Dysregulation of autophagy as a common mechanism in lysosomal storage diseases

Our perspective of the lysosome has shifted remarkably from its standing as a simple, terminal organelle for the degradation of cellular components to become a critical mediator of fundamental metabolic processes. Lysosomes coordinate signals from growth factors and cellular stressors and are sensitive to various metabolites, such as amino acids, glucose, lipids and cholesterol, to pivot cells between anabolic and catabolic processes, including autophagy [1–3]. The importance of the lysosome for cellular function is apparent from the large number of disorders associated with lysosomal failure: collectively known as the lysosomal storage disorders (LSDs), more than 50 inherited conditions affect lysosomal function and many are early-onset and fatal [4–7].

A key cellular homeostatic pathway implicated in several LSDs and myriad human diseases is autophagy. Autophagy is an intracellular degradation pathway essential for cellular survival and organismal health [8–10]. This process is vital for the maintenance of energy and tissue homeostasis by degrading damaged or excess intracellular components such as aggregation-prone proteins, lipids and organelles, and recycling the breakdown products [11]. There are three types of autophagy: macroautophagy (herein referred to as autophagy), microautophagy and chaperone-mediated autophagy (CMA); each type requires functional lysosomes in the final stage for degradation of the cargo [12]. Consequently, disruption of the hydrolytic functions of lysosomes impairs autophagic flux and, conversely, lysosomal function probably requires normal flux through autophagy [13]. Deregulation of autophagy is a common disease mechanism in many LSDs [14,15]. This review will describe the connections between autophagy and the LSDs, highlight common stages of autophagy disrupted in different disorders and discuss autophagy as a potential therapeutic intervention for treating certain LSDs.

Autophagy encompasses several vesicle fusion events leading to the eventual degradation of its cargo; a dynamic process termed autophagic flux (Figure 1). Multiple autophagy (Atg) genes encoding components of the autophagic machinery are required for the initiation of autophagy, which is marked by the de novo formation of double-membrane structures called phagophores. Two ubiquitin-like conjugation systems involving the Atg5–Atg12–Atg16 complex and phosphatidylethanolamine-conjugated microtubule‐associated protein 1 light chain 3 (LC3-II) are required for the initiation step [16,17]. Cytosolic components such as macromolecules (including specific substrates like p62) and organelles (including mitochondria) are sequestered in the expanding phagophores to form double-membrane vesicles called autophagosomes. LC3-II associates with autophagosome membranes throughout their lifespan and is hence used as a marker for autophagy [18]. Autophagosomes then undergo maturation into autolysosomes by one of two routes: the predominant route is a multi-step process, in which autophagosomes fuse first with late endosomes to form amphisomes, then subsequently with lysosomes; alternatively, autophagosomes can fuse directly with lysosomes [19,20]. This enables delivery of the autophagic cargo to the autolysosomes where these materials are degraded by acidic lysosomal hydrolases (Figure 1). The breakdown products are then exported via lysosomal transporters for recycling [21].

There are several molecular mediators of autophagosome maturation, including Rab7, Beclin1–Vps34–Vps15 complex and SNAREs (N-ethylmaleimide-sensitive factor-attachment protein receptors). Beclin1-interacting partners like Atg14L, Ambra1 and UVRAG promote autophagosome maturation whereas Rubicon inhibits this step [20,22,23]. Recently, gene knockout studies in mammalian cells have shown that the Atg8 family proteins (LC3 and GABARAP subfamilies) are crucial for autophagosome–lysosome fusion [24]. GABARAPs were found to preferentially recruit PLEKHM1 (Pleckstrin homology domain containing protein family member 1) [24], which associates with the homotypic fusion and protein sorting (HOPS) complex to mediate autophagosome maturation [25]. The SNARE proteins are critical for membrane tethering and fusion; for example, autophagosomal Syntaxin-17, Atg14 and SNAP-29 interact with late endosomal/lysosomal VAMP8 to mediate autophagosome maturation [26–28]. Upon autolysosome formation, the ATG conjugation system has been suggested to play a role in the degradation of the inner autophagosomal membrane [29]. The late stage of autophagy involving autophagosome maturation is primarily affected in many LSDs; the perturbations in the autophagy pathway in these conditions are described in subsequent sections (Table 1).

