A One Health Comparative Assessment of Antimicrobial Resistance in Generic and Extended-Spectrum Cephalosporin-Resistant Escherichia coli from Beef Production, Sewage and Clinical Settings

This study aimed to compare antimicrobial resistance (AMR) in extended-spectrum cephalosporin-resistant and generic Escherichia coli from a One Health continuum of the beef production system in Alberta, Canada. A total of 705 extended-spectrum cephalosporin-resistant E. coli (ESCr) were obtained from: cattle feces (CFeces, n = 382), catch basins (CBasins, n = 137), surrounding streams (SStreams, n = 59), beef processing plants (BProcessing, n = 4), municipal sewage (MSewage; n = 98) and human clinical specimens (CHumans, n = 25). Generic isolates (663) included: CFeces (n = 142), CBasins (n = 185), SStreams (n = 81), BProcessing (n = 159) and MSewage (n = 96). All isolates were screened for antimicrobial susceptibility to 9 antimicrobials and two clavulanic acid combinations. In ESCr, oxytetracycline (87.7%), ampicillin (84.4%) and streptomycin (73.8%) resistance phenotypes were the most common, with source influencing AMR prevalence (p < 0.001). In generic E. coli, oxytetracycline (51.1%), streptomycin (22.6%), ampicillin (22.5%) and sulfisoxazole (14.3%) resistance were most common. Overall, 88.8% of ESCr, and 26.7% of generic isolates exhibited multi-drug resistance (MDR). MDR in ESCr was high from all sources: CFeces (97.1%), MSewage (96.9%), CHumans (96%), BProcessing (100%), CBasins (70.5%) and SStreams (61.4%). MDR in generic E. coli was lower with CFeces (45.1%), CBasins (34.6%), SStreams (23.5%), MSewage (13.6%) and BProcessing (10.7%). ESBL phenotypes were confirmed in 24.7% (n = 174) ESCr and 0.6% of generic E. coli. Prevalence of bla genes in ESCr were blaCTXM (30.1%), blaCTXM-1 (21.6%), blaTEM (20%), blaCTXM-9 (7.9%), blaOXA (3.0%), blaCTXM-2 (6.4%), blaSHV (1.4%) and AmpC β-lactamase blaCMY (81.3%). The lower AMR in ESCr from SStreams and BProcessing and higher AMR in CHumans and CFeces likely reflects antimicrobial use in these environments. Although MDR levels were higher in ESCr as compared to generic E. coli, AMR to the same antimicrobials ranked high in both ESCr and generic E. coli sub-populations. This suggests that both sub-populations reflect similar AMR trends and are equally useful for AMR surveillance. Considering that MDR ESCr MSewage isolates were obtained without enrichment, while those from CFeces were obtained with enrichment, MSewage may serve as a hot spot for MDR emergence and dissemination.

entire food continuum using generic E. coli and cephalosporin-resistant E. coli in a defined geographic region of high beef production intensity in Canada.

Materials and Methods
Sampling points and details of study area (Figure 1), associated sample collection and AMU records on farms are published elsewhere [33,34]. Samples were collected and transported to the lab on ice, with E. coli isolated within 24 h after arrival at the laboratory.

Materials and Methods
Sampling points and details of study area (Figure 1), associated sample collection and AMU records on farms are published elsewhere [33,34]. Samples were collected and transported to the lab on ice, with E. coli isolated within 24 h after arrival at the laboratory. Figure 1. Details of sample type and source of Escherichia coli isolates investigated for antimicrobial resistance in a One Health study of the beef production system. Sample location/sites comprised composite fecal samples collected from penned cattle in feedlots A, B, C and D as well as water samples from catch basins. Environmental samples comprised constructed wetlands and a creek (adjacent to feedlot C). Wastewater influent and effluent were collected from two water treatment plants while samples at a beef processing facility were obtained from carcasses after hide removal, after final wash, from ground beef and retail meat. Human samples were obtained from blood, urine and abdomen samples collected from hospital patients in southern Alberta.

