Biological control of bacterial wilt in Arabidopsis thaliana involves abscissic acid signalling


  • Dong Xin Feng,

    1. INRA, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR441, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR2594, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    3. Department of International Cooperation, Chinese Academy of Agricultural Sciences, 12 Zhongguancun South Street, Beijing 100081, China
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Céline Tasset,

    1. INRA, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR441, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR2594, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Mathieu Hanemian,

    1. INRA, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR441, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR2594, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Xavier Barlet,

    1. INRA, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR441, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR2594, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    Search for more papers by this author
  • Jian Hu,

    1. College of Biological Sciences, China Agricultural University, Beijing 100193, China
    Search for more papers by this author
  • Dominique Trémousaygue,

    1. INRA, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR441, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR2594, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    Search for more papers by this author
  • Laurent Deslandes,

    1. INRA, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR441, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR2594, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Yves Marco

    1. INRA, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR441, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes–Microorganismes (LIPM), UMR2594, Chemin de Borde Rouge F-31326 Castanet-Tolosan, France
    Search for more papers by this author
    • These authors contributed equally to this work.

Author for correspondence:
Yves Marco
Tel: +33 5 61 28 55 09


  • Means to control bacterial wilt caused by the phytopathogenic root bacteria Ralstonia solanacearum are limited. Mutants in a large cluster of genes (hrp) involved in the pathogenicity of R. solanacearum were successfully used in a previous study as endophytic biocontrol agents in challenge inoculation experiments on tomato. However, the molecular mechanisms controlling this resistance remained unknown.
  • We developed a protection assay using Arabidopsis thaliana as a model plant and analyzed the events underlying the biological control by genetic, transcriptomic and molecular approaches.
  • High protection rates associated with a significant decrease in the multiplication of R. solanacearum were observed in plants pre-inoculated with a ΔhrpB mutant strain. Neither salicylic acid, nor jasmonic acid/ethylene played a role in the establishment of this resistance. Microarray analysis showed that 26% of the up-regulated genes in protected plants are involved in the biosynthesis and signalling of abscissic acid (ABA). In addition 21% of these genes are constitutively expressed in the irregular xylem cellulose synthase mutants (irx), which present a high level of resistance to R. solanacearum.
  • We propose that inoculation with the ΔhrpB mutant strain generates a hostile environment for subsequent plant colonization by a virulent strain of R. solanacearum.


Plants actively respond to a variety of chemical stimuli produced by soil- and plant-associated microbes. Such stimuli can either induce or condition plant host defenses that enhance resistance against subsequent infection by a variety of pathogens. Distinct defense pathways are activated by plants depending on the type of invader encountered.

Pre-treatment of plants with several types of bacteria including living and killed saprophytic bacteria, avirulent bacteria and hrp (hypersensitive reaction and pathogenicity) bacterial mutants induced a wide range of defense mechanisms. A well documented induced response, known as systemic acquired resistance (SAR), involves the production of the signal molecule salicylic acid (SA) that plays a central role in the activation of disease resistance responses in plants (Ryals et al., 1996; Vlot et al., 2008). Another type of induced resistance, referred to as induced systemic resistance (ISR), develops in response to colonization of plant roots by selected strains of non-pathogenic rhizobacteria (Van Loon et al., 1998; Shoresh et al., 2010). In contrast to SAR, SA is not required for ISR but components of the jasmonic acid (JA) and ethylene responses are necessary (Knoester et al., 1999; van Wees et al., 2000).

Bacterial wilt caused by the phytopathogenic bacteria Ralstonia solanacearum is one of the most important plant diseases worldwide (Hayward, 1991). Ralstonia solanacearum infects > 200 plant species including many species of agronomic importance, such as potato, tomato, peanut and banana. Wilt disease is endemic in all the tropical and subtropical areas and difficult to control because of the soil-borne nature of the bacterium. The most effective method is the use of resistant crop varieties, which can be effective in one environment but is often overcome by virulent pathogen isolates when environmental conditions change. Another strategy is based upon inoculation of tomato or potato with a R. solanacearum hrp mutant strain followed by subsequent inoculation with a virulent strain. Under standardized conditions, protection rates > 80% were recorded 4 wk after pathogen inoculation (Trigalet & Trigalet-Demery, 1990; Frey et al., 1993). The ability of the hrp mutants to protect tomato plants from a subsequent attack by a virulent strain of the pathogen was not caused by a direct antagonism between the two strains, both derived from the same parental GMI1000 strain (Frey et al., 1994). This protective capacity appeared to be correlated with the ability of hrp mutants to colonize xylem vessels (Frey et al., 1994). Competition for space between the two bacterial strains may therefore be involved in the mechanism of protection (Etchebar et al., 1998).

These observations do not however exclude the possibility of the induction of defense in susceptible hosts in response to invasion by the hrp strain (Kempe & Sequeira, 1983; Frey et al., 1993; Trigalet et al., 1994). Vascular browning of cells in the vicinity of invaded tomato xylem vessels observed after inoculation with an hrp strain as well as the fact that such mutants are capable of activating defense responses suggest the involvement of active plant defense mechanisms which may contribute to the protective effect of the hrp strain (Godiard et al., 1990; Jakobek et al., 1993; Etchebar et al., 1998).

