The nomenclature of autophagy-related genes was recently changed to the novel gene name ATG (for autophagy) (Klionsky et al., 2003). Consequently, autophagy genes from Arabidopsis thaliana are derivatively named AtATG. In the present study, the new ATG nomenclature has been adopted.
Starvation-induced expression of autophagy-related genes in Arabidopsis
Article first published online: 9 JAN 2012
2006 Société Française des Microscopies and Société Biologie Cellulaire de France
Biology of the Cell
Volume 98, Issue 1, pages 53–67, January 2006
How to Cite
Rose, T. L., Bonneau, L., Der, C., Marty-Mazars, D. and Marty, F. (2006), Starvation-induced expression of autophagy-related genes in Arabidopsis. Biology of the Cell, 98: 53–67. doi: 10.1042/BC20040516
- Issue published online: 9 JAN 2012
- Article first published online: 9 JAN 2012
- Received 26 November 2004/ 1 April 2005; Accepted 6 April 2005
Background information. Autophagy is a catabolic process for degradation of cytoplasmic components in the vacuolar apparatus. A genome-wide survey recently showed evolutionary conservation among autophagy genes in yeast, mammals and plants. To elucidate the molecular and subcellular machinery responsible for the sequestration and subsequent digestion of intracellular material in plants, we utilized a combination of morphological and molecular methods (confocal laser-scanning microscopy, transmission electron microscopy and real-time PCR respectively).
Results. Autophagy in Arabidopsis thaliana suspension-cultured cells was induced by carbon starvation, which triggered an immediate arrest of cell growth together with a rapid degradation of cellular proteins. We followed the onset of these responses and, in this report, provide a clear functional classification for the highly polymorphic autophagosomes by which the cell sequesters and degrades a portion of its own cytoplasm. Quantification of autophagy-related structures shows that cells respond to the stress signal by a rapid and massive, but transient burst of autophagic activity, which adapts to the stress signal. We also monitored the real-time expressions of AtATG3, AtATG4a, AtATG4b, AtATG7 and AtATG8a–AtATG8i genes, which are orthologues of yeast genes involved in the Atg8 ubiquitination-like conjugation pathway and are linked to autophagosome formation. We show that these autophagy-related genes are transiently up-regulated in a co-ordinated manner at the onset of starvation.
Conclusions. Sucrose starvation induces autophagy and up-regulates orthologues of the yeast Atg8 conjugation pathway genes in Arabidopsis cultured cells. The AtATG3, AtATG4a, AtATG4b, AtATG7 and AtATG8a–AtATG8i genes are expressed in successive waves that parallel the biochemical and cytological remodelling that takes place. These genes thus serve as early markers for autophagy in plants.
- poly(A)+ RNA
Most higher plants have to cope with a lack of carbohydrates at some time during their life history. Carbohydrate deprivation may occur in young seedlings during the critical transition to autotrophy (El Amrani et al., 1994; Escobar-Gutiérrez et al., 1998) or during environmental changes leading to a decrease in photosynthesis (Kraemer and Alberte, 1995; Brouquisse et al., 1998). To sustain respiratory activity and thus survive the lack of carbohydrates, plant cells must rapidly replace carbohydrate metabolism with both protein and lipid catabolism (Brouquisse et al., 1992, 1998; Aubert et al., 1996). Protein breakdown in plant cells is mainly mediated by two different proteolytic systems, namely the selective ubiquitin proteasome-dependent proteolysis and the non-selective vacuolar proteolysis known as self-cannibalism or autophagy (for reviews, see Vierstra, 1996; Brouquisse et al., 2001). The central plant vacuole that contains hydrolytic enzymes, like its animal equivalent, the lysosome, is involved in autophagy that is critical for survival during stress conditions (Marty, 1999).
The autophagic process is a bulk degradative pathway involving: (i) the sequestration of portions of the cytoplasm by endomembranes, thus forming an autophagic vacuole; and (ii) the subsequent degradation of the trapped cell components in order to recycle their constituents for maintaining essential functions and cellular homoeostasis.
In all eukaryotic organisms, autophagy is ubiquitous and operates either in normal cellular and developmental programmes (for reviews, see Moriyasu and Hillmer, 2001; Reggiori and Klionsky, 2002; Wang and Klionsky, 2003) or in response to various environmental stresses such as starvation (Dunn, 1994; Aubert et al., 1996; Moriyasu and Ohsumi, 1996; Codogno et al., 1997; Marty, 1997; Doelling et al., 2002; Hanaoka et al., 2002).
According to the size of the autophagic vacuoles, two types of autophagic processes have been described in eukaryotic cells (for a review, see Klionsky and Oshumi, 1999). Microautophagy is characterized by invaginations of the vacuolar membrane and direct engulfment of the cytoplasm by the vacuole where it is degraded by resident proteases. In contrast with microautophagy, macroautophagy involves the sequestration of large portions of the cytoplasm by double-membrane-bound compartments called autophagosomes. In yeast, the outer membrane of the autophagosome fuses with the vacuolar membrane. The autophagic body, composed of the sequestered material bound by the inner membrane of the autophagosome, is delivered to the lumen of the vacuole to be degraded. In mammalian cells, autophagosomes were described as fusing with lysosomes, thus forming new lytic compartments, formally called autolysosomes.
Both types of autophagy have been described in plant cells and are involved in various developmental processes, including the formation of the vegetative vacuole in meristematic cells (Marty, 1978), the formation of protein storage vacuoles (Levanony et al., 1992), the transformation of protein storage vacuoles to vegetative vacuoles (Van der Wilden et al., 1980; Herman et al., 1981) and senescence (Matile and Winkenbach, 1971; Bethke and Jones, 1998). During periods of nutrient starvation, the autophagic process can be reinitiated in plant cells that are already vacuolated (Chen et al., 1994; Aubert et al., 1996; Moriyasu and Ohsumi, 1996; Moriyasu and Klionsky, 2003; Moriyasu et al., 2003). Carbohydrate starvation-induced autophagy is associated with an increase in intracellular proteolysis (James et al., 1996; Moriyasu and Ohsumi, 1996) and a marked degradation of membrane polar lipids (Aubert et al., 1996).
