Present address: Mickaël Boyer, URMITE, CNRS 6236 – IRD 3R198, Faculté de Médecine, Université de la Méditerranée, 27 Boulevard Jean Moulin, 13385 Marseille Cedex 5, France.
Editor: Ian Head
Correspondence: Florence Wisniewski-Dyé, CNRS, UMR 5557, Ecologie Microbienne, Université Lyon 1, 43 boulevard du 11 Novembre 1918, 69622 Villeurbanne Cedex, France. Tel.: +33 472 44 58 89; fax: +33 426 23 44 68; e-mail: firstname.lastname@example.org
Bacterial signalling known as quorum sensing (QS) relies on the synthesis of autoinducing signals throughout growth; when a threshold concentration is reached, these signals interact with a transcriptional regulator, allowing the expression of specific genes at a high cell density. One of the most studied intraspecies signalling is based on the use of N-acyl-homoserine lactones (AHL). Many factors other than cell density were shown to affect AHL accumulation and interfere with the QS signalling process. At the cellular level, the genetic determinants of QS are integrated in a complex regulatory network, including QS cascades and various transcriptional and post-transcriptional regulators that affect the synthesis of the AHL signal. In complex environments where bacteria exist, AHL do not accumulate at a constant rate; the diffusion and perception of the AHL signal outside bacterial cells can be compromised by abiotic environmental factors, by members of the bacterial community such as AHL-degrading bacteria and also by compounds produced by eukaryotes acting as an AHL mimic or inhibitor. This review aims to present all factors interfering with the AHL-mediated signalling process, at the levels of signal production, diffusion and perception.
Bacterial populations can act co-operatively and do so by emitting and detecting small diffusing compounds, whose concentration is crucial for a coordinated behaviour. This process, known as quorum sensing (QS), allows a bacterial population to evaluate its size and to express specific genes at a high cell density, i.e. when a ‘quorum’ is reached (Fuqua et al., 1994). As the size of the bacterial population increases, so does the concentration of the autoinducing signal, which diffuses in and out of bacterial cells. Once a threshold concentration of the signal is reached, the compound is detected and triggers the activation or the repression of target genes. Some of the most-studied bacterial autoinducers are N-acyl-homoserine lactones (AHL) synthesized by LuxI-type proteins and these have been reported so far in over 70 genera belonging to the Proteobacteria; AHL interact with LuxR regulators and are used to regulate multiple functions, such as bioluminescence, synthesis of virulence factors, synthesis of antimicrobial compounds, production of exopolysaccharides, motility, biofilm development, etc. QS-controlled processes are often crucial for successful bacteria–host interactions, whether symbiotic or pathogenic.
The concept of AHL-mediated bacterial communication was initially based on studies carried out on clonal populations under laboratory conditions; however, natural bacterial populations live in complex environments where biodiversity and abiotic parameters are constantly fluctuating. Thus, the signalling process is directly influenced by abiotic factors (such as pH, temperature and medium composition) and biotic factors (such as other members of the bacterial community) that can modulate signal genesis, diffusion, interception and degradation and that can produce parasitic signals.
Many studies have also demonstrated that QS is a far more complex regulatory process than initially described, as luxI and luxR genes are regulated, directly or indirectly, by a number of transcriptional and post-transcriptional regulators responding to environmental conditions. Thus, QS is integrated in a global regulatory network in order to control AHL production according to various factors (such as trophic conditions) and to optimize the expression of QS-regulated genes.
This review aims to expose the different levels of regulation involved in AHL production inside the bacterial cell and the numerous environmental factors, whether abiotic or biotic, which can affect the fate of AHL outside the bacterial cell or interfere with signal perception.
Regulation of the synthesis and accumulation of AHL signals within bacterial cells
Genetic determinants of QS and QS cascade
Most of the AHL synthases described so far belong to the LuxI family, so called after the identification of the Vibrio fischeri AHL synthase that regulates bioluminescence. LuxI-type proteins direct the formation of an amide bond between S-adenosylmethionine and the acyl moiety of the cognate acyl carrier protein (ACP) (Hanzelka & Greenberg, 1996; Moréet al., 1996). Site-directed mutagenesis on LuxI of V. fisheri and RhlI of Pseudomonas aeruginosa revealed critical conserved amino acids for the synthase activity in the N-terminal part; the less conserved C-terminal part is involved in the recognition of acyl chains carried by ACP (Hanzelka et al., 1997). The nature of available substrates (amount of ACPs available in the cell), the specificity of AHL synthases for the different ACPs and the presence of several AHL synthases govern the nature of AHLs synthesized by a particular bacterium and the kinetics of AHL production. The majority of AHL synthases can accept several acyl-ACPs, leading to the synthesis of a set of different AHLs, as demonstrated for YtbI of Yersinia pseudotuberculosis, which can synthesize up to 24 different AHLs (Ortori et al., 2007).
AHL perception is mediated by a transcriptional regulator of the LuxR family, referring to the protein regulating the bioluminescence process initially evidenced in V. fisheri. Despite their low similarity, LuxR homologues all possess several conserved amino acids (Whitehead et al., 2001) and an identical architecture: an N-terminal moiety involved in interaction with the AHL signal, a C-terminal moiety with a helix-turn-helix motif involved in interaction with DNA, the central part of the protein allowing oligomerization (Stevens & Greenberg, 1997; Zhu & Winans, 1999; Fuqua & Greenberg, 2002). A few LuxR homologues, such as EsaR of Pantoea stewartii, function as repressors as they can bind DNA in the absence of AHLs, impeding the transcription of target genes; AHL binding reduces EsaR affinity for DNA, relieving the repression (von Bodman et al., 1998; Minogue et al., 2002).
