Identification of CTX-M Type ESBL E. coli from Sheep and Their Abattoir Environment Using Whole-Genome Sequencing

Widespread dissemination of extended-spectrum beta-lactamase (ESBL) Escherichia coli (E. coli) in animals, retail meats, and patients has been reported worldwide except for limited information on small ruminants. Our study focused on the genotypic characterization of ESBL E. coli from healthy sheep and their abattoir environment in North Carolina, USA. A total of 113 ESBL E. coli isolates from sheep (n = 65) and their abattoir environment (n = 48) were subjected to whole-genome sequencing (WGS). Bioinformatics tools were used to analyze the WGS data. Multiple CTX-M-type beta-lactamase genes were detected, namely blaCTX-M-1, blaCTX-M-14, blaCTX-M-15, blaCTX-M-27, blaCTX-M-32, blaCTX-M-55, and blaCTX-M-65. Other beta-lactamase genes detected included blaCMY-2, blaTEM-1A/B/C, and blaCARB-2. In addition, antimicrobial resistance (AMR) genes and/or point mutations that confer resistance to quinolones, aminoglycosides, phenicols, tetracyclines, macrolides, lincosamides, and folate-pathway antagonists were identified. The majority of the detected plasmids were shared between isolates from sheep and the abattoir environment. Sequence types were more clustered around seasonal sampling but dispersed across sample types. In conclusion, our study reported wide dissemination of ESBL E. coli in sheep and the abattoir environment and associated AMR genes, point mutations, and plasmids. This is the first comprehensive AMR and WGS report on ESBL E. coli from sheep and abattoir environments in the United States.


Introduction
Extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae are a serious public health threat and are increasing worldwide, including in the U.S. [1,2]. E. coli are commonly associated with gastro-intestinal, bloodstream, and urinary tract infections [2]. In addition, E. coli serves as a reservoir of transferrable antimicrobial resistance (AMR) genes, which can be passed to pathogenic organisms such as Salmonella spp. [3,4]. Other ESBL types such as SHV and TEM occurred prior to the emergence of CTX-M type ESBLs; however, CTX-M ESBLs became the leading type in clinical isolates in the early 2000s in the U.S. [5,6]. Later, community dissemination of CTX-M type ESBL E. coli, primarily due to bla CTXM- 15 and bla CTXM-14 , was reported among patients in the U.S. [7]. Additionally, CTX-M type ESBLs of food animal origin were first reported in fecal E. coli from sick and healthy dairy cattle in Ohio [8]. Nowadays, there are increasing reports of the dissemination of ESBL-producing E. coli in food animals, retail meat products, companion animals, and the environment in the U.S. and internationally, which in turn may increase public health risk [9][10][11][12][13][14][15].
Dissemination of ESBL E. coli in livestock farm-related environments such as soil, water, manure, air, dust, feed, etc., have recently been reviewed [16]. Although betalactamase genes including bla CTX-M-1 , bla CTX-M-2 , bla CTX-M-3 , bla CTX-M-8 , bla CTX-M- 14 and bla CTX-M-15 , bla SHV , bla TEM, and bla CMY-2 were detected in feces of sheep and retail lamb in other parts of the world [10,[17][18][19][20], there is no report available on AMR determinants of ESBL E. coli in small ruminants in the U.S. Therefore, to fill this gap in information, we conducted a study to detect and characterize AMR determinants using WGS in ESBL E. coli recovered from sheep and their abattoir environment in North Carolina.
Tetracyclines: From a total of 110 Tetracycline-resistant (MIC ≥ 16) ESBL E. coli, 103 (93.6%) carried at least one gene known to confer Tetracycline resistance (Table 1) (Table S3). One isolate that carried tet(M) was phenotypically sensitive to Tetracycline. Seven Tetracyclineresistant ESBL E. coli isolates did not carry any of the above Tetracycline-conferring genes (Tables 1 and S1).

