Subtype-Specific Selection for Resistance to Fluoroquinolones but Not to Tetracyclines Is Evident in Campylobacter jejuni Isolates from Beef Cattle in Confined Feeding Operations in Southern Alberta, Canada

ABSTRACT

Campylobacter jejuni was longitudinally isolated from beef cattle housed in four confined feeding operations (CFOs) in Southern Alberta, Canada, over 18 months. All of the cattle were administered a variety of antimicrobial agents (AMAs) nontherapeutically and metaphylactically during their time in the CFOs. In total, 7,966 C. jejuni isolates were recovered from cattle. More animals were colonized by the bacterium after >60 days in the CFO (interim) than were individuals upon entry at the CFO (arrival). Subtyping and resistance to seven AMAs were determined for 1,832 (23.0%) and 1,648 (20.7%) isolates, respectively. Increases in the proportion of isolates resistant to tetracycline were observed at all four CFOs between sample times and to ciprofloxacin and nalidixic acid at one or more CFOs. The vast majority of isolates resistant to tetracycline carried tetO, whereas ciprofloxacin resistance was predominantly attributed to mutations in the gyrA gene. Although considerable diversity was observed, a majority of C. jejuni isolates belonged to one of five predominant subtype clusters. There was no difference in subtype diversity by CFO, but the population structure differed between sample times. Selection for resistance to ciprofloxacin and nalidixic acid was subtype dependent, whereas selection for resistance to tetracycline was not. The findings indicate that a proportion of cattle entering CFOs carry resistant C. jejuni subtypes, and the characteristics of beef cattle CFOs facilitate transmission/proliferation of diverse subtypes, including those resistant to AMAs, which coupled with the densities of CFOs likely contribute to the high rates of cattle-associated campylobacteriosis in Southern Alberta.

IMPORTANCE A small proportion of cattle entering a CFO carry Campylobacter jejuni, including subtypes resistant to AMAs. The large numbers of cattle arriving from diverse locations at the CFOs and intermingling within the CFOs over time, coupled with the high-density housing of animals, the high rates of transmission of C. jejuni subtypes among animals, and the extensive use of AMAs merge to create an ideal situation where the proliferation of diverse antimicrobial-resistant C. jejuni subtypes is facilitated. Considering that Southern Alberta reports high rates of campylobacteriosis in the human population and that many of these clinical cases are due to C. jejuni subtypes associated with cattle, it is likely that the characteristics of beef cattle CFOs favor the propagation of clinically relevant C. jejuni subtypes, including those resistant to medically important AMAs, which constitute a risk to human health.

INTRODUCTION

Southern Alberta, Canada, possesses a very large number of beef cattle (≈3 million head), ≈1 million of which are in confined feeding operations (CFOs) (1). The rates of campylobacteriosis, primarily incited by Campylobacter jejuni, in people living in this region are substantially higher than the provincial and national averages (2). Our recent findings based on molecular epidemiological evidence have indicated that subtypes of C. jejuni originating from beef cattle may be contributing disproportionately to infections in Southern Alberta. Our research has also shown that many subtypes of C. jejuni associated with livestock, meat, wildlife, and the environment in Southern Alberta are not observed in diarrheic humans in the region, indicating that not all subtypes of the bacterium are a public health risk. In addition, we have observed that the prevalence of resistance to medically important antimicrobial agents (AMAs), such as fluoroquinolones, in C. jejuni is increasing in the region. Thus, it is important that studies on the development of resistance to AMAs and the proliferation and transmission of C. jejuni at a subspecies level of resolution are undertaken in important zoonotic reservoirs, such as beef cattle, to elucidate the risk posed to human beings by C. jejuni, including from subtypes resistant to AMAs, and to implement rationale-based mitigation strategies to reduce the burden of campylobacteriosis.

The intestines of beef cattle are readily colonized by Campylobacter species, including C. jejuni, and cattle chronically shed these bacteria in their feces in large numbers (3–5). However, the transmission of C. jejuni between individuals within a herd has not been fully elucidated. In most cases, calves are born and maintained on pasture in a relatively low-density situation until they are ≈9 to 11 months of age, at which point they are transferred to CFOs (Beef Farmers of Ontario, Guelph, Ontario, Canada). Research indicates that some calves are colonized by C. jejuni prior to entering CFOs, but the bacterium is more widespread among individuals within the CFO environment (6–8). It is thought that the high density of animals within CFOs facilitates the transmission of C. jejuni isolates among individuals, likely as a result of their continual contact with feces and close proximity to each other (6, 9, 10). In this regard, it is plausible that the continuous arrival of cattle at CFOs, coupled with high density and intermingling of individuals during their time at the CFO, provides a semiconstant population of individual animals that act as a reservoir for a large number of C. jejuni subtypes.

Beef calves on pasture are not typically administered AMAs, but once they arrive at the CFO, they are exposed to a variety of AMAs that are administered therapeutically, metaphylactically, and as antimicrobial growth promoters (AGPs) (11). It is estimated that more than 2 million kg of AMAs is administered to cattle in North America each year (12). The administration of AMAs, particularly those administered continuously in feed at relatively low concentrations, has been hypothesized to increase the likelihood that target and/or nontarget bacteria will develop increased resistance to those agents (13, 14). In previous research, we showed that resistance to tetracycline in Campylobacter species shed in beef cattle feces rapidly increases during the CFO period (6, 7), but our understanding of the development of antimicrobial resistance in C. jejuni in CFO cattle is hampered by the lack of information on population structure and strain dynamics in the CFO environment.

We hypothesized that (i) CFOs in Southern Alberta support high subtype diversity and resistance to AMAs in C. jejuni, (ii) resistance arises spontaneously in C. jejuni, and (iii) under selection pressure, resistance rapidly proliferates within a limited number of C. jejuni subtypes primarily through clonal expansion. To test these hypotheses, the following study objectives were created: (i) longitudinally sample cattle in four commercial CFOs in Southern Alberta, (ii) monitor AMA administration to cattle during the CFO period, (iii) longitudinally recover C. jejuni isolates from beef cattle upon entry to and after >60 days in the CFO, (iv) quantify/characterize antimicrobial resistance to AMAs and subtype a large number of C. jejuni isolates (>1,500), and (v) characterize the changes in the population structure (i.e., presence and abundance of specific subtypes) of both resistant and nonresistant C. jejuni strains over time.

RESULTS

All cattle were administered antimicrobial agents nontherapeutically and metaphylactically within the four commercial confined feeding operations examined in Southern Alberta.

Beef cattle were administered AMAs in feed at nontherapeutic concentrations as AGPs at all CFOs, according to standard industry practices (see Table S1 in the supplemental material). The majority of animals (77.6 to 89.7%) were administered chlortetracycline as an AGP. Animals administered chlortetracycline were also administered monensin in feed at 25 or 33 mg/kg dry weight (DW) of feed. Five hundred nineteen animals were administered tylosin phosphate at a concentration of 11 mg/kg DW of feed. At CFO1, 150 animals were administered ampolium at a dose of 16 g/head/day. The mean duration of in-feed chlortetracycline administration at the four CFOs ranged from 63.7 to 102.1 days. The in-feed administration duration of monensin ranged from 68.0 to 144.2 days, and the administration duration of tylosin phosphate ranged from 4.7 to 39.1 days.

