Neuronal macroautophagy: From development to degeneration

Abstract

Macroautophagy, a lysosomal pathway responsible for the turnover of organelles and long-lived proteins, has been regarded mainly as an inducible process in neurons, which is mobilized in states of stress and injury. New studies show, however, that macroautophagy is also constitutively active in healthy neurons and is vital to cell survival. Neurons in the brain, unlike cells in the periphery, are protected from large-scale autophagy induction because they can use several different energy sources optimally, receive additional nutrients and neurotrophin support from glial cells, and benefit from hypothalamic regulation of peripheral nutrient supplies. Due to its exceptional efficiency, constitutive autophagy in healthy neurons proceeds in the absence of easily detectable autophagic vacuole intermediates. These intermediates can accumulate rapidly, however, when late steps in the autophagic process are blocked. Autophagic vacuoles also accumulate abnormally in affected neurons of several major neurodegenerative diseases, including Alzheimer’s disease and Parkinson’s disease, where they have been linked to various aspects of disease pathogenesis including neuronal cell death. The build-up of autophagic vacuoles in these neurological disorders and others may reflect either heightened autophagy induction, impairment in later digestive steps in the autophagy pathway, or both. Determining the basis for AV accumulation is critical for understanding the pathogenic significance of autophagy in a given pathologic state and for designing possible therapies based on modulating autophagy. In this review, we discuss the special features of autophagy regulation in the brain, its suspected roles in neurodevelopment and plasticity, and recent progress toward understanding how dysfunctional autophagy contributes to neurodegenerative disease.

Introduction

Cell survival depends on the capacity to change metabolic states in response to external stimuli that vary from early development to senescence. Depending on the availability of nutrients (proteins, carbohydrates, and lipids) or trophic stimuli (growth factors and hormones), cells can switch metabolically from an anabolic state to one of catabolism. The turnover of cytoplasmic constituents for energy is regulated by the endosomal/autophagic/lysosomal (EAL) system, complementing the degradation of most short-lived proteins by the ubiquitin-proteasome system. Autophagy is the general term used to describe the “self-eating” or catabolism of cytoplasm, including organelles, which may occur by at least three distinct routes: chaperone-mediated autophagy (CMA); microautophagy; and macroautophagy. While all three forms of autophagy ultimately end with the degradation of substrates within lysosomes, each process has unique features. Chaperone-mediated autophagy involves the selective targeting of proteins containing a KFERQ-like peptide motif to lysosomes for degradation (Majeski and Dice, 2004), while microautophagy involves the pinocytosis of small quantities of cytosol directly by lysosomes (Muller et al., 2000). The third form of autophagy, known as macroautophagy, involves the sequestration of cytosolic regions containing proteins, sugars, lipids, RNA, and organelles, such as mitochondria, and perixosomes into double-membraned vacuoles that deliver their contents to late endosomal and lysosomal compartments for degradation (Shintani and Klionsky, 2004). While basal levels of autophagy ensure that intracellular protein aggregates and expired mitochondria are removed (Brunk and Terman, 2002, Komatsu et al., 2005), autophagy has been largely recognized as an inducible process due to its markedly heightened activity by nutrient or growth factor-deprivation, stress or pathogenic invasion (Levine, 2005, Kiffin et al., 2006). When macroautophagy is induced, the sequestration of large regions of cytosol into autophagic vacuoles (AVs), often with a diameter of 0.5 μm or larger, makes it the largest contributor to autophagic proteolysis, compared to CMA and microautophagy (Massey et al., 2006).

Although macroautophagy (hereafter referred to as autophagy) and protein synthesis are at opposite ends of the metabolic spectrum, both mechanisms are sensitive to intracellular amino acid levels and share a common signaling pathway, which is regulated by the growth regulating kinase, mTOR (mammalian target of rapamycin) (Beugnet et al., 2003). Phosphorylation of mTOR via the PI3 kinase/AKT-signaling pathway induces protein synthesis, while mTOR dephosphorylation inhibits protein synthesis and is linked with autophagy activation (Somwar et al., 1998). mTOR-mediated autophagy is particularly sensitive to the availability of specific amino acids (e.g. leucine and histidine), insulin, growth factors, and ATP, and if the levels of any of these regulators decrease autophagy is activated (Proud et al., 2001, Lum et al., 2005).

