Autophagic induction of amyotrophic lateral sclerosis-linked Cu/Zn superoxide dismutase 1 G93A mutant in NSC34 cells.

Previous studies have confirmed that the beclin 1 complex plays a key role in the initial stage of autophagy and deregulated autophagy might involve in amyotrophic lateral sclerosis. However, the mechanism underlying altered autophagy associated with the beclin 1 complex remains unclear. In this study, we transfected the Cu/Zn superoxide dismutase 1 G93A mutant protein into the motor neuron-like cell line NSC34 cultured in vitro. Western blotting and co-immunoprecipitation showed that the Cu/Zn superoxide dismutase 1 G93A mutant enhanced the turnover of autophagic marker microtubule-associated protein light chain 3II (LC3II) and stimulated the conversion of EGFP-LC3I to EGFP-LC3II, but had little influence on the binding capacity of the autophagy modulators ATG14L, rubicon, UVRAG, and hVps34 to beclin 1 during autophagosome formation. These results suggest that the amyotrophic lateral sclerosis-linked Cu/Zn superoxide dismutase 1 G93A mutant can upregulate autophagic activity in NSC34 cells, but that this does not markedly affect beclin 1 complex components.

Introduction

Most cases of amyotrophic lateral sclerosis are sporadic and about 10% of all cases are familial. Cu/Zn superoxide dismutase 1 was the first gene to be associated with amyotrophic lateral sclerosis and Cu/Zn superoxide dismutase 1 mutations account for about 20% of familial amyotrophic lateral sclerosis[12]. Several neurodegenerative diseases, including amyotrophic lateral sclerosis, are related to aggregation-prone proteins[3]. It is generally accepted that the aggregates formed by these proteins are relevant to toxicity and reducing the levels of these proteins may be beneficial.

There are two main degradation pathways acting on these aggregation-prone proteins: the ubiquitin-proteasome and autophagy pathways[4]. However, proteasomes frequently exhibit decreased activity in aggregate-related neurodegenerative diseases; moreover, protein aggregation can directly impair the ubiquitin-proteasome system in cell models[5]. When the ubiquitin-proteasome system is suppressed by aggregated proteins, autophagy can be activated in compensation[6]. The process of autophagy involves the formation of double-membrane structures called autophagosomes, which fuse with lysosomes to form autolysosomes, where engulfed cytoplasmic contents are degraded[7]. As the large protein aggregates cannot enter the narrow entrance of the proteasome barrel to be degraded, they may be routed to the autophagy pathway as a default[8]. Research on autophagy-related gene knockdown has demonstrated the inhibitory effect of autophagy on neurodegenerative diseases[910]. Previous data have also suggested that alteration of the autophagic process is involved in neurodegeneration, as affected neurons usually exhibit accumulation of autophagosomes or autolysosomes[11]. Mutant aggregation-prone huntingtin fragments and α-synuclein induced autophagy in cultured cell models and mouse models[1213]. These disease-causing aggregation-prone proteins could be cleared by autophagy[1415].

It has been demonstrated that autophagy and the proteasome pathway contribute equally to amyotrophic lateral sclerosis-linked mutant Cu/Zn superoxide dismutase 1 clearance in both COS-7 and N2a cells[16]. Sustained expression of a Cu/Zn superoxide dismutase 1 mutant resulted in decreased proteasomal activity[17]. In addition, emerging evidence suggests that altered autophagy may contribute to the pathogenesis of amyotrophic lateral sclerosis[18]. The spinal cord of an amyotrophic lateral sclerosis mouse model demonstrates an altered distribution of cathepsin B, L, and D, with concomitant elevated expression levels; these cathepsins are essential for proteolysis within autolysosomes[19]. Features of autophagy are often evident in the cytoplasm of degenerating motor neurons in amyotrophic lateral sclerosis patients and are closely associated with the characteristic aggregates[20]. These studies imply that autophagy might be upregulated in amyotrophic lateral sclerosis.

