Cross-Sectional Study on Prevalence and Molecular Characteristics of Plasmid Mediated ESBL/AmpC-Producing Escherichia coli Isolated from Veal Calves at Slaughter

Objectives The presence of ESBL/AmpC-producing E. coli in cattle has been reported previously, however information on veal calves is limited. This study describes the prevalence and molecular characteristics of E. coli with non-wild type susceptibility to cefotaxime in veal calves at slaughter. Methods Faecal samples from 100 herds, 10 individual animals per herd, were screened for E. coli with non-wild type susceptibility for cefotaxime. Molecular characterization of ESBL/AmpC genes and plasmids was performed on one isolate per herd by microarray, PCR and sequence analysis. Results 66% of the herds were positive for E. coli with non-wild type susceptibility for cefotaxime. Within-herd prevalence varied from zero to 90%. 83% of E. coli producing ESBL/AmpC carried bla CTX-M genes, of which bla CTX-M-1, bla CTX-M-14 and bla CTX-M-15 were most prevalent. The dominant plasmids were IncI1 and IncF-type plasmids. Conclusions A relatively high prevalence of various bla CTX-M producing E. coli was found in veal calves at slaughter. The genes were mainly located on IncI1 and IncF plasmids.


Introduction
The development of resistance to extended spectrum cephalosporins (ESC) has evolved rapidly world-wide in both clinical settings as well as in the community. Resistance to ESC is mainly caused by various extended spectrum ß-lactamases (ESBL) or AmpC ß-lactamases (AmpC) that are often found in Enterobacteriaceae and hydrolyse the ß-lactam ring of these antibiotics [1]. These enzymes are encoded by genes that are frequently located on mobile genetic elements (plasmids) [2], which have the ability to transfer horizontally within and between different bacterial species. Novel ESBL/AmpC genes or gene variants are reported on a regular basis (http://www.lahey.org/studies) and are classified based on either their functional characteristics [3] or primary structure [4].
Food-producing animals have been suggested as the primary reservoir of zoonotic foodborne pathogens, including antimicrobial resistant bacteria [5]. Faecal carriage of ESBL/AmpCproducing E. coli in cattle has been reported previously [6,7,8,9,10]. In addition, a recent study showed that in the period from 2005 to 2010, resistance to ESC in veal calves was mainly caused by plasmid mediated beta-lactamases, in contrast to the years before 2005 (Hordijk et al, submitted for publication). Furthermore, beef, chicken and pig meat products have also been found to be positive for ESBL/AmpC-producing bacteria [11], which may constitute a transmission route to humans. However, studies are difficult to compare, because isolation methods and sample sizes vary greatly. When cattle is studied, the type of cattle included is not always specified. For instance dairy farming is very different from veal calves by means of housing, life expectation and exposure to antibiotics. The majority of veal farms maintain an all-in, all-out system, while dairy farms generally maintain a closed production system. Almost each individual animal on a veal calf farm originates from another dairy farm. Furthermore, compared to dairy cattle, veal calves are more frequently exposed to antimicrobials [12]. These factors may influence the prevalence and genetic characteristics of antimicrobial resistance genes in isolates from these animals. Furthermore, antimicrobial susceptibility data from national surveillance studies have shown that isolates from veal calves are often multi resistant [12]. Therefore, the aim of this study was to determine the prevalence of E. coli with reduced susceptibility to ESC in veal calf herds. Moreover the ESBL/AmpC genes and plasmids on which they are located were identified to determine their molecular characteristics in relation to those found in other food-producing animals and humans.

