**2.1 Increasing inflammation in ALS**

To assess the effect of inflammation in ALS and thus to discover whether boosting the inherent inflammation would be beneficial, lipopolysaccharide (LPS) was daily administered to ALS mice (Nguyen et al., 2004). The effect of this treatment was a clear decrease in lifespan, implying that an increase of inflammation is detrimental in ALS (Nguyen et al., 2004). Another study, initially intended to decrease inflammation, administered macrophage colony stimulating factor (M-CSF) to ALS mice and observed an unexpected increase of microgliosis also leading to a decreased survival (Gowing et al., 2009). Although not directed specifically at astrocytes, this work has led to the understanding of the hazardous character of neuroinflammation in ALS.

### **2.2 Astrogliosis in ALS**

Reactive astrocytes alter gene expression including an upregulation of the intermediate filaments GFAP and vimentin that allow for visualisation of astrogliosis by increased immunoreactivity of these filaments in patient and ALS model tissue. Post mortem spinal

The Astrocytic Contribution in ALS: Inflammation and Excitotoxicity 381

also play a role in fALS patients with only 1 allele of mutant SOD1 or in fALS/sALS patients without disease causing SOD1 mutations. Remarkable work by Haidet-Phillips et al., (2011) shows that astrocytes collected post mortem from fALS and sALS are both able to induced motor neuron selective death which is not observed with astrocytes obtained from controls (Haidet-Phillips et al., 2011). Gene expression analysis of the the fALS and sALS astrocytes demonstrated increased expression chemokines, proinflammatory cytokines and components of the complement pathway (Haidet-Phillips et al., 2011). This study confirms that the astrocytic inflammatory effects found in vitro and in vivo in ALS mice may also

Attempts to minimise the inflammatory effect of astrocytes in ALS include genetic strategies targeting astrocytic knockdown of certain cytokines. Targeting of nuclear factor κB (NF-κB) activation specifically in astrocytes does not alter disease onset nor life span in ALS mice (Crosio et al., 2011). Which may be in part due to the mere decreased astrogliosis at presymtomatic stage (Crosio et al., 2011). Additionally, complete ablation of TNFα does not affect disease parameters (Gowing et al., 2006), also implying deletion of a single cytokine may not be sufficient to affect disease progression, as a general decrease in inflammation can be beneficial (see below). Instead of targeting cytokine production, ablating proliferating astrocytes was attempted in ALS, showing no effect on survival (Lepore et al., 2008a). This may be explained by the inability of astrocytes to proliferate in

A number of strategies have been utilized to diminish inflammation in ALS mice, though often not specifically targeting astrocytes. It is worthwhile to note that it is commonly unclear whether anti-inflammatory strategies truly exert an anti-inflammatory function due to the age-matched analysis of astrogliosis instead of disease stage-matched analysis with drugs that successfully extend lifespan. Among these therapeutic strategies are those intended to pharmacologically block the cyclooxygenase (COX) pathway by administration of celecoxib, a selective COX-2 inhibitor, and deletion of the prostaglandin E2 receptor. Both strategies diminish inflammation, postpone disease onset and prolong survival in ALS mice (Drachman et al., 2002; Liang et al., 2008). Similarly, celastrol administration also postponed disease onset extended lifespan, while decreasing TNFα, nitric oxide synthases (iNOS), cluster of differentiation 40 (CD40) immunoreactivity and astrogliosis when assessing age-matched spinal cord tissue (Kiaei et al., 2005). Similar effects are obtained when providing ALS mice with folic acid (Zhang et al., 2008) or bee venom (Yang et al., 2010). Thalidomide extends survival in ALS mice by destabilising cytokine mRNA including TNFα (Kiaei et al., 2006). Additionally, its analog lenalidomide also prolongs survival (Kiaei et al., 2006), even when administered after symptom onset (Neymotin et al., 2009). To conclude, minocycline has shown a dramatic increase in survival of ALS mice and postpones symptom onset (Kriz et al., 2002; Van Den Bosch et al., 2002; Zhu et al., 2002). Interestingly, this effect is moment of administration dependent, as administration of minocycline post disease onset increases the astrocytic

and microglial response in ALS mice, decreasing survival (Keller et al., 2011).

contribute to disease pathology in humans.

