N-acyl-L-homoserine lactone (AHL) signal molecules are utilized by Gram-negative bacteria to monitor their population density (quorum sensing) and to regulate gene expression in a density-dependent manner. We show that Serratia liquefaciens MG1 and Pseudomonas putida IsoF colonize tomato roots, produce AHL in the rhizosphere and increase systemic resistance of tomato plants against the fungal leaf pathogen, Alternaria alternata. The AHL-negative mutant S. liquefaciens MG44 was less effective in reducing symptoms and A. alternata growth as compared to the wild type. Salicylic acid (SA) levels were increased in leaves when AHL-producing bacteria colonized the rhizosphere. No effects were observed when isogenic AHL-negative mutant derivatives were used in these experiments. Furthermore, macroarray and Northern blot analysis revealed that AHL molecules systemically induce SA- and ethylene-dependent defence genes (i.e. PR1a, 26 kDa acidic and 30 kDa basic chitinase). Together, these data support the view that AHL molecules play a role in the biocontrol activity of rhizobacteria through the induction of systemic resistance to pathogens.
Rhizosphere bacteria multiply to high densities on plant root surfaces where root exudates and root cell lysates provide ample nutrients. Sometimes, they exceed 100 times those densities found in the bulk soil (Campbell & Greaves 1990). Certain strains of these plant-associated bacteria stimulate plant growth in multiple ways: (1) they may fix atmospheric nitrogen; (2) reduce toxic compounds; (3) synthesize phytohormones and siderophores; or (4) suppress pathogenic organisms (Bloemberg & Lugtenberg 2001). Research on the ‘biocontrol’ activity of rhizobacteria has seen considerable progress in recent years. Disease suppression of soilborne pathogens includes competition for nutrients and production of antimicrobial compounds or lytic enzymes for fungal cell walls or nematode structures (Persello-Cartieaux 2003). By contrast, systemic resistance can also be induced by rhizosphere-colonizing Pseudomonas and Bacillus species where the inducing bacteria and the challenging pathogen remained spatially separated excluding direct interactions (Van Loon, Bakker & Pieterse et al1998; Ryu et al. 2004).
A well-characterized system of rhizobacteria-induced resistance between Arabidopsis thaliana and Pseudomonas fluorescens strain WCS417 has been elaborated by Van Loon's group (Pieterse et al. 1996; Van Loon et al. 1998). This type of resistance was termed ‘induced systemic resistance’ (ISR) in order to differentiate it from the pathogen-induced and salicylate-mediated ‘systemic acquired resistance’ (SAR) (Van Loon et al. 1998; Shirasu & Schulze-Lefert 2000). During ISR, colonization with rhizobacteria leads to an enhanced expression (‘priming’) of jasmonate-inducible genes (Van Wees et al. 1999). With the help of various biosynthesis and perception mutants for plant signal molecules, it was possible to demonstrate that ISR requires functional jasmonate and ethylene signalling (Van Loon et al. 1998; Pieterse et al. 2001).
Bacteria have developed efficient systems to perceive their environment by means of diffusible signals. Well-characterized molecules are N-acyl-L-homoserine lactones (AHLs), which occur in various Gram-negative bacteria. They are used as signal molecules to control expression of various functions in a cell density-dependent manner, a phenomenon known as ‘quorum sensing’. The structures of AHL molecules discovered to date vary in size and composition of their acyl chains. Molecules with chain lengths of 4–16 carbon atoms, which often contain an oxo or hydroxyl group on the third carbon, have been described (Eberl 1999). The strain mainly used in this study, Serratia liquefaciens MG1, produces two AHL molecules [N-butanoyl (BHL) and N-hexanoyl homoserine lactones (HHL)] with the aid of the AHL synthase SwrI (Eberl et al. 1996). Swarming motility, a special form of bacterial surface locomotion, the production of extracellular proteolytic and chitinolytic activity as well as the ability to form biofilms on abiotic surfaces, has been shown to be regulated by quorum sensing in this strain (Eberl, Molin & Givskov 1999; Riedel et al. 2001; Labbate et al. 2004).
