SGM Journals
MICROBIAL PATHOGENICITY

Phenotypic diversification in vivo: Pseudomonas aeruginosa gacS− strains generate small colony variants in vivo that are distinct from in vitro variants

Lisa K. Nelson, M. Mark Stanton, Robyn E. A. Elphinstone, Janessa Helwerda, Raymond J. Turner, Howard Ceri

  • Biofilm Research Group, Department of Biological Sciences, University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada
  • Correspondence
    Howard Ceri
    ceri{at}ucalgary.ca

    • Microbiology 2010; 156(12):3699–3709 · https://doi.org/10.1099/mic.0.040824-0

    View at publisher PubMed

    Abstract

    Pseudomonas aeruginosa has long been known to produce phenotypic variants during chronic mucosal surface infections. These variants are thought to be generated to ensure bacterial survival against the diverse challenges in the mucosal environment. Studies have begun to elucidate the mechanisms by which these variants emerge in vitro; however, too little information exists on phenotypic variation in vivo to draw any links between variants generated in vitro and in vivo. Consequently, in this study, the P. aeruginosa gacS gene, which has previously been linked to the generation of small colony variants (SCVs) in vitro, was studied in an in vivo mucosal surface infection model. More specifically, the rat prostate served as a model mucosal surface to test for the appearance of SCVs in vivo following infections with P. aeruginosa gacS− strains. As in in vitro studies, deletion of the gacS gene led to SCV production in vivo. The appearance of these in vivo SCVs was important for the sustainability of a chronic infection. In the subset of rats in which P. aeruginosa gacS− did not convert to SCVs, clearance of the bacteria took place and healing of the tissue ensued. When comparing the SCVs that arose at the mucosal surface (MS-SCVs) with in vitro SCVs (IV-SCVs) from the same gacS− parent, some differences between the phenotypic variants were observed. Whereas both MS-SCVs and IV-SCVs formed dense biofilms, MS-SCVs exhibited a less diverse resistance profile to antimicrobial agents than IV-SCVs. Additionally, MS-SCVs were better suited to initiate an infection in the rat model than IV-SCVs. Together, these observations suggest that phenotypic variation in vivo can be important for maintenance of infection, and that in vivo variants may differ from in vitro variants generated from the same genetic parent.

    Edited by: W. J. Quax

    INTRODUCTION

    It has become increasingly evident that phenotypic diversification is a survival strategy that Pseudomonas aeruginosa employs. Under various laboratory conditions, P. aeruginosa produces phenotypic variants with altered colony morphologies (Boles et al., 2004; Davies et al., 2007; Déziel et al., 2001; Kirisits et al., 2005; Starkey et al., 2009). Interestingly, the production of these variants appears to be related to the biofilm mode of growth; P. aeruginosa produces far larger proportions of colony variants in biofilms compared with planktonic cultures (Boles et al., 2004). The promotion of variants in biofilms may, in part, account for the observation that biofilms are inherently more resistant to antimicrobials than their planktonic (free-living) counterparts (Ceri et al., 1999b; Costerton et al., 1999). Indeed, given that variant production in biofilms is increased when exposed to environmental stressors, it is likely that variants play a role in community survival. P. aeruginosa variants have been shown to be more resistant than wild-type (WT) to heavy metals, oxidants and antibiotics (Boles et al., 2004; Boles & Singh, 2008; Davies et al., 2007; Harrison et al., 2007).

    Given the diversity and resistance of biofilms observed in vitro, it follows that phenotypic diversification may be a survival strategy P. aeruginosa adopts during mucosal surface infections, such as those in cystic fibrosis (CF) patients, which are characterized by biofilm formation (Singh et al., 2000). Examinations of respiratory samples from CF patients with chronic P. aeruginosa infections tend to reveal numerous morphological variants, which supports this proposal (Drenkard & Ausubel, 2002; Govan & Deretic, 1996; Häussler et al., 1999, 2003; Zierdt & Schmidt, 1964). The most common morphological variants isolated include small colony variants (SCVs) and mucoid colonies that overexpress alginate (Govan & Deretic, 1996; Häussler et al., 1999). The presence of either of these variant types in sputa from CF patients correlates with decreased lung function and a poorer prognosis for the patient (Häussler et al., 1999; Pedersen et al., 1992; Schneider et al., 2008). Moreover, SCVs are better able to form biofilms, adhere to lung epithelia, resist antibiotic treatment and, in some cases, are more cytotoxic to macrophages than WT isolates (Davies et al., 2007; Drenkard & Ausubel, 2002; Häussler et al., 2003; von Götz et al., 2004). Together, these observations suggest that phenotypic variants play an important role in the disease process. Yet, it is unclear what triggers variant generation in vivo. Diversity may be initiated by stress from the host immune system, antibiotics prescribed to the patient or perhaps a combination of the two. Some recent work has begun to implicate antibiotics in variant production as long-term antibiotic use has correlated with the emergence of SCVs (Drenkard & Ausubel, 2002; Häussler et al., 1999). It is likely that selection pressure from the host mucosal environment also contributes to variant production. However, studying phenotypic variation within a host is largely an unexplored avenue of research. For instance, the question remains whether the genetic pathways in P. aeruginosa that lead to phenotypic variation in vitro are the same pathways that produce variants in vivo.

    In an exploration of this question, we chose to study phenotypic diversification within a host as mediated by the P. aeruginosa gacS (global activator of antibiotic and cyanide synthesis) gene. The gacS gene, along with gacA, constitutes an important two-component regulatory system involved in biofilm formation, environmental adaptation and chronic infections. Interestingly, it has been shown in vitro that stress applied to a P. aeruginosa gacS− strain results in phenotypic diversification of this strain into SCVs (Davies et al., 2007). These SCVs are hyper-biofilm formers that are more resistant to antimicrobials than either the gacS− parent or WT bacteria. We hypothesized that stress from the host encountered at a mucosal infection site would likewise result in the production of SCVs from P. aeruginosa gacS− bacteria. These in vivo-derived SCVs were predicted to share the same characteristics as in vitro SCVs in terms of their increased biofilm formation and antimicrobial resistance.

    In an examination of this hypothesis, we used our previously developed rat prostatitis model of acute and chronic P. aeruginosa infections as a model mucosal surface to test if gacS− bacteria would diversify to form SCVs under in vivo stress (Nelson et al., 2009). A P. aeruginosa PA14 gacS− strain previously characterized by our laboratory was used to infect the rat prostate so that in vivo-derived SCVs produced by this mutant would be comparable with in vitro SCVs from the same parent (Davies et al., 2007). By using the rat prostate model in conjunction with the PA14 gacS− mutant we show that diversity generation can be critical for maintaining an infection in vivo. Furthermore, we show that SCVs that arise in vivo exhibit both similarities to and differences from in vitro SCVs derived from the same gacS− parent.