Although the autophagic process in plant cells was investigated some time ago using cytological and biochemical approaches (Villiers, 1967; Matile and Winkenbach, 1971; Marty, 1972; Matile, 1975), the molecular mechanisms underlying this fundamental process remain unknown. Knowledge of the molecular machinery of autophagy has benefited from the study of autophagy defective mutants (apg or aut) in yeast (Tsukada and Oshumi, 1993; Thumm et al., 1994). Most of the autophagy defective mutants are impaired in autophagosome formation and cannot survive starvation conditions. Several mechanisms essential for yeast autophagosome formation have been characterized from studies of the gene products. Among these mechanisms, the Atg8 conjugation pathway is an ubiquitin-like system that allows the transient covalent binding of the soluble Atg8p protein to PE (phosphatidylethanolamine) (Ichimura et al., 2000; Kirisako et al., 2000). Soluble Atg8p is first processed by the cysteine protease Atg4p, thus removing the terminal arginine residue and exposing a glycine residue at the C-terminal end. Processed Atg8p is then activated by the E1-like enzyme Atg7p through a thioester bond, before being transferred to the E2-like enzyme Atg3p via a new thioester bond. Atg8p is finally covalently conjugated with PE on the autophagosome membrane via an amide bond. Upon completion of autophagosome formation, the amide bond linking Atg8p and PE may be cleaved by the Atg4p protease (Kirisako et al., 1999, 2000), thus allowing the free Atg8p to participate in a new cycle of autophagosome formation.
The sequencing of the Arabidopsis genome has allowed the identification of 25 AtATG genes orthologous to 12 of the 15 yeast autophagy genes (Hanaoka et al., 2002), suggesting that the autophagy pathway operates similarly in plants. Every gene involved in the Atg8 conjugation pathway in yeast has at least one orthologue in the Arabidopsis genome (Hanaoka et al., 2002). Both the E1-like protein- and the E2-like protein-coding genes ATG7 and ATG3 have only one orthologue in Arabidopsis, whereas the protease-coding gene ATG4 has two orthologues (AtATG4a and AtATG4b). Nine orthologues (AtATG8a–AtATG8i) of the yeast ubiquitin-like protein-coding gene ATG8 are present in Arabidopsis. Comparison of the four proteins involved in the Atg8 conjugation system in yeast (Atg3p, Atg4p, Atg7p and Atg8p) with their orthologous Arabidopsis genes reveals that the functional domains and essential amino acid sequences are well-conserved, further supporting the idea that the Atg8 system functions in a similar manner in plants and yeasts.
We combined structural and molecular methods to examine the physiological role of autophagy in cells subjected to sucrose starvation. We carried out a co-ordinated and systematic analysis of early cytological and molecular events in cells grown under nutrient-limiting conditions. Here, we describe the effects of sucrose starvation on growth and protein metabolism of cultured Arabidopsis cell cultures. We further characterize the formation and fate of autophagic vacuoles (autophagosomes), and report the kinetics of expression of five AtATG genes. To our knowledge, this is the first report that demonstrates that autophagy-related genes operate in a co-ordinated manner in plants. Description of these early molecular and structural events may provide important tools for understanding the pathway leading to autophagosome formation.
Effects of sucrose starvation on cell-suspension growth and cell morphology
Cells from a 4-day-old exponentially growing suspension culture were transferred to a sucrose-free medium or to a medium supplemented with 1.5% (w/v) sucrose. After 5 days of culture in the sucrose-containing medium, the cell number and dry weight increased by 25.6- and 4.7-fold respectively (Figures 1A and 1B), with a generation time of 3.4 days. During the same period, the mass of total cellular proteins increased by 1.9-fold (Figure 1C). In contrast, both cell divisions (as indicated by cell number; Figure 1A) and accumulation of biomass (Figure 1B) stopped immediately after the imposition of sucrose starvation. During the first day of starvation, 32% of the total protein was degraded (Figure 1C). Protein degradation was less rapid during the following days; the cells had lost 66% of their total protein content after 5 days of starvation. Cell viability was not altered during the first day despite the high protein degradation (Figure 1D) but significantly decreased between days 2 and 5, so that 34% of the cells had died after 5 days (Figure 1D).
The morphology of the cells after 1 day of culture was observed by Nomarski interference microscopy (Figures 2A–2D) and confocal laser-scanning microscopy (Figures 3A–3F). Under sucrose-supplemented conditions, numerous large transvacuolar strands of cytoplasm connecting the peripheral and perinuclear regions were observed (Figures 2A and 2B, arrows). In sucrose-free medium, the width of the cytoplasm was markedly decreased while the vacuole enlarged (Figures 2C and 2D). Concomitantly, the number of transvacuolar strands decreased dramatically.
Acidic compartments were detected in cells after incubation in fluorescent weak bases (Figures 3A–3F). Cells grown either in sucrose-rich or sucrose-depleted medium accumulated quinacrin, as detected by green fluorescence, in their large central vacuole (Figures 3B, 3C, 3E and 3F). In cells grown in sucrose-free medium, intensely fluorescent spots were also detected (compare Figures 3B and 3C with Figures 3E and 3F, arrowheads). These starvation-induced acidic compartments were most frequently seen in the dense perinuclear cytoplasm in close contact with the large central vacuole.
Formation of autophagic vacuoles is induced by sucrose starvation
Cells grown in sucrose-replete and sucrose-free media for 12, 24, 62 and 120 h were examined using EM (electron microscopy). There was a wide diversity of autophagy-related structures in the cytoplasm of cells deprived of sucrose for 12 h (Figures 4B and 5B–5E). Numerous (sometimes more than 100) vesicles limited by a single membrane, ranging from 100 to 300 nm in diameter, were clustered in starved cells (Figures 4B and 5A). Such clusters were not seen in control cells (Figure 4A).
Autophagic vacuoles were 0.6–2.5 μm in diameter (Figure 4B and inset and Figures 5B–5E), and were frequently located close to the Golgi stacks. Three classes of autophagic vacuoles were tentatively defined according to their internal content and membrane boundary. Class 1 autophagic vacuoles consist of double-membrane-bound organelles sequestering a portion of the intact cytoplasm, occasionally with various trapped organelles [e.g. mitochondria and ER (endoplasmic reticulum); Figure 5B]. Class 2 contains cytoplasmic material at different stages of degradation (Figure 5C) and remnants of membranes in various arrangements (Figure 5D). In class 3 autophagic vacuoles, the sequestered portion of the cytoplasm is totally digested, resulting in a clear vacuolar sap (Figure 5E). Internal membranes are the last structures to disappear. The resulting empty vacuoles finally protruded in the central vacuole, into which they were occasionally seen to be expelled.
