Sampling sites, collection and processing of samples
A total of 224 samples were collected during October, 2015 to May, 2016 from poultry sources, including 7 hatcheries, 9 broiler farms, and 4 LBMs in three sub-districts (Gazipur sadar, Sreepur and Tangail sadar) of Bangladesh (Fig. 1). Each of these sub-districts are among the regions with high number (>3,000) of poultry farms. The LBMs in these sub-districts are the primary commercial hub of trading poultry animals, which are mostly obtained from multiple farms at the nearby localities. The selection of the broiler farms was based on an inclusion criterion of a minimum flock size of 2000, while the hatcheries and LBMs located beyond 10 km2 of the selected broiler farms were excluded. In each sub-district, 3 broiler farms and at least 2 hatcheries, and one LBM were selected, keeping a minimum distance of 2 km for multiple sites of each category. Farm samples included broiler meats (BM-F, n = 28,), cloacal swabs (CS, n = 49), feed (F, n = 21), and drinking water (W, n = 24). Samples from hatcheries included chick meconium (CM, n = 33). LBM samples included broiler meats (BM-M, n = 27), drinking water (W, n = 21), and floor swabs (FS, n = 21) (Table 1). Aseptic measures were followed while collecting these samples, the amount of which varied type-wise: 100 g (wet weight) for meat and feed, 500 mL for water and 1-5 mL or mg swabbed materials in sterile cotton swab for faeces/cloaca and wet floor samples. The swab samples were preserved and transported in Cary-Blair transport media. Moreover, a total of 50 samples, including broiler meat (n = 26, each representing a composite of thigh, breast and drumstick) and liver (n = 24) were sampled from LBMs for screening of antimicrobial residues. To minimize sampling bias, three sub-samples of each sample were randomly collected and pooled together. The samples were immediately kept in sterile plastic containers, transported in an insulated foam box with cold chain (temperature, 4-6˚C) and processed within 6 h.
Culture based detection of Campylobacter spp.
The poultry-originated samples were screened for detection and isolation of Campylobacter spp. following filter-based selective culture method described by Shiramaru et al. (2012) [22]. In brief, the swab samples (1 g each) were suspended in sterile phosphate buffer saline (pH 7.4, 1 mL), while the meat samples (10 g each) were homogenized with the sterile saline (50 mL) using a tissue grinder (Seward Medical Ltd, London, UK). At least three replicates of a 100 µl portion of each suspended sample were spread onto a membrane filter (mixed cellulose ester type, 0.45 µm pore size, 47 mm diameter; Sartorius Stedim Biotech, Germany). In case of each water sample, a 100 mL portion in three replicates was filtered through the same kind of filter. Each of the membrane filters with inoculated samples was overlaid on blood agar medium (blood base agar no. 2; Oxoid, UK), supplemented with 5% defibrinated sheep blood, and kept at room temperature for 30 min. Afterwards, the filter was removed and the medium was incubated under microaerophilic conditions (5% O2, 10% CO2 and 85% N2) at 37ºC for 48 hours. Suspected colonies were further cultured on blood agar plate in microaerophilic conditions stated above to obtain pure isolates. Presumptive Campylobacter isolates were subjected to species-specific morphological and biochemical assays, including Gram's staining, motility test, catalase, oxidase, and hippurate hydrolysis, according to standard procedures [7].