Transcriptional regulation of autophagy occurs via the transcription factor EB (TFEB), which drives the expression of genes related to autophagy and lysosomal biogenesis [30,31]. In turn, Ca²⁺ stored in the lysosomal lumen has been shown to modulate autophagy by promoting the nuclear localization of TFEB through the activation of calcineurin [32]. Various signalling pathways influence autophagy by acting upstream of the autophagic machinery. These include the mechanistic target of rapamycin complex 1 (mTORC1) signalling pathway (through which growth factors, nutrients and energy status influence autophagy) and mTOR-independent pathways (including inositol and IP₃, cAMP, Ca²⁺, calpain) that negatively regulate autophagy [12,33–38]. Both mTOR-dependent and mTOR-independent signalling are amenable to chemical perturbations for modulating autophagy [34,37,39–41] (Figure 1). For instance, the mTORC1 inhibitors like rapamycin and torin1, and mTOR-independent compounds including trehalose and carbamazepine can stimulate autophagy [35,42–44]. Chemical inducers of autophagy have been tested in some LSDs as a possible therapeutic intervention (Table 2) and are described below.

LSDs are caused by abnormal lysosomal function leading to accumulation of undegraded metabolites [4–7]. The composition of accumulated materials in the lysosomes varies substantially between the LSDs and, while all LSDs are inherited conditions and in many cases mutations are found in genes encoding lysosomal proteins, the types of proteins affected are also varied. As a consequence, the underlying cell biology changes occurring in the LSDs may vary, but each results (directly or indirectly) in reduced clearance of aggregates and diminished cellular homeostasis and survival. Autophagic dysregulation is commonly found in LSDs but again, a spectrum of defects is seen at various stages of the autophagic pathway in different LSDs (Table 1) [14,15]. Here, we will concentrate on some of the better-studied LSDs and detail the different autophagic defects identified.

While defective autophagy has been implicated in the pathology of many different disorders, including cancer and various cardiovascular, metabolic, pulmonary and infectious diseases [10,45], the nervous system appears to be particularly susceptible [46,47]. This is likely due to a combination of the long-lived nature of post-mitotic neurons placing particular stress on protein-clearing processes, the extreme polarization of many neurons and the high metabolic requirements of neurons leading to higher levels of oxidative damage in lysosomes via the Fenton reaction. Consistent with this, dysregulated autophagy has been identified in almost every late-onset neurodegenerative disorder [34,47,48], and similar defects are now being recognized in the neuronal ceroid lipofuscinoses (NCLs), a sub-group of the LSDs that are collectively the most common causes of childhood-onset neurodegeneration [5,49,50]. To date, 13 disease-causing genes have been identified to cause NCL pathology. Several of these genes encode lysosomal proteins, including soluble enzymes/proteins (CLN1/PPT1, CLN2/TPP1, CLN5, CLN10/CTSD and CLN13/CTSF) and membrane proteins (CLN3, CLN7/MFSD8 and CLN12/ATP13A2), but others include endoplasmic reticulum (ER) membrane proteins (CLN6 and CLN8), cytosolic proteins (CLN4/DNAJC5 and CLN14/KCTD7) and one expressed in the secretory pathway (CLN11/GRN) [51–53]. The diverse nature of these proteins, their substrates or cargos and locations within the cell are reflective of the LSDs as a whole: multiple different cell biological processes are probably affected but with each terminating in common pathology. However, as autophagic perturbations start to be identified in NCL disease models (Figure 2), the possibility that each disease hinges on defective autophagy becomes more plausible.