Select animals were administered AMAs therapeutically or metaphylactically (Table S2). Ceftiofur sodium, a coccidiostat bolus, enrofloxacin, florfenicol (± flunixin meglumine), oxytetracycline, tilmicosin, trimethoprim-sulfadoxine, tulathromycin, and tylosin tartrate were administered to cattle at various frequencies. At the four CFOs, oxytetracycline at rates of 20 and 30 mg/kg of body weight were administered within 29.0 and 0.5 days of the animal's arrival at the CFO on average, respectively. Tilmicosin was only used extensively at CFO4 and was administered to cattle within 2.5 days of cattle arrival, on average. Tylosin tartrate was only administered at CFO3, 3.4 days, on average, after cattle arrived at the CFO. Tulathromycin was used extensively at CFO1 and CFO4, and to a lesser extent at CFO2; cattle were administered tulathromycin within 6.8 days, on average, after arrival. Enrofloxacin was administered to 11 cattle at 34.8 days, on average, after their arrival at CFO4.

Campylobacter jejuni and Campylobacter coli isolates were frequently recovered from beef cattle within confined feeding operations.

Campylobacteria were isolated from cattle upon arrival and after >60 days in the CFO (interim). A total of 10,665 presumptive Campylobacter isolates were recovered from feces collected from 4,427 individual cattle (arrival and interim) and from 640 composite fecal samples from pen floors. Of the isolated bacteria, 9,710 isolates (91.0%) were identified as Campylobacter species; C. jejuni was the most common species isolated (n = 7,966 [82.0%]), followed by C. coli (n = 1,724 [17.8%]), C. hyointestinalis (n = 18 [0.2%]), and C. fetus (n = 2 [0.02%]) (Table S3). The frequencies of isolation of C. jejuni and C. coli from individual cattle varied (P < 0.001) between arrival and interim samples, but there was no significant difference (P > 0.050) between CFOs (Table 1). Both C. jejuni and C. coli were more frequently (P < 0.001) isolated from fecal samples obtained from individual animals in interim samples than upon arrival (Fig. S1A), but only C. jejuni was more frequently isolated (P < 0.001) in interim composite pen manure samples (Fig. S1B).

TABLE 1.

Frequencies of Campylobacter jejuni and C. coli isolation from beef cattle housed in four commercial CFOs upon arrival and after >60 days in the CFO^a

	Frequency of isolation (%)
	Arrival		Interim
CFO	C. coli	C. jejuni	C. coli	C. jejuni
CFO1	3.1	36.0	16.4b	60.7b
CFO2	0.8	20.8	13.3b	57.3b
CFO3	4.9	25.7	20.7b	62.2b
CFO4	1.8	24.4	19.7b	63.5b
Mean ± SD for all CFOs	2.6 ± 1.7	26.7 ± 6.5	15.5 ± 3.4b	60.9 ± 2.7b

^aInterim is defined as after >60 days in the CFO.

^bArrival and interim samples differ (P < 0.050).

The majority of Campylobacter jejuni isolates recovered from beef cattle housed in confined feeding operations were highly resistant to tetracycline but not to nalidixic acid, ciprofloxacin, chloramphenicol, clindamycin, erythromycin, or gentamicin.

The sensitivities of C. jejuni to AMAs were determined for 1,648 isolates (25.5% of the total C. jejuni isolates recovered from individual fecal samples collected from cattle), which included 416 isolates from arrival samples and 1,232 isolates from interim samples (Table 2). Campylobacter jejuni isolates were characterized as resistant or susceptible based on MIC breakpoint values. The greatest number of isolates were resistant to tetracycline (82.4%), followed by nalidixic acid (9.2%) and ciprofloxacin (8.0%). Less than 1.0% of isolates were resistant to chloramphenicol (0.06%), clindamycin (0.5%), erythromycin (0.2%), and/or gentamicin (0.4%). An increase (P = 0.005) in the proportion of isolates resistant to ciprofloxacin from arrival to interim was observed at CFO1 and CFO2 (Fig. 1A); the average MIC increased from 1.32 μg · ml⁻¹ to 2.26 μg · ml⁻¹. Similarly, an increase (P = 0.004) in the proportion of isolates resistant to nalidixic acid from arrival to interim was observed at CFO1 (Fig. S2A); the average MIC increased from 21.83 μg · ml⁻¹ to 29.55 μg · ml⁻¹. A conspicuous increase (P < 0.001) in the proportion of isolates resistant to tetracycline from arrival to interim was observed at all CFOs (Fig. 1B), with an increase in the average MIC from 45.86 to 114.53 μg · ml⁻¹ and in the median MIC from 0.25 μg · ml⁻¹ to 128 μg · ml⁻¹.

TABLE 2.

MICs for ciprofloxacin, chloramphenicol, clindamycin, erythromycin, gentamicin, nalidixic acid, and tetracycline for Campylobacter jejuni isolates recovered from beef cattle housed in four commercial CFOs in Southern Alberta upon arrival and after >60 days in the CFO^c

Antimicrobial by sample typea	No. (%) of isolates by MIC (μg · ml−1)b													Avg MIC	MIC50
	0.06	0.125	0.25	0.5	1	2	4	8	16	32	64	128	256
Arrival samples
CIP		130 (31.3)	225 (54.10)	34 (8.2)	1 (0.2)	2 (0.5)	2 (0.5)	2 (0.5)	13 (3.1)	7 (1.7)				1.3	0.25
CHL						134 (32.2)	224 (53.9)	56 (13.5)	1 (0.2)	0 (0.0)	1 (0.2)	0 (0.0)	0 (0.0)	4.1	4.00
CLI				185 (44.5)	172 (41.4)	53 (12.7)	1 (0.2)	2 (0.5)	1 (0.2)	0 (0.0)	2 (0.5)			1.3	1.00
ERY				17 (4.1)	28 (6.7)	190 (45.7)	151 (36.3)	28 (6.7)	0 (0.0)	0 (0.0)	2 (0.5)			3.3	2.00
GEN				344 (82.7)	64 (15.4)	4 (1.0)	0 (0.0)	0 (0.0)	0 (0.0)	3 (0.7)	1 (0.2)	0 (0.0)		1.0	0.50
NAL						8 (1.9)	119 (28.6)	224 (53.9)	29 (7.0)	4 (1.0)	3 (0.7)	11 (2.6)	18 (4.3)	21.8	8.00
TET	21 (5.1)	116 (27.9)	81 (19.5)	14 (3.4)	4 (1.0)	0 (0.0)	2 (0.5)	2 (0.5)	4 (1.0)	20 (4.8)	54 (13.0)	80 (19.2)	18 (4.3)	45.9	0.25
Interim samples
CIP		373 (30.3)	660 (53.6)	80 (6.5)	11 (0.9)	0 (0.0)	2 (0.2)	7 (0.6)	47 (3.8)	52 (4.2)				2.3	0.25
CHL						313 (25.4)	734 (59.6)	178 (14.5)	7 (0.6)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	4.1	4.00
CLI				549 (44.6)	503 (40.8)	164 (13.3)	12 (1.0)	1 (0.1)	0 (0.0)	1 (0.1)	2 (0.2)			1.1	1.00
ERY				48 (3.9)	71 (5.8)	613 (49.8)	419 (34.0)	78 (6.3)	0 (0.0)	0 (0.0)	2 (0.2)			3.1	2.00
GEN				916 (74.4)	308 (25.0)	6 (0.5)	0 (0.0)	0 (0.0)	1 (0.1)	1 (0.1)	0 (0.0)	0 (0.0)		0.7	0.50
NAL						29 (2.4)	236 (19.2)	697 (56.6)	146 (11.9)	5 (0.4)	10 (0.8)	15 (1.2)	94 (7.6)	29.6	8.00
TET	15 (1.2)	7 (0.6)	5 (0.4)	1 (0.1)	3 (0.2)	6 (0.5)	3 (0.2)	10 (0.8)	15 (1.2)	86 (7.0)	397 (32.2)	489 (39.7)	195 (15.8)	114.5	128.00

^aCIP, ciprofloxacin; CHL, chloramphenicol; CLI, clindamycin; ERY, erythromycin; GEN, gentamicin; NAL, nalidixic acid; TET, tetracycline.