AHL-based quorum sensing systems have been identified in many plant-associated bacteria. These systems are involved in the regulation of Ti plasmid transfer in Agrobacterium tumefaciens, the production of wall-degrading enzymes in the soft-rot phytopathogen Erwinia carotovora, as well as the production of virulence factors, antibiotics and biofilm formation (Pierson III, Wood & Pierson 1998; Eberl 1999).
In recent years, evidence has accumulated that AHL molecules are also perceived by eukaryotes, which in turn often specifically respond to these bacterial signals. AHLs were found to stimulate inflammatory and immunogenic responses in mice or to affect IL-8 production in human cells, suggesting that AHLs may act as virulence factors (Smith et al. 2001). Mathesius et al. (2003) showed that Medicago truncatula roots respond to AHL exposure by significant changes in the accumulation of over 150 proteins, depending on AHL structure, concentration and time of exposure. AHL-mimicking compounds and AHL inhibitory activities were reported in exudates from pea seedlings (Teplitski, Robinson & Bauer 2000; Gao et al. 2003).
The aim of this study was to investigate whether AHL can contribute to the biocontrol activity of typical rhizobacteria. For this purpose, AHL-producing and AHL-deficient strains were compared in terms of their effects on systemic resistance of tomato to the fungal leaf pathogen Alternaria alternata. AHL-producing rhizobacteria as well as AHL molecules were then examined for interference with plant signaling pathways. In addition, expression of plant defence genes was analysed after treatment with AHL molecules. The results suggest that tomato plants are able to detect the presence of typical rhizobacteria using AHL molecules, and to respond by systemic induction of defence genes.
MATERIALS AND METHODS
Plant material and growth conditions
Tomato plants (Lycopersicon esculentum Mill. cv. Micro-Tom, Bruno Nebelung GmbH, Everswinkel, Germany) were either grown in potting substrate or in 250 g autoclaved quartz sand inside sterile plastic containers (PhytaTray, Sigma, Taufkirchen, Germany). Fifteen PhytaTrays containing five plants each were placed in a tray covered with a transparent lid to reduce contamination. The plants were watered with nutrient solution according to Steidle et al. (2001). Prior to sowing, the seeds were surface-sterilized with 5% (w/v) sodium hypochlorite for 20 min. The plants were incubated under 14 h light [from 0600 h to 2000 h, photosynthetic photon flux density (PPFD) 120 mol m−2 s−1, 25 ± 1 °C]/10 h dark (20 ± 1 °C) cycles and 70 ± 5% relative humidity (Tuomainen et al. 1997). Middle-aged leaves of 4- (quartz sand culture) to 6-week-old plants (potting substrate) were harvested, immediately frozen in liquid nitrogen and stored at −75 °C.
Inoculation with bacteria and challenge with pathogenic fungi
Green fluorescence protein (GFP)-tagged S. liquefaciens strains MG1 and MG44 (Eberl et al. 1996) as well as Pseudomonas putida strains IsoF (Gfp) and F117 (DsRed) (Steidle et al. 2002) were grown overnight at 37 ° and 30 °C, respectively, in Luria-Bertani (LB) medium. After centrifugation at 6000 g, the bacteria were resuspended in 1/10 volume 10 mm MgSO4. Ten milliliters of this suspension ([1010 cfu mL−1)] was added to 5-week-old non-sterile plants by pipetting the bacteria around the shoot onto the soil substrate. Contamination of leaves and shoots was carefully avoided. To determine the colony forming units present in tomato roots, the potting substrate was removed using 1 × phosphate-buffered saline (PBS). Serial dilutions were plated on LB agar supplemented with 50 mg kanamycin L−1, using kanamycin resistance as marker. Cell numbers were determined by counting green fluorescing cells of gfp -labelled S. liquefaciens compared to the total cell numbers after 4′-6-diamidino-2-phenylindole (DAPI) staining. For this purpose, rhizosphere root samples were homogenized in 4% (w/v) sucrose solution, and 5–10 µL of this suspension was immobilized on polycarbonate filters (GTTP 0.2 µm, Millipore, Eschborn, Germany). The filters were stained in 7 × 10−5% DAPI for 10 min, gently washed in distilled water and then air-dried. Cell counting was conducted using an epifluorescence microscope (Axioplan2, Zeiss, Oberkochen, Germany).
