OUP user menu

Protein disulphide isomerase protects against protein aggregation and is S-nitrosylated in amyotrophic lateral sclerosis

Adam K. Walker, Manal A. Farg, Chris R. Bye, Catriona A. McLean, Malcolm K. Horne, Julie D. Atkin
DOI: http://dx.doi.org/10.1093/brain/awp267 105-116 First published online: 10 November 2009

Summary

Amyotrophic lateral sclerosis is a rapidly progressing fatal neurodegenerative disease characterized by the presence of protein inclusions within affected motor neurons. Endoplasmic reticulum stress leading to apoptosis was recently recognized to be an important process in the pathogenesis of sporadic human amyotrophic lateral sclerosis as well as in transgenic models of mutant superoxide dismutase 1-linked familial amyotrophic lateral sclerosis. Endoplasmic reticulum stress occurs early in disease, indicating a critical role in pathogenesis, and involves upregulation of an important endoplasmic reticulum chaperone, protein disulphide isomerase. We aimed to investigate the involvement of protein disulphide isomerase in endoplasmic reticulum stress induction, protein aggregation, inclusion formation and toxicity in amyotrophic lateral sclerosis. Motor neuron-like NSC-34 cell lines were transfected with superoxide dismutase 1 and protein disulphide isomerase encoding vectors and small interfering RNA, and examined by immunocytochemistry and immunoblotting. Expression of mutant superoxide dismutase 1 induced endoplasmic reticulum stress, predominantly in cells bearing mutant superoxide dismutase 1 inclusions but also in a proportion of cells expressing mutant superoxide dismutase 1 without visible inclusions. Over-expression of protein disulphide isomerase decreased mutant superoxide dismutase 1 aggregation, inclusion formation, endoplasmic reticulum stress induction and toxicity, whereas small interfering RNA targeting protein disulphide isomerase increased mutant superoxide dismutase 1 inclusion formation, indicating a protective role for protein disulphide isomerase against superoxide dismutase 1 misfolding. Aberrant modification of protein disulphide isomerase by S-nitrosylation of active site cysteine residues has previously been shown as an important process in neurodegeneration in Parkinson's and Alzheimer's disease brain tissue, but has not been described in amyotrophic lateral sclerosis. Using a biotin switch assay, we detected increased levels of S-nitrosylated protein disulphide isomerase in transgenic mutant superoxide dismutase 1 mouse and human sporadic amyotrophic lateral sclerosis spinal cord tissues. Hence, despite upregulation, protein disulphide isomerase is also functionally inactivated in amyotrophic lateral sclerosis, which may prevent its normal protective function and contribute to disease. We also found that a small molecule mimic of the protein disulphide isomerase active site, (±)-trans-1,2-bis(mercaptoacetamido)cyclohexane, protected against mutant superoxide dismutase 1 inclusion formation. These studies reveal that endoplasmic reticulum stress is important in the formation of mutant superoxide dismutase 1 inclusions, and protein disulphide isomerase has an important function in ameliorating mutant superoxide dismutase 1 aggregation and toxicity. Functional inhibition of protein disulphide isomerase by S-nitrosylation may contribute to pathophysiology in both mutant superoxide dismutase 1-linked disease and sporadic amyotrophic lateral sclerosis. Protein disulphide isomerase is therefore a novel potential therapeutic target in amyotrophic lateral sclerosis and (±)-trans-1,2-bis(mercaptoacetamido)cyclohexane and other molecular mimics of protein disulphide isomerase could be of benefit in amyotrophic lateral sclerosis and other neurodegenerative diseases related to protein misfolding.

  • amyotrophic lateral sclerosis
  • superoxide dismutase 1
  • endoplasmic reticulum stress
  • protein disulphide isomerase
  • S-nitrosylation

Introduction

Amyotrophic lateral sclerosis (ALS) is a fatal, rapidly progressing neurodegenerative disorder primarily affecting motor neurons (Rothstein, 2009). The disease is characterized by muscle weakness leading to paralysis and death within 3–5 years of diagnosis, with no effective treatment. Similar to other neurodegenerative diseases, protein misfolding and abnormal intracellular protein inclusions are pathological hallmarks of ALS (Bruijn et al., 2004). Mutations in the gene encoding superoxide dismutase 1 (SOD1) account for a small proportion of cases of ALS (Rosen, 1993; Selverstone Valentine et al., 2005), but produce a disease that is clinically and pathologically similar to sporadic ALS. Also, transgenic rodents over-expressing mutant SOD1 are the most widely accepted animal models of ALS currently available, and they develop a disease similar to that seen in humans (Gurney et al., 1994; Turner and Talbot, 2008). Therefore, it is likely that cellular processes disrupted by mutant SOD1 are also impaired in other forms of ALS.

SOD1 contains four cysteine residues and the physiological SOD1 homodimer is stabilized by intrasubunit disulphide bonds; however ALS-linked mutant SOD1 proteins are prone to form disulphide-reduced monomers and high molecular weight oligomers containing non-native disulphide bonds (Furukawa et al., 2006; Wang et al., 2006). The proportion of disulphide-reduced SOD1 is increased in the brain and spinal cord of transgenic mutant SOD1 mice (Tiwari and Hayward, 2003; Jonsson et al., 2006), and mutant SOD1 aggregates containing aberrant, non-native disulphide bonds have also been detected in mutant SOD1 animal and cellular models (Furukawa et al., 2006; Niwa et al., 2007; Cozzolino et al., 2008). Recently it was found that even small amounts of disulphide reduced monomeric mutant SOD1 can trigger aggregation of other highly stable forms of SOD1 (Chattopadhyay et al., 2008). Aggregation and formation of mutant SOD1 inclusions is linked with neuronal toxicity (Turner et al., 2005; Soo et al., 2009) and the correlation between mutant SOD1 protein aggregation, neuronal death and disease progression suggests an association with pathogenesis (Johnston et al., 2000; Matsumoto et al., 2005; Kabuta et al., 2006).