In various bacteria, luxI homologues belong to the QS regulon; this creates a positive feedback loop that amplifies AHL synthesis, allowing an efficient coordination of regulation at the population level (Choi & Greenberg, 1991; Fuqua & Winans, 1994; Seed et al., 1995). Conversely, antiactivators such as TraM and TrlR of Agrobacterium tumefaciens, which interact with the TraR regulator, or QscR of P. aeruginosa, which can compete with LasR for AHL binding, can impede the premature expression of target genes at a low cell density and delay the establishment of the positive feedback loop (Fig. 1) (Luo et al., 2000; Chai et al., 2001; Lequette et al., 2006).
Several bacteria possess several luxR/luxI systems (Table 1) that are often organized in hierarchical networks. In the opportunistic pathogen P. aeruginosa, well known for its devastating effects on cystic fibrosis patients, QS is essential for chronic infection as it controls adhesion, biofilm formation and expression of virulence factors. The QS network consists of two LuxI/LuxR circuits arranged in series: LasI/LasR/3oxo,C12-HSL and RhlI/RhlR/C4-HSL. The LasR/3oxo,C12-HSL complex activates a variety of target genes involved in virulence and also exerts a transcriptional control of rhlR and rhlI (Latifi et al., 1996); thus, induction of the genes under control of the rhl system occurs subsequent to the induction of genes under control of the las system. Transcriptome analyses of P. aeruginosa revealed three classes of QS-regulated genes that are expressed at different times over growth: genes that respond to only one AHL, genes that respond to either AHL and genes that require both AHLs to be activated (Schuster et al., 2003; Wagner et al., 2003; Wagner et al., 2004). The hierarchical network of QS systems allows a chronologically ordered sequence of gene expression that might be decisive for a successful infection (Schuster et al., 2003).
In the mammalian enteropathogen Y. pseudotuberculosis that causes gastroenteritis in humans and chronic infections in immunocompromised patients, two QS systems, the YpsR/YpsI and YtbR/YtbI, modulate swimming motility via regulation of flhCD and fliA; flhCD and fliA encode, respectively, the motility master regulator and the flagellar sigma factor and fliA expression is flhCD dependent (Atkinson et al., 1999, 2008). The YpsR/YpsI system exerts a hierarchical positive regulation on the YtbR/YtbI and the AHLs synthesized via YtbI play a dual role, activating flhDC, but repressing fliA. Yersinia pseudotuberculosis is motile at 22 °C, but not at 37 °C, and QS is thought to prevent motility by repressing fliA, thus allowing FlhCD to control other genes unrelated to the flagellar cascade, but required for invasion of a mammalian host at 37 °C (Atkinson et al., 2008).
Rhizobium leguminosarum bv. viciae, a soil bacterium that can nodulate pea, vetch and lentil, possesses four QS systems, with each synthase being able to make a specific set of AHLs, and an additional orphan regulator (Table 1) (Wisniewski-Dyé & Downie, 2002; Sanchez-Contreras et al., 2007). The cinR/cinI/3OH,C14:1-HSL system, at the top of a regulatory cascade, regulates all other QS loci (Lithgow et al., 2000) and appears to act as an overall switch potentially influencing many aspects of rhizobial physiology, such as survival, transfer of symbiotic plasmid and association with specific legumes (Wisniewski-Dyé & Downie, 2002).
Transcriptional and post-transcriptional regulation
AHL synthesis can be modulated independently of cell density by various physiological factors (such as trophic conditions) through two-component regulatory systems, transcriptional regulators and post-transcriptional control. One of the earliest examples came from V. fischeri, the AHL synthesis of which is activated by the catabolic repressor CRP in response to specific substrates (Dunlap & Greenberg, 1985; Dunlap, 1989). One of the most-studied two-component systems is the GacS/GacA system, highly conserved in Pseudomonas species, which is interconnected with the Rsm (repressor of secondary metabolites) system allowing post-transcriptional control of QS (for a review, see Bejerano-Sagie & Xavier, 2007). The Rsm system involves RsmA, a protein able to complex with mRNAs to trigger their degradation by RNAses, and small noncoding RNAs (sRNAs). These sRNAs regulate gene expression either by binding to mRNA, affecting their stability or their translation, or by binding to regulatory proteins (such as RsmA) impeding their activity (Fig. 1). The use of sRNAs with a high turnover, rather than proteins, certainly allows the cells a swifter response.
In the opportunistic pathogen P. aeruginosa, inactivation of the GacA regulator was initially shown to reduce the synthesis of C4-HSL, and of the two LuxR-type transcriptional regulators, RhlR and LasR, and consequently to affect C4-HSL/RhlR-regulated functions (production of pyocyanin, cyanide and lipase) (Reimmann et al., 1997). GacA activates the expression of numerous genes, including rsmZ- and rsmY-encoding sRNAs (Kay et al., 2006). RsmA, whose synthesis is dependent on cell density, can activate rsmZ and rsmY expression; both RsmZ and RsmY can sequester RsmA, hindering its activity and thus creating a negative feedback loop. Thus, activation by the GacS/GacA triggers inactivation of RsmA via the sRNAs RsmZ and RsmY. In the absence of GacS/GacA activation, RsmA can interact with rhlI transcripts, inhibiting their translation and hence limiting C4-HSL synthesis (Pessi et al., 2001).
A second two-component regulatory system, the PprA/PprB, which controls membrane permeability and antibiotic sensitivity in P. aeruginosa, is also interconnected with QS; an insertion mutation in pprB (encoding the response regulator) causes a drastic reduction in lasI, rhlI and rhlR expression, and consequently a reduction in virulence factor production and cell motility (Dong et al., 2005). Furthermore, the 3oxo,C12-HSL influx is significantly reduced in the pprB mutant, hindering the establishment of the positive feedback loop of lasI and hence the expression of rhlI/rhlR that is partially under the control of the lasI/lasR/3oxo,C12-HSL (Dong et al., 2005). However, signals recognized by the cognate sensor proteins (GacA and PprA) remain to be characterized. RetS, an unusual hybrid sensor kinase-response regulator of P. aeruginosa, in concert with the GacS/GacA system and the LadS regulator, controls the expression of virulence factors (such as the type III secretion system) and is required for acute and chronic infection (Laskowski & Kazmierczak, 2006; Ventre et al., 2006).
