The symbiotic transcription factor MtEFD and cytokinins are positively acting in the Medicago truncatula and Ralstonia solanacearum pathogenic interaction

Authors

  • Sandra Moreau,

    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet-Tolosan, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Justine Fromentin,

    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet-Tolosan, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Fabienne Vailleau,

    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet-Tolosan, France
    3. Université de Toulouse, INP, ENSAT, Castanet Tolosan, France
    Search for more papers by this author
  • Tatiana Vernié,

    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet-Tolosan, France
    Search for more papers by this author
  • Stéphanie Huguet,

    1. Unité de Recherche en Génomique Végétale (URGV), INRA, UMR 1165, Université d'Evry Val d'Essonne, ERL CNRS 8196, Evry Cedex, France
    Search for more papers by this author
  • Sandrine Balzergue,

    1. Unité de Recherche en Génomique Végétale (URGV), INRA, UMR 1165, Université d'Evry Val d'Essonne, ERL CNRS 8196, Evry Cedex, France
    Search for more papers by this author
  • Florian Frugier,

    1. Institut des Sciences du Végétal (ISV), Centre National de la Recherche Scientifique (CNRS), Gif-sur-Yvette, France
    Search for more papers by this author
  • Pascal Gamas,

    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet-Tolosan, France
    Search for more papers by this author
  • Marie-Françoise Jardinaud

    Corresponding author
    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet-Tolosan, France
    3. Université de Toulouse, INP, ENSAT, Castanet Tolosan, France
    Search for more papers by this author

Summary

  • A plant–microbe dual biological system was set up involving the model legume Medicago truncatula and two bacteria, the soil-borne root pathogen Ralstonia solanacearum and the beneficial symbiont Sinorhizobium meliloti.
  • Comparison of transcriptomes under symbiotic and pathogenic conditions highlighted the transcription factor MtEFD (Ethylene response Factor required for nodule Differentiation) as being upregulated in both interactions, together with a set of cytokinin-related transcripts involved in metabolism, signaling and response. MtRR4 (Response Regulator), a cytokinin primary response gene negatively regulating cytokinin signaling and known as a target of MtEFD in nodulation processes, was retrieved in this set of transcripts.
  • Refined studies of MtEFD and MtRR4 expression during M. truncatula and R. solanacearum interaction indicated differential kinetics of induction and requirement of central regulators of bacterial pathogenicity, HrpG and HrpB. Similar to MtRR4, MtEFD upregulation during the pathogenic interaction was dependent on cytokinin perception mediated by the MtCRE1 (Cytokinin REsponse 1) receptor.
  • The use of M. truncatula efd-1 and cre1-1 mutants evidenced MtEFD and cytokinin perception as positive factors for bacterial wilt development. These factors therefore play an important role in both root nodulation and root disease development.

Introduction

Plants in their natural habitat are exposed to a number of microorganisms, only a few of which are pathogens or beneficial symbionts. Plant–pathogen interactions lead to the activation of signaling cascades, triggering defense mechanisms and the production of antimicrobial effectors that help the organisms ward off microbial attacks (Jones & Dangl, 2006). Symbiotic interactions are beneficial to both partners and an agriculturally and ecologically important interaction is the endosymbiosis between legumes (Fabaceae) and nitrogen-fixing bacteria (Rhizobiaceae). Rhizobia–legume beneficial symbiosis (referred to hereafter as symbiosis) begins with a molecular dialogue involving plant exudates and bacterial-specific lipochitooligosaccharides, the Nod factors. The perception and recognition of Nod factors by the plant leads to a sequence of highly regulated and coordinated events that ultimately result in the formation of a new root organ, the nodule (Oldroyd et al., 2011). In the nodule, bacteria are released and fix dinitrogen, which is assimilated by the host plant.

The first report of the activation of defense-related reactions in rhizobia–legume symbiosis was related to inefficient symbiotic interactions with Sinorhizobium meliloti (Niehaus et al., 1993). A hypersensitive reaction (HR), similar to that described in incompatible plant–pathogen interactions, was then reported to occur during abortive infection of alfalfa with the S. meliloti wild-type strain (Vasse et al., 1993). It is now generally believed that transient activation of defense-related mechanisms during the symbiotic interaction is necessary for a successful outcome of the symbiotic process. It is assumed that whilst symbionts suppress or avoid host defense responses in order to successfully colonize the roots, the plants needs to precisely tune certain defense-related mechanisms in order to control the number and localization of bacterial infections that are allowed to progress (Baron & Zambryski, 1995; Mithöfer, 2002). Many reports now describe the regulation of defense-related transcripts during rhizobia infection and nodule organogenesis. MtN1 and MtN13, structurally related to defense proteins, are both strongly expressed in Medicago truncatula during the infection process and in the nodule outer cortex, respectively (Gamas et al., 1998). Reactive oxygen species causing the HR in pathogenic interactions are produced from 24 to 48 h post-inoculation (hpi) during rhizobial infection, and seem to be involved in the relocalization of the NPR1 (Non-expresser of Pathogenesis-Related genes 1) protein within the cells, regulating its activity on Pathogenesis-Related (PR) gene expression (Peleg-Grossman et al., 2012). Global transcriptome analyses have confirmed that many defense-related genes are regulated in the rhizobia–legume symbiosis. During the early stages of the symbiotic interaction between Lotus japonicus and Mesorhizobium loti, genes encoding enzymes involved in phytoalexin biosynthesis, proteins involved in cell wall modification, PR proteins and HyperSensitivity-Related protein (HSR) are upregulated (Kouchi et al., 2004). Similarly, during the initial stages of the M. truncatulaS. meliloti interaction, putative defense genes are transiently activated (Lohar et al., 2006). Expression profiling in M. truncatula nodules has identified a large collection of downregulated genes coding for pathogenesis- and defense-related proteins (El Yahyaoui et al., 2004; Maunoury et al., 2010; Moreau et al., 2011; Boscari et al., 2013).

Most studies on the role of phytohormones in plant–pathogen interactions have focused on salicylic acid, jasmonic acid and ethylene (for a review, see Robert-Seilaniantz et al., 2011). Ethylene also plays an important role in nodulation by controlling the responses to Nod factors (Oldroyd et al., 2001), as well as the location and number of infections and, consequently, nodules (Heidstra et al., 1997; Penmetsa & Cook, 1997; Penmetsa et al., 2008). Jasmonic acid, like ethylene, is a negative regulator of rhizobia infections: it suppresses the expression of plant genes associated with the earliest steps of infection and regulates Nod factor-induced calcium spiking in infected root hairs (Sun et al., 2006). Salicylic acid also negatively affects the number of S. meliloti infections in M. truncatula roots (Stacey et al., 2006; Peleg-Grossman et al., 2009). Cytokinins (CKs) play a central positive role in the control of nodulation in both L. japonicus and M. truncatula (Frugier et al., 2008). CK signaling relies on a multi-step phospho-relay, consisting of Hybrid sensor Kinase receptors (HKs), Histidine Phosphotransfer proteins (HPs) and, ultimately, type-A and type-B Response Regulators (RRs) (Hwang et al., 2012). Type-B RRs positively control the expression of CK primary response genes, which include type-A RRs. Type-A RR transcription factors are negative regulators of CK signaling and are therefore involved in the homeostasis of the CK response. The M. truncatula CK receptor MtCRE1 (CK Response 1), one of the M. truncatula HKs, controls nodule formation (Gonzalez-Rizzo et al., 2006) from the initiation of cortical cell divisions to the regulation of nodule growth and differentiation (Plet et al., 2011). Several lines of evidence also support a possible link between CKs and pathogenicity (Choi et al., 2011). In Phaseolus vulgaris, the levels of active CKs are decreased after inoculation with the viral white clover mosaic potexvirus (Clarke et al., 1999) and, when seedlings are treated with exogenous CK, virus replication is reduced and PR gene expression is induced (Clarke et al., 1998). In wheat plants, infection with the fungal pathogen Tilletia caries leads to an increase in CK levels (Maksimov et al., 2002). More recently, a direct interaction between salicylic acid and CK signaling pathways was highlighted during the pathogenic interaction between Pseudomonas syringae and Arabidopsis, involving a type-B RR and the TGA3 salicylic acid response factor to regulate the PR1-dependent defense pathway (Choi et al., 2010). CKs are accumulated and also act synergistically with salicylic acid to activate defense gene expression in rice after infection by the blast fungus Magnaporthe oryzae (Jiang et al., 2013).

