Several lines of evidence indicate that some of the neurosupportive functions of the astrocytes can be lost during the progression of the disease . As mentioned in the previous paragraph, astrocytes were consistently reported to exhibit a dysfunction of the glutamate uptake system in ALS patients and transgenic rodents, and this was proposed to contribute to excitotoxic motor neuron cell death (Rothstein et al. 1992, 1995; Bruijn et al. 1997; Howland et al. 2002; Guo et al. 2003; Pardo et al. 2006) . In addition, astroglia were reported to modulate the intrinsic susceptibility of motor neurons to glutamate by fine-tuning the expression of the GluA2 subunit of the glutamatergic α-amino-3-hydroxy-5-methyl-isoxazole propionate receptors (AMPARs) (Van Damme et al. 2007) . Remarkably, the presence of GluA2 in the AMPAR subunit composition critically impacts the biophysical properties of the receptor and determines its permeability to calcium ions (Ca²⁺). In particular, the expression of GluA2 normally renders AMPARs impermeable to Ca²⁺, because of an RNA editing event of the Q/R site (Q, unedited; R, edited) in the GluR2 mRNA (Sommer et al. 1991) . Yet, astrocyte expressing the mutant hSOD1^G93A protein appear to lose their capacity to modulate the expression of the GluA2 subunit of the AMPARs in the neighbouring motor neurons, thereby exposing them to enhanced Ca²⁺ influx and cell death (Van Damme et al. 2007). Besides their involvement in excitotoxic mechanisms, ALS astroglia were described to hold additional functional deficits, including mitochondrial defects as well as a reduced capacity to release the metabolic substrate lactate (Cassina et al. 2008; Ferraiuolo et al. 2011b) . Furthermore, the early observations of ubiquitinated SOD1- and active caspase-3-immunopositive inclusions in these cells, in the two hSOD1^G93A and hSOD1^G85R mouse models of the disease, raised the original idea that mutant SOD1 can be toxic to astrocytes themselves (Bruijn et al. 1997; Pasinelli et al. 2000) . In subsequent experiments, our group investigated the significance of such astrocytic inclusions in hSOD1^G93A– expressing transgenic mice. We found that these unusual cells are specifically located in the motor neuronal microenvironment in the ventral horns of the spinal cord (Fig. 11.2) . Furthermore, they display morphological and biochemical features reminiscent of degenerating cells, including an atypical spheroid morphology with an increased diameter and a reduced number or even the absence of cell processes immunopositive for the Glial Fibrillary Acid Protein (GFAP) (Rossi et al. 2008; Martorana et al. 2012) . Dying astrocytes were first observed at the pre-symptomatic stage (Rossi et al. 2008), i.e. at a time when motor neurons exhibit axonal damage, but their somas are still vital (Pun et al. 2006) . The number of spheroid astrocytes significantly increased concomitant with the onset of neuronal degeneration and the appearance of ALS symptoms (Rossi et al. 2008; Martorana et al. 2012). The relevance of this phenomenon was more recently reinforced in the context of the human disease by the identification of cells with an analogous aberrant morphology in the spinal cord of ALS patients (Martorana et al. 2012). Interestingly, we realized that degenerating astrocytes, in the spinal cord of hSOD1^G93A mice, were surrounded by glutamatergic terminals (Fig. 11.2), as to indicate that these cells can sense the neurotransmitter glutamate spilled over during synaptic activity (Rossi et al. 2008; Martorana et al. 2012) . Mechanistic studies in vitro revealed that astrocytes expressing different SOD1 mutations, either G93A or G85R, displayed an enhanced susceptibility to physiological concentrations of glutamate. The glutamatergic mechanism responsible for the deleterious event was found to involve metabotropic glutamate receptor 5 (mGluR5) signalling, which triggers an apoptotic gliodegenerative pathway in culture (Rossi et al. 2008) . More recently, the signalling prompted by the activation of group I mGluRs, including mGluR5, was further characterized .

In normal astroglial cells, the activation of group I mGluRs is well known to cause the formation of inositol 1,4,5 trisphosphate (IP₃), followed by IP₃-mediatedrelease of Ca²⁺ from the endoplasmic reticulum (ER) stores, which results in intracellular Ca²⁺ oscillations (Zur Nieden and Deitmer 2006; Gunnarson et al. 2009) . Yet, mutant hSOD1^G93A-expressing astrocytes responded to the receptor stimulation with an aberrant and persistent Ca²⁺ release from the intracellular stores. This correlated with mitochondrial disarrangement, cytochrome c release from mitochondria , and astrocyte degeneration (Martorana et al. 2012). These results are in line with the mitochondrial dysfunction previously described by Cassina and colleagues in mutant SOD1-expressing astrocytes (Cassina et al. 2008) .

