Characterization of plasmid-mediated AmpC-producing E. coli from Swedish broilers and association with human clinical isolates


Corresponding author: S. Börjesson, National Veterinary Institute, SVA, Department of Animal Health and Antimicrobial Strategies, Stefan Börjesson, Travvägen 20, SE-751 89 Uppsala, Sweden



A selection of plasmid-mediated AmpC-producing Escherichia coli isolates carrying blaCMY-2 from Swedish broilers were characterized to establish their relatedness to and a possible overlap with human clinical E. coli isolates. The results showed diversity among the E. coli isolated from broilers, indicating that the spread in the population was not due to one strain. However, only one type of plasmid belonging to replicon type incK was identified. Furthermore, there were no indications of spread of blaCMY-2 E. coli isolates from broilers to human clinical settings, although Swedish broilers may be a source of blaCMY-2 and/or the plasmid carrying blaCMY-2.

The increasing trend worldwide of Enterobacteriaceae producing extended-spectrum beta-lactamases (ESBL) and plasmid-mediated AmpC (pAmpC) in clinical settings is of great concern, due to their ability to hydrolyse third-generation cephalosporins. Food-producing animals, poultry in particular, have been suggested to be a potential reservoir for these types of bacteria and the genes encoding ESBL/pAmpC. In a recent report the European Food Safety Authority concluded that there is indirect evidence that transmission of ESBL- and pAmpC-producing Enterobacteriaceae and genes encoding ESBL and pAmpC can occur between food-producing animals, especially broilers, and humans [1]. In Sweden, prescription of antimicrobials to animals is very limited compared with that in other European countries [2]. It was therefore an unexpected finding in 2010 that 34% of broilers in Sweden carried ESBL- or pAmpC-producing E. coli [3]. The only genes identified were blaCTX-M-1 and blaCMY-2, with blaCMY-2 clearly dominating. The cause of the high occurrence in broilers is not clear, but there are indications that imported breeding stocks may play a central role [3]. Furthermore, it is uncertain whether the high occurrence of blaCMY-2 has influenced the incidence among humans in clinical settings in Sweden. In this study we characterized a selection of blaCMY-2 E. coli isolated from the intestinal content of Swedish broilers and investigated the possibility of an overlap with human clinical blaCMY-2 E. coli isolates.

Twenty-two E. coli isolates carrying blaCMY-2 from broilers were obtained from the monitoring program SVARM 2010 [3] and were characterized using pulsed-field gel electrophoresis (PFGE) [4], multilocus sequence typing (MLST) ( and plasmid replicon (PR) typing [5]. Antimicrobial susceptibility was assessed using the VetMIC GN-mo plates (SVA, Uppsala, Sweden) interpreted with EUCAST epidemiological cut-off values. Transferability of blaCMY-2 was tested through conjugation to E. coli HMS174. Transformation was performed on isolates unable to transfer blaCMY-2 through conjugation and transconjugants carrying multiple PR-types. Plasmid DNA was transformed using ElectroMax™DH10B™. Transconjugants/transformants were tested for antimicrobial susceptibility, presence of AmpC genes [6] and PR types. The size of transferred plasmids was determined by nuclease-S1 treatment and PFGE according to Barton [7]. The 22 isolates belonged to 11 STs (Fig. 1; Table 1). Two isolates were defined as multiresistant (resistant to ≥ three classes) and 10 were resistant to beta-lactams only. There was an association between ST and resistance pattern (Fig. 1; Table 1). Isolates belonging to the same ST generally showed identical or closely related PFGE patterns (Fig. 1). Six of the isolates were non-typeable by PFGE (Table 1). The blaCMY-2 was identified on an IncK plasmid, except for one isolate where the blaCMY-2 could not be transferred. The blaCMY-2 + IncK plasmids transferred no additional resistance phenotype and were 60–100 kb in size.

Table 1. Characterization of pulsed-field gel electrophoresis non-typeable human clinical (H) and broiler (C) E. coli isolates carrying blaCMY-2
Isolate IDMLST (clonal complex)Plasmid replicon typesNon-beta-lactam resistance pattern
  1. C, Chloramphenicol; Ci, Ciprofloxacin; Na, Nalidixic acid; Gm, Gentamicin; Sm, Streptomycin; Su, Sulphamethoxazole; Tm, Trimethoprim; Tc, Tetracycline; ND, not determined; MLST, multilocus sequence typing.

