Abstract
The prevalence of extended-spectrum beta-lactamase (ESBL)– and plasmidic AmpC–beta-lactamase (pAmpC-BL)–producing Escherichia coli isolates has been studied in food-producing animals at the farm level in Tunisia, and recovered isolates were characterized for the presence of other resistance genes and integrons. Eighty fecal samples of food-producing animals (23 sheep, 22 chickens, 22 cattle, six horses, five rabbits, and two dromedaries) were obtained from 35 different farms in Tunisia in 2011. Samples were inoculated onto MacConkey agar plates supplemented with cefotaxime (2 mg/L) for cefotaxime-resistant (CTXR) E. coli recovery. CTXR E. coli isolates were detected in 11 out of 80 samples (13.8%), and one isolate per sample was further characterized (10 from chickens and one from a dromedary). The 11 CTXR isolates were distributed into phylogroups: B1 (five isolates), A (two isolates), D (three isolates), and B2 (one isolate). The following beta-lactamase genes were detected: bla CTX-M-1 (seven isolates), bla CTX-M-1+bla TEM-135 (one isolate), bla CTX-M-1+bla TEM-1b (one isolate), and bla CMY-2 (two isolates). All ESBL- and pAmpC-BL–producing E. coli strains showed unrelated pulsed-field gel electrophoresis patterns. Seven isolates contained class 1 integrons with four gene cassette arrangements: dfrA17-aadA5 (three isolates), dfrA1-aadA1 (two isolates), dfrA15-aadA1 (one isolate), and aadA1 (one isolate). All isolates showed tetracycline resistance and contained the tet(A) +/− tet(B) genes. Virulence genes detected were as follows (number of isolates in parentheses): fimA (10); aer (eight); papC (two); and papGIII, hly, cnf, and bfp (none). Chicken farms constitute a reservoir of ESBL- and pAmpC-BL–producing E. coli isolates of the CTX-M-1 and CMY-2 types that potentially could be transmitted to humans via the food chain or by direct contact.
Introduction
Most ESBLs are variants of CTX-M, TEM, or SHV families (Bonnet, 2004) and confer resistance to a variety of beta-lactam antibiotics, including penicillins, 2nd, 3rd, and 4th generation cephalosporins and monobactams (e.g., aztreonam), but usually not carbapenems or cephamycins (e.g., cefoxitin). pAmpC-BL confer resistance to penicillins, 2nd and 3rd generation cephalosporins, including β-lactam/inhibitor combinations, cefamycins (cefoxitin), but usually not 4th generation cephalosporins (cefepime, cefquinome) and carbapenems. The CMY type is the most frequently reported pAmpC-BL in E. coli (Jacoby, 2009; EFSA, 2011). ESBL- and CMY-producing E. coli have emerged as a community pathogen in many parts of the world (Pitout and Laupland, 2008). Recent reports indicate that food might be a source of human-acquired antimicrobial-resistant E. coli (Carattoli, 2008) due to the fact that similar ESBLs and plasmids encoding them have been detected in food-producing animals, food of animal origin, and humans (Leverstein–van Hall et al., 2011).
Our laboratory recently reported CTX-M-1–producing E. coli strains from food samples and healthy humans in Tunisia (Ben Sallem et al., 2011; Ben Slama et al., 2010; Jouini et al., 2007), but this type of beta-lactamase was very infrequent in clinical E. coli isolates in this country (Ben Slama et al., 2011). The present study was conducted to analyze the prevalence of ESBL- and pAmpC-BL–producing E. coli isolates in food-producing animals in Tunisia, to determine the type of enzymes produced by these isolates, to compare the obtained results with previous data on food, healthy human, and clinical samples in this country, and to deepen the knowledge of the routes of transmission of ESBLs in different ecosystems.
Materials and Methods
Isolates and susceptibility testing
Eighty fecal samples of food-producing animals (23 sheep, 22 chickens, 22 cattle, six horses, five rabbits, and two dromedaries) were obtained from 35 different farms in Tunisia from February to May 2011. The farms of chickens and rabbits were of intensive and extensive production, and the farms of the other tested animals were of extensive production (Table 1). In the case of chicken and rabbit farms of intensive production, only one fecal sample of each farm was obtained by mixture of four fecal samples recovered at different areas of the farm (samples were obtained from the floor in chicken farms, when animals were 35 days old). In the case of the farms of extensive production, more than one sample was obtained from each farm from fecal samples of selected animals (Table 1).
