Analysis of the cytosolic proteome in a cell culture model of familial amyotrophic lateral sclerosis reveals alterations to the proteasome, antioxidant defenses, and nitric oxide synthetic pathways.

Injury to motor neurons associated with mutant Cu,Zn-superoxide dismutase (SOD1)-related familial amyotrophic lateral sclerosis (FALS) results from a toxic gain-of-function of the enzyme. The mechanisms by which alterations to SOD1 elicit neuronal death remain uncertain despite intensive research effort. Analysis of the cellular proteins that are differentially expressed in the presence of mutant SOD1 represents a novel approach to investigate further this toxic gain-of-function. By using the motor neuron-like cell line NSC34 stably transfected with wild-type, G93A, or G37R mutant human SOD1, we investigated the effects of mutant human SOD1 on protein expression using proteomic approaches. Seven up-regulated proteins were identified as argininosuccinate synthase, argininosuccinate lyase, neuronal nitric-oxide synthase, RNA-binding motif protein 3, peroxiredoxin I, proteasome subunit beta 5 (X), and glutathione S-transferase (GST) Alpha 2. Seven down-regulated proteins were identified as GST Mu 1, GST Mu 2, GST Mu 5, a hypothetical GST Mu, GST Pi B, leukotriene B(4) 12-hydroxydehydrogenase, and proteasome subunit beta5i (LMP7). GST assays demonstrated a significant reduction in the total GST activity of cells expressing mutant human SOD1. Proteasome assays demonstrated significant reductions in chymotrypsin-like, trypsin-like, and post-glutamylhydrolase proteasome activities. Laser capture microdissection of spinal cord motor neurons from human FALS cases, in conjunction with reverse transcriptase-PCR, demonstrated decreased levels of mRNA encoding GST Mu 1, leukotriene B(4) 12-hydroxydehydrogenase, and LMP7. These combined approaches provide further evidence for involvement of alterations in antioxidant defenses, proteasome function, and nitric oxide metabolism in the pathophysiology of FALS.

Injury to motor neurons associated with mutant Cu,Znsuperoxide dismutase (SOD1)-related familial amyotrophic lateral sclerosis (FALS) results from a toxic gain-of-function of the enzyme. The mechanisms by which alterations to SOD1 elicit neuronal death remain uncertain despite intensive research effort. Analysis of the cellular proteins that are differentially expressed in the presence of mutant SOD1 represents a novel approach to investigate further this toxic gain-of-function. By using the motor neuron-like cell line NSC34 stably transfected with wild-type, G93A, or G37R mutant human SOD1, we investigated the effects of mutant human SOD1 on protein expression using proteomic approaches. Seven up-regulated proteins were identified as argininosuccinate synthase, argininosuccinate lyase, neuronal nitric-oxide synthase, RNA-binding motif protein 3, peroxiredoxin I, proteasome subunit ␤5 (X), and glutathione S-transferase (GST) Alpha 2. Seven downregulated proteins were identified as GST Mu 1, GST Mu 2, GST Mu 5, a hypothetical GST Mu, GST Pi B, leukotriene B 4 12-hydroxydehydrogenase, and proteasome subunit ␤5i (LMP7). GST assays demonstrated a significant reduction in the total GST activity of cells expressing mutant human SOD1. Proteasome assays demonstrated significant reductions in chymotrypsin-like, trypsin-like, and post-glutamylhydrolase proteasome activities. Laser capture microdissection of spinal cord motor neurons from human FALS cases, in conjunction with reverse transcriptase-PCR, demonstrated decreased levels of mRNA encoding GST Mu 1, leukotriene B 4 12-hydroxydehydrogenase, and LMP7. These combined approaches provide further evidence for involvement of alterations in antioxidant defenses, proteasome function, and nitric oxide metabolism in the pathophysiology of FALS.
Amyotrophic lateral sclerosis (ALS), 1 the most common form of motor neuron disease, is a fatal, adult-onset neurodegenera-tive disorder, characterized by selective loss of lower and upper motor neurons from the spinal cord and brain. Approximately 10% of ALS cases are inherited, and 20% of these familial ALS (FALS) cases result from dominantly inherited missense mutations in the gene encoding Cu,Zn-superoxide dismutase (SOD1) (1). As most FALS SOD1 mutants retain dismutase activity close to that of the wild-type enzyme (2), the injury to motor neurons associated with mutant SOD1 may result from a toxic gain-of-function of the enzyme rather than loss of its ability to catalyze the conversion of superoxide to hydrogen peroxide. Furthermore, mice with targeted deletion of the sod1 gene do not develop an ALS phenotype (3) in contrast to transgenic mice expressing mutant human SOD1 (4 -6).
Several non-mutually exclusive hypotheses have been proposed to describe the toxic gain-of-function of mutant SOD1 (7,8). These include altered free radical handling, altered copper/ zinc binding, and formation of high molecular weight protein aggregates. Evidence for altered free radical handling has come from numerous observations; for example, indices of free radical damage are increased in the transgenic mutant SOD1 mice (9) and human ALS cases (10,11). Cultured cells expressing SOD1 mutants have been shown to exhibit increased oxygen radical production and sensitivity to exogenously produced free radicals (12,13). Intracellular superoxide can react with nitric oxide to produce the oxidant peroxynitrite (14). Several studies have provided evidence for the role of peroxynitrite and nitric oxide in SOD1-related ALS. Increased nitrosylation of proteins by peroxynitrite has been suggested by elevated levels of 3-nitrotyrosine in transgenic mutant SOD1 mice (15,16). Reduced zinc binding has been demonstrated for several SOD1 mutants (17). Interestingly, zinc-deficient wild-type SOD1, as well as mutant SOD1, generates superoxide, which in turn increases peroxynitrite production. Apoptosis of cultured primary neu-rons induced by zinc-deficient SOD1 can be reduced by treatment with inhibitors of nitric-oxide synthase (18).
Protein aggregation as a toxic gain-of-function was initially proposed after the demonstration of anti-SOD1-reactive cytoplasmic inclusions in motor neurons and surrounding astrocytes in mutant SOD1 transgenic mouse models (19) and cell culture models (20) of human FALS. Aggregation may lead to toxicity in a number of ways. SOD1 aggregates may have altered free radical and metal ion chemistry as described above. They may also challenge the protein folding and degradative machinery of the cell, compromising housekeeping protein functions essential for cell viability (8). Indeed only modest inhibition of the proteasome complex is required to generate mutant SOD1 aggregates in transfected cell lines (21). Whatever the relative contributions of these potentially inclusive modes of SOD1 toxicity are, the molecular pathways they trigger that ultimately lead to neuronal degeneration are poorly understood. Studies using cell models (13) or transgenic mutant SOD1 mice (22) have demonstrated biochemical markers suggesting an apoptotic mode of programmed cell death for degenerating neurons (reviewed in Ref. 23).
