Protein and organelle turnover is essential to maintain cellular homeostasis and survival. Removing and recycling cell constituents is achieved by autophagy in all cells, including neurons. Autophagy contributes to various physiological processes, such as intracellular cleansing, cellular homeostasis, development, differentiation, longevity, tumor suppression, elimination of invading pathogens, antigen transport to the innate and adaptative immune systems, and counteracting endoplasmic reticulum (ER) stress and diseases characterized by the accumulation of protein aggregates . Autophagy plays a role in a number of infectious and inflammatory diseases, in addition to protein unfolding and misfolding diseases that lead to neuron, muscle and liver degeneration or heart failure [2-4]. Whereas autophagy has long been defined as a form of non-apoptotic, programmed cell death , recent findings suggest that autophagy functions primarily to sustain cells, and only defects in autophagy lead to cell death .
2. Autophagy in neuronal physiology
Autophagy was initially identified and characterized in a few cell types including neurons. The distinct vacuoles which feature this self-eating process were originally described at the ultrastructural level [7, 8]. The formation of autophagosomes was associated with chromatolysis of a restricted neuroplasm area, free of organelles, but filled with various types of vesicles . The function of autophagy in mature neurons , however, is still debated. In comparison with other organs, rodent brains show high expression levels of autophagy-related (Atg) proteins and low levels of autophagy markers such as autophagosome number and LC3-II. Indeed, even under prolonged fasting conditions, the number of autophagosomes does not increase in neurons, probably because their nutrient supply from peripheral organs is maintained . However, mice with CNS defects in their autophagic machinery exhibit neurological deficits, such as abnormal limb-clasping reflexes, locomotor ataxia, and lack of motor coordination, in addition to a significant loss of large pyramidal neurons in the cerebral cortex and Purkinje cells (PCs) in the cerebellar cortex [11-13].
Macroautophagy (hereafter referred to as autophagy) is initiated when a portion of the cytoplasm is sequestrated within a double-membrane organelle, the so-called autophagosome . The autophagic machinery has been extensively detailed at the molecular level in a number of reviews including several chapters of this book [14-17]. Atg and several non-Atg proteins have been identified as regulators of key steps leading to the degradation of cytosolic components in lysosomes: initiation and nucleation of phagophores, expansion of autophagosomes, maturation of autophagosomes into amphisomes/autolysosomes, and execution of autophagic degradation . The endosomal sorting complex required for transport (ESCRT) pathway functions in the sorting of transmembrane proteins into the inner vesicles of multivesicular bodies (MVB) during endocytosis. Also it is conceivably an essential part of the basal autophagy process in neurons because ubiquitin- or p62/SQSTM1 (p62)-labelled inclusions and autophagosomes accumulate in neurons deficient in ESCRT components . Increasing evidence suggests that autophagy is regulated in a cell type-specific manner and as such autophagy may serve a distinct function in neurons and may show difference in the molecular machinery underlying basal autophagy (Fig. 1).
2.1. Axonal autophagy
In neurons, autophagy occurs in axons, suggesting that it may be uniquely regulated in this compartment and specifically adapted to local axonal physiology . In primary dorsal root ganglion neurons, autophagosomes initiate distally in nerve terminals and mature during their transport toward the cell soma [21, 22]. In non-neuronal cells, the autophagosomal membrane has multiple possible origins, including endocytosed plasma membrane (amphisome), ER, mitochondria, and trans-Golgi membranes [18, 21, 23- 32]. In contrast, the origin of autophagosomes in the axons is likely to be restricted to the sources of membrane available in the terminals such as smooth ER and plasma membrane [33, 34], excluding rough ER and Golgi dictyosomes.
As observed in ultrastructural studies of axotomized neurons [9, 35], analysis of Purkinje cell (PC) degeneration in lurcher mutant GluRδ2Lc  demonstrates autophagosomes in their axonal compartment. In this study, an excitotoxic insult mediated by GluRδ2Lc triggered a rapid and robust accumulation of autophagosomes in dystrophic axonal swellings providing evidence that autophagy is induced in dystrophic terminals and that autophagosome biogenesis occurs in axons . The molecular scenario underlying the initiation of this axonal autophagy is unclear. Liang et al.  suggested that autophagy in lurcher PCs could be directly activated by an interaction between the postsynaptic GluRδ2Lc, nPIST and beclin 1 an important regulator of autophagy. Nevertheless, how activation of this postsynaptic signaling pathway in dendrites initiates autophagosome formation in axon compartments is uncertain. PC death that is correlated with early signs of autophagy appears to be independent of depolarization in the heteroallelic mutant Lurcher/hotfoot bearing only one copy of the lurcher allele and no wild-type GRID2 . However, in the lurcher mutant bearing only one copy of the lurcher allele and one copy of the wild-type GRID2 allele, the leaky channel of GluRδ2Lc depolarizes the neuron and this could transduce an electrical signal to the distal ends causing rapid physiological changes within axons. This effect combined with the local changes in postsynaptic signaling in dendrites may promote autophagosome biogenesis [37, 40].
2.2. Microtubule-dependent dynamics of neuronal autophagy
Previous data indicate that autophagy is a microtubule-dependent process. In cultured sympathetic neurons, autophagosomes formed in the distal ends of axon undergo retrograde transport along microtubules to the cell body where lysosomes that are necessary for the degradation step of autophagy are usually located . Consistent with these observations, prominent retrograde transport of GFP-LC3-labelled autophagosomes has been observed in the axons of primary cerebellar granule cells . In serum-deprived PC12 cells, autolysosomes formed by fusion of autophagosomes with lysosomes move in both anterograde and retrograde directions in neurites, and this trafficking requires microtubules . Furthermore, both pharmacological and siRNA-based inhibition of directional microtubule motor proteins kinesin and dynein partially block respectively, anterograde and retrograde neuritic transport of autophagosomes, indicating that they participate in this transport. Recent observations in primary dorsal root ganglionic neurons support a maturation model in which autophagosomes initiate distally, engulfing mitochondria and ubiquitinated cargo, and move bidirectionally along microtubules driven by bound anterograde kinesin and retrograde dynein motors . Fusion with late endosomes or lysosomes may then allow autophagosomes to escape from the early distal pool by robust retrograde dynein-driven motility. The involvement of the dynein-dynactin complex in the movement of autophagosomes along microtubules to lysosomes has also been demonstrated in non-neuronal cells . Consistent with the formation of an autolysosomal compartment, autophagosomes increasingly acidify as they approach the cell soma, thereby fueling the catalysis of the degradation of their engulfed contents. Fully acidified autolysosomes undergo bidirectional motility suggesting reactivation of kinesin motors [22, 41].
