Modulation of Pseudomonas aeruginosa gene expression by host microflora through interspecies communication



The change in gene expression patterns in response to host environments is a prerequisite for bacterial infection. Bacterial diseases often occur as an outcome of the complex interactions between pathogens and the host. The indigenous, usually non-pathogenic microflora is a ubiquitous constituent of the host. In order to understand the interactions between pathogens and the resident microflora and how they affect the gene expression patterns of the pathogens and contribute to bacterial diseases, the interactions between pathogenic Pseudomonas aeruginosa and avirulent oropharyngeal flora (OF) strains isolated from sputum samples of cystic fibrosis (CF) patients were investigated. Animal experiments using a rat lung infection model indicate that the presence of OF bacteria enhanced lung damage caused by P. aeruginosa. Genome-wide transcriptional analysis with a lux reporter-based promoter library demonstrated that ≈ 4% of genes in the genome responded to the presence of OF strains using an in vitro system. Characterization of a subset of the regulated genes indicates that they fall into seven functional classes, and large portions of the upregulated genes are genes important for P. aeruginosa pathogenesis. Autoinducer-2 (AI-2)-mediated quorum sensing, a proposed interspecies signalling system, accounted for some, but not all, of the gene regulation. A substantial amount of  AI-2  was  detected  directly  in  sputum  samples from CF patients and in cultures of most non-pseudomonad bacteria isolated from the sputa. Transcriptional profiling of a set of defined P. aeruginosa virulence factor promoters revealed that OF and exogenous AI-2 could upregulate overlapping subsets of these genes. These results suggest important contributions of the host microflora to P. aeruginosa infection by modulating gene expression via interspecies communications.


The occurrence of a bacterial infection relies on the pathogen's ability to regulate its gene expression in response to the host environment and is also directed by the host response. Bacterial diseases are often an outcome of the complex interactions between the pathogens and the host. In humans, bacterial pathogens not only interact with the host but also inevitably with the indigenous, generally avirulent microflora that is often a constituent of the hosts. The interactions between pathogenic microbes and the hosts have been the focus of intensive studies, and detailed molecular mechanisms have been revealed. The relationship between bacterial pathogenicity and the interactions of a pathogen with the resident microflora, however, is much less understood with the exception of the protective roles of the resident flora against invasion of pathogens (Falk et al., 1998; Guarner and Malagelada, 2003).

Pseudomonas aeruginosa is a major opportunistic pathogen in humans, capable of causing serious infections. In patients with cystic fibrosis (CF), P. aeruginosa chronic infection ultimately causes pulmonary failure resulting in premature mortality (Stover et al., 2000). The altered ion transport associated with CF patients gives rise to a more viscous pulmonary mucous layer, thereby impairing ciliary clearance, compromising the host immune response and permitting microbial colonization of the lungs (Bals et al., 1999). Chronic colonization and infection by P. aeruginosa occurs in ≈ 80% of CF patients by 18 years of age (Rajan and Saiman, 2002), and patients commonly experience periodic infections by other pathogens such as Staphylococcus aureus and Haemophilus influenzae. Burkholderia cepacia has also emerged as an important pathogen frequently associated with high mortality in CF patients (Hart and Winstanley, 2002; Soni et al., 2002). In addition to P. aeruginosa, B. cepacia and secondary pathogens, a variety of other microorganisms are also present in CF lungs (Coenye et al., 2002) including oropharyngeal flora (OF) bacteria such as viridans group streptococci and coagulase-negative staphylococci, which only colonize the upper respiratory tracts in healthy adults. To understand the pathogenesis of CF infection, it is necessary to learn the molecular interactions and cell–cell communications in this complex microbial community in the lungs of CF patients.

Quorum sensing is a cell density-dependent cell–cell signalling system that uses small extracellular molecules termed autoinducers to co-ordinate gene expression in a bacterial community. In Gram-negative bacteria, autoinducers are typically acyl-homoserine lactone (HSL) molecules and, in Gram-positive bacteria, typically oligopeptides (Miller and Bassler, 2001). P. aeruginosa possesses two acyl-homoserine lactone quorum-sensing systems, lasR/lasI (Passador et al., 1993) and rhlR/rhlI (Brint and Ohman, 1995; Latifi et al., 1995; Ochsner and Reiser, 1995) and one 2-heptyl-3-hydroxy-4-quinolone-based signalling system (denoted PQS for Pseudomonas quinolone signal) (Pesci et al., 1999). Quorum sensing in P. aeruginosa affects a broad range of genes and is central to its pathogenesis. Up to 11% of genes in the genome are controlled by quorum sensing (Schuster et al., 2003; Wagner et al., 2003). Quorum sensing regulates a range of important cellular and secreted virulence factors, and plays important roles in the development of biofilms (Fuqua et al., 2001; Miller and Bassler, 2001; Whitehead et al., 2001; Smith and Iglewski, 2003). Recent data suggest that P. aeruginosa lives in biofilms in the lungs of CF patients (Singh et al., 2000), and biofilms may be primarily anaerobic (Yoon et al., 2002), P. aeruginosa autoinducers are detectable in the lungs, suggesting that quorum sensing controls in vivo pathogenicity (Singh et al., 2000; Erickson et al., 2002; Favre-Bonte et al., 2002; Middleton et al., 2002).

The LuxS-dependent quorum-sensing system is a unique system, the signalling molecule of which, termed autoinducer-2 (AI-2), results from spontaneous cyclization of the LuxS product (4,5-dihydroxy-2,3-pentanedione) (Surette et al., 1999; Schauder et al., 2001). The luxS family of genes is spread widely among Gram-positive and Gram-negative bacteria, and AI-2 has been proposed to be an interspecies signalling molecule harbouring the information of both density and fitness of a bacterial community (Bassler, 1999; Schauder et al., 2001; Xavier and Bassler, 2003). Recent studies indicate that AI-2 is able to regulate a range of genes and cellular processes (DeLisa et al., 2001; Stevenson and Babb, 2002). It has been linked to virulence in a number of bacteria including Neisseria meningitidis (Winzer et al., 2002), Streptococcus pyogenes (Lyon et al., 2001), Vibrio cholerae (Miller et al., 2002; Zhu et al., 2002), Clostridium perfringens (Ohtani et al., 2002) and Actinobacillus actinomycetemcomitans (Fong et al., 2001). Evidence has shown that AI-2 is involved in mixed-species biofilm formation (McNab et al., 2003) and interspecies gene regulation (Fong et al., 2001). The P. aeruginosa PAO1 genome, however, lacks an identifiable luxS gene homologue (Stover et al., 2000) and therefore does not produce AI-2.

