Human Molecular Genetics, 2002, Vol. 11, No. 17 2061-2075
© 2002 Oxford University Press
Differential gene expression in a cell culture model of SOD1-related familial motor neurone disease
1Academic Neurology Unit, University of Sheffield, School of Medicine and Biomedical Sciences, Beech Hill Road, Sheffield S10 2RX, UK, 2Laboratory of Neurogenetics, National Institute on Aging, NIH, 9000 Rockville Pike, Bethesda, MD 20892, USA and 3Institute of Human Genetics, International Centre for Life, Central Parkway, Newcastle-upon-Tyne NE1 3BZ, UK
Received May 29, 2002; Accepted June 14, 2002
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Motor neurone disease is caused by mutations in Cu/Zn superoxide dismutase (SOD1) in 15–20% of familial cases, due to a toxic gain of function by the mutant enzyme. However, the underlying mechanism of SOD1-mediated neurodegeneration remains uncertain. By investigating alterations in gene expression in the presence of mutant Cu/Zn SOD, we aimed to identify pathways that contribute to motor neurone injury and cell death. Using a cellular model of familial motor neurone disease, the motor neuronal cell line NSC34 was stably transfected with either normal or mutant (G37R, G93A, I113T) SOD1 cDNAs, and the effect of the presence of these proteins on gene expression was analysed. This model allowed gene expression changes to be studied specifically in cells with a motor neurone phenotype, without interference from genes expressed by glia, astrocytes and other cell types located in the central nervous system. Using a commercially available cDNA membrane array, we investigated the expression levels of 588 genes from key biological pathways. Gene expression was studied in the cells under both basal culture conditions and following oxidative stress induced by serum withdrawal. Twenty-nine differentially expressed genes were identified, 7 of which were specifically downregulated in the presence of the mutant Cu/Zn SOD protein, and whose expression was further studied by real-time PCR. Presence of the mutant Cu/Zn SOD was confirmed to lead to a decrease in expression of KIF3B, a kinesin-like protein, which forms part of the KIF3 molecular motor. c-Fes, thought to be involved in intracellular vesicle transport was also decreased, further implicating the involvement of vesicular trafficking as a mode of action for mutant Cu/Zn SOD. In addition, a decrease was confirmed in ICAM1, a response in part due to the increased expression of SOD1, and decreased Bag1 expression was confirmed in two of the three mutant cell lines, providing further support for the involvement of apoptosis in SOD1-associated motor neurone death.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Motor neurone disease (MND) is an adult-onset neurodegenerative disorder characterized by progressive degeneration of the upper motor neurones in the motor cortex and lower motor neurones in the spinal cord and brainstem. Although the majority of cases are sporadic, 5–10% are familial, with an autosomal dominant mode of inheritance. Sporadic and familial cases are, however, clinically indistinguishable, suggesting similar underlying pathophysiological mechanisms of neurodegeneration. Over 90 mutations have been identified throughout the five exons encoding the Cu/Zn superoxide dismutase (SOD) gene (SOD1), and these account for 15–20% of familial cases (1).
The mechanism(s) by which mutant SOD1 causes selective motor neurone cell death has still not been clearly identified. However, a body of evidence indicates that the detrimental effect results from a toxic gain of function rather than a loss of normal SOD activity. Transgenic mice expressing mutant human SOD1, as well as the endogenous murine SOD1, develop a progressive motor neurone degeneration phenotype (2,3), whereas SOD1 knockout mice do not (4). Measurements of activity of the mutant Cu/Zn SOD proteins show that some retain full SOD activity (5), and no correlation between activity level and progression or duration of the disease has been identified (6). Key hypotheses for the toxic gain of function include aberrant free-radical handling/metabolism resulting in increased production of hydroxyl radicals (7), and/or toxic derivatives of peroxynitrite (8), metal toxicity (9) and abnormal protein aggregation (10,11).
A potentially important way of identifying the mechanisms of motor neurone injury would be to identify cell-specific changes in gene expression occurring in the presence of the mutant Cu/Zn SOD protein. To increase our understanding of the molecular mechanisms of SOD1-mediated neurodegeneration, we investigated changes in gene expression induced by mutant Cu/Zn SOD protein in a cell culture model of familial MND.
NSC34 cells are a hybrid mouse motor neurone/neuroblastoma cell line that retains the ability to proliferate whilst exhibiting many motor neurone characteristics without the addition of inducing agents. These characteristics include extension of neurites, generation of action potentials, expression of neurofilament proteins and choline acetyltransferase, synthesis and storage of acetylcholine, and induction of twitching in co-cultured muscle cells (12–14). These provide an ideal cellular model to investigate mechanisms of neurodegeneration associated with mutant Cu/Zn SOD specifically in cells with a motor neurone phenotype, without the dilution effect created by astrocytes, microglia, and other cell types found in the brain and spinal cord. The use of this immortilized cell line for such studies has been verified by increasing evidence corroborating findings from these cells with those in the SOD1 transgenic mice models and human MND cases. Studies investigating mitochondrial dysfunction in the presence of mutant G93A SOD1 in the NSC34 (15) cells showed abnormally swollen mitochondria similar to those seen in the G93A SOD1 transgenic mice, whilst structural changes in the mitochondria of human post-mortem central nervous system tissue have also been reported (16,17). These morphological changes were accompanied by decreases in the activities of complex II and IV of the mitochondrial electron transport chain (15); decreases in complex IV have been described previously in human spinal cord neurones (18,19). Evidence of an apoptotic element to the motor neurone cell death seen in MND (20) is also found in the cellular model. Annexin V binding, an early indicator of the apoptotic pathway, and activation of caspase 9 were identified in cells expressing the mutant Cu/Zn SOD cultured under basal conditions (21). Following oxidative stress induced by serum withdrawal, activation of the downsteam caspase 3 was significantly increased in the presence of the mutant Cu/Zn SOD. Most recently, the levels of neurofilament light mRNA and protein were found to be decreased in the NSC34 cells containing a G37R and G93A mutant Cu/Zn SOD. This finding was reproduced both in the G93A SOD1 transgenic mice and in familial SOD1-associated MND cases (22).
In this study, NSC34 cells were transfected with vector, normal human SOD1 or one of three mutant human SOD1 cDNAs (encoding the following amino acid substitutions: G37R, G93A and I113T). As well as reflecting mutations associated with a usual progression of MND, the use of G93A Cu/Zn SOD would allow data to be compared with the G93A SOD1 transgenic mouse colony, whilst the I113T Cu/Zn SOD results could be compared with human pathological tissue carrying the I113T SOD1 mutation in the Sheffield Brain Tissue Bank. Gene expression was investigated in these cells both under basal culture conditions, to analyse the response of the cells to the presence of the mutant protein, and following oxidative stress, induced by serum withdrawal. Previous work has shown that NSC34 cells produce increased levels of both nitric oxide and superoxide immediately upon serum withdrawal (23).
The use of cDNA array analysis to investigate changes in gene expression has been widely used in cancer research (24–26), and more recently in neurological disorders, such as Parkinson's disease, Huntington's disease and Alzheimer's disease (27–29). This technique enables the expression levels of a large number of genes to be determined in a single hybridization experiment, using cDNA probes derived from an RNA sample of interest. These expression levels can then be compared with other RNA samples following normalization. cDNA array analysis has identified patterns of gene expression that distinguish between subgroups of disease not previously identified, such as two distinct groups within the clinically heterogeneous population of patients with diffuse large B-cell lymphoma (25), as well as identifying potential targets for therapy (30,31).