PCR-based confirmation of the species identity
DNA template of each isolate of Campylobacter spp. was prepared by boiling method described by Hoshino et al. (1998) [23]. In brief, a pure bacterial colony grown on blood based agar was mixed gently in 250 µL distilled water and subjected to boiling, followed by immediate cooling on ice, for 10 minutes each. The tubes were then centrifuged at 10,000X g for 10 minutes and the supernatant was collected as DNA template for polymerase chain reaction (PCR). Initially, PCR screening targeting 16S rRNA gene was performed according to Samosornsuk et al. (2007) [24] to confirm whether the strains belonged to the genus Campylobacter. Afterwards, cdtC gene based multiplex PCR was done for species-specific (C. jejuni,C. coli and C. fetus) identification following methods described by Asakura et al. (2008) [25]. DNA templates of C. jejuni ATCC33560, C. coli ATCC33559 and C. fetus ATCC27374 strains were used as positive controls, and that of Escherichia coli ATCC 25922 was used as a negative control. Details of all primers and corresponding PCR amplicon sizes are shown in Additional File 1. PCR products were subjected to gel electrophoresis (1.5% agarose, Invitrogen, USA) and after staining with ethidium bromide (0.5 µg ml-1) and destaining with distilled water, each for 10 minutes, gel images were captured using a UV transilluminator (Biometra, Germany).
Determination of antimicrobial resistance
All Campylobacter strains were tested for their resistance pattern by standard disk diffusion method. Eight commonly used antimicrobials with standard doses (μg) were examined: amoxicillin (AMX, 30 μg), azithromycin (AZM, 30), ciprofloxacin (CIP, 5 μg), erythromycin (ER, 30 μg), gentamicin (GM, 10 μg), tetracycline (TET, 30 μg), streptomycin (STR, 10 μg) and norfloxacin (NOR, 10 μg). In brief, freshly grown broth culture (equivalent to 0.5 McFarland turbidity) of each strain was uniformly inoculated, using sterile cotton swab, over the entire surface of Muller Hinton agar (Oxoid, UK), supplemented with 5% defibrinated sheep blood. Afterwards, 3-4 antimicrobial discs were placed in each agar plate and incubated at 37 °C for 48 h under microaerobic conditions (5% O2, 10% CO2 and 85% N2). The zones of growth inhibition were compared with the zone size interpretative standards as described by the Clinical and Laboratory Standard Institute (2016) [26] and thereby interpreted as susceptible (S), intermediate resistant (I) or resistant (R) to the antimicrobials. E. coli strain ATCC 25922 was used as a quality control organism. All data were confirmed by conducting at least two replicates of the disc diffusion experiments.
Detection of antimicrobial residues
A total of 50 samples, including broiler meat (n = 26) and liver (n = 24) were tested for the presence of oxytetracycline, ciprofloxacin and enrofloxacin residues. Solid phase extraction of the samples was performed according to Popelka et al. (2005) [27]. A portion (4 gm) of grinded meat or liver tissues was homogenized in 10 ml phosphate buffer (pH 6.5), and treated with 2 mL trichloroacetic acid (30%, v/v) to fractionate the proteins, followed by centrifugation at 7000X g, and sonication in an ultrasonic bath, 15 min each. The supernatant (ca. 2 ml) was treated with formaldehyde (100 µL) at 100ºC for 45 min, transferred to a new tube and mixed with equal amount of diethyl ether for 10 minutes at 25ºC, and the oily top layer was discarded. Extraction with diethyl ether was repeated twice. Afterwards, the extracts were filtered (0.45 µm), dried by evaporation and reconstituted with 2 mL mixture of 99% methanol and acetone (1:1) to obtain the final extract, which was stored at 4ºC.
Standard solutions of residues of three antimicrobial agents, namely, oxytetracycline, ciprofloxacin and enrofloxacin, each >98% purity, were prepared by dissolving 1.0 g powder of each in 5 mL methanol (99%) and kept in the dark at 4ºC until analysis by thin layer chromatography (TLC) within a week following standard procedures [28]. Briefly, TLC silica plates of 0.25 mm thickness (Merck, Germany) were activated at 120°C for 2 hrs before inoculating 50 μl of sample extract or standard solution of antimicrobial residues. Acetone-methanol (1:1) was used as mobile phase solvent and chromatographs were observed at 256 nm. Each of the samples was analyzed in triplicate. An internal standard and a blank was included after every five samples during analysis by both methods.
