A Longitudinal Study of Antimicrobial Resistance in Enterococcus spp. Isolated from a Beef Processing Plant and Retail Ground Beef

Antimicrobial use in food-producing animals has come under increasing scrutiny due to its potential association with antimicrobial resistance (AMR). Monitoring of AMR in indicator microorganisms such as Enterococcus spp. in meat production facilities and retail meat products can provide important information on the dynamics and prevalence of AMR in these environments. In this study, swabs or samples were obtained from various locations in a commercial beef packing operation (n = 600 total) and from retail ground beef (n = 60) over a 19-month period. All samples/swabs were enriched for Enterococcus spp. and suspected enterococci isolates were identified using species-specific PCR primers. Enterococcus faecalis was the most frequently isolated species followed by Enterococcus hirae, which was found mostly on hides and ground beef. Enterococcus faecium (n = 9) and E. faecalis (n = 120) isolates were further characterized for antimicrobial resistance and resistant genes due to the clinical significance of these species. Twenty-one unique AMR profiles were identified, with 90% of isolates resistant to at least two antimicrobials, and two that were resistant to nine antimicrobials. Tetracycline resistance was observed most often in E. faecalis (28.8%) and was likely mediated by tet(M). Genomic analysis of selected E. faecalis and E. faecium isolates revealed that many of the isolates in this study clustered with other publicly available genomes from ground beef, suggesting that these strains are well adapted to the beef packaging environment. IMPORTANCE Antimicrobial resistance (AMR) is a serious challenge facing the agricultural industry. Understanding the flow of antimicrobial resistant-bacteria through the beef fabrication process and into ground beef is an important step in identifying intervention points for reducing AMR. In this study we used enterococci as indicator bacteria for monitoring AMR in a commercial beef packaging facility and in retail ground beef over a 19-month period. Although washing of carcasses post-hide removal reduced the isolation frequency of Enterococcus spp., a number of antimicrobial resistant-Enterococcus faecalis isolates were recovered from ground beef produced in the packaging plant. Genome analysis showed that several E. faecalis isolates were genetically similar to publicly available isolates recovered from retail ground beef in the United States.


.D). This couplet of 168
ARGs is present in many E. faecium and E. faecalis strains in NCBI, but can also be found in 169 Staphylococcus spp., Clostridium spp., and Campylobacter coli strains. E. faecium H112E 170 contained a gene region harboring the oxazolidinone resistance gene optrA in close proximity to 171 the macrolide resistance gene erm(A), ant(9)-Ia (aminoglycoside resistance), and xerC, a 172 tyrosine recombinase gene (Fig. 1E). This gene region aligned with complete coverage and 173 greater than 99% identity to both a plasmid in E. faecalis (GenBank CP042214.1) and an optrA 174 gene cluster in E. faecium (GenBank MK251151.1) suggesting that this gene array could have 175 originally been a plasmid that integrated into the chromosome of E. faecium H112E. Other 176 ARGs present that either assembled into single gene contigs or gene regions lacking other ARGs 177 were the lincosamide resistance gene lunG in E. faecalis H96E, the chloramphenicol resistance 178 gene catA, and msrC in E. faecium H134E and H112E. 179 Virulence genes 180 Genome assemblies were also screened for virulence genes using the VirulenceFinder 181 Enterococcus database. The virulence genes ace (collagen adhesin), camE, cCF10, cOB1 (sex 9 pheromones), ebpA, ebpB, ebpC (pili proteins), efaAfs (adhesion), elrA (enterococcal leucine rich 183 protein A), srtA (sortase), tpx (thiol peroxidase) were found in all E. faecalis genomes (Table  184 S4). The gelatinase-encoding gelE and hyaluronidase genes hylA and hylB were also detected in 185 74.5%, 68.8%, and 83.0% of E. faecalis genomes, respectively. Only two E. faecalis genomes 186 carried the cytolysin genes cylABLM but notably these were also the strains that had the greatest 187 number of ARGs, H11 and H22. The efaAfm gene, which encodes a cell wall adhesin, was found 188 in all eight E. faecium assemblies. The acm gene (collagen-binding protein) was the only other 189 virulence gene detected in the E. faecium genomes (75%). 190