**2.4 Genetic tools to minimise astrogliosis** 

**2.5 Therapeutic strategies targeting inflammation in ALS** 

ALS (Gowing et al., 2008).

cord tissue from fALS and sALS patients display astrogliosis (Schiffer et al., 1996), implying that reactivity of astrocytes is not limited to the familial form of ALS. Interestingly, astrogliosis levels are similar between long surviving and short surviving ALS patients, although this is not the case for microglial activation and the amount of dendritic cells (Sta et al., 2011). An extra facet of astrogliosis in ALS is an increased immunoreactivity of tolllike receptor 4 in astrocytes of sALS patients (Casula et al., 2011). Astrogliosis in ALS mice is present at symptomatic stages preceeding microgliosis (Kiaei et al., 2006; Keller et al., 2009; Yang et al., 2011). Interestingly, GFAP is not necessary for astrogliosis as GFAP deficient astrocytes can still become reactive and do not affect survival of ALS mice (Yoshii et al., 2011).

### **2.3 Mutant SOD1 affects astrocytic inflammatory behaviour**

The expression of mutant SOD1 in astrocytes alters their function in vivo and in vitro. To begin, deletion of mutant SOD1 in astrocytes in two distinct ALS models demonstrates the detrimental effect of mutant SOD1 in astrocytes mainly post onset, as deletion increased lifespan of ALS mice (Yamanaka et al., 2008; Wang et al., 2011a). Intriguingly, astrogliosis was unaltered, implying that the negative effect of mutant SOD1 in astrocytes is not due to altered levels of astrogliosis (Yamanaka et al., 2008), but potentially by astrocytes inducing microgliosis (Yamanaka et al., 2008; Wang et al., 2011a). An alternative approach arrives from the field of transplantation in which non-transgenic mesenchymal stem cells are transplanted into the spinal cord of ALS rats and differentiate into astrocytes, thus diluting the mutant SOD1 positive astrocytes in the spinal cord (Boucherie et al., 2009). This approach also shows unaltered astrogliosis, but also decreased microgliosis and cyclooxygenase 2 (COX2) expression, and extends murine ALS life span (Boucherie et al., 2009). The processes explaining this hazardous effect of mutant SOD1 in astrocytes has been investigated in vitro. To begin, an interesting approach of transducing human astrocytes with wild-type SOD1 or mutant SOD1 increases inflammation in mutant SOD1 cultures (Marchetto et al., 2008). In addition, the mutant SOD1 transduced astrocytes provide a less viable environment for human embryonic stem cell derived motor neurons (Marchetto et al., 2008). The latter was rescued by using a NADPH oxidase 2 (NOX2) inhibitor, apocynin (Marchetto et al., 2008). Other studies concur that mutant SOD1 primary astrocytes exhibit a higher gene expression of cytokines on baseline and when stimulated by interferon γ (IFNγ) or tumor necrosis factor α (TNFα) (Hensley et al., 2006), implying once again that mutant SOD1 expression may affect the threshold of astrocytes to produce proinflammatory cytokines. Accordingly, the expression of interferon simulated genes is detected in astrocytes of presymptomatic ALS mice (Wang et al., 2011b) and genetic ablation and knockdown of the interferon alpha receptor type 1 (IFNAR1) increase ALS mouse survival by 5% and 10%, respectively (Wang et al., 2011b). Intriguingly, Aebischer et al. stress the importance of interferon signalling in mutant SOD1 astrocytes by demonstrating that mutant SOD1 astrocytes trigger the selective death of motor neurons mediated by IFNγ (Aebischer et al., 2011). This mechanism is dependent on the activation of the lymphotoxin-β receptor by LIGHT (TNFSF14) and genetic ablation of LIGHT extends survival of ALS mice by 13%, but does not postpone disease onset (Aebischer et al., 2011). Although this is a large increase in disease survival, clearly other mechanisms remain to play a role.

