Distribution and Genomic Characterization of Third-Generation Cephalosporin-Resistant Escherichia coli Isolated from a Single Family and Home Environment: A 2-Year Longitudinal Study

Third-generation cephalosporin-resistant Escherichia coli (CREC), particularly strains producing extended-spectrum β-lactamases (ESBLs), are a global concern. Our study aims to longitudinally assemble the genomic characteristics of CREC isolates from fecal samples from an index patient with recurrent CREC-related urinary tract infections and his family and swabs from his home environment 12 times between 2019 and 2021 to investigate the distribution of antibiotic resistance genes. CREC identified using the VITEK 2 were subjected to nanopore whole-genome sequencing (WGS). The WGS of 27 CREC isolates discovered in 137 specimens (1 urine, 123 feces, and 13 environmental) revealed the predominance of ST101 and ST131. Among these sequence types, blaCTX-M (44.4%, n = 12) was the predominant ESBL gene family, with blaCTX-M-14 (n = 6) being the most common. The remaining 15 (55.6%) isolates harbored blaCMY-2 genes and were clonally diverse. All E. coli isolated from the index patient’s initial urine and fecal samples belonged to O25b:H4-B2-ST131 and carried blaCTX-M-14. The results of sequence analysis indicate plasmid-mediated household transmission of blaCMY-2 or blaCTX-M-55. A strong genomic similarity was discovered between fecal ESBL-producing E. coli and uropathogenic strains. Furthermore, blaCMY-2 genes were widely distributed among the CREC isolated from family members and their home environment.


Introduction
Escherichia coli is regarded as a top-ranked pathogen causing public health concerns and is widely distributed in both human and animal intestines. Most strains of E. coli are harmless, but some strains, which have acquired specific virulence attributes, are pathogenic and cause a variety of diseases in humans. Pathogenic E. coli strains are classified into categories based on the production of virulence factors and the site of infection. There are the following six well-described categories: enteroaggregative E. coli (EAEC), enterohemorrhagic E. coli (EHEC), enteroinvasive E. coli (EIEC), enteropathogenic E. coli (EPEC), enterotoxigenic E. coli (ETEC), extraintestinal E. coli (ExPEC), and Shiga toxin-producing E. coli (STEC) based on clinical characteristics and virulence factors [1]. In addition to enrolling CREC isolates from one urine and two fecal samples from the index patient in separate hospitalizations, 121 E. coli isolates were collected from fecal samples and 13 E. coli isolates from environmental surface samples for CREC screening during the study period. In total, 27 of the 137 samples contained the following CREC isolates: 25 fecal samples, 1 urine sample, and 1 environmental surface sample. The three initial CREC isolates (I-01-S, I-02-S, and I-01-U, Table 2) from the index patients, collected during two separate hospitalizations, shared the same antimicrobial susceptibility patterns.

ARGs Other Than β-Lactamase Genes
In addition to the aforementioned resistant β-lactamase genes, other ARGs were identified through WGS (Table 3). No carbapenem resistance gene was identified among the isolates. One isolate with a plasmid-mediated mobile colistin resistance gene, mcr-1.1, was discovered in one of the family members' samples.

Virulence Factors (VF)-Encoding Genes
As shown in Table 4, all 27 CREC isolates harbored a wide variety of VF-encoding genes. Accordingly, 55.6% of the isolates were classified as extraintestinal pathogenic E. coli (ExPEC), 33.3% as uropathogenic E. coli (UPEC), and 40.7% as avian pathogenic E. coli (APEC). In addition, isolates with ST131 were more likely to be classified as ExPEC or UPEC than non-ST131 isolates (p = 0.02 and < 0.001, respectively).    In addition to the aforementioned resistant β-lactamase genes, other ARGs were identified through WGS (Table 3). No carbapenem resistance gene was identified among the isolates. One isolate with a plasmid-mediated mobile colistin resistance gene, mcr-1.1, was discovered in one of the family members' samples.

