A tissue-specific rescue strategy reveals the local roles of autophagy in leaves and seeds for resource allocation Authors: Anne Marmagne, Fabien Chardon, Céline Masclaux-Daubresse Published: 2025 ABSTRACT AbstractAutophagy is a vesicular mechanism that plays a fundamental role in nitrogen remobilization from senescing leaves to seeds. The Arabidopsis (Arabidopsis thaliana) autophagy (atg) mutants exhibit early senescence, reduced biomass, and low seed yield. The atg seeds also exhibit major changes in N and C concentrations. During plant development, autophagy genes are expressed in the source leaves and in the sink seeds during maturation. We thus addressed the question of whether the seed composition defects in atg mutants are caused by defective N remobilization from source leaves or whether they are due to the absence of autophagy in seeds during maturation. To answer this question, we restored autophagy activity in the atg5 mutant by expressing the wild-type (WT) ATG5 allele specifically in source leaves using the senescence-associated gene 12 (SAG12) promoter or specifically in seeds using the Glycinin-1 promoter, or in both organs using both constructs. In atg5, N remobilization from the rosettes to seeds was almost completely reestablished when transformed with the pSAG12::ATG5 construct. However, transformation with the pSAG12::ATG5 construct only partially restored seed composition. In contrast, seed N and C composition was largely restored by transformation with the pGly::ATG5 construct, even though the early leaf senescence phenotype was maintained in the atg5 background. Cotransformation with pSAG12::ATG5 and pGly::ATG5 completely restored the WT remobilization and seed composition phenotypes. Our results highlight the essential role of autophagy in leaves for nitrogen supply and in seeds for the establishment of carbon and nitrogen reserves. Introduction Autophagy is a universal mechanism that facilitates the degradation of unwanted cell constituents in the lytic compartments of eukaryotic cells. Under normal conditions, autophagy operates at a basal level that constitutes housekeeping machinery and participates in cell homeostasis (for reviews, see Avila-Ospina et al. 2014; Bozhkov 2018; Chen et al. 2019a; Su et al. 2020). Under nutrient starvation and with aging, autophagy activity is enhanced, suggesting that autophagy plays a key role during leaf senescence in maintaining cell homeostasis through extensive recycling (Doelling et al. 2002; Hanaoka et al. 2002; Thompson et al. 2005; Xiong et al. 2005; Breeze et al. 2011). Guiboileau et al. (2012) demonstrated the prominent role of autophagy in nutrient recycling and in the remobilization of organic nitrogen from the Arabidopsis rosette leaves to the seeds, by monitoring 15N fluxes after labeling. Authors showed that N remobilization dramatically decreased in the atg18a RNAi line and in the atg5 and atg9 mutants compared with wild-type (WT). The decrease was more moderate when plants were grown under sufficient nitrate conditions compared with low nitrate conditions, but still substantial. In addition, Chen et al. (2019b) showed that the remobilization of nitrogen from the rosette leaves to the seeds was improved in the Arabidopsis transformants overexpressing the ATG8a or ATG8g genes. The role of autophagy in N remobilization was further demonstrated in maize and rice (Li et al. 2015; Wada et al. 2015). Mutations in the ATG single genes of the autophagy core machinery disrupt the normal development of autophagosomes and lead to hypersensitivity to N and C starvations, as well as to premature leaf yellowing (Thompson et al. 2005; Phillips et al. 2008; Yoshimoto et al. 2009; Chung et al. 2010; Lenz et al. 2011; Guiboileau et al. 2012, 2013; Masclaux-Daubresse et al. 2014; Floyd et al. 2015; Wang et al. 2015; Fahy et al. 2017). The atg mutants present an overall decrease in plant biomass with especially smaller rosettes and reduced seed yield (Doelling et al. 2002; Thompson et al. 2005; Phillips et al. 2008; Guiboileau et al. 2012). The elevated salicylic acid (SA) accumulation in autophagy mutants is partially responsible for their early leaf senescence phenotype (Yoshimoto et al. 2009). Using the atg5-1.sid2 and atg5-2.NahG double mutants impaired in both autophagy and SA signaling, Guiboileau et al. (2012) demonstrated that defects in N remobilization were as strong in the atg5-1.sid2 and atg5-2.NahG double mutants as in the atg5 single mutant. By contrast, the early yellowing phenotypes of atg5-1.sid2 and atg5-2.NahG were strongly attenuated compared with atg5. This demonstrated that the defects of the atg5-1.sid2 and atg5-2.NahG double mutants for N remobilization were related to the absence of autophagy activity and not to the premature senescence of their source leaves. Another notable phenotypic trait of the atg mutants is the change in their seed composition (Guiboileau et al. 2012; Di Berardino et al. 2018; Sera et al. 2019; Barros et al. 2023). While the decrease of N remobilization from the source leaves to the seeds in atg mutants would suggest lower N concentration in atg seeds, it was shown that the atg mutants displayed higher N concentrations in their seeds compared with WT. This higher seed N concentration in the atg mutants is associated with a lower C concentration and less seed production (Guiboileau et al. 2012). Thus, the imbalance in seed N-filling in atg mutants is also due to a lack of carbon allocation for seed production and lipid