The anti-apoptotic members of the Bcl-2 family, particularly Bcl-2 and Bcl-X_L, were largely reported to confer cell death resistance by fine-tuning intracellular Ca²⁺ signalling through direct interaction with the IP₃ receptor (IP₃R) channels (Chen et al. 2004; White et al. 2005; Zhong et al. 2006; Li et al. 2007; Rong et al. 2008, 2009) . Interestingly, structure-function analysis of Bcl-2 homologues revealed that their N-terminal homology domain 4 (BH4) is essential for inhibition of apoptosis (Hunter et al. 1996; Lee et al. 1996; Huang et al. 1998) . Our group thus investigated the impact of the BH4 domain of Bcl-X_L on astrocyte Ca²⁺ signalling by exploiting a biologically active BH4 peptide fused to the protein transduction domain of the HIV-1 TAT protein (TAT-BH4). We realized that TAT-BH4 modulates the IP₃R-dependent Ca²⁺ release from the ER, and restores spontaneous Ca²⁺ oscillations in mutant SOD1-expressing astrocytes. This tight control of IP₃Rs by the peptide prevents group I mGluR-driven aberrant release of Ca²⁺ from the intracellular stores, precludes the release of cytochrome c from mitochondria, and protects the cells from excitotoxic damage (Martorana et al. 2012) . Furthermore, chronic treatment of hSOD1^G93A transgenic mice with TAT-BH4 reduces degeneration of spinal cord astrocytes and shows a positive impact on the disease manifestations (Martorana et al. 2012) .

Consistent with a specific vulnerability of ALS astrocytes, functional astroglia deriving from human induced pluripotent stem cells (iPSC) expressing the mutant TDP-43^M337V protein were recently generated and shown to exhibit reduced cell survival (Serio et al. 2013) . Yet, in co-culture experiments, such cells were not harmful to wild-type motor neurons, possibly because in vitro differentiation may not have captured some of the detrimental properties that diseased astroglia harbor in vivo (Serio et al. 2013). In line with this conclusion, in transgenic rats expressing the same mutant form of TDP-43 in astrocytes, the loss of unhealthy astroglial cells paralleled progressive degeneration of motor neurons, denervation atrophy of skeletal muscles, and progressive paralysis (Tong et al. 2013) . Taken together, these findings support the view that glial cell sufferance can play a role in motor neuron degeneration also in ALS-TDP, though the evidence in this clinical subtype of the disease is still limited .

Another important function of the astrocytes that is worth mentioning concerns their association with the BBB, a complex structure tightly regulating the exchange of molecules between the brain parenchyma and the blood stream (reviewed in Daneman 2012) . On a structural standpoint, the BBB is made of endothelial cells, astroglia , pericytes and neurons, which globally establish a “neurovascular unit”. In this context, astrocytes project off their processes, and their endfeet ensheath the blood vessels. This strategic location suggests that they can importantly contribute to the maturation and maintenance of the BBB. Consistent with this vision, multiple studies in vitro have provided consistent evidence that astrocytes can influence the transendothelial resistance, a measure of BBB permeability; the organization of tight junctions between brain endothelial cells; and the luminal polarization of transporters, such the glucose transporter Glut-1 and the P-glycoprotein (Dehouck et al. 1990; Rubin et al. 1991; Hayashi et al. 1997; Sobue et al. 1999; Al Ahmad et al. 2011) . Although the integrity of the BBB is of outstanding importance for the maintenance of the optimal neuronal microenvironment, alterations of this structure have been described in a number of neurological disorders, including ALS (Daneman 2012). Early studies in the hSOD1^G93A ALS mice identified ultrastructural damage to astrocytes and endothelial cells, leading to vascular leakage of intravenously injected Evans Blue (EB) dye already at the pre-symptomatic stage (Garbuzova-Davis et al. 2007) . EB extravasation abnormalities were found in both the cervical and lumbar spinal cord (Garbuzova-Davis et al. 2007), and correlated with altered expression of critical proteins for the regulation of the efflux of catabolites from the brain parenchyma, such as Glut-1 and P-glycoprotein (Garbuzova-Davis et al. 2007). Evidence of BBB damage and microhemorrhages was later extended to other transgenic mice, harboring the G37R and G85R SOD1 variants (Zhong et al. 2008) , and to sporadic ALS cases (Garbuzova-Davis et al. 2012; Winkler et al. 2013) . Given the importance of the BBB in tuning the access of drugs to the CNS, these findings should be taken in consideration when designing new therapies for ALS.

