Salicylic acid is a key player of Arabidopsis autophagy mutant susceptibility to the necrotrophic bacterium Dickeya dadantii

Autophagy is a ubiquitous vesicular process for protein and organelle recycling in eukaryotes. In plant, autophagy is reported to play pivotal roles in nutrient recycling, adaptation to biotic and abiotic stresses. The role of autophagy in plant immunity remains poorly understood. Several reports showed enhanced susceptibility of different Arabidopsis autophagy mutants (atg) to necrotrophic fungal pathogens. Interaction of necrotrophic bacterial pathogens with autophagy is overlooked. We then investigated such interaction by inoculating the necrotrophic enterobacterium Dickeya dadantii in leaves of the atg2 and atg5 mutants and an ATG8a overexpressing line. Overexpressing ATG8a enhances plant tolerance to D. dadantii. While atg5 mutant displayed similar susceptibility to the WT, the atg2 mutant exhibited accelerated leaf senescence and enhanced susceptibility upon infection. Both phenotypes were reversed when the sid2 mutation, abolishing SA signaling, was introduced in the atg2 mutant. High levels of SA signaling in atg2 mutant resulted in repression of the jasmonic acid (JA) defense pathway known to limit D. dadantii progression in A. thaliana. We provide evidence that in atg2 mutant, the disturbed hormonal balance leading to higher SA signaling is the main factor causing increased susceptibility to the D. dadantii necrotroph by repressing the JA pathway and accelerating developmental senescence.

www.nature.com/scientificreports/ mammals and in plants. In plant, autophagy process was found to be compromised in all the atg mutants defective in the single ATG genes 23,24,31 . The ATG8 protein is encoded by nine genes in Arabidopsis and although the ATG8 protein is a key player of the autophagy core machinery, no phenotype had been reported so far for the different atg8 single mutants isolated in Arabidopsis, possibly due to functional redundancies. The conjugation of ATG8 to phosphatidyl-ethanolamine (PE) relies on a complex conjugation system that involves the ATG5 protein. The ATG2 protein is involved in lipid recruitment for autophagosome membrane elongation. Both ATG5 and ATG2 are encoded by single genes and their mutants display strong senescence and limited-growth phenotypes although phenotypes are more severe in atg2 than in atg5 [32][33][34][35] . Autophagy is post-transcriptionally regulated by the TOR protein kinase (Target of Rapamycin [36][37][38]. Autophagy genes are up-regulated under stress conditions, amongst which plant infection by pathogens 39,40 . In plants, recent studies show that autophagy is involved in plant pathogen interactions and the involvement of autophagy machinery in disease/resistance is highly dependent on the pathosystems as well as on the plant physiological status [41][42][43][44] . The fine tuning of the cell death related to the hypersensitive response (HR), which is a strong resistance mechanism, is altered in autophagy mutants 34,45 . In addition, several studies show that compromising autophagy results in an enhanced susceptibility to necrotrophic fungi [46][47][48] . However it remains unclear if alteration in plant immunity is directly due to the lack of autophagy degradation process or results from indirect effects of autophagy alteration.
Dickeya dadantii is a necrotrophic plant pathogenic bacterium that causes soft rot disease on a large host range of crops and is able to infect Arabidopsis thaliana 49,50 . As is is the case of many other necrotrophs, D. dadantii produces large amounts of plant cell wall degrading enzymes that cause soft rots of plant tissues. Several lines of defense allow the plants to limit infection by D. dadantii including the production of ROS via the disturbance of iron homeostasis and by the membrane located NADPH oxidase [50][51][52][53] . Several lines of evidence show that D. dadanti triggers JA defense pathway 50,53 .
The role of autophagy in plant tolerance to pathogens was mainly documented regarding the cell death hypersensitive response to bacteria such as Pseudomonas and in response to necrotrophic fungi 44,48 . So far, there is no report dealing with necrotrophic bacteria and autophagy. Here, we show that the defect of Arabidopsis autophagy mutants' tolerance to the bacterial necrotroph D. dadantii is not linked to the autophagy activity per se, but is an indirect effect of impaired fine tuning of SA defense signaling.  54 . The sid2 mutant was kindly provided by Pr. JP Métraux. The pUBI::ATG8a::GFP overexpressor Arabidopsis line was obtained from Chen et al. 55 .

