Variable within- and between-Herd Diversity of CTX-M Cephalosporinase-Bearing Escherichia coli Isolates from Dairy Cattle

Abstract

bla_CTX-M beta-lactamases confer resistance to critically important cephalosporin drugs. Recovered from both hospital- and community-acquired infections, bla_CTX-M was first reported in U.S. livestock in 2010. It has been hypothesized that veterinary use of cephalosporins in livestock populations may lead to the dissemination of beta-lactamase-encoding genes. Therefore, our objectives were to estimate the frequency and distribution of coliform bacteria harboring bla_CTX-M in the fecal flora of Ohio dairy cattle populations. In addition, we characterized the CTX-M alleles carried by the isolates, their plasmidic contexts, and the genetic diversity of the bacterial isolates themselves. We also evaluated the association between ceftiofur use and the likelihood of recovering cephalosporinase-producing bacteria. Thirty fresh fecal samples and owner-reported ceftiofur use data were collected from each of 25 Ohio dairy farms. Fecal samples (n = 747) yielded 70 bla_CTX-M-positive Escherichia coli isolates from 5/25 herds, 715 bla_CMY-2 E. coli isolates from 25/25 herds, and 274 Salmonella spp. from 20/25 herds. The within-herd prevalence among bla_CTX-M-positive herds ranged from 3.3 to 100% of samples. Multiple pulsed-field gel electrophoresis (PFGE) patterns, plasmid replicon types, and CTX-M genes were detected. Plasmids with CTX-M-1, -15, and -14 alleles were clonal by restriction fragment length polymorphism (RFLP) within herds, and specific plasmid incompatibility group markers were consistently associated with each bla_CTX-M allele. PFGE of total bacterial DNA showed similar within-herd clustering, with the exception of one herd, which revealed at least 6 different PFGE signatures. We were unable to detect an association between owner-reported ceftiofur use and the probability of recovering E. coli carrying bla_CTX-M or bla_CMY-2.

INTRODUCTION

A powerful class of plasmid-encoded extended-spectrum beta-lactamases (ESBLs) of emerging importance is the cefotaximases (CTX-M). As their name suggests, CTX-M enzymes exhibit heightened activity against cefotaxime compared to that against other cephalosporins and in particular are characterized by their ability to hydrolytically inactivate and thus convey resistance to extended-spectrum cephalosporin drugs, including cefepime. Since emerging in the late 1980s, bla_CTX-M alleles have rapidly become the most common genes to confer the ESBL phenotype among Enterobacteriaceae recovered from human health care settings (26). CTX-M-bearing isolates are also readily recovered from community-acquired infections in humans (35, 38) as well as from companion (34) and food (46, 49) animals in many parts of the world. The CTX-M genes are primarily associated with conjugative plasmids recovered from Escherichia coli, Salmonella spp., and Klebsiella pneumoniae (48). Although there is little information to account for the rapid dissemination of these genes, conjugal transfer of bla_CTX-M-bearing plasmids has been reported to be highly efficient (40). The global dissemination of this resistance genotype has been compared to that of the broad-spectrum TEM beta-lactamase seen in the 1960s in addition to the proliferation of bla_CMY-2 in the 1990s and 2000s (40).

Although only recently reported in U.S. livestock (52), bla_CTX-M has been reported in food animal populations (17, 46, 49), food products (11), and both domestic (34) and wild (9, 36) animal species throughout the world. While the zoonotic food-borne transmission of bacteria harboring bla_CTX-M has not yet been reported in the United States, a sampling of 20 packages of fresh retail meat from 7 Pittsburgh, PA, area supermarkets yielded an E. coli isolate carrying bla_CTX-M-1 with a pulsed-field gel electrophoresis (PFGE) pattern identical to that of a hospital-associated isolate identified during the same study period (11). In addition, the international spread of the highly virulent extraintestinal pathogenic E. coli (ExPEC) B2-O25:H4-ST131 is associated with the pandemic dissemination of bla_CTX-M-14 and bla_CTX-M-15 (20, 21, 44). E. coli O25:H4 ST131 has been linked to both community-acquired and nosocomial infections and reported in North America, Europe, Asia, Africa, and the Middle East, from multiple sources, including humans, their companion animals, and fresh meat products (22).

