Correspondence Isei Tanida, Department of Biochemistry and Cell Biology, National Institute of Infectious Diseases, 1-23-1, Toyama, Shinjyuku, Tokyo 162-8640, Japan. Tel: +81 3 5285 1111 (ext. 2126), fax: +81 3 5285 1157; email: email@example.com
Autophagy (macroautophagy) is a dynamic process for degradation of cytosolic components. Autophagy has intracellular anti-viral and anti-bacterial functions, and plays a role in the initiation of innate and adaptive immune system responses to viral and bacterial infections. Some viruses encode virulence factors for blocking autophagy, whereas others utilize some autophagy components for their intracellular growth or cellular budding. The “core” autophagy-related (Atg) complexes in mammals are ULK1 protein kinase, Atg9-WIPI-1 and Vps34-beclin1 class III PI3-kinase complexes, and the Atg12 and LC3 conjugation systems. In addition, PI(3)-binding proteins, PI3-phosphatases, and Rab proteins contribute to autophagy. The autophagy process consists of continuous dynamic membrane formation and fusion. In this review, the relationships between these Atg complexes and each process are described. Finally, the critical points for monitoring autophagy, including the use of GFP-LC3 and GFP-Atg5, are discussed.
wild-type human microtubule-associated protein 1 light chain 3
soluble unlipidated form of LC3
monomeric red fluorescent protein
WD-repeat protein interacting phosphoinositides
The term “autophagy” is derived from the Latin words for “self” and “eating.” Macroautophagy (here referred to simply as “autophagy”) is essential for tissue and cell homeostasis, and defects in autophagy are associated with many diseases, including neurodegenerative diseases, cardiomyopathy, tumorigenesis, diabetes, fatty liver, and Crohn's disease (1–3). Autophagy is a bulk degradation system that accompanies the dynamic processes of omegasome formation, initiation and elongation of the isolation membrane, autophagosome formation, autophagosome-lysosome fusion, and degradation of intra-autophagosomal contents by lysosomal hydrolases (Fig. 1) (4–6). Autophagy has an intracellular anti-viral function, the targeting of viral components or virions to degrade them via the lysosomes during viral infection; it also plays a role in the initiation of innate and adaptive immune system responses to viral infections (7–12). Some viruses encode virulence factors that interact with the host autophagy machinery and block autophagy. In contrast, other viruses utilize some autophagy components to facilitate their intracellular growth or cellular budding.
Taking advantage of yeast genetics, autophagy-defective (atg/apg/aut) mutants of Saccharomyces cerevisiae were isolated in 1993 (the nomenclature of autophagy related genes has been unified to ATG) (13, 14). The ATG (A uT ophaG y-related) genes were later isolated and characterized (Table 1) (5, 13, 15). Most ATG genes contribute to autophagosome formation, many being well conserved from yeast to mammals. Although the molecular mechanisms and cellular functions of mammalian autophagy were being elucidated within a decade, our molecular understanding of autophagy is still far from complete. In this review, we describe the molecular mechanism of action of mammalian Atg proteins and their cellular functions in autophagy.
Table 1. Atg genes essential for “core” autophagy in mammals
In mammals, the “core” Atg proteins are divided into five subgroups: the ULK1 protein kinase complex (16), Vps34-beclin1 class III PI3-kinase complex (17), Atg9-WIPI-1 complex (18–20), Atg12 conjugation system (21, 22), and LC3 conjugation system (23, 24). Autophagy is impaired without any of these “core” Atg gene products, indicating that a sequential reaction of many protein complexes, including kinases, phosphatases, lipids, and ATP-dependent conjugation, are indispensable for the whole process of autophagy. Upstream of the autophagy machinery, class I PI3-kinase and mTor kinase contribute to the induction of autophagy (25). The Vps34-beclin1 class III PI3-kinase complex is divided into at least three types, the Atg14-Vps34-Vps15-beclin1, UVRAG-Vps34-Vps15-beclin1, and Rubicon-UVRAG-Vps34-Vps15-beclin1 complexes (26–29). Each complex contributes to a different function during autophagy. The Atg9-WIPI-1 complex is composed of an Atg9 membrane-protein and WIPI-1 (18, 30). Two ubiquitylation-like reactions, the Atg12 and LC3 conjugation systems, are essential for the initiation and formation of autophagosomes (Fig. 1, Initiation, elongation, and maturation).
