Induction of the Nrf2-driven Antioxidant Response Confers Neuroprotection During Mitochondrial Stress in vivo

NF-E2 related factor (Nrf2) controls a pleiotropic cellular defense, where multiple antioxidant/detoxification pathways are up-regulated in unison. Although small molecule inducers of Nrf2 activity have been reported to protect neurons in vitro, whether similar pathways can be accessed in vivo is not known. We have investigated whether in vivo toxicity of the mitochondrial complex II inhibitor 3-nitropropionic acid (3-NP) can be attenuated by constitutive and inducible Nrf2 activity. The absence of Nrf2 function in Nrf2(-/-) mice resulted in 3-NP hypersensitivity that became apparent with time and increasing dose, causing motor deficits and striatal lesions on a more rapid time scale than identically treated Nrf2(+/+) and Nrf2(+/-) controls. Striatal succinate dehydrogenase activity, the target of 3-NP, was inhibited to the same extent in all genotypes by a single acute dose of 3-NP, suggesting that brain concentrations of 3-NP were similar. Dietary supplementation with the Nrf2 inducer tert-butylhydroquinone attenuated 3-NP toxicity in Nrf2(+/-) mice, but not Nrf2(-/-), confirming the Nrf2-specific action of the inducer in vivo. Increased Nrf2 activity alone was sufficient to protect animals from 3-NP toxicity because intrastriatal adenovirus-mediated Nrf2 overexpression significantly reduced lesion size compared with green fluorescent protein overexpressing controls. In cultured astrocytes, 3-NP was found to increase Nrf2 activity leading to antioxidant response element-dependent gene expression providing a potential mechanism for the increased sensitivity of Nrf2(-/-) animals to 3-NP toxicity in vivo. We conclude that Nrf2 may underlie a feedback system limiting oxidative load during chronic metabolic stress.


Introduction
The Cap 'n' Collar transcription factor NF-E2 related factor (Nrf2) controls the coordinated expression of important antioxidant and detoxification genes (Phase 2 genes) through a promotor sequence termed the antioxidant response element (ARE) (1)(2)(3). Phase 2 genes work in synergy to constitute a pleiotropic cellular defense that scavenges reactive oxygen/nitrogen species (ROS/RNS), detoxifies electrophiles and xenobiotics, and maintains intracellular reducing potential (4)(5)(6)(7)(8). Normally, Nrf2 is sequestered in the cytoplasm by the actin-bound regulatory protein, Keap1 (9,10). Multiple cysteine residues allow Keap1 to act as a molecular 'switch' by responding to electrophiles/ROS with a conformational change, which releases Nrf2 for nuclear translocation and activation of Phase 2 gene expression (11)(12)(13). Thus, Nrf2 provides an important mechanistic link between oxidative stress leading to cell death and antioxidant gene expression supporting cell survival.
Nrf2 -/mice are particularly maladapted when challenged with toxicity paradigms such as hyperoxic lung injury, butylated hydroxytoluene-induced acute pulmonary injury, acetaminophen-induced liver toxicity, benzo[a]pyrene-induced tumourigenesis in the forestomach, or cigarette smoke-induced emphysema (14)(15)(16)(17)(18)(19). Although the protective function of Nrf2 is apparent in peripheral tissues, the role of Nrf2 in the brain, where substantial ROS/RNS 2 production can occur due to a high rate of metabolism, remains unclear. We hypothesize Nrf2 function is critical for supporting neuronal survival during neurodegenerative disease when aberrant ROS/RNS production is known to be exacerbated (20). To study the effect of Nrf2 activity on in vivo neurodegeneration, we systemically administered 3nitropropionic acid , an irreversible inhibitor of succinate dehydrogenase (SDH), which causes striatal-specific cell loss leading to impairment of motor function 3-NP models (21). Metabolic inhibition by 3-NP produces oxidative stress in the brain through multiple mechanisms: 1.) ATP depletion, membrane depolarization, increased neuronal/glial glutamate release, over-activation of N-methyl-D-aspartate ionotropic glutamate receptors, accumulation of intracellular Ca 2+ , and excessive mitochondrial ROS production (22,23). 2.) Secondary activation of Ca 2+ -dependent enzymes such as phospholipase A 2 (production of inflammatory mediators) and nitric oxide synthase (peroxynitrite formation) (24,25). 3.) Excessive dopamine release, which can generate H 2 O 2 when metabolized (26).
Upregulation of Nrf2 activity in the brain is an attractive strategy for mitigating ROS/RNS overproduction during neurodegenerative disease. To test the hypothesis that augmentation of Nrf2 could protect neurons in vivo, we increased Nrf2 activity in the brain using dietary administration of the small molecule inducer, tert-butylhydroquinone (tBHQ) or adenoviral-mediated overexpression of Nrf2 (6). In previous studies, these treatments protected cultured neurons from toxicity paradigms that mimic aspects of in vivo neurodegeneration, such as glutamateinduced glutathione depletion, dopamine toxicity, direct exposure to H 2 O 2 , metabolic inhibition with mitochondrial toxins (rotenone), platelet activating factor-induced inflammatory responses, and increased intracellular Ca 2+ (6,(27)(28)(29)(30). Importantly, neuronal cultures derived from Nrf2 -/mice show increased susceptibility to neurotoxicity paradigms and were not protected by Nrf2 inducers, but can be rescued by overexpression of Nrf2 (28,31).
Our in vitro and in vivo data suggest that Nrf2 activation is a protective response to counter the toxic effects of metabolic inhibition and ROS/RNS production in the brain. Loss of Nrf2 function in Nrf2 -/mice exacerbated motor deficits and striatal lesions caused by 3-NP administration. Conversely, pre-activation of the Nrf2 response (using dietary administration of a Nrf2 inducer or viral gene transfer) attenuated 3-NP toxicity. Nrf2 activation may be a viable strategy to prime the antioxidant capacity of the brain, thereby decreasing injury caused by progressive neurodegeneration.

Materials and Methods
Chemicals: All chemicals were obtained from Sigma-Aldrich unless otherwise stated. In vivo 3-NP dosage regimen: All experiments were approved by the U.B.C. Animal Care Committee and were conducted in strict accordance with guidelines set by the Canadian Council on Animal Care. Adult male Wistar rats (Charles River, Canada) (250-350 g) and Nrf2 -/mice (C57Bl/6 / SV129 background), originating from the laboratory of Dr. Yuet Wai Kan (32), were maintained in a 12 h light-dark cycle with food and water ad libitum. 3-NP was prepared in 0.1 M phosphate buffered saline (PBS) and adjusted to pH 7.4 with NaOH and maintained at 4 o C for up to 1 week. For Nrf2 mice, a similar number of males and females (aged 10-16 weeks) were used. The 3-NP treatment regimen consisted of 9 total intraperitoneal (i.p.) injections with one injection given every 12 h at successively increasing doses: 20 mg/kg x 4, 40 mg/kg x 4, 60 mg/kg x 1 (33). No significant difference was observed between males and females with respect to behaviour, weight loss, or lesion size in this study. For rats, daily i.p. injections of 3-NP (n = 10) or PBS control (n = 10) were administered by one of two dosage regimens: (40 mg/kg, 40 mg/kg, 20 mg/kg) or (40 mg/kg, 20 mg/kg, 20 mg/kg, 20 mg/kg), each regimen resulting in the animal receiving a total of 100 mg/kg of 3-NP over 3 or 4 days. Both dosage regimens were tested in order to overcome possible preconditioning effects due to adenovirus injection. However, the average lesion volumes produced by the two dosage regimens were not different after statistical analyses (two-tailed ttest), and the groups were pooled. Since virus was injected unilaterally we conservatively only included in the study animals with measurable 3-NP lesions by histology. Animals were sacrificed 24 h after the last injection or upon displaying severe motor behaviour deficits such as recumbence (complete loss of motor control).