We used commercially available cDNA membrane arrays, spotted with 588 genes from key biological pathways, to obtain expression profiles from each cell line under basal conditions and following oxidative stress. By identifying the alterations in gene expression patterns induced by the presence of SOD1 mutations, we aimed to identify molecular pathways leading to motor neurone injury in familial MND.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Expression of human Cu/Zn SOD in the NSC34 single-cell clones
To confirm protein expression from the human SOD1 cDNAs in the single-cell clones, western blotting was carried out using an anti-human Cu/Zn SOD antibody, which also crossreacts with murine Cu/Zn SOD. As shown in Figure 1, those cells transfected with human SOD1—either normal (pCN) or mutant (pC37, pC93A, pC113)—are clearly expressing the human Cu/Zn SOD protein. The lower band identified in all cell lines is the endogenous mouse Cu/Zn SOD. The levels of expression are 20–60% of the endogenous murine Cu/Zn SOD. This is similar to the physiological situation found in humans, as compared with the SOD1 transgenic mice, where overexpression of human Cu/Zn SOD is much higher than endogenous murine Cu/Zn SOD (32).
|
PCR of mRNA using genomic DNA primers
Total RNA was DNase-treated, and the mRNA was then extracted using an oligo dT column (NucleoTrap mRNA Purification Kit, BD Clontech). However, to check that the mRNA used for the array analysis did not contain any contaminating genomic DNA, PCR containing 1 µl of mRNA and primers designed to amplify exon 4 of SOD1 (33) was carried out. No amplification of genomic DNA was obtained from any of the mRNA samples (data not shown).
Identification of differentially expressed genes under basal conditions
Under basal conditions, the expression of G37R, G93A and I113T mutant Cu/Zn SOD had no effect on the viability of the cells, although growth rates were observed to be slower than in cells transfected with vector only or normal SOD1. To identify changes in gene expression, the Atlas cDNA array membranes were hybridized with cDNA derived from the two control cell lines [vector only (pCEP4) and normal SOD1 (pCN)] and three mutant SOD1 cell lines (pC37, pC93A and pC113, encoding the G37R, G93A and I113T mutations). Following normalization of the hybridization levels with the housekeeping genes GAPDH and ß-actin, pairwise comparisons were conducted using the AtlasImage software (BD Clontech). An example of a pairwise comparison between pCN and pC93A under basal conditions is shown in Figure 2. A three-way pairwise comparison for each mutant-expressing cell line was then compiled (e.g. pCEP4 versus pCN; pCEP4 versus pC93A; pCN versus pC93A), and genes that gave a consistent 2-fold difference or more between the three comparisons were identified. This approach was employed to reduce the number of false-positive results and ensure consistent and robust results.
|
Pairwise analysis of filters hybridized with pCEP4-, pCN- and pC93A-transfected cells identified 134 differentially expressed genes, of which 35 were downregulated in the presence of the Cu/Zn SOD G93A mutant (Table 1). No genes were identified as upregulated. Two other mutant SOD1- expressing cell lines (pC113 and pC37) were used for cDNA array analysis, to identify common patterns of gene expression in the presence of different SOD1 mutations. The expression profiles from NSC34 cells transfected with pC113 and pC37 under basal conditions identified 146 and 80 differentially expressed genes respectively (Table 1). Cells transfected with pC113 SOD1 showed 49 genes differentially expressed in the presence of the mutant protein, 24 upregulated and 25 downregulated. Cells transfected with pC37 SOD1 showed 31 genes differentially expressed in the presence of the mutant protein, only 1 upregulated and 30 downregulated.
|
Data were combined from the three mutant SOD1-expressing cell lines, the rationale being that changes consistent between the three sets of data would more likely reflect changes due to the presence of mutant protein. This methodology identified 29 genes that were consistently differentially expressed (Table 2). Seven genes were consistently downregulated in the presence of the mutant Cu/Zn SOD protein. These covered a wide range of pathways, from cytoskeletal components such as kinesin family member 3B (KIF3B) to transcription factors such as neuronal-specific helix–loop–helix protein, NEX1, and the anti-apoptotic Bcl2-associated athanogene 1, Bag1. In addition, 14 genes showed decreased expression in cells transfected with either mutant or normal SOD1. These genes are primarily transcription factors, cell surface antigens and cell adhesion proteins (Table 2). Eight genes were also increased in the presence of normal Cu/Zn SOD only; half of these were involved in DNA synthesis, repair and recombination.
|
Identification of genes differentially expressed following oxidative stress
NSC34 cells transfected with mutant SOD1 cDNAs show a decrease in cell viability following serum withdrawal, compared with those transfected with vector only, whilst cells transfected with normal SOD1 are less sensitive to this form of oxidative stress, demonstrating a protective effect of normal Cu/Zn SOD (15). We therefore investigated changes in gene expression, which occur following oxidative stress, induced by serum withdrawal for 24 hours.
Initially, pairwise comparisons were carried out between each of the cell lines (pCEP4, pCN and pC93A) at basal conditions (+serum) and following serum withdrawal (-serum), and genes that showed a change of 2-fold or more were identified. The expression profiles of these genes in pCEP4, pCN and pC93A under basal conditions (+serum) and following serum withdrawal (-serum) were then compared to give a nine-way analysis (Table 3). As previously, the 2-fold difference had to be consistent between the analyses. This rationale allowed the expression levels of those differentially expressed genes to be compared under basal and serum withdrawal conditions across all the cell lines. This is important, since previous studies (15,21) have found that certain changes that occur in the mutant SOD1 cells under basal conditions occur in the control cells (pCEP and pCN) following serum withdrawal.
|
Nine-way analysis of pCEP4, pCN and pC93A identified 26 genes that were differentially expressed following serum withdrawal. The analyses were then repeated using array data from the cells transfected with pC113 SOD1 under basal conditions and following oxidative stress. This identified 23 differentially expressed genes. Compiling these results, 12 genes were found to be consistently differentially expressed following serum withdrawal, and they cover a wide range of pathways (Table 4). Surprisingly, only one gene was identified that was consistently decreased by at least 2-fold in all four cell lines following serum withdrawal, namely the c-Src proto-oncogene. No genes were found to be specifically up- or downregulated by at least 2-fold in the presence of the mutant Cu/Zn SOD protein. However, a number of genes that had shown a change in expression whilst under basal conditions also showed changes following serum withdrawal. For example, Bag1 expression, which was decreased in the presence of mutant Cu/Zn SOD under basal conditions, showed a decrease in the vector-only and normal SOD1-transfected cells following serum withdrawal.
|
Confirmation of differential expression
Of the 29 genes that were consistently differentially expressed in the presence of the three mutant Cu/Zn SOD proteins, confirmation of expression was carried out for those genes that showed a decrease specifically in the presence of mutant protein. These were thought to be of greatest interest with regard to the potential mechanisms of SOD1-mediated neurodegeneration.
KIF3B.