Extended-spectrum cephalosporin drugs are considered to be “critically important” to human medicine by the World Health Organization (53). It has been hypothesized that veterinary use of ceftiofur and cefquinome in livestock populations may provide selection pressure contributing to the dissemination of bla_CTX-M (23, 46). The presence of extended-cephalosporin-resistant organisms in food animal populations could serve as a reservoir of resistance genes for their food-borne transmission in fresh meat products (11, 31). However, the extent of the association of veterinary cephalosporin use in livestock with the emergence or dissemination of bla_CTX-M has not been established.

We previously reported the presence of bla_CTX-M in fecal E. coli isolates recovered from cattle in Ohio (52). However, basic epidemiologic information regarding the frequency, distribution, and predisposing factors for this important resistance gene in U.S. livestock populations is unknown. Therefore, our objectives were to estimate the frequency and distribution of Salmonella spp. and coliform species harboring bla_CTX-M among dairy cattle. In addition, we characterized the CTX-M alleles carried by the isolates, their plasmidic contexts, and the genetic diversity of the bacterial isolates themselves via PFGE. We additionally sought to evaluate the association between herd-level ceftiofur use and the likelihood of recovering bla_CTX-M-bearing bacteria.

MATERIALS AND METHODS

Source of the isolates.

A convenience sample of 25 Ohio dairy farms was recruited for voluntary participation in this cross-sectional study. At each farm, 30 fresh fecal samples were collected and placed in individual sterile 50-ml conical tubes. Samples were freshly voided feces that appeared to be from a single animal and were collected from free-stall barn and alleyway floors throughout each facility to minimize the possibility of repeated samples from the same animal. Samples were transported at ambient temperature to our research laboratory, where they were stored overnight at 4°C. The day following their collection, fecal samples were divided into duplicate 4-g aliquots for the culture and isolation of extended-spectrum-cephalosporin-resistant E. coli and Salmonella spp. Cow inventory and ceftiofur use data for each farm, including the number of cows treated therapeutically with ceftiofur (injectable or intramammary) during the previous month and the previous 6 months, were obtained from the owner or manager of each herd.

Bacterial culture and antimicrobial susceptibility testing.

For the recovery of E. coli resistant to extended-spectrum cephalosporins, 4-g fecal aliquots were homogenized into 36 ml nutrient broth containing 2 μg/ml cefotaxime. After overnight incubation, this broth was streaked onto MacConkey agar containing 4 μg/ml cefepime to identify isolates with a bla_CTX-M phenotype and onto MacConkey agar containing 4 μg/ml cefoxitin to identify isolates with a bla_CMY-2 phenotype. Characteristic lactose-positive and indole-positive isolates were confirmed as E. coli by PCR (1). We have previously used these methods to successfully recover fecal E. coli isolates harboring bla_CTX-M and bla_CMY-2 (52).

The remaining 4-g fecal aliquot was used for isolation of Salmonella spp. We used a two-phase enrichment in supplemented tetrathionate (TTB) and Rappaport-Vassiliadis R10 (RV) broths followed by differential selection on xylose-lysine-Tergitol 4 agar (XLT-4) (16, 31). From each sample, bacteria from a single black colony on XLT-4 were isolated on MacConkey agar and confirmed as Salmonella by using standard biochemical reactions, including triple sugar iron (TSI) agar, urea broth, and polyvalent antisera testing. Confirmed Salmonella sp. isolates were screened for extended-spectrum-cephalosporin resistance by inoculation onto MacConkey agar plates containing 4 μg/ml cefepime or 4 μg/ml cefoxitin with overnight incubation at 37°C.

Isolates with the expected phenotypes on selective media were further characterized to fully describe their resistance phenotypes. MICs of a panel of 26 antimicrobial drugs important to human and veterinary medicine were generated using a semiautomated broth microdilution system (CMV1AGNF and ESB1F MIC plates, TREK Diagnostic Systems, Cleveland, OH) following Clinical and Laboratory Standards Institute guidelines (8).

Isolate characterization.

