Beginning to Understand Autophagy, an Intracellular Self-Degradation System in Plants

Abstract

Autophagy is an evolutionarily conserved intracellular process for the vacuolar degradation of cytoplasmic components. There is no doubt that autophagy is very important to plant life, especially because plants are immobile and must survive in environmental extremes. Early studies of autophagy provided our first insights into the structural characteristics of the process in plants, but for a long time the molecular mechanisms and the physiological roles of autophagy were not understood. Genetic analyses of autophagy in the yeast Saccharomyces cerevisiae have greatly expanded our knowledge of the molecular aspects of autophagy in plants as well as in animals. Until recently our knowledge of plant autophagy was in its infancy compared with autophagy research in yeast and animals, but recent efforts by plant researchers have made many advances in our understanding of plant autophagy. Here I will introduce an overview of autophagy in plants, present current findings and discuss the physiological roles of self-degradation.

Introduction

As sessile organisms, the intracellular remodeling and physiological responses of plants to adverse environmental conditions is critical for their survival. A key process by which eukaryotic cells respond to and survive environmental stresses is vacuolar autophagy (Greek for self-eating), whereby cytoplasmic components are degraded in the vacuole to provide raw materials and energy, and also to eliminate damaged or toxic components, for the maintenance of essential cellular functions. Two types of autophagy, called macroautophagy and microautophagy, are well known in plants (Fig. 1; Bassham et al. 2006). Macroautophagy is the most extensively studied and is mediated by a special organelle termed the autophagosome. During macroautophagy, bulk cytosolic constituents and organelles are sequestered into an autophagosome which is a double-membrane structure. The outer membrane of the autophagosome then fuses with the vacuolar membrane and thus delivers the inner membrane structure and its cargo, namely the autophagic body, into the vacuolar lumen for degradation (Fig. 1). In animals, the subsequent fusion of the autophagosome with the lysosome, an acidic compartment containing hydrolytic enzymes, generates the autolysosome. Autolysosome-like small vacuole structures have also been found in tobacco suspension-cultured cells, but whether they exist in all plant species is still unclear. During microautophagy, a portion of cytoplasm or some organelle is delivered into the vacuolar lumen by invagination of the vacuolar membrane, causing the formation of intravacuolar vesicles, which are then digested by resident vacuolar hydrolases (Fig. 1). In addition to these two types of autophagy, different autophagic pathways have been reported in animals. One is chaperone-mediated autophagy (Arias and Cuervo 2011). This involves the direct translocation of cytosolic proteins across the lysosomal membrane, which requires protein unfolding by chaperone proteins. The second one is xenophagy which is a type of macroautophagy but is specific for destroying pathogens (Galluzzi et al. 2011). So far it is not known whether these types of autophagy exist in plants. Usually the term ‘autophagy’ indicates macroautophagy unless otherwise specified.

Fig. 1

The autophagic process in plants. There are two types of autophagy, called microautophagy and macroautophagy. In microautophagy, the cytoplasmic components are engulfed by an invaginated vacuolar membrane. During macroautophagy, bulk cytosolic constituents and organelles are sequestered into a double-membrane structure called an autophagosome. (The autophagosome is generated from a pre-autophagosomal structure and its intermediate is called a phagophore or isolation membrane.) The outer membrane of the autophagosome then fuses to the vacuolar membrane, thus delivering the inner membrane structure, the autophagic body, into the vacuolar lumen for degradation. Ordinarily this autophagic body is rapidly degraded by vacuolar hydrolases and its constituents are recycled. So far, in plants, both microautophagy and macroautophagy have been reported.

The existence of autophagy was first suggested around the late 1950s to the 1960s by the observation of cytoplasmic material, such as mitochondria and endoplasmic reticulum (ER), being engulfed by double-membrane vesicles and digested by lysosomal enzymes after fusion with lysosomes in the liver of rats starved of amino acids (De Duve and Wattiaux 1966). In plants, autophagic structures have also been reported since the late 1960s. For example, Villiers and Marty observed autophagosome-like structures engulfing cytoplasmic components in embryo cells and in root meristem cells during vacuole biogenesis, respectively (Villiers 1967, Marty 1978). Additionally inclusions of cytoplasmic constituents wrapped in a membrane were observed in vacuoles, and the entire sequence of invagination and formation of the intravacuolar vesicles, microautophagy, was documented in corn root meristem cells (Matile and Moor 1968). In cotyledons and senescing leaves, cytoplasmic components and chloroplasts were reported to be degraded in the vacuoles (Matile 1975, Van der Wilden et al. 1980, Wittenbach et al. 1982). The chloroplast degradation appeared to be executed by microautophagy (Wittenbach et al. 1982). These findings were largely based on morphological observations made by electron microscopy. Although these morphological studies provided initial insights into the structural components and characteristics of the process in plants, the results described above provided only static, one-shot images of autophagy; the physiological roles of plant autophagy were only speculative.

More recently, genetic analyses have greatly expanded our knowledge of the molecular mechanisms and physiological roles of autophagy. Notably, the identification in the yeast Saccharomyces cerevisiae of AuTophaGy (ATG) genes (Table 1), which are essential for autophagosome formation, has contributed significantly to the development of methods that have allowed the molecular dissection of autophagy in higher organisms. The knowledge gained from yeast research and advances in technology has contributed to the development of autophagy-monitoring systems, and has led to autophagy becoming one of the hottest fields in the life sciences [2012 (PubMed) publications with autophagy as a key word in 2011]. In addition to the rapid elucidation of the molecular mechanisms of autophagy, unexpected physiological and pathological roles for autophagy have been discovered, especially in the field of animal autophagy, attracting widespread attention. In contrast, our understanding of plant autophagy is much delayed. However, based on the knowledge gained from yeast autophagy research, plant autophagy genes and autophagy-defective plants were isolated, and accordingly the plant autophagy field has been gradually growing. These advances have recently helped to elucidate many physiological roles of plant autophagy. In this review I will give an overview of autophagy, an intracellular self-degradation system in plants, along with novel findings. I will also consider issues raised by recent plant autophagy research and discuss future perspectives.