The interaction of the autophagic membrane marker Atg8/LC3 with the microtubule-associated protein 1B (MAP1B)  implicates microtubule-dependent, axon-specific regulation of autophagosomes. Overexpression of MAP1B in non-neuronal cells reduces the number of LC3-associated autophagosomes without impairing autophagic degradation. The scarcity of LC3-labelled autophagosomes in CNS neurons under normal conditions may be explained by their high expression levels of MAP1B [10, 36]. By modifying microtubule function, the LC3-MAP1B interaction has been proposed to accelerate the delivery of LC3-autophagosomes to lysosomes, thereby promoting efficient autophagic turnover . The exact mechanism underlying the involvement of microtubule in autophagosome formation, as well as targeting and fusion with lysosomes is open to debate [44, 45]. Based on (i) the absence of obvious changes in LC3 autophagosomes when they are associated with phosphorylated MAP1B-P, (ii) the elevated level of MAP1B-P bound to LC3 in dystrophic terminals containing a large number of autophagosomes  and, (iii) the conserved role of MAP1B-P in axonal growth and repair during development or injury (which implicates autophagy in remodelling axonal terminals during regeneration) , the interactions of LC3 with MAP1B and MAP1B-P have been proposed to represent a regulatory determinant of autophagy in axons under normal and pathological conditions respectively .
2.3. Functions of neuronal autophagy
Neurons, as non-dividing cells, are more sensitive to toxic components than dividing cells. Therefore, their survival and the maintenance of their specialized functions under physiological and pathological conditions is crucial requiring a tight quality control of cytoplasmic components and their degradation. Autophagy is believed to be of particular importance in the synaptic compartments of neurons where high energy requirements and protein turnover are necessary to sustain synaptic growth and activity. The CNS displays relatively low levels of autophagosomes under normal conditions, even after starvation, but requires an indispensable turnover of cytosolic contents by autophagy even in the absence of any disease-associated mutant proteins [10, 47, 48]. The scarcity of immature autophagosomes in neurons is likely to reflect a highly efficient autophagic degradation in the healthy brain. Accordingly, inhibition of autophagy causes neurodegeneration in mature neurons suggesting that autophagy may regulate neuronal homeostasis [11, 12]. For example, abnormal protein accumulation and eventual neurodegeneration are observed in the CNS of mice lacking the atg5 or the atg7 genes. This implies that basal autophagy is normally highly active and required for neuronal survival [11, 12]. The cardinal importance of autophagy in central neurons is further supported by recent studies showing a rapid accumulation of autophagosomes in cortical neurons when lysosomal degradation is inhibited. Thus, constitutive autophagy apparently plays an active role in neurons even under nutrient-rich conditions [49, 50].
2.3.1. Axonal homeostasis
Constitutive autophagy is probably essential for axonal homeostasis. Suppression of basal autophagy by either deleting an atg gene or inhibiting autophagic clearance in neurons disrupts axonal transport of vesicles destined for lysosomal degradation, and causes axonal swelling and dystrophy [11, 12, 37, 50]. For examples, Atg1/Unc-51 mutants in C. elegans show defaults in axonal membranes , and Unc-51.1, the murine homologue of Unc-51 is necessary for axonal extension, suggesting a possible role for these proteins in axonal membrane homeostasis [20, 52, 53]. In the cerebellum, neuron-specific deletion of FIP200, a protein implicated in autophagosome biogenesis, causes axon degeneration and neuronal death . Altogether, these data suggest that autophagy is essential to maintain axonal structure and function through retrograde axonal transport . The degree of vulnerability and the formation of intracellular inclusions vary significantly among the different types of CNS neurons in mutant brains deficient in Atg5 or Atg7 suggesting disparate intrinsic requirements for autophagy and relative levels of basal autophagy . For example, while ubiquitinated inclusions are rare in the Atg5- or Atg7-deficient PCs, these cells are among the most susceptible neurons to Atg 5/7 gene deletion [54, 55]. ULK1, the human homologue of Atg1 is incorporated into the active NGF-TrkA complex after its K-63 polyubiquitination and association with p62 [52, 56]. The subsequent interaction of ULK1 with endocytosis regulators allows trafficking of NGF-bound TrkA receptors into endocytic vesicles  providing a possible mechanism of crosstalk between autophagy and endocytosis. By fusing with autophagosomes, some membrane compartments, including endosomes, can be removed from axons and degraded in lysosomes. This process maintains the homeostasis of the axonal membrane networks and as such is essential for axonal physiology [20, 53].
Indeed, dysfunctional autophagy has been implicated in axonal dystrophy. Axonal swellings occur in autophagy-deficient mouse brains [11, 12] and genetic ablation of Atg7 provokes cell-autonomous axonal dystrophy and degeneration, inferring that autophagy is crucial for membrane trafficking and turnover in axons . In Atg5- or Atg7-deficient PCs, axonal endings exhibit an accumulation of abnormal organelles and membranous profiles much earlier than the somato-dendritic compartment [53, 54]. Axonal degeneration is increasingly believed to precede somatic death by a non-apoptotic auto-destructive mechanism [58, 59]. The “dying-back” progressive retrograde degeneration of the distal axon is a likely model of the chronic injury observed in neurodegenerative diseases . NGF-deprivation induces autophagosome accumulation in the distal tips of neurites of PC12 cells, and knocking down Atg7 or beclin 1 expression delays neurite degeneration of NGF-deprived sympathetic neurons . This suggests that overactive or deficient autophagy contributes to axonal degeneration in a dying-back manner due to the fragility of the axonal tips .
2.3.2. Dendritic autophagy
Early autophagosomes have also been observed in dendrites and the cell body of neurons suggesting that axon terminals are not be the only sites where neuronal autophagosomes form, and that autophagy may play a regulatory function in dendrites under physiological and pathological conditions . Along this line, mTOR a key regulator of the autophagic pathway, modulates postsynaptic long-term potentiation and depression, suggesting that autophagy may critically control synaptic plasticity at the postsynaptic, dendritic compartment . Further investigations are required to determine the specific roles of autophagy in dendrites and axons.
Since autophagosomes can fuse with endosomes and form amphisomes, there is a link between autophagy and endocytosis . ESCRT proteins have recently been implicated in normal autophagy [19, 63, 64]. The endocytic pathways, in particular multi-vesicular bodies (MVBs) may serve as critical routes for autophagosomes to reach lysosomes, because defects in ESCRT function prevents fusion or maturation of autophagosomes. The ESCRT-MVB pathway could represent the primary, if not the only, route for delivering autophagosomes to lysosomes in some cell types . In neurons, a large part of the endocytosed cargo merges with the autophagic pathway prior to being degraded by lysosomes . Alterations in ESCRT function have also been linked to autophagy-deficiency in fronto-temporal dementia (FTD) and amyotrophic lateral sclerosis (ALS). In these cases, the particular vulnerability of the neurons appears to be associated with a dysfunction in the autophagosome-MVB pathway in the dendritic compartment .
