|Additional related information may be found at:|
|Neuropsychopharmacology: The Fifth Generation of Progress|
Amyotrophic Lateral Sclerosis, Glutamate, and Oxidative Stress
Andreas Plaitakis and P. Shashidharan
Amyotrophic lateral sclerosis (ALS) is a devastating human disease affecting approximately one to two persons per 100,000 population each year—an incidence similar to that of multiple sclerosis. Clinically, ALS is manifested by muscle weakness, wasting, spasticity, and weight loss; pathologically, it is characterized by degeneration and loss of the motor neurons in the spinal cord, brainstem, and cerebral cortex. Death usually occurs within 2–5 years after the onset of symptoms. While the overwhelming majority of cases are sporadic (primary or idiopathic ALS), about 5–10% of patients have a positive family history (familial ALS).
The cause of primary ALS is unknown, and no effective treatment is currently available. Recent studies, however, have shown that abnormal glutamate metabolism occurs in patients with sporadic ALS and is thought to cause motor neuron degeneration via neuroexcitotoxic mechanisms. Meanwhile, molecular genetic analyses have linked the familial form of ALS with mutations of Cu/Zn superoxide dismutase (SOD1), an enzyme that is part of the cellular antioxidant defense systems. Because there are indications that glutamate excitotoxicity is associated with increased oxidative stress, and that failure of antioxidant defense systems can lead to excitotoxicity, these observations suggest common final pathways to motor neuron degeneration. Because glutamate is central to the main theme of this chapter, the role of this amino acid in the biology of nerve tissue in health and disease is discussed in detail (see also A Critical Analysic of the Neurochemical Methords for Monitoring Transmitter Dynamics in the Brain, Exciatotry Amino Acid Neurotransmission, Galanin: A Neuropepptide with Important Central Nervous System Actions, and Luteinzing Hormone-Releasing Hormone Neuronal, this volume).
GLUTAMATE FUNCTION IN METABOLISM
Glutamate, a five-carbon skeleton dicarboxylic amino acid, is known to play a key role in mammalian intermediary metabolism. Glutamate is involved in the synthesis and/or catabolism of many compounds, including amino acids, ketoacids, and peptides. Gamma-aminobutyric acid (GABA), a major inhibitory neurotransmitter in the mammalian central nervous system (CNS), is known to be formed from glutamate (Fig. 1) by decarboxylation. Glutamate can be reversibly transaminated via glutamate oxaloacetate transaminase (GOT) with oxaloacetate to form a-ketoglutarate and aspartate (Fig. 1). Also, it can be oxidatively deaminated to a-ketoglutarate and ammonia via glutamate dehydrogenase (GDH). As such, glutamate is associated directly with aerobic metabolism via the Krebs cycle.
Glutamate plays an essential role in ammonia homeostasis. In mammalian liver, oxidative deamination of glutamate (via GDH) provides ammonia for urea synthesis. However, in brain, as in other organs that do not have an active urea cycle, formation of glutamate and glutamine via amination of a-ketoglutarate (GDH reaction) and glutamate (glutamine synthetase reaction), respectively, is essential for ammonia detoxification. Glutamate is an important building block for polypeptides and accounts for over 10% of amino acid residues present in most proteins. It is also a constituent of many biologically active oligopeptides such as glutathione, a tripeptide present at high concentrations intracellularly and thought to be involved in cellular mechanisms dealing with oxidative stress (see below).
GLUTAMATE AS EXCITATORY TRANSMITTER
Curtis and Watkins (17) demonstrated in 1960 that glutamate and other acidic amino acids can produce potent excitation of spinal neurons. However, the possibility that excitatory amino acids serve as neurotransmitters in mammalian CNS was initially discounted for a number of reasons. Glutamate and other acidic amino acids are present at high concentrations in all cells. These compounds show little regional variation in their distribution in mammalian brain and are also known to be involved in many aspects of intermediary metabolism. These characteristics are in contrast to those of other biologically active molecules (i.e., dopamine, norepinephrine, and acetylcholine) considered as classic neurotransmitters. In spite of these features, evidence has mounted over the past decade in favor of the neurotransmitter role. In fact, glutamate has been accepted as having satisfied the main criteria required for classification as an excitatory transmitter in mammalian CNS. (see The Psychopharmacology of Sexual Behavior, this volume).
A key feature that may allow glutamate to perform its many functions in nerve tissue is compartmentation of its distribution into distinct pools (21). The largest of these pools (about 50% of total) is related to the metabolism of neurons and is known as the metabolic pool. A smaller neuronal pool (20–30% of the total) is releasable from the nerve endings during neurotransmission and is thought to represent the glutamate neurotransmitter pool. A separate pool (10–30% of the total) is contained in glial cells and is believed to serve the recycling of transmitter glutamate (glial pool). The smallest glutamate pool (5% of the total) is believed to be involved in the synthesis of GABA (GABA precursor pool).
Transmitter glutamate is stored presynaptically in specific nerve endings (glutamatergic) where it can be released by a calcium-dependent mechanism (21). The extracellular concentration of glutamate is very low (1-3 µM) except during excitatory impulses when the concentration in the synaptic cleft can reach 1-2 mM (14). The synaptic action of the amino acid is believed to be terminated by rapid removal from the synaptic cleft via a high-affinity uptake system, which is sodium-dependent and which does not discriminate between glutamate or aspartate. This system is present both in the surrounding astrocytes and in the nerve terminals, with the glial uptake thought to be particularly efficient.
Glutamate uptake is accomplished via high affinity transport proteins (glutamate transporters or carriers) which are present on the plasma membrane of both neurons and glial cells. Initial efforts led to the cloning of three mammalian sodium-dependent excitatory amino acid transporters (EAAT). Two of these, originally designated GLAST (84) (present name: EAAT1) and GLT-1 (58)(present name: EAAT2), are expressed in brain and are localized in glial cells (32,58,84). However, more recent studies (40) revealed that GLT-1 (EAAT2) is also localized in neurons. The third one, originally designated EAAC1 (30) present name: EAAT3), is expressed both in brain (localized in neurons) and in peripheral tissues (30,76,77). We (75-77) and others (3)have cloned the corresponding human glutamate transporters. Recently, two additional glutamate transporters, designated EAAT4 (20) and EAAT5 (4), have been cloned: EAAT4 is predominantly expressed in the cerebellum and has the properties of a ligand-gated chloride channel (20), whereas EAAT5 is expressed predominantly in the retina (4).
