These authors have contributed equally to this work.
Address correspondence and reprint requests to Professor Jean-Pierre Julien PhD, Department of Anatomy and Physiology, Laval University, Research Centre of CHUL, 2705 boul. Laurier, Sainte-Foy, QC G1V4G2, Canada. E-mail: firstname.lastname@example.org
Recent studies suggest that superoxide dismutase (SOD1) may represent a major target of oxidative damage in neurodegenerative diseases. To test the possibility that oxidized species of wild-type (WT) SOD1 might be involved in pathogenic processes, we analyzed the properties of the WT human SOD1 protein after its oxidation in vivo or in vitro by hydrogen peroxide (H2O2) treatment. Using transfected Neuro2a cells expressing WT or amyotrophic lateral sclerosis-linked SOD1 species, we show that exposure to H2O2 modifies the properties of WT SOD1. Western blot analysis of immunoprecipitates from cell lysates revealed that, like mutant SOD1, oxidized WT SOD1 can be conjugated with poly-ubiquitin and can interact with Hsp70. Chromogranin B, a neurosecretory protein that interacts with mutant SOD1 but not with WT SOD1, was co-immunoprecipitated with oxidized WT SOD1 from lysates of Neuro2a cells treated with H2O2. Treatment of microglial cells (line BV2) with either oxidized WT SOD1 or mutant SOD1 recombinant proteins induced tumor necrosis factor-α and inducible nitric oxide synthase. Furthermore, exposure of cultured motor neurons to oxidized WT SOD1 caused dose-dependent cell death like mutant SOD1 proteins. These results suggest that WT SOD1 may acquire binding and toxic properties of mutant forms of SOD1 through oxidative damage.
Amyotrophic lateral sclerosis (ALS) is a lethal neurodegenerative disease characterized by progressive muscle weakness during adulthood. Pathological hallmarks are represented by the massive loss of motor neurons together with astrocytosis and microgliosis in the motor cortex, brainstem, and spinal cord (Cleveland and Rothstein 2001; Julien 2001). Approximately 10% of ALS cases are familial, the remainder ALS cases being diagnosed as sporadic (90%). For the vast majority of ALS cases, familial and sporadic, the etiology remains unknown. The discovery, a decade ago, of missense mutations in the gene coding for the Cu/Zn superoxide dismutase 1 (SOD1) in subsets of familial cases (Rosen et al. 1993) directed most ALS research to elucidating the mechanism of SOD1-mediated disease. To date, 114 different mutations have been discovered in the SOD1 gene that account for ∼20% of familial ALS cases. SOD1 is an abundant and ubiquitously expressed protein. Because of its normal function in catalyzing the conversion of superoxide anions to hydrogen peroxide (H2O2), it was first thought that the toxicity of different SOD1 mutants could result from decreased free radicals scavenging activity. However, different SOD1 mutants showed a remarkable degree of variation with respect to enzymatic activity. Mice expressing mutants SOD1G93A (glycine substituted to alanine at position 93) or SOD1G37R developed motor neuron disease despite elevation in SOD1 activity levels (Cleveland and Rothstein 2001). Moreover, SOD1 knockout mice did not develop motor neuron disease (Reaume et al. 1996). Gene disruption for the copper chaperone for SOD1 that delivers copper to SOD1 catalytic site had no effect on disease progression in mutant SOD1 transgenic mice (Subramaniam et al. 2002). Finally, transgenic mice overexpressing a mutant form of SOD1 lacking two of the four histidine residues coordinating the binding of the Cu at the catalytic site still developed motor neurodegeneration despite a marked reduction in SOD1 activity (Wang et al. 2002). Overall, these studies with genetically altered mice indicate that SOD1 mutants cause motor neuron disease through the gain of new toxic properties that is independent of the enzymatic activity involving the copper catalytic site.