Table 1

Arabidopsis Atg homologs

^a T-DNA insertional knock-out mutants of these genes have been published.

^b Unpublished data.

TORC1, TOR kinase complex 1; PI(3)P, phosphatidylinositol 3-phosphate; PE, phosphatidylethanolamine.

Conservation of ATG Protein-Dependent Autophagy in Plants

Progress in understanding the mechanistic basis of autophagy has been greatly facilitated by the discovery of yeast autophagy and the subsequent identification of the ATG genes which are essential for autophagosome formation. The molecular machinery of the autophagy pathway was discovered by screening yeast mutants which cannot accumulate autophagic bodies in the vacuolar lumen and are also hypersensitive to nutrient deficiency. Since the first ATG genes were identified in yeast in 1993 (Tsukada and Ohsumi 1993), by the unification of nomenclature and the isolation of new yeast autophagy-related genes, their number has increased to 32 (Barth et al. 2001, Klionsky et al. 2003, Kanki et al. 2009, Nakatogawa et al. 2009, Okamoto et al. 2009). The genes essential for the central autophagic machinery consist of 18 ATG genes: ATG1–10, 12–14, 16–18, 29 and 31. In Arabidopsis, about 30 ATG homologs, which correspond to the 18 yeast ATG genes, were identified, although no homologs have been identified for ATG14, ATG17, ATG29 and ATG31 (Table 1; Doelling et al. 2002, Hanaoka et al. 2002, Xiong et al. 2005). Although the Arabidopsis ATG proteins (AtATG) do not share high sequence homology with the yeast Atg proteins, most of the essential residues are well conserved, suggesting that the molecular basis of the core autophagic machinery is the same in plants and yeasts. In addition to Arabidopsis, ATG genes have also been identified in crop species such as rice and maize (Su et al. 2006, Ghiglione et al. 2008, Chumg et al. 2009, Shin et al. 2009, Xia et al. 2011, Kuzuoglu-Ozturk et al. 2012). Besides the canonic autophagy pathway described in yeasts and plants, a unique autophagy-related pathway dedicated to the transport of aminopeptidase I from the cytosol to the vacuole, called the cytoplasm-to-vacuole targeting (Cvt) pathway, exists in yeasts. This Cvt pathway involves the core machinery of autophagy and some additional proteins encoded by the genes ATG19–ATG21 and ATG23. However, there are no clear homologs of these genes in plant genomes and it therefore seems unlikely that plants have a functional Cvt pathway. Alternatively, functionally related proteins may exist even though their sequence similarity is quite low. Recently, nine novel regulatory proteins required for autophagosome formation were isolated in mammals by genome-wide small interfering RNA (siRNA) screening (McKnight et al. 2012). Although we do not know if the homologs of these proteins exist in plants, it is possible that plants also have additional autophagy-related proteins.

The function of Atg proteins and their regulation during induction of autophagy processes have been intensively investigated in yeast, leading to the division of the Atg proteins into four functional groups: (i) the Atg1 protein kinase complex; (ii) the phosphatidylinositol 3-kinase (PI3K) complex specific for autophagy; (iii) the Atg9 complex; and (iv) two ubiquitination-like systems, leading to Atg8 lipidation and Atg12 protein conjugation.

The Atg1 protein kinase complex consists of Atg1, Atg13, Atg17, Atg29 and Atg31, and is involved in induction and regulation of autophagy. It is well known that the target of rapamycin (TOR) complex, specifically TOR complex 1 (TORC1), is an upstream negative regulator of this complex (He and Klionsky 2009). In yeast, the active TORC1 hyperphosphorylates Atg13 under nutrient-sufficient conditions (Kamada et al. 2010). This prevents the association of Atg1 with the Atg13 subcomplex, that consists of Atg13, Atg17, Atg31 and Atg29, leading to the inhibition of autophagy induction. Under nutrient-limiting conditions TORC1 is inactivated and Atg13 is no longer phosphorylated. Subsequently Atg1 can associate with the Atg13 subcomplex and is autophosphorylated, leading to autophagy induction (Nakatogawa et al. 2009, Kamada et al. 2010). Recently Vierstra’s group clearly showed that the Arabidopsis ATG13 proteins AtATG13a and AtATG13b function as regulators of autophagy (Suttangkakul et al. 2011). It was also shown that AtATG1a interacts with AtATG13a. Therefore, the Atg1 protein kinase complex seems to play roles in plant autophagy. Although the four Atg1 homologs AtATG1a–1c and AtATG1t (which is a truncated version of ATG1 containing only the kinase domain but not the regulatory domain and seems to be specific to seed plants), and two ATG13 homologs AtATG13a and AtATG13b have been identified in Arabidopsis, the homologs of Atg17, Atg29 and Atg31, which are also part of the Atg1 kinase complex, have not been found (Doelling et al. 2002, Hanaoka et al. 2002, Suttangkakul et al. 2011). The isolation of proteins in plants that are functionally related to the yeast or animal proteins would help to elucidate the mechanisms controlling induction of plant autophagy. Homologs of subunits of the TORC1 complex such as TOR, RAPTOR and LST8 have been identified in Arabidopsis, and their knock-out and knock-down mutants have also been characterized (Menand et al. 2002, Anderson et al. 2005, Deprost et al. 2005, Deprost et al. 2007, Moreau et al. 2012). Using TOR RNA interference (RNAi) plants, it has been shown that TOR is a negative regulator of autophagy in Arabidopsis like in yeast and animals (Liu and Bassham 2010). Further investigation of autophagy using T-DNA insertion mutants or RNAi transgenic plants of other TORC1-related proteins would shed light on the dynamic molecular mechanisms involved in the induction of autophagy.