To understand the contribution of microbial interactions to P. aeruginosa pathogenesis and host diseases, animal studies have been carried out to analyse the effect of OF on P. aeruginosa virulence. The molecular interactions between pathogenic P. aeruginosa and avirulent strains isolated from sputum samples of CF patients have been investigated with an in vitro system. P. aeruginosa genes responding to OF bacteria have been identified by genome-wide gene expression profiling in the presence of OF strains. The role of AI-2-mediated quorum sensing in the bacterial interactions has also been assessed. Evidence is presented that suggests important contributions of host microflora to P. aeruginosa infection via modulation of its gene expression.


Mixed microbial species in CF lungs shown by clinical microbiology data

Present in the lungs of CF patients is a complex microbial community dominated by chronic infecting pathogens, most frequently P. aeruginosa. The constitution of this microbial community is influenced by the physiological status of the patients and antibiotic treatments. Because non-pathogenic species are often neglected in clinical laboratories (Shreve et al., 1999), many other species may be present in the CF lungs. This view is supported by recent findings that uncommon bacterial species live in the lungs of CF patients (Coenye et al., 2002).

The microbial dynamics in the lungs of CF patients over an extended period is reflected in the microbiological data gathered from sputum samples of CF patients. This is demonstrated by the data from the Adult CF Clinic in Foothills Hospital, Calgary, Alberta, Canada. Figure 1A shows the data collected on a single patient chronically infected with P. aeruginosa during the period from May 1990 to April 2001. Representatives of normal respiratory tract flora or OF strains such as viridans group Streptococcus and Staphylococcus spp. were consistently present as a significant fraction of the microflora in the sputa. Periodical colonization by H. influenzae, S. aureus and Candida albicans was also observed. The patient data presented in Fig. 1A are typical of most patients in this collection. Similar patterns were obtained in the sputa of patients that were not infected with P. aeruginosa, i.e. OF strains were consistently present at significant concentrations (data not shown). To generate a more comprehensive illustration of prevalence of these mixed communities in CF lungs, the average concentrations of Pseudomonas and normal oropharyngeal flora from cystic fibrosis patients with chronic Pseudomonas infections are presented in Fig. 1B. The average cfu ml−1 sputum was calculated for eight patients who had a minimum of 40 sputum samples during this time. The concentration of OF was regularly equivalent to or higher than that of Pseudomonas.

Figure 1.

Figure 1.

Microbiological analysis of sputum samples from CF patients.

A. Longitudinal microbiology data from a single CF patient colonized primarily by P. aeruginosa showing the transient infections by secondary pathogens and the persistent colonization of Pseudomonas and normal oropharyngeal flora. Bacterial levels are in colony-forming units (cfu) ml−1 sputum.

B. Concentration of Pseudomonas (solid bars) and normal oropharyngeal flora (open bars) from CF patients. The average cfu ml−1 sputum was calculated for eight patients (with a minimum of 40 sputum samples per patient) collected over a 10-year span. The error bars represent standard deviation in cfu.

Figure 1.

Figure 1.

Microbiological analysis of sputum samples from CF patients.

A. Longitudinal microbiology data from a single CF patient colonized primarily by P. aeruginosa showing the transient infections by secondary pathogens and the persistent colonization of Pseudomonas and normal oropharyngeal flora. Bacterial levels are in colony-forming units (cfu) ml−1 sputum.

B. Concentration of Pseudomonas (solid bars) and normal oropharyngeal flora (open bars) from CF patients. The average cfu ml−1 sputum was calculated for eight patients (with a minimum of 40 sputum samples per patient) collected over a 10-year span. The error bars represent standard deviation in cfu.

In vivo evaluation of the effect of OF bacteria on P. aeruginosa virulence

Association of P. aeruginosa and other bacteria in the lungs of CF patients suggests that interactions occur between different bacteria. To investigate the potential contributions of the associated bacteria to the pathogenicity of P. aeruginosa, the virulence of P. aeruginosa in the presence of an OF bacterium was tested in vivo using the agar bead rat lung infection model (Cash et al., 1979). Three groups of eight rats were inoculated intratracheally with P. aeruginosa PAO1 alone, PAO1 plus CF004 and CF004 alone. CF004 is a Streptococcus strain isolated from a CF sputum sample. Seven days after infection, the rats were sacrificed, and quantitative bacteriology and lung pathology were performed. The percentage of consolidation in the lung tissue was measured. Consolidated areas are tissues where accumulation of pulmonary oedema fluid and/or infiltration of inflammatory cells have occurred; therefore, more consolidation indicates more severe lung damage. As shown in Fig. 2A, significant enhancement of P. aeruginosa virulence in the presence of the OF strain was observed as indicated by increased lung damage (P < 0.0001, t-test). No significant change in P. aeruginosa loads was observed in the co-infected group compared with the group infected with P.  aeruginosa alone (P = 0.4) (Fig. 2B). Similarly, OF loads remained unchanged in the co-infection group and OF alone group (P = 0.9).

Figure 2.

Quantitative pathology and bacteriology of the rat lungs 7 days after infection using the agar beads model.
A. The rat lungs co-infected with P. aeruginosa and OF strain CF004 (group 2) showed significantly more consolidation than that infected with P. aeruginosa alone (group 1) or OF alone (group 3) (P < 0.0001, unpaired t-test). Consolidation indicates accumulation of pulmonary oedema fluid and/or infiltration of inflammatory cells. More consolidation denotes more severe lung damage.
B. P. aeruginosa (solid bars) and CF004 (grey bars) bacterial loads (cfu) in the lungs of the three groups.

