Diversity of Plasmids and Genes Encoding Resistance to Extended Spectrum Cephalosporins in Commensal Escherichia coli From Dutch Livestock in 2007–2017

Extended-spectrum β-lactamase (ESBL) and plasmid-mediated AmpC β-lactamase (pAmpC) genes confer resistance to extended spectrum cephalosporin’s. The spread of these genes is mostly facilitated by plasmid-mediated horizontal transfer. National surveillance activities to detect ESBL/pAmpC-producers in commensal bacteria from livestock are in place in the Netherlands since several years. This study aimed at reporting gene and plasmid diversity of commensal ESBL/pAmpC-producing Escherichia coli isolated from healthy animals during surveillance activities between 2007 and 2017. A collection of 2304 extended-spectrum cephalosporin-resistant (ESC-R) E. coli isolated from feces of broilers, dairy cattle, slaughter pigs, turkeys, ducks, and veal calves was investigated and ESBL/pAmpC genes were determined. Gene location of a selection of 473 E. coli isolates was determined and typing of plasmids linked to the ESBL/pAmpC genes was performed. Twenty-two different ESBL/pAmpC genes were identified with blaCTX-M-1 being the most prevalent gene in livestock (43.7%), followed by blaCMY -2 and blaSHV -12, independent of the animal source. Prevalence of typically human associated blaCTX-M-15 was highest in cattle. Less than 10% E. coli isolates owed their ESC-R phenotype to promoter mutations of the chromosomal ampC gene. Majority (92%) of ESBL/pAmpC genes analyzed were plasmid located, with IncI1α being the most represented plasmid family in isolates from all animals, followed by IncF (veal calves, dairy cattle and slaughter pigs), IncK (broilers and laying hens), IncX1 in broilers, and emerging IncX3 in broilers and dairy cattle. Prevalence and molecular diversity of ESC-R E. coli isolated from livestock over an 11-year period revealed a composite scenario of gene-plasmid combinations.


INTRODUCTION
Extended-spectrum β-lactamases (ESBLs) and plasmid-mediated AmpC β-lactamases (pAmpCs) are able to hydrolyse a large variety of β-lactam antibiotics, including cephalosporins and monobactams. The most clinically significant ESBL variants belong to the bla CTX−M , bla TEM , and bla SHV gene families together with pAmpC bla CMY gene family (Bush and Fisher, 2011). The successful spread of ESBL/pAmpC genes is mostly due to their localization on plasmids, resulting in easy transmission between bacteria (Rozwandowicz et al., 2018).
Extended-spectrum cephalosporin-resistant (ESC-R) Enterobacteriaceae have emerged globally in livestock animals during the last decades (Carattoli, 2008), with the consequent concern of animals being a putative source of ESBL/pAmpC-producing bacteria for humans either by direct contact or consumption of contaminated food products, as reviewed by (Ewers et al., 2012). Over the years, measures were implemented to reduce the use of third generation cephalosporins in livestock at national and European level (Speksnijder et al., 2015). Although the impact of transmission from livestock and the food chain on infections in humans is still debated (Madec et al., 2017;Dorado-Garcia et al., 2018), the ESBL/pAmpC reservoir in commensal bacteria from livestock has been increasingly investigated for its potential risk to public health (Michael et al., 2015).
Commensal ESC-R Escherichia coli randomly isolated from livestock feces have been monitored in the Netherlands since 1998, and phenotypic and genotypic results have annually been reported in the Monitoring of Antimicrobial Resistance and Antibiotic Usage in Animals in the Netherlands reports ( MARAN Reports). Since 2014, active monitoring through selective culturing and reporting of antimicrobial resistance in several bacteria, including ESC-R E. coli, has become mandatory for member states of the European Union (European Food Safety Authority, 2008. Results of these activities are yearly published (Maran Reports, 2002Reports, /2017 but lack detailed information on plasmid typing and epidemiology.
The aim of this study is to report gene and plasmid diversity observed in ESC-R E. coli isolated from healthy livestock from 2007 to 2017 during surveillance activities in the Netherlands.

