•Processes allowing the recycling of organic nitrogen and export to young leaves and seeds are important determinants of plant yield, especially when plants are nitrate-limited. Because autophagy is induced during leaf ageing and in response to nitrogen starvation, its role in nitrogen remobilization was suspected. It was recently shown that autophagy participates in the trafficking of Rubisco-containing bodies to the vacuole.
•To investigate the role of autophagy in nitrogen remobilization, several autophagy-defective (atg) Arabidopsis mutants were grown under low and high nitrate supplies and labeled with at the vegetative stage in order to determine 15N partitioning in seeds at harvest. Because atg mutants displayed earlier and more rapid leaf senescence than wild type, we investigated whether their defects in nitrogen remobilization were related to premature leaf cell death by studying the stay-green atg5.sid2 and atg5.NahG mutants.
•Results showed that nitrogen remobilization efficiency was significantly lower in all the atg mutants irrespective of biomass defects, harvest index reduction, leaf senescence phenotypes and nitrogen conditions.
•We conclude that autophagy core machinery is needed for nitrogen remobilization and seed filling.
Autophagy is a ubiquitous proteolytic process in eukaryotic cells that permits protein breakdown and amino acid recycling via nonselective lysosomal/vacuolar proteolysis (Nakatogawa et al., 2009; Reumann et al., 2010). The autophagy pathway is conserved across strata, and genetic screens in yeast have identified several AuTophaGy (ATG) genes that are required for the formation of the double membrane vesicles named autophagosomes that deliver products to be recycled into lytic vacuoles. Orthologs of ATG genes are present in plants, and their function has been assessed by using complementation of yeast atg mutants and by studying RNA interference (RNAi) and knock-out mutants (Doelling et al., 2002; Hanaoka et al., 2002; Yoshimoto et al., 2004; Xiong et al., 2005; Contento et al., 2005; Thompson et al., 2005). Using GFP-ATG8 fusions as marker of autophagosomes in Arabidopsis, it was also possible to monitor defects of mutants and RNAi lines in autophagic flux. In all the atg mutants isolated to date, early age-dependent leaf senescence symptoms and hypersensitivity to nitrogen and carbon starvation were observed (Hanaoka et al., 2002; Xiong et al., 2005; Thompson & Vierstra, 2005). Xiong et al. (2007) showed that the autophagy pathway plays a role in cleaning cells of oxidized proteins accumulated as a result of ageing or stress. More recently, Yoshimoto et al. (2010) proposed that the overaccumulation of salicylic acid (SA) or high concentrations of reactive oxygen species is involved in the early cell death and senescence phenotypes observed in old leaves of atg mutants. By overexpressing the bacterial salicylate hydroxylase in atg NahG mutants and by crossing atg mutants with the sid2 mutant, which is an SA biosynthesis-defective mutant, early leaf senescence symptoms were indeed reversed (Yoshimoto et al., 2009).
Plants are static organisms. Their ability to recycle and remobilize nutrients is crucial for adaptation to a fluctuating environment. Processes allowing the recycling of organic nitrogen and export from old senescing organs to sinks such as seeds are important determinants of plant productivity and yield, and organic nitrogen recycling becomes essential more than ever when plants are starved of nitrogen (Masclaux-Daubresse et al., 2010; for review). It has been known for a long time that the main source for nitrogen recycling and remobilization during leaf ageing is the chloroplast (Masclaux et al., 2000). The way in which chloroplast proteins are degraded during leaf senescence (induced or not by nitrogen starvation) is not fully understood. Stroma protein degradation could occur both inside and outside the chloroplast (Martinez et al., 2008; Reumann et al., 2010 for reviews). The oxidative environment existing in the chloroplast during senescence or in response to stress is suspected to catalyze Rubisco fragmentation (Ishida et al., 1997). The oxidized Rubisco peptides would then bind to the chloroplast envelope to be delivered into cytoplasmic vesicle-type compartments named ‘Rubisco-containing bodies’ (RCB) (García-Ferris & Moreno, 1994; Chiba et al., 2003; Prins et al., 2008). Recently, evidence that autophagy machinery participates in the traffic of RCBs to the central vacuole has been provided (Ishida et al., 2008; Wada et al., 2009; Izumi et al., 2010). The atg5 mutation, which compromises autophagy flux, indeed disrupted the accumulation of RCBs in vacuoles. Autophagy may then participate in chloroplast recycling.
