Autophagy is a regulated auto-degradative intracellular process highly conserved across many species, from yeast to mammals. Macroautophagy, hereafter referred to as “autophagy,” begins with the formation of a primitive membrane structure called the “omegasome,” which arises from intracellular membranous domains. The omegasome sequesters cytosolic targets (Lamb et al., 2013; Mizumura et al., 2014), which can include aggregates or insoluble proteins, damaged and/or oxidized organelles such as mitochondria (mitophagy), endoplasmic reticulum (ERphagy), peroxisomes (pexophagy), lipid droplets (lipophagy), and ferritin (ferritinophagy) (Iwata et al., 2006; Narendra et al., 2008; Lamb et al., 2013), and becomes known as an autophagosome. The autophagosome, in turn, fuses with a lysosome to form an “autophagolysosome,” where the cargo is degraded by hydrolytic lysosomal enzymes in either a bulk or selective manner (Lamb et al., 2013; Mizumura et al., 2014). This autophagic degradation results in the production of available metabolites that support key metabolic cellular requirements (Bento et al., 2016) (Figure 1). Nutritional stress (starvation) represents the most common condition to induce autophagy; however, other stress stimuli such as ER stress, hypoxia, mitochondrial damage, free radicals, and genotoxic stress can also induce it (Bento et al., 2016). Importantly, stress-induced autophagy and basal autophagy are crucial to maintain cellular homeostasis in vitro and in vivo (Fernandez and Lopez-Otin, 2015). Indeed, unbalanced autophagy, either increased or reduced, contributes to the progression of various human diseases, including neurodegenerative diseases, muscle disorders, cancer, metabolic dysfunctions, cardiovascular, and infectious diseases (Fernandez and Lopez-Otin, 2015).

How can autophagy be involved in the regulation of such a variety of human illnesses and disorders? As previously mentioned, autophagy functions in all cells of the body to remove toxic and damaged material. Thus, a decrement to autophagy promotes the accumulation of material that is normally removed from the cell and adversely affects cell survival.

During autophagy, the process of cargo selection is generally non-selective; however, specific autophagy receptors have been identified that mediate the clearance of specific cargoes (Bento et al., 2016). One of the most studied autophagy receptors is the protein p62/SQSTM1, which is both a selective autophagy substrate and a cargo receptor for autophagic degradation of polyubiquitinated proteins (Katsuragi et al., 2015; Bento et al., 2016). p62/SQSTM1 contains a light chain 3 (LC3)-interacting region (LIR), which targets the LC3 anchored in the omegasome membrane; cargoes are thus directed to the nascent autophagosome for clearance (Birgisdottir et al., 2013). Other LIR-containing autophagy receptors are the proteins B-cell lymphoma 2 (Bcl2) Interacting Protein 3 (BNIP3) and NIX, two homologous Bcl2 homology 3 (BH3)-only proteins, which localize in the outer mitochondrial membrane and mediate mitophagy (autophagic clearance of mitochondria) (Kanki, 2010; Hanna et al., 2012). Both BNIP3 and NIX contain a LIR motif in the amino terminus that faces the cytosol, leading to LIR-mediated mitophagy by interaction of the LIR motif with LC3 during autophagosome formation (Kanki, 2010; Hanna et al., 2012).

As mentioned above, the degradation of the autophagic cargo is dependent on lysosomal enzymes. Impaired fusion of the autophagosome with the lysosome, or inhibition of the acidification of the lysosomal compartment prevents the degradation of the autophagosome luminal content and promotes autophagosome accumulation, thus inhibiting “autophagic flux” (Klionsky et al., 2016). The term “autophagic flux” is defined as the difference between the rate of autophagosome formation and cargo clearance by lysosomes (Klionsky et al., 2016). A blockage in the autophagic flux can be observed in different human diseases, and is quite common in neurodegenerative diseases, where an abnormal accumulation of autophagosomes caused by inefficient fusion with lysosomes occurs (Ravikumar et al., 2005). The mechanisms responsible for impairing autophagic flux in neurodegenerative disease vary, and represent an active subject of investigation.

