An Intercellular Flow of Glutathione Regulated by Interleukin 6 Links Astrocytes and the Liver in the Pathophysiology of Amyotrophic Lateral Sclerosis

Oxidative stress has been proposed as a major mechanism of damage to motor neurons associated with the progression of amyotrophic lateral sclerosis (ALS). Astrocytes are the most numerous glial cells in the central nervous system and, under physiological conditions, protect neurons from oxidative damage. However, it is uncertain how their reactive phenotype may affect motor neurons during ALS progression. In two different ALS mouse models (SOD1G93A and FUS-R521C), we found that increased levels of proinflammatory interleukin 6 facilitate glutathione (GSH) release from the liver to blood circulation, which can reach the astrocytes and be channeled towards motor neurons as a mechanism of antioxidant protection. Nevertheless, although ALS progression is associated with an increase in GSH efflux from astrocytes, generation of reactive oxygen species also increases, suggesting that as the disease progresses, astrocyte-derived oxidative stress could be key to motor-neuron damage.


Introduction
Amyotrophic lateral sclerosis (or ALS) is a disease of the central nervous system (CNS) characterized by a progressive degeneration of motor neurons (MNs) in the cerebral cortex (upper MNs), brainstem and spinal cord (lower MNs) [1]. The consequence is muscle weakness, which can progress to paralysis, spreading from one body region to another [2]. Most cases (90-95%) have no genetic background (sporadic ALS, SALS), and their cause is unknown, whereas in 5-10% of cases, a specific genetic mutation (the most common are found in C9orf72, SOD1, FUS, or TARDBP) is the main underlying mechanism (familial ALS, FALS) [3]. In the pathophysiology of the disease, neuroinflammation and oxidative stress are main mechanisms leading to neurodegeneration and MN death [4].
Progression of the disease is associated with reactive glia-and immunity-dependent neuroinflammation [5]. As a consequence, an increase in levels of inflammation-related cytokines can be detected in the cerebrospinal fluid (CSF) and blood of murine models and ALS patients [6]. Interleukin 6 (IL6) is among these cytokines [6,7]. IL6 seems to have important function in the CNS, i.e., neurogenesis and response of neurons and glial cells following different types of injuries [8]. However, although it is difficult to evaluate inflammation of the CNS or the relationship between neuroinflammation and disease progression in ALS patients, it has been reported that the common IL6 receptor 358 Ala variant (IL6R358Ala) and trans-signaling are disease modifiers in ALS [7].
Glutathione (L-γ-glutamyl-L-cysteinyl-glycine; GSH), the most prevalent non-protein thiol in mammalian cells and a physiological antioxidant, is involved in many cellular functions [9]. Cys availability and γ-glutamyl-cysteine synthase activity are rate-limiting

Cytokine Levels
Simultaneous quantification of different cytokines in the blood serum or CSF was obtained using xMAP technology and a MAGPIX Luminex 200 platform (Thermo Fisher Scientific, Waltham, MA, USA), as recently described in detail [6]. The results were analyzed with xPONENT 4.2 ® software (Luminex, Austin, TX, USA) and expressed as pg/mL.

Isolation, Culture and Perifusion of Astrocytes
Isolation and culture of astrocytes from mouse adult spinal cords was based on the methodology described by Beaudet et al. [24]. A perifusion system was designed for cultured astrocytes, where a modified cultured flask was designed to have an inflow tube placed across the flask screw cap and an outflow tube placed on the flask wall opposite the screw cap. The system included a reservoir for the culture medium (including a gas diffuser to ensure its saturation using a mixture of O 2 /CO 2 (19:1)), a roller pump, a bubble trap and a filter (Amicon YM30, Bedford, MA, USA) placed at the screw cap. The culture flask was partially submerged in a thermostatized water bath at a constant temperature of 37 • C. Effluent flow was monitored continuously for O 2 and pH with Philips electrodes. The perifusion flow rate was of 2 mL/min.

