Annals of Clinical Microbiology and Antimicrobials Identification of Multi-drug Resistant Pseudomonas Aeruginosa Clinical Isolates That Are Highly Disruptive to the Intestinal Epithelial Barrier

Background: Multi-drug resistant Pseudomonas aeruginosa nosocomial infections are increasingly recognized worldwide. In this study, we focused on the virulence of multi-drug resistant clinical strains P. aeruginosa against the intestinal epithelial barrier, since P. aeruginosa can cause lethal sepsis from within the intestinal tract of critically ill and immuno-compromised patients via mechanisms involving disruption of epithelial barrier function.


Background
The human opportunistic pathogen, Pseudomonas aerugi-nosa, is a major cause of infectious-related mortality among the critically ill patients, and carriers the highest case fatality rate of all gram-negative infections [1]. Although the lungs have been traditionally considered to be a major site of P. aeruginosa infection among critically ill patients, a significant number of these infections arise as a result of direct contamination of the airways by the gastrointestinal flora or by hematogenous dissemination from the intestine to the lung parenchyma [2,3]. Yet even in the absence of established extraintestinal infection and bacteremia, the presence of highly virulent strains of P. aeruginosa within the intestinal tract alone can be a major source of systemic sepsis and death among immuno-compromised patients [4,5]. Extensive studies on the endemicity and prevalence of P. aeruginosa in the critically ill patients have identified the intestinal tract to be the single most important reservoir for this pathogen in cases of severe life-threatening sepsis [6,7]. Work from our laboratory has demonstrated that a major mechanism of the lethal effect of intestinal P. aeruginosa lies in its ability to adhere to and disrupt the intestinal epithelial barrier [8].
Within as little as 3 days in an intensive care unit, the feces of more than 50% of patients will culture positive for P. aeruginosa with up to 30% of these strains being antibiotic resistant [6]. In such patients, intestinal colonization by P. aeruginosa alone has been associated with a 3-fold increase in mortality in critically ill patients [4]. In fact the importance of intestinal P. aeruginosa as a cause of mortality in critically ill patients was recently demonstrated by a randomized prospective study in which selective antibiotic decontamination of the digestive tract (SDD) in critically ill patients with oral non-absorbable antibiotics decreased mortality associated with a decrease in fecal P. aeruginosa [9].
How multi-drug resistant (MDR) P. aeruginosa clinical isolates behave against the human intestinal epithelium is unknown. Therefore the purpose of this study was to determine the ability of MDR P. aeruginosa to disrupt epithelial integrity of Caco-2 monolayers and to correlate these findings to other relevant virulence features of P. aeruginosa including adhesiveness, motility, ability to form biofilm, and the presence of specific type III secretion related genes exoU and exoS.

Bacterial isolates
Under IRB protocol #11646B, University of Chicago, 35 strains of P. aeruginosa were consecutively obtained from the clinical microbiology laboratory from those selectively screened for gentamicin (Gm) resistance. We initially screened consecutive P. aeruginosa isolates that were resistant to Gm since Gm resistance has been shown to be the most common feature of MDR P. aeruginosa [10]. Among the 35 strains, three (# 3, 5, and 32) lost their resistance to Gm and one (#24) was re-identified not to be P. aeruginosa on subsequent culture. Therefore 31 clinical strains were available for phenotype and genotype analysis. Most isolates identified as P. aeruginosa were oxidase positive, hydrolyzed acetamide and arginine, oxidized glucose, and grew on cetrimide agar. Remaining isolates were identified by the Vitek 2 system (bioMérieux, Inc. Durham, NC). Additionally, isolates were verified by amplification of 16S DNA using primers forward 5'-GGACGGGTGAGTAATGCCTA-3' and reverse 5'-CGTAAGGGCCATGATGACTT-3', and genome DNAs of clinical isolates as templates. Susceptibility testing was performed by testing on the Vitek 2 or by disk diffusion. Susceptibility results were interpreted using Clinical Laboratory Standards Institute (CLSI) guidelines. Single colonies were picked up from Columbia SB agarized plates (Beckton Dickinson, Cockeysville, MD), grown in Pseudomonas broth containing Gm, 50 µg.ml -1 and kept at -80°C as frozen stocks containing 8% glycerol. The isolates were routinely subcultured from frozen stocks on Pseudomonas isolation agar (PIA) containing Gm, 50 µg.ml -1 . P. aeruginosa strains PAOI, ATCC 27853, PA103, and the environmental isolates PA190 and PA180 [11][12][13] were used as reference strains.