The site where autophagosome assembly occurs in the cell remains controversial, and different locations and organelles have been proposed as the site of assembly and source of the membranes composing the autophagic vacuoles. The ER (Hayashi-Nishino et al., 2009; Yla-Anttila et al., 2009), mitochondria (Hailey et al., 2010), the plasma membrane (Ravikumar et al., 2010; Moreau and Rubinsztein, 2012), the Golgi apparatus (Geng et al., 2010; Bodemann et al., 2011) and mitochondria-associated membranes (MAMs) (Hamasaki et al., 2013; Patergnani et al., 2015) have all been described as subcellular compartments from which the phagophore and the autophagosome might arise. More than 35 autophagy-related (atg) genes are involved in the process of autophagic vacuole formation (Bento et al., 2016). The primitive phagophore becomes an autophagosome thanks to the activity of the class III phosphoinositide 3-kinase (PI3KC3), which is a large macromolecular complex formed by a catalytic sub-unit the vacuolar protein sorting 34 (VPS34), a regulatory sub-unit VPS15 (p150), ATG6 (also known as Beclin 1 in mammals) and ATG14L (Russell et al., 2014). The PI3KC3 also binds, through Beclin 1, the activating molecule in Beclin 1-related autophagy 1 protein (AMBRA1), which stimulates the activation complex when autophagy is induced (Di Bartolomeo et al., 2010). The pro-autophagy role of Beclin 1 in the PI3KC3 can be also regulated by Bcl-2 or Bcl-XL located at the ER, which down-regulates autophagy by binding Beclin 1, specifically through the Bcl-2 homology type 3 (BH3) domain (Maiuri et al., 2007a,b).

The increased activity of PI3KC3 generates phosphatidylinositol 3-phosphate (PtdIns3P) in membrane microdomains where proteins with the PtdIns3P-binding (FYVE) and the conserved Phox-homology (PX) domains are recruited to form a protein elongation platform at the omegasome (Simonsen et al., 2004; Zhong et al., 2009). One of these proteins with FYVE domains is the double FYVE domain-containing protein 1 (DFCP1), which is highly accumulated in a PtdIns3P-dependent manner in conditions of starvation-induced autophagy (Axe et al., 2008). Other proteins whose recruitment is mediated by PtdIns3P are the WD-repeat proteins interacting with phosphoinositides, WIPI-49, which is four member protein family: WIPI1, WIPI2, WIPI3, and WIPI4 (Proikas-Cezanne et al., 2004).

The elongation and expansion of the phagophore requires the concerted actions of two ubiquitin-like conjugation systems (Polson et al., 2010). The first ubiquitin-like conjugation system is formed by ATG12–ATG5–ATG16L. ATG16L1, which is directly bound through WIPI2 to the membrane micro domains, allows the expansion of the phagophore (Itakura and Mizushima, 2010; Walczak and Martens, 2013). Specifically, the activity of this ubiquitin-like conjugation system begins with the conjugation of ATG12 to the lysine residue of ATG5 by the sequential reactions of the E1-like enzyme ATG7, and the E2-like enzyme ATG10. The resulting ATG12–ATG5 conjugate interacts non-covalently with ATG16L, and they oligomeryze, forming a large multimeric complex (Hamasaki et al., 2013). In the second ubiquitin-like system, the cytosolic forms of ATG8 (microtubule-associated protein 1 light chain 3 (MAP-LC3), GABARAP and GATE-16) are cleaved at their C-terminus by the cysteine protease ATG4B and then conjugated at the amino group with the lipid phosphatidylethanolamine (PE) by ATG7 and by the E2-like enzyme ATG3 (Tanida et al., 2004; Katahira et al., 2008). LC3 is the most characterized ATG8 family member and its lipidation with PE generates LC3-II (Otomo et al., 2013). LC3-II is specifically bound to the autophagic vesicle and is widely used as an autophagic marker (Klionsky et al., 2016). Once the autophagosome is formed, the ATG12-ATG5-ATG16L complex associated to the convex surface of the autophagosomal membrane dissociates and is released into the cytoplasm to be reused for the biogenesis of new autophagy vesicles (Lamb et al., 2013).