Materials and methods
Bacterial inoculation and quantification of disease severity. Inoculation experiments were performed with the D. dadantii 3937 strain as described in 56 . Bacteria were grown in Luria-Bertani medium. Plants used for RNA extraction were inoculated by leaf infiltration using a syringe without a needle with a bacterial suspension at 1 × 10 8 Colony Forming Unit/mL (CFU) made up in 10 mM MgSO 4 , mock inoculated controls consisted of leaves infiltrated with 10 mM MgSO 4 . Plants used for symptom scoring were inoculated by making a small hole with a needle in the leaf limb, and then spotting 5 μL of a bacterial suspension at a density of 1 × 10 8 CFU/mL made up in 50 mM potassium phosphate buffer (pH 7) on the top of the hole. Symptom severity scoring was performed according to the 0 to 5 severity scale described in Rigault et al. 56 57,58 Leaves were harvested 24 h post inoculation (H p.i.) and then frozen in liquid nitrogen. Total RNAs were purified with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. The total RNA concentration was determined using a NanoDrop ND-1000 spectrophotometer (NanoDropTechnologies Inc., Wilmington, DE, USA). RNA samples were treated with Turbo DNaseI (Ambion, Saint-Aubin, France) RNase-free to remove any DNA contamination. A total of 1 µg of DNase-treated RNA was reverse transcribed using the High Capacity cDNA Reverse Transcription Kit and 50 ng of random hexamers following the supplier's instructions. One microlitre of the 1:10 diluted cDNA was subjected to real-time quantitative PCR using SYBR Green PCR Mastermix (Applied Biosystems, Foster City, CA, USA) and gene-specific primers designed to amplify 100-150-bp fragments from each gene of interest and the reference genes APT and Clathrin. Primer sequences used for qRT-PCR are indicated in Supplementary Table 1.

Salicylic acid quantification.
For each sample, 2 mg of dry powder were extracted with 0.8 mL of acetone/ water/acetic acid (80/19/1 v:v:v). Salicylic acid stable labelled isotope used as internal standard was prepared as described in Le Roux et al. 59 . 1 ng of standard was added to each sample. The extract was vigorously shaken for 1 min, sonicated for 1 min at 25 Hz, shaken for 10 min at 10 °C in a Thermomixer (EPPENDORF, and then centrifuged at 8000g, 10 °C for 10 min). The supernatants were collected, and the pellets were re-extracted twice with 0.4 mL of the same extraction solution, then vigorously shaken (1 min) and sonicated (1 min; 25 Hz). After the centrifugations, the three supernatants were pooled and dried (final volume 1.6 mL).

Results
Arabidopsis tolerance to the bacterial necrotroph D. dadantii is altered in autophagy mutants but does not necessitate autophagy activity. The role of autophagy in Arabidopsis tolerance to necrotrophic phytopathogens was only reported for fungal plant pathogens (Table 1), using atg mutants or transgenic lines over-expressing ATG genes. Consistent data were obtained in all reports showing that atg mutants display increased susceptibility to necrotrophic fungi. It is not the case for biotrophs and hemibiotrophs for which the role of plant autophagy depends on the pathosystem 44,48 . To know whether autophagy is also involved in Arabidopsis tolerance to bacterial necrotrophs, we addressed the issue with the model bacterial necrotroph Dickeya dadantii 60 .