Several studies in NCL mouse models have indicated deregulation of autophagy in vivo. A recent study in Cln7-deficient mice reported accumulation of autophagosomes and autophagic substrates coupled with lysosomal dysfunction in the brain, suggesting that a block in autophagic flux may be occurring [54]. A similar build-up of autophagic compartments and substrates was also seen in mouse models of CLN5 and CLN6 diseases, as well as in CLN6 patient-derived fibroblasts [55–57]. The most common form of NCL is Batten disease, caused by mutations in the membrane protein, CLN3 [51]. Defective autophagosome maturation associated with increased autophagic and lysosomal compartments was observed in the Cln3-deficient mouse that models Batten disease, and also in patient fibroblasts as well as patient-specific induced pluripotent stem cell (iPSC)-derived neuronal cells [58–60]. The mechanism underpinning the changes to autophagy in CLN3-deficient cells is not fully understood, and the functions of the CLN3 protein remain unknown, but one potential mechanism may be due to altered Ca²⁺ homeostasis and deregulation of the ARF1–Cdc42 pathway identified in CLN3 mutant cells that can impact on vesicular trafficking [61,62].

Several forms of NCL are caused by mutations in lysosomal proteases, including Ppt1 (CLN1), Ppt2 (CLN2) and Cathepsin D (CLN10) [53,63]. Loss of cathepsin function commonly leads to dysregulated autophagy, presumably by reducing autophagic flux due to defective cargo clearance. Autophagosomes and mitochondrial proteins accumulate in cultured neurons of mice lacking cathepsin D or B and L [64–66], while in vivo, cathepsin D-deficient mice exhibit widespread accumulation of storage material in the lysosomes (indicating lysosomal dysfunction) and accumulation of autophagosomes. These changes likely lead to the neurodegeneration present in these mice that mimics the human disease [66]. However, a reduction in autophagic flux has been reported in CLN2 patient fibroblasts due to inhibition of autophagosome formation, which is linked to increased ROS and Akt–mTOR signalling pathway [60].

Sphingolipids are a major class of lipids enriched in the nervous system and are critical for neural development and function [67,68]. Sphingolipid turnover is therefore tightly regulated with their degradation mediated by a multi-step process requiring numerous lysosomal hydrolases [69]. The sphingolipidoses are a class of LSDs caused by deficiencies in functional hydrolases, culminating in the accumulation of wholly or partially undegraded sphingolipids [68]. Members of the sphingolipidoses include Niemann–Pick, Gaucher and Fabry diseases, mucolipidosis, and GM1/2 gangliosidoses such as Tay-Sachs and Sandhoff diseases [68]. Alterations in autophagy have been reported in some of the sphingolipidoses; the most extensively studied being Niemann–Pick type C1 (NPC1) disease.

Niemann–Pick disease is a recessively inherited neurodegenerative condition characterized by an accumulation of unesterified cholesterol deposits in various tissues, particularly the brain. The disease is subdivided into types A, B, C1 and C2. Types A and B are caused by mutations in SMPD1, leading to loss of acid sphingomyelinase activity, whereas types C1 and C2 are due to defective cholesterol transport [70]. Ninety-five per cent of cases are attributed to mutations in the lysosomal cholesterol transporter NPC1, with the lysosomal glycoprotein NPC2 that possibly aids in cholesterol trafficking affected in the remainder [71,72].

A number of studies have demonstrated a major role for autophagy in the pathology of NPC1 disease and potentially in treatment options. Various experimental platforms, including Npc1 mutant mice and disease-relevant cells (such as neurons) differentiated from human embryonic stem cells (hESCs) with NPC1 knockdown or patient-derived iPSCs, have revealed an accumulation of autophagosomes and lysosomes both in vivo and in vitro [73–80]. We previously reported that this autophagy phenotype is related to a block in autophagic flux (Figure 3). Consistent with this, we characterized a failure in the Syntaxin17/VAMP8 SNARE machinery that retards amphisome formation and significantly stalls the multi-step route of autophagosome maturation [73]. While it is possible that loss-of-function of the NPC1 protein, which normally resides on the late endosomal/lysosomal compartments, directly perturbs amphisome formation, accumulation of lysosomal cholesterol has also been reported to cause aberrant sequestration of SNAREs and prevent autophagosome maturation [81]. Another study, although consistent with dysfunctional autophagosome maturation, reported a distinct mechanism in which diminished sphingosine kinase activity and reduced levels of vascular endothelial growth factor (VEGF) lead to the accumulation of sphingosine that, in turn, impairs autophagic flux [80]. In addition to these mechanisms, depletion of lysosomal Ca²⁺ stores due to sphingosine storage was seen in NPC1 disease models, and this could possibly influence autophagy via calcineurin or calpain [32,36,82,83]. Similar to the autophagy phenotype in NPC1 disease, a block in autophagic flux has been reported in NPC2 disease as evident by the accumulation of autophagosomes and autophagic substrates as well as impaired lysosomal activity in Npc2-knockdown mouse adipocytes [84].

