Human primary brain astrocytes were purchased from the Applied Cell Biology Research Institute (Kirkland, WA, USA) and were cultured in DMEM supplemented with 10% FBS (Invitrogen, Carlsbad, CA, USA). Mouse primary cerebral astrocytes were purified from neonatal ICR mice according to standard procedure (⁹). Mouse embryonic fibroblasts (MEFs) were isolated from Nrf2^+/+ and Nrf2^?/? mice (which were generously provided by Dr. Jeffrey A. Johnson, University of Wisconsin, WI, USA). Tert-butylhydroquinone (tBHQ) and butylated hydroxyanisole (BHA) were purchased from Fisher Scientific (Fairlawn, NY, USA) and sodium arsenite (AS) and phenazine methosulfate (PMS) were purchased from Sigma (St Louis, MO, USA).

Equal amounts of proteins were separated on 11% SDS?PAGE gels, transferred onto nitrocellulose membranes, and incubated with antibodies specific for Nrf2, ATF3, HO1 and ?-actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA). The expressions of ATF3 and ?-actin were analyzed quantitatively using an image analysis software (NIH-Image J).

Cells were washed with phosphate buffered saline (PBS), fixed with 3.7% formaldehyde, and permeabilized with 0.5% Triton X-100 in PBS. After blocking with 1% blocking reagent (Roche) for 1 h, cells were incubated with the anti-ATF3 antibody overnight at 4?C, followed by incubation with an FITC-conjugated secondary antibody. Immunocytochemical analysis using the anti-phospho-histone H2A.X antibody (ser139) (Millipore, Billeria, MA, USA) was performed according to the manufacturer?s instructions. Human astrocytes were fixed with 5% acetic acid in ethanol for 5 min. After blocking with 3% BSA for 30 min, cells were incubated with anti-phospho-histone H2A.X antibody for 1 h, followed by incubation with an FITC-conjugated secondary antibody. After counterstaining with 2 ?g/ml of 4?,6?-diamidine-2-phenylindole dihydrochloride (DAPI) (Roche), images were acquired using an Axiovert M200 microscope (Zeiss, Oberkochen, Germany).

Total RNA was isolated from astrocytes using the TRIzol reagent (Molecular Research Center, Cincinati, OH, USA). cDNAs were synthesized using an oligo-(dT) primer and MMLV reverse transcriptase (Promega, Madison, WI, USA) and were used in subsequent PCR reaction using specific primer sets for human ATF3 (5?-ATGATGCTTCAACACCCAGG-3? and 5?-TTTCGGCACTTTGCAGCTG-3?), human Nrf2 (5?-AAACCAGTGGATCTGCCAAC-3? and 5?-GACCGGGAATATCAGGAACA-3?) (¹⁰), human HO1 (5?-GCTACCTGGGTGACCTGTCT-3? and 5?-GGGCAGAATCTTGCACTTTG-3?), human Sulfiredoxin1 (5?-CAAGGTGCAGAGCCTCGTGG-3? and 5?-AGCTGGTGAGAGGGGTGCCC-3?), human NQO1 (5?-CATTCTGAAAGGCTGGTTTG-3? and 5?-GGCTGCTTGGAGCAAAATAC-3?), and human ?-actin (5?-GACTACCTCATGAAGATC-3? and 5?-GATCCACATCTGCTGGAA-3?). PCR products were resolved on 1.5% agarose gels and visualized by ethidium bromide staining. Quantitative real-time PCR was carried out using commercially available SYBR green-based detection reagents (Roche, Branchburg, NJ, USA) and a Light Cycler apparatus (Roche).

siRNA duplexes targeting human Nrf2 (siNrf2, 5?-AAGAGUAUGAGCUGGAAAAAC-3?) (¹¹), human ATF3 (siATF3#1, 5?-UGAAGAAGAUGAAAGGAAA-3?, siATF3#2, 5?-GAAACAAGAAGAAGGAGAA-3?), and GFP (siGFP, 5?-GGCUACGTCCAGGAGCGCA-3?) were synthesized by Dharmacon (Lafayette, CO, USA) or Samchully Pharm (Seoul, Korea). siRNAs were transfected using Oligofectamine (Invitrogen).

