Colistin-Resistant Enterobacteriaceae Isolated From Process Waters and Wastewater From German Poultry and Pig Slaughterhouses

Due to the high prevalence of colistin-resistant Enterobacteriaceae in poultry and pigs, process waters and wastewater from slaughterhouses were considered as a hotspot for isolates carrying plasmid-encoded, mobilizable colistin resistances (mcr genes). Thus, questions on the effectiveness of wastewater treatment in in-house and municipal wastewater treatment plants (WWTPs) as well as on the diversity of the prevailing isolates, plasmid types, and their transmissibility arise. Process waters and wastewater accruing in the delivery and unclean areas of two poultry and two pig slaughterhouses were screened for the presence of target colistin-resistant bacteria (i.e., Escherichia coli, Klebsiella spp., Enterobacter cloacae complex). In-house and municipal WWTPs (mWWTPs) including receiving waterbodies were investigated as well. Samples taken in the poultry slaughterhouses yielded the highest occurrence of target colistin-resistant Enterobacteriaceae (40.2%, 33/82), followed by mWWTPs (25.0%, 9/36) and pig slaughterhouses (14.9%, 10/67). Recovered isolates exhibited various resistance patterns. The resistance rates using epidemiological cut-off values were higher in comparison to those obtained with clinical breakpoints. Noteworthy, MCR-1-producing Klebsiella pneumoniae and E. coli were detected in scalding waters and preflooders of mWWTPs. A total of 70.8% (46/65) of E. coli and 20.6% (7/34) of K. pneumoniae isolates carried mcr-1 on a variety of transferable plasmids with incompatibility groups IncI1, IncHI2, IncX4, IncF, and IncI2 ranging between 30 and 360 kb. The analyzed isolates carrying mcr-1 on transferable plasmids (n = 53) exhibited a broad diversity, as they were assigned to 25 different XbaI profiles. Interestingly, in the majority of colistin-resistant mcr-negative E. coli and K. pneumoniae isolates non-synonymous polymorphisms in pmrAB were detected. Our findings demonstrated high occurrence of colistin-resistant E. coli and K. pneumoniae carrying mcr-1 on transferrable plasmids in poultry and pig slaughterhouses and indicate their dissemination into surface water.


INTRODUCTION
Since the 1950s, colistin (polymyxin E) has been extensively used in the European animal production (Koyama et al., 1950) to prevent and treat gastrointestinal infections caused by Gram-negative bacteria (e.g., diarrhea in pigs caused by Escherichia coli and Salmonella spp. as well as colibacillosis in poultry) (EMEA, 2002). Moreover, it was also used in a lower dosage as a feed additive until the ban of antimicrobial growth promoters in the European Union (EU) in 2006 (EMA, 2016).
Despite its nephrotoxicity and neurotoxicity, colistin was re-introduced into human therapy to treat infections caused by multidrug-resistant Acinetobacter baumannii and Pseudomonas aeruginosa or carbapenemase-producing Enterobacteriaceae (CPE) (Azzopardi et al., 2013). Due to its high impact, the World Health Organisation (WHO) included colistin into the group of the "highest priority critically important antimicrobials" for human medicine (World Health Organization, 2019). Alongside with other antibiotics of the last resort (e.g., tigecycline, amikacin and the new combinations of ceftazidime-avibactam and ceftozolane-tazobactam), its use is restricted to clinical cases for which no alternative options are available (Nation and Li, 2009). However, in 2016, colistin was also classified as a highly important antimicrobial (VHIA) in the veterinary sector by the World Organisation for Animal Health (OIE, 2018). Data from Germany indicate a reduction of colistin sales between 2011 and 2016 by 45.7% from 127 to 69 tons. However, from 2016 its sales have been slightly increasing and reached 74 tons in 2018 making up 10.2% of the total amount of antimicrobials sold for the veterinary use in Germany (BVL, 2019).
In Gram-negative bacteria, colistin interacts with lipopolysaccharide (LPS) and phospholipids in the outer cell membrane. Due to the competitive displacement of divalent cations Ca 2+ and Mg 2+ from the phosphate groups of membrane lipids (Falagas and Kasiakou, 2005), both cell membranes are disrupted leading to the leakage of intracellular contents and subsequent bacterial death.