To the best of our knowledge, there has been no report that an amyotrophic lateral sclerosis-linked Cu/Zn superoxide dismutase 1 mutant can modulate autophagy in vitro. Here, we investigated the effect of mutant Cu/Zn superoxide dismutase 1 on autophagic activity in NSC34 cells, an immortalized motor neuron cell line. We validated the autophagic degradation of the Cu/Zn superoxide dismutase 1 G93A mutant in NSC34 cells and investigated whether there is increased autophagy in amyotrophic lateral sclerosis model mice.

Results

Quantitative analysis of experimental animals

Eighteen mice were assigned to one of three groups: 60 days (presymptomatic stage), 90 days (onset stage), and 125 days (end stage). Each group contained three Cu/Zn superoxide dismutase 1 G93A transgenic mice and three age-matched wild-type (WT) mice.

G93A mutant Cu/Zn superoxide dismutase 1 formed cytoplasmic aggregates

A prominent feature of Cu/Zn superoxide dismutase 1-linked amyotrophic lateral sclerosis is the appearance of Cu/Zn superoxide dismutase 1-rich cytoplasmic aggregates in the affected motor neurons[2122]. We compared the cellular expression of Cu/Zn superoxide dismutase 1 WT and G93A mutant. We transiently transfected NSC34 cells with enhanced green fluorescent protein (EGFP)-Cu/Zn superoxide dismutase 1 WT or G93A mutant constructs and took live cell images 48 hours later. As shown in Figure 1, EGFP-tagged Cu/Zn superoxide dismutase 1 G93A exhibited large bright fluorescence aggregates in the cytoplasm (right panel), whereas SOD WT presented uniform green fluorescence (left panel).

Figure 1

EGFP-SOD1 WT (A) and G93A mutant (B) in live NSC34 cells (fluorescence microscope, × 600).

EGFP-SOD1 G93A formed large aggregates in the cytoplasm. The arrows indicate an intracellular aggregate. SOD1: Cu/Zn superoxide dismutase 1; WT: wild type; G93A: glycine 93 changed to alanine. EGFP: enhanced green fluorescent protein.

G93A mutant Cu/Zn superoxide dismutase 1 was degraded by autophagy

It is postulated that the amyotrophic lateral sclerosis-linked Cu/Zn superoxide dismutase 1 mutant is cleared by autophagy because of its propensity to aggregate[23]. To investigate whether FLAG-tagged Cu/Zn superoxide dismutase 1 G93A was degraded by autophagy, we incubated NSC34 cells with or without 3-methyladenine (3-MA), a specific inhibitor of autophagy that functions at the sequestration stage by suppressing the activity of hVps34 (also called PI3KIII)[24]. Twenty-four hours after transfection, cell lysates were prepared and western blotted. As seen in Figure 2, there was obvious accumulation of mutant Cu/Zn superoxide dismutase 1 in the presence of 3-MA compared with the control group. Therefore, inhibition of the sequestration stage of autophagy blocked the degradation of Cu/Zn superoxide dismutase 1 G93A.

Figure 2

Inhibition of autophagosome formation by 3-MA caused the accumulation of SOD1 G93A (western blot analysis).

NSC34 cells were transfected with FLAG-tagged SOD1 G93A. Twenty-four hours after transfection, cells were incubated with or without 10 mmol/L 3-MA for 24 hours. Cell lysates were prepared and analyzed with the indicated antibodies. β-Actin was used as a loading control. SOD1: Cu/Zn superoxide dismutase 1; G93A: glycine 93 changed to alanine; 3-MA: 3-methyladenine, a specific inhibitor of autophagy at the sequestration stage.

Western blot analysis of microtubule-associated protein light chain 3II (LC3II) levels in the spinal cord of Cu/Zn superoxide dismutase 1 WT and G93A transgenic mice

To examine the extent of autophagy in Cu/Zn superoxide dismutase 1 G93A transgenic mice, a widely used animal model for amyotrophic lateral sclerosis[2526], we analyzed LC3II levels in the ventral lumbar spinal cord by western blotting. It is this location where motor neurons are primarily affected in amyotrophic lateral sclerosis. LC3 plays a critical role in autophagy. After pro-LC3 is translated, the C-terminal tail is cleaved to generate LC3I. LC3I becomes membrane-associated when it is converted to LC3II[27]. LC3II localizes in particular on the membrane of autophagosomes and can be used as a marker for autophagy[28]. At 60 days and 90 days (presymptomatic and disease onset stages, respectively), there was no difference in LC3II levels between Cu/Zn superoxide dismutase 1 G93A and WT control mice (data not shown and Figure 3A). However, at 125 days (end stage), LC3II levels were increased by 131 ± 10% in Cu/Zn superoxide dismutase 1 G93A mice compared with WT mice (Figure 3B, C; P < 0.05). This suggests that the autophagy status might alter at the end stage of the disease.