To date, seven autophagy-related “atg” proteins are known to directly regulate the formation of AVs (for a detailed review see (Ohsumi, 2001, Levine and Klionsky, 2004). Firstly, the ATG12 ubiquitin-like conjugation system, comprised of ATG 5, 7, 10, and 12, regulates the elongation of the AV isolation membrane, known as a phagophore (Mizushima et al., 1998, Mizushima et al., 2001). Secondly, the ATG8 conjugation system (ATG 3, 4, 7 and 8) regulates the attachment of the phospholipid, phosphatidylethanolamine (PE) to ATG8 (Ichimura et al., 2000). A mammalian homolog of ATG8, which was first characterized in yeast cells, is known as microtubule-associated protein light chain 3 (MAP-LC3) (Lang et al., 1998, Ohsumi, 2001). MAP-LC3 is normally cleaved from microtubule-associated proteins 1A and B by a cysteine protease, producing the cytosolic LC3-I isoform (Kabeya et al., 2000). The neuronal cytoplasm is densely packed with microtubules, and as a result, LC3-I is found abundantly throughout the neuronal cytoskeleton (Mizushima, 2004; Boland et al, unpublished observations, 2006).

Metabolic regulation of autophagic proteolysis was first proposed by de Duve and colleagues in the 1960s using cell fractionation and electron microscopy analyses of rat liver cells. In the 1980s, autophagy-signaling pathways were mainly characterized in yeast cells; however, the liver continues to be the most extensively used organ/cell type for research on mammalian autophagy. The fluctuating presence of autophagy modulators, such as free amino acids, insulin, and adrenaline, and activators, such as glucagon, in the liver’s blood supply makes hepatocytes an ideal cell type to study autophagy repression and activation (Seglen et al., 1991). Nutrient supply to the brain, by contrast, is tightly regulated and autophagy induction is not apparent in cortical regions of the mouse brain, even after 48 h of food deprivation (Mizushima, 2004). To understand the brain’s resistance to starvation-induced autophagy, it is important to consider the unique adaptive mechanisms that protect neurons from nutritional deprivation and the unwanted repercussions of large-scale autophagy activation. In this review, we will discuss why ongoing autophagy is less apparent in the healthy brain than in other tissues. We will also consider recent studies that suggest neuronal autophagy fulfills important survival functions in neurodevelopment and in mature neurons. Finally, we will discuss the implications of autophagy dysfunction in neurodegenerative disease and the importance of distinguishing among varying possible molecular bases for autophagy-related pathology to define new therapeutic interventions in these disorders.

Despite using the greatest percentage of total body energy, the brain contains minimal energy reserves and is highly dependent on an external supply of nutrients. The selective uptake of substrates, such as glucose and amino acids (Vannucci et al., 1997, Hawkins et al., 2006), and peptide hormones, such as insulin and leptin (Banks, 2004, Zhang et al., 2005), across the blood-brain-barrier ensures optimal nutrient entry into the brain. When nutrient availability is low, the brain maintains a sufficient nutrient supply by increasing the catabolism of glycogen, adipose fat, and muscle from peripheral sources. Glucose-sensitive neurons in the hypothalamus adjust peripheral glucose levels by autonomic stimulation of the pancreas to release glucagon and insulin, the adrenal medulla to release adrenaline and noradrenaline, and the pituitary gland to release hormones that regulate glucocorticoid and thyroid hormone levels (Uyama et al., 2004). In addition to raising peripheral glucose levels, the autophagy-stimulating peptide, glucagon, increases free amino acid levels by inducing both glycogen breakdown and autophagy in the liver (Deter and De Duve, 1967). Insulin also assumes the dual function of increasing glucose uptake into cells while inhibiting autophagy (Pfeifer, 1977). The hypothalamus, therefore, may be considered the hierarchical regulator of autophagy in vivo, in addition to its role as a regulator of blood glucose levels (Fig. 1). Phospho-mTOR is particularly abundant in the paraventricular and arcuate neurons of the hypothalamus and is responsive to peripheral nutrient fluctuation (Cota et al., 2006). Although mTOR activity is minimally affected in cortical and hippocampal regions of the brain by prolonged starvation for up to 48 h, mTOR activity declines after 24 h of starvation in hypothalamic arcuate neurons and influences food intake in mice (Cota et al., 2006).

