Intracellular proteinaceous inclusions are a hallmark neuropathological feature of ALS. Inclusions are found in both degenerating neurons and surrounding glia (Piao et al., 2003; Nishihira et al., 2008; Zhang et al., 2008) and are found not only in the brainstem and spinal cord, but also in the cerebellum, hippocampus, and the frontal and temporal lobes (reviewed in Al-Chalabi et al., 2012). The most common inclusions are of ubiquitinated proteins, which are found in both the upper and lower motor neurons (Neumann et al., 2006), and are suggestive of defects in protein turnover (Blokhuis et al., 2013). Based on their morphology, these ubiquitinated inclusions are subdivided into skein-like inclusions that are filamentous in structure, and rounded Lewy body-like inclusions (Leigh et al., 1988; Lowe et al., 1988; Kato et al., 1989). ALS-associated ubiquitinated inclusions are typically positive for p62/sequestosome-1 (King et al., 2011), a ubiquitin binding protein involved in autophagy (see below). In the vast majority of sporadic and familial ALS cases ubiquitinated protein inclusions are positive for Tar DNA-binding protein of 43 kDa (TDP-43) (Arai et al., 2006; Neumann et al., 2006). TDP-43 was also identified as the pathological protein in frontotemporal lobar degeneration (FTLD)-related ubiquitinated inclusions, supporting the idea that ALS and FTD reside on the same spectrum of disease (Arai et al., 2006; Neumann et al., 2006; reviewed in Swinnen and Robberecht, 2014). Not only are aggregates of wild type TDP-43 found in nearly all cases of disease, but mutations in the TARDPB gene that encodes TDP-43 are also causative for ALS (Kabashi et al., 2008; Sreedharan et al., 2008). ALS-associated mutations in TARDBP lead to cytoplasmic TDP-43 mislocalisation and its aberrant incorporation into neurotoxic ubquitinated cytoplasmic aggregates (Barmada et al., 2010). Inhibition of the proteasome or autophagy leads to the aggregation of TDP-43 (Urushitani et al., 2010; Wang et al., 2010) (see below).

Although the majority of ubiquitinated inclusions are immunoreactive for TDP-43, notable exceptions are the inclusions found in ALS patients with mutations in SOD1 or FUS, which are negative for ubiquitinated TDP-43 but immunoreactive for mutant aggregated Cu/Zn superoxide dismutase (SOD1) and fused in sarcoma protein (FUS) respectively (Watanabe et al., 2001; Wang et al., 2002; Mackenzie et al., 2007; Kwiatkowski et al., 2009; Vance et al., 2009). Similar to TDP-43, mutant FUS demonstrates abnormal cytoplasmic redistribution and aggregation (Kwiatkowski et al., 2009; Vance et al., 2009; Dormann et al., 2010). Further familial ALS-associated mutant proteins that are prone to aggregation are valosin containing protein (VCP), dynactin-1 (DCTN1), optineurin (OPTN) and ubiquilin-2 (UBQLN2) (Levy et al., 2006; Maruyama et al., 2010; Deng et al., 2011; Koppers et al., 2012).

C9orf72-related ALS presents an outlier to classical ALS pathology. In C9orf72-related ALS TDP-43 proteinopathy is present, but additional inclusions are p62/sequestosome-1 and ubiquitin positive, yet devoid of TDP-43 (Al-Sarraj et al., 2011; Cooper-Knock et al., 2012; Mackenzie et al., 2014). In addition, the GGGGCC repeat expansion in the C9ORF72 gene gives rise to five species of dipeptide protein (DPR) inclusions (GA, GR, GP, PR, and PA) by repeat-associated non-AUG translation (Mann et al., 2013; Mori et al., 2013; Mackenzie et al., 2014).