In the continuous process of autophagy, it is clear that each individual autophagic vacuole proceeds through the different functional stages depicted as class 1, class 2 and class 3 profiles. Because the EM pictures are ‘frozen’ stages of a continuous and dynamic sequence, the frequency of autophagic vacuoles in each class depends on the kinetics of each step, where the most frequent profiles correspond to the slowest step.
There is no precedent for this functional classification of autophagic structures. To our knowledge, this is the first report that presents the nomenclature this way.
Although polymorphic autophagic vacuoles were most frequently seen after 12–24 h of sucrose starvation, they were also occasionally present in control cells, but to a lesser extent. To determine whether the observed autophagy-related structures could specifically result from sucrose starvation, we compared the numbers of autophagic vacuoles and vesicles in starved cells with those in control cells (see the Materials and methods section). This quantitative study enabled a correlation between the number of autophagic vacuoles and sucrose starvation. After 12 h of sucrose starvation, the number of both autophagic vacuoles and clustered vesicles had risen dramatically (Figure 6). They were still abundant after 24 h of starvation, but then decreased progressively.
Sucrose removal from the growth medium triggers a rapid and massive but transient burst of autophagic activity in starving cells. Because carbon availability is crucial for survival, cells respond by a point-of-no-return mechanism that adapts to the stress signal. In the first 24 h after the onset of starvation, cells drive an intense autophagic recycling of intracellular substrates to meet their vital needs. Under our operational conditions, this autophagic burst makes carbon resources sufficient for the cell to survive for 8 days. Up to this time, the cells are able to recover if sucrose is added back to the sucrose-depleted medium. Should the release of carbon facilitated by intracellular digestion be insufficient, the cells cannot energetically sustain the autophagic process and will senesce.
Relative expressions of the ATG8 conjugation pathway genes
To determine whether genes encoding proteins from the ATG8 conjugation pathway are related to the starvation-induced autophagic process, 4-day-old exponentially growing cultured cells were transferred to either sucrose-supplemented or sucrose-free medium. After various time periods up to 5 days, cells were collected and specific mRNAs were analysed by real-time RT (reverse transcriptase)–PCR.
The relative expression patterns of genes encoding components of the ATG8 pathway were compared with that of the AS (asparagine synthetase) gene AtASN1. An increase in AS gene expression was previously reported to be related to a depletion of carbon resources (Davies et al., 1996; Downs and Somerfield, 1997).
A net increase in the relative expression levels of all genes involved in the ATG8 conjugation pathway was observed when cells were deprived of sucrose (Figures 7A and 7B). Although all these genes were expressed at a basal level in control cells, starvation conditions triggered a transient stimulation of all of them during the first 2 days of culture with subtle differences in expression patterns (Figure 7A).
The AtATG4a and AtATG4b genes were the first genes whose relative expressions were stimulated by starvation. Both genes have the same kinetics of expression. The maximum relative expression was observed after 9 h of culture and then rapidly decreased. Only small changes were observed from 12 h for AtATG4a or from 18 h for AtATG4b, up to 120 h. AtATG8a–AtATG8i orthologues were differentially expressed under starvation conditions. AtATG8a, AtATG8c and AtATG8g–AtATG8i genes were up-regulated with similar kinetics of relative expression during the first 12 h of culture, whereas AtATG8b and AtATG8d–AtATG8f showed little relative expression during the same period of time (Figure 7A). Among the expressed AtATG8 genes, AtATG8a and AtATG8i exhibited the highest relative expression levels with an 8-fold increase during the first 18 h of starvation, followed by a drastic decrease over the next 6 h. No significant changes were observed after 48 h in culture. In contrast, the relative expression of AtATG7 continued to increase until 24 h after the onset of sucrose starvation. It then decreased rapidly and was low after 48 h of culture. The relative expression of AtATG3 was not stimulated until 12 h after imposition of sucrose starvation. The increase in relative expression was transient with a maximum reached at 24 h. From 48 h onwards, no significant differential expression of AtATG3 was observed between control and starved cells.
The transcription of the AtASN1 gene was also stimulated in sucrose-starved cells, but its relative expression pattern was strikingly different (Figure 7A). The AtASN1 gene was weakly up-regulated during the first 48 h of culture in sucrose-depleted medium. Its relative expression rapidly increased between 48 and 72 h, then remained at a high level until at least 120 h.
Relative expression of all genes involved in the AtATG8 conjugation pathway thus peak as successive waves in cells in the early stages of starvation (Figure 7B). However, this pattern of gene expression is not repeated. Interestingly, the evanescent waves of gene expression immediately precede and in some cases overlap with the transient occurrence of autophagosomes as described in the previous subsection. Lastly, AtASN1 gene expression occurs at the end of the autophagic activity when deleterious amounts of NH4+ have accumulated.
These results strongly suggest that starvation conditions triggered the expression of a set of genes that could be involved in the formation of autophagosomes, as described previously in yeast and mammals (Mizushima et al., 1998; Kabeya et al., 2000; Doelling et al., 2002; Hanaoka et al., 2002; Yoshimoto et al., 2004). Although the amount of asparagine was not quantified in the cells, the accumulation of transcripts suggests that a de novo asparagine synthesis is probably induced to cope with the toxic effects of NH4+ that are produced after autophagic protein degradation.
Autophagy in physiological processes
Morphological, biochemical and genetic evidences have shown the importance of autophagy during plant development and in response to stress. Autophagic mechanisms are involved in cellular remodelling during normal cell differentiation and development (for a review, see Moriyasu and Klionsky, 2003). For example, different autophagosome-mediated pathways are likely to be responsible for the formation of vacuoles in differentiating meristematic cells (Marty, 1999; Moriyasu and Hillmer, 2001 for reviews) and cotyledon cells (Herman and Larkins, 1999; Chrispeels and Herman, 2000 for reviews). Vacuolar autophagy may also operate as a quality control mechanism for removal of organelles with abnormal functions in active or senescing organs (Niwa et al., 2004).
The role of autophagy in adaptation to starvation has been investigated mainly because of its physiological impact on basal metabolism and biomass production. The intracellular breakdown of cytoplasmic components by autophagy produces amino acids, phospholipids and other elements needed for basic metabolism and essential biosynthetic pathways. In suspension-cultured sycamore cells deprived of sucrose, the resulting shortage of respiratory substrates triggers an autophagic mechanism that is responsible for a massive degradation of the cytoplasm and a concomitant accumulation of free amino acids accompanied by a massive breakdown of membrane polar lipids.