The interconnection of regulators with QS networks was also reported in other Pseudomonas species. The GacS/GacA system controls the production of AHL required for the production of phenazines, antimicrobial compounds mediating the biocontrol activity of Pseudomonas aureofaciens 30–84 against fungal phytopathogens (Chancey et al., 1999); in Pseudomonas chlororaphis PCL1391, this regulation also involves PsrA (Pseudomonas sigma regulator), which positively regulates phzI/phzR expression via RpoS (Girard et al., 2006). GacA and PsrA act as master regulators of the virulence of plant-pathogenic Pseudomonas syringae pv. tomato, notably by controlling the expression of transcriptional activators (including LuxR regulators), alternate sigma factors (such as RpoS) and regulatory RNA (rsmB and rsmZ, for GacA) (Chatterjee et al., 2003, 2007).
In the plant-pathogen Erwinia carotovora ssp. carotovora, the GacS/GacA system regulates the synthesis of a unique sRNA, RsmB, which can bind to RsmA and inhibit its activity. RsmB is repressed by multiple components such as KdgR (a transcriptional regulator activated by plant signals) and HexA (a global regulator). In the absence of AHLs (i.e. at a low cell density), the AHL receptors, ExpR1 and ExpR2, activate the synthesis of RsmA by binding the rsmA promoter; this activation is inhibited in the presence of AHLs. Thus, at a high cell density (i.e. when the GacS–GacA two-component system is activated), the expression of QS-regulated phenotypes is no longer inhibited by RsmA (for a review, see Barnard & Salmond, 2007).
QS can also be regulated at the level of protein stability; the ATP-dependent Lon protease acts as a negative regulator of AHL production in Pseudomonas putida and P. aeruginosa, notably by degrading LasI and hence repressing the expression of lasR/lasI (Bertani et al., 2007; Takaya et al., 2008). In the absence of its AHL ligand, the LuxR regulator of A. tumefaciens, TraR, is susceptible to degradation by Clp and Lon proteases, indicating the importance of these proteases in QS timing and regulation (Zhu & Winans, 2001).
Endogenous AHL-inactivating enzymes
Two families of bacterial enzymes with AHL-degrading activity have been evidenced: AHL lactonases can hydrolyse the ester bond of the lactone ring, and AHL acylases can hydrolyse the amide bond of the molecule, releasing the homoserine lactone moiety and the corresponding fatty acid. Initially, such enzymes were identified, respectively, in Bacillus sp. and Variovorax paradoxus, bacteria that do not synthesize AHL signals (see Biotic factors intrinsic to the bacterial population) (Dong et al., 2000; Leadbetter & Greenberg, 2000). More recently, these enzymatic activities were found in AHL-producing bacteria belonging to the genera Agrobacterium and Pseudomonas.
In A. tumefaciens, attM encodes an AHL lactonase able to inactivate the TraI-made signal, 3oxo,C8-HSL. During the exponential phase, attM expression is repressed by the AttJ regulator, allowing 3oxo,C8-HSL accumulation and expression of the QS-regulated phenotype, i.e. conjugative transfer of the Ti plasmid (pTi). When cells enter the stationary phase, the inhibition of plasmid transfer is observed and coincides with a decrease in the AHL concentration due to AttM expression (Zhang et al., 2002). Under these conditions of starvation, AHL inactivation would save the energetic cost necessary for the QS-regulated conjugation. Moreover, attM expression is activated by γ-amino butyric acid (GABA), a compound synthesized by numerous organisms including plants (Chevrot et al., 2006). As GABA synthesis increases when the plant is wounded, it could stimulate AHL inactivation by AttM in A. tumefaciens cells located in the vicinity of the wound and modulate pTi transfer, and consequently affect bacterial virulence. However, a recent study showed that induction of attM resulted in only a twofold decrease in intracellular AHL levels, and mating experiments performed in developing tumours showed that AttM had no significant effect on plasmid transfer (Khan & Farrand, 2009). Two other enzymes homologous to AHL lactonases, AiiB and AiiC, are encoded by the A. tumefaciens genome, but AHL-degrading activity was evidenced only for AiiB (Carlier et al., 2003).
AHL-acylase and AHL-lactonase activities have been identified in numerous soil isolates belonging to the genus Pseudomonas, notably P. aeruginosa and P. syringae (Huang et al., 2003; Huang et al., 2006; Shepherd & Lindow, 2009). The PvdQ enzyme, homologous to AHL acylases, inactivates long acyl chain AHLs more efficiently than short acyl chain AHLs, and consequently its overexpression prevents 3oxo,C12-HSL accumulation. However, inactivation of pvdQ does not abolish the ability of P. aeruginosa to grow on a medium containing AHL as the sole carbon source, suggesting the existence of other AHL-degrading enzymes (Huang et al., 2003). The QuiP protein, displaying 21% homology with PvdQ, is also an AHL acylase that specifically degrades long acyl chain AHLs in P. aeruginosa; however, its role and its expression pattern remain to be determined (Huang et al., 2006). These AHL acylases would modulate the proportion of the two AHLs synthesized, in order to fine-tune their corresponding target genes.
Even if these endogenous enzymes can accelerate AHLs turnover and degradation, their activity is not solely dedicated to AHL inactivation as they can be involved in other metabolic pathways. In A. tumefaciens, attM belongs to the attKLM operon-encoding enzymes involved in an assimilative pathway of γ-butyrolactone; transcription of the attKLM operon is activated by γ-butyrolactone and also by GABA (Carlier et al., 2004). As for pvdQ of P. aeruginosa, this gene is also involved in the biosynthesis of the siderophore pyoverdine (Ochsner et al., 2002; Lamont & Martin, 2003).
