Superoxide dismutase 1 mutants related to amyotrophic lateral sclerosis induce endoplasmic stress in neuro2a cells

Authors


Address correspondence and reprint requests to Shin Jung Kang, Department of Molecular Biology, Sejong University, 98 Gunja-dong, Kwangjin-gu, Seoul 143-747, Korea. E-mail: sjkang@sejong.ac.kr

Abstract

One of the common features of damaged neurons in many neurodegenerative diseases is the presence of abnormal aggregates of the disease-related proteins. In amyotrophic lateral sclerosis (ALS) of both sporadic and familial forms, protein aggregates are found in the affected spinal cords. In familial ALS with mutations in copper–zinc superoxide dismutase 1 (SOD1), the propensity of SOD1 for aggregation is known to increase with the mutation. In the present study, we examined whether the aggregate-prone SOD1 mutants induce endoplasmic reticulum (ER) stress and the inhibition of the ER stress protects the cells. The ALS-related mutant G85R SOD1 and G93A SOD1 formed visible aggregates and caused cell death possibly by apoptosis when over-expressed in neuro2a cells. Interestingly, the rate of the mutant SOD1-induced cell death was greater than that of the visible aggregate formation. Expression of the mutant SOD1 caused signs of both early and late ER stress responses, namely, RNA-dependent protein kinase-like ER kinase and eukaryotic initiation factor α phosphorylation, Jun amino-terminal kinase activation, activating transcription factor 6-translocation, X-box binding protein 1 mRNA splicing, and caspase 12 activation. The X-box binding protein 1 mRNA splicing activation was also detected in the mutant SOD1-expressing cells even without the visible aggregates. The cell death induced by the mutant SOD1 over-expression looked like apoptosis as evidenced by nuclear morphology and terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate (dUTP) nick end labeling. Importantly, an ER stress inhibitor, salubrinal delayed the formation of insoluble aggregates of the mutant SOD1 and suppressed the mutant-induced cell death. In addition, over-expression of the ER-targeted Bcl-xL protected the cells from the mutant SOD1-induced cytotoxicity. These results suggest that the misfolding of ALS-related mutant SOD1 induces ER stress possibly prior to the formation of visible aggregates, which may contribute to the motor neuron degeneration in ALS pathogenesis.

Abbreviations used
ALS

amyotrophic lateral sclerosis

ATF6

activating transcription factor 6

BiP/Grp78

immunoglobulin heavy chain binding protein/glucose-regulated protein 78

dUTP

deoxyuridine triphosphate

eIF2α

eukaryotic initiation factor α

ER

endoplasmic reticulum

ERAI

endoplasmic reticulum stress-activated indicator

GFP

green fluorescent protein

JNK

Jun amino-terminal kinase

N2a

neuro2a

PAGE

polyacrylamide gel electrophoresis

PBS

phosphate-buffered saline

PERK

RNA-dependent protein kinase-like endoplasmic reticulum kinase

PI

propidium iodide

RIPA

ristocetin-induced platelet agglutination

SDS

sodium dodecyl sulfate

SOD1

superoxide dismutase 1

TUNEL

terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling

UPR

unfolded protein response

XBP1

X-box binding protein 1

One of the common features of degenerating neurons in many neurodegenerative diseases is the presence of abnormal aggregates of the disease-related proteins inside or outside the affected cells. For instance, Alzheimer’s, Parkinson’s, Huntington’s diseases and amyotrophic lateral sclerosis (ALS) exhibit aggregation and deposition of misfolded proteins such as amyloid β, tau, α-synuclein, polyglutamine-containing proteins, and superoxide dismutase 1 (SOD1) (Ross and Poirier 2004; Lansbury and Lashuel 2006). Proposed mechanisms of toxicity induced by these aggregates include oxidative stress and inhibition of specific organelle function (Shastry 2003; Gonatas et al. 2006), sequestration of essential cellular machineries like chaperones or ubiquitin-proteasome system by the aggregates (Ross and Pickart 2004; Chaudhuri and Paul 2006). However, it is not yet clear whether each toxicity mechanism is either necessary or sufficient for neurodegeneration. Furthermore, it is still under a debate whether the protein aggregate in its final form is a cause for neurodegeneration (Lansbury and Lashuel 2006). Recently, accumulating evidence supports a hypothesis that misfolded proteins exert their toxicity at oligomer or protofibrillar stage (Lansbury and Lashuel 2006; Haass and Selkoe 2007).

Amyotrophic lateral sclerosis is a fatal neurodegenerative disease characterized by selective loss of upper and lower motor neurons (Cleveland and Rothstein 2001). In addition to sporadic forms, the disease has familial forms where mutations in two genes, i.e. SOD1 and alsin, have been identified as causally related (Andersen 2006). Studies using transgenic mice expressing G37R, G85R, or G93A mutants of human SOD1 suggest that the toxicity of the mutant SOD1 is mediated via gain-of-function (Julien and Kritz 2006). In both patients and transgenic mice or transfected cells over-expressing the SOD1 mutants, affected cells exhibit abnormal aggregates containing SOD1 (Bruijn et al. 2004). As in other neurodegenerative diseases with protein aggregation, evidence is accumulating that misfolded aggregates of the mutant SOD1 occupy chaperones and/or protein degradation system (Kabashi and Durham 2006). Another possible mechanism of the mutant-induced toxicity is endoplasmic reticulum (ER) stress as an accumulation of misfolded protein induces ER stress (Rutkowski and Kaufman 2004). Indeed, Tobisawa et al. (2003) observed an up-regulation of immunoglobulin heavy chain binding protein/glucose-regulated protein 78 (Bip/GRP78), an ER-resident chaperone, in the SOD1G93A over-expressing COS7 cells. More recently, there has been a report that caspase 12 which is known to be involved in ER stress (Nakagawa et al. 2000) is up-regulated and cleaved during the disease progression in SOD1G93A transgenic mice (Wootz et al. 2004; Kikuchi et al. 2006). Furthermore, activation of ER stress-related transcription factors has been observed in the end stage SOD1G93A transgenic mice (Kikuchi et al. 2006).

