Selecting for development of fluoroquinolone resistance in a Campylobacter jejuni strain 81116 in chickens using various enrofloxacin treatment protocols
Anne Ridley, Veterinary Laboratories Agency Weybridge, Woodham Lane, Surrey KT15 3NB, UK. E-mail: email@example.com
Aims: To determine the effect of various enrofloxacin dose regimes on the colonization and selection of resistance in Campylobacter jejuni strain 81116P in experimentally colonized chickens.
Methods and Results: Two experiments were undertaken, in which 14-day-old chickens were colonized with 1 × 107–1 × 109 CFU g−1Camp. jejuni strain 81116P and then treated with enrofloxacin at 12–500 ppm in drinking water for various times. Caecal colonization levels were determined at various time-points after start-of-treatment, and the susceptibility of recovered isolates to ciprofloxacin was monitored. Resistance was indicated by growth on agar containing 4 μg ml−1 ciprofloxacin, MICs of 16 μg ml−1 and the Thr86Ile mutation in gyrA. Enrofloxacin at doses of 12–250 ppm reduced Camp. jejuni colonization over the first 48–72 h after start-of-treatment. The degree of reduction in colonization was dose, but not treatment time, dependent. In all cases, maximal colonization was re-established within 4–6 days. Fluoroquinolone-resistant organisms were recoverable within 48 h of start-of-treatment; after a further 24 h all recovered isolates were resistant. In contrast, a dose of 500 ppm enrofloxacin reduced colonization to undetectable levels within 48 h, and the treated birds remained Campylobacter negative throughout the remaining experimental period. By high pressure liquid chromatography, for all doses, the maximum concentrations of enrofloxacin and ciprofloxacin in the caecal contents were detected at the point of treatment completion. Thereafter, levels declined to undetectable by 7 days post-treatment withdrawal.
Conclusions: In a model using chickens maximally colonized with Camp. jejuni 81116P, treatment with enrofloxacin, at doses of 12–250 ppm in drinking water, enables the selection, and clonal expansion, of fluoroquinolone-resistant organisms. However, this is preventable by treatment with 500 ppm of enrofloxacin.
Significance and impact of the study: Treatment of chickens with enrofloxacin selects for resistance in Camp. jejuni in highly pre-colonized birds. However, a dose of 500 ppm enrofloxacin prevented the selection of resistant campylobacters.
Campylobacteriosis is now the major cause of food-poisoning in Europe and therefore of significant public health concern (Bronzwaer et al. 2009). Epidemiological studies indicate that most infections are attributable to the handling and consumption of chicken meat, and control throughout the food chain has been recommended.
Antimicrobial resistance in campylobacters, especially fluoroquinolone resistance, is also considered a public health concern in Europe (http://www.efsa.europa.eu/EFSA/efsa_locale-1178620753812_1178716893972.htm). Such resistance is typically associated with a single point mutation in the gyrA gene (Engberg et al. 2001). Fluoroquinolone-resistant Camp. jejuni infections have also been attributed to the veterinary use of these antimicrobials in poultry (http://www.efsa.europa.eu/cs/BlobServer/DocumentSet/aarestrup_presentation.pdf?ssbinary=true). Campylobacter is a common component of the avian gut flora with up to 83% of broiler flocks from individual EU member states colonized by the point of slaughter (EFSA 2007). Colonization is asymptomatic for the bird. Flocks may acquire fluoroquinolone-resistant campylobacters from the environment. Resistance may also be selected in an already Campylobacter-colonized flock during enrofloxacin treatment for clinical diseases in broiler production, such as colibacilliosis, pasteurellosis and mycoplasmosis. However, such treatments would normally be necessary within the first 3 weeks of age, which is generally prior to the period Campylobacter colonization is detected in a typical commercial flock. The therapeutic regimen recommended for broiler treatment of 50 ppm of enrofloxacin for 3–5 days (10 mg enrofloxacin kg−1 bodyweight) is defined on the basis of efficacy against such diseases. However, experimental studies (McDermott et al. 2002; van Boven et al. 2003) indicate that such therapeutic treatment of broilers results in the selection of resistance in Campylobacter to fluoroquinolones and subsequent clonal expansion of resistant organisms. Nevertheless, because such clinical diseases can cause significant flock losses and generate welfare problems, there is a need to maintain an adequate armoury of tools for disease control in the poultry industry while, of course, ensuring public health safety.
