The Pseudomonas aeruginosa Chp Chemosensory System Regulates Intracellular cAMP Levels by Modulating Adenylate Cyclase Activity

Summary

Multiple virulence systems in the opportunistic pathogen Pseudomonas aeruginosa are regulated by the second messenger signaling molecule adenosine 3’, 5’-cyclic monophosphate (cAMP). Production of cAMP by the putative adenylate cyclase enzyme CyaB represents a critical control point for virulence gene regulation. To identify regulators of CyaB, we screened a transposon insertion library for mutants with reduced intracellular cAMP. The majority of insertions resulting in reduced cAMP mapped to the Chp gene cluster encoding a putative chemotaxis-like chemosensory system. Further genetic analysis of the Chp system revealed that it has both positive and negative effects on intracellular cAMP and that it regulates cAMP levels by modulating CyaB activity. The Chp system was previously implicated in the production and function of type IV pili (TFP). Given that cAMP and the cAMP-dependent transcriptional regulator Vfr control TFP biogenesis gene expression, we explored the relationship between cAMP, the Chp system and TFP regulation. We discovered that the Chp system controls TFP production through modulation of cAMP while control of TFP-dependent twitching motility is cAMP-independent. Overall, our data define a novel function for a chemotaxis-like system in controlling cAMP production and establish a regulatory link between the Chp system, TFP and other cAMP-dependent virulence systems.

Introduction

Pseudomonas aeruginosa is an opportunistic bacterial pathogen responsible for a wide array of human diseases including ulcerative keratitis, soft tissue infection, otitis, urinary tract infection, pneumonia, bacteremia and endocarditis (Driscoll et al., 2007). Infection by P. aeruginosa is typically associated with disruption of physical barrier function (burn injury, wounds and medical devices) and defective immune defense mechanisms (immunosuppression and acquired immunodeficiency syndrome). Consequently, P. aeruginosa is a leading cause of gram-negative hospital-acquired infection (Ruiz et al., 1999; Arancibia et al., 2002). In addition infection with P. aeruginosa is the primary cause of morbidity and mortality in individuals with cystic fibrosis where it causes chronic lung infection (Brennan and Geddes, 2002).

Many virulence factors associated with P. aeruginosa infection are regulated by the small molecule second messenger adenosine 3’, 5’-cyclic monophosphate (cAMP or cyclic AMP) (Wolfgang et al., 2003). In P. aeruginosa, cAMP is believed to control gene expression through allosteric regulation of the transcription factor Vfr (Virulence factor regulator), which is a member of the cAMP receptor protein (CRP) family (West et al., 1994). Cyclic AMP and Vfr appear to be the central components controlling a global virulence gene response in P. aeruginosa through regulation of multiple virulence systems including type IV pili (TFP) (Beatson et al., 2002; Wolfgang et al., 2003), the type II secretion (T2S) system and secreted toxins (West et al., 1994; Beatson et al., 2002; Wolfgang et al., 2003; Ferrell et al., 2008), type III secretion (T3S) (Wolfgang et al., 2003), quorum sensing (QS) (Albus et al., 1997) and flagellar biogenesis (Dasgupta et al., 2002). Proteome and transcriptome analyses of Vfr-deficient mutants (Suh et al., 2002; Wolfgang et al., 2003) indicate the extent and complexity of the Vfr regulon and also suggest that there may be more systems under Vfr control than are currently recognized.

Cyclic AMP is produced by the enzyme adenylate cyclase (AC) and P. aeruginosa encodes two putative intracellular ACs, CyaA and CyaB (Wolfgang et al., 2003). P. aeruginosa mutants lacking either cyaA or cyaB have reduced intracellular cAMP; however, the contribution of CyaB is substantially greater than that of CyaA (Wolfgang et al., 2003). Furthermore, previous studies indicate that CyaB is the primary source of cAMP associated with P. aeruginosa virulence (Wolfgang et al., 2003; Smith et al., 2004). CyaB is predicted to be a member of the ubiquitous Class III family of ACs widely distributed in both eukaryotes and prokaryotes (Baker and Kelly, 2004). CyaB consists of an amino (N)-terminal MASE2 domain (membrane associated sensor 2) and a carboxy (C)-terminal AC catalytic domain (Nikolskaya et al., 2003). The MASE2 domain is predicted to function as a membrane anchor and regulator of CyaB catalytic domain activity (Nikolskaya et al., 2003).

Although functional regulation of CyaB has not been demonstrated, we predict that control of cAMP synthesis by CyaB may be an important step in coordinating virulence gene expression. As a first step in our investigation of cAMP regulation, we provide definitive biochemical evidence that P. aeruginosa CyaB is capable of synthesizing cAMP in vitro. We next performed a genetic screen for regulators of intracellular cAMP and identified genes encoding multiple components of the Chp chemotaxis-like chemosensory system. The Chp system was previously shown to control TFP production and twitching motility (Darzins and Russell, 1997; Whitchurch et al., 2004), a form of surface-associated bacterial movement mediated by the extension and retraction of TFP fibers (Skerker and Berg, 2001). In addition, the Chp system appears to play a critical role in P. aeruginosa virulence that cannot be accounted for by its regulation of TFP alone (D'Argenio et al., 2001). Given that cAMP and the cAMP-dependent transcriptional regulator Vfr have been implicated in TFP regulation (Beatson et al., 2002; Wolfgang et al., 2003), we examined the regulatory link between the Chp system, cAMP production, TFP biogenesis and twitching motility. We demonstrate that the Chp system controls intracellular cAMP levels by modulating CyaB activity and that the regulated production of TFP is cAMP-dependent, thereby providing a mechanistic link between the Chp system and TFP biogenesis. In addition, we provide evidence that the Chp system controls TFP function (twitching motility) through a cAMP-independent mechanism. Our finding that the Chp chemosensory system controls cAMP production suggests that this signaling system also plays a more global regulatory role in P. aeruginosa virulence via control of the cAMP/Vfr-dependent virulence regulon.

Results

Biochemical characterization of the AC activity of P. aeruginosa CyaB

Inactivation of cyaB affects the intracellular cAMP pool in P. aeruginosa (Wolfgang et al., 2003), but there is no direct evidence that CyaB is an AC. The predicted catalytic domain of CyaB is located in the C-terminal region of the protein, while the N-terminus encodes a putative membrane-anchoring domain (Nikolskaya et al., 2003). To characterize the AC activity of CyaB, we cloned and expressed the putative catalytic domain (amino acids 217–463) as an N-terminal hexahistidine fusion. The recombinant protein (CyaB_217–463) was purified (>99% homogeneity) under native conditions (Fig. 1A) and subjected to biochemical analysis. Purified CyaB_217–463 displayed AC activity in the presence of either Mn²⁺ or Mg²⁺ as the divalent metal co-factor (Fig. 1B). Enzyme parameters were determined in the presence of Mn²⁺-ATP due to the higher specific activity obtained versus Mg²⁺-ATP. CyaB_217–463 displayed pH and temperature optima of 8.5 and 40°C, respectively. The enthalpy of activation (E_A) derived from the linear arm of an Arrhenius plot was 53.5±0.4 kJ mol⁻¹. Kinetic parameters were determined using 250 nM enzyme at pH 7.5 and gave a K_m for ATP of 66.7±4.8 µM and a V_max of 230±4.6 nmol cAMP mg⁻¹ min⁻¹. A Hill coefficient of 1.01±0.03 indicated non-cooperative binding of ATP at the two predicted active sites (Linder and Schultz, 2003). All values are consistent with previously characterized bacterial and mammalian Class III ACs (Cann et al., 2003; Litvin et al., 2003; Hammer et al., 2006). Analysis of the amino acid sequence of the presumed active site revealed that CyaB most likely belongs to sub-Class IIIb (Linder and Schultz, 2003) and, consistent with this prediction, alanine substitution of substrate specificity-determining residues (K274 and T351) significantly (p<0.0001) reduced AC activity (Fig. 1C). These results provide definitive evidence that P. aeruginosa CyaB is an AC.

