**2. General characterization of mitochondria in ALS**

Clinical features of ALS, a severe neuromuscular degenerative disease, were described by Charcot in 1874 (Rowland & Shneider, 2001). Since that time clinical definitions (Shook and Pioro, 2009) and pathological features of the disease have been greatly expanded, aided in no small measure by advances in genetics of the disease (Rowland & Shneider, 2001) and development of transgenic animal models of ALS (Bendotti & Carri, 2004; Matsumoto et al., 2006; Howland et al., 2002). However, the etiology and pathogenesis of the disease remain poorly understood.

Clinical and experimental evidence showed that ALS is a systemic disease, with particular vulnerability of motor neurons due to some unique properties (Martin et al., 2007; von Lewinski & Keller, 2005; Panov et al., 2011a). Many ALS patients are hypermetabolic, an early and persistent phenomenon (Bouteloup et al., 2009, Desport et al., 2005). Muscular mitochondrial function in amyotrophic lateral sclerosis is progressively altered in ALS patients (Echaniz-Laguna et al., 2006, Krasnianski et al., 2005), and subtle ultrastructural changes of hepatocytes and liver dysfunction have also been described in biopsy specimens from ALS patients (Nakano et al., 1987). Significant changes were also found in skeletal muscle mitochondria of transgenic SOD1 (tgSOD1) animals (Dupuis et al., 2009, Krasnianski et al., 2005). It was suggested that increased ROS generation by skeletal muscle mitochondria (Muller et al., 2007) or mitochondrial uncoupling (Dupuis et al., 2009) may be primarily responsible for the loss of neuromuscular junctions and secondary distal degeneration of motor neurons in SOD1 mice. This is an interesting and important alternative hypothesis, which requires a more detailed and critical discussion.

The paper by Muller et al. (2007) presents data on generation of ROS by skeletal muscle mitochondria (SMM) from two lines of transgenic mice bearing different mutations in SOD1 gene, SOD1 knockout mice, and mice with denervated muscle. The authors

Role of Neuronal Mitochondrial Metabolic Phenotype in Pathogenesis of ALS 229

(Israelson et al., 2010); and abnormal axonal transport (Magrane & Manfredi, 2009). However, almost all of these proposed pathogenic mechanisms directly or indirectly involve mitochondria. The major problem in searching for alternative pathogenic mechanisms, in case of severe mitochondrial disability, is that practically all cellular and organ functions depend on energy and thus also become disturbed (Gershoni et al., 2009). Therefore, it is of paramount importance to precisely understand the nature and causes of mitochondrial

In the majority of studies of ALS, the mitochondrial abnormalities were assessed using morphological methods, which showed swollen brain and spinal mitochondria with dilated cristae, and vacuoles in matrix (Sasaki et al., 2004, Jaarsma et al., 2006). It was suggested that vacuolization may depend on accumulation amount, rather than from toxic property, of mutant SOD1 (Jaarsma, 2006). Morphologic methods, including quantitative histochemistry, and immunological methods, showed that both in ALS patients and tgSOD1 animals, the activities of respiratory chain complexes were diminished (Jung et al., 2002), and that the spinal cord mitochondria are predisposed to opening of the Ca2+-dependent permeability transition pores (mPTP), which trigger death of motor neurons (Martin et al., 2009). There is a large body of evidence that spinal cords in tgSOD1 animals are subjected to increased oxidative and nitrative stress, as indicated by accumulation of protein carbonyls and nitrated proteins (Martin et al., 2007, 2009). These data also demonstrated that mitochondria are involved early in the pathogenesis of ALS (Martin et al., 2007, 2009). Recently, Panov et al. (2011b) provided direct evidence that isolated mitochondria from the brain and spinal cord of tgSOD1 rats generate significantly more reactive oxygen species (ROS) as compared

