S. Sarkar, Department of Medical Genetics, University of Cambridge, Cambridge Institute for Medical Research, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 0XY, UK Fax: +44 1223 331206 Tel: +44 1223 331139 E-mail: firstname.lastname@example.org D. C. Rubinsztein, Department of Medical Genetics, University of Cambridge, Cambridge Institute for Medical Research, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 0XY, UK Fax: +44 1223 331206 Tel: +44 1223 762608 E-mail: email@example.com
Autophagy is a nonspecific bulk degradation pathway for long-lived cytoplasmic proteins, protein complexes, or damaged organelles. This process is also a major degradation pathway for many aggregate-prone, disease-causing proteins associated with neurodegenerative disorders, such as mutant huntingtin in Huntington’s disease. In this review, we discuss factors regulating the degradation of mutant huntingtin by autophagy. We also report the growing list of new drugs/pathways that upregulate autophagy to enhance the clearance of this mutant protein, as autophagy upregulation may be a tractable strategy for the treatment of Huntington’s disease.
Degradation of cellular proteins occurs by two pathways. The proteasomes predominantly degrade short-lived nuclear and cytosolic proteins. These substrates are generally selected for degradation after they are tagged with polyubiquitin chains. The narrow pore of the proteasome precludes entry of protein complexes and organelles. The bulk degradation of cytoplasmic proteins or organelles is mediated largely by macroautophagy, generally referred to as autophagy . Autophagy substrates generally have long half-lives and can include protein complexes or damaged cellular organelles. This process involves the formation of small double-membrane structures of unknown origin(s) called phagophores, which elongate to form autophagosomes. Autophagosomes ultimately fuse with mammalian lysosomes (or yeast vacuoles) to form autolysosomes, where their contents are degraded by acidic lysosomal hydrolases  (Fig. 1).
During autophagosome formation, the elongation of the phagophore involves a ubiquitin-like conjugation system, in which mammalian Atg12 is conjugated to Atg5. The Atg12–Atg5 conjugate then forms a complex with Atg16L. This complex associates with the isolation membrane for the duration of autophagosome formation, but dissociates upon its completion  (Fig. 1). The function of the Atg12 system is closely linked to another ubiquitin-like system involving microtubule-associated protein 1 light chain 3 (LC3), which is the mammalian ortholog of yeast Atg8 and the only known mammalian protein that specifically associates with the autophagosome membrane . LC3 is cleaved to form cytosolic LC3-I. After autophagy induction, LC3-I is conjugated with phosphatidylethanolamine, resulting in the LC3-II species that associates with autophagosomes . The membrane targeting of LC3 depends on Atg5 .
The formation of autophagosome precursors is prevented by 3-methyladenine (3-MA) or wortmanin, which are inhibitors of phosphatidylinositol-3-kinases, and class III phosphatidylinositol-3-kinase is required for autophagy [5–7] (Fig. 1). Autophagy is negatively regulated by the mammalian target of rapamycin (mTOR). Inhibition of mTOR by rapamycin induces autophagy, but its mechanism of action in mammalian cells is still unknown . At a physiological level, autophagy is induced by amino acid deprivation .
Autophagy regulates the clearance of aggregate-prone disease-causing proteins associated with various neurodegenerative disorders, such as mutant huntingtin [causing Huntington’s disease (HD)], ataxin-3 (causing spinocerebellar ataxia 3), forms of tau (causing frontotemporal dementias), the A53T and A30P α-synuclein mutants (causing familial Parkinson’s disease), and mutant forms of superoxide dismutase 1 [causing familial amyotrophic lateral sclerosis (ALS)] [10–14]. Two recent landmark studies highlighted the strong link between autophagy and neurodegeneration, where loss of basal autophagy in mouse neuronal cells mediated by knockdown of the essential autophagy genes, Atg5 or Atg7, resulted in progressive motor deficits, cytoplasmic aggregates, and neurodegeneration [15,16].