Recent developments in gene expression profiling technology have provided new impetus in the identification of molecular pathways activated by mutant ALS SOD1. Two independent studies using microarray analysis of transgenic mice expressing the human FALS-associated G93A SOD1 mutant (24,25) and microarray analysis of human ALS spinal cord (26) have detected some characteristic gene expression changes. Differential regulation of apoptosis-, inflammation-, and antioxidantrelated genes were findings common to all of these studies. The use of spinal cord tissue in these studies also underlined the possible involvement of protein expression changes in nonneuronal cells including microglia and astrocytes in motor neuron degeneration. Attributing any of these changes specifically to the motor neurons that account for a very small proportion of spinal cord tissue may not be possible. Primary alterations within degenerating motor neurons in response to mutant ALS SOD1 may go undetected among the protein changes occurring within the complex mixture of more abundant cell types in the central nervous system. We have addressed this issue by developing a cell culture model of FALS based upon a murine motor neuron-like cell line, NSC34, stably transfected with human FALS associated SOD1 mutants G93A and G37R (27)(28)(29). The NSC34 cells are a hybrid motor neuron ϫ neuroblastoma cell line that exhibits several features of motor neurons including neurofilament expression, the ability to generate action potentials, and induction of twitching in co-cultured muscle cells (30). Alterations in gene expression in this model were analyzed previously using a microarray approach. Expression of genes involved in the regulation of axonal transport, vesicular trafficking, and apoptosis were found to be altered by expression of ALS mutant hSOD1 (28). Here we extend these studies using proteomic techniques to analyze the alterations in protein expression. We have demonstrated that expression of both G37R and G93A hSOD1 results in the differential expression and altered function of proteins that regulate nitric oxide metabolism, intracellular redox conditions, and protein degradation.

EXPERIMENTAL PROCEDURES
Reagents-All two-dimensional gel electrophoresis reagents were ultra-pure grade and purchased from Bio-Rad and Sigma. COMPLETE EDTA-free TM protease inhibitor mixture was purchased from Roche Molecular Biochemicals. Cell culture media and reagents were purchased from Invitrogen. Enhanced chemiluminescence (ECL) kits and glutathione-Sepharose were purchased from Amersham Biosciences. Reduced glutathione assay kit, reduced glutathione, 7-amino-4-methylcoumarin (AMC), Suc-LLVY-AMC, Z-ARR-AMC, and Z-LLE-AMC were purchased from Calbiochem. Owl silver stain and rabbit anti-rat glutathione transferase-Mu antibody were purchased from Autogen Bioclear (Calne, UK). Sheep anti-bovine SOD1 antibody was purchased from The Binding Site (Birmingham, UK). Mouse anti-actin (AC-40), anti-mouse inducible nitric-oxide synthase monoclonal antibodies, and donkey anti-sheep IgG horseradish peroxidase conjugate were purchased from Sigma. Rabbit anti-rat neuronal nitric-oxide synthase antibody was purchased from Zymed Laboratories Inc. Rabbit antihuman nucleotide diphosphate kinase A antibody (nm23-H1 C-20) was purchased from Santa Cruz Biotechnology, Inc. Rabbit anti-bovine endothelial nitric-oxide synthase antibody was purchased from Bioquote Ltd. (York, UK). Rabbit anti-rat argininosuccinate synthase antibody was a gift from Masataka Mori (University of Kumamoto, Japan). Rabbit anti-rat argininosuccinate lyase antibody was a gift from Heinrich Wiesinger (University of Tuebingen, Germany). Rabbit anti-mouse LMP7 antibody was a gift from John Monaco (University of Cincinnati). Rabbit anti-human proteasome subunit X antibody was a gift from Klavs Hendil (University of Copenhagen, Denmark). Rabbit anti-mouse LMP2 antibody was purchased from Affiniti Research Products Ltd. (Exter, UK). Rabbit anti-porcine leukotriene B 4 12-hydroxydehydrogenase antibody was a gift from Takehiko Yokomizo and Takao Shimizu (University of Tokyo, Japan). Swine anti-rabbit IgG horseradish peroxidase conjugate and goat anti-mouse IgG horseradish peroxidase conjugate were purchased from Dako (Ely, UK).
Cell Lines and Cytosol Preparation-NSC34 single cell clones stably expressing pCEP4 expression vector only, wild-type hSOD1, G93A hSOD1, and G37R hSOD1 have been described previously (27). Cytosol was prepared from NSC34 cells using a modified version of the method of Yang and co-workers (31). NSC34 cells were seeded into T175 flasks at a density of 9 ϫ 10 5 cells per flask and maintained in Dulbecco's modified Eagle medium containing 10% v/v fetal bovine serum at 37°C in a humidified atmosphere with 5% CO 2 for 96 h. The medium was aspirated, and the cells were resuspended in 10 ml of ice-cold PBS, pH 7.4. Cells from two flasks were pooled for each cytosol preparation. The cells were centrifuged at 600 ϫ g for 5 min. The pellets were washed twice with 30 ml of ice-cold PBS and pelleted as above. The pellets were resuspended in 200 l of extract buffer (20 mM HEPES/KOH, pH 7.4, containing 10 mM KCl, 1.5 mM MgCl 2 , 1 mM dithiothreitol (DTT), 2 M sucrose, COMPLETE EDTA-free TM protease inhibitor mixture 1 tablet per 10 ml). The cells were homogenized with 30 passes of a mini homogenizer. The homogenates were centrifuged at 3000 rpm for 6 min at 4°C; the post-nuclear S1 supernatant was harvested; the pellet was resuspended in 100 l of extract buffer and homogenized as above; and the centrifugation step was repeated. The S1 supernatants were pooled and centrifuged at 13,000 ϫ g for 10 min at 4°C. The protein concentration of the resulting S2 supernatant was determined using a Coomassie G-250 assay (Pierce).
Two-dimensional Gel Electrophoresis-For analytical gels, cytosol containing 25 g of protein was made up to 300 l with 7 M urea, 2 M thiourea, 4% CHAPS, 30 mM DTT, and 0.2% v/v ampholyte, pH 3-10 (Bio-Rad). For preparative gels, 100 -500 g of cytosolic protein was used. Each 300-l sample was applied to a 17-cm, pH 3-10, immobilized pH gradient gel strip (IPG). The IPG strips were then rehydrated actively at 50 V for 16 h in a Bio-Rad Protean isoelectric focusing (IEF) cell. IEF consisted of 250 V for 15 min, linear ramping from 250 to 10,000 V over 3 h, followed by 10,000 for 60,000 V-h. The strips were then incubated at room temperature for 10 min in SDS-PAGE equilibration buffer (0.375 M Tris, pH 8.8, containing 6 M urea, 2% w/v SDS, 20% v/v glycerol) containing 2% w/v DTT followed by 10 min in SDS-PAGE equilibration buffer containing 2.5% w/v iodoacetamide. The strips were loaded onto 20 ϫ 18 cm 14% SDS-polyacrylamide gels, and overlaid with 0.5% low melting temperature agarose (Bio-Rad). Electrophoresis was performed in a Protean II xi electrophoresis cell (Bio-Rad). Analytical and preparative two-dimensional gels were stained with Owl silver stain and Biosafe Coomassie G-250 stain (Bio-Rad), respectively, according to the manufacturer's instructions. Molecular weight and pI values of proteins were estimated with two-dimensional standards (Bio-Rad).