With respect to the regional localization of the glutamate transporter proteins, immunocytochemical studies (22,23,32,78) using antibodies against synthetic peptides corresponding to specific regions of the deduced amino acid sequences for these proteins revealed that EAAT1 (GLAST) is preferentially expressed in the molecular layer of the cerebellum, whereas EAAT2 (GLT-1) is widely expressed in mammalian CNS. In the spinal cord, EAAT2 is expressed both in the ventral and dorsal horns, with the latter expression being particularly prominent. EAAT3 is also widely expressed in mammalian CNS where is localized in neuronal somata, dendrites and fine-caliber fibers. EAAT4 is predominantly expressed in the cerebellum where is localized in the Purkinje cells; lower levels of expression are found in the forebrain (23)
Synaptic glutamate, removed by uptake into the surrounding glial cells, is believed to be recycled via the glutamine/glutamate cycle (21). The first step of this cycle is amination of glutamate to glutamine via the action of glutamine synthetase, an enzyme of exclusive glial localization. Glutamine, thus formed, readily crosses cell membranes and is transported back to the nerve terminals where it can be converted to transmitter glutamate via the action of the neuronal enzyme glutaminase. In addition, glutamate taken up by glial cells may be oxidatively deaminated by GDH or transaminated by GOT to a-ketoglutarate, which can then enter the Krebs cycle and be oxidized to CO2 and H2O, or be recycled, serving as a precursor of transmitter glutamate.
Recycling of glutamate at the glutamatergic synapses seems to be essential for preventing depletion of the transmitter because the brain seems to have a rather limited capacity to synthesize glutamate de novo from glucose. Synthesis of glutamate from glucose removes a-ketoglutarate from the Krebs cycle which, if not replenished, will eventually result in the failure of the cycle. On the other hand, replenishment of five-carbon skeleton substrates of the Krebs cycle requires fixation of CO2 on pyruvate (Fig. 1) via either the pyruvate carboxylase or the malic enzyme reaction resulting in the formation of oxaloacetate and malate, respectively (anaplerotic reactions). These anaplerotic reactions are estimated to be only one-tenth as active in the brain as in the liver (89). However, enhanced CO2 fixation has been found in brain under conditions of increased ammonia load and attributed to de novo production of a-ketoglutarate, required for the synthesis of glutamate and glutamine as a means of ammonia detoxification.
Postsynaptic transduction of excitatory transmission is mediated by several classes of glutamate receptors . These include the NMDA (N-methyl-D-aspartate), the AMPA [amino-3-(3-hydroxy-5-methylisoxazol-4-yl)propionic acid]/kainate, and the metabotropic receptors. These receptors not only exhibit different ligand specificity, but they also differ in such features as duration and speed of postsynaptic transduction, desensitization, and other characteristics. The NMDA and the AMPA/kainate receptors are called ionotropic receptors since they are coupled to ion channels, whereas the metabotropic receptors mediate their action through G proteins.
The NMDA receptor is highly permeable to Ca2+ and can be blocked by magnesium in a voltage-dependent manner. NMDA, quinolinic and ibotenic acid are selective agonists, whereas ketamide and phencyclidine are selective antagonists at this receptor. Glycine has been shown to potentiate excitatory transmission by acting on a strychnine-insensitive allosteric site of the NMDA receptor (37) (Fig. 2). Transduction through the NMDA receptor produces slow but sustained physiologic responses. In addition to its role in glutamatergic transmission, the NMDA receptor is involved in neural development and activity-dependent synaptic plasticity. Long-term potentiation is a long-lasting increase in synaptic sensitivity that can be induced by high-frequency stimulation and that has been implicated in memory and learning processes. Molecular biological techniques have led to the cloning of two main subunits, designated NR1 and NR2, of the NMDA receptor in the rat (47). Subunit NR1 is ubiquitous and consists of seven isoforms generated by alternative mRNA splicing. Subunit NR2 is regionally localized and consists of four subtypes encoded by different genes.
The AMPA/kainate receptors seem to mediate predominantly fast excitatory synaptic transmission . Molecular cloning of these receptors has led to the realization that they exist physiologically at heteromeric combinations of multiple subunits (GluR1-GluR7; KA1-Ka2) with different ligand specificity (47). Depending on the particular subunit composition, the AMPA/kainate receptor may or may not be permeable to Ca2+ ions (47). CNQX and 2,3-benzodiazepine GYKI 52466 are known as nonselective and selective antagonists at the AMPA/kainate receptor, respectively. In contrast to the above receptors which are linked to ion channels, the metabotropic receptor mediates its action through G proteins. Activation of this receptor stimulates inositol 1,4,5-triphosphate (IP3) metabolism and is associated with mobilization of Ca2+ from intracellular stores. At the molecular biological level, the metabotropic receptor consists of at least eight subtypes (mGluR1 through mGluR8) (47,51). trans-(±)-ACPD [trans-(±)-1-amino-1,3-cyclopetane-dicarboxylic acid] is shown to be a selective agonist, and quisqualate a nonselective agonist, at the glutamate metabotropic receptor. (see Exciatotry Amino Acid Neurotransmission, this volume, for additional details)
The Neuroexcitotoxicity Concept
Over four decades ago Lucas and Newhouse (35) observed that the systemic administration of monosodium glutamate to experimental animals resulted in degeneration of retinal ganglion cells. Olney et al. (50) subsequently showed that excitatory amino acids given systemically to immature animals can cause neuronal degeneration in areas of the brain that lack an intact blood–brain barrier, such as the hypothalamus. They further noted that the neurotoxicity of acidic amino acids correlates with their ability to excite nerve cells and coined the term "neuroexcitotoxicity." These observations aroused considerable interest because the pattern of neuronal degeneration induced by the neuroexcitotoxic compounds are similar to that found in some disorders with system atrophy, such as in Huntington's disease. Whether the pathogenesis of these disorders relates to excitotoxic mechanisms, as originally suggested, remains unclear. The recent cloning of the gene responsible for Huntington's disease opens new avenues for testing this hypothesis and elucidating the primary pathogenetic process in this disease. Glutamate dysfunction has also been suggested in a variety of neuropsychiatric disorders, including schizophrenia (see Mood Disorders Linked to the Reproductive Cycle in Woman, this volume).