For this purpose, two different atg mutants, one affected in lipid recruitment for autophagosome formation (atg2) and the other affected in the conjugation system permitting ATG8 lipidation and anchorage to the membrane of autophagosomes (atg5) were inoculated with D. dadantii. Wild type Col-0 and the two atg2 and atg5 mutants were inoculated by spotting a bacterial suspension on leaves of 6 weeks old plants. The severity of the symptoms was scored based on 0-5 scale 56 . Severity of the symptoms on atg2 mutant was higher over time than that observed on the WT plants ( Fig. 1). Interestingly, the susceptibility level of the atg5 mutant was similar to that of the WT. The absence of difference between atg5 mutant and WT was confirmed on the two atg5 mutant allele (atg5-1 and atg5-2 33 ; data not shown). This indicates that only atg2 was more susceptible than Col-0 to D. dadantii while the atg5 mutant known to display a less severe yellowing phenotype than atg2, was not affected in its susceptibility to D. dadantii. Such difference between atg2 and atg5, then raised the question of the role of www.nature.com/scientificreports/ the autophagy machinery in the tolerance to D. dadantii, and the potential link with a senecence status of the mutants. We then used an Arabidopsis line with enhanced autophagic activity consisting in the overexpression of the ATG8a protein under the control of Ubiquitin promoter 55 . Following D. dadantii inoculation, plants overexpressing ATG8a displayed reduced symptom severity compared to the WT suggesting a positive role of autophagy activity on the tolerance of Arabidopsis to D. dadantii (Fig. 1).
Autophagy genes are not up-regulated upon D. dadantii infection. As disease symptoms observed on the different atg lines suggested a link between plant response to D. dadantii and autophagy, we monitored ATG gene expression following plant inoculation. RT-qPCR was performed to monitor the transcript level of 10 autophagy genes chosen for their positive response to stress in litterature 40 . None of these genes was found to be upregulated in response to infection (Fig. 2) by contrast with the defense gene marker PR1 that was upregulated  www.nature.com/scientificreports/ in the same plants as expected. These data suggest that transcriptional activation of autophagy genes was not required in the response of Arabidopsis to D. dadantii infection.

Salicylic acid dependent susceptibility of atg2 mutant to D. dadantii.
Our results indicate that functional ATG2 protein is essential to control disease symptoms in Arabidopsis upon D. dadantii infection but that working ATG5 protein and operational autophagy is not essential. Even though autophagy genes are not induced by infection, enhanced constitutive autophagic activity provides positive effect on plant tolerance to D. dadantii. If autophagy machinery was directly involved in Arabidopsis tolerance to the bacterium, then any knock-out mutant affected in autophagy should be more susceptible to the bacterium. Our results suggest that defect in autophagy activity was not the direct cause of Arabidopsis susceptibility to D. dadantii. Indirect effects of autophagy defects on plant tolerance to pathogens might be related to hormonal balance. It is well known that autophagy defect triggers hormonal disorders and especially exacerbates SA production and SA signaling, then enhancing fast and spectacular leaf senescence like symptoms 41,42 . Because senescence/yellowing phenotypes observed on the atg2 mutant is stronger than in the atg5 mutant, we suspected exacerbation of SA response in atg2 by comparison with atg5 and considered the possibility that SA accumulation in atg2 was the origin of its enhanced susceptibility to D. dadantii.
In order to determine whether SA is involved in the increased susceptibility of atg2 mutant, we monitored the susceptibility of the atg2.sid2 double mutant defective in SA synthesis as the SID2 gene encodes an isochorismate synthase which is involved in SA biosynthesis in Arabidopsis, in particular in response to pathogens 61 . Figure 3A shows the phenotypes of Col-0 wild type, atg2 and atg2.sid2 mutants infected with D. dadantii. Both soft rot symptoms and leaf yellowing phenotypes were clearly more severe on atg2 leaves than on Col-0 and atg2.sid2 leaves (Fig. 3). Although atg2 was early senescing by comparison with Col-0 and atg2.sid2 plants 34 (Fig. 3A), we observed enhanced chlorosis around the D. dadantii inoculation spots which were not present in uninfected plants or before inoculation (data not shown). This suggested that infection accelerated leaf senescence symptoms on atg2 plants. It can be noticed that disease severity was similar in WT, sid2 and atg2.sid2 double mutant. These data indicate that the higher susceptibility of the atg2 mutant can be restored to the WT level when SID2 gene is non-functional, suggesting the prominent role of SA in the increased susceptibility of atg2 mutant to D. dadantii. This confirms that autophagy degradation pathway is not essential for Arabidopsis tolerance to D. dadantii. Nevertheless, higher constitutive autophagic activity artificially enhanced by ATG8a overexpression may help increasing plant tolerance, possibly through positive effects on leaf longevity, meaning negative effect on SA production.

