Keywords

1 Introduction

Autophagy is a highly conserved catabolic process essential to maintain cell and tissue homeostasis. In most of the situations, it plays a pro-survival role and is induced in response to both external and intracellular cues, including amino acid deprivation, growth factor withdrawal, low cellular energy levels, ER stress, hypoxia, oxidative stress, infections, and organelle damage (He and Klionsky 2009; Klionsky 2005; Lum et al. 2005; Yorimitsu and Klionsky 2005). Autophagy has been considered for a long time a nonselective process for bulk degradation of either long-lived proteins or cytoplasmic components to both recycle building blocks and help restoring the cellular energy balance during nutrient deprivation. Recent evidences, however, have revealed the existence of numerous types of selective ­autophagy used to specifically eliminate unwanted structures, including organelles and invading microorganisms. As a result, under specific conditions, autophagosomes can exclusively sequester and turn over protein inclusions caused by aggregate-prone or misfolded proteins (a process named aggrephagy), peroxisomes (pexophagy), mitochondria (mitophagy), ER (­reticulophagy), ribosomes (ribophagy), secretory granules (zymophagy), and pathogens (xenophagy) (Reggiori et al. 2012).

1.1 The Autophagosomes

Autophagy is characterized by the formation of cytoplasmic double-membrane vesicles called autophagosomes with a diameter between 300 and 900 nm (Huang and Klionsky 2007) (Fig. 3.1). These carriers arise from a membranous cistern able to sequester cytoplasmic components while expanding. Complete autophagosomes subsequently fuse with lysosomes or plant and yeast vacuoles. In mammalian cells, this event is preceded by fusion with vesicles of the endocytic pathway and endosomes to form amphisomes (Hansen and Johansen 2011). During the fusion of autophagosomes with the lysosomes/vacuoles, the outer lipid bilayer of these carriers becomes part of the lysosome limiting membrane while the internal vesicles and their contents are exposed to lysosomal hydrolases and degraded. The metabolites resulting from this catabolic process are then recycled back to the cytoplasm by transporters present on the lysosomal membrane.

Fig. 3.1
figure 00031

Schematic representation of the proposed membrane sources for autophagosome biogenesis. All the organelles that have been implicated in providing membranes to one or more autophagosomal intermediates, i.e., phagophore or expanding phagophore, are represented in this draw. Continuous arrows highlight the existence of experimental data supporting the notion of a contribution of lipid bilayers whereas dashed arrows indicate a postulated or still to be firmly demonstrated involvement

Full size image

1.2 The Autophagosomal Precursor Structures

At the early stages of autophagosome formation, a portion of the cytoplasm is surrounded by a flat membrane sheet, which elongates by acquiring extra lipids and seals to sequester the cargo targeted for degradation (Fig. 3.1) (Tooze and Yoshimori 2010). This cistern has been called phagophore or isolation membrane. Phagophores appear to be formed to a particular location known as the phagophore assembly site or pre-autophagosomal structure (PAS) (Fig. 3.1). Most of the studies on the PAS have been done in the yeast Saccharomyces cerevisiae, where this structure is always found in proximity to the vacuole (Suzuki et al. 2001). The PAS is very likely an organizational site where probably a membrane acts at the very beginning as a docking platform for the hierarchical assembly of the Atg proteins (Yamamoto et al. 2012; Mari et al. 2010), the factors specifically involved in autophagosome biogenesis. Although recent studies in mammalian cells have indicated that a ­similar structure is present in high eukaryotes (Tooze and Yoshimori 2010), it remains largely unclear how the phagophores are generated. Two alternative mechanisms have been considered: The phagophore may be derived from a preexisting membrane (maturation model), or assembled from membranous constituents at the site of genesis (assembly model) (Juhasz and Neufeld 2006).