Figure 3

Altered autophagy status in SOD1 G93A transgenic mice.

(A, B) Western blot analysis of LC3II in the spinal cord of SOD1 G93A and age-matched SOD1 WT transgenic mice (90 and 125 days, respectively). (C) Quantitative analysis of LC3II levels in the spinal cord of SOD1 WT transgenic mice (125 days). The signal intensity in SOD1 G93A mice is shown as a percentage of the corresponding intensity in WT mice, which was set as 100% and expressed as the mean ± SEM (n = 3; two-sample t-test), aP < 0.05 vs. WT control. SOD1: Cu/Zn superoxide dismutase 1; WT: wild type; G93A: glycine 93 changed to alanine; LC3II: microtubule-associated protein light chain 3II.

Cu/Zn superoxide dismutase 1 G93A overexpression induced autophagy in NSC34 cells

Because we found increased LC3II in amyotrophic lateral sclerosis mice, we determined whether mutant Cu/Zn superoxide dismutase 1 could influence the autophagy status in vitro. Because LC3II itself is a substrate for autophagy, an increase in the amount of LC3II occurs in two conditions: upregulation of autophagosome formation or blockage of the autophagic process[2930]. Under the first condition, an increase in LC3II is regarded as autophagy induction. Under the latter condition, it is indicative of autophagy inhibition[29]. To distinguish these two conditions, we used bafilomycin A1, a vacuolar ATPase inhibitor that interferes with the autophagosome-lysosome fusion step[31], to assay lysosomal turnover of endogenous LC3II in Cu/Zn superoxide dismutase 1 WT or G93A-overexpressing cells.

As shown in Figure 4A, FLAG-tagged Cu/Zn superoxide dismutase 1 G93A mutant was present at a much lower level compared with Cu/Zn superoxide dismutase 1 WT, although the same amount of construct was transfected. This is consistent with previous studies that showed that the Cu/Zn superoxide dismutase 1 mutant has a high turnover rate because it is unstable[32]. The LC3II level in Cu/Zn superoxide dismutase 1 G93A-overexpressing cells was decreased compared with that in the SOD WT control in the absence of bafilomycin A1. However, after treatment with bafilomycin A1, the Cu/Zn superoxide dismutase 1 G93A cells showed a higher LC3II level than did the Cu/Zn superoxide dismutase 1 WT group (Figure 4A). Quantitative analysis showed that the protein level was increased by 133 ± 8% compared with the control (P < 0.05) (Figure 4B). Therefore, it was higher autophagic activity that led to the accelerated turnover of LC3II, reflected by a decrease in LC3II in Cu/Zn superoxide dismutase 1 G93A-overexpressing cells.

Figure 4

Overexpression of SOD1 G93A induced autophagy in NSC34 cells.

(A) NSC34 cells were transfected with FLAG-tagged SOD1 WT or G93A. Forty-four hours later, cells were treated with or without bafilomycin A1 (25 nmol/L, 4 hours). Cell lysates were prepared and analyzed with the indicated antibodies.

(B) Quantification of LC3II levels after bafilomycin A1 treatment. The intensity in SOD1 G93A-overexpressing cells is shown as a percentage of the corresponding intensity in WT-overexpressing cells, which was set as 100% (mean ± SEM, n = 3; aP < 0.05, vs. WT control (two-sample t-test)).

(C) NSC34 cells were co-transfected with FLAG-tagged SOD1 WT or G93A and EGFP-LC3. Forty-four hours after transfection, cells were treated with or without bafilomycin A1 (25 nmol/L, 4 hours). Cell lysates were prepared and analyzed with the indicated antibodies. SOD1: Cu/Zn superoxide dismutase 1; WT: wild type; G93A: glycine 93 changed to alanine.

