Linking wet-lab and genomic approaches for reliable detection of carbapenemase-producing Klebsiella pneumoniae in wastewater

Abstract

The World Health Organization (WHO) has designated carbapenemase-producing Klebsiella pneumoniae (CP-KP) as a critical priority pathogen due to its increasing importance for human health. Although wastewater-based surveillance (WBS) is discussed as a complementary tool for classical systems regarding hazard forecasting and early-warning, methods targeting a single AMR target remain scarce. Therefore, we applied an integrated “wet-lab to genomics” workflow to support reliable target detection of CP-KP in raw influent wastewater samples as a first step towards WBS. The CP-KP screening workflow was set up based on membrane filtration, selective chromogenic media for selective cultivation and the modified carbapenem inactivation method (mCIM) for confirming carbapenemase-production using 33 samples from four different wastewater treatment plants in North-Eastern Germany. All samples tested positive for CP-KP, with calculated concentrations ranging between 10² and 10⁴ colony-forming units (cfu) per 100 mL across the sample set. A total of 320 isolates belonged to the target group of enterobacteria (i.e. Klebsiella, Enterobacter, and Citrobacter), as defined by the manufacturer of the chromogenic media, with the majority being identified as KP (n = 297; 93%), including n = 253 (79%) verified CP-KP. Genotypic characterization of CP-KP by PCR revealed the predominance of bla_OXA-48-related genes (n = 83) among isolates from all WWTPs. As quality parameters, colony counts for viable Escherichia coli (EC) were employed as a proxy for valid wastewater samples and extended-spectrum beta-lactamase-producing E. coli (ESBL-EC) as indicator for AMR, with cfu/100 mL ranges from 10⁶ to 10⁷ and 10² to 10⁴, respectively. To verify the screening outcome, a subset of 58 CP-KP from two WWTPs were subjected to whole genome sequencing (WGS). As a result, eight different sequence types (STs), i.e., ST147 and ST273 (both: clonal group 147), ST258, ST35, ST15, ST37, ST307, and ST485 were identified. These include clinically relevant STs clustering closest with fecal isolates from Germany when compared with Pathogenwatch-database entries. Moreover, WGS data enabled the identification of antibiotic resistance genes (ARGs), and the detection of closely related isolates within the WWTP dataset.

1. Introduction

According to the World Health Organization (WHO), “antimicrobial resistance (AMR) threatens the effective prevention and treatment of infectious diseases caused by bacteria, parasites, viruses and fungi” (WHO, 2025). The WHO further declares that “AMR occurs when bacteria, viruses, fungi and parasites change over time and no longer respond to medicines making infections harder to treat and increasing the risk of disease spread, severe illness and death” (WHO, 2025). Thus, AMR represents an acquired phenotypic trait of a pathogen, which is directly linked to the treatment of infectious diseases in humans, animals, and plants (WHO, 2024a). According to Article 17 of the revised European Urban Wastewater Treatment Directive (2024), mandatory AMR monitoring in urban wastewater is required for agglomerations with 100,000 population equivalents or more in Europe (European Parliament and Council of the European Union, 2024).

Given the development of methods for wastewater-based surveillance (WBS) of AMR, it remains vital to acknowledge the natural evolution of antibiotics and antimicrobial resistance throughout the environment. Briefly, a wide range of antibiotics represent natural compounds, produced by environmental microorganisms, including—but not limited to—soil bacteria such as Streptomyces and other Actinomycetota, as well as multiple fungi (e.g., Penicillium spp.). Although their role within a specific habitat is still not fully understood (Clardy et al., 2009), the general concept of producing antimicrobial compounds appears to represent a successful evolutionary trait for microorganisms, since these active metabolites have likely existed for millions of years, according to current estimates (Hutchings et al., 2019; Wright et al., 2012). Antimicrobial resistance gene (ARG) evolution is closely tied to the natural biosynthesis of antibiotics, and “self-protection” motives of the producing cell have been discussed as a potential driver since the early 1970s (Hopwood, 2007).

However, many environmental bacteria possess intrinsic genetically determined resistance to different classes of antibiotics. The entirety of all intrinsic antibiotic resistances, i.e., the “intrinsic resistome,” is a naturally occurring phenomenon that predates antibiotic chemotherapy (Georgina Cox, 2013; Olivares et al., 2013). The mobile resistome, on the other hand, encompasses ARGs that gained mobility via horizontal gene transfer (Partridge et al., 2018), a process heavily influenced by anthropogenic activities (Gentile et al., 2024; Lobanovska and Pilla, 2017; Lowy, 2003). ARGs conferring reduced susceptibility to antibiotics may be spread between bacteria inhabiting different ecosystems, encompassing environmental, commensal, and, last but not least, (facultative) pathogenic bacterial species that cause infectious diseases (Tessa Marciniak et al., 2025).

Consequently, monitoring AMR within urban wastewater, which represents a classical interface within the One Health continuum, depends on several key considerations.

The target for AMR monitoring should be relevant to public health as defined by national and international health agencies (European Parliament and Council of the European Union, 2024; Sati et al., 2025; WHO, 2024b). In 2021, the WHO Tricycle Protocol adopted the One Health approach using an AMR indicator organism, namely extended-spectrum beta-lactamase (ESBL)-producing E. coli (ESBL-EC), to streamline global surveillance efforts (WHO, 2021). Likewise, monitoring of a single target seems feasible compared to an assortment of AMR pathogens and genes (Appling et al., 2023). The importance of sensitive methods for targeting rare subpopulations such as carbapenemase-producing K. pneumoniae (CP-KP) was addressed only recently (Bobis Camacho et al., 2024), and a systematic review on AMR wastewater surveillance declared clear and consistent reporting of study methods as a prerequisite to identify optimal practice (Chau et al., 2022).

Moreover, assessing AMR frequencies in terms of species exhibiting defined resistance phenotypes, e.g., carbapenemase-producing (CP) Enterobacterales, enables the identification of novel resistance mechanisms, variants, and combinations as well as sudden changes in resistance profiles (Blaak et al., 2021; Flach et al., 2021; Tiwari et al., 2022b). In recent years, WBS of AMR and especially its potential targets have been a subject of a broad scientific discussion, including intestinal bacteria such as Enterobacterales (Aguilar-Bultet et al., 2025; Blaak et al., 2021; Gómez-Sanz et al., 2023; Tiwari A. et al., 2024; Urzua-Abad et al., 2024), specifically Escherichia coli (EC) (Al-Mustapha et al., 2025; Conforti et al., 2024; Heljanko et al., 2024; WHO, 2021), and Klebsiella pneumoniae (KP) (Bobis Camacho et al., 2024; Heljanko et al., 2024; Radisic et al., 2023). In this context, sequencing approaches such as whole genome sequencing (WGS) of bacterial isolates and shotgun metagenomics are essential in supporting the detection and characterization of microbial species and ARGs from wastewater samples (Lepuschitz et al., 2019; Nguyen et al., 2021; Pillay et al., 2025; Pruden et al., 2021; Silvester et al., 2025; Tiwari et al., 2022a).

In addition, resident environmental microorganisms living in biofilm structures seem fully adapted to the challenges associated with the “wastewater system” habitat, including undulating nutrient accessibilities, the presence of selective agents (i.e., surfactants, biocides, antibiotics, heavy metals, pesticides, drugs and many others), variable oxygen and ammonia concentrations, as well as local temperature fluctuations. Consequently, resident (environmental) antibiotic resistant bacteria (ARB) such as Pseudomonas spp., Acinetobacter spp., and Aeromonas spp. harbor a broad range of mechanisms conferring tolerances and resistances, allowing for survival in these harsh conditions (Saati-Santamaría et al., 2022; Zhao et al., 2023; Zhong et al., 2019). Since bacteria of the human microbiota, as well as opportunistic pathogens capable of infecting humans, are typically well adapted to the conditions provided by warm-blooded hosts, their sheer survival in environments such as wastewater systems seems a considerable challenge (Fahrenfeld and Bisceglia, 2016; Guo et al., 2022; Korajkic et al., 2019).

Besides chemical stressors, a wide array of microorganisms, many of which are specifically adapted to the wastewater environment, likely induce continuous changes in the microbial community composition of wastewater systems (Fahrenfeld and Bisceglia, 2016; Ferreira et al., 2023; Korajkic et al., 2019; Poulsen et al., 2021). Lower temperatures are likely to favor the proliferation of cold-adapted resident environmental bacteria, which frequently carry ARGs, such as Pseudomonas spp. [reviewed in Tribelli and López (2022)], Aeromonas spp. [reviewed in Figueras Salvat and Ashbolt (2019), Lamy et al. (2022), and McLellan and Roguet (2019)], and Acinetobacter spp. [reviewed in Adewoyin and Okoh (2018), McLellan and Roguet (2019), and Muleshkova et al. (2025)], or Gram-positive spore-forming bacteria [reviewed in McLellan and Roguet (2019)]. Consequently, storage at 4 °C overnight prior to processing wastewater samples might introduce process-associated bias regarding detection rates of ARGs and human-associated bacteria (including antimicrobial resistant strains).