PFGE genotyping (CHEF-DRIII; Bio-Rad Laboratories, Hercules, CA) was performed on total genomic DNA by using SpeI (New England BioLabs, Ipswich, MA) following CDC-recommended procedures (7, 37). The genetic similarities of strains were compared by examining banding patterns after electrophoresis and applying generally accepted criteria to assign levels of similarity (45). In addition, the isolates were clustered into genotypic groups by using the Dice coefficient similarity index with clustering settings of 1.00% optimization and 1.00% band position tolerance via Bionumerics software (Applied Maths, Kortrijik, Belgium). To determine if isolates with the expected resistance genotype (i.e., bla_CTX-M-14 or bla_CTX-M-15) belonged to the B2-O25:H4-ST131 strain, sequencing (51) was performed on a subset of isolates by using adk (adenylate kinase), fumC (fumarate hydratase), gyrB (DNA gyrase), and mdh (malate dehydrogenase) housekeeping genes.

Plasmid characterization.

The plasmid content of each isolate was visualized by electrophoresis using a standard procedure (24). Conjugation experiments (14) to establish the transmissibility of plasmids harboring bla_CTX-M utilized wild-type E. coli donors with a rifampin- and nalidixic acid-resistant derivative of E. coli K-12 MG1655 as the recipient strain. Recipient acquisition of the expected plasmids and resistance genes was established with additional plasmid profiling, bla_CTX-M PCR, and sequencing of the resulting amplicons recovered from transconjugants. Individual plasmids were classified according to a PCR-based replicon typing procedure (PRT) that detects 18 replicon types based on incompatibility group loci (4, 5) by using boiled lysate as template DNA. Plasmids were characterized using restriction fragment analysis by digestion of 10 μl of extracted plasmid DNA (39) overnight with 1 μl of AccI (New England BioLabs) at 37°C.

Discrimination of bla_CTX-M alleles.

PCR, utilizing previously reported primer sets, was used to detect bla_CTX-M and to screen for other classes of beta-lactamase resistance genes, including CMY-2, TEM, SHV, and OXA (25, 26, 29, 43). The bla_CTX-M genes were sequenced using sequencing primers for CTX-M groups A1 and 9 (Table 1), which were designed in our laboratory (PrimerQuest; Integrated DNA Technologies, Coralville, IA). Amplicons were sequenced bidirectionally using the CEQ 8000 capillary electrophoresis system (Beckman Coulter, Palo Alto, CA) and analyzed using a BLAST search (http://blast.ncbi.nlm.nih.gov/). Additional plasmid-mediated beta-lactamase resistance gene sequencing was accomplished using the corresponding PCR amplification primers (GENEWIZ, South Plainfield, NJ) and analyzed using BLAST (http://blast.ncbi.nlm.nih.gov/).

Table 1.

Primers used for sequencing beta-lactamase genes from Escherichia coli isolates and their transconjugants

Target	Primer name	Oligonucleotide sequence (5′–3′)		Product size (bp)	Reference
		Forward	Reverse
blaCTX–M group A1	CTX–M–1 upstream	ATGTTGTTGTTAATTCGTCTC	CGTTATCGCTGTACTGTAG	446	52
	CTX–M–1 downstream	TTAACTATAATCCGATTGCG	TTTCTGCCTTAGGTTGAG	507	52
blaCTX–M group 9	CTX–M–4 upstream	ATTCCGCTGCTGCTGGGCAG	GCGGCGTGGTGGTGTCTCTC	511	This study
	CTX–M–4 downstream	CGGGAGGCGTGACGGCTTTT	TTGCCACGGAACGGTCTGCG	520	This study
blaCMY–2	AmpC	ATGATGAAAAAATCGTTATGC	TTGCAGCTTTTCAAGAATGCGC	1,143	25
blaTEM	TEM	ATAAAATTCTTGAAGACGAAA	GACAGTTACCAATGCTTAATCA	1,074	29

Localization of bla_CTX-M genes to plasmid DNA.

Southern hybridization was used for the detection of bla_CTX-M on plasmid fragments following AccI restriction fragment length polymorphism (RFLP). DNA was transferred from agarose gels to positively charged nylon membranes following standard procedures (41). Digoxigenin (DIG)-labeled probes were synthesized using previously reported CTX-M PCR primer sets with a panspecific primer used to detect group 1 bla_CTX-M and a group 9-specific primer to detect bla_CTX-M-14 by using the PCR DIG kit (Roche, Basel, Switzerland) (26). Following probe hybridization and stringent washing, anti-DIG antibody conjugated to alkaline phosphatase was colorimetrically visualized with nitroblue tetrazolium (NBT)/BCIP (5-bromo-4-chloro-3-indolylphosphate) by using the DIG High Prime kit (Roche), following the manufacturer's instructions.

