Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Oxidative Stress and Motor Neurone Disease

Oxidative Stress and Motor Neurone Disease Mark R. Cookson and Pamela J. Shaw. Department of Neurology, University of Newcastle upon Tyne, UK The effects of oxidative stress within post mitotic cells such as neurones may be cumulative, and injury by free radical species is a major potential cause of the age-related deterioration in neuronal function seen in several neurodegenerative diseases. There is strong evidence that oxidative stress plays an important role in the pathogenesis of motor neurone disease (MND). Point mutations in the antioxidant enzyme Cu,Zn superoxide dismutase (SOD1) are found in some pedigrees with the familial form of MND. How mutations in this ubiquitous enzyme cause the relatively selective cell death of specific groups of motor neurones is not clear, although a number of hypotheses have been forwarded. These include (1) the formation of hydroxyl radicals, (2) the catalysis of reactions of the nitrogen centred oxidant species peroxynitrite, (3) toxicity of copper or zinc and (4) protein aggregation. Some experimental support for these different hypotheses has been produced by manipulating cells in culture to express the mutant SOD1 proteins and by generating transgenic mice which over-express mutant SOD1. Observations in these model systems are, in some cases at least, supported by observations made on pathological material from patients with similar SOD1 mutations. Furthermore, there are reports of evidence of free radical mediated damage to neurones in the sporadic form of MND. Several lines of evidence suggest that alterations in the glutamatergic neurotransmitter system may also play a key role in the injury to motor neurones in sporadic MND. There are several important subcellular tar- gets, which may be preferentially impaired within motor neurones, including neurofilament proteins and mitochondria. Future research will need to identify the aspects of the molecular and physiological phenotype of human motor neurones that makes them susceptible to degeneration in MND, and to identify those genetic and environmental factors which combine to cause this disease in individuals and in familial pedigrees. Introduction Corresponding author: Pamela J. Shaw, Department of Neurology, University of Newcastle Upon Tyne, Ward 11, Royal Victoria Infirmary, Newcastle upon Tyne, NE1 4LP, UK; Tel.: +44 191 2325131; Fax: +44 191 261 0881; E-mail: pamela.shaw@ncl.ac.uk Motor neurone disease (MND, also referred to as amyotrophic lateral sclerosis) is one of the commonest adult-onset neurodegenerative disorders with a world wide incidence of 1-2 per 100,000. The most prominent pathological features in MND are the degeneration of lower motor neurones in the ventral horn of the spinal cord and brainstem and of upper motor neurones in the motor cortex which give rise to the corticospinal tract. The selective vulnerability of motor neurones is relative and clinical and pathological evidence indicating involvement outside the motor system is commonly seen in MND (Fig. 1). Thus MND can be regarded as a multi-system disease in which the motor system is affected earliest and most severely. MND is familial in 10 percent and sporadic in approximately 90 percent of cases. The disease typically progresses over several years and inevitably results in death, usually due to respiratory insufficiency. Current therapies have only a limited effect on clinical and pathological progression. The primary pathogenetic processes underlying motor neurone injury in MND are likely to be multifactorial and the precise molecular pathways leading to the cell death of motor neurones are at present unknown. In recent years, two particularly important research findings have led to progress in elucidating the cellular pathophysiology underlying MND. Firstly, in 1993, Rosen et al. reported the association of point mutations in the gene for Cu/Zn superoxide dismutase (SOD1) with some familial cases of MND (107). In view of the clinical and pathological similarities between the famil- rones, oxidative stress represents an important potential cause of the age-related deterioration in cellular function observed in adult-onset neurodegenerative diseases. There is particular interest in the role of oxidative stress in MND because genetically determined alterations in the free radical defence system caused by mutations in the SOD1 gene underlie 20 percent of cases of familial MND or two percent of cases of MND as a whole (101, 107). Cu/Zn Superoxide Dismutase (SOD1) Mutations Figure 1. Spinal cord from a case of MND showing degeneration in the posterior column sensory pathways (arrows) in addition to the motor pathways. Luxol fast blue/cresyl violet stain. Bar = 2 mm. ial and sporadic forms of MND, it is likely that these two subgroups share common pathways of cellular injury. Secondly, in 1995, Rothstein and colleagues showed that there is a specific loss of the glial glutamate transporter protein EAAT2 in pathologically affected areas of the CNS in sporadic MND patients (110). These two studies have been instrumental in the development of two major hypotheses for factors contributing to motor neurone injury in MND, namely oxidative stress and glutamate mediated toxicity (excitotoxicity). Both of these studies (107, 110) have proved to be robust, in that the findings have been replicated by other research groups. Over 60 mutations in the SOD1 gene have been found by different groups (101), and reduced expression of the EAAT2 protein in the spinal cord of MND cases has been confirmed by Fray et al (53). In this review we will discuss the normal biochemistry of SOD1 and the current state of knowledge regarding the cellular effects of the mutant SOD1 protein. The evidence that oxidative stress may contribute to motor neurone injury in the sporadic form of MND will be reviewed and the candidate subcellular targets for injury within motor neurones will be highlighted. Finally the potential links between oxidative stress and other mechanisms of motor neurone injury in MND will be discussed. Oxidative Stress Free radicals are a potential source of damage to DNA, lipids, membranes and proteins within cells. Any imbalance between the intracellular production of free radical species and anti-oxidant defence mechanisms or free radical clearance systems results in a state of oxidative stress. In relation to post-mitotic cells such as neu- SOD1 is a metalloenzyme comprising 153 amino acids with copper and zinc binding sites (Fig. 2). The enzyme functions as a homodimer, with a copper ion, essential for the dismutation reaction, held in position at the base of the active channel by four histidine residues. SOD1 is a cytosolic enzyme that is ubiquitous in distribution. There are two related superoxide dismutase enzymes. Manganese SOD is located in the mitochondria and SOD3 is extracellular. The primary role of all three enzymes is to catalyse the conversion (dismutation) of the superoxide radical to hydrogen peroxide, which in turn is converted to H2O by the action of glutathione peroxidase or catalase. In addition to its primary dismutase function, SOD1 has various subsidiary activities including peroxidase activity resulting in the generation of hydroxyl radicals from hydrogen peroxide (136) or superoxide (81), the production of nitronium species from peroxynitrite (7), and the protection of the enzyme calcineurin from inactivation (135). The active site channel of the normal SOD1 enzyme may impose spatial constraints on the access of molecules to the copper site. To date more than 60 different mutations in the SOD1 gene have been described in more than 250 ALS pedigrees, involving all five exons (113, 122) and rarely non-coding sequences (118) (Fig 2a). The majority of the mutations are single base-pair exonic substitutions. The sites of the SOD1 protein affected by the mutations tend to affect the dimer stability or beta-barrel folding of the SOD1 protein (17). In the presence of mutations the enzyme activity of SOD1 measured in red blood cells, in transformed lymphoblastoid cell lines and in brain tissue from MND cases is approximately 30 to 70 percent of the level measured in controls. This reduction in enzyme activity is thought to result from the lower stability and decreased half-life of the mutant protein (13). The precise sequence of events underlying motor neurone injury in the presence of SOD1 mutations has not yet been delineated, though some insights into the molecular pathophysiology have been gained through M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease Figure 2. Model of the SOD1 protein and its functional domains, and hypothesised toxic functions of mutations associated with familial MND. (a) A diagram of the Cu, Zn superoxide dismutase protein, comprising 153 amino acids with the 5 coding exons of the SOD1 gene as indicated. Above the protein are indicated some of the known mutations in SOD1 associated with familial MND, with the known functional domains indicated below. (b) Possible toxic actions of mutant SOD1. Normal SOD1 catalyses the conversion of superoxide to oxygen and hydrogen peroxide. At least four hypotheses have been proposed for the nature of the toxic function of mutant forms of SOD1 which are indicated by arrows; (1) enhanced peroxidase activity caused by increased access of copper at the active site to alternate substrates such as hydrogen peroxide; (2) enhanced reaction with peroxynitrite, catalysing the nitration of key protein residues; (3) metal toxicity, as a result of decreased affinity of the mutant enzyme for copper and/or zinc; and (4) protein aggregation due to misfolding of the mutant form of the protein. the use of experimental models. Given the widespread cellular expression of SOD1, the reasons why motor neurones are selectively injured in the presence of SOD1 mutations is also unclear. It is at present uncertain whether all SOD1 mutations have the same pathophysiological effects. It has been established that normal human motor neurones have a high expression of SOD1 in both the cell body and axons, compared to other groups neuronal groups (10, 96, 116). It has not yet been established whether a loss of function or a toxic gain of function of mutant SOD1, or both, is responsible for motor neurone injury, The loss of function hypothesis proposes that motor neurone injury occurs by the direct toxic effect of superoxide radicals inadequately scavenged by the mutant SOD1 protein. The following evidence has been put forward in support of the loss of function hypothesis: 1) In organotypic spinal cord cultures and cultures of PC12 cells, reduction of SOD1 activity by pharmacological means or the application of anti-sense oligonucleotides triggers cell death by apoptosis (109, 132). 2) Mutant SOD1 in Drosophila causes a dominant negative effect M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease on the normal SOD1 protein (97). 3) There are a large number of mutations in SOD1 which lead to a reduction in cytosolic enzyme activity, but are not predicted to produce similar structural changes in the protein (122). There is some experimental evidence that a loss of function of SOD1 could preferentially affect motor neurones. Thus, although motor function in mice where the SOD1 gene has been disrupted (SOD1 knockout) is normal, there is an increased susceptibility of motor neurones to injury and cell death following axotomy in these animals (102). However, an alternative hypothesis for which there is a compelling body of evidence, is that the mutant SOD1 protein may be causing motor neurone injury through a toxic gain of function. It is apparent that some mutant forms of SOD1 possess similar enzymatic activity to the normal enzyme (13). Furthermore, several groups have shown that transgenic mice carrying several tandem copies of the mutant human SOD1 gene develop motor neurone pathology remarkably similar to human MND (19, 62, 143). Transgenic mice with an equivalent copy number of the normal human SOD1 gene do not show any overt pathology. The key feature of these experiments is that they have been performed by introducing the mutant or wild type human SOD1 gene into mice with the normal complement of murine SOD1. Therefore, in this model the contribution of loss of SOD1 function must be minimal. This is also true in various cell culture models where the mutant forms of SOD1 are over-expressed in neuronal cells from rodents or in neural cell lines (46, 56, 100). Several hypotheses have been put forward to explain the potential toxic gain of function of mutant SOD1 (Fig. 2b). Many of the mutations could cause an alteration in the active site of the SOD1 enzyme. X-ray crytallographic studies have shown that the active channel of the mutant SOD1 protein has a slightly more open configuration than that of the wild-type enzyme, which could potentially allow greater accessibility of molecules to the active copper site (45). Such changes could allow mutant SOD1 to react with additional substrates such as H2O2 and peroxynitrite, as well as its normal substrate of superoxide anions (7, 17). Four main hypotheses have been put forward to explain the toxic gain of function of mutant SOD1. The first two of these hypotheses involve alterations in the metabolism of free radical species or their reaction products, but there are other possibilities that are not directly linked to oxidative stress (17). Hydroxyl radical formation. Hydrogen peroxide can, under certain conditions, interact with the active site of SOD1 to produce hydroxyl radicals. Under normal circumstances the charge profile of the enzyme channel and the local rate of production of hydrogen peroxide preclude the formation of significant quantities of hydroxyl radicals. However, it is possible that SOD1 mutations might result in unshielding of the active copper site, increasing the accessibility to hydrogen peroxide and/or may retard the egress of hydrogen peroxide, thereby augmenting the formation of hydroxyl radicals (136). Hydroxyl radicals may react in situ with SOD1 itself, thereby inactivating the enzyme, or may diffuse out into the cytosol to damage other cellular targets. Some of these pro-oxidant activities have been summarised by Singh et al., although these authors failed to find evidence of increased peroxidase activity for two SOD1 mutations (126). Nitration of protein tyrosine residues. Under normal physiological conditions the superoxide anion can combine with nitric oxide to from peroxynitrite. This process may be enhanced as the dismutation activity of SOD1 decreases, with a rise in the level of intracellular superoxide radicals. Normal or mutant SOD1 molecules can catalyse nitration of tyrosine residues of proteins by using peroxynitrite as a substrate, with the generation of nitronium ions. The wider active site channel of the mutant SOD1 protein may enhance the access of peroxynitrite to the active site. Neurofilament proteins and neurotrophic factor (tyrosine kinase) receptors are proteins particularly susceptible to nitrotyrosine damage and both are crucial for the normal function of motor neurones (7). Protein aggregation. It has been suggested that mutant SOD1 protein may accumulate and form toxic intracellular aggregates. This hypothesis is underpinned by immunocytochemical studies in autopsy material from MND patients and from in vitro models of SOD1 related familial MND using primary motor neurone cultures (46,120). Copper and zinc toxicity. Observations in yeast and bacteria indicate that certain mutant SOD1 proteins do not bind copper and zinc normally. However, many mutants do apparently bind copper normally (31). Reduced zinc binding in vitro has been confirmed using purified mutant SOD1 (37). Elevated intracellular levels of copper and zinc may be directly toxic to neurones (17). Copper chelators have been shown to have some M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease Model Observation Reference Cellular models: • Stable transfection in CSM14.1 rat nigral cells • Transfection in COS-1 or N2a cells • Stable transfection into human kidney cells • Microinjection into primary motorneurones • Adenoviral transfection into PC12 cells • Midbrain cultures from G93A transgenic mice • Stable transfection of human SH-SY5Y cells • Skin fibroblasts from familial MND patients Mouse models: • Over-expression of human G93A Increased cell death (apoptosis) after serum withdrawal Instability of mutant protein, no dominant negative effect on wild type SOD1 Instability of mutant (A4V) protein; reversed by proteasome inhibitors Increased cell death (apoptosis); aggregation of mutant SOD1 Increased cell death; rescue by anti-apoptotic and antioxidant strategies Increased sensitivity to cell death in oxidative paradigms Increased resting and stimulated intracellular free calcium Increased sensitivity to cell death induced by hydrogen peroxide or a peroxynitrite donor (100) (13) (63) (46) (56) (84) (21) (3) • Over-expression of G93A (low expressor) • Altered murine SOD1 to A4V • Over-expression of human G37R • Over-expression of human G85R Pathology similar to human disease Cytoskeletal alterations Fragmentation of Golgi apparatus Beneficial effects of antioxidants or riluzole Reduced orthograde transport and axonal neurofilaments Increased 3 nitrotyrosine Delayed onset by crossing with bcl-2 mice Increased calcium in motor nerve terminals Mitochondrial degeneration precedes axonal damage Slower progression of disease, possibly more similar to human MND Loss of motor function; spastic paralysis Age of onset related to transgene number; Vacuolar degeneration of mitochondria Reduced axonal transport of SOD1 and axonal abnormalities Increased free (not protein bound) 3-nityrotyrosine Motor neurone degeneration and astrocytic inclusions Decreased glial GLT-1 protein levels (62) (133) (88) (61) (148) (50) (74) (124) (73) (43) (106) (143) (14) (18) (19) (19) Table 1. Insights from cellular and mouse transgenic models of familial MND. neuroprotective effects in both cellular and animal models of SOD1 related MND (56, 64, 136). It has been shown recently that all mutant forms of SOD1 require a copper chaperone molecule called CCS to acquire Cu in vivo (31), although so far the relevance to SOD1 toxicity is speculative. There is no reason to assume that mutant SOD1 has only a single toxic function: none of the proposals for the toxic gain of function are mutually exclusive. As SOD1 mutations are heterogeneous, it is possible that specific mutations may have different cellular effects. It is clear that, whatever the mechanism(s) involved the cellular toxicity in the presence of mutant SOD1 protein, there must be a degree of selectivity for motor neurones, given that these are the most affected cells in both human patients and in transgenic mouse models. One reason for this may simply relate to the high level of expression of SOD1 by this cell group (10, 96, 116). Additionally, it has been suggested that motor neurones are particularly susceptible to oxidative stress given the high metabolic load imposed by their large cell size and long axonal processes. Insights from Cellular Models rather than transgenic mice. Firstly, the relative ease of manipulation of cultured cells means that the effects of several SOD1 mutations can be examined, rather than single mutations in transgenic animals. Secondly, purified neuronal cultures or clonal cell lines can be used to determine those effects of SOD1 mutations that are intrinsic to neurones, i.e., distinct from those which occur due to altered neuronal-glial interactions. Thirdly, application of putative neuroprotective strategies is more direct and easily controllable in culture than in vivo. (56). Apoptosis in the presence of oxidative stress. Rabizadeh showed that, when expressed in a neuronal cell line, the mutant SOD1 protein led to an increase in cell death under conditions of oxidative stress produced by serum or growth factor withdrawal. In contrast, overexpression of wild-type SOD1 inhibited apoptotic cell death (100). Other groups have prepared primary cultures of dopaminergic neurones from SOD1 transgenic mice (84) or have transfected cell lines with adenoviral constructs containing mutant or wild-type SOD1 (56). In these experimental paradigms mutant SOD1 also enhances apoptotic cell death, whereas normal SOD1 has a neuroprotective effect. Durham et al showed that microinjection of SOD1 expression constructs into primary neurones in culture enhances the naturally occurring cell death of motor neurones (46). This study is of interest as it demonstrated a differential effect of mutant The mechanisms involved in the neuronal toxicity of mutant SOD1 have been investigated in several laboratories using cell culture models. Table 1 summarises some of the main findings in such experimental systems. There are several advantages to using in vitro models M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease SOD1 on spinal cord motor neurones compared to cells cultured from other brain regions, including neurones from the hippocampus and the dorsal root ganglia. Altered affinity for hydrogen peroxide and increased formation of hydroxyl radicals. Yim et al have expressed the G93A and A4V SOD1 mutations, as well as wild-type SOD1, in Sf9 insect cells, purified the proteins, and studied their capacity for catalysing the dismutation of superoxide anions and for generating free radicals with H2O2 as a substrate (145, 146). Normal and mutant enzymes had identical dismutation activity. However, the free radical generating function of G93A SOD1 was enhanced relative to the wild-type enzyme, due to a small decrease in the value of the Km for H2O2. Furthermore, the A4V SOD1 had an even higher free radical generating function. Bredesen and co-workers have shown that at least two of the described SOD1 mutations can increase the rate of hydroxyl radical formation, probably due to increased availability of the copper at the active site of the enzyme to H2O2 (136). However, this data has recently been challenged by Singh et al., who failed to find any differences between mutant and wild type SOD1 molecules (126). Protein alterations. Durham et al suggested that the formation of intracellular aggregates of mutant SOD1 might be an important component of motor neurone toxicity. Following microinjection of mutant SOD1 expression constructs into primary motor neurones in culture aggregates of SOD1 were noted in cells which went into apoptosis (46). In vitro experiments have indicated that nitration of neurofilament proteins is catalysed by mutant SOD1 (37,38). These studies suggest that neurofilament proteins may be an important intracellular target for the toxic effects of mutant SOD1. Alteration in intracellular calcium homeostasis. Two studies have indicated that the presence of SOD1 may cause primary defects in the ability of cells to handle intracellular free calcium. Carri et al. have overexpressed the G93A SOD1 mutation in a human neuroblastoma cell line, SH-SY5Y (21). Two independent cell lines over-expressing the mutations at relatively modest levels (approximately double that of untransfected cells) had a significantly lower mitochondrial membrane potential and an increase in resting free cytosolic calcium. How these two key changes occur was not addressed, but it is possible that mutant SOD1 causes damage to the calcium homeostasis machinery of the cell via a free radical mediated mechanism. The suggestion that a derangement in intracellular free calcium underlies the cell death mechanism in motor neurone disease is strengthened by a report demonstrating similar increases in intracellular calcium in peripheral blood lymphocytes cultured from patients with sporadic MND (40). Increased intracellular free Ca2+ was seen in both basal conditions and after stimulation with an agent that uncouples oxidative phosphorylation. No alteration in a mitochondrial enzyme marker (cytochrome C oxidase) was found. The effects of SOD1 mutations on lymphocytic calcium handling has not yet been examined. There is good reason to consider that motor neurones vulnerable to pathology in MND may be relatively intolerant of changes in intracellular free calcium. Such motor neurones lack the calcium buffering proteins parvalbumin and calbindin D28k (66) but have a high expression of atypical calcium permeable glutamate receptors of the AMPA subtype (138), a class of glutamate receptors that has been implicated in MND (108, 115, 139). These differences in calcium handling may explain why motor neurones in cell culture are selectively vulnerable to cell death mediated via calcium influx after exposure to AMPA (22) or glutamate itself (104). If the selective vulnerability of motor neurones to toxic effects in the presence of SOD1 mutations is a relative phenomenon, then some excess toxicity in other cell groups might be expected. This may be true, as skin fibroblasts from patients with defined SOD1 mutations show an enhanced response to free radical mediated cell death, whether generated by hydrogen peroxide or by a peroxynitrite donor (3). Insights from SOD1 Transgenic Mice Several lines of transgenic mice have been produced which carry one or more copies of a mutant SOD1 gene, some salient features of which are given in Table 1. The first description of such a transgenic model was that reported by Gurney et al. (62), where several tandem copies of the G93A mutant form of the human SOD1 gene were introduced into mice. These mice develop an early onset (between 80 and 120 days), rapidly progressive motor dysfunction and become terminally paralysed approximately 6-12 weeks later (62, 73). Gurney and colleagues have also described another mouse line which have a lower copy number of the introduced G93A mutant SOD1. These animals develop a more M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease slowly progressive neuropathological phenotype (43). Subsequently, Ripps and colleagues have produced a transgenic mouse line by homologous targeting of one copy of the normal mouse SOD1 gene (106). Wong et al. have produced several transgenic mouse lines with the G37R mutation (141, 143). These have variable age of onset of disease, between 3.5 and 11 months, depending on the level of expression of the transgene (141). Finally, Bruijn and colleagues have produced mouse lines which express the human G85R mutant SOD1 at low levels, between 0.2x and 1x the endogenous mouse SOD1 levels (19). Transgenic mice with wild-type human SOD1 expressed at similar levels do not develop motor neurone disease. There are some differences in the pathology seen in each of these lines of transgenic mice, but damage to spinal cord and brainstem lower motor neurone groups is the dominant finding in all of them. All of the transgenic mouse models develop progressive weakness and wasting affecting the limb and bulbar muscles. A degree of caution must be exercised in extrapolating data from these transgenic mice to human MND. Nevertheless, the mice provide an important tool for examining the progression of cellular events underlying motor neurone injury longitudinally over the course of the disease, for testing hypotheses relating to the mechanisms of neuronal death and for evaluating novel therapeutic agents in vivo. Cellular pathology. The different mutations produce varying cellular pathology. The G93A and G37R mice tend to develop early changes of vacuolation in dendrites and axons of motor neurones, with reactive gliosis, followed by vacuolation of the perikarya and cell death. Prominent swelling and vacuolation of mitochondria and fragmentation of the Golgi apparatus are early features (73, 88, 143). The G93A mutant develops abnormal neurofilamentous accumulations in motor neurone perikarya similar to those present in human MND (133). The G85R mutant develops early morphological changes within glia, with inclusions that are immunoreactive with antibodies to SOD1 and ubiquitin (19, 20). These changes are accompanied by decreased expression of the glutamate transporter protein GLT-1. It has recently been demonstrated that G93A mice expressing a low copy number of the transgene develop pathology most closely resembling the changes found in human MND (41, 42). In these mice there is motor neurone degeneration associated with neuronal and filamentous inclusions, which show positive immunoreactivity for neurofilaments and ubiquitin, together with astrocytic inclusions. Kong and Xu showed in G93A mice that the onset of the disease involves a sharp decline in muscle strength and a transient explosive increase in vacuoles derived from degenerating mitochondria, but little motor neurone death (73). Significant cell death of motor neurones did not occur until the final stage of the disease, approximately nine weeks after the clinical features first became evident. These authors suggested that the motor neurone toxicity of mutant SOD1 is mediated by damage to mitochondria and that the absence of motor neurone death in the early stages of the disease indicates that the majority of motor neurones could potentially be rescued after clinical diagnosis. Axonal pathology has also been described in mutant SOD1 transgenic mice. There is strong evidence that both slow and fast axonal transport are affected in G93A mice (148), with a subsequent loss of axonal calibre. In the same study, a marked reduction in the neurofilament content of the L5 root was noted. It is likely that this loss of neurofilaments in the axonal compartment is linked with the appearance of neurofilament conglomerates in the cell bodies that has been noted by other authors (25,133) although this has not been tested directly. Axonal transport is also impaired in mice expressing the G37R mutant SOD1 enzyme (14). In this study, the authors used SOD1 itself as a marker of the slow component of anterograde axonal transport, and use the localisation of SOD1 to the proximal portion of the axons of spinal cord motor neurones to argue that the damage caused by SOD1 might localise to this area of the cell. Eyer et al. have recently shown that the progeny of crosses between mice expressing a NFL-lacZ transgene and those expressing SOD1 mutations develop motor neurone degeneration (48). The interesting property of NFL-lacZ transgenic mice is that neurofilaments are largely confined to the cell body compartment, and are restricted from entering the axon. As these mice develop both neuromuscular weakness and loss of motor axons, Eyer et al. argue that neurofilaments must not play a major role in the development of MND. However, these mice do not lack neurofilaments completely. Any effects that mutant SOD1 might have on perikaryal neurofilaments will still be maintained in this model. Thus, although it is possible that neurofilament alterations are a correlate of SOD1-mediated neurotoxicity rather than a crucial step in the development of the disease, it is also possible that the effects of mutant SOD1 on neurofilaments in the cell body is an important pathophysiological event. Williamson et al. have shown recently that M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease rofilaments is neuroprotective. NFH over-expression in the perikaryon may be neuroprotective because of the presence of high affinity calcium binding sites on neurofilaments (78, 79). Neurochemistry. Significant increases in the concentration of 3-nitrotyrosine, a marker of peroxynitritemediated nitration, have been demonstrated in the spinal cord of mice with the G93A SOD1 mutation (50). Malondialdehyde, a marker of lipid peroxidation, was increased in the cerebral cortex. Immunoreactivity for 3nitrotyrosine and malondialdehyde alteration of proteins were increased throughout the G93A mice spinal cord, the changes being particularly impressive within motor neurones. Bruijn and co-workers showed that 3-nitrotyrosine levels were elevated two to three fold in the spinal cords of G37R mice, coincident with the earliest pathological changes, and remained elevated in spinal cord throughout the progressive course of the disease (18). However, Bruijn et al. failed to find any increase in protein-bound nitrotyrosine using an immunoblotting technique (18) and found no evidence for increased formation of hydroxyl radicals, either by using salicylate spin trapping, or measuring malondialdehyde levels (18). Recently, Bogdanov et al. have demonstrated that there is an increased “hydroxyl-like” radical in G93A transgenic mice, using in vivo microdialysis (11). However, as these authors state this cannot be attributed unambiguously to hydroxyl radicals as peroxynitrite would be expected to have a similar oxidant activity in this system. Thus, there are indications that oxidative damage and increased nitration of protein residues are important aspects of the biochemical pathology of SOD1 mutations, though not all techniques used to examine these parameters have yielded consistent findings. Siklos et al. recently showed that spinal motor neurones of transgenic mice with the G93A SOD1 mutation show alterations in intracellular calcium (124). The cell bodies and proximal dendrites