OUP user menu

Regulation of virulence and antibiotic resistance by two-component regulatory systems in Pseudomonas aeruginosa

W. James Gooderham, Robert E. W. Hancock
DOI: http://dx.doi.org/10.1111/j.1574-6976.2008.00135.x 279-294 First published online: 1 March 2009

Abstract

The Gram-negative opportunistic pathogen Pseudomonas aeruginosa ubiquitously inhabits soil and water habitats and also causes serious, often antibiotic resistant, infections in immunocompromised patients (e.g. cystic fibrosis). This versatility is mediated in part by a large repertoire of two-component regulatory systems that appear instrumental in the regulation of both virulence processes and resistance to antimicrobials. Major two-component regulatory system proteins demonstrated to regulate these diverse processes include PhoP–PhoQ, GacA–GacS, RetS, LadS, and AlgR, among others. Here, we summarize the current body of knowledge of these and other two-component systems that provides insight into the complex regulation of virulence and resistance in P. aeruginosa.

Keywords
  • Pseudomonas aeruginosa
  • virulence
  • antimicrobial resistance
  • two-component regulatory system

Introduction

The bacterial world is dynamic. Bacterial survival under changing environmental conditions requires the capability to sense and quickly respond to many different stimuli. Often, these physiological responses are mediated at the level of gene expression and accomplished by regulatory proteins. One class of regulators, the bacterial two-component regulatory systems, are designed to sense diverse stimuli and enact an appropriate and rapid adaptive physiological response that can involve the altered transcription of a substantial number of genes.

Two-component regulatory systems classically comprise an inner membrane-spanning sensor histidine kinase and a cytoplasmic response regulator (Stock et al., 2000). The detailed mechanisms of two-component signal transduction systems have been reviewed previously (Stock et al., 2000; Mascher et al., 2006; Laub & Goulian, 2007). Functionally coupled sensor kinase and response regulator genes are often encoded adjacent to one another in the genome, forming an operon, and this arrangement is generally indicative of a cognate system which enacts a response to a specific signal detected by the sensor kinase, although in many cases the precise signal is unknown. The functional mode of both sensor kinases and response regulators is determined by reversible phosphotransfer reactions and ensuing protein conformational changes (Stock et al., 2000). In the archetypical system, the sensor kinase contains an N-terminal periplasmic input domain that detects a specific stimulus and a C-terminal cytoplasmic transmitter domain that binds ATP and has histidine kinase activity. The cognate response regulator contains a conserved aspartate receiver domain and a variable output domain that often binds DNA.

Classically, a membrane-bound dimeric sensor kinase detects an environmental stimulus/ligand in the periplasm via its input domain, and then undergoes trans-autophosphorylation at a conserved histidine residue in its transmitter domain (Mascher et al., 2006). This phosphoryl group is then transferred to and catalyzed by the conserved aspartate-containing receiver domain of the response regulator. Phosphorylation of the response regulator receiver domain often modifies the activity of the output domain, of which there are many types (Galperin et al., 2006; http://www.ncbi.nlm.nih.gov/Complete_Genomes/RRcensus). The output domain is most frequently a helix–turn–helix DNA-binding domain (e.g. OmpR, NarL, and NtrC subfamilies), but other types include those with enzymatic functions (e.g. GGDEF diguanylate cyclases, EAL c-di-GMP phosphodiesterases, and CheB-like methylesterases). In the former case, phosphorylation of the receiver domain changes the response regulator's affinity for specific DNA elements so as to modify gene expression and initiate the corresponding cellular response. Dephosphorylation of the response regulator by the sensor kinase serves to return the system to its preactivation state.

Structural and functional modifications of this basic functionality do exist. Sensor kinases that contain an aspartate-containing receiver domain at their C-terminus, analogous to that of the response regulator, are termed ‘hybrid’ sensor kinases. To accomplish histidine to aspartate phosphotransfer, it is therefore thought that these hybrid systems utilize small separate proteins that contain histidine phosphotransfer (Hpt) domains to facilitate phosphotransfer from the sensor's histidine to the terminal response regulator's aspartate (His→Asp→His→Asp; Rodrigue et al., 2000). Conversely, other nonclassical sensor histidine kinases are termed ‘unorthodox’ and contain an additional Hpt domain fused N-terminal to their receiver domain and are thought to participate in complex phosphorelay networks. Both types of systems may enable added flexibility relative to signal transduction mediated by the classical systems. In this review, we generally refer to each two-component system with the response regulator preceding its cognate sensor kinase.

Pseudomonas aeruginosa

Pseudomonas aeruginosa inhabits diverse environments that include soil and water habitats, and can also infect plants, insects, nematodes, and animals (Stover et al., 2000). Like other Pseudomonads, it exhibits nutritional and metabolic versatility. Together, this remarkable adaptability is driven by sophisticated and coordinated regulation of gene expression. Pseudomonas aeruginosa is also a major opportunistic human pathogen of immunocompromised patients. It is the third most common nosocomial (hospital) pathogen in North America, and a leading cause of chronic pulmonary infections and mortality in cystic fibrosis (CF) patients, ventilator-associated pneumonia, and sepsis in burn patients (Fridkin et al., 1999). A major reason for therapeutic failure in the treatment of these infections is its high intrinsic resistance to multiple classes of antibiotics due to low outer-membrane permeability and active antibiotic efflux systems (Hancock & Speert, 2000). The steady rise in adaptive and mutational resistance is increasingly impacting on therapeutic success and new antimicrobial therapeutic options are needed for resistant strains, some of which have developed resistance to virtually every type of antibiotic and have thus become hospital ‘superbugs’ (Falagas & Bliziotis, 2007).

Multideterminant virulence and high antibiotic resistance are two hallmarks of P. aeruginosa infections. A large number of two-component regulatory systems encoded in its genome have equipped P. aeruginosa with a sophisticated capability to regulate diverse virulence and resistance processes. In this review, we focus on the rapidly expanding recent body of knowledge regarding the roles of P. aeruginosa two-component regulatory systems participating in virulence and antibiotic resistance processes.

Pseudomonas aeruginosa virulence and antimicrobial resistance

An arsenal of virulence factors have been reported and appear to impact on the pathogenicity of this organism. Included among these are cell-associated factors such as lipopolysaccharide, flagella and type IV pili, the exopolysaccharide alginate, and secreted factors including toxins, elastase and other proteases, phospholipases and small molecules such as phenazines, rhamnolipids and cyanide. Quorum-sensing, type III secretion and many other regulatory systems are important for virulence and, as they have been reviewed in detail elsewhere, will not be a specific focus of this review (Kirisits & Parsek, 2006; Venturi et al., 2006; Yahr & Wolfgang, 2006).