Screening of P. aeruginosa genes modulated by OF bacteria

In order to understand the contribution of OF bacteria to P. aeruginosa virulence and to investigate the molecular interactions between P. aeruginosa and OF bacteria, an in vitro method was developed to screen for P. aeruginosa genes that are modulated in the presence of OF bacteria. The system uses a P. aeruginosa random promoter library constructed with the luxCDABE reporter carried on a low-copy-number plasmid pMS402. The activity of any individual promoter is thus represented by the amount of light generated by the clone containing the construct. By measuring luminescence in a multilabel plate counter, the P. aeruginosa library can be screened temporally under different conditions to identify differentially regulated promoters; hence, differentially regulated genes transcribed from these promoters.

Using this method, we screened 3456 P. aeruginosa (ATCC27853) clones for differentially expressed promoters in the presence of two Gram-positive OF bacteria, Streptococcus strain CF004 and Staphylococcus strain CF018. Both strains were isolated from a single sputum sample from a CF patient. The luminescence from each P. aeruginosa promoter clone, in both a monoculture of P. aeruginosa and a co-culture with CF004 or CF018, was measured. The luminescence data of the P. aeruginosa promoter clones at the 7.5 h time point in the initial screen are shown in Fig. 3A (CF004) and B (CF018). The initial screening identified 280 promoters potentially regulated by CF004 and 252 by CF018. Rescreening of these potential positives by temporally resolved gene expression profiling combined with growth evaluation confirmed 214 promoters affected by CF004 and 171 by CF018, representing ≈ 6% and 5% of the P. aeruginosa clones respectively (≈ 4% of P. aeruginosa operons assuming that the library is a random library). Among these promoters, 152 were common to both strains. The regulated P. aeruginosa promoters can be clustered into three classes: those regulated by both strains (class I), CF004 only (class II) or CF018 only (class III) (Fig. 3C). The different regulation of P. aeruginosa promoters by these two strains suggests that there are common as well as unique signals or pathways in the interactions between P. aeruginosa and these Gram-positive bacteria. Screening a subset of regulated promoter clones against other OF isolates also revealed varied levels of gene regulation by different isolates (data not shown). This may point to a more prominent role for specific OF strains in affecting P. aeruginosa gene expression, but may also simply be a reflection of the co-culture conditions.

Figure 3.

Scatter plot of the initial screen data at the 7.5 h time point. P. aeruginosa promoter clones were screened for differential expression in response to the presence of CF004 (A) or CF018 (B). Each point in the plots represents the activity of one individual promoter in the library. The promoters unaffected by the co-culture conditions (blue) fall along the diagonal (i.e. their expression is the same in both the monoculture and co-culture). Points in red distributed above and below the diagonal indicate promoters with a minimum 2.5-fold downregulation and upregulation, respectively, in the presence of CF004 or CF018.
C. Representatives of regulated promoters identified in initial screening are clustered by expression profiles. Red to green colour gradient indicates the levels of expression from high to low; data are normalized by the maximal value of each promoter. Three classes of regulated promoters are marked: those regulated by both strains (class I), CF004 only (class II) or CF018 only (class III).

Characterization of the modulated P. aeruginosa genes

A subset of the affected promoters was sequenced and compared with the annotated P. aeruginosa PAO1 genome (Stover et al., 2000) and GenBank data ( to identify the genes regulated. Table 1 lists the 48 operons with known or putative gene functions that are expressed from the regulated promoters. These genes can be classified into seven groups by function. An additional 33 characterized promoters are associated with genes encoding proteins with unknown function (data not shown). Seven promoters that are located at the 5′ end of the annotated genes but orientated in the opposite direction were also identified and designated orphan promoters (data not shown). These orphan promoters could potentially transcribe small open reading frames (ORFs) or small regulatory RNA molecules such as antisense RNA, thus having the potential of regulating gene expression on the other strand. In addition to the three with putative functions included in Table 1, another 14 promoters share no sequence homology with the PAO1 genome, neither do they share homology with any other bacterial sequences in the GenBank. They represent unique sequences in this clinical P. aeruginosa isolate. The identification of these unique clones probably reflects the plasticity of the Pseudomonas genome and suggests that a fraction of strain-specific genes may be associated with microbial interactions.

Table 1. .P. aeruginosa genes with known or putative functions expressed by OF-regulated promoters.
GroupaGene or
PA no.
DescriptionMaximum fold
  • a

    . Although multiple genes are often controlled by one promoter, only the first gene in each operon is listed. Promoters not listed include 33 promoters expressing genes with unknown functions, seven orphan promoters and 14 promoters not found in PAO or other sequenced bacteria.

  • b

    . Approximate maximum fold regulation in the presence of CF004 or CF018 at the 48 time points. (+) upregulation; (–) downregulation.

  • c

    . Promoters not present in the PAO1 genome.

  • d

    . Promoters represented more than once by non-identical clones or two identical clones (argF ).

  • e

    . Promoters regulated by only one OF strain.