Once the autophagosomes are fused with the lysosomes, hydrolytic lysosomal enzymes are activated, and they degrade both the luminal content of the autophagosome and its inner membrane. This process of degradation allows the formation of new metabolites that recycle back to the cytoplasm for being reused (Lamb et al., 2013). Autophagosomes fuse with lysosomes by the action of several soluble N-ethylmaleimide-sensitive fusion (NSF) attachment protein receptors (SNARE). These include the vesicle-associated membrane protein (VAMP)8, syntaxin-7, syntaxin-8, and t-SNAREs homolog 1B (Vti1B) (Moreau et al., 2013). Impairment in lysosome-autophagosome fusion induced by dysfunctions in SNARESs proteins decreases the size of the autophagic vesicles and inhibits their maturation into autophagolysosomes. This blockage in the fusion process leads to lysosomal storage disorders, which are characterized by an abnormal accumulation of substrates in the lysosomes (Fraldi et al., 2010; Moreau et al., 2013).

In the regulation of the autophagic process, the most characterized protein regulators are the mechanistic target of rapamycin, mTOR, and the AMP-activated protein kinase (AMPK), which are considered cellular nutritional sensors able to modulate different signaling pathways (Kim et al., 2011; Meijer et al., 2015; Chantranupong and Sabatini, 2016). mTOR is a constitutive Ser/Thr protein kinase sensitive to nutritional and energetic changes, and comprises two large protein complexes, the mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2) (Russell et al., 2014). mTORC1 is formed by mTOR, Regulatory-associated protein of mTOR (RAPTOR), mTOR associated protein, LST8 Homolog (mLST8), DEP domain containing mTOR-interacting protein (DEPTOR) and proline-rich Akt substrate of 40 kDa (PRAS40). In basal conditions the kinase activity of mTOR renders unc-51-like kinase 1 protein (ULK1, ortholog of Atg1 in humans) inactive by phosphorylation of Ser757, Ser758 and Ser638 (Kim et al., 2011; Meijer et al., 2015). In these conditions, the ULK complex formed by ULK1, ATG13 and FAK family kinase-interacting protein of 200 kDa (FIP200) remains inactive and autophagy is inhibited (Hara et al., 2008; Wong et al., 2013). In conditions of nutrient deprivation, a low ATP/AMP ratio activates AMP-activated protein kinase (AMPK), which up-regulates the Ser/Thr kinase activity of ULK by phosphorylation on Ser317, Ser777, and Ser555 (Wong et al., 2013). Once activated ULK is released from mTORC1 and relocates to phagophore assembly sites where it increases the activity of PI3KC3, which, as previously mentioned, generates PtdIns3P in the membrane microdomains (Axe et al., 2008). AMBRA can also regulate the relocalization of PI3KC3 to the autophagosome nucleation sites (Di Bartolomeo et al., 2010). Once autophagy is induced, ULK1 phosphorylates AMBRA, which releases PI3KC3 from the microtubule-associated dynein motor complex, initiating the process of autophagosome elongation (Di Bartolomeo et al., 2010) (Figure 2).

On the other hand, AMPK up-regulates autophagy through phosphorylation of tuberous sclerosis complex 2 (TSC2) at Thr1227, which, together with TSC1, forms the TSC complex (Inoki et al., 2003; Orlova and Crino, 2010). Active TSC2 induces autophagy by suppression of the RAS homolog enriched in brain (Rheb) protein, which is required to maintain mTOR constitutive activity (Orlova and Crino, 2010). AMPK can also directly phosphorylate the mTORC1 subunit RAPTOR, which inhibits mTORC1, thereby inducing autophagy (Gwinn et al., 2008) (Figure 2).

The protein kinase B (PKB, also known as AKT) is another negative regulator of autophagy (Heras-Sandoval et al., 2014). When nutrients and growth factors are provided, tyrosine kinase receptors are activated by autophosphorylation and stimulate the small GTPase Ras and class I Phosphatidylinositol 3 kinase (PtdIns3K) (Russell et al., 2014). Active class I PtdIns3K increases the levels of PtdIns(3)P in the membrane, promoting the recruitment and the subsequent activation of AKT (Russell et al., 2014). AKT phosphorylates TSC2 on multiple sites, leading to the disruption of the TSC1-TSC2 complex and promoting mTORC1 activation, thus inhibiting autophagosome formation (Inoki et al., 2002; Potter et al., 2002).