reserve deposition. As a consequence, the amount of the seed storage proteins is modified in the atg5 mutant (Di Berardino et al. 2018), and disruption of autophagy causes moderate decrease in the triacylglycerol content of the seeds (Fan et al. 2019). Interestingly, Minina et al. (2018) showed that, at the opposite, the overexpression of ATG5 or ATG7 under the control of the 35S promoter increases fatty acid concentration in the seeds. It remains to be determined whether the alteration in seed composition in atg mutants is due to a defect in autophagic activity in the source leaves or in the seeds. Indeed, transcriptomic data indicate that the expression of ATG genes is enhanced in the leaves with the senescence and in the seeds during the maturation phase that corresponds to the seed filling period. Di Berardino et al. (2018) confirmed the upregulation of all the ATG genes in both siliques and seeds during maturation. Using GFP-ATG8f reporter lines, authors showed that autophagosomes are present in the Arabidopsis embryo cells. In maize, the increase of the ATG8-PE form, which is involved in autophagy flux, during the development of endosperm and embryo, attested for a role of autophagy in seed maturation (Chung et al. 2009). More recently, Sera et al. (2019) confirmed the pivotal role of autophagy in regulating metabolism within the endosperm of rice. However, considering the well-documented impact of autophagy deficiency in leaves, which triggers substantial modifications in metabolic pathways (Guiboileau et al. 2013; Masclaux-Daubresse et al. 2014), including those related to photosynthesis, it becomes imperative to investigate whether alterations in the source leaves of autophagy mutants can influence seed composition. In a recent publication, Erlichman et al. (2023) addressed this query by conducting reciprocal crosses between WT and atg mutants in Arabidopsis. Their study highlights the role of autophagy in maternal tissues during seed development and its effect on protein concentration in the seed. In order to address this question, we decided to restore autophagy activity specifically in the source leaves, taking benefit that the early yellowing symptoms of atg mutants are accompanied by the upregulation of the senescence-related gene SAG12 (Noh and Amasino 1999; Masclaux-Daubresse et al. 2014). SAG12, which encodes a papain like cysteine protease, is one of the most highly induced genes in Arabidopsis and in Brassica napus during senescence (Guo et al. 2004; Poret et al. 2016). Unlike other senescence genes, SAG12 is not upregulated by environmental stresses but only responsive to leaf aging (Noh and Amasino 1999). The coding DNA sequence (CDS) of AtATG5 was subsequently expressed under the regulation of the SAG12 promoter in the atg5 mutant to specifically reinstate autophagy in senescing leaves when nitrogen remobilization is most active. Furthermore, autophagy activity in the sink seeds was restored by expressing the AtATG5 CDS under the control of the seed-specific soybean glycinin-1 promoter. To comprehensively reinstate autophagy in both senescing leaves and seeds, the 2 constructs were combined in the same atg5 plant. The phenotyping of all these complemented lines was conducted under the same low nitrate conditions as used by Guiboileau et al. (2012) for characterization of N remobilization and seed filling in autophagy mutants. Results Expression of pSAG12::ATG5 and pGlycinin::ATG5 in atg5 mutant Like many autophagy genes, ATG5 is expressed at higher levels in senescing leaves and seeds during the maturation phase, when protein storage deposition occurs (Avila-Ospina et al. 2014; di Berardino et al. 2018). Like many other autophagy mutants targeted to single ATG genes, atg5 mutants exhibit strong phenotypes such as early leaf yellowing, reduced biomass, and hypersensitivity to stress, which are exacerbated upon nutrient deficiency. Among all the Arabidopsis autophagy mutants described so far, atg5 mutants are those best characterized and described in the literature (Thompson et al. 2005; Guiboileau et al. 2012, 2013; Havé et al. 2018). We therefore chose in this study to restore autophagy activity in the seeds or in the senescent leaves of atg5 mutants by expressing the ATG5 coding sequence under the control of specific promoters. The Arabidopsis SAG12 and Glycinin G1 (GLYMA_03G163500) genes are well-known markers of senescing leaves and seed embryos, respectively (Noh and Amasino 1999; Dean and Finer 2023). To restore autophagy specifically in senescing leaves or maturing seeds, ATG5 CDS was cloned under their promoters. Several independent pSAG12::ATG5 and pGly::ATG5 transformed lines in the atg5 background were obtained. After confirmation of the expression of the ATG5 transgene in the different transformants, 2 atg5 lines carrying and expressing the pSAG12::ATG5 construct (named atg5-S1 and atg5-S2; Supplementary Fig. S1) and 2 atg5 lines carrying the pGly::ATG5 construct (named atg5-G1 and atg5-G2) were selected for further characterizations. In addition, the pGly::ATG5 construct was transformed into the atg5-S1 line to provide the 2 atg5-S1.G1 and atg5-S1.G2 double transformants expressing the ATG5 CDS in both senescing leaves and seed embryo. All these lines contain only one copy of the constructs, as verified by segregating transformants upon kanamycin and/or by DsRed selection. By definition, the lines transformed with pSAG12::ATG5 are denoted as rosette-complemented lines, while those transformed