Recent reports indicate that endoplasmic reticulum stress (ER) is an important pathway leading to cell death in ALS (Atkin et al., 2006, 2008; Saxena et al., 2009). The ER is an important organelle for post-translational modification and sorting of many proteins. An increased burden of unfolded proteins in the ER leads to ER stress, which triggers a homeostatic mechanism known as the unfolded protein response (Schroder, 2008). The unfolded protein response consists of three distinct signalling pathways, mediated by the chaperone immunoglobulin binding protein (BiP) (Romisch, 2005). In unstressed cells, BiP interacts with three unfolded protein response sensor proteins within the ER lumen: activating transcription factor 6, protein-kinase-like endoplasmic reticulum kinase (PERK), and inositol requiring kinase 1 (Schroder, 2008). During ER stress BiP is titrated away from these sensor proteins, leading to their activation. This suppresses general protein translation and induces expression of ER chaperones, such as protein disulphide isomerase (PDI) (Schroder, 2008), which is an important enzyme for the formation of native structures by isomerization and modification of protein disulphide bonds. Although usually protective, when the unfolded protein response is prolonged by disease states, such as in ALS, reactive oxygen species accumulate and apoptosis is triggered by the ER stress-specific transcription factor C/EBP homologous protein (CHOP) (Haynes et al., 2004; Schroder, 2008).

The unfolded protein response is induced in spinal cords of mutant SOD1 mice, rats and cell culture (Atkin et al., 2006; Kikuchi et al., 2006; Saxena et al., 2009) and also in patients with sporadic ALS (Atkin et al., 2008; Ilieva et al., 2007). ER stress and unfolded protein response induction are therefore features of all forms of ALS, not just mutant SOD1-associated disease. ER stress occurs prior to clinical signs in SOD1G93A rodents (Atkin et al., 2008; Saxena et al., 2009) implying a contribution to pathophysiology. PDI is increased in spinal cords of mutant SOD1 rodents and sporadic ALS patients, where it is also a constituent of cytoplasmic inclusions, providing evidence for a role in pathology (Atkin et al., 2006, 2008; Ilieva et al., 2007). ER stress is also a feature of other neurodegenerative diseases (Scheper and Hoozemans, 2009), and aberrant inactivation of PDI by S-nitrosylation of crucial active site cysteine residues allows misfolded proteins to accumulate in the brains of Parkinson's and Alzheimer's disease patients (Uehara et al., 2006).

Given the association between PDI, inclusion formation and ER stress, and the role of disulphide bonds in mutant SOD1 aggregation, we predicted that PDI could inhibit mutant SOD1 inclusion formation and prevent ER stress. We also predicted that S-nitrosylation of PDI in spinal cord could be involved in the pathology of mutant SOD1-expressing transgenic mice and non-mutant SOD1-linked ALS in humans. In the current study, we demonstrate that PDI expression modulates mutant SOD1 protein misfolding and toxicity, and identify PDI as a novel therapeutic target for prevention of protein misfolding, ER stress and toxicity in ALS, with implications for other neurodegenerative disorders.

Methods

Mouse tissue samples

Transgenic mice carrying the human SOD1G93A mutation were obtained from the TgN (SOD1-G93A)1Gur line (Jackson Laboratory). Controls were age-matched non-transgenic littermates. All methods conformed to the Australian National Health and Medical Research Council published code of practice for the use of animals in research, and were approved by the Howard Florey Institute animal ethics committee.

Human tissue samples

Human lumbar spinal cord segments (L3–L5) from 21 patients who died of respiratory failure caused by ALS were provided by the MND Research Tissue Bank of Victoria (Table 1). The clinical diagnosis of ALS was confirmed at post-mortem. Control samples were obtained from five individuals without evidence of neurological or psychiatric disease. Studies were approved by the Howard Florey Institute human ethics committee.

View this table:
Table 1

Patient information

Control patients
CaseSexAge (years)PMI (h)Cause of death
    1M5233Aortic rupture
    2M7550Acute myocardial infarction
    3M6468Ischaemic heart disease
    4M8568Ischaemic heart disease
    5M8671Multiple myeloma
ALS patients
CaseSexAge (years)PMI (h)Disease duration (months)Family history
    1F501132Mother: ALS
    2M531836No
    3M532984No
    4M7931126No
    5F551429No
    6M42842No
    7M792584NA
    8M593940No
    9F792329No
    10F552572Father: dementia
    11F564512No
    12M582514Father, brother: ALS
    13F609019No
    14M727104No
    15M424228No
    16F637217No
    17M726746No
    18M382187Father, brother: HD, maternal grandmother: a wasting muscular disease
    19M641424No
    20F5920168No
    21M647760Brother: possible dementia
  • M = male, F = female, Age = age at death, PMI = post-mortem interval, NA = information not available, HD = Huntington's disease.

Constructs

SOD1-EGFP vectors were as described previously (Turner et al., 2005). For the PDI-DsRed2 fusion construct, a SalI restriction enzyme site was introduced and the translation stop codon removed from a pCMV5 vector containing the full-length human PDI precursor protein cDNA (provided by Professor Neil Bulleid, University of Manchester) using QuikChange site-directed mutagenesis (Stratagene) with the primers 5′-GCTGTGAAAGATGAACTGTCGACCGCAAAGCCAGACCCGG-3′ (forward) and its reverse complement, according to the manufacturer's protocol. The resultant construct was digested and the product subcloned in-frame into the pDsRed2.N1 vector (Clontech), and the DNA sequence confirmed.

