B-cell Translocation Gene 2 (BTG2) Stimulates Cellular Antioxidant Defenses through the Antioxidant Transcription Factor NFE2L2 in Human Mammary Epithelial Cells*

Background: The putative tumor suppressor BTG2 is frequently down-regulated in human cancers. Results: BTG2 protects cells against oxidative stress, stimulates the activity of the antioxidant transcription factor NFE2L2, and associates with NFE2L2 at the antioxidant response element (ARE). Conclusion: BTG2 serves as a transcriptional coactivator for NFE2L2/ARE signaling. Significance: These findings suggest a novel mechanism to explain BTG2 function as a tumor suppressor. The B-cell translocation gene 2, BTG2, a member of the BTG/TOB (B-cell translocation gene/transducers of ErbB2) gene family, has been implicated in cell cycle regulation, normal development, and possibly tumor suppression. Previously, it was shown that BTG2 expression is lost or down-regulated in human breast cancers. We now report that BTG2 protects human mammary epithelial cells from oxidative stress due to hydrogen peroxide and other oxidants. BTG2 protection against oxidative stress is BRCA1-independent but requires the antioxidant transcription factor NFE2L2 and is associated with up-regulation of the expression of antioxidant enzymes, including catalase and superoxide dismutases 1 and 2. BTG2 stimulation of antioxidant gene expression is also NFE2L2-dependent. We further demonstrate that BTG2 is a binding partner for NFE2L2 and increases its transcriptional activity. In addition, BTG2 is detectable at the antioxidant response element (ARE) of several NFE2L2-responsive genes. Finally, we show that the ability of BTG2 to associate with NFE2L2, to protect cells against oxidative stress, and to stimulate antioxidant gene expression requires box B, a short highly conserved amino acid motif characteristic of BTG2/TOB family proteins, but does not require boxes A or C. These findings suggest a novel role for BTG2 as a co-activator for NFE2L2 in up-regulating cellular antioxidant defenses.

Murine Btg2 (also called Tis21 and Pc3) was identified as a gene that is rapidly and transiently induced in response to various stimuli (e.g. growth factors, phorbol esters, fetal calf serum) in 3T3 and other cell types (3). BTG2 is particularly expressed in noncycling cells, and its forced overexpression causes growth arrest in various rodent and human cell types (9 -13). The growth arrest occurs predominantly at the G 1 /S and G 2 /M boundaries (8, 9, 11, 14 -16). In this regard, BTG2 expression is regulated by the tumor suppressor p53, and BTG2 mediates p53-induced growth arrest (11,14). These findings suggest that BTG2 is a cell cycle-regulated gene that, in turn, functions as a negative regulator of the cell cycle.
The subcellular localization of BTG2 is primarily nuclear, but low levels of BTG2 can be detected in the cytoplasm (5,(17)(18)(19). It is expressed in a variety of normal human epithelial cell types, with expression levels higher in terminally differentiated than actively proliferating cells (19). In breast tissue, BTG2 expression ranged from moderate in ductal acini to low in myoepithelial cells, fibrous stroma, and adipose tissue (19). Estrogen and progesterone inhibited BTG2 expression, and BTG2 expression was highest during involution and remained steady during mammary differentiation (20). BTG2 expression was downregulated or lost during pregnancy and lactation, but recovered with cessation of lactation (20).
Immunohistochemical analyses of human cancers revealed down-regulation or loss of BTG2 in breast, prostate, hepatocellular, and renal cell carcinomas (20 -24). In estrogen receptorpositive breast cancers, loss of nuclear BTG2 expression was correlated with increased tumor size, higher tumor grade, and increased cyclin D1 expression (21). A recent study revealed that loss of BTG2 promotes mouse mammary tumor growth and metastasis, consistent with a role for BTG2 in mammary tumor progression (25).
Several studies suggest a link between stressful conditions and increased BTG2 expression (26 -28). Thus, BTG2 was strongly overexpressed in the pancreas, liver, and kidney during acute pancreatitis in rats (26,27). Following ischemic stroke, BTG2 was overexpressed in neurons within peri-infarct and infarct regions of brain but not in contralateral normal brain (28). These findings suggest a possible role for BTG2 in the response to stress. Herein, we document, for the first time, that BTG2 is a potent mediator of cytoprotection against oxidative stress, and we identify a potential mechanism for BTG2-mediated protection.