KIF3B is a component of the ubiquitously expressed KIF3 (kinesin II) molecular motor protein (34). KIF3B forms a heterodimer with KIF3A, which then associates with kinesin superfamily associated protein 3 (KAP3). Quantitative PCR using SYBR green, a fluorescent dye that binds double-stranded DNA, confirmed the differential expression of KIF3B. The three cell lines carrying the mutant SOD1 expressed KIF3B at a significantly lower level than in the cells transfected with either vector or normal SOD1 (P<0.05) (Fig. 3A). Expression of KIF3B in pC93A, however, was higher than in the other two mutants. Neither KIF3A nor KAP3 are represented on the cDNA array.
|
c-Fes.
c-Fes or Fps/Fes is a non-receptor protein tyrosine kinase that is expressed predominantly in myeloid, endothelial, epithelial and neuronal cells (35). It is expressed in the adult brain, in Purkinje cells and neuronal cells in the molecular layer of the cerebellum. Quantitative PCR using SYBR green confirmed the differential expression of c-Fes. All three mutants expressed c-Fes at significantly lower levels than in either of the control cell lines (P<0.01) (Fig. 3B).
ICAM1.
Intracellular adhesion molecule 1 is a ligand for lymphocyte function-associated antigens, which is constitutively expressed by haematopoetic cells, vascular endothelium, fibroblasts and some epithelial cells (36). Amplification of ICAM1 using SYBR green with the primers designed produced the correctly sized product when template was present, but formed primer dimers in the absence of ICAM1. Therefore, a Taqman probe for ICAM1 was used to quantify the level of gene expression, since this provides greater specificity. The level of fluorescence detected reflects the amount of amplified PCR product, rather than the presence of double-stranded DNA. Consequently, a Taqman probe was designed to ß-actin as well to serve as the control. Quantitative PCR showed a decrease of ICAM1 expression in cells transfected with normal SOD1, compared with those transfected with vector only (P<0.001). This decrease had been identified during the cDNA three-way analyses. More importantly, there was a further significant decrease in the expression of ICAM1 in all three mutant transfected cell lines (P<0.01), with no expression detectable in two of the three mutant transfections (Fig. 3C). This correlates with an absence of hybridization signal from the cDNA array membranes.
Bag1.
The Bcl2-associated athanogene 1 (Bag1) interacts with Bcl2 to enhance its anti-apoptotic activity (37) and also interacts with the Hsp70 family of molecular chaperones (38). Quantitative RT–PCR of Bag1 using SYBR green confirmed a decrease in expression in the pC37 and pC113 mutant SOD1-containing cell lines (P<0.05) (Fig. 3D). However, the decrease of Bag1 in cells expressing the pC93A mutant SOD1 could not be confirmed statistically (P>0.05).
MADR2.
The Drosophila mothers against decapentaplegic (MAD)-related gene 2, also known as SMAD2, acts downstream in both transforming growth factor ß (TGF-ß) and activin signalling pathways (39,40). The cDNA array analysis identified a decrease in MADR2 expression in all the cells expressing mutant Cu/Zn SOD. However, quantitative RT–PCR from at least six experiments did not confirm this finding (Fig. 3E).
GCM1.
Glial cells missing 1 (GCM1) is a homologue of a Drosophila protein that is involved in cell fate during neurogenesis/gliogenesis (41,42). However, the mammalian homologue is specifically expressed in placenta, and is involved in the formation of the labyrinth (43,44). Using the primers designed with SYBR green, numerous non-specific bands, in addition to what appeared to be the correctly sized fragment, were amplified, resulting in false levels of GCM1 expression. We therefore, again, used Taqman probes to increase the specificity of the quantitative PCR. However, during optimization of the experiment, no fluorescence was observed following 50 cycles of PCR. This suggests that none of the products previously amplified were specific to GCM1.
NEX1.
NEX1 is a neuronal-specific intronless gene containing a basic helix–loop–helix motif that is thought to play a role in glial/neuronal cell fate (45), similar to that of GCM in Drosophila. Primers designed to the cDNA of NEX1 were not able to amplify any product from the original cell lines using standard PCR conditions, following 50 cycles. Primers for NEX1, which were originally used to amplify the cDNA spotted onto the membrane, were obtained from BD Clontech. Although the correctly sized product was amplified from the mouse cerebellum control, no amplification was seen in the cell lines, following 50 cycles of PCR.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
This study represents the first cDNA array analysis of the effects of mutant SOD1 on gene expression in a cell culture model of familial MND. One of the advantages of this model is that changes that occur in cells with a motor neuronal phenotype can be studied without interference from genes expressed by astrocytes, microglia and other cells in the central nervous system. Although astrocytes may be involved in the pathogenesis of MND, through astrogliosis and dysfunction of astrocytic control of glutamate neurotransmission, it has been demonstrated in glial cell lines that the presence of mutant Cu/Zn SOD does not increase the toxicity of oxidative stress (46). This is in contrast to the evidence from expression of mutant Cu/Zn SOD in neuronal cell lines (47, 48). Therefore, since the site of death is the motor neurone, changes in gene expression induced by the presence of mutant SOD1 in this neuronal cell model should be relevant to SOD1 familial MND. Previous studies investigating alterations in mitochondrial function, expression of neurofilaments and involvement of apoptosis in neuronal cell death using this model have been reproduced in both SOD1 transgenic mouse and human MND cases (15,21,22), emphasizing the relevance of studying this model in relation to human MND.
Genes differentially expressed in the presence of the mutant protein at basal conditions
The three-way analysis performed on each of the mutant G37R, G93A and I113T Cu/Zn SOD-expressing cell lines identified 80, 134 and 146 differentially expressed genes respectively. The data were then combined, since changes consistent between the three sets of data would more likely reflect changes due to the presence of the mutant protein, rather than environmental factors. For example, subtle changes in cell density between the cell lines may contribute to some of the changes seen in only one cell line and not another. Cells containing the I113T Cu/Zn SOD mutant appeared to have a somewhat different transcriptional profile compared with that of the other two mutants, with upregulation of 24 genes. It is of interest that, in the human disease, there are some peculiarities observed in relation to the I113T Cu/Zn SOD mutations, where massive hyaline conglomerate neurofilamentous inclusions are observed in surviving motor neurones (49). It is possible that specific changes in transcriptosome might reflect these particular pathophysiological features.
Combining the data following cDNA array analysis, 29 genes were found to be differentially expressed in our cell lines, of which 7 were downregulated in the specifically presence of a mutant Cu/Zn SOD protein (carrying the amino acid substitutions G37R, G93A and I113T). Quantitative PCR confirmed the decrease in expression of KIF3B, c-Fes and ICAM1 in all three mutants, and the decrease in expression of Bag1 in two of the three mutant SOD1-expressing cell lines.
KIF3B.
This is a kinesin-like protein, which forms a heterodimer with KIF3A, and associates with KAP3 to form the KIF3 (kinesin II) plus-end directed molecular motor (34). It is involved in fast anterograde transport of membranous vesicles of 90–160 nm, carrying constituents required for neurite sprouting (50). A decrease in this protein may therefore affect the transport of vital constituents to the axonal terminal of the neurone. Both KIF3B and KAP3 have been shown to associate with fodrin, a cytoskeletal component that forms a major component of the subaxolemmal cytoskeletal network. It is suggested that fodrin may also mediate between the KIF3 motor and the cargoes, with specificity for KIF3 (51).
How molecular motors recognise their cargoes is still under investigation, although it has been suggested that small G proteins (GTPases) may be involved both in the association of the motors with the cargoes and in the activation of the motors (51). Rab proteins are members of the superfamily of monomeric GTPases, and Rab6 protein shows interaction with a kinesin-like protein termed Rabkinesin-6 (52). Rab proteins may therefore provide binding sites for motor proteins on the surface of organelles.
