Molecular Characteristics of Extended-Spectrum Cephalosporin-Resistant Enterobacteriaceae from Humans in the Community

Objective To investigate the molecular characteristics of extended-spectrum cephalosporin (ESC)-resistant Enterobacteriaceae collected during a cross-sectional study examining the prevalence and risk factors for faecal carriage of extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae in humans living in areas with high or low broiler density. Methods ESC-resistant Enterobacteriaceae were identified by combination disc-diffusion test. ESBL/AmpC/carbapenemase genes were analysed using PCR and sequencing. For E. coli, phylogenetic groups and MLST were determined. Plasmids were characterized by transformation and PCR-based replicon typing. Subtyping of plasmids was done by plasmid multilocus sequence typing. Results 175 ESC-resistant Enterobacteriaceae were cultured from 165/1,033 individuals. The isolates were Escherichia coli(n=65), Citrobacter freundii (n=52), Enterobacter cloacae (n=38), Morganella morganii (n=5), Enterobacter aerogenes (n=4), Klebsiella pneumoniae (n=3), Hafnia alvei (n=2), Shigella spp. (n=2), Citrobacter amalonaticus (n=1), Escherichia hermannii (n=1), Kluyvera cryocrescens (n=1), and Pantoea agglomerans (n=1). The following ESBL genes were recovered in 55 isolates originating from 49 of 1,033 (4.7 %) persons: bla CTX-M-1 (n=17), bla CTX-M-15 (n=16), bla CTX-M-14 (n=9), bla CTX-M-2 (n=3), bla CTX-M-3 (n=2), bla CTX-M-24 (n=2), bla CTX-M-27 (n=1), bla CTX-M-32 (n=1), bla SHV-12 (n=2), bla SHV-65 (n=1) and bla TEM-52 (n=1). Plasmidic AmpC (pAmpC) genes were discovered in 6 out of 1,033 (0.6 %) persons. One person carried two different E. coli isolates, one with bla CTX-M-1 and the other with bla CMY-2 and therefore the prevalence of persons carrying Enterobacteriaceae harboring ESBL and/or pAmpC genes was 5.2 %. In eight E. coli isolates the AmpC phenotype was caused by mutations in the AmpC promoter region. No carbapenemase genes were identified. A large variety of E. coli genotypes was found, ST131 and ST10 being most common. Conclusions ESBL/pAmpC genes resembled those from patients in Dutch hospitals, indicating that healthy humans form a reservoir for transmission of these determinants to vulnerable people. The role of poultry in the transmission to humans in the community remains to be elucidated.


Introduction
Extended-spectrum-β-lactamase/AmpC producing Enterobacteriaceae have been found among humans worldwide. Most large-scale studies in humans, however, report data of patients or travelers and/or focus on ESBL-producing bacteria and/or certain bacterial species only (e.g. Escherichia coli or Klebsiella pneumoniae) [1][2][3][4]. Consequently, data on the prevalence of fecal carriage of ESBL/AmpC/carbapenemase producing Enterobacteriaceae in healthy humans in the community are scarce. The major mechanism of resistance to extended spectrum cephalosporins (ESC) in the family Enterobacteriaceae is the production of an extendedspectrum β-lactamase (ESBL) or an AmpC β-lactamase [5]. ESBLs are often plasmid mediated, while the production of AmpC β-lactamases can result either from (over)expression of the chromosomal ampC gene or by the acquisition of a plasmid-mediated ampC determinant [5]. Initially ESBL/AmpC-producing organisms were associated with hospitals and institutional care in humans, but they are now increasingly found in the community and in food-producing animals [6]. A connection between ESBL/AmpC-producing bacteria in food animals and humans has been suggested [1,[7][8][9]. ESBL/AmpC-producing Enterobacteriaceae have frequently been reported in broilers and therefore they have been considered as a reservoir for ESBL/ AmpC-encoding resistance genes [7,10]. Transmission from broilers to humans through the food chain has been proposed [11][12][13], but could also occur through direct contact or through the environment [7]. In 2011, a cross-sectional study was performed to determine the prevalence of, and identify risk factors for, carriage of ESBL-producing Enterobacteriaceae in people living in municipalities with either high or low broiler densities [14]. The prevalence of carriage of ESBL-producing bacteria was 5.1% and this percentage was lower in municipalities with high broiler densities (3.6%) compared to municipalities with low broiler densities (6.7%) [14]. The aim of the present study was to analyse the isolates from this cross-sectional study, including isolates with an AmpC phenotype, with respect to molecular characteristics and compare them to published data on isolates from patients, broilers, and persons in contact with broilers.