Before 2015, colistin resistance in Enterobacteriaceae was assumed to be caused due to chromosomal mutations in genes (esp. pmrA/B and phoP/Q and mgrB) encoding regulatory proteins that influence transcription of enzymes that modify the lipopolysaccharide (Moskowitz et al., 2012;Quesada et al., 2015). But the description of the first plasmid-encoded, mobilizable colistin resistance gene (mcr-1) in E. coli from livestock in China and retail meat as well as in Chinese clinical Klebsiella pneumoniae isolates (Liu et al., 2016;Sun et al., 2018b) raised serious public health concern on the emergence of colistin-resistant bacteria.
Due to a high number of colonized animals, slaughterhouses might represent a significant source of introduction of mcr genes into the food chain, e.g., through possible contamination of carcasses and products (Irrgang et al., 2016;Inderbinen, 2017). Furthermore, slaughterhouse workers with occupational exposure to colonized animals and contaminated process water as well as employees of the wastewater treatment plants (WWTPs) might be exposed to an increased risk of colonization (Dohmen et al., 2017). Moreover, due to insufficient wastewater treatment by in-house and municipal WWTPs (mWWTPs), livestock wastewater might be an important route for dissemination of mcr-1-carrying bacteria into the environment (Hembach et al., 2017).
On the basis of the high prevalence of colistin-resistant Enterobacteriaceae in livestock feces, these bacteria might accumulate in process waters and wastewater from slaughterhouses. These waters might represent potential reservoirs that can contribute to a broad spread of the resistance to other environmental ecosystems including surface waters. So far, no data on the occurrence and characteristics of colistin-resistant Enterobacteriaceae in process waters and wastewater from German poultry and pig slaughterhouses have been reported. Furthermore, information on the impact of slaughterhouse wastewaters for the dissemination of this resistance is scarce and needs to be determined. Thus, this study aimed to evaluate their occurrence in the delivery and unclean areas of German poultry and pig slaughterhouses as well as in their in-house WWTPs. Moreover, their further spread into surface waters via municipal WWTPs was also investigated.
This hypothesis was tested using selective culture-dependent methods, followed by phenotypic and molecular characterization of the recovered isolates.

Sampling and Sample Preparation
Sampling and sample preparation of process waters and wastewater taken in poultry and pig slaughterhouses, their inhouse WWTPs as well as mWWTPs and on-site preflooders have been previously described (Savin et al., 2020a,b).
A total of 185 water samples were included in the study. Briefly, 82 samples of process waters and wastewater accruing in the delivery and unclean areas during slaughtering and cleaning operations were collected from two poultry slaughterhouses. Samples were taken at seven sampling sites: transport trucks (n = 5), transport crates (n = 10), stunning facilities (n = 10), scalders (n = 10), eviscerators (n = 10), production facilities (n = 5), influent (n = 16), and effluent (n = 16) of the in-house WWTPs. From each individual sample, one liter was collected using sterile Nalgene R Wide Mouth Environmental Sample Bottles (Thermo Fisher Scientific, Waltham, MA, United States). For more details please see Savin et al. (2020b).
Further 67 samples of process water and wastewater were collected from the delivery [animal transporters (n = 10), holding pens (n = 7)] and unclean areas [scalding and dehairing water (n = 10), aggregate wastewater from production facilities (n = 10)] as well as the in-house WWTPs (in-and effluent, each n = 15) of two pig slaughterhouses during slaughtering and cleaning operations. Additionally, 18 samples were collected from the influents (n = 9) and effluents (n = 9) of the mWWTPs receiving the wastewater from the investigated pig slaughterhouses for the final treatment. Their on-site preflooders upstream (n = 9) and downstream the discharge points (n = 9) were sampled as well. At each site, one liter was collected in sterile polyethylene Nalgene Wide Mouth Environmental Sample Bottles (Thermo Fisher Scientific, Waltham, MA, United States). For more details please see Savin et al. (2020a).