AHL diffusion and transport through membrane
AHLs, which are amphipathic molecules, are expected to freely diffuse from the inside to the outside, and vice versa, but this has been demonstrated only for the short acyl chain AHL 3oxo,C6-HSL using a 3H-labelled derivative (Kaplan & Greenberg, 1985). As the hydrophobicity is affected by the length of the acyl chain, the number of insaturations and the nature of the C3 substituent (H, O or OH), the diffusion speed is correlated with the nature of the acyl chain and long acyl chain AHLs, if they can diffuse at all, would diffuse more slowly than short acyl chain AHLs.
In P. aeruginosa, an active efflux of 3oxo,C12-HSL was evidenced, whereas C4-HSL freely diffuse across the cell membrane (Pearson et al., 1999). This efflux is mediated by a multidrug-efflux pump, encoded by the mexAB-oprM operon; a defined mutant lacking this pump accumulates 3oxo,C12-HSL with a higher concentration in the cytoplasm than in the extracellular medium. A second efflux pump, encoded by the mexGHI-opmD operon, is linked to QS; P. aeruginosa mutants no longer expressing this pump exhibit a reduced production of AHLs in the extracellular medium and reduced production of virulence factors (Aendekerk et al., 2002). Burkholderia pseudomallei KHW can produce six different AHLs and these molecules trigger transcription of the bpeAB-oprB genes coding for an active-efflux pump required for the extracellular secretion of all AHLs; AHL production is undetectable in a bpeAB-null mutant and its virulence is attenuated (Chan & Chua, 2005; Chan et al., 2007). This suggests the possibility of attenuating B. pseudomallei virulence using inhibitors of the BpeAB-OprB efflux pump.
So far, active efflux of AHLs has only been evidenced in P. aeruginosa and in B. pseudomallei; as a significant number of bacteria were shown to produce long acyl chain AHLs, it is likely that efflux systems are used by other bacteria. Via this type of transporter, the different AHLs produced by a bacterium can accumulate differentially inside and outside the cell; thus, the threshold concentration necessary to trigger QS-regulated genes will rely on cell number but also on the nature of the synthesized AHL. To our knowledge, no system involved in the entry of AHLs into the cytoplasm has been evidenced.
Regulation of AHL production by other endogenous signals
Gram-negative bacteria can produce signalling molecules other than AHLs. As a single bacterium can synthesize several types of signals, the synthesis of AHLs can be regulated by the production and perception of the other signals. For example, P. aeruginosa produces 2-heptyl-3-hydroxy-4-quinolones, collectively named PQS (Pseudomonas quinolone signal), whose synthesis is observed after the exponential phase of growth and relies on LasR/3oxo,C12-HSL (Pesci et al., 1999). The exogenous addition of PQS activates the expression of rhlI, the gene involved in C4-HSL synthesis, and consequently the expression of Rhl-regulated genes during the stationary phase. In return, C4-HSL would regulate negatively the production of PQS, but whether this control is direct or indirect remains to be demonstrated (McGrath et al., 2004; Wade et al., 2005). This complex regulatory network, relying on both AHLs and PQS, plays a central role in coordinating virulence, antibiotic resistance and fitness in P. aeruginosa.
In addition to AHLs and PQS, P. aeruginosa produces cyclic dipeptides (diketopiperazines) that could act as signalling molecules (Holden et al., 1999). These compounds could interfere with the signalling process, by competing with the AHL for binding to LuxR regulators. In this way, diketopiperazine could inhibit the feedback loop of lasI and rhlI and affect the accumulation of AHLs.
In the plant-pathogen Ralstonia solanacearum, cell–cell communication is integrated in a complex regulatory network, allowing the induction of virulence genes in response to plant signals and nutritional starvation. Ralstonia solanacearum synthesizes 3-hydroxypalmitic acid (3-OH-PAME) accumulating throughout bacterial growth (Flavier et al., 1997a). At a high cell density, 3-OH-PAME allows the expression of the transcriptional regulator PhcA, which induces the expression of virulence factors and the synthesis of AHLs (Flavier et al., 1997b). However, the QS regulon remains to be characterized, as well as the involvement of QS in virulence.
Furanosyl borate diesters, also named AI-2, are signalling molecules found in numerous Gram-negative and Gram-positive bacteria and are thought to act as interspecies signals (Waters & Bassler, 2005). In V. fischeri, AI-2 indirectly regulates C8-HSL synthesis and affects luminescence regulation (Lupp & Ruby, 2004).
Environmental factors affecting the synthesis, stability, diffusion and perception of AHLs signals
pH and temperature
pH and temperature are environmental factors affecting the half-life of AHLs. Indeed, alkaline pH and high temperature (>37 °C) favour the opening of the lactone ring, producing an N-acyl-homoserine compound devoid of signalling properties. Whereas 70% of N-propionyl-homoserine lactone (C3-HSL) is hydrolysed at pH 6, C4-HSL is completely hydrolysed at pH 8 (Yates et al., 2002; Decho et al., 2009). Consequently, AHLs must contain an acyl chain with at least four carbons for sufficient stability and activity under pH conditions encountered by most bacteria. Moreover, long acyl chain AHLs (>C8) are less sensitive to alkaline lysis and to elevated temperatures than short acyl chain AHLs.
Culture of E. carotovora ssp. carotovora in a rich laboratory medium is accompanied by alkalinization in the stationary phase. This alkalinization leads to inactivation of 3oxo,C6-HSL and reduced production of carbapenem, whose synthesis is QS regulated. In a buffered medium, 3oxo,C6-HSL and carbapenem can be detected during the stationary phase when bacteria are grown at 30 °C (optimal growth temperature), but remain at very low concentrations at 37 °C (Byers et al., 2002; McGowan et al., 2005). Interestingly, one of the plant responses following infection by E. carotovora is the activation of a proton pump allowing the alkalinization of the infection zone and hence the inactivation of AHLs; such a response is likely to interfere with QS regulation of virulence factors.