Studies of the genetic basis of resistance to R. solanacearum showed the existence of complex mechanisms in tomato (Thoquet et al., 1996a,b). To date, the Arabidopsis thaliana gene RRS1-R is the only characterized resistance gene (R) that confers resistance to R. solanacearum (Deslandes et al., 2002, 2003). Arabidopsis mutants unable to develop disease symptoms upon infection by the bacteria were also identified. Indeed, mutations (irx) in any of the three Arabidopsis cellulose synthases (CESAs) required for secondary cell wall formation (IRX1/CESA8, IRX3/CESA7 and IRX5/CESA4) conferred enhanced resistance to R. solanacearum (Hernandez-Blanco et al., 2007). Genetic and transcriptomic analyses revealed that IRX/CESA-mediated resistance was independent of SA, ET and JA signalling and identified a group of constitutively up-regulated genes, including abscissic acid (ABA)-responsive genes, in irx/cesa plants (Hernandez-Blanco et al., 2007).

In this report, A. thaliana was used as a model plant to better understand the molecular mechanisms involved in the biological control of bacterial wilt. Our data suggest that protection mediated by a hrp mutant strain relies on a novel molecular mechanism different from those involved in SAR and ISR and based on the manipulation of components of the ABA biosynthetic and response machinery.

Materials and Methods

Bacterial strains

The GMI1000 strain of Ralstonia solanacearum is a tomato wild type isolate (Boucher et al., 1987), which is virulent on Arabidopsis thaliana (L.) Heynh. ecotype Colombia (Col-0) and induces resistance on A. thaliana ecotype Niederzens (Nd-1). The R. solanacearumΔhrpB mutant strain used in this study is a derivative of strain GMI1000 (GMI1525, Genin et al., 1992). Both were grown at 28°C in BGT medium (Boucher et al., 1987). GMI1000 and GMI1525 bacteria were selected on rifampicin (50 μg ml−1) and streptomycin (40 μg ml−1), respectively.

Plant materials and inoculations

Four-week-old A. thaliana plants were root-inoculated by the ΔhrpB mutant strain in a suspension of 5 × 108 colony-forming units (CFU) ml−1, as previously described (Deslandes et al., 1998). Twenty-four hours after the first inoculation, plants were challenged by the virulent GMI1000 strain at a concentration of 2 × 107 CFU ml−1. Pathogenicity tests were conducted until complete wilting of unprotected plants. Heat-killed R. solanacearum bacteria were boiled for 30 s and used to inoculate plants in a suspension of 5 × 10CFU ml−1. Bacterial internal growth curves were performed according to the protocol described by Deslandes et al., 1998. Disease symptoms were scored according to the percentage of wilted leaves, using the following a 0–4 scale (0, no wilting; 1, 0–25%; 2, 25–50%; 3, 50–75%; 4, 75–100% of wilted leaves).

Inoculation with Pseudomonas syringae pv. tomato (Pst DC3000) were performed using ΔhrpB-inoculated or water-treated Arabidopsis plants as described above. After 24 h, plants were leaf-infiltrated with 2 × 105 CFU ml−1 of Pst DC3000 bacterial suspensions. Symptoms were scored at 5, 6, 7 and 8 d after infection. Plant inoculation with Pst DC3000 and in planta bacterial growth measured at 0 and 3 d after infiltration were performed as described in Raffaele et al. (2006).

Arabidopsis signalling mutants used in this study were obtained from the Arabidopsis Stock Center (Nottingham Arabidopsis Stock Center, Nottingham, UK). The pad2 and pad4 mutants were kindly provided by F. Ausubel (Harvard Medical School, Boston, USA).

Transcriptome analysis

Microarray analysis was performed on aerial parts of five plants harvested 24 h after water treatment or inoculation by the ΔhrpB strain and 5 d (25% of wilted leaves) after challenge inoculation with the GMI1000 virulent R. solanacearum strain. Two biological replicates corresponding to RNA extracted from different plants in two independent experiments were conducted. Total RNA was isolated from frozen tissues using the NucleoSpin RNAII kit (Macherey-Nagel; GmbH&Co.KG, Düren, Germany) according to the manufacturer’s recommendations. RNA concentrations were determined using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, and RNA integrity was confirmed by Bioanalyzer 2100 electrophoresis (Agilent Technologies, Probes were synthesized from RNA samples and hybridized to the Affymetrix Arabidopsis ATH1 GeneChip arrays (Affymetrix, according to the procedures provided by the manufacturer. Probes and hybridizations were performed by the microarray platform of Strasbourg Alsace-Lorraine Genopole at the IGBMC (

Expression measures were normalized by the Robust Multi-array Average (RMA) (Irizarry et al., 2003) implemented in Bioconductor packages. The pairwise comparison between the two biological replicates of water-treated sample and the two replicates of the hrp- inoculated samples was performed to identify differentially expressed genes. A similar comparison between Δhrp- and water-treated samples was performed 5 d after inoculation with the virulent strain.

A statistical analysis was performed with the LIMMA package using an empirical Bayes linear modelling approach (Smyth, 2004, 2005) and P-values were adjusted by the Benjamini and Hochberg method which controls the false discovery rate (FDR) (Benjamini & Hochberg, 1995). Significant up-regulated and down-regulated genes were selected using an adjusted FDR value of 0.01 and normalized ratios (Signal Log2 Ratio) of 0.8 relative to water-treated samples.