DNA fingerprint analysis
The clonality of P. aeruginosa isolates was determined using the random amplified polymorphic DNA (RAPD) PCR fingerprinting, described previously [14][15][16]. Primers 208 (5'-ACGGCCGACC-3') and 272 (5'-AGCGGGCCAA-3') were synthesized and used in PCR amplifications. Intact bacteria were used as a source of template chromosomal DNA. The following protocol was used: 45 cycles of 1 min at 94°C, 1 min at 45°C and 1 min at 72°C. After the last cycle, samples were maintained at 72°C for 10 min. The resulting amplified DNA fragments were separated on agarose gels (0.8%, w/v) containing ethidium bromide (0.5 µg.ml -1 ) and visualized using UV radiation. Fingerprints were considered distinct if they differed by at least three bands.

Human epithelial cells and transepithelial resistance (TER) assay
The Caco-2bbe (brush border-expressing) cell line was used in bacterial-cell culture experiments. Caco-2 cells were grown in 0.3 cm 2 transwells (Costar) in HEPES buffered (15 mM) DMEM media containing 10% FBS for 20 days, and electrophysiological measurements were done using agar bridges and Ag-AgCl-calomel electrodes and a voltage clamp (University of Iowa Bioengineering, Iowa City, IA) as previously described [17]. Fixed currents of 50 µA were passed across Caco-2 monolayers, and transepithelial resistance (TER) was calculated using Ohm's law. Fluid resistance was subtracted from all values. In order to assess the disrupting ability of P. aeruginosa strains against Caco-2 monolayers, overnight culture was added to the apical well (volume = 200 µl) to achieve a final bacterial concentration of ~10 7 CFU/ml. Media from the apical wells was then quantitatively cultured on PIA plates to determine the final bacterial count. Caco-2 monolayers were co-incubated with bacteria for up to 8 hours at 37°C, 5% CO 2 , and TER was measured each hour. All experiments were performed in triplicate.

Swimming motility
Swimming assay was performed as previously described by Rashid and Kornberg [18]. Briefly, swim plates prepared by using of 1% tryptone, 0.5% NaCl and 0.3% (wt/ vol) agarose, were inoculated with bacteria using a sterile toothpick. The plates were then wrapped to prevent dehydration and incubated at 37°C, overnight. The ability to swim was assessed by the radius of colony. All experiments were performed in triplicate.

Twitching motility
Twitching motility was determined by the method of Rashid and Kornberg [18]. Fresh prepared and briefly dried twitch plates (Tryptic soy broth solidified with 1% (wt/vol) Difco granulated agar) were stab inoculated with a sharp toothpick into the bottom of the Petri dish. After incubation at 37°C for 24 h, the halo zone of growth at the interface between the agar and the polystyrene surface was measured. All motility experiments were performed in triplicate.

Ability to form biofilm
Biofilm formation was assayed as described with modifications [19]. Briefly, P. aeruginosa strains were grown in 96-well plates in M63 supplemented with 0.5% casamino acids and 0.2% glucose. Plates were incubated at 37°C under mild shaking at 50 rpm (C24 Incubator Shaker, New Brunswick Scientific, Edison, NJ) for 8 hrs. The wells were then rinsed thoroughly with water and the attached material was stained with 0.1% crystal violet, washed with water, and solubilized in ethanol. Solubilized fractions were collected and absorbance measured at 550 nm with a Plate Reader. All experiments were performed in triplicate.

