Molecular communication in the rhizosphere

Abstract

This paper will exemplify molecular communications in the rhizosphere, especially between plants and bacteria, and between bacteria and bacteria. More specifically, we describe signalling pathways that allow bacteria to sense a wide diversity of plant signals, plants to respond to bacterial infection, and bacteria to coordinate gene expression at population and community level. Thereafter, we focus on mechanisms evolved by bacteria and plants to disturb bacterial signalling, and by bacteria to modulate hormonal signalling in plants. Finally, the dynamics of signal exchange and its biological significance we elaborate on the cases of Rhizobium symbiosis and Agrobacterium pathogenesis.

Introduction

This paper will exemplify molecular communications in the rhizosphere, especially between plants and bacteria, and between bacteria and bacteria. More specifically, we describe signalling pathways that allow bacteria to sense a wide diversity of plant signals, plants to respond to bacterial infection, and bacteria to coordinate gene expression at population and community level. Thereafter, we focus on mechanisms evolved by bacteria and plants to disturb bacterial signalling, and by bacteria to modulate hormonal signalling in plants. Finally, the dynamics of signal exchange and its biological significance we elaborate on the cases of Rhizobium symbiosis and Agrobacterium pathogenesis. For a complete overview of communication in the rhizosphere, we recommend other papers that illustrate plant-plant interactions, and that give additional insights about nitrogen-fixing microorganisms, plant-driven selection of microbes, plant growth promoting microorganisms, and plant pathogens.

Communication in the rhizosphere: mechanisms and functions

How bacteria sense plant signals

The bulk soil is generally a very poor, nutrient-diluted and therefore hostile environment in which nutrient bioavailability is often hampered by the soil biochemistry. Within this nutritional desert, the presence of plant roots provides the means for the formation of true oases with flourishing microbial populations because all roots have the ability to actively secrete low- and high-molecular-weight molecules into the rhizosphere. Root exudation is largely mediated by the root hairs, but also the root cap and apical epidermal cells make a significant contribution. These actively secreted compounds are composed of excretions—waste products from the plants’ internal metabolic processes without any identified function—and secretions—a mixture of compounds that facilitate external processes like lubrication or nutrient acquisition. Moreover, root growth is accompanied by sloughing-off of living cells, senescence, cell wounding and leakage from plant cells which represent more passive release mechanisms of diverse components that nonetheless are very important for the provision of carbon in the soil. The compounds released by these processes are termed mucilages and exudates, respectively. Finally, the microbial community actively participates in defining the composition of the rhizosphere by degrading and secreting complex organics compounds, and by lysing plant cells. These types of molecules are part of mucilages and lysates, respectively (Bertin et al. 2003; Somers et al. 2004). The whole of these root-associated components accumulating in the rhizosphere is termed rhizodeposit and it has a large impact on plant growth and soil ecology. Rhizodeposition is a dynamic process that is developmentally regulated and varies with the plant species and cultivar; it is also altered upon biotic and abiotic stress. Moreover, the microbial community influences the composition of the exudates to its advantage (Yang and Crowley 2000; Paterson et al. 2006; Shaw et al. 2006; Yoneyama et al. 2007).

From the above it is clear that root exudates are complex molecular mixtures and in Table 1 the diversity of molecules identified in rhizodeposits is illustrated. Generally, rhizodeposition is involved in primary and secondary plant metabolic processes, nutrient and water acquisition, plant defence and stimulatory or inhibitory interactions with other soil organisms (Bertin et al. 2003). However, depending on their relative abundance, the different components of rhizodeposits also affect the soil microorganisms. It is not difficult to envision that many of these compounds are chemoattractants and welcome nutrients for the microbes living in or nearby the rhizosphere (Somers et al. 2004; Brencic and Winans 2005). Whereas many micro-organisms can only utilise rather general plant metabolites, some bacteria have the capacity to catabolize certain plant secondary metabolites providing a selective advantage to colonize the rhizosphere of specific plants (Savka et al. 2002). Examples of such nutritional mediators are glycosides and aryl-glycosides (Faure et al. 1999, 2001), calystegin (Tepfer et al. 1988; Guntli et al. 1999), certain flavonoids (Hartig et al. 1991), proline (Jiménez-Zurdo et al. 1997), 1-aminocyclopropane-1-carboxylic acid (Penrose and Glick 2001), and homoserine and betaines (Boivin et al. 1990; Goldmann et al. 1991). Another well described effect of often unidentified components of rhizodeposits is the activation of bacterial gene expression culminating in more or less intimate interactions with the producing plant host (Stachel et al. 1985; Koch et al. 2002; Brencic et al. 2005; Brencic and Winans 2005; Cooper 2007; Reddy et al. 2007; Franks et al. 2008; Johnston et al. 2008). Recent genome-wide studies have shown that root exudates modulate the expression of a significant number of bacterial genes of which the function in rhizosphere colonisation and competitiveness had not been anticipated (Mark et al. 2005; Matilla et al. 2007; Yuan et al. 2008b). Moreover, plants have been shown to secrete components that interfere with quorum sensing (Teplitski et al. 2000; Dunn and Handelsman 2002; Gao et al. 2003), a cell-cell signalling mechanism in bacteria that is very important in group-coordinated processes that impact interactions with Eukaryotes (von Bodman et al. 2003; Waters and Bassler 2005). In order to trigger these diverse molecular, physiological and behaviour responses, soil bacteria first have to sense the presence of the root exudates via one- and two-component signal perception systems.

Table 1 Organic compounds and enzymes released by plants in root exudates and their function in the rhizosphere