However, other markers of ER stress, such as the chaperone BiP/Grp78 and the transcription factor CCAAT/enhancer binding protein homologous protein/growth arrest and DNA damage-inducible gene, were not altered in the SOD1G93A mice (Wootz et al. 2004), leaving a possible involvement of ER stress in the mutant SOD1-induced neurodegeneration inconclusive. Furthermore, it has not been documented whether suppression of ER stress response is protective against the mutant SOD1-induced toxicity.

As an adaptive response to ER stress because of accumulation of unfolded proteins in the ER lumen, cells activate unfolded protein response (UPR) and ER-associated degradation (Rutkowski and Kaufman 2004; Meusser et al. 2005). Initial stress response involves activation of three ER transmembrane proteins: the serine/threonine kinase Ire1 and RNA-dependent protein kinase-like ER kinase (PERK), and transcription factor 6 (ATF6) (Rutkowski and Kaufman 2004). Activation of PERK attenuates translation via phosphorylation of eukaryotic initiation factor α (eIF2α) to reduce the stress (Rutkowski and Kaufman 2004). ATF6 is cleaved during ER stress and translocates to the nucleus, where it initiates the transcription of chaperone genes and up-regulates X-box binding protein 1 (XBP1) mRNA (Shen et al. 2004). XBP1 mRNA is then spliced by Ire1α during late UPR (Yoshida et al. 2001). Based on this ER stress-dependent splicing, ER stress-activated indicator (ERAI) has been developed by Iwawaki et al. (Iwawaki et al. 2004). When the initial adaptive response fails, cells activate death program by activating Jun amino-terminal kinase (JNK) pathway and/or caspase 12 pathway (Rutkowski and Kaufman 2004; Shen et al. 2004).

In the present study, we present evidence G85R and G93A SOD1 mutants induce ER stress when over-expressed in neuro2a (N2a) neuroblastoma. Mutant SOD1 expression activated both early and late ER stress responses, namely, phosphorylation of PERK, eIF2α, and JNK, nuclear translocation of ATF6, XBP1 mRNA splicing, and caspase 12 activation. In addition, we observed that salubrinal, an ER stress inhibitor, suppressed the aggregation process of the mutant SOD1 and the mutant SOD1-induced cell death. Interestingly, we also found that the rate of cell death was much higher than that of visible aggregate formation. Consistently, signs of ER stress such as ERAI activation and XBP1 splicing were detected in the mutant SOD1-expressing cells even without the visible aggregates. Taken together, our results suggest that ER stress may be one of the toxicity mechanisms involved in neurodegeneration in ALS and support the hypothesis that protein aggregates are not the primary toxic species in ALS.

Materials and methods

Plasmid construction

The construct of human SOD1 (wild type) in pCDNA4 was a kind gift from Dr D. Cleveland (UCSD, San Diego, CA, USA). G85R and G93A mutations were introduced using site-directed mutagenesis kit from Stratagene (La Jolla, CA, USA). The SOD1 cDNAs were then subcloned into pEGFP-N1 or pdsRed-N1 expression vectors from Clontech (Mountain View, CA, USA). Mutations and the plasmid construction were confirmed by sequencing. The construct of ER-targeted Bcl-xL-FLAG was a generous gift from Dr R. S.-Olea (HMS, Boston, MA, USA). ERAI construct was a kind gift from Dr M. Miura (University of Tokyo, Tokyo, Japan).

Cell culture and transfection

Mouse neuroblastoma cell line N2a cells were maintained in Dulbecco’s modified Eagle’s medium (WelGENE, Daegu, Korea) supplemented with 5% fetal bovine serum (WelGENE) in a 37°C CO2 (5%) incubator. For the immunostaining, cells were grown on poly-l-lysine coated 12 mm coverslips in 24-well plates at a density of 2–5 × 105cells/mL. Transient expression of each plasmid (1–2 μg of DNA/35 mm dish) in N2a cells was accomplished with TransIT-LT1 transfection reagent according to the manufacturer’s protocol (Mirus, Madison, WI, USA). After overnight incubation with the transfection reagents, the medium was changed with fresh growth medium. Transfection efficiency at 24 h after incubation was between 20% and 30% in all experiments (Table 1). For propidium iodide (PI) staining, cells were incubated with PI (2 μg/mL) in Hank's buffered salt solution for 15 min, washed three times with Hank's buffered salt solution and then fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS). After washing three times with PBS, the fixed cells were mounted with Slowfade mounting medium from Promega (Madison, WI, USA).

Table 1.   Transfection efficiency of hSOD1-GFP in N2a cellsa
Time after transfection (h)% GFP-positive cells ± SD (n = 5)
Average number of cells per field ± SD (n = 5)
WTG85RG93A
  1. aFor the microscopic analysis of cells of 24 and 48 h transfection, cells were seeded at a density of 5 × 105 cells/mL and for the cells of 72 and 96 h transfection, at a density of 2 × 105 cells/mL.

  2. N2a, neuro2a; WT, wild type; hSOD1, human SOD1.

2420.9 ± 2.120.7 ± 3.119.6 ± 2.9
250.4 ± 28.2222 ± 11.4240.8 ± 42.8
4830.8 ± 2.429.8 ± 3.628.9 ± 2.7
343.2 ± 60.9293.8 ± 28.9296.8 ± 43.8
7224.9 ± 3.716.4 ± 2.715.8 ± 2.2
298.2 ± 44.2229 ± 69.6246.8 ± 29.6
9618.7 ± 1.611.52 ± 2.2210.2 ± 2.9
359 ± 36.3275.8 ± 98.1270 ± 124.9