The single polar flagellum of P. aeruginosa is an important and highly regulated virulence factor involved in both motility and surface attachment. Isolates from long-term chronic CF infections often lack flagella. One hypothesis addressing this is that these nonmotile cells are at some stage of a biofilm-like adaptation. Biofilm formation follows three distinct stages: surface attachment, microcolony formation, and differentiation into a mature biofilm community encased in an extracellular matrix (Kirisits & Parsek, 2006). Type IV pili-dependent twitching and flagellar-based swimming motility are important for colonization and initial surface attachment (O'Toole & Kolter, 1998), and many genes associated with these motility appendages are necessary for proper biofilm formation (Kirisits & Parsek, 2006). In addition, we demonstrated recently that there is a strong, but not obligate, correspondence in gene requirements between biofilm formation and swarming motility, which is dependent on both motility appendages and represents a complex adaptation to viscous environments (Overhage et al., 2007, 2008). Type III secretion system-mediated cytotoxicity is also considered to be a key virulence property of P. aeruginosa, while type II secretion mediates the secretion of a broad range of toxins and other virulence-related enzymes; interestingly, the structures involved in these secretion systems resemble flagella and pili, respectively.

Pseudomonas aeruginosa is capable of causing both chronic and acute infections and appropriately regulates the virulence determinants required for each of these types of infection (Furukawa et al., 2006). Chronic persistent infections can be microcolony or biofilm-like in nature, such as those present in CF lung infections (Singh et al., 2000), and on urinary catheters and other indwelling medical devices. Conversely, acute P. aeruginosa infections are typified by bacterial penetration of the host epithelium and systemic spread, such as in severe burn wound infections.

Antibiotic resistance is similarly complex. In addition to classical mutational or acquired resistance, P. aeruginosa antibiotic resistance can be triggered by environmental factors (a subset of adaptive resistance). Subinhibitory concentrations of antibiotics and other molecules can themselves induce resistance to subsequent exposure to otherwise lethal concentrations of antibiotics. During human infections, P. aeruginosa is likely exposed to endogenous β-defensin and cathelicidin host defense (antimicrobial) peptides at epithelial surfaces. Further, polymyxins (peptide antibiotics) and other types of conventional antibiotics also interact with P. aeruginosa when therapeutically administered. Of note, natural and synthetic antibacterial peptides are being pursed as novel therapeutics (Marr et al., 2006). However, adaptive resistance to cationic antimicrobial peptides and polymyxin B is known to occur in response to limiting extracellular concentrations of divalent Mg2+ and Ca2+ cations, but this is unlikely to be clinically meaningful as the body contains 1–2 mM divalent cations. However, as noted below peptides themselves are able to induce peptide resistance mechanisms through two-component regulation.

Overview of two-component systems in P. aeruginosa

The P. aeruginosa PAO1 genome encodes one of the largest complements of regulatory proteins found in bacteria representing just under 10% of all genes (Stover et al., 2000), and a major subset are the genes encoding 64 sensor kinases and 72 response regulators (Tables 1 and 2; Rodrigue et al., 2000; Chen et al., 2004; Galperin, 2006). These include four CheA-type sensor kinases and four CheB-type methylesterase response regulators that are part of chemosensory gene clusters. The most common type of gene organization (29 systems) comprises a response regulator upstream of its cognate sensor kinase. Almost all operons with this organization comprise an OmpR-type response regulator, and these systems likely represent coevolution through duplication from an ancestral OmpR-sensor kinase pairing (Chen et al., 2004). Twenty-one systems contain sensor kinases upstream of their cognate response regulator-encoding genes (separated by three or less genes). NarL-type response regulators can be associated with both classical and hybrid or unorthodox sensors and therefore, likely represent evolutionary recruitment during cognate pair formation (Chen et al., 2004). A further 13 sensors and 15 regulators are termed ‘orphans’ and are not physically linked to any other two-component gene, thus making the identification of cognate function pairs difficult (except GacA and GacS that are very well characterized). Additionally, there are three separate genes encoding HPt domains [hptA (PA0991), hptB (PA3345), and hptC (PA0033)] that possibly interact with hybrid kinases to form complex phosphorelay signaling networks (Rodrigue et al., 2000). HptB was recently shown to be capable of interacting with several hybrid kinases in a multistep phosphorelay pathway (Hsu et al., 2008). Complex CheA-type chemosensory signal transduction proteins such as ChpA (PA0413), which is involved in twitching and swarming motility, virulence and cytotoxicity (D'Argenio et al., 2001; Whitchurch et al., 2004; Leech & Mattick, 2006), will not be discussed in this review.

View this table:
Table 1

Pseudomonas aeruginosa virulence- and antibiotic resistance-associated classical-type response regulators and sensor histidine kinases

