High Prevalence of Fluoroquinolone-Resistant Campylobacter Bacteria in Sheep and Increased Campylobacter Counts in the Bile and Gallbladders of Sheep Medicated with Tetracycline in Feed

Campylobacter is a major cause of foodborne illness in humans, and antibiotic-resistant Campylobacter is considered a serious threat to public health in the United States and worldwide. As a foodborne pathogen, Campylobacter commonly exists in the intestinal tract of ruminant animals, such as sheep and cattle. Results from this study reveal the predominant genotypes and high prevalence of tetracycline (TET) and fluoroquinolone (FQ) resistance in sheep Campylobacter. The finding on fluoroquinolone resistance in sheep Campylobacter is unexpected, as this class of antibiotics is not used for sheep in the United States, and it may suggest the transmission of fluoroquinolone-resistant Campylobacter from cattle to sheep. Additionally, the results demonstrate that in-feed medication with tetracycline increases Campylobacter counts in gallbladders, suggesting that the antibiotic promotes Campylobacter colonization of the gallbladder. These findings provide new information on Campylobacter epidemiology in sheep, which may be useful for curbing the spread of antibiotic-resistant Campylobacter in animal reservoirs.

ABSTRACT

Campylobacter is a major foodborne pathogen in humans and a significant cause of abortion in sheep. Although ruminants are increasingly recognized as important reservoirs for Campylobacter species, limited information is available about the molecular epidemiology and antimicrobial resistance (AMR) profiles of sheep Campylobacter. Here, we describe a two-trial study that examined Campylobacter profiles in sheep and determined whether in-feed tetracycline (TET) influenced the distribution and AMR profiles of Campylobacter. Each trial involved 80 commercial sheep naturally infected with Campylobacter: 40 of these sheep were medicated with tetracycline in feed, while the other 40 received feed without antibiotics. Fecal and bile samples were collected for the isolation of Campylobacter. The bacterial isolates were analyzed for antimicrobial susceptibility and genotypes. The results revealed that 87.0% and 61.3% of the fecal and bile samples were positive for Campylobacter (Campylobacter jejuni and Campylobacter coli), with no significant differences between the medicated and nonmedicated groups. All but one of the tested Campylobacter isolates were resistant to tetracycline. Although fluoroquinolone (FQ) resistance remained low in C. jejuni (1.7%), 95.0% of the C. coli isolates were resistant to FQ. Genotyping revealed that C. jejuni sequence type 2862 (ST2862) and C. coli ST902 were the predominant genotypes in the sheep. Feed medication with tetracycline did not affect the overall prevalence, species distribution, and AMR profiles of Campylobacter, but it did increase the total Campylobacter counts in bile and gallbladder. These findings identify predominant Campylobacter clones, reveal the high prevalence of FQ-resistant C. coli, and provide new insights into the epidemiology of Campylobacter in sheep.

IMPORTANCE Campylobacter is a major cause of foodborne illness in humans, and antibiotic-resistant Campylobacter is considered a serious threat to public health in the United States and worldwide. As a foodborne pathogen, Campylobacter commonly exists in the intestinal tract of ruminant animals, such as sheep and cattle. Results from this study reveal the predominant genotypes and high prevalence of tetracycline (TET) and fluoroquinolone (FQ) resistance in sheep Campylobacter. The finding on fluoroquinolone resistance in sheep Campylobacter is unexpected, as this class of antibiotics is not used for sheep in the United States, and it may suggest the transmission of fluoroquinolone-resistant Campylobacter from cattle to sheep. Additionally, the results demonstrate that in-feed medication with tetracycline increases Campylobacter counts in gallbladders, suggesting that the antibiotic promotes Campylobacter colonization of the gallbladder. These findings provide new information on Campylobacter epidemiology in sheep, which may be useful for curbing the spread of antibiotic-resistant Campylobacter in animal reservoirs.

INTRODUCTION

Campylobacter species, particularly Campylobacter jejuni and Campylobacter coli, are a leading cause of bacterial foodborne gastroenteritis in humans around the world (1–3). Among all the causes of laboratory-confirmed bacterial foodborne illnesses, Campylobacter was the leading cause (19.2 per 100,000 population) in the year 2017, based on a report by the Centers for Disease Control and Prevention (CDC) FoodNet surveillance program in the United States (4). Although most Campylobacter-related illnesses in humans are characterized as self-limiting diarrhea (watery and/or bloody), antibiotic treatment is utilized in more severe clinical conditions, especially in young, elderly, or immunocompromised patients (5). However, antimicrobial resistance in Campylobacter is increasingly prevalent and has become a major public health concern in both developed and developing counties (6, 7). Of particular concern is the rising resistance to fluoroquinolone (FQ) and macrolide antibiotics, which are clinically important for the treatment of Campylobacter-induced diarrhea in humans, although in some cases tetracyclines (TET) and gentamicin (GEN) are also used to treat systemic infection caused by Campylobacter (6, 8, 9).

Food-producing animals are important reservoirs for Campylobacter (10). Epidemiologically, poultry meat is considered the major source of infection for human campylobacteriosis (11). C. jejuni and C. coli frequently colonize the intestines of many species of poultry, especially commercial chickens and turkeys, usually as part of the normal microbial flora without causing clinical disease (12). Recently, the role of ruminants (cattle and sheep) in Campylobacter ecology has been increasingly recognized in different countries (13–17). Ruminant Campylobacter strains can be transmitted to humans via contaminated milk and water, environmental contamination, or direct contact with animals (18). Source attribution studies indicate that ruminant Campylobacter is a significant source of infection for human campylobacteriosis (19–21).