(1) Protein relevant
to virulence
xcpP Secretion protein XcpP, type II protein
secretion apparatus
 +39 Akrim et al. (1993)
PA1882Probable efflux transporter, SugE family,
67% similar to SugE protein (E. coli )
shown to have cationic drug export
 +36 Stover et al. (2000);
Chung and Saier (2002)
cbpD Chitin-binding protein, possibly involved in
biofilm formation or pathogenicity
 +30 Folders et al. (2000)
PA4381Probable two-component response
regulator, 91% similar to ColR (P.
) which is crucial for
 +11 Stover et al. (2000)
lasB Elastase, one of the major virulence
  +7 Bever and Iglewski (1988)
orf5 c P. aeruginosa pathogenicity island PAGI-1
gene, a homologue of RpoN-dependent
transcriptional activators, similar to PrpR
(E. coli )
  +5 Liang et al. (2001)
wbpT c Putative α-d-alactosyltransferases,
involved in O-antigen biosynthesis of P.
serotype O6
  +4 Belanger et al. (1999)
ndk Nucleoside diphosphate kinase, producing
GTP from ATP, involved in alginate
synthesis via GDP-mannose production
  +4 Sundin et al. (1996)
phnB d Anthranilate synthase component II,
involved in phenazine biosynthesis
  +2 Mavrodi et al. (2001)
PA1282eProbable MFS transporter, similar to
multidrug efflux protein in S. aureus and
M. smegmatis
  +2 Stover et al. (2000)
phzM Phenazine-specific methyltransferase −29 Essar et al. (1990);
Mavrodi et al. (2001)
(2) Transporter or
PA4500Probable binding protein component of
ABC transporter
  +9 Stover et al. (2000)
PA4037Probable ATP-binding component of ABC
  +7 Stover et al. (2000)
PA3038Probable porin  +7 Stover et al. (2000)
PA2114Probable MFS transporter  −5 Stover et al. (2000)
lppL Lipopeptide LppL precursor−460 Jann et al. (1990)
(3) MetabolismPA1551Probable ferridoxin +19 Stover et al. (2000)
PA0182eProbable short-chain dehydrogenase +18 Stover et al. (2000)
PA5400Probable electron transfer flavoprotein
alpha subunit
 +14 Stover et al. (2000)
purD Phosphoribosylglycineamide synthetase,
purine metabolism
  +7 Stover et al. (2000)
dnaE DNA polymerase III, alpha chain  +5 Stover et al. (2000)
lysS Lysyl-tRNA synthetase  +5 Stover et al. (2000)
PA2953eElectron transfer flavoprotein-ubiquinone
  +5 Stover et al. (2000)
PA1919Probable radical-activating enzyme, 62%
similar to NrdG (Ralstonia eutroph a)
2000  +3Stover et al. (2000)
argF d Ornithine carbamoyltransferase, arginine
  +3 Stover et al. (2000)
PA5127Probable rRNA methylase  +2 Stover et al. (2000)
proS Prolyl-tRNA synthetase  +2 Stover et al. (2000)
thiG Probable thiamine biosynthesis protein,
thiazole moiety
  +2 Stover et al. (2000)
metG Methionyl-tRNA synthetase−152 Stover et al. (2000)
(4) Iron uptake
or storage
PA0112Membrane protein, 49% similar to a region
of putative haem O oxygenase in Aquifex
 +11 Stover et al. (2000)
PA4515Conserved hypothetical protein, iron
uptake factor PiuC
  +8 Stover et al. (2000)
hemE Uroporphyrinogen decarboxylase, haem
  +2 Stover et al. (2000)
(5) Transcriptional
lrp Global transcriptional regulator, leucine-
responsive regulatory protein
  +6 Stover et al. (2000)
PA1136Probable transcriptional regulator, 42%
similar to transcriptional activator LasR
  +4 Stover et al. (2000)
PA2489Probable transcriptional regulator  +2 Stover et al. (2000)
PA2766eProbable transcriptional regulator, TetR
  +2 Stover et al. (2000)
PA3898Probable AraC family transcriptional
  −9 Stover et al. (2000)
PA2879Probable transcriptional regulator, LysR
 −10 Stover et al. (2000)
PA3135Probable transcriptional regulator, LysR
 −30 Stover et al. (2000)
PA3604Probable two-component response
regulator, LuxR family
 −76 Stover et al. (2000)
PA1226Probable transcriptional regulator, TetR
−489 Stover et al. (2000)
(6) Secreted
PA3309Conserved hypothetical protein with
universal stress protein motif
 +67 Stover et al. (2000)
pra Protein activator, extracellular protein  +8 Hardegger et al. (1994)
(7) Other functional
merA c Mercuric ion reductase,  +2 Brown et al. (1983)
rpsT e 30S ribosomal protein S20 −16 Stover et al. (2000)
rpsQ e 30S ribosomal protein S17 −23 Stover et al. (2000)
pctA Chemotactic transducer −74 Kuroda et al. (1995)
PA2561Probable chemotaxis transducer −86 Stover et al. (2000)

Among the genes affected by OF are a relatively large number of well-characterized P. aeruginosa virulence factor genes or genes relevant to P. aeruginosa pathogenicity (Table 1, the first group). lasB, which encodes elastase, a major virulence factor that contributes to inflammatory damage of the respiratory epithelia and interferes with host immunological defences (Bever and Iglewski, 1988), was upregulated to a maximum of sevenfold during a 24 h time course. The most activated gene in this group was xcpP, which encodes a type II protein secretion apparatus responsible for secreting and chaperoning a number of virulence factors including elastase and exotoxin A (Akrim et al., 1993). Two probable multidrug efflux genes, PA1282 and PA1882, were also upregulated. PA1282 encodes a major facilitator superfamily (MFS) transporter protein similar to multidrug efflux proteins QacA and QacB in S. aureus and LfrA in Mycobacterium smegmatis (Stover et al., 2000). PA1882 encodes a small protein belonging to the SugE subfamily of the small multidrug resistance family (SMR). SugE has recently been shown to have cationic drug export function in Escherichia coli (Chung and Saier, 2002), and the same function has been proposed for the SMR family. Efflux pumps are not only important for bacterial antibiotic resistance but can also contribute to a pathogen's invasiveness by assisting export of virulence determinants (Hirakata et al., 2002). CbpD, encoding a chitin-binding protein, has only been found in clinical isolates of P. aeruginosa but not in soil pseudomonads. It has been suggested that CbpD may have a role as an adhesin mediating P. aeruginosa colonization of eukaryotic cells (Folders et al., 2000). ndk encodes a nucleoside diphosphate kinase involved in alginate production (Sundin et al., 1996), a process critical for P. aeruginosa pathogenicity. PA4381 encodes a probable two-component response regulator similar to ColR in Pseudomonas fluorescens, which is crucial for the bacterium's ability to colonize plants, although the mechanism by which ColR functions remains unclear (Dekkers et al., 1998). Two genes, wbpT and orf5, are present in P. aeruginosa ATCC27853 but not in PAO1. wbpT encodes a putative α-d-galactosyltransferase involved in O-antigen biosynthesis of P. aeruginosa serotype O6 (Belanger et al., 1999). orf5 is a P. aeruginosa pathogenicity island PAGI-1 gene, encoding a homologue of RpoN-dependent transcriptional activator (Liang et al., 2001). Identification of these upregulated genes that are relevant to pathogenicity might explain the observation in the animal studies that P. aeruginosa virulence was enhanced in the presence of OF.

