Autophagy takes flight in Drosophila

Abstract

Drosophila has been shown to be a powerful model to study autophagy, whose regulation involves a core machinery consisting of Atg proteins and upstream signaling regulators similar to those in yeast and mammals. The conserved role in degrading proteins and organelles gives autophagy an important function in coordinating several cellular processes as well as in a number of pathological conditions. This review summarizes key studies in Drosophila autophagy research and discusses potential questions that may lead to better understanding of the roles and regulation of autophagy in higher eukaryotes.

1. Introduction

Autophagy is a conserved process in which cytoplasmic proteins or organelles are non-selectively packaged into lysosomes (or vacuoles in yeast) for degradation. Autophagic substrates are broken down to small molecules that are recycled for macromolecular synthesis or used for generating energy, and therefore autophagy is considered as an adaptive system that allows cells to survive starvation. In addition to this non-selective form of autophagy, studies from the last decade have identified subsets of selective autophagic processes that specifically degrade intracellular organelles, such as mitochondria (mitophagy), endoplasmic reticulum (reticulophagy) or peroxisomes (pexophagy) [1]. These specific forms of autophagy provide an alternative way to clear damaged organelles, which can be toxic if accumulated to high levels. In mammals, autophagy has been implicated in several pathological conditions, such as neurodegenerative diseases, tumors and pathogenic infections. Collectively, autophagy is an important cellular process with multiple functions across species.

The delivery of autophagic substrates occurs through specialized double-membraned vesicles called autophagosomes. The formation of autophagosomes requires a tightly controlled mechanism involving a series of Atg proteins, first identified by systematic screens in yeast [2]. The core proteins for autophagy include three major groups, whose functions correspond to the steps of autophagosome formation [1]. The induction signal is transduced through an Atg1 kinase complex; this directs the membrane nucleation of autophagosomes through a second protein complex containing the PI(3)-kinase Vps34; finally, vesicle expansion is mediated by two ubiquitin-like groups, Atg8 and Atg5-Atg12-Atg16. The matured autophagosomes then fuse with lysosomes with the help of a set of general docking proteins to degrade components inside autolysosomes. Together with the target of rapamycin (TOR), a conserved regulatory kinase that inhibits autophagy, these molecules form a complex network for the regulation of autophagy [1].

The short life cycle and powerful genetics of Drosophila, along with a physiology comparable to mammals, has made this organism a handy model system for a wide variety of experimental questions. Together with yeast and mammalian cultured cells where autophagy is extensively studied, Drosophila has provided a useful model to dissect the molecular mechanisms and the physiological roles of autophagy in vivo. Autophagy is inducible by starvation in the Drosophila larval fat body, an analagous organ to mammalian liver, and studies of this response have contributed to our understanding of nutrient-dependent regulation of autophagy. In addition, high levels of autophagy are observed in certain dying cells during metamorphosis and oogenesis in Drosophila, and appear to act in concert with the apoptotic machinery in these contexts to promote cell elimination. The roles of autophagy in aging, neurodegeneration and oxidative stress have also been effectively addressed in this system. Through these studies, several Drosophila genes have been identified for their roles in regulating autophagy, including a group of upstream signaling molecules and the essential Atg homologs (Table 1). These genes all share evolutionary conservation across species and together they depict the molecular mechanism of autophagy, forming the basis for the applications of autophagy in human diseases in the Drosophila model. Therefore, studies in Drosophila can contribute considerably to our understanding of the autophagic process. Here, we summarize recent advances in our knowledge of autophagy function and regulation from experiments in the Drosophila system.

Table 1.