^bThe total number of C. jejuni isolates examined was 1,648.

^cInterim is defined as after >60 days in the CFO.

FIG 1.

Frequency of detection (%) of Campylobacter jejuni isolates recovered from beef cattle housed in four commercial confined feeding operations (CFOs) in Southern Alberta that were resistant to ciprofloxacin (A) or tetracycline (B) upon arrival and after >60 days in the CFO (interim). Asterisks indicate that the frequency of detection differed (P < 0.050) between the two sample times for individual CFOs.

Resistance to tetracycline in Campylobacter jejuni isolates was due to the carriage of tetO, and resistance to ciprofloxacin was the result of mutations in the gyrA gene.

Of the 484 C. jejuni isolates from cattle that exhibited phenotypic resistance to tetracycline that were examined, 97.8% and 99.3% of the isolates recovered at the arrival and interim sample times, respectively, carried the tetO resistance determinant gene. Four (2.2%) and two (0.7%) tetracycline-resistant isolates recovered from cattle at the arrival and interim sample times, respectively, were negative for tetO. None of the isolates carried tetB, tetC, tetL, tetM, tetQ, or tetW.

Of the 134 C. jejuni isolates that exhibited phenotypic resistance to ciprofloxacin and/or nalidixic acid, 115 (85.8%) isolates contained mutations in the quinolone resistance-determining region (QRDR) of the gyrA gene that conferred an amino acid substitution (Thr86Ile), 10 (7.5%) isolates contained mutations in the gyrA gene that did not confer an amino acid substitution, and nine (6.7%) isolates did not possess any mutations within the gyrA gene (Table 3). A total of 109 (94.8%) isolates that had the Thr86Ile mutation were resistant to both nalidixic acid and ciprofloxacin. Of the 19 C. jejuni isolates that did not have the Thr86Ile mutation, 13 (68.4%) isolates were resistant to nalidixic acid but not to ciprofloxacin. While the presence of a Thr86Ile substitution was sufficient to confer ciprofloxacin resistance, isolates with additional mutations at A21G and T72C had higher MIC values (Table 4). None of the 19 isolates that were resistant to ciprofloxacin and/or nalidixic acid without an amino acid substitution (Thr86Ile) exhibited plasmid-mediated quinolone resistance (PMQR), nor did they possess a C-to-T point mutation in the binding region of the transcriptional regulator of the CmeABC multidrug efflux pump.

TABLE 3.

Frequency of mutations in the gyrA for Campylobacter jejuni isolates recovered from beef cattle housed in four commercial confined feeding operations in Southern Alberta that exhibited phenotypic resistance to ciprofloxacin and/or nalidixic acid^a

Mutationb	No. (%) of isolates
None	9 (6.7)
Mutations resulting in a threonine to isoleucine substitution
C257T	8 (6.0)
C257T-T357C-G408A-C471T	34 (25.4)
T72C-C257T-T357C-G408A-C471T	1 (0.8)
A21G-C257T-T357C-G408A-C471T-T483C	1 (0.8)
C243T-C257T-T357C-C360T-C471T-T483C	16 (11.9)
T72C-C243T-C257T-T357C-C360T-C471T-T483C	1 (0.8)
A21G-T72C-C243T-C257T-T357C-C360T-C471T-T483C	54 (40.3)
No amino acid substitution
T357C-G408A-C471T	8 (6.0)
A21G-T72C-C243T-T357C-C360T-C471T-T483C	1 (0.8)
A64G-T72C-C243T-T357C-C360T-C471T-T483C	1 (0.8)

^aA total of 134 C. jejuni isolates were analyzed.

^bFor the mutation notation used, the first letter in the sequence refers to the wild-type nucleotide, the number (i.e., nucleotide 1 to nucleotide 483) refers to the mutation position, and the last letter in the sequence refers to the mutation nucleotide (e.g., for mutation C257T the nucleotide C becomes a nucleotide T at position 257 of the gyrA gene). All mutations, with the exception of C257T, are silent.

TABLE 4.

Degree of ciprofloxacin resistance in Campylobacter jejuni isolates recovered from beef cattle housed in four commercial confined feeding operations in Southern Alberta by gyrA mutation

Mutationa	No. (%) of isolates by MIC			Total no. of isolates
	<4 μg · ml−1	4–16 μg · ml−1	>16 μg · ml−1
C257T	0 (0)b	6 (75.0)	2 (25.0)	8
C257T-T357C-G408A-C471T	3 (8.8)	31 (91.2)	0 (0)	34
C243T-C257T-T357C-C360T-C471T-T483C	1 (6.3)	14 (87.5)	1 (6.3)	16
A21G-T72C-C243T-C257T-T357C-C360T-C471T-T483C	1 (1.9)	7 (13.2)	45 (84.9)	53

^aFor the mutation notation used, the first letter in the sequence refers to the wild-type nucleotide, the number (i.e., nucleotide 1 to nucleotide 483) refers to the mutation position, and the last letter in the sequence refers to the mutation nucleotide (e.g., for mutation C257T the nucleotide C becomes a nucleotide T at position 257 of the gyrA gene).

Although high rates of subtype diversity were observed, specific subtypes of Campylobacter jejuni were differentially selected for within confined feeding operations.

A total of 1,832 C. jejuni isolates were subtyped using a comparative genomic fingerprinting (CGF) method, and 329 subtype clusters were identified at 95% CGF profile similarity. Overall subtype diversity was high, and there was no difference between CFOs at the arrival or interim sample times (Fig. S3 to S5). Within individual CFOs, there were no changes (P = 0.235) in C. jejuni subtype diversity between arrival and interim samples. There was no difference (P = 0.103) in population structure between CFOs at arrival sampling, but at the interim sample time, the population structure at CFO1 differed from those at CFO2 (P = 0.005), CFO3 (P < 0.001), and CFO4 (P < 0.001), and the structure of the C. jejuni population at CFO2 differed from that at CFO3 (P = 0.001). Within individual CFOs, C. jejuni population structures differed between arrival and interim samples for CFO1 (P = 0.008), CFO3 (P = 0.002), and CFO4 (P = 0.001). Five subtype clusters (169.001, 238.002, 695.006, 735.001, and 853.011) represented the majority of C. jejuni isolates subtyped (n = 999 [54.5%]) (Fig. S6). These subtype clusters were predominant at all CFOs, but their relative abundances varied by CFO and sample time (Table 5). Overall, isolates from clusters 735.001 and 853.011 were more prevalent in interim samples than in arrival samples (P = 0.001 and 0.025, respectively), while isolates from clusters 169.001 and 695.006 were less prevalent in interim samples than in arrival samples (P = 0.022 and <0.001, respectively).

TABLE 5.

