International High-Risk Clones Among Extended-Spectrum β-Lactamase–Producing Escherichia coli in Dhaka, Bangladesh

Background: Escherichia coli is a major extended-spectrum β-lactamase (ESBL)–producing organism responsible for the rapid spread of antimicrobial resistance (AMR) that has compromised our ability to treat infections. Baseline data on population structure, virulence, and resistance mechanisms in E. coli lineages from developing countries such as Bangladesh are lacking. Methods: Whole-genome sequencing was performed for 46 ESBL–E. coli isolates cultured from patient samples at the International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b)-Dhaka. Sequence data were analyzed to glean details of AMR, virulence, and phylogenetic and molecular markers of E. coli lineages. Results: Genome comparison revealed presence of all major high-risk clones including sequence type 131 (ST131) (46%), ST405 (13%), ST648 (7%), ST410 (4.3%), ST38 (2%), ST73 (2%), and ST1193 (2%). The predominant ESBL gene and plasmid combination were blaCTX–M–15 and FII-FIA-FIB detected in diverse E. coli phylogroups and STs. The blaNDM–5 (9%) gene was present in prominent E. coli STs. One (2%) mcr-1–positive ST1011 E. coli, coharboring blaCTXM–55 gene, was detected. The extraintestinal pathogenic E. coli genotype was associated with specific E. coli lineages. The single nucleotide polymorphism (SNP)-based genome phylogeny largely showed correlation with phylogroups, serogroups, and fimH types. Majority of these isolates were susceptible to amikacin (93%), imipenem (93%), and nitrofurantoin (83%). Conclusion: Our study reveals a high diversity of E. coli lineages among ESBL-producing E. coli from Dhaka. This study suggests ongoing circulation of ST131 and all major non-ST131 high-risk clones that are strongly associated with cephalosporin resistance and virulence genes. These findings warrant prospective monitoring of high-risk clones, which would otherwise worsen the AMR crises.


INTRODUCTION
Infections with Escherichia coli that produce extended-spectrum β-lactamases (ESBLs) present an increasing clinical and public health threat (Mathers et al., 2015). These bacteria are resistant to several new-generation cephalosporin agents and render them ineffective for treating infections (Pitout and Laupland, 2008). Infections of ESBL E. coli causes increased morbidity, high mortality, longer hospital stays, and increased health care costs in comparison to infections caused by non-ESBL E. coli (Paterson and Bonomo, 2005). Treatment of such infections has been further complicated by the emergence of carbapenem resistance in ESBL-E. coli; these carbapenem-resistant isolates are often found resistant to all the available antibiotics (Doi, 2019).
Several strains of ESBL-producing E. coli are pathogenic; they commonly cause urinary tract infections (UTIs) in otherwise healthy people (Calbo et al., 2006;Franz et al., 2015). Moreover, when certain pathogenic bacterial clones horizontally acquire ESBL genes, they can emerge and reemerge rapidly within the population through clonal dissemination and thereby gain local or even global predominance as international high-risk clones (Price et al., 2013). Examples of such epidemiological successful extraintestinal pathogenic E. coli (ExPEC) lineages include sequence type 131 (ST131), ST410, ST38, ST73, ST405, and ST648, which are associated with both nosocomial and community-acquired infections and are being increasingly detected from multiple origins, worldwide (Baquero et al., 2013;Shaik et al., 2017;Manges et al., 2019). Since the last decade, CTX-M-type enzymes have become the most predominant ESBLs in E. coli and other Gram-negative bacteria of clinical significance (Cantón and Coque, 2006). Similarly, the transposable elements containing the New Delhi metallo-β-lactamase (NDM) gene have been identified to be responsible for the rapid dissemination of carbapenem resistance in E. coli (Kumarasamy et al., 2010).
Currently, a high number of ESBL-producing E. coli infections are linked to the pandemic E. coli lineage ST131 (Nicolas-Chanoine et al., 2008). Moreover, it is shown that E. coli ST131 strains are strongly associated with CTX-M-15-type ESBL and are predominantly responsible for causing bladder infections, kidney infections, and urosepsis worldwide including Southeast Asia (Nicolas-Chanoine et al., 2014;Chen et al., 2019). For instance, in India, it was reported that 93% of ST131 E. coli isolates carried CTX-M-15. Studies have also reported the presence of the H30 subclone that is reportedly responsible for the clonal dissemination of ST131 E. coli (Hussain et al., 2012(Hussain et al., , 2014Ranjan et al., 2015;Shaik et al., 2017). A study from Bangladesh reported that 71% of ESBL-producing E. coli was linked to the ST131 clone (Begum and Shamsuzzaman, 2016). Another study from Bangladesh identified a clonal cluster of clinical E. coli isolates belonging to serotype O25:H4, which indicates the widespread circulation of the ST131 E. coli lineage (Lina et al., 2014).
High-resolution studies on ESBL-producing E. coli strains from Bangladesh are lacking compared with the rest of the world, which limit our understanding of the population structure, emerging antimicrobial resistance (AMR) and their potential to spread. Herein, we report the findings of a genomic epidemiological investigation of the recent ESBL-E. coli isolates, especially the CTX-M-type ESBL producers associated with extraintestinal infections in Dhaka, Bangladesh. We analyzed their resistance genes, virulence genes, plasmid types, phylogenetic relatedness, and molecular features. Our study suggests ongoing circulation of ST131 and all major non-ST131 E. coli high-risk clones in Dhaka, Bangladesh. Studies such as these will provide important insights into the evolution of pathogens and will inform novel interventional strategies.

