INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Currently, there is increased public and scientific interest in the use of therapeutic and subtherapeutic antimicrobials in animals. This is due primarily to the possible emergence and dissemination of multiple-drug-resistant zoonotic bacterial pathogens (1, 8, 9, 22, 27, 32). Antimicrobial drug-resistant bacterial pathogens in animals pose a risk not only to animal health, but possibly to humans via transmission as food-borne pathogens (8, 9, 22, 27, 32).

Infections caused by antibiotic-resistant bacteria are a severe and costly animal health problem. These infections prolong illness and, if not treated in time with more costly, alternative antimicrobial agents, can lead to increased morbidity and mortality. One of the more challenging diseases facing cattle veterinarians and producers today is colibacillosis, a severe form of diarrhea, which causes significant financial loss to cow and calf producers (10, 37). There are multiple infectious etiological agents that can cause colibacillosis in cattle; however, Escherichia coli is recognized as the single most important bacterial cause (31, 36, 37). Numerous E. coli serotypes have been incriminated, but the majority are enterotoxigenic E. coli (ETEC) strains. These strains usually possess the K99 (F5) adhesion and produce heat-stable (STa or STb) and/or heat-labile (LT) enterotoxins (36, 37). The most important aspect of treating E. coli-related colibacillosis is to correct the accompanying electrolyte loss, dehydration, and acidosis. However, antimicrobial therapy is often initiated at the same time in an attempt to eliminate the pathogenic E. coli.

Florfenicol [d-threo-3-fluoro-2-dichloroacetamido-1-(4-methylsulfonylphenyl)-1-propanol] is a fluorinated structural analog of thiamphenicol and chloramphenicol approved by the Food and Drug Administration (FDA) in 1996 for treatment of bovine respiratory pathogens such as Pasteurella spp. However, it is not currently approved for treatment of E. coli-related cattle enteric diseases in the United States. Additionally, there are no approved NCCLS breakpoints for E. coli currently available; however, the resistance breakpoint for bovine respiratory pathogens (i.e., Pasteurella multocida) is 8 µg/ml (25).

Florfenicol is not approved for human use; however, it is related to chloramphenicol and can select for cross-resistance among bacterial pathogens. Florfenicol has been shown to have a spectrum of activity similar to that of chloramphenicol, except that it is active at lower concentrations than chloramphenicol against a variety of clinical bacterial isolates, including chloramphenicol-resistant bacteria (16, 26, 33). Florfenicol's mechanism of action is directed at disrupting bacterial protein synthesis by binding to the 50S subunit of the bacterial ribosome and is generally considered to be bacteriostatic (12, 23, 26, 33).

Neither chloramphenicol acetyltransferase (CAT), the enzyme responsible for most of the plasmid-mediated resistance to chloramphenicol (12), nor the known nonenzymatic chloramphenicol resistance gene (cmlA) confers resistance to florfenicol (14, 18, 23). Until recently, no genes conferring resistance to fluorinated derivatives of chloramphenicol or thiamphenicol were known. However, in 1996 researchers in Japan identified a novel plasmid-encoded gene (pp-flo) from Photobacterium piscicida that encoded resistance to both chloramphenicol and florfenicol (19). More recently, Bolton et al. described a gene with 97% homology to the pp-flo gene among Salmonella enterica serovar Typhimurium DT104 isolates, which they termed flo_St (9). Salmonella isolates possessing the flo_St gene also exhibited dual resistance to chloramphenicol and florfenicol (4, 9). The flo gene has also been identified recently among chloramphenicol-resistant Salmonella enterica serovar Agona isolates recovered from poultry in Belgium and florfenicol-resistant E. coli from poultry in the United States (13, 18).

The present study involves 48 antimicrobial-resistant strains of E. coli isolated from diarrheic calves and submitted to the North Dakota Veterinary Diagnostic Laboratory (ND-VDL) from 1997 to 1998. Since there was limited information regarding florfenicol resistance in the literature and virtually none concerning E. coli, a study was initiated to determine the mechanism of resistance among these bovine E. coli isolates.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Bacterial isolates. The present study focuses on 48 E. coli strains recovered from calf diarrhea cases submitted to the ND-VDL from 1997 to 1998. All calves were under 2 weeks of age, and several of the case histories indicated that florfenicol was used in an extra-label manner in an attempt to cure the diarrhea. E. coli bacteria were isolated from necropsy or fecal specimens from calves suffering diarrhea. The specimens were plated on blood and MacConkey agar plates. Indole and oxidase tests were performed on lactose-positive colonies. API 20E test strips (bioMérieux Vitek, Hazelwood, Mo.) were also used to confirm E. coli identification. Bovine E. coli isolates displaying decreased susceptibilities to florfenicol were subsequently collected for further analysis to determine the mechanism of resistance. Isolates were stored as 10% glycerol stocks at 80°C until analysis.