Typically, Campylobacter colonizes the intestinal tract in animals, but it may also translocate across the intestinal epithelial barrier and causes systemic infection, leading to bacteremia and abortion in ruminants and even occasionally in humans (1, 22). In fact, Campylobacter is the major cause of ovine abortions worldwide, including in the United States. Notably, a single hypervirulent tetracycline-resistant C. jejuni clone (named clone SA, for sheep abortion) is responsible for the majority of Campylobacter-associated ovine abortions in the United States (17, 23, 24). A previous survey of healthy sheep in a slaughterhouse revealed genetically diverse C. jejuni strains (including clone SA) present in the intestinal tract and gallbladder (25). Despite the fact that ruminants are increasingly recognized as important reservoirs for Campylobacter, limited information is available about the molecular epidemiology and antimicrobial resistance profiles of sheep Campylobacter in the United States.

For the control of abortion and other diseases, the tetracycline class of antibiotics is commonly used in the United States, and it is the only class of antibiotics approved for prevention and control of sheep abortion associated with Campylobacter (23). However, whether the administration of tetracycline in feed impacts the prevalence of Campylobacter in sheep is not known. Previously, it was shown that sheep harbored diverse strains of C. jejuni that were commonly resistant to tetracycline; however, resistance to FQ was seldomly detected (25). Recently, FQ resistance in cattle Campylobacter isolates has increased substantially in the United States (15, 26–28). However, detection of FQ resistance in sheep Campylobacter isolates is still limited. To address these knowledge gaps, we collected samples and isolated Campylobacter from controlled treatment (feed medication with tetracycline) studies using sheep that were derived from commercial operations and were naturally infected by Campylobacter. The goals were to examine the extent of FQ resistance in sheep Campylobacter and determine the effect of in-feed tetracycline on the prevalence, species distribution, and antimicrobial resistance profiles of Campylobacter in sheep.

RESULTS

Prevalence of Campylobacter in sheep.

In total, 461 sheep fecal samples and 160 bile samples were collected for isolation of Campylobacter in the two separate trials. The overall prevalence rate of Campylobacter in sheep feces was 87.0% (401/461). Among the positive samples, 68.6% (275/401) and 37.4% (150/401) were positive for C. jejuni and C. coli, respectively, and 24 fecal samples were positive for both C. jejuni and C. coli (Table 1). The overall prevalence of Campylobacter in fecal samples was comparable (71.8% versus 65.3% for C. jejuni; 33.2% versus 41.7% for C. coli) between the feed-medicated and nonmedicated groups (P > 0.05) (Table 1). Among the bile samples, 61.3% (98/160) were positive for Campylobacter (Table 2). Of the Campylobacter isolates, 98.0% (96/98) were C. jejuni, and 2.0% (2/98) were C. coli. The two C. coli isolates from bile were both isolated from the feed-medicated group in trial 2 (Table 2). Again, there was no significant difference (P > 0.05) between the groups in the prevalence of either C. jejuni (100% versus 95.8%) or C. coli (0% versus 4.2%). These results indicate that the in-feed tetracycline medication did not affect the overall prevalence of Campylobacter in fecal and bile samples.

TABLE 1.

Isolation rates of Campylobacter in sheep fecal samples

Isolate group	No. (%) of isolates in triala:
	1 (21 June 2017–6 July 2017)									2 (13 July 2017–26 July 2017)									All fecal samples
	1st week (21 June)			2nd week (29 June)			3rd week (6 July)			1st week (13 July)			2nd week (21 July)			3rd week (26 July)
	NM	M	T	NM	M	T	NM	M	T	NM	M	T	NM	M	T	NM	M	T	NM	M	Tb
All isolates	34 (52.3)	31 (47.7)	65 (100)	34 (50.0)	34 (50.0)	68 (100)	32 (51.6)	30 (48.4)	62 (100)	36 (51.4)	34 (48.6)	70 (100)	33 (50.0)	33 (50.0)	66 (100)	33 (47.1)	37 (52.9)	70 (100)	202 (50.4)	199 (49.6)	401 (100)
C. jejuni	30 (58.8)	21 (41.2)	51 (78.5)	27 (49.1)	28 (50.9)	55 (80.9)	20 (46.5)	23 (53.5)	43 (69.4)	19 (51.4)	18 (48.6)	37 (52.9)	24 (55.8)	19 (44.2)	43 (65.2)	25 (54.3)	21 (45.7)	46 (65.7)	145 (52.7)	130 (47.3)	275 (68.6)
C. coli	5 (29.4)	12 (70.6)	17 (26.2)	8 (47.1)	9 (52.9)	17 (25.0)	13 (56.5)	10 (43.5)	23 (37.1)	19 (52.8)	17 (47.2)	36 (51.4)	10 (40.0)	15 (60.0)	25 (37.9)	12 (37.5)	20 (62.5)	32 (45.7)	67 (44.7)	83 (55.3)	150 (37.4)

^aNM, nonmedicated group; M, medicated group; T, total.

^b24 fecal samples were positive both for C. jejuni and C. coli.

TABLE 2.

Overall prevalence of Campylobacter in sheep bile samples^a

Isolate group	No. (%) of isolates in trialb :
	1 (21 June 2017–6 July 2017)			2 (13 July 2017–26 July 2017)			All bile samples
	NM	M	T	NM	M	T	NM	M	T
Total	28 (50.9)	27 (49.1)	55 (100)	22 (51.2)	21 (48.8)	43 (100)	50 (51.0)	48 (49.0)	98 (100)
C. jejuni	28 (50.9)	27 (49.1)	55 (100)	22 (53.7)	19 (46.3)	41 (95.3)	50 (52.1)	46 (47.9)	96 (98.0)
C. coli	0 (0)	0 (0)	0 (0)	0 (0)	2 (100.0)	2 (4.7)	0 (0.0)	2 (100.0)	2 (2.0)

^aBile samples were collected at necropsy at the end of the two trials.