Fecal and Water Sampling from Cattle Feces, Catch Basins and Surrounding Streams
Four feedlots in Southern Alberta designated as A, B, and C used conventional production practices including the use of antimicrobials, while lot D was a commercial feedlot that employed both conventional and "raised without antimicrobials" practices. Feedlots had a one-time capacity of 15,000-40,000 cattle and were characterized by production conditions typical for western Canadian feedlots, with open air pens with dirt floors arranged side-by-side with central feed alleys. Animals used in this experiment were cared for in accordance with the Canadian Council of Animal Care [35]. All procedures and protocols used in this study were reviewed and approved by the University of Calgary's Animal Care Committee (Protocol number AC14-0029). Fresh pen-floor fecal samples were collected every two months from April 2014 to April 2016. Twenty pens within each feedlot containing between 100 to 300 cattle were randomly selected, and the same pens were sampled throughout the two-year study. For each pen, a combined sample of 10 g of feces was obtained from Figure 1. Details of sample type and source of Escherichia coli isolates investigated for antimicrobial resistance in a One Health study of the beef production system. Sample location/sites comprised composite fecal samples collected from penned cattle in feedlots A, B, C and D as well as water samples from catch basins. Environmental samples comprised constructed wetlands and a creek (adjacent to feedlot C). Wastewater influent and effluent were collected from two water treatment plants while samples at a beef processing facility were obtained from carcasses after hide removal, after final wash, from ground beef and retail meat. Human samples were obtained from blood, urine and abdomen samples collected from hospital patients in southern Alberta.

Fecal and Water Sampling from Cattle Feces, Catch Basins and Surrounding Streams
Four feedlots in Southern Alberta designated as A, B, and C used conventional production practices including the use of antimicrobials, while lot D was a commercial feedlot that employed both conventional and "raised without antimicrobials" practices. Feedlots had a one-time capacity of 15,000-40,000 cattle and were characterized by production conditions typical for western Canadian feedlots, with open air pens with dirt floors arranged side-by-side with central feed alleys. Animals used in this experiment were cared for in accordance with the Canadian Council of Animal Care [35]. All procedures and protocols used in this study were reviewed and approved by the University of Calgary's Animal Care Committee (Protocol number AC14-0029). Fresh pen-floor fecal samples were collected every two months from April 2014 to April 2016. Twenty pens within each feedlot containing between 100 to 300 cattle were randomly selected, and the same pens were sampled throughout the two-year study. For each pen, a combined sample of 10 g of feces was obtained from 20 fecal pats to generate composite pen samples that were transported to the lab in Cary-Blair enteric transport medium (BD Canada, Inc., Mississauga, ON, Canada).
In each feedlot, runoff from the sampled pens drained into an adjacent catch basin. In feedlot C, water samples were also obtained from accumulated runoff periodically transferred into a constructed wetland and from an adjacent ephemeral creek into which the wetland drained. From mid-depth (≈0.75 m), 1 L of water was collected using a polyethylene bottle attached to a telescopic pole at four different locations per site, which were combined to generate a composite sample. For the wetland, one sample was generated by collecting and combining samples from four consistent locations throughout the study. Water samples were collected monthly from catch basins, the wetland and creeks from April to October of 2014 and 2015. Subsequently, samples from the creeks were collectively referred to as surface streams, while the wetland samples were categorized together with samples from the catch basin. Samples were not collected from these sources from November to March as they were often frozen.

Wastewater Treatment Plants (WWTP)
In tandem to farm sample collection, a composite of influent (post-grit tank) and effluent (just prior to release) sewage samples (1 L) was collected bimonthly from WWTP in Calgary and Medicine Hat from April 2014 to April 2016.