Cell culture and transfection

Mouse NSC-34 cells (provided by Professor Neil Cashman, University of Toronto) and mouse neuroblastoma Neuro2a cells were maintained in Dulbecco modified Eagle's medium with 10% fetal calf serum, 100 µg/ml penicillin and 100 µg/ml streptomycin. Constitutive over-expressing NSC-34 cell lines were constructed by transfection of pDsRed2.N1 or pPDI-DsRed2.N1 and selection with G418 (Promega). Single cell clones were obtained by serial dilution. Transfections of NSC-34 cells were performed using TransFast reagent (Promega) according to the manufacturer's protocol. Cells were examined 72 h post-transfection by fluorescence microscopy and cells with prominent EGFP-positive inclusions were counted as a percentage of total EGFP-positive cells. Treatments with tunicamycin (Sigma) added to culture medium from a stock diluted at 200 μg/ml in dimethyl sulfoxide, and (±)-trans-1,2-bis(mercaptoacetamido)cyclohexane (BMC) (Toronto Research Chemicals) added to culture medium from a stock dissolved at 20 mM in methanol, were started 1 h following transfection as indicated, with respective vehicle treated controls.

Small interfering RNA

ON-TARGETplus SMARTpool small interfering RNA (siRNA) duplexes (Dharmacon) targeting mouse PDI (P4hb) or non-targeting control were co-transfected with SOD1-EGFP constructs into Neuro2a cells using DharmaFECT Duo transfection reagent (Dharmacon) according to the manufacturer's protocol. Cells were plated at 1 × 104 cells per well in 96 well plates 24 h prior to transfection with 100 nM siRNA and 100 ng plasmid per well. Protein levels, mRNA levels and SOD1 inclusion formation were analysed at 48 h post-transfection.

Real-time polymerase chain reaction

Cell pellets from siRNA and plasmid co-transfected Neuro2a cells were collected by trypsinization, and RNA isolated using the PicoPure siRNA Extraction Kit (Arcturus) and reverse transcribed using the SuperScript III qRT-PCR Kit (Invitrogen). Primers against the mouse PDI gene P4hb (forward 5′-CGTCCAACAGTGGTGTGTTC-3′ and reverse 5′-AAAGGCAGCTGATTGTGCTT-3′) and the housekeeping gene Hprt1 (forward 5′-CTTTGCTGACCTGCTGGATT-3′ and reverse 5′-TATGTCCCCCGTTGACTGAT-3′) were designed using Primer3 (Rozen and Skaletsky, 2000). Real-time PCR was carried out using the SYBR GreenER qPCR SuperMix Universal (Invitrogen) on a Rotor-Gene 6000 (Corbett Life Science) and analysed using the ΔΔCT method (Pfaffl, 2001).

Nuclear morphology apoptosis analysis

Neuro2a cells grown on coverslips were co-transfected with equal amounts of each plasmid as indicated, using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Cells were washed with phosphate buffered saline (PBS) 72 h post-transfection and fixed with 4% paraformaldehyde in PBS for 10 min. Cells were treated with Hoechst 33342 (1:10 000) prior to mounting in fluorescent mounting medium (Dako). Images were acquired using an Olympus Fluoview 1000 inverted confocal laser-scanning microscope. Apoptotic nuclei were defined as condensed (under ∼5 μm in diameter) or fragmented (multiple condensed Hoechst-positive structures in one cell), and counted as a percentage of non-apoptotic cells, from at least 100 cells expressing both transfected constructs per treatment in three independent experiments. Cells expressing only one transfected plasmid and those displaying abnormal morphology or undergoing cell division were excluded from analysis.

Immunofluorescence studies

NSC-34 cells grown on coverslips were washed with PBS and fixed with 4% paraformaldehyde in PBS for 10 min. Cells were permeabilized in 0.1% triton X-100 in PBS for 2 min, blocked for 30 min with 1% bovine serum albumin in PBS, and incubated with primary anti-CHOP antibody (1:50, Santa Cruz) for 16 h at 4°C. Secondary AlexaFluor-594 conjugated anti-mouse antibody (1:2000, Molecular Probes) was incubated for 1 h at room temperature, and cells were treated with Hoechst 33342 and mounted as above. Images were acquired using constant gain and offset settings for CHOP immunoreactivity using an Olympus Fluoview 1000 inverted confocal laser-scanning microscope.

Immunoblotting

Cell lysates were collected in TN buffer (50 mM Tris–HCl pH 7.5 and 150 mM NaCl) with 0.1% sodium dodecyl sulfate (SDS), 1% protease inhibitor cocktail (Sigma) and 1% phosphatase inhibitor (Sigma) by incubation on ice for 10 min. The supernatant was cleared by centrifugation at 16 100 g for 10 min (the SDS-soluble fraction). Pellets were sonicated for 10 s on ice in 250 µl TN buffer with 1% SDS and 1% protease inhibitor, then centrifuged at 16 100 g for 15 min. The resulting pellet (the SDS–insoluble fraction) was solubilized in 2× loading buffer with 8M urea and 10% β-mercaptoethanol. Protein concentrations were determined using the bicinchoninic acid assay method. Protein samples (10 or 20 µg) were electrophorezed through 9% or 12% SDS–polyacrylamide gels and transferred to nitrocellulose membranes. Membranes were blocked with 5% skim-milk in Tris-buffered saline pH 8.0 for 30 min then incubated with primary antibodies against the following proteins at 4°C for 16 h: SOD1 (1:4000, Calbiochem), PDI (1:2000, Stressgen or 1:1000, Abcam), β-actin (1:2000, Sigma-Aldrich), BiP (1:4000, Stressgen), phospho-PERK (1:500, Santa Cruz), CHOP/GADD153 (1:400, Santa Cruz), ubiquitin (1:1000, Dako), GAPDH (1:1000, Santa Cruz). Membranes were incubated for 1 h at room temperature with secondary antibodies (1:4000, HRP-conjugated donkey anti-sheep, goat anti-rabbit, or goat anti-mouse, Chemicon), and were detected using ECL reagent (Roche). Blots were stripped using ReBlot Plus solution (Chemicon) and re-probed as above. Quantitation of blots was performed by densitometry using ImageJ (NIH).

