Type III secretion systems (T3SS) function by translocating effector proteins into eukaryotic host cells and are important for the virulence of many Gram-negative bacterial pathogens. Although the secretion and translocation machineries are highly conserved between different species, each pathogen translocates a unique set of effectors that subvert normal host cell physiology to promote pathogenesis. The uniqueness of each pathogen is further reflected in the diversity of mechanisms used to regulate T3SS gene expression. Pseudomonas aeruginosa utilizes a complex set of signalling pathways to modulate T3SS expression in response to extracellular and intracellular cues. Whereas some pathways are dedicated solely to regulating the T3SS, others co-ordinately regulate expression of the T3SS with multiple virulence functions on a global scale. Emerging regulatory themes include coupling of T3SS transcription with type III secretory activity, global regulatory control through modulation of cAMP biosynthesis, repression by a variety of stresses, involvement of multiple two component regulatory systems, and an inverse relationship between T3SS expression and multicellular behaviour. Factors controlling activation of T3SS expression likely contribute to the environmental survival of the organism and to the pathogenesis of acute P. aeruginosa infections. Conversely, active repression of the T3SS might contribute to the persistence of chronic infections.
Pseudomonas aeruginosa is a Gram-negative opportunistic bacterial pathogen capable of causing a variety of diseases in individuals with compromised immune function. Individuals susceptible to P. aeruginosa infection include those with burn wounds, immunodeficiency, cystic fibrosis, or undergoing extended critical care (Richards et al., 1999; 2000). The ability of P. aeruginosa to avoid phagocytic clearance is a major virulence determinant. Successful evasion of phagocytosis depends primarily on the presence of a functional type III secretion system (T3SS). T3SSs are specialized secretion systems that facilitate the delivery (translocation) of bacterial effector proteins into eukaryotic host cells (for recent reviews see Ghosh, 2004; Journet et al., 2005). The P. aeruginosa T3SS is used to translocate four effectors (ExoS, ExoT, ExoU and ExoY) with antihost properties (Frank, 1997). ExoS and ExoT inhibit phagocytosis by disrupting actin cytoskeletal rearrangement, focal adhesins and signal transduction cascades important for phagocytic function (Barbieri and Sun, 2004). ExoU and ExoY are cytotoxins with phospholipase and adenylate cyclase activities respectively (Yahr et al., 1998; Sato and Frank, 2004). In addition to the antiphagocytic role, effector-mediated damage to host tissues enhances dissemination, which can lead to systemic infection and septic shock (Kurahashi et al., 1999). The importance of the T3SS in the pathogenesis of acute P. aeruginosa infections has been demonstrated through cell culture and animal infection experiments, and through epidemiological studies (Apodaca et al., 1995; Sawa et al., 1998; Holder et al., 2001; Roy-Burman et al., 2001).
Expression of the T3SS genes is finely tuned to environmental conditions. The two most potent inducing signals for T3SS gene expression are contact of P. aeruginosa with host cells and extracellular calcium concentrations in the micromolar range (hereafter referred to as ‘low Ca2+’) (Frank, 1997; Vallis et al., 1999). Host cell contact and low Ca2+ also serve as inducing signals for the highly homologous T3SSs of Yersinia sp. (Hueck, 1998). Signals that repress expression of P. aeruginosa T3SS genes include metabolic stress, DNA damage, high concentrations of extracellular Cu2+, and low osmolarity. The response to these signals appears to be stochastic, because only a subpopulation of bacteria express the T3SS genes under inducing conditions (Hornef et al., 2000; Rietsch and Mekalanos, 2006), leading to a phenomenon called bistability (Dubnau, 2006). Conditional activation of the T3SS might benefit P. aeruginosa by conserving energetic resources or by restricting expression of T3SS antigen genes to avoid antibody-mediated clearance within the host. With nearly 10% of all P. aeruginosa genes dedicated to transcriptional regulation, this effort has not been lost on the T3SS; in the last 5 years nearly 25 genes have been implicated in regulation of the T3SS. Although substantial efforts have centred on identifying these regulators, it is becoming clear that a complete understanding of T3SS regulation will require a global view of P. aeruginosa as a finely tuned environmental sensor. This microreview focuses on the growing complexity of the regulatory networks controlling T3SS expression.
