Autophagy is present at a basal level in all plant tissues and is induced during leaf ageing and in response to nitrogen (N) starvation. Nitrogen remobilization from the rosette to the seeds is impaired in autophagy mutants. This report focuses on the role of autophagy in leaf N management and proteolysis during plant ageing.
Metabolites, enzyme activities and protein contents were monitored in several autophagy-defective (atg) Arabidopsis mutants grown under low and high nitrate conditions.
Results showed that carbon (C) and N statuses were affected in atg mutants before any senescence symptoms appeared. atg mutants accumulated larger amounts of ammonium, amino acids and proteins than wild type, and were depleted in sugars. Over-accumulation of proteins in atg mutants was selective and occurred despite higher endopeptidase and carboxypeptidase activities. Specific over-accumulation of the ribosomal proteins S6 and L13 subunits, and of catalase and glutamate dehydrogenase proteins was observed. atg mutants also accumulated peptides putatively identified as degradation products of the Rubisco large subunit and glutamine synthetase 2 (GS2). Incomplete chloroplast protein degradation resulting from autophagy defects could explain the higher N concentrations measured in atg rosettes and defects in N remobilization.
It is concluded that autophagy controls C : N status and protein content in leaves of Arabidopsis.
Autophagy is a process found ubiquitously in eukaryotic cells that uses specific molecular machinery incorporating the products of AUTOPHAGY (ATG) genes for the formation of double-membrane vacuoles, named autophagosomes (Li & Vierstra, 2012; Yoshimoto, 2012). Autophagosome biogenesis starts with the nucleation of lipid membranes which enlarge and surround the cell material to be discarded. After enclosure, autophagosomes are targeted to lytic vacuoles or lysosomes. The outer membrane of the autophagosome then fuses to the vacuolar membrane to deliver its cargo into the lumen. The damaged proteins and unwanted organelles are then digested by proteases, lipases and other hydrolases. Present in all eukaryotic cells at a basal level, autophagy is induced in response to stress and during ageing (Liu et al., 2009; Breeze et al., 2011). Although it was considered for a long time as a nonselective mechanism, the recent discovery of protein motifs and of protein interactions between the membrane-bound ATG8 proteins and some protein adaptors provides evidence that autophagic vesicles can be selective for their cargos (Noda et al., 2010).
In plants, as in animals and yeast, ATG genes appear to be essential for cell longevity, especially under nutrient-limiting conditions (Hanaoka et al., 2002). Autophagy mutants indeed display early senescence symptoms and hypersensitivity to carbon (C) and nitrogen (N) nutrient shortages (Liu & Bassham, 2012). Because plants are static organisms, their ability to recycle and remobilize nutrients, and especially N, is crucial for adaptation to a fluctuating environment. Processes allowing the recycling of organic N at the whole-plant level are therefore essential for plant survival and fitness (Masclaux-Daubresse & Chardon, 2011). In a recent report, we provided evidence that the autophagy machinery is an important factor for N remobilization efficiency and for grain-filling in Arabidopsis (Guiboileau et al., 2012). Autophagy mutants are senescing earlier than the wild type, especially when grown under low nitrate nutrition. Their capacity to remobilize nitrogen efficiently and their fitness are also greatly affected. Using salicylic acid-deficient autophagy mutants (atg5.sid2) that presented a green phenotype similar to that of the wild type and similar defects in autophagy machinery to those of atg mutants, we showed that the double mutants remained inefficient for N remobilization. The N remobilization capacity of atg mutants therefore seems to be independent of their senescence phenotype.