The core genes for autophagy in Drosophila

Name	Functions	Confirmed Role in Autophagy	References
Insulin-TOR signaling pathway
InR	Insulin-like receptor	Yes (inhibition)	[3,38,45]
PI3K	PI3K class I	Yes (inhibition)	[3,27-28]
Rheb	Ras small GTPase family	Yes (inhibition)	[3]
TSC1/TSC2	GTPase-activating protein (GAP)	Yes	[3]
TOR	PIK-family Ser/Thr protein kinase	Yes (inhibition)	[3]
Autophagosome induction
Atg1	Ser/Thr protein kinase	Yes	[3,6,12,27]
Atg13	Atg1 complex	Yes	[6]
FIP200 (CG1347)	Atg1 complex	?
Atg101 (CG7053)	Atg1 complex	?
Autophagosome nucleation
Atg6	Vps34 complex	Yes	[3,27]
Vps15	Ser/Thr protein kinase Vps34 complex	Yes	[16]
Vps34	PI3K class III Vps34 complex	Yes	[16]
Atg14 (CG11877)	Vps34 complex	?
UVRAG (CG6116)	Vps34 complex	?
Rubicon (CG12772)	Vps34 complex	?
Autophagosome expansion
Atg3	E2-like enzyme for Atg8-PE	Yes	[12,27]
Atg4	Cysteine protease cleaves Atg8 C- terminus	Yes	[55]
Atg5	Atg5-Atg12-Atg16 complex	Yes	[3,12]
Atg7	E1-like enzyme for Atg8 and Atg12	Yes	[3,27,36]
Atg8a/ Atg8b	Ubiquitin-like, Conjugated to PE	Yes	[3,12,27,35]
Atg10 (CG12821)	E2-like enzyme for Atg12	?
Atg12	Ubiquitin-like, Atg5-Atg12-Atg16 complex	Yes	[3]
Atg16 (CG31033)	Atg5-Atg12-Atg16 complex	?
Recycling
Atg2	Atg2-Atg9 complex	Yes	[3,27]
Atg9	Atg2-Atg9 complex	?	[3]
Atg18	PI3P binding	Yes	[3,27]

2. Autophagy regulation: TOR and the Atg1 kinase complex

With its multiple functions, autophagy is a tightly regulated process under the control of several intracellular signaling networks. The highly conserved TOR pathway is an important component of these networks, integrating multiple cellular responses to growth factors, nutrients and energy levels (Fig. 1A). Recent work in a number of systems have identified the Ser/Thr protein kinase Atg1 as a central target of TOR in directing the formation of autophagosomes [1]. Loss of Atg1 blocks the formation of autophagosomes, and consensus observations across species have placed Atg1 downstream of TOR [3-5]. The ability of Atg1 to regulate autophagy relies on a group of interacting proteins without enzymatic activities. In yeast, Atg13 and Atg17 are two major components of a multi-protein Atg1 complex [5]. Atg1 activity is depleted in atg13 or atg17 mutant cells and autophagosome formation is greatly impaired in these lines. Whereas clear homologs of Atg17 have not been identified in Drosophila and other higher eukaryotes, Atg13 is indispensable for autophagy in both yeast and metazoans. The well-established yeast model has shown that phosphorylation of Atg13 by TOR signaling disrupts the interaction of Atg1 and Atg13. Upon starvation, Atg13 is dephosphorylated and quickly binds Atg1 to turn on autophagy [5]. In contrast to this yeast model, in which the interaction of Atg1 and Atg13 is limited to starved cells, Drosophila Atg1 and Atg13 interact constitutively regardless of nutrition conditions [6]. Similarly, the mammalian Atg1 homolog Ulk1 forms a 3MDa complex with Atg13, Atg101 and FIP200 that is stable under both fed and starved conditions [7,8]. These observations indicate a regulatory discrepancy in yeast and higher eukaryotes, in which the basal autophagy is constantly maintained (Fig. 2).

Fig 1.

Three major autophagy induction routs in Drosophila.

(A) Upon starvation or removal of growth factors, autophagy is induced through inhibition of TOR, which negatively regulates autophagy in well-fed organisms. The activity of TOR is under the control of Rheb and TSC1/2, both acting downstream of PI3K and AKT but with opposing effects on TOR. (B) Developmental autophagy is triggered by ecdysone signaling to eliminate tissues or organs, accompanied by the activation of cell death genes. The involvement of caspases during developmental autophagy is cell-specific. (C) JNK can direct the induction of autophagy in response to oxidative stress. Atg1 and other Atg genes are transcriptionally activated by JNK signaling, possibly through the dFOXO transcriptional factor.

Fig 2.

Comparison of Atg1 complexes in yeast, Drosophila and mammals.

(A) In yeast, the phosphorylation of Atg13 by TOR signaling prevents Atg13 from forming a complex with Atg1. No autophagy occurs until the phosphorylation on Atg13 is removed in response to starvation. (B) Drosophila Atg1-Atg13 complex is present constitutively in fed and starved conditions. Atg1 and Atg13 are both phosphorylated by Atg1 and TOR signaling; however, Atg1 is more sensitive to TOR signaling in fed animals while phosphorylation of Atg13 is highest under starved condition, where Atg1 activity is elevated. (C) Similar to Drosophila, mammalian Atg1 complexes show little change in composition in response to nutrient status, except that mTOR has higher affinity for the Atg1 complex under fed conditions. Although Atg1 and Atg13 are both substrates of mTOR and Atg1, similar to their Drosophila counterparts, starvation leads to decreased phosphorylation of Atg13 due to lower mTOR activity as well as higher Atg1-dependent phosphorylation of FIP200.