For real time RT-PCR analysis, the aerial parts of four plants were collected and RNA purified as described above for the transcriptional analysis. qRT-PCR reactions were performed in 384-well plates with a Lightcycler LC480 (Roche, using the LightCycler FastStart DNA MasterPlus SYBR Green I kit on 1 μg of total RNA (Roche applied Science, and cDNAs 10-fold diluted before use. The following conditions were used: one cycle of 9 min at 98°C followed by 45 cycles of 5 s at 95°C, 10 s at 65°C and 20 s at 72°C. The primer sets used in the experiments are listed in Supporting Information Table S6. The specificity of the amplification was systematically checked by melting curve analysis at the end of each run of real time RT-PCR. The relative expression of each gene of interest was calculated using At1g13320 as a reference gene for each sample.

Accession numbers

The microarray hybridization data have been submitted to ArrayExpress database (; accession number: E-MTAB-542).


Pre-inoculation of susceptible Col-0 plants with a ΔhrpB strain led to increased resistance to strain GMI1000 of R. solanacearum

Hrp genes encode components of type III secretory pathways and are required by many phytopathogenic bacteria to elicit a hypersensitive response (HR) on non-host or resistant host plants and for pathogenesis on susceptible hosts (Alfano & Collmer, 1997). Genetically engineered R. solanacearum mutants in these genes have been described as potential biocontrol agents of tomato and potato bacterial wilt (Genin et al., 1992; Frey et al., 1994; Smith et al., 1998). We reasoned that Arabidopsis may be a more suitable system to investigate the molecular mechanisms underlying plant protection mediated by hrp mutant strains. Four-week-old Arabidopsis susceptible Col-0 plants were either root-inoculated with an avirulent R. solanacearum strain mutated in the hrpB gene that encodes an AraC family regulator (Genin et al., 1992; Cunnac et al., 2004) or treated with water. Twenty-four hours later, these plants were subsequently challenged with the virulent GMI1000 strain of R. solanacearum. Pre-inoculation with the ΔhrpB strain protected the plants since only light wilting symptoms were observed 13 d after inoculation with the pathogenic strain (Fig. 1a,b). By contrast, when plants were pretreated only with water and later challenged with the virulent pathogen, wilt disease developed 4–6 d after inoculation with the GMI1000 strain, leading to plant death after 8–10 d. The protective effect was observed whether inoculation with GMI1000 was performed 24, 48, 72 h or 1 wk after pre-inoculation with the ΔhrpB mutant (data not shown).

Figure 1.

Pre-inoculation with a ΔhrpB strain protects plants from a subsequent challenge with a virulent Ralstonia solanacearum strain. (a) Phenotypic responses of Arabidopsis thaliana Col-0 plants treated with water (left panel) or inoculated with the ΔhrpB strain (right panel), 9 d after subsequent root inoculation by the R. solanacearum GMI1000 strain. (b) Disease symptom development curves. Plants were scored daily using a scale between 0 and 4. Means and standard errors (SE) were calculated from a total of 70 plants per assay (from three independent experiments). Open triangles, Col-0 plants inoculated with ΔhrpB and challenged with the virulent GMI1000 strain 24 h later; open circles, Col-0 plants co-inoculated with the ΔhrpB and GMI1000 strains; open squares, Col-0 plants treated with water and inoculated with GMI1000 strain after 24 h.

This protection might require an active plant metabolism or result from a direct physical competition for space within the xylem vessels between the two strains, which might lead to exclusion of the GMI1000 strain by the ΔhrpB mutant. In order to distinguish between these two hypotheses, both the virulent and avirulent strains were simultaneously co-inoculated using a 1 : 10 ratio. Even under these conditions that favored multiplication of the ΔhrpB strain, co-inoculated plants wilted in a similar manner than control plants pre-treated with water (Fig. 1b). This suggests that competition is not the main determinant of ΔhrpB -induced protection.

Various pretreatments induce different levels of resistance to a challenge inoculation with the pathogenic GMI1000 strain

Several pioneering studies showed that pretreatment with various compounds such as bacterial lipopolysaccharides (LPS), avirulent or heat-killed bacteria (Sequeira & Hill, 1974; Rathmell & Sequeira, 1975; Graham et al., 1977; Tanaka, 1983, 1985) protect plants, to a certain extent, from a subsequent inoculation with a virulent strain of R. solanacearum. To compare the efficiency of such treatments, susceptible Col-0 plants were pre-inoculated with Escherichia coli cells or heat-killed GMI1000 bacteria. A challenge inoculation with the virulent R. solanacearum strain was then performed 24 h later. A slight delay in symptom appearance (24–48 h) was observed in plants pre-inoculated with E. coli cells compared to water treated plants (Fig. 2). The protective effect was stronger in Arabidopsis plants pre-treated with heat-killed GMI1000 bacteria. None of these pre-treatments however conferred the high level of protection observed in plants protected by the ΔhrpB mutant strain (Fig. 2).

Figure 2.

Disease symptom development curves in response to different protective treatments on wilt disease development. Arabidopsis thaliana plants were first treated with water (open squares), or inoculated with Escherichia coli cells (filled squares), heat-killed GMI1000 bacteria (filled triangles) or the ΔhrpB strain (open triangles). They were then challenged with the virulent Ralstonia solanacearum GMI1000 strain 24 h later. Means and standard errors (SE) were calculated from a total of 70 plants per assay (from three independent experiments).