It is well known that the progressive degradation of chloroplast proteins occurring during leaf senescence and in response to N starvation is essential for N remobilization (Feller & Fischer, 1994). The ways in which proteins, especially chloroplast proteins, are degraded during leaf senescence are not fully understood (Martinez et al., 2008). There is evidence that the degradation of chloroplastic proteins such as Rubisco and chloroplastic glutamine synthetase 2 (GS2) can occur both inside and outside the chloroplast. When the chloroplast environment is oxidizing, the oxidation of a critical cysteine residue of Rubisco results in the fragmentation of its large subunit and in the binding of the oxidized peptides to the chloroplast envelope (García-Ferris & Moreno, 1993, 1994). Similarly, the chloroplastic glutamine synthetase has been shown to be rapidly degraded under conditions that cause photo-oxidative stress in leaves, and this has been linked to the oxidative carbonylation of histidine residues on the protein (Palatnik et al., 1999; Nakano et al., 2006). In addition to its localization in the chloroplasts, Rubisco was also detected in cytoplasmic vesicular compartments named ‘Rubisco-containing bodies’ (RCBs) or ‘Rubisco vesicular bodies’ (RVBs) depending on the authors (Chiba et al., 2003; Prins et al., 2008). It was shown that RCBs contain stroma proteins but not thylakoid proteins. Recently, several lines of evidence that RCB trafficking from the cytosol to the vacuole is autophagy dependent were provided (Ishida et al., 2008; Wada et al., 2009; Izumi et al., 2010). Tagging chloroplast stroma proteins with GFP, Wada et al. (2009) observed the appearance of many small vesicles in the vacuolar lumen which exhibited Brownian motion in the darkened leaves in which autophagic body degradation was blocked using the V-ATPase inhibitor concanamycin A. The accumulation of these vesicles, identified as RCBs, was impaired in autophagy mutants.
The over-expression of the cystatin protease inhibitor in tobacco (Nicotiana tabacum) leaves leads to the over-accumulation of RCBs and Rubisco in transgenic plants (Prins et al., 2008). This provides evidence that cysteine proteases can also be involved in the regulation of chloroplast protein degradation (Prins et al., 2008). Several cysteine proteases are induced during leaf ageing (Roberts et al., 2012), and it was shown that some protease inhibitors are down-regulated during leaf ageing and while N remobilization occurs. It is then likely that both proteases and protease inhibitors control the fine-tuning of protein degradation for N remobilization during leaf ageing (Etienne et al., 2007; Desclos et al., 2008). The cysteine proteases induced in response to leaf senescence are mainly described as vacuolar enzymes. Therefore, autophagy might play a role in the delivery of chloroplast components to the vacuole for degradation and then contribute to RCB proteolysis.
As we found that autophagy mutants have a low N-remobilization efficiency at the whole-plant level (Guiboileau et al., 2012), in this study we investigated the physiology of N management and remobilization in rosettes of autophagy mutants (the atg5 and atg9 mutants) and an RNAi line in which ATG18a expression is suppressed (RNAi18) (Supporting Information Fig. S1). From the results obtained, with respect to enzyme activities and metabolite contents in plants grown under nitrate-sufficient or nitrate-limiting conditions, we can provide a picture of the physiological symptoms presented by the autophagy mutants and compare them to the senescence symptoms observed in Arabidopsis leaves by Diaz et al. (2008). The protein profiles described in this report strongly suggest that protein degradation is incomplete and aberrant in autophagy mutants, and the identification of some proteins accumulating in the rosettes of atg mutants leads us to propose potential protein substrates for autophagy.
Materials and Methods
Plant material and growth conditions
Seeds of Arabidopsis thaliana (L.) Columbia wild type, the homozygous Arabidopsis mutants atg5 (SALK_020601) and atg9 (SALK_130796), and AtATG18a RNAi (RNAi18) (Hanaoka et al., 2002; Xiong et al., 2005; Yoshimoto et al., 2009; Inoue & Klionsky, 2010) were cultivated on sand under the same low-N (2 mM nitrate) and high-N (10 mM nitrate) nutrition rates used by Lemaitre et al. (2008). The experimental unit was a small pot (60 × 65 × 60 mm units) containing six plants positioned in a circle for the nitrate-limiting conditions (2 mM) or a single plant per pot for the high-nitrate conditions (10 mM). Plants were grown under short days (light : dark cycle 8 h : 16 h), with a photosynthetic photon flux density of 160 μmol m−2 s−1. The pots were watered three times per week, by immersion of the base of the pots. Harvests of whole rosettes started 30 d after sowing (DAS), and were performed every week. The fresh weight of individual rosettes was measured. At each harvesting time, three independent bulks containing six or 12 rosettes grown under high- or low-nitrate conditions, respectively, were collected for each genotype between 10:00 and 11:00 h, and stored at −80°C until use in further experiments. A total of five culture rounds were performed and the following analyses were carried out on at least two rounds of culture. The data are only presented for one round of culture, and it was verified that similar results were obtained for other rounds.
