Clinical isolates of Pseudomonas aeruginosa from superficial skin infections have different physiological patterns

Authors

  • Andrius Buivydas,

    1. Department of Biosciences, University of Helsinki, Helsinki, Finland
    2. Institute of Biotechnology, University of Helsinki, Helsinki, Finland
    3. Department of Biochemistry and Biophysics, Faculty of Natural Sciences, Vilnius University, Vilnius, Lithuania
    Search for more papers by this author
  • Tanja Pasanen,

    1. Division of Clinical Microbiology, Helsinki University Hospital, HUSLAB, Helsinki, Finalnd
    Search for more papers by this author
  • Ana Senčilo,

    1. Department of Biosciences, University of Helsinki, Helsinki, Finland
    2. Institute of Biotechnology, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • Rimantas Daugelavičius,

    1. Department of Biochemistry and Biophysics, Faculty of Natural Sciences, Vilnius University, Vilnius, Lithuania
    2. Department of Biochemistry and Biotechnologies, Faculty of Natural Sciences, Vytautas Magnus University, Kaunas, Lithuania
    Search for more papers by this author
  • Martti Vaara,

    1. Division of Clinical Microbiology, Helsinki University Hospital, HUSLAB, Helsinki, Finalnd
    Search for more papers by this author
  • Dennis H. Bamford

    Corresponding author
    1. Institute of Biotechnology, University of Helsinki, Helsinki, Finland
    • Department of Biosciences, University of Helsinki, Helsinki, Finland
    Search for more papers by this author

Correspondence: Dennis H. Bamford, Institute of Biotechnology, University of Helsinki P.O.Box 56(Viikinkaari 5), 00014 Helsinki, Finland. Tel.: +358 9 191 59100; fax: +358 9 191 59098;

e-mail: dennis.bamford@helsinki.fi

Abstract

Pseudomonas aeruginosa are known to have a wide physiological potential allowing them to constantly populate diverse environments leading to severe infections of humans such as septicemia, leg ulcers, and burn wounds. We set out to probe physiological characteristics of P. aeruginosa isolates from diabetic leg ulcers collected from Helsinki metropolitan area. A total of 61 clinical isolates were obtained. Detailed phenotypic (physiological) characteristics [outer membrane (OM) permeability, membrane voltage, and activity of multidrug resistance pumps] were determined in several growth phases leading to the division of the analyzed set of P. aeruginosa strains into five distinct clusters including cells with similar physiological properties. In addition, their antibiotic resistance patterns and genetic heterogeneity were determined. Multiple isolates from the same patient were genetically very closely related and belonged to the same phenotypic cluster. However, genetically close isolates from different patients expressed very different phenotypic properties. The characteristics of infected patients seem to determine the growth environments for microorganisms that adapt by changing their physiological and/or genetic properties.

Introduction

The human skin microbial communities are composed of commensal microorganisms together with (opportunistic) pathogens. The microbial cells associated with humans outnumber the human cells by one order of magnitude and, consequently, have a considerable effect on the human host. We chose to study P. aeruginosa strains originating from diabetic leg ulcers and decubitus wounds where P. aeruginosa has established a stable presence. Pseudomonas aeruginosa is a highly adaptable and metabolically versatile Gram-negative bacterium occupying a broad spectrum of ecological niches [e.g. soil, plants surface, coastal marine habitats, and surface of human and animal skin (Mathee et al., 2008; Wilson & Dowling, 1998)]. Importantly, in addition to be associated with leg ulcers, P. aeruginosa is one of the major causes of mortality and morbidity of burn victims, cystic fibrosis patients, and immunocompromised patients such as those suffering from AIDS (Harrison, 2007; Tohidpour et al., 2009).