Semi-qualitative Behaviour Scoring of Mice:
A detailed description for the assessment of 3-NP induced motor deficits was described previously (33). Briefly, the scoring system involved the evaluation of 5 major symptoms each with a 3-tiered scoring scale (0, 1, or 2), with 0 = normal behaviour and 2 = severe deficit. Symptoms included: a.) hind limb clasping, b.) reduced general locomotor activity, c.) hind limb dystonia (increased muscle tone leading to abnormal jerky movements), d.) truncal dystonia, and e.) postural instability. Animals received a cumulative score out of 10 for each trial. Behaviour was assessed by an experimenter blinded to the animal genotype and treatment before each scheduled injection. Histology and Fluorojade Staining: Mice were deeply anesthetized with euthanyl (Bimeda-MTC) and trans-cardially perfused with room-temperature PBS followed by ice-cold 4% paraformaldehyde in PBS (PFA). The brains were post-fixed overnight in PFA and then cryoprotected with 30% sucrose for 2 days before cryostat sectioning. Cresyl violet (CV) staining was performed with standard protocols on 40 m sections mounted on SuperFrost Plus slides (Fisher). Lesion volume was calculated according to the principle of Cavalieri (volume = s 1 d 1 + s 2 d 2 + s 3 d 3 + s 4 d 4 ) where s = lesion surface area and d = distance between 2 sections. Adjacent 20 m slide-mounted sections were examined for Fluorojade (Histochem Inc.) staining as previously described (34). CV sections were scanned at 600 DPI on an Epson desktop scanner and lesion area was quantified by an experimenter blinded to the animal genotypes using ImageJ software (National Institute of Health). Four anatomical levels of the striatum were examined (Bregma +1.10 mm, +0.40 mm, -0.26 mm, -0.92 mm) according to the mouse brain atlas (35). Fluorojadepositive neurons were only assessed in the core of the striatal lesion. Fluorojade images were collected using Northern Eclipse software (ver. 6.0). The widefield fluorescence microscopy (Zeiss Axiophot) setup was equipped with a Retiga Exi CCD camera (QImaging). For rats, perfusions cryosectioning and CV staining were performed as described for mice. Total lesion volumes were calculated using the following formula V t = V 1 +V 2 + …+ V n , where V n = h(r n-1 2 + r n-1 r n + r n 2 )/3, where h is the distance between 2 sections and r is the radius of the lesion area using sections from Bregma 1.70 mm, 1.2 mm, 0.70 mm, 0.2 mm, -0.26 mm, and -0.80 mm according the rat brain atlas (36). Succinate Dehydrogenase Assay: Male or female mice from each genotype were acutely injected with 60 mg/kg 3-NP and were always paired with PBSinjected littermates of the same sex and genotype. After 2 h, the mice were euthanized with halothane and then decapitated. Brains were washed with icecold PBS and striata and cortices were dissected from a 2 mm brain-matrix slice at approximately Bregma +1.0 mm. Crude mitochondrial samples were prepared by dounce homogenization in 0.25 M sucrose, 1 mg/ml bovine serum albumin (BSA), 1 mM EDTA, pH 7.4. To collect the mitochondria, the homogenate was centrifuged at 4°C, 600 × g for 5 min. The supernatant was then re-centrifuged at 7,200 × g at 4°C for 10 min. The supernatant was aspirated, and the mitochondrial pellet was resuspended in the same buffer. SDH activity was assayed immediately using a protocol based on studies by Pennington et al. (37). Briefly, the reaction consisted of 0.1 M potassium phosphate buffer (pH 7.4), 0.1 M succinate, 0.05 M sucrose, 2 mg/mL p-Iodonitrotetrazolium chloride (prepared fresh), and 125 g/mL crude mitochondria (final concentrations in 400 L reaction volume) and incubated for 20 min at 37 o C. The formazan mixture was vortexed before measurement on a microplate reader at 490 nm. Protein concentration was measured using the bicinchronic acid method (Pierce). Brain homogenates were serially diluted for the assay and data analyses were performed on values within the linear range of the SDH reaction. We also used 4 Nrf2 +/mice to examine the dose-response relationship between acute SDH inhibition and various injected 3-NP concentrations (20 mg/kg, 40 mg/kg or 60 mg/kg). Both 20 mg/kg and 40 mg/kg doses produced little or no change 2 h after injection, compared to PBS injection (data not shown), whereas 60 mg/kg produced an observable decrease in activity. The 60 mg/kg dose was used for further SDH bioassays. Preparation of tBHQ-supplemented diets: All rats and mice were normally fed Lab Diet 5001, which contains no tBHQ, but trace amounts of BHA for preserving animal fats (2 ppm). For tBHQ feeding, food pellets were powdered in a coffee grinder and dry mixed with 1% (w/w) tBHQ. Distilled water was added to the powder (equal v/w), mixed, and reshaped into food pellets. The pellets were then baked at 60 o C for 3 h. Control food was processed in the same fashion without the addition of tBHQ. Initially, we determined 5% tBHQ in diet was not consumed by Nrf2 +/mice (n = 3), and 0.5% tBHQ, although consumed, only slightly preserved motor function during 3-NP treatment (n = 3). For these reasons, a concentration of 1% tBHQ was used in all further feeding experiments. Enzyme Assays for brain tissue: Brain and liver tissue was dissected in ice-cold PBS and homogenized with 10 strokes of a dounce homogenizer in tissue buffer containing 25 mM Tris (pH 7.4) and 250 mM sucrose. Crude homogenates were centrifuged at 15,000 g for 10 min at 4 o C. Tissue homogenate supernatants were collected and promptly assayed for enzyme activity. NQO1 enzyme activity was determined by calculating the dicumarol-sensitive fraction of DCPIP reduction (38). Reactions consisting of 25 mM Tris-HCl buffer (pH 7.4) with 0.7 mg/mL BSA, 5 M FAD, 200 µM NADH, 30 g/mL protein, with and without 20 µM dicumarol were pre-incubated for 10 min at 25°C (final concentrations in 200 L reaction volume). DCPIP was then added to a final concentration of 36 M (20 L volume), and the reaction was monitored at 540 nm. The extinction coefficient for DCPIP was 2.1 x 10 4 M -1 cm -1 . The GST assay consisted of 1 mM 1chloro-2,4-dinitrobenzene (CDNB), 1 mM glutathione, and 100 g/mL protein at 37°C in 100 mM potassium phosphate buffer (pH 6.5) (final concentration in 150 L reaction volume) (39). The GST reaction was monitored at 340 nm and the spontaneous non-enzymatic slope was subtracted from the total observed slope. The extinction coefficient for CDNB was 9600 M -1 cm -1 . The LDH assay consisted of 3.3 mM pyruvate, 0.34 mM NADH, and 100 g/mL protein in PBS at 37°C (final concentrations in 150 L reaction volume) (40). The LDH reaction was monitored at 340 nm and the spontaneous non-enzymatic slope was subtracted from the total observed slope. Purified LDH enzyme standards were used to ensure that values were obtained within the linear range of the assay. The extinction coefficient for NADH was 6219 M -1 cm -1 . Protein concentration was determined using the bicinchronic acid method according to the manufacturer's protocol (Pierce). Stereotaxic Injections: Rats were anaesthetized by i.p. injection of Somnitol (MTC Pharmaceuticals, Cambridge, ON) and placed in a stereotaxic frame (David Kopf Instruments). A burr hole was drilled and injection was performed into the right striatum at coordinates: +1.00 mm anterior-posterior from Bregma, 2.6 mm medial-lateral, and both 4.5 & 5.5 mm dorsal-ventral, using a 10 l Hamilton gas-tight syringe with 26 gauge needle (type 2 tip). The virus was delivered by manually pressing the syringe plunger 3 times at 4 min intervals. Ad-GFP or Ad-Nrf2 was delivered at a concentration of 5 x 10 6 total plaque forming units in 3 l of PBS. Wounds were irrigated with 0.9% sterile saline solution and sutured. Animals were allowed to recover for 3 days before 3-NP treatment was initiated. Immunohistochemistry: For rats, frozen 10 m brain sections were thawed in PBS for 3 min at room temperature (RT). Mouse monoclonal anti-O4 (Chemicon) or rabbit polyclonal anti-GFAP antibodies were diluted 1:400 and 1:200 respectively in Ab buffer (3% BSA and 0.3% Triton X-100 in PBS) and incubated overnight at 4 o C in a humid chamber. Sections were rinsed 3 x 5 min in PBS and incubated with goat anti-mouse IgM Alexa546 conjugated or goat anti-rabbit Alexa546 conjugated secondary antibodies (both Molecular Probes, 1:500) in a humid chamber for 1 h at RT. Sections were rinsed 3 x 10 min in PBS prior to mounting with Fluoromount-G (Southern Biotechnology Associates Inc.) and No.1 glass coverslips (Corning). All GFP fluorescence from viral infection could be detected without using an anti-GFP antibody. To assess total striatal volume, a propidium iodide (PI) counterstain was used to label all cell nuclei. Sections were incubated in 8 M PI dissolved in PBS for 10 min at RT followed by a 10 min PBS rinse. To determine infectivity, GFP fluorescence was calculated by thresholding images to subtract background and calculating the area of fluorescence (ImageJ). In the same section, PI staining was used to locate the corpus callosum and caudate putamen. The area encompassed by these two regions was defined as the striatal area. Data was expressed as % GFP + volume within the total striatal volume. For NeuN staining, an antigen retrieval process was used following thawing. Sections were incubated at 80 o C for 30 min in 10 mM sodium citrate pH 8.5 and then rinsed for 5 min in PBS prior to anti-NeuN antibody incubation overnight at 4 o C (Chemicon, 1:100) and goat antimouse Cy3 secondary antibodies for 1 h at RT (Sigma, 1:200). Fluorescent images were captured using a Zeiss LSM Meta 510 confocal microscope and analyzed using NIH Image (for Mac) or ImageJ (for PC).