Sequence Types and Phylogenetic Analysis of ESBL E. coli Isolates
ClermonTyping of 113 ESBL E. coli isolates showed that most of the ESBL E. coli isolates belonged to phylogroup A (73/113, 64.6%) and phylogroup B1 (31/113, 27.4%). The remaining nine isolates were assigned to phylogroup C and D (two isolates each), phylogroup E (four isolates), and CladeI (one isolate). Distributions of phylogroups of ESBL E. coli isolates among the different sample types and seasons are shown in Figure 3.

Discussion
To our knowledge, this is the first report of molecular characterization of AMR determinants in ESBL E. coli from sheep and their abattoir environment in the U.S. The isolates were obtained from a year-round serial cross-sectional study between March 2019 and February 2020 in North Carolina. In this study, 95.6% (108/113) of the phenotypically confirmed ESBL E. coli carried CTX-M-type beta-lactamase genes as mechanisms of ESBL production. The most predominant beta-lactamase genes detected in our study were bla CTX-M-1 and bla CTX-M-32 followed by bla CTX-M-55 , bla CTX-M-65 , bla CTX-M-15 , bla CTX-M-27, and bla CTX-M-14 . In the U.S., bla CTX-M-1 was reported as the predominant CTX-M-type ESBL gene in E. coli recovered from environmental samples from dairy farms, livestock auction markets, and equine facilities [21]. However, bla CTX-M-15 is the predominant and widely disseminated ESBL gene carried by ESBL E. coli from dairy cattle farms in other locations and human urinary tract infections in the U.S. [7,[21][22][23]. In cattle and humans, bla CTX-M-27 and bla CTX-M-14 were also commonly reported in these studies. We detected bla CTX-M-15 in ESBL E. coli from 13 isolates recovered from cecal contents, sheep feces, lairage swabs, soil sample, and water, while bla CTX-M-14 and bla CTX-M-27 were less frequent and detected in only two and three isolates, respectively. Six ESBL E. coli isolates (O100:H32, ST10) recovered from both sheep and the abattoir environment in our study carried a combination of three beta-lactamase genes: bla CTX-M-1 (broad-spectrum ESBL gene), bla TEM-1A (narrow spectrum), and bla CMY-2 (AmpC type beta-lactamase gene). Such ESBL E.coli isolates were previously termed as mixed ESBL/AmpC phenotype [24]. CTX-M-type and SHV-type ESBL genes were found to coexist in ESBL E. coli from sheep meat in China [11]. bla SHV , bla OXA, and aac(6)-Ib-cr type beta-lactamase genes were not detected from both sources in our study, which may restate the current predominance of CTX-M and TEM-type ESBLs in E. coli [25]. A combination of CTX-M and TEM type beta-lactamase genes had been reported in ESBL E. coli isolates from sheep in Turkey [19], In this study, five ESBL E. coli isolates carried the AmpC type beta-lactamase gene, bla CMY-2 with bla TEM-1C (n = 4) or alone (n = 1) and did not carry the ESBL gene. The genes known for ESBL production were not detected in these isolates. This observation could be due to other undetected genes or false-positive results in the determination of ESBL status at the screening phase, as previously observed in other studies [26,27]. The other two ESBL producer isolates were resistant to Cefoxitin and Amoxicillin/Clavulanic acid in the absence of bla CMY-2 . These isolates carried ESBL genes bla CTX-M-1 and bla CTX-M-14 combined with bla TEM-1A and bla CARB-2 , respectively. This discrepancy of phenotypic and genotypic results could be the lack of expression of genes in the genotypically predicted resistant but phenotypically susceptible isolates to infer resistance, as previously noticed [28]. This is the first report of multiple beta-lactamase genes in ESBL E. coli from sheep in the United States. Wide dissemination of multiple types of beta-lactamase genes was previously reported from cattle and retail meats excluding lamb and goat in the U.S. [8,9,23] and companion animals (dogs and cats) [12]. From the U.S. public health sector, the most commonly reported CTX-M type genes in ESBL E. coli were bla CTX-M-15 and bla CTX-M-14 [5,7,22,29]. These studies also reported multiple types of beta-lactamase genes in patients with urinary tract and bloodstream infections and pneumonia, including were recently reported [31]. Hence, our study and others indicate the presence and dissemination of clinically important beta-lactamases in E. coli in sheep, their products, and the abattoir environment, and the necessity for routine surveillance of these pathogens.
