Pathogen espionage: multiple bacterial adrenergic sensors eavesdrop on host communication systems


  • Michail H. Karavolos,

    Corresponding author
    • Centre for Bacterial Cell Biology, Institute for Cell and Molecular Biosciences, The Medical School, Newcastle University, Newcastle, UK
    Search for more papers by this author
  • Klaus Winzer,

    1. School of Molecular Medical Sciences, Centre for Biomolecular Sciences, University of Nottingham, Nottingham, UK
    Search for more papers by this author
  • Paul Williams,

    1. School of Molecular Medical Sciences, Centre for Biomolecular Sciences, University of Nottingham, Nottingham, UK
    Search for more papers by this author
  • C. M. Anjam Khan

    Corresponding author
    • Centre for Bacterial Cell Biology, Institute for Cell and Molecular Biosciences, The Medical School, Newcastle University, Newcastle, UK
    Search for more papers by this author

For correspondence. E-mail; Tel. (+44) 191 222 7066; Fax (+44) 191 222 7424;

E-mail; Tel. (+44) 191 222 8147; Fax (+44) 191 222 7736.


The interactions between bacterial pathogens and their eukaryotic hosts are vital in determining the outcome of infections. Bacterial pathogens employ molecular sensors to detect and facilitate adaptation to changes in their niche. The sensing of these extracellular signals enables the pathogen to navigate within mammalian hosts. Intercellular bacterial communication is facilitated by the production and sensing of autoinducer (AI) molecules via quorum sensing. More recently, AI-3 and the host neuroendocrine (NE) hormones adrenaline and noradrenaline were reported to display cross-talk for the activation of the same signalling pathways. Remarkably, there is increasing evidence to suggest that enteric bacteria sense and respond to the host NE stress hormones adrenaline and noradrenaline to modulate virulence. These responses can be inhibited by α and β-adrenergic receptor antagonists implying a bacterial receptor-based sensing and signalling cascade. In Escherichia coli O157:H7 and Salmonella, QseC has been proposed as the adrenergic receptor. Strikingly, there is an increasing body of evidence that not all the bacterial adrenergic responses require signalling through QseC. Here we provide additional hypotheses to reconcile these observations implicating the existence of alternative adrenergic receptors including BasS, QseE and CpxA and their associated signalling cascades with major roles in interkingdom communication.


Prokaryotes and eukaryotes have intimately cohabited together in both commensal and pathogenic relationships. Consequently, during these complex interactions bacteria have encountered host hormones, and at the same time their eukaryotic hosts have been exposed to bacterial autoinducers. Through the pioneering work of Mark Lyte and colleagues, it has become recognized that organisms from different kingdoms have coevolved and acquired the tools to sense and respond to such signalling molecules, which are present within different niches (Lyte and Ernst, 1993). These interkingdom signalling processes between microbes and their hosts are central to the emerging field of ‘microbial endocrinology’ (Lyte, 1993; Lyte and Ernst, 1993). In recent years we have gained a better understanding of intrabacterial communication and also have begun to decipher various forms of interkingdom linguistics involving bacterial pathogens and their mammalian hosts. The molecules enabling this specialized communication may represent ancestral signalling espionage tools acquired and developed during co-evolution.

Indeed, bacterial pathogens use an array of molecular sensors to perceive and facilitate adaptation to changes in their environment. Mechanisms which allow bacterial pathogens to eavesdrop on mammalian host signalling systems such as neuroendocrine (NE) stress hormones may aid towards their successful adaptation and survival within the host (Lyte et al., 1996; 1997; Lyte, 2004; Pacheco and Sperandio, 2009). Upon entering the host, pathogens come in contact with a wide range of chemical signals including the NE stress hormone noradrenaline which is abundant in the gut as well as adrenaline which is found mostly in the bloodstream (Aneman et al., 1996; Eisenhofer et al., 1996; Furness, 2000). Interestingly, bacterial lipopolysaccharide induces the release of adrenaline and noradrenaline by macrophages in the bloodstream (Flierl et al., 2007; 2009). It has therefore been suggested that the phagocytic system represents a diffusely expressed adrenergic organ (Flierl et al., 2007; 2009).