Three days after the addition of bacteria, the plants were challenged with the leaf pathogenic fungus A. alternata. Fungal spores were suspended in 5.5 mm KHPO4, 62.5 mm glucose, 0.1% (v/v) Tween 20 (pH 6.0; modified after Leone & Tonneijk 1990) and sprayed onto the leaves (1000 spores mL−1). Leaf injury was determined independently by two persons using visual estimation of percentage necrotic leaf area.
Confocal laser scanning microscopy
Bacterial localization on the rhizoplane was studied using bacteria marked with GFP (green) or DsRed (red) fluorescent proteins. The chromosomal tagging of S. liquefaciens MG1 with GFP and P. putida IsoF with DsRed was described previously (Steidle et al. 2001). The GFP-based AHL monitor strain P. putida F117:Tn5-LAS was generated by random insertion of a transposon cassette based on components of the las quorum sensing system of Pseudomonas aeruginosa fused with the gfp gene (Steidle et al. 2001). Freshly harvested roots were washed with 1 × PBS and placed on a glass slide. After embedding in Citifluor (Citifluor Ltd, Canterbury, UK), the fluorescence was detected using a confocal laser scanning microscope (LSM510, Zeiss) (Hartmann et al. 1998).
Treatment with AHL and salicylic acid (SA)
Four-week-old plants grown in sterile quartz sand were treated with 1–10 µmN-BHL or HHL (Fluka, Seelze, Germany). A 100 × stock solution in methanol was diluted in 5 mL sterile water and pipetted on the sand to avoid contact with shoots and leaves. Control plants received equal amounts of methanol in water. Treatment with 0.5 mm SA (dilution of 100 × methanol stock in water) was conducted in the same way.
Extraction and quantification of free and conjugated SA were performed according to Meuwly & Métraux (1993). SA was detected using a Shimadzu RF 535 fluorescence detector (Shimadzu Europe, Duisburg, Germany) at excitation and emission wavelengths of 305 and 407 nm, respectively. The leaves from plants grown in non-sterile soil as well as shoots and roots from axenically grown plants were analysed.
Isolation of genomic DNA and quantification of fungal DNA in infected leaves
Fungal DNA of infected leaves was extracted from 100 mg homogenized plant material according to Bahnweg et al. (1998). The final DNA pellet was dissolved in 100 µL Tris-EDTA (TE buffer). A 1 µL part of this DNA solution was used for quantitative PCR (QPCR, real-time PCR), which was performed in 50 µL reactions with Absolute QPCR SYBR Green® ROX Mix (ABgene, Hamburg, Germany). A. alternata-specific primers (10 µm of each primer), A. alt-F3 (5′ to 3′ TCTAGCTTTGCTGGAGACTC) and A. alt.-R1.1 (5′ to 3′ AGACCTTTGCTGATAGAGAGT), derived from base-sequences of the internal transcribed spacers (ITS) of ribosomal RNA (rRNA) genes were used to generate short amplicons of 95 bp. The primers were specific for A. alternata and did not yield amplicons in PCR with DNA from 90 fungal strains, including all major tomato pathogens. The QPCR was performed with the ABI-PRISM 7700 Sequence Detection System, version 1.6 (Applera, Weiterstadt, Germany). Thermocycler conditions were 50 °C/2 min (elimination of uracil DNA), 95 °C/10 min (activation of enzyme), 40 cycles of 95 °C/15 s−60 °C/1 min (denaturation, annealing and amplification). A standard curve was generated for each run with a dilution series of purified and calibrated A. alternata genomic DNA (200 ng/20 ng/2 ng/200 pg/20 pg/2 pg/200 fg prepared in a 10 ng µL−1 DNA solution, the latter stabilizing highly diluted genomic DNA). The newly generated double-stranded DNA amplicon was quantified measuring the SYBR Green fluorescence intensity after each 60 °C step of the cycler program. Analysis of experiments [i.e. determination of Ct (cycle threshold) values to quantify initial amounts of target DNA] was conducted using the ABI-PRISM 7700 (Böhm et al. 1999).