S-nitrosylation biotin switch assay

Detection of protein S-nitrosylation was performed in decreased light as described previously (Jaffrey et al., 2001; Uehara et al., 2006), using fresh frozen human lumbar spinal cord and mouse spinal cord tissues. Briefly, ∼50 mg tissue were homogenized in HEN buffer [250 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) pH 7.5, 1 mM ethylenediaminetetraacetic acid (EDTA) and 0.1 mM neocuproine] and centrifuged at 2000 g for 10 min at 4°C. Protein concentrations were determined using the bicinchoninic acid method and equal amounts of protein were adjusted to a concentration of 0.8 μg/μl with HEN buffer and final 2.5% SDS. Free thiol groups were blocked by addition of 0.1% S-methyl methanethiosulfonate (Sigma) from 10% stock in dimethylformamide and incubation at 50°C for 20 min with frequent vortexing. Proteins were acetone precipitated and the pellet washed four times with −20°C 70% acetone. Pellets were dried at room temperature and re-suspended in HEN buffer with 1% SDS. Nitrosothiol groups were reduced by addition of 100 mM sodium ascorbate and labelled with 0.25 mg/ml biotin-HPDP (Pierce). Ascorbate and biotin-HPDP were omitted from some control reactions, as indicated. Samples were incubated with agitation for 1 h at room temperature, followed by acetone precipitation as above. Proteins were re-suspended in 250 μl 10% HEN buffer in H2O with 1% SDS and 750 μl neutralization buffer (25 mM HEPES, 100 mM NaCl, 1 mM EDTA and 1% triton X-100). Streptavidin–agarose slurry (40 μl, Sigma) was added and the samples vertically rotated at 4°C for 12 h. Beads were collected and washed six times with neutralization buffer (the first four washes with 600 mM NaCl). Proteins were released from dried beads with SDS sample buffer containing 5% β-mercaptoethanol and immunoblotted as above.

Statistics

Comparisons were performed as indicated by ANOVA followed by Bonferroni's post-test from three independent experiments, unless otherwise stated. P < 0.05 was considered statistically significant. Data are presented as mean ± standard error of the mean (SEM).

Results

Mutant SOD1 induces ER stress in motor neuron-like cells

We previously reported increased expression of the full range of ER stress proteins, including PDI and CHOP, in spinal cords of SOD1G93A rodents at presymptomatic (p60), symptomatic (p90) and end-stage (p120) disease (Atkin et al., 2006, 2008). To investigate whether mutant SOD1 inclusion formation is directly linked to ER stress, motor neuron-like NSC-34 cells were transiently transfected with vectors encoding either SOD1wt or mutant SOD1A4V tagged with EGFP. Mutant SOD1A4V produces the most aggressive form of SOD1-linked disease in human ALS patients and demonstrates a high propensity to aggregate in neuronal cell lines (Turner et al., 2005; Turner and Talbot, 2008). Seventy-two hours after transfection, individual cells were examined for nuclear CHOP immunoreactivity, indicating ER stress-induced apoptotic signalling. Cells were classified as either bystander (non-transfected), diffuse (expressing SOD1-EGFP without visible inclusions) or inclusion positive (expressing SOD1-EGFP with visible inclusions), where ‘inclusions’ were defined as structures visible by light microscopy (Fig. 1). In contrast, we define protein ‘aggregates’ as insoluble protein detected biochemically by immunoblotting. Nuclear CHOP immunoreactivity was rarely detected in untransfected, bystander or SOD1wt expressing cells. In contrast, nuclear CHOP immunoreactivity was detected in 21% of diffuse SOD1A4V expressing cells and 87% of inclusion positive cells, indicating that visible inclusion formation correlated with ER stress and apoptotic signalling (Fig. 1A–D). These data also argue that ER stress precedes inclusion formation, since nuclear CHOP was detected in some diffuse SOD1A4V-expressing cells. We then treated SOD1wt, SOD1A4V and SOD1G85R expressing NSC-34 cells with tunicamycin, an inhibitor of N-linked glycosylation which induces ER stress. Tunicamycin treatment strongly induced ER stress, as indicated by an increase in CHOP protein detected by immunoblotting (Fig. 1E), and also significantly increased visible inclusion formation in mutant SOD1-expressing NSC-34 cells (Fig. 1F). These data confirm the relationship between inclusion formation and ER stress (Yamagishi et al., 2007).