Subcellular fractionation

After transfection, the cells were washed in cold PBS and then scraped in cold PBS containing 1 mmol/L EDTA. The cells were harvested by centrifugation at 700 g for 5 min. After the centrifugation, the pellet was resuspended in a lysis buffer (10 mmol/L HEPES pH 7.9, 50 mmol/L NaCl, 0.1 mmol/L EDTA, 0.5 mol/L sucrose, 0.5% Triton X-100, and protease inhibitor cocktails) and incubated on ice for 5 min. The supernatants were centrifuged at 700 g for 10 min in a swinging bucket rotor and the resulting supernatants were designated as the extranuclear (cytosolic/membrane) fractions. After washing with the harvest buffer twice, the pellets were added with buffer C (10 mmol/L HEPES pH 7.9, 0.1 mmol/L EDTA, 0.1 mmol/L EGTA, 0.1% Nonidet-P40, 500 mmol/L NaCl, and protease inhibitor cocktails) and vortexed for 15 min at 4°C and then centrifuged at 15 000 g for 10 min. The resulting supernatants were designated as the nuclear extract. To verify the fractionation, each fraction was subjected to immunoblotting for histone H1 as a nuclear marker using anti-histone H1 (1 : 1000, Chemicon, Temecula, CA, USA) and α-tubulin as a cytosolic marker using anti-α-tubulin mouse monoclonal antibody (1 : 4000, Sigma, St Louis, MO, USA).

Fractionation of RIPA-soluble versus RIPA-insoluble materials

To localize SOD1 biochemically during the process of aggregate formation, SOD1-transfected cells at various times were washed three times with cold PBS and harvested with ristocetin-induced platelet agglutination (RIPA) buffer (50 mmol/L Tris–HCl, pH 7.4, 150 mmol/L NaCl, 0.1% sodium dodecyl sulfate (SDS), 1% Nonidet-P40, 0.5% sodium deoxycholate, 5 mmol/L EDTA, and protease inhibitor cocktails). After incubation on rotating shaker for 15 min at 4°C, the lysates were centrifuged at 12 000 g for 20 min at 4°C. The resulting pellet was designated as RIPA-insoluble and the supernatant, as RIPA-soluble fractions. The pellet was dissolved and boiled in SDS sample buffer and spun down briefly to remove undissolving materials before loading onto a SDS–polyacrylamide gel electrophoresis (PAGE) gel. One-tenth of the soluble fraction and a half of the insoluble fraction were processed for immunoblotting for the detection of SOD1. Percent soluble fraction was expressed as 100 × 10 × S/(10 × + 2 × I), where S is mean densitometric readings of band intensities of soluble fraction calibrated with those of tubulin and I is mean densitometric readings of insoluble fraction.

Immunoblotting

Neuro2a cells (106/mL) grown on 60 mm dish were washed with cold PBS and lysed by incubating in 2× SDS sample buffer and boiled at 95°C for 5 min. For the detection of phosphorylated proteins, Na3VO4 (1 mmol/L) plus NaF (1 mmol/L) were added in the washing and lysis buffers. Fifty micrograms of total protein was subjected to 10% SDS–PAGE. Proteins were then transferred onto immobilon-P membrane (Millipore, Bedford, MA, USA) and incubated with blocking solution containing 5% skim milk in Tris-buffered saline Tween-20 (50 mmol/L Tris, pH 7.5, 150 mmol/L NaCl, and 0.05% Tween 20) for 1 h at 25°C. The membrane was then incubated with primary antibody in blocking solution at 4°C overnight. After washing three times with Tris-buffered saline Tween-20 for 10 min each, the membrane was incubated with horseradish peroxidase-conjugated anti-rat, anti-mouse or anti-rabbit IgG (Sigma) for 40 min at 25°C. After washing three times, the bound antibody was revealed using the ECL western blotting reagent kit (Amersham, Arlington Heights, IL, USA). To evaluate the differences in protein levels, the exposed films were densitometrically analyzed by BIS303 imaging system using GelQuant software (DNR, Jerusalem, Israel).

Antibodies and other reagents

The primary antibodies used in this study are as follows: anti-eIF2α rabbit monoclonal antibody (1 : 2000, Cell Signaling, Beverly, MA, USA), anti-eIF2α phospho-specific mouse monoclonal antibody (1 : 2000, Stressgen, Ann Arbor, MI, USA), anti-JNK rabbit monoclonal antibody (1 : 2000, Cell signaling), anti-active JNK (1 : 5000, Promega), anti-PERK (Thr980) rabbit polyclonal antibody (1 : 1000, Cell signaling), anti-phospho-PERK (Thr980) rabbit polyclonal antibody (1 : 1000, Cell Signaling), anti-Bcl2 mouse monoclonal antibody (1 : 500, Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-Bax rabbit polyclonal antibody (1 : 500, Chemicon), anti-histone mouse monoclonal antibody (1 : 1000, Chemicon), anti-human SOD1 rabbit polyclonal antibody (1 : 2000, Chemicon), anti-α-tubulin mouse monoclonal antibody (1 : 4000, Sigma), anti-ATF6 rat monoclonal antibody (1 : 100, BD Bioscience, San Diego, CA, USA), anti-caspase 7 rabbit polyclonal antibody (1 : 1000, Chemicon) and caspase 12 rabbit polyclonal antibody (1 : 1000, Cell Signaling). Other antibodies and chemicals were purchased from Sigma unless stated otherwise.

Immunocytochemistry

After transfection, cells were fixed in 4% paraformaldehyde in PBS, followed by a permeabilization with 0.2% Triton X-100 in PBS for 15 min. The cells were then blocked with PBS containing 10% normal goat serum for 1 h at 25°C. After blocking, cells were incubated with primary antibodies diluted in PBS containing 5% normal goat serum at 4°C overnight. The samples were washed for 5 min in PBS containing 0.1% Triton X-100 for three times. They were then incubated with matching secondary antibodies conjugated with biotin. After washing, the biotin-decorated samples were incubated with Texas red- or FITC-conjugated streptavidin (Promega) in PBS for 30 min at 25°C. After washing, the samples were mounted with mounting medium (Slowfade kit with DAPI, Promega). Samples were examined under a fluorescence microscope (Axioplan 2, Zeiss, Oberkochen, Germany) and the images were analyzed using Axiovision software. Confocal images were obtained using LSM 510 microscope and analyzed using LSM image examiner (Zeiss).