Isolation of total RNA and Northern hybridization
For RNA isolation and Northern hybridization, leaf material from axenically grown plants was used after treatment with AHL molecules. Total plant RNA was isolated from 100 to 120 mg leaf material using the TRIZOL method according to the manufacturer's manual (Invitrogen, Karlsruhe, Germany). For Northern blot experiments, 5 µg total RNA was separated on a 1% (v/v) formaldehyde gel and transferred onto nylon membranes (Roche, Mannheim, Germany) by capillary blot (Sambrook & Russel 2001). The probes directed against tomato defence genes were cloned into the standard vector pGEM-T (Promega, Madison, WI, USA). The labelling was carried out in PCR reactions using the PCR Dig probe synthesis kit (Roche). Hybridization was performed overnight at 50 °C.
The array used in this study was developed in the course of the European Union (EU)-funded project TOMSTRESS, and will be described elsewhere. The 70 spotted PCR products were directed against tomato defence genes, genes of ethylene biosynthesis and perception, as well as genes involved in developmental processes such as ripening. Each probe was spotted in triplicate. Before using the filters in complex hybridizations, reference hybridization was performed to check the quality of the filters (Thimm et al. 2001). For this purpose, a 33P-labelled PCR product-specific primer (T7 oligonucleotide) was hybridized overnight at 38 °C. To remove radioactivity, the membranes were washed twice in 0.4 m NaOH for 30 min at room temperature and then neutralized in 2 × SSC.
The preparation of 33P-labeled complementary DNA (cDNA) using 4 µg total RNA was carried out with the Strip-EZ™ RT Kit (Ambion, Austin, TX, USA). Filters were pre-hybridized for at least 4 h at 65 °C in 5 × SSC, 5 × Denhardt solution, 0.5% (w/v) SDS and 100 µg mL−1 salmon sperm DNA. Hybridization of 33P-labelled cDNA was performed under the same conditions overnight. Arrays were then successively washed for 20 min at 65 °C in 20 × SSC, 0.1% SDS and 0.2 × SSC, 0.1% SDS. Radioactive signals were detected using a phosphoimager (FLA-3000, Fujifilm Europe, Dusseldorf, Germany). Images were directly imported into the ArrayVision software (Imaging Research Inc., St. Catharines, Canada). Normalization of the filters was achieved by dividing the signal intensity of each spot by the signal intensity of the mean of signal intensities of the whole filter. Each gene was spotted in triplicate, and the average of the three values was calculated. Then, the value for the treated sample was divided by the control value to obtain the induction factors. Only those signal intensities were considered, which were at least twice the background for treated sample or control. The methodical variability of macroarray experiments was determined by hybridizing three independent RNA samples extracted from the same plant material and ranged from 1 to 9% SE. The biological variability was investigated using three independent plant samples and was between 4 and 27%.
All experiments were performed at least twice with three replicate samples per test condition. The data shown are from representative experiments. When indicated, Tukey's multiple range test was used to determine differences among the treatment means (at P < 0.05; Statgraphics software, STSC, Inc., Rockville, MD, USA).
Systemic resistance to A. alternata is induced in tomato by AHL-producing rhizobacteria
When tomato plants (cv. Micro-Tom) were inoculated with the AHL-producing rhizobacterium S. liquefaciens MG1, the appearance of necrotic cell death after leaf infection with A. alternata was reduced by more than 70% 4 d after pathogen challenge (Figs 1a & 2a). In order to analyse the impact of AHL molecules in this interaction, an isogenic AHL-negative strain (MG44) was used in which the sole AHL synthase gene swrI was mutated by a gene replacement procedure (Eberl et al. 1996). Growth of the wild-type MG1 and the AHL-negative mutant MG44 in the rhizosphere was comparable during 9 d post-inoculation (Fig. 3). Inoculation with MG44 slowed down the development of A. alternata– induced cell death, but showed no significant alterations in response to the fungal pathogen when compared with non-inoculated controls after 4 d (Figs 1a & 2a).