While large-scale autophagy in neurons is undesirable, constitutive autophagy is now recognized as a vital housekeeping function (Komatsu et al., 2005). Although autophagy helps to maintain energy stores, it also requires ATP, the availability of which influences activity (Sakai and Ogawa, 1982). Therefore, in addition to regulating peripheral nutrient supply, the brain has evolved mechanisms that ensure optimal availability and utilization of energy substrates at a local level. Neurons ensure maximum glucose utilization by having superior glucose uptake and higher expression of the glucose-degrading enzyme, hexokinase, compared to glial cells (Vannucci et al., 1997, Cimino et al., 1998). Neurons are also able to self-regulate synaptic activity depending on ATP availability, which is mediated by ATP-sensitive potassium channels $(K_{\mathrm{ATP}}^{+})$ (Peters et al., 2004). If ATP levels progressively decline during profound hypoglycemia, reduced ATP inhibition of $K_{\mathrm{ATP}}^{+}$ channels allows K⁺ ion influx to hyperpolarize neurons, which decreases neurotransmission (Levin et al., 2001). The “global silencing” of the brain that occurs during extreme starvation, diabetes, and stroke, is believed to protect neurons from degenerating (Mobbs et al., 2001) and may have interesting implications for neuronal autophagy regulation.

Neurons are also protected from nutrient deprivation at a local level by glial cells, which provide more than just scaffold-based support for neurons. Glial cells release growth factors, cytokines, and other substrates that are essential for neuron survival. In the absence of vital growth factors such as NGF and/or GDNF, sympathetic ganglion neurons die with extensive autophagic involvement (Yu et al., 2003), highlighting roles of glial-derived growth factors in regulating neuronal autophagy as well as anabolic signaling. The brain stores minimal amounts of glycogen compared to other organs, and these stores are almost exclusively found in astrocytes (Phelps, 1972). When blood glucose levels are low, astrocytic glycogen is degraded into lactate, ketones and glucose, which are released at gap junctions to nearby neurons (Guzman and Blazquez, 2001, Ghosh et al., 2005). Unlike most cells, neurons have evolved the ability to metabolize either glucose, lactate, or ketones, depending on which source is available (Peters et al., 2004), thus ensuring neuronal energy supply is not compromised in times of low glucose availability.

After the discovery of lysosomes by de Duve in the 1960s, neurons became a focus of early morphological investigations of autophagy (Holtzman and Novikoff, 1965). Some of the first important insights into the dynamic mechanisms of this process, however, came several decades later by Hollenbeck and colleagues, whose studies characterized autophagosome-related compartments, their retrograde transport, and progressive maturation and fusion with lysosomes along neuritic processes (Hollenbeck, 1993). By reporting on active real-time AV formation at distal axonal regions, Hollenbeck provided evidence that constitutive neuronal autophagy may serve as both a key mechanism for remodeling neurite and growth cone structure during neurite extension and as a neuroprotective mechanism by removing damaged proteins and organelles that may otherwise accumulate within axons. Recent additional studies have emphasized the importance of efficient bi-directional vesicular transport and retrograde signaling for neuronal survival (Bannai et al., 2004, Reichardt and Mobley, 2004).

