More than 2,500 food samples of different origins (poultry, cattle, swine, vegetables) and matrices (meat and meat preparations, raw milk, cheese, vegetables) were taken by official food inspectors and investigated by German state laboratories (Saxony, Lower Saxony, Hesse, Bavaria). Food samples from processing plants, retail and raw milk samples were collected at the farm level. From each sample, 25 g were investigated by a non-selective pre-enrichment step for 18–24 h at 37°C in lysogeny broth (LB) following selective cultivation of 10 μl aliquots on MacConkey agar supplemented with 1 mg/L cefotaxime (CTX, Sigma-Aldrich, Munich, Germany) for 18–24 h at 37°C. The identification of E. coli was confirmed by MALDI-TOF (Biotyper, Bruker). From each sample one E. coli isolate phenotypically resistant to CTX was sent to the German National Institute for Risk Assessment (BfR). Positive samples were obtained from all analyzed matrices, even though only raw milk cheese was burdened and only one vegetable sample was tested positive. The ESBL genotype of 437 isolates was verified by PCR and Sanger sequencing as previously described (Rodriguez et al., 2009). Isolates positive for bla_CTX-M-15 were included in this study and further characterized. Phylogenetic groups were classified as previously described (Doumith et al., 2012). The antimicrobial resistance pattern was determined by microbroth dilution according to CLSI guidelines (CLSI M07-A9) at the National Reference Laboratory for Antimicrobial Resistance (NRL-AR, BfR). The used antimicrobial panel was in concordance to the decision 2013/652/EU of the European Union commission and was carried out with microtiter plates from TREK Diagnostic Systems (Thermo Fisher Scientific, Schwerte, Germany).

Molecular characterization was performed using pulsed-field gel-electrophoresis (PFGE) and multi-locus sequence typing (MLST). Phylogenetic relationship of the isolates was determined using XbaI-PFGE analysis according to the PulseNet protocol (https://www.cdc.gov/pulsenet/pathogens/protocols.html). MLST was performed using the Achtman scheme (adk, fumC, gyrB, icd, mdh, purA, recA; http://mlst.warwick.ac.uk/mlst/dbs/Ecoli).

The location of the bla_CTX-M-15 gene was determined by S1-nuclease PFGE (1–25 s 17 h, 120°, 6 V/cm) with subsequent southern blot hybridization using a bla_CTX-M-15 PCR-probe (Rodriguez et al., 2009). A chromosomal location of the bla_CTX-M-15 gene was assumed for isolates in which no positive signal had been detected. Plasmids harboring the bla_CTX-M-15 gene were isolated by alkaline lysis and transformed into E. coli DH10B™ competent cells (Invitrogen™, Thermo Fisher Scientific, Schwerte, Germany) (Birnboim and Doly, 1979; Rodriguez et al., 2009). Selection of transformed cells was carried out on LB agar supplemented with 1 mg/L CTX. Transformation of plasmids was confirmed by PCR. Incompatibility groups of the transferred plasmids were determined by PCR using the PBRT kit (Diatheva, Cartoceto PU, Italy). When transformation experiments were inconclusive, incompatibility group of the bla_CTX-M-15 harboring plasmid was determined using PFGE/southern blot hybridization with probes specific for IncF and IncI1. The ISEcp1 element was detected by using modified ALA3/ALA4 Primer (5′-TTTGCGCATACAGCGGCACAC-3′/5′-CTATCCGTACAAGGGAG-3′) (Rodriguez et al., 2014).

Additionally, whole genome sequencing of the isolates was performed. Therefore, genomic DNA was isolated from overnight cultures using the PureLink^®; Genomic DNA Mini Kit (Thermo Fisher Scientific, Schwerte, Germany). A NexteraXT library was generated and sequenced on a MiSeq benchtop sequencer (Illumina, CA, USA) with 2 × 300 bp paired-end reads. Raw reads were assembled using SPAdes (v 3.5.0) (Bankevich et al., 2012). Whole-genome-based phylogenetic analysis was performed using HarvestSuite (ParSNP) (Treangen et al., 2014).

Resistance genes, virulence genes, serotype and pMLST were predicted using the web-based tools of the Center for Genomic Epidemiology (Zankari et al., 2012; Carattoli et al., 2014; Joensen et al., 2014, 2015).