Neuroexcitotoxicity in Metabolic Encephalopathies
Around the time when the potential role of excitotoxicity in neurodegeneration was first suggested, evidence implicating altered glutamatergic mechanisms in nerve cell death induced by various metabolic insults began to accumulate. Initial studies on experimental thiamine deficiency encephalopathy revealed altered glutamate uptake and metabolism in the brains of deficient animals (59). Additional studies (31) have shown attenuation of brain lesions with the use of NMDA receptor antagonists. Also, metabolic alterations induced in the brain by the selective neurotoxin 3-acetylpyridine led to the finding that the glutamate-metabolizing enzyme glutamate dehydrogenase is reduced in patients with neurodegenerative disorders characterized by multiple system degeneration (60). The systemic metabolism of glutamate was found to be altered in these patients (60). Glutamatergic excitotoxic mechanisms have also been implicated in the nerve cell death that occurs in hypoxia and in hypoglycemia.
As indicated above, dysregulation of glutamate metabolism occurs in patients with multisystemic neurologic disorders associated with decreased glutamate dehydrogenase activity (60). Because motor neuron involvement is often encountered in such patients, it was suggested that innervation of anterior horn cells by the glutamatergic corticospinal fibers and interneurons renders these cells susceptible to the neurodegenerative process. The possibility was further raised (61) that abnormalities of glutamate metabolism may also occur in patients with primary ALS and be responsible for its neurodegeneration.
Plaitakis and Caroscio (61) accordingly measured amino acid levels in the fasting plasma from 22 patients with motor neuron disease (MND) (15 males, 7 females), 19 of whom had typical ALS (upper and lower motor neuron deficits). Results revealed that plasma concentrations of glutamate were selectively elevated (40.8 ± 12.7 mM; p < 0.001) in the ALS patients when compared to age-matched healthy controls (21.3 ± 7.9 mM) and to patients with other types of neurodegenerative or neuromuscular disorders (disease controls). Oral loadings with monosodium glutamate produced excessive elevations in plasma glutamate associated with proportional increases in plasma aspartate levels (61). Defective glutamate transport across cellular or mitochondrial membranes (glutamate-OH translocator linked to the oxidative deamination pathway) was thought to be responsible for the abnormal glutamate clearance. On the other hand, the rise in plasma aspartate was consistent with an intact transamination pathway, including a normally functioning glutamate/aspartate translocator (61,62). These metabolic abnormalities were similar to those observed in patients with GDH deficiency, although the activity of this enzyme was normal in leukocytes of ALS patients (61).
More extensive investigations were undertaken involving 88 patients with MND (64) . Of these, 62 had typical ALS, 23 had progressive bulbar palsy (PBP), and 3 had progressive muscular atrophy (PMA). As compared to control values, glutamate levels were increased by about 80% (p < 0.0001) in patients with typical ALS and by about 30% (p < 0.005) in those with PBP. No changes were found for PMA patients, but the number of such patients studied was very small. Nevertheless, it appeared that dysregulation of glutamate metabolism occurs primarily in patients with typical ALS.
Perry et al. (54) studied 28 patients with various types of motor neuron disease. Of these, 16 had typical ALS, four suffered from disease limited to lower motor neurons (i.e., PMA) (C. Krieger, personal communication), seven had PBP, and one had familial ALS. Glutamate values in normal controls were 25 ± 12 mM (N = 48), and those in the mixed group of 28 MND patients were 33 ± 19 mM (p < 0.05). Perry et al. (54) attributed these differences to patients' older ages, although another study by the same authors, which was reported concurrently and evaluated 98 controls (55), showed no effect of aging on plasma glutamate levels. If the data are analyzed according to disease type (66), patients with typical ALS have plasma glutamate levels of 38.2 ± 18.7 mM (N = 16), which are close to those obtained in our laboratory (61) as described above.
Blin et al. (7) studied 18 patients (12 males, 6 females) with primary ALS. They found significant glutamate elevations in the plasma of the ALS patients (168.3 ± 57.2 mM; p < 0.01) as compared to healthy controls (57.1 ± 31.7 mM; N = 16). Also, Iwasaki et al. (29) determined amino acid levels in the plasma of 10 ALS patients (6 males, 4 females) and compared them to 10 normal controls. Glutamate levels were significantly greater in the ALS (163.9 ± 117.3 mM; p < 0.001) than in the control group (34.1 ± 11.3 mM). Plasma aspartate and glycine levels showed lesser increases. Recently, Babu et al. (5) reported that blood glutamate levels were significantly higher in ALS patients than in controls. However, Shaw, et al. (81) found no significant differences in the fasting plasma levels of 22 amino acids between 37 patients with MND and 35 neurological controls.
Several investigators have measured amino acid levels in the cerebrospinal fluid (CSF) of ALS patients and have obtained seemingly conflicting results (54, 69, 70, 81). It is now well established that levels of glutamate in the CSF are extremely low (~0.3 mM), whereas those of glutamine are as high as those present in plasma (~660 mM). Therefore, measurement of CSF glutamate is much more difficult than that of plasma. Because glial cells have a tremendous capacity to eliminate transmitter glutamate, it remains uncertain whether any glutamate measured in CSF reflects that present in the CNS extracellular space. It has been our experience (unpublished data) that, in controls, CSF glutamate is present at trace levels or is nondetectable. Rothstein et al. (70), used a very sensitive analytical method to measure amino acids in the CSF and reported control glutamate values which were about 10-fold lower than those previously reported by the same authors utilizing different methodology (69). Both of these studies (69, 70) showed that glutamate and aspartate concentrations were significantly elevated in the CSF of ALS patients when compared to those of controls. Perry et al. (54), who have reported control CSF amino acid values similar to those of Rothstein et al. (70), did not find any differences in CSF glutamate levels between ALS patients and controls. However, Perry et al., (54) did not specify in their report the clinical syndromes of 17 ALS patients on whom CSF measurements were made. Recently, Shaw et al., (81) reported that CSF glutamate levels were increased in a subgroup of patients with MND. Hence, it is reasonable to conclude that the conflicting results on CSF glutamate in ALS may be attributable to the great technical difficulties involved in determining these levels and/or to the selection of patients.