Pre-inoculation of susceptible plants with the ΔhrpB strain limits bacterial multiplication of the virulent bacteria

We next tested whether reduced disease symptoms correlate with lower bacterial multiplication in protected plants. Bacterial growth was monitored on aerial parts of plants after root pre-inoculation with water or the ΔhrpB strain, followed by inoculation with the GMI1000 R. solanacearum strain. The ratio of inocula of the virulent vs the mutant strain was 0.1, which explains the differences observed between the concentrations of the two bacterial strains at time 0 (Fig. 3).

Figure 3.

 Pre-inoculation of Arabidopsis thaliana plants with a ΔhrpB strain of Ralstonia solanacearum limits multiplication of virulent bacteria in planta. Bacterial multiplication of ΔhrpB or GMI1000 strain was determined at 0, 3, 8 or 10 days after inoculation as follows: ΔhrpB, plants inoculated only with the ΔhrpB strain; ΔhrpB + GMI1000, plants pre-inoculated by the ΔhrpB strain and then challenged with strain GMI1000 24 h later; GMI1000, plants inoculated only with the GMI1000 strain; H2O + GMI1000, plants treated first with water and then challenged with strain GMI1000 24 h later; ΔhrpB + GMI1000, plants pre-inoculated by the ΔhrpB strain and then challenged with strain GMI1000 24 h later. Bacterial concentrations used in this experiment were 2 × 107 CFU ml−1 for the GMI1000 strain and 5 × 108 CFU ml−1 for the ΔhrpB strain. Triplicate assays were performed on three plants for each time point. This experiment was repeated twice, and reproducible results were obtained. Error bars represent SE of the mean.

Control plants were inoculated either with the GMI1000 or the ΔhrpB strain. As reported previously (Deslandes et al., 1998), ΔhrpB bacterial populations remained to low levels in Col-0 plants whereas virulent GMI1000 bacteria reached very high concentrations (1012 CFU g−1 FW; Fig. 3). Bacterial multiplication of the ΔhrpB strain was comparable in plants subsequently inoculated or not by the virulent GMI1000 strain. Pretreatment of plants with water before inoculation with GMI1000 did not affect bacterial multiplication that reached a comparable level to that of control plants. On the contrary, in plants pre-inoculated with the ΔhrpB strain and later challenged with the virulent strain, the bacterial density of the GMI1000 strain increased slowly and remained 4–5 orders of magnitude lower than that found in water-pretreated plants (c. 107 CFU g−1 FW). These data indicate that the pre-inoculation of susceptible plants with the ΔhrpB strain triggers an enhanced resistance response to the virulent strain that limits its multiplication in planta.

Pre-inoculation with the ΔhrpB strain protects plants against subsequent attack by a virulent strain of Pseudomonas syringae

We then checked whether protection caused by pre-inoculation of the R. solanacearumΔhrpB strain was specific of root pathogens or was also efficient against foliar pathogens. ΔhrpB pre-treated plants were therefore challenged with strain DC3000 of Pseudomonas syringae, a foliar bacterial pathogen, virulent on Col-0 plants. Water treated plants developed chlorotic symptoms after 3 d whereas plant pre-inoculation with the ΔhrpB strain led to a significant reduction of symptoms in response to bacterial infection (Fig. 4a), which correlated with a decrease in bacterial multiplication in protected plants 3 d after inoculation (Fig. 4b).

Figure 4.

Pre-inoculation of Arabidopsis thaliana plants with ΔhrpB protects plants against a foliar pathogen. (a) Phenotypic responses of Arabidopsis leaves treated with water (left panel) or inoculated with the ΔhrpB strain of Ralstonia solanacearum (right panel) and then leaf-infiltrated with a bacterial suspension of the Pseudomonas syringae DC3000 strain. Symptoms were scored at 3 d after infection. (b) In planta bacterial growth was measured at time 0 in plants pre-treated with water (black) or pre-inoculated with the ΔhrpB strain (white) and 3 d after inoculation in plants pre-inoculated with water (black with white dots) or with the ΔhrpB strain (white with black dots). Mean bacterial densities are calculated from two independent experiments (four individual plants per experiment). Error bars represent SE of the mean. *, < 0.05; **, < 0.01; using a t-test.

Resistance conferred by plant pre-inoculation with a ΔhrpB strain does not involve classical defense-associated signalling pathways

The molecular mechanisms involved in biological control against R. solanacearum remain unexplained although various hypotheses have been proposed (Trigalet & Trigalet-Demery, 1990). In order to identify the signalling pathways involved in the establishment of this type of resistance, the protective ability of hrpB mutant against a subsequent invasion by a pathogenic strain was estimated in various Col-0 mutants affected in plant defense responses. The Arabidopsis mutants tested and their responses to the virulent strain GMI1000 are listed in Table 1. NahG and sid2-2 plants altered in SA accumulation, as well as cpr1, ndr1 and npr1 mutants that are altered in the expression of pathogenesis-related (PR) proteins, were fully protected from the virulent GMI1000 strain after pre-inoculation with the ΔhrpB mutant (Fig. S1). Likewise, mutants affected in the ethylene signalling pathways such as ein2-1, shown previously to be more tolerant to the bacteria (Hirsch et al., 2002) or in the JA signalling pathways such as jar1-1 and coi1 displayed a similar protection to wild type plants. Finally, pad2 and pad4, phytoalexin-deficient mutants were not affected in terms of the ΔhrpB -induced protection (Fig. S1). These data suggest that neither SA, JA nor ethylene play a major role in the plant protection induced by the ΔhrpB strain.