FIG. 1. Biochemical characterization of P. aeruginosa CyaB.

A) Coomassie Blue-stained SDS-polyacrylamide gel of purified recombinant CyaB_217–463. A 1.5 µg portion of protein was applied and molecular-mass standards (in kDa) are indicated. B) CyaB_217–463 specific activity (n=4; ± standard error of the mean (SEM)) in the presence of 200 µM Mn²⁺-ATP (68 nM protein, 40°C, pH 8.5) or 1 mM Mg²⁺-ATP (13 µM protein, 40°C, pH 8.5). C) Wild type and mutant (K274A or T351A) CyaB_217–463 specific activities (n=8; ± SEM) assayed with 250 µM Mn²⁺-ATP (0.7 µM protein, 40°C, pH 7.5). Percentage indicates the portion of wild type activity. All values were significantly different (p<0.0001) compared to the value for wild type CyaB.

Construction of a quantitative cAMP reporter system

The activity of AC enzymes is typically regulated by cellular factors in response to intracellular and extracellular signals. However, there is currently no information regarding the signals or mechanisms that control activity of the major P. aeruginosa AC CyaB. To facilitate high-throughput screens for identifying candidate regulators of cAMP synthesis, we constructed a transcriptional reporter system for monitoring intracellular cAMP levels based on the E. coli lacP1 promoter. In E. coli, CRP activates transcription of lacZ in a cAMP-dependent manner by binding a CRP consensus sequence in the P1 promoter region of the lac operon (Ebright et al., 1984). P. aeruginosa Vfr was shown to interact with conserved nucleotides in the lacP1 promoter and restores cAMP-dependent lacZ transcription when expressed in an E. coli crp mutant (West et al., 1994; Suh et al., 2002; Kanack et al., 2006). Therefore we predicted that activity of the lacP1 promoter would be cAMP- and Vfr-dependent in P. aeruginosa. We created a transcriptional reporter by fusing the E. coli lacP1 promoter to the lacZ gene and then introduced the construct into the vacant ϕCTX phage attachment site on the chromosome of P. aeruginosa strain PAK (Fig. 2A). To test whether lacP1-lacZ transcription was sensitive to changes in intracellular cAMP, we assessed reporter activity in a set of isogenic mutant strains with an inactivated copy of cyaA, cyaB or both AC genes (cyaAB). The relative pattern of β-galactosidase activity in the AC mutants accurately reflected their respective intracellular cAMP levels as determined by direct measurement (Fig. 2B). We next compared intracellular cAMP and reporter activity in a vfr mutant. Reporter activity was eliminated in the vfr mutant, indicating that the lacP1 promoter is Vfr-dependent in P. aeruginosa. In contrast, the vfr mutant showed an approximately 50% reduction in intracellular cAMP compared to the wild type strain. This reduction suggests that Vfr may regulate expression of one or both AC genes or other factors that affect AC activity. The role of Vfr in regulating intracellular cAMP will be explored elsewhere. The correlation between lacP1-lacZ reporter activity and cAMP levels indicates that transcription from the E. coli lacP1 promoter can be used as a surrogate readout for intracellular cAMP production in P. aeruginosa.

FIG. 2. Activity of the lacP1-lacZ transcriptional reporter is cAMP-dependent and reflects the level of intracellular cAMP.

A) Schematic diagram of the lacP1-lacZ promoter reporter. The lacP1 promoter region of the E. coli lacZ gene contains two LacI binding sites (gray rectangles) and a CRP binding site (black rectangle). The transcriptional start site (+1) and −35 and −10 regions are indicated. The region of the lacP1 promoter extending 177 bp upstream of the transcriptional start site was fused to the lacZ gene. B) Direct measurement of cAMP by enzyme immunoassay (left axis, black bars) and β-galactosidase activity (right axis, gray bars) in the wild type strain (wt), the AC mutants (cyaA, cyaB and cyaAB), or the vfr mutant (vfr); all strains contained the lacP1-lacZ reporter. For the β-galactosidase assay, the data represent three independent experiments with four replicates each and values are reported as the mean ± SEM; values for all strains are significantly different (p<0.0001) when compared pair wise to the values for all other strains shown. For the cAMP assay, the data represent three independent experiments, each with three replicates and values are reported as the mean ± SEM. The asterisk (*) indicates that the values for the indicated strains are significantly different (p≤0.019) compared to the value for the wild type strain. C) A defect in cAMP synthesis can be restored by exogenous cAMP. β-galactosidase activity in the wild type strain (wt), AC double mutant (cyaAB) or vfr mutant (vfr) grown in the presence (+) or absence (−) of 50 mM cAMP. All strains contained the lacP1-lacZ reporter. The data represent two independent experiments with three replicates each and values are reported as the mean ± SEM.

Identification of P. aeruginosa genes encoding candidate CyaB regulators

To identify potential regulators of CyaB function we performed a transposon (Tn) mutagenesis screen in strain PAK carrying the cAMP-dependent transcriptional reporter system described above. The screen was carried out in a cyaA mutant strain such that CyaB was the only source of intracellular cAMP. We generated a library of approximately 60,000 random Tn insertion mutants and identified 125 candidates with reduced reporter activity by plating on LB agar plates containing 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal). Because the genome sequence of strain PAK is not available, the site of Tn insertion was determined for 121 of the mutants by comparing DNA sequence derived from each of the Tn-chromosomal junctions to the genome sequence of strain PAO1 (Stover et al., 2000; Winsor et al., 2009) (Table 1 and Fig. S1). Multiple Tn insertions occurred in cyaB and vfr, thus demonstrating that the screen produced mutants with the desired phenotype. The majority of the novel target genes identified in the Tn screen were previously implicated in the regulation of TFP biogenesis and/or twitching motility (Table 1). Both TFP production and twitching motility were previously shown to be Vfr-dependent (Beatson et al., 2002; Wolfgang et al., 2003). Multiple Tn insertions occurred in the pilGHIJKchpABC (Chp) gene cluster (Fig. S1), which encodes a chemotaxis-like chemosensory signal transduction system (Fig. S2)(Darzins and Russell, 1997; Whitchurch et al., 2004) predicted to regulate TFP. There were Tn insertions in the pilA and pilB genes, which encode pilin, the major structural subunit of TFP, and PilB, an ATPase required for TFP assembly, respectively (Strom and Lory, 1986; Turner et al., 1993). We also recovered mutants with Tn insertions in fimL and fimV, the gene products of which are required for wild type levels of twitching motility and TFP production (Semmler et al., 2000; Whitchurch et al., 2005). FimL is homologous to the N-terminus of the Chp system protein ChpA while FimV has no known homolog. The only targets identified in the screen not associated with TFP production or function were PA1821, grpE and mexF. PA1821 is immediately upstream of fimL (Fig. S1) and is predicted to encode an enoyl-CoA hydratase/isomerase. The grpE and mexF genes encode a cytoplasmic heat shock protein and an inner membrane component of a multidrug efflux pump, respectively.

Table 1.