The reduced cell viability seen in NPC1 mouse and patient-specific iPSC models has been attributed to dysregulated autophagy [73,74,78,80]. Importantly, since autophagy regulates the clearance of lipids including cholesterol (via lipophagy), autophagy-deficient Atg5^–/– cells also exhibited accumulation of these materials similar to the cellular phenotypes observed in NPC1 disease [73,85]. Therefore, the underlying autophagy defects in NPC1 disease can further aggravate the accumulation of lipids and cholesterol, and thus, defective autophagy should be considered a major disease mechanism in Niemann–Pick disease.

Gaucher disease is the most common form of sphingolipidosis [86], and is caused by a deficiency in glucocerebrosidase (GCase) activity which catalyses the final step in glycosphingolipid degradation [87]. The loss of enzyme function in Gaucher disease directly impairs this process leading to accumulation of glucocerebroside and widespread pathology that particularly affects the liver and, in severe cases, the CNS [88]. Gaucher disease is most frequently caused by mutations in the GBA1 gene which encodes GCase [87]; however, there are rare cases of Gaucher disease caused instead by mutations in the PSAP gene [89,90]. PSAP encodes the lysosomal glycoprotein prosaposin, the precursor saposins A-D, which are essential cofactors for various sphingolipid hydrolases. Saposin C is required for GCase activity, and thus mutations in PSAP that lead to saposin C deficiency also cause Gaucher disease [89,90].

Various lines of evidence suggest defective autophagy in Gaucher disease. Impaired autophagosome maturation and down-regulation of TFEB, including a reduction in lysosomal gene expression, were found in neurons differentiated from patient-specific iPSCs [91]. A similar impairment in autophagosome degradation was also seen in primary fibroblasts deficient in saposin C; in this case associated with reduced cathepsin B/D activity [92]. Mouse models of Gaucher disease have also revealed similar autophagy phenotypes: accumulation of various autophagic cargo such as dysfunctional mitochondria, ubiquitinated protein aggregates, insoluble α-synuclein and p62, together with autophagosomes and lysosomes were found in the brain or neurons and astrocytes cultured from mice deficient for Gba, Psap or glucosylceramidase [93–96]. Recently, a Drosophila model of neuropathic Gaucher disease generated by knocking out the Gba gene revealed severe lysosomal defects, GCase accumulation and a block of autophagic flux in the brain, resulting in reduced lifespan, neurodegeneration and age-dependent locomotor deficits [97]. Together, multiple model systems highlight deregulation of autophagy in Gaucher disease.

Mucolipidosis type IV (MLIV) is a neurodegenerative condition, caused by mutations in the MCOLN1 gene that result in the loss of TRPML1 function [98,99]. TRPML1 is an inward rectifying, non-selective cation channel of the transient receptor potential family that transports divalent cations, such as Ca²⁺, Fe²⁺ and Zn²⁺, from the lysosomal intraluminal space into the cytosol [100–102]. TRPML1 and TFEB positively regulate each other and together have the potential to impact on autophagy: Ca²⁺ release by TRPML1 activates the Ca²⁺-dependent phosphatase, Calcineurin, which mediates dephosphorylation-dependent nuclear translocation of TFEB leading, in turn, to increased expression of several lysosomal and autophagy-related proteins including TRPML1 [32] (Figure 4). Indeed, enhancing TRPML1 activity either by overexpression or by pharmacological stimulation increases autophagic flux [103].