Cultivation, Identification and Susceptibility Testing of Target Polymyxin-Resistant Lactose-Fermenting Enterobacteriaceae
Water samples were screened for polymyxin-resistant lactosefermenting Enterobacteriaceae (E. coli, Klebsiella spp., and Enterobacter cloacae complex) using SuperPolymyxin medium . For cultivation, aliquots of 100 µl and 1 ml of the original samples were applied onto SuperPolymyxin plates and incubated under aerobic conditions at 37 • C for 18-24 h. When possible, up to three colonies of lactose fermenters were picked based on their morphology and sub-cultured on Columbia Agar with 5% sheep blood (MAST Diagnostica, Reinfeld, Germany) at 37 • C for 18-24 h.
Identification of the isolates species was conducted by MALDI-TOF MS as previously described (Savin et al., 2020b).
The antimicrobial susceptibility testing of the isolates and transconjugants was performed by applying two different antibiotic susceptibility testing panels as well as epidemiological and clinical breakpoints. The first scheme (A) was based on broth microdilution according to CLSI guidelines (M07-A9) following application of epidemiological cut-off values of European Committee on Antimicrobial Susceptibility Testing (EUCAST) as recommended for isolates from livestock and food. The second one (B) was applied in order to assess the clinical relevance of recovered colistin-resistant isolates in human medicine. For this purpose, they were tested against clinically important antimicrobials for humans by microdilution method as previously described (Savin et al., 2020b). MICRONAUT MIC-Strips Colistin (MERLIN Diagnostika GmbH, Bornheim-Hersel, Germany) were used to test the colistin concentrations of up to 64 mg/L. Also, isolates of E. coli, K. pneumoniae, and E. cloacae complex that were cultivated from the same samples on CHROMagar TM ESBL plates (MAST Diagnostica, Reinfeld, Germany) as described previously by Savin et al. (2020b) and showed resistance to colistin, were included in this study.

Molecular Typing of Resistant Bacterial Isolates
Cell lysates prepared by boiling of bacterial suspensions (Aldous et al., 2005) were used as template for PCR. Determination of phylogenetic groups (A, B1, B2, C, D, E, F, clade I-V) of E. coli was conducted according to a previously published method (Clermont et al., 2013).
XbaI PFGE-Profiling of mcr-1-Positive E. coli and K. pneumoniae Isolates and mcr-1 Localization The phylogenetic relationship of the mcr-1-carrying E. coli and K. pneumoniae was assessed by XbaI macrorestriction via pulsed-field gel electrophoresis (PFGE) according to the PulseNet protocol (CDC, 2020). Plasmidal localization of the mcr genes was determined by S1-PFGE followed by Southern blotting and DNA-DNA hybridization against a digoxigenin-labeled PCR amplicon as previously described (Hammerl et al., 2018). The size of mcr-carrying plasmids was predicted on the basis of the S1-PFGE pictures with Bionumerics (Applied Math, Sint Marten-Latem, Netherlands; version 7.5) using Salmonella Braenderup (H9812) as size marker.

Conjugation Assays and Plasmid Analyses
In vitro conjugation experiments were conducted in liquid medium using the plasmid-free rifampicin-resistant E. coli recipient strain CV601 GFP at a donor:recipient ratio of 1:1 as previously described (Blau et al., 2018). Transconjugants were selected after incubation at 37 • C for 24-48 h under selective conditions on lysogeny broth (LB) agar (Sigma-Aldrich, St. Louis, MO, United States) containing colistin sulfate (1 µg/ml) and rifampicin (200 µg/ml) (w/v). Isolates that did not yield transconjugants were further subjected to filter mating assays with the rifampicin-resistant, lactose-negative E. coli recipient strain W3110 at a donor:recipient ratio of 10:1 . The selection of transconjugants was done on MacConkey agar (Sigma-Aldrich, St. Louis, MO, United States) containing colistin sulfate (1 µg/ml) and rifampicin (200 µg/ml) after incubation at 37 • C for 24-48 h under selective conditions. Potential transconjugants were subjected to PCR to confirm the presence of the mcr genes. Those transconjugants obtained with E. coli CV601 as recipient were additionally examined for GFP fluorescence using fluorescence microscope Axio Scope.A1 (Carl Zeiss Microscopy GmbH, Jena, Germany).