In some mouse models of Alzheimer's disease (AD) the gene encoding for Beclin 1 shows decreased mRNA levels in brain tissue, suggesting reduced autophagosome formation and increased development of AD (Pickford et al., 2008). Consistently, crossing mice that overexpress the human amyloid precursor protein, with those carrying a heterozygous deletion of Beclin 1, which per se leads to neurodegeneration, aggravates the AD phenotype (Pickford et al., 2008). Beclin 1 activity is also affected in Huntington disease (HD). Indeed, the mutant Huntington gene, which is critical for the onset of HD, recruits Beclin 1, inhibits its activity and, therefore, autophagosome formation (Shibata et al., 2006). Also, mTOR appears to be sequestered in cell and mouse models of HD, as well as in human brains of HD patients, indicating that the initiation of the autophagic process has been inhibited (Ravikumar et al., 2004).

Additionally, the first step of the autophagic process is compromised in some cases of Parkinson's disease (PD), where omegasomes formation is inhibited due to the mislocalization of the protein ATG9, preventing the formation of new autophagosomes. This reduces the clearance of the protein α-synuclein, which accumulates in intra-neuronal inclusions called Lewy bodies, a feature of the brain tissue of PD patients (Winslow et al., 2010).

Inefficient recognition of aggregate proteins during the autophagic process has been proposed as a cause of the development of neurodegenerative diseases (Sarkar et al., 2009). As previously described, recognition of certain posttranslational modifications by molecules that bind both cargo and components of the autophagic machinery drives the selectivity of the autophagic cargo (Sarkar et al., 2009). For instance, the protein p62/SQSTM1 is specifically recognized and degraded within autophagosomes (Bjorkoy et al., 2005). Mutations in p62/SQSTM1 have been found in ALS patients, which, as will be better explained in the next sections, leads to p62/SQSTM1 accumulation in inclusion bodies, a typical feature of the disease (Fecto et al., 2011). Also, some forms of HD are characterized by alterations in the process of cargo sequestration. Indeed, due to mutations in the Huntington protein, cellular and mouse models of HD show an inefficient engulfment of cytosolic components in the lumen of autophagosomes, which, even though they form normally, remain empty, thereby promoting the accumulation of the waste material in the cytoplasm of neurons (Martinez-Vicente et al., 2010).

The next step of the autophagic process is the fusion of autophagosomes with lysosomes or endosomes (to later fuse with lysosomes). Autophagosomes move along microtubules in the direction of the perinuclear microtubule-organizing center of the neuron, which is enriched in lysosomes (Lee et al., 2011). Microtubule-based vesicle transport is necessary for efficient autophagy and, when microtubules are disassembled, autophagosomes accumulate in axonal terminals where they cannot be degraded (Hollenbeck, 1993). Mutations that affect Dynein, a protein that promotes the movement on microtubules, impairs autophagosome-lysosome fusion and thus autophagic clearance of aggregate-prone proteins, causing HD in fly and mouse models (Ravikumar et al., 2005). Similarly, the dynein motor complex is mutated in the axonal Charcot-Marie-Tooth hereditary neuropathy type 2 (CMT2) and loses its function in ALS, promoting autophagosome accumulation (Ferrucci et al., 2011; Sasaki, 2011). Another protein that is necessary for autophagosome-lysosome fusion is RAB7, a late endosome protein, which influences lysosome positioning and thus the autophagosome-lysosome fusion process (Dodson et al., 2012; Uusi-Rauva et al., 2012). This protein has been found mutated in some forms of CMT2 (Spinosa et al., 2008). Additionally, mutations in the protein charged multivesicular body protein 2B (CHMP2B) affecting the activity of the endosomal sorting complex required for transport (ESCRT) have been identified in ALS and FTLD which prevent the formation of the amphisomes, the vesicles that arise from the fusion of the autophagosome with the endosome and ultimately fuse with the lysosome.