Whole-Genome Sequencing, Genome Assembly, and Annotation
Genomic DNA from an overnight bacterial culture was extracted and purified using QIAamp DNA Mini kit (Qiagen, Germany). The purity of the genomic DNA was assessed using a NanoDrop spectrophotometer (Thermo Fisher Scientific, United States), and quantification was performed using a Qubit 2.0 fluorometer (Life Technologies). DNA libraries were prepared using the Nextera XT DNA library Prep kit (Illumina) (Mazumder et al., 2020b). The pooled library was sequenced at the genomic sequencing facility of icddr,b on an Illumina Nextseq500 system using the 150-base-paired-end Mid-output v2.5 sequencing kit. The sequence reads were trimmed and filtered using Trimmomatic 0.39 (Bolger et al., 2014), and de novo assemblies of the resulting reads were generated using SPAdes 3.11.1 (Bankevich et al., 2012), and the quality check was done using QUAST v5.02 (Mikheenko et al., 2018). The genomes of our study were annotated using Prokka de novo (Seemann, 2014). The program was run on fast mode, and the genus-specific BLAST database was used. The genome features and associated metadata are listed in Table 1.

In silico Molecular Analysis
Phylotyping was done using the ClermonTyping online tool (Beghain et al., 2018). Other molecular features were  (Chen et al., 2005) belonging to different categories as described previously (Stoesser et al., 2015), a threshold, ≥ 90% identity; minimum coverage, ≥ 70% was used for both. AMR-related chromosomal point mutations were identified in the draft genomes using PointFinder (Zankari et al., 2017). The genetic context of bla CTX−M−15 , bla NDM−5 , and mcr-1 with respect to plasmid was determined by BLAST searching of particular contigs against the GenBank database. Genomes having three or more ExPEC-associated genes were classified as ExPEC according to Johnson's criteria (Johnson and Stell, 2000). From virulence and resistance gene presence and absence CSV files, heat plots were obtained, using the Python module seaborn and matplotlib.

Statistical Analysis
Using SPSS statistics software (version 25.0), continuous variables were compared with the non-parametric Mann-Whitney Utest, and proportions were analyzed using the χ 2 -test; p-values with a threshold of < 0.05 were considered statistically significant. Accordingly, the p-values are indicated within text where appropriate.