Faced with such a wide repertoire of host signals and environments, bacterial populations employ collective decision making or ‘quorum sensing’ (QS) to synchronize their responses efficiently, and circumvent host defences. QS systems depend on the production and sensing of diffusible signal molecules, sometimes referred to as ‘autoinducers’. Once a critical threshold concentration of autoinducer has been attained, a change in collective behaviour ensues through the activation of a sensor or regulator protein (Bassler et al., 1994; Fuqua et al., 1994; Rickard et al., 2006; Williams, 2007). Thus, organized bacterial attack against host defences ensures maximum chances of survival. Pathogen growth in serum containing media has long been known to be influenced by host NE stress hormones (Lyte, 2004; Freestone et al., 2008). Both adrenaline and noradrenaline are ferric iron chelators which promote the release of iron from the host iron transport proteins transferrin and lactoferrin, so enabling bacteria to acquire iron from the host (Freestone et al., 2000; 2008; Neal et al., 2001; Burton et al., 2002; Everest, 2007; Sandrini et al., 2010; Lyte et al., 2011). Host NE stress hormones have also been reported to mimic the effects of a bacterial autoinducer, termed AI-3, thus implying the existence of cross-talk involving both host and bacterial signal molecules (Sperandio et al., 2003).

The multi-faceted interplay between host and pathogen will determine the overall outcome of an infection. There is increasing evidence to suggest that bacteria can sense host NE stress hormones such as adrenaline and noradrenaline to modulate their virulence (Lyte, 1993; 2004; Lyte and Ernst, 1993; Sperandio et al., 2003; Freestone et al., 2008; Karavolos et al., 2008b; 2011a; Pacheco and Sperandio, 2009; Spencer et al., 2010; Lyte et al., 2011). In this MicroReview we will present the latest evidence supporting this interkingdom communication hypothesis and discuss the implications arising from the recent discovery of multiple bacterial adrenergic receptors for bacterial–host communication.

Quorum sensing and interkingdom communication

Quorum sensing is mediated via chemically diverse signal molecules which in Gram-negative bacteria include N-acylhomoserine lactones (AHLs), 2-alkyl-4-quinolones (AQs), fatty acids and furanones such as autoinducer-2 (AI-2) (Williams et al., 2007; Ng and Bassler, 2009; Pereira et al., 2012). QS provides an advantage to the bacterial population by facilitating the co-ordinated expression of mechanisms aiding bacterial survival, while simultaneously modulating metabolic fitness (Winzer and Williams, 2001; Winzer et al., 2003; Williams et al., 2007; Ng and Bassler, 2009; Pereira et al., 2012).

AHLs represent a class of autoinducers produced solely by Gram-negative bacteria which generally mediate signalling through activation of the LuxR family of transcriptional regulators although in Vibrio harveyi, the AHL, N-(3-hydroxybutanoyl)-homoserine lactone activates target gene expression via LuxN, a membrane-associated sensor kinase (Taga and Bassler, 2003; Xavier and Bassler, 2003). In Gram-positive bacteria, autoinducers are often peptides resulting from the cleavage of larger precursors, post-translationally modified and actively secreted. The secreted autoinducers are sensed through their interaction with membrane receptors via a classical two-component signal transduction system, or intracellularly following internalization by oligopeptides permeases (Dunny and Leonard, 1997; Lazazzera, 2000; Lyon and Novick, 2004; Ng and Bassler, 2009).