with pGly::ATG5 are referred to as seed-complemented lines. The tissue-specific expression of the ATG5 transgene under the control of pSAG12 and pGly in atg5 was confirmed using Western blots with antibodies raised against ATG5 (Supplementary Fig. S2). ATG5 protein was detected (i) in the leaves of atg5 pSAG12::ATG5 plants, but not in their seeds, and (ii) in the old siliques (containing maturing seeds) of atg5 pGly::ATG5 plants, but not in their leaves. Additionally, we detected ATG5 in the roots of atg5 pSAG12::ATG5 plants, although at much lower levels than in senescing leaves, consistent with the findings of the study by James et al. (2019). ATG5 transgene suppresses atg5 early leaf senescence when expressed under the control of SAG12 promoter, but not when expressed under Gly promoter Autophagy mutants such as atg5 display early leaf senescence phenotypes, hypersensitivity to nitrate limitation, defects in protein degradation, and nitrogen remobilization (Guiboileau et al. 2012). The rosette phenotype of complemented lines was examined all along plant growth, and we found that the early leaf yellowing phenotype of atg5 that is usually clearly observed at 60 d after sowing (DAS) under our low nitrate growth conditions was totally abolished in the atg5-S1, atg5-S2, atg5-S1.G1, and atg5-S1.G2 lines, but not in the atg5-G1 and atg5-G2 lines that display the same phenotype as atg5 (Fig. 1, A and B). Counting the number of senescing leaves (i.e. harboring yellow color on 50% of the leaf surface) showed that the rosette-complemented lines, atg5-S1 and atg5-S2, indeed displayed similar green phenotype as WT showing that overexpressing ATG5 under the control of the SAG12 promoter abolished early senescence in atg5 leaves (Fig. 1, B and C). To confirm this observation, different markers of leaf senescence were monitored. The SAG12 steady-state mRNA levels were quantified and, in order to avoid dilution due to mixed leaf tissues, we separated the leaves in 3 batches containing: the oldest leaves (from leaf 1 to 15), the intermediate leaves (from to 16 to 30), and the youngest still growing leaves (youngest leaves). In good accordance, with the yellowing phenotypes, the SAG12 steady-state mRNA levels were significantly and sharply lower in the oldest leaves of atg5-S1, atg5-S2, and WT than in the oldest leaves of atg5 (Fig. 2A). Chlorophyll concentrations were accordingly significantly lower in the oldest leaves of atg5 than in the oldest leaves of atg5-S1, atg5-S2, and WT (Fig. 2B). Western blots show that Rubisco and chloroplastic GS2 protein contents are maintained in atg5-S1 and atg5-S2 at the same level as in WT, while they are strongly lower in atg5 (Fig. 2C; Supplementary Fig. S3). By contrast, atg5 accumulated more GS1 protein than the other genotypes, which is consistent with its early senescence phenotype (Masclaux-Daubresse et al. 2014). Results then confirm that the complementation of atg5 using transformation with the pSAG12::ATG5 construct abolishes the early leaf senescence phenotypes currently found in autophagy defective mutants. The early leaf yellowing phenotype of atg5 mutant is suppressed by the pSA12::ATG5 construct. Plants were grown on sand with low nitrate supply (2 mm). Plants are grown on sand in growth chamber (21/17 °C day/night; 160 µmol m−2 s−1) under low nitrate conditions (2 mm nitrate). Plants are cultivated in long days (16 h light/8 h dark) for 60 d. A) Rosette phenotype of the atg5-S1, atg5-G1, atg5-G2, atg5-S1.G1, atg5-S1.G2 lines, atg5 mutant, and WT Col-0 at 60 DAS. Scale bar = 1 cm. B) Dissection of the first 30 leaves of atg5-S1, atg5-S2, atg5, and Col-0 shows the absence of leaf senescence phenotype on complemented lines. Observations were made on 4 different plants per genotypes. One representative plant is shown here for each genotye. Scale bar = 1 cm. C) Number of senescing leaves (mean and Sd are shown). Different letters indicate statistically significant differences according to Student’s t-test (P ≤ 0.05, n = 4 plants). Plant growth experiment was repeated and showed similar results. Senescence markers confirm the suppression of early leaf senescence phenotype in atg5-S1 and atg5-S2. A)SAG12 mRNA steady-state levels were monitored by RT-qPCR using specific primers and normalized to a synthetic reference gene that combines EF-1α and APC2, taking the reference Ct value as √(EF-1α × APC2). Data are the means and Sd of the expression values measured on the oldest leaves of 3 plants and the results are presented on a logarithmic scale. Different letters indicate statistically significant differences between genotype in the different leaves according to Student’s t-test (P ≤ 0.05, n = 3 plants). B) Chlorophyll concentrations in the oldest, intermediate, and youngest leaves of Col-0 WT, atg5-S1, atg5-S2, and atg5. Data are the means and Sd. Different letters indicate statistically significant differences between genotype in the different leaves according to Student’s t-test (P ≤ 0.05, n = 3 to 4 plants). C) The Rubisco large subunit and glutamine synthetases (GS1 and GS2) contents were monitored by Western blot using specific antibodies. Proteins were extracted from the oldest leaves of WT, atg5-S1, atg5-S2, and atg5 plants grown under low nitrate supply for 60 DAS in short days. The same protein amounts were loaded in each lane on a same gel and transfer on the same membrane. Plant growth, yield, and biomass allocation are partially restored by pSAG12::ATG5 transformation but not by