A widespread mechanism by which bacteria sense their environment and respond accordingly is the two-component system which is typically comprised of a usually membrane-bound sensor histidine protein kinase and a response regulator most often mediating differential gene expression. The structural genes are frequently organised as an operon, although many orphan kinases have been detected in the available bacterial genomes, and both encoded proteins consist of at least two domains. Via an N-terminal input domain the sensor perceives a specific stimulus, which upon interaction results in a conformational change of the cytoplasmic transmitter domain resulting in autophosphorylation at a conserved histidine residue. The activated transmitter domain will then in turn activate the N-terminal receiver domain of the cognate response regulator by phosphotransfer to a conserved aspartate residue. Next, the activated response regulator will mediate a cellular response via its C-terminal effector or output domain mainly by protein-protein interactions or protein-DNA interactions (see Fig. 1). Finally, dephosphorylation of the response regulator brings the system back to the pre-stimulus state (Laub and Goulian 2007). A common variation on this prototypical two-component system is the phosphorelay in which the histidine kinase has an additional receiver-like domain. Such hybrid histidine kinases will, upon signal perception and subsequent autophosphorylation, transfer their phosphoryl group intramolecularly to an aspartate residue in their receiver domain. This phosphoryl group is then transferred to a histidine residue of a cytoplasmic histidine phosphotransferase which finally shuttles it to the aspartate residue of the terminal response regulator (see Fig. 1a; Hoch and Varughese 2001). Whereas the histidine kinase domains of the sensors and the receiver domains of the response regulators comprise paralogous gene families that share considerable sequence and structural similarity, their input and output domains vary extensively although many conserved modules have been identified (Galperin 2006; Mascher et al. 2006; Szurmant et al. 2007). Classification of the sensor histidine protein kinases based on their domain architecture, reflecting the mechanism of sensing and signal transduction, revealed three major groups: the periplasmic- or extracellular-sensing histidine kinases, histidine kinases with sensing mechanisms linked to transmembrane regions, and cytoplasmic-sensing histidine kinases (see Fig. 1b; Mascher et al. 2006). The first class is the largest and signal detection occurs directly via binding of a small molecule to the sensor domain, indirectly through interaction with a periplasmic solute-binding protein, or via a conformational change of the input domain after a mechanical or electrochemical stimulus. The hybrid histidine kinases of phosphorelay systems belong to this class. Typically, the periplamic sensor kinases recognise solutes and nutrients, and so are part of many two-component systems involved in rhizosphere sensing. The second and smallest class of sensor kinases lack elaborate extracellular input domains and rely mainly on their transmembrane helices for perception of stimuli that are either associated with the membrane or occur within the membrane interphase. The third class groups the cytoplasmic-sensing histidine kinases which can be membrane-anchored or soluble and detect diffusible or internal stimuli (Mascher et al. 2006). A structural classification of the response regulators based on their domain architectures resulted in six major types that reflect their functionality: stand-alone receiver domains, and receiver domains combined with DNA- or RNA-binding, enzymatic, protein- or ligand-binding and uncharacterized output domains (see Fig. 1c; Galperin 2006). The transcriptional regulators with a DNA-binding output domain encompass 75% of all response regulators and typically have a important role in rhizosphere signal transduction.

Fig. 1

Different signal perception and transduction mechanisms in bacteria. a Overview of the signal transduction systems; b Classification of the sensor histidine protein kinases: the periplasmic- or extracellular-sensing histidine kinases (left), histidine kinases with sensing mechanisms linked to transmembrane regions (middle), and cytoplasmic-sensing histidine kinases (right); c Classification of the response regulators: stand-alone receiver domains (left), and receiver domains combined with an output domain (right). H, histidine residue; D, aspartate residue; P, phosphotransfer; HPT, histidine phosphotransferase; lightening flash, incoming signal

Although two-component systems have been considered as the paradigm signal perception and transduction systems in prokaryotes, large scale genome analyses have recently shown that a bacterial cell contains a plethora of the much simpler one-component systems (Ulrich et al. 2005). Typically, these systems are single proteins that contain input and output domains, but lack the phosphotransfer histidine kinase and receiver domains (see Fig. 1a). Another type of one-component systems resembles fusions of classical histidine kinases with full-length response regulators and consists of single proteins with input, transmitter, receiver and output domains (see Fig. 1a; Galperin 2006). The repertoire of input and output domains in one-component systems is much more diverse than in two-component systems, with many domains unique for the one-component systems. This finding suggests that one-component systems likely perceive similar stimuli and elicit similar responses as two-component systems; their greater variability however is related to their extensive involvement in cytoplasmic sensing (Ulrich et al. 2005).

The list of one- and mainly two-component systems involved in plant recognition by rhizospheric bacteria is obviously extensive and several of them will be described in detail throughout this chapter. The following examples illustrate that almost every class of sensors and response regulators are involved in rhizosphere sensing. The periplasmic-sensing histidine kinase VirA of Agrobacterium tumefaciens recognises acidic pH, phenolic compounds, and monosaccharides (the latter via the periplasmic sugar-binding protein ChvE) released by wounded plant cells and its cognate DNA-binding response regulator VirG activates vir gene expression initiating T-DNA transfer (Mukhopadhyay et al. 2004). The GacS hybrid histidine kinase of many proteobacteria recognises environmental signals and activates the transcription factor GacA that controls for instance the biosynthesis of extracellular enzymes and secondary metabolites involved in virulence (Heeb and Haas 2001). The CbrAB system of Pseudomonas aeruginosa senses the intracellular carbon/nitrogen ratio via the transmembrane sensor CbrA and CbrB adjusts its catabolism by modulating expression of catabolic operons (Nishijyo et al. 2002); in Pseudomonas putida, a CbrAB system is involved in the degradation of IAA (Leveau and Gerards 2008). The membrane-associated cytoplasmic sensor kinase FixL and its cognate response regulator FixJ mediate O₂-controlled gene expression in root-nodulating bacteria (Gilles-Gonzalez and Gonzalez 2004). The soluble sensor CheA with its response regulator CheY controls chemotaxis in many bacteria via protein-protein interactions (Szurmant and Ordal 2004). The best described one-component system implicated in perception of the plant is likely the cytoplasmic possibly membrane-associated NodD protein of rhizobia (Brencic and Winans 2005). It is a LysR type transcription factor that perceives flavonoids and then activates transcription of the nod genes that encode the biosynthesis of the lipochito-oligosaccharide Nod factor (Peck et al. 2006).

Although many signal transduction systems have been described and keep on being identified via genome-wide approaches (Mascher et al. 2006; Qian et al. 2008), a lot of the mechanistic details and the identity of many of the primary stimuli remain to be uncovered. Nevertheless, from the above it is clear that the simple and exchangeable modular design of one- and two-component systems combined with extensive cross-regulation permits bacteria to perform sophisticated information processing allowing them to survive in the dynamic rhizosphere environment.

How plants sense bacteria

Plants evolved complex and diverse mechanisms to sense and respond to bacterial presence. Morphogens such as cytokinines, auxins and Nod factors, can profoundly affect plants. Plants sense Nod factors via receptor kinases of the LysM family (primary Nod factor receptor MtLYK4/NFP, and secondary Nod factor receptor or entry receptor MtLYK3/HCL), upon which a complex signal transduction cascade is triggered involving other extracellular-domain-containing receptors (Jones et al. 2007 and references therein). Plants also respond to presence of bacterial quorum-sensing signals, but the mechanism involved is still unknown (Mathesius et al. 2003; Schuhegger et al. 2006; von Rad et al. 2008). However pathogen-associated molecular patterns (PAMPs) such as lipopolysaccharides (LPS), peptidoglycan, and abundant proteins like the translational factor EF-tu and flagellin, are perceived via specific receptors, the pattern recognition receptors (PRRs). Perception of PAMPS constitutes the primary immune response of plants and is referred as PAMP-triggered immunity (PTI) (Zipfel 2008). Regulatory cascades implicating several classes of kinases activate the PTI. Until today, FLS2 and EFR are the only known PRRs in Arabidopsis; other examples of plant PRRs are very scarce (Zipfel 2008). However, the genome of Arabidopsis possesses numerous (hundreds) potential PRRs (Schwessinger and Zipfel 2008). The availability of new genomic resources and novel tools should enable the discovery of additional PRRs from crop species.

