Extended spectrum beta-lactamase and fluoroquinolone resistance genes and plasmids among Escherichia coli isolates from zoo animals, Czech Republic

Authors

  • Hana Dobiasova,

    Corresponding author
    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    2. CEITEC VFU, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    • Correspondence: Hana Dobiasova, Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Palackeho tr. 1–3, CZ-612 42 Brno, Czech Republic. Tel.: +420 541 562647; fax: +420 541 562631; e-mail: dobiasova.hanka@gmail.com

    Search for more papers by this author
  • Monika Dolejska,

    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    2. CEITEC VFU, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Search for more papers by this author
  • Ivana Jamborova,

    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Search for more papers by this author
  • Eva Brhelova,

    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Current affiliation:
    1. CEITEC MU, Masaryk University, Brno, Czech Republic
    Search for more papers by this author
  • Lucie Blazkova,

    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Search for more papers by this author
  • Ivo Papousek,

    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Search for more papers by this author
  • Marketa Kozlova,

    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Search for more papers by this author
  • Jiri Klimes,

    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Search for more papers by this author
  • Alois Cizek,

    1. CEITEC VFU, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    2. Department of Infectious Diseases and Microbiology, Faculty of Veterinary Medicine, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Search for more papers by this author
  • Ivan Literak

    1. Department of Biology and Wildlife Diseases, Faculty of Veterinary Hygiene and Ecology, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    2. CEITEC VFU, University of Veterinary and Pharmaceutical Sciences Brno, Brno, Czech Republic
    Search for more papers by this author

Abstract

Commensal Escherichia coli isolates from healthy zoo animals kept in Ostrava Zoological Garden, Czech Republic, were investigated to evaluate the dissemination of extended-spectrum beta-lactamase (ESBL) and plasmid-mediated quinolone resistance (PMQR) genes. A total of 160 faecal samples of various animal species were inoculated onto MacConkey agar with cefotaxime (2 mg L−1) or ciprofloxacin (0.05 mg L−1) to obtain ESBL- or PMQR-positive E. coli isolates. Clonality of E. coli isolates was investigated by multilocus sequence typing and pulsed-field gel electrophoresis. Plasmids carrying ESBL or PMQR genes were typed by PCR-based replicon typing, plasmid multilocus sequence typing and restriction fragment length polymorphism. Forty-nine (71%, = 69) cefotaxime-resistant and 15 (16%, = 94) ciprofloxacin-resistant E. coli isolates harboured ESBL or PMQR genes. Isolates were assigned to 18 sequence types (ST) and 20 clusters according to their macrorestriction patterns by pulsed-field gel electrophoresis. The genes blaCTX-M-1 and qnrS1 were detected on highly related IncI1 plasmids assigned to clonal complex 3 (ST3, ST38) and on non-related IncN plasmids of ST1 and ST3, respectively. The gene qnrS1 was located on related IncX1 plasmids. Dissemination of antibiotic resistance is associated with spreading of particular E. coli clones and plasmids of specific incompatibility groups among various animal species.

Introduction

Emergence and dissemination of multidrug-resistant (MDR) gram-negative bacteria is a significant health problem worldwide. The impact of antibiotic resistance is associated with a considerable decrease in the number of efficient antimicrobials available for the treatment of various bacterial infections (Livermore et al., 2007; Strahilevitz et al., 2009; Platell et al., 2011). Cephalosporins and fluoroquinolones belong to critically important antimicrobials in human and veterinary medicine (FAO/WHO/OIE, 2007). The predominant mechanism of resistance to β-lactam antibiotics is mediated by extended-spectrum β-lactamases (ESBL) or AmpC-type β-lactamases (Bergenholtz et al., 2009). Plasmid-mediated quinolone resistance (PMQR) genes confer only a low degree of resistance but their expression can enable development of strains highly resistant to fluoroquinolones (Strahilevitz et al., 2009). The co-localization of ESBL and PMQR genes on the same mobile genetic elements and their association with other resistance genes is of particular concern (Literak et al., 2010; Dolejska et al., 2011ab).