The aim of this proof-of-concept study is to provide a “hands-on” procedure to quantify a selected AMR target (CP-KP) in wastewater samples alongside two indicators for urban wastewater (EC) and AMR viability (ESBL-EC) as sample quality parameters. In a second step, molecular methods were applied to characterize the primary target (CP-KP) including WGS data, in order to assess wet lab performance together with strain diversity and local dynamics. It should be noted that this study does not include epidemiological interpretation of WGS data in terms of public health surveillance purposes, i.e., target detection, assessment of results, communication and public health response (McNabb et al., 2002; WHO, 2026). Rather, the focus lies on the first step, the method development and workflow for reliable target detection (Liguori et al., 2022) by integrating different scientific disciplines.

2. Materials and methods

2.1. Sampling

Between July and November 2024, raw influent wastewater samples were collected from four different wastewater treatment plants (WWTP) in North-Eastern Germany. The sampling process was recently described in detail (Fretschner et al., 2025). Briefly, twenty-four-hour composite samples of primary clarified wastewater were transported via temperature-controlled shipment (4 °C) to the Microbiological Risks unit at the German Environment Agency (Berlin) and immediately further processed (within ten hours after sampling). Upon arrival, pH and temperature were measured, followed by sample homogenization via gentle inversion (overhead shaker reax20, Heidolph Scientific Products GmbH, Germany) for 15 min.

2.2. Pretesting: evaluation of two distinct temperature regimes for the detection of target colonies on selective chromogenic media

The study started with a pretest to identify the requirements for reliable KP detection in wastewater, a matrix that is, by nature, heavily loaded with non-target bacteria. Hence, a comparative analysis following previously published incubation regimes [ISO (2001) as cited in Blaak et al. (2021) and WHO (2021)] was conducted using influent wastewater samples collected throughout three consecutive weeks at two WWTPs. Briefly, wastewater-associated bacteria recovered by placement of loaded filters on selective media were subjected to either an incubation at 36 ± 2 °C for 18 ± 2 h, or a 4-5 h-incubation at 36 ± 1 °C followed by a 21 ± 3 h incubation at 44 ± 0.5 °C. For CP-KP detection as the primary target, 0.5 mL and 1 mL undiluted wastewater were used. For the detection of EC and ESBL-EC, a decadic serial dilution starting with 2 mL of influent wastewater (ad. 20 mL) was used as described in detail (Figure 1). Based on previous experiences regarding suitable amounts of wastewater resulting in countable plates, 0.1 and 0.5 μL volumes (≙ 1 mL 10⁻⁴ and 0.5 mL 10⁻³ wastewater dilutions) were used to assess concentrations for EC, and 50 and 100 μL for ESBL-EC (≙ 0.5 mL 10⁻¹ and 1 mL 10⁻¹ wastewater dilutions). Each volume was augmented with Phosphate-buffered saline (PBS) (Life Technologies Limited, Paisley, UK) to a final volume of 30 mL (pH ~ 7.4) to achieve an appropriate volume for membrane filtration. Filtration was performed in triplicate for each target and wastewater volume investigated, yielding six selective plates per target and incubation regime for each sample. To test for significant differences between the incubation regimes, a two-tailed, paired Student’s t-test was conducted. Statistical significance was defined as a p-value below 0.05, p-values below 0.001 were considered highly significant.

Figure 1.

Graphical summary of the experimental procedure identifying CP-KP, EC, and ESBL-EC in influent wastewater. A serial dilution was prepared from influent wastewater using PBS (10⁻¹ to 10⁻⁵) (A). Different volumes were filtered to detect presumptive CP-KEC, EC, and ESBL-EC (B). To ensure uniform bacterial distribution on the respective filters, volumes were supplemented with PBS to a final volume of 30 ml (C). Loaded filters were placed on chromogenic media as follows: CHROMagar^™ mSuperCARBA (CP-KP), CHROMagar^™ Orientation (EC), and CHROMagar^™ ESBL (ESBL-EC) (D). Following incubation, target colonies were quantified according to the manufacturer’s instructions. Presumptive CP-KEC grown on CHROMagar^™ mSuperCARBA were subjected to species identification and further characterization. Up to ten representative presumptive ESBL-EC colonies from at least two plates were used for species verification (E). PBS, phosphate-buffered saline; CARB, CHROMagar^™ mSuperCARBA; ESBL, CHROMagar^™ ESBL; ORIE, CHROMagar^™ Orientation; CP-KP, carbapenemase-producing K. pneumoniae; ESBL-EC, extended spectrum beta lactamase E. coli; EC, E. coli; V_D, filtrated volume of respective dilution; V_WW, absolute volume of raw wastewater on filter; CP-KEC; carbapenemase-producing Klebsiella-Enterobacter-Citrobacter species.

2.3. Screening procedure for the target organism (carbapenemase-producing Klebsiella pneumoniae) and fecal indicators (Escherichia coli, ESBL-Escherichia coli)

2.3.1. Carbapenemase-producing Klebsiella pneumoniae

A screening procedure was established, with KP positive for carbapenemase-production (CP) as the primary target (CP-KP; Figure 1). Samples representing four WWTPs were analyzed, with samples from WWTP 1 and 2 collected between weeks 31 and 48 of 2024 and samples from WWTP 3 and 4 collected between weeks 42 and 48 of 2024.

To identify CP-KP, 0.25, 0.5, 1, and 2 mL of wastewater were supplemented with PBS to a final volume of 30 mL. Subsequently, the samples were filtered using a SolarVac 601 MB filtration device (Rocker Scientific Co., Ltd., Taiwan) together with the Alligator 200 pump (Rocker Scientific Co., Ltd., Taiwan). The resulting four filters (cellulose nitrate filters 0.45 μm, Sartorius AG, Göttingen, Germany) were aseptically placed on CHROMagar^™ mSuperCARBA, a chromogenic selective agar targeting carbapenem-resistant Enterobacteriaceae (MAST Diagnostica GmbH, Reinfeld, Germany). Incubation at elevated temperatures was carried out as described above [ISO (2001) as cited in Blaak et al. (2021)].

According to the manufacturer’s guidelines, a metallic blue colony phenotype is a typical characteristic of enterobacteria belonging to the group consisting of Klebsiella spp., Enterobacter spp. and Citrobacter spp. (“KEC”-group; Figure 1E). Consequently, colonies associated with a metallic blue appearance on CHROMagar^™ mSuperCARBA plates were recorded as presumptive CP-KEC and selected for further processing.

Presumptive CP-KEC were enumerated following the principles for colony selection stated in DIN EN ISO 8199:2008 (DIN, 2008). In brief, 10 metallic blue colonies distributed across two of the four plates per sample were isolated by subculturing on columbia blood agar plates (Life Technologies GmbH, Darmstadt, Germany). Up to seven target colonies were isolated from a single plate. Plates were excluded if bacterial overgrowth occurred or if more than 200 colonies resembling the typical target colony were present, as described before (Blaak et al., 2021).

To verify species identity, presumptive CP-KEC (including CP-KP) were investigated by use of MALDI-TOF MS (Biotyper® Sirius, Bruker Daltronics GmbH & Co. KG, Bruker Corporation, Massachusetts, United States) in combination with flexControl (version 3.4) and MBT Compass HT software programs (version 5.1.300) along with the MBT Compass Library (version 12.0.0.0) and the BTyp 2.0-Sec-Library (version 1.0) according to the manufacturer’s instructions (Bruker Daltronics GmbH & Co. KG, Bruker Corporation, Massachusetts, United States).

2.3.2. Fecal and AMR quality control indicators

Two quality control indicators were used to complement the CP-KEC assessment, EC (wastewater) and ESBL-EC (viable AMR) following recommendations from the WHO in the “Tricycle protocol” (WHO, 2021). EC and ESBL-EC screening was performed using CHROMagar^™ Orientation and CHROMagar^™ ESBL, respectively (MAST Diagnostica GmbH, Reinfeld, Germany). For each indicator, three decadic dilution steps were utilized: 10⁻³ to 10⁻⁵ for EC, and 10⁻¹ to 10⁻³ for ESBL-EC. Since the difference between the decadic dilution steps is too high to achieve the required number of countable plates, two amounts of each of the three dilutions were taken for analysis, i.e., 0.5 mL and 1 mL, resulting in six plates per target and wastewater sample analyzed. These volumes (0.5 mL and 1 mL) were supplemented with PBS to a final volume of 30 mL to facilitate membrane filtration using 0.45 μm cellulose nitrate filters. Thus, the final wastewater volumes analyzed were 0.005, 0.01, 0.05, 0.1, 0.5, and 1 μL for EC, and 0.5, 1, 5, 10, 50, and 100 μL for ESBL-EC. Additional details are provided in Figure 1B.