In the co-evolution of host-microbe interactions (Chisholm et al. 2008), pathogens acquired the capacity to suppress PTI by interfering with recognition at the plasma membrane or by secreting effector proteins into the plant cell cytosol that alter resistance signalling and PTI. Remarkably, the ability to deliver proteins directly into plant host cells is a common feature among phytopathogens. Bacterial effectors that are released into plant cells can possess enzyme activities, such as proteases and phosphatases, which are responsible for modifying host protein to enhance pathogen virulence and evade detection. Some other effectors are protein chaperones protecting the pathogen itself from these potentially detrimental enzymatic activities or keeping the effector protein unfolded prior to secretion.

In response to the delivery of pathogen effector proteins, plants acquired surveillance proteins (R proteins) to recognize and either directly or indirectly block or modify the properties of bacterial effectors. This response constitutes the secondary immune response of plants and is referred to as effector-triggered immunity (ETI). The connection between PTI and ETI is an emerging field of research. Furthermore, the role of small RNAs in immunity and that of PTI in symbiosis are valuable areas to investigate.

How bacteria sense bacteria

Bacteria have evolved sophisticated mechanisms to coordinate gene expression at population and community levels via the synthesis and perception of diffusible molecules. Because the concentration of the emitted signal in a confined environment reflects the bacterial cell number per volume unit (commonly cell density), such a regulatory pathway was termed quorum sensing (QS) (Fuqua et al. 1994). In an open environment, however, the concentration of the signal reflects both the bacterial cell number and the signal diffusion coefficient. In such open environments, the term diffusion sensing was proposed (Redfield 2002). A recent tentative to unify quorum and diffusion sensing states that the perception of a signal by a cell (efficiency sensing) is modulated by three essential factors: cell density (quorum sensing), mass-transfer properties (diffusion sensing), and spatial distribution of the cells (Hense et al. 2007). However, additional environmental factors may directly modify the synthesis rate and stability of the signals in the rhizosphere, which will be discussed in a latter paragraph.

The nature of QS signals is highly diverse (Schaefer et al. 2008; Whitehead 2001). Oligopeptides and substituted gamma-butyrolactones have been described in Gram-positive bacteria, while other substituted gamma-butyrolactones, the N-acyl-homoserine lactones (AHLs), are synthesized by a large number of Gram-negative bacteria. In this latter bacterial group, 3-hydroxypalmitic acid methyl ester (Flavier et al. 1997), 3,4-dihydroxy-2-heptylquinoline (Holden et al. 1999), and a furanosyl borate diester (Chen et al. 2002) can also act as QS signals. The most studied QS signals among rhizobacteria are AHLs (Whitehead et al. 2001). The synthesis of AHL depends upon synthases generally belonging to two classes: the LuxI and the AinS homologs. The perception of the signal relies on a sensor protein, a LuxR homolog, which is also the transcriptional regulator controlling the expression of QS-regulated genes.

The rhizosphere is potentially favorable for QS signalling, because it is a spatially structured habitat that is colonized, at a high cell density, by diverse bacterial populations. Experimental evidence supports this assertion. Ten to twenty percent of the cultivable bacteria in soil and rhizospheric environments are AHL-producing (D’Angelo-Picard et al. 2004). They are able to communicate both at the intra- and inter-species level (Steidle et al. 2001, 2002). Moreover, AHL signalling is implicated in the manifestation of plant-associated phenotypes in pathogenic, symbiotic, and biocontrol bacterial strains. The functions controlled by QS are highly diverse, including the horizontal transfer of plasmids, and the regulation of rhizospheric competence factors such as antibiotics, as well as functions that are directly implicated in plant-bacteria associations, such as virulence factors (Whitehead et al. 2001).

The AHL QS-signals show variations in the length and side chains of a core structure, and each AHL receptor can recognize a specific AHL structure. Even though some correlation exists between the genetic position of a strain or a group of strains and their AHL production patterns (D’Angelo et al. 2005), most AHL profiles are not strictly conserved at the genus or species level. Indeed, some phylogenetically distant species exhibit similar AHL profiles, supporting inter-species communication. Several explanations may account for this phenomenon. At the molecular level, the amino acid sequences of the AHL synthases are sometimes more distant within one species than between distinct species (Gray and Garey 2001). At the ecological and evolutionary levels, the presence of multiple AHL synthase homologues in species such as in Rhizobium leguminosarum (González and Marketon 2003) and the fact that multiple luxI-luxR determinants in a bacterium may be acquired independently (Gray and Garey 2001), can explain the occurrence of these complex patterns of AHLs. As an example, in the genus Rhizobium, some strains produce a single AHL, while others synthesize several AHLs (González and Marketon 2003). Such heterogeneity within AHL profiles may result from a selective pressure that tends to stimulate the emergence of distinct molecular languages at sub-species level, especially when related organisms share common ecological niches. An alternative explanation calls for another selective pressure that would authorize inter-species cooperation. One can not exclude the possibility that bacterial populations use distinct communication pathways to discriminate different levels of genetic proximity (clone, population and community). One of the QS signals facilitating communication at community level would be a furanosyl borate diester (AI-2) that is synthesized and recognized by a large range of Gram-positive and Gram-negative bacteria (Chen et al. 2002). The multiplicity of QS-signals, their interconnection and their modulation by environmental factors, especially the plant host, as well as spatial and temporal constraints remain to be elaborated.

How bacteria and plants interfere with bacterial signals

Bacteria and plants, as well as their genetically modified derivatives generated for research and biotechnological purposes, can produce QS-signal biomimics or QS-interfering molecules, including QS-signal modifying enzymes (Dong et al. 2007). The term quorum quenching (QQ) encompasses various natural phenomena or engineered procedures that lead to the perturbation of the expression of QS-regulated functions.

QS-biomimics were discovered in plants and in bacteria; their function is still speculative (McDouglas et al. 2007). In contrast, numerous reports evaluated QQ mechanisms, their function in vivo, and their potential agricultural applications (Dong et al. 2007). The three main steps of QS regulation that seem to be targeted are signal synthesis, and the much better described signal stability and sensing. For instance, the red algae Delisea pulchra limits bacterial colonization (fouling) of its lamina by interfering with the QS-controlled motility and biofilm-formation ability. This process is mediated by halogenated furanones produced by the algae that bind the bacterial LuxR receptor, prevent the binding of or displace the AHL signal, and thereby accelerate the degradation of the LuxR protein (Rasmussen and Givskov 2006). Other inhibitors have been found in plants such as pea and soybean, Medicago, fruit extracts such as those from grape and strawberry, garlic, vanilla, lily and pepper, Clematis vitalba, Geranium molle, and Tropaeolum majusi (Rasmussen and Givskov 2006). Fungi such as Penicillum species also produce inhibitors of QS, identified as the lactones patulin and penicillic acid (Rasmussen et al. 2005). Interestingly, patulin naturally occurs in fruits such as apple, pear, peach, apricot, banana, pineapple, and grape, where the compound may also contribute to the inhibition of QS. The impact of these molecules on the behavior of rhizobacteria remains to be clarified. Aside from the investigations on natural inhibitors, efforts have been made to identify or design chemical compounds that may target the LuxR-like receptor(s). Most of the designs are based on actual AHL structures and analogues with either activating or inhibitory activity have been identified (Reverchon et al. 2002).