Transposon insertion mutants with altered cAMP reporter activity

number	locus*	function / pathway
39	cyaB	adenylate cyclase, cAMP synthesis
27	vfr	cAMP-dependent transcriptional regulator
13	fimV	TFP / biogenesis (?)
10	pilJ	TFP / biogenesis, chemosensing
9	chpA	TFP / biogenesis, chemosensing
8	fimL	TFP / biogenesis, chemosensing (?)
5	pilG	TFP / biogenesis, chemosensing
3	pilI	TFP / biogenesis, chemosensing
3	PA1821	probable enoyl-CoA hydratase/isomerase
1	pilA	TFP / biogenesis
1	pilB	TFP / biogenesis
1	grpE	chaperone, heatshock protein
1	mexF	multi-drug efflux

^*site of transposon insertion lies within the coding sequence or putative promoter region of the indicated gene.

To validate the Tn screen, we performed β-galactosidase assays to quantify reporter activity in a representative Tn insertion mutant for each candidate regulator (Fig. 3). All of the Tn insertion mutants displayed reduced reporter activity, with the majority having activity equivalent to that of the cyaAB mutant. As a control, we showed that a randomly selected Tn mutant (pirA∷Tn) that formed a blue colony on the primary selection plates had activity equivalent to that of the parent (cyaA) strain.

FIG. 3. Quantitation of cAMP reporter activity in Tn insertion mutants.

β-galactosidase activity of the cyaA mutant (parent), the AC double mutant (cyaAB) and representative Tn insertion mutants in the PAKcyaA∷lacP1-lacZ background grown in broth culture to mid-logarithmic growth phase. The gene disrupted by Tn insertion is indicated for each strain. The pirA mutant is a clone from the Tn insertion library screen that had a blue colony phenotype on X-gal indicator plates. The data represent three independent experiments with four replicates each and values are reported as the mean ± SEM. All values (except pirA∷Tn) were significantly different (p<0.0001) compared to the value for the parent strain.

Because the lacP1-lacZ reporter is Vfr-dependent, the regulatory effect of cAMP and Vfr cannot be distinguished. However, we discovered that the addition of cAMP to the growth media was sufficient to rescue reporter activity in an AC mutant (cyaAB) but not in a vfr mutant (Fig. 2C). This result indicates that P. aeruginosa, like other gram-negative bacteria, can transport extracellular cAMP (Saier et al., 1975), albeit by an unknown mechanism. To determine if reduced reporter activity in the Tn mutants was due to reduced intracellular cAMP, we tested whether the phenotype could be complemented by exogenous cAMP. With the exception of the Tn insertions in vfr, reporter activity was restored in all of the Tn mutants in the presence of exogenous cAMP (data not shown), indicating that the defect in each of the mutants specifically affects intracellular cAMP.

Multiple Chp chemosensing system and TFP regulatory genes affect intracellular cAMP

Given the established connection between the cAMP-dependent transcriptional regulator Vfr and TFP (Beatson et al., 2002; Wolfgang et al., 2003), we focused on the genes previously implicated in regulating TFP production and/or twitching motility for their role in cAMP regulation. To confirm that the targets identified in our screen affect intracellular cAMP, we constructed individual non-polar deletion mutants for the genes encoding PilA, PilB, FimL, FimV, PilG (CheY), PilI (CheW), PilJ (MCP) and ChpA (CheA). For the Chp chemosensory system components (Fig. S2) the E. coli Che system homolog is indicated in parentheses. In addition, we individually deleted the remaining Chp system genes encoding PilH (CheY), PilK (CheR), ChpB (CheB) and ChpC (CheW) to assess their contribution to intracellular cAMP levels. To facilitate complementation of the deletion mutants using a LacI-regulated plasmid-based gene expression system, it was necessary to modify the lacP1-lacZ cAMP reporter, which is subject to LacI repression. We eliminated the lacI binding sites within the lacP1 promoter and demonstrated that the resulting construct (lacP1ΔlacI-lacZ) is not responsive to LacI but is still cAMP-dependent (Fig. S3). Each of the non-polar deletion mutations was introduced into a cyaA mutant strain carrying a chromosomal copy of the lacP1ΔlacI-lacZ reporter. We determined the level of reporter activity and the level of intracellular cAMP in each mutant strain (Fig. 4A). In general, reporter activity mirrored direct cAMP measurements, with the exception of the pilH mutant, where reporter activity is likely to be at its maximum. A subset of mutants (pilG, pilI, pilJ, chpA, fimL and fimV) had cAMP levels equivalent to that of the adenylate cyclase (cyaAB) double mutant while the pilH and pilK mutants had higher cAMP levels than the parent (cyaA) strain. Intracellular cAMP levels in the pilA, chpB and chpC mutants were not statistically different relative to the parent strain, but their values for the reporter assay were. The values for the pilB mutant were not different than that of the parent strain in either assay. Together, these results indicate that fimL, fimV, pilA and the members of the Chp gene cluster (pilGHIJKchpABC) all affect intracellular cAMP and suggest that the individual components exert either a positive (pilG, pilI, pilJ, chpA, chpC, fimL and fimV) or negative (pilH, pilK and chpB) regulatory effect on intracellular cAMP.

FIG. 4. Inactivation of TFP/Chp genes affects intracellular cAMP.

A) Direct measurement of cAMP by enzyme immunoassay (left axis, black bars) and β-galactosidase activity (right axis, gray bars) in the cyaA mutant (parent), the double AC mutant (cyaAB), or the indicated non-polar deletion mutant in the PAKcyaA∷lacP1ΔlacI-lacZ background. For the β-galactosidase assay, the data represent three independent experiments with two replicates each and values are reported as the mean ± SEM; all values (except pilB) were significantly different (p≤0.0003) compared to the value for the parent strain. For the cAMP assay, the data represent three independent experiments with three replicates each and values are reported as the mean ± SEM; the asterisk (*) indicates that the values for the indicated strains are significantly different (p≤0.04) compared to the value for the parent strain. B) Complementation of the TFP/Chp mutants by expression of the corresponding gene in trans. Activity of the lacP1ΔlacI-lacZ reporter in the non-polar deletion mutant strains harboring either the empty pMMB67EH vector (V, black bars) or the pMMB-based plasmid (C, gray bars) encoding the corresponding complementing gene. The TFP/Chp genes were each expressed in the pMMB plasmid backbone listed in the experimental procedures. The cyaAB mutant (cyaAB) harboring either pMMB67EH (V) or pMMBV2-cyaA (C) serve as controls for complementation of lacP1ΔlacI-lacZ reporter activity. Strains were grown in broth culture containing IPTG at the concentration determined to be optimal for complementation of the mutant phenotype (see text) and assayed at mid-logarithmic growth phase. Relative reporter activity is reported as the β-galactosidase activity of an individual mutant strain divided by that of the cyaA mutant parent strain harboring pMMB67EH. The data represent a minimum of three independent experiments with two replicates each and values are reported at the mean ± SEM.

To confirm the role of the identified TFP/Chp genes in regulating cAMP, we determined whether the cAMP phenotype of the mutants could be complemented in trans. For the mutants with altered intracellular cAMP, we created expression plasmids by cloning the coding region of each gene into multiple derivatives of the P. aeruginosa expression vector pMMB67EH containing different versions of the isopropyl-β-D-thiogalactopyranoside (IPTG)-inducible tac promoter (see experimental procedures). We transferred the plasmids to their respective mutant strains in the PAKcyaA∷lacP1ΔlacI-lacZ background and determined the optimal induction conditions for each gene by assessing β-galactosidase activity. For each of the mutants, we established an expression level at which the corresponding gene either restored cAMP reporter activity (pilA, fimL, fimV, pilG, pilI, pilJ, chpA and chpC) or reduced elevated cAMP reporter activity to near wild type levels (pilH and chpB) (Fig. 4B). We were unable to complement the pilK mutant at any of the expression levels tested (data not shown). To investigate the possibility that deletion of pilK coding region affected distal gene expression, we made an alternative version of the pilK mutant by ablating the predicted translational start site. Both versions of the pilK mutant displayed equivalent levels of reporter activity, but plasmid expression of pilK failed to complement either of the mutants. Therefore we could not confirm the cAMP phenotype of the pilK mutant. Although the role of pilK in regulating cAMP remains inconclusive, the fact that mutants lacking the other TFP regulatory genes identified in our screen had altered cAMP and could be complemented in trans supports the conclusion that these genes contribute to the regulation of intracellular cAMP in P. aeruginosa.

