Comprehensive Molecular Characterization of Escherichia coli Isolates from Urine Samples of Hospitalized Patients in Rio de Janeiro, Brazil

Urinary tract infections (UTIs) are often caused by Escherichia coli. Their increasing resistance to broad-spectrum antibiotics challenges the treatment of UTIs. Whereas, E. coli ST131 is often multidrug resistant (MDR), ST69 remains susceptible to antibiotics such as cephalosporins. Both STs are commonly linked to community and nosocomial infections. E. coli phylogenetic groups B2 and D are associated with virulence and resistance profiles making them more pathogenic. Little is known about the population structure of E. coli isolates obtained from urine samples of hospitalized patients in Brazil. Therefore, we characterized E. coli isolated from urine samples of patients hospitalized at the university and three private hospitals in Rio de Janeiro, using whole genome sequencing. A high prevalence of E. coli ST131 and ST69 was found, but other lineages, namely ST73, ST648, ST405, and ST10 were also detected. Interestingly, isolates could be divided into two groups based on their antibiotic susceptibility. Isolates belonging to ST131, ST648, and ST405 showed a high resistance rate to all antibiotic classes tested, whereas isolates belonging to ST10, ST73, ST69 were in general susceptible to the antibiotics tested. Additionally, most ST69 isolates, normally resistant to aminoglycosides, were susceptible to this antibiotic in our population. The majority of ST131 isolates were ESBL-producing and belonged to serotype O25:H4 and the H30-R subclone. Previous studies showed that this subclone is often associated with more complicated UTIs, most likely due to their high resistance rate to different antibiotic classes. Sequenced isolates could be classified into five phylogenetic groups of which B2, D, and F showed higher resistance rates than groups A and B1. No significant difference for the predicted virulence genes scores was found for isolates belonging to ST131, ST648, ST405, and ST69. In contrast, the phylogenetic groups B2, D and F showed a higher predictive virulence score compared to phylogenetic groups A and B1. In conclusion, despite the diversity of E. coli isolates causing UTIs, clonal groups O25:H4-B2-ST131 H30-R, O1:H6-B2-ST648, and O102:H6-D-ST405 were the most prevalent. The emergence of highly virulent and MDR E. coli in Brazil is of high concern and requires more attention from the health authorities.

Urinary tract infections (UTIs) are often caused by Escherichia coli. Their increasing resistance to broad-spectrum antibiotics challenges the treatment of UTIs. Whereas, E. coli ST131 is often multidrug resistant (MDR), ST69 remains susceptible to antibiotics such as cephalosporins. Both STs are commonly linked to community and nosocomial infections. E. coli phylogenetic groups B2 and D are associated with virulence and resistance profiles making them more pathogenic. Little is known about the population structure of E. coli isolates obtained from urine samples of hospitalized patients in Brazil. Therefore, we characterized E. coli isolated from urine samples of patients hospitalized at the university and three private hospitals in Rio de Janeiro, using whole genome sequencing. A high prevalence of E. coli ST131 and ST69 was found, but other lineages, namely ST73, ST648, ST405, and ST10 were also detected. Interestingly, isolates could be divided into two groups based on their antibiotic susceptibility. Isolates belonging to ST131, ST648, and ST405 showed a high resistance rate to all antibiotic classes tested, whereas isolates belonging to ST10, ST73, ST69 were in general susceptible to the antibiotics tested. Additionally, most ST69 isolates, normally resistant to aminoglycosides, were susceptible to this antibiotic in our population. The majority of ST131 isolates were ESBL-producing and belonged to serotype O25:H4 and the H30-R subclone. Previous studies showed that this subclone is often associated with more complicated UTIs, most likely due to their high resistance rate to different antibiotic classes. Sequenced isolates could be classified into five phylogenetic groups of which B2, D, and F showed higher resistance rates than groups A and B1. No significant difference for the predicted virulence genes scores was found for isolates belonging to ST131, ST648, ST405, and ST69. In contrast, the phylogenetic groups B2, D and F showed a higher predictive virulence score compared to phylogenetic groups A and B1.