    METHODS

    Bacteria, media and culture conditions.

    P. aeruginosa PA14 strains used and isolated in this study are summarized in Table 1⇓. Notably, PA14 mucosal surface-derived SCVs (MS-SCVs) were recovered from both acute and chronic PA14 gacS− infections. For antimicrobial susceptibility testing, MS-SCVs recovered from acute and chronic infections behaved identically, and therefore data are shown only for the acute MS-SCV isolate. All strains were stored at −70 °C in Microbank vials (Pro-Lab Diagnostics) according to the manufacturer's instructions. To create an inoculum for the animal model, second P. aeruginosa subcultures grown on Miller Luria–Bertani (LB; Difco) agar were used to establish a culture of 1×106 cells (ml saline)−1, as determined by McFarland standards. Bacterial numbers were later reconfirmed by plate counts on both LB and Pseudomonas isolation agar (PIA; Difco). P. aeruginosa isolated from the animal model following infection was cultured on both LB and PIA. For antimicrobial susceptibility testing, all PA14 strains were grown and cultured on trypticase soy broth and trypticase soy agar (TSB and TSA, respectively; Difco).

    Table 1.

    Bacterial strains used in this study

    Animal model.

    The rat prostate served as a model for mucosal surface infections and was infected as described previously (Ceri et al., 1999a; Nelson et al., 2009; Nickel et al., 1990; Phan et al., 2008; Rippere-Lampe et al., 2001). Briefly, 300 g male Sprague–Dawley rats were obtained from the Life and Environmental Sciences Animal Resource Centre at the University of Calgary. The rats were maintained in polycarbonate box cages on aspen chip bedding and were provided rat chow and water ad libitum. Cages were housed at 20±2 °C and 40±10 % relative humidity with 12 h of daily illumination. During the infection procedure, rats were anaesthetized with halothane and catheterized through the urethra with a lubricated sterile PE10 polyethylene feeding tube. P. aeruginosa inocula, prepared as above, were injected into the catheter at a volume of 0.2 ml with the catheter placed at the base of the prostate. Four P. aeruginosa (PA14) strains were used to infect the rat prostate: WT, gacS, in vitro-derived SCVs (IV-SCVs) and MS-SCVs. The PA14 gacS− strain was instilled into 16 rats for both the acute and the chronic models (n=32). All other PA14 strains (WT, IV-SCV and MS-SCV) were instilled into eight rats for both the acute and the chronic models (n=48). The prostate was infected with P. aeruginosa for the lengths of time that most accurately emulate the pathology seen in human acute and chronic prostate infections (2 and 12 days, respectively), after which rats where killed by CO2 asphyxiation (Ceri et al., 1999a, c; Nelson et al., 2009). Approval for this study was granted by The Life and Environmental Sciences Animal Care Committee and all animal experiments were performed in accordance with guidelines of the Canadian Council of Animal Care.

    Tissue preparation.

    The ventral prostate was aseptically removed from the rat, photographed and weighed. The prostate tissue was then cut into four sections. One piece was homogenized in sterile saline to determine bacterial counts per milligram of tissue. The remaining three pieces were processed for histology, myeloperoxidase (MPO) assays and cytokine analysis, as described in the following sections.

    Histological sections.

    In accordance with previous studies, prostate tissue fixed in 10 % (v/v) neutral-buffered formalin was embedded in paraffin, cut to sections of 5 μm thickness and stained with standard haematoxylin and eosin (H&E) protocols (Nelson et al., 2009; Phan et al., 2008). Sections were scored blindly according to the criteria established by a veterinary anatomical pathologist, as described previously (Nelson et al., 2009). Acute sections (2 days) were scored as follows: Grade 0, a normal appearance; Grade 1, oedema and infiltrates of polymorphonuclear (PMN) leukocytes in the prostate stroma; Grade 2, PMN leukocytes in the prostate acini; Grade 3, loss of epithelial and/or basement membrane architecture in the prostate acini; Grade 4, haemorrhage; and Grade 5, tissue necrosis. Chronic sections (12 days) were graded as microscopically showing: Grade 0, a normal appearance; Grade 1, oedema and PMN or mononuclear (MN) leukocytes in prostate stroma; Grade 2, PMN or MN leukocytes in the prostate acini; Grade 3, loss of epithelial and/or basement membrane architecture in the prostate acini; Grade 4, necrosis accompanied by PMN and MN leukocytes and/or fibrosis; and Grade 5, necrosis with no evidence of tissue repair and absence of PMN and MN leukocytes.

    Myeloperoxidase assay.

    Prostate tissue was homogenized in hexadecyltrimethyl ammonium bromide (HTAB) buffer followed by centrifugation at 13 000 g for 2.5 min. Supernatants were mixed with O-dianisidine in phosphate buffer in a 96-well microtitre plate, followed by immediate readings of optical density at 450 nm (OD650) by using a TERMOmax microplate reader with Softmax Pro data analysis software (Molecular Devices). MPO activity was recorded per milligram of protein present in the prostate sample (Mullane et al., 1985).

    Cytokine [interleukin (IL)-1β, growth-regulated oncogene/cytokine-induced neutrophil chemoattractant (Gro/CINC)-1, IL-10 and transforming growth factor (TGF)-β] assays.

    Tissues to be screened for cytokine levels were homogenized in PBS containing 2 μl protease inhibitor cocktail ml−1 (Sigma). The homogenates were centrifuged at 1000 g for 10 min and supernatants were used for cytokine assays. Assays were performed by using commercial ELISA kits according to the manufacturer's instructions for rat IL-10, TGF-β, Gro/CINC-1 (R&D systems) and IL-1β (eBiosciences).

    Biofilm formation.

    In vitro biofilms of P. aeruginosa were grown in the Calgary Biofilm Device (CBD) (Innovotech) according to the methods previously developed in our laboratory (Ceri et al., 1999b). In brief, P. aeruginosa strains of interest were cultured twice on TSA from cryogenic stocks. Colonies from the second subculture were used to create a starting inoculum of 107 c.f.u. ml−1 by suspending colonies in 0.9 % (w/v) saline to match a 1.0 McFarland standard, followed by a 30-fold dilution in TSB. A 150 μl aliquot of this inoculum was then placed into each well of a 96-well micotitre plate, into which the sterile lid of the CBD was placed. The lid of the CBD contains 96 pegs on which biofilms grow following placement in the inoculum. Biofilm formation on the peg was facilitated by incubation at 37 °C with 95 % relative humidity on a gyrorotary shaker at 125 r.p.m. Following incubation of the inoculated device for 24 h, pegs from the device were collected for either confocal scanning laser microscopy (CSLM), for antimicrobial susceptibility testing or to evaluate biofilm cell density. To evaluate biofilm cell density, pegs were rinsed in 0.9 % (w/v) saline, broken off the lid with sterile pliers and placed in 0.9 % (w/v) saline containing Tween 20 (1 %, v/v). Pegs were sonicated for 30 min in an Aquasonic model 250HT ultrasonic cleaner (VWR Scientific) to disrupt the biofilms from the pegs (Harrison et al., 2010). To ensure biofilms were removed from the pegs, four sonicated pegs from each strain grown in the CBD were visualized by CSLM. In all cases, biofilms were eradicated from the pegs and only a few adherent bacterial cells remained (data not shown). The biofilms removed from the pegs were serially diluted in 0.9 % (w/v) saline and plated on TSA for viable cell counts.