TUNEL assay

To detect apoptotic nuclei, cells were fixed with PBS containing 4% paraformaldehyde for 10 min at 25°C and then post-fixed with ethanol : acetic acid (2 : 1) solution for 15 min. After washing with PBS, the cells were processed for terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) using Intergen ApopTag kit according to the manufacturer’s manual. After final washing, the samples were mounted with mounting medium (Slowfade kit with DAPI, Promega).

Caspase activity assay

Caspase 12 activity was measured using caspase 12 activity assay kit comprised of cell lysis buffer and fluorogenic caspase 12 substrate, ac-ATAD-afc according to the manufacturer’s manual (BioMax, Mountain View, CA, USA). For the assay, 100 μg of the lysis buffer-soluble cell extracts was used for each of 96 wells and the fluorescence was monitored using SpectraMax GEMINIEM spectrofluorometer (Molecular Devices, Sunnyvale, CA, USA). Relative activity was calculated from the slopes of readings for 20 min at 20-s intervals.

Statistics

For the statistical analysis, all the experiments were repeated at least three times. The results were expressed as mean ± SD of at least three independent experiments, unless stated otherwise. Paired data were evaluated by Student’s t-test. A value of p < 0.05 was considered significant.

Results

Aggregate formation of ALS-linked mutant SOD1 and cell death induction in N2a cells

First, we examined whether the mutant SOD1 forms aggregates when over-expressed in N2a cells. Aggregates were defined as bright non-homogeneities with the cytoplasm. To quantify aggregate formation, the intensity of the even and diffuse fluorescence was set as threshold and the number of cells in each field showing fluorescence intensities over the threshold was counted using imaging software. Following the thresholding analysis, the presence of aggregates was confirmed by eye in each field. The G85R and G93A mutant SOD1-green fluorescent protein (GFP) fusion proteins formed aggregates whereas the wild type SOD1 protein was distributed evenly in the cytoplasm (Fig. 1a). The aggregates were observed as early as at 8 h after the transfection. During the early incubation period, two types of aggregates were observed. One was scattered, irregular filamentous aggregate and the other was condensed, perinuclear aggregate (Fig. 1, lower panels with higher magnifications). After about 48 h, condensed aggregates were the major species. The amount of the total aggregates increased over incubation time up to 72 h (Fig. 1b). After that period, it was difficult to discriminate aggregates because most of the mutant SOD1-expressing cells were too shrunken. To examine whether the over-expression of the mutant SOD1-induces cell death, the transfected cells were incubated with PI that penetrates into only dying or dead cells. The rate of cell death was examined by counting PI-positive cells among GFP-positive cells. Transfection efficiency was between 20% and 30% and the expression of the hSOD1-GFP was relatively well maintained until 96 h after the transfection (Table 1). As shown in Fig. 1c, the mutant SOD1-expressing cells were shrunken and stained with PI whereas the wild type-expressing cells looked healthy. The cell death rate of the mutant SOD1-expressing cells increased over incubation time (Fig. 1d). At 96 h after the transfection, 70–80% of the mutant SOD1-expressing cells died while the wild type-expressing cells maintained the control viability. Interestingly, the rate of cell death was greater than that of the aggregate formation in the cells over-expressing the SOD1 mutants, suggesting the aggregates may not be the primal cause of cell death.

Figure 1.

 Aggregate formation of amyotrophic lateral sclerosis (ALS)-linked mutant superoxide dismutase 1 (SOD1) and cell death induction in neuro2a (N2a) cells. (a) Expression of the wild type (WT) and mutant SOD1 proteins (G85R and G93A) fused to GFP in N2a cells at 18 h after transfection is shown. The presence of aggregated proteins only in the mutant SOD1-expressing cells. Arrows indicate scattered filamentous aggregates and arrowheads, condensed aggregates. Top panels, 20×; bottom two panels, 100×. (b) The transfected cells containing aggregates were quantified at indicated post-transfection times by thresholding analysis. Here total aggregates including both filamentous and condensed forms were counted. The rate of aggregates was expressed as number of cells with aggregates versus total green cells. (c) At indicated times after the transfection, cell death was evaluated by staining the live cultures with propidium iodide (PI) that specifically stains dying or dead cells. Only the mutant SOD1-expressing cells showed the aggregates and PI-positive staining. Compare the arrowhead-indicated cells. 2× magnified view of the PI-positive cells are shown on the far right column (magnified). (d) The rate of cell death was calculated by counting PI-positive green cells versus total green cells.

Activation of kinases involved in ER stress by mutant SOD1 over-expression

Although the SOD1 is a cytosolic protein, we hypothesized the aggregate-prone mutant SOD1 may induce UPR where signaling events are shared with the UPR induced by proteins synthesized in the ER lumen. Furthermore, Kikuchi et al. (2006) observed the presence of mutant SOD1 in the ER in the G93A mice, raising the possibility that the mutant SOD1-induces ER stress in the N2a cells as well. The N2a cells were transfected with wild type or mutant SOD1 constructs and the status of kinases involved in UPR was monitored by immunoblot assay using phospho-specific antibodies against PERK, eIF2α, and JNK at 24 h after the transfection. Expression of the transgenic SOD1 was confirmed by immunoblot using anti-human SOD1 antibody. As shown in Fig. 2a, PERK that acts as a proximal sensor of UPR was more strongly phosphorylated in the mutant-expressing cells when compared with the wild type-expressing cells. A PERK substrate, eIF2α was also phosphorylated more strongly in the mutant-expressing cells (Fig. 2a). In addition, phosphorylation of JNK, another signaling molecule in ER stress was detected more strongly in the mutant SOD1-expressing cells (Fig. 2a). Considering the transfection efficiency and a possible non-synchronized UPR signaling induced by the mutant SOD1s among cells, small but consistent increase of phosphorylation of these signaling molecules in independent set of experiments suggests that these molecules are indeed activated by the mutant SOD1 over-expression. Taken together, these data indicate that some of the upstream components of UPR signaling pathway were activated by the over-expression of the mutant SOD1.