Confirmation of ten presumptive ESBL-EC per sample was conducted by means of MALDI-TOF MS as described before. To verify the EC counts on CHROMagar^™ Orientation, presumptive EC colonies were collected during an eight-week sampling period and identified using MALDI-TOF MS. The number of false-positives was low, i.e., 0.8% (2/240), indicating that the misclassification rate is negligible when colonies are counted based solely on their morphology on the chromogenic agar. Subsequent identification of EC was therefore limited to colony phenotype appearances.

2.4. Antimicrobial susceptibility testing of Klebsiella pneumoniae

Verified KP isolates were subjected to antibiotic susceptibility testing (AST) using a VITEK 2 COMPACT system (Biomérieux, Marcy l’Étoile, France), providing minimum inhibitory concentrations (MICs) for selected antibiotics representing cephalosporines (e.g., cefotaxime), carbapenems (e.g., imipenem, meropenem), quinolones (ciprofloxacin), aminoglycosides (gentamicin), and sulfonamides (trimethoprim-sulfamethoxazole). Further characterization included phenotypic confirmation of (i) carbapenemase-production of KP by use of the modified carbapenem inactivation method (mCIM) and (ii) ESBL-production confirmation according to CLSI guidelines (CLSI Clinical and Laboratory Standards Institute, 2025).

2.5. Molecular determination of carbapenem resistance genes by use of PCR screening

PCR initially was utilized to screen all CP-KP for the presence of common carbapenemase-encoding genes (i.e., bla_OXA-48-related genes, bla_NDM, bla_KPC, bla_VIM and bla_IMP) as previously described (details are provided in Supplementary Table S1) (Gröbner et al., 2009; Pfeifer et al., 2011; Pitout et al., 2005).

2.6. Correction of wet-lab screening data

In accordance with the procedures outlined in DIN EN ISO 8199:2018, enumeration of the target CP-KP was performed using the number of presumptive CP-KEC colonies grown on CHROMagar^™ mSuperCARBA, followed by a correction based on the results obtained by MALDI-TOF species verification and mCIM-confirmation of carbapenemase production for the representative selection of subcultures.

$$\begin{array}{l}\mathrm{cfu}\mathrm{CP}-\mathrm{KP}/100\mathrm{ml}=\left(\begin{array}{l}\text{colony counts of presumptive}\mathrm{CP}-\\ \mathrm{KEC}\mathrm{on}\text{mSuperCARBA}/100\mathrm{ml}\end{array}\right)\times \\ \frac{n_{\text{confirmed}\mathrm{CP}-\mathrm{KP}}}{n_{\text{selected subcultures}\mathrm{per}\text{sample}}}\end{array}$$

For ESBL-EC, cfu values were adjusted based on the proportion of subcultured colonies confirmed as EC.

2.7. DNA extraction and quantification for whole genome sequencing

Genomic DNA (gDNA) of 58 selected CP-KP isolates was extracted using the Monarch® Genomic DNA Purification Kit (New England Biolabs GmbH, Frankfurt am Main, Germany) according to the manufacturer’s instructions for gram-negative bacteria. DNA was quantified using a Qubit 4 Fluorometer and the Qubit^™ 1x dsDNA HS Assay Kit (both invitrogen by ThermoFisher Scientific, Massachusetts, United States), and stored at −20 °C until use.

2.8. Library preparation and whole genome sequencing

Dual-indexed or unique dual-indexed libraries were prepared from 1 ng of genomic DNA using the Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA, USA), following the manufacturer’s protocol with all reagent volumes scaled down to half. The library preparation procedure was fully automated on the Hamilton Microlab STAR liquid handling system (Hamilton company, Germany). Briefly, the gDNA was initially fragmented and tagmented in a single enzymatic step, followed by PCR amplification to enrich adapter-ligated fragments and incorporate index sequences. Library cleanup was performed using the MagSi-NGSPREP Plus magnetic beads (Steinbrenner Laborsysteme GmbH, Germany). The prepared libraries were quantified using the QuantiFluor® dsDNA System Kit (Promega) on an Agilent BioTek Synergy HTX multi-mode microplate reader. Fragment size distribution was assessed using the 5,300 Fragment Analyzer system (Agilent Technologies, Malaysia) with the High Sensitivity NGS Fragment Kit (Agilent, CA, USA).

The final libraries were normalized to 2 nM, pooled equimolarly, diluted to the recommended loading concentration, and sequenced on the Illumina NextSeq 2000 platform (Illumina Inc., Singapore) using v3 reagents with a 2 × 150 bp read configuration, following the manufacturer’s instructions. The obtained coverage was >150X with an average read length of 144 bp. Illumina raw read data sequenced for this study is available on the Sequence Read Archive (SRA) of the National Center for Biotechnology Information (NCBI) under Bioproject ID PRJEB107251.

2.9. Genomic characterization and comparison with publicly available data

Genome assembly and quality control were performed using the Generic Assembly and Reconstruction (GARI) pipeline (v1.0.0; https://github.com/rki-mf1/GARI). GARI employs fastp (v0.23.4; Chen, 2023) for adapter sequence removal and assessment of read quality statistics, and Shovill (v1.1.0; https://github.com/tseemann/shovill) for de novo genome assembly. The generated assemblies were evaluated based on multiple assembly statistics such as average coverage, assembly length, completeness, and contamination (see GARI pipeline results for a detailed overview) (Supplementary Table S2).

Assembled genomes in FASTA format were uploaded to the Pathogenwatch web application (v23.1.5), which integrates several community-driven tools for genomic antimicrobial resistance (AMR) prediction, multilocus sequence typing (MLST), core genome MLST (cgMLST) typing, and phylogenetic assessment.

To contextualize the wastewater isolates with respect to the question whether intestinal carriage of these strains by people connected to the wastewater system might be among the plausible sources, we compared them with publicly available Klebsiella pneumoniae genomes on Pathogenwatch (v23.1.5) as of February 20, 2025. After determining that the wastewater isolates encompassed eight different sequence types (STs), we created multiple collections on the Pathogenwatch web-based platform containing all publicly available genomes corresponding to one of those eight STs (Supplementary Table S3). From these Pathogenwatch collections, we downloaded the hashed cgMLST profiles and merged them into a single dataset comprising 14,590 publicly available Klebsiella pneumoniae genomes belonging to the same eight STs as the wastewater samples.

The combined cgMLST profile dataset was then utilized as input for a modified version of the cgmlst-dists tool (v0.4.0; https://github.com/KHajji/cgmlst-dists/tree/master), which calculates pairwise distances using hashed profiles (via the “-H” parameter), enabling further analysis without strictly requiring numeric allele names.

For 58 wastewater isolates, a neighbor-joining tree was constructed in Pathogenwatch based on pairwise single-nucleotide polymorphism (SNP) distances across a curated set of 1,972 core genome loci (Argimón et al., 2021). The resulting tree was exported in Newick format for visualization in R (v4.3.0; R Core Team, 2023), alongside the genomic AMR prediction output from Pathogenwatch. For data wrangling, the tidyverse package (v2.0.0; Wickham et al., 2019) was utilized. For visualization of the phylogenetic tree and heatmaps we used treeio (v1.26.0; Wang et al., 2020) and tidyheatmaps (v0.2.1; Engler, 2024), while additional plots were generated using ggplot2 (v3.5.1; Wickham, 2016).

2.10. Snippy analyses

To increase the resolution for the phylogenetic assessment of the wastewater isolates, we performed SNP-based analysis on assemblies using snippy (v4.6.0; https://github.com/tseemann/snippy). For each ST represented by at least two isolates, we selected the sample with the best assembly quality (based on metrics including N50, percentage of complete single-copy BUSCOs, percentage of reads mapped back to assembly, and contamination; see Supplementary Table S2) as the genomic reference sequence. All other isolate assemblies belonging to the same ST were then aligned and compared against this reference using snippy, enabling identification of core SNPs within each ST. Pairwise SNP distances were then visualized in R using the tidyheatmaps package.