Processing Plants
Protocols for beef processing plant and retail meat sample collection followed procedures published elsewhere [36,37]. Samples were obtained from~100 cm 2 areas of hides, beef trim and conveyers; 1000 cm 2 areas of washed carcasses; and from the whole of the distal surface of each chilled side (~12 × 500 cm 2 ). Carcasses or conveyer samples were collected by randomly swabbing the surface with a 2 × 2 sterile gauze pad (Millerdale Pharmacy, Dukal Corporation, Red Deer, AB, Canada) moistened with 0·1% w/v peptone water (Becton Dickinson Co., Sparks, MD, USA). For ground beef, 200 g was obtained by aseptically removing ground beef from 4-5 kg packs of coarsely ground chub. Samples were placed in sterile stomacher bags on ice and transported in a cooler to the lab. Processing plant samples and feedlot fecal samples did not arise from the same animals, although the feedlots sampled did send cattle to the same processing plant.

Humans
Human clinical samples were randomly selected from blood, urine and abdominal samples from anonymous individual patients and did not require patient approval or a clinical use permit. The samples were collected during the same time period as those described above by the Division of Medical Microbiology, Calgary Laboratory Services (CLS) biorepository. This laboratory services about 1.5 million people in Calgary and the surrounding rural area. Clinical strains were isolated and confirmed from samples as described by Pitout et al [38,39].

Isolation of E. coli
Processing of samples from cattle feces, catch basin, surrounding streams and sewage treatment followed procedures detailed by Adator et al. [40] and Tymensen et al. [41] to obtain generic E. coli and ESC r . To isolate ESC r from feedlot samples, 0.5 g of cattle feces was added to 4.5 mL E. coli broth-cefotaxime (2 µg/mL) and enriched overnight, followed by sub-culture onto MacConkey plates supplemented with 1 µg/mL of ceftriaxone (MilliporeSigma) [40]. Three distinct red/magenta lactose-fermenting colonies were subcultured onto tryptic soy agar (TSA)-ampicillin (32 µg/mL) (MilliporeSigma) and later archived at −80 • C. For catch basins and surrounding streams, ESC r were isolated by membrane filtration using the US EPA Method 1603 [34,40]. To obtain generic E. coli, the same procedure, excluding enrichment and selective plates, was used. For samples from the beef processing plant (BProcessing), swabs were mixed with 10 mL of buffered peptone water (BPW), stomached for 2 min and incubated overnight at 37 • C. For ground beef,~25 g of ground beef was incubated overnight at 37 • C in 225 mL of BPW. One mL from each resulting culture was added to 9 mL of E. coli enrichment broth containing a Durham tube and 2 µg/mL of cefotaxime, followed by overnight incubation at 37 • C and plating onto MacConkey agar containing 1 µg/mL of ceftriaxone. The plates were then incubated overnight at 37 • C. Samples from catch basin (CBasins), surface streams (SStreams), sewage treatment (MSewage) and humans (CHumans) were not enriched overnight prior to plating onto ceftriaxone-supplemented MacConkey plates.

Antibiograms
A comprehensive panel of antimicrobials was selected on the basis of: (i) use in beef production systems; (ii) category of medical importance to humans; (iii) adequate representation of antimicrobials in diverse classes; and (iv) use in phenotypic confirmation of extended-spectrum β-lactam -resistant E. coli. Resistance of ESC r and generic E. coli to an antibiotic panel (ampicillin, amoxicillin/clavulanic, ceftiofur, ceftazidime, ceftazidime/clavulanic acid, streptomycin, neomycin, oxytetracycline, florfenicol, trimethoprim/sulfamethoxazole and sulfisoxazole) was examined using the Kirby-Bauer disk diffusion susceptibility method according to documents M100-S26 and VET-01S of the Clinical and Laboratory Standards Institute (CLSI) [42,43]. Briefly, E. coli isolates were retrieved from −80 • C and streaked onto TSA supplemented with 5% sheep blood (PB75, Dalynn Biologicals, Calgary, AB) and incubated overnight at 37 • C. A single colony from each plate was subcultured onto PB75, incubated overnight at 37 • C and then used for ASTs. Isolates were resuspended in saline and density was calibrated to a 0.5 McFarland turbidity standard. The suspension was streaked onto Mueller Hinton Kirby agar plates (PM90K) (Dalynn Biologicals). BD BBL™ Sensi-Disc™ (BD) (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) or Thermo Scientific Oxoid (Dardilly, France) antimicrobial susceptibility disks were then placed onto the plates. The plates were incubated for 18 h and zone diameters of inhibition were recorded using BIOMIC V3 Microbiology System (Santa Barbara, CA) using standards set by the CLSI [42,43]. E. coli American Type Culture Collection (ATCC) 25,922, ATCC 35,218 and Staphylococcus aureus ATCC 29,213 were used as quality controls [42,44]. Putative extended-spectrum β-lactamase-resistant phenotypes were assigned if isolates showed an increased zone of inhibition of ≥5 mm for ceftazidime in combination with clavulanic acid versus the zone of inhibition obtained when tested against ceftazidime alone [42,45].