pGly::ATG5 It is well known that in autophagy mutants, rosettes are smaller and yield is lower than in WT (Thompson et al. 2005; Guiboileau et al. 2013). At the end of the plant cycle, when the seeds were mature, we harvested and separated the rosettes, stems, silique envelopes, and seeds for the different genotypes to measure dry weights (DWs) and calculate partitioning. The biomass and the yield of rosette-complemented lines, atg5-S1 and atg5-S2, were significantly higher than that of atg5 but still lower than that of WT (Fig. 3, A and D). Plant DWs between atg5 and the seed-complemented lines, atg5-G1 and atg5-G2, displayed marginal difference (Fig. 3B). The difference was mainly due to a weak increase of seed yields in the seed-complemented lines compared with atg5 (Fig. 3E). The atg5-S1.G1 showed significant improvement of both plant DW and seed yield compared with atg5-S1 (Fig. 3, C and F). This underlines the positive impact of combined supplementation on overall plant growth and seed yield. Complementation of atg5 by pSAG12::ATG5 partially restores plant biomass and yield as well as DW partitioning among the different organs. Least-square means from 2 independent experiments ± Se are shown (n ≥ 8 plants for each genotype). Different letters indicate significant differences between genotypes (Tukey’s test, P ≤ 0.05). The DW (g) of the whole plant A to C) and of seeds (seed yield; D to F) was measured at harvest. Biomass partitioning G to I) was calculated from the DW of the rosette, stem, silique envelopes (which are presented in Supplementary Fig. S4), and seeds for all the genotypes. All these results show that positive effects on the growth and yield production of atg5 were exclusively attained through the restoration of autophagy in rosettes by complementation with pSAG12::ATG5. The enhanced biomass production observed in atg5-S1 and atg5-S2 is reflected by the higher DW, compared with atg5, of all the organs taken individually as shown in Supplementary Fig. S4. As a result, the restoration of biomass partitioning in the different organs approaches near completeness in the atg5-S1 and atg5-S2 lines, reaching levels comparable with the WT in the atg5-S1.G1 lines (Fig. 3, G to I). Complementation of atg5 mutant by pSAG12::ATG5 restores nitrogen recycling and remobilization from the rosette to the seeds Isotope labeling using K15NO3 was used to estimate the remobilization and allocation of nitrogen in the different organs at harvest as described by Marmagne et al. (2022). In few words, K15NO3 was provided to plants at the vegetative stage for 1 d. Then plants were returned to unlabeled watering solution for the rest of the vegetative and reproductive development. Total nitrogen and 15N enrichment were then determined on the dry matter of the different organs collected at seed maturity (Fig. 3). From the DWs, N concentrations, and 15N enrichments measured in each organ, we calculated the partitioning of 15N in each organ (as % of the whole plant) in order to compare it with the partitioning of the biomass based on DWs (Figs. 3 and4). The 15N present in the new organs formed after labeling originates from N remobilization process. Of special interest is the partitioning of 15N in the seed compartment that is an indicator of the nitrogen remobilization efficiency (NRE), which is a component of the nitrogen use efficiency. Complementation of atg5 by pSAG12::ATG5 restores 15N partitioning and 15N remobilization efficiency. Least-square means from 2 independent experiments ± Se are shown (n ≥ 8 plants for each genotype). Different letters indicate significant difference between the genotypes (Tukey’s test, P ≤ 0.05). A to C) Partitioning of the whole plant 15N content in the rosette, stem, siliques, and seeds is presented as % of the whole plant. D to F) NRE was calculated as the ratio of 15NHI (15N in seeds as % of the whole plant) to HI (DW in seeds as % of the whole plant DW). Irrespective of the genotypes, the partitioning of 15N in the different organs does not parallel the partitioning of DW (Figs. 3, G to I and 4, A to C) and that discrepancies between the genotypes look much more dramatic for 15N partitioning than for DW partitioning. The large difference observed between genotypes and especially between atg5 and WT is due to the fact that autophagy has a prominent role in the control of nitrogen remobilization from source leaves to the seeds as shown by Guiboileau et al. (2012). Indeed, in atg5, up to 60% of the 15N isotopes absorbed at the vegetative stage remained sequestered in the rosette, and only 18% of the 15N reached the mutant seeds (Fig. 4, A and B). The WT performance for 15N remobilization is much better as only 17% of the 15N isotopes taken up at the vegetative stage remained in the WT rosette leaves and 60% reached the WT seeds (Fig. 4A). The 2 rosette-complemented lines, atg5-S1 and atg5-S2, performed significantly better for nitrogen remobilization than atg5 (Fig. 4A). Close to WT performance, <20% of the whole plant 15N remained in the rosettes of these lines, and more than 50% of the 15N absorbed was allocated to their seeds (Fig. 4A). Considering the variation of plant growth among the different genotypes and normalizing 15N partitioning in seeds (referred to as 15NHI according to Marmagne et al. [2022] by the harvest index [HI, DW partitioning in seeds]), we can see that the 2 rosette-complemented lines exhibit comparable NRE as WT (Fig. 4D). By contrast, 15N partitioning in the 2 seed-complemented