Catheter-Associated Urinary Tract Infection by Pseudomonas aeruginosa Is Mediated by Exopolysaccharide-Independent Biofilms

Abstract

Pseudomonas aeruginosa is an opportunistic human pathogen that is especially adept at forming surface-associated biofilms. P. aeruginosa causes catheter-associated urinary tract infections (CAUTIs) through biofilm formation on the surface of indwelling catheters. P. aeruginosa encodes three extracellular polysaccharides, PEL, PSL, and alginate, and utilizes the PEL and PSL polysaccharides to form biofilms in vitro; however, the requirement of these polysaccharides during in vivo infections is not well understood. Here we show in a murine model of CAUTI that PAO1, a strain harboring pel, psl, and alg genes, and PA14, a strain harboring pel and alg genes, form biofilms on the implanted catheters. To determine the requirement of exopolysaccharide during in vivo biofilm infections, we tested isogenic mutants lacking the pel, psl, and alg operons and showed that PA14 mutants lacking these operons can successfully form biofilms on catheters in the CAUTI model. To determine the host factor(s) that induces the ΔpelD mutant to form biofilm, we tested mouse, human, and artificial urine and show that urine can induce biofilm formation by the PA14 ΔpelD mutant. By testing the major constituents of urine, we show that urea can induce a pel-, psl-, and alg-independent biofilm. These pel-, psl-, and alg-independent biofilms are mediated by the release of extracellular DNA. Treatment of biofilms formed in urea with DNase I reduced the biofilm, indicating that extracellular DNA supports biofilm formation. Our results indicate that the opportunistic pathogen P. aeruginosa utilizes a distinct program to form biofilms that are independent of exopolysaccharides during CAUTI.

INTRODUCTION

Pseudomonas aeruginosa causes biofilm-mediated infections, including catheter-associated urinary tract infection (CAUTI), ventilator-associated pneumonia, infections related to mechanical heart valves, stents, grafts, and sutures, and contact lens-associated corneal infections (1, 2). P. aeruginosa is responsible for 12% of all nosocomial urinary tract infections (UTIs), making it the third most common organism after Escherichia coli and enterococci isolated from UTI patients in the hospital setting (3). Nosocomial UTIs are catheter associated, and the development of bacteriuria is directly related to the duration of catheterization (4, 5). Between 15 and 25% of hospital patients are catheterized for 2 to 4 days during their stays, while many nursing home patients remain catheterized for months or years (5, 6). Catheter-associated bacteriuria leads to an increased length of hospital stay, causing an estimated 900,000 additional hospital days per year (5), and complications associated with nosocomial UTIs cause or contribute to an estimated 7,500 deaths per year (7). Biofilms are a serious problem, as they are often refractory to antibiotic therapy. Antibiotic therapy can eliminate planktonic bacteria, but bacteria within biofilms survive antibiotic treatment (8–10). When antibiotic treatment ends, the biofilm can again shed planktonic cells, resulting in recurrent acute infection. This cycle of infection is difficult to stop and often requires the removal of the contaminated device to eliminate the bacterial biofilm (9, 11). However, removing the contaminated device is only a temporary solution, as replacement with a new device yet again provides a surface for biofilm formation. Using a suitable animal model to investigate the bacterial factors contributing to chronic infections will provide insights and a potential novel target for therapeutic intervention.

P. aeruginosa biofilms consist of an extracellular matrix that includes polysaccharides, proteinaceous components, and extracellular DNA (eDNA) (12–16). Nonmucoid strains of P. aeruginosa produce biofilms that are independent of alginate biosynthesis (17, 18). These biofilms can colonize solid surfaces and form mushroom structures in flow cells (19), rings in culture tubes and microtiter plates (20), or pellicles at the air-liquid interface (21). The main polysaccharide components include the PEL and PSL exopolysaccharides produced by the proteins encoded by the pel and psl genes, respectively (21–23). Mutations in either the pel or psl gene result in bacteria that produce less biofilm (11, 24, 25). The genome of P. aeruginosa strain PA14 lacks pslABC genes and does not produce the PSL polysaccharide. A PA14 pel operon mutant cannot produce the PEL polysaccharide and fails to form biofilm (21). In addition to the requirement of the pel and psl operons, the requirement of eDNA has been demonstrated by the ability of DNase I to reduce biofilm formation (13). Numerous other studies have demonstrated the participation of eDNA in P. aeruginosa biofilms (26, 27). Despite having identified a number of pseudomonal factors that contribute to biofilm formation in vitro, little is known about the contribution of each of these biofilm components during an animal model of CAUTI.

The relevance of biofilm genes identified from in vitro systems during animal infections has not been sufficiently studied because few models for chronic infections exist. One model system involves inserting an implant or catheter containing bacterial biofilm into either the lung (28, 29) or bladder (30, 31) of an animal to mimic catheter-induced chronic pneumonia or UTI, respectively. An advantage of the UTI model is that the biofilm formation and disseminating infection events can be separated. Animals either infected through the use of a catheter containing preformed biofilm or inoculated after the implantation of a sterile catheter had biofilm-based chronic infections (30). Here we used the murine model of CAUTI to test the contribution of extracellular polysaccharides to pseudomonal biofilm-mediated infection. We employed P. aeruginosa strains PAO1 and PA14 and their isogenic mutants lacking the pel, psl, and alg operons encoding the biosynthetic genes for the PEL, PSL, and alginate exopolysaccharides, respectively. Using these strains, we show that PEL, PSL, and alginate exopolysaccharides are dispensable for biofilm formation on the catheter and for further dissemination into the kidneys. The PA14 ΔpelD mutant could also form biofilm on the catheter during mixed infections with the parental PA14 strain. The PA14 ΔpelD mutant participated in in vitro mixed biofilms with PA14, indicating that the ΔpelD mutant can utilize the PEL polysaccharide produced by PA14. The ability of exopolysaccharide-deficient pseudomonas to form biofilm could be induced in vitro by mouse and human urine as well as urea, a major component of urine. Exposure of P. aeruginosa to urea induced a subpopulation of cells to round and lyse. The released eDNA could mediate biofilm formation of both PA14 and the ΔpelD mutant strains. Addition of DNase I degraded eDNA and reduced the biofilm formation of both PA14 and the ΔpelD mutant in the presence of urea. Together, these data demonstrate that P. aeruginosa has a mechanism independent of exopolysaccharides to mediate biofilm formation on catheters during CAUTI.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The strains and plasmids used in this study are listed in Table S1 in the supplemental material. The strains were grown in LB broth. Strain PA14 expressing the lux gene (PA14-lux) and PA14 ΔpelD-lux were constructed by mating recombination of pCTX-lac-lux into the att site, followed by Flp-based recombination to remove the residual pCTX vector sequence (32–34). Plasmid pMMB-PA3702 was maintained in 15 μg ml⁻¹ gentamicin, and 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was used to induce gene expression.