Several models of MLIV have highlighted a defect in autophagy. Accumulation of lysosomes, autophagosomes and autophagy substrates including polyubiquitinated protein aggregates was reported in patient-derived primary fibroblasts, in Mcoln1-deficient mice, and in Caenorhabditis elegans and Drosophila models [104–108]. These studies also implicated a role for wild-type TRPML1 in regulating lysosomal acidification and autophagosome maturation (alkalinization of lysosomes, for instance by inhibiting the vacuolar H⁺ATPase, inhibits hydrolase function). Furthermore, defects in CMA have been reported in MLIV, possibly through the interaction between TRPML1 with the molecular chaperones Hsc70 and Hsc40 that are required for protein translocation to the lysosome during CMA [109].

The glycogenoses are LSDs with profound autophagy defects that particularly affect skeletal and cardiac muscle [110,111]. As such, these diseases are often referred to as autophagic vacuolar myopathies (AVMs), although neuropathology is also present in some cases [112]. Diseases that lead to AVM are linked to genes involved in lysosome acidification, glycogen hydrolysis, and autophagosome maturation and fusion with lysosomes [113]. They include Pompe disease, Danon disease and X-linked myopathy with excessive autophagy (XMEA).

Pompe disease is caused by mutations in acid α-glucosidase (GAA), a lysosomal glycogen hydrolase [114]. Enzyme activity is completely absent in Pompe disease and consequently the accumulation of intra-lysosomal glycogen, as well as of autophagosomes, is a hallmark of the disease [115]. Primary myoblasts from Gaa-deficient mice have enlarged endosomes, lysosomes and autophagic vacuoles, in addition to delayed endosomal acidification and mobility [116]. These features have been attributed to a block in autophagic flux since elevated levels of the autophagosomal marker LC3-II and autophagic substrate p62 were observed in Gaa-deficient mice and in the myotubes of infantile-onset patients [117,118]. Interestingly, autophagy has been shown to play a role in the maturation of GAA [117] and the clearance of glycogen [119,120], and thus, defective autophagy is implicated as a disease mechanism in Pompe disease.

Danon disease is an X-linked disorder caused by mutations in the LAMP2 gene, which encodes a lysosomal membrane protein [121]. While LAMP2a is required for protein translocation in CMA [122], mutations in the LAMP2b isoform are associated with Danon disease [121]. Various tissues including liver, skeletal and heart muscle from Lamp2-deficient mice displayed accumulation of autophagic compartments, along with a decline in the lysosomal degradation of long-lived proteins in the hepatocytes that are unresponsive to amino acid starvation [123,124]. Muscle biopsies from patients also exhibit accumulation of LC3-II⁺ autophagic vesicles and large p62 aggregates [125]. These studies point to a likely block in flux affecting the late stage of autophagy. Consistent with this scenario, impaired autophagy associated with defective mitochondrial clearance (mitophagy) was found in cardiomyocytes differentiated from patient-specific iPSC cells with LAMP2 mutations as well as in Lamp2-deficient mice [126].

X-linked myopathy with excessive autophagy (XMEA) is characterized by progressive vacuolation and atrophy of skeletal muscle, and is caused by mutations in the VMA21 gene [127]. The VMA21 protein regulates the assembly of the vacuolar ATPase (v-ATPase) required for lysosomal acidification, and thus XMEA is associated with a reduction in the activity of lysosomal hydrolases [128]. Skeletal muscle biopsies show diminished lysosomal degradation and accumulation of autophagic compartments, suggesting a likely block in autophagic flux [128,129].

