Mutant SOD1 alters the motor neuronal transcriptome ... · Janine Kirby,1 Eugene Halligan,2 Melisa J. Baptista,1 Simon Allen,1 Paul R. Heath,1 Hazel Holden,1 Sian C. Barber,1 Catherine

doi:10.1093/brain/awh503 Brain (2005), 128, 1686–1706

Mutant SOD1 alters the motor neuronaltranscriptome: implications for familial ALS

Janine Kirby,1 Eugene Halligan,2 Melisa J. Baptista,1 Simon Allen,1 Paul R. Heath,1 Hazel Holden,1

Sian C. Barber,1 Catherine A. Loynes,1 Clare A. Wood-Allum,1 Joseph Lunec2 and Pamela J. Shaw1

1Academic Neurology Unit, University of Sheffield, School of Medicine and Biomedical Sciences, Sheffield and2Genome Instability Group, Department of Cancer Studies and Molecular Medicine, University of Leicester,Robert Kilpatrick Clinical Sciences Building, Leicester Royal Infirmary, Leicester, UK

Correspondence: Professor Pamela J. Shaw, Academic Neurology Unit, University of Sheffield, Medical School,Beech Hill Road, Sheffield S10 2RX, UKE-mail: [email protected]

Familial amyotrophic lateral sclerosis (FALS) is caused, in 20% of cases, by mutations in the Cu/Zn superoxidedismutase gene (SOD1). Although motor neuron injury occurs through a toxic gain of function, the precisemechanism(s) remains unclear. Using an established NSC34 cellular model for SOD1-associated FALS, weinvestigated the effects of mutant SOD1 specifically in cells modelling the vulnerable cell population, themotorneurons, without contamination from non-neuronal cells present in CNS. Using gene expression profiling, 268transcripts were differentially expressed in the presence of mutant human G93A SOD1. Of these, 197 weredecreased, demonstrating that the presence of mutant SOD1 leads to a marked degree of transcriptionalrepression. Amongst these were a group of antioxidant response element (ARE) genes encoding phase IIdetoxifying enzymes and antioxidant response proteins (so-called ‘programmed cell life’ genes), the expressionof which is regulated by the transcription factor NRF2.We provide evidence that dysregulation ofNrf2 and theARE, coupled with reduced pentose phosphate pathway activity and decreased generation of NADPH, rep-resent significant and hitherto unrecognized components of the toxic gain of function of mutant SOD1. Othergenes of interest significantly altered in the presence of mutant SOD1 include several previously implicated inneurodegeneration, as well as genes involved in protein degradation, the immune response, cell death/survivaland the heat shock response. Preliminary studies on isolated motor neurons from SOD1-associated motorneuron disease cases suggest key genes are also differently expressed in the human disease.

Keywords: amyotrophic lateral sclerosis; Nrf2; programmed cell life genes; SOD1

Abbreviations: Actb = b-actin; Akr1c13 = aldo-keto reductase family 1, member 13; ALS = amyotrophic lateral sclerosis;AP1= activator protein 1; ARE= antioxidant response element; Bag3=Bcl2-associated athanogene 3; Bnip3=E1B 19 kDa/Bcl2binding protein Nip3; B2m = b2-microglobulin; Ccl2 = chemokine (C-C motif) ligand 2; Cox4A = cytochrome c oxidasesubunit 4; Ddc = dopa decarboxylase; Erk = extracellular signal-regulated kinase; c-Fos = FBJ osteosarcoma oncogene;Gadd45a = growth arrest and DNA damage inducible 45 a; Gsn = gelsolin; GST = glutathione S-transferase;Gsta3 = glutathione S-transferase alpha 3; Gstm1/2 = glutathione S-transferase mu 1/2; G6pd = glucose-6-phosphatedehydrogenase; Hspa1b/4 = heat shock protein 1b/4; Idb2 = inhibitor of DNA binding 2; Jun = v-jun avian sarcoma virus 17oncogene homologue; Lmp7 = 20s proteasome b5 inducible subunit; Ltb4dh = leukotriene B4 12-hydroxydehydrogenase;c-Myc = myelocytomatosis oncogene; b-NF = b-napthoflavone; Nrf2 = nuclear factor erythroid 2-like 2;PA28a/b = proteasome activator 28a/b subunits; PA200 = proteasome activator 200 kDa; Pdcd6ip = programmed cell death6 interacting protein; PDTC = pyrrolidinedithiocarbamate; 6Pgd = 6-phosphogluconate dehydrogenase;Prdx3 = peroxiredoxin 3; Prdx4 = peroxiredoxin 4; Rgs2 = regulator of G-protein signalling 2; Scg2 = secretogranin II;Smn = survival motor neuron; SOD1 = Cu/Zn superoxide dismutase; S100a6 = calcyclin; t-BHQ = t-butylhydroquinone;Vegf = vascular endothelial growth factor

Received May 26, 2004. Revised March 9, 2005. Accepted March 15, 2005. Advance Access publication May 4, 2005

# The Author (2005). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]

IntroductionAmyotrophic lateral sclerosis (ALS) is one of the most

common adult onset neurodegenerative diseases, caused by

progressive degeneration of the upper and lower motor neur-

ons in the motor cortex, brainstem and spinal cord. Current

evidence suggests that multiple interacting factors contribute

to motor neuron injury in ALS. The four key pathogenetic

hypotheses comprise genetic factors (Hand and Rouleau,

2002), oxidative stress (Cookson and Shaw, 1999), glutama-

tergic toxicity (Heath and Shaw, 2002) and protein mis-

folding/aggregation (Wood et al., 2003). ALS is sporadic in

90–95% and familial in 5–10% of cases. Mutations in the

ubiquitiously expressed free radical scavenging enzyme,

Cu/Zn superoxide dismutase (SOD1), are causative in 20%

of familial cases (Rosen et al., 1993) and current insights into

the pathogenesis of ALS have arisen predominantly from

studying the effects of SOD1 mutations. Mutant SOD1 pro-

duces motor neuron injury by a toxic gain of function and

although the exact mechanism of action is unclear, several

hypotheses exist, including aberrant free radical handling,

abnormal protein aggregation and increased susceptibility

to excitotoxicity (Hand and Rouleau, 2002).

The availability of powerful genomics technologies pro-

vides the opportunity to unravel complex regulatory and

interactive pathways that govern neuronal phenotype and

changes in disease states. Several recent studies have attemp-

ted to investigate differential gene expression in relation to

motor neuron degeneration. One study of RNA extracted

from whole spinal cord homogenates from G93A mutant

SOD1 mice used Affymetrix Gene Chips to monitor differ-

ential gene expression (Olsen et al., 2001), whilst a similar

study used cDNA membrane arrays (Yoshihara et al., 2002).

Pooled RNA extracts from whole human spinal cord have also

been used to analyse gene expression changes in ALS cases

compared with controls (Malaspina et al., 2001). While these

studies confirmed that microglial and astrocytic activation,

and a neuro-inflammatory response all occur in SOD1-related

motor neuron injury, their limitation is that they have not

profiled gene expression specifically in the vulnerable cell

population: the motor neurons. Since motor neurons are

only a minority cell population within the spinal cord, any

changes in gene expression will have been diluted and poten-

tially masked by changes occurring in other cell types.

We have previously established a robust cellular model to

examine the molecular pathophysiology of motor neuron

injury associated with mutations in SOD1. NSC34 cells,

which are a hybrid mouse motor neuron/neuroblastoma

cell line, retain the ability to proliferate whilst exhibiting

many motor neuron characteristics (Cashman et al., 1992;

Durham et al., 1993). NSC34 cells have been stably transfected

with one of several mutant forms of human SOD1, wild-type

SOD1 or vector only and single cell clones derived by limiting

dilution (Menzies et al., 2002a). This cell line provides a good

model to investigate the effects of mutant SOD1 specifically in

cells with a motor neuronal phenotype. Using this model we

have previously demonstrated several important insights into

the toxicity produced by the mutant SOD1 protein, including:

(i) biochemical changes reflecting an increased tendency to

apoptosis with increased expression of cleaved caspase 9 and

annexin V staining on the cell surface under basal culture

conditions (Cookson et al., 2002; Sathasivam et al., 2004);

(ii) the development of morphologically abnormal mitochon-

dria, with impaired activity of complexes II and IV of the

respiratory chain and impaired cellular bioenergetic status

(Menzies et al., 2002a); and (iii) alterations in the cytosolic

proteome with specific changes in expression and function of

proteins involved in nitric oxide metabolism, antioxidant

defence and protein folding and degradation (Allen et al.,

2003). The likely importance of these changes is reinforced

by studies of apoptotic pathways, mitochondrial function and

expression of cellular proteins in murine SOD1 transgenic

models and in human CNS tissue (Wong et al., 1995; Martin,

1999; Phul et al., 2000).

In the present study we aimed to elucidate further the

pathophysiological alterations induced in motor neurons

by the presence of mutant SOD1 protein, using gene expres-

sion profiling. Although previously we have used cDNA

membrane arrays to screen 588 genes for changes in gene

expression (Kirby et al., 2002), by employing the Affymetrix

microarray system with the U74Av2 Murine GeneChip we can

perform simultaneous analysis of 6000 well characterized

genes and a further 6000 expressed sequence tag (EST) tran-

scripts. This provides a more powerful screening protocol

with the potential to identify not only individual genes that

are altered in response to mutant SOD1, but also specific

intracellular pathways. We wished initially to examine

changes occurring solely within motor neuronal cells both

uncontaminated by the presence of other cell types and with-

out the effects of interactions between the cells, while accept-

ing the role that non-neuronal astrocytes and microglia may

play in the generation and/or propagation of motor neuron

injury (Pramatarova et al., 2001; Lino et al., 2002).

Material and methodsReagentsTissue culture reagents were purchased from Invitrogen, as were

reagents for DNase treatment of RNA and cDNA synthesis. SYBR

Green PCR Master Mix was obtained from Applied Biosystems, PCR

primers from MWG Biotech and assay substrates from Sigma.

Reagents used for the microarray experiments were those recom-

mended by Affymetrix. The rabbit anti-mouse nuclear factor ery-

throid 2-like 2 (NRF2) polylclonal antibody was a gift from Professor

John Hayes, University of Dundee, the rabbit anti-human glucose-6-

phosphate dehydrogenase (G6PD) polyclonal antibody was from

Jose Bautista, University of Madrid and the rabbit anti-

peroxiredoxin 3 (PRDX3) polyclonal antibody was from Professor

Chi Dang, John Hopkins University School of Medicine. The sheep

anti-mouse SOD1 polyclonal antibody was purchased from Calbio-

chem, the mouse anti-cytochrome c oxidase subunit 4 (COX4A)

Mutant SOD1 induced transcriptome changes Brain (2005), 128, 1686–1706 1687

monoclonal antibody from Molecular Probes, mouse anti-survival

motor neuron (SMN) antibody from BD Transduction Laboratories

and mouse anti-b-actin (ACTB) monoclonal antibody from Sigma

(clone AC-40). Horseradish peroxidase (HRP)-conjugated second-

ary antibodies were obtained from Dako, except for the anti-sheep

HRP, which was purchased from Sigma.

Cell culture and RNA preparationThe NSC34 cell lines stably transfected with pCEP4 (pCEP), normal

human SOD1 (pCN) or G93A mutant human SOD1 (pC93) have

been described previously (Menzies et al., 2002a). Three replicate

sets of cells were grown in Dulbecco’s modified Eagle’s medium

(DMEM) with 10% fetal calf serum (FCS) to 80% confluency before

being harvested. Following a wash in Hank’s buffered saline solution

(HBSS), the cells were resuspended in RNAlater (Ambion), accord-

ing to the manufacturer’s instructions. Total RNA was isolated from

5–10 million NSC34 cells using TRI Reagent (Sigma), following

the manufacturer’s instructions. Following quantitation, 4 mg total

RNA was run on a formaldehyde denaturing agarose gel to assess the

RNA quality.

Microarray hybridization and analysisThe mRNA fraction of the total RNA was converted to cDNA by

reverse transcription (RT) using the SuperScript kit (Invitrogen) in

combination with a T7-(dT)24 oligomer [50-GCCAGTGAATTGTA-

ATCCGACTCACTATAGGGAGGCGG-(dT)24-30] (Genset Oligos),

according to the manufacturer’s instructions. The cDNA was used as

template for T7 RNA polymerase in vitro transcription (IVT), using

the BioArray High Yield RNA Transcript Labelling Kit (Enzo) to

incorporate fluorescent label, according to the manufacturer’s pro-

tocol. cRNA (10 mg) was applied to the GeneChip U74Av2A

array, according to Affymetrix protocols. This contains 6000 func-

tionally characterized genes and a further 6000 EST clusters. Three

chips were used for each of the triplicate NSC34 cell lines (pCEP,

pCN and pC93). Array washing and staining was performed in the

Fluidics Station 400 according to the Affymetrix protocol. Arrays

were scanned twice using the GeneArray laser scanner (Agilent

Technologies). The Microarray Analysis Suite (MAS v5.0) software

(Affymetrix) was used to monitor scanning and to convert the raw

image file into a cell intensity file (‘.CEL’).

