Effect of Subtherapeutic Administration of Antibiotics on the Prevalence of Antibiotic-Resistant Escherichia coli Bacteria in Feedlot Cattle^▿

Animals and treatments.

Three hundred crossbred steers (initial body weight, 198 ± 20 kg) were blocked by weight and then randomly assigned to 30 pens at the Lethbridge Research Centre feedlot. The steers were brought directly to the feedlot from a common range (Deseret Ranches, Raymond, Alberta, Canada) and received no antibiotics prior to the initiation of the experiment. In addition, AGP had not been administered to cattle in the feedlot prior to the present study. Five pens (10 steers per pen) were assigned to each of six treatments: (i) control, no antibiotics; (ii) chlortetracycline and sulfamethazine (each at 44 ppm; fed as Aureo S-700 G; Alpharma, Inc., Bridgewater, NJ; treatment denoted TET-SUL); (iii) chlortetracycline (11 ppm; fed as Aureomycin-100 G; Alpharma; treatment denoted TET); (iv) monensin (25 ppm, fed as Rumensin, Elanco Animal Health, Calgary, Alberta, Canada; treatment denoted MON); (v) tylosin phosphate (11 ppm, fed as Tylan, Elanco Animal Health; treatment denoted TYL); and (vi) virginiamycin (31 ppm, fed as V-Max, Pfizer Animal Health, New York, NY; treatment denoted VIR). Other than virginiamycin, these antibiotics and concentrations were selected on the basis of their commonality of use in the Canadian feedlot industry. Virginiamycin was included in the study because it is not registered for use in Canada and, as a result, neither calves nor their dams would have had prior exposure to this antibiotic. Adjacent pens in the feedlot were supplied by a common watering bowl, but assignment of treatments to pens was arranged so that only cattle in the same treatment group drank from the same bowl.

Steers were fed diets typical of feedlots in Western Canada. For the first 115 days in the feedlot, steers were fed a silage-based, background diet consisting of 70% barley silage, 25% barley grain, and 5% mineral and vitamin supplement (dry matter basis) (Fig. 1). Steers were then adapted to a barley grain-based finishing diet consisting of 85% barley, 10% barley silage, and 5% supplement (dry matter basis) over a 21-day transition period and were then maintained on this diet for an additional 179 days. The present study was approved by the Lethbridge Research Centre Animal Care Committee, with all cattle being cared for according to the guidelines of the Canadian Council on Animal Care (10).

The steers were fed once daily in a manner that ensured that all feed given was consumed each day. To avoid cross-contamination of feed, antibiotics were mixed with 5 kg of supplement and manually spread over the surface of the feed in each trough, immediately after delivery of the feed. The feed troughs were of sufficient length that all steers in a pen could attend at the same time. Steers assigned to the control treatment had no access to medicated feed at any time during the experiment.

Antibiotics were introduced into the diets 17 days after the steers’ arrival at the feedlot and were included in the silage-based diet for 61 days. Antimicrobial supplementation was discontinued for 86 days, and then antibiotics were reintroduced for an additional 42-day period during the feeding of the grain-based diet (Fig. 1). The removal of antibiotics was included in the study to determine whether withdrawal could be used as a management practice to reduce the prevalence of antibiotic-resistant E. coli should it arise. In addition, in some feedlots, withdrawal of medicated feed is used as a standard practice in growth performance and disease management.

Sample collection and microbiology.

Collection of fecal material from each animal was initiated on 11 sampling days (the total sampling time spanned 2 days) during the experiment (Fig. 1). Two clinical swabs (Starswap, Starplex Scientific, Ontario, Canada) per steer were inserted approximately 5 cm into the rectum and rotated until covered with a uniform amount of feces. Swabs were placed individually in sterile, capped test tubes and transported immediately to the laboratory (0.5 km). All samples were processed within 4 h of collection. At the lab, swabs were broken and mixed with 750 μl of brain heart infusion broth (BHI; Difco/Becton Dickinson, Sparks, MD) amended with glycerol (20% [vol/vol]). The tubes containing swabs and medium were vortex mixed thoroughly, and the slurries from both tubes for each steer were combined.