Whole genome sequences of the isolates have been deposited in the European Nucleotide Archive (ENA). Accession numbers of isolates RL16, RL25, RL36, RL40, RL63, RL162, RL195, RL212, RL224, RL230, RL239, RL330, RL331, RL345, RL346, RL364, RL379, RL406-0, RL452,RL464, and RL465 are summarized in the Table S2.

From 437 E. coli isolates phenotypically resistant to 3rd generation cephalosporins obtained from animal derived food, 404 isolates were confirmed as ESBL/AmpC-producing bacteria. Of these, 21 (5.2%) isolates harbored the bla_CTX-M-15 gene. This is in agreement with the observations, that while resistance to 3rd generation cephalosporins in Germany and other European countries is frequently mediated by CTX-M-15 enzymes in isolates from human origin, they are of low prevalence in bacteria from livestock (Pfeiffer et al., 2013; Brolund, 2014; Valentin et al., 2014). A comparable study from the UK even found no CTX-M-15-producing E. coli in food samples from animals and non-animal sources while there was an overall prevalence of ESBL-producing E. coli of 27.5% of the meat samples (Randall et al., 2017)

The main characteristics of the isolates is given in Table 1. There are distinct similarities regarding detected STs, pMLST of IncF plasmids, class 1 integrons or virulence between isolates obtained from food and animal origin. These results suggest a transmission from animal to food (Fischer et al., 2014).

There is a predominance of isolates belonging to clonal complex (CC) 10 (n = 8) and CC23 (n = 4) of the phylogenetic group A. Although the sequence types (ST) 38 and ST131 are typically observed in humans (Rodriguez et al., 2014),the frequent presence of ST167/ST617 (CC10) and ST410 (CC23) isolates from food samples in this study concurs with previous reports of isolates from animal samples within the same time period (2011–2013) and from human stool samples. In particular, those reported isolates harbored bla_CTX-M-15 and were members of the same phylogenetic group (Fischer et al., 2014; Ben Sallem et al., 2015). This underlines a possible transmission from animals to humans via contaminated food.

All CC10 isolates belong to ST167. Five of them showed an almost identical XbaI PFGE pattern (P1; Figure 1). These strains have been isolated over a period of 6 months from samples of raw milk (n = 4) and beef (n = 1) in Saxony (Eastern Germany). Isolates were obtained from four different samples taken at different time points from three different postal code locations. Milk samples were obtained from farms that were nearby, whereas the beef sample was taken about 200 km away. Therefore, a geographical spread of the clone might have occurred. In general, the phylogenetic group A clonal complex 10 (ST10/167/617) represents a successful clonal lineage, which can be found in humans, livestock, as well as in companion animals (Ewers et al., 2012). In this study two additional ST167 isolates obtained from turkey meat showed similar PFGE restriction patterns (P2; Figure 1) which are distinguishable from those of the raw milk isolates. Nevertheless, all eight ST167 isolates cluster when performing phylogenetic analysis based on whole genome sequences (Figure 2) and all isolates of this clade belong to the same serotype as shown by NGS data (Table S1).

There are several isolates (n = 4) belonging to ST410 harboring mainly a chromosomal location of the bla_CTX-M-15 gene. Only in one isolate (RL212) the bla_CTX-M-15 was located on a 90 kb multireplicon IncF plasmid as described above. However, NGS data revealed a close phylogenetic relationship of all isolates (Figure 2). In the Supplementary Material, a comparison based on whole genome sequences with ST410 isolates from German livestock, companion animals, humans and environment is shown (Supplementary Figure 1). The food-related isolates can be found in three of the five clades (B, C, D), which otherwise comprise of isolates from farm or farm environment-related samples as well as samples of human origin. In concordance with the other strains of clade B, bla_CTX-M-15 of RL25 integrated at a distinct location in the rhsE cassette, which is known as a hotspot for insertion sequences and recombination in E. coli (Saier, 2008). The bla_CTX-M-15 of RL465 (Falgenhauer et al., 2016b) and RL406-0 also integrated at the same location known for the other members of clade C, at a defective lambdoid prophage region. The results of the current study further extend previous findings of interspecies circulation of ST410 clones to include food and point out the potential risk of contaminated food as transmission vehicles for consumers (Falgenhauer et al., 2016a).

