Protection from mitochondrial complex II inhibition in vitro and in vivo by Nrf2-mediated transcription

Abstract

Complex II inhibitors 3-nitropropionic acid (3NP) and malonate cause striatal damage reminiscent of Huntington's disease and have been shown to involve oxidative stress in their pathogenesis. Because nuclear factor erythroid 2-related factor 2 (Nrf2)-dependent transcriptional activation by means of the antioxidant response element is known to coordinate the up-regulation of cytoprotective genes involved in combating oxidative stress, we investigated the significance of Nrf2 in complex II-induced toxicity. We found that Nrf2-deficient cells and Nrf2 knockout mice are significantly more vulnerable to malonate and 3NP and demonstrate increased antioxidant response element (ARE)-regulated transcription mediated by astrocytes. Furthermore, ARE preactivation by means of intrastriatal transplantation of Nrf2-overexpressing astrocytes before lesioning conferred dramatic protection against complex II inhibition. These observations implicate Nrf2 as an essential inducible factor in the protection against complex II inhibitor-mediated neurotoxicity. These data also introduce Nrf2-mediated ARE transcription as a potential target of preventative therapy in neurodegenerative disorders such as Huntington's disease.

Mitochondrial complex II inhibition with 3-nitropropionic acid (3NP) or malonate produces a characteristic striatal degeneration similar to that seen in Huntington's disease (HD) (1, 2). HD is an autosomal dominant neurodegenerative disorder that results from a polyglutamine repeat expansion in the first exon of the huntingtin gene (3). Hallmarks of the genetic disease include severe degeneration of striatal medium spiny neurons and progressive choreiform movements (4, 5). Similar behavioral deficits and selective damage to the medium spiny neurons of the striatum with sparing of the aspiny neurons are seen after complex II inhibition (6–10).

Furthermore, there is developing evidence that HD pathogenesis involves mitochondrial dysfunction, excitotoxicity, and subsequent reactive oxygen species (ROS) production (11–13). Mitochondrial deficiencies, including reduced overall respiration and reduced activities of complex II, III, and IV, have been measured in the striatum of postmortem HD brains (14, 15). Similarly, reduced mitochondrial activity has been observed in at least one genetic mouse model of HD (16), and enhancement of electron transport by coenzyme Q10 is effective in genetic models (17–20). HD patients also display increased ROS production in red blood cells and the striatum (21–24), which is reflected in in vitro and genetic mouse models of HD (18, 25–27).

Complex II inhibitors generate ROS (26, 27) as a direct consequence of disruption of the electron transport chain and excitotoxicity by means of calcium influx through the N-methyl-d-aspartate receptor (28–30). Additionally, the high concentration of striatal dopamine may contribute to ROS production and exacerbate the damage caused by complex II inhibition (31). Striatal dopamine depletion attenuates damage caused by either 3NP or malonate in vivo (32). Conversely, enhanced dopamine release by methamphetamine potentiates 3NP (33).

One mechanism by which cells respond to oxidative insult is through the antioxidant response element (ARE), a cis-acting enhancer sequence that regulates the transcription of many cytoprotective genes. Upon toxic insult, glutathione depletion, or chemical activation, the transcription factor nuclear factor erythroid 2-related factor 2 (Nrf2) translocates to the nucleus and dimerizes with small Maf proteins to form a transactivation complex that binds to the ARE. (Nrf2 regulation is reviewed in ref. 34.) Consequently, Nrf2-induced ARE activation coordinates the expression of many genes involved in combating oxidative stress and toxicity in a wide variety of tissues and cell types (35–41). In addition to protecting against chemical insults, carcinogenesis, and aging (41–44), Nrf2 has been shown to directly inhibit Fas-mediated apoptosis, a substrate for caspase-3-like proteases and an effector of PK-like ER kinase-mediated cell survival (45–48).

