Abstract
Extended-spectrum cephalosporin (ESC) resistance remains a threat since ESC are important antimicrobials used to treat infections in humans and animals. Escherichia coli is an important source of ESC-resistance genes, such as those encoding extended-spectrum β-lactamases (ESBLs). E. coli is a common commensal of lambs. Reports that contaminated food can be a source of ESC-resistant bacteria in humans and that ESBL-producing E. coli are found in sheep in Brazil led us to survey their presence in retail lamb meat. Twenty-five samples intended for human consumption were screened for ESC-resistant E. coli, and the isolates were characterized. IncI1-blaCTX-M-8 and IncHI2-blaCTX-M-2 were the main plasmids responsible for ESC resistance. The plasmids harbored common ESBL genes in Enterobacteriaceae from food-producing animals in Brazil. IncI1-blaCTX-M-14 and IncF-blaCTX-M-55 plasmids, associated with human infections, were also detected. Few CTX-M-producing E. coli have been clustered by typing methods, and some may be genetically pathogenic. The findings indicate the presence of diverse strains of E. coli, harboring important ESBL genes, in lamb meat in Brazil. Surveillance of ESC-resistant bacteria could reduce the spread of antimicrobial resistance through the food chain.
Introduction
Antimicrobial-resistant (AR) pathogens compromise the treatment of bacterial infections, especially, in humans.1–5 The potential link between the misuse of antimicrobials in food-producing animals and the emergence of resistant pathogens in humans has encouraged surveillance of AR worldwide. 6 Human exposure to AR Escherichia coli through handling and consumption of contaminated food may contribute to the spread of AR determinants.2,7,8 During the slaughter of food-producing animals, the carcass may be contaminated by AR commensal bacteria, which might reach humans through the food chain.7,9–12
Monitoring of commensal bacteria is important because they constitute reservoirs of AR genes, which allows tracking of emerging resistance in livestock and possible spread through animal-derived food. 13 These microorganisms are a public health concern since they are disseminated globally and cause treatment failure.14,15
Extended-spectrum cephalosporins (ESCs) are among the few available therapies for serious global Enterobacteriaceae infections in both humans and animals. The use of ESC can cause selection of resistant bacteria in animals. Thus, ESC have the highest priority and are critically important antimicrobials in human and veterinary medicine.16,17
Treatment with β-lactam antimicrobials select for bacteria that produce extended-spectrum β-lactamase (ESBL) and/or bacteria that are resistant to other antimicrobials due to coselection.18–22 CTX-M enzymes are the most prevalent ESBL in enterobacterial representatives isolated from clinical samples, food-producing animals, and the environment in Brazil.23–28 These enzymes, promote an important resistance phenotype to ESC. However, evidence from several studies argues against the role of animals in the dissemination of AR bacteria from food-production chains to humans.29,30 The resulting ambiguity surrounding the involvement of animals in this dissemination demands further study. Determining whether meat consumed by humans harbors AR bacteria is vital because foodstuff is the final and direct link between food-producing animals and humans.
In general, the majority of studies on ESBL-producing bacteria recovered from meat have been conducted on beef, pork, and chicken meat. Information for other meat, such as lamb meat, is lacking in countries including Brazil, where the consumption of lamb meat is increasing. 31 To provide clarity, we have been studying AR dissemination through feedlot lambs in the southeastern part of Brazil. We recently reported ESC-resistant E. coli as a commensal of these animals. 32 That study and another one 33 led us to continue tracking and comparing resistance to ESC in the lamb meat production chain.
Here, we present data concerning the characterization of CTX-M-producing E. coli isolated from retail lamb meat produced and sold in the State of São Paulo, Brazil, and the emergence of blaCTX-M variants of human clinical importance.
Materials and Methods
Meat samples and identification of isolates
We investigated the presence of ESC-producing E. coli in 25 lamb meat samples obtained between October 2018 and January 2019 from traditional marketplaces in the State of São Paulo in southeastern Brazil. All meat samples were obtained from Brazilian sheep herds as packed lamb shank and were immediately transported to the laboratory for microbiological analysis. Since this study analyzed meat samples, no Ethic Committee approval as needed.