In conclusion, taken together, these studies confirm a multifaceted role for astrocytes in ALS, even though they fail to provide conclusive evidence as to whether the multiple and diverse contributions of astroglia are due to distinct astrocyte subpopulations or to the same population receiving different external inputs (Molofsky et al. 2012; Oberheim et al. 2012) . We suggest that a better understanding of the impact of the distinct astrocytic mechanisms to ALS pathogenesis will be certainly beneficial to develop targeted therapeutics .

Microglia are the glial cell population deputed to the immune surveillance of the CNS. They typically respond to injury or disease with a massive activation in areas affected by neuronal degeneration. This condition is commonly known as “reactive microgliosis”. Reactive microglia have the capacity to release a wide variety of substances that can eitherlimit or exacerbate neuronal sufferance (reviewed in Aguzzi et al. 2013) . In both autoptic ALS cases and animal models, microgliosis was early recognized as a typical hallmark of the disease (Hall et al. 1998) . Yet, it was only with the recent advent of modern imaging technologies, coupled to the development of radiolabelled ligands, that the extent of this phenomenon could be assessed in vivo, in ALS patients, at the time of disease diagnosis. A number of positron emission tomography (PET) studies has been thus performed by neuroimaging the 18 kDa translocator protein (also known as the “peripheral benzodiazepine receptor”) (Turner et al. 2004; Corcia et al. 2012) , which is highly expressed in phagocytic inflammatory cells, including activated microglia (Papadopoulos et al. 2006) .

Several investigations, using mutant SOD1-expressing cellular and animal models of the disease, were then performed in order to elucidate the mechanism(s) that trigger the activation of such cells. These studies suggested that reactive microgliosis is prompted by the secretion of mutant SOD1 (Urushitani et al. 2006) , and the aggregated protein was described to be more efficient than the monomeric form in this activity (Roberts et al. 2013) . The relevance of microglia towards the disease manifestations was then tackled in vivo by genetic (Boillee et al. 2006; Gowing et al. 2008) and cell transplantation (Beers et al. 2006) approaches. Thus, reducing the expression of the mutant hSOD1^G37R protein selectively in microglial cells was reported not to affect the disease onset, although it slowed down ALS progression in transgenic mice (Boillee et al. 2006). In agreement with these results, transplantation of donor-derived microglia from the spinal cord of hSOD1^G93A mice into wild-type animals did not trigger ALS disease (Beers et al. 2006). Yet, transplantation of wild-type microglia induced neuroprotective effects on hSOD1^G93A-expressing motor neurons and prolonged the survival of transgenic mice (Beers et al. 2006). Altogether, these studies strengthen the idea that microglial cells modulate ALS progression, rather than influencing the development of the disease .