QS-signals are subject to enzymatic degradation. The AHL- lactonases catalyze a reaction that is identical to pH-mediated lactonolysis (opening the gamma-butyrolactone ring), while acylases/amidohydrolases convert AHL to homoserine lactone and a fatty acid. These enzymatic activities were observed in bacteria such as Variovorax (Leadbetter and Greenberg 2000) and Bacillus (Dong et al. 2000). Since these pioneer reports, numerous bacteria inactivating AHLs have been identified (Faure and Dessaux 2007). Some dissimilate AHL, i.e. use these substrates as growth substrates, and some do not (Leadbetter and Greenberg 2000; Uroz et al. 2003). To date, AHL inactivation has been described in α-proteobacteria (e.g. Agrobacterium, Bosea, Sphingopyxis and Ochrobactrum), β-proteobacteria (e.g. Variovorax, Ralstonia, Comamonas, and Delftia), and γ-proteobacteria (e.g. Pseudomonas and Acinetobacter). AHL inactivation also occurs in Gram-positive strains, both amongst low-G+C% strains or firmicutes such as Bacillus and high-G+C% strains or actinobacteria, e.g. Rhodococcus, Arthrobacter, and Streptomyces. Rhodococcus erythropolis has lactonase and acylase activies, as well as an oxidoreductase that converts 3-oxo-AHL to 3-hydroxy-AHL, which represents a different AHL-modifying activity that is not sensu stricto degrading (Uroz et al. 2005, 2008). Since the substitution at C3 is crucial for signal specificity, the oxidoreductase leads to a change in or loss of the signaling capability of the QS molecules. Aside from bacteria, AHL-degradation abilities have also been observed in animals (Chun et al. 2004) and plants (Delalande et al. 2005).

Several authors have proposed to take advantage of quenching to develop novel medical and animal therapies or novel biocontrol strategies for plant pathogens (Dong et al. 2007; Rasmussen and Givskov 2006). QQ applications therefore fall into the family of anti-virulence/anti-disease strategies. QQ-enzymes may be also used to identify the QS-regulated functions in bacteria (Smadja et al. 2004). For agricultural developments, the frequently proposed strategies imply the degradation of QS signal by plants and bacteria. They are illustrated by the following examples: (i) plants, which are genetically modified to gain the capacity to inactivate AHL because they express the AHL-lactonase AiiA of Bacillus, were more resistant to Pectobacterium carotovorum infection than the parental, wild-type plants (Dong et al. 2001); (ii) QQ bacteria were proposed as biocontrol agents to interfere with the virulence of plant pathogens (Uroz et al. 2003); (iii) chemicals that either directly interfere with QS-signalling or stimulate the growth of QQ-bacteria in the treated rhizosphere (Cirou et al. 2007). All QQ strategies were developed in vitro or under greenhouse conditions, so their efficiency in the field remains to be evaluated. However, QQ strategies may also prevent QS-regulated functions in plant benefic bacteria, such as antifungal synthesis by biocontrol strains (Molina et al. 2003).

How bacteria can interfere with plant hormones

Plant hormones control plant growth and development by acting as signal molecules. They affect the spatial and temporal expression of various phenotypes such as plant cell elongation, division, and differentiation. In addition, they play an important role in a plant’s response to biotic and abiotic stresses. Several plant-associated bacteria have evolved ways to tap into these hormone signalling pathways and to manipulate plant physiology accordingly and to their own advantage. One such way is stimulation of hormone synthesis by the plant itself. For example, the pathogenic bacterium Pseudomonas syringae pv tomato DC3000 is able to induce the biosynthesis of the hormones auxin (Schmelz et al. 2003) and abscisic acid (de Torres-Zabala et al. 2007) in Arabidopsis thaliana. Another intriguing example is the ability of bacterial quorum sensing molecules such as AHLs to downregulate auxin-induced genes (Mathesius et al. 2003). A different and well-known type of bacterial manipulation of plant hormone levels is the transfer, integration and expression of bacterial DNA coding for the biosynthesis of auxin and cytokinin in plant tissues, as described for Agrobacterium tumefaciens and A. rhizogenes (Francis and Spiker 2005).

Another route for exploitation of the plant hormone system is through bacterial synthesis or degradation of plant hormones (Costacurta and Vanderleyden 1995; Patten and Glick 1996; Tsavkelova et al. 2006; Spaepen et al. 2007; Glick et al. 2007). Table 2 shows examples for the five classical plant hormones (Kende and Zeevaart 1997), i.e. auxin (indole 3-acetic acid or IAA), ethylene, abscisic acid (ABA), cytokinin (zeatin) and gibberellin (gibberellic acid or GA). As is clear from the table, every one of these hormones can be synthesized and/or degraded by bacteria. Obviously, our understanding of the pathways, genes, and enzymes underlying bacterial synthesis and/or degradation is biased towards what is known about a small number of intensively studied cases. These include the synthesis of IAA (Patten and Glick 1996; Spaepen et al. 2007) and the activity of 1-aminocyclopropane-1-carboxylate (ACC) deaminase, which lowers ethylene concentrations through degradation of the ethylene precursor ACC (Glick 2005; Glick et al. 2007). Much less is known about other activities, such as the phenomenon of bacterial IAA degradation which has long been recognized but until recently (Leveau and Lindow 2005) did not receive serious attention as a means by which bacteria might affect plant physiology. Only very recently the first bacterial genes for IAA degradation were discovered in a Pseudomonas putida species (Leveau and Gerards 2008).

Table 2 Bacterial synthesis and degradation of plant hormones

Many bacteria are capable of producing more than one type of plant hormone (Boiero et al. 2007; Karadeniz et al. 2006). Moreover, some bacteria can produce and degrade the same hormone (Leveau and Lindow 2005), produce one and degrade the precursor of another (Patten and Glick 2002), or harbor the genes for more than one biosynthetic pathway, e.g. Pantoea agglomerans pv gypsophilae, which features an IAM as well as an IPyA biosynthetic pathway for IAA (Manulis et al. 1998). This potential of even single bacterial strains to interfere differently with plant hormone levels remains one of the challenges towards better understanding, predicting, and possibly controlling plant hormone manipulation in complex plant-associated bacterial communities.

