Fluoroquinolone resistance determinants in carbapenem-resistant Escherichia coli isolated from urine clinical samples in Thailand

Background Escherichia coli is the most common cause of urinary tract infections and has fluoroquinolone (FQ)-resistant strains, which are a worldwide concern. Objectives To characterize FQ-resistant determinants among 103 carbapenem-resistant E. coli (CREc) urinary isolates using WGS. Methods Antimicrobial susceptibility, biofilm formation, and short-read sequencing were applied to these isolates. Complete genome sequencing of five CREcs was conducted using short- and long-read platforms. Results ST410 (50.49%) was the predominant ST, followed by ST405 (12.62%) and ST361 (11.65%). Clermont phylogroup C (54.37%) was the most frequent. The genes NDM-5 (74.76%) and CTX-M-15 (71.84%) were the most identified. Most CREcs were resistant to ciprofloxacin (97.09%) and levofloxacin (94.17%), whereas their resistance rate to nitrofurantoin was 33.98%. Frequently, the gene aac(6′)-Ib (57.28%) was found and the coexistence of aac(6′)-Ib and blaCTX-M-15 was the most widely predominant. All isolates carried the gyrA mutants of S83L and D87N. In 12.62% of the isolates, the coexistence was detected of gyrA, gyrB, parC, and parE mutations. Furthermore, the five urinary CREc-complete genomes revealed that blaNDM-5 or blaNDM-3 were located on two plasmid Inc types, comprising IncFI (60%, 3/5) and IncFI/IncQ (40%, 2/5). In addition, both plasmid types carried other resistance genes, such as blaOXA-1, blaCTX-M-15, blaTEM-1B, and aac(6′)-Ib. Notably, the IncFI plasmid in one isolate carried three copies of the blaNDM-5 gene. Conclusions This study showed FQ-resistant determinants in urinary CREc isolates that could be a warning sign to adopt efficient strategies or new control policies to prevent further spread and to help in monitoring this microorganism.


INTRODUCTION
Urinary tract infections (UTIs) are one of the most common bacterial infections with ∼150 million cases per year globally, which pose a large burden on the healthcare system as the associated costs have reached USD 6 billion (Badamchi et al., 2019;Foxman, 2014).A UTI is one of the most common extraintestinal infections, with 80% of UTIs caused by Escherichia coli, which is the target of empirical therapy (Córdoba et al., 2017).Fluoroquinolones (FQs) are the drugs of choice to treat UTIs, especially complicated cases and catheter-associated UTIs, since this antibiotic class has a wide spectrum of antimicrobial activities with excellent bioavailability, good oral absorption, and good tissue penetrability (Sharma, Jain & Jain, 2009).Although serious side effects associated with fluoroquinolone treatment have been mentioned by the US FDA (US Food and Drug Administration, 2016), this antibiotic remains the drug of choice.
The progressive emergence of resistance to fluoroquinolones and other antibiotics commonly used for UTI treatment has been observed in several countries over the last few years (Córdoba et al., 2017;Lee, Lee & Choe, 2018;Cunha et al., 2016).The resistance rate against FQ has recently increased, particularly among the Enterobacterales, especially E. coli (Faine et al., 2022;Kibwana et al., 2023).FQ resistance in E. coli showed that the mutations in the quinolone resistance-determining regions (QRDRs) of the chromosomal genes gyrA, parC, and parE lead to alteration of the target proteins (DNA gyrase and topoisomerase IV) of these antimicrobial agents and the operons of the endogenous transmembrane efflux pump AcrAB-TolC (marR and acrR) (Chen, Erickson & Meng, 2020).The efflux pump (oqxAB), a variant of the aminoglycoside-modifying enzyme aac(6 )-Ib-cr and qnr determinants (qnrA) that encode DNA gyrase protection proteins, was also identified to have the potential to reduce the susceptibility to FQs and lead to resistance in E. coli (Cheng et al., 2020).
The most common mechanisms are the mutation of the gene that encodes type II topoisomerases (DNA gyrase and topoisomerase IV), which are enzymes that are essential for DNA replication and alter the fluoroquinolone binding affinity of the enzyme.It inhibits the activity of DNA gyrase and topoisomerase IV, which are enzymes that are essential for DNA replication (Hopkins, Davies & Threlfall, 2005).
In Thailand, there is insufficient data about FQ-resistant E. coli.In 2015, we began conducting the Emerging Antimicrobial Resistant Bacterial Surveillance Program (EARB).In this program, we focused on carbapenem-resistant Enterobacterles (CRE) and colistinresistant Enterobacterles isolated from patients in 11 hospital networks in Thailand (Takeuchi et al., 2022;Boueroy et al., 2022).