In this study, we have investigated the effect of enrofloxacin dose and length of treatment on the selection of resistance in Camp. jejuni colonizing the chicken gut using standard models of experimental colonization and previously published methods of detecting fluoroquinolone resistance in Campylobacter (McDermott et al. 2002; Randall et al. 2003). In addition, we used high pressure liquid chromatography (HPLC) to accurately determine the concentration of enrofloxacin and its metabolite ciprofloxacin in the chicken’s intestinal contents over time.
Materials and methods
A high-colonizing strain Camp. jejuni 81116P, generated by passage of strain Camp. jejuni 81116 through chickens (Cawthraw et al. 1996), was used for the in vivo studies. This strain had a minimal inhibitory concentration (MIC) ciprofloxacin of 0·125 μg ml−1. Control Campylobacter strains used for MIC studies were Camp. jejuni NCTC 11351 (ATCC 33560) and Campylobacter coli NCTC 11366 (ATCC 33559). These strains have been used as standards in other similar studies (Thwaites and Frost 1999).
In vivo studies
In Experiment 1, 1-day-of-age White Leghorn chicks (specific-pathogen-free (SPF) Lohmann, Germany; n = 88) that had been confirmed as Campylobacter free using cloacal swabbing were dosed at 1-day-of-age by gastric intubation with 3·7 × 106 CFU of Camp. jejuni 81116P. Caecal colonization (>1 × 107 CFU g−1) was confirmed in four of the birds at 14-days-of-age. The remaining birds were then divided into three groups (n = 28 per group) and treated with either 50, 125 or 250 ppm of enrofloxacin (Baytril®) in drinking water (Baytril 10% Oral Solution, Bayer plc, Newbury, UK) for 5, 3 or 1 days, respectively. Four birds were killed from each group at 1, 2, 3, 7, 10, 21 and 35 days following the start of antimicrobial treatment. These time-points were selected on the basis of preliminary experiments which indicated that resistance was observed in strains by 5 days post-treatment.
In Experiment 2, 1-day-of-age chicks (n = 220) that had been confirmed as campylobacter free using cloacal swabbing were orally dosed at with 1·6 × 106 CFU Camp. jejuni 81116P. Colonization (>1 × 107 CFU g−1) was confirmed in four of the birds at 14-days-of-age. The remaining chicks were then randomly separated into six groups (n = 36 per group). The birds were treated for 3 days with 12, 25, 50, 125, 250 or 500 ppm of enrofloxacin in drinking water. Four birds from each group were killed at 2 and 6 h, and at 1, 2, 3, 7, 10, 15 and 35 days post-start of antimicrobial treatment.
Recovery of Camp. jejuni isolates from in vivo model
Caecal contents were collected postmortem to determine colonization levels and for analysis of enrofloxacin and its metabolite ciprofloxacin by HPLC. Colonization levels per gram of caecal contents were determined for each of the four birds sacrificed at each time-point as described previously (Wassenaar et al. 1993) using the selective media 5% (w/v) blood agar containing Skirrow’s antibiotics (Oxoid, Basingstoke, UK) with 1% actidione (BASAC). The use of such selective media was necessary to recover campylobacters from the complex bacterial and fungal flora of caecal contents. Preliminary studies (data not shown) indicated that this medium had no effect on the growth of Camp. jejuni 81116P.
Measurement of antimicrobial resistance
The currently approved CLSI breakpoint for the human-use fluoroquinolone, ciprofloxacin, is 4 μg ml−1 (CLSI 2006). However, at the time the experiments were conducted no breakpoint had been approved, and we therefore started with R > 1 μg ml−1, which was consistent with the criteria used by the UK Health Protection Agency (HPA) (Thwaites and Frost 1999).
To identify and enumerate ciprofloxacin-resistant campylobacters in caecal contents, plates with clearly defined colonies (about 100 colonies per plate) were replica plated onto BASAC agar containing 0, 1 or 4 μg ml−1 ciprofloxacin (Bayer Plc). The choice of agar reflected that used in the initial isolation, and this approach has been used previously for the investigation into resistance in other enteric pathogens in poultry (Randall et al. 2006). The concentrations of antimicrobial selected accounted for the CLSI and UK HPA breakpoints as well as several breakpoints previously published (for example van Boven et al. 2003; Luo et al. 2003; Taylor et al. 2008).
In addition, the MICs for susceptibility against enrofloxacin and ciprofloxacin of 545 isolates (235 from Experiment 1, and 310 from Experiment 2), randomly selected from the original caecal culture plates containing discrete colonies (100–200), were compared to five randomly selected isolates from the plates containing the challenge strain. The MIC was determined using an agar doubling dilution method as previously described (Randall et al. 2003) with cut-off concentrations of 1 and 4 μg ml−1 used for reasons indicated previously.