Although autophagy is highly conserved from yeast to mammals, there are some differences regarding the PAS. In yeast, only one PAS per cell is formed upon autophagy induction and consequently autophagosomes arise one after the other. In mammalian cells and in most of the other eukaryotic organisms, multiple PAS and thus autophagosomes are simultaneously generated throughout the cytoplasm before being transported via microtubule-associated motor proteins to the perinuclear region, where lysosomes are concentrating (Rubinsztein et al. 2005). The reasons of this difference remain unknown.

1.3 The Key Actors: The Atg Proteins

The breakthrough discovery in the field of autophagy came with the isolation of the strains with a defect in this pathway in the yeast S. cerevisiae and Pichia pastoris (Nakatogawa et al. 2009). The cloning of the mutated genes led to the identification of the autophagy-related (ATG) genes. Crucially, these genes were highly conserved from yeast to mammals, and this finding has provided the molecular tool to investigate autophagy in all eukaryotic organisms (Nakatogawa et al. 2009). Within all the genes that have been shown to be involved in the autophagy, 16 of them are part of what is considered to be the minimal core machinery required to mediate the formation of a double-membrane vesicle (Nakatogawa et al. 2009). Almost all of these core Atg proteins are cytoplasmic and associate to autophagosomal membranes upon autophagy induction. This event appears to occur in a hierarchical manner leading to the organization and formation of the PAS (Suzuki et al. 2007; Itakura and Mizushima 2010). Although the precise molecular role of the Atg proteins in rearranging, fusing and expanding autophagosomal membranes remains largely mysterious, they have been classified into five functional groups: (1) The Atg1/Ulk kinase complex, (2) the Atg9 cycling system, (3) the autophagy-specific phosphatidylinositol 3-kinase (PtdIns3K) complex I, and (4 and 5) the Atg12 and Atg8/LC3 conjugation systems (see Chap. 2 of this book for details). While the current model is that there is a direct sequential order in the assembly and interaction of these functional groups at the PAS, recent evidences indicates that some of them could independently associate with this specialized site (Kageyama et al. 2011).

2 Where Are Autophagosomes Originating From?

2.1 The Challenge in Solving the Enigma of the Origin of the Autophagosomal Membranes

Despite the advances in understanding the molecular mechanisms of autophagy, the origin of the membranes composing autophagosomes remains largely mysterious. Since the late 1950s, when morphologists working in mammalian cells first recognized autophagosomes as a unique compartment related to lysosomes and de Duve coined the term autophagy, there have not been specific molecular markers to study autophagosome biogenesis until the 1990s, when genetic screens in yeast led to the isolation of the ATG genes. Pioneering studies using standard biochemical techniques such as subcellular fractionation did not allow the identification of the membrane source because autophagosomes contents reflect the composition of the cytoplasm, making difficult the enrichment of a specific marker protein. Additionally, phagophores as well as autophagosomes have a relatively protein-poor membrane, which makes difficult to detect marker proteins of other cellular compartments and thus determine their origin (Arstila and Trump 1968; Yokota 1993). While some studies have localized marker proteins of compartments such as the ER, Golgi, or mitochondria, to the isolation membrane and/or autophagosomes using these approaches, others failed to detect them (Reggiori 2006).

Another approach used to identify the source of the autophagosomal membranes has been the attempt to localize autophagosome protein markers to a specific subcellular compartment. Identification of the Atg proteins raised expectations. Most Atg proteins, however, are only transiently associated with the phagophores and/or autophagosomes (Chap. 2), and do not localize to other cellular compartments. Because of their immersion into lipid bilayers, transmembrane proteins are an optimal tool to follow membrane dynamics. To date, only two integral membrane proteins essential for autophagy, Vacuole Membrane Protein 1 (VMP1) and Atg9, have been identified and while Atg9 is present in all eukaryotes, VMP1 is only found in high eukaryotes, from worms to mammals (Tooze 2010). The study of both of them, however, has presented some challenges (Sects. 2.2 and 2.4).