Transformation Assays
mcr-1-positive isolates that did not generate any transconjugants were further submitted to transformation experiments using NEB R 10-beta electrocompetent E. coli cells (New England Biolabs, Ipswich, MA, United States) and MicroPulser Electroporator (Bio-Rad, Hercules, CA, United States) according to manufacturer's protocols. Plasmid DNA was extracted from overnight cultures of mcr-1-positive isolates using GeneJET Plasmid Miniprep Kit (Thermo Fisher Scientific, Waltham, MA, United States) according to manufacturer's protocol. The transformants were selected on LB agar (Sigma-Aldrich, St. Louis, MO, United States) containing colistin sulfate (1 µ g/ml).

Detection of Enterobacteriaceae in Samples From Poultry and Pig Slaughterhouses as Well as From mWWTPs
Due to the growth of accompanying bacterial microbiota that belongs to intrinsically colistin-resistant genera (e.g., Proteus, Providencia, Morganella) and colistin-susceptible isolates on the selective agar plates as well as absence of sample replicates, it was not possible to perform accurate quantification of target bacteria. This could be considered as a limitation of this study.
In the poultry and pig slaughterhouses the target bacteria were recovered at almost all sampling points as shown in Figures 2, 3, respectively. Interestingly, only one out of nine samples taken in the effluent of the mWWTPs was positive for target colistinresistant bacteria. Moreover, no colistin-resistant target bacteria were detected in the on-site preflooders upstream the discharge point (Figure 3).
Overall, 129 isolates were recovered from 185 samples. Of the isolates, 50.4% were determined as E. coli, 26.3% as K. pneumoniae and 23.3% as isolates of the E. cloacae complex. The most frequently isolated species in poultry and pig slaughterhouses was E. coli, whereas in mWWTPs K. pneumoniae was more abundant.

Resistance Patterns [Scheme A (EUCAST) and Scheme B (KRINKO)] and MIC of Colistin per Species
Isolates of E. coli, K. pneumoniae, and E. cloacae complex exhibited various resistance patterns. The resistance rates using epidemiological cut-off values (Figures 4A-C) were higher and different in comparison to those obtained with clinical breakpoints (Figures 5A-C).
According to the scheme A, the recovered isolates were either susceptible or expressed low resistance rates to gentamicin, tigecycline and with exception of E. cloacae complex to carbapenems (IMP and MEM). The resistance rates to 3rd generation cephalosporins (CTX and CAZ) varied between isolated species and were in the range of 23.5% for K. pneumoniae and 46.7% for E. cloacae complex. The highest level of multiple drug resistance (MDR, combined resistance to CST, CIP, and TET) shown isolates of E. coli (49.2%), followed by K. pneumoniae (35.3%), and E. cloacae complex (33.3%). MICs of antimicrobials with undefined epidemiological cut-offs for E. cloacae complex (AMP, CHL, NAL, SMX, TMP, ETP, and FOX) are shown in Table 1. MIC values of colistin for recovered E. coli and K. pneumoniae isolates are shown in Tables 2-4. Among isolates of E. cloacae complex, MIC values of colistin varied between 16 and >64 mg/L.
According to the scheme B, the isolates with exception of K. pneumoniae had lower resistance rates to 3rd generation cephalosporins (CTX and CAZ). The differences varied between 12.3% for CAZ by E. coli and 26.6% for CTX by E. cloacae complex (Figures 4, 5). Furthermore, they were susceptible to temocillin, ceftazidime-avibactam, imipenem, meropenem, amikacin and, with exception of some E. cloacae complex isolates, to tigecycline. The highest 3MDRO rates (multidrug-resistant organisms with combined resistance to PIP, CTX, and CIP) were exhibited by K. pneumoniae (26.5%), followed by E. cloacae complex (20.6%) and E. coli (13.8%). However, if using piperacillintazobactam instead of piperacillin for determination of the MDR status, as recommended by Magiorakos et al. (2012), the 3MDRO rates were lower at 5.9% for K. pneumoniae, 3.3% for E. cloacae complex, and 3.1% for E. coli.

Occurrence of mcr Genes
Of the mcr genes screened, only mcr-1.1 was detected in 70.8% of E. coli and 20.6% of K. pneumoniae isolates. Colistin MICs of mcr-1-positive E. coli isolates ranged from 4 to 8 mg/L, whereas mcr-1 carrying K. pneumoniae isolates expressed higher level of resistance from 4 to >64 mg/L.