ATF3 promoter regions were amplified from human genomic DNA by PCR using primer sets corresponding to ?3646 (5?-CGGGGTACCACCACTACTGAGGGCTGACTGG-3?), ?3150 (5?-CGGGGTACCACCACTACTGAGGGCTGACTGG-3?), 1250 (5?-CGGGGTACCAGGTCTCAGCTCGAATCTCGG-3?) and +3 (5?-CCCAAGCTTACTGTGGCTTGGAGAGCGTTG-3?) positions. PCR products were cloned into the pGL3-Basic vector. For construction of pATF3-ARE1-SV40 and pATF3-ARE2-SV40, fragments of approximately 200 bps that encompassed the putative ARE of the ATF3 promoter were amplified using the following primers: for pATF3-ARE1-luc, 5?-CGGGGTACCGCCACATCCTCAGAGAAATGTCAC-3? (forward) and 5?-GGAAGATCTATTGTCAGGAAACAGGGGCAG-3? (reverse); for pATF3-ARE2-luc, 5?-CGGGGTACCTAGATTTGGTTGCCACGAGTGG-3? (forward) and 5?-GGAAGATCTGCAGAGACTTGAAGATTGCTGAGG-3? (reverse). These PCR products were cloned into the pGL3-Promoter vector upstream of the SV40 promoter. Site-directed mutagenesis was performed by over-lapping PCR using the following primers: 5?-AAATAACTTACTTAAGGGTACACATAATCTAAGTGCTGAATC-3?, and reverse-complement primers (the mutated sequences were underlined). Mouse wild-type Nrf2, dominant-negative Nrf2 (Nrf2-DN, amino acids 296?581), and mouse Keap1 were amplified and cloned into the pCS2+ expression vectors.

HeLa cells and human astrocytes were transfected using Lipofectamine Plus (Invitrogen). All cells were cotransfected with the ?-galactosidase expression vector (pSV- ?-galactosidase). Luciferase activity was measured using the Luciferase assay kit (Promega) and ?-galactosidase activity was measured using O-nitrophenyl ?-galactopyranoside.

Chromatin immunoprecipitation (ChIP) was carried out using the ChIP kit (Upstate Biotechnology, Lake Placid, NY, USA). ARE of the ATF3 promoter was amplified by PCR using the following primers: for 1st PCR of ARE1, 5?-TACCACTACTGAGGGCTGACTGG-3?, and 5?-CATTGTCAGGAAACAGGGGC-3?; for 2nd PCR of ARE1, 5?-GTGCCACATCCTCAGAGAAATG-3?, and 5?-CACAGTAGGCACCAAAGTGTGG-3?; for PCR of ARE2, 5?-TAGATTTGGTTGCCACGAGTGG-3? and 5?-GCAGAGACTTGAAGATTGCTGAGG-3?. Primer pairs corresponding to the ARE of the NQO1 promoter were used as a positive control (¹²).

To assess the generation of ROS, human primary astrocytes were incubated with 5 ?M 2?,7?-dichlorodihydrofluorescein diacetate (DCFH-DA, Sigma) for 30 min at 37?C. Astrocytes were analyzed on a FACScan flow cytometer (Becton Dickinson, Bedford, MA, USA) and dichlorofluorescein (DCF) fluorescence was derived using the CellQuest? software. The Glutathione assay kit (Cayman Chemical, Ann Arbor, MI, USA) was used to measure total glutathione levels. Total glutathione [reduced glutathione (GSH) + oxidized glutathione (GSSG)] levels were measured at 405 nm, according to the manufacturer?s instructions.