ULK1 PROTEIN KINASE COMPLEX
The ULK1 protein kinase complex is composed of ULK1 (a protein kinase), Atg13, FIP200, and Atg101 (Fig. 1, Initiation) (16, 31–35). The mTOR kinase directly phosphorylates Atg13 to negatively regulate autophagy (33). Atg101 is important for the stability and basal phosphorylation of Atg13 and ULK1 (34, 35). FIP200 is important for the stability and phosphorylation of ULK1 (31). Considering that Atg13 is responsible for recruitment of Atg14 to the pre-autophagosomal structure in yeasts (36), it is possible that the ULK1-Atg13-FIP200-Atg101 complex interacts with the Atg14-Vps34 class III PI3-kinase complex in mammals.
VPS34-BECLIN1 CLASS III PI3-KINASE COMPLEX
The Vps34-beclin1 complex is a core complex of class III PI3-kinase (37). In mammals, at least three types of class III PI3-kinase complex contribute to autophagy (26–29, 38, 39). The Atg14-Vps34-Vps15-beclin1 complex is essential for autophagosome formation (Fig. 1, Initiation and elongation), and the UVRAG-Vps34-Vps15-beclin1 complex functions positively in autophagosome maturation and endocytic traffic (Fig. 1, Autophagosome-lysosome fusion) (27, 39). In contrast, the Rubicon-UVRAG-Vps34-Vps15-beclin1 complex negatively regulates autophagosome-maturation and endocytic traffic (Fig. 1, Autophagosome-lysosome fusion) (28). Ambra1, a protein containing a WD40 domain that activates beclin1-regulated autophagy, regulates autophagy and has a crucial role in embryogenesis (40). In sensory neurons, Vps34-independent autophagy has been reported as a non-canonical autophagy pathway (41).
Based on the findings in yeast, the Atg9-WIPI-1 complex is considered to be composed of Atg9, hypothetical Atg2 and WIPI-1 (PIP-binding protein) in mammals. Atg9 is the only integral membrane protein in yeasts (42, 43); its mammalian homologs are Atg9/mAtg9/Atg9L1 (ubiquitous expression) and Atg9L2 (expressed specifically in the placenta and pituitary gland) (18). Under nutrient-rich conditions Atg9 is localized to the trans-Golgi network and partial endosomes, whereas under starvation conditions it is localized to autophagosomes in a process dependent on ULK1 (18). WIPI-1 is also localized to the autophagosome during autophagy (Fig. 1, Elongation) (20, 44). Atg18, a yeast homolog of WIPI-1, constitutively interacts with yeast Atg2 in yeasts, and yeast Atg9 interacts with the Atg2-Atg18 complex during autophagy (45). According to the findings obtained with the yeast Atg9-model, mammalian Atg9 may interact with the Atg2-WIPI-1 complex during autophagy. Atg27 is required for autophagy-dependent cycling of Atg9 in yeasts (46). No mammalian homologs of Atg2 and Atg27 have yet been identified.
ATG12 CONJUGATION SYSTEM: THE FIRST UBIQUITYLATION-LIKE REACTION
The Atg12 conjugation system, the first ubiquitylation-like reaction, is essential for formation and elongation of the isolation membrane (Fig. 1, Initiation and elongation, Atg12-Atg5-Atg16 complex) (47). Although the amino acid sequences of Atg12 and ubiquitin are dissimilar, Atg12 does possess a ubiquitin fold (21). In the Atg12 conjugation system, Atg12 is activated by Atg7, an E1-like enzyme; transferred to Atg10, an E2-like enzyme, and conjugated to Atg5 to form Atg12-Atg5 conjugates (Fig. 2, Wild-type Atg12 and Atg5) (21, 22, 48–50). As in ubiquitin, the carboxyl terminal Gly of Atg12 is essential for the formation of thioester bonds with the active site Cys residues of Atg7 and Atg10, and is also essential for the formation of amide bonds with the Lys130 residues in Atg5 (21, 48, 49, 51–53). Therefore, Atg12 is a modifier that has a structural ubiquitin fold. Atg16 interacts with Atg5, forming a multimeric complex (54–56). In many tissues and cell lines, most endogenous Atg5 and Atg12 are present as the Atg12-Atg5 conjugate, and little increase in the amount of Atg12-Atg5 conjugate is observed during autophagy.