Moreover, ESBL E. coli from sheep and the abattoir environment carried AMR genes conferring resistance to Tetracyclines, Sulfonamides, Aminoglycosides, phenicols, Quinolones, Macrolides, Trimethoprim, and Lincosamide. AMR-associated point mutations at gyrA, parC, and parE that confer resistance to fluoroquinolones and at uhpT and cyaA that confer resistance to Fosfomycin were detected in these pathogens [32]. From all detected AMR genes in our study, ESBL E. coli from sheep carried a higher proportion of bla CTX-M-1 , bla TEM-1A , floR, qnrB19, and sul2, while those from the environment carried a higher proportion of bla CTX-M-15 and bla TEM-1C . Our study detected genotypic determinants of AMR in ESBL E. coli that were more diversified than in previous reports from cattle and retail meats in the U.S. [9] and sheep in Spain and Portugal [10]. The higher percentage of AMR genes in the sheep in our study could be due to inadequate biosecurity measures, including mixing of animals (sheep, goats, and cattle) from different farms and county fairs, sharing of contaminated feed and water from common sources at the abattoir resting area and prolonged time of duration for interaction, or sharing of AMR bacteria and the associated horizontal gene transfer between them [33]. Although our study did not evaluate these plausible reasons, it was reported that environmental samples from county fairs and livestock auction markets carried a higher level of Cephalosporin and fluoroquinolone-resistant E. coli than those from individual facilities for dairy cattle, equine, or companion animals [21]. At the study abattoir, sheep, goats, and cattle were allowed to roam around for a few hours to up to three days before slaughter. The abattoir operates year-round, receiving animals from different sources, which further increases the chance of introducing diversified genotypes of bacteria. We noticed that the abattoir routinely conducted proper cleaning and applied antiseptics on the lairage at the end of each slaughter day. However, the abattoir resting area was muddy and/or dusty, which might allow immediate contamination of the lairage. We detected a higher diversity of AMR genes in the abattoir environment and recovered a higher percentage of Salmonella and ESBL E. coli in abattoir environmental samples, which supports this observation (data not shown). Another contributing factor could be a large number of animals packed per waiting pens/cubicles as observed during the study.
From the 19 different types of plasmids detected in our study, about 70% of ESBL E. coli isolates carried two or more types. These were primarily incompatibility (Inc type) and colicinogenic (Col type) plasmids. Most plasmids detected in ESBL/AmpC E. coli were reported to be plasmid-mediated [10]. From all plasmids detected in this study, IncA/C, IncF, IncI1-Alpha, IncN, and IncH were previously found to be associated with MDR and commensal E. coli [34,35]. Combining all types of IncF plasmids (IncFIA, IncFIB, IncFIC, IncFIIpCoo, and IncFII), IncF was detected in more than two-thirds (76/113) of the ESBL E. coli isolates, indicating that they were the leading carriers of ESBL genes as previously noted [35]. IncR plasmids were the second abundant (57/113) types of plasmids in our study. IncR plasmid was described to carry genes belonging to many classes of antimicrobials, including beta-lactams and quinolones [35].
In this study, most of the isolates were phylogroups A (73/113) and B1 (31/113), followed by E (4/113), C (2/113), D (2/113), and CladeI (1/113), and all except phylogroup C were detected in isolates from sheep samples. Phylogroup A was detected at a higher proportion in isolates from all sample types except those from soil samples, where a higher proportion of phylogroup B1 was detected. ESBL E. coli isolates from cecal content had the most diversified phylogroups (A, B1, D, E, and CladeI). An abattoir-based study in Portugal indicated that 92.6% (50/54) of E. coli recovered from sheep were phylogroup A and B1 [39], the remaining two each from phylogroup B2 and D. However, the proportion of B1 was about twice the proportion of A1 in their study, contrasting the result in our study. Similarly, the predominance of phylogroups A and B1 in E. coli was reported in ruminants (cattle and sheep) in Turkey. In addition, they reported phylogroup D both from cattle and sheep but did not report other phylogroups [19]. Phylogroup B2 and D are considered pathogenic [40]. Two isolates in our study were phylogroup D.