Preparation of primary glial-enriched and COS-1 cultures:
Enriched glial cultures were prepared from 0-2 d post-natal Wistar rat pups as described previously (6). The conditions used largely results in a population of Type I and II astrocytes as assessed by anti-glial fibrillary acid protein (GFAP) staining. COS-1 cells (ATCC) were maintained in DMEM supplemented with 10% fetal bovine serum, 1 mM sodium pyruvate, 2 mM glutamine, 100 U/mL penicillin, and 0.1 mg/mL streptomycin. Plasmids and Adenoviruses: PEF mammalian expression plasmids carrying cDNA encoding mouse Nrf2 and Nrf2DN were a generous gift from Dr. Jawed Alam (Alton Ochsner Medical Foundation, New Orleans, LA) (1). Rat ARE sequences were obtained from the NQO1 promotor. The human placental alkaline phosphatase (hPAP) reporter plasmid (rQR51) was constructed as described previously (6,41). Recombinant adenoviral vectors were generated using the Cre-lox system (Canadian Stroke Network core facility, University of Ottawa), as described previously (6,42). The N-terminustagged Nrf2-GFP vector was made by PCRamplifying the Nrf2 coding sequence from the PEF vector with specific primer sequences: forward, 5'-ACTCAGATCTCGAGAAGATTTGATTGACATC CTT-3' and reverse, 5'-CCCGGGGGTACCCTAGTTTTTCTTTGTATC-3'. The resulting PCR product was restriction digested with Xho I and Kpn I for ligation into the pEGFP-N1 vector (Invitrogen). The Keap1 vector was a generous gift from Dr. Masayuki Yamamoto (University of Tsukuba), later modified to Keap1-FLAG by Laurie Zipper (University of Wisconsin) (9,43).

