Chlortetracycline alone and combined administration of chlortetracycline and sulfamethazine were selected as experimental treatments on the basis of their routine use in the Canadian feedlot industry. These antimicrobials are used to improve feed efficiency and prevent foot rot, liver abscesses and respiratory disease. Virginiamycin was included in the study as an antibiotic to which neither the steers nor their dams would have had prior exposure, given that it is not registered for use in cattle in Canada.
Resistance to amikacin, ceftriaxone (64 μg/ml), cefoxitin or nalidixic acid was not detected in any of the 531 E. coli isolates examined. Other researchers of E. coli from Canadian beef cattle have also reported the absence of resistance to these antibiotics  or, when resistance to nalidixic acid was found, it occurred in fewer than 2% of isolates studied . In the present study, the absence of resistance to these antibiotics in gut flora may be related to sole-source acquisition of the calves, and to the complete absence of antibiotic use prior to their arrival at the feedlot. Furthermore, our research feedlot had been constructed just prior to commencement of this experiment, thus there was no history of prior administration of subtherapeutic antibiotics at this site. Our results and those of others [30, 31] contrast with those of Hoyle et al. , who reported that all calves from a Scottish beef farm were found to shed nalidixic acid-resistant E. coli at least once during a 21-wk study.
Comparisons of AMR E. coli from steers in CON vs. T, TS and V groups suggests that subtherapeutic administration of these antimicrobials had only a limited impact on the nature of antimicrobial resistance in E. coli resident in these cattle. The resistances observed most commonly among these E. coli isolates were to tetracycline, sulfamethoxazole, ampicillin, chloramphenicol and streptomycin, which is consistent with the findings of other Canadian beef researchers [30, 31, 33].
In general, the antibiogram type and temporal point of isolation were more similar between isolates from CON and V groups than from those in T or TS. Virginiamycin, a streptogramin, that primarily targets Gram-positive bacteria , and appears to have had minimal influence on the nature of AMR in the non-target E. coli isolates obtained in this study. Similarly, dietary inclusion of monensin, another antibiotic that targets Gram-positive bacteria, also did not alter the nature of AMR E. coli isolated from beef cattle . These results suggest that antimicrobial suppression of Gram-positive bacteria does not give rise to unoccupied microbial niches that are filled via AMR E. coli.
Despite the fact that the E. coli characterized in this study were recovered from selective media, the fact that antibiotic resistance, particularly to tetracycline, streptomycin, sulfamethoxazole and ampicllin, was common in E. coli from cattle that were not administered tetracycline suggests that naturally occurring resistance determinants circulate in bovine gut microbial populations for reasons other than selection as a result of antimicrobial agents being included in the diet. Hoyle et al.  characterized bovine fecal E. coli from an organic farm and found that even with the restricted use of antimicrobials, ampicillin-resistant E. coli were readily isolated. In that study, age of the cattle and likely the diet they were provided, as opposed to subtherapeutic administration of antibiotics appeared to be an important factor for the acquisition and development of antibiotic-resistant commensal microflora. A higher prevalence of AMR E. coli in feces from younger than older animals within the same farm has been previously reported [37, 38]. A comprehensive longitudinal study of four feedlots in which antibiotics were only used therapeutically also found no difference in the nature of AMR among isolates collected from home pens compared with those from hospital pens in which antibiotics were administered . Our work as well as that of others has also observed that the presence and dissemination of AMR in E. coli during the feeding period may be a response to the diet rather than antimicrobial administration [12, 18, 40]. In the present study, short-term withdrawal of antibiotics appeared to have minimal impact on AMR in E. coli, given that AMR isolates were collected routinely on days C and E. Perhaps this is not surprising when one considers that even long term withdrawal of antimicrobials has in some cases had minimal impact on the nature of antimicrobial resistance . In the case of genetic determinants for tetracycline resistance, it has been proposed that these elements have established a steady state in E. coli populations, and that their presence is not necessarily related to antimicrobial usage .