Autophagy in HD
HD is a progressive, autosomal dominant, neurodegenerative disorder caused by the expansion of a CAG trinucleotide repeat (> 35 repeats) in the huntingtin gene, which is translated into an expanded polyglutamine tract in the N-terminus of the huntingtin protein. Mutant huntingtin toxicity is believed to be expressed after it is cleaved to form N-terminal fragments comprising the first 100–150 residues with the expanded polyglutamine tract, which are also the toxic species found in aggregates (also called as inclusions) . Although the polyglutamine disorders are associated with intraneuronal aggregates, it is debatable whether the aggregates are toxic or protective [18,19]. Recent studies and reviews have implicated the preaggregate oligomers as the most toxic species in neurodegenerative diseases [20–25]. However, induction of autophagy results in decreases of both aggregated and soluble ‘monomeric’ huntingtin species, and results in decreased toxicity in cell, fly and mouse models of HD . Phosphorylation of various mutant proteins, such as huntingtin, ataxin-1, and ataxin-3, may regulate neurodegeneration in these disease conditions [27–32], but does not primarily influence the process of autophagy, as far as we are aware. However, hyperphosphorylation of tau, causing neurofibrillary tangles in Alzheimer’s disease (AD) , may influence its location, dependence on autophagy, and accessibility to autophagy.
Increased autophagy has been reported in HD. Mouse clonal striatal cells transiently transfected with truncated and full-length human wild-type and mutant huntingtin show the presence of both normal and mutant proteins in dispersed and perinuclear vacuoles . Furthermore, huntingtin-labeled vacuoles display the ultrastructural features of early and late autophagosomes, and huntingtin-enriched cytoplasmic vacuoles appear to be more abundant in cells expressing mutant huntingtin . Similar features have been seen in brains from HD patients and transgenic mice, where there are excessive endosomal–lysosomal-like organelles, tubulovesicular structures, and multiple vesicular bodies [36,37]. Increased autophagosome–lysosomal bodies have also been found in primary striatal neurons from HD mice expressing truncated mutant huntingtin following dopamine-stimulated oxidative stress . Moreover, increased numbers of autophagosomes have been found in lymphoblasts of HD patients as compared to the control lymphoblasts .
Degradation of mutant huntingtin by autophagy
Previous work from our laboratory demonstrated that mutant huntingtin is an autophagy substrate . Inhibition of autophagy at the level of autophagosome formation by 3-MA , or at the level of autophagosome–lysosome fusion using bafilomycin A1 , slowed mutant huntingtin clearance and increased the levels of soluble and aggregated mutant huntingtin in HD cell models . Furthermore, rapamycin treatment increased mutant huntingtin clearance and decreased the levels of soluble proteins and aggregates  (Fig. 2). Yuan and colleagues have demonstrated that autophagy clears full-length mutant huntingtin . No discernible perturbation of wild-type huntingtin clearance was seen with these autophagy modulators [11,42]. These data suggest that the aggregate-prone mutant form of huntingtin, unlike the wild-type huntingtin, is strongly dependent on autophagy for its clearance.
Interestingly, we found that mTOR was sequestered in mutant huntingtin aggregates in HD cell models, transgenic mice, and patients’ brain. This sequestration impaired mTOR kinase activity, thereby inducing autophagy. Therefore, this study identified a new protective role for mutant huntingtin aggregates in inducing autophagy for their self-destruction by enhancing the clearance of the mutant protein . A recent study has shown that expanded polyglutamine with 72 repeats induced autophagy dependent on eukaryotic translation initiation factor 2α, and this protected against polyglutamine-induced endoplasmic reticulum stress-mediated cell death .
Inducing autophagy for enhancement of mutant huntingtin clearance
Autophagy upregulation may be a therapeutic strategy for HD and related conditions, where the mutant aggregate-prone proteins are autophagy substrates  (Fig. 2). The autophagic clearance of mutant huntingtin aggregates is likely to be a consequence of degrading the aggregate precursors (soluble and oligomeric species), rather than large aggregates that are much larger than typical autophagosomes [8,12]. In this review, we will restrict our discussion to studies investigating modulation of autophagy for mutant huntingtin degradation.
Inducing autophagy by mTOR inhibition
In addition to showing that rapamycin or its analog CCI-779 was protective in cells, Drosophila and mouse models of HD, it was also shown that raised intracellular glucose or glucose 6-phosphate induced autophagy by mTOR inhibition, thereby reducing mutant huntingtin aggregates/toxicity in HD cell models [11,12,44]. The mechanism by which mTOR regulates autophagy remains unclear, and this kinase controls several cellular processes besides autophagy, probably contributing to the complications seen with its long-term use over many months. mTOR is an important signaling molecule that regulates diverse cellular functions, such as initiation of mRNA translation, ribosome biogenesis, transcription, cell growth, and cytoskeletal reorganization . Inhibition of mTOR by rapamycin causes cell cycle arrest and leads to poor wound healing and mouth ulcers . Thus, compounds that induce autophagy by mTOR-independent mechanisms may be more suitable for the treatment of such neurodegenerative disorders, which may require drugs to be taken for decades.