Data analysisQuality control of each chip was carried out by confirming

the detection of spiked controls, low ratios of actin and GAPDH

signals to ensure detection of full length transcripts (mean actin 3–50

ratio 1.101 6 0.201; Gapdh 3–50 ratio 1.001 6 0.142) and that scaling

factor (mean 1.671 6 0.514), RawQ (mean 3.754 6 0.756), average

background (mean 113.5 6 39.28) and percentage of genes expressed

(mean 36.57 6 2.333), were similar for each chip.

The nine .CEL files generated by MAS 5.0 were converted into

‘.DCP’ files using dCHIP version 1.2 (www.dCHIP.org) (Li and

Hung Wong, 2001). The .DCP files were normalized, and the raw

gene expression data generated was then normalized using the

dCHIP system of model-based analysis (PM-MM). Comparative

analysis of global gene expression profiles was performed using

the dCHIP software. The three replicate murine NSC34 cell lines

transfected with the empty vector (pCEP) were designated as

‘baseline’ (B), and the three murine NSC34 cell lines transfected

with the normal human SOD1 (pCN) and the 3 murine NSC34

cell lines transfected with mutant human SOD1 (pC93) were

designated as ‘experiment’ (E). Genes differentially expressed

two-fold or higher in the two cell lines transfected with human

normal and mutant SOD1 versus control vector alone were then

identified by defining the appropriate filtering criteria in the

dCHIP software (E/B-2; E-B > 50, B-E > 50).

Genes were grouped functionally using the Gene Ontology (GO)

system available through NetAffx (www.affymetrix.com/analysis/

index.affx), taking into consideration the biological process, cellular

component and molecular function listed for each gene.

Quantitative RT–PCRThree sets of NSC34 cells (pCEP, pCN and pC93) were harvested

when flasks reached 80% confluency, with each set from a separate

passage. As an additional control for the specificity of the gene

changes, a further set of cells containing the G37R mutation were

also screened. Following a wash in HBSS, RNA was extracted using

TRIzol (Invitrogen), according to the manufacturer’s instructions.

RNA (2 mg) was DNase I treated and the sample divided into two.

cDNA synthesis was performed, with and without Superscript II

reverse transcriptase, following the manufacturer’s protocol. Primers

were designed to genes in key pathways to confirm their differential

expression, and where possible, crossed intron/exon boundaries.

Primer concentrations were optimized using 0.5 ml pCEP cDNA.

Quantitative RT–PCR (Q-PCR) was performed using 0.5 ml cDNA,

1· SYBR Green PCR Master Mix (Applied Biosystems) and optim-

ized concentrations of forward and reverse primers, to a total volume

of 20 ml. Primer sequences and optimized final primer concentra-

tions are given in the figure legends. Following an initial denatura-

tion of 95�C for 10 min, products were amplified by 40 cycles of 95�C

for 15 s and 60�C for 1 min, on an MX3000P Real Time PCR System

(Stratagene). Finally, a dissociation curve was performed to ensure

amplification of a single product and absence of primer dimers.

For each of the genes, a standard curve using 1, 0.5, 0.25, 0.125

and 0.0625 ml of cDNA was carried out to determine the efficiency

of the PCR was 100% 6 10%, such that values of pCEP, pCN,

pC37 and pC93, normalized to Actb expression, could be determined

using the ddCt calculation (SYBR Green PCR mix and RT–PCR

protocol; Applied Biosystems). The levels of expression of each

gene in pCN, pC37 and pC93 cells are shown as a percentage of

the gene expression seen in pCEP control cells. Paired t-tests were

used to analyse the data and determine the statistical significance of

any differences in gene expression.

Isolation of human motor neurons andamplification of RNAAn Arcturus Pixcell II laser capture microdissector was used to isolate

individual motor neurons from snap-frozen samples of spinal cord of

two neuropathologically normal control subjects (one male and one

female, mean age 50 6 3 years, post mortem delay 16.5 6 2.5 h) and

two subjects with familial ALS, carrying an I113T mutation in the

SOD1 gene (one male and one female, mean age 62 6 2 years, post

mortem delay 23.3 6 3.5 h), as previously described (Heath et al.,

2002). Following extraction of RNA using PicoPure RNA Isolation

Kit (Arcturus), a double round of RNA amplification was performed

using the RiboAmp RNA Amplification kit (Arcturus), according to

the manufacturer’s protocol. First strand cDNA synthesis was then

performed on 1 mg of cRNA, and semi-quantitative RT–PCR was

performed. Primers were designed to cross intron/exon boundaries.

1688 Brain (2005), 128, 1686–1706 J. Kirby et al.

Aliquots of PCR product were removed at 25, 30, 35 and 40 cycles

to ensure that densitometric analyses were carried out during the

logorithmic range of amplification. Densitometry was carried out

using the AlphaImager system and Alpha Innotech Spot Densito-

metry software (Alpha Innotech, San Leandro, CA, USA). Relative

level of expression for each of the genes was determined by normal-

izing to levels of ACTB expression in each of the cDNA samples, and

results combined to compare normals against FALS cases. Statistical

analysis was not performed, as data were only available from two

cases in each group.

Enzyme and metabolite assaysFor all enzyme and metabolite assays, the NSC34 cells were grown

for 96 h, harvested at 80% confluency, and S1 cytosolic fractions

prepared as described previously (Allen et al., 2003). All statistical

analyses were done using paired t-tests. G6PD assay reaction mix-

tures (Lee, 1982) consisted of 100 mM Tris/HCl pH 8.0 containing

1 mM glucose-6-phosphate and 1 mM NADP. 6-Phosphogluconate

dehydrogenase (6PGD) assay reaction mixtures consisted of 100 mM

Tris/HCl pH8.0 containing 1 mM 6-phosphogluconate and 1 mM

NADP (Lee, 1982). Malic enzyme assay reaction mixtures consisted

of 67 mM triethanolamine pH 7.4 containing 4 mM MnCl2·4H2O,

1 mM L-malate and 1 mM NADP (Hsu and Lardy, 1969). All reac-

tions were initiated by adding 200 mg of NSC34 post-nuclear S1

protein per ml assay reaction (Allen et al., 2003). The increase in

absorbance at 340 nm was measured at 1 min intervals for 5 min at

room temperature.

Total intracellular NADP(H) concentrations were determined

with modification to the method of Zerez et al. (1987). NSC34

cell monolayers previously seeded at a density of 3 · 105 cells per

T75 flask and incubated at 37�C for 96 h were resuspended in PBS

pH 7.4. The cells were centrifuged at 400 g for 5 min then resuspen-

ded in 1 ml per flask of an ice-cold solution of 10 mM nicotinamide,

20 mM NaHCO3 and 100 mM Na2CO3. The cells were immediately

snap-frozen in liquid nitrogen, rapidly thawed in a room temperat-

ure water bath and then centrifuged at 13 000 g at 4�C for 5 min to

obtain supernatants. Total NADP(H) assay reaction mixtures

consisted of 100 mM Tris/HCl pH 8.0 containing 5 mM EDTA,

2 mM phenazine ethosulphate, 0.5 mM MTT, 1.3 U/ml G6PD,

1 mM glucose-6-phosphate and either 20–200 nM standard

NADP or 100 ml lysate supernatant per ml assay reaction. Reactions

were initiated by addition of glucose-6-phosphate and the increase in

absorbance at 570 nm was measured as above.

Protein extraction and western blottingTotal cellular protein and the S1 cytosolic fraction were prepared

according to protocols described previously (Cookson et al., 1998;

Allen et al., 2003). Ten micrograms of protein was run on 14%

SDS–PAGE gels, and electroblotted onto Immobilon-P membranes.

Membranes were then blocked in TBS-Tween [20 mM Tris–HCl

pH 7.6, 137 mM NaCl, 0.1% (v/v) Tween-20] with 5% (w/v)

dried skimmed milk. The membranes were probed either with 1 in

5000 dilution of sheep anti-mouse polyclonal SOD1, 1 in 4000 dilu-

tion of rabbit anti-mouse polyclonal NRF2, 1 in 500 of rabbit anti-

human G6PD polyclonal or 1 in 10 000 of mouse anti-human SMN,

followed by peroxide-conjugated secondary antibody of 1 in 5000

for anti-sheep antibody, 1 in 2000 for anti-rabbit antibodies and 1 in

1000 for the anti-mouse antibody. To control for variation in protein

concentration of the samples, membranes were also probed with 1 in

2000 of mouse anti-ACTB monoclonal, followed by 1 in 1000 of

secondary antibody of rabbit anti-mouse.

The mitochondrially enriched preparations of NSC34 cell lines

were generated by differential centrifugation as previously described

(Menzies et al., 2002a). Ten micrograms of protein was run on 14%

SDS–PAGE gels, and electroblotted onto membranes. These were

probed with 1 in 1000 dilution of rabbit anti-PRDX3 and as a con-

trol, 1 in 10 000 of mouse anti-COX4A. Both secondary antibodies

were used at 1 in 1000.

All proteins were visualized using ECL western blotting detection

reagents kit (Amersham), and densitometry carried out using the

AlphaImager system and Alpha Innotech Spot Densitometry soft-

ware (Alpha Innotech). Following normalization to either ACTB for

cytosolic and COX4A for mitochondrial fractions, protein levels of

each gene in pC37 and pC93 cells were calculated as a percentage of

the gene expression seen in pCEP control cells. Paired t-tests were

used to analyse the data and determine the statistical significance of

any differences in gene expression.

Pharmacological manipulation of antioxidantresponse element (ARE)-driven geneexpressionNSC34 cells expressing human G93A SOD1 (pC93) or vector only

(pCEP) were seeded into T175 flasks at a density of 6 · 105 cells and

then grown for 96 h. The cells were then incubated for 24 h with

fresh medium containing either 10 mM pyrrolidinedithiocarbamate

(PDTC), 10 mM t-butylhydroquinone (t-BHQ), 10 mM b-naphtho-

flavone (b-NF), 15 mM sulforaphane or DMSO vehicle solvent only

at a 1:1000 dilution. S1 cytosolic fractions were then prepared as

described previously and relative specific activities for both G6PD

and total glutathione S-transferase (GST) were determined as above.

For MTT assays, NSC34 cells were seeded into 96-well plates at a

density of 2000 cells/cm2 and grown for 5 days in DMEM + 10% FCS.

Medium was replaced with fresh DMEM 6 serum 6 PDTC

(1–10 mM) for 48 h. MTT assays were performed as described pre-

viously (Cookson et al., 1998). Statistical analysis was performed

using the Wilcoxon matched pairs test.

ResultsMicroarray analysisThe transcription profiles of the motor neuron-like NSC34

cell line transfected with vector, normal human SOD1 or

mutant G93A SOD1 were generated using the murine Gene-

Chip U74Av2. Between 32% and 39% of the 12 000 transcripts

represented on the GeneChip were detected as present in the

three groups of NSC34 cells. To identify genes differentially

expressed in the presence of the mutant SOD1 protein, com-

parisons were initially made between the transcription profiles

obtained from the vector and the mutant SOD1 protein trans-

fected cell lines using the analysis software dChip version 1.2

(Li and Hung Wong, 2001). Transcripts that showed an

increase/decrease of at least two-fold, and a difference in sig-

nal intensity between the baseline and experimental arrays of

at least 50 units of fluorescence were selected for further

evaluation. Comparisons between the vector and the normal

SOD1 protein transfected cell line transcription profiles were


also obtained and transcripts identified according to the same

selection criteria.

The analyses identified 268 transcripts that were consist-

ently differentially expressed in the presence of G93A mutant

SOD1 protein: 197 transcripts were decreased and 71 tran-

scripts were increased (Tables 1 and 2). Thus, 268 of the 12 000

transcripts expressed on the array (2.2%), or 6–7% of genes

expressed within the NSC34 cells, are differentially expressed

in the presence of mutant SOD1. The genes were categorized

according to their molecular function, and included genes

involved in antioxidant and stress responses, apoptosis,

immunity and protein degradation. Unlike previous studies,

where heterogeneous cell populations were used, there were a

far greater number of genes showing decreases in expression

than increases. It appears that expression of mutant SOD1

within motor neuronal cells leads to a marked degree of

transcriptional repression.

In contrast to the numerous changes observed in the pres-

ence of mutant SOD1, only 16 transcripts were differentially

expressed specifically in the presence of normal SOD1, 10

increased and six decreased (Table 3).

