Genomic Characterization of Carbapenem-Resistant Bacteria from Beef Cattle Feedlots

Carbapenems are considered a last resort for the treatment of multi-drug-resistant bacterial infections in humans. In this study, we investigated the occurrence of carbapenem-resistant bacteria in feedlots in Alberta, Canada. The presumptive carbapenem-resistant isolates (n = 116) recovered after ertapenem enrichment were subjected to antimicrobial susceptibility testing against 12 different antibiotics, including four carbapenems. Of these, 72% of the isolates (n = 84) showed resistance to ertapenem, while 27% of the isolates (n = 31) were resistant to at least one other carbapenem, with all except one isolate being resistant to at least two other drug classes. Of these 31 isolates, 90% were carbapenemase positive, while a subset of 36 ertapenem-only resistant isolates were carbapenemase negative. The positive isolates belonged to three genera; Pseudomonas, Acinetobacter, and Stenotrophomonas, with the majority being Pseudomonas aeruginosa (n = 20) as identified by 16S rRNA gene sequencing. Whole genome sequencing identified intrinsic carbapenem resistance genes, including blaOXA-50 and its variants (P. aeruginosa), blaOXA-265 (A. haemolyticus), blaOXA-648 (A. lwoffii), blaOXA-278 (A. junii), and blaL1 and blaL2 (S. maltophilia). The acquired carbapenem resistance gene (blaPST-2) was identified in P. saudiphocaensis and P. stutzeri. In a comparative genomic analysis, clinical P. aeruginosa clustered separately from those recovered from bovine feces. In conclusion, despite the use of selective enrichment methods, finding carbapenem-resistant bacteria within a feedlot environment was a rarity.


Introduction
Carbapenems are β-lactam antibiotics that consist of a four-membered β-lactam ring fused with a secondary five-membered thiazolidine ring through the nitrogen and adjacent tetrahedral carbon atom. Unlike other β-lactams, carbapenems have two substitutions, at position one there is a substitution of sulfur for a carbon atom and at the fourth position of the thiazolidinic moiety, a carbon is substituted for a sulfone [1,2]. So far, four carbapenems, including ertapenem, meropenem, doripenem, and imipenem, have been approved for use in the US. These members differ in their side chains, influencing their antimicrobial activity. Carbapenems inhibit cell wall synthesis by preventing the formation of cross-linkages in peptidoglycan via binding to peptidoglycan binding protein (PBP), thus leading to cell lysis and death [3]. The ability of carbapenems to bind to diverse PBPs with high affinity and their stability against extended-spectrum β-lactamases (ESBLs) and AmpC β-lactamases account for their broad-spectrum activity against both Gram-negative and Gram-positive bacteria [2,4,5]. Ertapenem binds preferentially to PBPs 2 and 3 of Escherichia coli and

Recovery of Carbapenem-Resistant Isolates and Species Identification
A total of 116 presumptive carbapenem-resistant isolates were recovered from bovine fecal and catch basin samples. Among these 8 were E. coli, 100 were Pseudomonas spp., 5 were Acinetobacter spp., 2 were Ochrobactrum intermedium (Brucella intermedia), and 1 was identified as Stenotrophomonas maltophilia. Figure 1 shows the antimicrobial resistance profiles of 116 isolates tested using antibiotic discs. Of all tested E. coli (n = 8) isolates, one isolate was resistant to ertapenem and tetracycline, while all others exhibited intermediate resistance to ertapenem. All E. coli isolates were negative for carbapenemase production ( Figure 1).

Phenotypic Characterization
All Acinetobacter spp. isolates (n = 5) were resistant to both ertapenem and meropenem and positive for carbapenemase production. One isolate of Acinetobacter spp. was also resistant to tetracycline and trimethoprim-sulfonamide. Tobramycin resistance was also identified in two Acinetobacter spp. isolates recovered from bovine feces ( Figure 1).
A single O. intermedium isolate was resistant to ertapenem and meropenem, with both O. intermedium isolates being negative for carbapenemase production. The single S. maltophilia isolate identified in this study was resistant to all tested antibiotics and was positive for carbapenemase production ( Figure 1 and Table 1). A single O. intermedium isolate was resistant to ertapenem and meropenem, with both O. intermedium isolates being negative for carbapenemase production. The single S. maltophilia isolate identified in this study was resistant to all tested antibiotics and was positive for carbapenemase production ( Figure 1 and Table 1).
Hybrid genome assemblies allowed complete circular genomes to be constructed for some of the isolates (Supplementary Table S1). Moreover, some ARGs, including blaOXA-265 in A. haemolyticus, and blaPST-2 in P. saudiphocaensis, were only identified after hybrid assemblies (data not shown) were generated.