Virulence Factors (VF)-Encoding Genes
As shown in Table 4, all 27 CREC isolates harbored a wide variety of VF-encoding genes. Accordingly, 55.6% of the isolates were classified as extraintestinal pathogenic E. coli (ExPEC), 33.3% as uropathogenic E. coli (UPEC), and 40.7% as avian pathogenic E. coli (APEC). In addition, isolates with ST131 were more likely to be classified as ExPEC or UPEC than non-ST131 isolates (p = 0.02 and < 0.001, respectively).     a S and U indicate fecal (stool) and urinary origin, respectively; b Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box. Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box.  afaC  afaD  focC  hra  iha  lpfA  papA  papC  sfaD  yfcV  kpsE  kpsM II  kpsM III  chuA  fyuA  iroN  irp2  iucC  iutA  sitA  air  astA  ccI  cea  cib  cma  cnf1  hlyF  pic  sat  senB  tsh  vat  cvaC  eilA  etsC  gad  iss  katP  mcbA  mchF  ompT  terC  traT  Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box.  afaA  afaB  afaC  afaD  focC  hra  iha  lpfA  papA  papC  sfaD  yfcV  kpsE  kpsM II  kpsM III  chuA  fyuA  iroN  irp2  iucC  iutA  sitA  air  astA  ccI  cea  cib  cma  cnf1  hlyF  pic  sat  senB  tsh  vat  cvaC  eilA  etsC  gad  iss  katP  mcbA  mchF  ompT  terC  traT  Genes related to third-generation cephalosporin resistance; c Point mutation of gyrA (S83L). Colored boxes: Presence of predicted ARG. Antimicrobial resistance phenotype and corresponding ARGs are indicated with the same color of box.  afaA  afaB  afaC  afaD  focC  hra  iha  lpfA  papA  papC  sfaD  yfcV  kpsE  kpsM II  kpsM III  chuA  fyuA  iroN  irp2  iucC  iutA  sitA  air  astA  ccI  cea  cib  cma  cnf1  hlyF  pic  sat  senB  tsh  vat  cvaC  eilA  etsC  gad  iss  katP  mcbA  mchF  ompT  terC  traT  ExPEC:≥2 of 5 markers, including papAH and/or papC, sfa/focDE, afa/draBC, kpsM II, and iutA UPEC: ≥3 of 4 markers, including chuA, fyuA, vat, and yfcV APEC: ≥4 of 5 markers, including hlyF, iutA, iroN, iss, and ompT Presence of predicted VF-encoding genes