Adhesiveness
Caco-2 cells were grown to confluence in 24-well plates using HEPES-buffered DMEM media containing 10% fetal bovine serum. Overnight cultures of P. aeruginosa were added to the apical side of Caco-2 cells to a final concentration of 10 7 CFU/ml and co-incubated for 1 hour at 37°C, 5% CO 2 . Following the one hour incubation, the media was removed and ten-fold dilutions were plated on PIA plates to quantify non-adherent bacteria. Wells were then washed with a continuous flow of 35 ml of PBS. A final single washing with 200 µl was diluted and plated on PIA to quantify the final amount of remaining non-adher-ent bacteria. Caco-2 cells were then trypsinized with 200 µl Trypsin-EDTA (Gibco), incubated for 20 min at 37°C, 5% CO 2 , and lysed with 400 µl of a lysis mixture (PBS, EDTA 10 mM, Triton X-100 0.25%) [20] added directly to the trypsinized Caco-2 cells. The cells were vigorously pipetted for one minute, and released bacteria were plated on PIA to quantify adherent cells. The proportion of bacterial cells adhering to Caco-2 cells was calculated as (adherent cells -cells in last washing)/non-adherent + adherent cells. All experiments were performed in triplicate.

Effect of exposure of MDR P. aeruginosa clinical isolates to Gm on growth rate
Overnight culture of P. aeruginosa clinical isolate #1 was diluted as 1:100 in fresh M63 media supplemented with 0.5% casamino acids and 0.2% glucose and grown for 2 hours. After that, culture was spitted for control (no Gm) and Gm-variant that was added by Gm to a desirable concentration. 300 µl aliquots (in triplicates) were loaded in 96-well plate, and absorbance at OD550 nm was measured dynamically during growth at 37°C, 200 rpm. All experiments were performed in triplicate.

Cytotoxicity assay
Caco-2 cells were grown to confluence in 96-well plates, and inoculated apically by P. aeruginosa to the final concentration of 10 7 CFU/ml. Cells were incubated at 37°C, 5% CO 2 , for 8 hours, and released lactate dehydrogenase was determined by CytoTox 96 assay (Promega). All experiments were performed in triplicate.

Statistical analysis
Statistical analysis of the data was performed using Student t-test. Regression analysis was performed using Sigmaplot software.

Morphological and demographic analyses of MDR P. aeruginosa clinical isolates
Morphological and demographic data are displayed in Table 1. P. aeruginosa strains were consecutively collected based on their resistance to gentamicin (Gm), however most clinical isolates displayed multiple antibiotic resistances to various antibiotics clinical used against P. aerugi- nosa. Most strains were obtained from sputum and tracheal aspirates while few were from tissues and urine.
Significant variation was noted in colony morphology among the various strains. Environmental strains PA190 and PA180 were also tested for antibiotic resistance. Results indicated that PA190 was sensitive to all of the antibiotics routinely used for P. aeruginosa infection, whereas PA180 was resistant to Gm.

RAPD fingerprinting of consecutively obtained MDR P. aeruginosa clinical isolates
A total of 31 P. aeruginosa clinical isolates were typed by RAPD analysis with primers 208 (Fig. 1A) and 272 (Fig.  1B) [15]. RAPD fingerprints demonstrated that most clinical strains were of distinct RAPD type. More detailed demographic analysis of strains with similar RAPD revealed that strains 13 and 14 (G13) were from a single patient, strains 30 and 31 (G30) were also from a single patient, and 34 and 35 (G34) were also from a single patient. RAPD fingerprint G20 was similar for strains 4, 20, 21, and 25-28. All of these strains were obtained from specimens of tracheal aspirate, urine, and Foley catheter urine from the same patient during a 4 month period. As such, the total 31 clinical isolates contained 22 different genotypes.

