Summary

Introduction

One of the tenets of microbiology is that bacteria sense changes in their environments and alter their behaviour in order to better survive. When a bacterium’s environment changes, the response is often a programmed alteration in gene expression that results in a better chance at survival. Pathogens are no different. What we consider pathogenesis can easily be viewed as bacteria simply adapting their behaviour to survive, and thrive, in a new environmental niche: the host organism. This pathogenic response, however, comes at a cost to invading bacteria. During pathogenesis, numerous virulence factors, effector proteins and, occasionally, toxins are produced and used in various ways to help the bacteria survive within the host organism. Enterohaemorrhagic Escherichia coli, for example, uses type III secretion to translocate effector molecules to the cytoplasm of intestinal cells during colonization. Brucella, on the other hand, use type IV secretion to translocate effector molecules into the host cell that enable bacterial survival within macrophage vacuoles after phagocytosis. The bacteria in both of these cases affect their environment in such a way that they can survive where most bacteria would perish. However, the secretion machinery used by these bacteria is comprised of dozens of proteins that must first be assembled within the bacterial membrane before it can be used to translocate effector proteins across the bacterial and host membranes. The energy required to synthesize and maintain these molecular machines makes it impractical for them to be present when not needed. As a result, the genes responsible for these tasks, often found in pathogenicity islands or on plasmids, are tightly controlled at the transcriptional level and are not expressed until the bacteria sense they are in an environment in which the energy expense required for pathogenesis is justified. Once the appropriate stimulus is received, transcription factors direct the transcription of these operons.

The environmental stimulus that triggers pathogenesis can come in a number of forms: temperature, pH and host-derived molecules, to name a few. In addition to these factors contributed by the host, many pathogens can also detect signals originating from other bacteria. In the process known as quorum sensing, the bacterium produces a signal, called autoinducer, which is released into the environment. When the autoinducer reaches a critical concentration due to an increase in bacterial population, it triggers changes in gene expression resulting in a shift in the bacterium’s life cycle and metabolism. In the case of some pathogenic bacteria, quorum sensing is used to detect when the community to which it belongs has reached a population large enough to make a virulent lifestyle worthwhile, such as when biofilm formation is required. However, it is becoming clear that autoinducer-based signalling systems that do not fit into the strict quorum sensing model are also used by pathogenic bacteria. An example of this is when pathogenic bacteria use signalling molecules to detect the presence of host flora, thereby indicating that the pathogen is in a location suitable for colonization. In such cases, the detected signal is not provided by the invading bacterium, but rather by an established, high-density population. As a result, the invading bacterium does not need to produce and release signal, only to detect it.

To date, there are four main categories of cell-to-cell signalling systems. Two of these systems, which use autoinducer-1 (AI-1) and autoinducer-3 (AI-3), are found in Gram-negative cells while the Gram-positive cells use an autoinducing polypeptide (AIP) system. The fourth system, using autoinducer-2 (AI-2), is found in Gram-positive and Gram-negative cells and might represent a generalized signalling system. All of these systems begin with production and release of autoinducer into the environment, either by the pathogen or by the resident flora. The detection of these chemically distinct autoinducers, and the resulting alteration in gene expression, is specific to each system.

AI-1/LuxIR-based system

Since its initial discovery as the reason behind light production in Vibrio fisheri, the LuxI/LuxR system has become the model system of quorum sensing upon which the other systems have been based. It consists of LuxI, which synthesizes an N-acyl homoserine lactone (AHL) called AI-1, and LuxR, a transcription factor responsible for controlling gene expression in the presence of the autoinducer (Fig. 1A). LuxI and its homologues synthesize auto-inducers by transferring a fatty acid chain from an acylated acyl carrier protein (ACP) to S-adenosylmethionine (SAM), releasing the AHL and methylthioadenosine (Schaefer et al., 1996). AHLs produced by LuxI homologues possess different fatty acid moieties due to recognition of specific ACPs by the synthases and results in genus- and species-specific signals. After synthesis by LuxI, AI-1 is free to diffuse across the bacterial membrane, thereby being released into the surrounding environment. Since each individual bacterium within the growing population is producing AI-1 simultaneously, the concentration of AI-1 in the environment rises as the population increases. At high population density, the local concentration of AI-1 is high enough to diffuse back into the cell, where LuxR then binds it. When bound to AI-1, LuxR activates the transcription the luxCDABEGH operon by binding Lux boxes located within the promoter (Devine et al., 1989). The product of this operon, luciferase, catalyses a chemical reaction that results in luminescence.

