This study was conducted in a chicken farm with two different poultry houses (1000 m² each). The two houses were separated about 50 m to each other. After hatching, 50,000 chickens were equally and randomly allocated into two poultry houses (Day 0). Chickens in the medicated group (n = 25,000) were given successive prophylactic doses (0.5 mg/l) of apramycinsulfate (Shandong Qilu King-phar Pharmaceutical Co., Ltd., Shandong, China) in their drinking water from Days 1 to 5. In comparison, the un-medicated group (n = 25,000) was given drinking water without apramycin. No other antibiotics were used during the study period. Add antibiotic to drinking water for 5 days is the normal production behavior of the laying hens company. This study was carried out without any additional interference with the growth of the chickens. The protocol was approved by the Animal Ethics Committee of Sichuan University. We confirm that the best practice veterinary care and informed consent has been granted by the owners.

Samples were taken from each group as described in Table 1. Specifically, 15 cloacal swabs were collected from both the medicated and un-medicated groups at Day 0 and placed separately into sterile plastic bags. Fifteen sterilized plates were randomly placed under selected cages along two main diagonals of the poultry house containing both the medicated and un-medicated groups. Plates were placed at 12:00 am and withdrawn at 3:00 pm to allow for the collection of fresh fecal samples. Collections occurred on Days 1, 2, 3, 4, 5, 6, 8, 10, 15, 20, 30, and 40. Flies were captured using a sweep net on each sampling day from both of the two houses and approximately 30 flies were individually placed into sterile tubes for later morphological classification. All samples were placed into cool boxes containing ice packs and transported to the lab within 4 h for immediate bacterial isolation.

The cloacal swabs (n = 30) were separately put into 10 ml phosphate-buffered saline (PBS) and thoroughly vortexed. The resulting suspension was then 10-fold serial diluted with PBS and 100 μl of the dilution was plated onto eosin methylene blue (EMB) agar (Hangzhou Microbial Reagent Co., Ltd., Hangzhou, China) and incubated at 37°C overnight.

Fecal samples were collected from medicated (n = 15) and un-medicated groups (n = 15) at each sampling time. From these fresh fecal samples, 0.1 g was put into 10 ml PBS and thoroughly vortexed. The resulting suspension was 10-fold serial diluted with PBS and 100 μl was plated onto EMB agar and incubated at 37°C overnight.

Houseflies were collected at each sampling time, as previously described. Collected houseflies were morphologically identified using a stereomicroscope and 15 houseflies were randomly chosen for subsequent E. coli isolation. Each housefly was put into 10 ml PBS and thoroughly vortexed. The resulting suspension was 10 times gradient diluted with PBS, 100 μl was plated onto EMB agar, then incubated at 37°C overnight.

After overnight incubation, two colonies from each plate were selected for each sample. All isolates were then confirmed as being E. coli using a biochemical identification kit for Enterobacteriaceae (Hangzhou Microbial Reagent Co. Ltd., Hangzhou, China). All the confirmed E. coli isolates were kept frozen (-70°C) with 25% glycerol pending further analysis.

The minimum inhibitory concentration (MIC) of apramycinsulfate (China Institute of Veterinary Drugs Control, Beijing, China) for all E. coli isolates was determined using the agar dilution method following the guidelines of the Clinical and Laboratory Standards Institute [CLSI] (2012a). In short, E. coli strains were subcultured on Luria Bertani (LB) agar at 37°C for 12 h. A clearly separate colony of the E. coli isolate was picked and a suspension of each strain in saline solution was adjusted to match the 0.5 McFarland standard. Mueller–Hinton (MH) plates that contain different apramycinsulfate concentration (0.125–1024 μg/ml) were seeded with a multipoint inoculum replicator and incubated at 35°C for 16–18 h. E. coli ATCC 25922 was used as the quality control strain. MIC data were only accepted if MICs of the control strains were within the required reference ranges. MIC90 (the MIC that ≥90% tested bacteria were inhibited for each sampling group) was used to evaluate the changes trend of apramycin resistance.

For detection of apramycin resistance genes, genomic DNA was prepared using a QIAamp DNA Mini Kit according to the manufacturer’s instructions (Qiagen Inc., Valencia, CA, United States). Apramycin resistance genes aac(3)-IV and npmA were screened for all E. coli isolates as previously described (Yates et al., 2004; Zhou et al., 2010).