Table 1.   Responses of various defense-related Arabidopsis thaliana mutants to Ralstonia solanacearum GMI1000, 10 d after inoculation by the ΔhrpB strain
Arabidopsis linesAffected pathwaysProtectionReferences
  1. All the tested lines were in a Col-0 background except abi1-1 for which the appropriate control (Ler-0) was used. In the ‘Protection’ column, ‘Yes’ means that plants were as protected as wild type plants; ‘No’ means that plants developed some wilt symptoms.

aba1-5ABA (biosynthesis)YesLéon-Kloosterziel et al. (1996)
aba2-1ABA (biosynthesis)YesLéon-Kloosterziel et al. (1996)
aba3ABA (biosynthesis)YesLéon-Kloosterziel et al. (1996)
abi1-1ABA (signalling pathway)NoKoornneef et al. (1984)
abi4-101ABA (signalling pathway)YesLaby et al. (2000)
ein2-1EthyleneYesGuzman & Ecker (1990)
jar1-1JAYesStaswick et al. (1992)
coi1JAYesFeys et al. (1994)
Pap1JAYesBorevitz et al. (2000)
cpr1SA (constitutive)YesBowling et al. (1994)
sid2-2SA (biosynthesis)YesNawrath & Metraux (1999)
pad2SA, camalexin biosynthesisYesGlazebrook & Ausubel (1994)
pad4SA, camalexin biosynthesisYesGlazebrook et al. (1997)
NahGSA (accumulation)YesLawton et al. (1995)
npr1SA (response)YesCao et al. (1994)
AtMyb96SA-ABAYesSeo et al. (2009)
ndr1R gene-mediated resistanceYesCentury et al. (1995)

The same approach was used to show that these defense pathways were not involved in protection against P. syringae. Indeed, our data indicated that the phenotypical responses of protected mutants were similar to those of control Col-0 plants. Bacterial multiplication was estimated in the different mutants and confirmed this observation (Fig. 4b).

Transcriptional reprogramming of ΔhrpB -protected plants

In order to gain some knowledge on the mechanisms governing ΔhrpB-induced resistance, a transcriptome analysis was performed. Plants were pre-inoculated with the ΔhrpB strain or treated with water. Twenty-four hours later, they were inoculated with the virulent GMI1000 strain. Plant samples were harvested 24 h after pre-treatment, before inoculation with the virulent strain (P24 for ΔhrpB -inoculated plants, W24 for water-treated plants). Samples were also harvested 5 d after inoculation with the GMI1000 strain, at the appearance of the first wilting symptoms in unprotected plants (PD1 for protected plants, WD1 for wilting unprotected plants). Microarray data were generated using A. thaliana whole genome Affymetrix ATH1 chips. To confirm the reliability of the microarray results, 13 probe sets were selected to confirm expression differences with quantitative real time PCR (qRT-PCR). Selection of these probe sets was based on their statistically significant up- or down-regulated expression. The ratios from microarray and qRT-PCR were well correlated (R2 = 0.7541) and could be described by the linear regression equation = 0.6907x − 0.8496, indicating good consistency between the two methods (Fig. S2).The Classification Superviewer tool ( (Provart & Zhu, 2003) was used to define enriched functional classes in our analyses.

First we investigated transcriptional reprogramming associated to pre-inoculation with ΔhrpB strain after 24 h (P24/W24). Down- and up- regulated genes are listed in Tables S1 and S3, respectively. Among the 152 genes down-regulated in plants pre-inoculated with the ΔhrpB strain, a high enrichment of cell wall- and extracellular-related genes was observed. Cell-wall related genes were previously described as playing a role in the elaboration of the plant response to R. solanacearum. Indeed, Arabidopsis irx/cesa mutants impaired in cellulose synthase genes show an enhanced level of resistance to R. solanacearum and transcriptomic analysis of these mutants supported a direct role for ABA in resistance to R. solanacearum (Hernandez-Blanco et al., 2007). Interestingly, ΔhrpB-dependent gene expression changes affect 336 genes that are up-regulated and enriched in abiotic or biotic stimuli response genes and 91 of these genes (27%) were previously identified as being involved in ABA biosynthesis and signalling (Li et al., 2006) (Table S2a). A selection of marker genes involved in ABA biosynthesis, signalling and ABA-related transcription factors are listed in Table 2. Taken together, our data indicate that root inoculation with the ΔhrpB strain leads to the activation of ABA-related pathways.

Table 2.   Expression levels of selected transcripts up-regulated in protected Arabidopsis thaliana plants after 24 h classified according to functional groups
AGIGene descriptionFold change
  1. Columns on the right side of the table indicate the fold changes for plants inoculated with the ΔhrpB strain after 24 h (P24/W24) and challenged afterwards with the GMI1000 virulent Ralstonia solanacearum strain (PD1/WD1). Positive values represent up-regulation, negative values represent down-regulation, and asterisks denote FDR < 0.01. The complete list of genes up-regulated in protected plants after 24 h (SLR > 0.08 and FDR < 0.01) is included in Supporting Information Table S1.