Discussion
Increasing carbapenem resistance has threatened the clinical utility of these drugs in human medicine, leading to the challenge of "extreme drug resistant" bacteria [22]. Therefore, the Centers for Disease Control and Prevention (CDC) has declared carbapenem-resistant Enterobacteriaceae as an urgent and serious threat to human health [23]. In this study, we investigated if carbapenem-resistant E. coli could be recovered using carbapenem enrichment from bovine feces or catch basin water samples collected in intensive feedlots. We had extremely low recovery of carbapenem-resistant E. coli in bovine feces. The majority of isolates that were presumed to be E. coli were subsequently identified as other bacterial species, primarily Pseudomonas species. This is not surprising as we used etrapenem for sample enrichment, and Pseudomonas spp. are intrinsically resistant to this antimicrobial [24,25].

Discussion
Increasing carbapenem resistance has threatened the clinical utility of these drugs in human medicine, leading to the challenge of "extreme drug resistant" bacteria [22]. Therefore, the Centers for Disease Control and Prevention (CDC) has declared carbapenem-resistant Enterobacteriaceae as an urgent and serious threat to human health [23]. In this study, we investigated if carbapenem-resistant E. coli could be recovered using carbapenem enrichment from bovine feces or catch basin water samples collected in intensive feedlots. We had extremely low recovery of carbapenem-resistant E. coli in bovine feces. The majority of isolates that were presumed to be E. coli were subsequently identified as other bacterial species, primarily Pseudomonas species. This is not surprising as we used etrapenem for sample enrichment, and Pseudomonas spp. are intrinsically resistant to this antimicrobial [24,25].
In P. aeruginosa, we identified four different gene variants belonging to the blaOXA-50 class (blaOXA-50, blaOXA-902, blaOXA-486, and blaOXA-494) encoding carbapenemhydrolyzing oxacillinase (CHDLs). All of these variants were also present in clinical isolates in addition to seven other variants blaOXA-395, blaOXA-488, blaOXA-847, blaOXA-848 blaOXA-901, blaOXA-905, and blaOXA-906. It has been reported that blaOXA-50 naturally exists in all P. aeruginosa and does not appear to have been acquired based on the similar GC% content of the blaOXA-50 gene to the overall P. aeruginosa genome [24]. We did not find any insertion sequences, repeat elements, relaxases, integrases, within the genomic proximity of blaOXA-50, suggesting that this gene lacks mobility and is unlikely to be readily transferred to other bacterial species.
In the core-genome phylogenetic analysis, P. aeruginosa isolates from bovine sources clustered separately from those obtained from humans. Similar observations have been reported in Brazil, where P. aeruginosa recovered from healthy bovine urine samples segregated from human clinical urinary tract isolates [34]. These observations might be associated with a prolonged association and adaptation of these isolates within their respective environmental hosts. P. aeruginosa is known for its remarkable ability to adapt to diverse ecological niches, from soil to various living hosts [35,36]. Genes associated with metabolism and pathogenesis constitute the core-genome of P. aeruginosa, whereas genes required to adapt to various niches constitute the accessory genome. These genes were found to cluster in certain loci referred to as 'regions of genomic plasticity' [37,38]. In this study, P. aeruginosa recovered from the beef production system carried ARGs conferring resistant to chloramphenicol (catB7), β-lactams (blaOXA-50), fosfomycin (fosA), aminoglycosides (aph(3 )-Iib), and cephalosporins (blaPDC-55, blaPDC-374). These ARGs were also found in the genomes of human clinical isolates and were mapped on chromosomes in both bovine and human clinical isolates. In P. aeruginosa, the type IV pilus uptakes foreign DNA during transformation events [39,40]. The accessory genome can be distinguished from the core genome by its aberrant GC content, codon usage, and tetranucleotide usage [41]. The GC content of P. aeruginosa (~66.3%) is mostly higher than the foreign DNA. Over time, the acquired DNA may lose the sequence compositional difference that distinguishes it from the core genome of P. aeruginosa as it undergoes the same selective pressure as the core genome [42].
We observed variable phenotypic susceptibility profiles among P. aeruginosa isolates that harbored similar AMR profiles. Some isolates were resistant to all carbapenems, some were resistant to both meropenem and doripenem and some were only resistant to meropenem. Other than β-lactamase production, alterations or lack of porin OprD, and overexpression of resistance-nodulation-division (RND) efflux pumps (MexAB-OprM, MexCD-OprJ, MexEF-OprN, and MexXY-OprM) are also associated with carbapenem resistance in Pseudomonas species [43,44]. Not only does the combinations of these mechanisms confer reduced susceptibility to carbapenems, but the overexpression of these efflux pumps result in β-lactam, tetracycline, trimethoprim, aminoglycosides, and fluoroquinolone resistance [45,46]. It is more likely that the variation in susceptibility profiles in our isolates is associated with the varied number of MexAB-OprM operon and OprD found in each isolate [47,48]. Moreover, isolates with multiple MexAB-OprM operons also possessed a single copy of the MexR repressor gene. The presence of only one repressor gene in comparison to multiple efflux pumps as regulatory targets may disturb the molecular stoichiometry of the regulation affecting efflux pump expression, thus resulting in varied phenotypic resistance profiles among P. aeruginosa isolates.
P. stutzeri (Stutzerimonas stutzeri) is an opportunistic pathogen that rarely causes infection in humans [49]. In our study, we found the carbapenem-resistant gene, blaPST-2 in the P. stutzeri genome. The blaPST-2 encodes for a subclass B1 metallo-β-lactamase and was first identified on the chromosomes of P. stutzeri DSM 10701 [50]. From a genetic context, we did not find genes associated with mobility in proximity to this ARG. However, we suggest that blaPST-2 is not intrinsic to P. stutzeri as it was not present in all P. stutzeri genomes in the NCBI database (only 4 out of 19 P. stutzeri genomes had blaPST-2). Moreover, blaPST-2 was phylogenetically related to the previously characterized mobile subclass B1 metallo-β-lactamase families, including KHM, SIM, and IMP in P. aeruginosa and Citrobacter freundii, providing further evidence that this gene was more likely acquired [51][52][53].
Pseudomonas saudiphocaensis has only recently been classified as a new species [54] and has not been well characterized. There are only two studies where P. saudiphocaensis was recovered, first from air samples in the city environment of Makkah, Saudi Arabia, in 2012 [54] and then from a sheep dairy farm in New Zealand [55]. To the best of our knowledge, this is the first report of recovery of P. saudiphocaensis from bovine feces. With the hybrid genome assembly, we were able to construct a complete circular genome of this isolate. The genome of P. saudiphocaensis (3.6 Mbp) was small as compared to other Pseudomonas species, which ranged from 5.5 to 6.7 Mbp. These isolates also carried blaPST-2, which has not been identified in other P. saudiphocaensis genomes [54,55]. The blaPST-1 in P. stutzeri and P. saudiphocaensis showed 95% amino acid similarity. Despite limited knowledge of P. saudiphocaensis, it is suggested that it may have acquired this gene.

