Diversity of Multi-Drug Resistant Avian Pathogenic Escherichia coli (APEC) Causing Outbreaks of Colibacillosis in Broilers during 2012 in Spain

Avian pathogenic Escherichia coli (APEC) are the major cause of colibacillosis in poultry production. In this study, a total of 22 E. coli isolated from colibacillosis field cases and 10 avian faecal E. coli (AFEC) were analysed. All strains were characterised phenotypically by susceptibility testing and molecular typing methods such as pulsed-field gel electrophoresis (PFGE) and multi-locus sequence typing (MLST). The presence of 29 virulence genes associated to APEC and human extraintestinal pathogenic E. coli (ExPEC) was also evaluated. For cephalosporin resistant isolates, cephalosporin resistance genes, plasmid location and replicon typing was assessed. Avian isolates belonged to 26 O:H serotypes and 24 sequence types. Out of 22 APEC isolates, 91% contained the virulence genes predictors of APEC; iutA, hlyF, iss, iroN and ompT. Of all strains, 34% were considered ExPEC. PFGE analysis demonstrated a high degree of genetic polymorphism. All strains were multi-resistant, including those isolated from healthy animals. Eleven strains were resistant to cephalosporins; six contained bla CTX-M-14, two bla SHV-12, two bla CMY-2 and one bla SHV-2. Two strains harboured qnrA, and two qnrA together with aac(6’)-Ib-cr. Additionally, the emergent clone O25b:H4-B2-ST131 was isolated from a healthy animal which harboured bla CMY-2 and qnrS genes. Cephalosporin resistant genes were mainly associated to the presence of IncK replicons. This study demonstrates a very diverse population of multi-drug resistant E. coli containing a high number of virulent genes. The E. coli population among broilers is a reservoir of resistance and virulence-associated genes that could be transmitted into the community through the food chain. More epidemiological studies are necessary to identify clonal groups and resistance mechanisms with potential relevance to public health.


Introduction
Escherichia coli is a bacterium widespread in the intestine of animals and humans, and a pathogen that can induce enteric and extraintestinal infections. In particular, avian pathogenic E. coli (APEC) is the main cause of colibacillosis in poultry farms; a syndrome associated to airsacculitis, perihepatitis, pericarditis, and sometimes fatal septicemia. APEC strains are responsible for the mortality of 3-4% of the animals in a farm, and for the reduction of 2-3% of egg production [1], resulting in an economic burden to the poultry industry [2]. In many cases, the fundamental cause of the disease remains unclear, since the infection with E. coli is associated to the presence of Mycoplasma gallisepticum or respiratory viruses, such as Newcastle virus or Infectious Bronchitis virus [3].
Several virulence genes are implicated in avian colibacillosis such as adhesins, toxins, antihost defence factors, iron acquisition systems, autotransporters and the IbeA protein [4]. Subtractive hybridization studies have demonstrated sequence homology between specific DNA regions of APEC and human extraintestinal pathogenic E. coli (ExPEC) [5]. Additionally, the presence of similar virulence genes found in both, APEC and ExPEC strains, suggested that APEC strains may act as zoonotic pathogens and reservoir of virulence causing human infections [6][7][8]. According to Johnson et al. (2003), a strain could be considered ExPEC if exhibits two or more of the following virulence genes; pap (P fimbriae), sfa/foc (S/F1C fimbriae), afa/ dra (Dr binding adhesins), iutA (aerobactin receptor), and kpsM II (group 2 capsule synthesis) [9]. ExPEC strains are more often derived from virulence-associated B2 and D phylogroups [10].
The successful treatment of avian colibacillosis caused by APEC strains mainly depends on the use of antimicrobials. However, increasing resistance to critically important antimicrobials, such as third-generation cephalosporins and fluoroquinolones, is nowadays common in E. coli from poultry origin [11]. These resistances can be transmitted to humans via the food supply [12,13]. In particular, E. coli producing extended-spectrum beta-lactamases (ESBLs) and plasmid mediated AmpC beta-lactamases have increased considerably in the last years [14]. Normally, these genes are located on plasmids, and can be transferred by conjugation to other bacterial species [11]. Some of the virulence factors for APEC and ExPEC can also be harboured on plasmids. Particularly, ColV plasmids yield some virulence genes such as hlyF, ompT, iss and cvaC surrounding the replicon RepFIB [15].
Several studies have described APEC strains in the literature [16]. However, not many studies have combined extensive characterization at the serotype level, virulence-associated genes, molecular typing techniques, molecular determination of resistance mechanisms and mobile genetic elements involved in transfer of resistance. For this reason, the objective of this study was to discriminate and to perform such characterization of highly pathogenic E. coli causing outbreaks of colibacillosis in 13 different broiler farms throughout Spain, and compare them to avian faecal E. coli (AFEC) obtained from healthy animals. Additionally, the identification of clones more prone to cause disease has been assessed.

