A prospective point prevalence study was conducted in four of the largest hospitals (in total 3550 beds) in the Northern Netherlands between November 2012 and February 2013, covering a total population of approximately 2.85 million people. The Hospital Ethical Committee of the University Medical Center Groningen (UMCG) was informed and patients were approached to voluntarily participate in the study. Patients included in this study provided their written informed consent and a questionnaire concerning epidemiological and clinical data. The following high-risk wards for antibiotic-resistant microorganisms were selected: intensive care units (ICU), vascular surgery, internal medicine hematology/oncology, and dialysis wards (both for in- and outpatients). Gynecology and neurology (low-risk wards) were also included for comparison. From the largest German university hospital in the same (border) region, patients from four ICUs, a surgical ward, and a hematology/oncology ward were screened during October and November 2012 and included in the study. After consent agreement, all admitted patients from the studied wards were screened until completing a minimum of 100 samples per hospital.

The study in healthy people living in the Northern Netherlands was conducted retrospectively, using control patients included in a previous case–control study on microorganisms causing gastroenteritis. Control subjects were patients attending their general practitioner for a variety of medical questions, but no gastrointestinal problems, in the period between August 2010 and December 2012 (Bruijnesteijn van Coppenraet et al., 2015). No prevalence study was performed in the community in Germany.

A total of 445 rectal swabs (Copan ESwab^TM) were taken from hospitalized patients (median age = 66 years, range 18–99 years) in the Northern Netherlands, 51.7% (n = 230) from men and 48.3% (n = 215) from women. A total of 328 (73.7%) patients were screened at high-risk wards and 117 (26.3%) patients were screened at low-risk wards (Table 1). In the German university hospital, 242 patients (median age = 64 years, range 0–94 years) were included, 64.5% (n = 156) men and 35.5% (n = 86) women. Of these 242 patients, 140 were screened only for ESBL, 22 only for VRE, and 80 for both. From the Dutch community study, 400 frozen faeces samples were included; 41% (n = 164) from men and 59% (n = 236) from women, 12% of the samples were from children. The median age of the healthy individuals was 47.5 years (range 0–84 years).

Enterococci isolates from Netherlands were screened by in-house PCR for IS16 (a marker for specific hospital-associated strains), vanA and vanB genes as described previously (Clark et al., 1993; Werner et al., 2011). The GenoType Enterococcus (Hain Lifescience GmbH) was used in enterococci isolates from Germany, which detects species and genotypes vanA, vanB, vanC1, and vanC2. ESBL and VRE positive isolates were sent to our hospital for further characterization.

Enterobacteriaceae isolates resistant to third generation cephalosporins and natural chromosomal AmpC producers intermediate or resistant to cefepime were selected for DNA extraction using the UltraClean Microbial DNA Isolation Kit (Mo Bio Laboratories, Inc.) and further characterized for the presence of ESBL/AmpC genes using a DNA-array (Check-MDR CT103, Check-points, Wageningen, Netherlands) (Garcia-Cobos et al., 2015).

WGS was performed for all ESBL-E. coli and VRE isolates. For each isolate, several colonies (about 5 μL) of the culture were suspended in 300 μL microbead solution, which was subjected to DNA extraction with the Ultraclean Microbial DNA isolation kit (Mo Bio Laboratories, Carlsbad, CA, United States). The DNA concentration and purity were measured using a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States) and the Qubit double-stranded DNA (dsDNA) HS and BR assay kits (Life Technologies, Carlsbad, CA, United States). One nanogram of bacterial DNA was used for library preparation. The DNA library was prepared using the Nextera XT library preparation kit with the Nextera XT v2 index kit (Illumina, San Diego, CA, United States). The library fragment length was aimed at fragments with a median size of 575 bases and was assessed with the Genomic DNA ScreenTape assay with the 2200 TapeStation system (Agilent Technologies, Waldbronn, Germany). Subsequently, the library was sequenced on a MiSeq sequencer, using the MiSeq reagent kit v2 generating 250-bp paired-end reads. Sequencing was aimed at a coverage of at least 60-fold. MiSeq data were processed with MiSeq control software v2.4.0.4 and MiSeq Reporter v2.4 (Illumina, San Diego, CA, United States). Reads were quality-trimmed using the CLC Genomics Workbench software version 9.0.1 (CLC bio, Aarhus, Denmark) using default settings except for the following modifications: “trim using quality scores was set to 0.02” and “discard reads below length was set to 15.” Subsequently, trimmed-reads were de novo assembled with an optimal word size of 29 and a minimum contig length of 500. Metrics on raw read and assembly level are provided in Supplementary Table S1.