PA numberGene productBrief functional descriptionReferences
PA 0034Repressed during CF sputum medium in vitro growth. Directly upstream to hptC (PA0033)Palmer et al. (2005)
PA 0408PilGChemosensory pili (Pil–Chp) system, twitching motilityDarzins & Russell (1997)
PA 0409PilH
PA 0463/4CreB–CreCCatabolism. Swarming and swimming motilityWagner et al. (2007)
PA 0929/30PirR–PirSIron acquisitionVasil & Ochsner (1999)
PA 1099/8FleR–FleSFlagellar motility, adhesion to mucin. FleS likely cytoplasmic sensorRitchings et al. (1995); Dasgupta et al. (2003)
PA 1157/8PA1157 essential in rat chronic lung infectionPotvin et al. (2003)
PA 1179/80PhoP–PhoQLow Mg2+ signal. Polymyxin, antimicrobial peptide and aminoglycoside resistance. Virulence, swarming motility and biofilm formationErnst et al. (1999); Macfarlane et al. (1999, 2000); Ramsey & Whiteley (2004); McPhee et al. (2006)
PA 1135/6PA1136 encodes immunogenic peptides to cystic fibrosis serumBeckmann et al. (2005)
PA 2523/4CzcR–CzcSMetal and imipenem resistanceHassan et al. (1999); Perron et al. (2004)
PA 2586GacAGacA–GacS system. Multihost virulence, quorum-sensing-dependent regulation of exoproducts and virulence factors, biofilm formation and antibiotic resistance, swarming motility, type III secretionReimmann et al. (1997); Rahme et al. (2000); Parkins et al. (2001); Soscia et al. (2007)
PA 2686/7PfeR–PfeSIron acquisitionDean et al. (1996)
PA 2809/10CopR–CopSMetal and imipenem resistanceTeitzel et al. (2006); Caille et al. (2007)
PA 3045RocA2RocA2–RocS2. Fimbriae adhesin gene regulationKulasekara et al. (2005)
PA 3192/1GltRGlucose transport, type III secretion cytotoxicityWolfgang et al. (2003)
PA 3204/6PA3206 encodes highly immunogenic peptides to cystic fibrosis serumBeckmann et al. (2005)
PA 3346HptB-mediated phosphorelay, swarming motility and biofilm formationHsu et al. (2008)
PA 3702/4WspR–WspEWsp chemosensory system. Biofilm and cyclic-di-GMP level regulation, autoaggregation. WspR contains GGDEF output domain, WspE is CheA-type sensorD'Argenio et al. (2002); Hickman et al. (2005); Kulasekara et al. (2005)
PA 3879/8NarL–NarXNitrate sensing and respiration. Biofilm formation, swimming and swarming motilityVan Alst et al. (2007)
PA 3947RocR (SadR)RocS1/R/A1 (SadA/R/S) system. Cytotoxicity, virulence, regulation of fimbriae adhesins, type III secretion and biofilm formation. RocA1 contains EAL output domain, RocR is RocA1 antagonistKulasekara et al. (2005); Kuchma et al. (2005)
PA 3948RocA1 (SadA)Kulasekara et al. (2005); Ruer et al. (2007).
PA 4296/3PprB–PprAOuter-membrane permeability and aminoglycoside resistance. PprA sensor likely cytoplasmicWang et al. (2003); Wagner et al. (2007)
PA 4396Overexpression impairs type III secretion-mediated cytotoxicity. GGDEF output domainKulasakara et al. (2006)
PA 4547/6PilR–PilSType IV fimbriae expression, twitching and swarming motility, biofilm formationIshimoto & Lory (1992); Hobbs et al. (1993); Goodman et al. (2004); Overhage et al. (2007)
PA 4726/5CbrB–CbrACarbon and nitrogen storage, cytotoxicity, swarming motility, nematode virulenceGallagher & Manoil (2001); Rietsch et al. (2004); Wagner et al. (2007)
PA 4776/7PmrA–PmrBInduced by low Mg2+ and cationic antimicrobial peptides. Polymyxin B and antimicrobial peptide resistanceMcPhee et al. (2003); Moskowitz et al. (2004).
PA 4959FimXPhosphodiesterase (GGDEF and EAL domains), twitching motility, reduced in vitro cytotoxicityHuang et al. (2003); Kulasakara et al. (2006); Kazmierczak et al. (2006)
PA 5261/2AlgR–FimS (AlgZ)Virulence, alginate biosynthesis, twitching and swarming motility, biofilm formation, cyanide production, cytotoxicity and type III secretion system gene expressionDeretic & Konyecsni (1989); Whitchurch et al. (1996); Yu et al. (1997); Lizewski et al. (2002); Wu et al. (2004); Morici et al. (2007); Overhage et al. (2007); Belete et al. (2008).
PA 5360/1PhoB–PhoRPhosphate level regulation, low phosphate signal, quorum sensingFilloux et al. (1988); Jensen et al. (2006)
PA 5483/4AlgB–KinBAlginate biosynthesisWozniak & Ohman (1991); Goldberg & Dahnke (1992); Leech et al. (2008)
N/APvrRPhenotypic variation, antibiotic resistance, biofilm formationDrenkard & Ausubel (2002)
  • * Annotated gene identification number (PA) and protein names according to the PAO1 genome at http://www.pseudomonas.com. Response regulator-encoding genes are listed before their cognate sensor kinase-encoding genes; thus, this does not necessarily represent transcriptional direction.

  • Other classical-type RR and SK include PA0179/8, PA0267, PA0600/1, PA0756/7, PA1335/6 (aauR/aauS), PA1397/6, PA1437/8, PA1456/8 (cheY/cheA), PA1785 (nasT), PA1799/8, PA1978 (agmR)-PA1979/80, PA1637/6 (kdpE/kdpE), PA1785, PA1980/79, PA2376, PA2479/80, PA2572/1, PA 2657/6, PA2798, PA2881/2, PA2899, PA3077/8, PA3604, PA3714, PA4032/6, PA4080, PA4101/2, PA4117 (bphP), PA4296/7, PA4381/0 (colR/colS), PA4493/4, PA4781, PA4843, PA4885/6 (irlR), PA4983/2 (dmsR), PA4983/2, PA5124/5 (ntrC/ntrB), PA5166/5, PA 5200/9 (ompRenvZ), PA5364, PA 5511/2. Additional chemosensory-type systems: 4 CheB-like methylesterases: PA0173, PA0414 (chpB), PA1459, PA3703 (wspF). 4 CheA-like histidine kinases: PA0178, PA0413 (chpA), PA1458 (cheA), PA3704 (wspE).

  • Selected references generally do not contain original gene identification reference if this was unrelated to virulence or resistance.

View this table:
Table 2

Pseudomonas aeruginosa virulence-associated hybrid and unorthodox sensor histidine kinases

PA numberSensor kinaseBrief functional descriptionReference
PA 0928GacSGacA–GacS system. Plant, insect, animal virulence. Motility, biofilm, secondary metabolite regulation; aminoglycoside resistance. HPt domainRahme et al. (2000); Brinkman et al. (2001); Davies et al. (2007)
PA 1611PA3346–HptB-PA1611 phosphorelayLin et al. (2005); Hsu et al. (2008)
PA 3044RocS2RocA2–RocS2 system. Regulation of fimbriae adhesins. HPt domainKulasekara et al. (2005)
PA 3946SadS (RocS1)SadA–SadR–SadS system (RocS1–RocR–RocA1). Biofilm maturation, fimbrial genes, type III secretion; C. elegans virulence; HPt domainKulasekara et al. (2005); Gallagher & Manoil (2001); Kuchma et al. (2005)
PA 3974LadSVirulence, biofilm formation, type III secretion/cytotoxicityVentre et al. (2006)
PA 4112Encodes immunogenic peptides towards cystic fibrosis serum. HPT domainBeckmann et al. (2005)
PA 4856RetSVirulence, biofilm formation, type III secretion/cytotoxicityGoodman et al. (2004); Laskowski et al. (2004)
PA 4982Encodes peptides highly immunogenic to cystic fibrosis serum. HPt domain. Putative PA4983–PA4982 systemBeckmann et al. (2005)
  • * Annotated gene identification number (PA) and hybrid/unorthodox sensor kinase name according to the PAO1 genome at http://www.pseudomonas.com. Additional/uncharacterized hybrid sensors include PA1243, PA1396, PA1976, PA1992, PA2177, PA2583, PA2824, PA3271, and PA3462.

  • Where available each hybrid or unorthodox sensor is written following its cognate signaling protein(s) (e.g. response regulator). Sensors which include an HPt domain are unorthodox sensor kinases.

