Whole-Genome Sequence Analysis of CTX-M Containing Escherichia coli Isolates from Retail Meats and Cattle in the United States

In recent years, there have been increased reports on the detection of extended-spectrum beta-lactamase (ESBL)-producing Escherichia coli and Salmonella strains from food-producing animals and animal products in the United States. We characterized 18 ESBL E. coli isolates from cattle (n = 5), chicken breast (n = 5), ground turkey (n = 6), ground beef (n = 1), and pork chops (n = 1) that were collected by the National Antimicrobial Resistance Monitoring System (NARMS) between 2011 and 2015. In vitro antimicrobial susceptibility testing was done against a panel of 14 antimicrobials followed by a secondary panel of 9 β-lactam agents. Whole-genome sequencing was used to characterize the resistome, plasmids, and the genetic structures of the ESBL genes. All ESBL-producing E. coli isolates were resistant to at least three antimicrobial classes and carried various blaCTX-M genes. Most of the cattle and ground turkey isolates carried blaCTX-M-27. In chicken breast isolates, blaCTX-M-1 was present as part of an ISEcp1 transposition unit carried on a plasmid that shares sequence similarity with the backbone structure of the IncI plasmid. Isolates carrying the blaCTX-M-14 and blaCTX-M-15 genes, widely distributed in human clinical isolates, were also isolated. To our knowledge, this is the first report of the widely distributed blaCTX-M-14 and blaCTX-M-15 in E. coli isolates from retail meat samples in the United States. Different insertional sequences were identified upstream of these blaCTX-Ms, including ISEcp1, IS26, and IS903-D. CTX-M in E. coli from food animals and retail chicken breast were often present on plasmids with other resistance genes. Other resistance genes identified included aadA, strA, strB, aac(3)-IId, aac(3)-VIa, aph(3′)-Ic, blaTEM, blaHERA-3, floR, sul1, sul2, catA1, tetA, tetB, dfrA, and qacE. These data describe the emergence of CTX-M-carrying E. coli isolates in food animals and animal products monitored by NARMS program.

have become the predominant enzyme type in many parts of the world, 9 and have spread rapidly through clinical populations of Enterobacteriaceae. 10 Reports from several countries describe the presence of CTX-M-producing E. coli strains in apparently healthy food animals [11][12][13] and food animal products, 12 as well as pets 14 and wild birds. 15 Although CTX-Mproducing strains appear to have quickly spread worldwide, the increased prevalence of E. coli carrying these blactamases in numerous U.S. hospitals became apparent in the early 2000s. 7,10,16 Intestinal carriage of CTX-M-producing bacteria in foodproducing animals and contamination of retail meat might contribute to increased occurrences of infections with ESBL-producing bacteria in humans. A study on the presence of indistinguishable E. coli genotypes carrying CTX-M genes obtained from poultry, poultry products, and human clinical samples in the Netherlands has suggested the possible exchange of these genes through the food chain. 12 In the United States, few CTX-M ESBLs have been reported from food animals and animal products. 11,[17][18][19] The National Antimicrobial Resistance Monitoring System (NARMS) monitors changes in antimicrobial susceptibilities of zoonotic foodborne bacteria to medically important antimicrobials, including b-lactam antibiotics. Whole-genome sequencing has improved our ability to monitor resistomes and helps to identify and characterize emerging resistance genes and mobile genetic elements that facilitate the spread of these genes. The aim of this study was to investigate and characterize antimicrobial resistance genes and mobile genetic elements associated with phenotypically positive ESBL E. coli isolates recovered from cattle and retail meat samples collected through the NARMS program between 2011 and 2015. This information will help to characterize the molecular epidemiology of CTX-M carrying E. coli isolates in food animals and animal products monitored by the NARMS program.