Materials and Methods
A cross-sectional study was conducted between August and December 2011. A random sample of adults (>18 years), stratified according to age and gender was taken from eight municipalities across 4 provinces of the Netherlands: North-Brabant, Gelderland, Overijssel and Frisia. In each province the municipality with the highest respectively lowest number of broiler farms per km 2 was selected. This information was obtained from the Dutch Product Board for Poultry and Eggs. In total, 3,949 individuals were contacted by post and were asked to return a rectal swab and a questionnaire on demographics, contact with animals, lifestyle, medical history, eating habits and travel. For each respondent, distance to the nearest broiler farm was obtained using geographic data. Exclusion criteria were living or working on a commercial broiler farm [14]. The study was approved by the Medical Ethics Committee of University Medical Centre Utrecht, The Netherlands (protocol number 11-277). All participants provided written informed consent. Rectal swabs were obtained from 1,033 persons and were analysed to determine the presence of ESBL/AmpC/carbapenemase-producing Enterobacteriaceae. Bacteria were isolated by selective enrichment (Luria-Bertani broth (MP Biomedicals, Amsterdam, the Netherlands) supplemented with 1mg/L cefotaxime (Sigma-Aldrich, Zwijndrecht, the Netherlands), and cultured on selective plates (MacConkey agar no. 3, Oxoid, Badhoevedorp, the Netherlands) supplemented with 1 mg/L cefotaxime). Isolates (1-6 per person, depending on the numbers of different phenotypes) were tested phenotypically for ESBL/AmpC-production by a combination disc-diffusion test using cefotaxime and ceftazidime discs, with and without clavulanic acid (Becton Dickinson B.V., Breda, the Netherlands), according to CLSI guidelines [15]. A cefoxitin disc (Becton Dickinson B.V., Breda, the Netherlands) was used to detect AmpC phenotypes [7]. Genotypes of the ESBL/AmpC-positive isolates were determined by PCR and gene sequencing. For isolates with an ESBL phenotype, primers detecting CTX-Mgroup 1, CTX-M-group 2, CTX-M-group 9, CTX-M-group 8/25, OXA-1 like, SHV and TEM were used. In case of isolates displaying an AmpC phenotype, primers specific for ACC, ACT, BIL, CMY, DHA, FOX, LAT, MIR and MOX were used. In addition, all 71 isolates with an ESBL-phenotype were investigated using primers for the detection of carbapenemase genes of the KPC, NDM, OXA-48, and VIM families. For E. coli isolates with an AmpC phenotype, but negative in PCR for β-lactamase genes, chromosomal ampC promoter mutations were detected by PCR and sequencing analysis (Table 1).
DNA was extracted by Chelex-100 chelating resin (Bio-Rad Laboraties B.V., Veenendaal, the Netherlands). Published primer sets were used to screen for the group of ESBL/AmpC gene [22]. Complete ESBL/AmpC gene sequences were obtained by PCR using the primers as indicated in Table 1. Resulting amplicons were treated with ExoSAP-IT (Isogen Life Science, De Meern, the Netherlands) according to manufacturers' instructions. Aliquots of the purified PCR products were used in sequence reactions on an AB 3730 genetic analyser using the Big Dye Terminator technology (Applied Biosystems, Bleiswijk, the Netherlands). Each sequence was compared with known β-lactamase gene sequences (www.lahey.org/Studies) by multiplesequence alignment using the BLAST, BioNumerics and Seaview programmes.
Phylogenetic groups were determined for E. coli according to Doumith et al. [23]. Strains were sub-grouped according to Escobar-Páramo et al. [24]. For isolates identified as non-E. coli the bacterial species was identified by BBL (Becton Dickinson B.V., Breda, the Netherlands) and MALDI TOF MS on a Bruker Microflex LT instrument (Bruker Daltonics GmbH, Bremen, Germany).

Results
Out of 1,033 persons investigated, 165 (15.9%) carried ESC-resistant Enterobacteriaceae. Ten persons were positive for two types of ESC-resistant Enterobacteriaceae, yielding a total of 175 isolates with an ESBL/AmpC resistance phenotype.