Expression levels of endogenous mouseand transfected human SOD1 in theNSC34 cell linesQ-PCR demonstrated that there was no significant change in

the gene expression level of endogenous murine SOD1 in the

four groups of cell lines (Fig. 1A). Western blotting demon-

strated similar levels of murine SOD1 protein in all the cell

lines, and that normal human SOD1 or the G37R and G93A

mutant forms of the SOD1 protein were expressed at

comparable levels (Fig. 1B). The relative expression levels

of human SOD1 to endogenous murine SOD1 are compar-

able to the levels of mutant SOD1 expression expected in the

human disease.

Verification of microarray data byRT–PCR, western blotting andfunctional assaysOut of 268 genes diffentially expressed in the presence of

mutant G93A SOD1, 23 genes were selected for verification

based upon several criteria: (i) presence in a pathway of spe-

cific interest, particularly those related to the antioxidant

response; (ii) a potential involvement in ALS or neurodegen-

eration; and (iii) a high fold change. Initially, semi-

quantitative RT–PCR was used to verify the changes in the

NSC34 cell lines (data not shown) and in isolated motor

neurons from human SOD1-associated cases (see below).

This was followed up by Q-PCR, western blotting and func-

tional assays in the NSC34 cell lines to validate the changes

observed by microarray analysis, and to determine whether

the alterations correlated with protein and enzyme activity

levels. The additional G37R SOD1 mutant-containing cell line

was also included in the verification studies to provide further

confirmation of the importance of key genes in SOD1-

mediated neurodegeneration. A further five of the differen-

tially expressed genes had previously been identified as altered

in the presence of mutant SOD1 using proteomic approaches

[leukotriene B4 12-hydroxydehydrogenase (Ltb4dh), 20s

proteasome beta 5 inducible subunit (Lmp7), glutathione

S-transferase mu 1 (Gstm1), arginosuccinate synthase (Ass1),

neuronal nitric oxide synthase (nNos)]. These have been val-

idated in the NSC34 cell line and human SOD1-associated

familial ALS cases (Allen et al., 2003).

Antioxidant response proteinsOxidative stress is known to play a role in ALS (Cookson and

Shaw, 1999), and cellular oxidative stress results in the bind-

ing of the transcription factor NRF2 to ARE, located in the

promoters of phase II detoxifying enzymes (Chan et al., 1996).

In the presence of mutant SOD1, an unexpected three-fold

decrease in the level of Nrf2 was detected. This was associated

with decreases in multiple downstream phase II detoxifying

enzymes and antioxidant enzymes, including Gst enzymes

[Gstm1, 7.5-fold; Gstm2, two-fold; Gst omega 1 (Gsto1),

three-fold; Gst alpha 3, (Gsta3) four-fold], a 2.4-fold decrease

in G6pdx, a three-fold decrease in aldoketoreductase family 1,

member C13 (Akr1c13) and a 125-fold decrease of Ltb4dh.

Previous work has suggested that the ARE may be negat-

ively regulated by the activator protein 1 (AP1) component

FBJ osteosarcoma oncogene (c-FOS) (Venugopal and Jaiswal,

1996; Wilkinson et al., 1998), which was increased over three-

fold in the presence of mutant SOD1. Extracellular signal-

regulated kinase (ERK), also known as mitogen-activated

protein kinase 3 (MAPK3), reportedly phosphorylates both

NRF2, allowing its translocation into the nucleus, and mye-

locytomatosis oncogene (MYC) transcription factor (Owuor

and Kong, 2002). MYC induces many genes including the

peroxiredoxins (Prdx) (Wonsey et al., 2002). Both Erk1

and c-Myc were decreased, two- and three-fold, respectively,

in the presence of mutant SOD1, as well as the MYC target

genes Prdx3 and Prdx4, which decreased four- and two-fold,

respectively.

To confirm the differential expression of specific genes

involved in the antioxidant response element, Q-PCR was

carried out on RNA isolated from the four groups of NSC34

cells: pCEP, pCN, pC37 and pC93. Significant decreases in

expression were detected in Nrf2, Gsta3, G6pdx, Akr1c13 and

Prdx4, in both mutant containing cell lines compared with

vector-only transfected cells (Fig. 2A and B; P values given in

figure legends). Significant decreases were also detected in

Prdx3 and C-myc along with a significant increase in c-Fos,

in the presence of mutant G93A SOD1 compared with vector-

only transfected cells, although these changes were not

significant in the G37R SOD1 mutant cell line (Fig. 2B).

Western blotting was used to investigate the protein levels

of the key gene products NRF2, G6PD and PRDX3. Analysis

of total cell lysate clearly showed a significant decrease in

protein levels of NRF2 and G6PD, in the presence of both


Table 1 Genes found to be decreased in the presence of mutant G93A SOD1

GenBank accession No. Gene Fold changea

Apoptosis, cell survival and cell deathNM_013863 Bcl2-associated athanogene 3 �2.16NM_013492 clusterin �2.82NM_007836 growth arrest and DNA-damage-inducible 45 alpha �11.29

Calcium binding proteinsNM_008047 follistatin-like 1 �8.55NM_008175 granulin �2.44NM_008189 guanylate cyclase activator 1a (retina) �4.13NM_009242 secreted acidic cysteine rich glycoprotein �2.43NM_009129 secretogranin II �130.19

Cell adhesionU69176 laminin, alpha 4 �3.49NM_009369 transforming growth factor, beta induced, 68 kDa �3.14

Cell growthNM_011819 growth differentiation factor 15 �2.36NM_008344 insulin-like growth factor binding protein 6 �5.38NM_010784 midkine �2.30

CytoskeletalNM_009635 advillin �3.15 (�2.61)NM_010354 gelsolin �6.59NM_031170 keratin complex 2, basic, gene 8 �2.99M35131 neurofilament, heavy polypeptide �2.30NM_175155 putative adapter and scaffold protein �2.47

DevelopmentNM_010051 dickkopf homolog 1 (Xenopus laevis) �3.22NM_008595 manic fringe homolog (Drosophila) �3.47

ImmunityNM_009735 beta-2 microglobulin �7.71NM_011795 C1q related factor �80.81NM_008176 chemokine (C-X-C motif ) ligand 1 �4.61NM_013499 complement receptor related protein �4.25NM_008199 histocompatibility 2, blastocyst �2.44NM_010380 histocompatibility 2, D region locus 1 �3.30NM_008200 histocompatibility 2, D region locus 4 �2.24NM_010390 histocompatibility 2, Q region locus 1 �4.91NM_010392 histocompatibility 2, Q region locus 2 �3.16NM_010394 histocompatibility 2, Q region locus 7 �5.81NM_023124 histocompatibility 2, Q region locus 8 �4.03NM_010398 histocompatibility 2, T region locus 23 �3.67NM_010545 Ia-associated invariant chain �3.58NM_010708 lectin, galactose binding, soluble 9 (galectin 9) �2.01NM_018851 SAM domain and HD domain, 1 �2.12

Kinases/phosphatasesNM_021515 adenylate kinase 1 �4.21NM_007788 casein kinase II, alpha 1 polypeptide �2.13NM_013689 cytoplasmic tyrosine kinase, Dscr28C related (Drosophila) �14.23NM_026268 dual specificity phosphatase 6 �2.08NM_018869 G protein-coupled receptor kinase 5 �4.56NM_144554 induced in fatty liver dystrophy 2 �2.52NM_010693 lymphocyte protein tyrosine kinase �4.28NM_011952 mitogen activated protein kinase 3 (Erk1) �2.10NM_008696 mitogen-activated protein kinase kinase kinase kinase 4 �2.84NM_028444 protein kinase C delta binding protein �3.03NM_011212 protein tyrosine phosphatase, receptor type E �4.60 (�2.53)NM_011361 serum/glucocorticoid regulated kinase �13.96NM_030724 uridine-cytidine kinase 2 �2.01

Lipid-relatedNM_008509 lipoprotein lipase �5.94NM_011110 phospholipase A2, group V �3.76

MetabolismNM_007414 ADP-ribosylarginine hydrolase �2.17NM_016672 dopa decarboxylase �7.30NM_008813 ectonucleotide pyrophosphatase/phosphodiesterase 1 �3.87


Table 1 Continued


NM_016772 enoyl coenzyme A hydratase 1, peroxisomal �2.33NM_019477 fatty acid-coenzyme A ligase, long chain 4 �3.54NM_024243 fucosidase alpha-L-1 �2.06NM_013847 glycine C-acetyltransferase (2-amino-3-ketobutyrate-coenzyme A ligase) �2.04NM_010255 guanidinoacetate methyltransferase �2.02XM_181343 Riken cDNA 5730409F23Rik �2.12NM_009214 spermine synthase �2.67

Mitochondrial proteinsNM_025301 mitochondrial ribosomal protein L17 �2.17NM_013898 translocase of inner mitochondrial membrane 8 homolog a �2.70

mRNA processingNM_026611 ribonuclease T2 �2.00NM_027332 RNA processing factor 1 �2.28X60388 smn pseudogene �3.55NM_011420 survival motor neuron �2.02

Nuclear proteinsNM_026780 GCIP interacting protein p29 �2.75NM_013779 melanoma antigen, family L, 2 �3.46NM_010922 mitochondrial ribosomal protein L40 �2.34NM_031375 neugrin �2.29NM_010822 N-methylpurine-DNA glycosylase �3.16NM_019738 nuclear protein 1 �3.96NM_053086 nucleolar and coiled-body phosphoprotein 1 �2.15NM_022889 pescadillo homolog 1 �2.77NM_008894 polymerase (DNA directed), delta 2, regulatory subunit (50 kDa) �2.06NM_028696 RIKEN cDNA 5830411E10Rik �8.30NM_021336 small nuclear ribonucleoprotein polypeptide A0 �2.01NM_021521 trinucleotide repeat containing 11 (THR-associated protein, 230 kDa subunit) �2.81

Oxidoreductase activityNM_025801 6 phosphogluconate dehydrogenase �2.07NM_007410 alcohol dehydrogenase 5 (class III), chi polypeptide �2.31NM_009731 aldo-keto reductase family 1, member B7 �5.35NM_013778 aldo-keto reductase family 1, member C13 �3.12NM_008062 G6PD X-linked �2.42NM_019468 G6PD 2 �3.24NM_008162 glutathione peroxidase 4 �2.30NM_016763 hydroxyacyl-coenzyme A dehydrogenase type II �3.12NM_173011 isocitrate dehydrogenase 2 (NADP+), mitochondrial �2.03NM_025968 leukotriene B4 12-hydroxydehydrogenase �125.06NM_008810 pyruvate dehydrogenase E1 alpha 1 �2.29NM_011467 sepiapterin reductase �2.01NM_031201 tissue specific transplantation antigen P35B �2.62NM_009414 tryptophan hydroxylase �4.25

Protein degradationNM_009616 a disintegrin and metalloproteinase domain 19 (meltrin beta) �2.56NM_009984 cathepsin L �4.14NM_008788 procollagen C-proteinase enhancer protein �2.19XM_203562 proteasome activator PA200 �2.14NM_011189 proteasome 28 subunit, alpha �2.29NM_011190 proteasome 28 subunit, beta �2.18NM_010724 proteosome subunit, beta type 8 (lmp7) �2.30NM_011018 sequestosome 1 �3.06

SignallingNM_011333 chemokine (C-C motif) ligand 2 �120.64 (5.23)NM_009142 chemokine (C-X3-C motif) ligand 1 �2.04NM_008003 fibroblast growth factor 15 �2.12 (�2.50)NM_010276 GTP binding protein (skeletal muscle) �2.25NM_011183 presenilin 2 �2.57NM_009061 regulator of G-protein signaling 2 �12.07NM_008113 Rho GDP dissociation inhibitor (GDI) gamma �2.39NM_007486 Rho GDP dissociation inhibitor (GDI) beta �2.22NM_012026 Rho-guanine nucleotide exchange factor �3.48


Table 1 Continued


Stress responseNM_010356 glutathione S-transferase alpha 3 �4.02 (�2.88)NM_010362 glutathione S-transferase omega 1 �3.02NM_010358 glutathione S-transferase mu 1 �7.57NM_008183 glutathione S-transferase mu 2 �2.22XM_286803 heat shock protein 1B �22.93AF109906 heat shock protein 4 �2.85NM_007452 peroxiredoxin 3 �4.88NM_016764 peroxiredoxin 4 �2.40