Plant signalling and physiology are affected by bacterial hormone synthesis and/or degradation in different ways, depending on the physiological role of the hormone, on the recalcitrance of plant tissue to changes in the hormone pool, and on the magnitude of the hormonal sink or source that these bacteria represent. Bacterially produced IAA may be beneficial or detrimental to plants. In Azospirillum brasilense (Dobbelaere et al. 1999) and P. putida GR12-2 (Patten and Glick 2002) it enhances root proliferation which results in greater root surface area through which more nutrients and water can be absorbed from the soil. In P. syringae pv savastanoi (Robinette and Matthysse 1990), Erwinia chrysanthemi (Yang et al. 2007) and Rhodococcus fascians (Vandeputte et al. 2005), IAA synthesis has been shown to be necessary for pathogenesis. Bacteria with ACC deaminase activity are generally considered beneficial to plants, as they promote root elongation and increase root density (Glick 2005). For cytokinins, it was suggested that bacteria are indispensable to plant growth because they would represent the only source of this type of hormone in plants (Holland 1997). This hypothesis was later rejected however with the discovery of plant genes encoding cytokinin synthesis (Sakakibara and Takei 2002).

From the bacterial perspective, there are several advantages to invest in plant hormone production or degradation. It has been suggested (Robert-Seilaniantz et al. 2007) that plant pathogens benefit from the production of phytohormones as this suppresses plant defense responses. In galls and tumours, production of IAA and cytokinin stimulates cell division, which acts as a sink for exploitable nutrients from other parts of the plant. IAA production may also locally stimulate ethylene biosynthesis, which indirectly prevents water and nutrient losses to the shoot organs above the tumor (Aloni et al. 1995). IAA production or ACC deaminase activity by plant-growth promoting rhizobacteria results in increased root density and therefore more surface to colonize and greater return in root exudation. Several studies have shown that the ability to grow on or in plants is reduced in bacterial mutants unable to produce IAA (Brandl et al. 2001; Suzuki et al. 2003) or ethylene (Weingart et al. 2001), although the basis of this remains unclear.

There might also be other reasons for bacteria to produce or degrade plant hormones. Ethylene, for example, is a fungistatic (Smith 1973), the production of which might help bacteria to compete with fungi for plant-derived nutrients. Similarly, IAA has been shown to be inhibitory at high concentrations to plant-associated bacteria (Liu and Nester 2006). A less obvious reason to degrade plant hormones is that they represent sources of nutrition. For example, P. putida 1290 can use IAA as sole source of carbon and energy (Leveau and Lindow 2005). Given the relatively low concentrations of IAA and other hormones in the plant environment, it is doubtful that these compounds contribute greatly to bacterial biomass. However, it is noteworthy that three of the five classic hormones represent sources of nitrogen which might be of importance under conditions of nitrogen limitation. In fact, several of the degrading enzymes listed in Table 2 release readily available nitrogen from plant hormones or their precursors. For example, ACC deaminase produces ammonia, a property that has greatly facilitated the search for bacteria with ACC deaminase activity by selection for growth on ACC as sole source of nitrogen (Penrose and Glick 2003). Similarly, the transaminase enzyme involved in bacterial ethylene production from methionine releases the amino group from methionine as a source of nitrogen for growth (Ince and Knowles 1985). Several bacteria can use IAA as sole source of nitrogen (Leveau and Lindow 2005), but more than one enzymatic step is required for the release of nitrogen from the indole ring.

Degradation and utilization of plant hormones represent an extreme form of hormone inactivation, analogous to IaaL activity which conjugates and biologically inactivates IAA (Glass and Kosuge 1986). However, it is worth noting that the bacterial degradation products of some plant hormones are in turn signal molecules. For example, a Pseudomonas sp. from soil (Proctor 1958) was shown to convert IAA to catechol via salicylate, which is a plant hormone (Raskin 1992) involved in the plant response to pathogens. Thus, bacteria may have the potential to re-circuit certain plant signalling pathways by conversion of one hormone to another. Such bacterially induced re-circuiting may not be limited to plant signalling pathways. For example, the IAA degradation pathway described for Bradyrhizobium japonicum (Jensen et al. 1995) and an Alcaligenes sp. (Claus and Kutzner 1983) features isatin, which has a demonstrated signalling function in bacteria, e.g. in biofilm formation by strains of E. coli (Lee et al. 2007). Furthermore, there is a growing body of evidence to suggest that IAA can actually act as a signal molecule in bacteria and fungi (Spaepen et al. 2007). For example, IAA induces the expression of genes in E. coli related to survival under stress conditions (Bianco et al. 2006), stimulates by a positive feedback mechanism its own synthesis in Azospirillum species (Vande Broek et al. 1999), and provokes invasive growth in Saccharomyces cerevisiae (Prusty et al. 2004). Thus, the use of hormones as signalling molecules does not appear to be exclusive to plants, but may also underlie part of the communication between bacteria and other microorganism.

Integrative examples

In the rhizobia-plant interaction

The legume rhizosphere has a strong attractive power on rhizobia since abundantly secreted polycyclic aromatic compounds called flavonoids trigger chemotactic responses directing the bacteria to their compatible host (Reddy et al. 2007). Subsequently, specific flavonoids are perceived by the NodD protein, a LysR-type transcription factor, which initiates the transcription of nodulation genes that encode the biosynthetic machinery for the primary bacterial signal, the Nod factor. This lipochito-oligosaccharide consists of a β-1,4-linked N-acetylglucosamine backbone with four or five residues, carries an acyl chain at the C-2 position at the non-reducing end, and can be decorated at defined positions with acetyl, sulfonyl, carbamoyl, fucosyl or arabinosyl moieties depending on the rhizobial strain (reviewed by D’Haeze and Holsters 2002, 2005). Upon perception of the Nod factors by the plant multiple signal transduction pathways are redirected culminating in the initiation of nodule formation. However, the paramount role of legume flavonoids and rhizobial Nod factors in the initiation of the rhizobium-plant interaction has masked the appreciation of other signals derived from both partners in mediating the onset of a successful interaction. Moreover, it has become increasingly clear that flavonoids play several roles (in addition to nod gene induction), and likewise that Nod factors are not only essential for inducing plant responses like root hair curling and cortical cell division (Cooper 2007). The complexity of the molecular dialogue between both partners of the rhizobial symbiosis will be illustrated by two examples: the interaction between Sinorhizobium meliloti and Medicago, and between Rhizobium sp. NGR234 and one of its many hosts.