Figure 2.

 Activation of endoplasmic reticulum (ER) stress signaling by mutant superoxide dismutase 1 (SOD1) over-expression. (a) Neuro2a (N2a) cells were transfected with GFP constructs of wild type (WT) human SOD1 (hSOD1), mutant SOD1 (G85R and G93A) for 24 h. To examine the activation of early activating kinases involved in ER stress, the transfected cells were processed for immunoblot with antibodies specific for phosphorylated forms of RNA-dependent protein kinase-like endoplasmic reticulum kinase (PERK), Jun amino-terminal kinase (JNK), and eukaryotic initiation factor α (eIF2α). Loading and expression of the constructs were confirmed by probing with anti-tubulin and anti-hSOD1 antibodies, respectively. Consistent results were obtained in three independent experiments. Densitometric readings of the immunoblot bands from independent experiments as fold changes are shown in the lower three panels. The relative band intensities are expressed as mean ± SD (n = 3, *p < 0.05, wild type vs. G85R or G93A). (b and c) Cleavage of activating transcription factor 6 (ATF6) and translocation to the nucleus in the mutant SOD1-expressing cells. (b) The transfected cells as in (a) were fractionated into nuclear and extranuclear fractions. The fractionated samples were processed for immunoblot using antibodies specific for activated forms (ATF6 p50) and full-length (ATF6 p90) ATF6. Loading and nuclear fractionation were confirmed by probing with anti-tubulin and anti-histone H1 antibodies, respectively. Densitometric readings from three independent experiments are shown as in (a). (c) Representative confocal images showing the effect of mutant SOD1 on ATF6 translocation. ATF6 translocated into the nucleus in the mutant SOD1 – but not in the wild type-expressing cells. About 65% of the mutant SOD1-expressing cells showed nuclear ATF6 immunoreactivity. (d and e) Activation of endoplasmic reticulum stress-activated indicator (ERAI) in mutant SOD1-expressing cells. (d) N2a cells were double-transfected with ERAI (XBP1-venus) and pdsRed-hSOD1 plasmids. At 24 h after the transfection, cells were fixed, stained with DAPI and analyzed under a fluorescence microscope. The expression of Venus protein in the mutant SOD1 (G85R)-expressing cells. (e) Percentage of Venus/SOD1 double-positive cells among SOD1-positive cells. Significantly greater number of cells was double-positive in mutant SOD1 (G85R and G93A)-transfected cells compared with those of wild type SOD1 (WT). Results are expressed as mean ± SD (n = 4, *p < 0.05, wild type vs. G85R or G93A).

Translocation of ATF6 into nucleus in the cells expressing the mutant SOD1

It has been known that ATF6 on the ER is cleaved in the cytoplasmic side and the resulting fragment (50 kDa) translocates into the nucleus, where it functions as a transcription factor for the chaperones like BiP/GRP78 or other proteins that can relieve ER stress (Rutkowski and Kaufman 2004). Thus, we examined whether the mutant SOD1-induces cleavage of ATF6 into 50 kDa fragments and its translocation into the nucleus. The SOD1-expressing cells were fractionated into nuclear and extranuclear fractions and they were analyzed by immunoblot assay using anti-ATF6 antibody. As shown in Fig. 2b, the level of 50 kDa fragment ATF6 in the nuclear fraction significantly increased in the G85R or G93A SOD1-expressing cells when compared with the wild type expressing cells. In accordance with this result, the level of full-length ATF6 decreased in the mutant-expressing cells, suggesting a cleavage and subsequent translocation of the ATF6 into the nucleus. When examined by confocal microscopy, the nuclear translocation of the ATF6 was evident in the mutant SOD1-expressing cells (Fig. 2c). Of note, the translocation of ATF6 was detected in the mutant-expressing cells with or without the SOD1 aggregates. These results further suggest that UPR is activated by over-expression of the ALS-inducing SOD1 mutants.

Activation of ERAI in mutant SOD1-expressing cells

It is documented that the XBP1 mRNA splicing occurs as a result of Ire1α activation in the late stage of the UPR and enhances the further transcription of many proteins involved in the UPR (Yoshida et al. 2001). Based on this phenomenon, Iwawaki et al. (2004) developed ERAI construct composed of a part of XBP1 cDNA fused with a GFP variant, Venus cDNA. Under normal conditions, Venus is not expressed but during ER stress when the XBP1 mRNA undergoes splicing, a frame shift occurs in the ERAI mRNA and the Venus is expressed. Like XBP1 protein, Venus then translocates into the nucleus (Iwawaki et al. 2004). To examine whether XBP1 mRNA splicing occurs in the mutant SOD1-expressing cells, ERAI and pdsRed-SOD1 constructs were co-transfected into N2a cells and the expression of the Venus was examined under a fluorescence microscope. As shown in Fig. 2d and quantified in Fig. 2e, about 70% of the mutant SOD1-transfected cells expressed the Venus protein whereas only 10% of the wild type SOD1-expressing cells showed the expression of the Venus. The translocation of the Venus was also evident in the mutant SOD1-expressing cells. Interestingly, the mutant SOD1-expressing cells exhibited the expression of Venus regardless of the presence of the SOD1 aggregates (compare the middle and right panels of Fig. 2d).

Changes in the expression level of the ER chaperone proteins and cell death regulators

The ATF6 translocated into the nucleus functions as a transcription factor for ER chaperones like BiP/Grp78 and calreticulin (Rutkowski and Kaufman 2004). Thus, we examined whether the over-expression of the mutant SOD1 increases the ER chaperone, BiP/Grp78. N2a cells were transfected with either wild type or mutant SOD1 and the level of BiP/Grp78 was examined by immunoblot assay at 24 h after the transfection. The level of BiP/Grp78 slightly but consistently increased in the mutant SOD1-expressing cells compared with wild type-expressing cells (Fig. 3a). The ATF6 in the nucleus is also known to up-regulate the expression of CCAAT/enhancer binding protein homologous protein/growth arrest and DNA damage-inducible gene, which in return down-regulates Bcl-2 expression. In addition, the activated JNK is known to up-regulate the expression of Bax (Rutkowski and Kaufman 2004). To examine whether the levels of these cell death regulators change in the mutant SOD1-expressing cells, the wild type and the mutant SOD1-expressing cells were analyzed by immunoblot assay using anti-Bcl-2 and anti-Bax antibodies. As shown in Fig. 3b, the level of Bcl-2 slightly decreased while that of Bax increased during the 48–72 h transfection period.