Transfection, infection, and treatments for astrocytes and COS-1 cells:
Astrocytes and COS-1 cells seeded (1 x 10 5 cells/mL) in 24-well plates were transfected using Lipofectamine 2000 (Invitrogen), according to the manufacturer's protocol with some modifications. In particular, for each well of a 24well plate, 1 g of DNA and 1 L of lipofectamine were used. The transfection efficiency was typically ~ 20% for astrocytes and ~ 50% for COS-1 cells as assessed by galactosidase staining. For some experiments, adenovirus infection of astrocytes as described previously (6), was initiated immediately after lipofectamine transfection. All infected cultures were examined for adequate infection efficiency (~90 % of glia) as assessed by GFP fluorescence. 3-NP or tBHQ treatments were initiated 24 h after transfection. 3-NP was prepared in basic salt solution (BSS) consisting of (in mM) 3.1 KCl, 134 NaCl, 2.5 CaCl 2 , 1 MgSO 4 , 0.25 KH 2 PO 4 , 15.7 NaHCO 3 (pH 7.2), and filtered before use. The 3-NP stock was further diluted in BSS to the indicated concentration for experiments. Cells were washed with BSS twice and then exposed to 3-NP or BSS (vehicle) for 2 h. The cells were washed again 3 times with BSS and the original culture media was replaced. The Nrf2 inducer, tBHQ, was dissolved as a 1000x concentrated stock in pure DMSO (vehicle), sterile filtered, and then diluted to a final concentration of 10 or 20 M in the glial culture media. TBHQ was present in the media for the duration of the experiment until cells were harvested. Placental alkaline phosphatase assay: Human placental alkaline phosphatase (hPAP) activity was measured as described previously (44). Briefly, cultures were collected in lysis buffer consisting of 10 mM Tris-HCl (pH 8), 1 mM MgCl 2 , and 0.1% Triton X-100. Half of the sample was saved for determination of protein concentration andgalactosidase activity (for normalization of transfection efficiency). The remaining sample was heated to 65 o C for 30 min to inactivate endogenous phosphatase activity. The assay was initiated by mixing ~15 g protein with diethanolamine buffer (0.73 M diethanolamine with 0.36 mM MgCl 2 , pH 9.8), and 13.6 mM p-nitrophenyl phosphate (final concentrations in 150 L reaction volume). The reaction was monitored every 2 min over 30 min, at 405 nm. For all assays, hPAP activity from rQR51mut transfected astrocytes were subtracted as ARE-independent background. Semi-qualitative reverse-transcriptase PCR: Total RNA was isolated from infected cultures using Trizol Reagent (Life Technologies, GIBCO) following manufacturer's instructions. One µg of RNA was reverse transcribed using random hexamer primers and Thermoscript reverse transcriptase enzyme (Invitrogen). The resulting cDNA was then amplified by PCR using various primer sets (31) PCR conditions were as follows: initial denaturing step using 1 cycle at 95 o C for 3 min, followed by 26 cycles at 95 o C for 30 s, the appropriate primer annealing temperature (~55 o C) for 40 s, 72 o C for 1 min, and a final cycle at 72 o C for 5 min. For xCT and GCLC, 36 and 31 cycles were used, respectively. The PCR products were separated on a 1.4% agarose gel containing ethidium bromide. cDNA was then visualized using an ultraviolet light source. cDNA templates for each sample were serially diluted and band densities were plotted. These plots were used to ensure that data represented PCR products within the linear range of the reaction (Suppl. Fig. 2). Glutathione assay: Total glutathione was quantified in glial cell lysates and tissue homogenates (0.3 mg/mL) by the method of Tietze (45), as described previously (6).

Nrf2 -/mice are hypersensitive to systemic 3-NP treatment
Nrf2 -/mice were challenged with 3-NP to examine how loss of Nrf2 function affects both motor system impairment and toxicity attributed to metabolic inhibition in vivo. Our hypothesis was that Nrf2 -/mice would be more sensitive to the effects of 3-NP-induced metabolic inhibition. Thus in initial experiments we determined in vivo 3-NP dosing regimes that would provide significant impairment in Nrf2 -/mice and then assessed what impact these had on wild type animals. The extent of 3-NP toxicity was monitored through both manifestation of overt motor symptoms (i.e. reduced general locomotor activity, hind limb and truncal dystonia, and postural instability) and development of bilateral lesions restricted to the dorso-lateral striata (33). An advantage of using behavioral assessment as an adjunct to histology is that it can be performed repeatedly on the same animals providing a measure of 3-NP toxicity progression.
Lesions of Nrf2 -/mice contained mainly pyknotic cells which stained positive for fluorojade, a marker of neuronal degeneration (Fig. 1D center and right column, and 1F) (34). Only a sub-population of the remaining neurons in the Nrf2 +/lesions was fluorojade-positive, and little or no positive cells were observed in the striata of Nrf2 +/+ mice. No significant extra-striatal (i.e. hippocampal or cortical) fluorojade-positive staining was observed in 3-NP treated Nrf2 -/mice, consistent with the striatalspecific action of systemic 3-NP administration. Importantly, Nrf2 -/mice were specifically sensitive to 3-NP since injection of PBS control solution caused negligible motor abnormalities and no lesions/fluorojade-positive neurodegeneration (data not shown). Although the treatment regimen we used did not cause toxicity in Nrf2 +/+ mice, prolonged 3-NP administration (1-2 doses of 60 mg/kg 3-NP beyond current regimen) was able to provoke the development of motor deficits and lesions (data not shown).

Striatal SDH Inhibition by 3-NP is Not Different Among Genotypes
Since we used systemic administration of 3-NP, it was possible that increased toxicity in Nrf2 -/mice was due to compromised peripheral detoxification systems (i.e. liver, kidney, and intestines), leading to a higher circulating concentration of 3-NP in the brain. To address this possibility, we directly quantified SDH activity in the brain after an acute systemic injection of 3-NP (60 mg/kg). The cortex and striatum were collected after 2 h when peak brain SDH inhibition was shown to occur (47). SDH activity was reduced by 40-50% in total tissue from cortex and striatum of all genotypes, consistent with previous studies (Fig. 2)(47). Importantly, we observed no difference in SDH activity between genotypes, suggesting that increased damage caused by 3-NP in Nrf2 -/mice is most likely due to an intrinsic sensitivity in their brains.
Nrf2 induction by small electrophilic molecules represents a practical method to increase the Phase 2 response in vivo (48)(49)(50). Dietary administration of Nrf2 inducers, such as sulforaphane (isolated from broccoli sprouts) have been shown to induce Phase 2 genes in peripheral tissues and reduce toxicity associated with a number of pathologies, including stomach tumour formation and hypertension (18,51). To determine whether in vivo induction of Nrf2 could provide resistance to 3-NP toxicity, mouse diets were supplemented with the potent Nrf2 inducer, tert-butylhydroquinone (tBHQ) (1,52).
The in vivo phase 2 response was assessed by measuring the activities of two prototypic phase 2 enzymes, glutathione S-transferase (GST) and NAD(P)H:quinone oxidoreductase (NQO1), as well as total glutathione (GSH) content. Nrf2 -/mice exhibited lower basal striatal and cortical activity for GST and NQO1, when compared to Nrf2 +/+ littermate controls (Table 1, basal phase 2 enzyme levels). Interestingly, basal brain GSH levels were not altered, suggesting constitutive synthesis of brain GSH does not rely on Nrf2 function, at least in the young adult mice examined in here. As observed in previous studies, GST, NQO1 activities, and total GSH levels were markedly reduced in the livers of Nrf2 -/mice (53)(54)(55)(56). The housekeeping gene, lactate dehydrogenase (LDH), is not regulated by Nrf2 and accordingly, no significant differences in activity were observed between genotypes.
Primarily Nrf2 +/and Nrf2 -/mice were used in tBHQ feeding studies since we were interested in observing reduced motor deficit and lesion size during 3-NP treatment, and Nrf2 +/+ mice did not develop these phenotypes. The tBHQ-supplemented diet (1% w/w) was well tolerated by Nrf2 +/mice since no significant weight loss was observed over 1 week of feeding, except a transient loss of weight 2 d after initiation of feeding, and food consumption was similar with mice receiving control diet (Fig 3A,C). However, Nrf2 -/mice were less tolerant of dietary tBHQ with overall body weight ~ 10-20% lower than mice with control diet during 1 week of feeding (Fig.  3B, *** p < 0.0001 2-way ANOVA), and less food consumption than tBHQ-fed Nrf2 +/mice (Fig.  3B,D). In total, Nrf2 +/-, and Nrf2 -/mice consumed 12.9 + 0.9, and 7.6 + 0.7 mg tBHQ/g body weight over 1 week, respectively. The 1% tBHQ diet increased striatal and cortical GSH levels in Nrf2expressing mice (Nrf2 +/+ and Nrf2 +/-)( Table 1, tBHQinduced phase 2 enzyme levels). Importantly, the induction of brain GSH was not observed in Nrf2 -/mice, providing evidence that tBHQ-induced changes in GSH are Nrf2 dependent. In addition, tBHQinduced increase in liver GST and NQO1 activities was observed in Nrf2 expressing mice, but not Nrf2 -/mice, consistent with the effects of other Nrf2 inducers used in vivo (56,57). However, no induction of GST or NQO1 was detected in brain. Levels of the negative control enzyme, LDH, was unaffected by tBHQ in all tissues examined. Interestingly, we observed an increase in liver GSH levels in tBHQ-fed Nrf2 -/mice. We believe that this Nrf2-independent change may reflect an inherent hepatic toxicity associated with tBHQ feeding, perhaps due to inefficient detoxification from the body by the same pathways regulated by Nrf2 (Phase 2 genes). Together, these results highlight the sensitivity of the Nrf2 -/phenotype to tBHQ toxicity and suggest normal Phase 2 induction in response to tBHQ is abrogated in Nrf2 -/mice.
The tBHQ diet was given 1 week prior to the 3-NP regimen and maintained until sacrifice. TBHQfed Nrf2 +/mice had significantly attenuated 3-NP induced motor deficits compared to mice receiving the control diet (Fig. 3E, ** p < 0.001 two-way ANOVA). In contrast, tBHQ-fed Nrf2 -/mice exhibited an overall exacerbation of motor deficit (Fig. 3F, *** p = 0.0002 two-way ANOVA). The average lesion volume of Nrf2 +/given control diet was similar to that measure in our initial experiments (0.65 + 0.29 mm 3 , uncorrected for edema) (Fig.  3G,H). However, consistent with reduced behavioural scores, no tBHQ-fed Nrf2 +/mice exhibited detectable lesions indicating protection of neurons. TBHQ feeding did not reduce lesion size in Nrf2 -/mice, confirming the protective effect of tBHQ consumption was Nrf2-dependent (Fig. 3G,H).