Expression of the T3SS requires ExsA
The P. aeruginosa T3SS consists of 43 co-ordinately regulated genes encoding the type III secretion and translocation machinery, regulatory functions, type III effectors and effector-specific chaperones (Frank, 1997). The genes associated with secretion, translocation and regulation are located together on the chromosome within five operons; genes encoding the type III effectors and associated chaperones are located elsewhere at separate sites on the chromosome (Fig. 1A), or in the case of exoU, on a pathogenicity island (Kulasekara et al., 2006). All T3SS genes are under the direct transcriptional control of ExsA, a member of the AraC family of transcriptional activators. ExsA-dependent promoters lie upstream of pscN, popN, pcrG, exsC, exsD, spcS, exoS, exoT, exoY and exoU (Fig. 1A) and each promoter contains an ExsA consensus binding site (TNAAAANA) centred ∼15 base pairs upstream of the −35 RNA polymerase binding sites (Yahr and Frank, 1994; Hovey and Frank, 1995; Yahr et al., 1995; Frank, 1997). The lack of palindromic or repeated sequences in the ExsA binding sites suggests that ExsA might bind DNA as a monomer. It should be noted that ExsA autoregulates its own expression by activating transcription of the exsCEBA operon (Fig. 1A).
Shen et al. recently identified PsrA as a second positive transcriptional activator of the exsCEBA operon (Shen et al., 2006). PsrA bound the exsCEBA promoter region in electrophoretic mobility shift assays (EMSA), but the observed shift required relatively high PsrA concentrations (μM). PsrA also activates transcription of the RpoS stationary phase sigma factor gene, which has a negative regulatory effect on expression of the T3SS (Hogardt et al., 2004). It is therefore possible that loss of PsrA influences expression of the T3SS through regulation of RpoS production, rather than through direct activation of the exsCEBA operon.
Regulation of ExsA activity
Transcriptional regulation by AraC-like proteins is a feature common to many T3SSs. Although prototypical AraC family members are regulated by small molecule ligands, none of the T3SS-associated AraC homologues are regulated in this manner. An emerging concept, however, is that a number of T3SS-associated AraC homologues are regulated by protein ligands (Plano, 2004). These protein ligands function as either coactivators or antiactivators. Although a coactivator for ExsA has not been identified, ExsA transcriptional activation is antagonized by two antiactivators (ExsD and PtrA). Both ExsD and PtrA directly bind to ExsA and inhibit ExsA-dependent transcription (McCaw et al., 2002; Dasgupta et al., 2004; Ha et al., 2004). Whether these interactions with ExsA interfere with DNA-binding activity or inhibit recruitment of RNA polymerase remains to be determined. ExsD functions as part of a signalling cascade that couples T3SS transcription to type III secretory activity (discussed further below) (McCaw et al., 2002). ptrA was first identified in an IVET screen as being induced during P. aeruginosa burn wound infections (Ha and Jin, 1999). Subsequent studies by Ha et al. found ptrA to be highly induced in the presence of Cu2+ leading the authors to propose that Cu2+ released from burned tissues might be the in vivo signal leading to repression of T3SS gene expression during burn wound infections (Ha et al., 2004). Paradoxically, T3SS mutants are significantly attenuated for virulence in the same infection model (Holder et al., 2001). This apparent contradiction suggests that the T3SS might be expressed only at specific stages or in specific tissues during the pathogenesis of burn wounds.