3. Results

3.1. Description of the sample set

In total, 33 influent wastewater samples (i.e., 11 samples from WWTP 1, 10 from WWTP 2, 6 from WWTP 3, and 6 from WWTP 4) were used to set up the general procedure. The performance of the incubation regimen was evaluated during the pretesting phase (see below). The subsequent workflow, as described in Figure 1, resulted in 325 presumptive CP-KEC. Amongst these, 320 isolates belonged to the KEC group, with the majority being identified as KP (297 of 320; 93%), including 253 (79%) verified CP-KP.

3.2. Pretest results: evaluation of two different temperature regimes for detection of Klebsiella pneumoniae on chromogenic media

Two temperature regimes were initially evaluated with respect to their effect on detection frequencies of the primary target species (CP-KP) as well as fecal and AMR indicators (EC, ESBL-EC) on chromogenic media. In total, six influent wastewater samples obtained over a period of three consecutive weeks were analyzed, with up to 30 subcultures per target and incubation regime identified to species level. Median values obtained for confirmed KP were 15% (37 °C for 18 h) and 65% [36 ± 1 °C for 4 h followed by 21 h at 44 ± 0.5 °C as described by ISO (2001) as cited in Blaak et al. (2021)] (Figure 2). Differences for confirmed KP were highly significant (p < 0.001). EC and ESBL-EC exhibited a non-significant difference following species verification with median values for (i) EC of 95% (37 °C)/100% (44 °C) and (ii) ESBL-EC of 93% (37 °C)/100% (44 °C). More details are provided in Table 1.

Figure 2.

Colony count results of two different temperature regimes. Comparative assessment of two different temperature regimens regarding target species by orders of magnitude obtained from colony counts of wastewater-loaded filters on selective media [CHROMagar^™ mSuperCARBA (KP), CHROMagar^™ Orientation (EC), and CHROMagar^™ ESBL (ESBL-EC)]. Up to 30 subcultures per target and incubation regime were analyzed from six individual wastewater samples. Boxplots indicate the percentage [%] of species-confirmed subcultures of total isolates analyzed. KP, Klebsiella pneumoniae; EC, Escherichia coli; ESBL-EC, extended-spectrum beta-lactamase E. coli.

Table 1.

Comparison of two incubation regimes for the detection of carbapenemase-producing Klebsiella pneumoniae, Escherichia coli, and ESBL-E. coli.

Incubation temp. [°C]	Sample	KEC including KP					EC			ESBL-EC
		Subcultures	KEC (confirmed)	KP (confirmed)	% (KEC)	% (KP)	Subcultures	Species confirmed	%	Subcultures	Species confirmed	%
37	1	30	12	8	40.0	26.7	30	28	93.3	30	28	93.3
	2	30	5	1	16.7	3.3	30	29	96.7	30	24	80.0
	3	15	3	0	20.0	0.0	30	25	83.3	30	29	96.7
	4	30	17	16	56.7	53.3	30	29	96.7	30	25	83.3
	5	30	20	14	66.7	46.7	30	29	96.7	30	30	100
	6	30	4	0	13.3	0.0	30	28	93.3	30	28	93.3
44	1	27	27	24	100	88.9	30	30	100	30	30	100
	2	16	16	5	100	31.3	30	30	100	30	30	100
	3	21	21	10	100	47.6	30	30	100	30	30	100
	4	30	30	25	100	83.3	30	29	96.7	30	30	100
	5	30	29	29	96.7	96.7	30	30	100	30	30	100
	6	8	8	3	100	37.5	30	29	96.7	30	30	100

Results of the comparison of two incubation protocols regarding the isolation of target species in wastewater samples. Target species were suspect carbapenemase-producing K. pneumoniae (CP-KP) on CHROMagar^TM mSuperCARBA, total E. coli (EC) on CHROMagar^TM Orientation, and ESBL-E. coli (ESBL-EC) on CHROMagar^TM ESBL. The trial comprised influent wastewater samples from two WWTPs, respectively, in three consecutive weeks, i.e., six samples in total. For each target, two volumes of wastewater were filtered in triplicates, resulting in six agar plates per target and incubation regime. If possible, five colonies per plate were used for species-confirmation by MALDI-TOF MS, resulting in a maximum of 30 colonies analyzed per incubation regime. Percentages [%] indicate the ratio of species-confirmed isolates of total isolates analyzed (highlighted in bold values).

Abbreviations: temp., temperature; KEC, Klebsiella, Enterobacter, Citrobacter; KP, Klebsiella pneumoniae; EC, Escherichia coli; ESBL, extended-spectrum beta-lactamase.

Pretesting of different incubation protocols revealed significantly higher detection rates for the target organisms after incubation at 44 °C compared to 37 °C, incubation at elevated temperatures inhibits off-target-growth, including gram-negative non-fermenters such as Aeromonas spp. (Bobis Camacho et al., 2024; Hänninen et al., 1995), as well as other species of the KEC group, as shown previously for K. oxytoca and Enterobacter cloacae (Niemi et al., 2003). This allows for more precise target colony enumeration due to less unspecific growth on the respective filters. Thus, the 44 °C-incubation regime is suitable to detect a low abundance target (CP), as described previously regimes [ISO (2001) as cited in Blaak et al. (2021)]. However, more studies are necessary to assess putative loss of temperature-sensitive strains due to elevated incubation temperatures (Thompkins et al., 2008; Tiwari V. et al., 2024).

3.3. Screening results for target carbapenemase-producing Klebsiella pneumoniae, and indicators Escherichia coli and ESBL-Escherichia coli

3.3.1. Carbapenemase-producing Klebsiella pneumoniae concentrations in wastewater samples

As a result, concentrations of presumptive CP-KEC were, with few exceptions, in overall agreement with the corresponding values obtained for verified CP-KP across all samples and WWTPs (Figure 3), particularly with respect to their overall order of magnitude, with colony counts ranging from 10² to 10⁴ cfu/100 mL (Figure 3).

Figure 3.

Temporal variation of presumptive CP-KEC, CP-KP, EC, and ESBL-EC (cfu/100 mL) in wastewater samples. Temporal variation of target CP-KP, presumptive CP-KEC colony counts (left), and indicators (fecal & AMR) EC and ESBL-EC (right) across four WWTPs in 2024. Colony counts are depicted per 100 mL influent wastewater and represent mean values ± standard deviation (error bars) for each sampling point. Dotted lines indicate the mean CP-KP cfu/100 mL across the respective sample set (weeks 31–48/2024 for WWTP 1 and 2; weeks 42–48/2024 for WWTP 3 and 4). WWTP, wastewater treatment plant; CP, carbapenemase-producing; KEC, Klebsiella/ Enterobacter/ Citrobacter group; KP, Klebsiella pneumoniae; EC, Escherichia coli; ESBL-EC, extended-spectrum beta-lactamase-producing E. coli; cfu, colony-forming units.

Mean cfu/100 mL values obtained for CP-KP were 3.07 × 10³ cfu/100 mL (WWTP 1), 3.79 × 10³ cfu/100 mL (WWTP 2), 7.66 × 10³ cfu/100 mL (WWTP 3), and 1.22 × 10³ cfu/100 mL (WWTP 4). Spatio-temporal variation of CP-KP cfu counts was observed for the WWTPs investigated in this study. Exemplarily, CP-KP colony counts for week 45 exhibited a two-log₁₀ difference, ranging from 5 × 10² cfu/100 mL (minimum: WWTP 4) to 5 × 10⁴ cfu/100 mL (maximum: WWTP 3) (Figure 3). The minimum CP-KP counts were obtained for samples collected in week 41 (1.44 × 10² cfu/100 mL, WWTP 2) and the maximum was recorded in week 45 (3.44 × 10⁴ cfu/100 mL, WWTP 3). In WWTP 3, CP-KP counts were comparatively high in week 45, with cfu’s within a similar order of magnitude as ESBL-EC (10⁴ cfu/100 mL).

3.3.2. Escherichia coli and ESBL-Escherichia coli concentrations in wastewater samples

Colony counts of EC as fecal indicators were temporally stable across all sampling weeks regarding their orders of magnitude. Colony counts ranged between 10⁶ and 10⁷ cfu/100 mL with mean values of 1.7 × 10⁷ cfu/100 mL (WWTP 1), 1.5 × 10⁷ cfu/100 mL (WWTP 2), 1.5 × 10⁷ cfu/100 mL (WWTP 3), and 6.6 × 10⁶ cfu/100 mL (WWTP 4). The minimum EC colony counts were obtained for samples collected in week 48 (2.47 × 10⁶ cfu/100 mL, WWTP 4), peaking counts were recorded in week 45 (3.73 × 10⁷ cfu/100 mL, WWTP 1) (Figure 3).