AI-2 is a signal molecule derived from the rearrangement of 4,5-dihydroxy-2,3-pentanedione (DPD) which is itself a by-product in a reaction catalysed by the enzyme LuxS (Surette et al., 1999). It has been suggested that different bacteria use different forms of DPD as AI-2 (Xavier and Bassler, 2005). For example, in Vibrio harveyi AI-2 is a furanosyl borate diester (Chen et al., 2002), while Salmonella enterica serovar Typhimurium (S. Typhimurium) employs a different form of AI-2, (2R,4S)-2-methyl-2,3,3,4-tetrahydroxytetrahydrofuran (R-THMF), lacking boron (Miller et al., 2004). In S. Typhimurium, AI-2 is thought to freely diffuse and accumulate extracellularly where upon reaching a critical concentration it is internalized and activated by phosphorylation via the Lsr system (Xavier and Bassler, 2005). LuxS-dependent AI-2 activity has been detected in spent culture supernatants of a wide variety of Gram-negative and Gram-positive bacteria leading to the hypothesis that AI-2 represents a ‘universal’ bacterial communication signal (Miller and Bassler, 2001; Xavier and Bassler, 2005). However, not all bacteria possess LuxS nor produce AI-2 (Winzer et al., 2002a). In a number of species studied so far LuxS/AI-2 have been reported to affect the regulation of genes encoding a wide variety of virulence factors, motility, cell division, antibiotic production, biofilm formation and carbohydrate metabolism (Sircili et al., 2004; Vendeville et al., 2005; Ng and Bassler, 2009). However, since LuxS is a key metabolic enzyme involved in the activated methyl cycle not all of the phenotypes reported for luxS mutants are a consequence of the loss of AI-2 production (Winzer et al., 2002a,b; Redanz et al., 2012). For example, in Streptococcus sanguinis, complementation of the active methyl cycle via the gene encoding the S-adenosylhomocysteine hydrolase (SahH) restores biofilm formation independently of AI-2 and luxS (Redanz et al., 2012). In S. Typhimurium, AI-2 affects the expression of the Lsr system involved in its own uptake (Taga et al., 2003; Xavier and Bassler, 2003; De Keersmaecker et al., 2005). LuxS exhibits AI-2-independent activity in S. Typhimurium by modulating flagellar phase variation to favour expression of the more immunogenic phase-1 flagellin (Karavolos et al., 2008a; Kint et al., 2009). Formation of biofilms by S. Typhimurium is dependent on the small RNA micA whose expression is affected by luxS via a yet unknown mechanism (Kint et al., 2010). Depending on the stages of cell growth and presence of glucose, deletion of luxS may also exhibit pleiotropic effects on S. Typhimurium gene expression (Jesudhasan et al., 2010).

The LuxS enzyme due to its pleiotropic effects on bacterial metabolism has also been indirectly implicated in the production of another autoinducer (AI-3) in Escherichia coli and also Salmonella (Sperandio et al., 2003; Walters et al., 2006). In enterohaemorrhagic E. coli (EHEC), AI-3 was reported to act synergistically with adrenaline and noradrenaline to regulate motility and virulence via the two-component signal transduction systems QseBC and QseEF (Rasko et al., 2008; Pacheco and Sperandio, 2009).

In what will undoubtedly prove to be a landmark publication, Asano et al. have just demonstrated that bacteria in the mouse gut microbiome can produce biologically active catecholamine hormones (Asano et al., 2012). These observations vividly show that sufficient levels of NE stress hormones are present within the gastrointestinal tract to influence both the host and resident microbiota. Hence, it is likely that, the interactions of autoinducers and NE stress hormones with bacterial and host adrenergic receptors contribute to the maintenance of microbial endocrine homeostasis and avoidance of dysbiosis.

Although it is over a decade since AI-3 was first described by the Sperandio laboratory, neither its structure nor the identity of the corresponding synthase are known and no other groups have, to our knowledge, published data on this elusive signal molecule. We have not been able to detect AI-3 in Salmonella spent media using the published methodology and a variety of different growth conditions. This discrepancy could be due to subtle differences in growth conditions or even strain variation. Given that noradrenaline and adrenaline are both catechols and in common with AI-3 activate QseBC and QseEF, it is likely that all three are structurally related. Indeed both Salmonella and E. coli produce the catecholate siderophore, enterobactin which is metabolized to release 2,3 dihydroxybenzoylserine (DHBS, Fig. 1) which can also induce expression of the AI-3 reporter (our unpublished data) as well as mimicking the effects of noradrenaline on bacterial growth in serum containing media (Burton et al., 2002).

Figure 1.