GTPases act as molecular switches; GDP-bound forms of GTPase are inactive in the cytoplasm. Guanine nucleotide exchange factors (GEFs) associate with the inactive GTPases to accelerate the disassociation of GDP and binding of GTP to activate the GTPase. The gene that causes ALS2, an autosomal recessive form of MND, encodes a putative GTPase regulator, with homology to several GEFs (53,54). Therefore, abnormalities in fast axonal transport are implicated in ALS2, and in our SOD1 model of MND.
Other abnormalities in fast axonal transport have been reported in models of MND. In G86R-SOD1 transgenic mice, using subtractive hybridization, an increased expression of KAP3 was identified in spinal motor neurones during the asymptomatic stage of the disease (55). RT–PCR of KIF3A and KIF3B, however, showed no changes in gene expression in the G86R-SOD1 mice, compared with wild type. It is possible that the alteration in KIF3B is specific to motor neurones, and that extraction of RNA from whole spinal cord homogenates in these mice may have masked the presence of cell-specific alterations in gene expression. Since KIF3A and KAP3 are not on the cDNA array, we cannot comment on what effect the decrease in KIF3B may have on the expression of these interacting molecules.
Since KIF3B is involved in carrying constituents for neurite outgrowth, a decrease in expression could perhaps be associated with a change in the cellular morphology. However, a neuronal-specific protein, KIF3C, is also able to form a complex with KIF3A and KAP3 (56), and this may well be compensating for the decrease in KIF3B expression, such that no gross morphological change is observed.
c-Fes.
This is a non-receptor protein tyrosine kinase that is expressed predominantly in myeloid endothelial, epithelial and neuronal cells (35). Although described as a proto-oncogene because of its homology to a feline retrovirus gene, no association with human cancer or any other disease has been described. Potential biological roles for c-Fes include cell signalling from several different cytokine receptors (57,58), regulation of Ras or other related G proteins (59), and angiogenesis (60). A role has also been suggested in relation to intracellular vesicle transport, following localization of c-Fes to the trans-Golgi network, and observations of punctate staining in the cytoplasm and neurite outgrowths (35). More recently, use of a c-Fes–GFP fusion protein confirmed the perinuclear localization suggestive of the Golgi apparatus, and also indicated expression of c-Fes in apparent cytoplasmic vesicles (61). These vesicles also contained several Rab proteins, which are known markers for both excytosing (Rab1A and Rab3A), and endocytosing (Rab5B and Rab7) pathways. These monomeric GTPases regulate intracellular vesicle trafficking from the Golgi to the plasma membrane (62).
Therefore, there is a potential link between c-Fes and KIF3B, through vesicle trafficking; c-Fes co-localizes with Rab proteins, which are thought to be involved in the identification of molecular motors to their cargoes and/or activation of the molecular motors. Other evidence to support a role for GTPases and vesicle trafficking in MND includes the recently identified ALS2 gene. This encodes a novel GEF, a putative regulator of an as-yet unidentified GTPase (53,54). In addition, a newly identified GTPase has been identified as the gene responsible for SPG3A, an autosomal dominant form of the motor system degenerative disease hereditary spastic paraparesis (63).
ICAM1.
This is a heavily glycosylated protein (76–114 µkDa) composed of five extracellular Ig-like domains, a transmembrane spanning region and a cytoplasmic tail (36). It is a ligand for LFA-1, Mac1 and rhinovirus, and it participates in cellular interactions by binding to its ligands via different domains. ICAM1 was found to be downregulated in the presence of transfected human Cu/Zn SOD, although more significantly decreased in the presence of the mutant Cu/Zn SOD proteins. ICAM1 is constitutively expressed at low levels, but is upregulated by cytokines such as TNF-
(64), lipopolysaccharide (36) and oxidative stress (65). Studies investigating the levels of ICAM1 following treatment of melanoma cells with oxidative stress induced by paraquat, a generator of superoxide radicals (66) showed that ICAM1 activity was increased, in a concentration-dependent manner. The effect of Cu/Zn SOD on ICAM1 expression was investigated, since SOD activity also increased following paraquat treatment. By inhibiting Cu/Zn SOD with DTIC, the increase in ICAM1 expression was magnified. However, the addition of a SOD mimetic (MnTMPyP) inhibited the increase in ICAM1 expression. Therefore, it was suggested that the level of SOD1 expression plays a role in the modulation of ICAM1 expression and moderates its upregulation.
The overexpression of Cu/Zn SOD in the cellular model employed in this study is 20–60% of endogenous, and the mutant Cu/Zn SOD retains significant levels of superoxide dismutase activity [G37R was 100%; G93A was not determined, but G93C was 65% and I113T was 60% (5)]. Therefore, the changes in ICAM1 expression may be partially related to the activity of Cu/Zn SOD produced by transfection of the human genes.
In apparent contrast to our findings, a recent study demonstrated increased ICAM1 expression correlating with disease progression in G93A-SOD1 transgenic mice (67). However, immunohistochemistry localized the expression to microglia. Since the NSC34 cells have a motor neurone phenotype, expression of genes from cells such as microglia and astrocytes is not detected in this cellular model. This demonstrates that motor neurone-specific gene expression can be masked by the heterogeneous population of cells residing in the brain and spinal cord.
Bag1.
This is the Bcl2-associating athanogene (‘athano’ meaning anti-death) that enhances the pro-apoptotic activity of Bcl2, and also interacts with the Hsp70 family of molecular chaperones (37,38). There are two isoforms present in mouse, Bag1 and Bag1L, which are transcribed from a single mRNA and differ in the length of the N terminus (68). The upstream sequence contains a nuclear localization signal, and subcellular fractionation demonstrates that Bag1L is in the nuclear fraction whilst Bag1 is in the cytosol. Expression levels of Bag1 are reduced in olfactory neuronal cells undergoing apoptosis (69). The decrease in Bag1 expression in our cellular model would support the case for the involvement of an apoptotic cell death pathway in the SOD1-mediated neurodegeneration seen in MND (20). Of the downstream activators of cell death, caspases 1, 3, 7 and 11 were on the array. Although activation of caspase 3 has been demonstrated in this model (20), no change in gene expression was detected. This could be due to the fact that these enzymes are synthesized as procaspases, and that activation occurs primarily through protein cleavage rather than increased gene expression, or that any changes in gene expression were below 2-fold.
Confirmation of differential expression by quantitative RT–PCR was only obtained in two of the mutant-expressing cell lines: pC37- and pC113-transfected cells. Looking at the hybridization signals, little or no hybridization was seen on the arrays from these two cell lines. However, the fold differences for pC93-expressing cells compared with the two control cell lines only showed just over a 2-fold difference. This compares with the other genes, c-Fes, KIF3B and ICAM1, which showed fold changes of between 3 and 9. Other studies using the Clontech arrays have also experienced difficulty in confirming differential expression from identified level changes of between 2- and 3-fold (70). In addition, the use of ß-actin alone for normalization of the Taqman experiments compared with the use of both ß-actin and GAPDH for the array experiments may also contribute to this discrepancy.