PhoP–PhoQ

The PhoP–PhoQ system mediates in part the adaptive response to low extracellular Mg2+ concentrations in P. aeruginosa (Fig. 1; McPhee et al., 2006). PhoP–PhoQ controls resistance to aminoglycosides, polymyxin B, and antimicrobial peptides, albeit likely in different fashions (Macfarlane et al., 1999, 2000). In response to low external concentrations of Mg2+ that are detected by the PhoQ sensor kinase, the PhoP response regulator regulates polymyxin B and antimicrobial peptide resistance mainly through upregulation of arnBCADTEF (Fig. 1). This operon encodes a pathway for addition of 4-aminoarabinose to lipid A of lipopolysaccharide, thereby reducing the net negative charge of lipopolysaccharide and limiting its interaction with polycationic antibiotics such as polymyxin B and cationic peptides (McPhee et al., 2003; Moskowitz et al., 2004). Lipid A plays an important role in the pathogenesis of many Gram-negative bacterial infections by acting as a signature of such infections, activating the innate immune system, triggering the synthesis of host defense peptides, cytokines, clotting factors, and other immunostimulatory molecules. CF P. aeruginosa isolates contain lipid A structures that are equivalent to those formed upon activation of PhoP–PhoQ, and appear to generate increased host inflammatory responses (Ernst et al., 1999). A phoP response regulator mutant displays wild-type polymyxin B resistance in low Mg2+ media and remains sensitive when grown under high Mg2+ conditions (in contrast to the situation in Salmonella); however, a phoQ mutant displays super-resistance to polymyxin B and antimicrobial peptides in both low (wild-type resistance) and high Mg2+ (a constitutive resistance phenotype). These findings suggest that in noninducing high Mg2+ conditions, PhoQ dephosphorylates PhoP (inactivating it), and conversely in the phoQ mutant this inactivation is lost and possibly some other protein/mechanism allows PhoP to be activated through phosphorylation. Therefore, as for wild-type cells grown in low Mg2+, the phoQ mutation leads to massive upregulation of the arnBCADTEF operon and resistance to polymyxin B and cationic peptides (McPhee et al., 2003; W.J. Gooderham & R.E.W. Hancock, unpublished data). The role of PhoP in this resistance appears to be significant as the overexpression of PhoP in a phoP mutant gives rise to phoQ-like constitutive polymyxin B resistance (Macfarlane et al., 1999). Another group found that a phoP mutant displayed intrinsic super-susceptibility to polymyxin B but, in contrast, not to C18G antimicrobial peptides (Ernst et al., 1999). Consistent with PhoPQ Mg2+-dependent regulation of lipopolysaccharide modifications, lipid A from a phoP mutant was not modified with aminoarabinose and palmitate (Ernst et al., 1999). In addition, cells become resistant to gentamicin under limiting Mg2+ conditions when induction of PhoP–PhoQ occurs (Hancock et al., 1981; Macfarlane et al., 1999). However, this regulation is quite complex as, under high Mg2+ conditions, phoQ (and polar phoP mutants also deficient in phoQ) demonstrated increased resistance to the aminoglycoside antibiotics streptomycin, kanamycin, and amikacin (Macfarlane et al., 2000; McPhee et al., 2006).

Figure 1

A model for the Pseudomonas aeruginosa PhoP–PhoQ and PmrA–PmrB regulatory networks in resistance to cationic antimicrobial peptides. Mg2+ limitation leads to the activation (phosphorylation) of the PhoP and PmrA response regulators which positively autoregulate the transcription of their respective operons, as well as the arnBCADTEF operon. The ArnBCADTEF pathway modifies Lipid A with aminoarabinose, reducing the net negative charge on lipopolysaccharide, and consequently decreasing self-promoted uptake across the outer membrane, increasing resistance to polymyxins and cationic antimicrobial peptides. Conversely, under high Mg2+ conditions, PhoP and PmrA are dephosphorylated and presumed inactive (not shown). However, during growth in high Mg2+ plus subinhibitory antimicrobial peptides, an unidentified regulatory system is proposed to promote activation of arnBCADTEF and pmrAB operons, consequently increasing resistance to antimicrobial peptides. See text for details of these systems. Other genes regulated by PmrA and PhoP are not shown.

The presence of polyamines in the growth medium has been shown to increase P. aeruginosa resistance towards multiple antibiotics, including polymyxin B, colistin, aminoglycosides, and quinolones (Kwon & Lu, 2006). This is linked to the ability of the polyamine spermidine to induce both the phoPQ and arnBCADTEF operons (no effect is observed on pmrAB). The PhoP response regulator was necessary for this spermidine-induced resistance as a phoP mutant exposed to spermidine no longer showed increased resistance to cationic peptides, polymyxin B and quinolone antibiotics. Extracellular Mg2+ ion concentrations had no effect on polyamine-mediated antibiotic resistance. Further complicating the situation, spermidine uptake mutants lost spermidine-induced resistance to aminoglycosides and quinolones but retained resistance to the cationic peptide antibiotics. PhoQ also appears to be involved in quinolone resistance as a phoQ mutant displayed low-level quinolone super-susceptibility (Kwon & Lu, 2006). It is possible that the large number of two-component systems in P. aeruginosa, and their potential for cross-talk, allows a different response towards polyamines as polyamine-induced polymyxin B resistance was not observed in Escherichia coli and Salmonella enterica serovar Typhimurium (Kwon & Lu, 2007).

The sensor kinase PhoQ is also required for biofilm formation when using a dynamic (greater aeration) biofilm assay, and also for the maturation of biofilms in flow cells (Ramsey & Whiteley, 2004). Our unpublished observations have revealed a role for PhoQ in regulating twitching motility, which may help explain the biofilm phenotype, and phoQ mutants are swarming impaired, while in contrast a phoP mutant is a super-swarmer (Brinkman et al., 2001). Furthermore, phoQ mutants are less cytotoxic towards mammalian cells. Consistent with these observed phenotypes related to virulence, a phoQ mutant is significantly less virulent in a neutropenic mouse model (Macfarlane et al., 1999) and our recent observations indicate a major role for PhoQ in chronic rat lung infection.

PmrA–PmrB

PmrA–PmrB is another two-component system that responds to low Mg2+ and regulates resistance to polymyxin B and antimicrobial peptides (Fig. 1; McPhee et al., 2003, 2006; Moskowitz et al., 2004). Interestingly, the pmrAB operon was also shown to be strongly induced by a variety of cationic antimicrobial peptides, including polymyxins, bovine indolicidin, human LL-37, and other synthetic peptides (McPhee et al., 2003), in stark contrast to Salmonella where the binding of peptides to PhoQ mediates peptide resistance induction (Bader et al., 2005). The lipopolysaccharide aminoarabinose modification operon arnBCADTEF is also strongly induced by peptides, and this activation was at least partially dependent on PmrA–PmrB (McPhee et al., 2003). Transposon mutations in either the pmrAB or arnBCADTEF operons result in supersusceptibility to both polymyxin B and cationic antimicrobial peptides (Lewenza et al., 2005). Interestingly, peptide-induced activation of the pmrAB operon appeared to be independent of both the PhoP–PhoQ and PmrA–PmrB systems (McPhee et al., 2003), which indicates that another two-component regulator likely responds to cationic peptides (Fig. 1).