For bacterial isolation, subsamples of 25 g were dispensed in sterile flasks containing 225 mL Peptone Water (Oxoid, United Kingdom) with ceftiofur (4 mg/L), and incubated at 37°C for 18–20 hr. Serial dilutions made in sterile saline were inoculated onto MacConkey agar (Oxoid) with ceftiofur (4 mg/L), and incubated at 37°C for 18–20 hr. One colony of each morphology suggestive of E. coli from each meat sample was used for by biochemical identification using a commercial kit (NewProv, Brazil).
Antimicrobial susceptibility testing
Antimicrobial susceptibility tests and phenotypic detection of ESBL production were performed using the disc-diffusion method with commercially available discs (Oxoid) according to Clinical and Laboratory Standard Institute guidelines.34,35 The antimicrobials tested were amoxicillin/clavulanic acid, ceftazidime, cefotaxime, ceftiofur, cefoxitin, ertapenem, amikacin, gentamicin, enrofloxacin, nalidixic acid, tetracycline, trimethoprim/sulfamethoxazole, florfenicol, and fosfomycin. The minimum inhibitory concentration for polymyxin was determined using the Policimbac Commercial Kit (Probac, Brazil). Escherichia coli ATCC 25922, Klebsiella pneumoniae ATCC 700603, and Pseudomonas aeruginosa ATCC 27853 were used as quality control strains.
ESBL genes and plasmid identification
Genes encoding ESBLs were screened by PCR as previously described (Supplementary Table S1). The detected genes were sequenced using the BigDye Terminator v3.1 Kit (Applied Biosystems) and the 3130 Genetic Analyzer sequencer (Applied Biosystems-HITACHI) using specific primers (Supplementary Table S1). Raw sequences were aligned and analyzed using Accelrys DS Gene 2.0 software. The aligned sequences were compared to those in the GenBank database.
ESBL-harboring plasmids were transferred by broth-mating conjugation with the Salmonella M1744F- strain, which naturally lacks AmpC and is susceptible to cephalosporins. 36 Plasmid incompatibility groups were detected using the commercially available PBRT 2.0 kit (Diatheva, Italy). Transconjugants were used for plasmid linearization with S1 enzyme (Promega) followed by pulsed-field gel electrophoresis (PFGE) for 20 hr with an initial switch time of 1 sec and a final switch time of 30 sec, and an electric field of 6 V/cm. PFGE resolved that bands were transferred to a nylon membrane (Amersham, England) by capillary Southern blotting followed by hybridization (Roche, Swiss) with specific probes for CTX-M genes or the plasmid incompatibility groups detected.
Phylogenetic grouping and virulence genes
The phylogenetic group of isolates was accessed using a previously described protocol (Supplementary Table S1). Thirty-one genes codifying for virulence factors often related to extraintestinal infections or diarrheagenic E. coli pathovars were screened by PCR according to previously described protocols (Supplementary Table S1).
Typing of isolates
Isolates were typed by total DNA restriction with XbaI endonuclease (Thermo Fisher Scientific) followed by PFGE for 22 hr with an initial switch time of 2.2 sec and a final switch time of 54.2 sec, and an electric field of 6 V/cm. BioNumerics™ 7.6 (Applied Maths, Belgium) was used for dendrogram construction and clustering, based on the band-based Dice's similarity coefficient and the unweighted pair group method using arithmetic averages. Isolates were considered to belong to the same cluster when the similarity coefficient was ≥90%.
Isolates were also typed by multilocus sequence typing according to the Achtman scheme.
Sequence submission
The sequences obtained in this study were deposited in GenBank under the accession numbers MN816267–MN816292.