lines, atg5-G1 and atg5-G2, remained very close to that of atg5, but with a significant effect on 15N rosette and seed partitioning (Fig. 4B). However, the 15NHI to HI ratio shows that 15N remobilization efficiency was not different from atg5 for atg5-G1 and atg5-G2 (Fig. 4E). Considering that the transformation of atg5 by pSAG12::ATG5 provided a significant improvement of the 15NHI and that the complementation of atg5 by pGly::ATG5 weakly modified 15NHI, we can conclude that nitrogen remobilization from the rosette leaves to the seeds is mainly controlled by autophagy process in the senescing leaves. This is confirmed by the double transformation in which the introduction of pGly::ATG5 in the atg5-S1 line does not provide any added value for NRE (Fig. 4F). However, thanks to its better seed yield (Fig. 3F), the atg5-S1.G1 line shows an higher 15N partitioning to the seeds (15NHI) compared with atg5-S1, which is very close to that of the WT (Fig. 4C). Seed composition remains unchanged when atg5 is complemented with pSAG12::ATG5 but is restored by the pGly::ATG5 complementation Mutation of ATG5 strongly compromises N and C remobilization to the seeds and results in striking modifications of the N and C concentrations in the seeds (Guiboileau et al. 2012; Di Berardino et al. 2018). N and C concentrations measured as N% and C% (g of N or C per 100 g of DW) were analyzed in the different organs. Consistent with the findings of the study by Guiboileau et al. (2012), our results indicate that N% was higher in all the atg5 organs compared with WT, while C% is lower only in atg5 seeds (Figs. 5 and6). The rosette N% of atg5-S1 and atg5-S2 was significantly lower than in atg5, approaching levels similar to the WT rosettes (Fig. 5J). This suggests a robust and almost complete restoration of the rosette N% in atg5-S1 and atg5-S2. The restoration of N% in the stem of atg5-S1 and atg5-S2 was partial but still significant (Fig. 5H). N% in the silique envelopes of atg5-S1 and atg5-S2 was also partial (Fig. 5D), and values ranged between the WT and atg5 levels. In the seeds, the N% of atg5-S1 and atg5-S2 were close to the atg5 level, and N% of the atg5-S2 seeds was not different from atg5 seeds (Fig. 5A). Results show that the rosette complementation of atg5 by pSAG12::ATG5 efficiently restored N% in the rosette, partially in stem and silique envelopes, but not in seeds. By contrast, C% in the different organs of the rosette-complemented lines atg5-S1 and atg5-S2 was not different from atg5, except for seeds where it was higher in the atg5-S1 and atg5-S2 compared with atg5. This higher C%, significant only for atg5-S2, confirms the low effect of pSAG12::ATG5 on seed composition (Figs. 5A and 6A). Complementation of atg5 by pSAG12::ATG5 or pGly::ATG5 partially restores N% in seed. Least-square means from 2 independent experiments ± Se are shown (n ≥ 8 plants for each genotype). Different letters indicate significant difference between the genotypes (Tukey’s test, P ≤ 0.05). N concentrations (N%) was measured in the seeds A to C), silique envelopes D to F), stem G to I), and rosette J to L) in the 8 genotypes. Complementation of atg5 by pSAG12::ATG5 or pGly::ATG5 partially restores C% in seed. Least-square means from 2 independent experiments ± Se are shown (n ≥ 8 plants for each genotype). Different letters indicate significant difference between the genotypes (Tukey’s test, P ≤ 0.05). C concentrations (C%) was measured in the seeds A to C), silique envelopes D to F), stem G to I), and rosette J to L) in the 8 genotypes. While the N% of the rosette, stem, and siliques of the seed-complemented lines atg5-G1 and atg5-G2 were not significantly different from atg5 (Fig. 5, E, H, and K), N% of seeds was significantly lower in atg5-G1 and atg5-G2 than in atg5, but still much higher than in WT (Fig. 5B). This observation indicates that the embryo complementation of atg5 only partially restored N%. The most striking effect of embryo complementation was observed for the C% of the seeds. The seed C% in atg5-G1 and atg5-G2 was increased by 20% compared with atg5 (Fig. 6B). The seed C% of atg5-G1 was then not different from that of the WT seeds. These findings demonstrate that the restoration of autophagy activity in the embryo of atg5 not only restores partially N% in seeds but also brings C concentration to the level observed in the WT seeds. Such a feature is found also in the fully complemented lines atg5-S1.G1 and atg5-S1.G2, which exhibit higher C% in seeds than the atg5-S1 line (Fig. 6C). It is noteworthy that the combination of the 2 constructs pSAG12::ATG5 and pGLY::ATG5 in the atg5-S1.G1 line can restore the same N% and C% as in the WT for all the organs (Figs. 5 and 6). This highlights the complementary roles of rosette and seed autophagy in seed filling (Figs. 5C and 6C). Discussion The aim of this work was to determine the cause of the defects observed in autophagy mutants for source to sink resource allocation and seed composition. As atg mutants exhibit early leaf yellowing under low nitrogen conditions (Yoshimoto et al. 2009; Guiboileau et al. 2012), we focused first on senescence markers to characterize our different transgenic lines. As expected, only the pSAG12::ATG5 constructs abolished the early leaf senescence phenotype of atg5, while the atg5-G1 and atg5-G2 lines complemented in the seeds using the pGly::ATG5 construct did not display any change regarding leaf yellowing relative to atg5. The atg5-S1 and atg5-S2 lines recovered