of treatment with E 2 , TNF␣, or both. *, p Ͻ 0.05, TNF␣ compared with none (control); ns, not significant. transcription factors, signaling pathways, and growth factors can also regulate ABCG2 expression (23)(24)(25)(26). Moreover, the ABCG2 gene can be regulated through epigenetic mechanisms, including histone deacetylase inhibition and DNA methylation (27,28), as well as by post-transcriptional microRNA-dependent mechanisms (29,30). This high degree of regulation suggests that ABCG2 expression is tightly controlled and exquisitely sensitive to external stimuli. Furthermore, combinations of transcription factors, such as we find with ER and NFB, may work together in a combinatorial and cell-specific manner to fine-tune expression of ABCG2.
Our studies indicate that the combination of estrogen and proinflammatory cytokines can activate a novel transcriptional mechanism by which ER and NFB cooperate to up-regulate expression of ABCG2 in breast cancer cells. We have demonstrated that E 2 up-regulates ABCG2 expression in an ER-and ERE-dependent manner, whereas TNF␣ appears to have no effect on this gene in the absence of E 2 (Fig. 7). However, TNF␣ is capable of enhancing ER activity through a latent NFBRE that is functional only in the presence of ER. We propose that ER binding at the ERE allows p65 recruitment to the nearby non-consensus NFBRE. In turn, the presence of p65 appears to stabilize ER occupancy on the gene. The net result is the presence of two potent transcriptional activators, ER and p65, working together to cause a major increase in both ABCG2 mRNA and protein expression.
There are several unique features of ER and NFB crosstalk that underlie the regulatory mechanism for ABCG2. First, cooperativity between ER and NFB requires individual response elements for each transcription factor, which is in contrast to other genes that require only a single ERE or NFBRE for ER and NFB interaction (16,(31)(32)(33). Second, our studies show that the NFBRE of ABCG2 is a latent response element that requires ER for its ability to recruit p65. This is a surprising finding because ER has been shown to prevent NFB binding to NFBREs in various cellular backgrounds (34 -36). Third, we found that NFB can enhance ER occupancy at the ABCG2 promoter. Although stabilization of nuclear receptors on target gene promoters is associated with increased transcription (37), stabilization of ER by NFB has not been fully explored.
Cooperativity occurring between transcription factors at tandem response elements has been highlighted by numerous studies in the literature and is frequently associated with synergistic gene transcription (38,39). For example, previous studies have demonstrated cooperativity between multiple consensus or variant EREs or between an ERE and other nuclear receptor response elements (40 -42). ER has also been shown to act in a cooperative manner with Sp1. This interaction can be mediated by ER and Sp1 binding to independent sites, through various combinations of ERE half-sites and Sp1 sites, or through Sp1 sites alone (2). Cooperativity has also been observed between adjacent NFB sites, as well as between NFB sites and response elements for other transcription factors, such as STAT1, C/EBP, and NF-IL-6 (43-46), all of which act synergistically with NFB to increase transcription of target genes. Although cooperativity between transcription factors at adjacent response elements is fairly common, we have demonstrated here for the first time that such a reciprocal interaction can occur between ER and NFB at adjacent response elements and leads to the synergistic up-regulation of ABCG2.
The mechanism by which cooperativity between ER and NFB occurs is likely to involve a number of factors, including the kinetics of transcription factor-DNA interaction. Transcription factors have been shown to exhibit a rapid and dynamic interaction with DNA, termed "hit and run" (47). This model suggests that a transcription factor interacts with the promoter briefly (hit) and then is rapidly displaced from its binding site (run). The observed cooperativity in ER and NFB occupancy at the ABCG2 gene may be the result of increased frequency of transcription factor-DNA interactions, a decreased rate of transcription factor displacement, or both. How ER and p65 may alter each other's interaction with DNA is not known but could be explained by proteinprotein interactions, either directly between ER and p65 or indirectly through formation of a coregulator complex (32, 36, 48 -53).
In conclusion, our findings indicate that expression of ABCG2 can be synergistically up-regulated by a novel mechanism of cross-talk between ER and NFB at adjacent response elements. This finding could have important implications for patients with ER-positive breast tumors, in which both estradiol and proinflammatory cytokines are produced locally within the breast tumor microenvironment. Furthermore, our finding that ER and NFB are capable of working together to increase ABCG2 expression is consistent with reports that both of these transcription factors are associated with drug resistance in breast cancer (54,55) and that ER-positive breast tumors with constitutive activation of NFB may be more aggressive and resistant to tamoxifen (6 -8). 7. Proposed model for TNF␣ potentiation of E 2 -regulated ABCG2 expression involves cooperativity between ER and p65 at adjacent response elements. E 2 is capable of up-regulating ABCG2 expression through ER binding to an ERE in the ABCG2 promoter, whereas in the absence of E 2 , TNF␣ has no effect on ABCG2 transcription. However, in the presence of both E 2 and TNF␣, we found that ER is required to allow recruitment of the NFB subunit p65 to a latent NFBRE that displays no activity in the absence of E 2 but is required for TNF␣ potentiation of E 2 activity. Once recruited, p65 is capable of stabilizing ER occupancy on the gene. This reciprocity between ER and NFB at adjacent response elements ultimately leads to enhanced regulation of ABCG2.