The Chp chemotaxis-like system regulates intracellular cAMP levels by modulating CyaB activity

The candidate CyaB regulators affect intracellular cAMP, but the mechanism of regulation is not known. We hypothesized that these regulators could affect the intracellular cAMP pool via cAMP synthesis at the level of cyaB expression or CyaB activity or through altered cAMP degradation. P. aeruginosa possesses cAMP phosphodiesterase activity, which results in cAMP hydrolysis (Siegel et al., 1977), but this activity has not been genetically defined. To address cAMP regulation by the Chp system, we first evaluated the amount of CyaB protein produced in each of the deletion mutants by immunoblotting with CyaB-specific antibody (Fig. 5). Despite the fact that the mutants had altered intracellular cAMP (Fig. 4A), the level of CyaB produced was equivalent to that of the parent strain (cyaA). This result indicates that the amount of CyaB enzyme does not account for altered levels of cAMP in the TFP/Chp system mutants.

FIG. 5. CyaB protein levels do not account for altered intracellular cAMP in the TFP/Chp mutants.

CyaB immunoblot of whole cell lysates from the cyaA parent strain (cyaA), the double AC mutant (cyaAB) and the indicated mutants in the PAKcyaA∷lacP1ΔlacI-lacZ background. Samples were normalized based on total protein and blots were probed with affinity-purified CyaB-specific rabbit antiserum.

We further examined the regulatory effect of the Chp system on cAMP by evaluating the effect of PilG and PilH on cAMP pools. PilG and PilH, which are both CheY homologs (Fig. S2)(Darzins, 1993; Darzins, 1994), have opposite effects on intracellular cAMP accumulation (Fig. 4A). We tested the ability of different cAMP sources to restore wild type levels of cAMP reporter activity in strains lacking AC activity in addition to either pilG or pilH (Fig. 6A). Because of the potential regulatory effect of PilG and PilH on CyaB, we used plasmid-based overexpression of P. aeruginosa cyaA (pMMBV2-cyaA) and addition of exogenous cAMP as alternative means to restore intracellular cAMP. Both alternative sources of cAMP restored reporter activity in the cyaAB, cyaABpilG and cyaABpilH mutants to equivalent levels. The fact that the level of cyaA expression or exogenous cAMP that rescued the cyaAB mutant phenotype also complemented the triple mutants indicates that PilG and PilH do not affect intracellular cAMP levels via cAMP degradation. Plasmid-based cyaB expression restored reporter activity in the cyaAB mutant, but failed to restore reporter activity in the cyaABpilG mutant and resulted in higher reporter activity in the cyaABpilH mutant, suggesting that the differential effects of PilG and PilH on intracellular cAMP are specific for CyaB. There were equivalent levels of CyaB protein in each of the cyaB-expressing strains (Fig. 6B), eliminating the possibility that different amounts of the enzyme accounted for different levels of reporter activity. Taken together, these data support a model in which PilG and PilH affect CyaB activity but have specific and opposite effects.

FIG. 6. The Chp system regulates cAMP at the level of CyaB activity.

β-galactosidase activity of the wild type strain (wt), the double AC mutant (cyaAB), and the cyaAB, pilG (cyaABpilG) and cyaAB, pilH (cyaABpilH) triple mutants. All strains contained the lacP1ΔlacI-lacZ reporter and carried either the pMMB67EH empty vector or pMMBV2-cyaA (cyaA) or pMMBV2-cyaB (cyaB). Bacteria were grown in broth culture in the presence (+) or absence (−) of 50 mM cAMP. The data represent two independent experiments with four replicates each and values are reported as the mean ± SEM. The asterisk (*) indicates that the values for the indicated strains are significantly different (p<0.0001) when compared pair wise to the value for the cyaAB mutant carrying pMMBV2-cyaB. B) CyaB immunoblot of whole-cell lysates from the above strains. Samples were normalized based on bacterial number and blots were probed with affinity-purified CyaB-specific rabbit antiserum.

The Chp chemotaxis-like system controls TFP production but not twitching motility via regulation of intracellular cAMP

Based on the previous findings that Vfr affects TFP production and twitching motility, and that cAMP and Vfr are required for the expression of multiple TFP biogenesis genes (Beatson et al., 2002; Wolfgang et al., 2003), we hypothesized that the TFP regulators identified by our screen may affect TFP production and twitching motility through their control of cAMP. To test this hypothesis, we first established whether each of the mutants has a TFP production/function defect corresponding to its respective level of intracellular cAMP. The available information regarding the TFP phenotypes of the fimL, fimV and Chp locus mutants, however, comes from multiple studies using different P. aerguinosa strains and different methodologies (Darzins and Russell, 1997; Semmler et al., 2000; Whitchurch et al., 2004; Whitchurch et al., 2005). Therefore, we systematically evaluated the TFP phenotypes of the non-polar isogenic mutants in strain PAK. As an indicator of TFP production, we examined the amount of pilin, the major subunit of TFP, that could be recovered from the bacterial surface (Fig. 7A). To determine the sensitivity of our assay and to estimate relative TFP levels, we compared serial dilutions of the pilin-containing surface fractions recovered from the mutant and parental (cyaA) strains (Fig. S4). The level of surface pilin recovered from the pilH and chpB mutants was approximately 4-fold and 2-fold greater than that of the parent strain, respectively. The level of surface pilin recovered from the pilK mutant was elevated but less than 2-fold greater than that of the parent strain. In contrast, the parent strain produced approximately 2-fold more surface pilin than the chpC mutant and 50-fold more surface pilin than the cyaAB and fimL mutants. For the remainder of the mutants (fimV, pilG, pilI, pilJ and chpA) surface pilin levels were below the limit of detection of our assay (<1% of the parental strain), a phenotype similar to that of the pilA mutant, which lacks pilin and serves as a negative control. With the exception of the pilA mutant, all of the mutants produced intracellular pilin as assessed by immunoblot of whole bacterial cell lysates (Fig. 7B). This result indicates that altered pilin synthesis does not account for the differences in TFP production in the affected mutants and suggests that the corresponding defects are at the level of TFP assembly. In addition, altered TFP production in the pilA, pilG, pilH, pilI, pilJ, chpA, chpB, chpC, fimL and fimV mutants could be complemented by plasmid expression of the corresponding gene (data not shown). Although slightly more surface pilin was recovered from the cyaAB and fimL mutants than any of the Chp mutants lacking cAMP (pilG, pilI, pilJ and chpA), the overall trend was that mutants with reduced cAMP had a corresponding reduction in surface TFP production while mutants with increased cAMP (pilH, pilK and chpB) had greater levels of surface pilin (Fig. 4A and 7A).

FIG. 7. The role of the TFP/Chp genes in TFP production and twitching motility.

A) Coomassie Blue-stained SDS-polyacrylamide gel showing the amount of pilin in purified surface pilus fractions from the cyaA mutant (cyaA), a non-piliated pilA mutant (pilA), the double AC mutant (cyaAB) and the indicated mutants in the PAKcyaA∷lacP1ΔlacI-lacZ background. Samples for pilus preparations were normalized based on bacterial number. Pilus preps were performed three times and a representative gel is shown. B) Immunoblot of whole-cell lysates from the above strains probed with a pilin-specific monoclonal antibody. C) Twitching motility zones for the indicated strains. The data represent at least three independent experiments with a minimum of three replicates each and values are reported as the mean ± SEM.