The fact that atg mutants present hypersensitivity to nitrogen, reduced seed production, and reduced RCB accumulation in senescing leaves is accepted as evidence that autophagy participates in nitrogen recycling and mobilization in Arabidopsis. However, none of the studies reported to date has investigated nitrogen recycling and remobilization in atg mutants.
In order to investigate the role of autophagy in nitrogen remobilization at the whole-plant level, we first grew three atg mutant and RNAi lines (atg5-1, atg9-2 and atg18a RNAi lines) and explored their nitrogen remobilization efficiency (NRE) under high and low nitrate supplies, using 15N tracing as done by Masclaux-Daubresse & Chardon (2011). The ATG5, ATG9 and ATG18 genes disrupted or knocked down in these mutants are all involved in the first steps of autophagosome expansion and enclosure (Thompson & Vierstra, 2005). It is well documented that these mutants are defective in autophagic flux (Thompson et al., 2005; Xiong et al., 2005). Because atg mutants display early and fast leaf senescence phenotypes compared with wild type, and because leaf nitrogen remobilization needs time to proceed fully, we also monitored nitrogen remobilization in the atg5.sid2 and atg5.NahG lines previously characterized by Yoshimoto et al. (2009). Introgression of the sid2 mutation and overexpression of NahG in atg mutants were previously found to suppress the early leaf senescence phenotypes of atg mutants (Yoshimoto et al., 2009).
Materials and Methods
Plant material and growth conditions
Seeds of Arabidopsis thaliana (L.) Heynh. Columbia, atg5-1 (SALK_020601), atg5-2 (SAIL_129B07), atg9-2 (SALK_130796), atg5-1.sid2 and atg5-2.NahG have been characterized in previous reports (Hanaoka et al., 2002; Inoue et al., 2006; Yoshimoto et al., 2009). In their previous study, Yoshimoto et al. (2009) preferred atg5-2 to atg5-1 for overexpression of the NahG gene because exogenously overexpressed genes are often silenced in the atg5-1 background. Seeds of the AtATG18a RNAi transgenic plants (RNAi18) were kindly provided by Dr Diane C. Bassham (Iowa State University, Ames, IA, USA) and produced by Xiong et al. (2005).
Plants were cultivated on sand under low nitrogen nutrition (LowN; 2 mM nitrate) or under high nitrogen nutrition (HighN; 10 mM nitrate) as described in Chardon et al. (2010). Plants were grown in short days (8 h light) with 21°C day : 17°C night temperatures until 56 days after sowing (DAS) and then transferred to long days (16 h light), maintaining similar day : night temperatures. In the first set of experiments, seeds of atg5-1, atg9-2, RNAi18 and Col genotypes were sown on sand according to Masclaux-Daubresse & Chardon (2011). In the second set of experiments, seeds of atg5-1, atg5-2, atg5-1.sid2, atg5-2.NahG, sid2, NahG and Col genotypes were sown. Two independent biological repeats were carried out for each set of experiments, in two consecutive culture cycles in the same growth chamber. Each biological repeat included four plant repetitions for the first set of experiments, and six plant repetitions for the second set.
15N-labeling and tracing
Pulse 15N-labeling was performed according to Masclaux-Daubresse & Chardon (2011) and Chardon et al. (2010) on plants at the vegetative stage (40 DAS), a long time before flowering and harvest (which occurred c. 2.5 months later). For labeling, the unlabeled watering solutions were replaced by solutions that had the same nutrient composition except that plants were supplied with (10 atom% excess for remobilization and 2.5 atom% excess for nitrogen uptake measurements, respectively). For 15N uptake determination, plants were watered with solution for 24 h precisely, in order to monitor nitrogen uptake in rosette leaves and roots and to confirm that plants were homogenously labeled after a 24-h 15N-labeling period. For nitrogen remobilization, plant labeling was performed with the 10 atom% excess solution. After labeling, the pots containing sand and roots of the plants dedicated to nitrogen remobilization measurements were rinsed carefully several times with deionized water to discard 15N and unlabeled nutrient solutions were used for the rest of the culture cycle until seed harvest. At harvest, plants were separated into dry remains (DR; rosette + stem + cauline leaves + siliques) and total seeds (SEEDS). The dry weights (DWs) of the DR and SEEDS were determined.