INTRODUCTION
Urinary Tract Infections (UTIs) are one of the most important causes of community and healthcare-associated infections in many clinical onsets worldwide, including Brazil (Terpstra and Geerlings, 2016;Wurpel et al., 2016). Indeed 30-50% of healthcare-associated infections are due to UTIs. This high prevalence is linked to several risk factors, such as catheterization, surgical manipulation and disruption of the urinary tract, diabetes, immunosuppressant drug use, previous admissions, and other comorbidities (Saltoglu et al., 2015;Redder et al., 2016). The risk factors and antibiotic resistance profiles are different for infections acquired in the community or in the hospital environments (Saltoglu et al., 2015). Although in general the majority of UTI cases are uncomplicated, UTIs in hospitalized patients increase the risk for developing sepsis and lead to higher mortality rates (Melzer and Welch, 2013).
Escherichia coli is the main etiological agent responsible for 70-90% of all UTIs (Gurevich et al., 2016;Terpstra and Geerlings, 2016). The treatment of patients with UTIs has become increasingly difficult because of the rapid spread of antibiotic resistance (Can et al., 2015). Especially, extended spectrum beta-lactamase (ESBL)-producing E. coli are a problem, but an observed rise in fluoroquinolones and aminoglycosides resistance has also significantly contributed to problematic and reduced treatment options for infected patients (Tsukamoto et al., 2013;Bonelli et al., 2014). Several studies have already described the high prevalence of UTIs caused by ESBL-producing E. coli in the community and hospitals (Guzmán-Blanco et al., 2014;Gonçalves et al., 2016).
Recently, high antibiotic resistance rates have been associated with specific E. coli lineages, such as the multidrug resistant (MDR) sequence type (ST) 131 (Ben Zakour et al., 2016). Particularly, CTX-M beta-lactamase producing E. coli of serotype O25:H4 and ST131 is a successful spreading clone (Giedraitiene et al., 2017) strongly associated with the resistance to aminoglycosides and fluoroquinolones. In contrast, other E. coli lineages such as ST69, ST73, and ST95, also frequently found as a causative agent of community and hospital acquired UTIs, seem to persist as non-ESBL-producing isolates (Riley, 2014;Doumith et al., 2015).
Extra-intestinal pathogenic E. coli (ExPEC), including uropathogenic E. coli (UPEC) most commonly associated with human disease, consist of distinct phylogenetic groups with different sets of virulence genes. Previous studies have shown that most ExPEC isolates causing infections belong to phylogenetic groups B2 and D, while isolates in phylogenetic groups A and B1 were mostly identified as commensal E. coli isolates (Katouli, 2010). Moreover, pathogenic ExPEC isolates harbor specific virulence genes which confer their pathogenic potential (Cyoia et al., 2015) and are involved in every step in the pathogenicity of ExPEC. Thus, adhesins are a prerequisite to adherence and successful colonization, toxins are responsible for cell damage to urinary tract epithelial cells, and the iron uptake system allows colonization of the urinary tract thereby helping the bacteria to persist (Alizade et al., 2014).
Despite the diversity of ExPEC causing infections, previous studies have shown the connection between specific E. coli lineages and their particular resistance profiles, and severity of the infections (Can et al., 2015;Matsumura et al., 2016;Zhang et al., 2016). Thus, defining the genetic background of the pathogen by the identification of a particular ST, its serotype and the detection of resistance genes, can be useful not only for improving further patient treatment but also to allow an improved risk assessment of bacterial infections in the hospitals. The aim of this study is to comprehensively characterize the population structure of E. coli from urine samples collected from patients in four hospitals in Rio de Janeiro, Brazil using whole genome sequencing (WGS).