The clonal relatedness of aac(3)-IV-positive isolates were typed by PFGE as previously described (Gautom, 1997). Briefly, 145 aac(3)-IV-positive isolates were subcultured on LB agar at 37°C for 12 h. A single colony of each isolate was suspended with cotton swab in about 2 ml of TE buffer. The cell suspensions were adjusted to 20% transmittance by using a bioMérieux Vitek (Hazelwood, MO, United States). Proteinase K and lysozyme were added into 100 ml cell suspensions at final concentration of 1 mg/ml each and then incubated at 37°C for 10–15 min. Following the lysozyme–proteinase K incubation, 7 ml of 20% sodium dodecyl sulfate (50°C) and 140 ml of 1.2% InCert Agarose (50°C) were mixed with each bacterial suspension. Then the mixture was immediately added to plug molds (Bio-Rad Laboratories). After that, each solid plug was transferred to 2-ml round-bottom tubes with 1.5 ml of ESP buffer and incubated at 55°C for 2 h in a water bath. Then five times washes with 8–10 ml TE buffer (50°C) each in a shaker water bath for 15 min were carried out. For restriction endonuclease digestion, two 1-mm-thick slices of each plug were incubated at 37°C for 3 h with 50 U of XbaI enzyme. The plugs were then soaked in standard 0.5 Tris–borate–EDTA (TBE) prior to electrophoresis. The electrophoretic conditions used were as follows: initial switch time, 2.16 s; final switch time, 54.17 s; run time, 22 h; angle, 120°; gradient, 6.0 V/cm; temperature, 14°C; ramping factor, linear. PFGE profiles were analyzed using the BioNumerics Program (Applied Maths, Sint-Martens-Latem, Belgium) as previously described (Yates et al., 2004). The clonal clusters with a similarity cutoff value of 80% were used in this study.

To investigate the antimicrobial resistance patterns and resistance genes of aac(3)-IV-positive isolates belonging to different PFGE types, we tested one isolate of each PFGE type for susceptibility to 22 antimicrobial agents. This process was conducted using the disk diffusion method according to CLSI guidelines (Clinical and Laboratory Standards Institute [CLSI], 2012b). Briefly, MH agar plate was inoculated with suspensions of bacteria, equivalent to standard 0.5 McFarland. Subsequently, the disks of different antimicrobial agents were placed on media and then incubated at 35°C for 16–18 h. The tested antimicrobial agents were as follows: ampicillin (10 μg), piperacillin (100 μg), cefazolin (30 μg), ceftazidime (30 μg), cefotaxime (30 μg), ceftriaxone (30 μg), cefepime (30 μg), amoxicillin/clavulanic acid (20/10 μg), ampicillin/sulbactam (10/10 μg), piperacillin/tazobactam (100/10 μg), aztreonam (30 μg), imipenem (10 μg), meropenem (10 μg), tetracycline (30 μg), doxycycline (30 μg), ciprofloxacin (5 μg), levofloxacin (5 μg), gentamicin (10 μg), amikacin (30 μg), sulfamethoxazole/trimethoprim (1.25/23.75 μg), chloramphenicol (30 μg), and florfenicol (30 μg). All tested antimicrobial agents were obtained from Oxoid (Basingstoke, United Kingdom). E. coli ATCC 25922 was used as the control strain. The obtained data were interpreted according to CLSI recommendations (Clinical and Laboratory Standards Institute [CLSI], 2016).

Finally, we screened for the presence of 25 additional types of resistance genes and integron integrates genes in the 12 aac(3)-IV-positive isolates were screened using primers and PCR conditions as previously described: bla_TEM, bla_SHV, bla_OXA-1-like, bla_{CTX-M-group 1}, bla_{CTX-M-group 2}, bla_{CTX-M-group 9}, bla_{CTX-M-group 8/25} (Dallenne et al., 2010), tetA, tetB, tetM (Ng et al., 2001), qnrA, qnrB, qnrC, qnrD (Schink et al., 2012), aac(3)-IIa, aac(6′)-Ib, ant(3″)-Ia, aph(3′)-IIa (Zhang et al., 2012), sulI, sulII (Kerrn et al., 2002), cfr, cmlA, floR (Keyes et al., 2000; Kehrenberg and Schwarz, 2006), IntI, and IntII (Ishikawa, 2011).

Statistical analysis was performed using SPSS software for Windows, version 18.0 (SPSS Inc., Chicago, IL, United States). Data were analyzed using descriptive statistics and χ² tests. A P-value < 0.05 was considered statistically significant.

Over the course of the 40-day testing period, a total of 585 samples were collected. Two E. coli strains were selected from each sample. As shown in Table 1, a total of 1170 E. coli isolates from the medicated group (n = 390), un-medicated group (n = 390), and housefly group (n = 390) were obtained. Prior to apramycin administration (Day 0), 90 E. coli strains were collected from the included samples, 450 E. coli strains were collected during apramycin administration (Days 1–5), and 630 E. coli strains were collected after apramycin administration.