Expression Vectors, Transfections, and Small Interfering (si) RNA Treatments-Human wild-type BTG2 (wtBTG2) cDNA in the pcDNA3 vector was a gift from Dr. Saijun Fan (Georgetown University). Empty pcDNA3 vector was purchased from Invitrogen. For immunoprecipitation and chromatin immunoprecipitation experiments, we utilized BTG2 cDNA in the pCMV6 vector, which contains an NH 2 -terminal FLAG epitope tag (FLAG-BTG2) and the empty pCMV6 vector (OriGene Technologies, Rockville, MD). For domain mapping studies, we used expression vectors encoding HA-tagged human wildtype BTG2 (wt-HA-BTG2) and three HA-tagged deletion mutants: HA-BTG2-⌬A, HA-BTG2-⌬B, and HA-BTG2-⌬C, which were a gift from Dr. Shyamala Maheswaran (Massachusetts General Hospital, Boston, MA). These mutants are missing one of three conserved amino acid motifs designated box A, box B, and box C, as described previously (21) (see "Results"). Empty pCMV vector was purchased from Clontech. Expression vectors encoding wild-type nuclear factor erythroid 2-like 2 (wtNFE2L2) and dominant negative (DN)-NFE2L2 were described previously (30).
For transient transfections, subconfluent proliferating cells were transfected overnight using Lipofectamine (Invitrogen), with 2 g of plasmid DNA per 10-cm 2 well in 6-well dishes or 20 g of plasmid DNA per 100-mm dish. BTG2, BRCA1, NFE2L2, and control siRNAs were purchased from Thermo Scientific. Sequences for the siRNAs are provided in supplemental Table  1. For knockdown experiments, cells were pretreated with gene-specific or control siRNA (100 nM) for 48 -72 h, using the transfection reagent siPORT amine (Thermo Scientific).
MTT Dye Reduction Assays-The MTT assay measures the ability of viable mitochondria to reduce a tetrazolium dye to formazan. After the indicated cell treatments, subconfluent cells in 96-well dishes were assayed for MTT dye reduction, as described before (30). Each assay condition was tested in 10 replicate wells. Cell viability was calculated as the amount of dye conversion relative to sham-treated control cells and expressed as means Ϯ S.E. of three independent experiments.
Trypan Blue Dye Exclusion Assays-This assay measures the ability of viable cells with intact cell membranes to exclude trypan blue dye. After the indicated cell treatments, cells were assayed for trypan blue dye exclusion, as described earlier (30). Semiquantitative RT-PCR-Total cellular RNA was extracted using TRIzol reagent (Invitrogen), according to the manufacturer's instructions. The synthesis of first-strand cDNA template and the RT-PCR assays were conducted as described before (30,31). The PCR primers and reaction conditions are provided in supplemental Table 2. All assays were performed within the linear range of product amplification. PCR products were resolved on a 2% agarose gel containing ethidium bromide (0.1 mg/ml) and visualized under ultraviolet light using Quantity One 1-D Analysis software (Bio-Rad). Densitometry analyses were performed using ImageJ software (National Institutes of Health, Bethesda, MD). mRNA expression was normalized to ␤-actin and expressed as means Ϯ S.E. of -fold changes relative to control for three independent experiments.
Enzyme Activity Assays-Following the indicated transfections or siRNA treatments, cells were harvested and assayed using commercial kits to determine catalase (catalase assay kit, catalogue no. 707002, Cayman Chemicals, Ann Arbor, MI), total superoxide dismutase (SOD determination kit, catalogue no. 19160, Sigma), and total glutathione peroxidase (glutathione peroxidase assay kit, catalogue no. 353919, Calbiochem) activity. Enzyme activity values were normalized to the control values and expressed as -fold changes (mean Ϯ S.E.) for three independent experiments.
Assays of the Cellular Redox State-After the indicated transfections, cells were exposed to different concentrations of H 2 O 2 for 24 h and assayed for reduced (GSH) and oxidized (GSSG) glutathione using the glutathione assay kit (catalogue no. 703002, Cayman Chemicals). GSH/GSSG ratios were expressed as means Ϯ S.E. of three independent experiments.
Transcriptional Assays-NFE2L2 signaling through the antioxidant response element (ARE) was measured using the NQO1-ARE-Luc reporter plasmid, as described previously (30). The NQO1-ARE-Luc reporter contains the ARE of NQO1 (NAD(P)H dehydrogenase, quinone 1) driving a minimal promoter upstream of the luciferase gene. Briefly, subconfluent proliferating cells in 12-well dishes were transfected overnight with the indicated expression vector(s) (2 g/well), the NQO1-ARE-Luc or control (pGL3-Luc) reporter (0.2 g/well), and control plasmid pRSV-␤-gal (4 ng). The cells were then washed, postincubated for 24 h to allow luciferase expression, and assayed for luciferase activity using the luciferase assay system kit (catalogue no. E1501, Promega, Madison, WI). ␤-Galactosidase activity was determined as a control for transfection efficiency, using the ␤-galactosidase enzyme assay system (Promega, catalogue no. E2000). Total transfected DNA was kept constant by the addition of empty pcDNA3 vector. Luciferase values were normalized to ␤-gal and expressed as means Ϯ S.E. of the -fold change relative to a reporter-only control transfection, based on three independent experiments.