Two-dimensional Gel Image Analysis-Analytical two-dimensional gels were scanned using a Powerlook III scanner (UMax, Ascot, UK) and analyzed using Phoretix two-dimensional software (Non-linear Dynamics, Newcastle, UK). A previously described pairwise approach (32) was used to compare the intensities of protein spots between cell clones. Here we compared single cell clones of NSC34 stable transfectants (27) expressing pCEP4 vector only with those expressing wild-type hSOD1 and NSC34 stable transfectants expressing pCEP4 vector only with those expressing G93A hSOD1. Comparisons of spot intensities were made between a pair of gels electrophoresed and stained at the same time using samples prepared within the same experiment. Differences in spot intensities were tested with Wilcoxon t test using no less than six pairs of gels.
Glutathione-Sepharose Precipitation-NSC34 cells grown in T175 flasks were resuspended in ice-cold PBS, pH 7.4. The cells were centrifuged at 600 ϫ g for 5 min. The pellets were washed twice with ice-cold PBS and pelleted as above. The resulting pellets were resuspended in 0.5 ml per flask of GST extract buffer (20 mM potassium phosphate buffer, pH 7.0, containing 0.1% v/v Triton X-100 and protease inhibitors). The cells were homogenized with 30 passes of a mini homogenizer and then centrifuged at 13,000 ϫ g for 10 min at 4°C. The protein concentrations of the resulting supernatants were determined using the above Coomassie G-250 assay. Appropriate volumes containing 2.5 mg of protein were adjusted to 1 ml with GST extract buffer and incubated with 100 l of 25% v/v glutathione-Sepharose, equilibrated in 20 mM potassium phosphate buffer, pH 7.0, for 60 min at 4°C with end-over rotation. The glutathione-Sepharose precipitates were pelleted by centrifugation at 13,000 ϫ g for 30 s and then washed 6 times with 1 ml of 20 mM potassium phosphate buffer, pH 7.0, containing 50 mM NaCl followed by 3 washes with 1 ml of 20 mM potassium phosphate buffer, pH 7.0. The pellets were thoroughly aspirated and resuspended each in 300 l of 7 M urea, 2 M thiourea, 4% CHAPS, 30 mM DTT, and 0.2% v/v ampholyte, pH 3-10. The resuspended pellets were incubated at room temperature for 10 min to denature and dissociate the glutathioneprotein complexes and then centrifuged at 13,000 ϫ g for 30 s. The resulting supernatants were subjected to two-dimensional electrophoresis as described above. The relative expression levels of glutathione-Sepharose binding proteins between cell lines were analyzed using silver staining and Wilcoxon t test as described above. All glutathione-Sepharose binding proteins detectable by Coomassie Blue staining were selected for identification by MALDI-TOF-MS.
Western Blotting of One-and Two-dimensional SDS-PAGE-Samples were Western-blotted as described previously (27) using antibodies diluted as indicated in the figure legends. The proteins were visualized using an ECL kit according to the manufacturer's instructions. To determine the positions of spots corresponding to endogenous mouse SOD1 (mSOD1) and G93A hSOD1, two-dimensional gels were performed in duplicate using 25 g of cytosolic protein. Both gels were transferred onto polyvinylidene difluoride membrane. One of the membranes was probed with 1:1000 primary sheep anti-SOD1 polyclonal followed by 1:2000 secondary donkey anti-sheep IgG horseradish peroxidase conjugate. Antibody binding was visualized with 3,3Ј-diaminobenzidine (DAB). Both membranes were stained with Coomassie R-250. Spot positions of endogenous mouse SOD1 and G93A human SOD1 were located by comparing the DAB and Coomassie-stained spot pattern.
Peptide Mass Fingerprinting by MALDI-TOF-MS and Data Base Searching-Protein identification of Coomassie Blue-stained spots with tryptic peptide mass fingerprinting by matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry (MALDI-TOF-MS) and data base searching was performed at the Aberdeen Proteome Facility (University of Aberdeen, UK). In-gel trypsin digestion of proteins of interest was performed using an Investigator ProGest work station (Genomic Solutions, Huntingdon, UK). MALDI-TOF-MS was performed using a Voyager-DE STR instrument (Perspective Biosystems). Data base searching was performed using the MS-Fit (University of California, San Francisco) and Mascot Software (Matrix Science, London, UK) to search the Swiss Prot/TrEMBL (www.ca.expasy.org) and NCBI (ncbi. nlm.nih.gov) data bases.
Enzyme and Metabolite Assays-Glutathione S-transferase assay reaction mixtures (33) consisted of 100 mM potassium phosphate, pH 6.5, containing 1 mM EDTA, 1 mM 1-chloro-2,4-dinitrobenzene, and 2 mM reduced glutathione. Reactions were initiated by adding 100 g of NSC34 post-nuclear S1 protein per ml assay reaction. The increase in absorbance at 340 nm was measured at 0 -5 min at room temperature.
Glutathione reductase assay reaction mixtures (34) consisted of 50 mM HEPES/KOH, pH 8.0, containing 0.1 mM EDTA, 0.1 mM oxidized glutathione (Sigma), and 0.1 mM NADPH. Reactions were initiated by adding 100 g of NSC34 post-nuclear S1 protein per ml of assay reaction. The decrease in absorbance at 340 nm was measured at 0 -5 min at room temperature.
Proteasome assays were performed with modification to the method of Beyette and co-workers (35). NSC34 cells grown in T175 flasks were resuspended ice-cold PBS, pH 7.4. The cells were centrifuged at 600 ϫ g for 5 min. The pellets were washed twice with ice-cold PBS and pelleted as above. The resulting pellets were resuspended in 0.3 ml of proteasome extract buffer (20 mM Tris/HCl, pH 7.4, containing 0.1 mM EDTA, 1 mM 2-mercaptoethanol, 5 mM ATP, 20% v/v glycerol, and 0.04% v/v Nonidet P-40). The resuspended cells were homogenized by 25 passes through a 21-gauge needle. The resulting homogenates were centrifuged at 13,000 ϫ g for 15 min at 4°C. The protein concentrations of the supernatants were determined using the Coomassie G-250 assay. Proteasome assay reaction mixtures consisted of 50 mM HEPES/KOH, pH 8.0, containing 5 mM EGTA, 100 g of NSC34 extract protein per ml of assay reaction. The reactions were initiated by adding the appropriate proteasome substrate and incubating the reactions at 37°C for 45 min. Suc-LLVY-AMC was used at 50 M. Z-ARR-AMC and Z-LLE-AMC were used at 100 M. Hydrolysis of the peptides was measured at 340 nm excitation and 460 nm emission for AMC using a Denley spectrofluorimeter. Standard curves were constructed using 0 -50 M AMC to convert fluorescence units to AMC concentration.