Figure 1

Mutant SOD1 inclusion formation is linked with ER stress in NSC-34 cells. Immunofluorescence detection of nuclear CHOP (red, centre) in SOD1wt-EGFP (A) or diffuse (B) or inclusion-bearing (C) SOD1A4V-EGFP (green, left)-expressing motor neuron-like NSC-34 cells 72 h post-transfection determined by confocal microscopy. Nuclei are shown by Hoechst staining (blue) in the merged image (right panels). Asterisks indicate nuclei of non-transfected bystander cells, lacking CHOP induction. Note that only mutant SOD1-expressing cells form inclusions. Scale bars represent 10 μm. (D) Quantitation of frequency of nuclear CHOP induction in SOD1WT-EGFP- or SOD1A4V-EGFP-expressing NSC-34 cells 72 h post-transfection. At least 50 cells per treatment were analysed for nuclear CHOP immunoreactivity. Results are expressed as mean ± SEM, n = 3, *P < 0.05 versus respective bystander cells. (E) Immunoblot analysis of control and tunicamycin-treated NSC-34 cells, indicating a high level of CHOP induction following tunicamycin treatment. β-actin is shown as a loading control, and approximate molecular weight markers in kilodaltons are shown on the right for all blots. (F) Quantitation of frequency of visible EGFP-positive inclusion-bearing cells in EGFP, SOD1wt-, SOD1A4V- or SOD1G85R-EGFP-transfected vehicle control and 0.5 μg/ml tunicamycin-treated NSC-34 cells. At least 200 transfected cells per treatment were analysed per treatment. Results are expressed as mean ± SEM, n = 3, *P < 0.05 versus respective vehicle-treated controls by two-way ANOVA with Bonferroni's post-test.

PDI over-expression decreases mutant SOD1 aggregation and inclusion formation and inhibits ER stress

The presence of PDI within abnormal inclusions found in SOD1G93A rodents, motor neuronal cells (Atkin et al., 2006) and ALS patients (Atkin et al., 2008) implies that PDI may be a direct cellular defence against mutant SOD1 misfolding and aggregation. To assess this hypothesis, we established two clonal NSC-34 stable cell lines over-expressing different levels of human PDI tagged with fluorescent DsRed2 protein and transiently transfected these cell lines with SOD1-EGFP encoding constructs. We also examined the control parental NSC-34 cells, and a second control cell line over-expressing DsRed alone. Immunoblotting of cell lysates indicated that expression of endogenous mouse PDI was unchanged in the PDI-DsRed lines, which showed high levels of PDI-DsRed expression (Fig. 2A). In control cell lines 15% of cells transiently expressing SOD1A4V and 12% of cells expressing SOD1G85R formed visible inclusions. In contrast, the percentage of cells with inclusions was significantly less in PDI over-expressing cells than in controls, and was correlated with the level of expression of PDI-DsRed, suggesting a dose-dependent effect (Fig. 2B). The decrease in inclusion formation was not caused by decreased SOD1 expression, since soluble SOD1-EGFP levels were similar in each cell line (Fig. 2C). However, the levels of SDS-insoluble mutant SOD1 and ubiquitinated protein conjugates were lower in the PDI over-expressing cell lines than control lines, indicating that PDI over-expression attenuated aggregation and prevented proteasome disturbance (Fig. 2D). We then compared the level of ER stress between these cell lines. In PDI over-expressing cell lines, BiP and CHOP expression and PERK phosphorylation were lower than in controls (Fig. 2E), indicating a decrease in both ER stress and ER-stress induced apoptotic signalling by PDI over-expression.

Figure 2

PDI over-expression decreases mutant SOD1 aggregation and ER stress. (A) Immunoblot analysis of cell lysates of NSC-34 cells (cell line ‘1’) and modified clonal NSC-34 cells stably over-expressing either control DsRed (cell line ‘2’) or human PDI-DsRed (cell lines ‘3’ and ‘4’). Expression of endogenous mouse PDI was similar in all cell lines, and cell line 4 expresses ∼2-fold more PDI-DsRed than cell line 3. (B) Quantitation of frequency of inclusion formation in control and PDI-DsRed over-expressing cell lines, as indicated in A, transiently transfected with EGFP, SOD1wt-, SOD1A4V- or SOD1G85R-EGFP constructs. The percentages of transfected cells bearing visible EGFP-positive inclusions were calculated from at least 500 cells per treatment. Data are presented as mean ± SEM, n = 3, *P < 0.05 versus respective transfected control DsRed (‘2’) cell line by two-way ANOVA with Bonferroni's post-test. The incidence of inclusion formation in unmodified NSC-34 and the control DsRed cell lines was similar for both mutants. (C) Immunoblot analysis of soluble SOD1 levels in transfected cell lines analysed in B. SOD1 expression was similar in the control and PDI-DsRed over-expressing cell lines. Endogenous mouse SOD1 acts as an internal loading control. (D) Immunoblot analysis of SDS-insoluble protein from SOD1A4V-EGFP transfected cell lines analysed in B. The levels of insoluble SOD1A4V-EGFP (left panel) and high molecular weight (HMW) ubiquitinated proteins (right panel) were lower in the PDI-DsRed over-expressing cell lines than control cell lines. (E) Immunoblot analysis of BiP, phosphorylated PERK (p-PERK) and CHOP in EGFP, SOD1wt-, or SOD1A4V-EGFP-transfected control cell line 1 and PDI-DsRed over-expressing cell line 4. β-Actin is shown as a loading control.

PDI over-expression decreases mutant SOD1-induced cell death

We recently showed that visible mutant SOD1 inclusion formation is closely correlated with cell death (Soo et al., 2009). This implies that the decrease in mutant SOD1 aggregation and ER stress associated with PDI over-expression would also be expected to decrease cell death. We therefore co-transfected Neuro2a cells with SOD1wt- or SOD1A4V-EGFP with either DsRed or PDI-DsRed, and analysed apoptosis by examining nuclear morphology as previously described (Soo et al., 2009). Fragmented nuclei, indicative of apoptosis, were more common in cells expressing SOD1A4V than in those expressing SOD1wt; however, there were significantly fewer apoptotic nuclei in PDI-DsRed-expressing cells than in those expressing DsRed (Fig. 3F). This effect was also seen in non-inclusion bearing SOD1A4V-expressing cells, indicating that PDI over-expression protects against mutant SOD1-induced cell death both in the presence and absence of visible mutant SOD1 inclusions.