Human astrocytes were treated with oxidants. After 4 h, we examined the cellular reduction of 4-[3-(4-iodophenyl)-2-(4-nitrophenyl)-2H-5-tetrazolio]-1,3-benzene disulfonate (WST-1) to formazan using the EZ-Cytox Enhanced cell viability assay kit (ITSBIO, Seoul, Korea). Cells were incubated in a culture medium containing 10% WST-1 for 1 h and the absorbance was measured at 450 nm.

To detect apoptotic cells, TdT-mediated dUTP nick and labeling (TUNEL) assay was performed using an in situ cell detection kit according to the manufacturer?s instructions (Roche). After fixation and permeabilization, human astrocytes were incubated with TUNEL reaction mixtures for 1 h. After counterstaining with DAPI, apoptotic cells were observed using an Axiovert M200 microscope (Zeiss). The percentage of apoptotic cells was calculated as the number of TUNEL-positive apoptotic nuclei divided by the total number of nuclei stained with DAPI.

Dissociated cells were mixed with low-melting agarose, dropped onto a glass slide, and dried. The slide was placed in lysis buffer for 2 h and electrophoresed at 12 V for 20 min. The slides were stained with 2 ?g/ml ethidium bromide. The DNA was then visualized using a fluorescence microscope (Leica CTR5000, Cambridge, UK). The DNA damage in about 35 cells were analyzed using the Comet Assay ? program (Perceptive Instruments, Suffolk, UK).

All data are presented as mean ? standard error of the mean (SEM) of three independent experiments. Significance of differences was evaluated by unpaired Student?s t-test. ^NSP > 0.05, *P < 0.05, **P < 0.01.

To evaluate the role of Nrf2 in the regulation of ATF3 expression, we treated human and mouse primary astrocytes with the Nrf2 activators, tBHQ and BHA. Within the indicated concentration range, no apparent cytotoxicity was observed (data not shown). Immunoblotting results showed that tBHQ and BHA provoked a dose-dependent induction of ATF3 as well as heme oxygenase 1 (HO1) which is an Nrf2-target protein (Figure 1A). This result was confirmed by immunocytochemical analysis, which showed that the induced ATF3 was mainly localized in the nuclei of tBHQ-treated human astrocytes (Figure 1B). In contrast, unstimulated cells exhibited background staining. Moreover, RT?PCR and the quantitative real-time PCR analyses revealed that both tBHQ and BHA increased ATF3 mRNA levels significantly in human astrocytes (Figure 1C and D), indicating that the induction of ATF3 was caused by the upregulation of transcription.

To further confirm that Nrf2 participated in tBHQ-induced ATF3 expression, we used siRNA targeting Nrf2 (siNrf2). RT?PCR analysis showed that transfection of human astrocytes with siNrf2 reduced the levels of Nrf2 mRNA expression (Figure 2A). As shown in Figure 2B, the depletion of Nrf2 in human astrocytes resulted in the reduction of ATF3 and HO1 induction. Additionally, when MEFs isolated from Nrf2^+/+ and Nrf2^?/? mice were cultured in the presence tBHQ, ATF3 and HO1 induction were considerably reduced in Nrf2-deficient MEFs (Figure 2C).These results suggest that Nrf2 plays an essential role in ATF3 induction after treatment with tBHQ.