    CSLM.

    The 3D structure of biofilms was evaluated by CSLM. Pegs that were broken from the CBD were stained with 0.05 % (w/v) acridine orange (Sigma) for 5 min. The fluorescently labelled biofilms were examined by using a Leica DM IRE2 spectral confocal and multiphoton microscope with a Leica TCS SP2 acoustic optical beam splitter (Leica Microsystems), as described previously (Davies et al., 2007; Harrison et al., 2006). All images of biofilm slices were captured by using a ×63 water-immersion objective. Three-dimensional reconstruction of the confocal image stacks was accomplished by using IMARIS v6.3.1 software (Bitplane).

    Stock solution of antimicrobials and neutralizers.

    Antibiotics, metals, 30 % (w/v) hydrogen peroxide (H2O2) and hypochlorite (OCl−) were purchased from Sigma. Stock solutions of antibiotics were made to a concentration of 5120 μg ml−1 by using double-distilled water (ddH2O) as a dilutant, followed by storage in 0.5 ml aliquots at −70 °C. Metals were also diluted to starting concentrations in ddH2O, but were syringe-filtered and stored at room temperature prior to use. H2O2 and OCl− were diluted directly from the bottle supplied by the manufacturer immediately before use in the challenge plate. Challenge plates were made containing the antimicrobial of interest in Mueller Hinton broth (Difco) 30 min prior to use. For challenge plates containing metals, universal neutralizer was added to the recovery media (TSB, 1 % Tween 20, v/v). For the universal neutralizer, reduced glutathione (GSH) and l-cysteine were prepared at 0.25 M in ddH2O, syringe-filtered and stored at −20 °C.

    Antimicrobial susceptibility testing.

    Antibiotic, metal and biocide susceptibility testing was performed on the P. aeruginosa biofilms that had grown on the lid of the CBD as previously described (Ceri et al., 1999b; Davies et al., 2007; Harrison et al., 2004, 2010). All P. aeruginosa strains were challenged with antimicrobials after 24 h biofilm growth. At 24 h, the lid of the device was rinsed in 200 μl of 0.9 % (w/v) saline and then placed in a microtitre plate containing serial, twofold dilutions of the antimicrobial of interest (the ‘challenge plate’). The first and last rows of the challenge plate did not contain any antimicrobial and were used as either sterility controls or growth controls, respectively. After 24 h exposure to the challenge plate, the lid of the device was removed, rinsed twice in 0.9 % (w/v) saline and placed in 200 μl of recovery media (TSB, 1 % Tween 20, v/v). Biofilms were disrupted in the recovery medium by sonication in an ultrasonic cleaner. Following this, aliquots of 20 μl from the wells of the recovery medium were serially diluted tenfold in 0.9 % (w/v) saline and plated on TSA. Viable cells surviving antimicrobial exposure were enumerated on TSA plates that were incubated for 24 h at 37 °C.

    Statistical analysis.

    Group data are expressed as mean±sem. Figures and statistical analyses were compiled by using GraphPad Prism 5 software. Bacterial counts and MPO readings were log10-transformed. Following this, bacterial counts, MPO assays and cytokine ELISAs were analysed by one-way ANOVA and Tukey's multiple comparison test. P-values <0.05 were considered statistically significant.

    RESULTS

    Appearance of gacS− SCVs in vivo

    Since P. aeruginosa gacS− generates SCVs under stressful in vitro conditions (Davies et al., 2007), we believed the same would hold true under in vivo stress. Consequently, the rat prostate was infected with PA14 gacS− and the appearance of SCVs in prostate tissue was monitored for acute and chronic infections (2 and 12 days post-infection, respectively) (Fig. 1⇓). Notably, the total number of gacS−-infected rats differed between acute and chronic rats (n=12 vs n=14) as catheter leakage occurred from two rats in the acute group, disqualifying them from analysis.

    Figure image not available in archive
    Fig. 1.

    Bacterial colonization and prostate inflammation in rat prostates infected with P. aeruginosa PA14 strains. Solid grey bars reperesent gacS−-infected rats. This group was divided into two subgroups: one in which SCVs were produced from gacS− bacteria (checked) and one where no SCVs were isolated (striped). Black bars denote SCV-infected rats. Bacterial numbers in prostate tissue (a, e), weight of prostate (b, f), MPO levels in prostate tissue (c, g) and IL-1β levels in prostate tissue (d, h) during acute (a–d) and chronic (e–h) infections. Horizontal bars indicate a statistical difference between groups (P<0.05). For (a–d) and (e–h), respectively, gacS−, n=12 or 14; gacS− with SCV, n=8 or 10; gacS− without SCV, n=4.

    SCVs were isolated from prostate homogenates for the majority of gacS−-infected rats; 67 and 71 % of rats were positive for SCVs for acute and chronic infections, respectively. In acute infections, SCVs comprised a mean of 13 % of the bacteria isolated, whereas in chronic infections 24 % of the isolated colonies were SCVs. By means of comparison, in rats infected with PA14 WT (n=8) no SCVs were isolated. Whereas SCVs arose in vivo for the majority of rats infected with P. aeruginosa gacS−, it was interesting to note that SCVs failed to be produced in approximately 30 % of rats in both acute and chronic infections. When subsequent measures of bacterial counts and inflammation were performed, bacterial survivability and infectious capability were significantly reduced when SCVs were not present (Fig. 1⇑). In the subset of rats in which SCVs were not present, bacteria were nearly eradicated from the prostate in both acute and chronic infections (Fig. 1a, e⇑). In contrast, when SCVs were isolated from the prostate, bacterial counts similar to WT levels were found (Fig. 1a, e⇑). Similar trends were observed when inflammation was assessed by prostate weight (as a measure of oedema) and MPO levels (as a measure of inflammatory cell recruitment) (Mullane et al., 1985). Comparing rats which differed in SCV isolation, significantly lower prostate weight and MPO levels were seen in the non-SCV-producing group in both acute and chronic infections (Fig. 1b, c, f, g⇑).