Bacterial strains
Eighteen phenotypically positive ESBL E. coli isolates recovered from cattle fecal samples (n = 5) and retail meats (chicken breast [n = 5], ground turkey [n = 6], ground beef [n = 1], and pork chops [n = 1]) by the NARMS program between 2011 and 2015 were identified and selected for characterization. The isolates were identified from a total of 8,721 E. coli isolates recovered from fecal samples of healthy cattle and retail meat samples. The fecal isolates (n = 3,079) were recovered from healthy cattle as part of a NARMS on-farm pilot program to monitor antimicrobial resistance in foodborne pathogens. The retail meat E. coli isolates (n = 5,642) were recovered from chicken breast, chicken wing, pork chops, ground beef, and ground turkey.
A total 321 E. coli isolates with minimum inhibitory concentrations (MICs) ‡8 mg/mL for ceftiofur and/or ‡4 mg/ mL for ceftriaxone for isolates recovered before 2015 and ‡2 mg/mL for isolates recovered in 2015 were selected for further testing with a second panel of 9 b-lactam antimicrobials: aztreonam (ATM), cefquinome (CQN), imipenem (IMI), cefepime (FEP), piperacillin-tazobactam (TZP), ceftazidime (TAZ), ceftazidime-clavulanic acid (CAZ/ CLA), cefotaxime (FOT), and cefotaxime-clavulanic acid (CTX/CLA). The Clinical and Laboratory Standards Institute (CLSI) confirmatory test for ESBL production was used and is based on cefotaxime and ceftazidime MICs with and without clavulanic acid. Isolates showing a three or more twofold concentration decrease in the cefotaxime and ceftazidime MICs when tested in combination with clavulanate versus the MICs of cefotaxime and ceftazidime when tested alone are considered ESBL + . 20 E. coli ATCC 25922, Enterococcus faecalis ATCC 29212, Staphylococcus aureus ATCC 29213, Pseudomonas aeruginosa ATCC 27853, and Klebsiella pneumoniae ATCC 7000603 were used as quality control organisms for MIC determinations. Results were interpreted according to CLSI guidelines for broth microdilution methods with the exception of STR (NARMS resistance breakpoint, ‡32 mg/ mL), AZM (NARMS resistance breakpoint, ‡32 mg/mL), and CQN (NARMS resistance breakpoint, ‡32 mg/mL). 20

Conjugation
Conjugation experiments using a plate mating protocol were used to determine the transferability of resistance phenotypes and localize CTX-M genes to conjugative plasmids. We selected seven E. coli isolates that carried different CTX-M genes [N36254PS (bla CTX-M-32 ), N36410PS (bla CTX-M-27 ), N37058PS (bla CTX-M-32 ), N40513 (bla CTX-M-1 ), N40607 (bla CTX-M-1 ), N46045 (bla CTX-M-15 ), and N51980 (bla CTX-M-14 )] as donor cells. MAX Efficiency Ò DH5aÔ E. coli Competent Cells (Invitrogen, Carlsbad, CA) were used as recipients. The donors and recipients were grown in 2 mL LB medium (Becton Dickinson, Sparks, MD) at 37°C in a shaker incubator for 16-18 hours. Ten microliters of donor cells were spotted on top of 10 mL of recipient strain (DH5a) on blood agar plates and incubated at 37°C overnight. Each co-culture was then scraped from the plate and resuspended in 1 mL LB broth. Ceftiofur and nalidixic acid (Sigma-Aldrich, St. Louis, MO) were used as selective agents for the donor and recipient strains, respectively. Transconjugants were selected on LB agar containing nalidixic acid (30 mg/mL) and ceftiofur (4 mg/ mL). The MICs of donors, recipients, and transconjugants were determined using the Sensititre semiautomated antimicrobial susceptibility system. The b-lactam susceptibility testing panel was used to confirm the phenotype.

PCR-based plasmid replicon typing of transconjugants
Genomic DNA was extracted using the DNeasy Blood and Tissue kit (Qiagen, Valencia, CA) following the manufacturer's instructions. Amplification of plasmid replicon targets was carried out following the protocol described by Johnson et al. 21 with minor modifications for IncP characterization. For IncP, a simplex PCR with an annealing temperature of 65°C was used. The amplified products were separated by gel electrophoresis on 1.0% agarose gels.

Whole-genome sequencing
Whole-genome sequencing was used to characterize the resistome and plasmids in all strains (n = 18). Briefly, DNA was extracted using the DNeasy Blood and Tissue Kit (Qiagen) following the manufacturer's instructions. Wholegenome sequencing was performed on the MiSeq Desktop Sequencer using v2 sequencing reagent kits (Illumina, San Diego, CA). A de novo assembly was performed using CLC Genomics Workbench version 8.0 (Qiagen). Contigs of less than 200 bp were removed from analysis. The number of assembled contigs ranged between 78 and 258 with an average coverage of 50 · .

Resistome analysis
Resistance genes were identified using BLASTX a and the ResFinder resistance gene database. 22 The BLASTX results were processed with in-house PERL scripts to identify antimicrobial resistance genes using an 85% amino acid identity and 50% minimum sequence length.

Phylogenetic analysis
The Center for Food Safety and Applied Nutrition (CFSAN) SNP pipeline b was used to create the single nucleotide polymorphism (SNP) matrices from sequence data for the phylogenetic analysis. SNP redundancy by linkage disequilibrium was reduced and the phylogenetic tree was constructed with the maximum likelihood algorithm using the SNPhylo package. 23

Plasmid profiling
Identification of the plasmid type was done using the PlasmidFinder database. c The cutoff threshold for identity was set at 95% to determine the existence for a particular plasmid.
Multilocus sequence typing profiling E. coli multilocus sequence typing (MLST) allelic profiles and sequences were downloaded from the PubMLST database. d A total of 7,113 profiles for 7 different loci were used for the MLST. The SRST2 pipeline 24 was used to determine the MLST type for our E. coli isolates.