Deregulation of autophagy has been observed in myriad human diseases including many forms of cancer and neurodegeneration [10,45,47]. Despite appearing at first glance to have diverse changes in both the cell biology and the genetics underpinning them, deregulated autophagy has been reported in most (or possibly all) LSDs where investigated (Table 1) [14,15]. Potentially, dysfunctional autophagy is the common link between the LSDs but why is impaired autophagy such an important component of disease and how does it contribute to cell death? Clues have emerged from in vivo experiments aimed at depleting basal autophagy in a tissue-specific manner in mice. These result in degeneration and dysfunction of the targeted organ, indicating an essential role for autophagy in maintaining tissue homeostasis [130]. For example, brain-specific deletion of Atg5 or Atg7 in mice results in a neurodegenerative phenotype in the absence of any disease-causing, aggregation-prone proteins [131,132]. These genetic studies suggest that the decline in autophagic capability in the LSDs is likely a major contributory factor to the neuropathology that characterizes many of these disorders. Secondly, some of the non-degraded materials in LSDs such as lipids, cholesterol and glycogen are cleared by autophagy [85,119,120]. Hence, deficient autophagy will contribute to the build-up of these components with ever increasing impact on lysosomal function and autophagic flux. For example, as discussed in NPC1 disease where lipids and cholesterol are the primary non-degraded materials, impairment in autophagy will lead to a further accumulation of these macromolecules by reducing lipophagy [73]. This is because lipophagy is retarded due to loss of autophagy, as evident in autophagy-deficient mouse models [85,133]. Finally, impairment in functional autophagic flux can retard the clearance of autophagic cargo including damaged mitochondria and other undesirable materials, which leads to enhanced oxidative damage of lipids and proteins within the lysosomal membrane and a further reduction in efficiency of clearance [134–137]. Therefore, targeting this common pathway underlying several LSDs could be a promising therapeutic intervention.

Autophagy has been exploited for therapeutic benefits in diverse transgenic models of human diseases, including several neurodegenerative disorders, certain liver diseases, myopathies and infectious diseases [34,40,41,138,139]. In most of the conditions studied, induction of autophagy ameliorated the disease phenotypes and, in many cases, improved organismal longevity [140,141]. Multiple lines of evidence suggest that autophagy could also be targeted in LSDs as a treatment strategy [14,15] (Table 2). The most suitable disease contexts for this approach are likely to be LSDs where the lysosomal hydrolytic function is not overtly compromized (e.g., by direct mutation of an enzyme) so that the accumulated autophagic cargo can still be digested efficiently if autophagic flux is stimulated. Some of the mechanisms of how autophagy might be beneficial in LSDs are discussed below.

The dysregulated autophagy underpinning many of the LSDs occurs primarily due to failure of autophagosome maturation, which stalls autophagic flux (Table 1). This phenotype probably arises due to a block in the multi-step route of autophagosome maturation. However, this block may not be absolute because we have recently characterized a bypass mechanism in NPC1 disease models which restores functional autophagic flux [73]. We have shown that induction of autophagy facilitates direct autophagosome–lysosome fusion without requiring the formation of amphisomes in the multi-step route (Figure 3). Restoration of autophagic flux enables cargo clearance, which correlates with improved cell viability in primary mouse neurons with Npc1 knockdown and in NPC1 patient iPSC-derived neuronal and hepatic cells [73,74]. Interestingly, we observed cell-type specificity of small molecule autophagy enhancers in these patient-derived cells: rapamycin and carbamazepine rescued cell death and defective autophagy in both iPSC-derived neuronal and hepatic cells but other autophagy inducers including trehalose, verapamil, BRD2716, BRD5631 and BRD34009 were effective only in neurons [74,142]. Strikingly, enhanced autophagic flux did not reduce the cholesterol load in NPC1 mutant cells, meaning a combination strategy coupling autophagy induction with low doses of a cholesterol-depletion agent such as 2-hydroxypropyl-β-cyclodexterin (HPβCD) may be especially effective for NPC1 disease [73,74,143].

Treatment with autophagy inducers has also been found to be beneficial in other LSDs, including CLN3 disease (Batten disease). Administration of trehalose, an mTOR-independent autophagy inducer [44], in a mouse model of CLN3 disease promoted cellular clearance of undegraded materials, attenuated the neuropathology and extended lifespan [144]. Interestingly, although the autophagy-inducing property of trehalose has been recently attributed to the inhibition of the SLC2A family of glucose transporters [145], trehalose was found to activate TFEB by inhibiting Akt independently of mTORC1 [144]. Furthermore, inositol monophosphatase (IMPase) inhibitors such as lithium and L-690,330, which induce mTOR-independent autophagy by reducing inositol and IP₃ levels [35], rescued autophagic flux and improved cell viability in mutant Cln3 cerebellar cells [146]. However, further studies are required to determine whether the bypass mechanism of restoring autophagic flux is mediating the beneficial effects of autophagy induction in these contexts.