higher biomass and seed yield compared with atg5, certainly due to extended carbon fixation linked to their greener phenotype. By contrast, gain of biomass and yield was very weak in the seed-complemented lines relative to atg5. Nevertheless, transformation by the pGly::ATG5 construct, as well as by the pSAG12::ATG5, modified plant growth as evidenced by the biomass distribution. Indeed, DW partitioning in atg5-S1 and atg5-S2 was almost the same as in WT, and the atg5-G1 and atg5-G2 lines exhibited an intermediate pattern between atg5 and WT. It is noticeable that the DW partitioning of the double-complemented line atg5-S1.G1 was not different from that of WT. Complementation with the pSAG12::ATG5 and pGly::ATG5 constructs produced different changes in the partitioning of 15N. Expression of pSAG12::ATG5 in atg5 significantly increased 15N remobilization from the rosette leaves to the seeds and restored NRE to the same level as in the WT. Given that ATG5 was detected in the roots of atg5 pSAG12::ATG5 plants, it is likely that some 15N remobilized to the seeds also originates from the root tissues. However, the root contribution to nitrogen remobilization is likely minimal, as the root/shoot ratio is very low in Arabidopsis plants grown under short-day conditions. By contrast with pSAG12::ATG5, the pGly::ATG5 construct only provided a very weak restoration of 15N partitioning in atg5. NRE indicator was not different from that of atg5 indicating that the beneficial effect of the seed complementation was mainly related to the increase of the seed yield that enhanced the increase of 15N allocation to the seeds. In other words, the higher 15N allocation to the seeds in the pGly::ATG5 complemented lines was a consequence of their higher seed production that lead to enhanced sink strength for nitrogen. Higher N% in atg5 rosette leaves is due to defects in complete protein recycling (Guiboileau et al. 2013; Havé et al. 2018). While sharp differences in N% between atg5 and WT could be observed for rosette and stem, differences of N% in silique and seeds were more moderate, due to the fact that atg5 mutant produces less seeds and has lower HI than WT. As the sink strength is the primary driver of N remobilization (Masclaux-Daubresse and Chardon 2011; Marmagne et al. 2020), autophagy mutants mitigate the negative effects of limited nitrogen remobilization by producing fewer seeds. Nevertheless, theirs seeds are fairly properly filled. The atg5-S1 and atg5-S2 lines exhibited intermediate N% phenotypes between WT and atg5, irrespective of the organs, with a rather good restoration of N% in the rosette and the stems. By contrast, atg5-G1 and atg5-G2 were ineffective to restore N% in their rosette, stem, and siliques. However, these lines partially restored N% in their seeds. Despite their low NRE, the pGly::ATG5 complemented lines better controlled seed N% relative to atg5. Strikingly, the double complementation in both seeds and senescing leaves could fully restore N% in all the organs, as evidenced by the atg5-S1.G1 line. Altogether, results indicate that seed N concentration is controlled by both autophagy-dependent N remobilization from the rosette leaves and by autophagy-dependent seed N-filling. Carbon concentration is weakly affected in the atg5 rosette, stem, and silique but significantly lower in atg5 seeds. Complementation of atg5 by pSAG12::ATG5 poorly increased the C% of the seeds, despite prolonged leaf longevity and extended carbon fixation period. This indicates that autophagy occurring in the seeds during maturation plays an essential role for carbon influx in the seeds, irrespective of the photosynthesis activity of source leaves. Accordingly, complementation by pGly::ATG5 almost completely restored the C% of seeds, to a level similar to WT seeds, regardless of the strong leaf senescence phenotype and the shorter photosynthetic period of the atg5-G1 and atg5-G2 rosettes. Altogether, results show that autophagy in seeds is central for the control of the seed N% and C% and that autophagy in source leaves only partially control N% of seeds. These contributions are consistent with the fundamental role of autophagy in the maturation of seed storage proteins (Reyes et al. 2011; Di Berardino et al. 2018) and in the regulation of starch and sugar metabolism during seed maturation (Sera et al. 2019). In summary, our investigation underscores the critical role of autophagy in the leaves, contributing significantly to leaf longevity, to the maintenance of carbon fixation, and the effective remobilization of nitrogen to the seeds. However, when autophagy activity remains impaired in the seeds, leaf complementation fails at restoring the N% and C% to the same level as in WT. Conversely, the complementation of atg5 mutant for seed autophagy facilitates the recovery of N% and C% in the seeds, but it fails to ensure efficient N remobilization from the leaves. Even though seed composition is restored, lines with exclusive seed complementation still exhibit diminished biomass and low yield. The optimal recovery is achieved by combining the 2 constructs that complement mutant in both seeds and leaves during senescence, as it ensures both efficient N remobilization and seed N and C storage. Nevertheless, the double-complemented lines still exhibit slightly lower biomass and yield than WT. This may be due to the cost of transgene overexpression and the mismatch between the chosen promoters for complementation and the specific temporal