Surveillance Activities
All ESC-R E. coli isolates included in this retrospective study originated from fecal samples of livestock collected during different surveillance activities in the Netherlands. Because surveillance activities have changed over the years in terms of sampling and methodologies, full details can be found in the yearly reports (Maran Reports, 2002Reports, /2017. Main differences between monitoring activities are briefly described here. Non-selective culturing (2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017) was performed by isolation of one randomly selected E. coli colony from a directly inoculated MacConkey agar plate without supplemented antibiotics, each isolate representing one epidemiological unit as prescribed by EFSA guidelines (European Food Safety Authority, 2008). Selective culturing (2014)(2015)(2016)(2017) was performed by overnight incubation of fecal samples in Buffered Peptone Water (BPW) followed by sub-culturing on MacConkey agar plate supplemented with 1 mg/L cefotaxime, according to EURL-AR protocols 1 . Sampling of ESC-R E. coli via selective isolation was performed on fecal samples from broilers, veal calves, slaughter pigs and dairy cows (European Food Safety Authority, 2013). Outside of mandatory surveillance activities, additional sampling was performed for turkeys in 2011 and 2012 (usually excluded because of low production), laying hens in 2014 and 2016 (typically screened only for Salmonella), and ducks in 2016 (not included in the legislation). Furthermore, ESC-R E. coli isolates obtained during monitoring activities from 2011 to 2013 by the Netherlands Food and Consumer Product Safety Authority (NVWA) with selective culturing (O/N enrichment in BPW followed by selective isolation on MacConkey agar plate with 1 mg/L cefotaxime) of fecal samples from broilers, dairy cattle, slaughter pigs and veal calves were included.

Gene and Plasmid Typing
Along the years, different methods to identify ESBL/pAmpC gene families in ESC-R E. coli have been employed, including miniaturized DNA Microarrays (Identibac AMR-ve, Alere Technologies GmbH) (Batchelor et al., 2008), microarray analysis using the Check-MDR CT-101 array platform (Check-Points, Wageningen, Netherlands) or dedicated PCRs (Geurts et al., 2017). DNA was extracted by using the DNeasy Blood and Tissue kit (QIAGEN, Hilden, Germany) according to the manufacturer's recommendations or DNA lysate preparation (Veldman et al., 2018). Independent of the screening method applied, gene sequences were confirmed by PCR amplification and DNA sequencing (Liakopoulos et al., 2016). Nucleotide and deduced amino acid sequences were compared with sequences in the Lahey clinic database 2 and GenBank. Chromosomal mutations of promoters and attenuators of ampC genes were determined by sequencing and compared to GenBank (Mulvey et al., 2005).
A subset of 473 ESC-R E. coli was selected for genomic localization of ESBL/pAmpC genes: 63 from non-selective surveillance and 410 isolates from selective surveillance ( Table 2). Over the years, different selection criteria were applied with the aim of including all ESBL/pAmpC genes detected in each animal species, and taking into consideration existing knowledge of gene-plasmid epidemiology. The chosen number of isolates per gene type was dependent on how prevalent the gene was in a given year i.e., for selective surveillance of broilers in 2014, 43% of samples were bla CTXM−1 positive (n = 116 out of 269), and a third of them (n = 42) were analyzed for genomic localization of bla CTXM−1 gene. For non-selective surveillance based on the knowledge that bla CTXM−1 positive E. coli in broilers are usually associated with IncI1 plasmids , majority of these isolates were not typed. Further details can be found in the corresponding results sections. ESC-R E. coli from 2011 to 2012 were not typed because of a temporary change in research priorities. Plasmid typing results for ESC-R E. coli isolated in 2017 via selective surveillance were not available at the time of writing.
Transformation experiments to assess plasmid location of ESBL/pAmpC genes and plasmid classification by PCR-based replicon typing (Carattoli et al., 2005) were performed according to standard procedures, as previously described (Liakopoulos et al., 2016). When transformants could not be retrieved, chromosomal location of ESBL/pAmpC genes was confirmed by I-CeuI Pulsed-Field Gel Electrophoresis (PFGE) of total bacterial  DNA, followed by Southern blot hybridization, as previously described (Liu et al., 1993).
Selective surveillance from 2014 onward is based on the use of a harmonized protocol (European Food Safety Authority, 2008), a comparable number of isolates (300-400, depending on the year of sampling) and data are available for a four year period (2014-2017). Therefore, trends in ESC-R E. coli prevalence could be defined (Figure 2). A significant reduction (p < 0.001) from 67.3% (95% CI, 62.4-71.8) to 32.6% (95% CI, 27.3-38.2) was observed in broilers between 2014 and 2017. This trend is in accordance with decreasing prevalence of ESC-R E. coli from non-selective surveillance and in fresh chicken meat (Veldman et al., 2018). Prevalence in dairy cattle and slaughter pigs showed non-significant (p = 0.028,