^bNM, nonmedicated group; M, medicated group; T, total.

Statistically, the overall Campylobacter prevalence was higher in fecal samples than in bile samples (87.0% versus 61.3%, P < 0.05). However, the proportion of C. jejuni was significantly (P < 0.05) lower in feces than in bile (68.6% versus 98.0%), while the proportion of C. coli was higher in feces than in bile (37.4% versus 2.0%). Detailed information about the prevalence of Campylobacter and the distribution of C. jejuni and C. coli is summarized in Tables 1 and 2. We analyzed all of the C. jejuni isolates from feces and bile by PCR for identification of C. jejuni clone SA, and none were positive, indicating that clone SA was not present in the sheep used in this study.

Quantitative measurement of Campylobacter CFU in bile and gallbladder mucosa.

Significant differences in Campylobacter numbers (CFU/ml) between the feed-medicated and nonmedicated groups were detected for both the bile and gallbladder mucosal samples, with numbers higher in the medicated group (Fig. 1). In the bile samples, the mean log CFU/ml ± standard error of the mean (SEM) values were 5.041 ± 0.281 and 6.021 ± 0.108 in the nonmedicated and medicated groups, respectively, indicating an almost 10-fold difference. The difference between the two groups was statistically significant [t_(3.252,6) = 0.0174, P < 0.05]. For the gallbladder mucosal samples, the mean log CFU/ml ± SEM of gallbladder mucosal samples were 4.7239 ± 0.2932 and 5.823 ± 0.2611 in the nonmedicated and medicated groups, respectively, again indicating an approximately 10-fold difference. The difference was also statistically significant [t_(2.722,57) = 0.0086, P < 0.05]. These results indicated that Campylobacter titers are higher in the gallbladder and bile samples taken from the feed medicated group.

FIG 1.

Campylobacter counts (log CFU/ml) in bile (a) and gallbladder mucosal (b) samples in the nonmedicated and medicated groups. Each bar represents the average log CFU per ml (± standard error of the mean [SEM]). An asterisk (*) indicates a significant difference from the nonmedicated group (P < 0.05). The data include samples from both trials.

Antimicrobial susceptibility of the Campylobacter isolates.

For the tested fecal C. jejuni isolates (n = 236), all (100%) were resistant to tetracycline, but the rates of resistance to ciprofloxacin (CIP) and nalidixic acid (NAL) were low, at 1.7% (4/236) and 3.0% (7/236), respectively. All four ciprofloxacin-resistant (CIP^r) C. jejuni isolates were derived from trial 2. One C. jejuni isolate from the nonmedicated group in trial 1 showed resistance to gentamicin. The fecal C. coli isolates also showed a high resistance rate to tetracycline (99.0%; 100/101), with only 1 isolate sensitive to this antibiotic. In contrast to C. jejuni, the C. coli isolates were highly resistant to ciprofloxacin (95.0%; 96/101) and nalidixic acid (95.0%; 96/101). Resistance to the other five antibiotics was not observed. The key MIC results are shown in Table 3, while other relevant MIC data (MIC ranges, MIC₅₀, and MIC₉₀) are presented in Table S1.

TABLE 3.

Antimicrobial resistance profiles and rates of the tested Campylobacter isolates from fecal samples

Trial no.	Species	Medication group	Isolates (% [no./total tested]) resistant to:
			Tetracycline	Ciprofloxacin	Nalidixic acid	Gentamicin
1	C. jejuni	Nonmedicated	100.0 (60/60)	0.0 (0/60)	3.3 (2/60)	1.7 (1/60)
		Medicated	100.0 (60/60)	0.0 (0/60)	5.0 (3/60)	0.0 (0/60)
		Total	100.0 (120/120)	0.0 (0/120)	4.2 (5/120)	0.8 (1/120)
	C. coli	Nonmedicated	100.0 (25/25)	100.0 (25/25)	100.0 (25/25)	0.0 (0/25)
		Medicated	100.0 (25/25)	100.0 (25/25)	100.0 (25/25)	0.0 (0/25)
		Total	100.0 (50/50)	100.0 (50/50)	100.0 (50/50)	0.0 (0/50)
2	C. jejuni	Nonmedicated	100.0 (58/58)	1.7 (1/58)	3.4 (2/58)	0.0 (0/58)
		Medicated	100.0 (58/58)	5.2 (3/58)	0.0 (0/58)	0.0 (0/58)
		Total	100.0 (116/116)	3.4 (4/116)	1.7 (2/116)	0.0 (0/116)
	C. coli	Nonmedicated	96.0 (24/25)	88.0 (22/25)	88.0 (22/25)	0.0 (0/25)
		Medicated	100.0 (26/26)	92.3 (24/26)	92.3 (24/26)	0.0 (0/26)
		Total	98.0 (50/51)	90.2 (46/51)	90.2 (46/51)	0.0 (0/51)
Both trials	C. jejuni	Nonmedicated	100.0 (118/118)	0.8 (1/118)	3.4 (4/118)	0.8 (1/118)
		Medicated	100.0 (118/118)	2.5 (3/118)	2.5 (3/118)	0.0 (0/118)
		Total	100.0 (236/236)	1.7 (4/236)	3.0 (7/236)	0.4 (1/236)
	C. coli	Nonmedicated	98.0 (49/50)	94.0 (47/50)	94.0 (47/50)	0.0 (0/50)
		Medicated	100.0 (51/51)	96.1 (49/51)	96.1 (49/51)	0.0 (0/51)
		Total	99.0 (100/101)	95.0 (96/101)	95.0 (96/101)	0.0 (0/101)

Among the isolates from bile samples, all tested C. jejuni isolates were resistant to tetracycline (100.0%; 96/96), but susceptible to other antibiotics, while the two C. coli isolates were resistant to tetracycline, ciprofloxacin, and nalidixic acid. None of the bile C. jejuni or C. coli isolates showed resistance to gentamicin, azithromycin (AZI), erythromycin (ERY), florfenicol (FFN), telithromycin (TEL), or clindamycin (CLI).