Sampling, Isolation and Identification
The study includes bovine fecal and catch-basin water samples, collected from four feedlots (pens = 301) located in Alberta, over a period of two years (August 2016 to June 2019). Briefly, 20 g of feces were collected from 20 fresh fecal pats and placed in a sterile plastic container and thoroughly mixed. Each pen housed approximately 180 feedlot cattle. For enrichment, 0.5 g of the mixed fecal sample was inoculated into 4.5 mL of Escherichia coli (EC) broth containing 2 µg/mL of ertapenem (ETP, sigma Aldrich, ref: sml1238), followed by overnight incubation at 37 • C in a shaking incubator at 250 RPM. The enriched samples were then sub-cultured on MacConkey agar supplemented with 2 µg/mL ertapenem. From each sample, a maximum of three colonies were selected and sub-cultured on nutrient agar supplemented with 0.5 µg/mL ertapenem (Dalynn Biologicals, Calgary, AB, Canada). Water samples were collected from feedlot catch basins which collected run off water from the feedlot. Briefly, 1 L of catch-basin water was collected into a polyethylene bottle attached to a telescopic pole at two different locations per site, which were combined to generate a composite sample. Water samples were transported on ice within 4 h of collection for processing. Composite catch basin water (10 mL) was filtered through a 0.45 µm pore size filter of 47 mm diameter (S-Pak ® EMD Millipore Corp, Billerica, MA, USA) using a sterile vacuum-manifold filtration system (Pall Corporation, Port Washington, NY, USA). The filter was then submerged in 4.5 mL EC brothertapenem (0.5 mg/L) and incubated overnight at 37 • C with shaking, followed by sub-culture onto MacConkey Agar supplemented with 0.5 µg/L ertapenem at 37 • C (Dalynn Biologicals, Calgary, AB, Canada). A maximum of three colonies were selected from each sample and sub-cultured on nutrient agar supplemented with 0.5 µg/mL ertapenem at 37 • C for 24 h. For the identification of recovered isolates (n = 116), the 16S rRNA gene was amplified using universal bacterial 16S rRNA gene primers 27F (5 -AGAGTTTGATCCTGGCTCAG-3 ) and 1492R (5 -GGTTACCTTGTTACGACTT-3 ) followed by Sanger sequencing of the amplified PCR product [56]. Species were identified using BLAST search against the NCBI bacterial database.
Isolates were also tested for the production of carbapenemases using the chromogenic Carba NP test (RAPIDEC ® CARBA NP kit, BioMérieux, St-Laurent, QC, Canada), with Klebsiella pneumoniae ATCC 700603 and Klebsiella pneumoniae OLC2685 used as negative and positive controls, respectively. A bacterial colony (10-µL loop) was picked up from overnight-incubated Mueller-Hinton agar plates and mixed into API suspension medium. The bacterial suspension was then transferred to wells in a test strip and incubated at 37 • C. Test strips were read at 30 and 120 min. A 'positive' test corresponded to a color change from red to yellow-orange, while no-change in color was considered 'negative'.