Effect of multi-drug resistant (MDR) clinical isolates of P. aeruginosa on transepithelial resistance (TER) of Caco-2 monolayers
Among clinical isolates in our study, three isolates, #12, #22, and #23 showed resistance to Gm only, and two isolates, #18 and #21 showed resistance to only two antipseudomonas antibiotics (Table 1). Since multi-drug resistance is generally defined as resistance to three or more antimicrobial agents [10], we did not include these strains in any further experiments. Strains 13 and 14, 30 and 31, 34 and 35 were found to be repeat isolates based on RAPD analysis and demographic data; therefore, strains 14, 31, and 34 were not included in any further experiments.
The effect of MDR P. aeruginosa clinical isolates on TER of Caco-2 cells following apical inoculation is summarized in Figure 2. Dynamic tracking of TER following apical exposure of Caco-2 cells to P. aeruginosa ( Fig. 2A) (Fig. 2B) demonstrated that the strains 1, 13, and that of RAPD type G20 induced a rapid and profound decrease in TER similar to the highly cytotoxic strain PA103 [21]. Three isolates, 29, 7, and 15 had significant yet moderate effect on TER similar to the antibiotic sensitive reference strains ATCC 27853, PA01, and PA190. The remaining strains showed a minimal to negligible effect on TER as did the Gm R environmental isolate, PA180. Strain #1 was found to be most virulent strain based on the TER response of Caco-2 cells. TER decreased following apical exposure to as little as 10 3 CFU/ml (Fig. 2C) suggesting a profound ability of the organism to disrupt epithelial barrier function.

Adherence properties, motility patterns, and biofilm formation in relation to the epithelial barrier-disrupting phenotype
Regression analysis revealed that adherence (Fig. 3A) and swimming motility (Fig. 3B) significantly correlated with the TER changes in Caco-2 cells induced by MDR P. aeruginosa (r = 0.88, P < 0.0001, r = 0.57, P < 0.01, respectively). There was no correlation however between TER changes and twitching motility (r = 0.44) (Fig. 3C), or biofilm formation (r = 0.42) (Fig. 3D). High swimming motility and adherence to Caco-2 cells were the main phenotypic features of MDR barrier-disruptive strains 1, 13, and strains of G20 RAPD fingerprint. As a group, strains with a minimal effect on TER were characterized as having attenuated adherence, motility, and biofilm formation although several strains with a minimal effect on TER did display high motility behavior suggesting that motility alone is not predictive of the virulence of MDR P. aeruginosa against the intestinal epithelium.

Effect of exposure of MDR P. aeruginosa to Gm on growth rate
Strains #1, 13, and those of G20 RAPD genotype, the most virulent in terms of their effect on TER were tested for their ability to grow in the presence of Gm. We found that as much as 50 µg.ml-1of Gm had no effect on the growth of strains 13 and G20 RAPD genotype strains (data not shown), whereas strain #1 grown in the presence of Gm showed a dose-dependent stimulation (10-20 µg.ml -1 ) of growth (Fig. 4A). Dynamic tracking of strain #1 exposed  to 20 µg.ml -1 of Gm demonstrated this effect to be greatest during the exponential phase of growth (Fig. 4B).

Cytotoxicity of MDR P. aeruginosa clinical isolates, correlation with exoU/exoS genotype
The cytotoxic effect of the various clinical isolates following 8 hours of bacterial exposure is shown in Figure 5. Results demonstrated that most MDR clinical isolates with barrier-disruptive phenotypes harbored the exoU gene (except strain #33) and displayed cytotoxicity against Caco-2 monolayers. Clinical isolates harboring the exoS gene were not cytotoxic to Caco-2 cells.