Data analysis.

The relationship between owner-reported ceftiofur use and the probability of recovery of fecal bacteria carrying bla_CTX-M or bla_CMY-2 was investigated using multivariable logistic regression procedures (Proc GENMOD in SAS version 9.2 [2008]; SAS Institute Inc.) with the events/trial syntax. Generalized estimating equations were utilized to account for expected clustering within herds. Ceftiofur use in each herd was expressed as the proportion of cows that had been treated with ceftiofur during the 1-month and 6-month periods just prior to our sampling. These variables were included in independent models that also included herd size as a fixed effect so that the recent and longer-term ceftiofur uses could both be evaluated.

RESULTS

The 25 study herds were made up predominantly of Holsteins and ranged in size from 40 to 475 lactating cows, with a mean herd size of 150 lactating cows. Except for the single herd producing milk to organic standards, owners/managers of all other herds (96%) reported that it was their policy to include ceftiofur as one of their treatment options for sick cows. These herds had mean ceftiofur treatment rates of 3.6% (median, 2.3%; range, 0 to 10.9% within individual herds) over the previous month and 16.4% (median, 10.7%; range, 0 to 72.5% within individual herds) over the previous 6 months immediately prior to our sampling.

We collected the expected 30 fecal samples from each of 23 of the 25 study herds, but 29 samples were obtained from herd 21 and 28 samples from herd 25. E. coli isolates with bla_CTX-M were recovered from 70 samples among the 747 collected (9.4%), representing 5 bla_CTX-M-positive herds (20%). Susceptibility profiles revealed the characteristic bla_CTX-M phenotype with resistance or reduced susceptibility to all penicillins and cephalosporins except cefoxitin, as well as variable resistance to other drugs, including tetracycline, streptomycin, and chloramphenicol (Fig. 1). Within-herd recovery of E. coli with bla_CTX-M among the 5 positive herds ranged from 1 (3%) to 30 (100%) positive fecal samples (Table 2). In addition, 711 samples (94.8%), representing all 25 herds, yielded E. coli carrying bla_CMY-2, with a minimum within-herd recovery of 18 (60%) positive fecal samples. Salmonella spp. were recovered from 284 (37.9%) fecal samples, representing 19 (76%) of the 25 study herds. Within-herd Salmonella recovery for the 19 positive herds ranged from 1 (3%) to 30 (100%) positive samples. Salmonella isolates were susceptible to both third- and fourth-generation cephalosporins.

Fig 1.

MICs of 26 antimicrobial agents for 70 Escherichia coli isolates containing bla_CTX-M recovered from Ohio dairy cattle fecal samples (numbers of isolates are shown in the body of the table). Broken lines represent susceptible breakpoints and solid lines represent resistant breakpoints where available. Corresponding to the concentration listed at the top of each column (μg/ml), the included range of each antimicrobial is shown in gray.

Table 2.

Summary of E. coli isolates with bla_CTX-M recovered from the fecal flora of Ohio dairy cattle including specific CTX-M gene, additional β-lactamase genes carried by the isolates, plasmid incompatibility group, and replicon type of additional plasmids

Herd	No. of isolates	CTX-M gene type	Additional β-lactamase	CTX-M plasmida	Additional plasmid(s)
5	24	CTX-M-15		I1	F
	6	CTX-M-15	TEM–1	I1	F
6	4	CTX-M-14		F
	2	CTX-M-14		*
7	1	CTX-M-14		F
17	6	CTX-M-1	CMY–2	N
	1	CTX-M-1		N	A/C
19	6	CTX-M-1		N	Y
	6	CTX-M-1	TEM–1	N	Y
	7	CTX–M-1		N
	6	CTX-M-1		N	FIB
	1	CTX-M-1		*

^a*, plasmid could not be typed using our incompatibility group typing procedure.

