Autophagy Is Rapidly Induced by Salt Stress and Is Required for Salt Tolerance in Arabidopsis

1Soil salinity is a major abiotic factor that limits crop yield. Globally, about 900 million hectares of land were estimated to be saline, and more than 30% of the irrigated crops were salt-affected (FAO1). High concentrations of NaCl inhibits plant water uptake, and Na+ and Cl- accumulated in the cytosol lead to ion toxicity. As results, photosynthetic rates are reduced, leading to energy depletion and accumulation of excessive reactive oxygen species (ROS). In order to alleviate the osmotic and ionic stresses, plants close their stomata to minimize water loss, reduce their growth rates, and limit intracellular Na+ concentration by compartmentalization in the vacuole. Gradually, relocation of Na+ away from young tissues takes place to exclude Na+ from growing organs, and plants may enter a growth recovery phase to resume growth at a reduced steady rate.

2With engineering of salt-tolerant crops in mind, numerous studies have been carried out to elucidate salt stress response, with emphases on transcription regulation and ABA signaling. Enormous progress have also been made in understanding the ion transport mechanisms, such as the discovery of Salt Overly Sensitive (SOS) signaling cascade and the in-depth studies on the HKT and NHX transporters. Additionally, the molecular basis for tissue tolerance has been revealed, which includes biosynthesis of osmolytes, such as glycinebetaine, sugar alcohols, polyamines, and proline.

3Compared with the topics above, how salt-challenged plants manage to maintain their energy level and re-allocate the limited resources is less well understood. One pathway that may contribute to salt-elicited nutrient recycling is macroautophagy (hereafter autophagy).

4Autophagy is a bulk degradation pathway that helps maintain cellular homeostasis. In this pathway, obsolete proteins and damaged organelles are enveloped by an expanding double-membraned vesicle, the isolation membrane/phagophore, which matures to a sealed autophagosome before fusing with the lytic vacuole. The inner membrane of the autophagosome along with the cargo, termed the autophagic body, is then degraded by vacuolar hydrolases, and amino acids and other macro molecules are released back into the cytosol through transporters.

5The hallmark of autophagy is the formation of the autophagosome. As key players in this process, autophagy-specific ubiquitin-like proteins (UBLs) ATG8/LC3/GABARAP act as protein scaffolds to mediate phagophore expansion. Conjugation of ATG8 to the phagophore requires the activity of several other ATG proteins. Firstly, newly synthesized ATG8 is truncated by the cysteine protease ATG4 to expose the C-terminal Glycine residue. The Glycine is then conjugated to the amino group of Phosphatidylethanolamine (PE) in a ubiquitin-like conjugation reaction catalyzed by ATG7 as the E1, ATG3 as the E2, and the ATG12-ATG5-ATG16 complex as the E3 enzyme. PE-conjugated ATG8 stably associates with both phagophore and completed autophagosomes, hence is commonly used as a marker for microscopic study of autophagy. In addition, PE-conjugated ATG8 moves faster than the unconjugated ATG8 in SDS-PAGE gels, hence the amount of lipidated ATG8 is indicative of autophagic activity. Furthermore, by comparing the ATG8-PE levels in the presence and absence of vacuolar protease inhibitors, such as E-64d, or tonoplast H+-ATPase inhibitors, such as concanamycin A (Con A), autophagic flux can be quantified. The selective autophagic flux can be detected by measuring Neighbor of BRCA1 (NBR1) degradation. During selective autophagy, NBR1 binds both ATG8 and mono- and (especially) poly-ubiquitin, linking the ubiquitinated cargoes to the autophagy machinery. Then it is transported together with the cargoes inside the autophagosome to the lytic vacuole for degradation. Hence NBR1 serves as both a receptor and a selective substrate of autophagy, and the degradation of NBR1 is indicative of selective autophagic flux. Therefore, immunoblotting with antibodies against ATG8, NBR1, or epitope tags fused to ATG proteins provides more quantitative information for detecting autophagy.