Figure 3

PDI over-expression decreases mutant SOD1 toxicity. Neuro2a cells were co-transfected with either SOD1wt-EGFP or SOD1A4V-EGFP (green, far left panels) and DsRed or PDI-DsRed (red, middle left). Hoechst stain to show nuclei (blue, middle right) and merged images (far right) are shown for (A) SOD1wt-EGFP with DsRed, (B) SOD1wt-EGFP with PDI-DsRed, (C) inclusion-positive SOD1A4V-EGFP with DsRed, (D) diffuse SOD1A4V-EGFP with PDI-DsRed and (E) inclusion-positive SOD1A4V-EGFP with PDI-DsRed. Open arrow in C indicates a condensed apoptotic nucleus, whereas closed arrow in E indicates a normal nucleus, both in inclusion-positive SOD1A4V-EGFP expressing cells. Scale bars represent 10 μm. (F) Quantitation of apoptotic nuclei. At least 100 cells per group were analysed for the presence of condensed or fragmented nuclei. Shown are ratios of the rate of apoptosis in each group to SOD1wt-EGFP and DsRed expressing cells. Data are presented as mean ± SEM, n = 3, *P < 0.05 versus respective DsRed control by two-way ANOVA with Bonferroni's post-test.

Knockdown of PDI increases mutant SOD1 inclusion formation

Since PDI over-expression decreased mutant SOD1 aggregation and inclusion formation, we hypothesized that knockdown of PDI would have the opposite effect. Treatment of NSC-34 cells with a commonly used disulphide isomerase inhibitor, bacitracin, increased mutant SOD1 inclusion formation in our previous study (Atkin et al., 2006). Since bacitracin is a relatively non-specific pharmacological agent, we used siRNA to specifically deplete PDI from Neuro2a cells expressing EGFP alone, SOD1wt-EGFP or SOD1A4V-EGFP. The percentage of cells bearing SOD1A4V-EGFP inclusions was significantly increased from 14% in control non-targeting siRNA (siCON) co-transfected cells to 26% in siRNA targeting PDI (siPDI) co-transfected cells (Fig. 4A). However, EGFP (data not shown) or SOD1wt-expressing cells did not form inclusions with or without PDI siRNA transfection in Neuro2a cells (Fig. 4A), indicating that the protective effect of PDI is specific to the mutant protein. Real-time PCR showed that siPDI co-transfection caused a significant decrease in the level of mRNA of the mouse gene encoding PDI (P4hb), of 63% in SOD1wt-EGFP expressing cells and 54% in SOD1A4V-EGFP-expressing cells (Fig. 4B). Immunoblotting also showed a decrease in PDI protein level with siPDI transfection, whereas SOD1-EGFP expression was consistent between siCON and siPDI transfected cells (Fig. 4C). Quantitation of immunoblots by densitometry showed that siPDI transfection caused a significant decrease in PDI protein level, normalised to β-actin, of 58% in SOD1wt-EGFP expressing cells and 63% in SOD1A4V-EGFP-expressing cells (Fig. 4D). These data indicate that the increase in SOD1A4V inclusion formation with siPDI transfection was caused by the specific decrease in PDI expression.

Figure 4

PDI knockdown by siRNA increases mutant SOD1 inclusion formation. Neuro2a cells were co-transfected with either non-targeting siRNA (siCON) or siRNA targeting PDI (siPDI) and SOD1wt-EGFP or SOD1A4V-EGFP plasmids. (A) Quantitation of frequency of visible EGFP-positive inclusion bearing cells 48 h post-transfection showing a significant increase in SOD1A4V-EGFP inclusion bearing cells with siPDI co-transfection compared to siCON. At least 200 transfected cells per treatment were analysed. (B) Analysis of mouse PDI (P4hb) mRNA level by real-time PCR confirming significant decreases in PDI mRNA in siPDI co-transfected cells compared to siCON. (C) Immunoblot analysis showing decreased expression of PDI in cells co-transfected with siPDI compared siCON (top panel). SOD1-EGFP levels were unaltered by siPDI transfection compared to siCON (middle panel). β-Actin is shown as a loading control. (D) Densitometric quantitation of PDI protein level normalized to β-actin from immunoblots confirming significant decreases in PDI protein in siPDI co-transfected cells compared to siCON. All data are presented as mean ± SEM, n = 3–6, *P < 0.05 versus respective siCON co-transfected cells.

PDI is S-nitrosylated in human ALS and SOD1G93A mouse spinal cord

We next investigated whether or not aberrant inactivation of PDI by S-nitrosylation occurred in ALS. Using a biotin switch assay we found increased levels of S-nitrosylated PDI in lumbar spinal cord samples from 21 sporadic ALS patients, compared to five control patients without evidence of neurological disease, implying that ALS tissue contained elevated levels of enzymatically inactivated PDI (Fig. 5A and Supplementary Fig. 1). Immunoblotting of input samples showed that total PDI was increased in the ALS patients, confirming our previous observations (Atkin et al., 2008). However, quantification by densitometry revealed that despite the increased levels of total PDI, S-nitrosylated PDI levels in ALS patients were 4.8-fold greater than controls, with S-nitrosylated PDI expressed as a ratio to total PDI, normalized to GAPDH as a loading control (Fig. 5B). This result indicates that the large increase in S-nitrosylated PDI was not due solely to an increase in total PDI levels, but instead reflects an increase in the relative amount of PDI which is S-nitrosylated in ALS patients. S-nitrosylated PDI levels were not affected by or correlated to post-mortem interval in the ALS patients. The specificity of the biotinylation reaction was confirmed by the dramatically decreased detection of S-nitrosylated PDI in separate samples lacking treatment with ascorbate, which is required to enhance the chemical decomposition of nitrosothiol groups required for reaction with the biotinylating reagent biotin-HPDP (Jaffrey et al., 2001) (Fig. 5C). Similarly, no S-nitrosylated PDI was detected in the absence of biotin-HPDP, indicating the specificity of the final streptavidin precipitation step of the assay (Fig 5C). We then examined spinal cord tissue from transgenic SOD1G93A mice, which provide the most widely accepted model of ALS (Turner and Talbot, 2008). S-nitrosylated PDI was abundant in spinal cord tissue from disease end-stage SOD1G93A mouse but was virtually undetectable in non-transgenic littermate control tissue (Fig. 5D). These data demonstrate that aberrant inactivation of PDI by S-nitrosylation is a common feature of both mutant SOD1-linked and sporadic forms of ALS.