Despite indications of active autophagy in developing cultured neurons, detection of constitutive autophagy activity in neurons of the adult brain was elusive until recently. Unlike fibroblasts and hepatocytes that contain considerable numbers of AVs in nutrient-rich conditions, healthy neurons contain few AVs (Nixon et al., 2005) and express low levels of autophagosome-bound LC3-II relative to LC3-I, which is several orders of magnitude more abundant (Mizushima, 2004, Yu et al., 2005). The unchanging levels and distribution of LC3-II in the cortex after nutrient deprivation in mice also initially reinforced the idea that autophagy is minimally active in brain. However, this idea seemed at odds with the fact that macroautophagy is the only mechanism available to cells for the turnover of organelles: these structures have finite half-lives and neurons survive many decades. Indeed, constitutive autophagy has recently been shown to be an indispensable function in neurons based on studies of transgenic mice in which either of two ATG genes required for autophagy function were deleted specifically from neurons. Mice lacking either atg5 or atg7 exhibit motor and behavioral deficits by p14 as well as degeneration and loss of Purkinje cell neurons in the cerebellum and pyramidal neurons in the hippocampus (Hara et al., 2006, Komatsu et al., 2006). Diffuse protein aggregates appear in surviving neurons within these brain regions and others, culminating in the formation of toxic inclusion bodies. Because neuronal proteasome activity is not reduced in these mouse models, the results imply that autophagy is normally responsible for clearing aggregated proteins that are not degraded by the proteasome (Rideout et al., 2004). Severe neurodegeneration in mice that are deficient in neuronal autophagy underscores the importance of constitutive autophagy for neuronal survival.

The observation that AVs are generally scarce in brain tissue although autophagy is active in healthy neurons can now also be understood from the kinetics of AV formation and degradation. It has been shown that LC3-II levels and distribution at a single point in time do not accurately represent levels of autophagic activity: autophagosome markers, such as LC3-II, rise if autophagy is either induced or is blocked at a later autophagy step that slows or prevents degradation of the marker (Tanida et al., 2004). This concept, applied to the brain, suggests that a very low frequency of autophagy intermediates, such as autophagosomes, can reflect either a low rate of autophagy or the efficiency of autophagosome turnover (and LC3-II degradation) is very high. Recent studies of neurons (Boland et al., submitted) support this idea. Primary neurons in culture have few AVs but transiently accumulate many LC3-II-positive autophagosomes after induction with rapamycin or serum withdrawal. Inhibiting cathepsin activity, however, also rapidly and markedly elevates LC3-II and increases numbers of autophagosomes without changing the activity of mTOR, an index of autophagy induction. These observations imply that constitutive autophagy in neurons is quite active but is exceptionally efficient at eliminating autophagosomes and later stage autophagic vacuoles once they are formed. A highly efficient AV processing system in cells that are also largely protected from nutritional fluctuations could partly explain why few AVs are normally seen in healthy neurons of the brain.

These foregoing studies emphasize that ancillary criteria beyond the levels of autophagosome markers are required to accurately interpret the basis for any AV accumulation within cells. Distinguishing among factors that modulate the rate of AV formation and the rate of AV degradation (Fig. 2) is now critical for understanding the basis for AV accumulation in pathologic states and for interpreting how dyfunctional autophagy contributes to disease development (Chu, 2006). Indices of autophagy signaling and induction based on the activity of mTOR and indices of autophagic protein turnover efficiency, such as metabolic labeling analyses (Massey et al., 2006), have become essential to an assessment of autophagy in pathologic states. Although single timepoint measurements of LC3-II expression are difficult to interpret, levels measured at multiple timepoints can be more informative; a tapering over time of an initial rise in LC3-II, for example, is commonly seen with autophagy induction, while a progressive increase in LC3-II over time is more consistent with impaired degradation (Tanida et al., 2004). The morphology of different populations of AVs accumulating in neuropathological states is also informative when characterizing pathologic autophagy responses. As shown in Fig. 3, for example, the AVs associated with autophagy induction in primary cortical neurons are quite distinct from those that accumulate when AV fusion is impaired by vinblastine-induced disruption of the cytoskeleton for short or long periods, or when degradation of autophagic cargoes is blocked by inhibiting lysosomal enzymes (Henell and Glaumann, 1984, Tanida et al., 2004) (Fig. 3, unpublished observations).