expression of ATG5 during vegetative and reproductive development. Izumi et al. (2013) showed that growth retardation of atg mutants under short days is partly related to the night time energy availability. Autophagy provides free amino acids as an energy source during extended nights when sugar availability is limited. Many of the Arabidopsis ATG genes exhibit circadian rhythmicity, with increased expression in the night and fluctuations more pronounced in short days than in long days (Supplementary Fig. S5). ATG5 shares similar expression pattern with the nitrate reductases NIA1 and NIA2 genes, whose expression is known to be regulated by sugar availability. Like the NIA genes, ATG5 is likely controlled by the C/N balance and sugar availability (Huarancca Reyes et al. 2018). Since circadian rhythmicity is absent from the SAG12 expression pattern (Supplementary Fig. S5), we speculate that the day/night regulation of autophagy activity, required to provide cellular energy from amino acid catabolism, cannot be satisfied in the atg5 double-complemented line, which limits the production of biomass compared with WT. Our results then demonstrate that autophagy plays a crucial role in many plant growth processes, starting from the earliest stages of development. As expected, the strategy we employed, using promoter specific of leaf senescence or seed development, does not fully compensate for the defects associated with the absence of autophagy in early developmental stage or in other organs. The SAG12 and glycinin promoters highlight the importance of autophagy during senescence and seed maturation. Identifying other promoters more suitable to explore the role of autophagy in processes such as germination, leaf growth, or pollen maturation will facilitate the demonstration of the role of autophagy in these specific contexts. Materials and methods Plant material Arabidopsis thaliana Col-0 (WT) and atg5 T-DNA mutant (SALK_020601) were used in this study. The atg5-S1 and atg5-S2 lines were constructed by transformation of the atg5 mutant with the pSAG12::ATG5 construct. The atg5-G1 and atg5-G2 lines were constructed by transformation of atg5 mutant with the pGly::ATG5 construct. The atg5-S1.G1 and atg5-S1.G2 lines were constructed by transformation of atg5-S1 transformant with the pGly::ATG5 construct. The pSAG12::ATG5 construct was generated using the G50285 pENTR223 entry vector from ABRC containing the ATG5 CD. The ATG5 CDS was cloned downstream the SAG12 promoter into the pMAZ01 plasmid (James et al. 2019) using Gateway technology (Thermo Fisher Scientific, Courtaboeuf, France). The atg5 mutant plants were then transformed by the pMAZ01-pSAG12::ATG5 via Agrobacterium tumefaciens (GV3101::pMP90) using floral dipping (Clough and Bent 1998). Several primary transformants were selected for all the transformations on the basis of their resistance to antibiotics. Several independent homozygous lines presenting single insertion segregation rate were further analyzed for the presence of transgene (Supplementary Table S1) using PCR on genomic DNA as shown in Supplementary Fig. S1. Only the atg5-S1 and atg5-S2 lines were used in the analyses as they presented more homogeneous results. The ATG5 CDS was cloned using Gateway technology (Thermo Fisher Scientific) into the binary vector pBinGlyRed2 (Huai et al. 2015) containing a DsRed marker gene to generate the pGly::ATG5. The atg5 mutant and the atg5-S1 plants were then transformed by the pGly::ATG5 via A. tumefaciens (GV3101::pMP90) using floral dipping (Clough and Bent 1998). Growth conditions For remobilization assays, plants were grown on sand according to Masclaux-Daubresse and Chardon (2011) in growth chamber (21/17 °C day/night; 160 µmol m−2 s−1) under low nitrate conditions (2 mm nitrate). Plants were first cultivated in short days (8 h light/16 h dark) for 50 d and then transferred to long days (16 h light/8 h dark) to induce flowering. To measure N remobilization from rosette to sinks, plants were labeled at vegetative stage (42 DAS) with 2 mm K15NO3 (10 atom% excess) during 24 h. After labeling, the roots of plants and the sand in pots were carefully rinsed with clear water. Plants were then kept on growing until seed maturity. Remobilization assays were repeated for 2 times. For reverse transcription quantitative PCR (RT-qPCR), Western blots, and senescence profiling, plants were grown on sand in growth chambers and under low N conditions (2 mm nitrate). Plants were cultivated in short days (8 h light/16 h dark; 21/17 °C day/night; 160 µmol m−2 s−1), harvested 60 DAS, and then stored at −80 °C. The rosette leaves were separated to form 3 groups: the first 0 to 15 leaves represented the oldest leaves, the next 14 to 30 leaves represented intermediate leaves, and the rest of the rosette (≥31) represented the youngest leaves. For the Supplementary Fig. S2, plants were grown on potting soil in a greenhouse under standard nutritive conditions. All organs were harvested 50 DAS and then stored at −80 °C. DNA purification and PCR analysis Extraction of genomic DNA was realized using “Edwards” buffer (Edwards et al. 1991), and amplification by PCR, with ATG5 specific primers (Supplementary Table S1 and Fig. S1), was performed using dream TAQ polymerase (Thermo Fisher Scientific). RNA purification and RT-qPCR analyses Total RNA was isolated from frozen plant tissue with phenol/chloroform 5:1 (P1944, Sigma) according to the study by Di Berardino et al. (2018). Reverse transcription of extracted