TranscriptionNM_007679 CCAAT/enhancer binding protein (C/EBP), delta �2.56NM_008242 forkhead box D1 �3.04NM_010849 myelocytomatosis oncogene �3.24NM_010882 necdin �2.05NM_008668 Ngfi-A binding protein 2 �2.20NM_183248 NK6 transcription factor related, locus 2 (Drosophila) �2.17NM_010907 nuclear factor of kappa light chain gene enhancer in B-cells inhibitor alpha �2.90NM_010902 nuclear factor, erythroid derived 2, like 2 �3.14NM_009089 polymerase (RNA) II (DNA directed) polypeptide A �2.05NM_009031 retinoblastoma binding protein 7 �2.33NM_021525 RNA terminal phosphate cyclase like-1 �2.03NM_027139 TAF9 RNA polymerase II, TATA box binding protein (TBP)-associated factor, 32 kDa �2.06NM_011603 TATA box binding protein-like 1 �2.13NM_019916 T-cell leukemia, homeobox 3 �5.18NM_009372 TG interacting factor �2.37NM_011640 transformation related protein 53 �2.20

TranslationNM_025936 arginyl-tRNA synthetase �2.39NM_175937 cytoplasmic adenylation element binding protein 2 �2.39NM_133767 mitochondrial translation initiation factor 2 �2.08NM_025437 RIKEN cDNA 1500010B24 gene �2.07

TransportNM_013454 ATP-binding cassette, subfamily A, member 1 �3.63 (�2.18)NM_008761 FXYD domain-containing ion transport regulator 5 �6.51NM_011255 retinol binding protein 4, plasma �4.27NM_018861 solute carrier family 1 member 4 (glutamate transporter) �2.05NM_009194 solute carrier family 12 member 2 (Na/K/Cl transporter) �2.19NM_009437 thiosulfate sulfurtransferase, mitochondrial �3.06

MiscellaneousNM_007408 adipose differentiation related protein �2.78NM_010171 coagulation factor III �11.94AA_794108 developmental pluriopotency associated protein 2 �12.08NM_028610 developmental pluriopotency associated protein 4 �3.82NM_080595 Emu1 gene �4.34NM_133362 erythroid differentiation regulator �3.69NM_024169 FK506 binding protein 11 �2.10NM_025360 integral type 1 protein �2.31NM_134090 KDEL ER protein retention receptor 3 �5.75NM_025327 keratinocytes associated protein 2 �2.23NM_013599 matrix metalloproteinase 9 �2.03NM_030700 melanoma antigen, family D, 2 �4.29Z31362 neoplastic progression 3 �2.35NM_023456 neuropeptide Y �2.04NM_175329 Nurr77 downstream protein 2 �5.70NM_008756 occludin �2.78NM_011022 ovary testis transcribed �3.88NM_011150 peptidylprolyl isomerase C-associated protein �5.61NM_139198 placenta-specific 8 �13.95NM_009344 pleckstrin homology-like domain, family A, member 1 �12.61NM_011171 protein C receptor, endothelial �2.24NM_009057 recombination activating gene 1 gene activation �2.12NM_009052 reduced expression 3 �2.05NM_024226 reticulon 4 �2.02


mutant G93A and G37R SOD1, compared with both vector

only and normal human SOD1-transfected cells (Fig. 3A–C).

Since PRDX3 is a mitochondrial protein, a mitochondrial

enriched fraction of cell extract was used for western blotting.

The results show a low level of PRDX3 protein in both

the G37R and G93A SOD1 mutant-containing cells compared

with vector-only transfected cells. Whilst this decrease is only

significant for the pC93 cell, both mutants are significantly

decreased compared with the normal SOD1-transfected cells

(Fig. 3D and E). In addition, previous work with this cell

model in our laboratory identified decreased protein and

mRNA levels of GSTM1 and LTB4DH in the presence of

mutant SOD1 (Allen et al., 2003).

Enzyme activity of NADPH generatorsand levels of NADPHG6PD is an NADPH generator, which is required as a cofactor

to provide reducing power in numerous enzymatic reactions,

and the decrease in G6pdx expression in the presence of mut-

ant SOD1 is reflected in the protein level. 6PGD is also an

NADPH generator, and was also found by the microarray

analysis to be decreased in the presence of G93A mutant

SOD1. To investigate the effect of decreased levels of

G6pdx and 6Pgd on enzymatic activity and NADPH levels,

functional assays were carried out. Using cytosolic extracts,

G6PD and 6PGD activities were significantly decreased to

40% and 60%, respectively, in the presence of G93A mutant

SOD1, and to 66% and 76%, respectively, in the presence of

G37R mutant SOD1, compared with vector-only transfected

cells (Fig. 4A). Total NADP(H) levels were also reduced by

25% in the both mutant transfected cell lines (Fig. 4B). How-

ever, the activity of malic enzyme, another NADPH generator,

was identical in all three cell lines (Fig. 4A). Although there

was no change in gene expression of enzymes requiring

NADPH, [e.g. glutathione reductase (Gsr); thioredoxin

reductase (Txnrd)], there was a decrease in glutathione per-

oxidase 4 (Gpx4) (2.3-fold), and the peroxiredoxins

Prdx3 (4.8-fold) and Prdx4 (2.4-fold), which use the reduced

forms of glutathione and thioredoxin synthesized by gluta-

thione reductase and thioredoxin reductase, respectively.

There was also a decrease in several of the Gst enzymes

Table 1 Continued


NM_009254 serine (or cysteine) proteinase inhibitor, clade B, member 6 �2.27NM_009255 serine (or cysteine) proteinase inhibitor, clade E, member 2 �2.02NM_138684 single WAP motif protein 2 �7.31NM_009191 suppressor of K+ transport defect 3 �2.86NM_011576 tissue factor pathway inhibitor �4.72NM_011595 tissue inhibitor of metalloproteinase 3 �2.37 (2.58)NM_019634 transmembrane 4 superfamily member 2 �2.21NM_020026 UDP-Gal:betaGlcNAc beta 1,3-galactosyltransferase, polypeptide 3 �2.11NM_009529 Xlr-related, meiosis regulated �2.85

Unknown functionNM_177474 Chr 19, 1357 �2.21NM_138587 Chr 6, 176 �3.52AA763213 clone IMAGE:1229497 �7.06AI153421 clone IMAGE:1429340 �5.35AW047929 clone IMAGE:1529010 �2.36AW012491 clone IMAGE:2582523 �3.12NM_134139 expressed sequence AA407588 �2.04NM_134017 expressed sequence AI182287 �2.03NM_144847 hypothetical protein MGC12117 �2.93AI839653 RIKEN cDNA 1810008O21Rik �2.04AA646408 RIKEN cDNA 2210016F16Rik �3.29NM_173750 RIKEN cDNA 2700007P21Rik �2.81NM_026029 RIKEN cDNA 2700085E05Rik �2.56NM_027274 RIKEN cDNA 2810025M15Rik �2.57AI845988 RIKEN cDNA 2810404F18Rik �2.08NM_023323 RIKEN cDNA 2810470K21Rik �2.09NM_133684 RIKEN cDNA 2810484M10Rik �2.12NM_030240 RIKEN cDNA 2900092E17Rik �2.05NM_028447 RIKEN cDNA 3110038B19Rik �2.02NM_026582 RIKEN cDNA 5031439A09Rik �13.51NM_198004 RIKEN cDNA 5133401N09Rik �2.60

Functional groupings determined using NetAffx. aFold change gives the comparison between cells transfected with vector only andmutant SOD1. Figures in parentheses show the fold change in the normal SOD1-transfected cells compared with vectortransfected cells when it was greater than two.


Table 2 Genes increased in the presence of mutant G93A SOD1


AngiogenesisNM_010402 heart and neural crest derivatives expressed transcript 2 9.94NM_009505 vascular endothelial growth factor A 2.01

ApoptosisNM_009760 BCL2/adenovirus E1B 19 kDa-interacting protein 1, NIP3 2.53NM_011052 programmed cell death 6 interacting protein 2.10

Calcium binding proteinsNM_008791 Purkinje cell protein 4 2.10NM_011313 S100 calcium binding protein A6 (calcyclin) 10.45

Cell adhesionNM_023061 melanoma cell adhesion molecule 6.27

CytoskeletalNM_153058 microtubule-associated protein, RP/EB family, member 2 2.17NM_019410 profilin 2 2.21NM_021467 troponin I, skeletal, slow 1 7.31

DevelopmentNM_011839 mab-21-like 2 (Caenorhabditis elegans) 8.04

Kinases/phosphatasesNM_021605 NIMA-related expressed kinase 7 2.28NM_011074 PFTAIRE protein kinase 1 4.92NM_008862 protein kinase inhibitor, alpha 2.11NM_011218 protein tyrosine phosphatase, receptor type, S 2.50NM_009050 ret proto-oncogene 2.32NM_133819 protein phosphatase 1, regulatory (inhibitor) subunit 15b 2.58

MetabolismNM_007494 argininosuccinate synthetase 1 2.76NM_016690 heterogeneous nuclear ribonucleoprotein D-like 2.14NM_008549 mannosidase 2, alpha 1 4.10

mRNA processingNM_133195 bruno-like 4 RNA binding protein 2.80NM_019550 polypyrimidine tract binding protein 2 2.66 (2.18)

Nuclear proteinsNM_007840 DEAD/H box polypeptide 5 2.05NM_008197 H1 histone family, member 0 2.03NM_008252 high mobility group box 2 2.58NM_013548 histone 1, H3f 2.11NM_010496 inhibitor of DNA binding 2 8.28 (2.20)NM_008321 inhibitor of DNA binding 3 2.94NM_172458 RIKEN cDNA 9030612M13 gene 2.01

Oxidoreductase activityNM_008712 nitric oxide synthase 1, neuronal 2.33NM_013626 peptidylglycine alpha-amidating monooxygenase 2.05

Protein degradationXM_134935 ariadne ubiquitin-conjugating enzyme E2 binding protein homolog 1 (Drosophila) 2.35NM_008853 praja1 2.53NM_009481 ubiquitin specific protease 9, X chromosome 2.10NM_011665 ubiquitin-conjugating enzyme E2I 2.68

SignallingNM_007722 chemokine orphan receptor 1 139.67 (2.73)NM_011182 pleckstrin homology, Sec7 and coiled/coil domains 3 2.90NM_178376 Ras-related GTP binding A 2.50NM_009706 Rho GTPase activating protein 5 2.40NM_011368 src homology 2 domain-containing transforming protein C1 2.65NM_007706 suppressor of cytokine signaling 2 3.43NM_019491 v-ral simian leukemia viral oncogene homolog A (ras related) 2.98

TranscriptionNM_010096 early B-cell factor 3 2.52NM_010234 FBJ osteosarcoma oncogene 3.33NM_010723 LIM domain only 4 2.39NM_010788 methyl CpG binding protein 2 2.04NM_008709 N-myc 2.11NM_013665 short stature homeobox 2 4.31NM_021788 sin3 associated polypeptide, 30 kD 2.01


(two- to seven-fold), which eliminate compounds to which

reduced glutathione has been attached. Glutathione is

synthesized from cysteine and glutamate, and the glutamate

is transported into the cell by solute carrier 1, member 4

(Slc1a4). This gene also showed a two-fold decrease in

expression in the presence of mutant SOD1.

Genes previously implicated in ALS andneurodegenerationSeveral genes which have previously been implicated in ALS

or neurodegeneration were identified as being altered in our

cellular model of SOD1 associated ALS. Following deletion of

the hypoxia response element in the promoter of the vascular

endothelial growth factor (Veg f ), transgenic mice developed

a late onset, progressive degeneration of the motor neurons

in the spinal cord and brainstem (Oosthuyse et al., 2001).

Veg f was increased two-fold in the presence of mutant

G93A SOD1, whilst the calcyclin or S100 calcium binding

protein A6 (S100a6), which is overexpressed in reactive astro-

cytes of G93A SOD1 transgenic mice (Hoyaux et al., 2000),

is increased 10-fold. Whilst the increase in S100a6 was

confirmed by Q-PCR in the presence of both mutations,

the changes in VEGF were not (Fig. 5A and B). The

Smn gene, mutations in which cause spinal muscular atrophy

(Lefebvre et al., 1995), was decreased by two-fold in the

presence of mutant G93A SOD1, and this was confirmed at

the protein level in both mutant cell lines by western blotting

(Fig. 3F).

Genes altered in other pathways ofinterestOther genes of interest that were altered in the presence

of G93A SOD1 include those involved in protein degrada-

tion, immunity and apoptosis, cell survival, and cell death.

Involvement of the proteasome is implicated by the

presence of protein aggregates in neuronal cell bodies. The

20s proteasome associates with regulatory proteins that

function as proteasome activators, such as the ubiquitin-

independent proteasome activator 28 (PA28) complex and

the proteasome activator 200 kDa (PA200). The genes encod-

ing both PA28 subunits (PA28a and PA28b) and PA200 are

decreased two-fold in the presence of mutant SOD1, as is the

inducible 20s proteasomal subunit Lmp7. LMP7 was previ-

ously found to be decreased in the cytosolic proteome of this

model (Allen et al., 2003).