Importantly, the many of the details of the autophagosome-lysosome fusion process remain unknown. It is likely that additional proteins are involved in this step of autophagy and their mutation/deletion would likely affect autophagosome turnover.

Once autophagosomes have fused with lysosomes the process of degradation of the cytosolic cargo occurs. Various neurodegenerative diseases are characterized by reduced autophagic flux (increased numbers of autophagosomes). Inefficient autophagosomal clearance can be associated with reduced lysosomal acidification, which reduces the activity of lysosomal hydrolases and promotes the accumulation of undigested byproducts. This is typical of the so-called “lysosomal storage disorders” (LSD), which comprise more than 50 diseases and represent the most common forms of neurodegenerative disease in childhood. While the proteins and mechanisms involved in these diseases differ, impaired autophagy is unfailingly present (Menzies et al., 2015). For example, some forms of PD are characterized by lysosomal dysfunction, which is caused by loss of function mutations in the gene coding for the glucocerebrosidase. This promotes the accumulation of glucocerebroside together with α-synuclein in the brain, thus promoting neurodegeneration (Mazzulli et al., 2011; Murphy et al., 2014). In AD, however, decreased autophagosomal clearance is associated with mutations in the protein presenilin 1, which increases lysosomal pH, reduces lysosomal activity, thereby promoting amyloid deposition and the onset of the disease (Cataldo et al., 2004).

TAR DNA-binding protein (TARDBP), hereafter called TDP-43, was initially described as a transcriptional repressor factor of HIV (Ou et al., 1995). Since then, numerous studies have established that TDP-43 is a RNA binding protein that participates in multiple pathways of the “RNA metabolism,” which includes RNA splicing, RNA transport, mRNA stability, microRNA processing and stress granules assembly (Aulas et al., 2012; Buratti and Baralle, 2012; Ratti and Buratti, 2016) (Figure 3A). From a molecular point of view, the N-terminal of TDP-43 contains a nuclear localization signal (NLS), a nuclear exclusion signal (NES) and two RNA binding domains (RMM1, RMM2). The presence of NLS and NES allows the shuttling of TDP-43 between the nucleus and the cytoplasm (Ayala et al., 2008; Prpar Mihevc et al., 2016). The RRM1 is the region through which TDP-43 binds different RNAs targets (Buratti and Baralle, 2001; Ayala et al., 2011; Avendaño-Vázquez et al., 2012). Additional studies have suggested that RRM2 is required to bind the DNA and to maintain the protein structure (Kuo et al., 2009). Finally, the C-terminal region of TDP-43 is composed by a glycine rich domain (GRD), which has been involved in most of protein-protein interactions (Buratti et al., 2005; D'ambrogio et al., 2009; Buratti and Baralle, 2012). The GRD domain includes a glutamine/asparagine (Q/N) prion-like domain responsible for both, protein-protein interactions (Budini et al., 2012a) and TDP-43 aggregation (Fuentealba et al., 2010; Budini et al., 2012a,b). In addition to the Q/N region, also the N-terminal region of TDP-43 has been implicated in the aggregation process. Indeed, this domain is required to sequester the native TDP-43 into the aggregates, causing a loss-of-function (Figure 3B) (Budini et al., 2015; De Conti et al., 2015; Romano et al., 2015; Langellotti et al., 2016).

TDP-43 is the major component of inclusions or aggregates present in neuronal cells from patients affected by ALS and FTLD (Arai et al., 2006; Neumann et al., 2006). Studies have shown that more than forty disease-associated mutations have been discovered in TDP-43, principally in the C-terminal region of the protein (Buratti and Baralle, 2009; Mackenzie et al., 2010; Lattante et al., 2013). Even though experimental data suggest that some TDP-43 mutations favor its aggregation (Budini et al., 2012b), it is worth to note that around the 95% of people showing TDP-43 inclusions the protein is not mutated (Lattante et al., 2015). Hence, TDP-43 aggregation process does not necessarily require protein mutations.