Notably, for the fecal samples, the ciprofloxacin (CIP) resistance rate (95.0%) in C. coli is significantly (P < 0.05) higher than that in C. jejuni (1.7%). There is no significant difference (P > 0.05) in the overall antimicrobial susceptibility profiles, as tested in this study, between fecal and bile C. jejuni. The two C. coli isolates from bile samples also showed similar resistance patterns to the C. coli isolates from feces. In addition, there was also no significant difference (P > 0.05) in the antimicrobial resistance rates of C. jejuni and C. coli between the nonmedicated and feed-medicated groups.

Genetic diversity of the Campylobacter isolates from sheep.

In total, 24 C. coli and 39 C. jejuni isolates were selected for pulsed-field gel electrophoresis (PFGE) analysis. The 24 C. coli isolates consisted of twelve CIP^r C. coli isolates from feces (one representative isolate for each group and each sampling time in the two trials), the two CIP^r C. coli isolates from bile, the five ciprofloxacin-susceptible (CIP^s) fecal C. coli isolates, the TET-susceptible (TET^s) C. coli isolate, and the other four fecal C. coli isolates (isolated from the same sheep with the two CIP^r C. coli isolates from bile). The 24 C. coli isolates were grouped into three separate clusters, with the vast majority of isolates (75.0%; 18/24) grouped into type III (Fig. 2). The five CIP^s C. coli isolates were all type I, and the type II group included one CIP^r C. coli isolate from feces. The predominant cluster (type III) consisted of 18 CIP^r C. coli isolates, among which 16 were from fecal samples (including one TET-susceptible isolate and 4 from the same sheep from which the two CIP^r bile isolates were obtained) and two were from bile samples (Fig. 2). Multilocus sequence typing (MLST) analysis showed that the C. coli isolates grouped in types I, II, and III belong to ST827, ST1068, and ST902, respectively (Fig. 2), consistent with the classifications using PFGE. Thus, ST902 was the most prevalent ST in the sheep C. coli isolates analyzed in this study. Interestingly, the CIP^r bile and fecal isolates from the same sheep shared the same PFGE type (III) and MLST type (ST902), indicating that the bile and fecal isolates are genetically related.

FIG 2.

PFGE patterns of representative C. coli isolates in sheep. T1 and T2, trials 1 and 2, respectively; W1, W2, and W3, 1st, 2nd, and 3rd weeks, respectively; CIP, ciprofloxacin; NC, at necropsy; ND, not done. #, TET-susceptible C. coli. Boldface animal identifiers (IDs) indicate animals from which both bile and fecal isolates were typed.

The 39 C. jejuni isolates included 12 CIP^s C. jejuni isolates from bile (three representative isolates for each group in two trials), the four CIP^r C. jejuni isolates from feces, the three CIP^s fecal C. jejuni isolates (isolated from the same sheep with the CIP^r isolates), 20 CIP^s fecal C. jejuni isolates (from different sheep from which CIP^r isolates were isolated, representative isolates from each group and each sampling time in two trials, and including ten selected fecal C. jejuni isolates from the same sheep with the six bile C. jejuni isolates). The 39 C. jejuni isolates were clustered into five different clusters (Fig. 3), apart from four isolates that were nontypeable by PFGE because they were not digestible by KpnI. The four nontypeable isolates included one CIP^s bile isolate and three selected CIP^s fecal isolates from the same three sheep with bile isolates. Type A was the predominant genotype and accounted for 69.2% (27/39) of the analyzed isolates. This predominant cluster included both CIP^r and CIP^s C. jejuni isolates from fecal and bile samples. MLST results indicated that all but two type A C. jejuni isolates belonged to ST2862 (Fig. 2). One of the two non-ST2862 isolates was a CIP^r C. jejuni isolate (identifier [ID] 0993), which showed only one base difference in gltA from that of ST2862. After the sequence was submitted to the MLST database, the new gltA allele was assigned to gltA586 (in contrast to gltA2 in ST2862), and the isolate was assigned a new ST, ST9380. The other non-ST2862 isolate was also a new ST (ST9379) and was a bile isolate (ID 292). Detailed information for MLST profiles of the two isolates is presented in Table S2. The other four PFGE patterns, types B, C, D, and E, included 8 C. jejuni isolates. Type B belonged to ST982, while types C, D, and E all belonged to ST52 (Fig. 3). For the selected bile and fecal C. jejuni isolates from the same sheep, the situation was more diverse than that with the C. coli isolates (Fig. 3). Bile and fecal isolates from each of the animals with ID numbers 898, 346, and 604 shared the same PFGE and MLST profiles, while the bile and fecal isolates from three other animals were of different PFGE or MLST types (ID 292, the same PFGE with different STs; ID 961, different PFGE types but with the same ST; ID 610, different PFGE types with different STs).

FIG 3.

PFGE patterns of representative C. jejuni isolates in sheep. T1 and T2, trials 1 and 2, respectively; W1, W2, and W3, 1st, 2nd, and 3rd weeks, respectively; CIP, ciprofloxacin; NC, at necropsy; ND, not done. An asterisk (*) indicates newly assigned STs. Boldface animal IDs indicate animals from which both bile and fecal isolates were typed. Four of the C. jejuni isolates were nontypeable by PFGE and are not shown.

Together, the PFGE and MLST findings indicated predominant genotypes in both C. coli and C. jejuni, suggesting that clonal expansion may have been involved in the dissemination of both C. coli and C. jejuni in sheep.