Relative prevalences of predominant Campylobacter jejuni comparative genomic fingerprint subtype clusters recovered from beef cattle housed in four commercial CFOs upon arrival and after >60 days in the CFO^a

Cluster	Prevalence (%)
	CFO1		CFO2		CFO3		CFO4
	Arrival (n = 119)	Interim (n = 317)	Arrival (n = 83)	Interim (n = 288)	Arrival (n = 156)	Interim (n = 393)	Arrival (n = 101)	Interim (n = 375)
169.001	10.1	17.7	27.7	14.6b	20.5	8.1b	10.9	11.7
238.002	16.8	8.2b	4.8	9.4	3.2	7.4	9.9	8.5
695.006	22.7	6.9b	16.9	14.9	16.0	12.0	25.7	12.3b
735.001	5.0	7.9	2.4	9.4b	7.1	12.5	5.0	10.9
853.011	3.4	12.3b	7.2	11.5	10.3	10.7	5.9	8.3

^aInterim is defined as after >60 days in the CFO.

^bArrival and interim samples differ (P < 0.050).

Development of resistance to ciprofloxacin and nalidixic acid was subtype dependent in C. jejuni from beef cattle in confined feeding operations.

Overall, there was less (P < 0.001) subtype diversity for isolates resistant to ciprofloxacin (Shannon's index [H] = 2.26 ± 0.27) than for isolates susceptible to ciprofloxacin (H = 3.10 ± 0.22) and for isolates resistant to nalidixic acid (H = 2.56 ± 0.26) than for those susceptible to nalidixic acid (H = 3.05 ± 0.19; P = 0.003). This trend was maintained at CFO1, CFO2, and CFO3, but at CFO4, there was no difference in subtype diversity between resistant and susceptible isolates for ciprofloxacin (Fig. 2A) or nalidixic acid (Fig. S2B). There was no overall difference (P = 0.079) in subtype diversity between tetracycline-resistant (H = 3.28 ± 0.18) and tetracycline-susceptible (H = 3.05 ± 0.19) isolates, nor at individual CFOs (Fig. 2B).

FIG 2.

Genetic diversity of Campylobacter jejuni isolates susceptible and resistant to ciprofloxacin (A) or tetracycline (B) recovered from beef cattle housed in four CFOs in Southern Alberta. Error bars associated with the histogram bars represent 95% confidence intervals, and asterisks indicate a change (P < 0.050) in subtype diversity between the susceptible and resistant subtypes.

Increases in the proportions of C. jejuni isolates resistant to nalidixic acid (Fig. S7A and B) and ciprofloxacin (Fig. 3A and B) at the interim sample time compared to arrival were subtype and CFO dependent, whereas increases in the proportion of isolates resistant to tetracycline (Fig. 4A and B) were only CFO dependent (Table 6). Nalidixic acid- and ciprofloxacin-resistant isolates that belonged to CGF subtype cluster 169.001 were more prevalent (P ≤ 0.002) in interim samples at CFO2. Nalidixic acid- and ciprofloxacin-resistant isolates that belonged to subtype cluster 695.006 were more prevalent (P ≤ 0.011) in interim samples at CFO1, CFO3, and CFO4. Nalidixic acid-resistant subtype cluster 695.006 isolates were also more prevalent (P < 0.001) in interim samples at CFO2. At CFO3, a higher proportion of subtype cluster 735.001 isolates in arrival samples were resistant to ciprofloxacin (P = 0.002) but not to nalidixic acid (P = 0.113). Tetracycline-resistant isolates that belonged to subtype clusters 169.001 and 695.006 were more prevalent (P < 0.009) in interim samples at all four CFOs. Isolates resistant to tetracycline that belonged to subtype cluster 238.002 were more prevalent (P ≤ 0.004) in interim samples at three of the four CFOs (i.e., CFO2, CFO3, and CFO4). Tetracycline-resistant isolates that belonged to subtype cluster 853.011 were more prevalent (P ≤ 0.002) in interim samples at CFO1 and CFO3; the majority of 853.011 isolates in arrival samples at CFO2 and CFO4 were resistant to tetracycline (≥75%). The majority of isolates (≥75%) that belonged to subtype cluster 735.001 were also resistant to tetracycline in arrival samples at all four CFOs.

FIG 3.

Campylobacter jejuni comparative genomic fingerprint subtypes sensitive and resistant to ciprofloxacin recovered from beef cattle housed in four commercial CFOs upon arrival (A) and after >60 days in the CFO (interim) (B). Data are combined across the CFOs. The minimum spanning tree was generated in BioNumerics (version 6.6; Applied Maths), and connecting lines represent two or fewer mismatched loci (95% similarity) between respective subtypes.

FIG 4.

Campylobacter jejuni comparative genomic fingerprint subtypes sensitive and resistant to tetracycline recovered from beef cattle housed in four commercial CFOs upon arrival (A) and after >60 days in the CFO (interim) (B). Data are combined across the CFOs. The minimum spanning tree was generated in BioNumerics (version 6.6; Applied Maths), and connecting lines represent two or fewer mismatched loci (95% similarity) between respective subtypes.

TABLE 6.

Prevalences of Campylobacter jejuni isolates recovered from beef cattle housed in four commercial CFOs that were resistant to antimicrobial agents upon arrival and after >60 days in the CFO by comparative genomic fingerprint subtype cluster^a

Cluster by antibiotic	Prevalence (%)
	CFO1		CFO2		CFO3		CFO4
	Arrival	Interim	Arrival	Interim	Arrival	Interim	Arrival	Interim
Nalidixic acid
169.001	11.1	21.7	18.8	66.7b	26.7	3.8	0.0	7.3
238.002	0.0	4.3	0.0	6.3	20.0	7.7	0.0	3.7
695.006	16.0	83.3b	7.7	45.2b	13.0	58.1b	15.4	52.4b
735.001	0.0	8.3	0.0	0.0	20.0	2.1	0.0	10.3
853.011	0.0	5.9	0.0	0.0	16.7	8.6	25.0	3.4
Ciprofloxacin
169.001	11.1	21.7	6.3	59.3b	13.3	3.8	0.0	2.4
238.002	0.0	4.3	0.0	0.0	0.0	0.0	0.0	3.7
695.006	8.0	72.2b	0.0	0.0	4.3	32.6b	0.0	21.4b
735.001	0.0	0.0	0.0	0.0	20.0	0.0b	0.0	0.0
853.011	0.0	0.0	0.0	0.0	0.0	0.0	25.0	0.0
Tetracycline
169.001	44.4	95.7b	68.8	100.0b	63.3	96.2b	20.0	100.0b
238.002	70.6	95.7	0.0	87.5b	40.0	96.2b	33.3	100.0b
695.006	16.0	100.0b	38.5	93.5b	8.7	95.3b	11.5	100.0b
735.001	75.0	100.0	100.0	100.0	100.0	97.9	100.0	100.0
853.011	50.0	100.0b	80.0	96.6	41.7	97.1b	75.0	100.0

^aInterim is defined as after >60 days in the CFO.

^bArrival and interim samples differ (P < 0.050).

DISCUSSION

We longitudinally sampled beef cattle in four commercial CFOs in Southern Alberta for C. jejuni and C. coli over an 18-month period; the frequency of detection of both species in individual cattle increased after arrival, the average frequencies of detection of C. jejuni and C. coli in individual interim cattle samples increased from 26.7% ± 6.5% to 60.9% ± 2.7% and from 2.6% ± 1.7% to 17.5% ± 3.4%, respectively. The frequency of detection in composite pen manure samples also increased over time for C. jejuni (from 50.3% ± 8.9% to 77.1% ± 4.9%) but not for C. coli. The increase in prevalence of C. jejuni over time in individual cattle and composite pen samples coincides with previous studies (8, 15, 16). Campylobacter jejuni is shed periodically by cattle and survives for extended periods of time in cattle feces, which likely accounts for its higher rates of detection over time in CFO cattle, as Campylobacter-free cattle are placed in close proximity to contaminated individuals and feces and become infected (4, 5, 17). Our findings indicate that Campylobacter carriage rates increase in cattle housed in CFOs over time as a result of passive transmission from residual manure (i.e., upon arrival) and/or direct contact between individuals.