The influence of temperature on AHL synthesis was clearly demonstrated for the psychrotolerant bacterium Pectobacterium atrosepticum (causing soft rot of potato) whose pectate lyase production is under QS regulation. Maximal AHL synthesis is observed at the optimal growth temperature (24 °C) and is directly correlated with expI transcripts; at 12–15 °C, AHL production is reduced whereas pectate lyase production is optimal, suggesting a thermoregulation occurring downstream quorum signalling (Latour et al., 2007). The influence of temperature on AHL synthesis has also been studied in the food-borne human opportunistic pathogen Aeromonas hydrophila that regulates exoprotease production and biofilm development through QS; the concentration of C4-HSL is lower at 12 °C than at 22 or 30 °C (the optimal growth temperature) despite reaching dense populations, indicating that 12 °C is not inhibitory to C4-HSL production and that quorum signalling might occur under real food conditions (Medina-Martinez et al., 2006).
In E. carotovora ssp. carotovora, the nature of the carbon source (and also the temperature of growth) influences carbepenem production by modulating the level of carI transcription; the concentration of 3oxo,C6-HSL is high, intermediate or low in, respectively, a sucrose, a glucose or a glycerol-grown culture (McGowan et al., 2005). In A. hydrophila, C4-HSL synthesis occurs in Luria–Bertani broth supplemented with 0.1% and 0.5% glucose whereas no synthesis is detected in 1% glucose (Medina-Martinez et al., 2006); the QS regulation of exoproteases would allow this bacterium to use alternative carbon sources when the medium becomes deficient in glucose. In R. leguminosarum, where QS notably regulates transfer of a symbiotic plasmid, the quantity and the ratio of each AHL are also affected by the composition of the growth medium (Lithgow et al., 2001). Molecular mechanisms underlying the finely tuned production of AHLs according to the carbon source available remain unknown.
By studying the expression of the las and rhl systems of P. aeruginosa in different media differing in nutrient composition and oxygen composition, the expression of both systems appears to be significantly affected by nutritional growth conditions (Wagner et al., 2003; Duan & Surette, 2007). A differential regulation of lasI and rhlI is observed when bacteria are cultivated anaerobically (but not aerobically) in rich media: the transcription level of lasI is higher whereas rhlI transcription is repressed (Wagner et al., 2003). The expression level of AHL synthases (LasI and RhlI) is not always correlated with the expression level of the corresponding regulatory proteins (LasR and RhlR), suggesting that those genes are independently regulated, allowing fine-tuning of each system (Duan & Surette, 2007). LuxR-type regulators and their cognate AHL could thus play different roles according to growth culture conditions.
One of the nutrient deficiencies that can dramatically affect AHL production, at least in P. aeruginosa, is iron deficiency. Iron deficiency leads to enhanced formation of virulence factors and to inhibition of oxygen transfer, which may decrease the formation of oxidants or increase the solubility or availability of iron (Kim et al., 2003). The expression of the las system and markedly of the rhl system increases in response to iron limitation (Bollinger et al., 2001; Duan & Surette, 2007). Moreover, a strong correlation between the exhaustion of iron and lasR expression is observed under oxygen limitation (Kim et al., 2005). These findings are highly relevant as iron availability is often limited in biofilm, the primary mode of growth of this pathogen in the lung of cystic fibrosis patients, and as host cell defence mechanisms include iron sequestration and formation of oxidants.
Mass transfer processes, such as diffusion and advection, can strongly influence AHLs accumulation within a given environment. The number of bacterial cells required to reach the threshold concentration of AHL can thus vary considerably according to the diffusion rate of AHLs. The latter is dependent on the nature of a given AHL, on medium hydrophobicity, but can also be affected by bacteria themselves; indeed, several AHL-producing species are also capable of secreting polysaccharides constitutive of biofilm matrixes (Kolter & Greenberg, 2006). In a P. aeruginosa biofilm, it was suggested that the presence of hydrophobic exopolysaccharides in the biofilm matrix would limit the diffusion of AHL, because the 3oxo,C12-HSL concentration is higher within the biofilm than on the surface circulating fluid (Charlton et al., 2000). C4-HSL, being less hydrophobic than 3oxo,C12-HSL, have fewer interactions with the biofilm matrix and would be mainly responsible for the signalling process (Singh et al., 2000).
AHLs can also be transported by circulating fluids, a phenomenon termed advection, a process that will lead to signal washing and dilution (Horswill et al., 2007). This physical factor can play an important role at the local scale, notably when bacteria form biofilms. In P. aeruginosa, the speed of circulating fluids affects significantly the maturation of biofilm, whose structure depends on QS-regulated functions (Purevdorj et al., 2002). AHL accumulation would be greater within the biofilm than at the biofilm surface, leading to premature induction of QS-regulated functions in cells localized within the biofilm (de Kievit et al., 2001).
The plant-pathogen bacterium P. syringae, in which QS controls traits involved in virulence and epiphytic fitness, occurs on leaf surfaces in aggregates of various sizes. AHL accumulation within these aggregates is favoured on dry leaves, suggesting the influence of humidity on AHL diffusion outside aggregates (Dulla & Lindow, 2008). Thus, rain flowing out of leaves can disrupt cell–cell communication of epiphytic bacteria.
In addition to functioning as sensors of population density, low-cost autoinducers such as AHLs could be released as a proxy to determine whether secreted molecules will rapidly diffuse away from the cell, a phenomenon known as diffusion sensing (Redfield, 2002; Hense et al., 2007). Diffusion sensing could allow individual cells to minimize the loss of costly compounds (such as exoenzymes, siderophores, antibiotics) by extracellular diffusion.
Biotic factors intrinsic to the bacterial population
The initial function attributed to AHL signals was to estimate population density. The two factors determining population density (number of cells and volume available for growth) can vary independently, the number of cells increasing within a limited volume or the number of cells being constant while the volume becomes smaller. AHL concentration can increase accordingly, and when reaching a threshold concentration, allow the synchronized expression of specific genes at the population level. In certain bacteria, this phenomenon is amplified by a positive feedback loop on AHL production.