Transcription of the T3SS genes is coupled to type III secretory activity
Transcriptional induction of the T3SS by Ca2+ depletion involves two distinct physiological responses. The first response results in an increase in intracellular adenosine 3′,5′-cyclic monophosphate (cAMP) (Wolfgang et al., 2003). cAMP is a second-messenger signalling molecule that is required but not sufficient for transcription of T3SS genes (discussed further below). The second physiological response to low Ca2+ results in a dramatic increase in type III secretory activity through an unknown mechanism (McCaw et al., 2002). A recent study found that albumin and casein, both of which are low affinity Ca2+ binding proteins present in serum and Luria–Bertani broth, respectively, also contribute to activation of type III secretory activity (Kim et al., 2005). Serum albumin was previously shown to stimulate type III secretory activity in Yersinia enterocolitica (Lee et al., 2001).
Transcription of the P. aeruginosa T3SS is intimately coupled to type III secretory activity: transcription is repressed when the secretion machinery is inactive (high Ca2+) and derepressed when the secretion machinery is active (low Ca2+). The coupling of transcription to secretion is common among T3SSs and involves feedback mechanisms that block transcription in the absence of type III secretory activity (Miller, 2002; Plano, 2004). In P. aeruginosa, the coupling of transcription to secretion is mediated by three interacting proteins (ExsC, ExsE and ExsD) that regulate ExsA transcriptional activity (Fig. 1B). The ExsD antiactivator inhibits transcription of the T3SS at high Ca2+ levels or when the secretion machinery is inactivated by mutation in a gene coding for a structural component (Yahr et al., 1996; McCaw et al., 2002). ExsC functions as an anti-anti-activator by directly binding to and inhibiting the negative regulatory activity of ExsD (Dasgupta et al., 2004). Finally, ExsE binds to and antagonizes the regulatory activity of ExsC (Rietsch et al., 2005; Urbanowski et al., 2005). Based on these studies, and on the finding that ExsE is a secreted substrate of the T3SS, a model has been proposed for the coupling of transcription to secretory activity (Dasgupta et al., 2004). Under high Ca2+ conditions ExsE accumulates within the bacterium and sequesters the ExsC anti-anti-activator, leaving the ExsD antiactivator free to inhibit ExsA-dependent transcription (Fig. 1B). In contrast, ExsE is released extracellularly following activation of the type III secretion machinery by Ca2+ depletion. Decreased levels of intracellular ExsE allow ExsC to sequester ExsD thus liberating ExsA, which then activates transcription of the T3SS genes (Fig. 1B). Although the molecular details of the ExsE-ExsC, ExsC-ExsD and ExsD–ExsA binding interactions have not been determined, none of the associations are sensitive to excess Ca2+ or Ca2+ chelation (Dasgupta et al., 2004) (T.L. Yahr, unpublished). It is worth noting that coupled transcription and secretion is a common regulatory feature of T3SSs. The feedback mechanisms involved are diverse, however, and fall into one of three general categories: (i) secretion of a negative regulatory factor in Yersinia sp. (Pettersson et al., 1996), (ii) sequestration of a coactivator in Salmonella typhimurium and Shigella flexneri (Darwin and Miller, 2001; Mavris et al., 2002), and (iii) sequestration of an antiactivator (ExsD), as described above for P. aeruginosa. The presence of T3SS-associated ExsC, ExsD and ExsE homologues in Aeromonas hydrophila and Photorhabdus luminescens suggests that the regulatory mechanism utilized by P. aeruginosa is widely employed.
Contact-dependent expression of the T3SS genes
Like the low Ca2+ signal, contact of P. aeruginosa with host cells is also thought to activate the type III secretory machinery. Induction of T3SS-dependent cytotoxicity towards host cells is highly dependent upon the adherence properties of type IV pili (Kang et al., 1997). Interestingly, expression of heterologous adhesins restores cytotoxicity to a pilus mutant (Sundin et al., 2002). Whether adherence to host cells is sufficient for induction of T3SS gene expression is unclear as subsequent receptor-ligand interactions may also be required for induction.