Whole-Genome Sequencing, Assembly and Annotation
All Carba NP positive isolates (n = 28) and a subset (n = 14) of Carba NP negative isolates selected to be representative of identified species were subjected to whole genome sequencing using short and long reads sequencing technologies. High molecular weight genomic (HMW) DNA was extracted using a Genomic DNA preparation kit with Genomictip 20/G (Cat: 13323; QIAGEN, Germantown, MD, USA) according to manufacturer's instructions. DNA quality and quantity was estimated using a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Mississauga, ON, Canada) and a Qubit Fluorometer with PicoGreen (Q32850, Invitrogen, Carlsbad, CA, USA), respectively. The integrity of DNA was confirmed through agarose gel electrophoresis. Short-read sequencing was performed on the Illumina MiSeq platform. A genomic library was constructed using the Illumina NexteraXT DNA sample preparation kit (Illumina Inc., San Diego, CA, USA) followed by sequencing on Illumina MiSeq platform using the MiSeq Reagent Kit V3, generating 2 × 300 base paired-end reads. Long-read sequencing was performed on PromethION platform from Oxford Nanopore technologies (ONT . Adaptor-ligated DNA was bead-cleaned using a 0.4× volume of beads following the manufacturer's instructions using the supplied polyethylene glycol (PEG) based wash buffer. The DNA from the last step (200-400 ng) was loaded onto an ONT PromethION sequencing flow cell as directed by the manufacturer. MinKNOW Core 3.1.20 and guppy 2.0.10 were used for flow cell signal processing and base calling during each run, and reads were assembled de novo using Flye version 2.9.1 [57].

Comparative Genomic Analysis
Whole genome comparative genomic analysis was conducted between all the Pseudomonas aeruginosa isolates (n = 20) sequenced in this study to P. aeruginosa genomes (n = 76) (Supplementary Table S1) originating from North America, from the PathoSys-tems Resource Integration Center (PATRIC) (https://www.patricbrc.org; accessed on 20 February 2023). Although these same genomes were also available from the NCBI database, the metadata in the PATRIC database was more detailed. These isolates originated from human clinical sources, with the exception of a single isolate from cattle.
The core-genome phylogenomic tree was constructed using the SNVphyl pipeline v.1.2.3. The phylogenetic tree was generated by aligning paired-end reads against the P. aeruginosa POA1 reference genome (NC_002516.2) using SMALT (c.0.7.5; https://sourceforge.net/projects/smalt/ accessed on 10 March 2023). The generated read pileups were then subjected to quality filtering (minimum mean mapping quality score of 30), coverage cut-offs (15× minimum depth of coverage), and a single nucleotide variant (SNV) abundance ratio filter of 0.75 to obtain a multiple sequence alignment of SNV-containing sites. The SNV alignment, with no density filtering, was used to create a maximum likelihood phylogeny using PhyML version 3.0. The generated Newick file was visualized using the Interactive Tree Of Life (iTOL) v.6 [64].
We also compared variants of the blaOXA-50 of P. aeruginosa isolates used in this study to determine their genetic relatedness. For this, blaOXA-50 coding sequencing (CDS) was computationally extracted from each genome and aligned using MAFFT version 7.490. The resultant alignment-based tree was visualized using iTOL v.6 [64].

Conclusions
In conclusion, the recovery of carbapenem-resistant bacteria from beef production systems is quite rare. The majority of carbapenem-resistant bacterial species, including P. aeruginosa, A. haemolyticus, A. junii, A. lwoffii, and S. maltophilia, carried intrinsic carbapenemresistant genes and were only recoverable following enrichment. P. aeruginosa found in this study were multidrug-resistant, a possible reflection of the ability of this species to readily acquire foreign genes via transformation through the type IV pilus. Moreover, the phylogenetic analysis showed that bovine P. aeruginosa strains formed separate clusters from human clinical strains, indicating that they may have adapted to the cattle environment, and prudent management in feedlot systems has limited the spread of these isolates outside of cattle production systems.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/antibiotics12060960/s1, Table S1 Hybrid assemblies quality statistics determined using Quast; Table S2 List of Pseudomonas aeruginosa genomes retrieved from a public database for analysis; Table S3 AMR gene profiles of Pseudomonas aeruginosa genomes retrieved from a public database for analysis.