2.3.3. Protein homeostasis
Neurons deficient in Atg5 or Atg7 exhibit an accumulation of polyubiquitinated proteins in inclusion bodies even though the proteasome function is normal, suggesting that basal autophagy prevents spontaneous protein aggregation and plays an essential role in protein clearance and homeostasis in neurons. Such a function is even more critical in neurons expressing disease-related proteins like the aggregate-prone mutant α-synuclein and polyglutamine-containing proteins [66-68], although how autophagy selectively degrades these disease-related proteins is unclear. The ubiquitin-associated protein p62 is a likely candidate, providing a link between autophagy and selective protein degradation. Indeed, p62 binds numerous proteins through multiple protein-protein interacting motifs, including one for LC3 [55, 56] and the ubiquitin-associated C-terminal domain which binds ubiquitinated proteins. The relationship between p62 and autophagy is further supported by the observation that a marked accumulation of p62 and LC3 occurs only when lysosomes, but not proteasomes, are blocked. Furthermore, p62 protein levels are elevated in autophagy-deficient neurons [36, 55]. This argues that p62 is a specific substrate of autophagic degradation rather than a molecule involved in autophagosome formation since p62-knockout mice display intact autophagosomes and slower protein degradation. Autophagy-deficient cells and neurons accumulate ubiquitin- and p62-positive inclusions, and this accumulation is greatly reduced by ablating p62 [55, 69]. p62 with mutations in the LC3 recognition sequence escape autophagic degradation, leading to the formation of inclusions, whereas those with mutations in the self-oligomerizing domain PB1 are poorly degraded, but no protein inclusions form. Thus increased levels and oligomerization of p62 are required for the formation of inclusion bodies, and their degradation is facilitated by oligomerization. Ubiquitinated aggregates induced by proteasome inhibition are also reduced in p62-deficient cells suggesting that p62 is a general mediator of inclusion formation and normally functions as an adaptator targeting proteins for autophagic degradation .
2.3.4. Neuronal autophagy in synapse development, function and remodeling
Neuronal autophagy has been recently shown to play an important role in synapse development. The ubiquitin-proteasome system negatively regulates growth of the neuromuscular junction (NMJ) in Drosophila  whereas NMJ is positively regulated by neuronal autophagy; a decrease or an increase in autophagy correspondingly affects synapse size . Indeed, an overexpression or a mutation of Atg1, a gene involved in autophagy induction, respectively enhanced or decreased NMJ growth. Furthermore, this positive effect of autophagy on NMJ development is mediated by downregulating Hiw, an E3 ubiquitin ligase which negatively regulates synaptic growth by downregulating Wallenda (Wnd), a MAP kinase kinase [70- 72]. Although autophagy is considered as a nonselective bulk degradation process, it can regulate specific developmental events in a substrate-selective mode [73, 74]. In C. elegans, when presynaptic afferences are removed from postsynaptic cells, GABAA receptors are selectively targeted to autophagosomes . Accordingly, Hiw could traffic to autophagosomes via a still unknown mechanism, although Hiw could be unselectively degraded by autophagy along with other presynaptic proteins. Interestingly, the synaptic density in mice carrying an atg1 mutation is decreased due to excessive activity of the MAP kinase ERK, suggesting that activated ERK negatively regulates synapse formation and that Atg1 regulates synaptic structure by downregulating ERK activity . As pointed out by Shen and Ganetzky , autophagy is a perfect candidate to modulate synaptic growth and plasticity in function of environmental conditions, resulting in plausible consequences in learning and memory.
Autophagy has recently been shown to regulate neurotransmission at the presynaptic level . Besides enhancing protein synthesis via the mTORC1 complex, mTOR activity inhibits autophagy by an Atg13 phosphorylation-induced blockade of Atg1 . In the nervous system, mTORC1 promotes learning and synaptic plasticity dependent on protein synthesis [78- 80]. Conversely, the mTOR inhibitor rapamycin impedes protein synthesis and blocks cell injury-induced axonal hyperexcitability and synaptic plasticity, as well as learning and memory [81, 82]. In prejunctional dopaminergic axons, inhibition of mTOR induces autophagy as shown by an increase in autophagosome formation, and decreases axonal volume, synaptic vesicle number and evoked dopamine release. Similarly, non-dopaminergic striatal terminals also display more autophagosomes and fewer synaptic vesicles. Conversely, chronic autophagy deficiency in dopamine neurons increases dopaminergic axon size and evoked dopamine release, and promotes rapid presynaptic recovery. Thus mTOR-dependent axonal autophagy locally regulates presynaptic structure and function. In cultured brain slices, the occurrence of autophagosomes in presynaptic terminals isolated from their cell bodies confirms that autophagosomes are locally synthesized , and supports the view that this autophagy may serve to modulate presynaptic terminal function by sequestrating presynaptic components . The global stimulating effect of chronic autophagy deficiency on dopaminergic neurons is consistent with the implication of autophagy in neurite retraction of sympathetic neurons in vitro  and neuritic growth in developing neurons .
There are only a few other reports indicating that autophagy may participate in synapse remodeling. In the cerebellar cortex of the (Bax-/-;GluRδ2Lc) double mutant mouse (Fig. 2A), prominent autophagic profiles are evident in parallel fiber terminals subjected to intense remodeling in the absence of the PCs, their homologous target neurons . As mentioned above, endocytosed GABAA receptors are selectively targeted to autophagosomes in C. elegans neurons , whereas autophagy promotes synapse outgrowth in Drosophila . Autophagy may also modulate synaptic plasticity as recently demonstrated in mammalian hippocampal neurons . Here, neuronal stimulation by chemical LTD induces NMDAR-dependent autophagy by inhibiting the PI3K-Akt-mTOR pathway. Enhanced autophagosome formation in the dendrites and spines of these neurons targets internalized AMPA receptors to lysosomes suggesting that autophagy contributes to the NMDAR-dependent synaptic plasticity required to maintain LTD and assure certain brain functions . A possible mechanism for this formation of autophagosomes and autophagic degradation of AMPARs in dendritic shafts and spines may involve a change in endosome cycling. The formation of more amphisomes due to the fusion of endosomes with autophagosome [86, 87] would reduce the recycling endosome population, and direct more AMPAR-containing endosomes to autophagosomes for lysosomal degradation. Another alternative, but not exclusive actor is p62. This autophagosomal protein is important for LTP and spatial memory , interacts with AMPAR and is required for the trafficking of AMPAR . AMPAR via its interaction with p62 would be trapped in autophagosomes as their number increased . mTOR regulates protein turnover in neurons by functioning at the intersection between protein synthesis and degradation. During learning and reactivation in the amygdale and hippocampus, rapamycin inhibition of mTOR has recently been shown to impair object recognition memory , implicating signaling mechanisms involved in protein synthesis, synaptic plasticity and cell metabolism in this cognitive function.