Analysis of Amino Acids
Arterial blood was collected from the left ventricle into a heparinized syringe/tube. After centrifugation (10 min, 800× g), plasma was collected, and protein was precipitated by mixing 1 mL of plasma and 4 mL of 3.75% (w/v) sulphosalicylic acid in 0.3 M-lithium citrate buffer (pH 2.8). The supernatant (0.5 mL) was injected into an LA8080 high-speed amino-acid analyzer (Hitachi, Tokyo, Japan) for determination of amino acids.

RT-PCR
RNA was extracted using Qiagen RNAeasy mini kits (Hilden, Germany). Quantitative and qualitative analyses of RNA samples were performed using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). cDNA was obtained using a random hexamer primer and a MultiScribe reverse transcriptase kit as recommended by the manufacturer (Taq-Man RT Reagents; Thermo Fisher Scientific). PCR master mix and AmpliTaq Gold DNA polymerase were added to the specific primers (Sigma Genosys, Haverhill, UK), as previously reported [19], for GCL (γ-glutamylcysteine ligase, catalytic subunit), GGT (γ-glutamyl transpeptidase), SOD (superoxide dismutases) 1 and 2, CAT (catalase), GPX2 (glutathione peroxidase 2), GSR (glutathione reductase) and G6PDH (glucose-6-P dehydrogenase). For real-time quantification of mRNA relative to GAPDH, a SYBR Green I assay and an iCycler detection system (Biorad, Hercules, CA, USA) were used. Relative gene expression is expressed as fold change. The threshold cycle (CT) was determined, and the relative gene expression was expressed as fold change = 2 −∆(∆CT) , where ∆C T = C T target − C T GAPDH and ∆(∆C T ) = ∆C T treated − ∆C T control.

Enzyme Activities
GCL activity was determined by measuring the rate of ADP formation at 37 • C in a medium containing Tris/HCl buffer (pH 8.2), KCl, ATP, phosphoenolpyruvate, L-Glu, L-α-aminobutyrate, MgCl2, EDTA, NADH, pyruvate kinase and lactate dehydrogenase. GGT activity was measured using glycylglycine as a γ-glutamyl acceptor substrate and γ -glutamyl-4-nitroanilide and its carboxy derivative, γ-glutamyl-3-carboxy-4-nitroanilide, as donor substrates. SOD activity was measured using cyanide in the assay medium to distinguish the mangano-type enzyme (SOD2) from the cuprozinc type (SOD1). To measure CAT, activity samples were incubated in the presence of an H 2 O 2 solution for 2 min prior to rapid mixing of the incubation enzymatic reaction mixture with a cobaltbicarbonate reagent, which assesses non-reacting H 2 O 2 . Catalase activity is always directly proportional to the rate of dissociation of H 2 O 2 . GPX (selenium-dependent) activity was measured using H 2 O 2 as a substrate in Tris-HCl buffer with the addition of NaN 3 and EDTA (pH 8.5) at an incubation temperature of 37 • C. GSR activity follows the reduction of GSSG to GSH by monitoring the oxidation of NADPH monitored by a decrease in absorbance at 340 nm (see [25] and references therein).

Oxygen Consumption
O 2 concentration and rate of O 2 consumption in isolated astrocytes was continuously recorded using a high-resolution oxygraph (OROBOROS INSTRUMENTS Corp., Innsbruck, Austria). Isolated astrocytes were resuspended in respiration medium (DMEM 4500 mg/L glucose to which was added 5 mM pyruvate) that had been pre-warmed to 37 • C. A volume of 2.0 mL of cell suspension was added to the O 2 electrode chamber, where it was magnetically stirred and kept at 37 • C. The chamber was sealed, and the cells were incubated until a stable respiratory rate was reached.

Western Blotting
Western blots were run, as previously described [25]. Proteins were transferred to a nitrocellulose membrane and subjected to Western blotting with specific anti-human monoclonal antibodies (OriGene, Rockville, MD, USA; and Abcam, Cambridge, UK). Blots were developed using horseradish-peroxidase-conjugated secondary antibody and enhanced chemiluminescence (ECL system; GE HealthCare Life Sciences, Piscataway, NJ, USA). Protein bands were quantified using laser densitometry.

Expression of Results and Statistical Analyses
Data are presented as the means ± SD for the indicated number of different experiments. Statistical analyses were performed using Student's t test, and p values of <0.05 were considered significant. Survival data were analyzed with Kaplan-Meier curves and LogRank (Mantel-Cox) tests.