Effect of MDR P. aeruginosa clinical isolates on the intestinal epithelial barrier
Numerous reports have documented that the rise in multi-drug resistant nosocomial pathogens continues to threaten hospitalized patients despite various countermeasures including isolation techniques and antibiotic de-escalation therapy [22,23]. In the present study we focused on the effect of multi-drug resistant strains of P. aeruginosa on the intestinal epithelial barrier since intestinal P. aeruginosa has been shown to be a major cause of morbidity and mortality among immuno-compromised patients [4,24,25].
Caco-2 cells are an ideal cell model for these studies since they express several markers that are characteristic of normal intestinal epithelial cells including the presence of a brush border and the ability to maintain a highly resistant barrier to bacterial pathogens [17,26]. As previously mentioned, the ability of microorganisms to adhere to and alter the barrier function of intestinal epithelia is a key feature that defines their pathogenicity within the intestinal tract reservoir [27,28]. Conversely, the ability of the epithelium to resist the barrier dysregulating effect of a given pathogen through the release of mucus, IgA, defensins, etc, defines its innate defensive properties [29][30][31]. During host illness, especially under circumstances of critical illness, this delicate balance can be tipped in the favor of the microbe where the potential for a versatile pathogen like P. aeruginosa to subvert and erode an already compromised epithelial defense system exists [8,32].
Whether MDR P. aeruginosa [33][34][35][36] strains necessarily express a more virulent phenotype continues to remain a controversial issue. While the behavior of MDR P. aeruginosa against the intestinal epithelium is unknown, its high prevalence in the intestinal tract of critically ill and immuno-compromised patients begs a better understanding of the degree to which certain strains can disrupt the intestinal epithelial barrier. For example the apical side of the intestinal epithelium is highly resistant to various toxic and cytolytic exoproducts of P. aeruginosa including exotoxin A and elastase [8,11,37], whereas the lung is highly susceptible. As such, lung models of P. aeruginosa infection and pathogenesis cannot be directly extrapolated to the intestinal model. Interestingly, data from the present study establish that among the MDR P. aeruginosa isolates tested in the Caco-2 model, most display a minor to minimal ability to disrupt the intestinal epithelium in both motile and non-motile strains.

Phenotype and genotype analysis of P. aeruginosa isolates highly disruptive to the intestinal epithelium
We identified 8 MDR clinical isolates with 3 distinct RAPD fingerprints that display a disruptive phenotype against the intestinal epithelial barrier. The presence of such strains within the intestinal tract of critically ill patients has the potential to induce a state of gut-derived sepsis with a high mortality rate as their presence in this site is often difficult to detect and eradicate.
Common features of these highly disruptive strains include high swimming motility, increased adhesiveness   to intestinal epithelium, and the presence of the exoU gene. ExoU, an effector protein of the type III secretion machinery, has been previously shown to play a major role in mediating a cytotoxic phenotype of P. aeruginosa [38,39] against lung epithelial cells and HeLa cells [40]. That ExoU also plays an important role in disruption of the intestinal epithelial barrier and cellular cytotoxicity in this model suggests that intestinal colonization with MDR P. aeruginosa strains harboring the exoU-genotype may be associated with poor outcome in patients colonized by such strains. Although the presence of ExoS has been previously reported to play a role in the virulence of P. aeruginosa in a lung model [41], we found no correlation between exoS-genotype and the ability of strains to disrupt the intestinal epithelial barrier among our clinical isolates. As previously reported and confirmed by the results of the present study [42], motility and adhesion to host cells are important factors that appear to predict virulence.
As we and others have suggested, bacteria are fully capable of changing their virulence phenotype in direct response to host illness [43,44]. The frequent use of multiple antibiotics in the most severely ill patients could lead to the acquisition of, or alternatively the transformation to, highly virulent strains of P. aeruginosa that pose a significant threat to the patient. The ability of multi-drug resistant strains to persist for prolonged periods in such patients may allow for the development of extremely virulent phenotypes [45].
In conclusion, heterogeneity among critically ill humans, variability in immune response, and antibiotic use could explain the extremely polar phenotypes identified in the series of multi-drug resistant isolates collected in the present study: from phenotypes that are essentially inert with respect to the intestinal epithelium to highly motile, adhesive, and destructive phenotypes. Phenotypic assays such as motility and adhesiveness, and genotyping for the exoU gene could provide a significant prognostic input to identify multi-drug resistant P. aeruginosa strains most likely to place patients at risk for lethal gut-derived sepsis. Further characterization of strains 1, 13 and those of G20 RAPD genotype will be necessary to better understand the precise mechanism by which these strains disrupt the intestinal epithelium to a degree not previously reported for any intestinal pathogen.