Inositol-lowering agents trigger autophagy independently of mTOR
We previously showed that lithium induced autophagy by inhibiting inositol monophosphatase (IMPase; an intracellular target of lithium), leading to free inositol depletion, which, in turn, decreased inositol 1,4,5-trisphosphate (IP3) levels [47,48] (Fig. 2). This effect on autophagy was mimicked by a specific IMPase inhibitor, L-690,330. Induction of autophagy by these agents reduced the proportion of cells with mutant huntingtin aggregates and enhanced the clearance of soluble aggregate-prone proteins. Mood-stabilizing drugs such as carbamazepine and valproic acid, which deplete inositol levels, also enhanced the clearance of mutant proteins (Fig. 2). The autophagy-enhancing effect of lithium was most likely to be mediated at the level of, or downstream of, lowered IP3, as it was abrogated by pharmacological treatments that increased the level of IP3. Induction of autophagy by IMPase inhibition was mTOR-independent. Moreover, IP3 levels had no effect on the autophagy-inducing property of mTOR inhibition by rapamycin, suggesting that these two pathways are independent of each other . Therefore, agents that reduce inositol or IP3 levels may be possible therapeutic candidates where autophagy is a protective pathway.
The autophagy-inducing property of lithium has recently been suggested to contribute to its protective effects in ALS patients and mouse models, where the drug treatment increased survival and delayed disease progression . Remarkably, all the ALS patients on lithium treatment for 15 months survived, whereas approximately 30% of control patients matched for age, disease duration and sex receiving riluzole died . However, lithium may also be mediating its effects via autophagy-independent pathways.
Combination treatment with lithium and rapamycin has additive effects on autophagy
Although we demonstrated that lithium induced mTOR-independent autophagy by inhibiting IMPase , we have recently shown that glycogen synthase kinase-3β (GSK-3β), another intracellular target of lithium, has opposing effects on autophagy in an mTOR-dependent fashion  (Fig. 2). Inhibition of GSK-3β by SB216763 inhibited autophagy and resulted in increased mutant huntingtin aggregation; an effect that was also observed in GSK-3β knockout mouse embryonic fibroblasts. This effect was independent of the GSK-3β target, β-catenin. Indeed, inhibition of GSK-3β activated mTOR by phosphorylating the tuberous sclerosis complex protein TSC2 , which impaired autophagy. However, lithium or IMPase inhibitor (L-690,330) reduced the proportion of cells with mutant huntingtin aggregates even in GSK-3β null cells, suggesting that induction of autophagy by lithium due to IMPase inhibition occurred even in the absence of GSK-3β .
In order to counteract the autophagy inhibitory effects of mTOR activation resulting from lithium treatment due to GSK-3β inhibition, we used the mTOR inhibitor rapamycin in combination with lithium. This combination enhances autophagy by mTOR-independent (IMPase inhibition by lithium) and mTOR-dependent (mTOR inhibition by rapamycin) pathways [47,49] (Fig. 2). Combination treatment with lithium and rapamycin had additive protective effects on the autophagic clearance of mutant huntingtin, as compared to either drug alone. We have further demonstrated proof-of-principle for this rational combination treatment approach in vivo by showing greater protection against neurodegeneration in an HD Drosophila model with TOR inhibition and lithium, as compared to inhibition of either pathway alone [47,49]. Furthermore, this approach may also benefit from the cytoprotective effects of GSK-3β inhibition, due to activation of the β-catenin–Tcf pathway (Fig. 2). Although treatment with lithium on its own is also likely to mediate antiapoptotic effects in HD models [51,52], the autophagy-inhibitory effect of GSK-3β may explain the previous equivocal effects of lithium in an HD mouse model .
The rational combination treatment of HD or related disorders may be beneficial where the mutant aggregate-prone proteins are autophagy substrates. Combination therapy with more moderate IMPase and mTOR inhibition may also be safer for long-term treatment than using doses of either inhibitor that result in more severe perturbations of a single pathway. This alternative strategy may help to lessen the drug-specific side-effects.