What is not determined from the array analysis is whether there is an increase in a particular isoform of Bag1—either the cytosolic form, which interacts with Bcl2 and HSP70 molecular chaperones, or the Bag1L isoform, which is localized to the nucleus and stimulates transcription (71). Bag1 has a ubiquitin-like domain in the N terminus that associates with the 26S proteasome, linking the Hsc70 and Hsp70 molecular chaperones involved in protein folding, translocation and degradation to the proteasome (72). An abnormality in protein degradation may play a role in the aggregation of proteins. Aggregations of several proteins, including Cu/Zn SOD, neurofilament proteins and inclusion bodies containing proteins not yet identified, have been demonstrated both in human MND (49) and in transgenic mouse models of the disease (73). An increase in Hsp70 in cells containing mutant Cu/Zn SOD proteins reduced the formation of the aggregates (74), whilst Hsp70 was demonstrated to interact with mutant Cu/Zn SOD but not normal protein (75).
Interestingly, Bag1 expression was decreased in the two control cell lines (pCEP4 and pCN) following serum withdrawal. Comparative analysis of the array data suggests that Bag1 was decreased to levels similar to those found in the mutant SOD1-transfected cells under basal conditions. This suggests that the cells containing mutant Cu/Zn SOD are responding to an intracellular stress, and in the control cell lines, withdrawal of serum is required to obtain a corresponding intrinsic stress reaction.
MADR2.
This is a member of the MAD-related family required for serine/threonine kinase receptor signalling, and acts downstream in both TGF-ß and activin signalling pathways (39,40). Signalling occurs by TGF-ß or activin forming a complex with two serine/threonine receptor kinases. Type II receptors phosphorylate type I receptors, which in turn associate with and phosphorylate MADR2. MADR2 then translocates to the nucleus, where it associates with DNA-binding proteins and activates gene transcription (39). Although we identified a decrease in all three mutant Cu/Zn SOD-containing cell lines by cDNA array analysis, with no expression in pC93A and pC113 SOD1-transfected cells, we were not able to confirm this alteration by quantitative PCR.
TGF-ß is prominently expressed in large neurones, such as those in the hippocampus and spinal cord, and is an important regulator of neuronal differentiation and neurite outgrowth (76). Although MADR2 phosphorylation is stimulated by TGF-ß, no TGF-ß expression was detected in the cell lines, as determined by hybridization to the cDNA array. Activin, however, was not included on the array, so no comment on a correlation between MADR2 expression and its upstream activators can be made.
GCM1.
This is the mammalian homologue of the GCM Drosophila protein (41,42). Increased expression of GCM in Drosophila causes an increased number of glia and fewer neurones to develop, whereas loss of GCM expression leads to a decrease in glia and increase neurones, but no overall decrease in cell number (77,78). Although mammalian GCM1 shows high sequence conservation to GCM, and is able to rescue Drosophila GCM-null mutants, in mammals it is expressed specifically in the placenta (43). GCM1 is essential for the branching of the labyrinth, where exchange between maternal and fetal blood supplies occurs (44).
The GCM1 hybridization signals are clearly seen on the arrays from those cells transfected with vector-only or normal SOD1 cDNA, and are absent from those expressing a mutant Cu/Zn SOD. However, this differential expression could not be confirmed by quantitative PCR; no specific amplification products were obtained from the cell lines. This scenario has been found previously, using both the cDNA membrane arrays (79) and microarrays (80), and emphasizes that validation of the data is crucial.
NEX1.
This is a neuronal-specific gene containing a basic helix–loop–helix motif (45). It is expressed in differentiating neurones and glia during development, and continues to be expressed in the mouse adult brain with highest levels in the cerebellum. Similar to GCM in Drosophila, it is thought to play a role in glial/neuronal cell fate, since overexpression of NEX1 blocks neurogenesis in embryonic neural cells (81). However, the role that it plays during adulthood is, as yet, unknown. A human homologue for NEX1, NeuroD6 (GenBank accession no. NM_022728, NCBI), has been identified, but little is known about the function of this gene.
Hybridization signals are clearly present on the arrays in both the pCEP4 and pCN SOD1-transfected cell lines, with no signal detected in the pC37 SOD1 and pC113 SOD1 cell lines. There was a low expression level detected in the pC93A SOD1 cell line. However, no expression from these original cDNAs can be detected by quantitative PCR following 50 cycles. As mentioned previously, this has been found before (67,68), and suggests that there is cross-hybridization between the cDNA sequence attached to the array and another, closely related, cDNA that is transcribed in the cell lines.
Others genes differentially expressed under basal conditions
Although our primary interest is in genes differentially regulated in the presence of mutant Cu/Zn SOD, other genes also showed changes in expression in our cellular models. Genes that were increased in the presence of normal Cu/Zn SOD included several DNA-synthesis, -repair and -recombination genes. Genes that showed a decrease in expression in the presence of Cu/Zn SOD protein, whether normal or mutant, included plasma glutathione peroxidase precursor (GSHPX-P), which catalyses the reduction of hydrogen peroxide, working downstream of superoxide dismutase. However, these decreases may also be in response to the presence of ‘a protein’, rather than specifically to Cu/Zn SOD.
It is perhaps surprising that the consistent gene expression changes were decreased in the presence of a mutant Cu/Zn SOD, and no genes were found to be consistently increased. This could perhaps be due to some changes occuring post-translationally, such as the cascade pathways involved in apoptosis, in which the caspases are synthesized as procaspases and are then cleaved to become active. In addition, the balance between activators and inhibitors could be altered within a 2-fold increase in gene expression. Further, owing to experimental constraints with several genes being overexposed on the arrays, these were not able to be compared quantitatively between filters, and were excluded from the analysis.
Genes differentially expressed following oxidative stress
NSC34 cells transfected with mutant SOD1 constructs show a decrease in cell viability over increasing time periods of serum withdrawal, compared with those transfected with vector only (15). The cells transfected with normal SOD1 are less sensitive to serum withdrawal, demonstrating a protective effect of normal Cu/Zn SOD. Fetal calf serum is thought to contain neurotrophic factors and antioxidant enzymes in addition to salts and sugars. Serum withdrawal from the NSC34 cells results in the immediate release of the free-radical species nitric oxide and superoxide, as measured by a bioelectrode (23), indicating that the cells are under oxidative stress.
Therefore, we investigated the effects of the mutant Cu/Zn SOD proteins upon gene expression following serum withdrawal. Twelve genes were found to be consistently differentially expressed following nine-way analyses with the two mutant SOD1-transfected cell lines (pC93A and pC113) compared with the two control cell lines (pCEP4 and pCN). Only one gene, the c-Src proto-oncogene, was found to be consistently decreased in all four cell lines following serum withdrawal. Although this could suggest that the majority of gene expression changes following serum withdrawal (which would be detectable on this array) are less than a 2-fold change in expression level, the low number may also be explained by the highly stringent selection criteria imposed during the analysis. No genes were found to be specifically increased or decreased in the presence of mutant Cu/Zn SOD following serum withdrawal. However, keratinocyte growth factor (FGF-7) was decreased in all cells transfected with SOD1, whether normal or mutant.
cDNA array analysis was first applied to the study of MND in 2001, when human lumbar spinal cord (containing multiple cell types) from MND and control cases was applied to high-density gene discovery arrays, membrane arrays containing 18 400 non-redundant EST cDNAs (82). Significant differential expression was found for 14 genes, 13 of which increased in MND cases, and 1 decreased. None of these genes correlates with those identified in our neuronal culture model. However, these specific genes were not included on the mouse cDNA array employed in this study. In addition, one of the genes elevated was glial fibrillary acid protein, a marker of astrocytic activation, and as such is not likely to be expressed in a neuronal-specific cell line. Other genes identified included anti-oxidants (thioredoxin) and neuroinflammatory factors (interleukin-1 receptor accessory protein).