A number of studies have been focused on MDR Escherichia coli derived from humans, wild, domesticated or food animals and from the environment (Literak et al., 2010; Dolejska et al., 2011ab; Platell et al., 2011). Captive animals in zoological gardens can serve as a reservoir of antibiotic-resistant pathogenic bacteria such as Campylobacter spp., Salmonella spp., Yersinia spp., and pathogenic E. coli strains (Gopee et al., 2000; Stirling et al., 2008a). A high frequency of resistance including ESBL production and the presence of PMQR genes was found in E. coli isolated from zoo animals in Japan and China (Ahmed et al., 2007; Wang et al., 2012).

In this study we performed the detailed genotypic characterization of ESBL- and PMQR-positive E. coli isolates from animals kept in a zoo in the Czech Republic over a period of 1 year to assess the potential role of captive animals as a reservoir of emerging MDR bacteria and plasmids carrying antibiotic resistance genes.

Materials and methods

Ostrava Zoological Garden and antibiotic practice

Ostrava Zoological Garden, Czech Republic, houses about 3300 animals of 430 different species including 76 mammalian, 126 avian and 33 reptilian species. It occupies a wooded 1-km2 site in a suburb of the industrial city of Ostrava in the northeastern part of the Czech Republic. The zoo is visited by about 330 000 people per year.

Antibiotics were administered only on an individual basis under the guidance of the veterinary surgeon for a limited period of < 1–2 weeks in the case of suspected or documented bacterial infections, treatment of bite wounds or as surgical prophylaxis. In the year 2010, when our study was conducted, and during the preceding year, antimicrobials were therapeutically administered to only seven mammalian and two bird species. Antimicrobials used included cefovecin (3rd generation cephalosporin), cefprozil (2nd generation cephalosporin), cefadroxil (1st generation cephalosporin), amoxicillin–clavulanic acid, penicillin, streptomycin, enrofloxacin, marbofloxacin, tetracycline, oxytetracycline, doxycycline and neomycin.

The zoo feeds chickens to 19 animal species including eight mammalian and 11 bird species. All newly-hatched chickens were treated prophylactically with doxycycline for 5 days upon delivery and kept alive in the zoo until being used as feed within 1 month. A limited number of rabbits were also kept in the zoo and fed to zoo animals. The rabbits were treated with trimethoprim–sulphonamide irregularly in cases of digestive disorders. Carnivores were also fed meat from slaughtered food animals not suitable for human consumption. The meat might have contained antibiotic-resistant bacteria and antibiotic residues from recent antibiotic treatments.

Sampling

During the year 2010, Ostrava Zoological Garden was visited four times to obtain faecal specimens of defined individuals of various animal species kept in four distinct geographic areas of the zoo. Within each of four samplings, different animal species occupying a particular geographic area were sampled on one-time basis.

A total number of 160 faecal samples using sterile cotton swabs were obtained from 132 mammalian, avian and reptilian species. From the total of 160 faecal samples, 127 faecal samples were obtained from enclosures that were occupied by a single animal, 21 faecal samples were taken from enclosures occupied by two or more individuals of the same species and 12 faecal samples were taken from enclosures occupied by defined groups of animals belonging to two or more genera. The samples were placed into Amies transport medium (Oxoid, UK) and transported in an isothermal box to a laboratory for further processing. Samples were incubated in buffered peptone water (Oxoid) at 37 °C overnight.

Selective isolation of E. coli and antimicrobial susceptibility testing

All samples were enriched in MacConkey (MC) broth (Oxoid) at 37 °C for 24 h and subsequently cultured on MC agar (Oxoid) supplemented with cefotaxime (2 mg L−1) or ciprofloxacin (0.05 mg L−1) at 37 °C overnight. One lactose-positive colony grown on MC agar with antibiotics was selected per one sample and identified using the API 10S test kit (bioMérieux, France) and MALDI-TOF (MALDI Biotyper, Bruker Daltonics). Cefotaxime-resistant E. coli isolates were tested using the double-disk synergy test to identify ESBL-production (CLSI, 2008). Escherichia coli isolates selected on media with ciprofloxacin were tested for MIC of nalidixic acid and ciprofloxacin using the agar dilution method (CLSI, 2008). Both cefotaxime- and ciprofloxacin-resistant E. coli isolates were tested for susceptibility to 12 antimicrobial agents as described previously by Dolejska et al. (2011a) according to CLSI guidelines (CLSI, 2008).