The requirement for cell contact is well established, however, the mechanism of induction by cell contact is unclear. Chimeras of ExsE fused to the CyaA adenylate cyclase of Bordetella pertussis are known to be translocated into mammalian cells (M.L. Urbanowski and T.L. Yahr, unpublished). This finding suggests that host cell contact and low Ca2+ signals activate the ExsCEDA regulatory cascade by reducing intracellular ExsE levels through translocation and secretion respectively. Whether translocated ExsE contributes to pathogenesis outside of its regulatory role remains to be determined.
Although most studies on regulation of the T3SS have focused on induction by low Ca2+, it is unclear whether low Ca2+ is a physiologically relevant signal. A recent study examining the contribution of ExsC to host contact-dependent T3SS gene expression found induction of the T3SS genes to be highly dependent upon undefined host cell factors (Dasgupta et al., 2006). Mutants lacking exsC are defective for transcription of the T3SS genes under low Ca2+ conditions due to the unmitigated negative regulatory activity of ExsD (Dasgupta et al., 2004). As expected exsC mutants also fail to elicit T3SS-dependent cytotoxicity towards sheep erythrocytes, cultured Sf9 insect cells and the social amoeba Dictyostelium discoideum (Dasgupta et al., 2006). Surprisingly, an exsC mutant induced a wild-type level of T3SS-dependent cytotoxicity (albeit with a delayed response) towards mammalian epithelial and macrophage-like cell lines. Based on these data, Dasgupta et al. propose distinct ExsC-dependent and ExsC-independent regulatory pathways (Dasgupta et al., 2006). The ExsC-dependent pathway is important for the rapid response to all host cell types tested (generic cell contact) and might contribute to the survival of P. aeruginosa outside of the context of mammalian infections (Fig. 1C). In contrast, the ExsC-independent pathway is activated only in response to specific mammalian cell types and therefore might represent a specific adaptation to maximize expression of the T3SS genes during infections in mammalian hosts.
Transcriptional control of the T3SS through a cAMP-signalling cascade
Modulation of intracellular cAMP levels in response to environmental signals is another important mechanism controlling T3SS expression (Fig. 2). Recent studies indicate that conditions essential for the expression of the T3SS (low Ca2+ and high salt) also increase intracellular cAMP levels (Wolfgang et al., 2003; Rietsch and Mekalanos, 2006). cAMP influences P. aeruginosa gene expression by acting as an allosteric regulator of Vfr, a functional homologue of the Escherichia coli cAMP receptor protein (CRP). Vfr was initially described as a regulator of quorum sensing, exotoxin A production and type IV pilus-mediated twitching motility (West et al., 1994; Albus et al., 1997; Beatson et al., 2002). Subsequent studies using whole genome microarray analyses revealed that mutants lacking cAMP or vfr exhibited reduced expression of nearly 200 genes, including those involved in the T3SS, type IV pilus biogenesis and type II secretion (Wolfgang et al., 2003). This finding was the first indication that the T3SS is integrated into a global regulatory network that specifically controls genes implicated in pathogenesis. Regulation by cAMP has also been reported for the T3SS of Y. enterocolitica (Petersen and Young, 2002).