It is generally accepted that the use of AMAs at nontherapeutic doses in feed for prolonged periods to enhance growth in cattle (as AGPs) leads to an increase in prevalence of C. jejuni resistant to AMAs of the same family (6, 18, 19). The majority of cattle sampled at the four commercial CFOs were administered chlortetracycline in feed throughout the sampling period or metaphylactically via injection. The results showed that tetracycline resistance was present in the intestinal bacterial community of cattle entering the CFO (i.e., the majority of these animals and their mothers would not have been administered AMAs), but classical selection for resistance to tetracycline occurred during the feeding period when cattle were administered chlortetracycline in feed as an AGP. The current study was conducted in commercial CFOs, and cattle were administered a multitude of AMAs, including tetracyclines, in a noncontrolled manner. Thus, it is not possible to definitely conclude that the longitudinal increases that were observed in the prevalence of resistance to tetracycline were a direct result of selection due to tetracycline administration. However, we previously showed in a controlled-administration study that tetracycline resistance rates in C. hyointestinalis increased in beef cattle administered chlortetracycline relative to animals not administered AMAs (6). This suggests that the frequent administration of tetracyclines in the commercial CFOs directly selected for chlortetracycline resistance in the current study.

We used clinical breakpoints as opposed to epidemiological cutoffs to delineate resistant isolates in the current study. Epidemiological cutoff values have the advantage that they are established from MIC distributions that separate bacterial populations into those that are representative of a wild-type population and those that possess acquired or mutational resistance to AMAs (20). The epidemiological cutoff values for resistance to ciprofloxacin, nalidixic acid, and tetracycline in C. jejuni from beef cattle in the current study, as determined using the tool developed by Turnidge et al. (21), were 0.5, 16, and 0.5 μg · ml⁻¹, respectively. These values are less than the clinical breakpoints of 4, 64, and 16 μg · ml⁻¹ that were applied for these three AMAs, respectively. Nonetheless, the two methods provided comparable results for ascertaining resistance in C. jejuni isolates from cattle entering a CFO relative to resistance in isolates obtained from animals after 2 months or more in the CFO. For example, the values defining resistance to ciprofloxacin for C. jejuni isolates recovered from animals upon arrival were 6.5% and 5.8% based on epidemiological cutoff and clinical breakpoint values, respectively, and from cattle after >60 days in the CFO, resistance rates were 9.7% and 8.8%, respectively. For tetracycline, the resistance rates for isolates recovered from cattle upon arrival were 44.2% and 42.3% based on epidemiological cutoff and clinical breakpoint values, respectively, and from animals after >60 days in the CFO, the resistance rates were 97.7% and 95.9%, respectively. The similarity between the two methods is attributed to the bimodal distribution of MICs that was observed, particularly for tetracycline, which has been reported previously for C. jejuni isolates recovered from beef cattle (7).

A low proportion of cattle were colonized by C. jejuni resistant to ciprofloxacin (4.8% ± 1.6%) and nalidixic acid (7.1% ± 2.6%) upon arrival, but the proportion was significantly greater in interim samples at CFO1 (14.3%) and CFO2 (10.5%) for ciprofloxacin and at CFO1 (14.7%) for nalidixic acid. The reasons for this are uncertain; off-license administration of enrofloxacin to poultry as an AGP has been suggested as one possible explanation for increasing levels of resistance to fluoroquinolones observed in clinical C. jejuni isolates (22, 23), but enrofloxacin was not administered in feed to cattle in the current study. It is possible that the increased rates of ciprofloxacin resistance observed in C. jejuni from cattle at CFO1 and CFO2 are linked to other factors, such as the administration of tylosin phosphate in feed. Although the mechanisms are poorly understood, the use of tylosin phosphate has been associated with cross-resistance to ciprofloxacin in C. coli from pigs (24), and the highest levels of tylosin phosphate administered to cattle in feed as an AGP in the current study occurred at CFO1 and CFO2. It is noteworthy that we have observed a substantive increase in the prevalence of ciprofloxacin resistance in C. jejuni isolates from diarrheic humans living in southwestern Alberta in recent years (our unpublished data) and that C. jejuni subtypes of cattle origin are implicated in the high rates of campylobacteriosis in this region (our unpublished data). Minimal resistance (≤0.9%) to erythromycin was detected in C. jejuni from CFO cattle in the current study, despite the use of tylosin (and other macrolides) at therapeutic or nontherapeutic concentrations. These findings coincide with previous studies of Campylobacter species (C. hyointestinalis and/or C. jejuni) in cattle in CFOs (6, 25).

In the current study, nearly all of the C. jejuni isolates that exhibited phenotypic resistance to tetracycline carried the tetO resistance determinant, but none of the isolates carried tetB, tetC, tetL, tetM, tetQ, or tetW resistance determinants. These findings are similar to those of previous studies that demonstrated widespread carriage of tetO in tetracycline-resistant C. jejuni isolates in poultry (26–28) and cattle (9). As suggested previously, the high rate of carriage of the tetO resistance gene is likely due to its rapid, spontaneous, and plasmid-mediated transfer between C. jejuni in animal digestive tracts (29, 30). The gyrA gene was sequenced for 134 C. jejuni isolates that exhibited phenotypic resistance to ciprofloxacin and/or nalidixic acid. Ciprofloxacin resistance was largely conferred by a Thr86Ile mutation in the gyrA gene, but greater resistance was observed in those C. jejuni isolates that also possessed A21G and T72C mutations. Although the dominance of the gyrA Thr86Ile mutation in resistance to fluoroquinolones in C. jejuni is recognized, the role that silent mutations play in resistance is not well understood at present (31–33). A total of 19 isolates were resistant to ciprofloxacin and/or nalidixic acid but did not possess an amino acid substitution (Thr86Ile) in the QRDR of the gyrA gene, nor did they possess either of two other recognized factors known to confer fluoroquinolone resistance. Too few isolates were resistant to ciprofloxacin and/or nalidixic acid in the absence of recognized resistance factors for statistical analysis, but the unknown resistance factor may be primarily associated with resistance to nalidixic acid with only a secondary effect on ciprofloxacin resistance.