All 30 fecal samples from herd 5 yielded an E. coli isolate with bla_CTX-M. PFGE of a subset of these 30 isolates indicated that they represented a single E. coli strain (Fig. 2), and plasmid analysis of every isolate indicated that all carried both an IncI1 plasmid of approximately 95 kb, bearing bla_CTX-M-15, and an IncF plasmid of approximately 60 kb in size. In addition, a subset of 6 of these isolates also carried bla_TEM-1, although we did not determine if this gene was localized to a specific plasmid. All fecal samples from herd 5 also yielded an E. coli isolate harboring bla_CMY-2. Plasmid replicon typing found that 5 of the E. coli isolates with bla_CMY-2 carried only an IncI1 plasmid, while another 13 of the E. coli isolates with bla_CMY-2 carried an IncI1 plasmid in addition to IncA/C and IncFIB plasmids. One additional E. coli isolate with bla_CMY-2 carried only IncI1 and IncA/C plasmids, while plasmids from the remaining E. coli isolates with bla_CMY-2 from this herd were not typeable using our procedures. Restriction digestion with AccI yielded multiple bands among the bla_CTX-M-positive IncI1 plasmids, while bla_CMY-2-positive IncI1 plasmids recovered from the same samples yielded no multiband pattern, indicating that these plasmids, while belonging to the same incompatibility group, are polymorphic at the crude single-enzyme RFLP level. Screening of a subset of these isolates by using mdh indicated that isolates from this herd were not the pandemic B2-O25:H4-ST131 strain.

Fig 2.

Dendrographic analysis and SpeI pulsed-field gel electrophoresis (PFGE) profiles for a subset of 44 Escherichia coli isolates containing bla_CTX-M recovered from dairy cattle fecal samples on 5 Ohio farms. The isolate identification (ID), antimicrobial resistance phenotype, plasmid replicon type (PRT), CTX-M gene, and additional β-lactamase genes of each isolate are indicated. The dendrogram was assembled by using Dice coefficients and the unweighted-pair group method using arithmetic averages. Default clustering settings of 1.00% optimization and 1.00% band position tolerance were used. ^a, isolate IDs are identified as herd number-isolate number. ^b, resistance phenotype indicates antimicrobial resistance or decreased susceptibility. Antimicrobial names with corresponding abbreviations are presented in Fig. 1.

E. coli isolates with bla_CTX-M were recovered from 6 (20%) of the herd 6 fecal samples, all representing a single clonal strain (Fig. 2). The only E. coli isolate with bla_CTX-M recovered from herd 7 samples produced a PFGE banding pattern identical to that of the isolates from herd 6. All 7 of these isolates from herds 6 and 7 carried bla_CTX-M-14. Four of these 6 E. coli isolates from herd 6 and the herd 7 isolate each harbored bla_CTX-M-14 on an IncF plasmid approximately 80 kb in size. The remaining 2 E. coli isolates with bla_CTX-M from herd 6 also carried an 80-kb plasmid with bla_CTX-M-14, but it was not typeable using our plasmid replicon typing procedure. All 7 isolates also carried a second plasmid of approximately 30 kb that was not typeable. No additional beta-lactamase genes were detected among these 7 E. coli isolates from herds 6 and 7. The fecal samples from herds 6 and 7 each produced an E. coli isolate with bla_CMY-2; two samples from each of these two herds were evaluated for plasmid replicon type. Both isolates from herd 6 carried bla_CMY-2 on IncA/C plasmids, while only one of the herd 7 isolates carried bla_CMY-2 on an IncA/C plasmid and the second carried it on an IncFIA plasmid. Isolates from these herds were not the pandemic B2-O25:H4-ST131 strain based on screening of the mdh sequencing results from a subset of isolates.

Two distinct E. coli strains each carrying bla_CTX-M-1 were observed in herd 17 (Fig. 2). Six of the 7 isolates containing bla_CTX-M-1 recovered from herd 17 shared an identical PFGE banding pattern. This subset of isolates expressed the bla_CMY-2 phenotype in addition to the bla_CTX-M phenotype, including resistance or reduced susceptibility to cefoxitin and the beta-lactamase inhibitors (Fig. 1). Carriage of bla_CMY-2 by these 6 isolates was confirmed by PCR, although bla_CMY-2 was not present on either the transconjugants or transformants of these isolates, suggesting that the bla_CMY-2 was located either on the bacterial chromosome or on a plasmid that we were unable to isolate in our laboratory recipient E. coli. All 7 E. coli isolates from herd 17 carried bla_CTX-M-1 on an IncN plasmid of approximately 40 kb, although the IncN plasmid of the unique strain appeared to be slightly larger, at approximately 55 kb. Interestingly, the single E. coli isolate with a unique PFGE pulsotype was the only isolate that did not also carry bla_CMY-2, but it was the only isolate to also harbor an IncA/C plasmid (Table 2), the replicon type that is commonly associated with bla_CMY-2 (32). Restriction digests of the IncN plasmids carrying bla_CTX-M-1 resulted in similar banding patterns (Fig. 3). Southern hybridization of the restriction digests indicated that bla_CTX-M was located on common fragments of all plasmids (Fig. 3). In addition, because bla_CTX-M alleles contain no internal AccI restriction sites, the observed hybridization of multiple fragments indicates gene duplication or partial duplication within the plasmids, similar to what has been reported for bla_CMY-2 carried on Inc A/C plasmids (2).