Isolation
A total of 22 tissue swabs of culled-animals affected with colibacillosis arrived to the laboratory between January and March 2012. The samples were obtained as part of routine care. Samples were taken from chickens already sacrificed for diagnostic purposes following the procedures according to the requirements of the Ethics Committee of Animal and Human Experimentation of the Universitat Autònoma de Barcelona (Permit Number DMAH-4239 that specifically permits euthanasia of chickens). Animals were euthanized using intravenous sodium pentobarbital (100 mg/kg, Dolethal, Vétoquinol, Cedex, France) in the wing vein. The method to sacrifice the animals follows the welfare rules stated in the European Directive 86/609/CEE. None of the authors of this manuscript were involved in manipulating or sacrificing the chickens. All the samples were collected from clinical cases submitted to the Diagnostic Service of the Veterinary School of the Universitat Autònoma de Barcelona. Swabs were taken from 13 broiler farms located in nine different regions of Spain (Fig 1). Ten E. coli strains isolated from faeces of healthy animals collected in nine farms were also included in the study. The samples were plated onto MacConkey agar and incubated overnight at 37°C. Three lactose-positive colonies for each plate were selected and confirmed to be E. coli by PCR [17]. Subsequently, one representative was selected for further studies.

Serotyping
Determination of O and H antigens was carried out using the method previously described by Guinée et al. with all available O (O1 to O181) and H (H1 to H56) antisera [18]. Non-typeable isolates were denoted as ONT or HNT and non-motile isolates were denoted as HNM. All antisera were obtained and absorbed with the corresponding cross-reacting antigens to remove the nonspecific agglutinins. The O and H antisera were produced in the Laboratorio de Referencia de E. coli (LREC, Lugo, Spain). O25a and O25b subtypes were determined by PCR [19]. Phylogeny, pulsed-field gel electrophoresis (PFGE) and multilocus sequence typing (MLST) Isolates were separated in phylogroups (A, B1, B2, C, D, E or F) according to a method previously described [20,21]. PFGE was performed as described elsewhere [22]. The results were analysed by Fingerprinting II Informatix software (Applied Maths, Sint-Martens-Latem, Belgium). PFGE-types were separated based on differences of at least one band in the restriction profiles. The analysis of the bands generated was carried out using the Dice coefficient and unweighted pair group method with arithmetic averages (optimization of 1.25% and position tolerance 1.25%).

Detection of virulence-associated genes
All strains were tested by PCR for 29 ExPEC and APEC virulence-associated genes (Tables 1 and 2) [13,24]. The genes described previously by Johnson et al. (2008) as the minimal predictors of APEC virulence; iroN, ompT, hlyF, iutA and iss were detected by a multiplex PCR [24]. Virulence scores were calculated for each isolate as the sum of all virulence-associated genes detected; pap, sfa-foc and kpsM II were counted only once.

Statistical analysis
Differences in the prevalence between different groups were determined by Fisher's exact test as described before [25]. Virulence scores were compared by the use of Mann-Whitney U test. Statistical analyses were performed using GraphPad Prsim, version 3.1 software (GraphPad Software, Inc., San Diego, CA).

Resistance genes
All strains exhibiting resistance to third-generation cephalosporins (cefotaxime and ceftazidime) were tested by PCR methods for the presence of the bla CTX-M , bla SHV , bla TEM , bla CMY-1 and bla CMY-2 genes as described by Hasman et al. [27]. Detection of plasmid-mediated AmpC beta-lactamase genes was assessed by multiplex PCR [28]. Sequencing of both strands of amplicons was performed. The presence of the fluoroquinolone resistance genes aac(6')-Ib-cr, qnrA, qnrB, qnrS, qepA and oqxAB was also assessed [29,30].