Whereas the yeast Atg1 complex contains at least 8 proteins and mammalian Ulk1 can form a 3MDa complex, the number of Drosophila Atg1-interacting proteins for autophagy regulation remains to be determined. Among eighteen Drosophila proteins that have been identified as potential Atg1-interactors by yeast two-hybrid (http://www.thebiogrid.org/), thus far only Atg13 has been shown to play a role in autophagy [6]. Drosophila Atg1 has also been shown to form a complex with the kinesin heavy chain adaptor protein Unc-76, which has an important function in axonal transport that is distinct from the role of Atg1 in autophagy [9]. Collectively, Drosophila Atg1 may exert distinct functions by recruiting different partners, and in order to fully understand the role of Atg1 in autophagy control, discovering Atg1-interacting proteins specific to autophagy regulation will be a critical task.

Given that Atg1 is a protein kinase, how the kinase activity of Atg1 is involved in autophagy is important to address. Atg1 kinase activity increases after starvation both in yeast and mammalian cells, suggesting this activity is regulated by nutrition cues and contributes to autophagosome formation [5,10]. In addition, Atg1 kinase activity is decreased in yeast atg13 mutants, and co-expression of Atg13 enhances Atg1 kinase activity in both Drosophila and mammalian cells [5,6,11]. Together with the failure of kinase-defective Atg1 to rescue the lethality and autophagy defect of Drosophila Atg1 mutants [12], these findings support the notion that kinase activity of Atg1 is required for autophagy. Klionsky and co-workers further demonstrated two distinct functions of yeast Atg1: assembly of the pre-autophagosomal structure (PAS) requires a kinase-independent structural role of Atg1 in association with Atg13 and Atg17, whereas dissociation of Atg proteins requires Atg1 kinase activity [13]. This finding separates Atg1 kinase activity from the initiation of autophagy in yeast and raises the possibility that Atg/Ulk1 kinase activity may be required at one or more steps following the induction of autophagosome formation in higher eukaryotes.

Coexpression of Atg1 and Atg13 in Drosophila increases the phosphorylation of both of these proteins in a TOR- and Atg1 kinase- dependent manner [6]. This suggests that Atg13 and Atg1 itself are substrates of Atg1 kinase, although indirect phosphorylation by other kinases has not been excluded. Similar hyper-phosphorylation of Atg1 and Atg13 by TOR and Atg1 are also observed in mammals in vivo and in vitro [7,11]. A global, in vitro analysis of peptide phosphorylation identified 188 proteins as potential substrates of Atg1 kinase, including Atg8, Atg18 and Atg21 [14]. Identification of the direct substrates of Atg1 for autophagy regulation will be an important line of future investigation.

Overexpression of Drosophila Atg1 is sufficient to induce autophagy; in contrast, high levels of Ulk1 expression blocks starvation-induced autophagy in mammalian cells [4,12]. A comparable inhibitory effect on autophagy induction also occurs in response to Drosophila Atg13 overexpression [6]. These observations suggest that the Atg1-Atg13 complex can have both positive and negative roles in autophagy regulation. Considering that yeast Atg1 functions as a scaffold protein to initiate autophagy, it is possible that overexpression of either Atg1 or Atg13 makes molecules essential for autophagy unavailable by sequestering them away from their normal loci. Alternatively, autophagy induction may require a strictly balanced ratio of Atg1 and Atg13, and disruption of this balance by overexpression of either protein may lead to autophagic deficiency. This hypothesis is further supported by the observation that coexpression of Atg1 and Atg13 at low levels leads to autophagy induction under fed conditions [6].

In addition to its direct role in autophagosome formation, Atg1 induces autophagy partly through a negative feedback loop to TOR. The activity of TOR signaling is down-regulated in a dose-dependent manner when Atg1 is overexpressed, evident by reduced TOR-dependent phosphorylation of S6K [12]. Coexpression of low levels of Atg1 and Atg13 alters the intracellular localization of TOR from a diffuse perinuclear compartment to large cytoplasmic vesicles, which may indicate a disruption of the normal nutrient-dependent trafficking of TOR [6]. In addition, the sequestering of TOR from its normal loci may rely on the physical binding of TOR and Atg1 [6]. A similar dynamic interaction of TOR and Ulk1 complex is also evident in mammalian cells [7]. Taken together, the interaction of TOR and Atg1/Ulk1 complexes appear to involve several different levels, and the ultimate decision of autophagy induction likely reflects the balance of TOR and Atg1/Ulk1 activity.