Neuromotor Evaluation of ALS Mouse Models
Neuromotor functions were studied in SOD1 G93A and FUS-R521C mice using a standardized neurological score (designed to assess hind-limb function) ( Figure 1A) and the rotarod performance test (a measure of balance, coordination, physical condition and motor planning) ( Figure 1B). As shown in Figure 1, the onset of the symptomatology is around postnatal week 12 in SOD1 G93A mice and postnatal week 7 in the FUS-R521C model. An advanced state of disease progression was well established around postnatal weeks 17-18 (SOD1 G93A ) and 17-19 (FUS-R521C). The survival rate for both models is displayed in Figure 1C. Based on these data, further experiments were performed comparing measurements at the onset of the symptomatology and at an advanced state of progression. enhanced chemiluminescence (ECL system; GE HealthCare Life Sciences, Piscataway, NJ, USA). Protein bands were quantified using laser densitometry.

Expression of Results and Statistical Analyses
Data are presented as the means ± SD for the indicated number of different experiments. Statistical analyses were performed using Student's t test, and p values of <0.05 were considered significant. Survival data were analyzed with Kaplan-Meier curves and LogRank (Mantel-Cox) tests.

Neuromotor Evaluation of ALS Mouse Models
Neuromotor functions were studied in SOD1 G93A and FUS-R521C mice using a standardized neurological score (designed to assess hind-limb function) ( Figure 1A) and the rotarod performance test (a measure of balance, coordination, physical condition and motor planning) ( Figure 1B). As shown in Figure 1, the onset of the symptomatology is around postnatal week 12 in SOD1 G93A mice and postnatal week 7 in the FUS-R521C model. An advanced state of disease progression was well established around postnatal weeks 17-18 (SOD1 G93A ) and 17-19 (FUS-R521C). The survival rate for both models is displayed in Figure 1C. Based on these data, further experiments were performed comparing measurements at the onset of the symptomatology and at an advanced state of progression. Taking into account the % of animals that died before reaching the advanced progression status, to assess the neurological score and the rotarod test, their number was increased so that n (20 animals) was the same both at onset and at advanced progression. In wild-type controls (n = 20) the neurological score was equal to 0 and the time for the rotarod test was 1200 s in all cases. Both tests were performed twice a week. On the days of the week on which the animals were not subjected to tests, they were trained in the rotarod for 20 min. Control experiments performed in wild-type mice (n = 10) rendered a neurological score of 0 and a performance time of 1200 s in the rotarod test in all cases. * p < 0.01 comparing values obtained in amyotrophic lateral sclerosis (ALS) mice versus wild-type controls. Taking into account the % of animals that died before reaching the advanced progression status, to assess the neurological score and the rotarod test, their number was increased so that n (20 animals) was the same both at onset and at advanced progression. In wild-type controls (n = 20) the neurological score was equal to 0 and the time for the rotarod test was 1200 s in all cases. Both tests were performed twice a week. On the days of the week on which the animals were not subjected to tests, they were trained in the rotarod for 20 min. Control experiments performed in wild-type mice (n = 10) rendered a neurological score of 0 and a performance time of 1200 s in the rotarod test in all cases. * p < 0.01 comparing values obtained in amyotrophic lateral sclerosis (ALS) mice versus wild-type controls.