Adenoviral overexpression of Nrf2 attenuates 3-NP induced striatal lesioning in vivo.
To further address whether activation of Nrf2-dependent pathways is sufficient to protect the striatum during 3-NP toxicity, in parallel studies, we tested whether direct Nrf2 overexpression using adenovirus vectors (Ad-Nrf2 or Ad-GFP control) conferred neuroprotection to rats treated with 3-NP. Overexpression of Nrf2 causes its accumulation in the nucleus leading to even more robust AREdependent transcription than that observed with small molecule inducer treatment (43). In vivo intrastriatal adenovirus injection primarily led to infection of astrocytes (Fig, 4A top panel, and 4B), and almost no neurons or oligodendrocytes (Fig. 4A middle and  bottom panel, and 4B). Vector diffusion was observed throughout the striatum (Fig. 4C), infecting 0.48 ± 0.08% of the total striatal volume, as assessed by the presence of the GFP marker (Fig. 4D). After infection, protein expression was allowed to occur for 3 days before 3-NP treatment was initiated.
Following 3-NP challenge, lesion volumes in the infected hemisphere were significantly smaller in animals receiving Ad-Nrf2, compared to Ad-GFP control virus (9.6 ± 2.6 mm 3 versus 23.4 ± 4.2 mm 3 , p = 0.03, two-tailed t-test, Fig. 4E,F). Interestingly, a trend toward decreased lesion volume was also observed in the hemisphere contralateral to the virus injection (12.2 ± 3.8 mm 3 compared to 25.9 ± 4.9 mm 3 , p = 0.06), although this difference did not attain statistical significance. This effect is likely due to diffusion of secreted glial factors, possibly glutathione or glutathione precursors, by volume transmission along fiber bundles of the corpus callosum, or through the cerebral spinal fluid from the infected Nrf2 overexpressing astrocytes in the ipsilateral hemisphere (58,59).

