Tolerance of antibiotic susceptible and antibiotic resistant Escherichia coli, Enterococcus and Staphylococcus strains from clinical and wastewater samples against ozone was tested to investigate if ozone, a strong oxidant applied for advanced wastewater treatment, will affect the release of antibiotic resistant bacteria into the aquatic environment. For this purpose, the resistance pattern against antibiotics of the mentioned isolates and their survival after exposure to 4 mg/L ozone was determined. Antibiotic resistance (AR) of the isolates was not correlating with higher tolerance against ozone. Except for ampicillin resistant E. coli strains, which showed a trend towards increased resistance, E. coli strains that were also resistant against cotrimoxazol, ciprofloxacin or a combination of the three antibiotics were similarly or less resistant against ozone than antibiotic sensitive strains. Pigment-producing Enterococcus casseliflavus and Staphylococcus aureus seemed to be more resistant against ozone than non-pigmented species of these genera. Furthermore, aggregation or biofilm formation apparently protected bacteria in subsurface layers from inactivation by ozone. The relatively large variance of tolerance against ozone may indicate that resistance to ozone inactivation most probably depends on several factors, where AR, if at all, does not play a major role.
Surface water sources very often serve as drinking water reservoirs or as recreation facilities and therefore their water quality is of fundamental interest. This is manifested in WHO or EU guidelines, for instance in the European Union Water Framework Directive (EC 2000), which sets standards for acute and chronic toxicity in water for aquatic organisms to protect biological diversity and human health. To achieve these aims, improvements to sewage purification by ultrafiltration or ozonation as additional treatment stages in sewage treatment plants (STPs) are discussed and tested. The main goal of these additional wastewater treatment technologies is a reduction of the concentration of emerging contaminants in the effluent of STPs and ultimately in the receiving water bodies.
A positive side effect of, for example, ozonation is the inactivation of facultative pathogenic bacteria such as Escherichia coli, Enterococcus or Staphylococcus strains in effluents of STPs by several log units (Xu et al. 2002; Gehr et al. 2003; Lüddeke et al. 2015). The efficiency of ozone as a disinfectant depends on several parameters such as reaction rates of ozone or its decay products with compounds of sewage sludge, including bacteria (Gehr et al. 2003).
In batch assays Gehr et al. (2003) demonstrated that ozone decayed rapidly and formed hydroxyl radicals for non-selective oxidation, if the alkalinity was low and/or the organic concentration was high. Hunt & Mariñas (1999) reported that molecular ozone was the agent responsible for the inactivation of E. coli. To date, the exact mode of action and the specific target structures of ozone and of its decay products in microbial cells are not completely understood: ozone is expected to react rapidly with the unsaturated bonds of phospholipids and lipopolysaccharides in membranes and cell walls, but intracellular amino acids and proteins may also be affected (Dodd 2012). In addition DNA damages may also be caused by ozone (Gehr et al. 2003).
For advanced sewage treatment ozone is applied after secondary treatment either with a diffuser at the bottom of the tank or with a pump-injector system, injecting small gas bubbles into the influent water of the ozone reactor. Ozone concentrations between 2 and 10 mg/L, depending on the dissolved organic carbon (DOC) concentration, are used for final treatment of sewage. The minimally required contact time of ozone with sewage in STPs for a maximal oxidation is difficult to determine. Investigations in the STPs Bad Sassendorf, Duisburg-Vierlinden and Schwerte (Northrhine-Westphalia, Germany) indicated that the reactions triggered by ozone occurred within less than 30 minutes (Kommunalunternehmen der Stadt Warburg 2013).
In his recent review, Dodd (2012) raised the question whether antibiotic resistant bacteria have an advantage to cope with oxidative stressors such as ozone in comparison to bacteria that are not resistant against antibiotics. We addressed this problem and investigated if the treatment of conventionally purified sewage with ozone would favor a better survival of antibiotic resistant bacteria than of antibiotic sensitive bacteria. For this purpose, the sensitivity of antibiotic resistant and antibiotic susceptible E. coli, Enterococcus and Staphylococcus isolates from clinical samples, sewage and river water against ozone was determined for the first time in detail. Survival of antibiotic susceptible and antibiotic resistant bacterial strains was compared with data obtained from an ozonation unit in an STP.