GSH Synthesis and Release in Hepatocytes from ALS Mice
Under physiological conditions, the liver is the main reservoir and source of circulating GSH [27]. We studied the rates of GSH synthesis and efflux in isolated hepatocytes from wild-type (WT), SOD1 G93A and FUS-R521C mice. As shown in Table 1, in the presence of amino-acid precursors, GSH synthesis and efflux significantly increase in both models at an advanced state of disease progression, thus suggesting that an ALS-induced signaling mechanism likely promotes those changes. Table 1. Glutathione (GSH) synthesis and efflux in isolated hepatocytes from wild-type and ALS mice. The initial GSH concentration was 5.2 ± 0.3 µmol/g in isolated hepatocytes from wild-type mice, 5.0 ± 0.4 and 4.0 ± 0.3 µmol/g in isolated hepatocytes from SOD1 G93A mice (at onset and advanced state of progression, respectively) and 4.8 ± 0.4 and 3.7 ± 0.3 µmol/g in isolated hepatocytes from FUS-R521C mice (at onset and advanced state of progression, respectively). Hepatocytes were incubated in 10-mL Erlenmeyer flasks (final volume 2 mL) for 60 min (see Methods) in the presence or absence of amino-acid precursors for GSH synthesis (5 mM Gln, 2 mM Gly, 1 mM Ser, 1 mM N-acetylcysteine) [28]. Only L-amino acids were used. Glucose (5 mM) and bovine serum albumin (2%) were present in all incubations. Rates of GSH synthesis were calculated from total GSH content in incubations at 0, 20, 40 and 60 min. Rates of glutathione efflux were calculated from contents of GSH and glutathione disulfide (GSSG) in the extracellular medium at 0, 20, 40 and 60 min. All values are means ± SD for 7-8 observations. For both groups, the significance test refers to the comparison of rates in hepatocytes isolated from ALS mice versus those from the wild-type controls (* p < 0.05, ** p < 0.01).

GSH Efflux (nmol/g·min)
We also measured liver and blood GSH levels. As shown in Table 2 and as compared to values obtained in the wild-type mice, liver GSH levels decrease in both ALS mouse models at an advance state of progression, a fact that correlates with a decrease in the ratio of synthesis/efflux (Table 1). However, the increase in liver GSH efflux does not cause the consequent increase in circulating GSH; in fact, GSH decreases in the circulating blood in both models (Table 2), a fact suggesting that blood GSH may be bypassed for metabolism. This experimental evidence raises the question of how and where GSH flow is directed during ALS progression. Table 2. Blood and liver GSH in wild-type and ALS mice. Blood and liver samples were obtained and treated as described in Methods. Data are means ± SD for 10-12 different mice. * p < 0.01 comparing values obtained in ALS mice versus wild-type controls.