Chemical structures of the key ligands involved in bacterial adrenergic signalling. The hormones adrenaline and noradrenaline; the adrenergic blockers propranolol (β-blocker), phentolamine (α-blocker) and the bacterial breakdown product of siderophore enterobactin, dihydroxybenzoylserine (DHBS) are shown. LED209 is a QseC inhibitor that blocks both noradrenaline- and adrenaline-triggered QseC-dependent virulence gene expression. Chemical structures were drawn using ChemSpider (

Bacterial adrenergic sensors – signalling evolution

Mammalian adrenergic sensors are seven transmembrane helix G protein-coupled receptors (GPCRs) that bind host NE hormones noradrenaline and adrenaline (Kroeze et al., 2003). Adrenaline and noradrenaline associate with α1, α2 or β-adrenergic receptors. Receptor α1 couples to the Gq G-protein family, resulting in increased intracellular Ca2+ and smooth muscle contraction. Receptor α2, couples to the Gi G-protein family decreasing cAMP levels and resulting in smooth muscle contraction. The β receptors, on the other hand, couple to Gs G-protein family and increase intracellular cAMP activity, leading to heart muscle contraction, smooth muscle relaxation and glycogenolysis (Kroeze et al., 2003).

Interestingly, there is evidence to support the horizontal gene transfer of messenger, including adrenergic, signalling pathways between bacteria and ancestral eukaryotes (Lakshminarayan et al., 2004; Dunning Hotopp, 2011). Using bioinformatics there are no significant amino acid sequence similarities between mammalian adrenergic receptors and bacterial histidine kinase sensors involved in NE hormone sensing. However, direct binding of NE stress hormones has been demonstrated for the bacterial adrenergic sensor QseC of the bacterial two-component signal transduction system QseBC (Clarke et al., 2006), and suggests there are conserved structural motifs present in bacterial adrenergic receptors. Inhibition of binding to QseC can be brought about by mammalian α adrenergic blockers as well as synthetic molecules like LED209 (Fig. 1) signifying additional similarities in binding sites and modes of inhibition (Rasko et al., 2008). It is worth noting that LED209 will not block signalling through human β adrenergic receptor isoforms (Rasko et al., 2008).

Although, the roles of QseC and QseE have been well characterized in various pathogens there are increasing reports of alternative receptors which participate in host–pathogen interactions through NE hormones including the QseE, BasS and CpxA sensors of the QseEF, BasSR and CpxAR two-component signal transduction systems respectively (Reading et al., 2007; Karavolos et al., 2008b; 2011a; Spencer et al., 2010).

In silico analysis of selected E. coli and Salmonella putative bacterial adrenergic sensors reveals the presence of domains involved in ATP hydrolysis and signalling via the common histidine kinase mechanism (Fig. 2). There are no differences in sensor size with the exception of BasS which is significantly smaller (Fig. 2, Table 1). All sensors possess a HAMP (Histidine kinases, Adenylyl cyclases, Methyl-binding proteins, Phosphatases) linker domain which is closely associated with the histidine kinase region. Furthermore, all sensors carry a signal peptide apart from QseE in Salmonella. Differences are reflected in the presence and number of low complexity regions associated with the tertiary structure of these protein areas (Fig. 2). Despite extensive domain similarities, extensive bioinformatics analysis using BLAST-EXPLORER (Dereeper et al., 2008; 2010) ( reveals significant differences in the phylogenetic profile of these bacterial adrenergic sensors (Fig. 3). These differences are also reflected in the pathways controlled by these sensors and imply the existence of multiple bacterial adrenergic receptors in pathogens (discussed below).

Figure 2.

In silico sequence analysis of putative bacterial adrenergic receptor proteins. Comparisons involved domains of the respective receptors in E. coli O157:H7 (EHEC) and S. Typhi (Ty2). The domains shown in the diagram correspond to HATPase_c – histidine kinase-like ATPases; HisKA – His kinase A (phosphoacceptor); HAMP – Histidine kinases, Adenylyl cyclases, Methyl-binding proteins, Phosphatases; Pfam HAMP – HAMP linker domain found using the Pfam database; and BLAST – HAMP linker domain found using the Schnipsel database. Alignments were conducted using the SMART database (

Figure 3.

Phylogenetic tree of candidate bacterial adrenergic receptors. Similarities in the amino acid sequence of putative receptors were determined via multiple sequence alignment and analysis (BLAST-EXPLORER; All sequences analysed were obtained from the publically available E. coli O157:H7 (EHEC) or S. Typhi (Ty2) genome sequence. Numbers in red indicate branch confidence levels.