of spinal motor neurones showed small vacuoles filled with calcium, whereas these changes were not observed in oculomotor neurones. It was concluded that oxidative stress in the presence of mutant SOD1 resulted in derangement of intracellular calcium homeostasis in motor neurone populations lacking calbindin D28K and/or parvalbumin, while motor neurones possessing these calcium-binding proteins were more resistant to the stress. Figure 3. (a) A surviving motor neurone in the lumbar spinal cord of an MND patient showing a skein-like ubiquitinated inclusion. Immunoperoxidase staining for ubiquitin. (b) Lumbar motor neurone in a case of familial MND with the I113T SOD1 mutation showing hyaline conglomerate inclusions. These show positive staining with antibodies to phosphorylated and non-phosphorylated neurofilament proteins. Immunoperoxidase for neurofilament using monoclonal antibody SMI32. Scale bar in (b) indicates 50 mm and applies to both photomicrographs. NFL knockout mice also show a reduction in the other neurofilament subunits. When crossed with the G85R SOD1 transgenic line, lack of neurofilaments delays onset and increases survival, despite there being a reduction in motor neurone numbers attributable to an effect of NFL disruption directly (140). Increased sensory neuronal loss was seen, thus indicating that lack of neurofilaments decreases the selectivity of SOD1 mediated toxicity for motor neurones. Over-expression of NFH also increases survival in the G37R transgenic mouse (36), although neurofilamentous accumulations in the cell body were increased rather than reduced. Such a change might paradoxically reduce axonal neurofilaments whilst increasing the neurofilament content of the cell body. Thus, these two sets of results (36, 140) are both compatible with the idea that reducing axonal neu- SOD1 -Related Familial Motor Neurone Disease M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease Pathology. Examination of post-mortem tissue from patients with defined SOD1-linked familial cases has been reported by several groups (24, 67, 70, 71, 94, 98, 111, 114, 119, 129). The pathology of those cases where detailed descriptions have been published has been reviewed recently (67). Cases of MND with SOD1 mutations consistently show evidence of multisystem disease, with degeneration of the dorsal column pathways and spinocerebellar tracts usually being readily discernible. It is also common for these cases to show marked degeneration of the corticospinal tracts with little evidence of pathological change in the motor cortex, suggesting that the upper motor neurone injury may reflect a dying back axonopathy (67). Interestingly cases with the A4V mutation, which tend to have a rapidly progressive and short clinical course, have been reported to show little evidence of corticospinal tract pathology in the presence of florid lower motor neurone disease (39). The motor neurone inclusions present in SOD1 related familial MND cases are of interest. In some cases there are ubiquitinated inclusions with skein-like or Lewy body-like morphology identical to those observed in the sporadic form of the disease (Fig. 3a). Cases reported with A4V and I113T mutations have consistently shown the presence of dramatic argyrophilic hyaline conglomerate inclusions in the cell bodies and axons of motor neurones as well as in several other nonmotor neurone cell types (67) (Fig 3b). These hyaline conglomerate inclusions have been shown by confocal microscopic analysis and immunocytochemistry to contain equally abundant phosphorylated and non-phosphorylated neurofilament epitopes, indicating that phosphorylation is unlikely to be a key event in their formation. Such observations demonstrate that there are alterations in the distribution of neurofilament proteins in motor neurones in these patients. In contrast, neurofilament staining is virtually absent from the ubiquitinated inclusions typically found in sporadic MND. Inclusions in glial cells have been reported in two familial cases with unusually long duration of illness, 11 and 23 years (70, 71). One of these cases had a 2 bp deletion in SOD1 (71), but the nature of the other mutation was not reported. Neurochemistry. Neurochemical changes in patients with defined SOD1 mutations have been reported in relatively few cases. Ferrante et al. have shown that there is increased immunoreactive haemoxygenase-1 (a protein which is increased in response to oxidative stress) Figure 4. Spinal motor neurone from a case of sporadic MND showing diffuse immunoreactivity to nitrotyrosine in the motor neurone perikaryon and surrounding neuropil. Immunoperoxidase staining for nitrotyrosine. Scale bar indicates 50 m. and 8-hydroxy-2’-deoxyguanosine (49). However, no evidence for an increase in the lipid peroxidation marker (malondialdehyde) was found, although an increase in malondialdehyde-modified protein was seen. Beal et al. have demonstrated an increase in nitrotyrosine in two familial cases and a series of sporadic cases (5) (see also Fig. 4). As in mice carrying SOD1 mutations, increased free nitrotyrosine and nitrotyrosine staining was seen, although these authors did not assess protein-bound nitrotyrosine. Three familial cases with SOD1 mutations in exon 1 were included in a study of mitochondrial enzymes by Bowling et al. (15). An increase in mitochondrial complex-I activity was found in these patients, although no evidence of increased protein carbonyl groups was seen. Evidence that Oxidative Stress may be Playing a Role in Sporadic ALS Indices of free radical damage in post-mortem CNS tissue. The reactions of free radicals and other strong oxidant species with biomolecules leave modifications that can be estimated in biochemical assays. The interaction of oxygen free radicals with proteins can cause the formation of protein carbonyl residues. Increased protein-carbonyls have been found in the motor cortex (15) and spinal cord of sporadic cases of MND (117). Increased protein carbonyl formation can arise after reaction of proteins with peroxynitrite (68), as well as oxygen free radicals, and so the detection of this protein modification does not argue against the involvement of nitrogen centred species. Abe et al. have reported the presence of nitrotyrosine immunoreactivity within motor neurone cell bodies in MND patients (1,2), and this finding has been confirmed M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease subsequently by others (5). Fig. 4 shows an example of nitrotyrosine immunostaining in a lumbar spinal cord section from a sporadic MND patient. Chou and coworkers have reported iNOS (inducible nitric oxide synthase) immunoreactivity which co-localises with SOD1 within conglomerates in motor neurones (23,24). Therefore, there is some evidence that nitrergic stress may occur in MND patients, but at present the protein targets undergoing nitration have not been identified. Strong et al. have carefully examined protein nitration in a possible target molecule, namely neurofilament light, extracted from the spinal cords of a series of 15 sporadic and 11 control cases (127). No evidence for any alteration in the nitration of this protein was found. There is a great deal of evidence that biomolecules other than proteins can react with free radicals. DNA, for example, can be modified by free radical reactions, which can cause either strand breaks or modification of individual bases. There are approximately 20 oxidative DNA adducts which can be formed in such reactions. Fitzmaurice and colleagues have shown that there is an increase in one of these modified bases, 8-hydroxy-2’deoxyguanosine in spinal cord tissue from sporadic MND cases (52), although the number of samples in this study was very small. In a larger study, Ferrante et al. reported an increase in the same marker for oxidatively damaged DNA in the motor cortex of sporadic MND patients (49). Using immunohistochemical staining with an antibody raised against 8-hydroxy-2’-deoxyguanosine, the same authors demonstrated an increase in staining for this modified base in the spinal cord of familial and sporadic MND cases. Other modified DNA bases remain to be investigated, but such an approach might be useful in determining the possible contribution of nitrergic damage to DNA, as the reactions of nitrogen centred free radicals may produce a different pattern of DNA alterations. Oxidatively damaged DNA bases may come from both the nuclear compartment and from mitochondrial DNA. As discussed above, there is evidence of mitochondrial damage in the SOD1 transgenic mice (73, 143). However, in the studies of Ferrante et al., a nuclear fraction was used, indicating that there is damage to nuclear DNA in MND. Mammalian cells contain a number of enzymes that are activated by DNA repair mechanisms and measurement of these changes might provide an indirect way to assess if DNA damage has occurred in pathological samples. Oxidative stress may also lead to lipid modifications. Malondialdehyde is often used as a marker of lipid peroxidation. Although Ferrante et al. failed to see increased malondialdehyde in the motor cortex of sporadic MND patients (49), these authors did report a disease-related increase in staining for malondialdehyde modified proteins. Biochemical indices in CNS tissue which may indicate a compensatory response to oxidative stress. Several studies have indicated that there may be an increase in the expression and/or activity of free radical scavenging enzymes in sporadic MND. The activity of glutathione peroxidase (GPx), which serves to detoxify hydrogen peroxide, is elevated in the spinal cord of MND patients (65). In this study the content of selenium, a co-factor for GPx, was also increased in spinal cord tissue. However, another study has shown a decrease in GPx activity in the cortex of MND patients (99). An increase has been reported in the expression of SOD1 mRNA in individual surviving motor neurones from the spinal cord of sporadic MND cases (9). Increased SOD1 immunoreactivity has also been noted in glia in the vicinity of neuropathologically affected areas of the spinal cord in sporadic MND cases (116). Other free radical metabolising enzymes, namely SOD2 and catalase, also showed increased expression in the vicinity of the corticospinal tracts and/or in the neuropil of the ventral horn (116). Increased expression of metallothionein has also been demonstrated within spinal cord glia in cases of sporadic MND (125). This may also represent a compensatory response to ongoing oxidative stress, as metallothioneins are metal binding proteins, with free radical scavenging capabilities, which are also involved in the detoxification and storage of pro-oxidative metals. The above studies demonstrate an increased expression of molecules that may represent an attempt by cells within the CNS to respond to oxidative damage, presumably to limit damage to motor neurones. It is clear that in some circumstances, such responses can be inappropriate and enhance rather than limit cell death. As discussed earlier, there is evidence of DNA damage in pathologically affected tissue from sporadic MND cases. Nuclear DNA has a number of associated repair molecules, which can be activated by different types of DNA damage. For example, DNA strand breaks activate the enzyme poly(ADP-ribose) polymerase (PARP) which facilitates repair of DNA by increasing access to histone bound DNA by other repair enzymes. During this process PARP uses NAD+, thus depleting cells of antioxidant co-factors (128) (44). In vitro studies have shown that exposure of a motor neurone cell line to either peroxynitrite or hydrogen peroxide (two possible M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease reaction products of mutant SOD1) causes DNA damage and PARP activation (27). Importantly, in this experimental paradigm PARP inhibitors attenuated cell death. This is also true in mouse models of cerebral ischaemia, where PARP knockout mice are relatively resistant to cerebral arterial occlusion (47). Evidence has recently been uncovered suggesting an increase in PARP activation in sporadic MND cases, as measured by the accumulation of the reaction product poly(ADPribose) in spinal cord motor neurones and glia (28) (Fig. 5). This might, therefore, represent an attempt by cells to respond to oxidative DNA damage during the course of MND. Another DNA repair enzyme which may prove to be important in MND is the apurinergic/apyramidinergic endonuclease APE1, which catalyses the excision of abasic sites in DNA. Abasic sites can arise by the reaction of free radicals with DNA, as damaged nucleotides are removed by glycosylases (72). Cortical brain extracts from sporadic MND patients show reduced capacity for the repair of abasic sites and have a lower protein level of the enzyme APE1 (72). More recently, Olkowski has demonstrated a variety of point mutations in the gene for APE1 in 5 out of 7 patients with sporadic MND and these mutations were not present in a similar number of controls (92). These findings, which require confirmation in larger numbers of cases and controls, may represent the elucidation of an underlying genetic vulnerability factor in MND cases with no clear family history of the disease. DNA damage is linked to apoptosis via the p53 network (69), and motor neurones seem to retain the capacity for programming apoptotic cell death, even at maturity (93). As discussed earlier, there is some evidence for apoptosis in cellular models of SOD1 injury, and in SOD1-transgenic mice. In apoptotic cell death, the activation of a cascade that involves the caspase family of proteases occurs (121). This is controlled by a number of molecules, particularly the relative balance between the anti-apoptotic protein Bcl-2, which is located in the mitochondrial membrane (103) and its pro-apoptotic homologues such as Bax. Mu et al. (89) have demonstrated that there is decreased Bcl-2 and increased Bax expression within motor neurones in spinal cord from sporadic cases of MND. Troost et al. have demonstrated apoptotic cells within motor areas of the CNS in sporadic MND, but have also shown that there are changes in nearby brain areas, including an up-regulation of Bcl2 (131). DNA damage, as revealed by nick-end labelling, and another correlate of apoptosis, the LeY antigen, have also been shown to be increased in spo- Figure 5. Ventral horn of the lumbar spinal cord stained for poly(ADP-ribose), the reaction product of PARP activation, in (a) a control case and (b) a sporadic MND case. There is an increase in the intensity of staining for this marker in the MND case, where intense reactivity is seen within the nucleus of some shrunken neurones. Scale bar in (b) indicates 50 m and applies to both photomicrographs. radic MND patients (147). Therefore, it is possible that motor neurone death may occur via a programmed cell death pathway in sporadic MND cases. Candidate Cellular Targets for Oxidative Injury Neurofilament proteins. Neurofilament proteins form a major component of the cytoskeleton of neurones and are responsible for the maintenance of cell shape and axonal calibre as well as playing an important role in axonal transport. Motor neurones, with their very long axonal processes have a particularly high expression of neurofilament proteins. The smallest protein, neurofilament light, makes up the core of the neurofilament, while the two larger proteins, neurofilament medium and heavy, are arranged around this core and form M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease the projections or side arms radiating from the filament. The neurofilament subunits are produced in the motor neurone perikaryon and are transported down the axon by slow axonal transport, during which time progressive phosphorylation occurs. It has been shown that the tyrosine residues in the rod and tail domains of neurofilament light are more susceptible to SOD1-catalysed nitration compared to other proteins in the CNS (38). Features that may render neurofilament proteins vulnerable to oxidative stress include their long half-lives and their high content of lysine, an amino acid residue susceptible to oxidative modification Neurofilament proteins are of considerable interest as potential cellular targets for injury in MND. An alteration in the distribution of neurofilaments within surviving motor neurones is a characteristic feature of the pathology of MND (67, 111, 114) and such changes are also seen in mice carrying SOD1 mutations. These changes seem to reflect a decrease in the neurofilament content of distal axons (148) and