Determination of total nitrogen and carbon contents
After drying and weighing of each plant part, material was ground to obtain a homogenous fine powder. A subsample of 1000–2000 μg was carefully weighed in tin capsules to determine total N and C contents using a FLASH 2000 Organic Elemental Analyser (Thermo Fisher Scientific, Villebon, France).
Metabolite extraction and analysis
Contents of amino acids and were determined after extraction in a 2% (w/v) solution of 5-sulfosalicylic acid by the Rosen colorimetric method using glutamine as a reference (Rosen, 1957). Sugar contents were determined after ethanol extraction. Contents of sucrose and hexoses were determined in supernatants as described by Masclaux-Daubresse et al. (2002). Starch content was determined using pellets (Masclaux-Daubresse et al., 2002).
Glutamine synthetase activity measurement
Enzymes were extracted from frozen leaf material stored at −80°C. GS activity was measured according to Masclaux-Daubresse et al. (2002). The soluble protein content was determined in the crude leaf extracts used for GS activity measurement using a commercially available kit (Coomassie Protein assay reagent; BioRad, Hercules, CA, USA).
Protease activity assays
For analysis of endo- and exoproteolytic activities, protein extracts prepared as described in the previous section were desalted by centrifugation through Sephadex G-25 (coarse) columns equilibrated with 25 mM Tris/HCl, pH 7.5. All steps were performed at 0–4°C. Aminopeptidase activities were assayed using 2 mM L-leu-p-nitroanilide as substrate in 100 mM Tris HCl buffer, pH 7.5 with 1% (v/v) DMSO. The assays were performed kinetically by measuring absorbance at 405 nm, for 10 min at 25°C, using a microliter plate reader (Lab System iEMS Reader MF, Vienna, VA, USA). Carboxypeptidase assays were performed using 2 mM N-carbobenzoxy-L-phe-L-ala as substrate in 100 mM Na acetate, pH 5.0, containing 2% (v/v) DMSO. Substrate was omitted for blanks. Enzyme assays (in wells of microtitration plates) were incubated for 1 h at 37°C, and liberated α-amino groups were assayed with a TNBS reagent (150 ppm trinitrobenzene sulfonic acid in 50 mM Na borate buffer, pH 9.5) at 405 nm. The method was calibrated with 0–50 nmol glycine. Endopeptidases were quantified using 1% (w/v) azocasein as substrate, in 200 mM Na acetate buffer, pH 4.5, containing 0.2% (v/v) β-mercaptoethanol, 200 mM Na acetate buffer, pH 5.4, containing 0.2% (v/v) β-mercaptoethanol, or 200 mM Tris HCl, pH 7.5. After 3 h of incubation at 37°C, undigested azocasein and large fragments resulting from proteolytic digestion were precipitated with cold TCA (trichloroacetic acid; final concentration 5% w/v) before centrifugation (15 min at 3000 g). Absorbance of supernatants was determined at 450 nm after transfer to a new microtitre plate and addition of NaOH to a final concentration of 0.5 M.
Biomass was measured on six to 12 rosettes and on at least two culture rounds. Metabolite and enzyme activities were measured on three independent bulks of rosettes (each bulk contained six to 12 rosettes depending on N conditions). For all measurements, three technical repeats were performed. Means, standard errors and significances were determined using Excel software and Student's test was performed on the mean of the technical repeats of biological repeats (n = 3).