Autophagy, associated with an increase in endopeptidase activities, was also shown to contribute to intracellular protein degradation in tobacco suspension-cultured cells (BY-2) under sucrose starvation conditions (Moriyasu and Ohsumi, 1996; Takatsuka et al., 2004). However, in suspension-cultured BY-2 cells, net protein degradation was not observed during the first 24 h of starvation. Similarly, net protein degradation was reported to occur in sucrose-starved sycamore cells after a lag phase of at least 24 h when all intracellular carbohydrates had disappeared (Journet et al., 1986). In suspension-cultured Arabidopsis cells, we show that more than one third of the proteins was degraded during the first 24 h of sucrose starvation, indicating that proteolysis occurred sooner than in tobacco and sycamore cells. Vacuolar proteolysis was suggested to be involved in the degradative processes that take place in the root system of maize plants (Zea mays L.) subjected to prolonged darkness as well as in sugar-starved, excised roots (James et al., 1996; Brouquisse et al., 1998).
A non-specific proteome-wide breakdown can play an important survival role by maintaining the supply of amino acids under nutrient-limiting conditions. Indeed, plant cells degrade proteins by a variety of pathways. The ubiquitin/proteasome pathway is involved in the selective degradation of cytosolic and nuclear proteins (for a review, see Smalle and Vierstra, 2004). However, cells rely on vacuolar/lysosomal pathways of proteolysis for non-selective protein degradation in many instances, including starvation (Aubert et al., 1996; Moriyasu and Ohsumi, 1996). Vacuolar proteases are stable and active at acidic pH and thus are optimized to function in acidic intracellular compartments. We suggest that, following sucrose deprivation, bulk protein degradation in Arabidopsis cells involves vacuolar proteases. We found that small intracellular acidic vacuoles optimal for acid protease activity were formed rapidly after the removal of sucrose from the culture medium. They could play a major role in protein turnover that becomes enhanced during stresses such as starvation (for a review, see Kotyza and Krepela, 2002). The acidic vesicles were only seen in the presence of protease inhibitors in BY-2 cells (Moriyasu and Ohsumi, 1996). Because they are most frequently seen when their degradation is experimentally slowed down by inhibitors, we suggest that they are short-lived structures under natural (non-inhibitory) conditions.
Other vacuolar protein degradation pathways have been described (Majeski and Dice, 2004) and might operate in plants. The different mechanisms by which these degradation pathways might interact are unknown.
Although initially suspected by biochemical markers, evidence for cellular autophagy is best obtained by EM. Because the cells we studied were not grown synchronously, induced autophagic processes were not in phase and, consequently, autophagic vacuoles were highly polymorphic, ranging from double-membrane-bound autophagosomes with intact cytoplasmic content to empty single-membrane-bound vacuoles. A complete autophagic sequence can be reconstructed from the randomly observed profiles: (i) an early stage of cytoplasmic sequestration without noticeable digestion is seen in class 1 autophagic vacuoles; (ii) digestion occurs in class 2 autophagic vacuoles and the sequestered cytoplasm becomes degraded; and (iii) the cytoplasm is completely digested in class 3 autophagic vacuoles. Thus new starvation-induced lytic vacuoles are formed focally at the expense of the cytoplasm. In a given cell, autophagic vacuoles belong predominantly to the same class. We provide here, for the first time, a clear functional classification for the highly polymorphic autophagosomes encountered in such studies.
Homotypic fusion of the newly generated autophagic vacuole with the preexisting large central vacuole accounts for both tonoplast extension and increase in sap volume as observed in starving cells. The resulting gain in membrane surface area may accommodate an influx of water and additional vacuolar swelling without further membrane synthesis. Water channels in the tonoplast of the central vacuole (Reisen et al., 2003) and in the outer membrane of the autophagosome (Moriyasu et al., 2003) may facilitate water movement immediately after the fusion step.
In Arabidopsis cells, membrane fusion occurs generally after the materials inside the autophagic vacuole, including any inner membrane, are completely digested. Therefore inner membrane-bound autophagic bodies were rarely seen within the central vacuole in contrast with our earlier observations in sycamore cells (Aubert et al., 1996). It is assumed that substrates are being degraded in lysosomes with an invariable half-life of 5–10 min (for a review, see Marzella and Glaumann, 1987). This would imply that delivery and fusion of late autophagosomes to the central vacuole would be slower in Arabidopsis cells than in sycamore cells.
The results of the present study indicate that fusion of the autophagosome with the lytic central vacuole is not needed for the complete digestion of the sequestered substrates. Autophagosomes on their own are functionally self-sufficient to achieve the breakdown of the sequestered materials as documented by cytochemical studies (Marty, 1978). However, fusions may occur opportunistically between autophagosomes at various stages of their ontogeny.
Clusters of acidic vesicles were specifically induced in starving cells. Although their origin and fate were not elucidated, we suggest that they could be progenitors of autophagic vacuoles because they appeared only 12 h after deprivation of sucrose, and co-exist with autophagosomes in starving cells up to their decline. This distribution pattern suggests that they are continuously formed and used to build the autophagosome in a steady-state manner, consistent with their role as pre-autophagosomal structures.
Biological functions of ATG gene products are well conserved between yeast and plants, and orthologues acting in the Atg8 conjugation pathway could be good markers for membrane dynamics at the early step of autophagosome formation. If cellular autophagy contributes to the protein degradation process during sucrose starvation, we reasoned that genes involved in the formation of autophagosomal membranes would be expressed earlier than genes encoding enzymes involved in the recycling of the final degradation products. As an alternative to the difficult genetic approach of autophagy regulation in plants, we monitored the real-time expression of these genes in cells challenged with starvation.
Our results show that AtATG3, AtATG4a, AtATG4b, AtATG7 and AtATG8a–AtATG8i are expressed at a basal level in nutrient-supplemented cells that are growing, indicating that a functional series of autophagy-related genes, orthologues to those discovered in yeast and mammal cells, are expressed in plants. Furthermore, they indicate, as already shown by morphological studies, that autophagy is constitutive even in vacuolated cells (Marty, 1999). In plant cells, as in animal and yeast cells, this housekeeping function participates in cell homoeostasis by controlling the turnover of cytoplasmic components at a steady-state level. In contrast, the ATG genes encoding the ATG8 conjugation pathway were transiently up-regulated by starvation in a precise sequential manner. This supports the role of autophagy in the response of plant cells to starvation. The breakdown of proteins, carbohydrates, lipids, and RNAs by autophagy produces elements for intermediary metabolism and biosynthetic pathways that are required for survival.