A comprehensive analysis of gene expression changes in the G93A-SOD1 transgenic mouse was reported by Olsen and colleagues (83). Transcript profiles from spinal cord of G93A-SOD1 transgenic mice and their littermates were compared at 30, 60, 90 and 120 days using the Affymetrix murine 6500 oligonucleotide microarray. An additional comparison at 120 days with spinal cord of normal SOD1 transgenic mice was also performed. Prior to disease onset at 30 and 60 days, there were few significant changes in gene expression between the G93A-SOD1 transgenic mice and their littermates. However, at 90 and 120 days, gene expression changes in the G93A-SOD1 transgenic mice became significant, and changes in expression levels of 24 genes were reported. Four cDNAs were also on the cDNA array used for this study, although three of these are involved in astrocytic activation, and these changes in expression would not be expected in a neuronal cell line. Expression of vimentin and LFA-1 was not detected in any of the cell lines, whilst -integrin was expressed but no significant differences were identified. The fourth gene, the transferrin receptor, was increased at 120 days in the G93A-SOD1 transgenic mice, but low or undetectable in the littermates and normal SOD1 transgenic mice. This gene was not expressed in any of the transfected NSC34 cell lines.
A further report analysing gene expression changes in the G93A-SOD1 transgenic mice, compared with their littermates, used both the mouse Atlas array 1.2 (Clontech) containing 1176 genes on the membrane and the mouse Atlas Glass microarray containing 1081 genes (84). Lumbar spinal cords were obtained at 7, 11, 14 and 17 weeks (
49, 77, 98 and 117 days). At 17 weeks, 30 genes were increased and 7 decreased at least 3-fold, compared with littermates. Four of the genes increased were also identified by Olsen and colleagues (83).
Sixteen of the genes altered were also included on the cDNA array used here. Of these, expression of 13 could not detected in the NSC34 cell lines, including vimentin, a marker of astrocytic activation. Cathepsin D was increased 3-fold at 17 weeks in the mice, but was decreased in the NSC34 cells transfected with normal SOD1 and pC93A mutant SOD1. This protein, however, is also found to be upregulated during astrocytic activation (85), and therefore this increase would not have been detected in our neuronal cell line. The serine protease inhibitor 2–4 was expressed 3-fold higher in G93A-SOD1 transgenic mice of 17 weeks, but only 1.4-fold higher at 11 weeks. No significant difference in expression of the serine protease inhibitor 2–4 was detected in the transfected NSC34 cells, and this may reflect an age difference between the motor neurones in the cell lines and the transgenic mice models. The remaining gene, gastrulation brain homeobox 2 (gbx2), which was also increased 3-fold at 17 weeks in the G93A-SOD1 transgenic mice, was found to be decreased in the NSC34 cells transfected with normal SOD1 and pC93A mutant SOD1, compared with the vector-only-transfected cells. This again may be an age-related phenomenon.
Of interest, KIF3C was increased in the G93A-SOD1 transgenic mice at both 11 and 17 weeks. This is the kinesin-related motor protein, which is expressed primarily in brain and neurones, and is able to interact with both KIF3A and KAP3 in the place of KIF3B (56,86). KIF3C is most closely related to KIF3B, and its expression in the brain is similar to that of KIF3A and KIF3B. KIF3B expression is decreased in our neuronal cell line in the presence of mutant SOD1, and this leads to the possibility of the increase in KIF3C as a compensatory mechanism. The expression levels of KIF3C in the NSC34 cell lines will be investigated to determine if this is indeed the case.
This study has identified some of the genes that may be involved in triggering the neurodegenerative process in familial MND. Further studies are underway in this laboratory to investigate the expression levels of these genes in motor neurones from human MND cases. This will identify whether changes in the expression of genes in the cell model of familial MND are relevant to the human disease, and whether changes are specific to SOD1-related MND or present more generally in other forms of motor neurone degeneration.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Cell culture
NSC34 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% foetal calf serum (FCS) as described previously (87). The pCEP4 expression vectors containing normal or mutant SOD1 cDNAs were generated as described by Durham et al. (11). The NSC34 cells were transfected with vector only (pCEP4), normal human SOD1 cDNA (pCN) or one of four mutant SOD1 cDNAs encoding the G37R, G93A and I113T amino acid substitutions (pC37, pC93A and pC113), according to the method previously reported (15). From these parental cell lines, single-cell clones were selected by limiting dilution. Single-cell clones were grown until they were confluent, and then either received fresh media with FCS or without FCS for 24 h, after which time they were harvested for RNA extraction.
Western blotting for human Cu/Zn SOD
Protein extracts were harvested from confluent cells under basal conditions as described previously (87). Ten milligrams of protein was loaded on 14% gels and run for 1 h at 20 W/gel. The proteins were then electroblotted on to Immobilon-P membrane and the membranes were blocked in TBS/Tween (20 mM Tris–HCl, pH 7.6, 137 mM NaCl and 0.1% Tween-20) with 5% dried skimmed milk for 1 h. Anti-human Cu/Zn SOD primary antibody was used at 1 in 2500 (The Binding Site, UK). The secondary antibody, anti-sheep, was used at 1 in 1000 (Sigma). Antigen detection was carried out using enhanced chemiluminescence (ECL) (Amersham), according to the manufacturer's instructions.
RNA extraction
Cells were harvested, washed in PBS and resuspended in denaturing solution (BD Clontech). Total RNA was extracted using the Atlas Pure Total RNA Isolation Kit according to the manufacturer's instructions (BD Clontech) and quantified spectrophotometrically. Total RNA was DNase-treated prior to mRNA isolation using the NucleoTrap mRNA Purification Protocol (BD Clontech).
PCR of mRNA using genomic DNA primers
To ensure that there was no contaminating genomic DNA in the mRNA extract, 1 µl of mRNA from each sample was used in a PCR containing SOD1 genomic primers that amplify exon 4 (33); 12.5 pmol of each primer and Reddy Load PCR Mix (ABGene) to a total volume of 25 µl were used. Amplification parameters were as described previously (33).
Probe preparation and cDNA array hybridization
One microgram of mRNA was reverse-transcribed using gene-specific coding sequence (CDS) primers, according to the Atlas cDNA Expression Arrays User Manual. One exception was the use of [32P]dCTP (NEN), which required a 10x mix of 5 mM each of dATP, dGTP and dTTP (Boehringer Mannheim). The probe was purified by column chromatography using Nucleotrap-200 DEPC–H2O columns (BD Clontech). The membrane arrays were pre-hybridized for 30 min in hybridization solution, whilst the probe was denatured. Hybridization to the Atlas Mouse cDNA Expression Array I (BD Clontech) took place over 17 h at 68°C. The membranes were washed in 2x SSC/0.1%SDS at 68°C, three times for 30 min, followed by one wash of 0.1x SSC/0.5%SDS at 68°C, for 30 min. They were then exposed to X-ray film (Kodak Biomax MS) with intensifying screens at -80°C. To allow for quantification of both high and low transcript levels, short and long exposures of the membranes were obtained. Following autoradiography, the membranes were stripped in boiling 0.5% SDS for 10 min, allowed to cool, and then stored at -20°C until required.
