Characterization of the early response of Arabidopsis thaliana to Dickeya dadantii infection using expression profiling

To draw a global view of plant responses to interactions with the phytopathogenic enterobacterale Dickeya dadantii, a causal agent of soft rot diseases on many plant species, we analysed the early Arabidopsis responses to D. dadantii infection. We performed a genome-wide analysis of the Arabidopsis thaliana transcriptome during D. dadantii infection and conducted a genetic study of identified responses. A limited set of genes related to plant defence or interactions with the environment were induced at an early stage of infection, with an over-representation of genes involved in both the metabolism of indole glucosinolates (IGs) and the jasmonate (JA) defence pathway. Bacterial type I and type II secretion systems are required to trigger the induction of IG and JA-related genes while the type III secretion system appears to partially inhibit these defence pathways. Using Arabidopsis mutants impaired in JA biosynthesis or perception, we showed that induction of some IG metabolism genes was COI1-dependent but, surprisingly, JA-independent. Moreover, characterisation of D. dadantii disease progression in Arabidopsis mutants impaired in JA or IG pathways showed that JA triggers an efficient plant defence response that does not involve IGs. The induction of the IG pathway by bacterial pathogens has been reported several times in vitro. This study shows for the first time, that this induction does indeed occur in planta, but also that this line of defence is ineffective against D. dadantii infection, in contrast to its role to counteract herbivorous or fungal pathogen attacks.

S1. The wild type strain 3937 (our collection) was isolated from Saintpaulia ionantha 20 (African violet). Bacterial strains were grown at 30°C in Luria-Bertani medium. Liquid 21 cultures were grown in a shaking incubator (180 rpm). For plant inoculations, the bacterial 22 inoculum was prepared as follows: an aliquot of the -80°C glycerol stock culture was streaked 23 on solidified LB medium (1.5% Difco agar) and grown for 48 h at 30°C. Then, a single 24 colony was used to inoculate a liquid culture. After 8 h of growth at 30°C, 100 μl of the 1 culture was plated on LB agar medium and incubated overnight. The bacteria were then 2 suspended in inoculation buffer (50 mM KPO 4 , pH 7) and inoculum density was adjusted to 3 the value mentioned in each experiment. 4 Arabidopsis thaliana lines used in this study are listed in Supplemental Table S1. Arabidopsis 5 seeds were sown in non-sterile soil, synchronized by incubation for two days at 4°C and 6 allowed to germinate and grow for three weeks. They were then potted individually 7 (pathogenicity assays) or in threes (RNA isolation) into separate pots and incubated for a    Table S2). Of these genes, 61 were up and 1 3 was down-regulated at both 12 and 24 hpi. 4 We analysed the putative functions of these genes by using the Protein Sequence database of 5 the Munich Information Centre [41]. 64% of the differentially expressed genes are annotated 6 as encoding proteins of known or predicted functions, distributed among 3 main categories 7 ( Table 1, Fig 1B). Not surprisingly, nearly one third of the modulated genes are associated 8 with the "interaction with the environment, cell rescue and defence" category. Another third 9 corresponds to the "metabolism-related" category, with an over-representation of amino acid 10 and secondary metabolisms. The third main category (15%) is related to the "cellular 11 transport" category (electron transport, drug/toxin transport). Down-regulated genes do not 12 clustered in any special functional category.