In vitro catheter biofilm assay.

One-centimeter-long pieces of PE50 catheter material (polyethylene; outside diameter, 0.965 mm; inside diameter, 0.58 mm; Braintree Scientific, Inc.) were sterilized in 70% ethanol and allowed to air dry. Catheters were filled from end to end with bacterial suspensions (10⁹ CFU/ml) in either phosphate-buffered saline (PBS) with 1% tryptone (PBS-T), a 1:1 mixture of PBS-T and mouse urine, a mixture containing 10% PBS-T and 90% mouse urine, or PBS-T supplemented with 165 mM urea or 330 mM urea. Catheters were placed in sterile petri plates and incubated in a humid chamber for 1 to 16 h. After incubation, liquid was wicked from the tubing via capillary action on a paper towel. The biofilms were stained with crystal violet, and the unbound dye was washed away. The crystal violet was dissolved in 30% acetic acid and quantitated spectrophotometrically by reading the absorbance at 595 nm. Data were analyzed using GraphPad Prism software, with statistical significance, considered a P value of <0.05, being determined using Student's t test.

Murine model of CAUTI.

The CAUTI mouse model established by Kadurugamuwa et al. (30) was modified and is outlined below. All experimental animal procedures were performed in accordance with protocols approved by the IACUC at the University of Maryland at College Park. Segments of PE50 catheter material (polyethylene; length, either 1 or 5 cm; outside diameter, 0.965 mm; inside diameter, 0.58 mm; Braintree Scientific, Inc.) were sterilized in 70% ethanol and allowed to air dry. CF-1 female mice (Charles River, Wilmington, MA) weighing 26 to 35 g were anesthetized with ketamine (100 mg/kg of body weight) and xylazine (10 mg/kg) and placed on their backs. The periurethral areas were cleansed with 70% ethanol. For each mouse, a 5-mm catheter piece was loaded onto a sterile 7-cm metal stylet followed by a 5-cm piece of tubing. The stylet was placed in the urethral opening, and the tubing was advanced over the stylet and into the bladder until the 5-mm catheter piece was placed into the bladder. The metal stylet was removed, allowing the 5-cm tubing to serve as a conduit for inoculation of the bladder. Using an insulin syringe, 35 μl of the designated bacterial suspension, resulting in a final inoculum concentration of 10⁶ CFU, was injected transurethrally. The tubing was subsequently removed, while the 5-mm catheter was allowed to remain in the bladder. Data were analyzed using GraphPad Prism software, with statistical significance, considered a P value of <0.05, being determined using one-way analysis of variance (ANOVA).

Collection of human urine.

Urine was collected from volunteers in compliance with the Institutional Review Board at the University of Maryland at College Park. No personal identifiable information was collected about the participants at the time of urine collection. Urine samples were immediately sterile filtered with a 0.22-μm-pore-size filter and frozen at −20°C until use.

SEM.

Samples were obtained either from bacteria grown in vitro in the catheter or from catheters dissected from infected animals in vivo. In vitro, bacteria were grown in catheters with LB broth with and without 0.5 M urea and incubated for 2 days at 30°C. In vivo samples, including bladders and catheters, were dissected from animals infected with the indicated strains. All catheters and bladders were washed with sterile PBS and fixed with 2% glutaraldehyde in PBS overnight at 4°C, postfixed in 1% OsO₄ in PBS, and dehydrated through an ethanol series. The samples were critical point dried, mounted on stubs, and evaporatively coated with gold-palladium alloy. Stubs were mounted in an Hitachi S-4700 scanning electron microscope (SEM) to obtain micrographs.

Fluorescence microscopy.

Wild-type PA14 and PA14 ΔpelD were grown overnight in LB. The overnight culture was diluted 1:100 into LB, and 2 μl was transferred onto a sterile coverslip and covered with LB–1% agarose pads containing 5 mM MgCl₂, 0.001 mg/ml propidium iodide, and either 0 M urea, 0.5 M urea, or 0.1% SDS. Bacteria were incubated for 8 h in a humidified chamber and imaged using a Zeiss Axio Observer.Z1 inverted fluorescence microscope. Phase-contrast microscopy was used to image the bacteria, and eDNA was visualized via propidium iodide staining.

RESULTS

PEL, PSL, and alginate polysaccharides are not required in a murine model of CAUTI.