PmrA–PmrB was also identified by isolating spontaneous mutants with increased resistance to polymyxin B (Moskowitz et al., 2004). These mutations mapped to the PmrB H-box motif, likely allowing PmrB to constitutively activate PmrA by increasing its net phosphorylation. Thus under high Mg2+, the regulation of resistance to a variety of structural classes of cationic antimicrobial peptides and the addition of aminoarabinaose to Lipid A was shown to be dependent on pmrAB (Moskowitz et al., 2004).

CzcR–CzcS and CopR–CopS

Cross-resistance between metals and antibiotics exists and has raised concerns about overlaps in environmental and clinical resistance (Baker-Austin et al., 2006). The CzcRS (also termed CzrRS) two-component system was originally described in the context of regulation of the CzcBCA resistance-nodulation-division efflux pump and the control of intrinsic cadmium and zinc resistance (Hassan et al., 1999). Subsequently, coregulation of resistance to trace heavy metals and the carbapenem antibiotic imipenem was demonstrated (Perron et al., 2004). Activation of the CzcS sensor kinase led to imipenem and zinc resistance. It was demonstrated that exposure of P. aeruginosa to sublethal zinc ion concentrations activates czcRS transcription, which in turn leads to transcriptional activation of the czcCBA RND efflux pump system and zinc resistance. Concurrently, the increased levels of CzcR negatively regulate OprD levels, the porin that is the primary route by which carbapenems like imipenem enter the cell, thus explaining cross-resistance to imipenem (Perron et al., 2004). This is in agreement with findings from another study which found that zinc eluted from siliconized latex urinary catheters exerted a negative effect on the expression of OprD and carbapenem resistance (Conejo et al., 2003).

Copper treatment induces resistance to both copper and zinc. However, in a czcRS mutant, copper exposure was still capable of decreasing OprD levels and inducing imipenem resistance, indicating the presence of another regulatory system that responds to copper and negatively effected oprD expression. This new system involved the CopR response regulator that was demonstrated to link zinc, copper, and imipenem resistance by activating the czcRS operon, which consequently led to CzcR-mediated negative regulation of OprD levels and resistance to imipenem (Caille et al., 2007). Transcription of copRS was induced in copper-shocked and copper-adapted cells, and was also found to regulate copper tolerance (Teitzel et al., 2006).

GacA–GacS

The GacS–GacA two-component system forms one of the best characterized virulence regulatory pathways. Multihost P. aeruginosa pathogenicity studies identified the GacA response regulator as a master regulator of virulence and revealed conservation amongst virulence mechanisms in different hosts (Rahme et al., 2000). Both gacA and gacS mutants were demonstrated to be required for pathogenicity in plant (Arabidopsis thaliana), nematode (Caenorhabditis elegans), insect (Drosophila melanogaster, Galleria mellonella, Bombyx mori), and mouse models of acute infection (Rahme et al., 1997; Mahajan-Miklos et al., 1999; Tan et al., 1999; Jander et al., 2000; Brinkman et al., 2001; Chieda et al., 2005). GacA also contributes to chronic infection as demonstrated in a CF murine model (Coleman et al., 2003). Both GacA and the GacS sensor kinase are required for activation of multiple extracellular virulence factors as well as chronic persistence genes, including those required for biofilm formation and N-acyl-homoserine lactone autoinducers (Reimmann et al., 1997; Parkins et al., 2001). GacS mutants are also impaired in swarming motility (Brinkman et al., 2001).

The signal transduction pathway from GacA to individual virulence genes varies, as recently reviewed by Lapouge et al. (2008). Activated (phosphorylated) GacA is thought to positively regulate the transcription of two small (untranslated) regulatory RNAs, RsmZ and RsmY, which appear to act in a parallel and redundant posttranscriptional fashion to jointly antagonize RsmA, a small RNA-binding protein that itself negatively regulates the expression of certain quorum sensing signals and several extracellular products (Fig. 2; Reimmann et al., 1997; Pessi et al., 2001; Kay et al., 2006). Sequestration of RsmA by RsmY and RsmZ leads to derepression of RsmA-repressed promoters (Heurlier et al., 2004).

Figure 2

A model for the opposing influence of the RetS and LadS on the GacA–Rsm regulatory network in Pseudomonas aeruginosa. Signals that activate RetS repress the activation of GacA (possibly by dephosphorylation; step 6) and the expression of small noncoding regulatory RNAs RsmZ/Y (step 6), leading to free RsmA and promotion of the production of virulence factors associated with acute infections (step 7; and biofilm repression). Conversely, signals that activate GacS and LadS lead to phosphorylation of GacA (steps 1 and 2) which increases the levels of the noncoding RsmZ/Y RNA molecules (step 3) and these RNAs bind and sequester RsmA (step 4). The decrease in free RsmA results in an increase in the expression of the biofilm matrix polysaccharides, Psl and Pel, and the consequent promotion of biofilms (step 5) and repression of acute virulence factors. See text for details of this regulatory network. Unknown signals also activate several sensor kinases (including RetS) to transfer phosphoryl groups to HptB (step 8) and initiate the HptB-PA3346-PA3347 phosphorelay pathway (steps 9 and 10) to ultimately repress biofilm formation (step 10) and increase swarming, through an as-yet unknown mechanism. This figure is based on models proposed in the literature (Goodman et al., 2004; Ventre et al., 2006; Hsu et al., 2008).

RsmA normally negatively regulates N-butanoyl-homoserine lactone quorum sensing signal molecules and consequently a variety of extracellular products including pyocyanin (phz), hydrogen cyanide (hcnABC), elastase (lasB), and a lectin (lecA). RsmA also positively regulates the production of lipase (lipAH) and rhamnolipid (rhlAB). The RsmA/RsmZ system is thus able to positively control swarming motility, rhamnolipid and lipase biosynthesis (Heurlier et al., 2004).