In poultry and pig slaughterhouses the mcr-1.1 carrying isolates of E. coli and K. pneumoniae were detected at almost all sampling points including scalding water and effluents of the in-house WWTPs. Furthermore, mcr-1.1 positive isolates of E. coli were detected in on-site preflooders downstream the discharge point. Detailed information on the isolation source and phenotypic resistance of mcr-1.1 carrying isolates of E. coli and K. pneumoniae is given in Table 2. PFGE Patterns of Colistin-Resistant mcr-1 Carrying Isolates, Location of mcr-1 Gene Overall, the analyzed isolates (n = 53, 46 E. coli and 7 K. pneumoniae) exhibited a broad diversity as they were assigned to 25 different XbaI profiles (20 for E. coli and 5 for K. pneumoniae). S1 nuclease PFGE, followed by Southern blot hybridization revealed the presence of mcr-1 carrying plasmids ranging between 30 and 360 kb. Interestingly, the majority of the isolates exhibited a predominant plasmid type of 30 kb ( Table 2). However, we had also determined a substantial number of isolates exhibiting the same XbaI macrorestriction patterns and/or plasmid profiles.
The transformation experiments with 15 E. coli isolates carrying mcr-1 gene (15/46) and for which no transconjugants could be obtained, did not yield any transformants.
In 81.5% (22/27) of K. pneumoniae isolates the pmrA and pmrB genes revealed polymorphic positions that were non-synonymous at the protein level. Additionally, four nonsynonymous polymorphisms were found in pmrA (37Ala→Thr, 57Glu→Gly, 147Ala→Glu, and 217Ala→Val) and six in pmrB (2Ala→Ser, 73Pro→x, 74Ser→x, 112Thr→Pro, 157Thr→Pro, 203Ser→Pro). In one K. pneumoniae isolate recovered from the on-site preflooder downstream the discharge point, a yet unknown insertion of nine amino acids (Gln-Leu-Gln-Gln-Leu-Ala-Arg-Val-Gly) between amino acid residues Glu-201 and Gln-202 of pmrB was identified. Detailed information on non-synonymous polymorphisms of individual E. coli and K. pneumoniae isolates, their origin and resistance phenotypes is given in Tables 3, 4, respectively.

DISCUSSION
Our study provides data on the occurrence of colistin resistant Enterobacteriaceae (E. coli, K. pneumoniae, and E. cloacae complex) in process waters and wastewater along the slaughtering processes in poultry and pig slaughterhouses, their in-house and mWWTPs as well as receiving waterbodies.
The highest prevalence of colistin-resistant bacteria was detected in poultry slaughterhouses. This is in accordance with other studies indicating frequent occurrence of colistin-resistant Enterobacteriaceae in the poultry production chain in Germany (Irrgang et al., 2016;Inderbinen, 2017). Current data from official bodies on antimicrobial usage in different animal species in Germany are not available. However, the Report of the Federal Ministry of Food and Agriculture on the Evaluation of the Antimicrobials Minimization Concept introduced with the 16th Act to Amend the Medicinal Products Act (16th AMG Amendment) indicates a higher usage of colistin in German poultry production in comparison to other livestock production chains (BMEL, 2019). Moreover, between 2014 and 2017 consumption of polypeptide antibiotics in broiler production in Germany slightly increased from 11 to 13 tons. Whereas in pig production chain polypeptide antibiotics are mostly used to treat piglets and for the treatment of fattening pigs a decrease from 4 tons in 2014 to 0.5 tons in 2017 was observed (BMEL, 2019). Thus, the higher use of colistin in poultry may coincide with the frequent occurrence of colistin-resistant bacteria in this production chain. Furthermore, in comparison to poultry, a longer life span and time gap between administration of antibiotic and slaughtering among pigs may result in a decrease of colistin resistance when selection pressure is absent. Moreover, the kind of antibiotic treatment, e.g., treatment of individual pigs or small groups thereof in comparison to the whole flock treatment, may also be responsible for the lower occurrence of colistin resistance among pigs and accordingly in the pig slaughterhouses (BMEL, 2010). Furthermore, our results are in line with the EU summary report on Antimicrobial Resistance in zoonotic and indicator bacteria from humans, animals and food in 2017/2018 (ECDC, 2020) showing increased colistin resistance in E. coli isolates from broilers compared to those from pigs.