IL6 Induces GSH Efflux in Hepatocytes
It is well known that ALS is associated with neuroinflammation and an increase in proinflammatory cytokines in plasma and CSF (see e.g., Obrador [6]). We measured some of these cytokines in both biological fluids in our experimental models at an advanced state of disease progression. As shown in Figure 2, as compared to healthy wild-type mice, proinflammatory cytokine levels increase. As introduced above, IL6 signaling, in particular, could play a role in increasing the release of liver GSH. We tested this hypothesis in the ALS models (at an advanced state of progression) by administering anti-IL6 mAbs or anti-IL6R mAbs in vivo. As compared to controls treated with physiological saline, both types of antibodies decreased the rate of GSH efflux from hepatocytes (Table 3), thus proving, in two different models, that IL6 is a main cytokine responsible for the increase in hepatic GSH efflux during ALS progression. In parallel experiments, we found that treatment with anti-IL6 mAbs is also associated with an increase in GSH levels in the liver (Table 3). Blood GSH levels also increased (Table 3). Therefore, either the metabolic use of circulating GSH decreases or GSH is released to the blood by other organs/tissues as a consequence of anti-IL6 treatment. In order to answer this paradox, we looked at GSH levels, GCL activity and Cys/cystine uptake in the kidneys and lungs (two organs with a significant content of GSH). As shown in Table 4, anti-IL6R treatment induces a significant decrease in GSH in both organs (suggesting that more GSH is released to the blood) and an increase in GCL activity and Cys/cystine uptake (which implies a higher level of GSH synthesis). These results suggest a systemic adaptation to the treatment (involving kidneys and lungs, although it could involve more organs). This is an adaptation that would explain the increase observed in blood GSH (Table 3).
proving, in two different models, that IL6 is a main cytokine responsible for the increase in hepatic GSH efflux during ALS progression. In parallel experiments, we found that treatment with anti-IL6 mAbs is also associated with an increase in GSH levels in the liver (Table 3). Blood GSH levels also increased (Table 3). Therefore, either the metabolic use of circulating GSH decreases or GSH is released to the blood by other organs/tissues as a consequence of anti-IL6 treatment. In order to answer this paradox, we looked at GSH levels, GCL activity and Cys/cystine uptake in the kidneys and lungs (two organs with a significant content of GSH). As shown in Table 4, anti-IL6R treatment induces a significant decrease in GSH in both organs (suggesting that more GSH is released to the blood) and an increase in GCL activity and Cys/cystine uptake (which implies a higher level of GSH synthesis). These results suggest a systemic adaptation to the treatment (involving kidneys and lungs, although it could involve more organs). This is an adaptation that would explain the increase observed in blood GSH (Table 3). Figure 2. Levels of pro-inflammatory cytokines in cerebrospinal fluid (CSF) and plasma of wildtype and ALS mice. Cytokines were measured as described in Methods. To ensure that the results were comparatively homogeneous, all the mice were subjected to the tests and training indicated in Figure 1. Data in ALS mouse models correspond to an advanced state of progression. Data are mean values ± SD for n = 7. * p < 0.01 comparing values obtained in ALS models versus wild-type controls. Table 3. Administration of anti-IL6 or anti-IL6R mAbs decreases GSH efflux from hepatocytes. Antibodies were administered, as described in Methods, during the three days prior to isolation of hepatocytes or collection of blood or liver samples. See also the captions of Table 1; Table 2. Rates of GSH efflux were determined in hepatocytes incubated in the presence of amino-acid precursors (as in Table 1). Data in ALS mouse models correspond to an advanced state of progression. Data are mean values ± SD for n = 9-10. * p < 0.01 comparing values obtained in ALS models versus wild-type controls. + p < 0.01 values obtained in mice treated with anti-IL6 or anti-IL6R mAbs versus controls treated with physiological saline. Anti-IL6 8 ± 3 12 ± 3 + 11 ± 2 + 7.9 ± 0.5 7.0 ± 0.8 + 7.2 ± 0.8 + 7.5 ± 0.9 6.5 ± 0.5 + 6.9 ± 1.0 + Anti-IL6R 8 ± 2 10 ± 2 + 10 ± 3 + 8.0 ± 0.9 7.4 ± 0.7 + 7.5 ± 0.6 + 7.6 ± 1.0 6.9 ± 0.8 + 7.1 ± 0.5 +   TNF  IFN  IL1  IL2  IL6  GM- Table 3. Administration of anti-IL6 or anti-IL6R mAbs decreases GSH efflux from hepatocytes. Antibodies were administered, as described in Methods, during the three days prior to isolation of hepatocytes or collection of blood or liver samples. See also the captions of Table 1; Table 2. Rates of GSH efflux were determined in hepatocytes incubated in the presence of amino-acid precursors (as in Table 1). Data in ALS mouse models correspond to an advanced state of progression. Data are mean values ± SD for n = 9-10. * p < 0.01 comparing values obtained in ALS models versus wild-type controls. + p < 0.01 values obtained in mice treated with anti-IL6 or anti-IL6R mAbs versus controls treated with physiological saline.

GSH Synthesis in Astrocytes Is Highly Dependent on Extracellular GSH Supply
Based on the fact that astrocytes provide GSH to neurons [13], we investigated whether GSH levels in astrocytes are affected during ALS progression. For this purpose, we use a perifusion system (see Methods) to perfuse the cultured astrocytes with a constant flow of medium containing the same GSH levels measured in the blood of ALS mice ( Table 5). Supply of amino-acid precursors is not limited during ALS progression (Table   17 18 0