Fluoroquinolone Resistance
Using PointFinder, we identified two mutations in gyrA including S83L (serine to leucine) and D87N (aspartic acid to asparagine). The mutation S83L was detected in 13% (6/46) of isolates. In contrast, 80% (37/46) of isolates had both the mutations (S83L and D87N); the dual gyrA mutants were all resistant to ciprofloxacin. However, isolates (4/46) having a single mutation (S83L) were all susceptible to ciprofloxacin. Similarly, in the parC amino acid product, we identified two substitutions at codon position 80 (serine to isoleucine) and codon 84 (glutamic acid to valine). The mutation S80I was present in 33% (15/46) of isolates, and the mutation E84V was detected in one isolate, whereas 22 of 46 isolates harbored both of these mutations, and they were all resistant to ciprofloxacin. Three substitutions were also detected in parE gene, with I529L being predominant (20/46) followed by S458A (11/46) and L416F (1/46). All isolates (20/46) carrying I529L mutation were ciprofloxacin-resistant, and this mutation was found in the majority of ST131 strains (18/21 ST131 strains). Additionally, the plasmid-mediated quinolone resistance gene qnrs1 was detected in two isolates followed by qnrS4 and qnrS13 in one isolate each. Except for qnrs13, isolates carrying qnrS genes were susceptible to ciprofloxacin.

Carbapenem and Colistin Resistance
NDM-5 carbapenemase was detected in four E. coli isolates (8.5%). All four isolates belonged to the prominent clonal groups: two ST405 (phylogroup D), one ST648 (phylogroup F), and one ST167 (phylogroup A). All four bla NDM−5 genes were found to be located on plasmids (4/4 isolates). Except for one (ST167), all three isolates coharbored a bla CTX−M−15 gene. Phenotypic resistance to imipenem was detected in only two of four bla NDM−5 -positive isolates. Nonetheless, all four bla NDM−5 -positive E. coli were pan-drug-resistant as resistance was detected to at least 7 of 10 antibiotics tested. We detected one mcr-1-positive E. coli belonging to ST1011; the mcr-1 gene was detected on a plasmid. This strain was coharboring a bla CTXM−55 gene and showed phenotypic resistance to cotrimoxazole, ciprofloxacin, ceftriaxone, cefuroxime, and cefixime and was notably susceptible to imipenem, gentamicin, and amikacin.

Population Structure of Extended-Spectrum β-Lactamase Escherichia coli Isolates
To investigate the relatedness of 46 ESBL E. coli isolates, single nucleotide polymorphism (SNP)-based core genome phylogenetic tree was constructed (Figure 1). The 46 E. coli genomes clustered with a good concordance to their phylogroups. Particularly, strains belonging to phylogroups B2, D, and F formed distinct clades, whereas strains belonging to phylogroups A, B1, and C were clustered together under a large clade; 24 of these 46 E. coli isolates formed a large cluster of B2 phylogroup all qualified as ExPEC and had the most common plasmid combination of FII, FIA, and FIB; this clade mainly comprised (21/24) ST131 isolates. Further, it was identified that 16 of 21 ST131 E. coli belonged to the most virulent, widespread H30Rx subclone, three belonged to H30R, and two belonged to fluoroquinolone-susceptible fimH 41, CTX-M-15-positive ST131 E. coli. The 21 ST131 E. coli isolates showed little or no difference in the SNP analysis; these isolates were collected over 15 months, which indicates widespread dissemination of this particular lineage. The sister clade of ST131 clustered E. coli lineages that included ST1193, ST73, and ST421. When the branches of non-ST131 isolates were assessed closely, high clonality was observed between individual isolates in a few cases. For instance, six strains of ST405 formed an identical cluster; these were collected from November 2018 to July 2018, and five of six isolates carried bla CTX−M−15 and had identical serogroup and fimH types with almost common plasmid profile. However, two of these six isolates carried bla NDM−5 gene in addition to the shared profiles. Similarly, ST410 isolates formed identical clusters with shared molecular profiles. However, ST648 strains formed an identical cluster but demonstrated diverse molecular profiles.