Bacterial isolates
The present study used 103 urine isolates of CREc during 2015-2020, comprising 88 isolates from a previous study (Takeuchi et al., 2022) and 15 isolates from the present study from eight provinces in Thailand under the EARB program (Fig. 1 and Table S1).
CREc definition in the present study revealed E. coli strains that are resistant to at least one of the carbapenem antibiotics (ertapenem, meropenem, doripenem, or imipenem) or produce a carbapenemase, an enzyme that can destroy carbapenem antibiotics.

Ethical approval
Ethical approval was obtained from the Ethics Committee of Osaka University Graduate School of Medicine, Osaka, Japan, with approval number 14468-5.The present study was conducted following the principles of the Declaration of Helsinki; the need for informed consent was waived.

Antimicrobial susceptibility testing
All isolates were subjected to antimicrobial susceptibility testing using the broth microdilution method according to 2022 Clinical and Laboratory Standards Institute guidelines (Clinical Laboratory Standard Institute, 2022).The broth microdilution method was conducted using cation-adjusted Mueller-Hinton broth (Becton, Dickinson and Company; Sparks, MD, USA) for antimicrobial susceptibility testing of ciprofloxacin, levofloxacin, and carbapenem.Susceptibility to nitrofurantoin (NFT) was carried out via a disk diffusion technique and was interpreted based on Clinical Laboratory standard institute (2022).E. coli ATCC 25922 was used for quality control.

Detection of ESBL production
The production of ESBL was tested for the 103 urinary CREc isolates using the combined disk method with cefotaxime (30 µg) and ceftazidime (30 µg) with or without clavulanic acid (10 µg) (Clinical Laboratory Standard Institute, 2022).An increase in the zone size ≥5 mm for cefotaxime and ceftazidime with or without clavulanic acid was considered to indicate ESBL production (Clinical Laboratory Standard Institute, 2022).E. coli ATCC 25922 was used as a negative control.

Biofilm formation assay
The biofilm production of the urinary CREc isolates was determined using the Congo red agar (CRA) method, as previously described (Tajbakhsh et al., 2016).Briefly, the CREc isolates were cultured on CRA made of brain heart infusion agar with 36 g/L sucrose and Congo red dye 0.8 g/L, and they were incubated at 37 • C for 24-48 h.The biofilm production was characterized based on six color tones of colonies: very black, black, and almost black (interpreted as strong, moderate, and weak biofilm producers, respectively) and bordeaux, red, and very red (reported as non-biofilm producers).

Whole-genome sequencing and genome assembly
Bacterial DNA was extracted using the Applied Biosystems™ MagMAX™ DNA Multi-Sample Ultra 2.0 Kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions.All 103 isolates were sequenced using the short-read sequencing Illumina platform, HiSeq 3000 (Illumina, San Diego, CA, USA) (Takeuchi et al., 2022).
The QRDR mutations in gyrA, gyrB, parC, and parE were determined by performing a multiple sequence alignment of the protein sequence via Clustal Omega (https: //www.ebi.ac.uk/Tools/msa/clustalo/) using the genes of E. coli strain K-12 (MG1655) as a reference.