GSH Synthesis in Astrocytes Is Highly Dependent on Extracellular GSH Supply
Based on the fact that astrocytes provide GSH to neurons [13], we investigated whether GSH levels in astrocytes are affected during ALS progression. For this purpose, we use a perifusion system (see Methods) to perfuse the cultured astrocytes with a constant flow of medium containing the same GSH levels measured in the blood of ALS mice ( Table 5). Supply of amino-acid precursors is not limited during ALS progression (Table 6A). As shown in Table 5, GSH levels in astrocytes decrease at an advanced state of progression (approx. to 50%, as compared to the levels found in astrocytes isolated from WT mice or from ALS mice at the onset of symptomatology). Nevertheless, the decrease in GSH, paradoxically, is associated with an increase in GCL activity, the rate-limiting step in GSH synthesis (Table 5). GSSG levels also increase significantly, thus suggesting an increase in oxidative stress within astrocytes (Table 5). Since GSH synthesis is highly dependent on the availability of free Cys (the levels of which are extremely low inside and outside cells) [29], we also measured GGT activity and rates of Cys and cystine uptake. As shown in Table 6B, advanced ALS progression is associated with an increase in GGT activity, no changes in the rate of extracellular Cys uptake and a decrease in cystine uptake. Cells do not transport intact GSH from the extracellular space [30]. GGT is the only enzyme that cleaves the γ-glutamyl-cysteine peptide bond in GSH [31], thus releasing γ-Glu and cysteinylglycine, which is further cleaved into Cys and Gly by plasma-membrane-bound dipeptidases [32]. Free γ-glutamyl amino acids Cys and Gly can serve as GSH precursors [33]. Hence, our data suggest that extracellular GSH is the main source of Cys for astrocytes during advanced ALS progression. To further test this postulate, we investigated the effect on astrocyte GSH levels of specific inhibitors of GGT activity (acivicin) [34], Cys uptake by the ASC transport system for neutral amino acids [(1)-amino(1-(3,5-dichlorophenyl)-3,5-dimethyl-1H-pyrazol-4-yl)acetic acid, ACPP] [35] or cystine uptake by the Xccystine/glutamate antiporter (sulfasalazine, SSZ) [36]. These are the main carriers of Cys and cystine in mammalian cells (see Obrador [37]). As shown in Table 7, GSH depletion in astrocytes is more profound in the presence of specific inhibitors of GGT or ASC-mediated Cys transport. Table 5. GSH/GSSG levels and GCL activity in astrocytes isolated from wild-type and ALS mice. Data are mean values ± SD for n = 9. * p < 0.01 comparing values obtained in ALS models versus the wildtype controls. + p < 0.01 comparing advanced state of progression versus onset of symptomatology.    Cystine uptake (nmol/mg prot·min) 24.5 ± 3.5 10.6 ± 1.7 * 11.7 ± 2.2 *

In Astrocytes, Advanced ALS Progression Is Associated with an Increase in GSH Efflux and ROS Generation
As described in Table 5, at an advanced state of progression, GSH levels in astrocytes decrease in the two models of ALS studied, whereas GSSG levels increase, facts suggesting an increase in oxidative stress. We further studied the underlying mechanisms, and as shown in Table 8, ROS generation increases in astrocytes isolated from ALS mice at an advanced state of progression compared to astrocytes from WT controls. The increase in ROS generation is associated with an increase in O 2 consumption, which may reflect the high metabolic activity associated to the reactive astrocyte phenotype linked to the progression of the disease. However, in the group of enzymes studied, only SOD1 activity was found to be significantly higher in the astrocytes of the ALS mice compared to WT controls (Table 8, Figure 4). GSH efflux was also found to increase in both models, as compared to WT controls (Table 8). This increase in GSH efflux may also be facilitated by IL6 in astrocytes (as in hepatocytes), given that we observed that anti-IL6R mAbs prevents GSH depletion in perifused astrocytes (results not shown). This all indicates that the increase in ROS generation is not counteracted within the astrocytes. It is obvious that an increase in ROS generation by reactive astrocytes may cause damage to neighboring MNs. In a previous report, we observed that cytosolic GSH levels in MNs isolated from SOD1 G93A mice do not change as the disease progresses [6], which means that an increase in GSH efflux from astrocytes does not necessarily imply an increase in MNs. Table 7. Effect of inhibition of GGT activity or Cys/cystine uptake on GSH levels in astrocytes. Astrocytes isolated from wild-type and ALS mice at an advanced state of progression were incubated in the absence or presence of acivicin (0.05 mM), ACPP [(1)-amino(1-(3,5-dichlorophenyl)-3,5-dimethyl-1H-pyrazol-4-yl)acetic acid, 0.1 mM] or SSZ (sulfasalazine, 0.2 mM) during a 24 h-time period before measuring GSH levels. The culture medium contained extracellular GSH and cyst(e)ine at a concentration of 10 a 1 mM, respectively. Data are mean values ± SD for n = 5-6. * p < 0.01 comparing values obtained in ALS models versus wild-type controls. + p < 0.01 comparing the different inhibitors versus controls.