Figure 3.

 Induction of apoptosis by the mutant superoxide dismutase 1 (SOD1) over-expression. (a and b) Changes in the expression level of the endoplasmic reticulum (ER) chaperone proteins and cell death regulators. (a) Neuro2a (N2a) cells were transfected with wild type (WT) and the mutant SOD1 (G85R and G93A) and the expression of immunoglobulin heavy chain binding protein/glucose-regulated protein 78 (BiP/Grp78) (BiP) was examined by immunoblot assay at 24 h after the transfection. The level of BiP/Grp78 slightly increased in the mutant SOD1-expressing cells. (b) The expression levels of Bcl-2 and Bax were examined in the cells transfected as in panel (a) The level of Bcl-2 decreased in the G85R mutant expressing cells while that of Bax increased in cells expressing the mutant SOD1 at 72 h after the transfection. (c) N2a cells were transfected with GFP constructs of wild type (WT) human SOD1 (hSOD1), mutant SOD1 (G85R and G93A) for 24 h. To examine the activation of caspases involved in ER stress, the transfected cells were processed for immunoblot using anti-caspase 7 and anti-caspase 12 antibodies detecting both full length (FL) and active forms. Loading and expression of the constructs were confirmed by probing with anti-tubulin and anti-hSOD1 antibodies, respectively. Stronger activation of caspase 7 and 12 in the mutant expressing cells. (a–c) Consistent results were obtained in at least three independent experiments. Densitometric readings of the relative band intensities are shown at the bottom of each blot expressed as mean ± SD (n = 3, *p < 0.05, wild type vs. G85R or G93A). (d) To examine the mode of cell death induced by mutant SOD1, the transfected cells were processed for terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) and counter-stained with DAPI. The mutant SOD1-expressing cells are positive for TUNEL (arrow head) and show condensed and fragmented nuclei (asterisks, magnified).

Activation of caspases involved in ER stress and the induction of TUNEL-positive cell death in the mutant SOD1-expressing cells

It has been documented that the activated Ire1α binds to Tumor Necrosis Factor receptor-associated factor 2 and the complex induces the clustering and activation of caspase 12 (Yoneda et al. 2001). As previous data in this study indicate that the ER stress is induced by the over-expression of the mutant SOD1, we examined whether caspase 12 was activated in the mutant expressing cells. N2a cells were transfected with wild type or the mutant SOD1 and then the cell lysates were analyzed by immunoblot assay. As shown in Fig. 3c, the level of the caspase 12 active fragment significantly increased in the mutant-over-expressing cells. It is also known that the activated caspase 7 participates in the processing of caspase 12 (Momoi 2004). In accordance with this, activation of caspase 7 was detected in the mutant-expressing cells (Fig. 3c).

As the activation of caspase 12 and 7 was detected, it can be expected that the cells die by caspase-mediated apoptosis. Indeed, when the cells were examined for the signs of DNA fragmentation, the mutant-expressing cells were TUNEL-positive (Fig. 3d) and the nuclei showed typical apoptotic morphology, i.e. condensed and fragmented, as shown in the magnified pictures indicated by an asterisk in Fig. 3d.

Suppression of the aggregate-forming process and the cell death by an ER stress inhibitor, salubrinal

Since the previous data in this study indicate over-expression of the mutant SOD1-induces ER stress and subsequent apoptosis, there is a possibility that the inhibition of the ER stress may protect the mutant-expressing cells. To test this possibility, the effect of an ER stress inhibitor, salubrinal was examined in the mutant SOD1-expressing cells. Salubrinal is reported to specifically inhibit eIF2α dephosphorylation and thus inhibits the propagation of ER stress (Boyce et al. 2005). Interestingly, the mutant SOD1-expressing cells with salubrinal treatment showed mostly punctate filamentous aggregates whereas the untreated cells showed condensed perinuclear aggregates (Fig. 4a). The filamentous aggregates found in the salubrinal-treated cells looked similar to those of early time points (Figs 1a and 4a). Quantification of the filamentous versus condensed aggregates in Fig. 4b suggests that the mutant aggregates may develop from filamentous to condensed ones over incubation period. At 24 h after the transfection, about half of the cells with aggregates showed the punctate filamentous morphology but at 48 h, most of the aggregates were condensed ones. However, with the salubrinal treatment, the aggregates mostly remained as filamentous ones at 48 h after the transfection (Fig. 4b). The mutant SOD1s in the aggregates could be biochemically fractionated into RIPA-soluble and insoluble fractions. At 24 h after the transfection, more than 90% of the mutant SOD1 was RIPA-soluble but at 48 h, only about 50–60% was RIPA-soluble (Fig. 4c). Considering the amount of total aggregates are not much different in the cells between 24 and 48 h after the transfection (Fig. 1b), the decrease in the RIPA solubility may be due to the changes in the nature of aggregates, i.e. from filamentous to condensed forms. Of note, it is highly possible that the amount of insoluble fraction was underestimated as there were insoluble materials even after boiling in SDS–PAGE sample buffer that could not be loaded onto a gel. In that case, the solubility of the mutant SOD1 at 48 h may be even greater.

Figure 4.

