Molecular pathways involved in motor neuron degeneration

Protein aggregation and formation of cytoplasmic inclusions are characteristic of motor neuron degeneration in ALS, which is transversal to fALS and sALS. SOD1 and FUS

inclusions are observed in patients with each specific disease-causing mutation (Rosen 1993; Vance et al. 2009). The repeated expansion in C9ORF72 gene can sequester RNA-binding proteins that accumulate in RNA foci and originate dipeptide repeats (DPR) through repeat-associated non-AUG (RAN) translation forming toxic inclusions (Taylor et al. 2016).

Interestingly, TDP-43 aggregation is observed in almost all ALS cases (Ling et al. 2013).

Although the mechanisms are not fully understood, insights into the toxicity associated with protein aggregation have been mostly investigated in SOD1 models.

SOD1 is a ubiquitous small antioxidant protein which catalyses the dismutation of superoxide radicals in hydrogen peroxide and oxygen. More than 170 mutations in SOD1 were already described in ALS cases (http://alsod.iop.kcl.ac.uk). Conformational instability and inappropriate folding leading to the formation of protein aggregates is indicated as the cause of SOD1-mediated toxicity (gain-of-function) (Pasinelli and Brown 2006). In fact, dismutase activity did not correlate with disease pathogenesis. Interestingly, a recent study identified non-native SOD1 trimers as the toxic species involved in motor neuron death (Proctor et al. 2016). SOD1 aggregation seems to be an early event in the disease pathogenesis and SOD-containing aggregates increase in abundance with disease progression (Bruijn et al. 1998). As shown in Figure I.7, several other interrelated mechanisms have been implicated in motor neuron atrophy in ALS, including oxidative stress (Mitsumoto et al. 2008; Shaw et al. 1995), mitochondrial dysfunction (Igoudjil et al. 2011;

Sasaki and Iwata 2007; Vaz et al. 2015), ER stress (Atkin et al. 2008), impaired axonal transport and NF disorganization (De Vos et al. 2007), glutamate excitotoxicity (Spreux-Varoquaux et al. 2002; Vucic et al. 2008), impaired proteosome and autophagy (Kabashi et al. 2004; Tripathi et al. 2017), and Golgi fragmentation (Atkin et al. 2008). After the discovery of TDP-43 mutations and its presence in inclusions in about 90% of ALS patients, altered RNA metabolism, including biogenesis of miRNAs, gained a crucial attention (Paez-Colasante et al. 2015; Robberecht and Philips 2013). The majority of these mechanisms occur very early in ALS pathogenesis being observed before disease onset in mouse models.

A number of studies have been linking SOD1 accumulation with other disrupted mechanisms. For instance, SOD1 aggregates are known to sequester proteins such as the heat-shock proteins (Hsp) (Shinder et al. 2001), the poliubiquitine protein p62 (Gal et al.

2007) and the anti-apoptotic protein B-cell lymphoma 2 (BCL2) (Pasinelli et al. 2004), inducing reduced chaperone function, insufficient clearance of intracellular proteins and apoptotic-cell death, respectively. Mounting evidences showed that mSOD1 can accumulate at the intermembrane space and outer membrane of mitochondria causing vacuolation, alteration in morphology, mitochondrial dysfunction and degeneration (Igoudjil et al. 2011;

Tafuri et al. 2015).

Figure I.7 - Molecular mechanisms involved in motor neurodegeneration in ALS. Mutated superoxide dismutase 1 (mSOD1) accumulation and aggregation promotes the disruption of several interrelated molecular pathways within motor neurons. UPR is elicited due to the formation of aggregates causing ER stress, together with overwhelming of the proteasome and autophagy defects. Golgi fragmentation and alterations of vesicular trafficking have been reported as well. Accumulation of mSOD1 also occurs in mitochondria promoting energy impairment, as well as increased production of ROS, thus contributing to oxidative stress in the motor neuron.

Glutamate accumulation in the extracellular space can induce excitotoxicity with high Ca²⁺ entry, which further contribute to mitochondrial dysfunction. Impaired axonal transport is as well observed with the accumulation of neurofilaments, which may induce axonal retraction, and alterations in dynein- and kinesin-mediated transport.

Although not directly related to mSOD1-mediated motor neuron injury, altered RNA metabolism is observed in different ALS models being a transversal mechanism of toxicity in motor neurons. Importantly, it was reported that the expression of MMP-9 is determinant for motor neuron death in ALS. Finally, special features of motor neurons may confer increased vulnerability to injury, namely the existence of high metabolic rate and intense mitochondria activity in their long axons with consequent intrinsic oxidative stress, the presence of AMPA receptors lacking the GluR2 subunit (impermeable to calcium ions entry), and reduced heat shock proteins which increases motor neuron susceptibility to protein misfolding and formation of aggregates.