Phylogeny of enterococcal strains 191
Phylogenetic relationships among the 47 E. faecalis and 8 E. faecium strains and several 192 publicly available E. faecalis and E. faecium genomes were determined using the core genes 193 within each species. These additional E. faecalis and E. faecium strains included all publicly 194 available isolates from ground beef and several randomly selected human and cattle fecal isolates 195 also from Alberta (22). The core genome of E. faecalis contained 1,325 genes and the pan-196 genome 9,558. Among the 27 E. faecium genomes included for analysis, there were 1,417 genes 197 in the core genome and 7,848 in the pan-genome. 198 E. faecalis strains clustered by MLST type (Fig. 2). Interestingly, certain E. faecalis 199 strains that had been collected from retail ground beef in the United States had a MLST profile 200 (ST192, ST228, and ST260) that was shared with strains isolated from the conveyor belt, 201 carcasses after final washing, and retail ground beef in the present study. Six of the E. faecalis 202 isolates (G92, G127E, G149, H4, W97, and W133) had the same MLST profile as one of the 203 Alberta human isolates (HC_NS0077; leg wound). However, it should be noted that this human 204 isolate carried tet(M) and an additional virulence gene which was absent from the six isolates. 205 10 E. faecium isolates also clustered by MLST (Fig. 3). Three E. faecium isolates from retail 206 ground beef along with two isolates from the post-wash carcasses and one from US ground beef 207 had the same MLST (ST76). Unlike the E. faecalis genomes, there also appeared to be two 208 distinct clades of E. faecium with the two hide isolates (H134E and H112E) in a separate clade 209 from the other E. faecium isolates examined. 210