 Suppression of the aggregate-forming process by an endoplasmic reticulum (ER) stress inhibitor, salubrinal. (a) Morphology of the mutant superoxide dismutase 1 (SOD1) aggregates at 24 and 48 h after transfection with or without salubrinal treatment. Filamentous aggregate at 24 h is shown in the top left and condensed aggregate at 48 h, in the top right panel. In the salubrinal-treated cells, most of the aggregates were filamentous at 48 h (bottom panel). The aggregate species was discriminated by analyzing the captured images by eye. (b) Quantification of the suppression of the aggregation process by salubrinal. Asterisks; not detected. Salubrinal treatment increased the proportion of filamentous aggregates at 48 h. (c) Solubility of the mutant SOD1 aggregates was examined over incubation time. At 24 and 48 h after the transfection, the transfected cells were lysed in ristocetin-induced platelet agglutination (RIPA) buffer and one-tenth of the soluble and a half of the insoluble fractions were processed for immunoblot assay as shown in the upper panels. Percentage of soluble fractions in the total lysates was calculated from 10 times of the densitometric readings of the bands in soluble and two times of the readings in insoluble fractions (n = 3, *p < 0.05, wild type vs. G85R or G93A). The readings from the soluble fraction SOD1 were calibrated with the readings of the tubulin blots. The RIPA-solubility of the mutant SOD1 significantly decreased at 48 h after the transfection. Asterisks indicate non-specific band detected occasionally. (d) The suppression of the aggregation process by salubrinal was examined biochemically as in (c). A significantly greater amount of the mutant SOD1 was found in the soluble fraction with the salubrinal treatment.

As we observed more aggregates became filamentous upon salubrinal treatment, we examined the effect of the drug in the detergent solubility of the mutant SOD1s. Interestingly, when the cells incubated with salubrinal, considerable amount of the mutant SOD1 (80–90%) was still detected in the soluble fraction at 48 h after the transfection (Fig. 4d). These results suggest that the inhibitor of ER stress, salubrinal, suppressed the aggregate-forming process.

Then we examined whether the suppression of ER stress by salubrinal protects the cells from the mutant SOD1-induced cell death. To examine the effect of salubrinal on cell viability, N2a cells were transfected with the SOD1 constructs and treated with salubrinal at 4 h after the transfection. At 48, 72, and 96 h after the transfection, the cells were stained with PI and the rate of cell death was calculated. As shown in Fig. 5a and quantified in Fig. 5b, the mutant SOD1-expressing cells died as evidenced by PI uptake and cell shrinkage, which was significantly inhibited by salubrinal treatment. This result implies that the failure of ER stress management induced cell death when the mutant SOD1 was over-expressed.

Figure 5.

 Suppression of the mutant superoxide dismutase 1 (SOD1)-induced cell death by salubrinal. (a) To examine the effect of salubrinal on cell death induced by the mutant SOD1, cells transfected with wild type (WT) or mutant SOD1 (G85R and G93A) were treated with salubrinal at 4 h after the transfection. The cells were stained with propidium iodide (PI) for the visualization of the dying or dead cells at 48, 72, and 96 h after the transfection (cells at 72 h are shown). The rate of cell death with or without salubrinal is quantified in (b). Results are expressed as mean ± SD (n = 4, *p < 0.05, untreated vs. salubrinal). (c) To gain insight into the protective mechanism of salubrinal, changes in the eukaryotic initiation factor α (eIF2α) phosphorylations level (p-eIF2α) were examined with or without salubrinal treatment in the hSOD1-over-expressing cells at indicated time after the transfection. Densitometric readings of the relative band intensities are shown at the bottom expressed as mean ± SD (n = 3, *p < 0.05, wild type vs. G85R or G93A). (d) To examine the changes in the caspase 12 activity, the hSOD1-transfected cells incubated for 48 h were assayed for the ac-ATAD-afc cleavage. Expression of the mutant SOD1s induced the activation of caspase 12 but salubrinal suppressed the activation significantly. Results are expressed as mean ± SD (n = 3, *p < 0.05, untreated vs. salubrinal).

To gain insight into the protective mechanism of salubrinal against the mutant SOD1-induced toxicity, the status of eIF2α was examined with or without salubrinal treatment. As shown in Fig. 5c, the level of eIF2α phosphorylation induced by the mutant SOD1 over-expression was decreased back to the control level at 72 h after the transfection. In the salubrinal-treated cells, however, the phosphorylation of eIF2α was maintained high even at 72 h after the transfection. This result suggests that salubrinal indeed inhibited dephosphorylation of eIF2α, allowing the cells to attenuate translation and operate salvage programs successfully. In accordance with the protection from the mutant SOD1-induced cell death by salubrinal, the activation of caspase 12 as examined by fluorogenic substrate cleavage was also inhibited in the salubrinal-treated cells. Taken together, these results suggest that salubrinal protected the cells from the mutant SOD1-induced death by suppressing the UPR propagating into death signaling.

Suppression of the mutant SOD1-induced cell death by coexpression of ER-targeted Bcl-xL

It has been reported that the over-expression of Bcl-xL or ER-targeted Bcl-xL protects cells from ER stress-induced damage (Morishima et al. 2004). To examine whether over-expression of ER-targeted Bcl-xL shows protection against the mutant SOD1-induced toxicity, cell death rate was compared in the SOD1-expressing cells with or without coexpression of ER-Bcl-xL Following transfection with SOD1 only or SOD1 plus ER-Bcl-xL, cells were stained with PI at 48, 72, and 96 h after the transfection. As shown in Fig. 6a and quantified in Fig. 6b, ER-Bcl-xL significantly suppressed the cell death induced by mutant SOD1 expression. When we examined the effect of Bcl-xL that is not specifically targeted to ER, the protection rate was similar to that of ER-Bcl-xL (data not shown). These results further suggest that ER stress is one of the major route by which the mutant SOD1 exhibits its toxicity when over-expressed in N2a cells.

Figure 6.

 Suppression of the mutant superoxide dismutase 1 (SOD1)-induced cell death by coexpression of ER-targeted Bcl-xL. (a) To examine the effect of ER-Bcl-xL coexpression on the mutant SOD1-induced toxicity, cells were transfected with SOD1 only or SOD1 plus ER-Bcl-xL. The transfected cells were propidium iodide (PI)-stained at 48, 72, and 96 h after the transfection (cells at 72 h are shown) for the visualization of the dying or dead cells. The coexpression of ER-Bcl-xL protected cells from mutant SOD1-induced toxicity. The rate of cell death was quantified in panel (b). Results are expressed as mean ± SD (n = 3, *p < 0.05, untreated vs. salubrinal).