All these practical difficulties made and make the determination of the origin of the autophagosomal membranes an extremely challenging task. Nonetheless, the biogenesis of autophagosomes as well as the molecular function of the Atg proteins, some of which are probably involved in the delivery and assembly of the lipid bilayers composing autophagosomes, remains a fundamental knowledge to be understood. Key proteins acting at the first stages of autophagosome formation are involved in several physiological and pathophysiological processes. As a result, the Atg proteins are the principal candidates for the development of new therapies based on the modulation of autophagy. Thus during the last decade, an important effort has been made to try to find the origin of the lipid bilayers composing autophagosomes using advanced technologies like live-cell imaging and electron tomography (Tooze and Yoshimori 2010; Reggiori and Tooze 2009). Here we review the past and the more recent experimental evidences that have led to the implication of various subcellular organelles as the potential source of autophagosomal membranes (Fig. 3.1).

2.2 The Endoplasmic Reticulum

One of the first experimental evidences connecting the ER and autophagosomes was provided by the laboratory of Bill Dunn in 1990. Using morphological techniques combined with immunological reactions, the presence of organelle-specific proteins was explored on the limiting membranes of nascent or newly formed autophagosomes in amino acid-starved rat livers (Dunn 1990). Protein markers of the rough ER were detected on these intermediates while those of the Golgi, plasma membrane, and endosomes were absent. Other studies where rat liver membranes were fractionated showed that the fractions enriched in autophagosomal membranes contained marker proteins of the ER but not of the Golgi (Furuno et al. 1982; Kominami et al. 1983; Ueno et al. 1991). This notion was later indirectly supported by the observation that yeast mutants with a defect in COPII-mediated transport out of the ER display an impairment in autophagy (Ishihara et al. 2001), something also supported by studies in infected cells. Specifically, the Listeria monocytogenes-­containing compartment present in the cytoplasm of infected macrophages, which appears to have an autophagosomal origin, is positive for the rough ER marker protein disulphide isomerase (PDI) (Rich et al. 2003). Moreover, the expression of poliovirus proteins in COS-1 cells induced the formation of double-membrane vesicles from the ER that morphologically resemble autophagosomes (Suhy et al. 2000).

The study of Double FYVE domain-Containing Protein 1 (DFCP1), a PtsIns3P-­binding protein, provided the first convincing evidences that ER could be the source for autophagosomal membranes. DFCP1 localizes to the ER and Golgi in fed cells and upon amino acid deprivation, it translocates to cytosolic punctate structures, which are LC3- and ATG5-positive (Axe et al. 2008). This redistribution requires direct PtsIns3P recognition as well as proteins such as BECLIN 1 and PI3KC3 composing the autophagy-specific PtdIns3K complex responsible for PtdIns3P production. Importantly, live-cell imaging studies have shown that these punctate structures, formed 30 min after autophagy induction, are ring-shape membranous protrusions surrounding the autophagosomal marker protein LC3. Because they are frequently seen in association with the underlying ER assuming an Ω-like conformation, the authors named them omegasomes. Live-cell imaging experiments revealed the intimate connection between omegasomes and nascent autophagosomes (Axe et al. 2008). At the early stages of the formation of an autophagosome, a small amount of DFCP1 concentrates at the edge of an ER strand. As the DFCP1-positive region gets enlarged, LC3 begins to accumulate in very close proximity to it. Subsequently, the omegasome extends and fully encircles the LC3-positive central precursor structure before the LC3-positive autophagosome exits the omegasome and all the DFCP1 is reabsorbed into the ER. Recently, more ATG proteins have been localized to the omegasomes (Matsunaga et al. 2010; Polson et al. 2010). Together, all these data point to the omegasomes being the site where at least a subset of PAS and phagophores are formed (Fig. 3.2).