The above described altered functioning of mutant SOD1 expressing astrocytes is induced by an overexpression of multiple copies of mutant SOD1. It is unclear whether these effects also play a role in fALS patients with only 1 allele of mutant SOD1 or in fALS/sALS patients without disease causing SOD1 mutations. Remarkable work by Haidet-Phillips et al., (2011) shows that astrocytes collected post mortem from fALS and sALS are both able to induced motor neuron selective death which is not observed with astrocytes obtained from controls (Haidet-Phillips et al., 2011). Gene expression analysis of the the fALS and sALS astrocytes demonstrated increased expression chemokines, proinflammatory cytokines and components of the complement pathway (Haidet-Phillips et al., 2011). This study confirms that the astrocytic inflammatory effects found in vitro and in vivo in ALS mice may also contribute to disease pathology in humans.

### **2.4 Genetic tools to minimise astrogliosis**

380 Amyotrophic Lateral Sclerosis

cord tissue from fALS and sALS patients display astrogliosis (Schiffer et al., 1996), implying that reactivity of astrocytes is not limited to the familial form of ALS. Interestingly, astrogliosis levels are similar between long surviving and short surviving ALS patients, although this is not the case for microglial activation and the amount of dendritic cells (Sta et al., 2011). An extra facet of astrogliosis in ALS is an increased immunoreactivity of tolllike receptor 4 in astrocytes of sALS patients (Casula et al., 2011). Astrogliosis in ALS mice is present at symptomatic stages preceeding microgliosis (Kiaei et al., 2006; Keller et al., 2009; Yang et al., 2011). Interestingly, GFAP is not necessary for astrogliosis as GFAP deficient astrocytes can still become reactive and do not affect survival of ALS mice (Yoshii et al.,

The expression of mutant SOD1 in astrocytes alters their function in vivo and in vitro. To begin, deletion of mutant SOD1 in astrocytes in two distinct ALS models demonstrates the detrimental effect of mutant SOD1 in astrocytes mainly post onset, as deletion increased lifespan of ALS mice (Yamanaka et al., 2008; Wang et al., 2011a). Intriguingly, astrogliosis was unaltered, implying that the negative effect of mutant SOD1 in astrocytes is not due to altered levels of astrogliosis (Yamanaka et al., 2008), but potentially by astrocytes inducing microgliosis (Yamanaka et al., 2008; Wang et al., 2011a). An alternative approach arrives from the field of transplantation in which non-transgenic mesenchymal stem cells are transplanted into the spinal cord of ALS rats and differentiate into astrocytes, thus diluting the mutant SOD1 positive astrocytes in the spinal cord (Boucherie et al., 2009). This approach also shows unaltered astrogliosis, but also decreased microgliosis and cyclooxygenase 2 (COX2) expression, and extends murine ALS life span (Boucherie et al., 2009). The processes explaining this hazardous effect of mutant SOD1 in astrocytes has been investigated in vitro. To begin, an interesting approach of transducing human astrocytes with wild-type SOD1 or mutant SOD1 increases inflammation in mutant SOD1 cultures (Marchetto et al., 2008). In addition, the mutant SOD1 transduced astrocytes provide a less viable environment for human embryonic stem cell derived motor neurons (Marchetto et al., 2008). The latter was rescued by using a NADPH oxidase 2 (NOX2) inhibitor, apocynin (Marchetto et al., 2008). Other studies concur that mutant SOD1 primary astrocytes exhibit a higher gene expression of cytokines on baseline and when stimulated by interferon γ (IFNγ) or tumor necrosis factor α (TNFα) (Hensley et al., 2006), implying once again that mutant SOD1 expression may affect the threshold of astrocytes to produce proinflammatory cytokines. Accordingly, the expression of interferon simulated genes is detected in astrocytes of presymptomatic ALS mice (Wang et al., 2011b) and genetic ablation and knockdown of the interferon alpha receptor type 1 (IFNAR1) increase ALS mouse survival by 5% and 10%, respectively (Wang et al., 2011b). Intriguingly, Aebischer et al. stress the importance of interferon signalling in mutant SOD1 astrocytes by demonstrating that mutant SOD1 astrocytes trigger the selective death of motor neurons mediated by IFNγ (Aebischer et al., 2011). This mechanism is dependent on the activation of the lymphotoxin-β receptor by LIGHT (TNFSF14) and genetic ablation of LIGHT extends survival of ALS mice by 13%, but does not postpone disease onset (Aebischer et al., 2011). Although this is a large