RNA into cDNA was realized using RevertAid H Minus Reverse Transcriptase (EP0452, Thermo Fisher Scientific) according to the manufacturer’s protocol. RNA was treated with the DNase I, RNase-free kit (EN0521, Thermo Fisher Scientific). An inhibitor of RNase, RiboLock RNase Inhibitor (E00382, Thermo Fisher Scientific) was added to the reaction mixture according to the manufacturer’s protocol. The 10-µL reaction mixture contained 2.5 µL of cDNA (corresponding to 25 × 10−2 to 25 × 10−5 µg of cDNA), 0.3 µL of each primer (10 µm), 5 µL of a Takyon Rox SYBR MasterMix dTT Blue solution (UF-RSMT-B0710, Eurogentec, Liège, Belgium) containing the Taq polymerase, the deoxynucleotide triphosphates (dNTPs), and the Sybr Green in a reaction buffer, and 2.2 µL of water. The RT-qPCRs were run on a CFX 96 thermocycler (Bio-Rad, Marnes-la-Coquette, France) using a first step at 95 °C for 3 min and then 40 cycles of 10 s at 95 °C, 10 s at 58 °C, and 30 s at 72 °C. A final step consisted of an increase of 0.1 °C s−1 to reach 95 °C. The primers used for RT-qPCR had an average length of 20 bases and were designed in order to amplify fragments between 100 and 240 bp. All primers presented an efficiency of 100 ± 5%. For all RT-qPCR analyses, EF-1α (At5g60390) and APC2 (At2g04660) were used as reference genes (Di Berardino et al. 2018; Supplementary Table S1). They were combined as a synthetic reference gene in the calculation of relative expressions. Extraction of total proteins and Western blot Total proteins were extracted by homogenizing 100 mg of frozen material in 500 µL of buffer (50 mm Tris-HCl pH 7.5, 150 mm NaCl, 2 mm DTT, 1 mm phenylmethylsulfonyl fluoride, 50 mm MG132, and 1× plant protease inhibitor cocktail) and clarified by centrifugation at 16,000 × g for 5 min at 4 °C. Supernatant was filtered through 4 layers of Miracloth. Protein concentration was determined using the 2-D Quant kit (GE80-6483-56, GE Healthcare, Little Chalfont, UK) according to the manufacturer's protocol. Protein extract was then mixed with 0.5 volumes of 4X NuPAGE LDS sample buffer (NP008, Thermo Fisher Scientific) containing 0.1 m DTT. Five micrograms of proteins were separated by 12% SDS–PAGE. Gels were electroblotted to polyvinylidene difluoride membranes using the Trans-Blot Turbo transfer system (Bio-Rad). Antibodies were used at a 1:1,000 dilution in 3% milk in 0.1% phosphate-buffered saline Tween (PBS-T) buffer. Peroxidase-conjugated secondary antibody (1:10,000 dilution) was used for chemiluminescence detection using an ImageQuant LAS 4000 (GE Healthcare Life Science). The membranes were then stained with amido black to verify that an equal amount of protein was loaded in each lane. For the Supplementary Fig. S2, 6 µg of proteins was separated by 4% to 20% Mini-PROTEAN TGX Stain-Free Protein Gels (#4568094, Bio-Rad), and the ChemiDoc MP Imaging System was used for imaging the gels (Bio-Rad). For glutamine synthetases (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 (Lemaître et al. 2008). Antibodies against tobacco (Nicotiana tabacum) Rubisco large sudunit (LSU) were kindly provided by Dr A. Suzuki (IJPB, INRA, Versailles, France). Anti-ATG5 (AS15 3060, Agrisera Antibodies, Vannas, Sweden) and anti-APG8a (ab77003, Abcam, Cambridge, UK) are commercial antibodies. Chlorophyll determination Chlorophyll extraction and the determination of chlorophyll concentrations were performed as described by Diaz et al. (2008) using 50 mg of leaf fresh matter. Determination of total nitrogen and carbon concentrations and 15N abundance Labeled plants were collected at maturity, when they were completely dry. The shoots were then separated into seeds, stems, silique envelopes, and rosette. Collected samples were dried and weighed to determine the DW of each part of the plant (seeds, stem, silique, and rosette) and then ground into fine powders. A subsample of each powder sample (0.800 to 1.200 mg) was assayed for total N and C contents and 15N abundance using the FLASH 2000 Organic Elemental Analyzer coupled to an isotope ratio MS (Thermo Fisher Scientific). The N and C concentrations (N% and C%) of each sample were expressed as mg (of N or C) per 100 mg of DW. The 15N abundance was calculated as atom% (percentage of total N) and defined as Asample% = 100 × (15N)/(15N + 14N). The 15N natural abundance value (Acontrol%) was 0.3663 in this study. The 15N enrichment (E%) was calculated as Asample%−Acontrol%. The absolute quantity of 15N in each sample (µg of 15N per organ) was defined as Q = (DW × N%) × E%. 15N concentration was then calculated as Q/DW and expressed as µg 15N per g of DW. The partitioning of dry matter in seeds, stem, and rosette was calculated as a percentage of the whole plant (DWorgan/DWwhole plant). N partitioning in seeds, stem, and rosette was calculated as a percentage of the whole plant (mg of Norgan/mg of Nwhole plant). The 15N partitioning in seeds, stem, and rosette was calculated as a percentage of the whole plant (µg of 15Norgan/µg of 15Nwhole plant). Indicators used to monitor NRE were calculated as explained by Havé et al. (2017) and Marmagne et al. (2022). Statistical analyses ANOVA followed by Tukey’s honestly significant difference test and 2-sample t-tests were used in this study. All statistical analyses were performed using the free software environment R Version 4.1.0. Accession numbers Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers AT5G17290 (ATG5), AT5G45890 (SAG12), and Glyma13G18450 (GLYCININ). Supplementary Material