Array analysis
The X-ray films were scanned and the images analysed using AtlasImage software v1.01 (BD Clontech). Pairwise comparisons between filters were carried out following normalization with GAPDH and ß-actin expression levels. Alterations in expression levels of 2-fold or more were identified. In addition, genes that were undefined for one filter, but were detected on the other, were also reported. (Undefined genes were those where there was either no hybridization signal, or where the level of hybridization was <150% of the background level). Initially, the expression profiles of the cells under basal conditions were compared. pCEP4 and pCN were compared and then the mutant cell lines (pC93A, pC113 or pC37) were compared separately against the pCEP4 and pCN expression profiles, resulting in a three-way pairwise analysis. Only changes that were consistent between the three pairwise comparisons were noted. (If a 2-fold difference or more was seen between the pCEP4 and pC93A, and no difference between the pCEP4 and the pCN, a 2-fold difference had to be seen between the pCN and the pC93A for the result to be counted.) This reasoning was used to ensure consistent and robust results.
The expression profile following serum withdrawal was compared first against the cell line under basal culture conditions (e.g. pC93A+serum versus pC93A-serum). Of those genes that showed a difference, their expression profiles under basal conditions (pC93A+serum versus pCEP4 +serum, pC93A+serum versus pCN+serum and pCEP4 +serum versus pCN+serum) and following oxidative stress (pC93A-serum versus pCEP4-serum, pC93A-serum versus pCN-serum and pCEP4-serum versus pCN -serum) were compared, to give a nine-way analysis (Table 3). In each case, as previously, at least a 2-fold difference had to be consistent between the analyses for an alteration in gene expression to be considered robust.
Differentially expressed genes were divided into categories according to their expression patterns. Since we were primarily interested in changes due to the presence of mutant SOD1, changes where there was differential expression between the three cell lines—for example a decrease in the presence of pCN but a greater decrease in the presence of pC93A—were classed as decreased in the presence of pC93A (see Table 1).
Standard RT–PCR
Standard RT–PCR was carried out with each Taqman primer set prior to using SYBR green or the Taqman probes. RT–PCR reactions contained 1 µl of cDNA, 12.5 pmol of each primer (Tables 5 and 6) and Reddy Load PCR Mix (ABGene) to a total volume of 25 µl. The Taqman PCR programme of 95°C for 15 s and 60°C for 1 min, for 50 cycles was used, following an initial denaturation step of 95°C for 10 min. For amplifying NEX1, the primers were 5'-ATTCTGAGGATTGGCAAGAGACCG-3' and 5'-AAGGGACCTTCAAACTGAGGGCTGGC-3', and the PCR was set up as above. The PCR program used for NEX1 amplification was an initial denaturation step of 95°C for 5 min, followed by 95°C for 30 s, 66°C for 30 s and 72°C for 45 s, for 50 cycles. All PCR products were electrophoresed on 2% agarose gels.
|
|
Real-time quantitative RT–PCR
Aliquots of the original mRNA used for the cDNA array were reverse-transcribed using random hexamers and Multiscribe reverse transcriptase according to the manufacturer's instructions (Applied Biosystems). Primers were designed using Primer Express software (Applied Biosystems), where possible taking into account intron/exon boundaries to ensure specific amplification of cDNA (Table 5). Primers were designed to ß-actin as an internal control for normalization of starting cDNA levels. Quantitative PCR was carried out using SYBR Green PCR Master Mix according to the manufacturer's instructions (Applied Biosystems), with the exception that 25 µl reaction volumes were used, with 50 cycles of amplification. Each of the primer pairs was optimized to ensure amplification of the specific product and absence of primer dimers (Table 5). PCR was carried out on the Taqman 7700 (Applied Biosystems), and following amplification, RT–PCR reactions were electrophoresed on agarose gels to check for the correctly sized product. The real-time PCR results were analysed using the sequence detection system software v1.7 (Applied Biosystems). For each primer pair, a standard curve of 4, 2, 1, 0.5 and 0.25 µl of cDNA was run alongside the samples, from which a relative level of gene expression in the samples could be calculated. These levels were normalized to those of ß-actin, which were obtained in the same way as described, and then the expression levels were expressed as a percentage of that seen in pCEP4. At least three independent experiments for each of the genes were carried out. The results were analysed by one-way ANOVA followed by the Newman–Kuels post hoc test to determine their statistical significance.
Quantitative PCR using Taqman probes was used for measuring gene expression of ICAM1 and GCM1, because of the formation of non-specific products and primer dimers when using SYBR green. PCR was carried out using Taqman Universal PCR Master Mix according to the manufacturer's instructions (Applied Biosystems), with the exception that 25 µl reaction volumes were used, with 50 cycles of amplification. Primer pairs and probe concentrations were optimized (Table 6). Experiments were performed and data analysed in the same way as in the SYBR green experiments.
|
ACKNOWLEDGEMENTS |
|---|
Thanks go to Dr Denise Figlewicz for supplying the SOD1 cDNA constructs and to Dr Neil Cashman for the kind gift of the original NSC34 cells. We also thank Dr Ann Dalton and the North Trent Molecular Genetics Unit at the Sheffield Children's Hospital for use of the Taqman 7700. J.T., F.M.M. and P.J.S. are supported by the Wellcome Trust.
|
FOOTNOTES |
|---|
* To whom correspondence should be addressed. Tel: +44 1142712473; Fax: +44 1142261201; Email: jkirby{at}hgmp.mrc.ac.uk
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
1 Rosen, D.R., Siddique, T., Patterson, D., Figelwicz, D.A., Sapp, P., Hentati, A., Donaldson. D., Goto, J., O'Regan, J.P., Deng, H.-X. et al. (1993) Mutations in Cu/Zn superoxide dismutase gene are associated with familial amyotrophic lateral sclerosis. Nature, 362, 59–62.[Medline]
2
Gurney, M.E., Pu, H.F., Chiu, A.Y., DalCanto, M.C., Polchow, C.Y., Alexander, D.D., Caliendo, J., Hentati, A., Kwon, Y.W., Deng, H.X. et al. (1994) Motor neuron degeneration in mice that express a human Cu, Zn superoxide dismutase mutation. Science, 264, 1772–1772.
3 Wong, P.C., Pardo, C.A., Borchelt, D.R., Lee, M.K., Copland, N.G., Jenkins, N.A., Sisodia, S.S., Cleveland, D.W. and Price, D.L. (1995) An adverse property of a familial ALS-linked SOD1 mutation causes motor neuron disease characterised by vacuolar degeneration of mitochondria. Neuron, 14, 1105–1116.[ISI][Medline]
4 Reaume, A.G., Elliot, J.L., Hoffman, E.K., Kowall, N.W., Ferrante, R.J., Siwek, D.F., Wilcox, H.M., Flood, D.G., Beal, M.F., Brown, R.H. et al. (1996) Motor neurons in Cu/Zn superoxide dismutase-deficient mice develop normally but exhibit enhanced cell death after axonal injury. Nat. Genet., 13, 43–47.[ISI][Medline]
5
Borchelt, D.R., Lee, M.K., Slunt, H.H., Guarnieri, M., Xu, Z.-S., Wong, P.C., Brown, R.H., Price, D.L., Sisodia, S.S. and Cleveland, D.W. (1994) Superoxide dismutase 1 with mutations linked to familial amyotrophic lateral sclerosis possesses significant activity. Proc. Natl Acad. Sci. USA, 91, 8292–8296.