13 Table 1. Functional classification of Arabidopsis genes regulated in response to D.
14 dadantii infection.  19 conversion to OPC8-enoyl-CoA (Table 2). In addition to these biosynthesis genes, four of the The transcriptomic results were confirmed by qRT-PCR for a subset of Trp, IG and JA-4 related genes (Fig 3). All gene transcripts progressively accumulated from 12 hpi onwards, 5 with a relative fold change ranging from 1.9 to more than 80 24 hpi. In contrast to the 4 Activation of the two IG biosynthesis genes was not significantly different in the JA 5 biosynthesis mutants jar1 compared to the col-0 wild type plant 24 hpi (Fig 4). By contrast, in 6 the JA insensitive coi1 mutant no significant induction of the SOT16 gene was observed 24 7 hpi, while the induction of the CYP79B2 gene was not altered (Fig 4). These results reveal the 23 We quantified by qRT-PCR transcript accumulation levels of two IG-related genes ( Fig 5A  1 and B) and two JA-related genes (Figs 5C and D) after buffer inoculation, inoculation with 2 the wild type D. dadantii strain (3937) and with mutants altered in the three protein secretion 3 systems known to be involved in virulence (prtE, type I mutant, outC, type II mutant and 4 hrcC, type III mutant).
5 Figure 5A and B clearly show that IG-related gene transcript accumulation observed 30 hpi 6 with the wild type bacterial strain was considerably lower after infection with the mutant 7 strain with an altered protein secretion system I (T1SS prtE) and was similar to the buffer-8 inoculated control with the mutant strain with an altered protein secretion system II (T2SS 9 outC). Thus, T1SS and T2SS play a role in the induction of the IG biosynthesis pathway.
10 Conversely, inactivation of the secretion system III (T3SS hrp secretion system) leads to an 11 approximately two-fold IG-related transcript accumulation increase as compared to the wild 12 type strain.  [17]. In all JA mutants, we observed a significantly increased 10 number of plants with severe symptoms (scored 3) as compared to wild-type infected plants 11 three, four and five days post-inoculation ( Fig 7A). Moreover, symptoms appear earlier (2 and 12 3 dpi) in the coi1 and jar1 mutants. This confirms that the JA pathway is involved in partial 13 resistance of Arabidopsis to D. dadantii.
14 As IG-related genes were shown to be up-regulated during infection, we also investigated the  18 the related indole glucosinolate secondary metabolites (Fig 2). Moreover, our study also 19 revealed the modulation of JA-related genes expression. 4 the JA receptor COI1 (Fig 4). The occurrence of differential regulations along a biosynthetic 5 pathway is not unusual and could be related to the involvement of this pathway in a wide 6 metabolic network. Indeed, indole acetaldoxime -the product of CYP79B2 activity -serves as  25 systems involved in virulence to find out if a secreted virulence factor could trigger this gene 1 induction. We showed that the T1SS (PrtE) and T2SS (OutC) were individually necessary to 2 fully induce both pathways (Fig 5). Notably, induction impairment was more severe after 3 infection by the outC mutant than by the prtE mutant for both pathways. Of the hundred 4 proteins secreted by D. dadantii, pectinases are the predominant virulence factors secreted by 5 the T2SS, as revealed by the phenotype of the outC mutant that is totally unable to provoke 6 maceration symptoms [9]. We therefore extended our analysis to the ∆pel mutant that is 7 impaired in the synthesis of the five major pectate lyases PelABCDE. This analysis showed 8 that this limited set of five pectate lyases was required to trigger the observed inductions of 9 both JA and IG pathways (Fig 6).  4 However, these studies minimize the complexity of the glucosinolate response, as illustrated 5 by the diversity of glucosinolate species (more than 40 molecules, of which 4 known IGs) and 6 of nitrile specifier enzymes that generate various nitrile species in conjunction with 7 myrosinases. In addition, in vitro studies do not take into account well known bacterial 8 adaptations to various plant generated stresses encountered during infection, such as acidic 9 and oxidative stresses. Thus, we could potentially imagine a bacterial adaptive response to 10 toxic IG-breakdown products.
11 The phenotypic analysis of three Arabidopsis mutants impaired in IGs biosynthesis (atr1, 12 cyp79B2cyp79B3 and sot16) did not reveal any significant difference in symptom severity 13 after D. dadantii inoculation as compared to the wild type ( Fig 7B). This indicates that IGs do    22 Data deposition 23 Microarray data were deposited at Gene Expression Omnibus database (accession number