Genomic Localization of ESBL/pAmpC Genes in ESC-R E. coli From Non-selective Surveillance
Genomic location (plasmid or chromosome) of a subset of ESBL/pAmpC genes was determined in 63 ESC-R E. coli collected over the years with non-selective culturing ( Table 2). All ESBL/pAmpC genes were encoded on plasmids with different rep types (B/O, F, I1, K, and X1), in a few cases with multi-replicon plasmids (P/HI2 and P/I1), with distinctive prevalent gene-plasmid combinations per animal species (Supplementary Figure S2). Overall, the most common gene-plasmid combination was IncI1bla CTX−M−1 , detected in 30.1% of the isolates independently on the animal source. In broilers, IncX1-bla TEM−52c−Var and IncI1-bla SHV−12 were the most prevalent (36.8 and 39.5%, respectively) among the subset of analyzed isolates, excluding bla CTX−M−1 encoding E. coli that were not typed because typically associated with IncI1 plasmids . An E. coli isolate from broiler encoding genes bla CTX−M−1 , bla CMY−2 , and bla SHV−12 was associated to three IncI1, IncK, and IncX3 plasmids, respectively (Veldman et al., 2012). ESC-R Enterobacteriaceae encoding multiple ESBL/pAmpC genes have been described previously with various genomic settings both on plasmids and/or chromosome in livestock, meat, and clinical isolates (Dhanji et al., 2010;Veldman et al., 2010;Huang et al., 2017), depicting the complex plasmid scenario of cephalosporinresistance circulation among Enterobacteriaceae. Vast majority of ESC-R E. coli isolates from slaughter pigs were associated to IncI1 plasmids carrying bla CTX−M−1 (76.5%). The most common gene-plasmid combination in veal calves isolates was IncI1-bla CTX−M−1 , while ESC-R E. coli isolates from dairy cattle were associated with multireplicon plasmids IncP/HI2 encoding bla CTX−M−2 . All IncI1 plasmid subtyped (89%) were confirmed to be IncI1α (data not shown).