In this study, which aims to analyze the crosstalk between rhizobia–legume symbiosis and pathogenic interactions, we developed a biological system that involves a legume host plant, M. truncatula, its symbiotic partner, S. meliloti, and a pathogenic microbe, Ralstonia solanacearum, one of the best characterized bacterial pathogens that infects M. truncatula roots (Vailleau et al., 2007; Genin, 2010). Ralstonia solanacearum penetrates into the first 2–3 mm of the root tip, leading to the development of root symptoms, including root tip browning and swelling and a loss of viability of epidermal cells. Then, bacteria colonize the root stele and, in the susceptible A17 M. truncatula genotype, the xylem vessels (Turner et al., 2009), in which bacteria multiply extensively and produce large amounts of exopolysaccharides (Vasse et al., 2000). Exopolysaccharide accumulation in the vascular system and the subsequent collapse of the water flow cause wilting symptoms and, eventually, plant death (Vailleau et al., 2007).

The comparison of symbiotic (S. meliloti) and pathogenic (R. solanacearum) root transcriptomes performed in this study revealed that the ‘Ethylene response Factor required for nodule Differentiation’ gene (MtEFD) was upregulated in response to both microbes. MtEFD, encoding an APETALA2/Ethylene Responsive Factor (AP2/ERF) transcription factor, is expressed in nodule primordia and in the early infection zone of M. truncatula nodules, where it is required for the differentiation of plant cells and bacteria, and it may activate expression of the type-A RR MtRR4 (Vernié et al., 2008). During the pathogenic interaction between M. truncatula and R. solanacearum, we found that a number of CK-related genes were upregulated in response to R. solanacearum, and that both MtEFD and MtCRE1 were positive factors affecting disease development.

Materials and Methods

Biological material

The wild-type strain GMI1000 of R. solanacearum and regulatory mutant derivatives (GMI1525, hrpB and GMI1755, hrpG) have been described previously (Cunnac et al., 2004). A GMI1000 trans-zeatin synthase mutant (GRS428, tzs) was kindly provided by S. Genin (LIPM, INRA/CNRS, Castanet Tolosan, France). GMI1559 and GMI1485 are GMI1000 derivatives carrying constitutively expressed GUS (β-glucuronidase, UidA) and GAL (β-galactosidase) fusions, respectively (Etchebar et al., 1998). Conditions for the routine culture of R. solanacearum were as described by Cunnac et al. (2004). When required, antibiotics were used at the following concentrations: gentamycin, 10 μg ml−1; spectinomycin, 40 μg ml−1; tetracycline, 10 μg ml−1.

Seeds of M. truncatula genotypes A17 and F83005.5 were provided by INRA of Montpellier, SGAP Laboratory, Mauguio, France. The cre1-1 mutant (Plet et al., 2011) and the efd-1 mutant (Vernié et al., 2008) have been described previously. Seeds were surface sterilized, vernalized and germinated for 24 h using standard protocols (Boisson-Dernier et al., 2001). Germinated seeds were then grown on modified Fahraeus medium (Vailleau et al., 2007) covered with filter paper in square Petri dishes at 25°C (day : night, 12 h). Roots were kept in the dark with a mask on the Petri dish made of craft paper and aluminum foil.

Inoculation procedure

Inoculation was performed as described previously (Turner et al., 2009) with a bacterial concentration set at 107 colony-forming units (CFU) ml−1. Mock treatment was carried out using the same procedure with water.

RNA isolation

Sections (1.5 cm) of 20–30 root tips were ground in liquid nitrogen. RNA samples (three to five biological repetitions per sample) were isolated using the RNeasy Plant Mini Kit (Qiagen) with DNAse treatment following the manufacturer's procedure. Ribonucleic acid was quantified using a NanoDrop Spectrophotometer ND-100 (NanoDrop Technologies, Wilmington, DE, USA) and integrity was evaluated with a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA).

Microarray experiments, data processing, deposition and statistical analyses

Three independent biological replicates were performed for each condition, corresponding to three genotypes and three time points (0, 12 and 72 hpi). The ‘0 hpi’ sample material was collected immediately after inoculation. Total RNA was extracted using the RNeasy Kit (Qiagen). Four micrograms of total RNA per transcriptome were reverse transcribed and used to hybridize Affymetrix GeneChip® Medicago genome arrays at INRA-URGV (Evry, France), as described previously (Rey et al., 2013). Raw and normalized data are available from the Gene Expression Omnibus (GEO) repository (Barrett & Edgar, 2006) at the National Center for Biotechnology Information (NCBI) under the GSE accession number 18473 and through the CATdb database (AFFY_Ralstonia_Medicago (Gagnot et al., 2008)). Transcriptomes are also available in the M. truncatula Gene Expression Atlas (http://mtgea.noble.org/v3/).

We used ANOVA models to detect differentially expressed genes in the transcriptomes corresponding to A17 M. truncatula root tips exposed to R. solanacearum GMI1000 at 0, 12 and 72 hpi. We subjected all normalized data to an ANOVA model based on a randomized block design. ‘Biological repeat’ was considered as the blocking factor and time after inoculation as the factor of primary interest. For each gene model, the normality of the model residual error was assessed using a Shapiro and Wilk test (Royston, 1995) and homoscedasticity was verified using the Levene test (Hines & Hines, 2000). The false discovery rate (Benjamini & Hochberg, 1995) was calculated for each gene model. Gene models showing a significant effect for the factor of primary interest were subjected to Tukey's post hoc test for single-step multiple mean comparisons (Yandell, 1997).

Concerning the nodule transcriptomes (Benedito et al., 2008), normalized data were retrieved from the M. truncatula Gene Expression Atlas (http://mtgea.noble.org/v3/) and submitted to the same statistical procedure. All statistical tests were performed using R software v.2.15.1 (http://r-project.org).

Quantitative real-time reverse transcription-polymerase chain reaction (RT-PCR) analysis

Reverse transcription was performed with 2 μg of total RNA using the superscript reverse transcriptase II (Invitrogen Life Technologies) and anchored oligo(dT) or random primers for bacterial transcript quantification. Quantitative PCRs were conducted on 384-well plates using a LightCycler 480 (Roche) with the manufacturer recommended conditions and the primers shown in Supporting Information Table S1. Cycling conditions were as follows: 95°C for 5 min, 45 cycles at 95°C for 15 s and 60°C for 1 min. The ubiquitin encoding gene (MtUBIQ1) was used as an internal standard for sample comparisons, as its expression showed no significant regulation over the experimental time course in the pathogenic dataset (Mtr.48704.1.S1_x_at and Mtr.37220.1.S1_x_at; Table S2). The specificity and efficiency of the amplification were verified by analyses of melting curves and standard curves, respectively. The math formula method (Livak & Schmittgen, 2001) was used for the calculation of relative expression.