A large number of MHC encoded genes involved in anti-

gen presentation were decreased in the presence of mutant

SOD1. These included the MHC class I heavy chain variants

and b2-microglobulin (B2m). B2m, which was decreased

7.7-fold in the presence of mutant G93A SOD1, was verified

Table 2 Continued


TranslationNM_025824 basic leucine zipper and W2 domains 1 2.33

TransportNM_007581 calcium channel, voltage-dependent, beta 3 subunit 2.06NM_007711 chloride channel 3 2.20NM_008477 kinectin 1 2.26NM_011401 solute carrier family 2 member 3 (glucose transporter) 3.28NM_023908 solute carrier organic anion transporter family member 3a1 2.22

MiscellaneousNM_198018 active BCR-related gene 2.26NM_019875 ATP binding cassette, subfamily B, member 9 2.37NM_007657 CD9 antigen 3.39NM_021548 cyclic AMP phosphoprotein, 19 kDa 2.13NM_175009 DC6 protein 2.41NM_033612 elastase 1, pancreatic 2.40AI843046 tropomyosin 4 2.29NM_009517 wild-type p53-induced gene 1 2.07

Unknown functionNM_133791 Chr 8, D8Ertd594e 2.85NM_177643 clone IMAGE:3498439 2.03 (2.43)AA111360 clone IMAGE:554879 8.61AA589382 RIKEN cDNA 1110004P15Rik 6.50 (�3.24)AI840191 RIKEN cDNA 2210409B22Rik 2.33NM_025940 RIKEN cDNA 2610042L04Rik 2.34AW120755 RIKEN cDNA 2700010L10Rik 2.35AA822539 RIKEN clone A430054I22 5.46

Functional groupings determined using NetAffx. aFold change gives the comparison between cells transfected with vector only and mutantSOD1. Figures in parentheses show the fold change in the normal SOD1-transfected cells compared with vector transfected cellswhen it was greater than two.


by Q-PCR, with both mutants showing significant decreases,

compared with vector-only and normal human SOD1-

transfected cells (Fig. 5B). Only three genes involved in apop-

tosis showed alterations in expression of more than two-fold:

E1B 19 kDa/BCL2 binding protein Nip3 (Bnip3), a proapop-

totic member of the BCL2 family and programmed cell death

6 interacting protein (Pdcd6ip) were increased two-fold. The

anti-apoptotic protein, BCL2 associated athanogene 3 (Bag3),

which enhances the activity of BCL2 and also interacts with

HSC70/HSP70 proteins, was decreased two-fold, as were two

heat shock protein (HSP) 70 protein encoding genes Hspa1b

and Hspa4, decreased 22- and two-fold, respectively. The

decrease in Hspa1b was confirmed in both mutants by

Q-PCR (Figure 5B), although the increase in Bnip3 and

decreased in Bag3 expression were not (Figure 5C). Another

gene involved in cell survival which was decreased in the

presence of mutant G93A SOD1 was growth arrest and

DNA-damage inducible 45 a (Gadd45a), and the change

was confirmed in both mutants by Q-PCR (Fig. 5C).

Genes showing highest fold changesThere were 15 genes the expression levels of which were mark-

edly altered (>10-fold) in the presence of mutant SOD1, and

24 whose expression altered between five- and 10-fold. Of

these genes, those which were located in pathways of specific

interest were: (i) inhibitor of DNA binding 2 (Idb2) increased

by eight-fold; (ii) chemokine (C-C motif ) ligand 2 (Ccl2)

decreased by 120-fold; (iii) regulator of G-protein signalling

2 (Rgs2), which was decreased by 12-fold; (iv) dopa

decarboxylase (Ddc), decreased by seven-fold; (v) secreto-

granin II (Scg2) decreased by 130-fold; and (vi) gelsolin

(Gsn) decreased by six-fold. All these changes were confirmed

in both SOD1 mutant-containing cell lines by Q-PCR, except

Ccl2 and Gsn, where the decreases in G37R SOD1 expressing

cells were not significant compared with the vector-only

transfected cells but were to the normal human SOD1-

transfected cells (Figure 5A and D).

Expression in isolated human motorneurons from SOD1-associated ALS casesTo determine whether the changes observed in our cellular

model were applicable to the human disease state, preliminary

experiments were undertaken using semi-quantitative

RT–PCR on motor neurons isolated from spinal cord sections

Table 3 Genes found to be increased and decreased inthe presence of normal SOD1. Functional groupingsdetermined using NetAffx

GenBank accessionNo.

Gene Foldchangea

CytoskeletalNM_007392 actin, alpha 2, smooth

muscle, aorta475.31

NM_007791 cysteine and glycinerich protein 1

2.52

Kinases/phosphatasesNM_007463 aortic preferentially

expressed gene 12.18

mRNA processingNM_031249 cleavage stimulation

factor, 30 pre-RNAsubunit 2, tau

2.00

MiscellaneousNM_007725 calponin 2 2.61AI842649 myosin, light polypeptide 9,

regulatory8.11

NM_009052 reduced expression 3 3.06NM_015825 SH3-binding domain glutamic

acid-rich protein2.05

Unknown functionAA833425 RIKEN cDNA 3110003A17Rik 2.49NM_057172 D3Ertd330e 2.09

Cell cycleNM_008885 peripheral myelin protein,

22 kDa�3.00

Nuclear proteinNM_007841 DEAD box polypeptide 6 �2.02

OxidoreductaseNM_172291 cDNA BC024806 �2.12

TranscriptionNM_008887 paired-like homeobox 2a �2.29

MiscellaneousNM_018827 cytokine receptor-like factor 1 �4.01NM_009150 selenium binding protein 1 �2.29

aFold change gives the comparison between cells transfected

with vector only and normal SOD1.

A

B

Fig. 1 Q-PCR and western blotting results for the expression ofhuman and mouse SOD1 in NSC34 cells. (A) No changes in geneexpression levels of endogenous mouse Sod1 (Mm Sod1) wereidentified between the cell lines (n = 3). Q-PCR primers (and finalconcentrations) Mm Sod1 F 50 GGC CCG GCG GAT GA 30

(100 nM), R 50 CGT CCT TTC CAG CAG TCA CA 30 (300 nM)and Actb F 50 ATG CTC CCC GGG CTG TAT 30 (900 nM),R 50 CAT AGG AGT CCT TCT GAC CCA TTC 30 (300 nM).(B) Representative western blot showing the human SOD1protein migrating just above that of endogenous SOD1 in thenormal and mutant SOD1-transfected cells.


of two familial cases carrying the I113T SOD1 mutation and

two neuropathologically normal control cases, using laser

capture microdissection. These initial studies suggest the dif-

ferential expression of NRF2, B2M and VEGF is also present

in the human disease (Figure 6). However, the significant

decrease of SCG2 in the presence of mutant SOD1 was

not supported. In addition, our previous work has shown

significant decreases in the expression of GSTM1, LMP7

and LTB4DH compared with control motor neurons from

human SOD1-related ALS cases (Allen et al., 2003).

Reversal of mutant SOD1 down-regulationof antioxidant enzyme activityMultiple studies have demonstrated that the expression of

genes containing AREs can be promoted using small electro-

philic compounds that activate transcription factors such

as NRF2 and AP1. To investigate whether the mutant

SOD1-dependent downregulation of antioxidant enzymes

was reversible, NSC34 expressing either vector only or

G93A SOD1 were treated with either 10 mM PDTC,

10 mM t-BHQ, 10 mM b-NF or 15 mM sulforaphane,

then post-nuclear supernatants prepared from the cells

were assayed for G6PD and total GST activity. t-BHQ,

b-NF and sulforaphane had no effect upon antioxidant

enzyme activity (data not shown). In contrast, PDTC was

shown to significantly increase the G6PD and GST activities

in cells expressing G93A SOD1 from 44% to 63% and 72% to

88%, respectively (Fig. 7A). Similarly, PDTC also increased

G6PD and GST activities in cells expressing vector only from

100% to 123% and 100% to 125%, respectively.

NSC34 cells expressing G93A SOD1 have previously been

shown to be more susceptible than vector-only expressing

cells to cell death following oxidative stress induced by serum

withdrawal (Cookson et al., 2002; Menzies et al., 2002a). In

this model, PDTC is partly protective and can increase pC93

viability after 48 h serum withdrawal from 46 6 2.4% to

54 6 3.2% (Fig. 7B).

DiscussionMutations in SOD1 were identified as causative for ALS in

1993 (Rosen et al., 1993), and although there is substantial

evidence that motor neuron degeneration occurs through a

toxic gain of function of the mutant Cu/Zn SOD protein, the

underlying pathophysiological mechanism is still unknown.

In order to identify cellular pathways that are altered, specif-

ically in motor neurons, in the presence of mutant SOD1,

transcription profiles were obtained from our cellular model

of SOD1-associated familial ALS. Comparison of transcrip-

tion profiles from either mutant G93A SOD1 or normal SOD1

versus vector-only transfected cells resulted in the identifica-

tion of 268 transcripts altered by more than two-fold in the

presence of mutant SOD1; 197 transcripts were decreased and

71 transcripts increased. Although this is a large number of

genes, it became apparent that distinct pathways were affec-

ted. The genes could be categorized into several groups

including antioxidant response and related genes, protein

degradation, immunity, apoptosis, and cell survival/cell

death genes. There were also changes in several genes that

have been previously implicated in neurodegeneration.

A striking feature of the gene expression profile within

NSC34 motor neuronal cells in the presence of mutant

SOD1 was the marked degree of transcriptional repression.

A

B

Fig. 2 Q-PCR results for NRF2-regulated genes. In the presenceof mutant G93A and G37R SOD1, respectively, significantdecreases in the antioxidant response genes (A) Nrf2 (P = 0.0004;P = 0.0055), Gsta3 (P < 0.0001; P < 0.0001), G6pdx (P = 0.0006;P = 0.0097) and Akr1c13 (P = 0.0025; P = 0.0041) were detectedwhen compared with control cells expressing the vector only(n = 3). Significant decreases in (B) Prdx4 (P < 0.0001;P = 0.0141), Prdx3 (P = 0.0018) and c-Myc (P = 0.011) and anincrease in c-Fos (P = 0.014) gene expression were detected inpC93 cells compared with vector only, but only Prdx4 reachedsignificance in pC37 transfected cells (n = 3). *P < 0.05, **P < 0.01,***P < 0.001, in mutant cell lines compared with vector-onlytransfected cells. Q-PCR primer sequences (and finalconcentrations) were Nrf2 F 50 TGG AGG CAG CCA TGA CTGA 30 (100 nM), R 50 CTG CTT GTT TTC GGT ATT AAG ACACT 30 (100 nM); Gsta3 F 50 TGA ACT CCT CTA CCA TGT GGAAGA 30 (300 nM), R 50 TCT GGC TGC CAG GTT GAA G 30

(300 nM); G6pdx F 50 CAG CCC AAT GAG GCA GTA TAC A 30

(900 nM), R 50 CAT CAG GGA GCT TCA CAT TCT TG 30

(300 nM); Akr1c13 F 50 CTG CCT TGA TTG CAC TTC GAT 30

(100 nM), R 50 TCT CTC ATC TCA TTC TCT TTG AAA CTC T30 (100 nM); Prdx4 F 50 TTG GTT CAA GCC TTC CAG TAC A 30

(100 nM), R 50 TGG GAT TAT TGT TTC ACT ACC AGG TT 30

(100 nM); Prdx3 F 50 GCA GCT GCG GGA AGG TT 30 (300 nM),R 50 GGC AGA AAT ACT CCG GGA AAT 30 (100 nM); c-Myc F50 CGA GCT GAA GCG CAG CTT 30 (100 nM), R 50 GGC CTTTTC GTT GTT TTC CA 30 (100 nM); c-Fos F 50 CAT CAC TCCCGG CAC TTC A 30 (300 nM), R 50 GGA CTC TGA GGG CGACGA A 30 (300 nM).


This is in contrast to gene expression changes identified in

whole spinal cord homogenates, where an increase in expres-

sion of genes reflecting reactive gliosis and inflammatory

mechanisms have been observed, most likely arising

from numerically dominant non-neuronal cells and inter-

actions between the cell types (Malaspina et al., 2001;

Olsen et al., 2001). Recent evidence has highlighted the

potential importance of transcriptional repression in other

neurodegenerative disorders including Huntington’s disease

(Zuccato et al., 2003).