Plasmid DNA analysis
Isolates exhibiting resistance to cephalosporins were selected for plasmid characterization. Plasmid replicons tested were elected according to the presence of determinant resistance genes (HI1, HI2, I1, X, L/M, N, FIA, FIB, W, Y, P, FIC, A/C, T, FIIA and K) [31] and were identified using the PCR-based replicon typing method previously described [32,33]. Plasmids detection and sizing was performed on all the isolates by S1-nuclease PFGE of total DNA [34]. Restriction fragments from S1-PFGE gels were transferred onto a positively charged nylon membrane and hybridised with specific probes for bla CTX-M-14 , bla TEM , bla SHV , bla CMY and for each replicon that was previously identified.

Results
A total of 22 E. coli were recovered from 13 different farms distributed throughout Spain during 2012. Additionally 10 isolates from healthy animals collected in nine different farms were also included in the study to make a total of 32 E. coli isolates.  Table 1). The most prevalent serotypes were: O3:H26, O5:H10, O5:H51 and O78:H9. Additionally, the emergent clone O25b:H4 was detected in a commensal isolate.
XbaI-PFGE analysis showed a high degree of genetic polymorphism. A total of 31 different PFGE restriction profiles were identified among the 32 E. coli isolates (Fig 1). Only two isolates were epidemiologically related and belonged to the same farm.

Detection of virulence-associated genes
The prevalence of 29 virulence-associated genes is shown in Tables 1 and 2. Regarding the five virulence genes associated to APEC (Fig 1); 81%, 78%, 75%, 72% and 72% of the 32 E. coli strains yielded amplicons for iss, ompT, iutA, iroN and hlyF, respectively ( Table 1). The prevalence of these genes was higher in APEC isolates (91% of the APEC strains harboured all of the mentioned genes), when compared to the AFEC isolates. In general, the presence of virulenceassociated genes in AFEC strains was low, with 80% of the strains having from zero to three of the previously mentioned virulence genes (Table 1). According to the number of virulenceassociated genes, 34% of the isolates were considered ExPEC.