Morphological, histochemical and immunological methods provided important information about involvement of mitochondria in the pathology, however they do not tell us what caused the changes in mitochondrial morphology. Relatively few studies were performed on isolated BM and SCM mitochondria from control (Sullivan et al., 2004, Panov et al., 2011a), tgSOD1 animals (Kirkinezos et al., 2005, Damiano et al., 2006, Fischer et al., 2011, Panov et al., 2011b), Wobbler mice (Dave et al., 2003, Xu et al., 2001), or *mnd* mice, which spontaneously develop motor neuron pathology (Bertamini et al., 2002). In their pioneering work, Sullivan et al (2004) have shown that normal SCM have lower respiratory activity, lower threshold for calcium-induced mitochondrial permeability transition, and a higher degree of superoxide radical production (determined with dehydroethidium) and lipid peroxidation as compared with BM. However, the dehydroethidium method does not identify the origin of superoxide radical. Studies of isolated mitochondria from tgSOD1 mice showed that BM and SCM had decreased rates of respiration associated with the loss of complex IV activity (Kirkinezos et al., 2005). Very early in the course of the disease, the mitochondrial Ca2+ loading capacity was significantly diminished as compared with the age-matched control mice (Damiano et al., 2006). Respiratory activities were also diminished in SCM of Wobbler mice (Dave et al., 2003, Xu et al., 2001) and mice with motor neuron

A comparative study of isolated normal and tgSOD1 brain and spinal cord mitochondria was published recently by Panov et al. (2011a, b), and the unique metabolic features of BM and SCM were described (see also Panov et al, 2009). Normal SCM produced significantly more ROS than BM, and this was associated with the succinate-dependent reverse electron transport (Panov et al. 2011a). A rather unexpected finding was that BM and SCM isolated

dysfunctions, and whether they are early or late events in disease evolution.

with mitochondria from wild type animals.

degeneration disease (Bertamini et al., 2002).

concluded that "enhanced generation of mitochondrial ROS may be a common factor in the mechanism underlying denervation-induced atrophy". The authors suggested also that increased ROS generation in ALS mutant model may also cause the loss of neuromuscular junctions, and thus initiate the loss of motor neurons. There are, however, experimental facts, which oppose this conclusion. Muller et al (2007) isolated skeletal muscle mitochondria using a protease nagarse for digestion of the tissue. With the nagarse method, more than 90% of mitochondria are interfibrillar. Without the protease the yields of subsarcolemmal mitochondria are very low (A. Panov, unpublished observations). ROS generated *in situ* by interfibrillar mitochondria are unlikely to gain access to the sarcolemmal membrane where the neuro-muscular junctions are located. This is because superoxide radical has difficulty to leave mitochondria (Panov et al. 2005). Besides, both superoxide radicals and H2O2 are substrates for the antioxidant enzymes in the cytoplasm, such as SOD1, catalase, etc. Rather, it is more likely that increased ROS act locally to cause damage to interfibrillar mitochondria, particularly to mtDNA, and thus cause muscle atrophy. Data presented by Muller et al. (2007) do not show direct damage to neuromuscular junctions by increased ROS. Therefore this issue requires further study. There is also a quantitative aspect of ROS generation presented by Muller et al. (2007). A comparison of the rates of ROS production by skeletal muscle mitochondria (SMM) (Figure 1 in Muller et al., 2007) and brain (BM) and spinal cord mitochondria (SCM) (see Fig. 7 in Panov et al., 2011a) shows that BM and SCM generate several times more ROS than SMM. Therefore, it is more likely that the causes of degeneration of motor neurons in ALS are associated with primary pathological processes in neurons rather than in muscle.

Dupuis et al. (2009) also concluded that "a muscle restricted mitochondrial defect is sufficient to generate motor neuron degeneration". The authors based their conclusion on mitochondrial dysfunctions observed in animals with overexpression of uncoupling protein 1 (UCP1). Overexpression of UCP1 results in deenergization of the mitochondria due to increased conductivity of the inner membrane for protons. This condition results in complete inhibition of ROS production associated with the energy-dependent reverse electron transport. The major consequence of overexpression of UCP1 is inhibition of ATP production and suppression of functions, which depend on mitochondrial ATP levels. Because in neuronal tissue reverse electron transport is the major source of ROS production (Panov et al. 2007, 2009), and increased ROS generation is one of the leading pathogenic pathways in ALS, the mechanism proposed by Dupius et al. (2009), although interesting, probably has little to do with neurodegeneration in ALS.