Central Nervous System: Reduced Intracellular Glutamate Pools
In contrast to the elevated plasma and CSF levels, the concentration of glutamate was significantly decreased in all brain and spinal cord areas studied of ALS patients (36, 53, 62, 87). In absolute terms, the decrease in glutamate was the same in all brain areas studied (about 2 mM/gram wet tissue) (53, 62). However, there were greater proportional decreases in the spinal cord due to its normally lower glutamate content. Changes in glutamate levels were selective because other amino acids were not significantly altered except for aspartate, which was reduced in the cervical and the lumbar spinal cord of ALS patients (36, 62, 87).
Reductions in glutamate levels did not correlate with regional degenerative changes occurring in ALS. Thus, decreased glutamate content was found not only in pathologically affected areas such as the spinal cord, brainstem, and motor cortex (36, 53, 62, 87), but also in the basal ganglia, hippocampus, occipital cortex, and cerebellum (53, 62)—regions that are spared in the degenerative process. Also within the spinal cord, glutamate and aspartate decreases were observed even in the dorsal horns (36), which are pathologically spared in ALS. Since almost all glutamate measured in nerve tissue is intracellular (the extracellular levels are extremely small), these data are consistent with depletion of the intracellular CNS glutamate pools.
The above data are thought to suggest that the distribution of glutamate is altered between its intracellular and extracellular pools in ALS (61). Impaired glutamate transport across mitochondrial and/or cytoplasmic membranes may be implicated (62-63). Defective mitochondrial transport (glutamate-OH translocator) could affect the oxidation of glutamate by these organelles, whereas defective cytoplasmic transport, including impaired glial or neuronal high affinity uptake, as suggested by Plaitakis (63), could impair the detoxification of synaptic glutamate.
Rothstein et al. (71) have accordingly measured the high-affinity glutamate uptake in nerve tissue of ALS patients obtained at autopsy. They found decreased accumulation of [14C]glutamate by synaptosomes isolated from spinal cord, motor cortex, and somatosensory cortex . However, synaptosomes from visual cortex, hippocampus and striatum showed normal glutamate uptake. Kinetic analysis revealed decreased Vmax but normal Km. These findings are indicative of a decreased number of uptake sites but normal transport affinity. In addition, Shaw et al. (79), using quantitative autoradiography, showed that the specific binding of [3H]D-aspartate was decreased in the intermediate grey matter and in the substantia gelatinosa of the lumbar cord of ALS patients. Patients with progressive muscular atrophy (PMA) showed lesser changes. The authors suggested that these changes may be due to a loss of glutamatergic terminals of the corticospinal tract, which occurs primarily in ALS; involvement of this track is less conspicuous in PMA. Although is unclear whether these uptake changes are primary or secondary to the disease process, studies utilizing organotypic cultures of rat spinal cord have shown that chronic treatment with glutamate uptake inhibitors leads to loss of motor neurons (72).
Following the cloning of cDNAs encoding for three excitatory amino acid transporters, Rothstein et al. (73) used antibodies against synthetic oligopeptides corresponding to C terminal region of each of these transporters to study CNS tissue from ALS patients. Western blot analysis of tissue homogenates revealed a 30-95% decrease in EAAT2 immunoreactivity in 60-70% of patients with sporadic ALS. These changes were found in pathologically affected areas (spinal cord and motor cortex), but not in other CNS regions. The decrease in EAAT2 was selective since EAAT1 and EAAT3 immunoreactivity were not significantly altered. Patients with SOD1 mutations showed no significant decreases in EAAT2 immunoreactivity.
Rothstein’s group (9) originally reported that Northern blot analyses of brain and spinal cord tissue from ALS patients lacking EAAT2 showed no significant changes in the levels or the size of the mRNA encoding this protein. However, additional studies by the same group of investigators (34), using primarily the Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) method, led to the amplification of mRNAs that were distinct from the previously described human brain cDNAs specific for EAAT2 (3, 77). Two new mRNAs, designated “aberrant mRNA species” were detected in ALS patients: one that retains the 7th intron of the EAAT2 gene resulting in a truncated protein and another with exon 9-skipping. Surprisingly, these were limited to the pathologically affected CNS regions (34). The authors speculated that the “aberrant mRNAs” were the cause of the decreased EAAT2 in ALS, a contention supported by the possibility that the abnormal mRNAs could down regulate EAAT2 protein (34). However, Nagai et al. (46), who also used RT-PCR to study EAAT2 mRNA in ALS, have recently reported that such truncated transcripts are not disease-specific. In contrast to the data of Rothstein’s group (34), Nagai et al. (47) found truncated transcripts containing intronic sequences at the 3’ end of exon 7 and exon-8 skipping, both in ALS patients and in controls. Nagai et al. (47) attributed the presence of these truncated transcripts to alternative splicing.
Changes in N-Acetylaspartate and N-Acetylaspartylglutamate
In addition to glutamate, N-acetylaspartate (NAA) is also known to be present in high concentrations in the nerve tissue. In many regions of the mammalian CNS, NAA levels are second only to glutamate. As with glutamate, almost all NAA content is limited to the intracellular and, probably, neuronal compartment. N-acetylaspartylglutamate (NAAG), a dipeptide that can excite neurons, is also present at rather high concentrations intraneuronally. The contents of NAA and NAAG in human CNS exhibit a reciprocal distribution pattern consistent with rostrocaudal gradient of decreasing NAA and increasing NAAG concentrations (65). Because NAAG is thought to derive from NAA, this reciprocal distribution may reflect different regional turnover rates with the higher spinal cord NAAG, probably resulting from a high rate of conversion of NAA to NAAG.