Results
In vitro antimicrobial susceptibility testing of E. coli isolates The antimicrobial resistance profiles of phenotypically positive ESBL E. coli isolates are shown in Table 1. All the isolates were resistant to ampicillin, ceftiofur, ceftriaxone, and cefotaxime, as expected. Three of the 5 cattle and 2 of the 13 retail meat E. coli isolates were resistant to cefquinome, a fourth-generation cephalosporin. In addition, one of the cattle E. coli isolates and two of retail meat isolates showed resistance to aztreonam, a monobactam. Among the 18 strains producing ESBL, all had a three or more twofold concentration decrease in MIC for cefotaxime and ceftazidime in combination with clavulanic acid than the MIC when tested alone ( Table 2). Other non-b-lactam resistances observed were to sulfisoxazole (15/18), tetracycline (14/18), chloramphenicol (3/18) streptomycin (9/18), nalidixic acid (1/18), and trimethoprim/sulfamethoxazole (4/18).

Conjugation and plasmid typing
At least two plasmid types were detected using Plas-midFinder from each of the bla CTX-M + E. coli isolates (Fig. 1). The conjugation results showed that bla CTX-M genes can be transferred by broth mating. The transfer of bla CTX-M gene was confirmed by PCR. We further confirmed the plasmid replicon type of the transconjugant with the ESBL phenotype using PCR-based replicon typing. Based on the replicon typing, the two conjugative plasmid types that carried the CTX-M genes were IncI1 and IncF, present in isolates from chicken breast and cattle feces, respectively. One of the bla CTX-M genes identified from cattle isolates (bla CTX-M-32 ) was not transferable by conjugation. Other resistance phenotypes co-transferred by conjugation include Tet R and Smx R (Table 1).

Resistome analysis in ESBL E. coli isolates
Characterization of resistance genes was conducted using whole-genome sequencing. The distribution of resistance genes is shown in Fig. 2    Three of the four bla CTX-M-27 + cattle E. coli isolates clustered together and had identical MLST type (ST1508) (Fig. 2).
E. coli isolates recovered from ground turkey carried diverse CTX-M genes, including bla CTX-M-1 (n = 1), bla CTX-M-15 (n = 1), and bla CTX-M-27 (n = 4). A single E. coli isolate each from pork chops and ground beef carried bla CTX-M-14 and bla CTX-M-15 , respectively. The primary mechanisms responsible for the acquisition and mobilization of CTX-M genes are insertions sequences, transposons, and ISCR1. In our isolates, we identified ISEcp1, IS26, and IS903-D mobilization elements (Fig. 3). In chicken breast isolates, bla CTX-M-1 gene was present as part of an ISEcp1 transposition unit and shares sequence similarity with the backbone structure of the IncI plasmid. Conjugation results demonstrated that tet and sul resistance genes were carried on the same IncI plasmids harboring bla CTX-M-1 gene.
Two chloramphenicol-resistant isolates carried the floR gene and one contained catA1. All, but three E. coli isolates were resistant to sulfisoxazole and had either sul1 or sul2. Four also carried the dihydrofolate reductase gene dfrA.