Antimicrobial susceptibility determination. Antimicrobial MICs for E. coli isolates were determined using the Sensititre automated antimicrobial susceptibility system (Trek Diagnostic Systems, Westlake, Ohio) and interpreted according to the NCCLS guidelines for broth microdilution methods (24, 25). Sensititre susceptibility testing was performed according to the manufacturer's instructions. The following antimicrobials were assayed: amikacin, amoxicillin-clavulanic acid, ampicillin, apramycin, ceftiofur, ceftriaxone, cephalothin, chloramphenicol, ciprofloxacin, florfenicol, gentamicin, kanamycin, nalidixic acid, streptomycin, sulfamethoxazole, tetracycline, and trimethoprim-sulfamethoxazole. E. coli ATCC 25922, E. coli ATCC 35218, and Pseudomonas aeruginosa ATCC 27853 were used as controls in MIC determinations.

PCR, isolation, and sequencing of the E. coli flo and cmlA genes. Oligonucleotides were synthesized by Biosynthesis (Lewisville, Tex.). The PCR-derived flo probe was obtained using previously published primers (9) and yielded a 215-bp amplicon within the coding region. Two additional PCR primers were created flanking the entire pp-flo gene. E. coli Flo-F corresponded to nucleotides 981 to 1004 (TTGTTGTTGCGGCGCTCTGTAAGG), and E. coli Flo-R corresponded to nucleotides 2405 to 2384 (CGGCGACGGCGATGAACTGAAC) of the published sequence (GenBank accession no. D37826), yielding a predicted amplicon of 1,424 bp. The PCR-derived cmlA probe was amplified using previously published primers (18) and yielded a 698-bp amplicon. PCRs were carried out in a thermocycler 2400 (Perkin-Elmer Applied Biosystems, Norwalk, Conn.). Two hundred microliters of each PCR product was separated by 1.5% agarose gel electrophoresis at 75 V for 2 h. The appropriate bands were excised and purified utilizing GenElute Spin columns (Supelco, Bellefonte, Pa.). The purified PCR amplicons were sequenced at the University of Minnesota Advanced Genetic Analysis Center, St. Paul, Minn. Sequence comparisons were made using the NCBI-BLAST program (2).

PFGE. Five-milliliter broth cultures of each isolate were grown overnight at 37°C. The cell pellet was collected, embedded in agarose plugs, and lysed as previously described (5, 30). Agarose plugs were digested with 10 U of XbaI (Promega, Madison, Wis.) overnight at 37°C. Digested DNA was separated on a 1.2% agarose gel using the CHEF-DRII pulsed-field gel electrophoresis (PFGE) system (Bio-Rad, Hercules, Calif.). Electrophoresis was carried out for 25 h at 6 V with a ramped pulse time of 2 to 40 s in 0.5× Tris-borate-EDTA (TBE) buffer (14°C). Molecular weight standards were Saccharomyces cerevisiae YPH 755 chromosomes (Roche Biochemicals, Indianapolis, Ind.). Gel documentation and phylogenetic analysis of the E. coli isolates were performed using the RFLPScan and TREECON programs as previously described (21).

Mapping of the flo gene in bovine E. coli isolates. The flo gene was initially identified in a transferable R plasmid in the fish pathogen P. piscicida (18). A flo homolog was recently mapped within a multidrug resistance locus that is flanked by two class 1 integrons in serovar Typhimurium DT104 and serovar Agona (3, 11, 13). PCR primers were developed to determine if the E. coli flo gene had a genetic organization similar to that observed in either serovar Typhimurium DT104 or P. piscidida. The first PCR primers were designed to amplify a portion of flo and genes mapping downstream (tetR) within the Salmonella multidrug resistance (MDR) locus (GenBank accession no. AF071555). The second PCR set was targeted to flo and the broad-host-range plasmid RSF1010 replicon identified in P. piscicida (GenBank accession no. D37826). Primers flomap2F (GGGATCGGCGAAACTTTAC) and flomap2R (TGTGGTCGGTTCCGTTCTC) anneal at nucleotide positions 5330 of flo and 6073 of tetR within the serovar Typhimurium DT104 (GenBank accession no. AF07155), resulting in a 744-bp amplicon. Primers flomap2F (GGGATCGGCGAAACTTTAC) and pp-floMapR (TCCGCGTCCTTGCAATAC) anneal to nucleotide positions 1953 and 2723 of the P. piscicida plasmid-mediated flo sequence (GenBank accession no. D37826), resulting in a 771-bp amplicon. The PCR conditions used have been previously described (17).