LC3 LIPIDATION: SECOND UBIQUITYLATION-LIKE REACTION
The second ubiquitin-like conjugation system, the LC3 conjugation system, is unique in that its target is a phospholipid, PE (23, 24). Therefore, the LC3 conjugation system has been called LC3-lipidation. To date, at least four mammalian Atg8 homologs have been identified: LC3/MAP1-LC3/LC3B (microtubule-associated protein 1 light chain 3), GABARAP, GATE-16, and Atg8L (4, 57). LC3 is the best characterized of these proteins, and LC3-II is regarded as a promising autophagosome marker (Fig. 1, Maturation, LC3-II) (23). LC3 is synthesized as proLC3, which is cleaved by Atg4B to form LC3-I, with the carboxyl terminal Gly exposed (Fig. 2, Wild-type LC3) (23). LC3-I is activated by Atg7, transferred to Atg3, and finally conjugated to PE (51, 58). The carboxy-terminal Gly of LC3 is also essential for the formation of a thioester bond with the active site Cys residues of Atg7 and Atg3, and for the formation of an amide bond with PE (59, 60). With regard to GABARAP, GATE-16, and mAtg8L, the reactions mediated by Atg7 and Atg3 are similar to those of LC3. Both these Atg8 homologs and yeast Atg8 also have a ubiquitin fold, as is the case with Atg12, however their amino acid sequences are dissimilar from those of Atg12 and ubiquitin. Therefore, these Atg8 homologs are second modifiers activated by Atg7 and Atg10. Because LC3-I is localized in the cytosol and LC3-II to autophagosomes (Fig. 1, Elongation and maturation) (23), LC3-II is a promising autophagosomal marker in mammals. LC3-II on the cytoplasmic surface of autophagosomes is delipidated by Atg4B to recycle LC3-I for further autophagosome formation (Fig. 1, Autophagosome-Lysosome fusion). In contrast to what occurs with Atg12-Atg5 conjugate, the amount of endogenous LC3-II changes during autophagy.
RELATIONSHIP BETWEEN ATG12 CONJUGATION AND LC3 LIPIDATION
Atg12 conjugation is closely related to LC3 lipidation. Atg5 deficiency results in a defect in LC3 lipidation (47, 61). The yeast Atg12-Atg5 conjugate functions in vitro as an E3-like enzyme for Atg8 lipidation (62). Mammalian Atg16L determines the site of autophagosome formation (63). Therefore, the Atg12-Atg5-Atg16 complex may function as an E3-ligase complex to facilitate LC3 lipidation complex (Fig. 2, dashed arrow). Lack of Atg3 in mammals leads to a decrease in the Atg12-Atg5 conjugate as well as impairing LC3 lipidation (64), and is associated with defective autophagosome formation, including defects in elongation and complete closure of the isolation membranes, resulting in malformed autophagosomes. In cells lacking Atg3, Atg16L and Atg5 are localized to elongated isolation membranes/incomplete autophagosomes, suggesting that elongation of the isolation membrane can occur in the absence of LC3 lipidation (64). Thus, in addition to its potential E3-like function, the Atg12-Atg5-Atg16 complex may function in the elongation of isolation membranes.
