Genetic diversity and antimicrobial resistance among isolates of Escherichia coli O157: H7 from feces and hides of super-shedders and low-shedding pen-mates in two commercial beef feedlots

Background Cattle shedding at least 104 CFU Escherichia coli O157:H7/g feces are described as super-shedders and have been shown to increase transmission of E. coli O157:H7 to other cattle in feedlots. This study investigated relationships among fecal isolates from super-shedders (n = 162), perineal hide swab isolates (PS) from super-shedders (n = 137) and fecal isolates from low-shedder (< 104 CFU/g feces) pen-mates (n = 496) using pulsed-field gel electrophoresis (PFGE). A subsample of these fecal isolates (n = 474) was tested for antimicrobial resistance. Isolates of E. coli O157:H7 were obtained from cattle in pens (avg. 181 head) at 2 commercial feedlots in southern Alberta with each steer sampled at entry to the feedlot and prior to slaughter. Results Only 1 steer maintained super-shedder status at both samplings, although approximately 30% of super-shedders in sampling 1 had low-shedder status at sampling 2. A total of 85 restriction endonuclease digestion clusters (REPC; 90% or greater similarity) and 86 unique isolates (< 90% similarity) were detected, with the predominant REPC (30% of isolates) being isolated from cattle in all feedlot pens, although it was not associated with shedding status (super- or low-shedder; P = 0.94). Only 2/21 super-shedders had fecal isolates in the same REPC at both samplings. Fecal and PS isolates from individual super-shedders generally belonged to different REPCs, although fecal isolates of E. coli O157:H7 from super- and low-shedders showed greater similarity (P < 0.001) than those from PS. For 77% of super-shedders, PFGE profiles of super-shedder fecal and PS isolates were distinct from all low-shedder fecal isolates collected in the same pen. A low level of antimicrobial resistance (3.7%) was detected and prevalence of antimicrobial resistance did not differ among super- and low-shedder isolates (P = 0.69), although all super-shedder isolates with antimicrobial resistance (n = 3) were resistant to multiple antimicrobials. Conclusions Super-shedders did not have increased antimicrobial resistance compared to low-shedder pen mates. Our data demonstrated that PFGE profiles of individual super-shedders varied over time and that only 1/162 steers remained a super-shedder at 2 samplings. In these two commercial feedlots, PFGE subtypes of E. coli O157:H7 from fecal isolates of super- and low-shedders were frequently different as were subtypes of fecal and perineal hide isolates from super-shedders.


Background
Cattle shedding at least 10 4 CFU/g E. coli O157 in feces were first termed "super-shedders" by Matthews et al. [1], and the role of super-shedders in contamination of the food supply has been much investigated. As duration of super-shedding is unknown, Carlson et al. [2] proposed that persistent low-level shedders of E. coli O157: H7 were a greater food safety risk than were intermittent super-shedders. In contrast, 47% [3] to > 90% [4] of the shedding of E. coli O157:H7 within feedlot pens has been attributed to super-shedders, even though these animals are thought to represent < 10% of the cattle population [4,5]. As most previous studies of supershedders have evaluated a single point in time, it is also possible that many cattle are super-shedders for only a brief period. Accordingly, Robinson et al. [6] reported that within-animal variation of shedding E. coli O157 was greater than that among animals over time.
As the majority of cattle positive for E. coli O157:H7 shed < 100 CFU/g of feces [7], a super-shedder releasing up to 10 9 CFU/g feces [3] even for short durations could represent a significant point source of environmental and possibly food contamination. In pens of 8 cattle, Cobbold et al. [8] observed that isolates of E. coli O157 were similar among super-shedders and pen mates based on pulsedfield gel electrophoresis (PFGE) analyses, while Stanford et al. [9] demonstrated that contaminating the perineum of a steer with feces containing 10 6 CFU E. coli O157:H7 resulted in 7 of 8 steers in the pen acquiring and shedding this organism. In larger groups, presence of a supershedder in a load of 20 to 50 cattle during shipment to slaughter has increased carcass contamination with E. coli O157 [10,11]. However, it is not known if a similar level of transmission from super-shedders to pen mates occurs in commercial feedlot pens where 100 to 200 individuals may be housed.
Previously we reported that super-shedders increased the incidence of perineal swab (PS) contamination with E. coli O157 in pens of commercial cattle [3]. The objectives of the present study were to determine genetic relationships among subtypes of E. coli O157:H7: 1) shed in feces by super-and low-shedders (< 10 4 CFU/g feces) in commercial feedlot pens; 2) shed in feces and detected on PS of super-shedders. Pulsed-field gel electrophoresis was used to characterize super-shedder subtypes of E. coli O157:H7 as it has been used to evaluate genetic relationships and the transmission of E. coli O157 throughout the beef supply chain [12][13][14]. As antibiotic treatment can select for resistant bacteria and mutator alleles leading to increased virulence [15] and isolates from super-shedders may also have heightened virulence, antimicrobial resistance (AMR) profiles were also compared in isolates of E. coli O157:H7 from super-and low-shedder pen mates.