Determination of total nitrogen content and 15N abundance
Unlabeled samples were harvested in order to determine the 15N natural abundance. Determination of 15N abundance in all the samples collected (labeled and unlabeled) was performed on dry matter as described in Diaz et al. (2008). The 15N abundance was calculated as atom per cent and defined as A% = (100 × (15N)/(15N + 14N)) and nitrogen concentration (N%) as mg nitrogen per 100 mg DW. The 15N enrichment (E %) was calculated as (Asample%–Acontrol%). The absolute quantity of 15N contained in a sample was defined as Q = (DW × E % × N%).
Indicator used to monitor nitrogen use efficiency (NUE) and nitrogen remobilization efficiency (NRE)
The harvest index (HI), a key indicator for yield, was calculated as the (DWSEEDS)/(DWDR + DWSEEDS) ratio. DW and nitrogen concentrations (N%) were combined to determine the nitrogen harvest index (NHI) as (N%SEEDS DWSEEDS)/(N%DR DWDR + N%SEEDS DWSEEDS), which is a key indicator of grain filling with nitrogen. As plant morphology and HI varied depending on genotype, the ratio NHI : HI was calculated as the percentage of Col control in order to compare NUE performances between genotypes. In a similar manner, we combined DW, N% and 15N enrichments (E %) to determine the partition of 15N in seeds (15N harvest index; 15NHI), which is the proportion of 15N absorbed at the vegetative stage and remobilized to the seeds at the reproductive stage. 15NHI, calculated as (E %SEEDS × N%SEEDS × DWSEEDS)/((E %DR × N%DR × DWDR)+(E %SEEDS × N%SEEDS × DWSEEDS)), is an indicator for the NRE to the seeds. Again, because of plant morphology and HI genetic variations, the ratio 15NHI : HI was estimated as the percentage of Col control to compare the NRE performances between genotypes independently of biomass effects. C%SEEDS (carbon concentration in mg per 100 mg DW) and N%SEEDS are key indicators for seed filling. More details of the significance of all the indicators described can be found in Masclaux-Daubresse & Chardon (2011).
Experiments were carried out in two independent biological repeats (R1 and R2) performed as two consecutive culture cycles in the same growth chamber. An ANOVA Newman–Keuls (SNK) comparison was performed using xlstat (Microsoft, http://www.xlstat.com). ANOVA analyses were carried out using raw data for all the measured and calculated traits in order to analyze the effects of genotypes, nutrition and biological repeats as well as their interactions. ANOVA analyses are presented in Supporting Information Fig. S1 and show that the strongest effects were attributable to genotypes and nutrition, while repeat effects were smaller for most of the traits. Data were, however, normalized according to the mean of repeats R1 and R2 before being used for further analysis. The number of plant repeats was then eight and 12 for set one and set two of the experiments, respectively.
Autophagy is essential for vegetative biomass and seed production regardless of plant nutrition but does not modify the harvest index
In order to investigate the effect of the autophagy core machinery on plant yield and grain filling in response to nitrogen nutrition, two autophagy mutants (atg5-1 and atg9-2) and one RNAi line (RNAi18 for ATG18a RNAi) previously characterized in three different laboratories were initially studied (Hanaoka et al., 2002; Thompson et al., 2005; Xiong et al., 2005). As described previously, atg5-1, atg9-2 and RNAi18 senesced early compared with wild type and were hypersensitive to nutrient limitation (Fig. 1). The DWs of their DR (Fig. 2a,b) and SEEDS (Fig. 2c,d) were indeed lower than those of the wild type, and DW differences between mutants and wild type were higher under low (LowN) than under high (HighN) nitrogen nutrition. However, there was no significant difference in HI (HI = DWSEED/(DWSEEDS + DWDR)) between atg9-2, RNAi18, atg5-1 and wild type, except at LowN where atg5-1 displayed leaf senescence, biomass and yield phenotypes (Fig. 2e,f).