Other ALS-associated inclusions include Bunina bodies (found post-mortem in approximately 86% of sporadic ALS patients) and hyaline conglomerate inclusions (Bunina, 1962; Kato et al., 1989; Murayama et al., 1989). Bunina bodies are comprised of cystatin C, transferrin, peripherin, and sortilin-related receptor CNS expressed 2 (SORCS2) and are found in the surviving lower motor neurons within the brain stem and spinal cord (Okamoto et al., 1993; Piao et al., 2003; Mizuno et al., 2006, 2011; Mori et al., 2015). They also contain small organelle fragments such as vesicles and ER (Okamoto et al., 2008; Kimura et al., 2014). Bunina bodies and TDP-43 positive inclusions have been shown to co-localize, and TDP-43 inclusion prevalence increased with Bunina body presence suggesting a synergy between the two inclusion types (Mori et al., 2010, 2014). Hyaline conglomerate inclusions are comprised of phosphorylated and non-phosphorylated neurofilaments (Hirano et al., 1984; Munoz et al., 1988; Hays et al., 2006) and their formation is possibly linked to defective axonal transport (Munoz et al., 1988; Ackerley et al., 2000, 2003).

The typical occurrence of these protein aggregates in ALS patients strongly suggest a collapse of proteostasis in ALS. These proteinaceous inclusions are replicated in a number ALS-associated animal models, including mutant SOD1, FUS, and TDP-43 transgenic mice. The fact that these models are able to effectively replicate some of the most prevalent neuropathological features of ALS suggests that the proteinaceous inclusions and aggregates are major contributors to disease pathogenesis.

As mentioned above, about 10% of ALS cases are inherited, usually in an autosomal dominant fashion (reviewed in Renton et al., 2014). A number of familial ALS-associated proteins are known to be involved in the proteostasis network, including C9orf72, VCP, p62/sequestosome-1, ubiquilin-2, optineurin, dynactin, and TANK binding kinase 1 (TBK1) (Table 1 and Figure 1). The role of these genes in proteostasis is discussed in detail in the following sections.

Molecular chaperones assist protein folding and help maintain proteins in their native folded state. In addition, they function in proteostasis to facilitate protein unfolding and disaggregation, and the targeting of terminally misfolded proteins for degradation. In relation to neurodegeneration, perhaps the key function of chaperones is to prevent protein aggregation under conditions of stress. Altered chaperone function has been implicated in ALS (Figure 2).

The presence of chaperones, such as heat shock cognate protein of 70 kDa (Hsc70) and heat shock protein of 90 kDa (Hsp90) in detergent-insoluble neuronal aggregates in the SOD1G93A transgenic mouse model of familial ALS and post-mortem human sporadic ALS cases indicates that altered chaperone function may contribute to disease pathogenesis (Watanabe et al., 2001; Basso et al., 2009; Bergemalm et al., 2010). In cell based assays and mouse models, ALS mutant SOD1G93A and G85R show increased interaction with a number of chaperones, including Hsp70, suggesting that ALS-associated mutant species may result in a depletion of available chaperones and chaperone activity, therefore leading to cellular toxicity (Tummala et al., 2005; Ganesan et al., 2007). Furthermore, reduced levels of the chaperone alpha-B-crystallin (CRYAB) and increased incorporation of other molecular chaperones, including Hsc70, into the insoluble aggregate fraction are features of a faster progressing phenotype in SOD1G93A transgenic mice (Marino et al., 2015).

As discussed above, redistribution of wild type or mutant TDP-43 to the cytoplasm and its aberrant inclusion into ubiquitinated cytoplasmic aggregates is a hallmark of ALS (Arai et al., 2006; Neumann et al., 2006). Interestingly, TDP-43 aggregates have been shown to interact with Hsp70 and Hsp90 under conditions of heat shock stress or reactive oxygen species (ROS) insult (Chang et al., 2013; Udan-Johns et al., 2014; Budini et al., 2015). The finding that activating Hsp70 reduces the insoluble aggregates of TDP-43, suggests that chaperone dysregulation could be a contributing factor to TDP-43 aggregate formation and toxicity (Chang et al., 2013). In agreement with this, knock-down of the molecular chaperones Hsp70 and Hsp90, leads to increased aggregation of TDP-43 (Zhang et al., 2010), while increasing levels of the small heat shock protein B8 (HspB8) increased the solubility of mutant TDP-43 and reduced toxicity (Crippa et al., 2010). Similarly, upregulation of the small heat shock protein CG14207 was able to reduce neurotoxicity of full length TDP-43 and the C-terminal fragment of TDP-43 in a Drosophila model of ALS (Gregory et al., 2012). Overexpression of Hsc70 in Drosophila also prevents the aggregation of FUS protein into insoluble fractions (Miguel et al., 2012).