accumulation in the cell body (25, 133). Whether accumulation in the cell body is a consequence or a cause of impaired axonal transport remains to be established. Whilst within the cell body, neurofilaments may be modified in a number of reactions. Nitration of tyrosine residues. Although neurofilament proteins might be targets for peroxynitrite-mediated tyrosine nitration (37, 38), the evidence that this happens in vivo is not convincing, at least so far (18, 127). Such nitrated proteins, or neurofilaments that have been modified by oxygen free radicals, may be rapidly degraded. In vitro experiments have shown that exposure of neuronal cells to oxidative stress causes a reduc- tion in neurofilament content (29). Ubiquitination. Ubiqitinated inclusions with compact or Lewy body like morphology in surviving motor neurones in MND may show immunoreactivity for neurofilament epitopes (67, 91). In some cases of SOD1related familial MND, large hyaline conglomerate inclusions, showing strong immunoreactivity with antibodies to both phosphorylated and non-phosphorylated neurofilaments have been observed in the cell bodies and axons of motor neurones (67, 111). These inclusions tend to show only weak immunoreactivity to ubiquitin. Hyperphosphorylation. Hyperphosphorylation of neurofilaments within the cell body of motor neurones has also been observed in pathological specimens from human MND cases (85, 90, 120) and also from SOD1 transgenic mice (87). This type of modification can be stimulated by oxidative and glutamate mediated cellular stress in vitro (57). It has not yet been established whether phosphorylation of neurofilaments occurs because neurofilaments are already retarded from entering the axons, or if protein phosphorylation is a primary event. However, it is intriguing that some protein kinases that are capable of phosphorylating neurofilaments in vitro, such as the stress-activated protein kinases (59, 60), also control apoptosis in other systems (75). There is evidence that one of these kinases, GSK, is present in inclusions in the spinal cord of MND patients (4). Occasional cases of sporadic MND have deletions or insertions in the KSP repeat region of the neurofilament heavy gene (51, 130). In addition, motor neurone pathology develops in transgenic mice overexpressing NF-L, NF-M or NF-H subunits (35, 142, 144) or in mice expressing mutations in the NF-L gene (77). These in Figure 6. (Opposing page) Possible relationships between glutamate toxicity, raised intracellular free calcium, oxidative stress and mutant SOD1in MND. Astrocytic uptake of glutamate via excitatory amino acid transporters 1 and 2 (EAAT1 and EAAT2) has been suggested to be impaired in sporadic MND, as loss of EAAT2 protein occurs. This increased glutamate in the synaptic cleft may provoke rises in intracellular free calcium in the motor neurones via the NMDA subtype of glutamate receptors and through calcium-permeable AMPA receptors. Depolarisation of the motor neurone may also cause an influx of calcium via voltage dependant calcium channels (VDCC), whilst metabotropic glutamate receptors would cause release of calcium from intracellular stores via inositol triphosphate (IP3). Mitochondria may also contribute to the intracellular calcium pool. Above a toxic threshold, these calcium rises would stimulate a variety of calcium regulated enzymes such as the protease calpain, phospholipase A2 (PLA2) and nitric oxide synthase. These last two enzymes stimulate superoxide (O2-) and nitric oxide (NO) production respectively, which in turn may react to form the long-lived strong oxidant species peroxynitrite (ONOO). Superoxide can also be formed by leakage from the mitochondrial electron transport chain, and by the action of xanthine oxidase (XO) on xanthine in the cytosol. Superoxide is normally detoxified by SOD1 to hydrogen peroxide, which is in turn converted to water by catalase and glutathione peroxidase (GPx). Mutant SOD1 has been hypothesised to catalyse the formation or reactions of two strong oxidants, either hydroxyl radicals or peroxynitrite (see text for details): Cu or Zn toxicity and protein aggregation may also contribute to the toxic function(s) of mutant SOD1. The net effects of these reactions will be damage to critical cellular biomolecules, including DNA, protein and lipids. Damage to specific protein targets within motor neurones, such as neurofilaments, may result in neurone-specific effects including decreased axonal transport. Damage to DNA, on the other hand, would stimulate DNA repair enzymes including poly(ADP-ribose)polymerase (PARP) which could potentially deplete intracellular ATP and NAD, thus increasing the mitochondrial load and potentially increasing free radical production by secondary effects. The net effect of this cascade of reactions will be to cause the death of the motor neurone, whether by apoptosis or necrosis or a mixture of both. M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease vivo models demonstrate that disruption of neurofilament assembly can selectively injure motor neurones and emphasise the importance of neurofilament proteins in the normal health of motor neurones. The G93A SOD1 mutant transgenic mouse shows alterations in neurofilament organisation, with the development of spheroids within motor neurones which contain both phosphorylated and non-phosphorylated neurofilaments as well as a reduction in neurofilament protein and the rate of transport seen in the ventral root axons (133, 148). Mitochondria. Mitochondria represent another promising candidate for a cellular target that might be preferentially affected in MND. Mitochondria are a major source for intracellular free radical generation, as both superoxide and hydroxyl radicals tend to leak from the mitochondrial electron-transport chain. Mitochondria are also affected by oxidative stress and by peroxynitrite, which can inhibit the activity of key mitochondrial enzymes (12). Furthermore, the opening of the mitochondrial permeability transition pore (MPT) is a key event in the control of apoptosis and is sensitive to both prolonged elevations in intracellular free calcium and oxidative stress (86). Peroxynitrite can cause the efflux of calcium from mitochondria via the MPT (95). The anti-apoptotic molecule Bcl-2, mentioned above in relation to pathological changes in human MND, may be part of this pore (86). As mitochondria control intracellular ATP and NAD production, inhibition of mitochondrial function will decrease reserves of these key intermediates. Decreased energy production has been shown to enhance susceptibility of neurones to glutamate-mediated toxicity (58). Such a mechanism would presumably be exaggerated by depletion of NAD/ATP by other mechanisms including PARP activation. Therefore, mitochondrial energy metabolism is inextricably linked with other mechanisms that have been shown to be relevant to the pathophysiology of MND, including oxidative stress, glutamatergic toxicity and intracellular calcium homeostasis (Fig. 6). Age-related changes in mitochondrial function have been suggested as an important factor contributing to late-onset human neurodegenerative diseases. There is evidence for age-related deterioration in mitochondrial function (16), and for a decline in brain glucose metabolism with ageing (6). A significant factor contributing to the age-related decline in mitochondrial function is the effect of accumulating mutations in the mitochondrial genome (30, 80, 105). Mitochondrial abnormalities may also be important in the selective vulnerability of motor neurones to pathological changes in MND. As calcium plays an important role in controlling mitochondrial function and opening of the MPT, cells most vulnerable to prolonged increases in intracellular calcium might also be most likely to develop mitochondrial abnormalities. As discussed earlier, motor neurones appear to have cell-specific molecular features that may render them especially vulnerable to changes in intracellular free calcium. Also, motor neurones have a very high metabolic load, being large active cells that have, like all neurones, an obligatory dependence on oxidative phosphorylation. This may make them especially vulnerable to loss of mitochondrial function. Finally, calcium fluxes and neuronal activity are inextricably linked; active neurones with the highest firing rates will therefore show the greatest calcium fluxes over time, which will increase the likelihood of mitochondrial MPT opening. In this context, it is of great interest that mitochondrial abnormalities are seen early in the progress of the disease in mutant SOD1 transgenic mice (73, 143). These data imply that damage to mitochondria may be a primary event, with any other abnormalities occurring later. However, two critical sets of experiments have yet to be performed. First, the activity of mitochondrial enzymes has not yet been measured in these mice, and hence it has not been clearly established if the mitochondria are morphologically abnormal but retain their function. Secondly, it has not been established whether therapeutic interventions are able to prevent mitochondrial damage in these mice, and whether such therapies would also prevent motor neurone loss and neuromuscular weakness. Aggregations of mitochondria have been seen in the neurones which synapse on to motor neurones in the spinal cord ventral horn in samples from MND patients (112). Increased mitochondrial volumes as well as increased intracellular calcium in the motor nerve terminals of MND patients have also been reported (123). Measurements of mitochondrial enzymes have been made in post-mortem tissue samples from MND patients. Decreased cytochrome c oxidase activity in the spinal cord of MND cases compared to controls has been reported (55). No alteration in the activity of any of the mitochondrial complexes was seen in a study of the frontal cortex from sporadic MND cases (49), although an increase in complex I activity was seen in some familial cases. One study, in which mitochondrial enzyme activities were measured in muscle biopsy sam- M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease ples from MND patients in life, showed unaltered cytochrome c oxidase activity but reduced NADH: cytochrome c reductase activity compared to control cases (137). Comi et al. reported a severe reduction in cytochrome c oxidase activity and protein in a patient with motor neurone degeneration (26). Genetic analysis of the same patient indicated a small deletion in the gene encoding one of the subunits of cytochrome c oxidase, COI, which was predicted to lead to a truncated protein. This gene is particularly interesting as it is encoded by the mitochondrial genome, which lacks many of the repair mechanisms associated with the nuclear genome and may be more susceptible to damage. It is not yet clear if deletions in the genes encoding mitochondrial proteins are commonly present in MND patients. Therapeutics SOD1 transgenic mice Cellular models of SOD1 mutations Vitamin E Gabapentin Riluzole Penicillamine Over expression of BCl-2 Inhibition of ICE (Interleukin-1- converting enzyme) Copper chelators Glutathione Vitamin E Caspase inhibitors SOD mimetics Table 2. Motor neurone degeneration: neuroprotective therapies in experimental models. Anti-glutamate agents Neurotrophic factors Riluzole + CNTF BDNF IGF-1 USA + Europe - no significant benefit; + significant benefit; + trend trend towards benefit, not statistically significant. CNTF = ciliary neurotrophic factor. BDNF = brain derived neurotrophic factor. IGF-1 = insulin like growth factor 1. Antioxidant therapy N-acetyl cysteine + trend Lamotrigine Gabapentin + trend One of the great hopes for research into MND is that animal models such as the SOD1 transgenic mice will provide an effective way of testing therapies for the human disease. At the same time, identification of therapies that exert a positive neuroprotective action may also help to support or reject some of the hypotheses for the molecular and cellular events underlying motor neurone injury. Listed in table 2 are some of the therapeutic approaches that have been screened in models of SOD1 toxicity, both in vitro and in vivo. Gurney and colleagues have shown that one antioxidant compound, vitamin E, and two agents which antagonise the effects of glutamate, riluzole and gabapentin, are of benefit in SOD1 transgenic mice with the G37R mutation (61). In these mice vitamin E delays the onset of clinical disease but does not extend survival, whereas anti-glutamate therapy extends survival but does not delay the onset of the disease. These therapeutic findings suggest that oxidative stress may be contributing to motor neurone injury at an early stage in the disease with glutamate toxicity playing a role later in the cell death cascade. The effects of alterations in the expression of certain genes on the course of the murine motor neurone disease has been examined in several experiments in which SOD1 transgenic mice have been cross-bred with other mouse transgenic strains. Over expression of the antiapoptotic protein Bcl-2 delays onset (74) whilst overexpression of a dominant negative inhibitor of the interleukin -1- -converting enzyme slows disease progression (54) in the presence of mutant SOD1. Cell culture models are potentially a useful testing ground for novel therapeutic strategies. One of the few Table 3. Recent clinical trials in MND. studies so far to investigate neuroprotective strategies in relation to mutant SOD1-mediated neuronal death is that of Ghadge et al.(56). Similar to studies in transgenic mice, vitamin E was able to ameliorate cell death in PC12 cells transfected with mutant SOD1, as were anti-apoptotic strategies including inhibition of caspases and over-expression of Bcl-2. Some SOD mimetics, glutathione and copper chelators were also effective. In this system a nitric oxide synthase (NOS) inhibitor was not protective, perhaps not surprisingly given that PC12 cells do not express NOS. It is beyond the scope of this article to list all those studies using cell culture models that may be relevant to any of the putative pathways to cell death which may be active in MND, or even those limited to oxidative stress. However, using cell culture models, it may be possible to identify possible neuroprotective strategies that target cellular changes common to several types of neuronal cell death. An example of this type of approach is the use of PARP inhibitors. The attraction of this group of compounds is that they may reduce cell death associated with DNA damage, whatever the initial cause of strand breaks. The potential use of PARP inhibitors in combating neurodegeneration has been reviewed by Szabó and Dawson (128), along with the limitations of currently available inhibitors. One in vitro study has demonstrated that two potential reaction products from M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease mutant SOD1, hydrogen peroxide and peroxynitrite, cause cell death in a motor neurone cell line and that these toxic effects can be reduced using PARP inhibitors (27). Other studies have shown that PARP inhibitors can prevent excitotoxic cell death (34) and may be active in vivo, at least in acute neurotoxicity models (32, 33). Table 3 shows the therapeutic agents that have been recently tested in clinical trials involving patients with MND (for a review see reference (82)). The anti-glutamate agent riluzole has a modest effect in prolonging survival in patients with MND (8, 76). There have been no large-scale trials of anti-oxidant compounds in MND. One small- scale trial (involving only 110 patients) of N-acetyl cysteine showed a trend towards improvement in survival in the subgroup of patients with limb-onset disease, but this did not reach statistical significance (p=0.06) (83). Other small-scale trials of antioxidant therapies have been performed but have failed to reach statistical significance, although some improvement in mortality was noted (134). Conclusions with environmental factors) and the continuing extension in our understanding of the cell specific molecular and physiological phenotype of human motor neurones, are important priorities for future research. These strategies are likely to pay dividends in the development of more effective neuroprotective therapies for individuals afflicted by this devastating neurodegenerative disease. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Brain Pathology Wiley

Oxidative Stress and Motor Neurone Disease

Brain Pathology , Volume 9 (1) – Jan 1, 1999

Loading next page...