Two electron tomography-based studies have reinforced the idea that the omegasomes are specialized ER domains where autophagosomes are generated. In a first work, the three-dimensional reconstructions of areas with nascent autophagosomes have uncovered that the ER cisterns are positioned in parallel to the phagophores, inside and outside (Yla-Anttila et al. 2009) (Fig. 3.2). In addition, these ­tridimensional projections have revealed that the ER cisterns in the interior of the phagophore extend into the cytoplasm through the open end of the phagophore. In addition, they have also shown that the ER and the phagophore membranes are connected by narrow points of contacts suggesting the possibility that there is a lipid transfer between these two organelles (Yla-Anttila et al. 2009). A second investigation using the same ultrastructural approach and taking advantage of an inactive mutant of ATG4B, which causes the accumulation of phagophores, reached the same conclusions (Hayashi-Nishino et al. 2009). The authors also showed that the connections between the ER and the phagophore disappear when this later structure becomes an autophagosome. Finally, immuno-electron microscopy analysis demonstrated that these associations between the ER and nascent autophagosomes are omegasomes because positive for DFCP1 (Hayashi-Nishino et al. 2009).

Fig. 3.2
figure 00032

The putative mechanism for omegasome biogenesis. (a) Omegasome formation. Upon autophagy induction, one of the events occurring is the association of the PtdIns3K complex (composed by Beclin 1, Atg14L, hVps15 and hVps34) with membranes of the rough ER very likely through Atg14L. At this location, this complex generates PtdIns3P, a phosphoinositide that plays a key role in triggering the recruitment of several additional Atg proteins, including DFCP1 and the members of the WIPI protein family, which directly bind to this lipid. (b) Autophagosome biogenesis from omegasomes. At the omegasome, the phagophore is derived from the ER (step 1) and subsequently extends in between two ER cisterns by probably acquiring extra lipids through contact sites with the ER (step 2). The fusion of the two extremities of the expanding phagophore leads to its detachment from the ER and formation of an autophagosome (step 3). Adapted from Tooze and Yoshimori (2010)

Full size image

All these evidences indicating that the ER could be the principal origin of the autophagosomal membranes have also been highlighted by a study about ATG14, the autophagy-specific subunit of the PtdIns3K complex I (Itakura et al. 2008; Matsunaga et al. 2009). While PtdIns3P is thought to be restricted to endocytic compartments and being absent in the ER (Gillooly et al. 2000), omegasomes are PtdIns3P-enriched membranes (Axe et al. 2008). The work of Yoshimori and colleagues demonstrated that ATG14, a protein essential for omegasome formation, is the molecular connection between PtdIns3P and the ER. ATG14 exhibits both a punctate pattern and ER localization. The ATG14 puncta colocalize with marker proteins of the autophagosomal membranes, including DFCP1, indicating that those are autophagosomal intermediates. The formation of the ATG14-positive puncta is induced under autophagy conditions supporting a notion where ATG14, which constitutively resides in the ER, concentrates to specific sites on this organelle from which the phagophores/omegasomes will emerge. In agreement with this model, ATG14 knockdown impaired the formation of the DFCP1-positive omegasomes. Furthermore, ATG14 is essential for the recruitment of the PtdIns3K complex I to the ER, a crucial event for autophagosome formation either in basal or starvation conditions. Conversely, overexpression of ATG14 increases the amount of PtdIns3P-­positive puncta while the overexpression of a mutant form unable to localize to the ER does not (Matsunaga et al. 2010). ATG14 thus appears to provide the landmark for the recruitment of the PtsIns3K complex I and the local production of PtdIns3P essential for the recruitment of its effectors that lead to subsequent events required to form autophagosomes from the ER. On the same line, VMP1, a transmembrane protein only present in high eukaryotes and also essential for autophagosome biogenesis (Tian et al. 2010), can be found in the ER under certain circumstances (Dusetti et al. 2002) and it co-localizes with ULK1 and ATG14 upon autophagy induction (Itakura and Mizushima 2010).