Genotypic Characterization of ESBL Resistance
A simplex and two multiplex PCR assays were used to screen 705 ESC r using the primer sets and conditions previously validated elsewhere, with minor modifications. Multiplex 1 comprised bla TEM [46], bla SHV [46], bla OXA [47], and bla CMY [48], multiplex 2 comprised bla CTX-M 1,2,9 [47], while the simplex identified bla CTX-M [46]. DNA template was prepared by heat lysing each E. coli colony in TE (pH 7.4) at 99 • C for 10 min and the lysate was centrifuged 21,130× g for 5 min (5424 R Eppendorf Centrifuge). Each 25 µL multiplex PCR reaction comprised 12.5 µL of 1× Qiagen HotStar Plus Multiplex PCR Master Mix (Qiagen GmbH, Hilden, Germany), 2.5 µL CoralLoad Concentrate, primer-specific concentration and 2 µL DNA template. Simplex reactions were comprised of 12.5 µL of Qiagen PCR Master Mix, 2.5 µL coral-load solution, primer-specific concentration and 2 µL DNA template. Amplification conditions for multiplex 1 were as follows: 95 • C for 15 min; 30 cycles of 94 • C for 1 min, annealing at 53 • C for 1 min, 72 • C for 1 min with a final extension for 10 min at 72 • C. For multiplex 2 and simplex reactions, annealing temperature was 60 • C. PCR products were subjected to electrophoresis in 1.5% agarose gel in 1X TAE buffer. E. coli strains previously sequenced with known β-lactamase genes of interest were used as positive controls and water was used as a no template control.

Data Management and Analysis
Descriptive analysis comparing the distributions of resistance (R), intermediate (I) and susceptible (S) E. coli, MDR and ESBL-resistant phenotypes as well as β-lactamase genotypes was performed with PROC FREQ (SAS software, version 9.4 SAS Institute, Cary, North Carolina, USA). Isolates were categorized as MDR if they showed resistance to antimicrobial agents in ≥2 different antimicrobial classes [49,50]. To compare the effect of isolate source on AMR in both ESC r and generic E. coli sub-populations, univariate multinomial logistic regression models were fitted to the data using the SAS GLIMMIX procedure [13,51]. E. coli population (ESC r or generic E. coli) was modeled as a random block factor, while antibiotic isolate source and their interaction were considered fixed effects. Differences in AMR for each antibiotic per source were expressed as odds ratios (OR); source-specific ORs were specified with a 95% confidence interval and declared significant at p < 0.05. Antimicrobials which were not modeled due to low prevalence of resistance or inability for the model to converge included neomycin in ESC r and ceftiofur, ceftazidime, trimethoprim-sulfamethoxazole and neomycin in generic E. coli. Since not all bla genes are indicative of ESBLs, isolates were designated as cephalosporin-resistant, β-lactamase genotype-positive if they possessed one or more of bla SHV , bla TEM , bla OXA and bla CTXM and designated as a pAmpC genotype-positive if they possessed bla CMY . To examine the associations between the number or type of genes present and MDR phenotype per source, separate binary logit analyses were performed using logistic regression [52] and effects were considered significant at p < 0.05 [53]. For source effect analyses of AMR phenotypes and β-lactamase/pAmpC genes in ESC r , CHumans and MSewage ESC r isolates were categorized together as human-associated E. coli designated HM (total n = 123). ESC r BProcessing isolates were not included in the analyses due to the low number of isolates obtained.
Generally, the source of isolate was found to influence MDR phenotypes in the ESC r population (p < 0.004), although the exact or relative number of genes did not influence MDR. Additionally, when the effect of the presence or absence of a single gene was examined, statistical analyses revealed that MDR prevalence was not affected by the presence of just a single gene. Examination of the effect of specific bla/pAmpC gene types also revealed that the presence of the genes (SHV, TEM, OXA, CTX-M, CTXM 1, CTX-M 2, CTX-M 9 and CMY) did not influence the prevalence of MDR.