Previously, we have demonstrated that Nrf2-dependent transcription can prevent ROS-induced apoptosis in neurons and astrocytes in vitro (40, 49–52). Based on evidence of oxidative injury, we hypothesized that Nrf2-dependent ARE activation would play a role in preventing cell death resultant from complex II inhibition. Here we demonstrate that Nrf2 and ARE-dependent signaling are critical mediators of the cellular response to mitochondrial inhibitors in vitro and in vivo. Furthermore, we show that further ARE induction can protect against complex II inhibitor toxicity.

Materials and Methods

Animals. Nrf2^-/- and ARE-human placental alkaline phosphatase (hPAP) transgenic reporter mice were bred separately. Nrf2^-/- mice were created by targeted disruption of the Nrf2 gene (53). ARE-hPAP reporter mice were created by insertion of a 51-bp segment of rat NAD(P)H quinone oxidoreductase-1 promoter containing the core ARE into a minimal promoter upstream of the heat-stable hPAP (54).

Chemicals and Antibodies. 3NP and malonate were purchased from Sigma. Rabbit polyclonal anti-glial fibrillary acidic protein (GFAP) was purchased from DAKO, and monoclonal anti-β-III tubulin was purchased from Promega. Secondary antibodies, vector red alkaline phosphatase substrate, and nuclear fast red were purchased from Vector Laboratories. Fluorojade-B was purchased from Chemicon.

Primary Neuronal Culture. For primary cortical neuronal cultures, Nrf2^+/- mice were bred and cultures were prepared from individual embryonic day 15/16 embryos as described in ref. 49. Treatments were applied between days 3 and 5 in vitro. Immunostaining for β-III-tubulin and GFAP confirmed no difference in the ratio of neurons to astrocytes between genotypes (data not shown). hPAP⁺ cultures were prepared similarly. hPAP activity and histochemistry in primary cultures was measured as described in ref. 54.

Cytotoxicity Measurements. Lactate dehydrogenase (LDH) release into medium was measured by using the CytoTox96 Non-Radioactive Cytotoxicity Assay kit (Promega) according to the manufacturer's instructions. LDH activity was measured on both media and lysed cells. After normalizing to nontreated wells, the percentage of LDH in the media was calculated. Each measurement was made in at least triplicate on separate cultures from three individual pups. TUNEL staining (Roche Applied Science, Indianapolis) also was performed according to the manufacturer's instructions on cultures from at least two different pups per genotype.

Administration of 3NP. Nrf2^-/-, Nrf2^+/-, or Nrf2^+/+ mice received i.p. injections (50 mg/kg) of 3NP or vehicle every 12 h for a total of seven injections. 3NP in PBS (25 mg/ml) was prepared fresh and adjusted to pH 7.4 with 10 M NaOH. Individual doses were diluted so that injection volume remained constant at 200 μl. Six to eight hours after the last dose, mice were killed as described below. One Nrf2 knockout mouse died before the seventh dose, and three others had severe symptoms that were classified as stage III according to Gabrielson et al. (55).

Behavioral Assessment for 3NP-Treated Animals. To assess sensori-motor deficits in 3NP-treated animals, mice were trained up to five times per day on the rotarod (Columbus Instruments, Columbus, OH) for three days before 3NP administration (56). All mice were able to achieve 180 s (5 rpm) on the first trial by day 3. Rotarod function was measured before the first dose and after the last dose of 3NP.

Malonate Injections. Eighteen-week-old Nrf2^-/-, Nrf2^+/-, and Nrf2^+/+ mice received malonate lesions by intrastriatal stereo-taxic injection with contralateral vehicle injections. Mice were anesthetized with isoflurane. Malonate (1 μl, 0.5 M, pH 7.4 in 0.9% NaCl) was injected 0.5 mm anterior to bregma, 2.1 mm lateral to midline, and 3.8 mm ventral to dura. One minute after insertion of the Hamilton syringe, the solution was administered over 2 min, and the needle was withdrawn 2 mm/min. ARE-hPAP⁺ mice were injected similarly with 0.25 M malonate (1 μl) because of differences in background strain sensitivity.