Discussion
This study investigated fecal and urinary CREC isolates from a patient with recurrent UTIs and longitudinal collection of fecal samples from him and his asymptomatic family as well as samples from their home environment. To compare the genetics of these isolates, WGS was used for ARG and VF-encoding gene detection, phylogenetics, MLST, and serotyping.
Uniplex and multiplex PCR with specific primers were designed to investigate sequence types and characterize the antimicrobial resistance mechanisms of ESBL-carrying E. coli in the communities of southern Taiwan [7][8][9][10]20]. Traditionally, only certain sequence types, including ST69, ST73, ST95, and ST131, which are the major E. coli sequence types that cause community-acquired infection, were detected with target-specific primers [21]; O25b-ST131 was discovered to be the most common ESBL-producing E. coli clonal group in feces in our previous studies [8,20]. However, epidemiologic lineages might be underestimated due to a lack of corresponding primers. In this series, WGS was employed to overcome this limitation. Through nanopore sequencing, we found sequence types that might be neglected by traditional PCR, such as ST101 (25.9%, n = 7), which was even more than the pandemic ST131 (22.2%, n = 6). Similarly, WGS provided insight into the distribution of β-lactamase genes, including not only bla CTX-M (44%), which we used to detect with specific primers in our previous work, but also bla TEM (70.4%), bla AmpC (55.6%), and numerous ARGs.
WGS data revealed that fecal and urinary E. coli isolated from the index patient during the UTI had the following identical genetic features: both carried bla CTX-M-14 and ST131, and belonged to the B2 group, C1-nM27 subclade, and O25b:H4 serotype. This finding might suggest a potential relationship between intestinal colonization and UTIs. The digestive tract is a natural reservoir of E. coli, which may cause community infection and is a melting pot where resistance genes might be exchanged and cause resistant bacteria to rapidly increase under antimicrobial selective pressure. Among ESBL-producing E. coli, O25b:H4-B2-ST131 is a globally dominant clone characterized by simultaneous resistance to several classes of antimicrobials and contains numerous VF-encoding genes. In this study, all the ST131 clones were either ExPEC or UPEC. The combination of MDR and hypervirulence of O25b:H4-B2-ST131 E. coli represents a potential challenge to public health.
As displayed in Figure 4A, we found the identical plasmids with bla CMY-2 genes were mostly from index patients, which persistently exist for over 7 months, and from one environmental surface during the same collection, which might be associated with contaminated waste. Moreover, plasmid-mediated horizontal gene transfer plays a crucial role in the dissemination of antimicrobial resistance [12], which is shown in Figure 4B,C. Plasmid contigs containing bla CMY-2 that were isolated from the index patient and one of his cousins had high sequence similarity (>99% identity), which might suggest horizontal gene transfer or acquisition from common sources. In addition, identical plasmid-mediated bla CTX-M-55 genes were discovered among isolates from the family, which might indicate the same potential phenomena.
According to Reuland et al. (2015), the prevalence of CMY-2-producing E. coli is lower than that of CTX-M-producing strains in human carriers [14]. CMY-2-producing E. coli are prominent in poultry, with a prevalence of 30% among CREC isolates [22]; such strains should not be overlooked because of the possible dissemination of ARGs to human-borne E. coli through plasmid-mediated transmission after food consumption. Only approximately 1% of Europeans harbor E. coli with bla CMY-2 [22][23][24]. However, recent studies, mainly in Asia, have indicated an increasing prevalence of CMY-2-producing E. coli [25]. Yan et al. (2004) discovered the community spread of bla CMY-2 , with isolates harboring the gene found in the feces of food animals, retail ground meat, and the urine of patients with UTIs in Taiwan [26]. According to the results of the 7-year Study for Monitoring Antimicrobial Resistance Trends, the prevalence of CMY-2-producing E. coli, which causes intra-abdominal infection or UTI, had increased to 29.3% in Taiwan by 2014 [25].
In the present study, CMY-2-producing CREC was isolated from several family members, which might suggest long-term fecal colonization, especially for the index patients, and horizontal plasmid transfer between E. coli strains in southern Taiwan. However, the results of ARGs should be cautiously applied to general epidemiologic data due to the small sample size in this series. In addition, few studies have observed nonclonal dissemination of CMY-producing E. coli because the diversity of clonal lineages is high [27]. As shown in Figure 2, our CMY-2-producing E. coli isolates were genetically diverse, and ST101 was the most common (25.9%) among the nine sequence types found in the series.
As we know, AmpC β-lactamase leads to resistance to third-generation cephalosporins, especially to cephamycin [28]. On the basis of antimicrobial stewardship, a previous study proposed cephamycin as a candidate to replace carbapenem for treating ESBL-carrying E. coli infection [29]. The rising prevalence of CMY-2-producing E. coli in feces presents several public health concerns. If harboring bla CMY-2 , the pathogenic E. coli, such as ExPEC or UPEC, might eventually develop cephamycin resistance, which would limit the therapeutic options for MDR isolates, especially when an ESBL-producing strain is present [30]. Moreover, Drinkovic et al. (2015) reported that, although the prevalence of plasmid-mediated AmpC β-lactamase among E. coli was only 0.4% among urine isolates from an Auckland community, it still could affect the UTI treatment options [31].
Carbapenem remains the drug of choice for severe infections caused by ESBL-producing E. coli [32]; this is reasonable given that the susceptibility results of our series revealed no carbapenemase genes in any of the 27 CREC isolates. However, Chia et al. (2009) discovered that the concurrence of bla CMY-2 and porin deficiency led to carbapenem resistance, even in the absence of carbapenemase [33]. Thus, the emergence of isolates with bla CMY-2 genes may potentially suggest a greater distribution of MDR E. coli. Therefore, continued surveillance of both CTX-M-producing and AmpC-producing E. coli and other resistance mechanisms is warranted.
In our previous study, the prevalence of the fecal carriage of mcr-1-positive E. coli was low (2.4%) among community children in southern Taiwan [34]. Similarly, the prevalence in the present study was low, with mcr-1-positive E. coli encountered in only one of 27 CRECs from the 123 E. coli isolates.
This study has some limitations. First, because of the small sample size and the few colonies chosen per fecal sample, this study may not represent the general genomic background of CREC isolates in the feces of individuals in Taiwan. Although we ensured that all the E. coli isolates were chosen based on colony morphology and were validated by MALDI-TOF MS to reduce selection bias, the number of colonies directly influences the ability to accurately characterize E. coli strain diversity [35]. Instead, new approaches such as qPCR or real-time PCR might improve the sensitivity of detection [36] and will be applied to describe the E. coli population structure in future studies. Next, although our results revealed the abundance and distribution of ARGs in a single family, we could not explain the transmission network of ARG-harboring E. coli. Then, the MLST of CREC isolates from the index patient showed a drift from ST131 to ST101 in the study period, but only a few CREC isolates from family members had the same ST101 clone. More bioinformation, from CREC and non-CREC isolates, may be needed for a more comprehensive explanation. Our results address the emergence of CREC isolates producing either CTM-X or CMY-2 β lactamases that might cause difficult-to-treat infection and pose a threat to human health. Finally, we did not investigate the resistance mechanism of intrinsic efflux pumps, which requires more information from transcriptional analysis, overexpression of efflux pumpencoding genes, generation of mutants, for instance, leading to porin loss, etc., which are beyond the scope of our study, and are warranted in future studies.

Participants and Sample Collection
An uncircumcised male infant delivered vaginally at full term without any underlying disease experienced his first UTI episode at the age of 13 days in December 2019 and recurrent infection in April 2020. This infant was enrolled as our index patient during his second hospitalization for his recurrent UTI at Kaohsiung Veterans General Hospital (KVGH) in April 2020. After obtaining consent from all the family members in the same living space to participate in this study, each urine and fecal specimen of the index patient was collected during the two episodes of hospitalization. In addition, fecal samples were collected simultaneously from his family members, and household surfaces 12 times between May 2020 and September 2021 at intervals of 30-60 days. Swabs were also collected from the following household surfaces during the period: toilet rims; bathroom basins; the kitchen sink; living room, master bedroom, children's room, bathroom, and kitchen floors; toys. Drinking water was sampled. The ethics committee of KVGH (approval No. VGHKS19-CT3-20) granted study approval in 2019.