It is well known that oxidants induce the expression of Nrf2 target genes through the stabilization of Nrf2. To determine whether ATF3 was regulated by oxidants via Nrf2, human astrocytes were treated with the oxidants, sodium arsenite (AS) and phenazine methosulfate (PMS) which generate ROS inside cells. Figure 3A shows that both agents upregulated ATF3 and HO1 inductions in human astrocytes. Next, we determined whether Nrf2 was critical for oxidant-induced ATF3 expression. The depletion of Nrf2 in astrocytes reduced ATF3 and HO1 expression induced by treatment with AS or PMS (Figure 3B). Similarly, ATF3 expression was not induced by oxidants in Nrf2^?/? MEFs (Figure 3C). The reduction of ATF3 induction in siNrf2-transfected astrocytes (Figure 3B) was less drastic than that observed in Nrf2^?/? MEFs (Figure 3C). We suggest that, because transfection of human astrocytes with siNrf2 did not fully abolish the expression of Nrf2 mRNA, the reduction of ATF3 induction may be insufficient. Moreover, it is possible that ATF3 induction by oxidant treatment will be regulated by several pathways including the Nrf2/Keap1 pathway. Nevertheless, our findings showed that Nrf2 was one of the major pathways for the transcription of the ATF3 gene.

Nrf2 binds to ARE in promoter regions for the transcriptional activation of target genes. To determine whether the ATF3 promoter contains an ARE or not, we analyzed about 5-kb upstream region of the human ATF3 promoter using MOTIF search software. Thereafter, we found two potential ARE sites, termed ARE1 and ARE2, which were located from ?3541 to ?3533 and from ?2975 to ?2967, respectively (Figure 4A). ARE1 exhibited perfect match with the consensus ARE sequence, while ARE2 contained two mismatched bases.

To verify whether these two putative ARE sites were functional for Nrf2, the ATF3 promoter encompassing both ARE sites was cloned into the luciferase reporter vector (pATF3-3.6k-luc). Luciferase activity in tBHQ-treated human astrocytes was almost twice as high as that measured in non-treated cells (Figure 4B). In addition, luciferase activity was also increased in proportion to increasing doses of exogenous Nrf2 (Figure 4C), whereas Nrf2-DN lacking a transactivating domain strongly suppressed the luciferase activity that was increased by the overexpression of Nrf2 (Figure 4D). Keap1 prevents Nrf2 activation by sequestering Nrf2 in the cytoplasm. Consequently, the overexpression of Keap1 almost completely repressed the transcriptional activity increased by the overexpression of Nrf2 (Figure 4E). These results indicate that the ATF3 promoter contains Nrf2-responsive element(s).

Next, we investigated whether the two ARE sites of the ATF3 promoter were important for Nrf2-mediated activation. Figure 5A shows that the overexpression of Nrf2 failed to induce the increase in transcriptional activity of pATF3-1.3k-luc construct lacking both ARE sites and pATF3-3.1k-luc construct containing only ARE2, whereas Nrf2 increased the transcriptional activity of pATF3-3.6k-luc construct containing both ARE sites, suggesting that ARE1 may be important for Nrf2 mediation. The mutated ARE1 markedly reduced the Nrf2-mediated activation of the ATF3 promoter (Figure 5B). In addition, we showed that tBHQ treatment did not increase luciferase activity in human astrocytes transfected with mutant type ATF3 promoter (pATF3-3.6k-luc-mut) (Figure 5C). These results suggest that ARE1 was essential for the Nrf2-dependent transcriptional activation of the ATF3 gene. This finding was further confirmed in experiments using a luciferase-construct containing either ARE1 or ARE2 (pATF3-ARE1-luc or pATF3-ARE2-luc, respectively). The pATF3-ARE1-luc construct containing only ARE1 exhibited the Nrf2-mediated increase in luciferase activity, whereas the pATF3-ARE2-luc construct containing only ARE2 did not (Figure 5D). In addition, the site-specific mutation of the ARE1 sequence in the pATF3-ARE1-luc construct (pATF3-ARE1-luc-mut) completely abolished the Nrf2-mediated increase in luciferase activity (Figure 5D). Finally, we evaluated whether Nrf2 was able to bind ARE1 in vivo. ChIP analysis revealed that tBHQ treatment increased Nrf2 binding to ARE1 of the ATF3 promoters as well as ARE of the NAD(P)H: quinine oxidoreductase (NQO1) promoter which was used as a positive control (Figure 5E). These results suggest that Nrf2 binds to ARE1 located in the ATF3 promoter, thereby inducing ATF3 expression.