The GacA–GacS system also appears to be involved in antibiotic resistance. An early study showed that, relative to wild-type, a gacS mutant displayed supersusceptibility to the aminoglycosides, gentamicin and amikacin, and chloramphenicol (Brinkman et al., 2001). Subsequently it was shown that transcription of the sensor kinase-encoding gacS gene is repressed during growth in the presence of subinhibitory concentrations of three different classes of antibiotics, represented by tobramycin, ciprofloxacin, and tetracycline (Linares et al., 2006). This is apparently mediated through the RsmA/RsmZ system as rsmA was found to be transcriptionally induced (threefold) when P. aeruginosa cells were grown in the presence of subinhibitory concentrations of these three antibiotics (Linares et al., 2006). The mechanism(s) of resistance regulated by GacAS are poorly understood, and susceptibility to other antibiotics and cationic peptides are apparently unaffected by gacS mutations. However, mutation of gacS leads to an increased occurrence of antibiotic resistant small-colony variant cells in P. aeruginosa biofilms and it was suggested that in, some instances, GacS may be necessary for reversion of these variants to a wild-type state (Davies et al., 2007). Biofilms comprised of gacS mutant cells also produce phenotypically stable small colony variants with increased frequency when exposed to stresses such as hydrogen peroxide, human serum, or certain antibiotics (e.g. tobramycin, amikacin, oxacillin, piperacillin or rifampicin). These small-colony variants produce thicker biofilms and exhibited increased antimicrobial resistance compared to wild-type or parental gacS strains (Davies et al., 2007). Consistent with this, mutation of gacA leads to moderate, biofilm-specific decreased resistance, relative to wild-type, towards multiple antibiotics including azythromycin, chloramphenicol, erythromyin, piperacillin and tetracycline (Parkins et al., 2001).

RetS

The GacA/GacS/RsmZ pathway has been linked to two important two-component sensor/regulators of virulence, RetS and LadS (Fig. 2). RetS (regulator of exopolysaccharide and type III secretion) was identified during the initial screening of a panel of regulatory mutants for simple biofilm formation phenotypes (Goodman et al., 2004). This hybrid sensor kinase contains one sensor kinase domain and two tandem response regulator receiver domains. In a model of P. aeruginosa virulence adaptation, the hybrid sensor RetS was proposed to act in an opposite fashion to GacS (Goodman et al., 2004). Relative to wild type, a retS mutant led to massive increases in early biofilm formation that was linked to activation of the matrix exopolysaccharide psl operon. This phenotype correlated with hyperadhesion to mammalian CHO cells, but not increased cytotoxicity, as the retS mutant displayed impaired expression of the type III secretion system and associated effectors ExoS, ExoT and ExoY. In a mouse acute pneumonia model, the retS mutant was impaired in establishing lung infection and less able to disseminate systemically relative to wild type. Microarray analysis demonstrated the downregulation of several genes known to be required for acute infection, such as the type III secretion system, type II xcp secretion system, lipase (lipA), exotoxin A (toxA), and type IV pili genes. Conversely, the retS mutant was found exhibit derepression of specific genes associated with chronic infection such as the psl and pel exopolysaccharide-encoding operons. Thus, RetS acts normally to repress certain genes required for biofilm formation and activate expression of genes required for acute infection. The signal detected by RetS that governs this switch is as yet unknown. Screens and analysis of suppressor transposon mutants of retS mutant phenotypes revealed an important role for the GacS/GacA/RsmZ pathway in RetS signal transduction.

Another group similarly demonstrated that mutation of retS resulted in abolished expression of the type III secretion effectors ExoT and ExoU and a concomitant decrease in cytotoxicity towards epithelial cells and attenuated virulence in an acute infection model (Laskowski et al., 2004; Laskowski & Kazmierczak, 2006). This study however observed no differences between the retS mutant and wild type in epithelial cell attachment (Laskowski et al., 2004). Detailed mutational analysis of RetS indicated that the two tandem receiver domains are not equal and that the primary function of RetS may be to act as a phosphatase, as opposed to a kinase (Laskowski & Kazmierczak, 2006). This suggests a strong possibility of cross-talk with other sensor kinases.

Lost adherence sensor (LadS)

LadS was identified during the initial screening of transposon mutants (in a pilA mutant strain) for mutants with variant biofilm formation phenotypes (Ventre et al., 2006). Like RetS, LadS is a hybrid sensor kinase that shares similar domain organization with RetS, but unlike RetS it does not contain a second C-terminal response regulator receiver domain. A ladS mutant was unable to form biofilms and it appears LadS regulates both the early and later maturation stages of biofilm formation. This was traced to an ability of LadS to positively regulate the pel exopolysaccharide operon; indeed the opposite biofilm phenotypes of retS and ladS mutants likely reflect opposing influences on pel operon transcription (Fig. 2). Furthermore, transcriptional activation of exoT and resulting ExoT oversecretion in the ladS mutant correlated with increased cytotoxicity relative to wild type. Interestingly, a ladS retS double mutant displays enhanced biofilm formation and reduced cytotoxicity, phenotypes more reminiscent of a retS mutant, and possibly indicative of a signaling hierarchy in which LadS is upstream of RetS. Like RetS, LadS was found to be a component of the GacS/GacA/RsmZ signaling network in that deletion of ladS significantly decreased expression of rsmZ, whereas mutation of retS increased rsmZ transcription (Ventre et al., 2006). This has given rise to a model that attempts to explain the difference between acute infections (defined by expression and activation of the type III secretion system) and chronic infections (defined by a biofilm state), in which these three sensors are central in directing the expression of key virulence features (Fig. 2). In this model, LadS and GacS positively regulate rsmZ transcription leading to the chronic infection state, whereas RetS negatively regulates rsmZ transcription to promote an acute infection state. However, it is quite likely that this model understates the complexity of the switches involved.

It is not mechanistically known how LadS and RetS sensors accomplish signal transduction to modulate rsmZ and rsmY transcription. A simple explanation could be that RetS acts as a phosphatase to reverse the phosphorylation of GacA, whereas LadS acts to phosphorylate GacA (Laskowski & Kazmierczak, 2006) although this remains to be demonstrated experimentally and exactly how GacA modulates RsmZ levels is unknown. It is possible that Hpt proteins and other response regulators might be involved in the LadS and RetS modulatory pathways. Highlighting the prospective importance of these systems in controlling chronic infections, RetS and LadS regulate the HcpI secretion island I, which is required for chronic rat lung infection and thought to play an important role in chronic P. aeruginosa infections in CF by actively secreting HcpI (Potvin et al., 2003; Mougous et al., 2006). Discovery of the stimuli to which RetS, LadS, and GacS individually respond would provide substantial insight into how P. aeruginosa chronic and acute infection states are regulated. Interestingly, RetS and LadS share related 7TMR-DISMED2 periplasmic-sensing domains that have been predicted to bind carbohydrates, suggesting that they either bind host or bacterial-derived carbohydrates. We are unaware of any studies examining retS and lads mutants for possible antibiotic resistance phenotypes. These experiments would be of interest as these sensors are thought to be capable of influencing the GacA–GacS system, which has been implicated in antibiotic resistance and small-colony phenotypes (as discussed above).

