Comparative Studies of Antimicrobial Resistance in Escherichia coli, Salmonella, and Campylobacter Isolates from Broiler Chickens with and without Use of Enrofloxacin

This study investigated the effect of enrofloxacin (ENR) administration on the prevalence and antimicrobial resistance of E. coli, Salmonella, and Campylobacter isolated from broiler chickens under field conditions. The isolation rate of Salmonella was significantly lower (p < 0.05) on farms that administered ENR (6.4%) than on farms that did not (11.6%). The Campylobacter isolation rate was significantly higher (p < 0.05) in farms that administered ENR (6.7%) than in farms that did not (3.3%). The ratio of resistance to ENR was significantly higher (p < 0.05) in E. coli isolates from farms that used ENR (88.1%) than farms that did not (78.0%). The respective ratio of resistance to ampicillin (40.5% vs. 17.9%), chloramphenicol (38.0% vs. 12.5%), tetracycline (63.3% vs. 23.2%), and trimethoprim/sulfamethoxazole (48.1% vs. 28.6%) and the ratio of intermediate resistance to ENR (67.1% vs. 48.2%) were significantly higher (p < 0.05) in Salmonella isolates from the farms that used ENR than farms that did not. In conclusion, the use of ENR at broiler farms was an important factor in decreasing the prevalence of Salmonella but not Campylobacter and caused ENR resistance among E. coli and Salmonella but not Campylobacter. Exposure to ENR could have a co-selective effect on antimicrobial resistance in enteric bacteria in the field.


Introduction
In many countries, antimicrobial resistance in foodborne pathogens is recognized as an accumulative problem [1,2]. Systemically administered antimicrobial agents may cause the emergence and spread of antimicrobial resistance among zoonotic foodborne pathogens, such as non-typhoid Salmonella serotypes and Campylobacter [3,4]. Salmonella enterica and Campylobacter infections are the major causes of foodborne gastroenteritis in humans (known as salmonellosis and campylobacteriosis, respectively) in many countries [4][5][6][7]. Salmonella causes gastroenteritis with clinical symptoms such as headache, abdominal pain, and diarrhea in both adults and children; campylobacteriosis frequently occurs in children younger than five years of age, and Campylobacter species cause a wide range of syndromes, from asymptomatic infections to severe systemic infections. In addition, the major source of Salmonella and ENR leads to a rapid selection of Campylobacter, E. coli [42][43][44], and Salmonella [36] which are less susceptible to these antibiotics. Nevertheless, few studies have focused on the relationship between ENR use and ENR resistance in foodborne pathogens from broilers in the field [45]. In this field trial, we investigated the relationship between ENR use and the emergence of ENR resistance in Salmonella and Campylobacter, whereas E. coli was used as an indicator bacterium.

Studied Isolates
There were two groups of farms (Group 1-contained farms that use ENR; and Group 2-contained farms that do not use ENR). E. coli isolates (n = 516; 306 and 210 isolates from Group 1 and Group 2, respectively), Salmonella isolates (n = 255; 134 and 121 isolates from Group 1 and Group 2, respectively), and Campylobacter isolates (n = 165; 132 and 33 isolates from Group 1 and Group 2, respectively) were sequestered and assessed. All isolates were from 2418 clinical samples composed of fecal (1-day-old and 15-25-days-old broiler chickens) or from environmental source (litter, feed, and water) samples collected from broiler farms or meat samples collected from downstream retail markets (Table S1). All samples were collected between September and December in 2015. In Group 1, 0.5 g/L ENR was constantly added to the drinking water of 2-4-day-old chicks. Salmonella, E. coli, and Campylobacter were isolated using the standard method described previously [45][46][47][48].
No endangered population was involved, and no endangered species were used for the experiments. No chickens were touched or killed; fecal samples were collected by a veterinarian with prior approval from the farm managers. Therefore, no ethical approval was required for this study.
The MIC 50 and MIC 90 values, as well as the range of values obtained, are important parameters for reporting the results of susceptibility testing when multiple isolates of a given species are investigated. The MIC 50 represents the MIC value at which ≥50% of the isolates in a test population are inhibited; it corresponds to the mean MIC value. The MIC 90 represents the MIC value at which ≥90% of the strains within a test population are inhibited [53].