Reduced glutathione concentration in NSC34 cells was determined using a commercial assay kit (Calbiochem) according to the manufacturer's instructions.
RT-PCR Analysis of Gene Expression in NSC34 Cells and Human Motor Neurons-NSC34 cells expressing pCEP4 vector or G93A hSOD1 were harvested, washed in Hanks' buffered saline solution, and resuspended in TRIzol (Invitrogen). RNA was extracted according to the manufacturer's protocol. Following treatment with DNase I (Invitrogen), the sample was divided in two, and cDNA synthesis was performed both with and without the addition of Superscript II reverse transcriptase (Invitrogen), according to the manufacturer's instructions.
The Arcturus Pixcell II Laser Capture Microdissector was used to isolate motor neurons from 10 m toluidine blue-stained sections of human lumbar spinal cord. The material used was collected at autopsy from two neuropathologically normal control subjects (1 male and 1 female, mean age 50 Ϯ 3 years, post-mortem delay 16.5 Ϯ 2.5 h), and two subjects with FALS resulting from an I113T mutation in SOD1 (1 male and 1 female, mean age 62 Ϯ 2 years, post-mortem delay 23.5 Ϯ 3.5 h). RNA was extracted using TRIzol according to the manufacturer's instructions, except that 0.1 g/ml glycogen was added to assist in the recovery of RNA. As above, following DNase I treatment, samples were divided in two, and cDNA synthesis was carried out.
PCR was performed using 1 l of cDNA, 10 pmol of each primer (see Table I), 200 M total dNTPs, 0.5 units of Taq polymerase, and 1ϫ reaction buffer (75 mM Tris/HCl, pH 8.8, 20 mM (NH 4 ) 2 SO 4 , 1.5 mM MgCl 2 , 0.01% (v/v) Tween 20). NSC34 samples were denatured at 95°C for 5 min, followed by 30 cycles of 95°C for 15 s and 60°C for 1 min. Isolated human motor neurons required 35 cycles of amplification due to limited amounts of starting material. PCR products were electrophoresed on 3% agarose, and relative band intensities were determined by densitometry. The relative level of expression of each of the genes of interest was calculated relative to that of actin in the same sample.
Immunohistochemical Analysis of Motor Neurons-Serial sections were cut from paraffin-embedded lumbar spinal cords from normal neurological controls and sporadic ALS cases. Following de-waxing, immunohistochemistry was performed using standard techniques. Antigen retrieval (microwave for 10 min in 0.01 M citrate buffer, pH 6.0) was used prior to primary antibody incubation. Primary rabbit polyclonal antibodies to GST Mu, LMP7, and ASS were used at a dilution of 1:200 and that to LTB 4 12HD was used at 1:50. Incubations were performed overnight at 4°C. Staining was performed using the Vector ABC technique with DAB, except for LTB 4 12HD staining, which was performed using the Envision kit (Dako). Sections treated identically, except for omission of the primary antibody, were used as negative controls. The pattern of staining with each antibody was recorded using light microscopic examination.

Two-dimensional Gel Analysis and Identification of Differentially Expressed Proteins in NSC34
Cytosol-Cytosol from NSC34 cells stably transfected with pCEP4 vector only, wildtype hSOD1, and G93A hSOD1 was separated by two-dimensional gel electrophoresis. Representative silver-stained twodimensional gels from cells stably transfected with vector only and cells expressing G93A hSOD1 are shown in Fig. 1. Typically, up to 700 individual protein spots were detected in NCS34 cytosol within the 10 -100-kDa size range using pH 3-10 IPG strips. The positions of endogenous mouse SOD1 and G93A hSOD1 determined as described are shown with horizontal arrows (Fig. 1, A and B). Several landmark spots were identified as follows. Spots 1 and 2 (Fig. 1A) were identified as nucleotide diphosphate kinase A by Western blotting (results not shown). Spots 3-6 ( Fig. 1B) were all identified as cyclophilin A (cyph A) (Swiss-Prot accession number P17742) with 75, 68, 57, and 74% sequence coverage, respectively, by MALDI-TOF-MS analysis. Spot 7 (Fig. 1B) was identified as glyceraldehyde-3-phosphate dehydrogenase (Swiss-Prot accession number P16858) with 54% sequence coverage by MALDI-TOF-MS analysis.
To detect changes in cytosolic protein expression due to G93A hSOD1 expression, two-dimensional gels of cytosol from NSC34 cells transfected with pCEP4 vector only were compared with those of cytosol from NSC34 cells expressing G93A hSOD1 in a pairwise fashion. To control for changes resulting from hSOD1 expression and not the G93A amino acid change, two-dimensional gels of cytosol from NSC34 cells transfected with pCEP4 vector only were compared with those of cytosol from NSC34 cells expressing wild-type hSOD1. We detected 4 spots reduced (spots D1, D2, D3, and D4, Fig. 1A, Fig. 2, and Table II) and 4 spots increased (spots U1, U2, U3, and U4, Fig.  1B, Fig. 2, and Table II) in intensity in the G93A hSOD1 gels compared with pCEP4 vector only gels. Only one of these 8 spot changes (Fig. 2, spot D1) was found when pCEP4 vector only gels were compared with wild-type hSOD1 gels indicating that the other 7 spot changes were specific to G93A hSOD1 expression. This was the only change detected comparing pCEP4 vector only gels with wild-type hSOD1 gels; therefore, no spot changes were detected due to wild-type hSOD1 expression that were not seen due to G93A hSOD1 expression.
For identification of the differentially displayed spots by MALDI-TOF-MS, Coomassie Blue-stained two-dimensional preparative gels loaded with 100 -500 g of cytosolic protein were used. Only 6 of the 8 differentially displayed spots detected by silver staining were detected by Coomassie Blue staining even at the highest protein loads (spots D2, D3, D4, U2, U3, and U4). This prevented the detection and subsequent identification of the remaining 2 spots (spots D1 and U1) by MALDI-TOF-MS. The 6 Coomassie Blue-stained differentially displayed spots were identified by MALDI-TOF-MS analysis (Table III). Spot D2 that was reduced 2.3-fold in intensity in G93A hSOD1 cells (p Ͻ 0.02) ( Fig. 2 and Table II) was positively identified as glutathione S-transferase Mu 1 (GST Mu 1) (Swiss-Prot accession number P10649). Spots D3 and D4 that were undetectable in G93A hSOD1 cells (p Ͻ 0.02) ( Fig. 2 and Table II) were positively identified as hypothetical protein 251000C21Rik protein (Swiss-Prot accession number Q9CPS1) and 20 S proteasome ␤5i subunit (LMP7) (Swiss-Prot accession number P28063), respectively (Table III). The hypothetical 251000C21Rik protein sequence derived from the mouse RIKEN cDNA clone 251000C21RIK deposited in the Functional Annotation of Mouse (FANTOM) data base (36), at www.gsc.riken.go.jp/e/FANTOM, was BLAST searched. The three highest scoring protein sequences were porcine, human, and rabbit NADP-dependent leukotriene B 4 12-hydroxydehydrogenase (LTB 4 12HD), with 82.4, 79.7, and 76.9% sequence identity, respectively. This suggested that 251000C21Rik protein represented a mouse homologue of LTB 4 12HD.