In this study, we report that WT SOD1 acquires binding and toxic properties of mutant forms of SOD1 through oxidative damage. From these findings, we propose that sporadic ALS cases may share with familial ALS a common pathogenic pathway involving misfolding of abnormal SOD1 species.
Plasmids and antibodies
Mammalian expression plasmid carrying human SOD1 tagged with FLAG (pcDNA3-FLAG-SOD1) or mouse chromogranin B (CgB) tagged with hemagglutinin (HA) (pcDNA3-CgB-HA) were generated as previously described elsewhere (Urushitani et al. 2006), rabbit monoclonal anti-human SOD1 (SOD-100), and mouse monoclonal Hsp/Hsc70 antibodies were purchased from StressGen (Victoria, BC, Canada). Rat monoclonal anti-HA (3F10), mouse monoclonal non-phosphorylated neurofilament H (SMI32), and mouse monoclonal anti-actin (C4) antibodies were purchased from Roche (Basel, Switzerland), Steinberger Monoclonal Inc. (Baltimore, MA, USA), and Chemicon (Temecula, CA, USA). Anti-CgB (26102) and cyclooxygenase-IV (A-6431) antibodies were purchased from QED Bioscience (San Diego, CA, USA) and Molecular Probes (Eugene, OR, USA). Mouse monoclonal Anti syntaxin-1 (HPC1) and Akt1 (B-1) were from Santa Cruz (Santa Cruz, CA, USA). Rabbit polyclonal antibody against mouse/rat TGN-38 was generated as previously described (Urushitani et al. 2006).
Cultures, transfection, and drug treatment
Murine neuroblastoma cell line, Neuro2a cells were maintained in Dulbecco’s modified essential medium (DMEM) containing 10% fetal bovine serum. Transfections were performed using Lipofectamin PLUS (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. At 3 h after transfection, the medium was replaced with the nutrient medium containing 2 mmol/L dibutyryl cAMP. At 24 h after transfection, cells were exposed to 1.5 mmol/L H2O2 or sterile water (control) for 45 min for further analysis. Murine microglial cell line BV2 cells were maintained in DMEM/F12 Ham’s medium containing 10% fetal bovine serum. Except from the initial plating, antibiotics were not included in culture medium.
Immunoblotting and Immunoprecipitation
Cultured cells were washed twice in phosphate-buffered saline (PBS) and harvested in TNG-T buffer consisting of 50 mmol/L Tris–HCl pH 7.4, 150 mmol/L NaCl, 10% glycerol, 1% Triton X-100, and a protease inhibitor cocktail (Roche, Mannheim, Germany). After 1 h incubation on ice, cell suspension was centrifuged (20 000 g for 20 min) and the supernatant was collected. To investigate the effect of the treatment of the cell with H2O2, the lysates were incubated with anti-FLAG affinity gel (M2; Sigma, St Louis, MO, USA) for 1 h at 4°C. The immunoprecipitates were eluted in 4% sodium dodecyl sulfate sampling buffer and separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (denaturing-polyacrylamide gel electrophoresis) and transferred to polyvinylidene difluoride membrane (PerkinElmer, Boston, MA, USA) for western analysis. A western blot image was obtained using a western lightning chemiluminescence reagent (PerkinElmer).
In vivo ubiquitination
The modification of human SOD1 by polyubiquitin chain is investigated by in vivo ubiquitination experiment (Urushitani et al. 2002). Neur2a cells were co-transfected with FLAG-tagged hSOD1 (WT or G93A mutant) and HA-tagged ubiquitin. After exposure to H2O2, cells were harvested in TNG-T buffer containing protease inhibitor cocktail in the same protocol mentioned above. The lysates were immunoprecipitated using anti-FLAG affinity gel (M2; Sigma), and eluates from immunobeads were analyzed by western blotting using anti-HA (Roche) and anti-human SOD1 (StressGen) antibodies.