Statistical analysis
Data were analyzed using SPSS version 26.0 (Chicago, IL, USA).Fisher's exact test was used to establish the association between biofilm formation ability and adhesion factor genes in the 103 urine isolates of CREc.P < 0.05 was considered statistically significant.
To determine genetic relationships, a core-genome SNP-based phylogeny of the 103 CREc genomes is shown in Fig. 2. Most CREc urinary isolates were divided into six large clusters based on the main branches of the tree.The first cluster comprised ST361.The second cluster contained ST46.The third cluster consisted of ST34, ST167, and ST1702.The fourth cluster contained ST1193, ST131, ST2011, ST457, and ST354.The fifth cluster consisted of ST38 and ST405.The last cluster contained mainly ST410 strains and ST88 and ST448 (Fig. 2).The present study revealed that the CREc urinary isolates mainly belonged to the phylogroup C, which are considered to be associated with commensal status or intestinal pathotypes (Tenaillon et al., 2010).Nevertheless, the UTI E. coli isolates were significantly more frequent in the phylogroups B2 and D (Amarsy et al., 2019).Our study showed that phylogroup D had second prevalence and phylogroup B2 had rare status.The Clermont phylogroup C contained mainly ST410.This ST is an emerging and international high-risk clone worldwide, associated with a large number of clinical infections (Roer et al., 2018).In Thailand, ST410 has been identified in clinical isolates that co-harbored mcr and bla NDM (Boueroy et al., 2022;Paveenkittiporn et al., 2021).2).No APEC was found in the present study.Although most urinary CREc in this study could not be classified, they seem to be commensal or non-pathogenic pathotypes.Among the three investigated virulence factor classes, 20.39% (21/103) of the isolates were identified as having two UPEC-associated markers (  2).In total, 37 virulence genes were examined and categorized as adhesion molecules, invasion genes, toxins, iron uptake molecules, autotransporter systems, and protection factors, according to the genotypic profiles of the 103 urinary CREc isolates (Table S3).Overall, fim (100%), csg (100%), and ibeABC (100%) were the virulence genes with the highest distributions among isolates, followed by ecp (98.06%), espl (97.09%), hlyE (97.09%), eae (78.64%), cah (67.96%), upaG/ehaG (65.05%), tia (49.51%), and fyuA (47.57%) (Table S3).This was not surprising because these virulence factors, such as type 1 fimbriae (fim) and the major pilus subunit (ecpA), have also been reported in commensal E. coli (Pusz et al., 2014;Blackburn et al., 2009)  Almost all the ExPEC isolates in the presents study belonged mainly to the phylogenetic groups B2 and D and carried various virulence factors compared to the other phylogenetic groups (Table S4).Concordance with another report identified a higher number of virulence genes in phylogroups B2 and D compared to other groups in UTI E. coli isolates (Lee & Moon, 2016).In contrast, most urinary CREc isolates in the present study were not classified as any pathotype, which was consistent with them being Clermont types C and A (73.79%; 76/103).
In the biofilm assay, 50.49% (52/103) of the urinary CREc isolates produced weak-tostrong biofilm formation (Tables 1 and S5).Among these isolates, 20%, 9%, and 23% were strong, moderate, and weak biofilm producers, respectively.More than one-half (71.84%; 74/103) of these isolates could also be classified as weak and non-biofilm formation.In agreement with our findings, biofilm formation has been reported as low in prevalence in commensal E. coli strains (Karam, Habibi & Bouzari, 2018).According to another report, biofilm formation has greater carriage of adhesin genes (Fim, Pap, Sfa, and Afa) compared to non-biofilm formers (Katongole et al., 2020).Nevertheless, we did not observe any relationship between biofilm formation ability and the carriage of any adhesin genes (Table S6).Consistent with another report, the biofilm formation in carbapenem-resistant E. coli from the UTIs did not reveal any significant difference in the carriage rate of fimbriae genes (Chen et al., 2023).
Carbapenems are the drug of choice for complicated UTIs caused by multidrug-resistant Enterobacteriaceae, especially those due to ESBL-producing E. coli (Amladi et al., 2019).Although FQs are the drug of choice to treat UTIs treatment, FQ resistance is an increasing issue that causes concern in several countries, including Greece, Senegal, and Saudi Arabia (Yang et al., 2010;Chaniotaki et al., 2004).National Antimicrobial Resistance Surveillance Thailand (http://narst.dmsc.moph.go.th/) reported that E. coli urinary isolates had a high percentage of resistance against ciprofloxacin (67.6%) and levofloxacin (66.6%).
Our urinary CREc isolates exhibited high resistance to ciprofloxacin (97.09%) and levofloxacin (94.17%).The high frequency among FQ-resistant determinants in urinary CREc isolates in our study raises serious concerns and the need to identify alternative antibiotics for UTI therapy.One choice is nitrofurantoin as a suitable candidate for the treatment of UTIs caused by multidrug-resistant pathogens (Munoz-Davila, 2014).Several studies have reported a low resistance rate of urinary E. coli to NFT, including 18.4% in Turkey (Pullukçu et al., 2007), 6.2% in Spain, and3.7% in England (Farrell et al., 2003;Gobernado et al., 2007).Resistance rates to NFT in the CREc urinary isolates in the present study showed a low prevalence (33.98%) compared to other antibiotics.The resistance rate of NFT remained virtually unchanged, suggesting that it may be an important and economical option for UTI treatment (McKinnell et al., 2011).
In this study, CREc urinary isolates harboring blaNDM-5 and blaCTX-M-15 were the majority of carbapenemase and ESBL genes, respectively.Other studies have revealed that NDM, especially bla NDM-5 , is the main carbapenemase gene in CRE in Thailand (Takeuchi et al., 2022;Paveenkittiporn et al., 2021).The most widely distributed bla CTX-M enzyme worldwide is bla CTX-M-15 , which is commonly found in E. coli isolates causing UTI (Price et al., 2013).
In the present study, the aac(6 )-Ib-cr gene was predominant (58%, 58/100) among the ciprofloxacin-resistant CREc isolates.Similarly, this gene was highly frequent in ciprofloxacin-resistant E. coli from UTI patients in Nigeria (Eghieye et al., 2020).In contrast, qnrA, qnrB, and qnrS were reported as highly prevalent in other studies (Eghieye et al., 2020;Ramírez-Castillo et al., 2018) but were low in the present study.Furthermore, the oqxAB gene was identified in 5.92% of the FQ non-susceptible E. coli isolated from UTI patients in Taiwan (Kuo et al., 2022); however, this was not identified in the present study.
Additionally, strain no.AMR0278 carried the bla OXA-181 and qnrS1 genes located on a nonconjugative ColKP3-type plasmid that was 51,478 bp long (Fig. 4).This plasmid contained type IV secretory genes upstream and the IS6 family genes upstream and downstream of the bla OXA-181 and qnrS1 genes.In addition, this plasmid showed high similarity to the plasmids of E. coli and Klebsiella pneumoniae from Switzerland, the Netherlands, Ghana, Egypt, France, China, and the United States (Fig. 4).The ColKP3 plasmid type harbored almost all classes of AMR genes, such as bla OXA-181 and bla OXA-232 , and is common in K. pneumoniae (Ragupathi et al., 2019;Weng et al., 2020;Naha et al., 2021).The present study showed that a nonconjugative ColKP3-type plasmid carried the bla OXA-181 and qnrS genes in an E. coli isolate.
A limitation of the present study was that we selected only five isolates to do complete genomes using the ONT platform, which may not have been representative of all the isolates in this study.Although we chose from the representative strains in each cluster of the phylogenetic tree, it could not cover all the studied strains that may have presented new characteristics.Therefore, further study with current CRE isolates should be undertaken and the results compared.