2.3 Mitochondria

In a recent study, the group of Lippincott-Schwartz has provided evidences for a possible direct link between mitochondria and autophagosomes biogenesis under starvation conditions in mammalian cells (Hailey et al. 2010). They found colocalization between a fluorescence chimera targeted to the outer mitochondrial membrane and the autophagosomal protein markers LC3 and ATG5. Moreover, electron micrographs showing an association between mitochondria and autophagosomes, and fluorescence photobleaching techniques to study the dynamics distribution of fluorescence reporter proteins between these two organelles, revealed a membrane continuity between mitochondria and nascent autophagosomes (Hailey et al. 2010). As a result, it was proposed that these structural connections are key in the transfer of lipids required to support the phagophore expansion.

Conjugation of LC3 to PE is essential in the process leading to the biogenesis of an autophagosome (Mizushima et al. 2001; Nakatogawa et al. 2007) and mitochondria are one of the main locations where PE is synthesized via phosphatidylserine (PS) decarboxylation (Vance 2008). PS is principally synthesized in the mitochondria-­associated microdomains (MAMs), the interaction sites between the ER and ­mitochondria (Stone and Vance 2000). The rate-limiting step in the conversion of PS into PE is the transport of newly synthesized PS to mitochondria. Although the mechanism of this lipid translocation remains unknown, it is very likely to occur through the MAMs (Vance 2008). In yeast, components of these ER–mitochondria interaction sites, also known as the ER–Mitochondria Encounter Structures (ERMES), are functionally connected to phospholipid biosynthesis (Kornmann et al. 2009). It has been found that one protein involved in the regulation of mitochondrial dynamics, Mitofusin 2 (MFN2) is also mediating the tethering between mammalian ER and mitochondria but it remains unknown whether this protein is part of the ERMES (de Brito and Scorrano 2008). Nevertheless, cells where MFN2 has been depleted display a severe defect in autophagosome formation supporting a model where mitochondria may provide at least part of the autophagosomal lipids (Hailey et al. 2010).

2.4 The Golgi Complex

When analyzing the morphology and formation of the protein granules in the fat body cells of the butterfly Calpodes ethlius, Locke and Collins observed isolation membranes derived from the Golgi surrounding the protein granules targeted to degradation, which were finally leading to the formation of autophagosome-like compartments (Locke and Collins 1965). Later, the concept of the Golgi contributing to autophagy was reinforced by ultrastructural studies showing that the growing extremities of the phagophore and a section of the complete autophagosome can be decorated with lectins that recognize glycans exclusively present in post-Golgi membranes (Yamamoto et al. 1990). More recently, various molecular components of the Golgi have been linked to the process of autophagy further supporting the notion that this organelle could be involved in supplying at least part of the membranes composing autophagosomes.

The late compartments of the Golgi system have been linked to autophagosome biogenesis in various ways. In yeast, the late Golgi guanosine exchange factor (GEF) Sec7 and its downstream Arf GTPases are indispensable for autophagy (van der Vaart and Reggiori 2010; Reggiori et al. 2004a). Inactivation of Sec7 does not impair the formation of the PAS and phagophore but rather the expansion of this precursor structure suggesting a role of the Golgi in providing lipid bilayers required for the completion of autophagosomes (van der Vaart and Reggiori 2010). Another yeast GEF protein, Sec2, and its effector, the Rab GTPase Sec4, which are associated with secretory vesicles generated from the Golgi, are also essential for autophagy (Geng et al. 2010). The authors of this later work hypothesized that some components at the trans-Golgi network (TGN) such as Sec2 and Sec4 could redirect the membrane flow from the secretory pathway to the autophagosome biogenesis during autophagy-inducing conditions (Geng et al. 2010).