The antioxidant response and relatedgenesNRF2 is a bZIP transcription factor that is a master regulator

of ARE-driven gene expression, which includes phase II

A B

D E

F G

C

Fig. 3 Western blots of antioxidant response genes and SMN, confirming changes in mRNA levels are also reflected at the protein level forthese genes. (A) Representative western blots of NRF2, G6PD and ACTIN. There is a significant reduction in the protein level of (B) NRF2in pC37 (P = 0.0013) and pC93 (P = 0.0103) and (C) G6PD in pC37 (P = 0.0026) and pC93 (P = 0.0009) in the total cell lysate comparedwith pCEP, following normalization to the level of the control protein ACTB, which does not differ between the three sets of cells (n = 4).(D) Representative western blots of PRDX3 and COX4A. (E) A reduction of PRDX3 in the mitochondrially enriched fraction from pC93(P = 0.015) and pC37 cells was also detected compared with pCEP following normalization to the control protein COX4A, which does notdiffer in protein expression levels between the three sets of cells (n = 3). Although pC37 was not significantly decreased compared withpCEP, both pC37 and pC93 mutants were significantly different compared with pCN transfected cells (P = 0.0189 and P = 0.0159,respectively). (F) Representative blot of SMN. (G) Expression of SMN shows a reduction in the total cell lysate in both pC37 and pC93 cells(P = 0.016 and P = 0.0065), compared with vector-only transfected cells following normalization to ACTB. *P < 0.05, **P < 0.01, ***P < 0.001in mutant cell lines compared with vector-only transfected cells.


detoxification enzymes and antioxidant proteins, in a process

that has been referred to as ‘programmed cell life’ (Lee et al.,

2003b). NRF2 is post-translationally regulated by kelch-like

ECH-asssociated protein 1 (KEAP1), a cytosolic actin binding

protein localized to the cytoskeleton, which binds NRF2

within the cytosol under basal conditions. Oxidative stress

causes the release of NRF2 followed by translocation to the

nucleus, where it induces transcription by binding to the ARE

sequences in the promoters of specific genes.

Recent studies applying microarray analysis to Nrf2–/– cel-

lular models have identified genes that are either directly or

indirectly transcriptionally regulated by NRF2. Although

these studies vary with regard to the stimulus [sulforaphane

(Thimmulappa et al., 2002); mitochondrial toxins (Lee

et al., 2003b); t-BHQ (Lee et al., 2003a); 3H-1,2-dithiole-

3-thione (Kwak et al., 2003)], or specific pathways invol-

ved [phosphatidylinositol-3 kinase dependent/independent

(Li et al., 2002)], the expression of multiple genes, in addition

to the phase II detoxifying enzymes, have been shown to be

regulated by NRF2. Correlating these data with the results

obtained in the present experiments, NRF2 is not only

involved in the transcriptional regulation of Gsta3, Gstm1,

Gstm2, G6pdx, 6Pgd, Akr1c13, Ltb4dh and Prdx3 (Fig. 8),

but also Gadd45a, B2m, Rgs2, Ddc, Scg2 and Gsn, all of

which were decreased in the presence of mutant SOD1 in

motor neuronal cells.

The level of Nrf2 expression was decreased whilst the

AP1 complex component c-Fos was increased. It has been

reported that FOS negatively regulates ARE-driven expression

(Venugopal and Jaiswal, 1996; Wilkinson et al., 1998), acting

by heterodimerizing with another leucine zipper protein and

binding the ARE. However, it has also been shown that AP1

activity is repressed during oxidative stress due to direct

oxidation of specific cysteine residues in the v-jun avian sar-

coma virus 17 oncogene homologue (JUN) and FOS proteins

(Abate et al., 1990). Opposing regulatory effects of a single

transcriptional factor are not unprecedented, for example

regulation of urokinase is achieved by a heterodimer of JUN

and ATF-2 positively regulating this gene, whilst a JUN and

FOS heterodimer represses transcription (De Cesare et al.,

1995). Further work investigating the regulation of ARE trans-

cription is required to elucidate the role of FOS. However,

since NRF2 plays a role in basal cellular redox homeostasis

and in the mounting of a cellular cytoprotective response

to oxidative insults, we suggest that dysregulation of the

‘programmed cell life’ response may represent a key compon-

ent of the toxicity of mutant SOD1. This is supported by

evidence for: (i) the presence of oxidative stress in cellular

and animal models of ALS, as well as in human spinal cord

tissue (Cookson and Shaw, 1999); (ii) the partial restoration

of G6PD and GST activity of cells expressing mutant SOD1 to

levels of those expressing normal SOD1, and increased cell

viability following serum withdrawal, by PDTC, a compound

known to promote binding of both NRF2 and AP-1 (Meyer

et al., 1993; Wild et al., 1999); and (iii) the indication of

decreased expression of NRF2 in motor neurons from

SOD1-associated familial ALS cases. If chronic oxidative stress

provides a mechanism to explain the toxicity of mutant

SOD1, it also points to potential pharmacological and recom-

binant approaches aimed at reversing or preventing the dele-

terious effects of the mutant enzyme.

Dysregulation of the pentose phosphatepathway and NADPH synthesisThe significantly decreased expression and activities of the

pentose phosphate pathway enzymes G6PD and 6PGD are

likely to result in a significant lowering of the cell’s ability to

produce reduced NADPH, which is known to be crucial for

the regeneration of the antioxidant capacity within the CNS.

For example peroxiredoxin, responsible for eliminating

hydrogen peroxide, is dependent upon reduced thioredoxin,

which is regenerated by thioredoxin reductase at the expense

of NADPH. Hydrogen peroxide is also removed by gluta-

thione peroxide, which is dependent upon reduced gluta-

thione, and this is regenerated by glutathione reductase,

again at the expense of NADPH. In our cellular model,

A

B

Fig. 4 Enzyme and metabolite assays using S1 cytosolic fractions ofthe NSC34 cells. The activities of NADPH generators (A) G6PDand 6PGD were significantly decreased in pC37 and pC93compared with pCEP (P = 0.017 and P = 0.018 for G6PD,respectively; n = 3) and (P = 0.01 and P = 0.0002 for 6PGD; n = 4),whilst malic enzyme activity was unchanged between the threecells lines (n = 3). (B) Total cellular NADP(H) levels weresignificantly decreased in pC37 (P = 0.04), and pC93 (P = 0.04),compared with pCEP (n = 3). The activity of G6PD, 6PGD andmalic enzyme in pCN and pC93 is expressed as a percentage ofthe activities measured in the pCEP cells. *P < 0.05, **P < 0.01,***P < 0.001 in mutant cell lines compared with vector-onlytransfected cells.


there is reduced expression and function of NADPH-

generating enzymes, total NADP(H) levels are decreased,

and, as previously reported, there are decreased levels of

reduced glutathione in the presence of mutant SOD1

(Allen et al., 2003). Therefore, in the presence of mutant

SOD1, oxidative stress is likely to arise from dysregulation

of the pentose phosphate pathway, resulting in reduced avail-

ability of NADPH essential to maintain the major intracellular

antioxidants glutathione and thioredoxin in their reduced

states. Recent work supports a pivotal role for G6pd in the

cellular response to oxidative stress (Filosa et al., 2003).

Mouse embryonic stem cells carrying an exonic deletion in

the G6pd gene, under conditions of oxidative stress, fail to

upregulate the activity of the pentose phosphate pathway,

resulting in lowered NADPH/NADP ratio, decreased reduced

glutathione and ultimately cell death.

B

C D

A

Fig. 5 Q-PCR results for genes involved in neurodegeneration, immunity, stress response, apoptosis and those genes showing largeexpression changes in NSC34 cells. A significant increase in expression was seen in (A) S100a6 in both mutant containing cell lines(P = 0.0147 for pC37 and P = 0.0049 for pC93) compared with vector only, and in Idb2 (P = 0.03 for both pC37 and pC93), compared withcontrol-only transfected cells (n = 3), whilst Ccl2 was only significantly different in the presence of pC93 mutant SOD1 (P < 0.0001),compared with vector-only transfected cells, despite a decrease in pC37 cells. (B) The increase in Veg f expression was not verified,although significant decreases were seen in B2m (P = 0.0003 for pC37 and P < 0.0001 for pC93, compared with vector control) and Hspa1b(P < 0.0001 for pC37 and P < 0.0001 for pC93, compared with vector control). (C) Bag3 expression was only significantly increased in pC37transfected cells (P = 0.03) compared with vector only, whilst Bnip3 showed significant decreases in the two mutant cell lines(P = 0.04 for pC37 and P = 0.03 for pC93) compared with pCEP, but these were not significantly different in pCN. Gadd45a showed asignificant decrease in the presence of pC37 and pC93 (P = 0.0004 and P = 0.0006, respectively) compared with pCEP. (D) In the presenceof the G37R and G93A mutant SOD1, significant decreases in gene expression were detected for Rgs2 (P < 0.0001 for both pC37 andpC93), Ddc (P < 0.0001 for both pC37 and pC93) and Scg2 (P = 0.0016 for pC37 and P < 0.0001 for pC93). Gsn was decreased significantlyin the presence of pC93 mutant SOD1 (P = 0.0084) compared with vector-only transfected cells. The changes in S100a6, Idb2, B2m,Hspa1b, Gadd45a, Rgs2, Ddc and Scg2 in the mutant cell lines were also significantly different to those cells containing pCN (n = 3).*P < 0.05, **P < 0.01, ***P < 0.001 in mutant cell lines compared with vector-only transfected cells. Q-PCR primer sequences (and finalconcentrations) were S100a6 F 50 GAG CTG AAG GAG TTG ATC CAG AA 30 (100 nM), R 50 CAT CCA TCA GCC TTG CAATTT 30 (100 nM); Idb2 F 50 CCA GGA GGA CCC AGT ATT CG 30 (300 nM), R 50 GCA TTC AGT AGG CTC GTG TCA A 30 (900 nM);Ccl2 F 50 TGA TCC CCC AGC TGT GGT AT 30 (300 nM), R 50 TGA ACC CAC GTT TTG TTA GTT GA 30 (100 nM); Veg f F 50

TGG AGG CAG CCA TGA CTG A 30 (100 nM), R 50 CTG CTT GTT TTC GGT ATT AAG ACA CT 30 (100 nM); B2m F 50 CAT ACGCCT GCA GAG TTA AGC A 30 (300 nM), R 50 GAT CAC ATG TCT CGA TCC CAG TAG 30 (900 nM); Hspa1b F 50 GGG TTCGCT AGA GAG TAC GGA TT 30 (300 nM), R 50 CAC AGG GAC CCC CGA AGT TG 30 (300 nM); Bag3 F 50 CAG CCC ATG ACC CATCGA 30 (100 nM), R 50 CCT GGC TTA CTT TCT GGT TTG TTT 30 (100 nM); Bnip3 F 50 CGA AGT AGC TCC AAG AGT TCTCAC T 30 (100 nM), R 50 CTA TTT CAG CTC TGT TGG TAT CTT GTG 30 (100 nM); Gadd45a F 50 TCA GCA AGG CTC GGAGTC A 30 (100 nM), R 50 CAG CAG GCA CAG TAC CAC GTT 30 (100 nM); Rgs2 F 50 AAA AGC AAA CAG CAA ACT TTTATC AA 30 (200 nM), R 50 TTT AAA AAC GCC CTG AAT GCA 30 (200 nM); Ddc F 50 AGT CAC CAG GAC TCA GGA TTC ATC 30

(200 nM), R 50 CCG TAC ATT CTA AAA ACA AAC CAC AT 30 (200 nM); Scg2 F 50 GAC CGT CCA GAC ATG TTT CAAAG 30 (900 nM), R 50 TCA GGC AAG GCC TCT ACC AT 30 (300 nM); Gsn F 50 GCC CAT CCT CCT CGA CTC TT 30 (300 nM),R 50 CAT AGG CTC GCC AGG AAC CT 30 (300 nM).


Our findings are supported by earlier studies which

demonstrated that neurons, but not astrocytes, show mito-

chondrial dysfunction, glutathione oxidation, decreased

NADPH levels and cell death, following glucose deprivation,

an effect mediated by the superoxide ion (Almeida et al.,

2002). These authors suggest that NADPH generated by

the pentose-phosphate pathway prevents oxidative and

mitochondrial damage during oxidative stress specifically

in neuronal cells. Therefore, dysregulation of Nrf2 coupled

with reduced pentose-phosphate activity and decreased

generation of NADPH, may represent major and hitherto

unrecognized components of the toxic gain of function of

mutant SOD1.