Table 1. List of known putative bacterial adrenergic membrane receptors
NameSize (kDa)TMFunctionHormoneBlockerReference
  1. The information includes protein size, number of transmembrane (TM) segments, function, possible association with host NE stress hormones adrenaline (A) or noradrenaline (NA) as well as susceptibility to blocking by known adrenergic blockers. The bacterial TonB-dependent (TBD) transport system for catecholamines is also displayed for comparison. In silico analysis for the prediction of TM segments on proteins were accomplished using the TMHMM Server v2.0 (
TBDVariousVariousGrowth enhancementA/NAαFreestone et al. (2007)
QseC50.62VirulenceA/NAαClarke et al. (2006)
QseE51.62VirulenceAαReading et al. (2007)
BasS40.22Antimicrobial peptide resistanceA/NAβKaravolos et al. (2008b)
CpxA51.62Envelope stress/haemolysisA/NAβKaravolos et al. (2011a)

The QseC and QseE sensors

In EHEC adrenaline and noradrenaline can substitute for the bacterial autoinducer AI-3 implying the existence of cross-talk between the two signalling systems (Sperandio et al., 2003). This observation raised the possibility of the presence of adrenergic receptors in bacteria (Sperandio et al., 2003).

Indeed, the sensor kinase QseC is autophosphorylated on binding either adrenaline or noradrenaline, demonstrating the existence of adrenergic sensors in bacteria (Clarke et al., 2006). Furthermore, these responses can be inhibited by mammalian adrenergic antagonists (Table 1). Remarkably, there is strong specificity in the antagonistic effects, with QseC only being blocked by the α-adrenergic antagonist phentolamine (Clarke and Sperandio, 2005). In EHEC, high-throughput screening of a small library of organic compounds, yielded, LED209, a sulphonamide derivative which was reported to inhibit the binding of NE to QseC (Rasko et al., 2008). In S. Typhimurium, although the importance of QseC in virulence has been demonstrated no direct binding of NE stress hormones was shown (Moreira et al., 2010). Hence, in EHEC and Salmonella, the QseC sensor has been proposed as the adrenergic receptor.

The role of the QseEF system in conjunction with QseBC has also been investigated extensively in EHEC (Reading et al., 2007; Njoroge and Sperandio, 2012). QseC and QseE regulate expression of the locus of enterocyte effacement (LEE) genes in an adrenaline-dependent manner (Njoroge and Sperandio, 2012).

However, there is an increasing body of evidence from independent research groups which supports the existence of alternative bacterial adrenergic receptors. For example, it has been vividly demonstrated that QseBC does not play a role in the noradrenaline-enhanced enteritis or intestinal colonization in calves in S. Typhimurium (Pullinger et al., 2010a).

The putative BasS sensor

In S. Typhimurium the BasSR two-component system controls expression of the pmr locus and is implicated in the regulation of a variety of genes (Gunn et al., 1998; 2000; Marchal et al., 2004). The pmrHFIJKLM operon encodes a set of proteins involved in lipopolysaccharide modification and resistance to the cationic antimicrobial polypeptide polymyxin B (Noland et al., 2002; Yan et al., 2007).

Pre-treatment of Salmonella with adrenaline results in a significant reduction in bacterial survival during exposure to polymyxin B, an effect which is fully reversible by the β-adrenergic blocker propranolol (Karavolos et al., 2008b). This effect is dependent on the BasSR two-component signal transduction system which is the likely putative adrenaline sensor mediating the antimicrobial peptide response (Karavolos et al., 2008b). Adrenaline may exert its effect on the pmr locus of S. Typhimurium by signal transfer following interactions with the BasS membrane sensor, in a manner similar to the interaction of adrenaline with QseC in E. coli (Table 1). Direct binding of NE hormones to BasS remains to be demonstrated.

The low amino acid sequence identity (31%; between BasS and QseC implies the existence of conserved tertiary structural motifs produced by different sequences which are involved in the sensing and inhibition by adrenergic activators and blockers. Hence this difference may provide an explanation as to why we observe β-blockage in BasS as opposed to α-blockage in QseC (Karavolos et al., 2011a). The direct and reversible reduction of bacterial antimicrobial peptide resistance by a mammalian NE stress hormone via BasSR suggests a novel ‘antibacterial’ role for adrenaline of benefit to the host (Karavolos et al., 2008b).

The putative CpxA sensor

The cpxAR two-component signal transduction system is involved in regulating the expression of a broad range of genes following exposure to environmental signals and also during exposure to diverse envelope stresses. Further, CpxAR is important for the pathogenicity of Salmonella in vivo (Humphreys et al., 2004).