In addition, we used an EGFP-LC3 plasmid as a reporter for autophagy; EGFP-LC3 can be lipidated to a phosphatidylethanolamine-conjugated form (EGFP-LC3II) which specifically serves as a marker for autophagic membranes[3334]. We co-transfected NSC34 cells with FLAG-tagged Cu/Zn superoxide dismutase 1 WT or G93A and EGFP-LC3, then collected and western blotted the cell lysates. As shown in Figure 4C, in the absence of bafilomycin A1, there was a slightly higher level of EGFP-LC3II in Cu/Zn superoxide dismutase 1 G93A-overexpressing cells compared with WT controls. In contrast, in the presence of bafilomycin A1, EGFP-LC3II could not be degraded and thus accumulated, leading to a further increased content of EGFP-LC3II in Cu/Zn superoxide dismutase 1 G93A-overexpressing cells relative to that in Cu/Zn superoxide dismutase 1 WT-overexpressing cells. This indicates that Cu/Zn superoxide dismutase 1 G93A can enhance the autophagic flux and stimulate the conversion of EGFP-LC3I to EGFP-LC3II. Taken together, our results confirm that Cu/Zn superoxide dismutase 1 G93A increases autophagic activity in NSC34 cells.

Effect of Cu/Zn superoxide dismutase 1 G93A on the association of UVRAG, ATG14L, rubicon, and hVps34 with beclin 1

To investigate the mechanism of Cu/Zn superoxide dismutase 1 G93A mutant-induced autophagy in NSC34 cells, we focused on the initial step of autophagosome formation and investigated the influence of Cu/Zn superoxide dismutase 1 G93A on the binding capacity of ATG14L, rubicon, UVRAG, and hVps34 to beclin 1, which functions as a key regulatory complex of autophagy. Beclin 1 was the first identified autophagy-related protein in mammalian cells[35]. It induces autophagy by forming a complex with hVps34 (also called PI3KIII) and stimulating its kinase activity. At least three hVps34-beclin 1 complexes are involved autophagosome formation (Figure 5A)[36]. ATG14L coordinates with beclin 1 to increase the formation of double-membrane structures[37]. UVRAG competes with ATG14L for binding to the beclin 1-hVps34 complex and promotes autophagosome fusion with lysosomes[38]. Rubicon forms a complex with UVRAG-beclin 1-hVps34 and inhibits autophagy[37]. We co-expressed FLAG-Cu/Zn superoxide dismutase 1 WT or G93A with EGFP-vector or EGFP-beclin 1 in NSC34 cells, then 48 hours later used an anti-GFP antibody to co-immunoprecipitate the EGFP-tagged protein with its interacting proteins. As shown in Figure 5B, western blot validated the expression of transfected constructs and the indicated beclin 1-interacting proteins (left panel). EGFP and EGFP-beclin 1 were effectively immunoprecipitated by the anti-GFP antibody. However, there was no significant difference in binding ability of the autophagy modulators to beclin 1 between Cu/Zn superoxide dismutase 1 WT- and G93A-overexpressing cells (Figure 5B; right panel). These results demonstrate that Cu/Zn superoxide dismutase 1 G93A has little effect on the binding capability of the indicated proteins to beclin 1.

Figure 5

The effect of SOD1 G93A on the binding capacity of hVps34, ATG14L, UVRAG, and rubicon to beclin 1.

(A) Three hVps34-beclin 1 complexes that were involved in autophagosome formation[36]. The ATG14L and UVRAG complexes were required for autophagy, whereas the rubicon complex negatively regulated autophagy.

(B) NSC34 cells were co-transfected with FLAG-SOD1 WT or G93A and EGFP-vector or EGFP-beclin 1. Forty-eight hours after transfection, anti-GFP antibody was used to co-immunoprecipitate the EGFP-beclin 1 and its interacting proteins. Cell lysate (left panel) and immunoprecipitated products (right panel) were blotted with the indicated antibodies.