Purification of recombinant proteins
Recombinant glutathione S-transferase fused hSOD1 (GST-hSOD1) was generated as previously reported (Urushitani et al. 2002, 2004). Metallation of hSOD1 was performed by overnight incubation with 0.2 mmol/L of ZnCl2 and 3 h incubation with 0.2 mmol/L of CuCl2, followed by overnight dialysis against PBS. Moreover, the oxidization of the recombinant holo-hSOD1 was performed by 1 h incubation at 20°C with 0.1, 1, and 10 mmol/L of H2O2 followed by subsequent dialysis against PBS overnight. The recombinant proteins were stored at −80°C until use.
Neuro2a cells were transiently transfected with FLAG-tagged SOD1 (WT and G93A) in six-well culture plates. 24 h after transfection, cells were treated with 1.5 mmol/L of H2O2 in DMEM for 30 min and subsequently incubated in the nutrient medium eliminating H2O2 for further 30 min. Harvested cells were homogenized in homogenization buffer (250 mmol/L sucrose, 10 mmol/L Tris–HCl (pH 7.4), 1 mmol/L MgCl2, and protease inhibitor cocktail) and centrifuged for 15 min at 1000 g to exclude debris. The supernatant was further centrifuged at 8000 g for 15 min to separate pellet (heavy membrane fraction) and the supernatant. The supernatant was ultracentrifuged at 105 000 g for 60 min to obtain the cytosolic fraction (supernatant) and the light membrane fraction (pellet). Each pellet was resuspended in the homogenization buffer containing 1% Triton-X100 with brief sonication. Equal amounts of protein were analyzed by western blotting after determination of the protein concentration with Bradford assay (BioRad, Hercules, CA, USA).
Murine microglial cell-line, BV2 cells at 90% confluency in six-well culture plates, were treated with PBS, lipopolysaccharide (LPS) (10 μg/mL), WT SOD1 (10 μg/mL), oxidized WT SOD1 (10 μg/mL), or with G93A SOD1(10 μg/mL) for 24 h. All SOD1 recombinant proteins were previously metallated. Total RNA was extracted from total cell lysates using Trizol reagent (Invitrogen) according to manufacturer’s instructions. First strand cDNA was synthesized from total RNA using reverse transcriptase and oligo-dT primer (Invitrogen). The expression level of tumor necrosis factor (TNF-α), inducible nitric oxide synthase (iNOS), and glyceraldehyde 3-phosphate dehydrogenase was estimated by PCR. The primer pairs used in this experiment are 5′-TCAGTGAGACCACTGCAATG-3′ and 5′-GTGGAGTGAGACTTTGGATG-3′ for TNF-α; 5′-CCTTGTGTCAGCCCTCAGA-3′ and 5′-CACTCTCTTGCGGACCATCTC-3′ for iNOS, and finally 5′-GGCATTGTGGAAGGGCTCA-3′and 5′-TCCACCACCCTGTTGCTGT-3′ for glyceraldehyde 3-phosphate dehydrogenase.
Murine microglial cell-line, BV2 cells at 90% confluency in six-well culture plates, were treated with PBS, LPS (10 μg/mL), WT SOD1 (10 μg/mL), oxidized WT SOD1 (10 μg/mL), or with G93A SOD1 (10 μg/mL) for 24 h. All SOD1 recombinant proteins were previously metallated. Cell culture supernatants were collected and briefly centrifugated at 1000 g to eliminate particulates. TNF-α was measured using mouse TNF-α/TNFSF1A Quantikine immunoassay (MTA00; R & D Systems, Minneapolis, MN, USA) according to the instructions of the manufacturer. ELISA plate was then read in a SpectraMax 340pc plate reader (Molecular Devices, Sunnyvale, CA, USA) and analyzed using SoftMax Pro 3.1.1 software (Molecular Devices).