As an indicator of viable AMR bacteria, ESBL-EC illustrated roughly 100-fold lower colony counts compared to those of EC (Figure 3), with few deviations. For ESBL-EC, mean cfu/100 mL counts were 5.2 × 10⁴, 7.8 × 10⁴, 4.2 × 10⁴, 3.6 × 10⁴ for WWTP 1–4, respectively. The lowest colony counts were obtained for samples collected in week 47 (1.67 × 10³ cfu/100 mL, WWTP 1), and the highest counts were recorded in week 34 (1.28 × 10⁵ cfu/100 mL, WWTP 2) (Figure 3).

3.4. Carbapenemase-producing Klebsiella pneumoniae: antibiotic susceptibility profiles and PCR results

Overall, each of the wastewater samples investigated was positive for CP-KP (Table 2). AST results (MICs) were assessed for all subcultured CP-KP isolates as presented in Figure 4.

Table 2.

CP-KP screening results obtained for 33 wastewater samples from four WWTPs.

WWTP	Week (2024)	Sub-cultures (pheno-type)	Confirmed KEC	Confirmed KP	CP-KP	WGS	ESBL positive	PCR screening results
								blaOXA-48-related genes	bla VIM	bla NDM	bla KPC	blaOXA-48-related genes + blaVIM	blaOXA-48-related genes+ blaNDM	blaOXA-48-related genes + blaKPC	blaNDM+ blaKPC
1	31	10	10	9	5	3	0	4	0	0	0	0	0	0	0
	34	10	10	8	6	6	0	2	0	0	3	0	0	0	0
	36	10	10	9	8	7	3	5	0	0	3	0	0	0	0
	38	10	10	7	2	2	0	0	0	0	2	0	0	0	0
	41	10	10	10	10	4	4	3	0	0	7	0	0	0	0
	42	10	10	10	9	4	1	1	0	0	6	0	0	0	0
	43	10	10	10	10	5	0	0	0	0	3	0	0	0	2
	45	10	8	8	8	0	3	7	0	0	1	0	0	0	0
	46	10	10	6	6	0	0	4	0	0	2	0	0	0	0
	47	10	10	10	10	0	0	1	0	0	8	0	0	0	0
	48	10	10	10	10	0	4	1	0	0	9	0	0	0	0
2	31	10	10	9	7	7	3	3	0	0	0	0	4	0	0
	34	10	10	9	5	5	1	1	0	4	0	0	0	0	0
	36	10	9	9	6	6	6	3	0	3	0	0	0	0	0
	38	10	10	9	2	2	0	0	0	2	0	0	0	0	0
	41	9*	9	9	8	3	7	2	0	6	0	0	1	0	0
	42	10	10	10	10	0	0	0	0	9	0	0	0	0	0
	43	10	10	10	10	4	2	2	0	8	0	0	0	0	0
	45	9	9	9	9	0	0	0	0	9	0	0	0	0	0
	47	5*	5	5	5	0	0	1	0	4	0	0	0	0	0
	48	10	10	10	6	0	3	0	0	5	0	0	1	0	0
3	42	11	11	9	8		3	1	0	1	2	1	1	1	0
	43	10	10	8	8		2	2	0	1	0	0	2	0	0
	45	10	10	10	10		9	10	0	0	0	0	0	0	0
	46	10	10	10	10		8	9	1	0	0	0	0	0	0
	47	10	9	9	7		1	3	0	3	1	0	0	0	0
	48	10	10	10	5		3	5	0	0	0	0	0	0	0
4	42	10	10	10	10		6	2	0	4	1	0	1	0	0
	43	10	10	10	10		1	2	0	0	0	0	8	0	0
	45	11	11	8	6		2	2	0	0	0	0	2	0	0
	46	10	10	10	10		1	1	0	0	1	0	8	0	0
	47	10	10	10	10		0	2	0	1	0	0	7	0	0
	48	10	9	7	7		0	5	0	0	0	0	2	0	0

Information on samples, sampling period, subcultures, presumptive and confirmed CP-KP and PCR-screening results for four common carbapenemase-encoding genes. Grey areas: no data available. Asterisks (*) indicate samples with a maximum number of presumptive CP-KEC colonies below ten. WWTP, wastewater treatment plant; WGS, whole genome sequencing; KEC, Klebsiella-Enterobacter-Citrobacter group; KP, Klebsiella pneumoniae; CP-KP, carbapenemase-producing Klebsiella pneumoniae (species- and carbapenemase production verified); ESBL, extended-spectrum beta-lactamase.

Figure 4.

Distribution of MIC values for 253 CP-KP isolates across selected antimicrobial substances. The green-shaded areas of the graph indicate that, according to CLSI guidelines (CLSI Clinical and Laboratory Standards Institute, 2025), isolates would be classified as susceptible if isolated from human clinical cases. In contrast, the red-shaded areas denote the range at which clinical isolates would be classified as resistant for at least one specific indication (e.g., wound infection, pneumonia, blood stream infection).

Phenotypic AST was performed using antimicrobial substances from different antimicrobial families, including cephalosporines (cefotaxime, ceftazidime, cefuroxime), carbapenems (ertapenem, imipenem, meropenem), aminoglycosides (gentamicin), quinolones (ciprofloxacin), and folate pathway inhibitors (sulfamethoxazole/trimethoprim). Details on the distribution of MICs obtained for 253 CP-KP detected are given in Figure 4. AST profiles for each isolate are summarized in Supplementary Table S4.

Genotypic characterization of CP-KP by PCR revealed the predominance of bla_OXA-48-related genes (n = 83) among isolates from all WWTPs, followed by bla_NDM-1 (n = 60) in WWTPs 2, 3 and 4, bla_KPC (n = 49) in WWTP 1, 3 and 4, and bla_VIM (n = 1) in WWTP 3, while bla_IMP was not detected. Co-occurrence of carbapenemase-encoding genes was observed in 41 isolates with bla_OXA-48-related genes /bla_NDM-1 (n = 37; WWTPs 2, 3 and 4) being the most common combination, followed by bla_NDM-1 /bla_KPC (n = 2; WWTP 1), bla_OXA-48-related genes /bla_VIM (n = 1; WWTP 3) and bla_OXA-48-related genes /bla_KPC (n = 1; WWTP 3). Individual data for each isolate is provided in Supplementary Table S4. Due to the limited panel of PCR-targeted genes, 19 isolates lacked a positive PCR result for a specific carbapenemase-encoding gene (Table 2). Further characterization using WGS revealed carbapenemase-encoding genes (i.e., bla_KPC-2, bla_NDM-1, bla_NDM-5, bla_OXA-48, bla_OXA-181, bla_OXA-427) in each of the CP-KP isolates sequenced (Supplementary Table S4).

Fourteen CP-KP isolates recovered from samples of three WWTPs exhibited ertapenem, imipenem, and meropenem MICs below the CLSI-recommended thresholds for further testing for carbapenemase production (Figure 4; Supplementary Table S4). While PCR analysis revealed the presence of bla_OXA-48-related genes in 13 of these, the evaluation of whole genome data further revealed that the isolate lacking a specific PCR signal harbored the bla_OXA-48 gene bla_OXA-181 (Supplementary Table S4).

3.5. Genomic profiling of carbapenemase-producing Klebsiella pneumoniae wastewater isolates and comparison to public genomic data

To gain further insights into the genetic makeup of CP-KP isolated from wastewater samples, 58 isolates from WWTP 1 and 2 were selected for WGS, following a stepwise approach. The initial sequencing batch covering isolates from samples representing week 31, 34, 36, and 38 accounted for the majority of CP-KP isolates (n = 41). These isolates were selected to verify the sensitivity of the wet-lab screening process regarding CP-KP harboring carbapenemase-encoding genes (positive rate: 100%), as the PCR results (see Supplementary Table S4) were inconclusive due to limitations of the screening panel. Isolates for the second sequencing batch following week 41 were selected to reveal the phylogenetic background of CP-KP exhibiting divergent AST results obtained from a specific wastewater sample, while most presumptive isogenic isolates (i.e., isolates sharing a similar phenotype and AST profile) were mostly excluded to save resources. An overview of isolate IDs, STs, and ARGs detected from WGS per week across the 12-week sampling period is provided in Supplementary Table S4.

CP-KP belonging to different phylogenetic backgrounds were detected in most samples, including four different STs in WWTP 1 (weeks 36 and 42). Closely related genomes of isolates belonging to ST35 carrying bla_OXA-181 were detected in samples of both WWTPs (Figure 5). Further isolates obtained from samples of WWTP 1 were assigned as follows: ST37 (n = 10, bla_KPC-2), ST258 (n = 5, bla_KPC-2), ST273 (n = 1, bla_KCP-2), and ST485 (n = 10, bla_OXA-181). Isolates assigned to ST15 (n = 2, both isolates: bla_OXA-48 and bla_OXA-427, one isolate in combination with bla_NDM-1), and ST307 (n = 4, bla_NDM-1) were obtained from WWTP 2 samples. In addition, WGS analysis of isolates belonging to ST147 (n = 14) highlighted two genomic clusters. In the first cluster, four out of six isolates were found to carry bla_OXA-48 and bla_NDM-5, while 8 isolates from the second cluster were associated with bla_NDM-1 (Figure 5). WGS analyses revealed between one and three (isolate ID 8205) carbapenemase-encoding genes for CP-KP (Figure 5; Supplementary Table S4).