Secondary metabolismP24/W24PD1/WD1
 At5g38130Transferase family protein similar to transferase family protein2.43*1.05
 At2g37040Phe ammonia lyase 1 (pal1); phenylalanine ammonia-lyase1.74*− 2.21*
 At3g10340Phenylalanine ammonia-lyase 4 (PAL4); ammonia ligase/ammonia-lyase/catalytic1.82*− 4.25*
 At3g53260Phenylalanine ammonia-lyase (PAL2)2.19*− 1.67*
 At4g37980ELICITOR-ACTIVATED GENE 3-1 (ELI3-1); binding/catalytic/oxidoreductase/zinc ion binding1.79*− 1.01
 At3g212304-coumar ate : CoA ligase 5 (4CL5); 4-coumar ate-CoA ligase2.47*− 1.77*
 At3g14440NINE-CIS-EPOXYCAROTENOID DIOXYGENASE 3 (NCED3); 9-cis-epoxycarotenoid dioxygenase3.12*− 3.83*
 At4g26080ABA INSENSITIVE 1 (ABI1); calcium ion binding/protein serine/threonine phosphatase1.76*− 1.63*
 At5g57050ABA INSENSITIVE 2 (ABI2); protein serine/threonine phosphatase2.03*− 2.00*
 At5g59220Protein phosphatase 2C. putative/PP2C. Putative similar to protein phosphatase 2C4.03*− 5.28*
Abiotic stress
 At1g59860Heat shock protein. putative similar to heat shock protein GI:19617 from2.49*1.35
 At2g20560DNAJ heat shock family protein similar to DNAJ heat shock family protein2.16*− 1.92*
 At3g07770ATP binding similar to CR88 (EMBRYO DEFECTIVE 1956). ATP binding1.78*− 1.10
 At3g09440Heat-shock protein (At-hsc70-3) identical to (At-hsc70-3) (cytosolic Hsp70)2.17*− 1.84*
 At3g12580Heat shock protein 70 (HSP70); ATP binding5.19*− 1.91*
 At3g14200DNAJ heat shock N-terminal domain-containing protein similar to heat shock protein binding1.93*− 1.57*
 At5g09590MITOCHONDRIAL HSP70 2 (MTHSC70-2); ATP binding2.29*− 1.19
 At5g5144023.5 kDa mitochondrial small heat shock protein (HSP23.5-M) similar to ATHSP23.6-MITO2.27*− 1.80*
 At5g52640Heat shock protein 90.1 (ATHSP90.1); ATP binding2.94*− 1.50
 At5g56030Heat shock protein 81-2 (HSP81-2); ATP binding1.75*− 1.44*
 At1g13930Similar to nodulin-related1.75*− 1.14
 At3g56080Dehydration-responsive protein-related similar to dehydration-responsive family protein2.23*1.20
 At4g23680Major latex protein-related/MLP-related similar to major latex protein-related2.18*1.12
 At5g66400RESPONSIVE TO ABA 18 (RAB18)2.35*− 6.71*
 At2g22200AP2 domain-containing transcription factor encodes a member of the DREB subfamily A-6 of ERF/AP2 transcription factor family1.78*1.48
 At4g25480DEHYDR ATION RESPONSE ELEMENT B1A (DREB1A); DNA binding3.68*− 1.63*
 At4g28140AP2 domain-containing transcription factor. putative encodes a member of the DREB subfamily A-6 of ERF2.28*− 2.30*
 At5g05410DREB2A; DNA binding/transcription activator/transcription factor2.36*− 4.14*
 At5g11590TINY2 (TINY2); DNA binding/transcription factor1.94*1.21
 At5g67180AP2 domain-containing transcription factor. putative1.90*1.06

We then focused on the transcriptional reprogramming triggered by a subsequent infection with the virulent R. solanacearum GMI1000 strain (PD1/WD1). In ΔhrpB pre-treated plants, 194 and 991 genes were up- and down-regulated, respectively, in comparison to unprotected plants developing wilt disease (Tables S4, S5). Both sets of genes were enriched in abiotic and biotic stimuli responsive genes. Apart from the induction of two thionin genes that encode peptides with strong antimicrobial properties (Loeza-Angeles et al., 2008), very few up-regulated genes are associated to defense responses, thereby supporting the hypothesis that protection does not involve classical defense pathways. Interestingly, among down regulated genes, a large proportion of genes were previously described as being up-regulated during wilt disease development (Hu et al., 2008). Furthermore, this set of down-regulated genes comprises 25% of the genes (86 genes) that were up-regulated in ΔhrpB-pretreated plants (P24/W24), as shown in Fig. 5. Taken together, our data indicate that ΔhrpB pretreatment positively regulates in symptomless plants a set of genes that are normally induced at later stage during disease development.

Figure 5.

Number of differentially transcribed genes in Ralstonia solanacearumΔhrpB-protected Arabidopsis thaliana plants vs water-treated plants. The number of deregulated genes for the PD1/WD1 and P24/W24 stages (see the Results section for details) are presented on the x-axis and y-axis, respectively. Overlapping genes between PD1/WD1 and P24/W24 are shown in hierarchical clustering. Down-regulated and up-regulated genes are represented in green and in red, respectively. Hierarchical clustering (average linkage) was performed with Gene Cluster/Treeview (Eisen et al., 1998) on data presented in Supporting Information Tables S1, S3, S4 and S5.

Analysis of the 1 kb promoter regions of the co-regulated genes using the Promomer program (Toufighi et al., 2005) revealed that > 26% of the 991 down-regulated genes in protected plants (PD1/WD1) and > 35% of the 336 genes up-regulated 24 h after ΔhrpB strain inoculation (P24/W24) contained one or more ABRE (ABA responsive element)-related response element(s), (ACGTG(G-T)C; TACGTGTC; YACGTGGC). Another DRE/CRT response element (Dehydration Responsive Element/C repeaT, (A-G)CCGAC) (Tuteja, 2007) was also identified in 26% of the 991 and 24.7% of the 336 promoters. About 10% of these 336 and 991 promoters contained both ABRE and DRE elements. The significant over-representation of these motifs was confirmed using POBO (Kankainen & Holm, 2004).