Figure 5

PDI is S-nitrosylated in human ALS and transgenic SOD1G93A mouse spinal cord tissues. Fresh frozen spinal cord tissue was analysed using a biotin switch assay to detect S-nitrosylation. (A) Immunoblot analysis of biotin switch assay eluates, indicating an increase in S-nitrosylated (SNO)-PDI levels in 11 ALS patient samples, compared to three controls (top). Total PDI levels in input lysate samples are upregulated in patients compared to controls (middle). GAPDH levels in lysate samples (bottom) are shown as a loading control for total PDI. Representative immunoblots for an additional ten ALS patients and two controls are shown in Supplementary Fig. 1. (B) Densitometric quantitation of S-nitrosylated PDI levels in all 21 ALS and five control human spinal cord tissue samples relative to total PDI normalized to GAPDH. Data are presented as mean ± SEM, *P < 0.05 versus control by unpaired two-tailed t-test. (C) Control biotin switch analyses with samples lacking treatment with ascorbate or biotin-HPDP, indicating specificity of the assay. (D) Immunoblot analysis of biotin switch assay eluates of spinal cord tissue from disease end-stage SOD1G93A transgenic mouse and age-matched littermate control. Results are representative of three experiments for both human and mouse samples.

A small molecule mimic of PDI decreases mutant SOD1 aggregation and inclusion formation

The protection provided by PDI against ER stress and mutant SOD1 inclusion formation was tested further using BMC (also known as Vectrase-P), a small molecule mimic of the PDI active site. In cell-free in vitro systems and in yeast and bacterial cultures, BMC increases the folding and secretion of recombinant proteins (Woycechowsky et al., 1999; Kersteen and Raines, 2003), but BMC had not been tested previously in disease models. Treatment of SOD1A4V-EGFP transfected cells with BMC significantly decreased both the formation of mutant SOD1 inclusions and the level of SDS-insoluble mutant SOD1 in a dose-dependent manner, indicating that BMC is able to recapitulate the protective effects of PDI on mutant SOD1 aggregation (Fig. 6).

Figure 6

BMC, a small molecule mimic of the PDI active site, decreases mutant SOD1 aggregation and inclusion formation. NSC-34 cells were transfected with either EGFP or SOD1wt- or SOD1A4V-EGFP constructs and treated with BMC as indicated for 48 h. (A) Immunoblot analysis of expression of SOD1 in SDS-soluble and insoluble cell fractions from cells treated with or without 10 μM BMC. The amount of insoluble SOD1A4V was decreased following BMC treatment, as indicated in the panel on the right. Note that endogenous mouse SOD1 acts as an internal loading control. (B) Quantitation of frequency of inclusion formation in SOD1A4V-EGFP transfected cells treated with various concentrations of BMC as indicated. The percentages of transfected cells bearing EGFP-positive inclusions were calculated from at least 500 cells per treatment. Data are presented as mean ± SEM, n = 3, *P < 0.05 versus vehicle-treated control.

Discussion

The findings of this study demonstrate that PDI is protective against mutant SOD1 inclusion formation, which is linked directly with ER stress, in motor neuron-like NSC-34 cells, and that PDI decreases toxicity caused by mutant SOD1 expression. In addition, we have shown for the first time that PDI is modified by S-nitrosylation in transgenic mutant SOD1 mouse and human ALS spinal cord, indicating that the normal anti-aggregation properties of PDI are probably inactivated in ALS, providing new clues to the pathology of disease. This also indicates that the dysfunction of PDI is a common feature of sporadic ALS and is not restricted to those cases resulting from SOD1 mutations. Furthermore, we demonstrated that a small molecule mimic of the PDI active site, BMC, is also able to decrease mutant SOD1 inclusion formation and could be useful therapeutically.

Protection against mutant SOD1 by PDI is most likely modulated by disulphide bonding since treatment with BMC emulated the results of PDI over-expression, although it is possible that PDI also exerts protection by general inhibition of ER stress or its function as a protein chaperone. PDI contains two independent thioredoxin-like active sites consisting of the sequence CGHC, which are responsible for the disulphide isomerase activity of the protein (Ferrari and Soling, 1999), and BMC was designed based on this motif (Woycechowsky et al., 1999). Experimental use of BMC as a molecule to decrease protein misfolding in disease has not been previously reported, and this initial finding of protection against mutant SOD1 suggests that drugs similar to BMC could be of benefit in ALS and other neurodegenerative disorders.