INITIAL STEP OF AUTOPHAGY (LC3-II INCREASING STEP): OMEGASOME AND AUTOPHAGOSOME FORMATION
Autophagy is divided into six steps; omegasome formation, initiation of isolation membranes, elongation of the isolation membrane, autophagosome formation, autophagosome-lysosome fusion, and degradation (Fig. 1). The ULK1-protein kinase complex activates autophagic signaling via the mTor-signaling pathway when autophagy is induced (Fig. 1, Initiation) (33, 32). The omegasome, which is shaped like the Greek letter omega (Ω), is first formed from the ER. A PI(3)P-binding protein, DFCP1, is localized to PI(3)P on the omegasome under starvation conditions (Fig. 1, Initiation, DFCP1), but localizes to the ER and Golgi under nutrient-rich conditions. The Atg14-Vps34-beclin1 PI3-kinase complex positively regulates DFCP1-positive omegasome formation (Fig. 1, Initiation, omegasome) (65). After omegasome formation, the isolation membrane (also called the pre-autophagosome or phagophore) is formed inside the ring of the omegasome (Fig. 1, Initiation, isolation membrane), and the Atg12-Atg5-Atg16 complex is localized to the isolation membrane (Fig. 1, Elongation, Atg12-Atg5-Atg16 complex) (47, 54, 55). The protein Atg9, WIPI-1, the ULK1 protein kinase complex, and the Atg14-Vps34-beclin1 PI3-kinase complex are also localized to the isolation membrane (Fig. 1, Elongation). DFCP1 itself, however, is probably not required for autophagosome formation. Two PI(3)P-phosphatases (Jumpy [also known as MTMR14] and MTMR3) negatively regulate formation of the omegasome and the isolation membrane (Fig. 1, Elongation) (66, 67).
The Atg12-Atg5-Atg16 complex-localized isolation membrane elongates to engulf cytoplasmic components. In the later stages of isolation membrane elongation, the Atg12-Atg5-Atg16 complex progressively dissociates from the isolation membrane, whereas LC3-II is gradually localized to both sides of this membrane (Fig. 1, Elongation) (47). Finally, the isolation membrane closes to form the autophagosome (Fig. 1, Maturation). While LC3-II is localized to autophagosomes, most of the Atg12-Atg5-Atg16 complex dissociates from the autophagosome (47). During this process, LC3-II is increased. Rab32 and Rab33B also contribute to elongation of the isolation membrane (68, 69). Alfy, a PI(3)P-binding FYVE domain-containing protein, has been found to localize with autophagosomes and protein granules (70). Functional multivesicular bodies are required for Alfy-mediated clearance of protein aggregates via autophagy (71).
SUBSEQUENT STEP OF AUTOPHAGY (LC3-II DECREASING STEP): AUTOPHAGOSOME-LYSOSOME FUSION AND DEGRADATION
Soon after autophagosome formation, its outer membrane fuses with the lysosome to form the autolysosome, a process requiring Rab7 (Fig. 1, Autophagosome-lysosome fusion) (72, 73). Following autolysosome formation, Atg4B delipidates LC3-II on the cytosolic surface to recycle LC3-I (Fig. 1, Autophagosome-lysosome fusion, Atg4B). The FYVE and coiled-coil domain-containing protein FYCO1 functions as a Rab7 effector, binding to LC3 and PI3P and mediating microtubule plus end-directed vesicle transport (74). The fusion of autophagosomes and lysosomes is positively regulated by the UVRAG-Vps34-beclin1 PI3-kinase complex and negatively regulated by the Rubicon-UVRAG-Vps34-beclin1 PI3-kinase complex (Fig. 1, Autophagosome-lysosome fusion) (26–29, 38).
Following autolysosome formation, the lysosomal hydrolases, including cathepsins, lysosomal glycolytic enzymes, and lipases, degrade the intra-autophagosomal contents. In this step cathepsins degrade LC3-II on the intra-autophagosomal surface (Fig. 1, Degradation) (75, 76). In yeasts, Atg15, a vacuolar lipase, and Atg22, a vacuolar membrane protein, are indispensable for the specific degradation of autophagic bodies (77–79). No mammalian homologs of yeast Atg15 and Atg22 have yet been identified. During conversion by Atg4B of LC3-II to LC3-I on the cytoplasmic face of the autophagosome and degradation by lysosomal hydrolases of LC3-II on the luminal face of autophagosome, LC3-II decreases.
After digestion of intra-autophagosomal contents, a lysosomal-associated membrane protein 1 -positive and LC3-negative tubular structure, the protolysosome, is elongated from the autolysosome (Fig. 1, Protolysosome) (80). The protolysosome finally forms a vesicle, and matures into the lysosome by accumulating of lysosomal hydrolases.