From nine mcr genes tested, mcr-1 was the most prevalent one, which corroborates the study of Elbediwi et al. (2019) that emphasizes the global dissemination and high prevalence of mcr-1 gene among colistin-resistant bacteria isolated from animals and food products worldwide. With prevalences of 0.04 to 20.3%, mcr-1 is predominantly detected in Enterobacteriaceae isolates (E. coli, Klebsiella spp., Enterobacter spp., Salmonella spp., and Shigella spp.) from livestock, retail meat (1.4-19%) and to a lesser In Germany, colistin-resistant isolates from turkey and broilers food chains show the highest mcr-1 prevalence in comparison to pigs and cattle (Irrgang et al., 2016;Borowiak et al., 2020). Thus, livestock and poultry are considered as an origin of mcr-1 and is its important reservoir for transmission to humans (Liu et al., 2016). Based on the wide dissemination of mcr-1, EMA's (European Medicines Agency) Antimicrobial Advice Ad Hoc Expert Group (AMEG) advised to minimize sales of colistin for use in animals EU-wide to achieve a 65% reduction in 2016 (EMA, 2016). The genes mcr-2 to mcr-9 have not been detected in our study. This could be due to their limited geographical distribution and bacterial host range (Borowiak et al., 2020) as well as substantially low prevalence compared with mcr-1. While mcr-2 to mcr-8 are being detected mostly in E. coli and K. pneumoniae isolates from pigs and poultry in China and South Europe AbuOun et al., 2017;Borowiak et al., 2017;Carattoli et al., 2017;Yin et al., 2017;Wang et al., 2018;Yang et al., 2018), mcr-9 and mcr-10 were discovered in clinical strains of Salmonella enterica serotype Typhimurium (Carroll et al., 2019) and Enterobacter roggenkampii, respectively (Wang et al., 2020). Currently in Germany, mcr-3 was detected in Aeromonas spp. isolates of fish origin (Eichhorn et al., 2018). Furthermore, mcr-4 has been frequently identified in different Salmonella serovars from poultry meat and pork (Borowiak et al., 2020) as well as mcr-5 has been detected in E. coli and Salmonella isolates of livestock origin (Hammerl et al., 2018;Borowiak et al., 2020). Kneis et al. (2019) reported high abundances of mcr-3, mcr-4, mcr-5, and mcr-7 in German mWWTPs. However, the possible origin sources such as households, health care facilities, livestock farming sites or slaughterhouses as well as bacterial host ranges have not been identified (Kneis et al., 2019).   Escherichia coli isolates carrying mcr-1 on transferable IncHI2 plasmids were detected in on-site preflooder downstream the discharge point of mWWTP. Possible entry sources could be run-offs from the fields fertilized with contaminated manure (Guenther et al., 2017) and feces of wild animals (birds) (Lin et al., 2020). Previously, (Zurfluh et al., 2016) and (Falgenhauer et al., 2019) detected mcr-1 harboring E. coli in surface water and rivers in Switzerland and Germany, respectively. Moreover, in our study mcr-1-positive K. pneumoniae was recovered from poultry scalding water. This could be a possible source of contamination of carcasses and products and lead to the introduction of mcr-1 carrying K. pneumoniae into the food chain. (Schrauwen et al., 2017) reported that 24.8% of retail chicken meat in Netherlands were positive for mcr-1, carried mostly by E. coli and to a lesser extent by K. pneumoniae. Furthermore, 40.6% of poultry meat samples originating from Germany were contaminated with mcr-1 producing bacteria (Inderbinen, 2017). Some of the mcr-1 carrying isolates recovered from wastewater used for cleaning of stunning facilities and influents of in-house WWTP from poultry slaughterhouses belonged to ExPEC groups B2 and D, which are known to harbor more virulence factors than commensal strains and pose a zoonotic risk (Johnson et al., 2012). This enables the transmission of mcr-1-positive ExPECs of poultry origin to humans and represents a potential vehicle of mcr genes for human diseases, e.g., bloodstream and urinary tract infections (Izdebski et al., 2016;Zhong et al., 2019). Moreover, study of Zhuge et al. (2019) shown that mcr-1-positive E. coli of phylogroups B1 and F also possessed high virulence in rodent models for ExPEC-associated human infections and could therefore pose an elevated risk of infections for humans.