Gel electrophoresis and western blot analysis
For western blots, proteins were extracted at 4°C in Hepes (50 mM; pH 7.5), MgCl2 (2 mM), EDTA (0.5 mM), DTT (2 mM), leupeptin (2 μM) and Triton X-100 (0.1% v/v), and immediately denatured by boiling for 5 min after adding one volume of denaturing buffer (0.4 M Tris HCl, pH 6.8, 10% (w/v) SDS, 50% (v/v) glycerol, 25% (v/v) β-mercaptoethanol and 0.25% (w/v) bromophenol blue) to four volumes of protein extract. Proteins were then separated by SDS-PAGE; with an equal amount of protein in each lane. The percentage of polyacrylamide in the running gels was 10% for GS, glutamate dehydrogenase (GDH), Rubisco large subunit (LSU), isocitrate dehydrogenase (IDH), gamma-vacuolar processing enzyme (γ-VPE), chlorophyll binding protein (CAB), catalase (CAT) and thioredoxin-f, and 15% for Rubisco small subunit (SSU), ribosomal protein small subunit 6 (RPS6), ribosomal protein small subunit 14 (RPS14) and ribosomal protein large subunit 13 (RPL13). Denatured proteins were electrophoretically transferred to nitrocellulose membranes or stained directly with Coomassie blue for Rubisco detection. For GS, polyclonal antibodies were raised in rabbits against the synthetic peptide AYGEGNERRLTG (Eurogentec, Seraing, Belgium) and then used for detection of both GS1 and GS2 isoenzymes (Lemaitre et al., 2008). Antibodies against tobacco Rubisco LSU and SSU and tobacco Fd-dependent GOGAT (Fd-GOGAT, Glutamate OxoGlutarate Amino Transferase ferredoxin dependent) were kindly provided by Dr A. Suzuki (IJPB, INRA, Versailles, France), antibodies against Arabidopsis IDH by Dr M. Hodges (IBP, University Paris Sud, Orsay, France), antibodies against maize (Zea mays) phosphoenol pyruvate carboxylase (PEPc) by Dr J. Vidal (IBP, University Paris Sud, Orsay, France), antibodies against thioredoxin-f by Dr E. Issakidis-Bourguet (IBP, University Paris Sud, Orsay, France), antibodies against Arabidopsis CAT by Dr G. Noctor (IBP, University Paris Sud, Orsay, France), antibodies against γ-VPE by Dr N. Raikhel (University of California, Riverside, CA, USA), and antibodies against RPL13 by Dr J. Saez-Vasquez (LGDP, UPVD, Université de Perpignan, Perpignan, France). S14-1 (Ref. AS09477; Agrisera Antibodies, Vannas, Sweden) and S6 (Ref. 2317S; Cell Signaling Technology, Danvers, MA, USA) were used to detect the RPS14 and RPS6 families. All these antibodies had already been used against Arabidopsis proteins by the research groups from which they originated. Specificity was not verified in this study, but correct protein weight was verified.
Autophagy mutants display a lower C : N ratio than wild type
As reported previously, autophagy mutants displayed earlier leaf senescence symptoms under low nitrate than under high nitrate, and atg5 exhibited the strongest yellowing phenotype (Guiboileau et al., 2012). Yellowing symptoms were observed in atg5 from 40 DAS, and at 60 DAS all atg mutants and RNAi18 plants were senescing (Fig. S2). Under both high- and low-nitrate conditions, atg mutants and RNAi18 also displayed lower rosette biomass than wild type (Fig. 1). Reduced fresh weight was especially pronounced for atg5 (Figs 1, S3). Despite the strong leaf senescence phenotype observed in mutants at 75 DAS, there was no significant difference in water content between mutants or RNAi18 and wild type (Fig. S3). From within the time course of analyses of the mutant and wild-type rosettes for leaf yellowing, fresh weight and other parameters (e.g. C : N ratio and GS activities), we choose three time-points for further experiments. The first time-point was 30 DAS, when no visible yellowing symptoms could be detected in any mutant or wild-type plants (Fig. S2). At 60 DAS, all atg mutants displayed leaf senescence symptoms under low nitrate and at 75 DAS the first leaf-yellowing symptoms were visible on the wild type under the low-nitrate regime.
Under both low- and high-nitrate conditions, rosettes of the atg mutants and RNAi18 displayed a significantly lower C : N ratio than those of the wild type (Fig. 2a,b) as a result of both higher N concentrations (N%; Fig. 2c,d) and lower C concentrations (C%; Fig. 2e,f). In order to determine the nature of autophagy mutant C : N defects, we measured the concentrations of soluble and insoluble sugars, total amino acids, ammonium and proteins. Under low-nitrate conditions, relative to wild type, lower hexose concentrations were found at 60 and 75 DAS in atg mutants and RNAi18, and lower sucrose concentrations were found at 30 DAS in atg5 only and at 60 DAS in all atg mutants and RNAi18 (Fig. 3a–c). Starch concentration was only affected at 60 DAS in atg5 (Fig. 3d). Under high-nitrate conditions, lower concentrations of glucose and starch were observed in all atg mutants and RNAi18 at 60 and 75 DAS (Fig. 3e,h). A reduced starch content was also observed using lugol staining (Fig. S4). Sucrose and fructose concentrations were not significantly different from those of wild type (Fig. 3f,g). Despite the differences observed between the low- and high-nitrate regimes, and depending on the nature of the sugars considered, the results indicate that the lower accumulation of sugars in the atg mutants and RNAi18 relative to the wild type was consistent with the lower C concentration observed in mutants.