2.4. Few autophagosomes, a feature of basal neuronal autophagy
Neurons are highly resistant to large-scale induction of autophagy in response to starvation, probably due to the multiple energy sources available to assure their function . Interestingly, the activity of mTOR, a negative regulator of autophagy is significantly reduced in hypothalamic neurons from mice after a 48h starvation , although there are reports that autophagy in neurons can be regulated independently of mTOR [92, 93]. For example, insulin impairs the induction of neuronal autophagy in vitro, but in its absence induction of autophagy is mTOR-dependent. Furthermore, a potent Akt inhibitor provokes robust autophagy . Thus insulin signaling maintains a low level of autophagosome biogenesis in healthy neurons constituting a critical mechanism for controlling basal autophagy in neurons. In addition to insulin signaling, multiple parallel signaling pathways including the mTOR pathway can regulate autophagy in neurons. From these data, Yue and collaborators  have proposed that basal autophagy in CNS neurons is regulated by at least two mechanisms: (1) a non-cell-autonomous mechanism whereby regulators (nutrients, hormones and growth factors) are supplied by extrinsic sources (glia, peripheral organs), (2) a cell-autonomous mechanism controlled by intrinsic nutrient-mediated signaling or specific factors expressed in neurons.
Neurons may depend less on autophagy to provide free amino acids and energy under physiological conditions given their quasi exclusive use of blood-born glucose as a source of carbon and energy for protein synthesis. Accordingly, the primary function of neuronal autophagy may be different than a primary response to starvation, and autophagy regulatory mechanisms are likely to be specific in neurons. Furthermore, gender differences in autophagic capacity have been suggested by the faster autophagic response to starvation of cultured neurons from male rats compared to those from females . While in vivo evidence of neuronal autophagy mediated by nutrient signaling is still missing, a number of stress-related signals, neuron injuries and neuropathogenic conditions trigger prominent formation and accumulation of autophagosomes in neurons. During this process, neurons may undergo a significant change in autophagy regulation, involving a deregulation that allows neurons to switch from basal level (neuron-specific process featured by a low number of autophagosomes) to an activated state (well-conserved induced autophagy with large-scale biosynthesis of autophagosomes) . Hypoxic-ischemia [95, 96], excitotoxicity [97-99], the dopaminergic toxins, methamphetamine and MPP+ [65, 100, 101], proteasome inhibition [102-104], lysosomal enzyme/lipid storage deficiencies [105-108] are examples of these pathological inducers of neuronal autophagy (see below).
3. Autophagy in neuronal physiopathology
Autophagy normally protects effect against neurodegeneration, but defects in the autophagy machinery are sufficient to induce neurodegeneration. Indeed, neuron-specific disruption of autophagy results in neurodegeneration [11, 12]; for example PC-specific Atg7 deficiency impedes axonal autophagy via an important p62-independent axonopathic mechanism associated with neurodegeneration . Furthermore, specific defects in selective autophagic components or in the cargo selection process can induce neurodegeneration. This hypothesis is supported by the studies of cargo recognition and degradation components, such as p62, NBR1, or ALFY [109, 110]. Defects at any one of the autophagic steps can cause an abnormal accumulation of cytosolic components and lead to disease states. Therefore, each step of the autophagic process needs to be tightly regulated for efficient autophagic degradation.
The housekeeping role of neuronal autophagy is more evident when neurons are loaded with pathogenic proteins . In many neurodegenerative disorders, cytoplasmic, nuclear and extracellular inclusions composed of aggregated and ubiquitinated proteins are believed to contribute to organelle damage, synaptic dysfunction and neuronal degeneration. The autophagic process in diseased neurons participates in the clearance of abnormal aggregate-prone proteins such as the expanded glutamine (polyQ)-containing proteins (e.g. mutant huntingtin in Huntington’s disease (HD)), mutant forms of α-synuclein in familial Parkinson disease (PD), different forms of tau in Alzheimer’s disease (AD), tauopathies and FTD, mutant forms of SOD1 in motor diseases such as ALS, and mutant forms of PMP22 in peripheral neuropathies are cleared from diseased neurons by autophagy [19, 20, 55, 56, 66, 67, 111-115]. However, accumulation of these intracellular aggregates is believed to play a significant role in the etiology of neurodegenerative diseases including prion diseases (PrD) [3, 67]. One common feature is the dramatic cyto-pathological accumulation of autophagosomes in injured and degenerating neurons [116-121]. Such signs of defects in autophagy have been interpreted as a result of an “autophagic stress”, or in other words an imbalance between protein synthesis and degradation . This has traditionally been viewed as a highly destructive cellular mechanism, driving the cell to death . In these diseases, it is now accepted that autophagy eliminates aggregate-prone proteins and damaged organelles more efficiently than the proteasome machinery. Since the proteasome is unable to degrade them , the clearance of misfolded, aggregated proteins originating from neuropathologic deficits is highly dependent on autophagy. However, a blockade of the autophagic flux is likely to impede the clearance of these proteins. The accumulation of aggregated proteins and organelles within the diseased neurons then contributes to cell dysfunction and in the end results in cell death , (Fig. 3). Indeed, pharmacological upregulation of autophagy reduces neuronal aggregates and slows down the progression of neurological symptoms in animal models of tauopathy and HD , AD [41, 124, 125] and PrD [126, 127].
The mechanisms that determine the activation of autophagy for the removal of aggregated proteins are not clearly understood, but failure of the other proteolytic systems to handle the altered proteins seems to at least partly underlie autophagy activation. Thus oligomers and fibers of particular proteins can block the proteolytic activity of the ubiquitin-proteasome system and chaperone-mediated autophagy (CMA) that results in autophagy upregulation [128, 129]. In addition, sequestration of negative regulators of autophagy in the protein aggregates could also provoke activation of this pathway. Thus it has been shown that blockage of autophagy in neurons leads to the accumulation of aggregated proteins and neurodegeneration even in the absence of aggregate-prone proteins [11, 12]. Although the specific reasons for the failure of the proteolytic systems are unknown, factors such as enhanced oxidative stress and aging seem to precipitate entry into a late failure stage when the activity of all degradation systems are blocked or decreased, leading to accumulation of autophagic vacuoles and aggregates and finally cell death .