6
Ratovitski, T., Corson, L.B., Strain, J., Wong, P., Cleveland, D.W., Culotta, V.C. and Borchelt, D.R. (1999) Variation in the biochemical/biophysical properties of mutant superoxide dismutase 1 enzymes and the rate of disease progression in familial amyotrophic lateral sclerosis kindreds. Hum. Mol. Genet., 8, 1451–1460.
7 Wiedau-Pazos, M., Goto, J.J., Rabizadeh, S., Gralla, E.B., Roe, J.A., Lee, M.K., Valentine, J.S. and Bredesen, D.E. (1996) Altered reactivity of superoxide dismutase in familial amyotrophic lateral sclerosis. J. Neurol. Sci., 156, 65–72.
8 Beckman, J.S., Carson, M., Smith, C.D. and Koppenol, W.H. (1993) ALS, SOD and peroxynitrite. Nature, 364, 584.[Medline]
9 Brown, R.H. (1995) Amyotrophic lateral sclerosis—recent insights from genetics and transgenic mice. Cell, 80, 687–692.[ISI][Medline]
10 Shibata, N., Hirano, A., Kobayashi. M., Siddique, T., Deng, H.X., Hung, W.Y., Kato, T. and Asayama, K. (1996) Intense superoxide dismutase-1 immunoreactivity in intracytoplasmic hyaline inclusions of familial amyotrophic lateral sclerosis with posterior column involvement. J. Neuropathol. Exp. Neurol., 55, 481–490.[ISI][Medline]
11 Durham, H.D., Roy, J., Dong, L. and Figlewicz, D.A. (1997) Aggregation of mutant Cu/Zn superoxide dismutase proteins in a culture model of ALS. J. Neuropathol. Exp. Neurol., 56, 523–560.[ISI][Medline]
12 Hunter, D.D., Cashman, N., Morris-Valero, R., Bulock, J.W., Adams, S.P. and Sanes, J.R. (1991) An LRE (leucine-arginine-glutamate)-dependent mechanism for adhesion of neurones to S-laminin. J. Neurosci., 11, 3960–3971.[Abstract]
13 Cashman, N.R., Durham, H.D., Blusztajn, J.K., Oda, K., Tabira, T., Shaw, I.T., Dahrouge, S. and Antel, J.P. (1992) Neuroblastoma x spinal cord (NSC) hybrid cell lines resemble developing motor neurones. Dev. Dynam., 194, 209–221.[ISI][Medline]
14 Durham, D.H., Dahrouge, S. and Cashman, N.R. (1993) Evaluation of the spinal cord x neuroblastoma hybrid cell line NSC34 as a model for neurotoxictiy testing. Neurotoxicology, 14, 387–396.[ISI][Medline]
15 Menzies, F.M., Cookson, M.R., Taylor, R., Turnbull, D,M,, Dong, L., Figlewicz, D.A. and Shaw, P.J. (2002) Mitochondrial dysfunction in a cell culture model of familial amyotrophic lateral sclerosis. Brain, 125, 1522–1533.
16
Kong, J. and Xu, Z. (1998) Massive mitochondrial degeneration in motor neurones triggers the onset of amyotrophic lateral sclerosis in mice expressing a mutant SOD1. J. Neurosci., 18, 3241–3250.
17 Sasaki, S. and Iwata, M. (1996) Ultrastructural study of synapses in the anterior horn neurones of patients with amyotrophic lateral sclerosis. Neurosci. Lett., 204, 53–56.[ISI][Medline]
18 Fujita, K., Yamauchi, M., Shibayama, K., Ando, M., Honda, M. and Nagata, Y. (1996) Decreased cytochrome c oxidase activity but unchanged superoxide dismutase and glutathione peroxidase activites in the spinal cords of patients with amyotrophic lateral sclerosis. J. Neurosci. Res., 45, 276–281.[ISI][Medline]
19 Borthwick, G.M., Johnson, M.A., Ince, P.G., Shaw, P.J. and Turnbull, D.M. (1999) Ann. Neurol., 46, 787–790.[ISI][Medline]
20 Sathasivam, S., Ince, P.G. and Shaw, P.J. (2001) Apoptosis in amyotrophic lateral sclerosis: a review of the evidence. Neuropathol. Appl. Neurobiol., 27, 257–274.[ISI][Medline]
21 Sathasivam, S., Menzies, F.M., Cookson, M.R., Grierson, A.J. and Shaw, P.J. (2002) Characterisation of the cell death pathway in a cellular model of familial motor neurone disease. J. Neurol. Neurosurg. Psych., 73, 41.
22 Menzies, F.M., Grierson, A.J., Cookson, M.R., Heath, P.R., Tomkins, J., Figlewicz, D.A., Ince, P.G. and Shaw, P.J. (2002) Selective loss of neurofilament expression in Cu/Zn superoxide dismutase (SOD1) linked amyotrophic lateral sclerosis. J. Neurochem., in press.
23 Cookson, M.R., Menzies, F.M., Manning, P., Eggett, C.J., Figlewicz, D.A., McNeil, C.J. and Shaw, P.J. (2002) Cu/Zn superoxide dismutase (SOD1) mutations associated with familial amyotrophic lateral sclerosis (ALS) affect cellular free radical release in the presence of oxidative stress. Amyotrophic Lateral Sclerosis, in press.
24
Sgroi, D.C., Teng, S., Robinson, G., LeVangie, R., Hudson, J.R. and Elkahloun, A.G. (1999) In vivo gene expression profile analysis of human breast cancer progression. Cancer Res., 59, 5656–5661.
25 Alizadeh, A.A., Eisen, M.B., Davis, R.E., Ma, C., Lossos, I.S., Rosenwald, A., Boldrick, J.C. Sabet, H., Tran, T., Powell, J.I. et al. (2000) Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature, 403, 503–511.[Medline]
26 Liotta, L. and Petricoin, E. (2000) Molecular profiling of human cancer. Nat. Rev. Genet., 1, 48–56.[ISI][Medline]
27 Mandel, S., Grunblatt, E. and Youdim, M. (2000) cDNA microarray to study gene expression of dopaminergic neurodegeneration and neuroprotection in MPTP and 6-hydroxydopamine models: implications for idiopathic Parkinson's disease. J. Neural. Transm., 60, 117–124.
28
Luthi-Carter, R., Strand, A., Peters, N.L., Solano, S.M., Hollingsworth, Z.R., Menon, A.S., Frey, A.S., Spektor, B.S., Penney, E.B., Schilling, G. et al. (2000) Decreased expression of striatal signaling genes in a mouse model of Huntington's disease. Hum. Mol. Genet., 9, 1259–1271.
29 Ginsberg, S.D., Hemby, S.E., Lee, V.M.-Y., Eberwine, J.H. and Trojanowski, J.Q. (2000) Expression profile of transcripts in Alzheimer's disease tangle-bearing CA1 neurones. Ann. Neurol., 48, 77–87.[ISI][Medline]
30
Bertucci, F., Houlgatte, R., Benziane, A., Granjeaud, S., Adelaide, J., Tagett, R., Loriod, B., Jacquemier, J., Viens, P., Jordan, B. et al. (2000) Gene expression profiling of primary breast carcinomas using arrays of candidate genes. Hum. Mol. Genet., 9, 2981–2991.
31 Heller, R.A., Scena, M., Chai, A., Shalon, D., Bedilion, T., Gilmore, J., Woolley, D.E. and Davis, R.W. (1997) Discovery and analysis of inflammatory disease-related genes using cDNA microarrays. Proc. Natl Acad. Sci. USA, 94, 2150–2155.