Agrobacterium rhizogenes-mediated root transformation

A construct containing 2442 bp immediately upstream of the ATG start codon of MtEFD (Vernié et al., 2008) was introduced into Agrobacterium rhizogenes ARqua1 and used for M. truncatula root transformation. The transgenic roots were obtained after kanamycin (25 mg l−1) selection for 2 wk (Boisson-Dernier et al., 2001). Composite plants were then transferred into slanted agar in square Petri dishes with an interface of paper, as described previously (Vailleau et al., 2007).

Histochemical staining and microscopic analyses

The fluorescein diacetate (FDA) viability test was performed as described previously (Heslop-Harrison & Heslop-Harrison, 1970). Fluorescence was obtained using an excitation range of 450–490 nm, a dichroic mirror of 510 nm and a long-pass emission filter of 515 nm, observed with an inverted microscope (Leica DMIRBE, Wezlar, Germany), and images were acquired with a CCD camera (color cooled view; Photonic Science, Robertsbridge, UK). The root cell division zone and the vascular bundle initiation zone were localized on root tips at 3 and 5 dpi with R. solanacearum after clearing for 3–5 min with sodium hypochlorite 3.6%. They were observed in bright field (root cell division zone) or dark field (vascular bundles) using a DMI6000B microscope equipped with a DFC300 camera (Leica Microsystems). The images were then analyzed using Fiji software (Schindelin et al., 2012). Nuclei with DNA synthesis activity were visualized in M. truncatula root tips at 3 dpi with R. solanacearum using a Click-iT® EdU (5-Ethynyl-2′-deoxyUridine) cell proliferation assay (Alexafluor 647; Life Technologies, Carlsbad, CA, USA). Roots were incubated overnight in 10 μM EdU in liquid Farhaeus at 25°C. Samples were then fixed and treated according to the manufacturer's instructions, and observed with a Leica confocal SP2 AOTF (excitation, 633 nm; emission, 670 nm). Images were acquired and projected using Leica confocal software. Histochemical staining for GUS and GAL activity in plants or bacteria was performed as described previously (Vernié et al., 2008) with samples incubated for 8 h at 37 or 28°C. After staining, samples were cleared for 3–5 min with sodium hypochlorite 3.6%, and observed in a bright field using a DMI6000B microscope equipped with a DFC300 camera (Leica Microsystems).

In planta bacterial growth measurements

Leaves of inoculated plantlets were harvested and sterilized with 70% ethanol for 1 min, rinsed three times in sterile water, patted dry, weighed, crushed and taken up in water. Bacterial concentrations were determined by dilution plating on selective culture SMSA (Soil Isolation Medium) medium (Vailleau et al., 2007).

Results

MtEFD, an M. truncatula transcription factor involved in nodule differentiation, is also upregulated during pathogenic interaction depending on R. solanacearum pathogenicity

Susceptible A17 M. truncatula plants were inoculated with the R. solanacearum wild-type GMI1000 strain. Root tips (15 mm) were harvested immediately (0 hpi) and at 12 and 72 hpi. These last two time points were selected as they corresponded to the first root symptoms (epidermis cell death and root growth arrest) and vascular bundle colonization, respectively (Turner et al., 2009). Whole-genome gene expression of the samples was analyzed using the Affymetrix GeneChip® Medicago genome arrays.

A stringent ANOVA, considering the repetitions as blocks and the time after inoculation as the single variable, was performed. A set of 6049 M. truncatula probes showing a significant effect of the ‘time after inoculation’ parameter ( 0.01) was retained (Table S2). These probe sets were then submitted to a single-step multiple comparison procedure (Tukey test), and probes showing an adjusted  0.01 and an absolute log-ratio value ≥ 1 were selected. Of the 6049 initial probe set, 3229 probes were then considered as significantly regulated at 12 hpi and/or 72 hpi with R. solanacearum.

We then compared these root transcriptomes obtained in response to a pathogen with those resulting from the inoculation of A17 M. truncatula roots with the symbiotic S. meliloti GMI2011 strain (Benedito et al., 2008). The transcriptomes of non-inoculated roots, immature nodules (4 dpi) and mature nodules (10 dpi), two samples that strongly concentrate symbiotic transcripts, were analyzed using the same statistical analysis and thresholds. A set of 8176 M. truncatula probes was identified as differentially regulated at 4 dpi and/or 10 dpi relative to non-inoculated roots (Table S3).

The combination of all of these analyses highlighted 964 M. truncatula probes differentially regulated in both pathogenic and symbiotic interactions. Whereas 333 probes had an opposite regulation depending on the nature of the biotic interaction, 631 probes had a similar transcriptional response to both bacteria, with 500 downregulated and 131 upregulated probes (Table S4). Of the 131 latter probes, only 97 had biological or molecular function annotation according to Gene Ontology (GO) (http://www.geneontology.org/). However, no functional class was detected as significantly over-represented based on a Mann–Whitney test (not shown). Five annotated transcriptional regulators (GO:0006355) were identified: three shared homology to RNA polymerases and two were AP2/EREBP transcription factors (Mtr.43606.1.S1_at and Mtr.41581.1.S1_at). Medicago truncatula expression profiling data (http://mtgea.noble.org) showed that Mtr.43606.1.S1_at had a non-specific root expression. The Mtr.41581.1.S1_at probe, interestingly, corresponded to the MtEFD transcription factor previously described as only regulated during the M. truncatula–S. meliloti interaction (Vernié et al., 2008).

Although MtEFD expression was 13-fold higher (Tukey test, adjusted P = 2.9 × 10−4) at 72 hpi in A17 root tips challenged with R. solanacearum GMI1000, no significant difference was detected at 12 hpi (Fig. 1a). We took advantage of existing transcriptomes that we had generated (available at the GEO repository GSE 18473) to search for MtEFD expression in the resistant F83005.5 M. truncatula genotype (Vailleau et al., 2007). A 14-fold induction (Tukey test, adjusted P = 8.5 × 10−4) was similarly detected at 72 hpi (Fig. 1a), as well as no upregulation at 12 hpi. Quantitative RT-PCR experiments were performed in order to validate the Affymetrix data on both genotypes (Fig. 1b). A five time points, kinetics showed that the expression of MtEFD was at a maximum at 72 hpi, being 27-fold (Tukey test, adjusted P = 1.0 × 10−2) and 13-fold (Tukey test, adjusted P = 4.4 × 10−3) higher in A17 and F83005.5, respectively. However, no significant difference (Tukey test, adjusted P = 1.9 × 10−1) could be found between the two genotypes, indicating that transcriptional regulation of MtEFD expression is not linked to F83005.5 resistance or to the A17 susceptibility genetic determinism.

Figure 1.