A genome-wide gene-by-gene comparison approach was used to determine the genetic relatedness using SeqSphere⁺ version 3.4.0 (Ridom GmbH, Münster, Germany) (Leopold et al., 2014).

Genome assemblies from the VRE isolates were analyzed using the E. faecium cgMLST scheme previously published, considering a cluster alert distance of 20 different alleles (de Been et al., 2015).

An ad hoc cgMLST and whole-genome MLST (wgMLST) scheme was determined for E. coli isolates using the MLST⁺ target definer function with default parameters² and E. coli K-12 as a reference (GenBank accession no. NC_010473.1). The filters applied to reference genome were: “minimum length filter” that discards genes shorter than 50 bases; “start codon filter” that discards all genes that contain no start codon at the beginning of the gene: “stop codon filter” that discards all genes that contain no stop codon, more than one stop codon or if the stop codon is not at the end of the gene: “homologous gene filter” that discards all genes that have fragments that occur in multiple copies in a genome (with identity ≥90% and more than 100 bases overlap); “gene overlap filter” that discards the shorter of two overlapping flanking genes if these genes overlap >4 bp. The remaining genes were then used in a pairwise comparison using BLAST (Leopold et al., 2014) with 45 query genomes (Supplementary Table S2a). All genes of the reference genome that were common in all query genomes with a sequence identity of ≥90 and 100 overlap, and with the default parameter stop codon percentage filter turned on, formed the final cgMLST scheme; this discards all genes that have internal stop codons in >20% of the query genomes. Additionally, 26 plasmid sequences (Supplementary Table S2b) were added to exclude such genes are part of the cgMLST typing scheme. The final cgMLST scheme consisted of 1.771 targets/genes, and 2329 accessory genes were additionally included for the wgMLST scheme (Supplementary Tables S3, S4). The minimum coverage of the genome assemblies was 20 times (Supplementary Table S1) and the percentage of good targets/genes included in the cgMLST were 97.6% for E. coli and 98.6% for E. faecium (Supplementary Tables S5, S6).

Furthermore, to determine the genetic relatedness, the genetic distance for the E. coli isolates was calculated as the proportion of allele differences: dividing the number of allele differences between two genomes by the total number of genes commonly shared by those two genomes (Kluytmans-van den Bergh et al., 2016). In this study, thresholds for genetic distance were described to discriminate between epidemiologically related and unrelated E. coli isolates as 0.0095 when using wgMLST and 0.0105 for cgMLST.

Escherichia coli STs were determined uploading genome assemblies to SeqSphere+ software following the scheme of Wirth et al. (2006). Sequence genomes with no conclusive results for the 7-gene MLST were uploaded to the Enterobase database³. Additionally, E. coli major phylogenetic groups (A, B1, B2, and D) were analyzed in silico by using MLST⁺ Target Definer function of SeqSphere⁺, including the chuA, yjaA, and TSPE4.C2 loci (Clermont et al., 2000).

Genome assemblies were also uploaded to the Center for Genomic Epidemiology to extract information on resistance genes (ResFinder) and virulence factors (VirulenceFinder), and species confirmation for VRE and ESBL-E. coli (KmerFinder), and serotype (SerotypeFinder) and plasmid replicons (PlasmidFinder) for ESBL-E. coli (Zankari et al., 2012; Carattoli et al., 2014; Hasman et al., 2014; Joensen et al., 2014, 2015; Larsen et al., 2014).

In the Dutch hospital prevalence study, associations between ESBL and ARE carriage and the following variables were analyzed: length of hospital stay, antibiotic use, and (low- or high-risk) ward. Information was gathered by the questionnaires. Statistical analyses were performed using SPSS for Windows, v. 20.0. Univariate analyses were performed using the Fisher’s exact or Chi-square methods for categorical variables. The Mann–Whitney U-test was used as a non-parametric tests in variables with no normal distribution. Results with a p-value of ≤0.05 were considered to be statistically significant. All p-values are two-tailed.

Thirty-four isolates from 27 of the 445 included patients admitted to hospitals in the Northern Netherlands (6.1%) were confirmed ESBL and/or pAmpC positive. A total of 85.2% (23/27), 14.8% (4/27), and 3.7% (1/27) of these patients were positive for ESBL, pAmpC, and both, respectively. Among the 34 isolates, 32 were E. coli, of which 29 were ESBL positive and three were pAmpC producers. Resistance genes detected in the E. coli isolates are shown in Table 2. CTX-M-15 (n = 8) and CTX-M-14 (n = 8) were the most prevalent ones. The other two isolates were an Enterobacter cloacae, containing a CTX-M-1-like gene and a pAmpC CMY-II producing Proteus mirabilis. At high-risk wards, 19 patients (5.8%) were found with ESBL/pAmpC isolates compared to eight patients (6.8%) at low-risk wards (p = 0.68; NS). No association was found between ESBL/pAmpC carriage and antibiotic use, length of hospital stay or ward (Table 1).