Autophagy protects against cell death in the case of growth factor withdrawal, starvation and neurodegeneration, but it is required for some types of autophagic cell death [131-134]. However, the role of autophagy as a positive mediator of cell death is not well understood in mammalian systems, although many studies suggest that impaired autophagy sensitizes cells and organisms to toxic insults. Atg1-dependent autophagy restricts cell growth . Cells deficient in Pdk1, a positive regulator of mTOR pathway , display autophagy and reduced growth. The increased growth capacity that results from disrupting autophagy may contribute to the tumorigenicity of cells mutant for tumor suppressors [38, 137, 138]. Overexpression of Atg1 leads to apoptotic cell death . Cells undergoing autophagic cell death display signs of apoptosis , as do Atg1-null cells . Thus, elevated levels of autophagy promote cell death and the role of autophagy in cell death is likely to be context-dependent.
Neuronal autophagy is currently believed to constitute a protective mechanism that slows the advance of neurodegenerative disorders, and that its inhibition is associated with neurodegeneration . Substantial attention is currently being focused on the molecular mechanisms underlying the autophagic fight against neurodegeneration, the role of autophagy in early stages of pathogenesis and therapeutical approaches to upregulate protective neuronal autophagy. It is unclear whether accumulation of autophagic vacuoles in degenerating neurons results from increased autophagic flux or impaired flux. A chronic imbalance between autophagosome formation and degradation causes “autophagic stress” . Due to obvious therapeutic consequences, it is imperative to understand how autophagic stress occurs in each autophagy-associated neurodegenerative condition: either a cellular incapacity to support an excessive autophagic demand or a defective degradation (lysosomal) step .
4. Autophagy in prion diseases
4.1. Prion diseases
4.1.1. Infectious and familial prion diseases
Prion diseases (PrD) are transmissible spongiform encephalopathies (TSEs) which are fatal neurodegenerative diseases in humans (Creutzfeldt-Jakob disease (CJD), Gerstmann-Sträussler-Scheinker syndrome (GSS), variant CJD (vCJD), fatal familial insomnia (FFI) and kuru) and in animals (bovine spongiform encephalopathy (BSE), transmissible mink encephalopathy (TME), chronic wasting disease (CWD) and scrapie). In humans, PrD manifest after a long incubation period free of symptoms as a rapid progressive dementia that leads inevitably to death. Severe loss of neurons with extensive astrogliosis and moderate microglial activation, characteristic of all TSEs, results in a progressive spongiform degeneration of the brain tissue which is reflected by ataxia, behavioral changes and, in humans, a progressive cognitive decline [142-145]. According to the protein-only hypothesis , TSEs are caused by prions that are believed to be proteinaceous infectious particles mainly consisting of PrPSC, an abnormal isoform of the normal, host-encoded prion protein (PrPC), . Prions are able to catalyze a switch from PrPC conformation into an aggregated misfolded conformer PrPSC which collects throughout the brain according to a prion strain-specific anatomo-pathologic signature. These PrDs share a protein misfolding feature with other neurodegenerative diseases (e.g. AD, PD and HD), .
The central role played by PrPC in the development of PrD was first illustrated by the observation that disruption of the PrP gene (PRNP) in mice confers resistance to PrD and impairs the propagation of infectious prions , while PrP-overexpressing (tga20) mice exhibit reduced incubation periods when compared with wild-type mice . Overall, the current data argue for a primary role of the neuronal, GPI-anchored PrPC in prion neuropathogenesis .The subversion of PrPC function(s) as a result of its conversion into PrPSC is assumed to account for prion-associated toxicity in neurons . Whether PrPSC triggers a loss of PrPC physiological function (loss-of-function hypothesis) or promotes a gain of toxic activity (gain-of-function hypothesis), or both, is an ongoing debate in the TSE field . Elucidating the roles of PrPC in neurons should help to answer this question. Knockout experiments, however, have failed to reveal any obvious physiological role for PrPC. Mice devoid of PrPC are viable and display only minor phenotypic or behavioural alterations that vary according to the null strain, and hence, these results do not permit one to assign a specific function to PrPC. Ex vivo studies support the involvement of PrPC in copper homeostasis . In addition, the localisation of PrPC on the cell membrane and its affinity for the neuronal cell adhesion molecule (N-CAM), laminin, and the laminin receptor [154, 155] have implicated PrPC in cell adhesion. Such properties may reflect the involvement of PrPC in the outgrowth and maintenance of neurites, and even cell survival. Indeed, recent experimental evidence showing that PrPC interacting with β1 integrin controls focal adhesion and turnover of actin microfilaments in neurons substantiates a role for PrPC in neuritogenesis. Of note, integrins are well known inducers of autophagy (see review in chapter by Nollet and Miranti). Remarkably, during neuronal differentiation, the downregulation of Rho kinase (ROCK) activity by PrPC is necessary for neurite sprouting . A stress-protective activity has also been assigned to PrPC based on results obtained with primary neuronal cultures. Neuronal cells derived from PrP-knockout mice are more sensitive to oxidative stress and serum deprivation than wild type cells [157-159]. Moreover, after ischemic brain injury, PrPC-depleted mice revealed enlarged infarct volumes [160-162]. This neuroprotective role of PrPC has been linked to cell signaling events. The interaction of PrPC with the stress inducible protein (STI-1) generates neuroprotective signals that rescue cells from apoptosis . Previous studies of both neuronal and non-neuronal cells substantiate the coupling of PrPC to signaling effectors involved in cell survival, redox equilibrium and homeostasis (e.g. ERK1/2, NADPH oxidase , cyclic AMP-responsive element binding protein (CREB) transcription factor and metalloproteinases [165, 166]. According to these data, PrPC has been proposed to function as a « dynamic cell surface platform for the assembly of signaling modules » . Despite these overall advances, the sequence of cellular and molecular events that leads to neuronal cell demise in TSEs remains obscure [168, 169]. At present, one envisions that neuronal cell death results from several parallel, interacting or sequential pathways involving protein processing and proteasome dysfunction , oxidative stress [159, 171], apoptosis and autophagy .