MtEFD expression during the interaction between Medicago truncatula and Ralstonia solanacearum. (a) MtEFD expression levels corresponding to Affymetrix normalized data were extracted from our transcriptomic data available at the Gene Expression Omnibus (GEO) repository (GSE18473). A17 (closed circles) and F83005.5 (open circles) M. truncatula plants were inoculated with R. solanacearum GMI1000 (107 CFU ml−1) and root tips (15 mm from the apex) were harvested at 12 and 72 h post-inoculation (hpi). (b, d) Relative transcript levels of MtEFD were determined by quantitative reverse transcription-polymerase chain reaction, and results were normalized using the UBIQ1 reference gene. (b) Root tips were harvested from M. truncatula A17 (closed circles) and F83005.5 (open circles) genotypes inoculated with R. solanacearum GMI1000 (107 CFU ml−1) after the indicated times (hpi). (d) Root tips were harvested from M. truncatula A17 challenged with the R. solanacearum wild-type strain GMI1000 (107 CFU ml−1; closed circles), hrpB (107 CFU ml−1; open triangles) and hrpG (107 CFU ml−1; open squares). (a, b, d) The ‘0 hpi’ sample represents root tips collected immediately after inoculation. Three biological repetitions were analyzed, with 30 plants per biological repeat. Error bars, ± SE of the mean. Significant mean differences were detected using ANOVA followed by Tukey's post hoc test for single-step multiple mean comparisons. Means with different letters are significantly different. (a) Adjusted  0.01. (b, d) Adjusted  0.05. (c) Root tips of A17 composite plants expressing the β-glucuronidase gene (UidA/GUS) under the control of the MtEFD promoter region (2442 kb upstream of the predicted start codon). Transgenic roots were inoculated with water (mock control), R. solanacearum GMI1000 or GMI1000-β-galactosidase gene (107 CFU ml−1 each). The expression of β-glucuronidase (blue coloration) and β-galactosidase (magenta coloration) was analyzed at 3 dpi, and images are representative of the overall observations on n > 54 independent composite plants. Bars, 100 μm.

To better understand how MtEFD expression was spatially regulated in A17 M. truncatula root tips on R. solanacearum inoculation, composite plants expressing the pMtEFD-GUS reporter were generated (Vernié et al., 2008). MtEFD expression was localized in water-treated control root tips within two regions: in the root stem cell niche zone and at the meristematic transition zone between cell division and cell elongation (Fig. 1c). In R. solanacearum GMI1000-inoculated root tips (3 dpi), this specific expression pattern could not be observed and MtEFD (in blue) was instead expressed in all zones of the root tip, as well as in more mature regions. The inoculation of root tips with a lacZ-expressing GMI1000 R. solanacearum strain showed that MtEFD (in blue) was induced systemically, as expression did not co-localize with the bacteria (in magenta; Fig. 1c).

The bacterial Hrp (Hypersensitive response and pathogenicity) system is essential for the pathogenicity of R. solanacearum on host plants (Boucher et al., 1987). HrpG and HrpB are required for the root infection process: HrpG integrates environmental signals and, in the presence of plant cells, controls the type III secretion system via HrpB, as well as many other functions required for disease development (Genin & Denny, 2012). Medicago truncatula A17 root tips were comparatively challenged with the R. solanacearum GMI1000 wild-type strain or with GMI1000 hrpG and hrpB mutants lacking one or other of these pathogenicity determinants, and which are therefore unable to induce disease. No induction of MtEFD could be detected at 72 hpi in response to either of the mutant strains (Fig. 1d, Tukey test, adjusted > 5.0 × 10−2). This indicates that the regulation of MtEFD relies on the R. solanacearum type III secretion system.

MtEFD is a positive regulator of disease development during the susceptible interaction

An efd-1 mutant and its wild-type sibling (SWT; Vernié et al., 2008) were challenged with R. solanacearum GMI1000. Various root symptoms were comparatively followed in each genotype. Primary root growth arrest between 24 and 48 hpi (Fig. S1a), epidermal cell death evaluated in root tips at 3 dpi using an FDA viability test (Fig. 2a) as well as root tip browning and swelling at 5 dpi (Fig. 2b) were similarly observed in both genotypes. In response to colonization, no morphological difference with regard to the size of the root cell division zone (Fig. S1b), localization of the vascular bundle initiation zone (Fig. S1c) or number of dividing nuclei in the root tip (Fig. S1d) could be found between the control and efd-1. However, at 14 dpi in an in vitro assay, efd-1 plants showed delayed disease development relative to SWT (Fig. 3a), illustrated by greener aerial parts, branched primary roots and grown lateral roots.

Figure 2.

Comparison of root symptom development between the efd-1 mutant and its wild-type sibling (SWT) challenged with Ralstonia solanacearum. (a) efd-1 and SWT plantlets were water inoculated or inoculated with R. solanacearum GMI1000 (107 cfu ml−1) in vitro. The viability of the root tip epidermis was assayed 3 d post-inoculation (dpi) using a fluorescein diacetate (FDA) viability test. Images are representative of the overall observations on four biological repeats, with 25 plants per biological repeat. Bars, 500 μm. (b) efd-1 and SWT plants inoculated with R. solanacearum GMI1000 (107 CFU ml−1) expressing constitutively the β-glucuronidase gene (UidA/GUS; strain GMI1559). UidA expression (in blue) was analyzed at 5 dpi, and images are representative of the overall observations on four biological repeats, with 25 plants per biological repeat. Bars, 80 μm.

Figure 3.

Comparison of disease development between the efd-1 mutant and its wild-type sibling (SWT) challenged with Ralstonia solanacearum. (a) efd-1 and SWT plants inoculated with R. solanacearum GMI1000 (107 CFU ml−1). Plants were observed at 14 d post-inoculation (dpi), and images are representative of the overall observations on four biological repeats, with 30 plants per biological repeat. Bars, 1 cm. (b) efd-1 and SWT plants were inoculated with R. solanacearum GMI1000 (107 CFU ml−1) and scored at 21 dpi. A disease index was generated on Medicago truncatula A17. Score 0 corresponded to plantlets with no symptoms, score 1 to plants showing yellowing on one or two cotyledons, score 2 to plants with cotyledon wilting, score 3 to plants with first leaf wilting and score 4 to plant death. The percentage of plants showing a score ≥ 1 according to the disease index was calculated. Four biological repeats were performed, with 30 plants per biological repeat. Error bars, ± SE of the mean. Significant mean difference were detected using Student's t-test (*, < 0.05). (c) efd-1 and SWT plants were inoculated with R. solanacearum GMI1000 (107 CFU ml−1). The first two trifoliated leaves of 10 plants were collected at 21 dpi. The numbers of bacteria per gram of fresh leaf weight (FW) were then evaluated based on the number of ‘colony-forming units’ (CFU). Four biological repeats with three technical repeats each were performed, with 30 plants per biological repeat. Significant mean difference was detected using Student's t-test (**, < 0.01).

Disease symptoms on cotyledons and leaves were scored at 21 dpi according to a disease index generated on M. truncatula A17. Score 0 corresponded to plantlets with no symptoms, score 1 to plants showing yellowing on one or two cotyledons, score 2 to plants with cotyledon wilting, score 3 to plants with first leaf wilting and score 4 to plant death. Plantlets showing oozes (bacterial exudates) on hypocotyl or root were automatically scored as 4. The proportions of plantlets having a score ≥ 1 were significantly different (χ2 test, P = 2.0 × 10−2) between efd-1 and SWT genotypes, with 43% and 64% of wilting plants observed, respectively (Fig. 3b). The number of R. solanacearum bacteria in the leaves of the two genotypes was then assayed, and a significant 15-fold (t-test, P = 4.1 × 10−3) higher concentration of bacteria was detected in SWT relative to efd-1 (Fig. 3c).