The only negatively affected gene in this group was phzM. It encodes a phenazine-specific methyltransferase involved in the conversion of phenazine-1-carboxylic acid (PCA) to pyocyanin (Mavrodi et al., 2001). As phenazine compounds are not only virulence factors but also involved in bacterial competition (i.e. inhibiting or killing other bacteria in the vicinity), the downregulation of phzM in response to the presence of OF bacteria seems to contradict the antagonistic competition function of pyocyanin. However, as PCA itself is an active phenazine compound, downregulation of phzM may simply indicate a ratio change in the phenazine compounds. In contrast, phnB, a related promoter identified in our assay, was upregulated about twofold. A previous study showed that phnA and phnB encode an anthranilate synthase influencing pyocyanin production (Essar et al., 1990), but the precise role of the synthase in pyocyanin or PCA biosynthesis is not clear (Mavrodi et al., 2001).

The second group of genes affected by OF encodes five membrane proteins. lppL encodes a lipopeptide, and the other four encode probable transporters. The 13 genes in the third group are largely involved in the cellular processes of protein and DNA, and electron transfer. Three iron utilization-related genes were identified and classed in the fourth group.

Ten of the regulated genes (including the PA4391 in the first group) encode known or putative transcriptional regulators. This number represents 9.7% of the 103 characterized promoters, close to the ratio of 9.4% in the PAO1 genome for genes encoding either transcription regulators or two-component regulatory proteins (Stover et al., 2000). Except for the lrp gene encoding a global transcriptional regulator (Stover et al., 2000), the functions or targets of the remaining regulators in this group are unknown.

AI-2 is produced in sputum cultures and by most bacterial isolates from CF sputum

One possible contributor in the interactions between OF bacteria and P. aeruginosa is the AI-2-mediated signalling that is proposed as an interspecies communication pathway (Xavier and Bassler, 2003). In an effort to probe the role of AI-2 in the microbial community in the lungs of CF patients, the first question addressed was whether AI-2 was produced in this community. Thirty independent sputum collections were used to inoculate brain–heart infusion (BHI) and THY media. These cultures were grown for 16 h, and AI-2 activity in cell-free supernatants was measured using the Vibrio harveyi reporter system. All the samples tested exhibited AI-2 activity at different levels. Figure 4A shows the data from 10 independent patient samples.

Figure 4.

AI-2 detection.
A. AI-2 activities in the culture supernatants grown from sputum samples from 10 CF patients. AI-2 activity is represented as fold induction of luminescence above the medium control using the V. harveyi reporter system. The cultures were grown for 16 h in BHI, and culture supernatants were added at 5% in the assays. S. typhimurium LT2 was used as a positive control.
B. AI-2 production of the bacterial strains isolated from a single sputum sample. Nos 1, 7 and 9, Staphylococcus spp.; nos 2 and 12, Klebsiella spp.; nos 4, 5 and 11, Streptococcus spp.; no. 8, coagulase-negative Staphylococcus spp.; nos 3, 6 and 10, Pseudomonas spp. 5% cell-free overnight culture supernatants were added to the AI-2 assays.

To verify these results, AI-2 production by individual strains was tested. A sputum sample from a single patient was diluted and isolated on different solid media. Representative isolates were grown in BHI, and AI-2 activity was measured. As illustrated in Fig. 4B, the results indicate that the AI-2 signalling molecule was produced by most isolates, confirming that the bacteria present in the lungs of CF patients can produce AI-2. P. aeruginosa isolates were the only strains negative for AI-2 production. Consistent with this observation, the P. aeruginosa PAO1 genome does not contain luxS, a gene that is required for AI-2 production. Despite the possible difference in genetic composition, morphologically varied P. aeruginosa isolates from several sputum samples were also tested, and none produced detectable AI-2.

AI-2 activity can be detected directly in sputa and in BAL of rats co-infected with P. aeruginosa and a sputum isolate

To assess whether AI-2 activity is present in the lungs of CF patients, AI-2 was measured directly in the sputum samples. The insoluble materials were precipitated by centrifugation, and the macromolecule substances were reduced by methanol precipitation. As shown in Fig. 5, a substantial amount of AI-2 activity could be detected directly in the cleared sputum samples of CF patients, suggesting that AI-2 is produced in the lungs of CF patients. It is noteworthy that the AI-2 activity was detectable in the sample from the patient who was not colonized by P. aeruginosa (sample CFY in Fig. 5). The amount of AI-2 in sputum samples was comparable to that in the culture supernatant of Salmonella typhimurium LT2. In comparison, acyl-homoserine lactone autoinducers have also been detected in sputum samples from CF patients, but the concentrations are relatively lower compared with culture supernatants, i.e. 1–22 nM for 3-oxo-C12-HSL and 1–5 nM for C4-HSL (Erickson et al., 2002).

Figure 5.

AI-2 activity in sputum samples and in BAL of rats co-infected with P. aeruginosa and sputum isolate CFX1. AI-2 activity in four independent sputum samples was detected directly using sputum extracts at 5%. Sample CFY was from a CF patient who was not colonized by P. aeruginosa at the time of sample collection. Rat BAL 1 and BAL 2 represent BAL samples from rats infected with P. aeruginosa alone and those co-infected with P. aeruginosa and CF004 respectively.

One limitation of the above experiments is that sputum samples could potentially be contaminated with upper respiratory tract material during excretion. To address this issue and to determine whether AI-2 could accumulate in the lung, AI-2 activity was measured in rat bronchoalveolar lavage fluids. As exhibited in Fig. 5, a significant amount of AI-2 was detected in the bronchoalveolar lavage fluid of the animals co-infected with P. aeruginosa and Gram-positive isolate CF004, but not in those of the animals infected with P. aeruginosa alone. These data indicate that AI-2 can be produced and accumulate to a substantial level in the lung, supporting the involvement of AI-2 in the interactions in the microbial community. The high concentration of AI-2 detected in sputum samples would suggest that this could not all be accounted for by contamination of the sample during expelling. This same argument applies to the presence of normal oropharyngeal flora in sputum samples.