Sampling design and isolation of Escherichia coli
From January to December 2011, fecal samples from 10 individual veal calves were taken from 100 slaughter batches (1000 samples in total). The samples were taken as part of the Dutch national control program on antimicrobial resistance by the Netherlands Food and Consumer Product Safety Authority (NVWA). All slaughter batches originated from different herds. To ensure anonymity, the origin of samples was aggregated at province level. The sampled slaughter batches were equally distributed over the five slaughter houses in the Netherlands and the sampling took place equally distributed over the year. All fecal samples were individually screened for ESC reduced susceptible E. coli using selective enrichment broth. Fecal swabs were taken from the colon of the carcass after evisceration and subsequently transferred to the laboratory and placed in a tube with 1 ml Luria-Bertani broth (Beckton Dickinson) supplemented with 1 mg/L cefotaxime (Sigma-Aldrich, Germany) (LB+). LB+ tubes were incubated aerobically at 37uC for 2062 hours. Subsequently, the overnight LB+ culture was inoculated on a MacConkey agar plate (Becton Dickinson), which was supplemented with 1 mg/L cefotaxime (MC+). All MC+ plates were incubated aerobically at 37uC for 2062 hours. From each MC+ plate showing growth, one colony typical for E. coli was selected, subcultured on blood agar, and suspended the next day in 1 ml of buffered peptone water supplemented with 30% glycerol and stored at 280uC, pending analysis. All selected E. coli were inoculated in 5 ml Gersbach medium and incubated at 37uC for 2062 hours for E. coli confirmation. One droplet of Kovac's Indole reagent (Merck, Germany) was added to the Gersbach culture to check for tryptophan reduction following the requirements for E. coli identification stated by the European Food Safety Authority (EFSA) [13]. All indole-positive isolates were considered E. coli. From all isolates showing growth on MacConkey, one randomly picked isolate was selected per herd for antimicrobial susceptibility testing and molecular characterization.

Antimicrobial susceptibility testing
All selected isolates (n = 66) were tested for susceptibility to antimicrobials by broth micro-dilution according to ISO standard 20776-1:2006 using microtitre trays with a custom made dehydrated panel of antibiotics (Sensititre, Trek Diagnostic Systems, Basingstoke, UK). The following antibiotics were included: ampicillin, cefotaxime, ceftazidime, tetracycline, sulfamethoxazole, trimethoprim, ciprofloxacin, nalidixic acid, chloramphenicol, florfenicol, gentamicin, kanamycin, streptomycin and colistin. All results were interpreted using cut-off values defined for epidemiological purposes as recommended by the European Committee on Antimicrobial Susceptibility Testing (EUCAST; http://mic.eucast.org/Eucast2/), apart from sulfamethoxazole, for which the clinical break point defined by the Clinical and Laboratory Standards Institute (CLSI) was used [14]. Values above these cut-offs and break points were interpreted as resistant (sulfamethoxazole) or non-wild type (all other antimicrobials) [15].

Characterization of ESBL/AmpC genes and plasmids
All selected isolates were screened for ESBL/AmpC genes using the tube based amr-ve-05 microarray (Alere, Tilburg, the Netherlands) [16]. Beta-lactamase gene families identified by this microarray were subsequently characterized by PCR and sequence analysis as described previously [17]. For sequence analysis the following additional primers were used: TEM-Fseq: 59-GCCAACTTACTTCTGACAACG, CMY-F-838: 59-TGGCGTATTGGCGATATGTA and CMY-R-857: 59-TACA-TATCGCCAATACGCCA. Primers and conditions used to identify and sequence the modified aminoglycoside resistance gene aac(69)Ib-cr have been described previously [18]. Plasmids were isolated using a modified miniprep method as follows. One colony of interest was inoculated in 3 ml LB broth and incubated overnight. Subsequently 1.5 ml of the culture was transferred and spinned down at 14000 rpm for 5 minutes. The pellet was suspended in 60 ml TEG buffer (25 mM Tris/Hcl; 10 mM EDTA; 50 mM glucose). Then, 120 ml NaOH/SDS (0.2 M/1%) was added and the whole sample was placed on ice for 5 minutes. Subsequently, 90 ml NaOH (3 M) was added and the sample was placed on ice for 5 minutes. The suspension was spinned down at 14000 rpm for 5 minutes. The supernatant was transferred and 270 ml LiCl (5 M) was added. After 10 minutes, the suspension was spinned down at 14000 rpm for 5 minutes. The supernatant was transferred and 1 ml of EtOH (96%) was added. The suspension was spinned down at 14000 rpm for 10 minutes and the pellet was resuspended in 200 ml EtOH (70%). The suspension was spinned down at 13000 rpm for 10 minutes and the pellet was resuspended in 20 ml H 2 O. Plasmids were transformed into Electromax DH10B cells by electroporation (Invitrogen, USA) by mixing 2 ml of plasmid DNA suspension with 20 ml competent cells. The cells were electroporated under the following conditions: 1.25 kV, 200 ohm, 25 mFar. Transformants were subsequently plated on LB agar plates supplemented with 1 mg/L cefotaxime in order to selectively isolate transformants with ESBLs carrying plasmids. PCR-based replicon typing (PBRT) was conducted on the transformants to identify the replicon type of the plasmid inside the transformant [19]. Plasmid MLST (pMLST) [20] and replicon sequence typing (RST) were used to further characterize IncI1 and IncF plasmids, respectively [21]. Plasmids that were negative in the PBRT analysis were designated ''not typable'' (nt). The size of these plasmids was determined by PFGE with S1nuclease digestion as described previously [22]. Because this study focused on plasmid mediated genes, another ESBL/AmpC suspected isolate from the same herd was included as a replacement if the micro array results were negative. For isolates in which no plasmid mediated ESBL/AmpC gene was detected, the promoter region of the chromosomal ampC gene was sequenced using primers that have been described previously [23]. Specific ampC-types were designated as described by Mulvey et al [24]. In addition a combination disk test was performed as described previously on all isolates that were negative in the array [17].