Fecal slurry from each steer was plated onto MacConkey agar (Difco/Becton Dickinson) containing no antibiotics (MAC) or onto MAC amended with ampicillin (50 μg ml⁻¹; MAC+AMP), tetracycline hydrochloride (4 μg ml⁻¹; MAC+TET), gentamicin (2 μg ml⁻¹; MAC+GEN), or ciprofloxacin (2 μg ml⁻¹; MAC+CIP). Aliquots of 10 μl of slurry were plated onto each of the selective media. For plating onto nonselective agar (MAC), 100 μl of slurry was diluted with 2.9 ml of BHI broth, and 50-μl volumes were plated. Plates were incubated for 24 h at 39°C before lactose-positive colonies were counted, and those plates on which no colonies were evident were incubated for an additional 24 h to confirm the absence of bacteria. Colony counts from each of the selective media were compared against counts from nonselective (MAC) plates to estimate percentages of presumed antibiotic resistance by and across sampling days, treatments (antibiotic-supplemented feeding groups), and diet formulations (silage-based versus grain-based).

Concentrations of antibiotics used in the MAC+TET, MAC+GEN, and MAC+CIP plates were deliberately set below the standards for defining resistance described by the National Committee for Clinical Laboratory Standards (NCCLS) (40), now the Clinical and Laboratory Standards Institute (CLSI) (13), to maximize the likelihood of isolating resistant E. coli. The concentration of ampicillin selected for the MAC+AMP plates exceeded NCCLS standards (50 versus 32 μg ml⁻¹) but was chosen to prevent the overgrowth of plates from interfering with counting colonies. The antibiotics selected for study were chosen on the basis of commonality of use between cattle and humans, in the case of tetracycline, or their importance in treating human infections, in the case of ampicillin, gentamicin, and ciprofloxacin.

Two colonies were picked from each nonselective plate, and one or two were picked from each of the antibiotic-selective plates. Selected colonies were streaked onto Trypticase soy agar (Difco) and incubated at 39°C for 24 h. Colonies on the Trypticase soy agar plates were transferred to BHI and glycerol and stored at −80°C for further analysis (below). Also, 300 of these isolates (75 each from the MAC, MAC+AMP, MAC+TET, and MAC+GEN plates) were subsampled for confirmation of their identity as E. coli on the basis of lactose utilization, methyl red reaction, β-galactosidase and β-glucuronidase activities, and fatty acid methyl ester analysis (18).

Antimicrobial susceptibility testing.

Susceptibility to ampicillin (2, 4, 8, 16, and 32 μg ml⁻¹), gentamicin (0.5, 1, 2, 4, 8, and 16 μg ml⁻¹), and tetracycline (1, 2, 4, 8, and 16 μg ml⁻¹), as well as screening for reduced susceptibility to ceftazidime (0.5, 1, and 2 μg ml⁻¹) and cefpodoxime (0.5, 1, and 2 μg ml⁻¹), were analyzed by using the agar dilution method of CLSI (40) on 7,166 E. coli isolated from MAC (n = 3,512), MAC+TET (n = 2,205), MAC+AMP (n = 637), and MAC+GEN (n = 812) plates. Interpretations of isolates resistant to ampicillin, gentamicin, and tetracycline were determined by using antibiotic breakpoint concentrations on the basis of NCCLS (41) recommendations. Ceftazidime and cefpodoxime were included to screen for candidate extended-spectrum beta-lactamase (ESBL)-producing strains of E. coli (41). The reference strains E. coli (ATCC 25922), Enterococcus faecalis (ATCC 29212), Pseudomonas aeruginosa (ATCC 27853), and Staphylococcus aureus (ATCC 29213) were included with each susceptibility determination to confirm proper function of the assay.

PFGE.

After preliminary examination of prevalence data and shedding patterns, each of the ampicillin-resistant isolates collected from the control steers on days 15, 113, and 246 (80 isolates total) were selected from storage and subtyped by pulsed-field gel electrophoresis (PFGE) separation of XbaI-digested genomic DNA according to the Centers for Disease Control and Prevention (12) 1-day standardized laboratory protocol for molecular subtyping of E. coli. E. coli E318N (kindly made available by A. Borczyk, Enteric Reference Laboratory, Ministry of Health, Toronto, Ontario, Canada) was included in each PFGE gel as a control.

Definitions and statistical analyses.

Bacteria isolated from selective antibiotic plates were categorized as presumed resistant to the respective antibiotic. The percentage of E. coli resistant to the antibiotic concentrations in selective plates was calculated for each antibiotic, with consideration for dilutions, as follows: (number of colonies on the antibiotic-supplemented MAC plates/total colonies counted on nonselective MAC) × 100. Bacterial population estimates were expressed as the log₁₀ CFU swab⁻¹. The prevalence of steers shedding presumed resistant E. coli from the selective plates was calculated for each sampling day and feeding group as the number of steers from which antibiotic-resistant E. coli were isolated (on MAC+antibiotic plates), expressed as a percentage of the total number of steers in that feeding group. Overall, the prevalence was estimated as the percentage of steers from a particular feeding group that were positive for E. coli on the antibiotic-containing plates at least once over the course of the entire study.