The exclusively human pathogen S. enterica serovar Typhi (S. Typhi) encodes the HlyE haemolysin pore-forming toxin but is non-haemolytic on blood agar plates. Remarkably, exposure of S. Typhi to the NE stress hormones induces haemolysis on blood agar plates (Karavolos et al., 2011a).

The inhibition of NE stress hormone-mediated haemolysis by the adrenergic β-blocker propranolol in S. Typhi is another example of the existence of an additional putative novel bacterial adrenergic receptor. In S. Typhi, the NE stress hormone-mediated haemolytic phenotype is independent of the known E. coli O157:H7 adrenergic receptor QseBC and is mediated by the CpxAR two-component system (Karavolos et al., 2011a,b). Furthermore, haemolysis is not blockable by the α-adrenergic antagonist phentolamine and only by the adrenergic β-blocker propranolol (Karavolos et al., 2011a,b) (Table 1). It has to be noted that, similar to BasS, direct binding of NE hormones to CpxA remains to be demonstrated. However the phenotypes have been convincingly shown to be independent of QseC.

Emerging systems

There are constantly new reports of adrenergic involvement in a variety of bacterial processes (Peterson et al., 2011; Li et al., 2012). In another example of adrenergic receptor antagonist inhibition, increased expression of virK and mig14 in S. Typhimurium was reversed by the addition of the β-adrenergic antagonist propranolol (Spencer et al., 2010). Some adrenergic phenotypes in bacteria are associated with altered iron uptake via the siderophore enterobactin and its DHBS breakdown product (Burton et al., 2002; Freestone et al., 2003). A tonB mutant, defective in siderophore uptake, showed the same differential gene regulation upon exposure to NE stress hormones as the parent strain, suggesting that the adrenergic regulation is mediated through a mechanism independent of TonB.

Furthermore, in S. Typhimurium, QseBC, QseEF, CpxAR or BasSR are not involved the adrenergic signalling cascade leading to increased sensitivity to the antimicrobial peptide LL-37 (Spencer et al., 2010). Additionally, the LED209 inhibitor failed to affect these signalling pathways implying the existence of different adrenergic sensory mechanisms to QseC mediating this response (Spencer et al., 2010).

Bacterial adrenergic sensors – role in virulence

In EHEC adrenaline and noradrenaline signalling affects bacterial virulence and motility (Pacheco and Sperandio, 2009). In EHEC the membrane sensor QseC acts as a bacterial adrenergic receptor to activate virulence genes in response to interkingdom communication (Clarke et al., 2006; Rasko et al., 2008). It was also reported that S. Typhimurium lacking qseC were defective in motility and showed decreased colonization of the gastrointestinal tract in pigs (Bearson and Bearson, 2008). Consequently the qseC mutant was attenuated for systemic infection in 129x1/SvJ mice (Moreira et al., 2010). More recently, a new set of data highlights the importance of QseC and QseE whereby co-incubation of HeLa cells with adrenaline increases EHEC infectivity in a QseC- and QseE-dependent manner (Njoroge and Sperandio, 2012). Furthermore, the QseC inhibitor LED209, attenuated Salmonella virulence in mice, an effect that was attributed to the reduction in expression of the flhDC flagella regulator as well as the sifA virulence factor (Rasko et al., 2008). However, we have been unable to detect an LED209-induced change in flhDC or sifA expression either by qPCR or by luminescence expression analysis (our unpublished data) (Spencer et al., 2010). Nevertheless, it would be necessary to show extreme caution on assessing bacterial pathogenicity since bacterial receptor-blocking synthetic compounds may also have pleiotropic effects on mammalian or rodent adrenergic receptors and such interaction would potentially skew the results of studies (Smith et al., 1977).

Several groups have made alternative observations on the importance of qseBC. For example, transcriptional profiling of a S. Typhimurium qseB/qseC mutants revealed significant differences to the above observations including motility and responsiveness to host NE stress hormones or quorum sensing signal molecules (Merighi et al., 2009). Indeed the lack of qseBC in S. Typhimurium did not significantly affect expression of the flagellar operon genes and, importantly, adrenaline increased S. Typhimurium motility independently of qseBC (Merighi et al., 2009).