Open in a separate window

Fig. 1

Three autoinducing strategies in Gram-negative cells.

A. AI-1-based strategy using acylated homoserine lactones (red diamonds). AI-1 is synthesized by LuxI and released into the environment. As AI-1 diffuses back into the cell, it interacts with LuxR and results in a cellular response.

B. AI-2-based strategy using either R-THMF (green pentagons) or furanosyl borate diester (blue pentagons), both created from LuxS-derived DPD. In Vibrio spp., furanosyl borate diesters interact with LuxP in the periplasm and initiates the LuxQ/LuxO phoshporyl cascade, but in other organisms R-THMF is imported into the cell and phosphorylated by the Lsr system.

C. AI-3-based strategy using an autoinducer of unknown structure (yellow hexagons). Autoinducer is released into the environment and sensed by QseC as it diffuses back into the periplasm. QseC then phosphorylates the response regulator QseB.

IM, inner membrane; OM, outer membrane.

Homologous LuxI/LuxR systems have been identified in many Gram-negative bacteria, each capable of producing specific AHLs. In some bacteria, such as Pseudomonas aeruginosa and Serratia marcescens, both opportunistic pathogens, these signalling mechanisms control pathways responsible for expressing various virulence factors. Both of these organisms are capable of forming biofilms on medical implants and catheters, possibly resulting in persistent infections in hospital patients. Due to the nature of biofilms, these infections are less susceptible to treatment by antibiotics than they would be normally as individual cells. P. aeruginosa contains two systems homologous to LuxI/LuxR. LasI/LasR has been shown to control biofilm formation and the production of extracellular enzymes, as well as transcription of another quorum sensing system, RhlI/RhlR, adding an additional level of control through AHL signalling (De Kievit and Iglewski, 2000). Both systems play a role in virulence as lasI and rhlI strains, either as single or as double mutants, are less successful in colonizing the lungs of neonatal mice than the isogenic wild-type strain (Pearson et al., 2000). S. marcescens also has a homologous LuxI/LuxR system. This system, in the presence of the Serratia-produced AHL, has been implicated in the control of numerous pathways linked to pathogenesis in S. marcescens clinical strain 12 (Sma 12). When compared with wild-type Sma 12, a strain lacking the AHL synthase SmaI showed decreased swarming motility, less cell adhesion (presumed to be a defect in biofilm formation), and reduced secretion of extracellular caseinase, chitinase and haemolysin (Coulthurst et al., 2006). These defects were all corrected by addition of AHL, indicating that these traits are influenced, either directly or indirectly, through cell-to-cell signalling.

The intracellular pathogen Brucella melitensis uses AHL to regulate two macromolecular machines required during pathogenesis: a polar flagellum and a type IV secretion system (Delrue et al., 2005). However, AHL regulation of the operons responsible for these machines differs from the prototypic LuxI/LuxR system described above. The B. melitensis LuxR-like transcriptional regulator, VjbR, activates genes associated with flagellar biosynthesis and type IV secretion in the absence of Brucella-specific AHL. In the presence of AHL, VjbR is inactive and the genes are expressed at a lower level. Therefore, an increase in population density results in decreased expression of these virulence factors. The role of these two macromolecular machines in pathogenesis is still unclear, but quorum sensing appears to play a role in coordinating the expression of these complexes either during the onset of infection or during the intracellular life cycle of the bacterium. Interestingly, although a Brucella-specific AHL has been identified, the Brucella chromosome does not contain a LuxI homologue.

Although early studies suggested that these systems were designed to monitor the population density of an organism’s own population, it is becoming increasingly clear that this is not always the case. E. coli and Salmonella enterica serovar Typhimurium (S. Typhimurium), for example, do not have a LuxI homologue and therefore do not produce any AI-1, but they do encode a LuxR homologue named SdiA that when overexpressed has been shown to have a negative effect on genes involved in cell attachment in enterohaemorrhagic E. coli (EHEC) (Kanamaru et al., 2000), but positively regulates several genes located on the S. Typhimurium virulence plasmid including rck, a protein implicated in evasion of the host immune response (Ahmer et al., 1998). Although the exact role SdiA plays in pathogenesis is unclear, this protein allows EHEC and S. Typhimurium to alter gene expression in response to the presence of AI-1 produced by other bacteria (Michael et al., 2001). This would be of particular use to these two pathogens since they both colonize within the gastrointestinal (GI) tract where AI-1, produced by host’s natural flora, may be present.