The presence of the Thr86Ile mutation indicative of ciprofloxacin resistance was also investigated in a selection of isolates (n = 28) by sequence analysis of the quinolone resistance-determining region (QRDR) as previously described (Wilson et al. 2000).
Pulsed Field Gel Electrophoresis (PFGE)
Six discrete colonies were randomly selected for PFGE analysis from the original colonization plates of each of the treatment and control groups. Camp. jejuni chromosomal DNA was prepared and digested as previously described (On et al. 1998). Analysis of banding patterns was conducted to confirm recovered isolates were congruent with the original dosing strain.
Analysis of enrofloxacin and ciprofloxacin in caecal contents
Caecal content samples from Experiment 2 were prepared and analysed by HPLC for the enrofloxacin and ciprofloxacin recovered from each sample as described previously (Randall et al. 2006). The determined levels of enrofloxacin and ciprofloxacin were corrected for recovery by taking into account the extraction efficiency (rate of recovery) of the analytical method.
Spontaneous mutation rate for mutations conferring fluoroquinolone resistance in Camp. jejuni
A 48 h culture of Camp. jejuni 81116P was resuspended in phosphate-buffered saline (PBS) to give c. 1 × 109 CFU ml−1. The spontaneous mutation rate was determined by serially diluting the suspension and then plating 10 replicates onto BASAC and BASAC containing 4 μg ml−1 ciprofloxacin.
In Experiment 1, the caecal colonization level of Campylobacter was, on average, 7·8 × 107 CFU g−1 at the point of initiation of enrofloxacin treatment. This is about the maximal colonization level, and previous studies using strain 81116 have shown that such colonization levels are consistently maintained for up to 8 weeks (Newell and Wagenaar 2000). By 24 h following commencement of enrofloxacin treatment, colonization levels, regardless of dose, dropped by >4 log10 (Fig. 1). There was no growth on either 1 or 4 μg ml−1 ciprofloxacin-containing media, and all colonies recovered and tested by MIC (n = 30) at this time-point were susceptible to ciprofloxacin, with MIC values of 0·125 μg ml−1 consistent with that of the challenge strain.
Colonization levels in all groups dropped further at 48 h after start-of-treatment. However, at this time-point, 68% (n = 147/216), 76% (n = 222/292) and 71% (n = 68/96) of recovered isolates tested from the 50, 125 and 250 ppm treated groups, respectively, grew in the presence of 4 μg ml−1 ciprofloxacin. At 72 h after start-of-treatment, the proportion of recovered isolates growing at 4 μg ml−1 ciprofloxacin increased to at least 93% in each group. A random selection of these isolates (n = 10) was further investigated for MIC. All the isolates tested had a MIC to ciprofloxacin of 16 μg ml−1. At 7, 10, 21 and 35 days following the start-of-treatment, all isolates (n = 2468) recovered from each group grew in the presence of 4 μg ml−1 ciprofloxacin.
In Experiment 2, chicks colonized with Camp. jejuni were treated for 3 days with 12–500 ppm of enrofloxacin. For all groups, colonization levels dropped rapidly during the first 2 days after start-of-treatment (Fig. 1). However, by 4 days after the completion of treatment, colonization levels had returned to >107 CFU g−1 in all but the 500 ppm treatment group. Interestingly, the outcomes of treatment with 50, 125 or 250 ppm were independent of the length of that treatment. All caecal isolates (n = 739) tested after 24 h of treatment were susceptible to fluoroquinolones, and the MICs of isolates at this time were consistent with the original challenge strain (0·125 μg ml−1). However, within 48 h of start-of-treatment >99% of the recovered isolates from the 125 and 250 ppm groups (n = 146) grew in the presence of 4 mg l−1 ciprofloxacin, and all isolates (n = 20) examined further gave MIC levels of 16 μg ml−1. In contrast, only 50% of recovered isolates (n = 317) from the 25 and 50 ppm groups and 34% (n = 210) from the 12 ppm group displayed reduced susceptibility. MICs of the colonies from each of the susceptible and resistant isolates indicated MICs of 0·125 and 16 μg ml−1, respectively. At 72 h after start-of-treatment, 82% isolates (n = 171) from the chickens treated with 12 ppm were resistant. No susceptible Campylobacter isolates were recovered from any subsequent time-points up to 35 days from any of the treated groups. All isolates recovered and tested (n = 5116) after 72 h from the start-of-treatment had a MIC of 16 μg ml−1.