Results and Discussion
Meat samples and antimicrobial susceptibility
Twenty-six isolates were recovered from 15 lamb meat samples (60.0%). There were some samples from which more than one E. coli isolate was obtained, and there were some samples from which only one E. coli isolate was recovered. The source and date of isolation of each E. coli are shown in Fig. 1. All positive meat samples were inspected by federal service, which indicates that despite inspection for hygiene indicator microorganisms needed for international trade marketing, AR bacteria can be successfully transmitted by this kind of food. Several studies have reported a high percentage of raw meat contamination with ESC-resistant bacteria, especially in retail chicken meat.25,37–40 To our knowledge, the presence of AR-E. coli in lamb retail meat has not been previously reported.

Dendrogram constructed according to XbaI-pulsed-field gel electrophoresis typing of extended-spectrum β-lactamase-producing Escherichia coli isolated from retail lamb meat in southeastern Brazil. Antimicrobial susceptibility profile symbols are as follows: black squares, resistance; dark gray squares, intermediate resistance; and light gray squares, susceptibility to each antimicrobial. The bla genetic localization presents an approximation of the plasmids size in parenthesis. M, meat sample number; AMC, amoxicillin/clavulanic acid; CAZ, ceftazidime; CTX, cefotaxime; EFT, ceftiofur; AK, amikacin; CN, gentamicin; ENR, enrofloxacin; NA, nalidixic acid; TE, tetracycline; SXT, trimethoprim/sulfametoxazol; FFC, florfenicol; PhG, phylogenetic group; ST/CC, sequence type/clonal complex.
Regarding antimicrobial susceptibility (Fig. 1), resistance to at least one of the third-generation cephalosporins tested was detected in 22 (84.6%) isolates. The remaining four isolates presented only intermediate resistance to cefotaxime or ceftiofur. One isolate was resistant to ceftazidime. One isolate was resistant to amoxicillin/clavulanic acid but not to cefoxitin. Concerning the other classes of antimicrobials tested, 21 (80.8%) of the ESC-resistant E. coli isolates were resistant to tetracycline, 18 (69.2%) to trimethoprim/sulfamethoxazole, 15 (57.7%) to nalidixic acid (11 were also resistant to enrofloxacin), 6 (23.1%) to gentamicin and/or amikacin, and 5 (19.2%) to florfenicol. None of the isolates was resistant to cefoxitin, ertapenem, fosfomycin, or polymyxin B (data not shown). This is probably because these antimicrobials are not often used in sheep.
Multidrug-resistant (MDR) Enterobacteriaceae are considered to have resistance to at least one antimicrobial of three different classes. 41 By this criterion, 73.1% of the isolates were considered MDR. This finding is a concern because these bacteria could be a reservoir of such resistance genes and/or transfer them to other bacteria.
Recent studies evaluated AR in E. coli isolated from feces of feedlot lambs in Brazil.32,33 The reported rates of resistance to the tested antimicrobials differed from the rate found in the present study. This could reflect the different origins of the AR E. coli populations (feces) in the prior studies. A study by our group revealed a significant discrepancy between E. coli isolated from live chickens and chicken meat. 37 The finding means that at least some of the AR bacteria were introduced into the food chain through the handling of meat, and only a few representatives could occasionally be transferred from the gut of the animals, contaminate the carcass, and reach humans via the food chain.
Resistance genes and localization
The 26 isolates were ESBL producers. All harbored a variant of the blaCTX-M gene (Fig. 1). Half of the E. coli isolates carried blaCTX-M-8 (13/26, 50.0%), followed by blaCTX-M-2 (10/26, 38.5%). Two isolates (7.7%) harbored blaCTX-M-55, and one (3.8%) harbored blaCTX-M-14. All the blaCTX-M-8 genes were present in IncI1 plasmids ranging from 73 to 97 kb, while blaCTX-M-2 was harbored by IncHI2 plasmids ranging 242.5 to 290 kb or IncI1 (∼97 kb). The blaCTX-M-14 gene was also detected in an IncI1 plasmid ∼97 kb. One blaCTX-M-55 was detected in a small (∼29 kb) IncF plasmid. We could not detect the genetic localization of the second blaCTX-M-55 (Fig. 1).