The expression of the AtATG genes preceded the biochemical and cellular events conventionally related to autophagy. The kinetics of accumulation of the mRNAs is in good agreement with the time frame of the cellular and molecular events known for the Atg8 conjugation pathway in nutrient-starved yeast (Kirisako et al., 1999; Ichimura et al., 2000). We found that the proteases encoding AtATG4 genes are the earliest activated genes, followed successively by the ubiquitin-like encoding AtATG8a, the E1-like encoding AtATG7 and the E2-like encoding AtATG3 when Arabidopsis cells are transferred to sucrose-depleted medium.
In the yeast Saccharomyces cerevisiae, proteins of the Atg8 conjugation pathway have been shown to act in autophagosome formation. In plants, no direct evidence to date has shown how these proteins contribute to autophagosome assembly. The present study establishes for the first time, to our knowledge, a detailed correlation between the up-regulation of genes involved in the Atg8 conjugation pathway and the formation of autophagosomes in plant cells. From this relationship, we tentatively speculate on the molecular mechanisms underlying the structural processes. As its orthologue does in yeast, AtATG8 probably conjugates with PE by ubiquitination-like reactions involving AtATG4, AtATG7 and AtATG3 respectively. AtATG8 conjugated with PE (AtATG8–PE) would be membrane-anchored to yet unknown pre-autophagosomal structures. This proteinaceous membrane anchor could serve as a skeletal scaffold or primer for sophisticated organelle assembly such as tubules and cage-like structures. Phenotypic analysis of the mutants impaired in autophagosome assembly will help to dissect the early autophagic process.
Nitrogen resulting from sucrose starvation-induced proteolysis is primarily stored as asparagine whose synthesis is catalysed by AS (Brouquisse et al., 1992; Chevalier et al., 1996; Lam et al., 1996; Brouquisse et al., 1998). Asparagine accumulation occurs when plants have to detoxify large amounts of ammonia released from the deamination of soluble amino acids after proteolysis (for reviews, see Givan, 1979; Sieciechowicz et al., 1988). In Arabidopsis cells, AtASN1 mRNA began to accumulate steadily only 48 h after sucrose deprivation, clearly later than the transient accumulation of mRNA specific for AtATG3, AtATG4a, AtATG4b, AtATG7, and AtATG8a–AtATG8i. The presence of an increasing amount of AtASN1 mRNA after 48 h and during an extended period of starvation suggests that autophage-mediated protein degradation generating toxic amounts of NH4+ had already taken place.
Several groups have begun to explore molecular aspects of autophagy in plants (Doelling et al., 2002; Hanaoka et al., 2002; Surpin et al., 2003; Contento et al., 2004; Yoshimoto et al., 2004). Mutant plants defective in AtATG7 and AtATG9, two of the 12 autophagy genes described in Arabidopsis, display similar phenotypes (Doelling et al., 2002; Hanaoka et al., 2002). Under nutrient-limiting conditions, both mutants display premature leaf senescence with early chlorosis and a reduction in seed set. Even under nutrient-replete conditions, bolting and natural leaf senescence are accelerated in atatg9 mutant plants. Taken together with these previous results, our findings imply that autophagy, although not essential for normal growth and development, is required for efficient nutrient recycling on a whole plant scale and, therefore, for maintenance of the viability of the organism under nutrient-limited conditions.
The starvation-induced expression of autophagy genes encoding components of the conjugation pathway supports the hypothesis that autophagic mechanisms are well conserved in uni- as well as multicellular eukaryotic organisms.
Starvation-induced autophagy has been shown to occur in sink tissues earlier than in mature tissues (Brouquisse et al., 1998) and immunoblotting of duplicated gene products was reported to show an organ-specific pattern (Hanaoka et al., 2002). Here, we have reported that several gene isoforms were co-expressed, but at different rates, in undifferentiated suspension-cultured cells (e.g. AtATG8a–AtATG8i). We expect individual isoforms to be specifically expressed according to tissue, cell-developmental age or stress conditions in whole plants. Structurally related but not totally redundant proteins may be preferentially used for autophagy in specific physiological situations, illustrating the great adaptability of plants to changing conditions. Moreover, duplication of certain autophagy genes in plants may indicate several different autophagy pathways.
Materials and methods
Arabidopsis thaliana suspension-cultured cells (T87) (Axelos et al., 1992) were grown in modified Murashige and Skoog's (1962) medium containing 43.8 mM sucrose (1.5%; standard medium). Suspension cells were maintained by transferring 10 ml (1.6×106 cells·ml−1) of the cell suspension in stationary phase to 50 ml of fresh medium every 14 days. Cultures were grown in 250 ml flasks in the dark at 22±1°C with a rotation of 120 rev./min.
Exponentially growing cells (10 ml; 4-day-old) were collected by centrifugation at 100 g for 4 min. Cell pellets were washed in 100 ml of sucrose-free medium and, after two additional washing steps, the cells were finally resuspended in 50 ml of sucrose-free medium. Control cells were resuspended in standard sucrose-supplemented medium. Cells were then grown as described above.
Measurement of cell number, dry mass, total proteins and cell viability
Cells in 1 ml of suspension culture were collected and treated with 1 ml of enzymatic solution [2% cellulase (Onozuka R-10, Duchefa, Biochime B.V., Haarlem, The Netherlands) and 0.1% pectinase (Sigma—Aldrich, L'Isle-d'Abeau Chesnes, France) in 0.66 M sorbitol] for 45 min at 37°C in order to dissociate aggregates. Cell number was estimated by counting in a Fuchs—Rosenthal chamber under a light microscope. Duplicate measurements were performed on each sample and experiments were repeated for three different cultures.
For dry mass determination, cells in 50 ml of cell suspension were collected on a nylon filter (44 μm; Whatman, Maidstone, Kent, U.K.) and rapidly washed with water under vacuum filtration. The rinsed cells were transferred to a Petri dish and dried at 60°C for 48 h.