Occurrence of super-shedders
Although 153 super-shedders were detected in sampling 1 (Table 1), only 1 steer was a super-shedder at both sampling dates. The number of super-and low-shedders markedly declined at the second sampling, likely due to the prevalence of E. coli O157 generally declining seasonally from summer to fall [7,16] and after a month of acclimation of cattle within a feedlot [17] However, the proportion of super-as compared to low-shedders also declined at the second sampling, a result that may reflect the transition of super-shedders to low-shedder status as approximately 30% of low-shedders in the second sampling were super-shedders in sampling 1. Of the 21 supershedders which were low-shedders in sampling 2, only 2 steers had fecal isolates of E. coli O157:H7 in the same REPC at both samplings. Transition in PFGE subtypes within super-shedders was likely due to both transmission of E. coli O157:H7 among animals and mutation events. Transmission was likely for 3 steers that had isolates within a prevalent REPC at the second sampling and isolates from 3 different REPC specific to super-shedders in sampling 1. Mutation or transfer of E. coli O157:H7 from other sources was likely for 5 steers where isolates in the second sampling were from REPC that were specific to sampling 2 and a single pen within a feedlot.
As only one steer was a super-shedder at both samplings, our results support the contention of Robinson et al. [6] that individual cattle are super-shedders for short periods and that levels of shedding vary widely among sampling time points. However, as former supershedders constituted a large proportion of the lowshedder individuals identified at the second sampling (i.e., 30%), high-levels of shedding may also be associated with greater persistence of E. coli O157 within the gastro-intestinal tract [4,8,18], although results of the current study demonstrate transition of fecal PFGE subtypes by super-shedders over time (Additional file 1, Additional file 2). With only 2 samples collected per steer at least 1 month apart, estimates of the duration of high-level shedding by super-shedders were not possible in this study, a factor that will be investigated in future studies through a more frequent sampling routine. Robinson et al. [6] used an intensive regime which involved sampling individuals every 3 h, a protocol that would be almost impossible to implement under commercial feedlot conditions.
Pulsed-field gel electrophoresismost common profile A level of 90% similarity (< 3 band differences) was chosen to define REPCs as PFGE profiles differing by one or two bands are considered to be highly related [19]. The REPC most frequently detected (REPC A) was also the only PFGE profile that was present in all pens at     (Table 2), likely representing an E. coli O157:H7 subtype that is common to feedlots within the sampling area. In sampling 1, 28% of isolates belonged to REPC A, with 34% of isolates in REPC A at sampling 2. Within pens, membership in REPC A was not homogeneous and in 2 pens this PFGE profile was confined to low-shedders, while in 4 pens this PFGE profile was found only in super-shedders. As REPC A was not related to E. coli O157:H7 shedding status of the cattle (P = 0.94; Table 3), the abundance of isolates from this REPC is likely indicative of the fitness of this E. coli O157:H7 in the gastrointestinal tract or feedlot environment. Accordingly, Carlson et al. [2] determined that persistence and dominance of E. coli O157:H7 strains isolated from feedlot cattle was influenced by genotype and possibly related to the degree of adherence to intestinal epithelial cells As REPC A was of similar prevalence in both sampling 1 and 2 (P = 0.18; Table 3), there was no evidence of a transition over time to a single dominant subtype of E. coli O157 which would account for the majority of isolates, likely due to the wide diversity of PFGE subtypes in calves entering the feedlot [14]. Considering all PFGE subtypes, genetic diversity tended to increase slightly from sampling 1 (average of 4 isolates per subtype) to sampling 2 (average of 3 isolates per subtype), although any change in diversity was mostly likely a reflection of the disparity in numbers of isolates of E. coli O157:H7 collected (709 isolates in sampling 1 as compared to 86 in sampling 2).