P. aeruginosa encodes three exopolysaccharide biosynthesis operons, pel, psl and alg operons, that produce PEL, PSL and alginate, respectively. Biofilms formed in vitro require two extracellular polysaccharides, PEL and PSL, which have been synthesized previously (21–23, 26). We hypothesized that one or more of the PSL, PEL, and alginate exopolysaccharides are required for biofilm-mediated CAUTI in vivo. To test this hypothesis, we adapted a mouse model of urinary tract infection utilized in previous studies (30, 35). Mice were catheterized with 5-mm pieces of catheter tubing and subjected to transurethral challenge with P. aeruginosa. Mice were sacrificed at 14 days postinoculation, and the bladder, kidneys, liver, and spleen of each mouse were harvested and homogenized. The homogenates were plated on both LB agar and LB agar supplemented with triclosan, which is selective for Pseudomonas, for determination of the number of CFU. To confirm that the catheterization process itself did not negatively affect the mice, several mice were catheterized in the absence of bacteria. The organs from these mice were visually healthy and were free of bacteria (data not shown). Importantly, neither PA14 nor the PA14 ΔpelD strain was able to colonize mouse tissue in the absence of a catheter after 2 weeks, as determined from the CFU counts (data not shown). This underscores the importance of the catheter as a surface for initial colonization by the bacteria. To determine the requirement for PEL, PSL, and alginate polysaccharides, mice were implanted with catheters and challenged with 10⁶ CFU of PA14, PA14 ΔpelD, PA14 ΔpelA-G, or PA14 ΔpelA-G ΔalgD-A. No nonpseudomonads were recovered at the end of the 2-week CAUTI. The majority of animals exhibited chronic colonization by P. aeruginosa in the bladder after inoculation with PA14 (9/10 animals), PA14 ΔpelD (12/13 animals), PA14 ΔpelA-G (12/12 animals), or PA14 ΔpelA-G ΔalgD-A (7/8 animals) (Fig. 1). Furthermore, the organs harvested from mice inoculated with the PA14 ΔpelD, PA14 ΔpelA-G, and PA14 ΔpelA-G ΔalgD-A strains harbored bacterial numbers similar to those harbored by the organs of mice infected with the PA14 parental strain (Fig. 1A). To confirm these results for the PA14 strain, we also tested another commonly used laboratory strain (strain PAO1) and its isogenic Δpel Δpsl Δalg mutant lacking the exopolysaccharides. Both PAO1 (12/12 animals) and PAO1 Δpel Δpsl Δalg (6/10 animals) colonized the bladder in the murine model of CAUTI, but the average bacterial load of PAO1 Δpel Δpsl Δalg was less than that of the parental PAO1 strain (see Fig. S1A in the supplemental material). Nonetheless, PAO1 Δpel Δpsl Δalg was able to colonize the bladder, indicating that the exopolysaccharides are not absolutely required for biofilm formation in vivo. Of note, all of the kidneys with P. aeruginosa infections were from animals that also had bacteria in the bladder; this suggests that the kidney infections were the result of bacterial dissemination from the bladder. Only a few animals in each inoculation group had detectable numbers of CFU in the spleen and liver (see Fig. S1B and C in the supplemental material), indicating that the murine CAUTI is a model of localized and chronic P. aeruginosa infection that is restricted primarily to the urinary tract. These results demonstrate that P. aeruginosa lacking the exopolysaccharide genes is able to colonize the bladder of mice and indicate that P. aeruginosa establishes CAUTI independently of all three exopolysaccharides.

FIG 1.

P. aeruginosa colonization of catheters during CAUTI is independent of PEL, PSL, and alginate exopolysaccharides. The indicated organs from mice infected with PA14, PA14 ΔpelD, PA14 ΔpelA-G, or PA14 ΔpelA-G ΔalgD-A in the murine model of CAUTI were harvested, homogenized, and plated for counting of the numbers of CFU. Each symbol represents an individual mouse. Bars represent means. Points at 100 CFU per macerated organ represent samples with counts below the limit of detection. ns, differences were not significant, as determined by one-way ANOVA.

SEM of catheters reveals biofilm formation that is independent of extracellular polysaccharide genes in vivo.

We asked whether the bacteria within the infected bladder were located on the catheter, the bladder epithelium, or both surfaces. To test this, bladders and associated catheters were removed from infected mice, sectioned, and processed for scanning electron microscopy (SEM). The interior of the catheter isolated from mice infected with PAO1 had visible bacteria that formed elongated rods, many of which were septated in the interior surface of the catheter (Fig. 2A). We also asked whether strain PA14 ΔpelD/pMMB-PA3702, which lacks both the PEL and PSL polysaccharides, can also colonize the interior of implanted catheters. The plasmid pMMB-PA3702 was used to elevate the levels of cyclic-di-GMP, which enhances the in vitro biofilm phenotype between wild-type and ΔpelD mutant strains (36). We also observed PA14 ΔpelD/pMMB-PA3702 in the interior of the catheter embedded in a matrix material (Fig. 2B). This is in contrast to the uniform short rods that P. aeruginosa forms when grown in biofilms in vitro (see Fig. S2 in the supplemental material), which suggests that P. aeruginosa responds differently to the in vivo environment. The bacteria were encased in a granular matrix, but it was not apparent by SEM whether this matrix was of host or bacterial origin. The external surface of the catheter was devoid of bacteria (see Fig. S3C in the supplemental material). We performed SEM on the bladder epithelium and were unable to detect bacteria on the surface of the bladder (see Fig. S3 in the supplemental material), indicating that the interior of the catheter is the main reservoir of P. aeruginosa. These results show that the primary reservoir of P. aeruginosa during CAUTI is biofilms on the luminal surface of catheters. For the remainder of this study, PA14 and its isogenic pel mutants were used to study the requirement of PEL and PSL during biofilm formation in vitro and in vivo.

FIG 2.

Scanning electron micrographs of catheters isolated from mouse bladders. Catheters were isolated from mice infected with either PAO1 (A) or PA14 ΔpelD/pMMB-PA3702 (B) and visualized by SEM. Asterisks, divided bacteria; arrowheads, filamentous bacteria.

Competition studies reveal that neither PA14 nor PA14 ΔpelD is at a selective advantage in vivo.