It was hypothesized that sporadic and familial forms of ALS may share some final pathogenic mechanisms (Bendotti & Carri, 2004; von Lewinski & Keller, 2005). Several pathogenic factors have been proposed (reviewed in Bruijn et al., 2004): glutamate excitotoxicity (Heath & Shaw, 2002); mitochondrial dysfunction (Dupuis et al., 2009); increased production of reactive oxygen species (ROS) (reviewed in Barber & Shaw, 2010) and nitroxyradicals (NO- and ONOO-) (Martin et al., 2007); deregulation of Ca2+ homeostasis and triggering of permeability transition of mitochondria (von Lewinski and Keller, 2005; Martin et al., 2009; Morota et al., 2007); induction of pro-apoptotic pathways (Przedborski, 2004); misfolding and intracellular accumulation of mutant SOD1 (Tateno et al., 2004), including in brain and spinal mitochondria (Liu et al., 2004; Higgins et al., 2002); mutant SOD1-induced conformational changes in Bcl-2 (Pedrini et al., 2010) and VDAC1 (porin)

concluded that "enhanced generation of mitochondrial ROS may be a common factor in the mechanism underlying denervation-induced atrophy". The authors suggested also that increased ROS generation in ALS mutant model may also cause the loss of neuromuscular junctions, and thus initiate the loss of motor neurons. There are, however, experimental facts, which oppose this conclusion. Muller et al (2007) isolated skeletal muscle mitochondria using a protease nagarse for digestion of the tissue. With the nagarse method, more than 90% of mitochondria are interfibrillar. Without the protease the yields of subsarcolemmal mitochondria are very low (A. Panov, unpublished observations). ROS generated *in situ* by interfibrillar mitochondria are unlikely to gain access to the sarcolemmal membrane where the neuro-muscular junctions are located. This is because superoxide radical has difficulty to leave mitochondria (Panov et al. 2005). Besides, both superoxide radicals and H2O2 are substrates for the antioxidant enzymes in the cytoplasm, such as SOD1, catalase, etc. Rather, it is more likely that increased ROS act locally to cause damage to interfibrillar mitochondria, particularly to mtDNA, and thus cause muscle atrophy. Data presented by Muller et al. (2007) do not show direct damage to neuromuscular junctions by increased ROS. Therefore this issue requires further study. There is also a quantitative aspect of ROS generation presented by Muller et al. (2007). A comparison of the rates of ROS production by skeletal muscle mitochondria (SMM) (Figure 1 in Muller et al., 2007) and brain (BM) and spinal cord mitochondria (SCM) (see Fig. 7 in Panov et al., 2011a) shows that BM and SCM generate several times more ROS than SMM. Therefore, it is more likely that the causes of degeneration of motor neurons in ALS are associated with primary pathological

Dupuis et al. (2009) also concluded that "a muscle restricted mitochondrial defect is sufficient to generate motor neuron degeneration". The authors based their conclusion on mitochondrial dysfunctions observed in animals with overexpression of uncoupling protein 1 (UCP1). Overexpression of UCP1 results in deenergization of the mitochondria due to increased conductivity of the inner membrane for protons. This condition results in complete inhibition of ROS production associated with the energy-dependent reverse electron transport. The major consequence of overexpression of UCP1 is inhibition of ATP production and suppression of functions, which depend on mitochondrial ATP levels. Because in neuronal tissue reverse electron transport is the major source of ROS production (Panov et al. 2007, 2009), and increased ROS generation is one of the leading pathogenic pathways in ALS, the mechanism proposed by Dupius et al. (2009), although interesting,

It was hypothesized that sporadic and familial forms of ALS may share some final pathogenic mechanisms (Bendotti & Carri, 2004; von Lewinski & Keller, 2005). Several pathogenic factors have been proposed (reviewed in Bruijn et al., 2004): glutamate excitotoxicity (Heath & Shaw, 2002); mitochondrial dysfunction (Dupuis et al., 2009); increased production of reactive oxygen species (ROS) (reviewed in Barber & Shaw, 2010) and nitroxyradicals (NO- and ONOO-) (Martin et al., 2007); deregulation of Ca2+ homeostasis and triggering of permeability transition of mitochondria (von Lewinski and Keller, 2005; Martin et al., 2009; Morota et al., 2007); induction of pro-apoptotic pathways (Przedborski, 2004); misfolding and intracellular accumulation of mutant SOD1 (Tateno et al., 2004), including in brain and spinal mitochondria (Liu et al., 2004; Higgins et al., 2002); mutant SOD1-induced conformational changes in Bcl-2 (Pedrini et al., 2010) and VDAC1 (porin)

processes in neurons rather than in muscle.

probably has little to do with neurodegeneration in ALS.

