A number of seven different German WWTPs were investigated. These WWTPs possess different population equivalents with WWTP-1 440,000, WWTP-2 210,000, WWTP-3 23,500, WWTP-4 46,000, and WWTP-5 26,000, WWTP-6 45,000, and WWTP-7 8,000 (Table 1). All WWTPs performed mechanical separation, followed by activated sludge treatment, and sedimentation steps.

Whereas, WWTP-1 and WWTP-2 treat wastewater from urban areas including clinical, household, and industrial wastewaters, the wastewaters of the other 4 WWTPs are mainly affected by agricultural catchment areas including animal and food farming. WWTP 7 only treats municipal wastewater without affection of agriculture or industry.

Sampling points were arranged at the influents of WWTPs after mechanical separation and at the effluent sites after sedimentation tanks before the purified wastewater is released into the receiving bodies.

Volumes of about 100 mL from the 24 h-composite of the WWTP influent samples and about 300 mL of the WWTP effluent samples from the 24 h-composite sample were used for DNA extraction. Wastewater samples were filtered using polycarbonate membranes with a diameter of 47 mm and a pore size of 0.2 μm (Whatman Nuclepore Track-Etched Membranes, Sigma-Aldrich, Munich, Germany). DNA extraction was performed using the Fast DNA spin kit for soil (MP Biomedical, Illkrich, France) utilizing the lysing matrix E according to the manufacturer's protocol for wastewater. The quantities and purities of the DNA extracts were measured by means of the Qubit fluorimetric quantitation (Thermo Scientific, Waldham, USA).

The yield of DNA obtained from the influent water samples ranged from 71 to 244 μg per mL (n = 14) and from 15 to 143 μg per mL (n = 14) for the effluent water samples. To determine the abundances of 7 clinically relevant antibiotic resistant genes including the mcr-1 colistin resistance gene, and the 16S rRNA gene as well as the gene markers for some Enterobacteriaceae qPCR analyses were performed.

All primers and references are listed in Table 2, which do also contains the accuracy (R²) as well as efficiency (%) values for each qPCR detection system. More specifically, the Escherichia coli NRZ-14408 reference strain is carrying the mcr-1 colistin resistance gene and was kindly provided from National Reference Center in Bochum, Germany. For mcr-1 primer design the NCBI Primer BLAST software was used. For the quantification of the mcr-1 gene in environmental water samples a colistin-resistant E. coli NRZ-14408 (hospital isolate, mcr-1 positive) reference bacterium was used to create a calibration curve, which is shown in Supplementary Material. An R² value of 0.999 and an efficiency of 97.6% were determined, indicating the high specificity of the mcr-1 detection system.

The antibiotic resistance genes blaTEM, CTX-M, CTX-M-32, and CMY-2 are directed against ß-lactam antibiotics, whereas the tetM resistance gene is coding for the resistance against tetracycline and the ermB gene mediates the resistance against erythromycin. All primers, reference strains, and quality values of each detection system are listed in Table 2. In addition calibration and melting curves are given in Supplementary Material. The abundances of these ARGs were quantified in all WWTP effluent wastewater samples via qPCR approach using different reference strains carrying the mentioned resistance genes.

The yccT gene present in E. coli DSM 1103 was used as taxonomic gene marker as well as the gltA gene from K. pneumoniae DSM 30104 (Clifford et al., 2012). Total DNA of pure culture was extracted with the DNA extraction kit for soil (MP Biomedical, Illkrich, France) and subsequently used for the generation of target specific calibration curves. 16S rRNA primers were used to quantify eubacterial rDNA in the water samples for normalization. Here, an already cloned fragment of the eubacterial 16S rRNA gene in the pNORM plasmid of E. coli DH5α was used (Stalder et al., 2014). Plasmid-DNA was extracted with the GeneJet Plasmid Miniprep Kit (Thermo Scientific, Waldham, USA) and was used for the generation of the calibration curve which in turn was used for the calculation of the 16S rRNA gene copy number in water samples.

The mcr-1, ermB, tetM resistance genes, and the four ß-lactamase genes (CTX-M. CTX-M-32, CMY-2, blaTEM), the ribosomal 16S rRNA gene for Eubacteria, and the specific taxonomic gene markers of E. coli, K. pneumonia, and Acinetobacter baumannii were quantified in a SYBR Green qPCR approach. Reactions were run in volumes of 20 μL, containing 10 μL Maxima SYBR Green/ROX qPCR Master Mix (2x) (Thermo Scientific), 8.2 μL of nuclease-free water (Ambion, Life technologies, Karlsbad, Germany), 0.4 μL of the respective primers (stock concentration 10 μM, Table 2), and 1 μL of template DNA (20 ng μL⁻¹). The qPCR protocol comprised 10 min at 95°C for activation of the DNA polymerase followed by 40 cycles of 15 s at 95°C and 1 min at appropriate temperature for primer annealing and elongation (see Supplementary Material). Each water sample was analyzed in technical triplicates. To determine the specificity of the amplification, a melting curve was recorded by raising the temperature from 60 to 95°C (1°C every 10 s). Data analysis was performed by using the Bio-Rad CFX Manager software.