Ethics Statement
Samples were taken from the colon of the veal calves after slaughter as part of the Dutch national control program on antimicrobial resistance as required by the zoonosis directive of the European Parliament (2003/99/EC). Since the sampling of calves was not performed on live animals, no approval from the ethical committee was required. In line with 'Regulation (EC) No 854/ 2004 of the European Parliament and of the Council, article 4 section 8', concerning specific rules for the organisation of official controls on products of animal origin intended for human consumption, the NVWA is the designated authority in the Netherlands to take samples for surveillance purposes. The NVWA granted permission to analyse the samples as described in this paper.

Results
Prevalence of non-wild type cefotaxime susceptible E. coli From the 100 herds that were screened for the presence of E. coli with non-wild type susceptibility at slaughter, 66% were found positive for E. coli with non-wild type susceptibility to cefotaxime. The within-herd prevalence of E. coli with non-wild type susceptibility varied greatly, ranging from zero to 90%. However, in the majority of the positive farms, less than five out of ten fecal samples harbored E. coli with non-wild type cefotaxime susceptibility. From all positive farms 90% was located in the eastern and southern provinces. This is also where veal farming is concentrated. In addition, 86% of the ESBL/AmpC-producing E. coli was multi-drug resistant, showing non-wild type susceptibility to three or more antimicrobial classes, including ß-lactams (Tables 1 and  2).
The predominant plasmid types among all isolates harbouring ESBL/AmpC genes were IncI1 and IncF. Both IncI1 and IncF plasmids were present in 26% of the isolates depicted in Tables 1 and 2. The highest diversity of plasmids was observed among bla CTX-M-1 positive isolates, which was also the most abundant ESBL variant. In 9 isolates, bla CTX-M-1 was located on IncI1 with six different IncI1 pMLST types (Table 1). From these six different pMLST types, four types had not yet been reported in the pMLST database (http://www.pubmlst.org/plasmid, last accessed: 20 March 2013). Furthermore, in 7 isolates bla CTX-M-1 was located on IncF plasmids with four different IncF RST profiles. From these four different profiles, two had not yet been reported in the pMLST database. The remaining bla CTX-M-1 genes were located on IncB/O (three isolates), IncN (three isolates) and IncK (one isolate). In two isolates the plasmid was not typable. The second largest group of ESBL genes, bla CTX-M-15 , was predominantly carried on IncF plasmids, of which four different RST types were found among seven isolates (Table 1). From these seven isolates, five also harboured bla OXA-1 and aac(69)Ib-cr. The remaining bla CTX-M-15 genes were located on IncI1 (two isolates), IncHI2 (one isolate) and two on not typable plasmids. The bla CTX-M-14 genes were mainly carried on IncK plasmids ( Table 2), but also on IncI1, of which all three plasmids had the same pMLST sequence type (ST80), and one on IncF (F2; A-; B-)). From the isolates harbouring bla CTX-M-2/97 or bla CTX-M-3 , all genes were located on different plasmids (Table 1+2). All plasmids carrying bla CTX-M-32 , were not typable. Furthermore, four isolates harboured bla TEM-52c , of which three were carried on IncI1 plasmids with two different pMLST types (ST10 and ST36), both belonging to the same clonal complex (CC5) ( Table 2). From the fourth bla TEM-52c , which was found in an isolate also harbouring bla CTX-M-1 , transformation of the plasmid failed, so no replicon type was established. In the single isolate harbouring an ampC gene, bla CMY-2 was located on an IncK plasmid (Table 2). Finally, one transformant harbouring bla CTX-M-2 showed two replicon types, namely IncP and IncHI2 (Table 2).