Comparison of super-shedder hide swab and fecal grab isolates
Due to the volume of isolates collected and the laborious nature of PFGE, only one isolate was analysed per sample type which may have under-estimated the genetic diversity of E. coli O157:H7 compared to analysis of multiple isolates [20], although analysis of a single isolate is generally indicative of the dominant strain [21,22]. Consequently, the lack of congruity between simultaneously-collected fecal and PS isolates from individual super-shedders was surprising, as only 2 super-shedder steers had PS and fecal isolates in the same REPC (Table 1).
Most previous studies have used PFGE to characterize either hide or fecal isolates of E. coli O157:H7 [14,[22][23][24]. Avery et al. [12] found some commonality of PFGE subtypes from E. coli O157:H7 isolated from the hide and feces of the same animal, but only sampled 5 cattle. Arthur et al. [25] found similarities as 9 hide and 22 fecal PFGE profiles for E. coli O157:H7 from a single feedlot belonged to only 2 REPC. In contrast, from 795 E. coli O157:H7 isolates in the present study, a total of 85 REPCs and 86 unique isolates were detected (Table 1). Genetic diversity of E. coli O157:H7 varies widely by location, ranging from 100% of isolates in a single REPC [26], to the heightened diversity noted in our study and that of Sargeant et al. [14]. Similar to the present study, Childs et al. [27] found only occasional commonality among PFGE profiles of E. coli O157:H7 isolates onfarm collected from hides and those collected from the colon post-harvest.
Overall genetic diversity of super-shedder PS isolates across pens was high and PS isolates as a group were not more similar than E. coli O157:H7 isolates overall (P =0.77), although within the majority of pens, PS isolates had more similar PFGE profiles than all isolates from that pen (P < 0.01; Table 4). In contrast, all superand low-shedder fecal isolates across pens had more similar PFGE profiles compared to all isolates (P < 0.001), demonstrating a greater degree of similarity in fecal as compared to PS isolates. Genetic variation among fecal grab and PS isolates collected from the same animal may reflect the contamination of hides by E. coli O157:H7 originating from the feedlot environment or other cattle within the pen [27]. Escherichia coli O157:H7 on the hide would also be subjected to additional stresses such as irradiation, elevated temperatures and desiccation, all of which have been shown to increase phylogenetic diversity of E. coli O157:H7 [28]. As well, if cattle are carrying multiple strains of E. coli O157:H7 within the gastro-intestinal tract certain strains might preferentially survive on hides due to difference in environmental fitness among isolates [29].
Transmission of E. coli O157:H7 from super-shedders to pen mates In all feedlot pens monitored, PFGE profiles of fecal isolates from low-shedders were more similar (P < 0.05) than those from E. coli O157:H7 isolates overall (Table 4) and in 3 of 6 pens with sufficient isolates for comparison, were more similar than those of super-shedder fecal isolates. Sharing of PFGE profiles between low-and super-shedder isolates collected in the same pen was uncommon and 77% of super-shedder isolates were distinct from low-shedders within the same pen (Table 2). For 4/  . Thirty-four of the super-shedders had fecal isolates in REPCs unique to super-shedders with the remaining super-shedder isolates belonging to REPC A which colonized cattle regardless of shedding status. Similarly, pen 2 from feedlot A had a total of 36 super shedders, but 24 super-shedder fecal isolates belonged to REPCs unique to super-shedders and the remaining 12 belonged to REPC A. Pen 1 of feedlot B was notable as 8 of 9 super-shedders detected in the second sampling were housed in this pen (Table 1), but only 4/27 fecal isolates from super-shedders belonged to a REPC which also contained low-shedders (Table 2). In contrast to previously described pens, Pen 3 in feedlot A showed closer relationships among super-shedder and low-shedder PFGE profiles, as 11/12 super-shedder fecal isolates were in REPCs which also contained 46% of lowshedder fecal isolates.
In all pens of cattle with > 5 super-shedders, REPC exclusive to super-shedders were noted. In contrast, superand low-shedders did not share PFGE subtypes of E. coli O157:H7 in all pens of cattle evaluated. As multiple super-shedders shared PFGE subtypes, it is possible that transmission of these subtypes from an initial supershedder resulted in a gradual conversion of pen mates to super-shedding status. In pens where super-and lowshedders shared PFGE subtypes, it is possible that super-shedders were transmitting these subtypes to pen-mates. Due to the limited sharing of PFGE subtypes by superand low-shedders, these results would agree with those of Dodd et al. [13] where high and low-shedding cattle in truck-loads at slaughter shared identical PFGE subtypes in feces less than 25% of the time. The high degree of similarity in PFGE subtypes of super-and lowshedder E. coli O157:H7 isolates reported by Cobbold et al. [8] is likely a reflection of less diversity of E. coli O157:H7 in pens of 8 animals compared to that noted in the commercial feedlot pens of the present study.
If high-level shedding is transient in accord with Robinson et al. [6], certain REPC might be specific to super-shedders with low-shedders in these REPC either former or future super-shedders. Conversely, as hide contamination is of crucial importance in transmission of E. coli O157:H7 among pen mates [9,30], subtypes of E. coli O157:H7 carried on hides of super-shedders might be more critical for dissemination of the organism within a feedlot pen than those within feces. Accordingly, in our previous study and source of isolates for the current study [3], presence of super-shedders within a feedlot pen increased incidence of contamination of PS with E. coli O157:H7, although incidence or level of fecal shedding within pens was not uniformly impacted. Analysis of PFGE subtypes of PS from low-shedders would have helped to clarify the role of super-shedders in transmission of E. coli O157:H7 in commercial feedlot pens, but unfortuntately was not possible in the present study due to budget restraints. Transmission of E. coli O157:H7 within commercial feedlot pens is undoubtedly complex and additional study will be required to Anti-microbial resistance and PFGE profiles of E. coli O157:H7 Anti-microbial resistance of E. coli O157:H7 isolates was low (3.7%; Table 5) and did not differ in frequency among low-and super-shedders (P =0.69; Table 6). The 3 supershedder isolates with AMR all showed multi-drug resistance in contrast to the low-shedders where resistance to tetracycline predominated (10/14 resistant isolates). Qualitatively, no relationships among AMR and PFGE profiles were evident as resistant isolates were spread across REPC and unique isolates (data not shown). Resistance in the current study was lower than that reported for E. coli O157 by Rao et al. [31] in a survey of 21 Alberta feedlots, although those authors found higher levels of AMR in newly arrived cattle during the spring as compared to pre-slaughter. In accord with Alexander et al. [32], the most common AMR in the present study was to tetracycline. Antimicrobial resistance in the current study would be consistent with that for 93 E. coli O157:H7 isolates previously collected from feedlot B [33] where only 3 isolates exhibited AMR to either tetracycline or chloramphenicol, and one isolate demonstrated multi-drug resistance to tetracycline, amoxicillin/ clavulanic acid and ampicillin. Reasons for the low level of AMR in the present study are not known as antimicrobial use at feedlots A and B was similar to that of other feedlots in Alberta.
Multi-drug resistance has been linked to the presence of plasmids carrying multiple resistance determinants [34] and the presence of these plasmids has been demonstrated to confer fitness to environmental challenges such as acid tolerance and nutrient scarcity along with resistance to antimicrobials [35]. Accordingly, enterotoxigenic strains of E. coli have an increased prevalence of plasmid-mediated genes for antimicrobial resistance [36], although a relationship between virulence and AMR in E. coli O157:H7 has not been fully