Spots U2 and U4 that were increased 1.9-(p Ͻ 0.02) and 1.8-fold (p Ͻ 0.05) in intensity in G93A hSOD1 cells ( Fig. 2 and Table II) were positively identified as argininosuccinate synthase (ASS) (Swiss-Prot accession number P16460) and peroxiredoxin I (Prx I) (Swiss-Prot accession number P35700), respectively (Table III). The observed pI of Prx I (6.9) was more acidic than the theoretical value (8.26) (Table III). To investigate whether Prx I occupied an additional spot position, MALDI-TOF-MS analysis was performed on spot 8 (Fig. 1B) that had an apparent molecular mass of 25.5 kDa like spot U4, but with an apparent pI of 8.2. The resulting tryptic peptides exhibited 79% sequence coverage for Prx I, confirming that the protein also occupied a more basic position that matched the theoretical pI. MALDI-TOF-MS analysis of spot U3, prepared from cells expressing G93A hSOD1, detected tryptic peptides that exhibited 44 (9 matched peptides) and 42% (7 matched peptides) sequence coverage for cyph A (Swiss-Prot accession number P17742) and RNA-binding motif protein 3 (Rbm3) (Swiss-Prot accession number O89086), respectively (Table III and Fig. 3B). MALDI-TOF-MS analysis of spot U3 prepared from cells expressing pCEP4 vector alone detected tryptic peptides that exhibited 50 (9 matched peptides) and 22% (3  (Fig. 3A). As none of the matched peptide masses were shared between the two proteins, it was concluded that spot U3 contained both cyph A and Rbm3. The number and intensities of peptides matched to Rbm3 were significantly increased in spot U3 from cells expressing G93A hSOD1 (Fig.  3B) compared with those expressing pCEP4 vector only (Fig.  3A). In contrast the number and intensities of peptides matched to cyph A were similar between both cell lines (Fig. 3,  A and B). It was concluded that the increased intensity of spot U3 in cells expressing G93A hSOD1 was due to alteration of Rbm3 levels in the cytosol. We have also demonstrated that as well as spot U3, cyph A occupies spot positions 3-6 with pI values of 6.5, 6.7, 6.9, and 7.4 respectively (Fig. 1B and Table  III). Western Blotting of NSC34 Cytosol with Antisera to Identified Proteins-To investigate whether the cytosolic steadystate levels of these proteins were altered in the presence of G37R hSOD1 as well as G93A hSOD1, Western blotting analysis was performed on NSC34 cytosol with antisera to identi-fied proteins where available. Faint 46-kDa bands corresponding to ASS (Fig. 4A) were detected in cytosol from cells expressing pCEP4 vector only and wild-type hSOD1 (lanes 1  and 3), whereas strong bands corresponding to ASS (Fig. 4A) were detected in cytosol from cells expressing both G93A and G37R hSOD1 (lanes 2 and 4). The antiserum to GST Mu (Fig.  4B) detected a strong 25-kDa band corresponding to GST Mu in cytosol from cells expressing pCEP4 vector only and wild-type hSOD1 (lanes 1 and 3), whereas this band was reduced in intensity in cytosol from cells expressing both G93A and G37R hSOD1 (lanes 2 and 4). The antiserum to LTB 4 12HD (Fig. 4C) detected a strong 35-kDa band corresponding to LTB 4 12HD in cytosol from cells expressing pCEP4 vector only and wild-type hSOD1 (lanes 1 and 3). This band was undetectable in cytosol from cells expressing both G93A and G37R hSOD1 (lanes 2 and  4). This provided further evidence that the 251000C21Rik protein is the murine homologue of LTB 4 12HD. The antiserum to LMP7 (Fig. 4D) detected a strong 23-kDa band corresponding to LMP7 in cytosol from cells expressing pCEP4 vector only (lane 1). This band was slightly reduced in intensity in the presence of wild-type hSOD1 (lane 3). In marked contrast this band was weakly detected in cytosol from cells expressing both G93A and G37R (lanes 2 and 4). These results demonstrate that the cytosolic steady-state levels of ASS, LMP7, GST Mu, and LTB 4 12HD are altered in G37R hSOD1-expressing cells as well as G93A hSOD1 when compared with control cells expressing wild-type hSOD1 or pCEP4 vector only. Although antisera raised to Rbm3 and Prx I were not available to us, we have observed similar increases in intensities, to those seen in G93A hSOD1 cells, of the corresponding spots (U3 and U4, respectively) in cytosol from G37R hSOD1 cells on silverstained two-dimensional gels when compared with control cytosol (results not shown).
Argininosuccinate synthase (ASS) and argininosuccinate lyase (ASL) function to regenerate a pool of intracellular arginine that is dedicated to the synthesis of nitric oxide via the action of nitric-oxide synthase (NOS) (37,38). As steady-state levels of ASS were increased in cytosol from cells expressing G93A and G37R hSOD1, the effects of mutant SOD1 expression on the levels of other enzymes involved in arginine and nitric oxide cycling were determined. Western blots of NSC34 cytosol were probed with antisera to ASL, neuronal NOS (nNOS), endothelial NOS, and inducible NOS. There were significant increases in the intensities of both ASL bands (Fig. 4E) and nNOS bands (Fig. 4F) in the cytosol from cells expressing G93A and G37R hSOD1 (lanes 2 and 4) compared with those in the cytosol from cells expressing pCEP4 vector only and wildtype hSOD1 (lanes 1 and 3). Neither endothelial NOS nor inducible NOS were detected in any of the cytosol samples by Western blotting (results not shown). As well as an increase in the intensity of ASL bands in the cytosol from cells expressing G93A and G37R hSOD1 compared with control cells (Fig. 4E), we also observed an alteration in the pattern of ASL bands. An ASL doublet was present in cytosol from cells expressing pCEP4 vector only or wild-type hSOD1 (lanes 1 and 3), whereas an ASL triplet was present in cytosol from cells expressing G93A or G37R hSOD1 (lanes 2 and 4). This result indicated that mutant SOD1 expression resulted in either differential post-transcriptional or post-translational processing of the ASL mRNA or protein, respectively. The nature of this mutant hSOD1-dependent alteration to ASL expression is currently under investigation.