We also tested TFP-dependent twitching motility as an indicator of pilus function by measuring the extent of subsurface colony expansion (Fig. 7C). The cyaAB, pilK, chpB, chpC and fimL mutants produced twitching motility zones similar to that of the parent strain. However, in contrast to the pilK, chpB and chpC mutants, the cyaAB and fimL mutants had reduced cAMP and TFP (Fig. 4A and 7A). The pilG, pilH, pilI, pilJ, chpA and fimV mutants had a severe twitching motility defect; however, only the pilG, pilI, pilJ, chpA and fimV mutants had a corresponding defect in cAMP production and surface TFP, whereas the pilH mutant produced elevated levels of cAMP and TFP. For the mutants with altered twitching motility, we were able to complement the defect by plasmid-based expression of the corresponding gene (data not shown). The fact that most of the cAMP-lacking TFP/Chp mutants (pilG, pilI, pilJ, chpA and fimV) had a more severe twitching motility defect than the double AC (cyaAB) mutant and that the pilH mutant was defective for twitching motility despite having elevated cAMP (Fig. 4A and 7C) indicates that cAMP levels do not correlate with twitching motility. Overall, these results suggest that the TFP/Chp system may control TFP function in a cAMP-independent manner.

To further explore the hypothesis that the Chp system controls TFP biogenesis but not twitching motility via cAMP, we determined whether restoration of wild type levels of intracellular cAMP can complement the altered TFP phenotypes associated with the pilG and pilH Chp system mutants. As a first step, we determined that inclusion of 20 mM cAMP in LB agar restored TFP production (Fig. 8A), twitching motility (Fig. 8C) and cAMP-dependent reporter activity (data not shown) to wild type levels in the cyaAB mutant. We then assessed TFP production and twitching motility in the cyaABpilG and cyaABpilH triple mutant strains in the absence or presence of 20 mM cAMP (Fig. 8). In the absence of exogenous cAMP, the cyaABpilG mutant was similar to its cyaApilG parent strain in that it lacked TFP and was defective for twitching motility. Complementation of the cyaABpilG mutant with exogenous cAMP restored TFP production but not twitching motility. The cyaABpilH mutant (lacking endogenous cAMP) produced less TFP than its parental strain (cyaApilH), which synthesizes excess cAMP (Fig. 4A); however, there was no difference in twitching motility between these strains. Complementation of the cyaABpilH mutant with exogenous cAMP resulted in an increase in TFP production but not twitching motility. Immunoblot analysis of whole cell lysates showed that each of the mutant strains was capable of producing pilin (Fig. 8B), indicating that altered TFP production was due to a defect in assembly and not subunit availability. In summary, complementation of the cyaABpilG and cyaABpilH triple mutants with exogenous cAMP resulted in increased TFP levels (Fig. 8A), suggesting that the Chp system controls TFP production via cAMP. In contrast, adjusting intracellular cAMP to wild type levels failed to alleviate the twitching motility defects of either the cyaABpilG or cyaABpilH triple mutants (Fig. 8C), indicating that components of the Chp system regulate TFP function independent of cAMP. Taken together, these findings support a model in which cAMP is the primary mediator for Chp system control of TFP biogenesis, but not TFP function.

FIG. 8. The Chp system regulates TFP production but not twitching motility via control of intracellular cAMP.

A) Coomassie Blue-stained SDS-polyacrylamide gel showing the amount of pilin in purified pilus fractions from the wild type (wt) strain, the double AC mutant (cyaAB), the pilG and pilH mutants in a cyaA background (pilG and pilH, respectively) or the cyaAB, pilG (cyaABpilG) and cyaAB, pilH (cyaABpilH) triple mutants. All strains contained the lacP1ΔlacI-lacZ reporter. Bacteria were grown on LB agar plates in the presence (+) or absence (−) of 20 mM cAMP and were normalized based on bacterial number. B) Immunoblot of whole-cell lysates from the above strains probed with a pilin-specific monoclonal antibody. C) Twitching motility zones for the indicated strains grown in the presence (+) or absence (−) of 20 mM cAMP. The data represent two independent experiments with three replicates each and values are reported as the mean ± SEM. The asterisk (*) indicates that the values for the indicated strains are significantly different (p<0.001) compared to the value for the cyaAB mutant grown in the presence of cAMP.

Discussion

The results of this study provide evidence for new regulatory connections in the complex pathway controlling P. aeruginosa virulence factor production and suggest that regulation of intracellular cAMP by the Chp chemosensory system represents a mechanism by which environmental sensing is coupled to virulence gene expression. The regulation of intracellular cAMP by the Chp system represents an output not previously recognized for related bacterial chemosensory systems (Wadhams and Armitage, 2004; Berleman and Bauer, 2005; Kato et al., 2008; Mignot and Kirby, 2008; Porter et al., 2008). We hypothesize that the Chp system regulates TFP and other cAMP-dependent virulence systems in response to as yet unidentified signals in the host environment via the following steps: i) signal reception, ii) signal transduction, iii) modulation of AC activity and iv) cAMP/Vfr-dependent virulence gene regulation (Fig. 9).

FIG. 9. Model for coupling environmental sensing to virulence factor expression via the P. aeruginosa Chp chemosensory system.

1) In response to unknown signal(s), the Chp system controls cAMP synthesis by modulating activity of the AC enzyme CyaB, which is predicted to be integral inner membrane protein. The cAMP-dependent transcription factor Vfr mediates transcriptional activation of multiple virulence systems, including 2) T3S, QS, flagellar biosynthesis, multiple toxins and degradative enzymes and 3) the structural and regulatory factors responsible for TFP biogenesis. 4) The Chp system also exerts a regulatory effect on TFP function (twitching motility) that is cAMP-independent.

Signal reception

Currently the nature of the signal(s) received by the Chp system is not known. Previously, low calcium and high osmolarity in vitro culture conditions were shown to affect intracellular cAMP (Wolfgang et al., 2003; Rietsch and Mekalanos, 2006), but it is not known if these signals are relevant in vivo or if they act through the Chp system. Other possible signals include phosphatidylethanolamine and phosphatidylcholine, which are chemoattractants for TFP-dependent twitching motility in P. aeruginosa (Kearns et al., 2001; Barker et al., 2004; Miller et al., 2008). An intriguing possibility is that TFP fibers themselves relay signaling information for regulating cAMP, similar to the mechanism in Myxococcus xanthus where TFP are proposed to serve as sensors for the Dif chemosensory system that controls TFP-dependent social motility (Black et al., 2006).

Signal transduction

Our analysis of the individual Chp system components suggests that cAMP levels are responsive to a stimulatory signal transduced by the Chp system as well as a Chp-dependent signal attenuation/feedback mechanism. Using a combination of forward and reverse genetics we showed that pilG, pilI, pilJ, chpA and chpC had a positive effect on cAMP, while pilH and chpB had a negative effect. While preliminary, these results support a model in which Chp system regulation of cAMP involves both activation and adaptation pathways, which are conserved features among bacterial chemosensing systems (Fig. S2). Our finding that pilG, pilI, pilJ, chpA and chpC all had a positive effect on cAMP is consistent with these components acting sequentially in the signal relay. The fact that a strain lacking chpB, which encodes a putative methylesterase, has elevated cAMP suggests that the Chp system is subject to regulation by a traditional adaptation pathway involving MCP methylation/demethylation. However, further confirmation of the adaptation pathway is necessary given our inability to confirm the phenotype of the pilK (methlytransferase) mutant.