 
/lp/wiley/oxidative-stress-and-motor-neurone-disease-0LVVOKxIwI

References (157)

Publisher
Wiley
Copyright
Copyright © 1999 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1015-6305
eISSN
1750-3639
DOI
10.1111/j.1750-3639.1999.tb00217.x
Publisher site
See Article on Publisher Site

Abstract

Mark R. Cookson and Pamela J. Shaw. Department of Neurology, University of Newcastle upon Tyne, UK The effects of oxidative stress within post mitotic cells such as neurones may be cumulative, and injury by free radical species is a major potential cause of the age-related deterioration in neuronal function seen in several neurodegenerative diseases. There is strong evidence that oxidative stress plays an important role in the pathogenesis of motor neurone disease (MND). Point mutations in the antioxidant enzyme Cu,Zn superoxide dismutase (SOD1) are found in some pedigrees with the familial form of MND. How mutations in this ubiquitous enzyme cause the relatively selective cell death of specific groups of motor neurones is not clear, although a number of hypotheses have been forwarded. These include (1) the formation of hydroxyl radicals, (2) the catalysis of reactions of the nitrogen centred oxidant species peroxynitrite, (3) toxicity of copper or zinc and (4) protein aggregation. Some experimental support for these different hypotheses has been produced by manipulating cells in culture to express the mutant SOD1 proteins and by generating transgenic mice which over-express mutant SOD1. Observations in these model systems are, in some cases at least, supported by observations made on pathological material from patients with similar SOD1 mutations. Furthermore, there are reports of evidence of free radical mediated damage to neurones in the sporadic form of MND. Several lines of evidence suggest that alterations in the glutamatergic neurotransmitter system may also play a key role in the injury to motor neurones in sporadic MND. There are several important subcellular tar- gets, which may be preferentially impaired within motor neurones, including neurofilament proteins and mitochondria. Future research will need to identify the aspects of the molecular and physiological phenotype of human motor neurones that makes them susceptible to degeneration in MND, and to identify those genetic and environmental factors which combine to cause this disease in individuals and in familial pedigrees. Introduction Corresponding author: Pamela J. Shaw, Department of Neurology, University of Newcastle Upon Tyne, Ward 11, Royal Victoria Infirmary, Newcastle upon Tyne, NE1 4LP, UK; Tel.: +44 191 2325131; Fax: +44 191 261 0881; E-mail: pamela.shaw@ncl.ac.uk Motor neurone disease (MND, also referred to as amyotrophic lateral sclerosis) is one of the commonest adult-onset neurodegenerative disorders with a world wide incidence of 1-2 per 100,000. The most prominent pathological features in MND are the degeneration of lower motor neurones in the ventral horn of the spinal cord and brainstem and of upper motor neurones in the motor cortex which give rise to the corticospinal tract. The selective vulnerability of motor neurones is relative and clinical and pathological evidence indicating involvement outside the motor system is commonly seen in MND (Fig. 1). Thus MND can be regarded as a multi-system disease in which the motor system is affected earliest and most severely. MND is familial in 10 percent and sporadic in approximately 90 percent of cases. The disease typically progresses over several years and inevitably results in death, usually due to respiratory insufficiency. Current therapies have only a limited effect on clinical and pathological progression. The primary pathogenetic processes underlying motor neurone injury in MND are likely to be multifactorial and the precise molecular pathways leading to the cell death of motor neurones are at present unknown. In recent years, two particularly important research findings have led to progress in elucidating the cellular pathophysiology underlying MND. Firstly, in 1993, Rosen et al. reported the association of point mutations in the gene for Cu/Zn superoxide dismutase (SOD1) with some familial cases of MND (107). In view of the clinical and pathological similarities between the famil- rones, oxidative stress represents an important potential cause of the age-related deterioration in cellular function observed in adult-onset neurodegenerative diseases. There is particular interest in the role of oxidative stress in MND because genetically determined alterations in the free radical defence system caused by mutations in the SOD1 gene underlie 20 percent of cases of familial MND or two percent of cases of MND as a whole (101, 107). Cu/Zn Superoxide Dismutase (SOD1) Mutations Figure 1. Spinal cord from a case of MND showing degeneration in the posterior column sensory pathways (arrows) in addition to the motor pathways. Luxol fast blue/cresyl violet stain. Bar = 2 mm. ial and sporadic forms of MND, it is likely that these two subgroups share common pathways of cellular injury. Secondly, in 1995, Rothstein and colleagues showed that there is a specific loss of the glial glutamate transporter protein EAAT2 in pathologically affected areas of the CNS in sporadic MND patients (110). These two studies have been instrumental in the development of two major hypotheses for factors contributing to motor neurone injury in MND, namely oxidative stress and glutamate mediated toxicity (excitotoxicity). Both of these studies (107, 110) have proved to be robust, in that the findings have been replicated by other research groups. Over 60 mutations in the SOD1 gene have been found by different groups (101), and reduced expression of the EAAT2 protein in the spinal cord of MND cases has been confirmed by Fray et al (53). In this review we will discuss the normal biochemistry of SOD1 and the current state of knowledge regarding the cellular effects of the mutant SOD1 protein. The evidence that oxidative stress may contribute to motor neurone injury in the sporadic form of MND will be reviewed and the candidate subcellular targets for injury within motor neurones will be highlighted. Finally the potential links between oxidative stress and other mechanisms of motor neurone injury in MND will be discussed. Oxidative Stress Free radicals are a potential source of damage to DNA, lipids, membranes and proteins within cells. Any imbalance between the intracellular production of free radical species and anti-oxidant defence mechanisms or free radical clearance systems results in a state of oxidative stress. In relation to post-mitotic cells such as neu- SOD1 is a metalloenzyme comprising 153 amino acids with copper and zinc binding sites (Fig. 2). The enzyme functions as a homodimer, with a copper ion, essential for the dismutation reaction, held in position at the base of the active channel by four histidine residues. SOD1 is a cytosolic enzyme that is ubiquitous in distribution. There are two related superoxide dismutase enzymes. Manganese SOD is located in the mitochondria and SOD3 is extracellular. The primary role of all three enzymes is to catalyse the conversion (dismutation) of the superoxide radical to hydrogen peroxide, which in turn is converted to H2O by the action of glutathione peroxidase or catalase. In addition to its primary dismutase function, SOD1 has various subsidiary activities including peroxidase activity resulting in the generation of hydroxyl radicals from hydrogen peroxide (136) or superoxide (81), the production of nitronium species from peroxynitrite (7), and the protection of the enzyme calcineurin from inactivation (135). The active site channel of the normal SOD1 enzyme may impose spatial constraints on the access of molecules to the copper site. To date more than 60 different mutations in the SOD1 gene have been described in more than 250 ALS pedigrees, involving all five exons (113, 122) and rarely non-coding sequences (118) (Fig 2a). The majority of the mutations are single base-pair exonic substitutions. The sites of the SOD1 protein affected by the mutations tend to affect the dimer stability or beta-barrel folding of the SOD1 protein (17). In the presence of mutations the enzyme activity of SOD1 measured in red blood cells, in transformed lymphoblastoid cell lines and in brain tissue from MND cases is approximately 30 to 70 percent of the level measured in controls. This reduction in enzyme activity is thought to result from the lower stability and decreased half-life of the mutant protein (13). The precise sequence of events underlying motor neurone injury in the presence of SOD1 mutations has not yet been delineated, though some insights into the molecular pathophysiology have been gained through M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease Figure 2. Model of the SOD1 protein and its functional domains, and hypothesised toxic functions of mutations associated with familial MND. (a) A diagram of the Cu, Zn superoxide dismutase protein, comprising 153 amino acids with the 5 coding exons of the SOD1 gene as indicated. Above the protein are indicated some of the known mutations in SOD1 associated with familial MND, with the known functional domains indicated below. (b) Possible toxic actions of mutant SOD1. Normal SOD1 catalyses the conversion of superoxide to oxygen and hydrogen peroxide. At least four hypotheses have been proposed for the nature of the toxic function of mutant forms of SOD1 which are indicated by arrows; (1) enhanced peroxidase activity caused by increased access of copper at the active site to alternate substrates such as hydrogen peroxide; (2) enhanced reaction with peroxynitrite, catalysing the nitration of key protein residues; (3) metal toxicity, as a result of decreased affinity of the mutant enzyme for copper and/or zinc; and (4) protein aggregation due to misfolding of the mutant form of the protein. the use of experimental models. Given the widespread cellular expression of SOD1, the reasons why motor neurones are selectively injured in the presence of SOD1 mutations is also unclear. It is at present uncertain whether all SOD1 mutations have the same pathophysiological effects. It has been established that normal human motor neurones have a high expression of SOD1 in both the cell body and axons, compared to other groups neuronal groups (10, 96, 116). It has not yet been established whether a loss of function or a toxic gain of function of mutant SOD1, or both, is responsible for motor neurone injury, The loss of function hypothesis proposes that motor neurone injury occurs by the direct toxic effect of superoxide radicals inadequately scavenged by the mutant SOD1 protein. The following evidence has been put forward in support of the loss of function hypothesis: 1) In organotypic spinal cord cultures and cultures of PC12 cells, reduction of SOD1 activity by pharmacological means or the application of anti-sense oligonucleotides triggers cell death by apoptosis (109, 132). 2) Mutant SOD1 in Drosophila causes a dominant negative effect M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease on the normal SOD1 protein (97). 3) There are a large number of mutations in SOD1 which lead to a reduction in cytosolic enzyme activity, but are not predicted to produce similar structural changes in the protein (122). There is some experimental evidence that a loss of function of SOD1 could preferentially affect motor neurones. Thus, although motor function in mice where the SOD1 gene has been disrupted (SOD1 knockout) is normal, there is an increased susceptibility of motor neurones to injury and cell death following axotomy in these animals (102). However, an alternative hypothesis for which there is a compelling body of evidence, is that the mutant SOD1 protein may be causing motor neurone injury through a toxic gain of function. It is apparent that some mutant forms of SOD1 possess similar enzymatic activity to the normal enzyme (13). Furthermore, several groups have shown that transgenic mice carrying several tandem copies of the mutant human SOD1 gene develop motor neurone pathology remarkably similar to human MND (19, 62, 143). Transgenic mice with an equivalent copy number of the normal human SOD1 gene do not show any overt pathology. The key feature of these experiments is that they have been performed by introducing the mutant or wild type human SOD1 gene into mice with the normal complement of murine SOD1. Therefore, in this model the contribution of loss of SOD1 function must be minimal. This is also true in various cell culture models where the mutant forms of SOD1 are over-expressed in neuronal cells from rodents or in neural cell lines (46, 56, 100). Several hypotheses have been put forward to explain the potential toxic gain of function of mutant SOD1 (Fig. 2b). Many of the mutations could cause an alteration in the active site of the SOD1 enzyme. X-ray crytallographic studies have shown that the active channel of the mutant SOD1 protein has a slightly more open configuration than that of the wild-type enzyme, which could potentially allow greater accessibility of molecules to the active copper site (45). Such changes could allow mutant SOD1 to react with additional substrates such as H2O2 and peroxynitrite, as well as its normal substrate of superoxide anions (7, 17). Four main hypotheses have been put forward to explain the toxic gain of function of mutant SOD1. The first two of these hypotheses involve alterations in the metabolism of free radical species or their reaction products, but there are other possibilities that are not directly linked to oxidative stress (17). Hydroxyl radical formation. Hydrogen peroxide can, under certain conditions, interact with the active site of SOD1 to produce hydroxyl radicals. Under normal circumstances the charge profile of the enzyme channel and the local rate of production of hydrogen peroxide preclude the formation of significant quantities of hydroxyl radicals. However, it is possible that SOD1 mutations might result in unshielding of the active copper site, increasing the accessibility to hydrogen peroxide and/or may retard the egress of hydrogen peroxide, thereby augmenting the formation of hydroxyl radicals (136). Hydroxyl radicals may react in situ with SOD1 itself, thereby inactivating the enzyme, or may diffuse out into the cytosol to damage other cellular targets. Some of these pro-oxidant activities have been summarised by Singh et al., although these authors failed to find evidence of increased peroxidase activity for two SOD1 mutations (126). Nitration of protein tyrosine residues. Under normal physiological conditions the superoxide anion can combine with nitric oxide to from peroxynitrite. This process may be enhanced as the dismutation activity of SOD1 decreases, with a rise in the level of intracellular superoxide radicals. Normal or mutant SOD1 molecules can catalyse nitration of tyrosine residues of proteins by using peroxynitrite as a substrate, with the generation of nitronium ions. The wider active site channel of the mutant SOD1 protein may enhance the access of peroxynitrite to the active site. Neurofilament proteins and neurotrophic factor (tyrosine kinase) receptors are proteins particularly susceptible to nitrotyrosine damage and both are crucial for the normal function of motor neurones (7). Protein aggregation. It has been suggested that mutant SOD1 protein may accumulate and form toxic intracellular aggregates. This hypothesis is underpinned by immunocytochemical studies in autopsy material from MND patients and from in vitro models of SOD1 related familial MND using primary motor neurone cultures (46,120). Copper and zinc toxicity. Observations in yeast and bacteria indicate that certain mutant SOD1 proteins do not bind copper and zinc normally. However, many mutants do apparently bind copper normally (31). Reduced zinc binding in vitro has been confirmed using purified mutant SOD1 (37). Elevated intracellular levels of copper and zinc may be directly toxic to neurones (17). Copper chelators have been shown to have some M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease Model Observation Reference Cellular models: • Stable transfection in CSM14.1 rat nigral cells • Transfection in COS-1 or N2a cells • Stable transfection into human kidney cells • Microinjection into primary motorneurones • Adenoviral transfection into PC12 cells • Midbrain cultures from G93A transgenic mice • Stable transfection of human SH-SY5Y cells • Skin fibroblasts from familial MND patients Mouse models: • Over-expression of human G93A Increased cell