Overall and Longitudinal Antimicrobial Resistance
In this investigation, enrichment procedures were used to increase the likelihood of isolating extended-spectrum cephalosporin-resistant E. coli (ESC r ) within the beef production system. This study is the first to comprehensively assess AMR ESC r and generic E. coli in beef production systems from the perspective of a One Health continuum within the same geographical region.
In the present study, oxytetracycline, ampicillin and streptomycin resistance were the highest in both ESC r and generic E. coli. The generally high levels of AMR in ESC r as opposed to generic E. coli is a potential reflection of the enrichment procedure selecting for ESC r . Genes such as bla SHV , bla TEM and bla CTX-M are often borne on mobile genetic elements that code for resistance to multiple antimicrobials [54][55][56]. Although enrichment for ESC r was expected to only isolate β-lactamase resistant E. coli, 4.1% of isolates were susceptible to ampicillin. Factors such as silencing of resistance genes in E. coli [57] and the loss of plasmids in the absence of selective pressure may explain this finding [58][59][60].
Significant differences in AMR profiles associated with various isolate sources coincide with findings by Ojer-Usoz et al. [63] where they investigated ESBL E. coli from food, wastewater, rivers, farms and humans and found higher cefepime (a 4th GC) resistance in food and human isolates (94.2% and 94.6%, respectively), compared to farm isolates (2.0%). In our study, prevalence of AMR was higher in CFeces compared to CBasins, SStreams, BProcessing and HM (MSewage and CHumans combined), with the exception of ampicillin, ceftiofur and trimethoprim/sulfonamide. In North America and the United Kingdom, tetracyclines, penicillins and aminoglycosides tend to be among the major classes of antimicrobials used in beef production [7,64]. Widespread use of tetracyclines in livestock increases tetracycline-resistant E. coli in feces [62,65]. In this study, the conventional feedlots used tetracycline so it is not surprising that tetracycline resistance was the highest in E. coli isolated at the point source of utilization and declined with movement from CFeces to CBasins and surrounding water ways.
Runoff water from pens collects in CBasins, and from there it may enter the surrounding environments, resulting in the prevalence of resistance to oxytetracycline, ampicillin, and streptomycin and neomycin being similar in E. coli isolated from these environments. Considering that AMR prevalence of CFeces > CBasins > SStreams isolates, CFeces may serve as a point source from which antibiotic residues as well as AMR bacteria and genes may be disseminated. The decreasing AMR proportions in CBasins and SStreams could reflect decreasing selective pressure as a result of lower antibiotic residue concentrations in these environments as compared to CFeces [66,67]. According to Sayah et al. [13], similarities of AMR patterns in different farm sources suggest a common source of resistant bacteria, an assertion that agrees with our findings. Likewise, Ibekwe et al. [68] found that elevated levels of AMR E. coli in surface water coincided with nonpoint sources of fecal contamination from both agricultural and human sources. ESC r from MSewage and CHumans differed in the top-three resistant phenotypes (ampicillin > ceftiofur > oxytetracycline) as compared to ESC r (oxytetracycline > ampicillin > streptomycin) isolated from cattle. Symptomatic correlations between AMR and selective antibiotic pressures are expected when both antimicrobials and bacterial isolates are derived from a common source, as is the case when human feces enter surface waters in the form of sewage effluent [69]. The similarity in AMR phenotypes in MSewage and CHumans E. coli isolates is consistent with this observation.
Generic E. coli isolated from CFeces, CBasins, SStreams and MSewage were most often resistant to oxytetracycline, streptomycin, sulfisoxazole and ampicillin. Antibiograms of CFeces isolates, which aligned with those from CBasins suggests that catch basins restricted the flow of AMR E. coli into surrounding surface waters. Generic BProcessing isolates were resistant to ampicillin, oxytetracycline and amoxicillin-clavulanic acid, likely reflecting the frequent use of penicillins and tetracyclines in beef production [13].