Apart from the three ST410 isolates, a chromosomal localization is also likely for RL63 (ST176). The stable integration of the bla_CTX-M-15 genes into the chromosome is reported for different MLST variants (Falgenhauer et al., 2016a). These findings demonstrate that the chromosomal integration of the bla_CTX-M-15 gene occurred in several independent events and emphasize that a chromosomal location of this gene might be more common than anticipated (Rodriguez et al., 2014).

Apart from strain RL36, where the bla_CTX-M-15 seems to be located on the chromosome, the remaining seven ST167 isolates (PFGE pattern P1 and P2) harbor the bla_CTX-M-15 gene on 150–190 kb multireplicon IncFIA/FIB/FII plasmids. These large IncF plasmids (>150 kb), as well as the ST410 IncF plasmid, additionally harbored an aac(6)-1b-cr gene (plasmid mediated quinolone resistance gene). The further correlation of bla_CTX-M-15-encoding IncF plasmids with the detection of bla_OXA-1, catB3 and tet genes is also described by Lopez-Cerero et al. (2011). IncFII plasmids carrying bla_CTX-M-15 are known to be highly transferable (Carattoli, 2009). Except for plasmids from RL63 and RL379, all multireplicon IncF plasmids of this study belong to the pMLST F31:A4:B1, indicating plasmid-related spread of bla_CTX-M-15 carrying E. coli within different food production chains.

The bla_CTX-M-15 - aac(6)-1b-cr- harboring IncF plasmids were also associated with an 1,664 kb large class 1 integron containing a dfrA17/aadA5 gene cassette (Table 1) encoding for trimethoprim and aminoglycoside resistance. Similar class 1 integrons associated with bla_CTX-M-15 of phylogenetic group A E. coli were detected in isolates of livestock and companion origin as well as in samples of healthy humans worldwide (Dureja et al., 2014; Fischer et al., 2014).

The bla_CTX-M-15 gene was associated with an upstream located ISEcp1 element in all isolates, and has been frequently reported for bla_CTX-M-15 positive isolates (Lartigue et al., 2004; Smet et al., 2010). The association with ISEcp1 was even detected for chromosomally encoded bla_CTX-M-15 genes. This suggests that the resistance gene can be easily mobilized. Transposition of bla_CTX−M genes associated with the ISEcp1 element was demonstrated in vitro (Lartigue et al., 2004). The PCR for ISEcp1 was positive in all isolates except for one. For the isolate RL25 (ST410, chromosomal bla_CTX-M-15) insertion event of an IS1-element into the tnpA gene (transposase encoding) was detected at identical position to those found in ST410 isolates of different origin in the same clade (Fischer et al., 2014; Falgenhauer et al., 2016a). ISEcp1 elements, which are truncated by different IS elements, are occasionally reported and their effects on mobilization and expression of bla genes as well as their role in plasmid evolution have been discussed (Smet et al., 2010; Alonso et al., 2017).

In addition, the occurrence of virulence genes in the isolates was examined (Table S1). Most of the food isolates contained relatively few virulence genes. These included bacteriocins, glutamate decarboxylase, capsule synthesizing enzymes and serum survival genes. These virulence associated genes were also recognized in CTX-M-15-producing isolates from animals (Fischer et al., 2014). One isolate (RL346) harbored senB, which encodes an enterotoxin, that is responsible for enterotoxic activity of enteroinvasive E. coli (EIEC) and Shigella spp. (Nataro et al., 1995). Another isolate (RL452) carried two toxin genes (ncf1–cytotoxic necrotizing factor, involved in urinary tract infection (Mills et al., 2000), and vat–vacuolating autotransporter toxin, known to mediate increased fitness of uropathogenic E. coli (UPEC) during systemic infections (Nichols et al., 2016). This isolate belonged to phylogenetic group B2. These findings support the general assumption of low pathogenic potential in isolates of phylogenetic group A and B1 (major phylogenetic groups detected in this study) as compared to the higher virulence properties in isolates of the phylogenetic group B2 and D.

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