Thus, increased chaperone levels may provide a protective mechanism in ALS. The reported upregulation of HspB1 and HspB8 chaperone expression in the lumbar spinal cord of 39 ALS cases (4 of which were familial ALS cases) compared to 19 control samples may be an indication of such a protective response (Anagnostou et al., 2010).

Cells use stress sensors and inducible pathways to respond to a loss of proteostatic control. One such pathway is the UPR, an adaptive response to the accumulation of misfolded proteins in the lumen of the ER (i.e., ER stress) (Figure 3). ER stress induced by accumulation of unfolded proteins activates three UPR pathways mediated by three ER membrane resident stress sensors, protein kinase RNA-like ER kinase (PERK), inositol-requiring protein 1 (IRE1) and activating transcription factor 6 (ATF6) (reviewed in Hetz, 2012). The UPR tries to restore proteostasis by (i) attenuation of translation, (ii) induction of chaperones to aid protein folding, and (iii) upregulation of degradation pathways such as ER-associated degradation (ERAD) and autophagy to remove misfolded proteins. If ER stress is short-lived, the UPR restores proteostasis and the cell survives. If in contrast ER stress persists, as for example in ALS, the UPR triggers apoptosis and the cell is lost.

Multiple lines of evidence from ALS patients and models suggest that ER stress may be a contributing factor to the development of ALS (Figure 4). Amorphous, granular material indicative of unfolded proteins has been shown to accumulate in the ER of motor neurons of sporadic ALS patients (Sasaki, 2010). In line with ER stress and induction of the UPR, the ER chaperone binding immunoglobulin protein/glucose-regulated protein 78 (BiP/GRP-78) was upregulated in motor neurons of sporadic ALS patients (Sasaki, 2010) and the levels of all three UPR stress transducers were found to be elevated in cerebrospinal fluid (CSF) and spinal cord of sporadic ALS patients (Atkin et al., 2008). In addition, a number of UPR-related chaperones, including protein disulfide isomerase (PDI) were present in CSF and throughout the motor neurons of sporadic ALS patients (Atkin et al., 2006, 2008). PDI has also been found in association with SOD1-inclusions in SOD1G93A transgenic mice, and it has been suggested that PDI is involved in the removal of these aggregates (Atkin et al., 2006). In SOD1G93A transgenic mice elevated UPR stress sensor levels have been observed prior to disease onset, as early as postnatal day 5, suggesting ER stress may be an early pathogenic event in the development of ALS (Atkin et al., 2008; Saxena et al., 2009). In vitro, expression of mutant SOD1A4V in the NSC-34 motor neuron-like cell line induces ER stress and apoptosis (Walker et al., 2010).

Interestingly, exposure of primary motor neurons to the CSF of sporadic ALS patients leads to ER stress, activation of the UPR and neuronal degeneration (Vijayalakshmi et al., 2009, 2011). How CSF elicits ER stress and degeneration is not clear but the process appears to involve activation of Caspase-12. Caspase-12 is known to mediate ER stress-induced apoptosis, and is activated by the UPR (Nakagawa et al., 2000; Martinez et al., 2010).