ESBL, extended-spectrum beta-lactamase; pAmpC-BL, plasmidic AmpC–beta-lactamase.
Samples were seeded onto MacConkey agar plates supplemented with cefotaxime (CTX, 2 mg/L). After incubation at 37°C for 24 h, colonies showing E. coli morphology were recovered, and identified by classical biochemical methods and by species-specific polymerase chain reaction (PCR; amplification of uidA gene) (Jouini et al., 2007). One CTX-resistant (CTXR) E. coli isolate per sample was selected and screened for ESBL and AmpC phenotypes by double disk test (CLSI, 2010; Tan et al., 2009). Susceptibility testing to 18 antibiotics (ampicillin, cefoxitin, ceftazidime, cefotaxime, imipenem, aztreonam, gentamicin, amikacin, tobramycin, kanamycin, streptomycin, nalidixic acid, ciprofloxacin, sulphonamides, trimethoprim-sulfamethoxazole, tetracycline, rifampicin, and chloramphenicol) was carried out by disk-diffusion method (CLSI, 2010). The minimum inhibitory concentrations (MICs) of cefotaxime, ceftazidime, aztreonam, and cefoxitin were determined by agar dilution method (CLSI, 2010). E. coli ATCC 25922 was used as a control strain.
Pulsed-field gel electrophoresis (PFGE) analysis and phylogroup determination of CTXR E. coli strains
The clonal relationship among CTXR E. coli strains was determined by PFGE using XbaI enzyme as previously described (Sáenz et al., 2004). Patterns were visually analyzed and interpreted according to previously reported criteria (Tenover et al., 1995). The isolates were assigned to the phylogenetic groups A, B1, B2, or D using a PCR strategy with specific primers for chuA, yjaA, and TspE4.C2 determinants (Clermont et al., 2000).
Serotyping and virulence genotyping of E. coli isolates
All isolates were screened for O25b and O157 serotypes and for afa/dra operon (Blanco et al., 2009; Clermont et al., 2008). In addition, the sxt, fimA, papG allele III, hlyA, cnf1, papC, aer, eae, and bfp genes, encoding virulence factors often found in pathogenic E. coli (ExPEC) isolates, were tested by PCR (Ruiz et al., 2002).
Detection and characterization of beta-lactamase genes, genetic environment of blaCTX-M and blaCMY genes and other antibiotic resistance genes
The genes encoding TEM, SHV, OXA-1, CTX-M, and CMY type beta-lactamases and the genetic environment of bla CTX-M and bla CMY-2 genes were analyzed by PCR and sequencing (Vinué et al., 2008). The presence of genes associated with resistance to tetracycline [tet(A) and tet(B)], sulphonamides [sul1, sul2, and sul3], gentamicin [aac(3)-II, and aac(3)-IV], streptomycin [strA and strB], and quinolones [qnr, qepA, and aac(6’)-Ib-cr] was determined by PCR (Ben Slama et al., 2011).
Detection and characterization of integrons
The presence of intI1 and intI2 genes (encoding class 1 and class 2 integrases, respectively) and the 3’-conserved segment (qacEΔ1-sul1 genes) of class 1 integrons was examined by PCR. The variable regions of class 1 and class 2 integrons were characterized by PCR and sequencing in all intI1- or intI2-positive isolates (Ben Slama et al., 2011).
Results
CTXR E. coli isolates were detected in 11 out of 80 fecal samples of healthy food-producing animals analyzed (13.8%) that were recovered in 11 out of the 35 farms tested (31.4%). Nine of these samples contained ESBL-positive E. coli isolates, all of them harbored the bla CTX-M-1 gene, and two of these strains also harbored the bla TEM-135 or bla TEM-1b genes (Table 2). The ISEcp1-bla CTX-M-1-orf477 structure was found in all nine ESBL-positive isolates. The remaining two CTXR E. coli isolates contained the bla CMY-2 gene (encoding the beta-lactamase CMY-2). The ISEcp1-bla CMY-2-blc structure was identified in these two strains, but the IS10 sequence was demonstrated in one of them, truncating ISEcp1 (Table 2); this last structure is new and has been included in GenBank with the accession number JX440359.