Genomic Localization of ESBL/pAmpC Genes in ESC-R E. coli From Selective Surveillance
According to current guidelines (European Food Safety Authority, 2013), selective surveillance of ESC-R E. coli should be performed following an annual rotation system: broilers and turkeys (years 2014, 2016, 2018, 2020), pigs and bovines (years 2015, 2017, 2019). Although more animal species than the recommended ones are frequently analyzed in the Dutch surveillance program (Maran Reports, 2002Reports, /2017), the rotation system was followed to select a subset of ESC-R E. coli (n = 410) for further investigation on the genomic localization of ESBL/pAmpC genes ( Table 2). Because poultry ESC-R isolates for 2014 and 2016 were too many to include in the analysis (n = 783), 40-50% of all E. coli from broilers (n = 99 and n = 60, respectively) and laying hens (n = 24 and n = 28, respectively) were screened per year.
Results of this analysis are reported in Figure 3, except for thirteen ESC-R E. coli from ducks (2016) characterized by chromosomal bla CMY−2 (n = 8) or IncI1 plasmids carrying bla CTX−M−1 (n = 5). Transformants for 38 (9.3%) ESC-R E. coli could not be recovered. PFGE and Southern hybridization confirmed the chromosomal location of ESBL/pAmpC genes, mostly belonging to the CTX-M group (Figure 3): bla CTX−M−1 , bla CTX−M−14 , bla CTX−M−15 , bla CTX−M−32 , bla CTX−M−55 , bla CTX−M−9 , and bla CMY−2 . Although, the genetic surroundings of these genes were not investigated, it is known that ISEcp1 insertion sequence upstream of ESBL/pAmpC genes are associated with transposition and chromosomal integration of typically plasmid-encoded genes in E. coli, K. pneumoniae, and Shigella flexneri, among others, from animals or humans (Wang et al., 2013;Fang et al., 2015;Huang et al., 2017). Through chromosomal integration, ISEcp1 might contribute to lowering the fitness cost derived from harboring an entire plasmid,  while enhancing EBSL/AmpC gene expression under its own promoter (Poirel et al., 2003). Majority of ESBL/pAmpC genes (n = 372) were associated with plasmids (Figure 3). All IncI1 plasmid subtyped (86%) were confirmed to be IncI1α (data not shown). Gene-plasmid combinations in broilers did not show major differences between 2014 and 2016. IncI1-bla CTX−M−1 plasmids were the most common, followed by IncI1 encoding bla CMY−2 , bla SHV−12 or bla TEM−52c . IncK-bla CMY−2 (or multireplicon IncK/P) plasmids were also commonly detected, suggesting a relatively stable plasmid population in broilers in the Netherlands, as earlier described ). Yet, IncX3-bla SHV−12 plasmids, whose emergence in Dutch ESC-R E. coli of animal origin was recently revealed alongside a gradual decrease in the prevalence of IncI1-bla SHV−12 plasmids (Liakopoulos et al., 2018), were detected in both years. Plasmid IncX1 carrying bla SHV−12, bla TEM−52c or bla TEM−52c−Var followed in prevalence, the latter detected also in ESBL/pAmpC-producing E. coli from laying hens in 2016. Overall, plasmid-gene associations in isolates from laying hens were comparable to broilers, with IncI1-bla CMY−2, IncI1bla CTX−M−1 and IncK-bla CMY−2 being the most predominant in both 2014 and 2016. The presence of ESC-R E. coli at all levels of the Dutch broiler production pyramid has been demonstrated, as day-old chicks can inherit bacteria from their parents through contaminated egg shells or from the environment .
ESBL/pAmpC-producing E. coli isolated from slaughter pigs from both 2011 and 2015 were dominated by IncI1 plasmids encoding bla CTX−M−1 or bla TEM−52c , recognized as the most prevalent gene-plasmid combinations in Enterobacteriaceae from slaughter pigs worldwide (Geser et al., 2011;Randall et al., 2014;Biasino et al., 2018;Dang et al., 2018). ESC-R E. coli isolates from dairy cattle and veal calves showed a quite variable array of plasmid-gene combinations (Figure 3). Beside predominant IncI1 plasmids, IncF plasmids were detected in both animal reservoirs in association with bla CTX−M−1 and bla CTX−M−14 and bla CTX−M−15 genes. IncR-bla CTX−M−65 and IncR-bla CTX−M−55 were also identified in veal calves in 2015 but no R plasmid was detected in 2011. Geneplasmid combinations observed in veal calves are coherent with previous studies conducted in the Netherlands and in France (Hordijk et al., 2013a;Haenni et al., 2014) with relatively high prevalence of various bla CTX−M genes located on IncF and IncI1 plasmids. The more variable array of plasmid-gene combinations observed in veal calves compared to other livestock might be a consequence of international trade from different dairy farms to Dutch farms as well as high antimicrobial use and farm management.
In conclusion, the results of this study provide insight in the prevalence and molecular diversity of ESC-R E. coli, revealing a rather composite scenario of plasmidgene combinations circulating in livestock from the Netherlands over the last decade. Yet, the bias in the selection of isolates for plasmid typing should be kept in mind to avoid risky conclusions on prevalence of ESBLharboring plasmid types. Nevertheless, the study provides additional information on the occurrence of different plasmid types carrying ESBL/pAmpC-genes in E. coli from livestock in the Netherlands. These findings also demonstrate the added value of selective culturing of ESC-R E. coli and genotyping of genes and plasmids over random isolation for resistance determinants of public health concern.
AUTHOR CONTRIBUTIONS AK, AE-Z, JH, and BW acquired the data. CD, DC, KV and DM analyzed the data. DC and KV prepared the manuscript. All authors discussed, read, contributed to, and approved the final manuscript.

FUNDING
The Dutch Ministry of Economic Affairs funded surveillance activities over the years (WOT-01-002-003.02). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

ACKNOWLEDGMENTS
The authors would like to thank: Joop Testerink and Marga Japing (WBVR) for invaluable technical assistance; Michael Brouwer (WBVR) for insightful discussion on plasmid subtyping; Michel Rapallini and everyone at the Netherlands Food and Consumer Product Safety Authority involved in sampling activities.