4.1.2. Autophagy in prion-infected neurons
Prion propagation involves the endocytic pathway, specifically the endosomal and lysosomal compartments that are implicated in trafficking and recycling, as well as the final degradation of prions. Shifting the equilibrium between propagation and lysosomal clearance impairs the cellular prion load. This and the presence of autophagic vacuoles in prion diseased neurons [173, 174] suggest a role for autophagy in prion infection (reviewed in ). Indeed, the high numbers of autophagic vacuoles observed in neurons from experimentally prion-infected mice (Fig. 4A, B, Fig. 5) and hamsters is indicative of a robust activation of autophagy [175, 176]. Furthermore, autophagic vacuoles and multivesicular bodies have been detected in prion-infected neuronal cells in vitro .The formation of autophagic vacuoles has recently been observed in neuronal pericarya, neurites and synapses of neurons experimentally infected with scrapie, CJD and GSS , as well as in neuronal synaptic compartments in humans with certain PrD . PrDs are further correlated with autophagy given that the Scrg1 protein (encoded by the scrapie responsive gene1, Scrg1) is upregulated in scrapie and BSE-infected brains, as well as in brains of patients with sporadic CJD [178-180] and is associated with neuronal autophagosomes [181, 182]. This Scrg1 protein is thus, a new marker for autophagic vacuoles in prion-infected neurons (Fig. 4A, B). In the brains of CJD and FFI patients and experimentally scrapie-infected hamsters, increased cytoplasmic levels of LC3-II-immunostained autophagosomes have been demonstrated in neurons, again indicating autophagy activation. In addition, the decreased p62 and polyubiquitinated proteins levels in hamster and human brains infected with prion suggest an upregulation of autophagy with enhanced autophagic flux and protein degradation. Downregulation of the mTOR pathway and upregulation of the beclin 1 pathway in these infected tissues provide further evidence of autophagy activation . On the basis of these observations, Xu et al.  propose that neuronal autophagy is an intricate element of prion infections. They suggest that once PrPSC enters host cells and is delivered to endosomes, it accumulates in amphisomes via fusion with autophagosomes and then with lysosomes. At this initial stage of infection, PrPSC does not co-localize with autophagosomes, probably because PrPSC levels are too low to be detected due to their rapid degradation in autophagolysosomes. In agreement with this explanation, blocking the fusion of autophagosomes with lysosomes using bafilomycin A1 permits the visualization of PrP-PG14 and PrPSC in autophagosomes , as is the case for Aβ1-42 .
The role of lysosomes in PrDs is still controversial. Although autophagic lysosomal degradation of PrPSC in infected neurons is supposed to clear prion aggregates and inhibit PrPSC replication, there are indications that PrPSC may subvert the autophagic-lysosomal system to promote the conversion of PrPC into PrPSC. Lysosomal inhibitors prevent the build-up of PrPSC  and agonists of the autophagy-lysosome pathway enhance the clearance of PrPSC [185, 186, 126]. However, as PrPSC production increases, the accumulating PrPSC may saturate the clearance capacity of the system causing lysosomal disruption and release of PrPSC aggregates into the neuroplasm. In turn this would cause cell stress and over-activate autophagy, as has been reported in prion-diseased brain tissue .
The octapeptide repeats region of PrPC has been shown to negatively influence autophagy. As measured by LC3-II expression, autophagy induced by serum deprivation occurs earlier and to a greater extent in hippocampal neurons from ZH-I PrnP-/- compared with those from wild type mice. Reintroduction of PrPC, but not PrPC lacking its N-t octapeptide region, into ZH-I PrnP-/-neurons delays this upregulation of autophagy . The transconformation of PrPC into PrPSC could interfere with the function of this domain and as a consequence, upregulate autophagy. It is conceivable that the activation of autophagy observed in PrD models reflects a defense mechanism designed to degrade prions and resist oxidative stress. A reduction in autophagy combined with endosomal/lysosomal dysfunction has indeed been proposed to contribute to the development of PrD . Furthermore, the anti-cancer drug imatinib has been shown to activate lysosomal degradation of PrPSC  and is a potent autophagy inducer . When administered early during peripheral infection, imatinib delays both PrPSC neuroinvasion and the onset of clinical disease in prion-infected mice . Upregulation of autophagy has beneficial effects on the clearance of aggregate-prone proteins in PrD and other neurodegenerative diseases [66, 111-115, 191, 192]. Both lithium and trehalose enhance PrPSC clearance from prion-infected cells by inducing autophagy, as demonstrated by increases in LC3-II protein and the number of GFP-LC3 puncta [193, 126]. Furthermore, PrPSC can be cleared not only by mTOR-independent autophagy (lithium and trehalose), but also by the mTOR-dependent route because the mTOR inhibitor rapamycin also causes a decrease in cellular PrPSC. Lithium-induced autophagy also reduces PrPC levels. This treatment causes internalization of PrPC , and the consequent reduction of available membrane-bound PrPC is known to decrease its conversion into pathologic PrPSC [195-198]. This would provide an additional, indirect way to reduce PrPSC by reducing of PrPC with lithium treatment.
Whether autophagy-inducing compounds are candidates for therapeutic approaches against prion infection has recently been investigated in prion-infected mice. Starting in the last third of the incubation periods, treatment with rapamycin and to a lesser extent with lithium significantly prolonged incubation times compared to mock-treated control mice [126, 172]. Along this line, activation of the class III histone deacetylase Sirtuin 1 (Sirt1) has been shown to mediate the neuroprotective effect of resveratrol against prion toxicity  and prevent prion protein-derived peptide 106-126 (PrP106-126) neurotoxicity via autophagy processing . Moreover, Sirt1-induced autophagy protects against mitochondrial dysfunction induced by PrP106-126, whereas siRNA knockdown of Sirt1 sensitizes cells to PrP106-126-induced cell death and mitochondrial dysfunction. Finally, knockdown of Atg5 decreases LC3-II protein levels and blocks the effect of a Sirt1 activator against PrP106-126-induced mitochondrial dysfunction and neurotoxicity. Thus inducing Sirt1-mediated autophagy may be a principal neuroprotective mechanism against prion-induced mitochondrial apoptosis. Nevertheless, understanding the mechanisms underlying Sirt1-mediated autophagy against prion neurotoxicity and mitochondrial damage merits further investigation, in particular determining the Sirt1-mediated dowstream signaling network, including FOXOs, p53 and PGC-1α. More recently, the mTOR inhibitor and autophagy inducer rapamycin has been shown to delay disease onset and prevent PrP plaque deposition in a mouse model of the Gerstmann-Sträussler-Scheinker PrD . Here, the reduction in symtom severity and prolonged survival correlate with increases in LC3-II levels in the brains of treated mice, suggesting that autophagy induction enhances elimination of misfolded PrP before plaques form. This is in agreement with the well known neuroprotective effects of rapamycin in various models of neurodegenerative diseases with misfolded aggregate-prone proteins (e.g. PD , ALS , HD , spinocerebellar ataxia [66, 202], FTD  and AD [41, 124, 125].