Detection of high-molecular-weight plasmids by S1 nuclease treatment and PFGE. Due to their structure, large circular plasmids are not resolved by PFGE. This problem can be overcome by using S1 nuclease digestion prior to PFGE. S1 nuclease creates breaks in the plasmids, which allow them to enter the agarose gel during PFGE. Agarose plugs prepared for PFGE were treated with S1 nuclease (Roche Biochemicals) as previously described (6, 22). Plasmids were separated on a 1.5% agarose gel using the CHEF-DRII PFGE system (Bio-Rad). Plasmids were electrophoresed initially for 14 h at 6 V with a pulse time of 45 s, followed by a program of 6 h at 6 V with a pulse time of 25 s. Molecular weight standards were S. cerevisiae YPH 755 chromosomes (Roche Biochemicals).

Southern hybridization analysis. Megabase DNA was blotted to nylon membranes using the Bio-Rad procedure. Briefly, DNA in ethidium bromide-stained gels was nicked using a UV Transilluminator (Fisher Scientific) at 60 mJ of energy. Gels were soaked in 0.4 N NaOH-1.5 M NaCl for 15 min, and then DNA was transferred for 4 h to Magnagraph nylon (MSI, Westborough, Mass.) using the model 785 Vacuum Blotter (Bio-Rad) with NaOH as the buffer. After transfer, nylon membranes were washed briefly in 0.5 M Tris, pH 7.6, followed by 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) (29). Membranes were blocked and probed with a PCR-amplified portion of the florfenicol resistance gene probe as previously described (18).

Nucleotide sequence accession number. The sequence of the bovine E. coli flo gene has been assigned GenBank accession no. AF252855.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Antimicrobial resistance patterns in bovine E. coli. Forty-eight E. coli isolates recovered from calves with diarrhea were tested for their resistance to antimicrobial agents of human and veterinary significance according to NCCLS broth microdilution methods and guidelines (Table 1). Interestingly, 90% of bovine E. coli isolates were resistant to chloramphenicol (Table 1), an antibiotic that has been banned from veterinary use in food animals in the United States since the 1980s (15). Ninety-two percent of E. coli isolates were also resistant to florfenicol based on the NCCLS breakpoint for bovine respiratory pathogens, 8 µg/ml (25). In addition to chloramphenicol and florfenicol resistance, all bovine isolates were resistant to at least four antimicrobials and 37 (77%) isolates were resistant to at least nine antimicrobials. The most common antimicrobial resistance pattern observed in the bovine E. coli isolates included resistance to chloramphenicol, florfenicol, amoxicillin, ampicillin, ceftiofur, cephalothin, gentamicin, kanamycin, streptomycin, sulfamethoxazole, tetracycline, and trimethoprim-sulfamethoxazole (n = 14).

Identification of the flo resistance gene in bovine E. coli. PCR primers, based on the published pp-flo nucleotide sequence, amplified a 1.4-kb PCR product from chloramphenicol- and florfenicol-resistant bovine E. coli isolates. DNA sequence analysis of the E. coli flo gene demonstrated 98% identity to the published GenBank sequences of both serovar Typhimurium flo_St and P. piscicida pp-flo (AF118107 and D37826, respectively). DNA sequence analysis using the NCBI-BLAST program (2) predicted an open reading frame (ORF) of 1,215 bp, demonstrating the greatest identity (98%) with the pp-flo gene from P. piscicida (19), the cmlA-like gene described in the DT104 MDR locus (11), and the flo_St gene described by Bolton et al. (9). This 1,215-bp ORF translates into a hypothetical protein of 404 amino acids. Using BLAST-p (2), it was determined that this hypothetical protein demonstrated 98 and 89% amino acid sequence homology with the previously described flo_St and pp-flo gene products, respectively. There was also 49% identity between the E. coli hypothetical Flo protein and the previously described nonenzymatic chloramphenicol resistance gene product CmlA.