For the measurement of total protein content, cells were rinsed, immediately frozen in liquid nitrogen and ground with a prechilled mortar and pestle in homogenization buffer (50 mM Hepes/NaOH, pH 8, 1 mM EDTA, 1% Triton X-100, 100 mM PMSF, 1.5 μg·ml−1 aprotinin and 1 mM dithiothreitol). The homogenates were centrifuged at 1000 g for 5 min. The amount of protein in the supernatant was measured by the method of Bradford (Bradford, 1976) with BSA as a standard. A blank was prepared with complete homogenization buffer.
Dead and living cells were monitored by Evans Blue staining. The cell suspension (200 μl) was stained with 0.1% Evans Blue for 5 min at room temperature (20°C). Living cells were not stained, whereas dead cells were stained. Their respective percentages were estimated by counting under a light microscope.
Total RNA from suspension-cultured cells was purified by using the RNeasy mini kit (Qiagen, Courtaboeuf, France). Traces of DNA were removed using an RNase-free DNase set kit (Qiagen). The amount of total RNA was measured spectrophotometrically. The quality of the purified RNA was checked by visualizing the rRNA in the ethidium bromide-coloured agarose gel under UV light.
Poly(A)+ (polyadenylated) RNA was purified from total DNA-free RNA by using the poly(A)+ RNA kit (Qiagen).
Real-time RT—PCR amplification
The transcripts were quantified using fluorescence-based real-time RT—PCR assays. Purified poly(A)+ RNA (1 μg) from suspension-cultured cells was diluted with RNase-free water to a final volume of 14 μl and used for first-strand cDNA synthesis with oligo(dT)15. Briefly, all samples were supplemented with 1 μg of oligo(dT)15 and heat-treated for 5 min at 70°C, followed by 2 min incubation on ice. Samples were then mixed with 10 μl of reaction buffer (50 mM Tris/HCl, 75 mM KCl, 3 mM MgCl2 and 10 mM dithiothreitol; Promega, Charbonnières, France) containing 200 units of MMLV (Moloney-murine-leukaemia virus) RT (Promega) and 10 mM deoxynucleotides. cDNA synthesis was performed for 60 min at 42°C, followed by an enzyme inactivation step for 15 min at 70°C.
PCRs were performed using an iCycler iQ™ Multi-Color Real-Time PCR Detection System (Bio-Rad, Hemel Hempstead, Herts., U.K.) and qPCR™ Mastermix for Sybr™ Green I—No ROX (Eurogentec, Angers, France). The thermal cycling amplification conditions were 2 min at 50°C and 10 min at 95°C, followed by 40 cycles of two-step PCR (denaturing, 95°C for 15 s; and annealing, 60°C for 1 min). The fluorescent dye SYBR Green binds to the nascent double-stranded DNA, thus resulting in an increase in the fluorescence signal that can be monitored in real time. Fluorescence values were recorded during every cycle and represent the amount of product amplified to that point in the amplification reaction.
To quantify the results obtained by real-time RT—PCR, the comparative CT (threshold cycle) method (ΔΔCT) was used (Giulietti et al., 2001). CT corresponds to the point at which the fluorescent signal is first recorded as statistically significant above background (Gibson et al., 1996). This point always occurs during the exponential phase of the amplification. The more template present at the beginning of the reaction, the fewer number of cycles it takes to reach this point.
ΔCT values of the samples were determined by subtracting the average of triplicate CT values for the target gene in starved cells from the average of triplicate CT values for the target gene in control cells. To correct the sample-to-sample variations, the relative gene expression levels were normalized against the expression of the cyclophilin constitutively expressed gene AtROC5 (Hecht et al., 2001). AtASN1, a gene encoding type I AS in Arabidopsis, was chosen as a gene marker for protein degradation in cells (Genix et al., 1990).
Specific primers used for this quantification were: AtASN1 (At3g47340), 5′-GGGATTGATGCGATAGAGGA-3′ and 5′-TTGCGACAAGTTTCTTGGTG-3′; AtATG3 (At5g61500), 5′-TCATCCACACTTGCCTGGTA-3′ and 5′-CCGAGATCAAAGTCCATTGTG-3′; AtATG4a (At2g44140), 5′-GGCTGCATTGCAACTAGATTT-3′ and 5′-GAATCATGCAACCCCAGTTC-3′; AtATG4b (At3g59950), 5′-CTTTCACGTTCCCTCAAAGC-3′ and 5′-TTGCAATGGTAAGACGATGTG-3′; AtATG7 (At5g45900), 5′-GTACCGCTTGCTCTGAAACC-3′ and 5′-GTCTTCCCAGTCGAGGTTGA-3′; AtATG8a (At4g21980), 5′-CAATTTGTATACGTGGTTCGT-3′ and 5′-AGCAACGGTAAGAGATCCAA-3′; AtATG8b (At4g04620), 5′-TTGGCCAATTTGTGTACGTT-3′ and 5′-TCCACCAAATGTGTTCTCTCC-3′; AtATG8c (At1g62040), 5′-TGAGTGCCGAAAAGGCTATC-3′ and 5′-ACCAAACCAAAGGTGTTCTCT-3′; AtATG8d (At2g05630), 5′-TTTGACTGTTGGCCAGTTTG-3′ and 5′-AACCCGTCTTCGTCTTTGTG-3′; AtATG8e (At2g45170), 5′-TCTTTAAGATGGACGACGATTTC-3′ and 5′-CTCAGCCTTTTCCACAATCA-3′; AtATG8f (At4g16520), 5′-TGGGGCAGTTTGTGTATG-3′ and 5′-GGAACCCATCATCATCCTTTT-3′; AtATG8g (At3g60640), 5′-TGTGATTCGTAAGAGAATCCAAC-3′ and 5′-CCAAAAGTGTTTTCCCCACT-3′; AtATG8h (At3g06420), 5′-CCAAAGCTCTCTTTGTTTTCG-3′ and 5′-AAGAACCCGTCTTCTTCCTTG-3′; AtATG8i (At3g15580), 5′-TGTCAACAACACTCTCCCTCA-3′ and 5′-AACCAAAGGTTTTCTCACTGC-3′; AtROC5 (At4g34870), 5′-AAGCACGTTGTGTTTGGACA-3′ and 5′-AAGTCTCTCACTTTCTCACT-3′.
Living cells were examined in a Zeiss Axiophot photomicroscope (Carl Zeiss, Cologne, Germany) equipped with Nomarski differential interference contrast optics and a ×40 objective lens (NA=0.75).