Immunoprecipitation (IP)-Cells were transfected overnight with FLAG-BTG2 or pCMV6 (20 g per 100-mm dish) using Lipofectamine. Cell lysates were prepared as described before (32), and aliquots of 500 g of protein were incubated overnight with 20 l of agarose A/G beads (Santa Cruz Biotechnology) and 2 g of anti-FLAG antibody (M2, mouse monoclonal, Sigma), anti-NFE2L2 (C-20, rabbit polyclonal, Santa Cruz Biotechnology), or normal rabbit or mouse IgG (Santa Cruz Biotechnology) (negative controls). The beads were washed three times with radioimmune precipitation assay buffer supple-mented with the protease inhibitors mixture set II (Calbiochem) and boiled with the Laemmli buffer (Invitrogen) containing 2% SDS at 99°C for 10 min to release the immune complexes. The lysate (10 -30 l) was then subjected to Western blotting using anti-FLAG or anti-NFE2L2 antibodies (see above).
For domain mapping studies, MCF-7 cells were transfected overnight with the indicated HA-tagged BTG2 expression vector (20 g per 100-mm dish) using Lipofectamine, after which IPs were carried out as described above, except that the IP antibody was HA probe (3 g) (F-7, mouse monoclonal, Santa Cruz Biotechnology) or normal mouse IgG (Santa Cruz Biotechnology) (negative control). The resultant lysates (30 l) containing immune complexes were subjected to Western blotting using anti-HA or anti-NFE2L2 antibodies (see above).
Chromatin Immunoprecipitation (ChIP) Assays-ChIP assays were performed using the ChIP-IT express magnetic chromatin immunoprecipitation kit (catalogue no. 53009, Active Motif, Carlsbad, CA), according to the manufacturer's instructions. After the indicated transfections and/or H 2 O 2 treatments, cells were harvested, and nuclear extracts were crosslinked using formaldehyde. For each assay, 5 g of total cell DNA was incubated with 7 g of antibody (as above or BTG2 (Q-22, rabbit polyclonal; Santa Cruz Biotechnology)) overnight at 4°C. The precipitated protein-DNA complexes were reverse cross-linked, and the DNA released was separated from the protein, cleaned, and purified using a QIAquick PCR purification kit (catalogue no. 28104, Qiagen, Valencia, CA). The purified DNA was used as a template for PCR amplification of a sequence spanning the ARE within the NQO1 gene promoter (bp Ϫ467 to Ϫ446) and heme oxygenase (HO-1) enhancer regions (E1 (bp Ϫ4045 to Ϫ4024) and E2 (bp Ϫ9048 to Ϫ9027)). The PCR primers and reaction conditions were described previously (33). PCR products were resolved on a 3% agarose gel containing ethidium bromide and visualized under ultraviolet light.
BrdU Assays-Subconfluent proliferating cells were subjected to the indicated transfections or siRNA treatments, after which they were treated with different concentrations of H 2 O 2 for 24 h. During the final 12 h of H 2 O 2 treatment, the cells were incubated with the BrdU. BrdU incorporation into DNA was determined using the BrdU cell proliferation assay kit (catalogue no. QIA58, Calbiochem), according to the manufacturer's instructions. The percentage of growth inhibition relative to vehicle-treated control cells was calculated and expressed as means Ϯ S.E. from three independent experiments.
Cell Cycle Analysis-Cells were transfected with wtBTG2 or empty pcDNA3 vector and then treated Ϯ 500 nmol/liter H 2 O 2 for 24 h. Cells were harvested using trypsin, washed with PBS, and fixed in cold 70% ethanol. The samples were treated with RNase A, stained with propidium iodide (100 mg/ml), and analyzed by FACSort (BD Biosciences), using ModFit software (Verity Softwarehouse, Topsham, ME). At least 20,000 events were analyzed. Each assay condition was tested in three independent experiments.
Statistical Methods-Statistical comparisons were conducted using the two-tailed Student's t test.