GSK-3β is also known to hyperphosphorylate tau, and inhibitors of GSK-3β such as lithium may be used for preventing accumulation of hyperphosphorylated tau in AD [33,54]. Furthermore, GSK-3α has been shown to facilitate amyloid precursor protein processing at the γ-secretase step and thereby regulate amyloid-β (Aβ) production . Lithium reduced Aβ production by inhibiting GSK-3α . Thus, GSK-3 inhibition by lithium may be a tractable therapeutic strategy in AD, as it reduces the formation of both neurofibrillary tangles and amyloid plaques. Furthermore, lithium may also potentially enhance autophagic clearance of mutant tau, as autophagy induction with rapamycin has this effect .
Trehalose induces mTOR-independent autophagy
Trehalose, a disaccharide present in many nonmammalian species, functions as a chemical chaperone and protects cells against various environmental stresses by preventing protein denaturation . Trehalose has been shown to alleviate polyglutamine-induced pathology in an HD mouse model, and this protective effect was suggested to be mediated by trehalose binding to the expanded polyglutamines, thus stabilizing the partially unfolded mutant protein . We have recently reported a novel function of trehalose in inducing autophagy independently of mTOR  (Fig. 2). Trehalose increased autophagic flux in various cell lines, thereby enhancing the clearance of mutant huntingtin and α-synuclein mutants and reducing the toxicity of these mutant proteins. Furthermore, trehalose facilitated the clearance of endogenous autophagy substrates as assessed by reduced mitochondrial load, and this protected cells against proapoptotic insults by decreasing active caspase-3 levels . The dual protective properties of trehalose (‘autophagy induction’ for enhancing clearance and ‘chemical chaperone’ for inhibiting aggregation), coupled with its lack of toxicity, suggest that it may be a valuable drug for further development.
Screens for autophagy modulators
In order to identify further autophagy modulators, we recently carried out a primary small-molecule screen in yeast in collaboration with Schreiber and colleagues . First, novel small-molecule enhancers (SMERs) and small-molecule inhibitors of the cytostatic effects of rapamycin were identified in a yeast screen with 50 729 compounds. Three SMERs induced mTOR-independent autophagy in the absence of rapamycin, thereby enhancing the clearance of mutant huntingtin and A53T α-synuclein in mammalian cells, and attenuated mutant huntingtin fragment toxicity in HD cells and Drosophila models . These three SMERs also had additive effects with rapamycin, and the combined treatment facilitated greater clearance of mutant proteins than either of the treatments alone (Fig. 2). A further screen of structural analogs of these three SMERs identified 18 additional candidate drugs that reduced the proportion of cells with mutant huntingtin aggregates .
Yuan and colleagues recently performed an image-based screen for autophagy inducers by analyzing 480 bioactive compounds in a stable human glioblastoma cell line expressing green fluorescent protein (GFP)–LC3 . Analysis of autophagy was performed by using GFP–LC3 punctate structures with high-throughput fluorescence microscopy, and the screen hits were classified into three groups depending on the number, size and intensity of the GFP–LC3 vesicles. Further analysis of the hits was carried out, from which eight compounds were identified that induced autophagic degradation without notable cellular damage. These compounds are fluspirilene, trifluoperazine, pimozide, niguldipine, nicardipine, amiodarone, loperamide, and penitrem A, which did not affect mTOR activity and reduced the numbers of expanded polyglutamine aggregates in a cell-based assay with the exception of nicardipine. Some of these new targets may be beneficial for the treatment of HD, as seven out of the eight final hits were FDA-approved drugs .
Autophagy is a major degradation route for mutant huntingtin and other aggregate-prone proteins associated with neurodegenerative disorders. Furthermore, autophagy induction may also be a valuable strategy in the treatment of infectious diseases, including tuberculosis . Since the first discovery of autophagic clearance of mutant huntingtin by rapamycin was reported , studies have identified novel autophagy-inducing pathways/drugs. Although various small molecules have been identified since then, the key question now is to understand their targets regulating mammalian autophagy. This remains a daunting task, as it is still unclear how mTOR regulates autophagy.
We are grateful to the Wellcome Trust, Medical Research Council (MRC), EUROSCA and the National Institute for Health Research, Biomedical Research Centre at Addenbrooke's Hospital for funding.