Pseudomonas aeruginosa virulence factor genes are modulated by OF bacteria and AI-2

To investigate the effect of AI-2 on P. aeruginosa gene expression and pathogenicity, a group of 21 well-characterized P. aeruginosa genes that are related to pathogenicity (Table 2) was tested for differential expression in the presence of AI-2. Their response to OF strain CF004 was tested in parallel experiments. The promoter regions of these genes or their accommodating operons were polymerase chain reaction (PCR) amplified, cloned in pMS402 and transferred back into P. aeruginosa. The temporal expression of these genes was tested over a period of 24 h in the presence of exogenous AI-2 or CF004. As listed in Table 2, six genes were regulated by both AI-2 and the OF strain, and three were regulated only by the OF strain. These upregulated virulence factors include rhamnosyltransferase gene rhlA involved in rhamnolipid synthesis (Ochsner et al., 1994), elastase gene lasB (Bever and Iglewski, 1988), exotoxin genes exoT (Yahr et al., 1996), exoS (Kulich et al., 1994) and exoY (Yahr et al., 1998) and the phenazine synthesis genes phzA1 and phzA2 (Mavrodi et al., 2001). The stationary phase sigma factor rpoS (Tanaka and Takahashi, 1994) and flagellin type B gene fliC (Brimer and Montie, 1998; Feldman et al., 1998), both of which are relevant to P. aeruginosa virulence, were also upregulated. The results indicate that AI-2 was able to modulate P. aeruginosa gene expression patterns and pathogenicity. AI-2-mediated signalling represents one of the pathways that OF bacteria use to interact with P. aeruginosa.

Table 2. . Regulation of P. aeruginosa virulence-associated genes by OF strain CF004 and AI-2.
GenePA no.Function descriptionFold inductiona
  • a

    . Approximate maximum fold induction in the presence of CF004 and AI-2 respectively.

Regulated virulence factors
 phzA2PA1899Phenazine synthesis cluster 2, first gene7.14.3
 phzA1PA4210Phenazine synthesis cluster 1, first gene3.03.2
 rhlAPA3479Rhaminolipid (rhamnosyltransferase chain A)2.02.2
 exoTPA0044Exoenzyme T (99% similar to ADP-ribosyltransferase)1.91.9
 fliCPA1092Flagellin type B1.83.6
 exoSPA3841Exoenzyme S (ADP-ribosyltransferase)2.3n
 exoYPA2191Belongs to the exoenzyme S regulon, adenylate cyclase1.9n
 rpoSPA3622Stationary phase sigma factor1.8n
Unaffected virulence factors
 lasAPA1871Protease (staphylolytic protease preproenzyme LasA)
 aprAPA1249Alkaline protease (alkaline metalloproteinase precursor)
 adhPA2407Putative adhesion protein
 hemPA0041Putative haemagglutinin
 pilGPA0408Type IV fimbrial (part of the pilGHIJKL gene cluster)
 algDPA3540Alginate synthesis (GDP-mannose 6-dehydrogenase AlgD)
 plcHPA0844Haemolytic phospholipase C (haemolysin) precursor
 toxAPA1148Exotoxin A
 pvcAPA2254Pyoverdine chromophore biosynthesis protein PvcA, first of four ORF cluster
 rnrPA4937Exoribonuclease RNase R (virulence protein VacB)
 migAPA0705Probable glycosyl transferase (mucin-inducible gene)
 oprHPA1178Outer membrane protein H1 precursor and PhoP/Q operon


The usually non-pathogenic, indigenous microflora or ‘normal flora’ is a ubiquitous constituent of the animal or human host. In healthy individuals, the upper respiratory tracts are colonized by a variety of microorganisms comprising the normal flora or OF, and the lower tracts, although constantly inoculated, normally remain sterile. In contrast, a diverse OF is present along with the opportunistic pathogen P. aeruginosa in the lower respiratory tracts of adult CF patients. Analysis of clinical microbiology data from CF patients over a long period revealed that S. aureus and H. influenzae are periodically present in the lungs of CF patients. The normally avirulent oropharyngeal flora strains, such as coagulase-negative Staphylococcus spp. and viridans Streptococcus spp., are also present in the sputum samples in substantial concentrations (106−108 cfu ml−1). In typical clinical microbiology practice, the types of bacteria monitored are limited. Bacteria characterized in this study inevitably under-represent the actual bacterial diversity in the CF lungs, consistent with the recent findings that a large number of ‘unusual’ species are present in the lungs of CF patients (Coenye et al., 2002).

The animal experiments using the agar beads infection model showed that P. aeruginosa pathogenicity was significantly enhanced by the presence of OF, whereas the OF strain itself showed little virulence. The increased lung damage was not the result of changes in the levels of P. aeruginosa in the co-infections as there was no significant increase in bacterial load. This is reminiscent of some clinical situations in which no significant change in bacterial load is observed during periods of exacerbation and recovery (Jaffar-Bandjee et al., 1995; Wolter et al., 1999). Although resident flora can play an important protective role against the invasion of pathogens (Tancrede, 1992; Falk et al., 1998), in our experimental conditions, OF bacteria seem to contribute to disease progression.

Broad response in P. aeruginosa seems to be induced by the presence of OF. Although competition for nutrients was expected to be one of the factors, the observed response appears to be a result of more complex interactions between P. aeruginosa and OF bacteria. Only a relatively small number of genes involved in metabolism were affected (Table 1). As indicated by cfu counting, in the co-culture condition, the OF strains grew poorly and were unlikely to be able to compete effectively with the P. aeruginosa for nutritional resources. We expected competition for iron to be a factor in a mixed microbial community; however, the limited number of affected iron utilization genes suggests that iron competition, if any in the co-cultures, did not result in significant changes in P. aeruginosa gene expression. Consistent with this observation, a parallel screen for iron-regulated genes in P. aeruginosa identified a predominately non-overlapping set of promoters (data not shown). The results from the analysis of defined virulence factor genes also indicate that the pvcA (Stintzi et al., 1999) and toxA (Gray et al., 1984) genes, which are strongly affected by iron, were not affected by the presence of OF strains.