BTG2 Protects against Oxidative Stress in a BRCA1-independent
Manner-To test the ability of BTG2 to protect against oxidative stress, MCF-7 breast cancer cells, which have low to undetectable BTG2 protein (20), were transiently transfected with wtBTG2 or empty pcDNA3 vector, exposed to different concentrations of H 2 O 2 for 24 h, and assayed for cell viability using MTT assays. wtBTG2-transfected cells showed significantly higher survival than empty vector-transfected or vehicletreated control cells at all H 2 O 2 doses tested (p Ͻ 0.05, twotailed t tests) (Fig. 1A). Similar results were observed using T47D breast cancer cells (supplemental Fig. 1A) and using trypan blue dye exclusion assays instead of MTT assays (Fig. 1B), suggesting that the findings are not limited to one cell line or an artifact of the MTT assay. Protection of MCF-7 cells was also observed using two different oxidants, paraquat and nickel acetate (supplemental Fig. 1, B and C), suggesting that protection is not limited to H 2 O 2 .
To determine the effect of decreased BTG2 on cell survival, we used a nontumor mammary epithelial cell line that expresses easily detectable BTG2 protein levels (184A1). Here, cells treated with BTG2-siRNA showed a significant reduction in survival at all H 2 O 2 doses as compared with control (CON)-siRNA-or vehicle-treated cells (p Ͻ 0.05) (Fig. 1C). Studies in another nontumor human mammary epithelial cell line (MCF-10A) revealed that overexpression of wtBTG2 conferred resistance to H 2 O 2 , whereas knockdown of endogenous BTG2 caused increased sensitivity to H 2 O 2 (p Ͻ 0.05 at each H 2 O 2 dose) (supplemental Fig. 1, D and E, respectively).
As we previously showed that BRCA1 protects cells against oxidative stress (30), we tested whether BTG2-mediated protection is attributable to BRCA1. As expected, BRCA1-siRNA caused a reduction in cell survival in empty vector-transfected cells, but in wtBTG2-transfected cells, BRCA1-siRNA did not alter survival of H 2 O 2 -treated cells ( Fig. 1D and supplemental Fig. 2A). Furthermore, wtBTG2 protected HCC1937, a human breast cancer cell line that harbors a single mutant BRCA1 allele (5382insC) (34), to the same degree that it protected MCF-7 and T47D against H 2 O 2 (supplemental Fig. 2B), suggesting that BTG2 protection against oxidative stress is BRCA1-independent. Western blots confirming overexpression or underexpression of BTG2 or BRCA1 due to the transfections or siRNA treatments of MCF-7, T47D, and MCF-10A cells are shown in supplemental Fig. 3, A-F. BTG2 Protection against H 2 O 2 Requires NFE2L2-NFE2L2 (also called NRF2) is a basic leucine zipper transcription factor that mediates the cytoprotective cellular antioxidant response (35). To determine whether NFE2L2 is required for protection against H 2 O 2 by BTG2, cells were co-transfected with wtBTG2 and a DN-NFE2L2 expression vector and assayed for sensitivity to H 2 O 2 . In MCF-7 (Fig. 1E) and T47D (supplemental Fig. 2C) cells, DN-NFE2L2 somewhat reduced the survival of controltransfected cells and abolished protection due to wtBTG2. Similar to the results obtained using DN-NFE2L2, knockdown of NFE2L2 using siRNA caused a moderate reduction in survival of control-transfected cells but abolished the protective effect of wtBTG2 (Fig. 1F). Western blots confirming expression of DN-NFE2L2 in transfected cells and the knockdown of NFE2L2 in siRNA-treated cells are shown in supplemental Fig. 3, G-I. These findings suggest that BTG2 protection against oxidative stress requires functional NFE2L2.
BTG2 Stimulates Expression of Antioxidant Proteins in an NFE2L2-dependent Manner-To better understand the mechanism of BTG2 protection against oxidative stress, we tested whether BTG2 could stimulate expression and/or activity of several antioxidant proteins. MCF-7 cells transfected with wtBTG2 showed increased expression of catalase, superoxide dismutases 1 and 2 (SOD1 and SOD2), and BRCA1 mRNA ( Fig.  2A) and protein (Fig. 2B). Conversely, 184A1 cells treated with BTG2-siRNA showed decreased expression of catalase, SOD1, SOD2, and BRCA1 mRNA and protein (Fig. 2, C and D). To determine whether NFE2L2 is required for BTG-mediated stimulation of antioxidant gene expression, MCF-7 cells were both transfected with BTG2 and treated with NFE2L2-siRNA. As shown in Fig. 2E, knockdown of NFE2L2 reduced the basal protein expression of catalase, SOD1, and SOD2 and blocked or attenuated the BTG2 stimulation of antioxidant protein expression. Similar results were observed when MCF-7 cells were co-transfected with wtBTG2 and DN-NFE2L2 to block endogenous NFE2L2 activity (supplemental Fig. 4A). Densitometry results based on three independent experiments confirmed that the changes in mRNA and protein levels due to over-and underexpression of BTG2 were significant (p Ͻ 0.05) (supplemental Figs. 4B and 5, A-E). Similarly, BTG2-induced changes in antioxidant gene expression were observed in T47D cells (supplemental Fig. 6).