For quinacrine staining, cells in 100 μl of suspension medium [the Murashige and Skoog's (1962) medium containing 43.8 mM sucrose] were washed by centrifugation through 5 mM Hepes/NaOH (pH 7.5) containing 0.1 M sorbitol. The resuspended cells were stained in the same solution containing 40 μM quinacrin for 5 min at room temperature. The cells were washed again with the same solution and observed.
Fluorescence microcopy was performed using a confocal microscope (TCS-SP2-AOBS; Leica Microsystems, Wetzlar, Germany). Excitation was provided by the 488 nm argon laser beam line. We observed the green fluorescence through a 506–530 nm bandpass filter. Arabidopsis cells were mounted in Hepes/NaOH buffer under a coverglass for microscopy. The laser was focused on individual cells through a ×20 objective lens (NA=0.80).
Aliquots (20 ml) of cell suspension were placed in Falcon-type tubes and cells were sedimented using a hand centrifuge (Hettich-type 1011) to form a loose pellet (0.5 ml). The supernatant was discarded and 1.5 ml of fixative solution (2.5%, v/v, glutaraldehyde in 0.1 M NaH2PO4/Na2HPO4 phosphate buffer, pH 7.2) was added. After 4 h at room temperature, cells were washed in the same buffer for 30 min and post-fixed with 1% (v/v) OsO4 in phosphate buffer for 4 h at 4°C. After washing in phosphate buffer for 30 min, cells were incubated in phosphate buffer containing 1% (w/v) tannic acid for 30 min at room temperature in the dark. Cells were then washed in distilled water, dehydrated in ethanol followed by propylene oxide and embedded in Araldite/Epon resin mixture. Silver-grey sections were cut on a Reichert-Jung Ultracut E ultramicrotome mounted on copper grids and stained in 2.5% (w/v) uranyl acetate in 50% (w/v) methanol for 30 min, followed by lead citrate for 15 min. Sections were viewed in an H-600 EM (Hitachi, Tokyo, Japan) at an accelerating voltage of 75 kV with a 30 μm objective aperture.
Cells harvested from sucrose-rich and sucrose-free media were prepared for EM as described in the previous subsection. Ultrathin sections were examined at low magnification (×5000) to select for complete cell sections. Autophagosomes and putatively related vesicular profiles present on complete cell sections were counted under both sucrose (+) and sucrose (−) conditions. Random sections taken through 30 different cells were analysed at each time point of a given time-course experiment that has been reproduced three times with similar results. Autophagy structures specifically induced by sucrose starvation were defined as the difference between the total number of autophagy-related structures present in sucrose-deprived cells and the number of autophagy-related structures present in sucrose-supplemented cells (control).
We thank B. Lescure (UMR 215 INRA/CNRS, Toulouse, France) for kindly providing plant material, D. Garmyn [Microbiology, ENSBANA (Ecole Nationale Supérieure de Biologie Appliquée à la Nutrition et à l'Alimentation), Dijon, France] for generous help with real-time PCR. EM was performed at the Central Facility of Microscopy Applied to Biology [SERCO-BIO (Sercice Commun de Biologie—Centre de Microscopie Appliquée à la Biologie), Université de Bourgogne, Dijon, France]. This work was supported by the Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche (MENESR), the Centre National de la Recherche Scientifique (CNRS), the Institut National de la Recherche Agronomique (INRA) and the Conseil Régional de Bourgogne. Support from the AFIRST (Association Franco-Israélienne pour la Recherche Scientifique et Technologique) and the COFECUB (Comité Français d'Evaluation de la Coopération Universitaire avec le Brésil) is also acknowledged. T.L.R. received a fellowship from Coordenação de Aperfeiçoamento de Pessoal de Nivel Superior (Government of Brazil).
- Modifications of etioplasts in cotyledons during prolonged dark growth of sugar beet seedlings Plant Physiol. 1994 106 1555–1565
- Ultrastructural and biochemical characterization of autophagy in higher plant cells subjected to carbon deprivation: control by the supply of mitochondria with respiratory substrates J. Cell Biol. 1996 133 1251–1263
- A protocol for transient gene expression in Arabidopsis thaliana protoplasts isolated from cell suspension cultures Plant Physiol. Biochem. 1992 30 123–128
- Gibberellin signaling Curr. Opin. Plant Biol. 1998 1 440–446
- A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding Anal. Biochem. 1976 72 248–254
- Asparagine metabolism and nitrogen distribution during protein degradation in sugar-starved maize root tips Planta 1992 188 384–395
- Induction of a carbon-starvation-related proteolysis in whole maize plants submitted to light/dark cycles and to extended darkness Plant Physiol. 1998 117 1281–1291
- Regulation of protein degradation and protease expression by mannose in maize root tips. Pi sequestration by mannose may hinder the study of its signaling properties Plant Physiol. 2001 125 1485–1498
- Expression of α-amylases, carbohydrate metabolism, and autophagy in cultured rice cells is coordinately regulated by sugar nutrient Plant J. 1994 6 625–636
- Metabolic regulation of asparagine synthetase gene expression in maize (Zea mays L.) root tips Plant J. 1996 9 1–11
- Endoplasmic reticulum-derived compartments function in storage and as mediators of vacuolar remodeling via a new type of organelle, precursor protease vesicles Plant Physiol. 2000 123 1227–1234
- Signal transduction pathways in macroautophagy Cell Signal. 1997 9 125–130
- Transcriptome profiling of the response of Arabidopsis suspension culture cells to Suc starvation Plant Physiol. 2004 135 2330–2347
- Sugar regulation of harvest-related genes in asparagus Physiol. Plant 1996 111 877–883
- The APG8/12-activating enzyme APG7 is required for proper nutrient recycling and senescence in Arabidopsis thaliana J. Biol. Chem. 2002 277 33105–33114
- Asparagine synthetase gene expression increases as sucrose declines in broccoli after harvest N. Z. J. Crop. Hortic. Sci. 1997 25 191–195