Discussion

The appearance of protein aggregates is a hallmark of amyotrophic lateral sclerosis[2122]. Indeed, the Cu/Zn superoxide dismutase 1 G93A mutant forms large cytoplasmic aggregates in NSC34 cells. Increased autophagic structures have been found in neurons affected by neurodegenerative disease caused by aggregation-prone proteins[3940].

To investigate alterations in autophagy in amyotrophic lateral sclerosis, we measured the status of autophagy in the lumbar spinal cord of Cu/Zn superoxide dismutase 1 G93A transgenic mice using LC3II, whose levels are closely associated with the extent of autophagosome formation[28]. We used western blotting, as this is the most simple way to monitor autophagy[2741].

Our data indicated enhanced autophagic activity in Cu/Zn superoxide dismutase 1 G93A mice. However, it is still controversial whether increased autophagosome formation results from upregulated autophagic activity or impaired autophagic degradation[42], as an increase in LC3II level could be interpreted as a blockage of the autophagic process[30].

A previous study has reported defective retrograde axonal transport in amyotrophic lateral sclerosis mice, which suggests the possibility of inhibition of autophagosome transportation[43]. Moreover, we cannot conclude that the observed autophagosome accumulation is directly due to the mutant Cu/Zn superoxide dismutase 1 proteins. When the disease progresses to the pathology of the end stage, autophagy could be induced by nutritional stress and/or cell injury[40].

When we measured autophagic flux using bafilomycin A1, our data clearly demonstrated that Cu/Zn superoxide dismutase 1 G93A overexpression resulted in increased autophagic activity in NSC34 cells. There is mounting evidence for the activation of autophagy by aggregation- prone proteins. Overexpression of α-synuclein induced autophagy in cellular model and α-synuclein was degraded by autophagy[15].

Similarly, treatment with β-amyloid peptide activated a strong autophagic response and autophagy protected against β-amyloid-induced neuronal death[34]. Previous studies in cells, transgenic Drosophila, and mouse models have shown that mutant huntingtin fragments induce autophagy, whereas autophagy reduces the accumulation of mutant huntingtin and attenuates its toxicity[1244].

Here, we validated the autophagic degradation of Cu/Zn superoxide dismutase 1 G93A in NSC34 cells. Recent studies have demonstrated that autophagic clearance of mutant Cu/Zn superoxide dismutase 1 protects against amyotrophic lateral sclerosis in a mouse model[4546]. Taken together, we hypothesize that autophagy is upregulated in response to stimulation by aggregation-prone proteins including the amyotrophic lateral sclerosis-linked Cu/Zn superoxide dismutase 1 mutant. In turn, increased autophagy may contribute to degradation of the aggregated proteins as a defense against their toxicity, and may protect against cell death under certain conditions (Figure 6).

Figure 6

A model of the relationship between aggregation-prone proteins and autophagic activity.

The aggregation-prone proteins, including the amyotrophic lateral sclerosis-linked superoxide dismutase 1 mutant, can induce autophagic activity, although the mechanism is not fully understood. Meanwhile, increased autophagy may contribute to degradation of these proteins as a defense against their toxicity.

However, it is noteworthy that prolonged abnormal activation of autophagy may disturb the balance of cell homeostasis, shifting it from the clearance of unnecessary cellular constituents to triggering apoptosis or necrosis[474849]. For example, treatment of amyotrophic lateral sclerosis model mice with rapamycin, an inducer of autophagy, accelerated neuronal degeneration and disease progression accompanied by enhanced apoptosis, suggesting that augmenting autophagy could also be detrimental[50]. Thus, it will be of interest to determine whether enhanced autophagy induced by aggregation-prone proteins ultimately promotes cell death.

Autophagy is a highly regulated multi-step process[47]. Central to this process is the formation of the autophagosome[36], in which beclin 1 plays an essential role[51]. Beclin 1 forms a platform through binding to the mammalian homolog of Vps34[52]. The hVps34-beclin 1 interaction is modulated by recruiting activators or repressors of autophagy, consequently altering autophagic activity[3738].