    Tissue damage in rats infected with PA14 gacS− and WT bacteria was scored to assess whether the appearance of SCVs correlated with the severity of tissue damage. As emergence of SCVs correlates with decreased lung function and weight loss in CF patients (Häussler et al., 1999; Schneider et al., 2008), we reasoned that mucosal tissue damage would be more pronounced in cases where SCVs were isolated from prostate homogenates. In line with this, the majority of rats scored in the moderate to severe range of tissue damage when SCVs were isolated from either acute or chronic infections (Fig. 2⇓). In stark contrast, no rats scored in this range when SCVs were not isolated from acute infections (Fig. 2a⇓). In chronic infections, in the absence of SCVs, some rats did score in the moderate range, but this represented only 20 % of the rats infected (Fig. 2b⇓).

    Figure image not available in archive
    Fig. 2.

    Percentage of rats that displayed moderate to severe inflammation (graded as 3, 4 or 5 out of a score of 5) following prostate infection with P. aeruginosa PA14 strains. Prostate sections were processed, stained and graded as described in Methods. (a) Acute infections, (b) chronic infections.

    Effect of gacS− SCV infection on mucosal surfaces

    Since prostate infection was more apparent in cases where SCVs were produced from P. aeruginosa gacS−, we hypothesized that a mucosal surface infection consisting strictly of SCVs would result in considerable infection and inflammation. Therefore, a set of rats were infected with gacS− SCVs that were isolated from two different sources: the prostate mucosal surface (MS-SCV) and a biofilm grown in vitro (IV-SCV). We used both the MS-SCV- and the IV-SCV-derived SCVs to infect the rat prostate to see if pathogenicity was altered based on the source of SCV. Interestingly, differences were seen between IV-SCV and MS-SCV in acute infections only (Figs 1⇑ and 2⇑). In acute infections, IV-SCV prostate infections produced significantly lower bacterial counts, prostate weight and MPO levels than MS-SCV infections (Fig. 1a–c⇑). Surprisingly, for all measures of infection and inflammation, IV-SCV infections were statistically equivalent to gacS− infections that did not produce any SCVs (Fig. 1a–c⇑). Similar trends were seen for tissue damage where histological signs of inflammation were mostly absent in IV-SCV rats (Fig. 2a⇑). In chronic infections, no differences were observed between IV-SCVs and MS-SCVs, both of which caused statistically higher inflammation and tissue damage than the gacS− non-SCV-producing group (Figs 1e–g⇑ and 2b⇑). Of note, MS-SCVs and WT did not differ statistically with regard to inflammatory levels, except for measures of prostate weight and MPO levels in acute infections where MS-SCVs surpassed WT (Fig. 1b, c⇑).

    As an additional measure of inflammation in SCV-infected rats, both pro- and anti-inflammatory cytokine levels in rat prostate tissue were determined. The pro-inflammatory cytokines that were chosen for assessment were selected based on the fact that they are well-established markers of inflammation in the rat prostate (Nelson et al., 2009; Phan et al., 2008). Consistent with our previous observations, the pro-inflammatory cytokine IL-1β was elevated in acute WT and MS-SCV infections in comparison with IV-SCV- and non-SCV-producing groups (Fig. 1d⇑). As typical for IL-1β, levels were reduced considerably in chronic infections, such that differences between groups were no longer noticeable (Fig. 1h⇑). A second pro-inflammatory cytokine, Gro/CINC-1, showed similar trends to IL-1β; however, here differences between groups did not reach statistical significance (data not shown). Lastly, levels of the anti-inflammatory cytokines IL-10 and TGF-β were measured; levels were relatively consistent among groups, indicating that these cytokines do not play a role in the response to P. aeruginosa gacS− or P. aeruginosa SCVs (data not shown).

    Biofilm formation of gacS− SCVs from mucosal surface infections

    Biofilm formation has been shown to vary between WT, gacS− and in vitro-derived SCV P. aeruginosa morphotypes. IV-SCVs isolated from P. aeruginosa form exceptionally dense biofilms, whereas parental gacS− biofilms form very sparse and thin biofilms (Boles et al., 2004; Davies et al., 2007; Drenkard & Ausubel, 2002; Häussler et al., 2003). Thus, we hypothesized that dense biofilms would likewise be produced by the MS-SCVs isolated from the prostate. To test this, PA14 IV-SCVs, as well as PA14 WT, gacS− and MS-SCVs isolated from prostate homogenates, were grown in the CBD for 24 h and subsequently visualized by CSLM. At 24 h, similar cell densities of 5.5±0.4, 5.7±0.5, 6.0±0.7 and 6.1±0.6 log10 c.f.u. per peg were seen for PA14 WT, gacS−, IV-SCV and MS-SCV biofilms, respectively. However, the distribution of cell density was markedly different; P. aeruginosa gacS− formed flat microcolonies across the peg whereas WT and SCV strains displayed a concentrated biomass at the air–liquid interface. As a result, biofilm density varied between strains. Representative biofilms visualized with CSLM from each strain are shown in Fig. 3⇓. WT P. aeruginosa formed biofilms that displayed some degree of complexity (Fig. 3a⇓), whereas biofilms from gacS− strains sparsely covered the surface (Fig. 3b⇓). IV-SCVs formed dense biofilms with complicated 3D structure (Fig. 3c⇓); these were similar to MS-SCV biofilms (Fig. 3d⇓). Biofilms of MS-SCVs reached a height of 40 μm, about two and four times the height of WT and gacS− biofilms, respectively.

    Figure image not available in archive
    Fig. 3.

    Biofilm formation of P. aeruginosa PA14 wild-type (a), gacS– (b), IV-SCVs (c) and MS-SCVs (d). Biofilms were grown in TSB in the CBD, stained with acridine orange and imaged by using CSLM. The 3D reconstructions (created by using IMARIS v6.3.1 software) and side-views of each biofilm are shown.

    Antimicrobial susceptibility of gacS− SCVs from mucosal surface infections

    A number of studies have found that SCVs, whether isolated from in vitro biofilms or from CF sputum, are more resistant to antimicrobials than the WT (Davies et al., 2007; Drenkard & Ausubel, 2002; Häussler et al., 1999). Conversely, P. aeruginosa gacS− is particularly susceptible to antimicrobials (Davies et al., 2007). Thus, it was hypothesized that the MS-SCVs isolated from the prostate would be more resistant to antimicrobial agents than PA14 WT or gacS−. Additionally, antimicrobial susceptibility of IV-SCVs was evaluated to determine if differences existed been in vivo and in vitro SCVs derived from the same gacS parent.