6To see if the level of autophagic flux correlates with salt tolerance, Arabidopsis mutants defective in autophagy and mutants with reduced level of autophagy, as well as transgenic plants that have increased level of autophagy were selected for further analyses (Supplementary Figure S1). The five ATGs belong to the core autophagy machinery. ATG5, ATG7, and ATG10 are required for ATG8-lipid adduct and autophagosome formation, and ATG5 have been shown to localize at the outer surface of the cortical endoplasmic reticulum (ER) to recruit ATG8 for phagophore assembly and expansion. Both atg5 and atg7 are autophagy-deficient mutants. ATG2 supposedly forms a complex with ATG18 to mediate the shuttle of ATG9 vesicles, which are a source of autophagosomal membranes. Autophagic flux is known to be reduced rather than absent in atg9. Transgenic Arabidopsis over-expressing GmATG8c (ATG8-OX) was used to represent plants with high levels of autophagy.

7As a house-keeping pathway, autophagy is generally maintained as a basal level and can be quickly induced by nutrient deprivation and various stresses. Several lines of evidence suggest that autophagy is positively involved in plant salt stress adaptation. NaCl treatments have been shown to induce transcriptions of several Autophagy-related (ATG) genes, especially ATG8s and ATG18s in Arabidopsis, salt cress, rice, wheat, tobacco, pepper, and foxtail millet. In Arabidopsis, RNAi-AtATG18a plants were hypersensitive to salt and osmotic stress in germination and seedling growth. The autophagy-deficient mutants, atg5 and atg7, as well as the ATG8-interacting autophagy cargo adaptor mutant nbr1, exhibited sensitivity toward drought and salt treatments. Mutation in ATG8-Interacting Protein 1(ATI1) led to salt sensitivity during germination. Intriguingly, it was reported that over-expression of GFP–AtAtg8f-HA led to reduced tolerance toward mild osmotic and salt stresses, but not to stronger stresses.

8Many questions remain on how autophagy contributes to salt stress tolerance in plants. When does autophagic flux peak following salt stress? Does autophagy participate in the clearance of salt-induced oxidized proteins? Does it contribute to osmolyte production? Does it have a role in Na+ uptake or sequestration in the vacuole?

9Here we show that the autophagic flux is rapidly induced by salt treatment. Then autophagy deficient (atg5 and atg7, atg10), defective (atg9 and possibly also atg2), and enhanced (Pro35S:GmATG8c, ATG8-OX) lines were selected to see if the levels of autophagy correlate with salt stress tolerance. Physiological analyses revealed that the atg mutants accumulated less soluble sugars and proline, whereas ATG8-OX plants accumulated more osmolytes. Oxidized protein is alleviated after treated with NaCl in the wild-type within 3 h. We also show that Na+ sequestration in the lytic vacuole of salt-stressed root cortex cells is correlated with the level of autophagy. Our observations suggest that salt stress rapidly triggers autophagy to facilitate bulk protein turnover, thus providing macromolecules and energy required for plant survival.

10ATG2, At3g19190; ATG5, At5g17290; ATG7, At5g45900; ATG8a, AT4G21980; ATG9, At2g31260; ATG10, AT3G07525; NBR1, AT4G24690.

11Arabidopsis (ecotype Columbia-0) was grown as described. Generally, seeds were surface-sterilized with 75% ethanol for 5 min, 100% ethanol for 1 min, rinsed with ddH2O for five times, then stratified at 4°C for 2 days before plated on 1/2 Murashige and Skoog (1/2 MS) medium (Sigma–Aldrich, United States) containing 0.8% (w/v) agar (Sigma–Aldrich, United States), 1% (w/v) sucrose (Sigma–Aldrich, United States), pH5.7. Liquid 1/2 MS medium was prepared in the same way, only without agar. The plants were grown at 16 h (22°C)/8 h (18°C) with a photosynthetic photon flux density at 90 μE m-2 sec-1. The T-DNA insertion mutants atg2-1 (Salk_076727), atg5-1 (SAIL_129B07, CS806267), atg7 (SAIL_11H07, CS862226), atg9 (SAIL_527_A02, CS874564), and atg10-1 (Salk_084434) were obtained from ABRC, and transgenic lines carrying Pro35S:GmATG8c were as described. All mutants and transgenic plants were verified by genomic PCR. Primers used are listed in Supplementary Table S1. All lines have been freshly propagated to ensure wild-type level germination rates on control medium. Phenotypes were documented as described. All images were analyzed with Image J2, and statistical analyses (F-test, Student’s t-test) were done with Microsoft Excel 2010.