3. Type III PI3K complex

The double membrane of autophagosomes is a unique feature, making autophagosomes distinct from other vesicles. However, the origin of this double membrane is still debatable, and different origin sources have been suggested, such as ER or mitochondria [15]. A phosphatidylinositol-3-phosphate (PI3P)-enriched structure seems to be the site at which autophagosomes form. PI3P is the product of phosphoinositide 3-kinases (PI3K) and is known to play a critical role in autophagy. Treatment with Wortmannin or 3-methyladenine, general inhibitors of PI3Ks, potently blocks autophagy in mammalian cells, supporting the involvement of PI3P in autophagy formation. Although there is only one PI3K in yeast, three classes of PI3K have been characterized in Drosophila and mammals, and mutations in Vps34, the type III PI3K that generates PI3P, block the formation of autophagosomes in Drosophila [15,16]. These genetic results demonstrate the requirement of PI3K for autophagy, consistent with the effects of PI3K inhibitors in mammals. Interestingly, although over-expression of Drosophila Vps34 can increase the intensity of autophagy in starved animals, this is insufficient to induce autophagy under fed conditions [16]. These results indicate that generation of PI3P is not enough to induce autophagy without the coordinated effects of other Atg proteins or TOR-dependent signals.

In yeast, Vps34 regulates autophagy through a complex containing Atg6, Atg14 and Vps15 [1]. Both Drosophila Atg6 and Vps15 are required for autophagy and can interact with Vps34, suggesting that this conserved machinery is utilized in Drosophila [16]. Interestingly, a number of different Vps34 complexes have been observed in mammals, each containing the core proteins Atg6/Beclin 1, Vps15/p150 and Vps34, and different combinations of Atg14L, Ambra1, UVRAG or Rubicon [15]. Orthologs of Atg14L, UVRAG and Rubicon are also present in the Drosophila genome, indicating that Drosophila Vps34 may similarly form different complexes with specific functions in directing autophagosome formation.

4. Endocytic pathway and autophagy

The observation that Vps34 functions both in autophagy and endocytosis raises the question whether other components of the endocytic pathway are also involved in autophagy [16]. The endosomal sorting complex required for transport (ESCRT) complex contains 4 subgroups, including ESCRT-0, ESCRT-I, ESCRT-II and ESCRT-III. These subgroups function stepwise to control the delivery of ubiquitinated receptors to multivesicular bodies (MVB) [15]. Mutations in Drosophila vps28 (ESCRT-I), vps25 (ESCRT-II), vps32 (ESCRT–III), or vps4 (an AAA ATPase required for ESCRT-III function) each show increased levels of Atg8 punctate structures in fat body and ovarian follicle cells [16, 17]. Observation of these mutants by electron microscopy reveals the accumulation of autophagosomes but lack of autolysosomes or amphisomes, which result from fusion of autophagosomes and endosomal compartments. These results indicate that ESCRT components are required for an essential step in the maturation and fusion of autophagosomes with the endosomal compartment. Similar accumulations of autophagosomes in ESCRT mutations are evident in mammalian and nematode cells [18]. Interestingly, ESCRT components are not required for autophagy in yeast, as autophagosomes are apparently able to fuse directly with the yeast vacuole, without prior input from the endocytic pathway [15].

The fusion of autophagosomes with lysosomes requires a group of docking proteins acting on both sides of autophagosomes and lysosomes. These docking proteins include components of the homotypic fusion and protein sorting (HOPS) complex, consisting of the Vps-C complex (Vps18, Vps11, Vps33 and Vps16) together with Vps41 and Vps39. Mutation in Drosophila deep orange (dor), encoding a Vps18 homolog, causes accumulation of endosomes, suggesting a conserved role in endocytic trafficking. As observed in ESCRT mutants, autophagosomes accumulate in dor mutants in larval fat body cells, where developmental autophagy is induced to degrade fat bodies for pupation [19]. Similar accumulation of autophagosomes in mutants of dvps16A, one of two vps16 in Drosophila genome, supports the idea that the full HOPS complex is essential for autophagy in multicellular organisms, as in the yeast model [20]. Interestingly, UVRAG is able to associate with the HOPS complex, and overexpression of UVRAG enhances autophagosome fusion and autophagic flux in a Beclin 1-independent manner in mammals [21]. The function of this Vps34 component at a late step of autophagosome formation raises the question of how these dynamic endocytic proteins are regulated and integrated in autophagy regulation. For proteins with functions in both the endocytic pathway and autophagy, it is necessary to clarify whether and how these two functions overlap as well as their precise roles in autophagosome formation. As mentioned above, the function of Drosophila UVRAG has not yet been studied, and it will be interesting to determine whether Drosophila UVRAG has similar separate functions in autophagosome induction and maturation. Another Drosophila protein with dual roles in autophagy and endocytosis is liquid facets (lqf), a homolog of vertebrate epsin, whose mutation impairs endocytosis and developmental autophagy [22]. The roles of lqf in autophagy and endocytosis are reminiscent of ESCRTs and Vps34, and the lack of accumulation of autophagosomes in lqf mutants implies that lqf may function at early step of autophagy, similar to Vps34.