C30ST117 (N.D.)FIA, FIBCiNa
C33ST57 (CC350)K, FIB, F, I1 
C41ST2183 (N.D.)K, YCiNaTc
C53ST10 (CC10)K 
C54ST2184 (N.D.)K 
C65ST57 (CC350)K, FIB, F, I1 
H33N.D.K, FIA, FIBCiNaGmSmSuTmtc

Compared with data from other European countries [1], the Swedish isolates showed a more homogeneous population structure regarding both phenotype and genotype. The sampling was designed so that each isolate represented a unique slaughter batch. Furthermore, the Swedish broiler houses are cleaned and disinfected between batches. Therefore sampling bias is not the likely explanation for these results. Furthermore, the high prevalence in Sweden cannot be explained by antibiotic usage because this is very limited and no cephalosporins were prescribed [8]. A potential explanation could be transmission of cephalosporin-resistant E. coli through the production pyramid. This type of transmission has been suggested in earlier studies, which have described the occurrence of cephalosporin-resistant E. coli, mainly pAmpC producing, in the production pyramid [1, 8, 9]. However, other sources, such as feed, the environment and recirculation, may play a role, but this need to be further studied.

Figure 1.

Dendogram showing the genotypic relatedness of 16 Escherichia coli carrying blaCMY-2 isolates of broiler origin (C) and 17 E. coli carrying blaCMY-2 isolates of human patient origin (H) based on pulsed-field gel electrophoresis. For comparison, Dice coefficient and unweighted pair group method with arithmetic mean cluster analysis was performed, with position tolerance and optimization both set at 1%. The scale bar represents percentage similarity and the vertical dotted line indicates ≥80% similarity. Multilocus sequencing type (MLST) was not determined (ND) for isolates of human origin. Chloramphenicol (C), ciprofloxacin (Ci), nalidixic acid (Na), colistin (Cs), florfenicol (Ff), gentamicin (Gm), kanamycin (Km), streptomycin (Sm), sulphamethoxazole (Su), trimethoprim (Tm) and tetracycline (Tc).

To investigate the potential influence of broiler isolates in human clinical settings, 72 isolates carrying blaCMY-2 collected at the Swedish Institute for Communicable Disease Control (SMI) from clinical laboratories selected to mirror the sampling period for the broilers and six isolates from a 2009 point prevalence study were characterized [10] (Edquist P, Olsson Liljequst B, Tegmark Wisell K, Giske CG, Unpublished data ). Twenty-one of the isolates tested positive for the IncK plasmid and these were subjected to PFGE, with four found to be non-typeable (Table 1). None of the 17 typeable isolates clustered with broiler isolates (Fig. 1). Moreover, the human isolates had a higher heterogeneity than the broiler isolates, were generally resistant to a wider range of antimicrobials and carried more PR types. Through transformation, 20 isolates transferred blaCMY-2, 19 tested positive for IncK and one for IncI1. No other resistance phenotypes were transferred and the plasmids were in the same size-range as those from broilers, with one exception, a 200 kb plasmid. The study by Leverstein-van Hall et al. [11] clearly demonstrated a relationship between ESBL-producing isolates of broiler and human origin. In contrast, the Swedish human and broiler isolates with blaCMY-2 appeared very distinct both genotypically and phenotypically (Fig. 1), at least among the limited number investigated here. These results indicate that there has been very limited or no direct transmission of E. coli isolates in Sweden between broilers and humans. In addition, a point prevalence study that involved the majority of Swedish human clinical laboratories, performed in Sweden in 2009 by SMI, found that a minority (approximately 6%) of cephalosporin-resistant E. coli carried pAmpC (Edquist P, Olsson Liljequst B,Tegmark Wisell K, Giske CG, Unpublished data).

Despite the differences in isolate level, it is possible that transmission of the plasmid has occurred. In earlier studies on human clinical isolates, blaCMY-2 has mainly been described on other RTs [12], although there is a recent example of dissemination of IncK-blaCMY-2 among E. coli isolates in intensive care units in Canada [13]. The prevalence of IncK-blaCMY-2 plasmids in Europe appears also to have increased in the broiler population over the last few years [14], which in the future may influence occurrence among humans. Overall, the potential transfer appears very limited in Sweden, as the E. coli isolates with IncK-blaCMY-2 constituted a clear minority of the human clinical isolates, whereas the combination is almost universal in isolates from broilers. A limitation of this study was the small number of isolates included and the lack of temporal or spatial linkages to the human isolates. However, it is unlikely that the combination of genes in the broiler population has changed markedly over time. In addition, continued surveillance has also shown that blaCMY-2 continues to dominate [8].

No closely related human and broiler E. coli isolates were identified, indicating limited transmission of isolates between the two populations. However, because IncK-plasmids were identified in both populations, transmission of the plasmid cannot be ruled out. The high occurrence of pAmpC-producing E. coli in Swedish broilers is in any case distressing. Because Sweden is a low-prevalence country as regards ESBL and pAmpC, the occurrence in broilers may have a higher impact in the future.


Christina Greko, National Veterinary Institute (SVA), is acknowledged for her input into the project and the manuscript. The result in this article was partly presented at the 3rd ASM Conference on Antimicrobial Resistance in Zoonotic Bacteria and Foodborne Pathogens, 2012, Aix-en-Provence, France.

Transparency Declaration

The authors declare no conflict of interests. The work was supported by the Swedish Civil Contingencies Agency.