4.2. Doppel-expressing prion protein-deficient mice
Research efforts to determine the function of PrPC using knockout mutant mice have revealed that large deletions in the PrPC genome result in the ectopic neuronal expression of the prion-like protein Doppel (Dpl) causing late onset degeneration of PCs and ataxia in PrnP-/-mouse lines, such as Ngsk , Rcm0 , ZH-II  and Rikn .
Similar PC degeneration is observed when the N-terminal truncated form of PrP is expressed (ΔPrP) in Prnp-ablated mouse lines  and when Dpl is overexpressed [209, 210]. Of note, full-length PrPC antagonizes the neurotoxic effects of both Dpl and ΔPrP [208-212], but not PrPC lacking the N-terminal residues 23-88 . These results imply that Dpl and ΔPrP induce cell death by the same mechanism, likely by interfering with a cellular signaling pathway essential for cell survival and normally controlled by full-length PrPC [209, 214]. The mechanism underlying Dpl-induced neurotoxicity is still under debate. PrP-deficient neurons undergo Dpl-induced apoptosis in a dose-dependent, cell autonomous manner . Oxidative stress is a likely candidate to play a role in the death of these neurons because NOS activity is induced by Dpl both in vitro and in vivo [212, 216]. Endogenous, as well as exogenous PrPC has been shown to inhibit Dpl-induced apoptosis, a neuroprotective function that has been attributed to its BCL-2-like properties . Like BCL-2, PrPC antagonizes mitochondrial apoptotic pathways, thereby protecting neurons from cell death [217- 219]. In BAX-induced apoptosis [220, 221], PrPC probably acts by preventing the conformational changes in BAX that are necessary for its activation . In primary cultures, Dpl-induced apoptosis of Prnp+/+ as well as PrnP-/- granule cells, has recently been shown to be inhibited by BAX deficiency or pharmacologically blocking caspase-3 suggesting that it is mediated by Bax and caspase-3 . These results further confirm in vivo data concerning the effects of Bax expression on PC survival in the cerebellum of the Dpl-overexpressing Ngsk PrnP-/-mouse that we reported several years ago . In these mice, PC death is already significant as early as 6 months of age. During aging, quantification of PC populations shows that significantly more PCs survived in the Ngsk PrnP-/-:Bax-/- double mutant mice than in the Ngsk PrnP-/-mice. However, the number of surviving PCs is still lower than wild type levels and less than the number of surviving PCs in Bax-/- mutants. This suggests that neuronal expression of Dpl activates both BAX-dependent and BAX-independent pathways of cell death. Interestingly, a partial rescue of Ngsk PrnP-/-PCs is observed in Ngsk PrnP-/--Hu-bcl-2 double mutant mice, in a proportion similar to that found in Ngsk PrnP-/-:Bax-/- mice, strongly supporting the involvement of BCL-2-dependent apoptosis in Dpl neurotoxicity . The capacity of BCL-2 to apparently compensate for the deficit in PrPC by partially rescuing PCs from Dpl-induced death suggests that the BCL-2-like property of PrPC may counteract Dpl-like neurotoxic pathway in wild-type neurons. Although not exactly identical to BCL-2, PrPC may functionally replace BCL-2 as it decreases in the aging brain . The N-terminal domain of PrPC which is partially homologous to the BH2 domain of BCL-2 family of proteins [226, 227] is probably responsible for the neuroprotective functions of PrPC because BAX-induced apoptosis cannot be counteracted by N-terminally truncated PrP. BCL-2 antagonizes the pro-apoptotic effect of BAX by interacting directly with this BH2 domain [228-230], and this domain is missing in both Dpl and the neurotoxic mutated forms of PrP: ΔPrP [208, 214, 231] and Tg(PG14)PrP . Interestingly, expression of Dpl fused to a BH2-containing octapeptide repeat and the N-terminal half of the hydrophobic region of PrPC makes cells resistant to serum deprivation . Furthermore, N-terminal deleted forms of PrPC have been reported to activate both BAX-dependant and BAX-independant apoptotic pathways .
4.2.1. Autophagy in prion protein-deficient mice
The Dpl-activated, BAX-independent cell death mechanism may involve neuronal autophagy as we have detected the expression of Scrg1, a novel protein with a potential link to autophagy in the Ngsk PrnP-/-PCs (Fig. 4C), . Both neuronal Scrg1 mRNA and protein levels are increased in prion-diseased brains [179, 180], and Scrg1 is associated with dictyosomes of the Golgi apparatus and autophagic vacuoles in degenerating neurons of scrapie-infected Scrg1-overexpressing transgenic and WT mice (Fig. 4A, B), [181, 182]. Both before and during PC loss, protein levels of Scrg1 and the autophagic markers LC3-II and p62 are increased in Ngsk PrnP-/-PCs, whereas their mRNA expression is stable, suggesting that the degradation of autophagic products is impaired in these neurons [234, 235]. Autophagic profiles collect in somato-dendritic and axonal compartments of Ngsk PrnP-/- (Figs. 2B, 6), but not wild-type PCs. The most robust autophagy occurs in dystrophic profiles of the PC axons in the cerebellar cortex (Fig. 6D) and at their preterminal and terminal levels in the deep cerebellar nuclei (Fig. 6A-C) suggesting that it initiates in these axons. Taken together, these data indicate that Dpl triggers autophagy and apoptosis in Ngsk PrnP-/- PCs. As reflected by the abundance of autophagosomes in the diseased Ngsk PCs, Dpl neurotoxicity induces a progressive dysfunction of autophagy, as well as apoptosis. Whether this autophagy dysfunction triggers apoptotic cascades or provokes autophagic cell death independent of apoptosis remains to be resolved. In the Ngsk PrnP-/- PCs, the increased expression of LC3-II and p62 at the protein level, without any change in mRNA levels, suggests that the ultimate steps of autophagic degradation are impaired. This is further confirmed by the prominence of autophagolysosomes in these neurons which indicate that the fusion of autophagosomes with lysosomes occurs normally, but downstream, the autophagic flux is blocked.