5. Developmental autophagy and apoptosis

Although both autophagy and apoptosis are capable of leading cells to death as a final destiny, their relationship is still paradoxical. Diverse approaches have been applied to answer this question in different organisms, including yeast, Drosophila and mammals. The primary distinction of autophagy and apoptosis is based on the morphology of cells undergoing either process. Whereas the defining characteristic of autophagy is the formation of double-membrane vesicles containing organelles or cytoplasm, DNA fragmentation and cytoplasmic blebbing serve as fundamental morphological indicators of apoptosis. In Drosophila, the steroid hormone ecdysone controls larval molting and metamorphosis during the fruit fly life cycle. The level of ecdysone peaks before each molting in larval stage, and disruption of normal ecdysone levels can cause an arrest of larval growth [23]. A gradual rise in ecdysone synthesis at the end of the larval period induces developmental autophagy, allowing cellular reorganization in response to developmental timing. A peak of ecdysone at the end of the larval period triggers metamorphosis, the process to eliminate the larval tissues which are no longer necessary for adults and to prepare the maturation of adult tissues. Several larval tissues that undergo such elimination serve as excellent models to study the relationship between autophagy and apoptosis, and studies in Drosophila are beginning to elucidate general mechanisms by which steroid hormones can control both autophagic and apoptotic responses (Fig. 1B).

Dying larval midgut cells display several markers of apoptosis, such as DNA fragmentation, acridine orange staining and activated expression of proapototic genes. Mutation of E93, an early-acting ecdysone-regulated gene, blocks the destruction of the larval midgut; however, the surviving midgut cells still contain fragmented DNA, suggesting that induction of apoptosis is not sufficient for larval midgut cell death [24]. Accordingly, midgut degradation is not disrupted by expression of the pan-caspase inhibitor p35 nor by mutation of major caspases, further demonstrating that apoptosis is dispensable for developmental midgut degradation [25]. In contrast, mutation of E93 does inhibit the accumulation of autophagic vesicles normally observed in dying midgut cells. In addition, midgut destruction is blocked in animals lacking Atg1, Atg2 or Atg18 activity, directly implicating autophagy as a crucial process in ecdysone-induced degradation of midgut cells [24-25]. Caspase deficiency does not enhance the Atg mutant midgut phenotypes, indicating that autophagic cell death in the midgut is caspase-independent despite the high levels of caspase activity during this process [25].

The larval salivary gland, another tissue that is degraded during metamorphosis, also utilizes autophagy for its destruction [26]. The incomplete degradation of salivary glands in Atg mutant animals clearly indicates that salivary gland cell death is autophagy-dependent [27]. Ecdysone-mediated induction of E93 is also critical for autophagy-dependent salivary gland destruction. Expression of the class I PI3K catalytic subunit, or its target, AKT, inhibits salivary gland degradation [27], reminiscent of the requirement for PI3K down-regulation by ecdysone signaling during developmental autophagy in the larval fat body [28]. Caspase activity remains intact in these glands with high PI3K activity, in contrast to the low caspase activity, lack of DNA fragmentation and persistent autophagic vacuoles in glands expressing p35 [26]. Caspase activity is apparently normal and DNA fragmentation is also clearly observed in the salivary glands of a number Atg mutants. The combination of p35 expression with either elevated PI3K activity or Atg mutation enhances the malfunction of salivary gland destruction by either one, strongly suggesting a parallel regulation of salivary gland cell death by PI3K/autophagy and caspases [27]. Atg1 overexpression is sufficient to cause premature salivary gland degradation devoid of DNA fragmentation, and this is not suppressed by p35 expression, supporting the proposal that autophagic death of salivary gland cells is caspase-independent. This parallel model differs from observations made in Drosophila aminoserosa, fat body and wing disc cells, whose degradation induced by Atg1 is suppressed by p35 expression [12,29]. Further, DNA fragmentation is significantly reduced in dying ovarian cells in Atg1 or Atg7 mutants, indicating a strong epistatic connection between autophagic cell death and caspases [30-31]. It should be noted that caspases acts upstream of autophagy to direct the starvation-induced ovarian cell death, while autophagy is required to activate caspases during developmental ovarian cell death [30-31]. Together with findings in mammalian cells that autophagy can be induced as a backup mechanism when caspase activity is compromised [32], these differences in dependency on caspases of autophagic cell death may reflect differences in development stages and cell types.