Oxytetracycline
High prevalence of resistance to oxytetracycline in E. coli has been reported in several other studies related to beef cattle and the environment [13,62,66,68]. This finding is not surprising as tetracycline is a first-line antibiotic used in beef cattle to control liver abscesses and bovine respiratory disease [7,13]. Interestingly, oxytetracycline resistance in generic MSewage isolates in this study was less than in a study by Ibekwe et al. [68], where the highest tetracycline resistance was found in isolates from WWTPs (~27%), followed by urban and agricultural runoff.  2012) reported 100% ampicillin and 64.7% amoxicillin resistance in 3GC r E. coli from CFeces, hides, pre-evisceration and final carcasses. In our study, 96.3% and 71.7% of ESC r isolates from CFeces were resistant to ampicillin and amoxicillin/clavulanic acid, respectively, while all human isolates in our study were resistant to ampicillin. A USDA report examining E. coli isolates from 1950-2002 consistently found more resistance to antimicrobials such as penicillin, which have been used for decades in clinical and veterinary medicine [70]. In another report, ≤2% of 600 isolates from agricultural sources and urban runoff were resistant to amoxicillin-clavulanic [68]. Consequently, it can be speculated that there are other prevailing conditions such as antibiotic residues at the point of isolation which may impact ampicillin and amoxicillin/clavulanic acid resistance in specific populations.

Penicillins
Resistance to ampicillin and amoxicillin/clavulanic acid was consistently higher in both ESC r and generic BProcessing isolates. As such, it makes sense to surmise that irrespective of sub-population, β-lactamase-resistant E. coli could potentially contaminate beef products.

Aminoglycoside
Streptomycin resistance was found to range between 44% (CHumans) and 89.1% (CFeces) in ESC r . Coincidentally, high streptomycin (84.2%) and low kanamycin resistance (5.3%) were reported in generic E. coli from surface water. [66]. It is interesting to note that our isolates showed neomycin resistance ranging between 0% and 5.1% for ESC r , while very few generic E. coli were neomycin resistant. In generic isolates, streptomycin resistance ranged from 5.7% in BProcessing isolates to 41.5% in CFeces. These results are consistent with the observation of very low neomycin and higher streptomycin resistance in E. coli from feedlot cattle [62]. In the current investigation, CHumans ESC r exhibited low prevalence of streptomycin and neomycin resistance, a finding consistent with those of others [71]. E. coli from CFeces and CBasins in both sub-populations showed high resistance to streptomycin even though it was not administered to cattle, suggesting possible co-selection for this resistance when other antimicrobials such as tetracyclines are administered [70,72,73].

Trimethoprim/Sulfonamide
Trimethoprim-sulfamethoxazole is a first-line therapy for acute urinary tract infection (UTI), one of the most common infections in outpatients [74,75]. We found sulfamethoxazole-trimethoprim resistance to be rare in both generic and ESC r , as have others [13]. Sulfamethoxazole-trimethoprim resistances between 0.0% and 2.2% and sulfisoxazole between 0.0% and 13.98% have been reported in generic E. coli from livestock, companion animals, farm environments, surface water and human septage [13], compared to 1.9% (BProcessing) to 6.3% (CFeces) in generic E. coli in our study. Since this antibiotic is mainly used in human medicine, it is unsurprising that clinical human isolates demonstrated a higher prevalence of resistance than those from the beef production system.