At the molecular level, the mechanism(s) by which microglia contribute to motor neuron sufferance still needs to be elucidated in details. Yet, a transcriptional profile analysis recently performed on acutely isolated spinal cord microglia from hSOD1^G93A transgenic mice revealed that these cells exhibit an ALS-specific phenotype characterized by the concomitant up-regulation of both potentially neurotoxic and neuroprotective factors (Chiu et al. 2013) . Among the molecules presumably mediating microglial toxicity, there are for example reactive oxidant species. In both human and mouse autoptic tissues, the ROS-generating NOX2 enzyme was in fact shown to be up-regulated specifically in microglial cells (Wu et al. 2006) . This event correlated with the appearance of oxidation products and protein carbonylation adducts, including oxidative modifications of insulin-like growth factor 1 (IGF1) receptors, which are critically involved in neuronal survival. The relevance of NOX2 in disease progression was then confirmed by the evidence that genetic ablation of its catalytic subunit ameliorated the phenotype and extended the lifespan of hSOD1^G93A ALS mice (Wu et al. 2006). Further insights into the mechanism of NOX2- driven toxicity have been provided a few years later by Harraz and colleagues, who unveiled a new function for the SOD1 enzyme. Along with its dismutase activity, SOD1 was shown to participate to the regulation of NOX2 function by binding to its regulatory protein Rac1. Under reducing conditions, this interaction appears to stabilize Rac1 activation, thus leading to NOX2 activation and superoxide production. However, when the local concentration of oxidant species increases, SOD1 is displaced from Rac1, and NOX2 activity appears to be negatively regulated. Yet, certain mutant forms of SOD1 display a stronger interaction with Rac1, and this leads to both sustained NOX2 activation and excessive superoxide production. Because chronic treatment with the NOX2 inhibitor apocynin tremendously increased the lifespan of hSOD1^G93A ALS mice, this mechanism was proposed to contribute to disease progression (Harraz et al. 2008) . The promising outcome of this study subsequently encouraged further investigations on NOX2 inhibitors in the context of ALS. Thus, the neuroprotective properties of diapocynin, a potent NOX2 inhibitor, were addressed in vitro and in vivo (Trumbull et al. 2012) . Despite this drug proved to prevent motor neuron death at lower doses in culture experiments when compared to apocynin, the treatment of hSOD1^G93A transgenic mice failed to extend the animal lifespan. Moreover, the dramatic effect on mouse survival originally observed with apocynin could not be replicated in this study (Trumbull et al. 2012) , thus raising serious concerns about the therapeutic potential of these agents. Nonetheless, the pathway(s) leading to NOX2 activation, and their contribution to ALS pathogenesis, remain of great interest for the scientific community. In fact, the enzyme protein disulfide isomerase, which assists the oxidative protein folding in the ER, was recently shown to be up-regulated in microglia from hSOD1^G93A mice. This suggests that the unfolded protein response (UPR) does not occur only in ALS motor neurons, but also in microglial cells (Jaronen et al. 2013) . Importantly, UPR was reported to trigger the activation of microglial NOX2 and, thus, to increase the production of superoxide in culture experiments (Jaronen et al. 2013). These findings suggest that UPR, initiated by protein misfolding, may lead to NOX2 activation and ROS generation. However, recent studies propose alternative pathways leading to the activation of NADPH oxidase. For example, experiments in vitro showed that stimulation of the purinergic receptor P2 × 7 in mutant hSOD1^G93A-expressing microglia results in enhanced NOX2 activity and ROS production, thus suggesting a deleterious role for the P2 × 7 receptor in ALS (Apolloni et al. 2013b) . At variance with these results, genetic ablation of the P2 × 7 receptor in hSOD1^G93A mice, however, exacerbated gliosis and motor neuron cell death; anticipated the clinical onset of the disease; and worsened ALS progression (Apolloni et al. 2013a). The protective effect of P2 × 7 suggested by these results in vivo was completely unexpected, given that stimulation of the P2 × 7 receptor had been linked also to the induction of a neurotoxic phenotype in hSOD1^G93A astrocytes (Gandelman et al. 2010) and to motor neuron apoptosis (Gandelman et al. 2013). Altogether, these observations reveal that the P2 × 7 receptor exerts a complex dual role in ALS. Evidence in favor of a neuroprotective role for microglia in the context of ALS was provided by a recent study addressing the biological significance of the glycosaminoglycan keratan sulfate (KS) (Hirano et al. 2013) . In both autoptic ALS tissues and hSOD1^G93Atransgenic mice, KS was found to be highly expressed by a specific subpopulation of microglial cells exerting anti-inflammatory functions during the early phase of the disease. Noteworthy, genetic ablation of KS in the CNS of hSOD1^G93A mice resulted in marked reduction of anti-inflammatory microglia, and this correlated with acceleration of the clinical symptoms and shortening of the animal lifespan (Hirano et al. 2013) .

More recently, microglia was also describe to dialogue with the peripheral immunocompetent cells, an interaction that proved to be both beneficial and deleterious (reviewed in Appel et al. 2010) . Several lines of evidence indicate that T cells infiltrate the spinal cord of ALS patients (Troost et al. 1989; Kawamata et al. 1992; Engelhardt et al. 1993) and mutant SOD1 transgenic mice (Beers et al. 2008; Chiu et al. 2008) . Of note, infiltrating T lymphocytes were described to slow down disease progression in mice by modulating the microglial inflammatory response (Beers et al. 2008, 2011; Chiu et al. 2008; Henkel et al. 2013) . Yet, spinal cord microglia were also reported to recruit peripheral monocytes to the CNS, a process that critically impairs neuronal viability and the mouse survival (Butovsky et al. 2012) .

In conclusion, these findings globally suggest that the neuroinflammatory response driven by microglia includes both neuroprotective and neurotoxic aspects in ALS. Therefore, the possibility of establishing a successful therapeutic effect relies on the selective targeting of the toxic components of the immunoreaction, rather than on the non-specific suppression of the immunocompetent response .