12To construct ProATG8a:GFP-ATG8a, an 1199 bp fragment upstream of the start codon (ATG) of ATG8a (-1199 To 1) was PCR-amplified and inserted between Pst I and Nco I of pCAMBIA1302. Then a genomic fragment of ATG8a (942 bp) was inserted after GFP by homologous recombination with a ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China). The construct was verified by sequencing before introduced into Agrobacterium tumefaciens (GV3101) for floral dipping. Primary transformants were selected by antibiotic resistance and verified by PCR. The T3 homozygous marker line L5-1 was introduced into atg10 by crossing, and atg10/GFP-ATG8a plants were identified by genotyping of F2 individuals.

13To quantify autophagic flux, 10-day-old vertically grown seedlings (approximately 200 mg) were transferred to liquid 1/2 MS medium with 150 mM NaCl for 0, 0.5, 1, 3, and 6 h before harvested. To block autophagosome degradation in the lytic vacuole, a parallel set of seedlings were transferred to medium containing NaCl plus 0.5 μM Concanamycin A (ConA, Sigma–Aldrich, United States) and harvested at the same time points. Protein extraction, quantification, and immunoblotting were done as described. All SDS-PAGE gels were prepared with 6 M urea. Primary antibodies used were anti-GmATG8c (1:3000), anti-NBR1 (Agrisera, 1:3000), and anti-tubulin (Utibody, 1:5000). Each experiment was repeated for at least three times, and one representative result was shown. Quantification of immunoblots was done with Image J and statistical analyses (F-test, Student’s t-test) were done with Microsoft Excel 2013.

14Oxidized protein analysis was performed following a previous report. Ten-day-old seedlings grown on 1/2 MS vertical plates were transferred to liquid 1/2 MS medium with 150 mM NaCl for 0, 0.5, 1, 3, and 6 h before harvested for protein extraction. The extraction buffer contains 0.1 M Tris-HCl, pH 7.5, 0.3 M Sucrose, 1 mM EDTA, 0.1 mM phenylmethylsulphonylfluoride (PMSF), and 1% [v/v] β-mercaptoethanol. Extracts were centrifuged at 1,000 g for 10 min and supernatants were collected. Oxidized proteins were detected using an OxyBlot protein oxidation detection kit (Abcam, United Kingdom) according to the manufacturer’s instructions. Dinitrophenylhydrazine (DNP) signals (entire lane) were quantified by densitometry in Image J and normalized to the wild-type 0 h control value, which was set as 1. Each experiment was repeated for at least three times, and one representative result was shown. Coomassie blue staining of total proteins was used as the loading control.

15Ten-day-old seedlings grown on 1/2 MS vertical plates were transferred to liquid 1/2 MS containing 150 mM NaCl for 0, 8, and 24 h. Harvested samples were weighed, and free proline was quantified with ninhydrin assay at A520 nm with a kit (Comin Biotech, Suzhou, China) following manufacturers’ instructions. Total soluble sugar and reducing sugar contents were determined by the anthrone reagent method and dinitrosalicylic acid (DNS) method with a kit (Yuanye Biotech, Shanghai, China) following manufacturers’ instructions. The sugar contents were determined against a standard curve prepared with glucose (Sigma–Aldrich, United States). Three biological replicates were done in each case with consistent results, and representative results are shown.

16Ten-day-old seedlings grown on 1/2 MS vertical plates were transferred to liquid 1/2 MS containing 150 mM NaCl for 0, 8, and 24 h. A parallel set of seedlings were treated with NaCl plus 10 mM Chloroquine. Harvested samples were weighed and quantified for soluble amino acid on a membraPure Amino Acid Analyzer A300 (Germany) following manufacturer’s instructions.

17For germination assay, stratified seeds were sown on 1/2 MS medium with or without NaCl or mannitol. Four biological replicates were done, which are consist of four seed populations obtained from self-fertilization of independent parents. More than 100 seeds were used in each seed population. Radical protrusion was scored as germination at a 4-h interval during first 3 days, then daily until day 7.