After the assembly of the Atg1–Atg13 kinase complex, subsequent activation of the complex induces several steps during autophagosome formation, although we still do not know whether these steps are regulated by the Atg1 kinase complex directly or indirectly, or whether they occur in series or in parallel. From hierarchical analyses of Atg proteins it is thought that the PI3K and the Atg9 complexes act next (Table 1; Suzuki et al. 2007, Itakura and Mizushima 2010). The PI3K complex, which is essential for autophagy, is one type of Class III PI3K which consists of Vps34, Vps15, Atg6 and Atg14. One of the roles of the PI3K complex is to recruit the Atg18–Atg2 complex to autophagic membranes through an Atg18– phosphatidylinositol 3-phosphate (PI3P) interaction, although its precise function is not yet clear. A homolog of Atg14 that determines specificity for autophagy has not yet been isolated in Arabidopsis; however, it is likely that the PI3K complex plays a critical role in regulating plant autophagy since silencing of the ATG6 gene in tobacco and Arabidopsis plants resulted in reduced autophagy (Liu et al. 2005, Patel et al. 2008).

Because it is the sole transmembrane protein among Atg proteins, Atg9 is thought to be involved in delivering lipids to the pre-autophagosomal structure and the expanding phagophore. In yeast it is known that Atg9 makes oligomers and interacts with the Atg2–Atg18 subcomplex. Due to the difficulty in analyzing the transmembrane protein Atg9 and the relatively large protein Atg2, the actual function of this complex has not even been clearly elucidated in yeast. A similar ATG9 complex also seems to be involved in plant autophagy. The Arabidopsis T-DNA mutant Atatg9-1 shows autophagy-defective phenotypes such as early senescence (Hanaoka et al. 2002). The T-DNA insertion mutant Atatg2-1, Atatg18a-1 and transgenic plants carrying an RNAi construct that targets AtATG18a have defects in autophagosome formation (Xiong et al. 2005, Inoue et al. 2006, K. Yoshimoto, unpublished data). It is still unclear whether AtATG9 forms an oligomer that interacts with the AtATG2–AtATG18a complex, as its homolog does in yeast.

Two ubiquitination-like systems have been most extensively studied in many organisms. It is known that both the Atg8 lipidation system and the Atg12 protein conjugation system are important for the elongation and enclosure steps during autophagosome formation (Table 1). Atg8 covalently conjugates to the membrane lipid phosphatidylethanoleamine (PE) by a ubiquitin-related conjugation system consisting of Atg7 functioning as an E1-like enzyme and Atg3 acting as an E2-like enzyme. In this process, the C-terminal extension of the newly synthesized Atg8 must first be processed by Atg4, a cysteine protease, to expose the glycine residue at its C-terminus. This truncated Atg8 is then activated by Atg7 and Atg3 in turn, through a thioester bond between the glycine residue of Atg8 and a cysteine residue of each enzyme. Finally, Atg8 is conjugated to the head group of PE (Ichimura et al. 2000). This system is also conserved in plants. Phenotypes of the Arabidopsis T-DNA insertion mutant Atatg7 can be complemented by the wild-type AtATG7 protein but not by a mutant version of the protein AtATG7C/S which contains a substitution in a catalytically active cysteine residue. This suggests that AtATG7 functions as an E1 enzyme in ATG conjugation reactions (Doelling et al. 2002). In addition, the C-terminus of the nascent form of all nine AtATG8 homologs is cleaved in an AtATG4-dependent manner, and an alanine substitution of the C-terminal glycine residue of the proteins results in defects in their subcellular localization (Yoshimoto et al. 2004). It has been demonstrated that the PE-conjugated AtATG8s and the AtATG8–AtATG3 intermediates are present in wild-type plants but not in Atatg7 and Atatg4a4b, which is a double knockout mutant with T-DNA insertions in both the AtATG4a and AtATG4b genes (Yoshimoto et al. 2004, Chung et al. 2010). A second ubiquitin-like conjugation system using Atg7 as an E1-like enzyme and Atg10 as an E2-like enzyme covalently conjugates Atg12 to Atg5. In the conjugation reaction, Atg12 is activated by Atg7 and then transferred to the cysteine residue of the E2-like enzyme Atg10, before it is finally conjugated to Atg5 via an isopeptide bond between the C-terminal glycine residue of Atg12 and a lysine residue of Atg5. The Atg12–Atg5 conjugate then forms an oligomer with Atg16 and also stimulates the Atg8–PE conjugation reaction, acting as an E3 enzyme. Arabidopsis mutant analyses have shown that AtATG12–AtATG5 conjugates are essential for plant autophagy (Suzuki et al. 2005, Thompson et al. 2005, Chung et al. 2010). Using anti-AtATG12b antibodies and anti-AtATG5 antibodies AtATG12–AtATG5 conjugates were detected in wild-type plants, but not in Atatg5, Atatg7, Atatg10 and Atatg12a12b plants, which are deficient in the target of AtATG12, the E1-like and E2-like enzymes that target AtATG12 and AtATG12 itself, respectively. Instead, the levels of unconjugated AtATG12s were higher in these mutants than in control plants. Additionally, in the T-DNA mutant of AtATG10, Atatg10, wild-type AtATG10 was able to restore the generation of AtATG12–AtATG5 conjugates, whereas AtATG10C/S, in which the catalytically active cysteine residue was replaced with a serine residue, could not. This provided further evidence that AtATG10 functions as an E2 enzyme in the Atg12 conjugation reaction (Phillips et al. 2008). These results have been supported by the in vitro reconstitution of both the AtATG8 and AtATG12 conjugation systems using recombinant proteins (Fujioka et al. 2008). Whether the plant ATG16 homologs are involved in the formation of autophagosomes remains unclear.