ESBL/AmpC phenotype and genes of all isolates
Of these 175 isolates, 119 (68.0%) showed an AmpC-phenotype and were recovered from 116 persons. For most isolates, however, no AmpC gene was found. If an AmpC gene was detected that is specific for the species concerned (e.g. bla CMY-2 in C. freundii, bla ACC in H. alvei, bla DHA in M. morganii, and bla ACT/MIR-1 in Enterobacter species) it was considered as chromosomal and these isolates were excluded from further analysis. Six isolates carried plasmidic AmpC (pAmpC) genes: 4 E. coli isolates, 1 P. agglomerans isolate and 1 C. freundii isolate. The prevalence of pAmpC-producing Enterobacteriaceae was 5.0% ( Table 2).

Analysis of risk factors
After analysis of the questionnaires we found no clear evidence that certain genes were more often found in specific exposure categories. However, bla CTX-M-1 genes were relatively more often found than other genes among persons owning or in contact with a horse compared to persons not frequently exposed to a horse (p = 0,04) ( Table 3).

Discussion
In the present study the prevalence of persons carrying Enterobacteriaceae harboring ESBL and/or pAmpC genes was 5.2%. Response analysis with respect to age, sex, and province and broiler density showed that a representative sample of Dutch adults was obtained [14]. Medical histories of 1,025/1,033 persons were available and 7.9% reported admission to a hospital and 6.4% urinary tract infection in the 6 month prior to inclusion in the present study and neither of these two factors was identified as a risk factor for being ESBL-positive [14]. Therefore, this study population represents a predominantly healthy general population. Most studies on ESBL/AmpC producing bacteria include either hospitalized patients or persons visiting a general practitioner. To date only limited data are available on the prevalence of pAmpC-producing bacteria in the open population. The prevalence of persons carrying pAmpC genes in the present study (0.6%) was slightly lower than the 1.3% found in a study in community-dwelling individuals in the densely populated region of Amsterdam [31]. The disparity might be caused by the different study population: participants of the current study were living in rural areas, but dissimilarities in the methodology might also be an explanation. The prevalence of ESBL and pAmpC carriers in patients and out-patients of a Spanish University teaching hospital was 5.0% and 0.6%, respectively, which is similar to our findings [32]. The pAmpC gene discovered most frequently was bla CMY-2 [31,32] and this corresponds with the findings of the present study. Altogether, the findings indicate that healthy humans form a reservoir for transmission of these determinants to vulnerable people.
Interestingly, only two persons carrying ESBL-positive K. pneumoniae were found (0.2% of all persons tested), although the prevalence of ESC-resistant K. pneumoniae is increasing in Dutch hospitals. The European Antimicrobial Resistance Surveillance System that collects resistance data from invasive isolates throughout Europe showed that third-generation cephalosporin resistance in The Netherlands has increased from 3.5% in 2005 to 7.5% in 2013 in K. pneumoniae [33]. This might be explained by the fact that ESBL-producing K. pneumoniae have different transmission dynamics compared to ESBL-producing E. coli, the predominant ESBL-positive species in the present study. A recent study showed a higher rate of community acquisition among ESBL-producing E. coli compared to ESBL-producing K. pneumoniae in patients with bacteremia [34]. In addition, ESBL-producing E. coli isolates had more different genotypes and patients infected with ESBL-producing E. coli were more likely to come from high prevalence countries compared to ESBL-producing K. pneumoniae supporting the notion that ESBL-producing E. coli is more likely to be acquired in community settings while K. pneumoniae is more often associated with hospital outbreaks and clonal transmission within the hospital [34]. One third of all ESC-resistant isolates in the present study carried an ESBL-gene and in all but one isolate with an ESBL-phenotype an ESBL-gene was found. bla CTX-M-1 was found most frequently and exclusively in E. coli, followed by bla CTX-M-15 and bla CTX-M- 14 . In a study investigating ESBL-producing Enterobacteriaceae in Dutch community patients with gastrointestinal complaints, the most prevalent ESBL gene was bla CTX-M-15 , comprising 47% of all ESBLgenes and 85% of the genes of the CTX-M-group 1 [3]. In another study, bla CTX-M-15 was also most prevalent (39%), followed by bla CTX-M-1 (15%), among clinical Enterobacteriaceae obtained from Dutch patients [35]. Another