Previously, we developed GFP-based monitor strains that allow in situ visualization of AHL-mediated communication among individual cells in the plant rhizosphere (Steidle et al. 2001). Based on constitutive GFP production under the control of an LacI-repressible promotor (Andersen et al. 1998), it could be shown that S. liquefaciens MG1 colonized the tomato rhizosphere in gnotobiotic systems (Fig. 1b) and in standard substrate (not shown). The bacteria were distributed uniformly over the root surface. The colonization behaviour of the AHL-negative mutant MG44 was similar (not shown).
AHL-mediated cell-to-–cell communication was demonstrated in situ between a GFP-based AHL monitor strain (P. putida F117:Tn5-LAS) and an AHL-producing bacterial isolate from tomato roots (P. putida IsoF) tagged with the red fluorescent protein DsRed (Steidle et al. 2002). P. putida F117 is the AHL-negative derivate of IsoF. Given that the sensitivity of the monitor strains is within 0.1–1 µm, AHLs are produced by the microbial community colonizing the tomato rhizosphere in biologically relevant amounts, and the AHL concentrations in the tomato rhizosphere are in the µm–mm range (Fig. 1c). More importantly, no signals were detected when the sensor strains were inoculated alone, excluding the possibility that a plant derived a signal activating the AHL monitor strain (not shown). No GFP-labelled or kanamycin-resistant rhizobacteria (neither S. liquefaciens nor P. putida) could be detected in the shoots and leaves of the tomato cv. Micro-Tom, suggesting that endophytic growth did not occur.
To determine whether growth of A. alternata in tomato leaves was reduced by pre-inoculation of roots with rhizobacteria, we quantified A. alternata DNA by real-time PCR. To this end, primers for A. alternata, which were based on ITS regions of ribosomal DNA genes, were developed. Plants pre-treated with S. liquefaciens MG1 exhibited 79% lower amounts of A. alternata DNA when compared with non-inoculated controls (Fig. 2b). In contrast, 4 d after infection, the amount of A. alternata DNA was comparable in controls and in plants that were inoculated with the AHL minus mutant MG44. Leaf injury data were positively correlated (r = 0.85) with the leaf-–associated amounts of A. alternata DNA.
Inoculation with the AHL-producing P. putida strain IsoF isolated from the tomato rhizosphere (Steidle et al. 2001) also resulted in a marked reduction of leaf damage following challenge with A. alternata. This strain produces at least four different 3-oxo-AHL molecules with acyl side chains varying from C6 to C12 with the aid of a single AHL synthase (Steidle et al. 2002). There was no significant difference in root colonization between the wild-type strain IsoF (3.5 ± 0.8 × 105 cfu g−1 fresh weight (FG)) and the AHL-negative mutant F117 (5.0 ± 2.5 × 105 cfu g−1 FG) after 7 d. The quorum sensing system of IsoF is suggested to be involved in the regulation of biofilm development as the AHL-negative mutant F117 formed a structurally aberrant biofilm relative to the one formed by the wild type. In contrast to S. liquefaciens, however, the isogenic AHL minus P. putida mutant F117 still reduced necrotic cell death to about 50% (Fig. 2c), suggesting that additional factors which are not regulated by quorum sensing compounds contribute to the biocontrol activity of P. putida IsoF.
AHL induction of SA production and systemic expression of defence-related proteins
To investigate whether root inoculation with AHL-producing rhizobacteria induces systemic plant responses, we measured the levels of the signal molecule SA as well as gene expression in middle-aged tomato leaves. Free, as well as conjugated, SA levels were elevated up to threefold within 3 d after inoculation with S. liquefaciens MG1 (Fig. 4a). In addition, levels of conjugated gentisic acid (2,5-dihydroxy benzoic acid), a signal molecule leading to a different pattern of pathogenesis-related (PR) proteins in tomato (Bellés et al. 1999), increased in the leaves following rhizobacteria inoculation by 50% (data not shown). The AHL-negative strain MG44 did not induce a significant increase in systemic SA levels in free or conjugated form (Fig. 4a). Root application of N-HHL, the major AHL species produced by S. liquefaciens MG1 (Eberl et al. 1996), at concentrations of 10 µm also led to sustained accumulation (three- to fourfold) of SA in tomato roots (Fig. 4b) and less-efficient accumulation in leaves (data not shown). HHL concentrations of at least 1 µm were necessary to affect SA levels in tomato (data not shown).