Testing of antibiotic resistance genes, integrons and replicons

Genes responsible for ESBL phenotype (blaTEM, blaSHV, blaOXA and blaCTX-M) were identified by PCR and sequencing (Literak et al., 2010). The colonies grown on MCA with ciprofloxacin were tested for PMQR genes [qnrA, qnrB, qnrC, qnrD, qnrS, aac(6′)-Ib-cr, qepA and oqxAB] by PCR and sequencing (Kim et al., 2009; Literak et al., 2010). All isolates were screened for additional antibiotic resistance genes and integrons as described previously (Literak et al., 2010). Investigated resistance genes and integrons, primers, conditions of PCR and positive control strains used in the study are listed in Table S1 (available as Supporting Information on the FEMS Microbiology Ecology website). Replicons were tested using PCR-based replicon typing (Carattoli et al., 2005).

Escherichia coli molecular typing methods

All E. coli isolates were divided into phylogenetic groups (Clermont et al., 2000). Isolates belonging to phylogenetic groups B2 and D were tested by allele-specific rapid PCR to determine O25b-ST131 E. coli clone (Clermont et al., 2009). Epidemiological relatedness of E. coli isolates was analysed by XbaI-pulsed field gel electrophoresis (PFGE) (PulseNetUSA, 2004). All macrorestriction profiles were compared using the bionumerics fingerprinting software (Applied Maths, Belgium) dividing the isolates into epidemiologically related clusters (I–XV for cefotaxime- and 1–5 for ciprofloxacin-resistant E. coli isolates, respectively) because of a Dice similarity index ≥ 85%. The clusters I and V of cefotaxime-resistant isolates and the clusters 1 and 5 of ciprofloxacin-resistant isolates were divided into subgroups on the basis of intraclonal diversity of these clusters (Ia–d, Va–c of cefotaxime- and 1a–b, 5a–b of ciprofloxacin-resistant isolates). Multilocus sequence typing (MLST) was performed according to Wirth et al. (2006).

Plasmid characterization

Transferability of plasmids carrying ESBL or PMQR genes was achieved using conjugation and transformation experiments as described elsewhere (Dolejska et al., 2011b). Genomic DNA of transconjugants and transformants was isolated using the boiling technique described by Cattoir et al. (2007). Transconjugants and transformants with one plasmid were typed by PCR-based replicon typing (Carattoli et al., 2005), tested for the presence of additional antibiotic resistance genes using PCR (Table S1) and plasmid size was determined using S1-PFGE (Barton et al., 1995). Plasmids were isolated using the alkaline extraction method (Birnboim & Doly, 1979) and compared using restriction fragment length polymorphism performed by EcoRV digestion of plasmid DNA from transformants or transconjugants, dividing them into separate groups based on their restriction patterns. Plasmids showing an identical profile or one band difference were designated with the same letter (e.g. ‘A’) and those with two to three band differences were designated with a letter and a number of the particular type (e.g. ‘A1’). Plasmids belonging to incompatibility groups I1 and N were typed by plasmid multilocus sequence typing (pMLST) (Garcia-Fernandez et al., 2008, 2011) and assigned to sequence types (ST) using the on-line database http://pubmlst.org/plasmid/. Detection of insertion sequences ISEcp1 and IS26 located upstream of blaCTX-M-1 and ISEcl2 upstream of the gene qnrS1 was carried out by PCR mapping (Table S1). Investigation of the blaCTX-M-1 upstream region was conducted as previously described (Dolejska et al., 2011b). The upstream region of qnrS1 was investigated using forward primer ISEcl2 binding to orfB of the insertion sequence (Poirel et al., 2007) in combination with qnrS reverse primer (Cattoir et al., 2007) targeting a region of c. 1.5 kb.