Data were analyzed by using the “Proc Mixed” procedure of the SAS (52). Because antimicrobial agents were applied to feed provided to the pens of cattle, the pen was considered the experimental unit. The model for analyzing E. coli CFU swab⁻¹ included the fixed effects of diet (silage- versus grain-based), treatment (antibiotic regimen), and the interaction between diet and treatment. The random effect of a pen nested within treatment was used. The repeated statement was applied to the day of sampling, using the pen nested within treatment as the subject. Various error structures were tested, and the one giving the lowest Akaike information criterion was chosen for analysis. The effect of diet was analyzed by comparing two data sets generated by combining days on which animals consumed a silage-based diet (days 15, 36, 57, 71, 92, and 133) or a grain-based diet (days 134, 183, 204, 225, and 246). Both the silage- and the grain-based data sets contained one sampling point prior to administration of antibiotics (day 15 and day 134, respectively), at least two points from the period during which animals were administered antibiotics (days 36, 57, and 71 and days 183 and 204), and two sampling points that followed withdrawal of antibiotics from the diets (days 92 and 113 and days 225 and 246). For prevalence of resistant E. coli isolated from animals, the same statistical model was used, but with the fixed effects of antimicrobial treatment, day, and the interaction of antimicrobial treatment and day. In addition, analysis by diet was conducted. Where appropriate, the LSMEANS function in SAS was used to identify statistical differences.

Isolation of E. coli strains presumed resistant.

In total, approximately 3,300 fecal samples were processed for isolation of E. coli on plates supplemented with (MAC+AMP, MAC+CIP, MAC+GEN, and MAC+TET) or without (MAC) antibiotics. No isolates were recovered from any of the MAC+CIP plates. Of the 300 isolates tested for E. coli confirmation, more than 99% were identified as E. coli by biochemical and fatty acid methyl ester analyses (data not shown). The prevalence of isolation and resistant population data were analyzed as E. coli resistant to, and capable of growth on, the concentrations of antimicrobials in the selective plates. These E. coli strains were presumed resistant to antimicrobial agents in the selective plates.

(i) Prevalence of isolation.

On the first sampling day after arrival at the feedlot (day 15), more than 40% of the steers shed tetracycline-resistant E. coli (Fig. 2a), whereas fewer than 30% were shedding ampicillin- or gentamicin-resistant E. coli (Fig. 2b and c). With the exception of the TET-SUL feeding group, including antibiotics in the silage-based diet did not increase the prevalence of steers shedding tetracycline-resistant E. coli; in fact, the prevalence was lower (P = 0.05) on day 36, when diets other than the control treatment were supplemented with antibiotics, than on day 15, when no antibiotics were fed (Fig. 2a). In contrast, feeding TET-SUL resulted in a dramatic increase (P = 0.001) in the prevalence of tetracycline-resistant E. coli shedders. The prevalence of steers shedding tetracycline-resistant E. coli decreased slightly (P = 0.08) upon removal of TET-SUL from the silage-based diet (days 92 and 113). However, the high level of shedders from the TET-SUL treatment was generally sustained throughout the experiment. Withdrawing the other antibiotics did not affect the proportion of steers shedding tetracycline-resistant E. coli either on silage-based or on grain-based diets. Although the number of animals shedding tetracycline-resistant E. coli increased beyond levels recorded for day 134, when animals were fed grain-based diets without antibiotics, this effect was not attributable to antibiotic supplementation (days 164 to 206) during this feeding period (Fig. 2a). When animals were fed the grain-based diets, the prevalence of animals shedding resistant E. coli for animals treated with antibiotics was similar to that of the control group, with the exception of the TET-SUL treatment group. Furthermore, in comparing prevalences when animals were fed the silage- versus grain-based diets, the number of steers shedding tetracycline-resistant E. coli increased (P = 0.04) upon transition to the grain-based diet.

Including TET-SUL in the diet also increased (P = 0.02) the prevalence of steers shedding ampicillin-resistant E. coli throughout the trial (Fig. 2b). In general, prevalence of steers shedding ampicillin-resistant E. coli remained steady or tended to decline during the feeding period until the point that antibiotics were removed from the grain-based diet (days 225 and 246). After the withdrawal of antibiotics from the grain-based diet, the prevalence of steers shedding ampicillin-resistant E. coli increased (P = 0.04) in all treatment groups. A notable exception to this trend was observed on day 113, when the prevalence of shedding of ampicillin-resistant E. coli in the control group was dramatically increased over the day 92 level. The type of diet fed (silage- versus grain-based) had no effect on the prevalence of animals shedding ampicillin-resistant E. coli. The prevalence of cattle shedding gentamicin-resistant E. coli was not affected by the type of antibiotic fed or by the withdrawal of antibiotics from the diet, but it tended (P = 0.09) to be higher with the grain-based diet than with the silage-based diet (Fig. 2c).