Discussion
This study showed that 66% of the slaughter batches of veal calves were positive for fecal carriage E. coli with a non-wild type susceptibility to cefotaxime. In 68% of the positive herds, five or less animals out of ten were found carrier of these isolates. Since only one isolate per herd was included in the molecular analysis, no conclusions can be drawn towards clonal diversity of ESBL/ AmpC genes and plasmids in herds with multiple positive animals. Further in depth studies should be performed to assess whether a clonal spread within herds or a high diversity of ESBL/AmpC genes and/or plasmids exists within these herds. Factors that may influence the prevalence and spread of these resistance determinants within herds, including the use of antimicrobials and farm management, should be taken into account to analyze possible differences between farms.
Studies have been performed on fecal carriage of ESBL/AmpCproducing E. coli in cattle at slaughter in several countries, and the prevalence varied greatly. In Poland no ESBL/AmpC-producing E. coli were observed [25], in contrast to France (5.8%) [10], Switzerland (16%) [26], Hong Kong (3.1%) [27] and Japan (31.3%) [28]. However, while all studies were performed on individual animals, comparing these prevalence data should be performed with care since not all studies used the same selection methods (e.g. the use of enrichment broth and/or different selective media), which may have resulted in different screening sensitivities.
The vast majority of plasmid mediated ESBL/AmpC genes in the present study belonged to the bla CTX-M gene family, of which bla CTX-M-1 was most abundant, followed by bla CTX-M-15 and bla CTX-M-14 (Table 1+2). These findings are in line with the observed trend in the last five years of a retrospective study performed on faecal samples from veal calves collected at farms from 1997 to 2010 [29]. Studies from other countries, that focused on ESBLs in general or bla CTX-M genes in specific, confirm the relatively high abundance of bla CTX-M-1 , bla CTX-M-14 and/or bla CTX-M-15 genes in either healthy or sick cattle [6,8,10,30]. In contrast, in the United States, bla CMY-2 is the predominant gene found in cattle and bla TEM ESBL genes were found at a low level [31,32]. Furthermore, a study from Hong Kong reported that bla CTX-M-13 was the only ESBL observed in cattle at slaughter [27] and bla CTX-M-2 was reported to be predominant in cattle in Japan [33]. This indicates that there are geographical differences in the prevalence of resistance genes.
This study showed that ESBL/AmpC genes found in veal calves were predominantly located on IncI1 and IncF plasmids. These plasmids belong to the most commonly reported plasmid families in Enterobacteriaceae [34]. Interestingly, a relatively high proportion of IncI1 and IncF plasmid subtypes (mainly associated with bla CTX-M-1 ) had not yet reported in the pMLST database (Table 1). IncI1 sequence types (ST) in combination with bla CTX-M-1 have been reported either commonly (CC3, ST7) or occasionally (ST58) [34,35,36]. To our knowledge, the combination of bla CTX-M-1 with IncF type plasmids has not been reported previously. The IncF RST-type found in this study (F2;A-;B-) has frequently been associated with bla CTX-M-15 [34,37]. Furthermore, we have also  found bla CTX-M-1 in combination with IncB/O plasmids. This has also been reported in isolates from humans [38] and horses [11].
The bla CTX-M-1 associated with IncN was previously observed in human isolates [38,39], poultry [40], and in both porcine isolates and their farm workers, indicating transmission may occur between food-producing animals and their care takers [41].