Thus, emerging evidence suggests that the sensing of NE stress hormones by S. Typhimurium may not be exclusively mediated by QseC and QseE. Indeed, in S. Typhimurium QseBC, and also QseEF, do not contribute to the adrenergic responses leading to increased sensitivity to the antimicrobial peptides polymyxin B and LL-37 (Karavolos et al., 2008b; Spencer et al., 2010). Additionally, there is clear evidence that the QseC and QseE sensors are not required for noradrenaline-enhanced enteritis or intestinal colonization in calves (Pullinger et al., 2010a). In this model, S. Typhimurium-induced enteritis was significantly enhanced by addition of noradrenaline an effect which was associated with increased net bacterial replication (Pullinger et al., 2010a). Therefore, although NE stress hormones significantly impact on the establishment of enteritis in calves, they do so through alternative bacterial adrenergic sensors.

Such observations may also reflect differences in pathogenesis between S. Typhimurium, an invasive pathogen infecting macrophages and epithelial cells, and E. coli, a mainly non-invasive pathogen which remains in the host intestine. The significant divergence in niches occupied by these two pathogens requires different gene expression patterns for maximum infection efficiency; hence NE stress hormones may modulate different signalling pathways to the advantage or disadvantage of the pathogen.

Alternative strategies to investigate the importance of host produced NE hormones in the virulence of S. Typhimurium have been used by a number of groups. For example, transgenic mice defective in dopamine beta-hydroxylase, which are unable to produce adrenaline and noradrenaline, have been used as hosts. When they were infected with S. Typhimurium a modest positive effect on pathogenesis was evident suggesting that NE stress hormones reduce the virulence of S. Typhimurium or increase host resistance (Moreira et al., 2010). These transgenic mice also have defective Th1 T-cell responses important for immunity to intracellular pathogens and this may have complicated the findings. In a model of porcine salmonellosis, the effect upon bacterial virulence of injecting the neurotoxin 6-hydroxydopamine to force the release of noradrenaline was found to be mild and transient leading to increased faecal shedding of S. Typhimurium. Furthermore, the lack of qseC had no significant impact upon the outcome of the infection but reduced faecal shedding of qseC (Pullinger et al., 2010b).

The irregularities emerging in the importance of QseC and QseE in bacterial adrenergic sensing in Salmonella may reflect differences in in vitro growth conditions and also the specific host animal models employed. To add to this variation is the use of bacterial strains with differing genetic backgrounds. Therefore, any functional inferences proposed for the sensors are dependent on the in silico sequence similarities of the orthologues to qseC and qseE in EHEC, and extrapolating phenotypes. The de facto constantly changing levels of NE stress hormones also entails that the pathogen may have evolved sensors with varying levels of sensitivity. Moreover, one has to account for the pleiotropic phenotypic effects due to the ability of adrenaline and noradrenaline to modulate intracellular iron levels and hence growth (Freestone et al., 2008; Karavolos et al., 2008b). Such pleiotropic effects of qseC in EHEC and uropathogenic E. coli are described in a recent study which revealed perturbations in amino acid, nucleotide and carbon metabolism resulted in the downregulation of virulence factors (Hadjifrangiskou et al., 2011). Indeed, it was found that in the absence of QseC, the constitutively active QseB mediates the polyphenic effects on bacterial metabolism, and hence leads to virulence attenuation (Hadjifrangiskou et al., 2011).


Publications from a significant number of groups describe convincing observations that bacteria can sense and respond to neuroendocrine host stress signals. Collectively, these observations support the hypothesis that Salmonella and E. coli have evolved multiple specialized systems for directly sensing NE stress hormones (Lyte and Ernst, 1993; Lyte et al., 1997; 2011; Freestone et al., 2003; Hughes and Sperandio, 2008; Moreira et al., 2010; Njoroge and Sperandio, 2012). In EHEC expression of the LEE locus, Shiga toxin and motility genes appear to be regulated by NE stress hormones (Hughes and Sperandio, 2008). We have demonstrated that even a transient exposure of Salmonella to NE stress hormones can result in marked changes in bacterial physiology and expression of virulence factors (Karavolos et al., 2008b; 2011a,b; Spencer et al., 2010). A simplified illustration of the possible interactions of bacterial pathogens with mammalian NE stress hormones through membrane-bound bacterial adrenergic receptors is depicted in Fig. 4.