Overall, the prevalence of animals shedding resistant E. coli at least once throughout the experiment was similar across treatments. E. coli strains resistant to 4 μg tetracycline ml⁻¹ were isolated from all cattle within each treatment at least once during the study. Observations of shedding of gentamicin-resistant (2 μg ml⁻¹) E. coli were similar, with more than 90% of steers shedding these organisms at least once, irrespective of the presence of antibiotics in the diet. Over the course of the 11 sampling times, E. coli strains resistant to ampicillin at 50 μg ml⁻¹ were isolated from 70% of the steers in the study.

(ii) Antibiotic-resistant populations.

With the exception of the control (P = 0.80) and MON (P = 0.212) treatments, numbers of E. coli isolated on nonselective MAC plates from cattle were lower (P = 0.02) when the silage-based diet was fed, compared to the grain-based diet (Table 1). Supplementing the silage-based diet with TET (P = 0.024) and TET-SUL (P = 0.01) lowered the number of E. coli isolated in comparison to the control group (data not shown). Variability across treatments in the numbers of E. coli isolated was less pronounced when the grain-based diet was fed, but isolate numbers were numerically greater when tylosin or virginiamycin were administered, compared to the control.

In general, E. coli isolates culturable on antibiotic-supplemented MAC agar were more numerous when antibiotics were added to the grain-based diet than when they were added to the silage-based diet (Table 2). Changing from the silage- to the grain-based diet tended to increase (P = 0.09) the proportion of tetracycline resistance among isolated E. coli (MAC+TET plates) irrespective of the dietary antibiotic treatment. The TET-SUL treatment increased the percentage of E. coli resistant to tetracycline, compared to the control, for both silage-based (P = 0.001) and grain-based (P = 0.02) diets. The percentage of ampicillin resistance among the E. coli isolates was also increased by the inclusion of TET-SUL in both the silage-based (P = 0.001) and grain-based (P = 0.03) diets. The degree of resistance to ampicillin (at 50 μg ml⁻¹) among isolated E. coli was unaffected (P = 0.15) by the concentration of grain in the diet.

The overall percentages of E. coli resistant to antibiotic-supplemented MAC plates, across all sampling dates, are shown in Fig. 3. A significant proportion (>3%) of the E. coli population isolated from any treatment group (including the control) exhibited resistance to tetracycline. Including tetracycline in the diet increased the degree of tetracycline resistance in the E. coli population to 10% when tetracycline was fed alone (TET treatment; P = 0.041), and to 19.5% when it was fed in combination with sulfamethazine (TET-SUL treatment; P = 0.001). The TET-SUL treatment also resulted in increased (P = 0.05) proportions of ampicillin-resistant strains among the E. coli isolates. This was not observed with the TET treatment, which contained no sulfamethazine and lower tetracycline concentrations compared to the TET-SUL treatment (11 ppm versus 44 ppm, respectively). In all cases, less than 0.5% of the E. coli population exhibited resistance to gentamicin.

Antibiotic susceptibility.

A total of 7,166 E. coli isolates collected from the steers over the course of the study from MAC (n = 3,512), MAC+TET (n = 2,205), MAC+AMP (n = 637), and MAC+GEN (n = 812) plates were tested for susceptibilities. The MICs of ampicillin, ceftazidime, cefpodoxime, gentamicin, and tetracycline were determined for each isolate. Isolates were considered resistant if they were capable of growth at or above the resistance breakpoints determined by the CLSI.

(i) Antibiotic MICs for E. coli isolates.

The MICs of antibiotics for all isolates across antimicrobial treatments are reported in Table 3. The number of resistant E. coli strains was variable, depending on the selective media used for isolation. As expected, given that the ampicillin concentration in the MAC+AMP plates exceeded the NCCLS MIC standard of 32 μg ml⁻¹, breakpoint ampicillin resistance was confirmed in virtually all (635 of 637) of the isolates cultured on MAC+AMP. Similarly, a high proportion (2,183 of 2,205) of E. coli strains resistant to tetracycline were isolated from plates containing the same selective antibiotic (MAC+TET plates). There appeared to be a genetic linkage between determinants conferring resistance to tetracycline and ampicillin. Of the E. coli isolated from MAC+TET plates, 11.4% were resistant to ampicillin compared to 2.2% of those isolated from MAC plates. Conversely, a greater percentage of E. coli isolated from MAC+AMP plates were resistant to tetracycline compared to those from the MAC plates (77.7% versus 15.1%).