UPR, unfolded protein response; ER, endoplasmic reticulum; ROS, reactive oxygen species; Ca²⁺, calcium; MMP-9, matrix metalloproteinase-9; GluR2, glutamate receptor 2.

Three hypotheses have been proposed to describe the degeneration of both UMN and LMN in the corticospinal tracts: dying-back axonopathy, dying-forward and independent degeneration (Geevasinga et al. 2016; Nijssen et al. 2017). The dying-back phenomenon is well-known at the neuromuscular junction, where LMN axons are drawn back but no cell

communication with the muscles is loss leading to muscle wasting and atrophy. A pathway linking motor neuron dying-back and SOD1 accumulation is reflected in the altered axonal transport due to mitochondrial dysfunction, which causes energy depletion and altered local protein synthesis at the pre-synaptic terminal (De Vos et al. 2007). The dying-forward hypothesis, on the other hand, describes that the early hyperexcitability of UMN before disease onset (Vucic et al. 2008) induces glutamate excitotoxicity-induced LMN degeneration. Independent and random degeneration of both UMN and LMN has also been suggested (Geevasinga et al. 2016).

Early studies have described apoptosis as the typical cell-death pathway by which motor neurons degenerate, which is illustrated by the activation of caspases 3, 7 and 9 and release of cytochrome c (Cyt c) (Guegan et al. 2001; Ilzecka 2011; Pasinelli and Brown 2006). With the increased knowledge on the diverse cell death mechanisms, it was recently showed that necroptosis is activated in motor neurons in both fALS and sALS models (Ito et al. 2016; Re et al. 2014), as indicated by high levels of receptor-interacting serine/threonine-protein kinase 1 (RIP1) and mixed lineage kinase domain-like (MLKL), which are markers of this caspase-independent programmed cell death. The mechanisms triggering each specific cell death pathway needs, however, further elucidation.

Despite extensive research effort, the selective vulnerability of motor neurons to mutations in ubiquitously expressed proteins remains mysterious. One hypothesis relies on specific motor neuron features that distinguish them from all the other CNS neurons, as depicted in Figure I.7. Interestingly, it was recently indicated that among the motor neuron population, some types of them seem to be more resistant to ALS-linked damage, such as the ones in the oculomotor system or in the fast-twitch fatigue-resistant (FR) motor neurons (Nijssen et al. 2017; Rozas et al. 2016). For example, Kaplan and colleagues (2014) showed that high MMP-9 expression in fast motor neurons was determinant to induce their degeneration in the SOD1^G93A model, where enhancement of ER stress was a key feature. Additionally, oculomotor neurons were shown to be more resistant to calcium-induced mitochondrial stress as they express the glutamate receptor 2 (GluR2) in AMPA receptors, a subunit that is lacking in motor neurons.

The fact that selective death of motor neurons begin focally and asymmetrically, and then progressively spreads to contiguous groups of motor neurons, suggest the existence of specific disseminating mechanisms. Prion-like mechanisms have been reported in SOD-linked ALS, meaning that a misfolded SOD1 protein can be transferred and is able to induce misfolding and aggregation in recipient cells (Grad et al. 2015). It was shown that mSOD1 can be actively secreted through a chromogranin chaperone-like process (Urushitani et al.

2006), or via exosomes, as above mentioned, either in the outer leaflet or in the lumen (Grad et al. 2014b). While SOD1 aggregates are taken up by microglia macropinocytosis,

exosomes were also observed in naïve NSC-34 cells, though the intake mechanism was not explored (Grad et al. 2014a; Silverman et al. 2016). This spreading mechanism was maintained through several passages of conditioned media incubation. Interestingly, overexpression of chromogranin A was shown to accelerate disease progression by increasing the amount of misfolded SOD1 in the extracellular environment (Ezzi et al. 2010).

More recently, Ayers and colleagues (2016) showed that injection of spinal homogenates from paralyzed SOD1 mice into the sciatic nerve of adult SOD1^G85R-YFP mice (develop later onset disease at 20 months) precipitated disease onset, and the injury was progressively propagated from the lumbar to the cervical spinal cord likewise the process in humans.

Similar transmission mechanisms have been reported in TDP-43-ALS reinforcing the prion-like dissemination character of ALS pathogenesis (Grad et al. 2015; Silverman et al. 2016).

The well-described dissemination route of ALS also underlines an important role for glial cells that upon activation spread neuroinflammatory mediators to neighboring areas. In fact, ALS is a non-cell autonomous disease, where glial cells are actively involved in motor neuron injury, as it will be further addressed.