Alterations to Glutathione S-transferase Family Members-As GST Mu 1 expression was reduced due to mutant hSOD1 expression, we further investigated the expression and function of other GST family members expressed in NSC34 cells. The broad range GST substrate 1-chloro-2,4-dinitrobenzene was used to compare the overall GST activity of the NSC34 cell lines. The GST activity within NSC34 cell extracts was significantly reduced to ϳ60% of normal (p Ͻ 0.05) in cells expressing G93A and G37R hSOD1 but not in those expressing the wild-type enzyme (Fig. 5A). In contrast, glutathione reductase activity was not significantly different in any of the cell lines (Fig. 5B). The levels of reduced glutathione in the NSC34 cells were also measured and found to be significantly reduced to ϳ75% of normal (p Ͻ 0.05) in cells expressing G93A and G37R hSOD1 but not in those expressing the wild-type enzyme (Fig. 5C).
It was reasoned to be unlikely that reduced GST Mu 1 expression would be solely responsible for the dramatic reduction of glutathione conjugating capacity of the mutant hSOD1 expressing cell lines toward 1-chloro 2,4-dinitrobenzene. A more global loss of GST family members may have accounted for this markedly decreased activity. To investigate this further, detergent extracts of NSC34 cells expressing either pCEP4 vector only or G93A hSOD1 were prepared. These extracts were precipitated with glutathione-Sepharose, and the resulting precipitates were subjected to analytical two-dimensional gel electrophoresis to compare the relative expression levels of various GST sub-types between the two cell lines. Out of 11 spots detectable by Coomassie Blue staining (Fig. 5, D and  E), 10 were positively identified by MALDI-TOF-MS (Table  III). Spot E that was significantly reduced (p Ͻ 0.05) in cells   5 and Table IV). The most abundant GST precipitable by glutathione-Sepharose from NSC34 cells was GST Pi B that occupied 5 individual spot locations (spots F-J). Four of the GST Pi B spots were significantly less abundant (p Ͻ 0.05) in cells expressing G93A hSOD1 (Table IV and Fig. 5, D and E). In contrast to GST Mu and GST Pi, GST Alpha 2 (spot A) (Swiss-Prot accession number P10648) was found to be significantly increased (p Ͻ 0.05) in cells expressing G93A hSOD1 (Table IV and Fig. 5, D and E). Alterations to Proteasome Activity and Proteasome Subunit Expression-Up-regulation of the LMP7 proteasome subunit is known to promote cleavage after hydrophobic (chymotrypsinlike activity) and basic (trypsin-like activity) residues and suppress cleavage after acidic residues (postglutamyl cleavages) (39). As LMP7 expression was significantly reduced due to G93A and G37R hSOD1 expression, the effects of these mutant forms of SOD1 on proteasomal chymotrypsin-like activity, trypsin-like activity, and the postglutamyl hydrolase activity were investigated in the NSC34 cell lines using the model fluorogenic peptides Suc-LLVY-AMC, Z-ARR-AMC, and Z-LLE-AMC, respectively (Fig. 6, A-C). All three activities were significantly reduced (p Ͻ 0.02) in the presence of G93A and G37R hSOD1 expression, with chymotrypsin-like activity showing the most marked reduction to ϳ70% of normal. There was no significant change in chymotrypsin-like activity (Fig. 6A), but there were small reductions in trypsin-like (Fig. 6B) and postglutamyl hydrolase activity (Fig. 6C) in the presence of wildtype hSOD1 expression.
As there was significant reduction in the chymotrypsin-like activity (Fig. 6A) and the cytosolic levels of the inducible immunoproteasome subunit LMP7 (␤5i) due to mutant hSOD1 expression (Fig. 4), we investigated whether there were alterations in the level of expression of the constitutive ␤5 proteasome subunit, subunit X, and the other inducible immunoproteasome subunit LMP2 (␤1i). Western blotting of the proteasome assay extracts with anti-LMP7 serum (Fig. 6D) confirmed the dramatic reduction of LMP7 expression (Fig. 4) in cytosol of cells expressing G93A and G37R hSOD1 (24 and 14%, respectively) and a less dramatic reduction in cells expressing wild-type hSOD1 (66%). In the case of the constitutive subunit (Fig. 6E), Western blotting detected a significant increase (p Ͻ 0.05) in expression of subunit X (␤5) in cells expressing G93A and G37R hSOD1 (142 and 158%, respectively) that mirrored the decrease in LMP7. Paradoxically, this coincided with decreased postglutamyl hydrolase activity (Fig. 6C). We detected no significant change in the level of LMP2 expression (Fig. 6F) in cells expressing wild-type or G37R hSOD1, but we observed a small reduction (p Ͻ 0.05) in LMP2 expression in the cells expressing G93A hSOD1 (Fig. 6F). As there were substantial amounts of LMP2 remaining in the presence of hSOD1 expression (Fig. 6F), this probably accounts for the remaining chymotrypsin-like activity toward Suc-LLVY-AMC (Fig. 6A).
Alterations to Expression of Antioxidant Enzymes, Proteasome, and Nitric Oxide-related Enzymes of Motor Neurons from ALS Cases-To confirm whether the protein changes observed for GST Mu 1, LMP7, nNOS, ASS, and LTB 4 12-HD were accompanied by alterations in the expression levels of the corresponding mRNAs, RT-PCR analysis was performed on NSC34 cells expressing pCEP4 vector or G93A hSOD1 (Fig.  7A). Expression levels of ASS, nNOS, LMP7, and LTB 4 12-HD, but not GST Mu 1, were changed significantly (p Ͻ 0.05) due to G93A hSOD1 in NSC34 cells. To investigate whether the expression levels of these differentially regulated proteins were affected in motor neurons from human FALS cases, we performed RT-PCR analysis of gene expression in laser capture microdissected motor neurons from normal individuals or indi- FIG. 5. Alterations to glutathione handling due to ALS mutant SOD1 in NSC34 cells. A-C, assays. Lysates prepared from NSC34 cells stably transfected with pCEP4 vector only (white bars), expressing wild-type hSOD1 (black bars), expressing G37R hSOD1 (light gray bars), or expressing G93A hSOD1 (dark gray bars) were assayed for glutathione S-transferase activity (A), glutathione reductase activity (B), and reduced glutathione concentration (C). For the enzyme activities (A and B), assays were performed in triplicate on lysates extracted on 4 separate occasions. Specific activities are expressed as percentage of pCEP4 vector control. For reduced glutathione determinations (C), measurements were performed in triplicate on lysates extracted on 3 separate occasions. Reduced glutathione concentrations are expressed as nanomoles of GSH/1 ϫ 10 6 cells. Twodimensional gel analysis (D and E), lysates containing 2.5 mg of protein from NSC34 cells stably transfected with pCEP4 vector only (D) or expressing G93A hSOD1 (E) were precipitated with glutathione-Sepharose. The resulting precipitates were subjected to IEF on pH 3-10 IPG strips followed by SDS-PAGE on 14% polyacrylamide gels. Gels stained with Coomassie Blue are shown. viduals with FALS attributed to the I113T hSOD1 mutation (Fig. 7A). The expression levels of GST Mu 1, LMP7, and LTB 4 12-HD were significantly reduced (p Ͻ 0.05) (Fig. 7A) in FALS cases compared with normal human controls. In contrast, no significant changes were detected for ASS and nNOS expression (Fig. 7A).