To further investigate the neurodegenerative mechanisms induced by Dpl in Ngsk cerebellar PCs, we are using an organotypic cerebellar culture system which allows an easier way to approach mechanistic questions than in vivo models . For this purpose, we have assessed the growth and viability of PCs in cerebellar organotypic cultures from Ngsk and ZH-I PrnP-/- mice using morphometric methods to measure PC survival and development . The timing and amplitude of PC growth impairment and neuronal death are similar in Ngsk and ZH-I PrnP-/- cultures (Fig. 7). In addition, increased amounts of autophagic (LC3-II, Fig. 8) and apoptotic (caspase-3, Fig. 9) markers are detected in protein extracts from both cultures indicating that both apoptosis and autophagy (Fig. 2C, D) contribute to PC death in Ngsk  and ZH-I cultures. This suggests that PrPC -deficiency, rather than Dpl expression, is responsible for the neuronal growth deficit and loss in these cultures. For presently unknown reasons, the neurotoxic properties of Dpl do not seem to contribute to the degeneration of Ngsk PCs in these organotypic cultures. As the neurotoxicity induced by Dpl takes about 6 months to develop in vivo, it is possible that organotypic cultures are not mature enough to model 6-month-old cerebellar tissue. Nevertheless, ex vivo cerebellar organotypic cultures do provide a suitable system for analyzing the mechanisms underlying the neurotoxic effects of PrPC-deficiency and prion infections  using pharmacological and siRNA-based approaches.
Our results have shown that PrPC has a neuroprotective role in cerebellar PCs. As PCs survive in vivo in the cerebellum of the ZH-I mouse, the death of the ZH-I PCs in the organotypic cultures is likely to stem from the inherent stress of the ex vivo conditions. As mentioned above, PrPC negatively regulates autophagy as demonstrated by the upregulation of autophagy following serum deprivation in PrnP-/- hippocampal neurons when compared to PrPC-expressing neurons . Recent results suggest that PrPC can directly modulate autophagic cell death. Using antisens oligonucleotides targeting the Prnp transcript, the downregulation of PrPC expression in glial and non-glial tumor cells induces autophagy-dependent, apoptosis-independent cell death . Previous data have shown that PrPC acts as a SOD  and modulates the activity of Cu/Zn SOD by binding 5 Cu++ ions on its N-terminal octapeptide repeat domain [153, 157, 241]. A recent study of the effects of H2O2-induced oxidative stress on hippocampal neurons expressing PrPC or deficient in PrPC provides further support for the protective role of PrPC against oxidative stress . Although autophagy and apoptosis occur in both lines, the Prnp-/- neurons are less resistant to H2O2-induced oxidative stress than the Prnp+/+ neurons confirming the anti-oxidant activity of PrPC.
Furthermore, autophagy is more enhanced in Prnp-/- neurons than in Prnp+/+ neurons. In the latter, this is due to H2O2-induced enhancement of autophagic flux, and in the former due to H2O2-induced impairment of autophagic flux. Similarly, experiments using Atg7 siRNA to inhibit autophagy have revealed that the increased autophagic flux in Prnp+/+ neurons protects against H2O2 cytotoxicity. Thus a deficiency in Prnp may impair autophagic flux via H2O2-induced oxidative stress contributing to autophagic cell death . Since autophagic flux is apparently normal in both Prnp+/+ and Prnp-/- neurons in the absence of stress, the lack of PrPC only seems to contribute to autophagy impairment under stress-induced conditions, such as H2O2 treatment , stress-inducing in vitro conditions, as well as Dpl-induced toxicity.
4.2.2. Prion protein PrP-doppel Dpl chimeras
When overexpressed ectopically in neurons, mutations within the central region of PrPC are associated with severe neurotoxic activity, similar to that of Dpl [231, 243]. The absence of these segments, called central domains (CD) is believed to be responsible for neurodegeneration and ataxia. To understand the dual neurotoxicity vs. neuroprotective roles of PrPC, transgenic mice expressing a fusion protein made of the CD of PrPC inserted within the Dpl sequence have been generated . These mice failed to develop typical Dpl-mediated neurological disorder indicating that this N-terminal portion of PrPC reduces Dpl toxicity. To further investigate Dpl-like neurotoxicity, Lemaire-Vieille et al. recently generated lines of transgenic mice expressing three different chimeric PrP-Dpl proteins . Chi1 (Dpl 1-57 replaced with PrP 1-125) and Chi2 (Dpl 1-66 replaced with PrP1-134) abrogates the pathogenecity of Dpl confirming the neuroprotective role of the PrP 23-134 N-terminal domain against Dpl toxicity. However, when Dpl 1-24 were replaced with PrP 1-124, these Chi3 transgenic mice that express a very low level of the chimeric protein develop ataxia, as early as 5 weeks of age. This phenotype is only rescued by overexpressing PrPC, and not by a single copy of full-length PrPC, indicating the strong toxicity of the chimeric protein Chi3. The Chi3 mice exhibit severe cerebellar atrophy with significant granule cell loss and prominent signs of autophagy in PCs (Fig. 2E). We conclude that the first 33 amino acids of Dpl, that are absent in Chi1 and Chi2 constructs, confer toxicity to the protein. This is confirmed in vitro by the highly neurotoxic effect of the 25-57 Dpl peptide on mouse embryo cortical neurons. Since this chimeric transgene is not expressed by PCs in the transgenic mice expressing Chi3, the signs of autophagy displayed by these neurons in vitro could result from the neurotoxic effect of the exogenous Chi3 chimeric protein, as well as the deleterious effect of losing their primary afferences (i.e. the granule cells).
The beneficial effects that autophagy has on prion infections is currently supported by a growing bulk of evidence from in vivo and ex vivo data and is strongly promising for future mid-term therapeutic approaches. To further understand the fascinating interplay between autophagy and PrDs, further investigations are necessary to decipher their molecular interactions. Important issues remain. How are the different phases of prion infection physiopathology i.e. propagation, trafficking, recycling and clearance connected with autophagy? Which autophagic pathways are activated by prions - the mTOR-dependent, mTOR-independent or both? The biological function of autophagy per se in prion infection is obscure as the cellular levels of autophagy can apparently modify cell susceptibility to prion infection, although changes in autophagy may be a pre-requisite or a consequence of a prion infection.
Overall, the results point to a need to counteract cell stress and to eliminate toxic aggregate-prone proteins that eventually saturate the usual degradation pathways, including autophagy. These are common features of prion disesase and most of the other neurodegenerative diseases described in this review. Saturation of the autophagic machinery, loss or imbalance of autophagic flux is believed to lead to neurodegeneration. Understanding how autophagy relates to these diseases is a first step for developing autophagy modulation-based therapies for treating neurological disorders. This implies therapeutic consideration for each type of autophagic default at a precise step of the neurodegenerative disease concerned.
The authors are greatly indebted to Dr. Anne-Marie Haeberlé (CNRS UPR3212, Strasbourg) for excellent assistance in transmission electron microscopy and prion-infected tissue handling as well as to Dr. Catherine Vidal (Institut Pasteur, Paris) for intra-cerebellar inoculation of mice with 22L scrapie. A. R. is supported by a doctoral grant from the French Minister of Research and Technology and A. G. is supported by a grant from the French Centre National de la Recherche Scientifique and the AgroParisTech High School.