We investigated the mechanism of Cu/Zn superoxide dismutase 1-induced autophagy in vitro, by examining the effect of Cu/Zn superoxide dismutase 1 G93A on the binding capacity of several regulators (UVRAG, ATG14L, rubicon, and hVps34) to beclin 1. However, we did not observe any significant effects.

Therefore, Cu/Zn superoxide dismutase 1 G93A-induced autophagy in NSC34 cells may rely on autophagic machineries other than the beclin 1 complex. Apart from the beclin 1 complex, a subset of regulatory components is essential for autophagosome formation[36]. Moreover, several signaling regulation pathways are implicated in autophagy[3653]. The mammalian target of rapamycin (mTOR) kinase pathway has been reported to be involved in β-amyloid-activated autophagy[54]. Mutant huntingtin aggregates sequester mTOR and induce autophagy by impairing its kinase activity[44].

In addition, there are several steps that may cause the induction of autophagy, such as enhanced fusion of autophagosomes with lysosomes or the overexpression of hydrolase in lysosomes[19]. Further work is required to address the roles of these regulatory machineries in Cu/Zn superoxide dismutase 1 mutant-induced autophagy.

In conclusion, we have provided a link between alterations in autophagic activity and the amyotrophic lateral sclerosis-linked Cu/Zn superoxide dismutase 1 G93A mutant. We observed the possibility of increased autophagy in amyotrophic lateral sclerosis mice. Importantly, our results confirm that mutant Cu/Zn superoxide dismutase 1 protein can induce autophagy in NSC34 cells.

Materials and Methods

Design

Controlled, biochemical, observational research.

Time and setting

The experiments were performed at the University of Kentucky, KY, USA from October 2008 to April 2011.

Materials

Cu/Zn superoxide dismutase 1 G93A transgenic mice were sacrificed at age 60, 90, or 125 days (n = 3 for each group). Mice were anesthetized with an intraperitoneal injection of 0.1 mL pentobarbital (50 mg/mL) and transcardially perfused with 0.1 mol/L phosphate-buffered saline, pH 7.5, before the spinal cord was dissected. Age-matched WT Cu/Zn superoxide dismutase 1 transgenic mice were used as controls.

The experimental procedures were approved by the University Institutional Animal Care and Use Committee.

Methods

Plasmid construction

The construction of EGFP- and FLAG-tagged Cu/Zn superoxide dismutase 1 WT or G93A mutant plasmids was reported previously[55].

Amplified Cu/Zn superoxide dismutase 1 WT or G93A fragments were cloned into EcoRI and BamHI-digested p3×FLAG-CMV10 (Sigma, St. Louis, MO, USA) and the pEGFP-N3 expression vector (Clontech, Mountain View, CA, USA). EGFP-tagged beclin 1 plasmid was kindly provided by Dr. Qinjun Wang (University of Kentucky). A beclin 1-encoding cDNA was inserted into the EcoRI and BamHI sites of the pEGFP-N3 expression vector (Clontech). LC3 was amplified by PCR and was cloned into EcoRI-BamHI double-digested pEGFP-C1 (Clontech)[56].

Cell transfection and drug treatment

NSC34 cells were grown in Dulbecco's modified Eagle's medium (Invitrogen, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (Invitrogen) and penicillin/streptomycin at 37 °C in a humidified, 5% CO2 incubator. For transient transfection, cells were plated on a 6-well plate or 6-cm Petri dish to appropriately 80% confluence and transfected with plasmids using Lipofectamine reagent (Invitrogen) according to the manufacturer's protocol. For co-transfecting FLAG-Cu/Zn superoxide dismutase 1 WT or G93A and EGFP-vector or EGFP-beclin 1, an equal amount of each plasmid was used.

Cells were treated with 3-MA (10 mmol/L; Calbiochem, Billerica, MA, USA) or bafilomycin A1 (25 nmol/L; Calbiochem) as described.

Fluorescence microscopy

Forty-eight hours after transfection, live NSC34 cells expressing EGFP-Cu/Zn superoxide dismutase 1 WT or G93A were visualized and photographed by fluorescence microscopy (Nikon Eclipse E600, Chiyoda-ku, Tokyo, Japan) with identical camera settings.