Chemical induction of ER stress in NSC-34 cells led to increased TDP-43 cleavage (Suzuki et al., 2011), while overexpression of ALS-associated TDP-43A315T and Q331K mutants in Neuro2a cells led to ER stress and activation of a number of UPR stress response pathways, including increased CHOP protein levels, increased nuclear XBP1 and increased activation of ATF6 (Walker et al., 2013). This activation of the UPR by mutant TDP-43 could further promote the cytoplasmic mislocalisation of TDP-43, and therefore toxicity (Walker et al., 2013). Similarly, expression of ALS mutant FUS and C9orf72-associated poly-GA DPRs have been shown to induce ER stress in NSC-34 cells and primary neurons, respectively (Farg et al., 2012; Zhang Y.J. et al., 2014).

Several ALS-associated proteins appear to directly interfere with pathways required for proteostasis and/or UPR. ALS mutant SOD1 has been shown to interact with Derlin-1, an ER protein essential for ERAD, and perturb the ERAD pathway (Nishitoh et al., 2008). Dysfunctional ERAD induces ER stress which eventually triggers apoptosis via the ASK1 pathway (Nishitoh et al., 2008). Mutations in vesicle-associated membrane protein-associated protein B (VAPB), an integral ER protein that has been implicated in UPR activation via the IRE1 and ATF6 pathways cause ALS8 (Nishimura et al., 2004; Kanekura et al., 2006; Gkogkas et al., 2008; Suzuki et al., 2009). Overexpression of ALS mutant VAPBP56S has been shown to induce ER stress (Suzuki et al., 2009). However, VAPB expression is down-regulated in ALS8 patient-derived iPSC neurons due to reduced expression of the VAPBP56S mutant (Mitne-Neto et al., 2011), therefore it is more likely ALS-associated loss of function of VAPB may predispose motor neurons to ER stress (Kanekura et al., 2006; Gkogkas et al., 2008; Suzuki et al., 2009). Indeed, since knock-down of VAPB inhibits activation of the IRE/XBP1 pathway in response to chemical ER stressors, VAPB appears to promote the UPR in reaction to ER stress under physiological conditions (Kanekura et al., 2006). How VAPB regulates the UPR is not yet clear but may involve its role in maintaining ER/mitochondria contacts via interaction with the mitochondrial outer membrane protein PTPIP51 (De Vos et al., 2012).

ER/mitochondria contacts allow mitochondria and ER to communicate directly with each other via the exchange of calcium signals (reviewed in Paillusson et al., 2016). Under physiological conditions, mitochondrial calcium activates the rate-limiting enzymes of the Krebs cycle and thereby increases oxidative phosphorylation and ATP synthesis to match local energy demand. In turn, energized mitochondria influence ER calcium homeostasis and redox dependent ER processes such as oxidative protein folding (reviewed in Walczak et al., 2012). Disruption of ER/mitochondria contacts has been shown to induce ER stress and the UPR (Simmen et al., 2005), possibly by disturbing the variety of ER chaperones, such as BiP, calnexin, calreticulin, ERp44, ERp57, and Sigma non-opioid intracellular receptor 1 (Sig1R) that are present in mitochondria-associated ER membranes (MAMs) (Hayashi and Su, 2007). Reduced levels of VAPB have been reported in the spinal cord of sporadic ALS cases, suggesting that ER/mitochondria contacts and the UPR may be impaired as a consequence (Anagnostou et al., 2010). Consistent with this possibility, neuronal overexpression of wild type human VAPB has been shown to slow disease and increase survival in SOD1G93A transgenic mice (Kim et al., 2016) but whether this is related to the restoration of the UPR and ER/mitochondria contacts remains to be determined. Interestingly disruption of ER/mitochondria contacts appears to be a common phenomenon in ALS with reduced ER/mitochondria contact sites found in mutant SOD1, Sig1R, TDP-43, and FUS-related ALS (Lautenschlager et al., 2013; Stoica et al., 2014, 2016; Watanabe et al., 2016).