A contrario, ammonium, total amino acid and soluble protein concentrations were higher in atg mutants and RNAi18 than in the wild type (Fig. 4a–f), whatever the nitrate supply. Differences were clearly significant at 60 and 75 DAS. In addition, we observed that total RNA concentrations were significantly higher in atg mutants and RNAi18 than in the wild type (Fig. 4g), adding another potential N resource to be considered in the bulk of compounds responsible for the higher total N concentration observed in atg mutants (Fig. 2c,d).
Glutamine synthetase activity is maintained in autophagy mutants
GS activity is essential for inorganic N assimilation and amino acid biosynthesis. GS activity includes the activities of a unique chloroplastic GS2 isoform and several cytosolic GS1 isoforms. GS1 and GS2 gene regulation is quite complex and depends on leaf senescence and C availability (Cren & Hirel, 1999). In addition, while it is the case that the GS2 isoform is mainly involved in primary ammonium assimilation in young leaf tissues, GS1 isoforms can participate in both primary N assimilation (Lothier et al., 2011) and recycling of ammonium released from amino acid catabolism in old leaves (Masclaux-Daubresse et al., 2010). In order to determine whether differences in amino acid and ammonium concentrations in atg mutants and RNAi18 might be attributable to GSs, the total GS activity and the GS1 and GS2 protein contents were monitored. There were generally no differences in total GS activity among the wild type, atg mutants and RNAi18 plants grown under low nitrate, except for very slight increases at a few time-points (Fig. 5a). Western blots performed using polyclonal antibodies allowed us to identify the chloroplastic GS2 (40-kDa) and cytosolic GS1 (38-kDa) isoforms (Figs 5b, S5). At 30 and 37 DAS, GS1 and GS2 protein contents were not different among the four genotypes, and GS2 was more abundant than GS1, as is usually the case in young leaves (Fig. S5). At 44 DAS, GS1 became more abundant than GS2 proteins in atg5. From 60 DAS, GS2 became less abundant than GS1 in atg5, atg9 and RNAi18, although GS2 still remained more abundant in the wild type. Quantification of GS1 and GS2 protein contents showed that the higher percentage of GS1 than of GS2 observed in atg mutants was attributable both to a decrease in GS2 and to an increase in GS1 content. The amount of GS2 protein remaining in atg5 at 75 DAS was very low compared with GS1 (Fig. 5b). These results show that atg mutants and RNAi18 are able to maintain their GS activity as high as that of the wild type by balancing the disappearance of GS2 with an increase in GS1.
GS protein profiles and protein contents revealed differences among atg mutants, RNAi18 and wild type
Interestingly, the western blots we performed using GS antibodies revealed the presence of additional bands of lower mass in the samples from atg mutants and RNAi18 but not in wild-type samples. The decrease in GS2 protein content in atg5, atg9 and RNAi18, together with the presence of peptides (24- and 22-kDa) recognized by the GS antibodies and detected only in the atg5, atg9 and RNAi18 protein extracts, suggested that peptides were partial GS2 degradation products (Fig. 6). In order to determine if additional products could be observed for other proteins, several western blots were performed using the same protein extracts as before and various antibodies against cytosolic and chloroplastic proteins, such as PEPc, Fd-GOGAT and Rubisco LSU and SSU.
As found with GS antibodies, additional bands were observed using the LSU antibodies (Fig. 6). These bands (44, 42, 39, 35 and 31 kDa) were observed in atg5, atg9 and RNAi18 but not in the wild type, suggesting this time the presence of partial degradation products of the LSU in mutants. As no supplementary signal could be detected with the SSU and Fd-GOGAT antibodies, we concluded that the partial degradation was not a feature of all chloroplastic proteins but only of GS2 and Rubisco. In order to determine whether this disorder was age- and nitrate-dependent, other western blot analyses were performed using protein extracts from atg5 and wild-type plants grown under low or high nitrate, for 30 or 60 d (Fig. 7). For the LSU, additional bands were observed in old (60 DAS) rosettes of atg5 under both low- and high-nitrate conditions. For GS, additional bands were only observed in atg5 rosettes grown under low nitrate for 60 DAS. The presence of potential degradation products of LSU and GS2 was consistent with results previously obtained using wheat (Triticum aestivum) and pea (Pisum sativum), suggesting that reactive oxygen species can degrade specifically GS2 and LSU within the chloroplast (see Feller et al., 2008 for a review). Interestingly, additional bands were observed not only for the LSU and GS2, but also for the mitochondrial IDH.