DISCUSSION
ESBL-producing E. coli is responsible for a significant number of nosocomial and community-acquired infections (Pana and Zaoutis, 2018). In this study, we used a WGS approach to analyze 46 ESBL-E. coli isolates collected over a year (16 months) from a referral diagnostic center in Dhaka, Bangladesh, a Southeast Asian country from which data such as these are acutely lacking.
Our study revealed that the ST131 lineage is by far the most prevalent lineage (45%) represented particularly by H30Rx isolates (34%). This dominance of ST131 E. coli in our study isolates is possibly because of our focus on ESBL-E. coli. However, the non-ST131 (55%) E. coli isolates showed high diversity with respect to phylogroups, phylogenetic clustering, STs, serotypes, and fimH types, which is consistent with previous reports (Hussain et al., 2014;Musicha et al., 2017). Furthermore, we demonstrated that ESBLs, in particular, the CTX-M-type ESBLs, in Bangladesh have emerged across major high-risk international clones that comprise ST38, ST73, ST405, ST410, ST648, and ST1193 lineages, in addition to isolates of the pandemic E. coli ST131.
Molecular typing revealed that a majority of CTX-M-type ESBL producers belonged to the ST131 lineage, which constituted the most clonal group in the population structure of the studied isolates. The predominance of this lineage and its association with CTX-M-15 is in line with the reports from worldwide studies (Nicolas-Chanoine et al., 2014), including those from Asia (Hussain et al., 2012;Ranjan et al., 2015). Further, we documented a greater proportion (76%) of H30Rx subclones within the ST131 E. coli isolates, indicating a scenario of endemic circulation of H30Rx strains in the current setting, as these strains have been recovered between the span of April 2018 to June 2019. Clonal expansion of H30RX subclone might be the reason for the high prevalence and transmission of CTX-M-15-associated ST131 E. coli isolates in this setting. The H30Rx subclone is also reported in the neighboring country, India, in 2015 (Ranjan et al., 2015). The H30Rx isolates detected in this study were all qualified FIGURE 3 | Distribution of virulence gene profiles within 46 E. coli isolates are represented in the heatmap, with colored blocks indicating the gene presence and gray-colored blocks indicating the gene absence. Further, the virulence genes are vertically colored coded by functional categories (yellow, adhesins; blue, toxins; pink, miscellaneous; green, siderophores).
as ExPEC, which harbored a broad range of virulence genes with a relatively high virulence score, median (range): 27 (18-27). These isolates exhibited a high prevalence of a combination of IncFII, IncFIA, and IncFIB plasmid replicons, which is consistent with a recent WGS study from North Carolina (Kanamori et al., 2017). The gene aac(6')-lb-cr encoding the aminoglycoside and fluoroquinolone-modifying enzyme was strongly associated with H30Rx strains. In addition to the bla CTX−M−15 gene, they exhibited a high prevalence of bla OXA−1 and high resistance to third-generation cephalosporin antibiotics (ceftriaxone and cefixime). The association of H30Rx subclone with extendedspectrum cephalosporin and fluoroquinolone resistance might have contributed to their predominance and epidemiological success over other E. coli lineages.
In this study, we have further shown that the bla CTX−M−15 gene exists in genetically diverse strains including the prominent non-ST131 uropathogenic E. coli lineages such as ST38, ST73, ST405, ST410, ST648, and ST1193. Isolates of ST38 were reported to be evolving and is described to be increasingly detected in UTIs, which was previously considered to be a gut pathogen because they harbor genes for both EAEC and ExPEC (Chattaway et al., 2014). It is noticed that the ST38 isolate identified in this study was moderately lower in virulence and AMR gene content; particularly, it lacked the fluoroquinolone resistance and plasmids. Nonetheless, they were identified as ExPEC. ST73 is another frequently isolated ExPEC from UTI and bloodstream infections (Riley, 2014). The ST73 isolate in our study