Dutch study, however, found bla CTX-M-1 most often in faecal samples from persons admitted to the hospital, whereas bla CTX-M-14 was predominant in isolates from blood cultures, followed by bla CTX-M-1 [1]. In a German study investigating ESBL-producing E. coli in persons from the general population that had been in contact with patients with gastroenteritis, the majority of isolates belonged to CTX-M-type ESBL, with bla CTX-M-15 (46%) and bla CTX-M-1 (24%) as the most common types [4]. Therefore, in clinical isolates and persons in contact with patients, bla CTX-M-15 seems to be more prevalent than bla CTX-M-1 although both genes were found equally often in predominantly healthy persons in the community. Further research into the reason for this difference is needed. bla CTX-M-1 was most often associated with plasmids of the IncI1 family, while bla CTX-M-15 was more often associated with plasmids of the IncF family, underlining the different transmission dynamics.
Only few studies analyzed possible risk factors for carriage of ESBL/pAmpC-producers. Our results should be interpreted with caution, because of the small number of cases. Still, bla CTX-M-1 genes were relatively more often found among persons owning or in frequent contact with a horse. Interestingly, in a previous study contact with horses was identified as a risk factor for being ESBL-positive [14].
In this study, E. coli ST131, ST10, ST58 and ST38 were found most often. This is in accordance with the findings of Reuland et al. [3]: the predominant E. coli STs in their study were ST38, ST131, ST648 and ST10. E. coli ST131 has emerged as a global epidemic, multidrug-resistant clade. E. coli ST131 may cause extraintestinal infections, especially of the urinary tract, and its ESBL production is most often due to the presence of bla CTX-M-15 [36]. In an international study investigating 240 ESBL-producing E. coli with ST131 from nine countries, 193 (80%) contained bla CTX-M-15 [36]. In a Dutch study investigating clinical isolates most of the ST131 E. coli contained bla CTX-M-15 , and presence of this gene was associated with higher levels of resistance [35]. In the present study only one person carrying E. coli ST131 containing bla CTX-M-15 was found indicating that this ST131 clade does not seem to be endemic in humans in the community in the Netherlands.
Most E. coli isolates belonged to phylogroups A1 and B1. These phylogroups are less often recovered from extraintestinal body sites. Isolates belonging to phylogroups B2 and D, however, were also found. Virulent strains causing extraintestinal infections belong mainly to groups B2 and D [24,37]. This indicates that humans in the community carry E. coli isolates that have the potential to cause disease.
The prevalence of ESBL carriage was higher in areas with low broiler densities than in areas with high broiler densities and therefore living in areas with high broiler density was not identified as a risk factor [14]. The prevalence of ESBL/pAmpC carriage among people on broiler farms (19.1%) was higher than in the present study and an increased risk of carriage was shown among individuals having a high degree of contact with live broilers [7]. The most prevalent ESBL/AmpC genes in isolates from humans on broiler farms as well as broilers were bla CMY-2 , bla CTX-M-1 and bla SHV-12 , followed by bla TEM-52 , while bla CTX-M-15 was not found [7,10]. In contrast, in the present study, bla CTX-M-1 , bla CTX-M-15 and bla CTX-M-14 were among the most prevalent ESBL-genes identified and E. coli isolates carrying bla CMY-2 , bla SHV-12 and bla-TEM-52 were only found sporadically although 94.5% of the study participants reported eating chicken meat [14]. It has been postulated that humans acquire ESBL-producing bacteria by eating chicken meat, because Dutch chicken meat has been shown to be contaminated with E. coli strains containing ESBL-genes similar to those found in patients [1,12]. The same genes are, however, present in many different potential reservoirs, including cattle, companion animals, horses and pigs, and therefore conclusions regarding their origin cannot be drawn [6,38].
Genes encoding for the production of carbapenemases were not detected, signifying that the prevalence of carbapenemase producing Enterobacteriaceae in the community is low.

Conclusions
ESBL/pAmpC genes found in healthy humans in the community are similar to those in Dutch patients indicating that humans in the community could be a reservoir for these resistant determinants. While contact with broilers has previously been identified as a risk factor, the role of poultry in transmission to humans through the environment or the food chain remains to be elucidated.