Perhaps the most obvious impact of antimicrobial administration on the phenotype and genotype of E. coli was observed for isolates obtained from TS fed cattle, a response that may reflect the fact that two antimicrobials were administered to these animals. The MT isolates from the TS group exhibited a higher frequency of SMX resistance and as both sulfamethazine and sulfamethoxazole (SMX) are sulfonamides, this may reflect selection for strains resistant to SMX. Sharma et al.  recently reported similarities in the numbers of ampicillin-resistant and tetracycline-resistant isolates, as well as the types of resistance phenotypes observed, in E. coli collected from cattle fed chlortetracycline (44 ppm) alone or in combination with sulfamethazine at the same concentration. These results suggest that the administration of chlortetracycline, even in the absence of sulfamethazine, can lead to the emergence of resistance to SMX, as well as other antibiotics, including AMP and CHL.
E. coli exhibiting STRSMXTE and SMXTE resistance phenotypes have been frequently isolated from cattle . Enne et al.  documented that the prevalence of sulfonamide resistance among E. coli remained constant even with a 97% reduction in the clinical use of sulfonamides in the UK. Further work showed that a plasmid carrying the resistance determinants sul2, strA and strB enhanced host fitness even in the absence of antibiotic selective pressure . Linkages between CHL and TE phenotypes, sulphonamide resistance, and other resistance determinants have been described in plasmid profiling of human clinical isolates in Australia , but at this point it remains to be determined if similar linkages are responsible for the linked dissemination of these resistances in feedlot cattle. It is also possible that genes that confer fitness to environmental challenges (e.g., acid tolerance, nutrient limitations, metal concentration) other than those imposed by antibiotics are harboured on these plasmids and promote the acquisition of resistance determinants .
Detection of specific AMR E. coli frequently appeared to be transient over the duration of this study. Only in one steer (ID 99; group TS) was the same AMPCHLSMXTE E. coli clone obtained on all 4 sampling days. Others have also reported that the majority of E. coli O157:H7 subtypes occur intermittently within cattle and that few isolates persist for extended periods of time . Although isolates occurred transiently, there were instances where a particular isolate clearly occurred more frequently during specific phases of the feeding period. For example, E. coli isolates exhibiting STRTE phenotype were recovered almost exclusively on days D and E, particularly from CON, TS and V steers, and the majority of isolates were clones. This suggests that this particular isolate disseminated readily among pen mates within the feedlot or that this particular clone may have possessed fitness attributes that promoted its prevalence at these points during the feeding period.
In some instances, the occurrence of clones was clearly pen-associated. Some MT-isolated E. coli clones with specific PFGE profiles occurred exclusively or nearly exclusively within a single pen (e.g., STRSMXTE with PFGE type X in pen V-1). This same phenomenon was also observed for E. coli isolates with ampicillin resistance, i.e., cultured on MA (e.g., AMP with PFGE type F, pen V-5). The association of isolates with specific pens was not solely related to the administration of antibiotics, given that some pen associations were evident in the CON group as well (AMPSTRTE with PFGE type YY in pen CON-3; STRSMXTE with PFGE type W in pen CON-4). These findings suggest that the degree of transference of AMR E. coli in the feedlot depends on the subtype in question. A previous study in or laboratory used genotyping to document movement of E. coli strains from animal to animal within the feedlot environment . Others have reported that housing location can influence the nature of antimicrobial resistance in fecal coliforms from swine , but in that study, the pigs were housed in different barns as opposed to different pens within a common building. We determined previously that a rifampin-resistant strain of E. coli was transferred infrequently among feedlot cattle housed in adjacent pens even when it was inoculated (1010 CFU) into Trojan steers . In the present study, there was possible evidence of transmission of ampicllin-resistant E. coli among adjacent pens as identical AMPTE subtypes were recovered from TS steers in pens 3, 4, and 5 sampled on day E. Similarly, identical AMPSTRTE subtypes were obtained from V steers in adjacent pens 1 and 2 during this same sampling period. Our results suggest that the pen boundaries act as a significant impediment to the widespread dissemination of some AMR E. coli subtypes within the feedlot. At this point it is not known if a similar phenomenon would be observed in all feedlots as our feedlot only represented a single ecological unit.