Consistent with these findings, measurements of catalase and total SOD enzymatic activity revealed about 2-fold increases due to BTG2 overexpression (Fig. 3, A and B, and supplemental Fig. 7, A and B) and 2.5-fold decreases in activity due to BTG2 underexpression (Fig. 3, C and D). BTG2 overexpression also caused increases in glutathione peroxidase enzyme activity (Fig. 3E and supplemental Fig. 7C), although these increases (ϳ1.5-fold) were smaller than those for catalase and SOD.
Effect of BTG2 on Cellular Redox State-We tested the effect of BTG2 on redox status, indicated by the ratio of reduced to oxidized glutathione (GSH/GSSG). wtBTG2-transfected MCF-7 cells showed a similar basal GSH/GSSG ratio to control-transfected or untransfected cells (Fig. 3F). However, the wtBTG2-transfected cells showed significantly higher GSH/ GSSG ratios after treatment with H 2 O 2 (p Ͻ 0.05). Similar results were observed using T47D cells (supplemental Fig. 7D), suggesting that BTG2 can attenuate the loss of GSH in response to oxidative stress.
BTG2 Regulates NFE2L2 Signaling via Antioxidant Response Element-NFE2L2 stimulates cellular antioxidant defenses by up-regulating transcription of genes containing AREs in their regulatory regions (35). It can regulate the expression of catalase, SOD1, SOD2, glutathione synthase, NQO1, HO-1, and various other antioxidant genes (35). We investigated the role of BTG2 in NFE2L2/ARE signaling using an NFE2L2-responsive reporter driven by the ARE of the NQO1 promoter (NQO1-ARE-Luc). In MCF-7 cells, wtBTG2 enhanced wtNFE2L2-induced reporter activity by nearly 2-fold as compared with empty pcDNA3 vector (p Ͻ 0.05) (Fig. 4A). By itself (i.e. in the absence of wtNFE2L2 transfection), wtBTG2 increased reporter activity by about 1.6-fold (p Ͻ 0.05) (Fig. 4B). Expression of DN-NFE2L2 significantly reduced basal NQO1-ARE-Luc activity and abolished the ability of wtBTG2 to stimulate NQO1-ARE-Luc activity. Similar results were observed using T47D cells (supplemental Fig. 8). These results suggest that BTG2 can stimulate NFE2L2 signaling through the ARE.    Fig. 1D and supplemental Fig. 2A), BRCA1-siRNA caused down-regulation of catalase and SOD enzyme activity (p Ͻ 0.05) in MCF-7 and T47D cells (supplemental Fig. 9). However, BRCA1siRNA did not block the wtBTG2-induced stimulation of enzyme activity. Furthermore, BTG2-mediated stimulation of antioxidant enzyme expression and activity was observed in BRCA1-deficient HCC1937 cells and was similar in magnitude to that observed in MCF-7 and T47D cells (supplemental Fig.  10). Expression of DN-NFE2L2 also reduced basal catalase and SOD enzyme activity in MCF-7 cells, but co-expression of wtBTG2 did not substantially increase enzyme activity in NFE2L2-transfected cells (supplemental Fig. 11). These findings suggest that although basal catalase and SOD activity are regulated by BRCA1 and NFE2L2, stimulation of enzyme activity by BTG2 requires NFE2L2 but not BRCA1.

NFE2L2 (but Not BRCA1) Is Required for BTG2 Stimulation of Catalase and SOD Activity-Consistent with the MTT assays (
BTG2 Associates with Endogenous NFE2L2 Protein-Because NFE2L2 transcriptional activity is regulated by heterodimerization with several transcriptional co-regulators (see "Discussion"), we tested whether endogenous NFE2L2 can interact with an exogenous FLAG-tagged BTG2 protein in MCF-7 (Fig.  5) and T47D (supplemental Fig. 12) cells. An anti-FLAG IP of cells transfected with a FLAG-BTG2 expression vector co-precipitated NFE2L2, and vice versa. As negative controls, FLAG IP of empty pCMV6-FLAG vector-transfected cells did not yield FLAG-BTG2 or NFE2L2, and IPs performed using the appropriate nonimmune IgG gave no precipitated bands. These results suggest that the BTG2 and NFE2L2 proteins can associate in vivo.