We set out to investigate the cellular distribution of GST Mu, LMP7, ASS, and LTB 4 12-HD protein expression in normal human spinal cord. Immunohistochemical staining was performed on sections cut from paraffin-embedded lumbar spinal cords from normal individuals along with tissue from SALS cases (Fig. 7B) as a comparison. GST Mu, LMP7, and ASS were diffusely expressed in white and gray matter of the spinal cord. These enzymes were strongly expressed in the cytoplasm and in neurites of motor neurons. The intensity of neuronal cytoplasmic staining tended to be less for all three proteins in the SALS cases (Fig. 7B). Neuronal NOS has already been shown to be up-regulated at the protein level in human motor neurons from ALS cases (40). LTB 4 12-HD staining was only weakly detected in the neuronal cytoplasm in both SALS cases and controls. Interestingly, LTB 4 12-HD staining was observed in the nuclei of motor neurons and glia in SALS cases and controls.

DISCUSSION
Several recent studies (24 -26) employing genomic profiling technologies to identify gene expression changes in ALS have concentrated on whole spinal cord tissue from human ALS cases and transgenic mouse models of FALS. However, due to the low proportion of motor neurons compared with other cell types in spinal cord, the primary responses of motor neurons to mutant SOD1 toxicity that trigger cell death pathways may go undetected among the overall tissue transcriptional changes responding to motor neuron degeneration. To this end we have exploited a well characterized cell culture model of mutant SOD1-related FALS (27)(28)(29) to enable us to analyze the proteome changes that occur as a direct result of SOD1 toxicity in cells with a motor neuron phenotype.
By using a combination of two-dimensional electrophoresis, mass spectrometry, and Western blotting, we identified seven up-regulated proteins as ASS, ASL, nNOS, Rbm 3, Prx I, subunit X, and GST Alpha 2. Seven down-regulated proteins were identified as GST Mu 1, GST Mu 2, GST Mu 5, a hypothetical GST Mu homologue, GST Pi B, LTB 4 12HD, and LMP7. We also demonstrated that the mRNA expression levels of GST Mu 1, LTB 4 12HD, and LMP7 were similarly changed in motor neurons isolated from FALS cases. In the case of GST Mu 1 in the cell culture model, and nNOS in the model and isolated motor neurons, the protein changes were not reflected by the mRNA levels as determined by RT-PCR. This was probably due to the semi-quantitative nature of the RT-PCR technique, as recent studies in our laboratory employing the Affymetrix Murine Genome oligonucleotide array have demonstrated 2.3-fold up-regulation and 7.6-fold down-regulation of nNOS and GST Mu 1 mRNA respectively, 2 due to expression of G93A hSOD1 in NSC34 cells. Surprisingly, only one protein change (spot D1) was detected due to normal hSOD1 expression rather than the G93A mutant SOD1 protein. Further changes may have been detected using narrower pH ranges for the first dimension IEF along with a staining technique with a wider range of linearity than silver. However, it is of interest that our results using the Affymetrix oligonucleotide array also reveal only minor effects of wild-type hSOD1 on the mRNA expression profile of NSC34 cells compared with those of G93A hSOD1. 2 The differentially regulated proteins fall into four categories: (i) proteins involved in regulation of mRNA processing (Rbm3); (ii) proteins involved in NO metabolism (ASS, ASL, and nNOS); (iii) proteins involved in anti-oxidant defense (Alpha, Mu, and Pi class GSTs, LTB 4 12HD, and Prx I); and (iv) proteins involved in protein degradation (LMP7, subunit X). Our results provide further evidence for mutant SOD1-mediated alterations in the intracellular redox state and protein degradation machinery, which in turn supports the hypothesis that both altered free radical handling and abnormal protein aggregation are likely to be mechanisms contributing to motor neuron injury.
Rbm3 was shown to have elevated cytosolic levels in the presence of G93A hSOD1 expression. To date the exact function of this protein is unclear. It is a heterogeneous nuclear ribonucleoprotein that contains an N-terminal consensus sequence RNA binding domain and a C-terminal glycine-rich domain. Proteins with such domains have been shown to regulate mRNA stability and translation, mRNA splicing, and export of mRNA from the nucleus to the cytoplasm (41). The raised levels of this heterogeneous nuclear ribonucleoprotein in the cytosol in the presence of G93A hSOD1 expression may indicate alterations to protein biogenesis at the level of post-transcriptional mRNA processing and/or mRNA translation.
ASS acts in conjunction with ASL to regenerate arginine from citrulline for the purpose of nitric oxide production by NOS (37,38). As with ASS, we found that both ASL and nNOS were up-regulated by G93A and G37R hSOD1. Additionally, ASL was shown to undergo mutant hSOD1-specific alterations to either its post-transcriptional or post-translational processing. The human ASL gene product has been shown previously to undergo highly variable splicing (42). The nature of the alteration here to the mouse ASL gene product is currently under investigation. By using microelectrode biosensor measurements, we have determined previously that NSC34 cells expressing wild-type hSOD1 exhibit enhanced NO release, whereas those expressing ALS mutant hSOD1 exhibit reduced NO release following cell stress induced by serum withdrawal (29). Both groups of cells expressing wild-type or mutant hSOD1 show decreased super-oxide release in the same experimental paradigm. In addition, the mutant SOD1-expressing NSC34 cells are more sensitive to apoptosis stimulated by NO-releasing compounds (29). The mechanism by which NO release is reduced in the mutant SOD1-expressing NSC34 cells has yet to be determined. The reduction may result in a compensatory response by the cells in the form of up-regulation of the arginine/NO recycling pathway. The role of NO in ALS pathogenesis remains controversial due to its dual role as a neuroprotective and neurotoxic agent. At low concentrations NO can protect cells against oxidative stress, presumably by induction of adaptive responses in the form of up-regulation of antioxidant proteins (43,44). In contrast, inhibition of nNOS has been shown to protect rat motor neurons from cell death induced by oxidative stress (45). The findings that inhibition of nNOS in a cell culture model of FALS did not confer resistance to mutant SOD1 toxicity (46) and that blockade of NOS either by chemical means (47) or targeted deletion of the nNOS gene (48) had little effect on disease progression in a transgenic mouse model of human ALS have raised uncertainties regarding the importance of NO production in ALS pathogenesis. However, the relevance of our findings here to ALS is reinforced by studies reporting nNOS up-regulation in human ALS cases (40,49) and inducible NOS up-regulation in transgenic mice expressing ALS hSOD1 mutants (50). Furthermore, increased levels of nitrotyrosine have been shown in human cerebrospinal fluid in sporadic ALS (SALS) (51) as well the spinal cord of a transgenic mouse model (15). We anticipate that exploitation of recently developed "nitroproteomic" techniques (52, 53) will eventually clarify the contribution of altered nitric oxide metabolism to ALS.