An interesting feature of the Chp system is that the CheY-like proteins PilG and PilH have different effects on cAMP production by CyaB. CheY-like proteins are typically activated by CheA-dependent phosphorylation in response to signal detection and, in many γ-proteobacteria, signal termination via dephosphorylation is mediated by CheZ (Wadhams and Armitage, 2004). PilG and PilH both possess conserved phosphorylation sites; however, no CheZ-like phosphatase is associated with the P. aeruginosa Chp system. We hypothesize that signal termination for the Chp system may occur by a mechanism similar to that originally described in the Sinorhizobium meliloti chemotaxis system in which one CheY-like protein serves as a phosphate sink by accepting phosphoryl groups from a second CheY-like effector (Sourjik and Schmitt, 1998). Based on this model, we predict that phosphorylated PilG is a positive effector of CyaB activity and that PilH is not an effector but instead a phosphate sink, which is consistent with its negative effect on CyaB activity. An alternative hypothesis is that PilG and PilH are both true effectors, but each has a different rate of phosphorylation and/or spontaneous dephosphorylation, allowing them to act sequentially to first activate (PilG) and then inhibit (PilH) cAMP production.

Modulation of AC activity

This study provides evidence that the P. aeruginosa Chp system regulates cAMP through alteration of CyaB activity. Based on the structural similarities between P. aeruginosa CyaB and other Class III ACs, the enzyme is predicted to have at least two potential regulatory mechanisms. First, the adjacent membrane-spanning MASE2 domain, like other regulatory modules present in Class III ACs, is likely to be involved in the detection of intracellular or extracellular signals that are relayed to the catalytic domain (Nikolskaya et al., 2003; Linder, 2006). Second, CyaB belongs to the subset of Class III ACs (Class IIIb) in which the catalytic domain is subject to direct regulation by the bicarbonate ion (Cann et al., 2003; Linder, 2006). To our knowledge, regulation of P. aeruginosa CyaB by the Chp system is the first example of a bacterial Class III AC controlled by specific signaling proteins. However, it remains to be determined if Chp system regulation of CyaB is direct and whether signal detection involves interaction with the MASE2 and/or catalytic domains.

cAMP/Vfr-dependent virulence gene regulation

Although a direct role for cAMP in Vfr activation has not been demonstrated, we predict cAMP is required for Vfr activity in regulating Vfr-dependent virulence systems such as T3S, QS and the TFP biogenesis genes. Of the cAMP regulators identified in our screen, only FimL has been previously linked to Vfr regulation (Whitchurch et al., 2005). FimL was shown to control multiple virulence factors in strain PAO1, including TFP biogenesis and function, T3S, and biofilm formation, via its regulation of Vfr. The authors ruled out cAMP’s involvement because fimL had no effect on intracellular cAMP levels but the evidence that fimL twitching revertants had elevated cAMP makes the interpretation less clear. In contrast, we showed a definitive role for fimL in cAMP regulation in strain PAK. Although FimL is homologous to the N-terminus of ChpA, it remains to be determined if there is a functional relationship between FimL and the Chp chemosensing system.

The fact that multiple P. aeruginosa virulence factors are regulated by cAMP implicates FimL, FimV and the Chp system components in general virulence system regulation. In fact, other groups have reported significantly reduced virulence phenotypes associated with mutations in these genes. Tn insertion mutants in fimV and the chp gene cluster were attenuated in a Drosophila infection model, an effect that could not be accounted for by the loss of TFP function alone (D'Argenio et al., 2001). In cell culture assays, inactivation of fimV, fimL or pilJ had a greater effect on cytotoxicity than adherence (Kang et al., 1997; Ahn et al., 2004; Whitchurch et al., 2005), suggesting these mutants have defects in multiple virulence systems. The adherence and cytotoxic phenotypes of these mutants are similar to those of mutants lacking AC activity or Vfr (Wolfgang et al., 2003), which is consistent with the involvement of these genes in cAMP regulation. In addition, the fact that fimL and pilJ mutants were shown to have impaired T3SS activation (Whitchurch et al., 2005; Zolfaghar et al., 2005) provides direct evidence for regulation of cAMP-dependent virulence factors by the TFP/Chp components.

Our analysis of TFP function also revealed that although twitching motility is ultimately dependent on TFP biogenesis, the Chp system exerts cAMP-independent regulatory control over TFP function. Currently the mechanism by which the Chp system regulates twitching motility is not known but one possibility is via regulation of TFP extension and retraction (Winther-Larsen and Koomey, 2002), which is mediated by the ATPases PilB and PilT, respectively (Whitchurch et al., 1991; Turner et al., 1993; Chiang et al., 2008).

The majority of the Tn insertions in our screen for cAMP regulators mapped to genes previously implicated in TFP regulation, including those encoding TFP structural components (pilA) or regulators of TFP biogenesis and/or function (fimL, fimV, pilB, pilG, pilI, pilJ and chpA). We recovered multiple Tn insertions in the PA1821 locus, which is immediately upstream of fimL and encodes a probable enoyl-CoA hydratase/isomerase (Whitchurch et al., 2005). We believe the phenotype of the PA1821 Tn mutants is due to a polar effect on fimL but we did not confirm this hypothesis. The non-polar pilB mutant did not recapitulate the phenotype of the pilB Tn mutant, so it was not pursued further. However, this outcome did serve to demonstrate that not all TFP biogenesis factors affect cAMP, which is consistent with the fact that we recovered only a subset of the over 40 known TFP-related genes (Mattick, 2002; Huang et al., 2003; Whitchurch et al., 2005) in our screen. In contrast to the pilB mutant, which does not assemble TFP, the pilA mutant showed a modest reduction in cAMP. At this time it is unclear whether pilin (encoded by pilA) affects cAMP through the Chp system or through an independent mechanism; the same is true for FimV, another cAMP regulator with no known chemotaxis protein homolog. The fimV gene is required for TFP production and twitching motility in P. aeruginosa and is proposed to be involved in cell wall reorganization (Semmler et al., 2000; Ahn et al., 2004), a function that may directly or indirectly affect Chp system output.

In conclusion, this study has revealed a number of new regulators of the P. aeruginosa cAMP signaling pathway, including the members of the Chp chemosensory system. We also identified the specific regulatory effect of the Chp system in controlling AC activity and future studies will determine whether the relationship between the Chp system and CyaB is direct or indirect. Ultimately, defining the signal(s) that initiate the P. aeruginosa virulence programme via the Chp system will facilitate our understanding of the machinery for host environmental sensing in this pathogenic bacterium.

Experimental procedures

Bacterial strains and growth conditions

All mutant P. aeruginosa strains used in this study were derived from the prototypic laboratory strain PAK and are referenced in Table S1. For routine passage, E. coli and P. aeruginosa were grown at 37°C in Luria Bertani (LB) medium. pMMB-based expression plasmids were maintained in P. aeruginosa with 150 µg ml⁻¹ carbenicillin (Cb), except where noted. Bacterial growth in broth culture was assessed by optical density at 600 nm (OD₆₀₀).

Native purification of the CyaB catalytic domain

To create an expression construct for the C-terminal CyaB catalytic domain, we amplified the portion of the cyaB coding region corresponding to amino acid positions 217 to 463 by PCR using the 5’ cyaB K217 and 3’ cyaB T223 oligonucleotides (Table S2) and genomic DNA from P. aeruginosa strain PAK as a template. The resulting fragment contained engineered BamHI and PstI restriction endonuclease site at the 5’ end and at the 3’ end, respectively. The cyaB_217–463 fragment was cut and inserted into the pQE30 vector (Qiagen) between the BamHI and PstI sites, creating pQE30cyaB_217–463. The plasmid was transferred to E. coli M15[pREP4] for expression of 6×His-tagged CyaB_217–463. Bacteria were grown in LB broth at 30°C until OD₆₀₀ = 0.6, followed by induction with 20 µM IPTG for an additional 20 hours at 22°C. Purification of CyaB_217–463 using Ni-NTA affinity resin (Qiagen) was carried out using the methods described for purification of the Anabaena CyaB1 catalytic domain (Kanacher et al., 2002).