The involvement of autophagy in cell differentiation and development has been mainly studied in yeast, plants, nematodes, and flies (Levine and Klionsky, 2004), although early morphologic studies of brain development were the basis for defining a form of neuronal programmed cell death (PCD) distinct from apoptosis and necrosis, called autophagic cell death (ACD) (Cao et al., 2006), which is characterized by a marked proliferation of AVs and the progressive disappearance of organelles (Schweichel and Merker, 1973, Clarke, 1990). In ACD, neurons destined for elimination internalize cytoplasmic components into autophagic compartments to effect self-degradation. Death occurs by hyperactive autophagy and an important criterion for preventing or slowing ACD is by inhibiting autophagy (Nixon and Cataldo, 2006). While this concept has been validated for cells in some developing tissues (Baehrecke, 2005), few examples of ACD in brain tissue are well established by these criteria (Nixon and Cataldo, 2006). Insights into the importance of autophagy in PCD during brain development, however, will likely come from the studies of mice in which either the atg5 or atg7 genes required for autophagosome formation are deleted specifically in neurons (Hara et al., 2006, Komatsu et al., 2006). These mice survive to an early postnatal age indicating that brain development is still possible in the absence of atg5 or atg6, even though many populations later degenerate. Because neural circuitry has not yet been analyzed, however, the question of whether or not autophagy is essential for development, connectivity, or PCD of specific neuronal populations in developing mammalian brain is yet to be answered fully. It will also be important to establish in these future studies whether or not developing neurons compensate in some way for the loss of atg5 or atg7 function by inducing other forms of lysosomal degradation.

The importance of beclin 1 (AuTophagy Gene 6; ATG6), a component of a PI3 kinase protein complex essential for autophagy induction and cell growth control (Liang et al., 1999, Kihara et al., 2001), has also been investigated by generating mice lacking this gene. Although no report has been published to date on neuronal defects in these beclin 1 (−/−) mice, their death at approximately 7.5 days in utero indicates that embryonic development is severely impaired (Yue et al., 2003). The basis for a more profound effect on embryogenesis when beclin 1 is deleted, compared to that seen in atg5 and atg7 null mice is not clear but might reflect more pervasive effects of beclin 1 on cell function or, possibly, less effective compensatory mechanisms in neurons for the loss of beclin 1 function.

In the adult brain, long-term synaptic plasticity involves remodeling of nerve terminals not unlike the remodeling that occurs in developing neurons as they extend processes and form connections. This plasticity, which underlies long-term potentiation (LTP) and depression (LTD) (Tang et al., 2002; Hou and Klann, 2004), is maintained partly by local protein synthesis at synaptic regions (Steward and Schuman, 2001). The abundance of mTOR in hippocampal dendrites has been interpreted as evidence for pre-synaptic protein synthesis and is supported by additional observations that several translational proteins that are regulated by mTOR, such as 4E-BP1, 4EBP-2, and eIF-4E, are also present in post-synaptic hippocampal dendrites. Notably, mTOR activity appears to modulate post-synaptic LTP and LTD as well as NMDA and AMPA-mediated protein synthesis in dendrites (Gong et al., 2006). Autophagy, however, also regulates GABA_A receptor trafficking and degradation via the ATG8 homolog, GABARAP (GABA receptor associated protein) (Leil et al., 2004, Rowland et al., 2006). It is, therefore, interesting to speculate that the regulation at synapses of both autophagy and protein synthesis by mTOR activity is critical for different aspects of synaptic plasticity and that mTOR might act as a critical switch point that suppresses autophagy during synaptic strengthening (LTP) and induces it during LTD.