The phylogenetic analyses revealed that most of the unique sequence types tend to cluster around seasons but not around sample type or source of isolates. This may suggest close interaction between animals at the slaughter facility and the abattoir environment, facilitating the sharing of bacteria and AMR genes. Although only ST10 and ST398 were detected across all seasons and ST58 and ST2325 were detected in three seasons, these isolates were clonal, indicating persistence in the environment and animals throughout the year. This could be due to differences in bacterial fitness, previous environmental dissemination, and livestock farms and markets where the animals come from. It was interesting to see that these STs harbored diverse types of beta-lactamase genes. ST10 isolates harbored eight unique types of beta-lactamase genes (five CTX-M-types, AmpC type, and two TEM-types), ST58 and ST2325 harbored three CTX-M types, and the former had one TEM type beta-lactamase gene. However, isolates with ST398 harbored only bla CTX-M-32 and bla CARB-2 . This might need further investigation. A recent report indicated such fitness differences could be associated with plasmid-host adaptations [42].
Core genome phylogenetic analyses indicated that almost all types of beta-lactamase genes were scattered throughout the phylogenetic tree. Similar STs were detected in isolates recovered from both sheep and the environment. These may further indicate close interaction and mobile genetic transfer of acquired AMR genes between isolates from both sources. For example, six clonal ESBL E. coli isolates (O100:H32; ST10-A) that carried a combination of three beta-lactam genes were recovered from six different samples and detected in two seasons (fall and winter).
The study had limitations, as some important demographic information was not accessible such as the history of illnesses and antimicrobial use, geographical source of animals, history of transportation, dietary changes, and husbandry management. The study did not evaluate the possible contribution of cattle and goats at the same facility in the dissemination of ESBL E. coli and AMR genes. Additionally, we did not look into the effect of transportation and abattoir environment in acquiring AMR genes and their dissemination to sheep and their products.
In conclusion, this is the first comprehensive report of AMR determinants in ESBL E. coli from sheep and their abattoir environment in the U.S. Sheep are a significant reservoir of ESBL E. coli and AMR determinants, and this study notably indicated close interaction between ESBL E. coli from sheep and their abattoir environment. The abattoir environment might have played a significant role in the persistence and dissemination of these pathogens. We propose routine AMR surveillance of sheep and their products to prevent future public health risks.

Study Design and Bacterial Isolates
From the pool of ESBL E. coli isolates recovered during a serial cross-sectional study conducted between March 2019 and February 2020, we selected 113 ESBL E. coli isolates for molecular characterization of AMR determinants. The selected isolates were recovered from sheep samples (n = 65) and abattoir environment samples (n = 48). Break down of samples collected and sampling methodology are described in Table S4. Sources of ESBL E. coli isolates from sheep were carcass swabs (n = 10), feces (n = 28), cecal contents (n = 20), and abattoir resting area feces (n = 7), and those from the abattoir environment were lairage swabs (n = 21), soil (n = 10), feed (n = 8) and water (n = 9). The abattoir slaughtered sheep, goats, and cattle on a routine basis. These animals were allowed to roam around from a few hours to up to three days and share feed and water from the same troughs. Information on antimicrobial use, husbandry, and demography was not accessible to us. ESBL E. coli isolates were selected based on their AMR profile, the season of sampling, and the type (source) of samples. Confirmation of ESBL production was conducted using double-disk diffusion methods following Clinical and Laboratory Standards Institute (CLSI) guidelines [43]. Confirmed ESBL E. coli isolates had a zone of inhibition of ≥5 mm for either Cefotaxime or Ceftazidime with Clavulanic acid compared to without Clavulanic acid. The isolates' antimicrobial susceptibility was determined by broth microdilution methods using the NARMS Sensititre 14 antimicrobial drug panel. Data interpretation and categorization into susceptible, intermediate, and resistant were determined based on resistance breakpoints recommended by the CLSI of the U.S. [44,45], except for Streptomycin, which was determined based on resistance breakpoints recommended by the NARMS [46]. The number and percent resistance of ESBL E. coli isolates for the fourteen antimicrobials in the NARMS Sensititre panel are presented in Table 1.