The presence of additional bands was not the only feature of atg mutants. Western blots revealed that some native proteins also accumulated in atg5 compared with the wild type. IDH content was increased in atg5 grown under low-nitrate conditions at 30 DAS. Accumulation of another mitochondrial protein, GDH, was also observed in atg5 under low-nitrate conditions at 30 and 60 DAS. The peroxisome protein CAT also accumulated in atg5 rosettes under both low- and high-nitrate conditions as early as 30 DAS. Western blots using ribosomal protein (RP) antibodies revealed that RPL13 and RPS6 proteins were more abundant in atg5 than in the wild type, suggesting higher ribosome numbers in atg5. However, the absence of differences in RPS14 contents between atg5 and the wild type revealed that all RPs did not behave in the same way. As observed previously for Fd-GOGAT and PEPc, the chloroplastic thioredoxin-f, chloroplastic CAB and vacuolar -VPE protein contents were not different between atg5 and the wild type, regardless of the nitrate regime or time-point (Figs 6, 7).
Protease activities are higher in autophagy mutants and RNAi lines than in wild type
In order to determine whether the accumulation of some proteins and peptides in autophagy mutants could be associated with defects in protease activities, we measured amino-, carboxy- and endo-peptidase activities in wild-type, atg5, atg9 and RNAi18 plants grown under low- and high-nitrate conditions.
Aminopeptidase activity was higher in all the atg mutants and RNAi18 than in the wild type at low nitrate and in atg5 at high nitrate (Fig. 8a,b). Carboxypeptidase activity was also significantly higher in atg5 under low-nitrate conditions (Fig. 8c,d). However, no significant difference was observed among wild-type, atg mutant and RNAi18 plants grown under high nitrate.
Endopeptidase activity was assayed at acidic pH 4.5 (Fig. 8i,j) and pH 5.4 (Fig. 8g,h) and at slightly alkaline pH 7.5 (Fig. 8e,f). At pH 7.5 and pH 5.4, the endopeptidase activities were significantly higher in all the mutants than in the wild type when plants were grown under low-nitrate conditions (Fig. 8e–h). Under high nitrate, endopeptidase activity was only significantly higher for atg5 and RNAi18 at both pH 5.4 and pH 7.5. RNAi18 and atg9 endopeptidase activities did not follow the same trend as that of atg5, and were not different from that of wild type under high nitrate. At low nitrate, pH 5.4 and pH 7.5 endopeptidase activities in atg5 were very high.
The pH 4.5 endopeptidase activities measured in the rosettes of atg mutants and RNAi18 were not different from that of the wild type under low nitrate (Fig. 8i). However, pH 4.5 endopeptidase activity was significantly higher in atg mutants and RNAi18 compared with the wild type when plants were grown under high-nitrate conditions (Fig. 8j). Surprisingly, the endopeptidase activity measured at pH 4.5 was significantly higher in atg9 and RNAi18 than in atg5 for unknown reasons.
In plants, especially in Arabidopsis, most of the studies and reports dealing with autophagy have focused on the dissection of the molecular machinery involved in the regulation of the autophagy pathway, in the formation of autophagosome structures and more recently in the specificity of the autophagy process (for reviews, see Bassham et al., 2006; Li & Vierstra, 2012; Yoshimoto, 2012). These studies have provided a large set of atg mutants that, with very few exceptions, show similar phenotypes, such as early leaf senescence, hypersensitivity to N and C limitation, defects in plant immunity and hypersensitivity to a large panel of stresses such as drought, salt stress and pathogen attack, especially by necrotrophs. Because of the hypersensitivity of atg mutants to nitrate limitation, it was assumed that autophagy is essential for N recycling (Doelling et al., 2002; Hanaoka et al., 2002; Thompson et al., 2005), and we have recently documented the defect of N remobilization in autophagy mutants (Guiboileau et al., 2012). Although there are several lines of evidence that autophagy defects have strong impacts on the physiology of plants and on their response to the environment, no physiological data have been available until now. The combination of cytological approaches and mutant studies has provided evidence that the autophagy process could be important for the degradation of stroma proteins, especially Rubisco (Ishida et al., 2008; Wada et al., 2009). The aim of the work presented here was to gain insights into the physiological disorders affecting atg mutants.