    Antimicrobial susceptibility was evaluated by performing susceptibility testing on biofilms of PA14 WT, gacS−, IV-SCV and MS-SCV grown in the CBD for 24 h as similar cell densities were seen at this time point (see biofilm results above). The anti-biofilm activity of four antimicrobial agents, H2O2, OCl−, tobramycin and Ag+, was tested against each of these four strains (Table 2⇓ and Fig. 4⇓). Bactericidal concentrations of the antimicrobial agents were defined as the minimum biofilm eradication concentration required to kill 99.9 % of the bacteria (MBEC99.9). This definition is consistent with the American Clinical and Laboratory Standards Institute's definition of a bactericidal concentration of a 3 log10 reduction in viable bacterial cells (). The MBEC99.9 values of each of the antimicrobial agents against the WT, gacS−, IV-SCV and MS-SCV strains are summarized in Table 2⇓. As expected, and consistent with previously published work, the MBEC99.9 values were considerably higher for IV-SCVs than for WT and gacS− for all antimicrobials tested. Results for MS-SCVs were similar to those for IV-SCVs for the oxidants H2O2 and OCl−. Unlike IV-SCVs, however, biofilms of MS-SCVs did not show any increased resistance to the heavy metal Ag+. The same held true for Cu2+; IV-SCVs were highly resistant to this heavy metal, but MS-SCVs and gacS− isolates were very susceptible to killing (data not shown). Finally, whereas MS-SCVs were more resistant to tobramycin than the gacS− isolate, MBEC99.9 values were the same for MS-SCVs and WT.

    Figure image not available in archive
    Fig. 4.

    Susceptibility of P. aeruginosa gacS− (•), IV-SCV (▴) and MS-SCV (▪) strains to antimicrobials. Biofilms were exposed to various antimicrobials as detailed in Methods. Killing of biofilms against (a) H2O2, (b) OCl−, (c) tobramycin and (d) Ag+.

    Table 2.

    Biofilm antimicrobial susceptibility of P. aeruginosa PA14 WT, gacS−, IV-SCV and MS-SCV strains against various antimicrobials

    MBEC99.9, minimum biofilm eradication concentration required to kill 99.9 % of the bacteria. Values are the mean±sd of data from four independent trials.

    When resistance of biofilms from PA14 gacS−, IV-SCVs and MS-SCVs was examined against various antimicrobial concentrations, reproducible patterns of resistance were observed (Fig. 4⇑). In general, with the exception of Ag+, MS-SCVs were far more resistant to antimicrobials than gacS− strains (Fig. 4⇑). MS-SCVs and IV-SCVs were equally resistant to antimicrobials at all antimicrobial concentrations tested, except that MS-SCVs succumbed at a much lower concentration of Ag+ (Fig. 4⇑). As illustrated by heavy metals, the source of SCV (mucosal surface or in vitro) affected the resistance profile. Therefore, gacS− colonies isolated from the prostate were also tested for their antimicrobial resistance to see if they exhibited an altered resistance pattern from the original gacS− strain used to infect the prostate. However, in all cases, gacS− strains isolated from the prostate showed the same resistance pattern as the infecting PA14 gacS− strain (data not shown). Thus, MS-SCV biofilms were resistant to H2O2 and OCl− oxidative stressors, but not to heavy metals; gacS− biofilms were highly susceptible to all antimicrobials tested; and IV-SCV biofilms were, under the conditions tested here, universally resistant.

    DISCUSSION

    In humans, mucosal surfaces are often the target of opportunistic P. aeruginosa infections. It has been proposed that successful colonization of these mucosal environments, which present numerous and heterogeneous challenges, necessitates the evolution and diversification of bacteria (Oliver et al., 2000; Rainey & Travisano, 1998). The fact that numerous phenotypic variants of P. aeruginosa have been recovered from patients with mucosal infections supports this proposal (Govan & Deretic, 1996; Häussler et al., 1999, 2003; Zierdt & Schmidt, 1964). However, although these variants have been well characterized, factors leading to their production at the mucosal surface are not well understood. As such, it becomes difficult to speculate if mechanisms attributed to phenotypic variation in vitro are also at play in vivo. In this study, we addressed this issue by examining a gene (gacS) previously linked to P. aeruginosa variant production in vitro (Davies et al., 2007) to determine its ability to also produce variants in vivo. The gacS gene was chosen for two reasons. First, SCVs produced from deletions of the gacA and gacS genes have been well characterized in vitro (Davies et al., 2007; Parkins et al., 2001). This allows for a thorough comparison with the in vivo SCVs generated from gacS− bacteria in this study. Secondly, unlike WT bacteria which rarely produce stable SCVs, exposure of gacS− bacteria to in vitro stress causes stable SCVs to be generated with a high frequency. These SCVs exhibit characteristics similar to other SCVs described in the literature, including resistance to antimicrobials and dense biofilm formation (Boles et al., 2004; Boles & Singh, 2008; Kirisits et al., 2005; Starkey et al., 2009). Therefore, we believed that gacS− SCVs produced in vivo would be representative of P. aeruginosa SCVs.

    Having selected gacS− bacteria for use in our in vivo infection model (Nelson et al., 2009), we found that in the majority of rats infected with gacS− bacteria, MS-SCVs were produced. However, in approximately 30 % of animals, no MS-SCVs were isolated. The amount of colonizing bacteria, the level of inflammation and the extent of tissue damage in animal tissues were markedly different between the two groups, showing that infection is severely compromised when SCVs were not produced from a gacS− strain (Figs 1⇑ and 2⇑). Considering that neither acute nor chronic infections were very successful in the absence of MS-SCV, it is curious that PA14 gacS− bacteria failed to convert to SCVs in as many as 30 % of animals. Given that the input signals that GacS responds to are unknown, as are the mechanisms by which P. aeruginosa gacS− enhances SCV production, it is difficult to speculate on the reason for SCV appearance in some, but not all, PA14 gacS−-infected animals. As future research addresses these signalling pathways, it may become clearer why SCVs are produced in the infection model. Additionally, clues may be offered through recent and ongoing research concerning other genes and molecules linked to SCV production. For instance, the intracellular signalling molecule cyclic di-GMP (c-di-GMP) has recently been implicated in SCV production (Meissner et al., 2007; Starkey et al., 2009). Mutations in the wspF gene constitutively activate the diguanylate cyclase WspR, which elevates c-di-GMP levels and subsequently converts a WT P. aeruginosa strain into SCVs. Additionally, depletion of c-di-GMP in a wspF mutant strain converts the SCV back to WT morphology (D'Argenio et al., 2001; Hickman et al., 2005). The extrapolysaccharide biosynthetic locus has also been implicated through the pel and psl gene clusters, which are required for the SCV phenotype (Starkey et al., 2009). Thus, SCV production in an infection, or lack thereof, may be a result of various signalling pathways. Regardless, in any of these possible pathways, it is hypothesized that there is a race to convert gacS− bacteria to the more successful SCV form. In cases where the possible signal is recognized and the switch to the SCV phenotype occurs, the infection progresses to a typical chronic prostatitis; when the recognition of the switch does not occur, the gacS− cells are not able to withstand the host immune response, and therefore bacterial clearance and healing of the prostate takes place.