Histological Analysis. Mice were killed with CO₂ and perfused with 4% paraformaldehyde. Tissues were postfixed overnight at 4°C and cryoprotected in 30% sucrose/PBS. By using a cryostat (Leica, Deerfield, IL), adjacent coronal 20- and 50-μm sections were taken for staining with Fluorojade-B and cresyl violet, respectively, through the entire striatum. Degenerating neurons were detected with Fluorojade-B according to the manufacturer's suggested protocol. Lesion size was quantified by using the Cavieleri Estimator Stereo Investigator (MicroBrightField, Williston, VT) on sections 200 μm apart. hPAP histochemistry was measured as described in ref. 54.

Primary Astrocyte Cultures and Transplantation. Primary astrocyte cultures were prepared from postnatal day 1/2 ARE-hPAP or WT pups as described in ref. 49. Nearly all cells (>95%) stained for the astrocyte marker GFAP after 7 days in culture (data not shown). After 5–7 days in culture, astrocytes were infected at 50 or 200 multiplicity of infection (MOI) with recombinant adenovirus (Ad)-Nrf2-GFP or Ad-GFP by the protocol described in ref. 51. At 200 MOI, >95% of all cells expressed GFP after 24 h. At 50 MOI, ≈70% expressed GFP after 24 h.

Twenty-four hours postinfection, 100,000 cells (1 μl) were injected intrastriatally (by using the same coordinates as above) into 13- to 18-week-old ARE-hPAP reporter or WT mice. Two mice received Ad-Nrf2-GFP-infected astrocytes, and two received Ad-GFP-infected astrocytes bilaterally. Two other mice were injected with GFP-infected cells in one hemisphere and Nrf2-infected cells in the other. Mice were allowed to recover for 5 weeks and then were lesioned with 0.5 M malonate as described above. One mouse from each group (bilateral Nrf2, bilateral GFP, and GFP-left/Nrf2-right) received a malonate injection just lateral to the transplantation site (anterior-posterior = +0.5 mm, mediolateral = ± 2.4 mm, dorsoventral = -3.8 mm). Mice receiving bilateral Ad-GFP- or Ad-Nrf2-infected astrocyte transplants were lesioned in the right hemisphere.

Results

Nrf2^-/- Primary Cultures Are More Sensitive to 3NP than Nrf2^+/+ Cultures. To examine the differential sensitivity of Nrf2^-/- neurons to complex II inhibition, Nrf2^+/+ and Nrf2^-/- primary cortical neuronal cultures (3 days in vitro) were treated with 3NP for 48 h and assessed for LDH release. Vehicle-treated cultures showed no difference in LDH release, cellular morphology, or culture composition between genotypes or compared to non-treated controls. At all doses, LDH release was greater in the Nrf2^-/- cultures than in WT cultures. This trend was statistically significant at 2 mM 3NP (Fig. 1A). Nearly all Nrf2^-/- neurons were TUNEL-positive as a result of treatment with 2 mM 3NP, whereas a significantly lower percentage of Nrf2^+/+ neurons was TUNEL-positive, as determined by visual inspection (Fig. 1B). In both Nrf2^-/- and Nrf2^+/+ cultures, TUNEL-positive cells were exclusively neuronal as confirmed by colabeling with the neuronal marker NeuN (data not shown). This result is in agreement with previous reports that neurons are selectively vulnerable to 3NP toxicity (57).

Fig. 1.

Nrf2^-/- primary neuronal cultures are more vulnerable to 3NP. Nrf2^-/- and Nrf2^+/+ primary neuronal cultures were treated with 3NP. (A) LDH release was measured. (B) TUNEL staining was performed. **, P < 0.01 compared with WT mice.

hPAP Activation After Administration of 3NP in Primary Neuronal Culture. To examine whether the ARE is activated in response to complex II inhibition, ARE-hPAP⁺ primary cortical neuronal cultures were exposed to neurotoxic doses of 3NP. hPAP expression in neuronal cultures was significantly increased at 48 h (Fig. 2A). This activation was localized to the surviving astrocytes (Fig. 2B) as visualized by vector red staining and GFAP colabeling.