Whole-Genome Sequencing
The template DNA for whole-genome sequencing (WGS) was extracted from an overnight culture of all selected E. coli isolates using the Qiagen DNeasy PowerLyser Microbial Kit following the manufacturer's protocol. The purified DNA was quantified using a NanoDrop 2000 Spectrophotometer (Thermo Scientific, USA). The sequencing DNA library was prepared using the Nextera DNA Flex Library preparation kit (Illumina, San Diego, CA, USA) as previously described [47]. A Qubit 3.0 Fluorometer (ThermoFisher Scientific, Waltham, MA, USA) was used to quantify the library prep. WGS was performed on Illumina MiSeq with 300 bp paired-end reads. The average number of assembled contigs per sample was 96 (range 40 to 254), the average N50 was 201 kb (range 79 kb to 672 kb), and the total assembly length was 4.6 to 5.6 megabases (Mb).
Sequences were assembled using SPAdes 3.14.1 [48] and annotated with PROKKA [49] at default settings. The quality of genome assembly was assessed using Quast [50]. AMR genes, plasmids, and virulence genes were identified by the ABRicate pipeline, as previously described [51]. ABRicate included multiple databases including NCBI, CARD, ARG-ANNOT, ResFinder, MEGARES, EcOH, PlasmidFinder, Ecoli_VF, and VFDB. Reported AMR genes and plasmids were primarily based on summary results from ResFinder [52] and PlasmidFinder [53] databases of ABRicate program, respectively. The NCBI's AM-RfinderPlus database (version 3.10.5, Bethesda, MD, USA) [54] was used for the detection of AMR-associated point mutations. A gene was considered present in the assembled genome of an isolate when there was 90% nucleotide identity and 80% coverage of length match with the specific gene in the database. In silico serotyping of the E. coli isolates was carried out using the EcOH database [55] in the ABRicate program, whereas E. coli isolates were phylogrouped using ClermonTyping [56], which divides them into seven main phylogroups termed A, B1, B2, C, D, E, and F.

Phylogenetic Analysis
Prokka (version 1.14.6) was used to annotate isolate genomes [49], and pan-genome analyses were conducted using Roary (version 3.13.0) with a minimum percentage identity for blastp of 95% [57]. Within Roary, MAFFT [58] was used to create a core genome alignment of genes present in 99% of the isolates. The core genome alignment was used to generate a phylogenetic tree on RaxMLGUI2.0 (RaxML-NG version 1.0.1) [59]. The bestfitting model identified was general time-reversible substitution with a Gamma rate of heterogeneity and a proportion of invariable sites estimate (GTR + I + G) and used to generate the maximum-likelihood phylogenetic tree with 500 bootstrap replicates. The phylogenetic tree was visualized and annotated using iTOL version 6.3 (https://itol.embl.de/itol.cgi; accessed on 19 July 2021) [60].

Statistical Analyses
The frequency of detection of AMR genes in ESBL E. coli from sheep and the abattoir environment was estimated. Parameters of central tendency and dispersion, bar diagrams, contingency tables, and simple proportions were obtained. The statistical significance was set at the alpha value of ≤ 0.05. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/pathogens10111480/s1, Table S1: Phenotypic AMR profiles, AMR genes, and AMR associated point mutations detected in ESBL E. coli isolates (n = 113) from sheep and abattoir environment, Table S2: Frequency of AMR determinants detected in ESBL E. coli isolates (n = 113) among sample sources and seasons, Table S3: Number and percentage of AMR genes other than beta-lactamases in ESBL E. coli isolates (n = 113) from sheep and abattoir environment.