Autophagy modifies nitrogen allocation to the seeds and dry remains
The nitrogen concentration, measured in the DR and SEEDS of wild type and mutants, showed that, while autophagy defects weakly modified N%SEEDS, it dramatically increased N%DR at LowN (Table S1). The higher N%DR observed in atg5 in this experiment was confirmed in further experiments (see Table 1; discussed in subsequent paragraphs). Using N% and DW data, the NHI, which represents the partitioning of nitrogen in seeds ((N content in SEEDS)/(N content in whole plant)) and is a familiar agronomic indicator of nitrogen seed filling, was computed. The atg5-1 mutant was the only line to show a significant difference compared with the wild type for NHI, and only at LowN (Fig. 3a,b). In order to determine whether the NHI defect of atg5-1 is attributable to lower seed production, we compared the relative modifications in HI and NHI, considering NHI : HI as a good indicator of the variation of NUE in plants (Masclaux-Daubresse & Chardon, 2011). The significantly lower NHI : HI ratio observed in atg5-1, atg9-2 and RNAi18 compared with wild type at both LowN and HighN (Fig. 3e,f) indicated that autophagy plays a role in grain NUE independently of effects on seed productivity.
Table 1. The nitrogen and carbon concentrations of dry remains (DR) and total seeds (SEEDS) are different in autophagy single and double mutants of Arabidopsis thaliana compared with controls (shown in bold) in both nitrate-rich (HighN) and nitrate-limiting (LowN) conditions
N and C concentrations (N% and C%, respectively) are presented as mg per 100 mg DW. Means and SDs were computed from normalized values from two biological repeats with six plants each (n =12). The different letters indicate values significantly different at P <0.05 as determined using the xlstat ANOVA Newman–Keuls (SNK) comparison. Bold values are not significantly different from Col values.
In order to examine whether autophagy plays a role in plant nitrogen remobilization to the seeds, we used an improved method consisting of monitoring 15N flux by performing pulse chase experiments (Masclaux-Daubresse & Chardon, 2011). The labeled 15N isotope provided to plants as a pulse of in the early steps of rosette development was found afterwards in the seeds and in the DR at harvest. We estimated that the soil was rinsed enough after the labeling pulse to avoid further uptake of during the chase period. The partitioning of 15N in seeds (as a percentage of total 15N content in the whole plant) is representative of the ability of plants to remobilize the 15N assimilated in their rosette leaves at the vegetative stage. The abundances of 15N in seeds and DR were determined using isotopic ratio mass spectrometry and allowed us, by combining these values with the DW and N% data, to calculate the partitioning of 15N in seeds (15NHI). In contrast with HI and NHI, 15NHI was significantly and substantially lower in all the atg mutants and the RNAi line compared with the wild type at both LowN and HighN (Fig. 3c,d). Computing the 15NHI : HI ratio confirmed that NRE is impaired in atg mutants and showed that the effect of autophagy activity on nitrogen is independent of its effect on plant biomass and productivity (Fig. 3g,h). Comparison of 15NHI : HI (Fig. 3g,h) and NHI : HI (Fig. 3e,f) shows that NRE defects in atg mutants were more severe than NUE defects, especially at LowN.