death (apoptosis) after serum withdrawal Instability of mutant protein, no dominant negative effect on wild type SOD1 Instability of mutant (A4V) protein; reversed by proteasome inhibitors Increased cell death (apoptosis); aggregation of mutant SOD1 Increased cell death; rescue by anti-apoptotic and antioxidant strategies Increased sensitivity to cell death in oxidative paradigms Increased resting and stimulated intracellular free calcium Increased sensitivity to cell death induced by hydrogen peroxide or a peroxynitrite donor (100) (13) (63) (46) (56) (84) (21) (3) • Over-expression of G93A (low expressor) • Altered murine SOD1 to A4V • Over-expression of human G37R • Over-expression of human G85R Pathology similar to human disease Cytoskeletal alterations Fragmentation of Golgi apparatus Beneficial effects of antioxidants or riluzole Reduced orthograde transport and axonal neurofilaments Increased 3 nitrotyrosine Delayed onset by crossing with bcl-2 mice Increased calcium in motor nerve terminals Mitochondrial degeneration precedes axonal damage Slower progression of disease, possibly more similar to human MND Loss of motor function; spastic paralysis Age of onset related to transgene number; Vacuolar degeneration of mitochondria Reduced axonal transport of SOD1 and axonal abnormalities Increased free (not protein bound) 3-nityrotyrosine Motor neurone degeneration and astrocytic inclusions Decreased glial GLT-1 protein levels (62) (133) (88) (61) (148) (50) (74) (124) (73) (43) (106) (143) (14) (18) (19) (19) Table 1. Insights from cellular and mouse transgenic models of familial MND. neuroprotective effects in both cellular and animal models of SOD1 related MND (56, 64, 136). It has been shown recently that all mutant forms of SOD1 require a copper chaperone molecule called CCS to acquire Cu in vivo (31), although so far the relevance to SOD1 toxicity is speculative. There is no reason to assume that mutant SOD1 has only a single toxic function: none of the proposals for the toxic gain of function are mutually exclusive. As SOD1 mutations are heterogeneous, it is possible that specific mutations may have different cellular effects. It is clear that, whatever the mechanism(s) involved the cellular toxicity in the presence of mutant SOD1 protein, there must be a degree of selectivity for motor neurones, given that these are the most affected cells in both human patients and in transgenic mouse models. One reason for this may simply relate to the high level of expression of SOD1 by this cell group (10, 96, 116). Additionally, it has been suggested that motor neurones are particularly susceptible to oxidative stress given the high metabolic load imposed by their large cell size and long axonal processes. Insights from Cellular Models rather than transgenic mice. Firstly, the relative ease of manipulation of cultured cells means that the effects of several SOD1 mutations can be examined, rather than single mutations in transgenic animals. Secondly, purified neuronal cultures or clonal cell lines can be used to determine those effects of SOD1 mutations that are intrinsic to neurones, i.e., distinct from those which occur due to altered neuronal-glial interactions. Thirdly, application of putative neuroprotective strategies is more direct and easily controllable in culture than in vivo. (56). Apoptosis in the presence of oxidative stress. Rabizadeh showed that, when expressed in a neuronal cell line, the mutant SOD1 protein led to an increase in cell death under conditions of oxidative stress produced by serum or growth factor withdrawal. In contrast, overexpression of wild-type SOD1 inhibited apoptotic cell death (100). Other groups have prepared primary cultures of dopaminergic neurones from SOD1 transgenic mice (84) or have transfected cell lines with adenoviral constructs containing mutant or wild-type SOD1 (56). In these experimental paradigms mutant SOD1 also enhances apoptotic cell death, whereas normal SOD1 has a neuroprotective effect. Durham et al showed that microinjection of SOD1 expression constructs into primary neurones in culture enhances the naturally occurring cell death of motor neurones (46). This study is of interest as it demonstrated a differential effect of mutant The mechanisms involved in the neuronal toxicity of mutant SOD1 have been investigated in several laboratories using cell culture models. Table 1 summarises some of the main findings in such experimental systems. There are several advantages to using in vitro models M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease SOD1 on spinal cord motor neurones compared to cells cultured from other brain regions, including neurones from the hippocampus and the dorsal root ganglia. Altered affinity for hydrogen peroxide and increased formation of hydroxyl radicals. Yim et al have expressed the G93A and A4V SOD1 mutations, as well as wild-type SOD1, in Sf9 insect cells, purified the proteins, and studied their capacity for catalysing the dismutation of superoxide anions and for generating free radicals with H2O2 as a substrate (145, 146). Normal and mutant enzymes had identical dismutation activity. However, the free radical generating function of G93A SOD1 was enhanced relative to the wild-type enzyme, due to a small decrease in the value of the Km for H2O2. Furthermore, the A4V SOD1 had an even higher free radical generating function. Bredesen and co-workers have shown that at least two of the described SOD1 mutations can increase the rate of hydroxyl radical formation, probably due to increased availability of the copper at the active site of the enzyme to H2O2 (136). However, this data has recently been challenged by Singh et al., who failed to find any differences between mutant and wild type SOD1 molecules (126). Protein alterations. Durham et al suggested that the formation of intracellular aggregates of mutant SOD1 might be an important component of motor neurone toxicity. Following microinjection of mutant SOD1 expression constructs into primary motor neurones in culture aggregates of SOD1 were noted in cells which went into apoptosis (46). In vitro experiments have indicated that nitration of neurofilament proteins is catalysed by mutant SOD1 (37,38). These studies suggest that neurofilament proteins may be an important intracellular target for the toxic effects of mutant SOD1. Alteration in intracellular calcium homeostasis. Two studies have indicated that the presence of SOD1 may cause primary defects in the ability of cells to handle intracellular free calcium. Carri et al. have overexpressed the G93A SOD1 mutation in a human neuroblastoma cell line, SH-SY5Y (21). Two independent cell lines over-expressing the mutations at relatively modest levels (approximately double that of untransfected cells) had a significantly lower mitochondrial membrane potential and an increase in resting free cytosolic calcium. How these two key changes occur was not addressed, but it is possible that mutant SOD1 causes damage to the calcium homeostasis machinery of the cell via a free radical mediated mechanism. The suggestion that a derangement in intracellular free calcium underlies the cell death mechanism in motor neurone disease is strengthened by a report demonstrating similar increases in intracellular calcium in peripheral blood lymphocytes cultured from patients with sporadic MND (40). Increased intracellular free Ca2+ was seen in both basal conditions and after stimulation with an agent that uncouples oxidative phosphorylation. No alteration in a mitochondrial enzyme marker (cytochrome C oxidase) was found. The effects of SOD1 mutations on lymphocytic calcium handling has not yet been examined. There is good reason to consider that motor neurones vulnerable to pathology in MND may be relatively intolerant of changes in intracellular free calcium. Such motor neurones lack the calcium buffering proteins parvalbumin and calbindin D28k (66) but have a high expression of atypical calcium permeable glutamate receptors of the AMPA subtype (138), a class of glutamate receptors that has been implicated in MND (108, 115, 139). These differences in calcium handling may explain why motor neurones in cell culture are selectively vulnerable to cell death mediated via calcium influx after exposure to AMPA (22) or glutamate itself (104). If the selective vulnerability of motor neurones to toxic effects in the presence of SOD1 mutations is a relative phenomenon, then some excess toxicity in other cell groups might be expected. This may be true, as skin fibroblasts from patients with defined SOD1 mutations show an enhanced response to free radical mediated cell death, whether generated by hydrogen peroxide or by a peroxynitrite donor (3). Insights from SOD1 Transgenic Mice Several lines of transgenic mice have been produced which carry one or more copies of a mutant SOD1 gene, some salient features of which are given in Table 1. The first description of such a transgenic model was that reported by Gurney et al. (62), where several tandem copies of the G93A mutant form of the human SOD1 gene were introduced into mice. These mice develop an early onset (between 80 and 120 days), rapidly progressive motor dysfunction and become terminally paralysed approximately 6-12 weeks later (62, 73). Gurney and colleagues have also described another mouse line which have a lower copy number of the introduced G93A mutant SOD1. These animals develop a more M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease slowly progressive neuropathological phenotype (43). Subsequently, Ripps and colleagues have produced a transgenic mouse line by homologous targeting of one copy of the normal mouse SOD1 gene (106). Wong et al. have produced several transgenic mouse lines with the G37R mutation (141, 143). These have variable age of onset of disease, between 3.5 and 11 months, depending on the level of expression of the transgene (141). Finally, Bruijn and colleagues have produced mouse lines which express the human G85R mutant SOD1 at low levels, between 0.2x and 1x the endogenous mouse SOD1 levels (19). Transgenic mice with wild-type human SOD1 expressed at similar levels do not develop motor neurone disease. There are some differences in the pathology seen in each of these lines of transgenic mice, but damage to spinal cord and brainstem lower motor neurone groups is the dominant finding in all of them. All of the transgenic mouse models develop progressive weakness and wasting affecting the limb and bulbar muscles. A degree of caution must be exercised in extrapolating data from these transgenic mice to human MND. Nevertheless, the mice provide an important tool for examining the progression of cellular events underlying motor neurone injury longitudinally over the course of the disease, for testing hypotheses relating to the mechanisms of neuronal death and for evaluating novel therapeutic agents in vivo. Cellular pathology. The different mutations produce varying cellular pathology. The G93A and G37R mice tend to develop early changes of vacuolation in dendrites and axons of motor neurones, with reactive gliosis, followed by vacuolation of the perikarya and cell death. Prominent swelling and vacuolation of mitochondria and fragmentation of the Golgi apparatus are early features (73, 88, 143). The G93A mutant develops abnormal neurofilamentous accumulations in motor neurone perikarya similar to those present in human MND (133). The G85R mutant develops early morphological changes within glia, with inclusions that are immunoreactive with antibodies to SOD1 and ubiquitin (19, 20). These changes are accompanied by decreased expression of the glutamate transporter protein GLT-1. It has recently been demonstrated that G93A mice expressing a low copy number of the transgene develop pathology most closely resembling the changes found in human MND (41, 42). In these mice there is motor neurone degeneration associated with neuronal and filamentous inclusions, which show positive immunoreactivity for neurofilaments and ubiquitin, together with astrocytic inclusions. Kong and Xu showed in G93A mice that the onset of the disease involves a sharp decline in muscle strength and a transient explosive increase in vacuoles derived from degenerating mitochondria, but little motor neurone death (73). Significant cell death of motor neurones did not occur until the final stage of the disease, approximately nine weeks after the clinical features first became evident. These authors suggested that the motor neurone toxicity of mutant SOD1 is mediated by damage to mitochondria and that the absence of motor neurone death in the early stages of the disease indicates that the majority of motor neurones could potentially be rescued after clinical diagnosis. Axonal pathology has also been described in mutant SOD1 transgenic mice. There is strong evidence that both slow and fast axonal transport are affected in G93A mice (148), with a subsequent loss of axonal calibre. In the same study, a marked reduction in the neurofilament content of the L5 root was noted. It is likely that this loss of neurofilaments in the axonal compartment is linked with the appearance of neurofilament conglomerates in the cell bodies that has been noted by other authors (25,133) although this has not been tested directly. Axonal transport is also impaired in mice expressing the G37R mutant SOD1 enzyme (14). In this study, the authors used SOD1 itself as a marker of the slow component of anterograde axonal transport, and use the localisation of SOD1 to the proximal portion of the axons of spinal cord motor neurones to argue that the damage caused by SOD1 might localise to this area of the cell. Eyer et al. have recently shown that the progeny of crosses between mice expressing a NFL-lacZ transgene and those expressing SOD1 mutations develop motor neurone degeneration (48). The interesting property of NFL-lacZ transgenic mice is that neurofilaments are largely confined to the cell body compartment, and are restricted from entering the axon. As these mice develop both neuromuscular weakness and loss of motor axons, Eyer et al. argue that neurofilaments must not play a major role in the development of MND. However, these mice do not lack neurofilaments completely. Any effects that mutant SOD1 might have on perikaryal neurofilaments will still be maintained in this model. Thus, although it is possible that neurofilament alterations are a correlate of SOD1-mediated neurotoxicity rather than a crucial step in the development of the disease, it is also possible that the effects of mutant SOD1 on neurofilaments in the cell body is an important pathophysiological event. Williamson et al. have shown recently that M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease rofilaments is neuroprotective. NFH over-expression in the perikaryon may be neuroprotective because of the presence of high affinity calcium binding sites on neurofilaments (78, 79). Neurochemistry. Significant increases in the concentration of 3-nitrotyrosine, a marker of peroxynitritemediated nitration, have been demonstrated in the spinal cord of mice with the G93A SOD1 mutation (50). Malondialdehyde, a marker of lipid peroxidation, was increased in the cerebral cortex. Immunoreactivity for 3nitrotyrosine and malondialdehyde alteration of proteins were increased throughout the G93A mice spinal cord, the changes being particularly impressive within motor neurones. Bruijn and co-workers showed that 3-nitrotyrosine levels were elevated two to three fold in the spinal cords of G37R mice, coincident with the earliest pathological changes, and remained elevated in spinal cord throughout the progressive course of the disease (18). However, Bruijn et al. failed to find any increase in protein-bound nitrotyrosine using an immunoblotting technique (18) and found no evidence for increased formation of hydroxyl radicals, either by using salicylate spin trapping, or measuring malondialdehyde levels (18). Recently, Bogdanov et al. have demonstrated that there is an increased “hydroxyl-like” radical in G93A transgenic mice, using in vivo microdialysis (11). However, as these authors state this cannot be attributed unambiguously to hydroxyl radicals as peroxynitrite would be expected to have a similar oxidant activity in this system. Thus, there are indications that oxidative damage and increased nitration of protein residues are important aspects of the biochemical pathology of SOD1 mutations, though not all techniques used to examine these parameters have yielded consistent findings. Siklos et al. recently showed that spinal motor neurones of transgenic