If PDI interacts directly with mutant SOD1 to prevent aggregation, it is important to define the sub-cellular compartment in which this interaction occurs, since this could provide clues as to the cause of ER stress induction and ALS disease pathogenesis. PDI is conventionally recognized as being ER-resident; however, it is also located in the nucleus and extra-cellular matrix and on the cell surface (Ramachandran et al., 2001; Wilkinson and Gilbert, 2004). Similarly, ER stress can cause leakage of PDI from the ER to the cytoplasm (Tabata et al., 2007), suggesting that the cytoplasm is another possible site of interaction. PDI co-localizes with mutant SOD1 inclusions and binds to both SOD1wt and mutant SOD1 (Atkin et al., 2006), and an interaction between the typically ER-resident protein BiP with mutant SOD1 has also been reported (Kikuchi et al., 2006). Recently it has been suggested that SOD1 is actively recruited to microsomes via an ATP-dependent mechanism and that the ER is a site of increased mutant SOD1 misfolding (Urushitani et al., 2008). These data suggest that a proportion of cellular SOD1 protein could be found within the ER, and misfolding of mutant SOD1 in the ER lumen could account for the induction of ER stress in ALS. However, an interaction was found between mutant SOD1 and the cytoplasmic tail of the ER transmembrane protein Derlin-1, which led to an induction of ER stress (Nishitoh et al., 2008). The precise upstream cause of ER stress induction in both mutant SOD1-linked and other forms of ALS therefore remains to be determined.

We previously showed that ERp57, another abundant PDI family member, was also upregulated in transgenic SOD1G93A mice (Atkin et al., 2006), and here we note that although depletion of PDI by siRNA in Neuro2a cells caused an increase in mutant SOD1 inclusion formation, this increase was reasonably modest. The human PDI family includes nineteen proteins with varying function and specificities (Appenzeller-Herzog and Ellgaard, 2008), and hence compensation by one or more of the other PDI family members could be taking place, possibly including ERp57. Additionally, the PDI family member PDIA2 (also known as PDIp) is up-regulated in Parkinson's disease models and is present in human neuronal Lewy bodies (Conn et al., 2004). It is possible that different PDI family members could be involved in preventing misfolding of certain specific substrates in a cell-type specific manner, and further detailed investigation of the role of other PDI family members in ALS is therefore warranted.

The finding of PDI S-nitrosylation in sporadic ALS patients indicates that PDI loss of function occurs in the major form of disease. We predict that the role of PDI in sporadic ALS could be to decrease misfolding of other target proteins, possibly via disulphide bond modulation. Increased inhibition of PDI by S-nitrosylation has been found in Parkinson's and Alzheimer's disease brains, and this was postulated as a cause of pathology due to removal of the normal protective role of PDI in these diseases (Uehara, 2006, 2007). The findings of the current study suggest that a similar mechanism is involved in ALS, and this result could explain why upregulation of PDI does not prevent ALS disease progression. We previously showed that PDI protein levels in the spinal cord are highly abundant in large motor neurons in mice (Atkin et al., 2006), which could explain why motor neurons are particularly susceptible to PDI S-nitrosylation and inactivation in ALS. These data also suggest that prevention of nitrosative stress and S-nitrosylation of PDI could offer further therapeutic options for ALS.

The possibility that ER stress is an upstream mechanism in ALS also implies that ER stress inhibitors may be useful therapeutically. Inhibition of ER stress using salubrinal, a specific inhibitor of dephosphorylation of the PERK target protein eIF2α, protects against excitotoxic injury in rat brain (Sokka et al., 2007), and was most recently shown to delay disease progression and extend life span in three different mutant SOD1 mouse models of ALS (Saxena et al., 2009). In the same study, ER stress was detected very early in disease progression, and was evident in motor neurons before other cell types of the spinal cord. Genetic deletion of the BH3-only protein puma from SOD1G93A mice has also been shown to decrease ER stress and delay disease onset, although in the absence of any significant extension in life span (Kieran et al., 2007).

In summary, we propose that the propensity for mutant SOD1 misfolding involving aberrant disulphide bonding and a loss of PDI activity by S-nitrosylation could together lead to increased aggregation, inclusion formation and ER stress, eventually resulting in motor neuron death in ALS. PDI inactivation occurs not only in mutant SOD1-linked disease models, but also in sporadic human ALS patients, indicating that PDI mimics such as BMC and specific inhibitors of ER stress are potential therapeutic leads for treatment of ALS.

Funding

National Health and Medical Research Council of Australia (project grant 454749 and program grant 236805); Amyotrophic Lateral Sclerosis Association (USA); MND Research Institute of Australia; Bethlehem Griffiths Research Foundation; Henry H Roth Charitable Foundation Grant for MND Research; and Australian Rotary Health.

Supplementary material

Supplementary material is available at Brain online.

Acknowledgements

We thank Professor Neil Bulleid, University of Manchester, for the pCMV5 vector encoding human PDI, and Professor Neil Cashman, University of Toronto, for the NSC-34 cell line. We also thank Dr Bradley Turner for helpful input on the manuscript, and Kai Ying Soo and Professor Philip Nagley for technical assistance and access to unpublished data. Human spinal cord tissues were received from the MND Research Tissue Bank of Victoria, supported by the Victorian Brain Bank Network, the Mental Health Research Institute, Alfred Hospital, Victorian Forensic Institute of Medicine, The University of Melbourne, and funded by MND Research Institute of Australia and the Bethlehem Griffiths Research Foundation.

Footnotes

  • Abbreviations:
    Abbreviations
    ALS
    amyotrophic lateral sclerosis
    BMC
    (±)-trans-1,2-bis(mercaptoacetamido)cyclohexane
    CHOP
    C/EBP homologous protein
    ER
    endoplasmic reticulum
    PDI
    protein disulphide isomerase
    SOD1
    superoxide dismutase 1
    UPR
    unfolded protein response

References

View Abstract