Results

Isolation of ESBL- and PMQR-positive E. coli isolates and their antibiotic resistance phenotypes

From the 160 faecal samples of various animal species, 69 (43%) cefotaxime- and 94 (59%) ciprofloxacin-resistant E. coli isolates were found. Forty-nine (71%) cefotaxime-resistant and 15 (16%) ciprofloxacin-resistant E. coli isolates harboured ESBL or PMQR genes, respectively. All the ESBL-producing E. coli isolates were multiresistant (resistant to two or more antibiotic groups). ESBL-positive isolates showed resistance to tetracycline (98%) sulphonamides (96%), nalidixic acid (18%), streptomycin (16%), trimethoprim–sulphamethoxazole and gentamicin (both 4%). Of the PMQR-positive E. coli isolates, 12 isolates (80%) were multiresistant. PMQR-positive E. coli isolates showed resistance to ampicillin (73%), tetracycline (53%), sulphonamides and trimethoprim–sulphamethoxazole (both 47%), streptomycin (20%), cephalotin and amoxicillin–clavulanic acid (both 13%) and gentamicin (7%). Among the PMQR-positive E. coli isolates, MIC of nalidixic acid and ciprofloxacin varied from 8 to > 256 mg L−1 and from 0.06 to > 8 mg L−1, respectively.

Resistance genes and integrons detected in ESBL- and PMQR-positive E. coli isolates

In the 49 ESBL-producing E. coli isolates, the gene blaCTX-M-1 was detected in 48 isolates, and blaCTX-M-15 and blaSHV-27 in one isolate (130 CTX, Fig. 1). One CTX-M-1-producing isolate (116 CTX) also harboured PMQR gene qnrS1. In all PMQR-positive E. coli isolates, the gene qnrS1 was detected (Table 1). The additional antibiotic resistance genes were detected in the ESBL/PMQR-positive isolates (Fig. 1, Table 1): tet(A), tet(B) encoding resistance to tetracycline, sul1, sul2 encoding resistance to sulphonamides, and strA encoding resistance to streptomycin. The presence of class 1 integron was detected only in one isolate (13z CIP, Table 1).

Table 1. Characteristic of PMQR-positive Escherichia coli isolates from animals in the zoo
OriginStrain no.MIC (mg L−1)a NAL/CIPPFGE profilebSTPGPMQR genesAdditional AR genes and integrons
  1. AR, antibiotic resistance; PG, phylogenetic group; ST, sequence type.

  2. a

    Minimal inhibitory concentration of NAL, nalidixic acid and CIP, ciprofloxacin to donor strains.

  3. b

    Escherichia coli isolates were divided into clusters (1–5) consisting of clonally related isolates according to Dice similarity index of their XbaI macrorestriction profiles ≥ 85%. Escherichia coli isolates assigned to the clusters 1 and 5 were divided into subgroups (1a–b, 5a–b) owing to intraclonal diversity in these clusters.

  4. c

    Class 1 integron 1.5 kb: dfrA1-aadA1.

King Vulture (Sarcoramphus papa)143 CIP> 256/0.251a155B1 qnrS1
Amur Tiger (Panthera tigris)13z CIP> 256/> 81b3275B1 qnrS1 sul1, sul2, strA, tet(A),Int1c
Cottontop Tamarin - group (Saguinus oedipus)3 CIP> 256/> 821249B1 qnrS1 tet(A)
Serval (Leptailurus serval)32 CIP8/131434A qnrS1 blaTEM-32, strA, tet(A)
American Flamingo (Phoenicopterus ruber)71 CIP128/0.25410A qnrS1 blaTEM-1, sul2, tet(A)
Southern Screamer (Chauna torquata)101 CIP16/0.1255a1431B1 qnrS1 blaTEM-1, sul1, sul2, tet(A)
Indian Crested Porcupine (Hystrix indica)103 CIP8/0.55b1431B1 qnrS1 blaTEM-1, sul1, tet(A)
Figure 1.

Characteristics of ESBL-positive Escherichia coli isolates from animals in the zoo. AR, antibiotic resistance; PG, phylogenetic group; ST, sequence type. Dendrogram represents diversity of E. coli isolates carrying ESBL genes. Escherichia coli isolates were divided into clusters (I-XV) according to Dice similarity index of their XbaI macrorestriction profiles ≥ 85%. Escherichia coli isolates assigned to the clusters I and V were divided into subgroups (Ia–d, Va–c) owing to intraclonal diversity in these clusters.