BTG2 and NFE2L2 Are Present at the ARE-We performed ChIP assays to determine whether BTG2 can localize to ARE sites in NFE2L2 target genes, using primers that span the AREs of NQO1 and HO-1. In a control study, an NFE2L2 IP of untransfected MCF-7 cells revealed the presence of NFE2L2 at two known ARE sites of HO-1 (E1 and E2) and at an ARE in the NQO1 promoter (Fig. 6A). As negative controls, no NFE2L2 was detected at HO-1 exon 3 or NQO1 exon 2, which do not contain AREs, and an IP performed using nonimmune IgG showed no PCR bands. In Fig. 6B, MCF-7 cells were transfected with FLAG2-BTG2 or empty pCMV6-FLAG vector and subjected to an anti-FLAG IP. FLAG-BTG2 protein was found at all three AREs but not at control sites. Cells transfected with FLAG2-BTG2 and subjected to a control mouse IgG IP revealed no PCR bands. Input lanes corresponding to 10% of the amount of DNA used for ChIP assays are also provided. Similar results were observed in T47D cells (supplemental Fig. 13).
Next, we tested the effect of exposure to H 2 O 2 for different times on the abundance of FLAG-BTG2 and endogenous NFE2L2 at ARE sites, in FLAG-BTG2-transfected cells. Fig. 6C shows the input lanes for these experiments. Time-dependent increases in FLAG-BTG2 (Fig. 6D) and NFE2L2 (Fig. 6E) were observed at all three ARE sites, but no FLAG-BTG2 or NFE2L2 was detected at the control sites. Nearly maximal occupation of the ARE sites by BTG2 and NFE2L2 occurred by 8 h of H 2 O 2 exposure. These findings suggest that exogenous BTG2 and endogenous NFE2L2 co-localize at ARE sites, and the kinetics of recruitment of these two proteins to the ARE in response to oxidative stress are similar.
Finally, we tested for the presence of endogenous BTG2 at ARE sites using MCF-10A cells, which express easily detectable levels of the BTG2 protein. As shown in supplemental Fig. 14, A  and B, endogenous BTG2 can be detected at three different AREs under basal conditions, and exposure to H 2 O 2 caused time-dependent increases in BTG2 occupancy at these sites. A similar pattern of time-dependent recruitment of NFE2L2 to the ARE sites in response to H 2 O 2 was observed (supplemental Fig. 14C). Near maximal recruitment of BTG2 and NFE2L2 was observed after 8 h of exposure to H 2 O 2 . These findings suggest that endogenous BTG2 and NFE2L2 co-localize at ARE sites of NFE2L2-responsive genes in response to oxidative stress and that the kinetics of their recruitment are similar.
Effect of BTG2 on Proliferation and Cell Cycle Distribution-As BTG2 is known to inhibit cell proliferation (1, 2), we tested the effect of BTG2 overexpression on proliferation, using BrdU incorporation as a measure of the cell growth state. In untreated MCF-7  SEPTEMBER 7, 2012 • VOLUME 287 • NUMBER 37 cells, wtBTG2 caused significant growth inhibition (62%), relative to control-transfected or untransfected cells (Fig. 7A). Exposure to H 2 O 2 caused dose-dependent growth inhibition in control-and wtBTG2-transfected cells, but the degree of inhibition was greater in wtBTG2-transfected cells at all H 2 O 2 doses (p Ͻ 0.05). Conversely, knockdown of endogenous BTG2 in MCF-10A cells significantly attenuated the degree of H 2 O 2 -induced growth inhibition at each concentration of H 2 O 2 (Fig. 7B) (p Ͻ 0.05). These findings suggest that the endogenous BTG2 protein contributes to the growth inhibition observed in cells exposed to H 2 O 2 .

BTG2 Stimulates Antioxidant Defenses
We also tested the effect of BTG2 on the cell cycle distribution, determined by flow cytometry. In untreated MCF-7 cells, wtBTG2 caused a small but significant reduction in the percentage of S-phase cells and a corresponding increase in the percentage of G 0 /G 1 cells (Fig. 7C). In cells exposed to H 2 O 2 for 24 h, wtBTG2 caused a 20% reduction in S-phase cells with a corresponding increase in G 2 /M (Fig. 7D). Consistent with the protective effect of BTG2, wtBTG2-transfected cells also showed a reduction in the percentage of pre-G 1 (apoptotic) cells. Representative flow cytometry histograms are provided in supplemental Fig. 15. These results suggest antiproliferative and antiapoptotic effects of BTG2.