3-NP activates ARE-dependent gene expression in cultured astrocytes.
Our data show that both the Nrf2 inducer tBHQ and Nrf2 overexpression could protect neurons by pre-activating the Phase 2 response. Since increased susceptibility of Nrf2 -/mice may result from inability to activate expression of protective genes necessary for reducing toxicity associated with 3-NP, we hypothesized that metabolic inhibition by 3-NP was acting as a signal for Nrf2 activation. To test this hypothesis, we examined the effect of 3-NP on Nrf2 activity in cultured astrocytes, the primary site of ARE-mediated gene expression in brain cultures (6,41,60).
To mimic the effects of bolus in vivo injection, cultured astrocytes were treated with a wide range of 3-NP concentrations (1-30 mM) for 2 h and then ARE-mediated gene expression was assessed using a hPAP reporter assay regulated by an ARE-bearing NQO1 promoter 24 h later (Fig. 5A) (41). Astrocytes transiently treated (2 h) with 10-30 mM 3-NP exhibited a doubling of ARE-mediated gene expression. Over these time periods and with these 3-NP concentrations no apparent astrocyte toxicity was observed. Expression of ARE-driven hPAP, with and without 3-NP exposure, was reduced to sub-basal levels when astrocytes were infected with adenovirus overexpressing a dominant-negative form of Nrf2 (Ad-Nrf2DN), but not with overexpression of GFP control (Ad-GFP) (Fig. 5B) (6). Exposure to tBHQ (20 M for 24 h) induced hPAP expression to a similar extent as 3-NP treatment. Maximal hPAP activity was achieved with Nrf2 overexpression (Ad-Nrf2). Although the concentrations of 3-NP we use to activate Nrf2 mediated gene expression are high, the ability of dominant negative Nrf2 to block the induction indicates that it is nonetheless mediated by the ARE. Perhaps these high concentrations reflect the activated and toxin-resistant nature of astrocytes in vitro (61).
Semi-quantitative RT-PCR confirmed that astrocytes treated with 30 mM 3-NP for 2 h expressed higher levels of prototypic Nrf2 target gene mRNAs including NQO1, -glutamylcysteine synthetase (GCLC), and the cystine/glutamate exchanger (xCT) (Fig 5C). The expression of these mRNAs was increased by Ad-Nrf2 and suppressed by Ad-Nrf2DN infection, verifying the specificity of these target genes. For all RT-PCR experiments, DNA templates were serially diluted to ensure that band intensities were observed within the linear range of the PCR reaction (Suppl. Fig. 2). Since a number of Nrf2 target genes are involved in GSH synthesis (i.e. xCT, GCLC, and glutathione synthase), treatment with 30 mM 3-NP also increased intracellular GSH levels in cultured astrocytes (Fig.  5D) (5,6,54,55). Consistent with ARE-driven hPAP expression, 3-NP-mediated GSH induction was blocked with Ad-Nrf2DN infection, but not Ad-GFP control (Fig. 5D). Protein levels of heme oxygenase-1 (HO-1), a well characterized Nrf2 target, were also induced by 3-NP treatment (Fig. 5E,F) (1). Consistent with the GSH data both basal and induced expression of HO-1 was suppressed with Ad-Nrf2DN infection. Although this analysis does not encompass the full scope of Nrf2 gene targets, it nevertheless indicates the Phase 2 response is activated as a dose-dependent effect of 3-NP.
We further evaluated nuclear translocation of GFP-tagged Nrf2 (Nrf2-GFP) in response to 3-NP exposure. In initial experiments using astrocytes, Nrf2-GFP fluorescence was almost undetectable when co-expressed with its cytoplasmic regulator Keap1, possibly due to a faster rate of Nrf2 turnover by prote2some dependent pathways (62). As an alternative, COS-1 cells were used because of their high level of trans-gene expression and well-defined morphology for evaluating nuclear versus cytoplasmic localization. Consistent with previous studies, overexpressed Nrf2-GFP was almost exclusively localized to the nucleus unless coexpressed with Keap1 (Keap1-FLAG) (Fig. 6A, top row) (13,43). COS-1 cells co-expressing Nrf2-GFP and Keap1-FLAG were evaluated after treatment with 3-NP or BSS vehicle control for 2 h. As expected, the proportion of cells exhibiting nucleusenriched Nrf2-GFP fluorescence was significantly higher in 3-NP-treated cultures, when compared to vehicle control (Fig. 6A,B). Interestingly, 3-NP treatment also caused some aggregation of Keap1 and Nrf2-GFP in the cytoplasm, which we speculate is due to 3-NP-induced endoplasmic reticulum stress causing accumulation of unfolded proteins (63). This effect was transient since the aggregations were not observed 24 h after 3-NP washout (data not shown). Treatment with 20 M tBHQ for 4 h, as a positive control, also robustly increased nuclear localization of Nrf2-GFP (Fig. 6A,B). Both Nrf2-GFP and Keap1-FLAG constructs were functional, as assessed by their ability to augment and suppress ARE-driven hPAP expression, respectively (Fig. 6C).
Collectively, this in vitro data confirms Nrf2 translocates to the nucleus and initiates Phase 2 genes induction in response to 3-NP mediated metabolic inhibition. This effect can be mimicked using the Nrf2 inducer, tBHQ, and by adenoviral Nrf2 overexpression.

Discussion
This study provides evidence to support the neuroprotective function of Nrf2 during metabolic compromise in the in vivo brain. First, loss of Nrf2 function led to an increased susceptibility to 3-NP since Nrf2 -/mice developed motor deficits and striatal lesions more rapidly than Nrf2 expressing controls. Second, pre-activation of endogenous Nrf2 using the small molecule inducer, tBHQ, attenuated 3-NP toxicity in Nrf2 +/mice, but not Nrf2 -/-, confirming both the neuroprotective action and Nrf2-dependence of tBHQ in vivo. Third, direct overexpression of Nrf2 in the striatum is sufficient to reduce lesion size caused by 3-NP. Fourth, AREdependent gene expression in cultured astrocytes was activated by 3-NP, and could be completely suppressed by overexpression of a dominant-negative form of Nrf2.
The increased sensitivity of Nrf2 -/mice may result from an inability to induce Phase 2 genes in response to 3-NP toxicity. Indeed, a recent study by Calkins et al. (published while this work was submitted) showed using transgenic reporter mice, that ARE-dependent gene expression occurred in the immediate border of 3-NP induced striatal lesions, perhaps in reactive astrocytes commonly found within this region (64). Since Nrf2 is generally thought to respond to oxidative stress or signal transduction pathways associated with it (65,66), we speculate that 3-NP activates this protective pathway by indirectly generating ROS/RNS. 3-NP may increase brain oxidative stress through multiple mechanisms that may be implicated in our in vitro and in vivo experiments (6,(29)(30)(31). Also supported by our data, an alternative possibility for the increased sensitivity of Nrf2 -/mice may be the lower basal levels of phase 2 enzymes expressed in this genotype ( Table 1, basal phase 2 enzyme levels) (83).
Given that systemic 3-NP undergoes first pass metabolism before it reaches the striatum, it was possible that increased striatal damage in Nrf2 -/mice was in part due to impaired clearance of 3-NP from the body by peripheral detoxification. Indeed previous studies of Nrf2 -/mice have shown decreased Phase 2 enzyme activity in the liver and intestines (53,67), organs which may affect the removal of systemically injected 3-NP. However, we have shown that after a single high in vivo dose of 3-NP, striatal SDH was inhibited to an equal extent between Nrf2 +/+ and Nrf2 -/genotypes, suggesting that 3-NP produces similar metabolic inhibition on a dose-wise basis regardless of genotype. Thus, it appears that the brains of Nrf2 -/mice are intrinsically more sensitive to direct damage. This conclusion is also supported by a number of other findings. First, neurons and glia isolated from Nrf2 -/mice are more sensitive to in vitro toxicity paradigms involving oxidative stress or metabolic compromise (8,28,31). Second, direct injection of malonate (reversible SDH inhibitor) into the striatum produced larger lesions in Nrf2 -/mice, compared to Nrf2 expressing controls (64). Third, Nrf2 -/mice develop larger cortical infarcts than Nrf2 +/+ mice after experimental stroke (permanent distal middle cerebral artery occlusion), a model which also involves local brain metabolic inhibition (unpublished data) (84).