The expression of specific genes when the population has reached a ‘quorum’ is often crucial for bacteria–host interactions. Vibrio fischeri is able to colonize the light organ of the squid Euphrymna scolopes at a very high cell density (1010 cells mL−1); emission of light by V. fischeri is due to AHL accumulation in the limited volume of the light organ. If the light organ was smaller, a lower number of bacterial cells would be required to induce light, but the intensity of light would probably be insufficient for the squid to cheat its predators. A diffusion model showed that levels of bioluminescence are lower in planktonic V. fischeri cells compared with cells adhered onto glass surfaces; QS can occur locally in two-dimensional surface samples and is a function of cell population density as well as signal diffusion time (Parent et al., 2008).
Spatial distribution of cells
The concept of QS comes from studies undertaken on clonal populations, mainly from homogenous liquid cultures. However, these cultural conditions do not reflect natural fluctuating environmental conditions encountered by bacterial populations. In complex environments, such as a rhizosphere, the spatial distribution of cells is far from homogenous, cells are rather isolated or form clusters and are localized where diffusion rate and nutrient availability are temporally changing. A mathematical modelling, using a definite volume and a constant number of cells and taking into account the diffusion and production rates of AHLs, the existence of a positive feedback loop for AHL production and the spatial distribution of cells, shows that the latter is more important than cell density for sensing: the threshold AHL concentration required to induce QS-regulated functions is reached only when cells are clustered and when a positive feedback loop is present (Hense et al., 2007). Thus, it was suggested that the positive feedback loop acting on AHL synthesis is dedicated to coordinating more rapidly a concerted response at the population level, independent of the spatial distribution of cells.
Biotic factors within the bacterial community
AHL inactivation by bacterial enzymes
In natural environments, microorganisms form communities where AHL-producing bacteria interact with other organisms that are able to degrade AHLs. Processes interfering with cell–cell communication are known as quorum quenching. AHL-degrading enzymes fall into two categories: AHL-lactonases (AiiA family) and AHL-acylases/amidohydrolases (AiiD family) (Dong et al., 2000; Leadbetter & Greenberg, 2000; Zhang et al., 2002; Lin et al., 2003; Park et al., 2005). These enzymes have been identified in a huge number of bacterial isolates, notably soil isolates (Reimmann et al., 2002; Zhang, 2003). AHL lactonases, the most-studied AHL-degrading enzymes, were first identified in several Bacillus species (Dong et al., 2002; Lee et al., 2002). AHL lactonases were also evidenced in Klebsiella pneumoniae (ahlK), in the AHL producer A. tumefaciens (AttM and AiiB) and in Gram-positive bacteria such as Arthrobacter (AhlD) and Rhodococcus erythropolis (QsdA that does not display homology to AiiA) (Park et al., 2003; Uroz et al., 2008). Bioinformatic studies revealed that genes homologous to aiiA are present in other Rhizobiaceae such as Bradyrhizobium japonicum or Mesorhizobium loti (Carlier et al., 2003). Two metagenomic approaches identified new genes encoding lactonases displaying low similarity to previously characterized lactonases (Riaz et al., 2008; Schipper et al., 2009).
AHL acylases were identified in various Gram-negative bacteria (Ralstonia, V. paradoxus and P. aeruginosa) and in a high G+C Gram-positive rhizosphere isolate R. erythropolis W2. Two enzymatic activities implied in AHL inactivation were evidenced in this strain: an amidolytic activity, which cleaves the amide bond of AHL, and an oxidoreductase activity, which converts 3-oxo-AHLs to their corresponding 3-hydroxy derivatives (Uroz et al., 2005); growth on C6-HSL as the sole carbon source is likely due to such enzymatic activities.
Numerous bacterial species possessing the ability to inactivate AHLs have been isolated from the rhizosphere (d'Angelo-Picard et al., 2005; Jafra et al., 2006). AHL-producing and AHL-degrading bacteria can coexist within the same environment, as shown within the rhizosphere of tobacco (d'Angelo-Picard et al., 2005); thus, accumulation of AHLs is influenced by the presence of AHL-producing communities and AHL-degrading communities in a particular biotope.
The inactivation of AHLs can not only provide a source of nutrients but also prevents QS signalling in neighbouring bacteria and mitigates the bactericidal effect of 3-oxo-AHLs; such enzymes should provide a competitive advantage to the corresponding bacteria (Leadbetter & Greenberg, 2000; Kaufmann et al., 2005; Yang et al., 2006). Whether AHLs are the primary substrates for these enzymes is still unclear, and thus their physiological function(s) remain(s) to be elucidated. AiiA of Bacillus thuringiensis was recently shown to be involved in rhizosphere competence, as the survival rate of the aiiA mutant significantly decreased over time compared with that of the wild type (Park et al., 2008b).
A given AHL can be produced by several bacterial species; for example, P. aeruginosa, Serratia liquefaciens or A. hydrophila produce C4-HSL, a molecule controlling the expression of virulence factors in these pathogens (Winson et al., 1995; Eberl et al., 1996; Swift et al., 1997). Thus, in its natural habitat, an AHL-producing bacterial species can coexist with other(s) species producing identical AHLs or structurally related AHL, and can perceive these AHLs, leading to cross-talk (Fig. 2).
Spent culture supernatants of P. aeruginosa can activate the production of virulence factors in Burkholderia cepacia, an AHL-producing bacterium cohabiting with P. aeruginosa in the lungs of cystic fibrosis patients (McKenney et al., 1995). Moreover, P. aeruginosa-made AHLs are able to trigger the expression of several genes of B. cepacia, including cepI, in mixed biofilms formed in vitro or in vivo on mouse models (Riedel et al., 2001); this cross-talk seems to be unidirectional. Long-chain AHLs made by Mesorhizobium sp. can also restore the synthesis of protease and pyoverdin in an AHL-deficient P. aeruginosa (Krick et al., 2007).