Because the principal autophagy mutant phenotype, as observed in all studies performed by us and other groups, is early and severe leaf senescence, especially under nitrate limitation conditions, we decided to monitor some physiological and metabolic markers before and after the appearance of any leaf-yellowing phenotype, and three time-points were selected at which to perform the analysis.
At the time of the first harvest used in this study, 30 DAS, there was no visible yellowing phenotype in any of the mutants or in the wild type. As several autophagy genes were already significantly induced at this date in wild-type rosettes, by contrast with the SAG12 (senescence associated gene 12) senescence marker whose expression was very low (Fig. S6), we speculated that it may be possible to detect at 30 DAS precocious autophagy defects, unrelated to typical leaf senescence. Indeed, we observed that biomass and the C : N ratio were slightly lower in atg mutants and RNAi18 than in the wild type at 30 DAS under low-nitrate conditions. No significant difference in sugar and N compound concentrations was, however, detected at 30 DAS between mutants and the wild type. In contrast, immunoblotting allowed us to reveal differences in the abundance of certain specific proteins between the wild type and atg5 under low- and high-nitrate conditions. Taken together, these results provide evidence that physiological disorders are present in atg mutants even before the appearance of any visible senescence phenotype.
The differences between the wild type and atg mutants became more significant later, at 60 and 75 DAS. The C : N defect detected in atg mutants and RNAi18 was clearly linked to the accumulation of N compounds such as ammonium, amino acids and proteins, and to lower concentrations of sugars. Ammonium and amino acid over-accumulations in atg mutants and RNAi18 were not related to changes in GS activity, which was maintained at the same level in mutants and in the wild type throughout plant development. Surprisingly, we observed that both protein concentrations and protease activities are higher at 60 DAS in atg mutants relative to the wild type. This suggests that protein accumulation in mutants is not a result of protease deficiency. Immunoblots revealed that protein accumulation in atg mutants is selective. Relative accumulation of RPS6, RPL13, GDH and CAT was observed in atg5 from 30 DAS onwards; changes in GS1, GS2 and LSU contents and incomplete degradation products of GS2, LSU and IDH were clearly observed at 60 DAS in atg mutants and RNAi18.
All these results provide evidence that mutants accumulate physiological and metabolic disorders, which are on the one hand side effects of their early leaf senescence symptoms and on the other hand direct effects of autophagic activity defects. Indeed, comparison of the present data with those of our previous studies shows that, while the autophagy mutants display some modifications that are similar to the typical leaf senescence phenotypes as described in several plant species by Masclaux et al. (2000), Diaz et al. (2005, 2008) and Roberts et al. (2012), they also present modifications that are not related to senescence symptoms. The decreases in FW and sugar concentrations and the increases in GS1 protein content and protease activities observed in atg mutants and RNAi18, under low nitrate nutrition especially, are well-known markers of leaf senescence. However, while leaf senescence is characterized by decreases in total N, soluble proteins, total RNA and total amino acids, we observed exactly the opposite in atg mutants and RNAi18 irrespective of nitrate conditions. The autophagy defect is clearly characterized by the accumulation in leaves of various N compounds such as proteins, amino acids and ammonium. All other symptoms may be indirectly attributable to their early leaf senescence phenotype.
The higher concentration of total amino acids and soluble proteins in the rosettes of atg mutants relative to the wild type suggests that N resources are globally trapped in the leaves of mutants. This is in close accordance with the lower N remobilization efficiency we have reported for these mutants (Guiboileau et al., 2012). The over-accumulation of specific proteins and peptides in autophagy mutants suggests that mutants cannot properly degrade some of their proteins. The higher protease activities measured in mutants show that this is not attributable to defects in proteolysis. Regarding the pH responses of the endopeptidases and carboxypeptidases differentially active in mutants relative to the wild type, it seems that they are located in the vacuole and possibly also in the cytosol. Acidic endopeptidase and carboxypeptidase classes known to increase during leaf senescence are mainly located in the vacuole (Peoples & Dalling, 1988; Brouquisse et al., 2001; Roberts et al., 2012). It is, then, likely that the increase in peptidase activities in atg mutants is senescence-related.