Adenylate cyclase assay

AC assays were performed in a final volume of 100 µl and typically contained 50 mM buffer (pH 7.5, MOPS; pH 8.5, Tris/HCl), 2 mM [2,8-³H]cAMP (150 Bq), and [α-³²P] (25 kBq) as substrate (Salomon et al., 1974). Other assay conditions are as indicated in the figure legend. Protein concentrations were adjusted to maintain substrate utilization at less than 10%. Kinetic constants were determined over a substrate concentration range of 1 to 1000 µM.

P. aeruginosa strain construction

To create a cAMP-dependent transcriptional reporter for use in P. aeruginosa, we used a region of the E. coli lacP1 promoter extending from 175 bp upstream of the transcriptional start site to 104 bp downstream of the translational start site of lacZ; this region is predicted to contain elements necessary for lacZ transcription, including a CRP binding site (Busby and Ebright, 1999). A fragment containing the lacP1 promoter region was PCR amplified from E. coli strain HB101 genomic DNA with EclacP1 rep 5’ and EclacP1 rep 3’ oligonucleotides (Table S2) and digested with EcoRI and BamHI and cloned between the EcoRI and BamHI sites of the mini-CTX-lacZ plasmid, creating mini-CTX-PlacP1-lacZ. The resulting plasmid was integrated into the vacant ϕCTX phage chromosomal attachment site in strain PAK using previously described methods (Hoang et al., 2000). To construct isogenic AC and vfr mutants in the lacP1-lacZ reporter background, PAK∷lacP1-lacZ was subjected to allelic exchange with individual suicide vectors (pEXGmΔcyaA, pEXGmΔcyaB, or pEXGmΔvfr) using previously described methods (Wolfgang et al., 2003). To create the double AC (cyaAB) mutant, the PAKcyaA∷lacP1-lacZ strain was used in a subsequent round of allelic exchange with pEXGmΔcyaB. Construction of a version of the lacP1-lacZ reporter lacking LacI binding sites (named lacP1ΔlacI-lacZ) is described in the supplemental text.

Deletion alleles for pilA, pilB, pilG, pilH, pilI, pilJ, pilK, chpA, chpB, chpC, fimL and fimV were constructed using SOE-PCR as described (Wolfgang et al., 2003) using PAK genomic DNA as a template and the appropriate SOE oligonucleotide pairs (Table S2). Deletion alleles were cloned into pEXGmGW, pEXG2GW or pDONRX (plasmid construction described in supplemental text) by Gateway cloning and the resulting plasmids were used to introduce the various mutations into PAKcyaA∷lacP1ΔlacI-lacZ by allelic exchange as described (Wolfgang et al., 2003).

To create gene expression plasmids, the open reading frames (ORFs) of P. aeruginosa cyaA, cyaB, pilG, pilH, pilI, pilJ, pilK, chpA, chpB, chpC, fimL and fimV were amplified from PAK genomic DNA using the appropriate oligonucleotide pairs (Table S2; (Wolfgang et al., 2003) and cloned into Gateway vectors as described previously (Wolfgang et al., 2003). Briefly, PCR fragments were tailed with attB1 and attB2 sequences, cloned into the pDONR201 entry vector and then transferred to different versions of the pMMBGW destination vector. pMMBV1 and pMMBV2 have altered −35 and −10 regions resulting in reduced transcriptional expression relative to that of pMMB (S. Lory, Harvard Medical School, unpublished). The pMMBV3, pMMBV4 and pMMBV5 plasmids are versions of pMMB, pMMBV1, and pMMBV2, respectively, with an additional lac operator; this site was added by changing the sequence between the −35 and −10 sites from ATTAATCATCGGCTCG to TTGTGAGCGGATAACAA. These versions of the plasmids allow lower expression levels compared to the original series such that the maximal expression level of pMMB>pMMBV1>pMMBV2 and pMMBV3>pMMBV4>pMMBV5. All pMMB-based expression plasmids were transferred to the appropriate PAK strain by conjugation (Furste et al., 1986) followed by selection on LB agar plates containing 150 µg ml⁻¹ Cb and 25 µg ml⁻¹ irgasan (Irg). For complementation of cAMP reporter activity in the TFP/Chp mutants, the target genes were expressed in the following plasmids: pMMB (pilA, pilG and pilH), pMMBV1 (pilI, pilK, and fimV), pMMBV2 (chpC and fimL), pBMMBV4 (chpA), and pMMBV5 (chpB and pilJ).

β-galactosidase assay

Assays for β-galactosidase activity were conducted as described (Wolfgang et al., 2003) using bacteria grown to mid-logarithmic growth phase in LB broth or LB broth containing 30 µg ml⁻¹ Cb and the indicated amount of IPTG, or using bacteria grown on LB agar plates with or without 20 mM cAMP as indicated.

cAMP assay

For determination of intracellular cAMP concentration, bacteria were grown in LB broth at 37°C to OD₆₀₀ = 1.0. Bacteria were collected by centrifugation at 13, 000 × g for 2 minutes at 4°C and washed twice with 1 ml cold 0.9 M NaCl. Pellets were suspended in 100 µl 0.1N HCl and incubated on ice for 10 minutes with occasional vortexing to lyse bacteria. Lysates were centrifuged at 13, 000 × g for 5 minutes at 4°C to remove cellular material and the supernatant was assayed for cAMP using an enzyme-linked immunosorbant assay (Cayman Chemical) following the manufacturer’s protocol for sample acetylation. Duplicate bacterial pellets for protein determination were suspended in 100 µl phosphate buffered saline (PBS) and were lysed by three freeze/thaw cycles followed by centrifugation at 13,000 × g for 5 minutes to pellet unbroken cells. The protein concentration of the supernatant was determined by BCA protein assay (Pierce). Assay values for cAMP were converted to intracellular concentrations (µM) based on the estimated cellular volume per mg of protein (D'Souza-Ault et al., 1993).

Tn insertion library construction and screening

To create a P. aeruginosa Tn insertion mutant library, we performed Tn mutagenesis of PAKcyaA∷lacP1-lacZ using the plasmid pBTK30, which contains a Gm resistance-marked transposable element and a mariner C9 transposase (Goodman et al., 2004). Suspensions of E. coli strain SM10 harboring pBTK30 and the P. aeruginosa recipient (OD₆₀₀ = 40–50) were mixed at a ratio of 2:1 and pipetted onto LB plates and incubated for 2 hours at 37°C. Bacteria were then suspended in LB and dilutions were plated on LB plates containing 75 µg ml⁻¹ Gm, 25 µg ml⁻¹ Irg and 40 µg ml⁻¹ X-gal. Plates were incubated at 37°C for 24 hours. The cyaA mutant forms blue colonies on X-gal; colonies that appeared less blue (presumably due to reduced reporter activity) were selected from plates containing approximately 1000–2000 colonies per plate. Candidate clones were patched onto LB X-gal plates and grown at 22°C to confirm the color phenotype. The site of Tn insertion for each clone was determined by PCR with semi-random and Tn-specific oligonucleotides followed by sequencing of the Tn-chromosome junctions as described (Goodman et al., 2004). The Tn insertion location was identified by comparison to the genome sequence of P. aeruginosa strain PAO1 (Stover et al., 2000; Winsor et al., 2009). To test cAMP complementation of the Tn mutant phenotype, bacteria were patched on LB X-gal plates with either 50 mM cAMP or no cAMP and incubated at 22°C for 24 hours. Complementation by cAMP was scored by comparing the color of individual Tn mutants on plates with and without exogenous cAMP. The lacP1-lacZ reporter strains harboring the cyaA, cyaAB or vfr mutations were patched on each plate as controls to ensure consistent color development between plates with and without cAMP.