(Israelson et al., 2010); and abnormal axonal transport (Magrane & Manfredi, 2009). However, almost all of these proposed pathogenic mechanisms directly or indirectly involve mitochondria. The major problem in searching for alternative pathogenic mechanisms, in case of severe mitochondrial disability, is that practically all cellular and organ functions depend on energy and thus also become disturbed (Gershoni et al., 2009). Therefore, it is of paramount importance to precisely understand the nature and causes of mitochondrial dysfunctions, and whether they are early or late events in disease evolution.

In the majority of studies of ALS, the mitochondrial abnormalities were assessed using morphological methods, which showed swollen brain and spinal mitochondria with dilated cristae, and vacuoles in matrix (Sasaki et al., 2004, Jaarsma et al., 2006). It was suggested that vacuolization may depend on accumulation amount, rather than from toxic property, of mutant SOD1 (Jaarsma, 2006). Morphologic methods, including quantitative histochemistry, and immunological methods, showed that both in ALS patients and tgSOD1 animals, the activities of respiratory chain complexes were diminished (Jung et al., 2002), and that the spinal cord mitochondria are predisposed to opening of the Ca2+-dependent permeability transition pores (mPTP), which trigger death of motor neurons (Martin et al., 2009). There is a large body of evidence that spinal cords in tgSOD1 animals are subjected to increased oxidative and nitrative stress, as indicated by accumulation of protein carbonyls and nitrated proteins (Martin et al., 2007, 2009). These data also demonstrated that mitochondria are involved early in the pathogenesis of ALS (Martin et al., 2007, 2009). Recently, Panov et al. (2011b) provided direct evidence that isolated mitochondria from the brain and spinal cord of tgSOD1 rats generate significantly more reactive oxygen species (ROS) as compared with mitochondria from wild type animals.

Morphological, histochemical and immunological methods provided important information about involvement of mitochondria in the pathology, however they do not tell us what caused the changes in mitochondrial morphology. Relatively few studies were performed on isolated BM and SCM mitochondria from control (Sullivan et al., 2004, Panov et al., 2011a), tgSOD1 animals (Kirkinezos et al., 2005, Damiano et al., 2006, Fischer et al., 2011, Panov et al., 2011b), Wobbler mice (Dave et al., 2003, Xu et al., 2001), or *mnd* mice, which spontaneously develop motor neuron pathology (Bertamini et al., 2002). In their pioneering work, Sullivan et al (2004) have shown that normal SCM have lower respiratory activity, lower threshold for calcium-induced mitochondrial permeability transition, and a higher degree of superoxide radical production (determined with dehydroethidium) and lipid peroxidation as compared with BM. However, the dehydroethidium method does not identify the origin of superoxide radical. Studies of isolated mitochondria from tgSOD1 mice showed that BM and SCM had decreased rates of respiration associated with the loss of complex IV activity (Kirkinezos et al., 2005). Very early in the course of the disease, the mitochondrial Ca2+ loading capacity was significantly diminished as compared with the age-matched control mice (Damiano et al., 2006). Respiratory activities were also diminished in SCM of Wobbler mice (Dave et al., 2003, Xu et al., 2001) and mice with motor neuron degeneration disease (Bertamini et al., 2002).

