Genomic insights of high-risk clones of ESBL-producing Escherichia coli isolated from community infections and commercial meat in southern Brazil

During a microbiological and genomic surveillance study conducted to investigate the molecular epidemiology of extended-spectrum β-lactamase (ESBL)-producing Escherichia coli from community-acquired urinary tract infections (UTI) and commercial meat samples, in a Brazilian city with a high occurrence of infections by ESBL-producing bacteria, we have identified the presence of CTX-M (-2, -14, -15, -24, -27 and -55)-producing E. coli of international clones ST38, ST117, ST131 and ST354. The ST131 was more prevalent in human samples, and worryingly the high-risk ST131-C1-M27 was identified in human infections for the first time. We also detected CTX-M-55-producing E. coli ST117 from meat samples (i.e., chicken and pork) and human infections. Moreover, the clinically relevant CTX-M-24-positive E. coli ST354 clone was detected for the first time in human samples. In summary, our results highlight a potential of commercialized meat as a reservoir of high-priority E. coli lineages in the community, whereas the identification of E. coli ST131-C1-M27 indicates that novel pandemic clones have emerged in Brazil, constituting a public health issue.

Escherichia coli is a commensal of the human intestinal tract and most warm-blooded mammals, and an important pathogen for humans and animals [1][2][3] . In humans, urinary tract infections (UTIs) are the second most common bacterial infection managed in primary care, where uropathogenic E. coli (UPEC) is responsible for 75% to 95% of the cases 1,4 . The increasing antimicrobial resistance of UPEC has been of concern, with infections caused by antimicrobial-resistant (AMR) bacteria as extended-spectrum β-lactamase (ESBL)-producing E. coli representing a significant health care issue, since it compromises the effective treatment, being responsible for a large number of morbidity and mortality 1,3,5 .
Since E. coli can act as a large reservoir of resistance genes that directly impact treatment in human and veterinary medicine, the debate over the transmission of multidrug-resistant E. coli strains between animals and humans, through numerous pathways, has become increasingly discussed 6,7 . However, the relation between food-producing animals, humans, the environment, and the transmission of these resistant pathogens is not yet fully understood 8,9 .
In this study, 11 types of plasmid replicons were predicted, which belonged to the Col, IncA/C2, IncB/O/ K/Z, IncF, IncI, IncQ, IncR, IncY, IncX, p0111 and IncN families. In human isolates, the most frequent were  In total, 40 sequence types (STs) were identified, the most prevalent were ST131 (n = 13), ST38 (n = 8), ST648 (n = 7), and ST354 (n = 6). Some STs were detected in strains isolated from more than one source, confirming clonal dissemination between humans and chicken meat. Lineages of ST38, ST131, ST354, and ST1196 were found in both urine and chicken meat. The ST410 was the only observed in urine (n = 1) and pork (n = 1) strains. The ST117 was identified in samples collected from the three sources investigated, with two strains being isolated from urine, and one from chicken meat and pork, respectively. The clonal relatedness among isolates characterized in this study, and their distribution in Brazil, is shown in Fig. 1A,B. Additionally, it is possible to observe that Brazilian isolates of ST131 clustered with isolates from United Kingdom (https:// micro react. org/ proje ct/ 2mKg5 4AHdW j5xdJ 5VFej Y8). All information and genes detected in this study are quoted in supplementary material ( Phylogenetic analysis revealed genomic diversity among isolates. In fact, most of our isolates of the same ST grouped within different clades, closer to isolates from several different countries. For ST38, isolates grouped within two different clades, both mainly composed by strains of human sources (Fig. 2). While two chicken isolates grouped together within a clade with only strains of poultry sources, the other grouped within a clade mainly composed by strains from livestock and wild animal origin. For ST117, most of the isolates were from poultry origin, including closely related strains from human host (Fig. 3). For ST131, while the strain from chicken grouped with strains from various animal sources, human isolates grouped mainly with other human www.nature.com/scientificreports/ isolates, but isolates were relatively distant to each other, being closer to strains from several other countries (Fig. 4). For the ST354, three of human isolates grouped together within a clade shared only with human isolates, whereas other human isolates clustered into a clade, mainly composed by poultry strains. The isolate from food grouped within a clade with human, poultry, environment and food isolates (Fig. 5). For ST410, the human isolate grouped within a clade shared mainly with strains from human sources, and the isolate from swine grouped within a more diverse clade (Fig. 6). For ST648, human isolates grouped within three different clades, all of them composed mainly by isolates from human origin (Fig. 7).