The small molecule inducer, tBHQ, provided Nrf2-dependent amelioration of 3-NP toxicity in vivo.
Although studies indicate that loss of Nrf2 potentiates neurotoxicity (64), whether the in vivo effects of small molecule inducers of ARE-mediated gene expression are lost in Nrf2 knockout mice has not been previously addressed. In support of a neuroprotective role for Nrf2 in vivo, pre-activation of Nrf2 by the small molecule, tBHQ was capable of attenuating 3-NP toxicity. Previous in vitro studies have shown that tBHQ exposure augments AREdependent gene expression in cultured astrocytes and protects neurons from toxicity paradigms involving oxidative stress (6,28,68,69). We extend these findings to the in vivo situation since dietary consumption of tBHQ significantly reduced motor deficits and lesion developments in Nrf2 +/mice. Importantly, the protective effect of tBHQ was abrogated in Nrf2 -/mice. Furthermore, dietary tBHQ administration increased brain GSH levels in Nrf2 +/+ and Nrf2 +/mice, but not Nrf2 -/mice (Table 1, tBHQinduced phase 2 enzyme levels).
As expected from previous studies (70), dietary tBHQ treatment also increased Phase 2 enzyme activity (GST and NQO1) in the liver. Thus, it remains possible that attenuation of 3-NP toxicity in tBHQ-fed mice could be in part due to enhanced removal of 3-NP from the body by increased liver detoxification or conceivably liver derived antioxidants such as GSH released into the blood. However, our observed increase in brain GSH would be expected to contribute to local neutralization of ROS/RNS, leading to tissue preservation. In preliminary studies, we have found that even local infusion of tBHQ in brain ventricles using osmotic pumps can lead to increases in liver phase 2 enzyme levels suggesting that it is difficult to completely dissociate brain and peripheral organ detoxification. Future studies could more directly address the role of brain Nrf2 by developing tissue and organ specific Nrf2 knockouts. However, in support of the brain being the site of tBHQ action, overexpression of Nrf2 within the striatum using adenovirus was sufficient to reduce the effect of systemically administered 3-NP. Although some uncertainty remains as to the site of Nrf2 induction mediated by tBHQ in vivo, our study nonetheless indicates that striatal-specific damage due to the systemic administration of a metabolic stressor can be strongly attenuated by this dietary strategy. Given that many neurodegenerative diseases may be triggered by exposure to diet or lung-derived toxins (71)(72)(73), it is conceivable that a dietary strategy to increase antioxidant function in multiple organs may be an ideal prophylactic strategy. To our knowledge, this is the first study to show that a protective effect of a Phase 2 inducer on brain function in vivo is dependent on Nrf2.
Although tBHQ is widely used as a food antioxidant, and is well-tolerated by the body (74,75), it is also important to consider the detrimental action of tBHQ and other Nrf2 inducers. As observed in this study Nrf2 -/mice do not tolerate tBHQ administration as well as controls (as evidenced by weight loss and reduced food consumption) perhaps due to inefficient detoxification of tBHQ through Nrf2-regulated pathways (Fig. 3A,B). Other groups have also found that Nrf2 -/mice are less tolerant to dietary administration of known Nrf2 inducers (67). Recently, increased susceptibility to hypoxia has been linked to polymorphisms in the Nrf2 gene (76). The question arises as to whether such mutations affect Nrf2 function in humans, and could underlie sensitivity to human disease. Since many Nrf2 inducers are essentially reactive electrophiles, the inability to properly detoxify Nrf2 inducers from the body could be harmful to individuals with such a genetic predisposition.
With regard to practical application, recent studies have elucidated a variety of structurally related small molecules which are able to activate Nrf2 by reacting with thiol/disulfide groups on the cytoplasmic regulatory protein Keap1 (4,48,50,(77)(78)(79). Some of these inducers are found within our diet. For example, the isothiocyanate sulforaphane, a potent Nrf2 inducer abundant in cruciferous plants (i.e. broccoli sprouts), was found to inhibit gastric tumour formation in mice by administration of benzo[a]pyrene (78). Juurlink and colleagues have recently demonstrated a significant improvement of cardiovascular fitness in spontaneously hypertensive rats through chronic upregulation of Phase 2 enzymes by supplementing diets with broccoli sprouts (51,80). Such studies support the potential for mitigating oxidative tissue damage in chronic disorders involving oxidative stress, by applying a change in diet -a practical and realistic therapeutic approach from a clinical standpoint.

Overexpression of Nrf2 protein is sufficient to provide neuroprotection in vivo.
We have also conducted a proof of principle experiment showing that adenoviral overexpression of Nrf2 in the striatum is sufficient to reduce 3-NP toxicity in vivo. Adenovirus was found to mainly infect astrocytes throughout the striatum as well as in the corpus callosum. In our immunohistochemical analysis, we observed that some GFP-positive cells with astrocyte morphology (~35%) were not labeled by any cell-type marker tested However, these cells were labeled by a different astrocyte marker, S100 (total GFP and S100 co-localization = 91.8 + 0.2%, n = 2), confirming their identity (81). With regard to the extent of neuroprotection achieved with Nrf2 overexpression in vivo, we have previously demonstrated a small number of Ad-Nrf2 infected astrocytes can confer protection to many neighbouring neurons during an oxidative glutamate insult in vitro (1 astrocyte can protect 100 neurons) (6). These Nrf2 overexpressing astrocytes secrete high levels of glutathione, providing a means for a small number of astrocytes to affect the survival of a much larger number of neurons (6). In support of our findings, Calkins et al. observed significant neuroprotection in vivo by transplanting a relatively small number of Nrf2 overexpressing astrocytes (from culture) directly into the striatum prior to injection of the competitive complex II inhibitor, malonate (64). In addition, many factors contributing to oxidative stress (i.e. ROS/RNS) are cell permeable allowing them to be metabolized in a subpopulation of cells (82). Another factor to consider is the actual spatial localization of the astrocytes infected in situ. The patchy nature of the infection (Fig. 4C) suggests a wider area can be protected within the volume of the brain. Furthermore, there is precedent for a small number of cells playing a major role in striatal function. One such example are the cholinergic interneurons of the striatum which only make up 1-2% of all striatal neurons yet have profound impact on the functional output of the striatum influencing such diverse behaviours as sensorimotor function, sleep & arousal states, learning & memory, anxiety and pain sensations (83).

3-NP exposure activated ARE-mediated gene expression in vitro.
Our in vitro experiments show that 3-NP exposure upregulated Phase 2 gene expression in cultured astrocytes, providing a mechanistic link between 3-NP and Nrf2 activation. The effect of 3-NP was attributed to Nrf2 since a dominant negative version of Nrf2 was able to block the effect of 3-NP. However, further experiments are necessary to determine which component of 3-NP toxicity acts as a signal for Nrf2 activation. For example, Nrf2 activation may result from 3-NP induced ROS/RNS generation, loss of ATP, or even as a direct response to SDH inhibition. In preliminary studies using cultured astrocytes, we examined the level of SDH inhibition caused by the range of 3-NP concentrations used in vitro. After 2 h of exposure to non-Nrf2 inducing 3-NP concentrations (0.3-1 mM), SDH activity was reduced by ~ 90%. However, at Nrf2 inducing concentrations (10-30 mM), SDH activity was almost completely inhibited (~ 98%). Given this difference, it is possible that a non-linear relationship exists between SDH activity and activation of AREmediated gene expression. Conceivably, the requirement for high concentrations of 3-NP to induce Nrf2 mediated gene expression may in part be due the high intrinsic resistance of cultured astrocytes which show an activated phenotype to agents that induce metabolic inhibition and oxidative stress (61).

Conclusion.
Nrf2 plays an important role in regulating neuronal survival during metabolic compromise in vivo. The absence of Nrf2 function in Nrf2 -/mice exacerbates neurodegeneration caused by 3-NP administration. Augmentation of ARE-mediated gene expression in Nrf2-expressing animals using the small molecule inducer, tBHQ, attenuated tissue damage and preserved motor function. Dietary administration of Nrf2 inducers may have profound effects for neuronal viability in neurodegenerative disease, stroke and other related forms of energy deprivation.