MATERIALS AND METHODS
Laboratory assay simulating the ozonation of the STP in Eriskirch
In the pilot-scale ozonation unit (OCS system supplied by Wedeco, Herford, Germany) of the STP of Eriskirch (Federal State of Baden-Württemberg, Southern Germany), 34 m3 secondary-treated sewage with a mean DOC of 5.5 mg/L were separated per day and dosed with 4 mg/L ozone that was injected by a venturi injector (Xylem Water Solutions GmbH, Großostheim, Germany). After the ozone dosage, the sewage was pumped into a closed tank (volume: 0.5 m3). The contact time of ozone with sewage was 20 minutes, long enough to decompose organic material until ozone exhaustion (Tripathi et al. 2011).
Ozonation experiments in the laboratory were performed with 100 mL aqueous samples in 250 mL glass reactors, equipped with a diffuser and sterile 0.2 μm PTFE filters (Roth, Karlsruhe, Germany) at the inlet and outlet. Since the pH in the sewage-simulating, 20-fold diluted OECD-medium (Organization for Economic Cooperation and Development) (final DOC of 5.5 mg/L as in treated sewage) dropped below 5.5 after ozonation a phosphate-buffered medium (3.87 g/L K2HPO4, 2.42 g/L KH2PO4, 0.0138 g/L C6H12O6, 9 g/L NaCl dissolved in deionized water, pH 7.0, autoclaved for 15 minutes) with 5.5 mg DOC per L and a pH of 7 after ozonation was used instead. Compared to typical wastewater, the phosphate-buffered medium used did not contain carbonates which might serve as radical scavengers. An ozone-generator (Laborozonisator 301.7, Sander, Uetze-Eltze, Germany) was connected via Norprene tubings with the inlet filter unit of the glass reactor. Ozone was generated from hydrocarbon-free compressed air for 3 minutes and introduced via the diffuser into the glass reactor. A concentration of 4 mg O3/L was reached after 3 minutes. Then, 0.5 mL of an overnight culture of the respective strain, grown in DEV-nutrient broth, was inoculated. The used cell-concentrations were clearly higher than in secondary-treated sewage (e.g. Lüddeke et al. 2015) to be able to quantitatively detect surviving cells.
After 0.5, 1, 2, 5, 20 and 40 minutes of exposure to ozone, surviving cell numbers were determined. From E. coli cultures 10−2, 10−3 and 10−4 fold dilutions and from Enterococcus and Staphylococcus cultures 10−1, 10−2 and 10−3 fold dilutions were prepared and platted twice on DEV-nutrient agar (Roth, Karlsruhe, catalog number: CL02.1). To determine original cell densities in overnight cultures 100 μL of the dilution steps 10−6 and 10−7 were plated on DEV-nutrient agar (according to Deutsches Einheitsverfahren). After 20 hours of incubation at 37°C colonies were counted and inactivation with exposure time to ozone was calculated. The reproducibility of these assays was about +/− 0.3 log units.
Identification and antibiotic susceptibility testing of isolated strains
The tested 52 E. coli, 89 Enterococcus and 80 Staphylococcus strains were isolated within the SchussenAktivplus project during 2012 and 2014 from sewage and river water. This project, which was funded by the Federal Ministry for Education and Research (BMBF, Bonn), focused on the efficiency of advanced sewage treatment technologies to reduce the release of micropollutants and of facultative pathogenic and antibiotic resistant bacteria (for details see Triebskorn et al. 2013; Heß & Gallert 2014; www.schussen-aktivplus.de). Strains that probably have already survived ozonation (e.g. isolated from the effluent of the STP of Eriskirch) were excluded from this study. Depending on the contamination level of the respective samples, direct plating or the filtration method was used to obtain E. coli, Enterococcus and Staphylococcus isolates.
For cultivation of E. coli isolates, ECD-agar (Merck Millipore, Darmstadt, Germany) and ESBL CHROMagar™ (MAST Diagnostica GmbH, Reinfeld, Germany) were used, following the manufacturer's instructions. On ECD-agar, blue fluorescent and indole positive colonies (red-colored after reaction with 10 μL Kovac's reagent) were identified by polymerase chain reaction (PCR) targeting of a specific tuf-gene fragment (Maheux et al. 2009). Briefly, the DNA-amplification was performed in a total volume of 25 μL containing 0.625 units True Start HS Taq DNA Polymerase, 2.5 mM MgCl2 (Thermo Fisher Scientific, Waltham, MA, USA), 0.25 mM of each dNTP, 10 μM of both primers and 0.5 μL of template DNA (extraction with phenol/chloroform).