The upregulation of a significant number of virulence genes in P. aeruginosa by OF strains is intriguing and probably explains the increased lung damage observed in the presence of the OF strain in the animal experiments. The upregulation of a pathogen's virulence factors by microflora underscores the importance of bacterial interactions in pathogenicity. The interactions between bacteria and their host are believed to determine the evolution of many of the virulence factors that pathogens possess (Cotter and DiRita, 2000). The large number of P. aeruginosa virulence-related genes affected by other bacteria in this study suggests that the interactions among microbes in microflora may also contribute to the evolution of these virulence factors. Bacterial pathogens, especially those that inhabit both environmental niches as well as animal hosts, may rely on these interactions to maintain their virulence when out of the hosts. On the other hand, these virulence factors may play a role in the development of microbial communities.

The regulation of P. aeruginosa gene expression and virulence involves its own repertoire of cell–cell signalling molecules (Fuqua et al., 2001; Miller and Bassler, 2001; Whitehead et al., 2001; Smith and Iglewski, 2003). The pathways through which the OF strains affect P. aeruginosa gene expression are expected to be multifactorial and complex. One of the signals seems to be autoinducer-2 (AI-2) produced by non-pseudomonad strains. It seems that P. aeruginosa is modulating its behaviour by monitoring the environmental conditions and by eavesdropping on the other bacteria via AI-2 and probably other signals. AI-2 in the sputum samples could reach a substantial amount, which is readily detectable by the V. harveyi assay. The samples were processed to remove insoluble and macromolecule substances, but were not concentrated. The levels of AI-2 activity measured are unavoidably lower than actual levels because of the lost activity during sample preparation. A caveat to the use of sputum samples is that they may potentially be contaminated with upper respiratory tract material during production. Given the high concentration of OF bacteria and AI-2 present, it seems unlikely that this activity is exclusively caused by contamination. This is further supported by the presence of AI-2 in bronchoalveolar lavage fluid from chronically infected rat lungs.

The regulation of a number of virulence factor genes by AI-2 indicates that this signal produced by OF strains contributed, at least in part, to the observed modulation of P. aeruginosa gene expression and changes in its pathogenicity. The AI-2-regulated P. aeruginosa virulence factor genes partially overlap those modulated by OF strains, suggesting that, in the co-culture experiments, AI-2 was one of the signals but not the only signal produced by the OF that regulates P. aeruginosa gene expression. In the lungs of CF patients, this AI-2-mediated effect could also originate from secondary pathogens, suggesting that the virulence of secondary pathogens could also be delivered through enhancing the virulence of the primary pathogen, P. aeruginosa. It is also possible that P. aeruginosa could influence virulence factor genes of secondary pathogens or those potentially present in the OF strains.

Our evidence indicates that OF bacteria that are not normally thought of as serious problems in CF may contribute to lung pathology by influencing the gene expression of P. aeruginosa. Part of the influence of OF on P. aeruginosa may result from interspecies communication including AI-2-mediated signalling. The clinical efficacy of a course of antibiotic therapy does not always correlate with proven inhibition of replication of P. aeruginosa. P. aeruginosa bacterial density in sputum can remain high despite clinically effective treatment (Jaffar-Bandjee et al., 1995). The data presented here suggest that one contributing mode of action of these drugs may be through effects on OF strains. Similar effects on OF strains may also explain part of the efficacy of macrolide antibiotics such as azithromycin that have little antipseudomonal activity but have been proven to be of clinical benefit in CF (Peche’re, 2001; Saiman, 2002; Schöni, 2003). Protocols that take into consideration the potential significance of OF–P. aeruginosa interactions in CF may lead to more efficacious therapeutic interventions and improved clinical outcome.

Experimental procedures

Bacterial strains, plasmids and culture conditions

Pseudomonas aeruginosa PAO1 (Holloway, 1955) and the clinical isolate ATCC27853 that was used to construct the promoter library were grown on Luria–Bertani (LB) plates or LB broth at 37°C. Salmonella typhimurium LT2 (McClelland et al., 2001) was also grown in LB. Vibrio harveyi BB170 (Bassler et al., 1994) was grown on LB plates or in autoinducer bioassay (AB) medium (Greenberg et al., 1979) at 30°C. The culture conditions for bacteria in different assays are described separately.

The lux-based promoter reporter plasmid pMS402 was constructed by joining the PacI fragment of pCS26Pac (Bjarnason et al., 2003) with the EcoRI–BamHI fragment of pUCP28T (West et al., 1994), which was blunt-ended by filling in recessed termini and was attached with a blunt-end PacI linker.

Bacteria in sputum samples from CF patients were isolated by first diluting the sputum sample in THY (Todd–Hewitt broth supplemented with 0.5% yeast extract) and then grown on agar plates including blood agar, BHI agar, Todd–Hewitt agar and Pseudomonas isolation agar. After growth in a CO2 incubator or anaerobic jars, colonies were counted, and cell and colony morphology was examined. Distinct isolates were grown on blood agar plates and collected. Microbial species were characterized using conventional microbiological and biochemical procedures (Murray, 1995). The OF strains, CF004 and CF018, were isolated from a single sputum sample from a CF patient and classified as viridans group Streptococcus spp. and Staphylococcus spp., respectively, by microbiological and biochemical tests.

Pseudomonas aeruginosa was grown at 37°C on LB plates or in M9 minimal medium supplemented with casamino acid (0.1%), glucose (0.5%) and trimethoprim at 200 µg ml−1 where appropriate. CF004 and CF018 were grown at 37°C in BHI medium or on blood agar plates. Pseudomonas isolation agar (PIA) and BactoTM Columbia CNA agar were used as selection media for P. aeruginosa and Gram-positive strains respectively. Both monoculture and co-culture assays were set up in multiwell plates (384-well or 96-well) filled with the supplemented M9 medium (150 µl per well). The monocultures were inoculated with P. aeruginosa (1:500 dilution), and co-cultures were inoculated with P. aeruginosa (1:500 dilution) together with CF004 or CF018 (1:200 dilution). Under these co-culture conditions, P. aeruginosa grew normally and reached 2 × 109 cfu ml−1 during the 24 h of growth. The OF bacteria retained a stable but low growth in this condition and had 1 × 104−2 × 105 cfu ml−1 after 24 h growth. These co-culture conditions were selected to minimize competition for nutrients in order not to select for biosynthetic genes and presumably favour the detection of genes regulated by cell–cell signalling.