Fig. 2.

hPAP expression resulting from 3NP treatment in primary neurons. ARE-hPAP primary neuronal cultures were treated with 3NP. (A) LDH release and hPAP activation were measured. (B) hPAP activity was visualized by vector red, and GFAP was labeled with fluorescein.

In Vivo Sensitivity of Nrf2^-/-, Nrf2^+/-, and Nrf2^+/+ Mice to 3NP. We posited that this increased sensitivity would also extend to an in vivo model. Eighteen-week-old mice were dosed every 12 h with 3NP (50 mg/kg). After seven doses, a clear differential sensitivity had emerged, based on the subjective classification described by Gabrielson et al. (55). Seven of eight Nrf2^-/- mice rated at least stage I, three were ranked stage III (end stage), and one mouse died before receiving the seventh dose (Table 1). In the Nrf2^+/- group, only two of six mice rated stage I, and one of six mice from the Nrf2^+/+ group exhibited a stage I phenotype. No Nrf2^+/+ or Nrf2^+/- mice rated above stage I.

Table 1. Phenotypic scoring of mice after administration of 3NP.

Mouse type	Stage 0	Stage I	Stage II	Stage III
Nrf2+/+	5	1	0	0
Nrf2+/-	4	2	0	0
Nrf2-/-	1	3	1	3

Nrf2^+/+, Nrf2^+/-, and Nrf2^-/- mice were scored stage 0, I, II, or III based on the development of clinical symptoms. No. of mice receiving each score is reported.

Four hours after the last dose, motor skills were assessed by rotarod performance. Up to three trials were given for each mouse, and the longest time spent on the rotarod was used for analysis. All PBS-injected mice were able to maintain 5 rpm for 180 s (the maximum time allowed). The only group that exhibited significant deficits compared with the PBS-injected animals was the 3NP-injected Nrf2^-/- group. Furthermore, the knockout mice had significantly reduced function when compared with 3NP-injected Nrf2^+/+ and heterozygous mice (Fig. 3A). Likewise, the only group that differed significantly from their starting weight were the 3NP-injected Nrf2^-/- mice (Fig. 3B), which lost 12 ± 1.9% of their starting weight.

Fig. 3.

Nrf2^-/- mice are more vulnerable to 3NP in vivo. 3NP was administered (n = 6 for Nrf2^+/+ and Nrf2^+/-; n = 8 for Nrf2^-/-). (A and B) Latency to fall on rotarod (A) and weight loss as a percentage of starting weight (B) were measured 4 h after the final injection. (C) Cresyl violet (Left) and Fluorojade-B (Right) were used to visualize lesions. (D) Average lesion volume was calculated on sections 200 μm apart. All data are average ± SEM. *, P < 0.05 compared with WT mice; **, P < 0.01 compared with WT and heterozygous mice.

After mice were killed, brains were stained with cresyl violet or Fluorojade-B (Fig. 3C). Lesions in Nrf2^-/- mice were observable by using both staining methods. However, the lesions were more apparent in the Fluorojade B-stained sections; these sections were used for lesion volume quantification (Fig. 3D). No lesions were found in any of the Nrf2^+/+ mice, and only one Nrf2^+/- mouse had a measurable lesion. Five of the eight treated Nrf2^-/- mice were found to have measurable lesions.