Pilus preparation

TFP sheared from the bacterial surface were isolated from P. aeruginosa grown on LB agar plates or, where indicated, LB agar plates containing 20 mM cAMP. For complementation assays with plasmid-harboring strains, bacteria were grown on LB plates containing 30 µg ml⁻¹ Cb and IPTG ranging from 0 to 500 µM. Bacteria were suspended in 1 ml of 0.15 M NaCl and 0.2% formaldehyde and vortexed for 1 minute. Cells were removed by centrifugation at 12,000 × g for 5 minutes. The supernatant was transferred to a microfuge tube, adjusted to 0.1 M MgCl₂, and incubated at 4°C for 3 hours. Following centrifugation at 12,000 × g for 5 minutes, the resulting pellet was suspended in SDS-PAGE sample buffer, resolved on an 18% polyacrylamide gel and visualized by GelCode Blue Stain (Pierce). The relative amount of pilin monomer demonstrated by SDS-PAGE was used as an indicator of the abundance of TFP on the bacterial surface.

Immunoblotting

Whole cell lysates for detection of pilin and CyaB were prepared from bacteria grown in LB broth until OD₆₀₀ = 1. Lysates were normalized based on total protein. Strains harboring plasmids were grown in the presence of 30 µg ml⁻¹ Cb and the indicated amount of IPTG. Pilin samples were collected by centrifugation and suspended in 100 µl of SDS-PAGE sample buffer and incubated at 95°C for 10 minutes. The lysate was diluted 1:10 and 2.5µl was resolved on 18% SDS-polyacrylamide gels and transferred to nitrocellulose. For CyaB samples, bacterial pellets were resuspended in 100 µl H₂O and treated with 1 µl DNase (10 mg ml⁻¹) for 10 minutes at 37°C. SDS-PAGE sample buffer (100 µl) was then added followed by solubilization at 42°C for 45 minutes. CyaB samples were run on 12% SDS-polyacrylamide gels and transferred to nitrocellulose. For detection of pilin, membranes were probed with PKL1 anti-pilin monoclonal antibody (Yu et al., 1994) (gift of Randall Irvin, University of Alberta; 1:30,000 dilution) followed by an HRP-conjugated goat anti-mouse Ig antibody (Promega; 1:100,000 dilution). For detection of CyaB, membranes were probed with affinity purified anti-CyaB rabbit serum directed against a peptide determinant (DKERVIQALQQAAERLRDKVIL) in the C-terminal region of CyaB (Proteintech Group, Inc.; 1:5,000 dilution) followed by a horseradish peroxidase (HRP)-conjugated goat anti-rabbit Ig (Rockland; 1:25,000 dilution). Blots were developed with enhanced chemiluminescense reagents (Millipore) and visualized by autoradiography.

Twitching motility assay

Strains were grown overnight at 37°C on LB agar plates and bacteria were stab-inoculated to the bottom of 100 mm tissue culture-treated dishes (Corning) containing 5 ml LB/1% agar or LB/1% agar plus 20 mM cAMP. For complementation assays with plasmid-harboring strains, bacteria were grown in the presence of 30 µg ml⁻¹ Cb and IPTG ranging from 0 to 500 µM. Plates were incubated for 40 hours at 37°C in a humidified chamber and the zone of subsurface bacterial spread was measured.

Statistical analysis

Calculations to assess statistical significance for values obtained in all assays were made using an unpaired t test.

Supplementary Material

Supp Figure 01. FIG. S1. Location of Tn insertions.

Tns were mapped to the chromosome of P. aeruginosa strain PAK based on sequence homology of the insertion junction to the strain PA01 genome. The coding regions of individual genes (arrows) and intragenic sequences (black lines) are indicated. The Tn insertion site of each mutant is represented by a circle. The reporter activity of a representative Tn mutant from each locus, indicated by asterisk (*), was quantified by β-galactosidase assay (Fig. 3).

Supp Figure 03. FIG.S2. Model of the components of the P. aeruginosa Chp chemosensory system.

The P. aeruginosa Chp proteins are predicted to be homologs of the E. coli Che system proteins controlling flagellar motility (Darzins and Russell, 1997; Whitchurch et al., 2004). The corresponding E. coli homolog for each Chp protein is indicated in parentheses: i) PilJ, a member of the transmembrane methyl-accepting chemotaxis protein (MCP) receptor family, ii) ChpA, a complex signal transduction protein with predicted histidine kinase activity (CheA), multiple phosphotransfer domains, a CheW-like domain, and a CheY-like receiver domain, iii) PilI and ChpC, CheW-like accessory proteins, iv) PilG and PilH, CheY-like response regulators, v) PilK, a methyltransferase, and vi) ChpB, a methylesterase. The predicted flow of information in the Chp system is from the PilJ receptor to ChpA (a process mediated by PilI and/or ChpC) and then from ChpA to PilG and PilH. The phosphorylated response regulators are predicted to impart their regulatory effect on the machinery responsible for motility; however, the mechanical basis for TFP-dependent motility in P. aeruginosa is not understood. Methylation of the MCP by PilK promotes signal transduction while ChpB is involved in signal attenuation by mediating MCP demethylation.

Supp Figure 04. FIG. S3. Creation of lacP1ΔlacI-lacZ, a version of the cAMP-dependent lacP1-lacZ transcriptional reporter that is not subject to LacI repression.

A) Schematic diagram of the lacP1ΔlacI-lacZ promoter reporter, which was derived from the lacP1-lacZ reporter construct. The lacP1ΔlacI-lacZ promoter reporter extends 78 bp upstream of transcriptional start site (+1) of lacP1, thereby eliminating the upstream LacI binding site (gray rectangle) while retaining the CRP binding site (black rectangle). The downstream LacI binding site contains a region of synthetic sequence (indicated by the gray hatched box) that substitutes the nucleotides required for LacI binding. B) β-galactosidase activity (right axis, grey bars) in the wild type strain (wt), the AC mutants (cyaA, cyaB and cyaAB), or the vfr mutant (vfr); all strains contained the lacP1ΔlacI-lacZ reporter. Values for direct measurement of cAMP by enzyme immunoassay (left axis, black bars) are from Fig. 2. For the β-galactosidase assay, the data represent two independent experiments with three replicates each and values are reported as the mean ± SEM; values for all strains are significantly different (p≤0.0004) when compared pair wise to the values for all other strains shown. C) β-galactosidase assay of the lacP1-lacZ (lacP1) and lacP1ΔlacI-lacZ (lacP1ΔlacI) reporter strains carrying either the pMMB67EH plasmid encoding LacI (+) or pMMBV1ΔlacI (described in supplemental text), in which lacI was inactivated (−). The data represent two independent experiments with three replicates each and values are reported as the mean ± SEM.

SuppFigure 02. FIG. S4. Relative quantitation of TFP from the TFP/Chp mutants.

Coomassie Blue-stained SDS-polyacrylamide gel showing dilutions of purified surface pilus fractions from the cyaA mutant (cyaA), the double AC mutant (cyaAB) and the indicated mutants in the PAKcyaA∷lacP1ΔlacI-lacZ background. Samples for pilus preparations were normalized based on bacterial number and were analyzed on separate gels: A) pilH, pilK and chpB, B) chpC and C) cyaAB and fimL. Diluted TFP samples are indicated by their corresponding dilution factor. Fold difference determinations were based on visual estimation of band intensity relative to the appropriate cyaA standards. Relative TFP quantitation was performed three times and a representative gel is shown.