Box B of BTG2 Is Required for the Association of BTG2 with NFE2L2-Boxes A, B, and C are short amino acid sequences characteristic of the BTG/TOB family proteins (21). Boxes A and B are present in all family members, whereas box C is present in only BTG1 and BTG2. Here, we sought to test the requirement for these conserved sequences by the use of expression vectors encoding wild-type BTG2 and three mutant proteins each deleted of a different box and each with an NH 2terminal HA tag (21). The proteins encoded by these vectors are illustrated schematically in Fig. 8A. All four HA-tagged proteins were well expressed in transfected MCF-7 cells (supplemental Fig. 16A). Based on IP-Western blotting experiments utilizing an anti-HA antibody for IPs, endogenous NFE2L2 co-precipitated with the wt-HA-BTG2, HA-BTG2-⌬A, and HA-BTG2-⌬C proteins but failed to co-precipitate with HA-BTG2-⌬B (Fig. 8, B-E). As negative controls, IPs FIGURE 6. BTG2 and NFE2L2 are present at the ARE. A, untransfected MCF-7 cells were subjected to ChIP assays using anti-NFE2L2 antibody or nonimmune rabbit IgG (negative control). The PCR primers correspond to two different AREs (E1 and E2) of the HO-1 gene, HO-1 exon 3 (negative control), an ARE from the NQO1 gene promoter (NQO1 pr), and NQO1 exon 2 (negative control). B, MCF-7 cells were transfected with FLAG-BTG2 or empty pCMV6-FLAG vector, and ChIP assays were performed using anti-FLAG or nonimmune mouse IgG (negative control). C-E, MCF-7 cells were transfected with FLAG-BTG2; exposed to H 2 O 2 (250 nmol/liter) for the indicated times; and harvested for ChIP assays. Panel C shows input DNA, whereas panels D and E show ChIP assays using anti-FLAG (D) or anti-NFE2L2 (E) antibody. These data are representative of three independent experiments.  performed with nonimmune mouse IgG did not yield any HA-BTG2 or NFE2L2 bands, and empty vector-transfected cells showed no HA-BTG2 protein band. These findings suggest that box B is required for the in vivo association of BTG2 with NFE2L2 but that boxes A and C are dispensable for this function.
BTG2-mediated Antioxidant Response Functions Require Box B-To test the significance of the finding that a BTG2 protein deleted of box B does not associate with NFE2L2, MCF-7 cells were transfected with each HA-BTG2 expression vector or empty vector and tested for protection against H 2 O 2 using MTT assays. Here, we found that cells transfected with wt-HA-BTG2 were significantly protected against H 2 O 2 relative to empty vector-transfected cells (p Ͻ 0.05 at each H 2 O 2 concentration tested), and the survival levels among cells transfected with wt-HA-BTG2, HA-BTG2-⌬A, or HA-BTG2-⌬C were similar (Fig. 8F). On the other hand, MCF-7 cells transfected with HA-BTG2-⌬B showed little or no protection relative to empty vector-transfected cells at any concentration of H 2 O 2 . Finally, we found that transfection of MCF-7 cells with wt-HA-BTG2, HA-BTG2-⌬A, or HA-BTG2-⌬C exhibited higher levels of antioxidant proteins (catalase, SOD1, and SOD2) relative to empty vector-transfected cells, whereas cells transfected with HA-BTG2-⌬B showed similar levels of antioxidant proteins to empty vector-transfected cells (supplemental Fig. 16). A representative Western blot is shown in supplemental Fig.  16A, and densitometry results based on three independent experiments are provided in supplemental Fig. 16B. These findings suggest that the antioxidant-related functionality of BTG2 requires box B but does not require box A or box C.

DISCUSSION
Here, we have demonstrated that BTG2 mediates protection of human mammary epithelial cells against oxidative stress due to H 2 O 2 and other oxidizing agents. In this regard, BTG2-mediated cell protection requires functional NFE2L2 (but not BRCA1) and may be due, in part, to the ability of BTG2 to function as a co-activator for antioxidant transcription factor NFE2L2. The cytoprotective activity of BTG2 against oxidative stress and other antioxidant-related functions were demonstrated in both breast cancer cell lines (MCF-7, T47D, and HCC1937) and nontumor human mammary epithelial cell lines (184A1 and MCF-10A). In the latter two cell lines, knockdown of endogenous BTG2 sensitized the cells to oxidative stress, suggesting that endogenous BTG2 contributes to basal resistance to oxidative stress.