Visualization of Autophagic Activity in Plants

As stated earlier, electron microscopy has been used for many decades to visualize autophagic processes in plant cells. Electron microscopy is one of the most reliable methods for monitoring autophagic structures in many organisms, and was in fact originally used to discover autophagy in the late 1950s. At the ultrastructural level, an autophagosome is defined as a double-membrane vesicle containing undigested cytoplasmic constituents, including organelles. However, this method only provides static, one-shot images of autophagy. Since the autophagic process is a very dynamic and rapid event, the interpretation of a result based on morphological analysis requires caution. It is therefore recommended that a combination of several techniques be used to conclude whether a structure is indeed an autophagic structure. At present immunoelectron microsocopy and tomographic analyses would certainly be the most powerful tools for that.

Nowadays the assessment of autophagosome structures by electron microscopy has been replaced by light microscopy methods. As noted above, Atg8 is modified by conjugation with the lipid PE molecule via ubiquitination-like reactions. Consequently, Atg8 behaves as an intrinsic membrane protein, and is involved in autophagosome formation in its PE-conjugated form. PE-conjugated Atg8 is localized to the phagophore and autophagosome, and, eventually, a proportion of Atg8 that is enclosed in autophagosomes is transferred to the vacuolar lumen during the autophagic process. Therefore, in yeast and mammals, Atg8 and its ortholog LC3 (light chain 3) are useful molecular markers for monitoring the autophagic process (Kirisako et al. 1999, Kabeya et al. 2000). In plants the ATG8 protein also serves as a suitable molecular marker for autophagosomes (Yoshimoto et al. 2004, Contento et al. 2005, Thompson et al. 2005). In wild-type roots expressing green fluorescent protein (GFP)–AtATG8, GFP fluorescence can be observed in punctate and ring-shaped structures, which probably correspond to pre-autophagosomes, phagophore and autophagosomes (see Fig. 1). These structures increase under nutrient starvation or stress conditions. Although the number of such structures per cell is usually an accurate measurement of autophagic activity, this measure is potentially problematic. Even in autophagy-defective plants such as Atatg4a4b, dot-like structures have been detected in the roots of plants expressing GFP–AtATG8. The structures observed in the atg mutant seem to be protein aggregates because under conditions of nitrogen starvation the GFP–AtATG8 protein is delivered to the vacuolar lumen in wild-type roots, while no GFP–AtATG8 proteins were delivered to the vacuoles in Atatg4a4b plants. Such aggregation often occurs when proteins are exogenously overexpressed in plant cells. Initially, it is difficult to distinguish whether the punctuate structures stained with GFP–AtATG8 are actual autophagic structures or protein aggregates. To evaluate autophagic activity correctly, therefore, autophagic flux such as transport of GFP–ATG8 to the vacuolar lumen should be confirmed in addition to the observation of the punctuate structures.

In mammalian cells the Atg12–Atg5 conjugate primarily decorates the outer membrane of the phagophore, but is absent from mature autophagosomes, and dissociates from the autophagosome when its formation has been completed. Therefore Atg12–Atg5 has been used as a good molecular marker for monitoring intermediate processes of autophagosome formation. As in mammalian cells, using GFP–AtATG5 it may be possible to investigate the process of plant autophagosome formation in detail.

Approaches that utilize chemical inhibitors of autophagic processes have also been developed in plants. In yeast, phenylmethylsulfonyl fluoride (PMSF), a serine protease inhibitor, has been used for monitoring the autophagic process (Tsukada and Ohsumi 1993). This chemical inhibits vacuolar hydrolases leading to accumulation of autophagic bodies in vacuolar lumens. A similar approach has been taken to detect the autophagy process in plant cells. E-64 {[l-3-trans-ethoxycarbonyloxirane-2-carbonyl]-l-leucine (3-methylbutyl) amide}, a cysteine protease inhibitor of the papain family, was first used in tobacco cells to inhibit a process of plant autophagy (Moriyasu and Ohsumi 1996). In cultured tobacco cells under sucrose starvation conditions, E-64 treatment caused the accumulation of single membrane vesicles around the nucleus, not in the vacuolar lumen. These vesicles were acidic and contained partially degraded cytoplasmic constituents, cysteine protease activity, acid phosphatase activity and vacuolar H⁺-ATPase subunits. Their accumulation was inhibited by the PI3K inhibitor 3-methyladenine (3-MA), a well-known autophagy inhibitor in yeast and mammals, indicating that the vesicles were autophagic structures. Additionally, other PI3K inhibitors such as wortmannin and LY294002 also inhibited the accumulation of these vesicles. Since these vesicles were single membrane, located in the cytoplasm and contained acid phosphatase, which is a marker enzyme of lysosomes, the structures observed in tobacco cells after E-64 treatment seemed to be autolysosomes rather than the autophagic bodies observed in yeast treated with PMSF (Moriyasu and Ohsumi 1996, Takatsuka et al. 2004, Takatsuka et al. 2011). When Arabidopsis roots were treated with E-64, many aggregated (not spherical) structures accumulated in the vacuolar lumen of wild-type cells, but not in that of Atatg cells (Inoue et al. 2006, Yoshimoto et al. 2009). This result indicated that these aggregated structures were autophagic bodies that were probably partially degraded. It seems that E-64 does not fully prevent degradation of the membranes of the autophagic bodies in vacuolar lumens. The difference in E-64-treated cells between tobacco and Arabidopsis suggests that different autophagic pathways exist in different plant species.