Pseudomonas aeruginosa has a high intrinsic and acquired antibiotic resistances, which make it difficult and sometimes even impossible to control the infection (Bonomo & Szabo, 2006). There are clinical P. aeruginosa strains resistant to all available antibiotics in clinical use (Livermore, 2002; Bonomo & Szabo, 2006). Pseudomonas aeruginosa employs all known antibiotic resistance strategies common to Gram-negative bacteria. However, active efflux pumps play the major role for survival under antibiotic pressure (Lomovskaya et al., 2007; Strateva & Yordanov, 2009). To understand and evaluate the heterogeneity and resistance patterns, a number of comparative pheno- and genotype studies on the clinical and environmental isolates have been performed such as pheno (e.g. antimicrobial resistance, antigen determinations, presence of virulence factors)- and genotypic tests [pulsed field electrophoresis (PFGE), amplified fragment length polymorphism [AFLP (Finnan et al., 2004; Pirnay et al., 2002)], random amplification of polymorphic DNA [RAPD (Deligianni et al., 2010)], repetitive-element-based PCR [rep-PCR; (Doleans-Jordheim et al., 2009; Kiewitz & Tummler, 2000; Syrmis et al., 2004)]].

We have recently shown that a variety of physiological membrane properties of Gram-negative bacteria can be studied in real time using ion-selective electrodes and membrane-penetrating tetraphenylphosphonium (TPP+) ions generating information on outer membrane (OM) permeability, membrane voltage, and activity of antibiotics-extruding pumps (Daugelavicius et al., 2010). In this study, we apply both TPP+ and rep-PCR measurements on a set of P. aeruginosa strains obtained from diabetic leg ulcers and decubitus wounds. The phenotypic together with genotypic results propose that each individual supports a distinct P. aeruginosa population with characteristic physiological properties.

Materials and methods

Bacterial strains and susceptibility testing

Sixty-one clinical P. aeruginosa isolates from 51 different patients with superficial skin infections (leg and foot ulcers and decubitus wounds) were collected by HUSLAB (Laboratory of Helsinki University Central Hospital) in 2007. The samples originated from 18 outpatient health centers and 11 different hospitals. Pseudomonas aeruginosa strains EryR (nfxB), MutGR1 (mexZ), PAO7H (nfxC), and PT629 (nalB) overproducing multiple drug efflux pumps MexCD-OprJ, MexXY-OprM, MexEF-OprN, and MexAB-OprM, respectively, and wild-type strain PAO1 were obtained from Prof. Patrick Plésiat's laboratory (CHU Minjoz, Besanc¸ France).

All the strains were tested for susceptibility to amikacin, ceftazidime, ciprofloxacin, meropenem, piperacillin+tazobactam, and tobramycin by disk diffusion method according to the CLSI guidelines (http://www.clsi.org).

Electrochemical assessment of phenotypes of P. aeruginosa cells

To assess the OM permeability, membrane voltage, and activity of efflux pumps in real time, TPP+ ions and TPP+-selective electrodes were employed as described previously (Daugelavicius et al., 2010). Briefly, cells were grown in LB medium (Sambrook & Russell, 2001) containing diminished NaCl concentration (0.5%) at 37 °C with aeration. When an appropriate OD550 was reached, 2.5 mL of suspension was directly transferred to 10 mL thermostated (37 °C) vessels containing 2.5 mL of 400 mM sodium phosphate buffer, pH 8. TPP+-selective electrodes were connected to the electrode potential–amplifying system with an ultralow-input bias current operational amplifier AD549JH (Analog Devices). The amplifying system was connected to a computer through the data acquisition board AD302 (Data Translation, Inc.). Ag/AgCl reference electrodes (Thermo Inc.; Orion model 9001) were indirectly connected to the cell suspension through agar salt bridges. Electrodes were calibrated at the end of each measurement. Typical curves from at least three independent measurements were analyzed.