In the German hospital, a total of 18 isolates from 17 patients (17/220; 7.7%) were ESBL positive. Sixteen isolates were E. coli and two were Klebsiella pneumoniae. Of these, 12 E. coli and one K. pneumoniae isolates were available for molecular testing. Six out of 12 (50%) E. coli isolates and the K. pneumoniae isolate had a CTX-M-15 gene (Table 2).

In the retrospective Dutch community study, 11 patients (11/400; 2.75%) were ESBL/pAmpC positive: 10 ESBL-E. coli (CTX-M-1 being the most prevalent gene) and one pAmpC E. coli (Table 2). Overall, no carbapenem resistance was observed neither in the community nor in the hospitals.

Among ESBL/pAmpC-E. coli isolates from Dutch hospitals, the most prevalent ST was ST131 (CC, ST131; n = 5, 15.6%), all of them belonging to phylogroup B2 (Table 2). In the Dutch community isolates, 10 different STs were found, most of them belonging to CC ST10 (n = 3, 27.3%) and one isolate to ST131 (phylogroup B2). In the German hospital, the most prevalent STs were ST38 (33.3%) and ST10 (33.3%) (Table 2).

In the Dutch hospitals, 105 patients (105/445; 23.6%) were colonized with ARE, including six patients (6/445; 1.3%) with VRE. All ARE were positive for IS16 and all VRE were vanB positive. Colonization of ARE (and VRE) was associated with high-risk wards (p < 0.001), prolonged hospitalization (p < 0.001), and use of antibiotics (p = 0.05), especially penicillins and fluoroquinolones (p < 0.001) (Table 3).

In the border German university hospital four (4/102; 3.9%) VRE isolates were isolated. Three of them were vanA positive and one was vanB positive.

In the retrospective Dutch community study, six ARE (6/400; 1.5%) were found, three of them were IS16 positive. Only one vanA-VRE (1/400; 0.25%) was found, this strain was amoxicillin susceptible and IS16 negative.

Genome assemblies of 55 ESBL-E. coli [Dutch community (n = 11), Dutch hospitals (n = 32), and German hospital (n = 12)] of this study were analyzed by a gene-by-gene approach and the allelic distance from the cgMLST and wgMLST were visualized in a minimum spanning tree (Figure 1 and Supplementary Figure S1, respectively).

Six groups of isolates with a lower number of different alleles (≤35) by cgMLST were further analyzed. Supplementary Table S7 summarizes the origin of the isolates in every group and the core and whole-genome genetic distance. Those groups formed by isolates with an epidemiological link (isolated from the same patient; group 1, 4, 5a, 6a, and 7), showed a core and whole-genome genetic distance lower than 0.0030 and 0.0046, respectively. In addition, isolates of group 5b, although with unknown epidemiological link, had a core genetic distance of 0.0063 and a whole-genome genetic distance of 0.0076. Both isolates were positive for CTX-M-14, however, no plasmid replicons were found in one of them (isolate 38_Esco_HA-DE) (Supplementary Table S7).

Among those groups including isolates with non (or unknown) epidemiological link, the core genome genetic distance was between 0.0122 and 0.0199 and the whole-genome genetic distance was between 0.0104 and 0.0208 (groups 2, 3, 6b, and 6c; Figure 1). Resistance and virulence profiles of the isolates are shown in Supplementary Table S8.

A minimum spanning tree was created for the 11 VRE isolates [Dutch community (n = 1), Dutch hospitals (n = 6), and German hospital (n = 4)]. Two clusters of isolates from different patients were observed (Figure 2). One cluster of four vanB-VRE isolates from the Dutch hospital belonged to cluster type (CT) 110 (ST17); two isolates were from the same ward in hospital A and the other two isolates were isolated from different wards in hospital B. The other cluster of two vanA-VRE isolates were isolated from different wards from the German hospital (CT 20, ST203). The resistance and virulence genotypes of VRE isolates are shown in Supplementary Table S8.

Sequence data obtained in this study has been deposited at the National Center for Biotechnology Information under BioProject no. PRJNA352198.