The presence of both blaCTX-M-8 and blaCTX-M-2 was expected because these genes are common in bacteria isolated from humans and food-producing animals in Brazil.23,26,37,42–46 blaCTX-M-2 is often detected in the chromosome,37,43,47,48 but few studies have reported its mobilization by IncHI2 plasmids.49–52 On the contrary, blaCTX-M-8 is commonly detected in IncI1 plasmids,28,37,53–56 which agrees with our results. The findings corroborate IncI1 as a consistent plasmid in bacteria of animal origin. 57
The blaCTX-M-14 gene has not been detected often since its first description in Brazil. 58 However, some studies have reported this gene in Enterobacteriaceae isolated from humans and food-animals in the country.32,46,59 The presence of this gene in food-producing animals and subsequently in foodstuffs is a public health concern because it is still a dominant worldwide ESBL-variant in important infections, especially in Asian countries.60–63 Recent studies support that concern. These studies described bacteria from human infections harboring IncI1-blaCTX-M-14 or IncHI2-blaCTX-M-1464,65 and E. coli from retail meat carrying IncF-blaCTX-M-14.63,66 Several reports have identified this gene harbored by different incompatibility groups in Enterobacteriaceae members isolated from food-producing animals or clinical specimens in China, such as the IncF family, IncK, IncI1, IncHI2, and even nontypeable plasmids.67–70 In our study, blaCTX-M-14 was carried by an IncI1 plasmid of ∼97 kb, in accordance with Chinese reports.
Finally, the presence of blaCTX-M-55 was remarkable since it was only recently identified in Brazil in few studies,37,71,72 a common gene among E. coli isolated from animals in Asian countries. 73 To our knowledge, this is the first report of blaCTX-M-55 in E. coli recovered from food products in Brazil. The aforementioned reports from Brazil associated this gene with 70–100 kb IncF plasmids. However, other plasmids reported globally are also responsible for the mobilization of blaCTX-M-55.74,75 In our study, the blaCTX-M-55 gene was harbored by a small IncF plasmid (Fig. 1). This was different from the others detected in the country. One study reported the insertion of the blaCTX-M-55 gene into the chromosome of Salmonella recovered from food-animals in China. 76 This could be the explanation for why we could not determine the genetic localization or conjugate the blaCTX-M-55.
E. coli phylogroups and virulence genes
E. coli phylogenetic group B1 was the most detected (n = 15, 57.7%), followed by phylogroups D (6/26, 23.1%) and A (5/26, 19.2%) (Fig. 1). Thus, a majority (76.9%) of the isolates were considered as commensal or not potentially pathogenic because they belonged to phylogroups B1 or A. 77
The results agree with those obtained for virulence genes, in which only 7 of 31 screened genes were detected. Genes codifying for fimbrial adhesins were the most common, with fimH present in 13 isolates (50.0%) and papEF in 6 (23.1%). Genes responsible for siderophores, polysaccharide capsule, and toxin production were also detected. The iutA and iroN genes were present in eight (30.8%) and four (15.4%) isolates, respectively. The kpsMT II gene for capsule serotype K5 was detected in three (11.5%) isolates. The vat gene was detected in one (3.8%) isolate (Fig. 1).
With the exception of vat, the other genes could be considered to function in the adherence and persistence of commensal bacteria in a niche like the gastrointestinal tract of mammals. Thus, they would not confer appreciable pathogenicity. The presence of vat must be considered a concern since the cytotoxin it encodes induces host cell vacuolization. 78 Isolate LM5, which belongs to the phylogenetic group D, could be classified as a potentially pathogenic ESBL-producing E. coli. In addition, the LT gene, responsible for heat-labile enterotoxin of enterotoxigenic E. coli pathovar, 79 was detected in 14 (53.8%) isolates (Fig. 1). This is further evidence of the possibility of pathogenicity for the CTX-M-producing isolates recovered from retail lamb meat. Additional studies are required to verify the expression of genes and phenotypic pathogenicity of these isolates.