To further assess plant defence responses to the inoculation with AHL-producing rhizobacteria, we used a macroarray approach with 70 defence and signalling genes. The gene probes spotted on the array were directed against PR proteins and defence-associated genes, genes involved in jasmonate or ethylene biosynthesis and perception as well as signalling. In addition, genes involved in developmental and metabolic processes were analysed (see additional material).
As summarized in Table 1, root application of N-BHL and N-HHL for 5 h led to an enhanced systemic expression of specific genes involved in defence, the antioxidant system, ethylene signalling as well as in other functions in the leaves. Inoculati on with both AHL molecules induced marked expression of PR1a (P4), PR1 (P6), as well as 26 kDa acidic chitinase with induction factors between 4 and 18 (Table 1). A putative ascorbate peroxidase and the proteinase inhibitor CEVI57 were also induced by both molecules. Interestingly, the spectrum of responding defence genes was broader in response to HHL than to BHL. This included, in particular, a 30 kDa basic chitinase, which is ethylene responsive (Van Kan et al. 1995), P69, a subtilisin-like endoproteinase, the ethylene receptor Le-ETR1 as well as InvLp11, a vacuolar invertase.
Table 1. Macroarray analysis of defence genes in tomato leaves after application of 10 µm N-hexanoyl (HHL) and N-butanoyl (BHL) homoserine lactones to roots of plants growns in quartz sand. Leaves were harvested after 5 hours.
The systemic expression of PR proteins in the leaves 5 h after applying BHL and HHL to the roots was confirmed by Northern blot hybridization, and was compared to SA induction in terms of its kinetics. PR1a (P4) gene expression peaked after 5 h of BHL or HHL treatment (10 µm each), which was markedly faster when compared with SA induction (at 0.5 mm; Fig. 5). In line with this result, the expression of the 30 kDa chitinase was also induced after 5–10 h by BHL and HHL, while the response to SA was markedly lower (induced expression after 24 h). Control plants showed only minor expression of both PR1a (between 5 and 24 h) and the 30 kDa chitinase (after 5 h). The induction of the 30 kDa chitinase in control plants after 10–24 h is unresolved and may be a result of the elevated humidity in the aseptic plant system used.
AHL compounds in the tomato rhizosphere lead to systemic resistance against A. alternata
Using GFP-based monitor strains for in situ visualization of AHL-mediated communication among individual cells (Steidle et al. 2001), we could demonstrate that the ubiquitous rhizobacterium P. putida produces AHL molecules in the tomato rhizosphere. The coinoculation experiments and the application of pure AHL compounds were performed on roots of gnotobiotically grown plants, which excludes contamination with other AHL-producing or metabolizing bacteria. As AHL signal molecules are produced and recognized by numerous microorganisms (Pierson III et al. 1998), cross-communication between bacterial populations in complex communities can occur. In addition, root-associated bacteria produce AHL molecules more often than soil-borne bacteria (Pierson III et al. 1998; Elasri et al. 2001; Steidle et al. 2001).
Laser scanning microscopy of red fluorescent ‘AHL producer’ and green fluorescent ‘AHL sensor’ strains revealed that the bacteria conducted AHL-dependent cell-to-cell communication while colonizing the rhizoplane (Fig. 1c). Based on the sensitivity of the sensor strain, it was estimated that the indigenous rhizosphere community produces AHL in overall concentrations ranging from 0.1 to 10 µm, while local hot spots (i.e. microcolonies) might be present with much higher AHL concentrations. No signals were detected when the sensor strains were inoculated alone in aseptically grown tomato, eliminating the possibility that plant-derived signals activated the AHL monitor strain.