Phenicol
It has been reported that florfenicol use in livestock co-selects for chloramphenicol resistance [76][77][78]. Florfenicol is approved for use in treating BRD. Donaldson et al. [76] reported 93% florfenicol and 94% chloramphenicol resistance in 3GC r E. coli isolated from dairy cattle, while 92% of E. coli isolated at necropsy or from fecal samples of calves with diarrhea were florfenicol resistant [79]. In the present study, florfenicol resistance in ESC r was 77.4% in CFeces, while it was low ≤ 45% in isolates from other environments. Similarly, Escherichia/Shigella, Pseudomonas, Sphingobactaria, and Brevundimonas obtained from a university hospital outlet and municipal sewage inlet on sulfamethoxazole/trimethoprim and streptomycin-selective plates exhibited florfenicol resistance in ≤45% of the isolates [80]. Results from our study suggest that florfenicol resistance in E. coli was generally infrequent in human-associated sources, likely due to limited use of phenicols in humans.

Third-Generation Cephalosporins and ESBL Phenotype Distribution
In cattle feedlots, commensal 3GC r E. coli represent a sub-population of generic E. coli associated with hide and fecal populations [81,82]. We observed that ceftiofur resistance (93.9%) in MSewage ESC r isolates was substantially higher than in all other sources except for CHumans isolates (92%). Ceftiofur (38.9%) and ceftazidime (27.1%) resistance in SStreams ESC r isolates was comparably lower than cefotaxime-resistant (77%) ESBL E. coli isolated from rivers and lakes in Switzerland [83]. Schmidt et al. [81] and Volkova et al. [82] also reported ceftiofur and ceftriaxone resistance of 100% in 3GC r E. coli from cattle feces, hide and carcasses as compared to 66.7% of isolates from BProcessing and 65.5% from CFeces being resistant to ceftazidime in our study. It is possible that differences in the prevalence of 3GC resistance within ESC r populations are attributable to differences in isolation methods, including the specific type and concentration of antimicrobial used for enrichment. The use of enrichment broths containing cephalosporins may also enhance bacterial conjugation and exchange of resistance plasmids between bacteria [28].
The β-lactam antimicrobials including cephalosporins are routinely used for treating bacterial infections in humans so that hospitals are a significant source of cephalosporin antimicrobials in wastewater [84]. Korzeniewska et al. [84] found that the majority of generic E. coli in three different hospital sewage plants were resistant to cefotaxime and ceftazidime, a result similar to ours with MSewage ESC r isolates, although this contrasted the low resistance in our generic E. coli isolates. ESBL E. coli were readily obtained from MSewage even without enrichment with ceftriaxone. Discharge of untreated sewage containing antimicrobial residues and resistant bacteria from humans and animals may contribute to the emergence and spread of AMR bacteria in aquatic environments [80]. It is possible that antimicrobial residue levels in MSewage exert sufficient selective pressure for the development of MDR phenotypes. It was not surprising that clinical isolates obtained without enrichment showed the highest ESBL phenotype, as β-lactams are widely used in Canadian clinics.
Phenotypic ESBL isolates were not detected in BProcessing samples, an observation that agrees with others that did not detect ESBL E. coli in raw milk or minced beef, despite enrichment [85]. Furthermore, these observations suggest that there is a decrease in ESBL bacteria after defecation as one moves down the production chain [86]. The overall low levels of phenotypic ESBL E. coli reported suggests that without enrichment, negligible levels would have been detected within the beef production environment.