18To observe NaCl-induced autophagosome formation, 4- or 5-days-old vertically grown GFP-AtATG8a seedlings were immersed in 100 mM NaCl, or 0.5 μM ConA, or both NaCl and ConA, for 30 min and 1 h, respectively, and scanned with a Leica SP5 (Leica, Germany) with a same set of scanning parameters. For each condition, at least 30 seedlings from three to five biological replicates were imaged, and >30 root cortex cells in the root hair zone were quantified for the numbers of autophagosomes/autophagic bodies.

19For imaging of Na+, 5-days-old Arabidopsis seedlings were transferred to liquid 1/2 MS (control) or 1/2 MS containing 100 mM NaCl (salt stress) for 6 h, and then CoroNa Green AM (Thermo Fisher Scientific) was added into the medium to a final concentration of 5 μM. Two hours later, the seedlings were stained with 5 μM FM4-64 (Thermo Fisher Scientific) for 5 min, washed thoroughly, and scanned with a SP5 (Leica, Germany) confocal microscope following previous reports. At least 20 seedlings from three biological replicates were imaged for each line in each condition, and >30 root cortex cells in the meristematic zone were quantified for the CoroNa Green AM fluorescent intensities in Image J.

20Ten-day-old seedlings were transferred to liquid 1/2 MS medium containing 150 mM NaCl for 0 (control), 30 min, and 3 h. RNA extraction, reverse transcription, RT-PCR and q-RT-PCR were performed as described. EF1a (At5g60390) was used as an internal control. Three biological replicates consisting of three technical repeats each were done. Transcript levels of the nine salt-inducible genes following NaCl treatment in ATG8-OX, atg5, and atg9 were compared with both WT control and with the same germplasm control using Student’s t-test. Primers used are listed in Supplementary Table S1.

21To see if the autophagic flux is affected by salt stress, we firstly observed autophagosome formation and autophagic body turnover by looking at the autophagosome marker GFP-ATG8a in transgenic Arabidopsis seedlings carrying ProATG8a:GFP-ATG8a (Supplementary Figure S2). The marker line has a similar growth rate to the WT both under normal growth conditions and under nitrogen or carbon limitation, or salt and osmotic stresses (Supplementary Figure S2). The vacuolar proton pump inhibitor Concanamycin A (ConA) was used to preserve the autophagic bodies in the central vacuole. Under controlled conditions, GFP-ATG8a signals were mainly observed in the cytoplasm, with sporadic, relatively large (1–2 μm in diameter) puncta representing autophagosomes also detected (Figure 1A, DMSO). In the presence of ConA, additional smaller puncta representing autophagic bodies were observed in the vacuole (Figure 1A, +ConA 0.5 h). When treated with 100 mM NaCl for 30 min, large puncta accumulated (Figure 1A, +NaCl 0.5 h). In the presence of ConA, the numbers of autophagosomes and autophagic bodies increased even more, indicative of induced autophagic flux (Figure 1A, +NaCl +ConA 0.5 h). Interestingly, the number of ATG8a puncta was slightly reduced at 1 h of salt treatment (Figure 1A, lower panel), suggesting that the autophagic flux peaked shortly after salt stress. Quantification of the number of autophagic bodies confirmed that the autophagic flux, represented by the difference between the numbers of autophagic bodies treated with NaCl only and that of NaCl plus ConA treatment, peaked at 0.5 h (Figure 1C). Imaging of an atg10 mutant carrying GFP-ATG8a under the same conditions showed that few autophagosome/autophagic bodies could be detected in the mutant with or without NaCl or NaCl plus ConA (Figure 1B). The imaging results indicated that autophagic flux can be rapidly induced by NaCl treatment in Arabidopsis.