We used the CGF method (34) to subtype a large number of C. jejuni isolates (n = 1,832) collected from four CFOs over the 18-month study period. We found high subtype diversity among C. jejuni isolates from cattle at each CFO, and there was no significant change in subtype diversity between or within CFOs at the arrival and interim sample times. Others have shown that C. jejuni strains present in cattle are highly genetically diverse and that a relatively small number of highly prevalent subtypes predominate (8, 35, 36); the findings of the current study confirm this observation across a much larger data set. We identified five CGF subtype clusters that were most prevalent among C. jejuni strains isolated from beef cattle. The predominant CGF clusters 169.001 and 238.002 roughly correspond to multilocus sequence typing (MLST) clonal complex sequence type 21 (ST-21), cluster 695.006 with clonal complex ST-61, cluster 735.001 with clonal complex ST-42, and cluster 853.011 with clonal complex ST-403. Clonal complex ST-21 is considered a generalist group because it is commonly detected in multiple host species and environments, while ST-42, ST-61, and ST-403 are all clonal complexes predominantly associated with C. jejuni isolated from cattle (36, 37). The population structures of C. jejuni in arrival samples were similar across CFOs, but the population structures for CFO1 and, to a lesser extent, CFO2, differed significantly from other CFOs at the interim sample time. In the current study, all samples were stored at 5°C after collection, but the time from collection to processing varied (≤9 days). Although we did not detect a conspicuous difference in C. jejuni isolation prevalence among samples, it is not known what impact if any the differential duration of storage at 5°C biased isolation of subtypes, including those resistant to AMAs. Although all CFOs included in the current study follow industry guidelines for animal care, the uneven administration of different AMAs (both therapeutically and nontherapeutically) at different doses and rates to different individuals in an ever-shifting CFO cattle population hindered identification of the factors involved in shaping the unique population structure of C. jejuni at CFO1. Nonetheless, study findings provide important information that can be used in the design of future studies to elucidate the transmission dynamics of C. jejuni subtypes in beef cattle production and transmission to human beings.

Previous studies indicated that the use of AMAs as AGPs in livestock increases the antimicrobial resistance of C. jejuni over time (6, 18, 19), but few examined changes in antimicrobial resistance over time at a subspecies level. We found no difference in subtype diversity between isolates resistant or susceptible to tetracycline. However, the proportion of isolates from four of the five predominant subtype clusters that were resistant to tetracycline increased greatly from the arrival to interim sample time, and the vast majority of isolates from the fifth predominant cluster were already resistant to tetracycline upon arrival. In comparison, isolates resistant to ciprofloxacin and/or nalidixic acid were less diverse than their susceptible counterparts, and only predominant C. jejuni subtype cluster 234 demonstrated a significant increase in the overall proportion of isolates resistant to ciprofloxacin and nalidixic acid over time. These findings suggest that the rate at which C. jejuni strains develop resistance to tetracycline is subtype independent, and resistance to ciprofloxacin and nalidixic acid is subtype dependent. A previous study reported evidence for plasmid-mediated horizontal transfer of the tetO resistance determinant between C. jejuni strains from chickens (29), and tetracycline resistance in C. jejuni strains from cattle is likely conferred in a similar fashion. A recent cross-sectional study of beef and dairy cattle in Michigan also examined genetic diversity and identified predominant subtypes within cattle herds consistent with our findings (9); however, the longitudinal nature of our study allowed us to further compare the change in population subtype diversity, prevalence of specific predominant subtype clusters, and cluster specific antimicrobial resistance over time. Notably, Cha et al. (9) observed subtype-specific C. jejuni resistance to ciprofloxacin, nalidixic acid, and tetracycline, which we did not observe for tetracycline in our study. It is possible that our different result regarding tetracycline resistance is due to variations in sample size, sample period length, and/or cattle production systems between the two studies (e.g., density of animals, husbandry, and AMA administration). We subtyped 1,832 isolates by CGF and characterized antimicrobial resistance (AMR) in 1,648 C. jejuni isolates by agar dilution, the gold standard method (compared to the 110 isolates that were examined by Cha et al. [9]), which allowed us to perform a more robust comparison of subtype-dependent C. jejuni AMR in both time (arrival versus interim) and space (individual CFO). We not only observed subtype-specific differences in C. jejuni AMR but also increases in the proportion of individuals resistant to AMAs for specific subtypes. These findings highlight how longitudinal sampling at a subspecies level in a relatively large number of isolates can provide additional insights on antimicrobial resistance in C. jejuni within cattle production systems. Importantly, these data can help ascertain risk and create effective guidelines governing the use of AMAs in specific production systems (e.g., in CFO cattle).

In conclusion, we observed high subtype diversity in C. jejuni strains isolated from beef cattle in Southern Alberta. Although the use of AMAs, particularly in feed as AGPs, was expected to select for antimicrobial resistance in C. jejuni colonizing the intestine of CFO cattle, selection did not reduce the overall subtype diversity nor bias the C. jejuni population structure toward the presence, abundance, or dominance of specific C. jejuni subtypes. Evidence indicated that phenotypic and genetic resistance to nalidixic acid and ciprofloxacin was subtype dependent. Although we cannot exclude the possibility that spontaneous mutations conferred resistance, evidence suggests that the majority of C. jejuni subtypes resistant to nalidixic acid and ciprofloxacin entered the CFO at low frequencies in the intestines of cattle and proliferated during the CFO maintenance period in the absence of selection pressure. Consistent with this observation, some reports indicate that there may be no fitness cost to carrying mutations in the gyrA gene that confer resistance to fluoroquinolones (38). Similarly to nalidixic acid and ciprofloxacin, a portion of animals entering the CFO shed C. jejuni subtypes resistant to tetracycline at low frequencies. In contrast, the increased prevalence of tetracycline resistance within the CFOs was not subtype dependent, consistent with widespread horizontal transmission of tetO determinants and subsequent proliferation of diverse C. jejuni subtypes under selection pressure (contrary to our hypothesis). We conclude that a small proportion of cattle entering a CFO carry C. jejuni, including subtypes resistant to AMAs. The large numbers of cattle arriving from diverse locations at the CFOs and intermingling within the CFOs over time, coupled with the high-density housing of animals, the high rates of transmission of C. jejuni subtypes among animals, and the extensive use of AMAs therapeutically, metaphylactically, and as AGPs, merge to create a situation where the proliferation of diverse antimicrobial-resistant C. jejuni subtypes is facilitated.

MATERIALS AND METHODS

Beef cattle.

Candidate crossbred beef steer and heifers utilized in the study were typically procured through the auction market system across western Canada and were maintained in one of four commercial CFOs in south central Alberta (designated CFO1, CFO2, CFO3, and CFO4) under production conditions that are typical of those used at large commercial cattle CFOs throughout western Canada and the United States. The four CFOs included in the study possessed feeding capacities ranging from 15,000 to 20,000 animals. Animals were housed in open-air pens with soil floors arranged side-by-side with central feed alleys, and individual pens held 50 to 350 animals. Upon arrival at a CFO, a two-stage random sampling plan was used to determine pen assignments and the selection of individual animals for enrollment in the study. Once the study began at each CFO, 30% (three pens selected out of every 10 pens) of all new pens were randomly selected for the survey using a pen randomization table provided by Feedlot Health Management Systems (FHMS; Okotoks, Alberta, Canada).

Antimicrobial agent administration.

The dosages of all AMAs used in the study were those recommended by the drug manufacturer and approved by the Veterinary Drugs Directorate of Health Canada. At all CFOs, animals were administered AMAs according to standard industry practices, and AMA administrations were recorded.

Fecal sample collection.

Within selected pens, ≈10% of the animals were randomly selected using a randomization table. The collection of fecal samples from individual animals commenced on 27 August 2008 and ended on 21 April 2010. A total of 4,427 beef cattle (9.9% of 44,760 cattle) in four commercial CFOs (216 pens) were sampled upon arrival (termed “arrival”) and after >60 days in the CFO (termed “interim”). The average numbers of animals per pen were 267.2, 112.3, 206.4, and 365.0 at CFO1, CFO2, CFO3, and CFO4, respectively. In total, 9,557 individual fecal samples from beef cattle were processed (4,519 for arrival samples and 4,138 for interim samples) over a 602-day period. The average times between collection of arrival and interim fecal samples from cattle were 104.4 ± 35.1, 106.9 ± 34.5, 69.8 ± 20.3, and 80.4 ± 20.4 days at CFO1, CFO2, CFO3, and CFO4, respectively, with an overall mean time of 90.3 ± 9.1 days. A new palpation sleeve (no. 33; Almedic, Montreal, Quebec, Canada) was used to collect feces from the rectum of each animal, and the collected feces (minimum of 4 g) was transferred into a labeled sterile sample container and refrigerated at ≈5°C. Fecal samples were transported to FHMS within 7 days of collection. At FHMS, sample information was recorded and shipped once a week (on ice in a cooler) by overnight courier (Greyhound Courier Express, Greyhound Canada Transportation Corp., Calgary, Alberta, Canada) to the Agriculture and Agri-Food Canada Research and Development Centre at Lethbridge, Alberta, Canada.