In a reciprocal experiment, composite efd-1 and A17 plants were generated with a construct bearing the MtEFD coding sequence under the control of the 35S promoter (efd-1:MtEFDox and A17:MtEFDox). As controls, efd-1 and A17 composite plants were generated with a construct bearing no coding sequence (efd-1:control and A17:control). MtEFD expression, quantified in 3-wk-old transgenic roots, was significantly higher (8.5-fold, Tukey test, adjusted P = 1.9 × 10−2) in A17:control relative to efd-1:control, as expected. Significant 254-fold (Tukey test, adjusted P = 3.1 × 10−5) and 32-fold (Tukey test, adjusted P = 8.7 × 10−4) MtEFD overexpression levels were detected in efd-1:MtEFDox and A17:MtEFDox roots, respectively (Fig. 4a). The composite plants efd-1:MtEFDox, A17:MtEFDox and their respective controls were then challenged with the R. solanacearum GMI1000 strain expressing constitutively the GUS reporter gene (Cunnac et al., 2004). Because of the 3 wk in vitro culture, composite plants have yellowish aerial parts, and cotyledon or leaf symptoms could not be scored. Colonization of transgenic roots was followed quantitatively using real-time PCR of the UidA/GUS bacterial reporter. When compared with their respective controls, significant 11.5-fold and 14.5-fold GUS overexpression was detected (Tukey test, adjusted P = 4.5 × 10−6 and adjusted P = 1.2 × 10−6) at 5 dpi in efd-1:MtEFDox and A17:MtEFDox roots, respectively (Fig. 4b). These results indicate that MtEFD overexpression increased quantitatively root bacterial colonization in A17 and the efd-1 mutant.

Figure 4.

Effect of MtEFD overexpression in A17 and efd-1 mutant on bacterial colonization. (a) efd-1 and A17 Medicago truncatula composite plants were generated to express the MtEFD coding sequence under the control of the 35S promoter (efd-1:MtEFDox and A17:MtEFDox). A17:control and efd-1:control correspond to A17 and efd-1 plants transformed with a construct in which no coding sequence was included. Relative transcript levels of MtEFD were determined by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) using non-inoculated roots. (b) efd-1 and A17 composite plants overexpressing MtEFD were inoculated with a Ralstonia solanacearum GMI1000 (107 CFU ml−1) derived strain expressing constitutively the β-glucuronidase gene (UidA/GUS; strain GMI1559). Relative UidA transcript levels were determined by qRT-PCR in root tips harvested from composite plants at 5 d post-inoculation. (a, b) Results were normalized using the UBIQ1 reference gene. Four biological repetitions were analyzed, with 30 plants per biological repeat. Significant mean differences were detected using ANOVA followed by Tukey's post hoc test for single-step multiple mean comparisons. Means with different letters are significantly different (adjusted  0.05).

Medicago truncatula CK metabolism and signaling genes are transcriptionally regulated in response to R. solanacearum

Because the MtEFD transcription factor regulates the expression of the MtRR4 CK primary response gene during the symbiotic interaction (Vernié et al., 2008), expression of MtRR4 was followed during the pathogenic interaction. MtRR4 (Mtr.9656.1.S1_at) was not part of the list of M. truncatula probes ultimately retained in the global transcriptomic analysis (Table S2) because of an adjusted P value (7.0 × 10−2) over the threshold (1.0 × 10−2). However, the transient upregulation of MtRR4 at 12 hpi vs 0 hpi seemed to be reproducible, independent of biological repetition variability (Fig. 5a), with a 5.8-fold induction for A17 M. truncatula (Tukey test, adjusted P = 7.0 × 10−2) and a seven-fold induction for F83005.5 M. truncatula (Tukey test, adjusted P = 3.0 × 10−2). Validation by quantitative PCR experiments at 12 hpi (Fig. 5b) indeed revealed 11.3-fold induction (Tukey test, adjusted P = 1.6 × 10−6) and 7.5-fold induction (Tukey test, adjusted P = 8.8 × 10−7) for A17 and F83005.5 M. truncatula genotypes, respectively. These inductions were not sustained at 72 hpi, confirming the transient induction observed previously (Fig. 5a). Although, during the nitrogen-fixing symbiotic interaction, MtEFD regulates MtRR4, no such relationship could be demonstrated during the pathogenic interaction: indeed, the induction of MtRR4 (12 hpi) preceded MtEFD induction (72 hpi). Accordingly, the efd-1 mutant showed a 6.9-fold induction of MtRR4 expression at 12 hpi (Tukey test, adjusted P = 2.8 × 10−10; Fig. 5b). A17 M. truncatula plantlets were additionally comparatively challenged with the R. solanacearum GMI1000 wild-type strain and the two mutant strains used previously, hrpB and hrpG. No induction of MtRR4 could be detected at 12 hpi in response to the hrpG mutant, but a 9.5-fold induction could still be detected with the hrpB mutant (Tukey test, adjusted P = 1.5 × 10−8; Fig. 5c). This indicates that the MtRR4 transient upregulation at 12 hpi does not rely on the type III secretion system of R. solanacearum.

Figure 5.

MtRR4 expression during the interaction between Medicago truncatula and Ralstonia solanacearum. (a) MtRR4 expression levels corresponding to Affymetrix normalized data were extracted from our transcriptomic data generated in this study and available at the Gene Expression Omnibus (GEO) repository (GSE18473). A17 (closed circles) and F83005.5 (open circles) M. truncatula plants were inoculated with R. solanacearum GMI1000 (107 CFU ml−1), and root tips (15 mm from the apex) were harvested at 12 and 72 h post-inoculation (hpi). (b, c) Relative transcript levels of MtRR4 were determined by quantitative reverse transcription-polymerase chain reaction, and the results were normalized using the UBIQ1 reference gene. (b) Root tips (15 mm from the apex) were harvested from M. truncatula A17 (closed circles), F83005.5 lines (open circles) and efd-1 mutant (open diamonds) inoculated with R. solanacearum GMI1000 (107 CFU ml−1) at 12 and 72 hpi. (c) Root tips were harvested from M. truncatula A17 and challenged with R. solanacearum wild-type strain GMI1000 (107 CFU ml−1; closed circles), hrpB (107 CFU ml−1; GMI1525; open triangles) and hrpG (107 CFU ml−1; GMI1755; open squares). (a, b, c) The ‘0 hpi’ sample represents root tips collected immediately after inoculation. Three biological repetitions were analyzed, with 30 plants per biological repeat; error bars, ± SE of the mean. Significant mean differences were detected using ANOVA followed by Tukey's post hoc test for single-step multiple mean comparisons. Means with different letters are significantly different, with: (a) adjusted  0.01; (b, c) adjusted  0.05.