As previously mentioned, TDP-43 shuttles between the nucleus and the cytoplasm (Ayala et al., 2008), however, in pathological conditions, it has been observed that TDP-43 mainly remains in the cytoplasm where it can aggregate. Indeed, affected cells can show cytoplasmic TDP-43 aggregates and complete depletion of TDP-43 in the nuclear compartment (Arai et al., 2006; Neumann et al., 2006). TDP-43 aggregates are mainly composed by TDP-43 full-length and by C-terminal protein fragments (25 and 35 kDa) that can be distinguished by ubiquitination and specific phosphorylation (Arai et al., 2006; Neumann et al., 2006; Hasegawa et al., 2008; Inukai et al., 2008).

On the other hand, mitochondrial dysfunction, observed in samples from patients with ALS, may be regulated by TDP-43. Indeed, overexpression of TDP-43 and its C-terminal fragments induced mitochondrial damage in NSC-34 motor neuron cell line (Hong et al., 2012). In addition, full-length TDP-43 and truncated TDP-43 localize in the mitochondria, suggesting that TDP-43 and its C-terminal fragments cause mitochondrial dysfunction and enhance mitophagy (Hong et al., 2012). These observations were also confirmed by Onesto et al. (2016), who reported that the overexpression of a pathological A382T TDP-43 mutant in fibroblasts, induced changes in mitochondria morphology which caused the fragmentation of the mitochondria network (Onesto et al., 2016). Thus, since mitochondria are key in neuronal metabolism, TDP-43 mutations can induce cell death by impairing not only TDP-43-dependent RNA metabolism, but also mitochondria function in ALS. In line with this, it was demonstrated that lithium prevents most ALS pathological changes in vitro an in vivo models by mechanisms that involve mitochondrial protection, induction of autophagy, mitophagy and mitochondriogenesis (Sarkar et al., 2005; Tasdemir et al., 2007; Pasquali et al., 2009; Natale et al., 2015).

Currently, the role of TDP-43 aggregates in neurodegenerative diseases is unknown and two hypotheses have been proposed (Lee et al., 2012): (1) the “gain-of-function” hypothesis, which suggests that cytoplasmic TDP-43 aggregates, or the presence of missense mutations in TDP-43, provokes cell death by cytotoxic effects (Johnson et al., 2008, 2009; Zhang et al., 2009); (2) the “loss-of-function” hypothesis, which suggests that TDP-43 aggregates act as a “sink” by additionally sequestering functional TDP-43 (nuclear and cytoplasmic). This causes a generalized impairment in TDP-43 activity that affects the whole cell physiology (Figure 3B). Despite studies strongly support the idea that the aggregation induces TDP-43 loss-of-function (Budini et al., 2015; De Conti et al., 2015; Romano et al., 2015), it is still not clear which hypothesis is correct and, it is highly possible that a combination of both mechanisms, gain- and loss-of-function, contributes to the onset and disease progression.

In addition to TDP-43, other genetic markers have been associated with the onset of ALS and FTLD and, among those, the most studied is the genetic factor C9orf72 (Lattante et al., 2015). This factor, was described in 2011 by two independent groups and consists of a polymorphism composed by different repeats of the hexanucleotide GGGGCC [(G₄C₂)n]. These repeats locate in a non-coding region of C9orf72, specifically within the intron that separates exons 1a and 1b (DeJesus-Hernandez et al., 2011; Renton et al., 2011). It has been shown that healthy people carry 20 or less G₄C₂ repeats (n ≤ 20), whereas affected individuals incorporate between 30 and 250 hexanucleotide repeats (n ≥ 30) (Renton et al., 2011). The C9orf72 gene can generate three different variants by alternative splicing of its pre-mRNA (DeJesus-Hernandez et al., 2011). By analyzing brain samples from patients, studies indicated that the pathological mechanism of (G₄C₂)n repeats may involve both, a reduction in the expression of some C9orf72 variants and/or the aberrant formation of RNA foci (DeJesus-Hernandez et al., 2011). Additionally, studies performed during the last 5 years indicated that RNA foci are formed by stable multimeric G-quadruplexes, which can abnormally sequester RNA-binding proteins (RBPs) impairing their physiological functions. In addition, RNA foci can also suffer repeat-associated non-ATG (RAN) translation. This translation occurs in sense and antisense of the (G₄C₂)n repeats, in all reading frames, and in the absence of an initiating ATG codon. Thus, this mechanism can generate polypeptides formed by different numbers of dipeptide units originated from one (G₄C₂)n repeat. These polypetides, known as DRPs (Dipeptide Repeat Proteins) or C9RANT proteins, have shown to be toxic for the cell, since they impair RNA biogenesis (Zu et al., 2013; Kwon et al., 2014; Gendron et al., 2015). Studies have evaluated the effects of proline/arginine repeated 20 times (PR20) RAN translation products in primary neuronal cultures of rat spinal cord. It has been shown that PR20 induces cell death in motor neurons without causing mitochondrial alterations, ER stress or mTOR inhibition (Gupta et al., 2017). Additionally, PR20 accumulation inhibits protein degradation pathways, such as autophagy and the ubiquitin proteasome system (Gupta et al., 2017). The pathological effects of (G₄C₂)n repeats have been recently reviewed by Todd and Petrucelli (2016).