According to the classification of Magiorakos et al. (2012) and applying epidemiological cut-off values, target isolates showed high percentage of multidrug resistance (combined resistance to CST, CIP, and TET) with the highest rate of 49.2% for E. coli. However, it is important to note that from a human clinical perspective, the antibiotic groups are not considered to be equally clinically relevant (Exner et al., 2017). Thus, from the point of view of KRINKO, the multidrug resistance rates (3MDRO rates, combined resistance to TZP, CTX, and CIP) were low with the highest percentage of 5.9% for K. pneumoniae. For this evaluation the combination of piperacillin/tazobactam instead of piperacillin is used, as in the clinical practice in Germany piperacillin is administered only in combination with β-lactamase inhibitors. Furthermore, applying clinical breakpoints, isolates were completely susceptible to reserve antibiotics ceftazidime-avibactam and tigecycline as well as carbapenems (IMP and MEM). Moreover, temocillin, which was introduced in 2019 for therapy of extended spectrum βlactamase (ESBL) and AmpC producers, and amikacin, classified by WHO as reserve second−line drug, were also effective against all isolates. Thus, these substances could be still effective in antimicrobial therapy in case of infection.
It was already reported that mcr-1 gene occurs frequently in isolates that are susceptible to most classes of antimicrobials (EMA, 2016). However, this contrasts with the findings of our study, as considerable percentage of colistin-resistant mcr-1 positive isolates from our study showed resistance of up to eleven  antibiotics, including clinically relevant ones. This reinforces the theory that possible transmission of mcr-1 gene to highly virulent bacteria carrying other antimicrobial resistance genes, e.g., ESBL or carbapenemases would narrow clinical therapeutic options (Forde et al., 2018). Zheng et al. (2017) and Zheng et al. (2016) reported on E. coli isolates from blood stream infections which co-produce NDM-1 and MCR-1. In our study mcr-1 gene was detected in a wide range of plasmid types such as IncI1, IncHI2, IncX4, IncF, and IncI2, which is in consent with other reports. mcr-1 located on IncI1 plasmids was detected in E. coli recovered from pig manure (Guenther et al., 2017) and chicken feces (Hassen et al., 2020). E. coli isolates recovered from pigs in Portugal carried mcr-1 on IncHI2 and IncX4 plasmids . In another study, mcr-1 gene was located on IncX4 and IncHI2 plasmids in E. coli from broilers and veal calves in Netherlands (Veldman et al., 2016). Furthermore, Gelbíčová et al. (2019) isolated E. coli, K. pneumoniae, and Citrobacter   braakii from raw turkey meat and liver which harbored mcr-1 gene on IncX4, IncHI2, and IncI2 plasmids. In addition to livestock and food products, MCR-1-producing E. coli which carry the resistance on IncX4, IncHI2, and IncI1 types of plasmids, were isolated from different environmental sources such as surface water in Germany (Falgenhauer et al., 2019) and public seawater beach in Norway (Jørgensen et al., 2017). The association of mcr-1 gene with insertion sequence ISApl1 might play a major role in its mobilization, its further successful establishment in BHR plasmids and subsequent dissemination among Enterobacteriaceae (Snesrud et al., 2016;Poirel et al., 2017). On the other hand, without colistin exposure, ISApl1 is able to facilitate the deletion of resistance genes, as described by Zhang et al. (2019) for mcr-1 and mcr-3.19.