Prevalence of Multidrug Resistance
Of the 88.8% MDR ESC r in the current study, most (45.1%) were resistant to antimicrobials in six different drug classes. The most common MDR patterns in ESC r involved combinations of oxytetracycline, ampicillin, streptomycin, ceftiofur, florfenicol and sulfisoxazole, whereas oxytetracycline, streptomycin and sulfisoxazole were common in generic E. coli. Others have also found high levels of oxytetracycline, ampicillin and streptomycin resistance in generic E. coli isolates from feedlot cattle, farm environments, human septage and surface water [13,62]. In concordance with results from Ojer-Usoz et al. [63], MDR ESC r exceeded 60% in all sources. Ojer-Usoz et al. [63] found levels of MDR in ESBL E. coli ranging between 58% and 86.2% in food, WWTP, rivers, farms and humans. Numerous investigators note that the administration and presence of even a single antimicrobial can select for MDR strains in both humans and animals [87][88][89]. Consequently, it is likely that the high MDRs reflect long-term exposure of bacteria to specific antimicrobials [87]. Long-term tetracycline use confers resistance to other antimicrobial agents via co-selection, as tet genes and other resistance genes often share common integrons, plasmids or transposons [65,90].
The SHV-and TEM-type genes have also been detected in livestock and meat products [63,103], while OXA has been rarely reported in livestock. Brinas et al. [104] detected SHV-or OXA-type β-lactamases in only 3% of 124 ampicillin-resistant E. coli recovered from food and feces of healthy animals and humans. This is only slightly higher than the level of SHV (1.4%) we observed in ESC r . Interestingly, the OXA-type genes in our study were unique to MSewage and clinical isolates, while SHV-type were specific to CFeces and MSewage. Agga et al. [11] also reported a higher diversity of resistance genes in municipal wastewater than in livestock environments. In wastewater, the commonality of homologous regions within complex DNA elements enhances recombination [105,106], a response that may account for the high diversity of bla genes in MSewage. The complete absence of bla OXA in our isolates from the beef production environment agrees with Ojer-Usoz et al. [63], who found bla OXA-1 (20%) only in clinical E. coli isolates and not in isolates from food, farm or aquatic environments. OXA-type β-lactamases have often been reported in clinical Enterobactreacea around the world as well as from WWTP, with some variants implicated in hospital outbreaks [107][108][109].

Occurrence of AmpC Gene
Few investigations have assessed the frequency of bla CMY genes in E. coli isolated from food and environmental matrices as compared to isolates obtained from clinical settings. Of ESC r from cattle sources, we found bla CMY ranging from 87.2% to 93.2%, in contrast to the lower prevalence of 32% to 50% in human sources. In Canada, bla CMY has mostly been detected in clinical or hospital-associated isolates and infrequently in food-associated Enterobacteriaceae, while others have suggested that food sources may be an emerging concern [110][111][112].

Conclusions
Overall, AMR profiles of E. coli from the various sources reflected corresponding antimicrobial use in those segments of the continuum. A point source effect on AMR occurrence, where sources with antibiotic use reflected high AMR E. coli was further underscored by the similarity of AMR patterns between CFeces and CBasins. As such, continuing to challenge both the human medical and veterinary communities to monitor use protocols and improve antimicrobial stewardship practices in line with community, national and global AMR reduction goals. Continuous monitoring of critically important antimicrobials such as neomycin should be considered as a means of detecting early changes in resistance trends and onset of AMR emergence. E. coli isolation with enrichment enhanced the sensitivity of detecting ESBL-producing bacteria. ESBL phenotypes within ESC r were more frequently associated with human than cattle sources. In generic E. coli, MDR was lowest in BProcessing isolates, suggesting that strategies employed during beef processing reduce the risk of MDR isolates in final meat products.  The funding agencies had no role in study design, sample collection, analysis, interpretation, and in writing of the manuscript.
Conflicts of Interest: C.W.B. would like to declare that he is part owner and managing partner of Feedlot Health Management Services and Southern Alberta Veterinary Services. S.J.H. is an employee at Feedlot Health Management Services, Okotoks, Alberta, Canada. Feedlot Health is a private company that provides expert consultation regarding production and management of feedlot cattle and calf grower calves, including developing veterinary protocols to support animal health. They also conduct in-house and contract research related to dairy calf grower and feedlot production. All authors declare that the research was conducted without conflict of interest.