Primary culture of mouse embryonic spinal cord
Primary dissociated cultures from embryonic mouse spinal cord were prepared as described elsewhere (Urushitani et al. 2002). Twelve days after plating, cultures were treated with recombinant proteins (human WT SOD1, human oxidized WT SOD1, and human G93A SOD1) for 24 h, followed by fixation in 4%p-formaldehyde. The motor neuron viability was estimated by immunocytochemistry using anti-non-phosphorylated neurofilament H (SMI32; 1 : 500). Motor neurons were identified as SMI32-immunoreactive large neurons (>20 μm) with a single long extending axon. Anti-mouse IgG conjugated with Alexa 488 (Invitrogen, Burlington, ON, Canada) was used as secondary antibody. Cultures were observed under fluorescent microscope and four images from randomly selected fields were obtained from three sister cultures. The number of motor neurons was obtained from each field.
Misfolding of WT SOD1 by oxidative stress
To test the hypothesis that SOD1 may be a target protein of oxidative stress in neurodegeneration, we treated bacterially purified recombinant SOD1 proteins (WT, G85R, and G93A) with increasing concentrations (0, 0.1, 1, and 10 mmol/L) of H2O2 for 1 h and analyzed its migration pattern and solubility. Western analysis of total fractions of H2O2-treated recombinant SOD1 showed remarkable change in migration pattern, fragmentation, and high molecular aggregate formation. Without H2O2 or low concentration of H2O2, WT SOD1 was migrated at 17 and 34 kDa, corresponding to monomer and dimer, respectively (Fig. 1a left, lanes 1, 2). However, exposure to higher concentration of H2O2-induced high molecular smear (Fig. 1a left, lanes 3 and 4), which was detected in G85R and G93A mutants exposed to lower concentration of H2O2 (Fig. 1a right). Moreover, ultracentrifugation of the recombinant protein after H2O2 treatment revealed that such oxidation-related species were exclusively detected in pellet fraction, but not in supernatant (Fig. 1b). Only monomer and dimer SOD1 species were detected in supernatant. Although WT SOD1 is less susceptible to oxidative stress than mutant SOD1, the molecular change obtained from western analysis is definitely similar to mutants. These in vitro results indicate that oxidation affects the misfolding and aggregation of WT SOD1 as well as mutant SOD1. Our results are consistent with previous report showing that oxidation of WT SOD1 promotes its aggregation in vitro analyzed by light scattering assay (Rakhit et al. 2004). To further investigate the oxidation of WT SOD1 in vivo, Neuro2a cells were transfected with WT and G93A SOD1 tagged with FLAG peptides at N′-terminus and were exposed to 1.5 mmol/L H2O2 for 45 min at 24 h after transfection. Then, pull-down assay of the transfected cell lysates using anti-FLAG affinity gel revealed that Hsp/Hsc70 was co-immunoprecipitated with mutant G93A, oxidized WT SOD1, but not with non-oxidized WT SOD1 (Fig. 2a). These results suggest that oxidation by H2O2 can cause misfolding of WT SOD1 with ensuing interaction with Hsp/Hsc70.
Oxidized WT SOD1 can be conjugated to poly-ubiquitin
Most types of ALS-related SOD1 mutant proteins are degraded by the ubiquitin-proteasomal pathway (Ciechanover and Brundin 2003; Urushitani et al. 2006). Based on this notion, we performed in vivo ubiquitination experiment to investigate whether oxidation transforms WT SOD1 protein to misfolded species suitable for polyubiquitination. Neuro2a cells were transfected with FLAG-tagged WT or G93A SOD1s together with HA-tagged ubiquitin and then exposed to 1.5 mmol/L H2O2 for 45 min before harvesting. Western blot analysis of total cell lysates and anti-FLAG immunoprecipitates shows that WT SOD1 in H2O2-treated cells was conjugated with multi-ubiquitin chain unlike WT SOD1 from untreated cells. The same phenomenon was observed with G93A mutant SOD1 (Fig. 2b).