AI-2/LuxS-based system

Although considered the prototypical mechanism for cell-to-cell signalling, the AI-1 system is not the only strategy available. Many bacteria, Gram-negative and Gram-positive alike, have a system that detects an extracellular signal named autoinducer-2 (AI-2). This signal is different from the AHLs described above and is synthesized from a by-product of SAM metabolism. LuxS converts ribosehomocysteine into homocysteine and 4,5-dihydroxy-2,3-pentanedione (DPD), a compound that, in the presence of water, cyclizes into several furanones (Schauder et al., 2001). The structure of two AI-2 signals have been determined by co-crystalization with two different AI-2-binding proteins: a furanosyl borate diester used by Vibrio harveyi to control luminescence, and a furanone [(2R,4SL)-2-methyl-2,3,3,4-tetrahydroxytetrahydrofuran (R-THMF)] used by S. Typhimurium (Chen et al., 2002; Miller et al., 2004).

Like AI-1, AI-2 in either form is released by the bacterium where it accumulates in the cell’s environment. Detection of AI-2, however, is different from AI-1 detection. In fact, two separate mechanisms resulting in the detection of AI-2 have been described. The first, described in V. harveyi, is specific to the borate diester form of AI-2. This mechanism detects the presence of AI-2 in the periplasm by first binding the signal with an autoinducer-specific binding protein, LuxP. This AI-2/LuxP complex then interacts with a sensor kinase, LuxQ, initiating a phospho-transfer cascade that results, through deactivation of the negative response regulator LuxO, in luciferase production and luminescence (Fig. 1B). In pathogenic Vibrio cholerae LuxO is believed to repress the production of transcriptional regulator HapR at low cell density, which in turn represses production of several virulence factors (Zhu et al., 2002). As a result, this bacterium has the counterintuitive trait of expressing virulence factors at low population density while down-regulating these virulence factors as the level of autoinducer increases.

To date, the LuxP/LuxQ cascade has only been identified in Vibrio spp. AI-2 is handled by a different mechanism in E. coli and S. Typhimurium. These organisms import AI-2 by a mechanism homologous to ribose import by the rbs system. Unlike the LuxP/LuxQ system, the Lsr (LuxSregulated) system induces a cellular response by transporting AI-2 into the cytoplasm of the cell. This process starts with recognition of the signal by a periplasmic protein, LsrB, which binds the R-THMF form of AI-2. Once bound, the Lsr ABC transporter, comprised of LsrA and LsrC, imports AI-2 into the cell where LsrK phosphorylates it. It is believed that the phosphorylated form of AI-2 interacts with the transcriptional repressor LsrR to relieve repression of the lsr operon (Taga et al., 2001). Although it has not been shown directly, it is possible that this mechanism, or one similar to it, might result in the upregulation of additional operons in the presence of phosphorylated AI-2.

LuxS/AI-2 systems have been discovered in a wide range of Gram-positive and Gram-negative bacteria leading to the suggestion that the AI-2 system represents a method of cross-species signalling (Xavier and Bassler, 2003). Considering the mixed nature of most natural bacterial communities, such as that found in the GI tract of mammals, this common signal might be used to regulate genes needed to survive in the presence of other bacteria. Along these same lines, pathogenic bacteria may use these same signals to activate genes that would allow it to gain an advantage in this same environment. However, the role of AI-2 in non-Vibrio pathogenesis is unclear. Early studies using either luxS mutants or preconditioned media reported increased expression of flagella and type III secretion machinery in EHEC only under conditions where autoinducer was present (Sperandio et al., 1999; 2001). However, it should be noted that a direct influence on these genes by phosphorylated AI-2 or an AI-2-bound response regulator was never shown. In fact, the discovery of a third autoinducer, described below in more detail, presents the possibility that upregulation of these genes was due to a compound other than AI-2 in the preconditioned media. The data from the luxS mutant might have been misleading, as well. Because LuxS is part of the pathway responsible for synthesis of SAM and methionine, a luxS mutant would be at a metabolic disadvantage compared with the wild-type strain. The lack of available methionine in the cell caused by this break in a metabolic pathway might have pleotropic effects caused by unreliable protein translation (Winzer et al., 2002).