Point mutations in gyrA.

Based on different PFGE and MLST types, eight CIP^r C. coli and four CIP^r C. jejuni isolates were selected for determination of the point mutations in gyrA. All of the CIP^r C. coli isolates harbored a single Thr-86-Ile mutation in GyrA without any other amino acid changes in this region (Fig. 2). For the CIP^r C. jejuni isolates, three had the Thr-86-Ile point mutation alone, and one carried the Thr-86-Ile mutation plus the Ser-22-Gly, Asn-203-Ser, and Arg-285-Lys changes (Fig. 3). These results are consistent with the known role of the Thr-86-Ile mutation in mediating CIP resistance in Campylobacter.

DISCUSSION

Results from this study revealed high prevalences of TET-resistant Campylobacter spp. and CIP^r C. coli strains in sheep derived from two commercial farms. These animals were naturally infected and already carried antibiotic-resistant Campylobacter on the day of arrival (Tables 1, 2, and 3). Subsequent treatment with in-feed tetracycline in a research setting did not affect the overall prevalence, antimicrobial susceptibility profiles, and species distribution of Campylobacter, as there were no statistically significant differences (P > 0.05) between the feed-medicated and nonmedicated groups. However, the quantitative CFU counts in bile and gallbladder mucosa collected at the end of the trials revealed higher total CFU counts in the medicated group than in the nonmedicated group (P < 0.05), suggesting that tetracycline treatment may promote Campylobacter colonization in the gallbladder. The exact reason for this effect is unknown, but it is known that tetracyclines undergo significant enterohepatic circulation, resulting in significant concentration of tetracyclines in bile, which may inhibit competitive bacterial organisms and confer a colonization advantage for tetracycline-resistant Campylobacter. Since similar numbers of animals received therapeutic treatment in both groups, it is likely that the effect is attributable to in-feed use of tetracycline. Regardless of the reasons, the enhanced colonization of the gallbladder may give Campylobacter an advantage in the spread to other organs or dissemination back to the intestinal tract.

In fecal samples, C. jejuni accounted for 68.6% of the total isolates (Table 1), while in bile samples, the isolates were almost exclusively C. jejuni (98.0%), suggesting that C. jejuni has a higher tropism to bile or a higher ability to reach the gallbladder and adapt within it. Previous studies observed similar prevalence of C. jejuni and C. coli in sheep gallbladders (29, 30); however, these previous studies were conducted in a different country, and it is possible that differences in sheep production practices or genetic variability of Campylobacter strains may influence gallbladder colonization by the organism. Bile is harmful to bacteria; however, Campylobacter utilizes the multidrug efflux pump CmeABC for bile resistance (31–34). The exact reason for the predominance of C. jejuni in the gallbladder, as observed in this study, is unknown and remains to be investigated in future studies. Interestingly, C. jejuni clone SA was not detected in the sheep examined in this study. This clone is a predominant cause of Campylobacter-associated ovine abortions and is also distributed in feedlot and dairy cattle in the United States (23, 35). A recent study also reported identification of clone SA in Grenada (36). The lack of clone SA detection in this study is not entirely surprising, as we have previously found that the distribution of clone SA is highly variable from farm to farm (25). Thus, it was likely that the two farms from which the sheep isolates were derived did not harbor this particular C. jejuni strain.

Except for FQs and tetracycline, the Campylobacter isolates examined in this study were generally susceptible to other tested antimicrobials (Table 3). All but one of the tested Campylobacter isolates from fecal and bile samples were resistant to tetracycline. This finding further strengthens the result of our previous study on Campylobacter carriage in sheep slaughterhouses, which revealed that the rate of resistance to tetracycline was already 83.3% (25). In this study, the in-feed tetracycline medication did not further influence the prevalence of tetracycline-resistant Campylobacter due to the fact that tetracycline resistance in the sheep was already very high (virtually 100%) at the beginning of the treatment. Thus, the result should be interpreted cautiously. Given the high prevalence of tetracycline-resistant Campylobacter on sheep farms in the United States, the use of tetracycline as a means to control Campylobacter-induced sheep abortion would have a limited impact. In fact, tetracycline medication may have facilitated the persistence of tetracycline-resistant Campylobacter strains on sheep farms. Results from this study further underline the need for improved tetracycline stewardship in sheep to reduce the prevalence of tetracycline-resistant Campylobacter.

A significant finding of this work is the high detected prevalence of CIP^r C. coli (95%), whereas CIP resistance in the sheep C. jejuni isolates remained low (1.7%) (Table 3). The difference between C. coli and C. jejuni CIP resistance rates is unlikely due to variation in their susceptibility to fluoroquinolones, as CIP^r C. jejuni is also common in other animal species (6, 7). The high-level prevalence of CIP^r C. coli in sheep was unexpected, as this class of antibiotics is not used in sheep production in the United States. A recent study on cattle Campylobacter reported CIP resistance rates of 35.4% in C. jejuni and 77.3% in C. coli (15). This could be explained by the fact that FQ antibiotics are labeled for treatment of respiratory disease in cattle in the United States. The reason for the detected high incidence of CIP^r C. coli in sheep in this study is unknown, but there is a possibility of transmission of CIP^r Campylobacter from cattle to sheep. Although the sampled sheep farms were not adjacent to other animal species, sheep could get Campylobacter through birds, flies, or other transmission vehicles. However, if this were the sole case, we would have also observed an increased prevalence of CIP^r C. jejuni in sheep, as it is also common in cattle. It is possible that the CIP^r C. coli strains were introduced to sheep by chance and something on these farms favored the selection of CIP^r C. coli, contributing to its overall high prevalence in sheep. It should be pointed out that fluoroquinolones are also approved for use in swine and were used in poultry prior to 2005 in the United States. Thus, the possibility that sheep acquired CIP^r C. coli from other animal species rather than from cattle cannot be totally excluded.