Minimum inhibitory concentration for apramycin was tested for all 1170 E. coli isolates. MIC90 was used to evaluate the changes trend of apramycin resistance (Figure 1).

For E. coli isolates obtained from the medicated group, apramycin MIC90 was at a low level (8 μg/ml) prior to apramycin administration (Day 0). After the addition of apramycin, MIC90 increased significantly from Days 2 to 6 and was maintained above 512 μg/ml compared to that in Day 0 and Day 1 (P < 0.05). This was with the exception of Day 5, which sustained a level of 256 μg/ml. However, ending apramycin administration resulted in a substantial decrease in MIC90 (8–16 μg/ml) from Days 8 to 15. To our surprise, MIC90 increased again (above 512 μg/ml) from Days 20 to 40.

For E. coli isolates obtained from the un-medicated group, apramycin MIC90 was remained at low level (8–16 μg/ml) from Days 0 to 8. This was with the exception of Day 3, which sustained a level of 64 μg/ml. Days 10–20 saw a dramatic increase (128–1024 μg/ml), but a subsequent decrease to 8 μg/ml from Days 30 to 40. Significant difference was found for the MIC90 values between E. coli isolates from the un-medicated group and medicated group (P < 0.05).

For E. coli isolated from houseflies, apramycin MIC90 remained at a low level (8–16 μg/ml) from Days 0 to 6, then increased and fluctuated between 256 and 1024 μg/ml from Days 8 to 40. MIC90 values for apramycin were significantly different between 1–6 days and 8–40 days for E. coli isolated from houseflies (P < 0.05).

Apramycin resistance genes aac(3)-IV and npmA were screened for all 1170 E. coli isolates. Aac(3)-IV was detected in 32, 71, and 42 E. coli isolates from the un-medicated, medicated, and housefly groups, respectively. npmA gene was not detected in any samples from this study. The change of aac(3)-IV frequency is shown in Figure 2.

For the medicated group, aac(3)-IV detection rate was 6.67% before treatment (Day 0) and showed a steady increase from Day 1 (3.33%) to Day 4 (63.33%). Rates then decreased and fluctuated between 0 and 23.33% from Days 5 to 40. Noticeably, aac(3)-IV detection rates were still higher than Day 0. This rate held even 35 days after treatment (Day 40).

For the un-medicated group, aac(3)-IV detection rate showed no drastic change when compared to Day 0. Rates fluctuated between 3.33 and 16.67% for the entirety of the experiment.

For the housefly group, aac(3)-IV detection rate was low from Days 0 to 6 (0–3.33%), then increased and fluctuated between 13.33 and 36.67% from Days 8 to 40.

The aac(3)-IV detection rate was significantly different between medicated group and un-medicated group from days 3 to 4 (P < 0.05). No significant difference was found between un-medicated group and housefly group (P > 0.5).

A total of 145 aac(3)-IV-positive E. coli isolates from the un-medicated (n = 32), medicated (n = 71), and housefly groups (n = 42) were analyzed using PFGE and 12 PFGE types were characterized (Figure 3). Among these, the three predominant PFGE types that emerged in the un-medicated group were types A (n = 12), B (n = 4), and D (n = 5). In the medicated group, the three major types were types A (n = 8), E (n = 39), and G (n = 9) and the housefly group were types A (n = 7), E (n = 11), and G (n = 19). PFGE types A, E, and G were the predominant types in both the medicated and housefly groups, suggesting houseflies play an important role in the spread of antibiotic-resistant E. coli.

Antimicrobial resistance profiles of the 12 E. coli isolates from each PFGE type are shown in Table 2. All tested isolates were multi-resistant, showing an antimicrobial-resistant phenotype to 10–18 antibiotics. Furthermore, all 12 isolates were co-resistant to the following antibiotics: ampicillin, tetracycline, doxycycline, ciprofloxacin, levofloxacin, gentamicin, and sulfamethoxazole/trimethoprim. They showed sensitivity to piperacillin/tazobactam, imipenem, meropenem, and amikacin. The number of isolates resistant to other antimicrobials ranged from 4 to 11 (Table 2).

Resistance gene screening results showed multiple resistance genes co-existed in all 12 different E. coli isolates from each PFGE type (Table 2). The isolates among the 12 different PFGE types harboring resistance genes other than aac(3)-IV are shown in Table 2. Remarkably, 10 isolates harbored at least one ESBL genes (bla_{CTX-M-group 1or9}). Moreover, among the 12 isolates, 11 were positive for the type I integrase gene intI.