AI-3/QseC system

Of the three Gram-negative cell-to-cell signalling systems described to date, the AI-3 system is the least studied. AI-3 was first described as a compound found in spent media, different from AI-2, which activated the expression of genes involved in attachment of EHEC to, and subsequent actin rearrangement in, eukaryotic cells (Sperandio et al., 2003). The structure and synthesis of this signal is unclear and might, in fact, represent a family of molecules as AI-1 does. Early data suggested that LuxS was, once again, involved in some way because AI-3 production is impaired in luxS mutants. Further studies, however, proved that the lack of AI-3 production in these mutants was caused by a shift in the cell’s metabolism as it uses oxaloacetate, instead of SAM, as a precursor for methionine and that addition of L-aspartate to the growth medium, thereby relieving the demand for oxaloacetate, restored AI-3 production but had no effect on AI-2 production (Walters et al., 2006). This study also found that a number of commensal bacteria such as non-pathogenic E. coli and Enterobacter cloacae, as well as pathogenic Shigella, Salmonella and Klebsiella species, all produce AI-3. This suggests that AI-3 might represent another cross-species signal but, unlike AI-2, this new signal has not yet been detected in Gram-positive bacteria. However, the role of AI-3 in commensal communities has not yet been determined.

Detection of AI-3 is accomplished through a two-component system comprised of the sensor kinase QseC and response regulator QseB (Fig. 1C). In the presence of periplasmic AI-3, QseC first undergoes autophosphorylation and then transfers this phosphate to QseB, which activates genes responsible for flagella biosynthesis and motility by upregulating the master flagellar regulator genes flhDC (Clarke et al., 2006). As mentioned above the presence of AI-3 is also linked to the formation of attaching and effacing (AE) lesions by EHEC, a feat accomplished through upregulation of five separate loci of enterocyte effacement (LEE) operons located within the EHEC chromosome (Sperandio et al., 2003). The complete cascade responsible for regulation of these genes remains unclear, but likely involves QseA, a LysR-family regulator that is influenced by cell-to-cell signalling and directly upregulates LEE genes (Sperandio et al., 2002). Based on these observations, it has been proposed that enteric pathogens might use AI-3 produced by the host’s flora as an indicator that they are in a habitable location in the GI tract and that upregulation of flagella and motility are needed to penetrate the mucosal lining of the colon in order to reach underlying epithelial cells. Once in contact with epithelial cells, proteins encoded by the LEE genes enable the pathogen to attach to the eukaryotic cells for the purpose of colonization.

Intriguingly, the QseBC cascade also responds to the adrenergic signals adrenaline and noradrenaline, both found in the GI tract (Clarke et al., 2006). This suggests that QseC responds to bacterial and host signals simultaneously. Conversely, this presents the possibility that adrenergic receptors on eukaryotic cells might respond to AI-3 as well as adrenaline or noradrenaline. If this is the case, enteric bacteria might take advantage of this communication to prepare epithelial cells of the host GI tract for colonization, thereby preconditioning the GI tract for subsequent colonization by pathogens.

AIP/Agr system

This cell-to-cell signalling system, found exclusively in Gram-positive bacteria, is based on the prototypic Agr system first identified in Staphylococcus aureus. Unlike the previously described systems, the Gram-positive system uses a polypeptide signal instead of a smaller molecule (Fig. 2). These polypeptides play a dual role: they act as an autoinducer for the organism that produced it but an inhibitor to other organisms. This signal is referred to as an AIP and is encoded by the agrD gene. After translation, the AgrD propeptide is targeted to the membrane by an N-terminal signal sequence. Once at the membrane, AgrB, a membrane-bound endopeptidase, cleaves the C-terminus of the propeptide. The N-terminus of the propeptide, including the signal sequence, is removed by the signal peptidase SpsB. Finally, the C-terminus of this processed polypeptide is covalently linked to a conserved, centrally located cysteine to form a thiolactone ring with a free N-terminal tail. In many cases, both of these structures are required for proper function of the AIP. After release into the environment, the AIP is recognized by a signal receptor, AgrC. This protein contains an N-terminal transmembrane domain responsible for recognizing specific AIPs and a C-terminal histidine kinase domain that, in the presence of the correct AIP, phosphorylates a response regulator called AgrA. Phosphorylated AgrA activates transcription of select genes by binding direct repeats located in the promoter regions. One aspect that is unique to the AIP/Agr system is the fact that an AIP produced by one strain of Staphylococcus will simultaneously interfere with another strain’s Agr system. This dual role as activator and inhibitor is related to the interaction between the AIP and ArgC. The cyclic structure of AIP is required for interaction with AgrC, but it is the N-terminal tail that is responsible for AgrC activation. Removal of this tail, in fact, results in a universal inhibitor that binds to AgrC but is unable to activate the Agr system (Lyon et al., 2000).