Mutations in the SIGMAR1 gene that encodes Sig1R cause a juvenile form of ALS (ALS16) (Al-Saif et al., 2011). Sig1R1 is an ER protein that resides at ER/mitochondria contacts where it interacts with BiP and regulates calcium exchange by acting as a ligand-operated receptor chaperone for the inositol 1,4,5-trisphosphate receptor 3 (IP3R3) (Hayashi and Su, 2007). ALS-associated loss of function of Sig1R has been linked to dysfunctional ER/mitochondria communication calcium dysfunction and ER stress (Prause et al., 2013; Vollrath et al., 2014).

Mutations in the ATPase VCP cause of 1–2% of familial ALS cases as well as inclusion body myopathy (IBM) with Paget’s disease (PDB) and FTD (Johnson et al., 2010; Koppers et al., 2012). Among other functions, VCP is involved in protein degradation, ERAD, ER stress and autophagy (reviewed in Meyer and Weihl, 2014), and pathogenic mutations have been shown to impair mitochondrial proteostasis, attenuate ERAD and lead to an impaired stress response (Fang et al., 2015; Wang et al., 2016). Mutant VCP has been linked to altered TDP-43 metabolism in Drosophila (Ritson et al., 2010) and spinal cord motor neurons of mutant VCP transgenic mice exhibit TDP-43 pathology (Custer et al., 2010). Mutant VCP expression also leads to redistribution of wild type TDP-43 from the nucleus to the cytoplasm in vitro and in vivo (Gitcho et al., 2009; Custer et al., 2010; Ritson et al., 2010). Thus, the typical TDP-43 pathology observed in ALS may be a direct consequence of dysfunctional proteostasis.

The characteristic pathological protein accumulations seen in ALS are indicative of defective protein clearance. In eukaryotes, the majority of misfolded proteins are degraded by the UPS in which poly-ubiquitin-tagged substrates are directed to and degraded by the 26S proteasome, a multimeric ATP-dependent protease complex comprised of the core 20S protease capped with two 19S regulatory subunits (reviewed in Eytan et al., 1989; Driscoll and Goldberg, 1990; Hoffman et al., 1992; Voges et al., 1999; Ciechanover and Kwon, 2015). Evidence from both familial and sporadic cases of ALS indicates proteasome dysfunction may be involved in disease pathogenesis (Figure 5).

ALS mutant SOD1 has been shown to directly interact with the 19S regulatory subunits of the proteasome, which could contribute to the observed proteasomal inhibition seen in multiple models of ALS (Urushitani et al., 2002; Kabashi et al., 2004; Cheroni et al., 2005, 2009). Furthermore, reduced expression of UPS components in the spinal cord of SOD1G93A transgenic mice has been reported (Basso et al., 2009; Marino et al., 2015), and ALS mutant SOD1 itself is poly-ubiquitinated and cleared by the proteasome (Niwa et al., 2002; Urushitani et al., 2004). Possibly, age-related reductions in UPS activity or high demand, leads to the formation of cytotoxic mutant SOD1 inclusions (Kitamura et al., 2014).

Other familial ALS-associated proteins also support the role of a defective UPS in the development of ALS. In addition to its role in ERAD and ER stress discussed above, VCP is involved in substrate delivery to the proteasome (Dai and Li, 2001; Meyer and Weihl, 2014) and regulation of proteasome activity (Wojcik et al., 2004; Clemen et al., 2015). ALS-associated missense mutations in VCP have been shown to disrupt VCP-proteasome interaction, possibly resulting in defective proteasomal clearance of ubiquitinated proteins resulting in aggregation (Barthelme et al., 2015).