In Vivo Sensitivity of Nrf2^-/-, Nrf2^+/-, and Nrf2^+/+ Mice to Malonate. Because of concerns about possible differences in systemic toxicity, metabolism, clearance, and blood–brain barrier permeability in Nrf2^-/- and Nrf2^+/+ mice, we also examined sensitivity to intrastriatal malonate injection. Ten days postinjection, mice were killed and lesion size was measured by cresyl violet staining (Fig. 4A). None of the mice exhibited an overt behavioral phenotype; however, both Nrf2^-/- and Nrf2^+/- mice had significantly larger lesions than Nrf2^+/+ mice. Average lesion size was increased >21-fold in Nrf2^-/- vs. Nrf2^+/+ mice and 4-fold in Nrf2^-/- vs. Nrf2^+/- mice (Nrf2^+/+, 0.03 ± 0.07 mm³; Nrf2^+/-, 0.16 ± 0.07 mm³; Nrf2^-/-, 0.65 ± 0.27 mm³). Only one Nrf2^+/+ mouse had a measurable lesion. No lesions were found in the contralateral vehicle-injected hemisphere of any mouse (Fig. 4B).

Fig. 4.

Nrf2^-/- mice are more vulnerable to malonate in vivo. Malonate was administered to the right striatum with a contralateral saline control (n = 6 for Nrf2^+/+; n = 8 for Nrf2^+/-; n = 7 for Nrf2^-/-). (A) Representative sections of mice with lesions are shown. (B) Average lesion size was quantified. Data are average ± SEM. *, P < 0.05 compared with WT mice.

Activation of ARE-hPAP Reporter as a Result of Toxin Administration. Malonate was injected intrastriatally into ARE-hPAP reporter mice. After 48 h, the mice were killed, and hPAP histochemistry was performed. Fig. 5 demonstrates that hPAP activity is present near the edges of the lesion (circumscribed by the dashed circles) as determined by cresyl violet and 5-bromo-4-chloro-3-indolyl phosphate/tetranitro-blue tetrazolium staining of serial sections (Fig. 5 A and B). ARE-activated astrocytes, as visualized by GFAP immunohistochemistry, occur in a similar pattern (Fig. 5C)

Fig. 5.

hPAP reporter is expressed in penumbra of malonate-induced lesions. Serial sections from malonate-lesioned ARE-hPAP reporter mice were stained with cresyl violet (A), hPAP histochemistry and nuclear fast red (B), and fluorescein-labeled GFAP (C).

Nrf2 Overexpression in Transplanted Astrocytes Protects from Malonate Lesions. We tested the hypothesis that preactivation of Nrf2 in vivo could protect from lesions caused by complex II inhibitors. To maintain Nrf2 overexpression long term, primary astrocytes infected with Ad-Nrf2-GFP or the GFP control vector were injected into the striatum. Infection rates of the astrocytes approached 100% at 200 multiplicity of infection, as visualized by GFP expression. Only those astrocytes infected with the Nrf2 Ad demonstrated hPAP activity (Fig. 6A). Postinjection survival and migration of astrocytes was monitored by GFP expression. Migration of infected astrocytes away from the needle tract was limited, which is in agreement with published accounts of cortical astrocyte injections (58). After 5 weeks, mice were dosed with malonate as described above. Strikingly, hemispheres receiving Nrf2-infected astrocyte transplants were virtually resistant to malonate, whereas hemispheres receiving control astrocytes were no different from untransplanted controls (Fig. 6 B and C).

Fig. 6.

Ad-Nrf2-infected astrocyte transplants protect from malonate-induced lesions. hPAP⁺ astrocytes were infected with Ad-GFP or Ad-Nrf2-GFP. (A) GFP expression and hPAP histochemistry were visualized. Mice were lesioned 5 weeks posttransplant with malonate. Lesions were visualized by cresyl violet (B) and then were quantified (C). *, P < 0.05 compared with hemispheres receiving GFP-infected astrocytes.

