Molecular Epidemiology of bla_CMY-2 Plasmids Carried by Salmonella enterica and Escherichia coli Isolates from Cattle in the Pacific Northwest

Nontyphoid Salmonella enterica subsp. enterica (NTS) causes approximately 36,000 confirmed cases of food-borne illness in the United States and an estimated 1.4 million unreported cases each year (9, 17, 25). A trend toward increasing antimicrobial resistance to broad-spectrum cephalosporins in some NTS serovars has raised concern because these drugs are regarded as first-line therapy in pediatric salmonellosis (11). The most recent surveillance data indicate that 0.6% of U.S. human NTS isolates are resistant to ceftriaxone and 3.4% are resistant to ceftiofur (a veterinary broad-spectrum cephalosporin) (7). The predominant mechanism of cephalosporin resistance in NTS in the United States is a cephamycinase encoded by plasmid-borne bla_CMY-2 genes (11, 27, 28, 30). bla_CMY-2, which likely originated from the chromosomal AmpC locus of Citrobacter freundii (1, 19), has been observed in plasmids in several species of the Enterobacteriacea (2-4, 15, 16, 18, 27, 29).

Commensal bacteria may serve as a reservoir of plasmid-borne antimicrobial resistance genes for pathogens, and there is evidence that plasmid transfer occurs readily between Escherichia coli and S. enterica. For example, a phylogenetic analysis of F plasmid-specific genes from reference collections of S. enterica and E. coli found several examples in which finO and traD sequence variants were shared between the two species (6). Furthermore, bla_CMY-2 Southern blot experiments with plasmids from E. coli and S. enterica have revealed similarities among isolates, suggesting that sequences (in addition to the bla_CMY-2 open reading frame) are shared among the plasmids harbored by these microbial genera (27, 28).

Cattle could represent an important niche for transfer of bla_CMY-2 plasmids between E. coli and NTS. Both bacterial genera inhabit the bovine gastrointestinal tract, and selection pressure favoring cephalosporin resistance is ubiquitous in some cattle production systems due to the frequent use of ceftiofur (13). For example, between 2001 and 2003, the percentage of ceftiofur-resistant NTS rose more rapidly among isolates from U.S. cattle than in those from human, chicken, turkey, and swine hosts (8). The predominance of bla_CMY-2--mediated cephalosporin resistance among NTS and E. coli isolates from cattle (20, 24, 30) led us to investigate the relationship between bla_CMY-2 plasmids and the two microbial genera. Assessing the diversity of host chromosomal and plasmid DNAs from commensal isolates of E. coli and clinical isolates of S. enterica permitted us to evaluate whether bla_CMY-2 dissemination in this ecological niche is clonal or due to epidemic plasmid spread and whether the nature of this process differed in a pathogen (S. enterica) and a potential reservoir of antimicrobial resistance genes (E. coli).

E. coli isolates from 46 animals originating in 14 herds and S. enterica isolates from 48 animals with salmonellosis originating in 47 herds were chosen to represent the bovine commensal flora and a major bovine pathogen, respectively. All isolates were obtained from cattle in Washington state or Idaho between 2001 and 2003. All isolates gave an amplicon of the appropriate length when tested by PCR using bla_CMY-2-specific primers described by Zhao et al. (30) (Table 1).

E. coli and S. enterica serotypes were determined by the Gastroenteric Disease Center (University Park, PA) and the National Veterinary Services Laboratory (Ames, IA), respectively. Isolates were assessed for pulsed-field gel electrophoresis (PFGE) type in accordance with PulseNet protocols (23). Plasmids were isolated by electroporation into E. coli DH10B and prepared for PstI restriction fragment length polymorphism (pRFLP) typing using previously described methods (12, 21). pRFLP typing and bla_CMY-2 Southern blotting were performed as described by Giles et al. (12) but using continuous voltage (7.2 V/cm for 1.5 h). Agar diffusion susceptibility testing was performed in accordance with CLSI standards (10). Gel images were analyzed using Bionumerics (Applied Maths, Belgium) with optimization and tolerance settings determined by the minimum values required to classify a standard plasmid (which was included with each gel) as indistinguishable from itself in an unweighted pair group method with arithmetic mean (UPGMA) analysis.

S. enterica included four serotypes: Newport (n = 35), Typhimurium (n = 5), Dublin (n = 7), and Muenster (n = 1); PFGE patterns were highly similar within serotypes (Fig. 1). Sixteen serotypes were identified among the 28 E. coli isolates that were typeable for both O and H antigens, but PFGE patterns were markedly more diverse than in the Salmonella spp.; each of the 46 isolates displayed a unique pattern (data not shown). The differences in serotype and PFGE diversities between the two genera may reflect the sources of the isolates: clinical versus commensal bacteria from healthy animals. Pathogenic Salmonella spp. have been described as inherently clonal (5, 14), whereas relatively little is known about the genetic diversity of nonpathogenic E. coli isolates from animal sources. Winokur et al. found diverse PFGE patterns among 55 bla_CMY-2 commensal E. coli isolates from clinical veterinary specimens (28). Taken together, these findings suggest that bla_CMY-2-bearing commensal E. coli isolates are not strongly clonal at the serotype and PFGE levels, regardless of the clinical status of the source.