- Autophagy and related mechanisms of lysosome-mediated protein degradation Trends Cell Biol. 1994 4 139–143
- Modeling of allocation and balance of carbon in walnut (Juglans regia L.) seedling during heterotrophy-autotrophy transition J. Theor. Biol. 1998 194 29–47
- Transient accumulation of asparagine in sycamore cells after a long period of sucrose starvation Plant Physiol. 1990 94 717–722
- A novel method for real time quantitative RT—PCR Genome Res. 1996 6 995–1001
- An overview of real-time quantitative PCR: applications to quantify cytokine gene expression Methods 2001 25 386–401
- Metabolic detoxification of ammonia in tissues of higher plants Phytochemistry 1979 18 375–382
- Leaf senescence and starvation-induced chlorosis are accelerated by the disruption of an Arabidopsis autophagy gene Plant Physiol. 2002 129 1181–1193
- The Arabidopsis somatic embryogenesis receptor kinase 1 gene is expressed in developing ovules and embryos and enhances embryogenic competence in culture Plant Physiol. 2001 127 803–816
- Protein storage bodies and vacuoles Plant Cell 1999 11 601–614
- Uptake and apparent digestion of cytoplasmic organelles by protein bodies (protein storage vacuoles) in mung bean cotyledons Eur. J. Cell Biol. 1981 24 226–235
- A ubiquitin-like system mediates protein lipidation Nature (London) 2000 408 488–492 et al
- Purification and biochemical characterization of a vacuolar serine endopeptidase induced by glucose starvation in maize roots Biochem. J. 1996 320 283–292
- Biochemical changes during sucrose deprivation in higher plant cells J. Biol. Chem. 1986 261 3193–3199
- LC3, a mammalian homologue of yeast Apg8p, is localized in autophagosome membranes after processing EMBO J. 2000 19 5720–5728
- Formation process of autophagosome is traced with Apg8/Aut7p in yeast J. Cell Biol. 1999 147 435–446
- The reversible modification regulates the membrane-binding state of Apg8/Aut7 essential for autophagy and the cytoplasm to vacuole targeting pathway J. Cell Biol. 2000 151 263–276
- Vacuolar import of proteins and organelles from the cytoplasm Annu. Rev. Cell Dev. Biol. 1999 15 1–32
- A unified nomenclature for yeast autophagy-related genes Dev. Cell 2003 5 539–545 et al
- Proteases and antiproteases in health and disease—a review III. Cysteine proteases and their natural inhibitors Biomarkers Environment 2002 5 1–21
- Impact of daily photosynthetic period on protein synthesis and carbohydrate stores in Zostera marina L. (eelgrass) roots: implications for survival in light-limited environments J. Exp. Mar. Biol. Ecol. 1995 185 191–202
- The molecular-genetics of nitrogen assimilation into amino acids in higher plants Annu. Rev. Plant Physiol. Plant Mol. Biol. 1996 47 569–593
- Evidence for a novel route of wheat storage proteins to vacuoles J. Cell Biol. 1992 119 1117–1128
- Mechanisms of chaperone-mediated autophagy Int. J. Biochem. Cell Biol. 2004 36 2435–2444
- Distribution des activités phosphatasiques acides au cours du processus d'autophagie cellulaire dans les cellules du méristème radiculaire d'Euphorbia characias L C.R. Acad. Science (Paris) 1972 274D 206–209
- Cytochemical studies on GERL, provacuoles, and vacuoles in root meristematic cells of Euphorbia characias Proc. Natl. Acad. Sci. U.S.A. 1978 75 852–856
- The biogenesis of vacuoles: insights from microscopy Adv. Bot. Res. 1997 25 1–42
- Plant vacuoles Plant Cell 1999 11 587–599
- Glaumann H. Ballard F.J. Autophagy, microautophagy, and crinophagy as mechanisms for protein degradation Lysosomes: Their Role in Protein Breakdown 1987 New York Academic Press 319–367
- The Lytic Compartment of Plant Cells, Cell Biology Monographs, vol. 1 1975 Wien, Germany Springer-Verlag 183
- Function of lysosomes and lysosomal enzymes in the senescing corolla of the morning glory (Ipomoea purpurea) J. Exp. Bot. 1971 22 759–771
- A new protein conjugation system in human. The counterpart of the yeast Apg12p conjugation system essential for autophagy J. Biol. Chem. 1998 273 33889–33892
- Robinson D.G. Rogers J.C. Autophagy and vacuole formation Vacuolar Compartments 2000 71–89 Annual Plant Reviews
- Klionsky D.J. Autophagy in plants Autophagy 2003 Georgetown, TX Landes Bioscience 208–215
- Autophagy in tobacco suspension-cultured cells in response to sucrose starvation Plant Physiol. 1996 111 1233–1241
- Alpha tonoplast intrinsic protein is specifically associated with vacuole membrane involved in an autophagic process Plant Cell Physiol. 2003 44 795–802
- A revised medium for rapid growth and bioassays with tobacco tissue cultures Physiol. Plant 1962 15 473–497
- Disposal of chloroplasts with abnormal function into the vacuole in Arabidopsis thaliana cotyledon cells Protoplasma 2004 223 229–232
- Autophagy in the eukaryotic cell Eukaryotic Cell 2002 1 11–21
- Expression of a cauliflower tonoplast aquaporin tagged with GFP in tobacco suspension cells correlates with an increase in cell size Plant Mol. Biol. 2003 52 387–400
- The metabolism of asparagine in plants Phytochemistry 1988 27 663–671
- The ubiquitin 26S proteasome proteolytic pathway Annu. Rev. Plant Physiol. Plant Mol. Biol. 2004 55 555–590
- The VTI family of SNARE proteins is necessary for plant viability and mediates different protein transport pathways Plant Cell 2003 15 2885–2899 et al
- 3-Methyladenine inhibits autophagy in tobacco culture cells under sucrose starvation conditions Plant Cell Physiol. 2004 45 265–274
- Isolation of autophagocytosis mutants of Saccharomyces cerevisiae FEBS Lett. 1994 349 275–280
- Isolation and characterization of autophagy-defective mutants of Saccharomyces cerevisiae FEBS Lett. 1993 333 169–174
- Protein bodies of mung bean cotyledons as autophagic organelles Proc. Natl. Acad. Sci. U.S.A. 1980 77 428–432
- Proteolysis in plants: mechanisms and functions Plant Mol. Biol. 1996 32 275–302
- Cytolysomes in long-dormant plant embryo cells Nature (London) 1967 214 1356–1357
- The molecular mechanism of autophagy Mol. Med. 2003 9 65–76
- Processing of ATG8s, ubiquitin-like proteins, and their deconjugation by ATG4s are essential for plant autophagy Plant Cell 2004 16 2967–2983