Phenotyping of clinical isolates of P. aeruginosa

For the detailed physiological characterization, we dissected the TPP+ fluxes to discrete component reactions for three different cell growth phases as follows: (1) initial TPP+ accumulation into the cells [depends on the OM permeability, the activity of TPP+ extruding pumps, and the level of membrane voltage (Δψ)]; (2) EDTA effect [diminishes outer membrane (OM) barrier]; (3) PAβN effect [inhibits resistance nodulation division (RND)-type efflux pump activities]; (4) maximal accumulation of TPP+ (reflects the level of Δψ); (5) the effects of gramicidin D [GD; intercalates into the cytoplasmic membrane (CM) and forms channels for monovalent cations when OM is highly damaged or absent]; (6) polymyxin B (PMB) titration effects (PMB disrupts OM barrier and diminishes Δψ). For the clustering of different phenotypes, we utilized numerical values obtained as explained in the legend for Fig. 1. All the strains were grouped into phenotypic clusters based on the similarity of above-described factors.

Figure 1.

Schematic representation of the evaluation of phenotypic data of clinical isolates of Pseudomonas aeruginosa. To compare physiological phenotypes of the isolates, TPP+ concentration changes in the medium were given particular values. The values for the initial and maximal TPP+ accumulation were assigned lower (three categories instead of five) as both processes are influenced by two indistinguishable factors: (1) the level of the membrane potential that varies among different strains or even among the same strain grown under different conditions; (2) the amount of TPP+ bound to the envelope of analyzed cells. For the evaluation of the initial and maximal accumulations of TPP+, three categories were assigned. Value ‘1’ was given in the case when the initial TPP+ concentration in the medium was not more than 0.25 μM different from the final one. Values ‘2’ and ‘3’ were assigned in the cases when the initial TPP+ concentration in the medium differed from the final one up to 1 μM and more than 1 μM, respectively. For all other analysis stages, values from 1 to 5 were assigned. Value ‘1’ was assigned in cases when the cells responded to additives in a reverse manner, for example, EDTA caused a leakage of TPP+ (see Fig. 2b, the representative isolate of the cluster II). The value ‘2’ was assigned in cases when the response to additives was < 0.25 μM. In other cases, the values ‘3’, ‘4’, and ‘5’ were assigned when TPP+ concentration in the medium changed < 0.5 μM, 0.5–1 μM, and > 1 μM, respectively.

Genotyping of P. aeruginosa strains

DNA was extracted from strains grown on CLED (cysteine lactose electrolyte deficient) plates using the UltraClean microbial DNA isolation kit (Mo Bio Laboratories, Solona Beach, CA) and diluted to 35 ng μL−1. The DNA was amplified using the DiversiLab Pseudomonas kit (Bacterial Barcodes, Inc. cat no PL-PA01, Athens, GA) for DNA fingerprinting following the manufacturer's instructions. Briefly, 2 μL of genomic DNA, 18 μL the rep-PCR master mix, 2 μL primer mix provided in the kit, 0.5 μL AmpliTaq polymerase, and 2.5 μL 10 X PCR buffer (Applied Biosystems Roche, Branchburg, NJ) were mixed to a total volume of 25 μL per reaction. PCR was carried out on preheated thermal cycler (DNA Engine Tetrad 2, Peltier Thermal Cycler Bio-Rad, Hercules, CA). The thermal cycling parameters were as follows: initial denaturation at 94 °C for 2 min, followed by 35 cycles of denaturation at 95 °C for 30 s, annealing at 50 °C for 30 s, extension at 70 °C for 1.5 min, and final extension at 70 °C for 3 min. The kit-specific positive and negative controls were included within each reaction set for the validity of amplification. The rep-PCR products were detected, and the amplicons were separated using microfluidics lab-on-a-chip technology and analyzed using the DiversiLab system (Bacterial Barcodes, Inc.). Further rep-PCR analysis was performed with the web-based DiversiLab software (version 3.4) using the band-based modified Kullback–Leibler distance for the calculation of percent similarities. The automatically generated dendrogram and the virtual gel image were used for interpretation. The manufacturer-provided guidelines for strain-level discrimination were: similarity more than 97% is considered as indistinguishable (no differences in fingerprints); similarity more than 95%, as similar (1–2 band difference in fingerprints); and similarity less than, 95% as different.