22To confirm that NaCl treatment can rapidly induce autophagy, immunoblotting of ATG8s was performed. We first examined whether the anti-GmATG8c antisera could detect both the lipidated and the non-lipidated ATG8s following a previous report. Total membrane fraction from the wild-type and atg7 seedlings were collected, and the solubilized samples either treated or not with phospholipase D (PLD) were analyzed with immunoblotting. Unfortunately, the antibodies preferentially recognize un-lipidated ATG8s (Supplementary Figure S3). Hence the induction of autophagy was analyzed by comparing the difference in ATG8 protein levels between Concanamycin A (ConA)-treated and untreated samples at the same time points. Consistent with the imaging results (Figure 1), the levels of ATG8s were significantly induced at 0.5 and 1 h of NaCl plus ConA treatment (Figure 2).

23Since the autophagic flux can also be detected by measuring NBR1 degradation, changes in NBR1 levels following salt stress were monitored. The clear reduction in NBR1 protein level at 0.5 h of salt stress, in combination with the relative constant levels of NBR1 at the same time points in the presence of ConA (Figure 2), indicated that NBR1-dependent, selective autophagy peaked at 0.5 h following NaCl treatment.

24Salt-induced ATG8 flux, represented by an elevation in ATG8 levels following NaCl treatment and an even higher induction in NaCl plus ConA treatment, was not observed in atg5 or atg7 mutants (Supplementary Figure S4). ATG8 levels were induced in atg2 and atg9 only at 0.5 h (Supplementary Figure S4). NBR1 degradation was observed at 0.5 and 1 h in atg9, and at 6 h in atg2, however, not in atg5 or atg7 (Supplementary Figure S4).

25Oxidized proteins induced by ROS are known substrates for autophagy. To see whether autophagy may contribute to the clearance of oxidized proteins generated upon salt stress, levels of oxidized proteins were analyzed over a time-course of 6 h. In the WT, oxidized proteins accumulated significantly within 0.5 h of salt treatment, got back to the control level at 3 h, before returning to the 0.5–1 h level by 6 h. atg9 performed better than the WT in the clearance of oxidized proteins, with a reduction clearly observed at 1 h. atg2 and atg7 both had higher-than-WT levels of oxidized proteins before stress, which then got further induced by salt stress. Only at 6 h did the levels of oxidized proteins drop slightly in atg2 and atg7. Overall, the results indicate that autophagy could have a positive role in transiently reducing the level of oxidized proteins generated by salt stress (Figure 3).

26To see if autophagy is positively involved in salt stress adaptation, germination percentage of the autophagy mutants and ATG8-OX plants were documented in a time course on control and NaCl-containing plates. Under controlled condition, ATG8-OX germinated significantly faster than the wild-type (p < 0.01, paired student’s t-test), whereas the autophagy mutants were statistically similar to the wild-type (Figure 4A). In the presence of NaCl, germination was slowed down in all lines, especially at higher concentrations (Figures 4B–D). The two autophagy-deficient lines, atg5 and atg7, germinated significantly slower than the wild-type on all salt stress conditions (Figures 4B–D). atg9 and atg2 germinated faster than the wild-type on 50 mM NaCl, but not on higher concentrations (Figures 4B–D).

27Similar germination curves were observed in the atg mutants on mannitol-containing plates (Supplementary Figure S5). These physiological data suggested that the level of autophagy is positively correlated with the germination rate at salt- and osmotic- stress conditions.

28Consistent with the germination phenotypes, root-bending assay performed on 150 mM NaCl showed that the WT and the ATG8-OX lines responded normally to gravity stimulation on vertical plates supplemented with 150 mM NaCl. In contrast, atg5 and atg7 had less bending in their primary roots on NaCl plates (Supplementary Figure S6). Altogether, the phenotypes confirmed that autophagy is required for the adaptation of seedlings toward salt stress.

29Autophagy may cater to the timely production of salt-induced osmolytes, such as proline and soluble sugars. The levels of proline, soluble sugar, and reducing sugar were measured in control and salt-stressed (0, 8, and 24 h) seedlings (Figures 5A–C). Before salt treatment, ATG8-OX had significantly higher levels of proline and reducing sugar than the wild-type, whereas several atg mutants had significantly lower levels of the osmolytes compared to the wild-type. Following salt treatment, all lines accumulated more osmolytes. ATG8-OX generally had much higher levels of proline and sugars, especially at 24 h (Figures 5A–C). In contrast, the atg mutants accumulated significant less osmolytes compared with the wild-type (Figures 5A–C), suggesting that the production of proline and soluble sugars indeed relies on autophagy.