Discussion

In the present study, we present evidence that the ALS-related mutant SOD1-induces UPR and ER stress, ultimately leading to the apoptotic cell death when over-expressed in N2a cells. We observed both early events of UPR, namely the activation of PERK, eIF2α, JNK and the nuclear translocation of ATF6 and the late event like XBP1 mRNA splicing. As a result of ER stress, an up-regulation of BiP/Grp78 and Bax and a down-regulation of Bcl-2 could be detected. Both activation of early UPR kinases and up-regulation of BiP/Grp78 indicate that cells are initially trying to unload the burden induced by the accumulation of the misfolded mutant SOD1. However, an up-regulation of proapoptotic Bax and down-regulation of anti-apoptotic Bcl-2 suggest that the UPR induced by the mutant SOD1s is not properly overcome, leading to the activation of a suicide program. This was evident as the activation of both caspase 12 and 7 and TUNEL-positive apoptotic nuclei were detected in the mutant SOD1-expressing cells. Indeed, the phosphorylation of eIF2α induced by the mutant SOD1 over-expression returned to control level at later period of over-expression, but salubrinal maintained the phosphorylation of eIF2α and protected the cells from the mutant SOD1-induced toxicity. This suggests that the initial UPR signaling caused by the mutant SOD1 expression was a protective one but turned into a death signal as the ER stress accumulated beyond the cellular capacity to manage the stress properly.

Interestingly, UPR induced by the mutant SOD1 that is a cytosolic protein shared the same signaling events as the UPR induced by proteins synthesized in the ER lumen. It remains to be studied how the accumulation of the misfolded proteins in the cytoplasm can induce the activation of ER-embedded kinases and signaling molecules. Alternatively, as Kikuchi et al. (2006) reported that the mutant SOD1 was detected in the ER fraction in the G93A mice, it is possible that the mutant SOD1 could accumulate in the ER in the N2a cells as well and induced the ER stress. Another possibility is that the accumulation of the misfolded proteins in the cytoplasm consumes cytoplasmic chaperones like heat shock protein family members and this somehow triggers the activation of UPR in the ER. The fact that most parts of the ER-embedded UPR kinases and signaling molecules are facing cytoplasmic side (Rutkowski and Kaufman 2004) may support the idea that the UPR signaling events are shared in proteins synthesized in ER lumen and cytoplasm. An occupation of proteasome by the misfolded mutant SOD1 interferes with the housekeeping ER-associated degradation and again this in return might induce UPR in the ER.

In many neurodegenerative diseases, it is a common pathological hallmark that the disease-related mutant proteins form insoluble aggregates (Ross and Poirier 2004). However, it is still under a debate whether the aggregate is a cause or consequence of the cellular damage induced by misfolded proteins. The idea that the aggregate is toxic is supported by the observations that many essential components like ubiquitin proteasome components or chaperone proteins are recruited to the aggregates, which compromises vital functions in the cytoplasm (Chaudhuri and Paul 2006).

However, there is accumulating evidence that the toxic species is the immature protofibrils composed of misfolded protein oligomers and the formation of the insoluble aggregate is rather a cellular defense mechanism to sequester these toxic oligomers. Indeed, visible aggregates of the mutant SOD1 is not detected prior to the onset of muscle weakness in a mouse model of ALS (Brown 1998). In polyglutamine disease where polyglutamine tract of the mutant protein forms insoluble aggregates, the correlation between aggregate formation and induction of apoptosis has been challenged by many studies (Lansbury and Lashuel 2006). Furthermore, using over-expression system on cultured neuronal cells, Lee et al. (2002) reported that there is no correlation between the mutant SOD1 aggregate formation and cell death. In support of this report, we observed in the present study that the rate of cell death is greater than that of visible aggregate formation (Fig. 1). In addition, the splicing of XBP1 and the activation of ERAI were observed in the mutant SOD1-expressing cells without the aggregates (Fig. 2d). These results imply that the misfolded mutant SOD1-induces ER stress before forming into insoluble aggregates and can induce cell death. One can argue against this conclusion as the ER stress inhibitor salubrinal seemed to delay the aggregate formation process and thus the cytoprotective effect of this drug may be due to the suppression of visible aggregate. However, it is still possible that the protective effect of salubrinal is due to the inhibition of the ER stress response caused by the mutant oligomers and the delay in the aggregate formation is the result of the successful management of the misfolded species by UPR. Taken together, the data in the present study supports the idea that the toxicity of the aggregate-prone mutant proteins in neurodegenerative diseases.

Until now there is no effective therapeutics for ALS. Results of drug studies in ALS mice indicate only a modest effect of anti-oxidant therapy on the clinical course of the disease. A number of trophic factors and anti-inflammatory agents have been reported to prolong survival in mouse models and some are now in clinical trials (McGeer and McGeer 2005). However, these therapies are targeted to the late events of the cellular damage and there is currently no intervention at earlier events. The data in the present study suggest that ER stress occurs before the formation of visible aggregates and the inhibition of this ER stress signaling protects cells against the misfolded protein-induced death. However, further investigation on the ER stress signaling is required to understand how the initial survival signal turns into death signal if one does not want to interfere with the survival signal while trying to suppress the death signal. Therefore, it will be of great interest to examine the effect of a pharmacological inhibition of ER stress in animal models. As signs of ER stress are observed in other neurodegenerative diseases (Lindholm et al. 2006), therapeutics targeting ER stress may be of great value in other forms of neurodegenerative diseases accompanying misfolded protein aggregates.

Acknowledgements

This work was supported by a grant from Korea Research Foundation (2003-042-C00102). We thank Dr D. Cleveland, Dr M. Miura, and Dr R. S.- Olea for the generous gifts.

Ancillary