Antimicrobial susceptibility testing and resistance genes
All the analyzed strains were multi-resistant (resistant to more than 3 antimicrobial families), including those isolated from healthy animals. Furthermore, 50% were resistant to more than eight antimicrobials. Susceptibility testing detected 11 strains resistant to cephalosporins (34%); six bla CTX-M-14 , two bla SHV-12 , two bla CMY-2 and one bla SHV-2 . Two of these strains belonged to AFEC isolates. Two isolates were resistant to cefoxitin, and the resistance mechanism involved could not be determined.
The presence of the qnrS gene was only confirmed in the isolate belonging to O25b: H4-B2-ST131. Finally, two of the APEC strains exhibited qnrA and two qnrA together with aac (6')-Ib-cr. The genes qnrB, qepA and oqxAB were not found in this strain collection.
fluoroquinolones. Several studies have demonstrated that the E. coli population of broilers are a reservoir of antimicrobial resistance genes that may be transferred by mobile genetic elements to the community via the food chain [12]. Although efforts have been implemented to reduce the use of antimicrobials in the poultry industry, these are sometimes overused in food producing animals, and particularly in broiler farms [35]. The restriction of fluoroquinolones and cephalosporins usage in livestock for human consumption, and the implementation of measures to limit the dissemination are needed [36]. This study have also demonstrated the presence of virulence genes associated to APEC strains in commensal E. coli, indicating a possible reservoir of virulence-associated genes in this population. However, the presence of the minimal predictors of APEC virulence was much higher in APEC isolates than AFEC isolates. Furthermore, the similarity of avian pathogenic strains with human pathogenic E. coli (ExPEC) based on virulence-associated genes was confirmed in this study, since a total of 34% of the isolates could be considered ExPEC. As also suggested by other studies, certain APEC subgroups, specifically a large proportion of phylogroup A isolates, may be considered potential zoonotic agents [6].
The information provided by different typing methods is usually unrelated. MLST results, phylogenetic groups and serotyping showed similarities between strains that with PFGE were not significant. For instance, the two highly pathogenic strains O78:H9-C-ST23 shared 80% of identity by PFGE, as well as the two strains O5:H10-A-ST93, which shared 70% similarity. Additionally, two AFEC strains O3:H26-A-ST165 exhibited 90% similarity by PFGE. In these cases, the virulence gene content and the results of the susceptibility testing were equal or almost equal. However, the combination of all these techniques may be useful to discriminate between strains causing outbreaks and those with potential risk to cause disease in animals or humans. For instance, combining serotyping, phylotyping and MLST the strain GN-2221 (O25b:H4-B2-ST131) could be identified as a potential zoonotic agent causing infections in humans. Still, further studies based in animal models are necessary to properly confirm the pathogenicity of the strain. All the serotypes found within this strain collection have been well described in previous studies [8,16,[37][38][39]. In contrast, different results have been observed from the ST data, where several ST types (ST10, ST23, ST48, ST57, ST93, ST117, ST131, ST156, ST350, ST429 and ST648) have been previously associated to APEC strains [8,16,[40][41][42], whereas other STs (ST101, ST165, ST189, ST297, ST533, ST539, ST624, ST650, ST889 and ST3161) have not been related to infections cause by APEC before.
The phylogroup B2 is known to harbour many more virulence-encoding genes than the rest of the E. coli phylogroups [10]. In our study, the two B2 strains found were O25b: H4-B2-ST131, which previously has been associated to human infections [43], and O2: H1-B2-ST429 which is frequently associated with poultry disease [44]. They exhibited 14 and 16 of the 29 virulence genes tested, respectively. Additionally, both strains were resistant to cephalosporins by production of CMY-2. The typical highly pathogenic APEC clonal group O78-H9-C-ST23 was found twice in this strain collection. Studies on sequencing and phylogenetic analysis of this clonal line revealed that O78 is more closely related to human strains (i. e. ST23 enterotoxigenic E. coli (ETEC)) than other APEC strains [44]. These results suggest that the E. coli population of broilers may be a potential reservoir of virulence-associated genes that could be transferred to humans through the food chain.
Our results are in line with the most commonly described ESBLs and AmpC producing E. coli in poultry production, which are CTX-M-14, CTX-M-1, CMY-2 and SHV-12 [45]. Also, these data provided evidence to the known genetic heterogeneity among ESBL-harboring E. coli isolates in broilers. In the last years, an increase in the presence of E. coli O25b: H4-B2-ST131 producing CTX-M-15 with a high virulence potential has been reported in human infections [46][47][48]. The high prevalence of this clonal group among multi-drug resistant isolates has important clinical and public health implications, due to the risk of treatment failure. The first time that the human ExPEC clonal group O25b:H4-B2-ST131 was detected in poultry was in Spain in 2010 [13,49]. In both cases, the genes encoding resistance to cephalosporins were CTX-M-9. Additionally, other studies have described ST131 isolates of human origin carrying CTX-M-15 and qnrS in IncN plasmids [50]. It is noteworthy that our study describes for the first time a poultry E. coli isolate clonal group O25b:H4-B2-ST131 producing CMY-2 with co-resistance to flouroquinolones (qnrS). Interestingly, this isolate presented an exceptional phenotype, since it was resistant to fluoroquinolones (MIC = 0.25) but was susceptible to nalidixic acid (MIC = 4) according to epidemiological cut-off values. This isolate was isolated from a healthy animal, corroborating the true zoonotic potential.
Up to date, the most common plasmids carrying cephalosporin resistance genes in E. coli isolated in poultry farms belong to the IncI1, IncFIB and IncN families [51]. However, in this study, cephalosporin resistance genes are mostly associated to IncK plasmids. Moreover, in some occasions two replicas of the same ESBL gene were harboured in two different plasmids. Interestingly, isolate O25b:H4-B2-ST131 harboured two plasmids from the same incompatibility group (IncN), in the same host cell. Therefore, it is probable that another unknown replicon is present and expressed in at least one of these plasmids.
In conclusion, this study demonstrated a very diverse population of multi-drug resistant E. coli in broiler farms containing a high number of virulence-associated genes. This is probably the combination of virulence and resistance genes transferring from one strain to another via mobile genetic elements creating a multi-clonal scenario, together with E. coli strains acquiring the genes and becoming clonally successful. More epidemiological studies are necessary to identify clonal groups and resistance mechanisms with potential relevance to public health. Additionally, prudent use of antimicrobials in animal production should be implemented to reduce the burden of resistance organisms entering the food chain.