Typing of isolates
XbaI-PFGE typing revealed an overall dissimilarity among the studied E. coli, with the exception of nine (34.6%) isolates that clustered into four distinct groups (Fig. 1). Interestingly, the clustered E. coli were recovered from different meat samples.
The observation that some were isolated from samples purchased on different days negates the idea of cross-contamination of samples by the laboratory staff during meat processing. Furthermore, some clusters were formed by isolates with important different traits, such as the blaCTX-M gene or the sequence type (ST) lineage. However, despite the examination of different meat batches, all commercially available packages were prepared from animals slaughtered at the same slaughterhouse. This suggests a focus on CTX-M-producing E. coli dissemination. In addition, this could indicate the use of insufficient hygiene methods that allowed contamination of different meat products with bacteria.
Mistakes in carcass preparation can make the product a potential carrier for AR bacteria, since bacteria in the gastrointestinal tract of sheep or in the abattoir environment could contaminate the meat during handling.80–85 Considering that the analyzed meats were inspected by federal services, we suggest that inspection of AR bacteria should also be performed in addition to coliforms as determinants of health safe food at the establishments.
Concerning MLST, the following STs/clonal complexes (CC) were identified: ST 23/CC 23, ST 58/CC 155, ST 106/CC 69, ST 117, ST/CC 155/155, ST 205/CC 205, ST 349/CC 349, ST 398/CC 398, ST/CC 448/448, ST/CC 761/10, ST 1169, ST 1204, ST 1290, ST/CC 1308/86, ST 1653, ST 2612, ST 4680, and ST 6228, with up to three isolates per ST/CC (Fig. 1). These data agree with the XbaI-PFGE results. The considerable distinction among the AR isolates supports the idea that the blaCTX-M genes detected are spread by plasmids instead of specific strains or clones. This idea is also corroborated by the common plasmids detected in this study that can mobilize the blaCTX-M-2 and blaCTX-M-8 genes.
The lineage E. coli ST 58/CC 155 was previously reported as a commensal in feedlot lambs by our group. 32 The lineage is also present in humans and poultry in Brazil, according to the EnteroBase website. It is interesting to note that most of the STs detected are already described as pathogenic and nonpathogenic to humans and animals in several countries.
Thus, instead of presenting a general low pathogenic level, several lineages found in lamb meat in Brazil have already been associated with hard-to-treat human infections. This is particularly possible for the LM5 phylogroup D-ST 117 isolate, which harbors the vat and LT genes. In addition, isolates LM9 and LM16 were positive for the LT gene. They belong to phylogenetic group D and ST/CC 349/349 and have been reported as the cause of infections associated with extraintestinal pathogenic E. coli, according to the EnteroBase website.
Other isolates belonging to phylogenetic group D, such as LM11, LM25, and LM34, are also representatives of important ST lineages, such as ST/CC 349/349, ST 2612, and ST/CC 106/69, respectively. The isolates from this study were a clear source and reservoir of ESC-resistance genes. The presence of genes frequently detected in bacteria isolated from humans and animals in other regions, such as blaCTX-M-14 and blaCTX-M-55, corroborates the significance of this spread.
In conclusion, despite the small amount of lamb meat that was sampled, it is a source of potentially pathogenic MDR-E. coli. These bacteria carry the same blaCTX-M genes in plasmids that are present in food-producing animals or humans. The data indicate the emergence of blaCTX-M-55 in bacteria present in food in Brazil. The findings bring up issues concerning the global spread of antimicrobial resistance from the One Health perspective.
Footnotes
Acknowledgments
We are very grateful to Juliana Rodrigues Froes, Luciana Moran Conceição, and Valéria D'Artibale Fraga for technical support. We thank Editage for English language editing.
Disclosure Statement
No competing financial interests exist.
Funding Information
This work was supported by the São Paulo Research Foundation (FAPESP) (grant numbers #2018/02691-4, #2018/16147-4, #2018/16343-8, #2019/16003-5) and the Brazilian Coordenação de Aperfeiçoamento de Pessoal de Nível Superior.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