Figure 5.

WGS AMR profiling of 58 CP-KP isolates from two WWTPs. (A) A phylogenetic tree of 58 wastewater isolates from WWTP 1 and 2. (B) Sequence types (STs). (C) The presence of ARGs in the isolate assemblies, color-coded by the antibiotic class to which they confer resistance. (D) The number of publicly available samples in Pathogenwatch belonging to the same ST as the focal sample. (E) The minimal cgMLST allelic distance to the closest publicly available Pathogenwatch genomes. (F) The number of publicly available genomes from Pathogenwatch with a core genome MLST (cgMLST) allelic distance (AD) of fewer than 15 alleles from the corresponding wastewater isolate.

Closely related Pathogenwatch genomes, with an allelic distance (AD) of less than 15 to the genomes of wastewater isolates, were predominantly associated with STs that are more frequently observed in the Pathogenwatch dataset. Even though only one out of four isolates belonging to ST307 had two Pathogenwatch neighbors with an AD < 15, all WGS of isolates illustrated low minimal ADs (14–17) to their closest publicly available sample in the Pathogenwatch dataset. Genomes of ST147 wastewater isolates associated with bla_NDM-1 showed up to six closely related Pathogenwatch samples with AD < 15 and minimal ADs as low as seven alleles, while the ST147 isolates associated with bla_OXA-48 and bla_NDM-5 showed greater distance to publicly available samples, with a minimal AD of 23. For ST258, the minimal AD to Pathogenwatch genome entries ranged between 4 and 8. Such close matches were also reflected in the number of Pathogenwatch genome entries with AD < 15, which ranged from 29 to 125. The threshold of an AD < 15 did not show any hits with Pathogenwatch genomes for the other STs, for which the minimal AD to database genomes were instead as follows: 48–57 (ST15), 19–25 (ST35), 51–53 (ST37), 35 (ST273), and 16–20 (ST486) (Figure 5E).

It is noteworthy that most Pathogenwatch genomes closely related to the wastewater isolates from this study were from Germany (Supplementary Figure S1). Moreover, most hits are from fecal isolates, followed by clinical and human-associated samples (Supplementary Figure S2).

3.6. Tracing temporal sequence type occurrence and SNP dynamics across the sample set

A detailed comparative whole genome SNP analysis was performed in order to investigate genomic distances between isolates assigned to a particular ST and to facilitate further tracing of putative temporal dynamics across 13 weeks of sampling (Figure 5). A further analysis, including isolates sharing the ST35 background present in samples of WWTP 1 and 2, is available in Supplementary Figures S3, S4.

Closely related ST485 isolates were detected in WWTP 1, with a genomic distance range between one and 14 SNPs from week 31 to 42. Samples belonging to ST37 showed a minimum genomic distance of 7 SNPs between isolates from weeks 31 and 43, indicating limited genomic diversity within this sequence type across the sampling period. The range of differences between isolates obtained on a specific sampling day is two (week 38) to 21 (week 36). Genomes assigned to ST35 were detected at multiple time points in both WWTPs (Figures 6, 7). SNP distances revealed distinct clusters among WWTPs differing from each other by at least 14 SNPs (see Supplementary Figure S4). A group of closely related ST35 isolates obtained from WWTP 1 samples showed a genomic distance range of 1–24 SNPs. One sample (8220, WWTP 1, week 41) exhibited a minimum difference of 19 SNPs to other genomes from WWTP 1, and a minimum difference of 32 SNPs to genomes from WWTP 2 (see Supplementary Figure S4). The range of SNP differences between isolates assigned to ST258 is 2 to 106 SNPs, with isolate 8290 (week 42) being an outlier (see Supplementary Figure S4).

Figure 6.

Occurrence, temporal variation, and range of SNP distances of CP-KP WGS per ST detected in WWTP 1. Color-coded STs are assigned to the wastewater sample (week of isolation) indicated on the x-axis, along with the number of isolates subjected to WGS per sampling week. Four-digit strain IDs represent the individual isolates (Figure 5; Supplementary Figure S4; Supplementary Table S4). Single-nucleotide polymorphism (SNP) distances between subsets of isolates and/or sampling weeks are indicated with bold numbers (see Supplementary Figure S4 for full overview of pairwise SNP distances between isolates). WWTP, wastewater treatment plant; ST, sequence type; WGS, whole genome sequencing; SNP, single nucleotide polymorphism.

Figure 7.

Occurrence, temporal variation, and range of SNP distances of CP-KP WGS per ST detected in WWTP 2. Color-coded STs are assigned to the wastewater sample (week of isolation) indicated on the x-axis, along with the number of isolates subjected to WGS per sampling week. Four-digit strain IDs represent the individual isolates (Figure 5; Supplementary Figure S4; Supplementary Table S4). Single-nucleotide polymorphism (SNP) distances between subsets of isolates and/or sampling weeks are indicated with bold numbers (see Supplementary Figure S4 for a full overview of pairwise SNP distances between isolates). WWTP, wastewater treatment plant; ST, sequence type; WGS, whole genome sequencing; SNP, single nucleotide polymorphism.

Further investigation of WWTP 2 isolates (Figure 7) revealed that the two ST15 isolates (8198 and 8205 from week 41) differed by 30 SNPs. Genomes assigned to ST307 sampled in week 36 and 43, have 10 to 12 SNPs differences between three isolates from week 36, and a minimum distance of 23 SNPs to the sample 8381 from week 43. Genomes assigned to ST147 showed temporal differences regarding the detection of genomes assigned to two distinct clusters (Supplementary Figure S4), including different carbapenemase-encoding gene profiles (Figure 5). ST35 genomes obtained from isolates of WWTP 2 seem closely related, with a SNP differences between 0 to 7, regardless of the sampling week.

4. Discussion

Here we present a straightforward filtration-based “wet-lab” approach accompanied by WGS analysis to quantify and characterize a selected AMR target (CP-KP) in wastewater samples as a proof-of-concept study. Two indicators for wastewater (EC) and AMR (ESBL-EC) viability were used as quality parameters for quantification.

4.1. Quantification of AMR screening target

The workflow was designed based on a lab regime described previously [ISO (2001) as cited in Blaak et al. (2021)] and has been specifically optimized for the detection and quantification of CP-KP. The resulting total verification rate for KP was 93% (297 of 320 identified KEC) including 253 verified CP-KP, which equates to a positive predictive value of 85.1% for CP-KP across all WWTP samples, when using the outlined procedure.

Care should be taken before comparing different approaches aiming at AMR assessment in wastewater samples due to divergent study designs, target choices and abundances. Here, the workflow was specifically amended to target CP-KP. However, CP-Enterobacterales (CP-E) and other carbapenemase-producing gramnegative bacteria isolated from wastewater samples have been quantified before. Thus, findings from previous studies may be useful for broadly classifying procedure performance. In this study, each of the 33 wastewater samples from four distinct WWTPs tested positive for CP-KP. Previous studies reported CP-KP detection at least once in eight out of ten WWTPs in Finland using the same selective agar (25/89 samples tested positive in total = 28.1%) (Heljanko et al., 2024). In comparison, CP-E were isolated from 89 of 100 samples representing 100 urban WWTPs in the Netherlands (Blaak et al., 2021).

The mean colony forming units for CP-KP reported in this study (3.07 × 10³ cfu/100 mL for WWTP 1, 3.79 × 10³ cfu/100 mL for WWTP 2, 7.66 × 10³ cfu/100 mL for WWTP 3 and 1.22 × 10³ cfu/100 mL for WWTP 4) appear slightly higher than values reported in other studies, which often assessed a broader group of species (i.e., CP-E). A previous study across 100 Dutch WWTPs, for instance, reported CP-E values ranging between 0 and 2 × 10³ cfu/100 mL, with a mean value of 7.9 × 10¹ cfu/100 mL (Blaak et al., 2021). Another study conducted in northern Spain investigating influent samples from two WWTPs, reported mean values ranging between 1 × 10³ and 2 × 10⁵ cfu/100 mL CP-E (Pino-Hurtado et al., 2024). Presumptive CP-E values obtained for raw wastewater samples of a Japanese WWTP were between 10³–10⁴ cfu/100 mL (Urase et al., 2022). For Germany, reported figures range between 10² and 10⁵ cfu/100 mL CP-E for influent wastewater of treatment plants not influenced or influenced by hospital wastewater (n = 3/n = 3), respectively (LANUV, 2024). The authors state that the performance of CHROMagar^™ ESBL for the quantification and isolation of CP-E is inferior to that of CHROMagar^™ mSuperCARBA (LANUV, 2024), which was also employed in this study. The low standard deviation values across all WWTPs indicate the reliability and robustness of the target screening procedure established for presumptive CP-KEC and confirmed CP-KP (Figure 3). Moreover, a recent German study suggested that CHROMagar^™ mSuperCARBA improves the detection of CP-KEC harboring OXA-48 variants (i.e., OXA-244 and OXA-181) in wastewater samples (LANUV, 2024).