Genetic evidence indicates that cAMP, Vfr and ExsA act on a common pathway to regulate T3SS gene expression, because mutants lacking either vfr or cAMP biosynthesis can be complemented by ExsA overproduction (Wolfgang et al., 2003). cAMP and Vfr appear to act upstream of or at the same step as ExsA, because an exsA mutant can not be complemented by overproduction of either cAMP or Vfr. Shen et al. recently showed that purified Vfr combined with cAMP failed to recognize the exsCEBA promoter in EMSA, suggesting that the cAMP/Vfr complex does not directly regulate exsA expression (Shen et al., 2006). Despite evidence that cAMP/Vfr can recognize E. coli CRP-like DNA binding sites, it is not clear how cAMP and Vfr exert transcriptional control over the T3SS genes. To date, only two Vfr DNA binding sites have been identified in P. aeruginosa. These sites closely resemble E. coli CRP binding sites and map to the promoters of lasR, the global regulator of the las quorum sensing system, and fleQ, the master regulator of flagellar biogenesis (Albus et al., 1997; Dasgupta et al., 2002). It is important to note that neither LasR nor FleQ is known to influence expression of the T3SS genes (Dasgupta et al., 2003; Hogardt et al., 2004; Bleves et al., 2005). The apparent lack of additional CRP-like binding sites in the P. aeruginosa PAO1 genome indicates that the cAMP/Vfr transcriptional complex might recognize a second non-CRP-like binding site and/or that Vfr might function as part of a protein complex in which DNA binding is Vfr-independent. The existence of alternative recognition sequences is a distinct possibility given subtle structural differences in the DNA binding helix of Vfr relative to E. coli CRP and the fact that production of CRP in P. aeruginosa vfr mutants does not restore expression of Vfr-dependent genes (West et al., 1994). Because CRP family members can act synergistically with AraC-like transcriptional activators, we favour a model in which the cAMP/Vfr complex directly enhances ExsA binding or activity as part of a transcriptional complex.
Cyclic AMP control mechanisms
Three general mechanisms control intracellular cAMP levels in P. aeruginosa: synthesis, degradation and excretion (Siegel et al., 1977). P. aeruginosa produces two adenylate cyclases (CyaA and CyaB) responsible for cAMP synthesis. Mutants lacking both cyaA and cyaB are devoid of measurable intracellular cAMP, but the contribution of CyaB is substantially greater than that of CyaA (Wolfgang et al., 2003). Mutants lacking cyaB are defective for T3SS gene expression in vitro and T3SS activity in vivo (Wolfgang et al., 2003; Smith et al., 2004). The predicted domain structure of CyaB consists of six transmembrane helices that form a series of loops in the inner membrane, thereby serving as an anchor for the catalytic domain and as a potential sensor for extracellular signals. The catalytic domain of CyaB shares features common to a subset of adenylate cyclases that are regulated by inorganic carbon (CO2/HCO3–) (Cann et al., 2003; Linder and Schultz, 2003). This raises the possibility that CyaB senses HCO3–, allowing the bacteria to respond to fluxes in inorganic carbon by modulating cAMP levels. This is particularly relevant given the recent findings that mutations altering metabolism strongly influence expression of the T3SS (Dacheux et al., 2002; Rietsch et al., 2004; Rietsch and Mekalanos, 2006).
A transposon mutagenesis screen for insertion mutants defective in CyaB-dependent cAMP synthesis identified a set of strains carrying insertions in the type IV pilus chemosensory signal transduction system (pilGHIJKchpA) (N. Fulcher and M.C. Wolfgang, unpublished). The pilus chemotaxis system (chp) is highly homologous to the chemotactic signalling system (che) that controls flagellar rotation and swimming motility (Mattick, 2002). Previous studies have shown that the chp system controls the surface localization of type IV pili and pilus-mediated twitching motility by an unknown mechanism (Darzins, 1994; 1995; Whitchurch et al., 2004). Transposon insertion mutants of pilG, pilI, pilJ and chpA exhibit not only reduced pilus expression and twitching motility but also drastically reduced cAMP levels and T3SS gene expression without significantly altering cyaB transcription (N. Fulcher and M.C. Wolfgang, unpublished). In all cases T3SS gene expression was rescued by the addition of exogenous cAMP. Although the mechanism by which the chp system influences cAMP levels is unclear, the most likely scenario is that it directly or indirectly alters CyaB activity in response to an unknown extracellular signal. These results partially explain the published phenotypes for several chp system mutants that could not be accounted for by loss of twitching motility alone (D'Argenio et al., 2001; Whitchurch et al., 2004; Zolfaghar et al., 2005).