To calculate the gene copy number of the respective antibiotic resistance gene, the 16S rRNA gene, and taxon-specific gene marker, reference strains carrying the genetic targets of interest were used. With regard to the already known genome sizes of the reference bacteria it is possible to calculate the cell copies. The following equation with an average molecular weight for one base pair of about 650 g/mol, Avogardo's number with 6.022 × 10²³ molecules/mol, and a converting factor of 10⁹ ng/g was used.

Serial dilution of DNA stocks were prepared and used for calculation of the calibration curves starting with 500 ng DNA (see Supplementary Material). The coefficient of determination of all standard curves was above 0.994 in all experiments (Table 2), indicating minimal variability within the linear data range.

Using these curves, the measured Ct values of the mcr-1 gene, the 16S ribosomal RNA gene or the taxon-specific gene markers from water samples can be used for copy number calculations (Alexander et al., 2015, 2016). The abundances of antibiotic resistance genes (ARGs) and opportunistic bacteria were quantified in each water sample and normalized to 100 ng total extracted DNA (cell equivalents per 100 ng DNA). In addition, the qPCR data was also normalized to the 16S rRNA gene abundances of the corresponding water samples.

In total 14 biological samples from the influents and effluents of 7 WWTPs were obtained from two individual sampling campaigns in autumn and spring time. For data presentation the standardized box plot diagram was used displaying the distribution of data based on the five number summaries: minimum, first quartile, median, third quartile, and maximum. In the box plot the central rectangle spans the first quantile (P = 25) to the third quantile (P = 75). A segment inside the rectangle shows the median and whiskers above and below the box show the minimum and maximum values. Thus, the box plot displays the full range of variation (from min to max), the likely range of variation, and a typical value (the median). In addition, student's t-test calculations were performed to identify the significance of reduction of gene targets during conventional wastewater treatments.

PMA can enter dead or injured bacteria due to the loss of their membrane integrities, intercalate with the intracellular and extracellular DNA. The presence of the PMA-DNA complex blocks the polymerase activity at the target sites. In consequence, no PCR product is generated. The protocol is in accordance with that of our previous study (Villarreal et al., 2013) and based on the publications of Nocker et al. (2007a,b) and Nocker and Camper (2009), where the method was optimized also for natural mixed populations. To separate the living bacterial population from dead or injured bacteria during conventional wastewater treatment, the PhAST Blue Photo-Activation System (GenIUL, Barcelona, Spain) was used. After filtration of the influent and effluent wastewater samples originating from WWTP 1 (n = 3), polycarbonate membranes (Nucleopore) were submerged in 300 μL of a 25 μM propidium monoazide (PMA; Biotium, Hayward, California, USA) solution and put into a 1.5 mL colorless tube (SafeSeal tubes, Carl Roth, Karlsruhe, Germany). After incubation in the dark at 4°C for 5 min, samples were exposed to the LED light of the photoactivation system at 100% intensity for 15 min. After PMA treatment, samples were prepared for DNA extraction using the Fast DNA spin kit for soil according to the manufacturer's protocol for wastewater.

The mcr-1 amplicons were separated on a 1% agarose gel to examine the primers and PCR protocols for proper specificity. The results demonstrated that mcr-1 positive water samples generated amplicons with the expected length of 183 bp (Figure 2). Sequence analyses confirmed the mcr-1 identity (see Supplementary Material). For that, the mcr-1 PCR products were purified with ExoSAPit (GE Healthcare, Munich, Germany) before sequencing. The sequencing reactions were performed in a 10 μL reaction volume, including 2 μL of Premix (BigDye® Terminator v1.1 Cycle Sequencing Kit, Applied Biosystems), 0.5 μL primer 517R (10 pM), and 1 μL of purified, 1:3 diluted PCR product. This volume was filled up to 10 μL with sterile water. The temperature profile of the sequencing reaction was as followed: 5 min at 96°C, 25 cycles of 10 s at 96°C, 5 s at 55°C, and 1 min at 60°C, afterwards cooling down to 4°C. For deleting excessive dye, the sequence products were purified with DyeEx2.0 Spin Kit (Qiagen, Hilden, Germany). Three microliters of this product were added to 12 μL of Hi-Di-Formamid (Applied Biosystems) and analyzed in the ABI Prism 310 Genetic Analyser (Applied Biosystems) using POP4 Polymer (Applied Biosystems) and 47 cm × 50 μm capillaries (Applied Biosystems) according to the manufacturer‘s instructions. For database analyses, NCBI (National Center for Biotechnology Information) Blast alignments were performed.