Statistical Analysis
Statistical analysis was performed using the SPSS program version 19.0 for Windows. The Chi-square test was used to analyze significant differences in isolation rates of Salmonella and Campylobacter and AMR rates of E. coli, Salmonella, and Campylobacter between broilers from Groups 1 and 2. p-values < 0.05 were considered indicative of statistical significance. Results are reported as the mean with 95% confidence interval (CI).

Prevalence of Salmonella and Campylobacter in the Presence or Absence of ENR Use
Salmonella isolation rates from 15-25-days-old broilers and from retail meat were significantly lower (p < 0.001) in Group 1 than in Group 2 ( Table 1). There was no significant difference in the Salmonella isolation rate of 1-day-old broilers between Group 1 and Group 2. Campylobacter isolation rates of 15-25-days-old broilers and retail meat were significantly higher (p < 0.05) in Group 1 than in Group 2. The Campylobacter isolation rate of 1-day-old broilers was similar between Group 1 and Group 2. Bold type indicates p-value < 0.05; ENR: enrofloxacin; results are expressed as the mean (95% confidence interval). Group 1-contained farms that use ENR; and Group 2-contained farms that do not use ENR. In Group 1, 0.5 g/L ENR was constantly added to the drinking water of 2-4-day-old chicks.

Antimicrobial Resistance of E. coli, Salmonella, and Campylobacter Isolates in the Presence or Absence of ENR Use
The prevalence of antimicrobial resistance differed between Group 1 and Group 2 for 16 antimicrobials assessed (Tables 2-4). E. coli isolates from 15-25-days-old broilers from Group 1 were significantly more likely to be resistant to cephalothin (p = 0.038) and ENR (p = 0.048) ( Table 2). The prevalence of multidrug resistance in E. coli isolates did not differ significantly between Group 1 and Group 2. A higher frequency of Salmonella isolates from 15-25-days-old broilers from Group 1 were resistant to ampicillin (p = 0.005), chloramphenicol (p = 0.001), tetracycline (p < 0.001), and intermediate resistance (IR) to ENR (p = 0.028) ( Table 3). The prevalence of multidrug resistance in Salmonella isolates was also significantly higher in 15-25-days-old broilers in Group 1 (p = 0.016). Campylobacter isolates from 15-25-days-old broilers from Group 1 had significantly lower resistance to tetracycline (p < 0.001). The prevalence of multidrug resistance in Campylobacter isolates was also significantly lower on Group 1 (p = 0.005) ( Table 4).     The MIC 50 /MIC 90 values of E. coli, Salmonella, and Campylobacter isolates from the broilers with or without ENR treatment are illustrated in Tables S3-S5. The MIC 50 of ciprofloxacin and ENR were 8 µg/mL and 16 µg/mL against E. coli isolates from both 1-day-old broilers and retail meat in Group 1, respectively, which was higher than the 4 µg/mL and 8 µg/mL, respectively, in Group 2 (Table S4). Similarly, the MIC 50 of ENR against the Salmonella isolates from ENR-treated and untreated 15-25-days-old broilers were 0.5 and <0.25 µg/mL, respectively. The MIC 90 of ciprofloxacin against Salmonella isolates from ENR-treated and untreated 15-25-days-old broilers was 0.5 and 0.25 µg/mL, respectively (Table S5). There was no enhanced MIC 50 /MIC 90 value against Campylobacter isolates from broilers in Group 1 as compared with isolates from broilers in Group 2 (Table S6). On the contrary, the MIC 50 of ciprofloxacin and ENR were 16 µg/mL and 8 µg/mL against E. coli isolates from retail meat of broilers in Group 2, which was higher than that of 8 µg/mL and 4 µg/mL against the isolates from in Group 1, respectively. The MIC 90 of ENR were 32 µg/mL and 16 µg/mL against Campylobacter isolates from 15-25-days-old broilers and retail meat of broilers in Group 2 and 16 µg/mL and 8 µg/mL against broilers in Group 1, respectively.