Degradation of cAMP is catalysed by the enzyme cyclic-3′, 5′-AMP phosphodiesterase (CpdA). The P. aeruginosa genome has a single cpdA homologue (PA4969); however, its role in controlling cAMP-dependent T3SS gene expression remains to be determined. In addition, the mechanism by which cAMP is excreted by bacteria is unknown, as is whether excretion plays a role in regulation of T3SS genes in P. aeruginosa.
Vfr control mechanisms
Mutants unable to synthesize cAMP show a twofold reduction in vfr transcription (Wolfgang et al., 2003) and a disproportionate reduction in Vfr protein levels (greater than 10-fold; M.C. Wolfgang, unpublished). Given the fact that CRP family members are intrinsically unstable (Ohki et al., 1992), we propose that cAMP might protect Vfr from degradation by facilitating proper folding. Alternatively, Vfr levels might be controlled by a cAMP-dependent post-transcriptional mechanism. The vfr gene is transcribed with an unusually long (146 base pair) 5′ untranslated leader sequence containing an imperfect 52 base pair inverted repeat (Runyen-Janecky et al., 1997). Similar leader sequences have been implicated in post-transcriptional regulation and mRNA stability.
Whitchurch and colleagues recently showed that deletion of fimL, encoding an unusual signalling protein, resulted in reduced Vfr expression (Whitchurch et al., 2005). While the overall effect on Vfr levels was subtle (30–50% reduction in transcription relative to wild type), expression of the T3SS genes and other cAMP/Vfr-dependent phenotypes including piliation and twitching motility were reduced in the fimL mutant. The defect in T3SS-mediated host cell cytotoxicity displayed by the fiml mutant could be overcome by providing Vfr in trans. Based on these results, Whitchurch et al. concluded that FimL directly or indirectly regulates vfr expression. Interestingly, FimL shares a high degree of sequence similarity with the amino-terminus of ChpA (Whitchurch et al., 2005). Given the role of ChpA in regulating cAMP production, Fiml might have a similar function in regulating cAMP synthesis through the Chp pilus chemotaxis system. Despite this intriguing connection, the fimL mutant showed no measurable reduction in intracellular cAMP under the conditions examined (Whitchurch et al., 2005). Extragenic suppressors of the fimL mutation, however, resulted in the hyper-production of cAMP and restoration of Vfr and T3SS function, which further implicates cAMP as a regulator of the T3SS.
Stress as a regulator of the T3SS
In addition to the low Ca2+ and host contact signals, expression of the T3SS genes is also influenced by osmolarity, DNA damage and metabolic stress (Fig. 2). Transcription of the T3SS genes is activated by elevated extracellular salt concentrations with maximal expression occurring at 200 mM NaCl (Hornef et al., 2000; Rietsch and Mekalanos, 2006). Microarray analyses of the P. aeruginosa osmotic stress response found genes encoding the T3SS to be significantly induced under hyper-osmotic stress conditions (Aspedon et al., 2006). The salt effect has been attributed to osmolarity, as similar results can be achieved using other solutes such as sucrose (Aspedon et al., 2006; Rietsch and Mekalanos, 2006). Interestingly, intracellular cAMP levels are elevated following growth in 200 mM NaCl when compared with growth in 5 mM NaCl, suggesting a possible link between osmotic stress, the cAMP signalling cascade and regulation of the T3SS (Rietsch and Mekalanos, 2006).
The stress associated with DNA damage is a regulatory signal in many organisms. In P. aeruginosa DNA damage activates the SOS response signalling cascade, which results in the induction of DNA repair systems and the production of bacteriocins termed pyocins (Michel-Briand and Baysse, 2002). Wu and Jin recently identified PtrB as an additional SOS-induced gene product (Wu and Jin, 2005). PtrB acts as a repressor of T3SS gene expression through an unknown mechanism. Repression of the T3SS genes in response to DNA damage might allow cells to redirect energetic resources to the SOS stress response.