Bacterial Isolates
E. coli isolates were collected from urine samples of patients admitted to different wards of the Hospital Universitário Pedro Ernesto (HUPE; a 600-bed university hospital) or to one of three small private hospitals (coded Hospital A, Hospital B and Hospital C; see Data Sheet S1). All four hospitals are located in the city of Rio de Janeiro, Brazil. Patients were included regardless the presence of risk factors or observed UTI symptoms. In this study, 107 isolates were collected between November 2015 and November 2016 from the patients (50.60% were from the private hospitals and 49.40% from the public hospital). Eightyeight percent of the isolates were from female patients. Bacterial isolates were cultured on cysteine lactose deficient medium agar plates (CLED, BD, Germany) till a cell density higher than 10 5 colony-forming units was obtained. Bacterial cells were stored at −80 • C in a Luria-Bertani Broth (LB, Merck, S.A.) with 20% glycerol.

DNA Extraction and Whole Genome Sequencing
Total bacterial DNA was extracted from each isolate using the UltraClean R microbial DNA isolation kit (MO BIO Laboratories, Carlsbad, CA, US) following the manufacturer's protocol. A DNA library was prepared for individual samples using the Nextera XT kit (Illumina, San Diego, CA, US) following the manufacturer's instructions. Whole genome sequencing was performed on the Miseq (Illumina) to generate 250-bp pairedend reads to obtain a coverage of at least 60-fold as previously described (Ferdous et al., 2015).

Assembly and Data Analysis
De novo assembly was performed using CLC Genomics Workbench v10.0.1 (Qiagen, CLC bio A/S, Aarhus, Denmark) using default settings and an optimal word-size. The assembly quality data for all isolates is available in the supplementary data table (Data Sheet S2). Annotation was performed by uploading the assembled genomes onto the RAST server version 2.0 (Aziz et al., 2008). The ST was identified by uploading the assembled genomes in fasta format to the Center for Genomic Epidemiology (CGE) MLST finder website (version 1.7) (Larsen et al., 2012). Presence of antibiotic resistant genes was determined by uploading assembled genomes in fasta format to ResFinder 2.1 (Zankari et al., 2012), the serotyping by using the SerotypeFinder tool (Joensen et al., 2015), and the fimH type by uploading the genomes to FimTyper (version 1.0) (Roer et al., 2017) all present through the CGE website.
Virulence Genes, Virotype, Phylogenetic Typing, and Analysis The virulence genes were identified by blasting them against known virulence reference genes (see Data sheet S3) downloaded from the NCBI or ENA database into the CLC Genomics Workbench v10.0.1 (Qiagen, CLC bio A/S, Aarhus, Denmark). In total, 64 virulence genes were investigated and the predictive virulence score was determined using the number of genes found in each isolate. Predictive virulence genes scores were also used to characterize the isolates as ExPEC or UPEC as described by Johnson et al. (2015). The virotype of the ST131 isolates was defined as described by Dahbi et al. (2014). Phylogenetic groups were defined as described by Clermont et al. (2013). To determine the phylogenetic relationship the isolates were uploaded into SeqSphere v.4.1.9 (Ridom, Munster, Germany) and a gene-bygene typing approach using a 2764-genes core genome (cg) MLST scheme was used as previously described (Ferdous et al., 2016).

Statistical Analysis
The Mann-Whitney test was used to compare the mean of predictive virulence scores (PVS) between the phylogenetic and ST groups. Analysis was performed using GraphPad Prism v7.03 (GraphPad Software, La Jolla, US).

Nucleotide Sequence Accession Number
The raw data of all whole genome sequenced isolates were deposited in the European Nucleotide Archive under the project number PRJEB23420. See the supplementary data table (Data Sheet S2) for individual accession numbers.