30To see if the salt-induced transcription is affected by the level of autophagy, we compared the expression patterns of salt-inducible genes in atg5 and atg9 mutants, ATG8-OX, and the wild-type with quantitative RT-PCR (Figure 6). The selected genes are known to be induced by salt stress either dependent or independent of ABA. Since we observed an induction of autophagic flux at 0.5 h of NaCl treatment, two relatively early time points (0.5 and 3 h) were selected for the transcript analysis. Expression of most markers was induced at 0.5 h in the wild-type, atg9, and ATG8-OX, and was further induced at 3 h (Figure 6). The expression patterns observed in the wild-type, atg9, and ATG8-OX indicated that autophagy may not affect salt-inducible gene expression directly. The extra high induction in gene expression observed in atg5 (Figure 6), however, is unexpected, and might have resulted from a yet unknown regulatory mechanism.

31To see whether autophagy might play a role in Na+ compartmentation upon salt stress, sodium ions were visualized with CoroNa Green, a fluorescent dye specific for Na+-imaging, with a same set of scanning parameters on a confocal microscope. The lipophilic dye FM4-64 was used to stain the PM, outlining the cells. Without NaCl, the fluorescence was barely visible (Supplementary Figures 7, 8) as reported. After 8 h of 100 mM NaCl treatment, striking differences in fluorescent intensity was observed among the lines (Figure 7 and Supplementary Figure S7). As described, Na+ accumulated in the vacuoles of root cortex cells in all lines, however, at very different quantities. Compared with the wild-type, very weak signals were detected in atg2, atg5, and atg7, whereas much stronger signals were observed in ATG8-OX (Figure 7 and Supplementary Figure S8). atg9 had lower-than-WT yet clearly visible signals (Figure 7 and Supplementary Figure S9). The observation suggested that sodium compartmentation in the root cortex cells positively correlates with the level of autophagy.

32In the past decade, several methods have been established for observing autophagy in plants, including staining acidic vesicles with the fluorescent dye monodansylcadaverine (MDC), electron microscopy of cells, etc. The most widely used method is to observe the presence and distribution of Pro35S:GFP-ATG8s and ProUBQ10:GFP-ATG8s in plant cells. Studies had revealed, however, that ectopic expression of ATG8s could have measurable impact on plant development. The Pro35S:GFP-ATG8 plants generally grow faster to larger sizes. In this study, we constructed GFP-ATG8a driven by its own promoter, and generated T3 homozygous lines. Line 5-1 had phenotypes indistinguishable from the WT under normal growth conditions, and shared the same starvation- and stress- induced phenotypes with the WT. A bonus is that the T-DNA insertion site in this line had been revealed from TAIL-PCR, so that the marker line can be easily genotyped after crossing with other lines. We also noticed that in this line, the strong nuclear ATG8 signals commonly observed in both Pro35S:GFP-ATG8 and ProUBQ10:GFP-ATG8 under controlled conditions is absent (Figure 1 and Supplementary Figure S2), indicating that the ATG8a level in this line is closer to the endogenous level.

33An interesting finding in this study is that the autophagic flux peaks as early as 0.5 h upon salt stress in Arabidopsis seedlings (Figures 1, 2). Similarly, it has been reported that autophagic flux was induced at 30 min following hypertonic stress (350 and 500 mOsmol/kg NaCl, approximately 150 to 250 mM NaCl) on in LLC-PK1 renal proximal tubule-like cells. Such speedy induction of autophagy is more likely based on post-translational modification rather than de novo synthesis of core autophagy proteins. What the possible modifications are remain to be explored. Apart from the possible regulation on ATG8 deacetylation, a known determinant on ATG8 activity, the redox-controlled ATG4 activity, might also regulate salt-induced autophagy. The ATG4 protease not only cleaves nascent ATG8 at its conserved C-terminal Glycine residue to promote conjugation of ATG8 to PE during autophagosome formation, but cleaves the amide bond between ATG8 and PE on completed autophagosome to recycle ATG8. The two Arabidopsis ATG4s have been demonstrated to have different substrate preference in vivo and in vitro, and the activity of Chlamydomonas ATG4 has recently been shown to be inhibited by oxidation at a Cysteine residue (C400) conserved between plants and yeasts. It can be postulated that the oxidative stress generated during salt stress may modulate ATG8-PE formation by regulating ATG4 activity.