Phenethylamide metabolites produced by the marine Gram-positive bacterium Halobacillus salinus inhibit QS-regulated phenotypes, such as bioluminescence in Vibrio harveyi and violacein production in Chromobacterium violaceum; these nontoxic secondary metabolites may act as QS antagonists by competing with AHL for receptor binding (Teasdale et al., 2009). Diketopiperazines have also been isolated from supernatants of Proteus mirabilis, Citrobacter freundii, Enterobacter agglomerans and P. putida (Holden et al., 1999; Degrassi et al., 2002). Some diketopiperazines are capable of activating or antagonizing some LuxR-based QS systems, such as the C4-HSL-dependent swarming of S. liquefaciens (Holden et al., 1999; Degrassi et al., 2002). Although the physiological role of these diketopiperazines is yet to be established, their activity suggests the existence of cross-talk among bacterial signalling systems.
Some bacteria, although they do not produce any AHL, can detect these signals and induce specific genes accordingly, and so act as eavesdroppers. For example, Escherichia coli does not make AHL, but possesses a LuxR homologue, SdiA, able to interact with exogenous AHLs (Yao et al., 2006). SdiA is an orphan LuxR, a term used to specify that the bacterium has no AHL synthase or that the LuxR regulator has no cognate AHL synthase. The SdiA/AHL complex regulates specific target genes, notably genes involved in acid tolerance (Van Houdt et al., 2006). This strategy allows E. coli to save the energetic cost linked to AHL synthesis, but the SdiA/AHL regulation only occurs in the presence of AHL producers. Thus, E. coli is able to intercept the signals and could consequently interfere with the accumulation of AHLs produced by other bacterial populations.
Barrier to AHL diffusion
Whether the presence of non-AHL-producing and non-AHL-degrading bacteria affects the induction of AHL-mediated gene expression was investigated. Using artificial microcolonies of mixed species, containing an AHL producer and an AHL biosensor, covered with layers of non-AHL-producing cells, it was shown that the presence of the non-AHL-producing cells enhanced green fluorescence, the biosensor response (Mason et al., 2005). Thus, it is likely that non-AHL-producing cells can act as a barrier to AHL movement, allowing AHL to accumulate within the microcolony, and can thus affect AHL-mediated responses.
Influence of eukaryotes on bacterial communication
Microorganisms residing within biofilms are more resistant to antibacterial agents; the finding that AHLs are involved in biofilm development has thus provided a novel mechanism of biofilm control. 3oxo,AHLs were found to rapidly react with halogenated compounds, such as HOCl and HOBr, extensively used for microbial control in industrial systems (Michels et al., 2000; Borchardt et al., 2001). These compounds can also be generated by a variety of organisms; the natural haloperoxidase systems of the marine alga Laminaria digitata are capable of mediating the inactivation of 3oxo,AHLs, and may prevent biofouling on the algal surface (Borchardt et al., 2001).
Halogenated furanones, compounds naturally produced by the red macroalga Delisea pulchra, were the first molecules found to interfere with QS-regulated phenotypes; these compounds, which display structural homologies with AHLs, can inhibit swarming motility in S. liquefaciens, reduce virulence and bioluminescence in V. harveyi and alter the architecture of P. aeruginosa biofilms (Givskov et al., 1996; Manefield et al., 2000; Hentzer et al., 2002). Halogenated furanones inhibit QS by directly interacting with LuxR homologues, triggering LuxR turnover (Manefield et al., 2002). This type of secondary metabolite is thought to protect D. pulchra from bacterial colonization (Kjelleberg et al., 1997). This initial finding that natural compounds could interfere with AHL signalling accelerated the screening for QS inhibitors with the aim of blocking QS in bacterial pathogens, without the risk of emergence of resistance, in contrast to antibiotics.
The unicellular soil-freshwater alga Chlamydomonas reinhardtii also secretes substances that mimic bacterial AHLs (Teplitski et al., 2004). More than a dozen unidentified substances are capable of specifically stimulating the LasR or CepR AHL-biosensor strains, but not the LuxR, AhyR or CviR biosensors, suggesting that these compounds will affect QS signalling differently. Moreover, treatment of Sinorhizobium meliloti, a soil bacterium establishing a symbiosis with some leguminous plants, with a C. reinhardtii partially purified AHL mimic affects the accumulation of some proteins that are altered in response to the bacterium's own AHL signals (Teplitski et al., 2004). As there is no obvious purpose for C. reinhardtii to interfere with bacterial signalling, the effect of these compounds on signalling might be coincidental.
Plants such as pea, vetch, soybean, rice and Medicago truncatula can produce compounds acting as AHL mimics as they can interfere with some AHL biosensors (Teplitski et al., 2000). A large screening aimed at identifying QS inhibitors revealed that garlic extracts contain such compounds (Rasmussen et al., 2005a). The QS-regulated production of exopolysaccharides of S. meliloti is inhibited by l-canavanine, an arginine analogue that is exclusively produced by leguminous seeds, such as those of Medicago sativa (Keshavan et al., 2005). Although the structure of the active compounds has not been elucidated in most studies, the existence of such compounds suggests that plant–bacteria interactions can be manipulated by plants.
Another level of QS-interference by plant metabolites is exemplified by the plant–A. tumefaciens interaction. This crown-gall-causing agent transfers a part of its pTi, the T-DNA, into the genome of plant cells, leading to anarchic growth of plant cells; T-DNA also codes for proteins involved in the biosynthesis of opines, compounds that are specifically catabolized by A. tumefaciens due to genetic determinants localized on the pTi. The conjugative transfer of pTi, which is QS regulated via the traI/traR system, is also influenced by opines. Thus, the expression of the tra operon requires the presence of two signals: a plant signal, an opine, indicates that the nutrient-rich place is favourable to conjugation, and a signal produced by the recipient bacterial cells, an AHL, triggers conjugation at a high cell density (Zhang et al., 1993; Fuqua & Winans, 1994; Piper & Farrand, 2000). Plant-made GABA and salicylic acid also act on the pTi transfer by inducing the expression of the AttM endogenous lactonase (see AHL diffusion and transport through membrane) (Chevrot et al., 2006; Yuan et al., 2008).