Fig 3.

(A) AccI restriction analysis of IncN plasmid DNA isolated from Escherichia coli transconjugants containing bla_CTX-M recovered from Ohio dairy cattle fecal samples in two herds. Lanes: 1, Fisher exACTGene 1kb Plus DNA ladder; 2, Roche Applied Science DNA molecular weight marker II (DIG labeled); 3, 17-7; 4, 17-12; 5, 19-4; 6, 19-7; 7, 19-10; 8, 19-12; 9, 19-13; 10, 19-15; 11, 19-17; 12, 19-19; 13, 19-29; 14, negative-control CMY-2 plasmid DNA; 15, Roche Applied Science DNA molecular weight marker II (DIG labeled); 16, Fisher exACTGene 1kb Plus DNA ladder (lanes 3 to 13 are identified as herd number-isolate number). (B) Southern blot hybridization using CTX-M probe of AccI restriction analysis of IncN plasmid DNA isolated from Escherichia coli transconjugants containing bla_CTX-M recovered from Ohio dairy cattle fecal samples in two herds. Lanes: 1, Fisher exACTGene 1kb Plus DNA ladder; 2, Roche Applied Science DNA molecular weight marker II (DIG labeled); 3, 17-7; 4, 17-12; 5, 19-4; 6, 19-7; 7, 19-10; 8, 19-12; 9, 19-13; 10, 19-15; 11, 19-17; 12, 19-19; 13, 19-29; 14, negative-control CMY-2 plasmid DNA; 15, Roche Applied Science DNA molecular weight marker II (DIG labeled); 16, Fisher exACTGene 1kb Plus DNA ladder (lanes 3 to 13 are identified as herd number-isolate number).

While isolates from the other 4 bla_CTX-M-positive study herds exhibited a high degree of within-herd clonality, the 26 bla_CTX-M-carrying E. coli isolates from herd 19 produced 6 different bacterial fingerprints on PFGE analysis (Fig. 4). We detected bla_CTX-M-1 on IncN plasmids ranging in size from approximately 30 to 60 kb for 25 of the 26 isolates. The remaining isolate carried bla_CTX-M-1 on a plasmid of similar size that could not be typed using our replicon typing procedure. Twelve of the 26 bla_CTX-M-carrying E. coli isolates also carried an IncY plasmid, and 6 other isolates contained an IncFIB plasmid. A subset of 6 isolates carried bla_TEM-1 in addition to bla_CTX-M-1 (Table 2). Restriction analysis and southern hybridization of AccI restriction digests of plasmid DNA from a subset of transconjugants containing only a single plasmid found that IncN plasmids harboring bla_CTX-M-1 produced nearly identical banding patterns with the bla_CTX-M-1 on a common fragment, suggesting a clonal IncN plasmid present in multiple herds and in multiple E. coli strains within the same herd (Fig. 5). Restriction digest banding patterns and fragments hybridizing the CTX-M probe were unique for plasmids of different incompatibility groups (Fig. 5).

Fig 4.

SpeI PFGE of 13 Escherichia coli isolates containing bla_CTX-M-1 on IncN plasmids recovered from dairy cattle fecal samples in a single Ohio dairy herd (herd 19). Lanes: 1, Salmonella enterica serovar Braenderup H9812; 2, 19-4; 3, 19-6; 4, 19-8; 5, 19-10; 6, 19-11; 7, 19-12; 8, 19-13; 9, 19-18; 10, 19-19; 11, 19-20; 12, 19-21; 13, 19-22; 14, 19-23; 15, Salmonella Braenderup H9812 (lanes 2 to 14 are identified as herd number-isolate number).

Fig 5.