For cultivation of Enterococcus isolates bile-esculin-azide agar (Roth) and VRE CHROMagar™ (MAST Diagnostica GmbH), supplemented with 0.15 g/L sodium azide were used, following the manufacturer's instructions. Black colonies with a halo on bile-esculin-azide agar and pink colonies on VRE CHROMagar™ were identified on species level based on their biochemical properties with Micronaut-Strep2®-microtiter plates according to the manufacturer's instructions (MERLIN, Gesellschaft für Mikrobiologische Diagnostika mbH, Bornheim, Germany).
For cultivation of Staphylococcus isolates Chapman-Stone agar, containing 0.05 g/L sodium azide, was used. Colonies grown after incubation for 48 hours at 37°C were streaked on Mannitol-Salt agar before they were identified by their physiological reactions on Micronaut-Staph®-microtiter plates (MERLIN). S. aureus and S. saprophyticus isolates from clinical specimen were obtained from Dr A. Becker, Städtisches Klinikum Karlsruhe.
Antibiotic resistance (AR) of the E. coli isolates against ampicillin (10 μg), ciprofloxacin (5 μg), cotrimoxazol (1.25 μg trimethoprim/23.75 μg sulfamethoxazol; SXT) was tested using the agar diffusion test according to Deutsche Industrie Norm (DIN 58940, 2011). For testing AR against cefotaxim (5 μg) clinical breakpoints according to EUCAST (2011) were applied. All ampicillin resistant isolates were additionally tested against ceftazidim (30 μg) and cefpodoxim (10 μg) according to Clinical and Laboratory Standards Institute (CLSI 2011); the inhibitory effect of clavulanic acid on ß-lactamase was checked as described by Bradford (2001). According to the definition of Robert-Koch-Institut (2007), extended-spectrum-ß-lactamase (ESBL) producers phenotypically were resistant against cefpodoxim as well as against ceftazidim and/or cefotaxim. All isolates that were resistant against the above-mentioned antibiotics were classified as ESBL producers.
AR of the Enterococcus isolates was tested against ampicillin (10 μg), chloramphenicol (30 μg), ciprofloxacin (5 μg) and erythromycin (15 μg) according to DIN 58940 (2011). The susceptibility of E. faecium and E. faecalis isolates against vancomycin (VRE) was tested according to CLSI (2011). Vancomycin resistance (VR) was confirmed by the presence of vanA-E and vanG genes using primers and PCR conditions previously described by Depardieu et al. (2004). The DNA-amplification was performed in a 25 μL assay containing 0.625 units True Start HS Taq DNA Polymerase, 2.5 mM MgCl2, 0.25 mM of each dNTP, 10 μM of both primers and 0.5 μL of template DNA (extraction with phenol/chloroform). Intrinsic low-level vancomycin resistances of E. gallinarum and E. casseliflavus (vanC1- and vanC2-type) were not considered.
Susceptibility of the Staphylococcus isolates against oxacillin (5 μg), ciprofloxacin (5 μg) and erythromycin (15 μg) was tested by the disc-diffusion test with Mueller-Hinton agar according to DIN 58940 (2011) and against clindamycin (2 μg) according to CLSI (2011). Oxacillin resistance was confirmed by the presence of the mecA gene using previously described primers and PCR conditions (Predari et al. 1991). DNA-amplification was performed in a 25 μL assay containing 0.625 units True Start HS Taq DNA Polymerase, 2.5 mM MgCl2, 0.25 mM of each dNTP, 10 μM of both primers and 0.5 μL of template DNA (extraction with phenol/chloroform).
Pigment extraction and quantification
Cells of an overnight culture of the respective S. aureus strains, grown in DEV-nutrient broth, were pelleted by centrifugation. The cells were re-suspended in a 0.9% sodium chloride solution to obtain a turbidity of McFarland 2.0. Two mL of the suspension were used to extract staphyloxanthin and to quantify pigment production as described by Morikawa et al. (2001).
For descriptive analysis of the obtained data, box plots created with Microsoft Excel were used displaying variation within the respective ‘clusters’: the line dividing the rectangle built by the first and third quartile displays the median of the obtained data. Minimal and maximal values are indicated by the whiskers.