Phylogenetic Analysis
In the present study, the most frequently found phylogenetic group was B2 (n = 52; 49.53%), followed by phylogenetic groups A (n = 20; 18.69%), D (n = 14; 13.08%), B1 (n = 14; 13.08%), and F (n = 4; 3.74%; Figure 2C). For 1.87% of the isolates it was not possible to identify the phylogenetic group ( Figure 2C). All ST131, ST73, and ST648 isolates belonged to The frequency for all antibiotic tested, showing the high resistance rate to antibiotics most frequently used in the treatment of UTIs such as aminoglycosides, fluoroquinolones, trimethoprim, and trimethoprim-sulfamethoxazole and a low frequency of resistance to fosfomycin and nitrofurantoin.
FIGURE 2 | Distribution of sequence types (ST), serotypes, and phylogenetic groups extracted from the whole genome sequence data. (A) Percentage of ST lineages found in this study, showing the high prevalence of ST131, ST69, ST10, ST648, ST450, and ST73. Isolates belonging to singleton STs comprise more than one third of the isolates; (B) Frequencies of serotypes found showing O25:H4 to be the most frequent serotype; (C) Frequencies of the five phylogenetic groups, showing the high prevalence of B2, followed by A, D, and B1 and the low prevalence of isolates belonging to phylogenetic group F. phylogenetic group B2 while ST69 and ST405 isolates belonged to phylogenetic group D. The isolates of ST10, ST1703, ST744 were classified in the phylogenetic group A and the ST354 isolates were classified in phylogenetic group B1 (see Data Sheet S4). The other isolates represented by a diversity of ST groups were classified into different phylogenetic groups. We investigated the genetic relationships of the sequenced isolated based on their core genome. Not surprisingly, the isolates of the same ST were genetically related and formed ST specific cgMLST clusters (Figure 3). The ST131 isolates with serotype O25:H4 showed less genetic diversity and clustered closely to each other in the cgMLST phylogenetic tree. In general, the ST131 isolates were more closely related with each other while the isolates within ST69 were more diverse. On the other hand, the ST131 isolates could be separated by their serotype and O16:H5/ST131 isolates clustered separately from O25:H4/ST131 ones. Based on the core genome analysis the same was observed for isolates belonging to ST405, ST1703, and ST648 that clustered according to their ST and within such cluster isolates showed a high degree of genetic relatedness. Observed genetic relationships between isolates were independent from their hospital origin.

Clonal Associations of blaCTX-M
Whole genome sequencing data was used to screen for the presence of genes responsible for the ESBL phenotype. This analysis revealed that 30 of the 31 (96.77%) ESBL-producing isolates contained a gene encoding a beta-lactamase of the bla CTX−M type. In addition, two isolates were AmpC betalactamase producing and contained the bla CMY−2 gene. In the CTX-M positive isolates, bla CTX−M−15 was the most frequently found variant (n = 17; 53.12%) followed by bla CTX−M−8 (n = 5; 15.62%). The majority of bla CTX−M−15 isolates belonged to O25:H4/ST131, and all the isolates that were CTX-M-15-producing belonged to high risk clonal groups (O25:H4/ST131, O1:H6/ST648, or O102:H6/ST405). Among the singleton isolates 17.24% (n = 5) were ESBL-producing, and carried different CTX-M genes (Table 1). Interestingly, the CTX-M-producing isolates were also frequently found to carry FIGURE 3 | Neighbor-joining (NJ) phylogenetic tree of Escherichia coli isolates based on a 2764-genes core genome MLST scheme. High-risk clonal groups are indicated by red doted boxes. For all isolates the phylogenetic groups, serotype and ST group is indicated unless the typing could not be identified from the whole genome sequence data.

Association of ST and Phylogenetic Groups with Resistance Pattern
The majority of the MDR isolates belonged to ST131, ST648, or ST405 while most non-MDR isolates belonged to ST69, ST10, ST73, or singleton STs. The ST131, ST648, and ST405 isolates also showed a higher resistance rate to other antibiotic classes as ampicillin and amoxicillin/clavulanate ( Figure 4A). Among the singleton STs, the number of MDR isolates was low. The phylogenetic groups B2, D, and F were more often found to be resistant to ampicillin, amoxicillin/clavulanate, ciprofloxacin, and trimethoprim than phylogenetic groups A and B1 ( Figure 4B).