Typically, the rhizodeposits of alfalfa (Medicago sativa) and one of the model legumes, barrel medic (M. truncatula), are complex and consist of flavonoids, sugars, amino acids, dicarboxylic acids, hydroxy-aromatic acids, biotin and other vitamins that trigger chemotactic responses in and support growth of their microsymbiotic partner Sinorhizobium meliloti (Cooper and Rao 1995; Streit et al. 1996; Heinz et al. 1999). These plant metabolites are sensed and appropriate responses generated via one- and two-component systems, but recently a downstream role for trans-acting riboregulators has been revealed (del Val et al. 2007). The nutritional advantage for the bacteria inhabiting the rhizosphere is reinforced by the secretion of a riboflavin degradation product, lumichrome, by S. meliloti. It is suggested that lumichrome enhances root respiration and that the root-evolved CO₂ increases net carbon accumulation improving both plant and bacterial growth, but alternative mechanisms explaining the plant growth stimulatory effect have not been ruled out (Phillips et al. 1999; Matiru and Dakora 2005). Moreover, once a functional nodule is established, bacteroids synthesize rhizopines that are secreted into the rhizosphere and can be utilised by some S. meliloti strains, further strengthening the nutritional relation between both partners (Galbraith et al. 1998). At high cell densities long chain acyl homoserine lactones (AHLs), quorum sensing signals secreted by S. meliloti, accumulate beyond a threshold level and trigger responses in the population that positively affect the efficiency of root colonisation and nodule invasion, such as the down-regulation of bacterial motility (Hoang et al. 2008) and the production of symbiotically active galactoglucan (Marketon et al. 2003). Unexpectedly it was shown that the AHLs produced by S. meliloti had a strong impact on the proteome of M. truncatula, modulating 7% of the total resolved proteins affecting diverse functions such as primary metabolism, protein processing, transcriptional regulation, host defence, hormone responses and cytoskeletal activity (Gao et al. 2003; Mathesius et al. 2003). M. truncatula itself produces quorum sensing mimics that can potentially modulate the bacterial behaviour in the rhizosphere (Teplitski et al. 2000), and interestingly, exposure of the roots to AHLs of S. meliloti altered the amounts and types of AHL mimics secreted by M. truncatula (Mathesius et al. 2003), illustrating a strong interplay between both partners. At this point of the interaction the bacterial population is located close to the root, sufficiently dense and not motile which allows it to colonize the root hairs. Biofilm formation represents the “natural way of life” for bacterial populations because it offers a protective environment and the possibility for co-operative behaviour (Morris and Monier 2003; Lasa 2006). Typically surface polysaccharides play an important role in biofilm maturation (Branda et al. 2005), and in S. meliloti cyclic β-glucans are mediating efficient root hair attachment (Dickstein et al. 1988), a first and essential step in biofilm formation. The nodD-like gene syrM is involved in controlling biosynthesis of succinoglycan, which contributes to the capacity to form highly structured biofilms (Fujishige et al. 2006). Interestingly, it was discovered that core Nod factors synthesized by the common nod genes nodABC, and regulated by NodD1 but independent of nod gene-inducing plant flavonoids, are also required for biofilm formation and efficient attachment to roots. The core Nod factors apparently facilitate cell-to-cell adhesion which is thought to allow the bacteria to remain closely attached to the roots until, in response to plant inducers, a sufficient localized concentration of the host-specific signalling Nod factor is produced, required for triggering plant developmental processes that mark the onset of the symbiotic interaction (Fujishige et al. 2008). Indeed, upon perception of luteolin by NodD1 (Peck et al. 2006), or non-flavonoid inducers by NodD2 (Phillips et al. 1992; Gagnon and Ibrahim 1998), expression of both the common and the host-specific nod genes is activated and fully decorated Nod factors are synthesized (Lerouge et al. 1990). However, the rhizobial response to plant flavonoids goes far beyond the synthesis of host-specific Nod factors. Several genome-wide studies have identified multiple luteolin- or apigenin-induced genes that have no nod-box in their promotors and hence do not belong to the nod gene family (Barnett et al. 2004; Zhang and Cheng 2006). The function of most these genes awaits elucidation, but these results strongly suggest that the early stages of symbiosis are likely to be more complex than originally anticipated. In a last step of the rhizospheric signalling between S. meliloti and its legume host, the localized production of host-specific Nod factors is perceived by the plant via the LysM-type receptor kinases and the subsequent complex signal transduction cascade that culminates in early plant responses such as initiation of cortical cell division, calcium spiking and formation of colonized curled root hairs (Jones et al. 2007 and references therein). From the latter, the bacteria induce inward tip growth of the root hair and via these infection threads gain access to plant tissues, start their endophytic life phase and initiate their journey to the nodule primordium. Although beyond the scope of this chapter, clearly, during this endophytic part of the infection process many signals are exchanged, some of which are identified and known to be involved for instance in formation and progression of the infection threads (Nod factors, EPS and LPS; Jones et al. 2007, 2008), suppression of and protection against plant defence (SPS; Campbell et al. 2002; Ferguson et al. 2005; Jones et al. 2008) and activation of cortical cell division (flavonoids and cytokinins; Gonzalez-Rizzo et al. 2006; Wasson et al. 2006); many signals however remain to be discovered.