    Although the reasons for SCV production in select gacS− prostate infections cannot yet be accounted for, it can be said that when MS-SCVs were produced, colonization and inflammation occurred at the mucosal surface (Figs 1⇑ and 2⇑). Likewise, when MS-SCVs infected the prostate, markers of colonization and inflammation were noticeably present in both acute and chronic infections (Figs 1⇑ and 2⇑). Interestingly, apart from measures of prostate weight and MPO in acute infections, MS-SCV infections were similar to WT in terms of virulence (Fig. 1b, c⇑). This response is slightly different from that reported by Starkey et al. (2009), which showed that clinical and laboratory SCVs from other genetic backgrounds were less immunostimulatory than WT. The difference could reflect that the study by Starkey et al. (2009) was performed on human epithelial cell lines for 4 h, whereas the current study was undertaken in a rat mucosal surface model for a period of either 2 or 12 days. It could also be a phenomenon particular to MS-SCVs in a gacS− background where these SCVs match the WT in their immune stimulation. In either case, the data suggest that MS-SCV production in a gacS− mutant produces an infection comparable with WT and therefore the gacS gene is not necessarily required for an infection. This would stand in contrast with the other member of the GacS–GacA two-component system, the gacA gene, which is required for virulence in both acute and chronic infections in numerous model systems (Chieda et al., 2005; Coleman et al., 2003; Mahajan-Miklos et al., 1999; Rahme et al., 2000; Tan et al., 1999). This difference is probably accounted for by the fact that, unlike gacS− mutants, gacA− mutants do not produce stable SCVs (Parkins et al., 2001).

    It is clear that SCVs facilitate infection (Figs 1⇑ and 2⇑) and that the GacS system is involved in their production. When the gacS gene is supplied back to a SCV isolate from a gacS− strain, the SCV reverts back to its larger colony phenotype (Davies et al., 2007). Additionally, P. aeruginosa gacS− strains produce stable SCVs at a high frequency in vitro (Davies et al., 2007) and in vivo, as shown in this work (13 and 24 % of bacteria in the prostate are MS-SCVs in acute and chronic infections, respectively). As such, the question becomes: could losing the gacS gene in an effort to facilitate high-frequency production of SCVs be advantageous? The present data suggest that this may be the case as the gacS gene seems to represent a fitness trade-off. While the gacS− mutants were highly susceptible to antimicrobial agents, this cost was counter-balanced by the production of MS-SCVs that were exceptionally more resistant to antimicrobial stressors than WT (Table 2⇑ and Fig. 4⇑). Additionally, MS-SCVs were generally similar to WT in terms of virulence (Fig. 1⇑) and therefore MS-SCVs were not easily cleared nor excessively damaging to the host, neither of which would benefit survivability of the bacteria. Together, this implies that a deletion in gacS affords P. aeruginosa the ability to produce a high frequency of phenotypic variants that can still maintain infection and strongly resist antimicrobials. Moreover, the MS-SCVs produced from gacS− strains were very stable, which would mean a resistant population would remain even when the initial stressor was removed. Further supporting that a deletion of the gacS gene could offer a survival advantage, studies with Pseudomonas fluorescens and Pseudomonas chlororaphis have shown that mutations in gacS are often selected for during phenotypic diversification of these pseudomonads (Sánchez-Contreras et al., 2002; Schmidt-Eisenlohr et al., 2003). However, a cautionary note in interpreting gacS− mutations as favourable in the current study is required, as it is acknowledged that the MS-SCVs arising from gacS− mutants may have accumulated other genetic mutations that contributed to the observed phenotype. Also, it is worth noting that gacS− mutations at the onset of infection do not appear overly advantageous, as gacS− bacteria were cleared in 30 % of rats in the study (Fig. 1⇑). A survival advantage may be more likely if the gacS mutations accumulated over time as they would in a natural infection. In this case, the clearance rate of bacteria may be considerably lower than the one observed in this study where all infecting bacteria already possessed the gacS deletion.

    Subsequent to determining that MS-SCVs arise in vivo, these MS-SCVs were compared with in vitro SCVs from the same gacS− parent. This comparison was undertaken as there has previously been some question as to whether laboratory-derived and clinically derived variants are comparable. Although altered phenotypes between CF and laboratory SCVs were detected by von Götz et al. (2004), others have shown that motility traits and transcriptional profiles between the two types of SCVs were similar (Kirisits et al., 2005; Starkey et al., 2009). To our knowledge, this study is the first to investigate this issue by comparing an in vitro (IV-SCV) and in vivo (MS-SCV) -derived variant arising from the same genetic mutation. When comparing the IV-SCVs and MS-SCVs derived from PA14 gacS− bacteria, it was revealed that they behaved similarly, in that both were able to cause infection (Figs 1⇑ and 2⇑), form dense biofilms (Fig. 3⇑) and resist killing by antimicrobials (Table 2⇑ and Fig. 4⇑). Despite these general similarities, there were notable differences. One clear difference between IV-SCVs and MS-SCVs was their behaviour in acute infections. IV-SCVs and MS-SCVs were equally adept at causing chronic infections, but IV-SCVs were impaired in their ability to cause acute infections (Figs 1⇑ and 2⇑). These data may suggest that passage through a mucosal infection creates MS-SCVs that are capable of colonization and inflammation immediately upon reinfection. In contrast, IV-SCVs can only establish a chronic infection given sufficient time to adapt as they lacked the benefit of a previous exposure to the host system.

    In addition to their ability to cause infection, MS-SCVs and IV-SCVs also differed in the specificities of their antimicrobial resistance. Whereas IV-SCVs showed increased resistance to numerous heavy metals, antibiotics and oxidants, MS-SCVs appeared to be selectively resistant. MS-SCVs were resistant to H2O2 and OCl−, but were similar to WT in terms of their eradication threshold to tobramycin and were not at all resistant to heavy metals (Table 2⇑ and Fig. 4⇑). Thus, MS-SCVs were not indiscriminately resistant to antimicrobials, but rather were resistant to the oxidants H2O2 and OCl−, which they would have encountered through exposure to host neutrophils. Given that MS-SCVs were not resistant to silver and copper, which also kill by inducing oxidative stress (Harrison et al., 2009), it appears that MS-SCVs are resistant to host-specific oxidative stressors. As such, oxidative stressors in the host may prompt phenotypic variation. It has previously been shown that oxidative stress selects for diversification in vitro, and it has since been speculated that oxidative stress similarly causes phenotypic variation in vivo (Boles & Singh, 2008). Our results support this hypothesis. They further suggest that previous exposure to the immune system is not necessarily sufficient for increased resistance to oxidants present in neutrophils as the gacS− colonies recovered from the prostate were not resistant to H2O2 and OCl− (Table 2⇑ and Fig. 4⇑). Therefore, it is likely that the properties of MS-SCVs, such as dense biofilm formation (Fig. 3⇑), are responsible for the increased resistance to neutrophil stressors. Apart from the oxidative stressors present in neutrophils, antibiotics may also be playing a role in the selection process. Unlike the MS-SCVs studied here, SCVs isolated from CF patients are considerably more resistant to tobramycin than WT counterparts (Drenkard & Ausubel, 2002; Häussler et al., 1999). Tobramycin is an antibiotic commonly prescribed to CF patients, and therefore SCVs isolated from these patients may have been selected for because of their increased tolerance to tobramycin. Overall, it would appear that SCVs arising from infection sites exhibit resistance to the stressors they encountered. MS-SCVs differ from IV-SCVs in this regard as IV-SCVs appear to be resistant to most antimicrobials, regardless of the initial stressor that selected the IV-SCV (Foglia et al., 2005). It is currently unclear why IV-SCVs appear to exhibit resistance to most antimicrobials whereas MS-SCVs do not. It is possible that laboratory conditions encourage resistance to numerous stressors whereas host conditions are more likely to select for resistance to a particular stressor. Nevertheless, we can conclude that given the dissimilarities in resistance profiles between MS-SCVs and IV-SCVs, especially in terms of metal tolerance, in vivo and in vitro variants from the same parent can be physiologically different.