While the brain is relatively protected from autophagy induction by nutritional deprivation compared to other tissues, neurons are, nevertheless, responsive to other stimuli that induce autophagy, particularly in pathologic states. Neurotrophin deprivation, often a factor in neurodegeneration, is a strong stimulus for autophagy induction (Xue et al., 1999). Toxic factors, such as abnormal protein aggregation, also stimulate autophagy in disease states (Fortun et al., 2003, Yamamoto et al., 2006), using in some cases a route that bypasses the nutrient-regulated mTOR pathway (Yamamoto et al., 2006). In addition, other pathologic factors impair the execution of autophagy during its normal housekeeping role or after its induction in response to stress. As discussed earlier, either induction or impaired autophagy execution in neurons can promote AV accumulations, which are increasingly reported as pathological findings in certain neurodegenerative disorders. Although AV proliferation in degenerating cells has often been interpreted as an indication that autophagy over-activation is mediating neurodegeneration or ACD, an alternative view is emerging that autophagy is induced in most of these disease settings to protect neurons by eliminating abnormal and potentially toxic proteins and organelles that might trigger injury or apoptosis (Levine and Yuan, 2005, Nixon and Cataldo, 2006). Degeneration and cell death can subsequently develop when etiologic factors in the disease directly or indirectly impair the efficacy of autophagy in its neuroprotective role. In some cases, progressive damage and destabilization of the membranes of the accumulating AVs and lysosomes can lead to the cytoplasmic release of lysosomal hydrolases and cell death (Nixon and Cataldo, 2006, Kroemer and Jaattela, 2005).

While the necessary criteria to establish ACD have been applied to only a few neuropathological states so far, overactive autophagy seems to be a key factor mediating the degeneration of Purkinje neurons in Lurcher mice, which express a mutant form of the ionotropic glutamate receptor, Grid2 (Yue et al., 2002). An interaction between Grid2 and Beclin1 is the first example of a direct link between neurotransmitter receptor function and autophagy regulation (Yue et al., 2002). The possible modulation of autophagy by a Grid2-Beclin 1 interaction, and the role of GABARAP in autophagy regulation and GABA receptor trafficking (Kabeya et al., 2000), suggest important relationships between neurotransmission and autophagy activation in neurons, which might provide insights into the role of autophagy in excitotoxin-mediated neurodegeneration. Autophagy is also activated in neurons exposed to hypoxic (Zhu et al., 2005) or excitotoxic stimuli (Borsello et al., 2003); however, an incisive role of autophagy over-activation in the degenerative process is not yet established. A pharmacological inhibitor of autophagy, 3-methyladenine (3MA), has also been shown to protect against cell death in certain in vitro settings, including chloroquine-treated cortical neurons (Zaidi et al., 2001), nerve growth factor (NGF)-deprived sympathetic neurons (Xue et al., 1999), and cerebellar granule cells deprived of serum and K+ (Canu et al., 2005). Autophagy in these examples is considered to be either a triggering stimulus or a partner with either apoptotic or necrotic mechanisms in the cell death process (Levine and Yuan, 2005).

In other neurodegenerative disease states in which autophagy has been implicated, the evidence suggests that neuronal cell death stems from a failure of autophagy to sustain an adequate level of degradative activity rather than overactive autophagy. Studies of atg5 and atg7 gene deletion underscore the importance of a constitutive level of autophagosome formation for neuronal survival even in the absence of any disease factor that could elicit an autophagic response. Further supporting the pathogenic significance of autophagy failure are observations that impeding the actions of one or more cathepsins essential for degrading autophagosome cargoes is associated with progressive neurodegeneration as shown in cathepsin D (−/−) mice and cathepsin B and L (−/−) mice (Koike et al., 2005). Autophagy dysfunction associated with AV accumulation in neurons can also arise from impaired vesicular transport (Stokin et al., 2005), defective fusion of autophagosomes with lysosomes (Ravikumar et al., 2005), or reduced lysosomal activity (Bi et al., 1999). Autophagosome–lysosome fusion is slowed in aging cells (Terman, 1995) and the accumulation of dysfunctional mitochondria in aging neurons (Brunk and Terman, 2002) lends additional support to the idea that age-related impairments of autophagy are likely to be an important factor in the development of some late-age-onset neurodegenerative diseases. The appearance of oxidative stress in the brain in many of these aging-related disorders may involve, in part, the impaired degradation of mitochondria (“mitophagy”), a major contributing factor to oxidative stress (Lemasters, 2005).