Based on the results of our pilot study, the establishment of a structured approach for subsequent method validation according to DIN EN ISO 13843 (DIN, 2018) is required as a next step in advancing the method. This validation includes the categorical performance characteristics (sensitivity, specificity, efficiency, selectivity, false positive rate and false negative rate), determination of the upper limit and consideration of the lower limit of detection, assessment of precision (repeatability and reproducibility), robustness, relative recovery and uncertainty of counting.

4.2. Reliability indicators for wastewater (Escherichia coli) and AMR (ESBL-Escherichia coli)

Regarding overall ranges of cfu/100 mL concentration levels, EC and ESBL-EC counts showed no explicit temporal variation while the observed values were generally consistent with those reported in studies from various European countries (Blaak et al., 2021; Blaak et al., 2015; Conforti et al., 2024; Pallares-Vega et al., 2021; Tiwari et al., 2022b; van Heijnsbergen et al., 2022), North America (Haberecht et al., 2019), and Africa (Banu et al., 2021). This consistency indicates the suitability of these indicators as proxy for fecal load and AMR viability in wastewater samples, respectively. The WHO Tricycle Protocol has previously stated that ESBL-EC is a relevant and representative proxy for AMR (WHO, 2021). Including independent quality control indicators is therefore essential to ensure sample validity and comparable quantification of target organisms.

A recent study compared culture-based detection with real-time (RT)-PCR quantification of pathogens in wastewater samples and demonstrated that RT-PCR based results depend strongly on primer specificity and the presence of conserved target regions. In that study, two of six target species were not detected by RT-PCR (Mabeo et al., 2026). The authors attributed this limitation to genetic variability in environmental strains, including mutations in primer-binding regions or the absence of specific housekeeping or virulence genes, which can result in non-amplification despite the presence of viable cells. The study further demonstrated that integrating culture-based and molecular methods improved detection sensitivity and provided complementary insights (Mabeo et al., 2026).

4.3. Characterization of a selected AMR target: carbapenemase-producing Klebsiella pneumoniae

Following quantification and confirmation of presumptive CP-KP, verified isolates were further phenotypically and genotypically characterized, revealing that a variety of different AST profiles and carbapenemase-encoding genes are associated with the CP-KP wastewater collection.

Since clinical breakpoints suitable for classification of isolates of human or animal origin (as either resistant, intermediate or susceptible) are not applicable to isolates from environmental samples (Smith, 2020), raw MIC values for selected antibiotics representing cephalosporines (e.g., cefotaxime), carbapenems (e.g., imipenem, meropenem), quinolones (ciprofloxacin), aminoglycosides (gentamicin), and sulfonamides (trimethoprim-sulfamethoxazole) are presented in order to facilitate comparability with other studies, especially with respect to One Health integrated surveillance efforts (JPIAMR, 2022).

Interestingly, 14 CP-KP yielded MICs for different carbapenems (Figure 4) below the recommended thresholds for further testing for carbapenemase production (CLSI, 2025). Further characterization of these CP-KP revealed the presence of bla_OXA-48 and/or bla_OXA-48-like genes (e.g., bla_OXA-181). This finding agrees with previous observations for clinical isolates [reviewed in Boyd et al. (2022) and Peirano and Pitout (2025)], and wastewater isolates (Zurfluh et al., 2017). However, the screening procedure described here detected these phenotypes, since all confirmed CP were subjected to mCIM testing. By applying the screening procedures described in the present study (which involved mCIM testing for all confirmed CP), we were able to detect phenotypes which might have been missed otherwise. Thus, our approach allows for the detection of a broad range of different genotype/phenotype combinations among CP-KP populations, which is also reflected by the within-sample heterogeneity of the wastewater CP-KP (Figures 6, 7).

Besides carbapenems, most CP-KP isolates yielded MICs above clinical breakpoints for cefuroxime (second generation cephalosporines), cefotaxime and ceftazidime (third generation cephalosporines), sulfonamide and antifolate classes (sulfamethoxazole and trimethoprim combination) and fluoroquinolones (ciprofloxacin), indicating the presence of a broad range of resistance-encoding genes.

4.3.1. Carbapenemase-producing Klebsiella pneumoniae representing two different WWTPs include sequence types of global concern

Unless the entry pathway is exclusive, it is generally not appropriate to draw conclusions about a single source from communal wastewater samples. Unlike samples subjected to microbiological diagnostics from well-defined origins, such as human, food, animal, or process-hygiene sources, this complex matrix comprises multiple concurrent entries. The phenotypical and genotypical diversity of a specific bacterial species obtained from wastewater screening procedures is difficult to predict, as it can vary between sewage treatment plants, sewer systems, the population connected to the system, industrial discharges and sampling days.

Comparative analysis of samples across the wastewater network based on 16S rRNA gene metabarcoding revealed that bacterial ß-diversity was more similar between WWTP influent samples and samples from a community wastewater collection point than between WWTP influent samples and hospital wastewater samples. This finding suggests the potential to detect relevant, and potentially novel pathogen-resistance phenotypes within the population connected to the system (Verburg et al., 2021).

A total of 8 different STs, i.e., ST147 and ST273 (both: clonal group 147), ST258, ST35, ST15, ST37, ST307, ST485 were identified in this study. Due to the limited number of isolates subjected to WGS and the lack of long-term data, discussion on distinct ST frequency occurrences is out of scope. Here, WGS was primarily used as a complementary method to confirm the reliability of the wet lab protocol targeting CP-KP and to capture overall CP-KP diversity within individual samples across the sampling period. The isolation of bacterial species from wastewater samples, including their cultivation, sequencing, subsequent genomic reconstruction, and annotation can contribute to a better understanding of bacterial phenotype characteristics by providing molecular insights, including—but not limited to—the presence of resistance genes. On the other hand, metagenomic approaches may provide insights into the general population in wastewater and the broader resistome of the entire bacterial community within the wastewater system, in which sequencing depth, among other factors, is a particularly decisive parameter for detecting rare targets (Abdulkadir et al., 2024; Kim et al., 2024). More research on the subject, especially comparison with metagenomic data, will provide insights about the ability of the proposed approach to capture phylogenetic diversity of CP-KP. Multiple STs (up to four) were identified across the majority of the wastewater samples, indicating that the presented approach allows for simultaneous detection of genetically diverse CP-KP (Figures 5–7).

Besides being identified in wastewater samples of this study, CP-KP belonging to ST15 (Spadar et al., 2022), ST147 (and ST273) (Rodrigues et al., 2022; Spadar et al., 2022), ST35 (Haak et al., 2025), ST307 (Heiden et al., 2020), ST258 (Becker et al., 2018) seem of general clinical importance, whereas isolates belonging to ST485 are less frequently reported (Cao et al., 2021; Kang et al., 2022; Liao et al., 2021). Except for ST35, all detected CP-KP STs have been previously described in studies dealing with influent wastewater samples: ST15 (Heljanko et al., 2024; Spadar et al., 2022), ST37 (Heljanko et al., 2024), ST147 (Nasser-Ali et al., 2024; Teban-Man et al., 2021), ST258 (Surleac et al., 2020; Teban-Man et al., 2021), ST273 (Ilabaca-Carrasco et al., 2025), ST307 (Ekwanzala et al., 2019; Heljanko et al., 2024), ST485 (Surleac et al., 2020). KP belonging to ST35 were previously detected in wastewater, but the respective isolates were negative for carbapenemase production (Di Cesare et al., 2024; Surleac et al., 2020).