Induction of autophagy via the classical mTORC1 pathway also had some beneficial effects in LSD models. Although TORC1 activity is dysregulated in a Drosophila model of Gaucher disease [97], the mTORC1 inhibitor rapamycin [42] partially corrects the shortened lifespan and locomotor defects [97], whereas reactivation of dysregulated mTORC1 by knockdown of the Rheb inhibitor, TSC2, in a mouse model of Pompe disease rescued the muscular atrophy and autophagy defects [147]. Rapamycin, while commonly used in research contexts, particular in cell-based assays in vitro, is unlikely to be useful clinically in all LSDs, in part because its efficacy appears to be context-dependent in vivo and subject to feedback loops. Even in vitro rapamycin has differential effects. For example, it reduces mitochondrial ROS and lipid droplets and induces lysosomal exocytosis in Niemann–Pick type B lymphocytes [148], and improves autophagic flux and GAA maturation in Pompe disease patient myotubes [117], but is toxic in neuronal cells differentiated from Gaucher disease patient-derived iPSCs [91]. Since mTOR governs critical cellular function like cell growth and translation [38], the use of mTOR inhibitors for induction of autophagy can produce potential side-effects [149]. Consequently, the mTOR-independent autophagy inducers are considered more desirable for clinical applications in patients [34,40,41,138].

Genetic up-regulation of autophagy via overexpression of TFEB has shown potential as a therapeutic strategy. Overexpression of TFEB enhances the clearance of proteinaceous aggregates in transgenic models of late-onset neurodegeneration disorders [150–152], and improves autophagy and lysosomal defects while rescuing disease severity in some LSDs [153,154]. Apart from mediating the transcription of genes required for autophagy and lysosomal biogenesis [30,31], TFEB also facilitates lysosomal exocytosis via the recruitment and fusion of the lysosomes to the plasma membrane by increasing local Ca²⁺ concentrations through its target, the TRPML1/MCOLN1 cation channel (Figure 4) [32,155]. This mechanism may underpin some of the beneficial effects of overexpression because enhanced secretion may enable the clearance of pathologically enlarged lysosomes and undegraded materials associated with LSDs that have an impact on autophagic flux.

Experimentally, overexpression of TFEB reduces glycogen load and the size of lysosomes, and promotes autophagosome maturation and lysosomal exocytosis in immortalized myogenic cells from Pompe disease patients [156] and rescues tissue pathology in vivo in autophagy-dependent manner [155,156]. In addition, overexpression of TFEB together with GAA cooperatively facilitated glycogen clearance and improved pathology in skeletal muscles differentiated from Pompe disease patient-derived iPSCs [157]. Similarly, lysosomal exocytosis and reduction in cellular vacuolization occur after TFEB overexpression in various cell and mouse models of LSDs, including multiple sulfatase deficiency (MSD) and mucopolysaccharidosis type IIIA (MPS-IIIA) where a block in autophagy was previously reported [155,158]. Overexpression of TFEB also causes reduction in substrate accumulation in various cell models of Pompe disease, MSD, MPS-II, MPS-IIIA and Batten disease [150,155,156], although it did not rescue the lysosomal depletion phenotype in neuronal cells differentiated from Gaucher disease patient-derived iPSCs [91]. However, TFEB did enhance lysosomal biogenesis in the presence of recombinant GCase in these cells [91]. Taken together, these studies highlight TFEB as a promising therapeutic target with the potential to be manipulated in multiple LSDs.