Based on the initial observation that the CK primary response gene MtRR4 was upregulated in response to R. solanacearum, a survey of other M. truncatula probes corresponding to transcripts involved in CK metabolism, signaling or response, and differentially regulated on the basis of ANOVA of pathogenic transcriptomes (Table S2), was conducted (Table 1). Concerning CK metabolism, three CK oxidases/dehydrogenases (CKX) catalyzing the degradation of CK and a CK hydrolase involved in CK conjugation were upregulated at 72 hpi, whereas a LOG-like transcript allowing the conversion of CK ribonucleotides into active CKs (Kuroha et al., 2009) was downregulated at 72 hpi. However, genes encoding rate-limiting enzymes of CK biosynthesis, the isopentenyltransferase (IPT) and cytochrome P450 monooxygenase, were not differentially regulated at 12 or 72 hpi. Concerning CK signaling, no sensor kinase receptors or histidine phosphotransfer proteins were significantly regulated during the pathogenic interaction. Nevertheless, several other members of the type-A RR family were detected as upregulated. Three groups could be identified on the basis of their temporal expression pattern: two genes transiently induced at 12 hpi, four early-induced genes at 12 hpi with an induction sustained to 72 hpi, and two late-induced genes at 72 hpi. Two members of the type-B RR transcription factor family, acting as positive regulators of CK signaling by controlling the transcription of a subset of ‘CK primary response genes’ (Hwang et al., 2012), were also upregulated, one at 12 hpi and the other at 72 hpi (Table 1). Among the 39 genes proposed by Ariel et al. (2012) as candidate CK primary response genes in M. truncatula root apices, 11 showed a differential expression in response to R. solanacearum, consisting mainly of upregulations (nine of 12 genes). In addition, two members of the CRF (Cytokinin Responsive Factor; Rashotte & Goertzen, 2010) family were upregulated at 72 hpi. As some expansin genes have been shown to be CK upregulated during parasitic interactions (O'Malley & Lynn, 2000), we specifically analyzed this gene family and found that five M. truncatula expansin transcripts were upregulated at 72 hpi with R. solanacearum. Similarly, ACC synthase encoding transcripts have been described as being repressed by CKs (Bhargava et al., 2013), and we observed a downregulation at 72 hpi. Finally, GH3 genes have been reported to be induced by CKs (Bhargava et al., 2013), and a 27.5-fold induction of a GH3 gene was detected at 72 hpi in our dataset. The majority of these genes were retained in the overall comparative pathogenesis/symbiosis analysis (Table S4), but not in the limited set of similarly upregulated genes in both pathogenic and symbiotic conditions, indicating that the regulation of these genes is dependent on the type of biotic interaction.

Table 1. Differential expression of cytokinin-related transcripts during the A17 Medicago truncatula and GMI1000 Ralstonia solanacearum pathogenic interaction
Affymetrix identifierAnnotationReferenceA17 M. truncatula–GMI1000 R. solanacearum interaction
Fold 12 vs 0 hpiAdj. P value 12 vs 0 hpiFold 72 vs 0 hpiAdj. P value 72 vs 0 hpi
  1. Probe sets corresponding to cytokinin metabolism, signaling or candidate primary response genes were extracted from Supporting Information Table S2. Yellow cells, fold ratio ≥ 2 and conditions significantly different (adjusted  0.05); blue cells, fold ratio ≤ −2 and conditions significantly different (adjusted  0.05). Annotations and gene names were extracted from the references cited in the table. –, Gene Ontology annotation; *, transcript for which quantitative reverse transcription-polymerase chain reactions (qRT-PCRs) in root tips (15 mm from the root apex) were conducted and which validated the Affymetrix data. hpi, hours post-inoculation.

Cytokinin metabolism
Mtr.24597.1.S1_atCytokinin oxidase10.9882.80.037
Mtr.14413.1.S1_atCytokinin oxidase *2.90.25260.009
Mtr.38799.1.S1_s_atCytokinin oxidase20.2435.70.02
Mtr.29566.1.S1_atCytokinin hydrolase−1.10.7655.40.001
Mtr.50458.1.S1_atLOG-like *−2.30.061−6.20.004
Cytokinin signaling
Positive response regulators
Mtr.44060.1.S1_atB-type RR3.60.0420.80.761
Mtr.14613.1.S1_atB-type ARR6-like−1.20.0832.10.001
Negative response regulators
Mtr.43256.1.S1_atA-type RR2.30.0071.10.7
Mtr.9656.1.S1_atA-type MtRR4*Plet et al. (2011)5.20.0731.80.573
Mtr.2193.1.S1_atA-type RR40.0013.90.001
Mtr.5335.1.S1_atA-type RR12.80.00312.40.003
Mtr.174.1.S1_atA-type MtRR11*Den Camp et al. (2011)12.60.0024.60.014
Mtr.51829.1.S1_s_atA-type ARR8-like*15.40.00328.80.001
Mtr.31738.1.S1_atA-type MtRR8Saur et al. (2011)1.80.1668.30.002
Cytokinin response genes
Candidate M. truncatula cytokinin primary response genes
Mtr.12473.1.S1_atUDP-glucuronosyl UDP-glucosyltransferaseAriel et al. (2012)2.10.02910.998
Mtr.9007.1.S1_atUDP-glucuronosyl UDP-glucosyltransferaseAriel et al. (2012)4.80.047−1.10.987
Mtr.44630.1.S1_atUDP-glucuronosyl UDP-glucosyltransferaseAriel et al. (2012)40.0372.90.077
Mtr.10628.1.S1_atCytochrome P450Ariel et al. (2012)1.40.443380
Mtr.13370.1.S1_at2OG-Fe(II) oxygenaseAriel et al. (2012)40.0969.40.021
Mtr.24815.1.S1_atGlutaredoxin-like plant IIAriel et al. (2012)2.20.3029.50.018
Mtr.38111.1.S1_atGlutathione S-transferase GST 14Ariel et al. (2012)2.80.40354.50.011
Mtr.33011.1.S1_atCtr copper transporterAriel et al. (2012)−20.0095.70
Mtr.39997.1.S1_atHypothetical proteinAriel et al. (2012)1.60.1432.50.018
Mtr.4878.1.S1_atGlyoxalase bleomycin resistance protein/dioxygenaseAriel et al. (2012)−1.20.667−3.20.017
Mtr.34024.1.S1_atPathogenesis-related transcriptional factor and ERF transcriptional factor B3Ariel et al. (2012)−2.10.115−6.20.006
Other cytokinin response genes
Mtr.43606.1.S1_atCRFRashotte & Goertzen (2010)2.40.2166.70.003
Mtr.45912.1.S1_atCRFRashotte & Goertzen (2010)1.80.3557.90.033
Mtr.41561.1.S1_atExpansin 4*O'Malley & Lynn (2000)1.80.1737.40.003
Mtr.19007.1.S1_atExpansinO'Malley & Lynn (2000)4.70.04350.037
Mtr.36209.1.S1_atExpansin B3O'Malley & Lynn (2000)1.70.6147.60.035
Mtr.8532.1.S1_s_atExpansin A1O'Malley & Lynn (2000)6.80.05812.90.023
Mtr.33549.1.S1_atExpansinO'Malley & Lynn (2000)−1174.50.005
Mtr.38111.1.S1_atGSTBhargava et al. (2013)2.80.40354.50.011
Mtr.40263.1.S1_atGH3.1*Bhargava et al. (2013)1.10.98927.50.006
Mtr.9002.1.S1_atACC oxidaseBhargava et al. (2013)−1.20.737−2.50.017

MtEFD induction during the pathogenic interaction between M. truncatula and R. solanacearum is MtCRE1 dependent

In order to evaluate whether CKs had a positive or negative role on M. truncatula infection by R. solanacearum GMI1000, a mutant in the MtCRE1 CK receptor essential for symbiotic nodulation (Plet et al., 2011) was challenged by the pathogenic bacteria. The cre1-1 mutant showed significantly delayed symptoms at 21 and 28 dpi when compared with the wild-type (Tukey test, adjusted P < 2.1 × 10−3). The cre1-1 response was thus similar (Tukey test, adjusted P > 8.9 × 10−1) to that of the efd-1 mutant (Fig. 6). These results indicate that, like MtEFD, CK signaling mediated by the MtCRE1 receptor is a positive regulator of R. solanacearum disease development. Ralstonia solanacearum GMI1000 has a trans-zeatin synthase (tzs) gene which encodes an isopentenyl transferase protein involved in the CK biosynthesis first committed step (Salanoubat et al., 2002). In order to check whether the CKs involved in disease development were of bacterial origin, wild-type roots were inoculated with a R. solanacearum GMI1000 tzs mutant strain. Symptom appearance was similar to that observed after inoculation with the wild-type GMI1000 strain (Fig. 6). Therefore, CK of bacterial origin could not be involved in disease development in our experimental conditions.