Measurement of these compounds in postmortem brain and spinal cord tissue from ALS patients showed that both NAA and NAAG were significantly reduced in the cervical spinal cord of these patients by 40% and 48% (p < 0.001), respectively (16, 65, 87). In contrast, NAA and NAAG levels were not altered in postmortem frontal and cerebellar cortex of ALS patients (65). However, Pioro et al. (56)) using 1H-magnetic resonance spectroscopy detected decreased NAA levels in the cerebrum of ALS patients in vivo. Similar findings have been reported recently by Block et al. (8), who also used proton magnetic resonance spectroscopy to show that NAA levels are decreased in the motor cortex of ALS patients
More recently, Pioro ( 57) was able to detect glutamate and glutamine in vivo using short time (TE) proton magnetic resonance spectroscopy and spectroscopic imaging. He used creatine/phosphocreatine (Cr) as a reference point since these compounds (in contrast to neuron-specific NA groups) are distributed homogeneously throughout the brain. He reported preliminary data showing that, although the NA/Cr values were decreased in sensorimotor cortex and brain stem, Glu+Gln/Cr were increased in the medulla of these patients. The author could not determine whether elevation of Glu+Gln precedes or follows neuronal degeneration in these patients.
Glutamate Receptors in ALS
We (61-63) and others (69-71) have hypothesized that the abnormality in glutamate metabolism is associated with altered presynaptic glutamatergic mechanisms leading to an abnormally enhanced excitatory transmission mediated by glutamate receptors and selective degeneration of postsynaptic motor neurons. Disappearance of neurons that bear glutamate receptors is expected to cause a decrease in the density of these receptors. Allaoua et al. (1) have accordingly measured the density of glutamate receptors in the spinal cord of ALS patients and reported that the NMDA receptor was decreased in these patients. Virgo and de Belleroche (88) also found that mRNA levels for the NR-1 subunit of the NMDA receptor were significantly decreased (by 50%) in the ventral spinal cord region, but not in the dorsal region. However, the mRNA levels for the NR-2 subunit were decreased both in the dorsal horns (55%) and ventral horns (78%) of ALS patients (74).
Shaw et al. (80) studied the binding of [3H]MK-801, a selective agonist at the NMDA site. Results showed that [3H]MK-801 binding was decreased in the spinal ventral horns of ALS patients. In contrast, this binding was increased in the intermediate spinal gray matter and deeper layers of the motor cortex. Gredal et al. (25) also found that the increased [3H]MK-801 binding in several neocortical regions. This was associated with an enhanced binding affinity of the ligand for the receptor
In contrast to the changes of the NMDA receptor in ALS found by Allaoua et al. (1), the AMPA/kainate receptors were not altered (1). Morrison et al. (42) recently used SOD1 transgenic mice to study glutamate receptor changes. They reported that GluR2-immunoreactivity was not altered in these animals, making it unlikely that the motor neuron loss in this animal model is mediated via the AMPA/kainate receptor.
PATHOPHYSIOLOGY OF GLUTAMATE ALTERATIONS IN ALS
A Diffuse Metabolic Defect?
As indicated above, there are solid data showing that the glutamate content of brain and spinal cord is selectively decreased in ALS patients. In absolute terms, the decrease in glutamate levels is the same in all brain areas studied (about 2 mM/gram wet tissue). These changes do not correlate with regional degenerative changes, since they occur both in pathologically affected areas (spinal cord, brainstem, and motor cortex) and in CNS regions that are spared in the degenerative process (basal ganglia, hippocampus, occipital cortex, and cerebellum). In contrast, changes in the neuronal markers NAA and NAAG were limited to histologically affected CNS regions. These widespread decreases of nerve tissue glutamate cannot be attributed to neuronal loss since they occur in CNS regions that are histologically normal and maintain a normal content of the neuronal marker NAA.
The magnitude of glutamate decrease (approaching 50% in some CNS regions) is consistent with an involvement of the metabolic pool of the amino acid. This is the largest of all glutamate pools (estimated to be 50% of total). In contrast, changes limited to the neurotransmitter glutamate pool, resulting from degeneration of the glutamatergic nerve terminals, are not sufficient to account for these decreases. As mentioned above, Pioro (57) reported preliminary data based on proton magnetic resonance spectroscopy that showed that Glu+Gln/Cr are increased in the medulla of living ALS patients. These interesting data, if confirmed by additional studies, suggest that dysregulation of glutamate metabolism in brain in vivo is a dynamic process. It remains to be seen whether CNS glutamate increases during the active phase of neurodegeneration and whether this is followed by depletion of the amino acid at the end stages of the disease (as detected in post mortem tissues).
The CNS glutamate alterations, taken together with the systemic glutamate abnormalities, suggest that a generalized abnormality in glutamate metabolism occurs in ALS (61-63). To this date, however, the nature of this metabolic defect remains unclear. In this regard, observations made by Rothstein’s group are of great interest. The changes described by these investigators (2, 34, 71-73), involving the high affinity glutamate uptake, the EAAT2 protein immunoreactivity and the EAAT2 mRNA species were all limited to histologically affected CNS regions. In contrast, decreased glutamate content occurs throughout the CNS (see below). Meanwhile, detection of mutations involving the Zn/Mn SOD in familial ALS cases (see below), a metabolic enzyme showing widespread distribution in human tissues, has provided direct evidence that diffuse metabolic defects are capable of damaging motor neurons selectively.
A Glutamate Transport Defect?
Because almost all glutamate measured in nerve tissue is intracellular and the glutamate present in the plasma and CSF may reflect the extracellular concentrations, the data described above are consistent with an altered distribution of the amino acid between its intracellular and extracellular pools (62). Plaitakis et al (63) further suggested that a defect in the transport of glutamate or an increased release of intracellular pools to extracellular space could be operational. The finding of decreased accumulation of [14C]glutamate by spinal cord and brain synaptosomes of ALS patients, as reported by Rothstein et al.(71 ), is consistent with this postulated transport defect. Decreased glutamate transport is expected to disrupt the recycling of this transmitter with resultant depletion of its intracellular stores (decreased nerve tissue glutamate content) and increased synaptic (extracellular) levels. However, a careful consideration of the available data reveals that a correlation between regional changes in glutamate content and in the high-affinity uptake is lacking. Thus, in the striatum and visual cortex, areas showing decreased glutamate levels (53), synaptosomal glutamate uptake was found to be normal (71). Hence, defective uptake cannot adequately account for the widespread depletion of CNS glutamate content in ALS.