Preparation of cell lysates and tissue homogenate

Cells were harvested and lysed in 1 × radioimmune precipitation assay buffer (1% NP-40, 50 mmol/L Tris-HCl/pH 7.4, 0.25% sodium deoxycholate, 150 mmol/L NaCl, 1 mmol/L ethylenediaminetetraacetic acid (EDTA) containing protease inhibitor mixture (P8340; Sigma; 1:1,000 dilution), and 0.625 mg/mL N-ethylmaleimide (E3876; Sigma). After incubation on ice for 20 minutes, the lysates were cleared by centrifugation at 1,000 × g for 10 minutes at 4°C and the protein concentrations were determined using a BCA assay (Thermo Scientific, Waltham, MA, USA).

Dissected mouse spinal cords were lysed in radioimmune precipitation assay buffer supplemented with protease inhibitor mixture (P8340; Sigma; 1:1,000 dilution), 0.2 mmol/L phenylmethylsulfonyl fluoride, and 0.625 mg/mL N-ethylmaleimide. Lysates were cleared by centrifugation at 1,000 × g for 10 minutes at 4°C, and the protein concentration was determined by BCA assay.

Proteins were denatured by boiling in 6 × SDS sample buffer for 5 minutes at 95°C to prepare for SDS-polyacrylamide gel electrophoresis (PAGE).

Co-immunoprecipitation

For co-immunoprecipitation experiments[37], NSC34 cells were transfected simultaneously with FLAG-SOD WT or G93A and EGFP-vector or EGFP-beclin 1 in equal amounts. Forty-eight hours later, cells were lysed in immunoprecipitation lysis buffer (20 mmol/L HEPES/pH 7.4, 1 mmol/L MgCl2, 0.25 mmol/L CaCl2, 0.2% triton X-100, 150 mmol/L NaCl, EDTA-free protease inhibitor cocktail, 200 µg/mL phenylmethylsulfonyl fluoride and DNase I). Dynabeads M-270 E-proxy (Invitrogen) were conjugated with anti-GFP antibody and incubated with cell lysate at 4°C for 2 hours. After the beads were washed five times in immunoprecipitation lysis buffer, proteins were eluted by incubating the beads in elution buffer (0.5 mmol/L EDTA, pH 8, and 0.5 mol/L NH3·H2O) at room temperature for 20 minutes and dried in a vacuum speed centrifuge.

Proteins were denatured by boiling in 2 × Laemmli loading buffer for SDS-PAGE.

Western blot analysis

Cell lysates, tissue homogenates, and co-immunoprecipitated proteins were subjected to SDS-PAGE and subsequent western blot analysis.

The primary antibodies used in western blotting were anti-FLAG monoclonal antibody (F1804; Sigma), rabbit anti-LC3 polyclonal antibody (L8929; Sigma), rabbit anti-Cu/Zn superoxide dismutase 1 polyclonal antibody (sc-11407; Santa Cruz Biotechnology, Santa Cruz, CA, USA), rabbit anti-GFP polyclonal antibody (sc-8334; Santa Cruz Biotechnology), rabbit anti-UVRAG polyclonal antibody (AP1850e; Abgent, San Diego, CA, USA), goat anti-β-actin polyclonal antibody (sc-1616; Santa Cruz Biotechnology). Rabbit anti-ATG14L, anti-rubicon, and anti-hVps34 polyclonal antibodies were generously provided by Dr. Qingjun Wang (University of Kentucky)[37]. After overnight incubation with primary antibodies at 4°C, each blot was probed with the corresponding anti-rabbit, anti-mouse, or anti-goat secondary antibody coupled to horseradish peroxidase (Santa Cruz Biotechnology).

Protein bands were visualized using enhanced chemiluminescence (Pierce, Rockford, IL, USA). The signal intensity was quantified using Image J (National Institutes of Health, Bethesda, MD, USA).

Statistical analysis

All data were statistically analyzed using Excel 2007 software (Microsoft Corporation, Redmond, WA, USA) and expressed as mean ± SEM. The two-sample t-test was used for difference comparison between two groups. P < 0.05 was considered statistically significant.