mice with the G93A SOD1 mutation show alterations in intracellular calcium (124). The cell bodies and proximal dendrites of spinal motor neurones showed small vacuoles filled with calcium, whereas these changes were not observed in oculomotor neurones. It was concluded that oxidative stress in the presence of mutant SOD1 resulted in derangement of intracellular calcium homeostasis in motor neurone populations lacking calbindin D28K and/or parvalbumin, while motor neurones possessing these calcium-binding proteins were more resistant to the stress. Figure 3. (a) A surviving motor neurone in the lumbar spinal cord of an MND patient showing a skein-like ubiquitinated inclusion. Immunoperoxidase staining for ubiquitin. (b) Lumbar motor neurone in a case of familial MND with the I113T SOD1 mutation showing hyaline conglomerate inclusions. These show positive staining with antibodies to phosphorylated and non-phosphorylated neurofilament proteins. Immunoperoxidase for neurofilament using monoclonal antibody SMI32. Scale bar in (b) indicates 50 mm and applies to both photomicrographs. NFL knockout mice also show a reduction in the other neurofilament subunits. When crossed with the G85R SOD1 transgenic line, lack of neurofilaments delays onset and increases survival, despite there being a reduction in motor neurone numbers attributable to an effect of NFL disruption directly (140). Increased sensory neuronal loss was seen, thus indicating that lack of neurofilaments decreases the selectivity of SOD1 mediated toxicity for motor neurones. Over-expression of NFH also increases survival in the G37R transgenic mouse (36), although neurofilamentous accumulations in the cell body were increased rather than reduced. Such a change might paradoxically reduce axonal neurofilaments whilst increasing the neurofilament content of the cell body. Thus, these two sets of results (36, 140) are both compatible with the idea that reducing axonal neu- SOD1 -Related Familial Motor Neurone Disease M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease Pathology. Examination of post-mortem tissue from patients with defined SOD1-linked familial cases has been reported by several groups (24, 67, 70, 71, 94, 98, 111, 114, 119, 129). The pathology of those cases where detailed descriptions have been published has been reviewed recently (67). Cases of MND with SOD1 mutations consistently show evidence of multisystem disease, with degeneration of the dorsal column pathways and spinocerebellar tracts usually being readily discernible. It is also common for these cases to show marked degeneration of the corticospinal tracts with little evidence of pathological change in the motor cortex, suggesting that the upper motor neurone injury may reflect a dying back axonopathy (67). Interestingly cases with the A4V mutation, which tend to have a rapidly progressive and short clinical course, have been reported to show little evidence of corticospinal tract pathology in the presence of florid lower motor neurone disease (39). The motor neurone inclusions present in SOD1 related familial MND cases are of interest. In some cases there are ubiquitinated inclusions with skein-like or Lewy body-like morphology identical to those observed in the sporadic form of the disease (Fig. 3a). Cases reported with A4V and I113T mutations have consistently shown the presence of dramatic argyrophilic hyaline conglomerate inclusions in the cell bodies and axons of motor neurones as well as in several other nonmotor neurone cell types (67) (Fig 3b). These hyaline conglomerate inclusions have been shown by confocal microscopic analysis and immunocytochemistry to contain equally abundant phosphorylated and non-phosphorylated neurofilament epitopes, indicating that phosphorylation is unlikely to be a key event in their formation. Such observations demonstrate that there are alterations in the distribution of neurofilament proteins in motor neurones in these patients. In contrast, neurofilament staining is virtually absent from the ubiquitinated inclusions typically found in sporadic MND. Inclusions in glial cells have been reported in two familial cases with unusually long duration of illness, 11 and 23 years (70, 71). One of these cases had a 2 bp deletion in SOD1 (71), but the nature of the other mutation was not reported. Neurochemistry. Neurochemical changes in patients with defined SOD1 mutations have been reported in relatively few cases. Ferrante et al. have shown that there is increased immunoreactive haemoxygenase-1 (a protein which is increased in response to oxidative stress) Figure 4. Spinal motor neurone from a case of sporadic MND showing diffuse immunoreactivity to nitrotyrosine in the motor neurone perikaryon and surrounding neuropil. Immunoperoxidase staining for nitrotyrosine. Scale bar indicates 50 m. and 8-hydroxy-2’-deoxyguanosine (49). However, no evidence for an increase in the lipid peroxidation marker (malondialdehyde) was found, although an increase in malondialdehyde-modified protein was seen. Beal et al. have demonstrated an increase in nitrotyrosine in two familial cases and a series of sporadic cases (5) (see also Fig. 4). As in mice carrying SOD1 mutations, increased free nitrotyrosine and nitrotyrosine staining was seen, although these authors did not assess protein-bound nitrotyrosine. Three familial cases with SOD1 mutations in exon 1 were included in a study of mitochondrial enzymes by Bowling et al. (15). An increase in mitochondrial complex-I activity was found in these patients, although no evidence of increased protein carbonyl groups was seen. Evidence that Oxidative Stress may be Playing a Role in Sporadic ALS Indices of free radical damage in post-mortem CNS tissue. The reactions of free radicals and other strong oxidant species with biomolecules leave modifications that can be estimated in biochemical assays. The interaction of oxygen free radicals with proteins can cause the formation of protein carbonyl residues. Increased protein-carbonyls have been found in the motor cortex (15) and spinal cord of sporadic cases of MND (117). Increased protein carbonyl formation can arise after reaction of proteins with peroxynitrite (68), as well as oxygen free radicals, and so the detection of this protein modification does not argue against the involvement of nitrogen centred species. Abe et al. have reported the presence of nitrotyrosine immunoreactivity within motor neurone cell bodies in MND patients (1,2), and this finding has been confirmed M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease subsequently by others (5). Fig. 4 shows an example of nitrotyrosine immunostaining in a lumbar spinal cord section from a sporadic MND patient. Chou and coworkers have reported iNOS (inducible nitric oxide synthase) immunoreactivity which co-localises with SOD1 within conglomerates in motor neurones (23,24). Therefore, there is some evidence that nitrergic stress may occur in MND patients, but at present the protein targets undergoing nitration have not been identified. Strong et al. have carefully examined protein nitration in a possible target molecule, namely neurofilament light, extracted from the spinal cords of a series of 15 sporadic and 11 control cases (127). No evidence for any alteration in the nitration of this protein was found. There is a great deal of evidence that biomolecules other than proteins can react with free radicals. DNA, for example, can be modified by free radical reactions, which can cause either strand breaks or modification of individual bases. There are approximately 20 oxidative DNA adducts which can be formed in such reactions. Fitzmaurice and colleagues have shown that there is an increase in one of these modified bases, 8-hydroxy-2’deoxyguanosine in spinal cord tissue from sporadic MND cases (52), although the number of samples in this study was very small. In a larger study, Ferrante et al. reported an increase in the same marker for oxidatively damaged DNA in the motor cortex of sporadic MND patients (49). Using immunohistochemical staining with an antibody raised against 8-hydroxy-2’-deoxyguanosine, the same authors demonstrated an increase in staining for this modified base in the spinal cord of familial and sporadic MND cases. Other modified DNA bases remain to be investigated, but such an approach might be useful in determining the possible contribution of nitrergic damage to DNA, as the reactions of nitrogen centred free radicals may produce a different pattern of DNA alterations. Oxidatively damaged DNA bases may come from both the nuclear compartment and from mitochondrial DNA. As discussed above, there is evidence of mitochondrial damage in the SOD1 transgenic mice (73, 143). However, in the studies of Ferrante et al., a nuclear fraction was used, indicating that there is damage to nuclear DNA in MND. Mammalian cells contain a number of enzymes that are activated by DNA repair mechanisms and measurement of these changes might provide an indirect way to assess if DNA damage has occurred in pathological samples. Oxidative stress may also lead to lipid modifications. Malondialdehyde is often used as a marker of lipid peroxidation. Although Ferrante et al. failed to see increased malondialdehyde in the motor cortex of sporadic MND patients (49), these authors did report a disease-related increase in staining for malondialdehyde modified proteins. Biochemical indices in CNS tissue which may indicate a compensatory response to oxidative stress. Several studies have indicated that there may be an increase in the expression and/or activity of free radical scavenging enzymes in sporadic MND. The activity of glutathione peroxidase (GPx), which serves to detoxify hydrogen peroxide, is elevated in the spinal cord of MND patients (65). In this study the content of selenium, a co-factor for GPx, was also increased in spinal cord tissue. However, another study has shown a decrease in GPx activity in the cortex of MND patients (99). An increase has been reported in the expression of SOD1 mRNA in individual surviving motor neurones from the spinal cord of sporadic MND cases (9). Increased SOD1 immunoreactivity has also been noted in glia in the vicinity of neuropathologically affected areas of the spinal cord in sporadic MND cases (116). Other free radical metabolising enzymes, namely SOD2 and catalase, also showed increased expression in the vicinity of the corticospinal tracts and/or in the neuropil of the ventral horn (116). Increased expression of metallothionein has also been demonstrated within spinal cord glia in cases of sporadic MND (125). This may also represent a compensatory response to ongoing oxidative stress, as metallothioneins are metal binding proteins, with free radical scavenging capabilities, which are also involved in the detoxification and storage of pro-oxidative metals. The above studies demonstrate an increased expression of molecules that may represent an attempt by cells within the CNS to respond to oxidative damage, presumably to limit damage to motor neurones. It is clear that in some circumstances, such responses can be inappropriate and enhance rather than limit cell death. As discussed earlier, there is evidence of DNA damage in pathologically affected tissue from sporadic MND cases. Nuclear DNA has a number of associated repair molecules, which can be activated by different types of DNA damage. For example, DNA strand breaks activate the enzyme poly(ADP-ribose) polymerase (PARP) which facilitates repair of DNA by increasing access to histone bound DNA by other repair enzymes. During this process PARP uses NAD+, thus depleting cells of antioxidant co-factors (128) (44). In vitro studies have shown that exposure of a motor neurone cell line to either peroxynitrite or hydrogen peroxide (two possible M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease reaction products of mutant SOD1) causes DNA damage and PARP activation (27). Importantly, in this experimental paradigm PARP inhibitors attenuated cell death. This is also true in mouse models of cerebral ischaemia, where PARP knockout mice are relatively resistant to cerebral arterial occlusion (47). Evidence has recently been uncovered suggesting an increase in PARP activation in sporadic MND cases, as measured by the accumulation of the reaction product poly(ADPribose) in spinal cord motor neurones and glia (28) (Fig. 5). This might, therefore, represent an attempt by cells to respond to oxidative DNA damage during the course of MND. Another DNA repair enzyme which may prove to be important in MND is the apurinergic/apyramidinergic endonuclease APE1, which catalyses the excision of abasic sites in DNA. Abasic sites can arise by the reaction of free radicals with DNA, as damaged nucleotides are removed by glycosylases (72). Cortical brain extracts from sporadic MND patients show reduced capacity for the repair of abasic sites and have a lower protein level of the enzyme APE1 (72). More recently, Olkowski has demonstrated a variety of point mutations in the gene for APE1 in 5 out of 7 patients with sporadic MND and these mutations were not present in a similar number of controls (92). These findings, which require confirmation in larger numbers of cases and controls, may represent the elucidation of an underlying genetic vulnerability factor in MND cases with no clear family history of the disease. DNA damage is linked to apoptosis via the p53 network (69), and motor neurones seem to retain the capacity for programming apoptotic cell death, even at maturity (93). As discussed earlier, there is some evidence for apoptosis in cellular models of SOD1 injury, and in SOD1-transgenic mice. In apoptotic cell death, the activation of a cascade that involves the caspase family of proteases occurs (121). This is controlled by a number of molecules, particularly the relative balance between the anti-apoptotic protein Bcl-2, which is located in the mitochondrial membrane (103) and its pro-apoptotic homologues such as Bax. Mu et al. (89) have demonstrated that there is decreased Bcl-2 and increased Bax expression within motor neurones in spinal cord from sporadic cases of MND. Troost et al. have demonstrated apoptotic cells within motor areas of the CNS in sporadic MND, but have also shown that there are changes in nearby brain areas, including an up-regulation of Bcl2 (131). DNA damage, as revealed by nick-end labelling, and another correlate of apoptosis, the LeY antigen, have also been shown to be increased in spo- Figure 5. Ventral horn of the lumbar spinal cord stained for poly(ADP-ribose), the reaction product of PARP activation, in (a) a control case and (b) a sporadic MND case. There is an increase in the intensity of staining for this marker in the MND case, where intense reactivity is seen within the nucleus of some shrunken neurones. Scale bar in (b) indicates 50 m and applies to both photomicrographs. radic MND patients (147). Therefore, it is possible that motor neurone death may occur via a programmed cell death pathway in sporadic MND cases. Candidate Cellular Targets for Oxidative Injury Neurofilament proteins. Neurofilament proteins form a major component of the cytoskeleton of neurones and are responsible for the maintenance of cell shape and axonal calibre as well as playing an important role in axonal transport. Motor neurones, with their very long axonal processes have a particularly high expression of neurofilament proteins. The smallest protein, neurofilament light, makes up the core of the neurofilament, while the two larger proteins, neurofilament medium and heavy, are arranged around this core and form M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease the projections or side arms radiating from the filament. The neurofilament subunits are produced in the motor neurone perikaryon and are transported down the axon by slow axonal transport, during which time progressive phosphorylation occurs. It has been shown that the tyrosine residues in the rod and tail domains of neurofilament light are more susceptible to SOD1-catalysed nitration compared to other proteins in the CNS (38). Features that may render neurofilament proteins vulnerable to oxidative stress include their long half-lives and their high content of lysine, an amino acid residue susceptible to oxidative modification Neurofilament proteins are of considerable interest as potential cellular targets for injury in MND. An alteration in the distribution of neurofilaments within surviving motor neurones is a characteristic feature of the pathology of MND (67, 111, 114) and such changes are also seen in mice carrying SOD1 