Discussion
This study presents the first reports of E. coli ST131-C1-M27 in human infection and CTX-M-24-positive E. coli ST354 from UTI, in Brazil. In this country CTX-M-producing E. coli are endemic. Our data show a wide distribution of these isolates belonging to international clones in livestock and community patients. The extensive presence of CTX-M enzyme-producing strains in several sources raises the hypothesis that the spread occurs with greater frequency and efficiency, especially among Enterobacterales 10 . E. coli ST131 is globally disseminated and related to the spread of resistance genes, including specific CTX-M variants 17 . Recent studies have shown that ST131 is rare among animal isolates, becoming almost exclusively a human pathogen, as demonstrated by our results, where ST131 was predominantly found a human lineage isolated from urine 18 . The bla CTX-M-15 gene is frequently found in E. coli ST131 UTI and is strongly related to the C2 subclade 11 . Different groups of plasmids can carry the bla CTX-M-15 gene, here we observe that this gene is involved with the incompatibility group IncF. In a previous study was shown that clade C was related to highest rates of UTI, with subclade C2 being the most common and associated with incompatibility group IncFII 19 . Besides, CTX-M-15-producing E. coli ST131 has already been shown to be involved in outbreaks in health institutions and is the most prevalent ESBL-producing E. coli worldwide 20 .
The CTX-M-27-producing ST131-C1 has been considered a new epidemic clone, and it has not been reported in human infections so far, in Brazil. Clade C1-M27 is associated with CTX-M-27 and was first observed colonizing children in France in 2012. Birgy et al. (2019) 21 suggests that the C1-27 subclade was recently emerged, because it was observed that the difference in SNPs between the C1 subclade isolates was smaller compared to the SNPs difference between the C2 subclade and the A clade isolates. In addition, the plasmid predominantly Although the first one is widely distributed, especially in China, South-East Asia, Japan, South Korea, and Spain, microorganisms producing CTX-M-24 remain relatively rare, being reported with greater incidence in countries such as Peru and Bolivia 19,23,24 . This study found an association between CTX-M-24 and E. coli ST354 detected in two human isolates, never before reported in UTI in Brazil. Strains of ST354 isolates were positive to bla CTX-M-24 and displayed further resistance to ciprofloxacin, being associated with extra-intestinal infections, animals and humans, reinforcing the zooanthroponotic hypothesis of these clones 25 23 . In the last ten years, IncI1-type plasmids have had a high spread, mainly in animal reservoirs. There are reports of bla CTX-M-2 , bla CTX-M-8 and bla CTX-M-55 genes frequently found on IncI plasmids from E. coli isolated from chickens and pigs several countries, such as China, France, the United States of America, and the United Kingdom [27][28][29] .
The international clone ST117 was the only ST found in all our three sources studied. This clone is often associated with AMR and found in animals and human isolates, having a zoonotic profile [30][31][32] . E. coli ST117 isolates have been identified in pigs, with the bla CTX-M-55 gene being carried by IncN-types plasmids. These data found in our study raises concern, as they may indicate a possible new epidemic transmission of CTX-M-55 in Brazil 31,33 .
ST38 was found in chicken meat and urine samples. Considered a minority among ESBL-producing E. coli in Europe, Middle East and Asia, the ST38 E. coli has been previously associated with bla CTX-M-2 , bla CTX-M-9 and bla CTX-M-14 genes, but rarely with the bla CTX-M-27 gene 34,35 . The bla CTX-M-27 gene has been associated with IncF plasmids, with the most commonly identified being IncF[F-:A-:B33] 35 . Interestingly, in our study a ST38 E. coli isolate from human urine carried the bla CTX-M-27 and had plasmid IncF[F2:A-:B10]. Furthermore, recent reports linked CTX-M-types producing E. coli ST38 to UTI 36 . These data may indicate an increased frequency with which ST38 isolates are being found, being a prominent ESBL lineage worldwide [34][35][36] . www.nature.com/scientificreports/ In summary, E. coli carrying bla CTX-M genes from different sources seem to be related to the spread of internationally known clones (ST38, ST117, ST131, ST354, ST410, ST648). Some clones associated with CTX-M variants are more prevalent in some sources than others, which does not exclude the possibility that new clones are entering and establishing themselves in different niches. In fact, we detected E. coli ST131-C1-M27 in human infections, and CTX-M-24 positive E. coli ST354 from UTI, which could be a public health issue. Furthermore, the presence of CTX-M (-55, -27, -24, -15, -14 and -2)-producing E. coli among ST38, ST117, ST131 and ST354 clones from humans and meat for consumption reinforces the need for additional surveillance studies, in order to study dynamics of dissemination of ESBL-producing E. coli clones at the human-animal-food-environmental interface. As screening criteria for performing the ERIC-PCR, we selected 1,389 E. coli isolates from human urine that presented ESBL production and/or were resistant to quinolones, previously identified by the automated VITEK Ⓡ 2 system.