Concanamycin A, a V-ATPase inhibitor, has also been used to monitor autophagic processes in plant cells. When Arabidopsis roots are treated with concanamycin A, many spherical structures can be observed in the vacuolar lumen of wild-type cells by conventional light microscopy. These structures are absent in Atatg4a4b-1 cells (Yoshimoto et al. 2004). Electron microscopy showed that these spherical bodies contain cytosolic constituents including organelles such as mitochondria, ER and Golgi bodies, indicating that these structures are plant autophagic bodies. Concanamycin A probably raises the pH of the vacuolar lumen when exogenously added to plant cells. It is expected that under such high pH conditions vacuolar hydrolases cannot act, leading to the accumulation of autophagic bodies in the vacuoles. Treatment with concanamycin A is therefore a very easy way to detect whether autophagy occurs in plants and is now used like PMSF treatment in yeast.

Recent advances in technology allow us to use fluorescent chemicals to monitor autophagosomes. LysoTracker and monodansylcadaverine (MDC) staining methods have been used for monitoring autophagy in plants as well as in mammals (Contento et al. 2005, Liu et al. 2005, Patel et al. 2008). However caution is necessary in applying these methods. LysoTracker is a fluorescent probe that functions as a label of acidic organelles in living cells, therefore it is not specific for autophagy. In plants, LysoTracker labels senescence-associated vacuoles (SAVs) as punctate structures, which increase during senescence but are not autophagic compartments because they are formed even in autophagy-defective plants such as the Atatg7 mutant (Otegui et al. 2005). In mammalian cells, MDC was originally proposed to be an indicator of autophagosomes and autolysosomes; however, recent studies revealed that it was not a suitable marker for autophagy-specific structures. MDC-positive structures apparently did not always co-localize with autophagosomes labeled by the GFP–ATG8 homolog, which is the most reliable molecular marker of autophagosomes (Mizushima 2004). It seems that MDC labels acidic organelles such as lysosomes and late endosomes rather than autophagosomes. Therefore, careful observations are also required in plants.

In addition to fluorescence microscopy assays, biochemical assays have also been utilized to assess autophagic activity. As I described above, GFP–ATG8 is delivered to vacuolar lumens as a consequence of autophagy. It is partially degraded upon reaching the vacuole, resulting in the appearance of a free GFP fragment. Therefore the detection of the free GFP fragment by immunoblotting with anti-GFP antibody is one approach to the measurement of autophagic flux. The GFP–ATG8 processing assay has also been tested in Arabidopsis and was confirmed to be robust in plants (Suttangkakul et al. 2011, Klionsky et al. 2012). In mammals, conversion from LC3-I to a PE-conjugated form LC3-II can be detected by immunoblotting using antibodies against LC3. The LC3-II migrates faster than LC3-I because of its hydrophobicity, and the amount of LC3-II usually correlates well with the number of autophagosomes. In plants, PE-conjugated ATG8 can also be detected by immunoblotting with anti-AtATG8 antibodies using a urea-containing gel (Yoshimoto et al. 2004, Chung et al. 2009, Chung et al. 2010, Suttangkakul et al. 2011). As in mammals, it may be possible to use the ratio of free AtATG8 to PE-conjugated AtATG8 as an indicator of plant autophagic activity. However, there are some cases when the amount of the PE-conjugated form does not correlate with autophagic activity. In fact, even in Atatg mutants ATG8 lipidation still occurs. In other words, it is still possible that autophagy is suppressed even if the PE-conjugated form is detected (Yoshimoto et al. 2004, Suttangkakul et al. 2011). Acknowledgement of these limitations is necessary for the interpretation of results from biochemical assays, which probably need to be combined with other approaches to assess autophagic activity correctly.

Many Aspects of Autophagy in Plant Life

Since the middle of the 20th century, morphological studies have proposed many physiological roles for autophagy in plants. With the recent development of systems for monitoring plant autophagy, it has been clearly shown that Arabidopsis atg mutants and transgenic plants expressing ATG gene silencing constructs are all defective in autophagy. Using these plants, now it is finally possible to elucidate the functions of autophagy in plant life.

Early studies by electron microscopy proposed that autophagy involves the formation of vacuoles (Marty 1978, Marty 1997). Contrary to our expectations, Atatg mutants defective in autophagy are able to complete normal embryogenesis, germination, cotyledon development, shoot and root elongation, flowering and seed production, not showing a defect in vacuole formation during development. However, there is still a possibility that autophagy is partially involved in vacuole formation because the outer membrane of the autophagosomes fuses to the vacuolar membrane, thus supplying membranes to the vacuole and supporting its formation. Another possibility is that non-canonical autophagy that does not involve the ATG genes exists in plants and functions in the formation of vacuoles (Yano et al. 2007).