Prevalence of PMQR and QRDR Mutations among the Selected E. coli, Salmonella, and Campylobacter Isolates from Untreated or ENR-Treated Broilers
The prevalence of PMQR and QRDR mutations in E. coli isolates (ENR, MIC ≥ 1) is displayed in Table 5. PMQR genes were not detected in any isolate; qnrB and qnrS1 genes were revealed in E. coli isolates from broilers with ENR treatment. The most frequent QRDR pattern in E. coli isolates (ENR MIC, 16-32 µg/mL) from broilers with ENR treatment was Ser-83-Leu and Asp-87-Asn in gyrA and Ser-80-Ile and Ser-83-Tyr in parC (8/28, 28.6%). The most common QRDR patterns observed in E. coli isolates (ENR MIC, 8-16 µg/mL) from broilers without ENR treatment were Ser-80-Ile and Ser-83-Tyr in parC (6/17, 35.3%). Alterations in the gyrA gene were frequently observed in 92.9% (26/28) and 76.5% (13/17) of E. coli isolates from broilers in Group 1 and Group 2, respectively. Alterations in the parC gene were noted in all E. coli isolates. Substitutions in the parE gene occurred less often, with 85.7% (24/28) and 82.4% (14/17) of isolates from broilers in Group 1 and Group 2, respectively, having wild type parC. Table 5. Prevalence of PMQR and amino acid changes in the gyrA, parC, and parE genes of E. coli isolates (n = 45) from Group 1 and Group 2 or the corresponding MICs of ENR a . The gyrA gene sequences of targeted isolates were submitted to the GenBank database and are available under the accession numbers OR047806-OR047961; the parC gene sequences of targeted isolates were submitted to the GenBank database and are available under the accession numbers OR047672-OR047805.
The prevalence of PMQR and QRDR mutations in Salmonella isolates (ENR, MIC ≥ 0.25) is indicated in Table 6. PMQR genes were not observed in any Salmonella isolate. The most common QRDR pattern detected in Salmonella isolates (ENR MIC, 0.25-1.00 µg/mL) from broilers in Group 1 was Asp-87-Gly in gyrA and Tyr-57-Ser in parC (28/50, 56.0%); the same QRDR pattern was observed in Salmonella isolates (ENR MIC, 0.25-0.50 µg/mL) from broilers in Group 2 (33/49, 67.3%). Changes in the gyrA gene were observed in 90.0% (45/50) and 98.0% (48/49) of Salmonella isolates from broilers in Group 1 and Group 2, respectively. In Salmonella isolates from broilers in Group 1 and Group 2, 92.0% (46/50) and 98.0% (48/49) of the changes in the parC gene were observed, respectively. Table 6. Prevalence of PMQR and amino acid changes in the gyrA and parC genes of Salmonella isolates (n = 99) from Group 1 and Group 2 and the corresponding MICs of ENR a . The prevalence of PMQR and QRDR mutations among Campylobacter isolates (ENR, MIC ≥ 4) is presented in Table 7. PMQR genes were not detected in any Campylobacter isolate. The only QRDR pattern of Thr-86-Ile in gyrA was observed in all Campylobacter isolates (ENR MIC, 4-32 µg/mL) from broilers in Group 1 and Group 2.