Nitrogen remobilization defect in autophagy mutants is not a result of premature leaf cell death
Despite nitrogen remobilization being highly active during leaf senescence, this process needs time to occur and has to be completed before the death of the leaf structures. Because the autophagy lines studied were senescing earlier and faster than in the wild type, we suspected that their nitrogen remobilization defects might have been a result of premature cell death in their leaf tissues. In order to determine the role of autophagy machinery in nitrogen resource mobilization independently of leaf senescence side effects, we then used the atg5-1.sid2 and atg5-2.NahG double mutants previously described by Yoshimoto et al. (2009). Lowering SA production by overexpressing the NahG gene in atg5-2 and by introgressing the sid2 mutation into atg5-1 indeed suppressed the senescence phenotypes actually observed in the atg5-1 and atg5-2 mutants. Consistent with the findings of Yoshimoto et al. (2009), we showed that under our conditions the atg5-2.NahG and atg5-1.sid2 double mutants senesced later than atg5 at LowN and HighN (Fig. S2). In addition, the comparisons of atg5-1, atg5-2, Col, sid2 and NahG with atg5-1.sid2 and atg5-2.NahG biomasses showed that, at both LowN and HighN, atg5-2.NahG and atg5-1.sid2 mutants recovered partially but not fully the DR and seed biomasses of the wild type (Fig. S3). The HI of atg5-2.NahG and atg5-1.sid2 was therefore significantly higher than that of atg5-2 and atg5-1, albeit significantly lower than that of Col, sid2 and NahG control lines (Fig. 4a,b).
As described previously, 15N-labeling and tracing were performed in order to monitor the NHI and nitrogen remobilization to the seeds. This time, 15N uptake was also monitored after labeling, with 15N concentration measured in plants at the end of the 24-h labeling period. Results showed that there was no difference in uptake between genotypes at the vegetative stage (Fig. S4).
From data obtained at seed harvest, NHI : HI and 15NHI : HI were computed to compare grain NUE and NRE, respectively, between the atg5-1.sid2 and atg5-2.NahG double mutants, the atg5-1 and atg5-2 single mutants and the Col, sid2 and NahG controls. At LowN, both NHI : HI and 15NHI : HI were significantly lower in the atg5-1, atg5-2, atg5-1.sid2 and atg5-2.NahG single and double mutants than in Col, sid2 and NahG controls (Fig. 4c,e). At HighN, a similar trend was observed for 15NHI : HI which was significantly reduced in both the atg5-2 and atg5-2.NahG mutants compared with the Col control and in atg5-1 and atg5-1.sid2 compared with the Col and sid2 controls (Fig. 4f). The large standard deviation obtained for the 15NHI : HI data for NahG did not allow us to observe a significant difference compared with atg5-2. Despite the large 15NHI : HI decrease observed in all the atg5 single and double mutants compared with the controls, the effect of the atg5 mutation on NUE (i.e. NHI : HI) was weak at HighN (Fig. 4d). The nitrate concentrations provided to the plant throughout its lifespan were certainly sufficient to maintain atg5 mutant NHI, despite the defect of these mutants in nitrogen remobilization.
The absence of nitrogen remobilization recovery in the atg5-1.sid2 and atg5-2.NahG double mutants compared with the atg5-1 and atg5-2 single mutants at LowN and HighN then indicated that autophagy activity by itself and not premature leaf senescence was the main factor affecting nitrogen remobilization to the seeds. Under nitrogen-limiting conditions, such a decrease in nitrogen remobilization significantly affected grain NUE.
Autophagy modifies nitrogen and carbon concentrations in plants
The determination of N%DR, C%DR, N%SEEDS and C%SEEDS revealed that, in all the single and double atg5 mutants, nitrogen concentration in DR was higher than in controls (Table 1). Nitrogen concentrations in seeds (N%SEEDS) were also higher in the atg5-1, atg5-1.sid2, atg5-2 and atg5-2.NahG mutants than in Col, sid2 and NahG. This higher N%SEEDS in atg5 mutants was accompanied by a lower C%SEEDS. While measurement of 15N uptake over 24 h at the vegetative stage did not reveal any difference between atg5 mutants and control lines at both LowN and HighN, the computing of the global plant growth rate and the global plant nitrogen accumulation rate between labeling time and seed harvest time showed that both rates were lower in all the atg5 mutants compared with the control lines (Table 2). However, comparing growth rate and nitrogen accumulation rate discrepancies between mutants and controls, we observed that the nitrogen acquisition rate defects were less marked than the growth rate defects. This certainly explains why, at the end of the plant cycle, nitrogen concentrations were higher in mutants than in the wild type. The lower C%SEEDS observed in all the atg5 mutants suggests that autophagy defects have in some way limited the duration of carbon fixation dedicated to plant growth and seed filling. Therefore, the higher nitrogen concentration observed in atg5 mutants might also be related to lower carbon assimilation and lower nitrogen dilution during plant ageing (Greenwood et al., 1990).