The emerging view that autophagy is initially induced as a neuroprotective response in stressed or injured neurons but is subsequently overwhelmed or impaired by disease-related factors could partly account for evidence that autophagy seems to be both induced and impaired in several major neurodegenerative diseases, including Alzheimer’s disease (AD). Impaired autophagic function in AD is evidenced by a massive build-up of autophagy intermediates especially within dystrophic dendrites of affected neurons, indicating that the usually efficient progression of autophagosomes to lysosomes is impeded (Nixon et al., 2005). Autophagosome–lysosome fusion is already known to be slowed by normal cell aging (Martinez-Vicente et al., 2005) and additional risk factors for AD, including mutations of presenilin, impair autophagy (Nixon et al., in press, Cataldo et al., 2004). Autophagic vacuoles are highly enriched in γ-secretase and actively generate the toxic amyloid-β peptide (Aβ) during autophagy (Yu et al., 2005). Although normally most of the generated Aβ would be degraded within lysosomes, in AD and transgenic AD models, the marked buildup of autophagic intermediates within an impaired autophagy pathway is a significant source and intracellular reservoir of Aβ (Yu et al., 2005). Aβ within lysosomal compartments can destabilize AV membranes and trigger release of hydrolytic enzymes into the cytoplasm (Glabe, 2001). The ε4 variant of the cholesterol carrier apolipoprotein E (ApoE), a strong risk factor for AD, compounds this membrane injury (Ji et al., 2002). Therefore, in light of evidence that constitutive autophagy is required for neuron survival, a strong link seems to exist between AD, autophagic failure, neurodegeneration, amyloidogenesis, and possibly the intracellular accumulation of other disease-related proteins such as tau.

Autophagy failure in the presence of autophagy induction is also evident in other neurological disorders. Juvenile neuronal ceroid lipofuscinosis caused by mutations of an endosomal/lysosomal membrane protein encoded by CLN3, causes mental retardation, dementia and extensive neurodegeneration (Cao et al., 2006). A genetic model of the disease, the CLN3 (−/−) mouse, exhibits neurodegeneration associated with accumulations of AVs containing incompletely degraded proteins, which is indicative of impaired autophagic proteolysis. Metabolic studies of CLN3 (−/−) mice confirm that autophagic protein turnover is slowed in cerebellar cells even though levels of autophagy induction are higher than normal. Further inhibiting autophagy promoted neurodegeneration in this mouse model (Cao et al., 2006). Similarly, in a model of AD, the PS/APP mouse, inhibiting autophagic proteolysis with leupeptin promotes neurodegeneration at inhibitor doses that do not similarly affect neurons in wild-type mice (Nixon et al., 2001). Stimulating proteolysis within AVs may be beneficial in states of autophagy impairment (Butler et al., 2005). Huntington’s disease is a neurodegenerative disorder caused by gene mutations that encode abnormally long sequences of polyglutamine in the protein, Huntingtin (htt). When overexpressed in cells, mutant htt accumulates in autophagic compartments in amounts proportionate to the length of the polyglutamine sequence in the protein (Kegel et al., 2000), suggesting that its turnover by autophagy is impeded. Inhibiting the formation of autophagosomes or impeding their fusion with lysosomes (Ravikumar et al., 2005) increases htt aggregation in cells in vitro and in vivo. By contrast, stimulating autophagy with rapamycin treatment reduces htt accumulation and neurodegeneration in cell and fly models of polyglutamine disease and reduces neurological deficits and htt aggregation in a mouse model of Huntington’s disease (Ravikumar et al., 2004); this suggests that inadequate levels of autophagy contribute to the complex neuronal cell death pattern in this disease (Hickey and Chesselet, 2003).