To determine the genetic relatedness of the Campylobacter isolates, we conducted PFGE and MLST analyses, which are commonly used for genotyping of Campylobacter (20, 37–39). PFGE typing of 24 representative C. coli isolates revealed a limited number of types and demonstrated that the majority (18/24) of the CIP^r C. coli isolates were grouped into a single type (type III; Fig. 2), which was confirmed by MLST to be a single ST (ST902). It is worthwhile to note that type III included isolates from different and geographically distant farms and from different sampling times, suggesting that this genotype is stable and is prevalent on different sheep farms. Also, bile and fecal isolates from the same sheep shared the same PFGE type and ST. The other two PFGE types belonged to ST827 (type I) and ST1068 (type II). C. coli ST827 has been previously reported as sometimes the most common ST in both sheep and cattle (40, 41), while ST1068 is more often found in cattle (15, 41, 42). Moreover, ST1068 was also identified among swine C. coli isolates, sometimes as the most predominant ST (42–44). In addition, the three identified STs (ST902, ST827, and ST1068) all belong to the ST828 complex (Fig. 2). Although the specific prevalent STs vary in different species, C. coli isolates of the ST828 complex, the most prevalent CC (clonal complex) for C. coli, have been frequently reported in published studies from different countries and different sources, including poultry, cattle, sheep, swine, and even humans (15, 45–48). Our results confirmed the prevalence of this CC and the existence of specific endemic STs among animals in different regions.

The 39 representative C. jejuni isolates examined by PFGE typing in this study were grouped into five clusters (Fig. 3), with the exception of four isolates that were not typeable by PFGE because KpnI failed to digest appropriately. Similar to the C. coli result, the majority (27/39) of the tested C. jejuni isolates grouped into one cluster (type A) and were confirmed by MLST to be a single ST (ST2862) (except for two isolates), suggesting it was the predominant C. jejuni clone in sheep. This clone included both CIP^s and CIP^r C. jejuni isolates from both fecal and bile samples that were collected from different farms (trials 1 and 2) and at different sampling times (Fig. 3), indicating this genotype was commonly disseminated in sheep. The other four PFGE types belonged to ST982 (type B) and ST52 (types C, D, and E). ST982 accounted for 20.8% (10/48) of the C. jejuni isolates in a previous study involving a lamb slaughterhouse, and it was also reported in cattle (15, 25, 49). ST52 was reported sporadically in C. jejuni isolates from different countries and sources, including from ruminants (36–38, 48, 50, 51). Both ST2862 and ST982 belong to CC21, which is the most prevalent CC in C. jejuni.

To examine the possible transmission of Campylobacter between sheep and cattle, we selected some cattle Campylobacter isolates identified in our previous study (15) to compare with the sheep Campylobacter isolates identified in this study. The PFGE and MLST results revealed that ST902 (the predominant C. coli genotype identified in this study in sheep) with the same PFGE pattern was also found in the cattle C. coli isolates (Fig. S1a), although the predominant ST in cattle was ST1068 (15). Here, we also identified ST1068 in sheep C. coli (Fig. 2), which had a PFGE pattern similar to that of cattle C. coli (Fig. S1b). These findings further suggest the possible transmission of fluoroquinolone-resistant (FQ^r) C. coli (e.g., ST902 and ST1068) between cattle and sheep. In contrast to the C. coli isolates, the predominant C. jejuni genotype in sheep, ST2862 (Fig. 3), was not previously identified in cattle in the United States (15, 26, 49). Although C. jejuni ST982 has been found in cattle (15), PFGE patterns of cattle isolates were quite different from those of the sheep isolates (Fig. S1c). Thus, the genetic relatedness between sheep and cattle C. jejuni isolates was not as clear as that for C. coli.

In Campylobacter, FQ resistance is mediated by point mutations in the quinolone resistance-determining region (QRDR) of DNA gyrase (GyrA) and by the function of the CmeABC multidrug efflux pump (8, 52). The most frequent mutation associated with FQ resistance in Campylobacter is Thr-86-Ile, followed by Asp-90-Asn, Thr-86-Lys, Thr-86-Ala, Thr-86-Val, Asp-90-Tyr, and Ala-70-Thr (8, 53). Additionally, double mutations, including Thr-86-Ile plus Pro-104-Ser and Thr-86-Ile plus Asp-90-Asn, have also been associated with FQ resistance in Campylobacter (54). The mutation of Asn-203-Ser is also known to confer FQ resistance along with the Thr-86-Ile mutation (55). Ser-22-Gly has not been linked to FQ resistance in Campylobacter, and there was no evidence that Asn-203-Ser and Arg-285-Lys alone were associated with the FQ resistance phenotype (15). In this study, all examined FQ^r Campylobacter isolates harbored the Thr-86-Ile mutation in GyrA, while one isolate also carried the Ser-22-Gly, Asn-203-Ser, and Arg-285-Lys mutations, consistent with the known role of the Thr-86-Ile change (alone or along with other mutations) in CIP resistance.