    Overall, these differences between MS-SCVs, IV-SCVs and gacS− strains begin to reveal the importance of in vivo diversification for P. aeruginosa and the ways in which it differs from observations made in vitro. However, future studies revealing the genetic mechanisms underlying these differences are required to fully understand the processes involved in successful adaptation to the host.

    Acknowledgments

    This work was supported by a grant from the Natural Science and Engineering Research Council of Canada to H. C. and R. J. T. L. K. N. was funded on an NSERC graduate student fellowship. R. E. A. E. was funded as a summer student by the Markin USRP in Health and Wellness.

    References

    1. Boles, B. R. & Singh, P. K. (2008). Endogenous oxidative stress produces diversity and adaptability in biofilm communities. Proc Natl Acad Sci U S A 105, 12503–12508.
    2. Boles, B. R., Thoendel, M. & Singh, P. K. (2004). Self-generated diversity produces “insurance effects” in biofilm communities. Proc Natl Acad Sci U S A 101, 16630–16635.
    3. Ceri, H., Olson, M. E. & Nickel, J. C. (1999a). Prostatitis: role of the animal model. In Textbook of Prostatitis, pp. 109–114. Edited by Nickel, J. C.. Oxford. : Isis Medical Media.
    4. Ceri, H., Olson, M. E., Stremick, C., Read, R. R., Morck, D. & Buret, A. (1999b). The Calgary Biofilm Device: new technology for rapid determination of antibiotic susceptibilities of bacterial biofilms. J Clin Microbiol 37, 1771–1776.
    5. Ceri, H., Schmidt, S., Olson, M. E., Nickel, J. C. & Benediktsson, H. (1999c). Specific mucosal immunity in the pathophysiology of bacterial prostatitis in a rat model. Can J Microbiol 45, 849–855.
    6. Chieda, Y., Iiyama, K., Yasunaga-Aoki, C., Lee, J. M., Kusakabe, T. & Shimizu, S. (2005). Pathogenicity of gacA mutant of Pseudomonas aeruginosa PA01 in the silkworm, Bombyx mori. FEMS Microbiol Lett 244, 181–186.
    7. Coleman, F. T., Mueschenborn, S., Meluleni, G., Ray, C., Carey, V. J., Vargas, S. O., Cannon, C. L., Ausubel, F. M. & Pier, G. B. (2003). Hypersusceptibility of cystic fibrosis mice to chronic Pseudomonas aeruginosa oropharyngeal colonization and lung infection. Proc Natl Acad Sci U S A 100, 1949–1954.
    8. Costerton, J. W., Stewart, P. S. & Greenberg, E. P. (1999). Bacterial biofilms: a common cause of persistent infections. Science 284, 1318–1322.
    9. D'Argenio, D. A., Gallagher, L. A., Berg, C. A. & Manoil, C. (2001). Drosophila as a model host for Pseudomonas aeruginosa infection. J Bacteriol 183, 1466–1471.
    10. Davies, J. A., Harrison, J. J., Marques, L. L., Foglia, G. R., Stremick, C. A., Storey, D. G., Turner, R. J., Olson, M. E. & Ceri, H. (2007). The GacS sensor kinase controls phenotypic reversion of small colony variants isolated from biofilms of Pseudomonas aeruginosa PA14. FEMS Microbiol Ecol 59, 32–46.
    11. Déziel, E., Comeau, Y. & Villemur, R. (2001). Initiation of biofilm formation by Pseudomonas aeruginosa 57RP correlates with emergence of hyperpiliated and highly adherent phenotypic variants deficient in swimming, swarming, and twitching motilities. J Bacteriol 183, 1195–1204.
    12. Drenkard, E. & Ausubel, F. M. (2002). Pseudomonas biofilm formation and antibiotic resistance are linked to phenotypic variation. Nature 416, 740–743.
    13. Foglia, G. R., Marques, L. L. R., Harrison, J. J., Turner, R. J., Olson, M. E. & Ceri, H. (2005). GacS as a phase dependent on/off switch for stress tolerance in pseudomonads? Phytopathology 95, S30.
    14. Govan, J. R. & Deretic, V. (1996). Microbial pathogenesis in cystic fibrosis: mucoid Pseudomonas aeruginosa and Burkholderia cepacia. Microbiol Rev 60, 539–574.
    15. Harrison, J. J., Ceri, H., Stremick, C. A. & Turner, R. J. (2004). Biofilm susceptibility to metal toxicity. Environ Microbiol 6, 1220–1227.
    16. Harrison, J. J., Ceri, H., Yerly, J., Stremick, C. A., Hu, Y., Martinuzzi, R. & Turner, R. J. (2006). The use of microscopy and three-dimensional visualization to evaluate the structure of microbial biofilms cultivated in the Calgary Biofilm Device. Biol Proced Online 8, 194–215.
    17. Harrison, J. J., Ceri, H. & Turner, R. J. (2007). Multimetal resistance and tolerance in microbial biofilms. Nat Rev Microbiol 5, 928–938.
    18. Harrison, J. J., Tremaroli, V., Stan, M. A., Chan, C. S., Vacchi-Suzzi, C., Heyne, B. J., Parsek, M. R., Ceri, H. & Turner, R. J. (2009). Chromosomal antioxidant genes have metal ion-specific roles as determinants of bacterial metal tolerance. Environ Microbiol 11, 2491–2509.
    19. Harrison, J. J., Stremick, C. A., Turner, R. J., Allan, N. D., Olson, M. E. & Ceri, H. (2010). Microtiter susceptibility testing of microbes growing on peg lids: a miniaturized biofilm model for high-throughput screening. Nat Protoc 5, 1236–1254.
    20. Häussler, S., Tummler, B., Weissbrodt, H., Rohde, M. & Steinmetz, I. (1999). Small-colony variants of Pseudomonas aeruginosa in cystic fibrosis. Clin Infect Dis 29, 621–625.
    21. Häussler, S., Ziegler, I., Lottel, A., von Gotz, F., Rohde, M., Wehmhohner, D., Saravanamuthu, S., Tummler, B. & Steinmetz, I. (2003). Highly adherent small-colony variants of Pseudomonas aeruginosa in cystic fibrosis lung infection. J Med Microbiol 52, 295–301.