The collection of composite fecal samples from the floors of pens commenced on 27 August 2008 and ended on 26 May 2010. Once a pen of animals had been filled and individual animals enrolled in the study, pen-level composite fecal samples were collected at three time points: at arrival, at interim, and ≈30 days or less prior to shipment (termed “exit”). A total of 640 composite pen manure samples from the 216 pens were processed. The average times between collection of the arrival and interim pen samples and between collection of the interim and exit samples were 89.1 ± 10.2 and 66.9 ± 11.2 days, respectively. The collection of the composite arrival and interim fecal samples from pens was timed to be as close as possible to the collection dates for individual animals. To collect composite fecal samples from pens, a clean spoon was used to combine feces from 20 fresh pen floor pats (minimum of 10 g from each pat) into a plastic container for each composite sample. Samples were stored ≈5°C and shipped as described above.

Isolation of Campylobacter species.

Upon arrival at the laboratory, fecal samples from cattle were stored at 4°C and processed within 1 day. A subsample of feces (≈1 g) was suspended in Columbia broth (Oxoid, Nepean, Ontario, Canada), tubes were thoroughly mixed using a vortexer (high setting), and 25 μl of the resultant slurry was spread onto Campylobacter blood-free selective agar base (modified charcoal cefoperazone desoxycholate agar [CCDA]; Oxoid) containing selective supplement SR155 (Oxoid). All cultures were incubated at 42°C in a microaerobic environment consisting of 10% CO₂, 5% O₂, 3% H₂, and 82% N₂. At 48 h, two representative colonies for each unique colony type per culture were transferred to CCDA not containing the selective supplement and streaked for purity. All cultures were incubated for 48 to 72 h, and the biomass of each isolate was suspended in Columbia broth containing 30% glycerol and placed at −80°C.

Identification of Campylobacter species.

The genomic DNA of the isolates was extracted using an AutoGen 740 robot (Holliston, MA), according to the manufacturer's protocol, and genomic DNA was subjected to diagnostic PCR for the Campylobacter genus (16S rRNA gene), C. jejuni (hipO), C. coli (aspartokinase gene), C. fetus (23S rRNA gene), and C. hyointestinalis (23S rRNA gene) (Table 7). The partial 16S rRNA gene of isolates not definitely identified by taxon-specific PCR was sequenced and subjected to BLASTN analysis (National Center for Biotechnology Information, Bethesda, MD) (39).

TABLE 7.

Primer sequences for the identification of Campylobacter species and for the amplification of DNA encoding tetracycline resistance determinants and the gyrase A gene in Campylobacter jejuni isolates recovered from beef cattle

Taxon	Target	Primer	Ta (°C)a	Sequence (5′→3′)	Length (bp)	Reference or source
Campylobacter genus	16S rRNA	C412F	58	GGATGACACTTTTCGGAGC	816	50
		C1228R		CATTGTAGCACGTGTGTC
C. jejuni	hipO	CjhipOF	60	AAATAGGAAAAACAGGCGTTGT	695	51
		CjhipOR		TATCATTAGCCTGTGCAAGACC
C. coli	Aspartokinase gene	CC18F	60	GGTATGATTTCTACAAAGCGAG	500	52
		CC519R		ATAAAAGACTATCGTCGCGTG
C. fetus	23S rRNA	CF441F	54	GTTAGGGAAGAACAATGACGG	554	53
		CF995r		TTATCTCTAAGAGATTAGTTGG
C. hyointestinalis	23S rRNA	HYO1F	54	ATAATCTAGGTGAGAATCCTAG	611	54
		HYOFET23SR		GCTTCGCATAGCTAACAT
C. jejuni	tetB	tetB1-f	54	ACACTCAGTATTCCAAGCCTTTG	208	55
		tetB1-r		GATAGACATCACTCCCTGTAATGC
C. jejuni	tetC	tetC-fw	55	CTTGAGAGCCTTCAACCCAG	418	56
		tetC-rv		ATGGTCGTCATCTACCTGCC
C. jejuni	tetL	tetL-sense	56	GGTTTTGAACGTCTCATTACCTGAT	106	55
		tetL-antisense		CCAATGGAAAAGGTTAACATAAAGG
C. jejuni	tetM	tetM-fw	50	ACAGAAAGCTTATTATATAAC	171	56
		tetM-rv		TGGCGTGTCTATGATGTTCAC
C. jejuni	tetO	tetO-fw	45	ACGGARAGTTTATTGTATACC	171	56
		tetO-rv		TGGCGTATCTATAATGTTGAC
C. jejuni	tetQ	tetQ-fw	55	AGAATCTGCTGTTTGCCAGTG	169	56
		tetQ-rv		CGGAGTGTCAATGATATTGCA
C. jejuni	gyrA	GyrAFcj	56	GCTCTTGTTTTAGCTGATGCA	620	This study
		GyrARcj		TTGTCGCCATACCTACAGCTA

^aT_a, annealing temperature.

Susceptibility to antimicrobial agents.

The susceptibilities of representative C. jejuni strains from cattle (target of 20% of isolates recovered from individual cattle at arrival and interim sample times) to seven AMAs were quantified. MICs to ciprofloxacin, chloramphenicol, clindamycin, erythromycin, gentamicin, nalidixic acid, and tetracycline were determined using the agar dilution methodology according to the Clinical and Laboratory Standards Institute (40), with the exception that the Mueller-Hinton agar (Difco, Sparks, MD) was not amended with 5% defibrinated horse blood. Campylobacter cells were harvested from the surface of the medium after 48 h growth under microaerobic conditions at 37°C. Cells were suspended in sterile NaCl (0.075%), and the density of cells was adjusted to a 0.5 McFarland standard by spectrophotometry at 625 nm (Genesys 20; Thermo Scientific, Rockford, IL). Aliquots (300 μl) of the saline suspension were pipetted into the seeding wells of a Cathra replicator (Oxoid). Freshly prepared plates of Mueller-Hinton agar amended with AMAs were then inoculated using 1-mm pins in the inoculating head of the replicator. Cultures were incubated microaerobically at 37°C for 48 h, and the MIC was defined as the lowest concentration resulting in complete inhibition of visible growth on the medium. Campylobacter jejuni (ATCC 33560) was utilized as a quality control strain. The breakpoint values for resistance (intermediate resistance breakpoint in parentheses) were ≥4 μg · ml⁻¹ (2 μg · ml⁻¹) for ciprofloxacin, ≥32 μg · ml⁻¹ (16 μg · ml⁻¹) for chloramphenicol, ≥8 μg · ml⁻¹ (4 μg · ml⁻¹) for clindamycin, ≥32 μg · ml⁻¹ (16 μg · ml⁻¹) for erythromycin, ≥8 μg · ml⁻¹ (4 μg · ml⁻¹) for gentamicin, ≥64 μg · ml⁻¹ (32 μg · ml⁻¹) for nalidixic acid, and ≥16 μg · ml⁻¹ (8 μg · ml⁻¹) for tetracycline (41). As a confirmation, the MICs for isolates deemed resistant were repeated.