Discussion 211
Antimicrobial resistance continues to be a serious public health threat and there are 212 concerns that antimicrobial-resistant bacteria in food-producing animals may be transferred to 213 humans through the food production system. In this study we used culturing and whole genome 214 sequencing to monitor AMR and enterococci distribution in beef production from slaughter 215 through to the retail sector over a nineteen-month period. Although 10 different Enterococcus 216 spp. were isolated at least once during the study, only E. faecalis was found in all sampling 217 locations. This is consistent with previous surveys that sampled from beef plants (4) or retail 218 ground beef (5). E. hirae was isolated most frequently from post-hide removal swabs, which was 219 expected given that E. hirae has been reported to be the most prevalent Enterococcus spp. in 220 cattle feces (2,22,24) and there is greater likelihood of contamination from feces at the hide 221 removal step (25). 222 The number of enterococci-positive samples recovered from the carcass post-washing 223 and the conveyor belt area was substantially lower than in any other sample type. Carcasses are 224 subjected to washing with hot water and spraying with organic acids after hide removal which 225 reduces the microbial load on the carcasses. The proportion of enterococci isolated from the 226 conveyor belts was lower than an earlier study at the same plant (10.7% vs. 48%) (4). This may 227 represent improvements in sanitation within the conveyor area or possibly variation in the 228 11 prevalence of enterococci. However, 82.7% of the ground beef produced within the plant was 229 positive for Enterococcus spp., most of which were E. faecalis, suggesting that the conveyor area 230 is not a reflection of the prevalence of enterococci in the ground beef produced. Enterococci 231 were also isolated from the majority of ground beef samples taken from retail stores in Alberta 232 which was similar to previous surveys of enterococci in retail ground beef in Alberta (4, 26) and 233 the United States (5). 234 We subjected 120 E. faecalis and 9 E. faecium isolates to antimicrobial susceptibility 235 testing due to their relevance to human health. Of the antimicrobials classified by the World 236 Health Organization (WHO) as critically important in human medicine (27), infrequent 237 resistance to ciprofloxacin, daptomycin, erythromycin, gentamicin, kanamycin, and tigecycline 238 was noted. None of the isolates were resistant to vancomycin or linezolid, antimicrobials often 239 used to treat VRE strains (28). Resistance to lincomycin and quinupristin-dalfopristin is intrinsic 240 in E. faecalis and mediated by the chromosomally-encoded lsa(A) gene (29), thus explaining the 241 widespread resistance of E. faecalis to these antimicrobials. Tetracycline resistance was observed 242 in 30% of E. faecalis and 33.3% of E. faecium isolates, which may have been due to the tet(M) 243 gene which was detected in 83.3% of tetracycline-resistant E. faecalis isolates and was absent in 244 tetracycline-susceptible ones. Feedlot cattle in Western Canada have historically received 245 tetracyclines such as chlortetracycline and oxytetracycline in feed or via injection for treatment 246 and prevention of disease, possibly accounting for the prevalence of tetracycline resistance noted 247 here (13). 248 Ionophores are one of the most widely used classes of antimicrobials in livestock 249 production. Because they are only employed in veterinary medicine it is assumed that their use 250 does not impact human health (30). As a potential human pathogen that inhabits the 251 12 gastrointestinal tract of food-producing animals, several studies have examined ionophore 252 resistance in Enterococcus spp. but reported little or no concern for its development (31). If any 253 degree of resistance was observed it was attributed to thickening of the cell wall, or glycocalyx; 254 traits that were considered to be genetically unstable and reversible upon removal of selective 255 pressure (32). Recently, enterococci isolated from various locations around the world and from 256 both humans and animals, contained both the narasin gene which encodes for ionophore-257 resistance and the vanA gene, raising the possibility that ionophore use may co-select for 258 vancomycin resistance in these strains (30) However, the risk that such strains found on the hides may pose to the food production system is 278 unknown as they were not isolated in the downstream processing environment. 279

Materials and methods 280
Sampling and isolation of Enterococcus spp. 281 Samples were collected a total of 15 times from July 2014 through February 2016 from a 282 commercial beef processing facility in Alberta, Canada that processed more than 3,000 carcasses 283 per day. During each visit 10 samples were obtained from each of four different areas within the 284 plant: carcasses after hide removal (H), carcasses after final washing (W), conveyer belts (C), 285 and ground beef made in the plant (G). A 2 cm x 2 cm gauze swab was used to sample a 286 randomly selected 10 cm x 10 cm area on the surface of the carcasses and conveyor belts. In 287 total, 150 samples were obtained from each sample type or location. During the same time 288 period, 60 samples of retail ground beef (R) were collected from various retail locations in 289 Alberta. The exact origin of these retail ground beef samples was unknown. All samples were 290 transported to the lab on ice and immediately processed. The swabs and 25 g of each ground 291 product and retail ground beef sample were transferred to a stomacher bag for homogenization 292 and pre-enrichment with 10 ml (swabs) or 225 ml (ground product/beef) of buffered peptone 293 water. These samples were then stomached at 260 rpm for 2 min in a Stomacher 400 Circulator 294 (Seward, Norfolk, UK) and incubated overnight at 37°C. 295 One milliliter of this mixture was then added to 9 ml of Enterococcosel broth (BD,  296 14 Oakville, ON, USA) and incubated overnight at 37°C for the enrichment of enterococci. 298 Erythromycin was chosen since macrolides are important in human and veterinary medicine and 299 enterococci are not intrinsically resistant to this antimicrobial. Enterococcosel broth tubes 300 displaying evidence of esculin hydrolysis (black) were streaked onto Enterococcosel agar with 301 and without 8 µg ml -1 erythromycin and incubated at 37°C. After 48 h the plates were examined 302 for colonies with black zones (esculin hydrolysis) and three colonies from each plate were re-303 streaked onto Enterococcosel agar and incubated for 48 h at 37°C. One positive colony from 304 each agar plate was then transferred to 1 ml of brain heart infusion (Dalynn Biologicals, Calgary, 305 AB, Canada) containing 15% glycerol and frozen at -80°C. Confirmation and species 306 identification of presumptive enterococci isolates was done via PCR with the Ent-ES-211-233-F 307 and Ent-EL-74-95-R primers (33) to amplify the groES-EL spacer region as previously described 308 (2).