    22. Hickman, J. W., Tifrea, D. F. & Harwood, C. S. (2005). A chemosensory system that regulates biofilm formation through modulation of cyclic diguanylate levels. Proc Natl Acad Sci U S A 102, 14422–14427.
    23. Kirisits, M. J., Prost, L., Starkey, M. & Parsek, M. R. (2005). Characterization of colony morphology variants isolated from Pseudomonas aeruginosa biofilms. Appl Environ Microbiol 71, 4809–4821.
    24. Mahajan-Miklos, S., Tan, M. W., Rahme, L. G. & Ausubel, F. M. (1999). Molecular mechanisms of bacterial virulence elucidated using a Pseudomonas aeruginosa Caenorhabditis elegans pathogenesis model. Cell 96, 47–56.
    25. Meissner, A., Wild, V., Simm, R., Rohde, M., Erck, C., Bredenbruch, F., Morr, M., Romling, U. & Haussler, S. (2007). Pseudomonas aeruginosa cupA-encoded fimbriae expression is regulated by a GGDEF and EAL domain-dependent modulation of the intracellular level of cyclic diguanylate. Environ Microbiol 9, 2475–2485.
    26. Mullane, K. M., Kraemer, R. & Smith, B. (1985). Myeloperoxidase activity as a quantitative assessment of neutrophil infiltration into ischemic myocardium. J Pharmacol Methods 14, 157–167.
    27. Nelson, L. K., D'Amours, G. H., Sproule-Willoughby, K. M., Morck, D. W. & Ceri, H. (2009). Pseudomonas aeruginosa las and rhl quorum-sensing systems are important for infection and inflammation in a rat prostatitis model. Microbiology 155, 2612–2619.
    28. Nickel, J. C., Olson, M. E., Barabas, A., Benediktsson, H., Dasgupta, M. K. & Costerton, J. W. (1990). Pathogenesis of chronic bacterial prostatitis in an animal model. Br J Urol 66, 47–54.
    29. Oliver, A., Canton, R., Campo, P., Baquero, F. & Blazquez, J. (2000). High frequency of hypermutable Pseudomonas aeruginosa in cystic fibrosis lung infection. Science 288, 1251–1254.
    30. Parkins, M. D., Ceri, H. & Storey, D. G. (2001). Pseudomonas aeruginosa GacA, a factor in multihost virulence, is also essential for biofilm formation. Mol Microbiol 40, 1215–1226.
    31. Pedersen, S. S., Hoiby, N., Espersen, F. & Koch, C. (1992). Role of alginate in infection with mucoid Pseudomonas aeruginosa in cystic fibrosis. Thorax 47, 6–13.
    32. Phan, V., Belas, R., Gilmore, B. F. & Ceri, H. (2008). ZapA a virulence factor in a rat model of Proteus mirabilis induced acute and chronic prostatitis. Infect Immun 76, 4859–4864.
    33. Rahme, L. G., Stevens, E. J., Wolfort, S. F., Shao, J., Tompkins, R. G. & Ausubel, F. M. (1995). Common virulence factors for bacterial pathogenecity in plants and animals. Science 268, 1899–1902.
    34. Rahme, L. G., Ausubel, F. M., Cao, H., Drenkard, E., Goumnerov, B. C., Lau, G. W., Mahajan-Miklos, S., Plotnikova, J., Tan, M. W. & other authors (2000). Plants and animals share functionally common bacterial virulence factors. Proc Natl Acad Sci U S A 97, 8815–8821.
    35. Rainey, P. B. & Travisano, M. (1998). Adaptive radiation in a heterogeneous environment. Nature 394, 69–72.
    36. Rippere-Lampe, K. E., Lang, M., Ceri, H., Olson, M., Lockman, H. A. & O'Brien, A. D. (2001). Cytotoxic necrotizing factor type 1-positive Escherichia coli causes increased inflammation and tissue damage to the prostate in a rat prostatitis model. Infect Immun 69, 6515–6519.
    37. Sánchez-Contreras, M., Martin, M., Villacieros, M., O'Gara, F., Bonilla, I. & Rivilla, R. (2002). Phenotypic selection and phase variation occur during alfalfa root colonization by Pseudomonas fluorescens F113. J Bacteriol 184, 1587–1596.
    38. Schmidt-Eisenlohr, H., Gast, A. & Baron, C. (2003). Inactivation of gacS does not affect the competitiveness of Pseudomonas chlororaphis in the Arabidopsis thaliana rhizosphere. Appl Environ Microbiol 69, 1817–1826.
    39. Schneider, M., Muhlemann, K., Droz, S., Couzinet, S., Casaulta, C. & Zimmerli, S. (2008). Clinical characteristics associated with isolation of small-colony variants of Staphylococcus aureus and Pseudomonas aeruginosa from respiratory secretions of patients with cystic fibrosis. J Clin Microbiol 46, 1832–1834.
    40. Singh, P. K., Schaefer, A. L., Parsek, M. R., Moninger, T. O., Welsh, M. J. & Greenberg, E. P. (2000). Quorum-sensing signals indicate that cystic fibrosis lungs are infected with bacterial biofilms. Nature 407, 762–764.
    41. Starkey, M., Hickman, J. H., Ma, L., Zhang, N., De Long, S., Hinz, A., Palacios, S., Manoil, C., Kirisits, M. J. & other authors (2009). Pseudomonas aeruginosa rugose small-colony variants have adaptations that likely promote persistence in the cystic fibrosis lung. J Bacteriol 191, 3492–3503.
    42. Tan, M. W., Rahme, L. G., Sternberg, J. A., Tompkins, R. G. & Ausubel, F. M. (1999). Pseudomonas aeruginosa killing of Caenorhabditis elegans used to identify P. aeruginosa virulence factors. Proc Natl Acad Sci U S A 96, 2408–2413.
    43. von Götz, F., Häussler, S., Jordan, D., Saravanamuthu, S. S., Wehmhöner, D., Strüssmann, A., Lauber, J., Attree, I., Buer, J. & other authors (2004). Expression analysis of a highly adherent and cytotoxic small colony variant of Pseudomonas aeruginosa isolated from a lung of a patient with cystic fibrosis. J Bacteriol 186, 3837–3847.
    44. Zierdt, C. H. & Schmidt, P. J. (1964). Dissociation in Pseudomonas aeruginosa. J Bacteriol 87, 1003–1010.