Mutations in UBQLN2 which encodes the ubiquitin-like protein, ubiquilin-2, are associated with X-linked ALS and ALS-dementia (Deng et al., 2011). Ubiquilin-2 is a member of the ubiquilin family, which regulates the degradation of ubiquitinated proteins. Patients with ubiquilin-2 mutations display the classical ALS-associated ubiquitinated protein aggregates, which, interestingly, are also positive for ubiquilin-2 (Deng et al., 2011). Like VCP, ubiquilin-2 is able to bind poly-ubiquitin chains and aids in substrate delivery to the proteasome (Ko et al., 2004). ALS-associated ubiquilin-2 mutants are defective in proteasome binding, resulting in defective substrate delivery to the proteasome and the accumulation of poly-ubiquitinated proteasome substrates (Chang and Monteiro, 2015). Ubiquilin-2 has also been implicated in an autophagy independent pathway that clears protein aggregates via the proteasome. In this pathway, ubiquilin-2 shuttles Hsp70-bound aggregated proteins to the proteasome to allow degradation. ALS mutant ubiquilin-2 has lost the ability to bind to Hsp70 and this sensitizes cells to protein stress (Hjerpe et al., 2016).

As discussed above, ALS mutant VAPB is linked with ER stress and defective UPR. However, there is also evidence that ALS mutant VAPB may impair the UPS. VAPBP56S is found in cytoplasmic aggregates, which by itself could be an indication of defective proteasome function (Suzuki et al., 2009; Moumen et al., 2011). Expression of mutant VAPBP56S in cells promotes the formation of ubiquitinated aggregates, as well as the accumulation of other proteasomal substrates (Chen et al., 2010; Moumen et al., 2011). Similarly, ubiquitinated inclusions have been found in the spinal cord motor neurons of VAPBP56S transgenic mice (Tudor et al., 2010). Furthermore, mutant VAPBP56S has been demonstrated to interact with the 20S subunit of the proteasome, suggesting proteasome sequestering and trapping, and therefore UPS dysfunction, could be a contributing factor in ALS8 (Moumen et al., 2011).

Impaired proteasome function has also been reported in sporadic ALS cases (Kabashi et al., 2012). Levels of the 20S proteasome subunit were significantly reduced in motor neurons of sporadic ALS cases compared to controls, and proteasomal activity was found to be impaired in the spinal cord (Kabashi et al., 2012). As sporadic ALS cases make up the vast majority of clinically diagnosed ALS cases, this finding suggests that defective proteasomal activity may explain ALS pathology and be a contributing factor in a high percentage of ALS cases. In this context, it is noteworthy that conditional knockout of the proteasome subunit Rpt3 in motor neurons in mice results in motor neuron degeneration and locomotor defects accompanied by the accumulation of a number of ALS-related proteins, including TDP-43, FUS and optineurin in intraneuronal inclusions (Tashiro et al., 2012). Such inclusions are obviously reminiscent of the classical ALS pathology, suggesting that proteasomal dysfunction alone may be sufficient to drive the development of ALS.

Autophagy is an intracellular lysosomal degradation pathway responsible for the bulk clearance of cytoplasmic components such as misfolded proteins and damaged organelles. Autophagy requires the formation of a double membrane structure called the autophagosome, which encapsulates autophagic substrates prior to their transport to, and fusion with the lysosome (reviewed in Bento et al., 2016). The autophagic process can be divided into four distinct stages: (i) translocation and initiation, (ii) elongation and recruitment, (iii) completion, and (iv) lysosome fusion and degradation. The stages of autophagy are detailed in Figure 6.

Autophagy is essential for neuronal health. Inhibition of autophagy in neurons by neuronal-specific knockout of essential autophagy genes such as Atg7, Atg5, and RB1CC1 (FIP200) causes neurodegeneration in mice in absence of any other contributory factors (Hara et al., 2006; Komatsu et al., 2006; Liang et al., 2010). Loss of autophagy in these mice is accompanied by progressive deficits in motor function, including abnormal limb-clasping reflexes (also observed in ALS mutant SOD1 transgenic mice) and a reduction in coordinated movement. Furthermore, reminiscent of many neurodegenerative diseases including ALS, loss of basal autophagy causes accumulation of neuronal ubiquitin-positive cytoplasmic inclusion bodies (Hara et al., 2006; Komatsu et al., 2006; Liang et al., 2010). Basal neuronal autophagy appears to be especially required for the maintenance of axons as loss of autophagy causes axonal dystrophy (Komatsu et al., 2006).