Enzyme replacement therapy (ERT) to replace the absent or non-functional hydrolytic enzyme is approved for patients with Gaucher, Pompe and Fabry diseases and some mucopolysaccharidoses, including MPS type I [159,160]. For Gaucher disease, ERT for the GCase enzyme is a relatively mature therapy with several drugs on the market, including imiglucerase, velaglucerase alfa and taliglucerase alfa [161–163], along with two substrate reduction therapies, miglustat and eliglustat [164,165]. Recent data demonstrate that ERT for the GAA enzyme deficient in Pompe disease improves muscle mass and autophagic flux in muscle biopsies of some patients after treatment [117,166]. As with the Gaucher disease experiments described above [91], combination strategies that include induction of autophagy may prove more effective in Pompe disease and similar LSDs than ERT alone, in this case possibly by stimulating clearance of the accumulating glycogen and facilitating GAA maturation [117,157]. Indeed, experiments with a Pompe mouse model in vivo demonstrated increasing autophagic flux with rapamycin or its analogue CCI-779 alongside ERT for the GAA enzyme reduced muscle glycogen levels compared with either treatment alone [167]. These emerging data highlight the potential benefits of combinatorial treatment strategies in the LSDs.

The LSDs are characterized by accumulation of undegraded macromolecules due to defective or absent lysosomal hydrolases, or because of dysregulated endosomal–lysosomal processes, including altered vesicular trafficking, autophagosome maturation, lysosomal acidification or transport of molecules across the lysosomal membrane. The separate classes of LSDs represent several different defects in lysosomal biology and to some extent require tailored therapies to overcome them. However, one feature apparent in many of the LSDs is dysregulated autophagy (Table 1), and this may be the common therapeutic target that can be exploited similarly in different disorders. Based on the common blockage of autophagic flux, therapies designed to stimulate autophagy and alleviate the block are a logical option to trial. While autophagy inducers have already shown benefit in a few LSDs (Table 2), a broader evaluation of autophagy modulation in the LSDs is urgently required. Autophagy induction shows particular promise in combination with existing therapeutic options such as ERT. However, ERT is applicable to only a subset of the LSDs and the treatment costs are high, and therefore, identifying novel therapeutic interventions for the LSDs that are cheaper and more widely useful is essential. Ultimately, different combinations of therapies are likely to be required for each of the LSDs but autophagy regulators may play a significant role in many.

• Autophagy is a vital cellular process requiring the degradative function of lysosomes.

• Abnormal lysosomal function leading to accumulation of undegraded metabolites occurs in the LSDs.

• Defects in autophagy are emerging to be a common disease mechanism underlying LSDs.

• Stimulation of autophagy is a potential therapeutic intervention in LSDs.

The authors thank the funding agencies for supporting our research. S.S. is funded by Wellcome Trust Seed Award [109626/Z/15/Z], UKIERI (UK-India Education and Research Initiative) DST Thematic Partnership Award [2016-17-0087] and Birmingham Fellowship. R.I.T. is funded by a New Investigator Award from the Biotechnology and Biological Sciences Research Council [BBSRC; BB/N008472/1]. T.B. is funded by the Medical Research Council (MRC) Biomedical Catalyst: Developmental Pathway Funding Scheme (DPFS) Award [MR/P007732/1]. T.R.R. is funded by Fundação de Amparo à Pesquisa do Estado de São Paulo [FAPESP; 2015/02041-1], Fundação de Apoio à Pesquisa da Faculdade de Ciências Médicas da Santa Casa de São Paulo (FAP FCMSCSP) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)/Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES). S.S. is also a Former Fellow for life at Hughes Hall, University of Cambridge, U.K.

The authors declare that there are no competing interests associated with the manuscript.

ERT

enzyme replacement therapy

GAA

acid α-glucosidase

GCase

glucocerebrosidase

iPSC

induced pluripotent stem cells

LAMP2

lysosome-associated membrane protein 2

LSD

Lysosomal storage disorder

MLIV

mucolipidosis type IV

MPS

mucopolysaccharidosis

MSD

multiple sulfatase deficiency

mTORC1

mechanistic target of rapamycin complex 1

NCL

neuronal ceroid lipofuscinosis

NPC1

Niemann–Pick type C1 disease

ROS

reactive oxygen species

SNARE

N-ethylmaleimide-sensitive factor-attachment protein receptors

TFEB

transcription factor EB

XMEA

X-linked myopathy with excessive autophagy