Discussion
The development of antimicrobial resistance indicates a clear link between the use of antimicrobials in veterinary medicine and animal production [56]. Administration of FQs such as ENR to poultry can lead to rapid selection of E. coli, Salmonella, and Campylobacter, which are less susceptible to such antimicrobials. In the present study, the interaction between the use of ENR and antimicrobial resistance in E. coli, Salmonella, and Campylobacter of broiler chickens was investigated under field conditions.
As expected, the isolation rate of Salmonella from samples of 15-25-days-old broilers was significantly lower (p < 0.05) in Group 1 than in Group 2 (56/485, 11.6%) ( Table 1). The isolation rate of Salmonella from broiler meat was also significantly lower (p < 0.05) in Group 1 (30/105, 28.6%) than in Group 2 (39/64, 60.9%). Therefore, the use of ENR was an important factor leading to a lower prevalence of Salmonella in both broiler farms and the retail market, which is consistent with a previous study [46]. In contrast, Campylobacter isolation rates from samples of 15-25-days-old broilers were significantly higher (p < 0.05) in Group 1 (83/1236, 6.7%) than in Group 2 (16/485, 3.3%). Similarly, Campylobacter isolation rates from broiler meat were significantly higher (p < 0.05) in Group 1 (48/105, 45.7%) than in Group 2 (17/64, 26.6%). Instead, the use of ENR increased Campylobacter prevalence in both broiler farms and the retail market. That could be because Campylobacter resistance to FQs has already reached a high level, and the use of ENR did not reduce prevalence as it did for Salmonella [55]. However, a previous study demonstrated a low prevalence of Campylobacter in retail chickens after the ENR ban [37]. In the present study, the use of ENR at the broiler stage had a different effect on the prevalence of Salmonella and Campylobacter in broilers and their meat products such that ENR use resulted in a high positive rate of Campylobacter but a low positive rate of Salmonella. The reasons for the above phenomena might be that the ENR resistance of Salmonella and Campylobacter isolates was different [47,55]. Moreover, low and high ENR resistance values were observed in Salmonella (Table 3) and Campylobacter isolates (Table 4), respectively. Therefore, the use of FQs in chickens rapidly leads to the selection of resistant Campylobacter organisms. However, FQ-resistant Campylobacter isolates are counter-selected under field conditions in the absence of selection pressure [25,29].
After administration of ENR, a significant change in resistance to ENR was observed in E. coli isolates from 15-25-days-old broilers. In the present study, we discovered a significant difference (p = 0.048) in resistance to ENR in E. coli isolates from 15-25-days-old broilers between Group 1 (89/101, 88.1%) and Group 2 (92/118, 78.0%) farms (Table 2). Therefore, ENR use may be an important factor in ENR resistance of E. coli in broiler farms. Similarly, antimicrobial resistance in E. coli is mainly drug-dependent [45]. Resistance to cefoxitin and tetracycline was significantly higher in E. coli isolates from broiler meat in Group 2 (45.0%; 100%) than in broilers in Group 1 (20.5%; 64.1%). This observation could be due to cross-contamination by E. coli from resistant carrier chickens slaughtered on the same day or contamination by resident flora with AMR in the slaughterhouse [47]. These data are consistent with previously published studies indicating that antimicrobial resistance in broilers not exposed to ENR showed changes over time [44,57].
There was no significant difference (p > 0.05) in resistance to FQs in Salmonella isolates from 15-25-days-old broilers between Group 1 (ciprofloxacin, 1.3%; ENR, 2.5%) and Group 2 (0%) farms (Table 3). This result could be due to low FQ resistance of Salmonella isolates from both Group 1 and Group 2, likely because of the short growth cycle of modern broilers, which does not allow sufficient time for resistance to develop, even after antimicrobial treatment. Salmonella isolates from 15-25-days-old broilers in Group 1 revealed a significant increase (p = 0.028) in intermediate resistance to ENR after ENR administration. A significant escalation (p < 0.05) in resistance to ampicillin, chloramphenicol, tetracycline, and trimethoprim/sulfamethoxazole was also observed in Salmonella isolates from 15-25-days-old broilers in Group 1. Moreover, the results of E. coli resistance to ampicillin, chloramphenicol, tetracycline, and trimethoprim/sulfamethoxazole were also consistent. The plausible explanations of the above phenomena were that the four antimicrobials were commonly used in broiler farms, and resistance to these antimicrobials was most frequently observed in isolates from chickens. In fact, selection pressure exerted by the antimicrobial agent contributed to the co-selection of this antimicrobial resistance pattern; the co-selection of resistance to more than one antimicrobial agent is a common feature of resistance acquired by horizontal gene transfer due to the genetic linkage of resistance genes [45,58,59].
There was no significant difference (p > 0.05) in resistance to FQs in Campylobacter isolates from broilers between Group 1 (78.4%) and Group 2 (100%) farms (Table 4). This may be due to the high FQ resistance of Campylobacter isolated from both the ENR-treated and untreated broilers, making it difficult to discern differences in FQ resistance between the two populations. A previous study demonstrated a low rate of ciprofloxacin resistance in Campylobacter in retail chickens after the ENR ban [36]. In other previous studies, therapeu-tic administration of ENR to mice did not result in FQ resistance in Campylobacter jejuni [60]. However, treatment of pigs with FQs selected and spread FQ-resistant Campylobacter in the house [61].
In general, PMQR genes confer low resistance to quinolones but may play an important role in the emergence of resistance mutations and the spread of antimicrobial resistance [62]. In the present study, the PMQR genes (qnrB and qnrS1) were detected in E. coli isolates from the broilers in Group 1, whereas these genes were absent in E. coli isolates from the broilers in Group 2 (Table 5). Accordingly, the qnrS1 gene was uncovered in E. coli isolates from chicken meat produced in integrated broiler operations in Korea [63]. In the present study, more amino-acid mutations in the gyrA, parC, and parE genes were discovered in broiler isolates from Group 1 than in isolates from Group 2. Thus, the use of ENR resulted in the emergence of PMQR genes and higher resistance to ENR with more mutations in QRDR genes.
Although PMQR has been extensively studied, QRDR mutations seem to represent the main mechanism of quinolone resistance in animal isolates [64]. Moreover, PMQR was commonly detected in Enterobacteriaceae, particularly in E. coli, and the prevalence of PMQRs in Salmonella remains extremely low [65]. In the present study, no PMQR gene was realized in all Salmonella (n = 99) and Campylobacter (n = 76) isolates tested (Tables 6 and 7). The most common QRDR pattern, an exchange of a single amino acid in both gyrA (Asp-87-Gly) and parC (Tyr-57-Ser), was identical in Salmonella isolates from broilers in Group 1 and Group 2 ( Table 6), and similar QRDR patterns were noted, except for the Ser-83-Ala exchange in the gyrA gene in Salmonella isolates from Group 1. Only one QRDR pattern (Thr-86-Ile in gyrA) was revealed in the Campylobacter isolates from both Group 1 and Group 2 ( Table 7). This is consistent with our previous findings that the Thr-86-Ile substitution in the gyrA was the primary contributor to the high-level quinolone resistance in Campylobacter isolates from duck meats [66]. Therefore, the use of ENR may not lead to the emergence of PMQR genes and QRDR genes with amino acid exchange in Salmonella and Campylobacter isolates [67].