(A) AccI restriction analysis of plasmid DNA from Escherichia coli transconjugants that contained a single plasmid carrying bla_CTX-M representing 5 Ohio dairy herds. Lanes: 1, Fisher exACTGene 1kb Plus DNA ladder; 2, Roche Applied Science DNA molecular weight marker II (DIG labeled); 3, 5-5; 4, 6-17; 5, 7-23; 6, 17-9; 7, 17-29; 8, 19-1; 9, 19-2; 10, 19-7; 11, 19-10; 12, 19-14; 13, 19-18; 14, negative-control CMY-2 plasmid DNA; 15, Roche Applied Science DNA molecular weight marker II (DIG labeled); 16, Fisher exACTGene 1kb Plus DNA ladder (lanes 3 to 13 are identified as herd number-isolate number). (B) Southern blot hybridization using CTX-M probe of AccI restriction analysis of plasmid DNA from a subset of Escherichia coli transconjugants that contained a single plasmid carrying bla_CTX-M representing 5 Ohio dairy herds. Lanes: 1, Fisher exACTGene 1kb Plus DNA ladder; 2, Roche Applied Science DNA molecular weight marker II (DIG labeled); 3, 5-5; 4, 6-17; 5, 7-23; 6, 17-9; 7, 17-29; 8, 19-1; 9, 19-2; 10, 19-7; 11, 19-10; 12, 19-14; 13, 19-18; 14, negative-control CMY-2 plasmid DNA; 15, Roche Applied Science DNA molecular weight marker II (DIG labeled); 16, Fisher exACTGene 1kb Plus DNA ladder (lanes 3 to 13 are identified as herd number-isolate number).

All bla_CTX-M gene-harboring plasmids were transferred in conjugation experiments. PCR for bla_CTX-M performed on DNA isolated from MG1655 transconjugants yielded amplicons of the expected size of 544 bp (26, 29, 43). Subsequent plasmid replicon typing and sequencing confirmed carriage of the expected replicons and bla_CTX-M alleles that were identified in their respective parental donor strains. SHV and OXA genes were not detected in this group of E. coli isolates with bla_CTX-M.

Mean owner-reported herd-level ceftiofur treatment rates for bla_CTX-M-positive herds were 3.3% and 13.5% for the previous 1 and 6 months, respectively. Ceftiofur treatment rates of 3.6% and 17.2% for bla_CTX-M-negative herds were reported over the same 2 time periods. Neither ceftiofur treatment rate for the prior 1 month or 6 months was associated with the probability of recovery of E. coli harboring bla_CTX-M or bla_CMY-2 from fecal samples in these herds by using multivariable logistic regression models that adjusted for herd size as a continuous independent variable.

DISCUSSION

We recovered E. coli carrying bla_CTX-M from 70 (9.4%) of the 747 dairy cattle fecal samples collected in this study. bla_CTX-M-carrying isolates were present in 5 (20%) of 25 participating herds, with recovery rates within herds ranging from 3 to 100% of samples. Bovine fecal E. coli isolates harboring bla_CTX-M were first reported in U.S. livestock in 2010 (52). While the prevalence of bla_CTX-M-positive E. coli in the enteric flora of U.S. food animals has been previously unaddressed, studies from both Europe and Asia have shown the recovery of bla_CTX-M from livestock with increasing frequency (27, 46). Our results suggest that the prevalence of bla_CTX-M in U.S. food animal populations may be similar to those seen in Europe and Asia.

We found no association between the owner-reported therapeutic use of ceftiofur and the recovery of bla_CTX-M-positive E. coli from bovine fecal flora. It is important to note that all herds, with the exception of the single organic dairy, reported ceftiofur as part of their treatment policy options. Mean ceftiofur treatment rates of 3.6% and 16.4% were reported for the previous 30 days and 6 months prior to sampling, respectively, in these herds. Ceftiofur use in swine has been associated with higher recovery of fecal coliform bacteria with bla_CTX-M from individual treated animals than from untreated controls (6), but similar data for cattle are not available. At the herd level, there is a paucity of information regarding a potential association between ceftiofur use and the recovery of bla_CTX-M-mediated resistance. Amplification of another extended-spectrum-cephalosporin resistance gene, bla_CMY-2, in fecal E. coli has been associated with ceftiofur use in livestock (19, 28, 47). However, others have reported no association of ceftiofur use with extended-spectrum-cephalosporin resistance or simply a reduction in susceptible bacteria allowing for these rare resistant phenotypes to be detected (10, 16, 30, 42). The frequent use of ceftiofur in dairy herds may make the detection of a true association between ceftiofur administration and the recovery of bla_CTX-M-positive fecal E. coli difficult. Our characterization of herd ceftiofur use as the proportion of treated cows also may not be an appropriate measure to fully capture the complex interactions of management, environment, and antimicrobial selection pressure in the dissemination of bacterial resistance genes.