Our in vivo studies suggested that PelD is dispensable for biofilm formation on catheters in the mouse. One possibility is that the ΔpelD mutant strain occupies a niche that would be otherwise occupied by the parental PA14 strain during CAUTI. We tested this hypothesis by competing parent PA14 and the isogenic ΔpelD strain in the murine model of CAUTI. Modified versions of both PA14 and PA14 ΔpelD were prepared by incorporation of a constitutive lux marker at the chromosomal ctx phage attachment site (att) of each strain. The PA14-lux and PA14 ΔpelD-lux colonies were easily identifiable due to the bioluminescence resulting from constitutive expression of the lux gene. Bacteria were mixed in a 1:1 ratio in two different combinations: PA14 plus PA14 ΔpelD-lux or PA14 ΔpelD plus PA14-lux. Each combination was tested at 10⁶ CFU in the murine model of CAUTI in groups of five mice, as described above. At 14 days postinoculation, mice were sacrificed and the bladder and kidneys of each mouse were harvested, homogenized, and enumerated by plating serial dilution of the tissue homogenates and counting the numbers of luminescent (Lux⁺) and nonluminescent (Lux⁻) colonies. Colony counts from homogenized bladders and kidneys revealed that the competitive index for PA14 ΔpelD-lux plus PA14 was 0.63 ± 0.42 and that that for PA14 ΔpelD plus PA14-lux was 1.95 ± 1.57 (Fig. 3). These results indicate that there is no competitive disadvantage for PEL and PSL polysaccharide-deficient bacteria in the murine model of CAUTI. Since the ΔpelD mutant can form biofilms on catheters equally well when inoculated alone and during coinfections with PA14, we asked whether the PA14 strain could promote the ability of the ΔpelD strain to participate in mixed biofilms in vitro. Using the same lux-labeled strains in mixed biofilms in vitro, we found that the ΔpelD strain can participate in mixed biofilms when used in a 1:1 ratio with the parental PA14 strain (see Fig. S4 in the supplemental material). The simplest explanation for these results is that the ΔpelD mutant can utilize a product produced by the parental PA14 strain, presumably the PEL polysaccharide, to participate in the biofilm, suggesting that PEL is likely a public good that is shared by pseudomonads in a biofilm. Taken together, these results show that exopolysaccharide-deficient P. aeruginosa can cause CAUTI alone or in combination with wild-type P. aeruginosa.

FIG 3.

PA14 and PA14 ΔpelD do not compete in the murine CAUTI model. Mice were inoculated with either PA14 and PA14 ΔpelD-lux or PA14-lux and PA14 ΔpelD. Each symbol represents the ratio of PA14 ΔpelD/PA14 in either the bladder or kidneys of an individual mouse. Bars represent means.

PA14 ΔpelD and PA14 ΔpelA-G retain phenotype and genotype after CAUTI in a murine model.

Since pel mutants alone can infect mice in the catheter model of infection, we asked whether this ability is due to a heritable suppressor mutation. To test if genetic mutations caused a permanent change, P. aeruginosa strains were isolated after urinary tract infection with PA14, PA14 ΔpelD, PA14 ΔpelA-G, or PA14 ΔpelA-G ΔalgD-A. Each of the isolates was tested for its ability to form biofilm in 96-well polystyrene plates and subsequently genotyped for pelD. Of the 36 PA14 ΔpelD isolates collected, all isolates retained the ΔpelD genotype and were unable to form biofilms. Similarly, 16 PA14 ΔpelA-G mutants, which lacked the entire pel operon, were unable to form biofilms after isolation from mouse and lacked the pelD gene (Table 1). Of the 24 PA14 isolates, all isolates retained the wild-type pelD allele and the ability to form biofilm. Similarly, all 42 of the isolates from PA14 ΔpelA-G ΔalgD-A-infected mice retained both the ΔpelA-G and ΔalgD-A mutations. Our results differ from those obtained for other chronic pseudomonal infections, such as infections in cystic fibrosis (CF) patients, in which mutations are known to occur (37). These results suggest that there is not sufficient selective pressure to alter the pel locus, the alg locus, or biofilm formation in vivo during the 2-week CAUTI.

TABLE 1.

Pseudomonas strains maintain genetic identity after murine UTI

Isolatea	PCR result for pelDb	No. of isolates
		Biofilm positivec	Biofilm negatived
PA14 (24)	WT	24	0
	ΔpelD	0	0
PA14 ΔpelD (36)	WT	0	0
	ΔpelD	0	36
PA14 ΔpelA-G (16)	WT	0	0
	ΔpelD	0	16
PA14 ΔpelA-G ΔalgD-A (42)	WT	0	0
	ΔpelD	0	42

^aStrains were isolated from catheters used for murine UTIs. Numbers in parentheses indicate the number of isolates of each strain tested.

^bIsolates were genotyped using primers specific to the pelD gene, vl491 (5′-AGTTAACGGAGTGGGCCCACACACTGTTCT-3′) and vl494 (5′-AAAGCTTTGTGTTCGGTCATGTCCAGTATCT-3′). WT, wild type.

^cBiofilm-positive isolates had an A₅₉₅ of ≥0.5 and an average A₅₉₅ of 1.35 ± 0.36.

^dBiofilm-negative isolates had an A₅₉₅ of <0.5 and an average A₅₉₅ of 0.09 ± 0.08.

PA14 ΔpelD forms biofilm in vitro in the presence of urine.