Table 2. Comparison of nitrogen accumulation rate and growth rate between autophagy mutants and control lines of Arabidopsis thaliana grown under low (LowN) and high (HighN) nitrate nutrition
(N(Th) − N(Te)):(DW(Th) − DW(Te))
N (mg per plant)
DW (mg per plant)
As % of Col
The nitrogen content (N as mg per plant) and dry weight (DW as mg per plant) of plants collected at labeling time (Tl, 42 d after sowing (DAS)) and at harvest (seed maturity, Th, 100 DAS) were computed in order to estimate nitrogen accumulation (N(Th) − N(Tl)) and growth (DW(Th) − DW(Tl)) rate ratios. The rate ratio is expressed as an absolute value and as a percentage of the value for Columbia (Col). Control lines are shown in bold.
In the present report, we provide evidence for a role of the autophagy machinery in nitrogen remobilization at the whole-plant level for seed production. It is not known how to enhance autophagy activity genetically through manipulating gene expression of repressor/activator for example. We then used autophagy mutants and nitrogen limiting and ample conditions known to modulate autophagy level to examine effects on NUE related traits. Two sets of experiments were performed. In the first set, mutants and RNAi lines in three different ATG genes were used. All of these ATG genes have already been described as part of the core machinery functioning in the formation of pre-autophagosomes and in autophagosome membrane expansion and enclosure (Thompson & Vierstra, 2005 for review). The severity of the leaf yellowing phenotypes, and the defects in biomass and HI observed were greater in the atg5-1 mutant than in atg9-2 and RNAi18 lines. These differences were consistent with the severity of autophagic activity defects described for these lines in previous studies. Autophagic activity monitored using GFP-ATG8 fusion indeed showed that atg5 mutants do not produce any autophagosomes, whatever the environmental conditions studied. By contrast, few remaining autophagic bodies can be detected in atg9-2 and RNAi18, leading authors to consider these lines ‘leaky’ (Yoshimoto et al., 2004; A. Guiboileau pers. comm. for RNAi18). Despite such differences, we show here that, compared with wild type Col, all these autophagy defective mutants (atg5-1, atg5-2, atg9-1 and RNAi18) present lower NRE under both low- and high-nitrate conditions. It is usually accepted, for instance in cereals, that senescence is associated with high seed harvest and high nitrogen remobilization indices (Kichey et al., 2007). Our previous studies showed that, in Arabidopsis, correlations between NRE for grain production and plant senescence are not as obvious as in cereals (Diaz et al., 2008; Masclaux-Daubresse & Chardon, 2011). The diminution of nitrogen remobilization observed here in the early-senescing autophagy-defective lines again illustrates this feature.
Interestingly, our results showed that the lower NRE of the autophagy mutant was not linked to the reduction in seed dry matter accumulation. Mutants’ NRE defects were actually independent of defects in seed production (i.e. independent of HI). Therefore, it appeared that defects observed for NRE were not attributable to a reduction in plant fertility or plant yield. The computing of the NHI : HI and 15NHI : HI ratios, which facilitates the comparison of grain NUE and NRE in the different genotypes, showed that the NUE of atg mutants was less affected than the NRE. It was concluded that atg mutants compensated the lack of nitrogen remobilization from rosette leaves using nitrogen uptake at the reproductive stage and assimilating exogenous nitrogen sources during seed development, as reported by Masclaux-Daubresse et al. (2010).