Open in a separate window

Fig. 2

AIP-based autoinduction and inhibition in Gram-positive cells using polypeptides (green hexagons). Propeptide, encoded by AgrD, is processed by AgrB and released into the environment where it can either activate the AgrA/AgrC cascade or inhibit the same cascade of a different species. M, membrane; P, peptidoglycan.

The Agr system has been linked to pathogenesis in numerous Gram-positive organisms. For example, S. aureus strains lacking the Agr system are impaired in their ability to cause osteomyelitis in both rabbit and murine models (Gillaspy et al., 1995; Blevins et al., 2003). AIP-directed regulation of genes by AgrA results in production and release of numerous toxins by S. aureus such as alpha-, beta- and delta-haemolysins, serine proteases and toxic shock syndrome toxin 1 (TSST-1). The release of haemolysin and protease disrupts localized eukaryotic cells and aids in Staphylococcal adhesion whereas TSST-1 is a superantigen that binds non-specifically to T-cell receptors ultimately resulting in a strong immunological response due to systemic release of cytokines. In addition to upregulation of toxins, however, high concentration of AIP is also responsible for downregulation of surface-exposed proteins such as fibronectin-binding protein and protein A, two proteins involved in Staphylococcal adhesion. This reduction in cell adhesion would have a negative effect on the architecture of any biofilm-associated communities. As a result, the bacterial cells are released from their communities and are free to migrate to other locations within the host. Once a new, suitable location is found, the cell can reattach and begin multiplying again, possibly resulting in a secondary infection.

Enterococcus faecalis is another Gram-positive cell that benefits from the use of AIP signal sensing. This organism uses a two-component system homologous to Agr system of Staphylococcus to detect the presence of AIP. When AIP is detected, the cell produces and releases two extracellular proteases: gelatinase (GelE) and serine protease (SprE) (Qin et al., 2000). The role of these two proteins in pathogensis is still unclear, but gelatinase has been implicated in cell migration within hosts. In an in vitro assay measuring E. faecalis translocation across a monolayer of polarized human colon cells, the presence of gelatinase was found to be important (Zeng et al., 2005). Translocation of E. faecalis across the epithelial cells of the colon might allow this organism to enter the vascular or lymphatic systems and migrate to other locations within the host where secondary infections might occur.

Interestingly, the AIP-based system is not the only quorum sensing system used by E. faecalis. Cytolysin production by this bacterium is repressed at a transcriptional level by two proteins: a membrane-bound sensor, CylR1, and a DNA-binding protein, CylR2 (Haas et al., 2002). When the concentration of cytolysin outside the cell is low, CylR2 represses the genes encoding the cytolysin subunits, cylL_L and cylL_S, by binding the cylL promoter. CylR1 monitors the concentration of cytolysin outside the cell and when a threshold concentration is reached, CylR1 alters the DNA-binding capacity of CylR2. This results in the release of CylR2 from the cylL promoter and an increase in cytolysin production. As a result, cytolysin is upregulated in a population-dependent manner through its own detection rather than the through the detection of AIP.

Concluding remarks

It is becoming more and more clear that interbacterial communication plays an important role in bacterial interactions with host organisms. This is true for both natural, symbiotic flora and invading pathogenic bacteria. In the case of host flora, signalling molecules are used to monitor population densities to better coordinate population-wide responses, often resulting in the formation of complex biofilms. Some pathogens also use signalling molecules to coordinate population-dependent responses in cases where large number of bacteria are required for colonization, but can also use signal molecules produced by other populations, such as the existing flora, as an indicator of their location within the host. Using this strategy, invading bacteria switch on pathogenic genes only when they are in an environment where pathogenesis is likely to succeed. This insures that the bacterium does not expend the extra energy required for pathogenesis unless it is beneficial to do so.

Acknowledgements

References