Our murine CAUTI studies revealed that infection with PA14 ΔpelD alone resulted in persistent biofilm formation. This suggests that the host environment allows PA14 to bypass the requirement for pelD during biofilm formation in vivo. We reasoned that urine represents one of the major differences between the in vitro biofilm conditions and the conditions in the mouse model. Therefore, we conducted in vitro catheter studies in the presence of mouse urine. To enhance the difference between the PA14 and PA14 ΔpelD strains, we overexpressed the WspR diguanylate cyclase (pMMB-PA3702), which activates the transcription of the pel operon and the production of the PEL polysaccharide. Catheters were filled with suspensions of either of the two strains in the presence of PBS-T, a 1:1 mixture of PBS-T and mouse urine, or a mixture containing 10% PBS-T and 90% mouse urine. The catheters were incubated at 37°C overnight, and bacterial biofilm formation was quantified by the crystal violet assay. PA14/pMMB-PA3702 formed biofilms on catheters in PBS-T (absorbance at 595 nm [A₅₉₅], 0.17 ± 0.01), while the ΔpelD/pMMB-PA3702 strain did not (A₅₉₅, 0.0035 ± 0.001) (Fig. 4A). In the presence of 50% mouse urine, PA14 biofilm formation was similar to that in PBS-T (A₅₉₅, 0.177 ± 0.001). However, the A₅₉₅ for the ΔpelD strain was 0.023 ± 0.009, a 6.6-fold increase compared to that for PA14 ΔpelD biofilm formation in the presence of PBS-T alone. PA14 ΔpelD biofilm formation was further increased in the presence of 90% urine (A₅₉₅, 0.075 ± 0.004; a 21.6-fold increase over the level in PBS-T). Interestingly, biofilm formation by PA14 decreased in the presence of 90% urine (A₅₉₅, 0.069 ± 0.001; a 2.5-fold decrease from the level in PBS-T). Despite this reduction, PA14 still formed biofilm in urea, whereas in the absence of urea the PA14 ΔpelD strain did not.

FIG 4.

Mouse and human urine induce biofilm formation by PA14 ΔpelD. (A) Quantification of the crystal violet eluted from catheters after staining for colonization by PA14/pMMB-PA3702 or PA14 ΔpelD/pMMB-PA3702 in the presence of PBS-T with or without the indicated amounts of mouse urine. (B and C) Quantification of PA14 ΔpelD/pMMB-PA3702 (B) or PA14/pMMB-PA3702 (C) biofilms in PBS-T with or without the indicated amounts of human urine. (D) PA14 ΔpelD biofilm formation in LB or artificial urine medium (AUM). Statistical significance (*, P < 0.05) was determined by using Student's t test. Abs, absorbance.

To determine if the induction of PA14 ΔpelD biofilm development applies to human urine, we conducted a biofilm assay in 1× PBS-T supplemented with 25%, 50%, or 75% human urine. We performed an in vitro biofilm experiment using PA14 and PA14 ΔpelD strains. Similar to the results from mouse urine, the A₅₉₅ of PA14 ΔpelD/pMMB-PA3702 in 50% (A₅₉₅, 0.20 ± 0.05) and 75% (A₅₉₅, 0.20 ± 0.03) human urine was enhanced compared to that for the PBS-T control (A₅₉₅, 0.09 ± 0.03) (Fig. 4B). Unlike mouse urine, the addition of human urine enhanced PA14/pMMB-PA3702 biofilm formation. With the addition of either 25% or 50% human urine, there was a 3-fold increase in the A₅₉₅ (3.73 ± 0.25 and 3.30 ± 0.23, respectively) compared to that for the PBS-T control (A₅₉₅, 1.10 ± 0.05). Furthermore, the addition of 75% human urine resulted in a 4-fold increase in biofilm formation (A₅₉₅, 4.39 ± 0.44) (Fig. 4C). Urine is a complex medium consisting of more than 60 different compounds that compose 99% of the solutes of urine (38). We tested the ability of chemically defined artificial urine medium (AUM) to induce biofilm formation by the PA14 ΔpelD strain (39). The level of induction of biofilm formation by PA14 ΔpelD in AUM (A₅₉₅, 0.68 ± 0.23) increased 6-fold compared to that with growth in LB (A₅₉₅, 0.11 ± 0.05) (Fig. 4D). On the basis of these data, we conclude that there are components in both human and mouse urine that alter the program of in vitro biofilm development to one in which biofilm is formed independently of exopolysaccharides.

Urea promotes polysaccharide-independent biofilms in vitro.

In order to identify a specific component of urine that induces PA14 ΔpelD/pMMB-PA3702 biofilm formation, we performed biofilm assays in LB supplemented with biologically relevant amounts of six of the most concentrated nitrogen- and carbon-containing compounds identified in human urine. These compounds included urea (500 mM), citric acid (2 mM), creatinine (17.7 mM), glycine (49.4 mM), histidine (1.2 mM), and glutamic acid (1.1 mM). Uric acid (1.7 mM) was not included in these studies, as it is insoluble in LB. Our biofilm results showed that urea was the only compound able to significantly enhance biofilm formation by the ΔpelD mutant strain (A₅₉₅, 0.43 ± 0.06) compared to the level of biofilm formation for the control (A₅₉₅, 0.26 ± 0.06) (Fig. 5A). Interestingly, PA14/pMMB-PA3702 biofilm formation decreased in the presence of 0.5 M urea (A₅₉₅, 0.28 ± 0.001) compared to the level of biofilm formation for the control (A₅₉₅, 0.99 ± 0.11) (Fig. 5B). Although the level of biofilm formation was reduced, the parental PA14 strain still formed biofilm in urea, whereas the PA14 ΔpelD strain did not form biofilm in the absence of urea. We determined the concentration of urea in the pooled mouse and human urine to be 786 mM and 243 mM, respectively (see Materials and Methods in the supplemental material). To determine the dose-response of PA14 ΔpelD to urea, we repeated our biofilm assays with 0.165 M urea or 0.330 M urea in catheters. In PBS-T, PA14 formed biofilm on the catheters (A₅₉₅, 0.53 ± 0.03), while the ΔpelD strain did not (A₅₉₅, 0.011 ± 0.007) (Fig. 5C). As was the case with the mouse urine, increasing concentrations of urea reduced the levels of PA14 biofilm formation on catheters to A₅₉₅s of 0.52 ± 0.05 and 0.441 ± 0.009 in the presence of 0.165 M and 0.330 M urea, respectively (Fig. 5C). Biofilm formation by the ΔpelD strain was enhanced by the presence of urea. In the presence of 0.165 M and 0.330 M urea, the A₅₉₅ readings for PA14 ΔpelD were 0.0071 ± 0.0004 and 0.25 ± 0.07, respectively. These results indicate that urea, the main nitrogenous waste component in mammalian urine, alters the program of biofilm formation by P. aeruginosa to be exopolysaccharide independent.