RESULTS AND DISCUSSION
Exposure time for inactivation of bacteria by ozone
A Gram-negative E. coli strain, a Gram-positive Enterococcus isolate and a Gram-positive Staphylococcus isolate were exposed to 4 mg/L ozone. Inactivation of cells by more than 3 log units occurred within the first 30 seconds contact time from 1–4 × 106 initially to 5 × 102–5 × 103 bacteria per mL (Figure 1). No further inactivation occurred later on, although even after 5 minutes more than 2 mg/L ozone was still present, an ozone concentration that was higher than the initial concentration in some pilot-scale ozonation systems. A consistently higher ozone concentration and a longer exposition time might presumably result in a higher inactivation rate. The phenomenon of a rapid initial inactivation of bacteria by ozone, however, seemed to be independent of the applied ozone concentration since Xu et al. (2002) also reported no difference in inactivation of fecal coliforms between 2 and 10 minutes hydraulic retention time for a given ozone dose. Differences in cell wall and membrane architecture between Gram-positive and Gram-negative bacteria seemed not to be the decisive structures that were responsible for survival in the presence of ozone.
Inactivation of E. coli by ozone
In the STP of Eriskirch, the percentage of ampicillin resistant isolates, which were additionally resistant against at least one more class of antibiotics, decreased from 8.3 to 3.8% after ozonation, as noted by Lüddeke et al. (2015). This was in line with the laboratory approach, where isolates that, in addition to ampicillin, were resistant against cotrimoxazol, ciprofloxacin or cefotaxim (ESBL) apparently were more sensitive against ozone, manifested in higher median values for inactivation of living cells (2.7–3.4 log units compared to 1.9 log units of only ampicillin resistant isolates; Figure 2). Händel et al. (2013) demonstrated that E. coli cells apparently compensated ‘metabolic costs’ for AR by physiological adaption. Comparing antibiotic susceptible and antibiotic resistant strains, changes in gene expression levels, mainly of genes for cell wall maintenance, DNA metabolic processes, cellular stress and respiration as well as for the electron transport, were observed. Overall the acquisition of AR primarily seemed not to require much extra energy, but seemed to cause a reduced ecological versatility (Händel et al. 2013). Nevertheless there were apparently differences between mechanisms for ARs: on average, the ampicillin-only resistant E. coli strains were less susceptible against ozone than E. coli strains that were additionally resistant against ciprofloxacin and cefotaxim, as judged from results of laboratory assays (Figure 2) and of pilot plant operation in Eriskirch. The tested ampicillin-only resistant E. coli isolates showed a very broad range of susceptibility against ozone. The inactivation of living cells varied between 0.8 and 3.9 log units (Figure 2). The reason could be that several mechanisms for resistance against ß-lactam-antibiotics are expressed, that are more or less energy intensive, e.g. overexpression of the intrinsic ampC-gene or the expression of one of the bla-genes (e.g. TEM, SHV, CTX-M; Robert Koch Institut 2007).
Inactivation of Enterococcus isolates by ozone
Nevertheless, concerning the inactivation of E. casseliflavus strains by ozone, which varies over 0.9–4.3 log units, pigment synthesis itself could not explain the broad range of susceptibility against ozone within the strains of this species. In the effluent of the STP Eriskirch the percentage of isolates that were identified as E. casseliflavus was higher after ozonation (18.4%) than before ozonation (12.3%; as noted by Lüddeke et al. 2015), indicating a higher ‘resistance’ of the respective species against ozone. A notable reduction of 15.8% of E. faecium strains in the effluent of the STP Eriskirch after ozonation (as noted by Lüddeke et al. 2015) was in accordance with results of the laboratory experiments (Figure 3). Lüddeke et al. (2015) assumed that the recorded shift might explain the drastic decrease of antibiotic resistant enterococci of about 25.4% after ozonation.
