Carriage of Antibiotic-Resistant Escherichia coli Among Healthy Children and Home-Raised Chickens: A Household Study in a Resource-Limited Setting

Study population

The study was carried out in a medium-sized urban area (approximately 30,000 inhabitants) of the Bolivian Chaco (Villa Montes, Tarija Department, Bolivia) and involved 12 households, randomly selected among those that had been found to rear domestic chickens in a previous survey (ANTRES study, see www.unifi.it/infdis/antres/default.htm). In that setting, chickens were reared in close household proximity, in conditions of high human–animal promiscuity.

For each household, rectal swabs were obtained from one or two healthy children aged 6–72 months (n = 21), and from one or two chickens (n = 21). Full ethics clearance was obtained from the qualified local authorities that had revised and approved the study design. Informed consent was obtained from children's parents or other legal guardians before collecting the samples.

Screening for fecal carriage of antibiotic-resistant E. coli

Screening for fecal carriage of antibiotic-resistant E. coli was carried out by a direct-plating method (DPM), as described previously.^3,4 Briefly, each rectal swab was spread on a MacConkey Agar No. 3 (MCA) plate (Oxoid) to yield uniform growth, and antibiotic-containing disks were directly placed onto the seeded plate. Antibiotics tested included ampicillin, ceftriaxone, tetracycline, chloramphenicol, streptomycin, kanamycin, gentamicin, amikacin, trimethoprim-sulphametoxazole, nalidixic acid, and ciprofloxacin (Oxoid). After incubation at 37°C for 12–14 hours, DPM plates were inspected for growth, and inhibition zone diameters were measured and interpreted according to previously described breakpoints. Only bacterial growth exhibiting an E. coli–like morphology was considered valid for the analysis.^3,4

E. coli isolates selected for molecular analysis

Molecular analysis was focused on isolates exhibiting a resistance phenotype to tetracycline or quinolones, neither of which is routinely used for children. Moreover, resistance to these two classes of drugs is mainly plasmid-mediated or mutational, respectively.^9,10

Three households in which carriage of tetracycline- or quinolone-resistant isolates had been observed in both humans and animals were selected. Rectal swabs from children (n = 6) and chickens (n = 6) were streaked on two MCA plates containing tetracycline 5 μg/mL (TET-MCA) or nalidixic acid 40 μg/mL (NAL-MCA), respectively. After incubation at 37°C for 12–14 hours, three isolated colonies from each plate were picked up and subjected to biochemical identification by the API20E system (BioMérieux) and in vitro susceptibility testing by the disk-diffusion method.^7,8 A total of 66 E. coli isolates were collected, of which 36 from TET-MCA (18 from chickens and 18 from children) and 30 from NAL-MCA (18 from chickens and 12 from children, as children from one of the three selected households did not carry nalidixic acid-resistant E. coli).

Genotyping of E. coli isolates

E. coli phylogroups (A, B1, B2, and D) were determined by the multiplex polymerase chain reaction (PCR) methods developed by Clermont et al. ⁶ Random amplification of polymorphic DNA (RAPD) genotyping was performed using, separately, the decamer primers 1290 and 1254, as previously described. ¹³ RAPD patterns were considered to be different when the profiles differed by at least one band.

Characterization of plasmids and of plasmid-mediated resistance genes

Tetracycline resistance genes [(tet(A), tet(B), tet(C), and tet(D)] were investigated by PCR, as described previously. ¹⁰ Plasmid-mediated quinolone resistance (PMQR) genes were investigated by PCR and sequencing as described previously (qnrA, qnrB, and qnrS ¹⁵ ; aac(6′)-Ib-cr ¹⁶ ; qepA ¹⁸ ). Controls for qnr and qepA genes were kindly provided by Prof. Patrice Nordmann and Dr. Laurent Poirel (Université Paris-Sud, K.-Bicêtre, France), and by Prof. Patrice Courvalin (Institut Pasteur, Paris, France), respectively. Nucleotide sequences were determined on both strands of PCR amplification products at the Macrogen sequencing facility (Macrogen Inc.).

Plasmid replicon typing was carried out by a multiplex PCR method, ⁵ and by Southern blotting using a probe specific for ColE-like replicons (not included in the multiplex PCR approach), as described elsewhere. ¹⁵ Plasmid restriction profiles were analyzed by agarose gel electrophoresis after digestion with HaeIII and EcoRI (Promega). Conjugal transfer of tet genes was assayed by mating experiments in Mueller Hinton (MH) broth (Difco Laboratories), using E. coli J53 (pro met [Rif^r] [Nal^r]) as the recipient and MH agar plates containing rifampin (400 μg/mL) and tetracycline (5 μg/mL) for selection of transconjugants, as described previously. ¹⁴ qnrB genes were transferred by electroporation into E. coli HB101 (F⁻ hsdS20 recA13 ara-14 proA2 lacY1 galK2 rpsL20 [Str^r] xyl-5 mtl-1 supE44 leuB6 thi-1), using MH agar plates containing nalidixic acid (8 μg/mL) for selection of transformants. Sequence analysis of gyrA and parC of transformants was carried out as described previously, ¹⁵ to exclude the occurrence of chromosomal mutations after selection on nalidixic acid.