While under normal growth conditions, the small GTPase RAB33B is present in the cis-Golgi and functions in the Golgi-to-ER retrograde transport, starvation conditions trigger its recruitment to LC3- and ATG16L1-positive autophagosomal membranes where it is in association with the ATG12–ATG5–ATG16L1 complex through direct binding to ATG16L1 (Itoh et al. 2008). This interaction appears to be involved in modulating the fusion of autophagosomes with lysosomes, and it also requires the activity of OATL1, a Rab GTPase-activating protein (GAP) specific for RAB33B, to be recruited to autophagosomal membranes via the binding to LC3 (Itoh et al. 2011). It remains to be established whether RAB33B modulates the assembly of Golgi-derived membranes with the autophagosomal ones.

RAB24 localizes to the ER, cis-Golgi and the ER–Golgi intermediate ­compartment in presence of nutrients but when those are removed, it relocalizes to autophagosomes labeled with both LC3 and the dye monodansylcadaverine (Munafo and Colombo 2002). Further studies will be necessary to precisely define the role of this component and allow to uncover whether it is related with the Golgi membranes transport to form the autophagosome. In this situation as well, it is unclear whether the change in RAB24 subcellular distribution reflects a variation in the direction of the membrane flux through or from the Golgi.

Finally, it has recently been discovered that the Golgi transmembrane protein Ema and the peripheral membrane protein Lva are associated with autophagosomes generated in response to starvation in Drosophila fat body cells (Kim et al. 2012). In absence of Ema, autophagosome are still formed but their size is strongly reduced suggesting that Golgi membranes are necessary for the phagophore elongation in this tissue (Kim et al. 2012).

Indirectly, the Golgi is certainly involved in supplying at least part of the lipid bilayers composing autophagosomes. The Atg9-positive membranes are playing a pivotal role in the formation of the PAS (Yamamoto et al. 2012; Mari et al. 2010; Noda et al. 2000; Orsi et al. 2012) and because these structures cycle to and from this site (Reggiori et al. 2004b), it has been also hypothesized that the Atg9-­containing membranes could relevantly contribute to the autophagosome biogenesis. While this could be true in high eukaryotes because ATG9 dynamically associate and dissociate from the sites where autophagosomes are formed (Orsi et al. 2012), in yeast it appears that the initial pool of Atg9 forming the PAS is retrieved only when autophagosomes are completed (Yamamoto et al. 2012). Importantly, the Atg9-containing membranes are derived from the Golgi in yeast (Yamamoto et al. 2012; Mari et al. 2010; Ohashi and Munro 2010) and part of them are associated with the TGN in mammals (Orsi et al. 2012; Young et al. 2006). Thus, it cannot be excluded that some of the Golgi proteins that have been implicated in autophagosome biogenesis mediate the delivery of Atg9-containing membranes to the nascent autophagosomes. For example, Ypt1 and its autophagy-specific GEFs, i.e., the TRAPIII complex, are thought to be involved in the tethering of the Atg9 containing structures with the PAS, as Ypt1 was shown to colocalize with Atg9-containing membranes and the ATG9 deletion decreases the colocalization degree between Trs85 and Ypt1 (Lynch-Day et al. 2010). Another observation supporting this notion was the impairment of Atg9 containing membranes anterograde movement to the PAS, in yeast, in sec2 and sec4 mutants (Geng et al. 2010). Finally Bif-1, a protein interacting with Beclin 1 and essential for autophagy (Takahashi et al. 2007), appears to regulate Atg9 trafficking by mediating the fission of Golgi membranes during autophagy (Takahashi et al. 2011).

2.5 The Endosomes

The fusion of early endosomes with early autophagosomal intermediates was firstly observed in the 1990s in exocrine pancreas cells (Tooze et al. 1990). Later this event was shown to take place at different stages of the endocytic and autophagic pathways, but it occurs predominately with initial autophagosomal structures at least in hepatocytes (Liou et al. 1997). Recently, Longatti and colleagues have found that ULK1 localizes to recycling endosomes positive for both RAB11, a Rab GTPase essential for the regulation of the recycling of the endocytosed proteins, and the transferrin receptor (TfnR). Additionally they showed that RAB11 and TBC1D14, which acts as a RAB11 effector, direct the recycling endosomes to merge with nascent autophagosomes labeled with LC3 (Longatti et al. 2012). Interestingly, ATG9 is also found in the recycling endosomes, where it interacts with the TfnR (Longatti et al. 2012). Together, these data indicate that the recycling endosomes play a key role in the early events of the PAS and/or phagophore formation.