FIG 5.

Urea induces biofilm formation by PA14 ΔpelD. Quantification of crystal violet biofilm formation assays for colonization of either PA14/pMMB-PA3702 (A) or PA14 ΔpelD/pMMB-PA3702 (B) in the presence of LB or LB supplemented with urea, citric acid, creatinine, glycine, histidine, or glutamic acid. (C) Biofilm formation assays with catheters inoculated with either PA14/pMMB-PA3702 (closed bars) or PA14 ΔpelD/pMMB-PA3702 (open bars) in PBS-T with or without urea at the indicated concentrations. Statistical significance (*, P < 0.05) was determined by using Student's t test.

Urea promotes polysaccharide-independent biofilms on catheters in vitro.

Results from the crystal violet assays for PA14 and PA14 ΔpelD suggested that the ΔpelD mutant bacteria can form biofilm in the presence of urine and urea. We asked whether these in vitro biofilms on catheters are morphologically similar when visualized by SEM. Catheters incubated in vitro with either PA14 or PA14 ΔpelD in the presence or absence of 0.5 M urea were washed, fixed, and evaporatively coated, and images were revealed by SEM. Representative micrographs are shown in Fig. 6. As expected, the catheters inoculated with wild-type strain PA14 in PBS-T contained a layer of bacteria that covered most of the surface of the catheter (Fig. 6A). Greater magnification of the image revealed that wild-type PA14 had extracellular filaments that connected the individual cells to form a biofilm (Fig. 6A, inset). Also as expected, PA14 ΔpelD could not form biofilm on catheters in PBS-T, just as the mutant was unable to form biofilm on plastic and glass materials in vitro (Fig. 6B). In contrast, treatment in the presence of urea caused a reduction in the coverage of the PA14 biofilm on catheters, which agrees with the results from the crystal violet assays (Fig. 6C). Urea treatment enabled the PA14 ΔpelD mutant to form biofilm on the catheter surface (Fig. 6D). Increased magnification showed that the biofilm formed by the PA14 ΔpelD mutant in the presence of urea appeared to have similar filamentous material that connected the bacteria (Fig. 6D, inset). Together, these results support the suggestion that urea alters the morphology of P. aeruginosa biofilm to a heterogeneous form that is independent of exopolysaccharides.

FIG 6.

Scanning electron microscopy reveals biofilm formation by PA14 ΔpelD/pMMB-PA3702 in the presence of urea. PA14/pMMB-PA3702 (A and C) and PA14 ΔpelD/pMMB-PA3702 (B and D) were visualized after incubation for 16 h on catheter tubing in the absence or presence of 0.5 M urea. (Insets) Photos taken at higher magnification.

Some clinical isolates of P. aeruginosa from human urinary tract infections demonstrate a mucoid phenotype (40, 41). We asked whether urine or urea would induce the production of alginate in vitro. We quantitated the amount of uronic acids present in cultures grown with and without added urine or urea. We found that neither the parental PA14 nor the PA14 ΔpelD mutant make detectable levels of alginate in response to urea (see Table S2 in the supplemental material). We then asked if urine or urea would affect alginate production from a mucoid clinical isolate (CF27). We found that urea and urine inhibited the production of alginate from CF27, suggesting that the biofilms formed during CAUTI are occurring independently of alginate biosynthesis (see Table S2 in the supplemental material). These results corroborate our findings that PA14 ΔpelA-G ΔalgD-A, similar to the parental PA14 strain, is capable of causing CAUTI.

Urea effects on biofilm formation by PA14 and PA14 ΔpelD are mediated by cell death.

One possible explanation for the ability of the ΔpelD mutant bacteria to form biofilm in the presence of urea is that some cells lysed and released eDNA. To test this possibility, we placed either PA14 or the ΔpelD mutant under agarose pads containing propidium iodide to stain eDNA. In the absence of additional chemicals, neither PA14 nor PA14 ΔpelD released eDNA after 8 h of growth under an LB agarose pad (Fig. 7A and B). When placed under an LB agarose pad containing 0.5 M urea, the urea had little effect on P. aeruginosa growth and replication. Exposure to 0.5 M urea led to cell rounding in a subset of cells and their inability to exclude propidium iodide in both PA14 and the ΔpelD mutant (Fig. 7A and B and insets). Quantitation of these images revealed that 0.07% of PA14 cells and 0.28% of PA14 ΔpelD cells had rounded or lysed. When placed under an LB agarose pad with 0.1% SDS, P. aeruginosa also grew to high density, indicating that not all cells were lysed. Nonetheless, SDS led to the release of eDNA in a subset of cells for both the PA14 and ΔpelD strains (Fig. 7A and B). We asked if the lysis of pseudomonads by urea is due to inhibition of growth. We performed growth curves of PA14 and PA14 ΔpelD in the presence of various concentrations of urea and urine. We found that the bacteria could reach the same numbers of CFU/ml in overnight cultures. Although the final number of CFU/ml was not affected in overnight cultures, higher concentrations of urea (0.4 and 0.5 M) reduced the growth rate of PA14 (see Fig. S5A to C in the supplemental material). The reduction in growth was not due to the urease genes, as transposon mutants with mutations of ureB, ureC, ureE, and ureG had growth curves similar to those for PA14 (see Fig. S5D in the supplemental material). Together, these results support the hypothesis that the ΔpelD mutant lacking extracellular polysaccharide can be induced to form biofilm via release of eDNA from a subpopulation of rounded and lysed cells.