The co-transfer of the decreased susceptibility to fluoroquinolones (MIC of CIP 0.25 mg/L) by the majority of the isolates recovered in the poultry slaughterhouses could be due to plasmid-mediated quinolone resistance (PMQR) genes. They are known to provide only low-level resistance that by itself does not exceed the clinical breakpoint of >0.5 mg/L for susceptibility (Jacoby et al., 2014). Furthermore, resistance to tetracyclines was co-transferred by the isolates from pig slaughterhouses, as tetracycline resistance genes are often located on mobile genetic elements such as plasmids, transposons, conjugative transposons, and/or integrons (Roberts, 2003). Thus, fluoroquinolones and tetracyclines, which make up 25.7% of the total antimicrobial usage in the veterinary medicine in Germany (BVL, 2019), may impose a selective pressure that could favor the selection of mcr genes, even without use of colistin and vice versa. Moreover, Savin et al. (2020a) reported on antimicrobial residues of ampicillin, ciprofloxacin, and ofloxacin detected in German mWWTPs which exceeded their PNECs (Predicted No Effect Concentration) (Bengtsson-Palme and Larsson, 2016). Ofloxacin exceeded its PNEC even after dilution of the treated wastewater with the recipient water. This may contribute to the co-selection of mcr-1 carrying bacteria in surface water, whereas the residues of ampicillin may promote the dissemination of mcr-carrying strains of species with intrinsic resistance to this antimicrobial (e.g., Klebsiella spp., E. cloacae complex).
The great majority of colistin-resistant E. coli and K. pneumoniae which were tested negative for known mcr genes harbored chromosomal point mutations in the pmrAB coding regions. For E. coli, a mutation at the amino acid position 10 in pmrB has been detected by Cannatelli et al. (2017) leading to the substitution 10Leu→Pro that confers resistance to colistin. However, in our study, the polymorphisms at this position resulted in leucine to arginine substitution. One K. pneumoniae isolate recovered from the wastewater used for cleaning of poultry stunning facilities demonstrated mutation 157T→P (PmrB) that has been previously reported in K. pneumoniae from patients and healthy humans (Olaitan et al., 2014) as well as in clinical colistin-resistant K. pneumoniae carbapenemase (KPC)-producing isolates (Leung et al., 2017). Furthermore, a substitution 217A→V (PmrA) that has been already described in colistin-resistant isolates from clinical blood cultures (Esposito et al., 2018) was found in isolates recovered from the influent of mWWTPs and on-site preflooders. To determine whether other detected polymorphisms in E. coli and K. pneumoniae cause resistance to colistin, complementation assays are needed.
We are not aware of other studies in Germany that investigated such environmental samples (i.e., process waters and wastewater) which have been taken directly in the slaughterhouses and their on-site WWTPs that underlines the novelty of our study. In conclusion, our results indicate high prevalence of E. coli isolates which carry mcr-1 on a wide variety of transferable plasmids in process water accruing along the slaughtering process in German poultry slaughterhouses. This may pose an elevated risk of colonization for slaughterhouse employees with occupational exposure to process water and wastewater. Furthermore, despite strict hygiene rules established in German slaughterhouses, mcr-1 carrying bacteria could be introduced into the food chain through cross-contamination (e.g., scalding water). Moreover, due to insufficient treatment of wastewater, such strains were discharged into the environment. In order to determine the persistence of mcr-1 carrying E. coli isolates in the receiving water bodies, further investigations are needed. Furthermore, besides colistin, overall reduction of the use of antibiotics in livestock is required, as it was shown that mcr-1 can be also co-selected by fluoroquinolones and tetracyclines. In this way, the input of resistant bacteria into the slaughterhouses can be reduced. Additionally, as mcr-1 carrying isolates were detected in the effluent of the WWTPs, a broad dissemination the environment can be expected. Thus, this study supports the necessity of the implementing of advanced wastewater treatment technologies to limit the exposition of the environment with bacteria expressing resistances against last resort antimicrobials.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
MS: project administration, conceptualization, methodology, investigation, writing -original draft, and visualization. GB, KS, and RS: writing -review and editing. KB, CH, and JH: investigation, writing -review and editing. MP and ES: investigation. JK: conceptualization, writing -review and editing, supervision, and funding acquisition. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by the BMBF (Federal Ministry of Education and Research) funding measure HyReKA (02WRS1377) to MS, MP, ES, RS, and JK. The scientific work of Jens A. Hammerl was supported by grants of the Bundesinstitut für Risikobewertung (43-001 and 1322-648). JH received further grants of the European Joint Programme One Health EJP (ARDIG and Full_Force) and the BMG project GÜCCI. The contribution of KB was funded by the BMBF funding measure WavE (O2WAW1402). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.