Nitrogen remobilization is usually associated with leaf senescence, because many proteases and enzymes involved in this process are induced during ageing (Masclaux et al., 2000; Diaz et al., 2008; Desclos et al., 2009). It was nevertheless observed that remobilization needs time to proceed and is compromised when cell death occurs too rapidly to permit complete protein recycling (Guiboileau et al., 2010; Masclaux-Daubresse et al., 2010). Because lower nitrogen remobilization in autophagy mutants might have been a side effect of the earlier and too-rapid leaf senescence, the atg5-1.sid2 and atg5-2.NahG lines modified for SA signaling were also studied (Yoshimoto et al., 2009). As previously described, we observed that the senescence phenotypes of the atg5-1.sid2 and atg5-2.NahG lines were significantly delayed and attenuated compared with those of the atg5-1 and atg5-2 single mutants. The labeling and tracing experiments showed that, despite having recovered their green phenotype and leaf longevity, both the atg5-1.sid2 and atg5-2.NahG double mutants showed defects in NRE and grain NUE. Their nitrogen management phenotype was not significantly different from that of atg5-1 and atg5-2, albeit a small compensation effect was noted. This led us to the conclusion that the autophagy machinery defect was in itself the main factor affecting nitrogen remobilization in atg mutants.
Despite their defects in nitrogen remobilization, the nitrogen concentrations of the seeds and DR of the atg5 single and double mutants were slightly – by reference to the large natural variation observed by Masclaux-Daubresse & Chardon (2011)– but significantly higher than those of controls. The higher nitrogen concentrations measured in mutant seeds were correlated to lower carbon concentrations. While the 15N-uptake experiment performed at the vegetative stage did not show any difference for nitrate uptake between mutants and controls, the nitrogen budget approach, which compared nitrogen contents at labeling and harvesting times, showed that the rate of nitrogen accumulation in atg plants was less affected than their growth rate. Nitrogen dilution occurring in plant shoots with ageing was less pronounced in atg mutants than in controls and this explains why atg mutants presented higher nitrogen concentrations at seed harvest (Greenwood et al., 1990). The lower nitrogen dilution in atg5 mutants, together with their defects in seed filling with carbon and their reduced rosette size, suggest defects in photosynthetic carbon assimilation. It is indeed possible that the autophagy machinery, by controlling the maintenance of the chloroplast photosynthetic machinery, has impacts on photosynthetic activity.
In Fig. 5, a model is proposed for the role of autophagy in nitrogen recycling and remobilization in leaves. Because autophagy is involved in long-life protein and organelle recycling (Maiuri et al., 2007; Ishida et al., 2008), as well as in cell waste degradation and SA signaling regulation (Yoshimoto et al., 2009), autophagy machinery facilitates nutrient recycling and exports as well as cell longevity (Fig. 5a). The defects of autophagy mutants in nitrogen remobilization are certainly attributable to their inability to digest and recycle proteins and other nitrogen resources in vegetative tissues. In addition, cell waste accumulating with ageing in autophagy mutant leaves certainly promotes premature cell death and indirectly affects remobilization of some amino acids produced through alternative proteolytic pathways (Fig. 5b). This effect of premature leaf senescence is confirmed by the partially recovered nitrogen remobilization flux observed in atg5-1.sid2 and atg5-2.NahG (Fig. 5c).
The main finding of this study is that autophagy plays an important role in nitrogen management at the whole-plant level through the control of nitrogen remobilization. This role explains the importance of autophagy in adaptation to nutrient restriction and in plant longevity. The significant and strong effect of autophagy mutation on NRE observed here provides the first evidence of the importance of the autophagy machinery in nitrogen management and more globally in NUE at the whole-plant level. This is also the first time that mutants affected in nitrogen remobilization have been described in Arabidopsis. However, it is also likely that autophagy does not affect only nitrogen management, and the effects observed on plant biomass, HI and seed composition suggest that carbon fixation is also impaired. The photosynthetic machinery, located in the chloroplast, actually recruits a large proportion of the leaf protein resources used for nitrogen remobilization. The possible role of autophagy in chloroplast and photosynthesis maintenance remains to be elucidated.
The authors thank Dr Diane Bassham (Iowa State University, IA, USA) who kindly provided the AtATG18a RNAi transgenic plants (RNAi18). Collaboration between IJPB and RIKEN was facilitated by the SAKURA program no 21124QA for bilateral collaborations supported by the French Ministère des Affaires Etrangères et Européennes and the Japan Society for the Promotion of Science.