The accuracy of the novel qPCR detection system targeting the mcr-1 colistin resistance gene in wastewater samples was demonstrated with the extracted total DNA of original water samples. Here, total DNA from the WWTP-1 with the highest population equivalent of 440.000 was used. Neither irregular amplification results were observed during qPCR nor unexpected shoulders seen during melting curve analyses (Figures 1A,B). After qPCR amplification the amplicons were separated in an agarose gel and resulted in one specific DNA band positioned at the expected product size of 183 bp (Figure 2). The amplicons were also sequenced. The evaluation of the BLAST search confirmed the 99% identity with the mcr-1 resistance gene from E. coli (LC191581), K. pneumoniae (KX377410), Salmonella enterica (KX257482), and Cronobacter sakazakii (KX505142) strain. As a first result of this study it could be shown that this newly developed qPCR system is suited for the reliable quantification of the mcr-1 gene in wastewater populations.

The Ct-values from qPCR were used for cell equivalent calculations normalized to 100 ng total DNA, which is a relative quantification related to the population microbiome. It is independent from the sample volume. Additionally, the cell equivalents from the calibration curve were normalized to the 16S rRNA gene copies as an alternative eubacterial biomass gene marker. It became obvious that the mcr-1 gene was present in influent samples of WWTP-1 with a calculated abundance of 8.11 × 10¹ cell equivalents per 100 ng total DNA. The median values of cell equivalent per 100 ng DNA of the effluent water samples were quantified with 7.0 × 10¹. Alternatively, the cell equivalents were also referred to 16S rRNA gene copies and corresponded with 2.64 × 10⁻⁸ cell equivalents in the influent and 1.30 × 10⁻⁹ cell equivalents in the effluent water samples, respectively. The data from WWTP-1 is indicating the stable presence of mcr-1 gene within the microbiome of the wastewaters of this large WWTP.

Additionally, influent and effluent wastewater samples of WWTP-1 were treated with PMA to discriminate living bacteria from dead bacteria or extracellular DNA (eDNA) prior to DNA extraction. The presence of eDNA, which might be released from dead or injured bacteria, is known to be a promoting factor for horizontal gene transfer (HGT) (Davies et al., 2006; Aminov, 2011). In case of the colistin resistance the mcr-1 gene is located on a conjugative plasmid, which is already described to transfer the colistin resistance among Enterobacteriaceae (e.g., Liu et al., 2016). After normalization to 100 ng DNA and 16S rRNA gene copies slightly reduced cell equivalents were found in case of PMA treatment in influent and effluent samples (influent PMA treated: 6.2 × 10⁰ cell equivalents per 100 ng DNA and 8.92 × 10⁻⁸ per 16S rRNA gene copy; effluent PMA treated: 6.2 × 10⁰ cell equivalents per 100 ng DNA; and 3.66 × 10⁻⁸ cell equivalents per 16S rRNA gene copy). Comparing PMA-treated with untreated samples of the influent and effluent in maximum one order of magnitude difference became obvious. These results are indicating the presence of low amounts of eDNA or distinct numbers of injured/dead bacteria in wastewater samples, which might have a relevance for transformation and therefore HGT.

Nevertheless, it could be shown that the mcr-1 colistin resistance gene is present in water samples from WWTP-1 and still living bacteria carrying the mcr-1 gene are released to the receiving adjacent body.

As mentioned before, the influent and effluent wastewaters of 6 additional WWTPs were investigated for the abundance of the mcr-1 gene coding for the colistin resistance in Enterobacteriaceae. In contrast to WWTP-1 the population equivalents of these WWTPs were much lower ranging from 8,000 to 210,000 p.e. The mcr-1 gene was quantified in the influent wastewater samples of all six WWTPs. These mcr-1 positive tested WWTPs treat urban and rural wastewaters including hospitals, livestock, and food industries. Here, the cell equivalents ranged from 4.45 × 10¹ to 2.01 × 10² in 100 ng total DNA, and 9.89 × 10⁻⁸ to 3.44 × 10⁻⁷ per 16S rRNA gene copy number (Figure 3).

Compared to WWTP-1 (440,000 p.e.), the influent waters of these WWTPs showed higher abundances of the mcr-1 gene (Figure 3), which might result from the agricultural or food industry impacts. Even in influent samples of the smallest WWTP-7 (8,000 p.e.) mcr-1 gene copies were detected despite the fact that no hospitals and no intensive animal farms or food industry released wastewaters to this WWTP.