The colonization levels in chickens treated with 500 ppm enrofloxacin for 3 days decreased substantially within 6 h of treatment to ≤1·2 × 105 CFU g−1 caecal contents and continued to fall over the following 12 h (Fig. 1). At this time-point, all recovered campylobacters (n = 272) remained susceptible. No campylobacters were recovered from this group after 48 h of enrofloxacin treatment.
There was substantial variation in the kinetics of the concentration of enrofloxacin in the caecal contents. For all doses, there was a rapid increase over the first 24 h of treatment. Thereafter some doses, such as 50 ppm, plateaued, while others, such as 500 ppm, continued to rise until treatment withdrawal at day 3 (Fig. 2). The increasing levels of ciprofloxacin in the caecal contents reflected those of enrofloxacin, albeit substantially lower. Following treatment withdrawal, levels of both fluoroquinolones dropped rapidly and were undetectable (<0·05 mg kg−1) by day 10 post-termination of treatment.
The PFGE profiles of all recovered isolates investigated (n = 6) (from each group in both experiments) were indistinguishable from the challenge strain. Resistance induction was consistent with the acquisition of the Thr86Ile mutation as detected by sequence analysis in 28 randomly selected isolates, all with resistance to ciprofloxacin as indicated by an MIC value of 16 μg ml−1. On the basis of serial dilution experiments, the spontaneous mutation rate for this mutation in Camp. jejuni 81116P was estimated to be 1–2 × 10−7.
Enrofloxacin is approved for therapeutic use in the control of bacterial infections in broilers in Europe. However, the contribution of the veterinary use of enrofloxacin to the development of resistance and increasing trend of resistance in animal and human populations in potential pathogens in the food chain remains a concern (EFSA 2009). Although there is increasing new evidence from analyses of case–control studies to indicate that infection of humans with fluorquinolone-resistant campylobacters does not result in more severe or prolonged illness in humans (Wassenaar et al. 2007; Evans et al. 2009), studies following the enrofloxacin ban in the United States have shown persistence of fluoroquinolone resistance in campylobacters recovered from poultry products, a few years after cessation of the ban (Price et al. 2007; Nannapaneni et al. 2009).
The recommended treatment with enrofloxacin, which equates to about 50 ppm for 5 days, is effective in controlling Escherichia coli (van Boven et al. 2003), but this regimen can rapidly select for, and enables, clonal expansion of fluoroquinolone resistance in Camp. jejuni. Resistant Camp. jejuni then persist in the chicken gut, contaminate poultry carcasses during processing and, thereby, may constitute a risk to public health. In this study, we have demonstrated that, although various alternative enrofloxacin treatment regimens also select for resistance, treatment of 500 ppm for 3 days terminated Camp. jejuni colonization in the gut.
The dosing of chickens with 12–500 ppm enrofloxacin led to a rapid increase in concentration of the drug in the caecal contents within 24 h of start-of-treatment. In addition to enrofloxacin, its major metabolite ciprofloxacin, which is widely used in human medicine, was also recovered by HPLC. Not surprisingly, as the levels of administered drug increased, the caecal concentrations of both enrofloxacin and ciprofloxacin also increased. Consequently, this will have an impact on the total concentration of fluoroquinolones found in the gut. By HPLC, levels were maximal at the completion of treatment and declined rapidly thereafter becoming undetectable 7 days after withdrawal of treatment. Similar observations have been made previously (Randall et al. 2006) during studies on Salmonella. This reduction is largely attributable to the short half-life (4·4–5·8 h) of enrofloxacin in broilers (Knoll et al. 1999).
Although the preliminary identification of resistance in our study was defined as growth in the presence of 1 μg ml−1 ciprofloxacin all resistant isolates recovered and further tested were also resistant at the CLSI breakpoint of 4 μg ml−1 (CLSI 2006), each having an MIC value of 16 μg ml−1. Thus, the resistant isolates recovered from treated chickens were resistant to ciprofloxacin. Moreover, the susceptible isolates recovered and tested all had the same MIC as the original challenge strain of 81116P. The presence of the Thr86Ile mutation determined by the sequencing of the QRDR of the 28 resistant isolates taken from the different groups of birds was the only mutation observed. Although a mutation in Asp90 has been indicated for Camp. jejuni in other studies, this was only observed in resistant isolates with MIC values of ≤8 μg ml−1 (Payot et al. 2002; Takahashi et al. 2005).