The ABI1 gene is required for full ΔhrpB -mediated protection

To estimate the importance of ABA in the establishment of ΔhrpB-mediated protection, five knock-out lines in ABA biosynthesis and signalling related genes were tested. Upon challenge with the virulent strain, the level of protection of aba1-5, aba2-1, aba3 and abi4-101 mutants was comparable to that of wild type plants. Interestingly, abi1-1 mutant developed however significantly more disease symptoms than, wild type, Ler-0 pre-inoculated plants (Fig. 6), even if Ler-0 and abi1-1 mutant plant displayed heterogeneous responses to the virulent GMI1000 strain, with some plants wilted completely and rapidly (within a day) while others were fully resistant to the pathogen. This result suggests that ABI1-1 plays a key role in ΔhrpB-induced protection.

Figure 6.

The ABI1 gene is required for ΔhrpB-associated protection. Symptom development of Ler-0 (grey) and abi1-1 (black) Arabidopsis thaliana plants 9 d after Ralstonia solanacearum GMI1000 inoculation. Plants were first inoculated with water or with the ΔhrpB strain and 24 h later, were then challenged with the virulent GMI1000 strain. Mean disease indexes are calculated from three independent experiments (minimum 30 plants per condition and per experiment). Error bars represent SE of the mean. **, < 1.0 × 10−8; using a t-test.


Bacterial wilt caused by R. solanacearum is a serious disease of many valuable plant species in tropical and subtropical areas. Strategies for controlling this disease are limited and biological control of bacterial wilt using hrp mutant strains of the pathogen constitutes an alternative that has been developed for a decade, particularly in tomato. The molecular mechanisms underlying this resistance are completely unknown although several hypotheses have been proposed. In this study, we demonstrate that Arabidopsis plants can be protected by an initial root-inoculation with a ΔhrpB strain against a subsequent challenge by a virulent strain of the pathogen and that ABA-related signalling pathways intervene in this process.

Various treatments including root-inoculation of Arabidopsis plants with E. coli, heat-killed GMI1000 and ΔhrpB bacteria induced different levels of protection against the virulent R. solanacearum strain. Heat-killed virulent GMI1000 bacteria induced a higher level of protection than E. coli. This observation may be explained by the important role played by surface components of the pathogen in the recognition and induction of disease resistance. LPS is a cell surface component involved in the interaction with eukaryotic organisms (Newman et al., 2000) and may therefore elicit plant defense responses (Leach et al., 1983; Barton-Willis et al., 1984; Leeman et al., 1995; Milling et al., 2010). When injected into tobacco leaves, a purified LPS fraction extracted from R. solanacearum was shown to strongly induce disease resistance (Graham et al., 1977). Apart from LPS, proteinaceous compounds may be involved in this protective effect since boiled extracts from R. solanacearum possess also a strong elicitor activity attributable to one or more proteins (Pfund et al., 2004). However, heat-killed bacteria protect plants from a subsequent challenge with the virulent pathogen to a lesser extent than living bacteria, which supports the hypothesis of an active participation from both the host and the pathogen.

Two major reasons may explain the increased resistance observed after pre-inoculation with a hrp strain: the resistance response displayed by the protected plant may arise from a direct competition for vascular colonization between the protective and virulent strain. It was previously shown that mutant and pathogenic strains compete to invade xylem vessels in tomato (Etchebar et al., 1998). Our data suggest that physical competition plays a minor role in protection. Indeed, simultaneous co-inoculation of a 10-fold excess of the ΔhrpB strain with the virulent strain does not lead to an increased resistance of the plants, which should be the case if the protecting bacteria hindered the colonization of a particular niche by the virulent strain. Alternatively, it may be caused by the induction of defense responses of the host plant. Induced resistance has been cited as a possible mechanism of biocontrol involving spontaneous nonpathogenic mutants of R. solanacearum (Kempe & Sequeira, 1983). A major role of the SA- or ethylene/JA acid-associated signalling pathways in the establishment of the hrp-induced resistance can probably be ruled out: abrogation of these pathways had no effect on this resistance and few genes related to SA, JA and ethylene showed a differential expression in protected plants. Therefore, protection by the ΔhrpB mutant is probably related neither to systemic acquired resistance, a SA-dependent response to a localized infection nor to the JA/ethylene-dependent ISR (Shoresh et al., 2010). This protection affects also the multiplication of the bacterial pathogen P. syringae, which suggests that activated protective mechanisms are efficient both against foliar and root pathogens.

A transcriptome analysis was performed to decipher the mechanisms underlying protection by the ΔhrpB strain. Two time points were analysed, a first one, 24 h after pre-treatment with ΔhrpB strain or water (P24/W24) and a second one following a subsequent inoculation with the virulent strain (PD1/WD1). Already 24 h after pre-treatment with the ΔhrpB strain, expression of almost 500 plant genes is modified, including many up-regulated genes that were previously shown to be induced by a virulent strain in wilting plants (Hu et al., 2008). In ΔhrpB protected plants subsequently inoculated with a virulent strain (PD1/WD1), 25% of these up-regulated genes become down regulated. Our data indicate that, in healthy plants, pre-treatment with ΔhrpB strain brings forward a transcriptional reprogramming of genes normally induced lately during wilt disease development (Fig. 5).