Discussion
ALS progression is associated with severe diffuse astrogliosis, which can be a consequence of neurotoxic insults, diffuse trauma, diffuse ischemia, or different types of infection [38]. Reactive astrocytes show changes in gene expression, with pronounced upregulation of glial fibrillary acidic protein, hypertrophy and dispersed proliferation and can cause inhibition of axonal regeneration [39] and favor the arrival of inflammatory cells [40]. Therefore, mechanisms that can trigger chronic neurodegeneration. Indeed, reactive astrocytes surround degenerating MNs in ALS patients and ALS rodent models [41]. Nevertheless, the mechanisms underlying the potential damaging role of astrocytes still are a question of debate.
IL6 is upregulated in the CNS of patients affected by neuroinflammation of different etiologies [8], as well as in murine models of brain injury [42]. Its pro-inflammatory transsignaling pathway, shown to mediate neurodegeneration in mice, depends on a soluble form of IL-6R capable of binding IL-6 and stimulating a response on distal cells that express β-receptor glycoprotein 130 [43]. IL6 knockout mice show a lower inflammatory response and neuroglial activation [44,45], increased oxidative stress [46], decreased lymphocyte recruitment [47] and a low rate of recovery and healing [48], all suggesting that IL6 may be a double-edged sword, particularly in chronic pathological processes, as is the case with ALS. As previously suggested concerning murine models of metastatic melanoma [14], we found that IL6 also promotes the release of GSH from the liver in the two ALS models assayed (Table 3). GSH synthesis and efflux are significantly increased, as compared to WT controls, in the hepatocytes of ALS mice at an advanced state of disease progression (Tables 1 and 3). The consequent increase in blood GSH facilitates an interorgan/cellular flow that can be potentially used by astrocytes and transferred to MNs. Overall, this mechanism represents a protective effect for MNs, promoted by the increased levels of IL6 in blood and CSF (Figure 2).
Binding of IL6 to its receptor initiates a signal-transduction cascade through different transcription factors, Janus kinases (JAKs) and signal transducers, as well as activators of