SadA–SadR–SadS (RocA1–RocR–RocS)

Another system that appears to be a player in the acute/chronic infection switch is the SadA–SadR–SadS system, which regulates late biofilm maturation and the type III secretion (Kuchma et al., 2005). This tripartite system comprises a single sensor kinase (SadS) that interacts with two response regulators, SadA, a DNA-binding response regulator, and SadR, which contains an EAL output domain. Although they show no differences in twitching or swimming motility, or even in initial stages of biofilm formation, mutations in any one of sadA, sadR, or sadS lead to altered mature biofilm architecture. Furthermore, SadARS represses the type III secretion system specifically under conditions promoting biofilm formation. Again the nature of the signal involved is unknown and the sadARS locus itself does not appear to be induced during biofilm formation. Transposon mutants in sadS resulted in loss of virulence in a C. elegans-killing model (Gallagher & Manoil, 2001).

The sadARS genes were also simultaneously identified as rocS1–rocR–rocA1 and found to contribute to regulation of the cupB and cupC genes, which encode fimbrial adhesion factors involved in biofilm and pellicle formation (Kulasekara et al., 2005; Ruer et al., 2007). Examination of cupB and cupC expression indicated that the SadS sensor kinase and SadA response regulator form a cognate pairing to activate cup expression, whereas SadR represses cup expression by antagonizing SadAS signaling. Pseudomonas aeruginosa contains another rocS1rocA1 (sadAS) paralog, rocS2rocA2, and both systems share significant homology to the bvgAS system of Bordetella (Kulasekara et al., 2005).

AlgR–FimS (AlgZ)

In P. aeruginosa, the AlgR response regulator (Deretic et al., 1989) regulates several important virulence factors including alginate, lipopolysaccharide, and hydrogen cyanide production (Carterson et al., 2004), and twitching (Whitchurch et al., 1996) and swarming motility (Overhage et al., 2007). AlgR likely forms a cognate system with the atypical sensor FimS (=AlgZ) which regulates alginate production (Yu et al., 1997). Highlighting the importance of these factors in virulence, an algR mutant displayed attenuated pathogenesis relative to the wild type in acute septicemia (neutropenic mouse model), but was cleared more rapidly in a murine pneumonia model (Lizewski et al., 2002).

Alginate production (mucoidy) has been proposed to bestow a selective advantage on P. aeruginosa during chronic CF lung infections by protecting it from phagocyte killing. AlgR regulates mucoidy (Deretic & Konyecsni, 1989) and is an activator of the algD and algC promoters. AlgC is involved in the production of alginate, lipopolysaccharide, and rhamnolipids whereas algD is the first gene in the alginate biosynthetic operon.

In addition to AlgR, the AlgB response regulator regulates alginate production by activating algD transcription through binding to sites in the promoter (Wozniak & Ohman, 1991; Goldberg & Dahnke, 1992; Leech et al., 2008). AlgB apparently forms a cognate system with the KinB sensor kinase (Ma et al., 1997). Interestingly, neither AlgB nor AlgR appear to require phosphorylation of conserved aspartate residues in their receiver domains to positively regulate alginate production (Ma et al., 1998). Activation of these systems may be a response to stress, since exposure of nonmucoid P. aeruginosa to cell wall-damaging antibiotics (e.g. d-cycloserine) caused strong activation of algD transcription (Wood et al., 2006). This stress effect was shown to be dependent on the regulators AlgB and AlgR, but independent of KinB, partially mimicking the transcriptional effects resulting from mucoidy in P. aeruginosa (Wood et al., 2006).

AlgR also regulates biofilm formation. AlgR was shown to have a role in early biofilm formation because an algR mutant was deficient in static biofilm formation (Whitchurch et al., 2002). In contrast, flow-cell analysis showed no difference in biofilm formation by this mutant relative to wild type at 24 h, but by days 3 and 6 the algR mutant exhibited reduced biofilm suggesting a possible role for AlgR in regulating biofilm maturation in this model system (Morici et al., 2007). Gene expression analysis of an algR mutant indicated that in mature continuous culture biofilms, but not in planktonic algR cells, AlgR selectively represses Rhl, but not Las quorum sensing (Morici et al., 2007).

Both algR and fimS mutants display defective twitching motility phenotypes, and phosphorylation of AlgR is necessary for twitching motility (Whitchurch et al., 1996, 2002). Similarly, AlgR mutants demonstrate reduced swarming motility (Overhage et al., 2007). Activated AlgR positively regulates the fimU-pilVWXY1Y2E prepilin cluster during the stationary phase of growth by physically binding to this promoter (Lizewski et al., 2002; Belete et al., 2008). Activation of this prepilin operon leads to the assembly and export of a functional type IV pilus. However, although complementation with the fimTU-pilVWXY1Y2E operon was able to restore twitching in an algR mutant, this operon was unable to complement the mature biofilm deficiency, indicating that twitching does not play a role in the observed algR mature biofilm defect (Morici et al., 2007). Instead the algR biofilm defect could be linked to AlgR-mediated derepression of Rhl quorum sensing, which in turn resulted in increased rhamnolipid production and active motility (Morici et al., 2007). Therefore, normal development into mature biofilms appears to require AlgR-mediated repression of the Rhl quorum-sensing system during growth as biofilms.

FleR–FleS and PilR–PilS

Microarray analysis has provided an insight into flagella gene regulation. At the top of the hierarchy is FleQ, the master regulator of flagellar biosynthesis that controls regulation of various promoters due to differential binding affinity during various stages in flagella assembly (Dasgupta et al., 2003). FleQ is NtrC-like, but is missing the key conserved aspartate in the receiver domain. FleQ can transcriptionally regulate the FleS–FleR two-component system, which also regulates the flagella gene transcriptional network. Together with the alternative sigma factor RpoN, activated FleR positively regulates class III genes in mucoid strains (Dasgupta et al., 2003). The FleS–FleR system also controls both flagellar synthesis (motility regulation) and adhesion to mucin (Ritchings et al., 1995). We demonstrated recently that FleS mutants are modestly supersusceptible to ciprofloxacin and have a substantial defect in biofilm formation and swarming and swimming motility compared with the wild-type and a complemented mutant, indicating a role for this sensor in intrinsic antibiotic resistance (Breidenstein, in press).

The PilR–PilS system was first identified as capable of transcriptionally activating the pilin gene pilA and consequently controlling the expression of type IV pili (Ishimoto & Lory, 1992; Hobbs et al., 1993; Boyd & Lory, 1996). Interestingly, the inner membrane-embedded PilS sensor kinase was shown to be retained at both poles of the cell, despite localization of pili to one cell pole, which suggests an added role for PilS in the maintenance of cell polarity after cell division (Boyd et al., 2000). More recently, purified PilS was paired with a panel of response regulators in order to identify potential acceptor proteins in phosphorelay. In addition to its cognate regulator, PilR, both the conventional PA4843 and the hybrid PA2824 response regulators were identified as putative PilS substrates (Labaer et al., 2004). Although preliminary in nature, this suggests that P. aeruginosa histidine kinases can phosphorylate alternative response regulator substrates at rates similar to their cognate substrates, a phenomenon known in the field as cross-talk.