Oxidized WT-SOD1 interacts with CgB
We previously identified CgB as a binding partner of mutant SOD1 (Urushitani et al. 2006). Chromogranins were found to interact and to promote secretion of mutant SOD1 species. To test whether oxidized WT SOD1 species can interact with chromogranins, Neuro2a cells were transfected with FLAG-tagged WT or G93A SOD1s together with mouse HA-tagged CgB. At 24 h post-transfection, cells were exposed to 1.5 mmol/L H2O2 for 45 min before harvesting. Western blot analysis of fractionated Neuro2a cell lysates showed that both WT and mutant SOD1 distributed in the microsomal fraction where CgB is abundantly expressed (Fig. 3a). Treatment with H2O2 did not affect protein levels of WT and mutant SOD1 in subcellular fractions. Total cell lysates were immunoprecipitated with anti-FLAG affinity gel and analyzed by western blotting. As shown in Fig. 3b, CgB was co-immunoprecipitated with either G93A SOD1 or oxidized WT SOD1, but not with intact WT SOD1. This result indicates that oxidation of WT SOD1, which distributes in endoplasmic reticulum–Golgi compartments, can induce its binding to chromogranins.
Oxidized WT SOD1 can induce microglial activation and death of motor neurons
There is evidence that both WT and mutant SOD1 species can be secreted (Turner et al. 2005; Urushitani et al. 2006). However, unlike WT SOD1, secreted mutant SOD1 induces proinflammatory molecules such as TNF-α, iNOS, and COX2 (Urushitani et al. 2006). To examine whether oxidized WT SOD1 can mimic mutant SOD1 in activating microglia, murine microglial BV2 cells were exposed to bacterially purified recombinant holo-SOD1 that was treated or not with H2O2. Semi-quantitative RT-PCR experiment revealed that exposure of BV2 cells to either G93A SOD1 or oxidized WT SOD1 induced the expression of TNF-α and iNOS (Figs 4a and b). Moreover, ELISA immunoassay revealed that oxidized SOD1 species-induced TNF-α secretion in the medium of BV2 cells, as well as mutant SOD1 did (Fig. 4c).
Unlike WT SOD1, extracellular mutant SOD1 can induce the death of cultured motor neurons even in absence of microglia (Urushitani et al. 2006). In this study, embryonic spinal cord cultures were exposed to either recombinant WT SOD1, oxidized WT SOD1, or G93A SOD1 (0.5 and 1.0 μg/mL) for 24 h. As expected, WT SOD1 was not toxic to motor neurons in this concentration range. However, oxidized WT SOD1 exhibited toxicity to cultured motor neurons in a dose-dependent manner like G93A mutant SOD1 (Fig. 5). We also observed microglial activation in the mixed culture characterized by large ameboid cell morphology (data not shown). These results indicate that oxidized WT SOD1 acquires the neurotoxic property of mutant SOD1 species.
From the results presented in this study, we conclude that WT SOD1 can acquire through oxidation many of the binding and toxic properties of ALS-linked mutant SOD1. This conclusion is supported by the following results: (i) H2O2-treated recombinant WT SOD1 yielded aggregates similar to those of mutant SOD1 species, (ii) oxidation induces misfolding of WT SOD1 as revealed by the interaction with Hsp/Hsc70 and by the multi-ubiquitination, (iii) oxidized WT SOD1 distributed in membrane fractions and interacted with neurosecretory protein CgB in the transfected neuronal cells, and (iv) extracellular oxidized WT SOD1 triggered microglia activation and death of cultured motor neurons.