Conclusions
In conclusion, ENR use in broiler farms is an important factor in reducing the prevalence of Salmonella but not Campylobacter, as well as causing ENR resistance in E. coli but not in Salmonella or Campylobacter. ENR use in broiler farms may lead to the emergence of PMQR genes and high levels of QRDR mutations and ENR resistance in E. coli and less so in Salmonella and Campylobacter. Understanding the relationship between antimicrobial use in food-producing animals and the emergence of resistant commensal bacteria may facilitate the prudent use of antimicrobials in these animals. Therefore, because of its selective effect, ENR must be used judiciously.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods12112239/s1, Table S1: Sampling of feces and isolation information. Table S2: Specific guideline used for each antimicrobial substance; Table S3: Strains chosen as specific positive control with sequence accession numbers; Table S4: Distribution of the MIC 50 /MIC 90 values of (fluoro)quinolones among E. coli isolates from broiler chickens with or without ENR treatment; Table S5: Distribution of the MIC 50 /MIC 90 values of (fluoro)quinolones among Salmonella isolates from broiler chickens with or without ENR treatment; Table S6: Distribution of the MIC 50 /MIC 90 values of (fluoro)quinolones among Campylobacter isolates from broiler chickens with or without ENR treatment.

Conflicts of Interest:
Author Hyung-Kwan Jang and Min Kang were employed by the company of Bio Disease Control (BIOD) Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.