Discussion

In the current study, we have demonstrated the importance of Nrf2-mediated ARE induction due to complex II inhibition. The ARE is a cis-acting sequence in the promoter of many cytoprotective genes. In response to a variety of insults, the transcription factor Nrf2 interacts with the ARE to induce the expression of a multitude of genes, including thioredoxin reductase-1, ferritin, heme-oxygenase-1, and peroxiredoxin (49, 50). These enzymes consequently increase levels of glutathione and NADPH, free-radical scavenging, and other protective pathways. Nrf2^-/- mice, lacking the ability to induce ARE-driven gene expression, are more susceptible to a variety of toxic insults in vivo (35, 38, 59, 60). Furthermore, Nrf2^-/- mice are known to spontaneously develop hemolytic anemia and a lupus-like syndrome relatively late in life (52, 61, 62). The wide array of tissues affected by Nrf2 deficiency suggests a strong role in general cellular protection. This study demonstrates that Nrf2^-/- mice are more susceptible to neurotoxins in vivo.

Previously, we have shown that Nrf2 is a critical determinant of vulnerability to mitochondrial complex I inhibitors and calcium toxicity in vitro (40). Like complex I inhibition, complex II toxicity is known to involve oxidative stress and excitotoxicity (63, 64). Consequently, we hypothesized that ARE-mediated transcription would also be important in protecting against complex II inhibition. Indeed, we found that Nrf2^-/- neurons are more vulnerable to 3NP in vitro. This result is likely due to Nrf2-dependent gene expression changes (40, 49–51). In vivo, Nrf2^-/-, Nrf2^+/-, and Nrf2^+/+ mice revealed vulnerability to 3NP exposure that inversely correlated with the number of intact Nrf2 alleles present.

3NP is typically administered systemically. To ensure that differential sensitivity in the knockout mice was specifically due to lack of Nrf2 and not to other factors that influence toxin delivery to the brain or systemic toxicity, we also assessed vulnerability to local malonate administration. Malonate produced a more uniform lesion while exhibiting the same inverse correlation between lesion size and number of Nrf2 alleles. We have found no indication that metabolism of either 3NP or malonate is influenced by Nrf2 deficiency. It is known that malonate is incorporated into the fatty acid biosynthesis pathway (65). Because fatty acid synthesis genes have not been identified as Nrf2-dependent targets in astrocytes or neurons (40, 49, 51), it is unlikely that malonate metabolism is affected. Very little is known about the clearance of 3NP.

A differential sensitivity to complex II inhibitors between Nrf2^-/- and Nrf2^+/+ mice may exist because of baseline differences in Nrf2-driven gene expression or lack of an inducible protective response. In primary neuronal cultures, we saw that ARE-dependent transcription occurs in surviving astrocytes at toxic doses of 3NP. Furthermore, we found that ARE-dependent transcription also is induced in vivo and is localized to the penumbra of the lesion formed by malonate injection; activated astrocytes also are seen in the penumbra. This observation suggests that in close association with neuronal death due to complex II inhibition, astrocyte populations mount a response that may involve activation of Nrf2 and ARE-dependent signaling.

We proposed that additional ARE activation may provide further protection against complex II inhibition. Astrocytes engineered to express Nrf2 were injected into the striatum 5 weeks before lesioning. A proportion of transplanted cells survived and provided significant protection from malonate lesions. Only the transplanted cells had up-regulated hPAP activity, indicating that a relatively small number of astrocytes overexpressing Nrf2 can protect against an acute insult. After 5 weeks, the ARE activity seen in the transplanted brains is principally due to the transplanted cells and not the transient trauma of injection, which can activate the ARE transiently (data not shown).

The induction of ARE-dependent transcription is an exciting potential tool in the prevention of neurodegeneration. Further study as to the utility and mechanism of Nrf2-mediated protection by cell transplants certainly is warranted. The overwhelming protection seen in acute toxin exposure suggests that these transplants may be beneficial in genetic models of HD, where the insult is chronic and multifaceted. In addition, chemical ARE activators also may be useful in chemical and genetic models of neurodegeneration.

Conclusions

We have shown that Nrf2 deficiency renders mice more susceptible to complex II inhibition, an insult that can activate ARE-dependent transcription. Furthermore, preactivation of the ARE in transplanted astrocytes can dramatically protect against complex II inhibition. Taken together, these data confirm Nrf2 and ARE-dependent signaling as a critical determinant of neurotoxicity both in vitro and in vivo.