Molecular typing of E. coli isolates

Forty-nine ESBL-producing E. coli isolates were divided into phylogenetic groups A (19 isolates), B1 (18), B2 (1) and D (11). Fifteen PMQR-harbouring E. coli isolates belonged to phylogenetic groups A (6) and B1 (9). None of the isolates assigned to phylogenetic groups B2 and D belonged to the O25b-ST131 type.

All E. coli were typed using XbaI-PFGE. Sixteen ESBL-producing and four PMQR-positive isolates were non-typeable using this method. Cluster analysis of macrorestriction patterns was performed on 33 ESBL-producing E. coli isolates and 11 PMQR-harbouring E. coli isolates. For MLST analysis, one to four representative ESBL- and PMQR-positive E. coli isolates of each cluster were chosen at random.

ESBL-producing isolates were grouped into 15 clusters (I–XV). Seventy per cent of them belonged to one of five clusters of clonally related isolates. Four of these clusters (I, V, VIII, X) consisted of three to eight isolates that were clonally related. Isolates belonging to the same cluster of clonally related isolates were obtained from different animal species. Ten E. coli isolates had unique macrorestriction profiles. Twenty representative ESBL-positive E. coli isolates of each cluster were assigned to 13 different STs (Fig. 1).

Eleven PMQR-harbouring E. coli isolates were grouped into five clusters (1–5). In all, 82% of PMQR-positive E. coli belonged to one of three clusters of clonally related isolates. Two of these clusters (clusters 2 and 5) consisted of three and four isolates originating from different animal species, respectively. Two PMQR-positive isolates had unique macrorestriction profiles. Seven representative PMQR-positive E. coli isolates of each cluster were assigned to six distinct STs (Table 1).

Characterization of plasmids carrying ESBL or PMQR genes

From a total of 33 CTX-M-producing and 11 PMQR-positive isolates, conjugation to E. coli or Salmonella enterica was demonstrated in 31 and 7, respectively. Only two CTX-M-carrying and none of the PMQR-harbouring plasmids were transferred through transformation. All transformants carried multiple plasmids and these plasmids were therefore not investigated further. For plasmid analysis, one to four transconjugants carrying a single plasmid with ESBL and/or PMQR gene were chosen per one cluster. A total of 14 blaCTX-M-1- and 4 qnrS1-carrying plasmids were characterized (Table 2).

Table 2. Characteristic of ESBL- and PMQR-positive plasmids from Escherichia coli isolates
E. coli isolatesESBL-/PMQR-positive plasmids
Strain no.PFGE profileSTInc groupSize (kb)RFLP profileaST/CCESBL-/PMQR-genesAdditional AR genesUpstream region of blaCTX-M-1/qnrS1
  1. AR, antibiotic resistance; CC, clonal complex; Inc, incompatibility group; NT, non-typeable; ST, sequence type.

  2. All plasmids were successfully transferred via conjugation to E. coli MT102RN and Salmonella Typhimurium SL5325 or Salmonella Enteritidis Fa8065 recipient strains.

  3. a

    RFLP profile determined by EcoRV digestion.

73 CTXIc58I1100A1NT bla CTX-M-1 sul2, tet(A)ISEcp1
72 CTXId58I1100A138/CC-3 bla CTX-M-1 sul2, tet(A)ISEcp1
71 CTXVIII1146I1105A13/CC-3 bla CTX-M-1 sul2, tet(A)ISEcp1
85 CTXX48I1105A13/CC-3 bla CTX-M-1 sul2, tet(A)ISEcp1
80 CTXIV746I1100A23/CC-3 bla CTX-M-1 sul2, tet(A)ISEcp1
76 CTXIX58I1100A23/CC-3 bla CTX-M-1 sul2 ISEcp1
86 CTXXII2325I1105A23/CC-3 bla CTX-M-1 sul2 ISEcp1
29 CTXIa58I1100A33/CC-3 bla CTX-M-1 sul2, tet(A)ISEcp1
30 CTXIb58I1100A33/CC-3 bla CTX-M-1 sul2, tet(A)ISEcp1
116 CTXII1324I1100A33/CC-3 bla CTX-M-1 sul2, qnrS1 ISEcp1
12z CTXVII58I1140A33/CC-3 bla CTX-M-1 sul2, tet(A)ISEcp1
74 CTXXI3274I1100A33/CC-3 bla CTX-M-1 sul2, tet(A)ISEcp1
71 CIP410I1125A438/CC-3 qnrS1 blaTEM-1, sul2, tet(A)ISEcl2
78 CTXXIII1288N40a1 bla CTX-M-1 ISEcp1::IS26
34 CTXXIV410N40a1 bla CTX-M-1 sul2, tet(A)ISEcp1::IS26
32 CIP31434N50b3 qnrS1 blaTEM-32, strA, tet(A)ISEcl2
101 CIP5a1431X150α qnrS1 bla TEM-1 ISEcl2
13z CIP1b3275X150α qnrS1 ISEcl2