Figure 2. Inhibition of brain SDH is not different among Nrf2 genotypes 2 h after acute 3-NP injection.
Four mice from each genotype were injected with a single dose of 60 mg/kg 3-NP or PBS. 3-NP-injected mice were compared with PBS-injected littermates for each experiment. Animals were sacrificed 2 h after 3-NP injection when the peak of SDH inhibition has been shown to occur (47). Mitochondria samples were prepared from striatum and cortex and used in the SDH assay. A 490 was normalized to mitochondrial protein content. Data represent the mean + SEM from n = 4 mice for each genotype and treatment group. * p < 0.05, compared to PBS injected controls from same genotype, one-way ANOVA with Bonferroni post-hoc test. Mice received a tBHQ-supplemented diet for 1 week prior to and during 3-NP administration. Only Nrf2 +/and Nrf2 -/mice were studied since we were interested in observing reduced motor deficit and lesion size with tBHQ treatment, and Nrf2 +/+ mice did not develop these phenotypes. A.) No significant differences were observed in body weight of Nrf2 +/mice between tBHQ and control diet. B.) Nrf2 -/mice never fully adjusted to the tBHQ-supplemented diet where body weight of tBHQ-fed Nrf2 -/mice dropped by 10-20% during 1 week of tBHQ feeding. C.) Control diet and tBHQ-fed Nrf2 +/mice consumed similar amounts of food throughout the experiment. D.) Nrf2 -/mice consumed less food than Nrf2 -/mice with control diet. E.) tBHQ-fed Nrf2 +/mice exhibited reduced motor deficit at the end of the 3-NP treatment regimen. F.) An overall exacerbation of 3-NP induced motor deficits was associated with tBHQ-feeding of Nrf2 -/mice. G, H) tBHQ-feeding prevented lesion development in Nrf2 +/mice, but had no effect on Nrf2 -/mice. Lesions are outlined by dotted lines. For the Nrf2 +/group, n = 8 and n = 6 animals were used for control and tBHQsupplemented diets respectively. For the Nrf2 -/group n = 6 and n = 5 animals were used for control and tBHQsupplemented diets respectively. Data represents mean + SEM. * p < 0.05, ** p < 0.01, compared to Nrf2 +/mice receiving control diet, one-way ANOVA with Bonferroni post-hoc test. . Ad-Nrf2 infected animals developed smaller lesions than Ad-GFP control animals following 3-NP administration. A.) Immunostaining of GFP with the cell-type specific markers for astrocytes (GFAP, top panel), neurons (NeuN, middle panel) and oligodendrocytes (O4, bottom panel) was first performed to characterize the specificity of viral infection in rats in vivo. Scale bars = 10 m. B.) Incidence of co-labeling between GFP and the different markers. N represents the total number of GFP + cells counted for 2 animals. C.) Representative images of GFP expression in the striatum of Ad-GFP infected animals at Bregma (clockwise from top left) 1.00 mm, 0.48 mm, -0.80 mm, and -0.26 mm. Scale bars: 500 m. D.) Analysis of GFP fluorescence as a percentage of total striatal volume (as determined by the area outlined by a propidium iodide counterstain, n = 11 animals) E.) Quantification of lesion volumes in the ipsilateral (right) and contralateral (left) hemispheres. Data represents mean + SEM of n = 5 animals per group. * p < 0.05, compared to Ad-GFP infected controls, two-tailed t-test. F.) Representative cresyl violet stained sections at Bregma 0.70 mm for all indicated treatment groups. Arrow indicates hemisphere receiving the adenoviral vector. Scale bar = 2.5 mm. A.) Cultured cortical astrocytes were transiently transfected with a heat-stable placental alkaline phosphatase (hPAP) reporter construct regulated by an ARE-bearing NQO1 promoter (41). The astrocytes were then treated with varying concentrations of 3-NP (1-30 mM) for 2 h in a glucose-free basic salt solution (BSS vehicle), followed by washout. A doubling of hPAP activity was observed 24 h after 10 or 30 mM 3-NP treatment. B.) Expression of ARE-driven hPAP, with vehicle or 3-NP exposure, was reduced to sub-basal levels when astrocytes overexpressed a dominant-negative form of Nrf2 (Ad-Nrf2DN), but not with overexpression of a GFP control (Ad-GFP) (6). ARE-driven hPAP was induced by 24 h of tBHQ exposure and maximally increased with Ad-Nrf2 infection. Data represent the mean + SEM from three independent experiments. C.) Semi-quantitative RT-PCR confirmed that astrocytes treated with 30 mM 3-NP expressed more mRNAs from prototypic Nrf2 target genes including NAD(P)H:quinone oxidoreductase (NQO1), -glutamylcysteine synthetase (GCLC), and the cystine/glutamate exchanger (xCT). DNA templates were serially diluted to ensure that band intensities were observed within the linear range of the PCR reaction (Suppl. Fig. 2). As controls, Ad-Nrf2 and Ad-Nrf2DN infected cultures were also evaluated with RT-PCR. In all cases, the mRNA levels were induced with Nrf2 overexpression and suppressed with Nrf2DN overexpression. RT-PCR experiments were reproduced three times with similar results. D.) Cultures treated with 30 mM 3-NP produced more total intracellular GSH than vehicle treated controls. GSH induction was blocked with Ad-Nrf2DN infection, but not Ad-GFP control. E.) Western blot analysis shows protein expression of heme oxygenase-1 (HO-1) was induced with 30 mM 3-NP treatment in Ad-GFP infected cultures. Both basal and induced expression of HO-1 was largely blocked with Ad-Nrf2DN infection. D.) Densitometric analyses of HO-1 protein levels. GSH and HO-1 data represent the mean + SEM from three independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001, two-tailed paired t-test. Consistent with previous studies, overexpressed Nrf2-GFP was almost exclusively localized to the nucleus unless co-expressed with its cytoplasmic regulator Keap1 (top row) (13,43). COS-1 cells co-expressing Nrf2-GFP and Keap1-FLAG were evaluated after treatment with 3-NP or BSS vehicle control. The proportion of cells exhibiting nuclear-enriched Nrf2-GFP was significantly higher in 3-NP-treated cultures, when compared to vehicle control. 3-NP also appeared to cause a transient aggregation of some Keap1-FLAG and Nrf2-GFP in the cytoplasm (bottom row). Treatment with 20 M tBHQ for 4 h, as a positive control, also robustly increased nuclear localization of Nrf2-GFP. B.) Summary of Nrf2-GFP localization in response to tBHQ and 3-NP. C.) Both Nrf2-GFP and Keap1-FLAG constructs were functional, as assessed by their ability to augment and suppress ARE-driven hPAP expression, respectively. Data represents three separate experiments by an experimenter blinded to the treatment conditions (total cells evaluated: n = 119 and n = 158, BSS and 3-NPtreated cells, respectively). * p < 0.05, two-tailed paired t-test.  by guest on July 10, 2020