The results of our laboratory experiments supported this hypothesis since the median for inactivation of active cells of antibiotic resistant E. faecium strains was higher (3.4–4.2 log units; Figure 3(b)) compared to that of susceptible strains (1.7 log units; Figure 3(b)). Thus, antibiotic resistant strains may reveal a higher sensitivity to ozone. Luczkiewicz et al. (2011) also reported an increased percentage of erythromycin, ciprofloxacin and chloramphenicol susceptible Enterococcus strains after ozonation. They also observed a decreased percentage of isolates that were identified as E. faecium and E. faecalis (10.2% and 25.2% decrease, respectively), mainly in favor of E. hirae (36.3% increase; Luczkiewicz et al. 2011). Händel et al. (2013) reported that E. faecium strains with ampicillin resistance invested the same maintenance energy as antibiotic sensitive E. faecium strains, whereas vancomycin resistant strains required extra energy. High-level vancomycin resistance (e.g. encoded by vanB), associated with a metabolic burden, might result in a higher sensitivity against ozone compared to vancomycin susceptible strains in the absence of the antibiotic (Figure 3(b)). E. casseliflavus and E. gallinarum strains intrinsically are low-level vancomycin resistant (vanC1- and vanC2-type). Comparing ozone susceptibility of isolates of the two Enterococcus species in laboratory assays, E. casseliflavus strains (median inactivation of living cells 1.5 log units; Figure 3(a)) were more resistant against ozone than E. gallinarum strains (most sensitive Enterococcus species: median inactivation of living cells 3.8 log units; Figure 3). Therefore, low-level vancomycin resistance seemed not to be ‘the’ determining factor for ozone-susceptibility. Pigmentation of E. casseliflavus might be one possibility to explain this difference.
Inactivation of Staphylococcus isolates by ozone
The median inactivation of active cells of S. epidermidis isolates (2.3 log units) was only slightly higher compared to that of S. aureus (2.1 log units, Figure 4(a)). S. epidermidis is a well-known biofilm-forming species (e.g. Cerca et al. 2005) and cells of an overnight culture already aggregated and formed visible flocs (data not shown).
Although cell flocs in the S. epidermidis culture could be disaggregated to single cells by high shear forces, extracellular polymeric substances (EPS) may have surrounded single cells and thus have protected cell walls or cell membranes from oxidation by ozone. The clearly higher resistance against ozone of cells embedded in biofilms has already been described and attributed to an EPS matrix (e.g. Hems et al. 2005). The same observation as for S. epidermidis was made with aggregating strains of S. xylosus, resulting in a median inactivation by ozone of 2.7 log units, whereas isolates of S. sciuri and S. saprophyticus, both of which neither produced pigment nor grew in aggregates were less resistant to ozone (Figure 4(a)).
Regarding the inactivation of S. saprophyticus isolates of clinical and environmental samples by 4 mg/L ozone, there were no significant differences with respect to their origin (Figure 4(b)). The median for inactivation, independently of origin or AR, e.g. against erythromycin was almost identical (Figure 4(b)).
Taking the results of E. coli strains, Enterococcus and Staphylococcus isolates from hospital sources, sewage and river water as a whole, then single isolates, that were resistant against at least one of the tested antibiotics, were not per se more resistant against oxidative stress by 4 mg/L ozone than antibiotic sensitive strains. Some species seemed to be a little more ozone-resistant than others and thus might be ‘positively selected’ by ozonation with the consequence that the percentage of such strains increases. Pigment and biofilm production may be factors affecting sensitivity against ozone but further tests particularly focusing on these aspects are necessary to estimate their influence on ozone sensitivity.
From this study we have established that AR per se did not lead to a reduced sensitivity of E. coli, Enterococcus and Staphylococcus strains against 4 mg/L ozone. Also, pigment-producing E. casseliflavus and S. aureus were a little less sensitive against ozone compared to non-pigmented species of the respective genera.
Within the same species, the susceptibility against ozone expressed as inactivation of active cells differed over a wide range up to 3.8 log units. And finally, cell wall and membrane architecture seemed not to be ‘the’ decisive structures that were responsible for survival in the presence of ozone. Several other factors may influence resistance against ozone within species or genera, including the ability to form aggregates or biofilms.
SchussenAktivplus was funded by the Federal Ministry for Education and Research BMBF (02WRS1281I) and co-funded by the Ministry of Environment Baden-Württemberg. In addition, Jedele & Partner GmbH, Ökonsult GbR, the City of Ravensburg, the AZV Mariatal and the AV Unteres Schussental contributed financially to this research project. The authors are very grateful to Dr A. Becker and her team at Städtisches Klinikum Karlsruhe for providing clinical strains and helpful discussions. We thank Dr S. Schmidt for providing E. coli and Enterococcus strains. Furthermore, we thank the team at University of Tübingen under the guidance of Professor Dr R. Triebskorn and the SchussenAktivplus project team of ISF at LUBW in Langenargen for organizing sampling campaigns and for supplying sewage and river water samples. We thank Professor Dr J. Winter, KIT, for providing laboratory space and helpful discussions during experimentation and preparation of the manuscript.