While astrocytes and microglia have been at the center of the investigations on ALS for almost a decade, other glial cell populations have attracted the attention in more recent times. Among others, these latter include oligodendrocytes and Schwann cells , the myelin-forming cells of the central and peripheral nervous system, respectively. Although early analyses of mutant SOD1 transgenic mice revealed SOD1-immunopositive inclusions in oligodendrocytes (Stieber et al. 2000) , this observation was not pursued immediately thereafter, and these cells have been involved in ALS pathogenesis quite recently. In both ALS patients and mutant hSOD1^G93A transgenic mice, it has been consistently reported a loss of the monocarboxylate transporter 1 (MCT1), a protein that is highly expressed in oligodendrocytes, and which is deputed to provide motor neurons with the metabolic substrate lactate (Lee et al. 2012; Philips et al. 2013) . Noteworthy, ubiquitous genetic ablation of MCT1 in mice was described to cause axonopathy (Lee et al. 2012). Furthermore, selective reduction of its expression in oligodendrocytes resulted in axonal sufferance (Lee et al. 2012) . This suggests that the shuttling of lactate from oligodendrocytes, through MCT1, is crucial for the energy supply to axons, and any disturbance of this transport can lead to axon dysfunction and, eventually, to neuronal cell death.

More recently, oligodendrocytes themselves were described to be a direct target of the disease. An apoptotic oligodendrocytic phenotype was, in fact, detected in the ventral grey matter of the spinal cord from mutant hSOD1^G93A transgenic mice before actual motor neuron loss became evident. Surprisingly, the overall number of oligodendrocytes was fully preserved (Philips et al. 2013). In demyelinating diseases, mature oligodendrocytes are typically replaced by the differentiation of NG2⁺ cells , which behave as precursors committed to the oligodendrocyte lineage (Chang et al. 2000) . In keeping with this, investigations of the NG2⁺ cell fate in symptomatic mutant SOD1 mice revealed that they exhibit an increased proliferation rate (Magnus et al. 2008; Kang et al. 2010, 2013; Philips et al. 2013) . The occurrence of this event is accompanied by a more frequent differentiation of NG2⁺ cells into oligodendrocytes (Magnus et al. 2008; Kang et al. 2010, 2013; Philips et al. 2013). Yet, at the time when mice show overt signs of the disease, the concomitant degeneration of early-born oligodendrocytes and, thus, the accelerated turnover of these cells, result in grey matter demyelination in ALS mice and human CNS (Kang et al. 2013). In addition, newly generated oligodendrocytes display reduced MCT1 expression, and thus fail to provide motor neuron with metabolic support (Kang et al. 2013; Philips et al. 2013). Since oligodendrocyte myelination is regulated by lactate (Rinholm et al. 2011) , it was proposed that reduced levels of MCT1 in oligodendrocytes may affect the process of myelination, in addition to depriving motor neurons of critical metabolic substrates (Magnus et al. 2008; Kang et al. 2010, 2013; Philips et al. 2013). A role for NG2⁺ cells and their oligodendrocyte progeny in ALS development was corroborated also by gene targeting experiments. Thus, diminishing mutant SOD1 expression in these cells was shown to delay the onset of the disease and to extend the life span of mutant hSOD1^G37R transgenic mice (Kang et al. 2013).

In the peripheral nervous system (PNS), individual axons are myelinated by the Schwann cells , the major glial component of the PNS. An initial indication of the involvement of this glial cell population in the pathogenesis of ALS was provided by the early observation that femoral nerves from post-mortem cases displayed a certain degree of myelin disruption (Perrie et al. 1993) . Later studies in mutant hSOD1^G93A mice provided further insights into the role of these cells in ALS by revealing that they exhibit signs of distress at the asymptomatic stage (Keller et al. 2009) . Yet, investigations addressing the impact of Schwann cells on ALS pathogenesis in vivo by cell-specific expression or ablation of mutant SOD1 provided slightly divergent outcomes. Thus, transgene-driven expression of mutant hSOD1^G93A protein in Schwann cells resulted neither detrimental for motor neurons nor deleterious to the disease manifestations (Turner et al. 2010) . On the other hand, removal of mutant hSOD1^G37R from Schwann cells by gene excision experiments accelerated disease progression in vivo (Lobsiger et al. 2009) . Considering the intimate relationship between motor neuronal axons and myelin, these results indicate that Schwann cells certainly deserve further attention in order to definitely elucidate their contribution to the disease.