Environmental water samples were collected during October 2015 from Svartån River and Hjälmaren Lake near Örebro, Sweden, a city that reflects urban areas in Nordic countries. The Svartån River flows through Örebro city and is the recipient of the effluent water from Örebro wastewater treatment plant (WWTP) before it flows into Hjälmaren Lake. The WWTP serves a population of 140,000 people and processes an average of 45,000 m3 of wastewater per day (www.orebro.se). The untreated wastewater from Örebro hospital, veterinary clinics, household, and industries is transported via the sewage system to the WWTP. There are no pharmaceutical industries in Örebro city, and most of the agricultural activities are outside of the urban area. Surface water samples from the river were collected from upstream (59° 16′ 03.5″ N, 15° 08′ 49.9″ E) and downstream (59° 16′ 42.5″ N, 15° 15′ 41.5″ E) of Örebro city and WWTP. Surface water samples from Hjälmaren Lake (59° 16′ 40.2″ N, 15° 17′ 31.1″ E) were collected approximately 1 nautical mile downstream of WWTP effluent point. The influent and effluent water from Örebro WWTP was collected to analyze for the presence of CPE. Three separate water samples were collected in sterile 1-litre glass bottles from each location at different times of the day, transported and stored at 4 °C, and analyzed within 24 h.
CPE from a pre-existing collection at Örebro University Hospital were included in the study. The isolates were collected from patients during the years 2008 to 2015.
Bacterial isolation and identification
Water samples were filtered through 0.45 μm polyethylene sulfonate membrane filters (Sartorius Stedim Biotech, Sweden) and placed on selective chromogenic agar medium for carbapenem-resistant bacteria, chromID™ CARBA (bioMérieux, Marcy-l’Etoile, France) and chromID™ OXA-48 (bioMérieux, Marcy-l’Etoile, France). The agar plates were incubated for 18 h at 37 °C, and suspected CPE colonies were picked and streaked onto Chromocult® Coliform Agar (Merck, Darmstadt, Germany) to further identify the isolates as Enterobacteriaceae and to obtain pure cultures.
Isolates were identified using matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) with Microflex LT system (Bruker Daltonik GmbH, Bremen, Germany) according to manufacturer’s instructions.
Analysis of carbapenemase production
Putative carbapenemase producers were screened for their susceptibility to meropenem and imipenem using the disk diffusion method according to the EUCAST specifications. The isolates with inhibition zone diameters of ˂ 25 mm and ˂ 23 mm for meropenem (10 μg) and imipenem (10 μg), respectively, were considered presumed carbapenemase producers .
Presumed carbapenemase producers were further tested for their carbapenemase production using RAPIDEC® CARBA NP test (BioMérieux, Marcy-l’Etoile, France) according to manufacturer’s instructions .
Antibiotic susceptibility testing
The antibiotic susceptibility of isolates was tested for six different classes of antibiotics: beta-lactams (imipenem, meropenem, cefotaxime, ceftazidime, piperacillin-tazobactam), aminoglycosides (gentamicin, amikacin, tobramycin), fluoroquinolone (ciprofloxacin), trimethoprim-sulfamethoxazole, tetracycline (tigecycline), and polymyxin B (colistin). The disk diffusion method was used according to the specifications of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) . Minimum inhibitory concentrations were determined using ETEST® (bioMérieux, Marcy-l’Etoile, France) following manufacturer instructions. The plates were incubated at 37 °C for 18 h. EUCAST clinical breakpoint values for zone diameters (mm), and the MIC values were used to categorize the isolates as susceptible (S), intermediate (I), and resistant (R) .
DNA isolation and whole genome sequencing
Genomic DNA of Klebsiella spp. was isolated from an overnight culture in nutrient broth (Merck, Germany) using guanidinium thiocyanate-phenol-chloroform extraction method . DNA samples were sent to GATC Biotech (Konstanz, Germany) for whole genome sequencing (WGS). Short paired-end reads of 150 bp were generated using Illumina HiSeq. Quality trimming of the paired-end reads was performed with Trimmomatic version 0.32.3 . De novo genome assembly was performed using SPAdes Genome Assembler version 3.10.1 . The sequences have been submitted to GenBank ID: 451179 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA451179).