The physiological role of C9orf72 has been poorly studied. Two different groups found that C9orf72 shares structural similarities with proteins containing DENN modules (Zhang et al., 2012; Levine et al., 2013). Proteins containing DENN modules work as GDP-GTP exchanging factors for RAB GTPases, which are involved in endosomal traffic and in the formation of autophagic vacuoles (Zerial and McBride, 2001; Amaya et al., 2015). Thus, it was proposed that C9orf72 has a role in both endosomal traffic and autophagy. Additional studies showed that C9orf72 interacts with different RAB proteins (Farg et al., 2014) involved in the autophagic process, confirming the hypothesis that C9orf72 plays an important role in the autophagic pathway (Amaya et al., 2015).

Studies have demonstrated that mutations in autophagy-related genes are associated with familial ALS and FTLD (Skibinski et al., 2005; Parkinson et al., 2006; Lee et al., 2009; Maruyama et al., 2010; Le Ber et al., 2013; Shimizu et al., 2013; Teyssou et al., 2013; Wong and Holzbaur, 2014; Cirulli et al., 2015; Krasniak and Ahmad, 2016). In fact, mutations in the ubiquitin-binding site on p62/SQSTM1 have been associated with ALS and FTLD caused by the inhibition of protein aggregates clearance mediated by autophagy (Le Ber et al., 2013; Shimizu et al., 2013; Teyssou et al., 2013). Other genes, whose mutations are associated with impaired autophagy regulation, are Optineurin (optn) and CHMP2B. OPTN, which is involved in the mitophagy machinery, is proposed to impact on ALS as its mutations promote severe mitochondria degradation (Maruyama et al., 2010; Wong and Holzbaur, 2014). CHMP2B, which is part of the ESCRTIII protein complex, is involved both in the endosomal-sorting pathway (Skibinski et al., 2005; Parkinson et al., 2006) and in the formation of autophagolysosomes (Filimonenko et al., 2007). In agreement with this, it has been observed that mutations in CHMP2B promote autophagic vesicles accumulation in brain tissues of FTLD patients (Lee et al., 2009; Krasniak and Ahmad, 2016).

Furthermore, new genetic factors that associate the autophagic machinery and neurodegenerative diseases have been identified. A recent exome sequencing assay performed on more than 2,000 ALS patients identified the TANK-Binding Kinase 1 (Tbk1) as a new gene associated with ALS (Cirulli et al., 2015). TBK1 protein belongs to the IKK-kinase family, which is involved in immune signaling pathways in neurons and other cell types (Oakes et al., 2017). In this regard, studies have suggested that TBK1 may be implicated in protein aggregations-induced neuroinflammation (Oakes et al., 2017). Other studies have shown that TBK1 phosphorylates OPTN and p62/SQSTM1 to promote autophagy, and, when non-functional, TBK1 can impair autophagy and mitophagy, affecting neuronal homeostasis (Matsumoto et al., 2015; Richter et al., 2016).

Altogether, these studies suggest that mutations in autophagy-related genes and, as consequence, impaired autophagy in the brain tissue might contribute to the onset or progression of ALS and FTLD.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.