In summary, we observed high prevalences of tetracycline-resistant Campylobacter and FQ^r C. coli in commercial sheep operations, regardless of in-feed medication with tetracycline. Although the medication increased the total Campylobacter CFU counts in the sheep gallbladder, it did not affect overall prevalence, genotypes, or antimicrobial susceptibility profiles of the Campylobacter isolates under the conditions employed in this study. The sheep harbored predominant genotypes of C. jejuni (ST2862) and C. coli (ST902). Of particular note, the FQ^r C. coli clones in the sheep (both ST902 and ST1068) were genetically linked to the isolates previously identified in cattle, suggesting the possibility of dynamic transmission of FQ^r C. coli between sheep and cattle. However, there are several limitations in this study. First, the work was a controlled laboratory treatment study, and the preexisting high prevalence of tetracycline-resistant Campylobacter strains limited our ability to analyze the effect of the in-feed medication on development of tetracycline resistance. Second, the sheep were derived only from two commercial farms, and the sample sizes were not particularly large. Thus, the findings may not represent the overall situation with sheep Campylobacter in the United States. Despite these limitations, the findings provide new and useful information on sheep Campylobacter. Since ruminants are important reservoirs for Campylobacter and are a significant source of foodborne campylobacteriosis in humans, enhanced efforts should be directed toward the development of interventions to reduce prevalence and transmission of FQ^r Campylobacter in sheep and cattle.

MATERIALS AND METHODS

Sample collection and Campylobacter isolation.

Two separate trials were conducted, and each involved 80 feeder lambs, with 40 each in the feed-medicated group and feed-nonmedicated group. These sheep were purchased from two commercial farms and then housed in the laboratory animal facility at Iowa State University during the treatment study. Trial 1 was conducted using animals from farm 1, while trial 2 was performed using animals from farm 2. The two farms are in the Midwest region and are geographically separated by approximately 86 miles. The ages of the sheep at enrollment varied, but all were spring-born feeder lambs and were approximately 2 months old. Both source farms raise sheep in a semi-intensive management program with outdoor access. Treatment of sheep with FQ antibiotics is expressly prohibited in the United States, and there was no history of use of these antibiotics on these farms.

In each trial, the animals were randomly assigned into the feed-medicated and nonmedicated groups. Chlortetracycline (Aureomycin; Zoetis) was used for the feed medication, and the medication was conducted to determine the effect of the antibiotic on prevention and control of bacterial pneumonia in feedlot lambs. The antibiotic was fed extralabel (350 mg/head/day; 160 mg/kg body weight or 160 ppm in feed) in accordance with FDA/Center for Veterinary Medicine (CVM) Guidance 615.115. Regardless of the feed medication, clinically sick animals (19 in each group for trial 1; 17 and 18 in each group, respectively, for trial 2) received a chlortetracycline injection (9 mg/lb), and the injection was repeated once in 72 h if needed. The treated sheep remained in their respective groups.

Trial 1 was performed from 21 June 2017 to 6 July 2017, while trial 2 was carried out from 13 July 2017 to 26 July 2017 (Table 1). Fecal samples were collected at the beginning of the study and once a week thereafter. Bile and gallbladder samples were collected at necropsy. The samples were then cultured for Campylobacter using Mueller-Hinton (MH) agar plates (Difco) with selective antimicrobials and growth supplements (SR084E and SR117E; Oxoid). The plates were incubated at 42°C for 48 h under microaerobic conditions (85% N₂, 5% O₂, and 10% CO₂). Up to three Campylobacter-like colonies from each sample were subcultured onto MH agar plates, and the pure cultures were stored in glycerol stocks at −80°C.

Enumeration of Campylobacter in bile and gallbladder mucosa.

In addition to isolation of Campylobacter from individual bile samples as described above, we also determined the total Campylobacter CFU in bile and gallbladder mucosa. For this purpose, in the two trials, bile and gallbladder samples were randomly collected from sheep at necropsy. In total, 37 and 35 bile/gallbladder mucosal samples were selected in the two trials from the feed-nonmedicated and feed-medicated groups, respectively. For each group, the sheep were housed in 4 separate pens, and the collected bile from each gallbladder (5 to 15 ml) was pooled based on treatment group and pen number. In total, 4 pools were obtained per group in each trial. For collection of gallbladder mucosal samples, 5 ml of phosphate-buffered saline (PBS) was used to wash any residual bile off each gallbladder mucosa. Then, a sterile razor blade was used to scrape mucosa from each gallbladder onto a bile-free weigh boat. Next, the blade was washed with PBS to rinse any mucosa on the blade onto the weigh boat, which then was rinsed with roughly 2 ml of PBS, and the collected gallbladder mucosa was decanted into a 15-ml conical tube. The final volume was adjusted to 3 ml, and the 15-ml conical tube was vortexed vigorously prior to culture for Campylobacter.

For both the bile and gallbladder mucosal samples, a 10-fold dilution series was performed using 1.5-ml snap cap tubes filled with 900 μl of MH broth. Each dilution was spread onto MH agar plates containing selective supplement (SR084E and SR117E; Oxoid), and standard culture conditions for Campylobacter were utilized as for the fecal samples (described above). After culturing, the Campylobacter CFU were counted for each sample and the log₁₀ averages were computed for both bile and gallbladder mucosa. The detection limit of the plating method was 100 CFU/ml of bile or gallbladder mucosal suspension. All of the 16 pooled bile samples demonstrated visually pure growth of Campylobacter spp. and were utilized in the final calculations. For the gallbladder mucosal samples, 4 in the nonmedicated group and 9 in the medicated group were not usable because of high background bacterial contamination after culturing. The remaining plates were visually determined to have pure growth of Campylobacter spp. and thus were used in the final calculations. In total, Campylobacter CFU counts were obtained from 33 and 26 gallbladder mucosal samples in the nonmedicated and medicated groups, respectively. Representative colonies from each group were selected and subjected to species confirmation and identification (see “Bacterial species identification,” below).

Bacterial species identification.

Matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry (Bruker) was used for bacterial species identification. All isolates were identified to the species level. Sample preparation and analysis were done as described previously (56). Mass spectra were acquired and analyzed using a microflex LT mass spectrometer (Bruker Daltonics) in combination with research use only version of the MALDI Biotyper Compass software 4.1 and the reference database MBT 7311 MSP Library (no. 1829023) at Iowa State University. Data were interpreted in accordance with the manufacturer’s (Bruker Daltonics) standard criteria, as follows: (i) high-confidence identification when the score was between 2.00 and 3.00, (ii) low-confidence identification when the score was between 1.70 and 1.99, and (iii) no organism identification possible when the score was 1.69 and lower.