In healthy protected plants, ΔhrpB-dependent gene expression changes mainly concern biotic and abiotic-stress related genes. Interestingly, strong induction of two thionin genes may partly explain the observed resistance. Thionins are small basic, cystein-rich peptides that display a broad in vitro antifungal and antibacterial activity resulting from permeabilization of cell membranes (Carrasco et al., 1981; Molina et al., 1993). Indeed, tomato transgenic plants expressing the Arabidopsis THI2.1, a gene strongly activated in protected plants challenged by the virulent strain, exhibit enhanced resistance to R. solanacearum (Chan et al., 2005).

Additionally, this study uncovers the involvement of genes associated to biosynthesis and signalling of ABA after a ΔhrpB pre-treatment (Tables 2, S2). To evaluate the importance of this hormone in ΔhrpB-induced protection, mutants affected in genes related to ABA biosynthesis and signalling were tested. Among all of them (Table 1), ABA insensitive abi1-1 plants developed significantly more disease symptoms than wild type plants. ABA responsive genes that are still ABA regulated in abi1-1 mutant (Hoth et al., 2002) might explain the partial loss of protection of the abi1.1 mutant. Additionally, this partial loss of protection suggests that ΔhrpB-induced resistance probably requires the activation of a whole set of genes and that inactivation of a single gene is not sufficient to prevent protection. Identification of ABRE- and DRE-related ABA response elements, depicted in abiotic stress responses (Tuteja, 2007), within many co-regulated gene promoters 24 h after protection suggests a regulation of gene expression through DREB or ABF transcription factors in the establishment of protection. DREB1A (At4g25480) and DREB2A (At5g05410) that are up-regulated 24 h after inoculation with the ΔhrpB strain, are therefore potentially involved in ABA signalling during protection. This result highlights the importance of ABA signalling in the biocontrol of R. solanacearum. A previous study already showed the direct involvement of ABA signalling in resistance to R. solanacearum (Hernandez-Blanco et al., 2007) thereby demonstrating that ABA plays a key role in the establishment of the plant response to this pathogen.

In plant–pathogen interactions, a complex role for ABA is emerging. In many cases, ABA involves cross-talks with SA- or ethylene/JA-associated signalling pathways (Mauch-Mani & Mauch, 2005; Melotto et al., 2006; de Torres-Zabala et al., 2007; Asselbergh et al., 2008; Fan et al., 2009; Cao et al., 2011). Our transcriptomic and genetic data reveal that biological control of bacterial wilt is not associated with SA and JA-signalling pathways. Two main hypotheses may explain the activation of ABA-/abiotic stress-associated gene expression in protected plants. Inoculation of the ΔhrpB strain may provoke a reduction of water transport in xylem vessels leading to a hydric stress and to the subsequent activation of ABA-related signalling. Alternatively, some hrp-dependent genes may play a role in the activation of the ABA-related signalling, for example by disrupting the complex interplay between hormone-mediated signalling events occurring during plant response to the pathogen. Indeed, in addition to hrpB, a regulatory switch that controls multiple virulence pathways including Type III secretion effectors (Occhialini et al., 2005), R. solanacearum possesses another master pathogenicity component, HrpG. This regulator controls major steps during the interaction with plant cells by regulating genes involved in secondary metabolic pathways, detoxification of various antimicrobial compounds as well as genes directing the biosynthesis of phytohormones and many genes of yet unknown function (Genin, 2010). Future studies will be aimed at the identification of bacterial genes potentially involved in the induction of biological control.

In conclusion, the present study demonstrates an active participation of the host plant through the activation of an ABA-dependent defense system independent of SA, JA and ethylene in the biological control exerted by the ΔhrpB mutant of R. solanacearum. The molecular mechanisms underlying this form of resistance are still hypothetical: it is conceivable that dehydration stress caused by root-inoculated ΔhrpB bacteria and/or some yet unknown bacterial determinants triggers a stress leading to the activation of signalling components common to both biotic and abiotic stress. Some of these components, including ABA-related genes, would generate directly or indirectly a hostile environment for bacteria as previously shown for the irx mutants. Alternatively, this environment may prime resistance: expression of Thi2.1, a peptide whose overexpression renders tomato plants resistant to R. solanacearum, was indeed significantly activated in protected plants challenged by the virulent bacteria. Presumably, the combinatory effect of these different mechanisms constitutes the basis of the ΔhrpB-mediated protection. Whether these mechanisms are also efficient against other vascular pathogens remains an open question that we would like to address in future studies. In any case, their deciphering is a prerequisite for rational development of efficient biological control for agriculture.


Quantitative RT-PCR experiments were carried out at the Toulouse Genopole ‘PLAGE’ platform. The authors thank Pariente J-L., Icher C. and Bosc S. for the plant production. J.H. was supported by the National Science Foundation of China (Grants 31071675 and 30671179). This work was supported by a PhD fellowship from the French Ministry for Higher Education and Research CNRS (C.T.) and by a PhD fellowship from the Agence Nationale de la Recherche, Grant PCS-08-GENO-SCRIPS (M.H.). This work is part of the Laboratoire d’Excellence (LABEX) entitled TULIP (ANR-10-LABX-41).