Systemic pathogen resistance was analysed with A. alternata, a member of the genus Alternaria, which includes widespread necrotrophic pathogens causing rot diseases on leaves and fruits of a variety of important agronomic host plants including tomato, potato, cauliflower and citrus fruits (Thomma 2003). The analysis showed a clear dependence on the production of AHL molecules by the rhizobacterium S. liquefaciens MG1. Tomato plants inoculated with the isogenic AHL-negative mutant MG44 were not significantly different from control plants with regard to susceptibility (Figs 1a & 2a). The contribution of AHL molecules to enhance plant defence responses was confirmed with another AHL-producing strain, P. putida IsoF, an isolate from tomato roots (Steidle et al. 2001), which also induced resistance against A. alternata. However, the difference between wild-type and AHL-negative strain was not as clear-cut as in the case of S. liquefaciens although there was no difference in root colonization. Probably, other bacterial determinants such as specific lipopolysaccharides or flagellin structures (Duijff, Gianinazzi-Pearson & Lemanceau 1997; Felix et al. 1999) were effective in eliciting biocontrol activity in addition to the AHL production by P. putida IsoF.
Both visible damage and fungal DNA amounts of A. alternata were reduced after pre-inoculation with S. liquefaciens MG1 (Fig. 2). Because direct antagonism between rhizobacteria and fungal pathogens is well described (Berg et al. 2002), it was important to ensure that biocontrol active bacteria and the pathogen remained spatially separated in our studies. Because GFP-labelled and kanamycin-resistant derivatives of S. liquefaciens MG1 or P. putida IsoF could not be detected in leaves and shoots, we propose that the plant defence machinery is activated by bacterial components or metabolites in the roots and by subsequent signal translocation to the above-ground parts of the plant.
The role of SA in AHL-induced resistance
Resistance in plants is a highly regulated phenomenon depending on several signalling pathways, each activated by a set of different biotic and abiotic stimuli (Pieterse & van Loon 1999). SA, one of the best characterized signal molecules, seems to be involved in systemic resistance against A. alternata induced by S. liquefaciens MG1 because elevated levels were systemically measured in leaves after inoculation with the bacteria. In addition, conjugates of gentisic acid were induced by rhizobacteria inoculation as well. Gentisic acid is also pathogen-inducible, and stimulates a pattern of PR proteins in tomato, which is different to that induced by SA (Bellés et al. 1999), thereby putatively complementing the response.
Many rhizobacteria of the genus Pseudomonas or Serratia produce SA under iron-limiting conditions as a precursor of siderophores (De Meyer & Höfte 1997; Press et al. 1997; Mercado-Blanco et al. 2001). Resistance of tomato or tobacco against Botrytis cinerea induced by the biocontrol strain P. aeruginosa 7NSK2 depends on in planta accumulation of SA as revealed by experiments with transgenic nahG plants, which are not able to accumulate SA (De Meyer, Audenaert & Höfte 1999; Audenaert et al. 2002). S. liquefaciens MG1 produces picomolar amounts of SA in liquid culture (data not shown), and these low amounts may potentially contribute to induce systemic resistance in plants. However, as the AHL-deficient mutant MG44 failed to effectively induce systemic resistance (Figs 1a & 2a) and did not result in increasing SA levels in leaves (Fig. 4a), a major contribution of SA of bacterial origin can be ruled out in our system. In addition, AHL application to tomato roots led to an induction of SA in roots and, to a lower extent, in leaves (Fig. 4b, data not shown). Therefore, we suggest that AHL-induced SA production is an additional new route to systemic signal transduction in tomato. The possibility that AHL can activate the SAR pathway by damaging cells and tissues can be excluded, as tomato cell suspension cultures were not affected in viability by AHL molecules at the concentrations used. Furthermore, roots and shoots of AHL-treated plants showed no signs of tissue damage. Further experiments with NahG tomato plants should reveal whether MG1 is still able to induce resistance.