Multilocus sequence typing and phylogenetic analysis
MLST was performed using MLST-1.8 Server provided by Center for Genomic Epidemiology , and sequence types (STs) were confirmed using SRST2 . For genomic comparisons, we performed core genome SNP analysis and core genome gene-by-gene analysis. To increase the shared sequences for analysis, only K. oxytoca were included in the analysis. Phylogenetic maximum likelihood tree based on core genome SNPs analysis was generated in Parsnp  using the default parameters. K. oxytoca JKo3 was selected as the reference genome. Neighbor-joining tree based on core genome MLST (allelic comparison) was performed in SeqSphere+ (Ridom Muenster, Germany) using default parameters, and K. oxytoca JKo3 was selected as the reference genome. Phylogenetic trees were reconstructed in FigTree version 1.4.3 (Institute of Evolutionary Biology, University of Edinburgh). To provide a wider phylogenetic context, we included nine epidemiologically unrelated K. oxytoca genomes in the analysis. The assembled genomes were randomly selected from Ensembl bacterial genome database (bacteria.ensembl.org). The strain names and archive numbers are in Supplementary Table S1.
Antibiotic resistance gene profiling
The antibiotic resistance gene profiling of isolates was performed with oligonucleotide microarray-based assay developed by Alere Technologies [28, 29]. This was complemented with allelic identification of the antibiotic resistance genes (ARGs) from the assembled genomes using Resfinder, server 2.1 . For microarray-based identification of ARGs, fresh cultures from Mueller Hinton agar (BD, LePont de Claix, France) were inoculated to AMIES agar gel tubes (Copan, Brescia, Italy) and sent to Alere Technologies (Germany) for genotyping. The antibiotic resistance gene profiles of both clinical and environmental isolates were analyzed for clinically important carbapenemase genes along with narrow spectrum β-lactamase, ESBL, and other antibiotic resistance genes. The assay included the following carbapenemase genes, blaKPC, blaVIM, blaNDM, blaBIC, blaDIM, blaGES, blaGOB, blaPAM, blaSFH, blaSMB, blaSME, blaSPM, blaTMB, blaGIM, blaIND, blaKHM, as well as blaOXA-23, blaOXA-40, blaOXA-48, blaOXA-50, blaOXA-51, blaOXA-54, and blaOXA-58, and the following ESBL genes, blaCTX-M-1/M-15, blaSHV and blaTEM. In addition, genes conferring resistance to aminoglycosides (e.g., aac, aad, ant2, aphA, strA, strB), quinolones (e.g., qepA, qnrA, qnrB, qnrC, qnrD, qnrS), macrolides (mph, mrx), sulfonamides (sul1, sul2, sul3), and trimethoprim (e.g., dfrA genes) were evaluated (Alere Technologies, http://www.alere-technologies.com). Four genes related to two bacterial toxin-antitoxin systems higB-higA and splT-splA were also included in the assay.
Mass spectra analysis
A loop (approx. 1 μl) of freshly grown bacteria was suspended in ethanol (900 μl, 99.5%), centrifuged at 11,000×g for 2 min, and the pellet was air-dried at room temperature for 3 min. The pellet was incubated in 70% formic acid (20 μl) for 3 min at room temperature to lyse the cells and the released proteins were isolated with acetonitrile (20 μl). The mixture was vortexed and centrifuged at 11,000×g for 2 min. The supernatant was spotted onto a MALDI-TOF MS target plate with Bruker Matrix solution (1 μl). MALDI-TOF MS plates were air-dried at room temperature, and mass spectrometry was performed. Two replicate analyses were performed for each bacterial isolate.
The mass spectra of the bacterial proteins of 2000–20,000 mass-to-charge ratio (m/z) were analyzed using BioNumerics version 7.5 created by Applied Maths NV (http://www.applied-maths.com). Default parameters for strict pre-processing of spectrum data using baseline subtraction, noise elimination, and curve smoothing were selected. Similarity comparison was performed with peak-based Pearson coefficient using default parameters, and phyloproteomic dendrogram was created.
The datasets generated during the current study have been deposited in GenBank ID: 451179 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA451179).