Questionnaire Design
Questionnaires were completed after the first specimen collection. The questionnaire included items on travel history, animal contact, dietary habits, cleaning habits, weekly food consumption, weekly probiotic use, and medical history. Weekly food consumption and probiotic use were quantified using a Likert-type scale.

Microbiological Analysis and Antimicrobial Susceptibility Testing
Each fecal sample was streaked on a CHROMagar ECC plate (CHROMagar, Paris, France) and incubated at 37 • C for 24 h. Up to two E. coli isolates based on colony morphology from each specimen were subsequently validated through matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF) mass spectrometry (MS) and subjected to antimicrobial susceptibility testing with the VITEK 2 automated system and AST-N320 cards (bioMérieux, Marcy-l'Etoile, France). Aside from tigecycline and colistin, the breakpoints of antimicrobials were based on the M100-S30 Clinical and Laboratory Standards Institute (2020) standard [37]. Classifications of 18 antimicrobials are listed in Supplementary Table S2. Minimum inhibitory concentrations of tigecycline were interpreted using the US Food and Drug administration (FDA) tigecycline susceptibility breakpoints for Enterobacteriaceae (https://www.fda.gov/drugs/development-resources/tigecyclineinjection-products, accessed on 15 July 2022 [38]), while susceptibility tests for colistin were not shown due to unreliable results based on VITEK 2 methodology [39]. The validation of the antimicrobial resistance phenotype was based on the broth microdilution test for colistin. E. coli isolates that were non-susceptible to three or more antimicrobials from different classes were defined as exhibiting MDR [40]. PFGE was performed to identify E. coli isolates from index patient's initial urine and fecal samples [41]. Isolates resistant to third-generation cephalosporin were used for WGS.

Nanopore Sequencing and De Novo Assembly
Total genomic DNA from 27 CREC strains was extracted with the QIAamp PowerFecal Pro Kit (Qiagen, Hilden, Germany); subsequently, DNA fragments longer than 3 kb were enriched with KAPA Hyper Beads (Roche, Wilmington, MA, USA) according to the manufacturer's recommended procedures. WGS was performed on an ONT MinION sequencer (Oxford Nanopore Technologies, Oxford, UK) as previously described [42]. Briefly, each DNA sample was tagged with a unique barcode by using Rapid Barcoding Kit 96 (Oxford Nanopore Technologies). The barcoded DNAs were mixed and sequenced on a primed MinION SpotON Flow Cell (FLO-106MIN). Genome sequences were assembled de novo through a sampling strategy as previously described [42], and the assembled sequences were polished with Homopolish [17].

Pathogenic Types of E. coli
WGS was applied to detect VF-encoding genes due to high concordance compared with conventional PCR tools [43,44]. Established criteria were used to classify pathogenic E. coli based on VF-encoding genes. Isolates were categorized as extraintestinal pathogenic E. coli (ExPEC) if they carried two or more of the five virulence genes papAH and/or papC, sfa/focDE, afa/draBC, kpsM II, and iutA [45]; as uropathogenic E. coli (UPEC) if they carried three or more of the four virulence genes chuA, fyuA, vat, and yfcV [46]; avian pathogenic E. coli (APEC) if they carried four or more of the five genes hlyF, iutA, iroN, iss, and ompT [47].

Conclusions
We observed a strong genomic association between fecal ESBL-producing E. coli and uropathogenic strains. WGS may be useful in the analysis of epidemiological types and trends, characterization of antimicrobial resistance mechanisms, and detection of ARGs and VF-encoding genes. Although a single-family study may not be representative of the general genomic backgrounds of CREC isolates in the feces of individuals for Taiwan, some unique ARGs, including bla CTX-M and bla AmpC genes, were found. To our knowledge, this is the first study in Taiwan to suggest bla CMY-2 gene distribution among asymptomatic individuals and to investigate the genomic characteristics of CREC isolates through nanopore sequencing. Furthermore, a longitudinal nationwide multi-center investigation of other resistance mechanisms, such as transcriptional regulation of intrinsic efflux pumps, for both CREC and non-CREC isolates in the human carriage, household spread, and environmental exposure is warranted.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10.3 390/antibiotics11091152/s1, Figure S1: Pulsed-field gel electrophoresis (PFGE) of Escherichia coli isolates from index patient's initial urine and stool samples, Table S1: Genomic backgrounds of 15 bla CMY-2 plasmids, Table S2: Classifications of eighteen antimicrobials applied in VITEK2 AST-N320 cards.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.