Alterations in gene expression induced by AHL molecules
To test whether AHL signal molecules could lead to systemic changes in gene expression, tomato leaves were analysed by macroarray and Northern blot analysis. As shown in Table 1, the expression of several PR proteins was altered by BHL and HHL, two short chain AHL molecules produced by S. liquefaciens MG1. The spectrum of genes induced by HHL was broader than that induced by BHL. Both AHL molecules induced isoforms of PR1. In tobacco, acidic and basic isoforms are differentially regulated during defence responses (Uknes et al. 1993). PR1a in tobacco and A. thaliana is considered as a marker of the SA-dependent defence pathway SAR (Durner, Shah & Klessig 1997). In our studies, expression of the tomato isoform PR1a (P4) was up-regulated very rapidly following HHL and BHL application within the first 5 h, and more slowly after the addition of SA. This may be a result of the adsorption of SA to the soil substrate, thereby lowering its active concentration. Genes encoding other PR proteins that were induced by AHL were chitinases and an acidic β-1,3-glucanase. In tomato, acidic and basic isoforms of chitinases are differentially regulated. The 26 kDa acidic chitinase is induced by SA, whereas the 30 kDa basic chitinase is considered ethylene-dependent because it responds to ethephon, an ethylene-releasing compound (Van Kan et al. 1995).
Our results suggest that AHL molecules play an important role in the biocontrol activity of S. liquefaciens and other rhizobacteria in tomato, and act as mediators of communication between prokaryotes and eukaryotes. In line with these findings, Mathesius et al. (2003) reported specific responses of more than 150 proteins in roots of M. truncatula exposed to nm–µm micromolar concentrations of long chain (C12 to C16) AHL molecules. Even though SA contents were up-regulated systemically by inoculation with AHL-producing rhizobacteria, the pattern of AHL-induced defence genes was not confined to salicylate-dependent genes. In addition, the accumulation of specific transcripts depended on the AHL species. Hence, it may be speculated that tomato plants are able to perceive AHL molecules with yet unidentified receptor-based recognition mechanisms.
Experiments on producing AHLs using transgenically modified plants showed different effects on susceptibility after challenging tobacco and potato plants with soft-rot erwiniae (Mäe et al. 2001; Toth et al. 2004). It was found that AHLs derived from modified plants are able to interfere with quorum sensing regulated traits of pathogenic bacteria. It would be interesting to analyse the resistance of these plants against other pathogens, which do not use AHL-dependent quorum sensing systems, and to investigate their signalling systems for systemic responses in order to find out whether or not these plants are more resistant to plant pathogens.
The AHL-induced resistance by S. liquefaciens MG1 in tomato plants differs from that induced by P. fluorescens WCS417r (Van Loon et al. 1998) in several respects. Firstly, P. fluorescens WCS417r does not produce AHLs in detectable amounts. Secondly, ISR in A. thaliana is not accompanied by elevated levels of plant signalling molecules, whereas SA contents are systemically induced by S. liquefaciens MG1 and AHL molecules. Thirdly, root application of two AHL molecules led to marked systemic inductions of defence gene expression, especially of pathogenesis-related proteins, whereas gene expression of defence-related proteins is not markedly altered in response to ISR-inducing rhizobacteria (Van Loon et al. 1998). Preliminary results show that AHLs are also effective in systemic induction of pathogen resistance in barley (Uta von Rad et al., unpublished results). AHLs may therefore be considered as potential candidates for a new group of general elicitors for plant defence because they are produced in the tomato rhizosphere in effective amounts, induce SA systemically, lead to enhanced gene expression for typical defence-related proteins and result to increased resistance against fungal pathogens. Interactions of this AHL-induced resistance with ISR and SAR require further analysis in future studies.
We thank Lucia Gößl, Rosina Ludwig and Judith Mayr for expert technical assistance. We gratefully acknowledge our colleagues for critically reading our manuscript. This research was supported by grants from the Federal Ministry for Research and Education (BMBF), the EU-FAIR project TOMSTRESS and partly by the Fonds der Chemischen Industrie. This manuscript is dedicated to Prof. H. Sandermann on the occasion of his 65th birthday.