Quantification of residual antimicrobials were done by ultra-high performance liquid chromatography (UHPLC) for the meat and liver samples, which showed positive by TLC, following procedures described by Cooper et al. (1998) [29]. Stainless steel column Acclaim 120, C18 (5 μm, 120Å, 4.6 X 250 mm) was used for chromatography (Dionex ultimate 3000 UHPLC). Phosphate buffer solution was prepared by adding di-sodium hydrogen phosphate to 0.2 M potassium di-hydrogen phosphate solution until pH 5.0. HPLC mobile phase constituted of distilled water and acetonitrile (85:15, v/v) for oxytetracycline, and 0.01 M phosphate buffer and acetonitrile (80:20, v/v) for others. Acetonitrile was HPLC grade (Panreac, Italy) and other reagents were of p.a. grade (Merk, Germany). A portion (25 μl) of each sample was injected into chromatographic column, equilibrated with mobile phase, and run at 0.8 ml min-1 until the mobile phase ascended 7 cm. Afterwards, the column was air dried and visualized under UV light (λ = 254 nm and 366 nm). Standard controls (2 to 200 µg ml-1) were prepared by serially (2-fold) diluting the stock solution of each antimicrobial. Six replicates of each concentration were assayed to standardize the regression equation (coefficient value >0.99). Identification was done by comparing Rf values of antibiotic standards, i.e., 0.35, 0.80, and 0.97 for oxytetracycline, ciprofloxacine, and enrofloxacin, respectively. Peak area was used for antimicrobial quantification by regression analysis, Y = aX + b, where Y = component area or height, a = slope and b = intercept of the regression line, and X= estimated amount of antimicrobial. Extraction recovery was evaluated with comparison of peak areas for standard antimicrobial solution to that of the TLC-negative broiler tissue homogenates, spiked with the same standard solution.
Data collection on health safety and hygiene practices
A total of 14 broiler farms, including the 9 sampled farms and additional 5 farms in the same geographical locations were enrolled for survey using semi-structured interviews through participatory methods to understand poultry husbandry associated risks, personal and environmental hygiene, and vulnerabilities to campylobacter infections, antimicrobial usages, and occupational safety. Each semi-structured interview was conducted using a questionnaire (see Additional File 2) and involving at least a couple of representatives from each of the selected farms (n = 14). Two teams (A and B), each comprising three experienced veterinarians from the Bangladesh Agricultural University, conducted the interviews at two phases upon prior written consent from each of the key informants, including farmer, farm manager or owner. Team A collected data from half of the selected farms (n = 7) in the first phase, which were verified by Team B in the second phase, and vice versa for the rest farms (n = 7). A total of 56 semi-structured interviews, including two from each farm at each of the two phases (3-month interval), were accomplished.
Data management and Statistical analyses
Data were recorded into Microsoft Excel 10 (Microsoft Corporation, Redmond, WA, USA) spreadsheet from the hard copies and statistically analysed using ‘Xact’ (ver. 7.21d, SciLab GmbH, St. Yrieix, France) and Statistica (ver. 10.0, StatSoft Inc., USA). Significant difference or association (p <0.05) between the prevalence of Campylobacter species and any individual geo-socio-anthropogenic variables (study sites, and sample sources and types) was determined by Fisher’s exact test since the group-wise sample number was small. Antimicrobial susceptibility profile of the bacterial isolates was evaluated according to their differentiation into three independent groups, i.e., resistant, intermediate, and susceptible. MDR trait was defined as bacterial strains showing resistance against at least one antimicrobial agent in three or more antimicrobial classes [30]. Descriptive comparison of resistant patterns of Campylobacter strains with respect to diversity in sources and/or anthropogenic factors was performed using mean/ median and standard deviation, and also in the form of the box plots. Differences in the patterns and/or occurrence of antibiotic resistance in Campylobacter spp. (C. coli and C. jejuni) were calculated by the Paired Samples t-test. A ‘p’ value of <0.05 was considered significant.