PvrR and PprB–PprA

The P. aeruginosa phenotypic variant response regulator PvrR plays a role in the formation of small colony variants that are hyper-adherent and antibiotic resistant (Drenkard & Ausubel, 2002). In the presence of high concentrations of the aminoglycoside kanamycin, pvrR mutation increases the frequency of appearance of these small colony variants (Drenkard & Ausubel, 2002). Interestingly, PvrR contains an EAL output domain that possesses bis-(3′,5′)-cyclic-dimeric-guanosine monophosphate (c-di-GMP) phosphodiesterase activity. This type of domain, together with the reciprocal activity of the GGDEF diguanylate cyclase c-di-GMP-synthesizing domain, regulates the cellular level of the second messenger c-di-GMP. To note, PvrR resembles the SadR (RocR) response regulator in terms of their common EAL output domain architecture. Phenotypic analysis of a variety of P. aeruginosa mutants in genes possessing GGDEF and EAL domains has strongly implicated c-di-GMP metabolism in virulence (Kulasakara et al., 2006)

The PprA–PprB two-component system regulates cell membrane permeability and influences susceptibility to antibiotics (Wang et al., 2003). The PprA sensor kinase was initially identified as a spontaneous mutant displaying increased resistance to aminoglycoside antibiotics. In a similar manner, introduction of a plasmid overexpressing the pprB response regulator into clinical isolates increased their susceptibility to aminoglycosides (Wang et al., 2003).

NarL–NarX

As the CF lung is supplied with a low but continuous level of oxygen, P. aeruginosa might preferentially respire in this environment using oxygen. However in regions deep within the lungs of CF patients that contain thick mucous it is likely that P. aeruginosa respires anaerobically using nitrate. Nitrate metabolism has been linked to motility, biofilm formation, and virulence. The NarX sensor kinase is necessary for swarming motility, although in contrast, the narL response regulator mutant displayed a super-swarming phenotype that was proposed to result from overproduction of rhamnolipids (Van Alst et al., 2007). Although the individual narX and narL mutants displayed normal swimming motility, a narXL double mutant, displayed reduced swimming motility, consistent with the possibility that both the sensor kinase and response regulator are jointly required for flagella-dependent movement. Furthermore, the hyper-biofilm phenotype displayed by a narXL double mutant implies that altered swimming might lead to impaired dispersal from the biofilm (Van Alst et al., 2007). Thus, the NarX–NarL regulon likely includes genes that are able to modulate motility and biofilm formation.

Concluding remarks

As discussed in this review, P. aeruginosa contains a variety of two-component signal transduction systems that regulate complex antibiotic resistance and virulence processes. There are still, however, many putative two-component systems encoded in the genome that await detailed characterization. Given the important functions of those two-component systems already characterized, it is likely that some of these uncharacterized systems have interesting roles in antibiotic resistance and virulence processes, and our preliminary screening supports this possibility. However as P. aeruginosa is environmentally ubiquitous, some of the signals and physiological adaptations regulated by these uncharacterized systems may be related to niche-specific environmental adaptations and might therefore be tricky to fully elucidate.

Presently, all phenotypically characterized P. aeruginosa hybrid kinases have been shown to have roles in virulence and associated processes (Table 2). This could be due to the enormous emphasis on understanding the regulation of virulence in this organism. In contrast, to date only hptB, one of three Hpt phosphorelay proteins, has been ascribed a preliminary signal transduction role in interconnecting the hybrid sensors PA1611, PA1976, PA2824, and RetS with the response regulator PA3346 (Hsu et al., 2008). The other two-phosphorelay proteins HptA and HptC, that are presumably required by a subset of hybrid sensors, have not been as well studied. For example it was observed that the expression of hptC was found to be transcriptionally repressed during growth in vitro in CF sputum medium (Palmer et al., 2005), but the consequences thereof were not further investigated. It will be interesting to see which hybrid sensor kinases and response regulators form phosphorelay networks with HptA and HptC as well as the adaptations that are regulated by these signaling modules. Elucidating the cognate sensor pairings for orphan response regulators would greatly facilitate our understanding of these systems and would shed light on the cross-talk that is likely to occur.

There is an apparent deficiency in our understanding of the signals/ligands sensed by the well-characterized histidine kinases, reflecting a general problem in the field. For example, the input signals for major regulators of virulence like GacS, LadS, and RetS, are unknown. Moreover, some of these signals might act indirectly with respect to transcriptional activation of promoters of two-component systems. For example, cationic antimicrobial peptides and low Mg2+ transcriptionally activate the pmrAB operon, enabling lipopolysaccharide modifications and resistance to antimicrobial peptides (McPhee et al., 2003). However, it seems likely that another regulator accomplishes the transcriptional activation of pmrAB (Fig. 1). When the specific inducing signal or ligand is unknown it makes phenotypic analyses of uncharacterized two-component systems extremely difficult in basic high-throughput screens. For example, PhoP and PmrA mutants do not demonstrate any observable antibiotic resistance phenotype, possibly due to redundancy; however knowing that Mg2+deficiency is the inducing signal permits one to observe a role for these regulators in antimicrobial peptide supersusceptibility.

It is known that P. aeruginosa strains present at the later stages of CF infections differ remarkably in phenotype from their early environmental colonizer strain predecessors. Some of the most commonly observed phenotypes include antibiotic resistance, loss of motility, appearance of small-colony variants, increased mutation rate, and decreased virulence factor production. Whole-genome analysis documenting P. aeruginosa genetic adaptations in CF airways has demonstrated that the virulence factors required for acute infections are often mutated (i.e. selected against) in chronic CF infections (Smith et al., 2006). Intriguingly no two-component system was found to be selected against (apart from FleQ, which lacks a key aspartate residue). Conversely, there is evidence that certain two-component systems might mediate the persistent adaptations required for chronicity, as described above for RetS, LadS and SadARS. This, therefore, lends added support to the idea that these systems could function as adaptive factors required to establish and maintain chronic infections. In this regard, the development of improved chronic infection models, and the increased use of existing models, should help in the elucidation of how P. aeruginosa establishes persistence in CF chronic infections.

It appears evident that two-component regulatory systems, and especially their conserved histidine and aspartate features, could represent novel therapeutic targets due to their widespread distribution in prokaryotic genomes (Stephenson & Hoch, 2002). Although this avenue of research has received substantial interest, antimicrobial therapeutics for human use that harness this insight are yet to be developed.

Acknowledgements

We are grateful for funding from the Canadian Cystic Fibrosis Foundation (CCFF) and the Canadian Institutes of Health Research. W.J.G. held graduate studentships from the CCFF and the Michael Smith Foundation for Health Research-Genome British Columbia. R.E.W.H. is a Canada Research Chair.

Footnotes

  • Editor: Miguel Camara

References

View Abstract