Plasmids carrying ESBL or PMQR genes belonging to three incompatibility (Inc) groups IncI1, IncN and IncX1 were identified. The genes blaCTX-M-1 and qnrS1 were located on IncI1 plasmids varying in size from 100 to 140 kb identified in 13 transconjugants. Plasmids showed related restriction profile (A1–A4) and belonged to clonal complex 3 (CC-3) except for IncI1 plasmid from the isolate 73 CTX, which could not be typed using pMLST. IncN plasmids of ST1 and ST3 carrying blaCTX-M-1 or qnrS1 genes were found in three transconjugants. Two qnrS1-harbouring IncX1 plasmids showed identical restriction profiles. ISEcl2 was located upstream of qnrS1 at the identical distance on all IncX1, IncI1 and IncN plasmids carrying this gene, whereas the gene blaCTX-M-1 was found downstream of the continuous or truncated ISEcp1 element (Table 2).

Discussion

Pepperell et al. (2002), Lavigne et al. (2007) and Cavaco et al. (2008) proposed that commensal microflora could represent a long-term reservoir of resistance genes that could be transferred horizontally to other bacteria. In this study, we describe the high prevalence of resistance to third generation cephalosporins and fluoroquinolones in commensal E. coli isolates from healthy captive animals in a zoo in the Czech Republic. ESBL-producing and PMQR-harbouring isolates were found in 31% (= 160) and 9% of sampled animals belonging to different avian, mammalian and reptilian species, respectively. Similar results were obtained by Wang et al. (2012), who found 32% (= 206) ESBL-producing E. coli isolates including strains carrying various PMQR genes in primates kept in six zoos in China.

The high occurrence of ESBL-producing and PMQR-harbouring isolates found in our study could be associated with antibiotic therapy used in animals for treatment of various infections. Beta-lactam antibiotics including the second and the third generation cephalosporins, fluoroquinolones, tetracyclines and aminoglycosides were applied, however, only to limited number of animals. Selection of ESBL-producing E. coli in the gut microflora of animals induced by cephalosporin treatment has been demonstrated (Cavaco et al., 2008). Most isolates showed resistance to multiple antibiotics including tetracyclines, sulphonamides, aminoglycosides and quinolones; therefore, the drugs of other antibiotic groups might have played a role in co-selection of ESBL- and PMQR-positive E. coli clones or plasmids carrying particular resistance genes.

Several studies have concentrated on the possible transmission of zoonotic ESBL-producing E. coli through the food chain (Livermore et al., 2007; Leverstein-van Hall et al., 2011; Platell et al., 2011). In the zoological garden, 19 sampled animals were fed chickens and rabbits treated with antibiotics. These chickens and rabbits are a plausible source of ESBL-producing and PMQR-harbouring E. coli isolates found in the zoo animals. Escherichia coli clones belonging to particular sequence types (ST10, ST58, ST117) harbouring blaCTX-M-1 or qnrS1 isolated from different avian and mammalian species kept in the Czech zoo have been identified previously in poultry, poultry meat and humans in the Netherlands (Leverstein-van Hall et al., 2011), supporting our theory of zoonotic transmission of antibiotic-resistant E. coli clones via the food chain. However, samples of food (chickens, rabbits or other food sources) and water sources of zoo animals or faecal samples of possible vectors (e.g. zoo-keepers) were not analysed in our study.