Because experimental lesioning of glutamatergic pathways is known to lead to decreased glutamate uptake, it is possible that loss of uptake, as described by Rothstein et al. (71), reflects disappearance of glutamatergic nerve endings in ALS. Shaw et al. (79), who studied the spinal cord of patients with ALS and PMA, reported that the decreased [3H]-aspartate binding in these disorders correlates with the loss of glutamatergic terminals of the corticospinal tract. Observations on two studies of mouse models of ALS are consistent with this possibility. In the Mnd mouse, a genetic mutant used as a model for adult-onset MND disease (6), decreased glutamate uptake was found in CNS regions analogous to those of ALS patients (spinal cord and motor cortex but not in the striatum). These uptake changes did not precede the onset of the neurologic abnormalities but, instead, they occurred after the development of the neuropathological changes. As with the human data, the high-affinity uptake of other neurotransmitters was not altered (6). Hence, a primary genetic defect affecting the glutamate/aspartate transport system seems unlikely. Similar data have been obtained in mice expressing the dominant mutation of human copper/zinc superoxide dismutase (SOD1). In this animal, a model for familial ALS caused by a specific genetic defect, decreased glutamate uptake in the spinal cord occurs late in the course of their disease (13), probably reflecting loss of motor neurons. Alternatively, oxidative damage to cellular membrane components, such as EAAT2, could be responsible for the decreased glutamate uptake. However, the EAAT2 was found to be normal in patients with the SOD1 mutation (73).
A Primary EAAT2 (GLT-1) Transporter Defect?
As described above, Rothstein et al. (73) reported that Western blot analysis of postmortem tissues revealed that EAAT2 immunoreactivity was significantly decreased in the motor cortex and spinal cord of ALS patients. However, recent immuno-cytochemical studies by Fray et al. (22) revealed that, although EAAT2 immunoreactivity was decreased in the spinal cord, this transporter was increased in the middle laminae of the motor cortex. The authors suggested that glutamate pathology in MND may be a more complex phenomenon than previously thought (22). It is of interest that changes in EAAT2 immunoreactivity, as reported by Fray et al. (22 ), are similar to those of the NMDA receptor (25,80) and may imply that parallel alterations in synaptic elements occur in the CNS of ALS patients.
Although decreases in EAAT2 immunoreactivity, as described by Rothstein et al. (73), appeared to be marked for some sporadic ALS cases, the implications of these findings remain uncertain. The fact that the EAAT2 alterations are limited to the pathologically affected CNS regions raises the possibility that they may represent a consequence of glutamatergic nerve terminal degeneration. There is already ample experimental evidence showing that lesioning of glutamatergic pathways decreases the expression of EAAT2 protein by glial cells. Levy et al., (33) demonstrated that cortical lesions capable of decreasing striatal glutamate uptake lead to reduced expression of the EAAT2 (GLT-1). Gegelashvili et al. (24) described that neuronal soluble factors regulate the expression of this transporter by cultured astroglia. Hence, disappearance of nerve terminals, as occurring in ALS, could eliminate the chemical signal that induces expression of this transporter by the surrounding astrocytes.
Although Northern blot analyses of ALS tissues by Rothstein’s group (9) revealed no significant changes in the levels or in the size of the mRNA for EAAT2, additional studies by the same group of investigators (34) revealed the presence of EAAT2 mRNAs that were distinct from the previously described human brain cDNAs specific for EAAT2 (3, 77). These were designated “aberrant mRNA species” and thought to encode truncated proteins. However, previous studies by these investigators (73) using Western blots failed to show the presence of truncated EAAT2 proteins in ALS tissues. Nagai et al. (46) found truncated transcripts (mRNAs) both in ALS patients and controls and attributed these changes to alternative splicing of GLT-1 mRNA. These authors concluded that these mRNAs are the result of physiological processes and, as such, they do not play a pathogenetic role in ALS.
The possibility of a primary genetic defect involving the EAAT2 is not supported by recent data obtained by Aoki et al. (2), who performed structural analyses of the gene encoding EAAT2. These authors used genomic DNA isolated from ALS patients on whom the Rothstein’s group had detected abnormalities both in EAAT2 protein and mRNA (34,73). Results revealed that these patients had no structural abnormalities of the gene encoding EAAT2. Hence, the mechanism by which the “aberrant EAAT2 mRNAs” are generated [if not through the physiological process of alternative splicing, as suggested by Nagai et al. (46)] remains a mystery. In addition, data in EAAT2 gene null mice do not support a primary role for this transporter in motor neuron degeneration. Tanaka et al. (85), who generated mice lacking EAAT2, reported that the clinical manifestation of these animals is generalized seizures rather than ALS. It is known from clinical observations that epilepsy is not a feature of ALS.
In conclusion, the new data generated by Rothstein and his co-workers on EAAT2 are quite exciting and may lead to a better understanding of ALS. However, more extensive studies are needed to test whether the abnormalities described by these investigators are causally related to this disorder or represent consequences (epiphenomena) of the neurodegenerative process.
A Primary Membrane Abnormality?
An alternative possibility that could account for the altered glutamate distribution in nerve tissue is an increased release of glutamate from the nerve terminals and/or increased leakage of the amino acid through a defective cell membrane. This is expected to lead to depletion of the intracellular stores and to increased extracellular glutamate levels. The finding that, in addition to glutamate, other compounds with a high intracellular/extracellular gradient, such as aspartate, NAA and NAAG, are also reduced in nerve tissue (16, 62, 87) and increased in the CSF (69) is consistent with this possibility. NAA is an anion that may contribute substantially to intraneuronal osmotic pressure. Hence, reduction in the nerve tissue levels of NAA is indicative of a diffuse membrane abnormality that may permit compounds with high intracellular concentrations to leak out of the cell (62). This is expected to increase the workload of the various membrane pumps responsible for transporting these substances against a concentration gradient. Ultimately, failure of these systems may lead to an accumulation of toxic amounts of glutamate at the synapses and degeneration of the postsynaptic motor neurons. Detection of mutations affecting the Cu/Zn superoxide dismutase in familial ALS (68), a defect capable of affecting membrane permeability through lipid peroxidation (see below), is consistent with this possibility. Recent studies by Block et al. (8) using proton magnetic resonance spectroscopy to evaluate the primary motor cortex of ALS patients revealed alterations in the inositol and choline, two compounds associated with plasma membrane metabolism.
A Problem of Energy Metabolism?