Animal infection test

Animal studies were carried out by Pathoprobe R and D using an established rat lung infection model (Cash et al., 1979). Three groups of male Sprague–Dawley rats (eight in each group) weighing between 180 g and 200 g were inoculated intratracheally with 0.05 ml of bacterium suspensions of 106 cfu ml−1 embedded in agar beads (P. aeruginosa alone, OF strain alone and a mixture of equal amounts of both). The rats were sacrificed for quantitative bacteriology and histopathology at the end of 7 days after infection. The pathogenicity of P. aeruginosa was assessed by the lung damage indicated as percentage of consolidation.

Random promoter library construction

The promoter library was constructed and validated according to procedures developed in our laboratory (Bjarnason et al., 2003). Briefly, Sau3A partially digested P. aeruginosa chromosomal DNA fragments with the sizes of 1–2 kb were selected by sucrose gradient centrifugation and ligated into the BamHI restriction site upstream of the promoterless luxCDABE operon in the low-copy-number plasmid pMS402. The ligated DNA was transformed into P. aeruginosa by electroporation. The transformants were picked into 384-well plates, and promoter-containing clones were selected by measuring luminescence production in LB and M9 media at several time points during growth. A library of 3456 promoter clones was constructed and used for screening CF004- and CF018-regulated promoters. Using the 6.26 Mbp P. aeruginosa PAO1 genome size (Stover et al., 2000) as reference and the estimation that every 2–2.5 kb of DNA may contain one transcription unit, ≈ 2500–3130 promoter regions are predicted in the P. aeruginosa genome.

Promoter screening and clustering

In the initial screening, the luminescence of each promoter clone in a monoculture of P. aeruginosa or co-culture with CF004 or CF018 was measured in a Wallac Victor2 model 1450 multilabel counter (Perkin-Elmer Life Sciences) at 1.5 h intervals for 10.5 h and then at 24 h. The luminescence was measured as counts per second (c.p.s.). After rearraying the potential regulated clones in new multiwell plates, rescreening was carried out in a similar manner. However, the luminescence of each well was measured every 30 min for 24 h, and bacterial growth in the wells was monitored at the same time by measuring optical density (OD600). The growth of the bacteria was also confirmed by plating selected cultures on selective media and cfu counting.

The promoter clustering was performed according to the similarity in their expression profiles using the cluster program and visualized using treeview (Eisen et al., 1998).

Sequence analysis of the modulated promoters

A subset of OF-regulated promoters was PCR amplified using  pZE.05  (CCAGCTGGCAATTCCGA)  and  pZE.06 (AATCATCACTTTCGGGAA) primers, which flank the BamHI site of pMS402. The PCR products were sequenced using an automatic sequencer, and DNA sequences were analysed using the blast program from the National Center for Biotechnology Information ( and vector NTI 7.1 (Informax) programs.

Measurement of AI-2 activity

AI-2 activity was measured using the V. harveyi reporter assay as described previously (Surette and Bassler, 1998). Briefly, an overnight culture of V. harveyi was diluted 1:5000 in AB medium (Greenberg et al., 1979) on a 96-well microtitre plate, and 10 µl of samples (or diluted samples) to be tested was added to 90 µl of the diluted culture. The plates were incubated at 30°C with shaking at 200 r.p.m. The luminescence values of individual cultures (wells) were measured every hour for a total of 12 h in a Victor2 Multilabel counter, and AI-2 activity was reported as fold induction of luminescence by the reporter strain above the medium control.

AI-2 activity in sputa was measured after the sputum samples were processed by the following procedures. The sputa were passed through an 18-gauge needle 20 times and a 21-gauge needle five times. Then, the samples were cleared by centrifugation at 10 000 g for 15 min. Macromolecules in the cleared fluid were reduced via precipitation by an equal volume of methanol and filtration through a 0.22-µm pore size filter. Methanol in the extracts was removed by evaporation in a SpeedVac, and samples were adjusted to the original volume by adding sterile water. AI-2 in culture supernatant or bronchoalveolar lavage (BAL) fluid was measured directly after the samples were subjected to centrifugation at 15 000 g for 10 min and sterilization by passing through a 0.22-µm pore size filter.

Construction of a P. aeruginosa virulence factor gene set

The promoter regions of selected P. aeruginosa virulence factors were amplified by PCR and cloned at the XhoI–BamHI sites of pMS402. The expression of the virulence factors in a monoculture of P. aeruginosa alone or co-culture with CF004 was measured as luminescence production taken at 30 min intervals for 24 h using a Wallac Victor2 Multilabel counter.

In vitro synthesis of AI-2

AI-2 was synthesized as described previously (Schauder et al., 2001). The reaction was carried out for 1 h at 37°C in 10 mM sodium phosphate buffer (pH 7.5) containing 1 mM substrate S-adenosylhomocysteine (SAH) and 1 mg ml−1 purified E. coli LuxS– and Pfx–GST fusion proteins. After incubation, the reaction mixture was filtered through Biomax-5 Centricon Plus-20 centrifugation filters (Millipore) to remove proteins. The amount of AI-2 in the preparation was estimated by fold induction of luminescence using the V. harveyi system described above.


We thank our colleagues and members of the Surette laboratory for helpful discussions and critical reading of the manuscript. M.G.S. is supported by a Canada Research Chair in Microbial Gene Expression and a Senior Scholar Award from the Alberta Heritage Foundation for Medical Research (AHFMR). K.D. is a recipient of an AHFMR postdoctoral fellowship. This research was supported by the Canadian Institutes of Health Research and Quorex Pharmaceuticals.