Association of ST and Phylogenetic Group with Virulence Genes
The main six ST groups identified in this study were compared to evaluate their urovirulence-potential, using the 64 identified virulence genes (Data Sheet S7). Based on the predictive virulence score (PVS) no statistically significant difference was found for ST131 (PVS = 18.3) and ST648 (PVS = 17.6) isolates compared to ST69 (PVS = 17.8) isolates (p = 0.2444 and p = 0.9993, respectively). In contrast, the ST405 (PVS = 13.0) and ST10 (PVS = 12.7) isolates had lower PVS compared to other STs groups (p < 0.0001). The ST73 isolates appeared to have the highest PVS (24.0) compared to other groups (p < 0.0001). Interestingly, the PVS for isolates belonging to singleton ST groups scored slightly higher (PVS = 19.0) than isolates belonging to ST131, ST648, ST405, ST69, and ST10 (p = 0.0439). When the same analysis was performed on different phylogenetic groups, phylogenetic groups B2, D, and F had higher PVSs than phylogenetic groups A and B1 (p = 0.2190), although this was not statistically significant ( Table 3).

DISCUSSION
In this study, a comprehensive molecular characterization of E. coli isolated from urine samples of hospitalized patients in hospitals in Rio de Janeiro was performed and showed the presence of successful MDR clones similar to those found in other parts of the world (Riley, 2014). In general, high resistance rates to antibiotics such as cephalosporin, aminoglycosides, fluoroquinolones and trimethoprim often used to treat patients with UTIs were found. The emergence of MDR E. coli complicates the treatment of UTIs and is a major concern for hospitals (Flores-Mireles et al., 2015). Our results are in agreement with previous reports from Brazil, showing an increase of resistance rates of E. coli to aminoglycosides and fluoroquinolones (Correal et al., 2014;Rodrigues et al., 2016). In addition, the resistance rates to fosfomycin and nitrofurantoin, antibiotics used to treat uncomplicated UTIs, were found to be low in the investigated isolates, consistent with results from previous studies (Michalopoulos et al., 2011;Derakhshandeh et al., 2015). In our study, 49.53% of the isolates were identified as UPEC and 9.34% were classified as ExPEC (non-UPEC) based on predictive virulence genes score. The other 41.13% could not be typed as ExPEC using this method, indicating that the predictive virulence genes score is not always sufficient for classification of ExPEC as has also been reported before (Berman et al., 2014). In general, ExPEC can be classified into five phylogenetic groups, i.e., A, B (subgroups B1 and B2), D, E, and F, and the majority of the isolates in our study belonged to phylogenetic groups B2 and D. Indeed, other studies, as the ones from Iran and China, show that human pathogenic ExPEC predominantly belong to these two groups (Kazemnia et al., 2014;Tong et al., 2014), that are also considered to be more virulent and more associated with infections than, e.g., phylogenetic groups A and B1 (Lee et al., 2016). In our study, two isolates could not be assigned to any of the phylogenetic groups. This is in agreement with findings of others that assigning isolates to a specific phylogenetic group based on the current guidelines is not always possible (Clermont et al., 2013). The phylogenetic groups B2 and D were more often found to be MDR than the isolates of phylogenetic groups A and B1, which is agreement with other studies (Lee et al., 2016).
In our study population, the two most frequently found E. coli lineages were ST131 and ST69, which is in line with previous studies showing the worldwide spread of these STs and their association with UTIs (Peirano et al., 2014;Doumith et al., 2015). ST69 has previously been associated with both community acquired and healthcare associated UTIs (Riley, 2014) and appears to be frequently MDR, due to the presence of a resistance gene cassette (dfrA17-aadA5) that confers resistance to aminoglycosides and trimethoprim (Riley, 2014). Interestingly, our results showed that ST69 isolates were susceptible to aminoglycosides but had a high resistance rate to trimethoprim. As ST131 has emerged as the most prevalent high-risk lineage among infections caused by E. coli (ExPEC), its high prevalence in this study is not surprising. Moreover, the high frequency of the O25:H4/ST131 clonal group was also similar to findings of others in Brazil, Lithuania and the Netherlands (Dias et al., 2009;Overdevest et al., 2015;Giedraitiene et al., 2017). Other ST groups found in this study include ST648, ST405, ST73, and ST10, previously