PFGE analysis of the 70 bla_CTX-M-positive E. coli isolates indicated that the clonal dissemination of a single or predominant strain was common within herds 5, 6, and 17. Of note, the PFGE pulsotype of the single herd 7 CTX-M-producing E. coli isolate was identical to those from herd 6, which is located in the same county in Ohio but is otherwise epidemiologically unrelated. This suggests between-herd transmission of this resistant strain, which could result from the movement of people, domestic animals, and wildlife between farms. The fact that these two herds were also sampled for this research project on the same day suggests the possibility of cross-contamination of samples at the farm or in the laboratory by the investigators. The total plasmid content of the single herd 7 isolate was also identical to that of the herd 6 isolates.

While the E. coli isolates harboring bla_CTX-M in the other herds displayed notable within-herd homogeneity, the E. coli isolates from herd 19 revealed 6 different bacterial fingerprints on PFGE. Multiple pulsotypes of bla_CTX-M-positive E. coli isolates from individual dairy herds have been previously reported in Europe (12, 27). Differences in the environment, management, or antimicrobial use practices that might have resulted in the exchange of a single plasmid among multiple E. coli strains are not clear. The owner of this herd did not report unusual ceftiofur use compared to other herds in the study, as 1.7% and 22% of cows were treated with ceftiofur in the previous 1 month and 6 months, respectively.

We found that the bla_CTX-M gene was carried by plasmids of incompatibility groups I1, F, and N with a high degree of herd-level clonality (Table 2). In herd 5, the CTX-M-15 gene was located on an IncI1 plasmid; the CTX-M-14 gene of herds 6 and 7 was carried on an IncF plasmid, and the CTX-M-1 gene from herds 17 and 19 was located on an IncN plasmid. Both IncI1 and IncN plasmids have been suggested as the animal reservoir for the bla_CTX-M-1 gene, with a high prevalence reported in livestock and poultry as well as in retail meat products (33). The IncF plasmid family is often carried by the enteric floras of both humans and animals worldwide and has been reported to harbor various resistance genes (3). These 3 major plasmid families have been previously associated with specific bla_CTX-M genes, illustrating that the dissemination of this resistance gene is strongly linked to its association with a plasmid rather than an epidemic bacterial strain (13).

Individual CTX-M-1, -14, and -15 alleles were clonally conserved within herds. CTX-M-1 and -15 of group A1 and the CTX-M-14 of group 9 are among the most common of the CTX-Ms, often cultured from both humans and animals globally, and have all been identified with highly virulent multidrug-resistant (MDR) E. coli ST131-O25:H4-B2 (33). Our recovery of these three specific alleles from dairy cattle corresponds with the range of bla_CTX-M genes reported from human isolates in North America (15). The global emergence of this mobile resistance determinant has been compared to that of the broad-spectrum TEM beta-lactamase seen in the 1960s (18) and the proliferation of the bla_CMY-2 enzyme in the 1990s, with animal populations serving as a potential genetic reservoir (50).

In addition to recovering bla_CTX-M, we found that 100% of herds and 95% of fecal samples had E. coli carrying bla_CMY-2, with owner-reported ceftiofur use on 24 (96%) of the 25 farms. We previously reported that ceftiofur was used by 61% of Ohio dairy farms in 2006 (47) and 88% of such farms in 2009 (16), suggesting that ceftiofur use by dairy producers may have increased. Over the same time period, we recovered E. coli with bla_CMY-2 from 67% of herds and 34% of individual cow fecal samples in 2006 (47) and from 92% of herds and 70% of cows in 2009 (16). With on-farm use of ceftiofur potentially contributing to the emergence, maintenance, and dissemination of extended-spectrum-cephalosporin resistance genes within dairy herds, our observed increase in bla_CMY-2 prevalence in dairy populations concurrent with increasing ceftiofur use may portend a similar dissemination of bla_CTX-M in U.S. livestock.