Morphological observations indicate that autophagic structures are often found under nutrient-starved conditions in plant cells, suggesting a nutrient recycling function for plant autophagy. When Atatg mutants and ATG gene-silenced plants are grown under nitrogen- and carbon-depleted or limited conditions, they predictably exhibit a drastic acceleration in starvation-induced chlorosis and artificially induced senescence, and a reduced growth rate of roots, indicating that when nutrient supply from the environment is limited, autophagy is required for nutrient mobilization in plant cells (Doelling et al. 2002, Hanaoka et al. 2002, Yoshimoto et al. 2004, Thompson et al. 2005, Phillips et al. 2008, Chung et al. 2010, Suttangkakul et al. 2011). Recently, using ¹⁵N tracing, it has been proved that autophagy is needed for nitrogen remobilization (Guiboileau et al. 2012). Natural senescence is also accelerated in the autophagy-defective plants; however, the phenotype is not likely to be caused by nutrient starvation because it is observed even under nutrient-rich conditions and is suppressed by inactivation of salicylic acid (SA) signaling (Yoshimoto et al. 2009). In addition to a function in nutrient recycling, autophagy seems to have a role in the elimination of excessive SA signaling during senescence. How autophagy negatively regulates SA signaling remains unclear.

Under nutrient-starved conditions, chloroplasts can be partially degraded by autophagy via spherical bodies named Rubisco-containing bodies (RCBs), which contain stromal proteins but not thylakoid membranes. When wild-type plant leaves expressing chloroplast stroma-targeted GFP were incubated with the V-ATPase inhibitor concanamycin A in the dark, many small vesicles exhibiting Brownian motion were observed in the vacuolar lumen. These vesicles were not co-localized with Chl, suggesting that they were RCBs. The accumulation of RCBs in the vacuolar lumen is disrupted in the Atatg mutants, indicating that RCBs are transported to the vacuole by autophagy for degradation (Ishida et al. 2008). Small vesicles structurally similar to RCBs, SAVs or RVBs that contain stromal proteins, have been discovered as cytoplasmic bodies in tobacco plants (Otegui et al. 2005, Prins et al. 2008). Although these vesicles are involved in degradation of stromal proteins, it is still not known whether they are related to the autophagic process. In Arabidopsis, the transport of RCBs from the cytoplasm to the vacuole by autophagy appears dependent on carbon status (Izumi et al. 2010). The accumulation of RCBs in the vacuolar lumen was suppressed by addition of metabolic sugars such as sucrose, glucose and fructose, and also illumination by light. Light-mediated suppression was cancelled by addition of DCMU, an inhibitor of photosynthetic electron transport. Consistently, starch excess mutants produced fewer RCBs than wild-type plants, while starch-less mutants produced a large number of RCBs (Izumi et al. 2010). These results suggest that plants sense carbon status and, when it is limited, autophagy degrades RCBs to supply energy from stromal proteins and the chloroplast envelope as a carbon source without destruction of whole chloroplasts. Alternatively, via RCBs autophagy might degrade a portion of the stromal proteins damaged under starvation conditions as part of a quality control system.

Roles for autophagy under environmental stresses such as oxidative, drought and salinity stress have also been suggested by experiments involving autophagy-defective plants (Xiong et al. 2007, Liu et al. 2009, Shin et al. 2009). Autophagy-defective Arabidopsis and rice (Oryza sativa) seem to be hypersensitive to various abiotic stresses. In response to oxidative stress, AtATG18a knock-down transgenic plants accumulated more oxidized proteins than wild-type plants, suggesting that autophagy can transport oxidized or damaged cellular components to the vacuole for degradation to protect plant cells from these toxic materials (Xiong et al. 2007). However, whether this transport is selective is still unclear.

Recently some studies have proposed that autophagy is involved in plant immunity (Liu et al. 2005, Hofius et al. 2009, Yoshimoto et al. 2009, Lai et al. 2011, Lenz et al. 2011, Wang et al. 2011). The innate immune response triggered by incompatible plant–microbe interaction leads to hypersensitive response cell death which is a form of programmed cell death (HR-PCD) characterized by the rapid death of plant cells at the site of pathogen infection to prevent the spread of pathogens (Table 2). There is evidence that autophagy suppresses the spread of pathogen-induced cell death to uninfected tissue from the infection site (Liu et al. 2005, Yoshimoto et al. 2009). However, the role of autophagy in HR-PCD is controversial because it has also been reported that autophagy can have a pro-death function at the infection site during HR-PCD (Hofius et al. 2009). Depending on the plant growth conditions and pathogen lifestyle, autophagy seems to be able to have either pro-survival or pro-death functions (Hofius et al. 2009, Yoshimoto et al. 2009, Lai et al. 2011, Lenz et al. 2011, Wang et al. 2011). When autophagy-defective Arabidopsis plants were infected with virulent biotrophic pathogens, they exhibited increased resistance, suggesting a negative regulatory role for autophagy in plant basal immunity (Table 2; Lenz et al. 2011). In contrast, autophagy-defective plants develop spreading necrosis and enhanced disease susceptibility upon infection with necrotrophic pathogens, suggesting that autophagy contributes to resistance to necrotrophs (Table 2; Lenz et al. 2011). The exact roles of autophagy in plant immunity are still unclear, and may vary depending on the nature of the plant species and pathogen strains. The negative regulatory role of autophagy in pathogen-induced cell death is probably caused by highly accumulated SA in autophagy-defective plants (Yoshimoto et al. 2009). Thus, it has been proposed that autophagy has a function in the elimination of excessive SA signaling during pathogen-induced cell death (Yoshimoto 2010). The susceptibility of Atatg mutants to necrotrophic pathogens might also be explained by the suppression of jasmonic acid signaling, which is required for resistance against necrotrophic pathogens, by excessive SA signaling/accumulation (Table 2; Koornneef et al. 2008, Lenz et al. 2011).