ESTIMATION OF AUTOPHAGY: YIN AND YANG
It is necessary to estimate autophagic activity accurately and quantitatively when studying autophagy in infection and immune responses. LC3-II and LC3-positive puncta are recognized as promising autophagosome and autolysosome markers (but not “autophagy” markers). However, autophagosomes and autolysosomes are transient structures during autophagy. Therefore, the amount of LC3-II (or number of LC3-positive puncta) alone does not always reflect autophagic activity. Production of LC3-II is increased when autophagy is activated (Fig. 1, Maturation), in addition lysosomal degradation of LC3-II and delipidation of LC3-II by Atg4B are simultaneously activated (Fig. 1, Autophagosome-lysosome fusion). Many methods for monitoring autophagy, including GFP-LC3, tf-LC3, and LC3-II turnover assay, have been proposed, these have both advantages and disadvantages. Recently, critical issues and guidelines for monitoring autophagy have been described (81–83).
LC3 FUSED TO GREEN FLUORESCENT PROTEIN AND ITS NEGATIVE CONTROL FOR ESTIMATION OF AUTOPHAGY
LC3 fused to green fluorescent protein is useful for in vivo imaging of autophagosome formation (84, 85). However, caution must be exercised due to the limitations of GFP-LC3 (86, 87). GFP-LC3 tends to form puncta in cells independent of autophagy, and GFP fluorescence in lysosomes may occur even after degradation of the LC3 moiety. Therefore, this method tends to overestimate the number of autophagosomes. These problems may be avoided by using a mutant, GFP-LC3ΔG which lacks the essential carboxy-terminal Gly of LC3, as a negative control (Fig. 2, LC3ΔG). GFP-LC3 transgenic mice, however, can be used to study autophagy in many tissues outside the brain (84, 88).
ATG5, ATG12, AND NEGATIVE CONTROLS FOR STUDYING THEIR FUNCTIONS IN AUTOPHAGY
Endogenous Atg5 and Atg12 are mainly present as the Atg12-Atg5 conjugate, this conjugate being essential for autophagy. Therefore, when Atg5 and Atg12 are analyzed using an expression plasmid(s), negative controls should be used. The Lys130 within human Atg5 is essential for Atg12 conjugation (Fig. 2, Wild-type Atg12 and Atg5). An Atg5K130R mutant, in which essential Lys130 has been changed to Arg, has a defect in conjugate formation resulting in a defect of autophagosome formation (Fig. 2, Atg5K130R) (47). Therefore, mutant Atg5K130R is suitable as a negative control for Atg5. The carboxy-terminal Gly within Atg12 is also essential for formation of the Atg12-Atg5 conjugate. A mutant Atg12ΔG lacking the carboxy-terminal Gly within Atg12 has defects in E1-like and E2-like reactions with Atg7 and Atg3, respectively (Fig. 2, Atg12ΔG) (58, 51). Therefore, mutant Atg12ΔG is also suitable as a negative control for Atg5. It is necessary to use these negative controls to clarify whether the functional interaction between Atg5 (or Atg12) and a target protein is related to the conjugate, that is, to autophagy.
TANDEM FLUORESCENT PROTEIN-LC3-COLOR CHANGE ASSAY
The mRFP-GFP-tandem fluorescent protein-LC3-color change assay is based on a difference between GFP and mRFP in pH stability (89, 90). Autophagosomes have a pH similar to that of the cytosol, while autolysosomes have an acidic pH. At an acidic pH, the fluorescence of mRFP is stable, while that of GFP decreases. Therefore, the merged color of mRFP-GFP-LC3 in autophagosomes is yellow, while that in autolysosomes is red (89). This assay is suitable for real-time (and short-term) monitoring of autophagy, but care should be taken when using it in long-term monitoring of this process. Fluorescence derived from GFP in the lysosomes has been observed even after degradation of LC3 (87).