Finally, metabolic signals/stresses have a profound influence on expression of the T3SS genes, possibly resulting from the overproduction or depletion of specific metabolites. Mutants lacking pyruvate dehydrogenase (aceA or aceB) or a glucose transport regulator (gltR) fail to express the T3SS genes under low Ca2+ conditions (Dacheux et al., 2002; Wolfgang et al., 2003). Similarly, overproduction of MDR efflux pumps or overexpression of genes involved in histidine transport and metabolism prevents expression of the T3SS genes (Rietsch et al., 2004; Linares et al., 2005). Each of these mutations is thought to result in metabolic imbalance, possibly through excretion or increased demand for specific metabolites. A recent study suggests that one metabolite controlling expression of the T3SS is derived from acetyl-CoA (Rietsch and Mekalanos, 2006). The molecular mechanisms linking metabolic stresses to expression of the T3SS genes will be an area of future research.
Reciprocal regulation of the T3SS and chronic virulence factors
P. aeruginosa infections can be classified as either acute or chronic. Virulence determinants activated by the cAMP/Vfr system are typically associated with acute P. aeruginosa infections. In contrast, recent studies have linked negative regulation of the T3SS with factors thought to contribute to chronic infections of the cystic fibrosis (CF) airways. These factors include the exopolysaccharide alginate, the pel and psl loci involved in polysaccharide biosynthesis and biofilm formation, and quorum sensing. Alginate production is regulated by the alternative sigma factor AlgU, which is normally antagonized by the MucA antisigma factor. Many P. aeruginosa isolates from the airways of older CF patients, however, carry mucA mutations and overproduce alginate, resulting in a mucoid phenotype (Martin et al., 1993). Emergence of the mucoid phenotype is a clinical indicator for the onset of chronic infection and is thought to contribute to persistence of P. aeruginosa in the CF lung. Interestingly, most CF isolates fail to express the T3SS genes when cultured ex vivo under low Ca2+ growth conditions (Dacheux et al., 2000). A recent study by Wu et al. found that expression of the T3SS genes was repressed in a mucA mutant, suggesting that emergence of the mucoid phenotype and loss of T3SS expression coincides with the acquisition of mucA mutations (Wu et al., 2004). The T3SS defect in the mucA mutant was in part dependent on the MucA-repressed regulators AlgU (AlgT) and AlgR. The absence of either of these downstream regulators (which prevents alginate production) restored low Ca2+-induced expression of the T3SS genes (Wu et al., 2004). The mechanism linking expression of the T3SS genes and alginate production, however, remains unclear. It is possible that production of large amounts of alginate imposes a metabolic stress on the bacterium (see above). Interestingly, microarray analyses of mucA mutants show altered expression of numerous genes involved in metabolism and osmotic stress (Firoved and Deretic, 2003; Wu et al., 2004). While mucA mutations represent one mechanism for loss of T3SS gene expression, Smith and colleagues recently showed that P. aeruginosa strains accumulate mutations in other regulatory genes (exsA, vfr, cyaB) during chronic CF infection that could also account for the lack of T3SS gene expression (Smith et al., 2006). Consistent with this finding, Dacheux et al. found that expression of exsA in trans restored T3SS gene expression to most but not all CF isolates (Dacheux et al., 2001).