C. jejuni clone SA detection.

PCR was used for identifying C. jejuni clone SA. DNA template was extracted from Campylobacter colonies using single-cell lysis buffer (57). Primers specific for C. jejuni clone SA were designed as described previously (58). C. jejuni IA3902, a clinical isolate of clone SA, was used as the positive control, while a reaction without template DNA served as a negative control.

Antimicrobial susceptibility testing.

A total of 236 C. jejuni and 101 C. coli nonduplicate isolates from fecal samples were tested for antimicrobial susceptibility. Those isolates were representatives of individual animals at different sampling times. Also, 98 C. jejuni and 2 C. coli nonduplicate isolates (a single isolate from each animal) from bile samples were included in the susceptibility testing. The MICs of nine antibiotics were determined using a standard broth microdilution method as recommended by CLSI and the National Antimicrobial Resistance Monitoring System for Enteric Bacteria (NARMS) (59–61). The tested ranges and breakpoints of the nine antibiotics are listed in Table S1. Commercially available Sensititre Campylobacter plates (Thermo Fisher Scientific, Waltham, MA) were used for the tests. The nine antibiotics were azithromycin (AZI), ciprofloxacin (CIP), erythromycin (ERY), gentamicin (GEN), tetracycline (TET), florfenicol (FFN), nalidixic acid (NAL), telithromycin (TEL), and clindamycin (CLI). After incubation in a microaerobic environment for 24 h at 42°C, the results were interpreted. For each isolate, the MIC value was set as the lowest antimicrobial concentration at which no bacterial growth was observed. The antimicrobial resistance breakpoints were chosen according to the standards established by CLSI for bacteria isolated from animals (59–61). C. jejuni ATCC 33560 and C. coli ATCC 33559 were used as quality control strains.

PFGE and MLST typing of Campylobacter isolates.

Representative Campylobacter isolates were analyzed using pulsed-field gel electrophoresis (PFGE) and multilocus sequence typing (MLST). PFGE analysis of the macrorestriction fragment patterns of genomic DNA using KpnI enzyme was performed following the CDC’s standardized PulseNet protocol for Campylobacter, with minor modifications (23). Briefly, fresh cultures of Campylobacter were embedded in 1% Seakem Gold agarose (Fisher Scientific, Fair Lawn, NJ) and lysed with proteinase K for 30 min at 50°C in a water bath shaker. The gel plugs were digested with KpnI for 4 h at 37°C. Digested plugs were embedded into 1% agarose and separated by electrophoresis in 0.5 × Tris-borate-EDTA (TBE) buffer (Promega) at 14°C for 18 h using a Chef Mapper electrophoresis system (Bio-Rad, Hercules, CA). The gels were stained with ethidium bromide for 30 min and then photographed using a digital imager (Alpha Innotech, Santa Clara, CA). The PFGE patterns were analyzed by GelCompare II v.6.5 software (Applied Maths, Kortrijk, Belgium) using the Dice similarity coefficient and unweighted-pair group method with arithmetic averages (UPGMA) with 0.5% optimization and 1.5% position tolerance. A Lambda DNA ladder (Bio-Rad) was used as the molecular size marker. PFGE patterns with ≥90% similarity to each other were defined as a cluster.

To further confirm the genetic diversity of the isolates, MLST was performed for selected representative isolates of different PFGE patterns, as described previously (62). The seven housekeeping genes were amplified and sequenced using the primers recommended by the Campylobacter MLST website (https://pubmlst.org/campylobacter/), which was developed by Keith Jolley and Man-Suen Chan at the University of Oxford (63). All PCR products were purified using the QIAquick PCR purification kit (Qiagen, Hilden, Germany) and then sequenced at the DNA Core Facility of Iowa State University using an Applied Biosystems 3730xl DNA analyzer. The sequences were submitted to the MLST database to determine allele numbers and STs. The new alleles and STs were submitted to the database and were assigned new numbers.

gyrA mutation determination.

In Campylobacter, FQ resistance is conferred by point mutations in the gyrA gene in conjunction with the CmeABC efflux pump (64). To confirm the mechanism of FQ resistance in this study, four FQ^r C. jejuni isolates and eight FQ^r C. coli isolates of different PFGE and MLST types were selected for determination of the point mutations in the quinolone resistance-determining region (QRDR) of gyrA. The primers GyrAF1 (5′-CAACTGGTTCTAGCCTTTTG-3′) and GyrAR1 (5′-AATTTCACTCATAGCCTCACG-3′) were used for C. jejuni isolates, while GyrAF2 (5′-TTATTTAGATTATTCTATGAGCGT-3′) and GyrAR2 (5′-CTTGAGTTCGATTACAACAC-3′) were used for C. coli isolates, as described previously (15). All PCR products were purified using the QIAquick PCR purification kit (Qiagen, Hilden, Germany) and then sequenced at the DNA Core Facility of Iowa State University using an Applied Biosystems 3730xl DNA analyzer.

Statistical analysis.

Chi-square and Fisher’s exact tests were used to compare the prevalences and the antimicrobial resistance rates of the Campylobacter isolates from different samples (fecal and bile samples) and different groups (feed-nonmedicated and feed-medicated groups). An unpaired t test was used to compare the average Campylobacter CFU/ml results from the bile and gallbladder mucosal samples between the nonmedicated and medicated groups in two trials. The data were analyzed using GraphPad (Prism). P values of <0.05 were deemed to be statistically significant.