Figure 4.

Diagrammatic representation of signalling interactions between pathogenic bacteria and host NE stress hormones. NE hormones are naturally being produced by the host neuroendocrine system at synapses where they are sensed by G protein-coupled receptors (GPCR). The activated GPCR facilitates release of the G-protein β and γ subunit complex from α with subsequent GTP/GDP exchange and association with its effector partner (E). During infection, bacterial lipopolysaccharide (LPS) on the bacterial cell wall (CW) induces secretion of NE stress hormones by host neutrophils and macrophages. Bacterial adrenergic receptors (BARs) in the cell membrane (CM) are also able to sense and respond to these heterologous signals through altering the phosphorylation (P) status of the cognate response regulator (RR). NE hormone signalling is also accomplished independently of BARs via siderophore receptor proteins mediating the provision of iron and subsequent bacterial growth enhancement (Fe signalling).

The effects of adrenaline and noradrenaline are likely to be complex, involving multiple effects on both bacteria and possibly host gene expression signatures to actively influence the outcome of the infection. It is likely that the adrenergic modulation of target genes may confer an advantage to the bacteria under certain in vivo conditions but an unavoidable disadvantage in others; for example downregulation of the lipopolysaccharide (LPS) modifying enzymes PmrF and PagL causes an increase in sensitivity to polymyxin B, but also concomitantly reduces activation of the TLR-4 Toll-like receptors, reducing the host inflammatory response to infection (Kawasaki et al., 2004; Miller et al., 2005). Timing is also crucial for the efficient assimilation of host-derived signals during host infection. Indeed, it was shown that type 1 pili and QseC become critical in different infection stages, and hence these factors have an additive effect on weakening UPEC virulence (Kostakioti et al., 2012). The existence of multiple bacterial adrenergic receptors may hence reflect the culmination of many millions of years of receptor structural evolution to accommodate these signals and orchestrate the best survival response.

Thus, host NE stress hormones can provide vital environmental cues for bacterial pathogens to navigate their way through their specific infectious cycle. On the other hand, NE stress hormones can provide the host with a unique tool to manipulate bacterial pathogens. These conflicting functions of NE hormones may have evolved from the presence of primitive bacterial adrenergic receptors with variable adrenergic sensing domains thus providing crucial versatility in the pathogen's survival tactics during host infection.

Conclusions and future perspectives

There is a strong body of evidence suggesting interkingdom communication by bacterial pathogens eavesdropping on the host adrenergic network. This ‘pathogen espionage’ is mediated through multiple bacterial receptors sensing host signalling molecules including NE stress hormones. These interactions are delicately balanced and it is clear that in most circumstances they can benefit the pathogen. However, there are some instances where the balance is tilted in favour of the host suggesting ‘counter-espionage’ by the host using these signals to its advantage.

Urgent future priorities include the chemical characterization of AI-3, the identification of the corresponding synthase and their conservation and function in other pathogenic bacteria which in turn will provide further insights into AI-3 cross-talk with adrenergic signalling in the host.

There is a need to identify the complete repertoire of bacterial adrenergic receptors and improve our understanding on why there are multiple sensors responding to one or more bacterial or host-derived signals, hence, defining the prokaryotic adrenergic signalome. Why do pathogens use more than one bacterial adrenergic receptor? Perhaps there is a specialized need for certain receptors in specific niches.

Furthermore, it is essential to demonstrate binding of NE stress hormones to the increasing number of identified putative bacterial adrenergic receptors. It would be immensely valuable to determine the crystal structures of bacterial adrenergic receptors with bound agonist and antagonist ligands in order to elucidate conserved structural motifs across the spectrum of receptors. Finally, the search for new interkingdom communication modules should be intensified to provide not only new biological insights into pathogenicity but also novel avenues for antimicrobial research.


The authors were supported by research grants from the Medical Research Council, UK. We wish to thank the anonymous reviewers for their valuable comments and suggestions. We sincerely apologize to all those colleagues whose important work we have not been able to cite due to space limitations. We would also like to thank Dr Simon Cockell at the Bioinformatics Support Unit (Newcastle University) for help with sequence analysis.