Salicylic acid signaling rather than SA amount is involved in atg2 mutant enhanced susceptibility to D. dadantii. Previous reports indicate that the level of salicylic acid in autophagy mutants is higher
than that of WT 54,62 . To determine the mechanism by which high SA level could lead to enhanced susceptibility, we hypothesized that high amounts of SA could repress the JA signaling pathway which is known to limit D. dadantii infection 50,57 . Indeed, it is commonly reported that SA signaling represses JA signaling and that ET and JA signaling are commonly synergistic [17][18][19][20]63 .
To further investigate the possibility that SA and JA signaling were modified in atg2 mutants, we determined the status of SA and JA mediated defenses in Col-0, sid2, atg2, atg5, atg2.sid2 and atg5.sid2 plants. For this purpose, we monitored the expression of defense marker genes of the SA (PR1) and the JA (PDF1.2) pathways in infected plants. Controls consisted of plants inoculated with 10 mM MgSO 4 and naïve plants that were untreated. The latter naïve plants were used to determine the initial immunity level of plants before inoculation. Transcript levels of both the SA defense marker gene PR1 and of the JA defense marker gene PDF1.2 were increased in WT Col-0 plants following D. dadantii infection compared to controls (Fig. 4). While no PR1 transcript was detected in the sid2 mutant line, higher levels of PDF1.2 transcripts were detected in these plants compared to Col-0 confirming the repressive action of the SA pathway on the JA pathway. Strikingly, in the atg2 mutant, higher levels of PR1 transcripts were detected compared to the WT (10 times higher in atg2 compared to Col-0 infected plants) and an up-regulation of this level was observed in response to D. dadantii compared to the mock inoculated plants. The level of PR1 transcripts in atg5 mutant was unchanged compared to the WT indicating that SA signaling in the atg5 mutant was not affected. The expression level of PDF1.2 was reduced in the atg2 mutant compared to that of Col-0 which is consistent with a high SA signaling. The transcript level of PR1 was undetectable in both atg2.sid2 and atg5.sid2 double mutants, indicating that SA signaling was totally abolished by introducing the sid2 mutation. Interestingly, in naïve atg simple and double mutant plants, the level of PDF1.2 transcripts was reduced compared to that of mock inoculated plants. Our data suggest that mechanical stress triggers JA signaling in atg mutant backgrounds, a process which is repressed by D. dadantii infection. To know whether senescence was also involved in this interaction, the transcript level of the senescence marker gene SAG12 was monitored. SAG12 was only expressed in atg2 mutant plants (Fig. 4). Intriguingly, it was down regulated by mock inoculation and up-regulated by infection confirming the accelerated senescing phenotype we observed on atg2 infected plants compared to mock atg2 plants.
To investigate the role of SA content in the response of autophagy mutants to D. dadantii, SA content was quantified in naïve plants (Fig. 5). Both atg2 and atg5 mutants accumulate higher SA levels than Col-0 plants. The level of SA was reduced in both double mutants atg2.sid2 and atg5.sid2 relative to single mutants confirming the role of SID2 in SA production in Arabidopsis. www.nature.com/scientificreports/

Discussion
The mechanisms by which plant autophagy affects immunity are complex and vary depending on the considered pathosystems. In some cases, opposite mechanisms can take place depending on plant age or infectious stage. For example, in young Arabidopsis leaves, autophagy has a pro-death activity enhancing HR mediated cell death 45 while in older leaves, autophagy seems to counteract cell death 34 . Plant autophagy was shown to counteract the bacterial pathogen Pseudomonas syringae pv. tomato which, surprisingly, secretes an effector able to trigger autophagy 64 .