Table 2

Plant immune responses defined by plant–microbe interactions

Plant–microbe interaction	Plant response	Disease	Phenotype of atg
Incompatible	HR-PCD	−	Spread of cell deatha,b, increased cell death ratec
Compatible (biotroph)	SA signaling	+	Resistanced,e
Compatible (necrotroph)	JA signaling	+	Susceptibled,f

If a plant possesses a resistance (R) gene that corresponds to a pathogen’s avirulence gene (Avr), then the interaction is said to be incompatible; infection is prevented by HR-PCD and no disease develops. Alternatively, if a plant does not possess the matching R gene for a pathogen’s avirulence gene, the interaction is termed compatible and infection proceeds, causing disease. During compatible interactions, salicylic acid (SA) or jasmonic acid (JA) signaling is activated as part of a plant’s innate immune response to infection by biotrophic or necrotrophic pathogens, respectively. Biotroph: a pathogen that derives nutrients from the living tissues of its host. Necrotroph: a pathogen feeding only on dead host tissues. Phenotypes of atg mutants are based on ^aLiu et al. (2005), ^bYoshimoto et al. (2009), ^cHohius et al. (2009), ^dLenz et al. (2011), ^eWang et al. (2011) and ^fLai et al. (2011).

Selectophagy—Selective Disposal of a Protein/Organelle by Autophagy in Plants

Although autophagy was generally considered as a non-selective pathway capable of degrading a portion of cytoplasm, several recent findings have indicated that autophagy can be selective like the ubiquitin–proteasome system (Johansen and Lamark 2011).

In plants, selective autophagy was first experimentally suggested in tobacco suspension-cultured BY-2 cells overexpressing Cyt b₅ fused with red fluorescent protein (Cytb5–RFP) (Toyooka et al. 2006). The ectopically overexpressed Cytb5–RFP was observed as punctuate aggregates in the cytoplasm and transported to the vacuoles by autophagy under nutrient-limiting conditions. Since the degradation rate of Cytb5–RFP aggregates was faster than that of other organelle proteins such as BiP, Sec22, F1-ATPase and CP29, the transport of the Cytb5–RFP aggregates to the vacuoles appeared to be selective.

Recent studies in yeast and animals have revealed that selectivity for target recruitment is mediated by the autophagosomal protein ATG8 and autophagic adaptors. In addition to a function in membrane tethering (hemi-fusion) during elongation of the phagophore, the ATG8 protein has an additional function as a recruiter for targets in selective autophagy. Through the ATG8-interacting motif (AIM in yeast) or LC3-interacting region (LIR in animals) which consists of the consensus sequence W/Y/F-XX-L/I/V surrounded by several acidic residues, the autophagosomal protein ATG8/LC3 binds to adaptors such as yeast mitochondrial membrane protein ATG32 and mammalian Nix, which recruit mitochondria. Similarly via interaction with the ATG8 homolog mammalian NBR1 (neighbor of Brca1 gene) and p62/SQSTM1 (sequestosome 1), Caenorhabditis elegans SEPA-1 and mammalian NDP52 (nuclear dot protein 52kDa) recruit ubiquitinated protein aggregates, P granules and invaded bacteria to the autophagosomes, respectively (Johansen and Lamark 2011). This molecular mechanism for selective autophagy via AIM/LIR is also conserved in plants. Quite recently four studies showed selective degradation of a protein/organelle in plants through AIM/LIR (Svenning et al. 2011, Vanhee et al. 2011, Zientara-Rytter et al. 2011, Honig et al. 2012). Although there is no p62 homolog in Arabidopsis based on sequence similarity, one NBR1 homolog with hybrid properties of mammalian NBR1 and p62 has been found in Arabidopsis and tobacco (Svenning et al. 2011, Zientara-Rytter et al. 2011). In Arabidopsis, NBR1 forms a homo-oligomer via the PB1 domain, binds to ubiquitin via the C-terminal UBA domain, and interacts with AtATG8 through the conserved AIM/LIR. Consequently it is selectively engulfed by autophagosomes and delivered to the vacuolar lumen for degradation, as occurs in mammals. It is well described that both p62 and NBR1 act as cargo receptors for degradation of ubiquitinated protein aggregates in mammals (Lamark and Johansen 2012). These proteins are involved in elimination of undesired aggregated proteins by autophagy and are thus important for the maintenance of homeostasis inside cells. The physiological function of the NBR1 homolog in plants is still unclear because cargo for selective autophagy has not been clearly identified and knock-out plants of the NBR1 homolog are unavailable (a homozygous T-DNA insertion mutant for the NBR1 gene cannot be isolated, probably due to embryonic lethality or male sterility; K. Yoshimoto, unpublished data). We can, however, hypothesize that after ubiquitination, protein aggregates such as Cytb5–RFP described above might be a target for the NBR1 homolog in plants. Tobacco NBR1 homolog, Joka2, was isolated as an interactor of a small coiled-coil protein UP9C, a member of the UP9/LSU family. Although the function of UP9C is still unknown, the increase of the transcript levels of the UP9C gene and the reduced root growth of UP9C silenced plants during sulfur deficiency suggest that NBR1 has an important role for selectophagy and contributes to the maintenance of cellular homeostasis during sulfur deficiency in plants (Lewandowska et al. 2010, Zientara-Rytter et al. 2011).