To assess the importance of ATF3 induction in astrocytes, we used two siRNAs targeting ATF3 (siATF3#1 and siATF3#2) for the knockdown of tBHQ-induced ATF3 expression (Figure 6A). It was previously reported that Nrf2 has an antioxidative function via the modulation of cellular ROS levels (¹³). Therefore, we assessed whether the Nrf2-dependent induction of ATF3 affected cellular ROS levels employing 2?,7?-dichlorodihydrofluorescein diacetate (DCFH-DA) which is sensitive to oxidation by cellular ROS. Figure 6B shows that tBHQ decreased ROS levels in astrocytes. However, the tBHQ-mediated reduction of cellular ROS levels was decreased in ATF3-depleted astrocytes, which implies that ATF3 may be required for the antioxidative functions of Nrf2. Furthermore, we examined the effect of ATF3 induction on the expression of Nrf2 target genes because ATF3 could exert its antioxidative effect via the transcriptional modulation of the expression of Nrf2 target genes. However, the absence of ATF3 did not affect the mRNA expression of Nrf2 and Nrf2 target genes including antioxidant enzymes HO-1 and sulfiredoxin 1 (SRXN1), and the detoxifying enzyme NQO1 (Figure 6C). Additionally, the immunocytochemical analysis using anti-Nrf2 antibody showed that the depletion of ATF3 did not affect the immunoreactivity for Nrf2 which was stabilized by treatment with tBHQ (Supplementary Figure S1). These findings suggest that ATF3 contributed to the antioxidative function of Nrf2 in astrocytes although ATF3 did not modulate the transcription of Nrf2 and Nrf2-target genes.

Next, we speculated that ATF3 might affect cellular viability against oxidative stress. To address this, we used siATF3 to silence the expression of ATF3 induced by treatment with AS and PMS. WST assay was performed to examine cytotoxicity after treatment with AS or PMS for 24 h. As shown in Figure 7A, the absorbance was decreased in ATF3-deficient astrocytes treated with oxidants, which implies that oxidant-induced cytotoxicity was increased by the knockdown of ATF3. To examine apoptosis, we performed a TUNEL assay which detects DNA fragmentation resulting from apoptosis. The number of apoptotic cells generated by treatment with oxidants was increased in ATF3-depleted cells (Figure 7B and Supplementary Figure S2). Next, we determined the effect of ATF3 on DNA damage by performing a single-cell gel electrophoresis assay to visualize single- and double- strand DNA breaks. As illustrated in Figure 7C, DNA damage was enhanced in ATF3-depleted cells. This result was confirmed by the immunocytochemical analysis using the anti-phospho histone H2A.X antibody. Phosphorylation of histone H2A.X is an indication of the early DNA damage response (¹⁴). We found that the number of foci generated by the treatment with oxidants was increased in ATF3-depleted cells (Figure 7D and Supplementary Figure S3). All together, these results demonstrated that the knockdown of ATF3 enhanced DNA damages under oxidative stress conditions in human astrocytes.

GSH is the most abundant antioxidant. It protects cells by scavenging ROS in stress conditions. Next, we asked whether ATF3 was able to regulate the amount of GSH in response to oxidants. The levels of GSH increased after AS and PMS treatments (Figure 7E), suggesting that the upregulation of GSH has a counter-effect on oxidant-induced damage. However, the depletion of ATF3 reduced the upregulation of GSH, which implies that ATF3 is required for the generation of intracellular GSH. Therefore, these data suggest that ATF3 induction in astrocytes may contribute to the cytoprotective function of Nrf2 against oxidative insults by modulating GSH levels. The experiments depicted in Figure 7 were performed using non-transfected control cells, in which the values were not different from those observed in siGFP-transfected control cells (data not shown).