A second global regulatory network affecting the T3SS is controlled by the RetS and LadS sensor proteins (Fig. 2). RetS is required for expression of the T3SS genes and for repression of the pel and psl loci (Goodman et al., 2004; Laskowski et al., 2004; Zolfaghar et al., 2005; Ventre et al., 2006). Mutants lacking RetS have reduced T3SS-dependent cytotoxicity to host cells and an increased capacity to form biofilms. In contrast, LadS has the opposite effect on T3SS gene expression and biofilm formation (Ventre et al., 2006). RetS and LadS are members of the two-component regulatory family with unusual domain organization. Both proteins contain amino-terminal 7-transmembrane and 7-transmembrane-associated periplasmic sensing domains, internal domains encoding a sensor histidine kinase and an ATPase, and a carboxy-terminal region encoding either a single (LadS) or tandem (RetS) response regulator-like receiver domains. The role of the tandem receiver domains in RetS was recently examined by site-directed mutagenesis. Whereas the phosphoacceptor site in the first domain appears to inhibit RetS activity, the phosphoacceptor site in the second domain is essential for RetS activity (Laskowski and Kazmierczak, 2006). Ventre et al. suggested that RetS and LadS might be involved in sensing host cell-associated carbohydrates (Ventre et al., 2006). Neither retS nor ladS are genetically linked to a typical response regulator and it is thought that RetS and LadS regulatory activities are mediated through the GacS/GacA/rsmZ pathway. GacAS is a two-component regulatory system that activates transcription of rsmZ, a small untranslated regulatory RNA (Heurlier et al., 2004). rsmZ controls gene expression by sequestering RsmA, an RNA-binding protein involved in post-transcriptional control of gene expression. The current model indicates that RetS inhibits rsmZ expression, resulting in elevated levels of free RsmA, whereas LadS and GacS have the opposite effect on RsmA (Ventre et al., 2006). Consistent with this hypothesis, the purified cytoplasmic domain of RetS blocks the phosphorylation of purified GacA in vitro (A.L. Goodman and S. Lory, pers. comm.).
Although it is unclear how RsmA regulates expression of the T3SS genes, many of the genes regulated by the LadS and RetS systems are also under the transcriptional control of the cAMP/Vfr complex (Goodman et al., 2004). This observation, combined with the findings that Vfr overexpression complements a retS mutant, and that vfr and cyaB transcript levels are significantly reduced in an rsmA mutant, suggests that RetS/LadS/GacSA regulates expression of the T3SS genes through the cAMP/Vfr pathway (Mulcahy et al., 2006). Important questions for future research include the biochemical nature of the signals recognized by the RetS, LadS and GacS sensors, and the target of regulation by RsmA.
In contrast to the findings described above, P. aeruginosa small colony variants (SCVs), isolated from chronically infected CF patients have both an increased capacity to form biofilms and increased T3SS gene expression and cytotoxicity (Drenkard and Ausubel, 2002; von Gotz et al., 2004). These findings suggest that reciprocal regulation of the T3SS and chronic virulence factors is not absolute.
Role of quorum sensing
Mutants lacking the Rhl quorum sensing system show increased expression of T3SS genes and secretion of ExoS at an earlier stage during exponential growth (Hogardt et al., 2004; Bleves et al., 2005). Similarly results were found for mutants lacking the rpoS stationary-phase sigma factor (Hogardt et al., 2004). As RpoS regulates the Rhl system (Whiteley et al., 2000), the mechanism of regulation by RpoS is likely mediated through quorum sensing.
The past 5 years have seen a tremendous increase in our understanding of the pathways controlling expression of the T3SS. Despite these advances, many more questions have been generated than have been answered. One key to answering these questions will lie in understanding the basis of global regulation by Vfr and RsmA. The most difficult tasks for the future, and possibly the most rewarding, will be identification of the host and environmental signals that induce or repress T3SS gene expression, and integrating each of the signalling pathways into a coherent regulatory model. Armed with this knowledge it should be possible to design therapeutic strategies that interfere with expression of the T3SS genes.
Support for this work was provided by the Howard Hughes Medical Institute Biomedical Research Support Faculty Start-up Program (T.L.Y.), the University of Iowa W.M. Keck Microbial Communities and Cell Signalling Program (T.L.Y.), and the National Institutes of Health (RO1-AI055042 to T.L.Y.). M.C.W was supported by Faculty Start-up funds from the UNC School of Medicine, Department of Microbiology and Immunology and Cystic Fibrosis/Pulmonary Research and Treatment Center. We thank Nanette Fulcher in the Wolfgang Laboratory and members of the Yahr Laboratory for critical reading of the manuscript.