Impaired energy metabolism or increased oxidative stress (see below) could render the nerve cells incapable of maintaining their high intracellular/extracellular gradients for many biologically important compounds. Decreased glucose utilization has been shown by the use of the PET technology in ALS patients (18). Defective glial uptake or metabolism could also lead to decreased detoxification of transmitter glutamate and impaired recycling of the amino acid at the nerve terminals. Recently, Wiedemann et al. (90) studied mitochondrial function in skeletal muscle homogenates of 14 patients with sporadic ALS. They reported that mitochondrial oxidative phosphorylation and the specific activity of NADH:CoQ oxidoreductase were significantly decreased in the ALS patients as compared to 28 age matched controls. The authors considered unlikely that these abnormalities are due to neurogenic atrophy, since they were not present in muscle homogenates of patients with spinal muscular atrophy.
MUTATIONS OF Cu/Zn SUPEROXIDE DISMUTASE IN FAMILIAL ALS
The use of the "reverse genetic approach" has led to a major breakthrough in elucidating the primary genetic abnormality of a subgroup of familial ALS cases. The initial step was the linkage of dominant ALS to human chromosome 21q22.1, a region containing the gene for the Cu/Zn superoxide dismutase (SOD1). A tight linkage was found between the disease in several ALS families and the SOD1 gene, while direct sequencing of this gene revealed the presence of 11-point mutations in 13 families (68). Further investigations revealed that over 50 missense mutations occur in patients with familial ALS. About 20% of all familial ALS cases are known to have mutations involving the SOD1 gene. Enzyme activity was reduced (by about 50%) in the red cells and in the brain and spinal cord (10) of patients with these mutations. Given the presence of SOD1 in almost all cellular systems, the selective degeneration of motor neurons is a remarkable phenomenon as yet unexplained.
Defective Handling of Oxidative Stress
Cu/Zn SOD is a cytosolic enzyme that is known to dismutate superoxide () generated intracellularly during oxidation of various compounds, such as hypoxanthine to xanthine via xanthine oxidase (83). This results in the formation of H2O2, which is then rapidly converted to H2O and O2 via the action of catalase, an enzyme present at high levels in most cells. Also, oxidation of glutathione via glutathione peroxidase is another way by which H2O2 can be converted to H2O. Given the high intracellular levels of glutathione, this tripeptide may play an important role in cellular mechanisms protective against oxidative stress.
Malfunction of the cytosolic SOD is expected to impair the ability of the cell to eliminate produced during certain oxidation reactions. The nondetoxified superoxide radical can oxidize a number of cellular systems, particularly lipids that are constituents of the many membrane systems present in each cell (83). Peroxidation of cell membrane lipids, in particular, is expected to alter the membrane properties. Membrane fluidity may be affected along with membrane permeability to various compounds. Because the cytoplasmic membrane is essential for the transport of many biologically important substances, these processes are expected to be impaired.
On theoretical grounds (66), an increased membrane permeability to substances with high intracellular concentration, such as glutamate, is expected to lead to leakage of these substances out of the cell, resulting in an increased workload for the various membrane systems responsible for transporting these substances against a concentration gradient. This, in turn, is expected to enhance the metabolic demands of the cells and perhaps lead to increased superoxide formation. Peroxidation of membrane lipids may also occur in other membranous structures of the cell such as the endoplasmic reticulum (83), the Golgi apparatus, and the outer mitochondrial and nuclear membranes. In this regard, it may be relevant that fragmentation of the Golgi apparatus has been shown to occur in ALS (43). Also, oxidation of enzymes may impair energy metabolism or other metabolic processes (83). Given the limited ability of to penetrate membranes, this oxidative damage may involve primarily cytosolic enzymes.
In light of the above considerations, it is of interest that Ghadge et al. (26 ) showed that expression of mutant SOD1 in PC12 cells is associated with higher rates of superoxide production under various conditions. Hall et al. (27) detected an increase in spinal cord lipid peroxidation in transgenic mice with SOD1 mutations. In addition, Shaw et al. (82) detected increased protein carbonyl levels in the spinal cord of patients with sporadic motor neuron disease, thus indicating increased oxidative damage in these patients.
With respect to the effects of increased oxidative stress on energy metabolism, Browne et al. (10) recently reported that ALS patients with SOD1 mutations show marked increases in the mitochondrial Complex I and II-II activities. Similar findings were also reported in transgenic mice overexpressing human mutant SOD1 (26). Whether or not these abnormalities involving mitochondrial energy metabolism reflect an adaptation to enhanced energy demand, as postulated above, it remains to be further studied.
Other investigators have focused their attention on structural proteins. Crow et al. (15) suggested that SOD1-catalyzed nitration of neurofilament-L may play a role in the ALS pathogenesis. Neurofilament pathology has been identified in familial ALS cases associated with mutations of neurofilament proteins. Bruijn et al. (11) recently suggested that SOD1 mutations cause aggregation of the mutated protein. This, along with coaggregation of unidentified essential components of the cell and/or aberrant catalysis by misfolded SOD1 mutant enzyme could account for the MND pathology (11).
Although the SOD1 defect is limited to a small percentage of ALS patients, its recognition is of paramount importance for the following reasons: (a) A motor neuron disease, similar in all respects to sporadic ALS, can result from a generalized metabolic abnormality, thus indicating that such defects are capable of damaging motor neurons selectively; (b) the consequences of SOD1 abnormalities on cellular functions are precisely those implicated in the pathogenesis of primary ALS, which have been associated with altered glutamate metabolism (increased membrane permeability, impaired transport, and/or energy metabolism); and (c) there are probably myriad abnormalities with pathophysiological consequences similar to those induced by SOD1 malfunction, and this is consistent with the proposed multifactorial origin of ALS.
THE PROBLEM OF SELECTIVE MOTOR NEURON VULNERABILITY
Although the defect in glutamate metabolism in primary ALS is generalized, motor neurons are the only targets of the degenerative process. The same is also true for the familial ALS, which has been linked to Cu/Zn SOD mutations because this enzyme is expressed in many tissues. As such, there is no reason to believe that the above metabolic abnormalities are limited to the motor pathways. If this is so, why are only motor neurons affected?