mutations. These changes seem to reflect a decrease in the neurofilament content of distal axons (148) and accumulation in the cell body (25, 133). Whether accumulation in the cell body is a consequence or a cause of impaired axonal transport remains to be established. Whilst within the cell body, neurofilaments may be modified in a number of reactions. Nitration of tyrosine residues. Although neurofilament proteins might be targets for peroxynitrite-mediated tyrosine nitration (37, 38), the evidence that this happens in vivo is not convincing, at least so far (18, 127). Such nitrated proteins, or neurofilaments that have been modified by oxygen free radicals, may be rapidly degraded. In vitro experiments have shown that exposure of neuronal cells to oxidative stress causes a reduc- tion in neurofilament content (29). Ubiquitination. Ubiqitinated inclusions with compact or Lewy body like morphology in surviving motor neurones in MND may show immunoreactivity for neurofilament epitopes (67, 91). In some cases of SOD1related familial MND, large hyaline conglomerate inclusions, showing strong immunoreactivity with antibodies to both phosphorylated and non-phosphorylated neurofilaments have been observed in the cell bodies and axons of motor neurones (67, 111). These inclusions tend to show only weak immunoreactivity to ubiquitin. Hyperphosphorylation. Hyperphosphorylation of neurofilaments within the cell body of motor neurones has also been observed in pathological specimens from human MND cases (85, 90, 120) and also from SOD1 transgenic mice (87). This type of modification can be stimulated by oxidative and glutamate mediated cellular stress in vitro (57). It has not yet been established whether phosphorylation of neurofilaments occurs because neurofilaments are already retarded from entering the axons, or if protein phosphorylation is a primary event. However, it is intriguing that some protein kinases that are capable of phosphorylating neurofilaments in vitro, such as the stress-activated protein kinases (59, 60), also control apoptosis in other systems (75). There is evidence that one of these kinases, GSK, is present in inclusions in the spinal cord of MND patients (4). Occasional cases of sporadic MND have deletions or insertions in the KSP repeat region of the neurofilament heavy gene (51, 130). In addition, motor neurone pathology develops in transgenic mice overexpressing NF-L, NF-M or NF-H subunits (35, 142, 144) or in mice expressing mutations in the NF-L gene (77). These in Figure 6. (Opposing page) Possible relationships between glutamate toxicity, raised intracellular free calcium, oxidative stress and mutant SOD1in MND. Astrocytic uptake of glutamate via excitatory amino acid transporters 1 and 2 (EAAT1 and EAAT2) has been suggested to be impaired in sporadic MND, as loss of EAAT2 protein occurs. This increased glutamate in the synaptic cleft may provoke rises in intracellular free calcium in the motor neurones via the NMDA subtype of glutamate receptors and through calcium-permeable AMPA receptors. Depolarisation of the motor neurone may also cause an influx of calcium via voltage dependant calcium channels (VDCC), whilst metabotropic glutamate receptors would cause release of calcium from intracellular stores via inositol triphosphate (IP3). Mitochondria may also contribute to the intracellular calcium pool. Above a toxic threshold, these calcium rises would stimulate a variety of calcium regulated enzymes such as the protease calpain, phospholipase A2 (PLA2) and nitric oxide synthase. These last two enzymes stimulate superoxide (O2-) and nitric oxide (NO) production respectively, which in turn may react to form the long-lived strong oxidant species peroxynitrite (ONOO). Superoxide can also be formed by leakage from the mitochondrial electron transport chain, and by the action of xanthine oxidase (XO) on xanthine in the cytosol. Superoxide is normally detoxified by SOD1 to hydrogen peroxide, which is in turn converted to water by catalase and glutathione peroxidase (GPx). Mutant SOD1 has been hypothesised to catalyse the formation or reactions of two strong oxidants, either hydroxyl radicals or peroxynitrite (see text for details): Cu or Zn toxicity and protein aggregation may also contribute to the toxic function(s) of mutant SOD1. The net effects of these reactions will be damage to critical cellular biomolecules, including DNA, protein and lipids. Damage to specific protein targets within motor neurones, such as neurofilaments, may result in neurone-specific effects including decreased axonal transport. Damage to DNA, on the other hand, would stimulate DNA repair enzymes including poly(ADP-ribose)polymerase (PARP) which could potentially deplete intracellular ATP and NAD, thus increasing the mitochondrial load and potentially increasing free radical production by secondary effects. The net effect of this cascade of reactions will be to cause the death of the motor neurone, whether by apoptosis or necrosis or a mixture of both. M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease vivo models demonstrate that disruption of neurofilament assembly can selectively injure motor neurones and emphasise the importance of neurofilament proteins in the normal health of motor neurones. The G93A SOD1 mutant transgenic mouse shows alterations in neurofilament organisation, with the development of spheroids within motor neurones which contain both phosphorylated and non-phosphorylated neurofilaments as well as a reduction in neurofilament protein and the rate of transport seen in the ventral root axons (133, 148). Mitochondria. Mitochondria represent another promising candidate for a cellular target that might be preferentially affected in MND. Mitochondria are a major source for intracellular free radical generation, as both superoxide and hydroxyl radicals tend to leak from the mitochondrial electron-transport chain. Mitochondria are also affected by oxidative stress and by peroxynitrite, which can inhibit the activity of key mitochondrial enzymes (12). Furthermore, the opening of the mitochondrial permeability transition pore (MPT) is a key event in the control of apoptosis and is sensitive to both prolonged elevations in intracellular free calcium and oxidative stress (86). Peroxynitrite can cause the efflux of calcium from mitochondria via the MPT (95). The anti-apoptotic molecule Bcl-2, mentioned above in relation to pathological changes in human MND, may be part of this pore (86). As mitochondria control intracellular ATP and NAD production, inhibition of mitochondrial function will decrease reserves of these key intermediates. Decreased energy production has been shown to enhance susceptibility of neurones to glutamate-mediated toxicity (58). Such a mechanism would presumably be exaggerated by depletion of NAD/ATP by other mechanisms including PARP activation. Therefore, mitochondrial energy metabolism is inextricably linked with other mechanisms that have been shown to be relevant to the pathophysiology of MND, including oxidative stress, glutamatergic toxicity and intracellular calcium homeostasis (Fig. 6). Age-related changes in mitochondrial function have been suggested as an important factor contributing to late-onset human neurodegenerative diseases. There is evidence for age-related deterioration in mitochondrial function (16), and for a decline in brain glucose metabolism with ageing (6). A significant factor contributing to the age-related decline in mitochondrial function is the effect of accumulating mutations in the mitochondrial genome (30, 80, 105). Mitochondrial abnormalities may also be important in the selective vulnerability of motor neurones to pathological changes in MND. As calcium plays an important role in controlling mitochondrial function and opening of the MPT, cells most vulnerable to prolonged increases in intracellular calcium might also be most likely to develop mitochondrial abnormalities. As discussed earlier, motor neurones appear to have cell-specific molecular features that may render them especially vulnerable to changes in intracellular free calcium. Also, motor neurones have a very high metabolic load, being large active cells that have, like all neurones, an obligatory dependence on oxidative phosphorylation. This may make them especially vulnerable to loss of mitochondrial function. Finally, calcium fluxes and neuronal activity are inextricably linked; active neurones with the highest firing rates will therefore show the greatest calcium fluxes over time, which will increase the likelihood of mitochondrial MPT opening. In this context, it is of great interest that mitochondrial abnormalities are seen early in the progress of the disease in mutant SOD1 transgenic mice (73, 143). These data imply that damage to mitochondria may be a primary event, with any other abnormalities occurring later. However, two critical sets of experiments have yet to be performed. First, the activity of mitochondrial enzymes has not yet been measured in these mice, and hence it has not been clearly established if the mitochondria are morphologically abnormal but retain their function. Secondly, it has not been established whether therapeutic interventions are able to prevent mitochondrial damage in these mice, and whether such therapies would also prevent motor neurone loss and neuromuscular weakness. Aggregations of mitochondria have been seen in the neurones which synapse on to motor neurones in the spinal cord ventral horn in samples from MND patients (112). Increased mitochondrial volumes as well as increased intracellular calcium in the motor nerve terminals of MND patients have also been reported (123). Measurements of mitochondrial enzymes have been made in post-mortem tissue samples from MND patients. Decreased cytochrome c oxidase activity in the spinal cord of MND cases compared to controls has been reported (55). No alteration in the activity of any of the mitochondrial complexes was seen in a study of the frontal cortex from sporadic MND cases (49), although an increase in complex I activity was seen in some familial cases. One study, in which mitochondrial enzyme activities were measured in muscle biopsy sam- M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease ples from MND patients in life, showed unaltered cytochrome c oxidase activity but reduced NADH: cytochrome c reductase activity compared to control cases (137). Comi et al. reported a severe reduction in cytochrome c oxidase activity and protein in a patient with motor neurone degeneration (26). Genetic analysis of the same patient indicated a small deletion in the gene encoding one of the subunits of cytochrome c oxidase, COI, which was predicted to lead to a truncated protein. This gene is particularly interesting as it is encoded by the mitochondrial genome, which lacks many of the repair mechanisms associated with the nuclear genome and may be more susceptible to damage. It is not yet clear if deletions in the genes encoding mitochondrial proteins are commonly present in MND patients. Therapeutics SOD1 transgenic mice Cellular models of SOD1 mutations Vitamin E Gabapentin Riluzole Penicillamine Over expression of BCl-2 Inhibition of ICE (Interleukin-1- converting enzyme) Copper chelators Glutathione Vitamin E Caspase inhibitors SOD mimetics Table 2. Motor neurone degeneration: neuroprotective therapies in experimental models. Anti-glutamate agents Neurotrophic factors Riluzole + CNTF BDNF IGF-1 USA + Europe - no significant benefit; + significant benefit; + trend trend towards benefit, not statistically significant. CNTF = ciliary neurotrophic factor. BDNF = brain derived neurotrophic factor. IGF-1 = insulin like growth factor 1. Antioxidant therapy N-acetyl cysteine + trend Lamotrigine Gabapentin + trend One of the great hopes for research into MND is that animal models such as the SOD1 transgenic mice will provide an effective way of testing therapies for the human disease. At the same time, identification of therapies that exert a positive neuroprotective action may also help to support or reject some of the hypotheses for the molecular and cellular events underlying motor neurone injury. Listed in table 2 are some of the therapeutic approaches that have been screened in models of SOD1 toxicity, both in vitro and in vivo. Gurney and colleagues have shown that one antioxidant compound, vitamin E, and two agents which antagonise the effects of glutamate, riluzole and gabapentin, are of benefit in SOD1 transgenic mice with the G37R mutation (61). In these mice vitamin E delays the onset of clinical disease but does not extend survival, whereas anti-glutamate therapy extends survival but does not delay the onset of the disease. These therapeutic findings suggest that oxidative stress may be contributing to motor neurone injury at an early stage in the disease with glutamate toxicity playing a role later in the cell death cascade. The effects of alterations in the expression of certain genes on the course of the murine motor neurone disease has been examined in several experiments in which SOD1 transgenic mice have been cross-bred with other mouse transgenic strains. Over expression of the antiapoptotic protein Bcl-2 delays onset (74) whilst overexpression of a dominant negative inhibitor of the interleukin -1- -converting enzyme slows disease progression (54) in the presence of mutant SOD1. Cell culture models are potentially a useful testing ground for novel therapeutic strategies. One of the few Table 3. Recent clinical trials in MND. studies so far to investigate neuroprotective strategies in relation to mutant SOD1-mediated neuronal death is that of Ghadge et al.(56). Similar to studies in transgenic mice, vitamin E was able to ameliorate cell death in PC12 cells transfected with mutant SOD1, as were anti-apoptotic strategies including inhibition of caspases and over-expression of Bcl-2. Some SOD mimetics, glutathione and copper chelators were also effective. In this system a nitric oxide synthase (NOS) inhibitor was not protective, perhaps not surprisingly given that PC12 cells do not express NOS. It is beyond the scope of this article to list all those studies using cell culture models that may be relevant to any of the putative pathways to cell death which may be active in MND, or even those limited to oxidative stress. However, using cell culture models, it may be possible to identify possible neuroprotective strategies that target cellular changes common to several types of neuronal cell death. An example of this type of approach is the use of PARP inhibitors. The attraction of this group of compounds is that they may reduce cell death associated with DNA damage, whatever the initial cause of strand breaks. The potential use of PARP inhibitors in combating neurodegeneration has been reviewed by Szabó and Dawson (128), along with the limitations of currently available inhibitors. One in vitro study has demonstrated that two potential reaction products from M.R. Cookson and P.J. Shaw: Oxidative Stress and Motor Neurone Disease mutant SOD1, hydrogen peroxide and peroxynitrite, cause cell death in a motor neurone cell line and that these toxic effects can be reduced using PARP inhibitors (27). Other studies have shown that PARP inhibitors can prevent excitotoxic cell death (34) and may be active in vivo, at least in acute neurotoxicity models (32, 33). Table 3 shows the therapeutic agents that have been recently tested in clinical trials involving patients with MND (for a review see reference (82)). The anti-glutamate agent riluzole has a modest effect in prolonging survival in patients with MND (8, 76). There have been no large-scale trials of anti-oxidant compounds in MND. One small- scale trial (involving only 110 patients) of N-acetyl cysteine showed a trend towards improvement in survival in the subgroup of patients with limb-onset disease, but this did not reach statistical significance (p=0.06) (83). Other small-scale trials of antioxidant therapies have been performed but have failed to reach statistical significance, although some improvement in mortality was noted (134). Conclusions with environmental factors) and the continuing extension in our understanding of the cell specific molecular and physiological phenotype of human motor neurones, are important priorities for future research. These strategies are likely to pay dividends in the development of more effective neuroprotective therapies for individuals afflicted by this devastating neurodegenerative disease.

Journal

Brain PathologyWiley

Published: Jan 1, 1999

There are no references for this article.