Origin of the samples: C, chicken; D, dromedary.
SXT, trimethoprim-sulfamethoxazole; SUL, sulphonamides; TET, tetracycline; NAL, nalidixic acid; CIP, ciprofloxacin; KAN, kanamycine; STR, streptomycin; (i), intermediate resistance.
It is of interest that all except one of the CTXR E. coli isolates were recovered from chicken samples. The remaining one was from a dromedary sample. No CTXR E. coli isolates were detected in fecal samples of other farm animals such as sheep, cow, horse, or rabbit. Two types of chicken farms were tested (of intensive and extensive farming), and all CTXR isolates (ESBL- or pAmpC-BL–producing isolates) were detected in eight out of 14 farms with intensive production, but none in those of extensive production (Table 1).
Seven ESBL-positive isolates contained class 1 integrons with the following gene cassette arrangements: dfrA17-aadA5 (three isolates), dfrA1-aadA1 (two isolates), dfrA15-aadA1 (one isolate), and aadA1 (one isolate). The dfrA17-aadA5 was detected inside a class 1 integron lacking the qacEΔ1 and sul1 genes in the three E. coli isolates (Table 2). No class 2 integron was detected among the studied strains. The phenotypes of resistance of all CTXR isolates as well as the MICs of beta-lactams are shown in Tables 2 and 3. As expected, all bla CTX-M-1-positive isolates exhibited very high MIC values for cefotaxime (≥128 μg/mL) and the two bla CMY-2-positive isolates high MIC values for cefoxitin (64 μg/mL).
PFGE analysis demonstrated unrelated pulsotypes among all 11 CTXR isolates (Table 2). Phylogenetic analysis revealed that these strains were classified into the following phylogroups: B1 (five strains), A (two strains), D (three strains), and B2 (one strain; Table 2).
A variety of resistance genes located outside integrons were observed among our strains: tet(A) or tet(B) (in the 11 tetracycline-resistant strains), strA/B with/without aadA1 (in two streptomycin-resistant strains), and sul2 (in 11 sulphonamide-resistant strains). The virulence genes fimA, aer, and papC were detected in 10, eight, and two isolates, respectively, but none of the ESBL-producing isolates harbored the virulence genes sxt, papG-III, hly, cnf1, eae, afa/dra, and bfp, or were ascribed to the serotypes O25b or O157.
Discussion
To our knowledge, this is the first study of the genetic background of cefotaxime resistance in commensal E. coli isolates recovered from food-producing animals in Tunisia and one of the first reports in Africa. There is only one previous report about the detection of ESBL in food-producing animals in the African continent, and it concerned the detection of a CTX-M-15–producing E. coli strain among 89 ampicillin-resistant isolates obtained from fecal microbiota of healthy food-producing animals in Nigeria (Fortini et al., 2011). Nonetheless, numerous reports are available from other continents and countries in food-producing animals (Briñas et al., 2005; Carattoli, 2008; EFSA, 2011; Smet et al., 2008).
Our findings show a high percentage of fecal carriage of ESBL-positive E. coli isolates from healthy food-producing animals (11.2%) in samples obtained in 2011. If we consider the percentage of carriage in particular animal species, it is of interest that 10 of 22 tested samples of chicken origin carried CTXR E. coli isolates (45.5%): all of them obtained in farms of intensive production and eight of them (36.4%) carrying ESBL-producing isolates. These resistant isolates were not detected in fecal samples of sheep, cattle, horses, or rabbit, in contrast with other studies (Blanc et al., 2006; Ho et al., 2011; Horton et al., 2011; Zhao et al., 2001). The number of samples tested in our work was low, but these results could indicate an enrichment of CTXR E. coli isolates (with ESBL or pAmpC-BL) in chicken farms, which might reflect a high antibiotic pressure for selection of resistant bacteria in this ecosystem. These results are in agreement with the high incidence of ESBL-positive E. coli carriage on raw chicken meat in Tunisia (Ben Slama et al., 2010; Jouini et al., 2007). In a previous study performed by our group in 2007 with a low number of fecal samples of food-producing animals at the farm level (Jouini et al., 2007), no CTXR isolates were detected, which could reflect an increase in the prevalence of these resistant microorganisms in past years.