6. Oxidative stress and the JNK pathway

The versatile Jun-N-terminal kinase (JNK) pathway is best known for its role in apoptosis. As a branch of the mitogen-activated protein kinase (MAPK) pathway, the activity of JNK is regulated via a kinase cascade. Drosophila JNK and its upstream kinase (JNKK) are both encoded by single genes, basket (bsk) and hemipterous (hep), respectively. After activation by Hep, Bsk phosphorylates two transcriptional factors, Jun-related antigen (Jra) and Kayak (Kay) (Drosophila Jun and Fos, respectively). Jra and Kay facilitate the transcriptional induction of an array of JNK target genes, including the phosphatase Puckered (Puc). Following activation by JNK, Puc down-regulates JNK signaling through negative feedback to Bsk, which is dephosphorylated and inactivated by Puc [33]. This feedback loop activates JNK signaling in a precise timeframe, by which Drosophila JNK is highly regulated and has been implicated in several cellular process, such as dorsal closure, wound healing and longevity [33].

Treating wild type larvae with H₂O₂ or paraquat, a chemical inducer of oxidative stress, simultaneously induces autophagosome formation and activates JNK signaling, suggesting a connection between autophagy and JNK (Fig. 1C) [34]. Accordingly, paraquat-induced autophagosome formation is suppressed in bsk mutant animals, indicating that autophagy is a downstream effector of JNK signaling. Flies with higher JNK activity have an increased survival rate when challenged with paraquat, and this advantage is lost when Atg1 and Atg6 levels are compromised, indicating that the anti-oxidative stress ability of JNK signaling requires intact autophagy machinery. Consistent with these findings, overexpression of Atg6 or Atg8 increases the resistance of flies to oxidative stress, whereas flies bearing Atg7 or Atg8 mutations are more sensitive to oxidative stress [34-36].

Following paraquat treatment, expression of Atg1 and Atg18 rises transiently in concert with the peak of JNK activation, implying that Atg genes may be direct transcriptional targets of the JNK pathway [34]. Indeed, constitutive activation of JNK signaling by expression of activated Hep leads to increased expression of Atg1 and Atg8, and subsequent autophagy induction [34]. However, JNK signaling is dispensable for developmental or starvation-induced autophagy, evident by the observation that both autophagic processes proceed normally in the absence of JNK activity [34]. In contrast to these results in Drosophila, JNK is activated by starvation in mammalian cells [37]. In fed cells, Bcl-2 is predominantly partnered with Beclin 1. Upon the stimulus of starvation, phosphorylation of Bcl-2 by JNK disrupts its association with Beclin 1, allowing Beclin 1 to interact with Vps34 and initiate autophagosome formation [37]. Together, these observations imply a unique role of Drosophila JNK in autophagy induction, and suggest the effect of JNK on autophagy induction may be limited to non-nutritive stress in Drosophila.

Drosophila dFOXO is a member of the FOXO (Forkhead Box O) family of transcriptional factors, which are important for stress resistance. Genetic interaction experiments in Drosophila demonstrate a strong connection between JNK signaling and dFOXO. Targeted overexpression dFOXO in the developing eye results in a small, rough eye phenotype, which is suppressed by reducing JNK activity; similarly, removing one copy of dFOXO suppresses an eye defect caused by expression of activated JNK [38]. High JNK signaling up-regulates the expression of dFOXO target genes, including growth controlling effector eIF4E binding protein (4E-BP) and oxidative stress protective small heat shock proteins (sHsps) [38]. Thus, JNK positively regulates the activity of dFOXO, suggesting that the anti-oxidative stress effect of JNK may partly be accounted for by the elevated expression of sHsps through dFOXO. Recently, Juhasz et al reported that dFOXO is essential and sufficient for autophagy induction, establishing a direct connection between dFOXO and autophagy [39]. Given the connections between FOXO and JNK pathways and their roles in autophagy regulation, it is reasonable to speculate that the effects of JNK on autophagy are mediated through FOXO-dependent transcription of Atg genes. If so, it will be important to determine how these signals are integrated with Fos/Jun-dependent outputs and non-transcriptional branches of this pathway.