Genes implicated in ALS andneurodegenerationOne of the most interesting changes in a gene implicated

in ALS was the increase in Vegf expression, and although

this was not confirmed by Q-PCR in the cellular model,

the expression levels of the human gene were investigated

during our preliminary studies of human SOD1 cases using

semi-quantitative RT–PCR. These data suggested an increase

is also present in the isolated human motor neurons from the

two SOD1 cases, compared with neurologically normal con-

trols. VEGF is neuroprotective in cultured primary neurons

(Oosthuyse et al., 2001), and therefore, increases of Vegf may

represent a neuroprotective cellular response. This effect of

Fig. 6 Semi-quantitative RT–PCR results using isolated motorneurons from two neurologically normal cases and two familialI113T SOD1-associated ALS cases. Results show a decrease inhuman NRF2 gene expression in the FALS cases (n = 7), anincrease in VEGF expression in the FALS cases (n = 11) and adecrease in B2M (n = 14). SCG2 (n = 9) shows no change. Semi-quantitative RT–PCR primers were NRF2 F 50 CCC CTG TTGATT TAG ACG GTA TG 30, R 50 AAG ACA CTG TAA CTCAGG AAT GGA TAA TAG 30; VEGF F 50 GCC GAC TGA GGAGTC CAA CA 30, R 50 TGT TGG TCT GCA TTC ACA TTT G 30;B2M F 50 GTG ACT TTG TCA CAG CCC AAG ATA 30, R 50

AAT GCG GCA TCT TCA AAC CT 30; SCG2 F 50 CCT CCCACC CCA AGC AA 30, R 50 CAA GAT AAC AGC TCA GAGGAA ATG AA 30; ACTB F 50 GAG CTA CGA GCT GCC TGA CG30, R 50 GTA GTT TCG TGG ATG CCA CAG 30.

A

B

Fig. 7 Pharmacological manipulation of ARE-driven geneexpression. (A) NSC34 cells show decreased GST and G6PDactivity in the presence of mutant SOD1. Addition of 10 mMPDTC for 24 h significantly increases the activity of both GST inpC93 (P = 0.005) and pCEP (P = 0.042), and G6PD in pC93(P = 0.04) and pCEP (P = 0.016) (n = 3 for GST, n = 4 for G6PD).Specific activity measured relative to untreated pCEP. *P < 0.05,**P < 0.01. (B) PDTC partly protects pC93 cells and pCEP against48-h serum withdrawal in an MTT assay. Cells maintained inserum are defined as 100% and results are expressed as apercentage of this (data are mean 6 SEM from five experiments,each with three wells; *P = 0.04, **P = 0.002).

Fig. 8 Diagram illustrating the regulation of Nrf2 and its influence,whether directly or indirectly, on the transcription of other genes.Arrows signify positive regulation whilst lines signify negativeregulation. Genes in green boxes show decreased expression inthe presence of mutant G93A SOD1, whilst those in yellow boxesare increased.


mutant SOD1 on Vegf was recently demonstrated in the G93A

SOD1 transgenic mice. The basal level of Vegf expression was

increased in the mutant SOD1 mice compared with litter

mates, but following hypoxic stress, there was impaired

Vegf up-regulation (Murakami et al., 2003).

Protein degradationThe 20s proteasome is the catalytic core of the proteasome

complex and associates with regulatory proteins that func-

tion as proteasome activators. Activation of the ubiquitin-

independent pathway is completed by the association of

the PA28 complex, which is thought to be the mechanism

by which oxidized proteins are degraded (Grune et al., 1997).

However, heavily oxidized proteins become extensively

crosslinked and aggregate such that they are poor substrates

for degradation. In this report we describe a decrease in

the expression of both PA28 activator subunits, PA200 and

LMP7. We have previously described functional alterations

in the proteasome activities in this cell model of ALS

(Allen et al., 2003). These changes may impair the motor

neuron’s ability to remove oxidized proteins and may con-

tribute to the formation of abnormal intracellular protein

aggregates.

Genes associated with immunityAlthough expression of MHC class I genes in the various

subgroups of neurons in the healthy CNS is either absent

or very low (Mucke and Oldstone, 1992), motor neurons

of the spinal cord and brainstem exhibit significant expression

of MHC class I (Linda et al., 1998). Cell surface expression of

antigen presenting class I MHC is dependent upon the non-

covalent association of B2M with the heavy chain and there

are decreases in expression of the B2m and MHC class 1 heavy

chain molecules in the presence of mutant SOD1. This cor-

responds with a previous study where motor neurons from

ALS patients did not display any MHC class I or B2M immun-

oreactivity (Lampson et al., 1990). Our RT–PCR experiments

suggest this may be due to reduced gene expression levels,

as in both mutant SOD1-transfected cell lines, there was

a significant decrease in B2m expression compared with

controls, and the preliminary studies in isolated motor

neurons also showed decreased B2M expression in the

SOD1-associated cases.

In addition to the role LMP7 and PA28a and -b subunit

may play in forming protein aggregates, they are also involved

in antigen presentation (Preckel et al., 1999). Previously, it

was proposed that the decrease in LMP7 was a neuroprotect-

ive response, reducing the antigen presentation function of

the proteasome (Allen et al., 2003). However, in the light of

our current findings, the down-regulation of the immuno-

proteasome subunits may be part of a broader survival strat-

egy to reduce the repertoire of MHC-restricted peptides,

which could potentially increase both in quantity and variety

in cells that are challenged by oxidative stress.

Apoptotic, cell death and cell survivalproteinsA body of evidence indicates that apoptosis plays a role in

motor neuron degeneration, and previous studies in the

NSC34 cellular model have shown characteristic mitochon-

drial swelling (Menzies et al., 2002a), cell surface annexin V

binding (Cookson et al., 2002) and activation of several

caspase proteins (Sathasivam et al., 2004). In the presence

of mutant G93A SOD1, we identified an increase in the pro-

apoptotic Bnip3 and Pdcd6ip and a decrease in the anti-

apoptotic Bag3 genes, although these were not confirmed

by Q-PCR. However, although relatively few genes encoding

apoptosis regulating protein were altered in our cellular

model, it is noteworthy that many of the molecular effectors

of apoptosis are regulated by alteration in subcellular local-

ization or by cleavage of a precursor protein, rather than by

alterations in gene expression levels.

HspA1B and HspA4, members of the HSP70 multigene

family, which are expressed in response to heat shock, oxid-

ative free radicals and toxic metal ions, were both decreased in

the presence of G93A mutant SOD1. HSPs are sequestered by

mutant SOD1 in the characteristic protein aggregates seen in

both the motor neurons of human ALS cases and transgenic

mice models of ALS (Okado-Matsumoto and Fridovich,

2002), and overexpression of Hsp70 in a cultured neuronal

cell model reduced both aggregate formation and cell death

(Takeuchi et al., 2002). Thus, reduced expression of two key

Hsp70 protein family members is likely to be detrimental to

the ability of the cell to refold and/or eliminate abnormal

proteins.

Comparison of microarray studiesIn contrast to the microarray studies reported previously

using whole spinal cord extracts, the majority of altered

genes identified in our motor neuronal cell model were

decreased. Since several of the increases in gene expression

identified previously were due to reactive gliosis, which occurs

in the spinal cord during neurodegeneration, it is suggested

that the presence of non-neuronal cells dilutes and potentially

masks changes occuring in motor neurons. Comparisons

between studies are difficult given variability in starting

material and array formats. The cellular model has little bio-

logical variation, and does not possess CNS supporting cells,

whilst whole spinal cord from transgenic mice show little

biological variation, but do possess heterogeneous cell types

and human spinal cord sections possess both biological vari-

ation and heterogeneous cell types. In array studies using

spinal cord from the G93A SOD1 transgenic mice, one

used GeneChip microarrays and the other cDNA membrane

arrays, resulting in only four common gene changes (Olsen

et al., 2001; Yoshihara et al., 2002). In our previous study

using cDNA membrane arrays, five genes were identified as

differentially expressed (Kirby et al., 2002). However, in this

study, none of those genes were identified and subsequent

Q-PCR experiments support the current microarray data that


they are not differentially expressed in this set of transfected

cells. With this in mind, it highlights the importance of val-

idating data obtained by microarray analysis, not only at the

RNA and protein level, but also in additional samples to those

used to conduct the experiment. In the previous experiment,

the RNA template for hybridization to the array and for

Q-PCR was from a single source, whereas in the current study

the Q-PCR experiments were performed on three further

separate samples, distinct from the three used for the original

microarray experiments. Discrepancies could also be attrib-

utable to the array formats, as highlighted above, as the macro-

array sample gene expression levels by hybridization to a

duplicate spotted cDNA, whereas the GeneChip samples

gene expression by hybridization to 16 probe pairs. Each probe

pair is a 25mer oligo consisting of a perfect match probe and a

mismatch probe, containing a single nucleotide substitution.

The mismatch probes serve as specificity controls for their

perfect match probe, and the levels of non-specific hybrid-

ization are taken into account by the software during analysis

of the data. The GeneChip protocol is also more consistent in

stringency washes between the three hybridizations, owing to

use of the fluidics station, compared with manual washes. In

addition, when using a model of a disease, changes should be

correlated back to the human disease. Our preliminary find-

ings suggest the changes identified in NSC34 cells are relevant

to human SOD1-related motor neuron degeneration. How-

ever, these studies, in collaboration with other brain tissue

bank centres, need to be extended in a larger sample size, with

different SOD1 mutations, to determine the true relevance to

the human disease of the results from the cellular model.

In summary, gene expression profiling in our cellular

model of motor neuron degeneration has identified key cel-

lular pathways specifically altered in the presence of mutant

SOD1. To our knowledge, this is the first microarray analysis

that has attempted to dissect changes in cells with a motor

neuron phenotype uncontaminated by the multiple other

cellular groups present in spinal cord. We have verified key

differentially expressed genes in the cellular model by Q-PCR

in both the G37R and G93A SOD1 transfected NSC34 cell

lines, and preliminary studies suggest these changes are also

present in motor neurons microdissected from human SOD1-

associated ALS spinal cord. Previous work from our group

and others indicates that motor neurons in the presence of

mutant SOD1 are under stress, with abnormal mitochondria,

cytoskeletal dysregulation, proteasomal dysfunction,

increased sensitivity to oxidative, glutamatergic and nitric

oxide insults, as well as basal activation of early components

of the apoptotic cascade (Eggett et al., 2000; Cookson et al.,

2002; Menzies et al., 2002a, b; Allen et al., 2003; Sathasivam

et al., 2004). We have now added significantly to the existing

knowledge of the toxic cellular effects of mutant SOD1 by

comparative gene expression profiling. Our data suggest that

failure to mount a ‘programmed cell life’ or ARE-driven

response may play an important role in mutant SOD1-

induced motor neuron injury. The pathways showing dys-

regulated gene expression in the presence of mutant SOD1 are

potentially amenable to therapeutic manipulation and may

lead to new treatment approaches for human ALS.

AcknowledgementsWe wish to thank Neil Cashman for supplying the original

NSC34 cell line and Denise Figlewicz for the SOD1 constructs.

J.K. and P.R.H. are funded by the Motor Neurone Disease

Association, S.A., M.J.B., S.C.B., C.A.W.-A. and C.A.L. are

funded by the Wellcome Trust, and P.J.S. is supported

by both. E.H. and J.L. would like to acknowledge the Food

Standards Agency and the Arthritis Research Campaign.

Note added in proofRecently, Jiang and colleagues have published a report detail-

ing the gene expression profile of spinal motor neurons in

sporadic amyotrophic lateral sclerosis (Jiang et al., 2005).

They compared the gene expression profiles of isolated

motor neurons with whole spinal ventral horn. Interestingly,

their study showed evidence of transcriptional repression in

the gene expression profile of motor neurons. Comparison of

the 60 reported genes altered in sporadic ALS motor neurons

with our cellular model provided only a few correlates, and

this is perhaps not surprising as these are sporadic ALS cases,

rather than SOD1-related motor neuron disease, the changes

reflect those at end stage of the disease, the genes expressed

have been influenced by the presence of astrocytes and other

surrounding CNS tissue, and an alternative array platform

(glass microarrays with Cy3- and Cy5-labelled probes) has

been used.

References

Abate C, Patel L, Rauscher FJ 3rd, Curran T. Redox regulation of fos and jun

DNA-binding activity in vitro. Science 1990; 249: 1157–61.

Allen S, Heath PR, Kirby J, Wharton SB, Cookson MR, Menzies FM, et al.

Analysis of the cytosolic proteome in a cell culture model of familial

amyotrophic lateral sclerosis reveals alterations to the proteasome, anti-

oxidant defenses, and nitric oxide synthetic pathways. J Biol Chem 2003;

278: 6371–83.

Almeida A, Delgado-Esteban M, Bolanos JP, Medina JM. Oxygen and glucose

deprivation induces mitochondrial dysfunction and oxidative stress in

neurones but not in astrocytes in primary culture. J Neurochem 2002;

81: 207–17.

Cashman NR, Durham HD, Blusztajn JK, Oda K, Tabira T, Shaw IT, et al.

Neuroblastoma x spinal cord (NSC) hybrid cell lines resemble developing

motor neurons. Dev Dyn 1992; 194: 209–21.

Chan K, Lu R, Chang JC, Kan YW. NRF2, a member of the NFE2 family of

transcription factors, is not essential for murine erythropoiesis, growth, and

development. Proc Natl Acad Sci USA 1996; 93: 13943–8.