Discussion
CTX-M-producing strains appear to have quickly spread worldwide, with the notable exception of the United States where TEM-and SHV-type ESBL have appeared to predominate until recently. CTX-M ESBLs have been reported in the United States mainly from human clinical isolates of Enterobacteriaceae encoding for CTX-M group 1 and 9. 7,10,16,25,26 Infections caused by bacteria producing CTX-M enzymes are not limited to the hospital setting. 27 Intestinal carriage of CTX-M-producing bacteria in food-producing animals and contamination of retail meat may contribute to increased incidences of infections with ESBL-producing bacteria in humans. Various reports have documented dissemination of ESBL-producing E. coli in healthy food-producing animals and animal products in several countries, 12,13,17,28,29 and the potential of wild birds as possible reservoirs and vehicles for dissemination of CTX-Ms in the United States. 15 In this study, we are reporting the first bla CTX-M-14 and bla CTX-M-15 gene carrying E. coli isolate from NARMS retail meat program. All E. coli isolates obtained before 2011 from NARMS were phenotypically and genotypically negative for bla CTX-M . 30 However, McDermott et al. recently identified the first bla CTX-M-1 -positive Salmonella isolate recovered from NARMS retail meat samples in the United States. 18 Salmonella enterica serovar Infantis isolates containing bla CTX-M-65 obtained from chicken, cattle, and human sources collected between 2012 and 2015 in the United States through routine NARMS surveillance have been reported. 19 Davis et al., reported bla CTX-M -carrying E. coli strains among isolates collected from Washington State cattle in 2011, while none from those collected in 2008. 13 Investigations of nontyphoidal Salmonella isolates of human origin submitted to Center for Disease Control and Prevention (CDC) as part of the NARMS program between 2005 and 2007 identified Salmonella isolates producing CTX-M enzymes (bla CTX-M-15 , bla CTX-M-5 , and bla CTX-M-55/57 ). 31,32 The successful spread of CTX-M genes depends on the clonal nature of strains carrying the resistance genes, and mobile genetic elements responsible for its capture and spread. E. coli ST131 and ST405 are by far the most important sequence types (STs) associated with the sudden worldwide increase of CTX-M genes, including bla CTX-M-15 33 and other CTX-M genes such as bla CTX-M-1 , bla CTX-M-3 , bla CTX-M-10 , and bla CTX-M-14 . 26 None of our bla CTX-M + E. coli belonged to ST131 or ST405, indicating that the spread of CTX-M genes is not associated with the established clonal strains. We observed diverse STs carrying the same bla CTX-M gene, and in some instances, different bla CTX-M genes carried by the same ST. For example, E. coli isolates recovered from two ground turkey isolates carrying bla CTX-M-27 and two chicken breast isolates carrying bla CTX-M-1 were ST117, indicating the potential of the same ST to spread multiple CTX-M genes.
The most commonly reported mobilization elements mediating the spread of CTX-M genes include ISEcp1, ISCR1, and IS26 34-36 and phage-related sequences. 37 In our isolates, different IS elements, including ISEcp1, IS26, and IS903-D were identified upstream of bla CTX-M-1 , bla CTX-M-27 , and bla CTX-M-32 genes as previously reported elsewhere. [38][39][40] All phenotypically positive ESBL E. coli isolates recovered from chicken breast encoded bla CTX-M-1 as part of an ISEcp1 transposition unit and shares sequence similarity with the backbone structure of the IncI plasmid. IncI1 has been shown to be one of the main plasmid lineages that contribute to the dissemination of bla CTX-M-1 genes in the food chain, including chicken retail meat, the environment, and humans. 41 In the Netherlands, bacteria producing ESBL isolated from chicken meat and gut of broilers predominantly carried bla CTX-M-1 located on IncI1 plasmids. 42 Similarly, Day et al. demonstrated a widespread distribution of IncI1 plasmids carrying bla CTX-M-1 gene among E. coli recovered from humans, animals, and food products in Germany, the Netherlands, and the United Kingdom. 43 Furthermore, a study in the Netherlands revealed the presence of indistinguishable genotypes, CTX-M genes and plasmids, in E. coli obtained from poultry, retail chicken meat, and human clinical samples, suggesting possible exchange through food chain. 12 The bla CTX-M-27 genes identified in our cattle and ground turkey E. coli isolates were associated with IncF plasmids. Horizontal transfer is important in the dissemination of bla CTX-M-27 gene, as evidenced by the fact that bla CTX-M-27 genes in our isolates were transferred by conjugation, confirming the location of the gene on a conjugative plasmid. IncF plasmids encode numerous addiction systems that ensure and contribute to the maintenance of antimicrobial resistance determinants and virulence factors even in the absence of antibiotic selection pressure. 44 IncF replicon-type plasmids carrying bla CTX-M-27 have been documented in cefotaximaseproducing E. coli clinical isolates from Dublin, Ireland. 45 Plasmids carrying bla CTX-M genes are often self-conjugative and carry additional resistance determinants, 46 greatly facilitating widespread distribution of alleles in different environments. A recent report has documented a case of ceftriaxone treatment failure caused by Salmonella Typhimurium due to the in vivo acquisition of a bla CTX-M-27 -encoding IncFII group transmissible plasmid. 47 The presence of bla CTX-M-15 and bla CTX-M-14 was reported in most U.S. medical centers participating in the Meropenem Yearly Susceptibility Test Information Collection (MYSTIC) program in 2007. 10,25 In this study, we are reporting for the first time bla CTX-M-15 -and bla CTX-M-14encoding E. coli from retail meat samples collected in 2013. CTX-M-15 and CTX-M-14, the two most frequently identified CTX-M enzyme worldwide, have been detected in bacteria isolated from humans, animals, and the environment. 17,42,48 A recent study from six community hospitals in North Carolina and Virginia from 2010 to 2012 demonstrated that 80% of ESBL-producing isolates contained CTX-M enzymes. In these isolates, ST131 was associated with 48% of bla CTX-M-15 -producing E. coli isolates and 66% of the bla CTX-M-14 -producing E. coli isolates. 49 While the prevalence of these two successful CTX-M enzymes is low from domestic food animal sources, monitoring will continue to help determine whether this mechanism is becoming more widespread among animal and food strains of E. coli in the United States.