Antimicrobial resistance screening of enterococci isolates 314
Due to their relevance to human health, isolates with a groES-EL spacer region that was 315 100% identical to E. faecalis or E. faecium were screened for antimicrobial resistance genes 316  (Table S5). 324

Sequencing of selected Enterococcus faecalis and Enterococcus faecalis isolates 325
Forty-seven E. faecalis and eight E. faecium isolates were selected for whole genome 326 sequencing based on their AMR profiles and sample origin. Briefly, the isolates were re-cultured 327 from the frozen glycerol on BEA and incubated for 24 h at 37°C to obtain isolated colonies with 328 typical morphology and colour. A single colony was then streaked onto BHI agar (Dalynn 329 Biologicals), grown overnight at 37°C, and colonies from this plate were suspended in 10 mM 330 Tris-1mM EDTA (TE) (pH 8.0) buffer to obtain an OD 600 of 2.0 (2 x 10 9 cells ml -1 ). One 331 milliliter of this suspension was pelleted via centrifugation at 14,000 x g for 2 min. Genomic 332 DNA was extracted from the pellet using the DNeasy Blood and Tissue kit (Qiagen,333 Mississauga, Ontario, Canada) with the modification that cells were incubated with agitation 334 Trimmomatic v. 0.39 (34) was used to remove sequencing adapters, reads with a quality 344 score of less than 15 over a 4-bp sliding window, and reads that were less than 50 bp in length. 345 Genomes were assembled with SPAdes v. 3.15.1 (35) in "isolate mode" and the quality of the 346 assemblies was assessed with QUAST v. 5.0.2 (36). Potential contamination within each 347 assembly was determined using Kraken 2 v. 2.1.1 and the minikraken2 database v. 2 (37) as 348 well as CheckM v. 1.1.3 (38). GTDB-tk v. 1.3.0 (39) was also used to confirm the taxonomic 349 assignments of the assemblies and Prokka v. 1.14.6 (40) was used to annotate the assemblies. 350 Determination of MLST was done on the assembled genomes using the E. faecalis 351 (https://pubmlst.org/efaecalis) and E. faecium (https://pubmlst.org/efaecium/) MLST databases 352 (41,42). 353 The accessory, core, and pan-genome of the E. faecalis and E. faecium genomes were 354 identified using Roary v. 3.13.0 (43) with a BLASTp identity cut-off of ≥95%. The core genome 355 is defined as genes present in ≥ 99% of genomes. The core genes for both species were aligned 356 in Roary using MAFFT v. 7.475 (44) and a maximum likelihood phylogenetic tree was inferred 357 from this alignment using RAxML v. 8.2.12 (45) and viewed with ggtree v. 2.4.1 (46) in R 358 3.6.1.. Several publicly available E. faecalis and E. faecium assemblies from various isolation 359 sources, including from humans and cattle in Alberta, were also included in the core and pan-360 genome analysis as listed in Table S6. The genome assemblies were screened for virulence genes 361 using the VirulenceFinder 2.0 database (47) and BLASTn (≥90% identity) and for ARGs using 362 the CARD v. 3.0.9 (48) Resistance Gene Identifier (RGI). The depicted gene regions containing 363 ARGs were constructed and validated using contig alignments in Geneious v. 11.0.9. BLAST 364 was used to identify highly similar regions with >80% pairwise identity in bacterial strains 365 present in NCBI. 366    (3)