Results

Isolates obtained

We analyzed sixty-one P. aeruginosa isolates originating from diabetic superficial skin infections from the Helsinki metropolitan area collected in HUSLAB (Laboratory of Helsinki University Central Hospital) during the period of January 31–June 26, 2007. Nineteen of the isolates were obtained from samples taken repetitively from the same patient during their treatment periods (eight patients were repetitively sampled twice, and one patient, three times; Fig. 2, colored isolates). In addition, five control strains (PAO1 and its derivates harboring different RND-type efflux pumps) were included in our studies. All the strains were tested for antibiotic resistance, physiological phenotypes, and rep-PCR.

Figure 2.

Genotypic (a) and phenotypic (b) clustering of clinical isolates of Pseudomonas aeruginosa. (a) dotted and solid red lines indicate genetic distances for genetically indistinguishable and similar isolates, respectively; (b) on the right isolates falling into the cluster are listed; the left panel shows physiological patterns of a representative isolate where different curve color represents samples taken from different growth phases, and color codes indicate OD550; TPP+, EDTA, phenylalanyl-arginyl-β-naphthylamide (PAβN), and gramicidin D (GD) were added to the cell suspensions to obtain final concentrations of 3 μM, 0.4 mM, 60 μg mL−1, and 4 μg mL−1, respectively, polymyxin B (PMB) was added to the concentrations indicated in the figure (μg mL−1). Repetitive isolates from a single patient are represented using the same color. Standard PAO1 strains having different expression level of a particular MDR pump are indicated with arrows together with the name of the pump. cI, intermediate resistance to ciprofloxacin; cR, resistance to ciprofloxacin; mR, resistance to meropenem; tI, intermediate resistance to tobramycin; tR, resistance to tobramycin. For the rep-PCR gel see Fig. S1.

Antibiotic resistance

Resistance to clinically relevant antibiotics (see 'Materials and methods') was determined by the disk diffusion method, according to CLSI standards. Seventeen of the clinical isolates were resistant to at least one antibiotic, and six of them were resistant to more than one antibiotic. The resistance to particular antibiotics is indicated in Fig. 2, next to the sample ID. All the resistant isolates showed resistance to ciprofloxacin (an antibiotic belonging to fluoroquinolones), except for the isolate key 34 (Fig. 2a), which was resistant only to meropenem (β-lactam). Two repetitive isolates from the same patient were resistant to the same antibiotic (Fig. 2a, keys: 57 and 58).

Physiological phenotypes of P. aeruginosa clinical isolates

The P. aeruginosa strains were also analyzed by comparing their phenotypic properties: (1) OM permeability, (2) activities of their antibiotics-extruding pumps, and (3) resistance of the cell envelope to permeabilizing/disrupting agents (PMB and GD). All these phenotypic properties (comprehensive cell physiology indicators) were analyzed using real-time activity measurements employing tetraphenylphosphonium (TPP+) and TPP+-selective electrodes, as described in 'Materials and methods'. All the 66 P. aeruginosa strains were analyzed when the culture turbidity (OD550) reached 0.55, 1, and 1.45, representing different growth phases.

Phenotypic grouping

We divided our P. aeruginosa collection into five distinct phenotypic clusters using the above-described principles (Fig. 2b). The characteristic properties of the 23 strains belonging to cluster BI (Fig. 2b) were low initial accumulation of TPP+, noticeable OM permeabilization with EDTA, and a strong PAβN inhibitory effect. These cells also had a high membrane voltage ensuring significant accumulation of TPP+ and no significant GD effects. Only four strains were prescribed to the cluster BII (Fig. 2b), which had unique characteristics such as high initial permeability to TPP+ and an unusual EDTA effect resulting in the release of accumulated TPP+. Cells of the cluster BII also showed low response to PAβN addition, were sensitive to GD, and were disrupted by low PMB concentrations reflecting to exceptional permeability of the cell OM to various compounds. The strains belonging to the cluster BIII (Fig. 2b) had low initial permeability to TPP+. However, very strong EDTA and GD effects were the main characteristics of this cluster containing seven isolates. Cluster BIV isolates (Fig. 2b) showed the lowest responses to all of the parameters tested. There were five isolates in this cluster. The distinguishing set of characteristic properties of cluster BV (Fig. 2b) containing 27 strains were low initial permeability to TPP+, low EDTA effects on OM permeability, and strong PAβN inhibitory effect. Pseudomonas aeruginosa strains belonging to the BV cluster showed no responses to GD but were remarkably sensitive to PMB in any of the three analyzed cell growth phases.