Resource constraints limited our characterizations to only single isolate from each selective plate from each steer during later samplings. It further restricted our ability to study isolates from all steers on all treatments It is possible that this approach may not have given a complete picture of the genetic diversity of tetracycline- and ampicillin-resistant E. coli present in feedlot steers. Ensuring representative sampling is always a challenge considering the voluminous nature of digesta within the bovine intestinal tract and the number of cattle that are typically housed within a feedlot. Others have reported that examining single vs multiple isolates did not compromise interpretation of the temporal trends or the nature of diversity of E. coli within cohorts [50, 51]. In early samples, where we did select two isolates, PFGE frequently identified both isolates as clones. That finding is perhaps not surprising, given the frequency with which we isolated clones from individual pen mates. This pattern may have been amplified by the use of selective plates for establishing the isolate collections, a practice that obviously selects for less diverse subpopulations. In the present study, the degree of diversity was clearly related to the nature of the resistant phenotype. Some phenotypes such as TE, SMXTE and STRSMXTE exhibited a high degree of diversity whereas others, such as AMPCHLSMXTE were solely of a clonal nature suggesting the resistance genes may be chromosomally encoded while others may be plasmid mediated both of which could contribute to the varying degrees of diversity among isolates examined.
Screening for resistance determinants showed that the majority of tetracycline-resistant isolates harboured the tet(B) efflux gene, followed less frequently by tet(A) and tet(C). These findings are consistent with those of Walk et al.  who reported that 64.8%, 28.1 and 4.6 of tetracycline-resistant isolates from conventional and organic dairies possessed tet(B), tet(A) and tet(C) determinants. The prevalence of tet efflux genes in E. coli is likely related to their occurrence on mobile conjugative plasmids and transposons, although tet(B) has recently been reported also to integrate into chromosomal DNA . Tet(B) has been reported in a variety of other Gram-negative bacteria, including Enterobacter, Proteus, Salmonella, Actinobacillus, Haemophilus, Morazella and Treponema spp. This distribution is thought to reflect frequent gene transfer . In the present study, isolates from MT were screened for other efflux, ribosomal protection, and tetracycline catabolism determinants that included tet(K), tet(L), tet(M), tet(O), tet(S), tetA(P), tet(Q), and tet(X). This group of tet genes are normally present on mobile conjugative plasmids or chromosomally located in Gram-positive bacteria , but there has been reports of their transfer to phylogenetically distant bacteria, as tet(K) and tet(L) have been reported in Gram-negative bacteria . Our screening failed to detect these genes, and to our knowledge, there have been no reports of these determinants occurring in E. coli.
During screening of the ampicillin-resistant isolates for three β-lactamase genes the bla
TEM1 determinant was detected in 50 to 100% of isolates from the four treatment groups. Amplicons for bla
OXA1 or bla
PSE1 were not produced in any of the remaining MA isolates. Other research teams have also failed to detect bla
SHV and bla
PSE1 in ampicillin-resistant E. coli isolates recovered from cattle [20, 22]. We are presently in the process of screening for additional β-lactamase determinants in ampicillin-resistant E. coli isolates that were not equated with bla
TEM1. A close association of bla
TEM1 with class I integrons has been reported, which likely accounts for the wide dissemination of this determinant among Gram-negative bacteria . Others in Denmark and Spain also found bla
TEM1 to be the most common determinant observed in ampicillin-resistant E. coli of animal origin, with bla
OXA1 detected only occasionally [53, 54].