**2.3 Mutant SOD1 affects astrocytic inflammatory behaviour** 

increase in disease survival, clearly other mechanisms remain to play a role.

The above described altered functioning of mutant SOD1 expressing astrocytes is induced by an overexpression of multiple copies of mutant SOD1. It is unclear whether these effects

2011).

Attempts to minimise the inflammatory effect of astrocytes in ALS include genetic strategies targeting astrocytic knockdown of certain cytokines. Targeting of nuclear factor κB (NF-κB) activation specifically in astrocytes does not alter disease onset nor life span in ALS mice (Crosio et al., 2011). Which may be in part due to the mere decreased astrogliosis at presymtomatic stage (Crosio et al., 2011). Additionally, complete ablation of TNFα does not affect disease parameters (Gowing et al., 2006), also implying deletion of a single cytokine may not be sufficient to affect disease progression, as a general decrease in inflammation can be beneficial (see below). Instead of targeting cytokine production, ablating proliferating astrocytes was attempted in ALS, showing no effect on survival (Lepore et al., 2008a). This may be explained by the inability of astrocytes to proliferate in ALS (Gowing et al., 2008).

### **2.5 Therapeutic strategies targeting inflammation in ALS**

A number of strategies have been utilized to diminish inflammation in ALS mice, though often not specifically targeting astrocytes. It is worthwhile to note that it is commonly unclear whether anti-inflammatory strategies truly exert an anti-inflammatory function due to the age-matched analysis of astrogliosis instead of disease stage-matched analysis with drugs that successfully extend lifespan. Among these therapeutic strategies are those intended to pharmacologically block the cyclooxygenase (COX) pathway by administration of celecoxib, a selective COX-2 inhibitor, and deletion of the prostaglandin E2 receptor. Both strategies diminish inflammation, postpone disease onset and prolong survival in ALS mice (Drachman et al., 2002; Liang et al., 2008). Similarly, celastrol administration also postponed disease onset extended lifespan, while decreasing TNFα, nitric oxide synthases (iNOS), cluster of differentiation 40 (CD40) immunoreactivity and astrogliosis when assessing age-matched spinal cord tissue (Kiaei et al., 2005). Similar effects are obtained when providing ALS mice with folic acid (Zhang et al., 2008) or bee venom (Yang et al., 2010). Thalidomide extends survival in ALS mice by destabilising cytokine mRNA including TNFα (Kiaei et al., 2006). Additionally, its analog lenalidomide also prolongs survival (Kiaei et al., 2006), even when administered after symptom onset (Neymotin et al., 2009). To conclude, minocycline has shown a dramatic increase in survival of ALS mice and postpones symptom onset (Kriz et al., 2002; Van Den Bosch et al., 2002; Zhu et al., 2002). Interestingly, this effect is moment of administration dependent, as administration of minocycline post disease onset increases the astrocytic and microglial response in ALS mice, decreasing survival (Keller et al., 2011).