Distribution of isolates in phenotypic clusters

Repetitive isolates from the same patient always fell to the same cluster (see color-coded strains, Fig. 2b) except isolates MX1987 and MX2069 (Fig. 2b, orange) which diverged into clusters BI and BIV, respectively. All clusters contained pairs of repetitive isolates except cluster BIV. The highest number of repetitive isolates (four pairs) was observed in cluster BI. Clusters BII and BIII contained a single pair of repetitive isolates each.

Antibiotic-resistant isolates were distributed to all of the clusters. The percentages of antibiotic-resistant strains within the clusters were as follows: BI – 13%; BII – 75%; BIII – 29%; BIV – 60%; BV – 22%. Clusters BI, BIII, and BV contained isolates (two, one, and three, respectively) resistant to more than one antibiotic (Fig. 2b). The clusters BII and BIV contained only ciprofloxacin resistant isolates.

Five genetically highly similar PAO1 strains but having a different expression level of particular MDR efflux pump (Fig. 2a, cluster AII) were distributed among three phenotypic clusters (Fig. 2b, clusters: BI, BIII and BV). Strain PT629, harboring an increased expression of MexAB-OprM, was a member of cluster BI together with wt strain PAO1. EryR and PAO7H, harboring an increased expression of MexCD-OprJ and MexEF-OprN, respectively, were included in the cluster BIII. Strain MutGR1, harboring an increased expression of MexXY-OprM, was classified to the cluster BV (Fig. 2b).

Genotyping of the P. aeruginosa isolates

DiversiLab system was used to genotype our set of P. aeruginosa isolates (the obtained genetic dendrogram is shown in supplemental Fig. S1). Such genotyping has been successfully applied to detect and investigate multidrug resistant P. aeruginosa clones in hospital wards (Deplano et al., 2011). Our rep-PCR data revealed that the P. aeruginosa populations studied were highly diverse. The criteria used for similarity were the following: indistinguishable, when genetic similarity of the strains was > 97%; similar – 95%; dissimilar – < 95%. Four clusters could be observed in the dendrogram (Fig. 2a). As anticipated, all PAO1 strains formed a single cluster (Fig. 2a, AII) and were designated as indistinguishable. Four, five, and six isolates formed clusters AI, AIII, and AIV, respectively (Fig. 2a). Four and three antibiotic-resistant isolates accumulated in clusters AIII and AIV, respectively. However, there were no resistant strains in the clusters AI and AII.

Genetically indistinguishable strains were repetitively isolated from the same patient in four cases (Fig. 2a, keys: 1 and 2; 12 and 13; 14 and 15; 57 and 58). However, in one case with three isolates from a single patient (Fig. 2a, keys: 26, 27 and 28), two were indistinguishable and one (Fig. 2a, key 28) was clearly dissimilar to the two others. Two repetitive isolates (Fig. 2a, keys: 64 and 65) were similar. The rest six of repetitive isolates (Fig. 2a, keys: 10 and 11; 18 and 19; 39 and 40) were determined as dissimilar. It is worth mentioning that genetic similarity/difference of the repetitive isolates was not connected to the time of isolation (not shown).

In addition, a few sporadic genetically similar pairs of the clinical P. aeruginosa isolates neither belonged to any of the clusters nor originated from the same patient (Fig. 2a, keys: 6 and 7; 29 and 30; 35 and 36). Two of them were resistant to ciprofloxacin (Fig. 2a, keys: 35 and 36). It is also worth noting that there was no apparent connection between the isolation places and the genetic similarity observed (not shown).