FIG 7.

Urea causes the release of DNA from a subset of P. aeruginosa cells. PA14 (A) or PA14 ΔpelD (B) overnight cultures were subcultured 1:100 and placed under agarose pads containing LB with the indicated concentrations of urea or SDS. Samples were incubated for 8 h at 37°C in a humidified chamber. Cells were imaged by phase-contrast microscopy, and eDNA was detected by propidium iodide staining (red). (Insets) Magnified images of rounded cells (arrowheads) that stain positive with propidium iodide.

eDNA allows biofilm formation that is independent of exopolysaccharides.

To test our hypothesis that eDNA is a matrix component that contributes to biofilm formation in the absence of PEL and PSL exopolysaccharides, we asked if DNase I treatment reduces the level of urea-induced biofilm formation. PA14 and PA14 ΔpelD were grown in the presence of 0.5 M urea for 48 h to develop a biofilm. These biofilms were washed and treated with buffer or buffer containing DNase I. Following DNase I treatment, biofilms were washed a second time and immediately stained with crystal violet (Fig. 8A). Significantly less pellicle was left in tubes treated with DNase I than in the control tubes for both PA14 and PA14 ΔpelD. This suggests that eDNA is a key component in the matrix that maintains biofilm structure. However, exopolysaccharides still play a role in these biofilms formed in the presence of urea, as PA14 biofilms were more robust than PA14 ΔpelD biofilms (Fig. 8B). While DNase I treatment significantly reduced the levels of biofilm formation for both PA14 and PA14 ΔpelD, there were still residual biofilms remaining (Fig. 8B), suggesting that there are additional factors other than eDNA that are contributing to biofilm formation in the presence of urea. These results indicate that eDNA from lysed bacteria is a major component of urea-mediated biofilms in vitro which allows PA14 ΔpelD mutants to form biofilms independently of both PEL and PSL exopolysaccharides. In addition, exposure to urea induces the parental PA14 strain to produce a DNase I-sensitive biofilm.

FIG 8.

DNase I treatment reduces the amount of biofilms formed in the presence of urea. (A) PA14 and PA14 ΔpelD biofilms in borosilicate tubes were treated with 5 μg of DNase I or buffer as a control at 37°C for 90 min and stained with crystal violet. (B) Quantification of eluted crystal violet after DNase I or control treatment. Statistical significance (*, P < 0.05) was determined by using Student's t test.

DISCUSSION

Exopolysaccharides in P. aeruginosa biofilm-mediated infections.

Previous investigations of the role of P. aeruginosa biofilms during infection were assessed by two primary methods: (i) analysis of tissue and isolates from infected patients and (ii) the use of animal infection models. Wound infection biopsy specimens from patients were stained with P. aeruginosa- and Staphylococcus aureus-specific peptide-nucleic acid stains and analyzed by confocal microscopy (42). These studies revealed that S. aureus occupies the wound surface, while P. aeruginosa resides deeper within the tissue beneath the wound site (42). As both P. aeruginosa and S. aureus are detected in microcolonies that resemble biofilms, these observations indicate that biofilm formation may be relevant in wound infections. Recent studies that sequenced bacteria isolated from cystic fibrosis (CF) patients demonstrated that genetic mutations in P. aeruginosa lead to defective quorum signaling, increased production of PEL and PSL polysaccharides, and the appearance of small-colony variants (SCVs) (18, 37, 43). In previous studies and our current study, we and others have shown that P. aeruginosa forms biofilms during CAUTIs (30). Taken together, these studies show that P. aeruginosa biofilms participate in infection at a number of different sites, including wounds, the lungs of CF patients, and the urinary tract.

The question of the matrix components involved in P. aeruginosa biofilm formation during infection has been addressed using a number of animal model systems. These studies reveal a diverse role for exopolysaccharide in different infection models. In the murine subcutaneous catheter implant model, the ΔyfiR mutant, which has an SCV phenotype and promotes biofilm formation via induction of the pel and psl genes, had a competitive disadvantage in the catheter and surrounding tissue compared to the PAO1 parental strain (44). The authors attributed this defect to the reduced growth rate of the bacteria. The direct contribution of the pel and psl genes in an animal infection model was also assessed. In an acute pneumonia model of competitive infection, the Δpsl mutant had a 3-fold decrease in the competitive index compared to that for the parental PAO1 strain at the 12-h time point, suggesting that extracellular polysaccharides can contribute to acute and chronic infections (45). In a Drosophila model of oral P. aeruginosa infection, both the PAO1 parent and the ΔrsmA mutant, which is constitutively induced for pel and psl operons, formed biofilm in the fly crop (46). However, the ΔrsmA mutant had a slower progression of lethality than the parental PAO1 strain, suggesting reduced virulence (46). In the Drosophila model, the PEL polysaccharide is required for biofilm formation, as a pelB insertion mutant did not form microcolonies and larger aggregates (46). However, mutations in the gacAS, retS, and rsm pathway prevent the expression of the type III secretion system, a potent system for the delivery of cytotoxic effector molecules for acute infections, suggesting that biofilm formation and lethality are reciprocally regulated by this pathway (46, 47). In a chinchilla model of otitis media, both the parental PAO1 strain and an isogenic ΔwspA mutant with constitutive overproduction of cyclic di-GMP and induction of pel and psl genes colonized the bullae (48). However, the ΔwspA mutant strain showed delayed pathogenesis, suggesting that constitutive biofilm formation reduced the progression of disease symptoms. Further studies that compared the ΔwspA strain with ΔwspA Δpsl and ΔwspA Δpel strains revealed that neither polysaccharide alone altered the outcome of the infection (48). However, eDNA was visualized by staining with propidium iodide, suggesting that eDNA may also serve as a matrix component during the infection (48). Similarly, studies in a Drosophila model of infection have revealed another alternative pseudomonal biofilm pathway that is regulated by the PprAB two-component regulatory system (49). Our studies using P. aeruginosa PA14 and PAO1 strains lacking PEL, PSL, and alginate polysaccharides in the murine model of CAUTI suggest that these exopolysaccharides are dispensable for CAUTIs, as eDNA can compensate for these polysaccharides during biofilm formation. Our results indicate that P. aeruginosa utilizes a distinct program to form biofilms during CAUTI of mammalian hosts and that eDNA can act as the primary matrix component of these biofilms. Furthermore, P. aeruginosa biofilms form in response to different local host conditions, depending on the site of infection.