BTG2 was found to up-regulate expression of antioxidant enzymes known to be regulated by NFE2L2, including catalase, SOD1, and SOD2 (35), and it increased enzyme activity to a similar degree, suggesting that the increases in activity are due to increased protein levels. Inhibition of NFE2L2 by expression of a dominant negative protein or by knockdown using siRNA abolished or attenuated BTG2-stimulated cell protection and enzyme expression but also reduced cell survival in the absence of exogenous BTG2, suggesting that NFE2L2 functions through BTG2-dependent and -independent pathways. The ability of BTG2 to stimulate NRF2/ARE signaling, physically associate with NFE2L2, and localize to ARE sites of NFE2L2-responsive genes (HO-1 and NQO1) suggest that BTG2 acts as a transcriptional co-activator for NFE2L2. It should be emphasized that our studies do not rule out the possibility of other non-NFE2L2-related mechanisms of BTG2 protection.
In this regard, it was previously reported that BTG2 may function as a binding partner and co-regulator for several other transcription factors or regulatory proteins. Thus, BTG2 acts as a co-activator for HoxB9 (36), a transcription factor that regulates normal development, including mammary development and differentiation (37). BTG2 also binds to and modulates the activity of PRMT1, a histone H4-specific methyltransferase that participates in a variety of transcriptional pathways (38). Also, BTG2 can bind to transcriptional co-regulators CCR4associated factor 1 (CAF1) (4) and repressor of estrogen receptor activity (REA) and regulate estrogen receptor transcriptional activity (39).
As noted earlier, the cellular response to oxidative stress is mediated, in part, through activation of NFE2L2, which stimulates expression of antioxidant and detoxification proteins (33,35,40). NFE2L2 activation is due to its release from a cytoplasmic inhibitor (KEAP1), nuclear translocation, and formation of heterodimers with various binding partners (e.g. musculoaponeurotic fibrosarcoma oncogene homolog (MAF) proteins, JUN proteins, ATF2/4) at ARE sites. Our findings suggest that BTG2 may also serve such a function. Interestingly, NFE2L2 has been identified as a potential target for cancer prevention, based on studies showing that: 1) NFE2L2 knock-out mice are more sensitive to colitis-induced colon cancer and carcinogeninduced mammary cancer (41,42); and 2) various proposed cancer prevention agents (e.g. diindolylmethane and sulforaphane) stimulate NFE2L2 signaling (29,(43)(44)(45)). It would be interesting to determine whether these agents stimulate BTG2 expression as a potential mechanism for stimulating NFE2L2 activity.
Studies utilizing a set of deletion mutants revealed that several antioxidant-related functions of BTG2 (protection against H 2 O 2 , association with the NFE2L2 protein, stimulation of antioxidant gene expression) were found to be dependent upon the presence of box B, a highly conserved amino acid sequence within the BTG/TOB protein family. However, these functions did not require boxes A or C. Interestingly, all three boxes were found to be required for or to contribute to the growth inhibitory function of BTG2. The fact that box B contributes to the antioxidant function of BTG2 suggests that this function may be conserved among the different members of the BTG/TOB protein family as is the growth inhibitory function, although this hypothesis remains to be proven.
It was noted in the Introduction that BTG2 expression is frequently down-regulated or lost in various types of human cancers, including breast, raising the possibility that BTG2 functions as a tumor suppressor. Interestingly, recent studies have revealed that BTG2 is one of a small number of genes that differs in both copy number and gene expression in node-negative breast cancers from long-term (10-year) cancer survivors and deceased patients, and up-regulation of BTG2 protein expression was found to correlate with increased patient survival (46,47). Importantly, our findings raise the possibility that the proposed tumor suppressor activity of BTG2 may be due not only to its function as a p53-inducible antiproliferative gene but also to its ability to protect cells against oxidative stress. In this regard, oxidative free radicals may produce a variety of DNA lesions, some of which are cytotoxic and others of which are not cytotoxic but may be mutagenic and, ultimately, carcinogenic, if left unrepaired (e.g. by the base excision repair process) (48). A gene that protects against oxidative stress by stimulating antioxidant defenses to detoxify free radicals may be potentially anticarcinogenic. Thus, the ability of BTG2 to stimulate the expression of enzymes that detoxify oxidative radicals (e.g. catalase and superoxide dismutases) may contribute to a tumor suppressor function.