Some of the CP-KP STs (e.g., ST258, ST307) have been recognized as high-risk clones in human hospitals that are associated with a broad range of ARGs (ECDC, 2019; Heiden et al., 2020; Weingarten et al., 2018). CP-KP ST258, for instance, is of public health concern because of its widespread distribution, its frequent multidrug resistance, and its ability to rapidly exchange plasmid borne resistance determinants with other Enterobacteriaceae (Conlan et al., 2014) and reviewed in Navon-Venezia et al. (2017). It has been discussed that adaptive mutations likely contribute to ST258 persistence among hospitalized patients (Marsh et al., 2019), and mutations have been detected among a single patient colonized by CP-KP over a 4.5-year period (Jousset et al., 2018). Thus, long-term enteral colonization of humans, either in terms of hospitalized patients or even among the general population, needs to be considered as a possible source for CP-KP in wastewater samples. Furthermore, environmental sources need to be addressed as well. Closely related CP-KP genomes belonging to ST147 have been isolated from different sampling sites influenced by hospital wastewater. These sites include a path from patient rooms (toilets and showers) to the WWTP effluent. In particular, the limited genomic diversity together with the stable presence of the bla_NDM-1 (often in association with bla_OXA-48) gene among all isolates investigated was discussed as a sign for long-term persistence of this particular clone in the respective sewage system (Kehl et al., 2022). The ST147 genomes from isolates of WWTP 2 in the present study carry either bla_OXA-48 with or without bla_NDM-5 genes, or bla_NDM-1. Genome distances of >500 SNPs (Figure 7; Supplementary Figure S4), indicate the presence of at least two different clusters, which were detected simultaneously in week 34.

Of note, one isolate (8205) belonging to ST15 revealed the presence of three distinct carbapenemase-encoding genes (bla_OXA-48, bla_OXA-427, bla_NDM-1). Such accumulation of resistance genes appears to be rare, but not unprecedented, as reported for a clinical ST147 isolate (Mataseje et al., 2022).

4.3.2. Phylogenetic relationship between WGS

So far, only limited genomic diversity has been reported for CP-KP isolates from wastewater samples (LANUV, 2024), hospitalized patients (Marsh et al., 2019), and hospital environments (Haak et al., 2025), even from sampling campaigns spanning several years. These findings raise questions regarding the expected mutation rates of CP-KP in its natural habitat as part of the gut-colonizing microbiota. Additional uncertainty exists regarding mutation rates in more challenging environments, such as in patients receiving treatment (antibiotics) or in the wastewater system (Kehl et al., 2022).

Of note, the majority of the Pathogenwatch samples closely related to wastewater isolates from this study are submissions from Germany. These samples are mostly of fecal origin, followed by clinical and other human-associated sources (Supplementary Figures S1, S2), indicating the influence of the population connected to the wastewater system on the bacterial composition in influent wastewater (Knight et al., 2025; Newton et al., 2015).

When analyzing the phylogenetic relationship or clonality of KP using SNPs, the expected mutation rate for that particular species is an important parameter. Current literature provides, due to different methods, i.e., core genome (cg), maximum common genome (mcg), and study goals, a variety of mutation rates for KP, including CP-KP and different STs (Yang et al., 2017). Recently, genome-wide nucleotide substitutions of multidrug-resistant KP sharing the clonal group 147 (i.e., ST147, ST273 and ST392) revealed an evolutionary rate range from ~1×10⁻⁶ to 2×10⁻⁶ substitutions/site/year (≈ 3–8 SNPs/genome/year) (Rodrigues et al., 2022). The genomes presented here span a sampling period of 13 weeks, thus a range of 0.75 to 2 SNPs per genome seems a reliable estimation for clonally related isolates. A further study on pairwise SNP distances between CP-KP ST35 outbreak strains in Germany showed very limited diversity within clonally related clusters (i.e., 0 to 6 SNPs) over a period of 30 weeks (Haak et al., 2025), suggesting careful interpretation of SNP distances between genomes from wastewater-associated isolates.

Taken together, further research and development efforts are required to increase our ability to trace back the source of CP-KP in wastewater samples and to address significant knowledge gaps regarding survival and residence of clinical important CP-KP in wastewater systems, including biofilms. However, based on the Pathogenwatch matches, feces from enteral colonized people appear to be a plausible source of the isolates reported in this study. The inclusion of the wastewater pathway may provide early insight into the emergence of novel pathogen-resistance combinations and thus could further strengthen integrated genomic surveillance efforts and public health response.

4.4. Study limitations

This study was designed and implemented as a proof-of-concept approach, confirming its ability to detect different CP-KP in wastewater, and the procedure was rigorously tested within the scope of this study, but has several important limitations. Those include the restricted geographic coverage and number of WWTPs included, and the possible loss of clinically relevant CP-KP isolates due to the chosen incubation regimen and selective media. Moreover, there is a putative process bias associated with the use of filter membranes used for bacterial concentration due to different adsorption characteristics of bacterial cells to the filter material, or shielding effects of the filter leading to decreased antibiotic concentrations or access to nutrients (DIN, 2023). A validation process according to international standards is required. These limitations need to be addressed in subsequent studies to further improve targeted CP-KP detection in wastewater samples.

4.5. Outlook

In the absence of ‘gold standards’, the assessment of method performance indicators, including sensitivity and specificity, for AMR quantification in wastewater samples requires a comparative approach that bridges disciplines to ensure reliability of results. While culture-based WGS allows for detailed characterization of individual isolates, including precise identification of ARGs and a deep understanding of phylogeny, it might introduce selection biases (Jarocki et al., 2024). Such selection bias can be addressed by incorporating culture-independent metagenomic approaches alongside culture-based methods. While sequencing depth and related parameters are critical for detecting rare genetic determinants, metagenomic data from the same samples enable direct comparison between culture-based and culture-independent results and provide complementary insights into the diversity and abundance of ARGs and ARBs in the wastewater system (Rolbiecki et al., 2024). Future studies should continue to investigate the complementary nature of isolate-based sequencing and metagenomic analyses to assess the broader resistome in wastewater.

Here, we provide a robust and conclusive workflow for the detection of CP-KP, which is currently considered a suitable target for WBS due to its relevance to human health according to the WHO bacterial priority pathogens list (Sati et al., 2025; WHO, 2024b). Taken together, this proof-of-concept approach demonstrates the feasibility of detecting CP-KP in wastewater. In the context of regulatory requirements, the availability of validated and robust methods is essential to ensure comparability and reliability within a regulatory framework. Therefore, a validation process that adheres to international standards is imperative.

5. Conclusion

This study presents a wet-lab-based workflow for the monitoring of CP-KP in wastewater, alongside EC and ESBL-EC as fecal and AMR indicators, respectively. CP-KP were detected in all samples analyzed with concentrations between 10² and 10⁴ cfu/100 mL. EC and ESBL-EC range from 10⁶ to 10⁷ and 10² to 10⁴, respectively, which is in line with findings from other studies analyzing influent wastewater. The majority of isolates from a chromogenic, selective agar was confirmed as KP (93%), and 79% were further confirmed for carbapenemase production. Genomic analyses identified bla_OXA-48-like genes as the predominating gene within the isolates. Furthermore, WGS of a subset of isolates reveales eight different sequence types: ST15, ST35, ST37, ST147, ST258, ST273, ST307, and ST485. Most of these STs are of clinical relevance, and sequences express high similarity with fecal isolates from Germany, which suggests that the isolates originate from the people connected to the WWTPs. Acknowledging that more research on the topic, as well as a validation process are needed, these findings indicate that the method described is reliable in detecting CP-KP strains with different genetic backgrounds in influent wastewater.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. AMELAG (“Wastewater monitoring for epidemiological situation assessment”) is financed by the German Federal Ministry of Health (duration: November 2022 to December 2025).

Data availability statement

The Illumina sequencing data generated in this study have been deposited in the European Nucleotide Archive (ENA) under the BioProject accession number: PRJEB107251.

Author contributions

KW: Conceptualization, Methodology, Investigation, Formal analysis, Data curation, Visualization, Writing – original draft, Writing – review & editing. VB: Methodology, Investigation, Formal analysis, Data curation, Visualization, Writing – review & editing. CB: Methodology, Investigation, Formal analysis, Data curation, Writing – review & editing. DB: Investigation, Writing – review & editing. DD: Project administration, Formal analysis, Supervision, Writing – review & editing. SS: Investigation, Data curation, Writing – review & editing. JB: Investigation, Writing – review & editing. AL: Investigation, Writing – review & editing. CF: Formal analysis, Writing – review & editing. SF: Formal Analysis, Writing – review & editing. SW: Formal analysis, Writing – review & editing. AB: Conceptualization, Investigation, Formal analysis, Writing – review & editing. MH: Conceptualization, Methodology, Project administration, Formal analysis, Supervision, Writing – review & editing. BW: Conceptualization, Methodology, Project administration, Formal analysis, Supervision, Visualization, Writing – original draft, Writing – review & editing.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb.2026.1821458/full#supplementary-material