The Glycine-Potentiated Excitotoxic Hypothesis
Based on the pattern of neuronal connectivity and the characteristics of glutamate receptors, Plaitakis (63) suggested that the glycinergic co-innervation renders motor neurons the selective targets of a glutamate-mediated neurodegenerative process. In motor neurons, unlike other neuronal systems, inhibition is in large part mediated through glycinergic interneurons. Glycine released from such glycinergic nerve terminals is known to inhibit motor neurons by binding to a Cl- channel-linked glycinergic receptor, which is strychnine-sensitive. In addition, glycine has been shown to potentiate glutamatergic transmission by acting on a strychnine-insensitive allosteric site of the NMDA subtype of glutamate receptor (37). Glycine increases the frequency of the NMDA receptor channel openings by shortening the desensitization period that follows glutamate action on this receptor (37). Desensitization of glutamate receptors seems to be of importance, since it may represent the main mechanism by which the neurotransmitter action of glutamate is terminated. Because both dorsal and ventral horn neurons express NMDA receptors in high density (80), the differential sensitivity of spinal cord neurons to ALS neurodegeneration cannot solely relate to the presence of NMDA receptors on these cells.
Studies on cultured chick motor neurons have shown that innervation by interneurons potentiates glutamate transmission (49). A substantial proportion of such interneurons are glycinergic, projecting to motor neurons with glycine levels known to be greater in the ventral than in the dorsal horns. There are also indications that motor neurons receive a dense glutamatergic innervation (39). Mitchell et al. (39) found that Na+-dependent glutamate transport, expected to be associated with glutamatergic synaptic elements, is more densely expressed in the ventral than the dorsal horns of the cat. Also, motor neurons that degenerate in ALS are surrounded by astrocytes expressing high levels of the EAAT2 (39).
Although the glycine allosteric site at the NMDA receptor is thought to be fully saturated by normal levels of nerve tissue glycine, Thomson et al. (86) obtained enhancement of excitatory postsynaptic potentials in cerebral cortex slices by applying increased concentrations of glycine. Similarly, Budai et al. (12) obtained enhancement of NMDA-evoked neuronal activity by glycine in the rat spinal cord in vivo and concluded that the glycine sites on NMDA receptors were not saturated.
Under conditions of enhanced synaptic glutamate, postsynaptic transduction seems to be mediated primarily by the NMDA receptor. As such, desensitization of this receptor may be of particular importance for protecting postsynaptic neurons from neurotoxicity. In the motor neurons, however, the presence of high glycine levels may lead to prolonged openings of the NMDA-linked channels, with resultant excessive entrance of Ca2+ into these cells. This, in turn, is expected to activate intracellular proteases and lipases, which can damage the cell by inducing a variety of secondary changes. Also, prolonged depolarization of postsynaptic neurons by glutamate analogues is shown to cause ATP depletion and accumulation of intracellular purine metabolites (48). Depletion of ATP is thought to impair the function of ion-dependent ATPases and thus membrane fluxes. Studies on kainate toxicity in cerebellar slices showed that, in addition to ATP depletion, an increased leakage of glutamate and aspartate occurred into the medium (48). However, the glycine hypothesis does not readily explain the degeneration of corticospinal neurons as well as the sparing of the oculomotor neurons that occur in ALS. In this regard, the pattern of NMDAergic and glycinergic synapses in the motor cortex in health and disease remains to be further studied.
The Glutamate Transporter Hypothesis
An alternative hypothesis suggested by Medina et al (39) and Milton et al (40) is that neuronal vulnerability in ALS relates to the level of expression of glutamate transporters. These investigators (39,40) found that motor neurons vulnerable to ALS neurodegeneration are surrounded by astrocytes, which show a higher level of expression of EAAT2 than glial cells present around motor neurons resistant to this neurodegeneration (oculomotor nuclei). However, this hypothesis cannot explain the overall pattern of CNS degeneration in ALS, since the highest levels of EAAT2 expression occurs in the caudate nucleus, the nucleus basalis of Meynert and hippocampus (32). These brain regions are not ordinarily involved in ALS.
Glutamate or Oxidative Stress: Which Is Primary?
As already discussed, altered glutamate metabolism has been shown in patients with primary ALS, which accounts for the overwhelming majority of patients with MND. These patients do not seem to have mutations of the SOD1 gene, which are specifically associated with familial ALS. Since the glutamatergic abnormalities detected in primary ALS suggest a defect in membrane permeability and since SOD1 malfunction could damage membranes via lipid peroxidation (see above), it is possible that membrane abnormalities induced by diverse etiologies may cause altered presynaptic glutamatergic mechanisms and selective degeneration of motor neurons.
As indicated above, glycine-induced potentiation of NMDA function, could increase the metabolic demands of motor neurons and/or make them bear, perhaps temporarily, increased Ca2+ loads. This may lead to an increased production of superoxide radicals (38). ATP breakdown is thought to generate increased amounts of hypoxanthine (38, 48), oxidation of which may lead to superoxide formation (83). Normal cells may compensate by enhancing their defenses against oxidative stress. However, motor neurons with SOD1 malfunction may be less well able to handle oxidative stress. Over the years, accumulation of superoxide radicals may damage cell membranes through lipid peroxidation, causing altered glutamate distribution and thus initiating a vicious cycle. Whether this explains the rapid downhill course patients often experience after remaining disease-free for decades remains to be established.
There is evidence that neuroexcitotoxicity is associated with increased oxidative stress that may contribute to nerve cell death (44). Kainate-induced degeneration of cultured cerebellar neurons was shown to be prevented by inhibiting xanthine oxidase or the formation of this enzyme from xanthine dehydrogenase (38). Another mechanism by which glutamate toxicity may lead to increased oxidative stress is inhibition of cystine transport with a resultant reduction in intracellular glutathione levels (44). Conversely, failure of antioxidant mechanisms seems to potentiate neuroexcitotoxic cell damage. Also, the pattern of motor neuron degeneration induced by excitotoxin agonists is similar to that seen in the SOD-1 transgenic mouse model of ALS (28). As such, multiple metabolic abnormalities capable of interfering with different aspects of glutamate transmission and/or cellular defenses against oxidative stress may underlie primary ALS. The primary abnormality(ies) remains to be defined, and therefore the questions raised by these observations are an important challenge to modern neuroscience.
This work was supported by NIH grant NH-16871 and by Mount Sinai Research Center grant RR00071.