Genetic alterations in biofilms formed in vitro and in vivo.

Some intriguing possible explanations for the ability of the Δpel mutant to participate in biofilm formation include either the acquisition of the necessary genes in trans via horizontal gene transfer or the acquisition of a suppressor mutation at a second site. Horizontal gene transfer is a likely possibility, as P. aeruginosa is known to have a large number of regions of genome plasticity on the chromosome (50). Mutagenesis of P. aeruginosa in chronic infections, such as those in cystic fibrosis patients, is well-known and has been demonstrated in sequencing studies of strains longitudinally collected from a single patient (37). In the mixed biofilm assays, we found that the PA14 strain maintains its ability to form biofilm in vitro, while the ΔpelD strain retains its inability to form biofilm. These results suggest that the ability of the ΔpelD mutant to participate in a mixed biofilm is likely due to the expression of a surface protein that specifically binds to the PEL polysaccharide produced by the parental PA14 strain. Bacterial isolates from the mice infected with either PA14 or the ΔpelD strain retained their respective pel locus and their ability or inability to form biofilm in vitro. These results suggest that during a 14-day infection, P. aeruginosa has not begun to mutate the pel genes, suggesting that these genes are not under negative selection during CAUTI. Extending the infection time may reveal whether P. aeruginosa accumulates mutations in chronic, non-CF patient infections. Future studies in other chronic models of P. aeruginosa infection will reveal whether the PEL and PSL polysaccharides are required for biofilm infections at other anatomical locations. In addition to the requirement for the PEL and PSL polysaccharides in vitro, our results also suggest that P. aeruginosa mutants that are unable to produce PEL and PSL can nonetheless participate in biofilm formation by wild-type bacteria. This raises an important question as to the evolutionary pressure to prevent P. aeruginosa from cheating and losing the pel genes. Likely, there are environmental conditions in which P. aeruginosa forms planktonic single cells that have to be able to form biofilm without assistance from other bacteria. Those cells that lost the ability to form biofilm from individual planktonic cells would be at a great disadvantage. Another alternative is that participation in the biofilm alone is not sufficient to confer the same physiological advantages that PEL polysaccharide-producing bacteria have. Future experiments will delineate between these possibilities.

Urea-mediated exopolysaccharide-independent biofilms.

The urinary tract is a dynamic environment with physical, chemical, and cellular changes that can influence CAUTI. The bladder alone must expand and contract in order to store and expel urine. The concentrations of components that make up urine are constantly changing in response to the diet and hydration level of the host over the course of a day (51). The cells of the bladder epithelium are repeatedly being turned over, a process that occurs approximately every 6 weeks in guinea pigs (52). Also, during infection, immune cells, such as neutrophils, can infiltrate into the lumen of the bladder. Here, we demonstrate that human urine can induce P. aeruginosa biofilm formation and is an important environmental factor in P. aeruginosa-mediated CAUTI. Furthermore, to determine a component in urine that enhances biofilm formation in the absence of exopolysaccharide, we individually tested the most abundant nitrogen- and carbon-containing compounds detected in human urine. This does not exclude the possibility that other nonnitrogen or noncarbon compounds enhance P. aeruginosa biofilm formation. Of the components tested, only urea, the primary solute of urine, was able to induce P. aeruginosa biofilm formation in the absence of exopolysaccharides. Urea may trigger a number of possible changes to P. aeruginosa to promote biofilm formation by altering the existing cell envelope or production of additional extracellular matrix. Here we show that urea induces the release of eDNA from a subset of bacteria to enhance biofilm formation by both parental and exopolysaccharide-deficient P. aeruginosa strains. These results suggest that P. aeruginosa has a genetic program to respond to host urine that permits the formation of an alternative eDNA-based biofilm. What is the basis for this urea-dependent activation of DNA release? Several studies have established that quorum sensing induces eDNA release (53–55). In one study, PAO1 released increasing amounts of eDNA when the culture reached stationary phase (55). Supporting the requirement of the quorum-sensing pathway, addition of the quorum-sensing inhibitor furanone C30 prevented eDNA release (55). PAO1 ΔlasI, ΔrhlI, and ΔpqsA mutants released less eDNA than the parental PAO1 strain and formed thinner biofilms that lacked characteristic mushroom formations (55). Furthermore, the reduction in eDNA release by these quorum-sensing mutants could be complemented by supplementing the medium with exogenously added acylated homoserine lactones, N-3-oxo-dodecanoyl-l-homoserine lactone, and N-butyryl-l-homoserine lactone (54, 55). Together, these studies demonstrate that the quorum-sensing system activates the release of eDNA to promote biofilm formation in vitro. In our system, urine and urea can trigger eDNA release through the quorum sensing-regulated process or a novel mechanism that is independent of quorum regulation. In future studies, we will resolve the influence that urea has on quorum sensing and in turn establish the role that quorum sensing plays during CAUTI.