7. Effects of autophagy on lifespan

Aging is the ultimate path for all organisms, usually accompanied by signs of accumulation of cellular damage, increased sensitivity to stresses, and reduced fitness to the environment. The role of autophagy in degrading defective cellular components and assisting cells against stresses suggests that this process may have beneficial effects on lifespan. The expression levels of several Drosophila Atg genes, including Atg2, Atg8a and Atg18, decline as flies age, consistent with a role of autophagy in anti-aging [35]. Similarly, Beclin 1 levels are reduced in elder human brains [40], and the rate of autophagy has been suggested to decrease as organisms age.

Although flies bearing Atg8a or Atg7 mutations can survive to adult stage, they have a reduced lifespan, increased levels cellular damage and sensitivity to oxidative stress, and perform poorly in aging-related mobility tests [35-36]. Mice lacking atg7 or atg5 progress through embryogenesis with no apparent developmental abnormality, but die soon after birth [41-42]. Similarly, mutations in C. elegans atg7 and atg12 shorten lifespan, and down-regulation of bec-1 suppresses the extended lifespan caused by mutant daf-2, the C. elegans ortholog of insulin/IFG-1 receptor tyrosine kinase [43-44]. Interestingly, overexpression of Atg8a in the Drosophila central nervous system is sufficient to significantly increase lifespan and reduce accumulation of ubiquitinated and oxidized protein [35]. Pan-neuronal overexpression of Atg8a early in development had no beneficial effect in this study. These results suggest that although Atg7 and Atg8a are largely dispensable for embryonic and larval development, survival during adulthood is closely tied to the levels of autophagic proteins, and, presumably, to autophagic capacity or rate. Thus, therapies aimed at maintaining autophagy at higher levels late in adult life may have a beneficial effect on lifespan.

The aging process is also controlled by insulin-like signaling in Drosophila. Reduced insulin-like signaling, through mutations in insulin-like receptor (InR) or the InR substrate chico, is beneficial to longevity. dFOXO appears to be a critical factor downstream to insulin-like signaling for longevity control. Phosphorylation of dFOXO by insulin-like signaling causes its translocation from nucleus to cytosol, thereby inhibiting expression of dFOXO target genes. Specific expression of dFOXO in adult head fat body significant prolongs lifespan [45]. More strikingly, this localized expression of dFOXO induces systemic down-regulation of insulin-like signaling throughout the organism, evident by the overall increased nuclear retention of dFOXO. The level of dFOXO is inversely correlated with the expression of Dilp2, one of seven insulin-like molecules in Drosophila [45]. Together, these findings suggest that the longevity effect of dFOXO is specific to adult head fat body and acts cell non-autonomously through Dilp2 [45].

As discussed above, JNK protects against oxidative stress in part through dFOXO-mediated transcription. Further, several lines of evidence suggest that JNK may also promote longevity through dFOXO-mediated inhibition of insulin-like signaling. Flies with increased JNK activity live longer, and this advantage is suppressed by loss of one copy of dFOXO. Activation of JNK signaling specifically in insulin-like peptide-generating cells significantly extends lifespan and down-regulates the level of dilp-2 in a dFOXO-dependent manner [38]. As mention above, both JNK signaling and dFOXO are essential and sufficient for autophagy induction, raising the possibility that the benefical effect of these factors on lifespan is via autophagy [34,39]. Further investigation of the JNK-FOXO-autophagy connection in Drosophila should address whether the lifespan effects of localized dFOXO and JNK expression reflect local benefits of autophagy in the head or non-autonomous effects in the peripheral tissues.

8. Autophagy in Drosophila neurodegeneration models

Neurodegenerative diseases are progressive disorders that affect millions of people worldwide. The loss of specific neuronal populations is the classic pathology of neurodegeneration. A wide range of studies have converged toward the concept that the misfolding and accumulation of specific proteins in neurons is the root cause of neuronal cell degeneration and other symptoms of these diseases such as uncontrolled movement [46]. For example, patients with Huntington’s disease express a toxic form of huntingtin protein with an expanded run of glutamine repeats, which forms aggregates in neurons, a typical pathological feature of this disease. The severity of neurodegenerative diseases usually correlates with the expression levels of these specific mutant proteins. Therefore the clearance mechanism of toxic proteins and aggregates in neuronal cells is of high clinical interest [46].