Isolates exhibiting resistance to gentamicin were recovered only from MAC+GEN plates, and at a low rate. Of the 812 isolates tested from MAC+GEN plates, which contained 2 μg of gentamicin ml⁻¹, only 7 were confirmed as gentamicin resistant. However, of the MAC+GEN isolates, 5.4% were resistant to ampicillin, and 25.1% were resistant to tetracycline. From the NCCLS screening procedures, 131 and 142 candidate ESBL-producing E. coli isolates were identified by growth on plates containing 2 μg of ceftazidime or cefpodoxime ml⁻¹. With the exception of the MAC+AMP plates, the type of selective medium used to isolate E. coli did not affect those with reduced susceptibility to ceftazidime or cefpodoxime. For E. coli isolated from MAC+AMP plates, 14.3 and 16.2% of the isolates were capable of growth on plates containing 2 μg ceftazidime ml⁻¹ and 2 μg cefpodoxime ml⁻¹, respectively. This accounted for the majority of total candidate ESBL-producing E. coli with reduced susceptibility to ceftazidime (91 of 131 isolates) and cefpodoxime (103 of 143 isolates).

(ii) Breakpoint resistance within diet treatments.

With the exception of the TET-SUL treatment, the administration of antibiotics to cattle did not affect the development of resistance in E. coli to ampicillin, tetracycline, or gentamicin compared to the control group (Table 4). Administration of TET-SUL to the steers increased (P = 0.06) the recovery of ampicillin-resistant E. coli on MAC, MAC+TET, and MAC+AMP plates. Similarly, the recovery of isolates resistant to tetracycline was higher for the TET-SUL group versus all of the other treatments (P = 0.01), when E. coli was isolated on MAC, MAC+AMP, or MAC+GEN media. Similar rates of tetracycline resistance (97.4 to 99.7%) were observed among isolates from all treatments, when isolated on MAC+TET plates. E. coli strains resistant to gentamicin were recovered from steers in the control, TET-SUL, TET, and VIR treatments. These isolates were only recovered on MAC+GEN plates, and at equally low rates (0.6 to 2.2%) across these treatments. The prevalence of candidate ESBL-producing isolates was evenly distributed among samples from steers in all feeding groups except the MON treatment (data not shown).

Isolates exhibiting coresistance to both ampicillin and tetracycline were isolated from steers within each of the antimicrobial treatments (Table 4). Recovery of tetracycline and ampicillin coresistance was greatest from the TET-SUL treatment. Almost all of the isolates from the TET-SUL treatment that displayed ampicillin resistance were also resistant to tetracycline, regardless of isolation media (83.3 to 100%). Across all antimicrobial treatments, every E. coli isolate selected from MAC+TET plates that was resistant to ampicillin was also resistant to tetracycline. In addition, a high proportion (57.1 to 95.2%) of E. coli isolates selected from MAC+AMP plates showed resistance to tetracycline. These data suggested that in many of the isolates, determinants encoding ampicillin resistance were linked to determinants encoding tetracycline resistance. One isolate from the control treatment was resistant both to tetracycline and to gentamicin. Two isolates from the TET and one from the VIR treatments showed coresistance to ampicillin, tetracycline, and gentamicin.

PFGE patterns.

Pulsed-field gel electrophoresis of ampicillin-resistant E. coli isolated from the control steers on days 15, 113, and 246 provided further insight into the observations of ampicillin resistance. The majority of steers shedding ampicillin-resistant E. coli on days 113 and 246 were from pens 19 and 21, respectively (Fig. 4) . These groups of animals accounted in large part for the increase in prevalence of steers shedding ampicillin-resistant E. coli on those days (Fig. 2). The PFGE patterns of the ampicillin-resistant isolates collected on day 113 from each of the 10 steers in pen 19 were identical to one another (representative data in Fig. 5A) and to those obtained from the four steers in adjacent pen 20 on the same day (data not shown). Similarly, the PFGE patterns of ampicillin-resistant E. coli isolates from each of the 10 animals in pen 21 on day 246 were identical (Fig. 5B). Of note, however, is that the patterns produced by pen 19 isolates from day 113 or by the pen 21 isolates from day 246 were not the same as the patterns observed for ampicillin-resistant isolates collected from these same steers (one only in pen 21) on day 15 of the experiment (Fig. 5C and D). The dissimilar PFGE patterns between pens 19 and 21 illustrated that the ampicillin-resistant E. coli isolated from these pens were likely different strains.