34The germination assay clearly indicates that autophagy plays a positive role in ensuring timely germination upon salt stress treatment. There are at least two explanations for such observation. Firstly, autophagy is known to contribute to nutrient remobilization from source to sink, and autophagy mutants had insufficient protein degradation in their rosette leaves. Therefore, during seed maturation, the acquisition of seed storage proteins, free amino acids, fatty acids, and other macromolecules is likely defective in the autophagy mutants. When challenged by salt stress, the germinating atg mutants could have insufficient osmolyte production, thus exhibiting sensitivity toward osmotic challenge. In contrast, the ATG8-OX seeds could have benefited from higher levels of storage proteins and other macromolecules during germination. Such differences become clearer as the stress conditions become more severe (Figure 4 and Supplementary Figure S5), especially on the two autophagy deficient mutant, atg5 and atg7. Indeed, a global analysis on etiolated (carbon starvation) autophagy mutant seedlings showed that, amino acids, organic acids, and protein levels were significantly decreased in atg5. Secondly, autophagy participates in the clearance of damaged organelles such as peroxisomes, and salt stress is known to generate such damages. Salt stress also leads to accumulation of protein aggregates and oxidized proteins, both of which have been reported to be substrates of autophagy. On the other hand, autophagy is certainly not the only degradation pathway responsible for the elimination of damaged organlles, oxidized proteins, and protein aggregates. Our observation that the level of oxidized proteins are only transiently reduced in the wild-type indicated that autophagy may not be the major pathway involved in the long-term clearance of oxidized proteins. It has been reported that proteins oxidized in salt-stressed Arabidopsis are versatile not only in their functions but also in their sub-cellular localizations. A number of them are predicted to localize to the apoplast, the plasma membrane, and the nucleus, thus are likely to be degraded by other pathways. A recent report showed that in severely salt-stressed tomato roots (250 mM NaCl, 6 h), two of the three catalytic subunits, β2 and β5, of the proteasome are transiently modified to generate the stress proteasome. At 24 h, a normal proteasome profile reappeared. Such a transient apperance of the stress-induced active proteasome coincites with oxidized protein degradation and is likely a parallel pathway to autophagy in short-term salt stress response.

35Why does the autophagic flux, reflected by the turnover of both ATG8 and NBR1 in the lytic vacuole, take place at such an early time point? Considering that salt-induced transcription generally starts at a similar time point, it can be hypothesized that the efficient translation of stress-responsive proteins may rely on a reliable pool of amino acids. Indeed, the quantification of free amino acids following salt stress supported this hypothesis (Supplementary Figure S9). At the 8-h time point, the ATG8-OX seedlings had significantly lower levels of free amino acids, indicating that it was able to remobilize more amino acids for protein synthesis at an earlier time point. As a bulk degradation pathway, autophagy is likely a good candidate for maintaining such a pool. One may wonder which organelles or proteins are preferentially “eaten” during salt stress, since various substrates have been discovered in autophagy induced by different abiotic stresses. For instance, upon heat stress, autophagy is required for the clearance of the ubiquitinated protein aggregates. Hypoxia-induced autophagy, on the other hand, may use chloroplast proteins as substrates, as the autophagy mutants die quickly following submergence and their detached leaves turn yellow faster upon ethanol treatment. Interestingly, autophagy also regulates the compartmentation of sodium ions in root cells, although the underlying mechanism remains uncovered. A model summarizing our findings is presented (Figure 8). The interesting fact that autophagy gets induced very rapidly following salt treatment and is required for successful salt adaptation may be explored in molecular breeding of salt tolerant crops in future.

36QG, LL, and PZ conceived and designed the study. LL, PZ, DW, and QG analyzed the data and drafted the manuscript. LL, PZ, RZ, JF, JZ, JS, and ZW carried out the experiments. All authors have read and approved the final manuscript.

37The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.