Expression of mutant hSOD1 resulted in a complete loss of the LTB 4 12HD protein from NSC34 cells. LTB 4 12HD may have a wider role in antioxidant defense as well as in lipid messenger metabolism. A recent study (54) demonstrated that LTB 4 12HD was effective at reducing a wide variety of cytotoxic ␣,␤-unsaturated aldehydes and ketones including products of lipid peroxidation such as 4-hydroxy-2-nonenal (HNE). HNE has been shown to be elevated in the cerebrospinal fluid of SALS cases (55) and SALS spinal cord (56,57). A potential chemoprotective role for the enzyme was suggested by its ability to confer resistance of LTB 4 12HD-transfected cell lines to HNE-induced apoptosis (54). The association of a significant amount of LTB 4 12HD with membranes (54) and the nuclear associated staining of motor neurons in this study supports the notion that the enzyme may help to protect membranes from oxidative damage.
Other antioxidant enzymes that were down-regulated were GST Mu 1, GST Mu 2, GST Mu 5, GST Pi B, and a hypothetical Mu class GST Rik061005A07. The GST family catalyzes the conjugation of reduced glutathione to toxic compounds to allow their elimination from cells by glutathione conjugate transporters (58). Here we demonstrated this function of GST to be significantly impaired in G93A and G37R hSOD1-expressing NSC34 cells. Previously, human neuroblastoma cells overexpressing mouse GST Mu showed increased resistance to cell death induced by hydrogen peroxide, peroxynitrite, and HNE (59). Human GST Mu when added to rat neuronal cultures protected against HNE cytotoxicity (60). Additional to their detoxification role, GST family members have also been shown to regulate apoptotic pathways. Mouse GST Mu has been shown recently to interact physically with and inhibit apoptosis signal-regulating kinase 1 (ASK1) that is known to activate the stress-activated protein kinase/c-Jun N-terminal kinase and p38 pathways (61).
Prx I was shown to occupy at least two spot positions on our two-dimensional gels. The most abundant of the two forms, which was unchanged on our gels, had an apparent pI (8.2) in close agreement with the theoretical value (8.26) for Prx I. The less abundant form, which was up-regulated in the presence of mutant SOD1 expression, had a more acidic pI of 6.9. Prx I, II, and III (62,63) have all been shown to undergo acidic shifts when cultured cells are subjected to hydroperoxide-mediated oxidative stress. A recent study (64) has demonstrated that the acidic shift of Prx II produced under conditions of oxidative stress was due to conversion of an essential active site cysteine to cysteic acid. The possibility that Prx I undergoes over-oxidation of its cysteine residues due to mutant SOD1 expression FIG. 7. RT-PCR and immunohistochemical analysis of human motor neurons. RT-PCR (A), relative expression levels of ASS, nNOS, GST Mu 1, LMP7, and LTB 4 12-HD in NSC34 cells expressing pCEP4 vector (white bars) or G93A hSOD1 (black bars), and in laser captured human motor neurons from normal controls (white bars) or I113T SOD1 FALS cases (black bars) were measured by densitometry of RT-PCR products generated on at least 5 separate occasions. The levels are expressed relative to that of actin. B, immunohistochemistry. Human lumbar spinal cord sections from normal neurological controls and sporadic ALS cases were stained using antisera raised to ASS, GST Mu, LMP7, and LTB 4 12-HD. Cytoplasmic staining in anterior horn motor neurons is indicated with an asterisk. Focal granular brown staining in the cytoplasm of neurons from SALS cases represents lipofuscin, indicated with a chevron. Nuclear staining of LTB 4 12-HD, in anterior horn motor neurons and glia, is indicated with large and small arrows, respectively. Black bar indicates 100 m. therefore warrants further investigation. All of the antioxidant proteins shown here to be down-regulated (GST family and LTB 4 12HD) or post-translationally modified (Prx I) have been shown previously to either contain the antioxidant-response element in their genes, undergo nuclear factor E2 p45-related factor (Nrf2)-dependent transcriptional regulation via antioxidant-response element binding, and/or undergo up-regulation by chemical agents known to activate Nrf2 (65,66). It is tempting to speculate that this down-regulation is due to alterations in Nrf2 status. The downregulation of genes encoding antioxidant proteins is likely to contribute significantly to the increased sensitivity of the mutant SOD1-expressing NSC34 cells to oxidative stress (27). Identification of the primary alteration to SOD1 biochemistry that drives this transcriptional down-regulation of antioxidant proteins may prove to be insightful in determining how the mutant enzyme triggers cell death in motor neurons.
The expression of both G93A and G37R hSOD1 resulted in a significant reduction in amount of the LMP7 proteasome subunit (␤5i) that coincided with a reduction in chymotrypsin-like activity and an increase in the amount of subunit X (␤5). The 20 S proteasome can contain three interferon-␥-inducible ␤ subunits LMP7, LMP2, and MECL that are homologous and interchangeable with three constitutive ␤ subunits X, Y, and Z, respectively (39,(67)(68)(69). It is likely that interferon-␥-inducible ␤ subunits favor the production of peptides with basic or hydrophobic C termini that are most suited to transport into the endoplasmic reticulum via the transporter associated with antigen presentation prior to major histocompatibility complex class I presentation. Several studies have reported proteasome activity and subunit content alteration in response to diseaseassociated protein aggregation and/or oxidative stress (70 -73). SOD1-containing aggregates have been observed in both transgenic mouse (19) and cell culture models of FALS (20,21). Similarly ubiquitinated protein aggregates have been observed in human SALS (74). The down-regulation of LMP7 and upregulation of its displacing partner, subunit X, may suggest an adaptive response to a challenge to the proteasome resulting from ALS mutant SOD1 expression. The observed reduction in proteasome activity may contribute over time to the abnormal aggregation of intracellular proteins observed in motor neurons, in cellular and animal models as well as human disease.
In conclusion we have used proteomic approaches to identify protein alterations in a cell culture model of SOD1-related FALS. The novel approach in this study is that we have applied proteomic technology to motor neuronal cells expressing mutant SOD1 at levels approximating those seen in human FALS. Several of the protein changes were corroborated at the transcriptional level in motor neurons isolated from human FALS cases. The proteome alterations we have identified are accompanied by functional consequences and may contribute to or represent responses to cellular changes that ultimately trigger cell death pathways resulting in neurodegeneration. These protein changes provide further evidence for the altered free radical handling and protein aggregation hypotheses to explain the toxic gain-of-function of FALS-associated mutant SOD1.