The Astrocytic Contribution in ALS: Inflammation and Excitotoxicity 383

intracellular calcium), glutamate is recycled for further use by the glial and endothelial cells, including astrocytes. Astrocytic glutamate re-uptake occurs by the glutamate transporters excitatory amino acid transporter 1 (EAAT1) and excitatory amino acid transporter 2 (EAAT2; also known as glutamate aspartate transporter (GLAST1) and glutamate transporter 1 (GLT-1), respectively). These transporters internalise glutamate, eg. into the astrocyte, for conversion to glutamine that is returned to the pre-synaptic neuron to be

Decreased glutamate uptake and EAAT2 protein levels are a common feature in both fALS and sALS and both in vitro and in vivo model systems (Staats and Van Den Bosch, 2009). In vitro transfection of primary cultured astrocytes with either mutant SOD1 or wild type human SOD1 down-regulates EAAT2 post transcriptionally (Tortarolo et al., 2004) and decreases EAAT2 transcription (Yang et al., 2009). Accordingly, glutamate transport is decreased in a neuronal cell line by mutant SOD1 transfection (Sala et al., 2005). Interestingly, this down-regulation also occurs in ALS model rats at pre-symptomatic stages through to end stage (Howland et al., 2002), at end stage only (Warita et al., 2002), in ALS model mice at end stage (Bendotti et al., 2001; Guo et al., 2010) and in post mortem patient spinal cords (Sasaki et al., 2001). In addition, in patient material the loss of EAAT2 and decreased tissue glutamate transport does not coincide with decreased levels of gene expression (Bristol and Rothstein, 1996), indicating that the loss is induced post transcriptionally, also in humans. Interestingly, a decrease of EAAT2 protein levels is not only in mutant SOD1 ALS models, but also a model of ALS/PDC (Wilson et al., 2003). In this model wild-type mice are fed with washed cycad flour containing β-methylaminoalanine (BMAA), which causes an ALS-like phenotype (Wilson et al., 2002). Although the loss of EAAT2 in ALS is apparent, it remains unclear whether this post transcriptional loss

To assess whether the loss of glutamate transport or the loss of EAAT2 specifically results in motor neuron loss, pharmacological and genetic tools have been used. To begin, research conducted by pharmacologically inhibiting glutamate transport in the rat spinal cord, failed to show any motor neuron loss despite the increased levels of glutamate (Tovar et al., 2009). In contrast, a similar experiment has been performed to address whether EAAT2 loss specifically induces motor neuron loss. EAAT2 null mice live for approximately 6 weeks before they succumb to epileptic seizures and are vulnerability to acute brain injury (Tanaka et al., 1997). To this end, heterozygous mice demonstrated the effect of approximately 40% knockdown of EAAT2 in the spinal cord in ALS mice (Pardo et al., 2006). This knockdown resulted in a non-significant decrease of symptom onset and significant, but moderate,

To assess the expected beneficial role of EAAT2 in ALS, transgenic mice overexpressing human EAAT2 in specifically astrocytes were crossbred with mutant SOD1 mice. Although glutamate uptake is increased in vivo and an overexpression of human EAAT2 is protective on cortical neurons in vitro, an effect on symptom onset or lifespan was absent (Guo et al., 2003). Possibly, the expression levels were insufficient to induce an effect or human EAAT2 is not as efficient as murine EAAT2 in mouse, as administration of ceftriaxone (a β-lactam antibiotic) and GPI-1046 (a synthetic, non-immunosuppressive derivative of FK506) increase EAAT2 protein levels and extend lifespan of ALS mice (Rothstein et al., 2005; Ganel et al.,

release again as glutamate (Laake et al., 1995).

of EAAT2 proceeds or follows the loss of motor neurons.

decrease of lifespan in ALS mice (Pardo et al., 2006).

**3.3 Targeting (astrocytic) EAAT2** 

Fig. 1. Mutant SOD1 in astrocytes affecting (motor) neuron survival.