shown to be associated with urinary and bloodstream infections (Peirano et al., 2014;Doumith et al., 2015; Gonçalves et al., 2016;Hertz et al., 2016;Matsumura et al., 2016). Interestingly, in contrast to other studies performed in the UK and Denmark, the high virulent lineage ST73 was found less frequently than ST10, i.e., only in 3.7 and 6.7% of the collected isolates, respectively (Gibreel et al., 2012;Hertz et al., 2016). ESBL-producing bacterial isolates are of great medical concern in Latin American countries such as Brazil (Bonelli et al., 2014;Sampaio and Gales, 2016). The majority of ESBL-producing isolates in this study carried the bla CTX−M−15 gene, different from previous studies, in which bla CTX−M−2 and bla CTX−M−8 were found most frequently (Bonelli et al., 2014;Guzmán-Blanco et al., 2014). The majority of ESBL-producing isolates in O25:H4/ST131 clonal group were CTX-M-15 producing. The E. coli O25:H4/ST131 CTX-M-15 producing isolates were detected in other countries worldwide (Yumuk et al., 2008;Merino et al., 2016) and are known to be associated with increased capacity of plasmid uptake which results in high plasmid diversity despite showing a similar phenotype (Petty et al., 2014). In addition, the O25:H4/ST131 CTX-M-producing isolates in this study were also found to be resistant to gentamicin, tobramycin, and ciprofloxacin. This is similar to data presented in studies worldwide that showed that CTX-M-producing isolates are often MDR (Pitout and Laupland, 2008;Ewers et al., 2014;Ciesielczuk et al., 2015).
In general, higher resistance rates for more than three antibiotic classes were found in isolates belonging to ST131, ST648, and ST405. These results are in agreement with previous studies in the UK and Denmark that showed a broad-spectrum resistance of ST131 E. coli (Ferjani et al., 2014;Hertz et al., 2016) and that ST648 and ST405 have mobile elements containing genes that confer resistance to aminoglycosides, sulfonamides, and trimethoprim (Matsumura et al., 2013;Zhang et al., 2016). In addition, the successful spread of the high-risk clone O25:H4/ST131 is largely responsible for the increased prevalence of ESBL-producing isolates. Other examples of E. coli high-risk clones include isolates that belong to ST405 and ST648 Mathers et al., 2015). Our results showed that all ST131 isolates belong to phylogenetic group B2 and that all ST405 isolates belong to phylogenetic group D. These groups, often CTX-M-ESBL producing, have been reported as high-risk pandemic clones Shaik et al., 2017). Patients carrying such a high-risk isolate that easily spreads can be the cause of outbreaks in hospital settings and should be put into isolation upon admission.
In contrast to findings of others who reported that ST648 isolates belong to phylogenetic group D (Gonçalves et al., 2016;Müller et al., 2016), we found that the ST648 isolates in this study belong to phylogenetic group B2. This classification was based on the observation that in the whole genomes of our ST648 isolates the yjaA and arpA genes were absent, whereas the tspE4.C2 and chuA genes were present. Therefore, they belong to phylogenetic group B2 based on the phylo-typing method described by Clermont et al. (2013). In addition, our ST648 isolates contained a mutation (G → C) in the primer binding site of primer TspE4C2.1b at the position where the most 3' nucleotide of this primer should anneal. This may lead to misclassification of the isolate as belonging to phylogenetic group F instead of B2 when using the PCR-based method for phylo-typing described by Clermont et al. (2013).
The results of this study, show that the majority of O25:H4/ST131 isolates belong to subclone H30-R, whereas part of these isolates belong to subclone H30-Rx (classified as virotype C or A). The rise in fluoroquinolone resistance in the last years is associated with the rapid emergence of this latter subclone that is often MDR (Peirano et al., 2014). It has also been associated with upper UTIs and primary sepsis, and often contains the aac(6')-Ib-cr gene (responsible for fluoroquinolone resistance) (Peirano et al., 2014). The evolutionary history of subclone H30-Rx is unclear. The most accepted theory to explain the success of its emergence is that it has, as other high-risk bacterial clones, acquired certain adaptive traits and survival skills while acquiring antibiotic resistance and virulence genes located on mobile elements (Woodford et al., 2011;Mathers et al., 2015). Therefore, detailed molecular characterization studies are required to increase the knowledge about the evolution of this subclone (Petty et al., 2014;Matsumura et al., 2016) and to identify specific molecular markers (including resistant/virulence genes and/or specific plasmids) to optimize diagnostics and subsequent antibiotic therapy.