The presence on western blot membranes of additional bands recognized by the LSU and GS2 proteins suggests that the LSU and GS2 proteins have undergone the first steps of their degradation process but that proteolysis of their peptides has not been completed. The release of peptides from the GS2 and Rubisco LSU proteins by reactive oxygen species (ROS) in chloroplasts is well documented in several plant species (see Feller et al., 2008 for a review). We know that autophagy mutants suffer from oxidative stress (Yoshimoto et al., 2009). The peptides observed on our western blots could have been released from GS2 and LSU by the effect of ROS within the chloroplast. Considering that the increased peptidase activities are mainly located in the vacuole and not in the same compartment as LSU, GS2 or IDH products, we can speculate that they cannot complete proteolysis unless the autophagy pathway causes the targeting of these substrates to the vacuole.
In addition to GS2 and LSU peptides, atg mutants also specifically accumulated several native proteins, such as ribosomal RPS6 and RPL13, IDH, GDH and CAT. Together with the higher total RNA concentrations (mostly represented by rRNA), the over-accumulation of RPS6 and RPS13 observed in atg mutants and RNAi18 suggests defects of ribophagic activity (MacIntosh & Bassham, 2011). However, the fact that the RPS14 content was not different between the wild type and atg mutants suggests that the role of autophagy in the degradation of the whole ribosome complex or in the degradation of each RP protein individually is more complex. Also, the fact that two mitochondrial proteins, IDH and GDH, accumulate in atg5 under low-nitrate conditions does not mean that mitophagy is globally affected in atg mutants. The mechanisms by which IDH and GDH accumulate in atg mutants can be different. The detection on IDH western blot membranes of peptides of lower mass than the native IDH protein suggests that, like that of GS2 and LSU, IDH protein degradation needs autophagy to be complete. The absence of additive bands on GDH western blot membranes and a recent transcriptome analysis performed in our laboratory (C. Masclaux-Daubresse, unpublished) suggest that, unlike IDH, GDH protein accumulates in atg mutants as a result of higher GDH gene expression.
Thus, the present study provides several lines of evidence of protein turnover defects in atg mutants. However, imperfect protein degradation cannot explain why atg mutants over-accumulate amino acids and ammonium. Ammonium and amino acid accumulation does not seem to be a result of defects in N assimilation, as GS activity was not different between the wild type and mutants and nitrate did not accumulate in mutants or in wild type (not shown). One explanation for amino acid and ammonium accumulation is the early and severe senescence symptoms observed on the leaves of atg mutants. Export of amino acids and ammonium requires efficient phloem loading, and vein tissue longevity is needed for that (Tegeder & Rentsch, 2010). One can suppose that phloem loading is affected when leaf senescence occurs too rapidly and too early. Consistent with this idea is the finding that ammonium and amino acid accumulation occurred at 60 DAS, when senescence symptoms were observed on leaves of mutants.
The accumulation of soluble proteins, total amino acids and ammonium in the rosettes of autophagy mutants can explain why they are less efficient at remobilizing N to their seeds (Guiboileau et al., 2012). Interestingly, we provide here some evidence that the nature of the proteins and peptides that accumulate in rosettes of atg mutants is specific and that autophagy is involved in the degradation of chloroplast protein cleavage fragments (Ono et al., 2013). Although it is probable that the increase of protease activities in leaves of atg mutants is related to their early senescence phenotype (Masclaux et al., 2000), we can also hypothesize that their respective genes are overexpressed in atg mutants as a response to the over-accumulation of unwanted proteins and peptides. The future characterization of the proteome and transcriptome of atg mutants and the study of the specific protease activities that increase in atg mutants are promising lines of enquiry that will enable identification of components of the machinery involved in the recycling of inorganic N for remobilization.
The authors thank Prof. Andreas Fischer (University of Montana, Bozeman, MT, USA) for providing procedures with which to measure protease activities, and Sascha Ludwig for carrying out preliminary experiments while in receipt of an FP6 Marie Curie fellowship (no. A02425; VERT). The authors also thank Prof. David Logan (University of Angers, Angers, France) and Dr Michèle Reisdorf-Cren (UVSQ, Versailles, France) for help with revision of the manuscript. This work was mainly supported by INRA, and by Ministère de l'Education et de la Recherche (funding received by A.G.) and Université de Versailles Saint-Quentin en Yvelines (funding received by J.L.).