As for other legume-rhizobium examples, the signal exchange occurring at the onset of the symbiotic interaction between the promiscuous nodulator Rhizobium sp. NGR234 (hereafter NRG234) and one of its over 112 hosts (Pueppke and Broughton 1999) overlaps with the one described above for the S. meliloti-Medicago interaction. Indeed, flavonoid and non-flavonoid nod gene inducers (Le Strange et al. 1990), rhizopines, bacterial surface polysaccharides (Broughton et al. 2006; Staehelin et al. 2006) and Nod factors are important players in the communication between this bacterium and its host, but other signals play a role also. The Nod factors secreted by NGR234 activate flavonoid release in soybean (Schmidt et al. 1994), and the flavonoids activate transcription of 19 nod-box-containing promotors and 147 other genes in a nod-box-independent way. Whereas the functions of the latter largely remain to be discovered, the nod-box controlled genes encode typical pathways involved in Nod factor biosynthesis, rhizopine catabolism, SPS synthesis and modification, and nitrogen fixation, but also in transcriptional control, hopanoid synthesis, auxin (IAA) production, and type III secretion (Kobayashi et al. 2004). The presence of nod-boxes in the promotors of transcriptional regulators creates a complex regulatory network that allows sequential activation of gene expression. In this network, NodD1 is the key regulator of all 19 flavonoid-inducible loci including syrM2. SyrM2 in its turn controls the delayed flavonoid-induction of a number of loci that have SyrM binding sites in their promotors. One of these is nodD2 of which the gene product is required for the optimal activation of specific-nod boxes that control the expression of genes involved in the later stages of the symbiotic interaction. NodD2 also represses nodD1 expression, which results in a self-attenuation of the flavonoid-induced regulatory cascade (Kobayashi et al. 2004). Expression of hopanoid biosynthetic genes is NodD1 dependent and thus flavonoid inducible (Kobayashi et al. 2004). These lipids function as membrane reinforcers and could mediate resistance to environmental stress in the soil. However, hopanoids have been discovered in a number of nitrogen-fixing soil bacteria (Kannenberg et al. 1996), and in the actinomycete Frankia they are located in the envelope of specialised nitrogenase-containing vesicles possibly reducing oxygen diffusion and thereby protecting the nitrogenase (Rosa-Putra et al. 2001; Alloisio et al. 2007). Hence, hopanoids might function either during the rhizospheric or the endophytic phase of the symbiotic interaction. NodD1-controlled expression of the response regulator TtsI results in the activation of genes that carry a tts-box in their promotors and, amongst others, code for part of a type III secretion system, nodulation outer proteins (Nops) and homologs of effectors of pathogens, and the rhamnan component of LPS (Marie et al. 2004). The proteins secreted via the type III secretion system are rhizobial keys that are needed when the bacteria have entered the root hairs and, upon injection into the plant cells, they are thought to interfere with the eukaryotic cellular metabolism, altering plant defence or signalling networks permitting the continuation of nodule development (Marie et al. 2004; Skorpil et al. 2005). The rhamnose-rich LPS is likely also only implicated in the later stages of the interaction, and could be required for protection against plant defence molecules and for bacterial release from infection threads (Marie et al. 2004; Broughton et al. 2006). Auxin production is widespread amongst plant-associated bacteria including rhizobia, and it is often related to epiphytic fitness and suppression of defence (Prinsen et al. 1991; Robert-Seilaniantz et al. 2007; Spaepen et al. 2007). NGR234 synthesizes IAA via three independent pathways: the indole-3-acetamide, the tryptamine and the indole-3-pyruvic acid pathway. The latter is predominant and expression of the genes encoding this pathway is controlled by the NodD1-SyrM2-NodD2 regulatory circuit implying a function during the later stages of the interaction when a more intimate contact between both partners has been established (Theunis et al. 2004). Although no obvious nodulation phenotype was obtained upon mutation of the indole-3-pyruvic acid pathway, a putative role has been postulated in vascularisation of the nodule tissue, facilitating carbon and nitrogen exchange, or acting as a synergistic factor for other signals (Theunis et al. 2004).

From the above it is clear that the action radius of flavonoids and Nod factors has been underestimated. Moreover, the molecular dialogue between legumes and rhizobia has proven to go far beyond these two established signals. Instead a true communication network is established between both partners reflecting the complexity of setting up a successful interaction in the rhizosphere.

In the agrobacteria-plant interaction

Agrobacterium tumefaciens is a soil α-proteobacterium that can infect a broad range of dicotyledonous plants and transfers an oncogenic DNA fragment, the T-DNA, from its tumour-inducing (Ti) plasmid to the nuclear genome of plants (Gelvin 2000). This natural engineer largely contributed to the enormous advances in plant sciences. In the transformed plant tissues, the expression of T-DNA genes leads to the uncontrolled synthesis of growth regulators, auxin and cytokinins, resulting in the formation of tumours, a phenomenon known as crown gall disease. Three main steps could be proposed to describe the dynamics of the A. tumefaciens-plant interaction: (1) the colonization of rhizosphere and plant tissues by virulent and avirulent (free of Ti plasmid) agrobacteria; (2) the transfer of T-DNA from virulent agrobacteria to plants; (3) the emergence and development of a tumour in which avirulent bacteria may be converted into virulent ones by horizontal transfer of the Ti plasmid. In the course of their interaction, plants and agrobacteria exchange a wide variety of signals including, sugars, amino acids, phenolics, and lactones.

The number of agrobacteria increases (from 100 to 1,000 fold), as the structure of these populations varies, when the plant environment was compared to bulk soil (Sanguin et al. 2006). Agrobacteria can survive inside roots and root nodules (Wang et al. 2006), and invade the plants via vessels and apoplasm (Cubero et al. 2006). Microarray analysis of bacterial diversity revealed the predominance of agrobacteria in rhizosphere of maize (Sanguin et al. 2006). A high diversity of agrobacteria can coexist in one cubic centimetre of soil (Vogel et al. 2003). Commonly, most of the agrobacteria recovered from soil and rhizospheric samples are avirulent, lacking the Ti plasmid (Mougel et al. 2001). However, in conductive soils, virulent strains may dominate (Krimi et al. 2002). Several functions contribute to the capacity of agrobacteria to colonize the root, including motility, chemotaxis, surface characteristics and assimilation of a large spectrum of plant compounds. The genome of A. tumefaciens C58 is rich in ABC-genes that would participate in the sensing and transport of a large range of organic and inorganic compounds (Wood et al. 2001).

A complex machinery is required for the transfer of T-DNA to a plant cell. The A. tumefaciens VirB/D4 system is an archetypal Type IV secretion system composed of 11 VirB mating pair formation subunits and a VirD4 substrate receptor that form a transenvelope secretion channel (Christie et al. 2005). The substrate of translocation is a single-stranded copy of the T-DNA that becomes integrated into the plant nuclear genome. Transfer of T-DNA operates in a few of hours (Sykes and Matthysse 1986). The transcription of the vir regulon is induced by specific plant-released phenolic compounds in combination with several other stimuli, such as monosaccharides, acidic pH and temperature below 30°C (Brencic and Winans 2005). The VirA-VirG two-component system and ChvE sugar binding protein are involved in the perception of these stimuli. Activation of vir genes and T-DNA transfer were observed in wounded and unwounded plant tissues. In unwounded transformed plant tissues, the synthesis of opines from T-DNA genes is observed even in the absence of tumour (Brencic et al. 2005), suggesting that cell division during wound healing may play a role in tumour formation.

T-DNA encodes the synthesis of the plant growth factors, cytokinines and auxin, as well as opines, which are specific growth substrates and signals for the bacteria colonizing the plant host. The cytokinine biosynthesis enzyme, which is encoded by the T-DNA, is targeted to and functions in plastids to shunt the original cytokinine pathway (Sakakibara et al. 2005). This feature illustrates that agrobacteria manipulate several compartments of the plant cells. The emergence and development of a tumour is a complex process in which overproduction of auxin and its gradual, flavonoid-dependent retention in the tissue, play an essential role (Schwalm et al. 2003). Furthermore, high vascularisation and epidermal disruption are associated with the establishment of tumours. These phenomena are linked to the redirection of the nutrient-bearing water flow and carbohydrate delivery for growth of the tumour tissues and the inhabiting bacteria (Wächter et al. 2003).