A comparative study of isolated normal and tgSOD1 brain and spinal cord mitochondria was published recently by Panov et al. (2011a, b), and the unique metabolic features of BM and SCM were described (see also Panov et al, 2009). Normal SCM produced significantly more ROS than BM, and this was associated with the succinate-dependent reverse electron transport (Panov et al. 2011a). A rather unexpected finding was that BM and SCM isolated

Role of Neuronal Mitochondrial Metabolic Phenotype in Pathogenesis of ALS 231

mm3 in all cortices (the density of synapses is not very much different among species) it is clear that in the brain most energy expenditures are associated with synaptic activity, where most mitochondria are located (Abeles, 1991). The close spacing of terminals along axons (5 μm, implying a diffusion time of only 25 milliseconds) make terminal and axonal

Neurons communicate by means of signals that lead to depolarization of their excitable membranes. The degree of depolarization of the dendritic or soma membranes is determined by complex interactions between excitatory and inhibitory transmitters. The depolarization spreads to the axon, leading to release of an excitatory or inhibitory transmitter at the axonal endings of the cell, thus allowing information to be transferred to the next cell. When impulses are propagated, the cell loses K+ and gains Na+, and in order to preserve excitability the cell must extrude Na+ and accumulate K+, using metabolic energy derived from oxidative

consume in grey matter approximately 75% of the brain's energy (Abeles, 1991; Attwell & Laughlin, 2001). Because depolarization and restoration of ionic composition across the neuronal membrane occur in the millisecond scale, neuronal mitochondria must provide large quantities of ATP in a very short time. This distribution of energy use reflects the fact that a chemical synapse is an amplifier: for each glutamate released approximately 100 ions enter the postsynaptic terminal (Abeles, 1991). The rest (25%) of the brain's energy is used for the basic "housekeeping" functions, such as maintenance of pH gradients, asymmetrical distributions of phospholipids in membrane bilayers, protein synthesis (about 2%), anaplerotic functions and recycling of glutamate from synaptic clefts (Attwell & Laughlin, 2001). It was suggested that white and grey matter require similar amounts of energy on basic (nonsignaling) cellular

It is widely accepted that nerve tissues of adult mammals use glucose as the major source of energy. To fully understand the specifics of energy metabolism in brain, it is important to take into account that, in dendrites and synaptic terminals, there is limited space for glycolytic enzymes that would provide mitochondria with enough pyruvate derived from glucose. Therefore, neurons cannot be functionally and metabolically regarded without astroglia. Astrocytes provide neurons with lactate, as the major source of energy, replenish the neuronal pool of glutamine and recycle glutamate (Ebert et al., 2003; Pellerin et al., 2007; Suzuki et al., 2011). Of the total energy consumed by the brain, approximately 80% of energy expenditure is satisfied by oxidation of glucose and 20% by oxidation of fatty acids (Ebert et al., 2003). As we will see later, neurons also oxidize significant amounts of glutamate and GABA as a source of energy. The neurotransmitter GABA is formed from glutamate by the action of glutamate decarboxylase. It appears that glutamine serves as the precursor for glutamate and GABA. Therefore, the pool of glutamine must be constantly replenished in anaplerotic reactions, which occur

Fig. 1. Formation of GABA from glutamate. GAD – glutamate decarboxylase.

) and presynaptic Ca2+ influx

mitochondria functionally indistinguishable (Attwell & Laughlin, 2001).

metabolism (Abeles, 1991). Postsynaptic ion fluxes (Na+, K+, Cl-

activities (Attwell & Laughlin, 2001).

predominantly in astroglia.

from presymptomatic tgSOD1 rats showed almost normal rates of oxidative phosphorylation. Panov et all. (2011b) purified mitochondria in a Percoll gradient, and thus eliminated damaged mitochondria. This resulted in significantly lower yields of mitochondria per 1 gram of tissue of tgSOD1 rats: the yields were diminished by 27% for the brain and by 58% for spinal cord as compared with the normal Sprague Dawley rats (Panov et al., 2011b). Avossa et al. (2006) used immunocytochemistry and electron microscopy techniques to compare wild type and G93A spinal cord tissues after 14 days of growth under standard *in vitro* conditions. Wild type and mutant cultures displayed no differences in the analyzed parameters as well as in the number of motor neurons, and there were no signs of mitochondria vacuolization or protein aggregate formation in G93A ventral horns. Together, the results of Panov et al. (2011b) and Avossa et al. (2006) suggest that in tgSOD1 rats the BM and SCM are primarily normal but suffer from some pathological events, which rapidly deteriorate mitochondria and cause death of neurons.

In order to better understand the metabolic events involved in putative pathological mechanisms of neurodegeneration, we have to discuss the distinctive features of brain and spinal cord energy metabolism and their mitochondria.