In addition to NBR1, two other cargo receptors possessing AIM/LIR have been reported in Arabidopsis. One is At-TSPO, a tryptophan-rich sensory protein (TSPO)-related membrane protein, which has a strong affinity for heme (Vanhee et al. 2011). Vanhee et al. found that ABA-induced At-TSPO proteins are gradually degraded and that this degradation needs the AIM/LIR-like sequence in At-TSPO, which is suppressed in the atg5 mutant. Therefore autophagy seems to be involved in degradation of undesired toxic heme under ABA-dependent environmental stress conditions. However, whether the degradation of At-TSPO is directly mediated by physical interaction between ATG8 proteins and the AIM/LIR sequence is still unclear. The other cargo receptors are ATI1 and ATI2 which were isolated as Arabidopsis ATG8f-interacting membrane proteins by yeast two-hybrid screening (Honig et al. 2012). Although there are two putative AIM/LIR sequences in ATI1 and ATI2, a precise interaction site has not been elucidated. Under favorable conditions, both ATI1 and ATI2 are partially localized on the ER, probably at the ER membrane because these proteins posses a putative single transmembrane domain. Under carbon-starved conditions these proteins are interestingly localized at the periphery of carbon starvation-induced spherical bodies. The newly identified spherical structures named ATI1 bodies seem to be associated with the ER network and are transported to the central vacuole, suggesting that ATI1 is a cargo receptor for degradation of ATI1 bodies, which are induced under carbon-starved conditions.

Conclusions and Future Perspectives

The recent availability of systems for monitoring autophagy and of organisms that are defective in autophagy has greatly increased our understanding of autophagy in plants as well as in yeasts and animals. Now, in addition to analogous phenomena, many plant-specific autophagy phenomena have been reported, signifying the dawn of plant autophagy research. However, there are some challenges that have to be addressed in plant autophagy research. Using GFP–ATG8 transgenic plants, we are now able to monitor autophagic processes; however, this has been predominantly in root cells. Monitoring autophagy in all plant tissues is still difficult because of the thickness of plant organs and also due to the probably inadvertent silencing of GFP–ATG8 in aerial parts of the transgenic plants. Novel autophagy-assessing systems such as the alkaline phosphatase assay, a quantification method in yeast (Noda et al. 1998), are required to further our understanding of biological phenomena involving autophagy in plants. It should also be noted that certain approaches and methods for analyzing autophagy reported in the literature are not always appropriate. To expand our knowledge of plant autophagy correctly, it will be necessary to observe autophagic processes with care using several methods and also to pay close attention to the interpretation of data. In addition, caution is necessary for the interpretation of phenotypic analyses in autophagy-defective plants as there is the possibility that a phenotype is caused by a secondary effect or that some ATG proteins have additional functions independent of autophagy. Indeed ATG6, one of the constituents of the PI3K complex, has multiple functions in addition to its role in autophagy, such as vacuolar protein sorting. Consequently, the Atatg6 mutant shows defects in pollen germination, indicating that AtATG6 plays a role in phosphoinositide signaling during pollen germination (Fujiki et al. 2007, Qin et al. 2007, Harrison-Lowe and Olsen 2008).

Although current reports suggest that plant autophagy has a variety of functions in selective degradation of specific protein(s)/organelle(s) as well as non-selective degradation, there are still many questions that have to be addressed. Of particular interest is the question of how plant cells determine that degradation of protein(s)/organelle(s) is necessary, recognize the structures that are to be degraded, and degrade these structures to the appropriate extent while maintaining intracellular homeostasis. Additionally the source of the plant autophagosomal membrane during the degradation process also remains to be elucidated. Further studies combined with morphological, biological, genetic and comparative approaches will provide new insights into plant autophagy.

Funding

This study was supported by the Ministry of Education, Culture, Sports, Science and Technology, Japan [a Grant-in-Aid for Young Scientists (B) (grant No. 22770049) to K.Y.] and by INRA, France [an INRA package program to K.Y.].

Acknowledgments

The author would like to thank Dr. Céline Masclaux-Daubresse and Dr. Ivana Saska for critical reading of the manuscript.

Abbreviations

Abbreviations
 
• AIM

  ATG8-interacting motif

 
• ATG

  autophagy-related

 
• ATI

  ATG8-interacting protein

 
• Cvt

  cytoplasm-to-vacuole targeting

 
• E64

  [l-3-trans-ethoxycarbonyloxirane-2-carbonyl]-l-leucine (3-methylbutyl) amide

 
• ER

  endoplasmic reticulum

 
• GFP

  green fluorescent protein

 
• HR-PCD

  hypersensitive response programmed cell death

 
• LC3

  light chain 3

 
• LIR

  LC3-interacting region

 
• MDC

  monodansylcadaverine

 
• NBR1

  neighbor of Brca1 gene

 
• PB1 domain

  Phox and Bem1 domain

 
• PE

  phosphatidylethanolamine

 
• PI3K

  phosphatidylinositol 3-kinase

 
• PI3P

  phosphatidylinositol 3-phosphate

 
• PMSF

  phenylmethylsulfonyl fluoride

 
• RCB

  Rubisco-containing body

 
• RFP

  red fluorescent protein

 
• RNAi

  RNA interference

 
• SA

  salicylic acid

 
• SAV

  senescence-associated vacuole

 
• TOR

  target of rapamycin

 
• TORC1

  TOR complex 1

 
• TSPO

  a tryptophan-rich sensory protein

 
• UBA domain

  ubiquitin-associated domain.