The synthesis of opines defines a specific micro-habitat in the plant host. The assimilation of opines as carbon and nitrogen sources confer a selective advantage to the Ti plasmid harbouring bacteria in plant tumours, the so called opine niche. Some opines, termed conjugative opines, are required for high-rate of synthesis of 3-oxo-octanoyl-homoserine lactone (OC8HSL), a cell-to-cell signal implicated in the QS regulation of the conjugative transfer of the Ti plasmid (Piper et al. 1993). The recipient bacteria for the Ti plasmid may be Ti plasmid free agrobacteria, which represent up to 1% of the total cultivable bacteria in the rhizosphere, as well as other rhizobacteria belonging to different genera, such as Sinorhizobium, Rhizobium, and Phyllobacterium (Teyssier-Cuvelle et al. 1999, 2004). The Ti plasmid confers to these non-Agrobacterium hosts the capacity to assimilate opine and, in some instances, to induce tumours on the plant hosts; it also remains transferable to other bacteria (Teyssier-Cuvelle et al. 2004). These data strongly suggest that the Agrobacterium populations may not be unique reservoirs for the maintenance and propagation of the Ti plasmid in the rhizosphere. In addition to conjugation, Ti plasmid copy-number (Li and Farrand 2000) and severity of tumour symptoms are also subjected to QS regulation (Pappas and Winans 2003; Chevrot et al. 2006). Even though the mechanism that places emergence of tumours under QS regulation remains unknown, anti-virulence strategies targeting QS, termed quorum-quenching, have been proposed to decrease the Agrobacterium-induced symptoms on plants (Molina et al. 2003; Chevrot et al. 2006).

In A. tumefaciens C58-induced tumours, the conjugative opines, agrocinopines A and B, tightly control the synthesis of the OC8HSL signal at the transcriptional level. The AccR-mediated transcriptional repression of the arc (agrocinopine catabolism) operon (orfA-orfB-splA-traR-mcpA) of the Ti plasmid is released in the presence of agrocinopines A and B (Beck von Bodman et al. 1992; Piper et al. 1999). The traR gene of the arc operon encodes the transcriptional regulator TraR that binds OC8HSL and permits the expression of the OC8HSL synthase encoded by the traI gene. This latter gene belongs to the trb operon, located on the Ti plasmid. The TraR/OC8HSL complex also activates the expression of the tra and rep operons that are required for conjugative transfer and copy-number amplification, respectively, of the Ti plasmid. However, TraR activity is modulated at the post-translational level by TraM, which directly interacts with TraR (Luo et al. 2000) and thereby prevents the interaction between the TraR/OC8HSL complex and target DNA-sequences of QS-regulated promoters. In the presence of conjugative opines, the antagonistic effect of TraM would be compensated by the high synthesis rate of TraR.

The enzymatic inactivation of OC8HSL by lactonases AttM (Zhang et al. 2002) and AiiB (Carlier et al. 2003) also participates in the fine tuning of QS-controlled functions in A. tumefaciens C58. The expression of the lactonase AttM is regulated at the transcriptional level by plant signals, such as gamma-aminobutyrate (GABA) and its by-products such as gamma-hydroxybutyrate (GHB) and succinic semialdehyde (SSA) (Carlier et al. 2004; Chevrot et al. 2006). In wounded tissues and in A. tumefaciens-induced plant tumours GABA accumulates to high levels (Deeken et al. 2006). Noticeably, increasing evidences would suggest that GABA plays a key-role in interactions between plants and other organisms, including bacteria, fungi and insects (Shelp et al. 2006). In A. tumefaciens, the lactonase-encoding gene attM is part of the attKLM operon, the expression of which is controlled by the transcription factor AttJ (Zhang et al. 2002). In the presence of SSA and GHB, the repressing activity of AttJ is altered and the attKLM operon is expressed (Chai et al. 2007). Although GABA and gamma-butyrolactone (GBL) do not directly alter the repressing activity of AttJ, the expression of attKLM is also observed in the presence of these compounds. It is assumed that GABA and GBL are converted to SSA and GHB by A. tumefaciens and/or the plant host. In addition to the implication of AttM in the GBL-ring cleavage of OC8HSL, the attKLM operon encodes a complete degradation pathway of GBL into succinate, with GHB and SSA as intermediates (Carlier et al. 2004; Chai et al. 2007).

Plants recognize agrobacteria as invaders, and induce plant defense genes; in parallel agrobacteria have developed strategies to avoid plant defenses, including phenolics and reactive oxygen species (Kalogeraki et al. 1999; Citovsky et al. 2007; Saenkham et al. 2007). Noticeabely, benefic bacteria are also able to induce and avoid some chemical plant-defenses (examples in Faure et al. 1995, 1996; Dombrecht et al. 2005; Madhaiyan et al. 2006). A. thaliana detects different A. tumefaciens effectors, such as a conserved domain of flagellin and the transcriptional factor EF-Tu of A. tumefaciens. Specific receptors belonging to the Leu-rich repeat transmembrane receptor (LRR) family are implicated in perception of these effectors, such as EFR for EF-Tu and FLS2 for flagellin in A. thaliana (Chinchilla et al. 2006; Zipfel et al. 2006). Remarkably, Nicotiana benthamiana, a plant unable to perceive EF-Tu, acquires EF-Tu binding sites and responsiveness upon transient expression of the EFR receptor of A. thaliana. The LRR receptor kinase activates the mitogen-activated protein kinases (MAPK) to activate the immune response. One of the phosphorylated targets of MAPK3 is the transcription factor VIP1 that relocalizes from the cytoplasm to the nucleus and regulates the expression of the PR1 pathogenesis-related gene. A. tumefaciens uses a Trojan horse strategy by hijacking VIP1 to import the VirE2 protein (associated with the T-DNA) into the nucleus (Djamei et al. 2007). Finally, two recent studies described the essential role of salicylic acid (SA) and auxin (IAA) in the control of virulence. IAA inhibits the expression of vir genes and the growth of A. tumefaciens (Liu and Nester 2006). This feature suggests a retro-control of T-DNA transfer by a product encoded by the T-DNA; therefore IAA avoids the cost—for plant and bacteria— of an additional transformation. However, SA, which accumulates upon bacterial infection, also shuts down the expression of the vir regulon (Yuan et al. 2007). Recently, multidisciplinary approaches are taken to give an integrative view of the fascinating A. tumefaciens-plant host interaction (Deeken et al. 2006; Yuan et al. 2008a).

Conclusions and perspectives

A multiplicity of signals controls the responses of plants and their associated organisms in the rhizosphere. The deciphering of the interconnections between all these signals is a future challenge that will be supported by global and fine analytic tools including transcriptomics, proteomics and metabolomics. Moreover, the analysis of temporal and spatial factors in these processes will give more precise insights into the dynamics of the interactions in the rhizosphere. Finally, in addition to model organisms, approaches such as metagenomics (Leveau 2007; Riaz et al. 2008), will take into account the diversity of organisms that communicate in the rhizosphere and the mechanisms implicated in this communication.