was found to carry an extensive array of virulence genes, whereas it was susceptible to most of the antibiotics and carried only IncCoI plasmid type, in contrast to other studies, which detected FII, FIA, and FIB plasmids; the reason for this could be that the ST73 strains are not expanding clonally as also suggested by others (Bogema et al., 2020). ST405 E. coli lineage has been implicated as drivers of bla CTXM−15 and is often associated with bla NDM genes and extensive virulence repertoire similar to the ST131 (Devanga Ragupathi et al., 2020). All six ST405 isolates from our study demonstrated high AMR rates and specifically carried a set of common plasmid replicons, including IncFII, IncFIA, IncFIB, and IncCoI. Five of these six ST405 strains qualified as ExPEC with extensive virulence gene profiles, consistent with other reports (Zhang et al., 2018;Devanga Ragupathi et al., 2020). Two of these six ST405 isolates also harbored bla NDM−5 gene. ST410 is considered another emerging "highrisk" clone that is resistant to fluoroquinolones, cephalosporins, and sometimes carbapenems (Roer et al., 2018). The two ST410 isolates in our collection showed moderate virulence and resistance profiles; they harbored bla CTX−M−15 gene with a set of five plasmid replicon types and fimH 24 allele (Figure 1). The lineage ST648 has been predicted to become another internationally circulating clone that will worsen infection treatment possibilities because of its AMR (Schaufler et al., 2019). The three ST648 isolates detected in this study carried bla CTX−M−15 and demonstrated high AMR rates; all the three strains consistently carried IncCoI plasmids. One ST648 strain was also positive for the bla NDM−5 gene. ST1193 is the latest pandemic multidrug-resistant uropathogenic clonal group (Johnson et al., 2019), which is usually detected among nonlactose fermenters (Johnson et al., 2019). The ST1193 isolate in our collection was qualified as ExPEC; it showed high resistance rates, with particular resistance to fluoroquinolones, cephalosporins, and cotrimoxazole. It harbored IncFIA, IncFIB, and IncCoI plasmid types. Several WGS studies have described the features of E. coli ST131. In contrast, not many have addressed the molecular epidemiology of non-ST131 E. coli worldwide. In this study, we have described that the ST131 E. coli is currently the most prevalent CTX-M-type ESBL producer in our community. However, we have also drawn attention to the emergence of significant clonal groups among non-ST131 E. coli. Further work is warranted in Bangladesh to accurately estimate the prevalence of these clonal groups and systematically analyze them in a global context, which is pertinent from public health and clinical standpoint.
Taken together, our study suggests that the CTX-M-type ESBLs and particularly the CTX-M-15 are prevalent among diverse E. coli STs circulating in Dhaka, Bangladesh. Further, our findings confirm the striking predominance of ST131 lineage and also revealed the presence of several major high-risk non-ST131 ExPEC clonal lineages, such as ST38, ST73, ST405, ST410, ST648, and ST1193, all associated with cephalosporin resistance and virulence. Further genomic epidemiological studies are needed to keep track of significant virulent/multiresistant E. coli clones in Bangladesh. Such studies are needed to inform us about the evolutionary dynamics of pathogenicity and resistance in emerging E. coli lineages of public health concern which will aid in evidence-based infection control and antibioticprescribing policies.

DATA AVAILABILITY STATEMENT
The datasets generated for this study can be found in the GenBank (Bioproject Accession: PRJNA654992). The GenBank accession numbers of 46 genomes sequenced in this study can be found in Table 1.

AUTHOR CONTRIBUTIONS
AH and RM designed the study, performed genome sequencing, and drafted the manuscript. AA, AH, and RM carried out the bioinformatics analyses and interpretation of results. MI, MS, SM, AT, FH, and NA were involved in the culture and AST of bacteria isolates. DA and DM supervised the study. All authors have read and approved the final manuscript.