Evidence that the flo resistance gene is widely disseminated among bovine E. coli isolates. Forty-one (85%) of the E. coli isolates assayed were resistant to both chloramphenicol (MIC, 32 µg/ml) and florfenicol (MIC, 16 µg/ml). For 36 of these E. coli isolates, florfenicol MICs were 128 µg/ml, higher than those previously reported for florfenicol-resistant E. coli (9) but comparable to MICs observed for serovar Typhimurium DT104 (18). These isolates were further analyzed for the presence of both flo and cmlA efflux genes. Forty-two of 44 isolates that displayed decreased susceptibilities to florfenicol (MIC, 16 µg/ml) were positive by PCR for the flo gene (Table 2). All E. coli isolates for which florfenicol MICs were 8 µg/ml were negative for the flo gene (n = 4).

View larger version (42K): [in this window] [in a new window]   FIG. 1.   Phylogenetic tree of florfenicol-resistant bovine E. coli. Total DNA was digested with XbaI and separated by PFGE. Molecular weight markers were S. cerevisiae strain YPH 755 chromosomes. Similarities among E. coli PFGE patterns were identified by cluster analysis using the neighbor-joining method to draw a phylogenetic tree (21). Phylogenetic analysis identified 44 PFGE patterns in the 45 florfenicol-resistant bovine E. coli isolates assayed.

The flo resistance gene maps to high-molecular-weight plasmids in bovine E. coli. Recently, Briggs and Fratamico characterized the MDR locus of the penta-antibiotic-resistant serovar Typhimurium DT104 (11). This locus contains the florfenicol resistance gene flo, which is flanked on one side by a class 1 integron containing the streptomycin/spectinomycin resistance gene aadA2 and on the other by the tetracycline resistance gene, tetA (3, 11, 13). To determine whether flo-positive bovine E. coli isolates have a similar genetic organization, PCR primers targeted to flo and genes downstream in the MDR locus of serovar Typhimurium DT104 were designed (Fig. 2). PCR analysis of a serovar Typhimurium DT104 isolate resulted in a PCR product of approximately 750 bp, corresponding in size to the predicted amplicon between flo and tetR (Fig. 2). No PCR amplicons were observed for primers anchored in DNA sequences downstream of the DT104 MDR locus and flo in the florfenicol-resistant bovine E. coli isolates. However, several florfenicol-resistant bovine E. coli isolates yielded 770-bp amplicons using PCR primers within flo and downstream within the P. piscicida R plasmid RSF1010 replicon (Fig. 2). This suggests horizontal transfer of the E. coli flo gene via acquisition of large broad-host-range plasmids.

View larger version (20K): [in this window] [in a new window]   FIG. 2.   Comparison of the location of the flo gene in serovar Typhimurium DT104, P. piscicida, and E. coli. PCR primers were designed to amplify a portion of flo and genes mapping downstream (tetR) within the Salmonella MDR locus or to amplify flo and the RSF1010 replicon identified in P. piscicida. Primers flomap2F and pp-floMapR anneal to nucleotide positions 1953 and 2723 of the P. piscicida plasmid-mediated flo sequence, resulting in a 771-bp amplicon. Primers flomap2F and flomap2R anneal at nucleotide positions 5330 of flo and 6073 of tetR within the serovar Typhimurium DT104 MDR locus, resulting in a 744-bp amplicon.

View larger version (44K): [in this window] [in a new window]   FIG. 3.   Location of the flo resistance gene in bovine E. coli genomes by PFGE and Southern analysis. Total E. coli genomic DNA, plasmid and chromosome, was cut with XbaI, separated by PFGE, and probed with labeled flo. Lanes 2 to 15, bovine E. coli isolates. Avian E. coli strain 5790 served as the positive control for flo (lane 1).

Is the widespread dissemination of florfenicol resistance in bovine E. coli due to a common flo-containing plasmid? A common 225-kb flo-containing plasmid was identified in several genetically distinct bovine E. coli isolates. Southern analysis was used to determine if the florfenicol resistance gene, flo, mapped to a common XbaI DNA fragment. This would indicate that florfenicol resistance emerged in bovine E. coli due to dissemination of a common resistance plasmid containing flo. On the contrary, we found that flo mapped to a diverse array of XbaI DNA fragments in bovine E. coli, ranging in size from 72 to 529 kb (Fig. 3, lanes 2 to 15). The flo gene mapped to 14 distinct XbaI DNA fragments in 32 bovine E. coli isolates that were positive for flo. A common 124-kb XbaI DNA fragment containing flo was identified among nine genetically distinct bovine E. coli isolates. However, the 124-kb XbaI fragment was not located on a common plasmid in those isolates. The flo probe also hybridized to two XbaI fragments in four of the E. coli isolates, indicating that multiple copies were present (Fig. 3, lanes 2 to 5). These data suggest that the dissemination of florfenicol resistance is not attributed to one common plasmid.