Forty pRFLP patterns were observed among the 94 isolates; however, repeatability analyses using a subset of 14 plasmids demonstrated that consistent self-grouping was observed only at the 90% similarity level. Thus, we considered plasmids with ≥90% similarity indistinguishable, resulting in the designation of 34 unique patterns. Four reference plasmids (A, B, C, and D), described by Giles et al. (12), were included as positive controls. Twelve pRFLP patterns were observed in more than one bacterial isolate, and two patterns were observed in both genera, consistent with exchange of some plasmids among commensal E. coli and S. enterica isolates (Fig. 2). Eighty-five plasmids had bla_CMY-2 Southern blot fragments identical to the A or C patterns previously described and were conserved within pRFLP types, consistent with horizontal-transfer activity (Fig. 3).

Each Salmonella serotype tended to be associated with a specific plasmid variant. Notably, for S. enterica serovar Newport, 26 of 35 isolates originated from 26 different herds but shared a single pRFLP type (type 18). Using the model of plasmid-bacterial-host associations proposed by Souza and Eguiarte (22), the relationship between S. enterica serovar Newport and its plasmids could be described as clonal, implying that bla_CMY-2 plasmids in this serotype were largely disseminated with an epidemic host bacterium. In contrast, the diversity of plasmids from E. coli was high, reflecting the high level of PFGE and serotype diversity observed between the isolates. The exceptions to this observation were five pRFLP types (10, 13, 16, 17) associated with multiple serologically distinguishable host strains, consistent with the idea of epidemic plasmids (22).

Although there was evidence of interspecies sharing of plasmids, the predominance of only two plasmid variants (A and C) in Salmonella isolates from an animal niche containing a plethora of E. coli-borne plasmid variants (presumably available for transfer to Salmonella) was conspicuous, suggesting that the major mechanisms of bla_CMY-2 dissemination differ between S. enterica and E. coli. This pattern is consistent with a recent study by Welch et al., who found greater diversity of plasmids among E. coli isolates than among the S. enterica isolates, using PCR primer sets representing 13 widely spaced loci from an entirely sequenced IncA/C bla_CMY-2 plasmid from S. enterica serovar Newport (26). Our observation that isolates limited to a solitary niche (cattle) are similarly diverse suggests a model of bla_CMY-2 dissemination in which insertions and deletions that occur during promiscuous plasmid sharing among E. coli isolates occasionally result in plasmids that are successful in a Salmonella host (such as pRFLP types 13 and 18). Also consistent with this model, conjugation experiments revealed that 40% of E. coli plasmids (versus 2% of Salmonella plasmids) were able to transfer or be mobilized to a Nal^r DH5α recipient. The subsequent success of a Salmonella host/plasmid clone, then, is likely modulated by diverse factors, including virulence, infectivity, and environmental persistence, as well as antimicrobial selection pressures. Factors that promote interspecies exchange of antimicrobial-resistance plasmids and enhance dissemination of S. enterica clones merit further study.

• Open in new tab
• Download powerpoint

FIG. 1.

UPGMA dendrograms with negative images of ethidium bromide-stained gels from XbaI PFGE of bla_CMY-2-positive S. enterica from cattle following normalization and analysis with Bionumerics software. The PFGE type designations correspond to those referred to in Table 1. Ser., serotype.

• Open in new tab
• Download powerpoint

FIG. 2.

UPGMA dendrogram with a negative image of an ethidium bromide-stained PstI plasmid RFLP gel after normalization in Bionumerics software. The bracketed strains indicate identical RFLP patterns from plasmids isolated from both E. coli and S. enterica. Plasmid type designations correspond to those in Table 1. Reference A, B, C, and D plasmids (12) were included as controls.

• Open in new tab
• Download powerpoint

FIG. 3.

(A) pRFLP of A, B, C, and D reference plasmids, followed by three examples of bla_CMY-2 plasmids from S. enterica isolates from Washington state cattle. Lanes: 1, 8, and 12, 12-kb ladder; 2, λ\HindIII markers; 3, 4, 5, and 6, A, B, C, and D reference plasmids (12); 9, 10, and 11, pS7907, pS7909, and pS8129; 7, undigested reference plasmid D. (B) Southern hybridization of the gel from panel A with a full-length CMY-2 probe.

• Copyright © 2007 American Society for Microbiology