Relating genetic diversity to phenotypic properties

We investigated genetic and phenotypic relationships by comparing the genetic heterogeneity of the P. aeruginosa isolates (Fig. 2a) to the phenotypic one (Fig. 2b). We considered analyzing only genetically similar or indistinguishable isolates together with the repetitive single-patient isolates.

All the isolates from the genotypic cluster AI (Fig. 2a) were found in the phenotypic cluster BI (Fig. 2b). However, it was the only robust genotypic–phenotypic connection observed among isolates. Genetically indistinguishable pairs of isolates (Fig. 2a, keys: 49 and 50; 52 and 53) having the same antibiotic resistance patterns were singled into different phenotypic clusters (Fig. 2b, clusters: BV and BI; BIII and BV, respectively). In addition, antibiotic-sensitive isolate ET1150 (Fig. 2a, key 51) showed phenotypic similarity (Fig. 2b, cluster BI) only to one (Fig. 2b, cluster BI, ME1343) of the two genetically similar multidrug resistant strains (Fig. 2a, keys: 49 and 50).

Discussion

There are well-developed genetic tools to analyze bacterial populations obtained from various clinical settings (Ibrahem et al., 2008; Doleans-Jordheim et al., 2009; Thomsen et al., 2010; Stewart et al., 2011). However, there is much less information on the physiology of the microorganisms residing in the human body, while even gentle changes in the environment may have an impact on the observable phenotypic changes leading to genetically indistinguishable or very similar bacteria possessing very different physiological phenotypes as observed here.

We have studied populations of P. aeruginosa from a common clinical source (diabetic leg ulcers and decubitus wounds) together with well-defined standard strains and applied established membrane voltage (Daugelavicius et al., 2007; Krupovic et al., 2007ab; Kivela et al., 2008) and multiple drug resistance efflux pumps (Daugelavicius et al., 2010) monitoring systems. The genetically indistinguishable standard strains overexpressing different Mex pumps resulted in very different phenotypic patterns (Fig. 2), which confirmed the delicate distinctive power of the applied physiological assay. rep-PCR-indistinguishable isolates harboring the same multiple antibiotic resistance patterns (Fig. 2a, keys: 49 and 50; 52 and 53) had phenotypic difference emphasizing the high physiological variability of P. aeruginosa. The presence of several different resistance mechanisms is also supported by our finding that antibiotic-sensitive strains (showing genetic similarity to resistant ones) share physiological similarity (Fig. 2a, keys: 50 and 51; 2B, cluster BI). Recent findings and as also observed here support the observation that each human possesses a unique set of steady microbial populations (Fierer et al., 2010), including wounds with distinct microbiota fingerprints (Thomsen et al., 2010). When we compared results from multiple ulcer samples from the same patient, it was observed that they belonged to a specific phenotype indicating the sensitivity and reproducibility of our physiological assays. Practically, all samples from the same patient had the same physiological phenotype (see Fig. 2b). The observed phenotypic variability between individuals could be related not only to genetically determined physiological differences among isolates, but also reflect diverse conditions in the same occupational niches (the same isolation origin) in different humans. The time points of repetitive isolations varied from one week to three months (not shown), indicating that a stable adjusted genotype together with stable phenotype is established soon leading to unique wound populations with their own properties as observed here.

Monitoring the physiology of clinical isolates extends the understanding of bacterial population behavior in response to even small changes in the environment. Moreover, it unravels physiological variability and could be highly beneficial when choosing more precise treatments of particular infections taking into account the huge physiological capacity of P. aeruginosa.

Acknowledgements

Funding: This work was supported by the Academy Professor (Academy of Finland) funding grants 255342 and 256518 (D.H.B.), the Finnish Center of Excellence in Virus Research Program (2006–2011) of the Academy of Finland (grant 1129684 to DHB), Research Council of Lithuania (grant MIP 128/2010 to RD), and COST action BM0701 (ATENS).

Ancillary