The pathogenicity of UPEC is based on virulence and fitness factors that allow the bacteria to entry, adhere, acquire essential nutrients such as iron, multiply, cause tissue damage, and disseminate in the urinary tract (Subashchandrabose et al., 2014). The most frequently found virulence genes in our isolates were associated with the iron uptake system and adhesins, whereas fimbriae and toxins were less frequently found. These results differ from previous studies where a high frequency of adhesins and toxins genes among UPEC isolates were found (Alizade et al., 2014). Whereas, several studies showed the association between the presence of adhesins and toxins with more complex UTIs (Wiles et al., 2008;Tarchouna et al., 2013), others could not correlate the presence of these virulence genes with the complexity of UTIs (Kudinha et al., 2013;Firoozeh et al., 2014). Most likely, the complexity of a UTI is defined by a combination of virulence genes, including those associated to the iron uptake system and adhesins. Indeed, efficient iron uptake is essential for the bacteria to survive and colonize in a poor iron environment as the urinary tract (Lee et al., 2016). In addition, the presence of adhesins such as afa, pap, sfa has been described to be important for invading urinary epithelial cells and in our isolates identified virulence genes cnf-1 and hlyA are essential subsequent dissemination (Lee et al., 2016). Other genes frequently found in our isolates were ompT, malX, kpsM, and traT. These genes are common virulence genes found in isolates associated with cystitis and pyelonephritis (Firoozeh et al., 2014;Derakhshandeh et al., 2015).
Overall, in our study, virulence genes were most prevalent among B2 isolates, followed by group D and F. In addition, their prevalence among sequence types ST131, ST69, ST1703, ST405, and ST648 was similar. ST73 isolates had a higher PVS compared to the other investigated groups. This is in agreement with findings of others that described E. coli ST73 to be a high virulent clone (Alhashash et al., 2016). In addition, ST131-B2 strains have emerged globally causing MDR resistant extraintestinal infections . Therefore, MDR isolates belonging to phylogenetic group B2 and clonal group O25:H4/ST131 are considered to form a double threat, because of their high resistance rate and substantial extraintestinal virulence capacity (Ferjani et al., 2014).
In conclusion, a large diversity of E. coli isolates causing UTIs was found in urine samples obtained from patients in Rio de Janeiro. The identified STs belonged to the most prevalent clonal groups reported worldwide. Among the investigated isolates the antibiotic resistance rate was high, as was the prevalence of ESBL-producing isolates. This result is associated with the presence of high-risk clones, often MDR, that mainly belong to phylogenetic group B2 D and F, containing a high number of virulence genes. The presence of highly virulent and MDR E. coli in Brazilian hospitals is of high concern for health care institutions and requires more attention from the health authorities. Clearly, it has consequences for the treatment of the patients and the outcome of the disease. Therefore, standard implementation of molecular methods to characterize E. coli isolates from urine in hospitalized patients is required to optimize diagnostic stewardship, patient treatment and infection control measures.

ETHICS STATEMENT
This study was approved by the Pedro Ernesto University Hospital ethical committee according and with Brazilian legislation and receive this register number: CAAE:45780215.8.0000.5259.

AUTHOR CONTRIBUTIONS
AC: drafting the article, data analysis, and interpratation; NA, CS, and JC: data collection and sample collection; MF and MC: data analysis and interpretation; JL: revision of the article; AR: conception and design of the work; PD: data collection and revision of the article; AF: final approval of the version to be published; JR: critical revision of the article.

ACKNOWLEDGMENTS
The author has received financial support from the Abel Tasman Talent Program for biomedical research talent of the University Medical Center Groningen which aims to support international high-quality research at the University of Groningen/University Medical Center Groningen, the Netherlands. This work was also supported by grants from FAPERJ and CAPES.