Genetics of tetracycline and fluoroquinolone resistance.

All isolates resistant to tetracycline from the arrival sample time were assessed, whereas a subset of 306 arbitrarily selected isolates (≈25% of the isolates) were evaluated from the interim sample time, ensuring representation across the four CFOs and collection times. Three C. jejuni strains were included as controls; one strain was resistant to ciprofloxacin and tetracycline, one strain was resistant to tetracycline only, and one strain was susceptible to both ciprofloxacin and tetracycline. The presence of tetB, tetC, tetL, tetM, tetO, tetQ, and tetW was determined by PCR. Briefly, each 20-μl PCR mixture contained 2 μl of DNA (20 to 50 ng), 2 μl of 10× PCR buffer, 0.4 μl of 10 mM deoxynucleoside triphosphates (dNTP), 0.4 μl of 25 mM MgCl, 1 μl each of forward and reverse primers (10 μM; Integrated DNA Technologies, Coralville, IA) (Table 7), 2 μl of bovine serum albumin (BSA; Life Technologies, Inc., Burlington, Ontario, Canada), 0.1 μl of HotStar Taq polymerase (Qiagen, Inc., Toronto, Ontario, Canada), and 11.1 μl of Optima water (Fisher Scientific, Ottawa, Ontario, Canada). The cycle conditions were one cycle at 95°C for 15 min; 30 cycles at 94°C for 30 s, at the appropriate annealing temperature (T_a) for 1 min, and at 72°C for 1 min; one cycle at 72°C for 5 min; and a hold at 4°C. Amplicons were visualized by capillary electrophoresis using a QIAxcel system (Qiagen, Inc.).

The genetics of resistance to quinolones and fluoroquinolones was examined in all C. jejuni strains that were resistant to ciprofloxacin and/or nalidixic acid. Three C. jejuni strains were included as controls, as described above. Amplification and sequencing of the gyrA gene corresponding to codons 1 through 177 were conducted. Primers for the gyrA gene were modified from those described by Jesse et al. (42). Modifications to the primers were made to encompass more C. jejuni strains based on new sequence data from the National Center for Biotechnology Information (Table 7); the software program Geneious (Biomatters Ltd., Auckland, New Zealand) was used. The conditions for PCR amplification were 1 cycle at 95^o for 15 min; 35 cycles at 94^o for 30 s, 56°C for 60 s, and 72^o for 60 s; and extension for 10 min at 72°C. The same PCR mixture as described above was used. Each PCR was performed with 2 μl of a 10⁻¹ dilution of genomic DNA. The resulting gyrA amplicons were sequenced in-house using an ABI 3130 genetic analyzer (Life Technologies, Carlsbad, CA) with POP7 and BigDye 3.1 chemistry. Further sequencing was done off-site by Macrogen USA (Rockville, MD). Sequence alignments and analyses, including the identifications of mutations that would result in amino acid substitutions (Thr86Ile), were completed using Geneious (Biomatters Ltd.).

In isolates that did not exhibit mutations that would confer an amino acid substitution in the gyrA gene, other genetic resistance determinants were investigated. The presence of PMQR was investigated with methods designed for non-Campylobacter bacteria (43–47); slight modifications to primers were made to enhance comprehensiveness. In addition, potential mutations in the cmeABC operon were investigated; CmeABC is a resistance-nodulation-division type of efflux pump, and a C-to-T point mutation in the binding site for the cmeR transcriptional regulator may lead to increased fluoroquinolone resistance in Campylobacter (48).

Comparative genomic fingerprinting.

Campylobacter jejuni isolates from cattle were subtyped using a semiautomated 40-locus CGF method (34). Briefly, eight-multiplex PCRs were performed for each C. jejuni isolate; each five-multiplex reaction mixture contained 1 U Fisherbrand Taq DNA polymerase (Fisher Scientific), 1× buffer, 2.5 mM MgCl₂, 0.2 mM each deoxynucleoside triphosphate (dNTP), 0.12 to 0.74 μM the 10 primers (of note, primer concentration was optimized to produce a strong amplicon for each primer set in the multiplex), and 1 μl of DNA template (20 to 100 ng) in a 25-μl reaction mixture. An EP Gradient Mastercycler (Eppendorf, Mississauga, Ontario, Canada) was used, and PCR conditions were an initial denaturation at 94°C for 5 min; 35 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s; and a final extension step at 72°C for 5 min. Amplicons were resolved using a QIAxcel high-throughput capillary electrophoresis system with DNA screening cartridges (Qiagen, Inc.), using the AM320 separation method and a 20-s injection time. The 15- to 3,000-bp alignment marker and a 100- to 2.5-kb size ladder were used as size standards (Qiagen, Inc.).

Data analysis.

Most analyses were conducted using Statistical Analysis Software (SAS; Cary, NC). In order to determine if significant count shifts occurred among the two sample times, the Genmod procedure from SAS was used to perform a log-linear analysis for each AMA using frequency counts as the dependent variable and MIC level, sample time, and their interaction as the factors in the model; only MIC levels that had at least one frequency for each sample time were used in these analyses. When a significant treatment effect was observed, contrast statements were used to evaluate differences among means of interest. Median MICs for each AMA were calculated from cumulative susceptibility data. Log-linear analysis also was performed to determine if significant differences existed in resistance (as defined by breakpoints) to each AMA between the sampling times; natural-log transformation also was used as an offset variable for total counts, and contrast statements were used to evaluate differences among means of interest when a significant treatment effect was observed. The frequency procedure of SAS was used to perform the chi-square test with Fisher's exact test in order to assess the relationship among categorical variables at the sample times; analyses were restricted to ciprofloxacin, nalidixic acid, and tetracycline.

To analyze the subtype diversity of C. jejuni, isolates were assigned arbitrary subtype cluster designations based on 95% or greater CGF profile similarity, as calculated using the simple matching analysis coefficient with unweighted pair group method using average linkages (UPGMA) clustering in BioNumerics (version 6.6; Applied Maths, Austin, TX). Randomized resampling was performed to normalize sample size, and cluster richness and abundance were used to calculate the Shannon diversity index. Hutcheson's t test was used to test the significance of differences in subtype diversity (49). Subtype clusters shared by a large number of isolates (i.e., the five most predominant clusters) were identified, and associations were measured using the chi-square test of independence; in cases where expected frequencies were too low for the traditional chi-square test, Yates' correction was employed.

To compare the structure of C. jejuni populations (i.e., the presence and abundance of specific subtypes) within and between CFOs, a similarity matrix was generated for isolates using the simple matching coefficient in BioNumerics (version 6.6; Applied Maths), and permutational multivariate analysis of variance (PERMANOVA) analyses were used to determine significant differences in population structure in Primer 7 (version 7.0.11, Quest Research Limited, Auckland, New Zealand). PERMANOVA analyses consisted of 9,999 main test and pairwise unrestricted permutations of raw data with partial sums of squares. Population structure was visualized using a minimum spanning tree in BioNumerics (version 6.6; Applied Maths).

For comparison with previous studies, representative CGF profiles of predominant clusters were queried against the Canadian Campylobacter CGF database in order to convert the arbitrary cluster identifiers into reference CGF and MLST database identifiers.