The short life cycle, powerful genetics, and visible morphological defects make Drosophila a useful system for studying neurodegeneration. Several neurodegenerative disease models have been successively developed in Drosophila, such as Huntington’s, Parkinson’s and Alzheimer’s diseases. For example, age-dependent neurodegeneration of the fly retina is observed in eyes expressing pathogenic versions of huntingtin, ataxin-1, or other aggregate-prone proteins carrying poly-glutamine or poly-alanine extensions [47-48]. Rapamycin treatment reduces the severity of these neurodegeneration phenotypes, in an autophagy-dependent manner [48]. Similarly, inhibition of TOR in mouse models of Huntington’s disease significantly increases the clearance of hungtingtin aggregates, whereas over-expression of Rheb increases huntingtin aggregation [47]. Interestingly, TOR protein is sequestered into pathogenic huntingtin aggregates, leading to decreased TOR signaling and induction of autophagy [47]. Sequestering effects on TOR protein are also observed with intranuclear ataxin-1 and in brains from patients with spinocerebellar ataxia type 2, 3 and 7 [47]. An independent study described a similar induction of autophagy by Ataxin-3 in Drosophila, suggesting that induction of autophagy by pathogenic aggregates is a common phenomenon in neurodegenerative diseases [49]. Thus, aggregate-prone proteins appear to protect cells from their own toxicity in part by recruiting and sequestering TOR into the aggregates, leading to autophagy induction and increased protein clearance.

The autophagy-lysosomal pathway functions in parallel to the ubiquitin-proteasome system, the other major pathway of cellular degradation. In degenerative neuronal cells, ubiquitinated proteins that are marked for proteasomal degradation often accumulate and form aggregates. Accumulation of ubiquitinated protein aggregates is also a common observation in Drosophila and mice lacking Atg5, Atg7 or Atg8a [35-36,41,50], indicating an intriguing interaction between these two systems. A recent study showed that aging flies have increased expression of Ref(2)P, the Drosophila homolog of P62, accompanied by an increased level of ubiquitinated protein [51]. Ref(2)P was shown to interact with ubiquitinated protein aggregates through its ubiquitin-associated (UBA) forming detergent-insoluble aggregates. Similar to huntingtin aggregates, autophagy is required for the clearance of these p62 and ubiquitinated protein aggregates, which are also found in organisms with neurodegenerative diseases. Disruption of either proteasomal or autophagy activity significantly increases the level of these aggregates and enhances their colocalization in young wild type flies. However, deletion of either the PBI multimerization domain or the UBA domain of p62 suppressed aggregate accumulation caused by Atg8a mutation, suggesting that binding of p62 to ubiquitin is crucial for aggregate formation. The ability of p62 to bind both Atg8/LC3 and ubiquitin brings the autophagy machinery to p62-ubiquitinated protein aggregates for their degradation, which may exemplify how autophagy ameliorates neurodegeneration [52].

Another recent study further demonstrates the intersection of the autophagy and proteasome systems in controlling neurodegeneration [53]. Inhibition of proteasomal activity by DTS7, a temperature-sensitive dominant-negative mutation of the beta2 subunit of the proteasome, causes a degenerative eye morphology. The DTS7-induced eye phenotype is enhanced in Atg mutants and strongly suppressed by rapamycin treatment. The suppression by rapamycin is impaired by loss of Atg12 or Atg6, indicating that deficient proteasomal activity causes neuronal degeneration in an autophagy-dependent manner.

9. Conclusions

The versatility of autophagy as a catabolic process with a variety of substrates allows it to play unique roles in the control of cell death, cell survival, organism development and disease control. These functions rely on a complex regulatory network, whose components are still being identified. The conserved morphology and regulation of autophagy allows researchers to study this process in different model organisms; among them, the advantages of Drosophila as a model to study the functions and mechanism of autophagy are evident. A series of Drosophila proteins involved in the autophagic process have been identified, including the core proteins consisting of Atg proteins and TOR-related signaling regulators, as well as proteins with functions in other processes, such as the endocytic pathway. An important future goal for researchers working in Drosophila will be to use the powerful genetics of this system to identify new factors acting in autophagy through forward genetic screens, and to piece together the mechanisms by which these components function together. Fruits from the first of such screens are beginning to be realized, and suggest a wide variety of proteins impact this process through distinct mechanisms [23,49,54]. The evolutionary conservation of autophagy suggests that studies in Drosophila will provide useful resources to understanding the overall mechanism of autophagy across species.

Abbreviations

Atg1

Autophagy-related gene 1

TOR

target of rapamycin

PI3K

phosphoinositide 3-kinase

Ulk1

Unc-51 like kinase 1

S6K

RPS6-p70-protein kinase

JNK

Jun-N-terminal kinase