Figure 6.

Role of the CRE1 cytokinin signaling pathway and Ralstonia solanacearum tzs (trans-zeatin synthase) gene on disease symptom development. Medicago truncatula A17 (closed circles), efd-1 mutant (closed squares) and cre1-1 mutant (closed triangles) plants were inoculated with R. solanacearum GMI1000 (107 CFU ml−1). Medicago truncatula A17 plants (open triangles) were also inoculated with R. solanacearum tzs mutant strain (107 CFU ml−1). Plants were scored at 14, 21 and 28 d post-inoculation (dpi). A disease index was generated on M. truncatula A17. Score 0 corresponded to plantlets with no symptoms, score 1 to plants showing yellowing on one or two cotyledons, score 2 to plants with cotyledon wilting, score 3 to plants with first leaf wilting and score 4 to plant death. The percentage of plants showing a score ≥ 1 according to the disease index was calculated. Three biological repeats were performed, with 30 plants per biological repeat; error bars, ± SE of the mean. Significant mean differences were detected using ANOVA followed by Tukey's post hoc test for single-step multiple mean comparisons. Means with different letters are significantly different (adjusted  0.05).

To test the possible relationships between MtCRE1-dependent CK signaling and the MtEFD transcription factor during the pathogenic interaction, cre1-1 mutant root tips challenged with R. solanacearum GMI1000 were sampled just after inoculation (0 hpi) and at 12 and 72 hpi. MtRR4 and MtGH3.1, used as CK-related early and late pathogenic response genes, respectively (Fig. 7a,b), were no longer detected in the cre1-1 mutant in response to R. solanacearum (Tukey test, adjusted P > 5 × 10−2), but were detected at the same levels in the efd-1 mutant and the wild-type. These results indicate that the regulation of these two genes during the pathogenic interaction is dependent on CK perception by the MtCRE1 receptor. Interestingly, the induction of MtEFD expression by R. solanacearum was similarly abolished in the cre1-1 mutant (Fig. 7c).

Figure 7.

Expression pattern of three transcripts responding to Ralstonia solanacearum GMI1000 in cre1-1 and efd-1 Medicago truncatula mutants. Relative transcript levels of MtRR4 (a), MtGH3.1 (b) and MtEFD (c) were determined by quantitative real-time reverse transcription-polymerase chain reaction (qRT-PCR) in root tips (15 mm from the root apex) harvested from M. truncatula A17 (closed circles), efd-1 mutant (closed squares) and cre1-1 mutant (closed triangles) plants at 12 or 72 h post-inoculation (hpi) with R. solanacearum GMI1000 (107 CFU ml−1). Results were normalized using the UBIQ1 reference gene. Three to five biological repetitions, depending on the genotypes, were analyzed, with 30 plants per biological repeat. Error bars, ± SE of the mean. Significant mean differences were detected using ANOVA followed by Tukey's post hoc test for single-step multiple mean comparisons. Means with different letters are significantly different (adjusted  0.05).

Overall, transcriptional regulation of selected CK-related transcripts therefore seems to be independent of MtEFD during the interaction between M. truncatula and R. solanacearum GMI1000, whereas MtEFD induction is dependent on MtCRE1 in the same biological conditions.

Discussion

The characterization of the interplay between responses to infection with beneficial and detrimental microorganisms requires biological experimental systems in which plant roots can be colonized by symbiotic and pathogenic microbes (Rey & Schornack, 2013). The legume species M. truncatula and L. japonicus are model plants for endosymbioses, and therefore excellent plant candidates for such systems. Dual symbiotic vs pathogenic systems have been set using various types of pathogen, such as oomycetes (Aphanomyces euteiches, Rey et al., 2013), fungi (Colletotrichum trifolii, Genre et al., 2009; Rhizoctonia solani, Anderson et al., 2010; Verticillium spp., Ben et al., 2013) or nematodes (Meloidogyme spp., Damiani et al., 2012), but none has been described with a bacterial pathogenic partner. The experimental system developed in this study between M. truncatula and R. solanacearum or S. meliloti presents several advantages. The first is that R. solanacearum, a soil-borne bacterium causing the widespread disease known as bacterial wilt, is a natural root colonizer (Vailleau et al., 2007). Second, this plant pathogen has been extensively studied using a combination of genetic, genomic and post-genomic approaches (Peeters et al., 2013). Finally, as S. meliloti, R. solanacearum first infects the young part of the root, that is, the region of the initiation and development of root hairs (Turner et al., 2009).

This dual system allowed us to undertake a global comparative analysis between symbiotic and pathogenic transcriptomes using Affymetrix GeneChip® Medicago genome arrays. This comparative analysis was rather crude as it was based on whole-organ analyses with different temporal scales, but, nevertheless, it allowed us to identify a set of 964 genes deregulated in both interactions. Among these, 34% had specific behavior depending on the nature of the interaction, whereas 52% and 14% were commonly downregulated and upregulated, respectively. The proportions of transcripts belonging to the three categories were surprisingly close to those observed by Damiani et al. (2012), where nodule infection zone cells were compared with giant cells induced by root-knot nematodes in Mtruncatula following laser dissection. This could support the hypothesis that downregulation mechanisms are more conserved than those involved in the induction of genes between symbiotic and pathogenic interactions (Damiani et al., 2012).

It is well established that defense reactions need to be prevented to allow rhizobia–legume symbiotic interactions to take place (for a review, see Jones et al., 2007). Indeed, although rhizobia or Nod factors can transiently induce defense gene expression (Lohar et al., 2006; Nakagawa et al., 2011), steady expression during symbiotic interactions of defense genes and transcription factors, normally associated with pathogenic interactions, has only been described in defective nodules, for example induced by mutant strains of rhizobium, in particular exopolysaccharide- or lipopolysaccharide-defective strains (Niehaus et al., 1993; Tellström et al., 2007; Moreau et al., 2011). In this study, we observed a very unusual situation in which a transcription factor gene which plays an essential positive role for wild-type nodule development, MtEFD, is also induced by a plant pathogen, R. solanacearum, and plays a positive role in disease development. Another key symbiotic gene, MtNFP (Nod Factor Perception), has recently been reported to contribute to the regulation of pathogenic interactions between M. truncatula and A. euteiches (Rey et al., 2013). However, in this case, MtNFP is positively involved in the plant immunity against the pathogen.

We also identified that CKs, phytohormones which play a key positive role during the early and late stages of nodulation (Frugier et al., 2008), are strongly affected at the transcriptional level during the pathogenic interaction between M. truncatula and R. solanacearum. Indeed, transcriptomics data indicated that many transcripts involved in CK metabolism, signaling and response are differentially regulated already at 12 hpi and, in several cases, sustained at 72 hpi. The importance of CKs during this interaction was further strongly reinforced by genetic evidence, as the CK receptor essential for symbiotic nodulation, MtCRE1 (Plet et al., 2011), was shown to play a positive role in disease development. Finally, we established a link between MtEFD and CKs in the context of M. truncatulaR. solanacearum interactions, as MtEFD induction was abolished in the cre1-1 mutant.