Identical or closely related IncI1 plasmids of clonal complex 3 were the most prevalent among blaCTX-M-1-harbouring E. coli isolates of different clonal lineages originating from several animal species fed chickens and rabbits. IncI1 plasmids belonging to ST3 carrying blaCTX-M-1 are widely disseminated in E. coli and S. enterica isolates from poultry throughout Europe (Garcia-Fernandez et al., 2008; Cloeckaert et al., 2010; Leverstein-van Hall et al., 2011) but they have been also found in other food-producing and companion animals as well as humans (Madec et al., 2011). Results of plasmid analysis might indicate that CTX-M-1-positive E. coli harbouring IncI1 plasmids isolated from zoo animals could originate in poultry.

Plasmids of incompatibility group IncN harbouring ESBL or PMQR genes identified in our study have been previously demonstrated in E. coli and S. enterica isolates from humans, animals and the environment in various European countries (http://pubmlst.org/plasmid, Dolejska et al., 2013). IncN plasmids belonging to ST1 harbouring blaCTX-M-1 are broadly disseminated in E. coli isolates from poultry, pigs, horses, cattle, wild waterbirds and humans (http://pubmlst.org/plasmid, Dolejska et al., 2013). Plasmids of IncX1 group carrying qnrS1 genes have been identified in E. coli from poultry (Cerquetti et al., 2009), wild waterbirds (Literak et al., 2010), horses (Dolejska et al., 2011a) and humans (Literak et al., 2011).

The abundance of MDR E. coli isolates in respective faecal samples of zoo animals was not determined in our study. ESBL- or PMQR-positive E. coli might have constituted either the predominant population of E. coli in the intestine of particular zoo animals or might have been transient flora. Duval-Iflah et al. (1981) showed that indigenous flora that was adapted to host intestine was the dominant population and exerted a barrier against the establishment of externally introduced E. coli strains. However, even the transient presence of MDR E. coli harbouring conjugative ESBL- or PMQR-carrying plasmids might be of particular concern. Licht et al. (1999) showed that conjugation in the intestine of a streptomycin-treated mouse model occurred initially at high rates after introduction of high numbers of the E. coli donor cells. Thereafter the rate of horizontal transfer of plasmids dropped but transconjugants remained present in the faeces of investigated mice at rather high levels.

In our study, we showed that zoo animals can represent a reservoir of resistance genes to critically important antimicrobials. Dissemination of these resistance genes is of particular concern. Multiresistance can be transmitted via direct contact between animals and humans, as has been shown by Moodley & Guardabassi (2009) and Dolejska et al. (2011a). Moreover, there has been a rise in the popularity of petting farm areas of zoos with direct contact between animals and visitors, especially children (Stirling et al., 2008b).

In conclusion, our study showed a high prevalence of E. coli harbouring ESBL or PMQR genes among various animal species in the Czech zoological garden. The extensive distribution of MDR E. coli is associated with successful spread of particular clones as well as horizontal gene transfer of related plasmids. Results of molecular analyses might indicate a common source of MDR E. coli possibly spreading via the food chain. However, further investigation is warranted to assess the potential impact of transmission of MDR E. coli via the food chain in dissemination of antimicrobial resistance in zoo animals.

Acknowledgements

The authors are grateful to the zoo director Petr Colas, the head of zoological department II Ivo Firla, and the zoo veterinarian Petr Gajdosik for their co-operation. We thank George Jacoby (Lahey Clinic Inc., Burlington, MA), Lina Cavaco, Henrik Hasman (National Food Institute, Copenhagen, Denmark), Ming-Gui Wang (Institute of Antibiotics, Huashan Hospital, Shanghai, China), Surbhi Malhotra-Kumar (University of Antwerp, Antwerp, Belgium) and Alessandra Carattoli (Instituto Superior di Sanita, Rome, Italy) for positive control strains. We thank Eva Suchanova and Martina Masarikova for their excellent laboratory work. This study drew on the pMLST website (http://pubmlst.org/plasmid/) developed by Keith Jolley and sited at the University of Oxford, UK. This study was funded by the project ‘CEITEC – Central European Institute of Technology’ (CZ.1.05/1.1.00/02.0068) from the European Regional Development Fund and Grant no. P502/10/P083 of the Czech Science Foundation.

Ancillary