Besides WWTP-1, the mcr-1 gene was detected in the effluent water samples of WWTP-2 and WWTP-3 with 8.4 × 10⁰ to 1.3 × 10¹ cell equivalents per 100 ng total DNA and 3.88 × 10⁻⁷ and 1.55 × 10⁻⁷ per 16S rRNA.

Hence, the mcr-1 gene was detected both in the influent waters of all seven WWTPs and in the effluent water samples of WWTP-1, WWTP-2, and WWTP-3.

These WWTPs differed in population equivalents (Table 1), which gives hints that the dimension of the WWTPs does not significantly impact the survival and persistence of mcr-1 gene carrying bacteria. For the mcr-1 gene a maximum reduction of 1 Log scale was determined in WWTP-1, WWTP-2, and WWTP-3.

Besides the mcr-1 colistin resistance gene, 6 other clinically relevant antibiotic resistance genes were quantified in effluent wastewater samples of the 7 WWTPs. Both the relative quantification and the cell equivalents per 16S rRNA gene copy number are shown in Figures 4, 5. The targeted genes are the ermB gene coding for the erythromycin resistance, the tetracycline resistance gene tetM, and 4 different ß-lactam resistance genes (CTX-M-32, blaTEM, CMY-2, and CTX-M). All resistance genes were distinctly present in all wastewater effluents samples. The most abundant resistance gene was ermB with a median value of 2.39 × 10⁵ cell equivalents in 100 ng DNA or 3.08 × 10⁻³ cell equivalents per 16S rRNA gene copy number. The second most abundant gene was the tetracycline resistance gene tetM with 1.26 × 10⁴ cell equivalents in 100 ng DNA and 1.68 × 10⁻⁴ per 16S rRNA gene copy number. The abundances of the blaTEM, and CTX-M-32 genes were found in a similar range slightly below the tetM gene. The median values for the CMY-2 and CTX-M ß-lactamase genes were detected in a lower abundance of 1.0 × 10² in 100 ng DNA and 8.7 × 10⁻⁷ per 16S rRNA gene copy number.

In comparison to mcr-1 gene abundances, these resistance genes were much more frequently found and were quantified in higher concentrations in all wastewater effluents released to the aquatic environment. Data from the influent samples of WTTPs demonstrated a reduction of the mentioned resistance genes ranging from 1 Log to maximum of 2 Logs during wastewater treatment (data not shown), which is in accordance with the data targeting the mcr-1 gene.

The abundance of specific taxonomic gene markers was quantified targeting A. baumannii and the Enterobacteriaceae K. pneumoniae and E coli. These three bacterial species are described to belong to the network of horizontal gene transfer with clinical relevance. All of them are opportunistic pathogens and were shown to hold resistances to colistin by previous studies (Hua et al., 2017; Jeannot et al., 2017). To address the growing concern of emerging colistin resistant pathogens, the abundance of these possible carrier bacteria are also analyzed in the German WWTP of this study.

The Figures 6, 7 summarize the data of all seven investigated WWTPs (influent/effluent) according to the two different normalization approaches, i.e., per 100 ng total DNA (Figures 6A–C) and per 16S rRNA gene copy (Figures 7A–C). Both figures show the abundances of taxonomic gene markers specific for A. baumannii (secE), K. pneumoniae (gltA), and E. coli (yccT). Calibration curves and melting curve analyses for each parameter are documented in Supplementary Material. The obtained Ct values matched with the linear ranges of the calibration curves.

The median values of the influent samples were calculated for the specific taxonomic genes with 1.12 × 10⁴ for A. baumannii, 2.15 × 10⁴ for K. pneumoniae, and 2.96 × 10⁴ for E. coli cell equivalents per 100 ng total DNA. Decreased median values per 100 ng DNA resulted from the effluent samples for all three target genes, i.e., 1.05 × 10² for A. baumannii, 1.60 × 10³ for K. pneumoniae, and 1.80 × 10³ for E. coli (Figure 6). In total, the abundances of these taxonomic gene markers decreased during conventional wastewater treatment with 1–2 orders of magnitudes, but are still present in significant amounts in effluent samples.

Figure 7 illustrates the abundances of taxonomical marker genes referred to the 16S rRNA copy number. The median values were analyzed and resulted in 2.17 × 10⁻⁵ cell equivalents for A. baumannii, 8.25 × 10⁻⁵ for K. pneumoniae, and 1.3 × 10⁻⁴ for E. coli in the influents samples. The investigation of cell equivalents in the effluent samples resulted in a median of 2.60 × 10⁻⁶ for A. baumannii, 5.86 × 10⁻⁵ for K. pneumoniae, and 3.29 × 10⁻⁵ for E. coli.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.