All doses of enrofloxacin tested in this chicken model caused a reduction in Campylobacter colonization of between 3 and 5 log10. The extent and persistence of this reduction was dose dependent. Despite the evidence that for all treatments the fluoroquinolone levels in the caecal contents exceeded the MIC of Camp. jejuni 81116P within 6 h this level of antibiotic appears insufficient to clear the colonization.
At doses of 12–250 ppm, the decline in colonization levels stabilized by 3–4 days after the start-of-treatment and by 4–6 days later had returned to the original maximal level (about 1 × 108 CFU g−1 caecal contents). This recovery of colonization was concomitant with the selection of fluoroquinolone resistance. The proportion of resistant organisms was both dose- and time dependant, indicating that the selection of resistant campylobacters was ongoing throughout the treatment period. It has been suggested (Zhang et al. 2006) that fluoroquinolone resistance in Campylobacter has no ecological fitness cost and may even be advantageous. Once resistance was established in our studies, it persisted until slaughter which was at least 35 days post-treatment.
On the basis of the serial dilution studies, it was estimated that the spontaneous mutation rate for Camp. jejuni 81116P in vitro was between 1 and 2 × 10−7 which is consistent with findings of others (Zhang et al. 2003). Although this estimation cannot account for the competitive pressures of the intestinal habitat, it was used to give an indication of the presence of such mutants in the dose. Given that the caecal contents of a chicken can contain over 109 CFU g−1 of Campylobacter (Cawthraw et al. 1996), it seems likely that up to 103 CFU g−1 of naturally occurring resistant Camp. jejuni may be present in each chicken gut. It appears to be the clonal expansion of these organisms under the selective pressure of the enrofloxacin treatment that enables the repopulation of the chicken gut. However, treatment with 500 ppm enrofloxacin reduced Camp. jejuni colonization to undetectable levels within 48 h and, unlike Salmonella (Randall et al. 2006), there was no evidence of recolonization from previously shed organisms. Interestingly, at this dose level the concentration of enrofloxacin in the caecal contents within 2 h of treatment was nearly fivefold greater (at 23 μg g−1 caecal contents) than that observed with a dose of 250 ppm. These results suggest that achieving a high initial concentration in the chicken gut is essential to eliminate all naturally susceptible and -resistant campylobacters. With such an approach, the risk of transfer to humans via the food chain of resistant Camp. jejuni selected by enrofloxacin may be reduced.
In conclusion, these studies confirm that treatment with enrofloxacin selects for fluoroquinolone resistance in Camp. jejuni in already-colonized chickens. Enrofloxacin is an important therapeutic drug for the poultry industry. The outcomes of bans in its use for this purpose in some countries are as yet unclear (http://www.efsa.europa.eu/EFSA/efsa_locale-1178620753812_1178716893972.htm), although recent evidence suggests that such bans may have a limited impact on ciprofloxacin resistance in Campylobacter (Barzilay et al. 2009). Nevertheless, modification of application protocols might be one strategic approach to reduce or prevent selection of resistance in campylobacters. One strategy would be to only use enrofloxacin in Campylobacter-negative flocks. Fluoroquinolones are administered in a relatively small percentage of all poultry flocks for highly pathogenic E. coli conditions, such as clinical outbreaks of collibacillosis, which can occur during the first few weeks of life. Most flocks are Campylobacter negative for the first 2–3 weeks of life, and improved biosecurity appears to be extending this lag phase in some countries (Newell and Wagenaar 2000), thereby reducing the risk of selecting resistant mutants, so this may be a realistic strategy. However, the need for a rapid on-farm test for Campylobacter is highlighted. In addition, our results indicate that high levels (500 ppm) of enrofloxacin for short treatment periods will provide sufficient bioavailability of active drug in the caecal contents to prevent the colonization of the intestinal tract even with highly colonizing Camp. jejuni strains.
There were no observable adverse effects on chickens administered such high doses of enrofloxacin, and this is consistent with toxicity data for this drug. Clearly, further work is required to determine the effects of high doses of enrofloxacin on different strains of Campylobacter as well as on the other gut microflora and to determine the risk of recolonization of the treated birds following completion of treatment. Collaboration of researchers with the industry is required to assess the validity of such an approach under field conditions.
This work was funded by the Department for Environment, Food and Rural Affairs, Great Britain. We acknowledge the advice of Dr Luke Randall for technical expertise.