Collection of samples
Isolates were those collected in 2007 in the study of Stephens et al. [3] (excluding samples from 1 pen lost to a freezer malfunction) and were obtained from a total of 11 pens in two commercial feedlots, with an average of 181 steers per pen ( Table 1). The feedlots were both located in southern Alberta and were separated by a distance of 62 km. Steers were sampled twice: at entry to the feedlot during the months of May through October, 2007 and prior to shipment to slaughter during the months of September through December, 2007. The animal care committee at the Lethbridge Research Centre did not evaluate studies done under commercial conditions until after publication in 2009 of revised Canadian Council of Animal Care guidelines on the the care and use of farm animals in research, teaching and testing. Therefore, this study was exempt from requiring ethical approval. However, anything more than a minimally invasive study would not have been acceptable to our commercial collaborators.
Fecal grab and perineal hide swab (100 cm 2 area around the anus in the center of the perineum) samples were simultaneously obtained from each steer. Fecal samples were obtained by rectal palpation using a clean glove for each animal. Feces were placed in Whirl-Pak W bags and transported to the laboratory on ice. Perineal swab samples (PS) were obtained using a sterile Sponge-Sicle W hydrated with 25 mL of PBS with a new Sponge-Sicle W used for each animal. Each SpongeSicle W was placed in a separate Whirl-Pak W bag along with 45 mL of modified E. coli broth with 20 mg/L novobiocin (mEC-nov) and transported to the laboratory at ambient temperature. All samples were delivered to the laboratory for analysis within a period of 12 h and refrigerated at 5°C until completion of E. coli O157:H7 detection and enumeration.

Escherichia coli O157:H7 Detection
Bags of feces were manually blended prior to subsampling and 1 g of feces was added to 9 mL of mECnov and incubated at 37°C for 6 h. Perineal swab samples were incubated in the original transport media for 18 h at 37°C. After enrichment, a 1 mL aliquot of each sample type (fecal grab and PS) was subjected to immunomagnetic separation (IMS) using Dynabeads anti-O157 and a PickPen W magnetic particle separation device as per manufacturers' instructions. Fifty-μL of the bead-bacteria mixture was plated onto sorbitol MacConkey agar supplemented with 2.5 mg/L potassium tellurite and 0.05 mg/L cefixime (CT-SMAC) and plates were incubated for 16 h at 37°C. Up to 3 sorbitol negative (clear) colonies per plate were subjected to agglutination using an E. coli O157 latex kit. One isolate per sample was further subjected to multiplex PCR assays for the detection of the stx1, stx2, eaeA, and flicC (H7) genes and isolates with eaeA, flicC and either or both of stx1 and stx2 were confirmed as E. coli O157:H7 [37].

Escherichia coli O157 Enumeration
Fecal grab samples that were positive for E. coli O157 were serially diluted (1 g of feces in 9 mL of mEC-nov)