Material and methods
Through the analysis of the dendrogram constructed by the ERIC-PCR technique, we selected 59 E. coli isolates from human urine that presented an interesting genetic similarity profile to perform the complete genome sequencing.
All participants provided the written informed consent for this study and the study was approved by the Ethics and Research Committee of the State University of Londrina by a document numbered as CAAE 56869816.0.0000.5231. Besides, we confirm that all experiments were performed in accordance with ethical regulations.

E. coli isolated from chicken meat and pork samples.
A surveillance study from January to May 2019 was carried out, to research ESBL-producing E. coli, in chicken and pork meat, bought at markets and butcher We performed ERIC-PCR on all 169 E. coli isolates derived from chicken meat and pork that showed ESBL production and/or quinolone resistance. Through dendrogram analysis, we selected 32 E. coli isolates from chicken meat (n = 24) and pork (n = 8) that presented an interesting genetic similarity profile to perform the complete genome sequencing. The samples of chicken meat (n = 50) and pork (n = 50) were dipped in Brain Heart Infusion (BHI) broth (Oxoid) with cefotaxime (4 µg/mL), ciprofloxacin (4 µg/mL), and both (Sigma-Aldrich, Munich, Germany) to selected resistant E. coli strains. After incubation, the solution was inoculated in the same way used for urine samples. We evaluated the growth of E. coli from the three BHI solutions and we preferentially selected the isolates that grew in broth with both antimicrobials (cefotaxime and ciprofloxacin). In case there was no growth of any isolate on that broth, we selected from the broth containing separate cefotaxime and ciprofloxacin. All the isolates selected were stored in Tryptic Soy Broth (TSB) with 15% glycerol (− 20•C).
The identification and bacterial susceptibility were performed by the automated VITEK Ⓡ 2 system, using the VITEK Ⓡ 2 AST 239 card and the VITEK Ⓡ 2 GN ID card (BioMérieux, USA). The bacterial susceptibility was tested for 14 antibiotics: ampicillin, amoxicillin/clavulanate, ceftriaxone, cefepime, ertapenem, meropenem, nalidixic acid, ciprofloxacin, norfloxacin, gentamicin, amikacin, nitrofurantoin, trimethoprim-sulfamethoxazole, and piperacillin-tazobactam. The CLSI 2020 (Clinical and Laboratory Standards Institute) criteria were used for interpretation. E. coli ATCC Ⓡ 25922 strain was used as quality control. DNA isolation and whole-genome sequencing. For DNA extraction, strains were grown on Mueller-Hinton Agar overnight at 37 °C. Subsequently, a single colony was inoculated in 2 mL of Luria-Bertani broth for 12 h at 37 °C. The suspension was used to continue extraction and purification by the DNA extraction kit (Invitrogen, Carlsbad, CA). The extracted DNA was quantified by Qubit dsDNA (double-stranded DNA) BR assay kit (Invitrogen, Carlsbad, CA). After quantification, the DNA was used to construct a paired-end library (150 bp), sequenced using the NextSeq platform (Illumina). The instructions of each manufacturer were followed in all steps. In order to compare our strains belonging to ST38, ST117, ST131, ST354, ST410 and ST648 with other genetically related strains from different sources and countries, we performed a search for each ST on the Enterobase Escherichia/Shigella database 46 (https:// enter obase. warwi ck. ac. uk), then we downloaded all genome assemblies with data for country, source and collection year. Selection of assemblies to phylogenetic analysis was based on fimH type, obtained from Enterobase, and on average nucleotide identity (ANI), assessed with FastANI v1.32 47 (https:// github. com/ ParBL iSS/ FastA NI).