In this study, bla CTX-M-15 was mainly located on IncF plasmids, a combination that has been reported frequently [34,39,42,43]. In addition, the combination of bla CTX-M-15 with bla OXA-1 and aac69-Ib-cr (the latter causing reduced susceptibility to aminoglycosides and fluoroquinolones), has been reported previously [44]. This combination was observed in the predominant IncF type found in this study carrying bla CTX-M-15 (F31;A4;B1). Furthermore, both the IncF RST types F2;A-B-and F31;A4;B1 and the IncI1 pMLST type ST31 that carry bla CTX-M-15 that were reported in this study have also been isolated from cattle in France [37]. The bla CTX-M-14 genes in our study were mainly located on IncK plasmids. This has also been reported in calves in the UK [45], as well as in isolates from humans and turkey [46,47]. None of the plasmids harbouring bla CTX-M-32 were typable by PBRT, making it difficult to draw any conclusions towards genetic characteristics. For these isolates, we determined the size of the ESBL harbouring plasmid. The combination of bla TEM-52 with IncI1 (ST36) has been reported in cattle in France [48], but both ST36 and ST10 in combination with bla TEM-52 have also been reported in humans and poultry isolates [38] and are the most commonly found types [34]. This shows that various combinations of ESBL/AmpC genes and plasmids are widely distributed in isolates of both animal and human origin. The IncHI2/P multi replicon plasmid harboring bla CTX-M-2 was not confirmed as a multi replicon by Southern blot hybridization. However, similar IncHI2/P plasmids harboring bla CTX-M-2 were reported previously and have been confirmed as multi replicons [49]. Additional experiments such as PFGE, RFLP or sequence analysis of the whole plasmid are required to determine whether the plasmids with similar replicon types, pMLST-type or RST-type are clonally related. Based on the data presented in this study we cannot determine whether the overlapping genes and plasmids reside in E. coli that are host specific (animal or human) or that they proliferate well in both human and animal hosts.
sulfamethoxazole, Tmp = trimethoprim, Str = streptomycin, Kan = kanamycin, Gen = gentamicin), Col = colistin. 4 qnrS and aac(69)Ib-cr are not ß-lactamase genes, but cause reduced susceptibility to quinolones. aac(69)Ib-cr also causes reduced susceptibility to aminoglycosides. 5 Plasmid size in kilo bases (kb) of non-typeble plasmids. doi:10.1371/journal.pone.0065681.t001 Table 2. E. coli carrying ESBL/AmpC genes other than bla CTX-M group 1 with their corresponding plasmids and resistance profile. In six herds, only promoter mutations of the chromosomal ampC gene were detected (Table 2) and no isolates harbouring plasmid mediated ESBL/AmpC genes were found within the same herd. Since this study focused on the dissemination of plasmid mediated ESBL/AmpC genes, this prevalence is likely to be an underestimate of the actual number of herds that are positive for chromosomal ampC genes with promoter variations that lead to non-wild type susceptibility to cefotaxime. Furthermore, we based the absence of plasmid mediated ESBL/AmpC genes on array data. Novel genes or ESBL/AmpC variants not present on the array will have been missed. The phenotypic combination disk test was performed on all isolates negative in the array in order to reduce the chance of missing novel genes.
The fact that international trafficking of live calves from many different dairy farms to Dutch veal calf farms happens on a large scale may contribute to the high dissemination observed in this study. This may also explain the high diversity in gene/plasmid combinations compared to Dutch poultry, in which bla CTX-M-1 is highly prevalent and commonly associated with IncI1 [36].

Conclusion
A relatively high percentage of slaughter batches (66%) were found positive for E. coli with a non-wild type susceptibility to cefotaxime. The within-herd prevalence varied greatly from zero to 90% positive. In the majority of herds positive for ESBL/ AmpC-producing E. coli, the within-farm prevalence was below 50%. Furthermore, plasmid mediated resistance to cefotaxime was predominantly caused by enzymes encoded by the bla CTX-M gene family. Many gene-plasmid combinations found in this study have also been found in cattle in other countries, indicating that there is a non-local dissemination of resistance determinants. However, since the number of ESBL/AmpC-producing E. coli was not quantified, no conclusions can be drawn about the actual risk of presence.