RAB5, a small GTPase that acts on the regulation of the early endocytic pathway in mammalian cells, is an activator of PI3KC3 by being part of the complex that comprises PI3KC3 and BECLIN 1 that localizes to ATG5-positive autophagosomal precursors. The important role of this protein is also elicited by the fact that inhibition of RAB5 activity results in both a decrease in the formation of LC3-positive autophagosome and a concomitant accumulation of ATG5-positive autophagosomal precursors, supporting the idea that early endosomal functions may be connected with the biogenesis of autophagosomes (Ravikumar et al. 2008).

2.6 The Plasma Membrane

The plasma membrane (PM) has recently been shown to contribute to the formation of early autophagosomal intermediates. In particular, ATG16L1 was shown to be present in vesicles derived from the PM, positive for other early autophagosomal marker proteins such as ATG5 and ATG12, and therefore considered to represent autophagosomal precursor structures. These vesicles are formed by clathrin coat-mediated endocytosis through a process that requires the small G protein ARF6 and the local generation of phosphatidylinositol-4,5-biphosphate, and they mature into autophagosomes through a mechanism that remains to be characterized (Ravikumar et al. 2010; Moreau et al. 2012). Interestingly, the same laboratory has also revealed that these ATG16L1-postive vesicles must undergo SNARE-mediated homotypic fusion to generate what appears to be a successive autophagosomal precursor of larger size (Moreau et al. 2011). The authors hypothesized that the PM contribution to the autophagosome biogenesis may be crucial especially during high autophagy activity, because the PM surface could represent an important reservoir to avoid interfering with processes carried out by other potential membrane source compartments.

3 How to Rationalize All the Findings

In this chapter we have reviewed the different hypothetical origin of the lipid bilayers composing autophagosomes. While a lot of effort has been invested in solving this central question in the field of autophagy, the fact that several organelles have been implicated in being the source of the autophagosomal membranes has create somehow confusion. Are all these observation in contradiction? Obviously, all the studies supporting that a specific organelle provides the membranes necessary for the formation of the autophagosomes have to be the subject of additional examinations but it could also be possible that there are in fact multiple origins.

One potential scenario could be that organelles such as the plasma membrane, the endosomes, and the Golgi (or Golgi derived Atg9-containing membranes) all provide membranes to form the initial PAS and/or phagophore whereas other organelles like the ER and the mitochondria provides the extra lipids necessary to expand the phagophore into an autophagosome. This notion is supported by the observation that at least in certain situations, clusters of Atg proteins are independently recruited and assembled to the site where the PAS and/or an early autophagosomal precursor structure will be formed (Kageyama et al. 2011; Itakura et al. 2012; Noda et al. 2012).

Another possibility could be that the membrane source could vary depending on the cellular cue or stress inducing autophagy (Mari et al. 2011). The observed differences could also reflect tissue- and organism-specific diversities. On this line, cells could also draw from one or more additional membrane reservoirs when there is need to sustain an intense autophagy activity or generate autophagosomes of huge dimensions like during the invasion of specific bacteria such as Streptococcus aureus (Nakagawa et al. 2004).

Finally, the functional connections between autophagosomes and a specific organelle could also be dictated by the type of selective autophagy that is triggered. For example, one would imagine that the cell is not using membranes from damaged organelles that have to be turned over by autophagy.

The mystery of the origin of the autophagosomal membranes has intrigued researchers working in the field of autophagy for decades. The recent data have revealed the complexity of this issue in its entirety and highlighted the necessity of additional investigations. These future studies will also be pivotal in helping us understanding better the regulation and mechanism underlying autophagosome biogenesis.