During nodulation, MtEFD has been proposed to negatively control CK pathways via the type-A RR, MtRR4, thereby impacting on both nodule initiation and development (Vernié et al., 2008). An interesting question is therefore whether the same regulatory module is recruited during a pathogenic interaction. It is not yet possible to unequivocally answer this question. Indeed, several observations support a functional disconnection between MtEFD and MtRR4: (1) in contrast with MtEFD, MtRR4 induction in response to R. solanacearum does not require HrpB, that is the bacterial type III secretion system; (2) MtRR4 induction takes place much earlier than MtEFD induction (12 vs 72 hpi, respectively); and (3) MtRR4 induction is still detected in the efd-1 mutant. This may indicate that the ‘ins and outs’ of MtEFD regulation differ strikingly according to the type of plant–microorganism interaction: MtEFD induces the expression of the MtRR4 type-A RR during nodulation, and thereby negative feedback on CK signaling, whereas MtEFD expression is dependent on CK perception by the MtCRE1 receptor during the pathogenic interaction.

Nevertheless, MtRR4 has been shown to be transcriptionally activated by two independent pathways depending on the symbiotic nodulation stage: very early, when MtEFD is not yet upregulated (24 hpi; Lohar et al., 2006), depending on the CK phosphotransfer signaling pathway; and later (at 3–4 dpi) in an MtEFD-dependent manner (Vernié et al., 2008). It is therefore possible that, similarly, during the interaction with R. solanacearum, there are different temporal mechanisms controlling MtRR4 activation, first dependent on the CRE1 CK signaling pathway and later on the MtEFD pathway (Fig. 8).

Figure 8.

Model of cytokinins (CKs) and MtEFD interconnections during Ralstonia solanacearum pathogenic interaction and Sinorhizobium meliloti symbiotic interaction in the legume Medicago truncatula. In response to R. solanacearum (left panel in green) and dependent on the hrpG gene, CK perception via the CK receptor MtCRE1 activates CK response genes and, among them, the type-A response regulator MtRR4 (shown in this article). Whether R. solanacearum produces CKs or manipulates plant CKs remains unknown, but our microarray analyses did not show any plant CK biosynthesis gene upregulated in response to R. solanacearum. CK signaling therefore induces indirectly MtEFD expression (this article and Vernié et al., 2008; Ariel et al., 2012). MtEFD therefore positively acts on disease development via its target genes which need to be identified. We cannot exclude an EFD-independent pathway activating the disease via CK response genes (dotted purple line). In response to S. meliloti (right panel in blue), CK signaling is activated via the Nod factor pathway (Frugier et al., 2008; Den Camp et al., 2011). CKs are perceived via MtCRE1 and CK signaling genes are upregulated (Frugier et al., 2008; Den Camp et al., 2011), among them MtRR4. These CK responses positively regulate nodule development (Plet et al., 2011; solid line). In response to S. meliloti, MtEFD is expressed in the dividing cortical cells of young nodule primordia (Vernié et al., 2008). As CK signaling is essential for nodule primordia formation, we hypothesize that MtEFD is indirectly induced by CK response genes (blue dotted line). Therefore, MtEFD induces MtRR4 expression (solid blue line), controlling CK responses (purple dotted line) and nodule development. Other MtEFD targets are also probably involved in nodule development. Whether or not MtEFD activates the same targets in response to both pathogenic and symbiotic bacteria remains unknown. Common elements between R. solanacearum and S. meliloti responses are indicated in purple.

Transcriptomic analyses have revealed that CKs are profoundly affected by the pathogen, and it is possible that R. solanacearum exploits this alteration of hormone balance to improve its colonization efficiency. Although we could not obtain microscopic evidence for this, it is conceivable, for example, that modification of CK metabolism and response may help the bacteria to invade the stele region and the vasculature, as CKs are known to play a role in the regulation of vascular differentiation (Mahonen et al., 2006) and to improve nutrient allocation within the root (Walters et al., 2008). Alternatively, plant CK responses may also primarily help the plant to cope with the stress induced by R. solanacearum infection, and this response may indirectly benefit R. solanacearum, which behaves as a biotrophic pathogen during the first days of invasion.

A bacterial origin of CKs can be relevant for pathogenic interactions (e.g. in the Rhodococcus fascians–Arabidopsis interaction; Pertry et al., 2009). A tzs gene encoding an isopentenyl transferase protein has been identified in R. solanacearum K60 (Akiyoshi et al., 1987) and in the GMI1000 genome (Salanoubat et al., 2002), the expression of which is HrpG independent in GMI1000 (Valls et al., 2006). Akiyoshi et al. (1987) demonstrated that several strains of R. solanacearum could secrete CKs in liquid axenic cultures. In our study, preliminary results demonstrated that inoculation of M. truncatula wild-type roots with a tzs mutant led to the same disease development as inoculation with the GMI1000 wild-type strain. Even though the role of CKs and of an isopentenyl transferase gene in R. fascians has been demonstrated to be central for bacterial virulence on Arabidopsis thaliana (Pertry et al., 2009), we could not find such a correlation in our biological system. By contrast, a recent study has reported that the HopQ1 type III effector of Pseudomonas syringae pv. tomato can induce CK signaling, leading to the suppression of flagellin receptor FLS2 accumulation and, consequently, to the suppression of plant innate immunity (Hann et al., 2013). Therefore, it could be interesting to look for a putative HrpG-dependent CK signal in the M. truncatulaR. solanacearum interaction.

In the model proposed (Fig. 8), MtEFD transcription is not upregulated directly by CKs. Indeed, we have no evidence for a direct impact of CK signaling on MtEFD, as this gene is not transcriptionally induced by the exogenous treatment of roots with CKs (Ariel et al., 2012). MtEFD transcriptional activation by R. solanacearum is intimately connected with bacterial pathogenicity, as it is dependent on the type III secretion system regulator HrpB. The type III secretion system translocates some 70 bacterial effector proteins directly into the host cells to suppress defense responses and to facilitate bacterial multiplication during the first stages of infection (Poueymiro & Genin, 2009). The bacterial effectors triggering the MtEFD response have yet to be uncovered, taking advantage of the existence of 67 loss-of-function type III effector mutants in the GMI1000 background (Poueymiro & Genin, 2009). In the future, it will also be very interesting to identify the set of MtEFD target genes regulated in the context of pathogenic vs symbiotic interactions.

In conclusion, this work adds a new link between CKs and pathogenicity in an original legume root pathosystem. By showing the importance of the symbiotic transcription factor MtEFD in bacterial wilt, it also adds an important piece to the complex crosstalk existing between symbiotic and pathogenic responses.

Acknowledgements

This work was supported by the Agence Nationale de la Recherche-funded grant ‘GENOPEA’ (ANR-08-PXXX-0-03), a doctoral grant from EU-CNRS (Fonds Social Européen) to T. Vernié and a doctoral grant from INRA-CJS to J. Fromentin. Plant material, Affymetrix array and URGV services for transcriptome analyses were funded by Ecole Nationale Superieure d'Agronomie, part of the Institut National Polytechnique de Toulouse (ENSAT-INP). This research, initiated in ENSAT-INP, was performed in the Laboratoire des Interactions Plantes-Microorganismes, part of the Laboratoire d'Excellence (LABEX) entitled TULIP (ANR-10-LABX-41). Quantitative RT-PCR experiments were carried out at the Toulouse Genopole ‘PLAGE’ platform.

Ancillary