LC3-PHOSPHOLIPID CONJUGATE TURNOVER ASSAY
The amount of LC3- II increases during autophagosome formation, an initial step in autophagy, while LC3-II decreases during autophagosome-lysosome fusion and degradation of intra-autophagosomal contents by lysosomal hydrolases. Therefore, it is difficult to judge whether a transient assessment of LC3-II by immunoblotting represents activation or impairment of autophagy. To resolve this issue, the LC3-II turnover assay, a measure of autophagic flux in which LC3-II is assayed by immunoblotting with anti-LC3 antibody in the presence and absence of lysosomal inhibitors, is employed (76). A mixture of E64d (a membrane-permeable inhibitor of cathepsins B, H, and L) and pepstatin A (a membrane-permeable inhibitor of cathepsins D and E) is used to inhibit lysosomal function (91). Treatment of cells with this inhibitor cocktail results in significant accumulation of autolysosomes (and LC3-II dots) because there is little degradation of their contents. Thus, the accumulation of LC3-II reflects the activity of the process of delivering LC3-II into lysosomes, that is, autophagic flux. Fluorescent microscopic assays of LC3-positive puncta in the presence and absence of the lysosomal inhibitor cocktail may also be useful. Methods for immunoblotting and immunostaining of endogenous LC3 have been described (76). Bafilomycin A1 (an inhibitor of V-ATPase) is also used to inhibit autophagy and to estimate the autophagic flux of LC3-II. As V-ATPase contributes to the acidification of other organelles, including the Golgi and endosomes, bafilomycin A1 may show multiple off-target effects (92, 93).
INCREASE OF UBIQUITYLATED PROTEINS AND A PROTEIN ESSENTIAL FOR SELECTIVE AUTOPHAGY, P62/SQSTM1, IN IMPAIRMENT OF AUTOPHAGY
p62 has ubiquitin-binding and LC3-binding domains, and binds to ubiquitylated protein aggregates to degrade them selectively via autophagy (94–96). When autophagy is impaired, p62 increases in cells and tissues (94, 97). At the same time, ubiquitin-positive aggregates accumulate. Ubiquitin-positive and p62-positive aggregates are observed in brains in some neurodegenerative diseases and in other autophagy-defective tissues. Therefore, accumulation of p62 and ubiquitin-positive proteins suggests the possibility of impairment of autophagy.
A DOMINANT NEGATIVE ATG4B MUTANT FOR INHIBITION OF AUTOPHAGY
Atg4B is a cysteine protease which is essential for conversion of proLC3 to LC3-I and for delipidation of LC3-II (Figs 1 and 2) (98). A mutant Atg4BC74A, in which the active site Cys74 is changed to Ala, produces defects in conversion and delipidation (Fig. 2, Atg4BC74A) (99, 100). Because overexpression of the mutant Atg4BC74A results in inhibition of LC3 lipidation, that is, in autophagy, the mutant is employed as a dominant negative mutant.
Autophagy is a bulk process of degradation of cytoplasmic components, including organelles. The pathophysiological functions of autophagy are becoming clear; however, our understanding of autophagy machinery, and methods for monitoring autophagy, are somewhat less than perfect. We have reviewed both the “core” Atg complexes essential for autophagosome formation, and assays of autophagy. Mammalian cells have mammalian-specific Atg proteins and more complicated mechanisms than yeast, probably because mammalian cells utilize autophagic machinery for tissue- and cell-specific functions as well as for self defense mechanisms against intracellular and extracellular stresses. In addition to so called “autophagy” as a non-selective function, the presence of selective autophagy has been reported; mitophagy is a type of autophagy specific for degradation of mitochondria, reticulophagy for the endoplasmic reticulum, ribophagy for ribosomes, piecemeal autophagy for the nucleus, and xenophagy for pathogens. Selective autophagy-specific genes are now being isolated and characterized. For future clinical applications based on autophagy, it will be necessary to screen for compounds which inhibit or activate autophagy.
This study was supported in part by grants-in-aid from the Ministry of Health, Labor, and Welfare of Japan, and by Grants-in-Aid for Scientific Research on Priority Areas “Proteolysis in the Regulation of Biological Processes” from the Ministry of Education, Science, Sports and Culture of Japan.