Statistical analysis

Statistical differences in the prevalence of antibiotic resistance were determined by the Chi-square test. Confidence intervals were calculated by Stata Software release 8.0 (StataCorp., 2003).

Prevalence of fecal carriage of antibiotic-resistant E. coli

All children (n = 21) and chickens (n = 21) of the 12 households investigated were found to carry commensal E. coli with at least one acquired resistance trait. Except for resistance to quinolones, which was significantly higher in chickens (p < 0.001 and p = 0.01 for nalidixic acid and ciprofloxacin, respectively), resistance prevalence detected in children and chickens was overall comparable and characterized by very high resistance rates (≥50%) to ampicillin, tetracycline, chloramphenicol, streptomycin, and trimethoprim-sulphametoxazole (Table 1). Remarkable rates of resistance to these drugs have been previously reported in commensal E. coli from healthy individuals living in urban and rural areas of the Bolivian Chaco.^2,3

Phenotypic and genotypic characterization of E. coli isolates with acquired resistance to tetracycline or quinolones

To investigate the possible spread of resistant strains between children and home-raised chickens, E. coli isolates of human and animal origin exhibiting acquired resistance to tetracycline or quinolones were characterized. A total of 66 isolates were obtained by streaking rectal swabs from children (n = 6) and chickens (n = 6) of three selected households on TET-MCA and NAL-MCA plates, and picking up three isolated colonies from each plate (36 and 30 isolates from TET-MCA and NAL-MCA, respectively) (Table 2).

Susceptibility testing showed that all isolates selected onto NAL-MCA were also resistant to tetracycline, whereas 50% of those selected onto TET-MCA were also resistant to nalidixic acid. Moreover, a multidrug-resistant phenotype (resistance to more than three different antibiotic classes) was more common among isolates from children (70%) than among those from chickens (42%) (p = 0.02) (Table 2).

Resistant isolates belonged to phylogenetic group A (67%), B1 (18%), and D (15%): group A was significantly more prevalent in children than in chickens (80% vs. 56%, p = 0.03), group B1 was present only in chickens (33%, p < 0.001), whereas group D was equally distributed (20% vs. 11%, p = 0.3) (Table 2). A diverse distribution of the four main phylogenetic groups among human and animal commensal E. coli has been observed in a number of studies, and reflects adaptation to both host characteristics (e.g., diet, gut morphology, and body mass) and environmental forces. ¹⁷

RAPD genotyping demonstrated a remarkable clonal diversity of resistant isolates of both human and animal origin (12 and 14 different RAPD types, respectively), with most children (n = 4) and chickens (n = 5) being colonized by more than one resistant clone. Transmission of resistant strains was observed between children (family number 2, RAPD type 12) or chickens (family number 1, RAPD type 8) living in the same household, but in no case the same resistant strain was found to be shared by both humans and animals, or by members of different households (Table 2).

The overall genetic heterogeneity of resistant isolates and the observed intra host diversity of resistant E. coli suggested a possible role of horizontal gene transfer in resistance dissemination. To verify this hypothesis, the 66 isolates were investigated for the nature and transferability of plasmid-mediated tetracycline and quinolone resistance genes.

Nature and transferability of plasmid-mediated tetracycline and quinolone resistance genes

Resistance to tetracycline was related to carriage of either tet(A) or tet(B) genes (46% and 54% of RAPD types, respectively), with no significant association between nature of tet genes and human or animal origin of resistant isolates (p = 0.71 for tet(A) and tet(B)) (Table 2). tet genes were often located on conjugative multidrug resistance plasmids harboring different replicon combinations (F, HI1, I1, Iγ, and N). In no case inter-clonal plasmid dissemination was documented, based on replicon typing and RFLP analysis.

PMQR genes, investigated in all the 66 isolates regardless of their susceptibility phenotype, were detected only in isolates collected from children: aac(6′)-Ib-cr was found in a ciprofloxacin-resistant clone from a single child (RAPD type 2), and qnrB19 was found in two different ciprofloxacin-resistant clones from children living in the same household (RAPD type 11, colonizing only one child; RAPD type 12, shared by the two children) (Table 2). Analysis of transformants obtained with plasmid DNA from the qnrB19-positive clones revealed two small ColE-like plasmids (8 kb in RAPD type 11 and 2.7 kb in RAPD type 12), harboring qnrB19 as the sole antibiotic resistance gene. By restriction mapping with HaeIII, the 2.7 kb plasmid was found to be identical to pECY6-7, a qnrB19 harboring ColE-like plasmid that has been recently found to be widespread among commensal enterobacteria of healthy children in Bolivia and Peru. ¹⁵

Dissemination of PMQR genes is believed to be an important promoter for evolution of fluoroquinolone resistance in Enterobacteriaceae. ¹² The finding of PMQR genes in all the isolates exhibiting a resistance phenotype to ciprofloxacin would support this hypothesis.