Several reports indicate that the susceptibility of atg mutants was increased in response to fungal necrotrophic pathogens (Table 1). To determine the effect of atg mutations on Arabidopsis susceptibility to the bacterial necrotroph D. dadantiii, we used an atg2 and atg5 mutants. The ATG2 and ATG5 genes are single genes in Arabidopsis as in many plant species, and are essential for autophagic activity as their mutants cannot form autophagosomes. The atg2 and atg5 mutants were previously characterized by many groups amongst which Pr. R. Vierstra and Pr. K. Yoshimoto groups, and display the typical hypersensitivity to abiotic stresses described for many autophagy  www.nature.com/scientificreports/ mutants in Arabidopsis 33,34 that is related to leaf yellowing phenotypes and necrotic spots. It might be noticed that atg2 mutant phenotypes are stronger than those of atg5 mutant. Our findings showing different behaviors of two autophagy mutants are not surprising with regard to the literature. Indeed, it appears from the literature that the susceptibility of atg mutants to different pathogens is not always similarly affected compared to the WT plants. For instance, although we found that in response to D. dadantii, the atg2 mutant showed enhanced susceptibility, the susceptibility of the same atg2 mutant to the necrotrophic fungus Botrytis cinerea has not been found affected 65 . In atg7 and atg12 Arabidopsis mutants, the susceptibility to the necrotrophic fungus Sclerotinia sclerotiorum was shown unaffected 66 . All these reports indicate that although a general trend observed in autophagy mutants is the susceptibility to necrotrophic pathogens, some mutants are likely not affected. A common trend in necrotrophy is cell wall degrading enzyme production by pathogens. Cell wall homeostasis is tightly linked to membrane stability which could be in part under the control of autophagy 29,67,68 . The mechanisms explaining the higher susceptibility of some atg mutants to necrotrophic pathogens remained unknown 44,48 . One hypothesis is that as a pro-survival mechanism, autophagy hampers the growth and development of necrotrophs which proliferate preferentially on dead cells. In this situation, autophagy would ensure the clearance of degraded cellular components thus protecting plant tissues from activating stress responses such as the SA signaling pathway 44 . Another hypothesis is that the hormonal balance which is disturbed in atg mutants favors necrotrophs. Both hypotheses are not mutually exclusive and can depend upon the the pathosystem considered. In the interaction between Arabidopsis and D. dadantii, we demonstrate that the enhanced susceptibility of the atg2 mutant relies on SA signaling Our data also show that the senescing state of the plants correlates with disease development since SAG12 expression and chlorotic phenotype were enhanced in atg2 plants and lost in the atg2.sid2 mutant. It is however intriguing that there is no correlation between SA content and PR1 expression level in atg2 and atg5 mutants. One interpretation is that senescence and SA form an amplification loop in atg2 and that both are required to trigger high susceptibility 34 . Such amplification loop between senescence and SA was indeed recently described 69,70 .
We showed that in atg2 higher SA content and signaling coincided with low JA signaling thus explaining increased susceptibility, as JA signaling has been shown to contribute to Arabidopsis tolerance to D. dadantii 50,57 . The disease phenotype of the atg5 mutant, which is similar to that of the WT plants, may be explained by the fact that atg5 is senescing later than atg2 mutant. This may be due to a differentially active senescence and immunity molecular machinery in each atg mutant. For instance, transcript levels of genes encoding transcription factors related to immunity and senescence (WRKY, ERF and JAZ) are different in an atg5 compared to an atg9 mutant 54 . Interestingly, ATG9 interacts with ATG2 in the phagophore expansion process 23 suggesting that atg9 and atg2 mutants could harbor the similar immunity and senescence defects that differ from those of the atg5 mutant. In addition to the differential expression of transcription factor encoding genes in different atg mutants, it is possible that different proteolytic activities reside in each mutant. Indeed, it was recently shown that autophagy deficiency in the atg5 mutant resulted in an alteration of cellular proteolytic activities 67 . The actors of immunity including transcription factors may be targeted by proteolysis which ensures fine tuning of adequate defense responses 71,72 . It would be interesting to compare the proteolytic activities in atg2 and atg5 mutants to investigate the lifetime of key immunity and senescence related transcription factors.
Altogether, we provide here the evidence that SA plays a pivotal role in the enhanced susceptibility of atg2 mutant to D. dadantii through the repression of JA-dependent defenses, and that autophagy degradation function is not directly involved in plant tolerance to D. dadantii. While our study argues against the involvement of plant autophagy in the tolerance to this type of aggressors, it confirms the importance of leaf senescence status in the susceptibility of plants and points to the hormonal balance as a key player in this process. In addition, overexpression of ATG8 proteins, that had been shown to provide positive effects on plant tolerance to many stresses such as drought in several plant species (Chen et al. 73 for a review), significantly increased the tolerance of our Arabidopsis plants to D. dadantii infection.