CONCLUSIONS
NDM-5 and CTX-M-15 were the predominant carbapenemase and ESBL genes, respectively, among the 103 urinary CREc isolates.The mutations in the QRDR of the gyrA and parC genes were the most predominant, followed by the presence of the PMQR determinant aac(6 )-Ib.The coexistence of bla CTX-M-15 and aac(6 )-Ib was widely observed.
Among the CREc isolates, resistance to ciprofloxacin and levofloxacin was greater than 90%.
NFT maintained high sensitivity rates greater than 60%.They can be chosen as empirical antimicrobial treatments for uncomplicated UTIs.The results of this study highlighted the significant frequency of FQ resistance in CREc urinary isolates in Thailand, which is reflected in infection control and must be adopted to prevent further spread, emphasizing the need to expand antimicrobial drug resistance screening at hospitals.Therefore, to decrease the prevalence of FQ-resistant determinants in urinary CREc organisms in the future, continual epidemiologic surveillance and monitoring of antimicrobial prescriptions and consumption are required.• Shigeyuki Hamada analyzed the data, authored or reviewed drafts of the article, and approved the final draft.
• Anusak Kerdsin conceived and designed the experiments, authored or reviewed drafts of the article, and approved the final draft.

Human Ethics
The following information was supplied relating to ethical approvals (i.e., approving body and any reference numbers): Ethical approval was obtained from the Ethics Committee of Osaka University Graduate School of Medicine, Osaka, Japan.The ethics approval number was 14468 and 22014.This study was conducted according to the principles of the Declaration of Helsinki.The need for informed consent was waived.

Figure 1
Figure 1 Provinces where samples were collected between 2015 and 2020.Samples were collected from eight provinces nationwide: Surin, Udon Thani, Sakon Nakhon, Nakhon Ratchasima, Nakhon Phanom, Tak, Surat Thani, and Chumphon.(A geographical information system (GIS) software QGIS (version 2.18.28) was used to create a study map).The total number of isolates is shown in the map, and the number and percentage of ST is presented in the pie chart.Full-size DOI: 10.7717/peerj.16401/fig-1

Figure 3
Figure 3 Protein level alignment of the bla NDM carrying plasmid in five CREc isolates.Two different plasmid replicon types including IncFI and IncFI/IncQ were identified (A).Protein level alignment of the IncFI plasmid (no.AMR0278) carried NDM-5 with previously found in strains isolated in other country (B).Genes of interest are colored and defined in the legend.Alignments were made using clinker and clustermap.js and visualized using ApE v3.0.8.(C).Horizontal arrows indicate location, size, and direction of transcription, and visualized Proksee.Full-size DOI: 10.7717/peerj.16401/fig-3

Figure 4
Figure 4 Protein level alignment of the ColKP3-type plasmid carrying bla OXA-181 and qnrS (no.AMR0278) with previously found plasmids in strains isolated in other country.Horizontal arrows indicate location, size, direction of transcription, and orientation of open reading frames.Alignments were made using clinker and clustermap.js.Full-size DOI: 10.7717/peerj.16401/fig-4