There are multiple lines of evidence for involvement of oxidative stress in the pathogenesis of neurodegenerative diseases including Parkinson’s, Alzheimer’s diseases, and ALS (Dauer and Przedborski 2003; Andersen 2004; Lahiri and Greig 2004; Shaw 2005). Considering the abundance of SOD1 comprising 1% of the total protein (Pardo et al. 1995), and its role as antioxidant, it is plausible that SOD1 may be a target of oxidative stress in neurodegenerative disorders. Actually, the oxidation of WT SOD1 is a well-established phenomenon. H2O2 can interact with the active site of WT SOD1 and may inactivate the enzyme through hydroxyl radical production (Hodgson and Fridovich 1975; Yim et al. 1990). Rakhit et al. (Rakhit et al. 2004) found that WT SOD1 possesses four oxidation-prone amino acids (His48, 80, 120 and Phe20) and that their oxidation triggers SOD1 aggregation. In addition, oxidation of cysteine residues in WT SOD1 can also provoke its misfolding and aggregation via intermolecular disulfide (Furukawa et al. 2006).
In this study, we show that Hsp70 interacts with the oxidized WT SOD1 and that multi-ubiquitination of oxidized WT SOD1 species may take place in order to degrade it in the ubiquitin-proteasomal pathway. This result indicates that both oxidized WT SOD1 and mutant SOD1 are misfolded. Such misfolded proteins can form aggregates with inherent cytotoxic properties (Bucciantini et al. 2002). A significant proportion of WT SOD1 or mutant SOD1 species can be translocated into the endoplasmic reticulum–Golgi and secreted (Turner et al. 2005; Urushitani et al. 2006). Our results demonstrate that oxidized WT SOD1 can interact with the neurosecretory protein CgB like the mutant SOD1 forms. Once secreted in the milieu, the extracellular oxidized WT SOD1 may activate microglial cells (Fig. 4) and induce motor neuron death (Fig. 5). Such model of toxicity based on secreted oxidized WT or mutant SOD1 would be compatible with the view that the disease is not strictly autonomous to motor neurons and that multiple cell types may contribute to the disease including motor neurons, interneurons, microglia, and astrocytes (Clement et al. 2003; Boillee et al. 2006). It may also explain how the damage can be propagated from one cell to another, as suggested from the analysis of chimeric mice expressing mutant SOD1 (Clement et al. 2003).
The view that WT SOD1 may acquire toxic properties upon oxidative damage are consistent with the recent report that WT SOD1 expression exacerbated disease in transgenic mice expressing mutant SOD1 forms such as A4V, L126Z, and G93A SOD1 mutants (Deng et al. 2006). It is noteworthy that overexpression of WT SOD1 conferred ALS disease to unaffected A4V SOD1 mice. Accordingly, the authors proposed a model of disease based on the aggregation of WT SOD1 in mitochondria compartment due to oxidation of cysteine residues. However, the view that mitochondria is the primary target of mutant SOD1 cytotoxicity has been questioned by another study indicating that high expression levels of human SOD1 may cause artificial loading of this protein in this organelle (Bergemalm et al. 2006). An alternative explanation would be, as suggested above, that oxidized WT SOD1 species might contribute via chromogranin-mediated secretion to a neurotoxic environment.
Based on these results, the possibility that WT SOD1 may be a contributor of pathogenesis in sporadic ALS must be considered. Moreover, at the light of results presented here, a role for SOD1 in pathogenesis of other neurodegenerative diseases cannot be excluded either. A recent report by Choi et al. (Choi et al. 2005) showed that SOD1 is oxidized in the brain lysates from patients with Alzheimer’s and Parkinson’s diseases. Oxidized SOD1 proteins are also present in senile plaques and Lewy bodies (Choi et al. 2005). Yet, it remains to be resolved how oxidized or mutant SOD1 species can trigger through protein misfolding and aggregation selective neuronal death pathways.
This work is supported by Research grant of Canadian Institute of Health Science (CIHR), ALS association (USA), the ALS Society of Canada and Robert-Packard ALS Research Center. MU is a recipient of Research Fellowship of Uehara Memorial Foundation and post-doctoral fellowship of CIHR. JPJ is recipient of a Canada Research Chair (Tier 1) on neurodegeneration.