Our results increase the number of hosts of bla CTX-M-1-producing E. coli isolates in Tunisia (food samples and healthy humans) (Ben Sallem et al., 2011; Ben Slama et al., 2010) and now in farm animals and might reflect the successful spread of an epidemic plasmid. This possibility is also supported by our PFGE result, which showed unrelated patterns among all strains (Table 2). E. coli with bla CTX-M-1 have been identified in food-producing animals in various European countries (Aarestrup et al., 2006; Bortolaia et al., 2010; Briñas et al., 2005; Girlich et al., 2007; Moodley et al., 2009), and the bla CTX-M-1 gene has been frequently detected in IncI1 and IncN plasmids (Blanc et al., 2006; Girlich et al., 2007; EFSA, 2011; Moodley and Guardabassi, 2009; Bortolaia et al., 2010).
The ISEcp1 insertion sequence has been observed upstream of the ORFs encoding the CTX-M-1 and the CMY-2 enzymes in all CTXR strains. This ISEcp1 element contains typical –35 and −10 putative promoter regions and could mobilize such genes (Eckert et al., 2006). Thus, the two different genetic environments detected in this study for the bla CMY-2 gene are of interest.
There are reports of bla CMY-2 genes in E. coli isolates from humans, food animals, and companion animals (Ben Slama et al., 2010; Briñas et al., 2005; Carattoli et al., 2005; Mataseje et al., 2010; Murphy et al., 2009; Yan et al., 2004).
Interestingly, one of the strains (C4333) harbored both the bla CTX-M-1 and the bla TEM-135 genes. This association of genes in the same strain was previously observed in an E. coli strain of pet origin in Tunisia (data not shown). The bla TEM-135 gene was first found in Salmonella enterica serovar Typhimurium and recently was found in penicillinase-producing N. gonorrhoeae isolates (Nakayama et al., 2011; Ohnishi et al., 2010; Pasquali et al., 2005). Although the TEM-135–associated resistance phenotype does not correspond to an ESBL phenotype, its detection in strains of animal origin is worrisome, since it is considered a possible direct precursor of an ESBL (Nakayama et al., 2011).
Most ESBL-producing isolates exhibited resistance to antibiotics used in intensive animal production, mainly streptomycin, tetracycline, sulphonamides, and trimethoprim, and these resistances could play an important role in the co-selection of ESBL-producing bacteria (Carattoli, 2008).
In agreement with other studies, integrons were commonly identified, and they corresponded to a few integron types, especially class 1 integrons lacking the qacEΔ1 and sul1 genes (Ben Sallem et al., 2011; Ben Slama et al., 2010, 2011). Transfer of plasmids with class 1 integrons between bacterial isolates from food-producing animals and humans has been suggested previously (Kang et al., 2005; Leverstein-van Hall et al., 2002; Singh et al., 2005)
Most of our ESBL–producing E. coli isolates belonged to phylogroups B1 and A (n=7), which are more often associated with animal or human commensal E. coli isolates; phylogroups D and B2 were less represented among our isolates (n=2). None of our ESBL-positive isolates were ascribed to phylogroup B2, though this group is very frequently detected among clinical ESBL-positive isolates. Nevertheless, one of our CMY-2–producing isolates belonged to phylogroups B2. More studies should be performed in the future in order to determine if the prevalence of the B2 phylogroup is higher among pAmpC-BL– than among ESBL-producing E. coli isolates of animal origin.
Conclusion
Chickens have become an important reservoir of CTXR E. coli isolates. Our study reports the dissemination of the genes bla CTX-M-1 and bla CMY-2 in E. coli isolates of fecal samples of chickens at the farm level in Tunisia. Detailed molecular comparison of plasmids and genomes of isolates from various sources will help to better define the transmission dynamics of bla CTX-M between humans and food-producing animals.
Footnotes
Acknowledgments
This study was financed by Integrated Action of the Spanish Agency of International Collaboration (AECID, reference A1/038210/11) of Spain and the Tunisian Ministry of Higher Education and Scientific Research. V.E. has a predoctoral fellowship from the University of La Rioja (Spain).
Disclosure Statement
No competing financial interests exist.