Extracts of 50 members of the genus Penicillium were screened for QS-inhibitory compounds; several fungi were found to produce such inhibitory activities and two of the compounds were identified: penicillic acid and patulin (Rasmussen et al., 2005b). These two compounds can inhibit expression of QS-controlled genes in P. aeruginosa, as revealed by DNA microarray transcriptomics; patulin can enhance biofilm susceptibility to tobramycin treatment. Patulin is thought to accelerate LuxR turnover, as observed previously with halogenated furanones (Rasmussen et al., 2005b).
Pseudomonas aeruginosa seems to be able to perceive its host immune activation. Interferon-γ (IFN-γ), signalling proteins synthesized during the host immune response, can activate the production of virulence factors via the rhlI/rhlR system (Wu et al., 2005; Wagner et al., 2006). IFN-γ are recognized by a major outer membrane protein, OprF, which, by an unknown mechanism, drives rhlR overexpression. rhlI expression triggers the expression of QS-dependent virulence factors, notably PA-I lectin and pyocyanin. In contrast, 3oxo,C12-HSL stimulate the production of IFN-γ by T lymphocytes (Smith et al., 2002). This molecular dialogue illustrates the pivotal role of QS in the establishment of the interaction between a bacterium and its eukaryotic host.
Cyclic dipeptides (diketopiperazines), like the ones produced by some bacterial species (see Regulation of AHL production by other endogenous signals and Biotic factors within the bacterial community), have also been evidenced in yeasts, lichens, fungi and in some mammalian tissues, suggesting that these eukaryotic diketopiperazines could also interfere with bacterial communication (Prasad, 1995).
Recently, the alkaloid solenopsin A, produced by the fire ant Solenopsis invicta, was shown to efficiently disrupt QS in P. aeruginosa, by targeting the rhl QS system (Park et al., 2008a).
Exploitation of AHLs by eukaryotes?
Whether AHL are dedicated only to bacterial signalling is certainly questioned, with several studies reporting that some eukaryotes are able to perceive and to respond to AHLs. Indeed, 3oxo,C12-HSL produced by P. aeruginosa has an immunomodulatory activity and might contribute to bacterial pathogenesis as a virulence determinant per se (Telford et al., 1998; Smith et al., 2002). In the marine environment, the zoospores of the green seaweed Ulva are attracted by synthetic AHLs, and by AHLs released by bacterial biofilms; when zoospores detect AHLs, their swimming rate is reduced and this results in accumulation of cells at the source of the AHL. It is probable that AHLs act as cues for the settlement of zoospores influencing their biogeography, rather than being directly involved as a signalling mechanism (for a review, see Joint et al., 2007).
In the rhizosphere, several AHLs were shown to trigger significant changes in the accumulation of >150 proteins of the model leguminous plant M. truncatula (Mathesius et al., 2003); some of these proteins are involved in defence, stress response, transcriptional regulation and hormonal response. In addition, exposure to AHLs was found to induce changes in the secretion of compounds by the plants that mimic QS signals and thus have the potential to disrupt QS in associated bacteria (Mathesius et al., 2003).
Inoculation of AHL-producing S. liquefaciens on tomato plants increases their systemic resistance against the fungal leaf pathogen, Alternaria alternata, whereas the AHL-negative mutant S. liquefaciens is less effective in reducing symptoms. Moreover, AHL molecules systemically induce salicylic acid and ethylene-dependent defence genes (Schuhegger et al., 2006). Thus, AHL molecules play a role in the biocontrol activity of rhizobacteria through the induction of systemic resistance to pathogens. The contact of Arabidopsis thaliana roots with C6-HSL results in distinct transcriptional changes in roots and shoots, alterations of the auxin to cytokinin ratio and increase of root elongation (von Rad et al., 2008). C6-HSL may contribute to tuning plant growth to the microbial composition of the rhizosphere.
Thus, eukaryotes have an extensive range of functional responses to AHLs that may play important roles in the beneficial or the pathogenic outcomes of eukaryote–prokaryote interactions.
Initially, regulation through QS had been perceived as a relatively simple model involving an AHL synthase, an AHL signal and a LuxR-type regulator activating specific genes at a high cell density. The overall picture is far more complex: QS genes are embedded in a network of global regulation, where the synthesis of the AHL signal is highly responsive to the growth phase and to environmental factors; intertwined QS systems can coexist within a single bacterium allowing the fine tuning of AHL synthesis and consequently the expression of QS-regulated phenotypes. QS regulation networks seem to be all the more complex as they appear to be strain specific with various AHL patterns and distinct QS regulons displayed by strains of the same species (Boyer et al., 2008; Steindler et al., 2008).
In natural biotopes, bacterial signalling can be compromised by many factors, whether abiotic or biotic; indeed, AHL accumulation relies on the stability, diffusion, sequestration, inactivation of AHL and the spatial distribution of cells. Moreover, the signalling process can undergo interferences such as cross-talk events and various molecules mainly produced by eukaryotes, suggesting that eukaryotes have evolved strategies to interfere with bacterial signalling in order to protect themselves from pathogenic bacteria. Hence, it is clear that AHL signalling is far more than a matter of quorum.
Given the complexity of regulatory networks and the number of factors affecting QS signalling, modelling approaches are undertaken to study the transition to QS (Goryachev et al., 2005) or to follow the evolution of AHL signals in conjunction with the spatial distribution of cells and environmental factors (Hense et al., 2007). In this context, the concept of efficiency sensing, a concept that unifies both QS and diffusion sensing, was proposed; whereas QS postulates that bacteria sense their density to coordinate a concerted behaviour, diffusion sensing proposes that sensing is an autonomous activity used to detect mass-transfer limitation. Efficiency sensing would enable cells to sense cell density, diffusion limitation and cell distribution (clustering), and includes the potential for co-operation as clusters may help protect against interference by other species and cheaters (Hense et al., 2007).