Discussion
ALS progression is associated with severe diffuse astrogliosis, which can be a consequence of neurotoxic insults, diffuse trauma, diffuse ischemia, or different types of infection [38]. Reactive astrocytes show changes in gene expression, with pronounced upregulation of glial fibrillary acidic protein, hypertrophy and dispersed proliferation and can cause inhibition of axonal regeneration [39] and favor the arrival of inflammatory cells [40]. Therefore, mechanisms that can trigger chronic neurodegeneration. Indeed, reactive astrocytes surround degenerating MNs in ALS patients and ALS rodent models [41]. Nevertheless, the mechanisms underlying the potential damaging role of astrocytes still are a question of debate.
IL6 is upregulated in the CNS of patients affected by neuroinflammation of different etiologies [8], as well as in murine models of brain injury [42]. Its pro-inflammatory transsignaling pathway, shown to mediate neurodegeneration in mice, depends on a soluble form of IL-6R capable of binding IL-6 and stimulating a response on distal cells that express β-receptor glycoprotein 130 [43]. IL6 knockout mice show a lower inflammatory response and neuroglial activation [44,45], increased oxidative stress [46], decreased lymphocyte recruitment [47] and a low rate of recovery and healing [48], all suggesting that IL6 may be a double-edged sword, particularly in chronic pathological processes, as is the case with ALS. As previously suggested concerning murine models of metastatic melanoma [14], we found that IL6 also promotes the release of GSH from the liver in the two ALS models assayed (Table 3). GSH synthesis and efflux are significantly increased, as compared to WT controls, in the hepatocytes of ALS mice at an advanced state of disease progression (Tables 1 and 3). The consequent increase in blood GSH facilitates an interorgan/cellular flow that can be potentially used by astrocytes and transferred to MNs. Overall, this mechanism represents a protective effect for MNs, promoted by the increased levels of IL6 in blood and CSF (Figure 2).
Binding of IL6 to its receptor initiates a signal-transduction cascade through different transcription factors, Janus kinases (JAKs) and signal transducers, as well as activators of transcription (STATs) [49]. Whether treatment with anti-IL6 or anti-IL6R alter the expression of the mutant SOD1 and/or the mutant FUS is unknown and should be investigated. What is remarkable is the fact that despite the different phenotypes of the two ALS models, the effect of anti-IL6 treatment on GSH levels is quite similar (Table 3).
It has been demonstrated that astrocytes provide Cys to neurons by releasing GSH [13] via the multi-drug-resistant protein 1 [50]. Astrocytes can degrade extracellular GSH via GGT and release Cys-Gly, which is broken down by the action of a plasma-membranebound dipeptidase [37]. Then, free Cys, which is rate-limiting for GSH synthesis, can be taken up through the ASC system [37] (Table 7). The blood supply of all other amino-acid precursors for GSH synthesis is not limited in ALS mice (Table 6A). Importantly, although the rate of cystine uptake by astrocytes decreases as the disease progresses, the rate of Cys uptake is not affected ( Table 6B). The effect of the inhibitors ( Table 7) seems to be relative to the initial control GSH levels in each case. What is important is the fact that inhibition of GGT activity or Cys uptake causes a higher GSH depletion than inhibition of cystine uptake. Cys levels are extremely low (outside and inside cells), and consequently, Cys uptake is highly dependent on the generation of Cys from extracellular GSH (via the reaction catalyzed by GGT). GGT activity in astrocytes was found to increase, as compared to WT controls, in ALS mice (Table 6B). However, GGT activity in MNs isolated from SOD1 G93A mice at an advanced state of progression is lower than its equivalent in astrocytes [6]. Since astrocytes outnumber neurons by over fivefold [51], it is obvious that astrocytes have the highest capacity to metabolize extracellular GSH. Taken together, these experimental facts indicate that GSH synthesis in MNs is highly dependent on GSH supply from astrocytes, which, in turn, depends on the interorgan flow from the liver controlled by IL6.
As shown in Table 8, in both ALS models, generation of ROS by astrocytes increases at an advanced state of disease progression. Previously, we found that GSH levels in MNs isolated from SOD1 G93A mice at an advanced state of progression are slightly higher than the levels measured in MNs from WT control mice [6]. However, GSSG levels (which indirectly reflect a higher exposure to ROS) increase by approx. 100% [6]. An increase in O 2 consumption and GSH efflux (Table 8), as well as the consequent decrease in GSH levels (Table 5), may explain the increase in ROS generation by astrocytes. ROS could be a direct cause of oxidative damage in neighboring MNs. The fact that SOD1 and SOD2, two key antioxidant enzyme activities, decrease in MNs of SOD1 G93A mice as the disease progresses [6] sets up a cumulative-damage scenario. As shown in Table 5, GSH levels decrease in astrocytes isolated from ALS mice at an advanced state of progression, compared to onset or WT. However, GCL activity increases (Table 5), and Cys uptake remains unchanged (Table 6B). Nevertheless, the increases in GSSG (Table 5) and ROS generation (both indicating more oxidative stress), as well as the increase in GSH efflux (Table 8), explain why GSH decreases. An increase in ROS exposure and a decrease in its antioxidant-defense capacity may be key factors in the death of MNs.

Conclusions
An increase in circulating levels of IL6 (released by glial and inflammatory cells) promotes the release of GSH from the liver (a physiological reservoir of this tripeptide). The increased availability of extracellular GSH facilitates its transfer, via astrocytes, to MNs and favor its antioxidant defense. However, in advanced stages of the disease, this process is associated with an increase in ROS generation by astrocytes. Oxidative stress can cause cumulative damage to MNs, as well as their death. From the results described herein, it seems obvious that maintaining GSH levels in astrocytes, and consequently in MNs, can help to slow down the progression of the disease. Since the availability of Cys is a limiting factor for the synthesis of GSH, a direct donor of Cys, e.g., N-acetylcysteine, which easily crosses the plasma membrane, could be a direct drug solution. Nevertheless, maintenance of GSH homeostasis in ALS is only part of the potential benefits of combined strategies targeting different pathways/molecules at the same time [52].