As discussed above, molecular chaperones play an essential role in the proteostasis network. They are upregulated in response to protein stress (e.g., by the UPR), and thus are amenable to regulation. Furthermore, evidence suggests that inhibition of chaperone activity may contribute to ALS.

Hydroxylamine derivatives, such as Arimoclomol stabilize the transcription factor heat shock factor protein 1 (HSF-1) in its active state leading to its prolonged activation and the upregulation of a number of heat shock family chaperones, including Hsp60, Hsp70, and Hsp90 (Vígh et al., 1997). Arimoclomol has been shown to delay disease progression, extend lifespan, increase muscle function, and prevent aggregation of mutant SOD1 in SOD1G93A mouse models of ALS (Kieran et al., 2004; Kalmar et al., 2008, 2012). Importantly, HSF-1 does not bind its corresponding DNA elements in the absence of cellular stress. Thus, the prolonged activation of HSF-1 by hydroxylamine derivatives only enhances the heat shock response in already stressed cells, therefore mitigating the potentially toxic induction of the heat shock response and chaperone upregulation in otherwise healthy cell populations (Hargitai et al., 2003).

In a similar fashion to Arimoclomol, treatment with the Hsp90 inhibitor 17-AAG activates Hsp70 reducing the number of pathological TDP-43 aggregates in HEK293T cells overexpressing aggregation-prone TDP-43 C-terminal fragments (Chang et al., 2013).

ER stress appears to be involved in ALS pathogenesis. Accordingly, the UPR which alleviates ER stress under physiological conditions may be an attractive therapeutic target.

The PERK branch of the UPR leads to phosphorylation of the eukaryotic translational initiation factor eIF2α which decreases global protein synthesis and up-regulates the translation of selected stress-induced mRNAs such as the transcription factor ATF4. ATF4 induces expression of genes involved in amino acid metabolism, resistance to oxidative stress, and the proapoptotic transcription factor CHOP (Figure 3). The ER stress inhibitors salubrinal and guanabenz prevent eIF2α dephosphorylation and keep eIF2α in its inactive state (Boyce et al., 2005; Tsaytler et al., 2011). Salubrinal treatment protected SOD1G93A transgenic mouse motor neurons from ER stress (Saxena et al., 2009) and treatment of SOD1G93A transgenic mice with guanabenz delayed onset of disease, attenuated motor neuron loss and significantly extended lifespan (Jiang et al., 2014; Wang et al., 2014). In contrast, others have shown that guanabenz treatment accelerated disease progression in SOD1G93A transgenic mice (Vieira et al., 2015). The reason for these opposing findings is not clear, but may be attributable to differences in study design (discussed in Vieira et al., 2015). Both guanabenz and salubrinal attenuated ER stress and reduced paralysis and neurodegeneration in mutant TDP-43 Caenorhabditis elegans and zebrafish models of ALS (Vieira et al., 2015).

Perhaps counter-intuitively, conditional knockout of XBP-1 was shown to extend the survival of female SOD1G86R transgenic mice (Hetz et al., 2009). However, in this model, abrogation of the UPR coincided with increased autophagic clearance of SOD1 aggregates, suggesting that a lower aggregated protein burden elicits the protective effect of XBP-1 knockout (Hetz et al., 2009). Similarly, knockout of ATF4 also delayed disease onset and prolonged life span in SOD1G86R transgenic mice. ATF4 deficiency completely prevented the induction of pro-apoptotic genes, but increased SOD1 aggregation suggesting that, in this case, prevention of UPR-induced apoptosis mediated protection (Matus et al., 2013).

While increasing protein folding may help to reduce the pathological protein accumulations in ALS, an alternative approach is to promote their degradation by activation of the UPS and/or autophagy.