Cookson MR, Shaw PJ. Oxidative stress and motor neurone disease. Brain

Pathol 1999; 9: 165–86.

Cookson MR, Ince PG, Shaw PJ. Peroxynitrite and hydrogen peroxide

induced cell death in the NSC34 neuroblastoma x spinal cord cell

line: role of poly (ADP-ribose) polymerase. J Neurochem 1998; 70:

501–8.

Cookson MR, Menzies FM, Manning P, Eggett CJ, Figlewicz DA, McNeil CJ,

et al. Cu/Zn superoxide dismutase (SOD1) mutations associated with

familial amyotrophic lateral sclerosis (ALS) affect cellular free radical


release in the presence of oxidative stress. Amyotroph Lateral Scler Other

Motor Neuron Disord 2002; 3: 75–85.

De Cesare D, Vallone D, Caracciolo A, Sassone-Corsi P, Nerlov C, Verde P.

Heterodimerization of c-Jun with ATF-2 and c-Fos is required for positive

and negative regulation of the human urokinase enhancer. Oncogene 1995;

11: 365–76.

Durham HD, Dahrouge S, Cashman NR. Evaluation of the spinal cord

neuron · neuroblastoma hybrid cell line NSC-34 as a model for neuro-

toxicity testing. Neurotoxicology 1993; 14: 387–95.

Eggett CJ, Crosier S, Manning P, Cookson MR, Menzies FM, McNeil CJ, et al.

Development and characterisation of a glutamate-sensitive motor neurone

cell line. J Neurochem 2000; 74: 1895–902.

Filosa S, Fico A, Paglialunga F, Balestrieri M, Crooke A, Verde P, et al. Failure

to increase glucose consumption through the pentose-phosphate pathway

results in the death of glucose-6-phosphate dehydrogenase gene-deleted

mouse embryonic stem cells subjected to oxidative stress. Biochem J 2003;

370: 935–43.

Grune T, Reinheckel T, Davies KJ. Degradation of oxidized proteins in mam-

malian cells. FASEB J 1997; 11: 526–34.

Hand CK, Rouleau GA. Familial amyotrophic lateral sclerosis. Muscle Nerve

2002; 25: 135–59.

Heath PR, Shaw PJ. Update on the glutamatergic neurotransmitter system

and the role of excitotoxicity in amyotrophic lateral sclerosis. Muscle Nerve

2002; 26: 438–58.

Heath PR, Tomkins J, Ince PG, Shaw PJ. Quantitative assessment of AMPA

receptor mRNA in human spinal motor neurons isolated by laser capture

microdissection. Neuroreport 2002; 13: 1753–7.

Hoyaux D, Alao J, Fuchs J, Kiss R, Keller B, Heizmann CW, et al. S100A6, a

calcium- and zinc-binding protein, is overexpressed in SOD1 mutant mice,

a model for amyotrophic lateral sclerosis. Biochim Biophys Acta 2000;

1498: 264–72.

Hsu RY, Lardy HA. Malic enzyme. Methods Enzymol 1969; 13: 230–5.

Jiang YM, Yamamoto M, Kobayashi Y, Yoshihara T, Liang Y, Terao S, et al.

Gene expression profile of spinal motor neurons in sporadic amyotrophic

lateral sclerosis. Ann Neurol 2005; 57: 236–51.

Kirby J, Menzies FM, Cookson MR, Bushby K, Shaw PJ. Differential gene

expression in a cell culture model of SOD1-related familial motor neurone

disease. Hum Mol Genet 2002; 11: 2061–75.

Kwak MK, Wakabayashi N, Itoh K, Motohashi H, Yamamoto M, Kensler TW.

Modulation of gene expression by cancer chemopreventive dithiolethiones

through the Keap1–Nrf2 pathway. Identification of novel gene clusters for

cell survival. J Biol Chem 2003; 278: 8135–45.

Lampson LA, Kushner PD, Sobel RA. Major histocompatibility complex

antigen expression in the affected tissues in amyotrophic lateral sclerosis.

Ann Neurol 1990; 28: 365–72.

Lee CY. Glucose-6-phosphate dehydrogenase from mouse. Methods Enzymol

1982; 89 Pt D: 252–7.

Lee JM, Calkins MJ, Chan K, Kan YW, Johnson JA. Identification of the NF-

E2-related factor-2-dependent genes conferring protection against oxidat-

ive stress in primary cortical astrocytes using oligonucleotide microarray

analysis. J Biol Chem 2003a; 278: 12029–38.

Lee JM, Shih AY, Murphy TH, Johnson JA. NF-E2-related factor-2 mediates

neuroprotection against mitochondrial complex I inhibitors and increased

concentrations of intracellular calcium in primary cortical neurons. J Biol

Chem 2003b; 278: 37948–56.

Lefebvre S, Burglen L, Reboullet S, Clermont O, Burlet P, Viollet L,

et al. Identification and characterization of a spinal muscular atrophy-

determining gene. Cell 1995; 80: 155–65.

Li C, Hung Wong W. Model-based analysis of oligonucleotide arrays: model

validation, design issues and standard error application. Genome Biol 2001;

2: RESEARCH0032.

Li J, Lee JM, Johnson JA. Microarray analysis reveals an antioxidant respons-

ive element-driven gene set involved in conferring protection from an

oxidative stress-induced apoptosis in IMR-32 cells. J Biol Chem 2002;

277: 388–94.

Linda H, Hammarberg H, Cullheim S, Levinovitz A, Khademi M, Olsson T.

Expression of MHC class I and beta2-microglobulin in rat spinal

motoneurons: regulatory influences by IFN-gamma and axotomy. Exp

Neurol 1998; 150: 282–95.

Lino MM, Schneider C, Caroni P. Accumulation of SOD1 mutants in

postnatal motoneurons does not cause motoneuron pathology or

motoneuron disease. J Neurosci 2002; 22: 4825–32.

Malaspina A, Kaushik N, de Belleroche J. Differential expression of 14 genes

in amyotrophic lateral sclerosis spinal cord detected using gridded cDNA

arrays. J Neurochem 2001; 77: 132–45.

Martin LJ. Neuronal death in amyotrophic lateral sclerosis is apoptosis: pos-

sible contribution of a programmed cell death mechanism. J Neuropathol

Exp Neurol 1999; 58: 459–71.

Menzies FM, Cookson MR, Taylor RW, Turnbull DM, Chrzanowska-

Lightowlers ZM, Dong L, et al. Mitochondrial dysfunction in a cell

culture model of familial amyotrophic lateral sclerosis. Brain 2002a;

125: 1522–33.

Menzies FM, Grierson AJ, Cookson MR, Heath PR, Tomkins J, Figlewicz DA,

et al. Selective loss of neurofilament expression in Cu/Zn superoxide dis-

mutase (SOD1) linked amyotrophic lateral sclerosis. J Neurochem 2002b;

82: 1118–28.

Meyer M, Schreck R, Baeuerle PA. H2O2 and antioxidants have

opposite effects on activation of NF-kappa B and AP-1 in intact cells:

AP-1 as secondary antioxidant-responsive factor. EMBO J 1993; 12:

2005–15.

Mucke L, Oldstone MB. The expression of major histocompatibility complex

(MHC) class I antigens in the brain differs markedly in acute and persistent

infections with lymphocytic choriomeningitis virus (LCMV). J Neuro-

immunol 1992; 36: 193–8.

Murakami T, Ilieva H, Shiote M, Nagata T, Nagano I, Shoji M, et al.

Hypoxic induction of vascular endothelial growth factor is selectively

impaired in mice carrying the mutant SOD1 gene. Brain Res 2003; 989:

231–7.

Okado-Matsumoto A, Fridovich I. Amyotrophic lateral sclerosis: a proposed

mechanism. Proc Natl Acad Sci USA 2002; 99: 9010–4.

Olsen MK, Roberds SL, Ellerbrock BR, Fleck TJ, McKinley DK,

Gurney ME. Disease mechanisms revealed by transcription profiling

in SOD1-G93A transgenic mouse spinal cord. Ann Neurol 2001; 50:

730–40.

Oosthuyse B, Moons L, Storkebaum E, Beck H, Nuyens D, Brusselmans K,

et al. Deletion of the hypoxia-response element in the vascular endothelial

growth factor promoter causes motor neuron degeneration. Nat Genet

2001; 28: 131–8.

Owuor ED, Kong AN. Antioxidants and oxidants regulated signal transduc-

tion pathways. Biochem Pharmacol 2002; 64: 765–70.

Phul RK, Shaw PJ, Ince PG, Smith ME. Expression of nitric oxide synthase

isoforms in spinal cord in amyotrophic lateral sclerosis. Amyotroph Lateral

Scler Other Motor Neuron Disord 2000; 1: 259–67.

Pramatarova A, Laganiere J, Roussel J, Brisebois K, Rouleau GA.

Neuron-specific expression of mutant superoxide dismutase 1 in transgenic

mice does not lead to motor impairment. J Neurosci 2001; 21: 3369–74.

Preckel T, Fung-Leung WP, Cai Z, Vitiello A, Salter-Cid L, Winqvist O, et al.

Impaired immunoproteasome assembly and immune responses in

PA28–/– mice. Science 1999; 286: 2162–5.

Rosen DR, Siddique T, Patterson D, Figlewicz DA, Sapp P, Hentati A, et al.

Mutations in Cu/Zn superoxide dismutase gene are associated with familial

amyotrophic lateral sclerosis. Nature 1993; 362: 59–62.

Sathasivam S, Grierson AJ, Shaw PJ. Characterisation of the caspase cascade

in a cell culture model of SOD1-related FALS; expression, activation and

therapeutic effects of inhibition. Neuropathol Appl Neurobiol 2005;

In press.

Takeuchi H, Kobayashi Y, Yoshihara T, Niwa J, Doyu M, Ohtsuka K, et al.

Hsp70 and Hsp40 improve neurite outgrowth and suppress intracytoplas-

mic aggregate formation in cultured neuronal cells expressing mutant

SOD1. Brain Res 2002; 949: 11–22.

Thimmulappa RK, Mai KH, Srisuma S, Kensler TW, Yamamoto M, Biswal S.

Identification of Nrf2-regulated genes induced by the chemopreventive

agent sulforaphane by oligonucleotide microarray. Cancer Res 2002; 62:

5196–203.


Venugopal R, Jaiswal AK. Nrf1 and Nrf2 positively and c-Fos and Fra1

negatively regulate the human antioxidant response element-mediated

expression of NAD(P)H:quinone oxidoreductase1 gene. Proc Natl Acad

Sci USA 1996; 93: 14960–5.

Wild AC, Moinova HR, Mulcahy RT. Regulation of gamma-glutamylcysteine

synthetase subunit gene expression by the transcription factor Nrf2. J Biol

Chem 1999; 274: 33627–36.

Wilkinson JT, Radjendirane V, Pfeiffer GR, Jaiswal AK, Clapper ML.

Disruption of c-Fos leads to increased expression of NAD(P)H:quinone

oxidoreductase1 and glutathione S-transferase. Biochem Biophys Res

Commun 1998; 253: 855–8.

Wong PC, Pardo CA, Borchelt DR, Lee MK, Copeland NG, Jenkins NA, et al.

An adverse property of a familial ALS-linked SOD1 mutation causes motor

neuron disease characterized by vacuolar degeneration of mitochondria.

Neuron 1995; 14: 1105–16.

Wonsey DR, Zeller KI, Dang CV. The c-Myc target gene PRDX3 is required

for mitochondrial homeostasis and neoplastic transformation. Proc Natl

Acad Sci USA 2002; 99: 6649–54.

Wood JD, Beaujeux TP, Shaw PJ. Protein aggregation in motor neurone

disorders. Neuropathol Appl Neurobiol 2003; 29: 529–45.

Yoshihara T, Ishigaki S, Yamamoto M, Liang Y, Niwa J, Takeuchi H, et al.

Differential expression of inflammation- and apoptosis-related genes in

spinal cords of a mutant SOD1 transgenic mouse model of familial amyo-

trophic lateral sclerosis. J Neurochem 2002; 80: 158–67.

Zerez CR, Lee SJ, Tanaka KR. Spectrophotometric determination of oxidized

and reduced pyridine nucleotides in erythrocytes using a single extraction

procedure. Anal Biochem 1987; 164: 367–73.

Zuccato C, Tartari M, Crotti A, Goffredo D, Valenza M, Conti L, et al.

Huntingtin interacts with REST/NRSF to modulate the transcription of

NRSE-controlled neuronal genes. Nat Genet 2003; 35: 76–83.


Mutant SOD1 alters the motor neuronal transcriptome ... · Janine Kirby,1 Eugene Halligan,2 Melisa J. Baptista,1 Simon Allen,1 Paul R. Heath,1 Hazel Holden,1 Sian C. Barber,1 Catherine

Documents