MinION nanopore sequencing identifies the position and structure of bacterial antibiotic resistance determinants in a multidrug-resistant strain of enteroaggregative Escherichia coli

The aim of this study was to use single-molecule, nanopore sequencing to explore the genomic environment of the resistance determinants in a multidrug-resistant (MDR) strain of enteroaggregative Escherichia coli serotype O51 : H30, sequence type (ST) 38. Sequencing was performed on the MinION Flow cell MIN-106 R9.4. Nanopore raw FAST5 reads were base-called using Albacore v1.2.1, converted to FASTA and FASTQ formats using Poretools v0.6.0, and assembled using Unicycler v0.4.2, combining the long-read sequencing data with short-read data produced by Illumina sequencing. The genome was interrogated against an antimicrobial resistance (AMR) gene reference database using blast. The majority of the 12 AMR determinants identified were clustered together on the chromosome at three separate locations flanked by integrases and/or insertion elements [region 1 –catA, bla OXA-1, aac(6′)-Ib-cr, tetA and bla CTX-M-15; region 2 – dfrA1 and aadA1; region 3 – catA, bla TEM-1, tetA and sul2]. AMR determinants located outside these three regions were a chromosomally encoded bla CMY-16, mutations in gyrA and parC, and two plasmid-encoded AMR determinants, bla OXA-181 and qnrS1 located on the same IncX3 plasmid. Long-read analysis of whole genome sequencing data identified mobile genetic elements on which AMR determinants were located and revealed the combination of different AMR determinants co-located on the same mobile element. These data contribute to a better understanding of the transmission of co-located AMR determinants in MDR E. coli causing gastrointestinal and extra-intestinal infections.


INTRODUCTION
Enteroaggregative Escherichia coli (EAEC) cause a range of gastrointestinal symptoms including acute or persistent, watery or mucoid diarrhoea, often accompanied by severe abdominal pain [1]. Studies have shown that EAEC make a significant contribution to the burden of diarrhoeal disease globally, and are a common cause of travellers' diarrhoea in the UK [2]. Certain sequence types (STs), such as ST38, also cause extra-intestinal infections, including sepsis and urinary tract infections [3]. There are no known animal reservoirs and transmission is most likely person-to-person via the faecal oral route.
From the analysis of the short-read sequencing data, it was not possible to determine the whether the resistance determinants in this multidrug-resistant (MDR) strain of EAEC were plasmid-encoded or if they had been incorporated into the chromosome. The aim of this study was to use singlemolecule, nanopore sequencing to explore the genomic environment of the resistance determinants in this MDR strain.

METHODS
DNA extraction and nanopore sequencing DNA was extracted using the Wizard Genomic DNA Purification kit (Promega). Library preparation was performed using the 1D Genomic DNA sequencing kit SQK-LSK108 (Oxford Nanopore Technologies) with the omission of DNA shearing and DNA repair steps to prevent further DNA fragmentation. Library preparation was initiated at the DNA end-prep step using NEB repair modules (New England Biolabs). All bead washing steps were performed using AMPure XP beads (Beckman Coulter). The final 80 µl prepared library was sequenced. Sequencing was performed on the MinION using a FLO-MIN-106 R9.4 Flow cell (Oxford Nanopore Technologies) using the MinKNOW software for the full 48 h run time with no alterations to any voltage scripts.
Illumina sequencing and detection of AMR determinants from the short-read data were performed as described previously [2].

RESULTS AND DISCUSSION
The genome of EAEC O51 : H30 ST38 assembled into a single chromosomal contig of 5 492 922 bp and five plasmids ( Table 1). Located on the chromosome at positions 2 087 027-2 109 149 were genes encoding pap pili/fimbriae, characteristic of extra-intestinal E. coli. The largest plasmid was 129 627 bp (p266917_2_01) and belonged to replicon type IncFIB based on the repA sequence. This plasmid carried known EAEC virulence genes, including the universal regulator aggR, the dispersin-encoding aap, the aggregative transported aat and the aggA genes encoding fimbriae type I. The remaining four plasmids (p266917_2_02, p266917_2_03, p266917_2_04 and p266917_2_05) were smaller in size ranging from 33 288 to 97 124 bp (Table 1). These plasmids have been previously described in E. coli and Klebsiella species, and isolated in globally dispersed regions from clinical samples, food, animals and the environment (Table 1).
Despite the number of plasmids harboured by this isolate, analysis of the assembled genome indicated that the majority of the AMR determinants were clustered together on the chromosome in three separate regions flanked by integrases and/or insertion elements. AMR region 1 was located on the chromosome at positions 1 686 326-1 719 287 (32 961 bp) between a transfer messenger RNA gene (ssrA) and a tyrosine recombinase (xerC) (Fig. 1a). The AMR determinants located on this integron were catA, bla OXA-1 , aac(6¢)-Ib-cr, tetA and bla CTX-M-15 , conferring reduced susceptibility to chloramphenicol, ampicillin, streptomycin and ciprofloxacin, tetracycline and the third-generation cephalosporins. bla CTX-M-15 is the most abundant CTX-M gene in ESBL (extended-spectrum beta-lactamase)-producing E. coli causing human infections [8]. Analysis using BLASTn indicated that the AMR determinants on this transposon originated from two different fragments of DNA originating from different plasmids (Accessions: NC_014384.1 and

IMPACT STATEMENT
The implementation of whole genome sequencing (WGS) for routine public health surveillance has enabled us to monitor trends in antimicrobial resistance (AMR) gene content in Escherichia coli and provides real-time data on emerging resistance patterns nationally and internationally. Understanding the genomic environment of these AMR determinants, with respect to whether they are chromosomally encoded or plasmid-encoded, and whether they are co-located is essential for monitoring transmission of AMR and assessing the risk to public health. Short-read WGS data can identify the presence or absence of AMR determinants but not their genomic architecture. In this study, long-read analysis of WGS data enabled us to determine the mobile genetic elements on which the key AMR determinants are located, and to characterize the combination of different AMR determinants co-located on the same mobile element. A combination of short-and long-read WGS data contributes to a better understanding of the transmission of colocated AMR determinants in multidrug-resistant E. coli. MF353155.1) inserted into a fragment of DNA that originated from a strain of Enterobacter cloace (Fig. 1a).
In addition to the PMQRs qnrS1 and aac(6¢)-Ib-cr, which exhibit the potential to induce reduced susceptibility to the fluroquinolones, EAEC O51 : H30 also had one mutation in gyrA[83 : S-L] and two in parC[80 : S-I;84 : E-V]. The combination of all these fluroquinolone resistance determinants resulted in the isolate exhibiting an MIC>0.5 mg l À1 to ciprofloxacin.
Of the two plasmid-encoded AMR determinants, bla OXA-181 is a bla OXA-48 -like carbapenemase conferring resistance to penicillins and carbapenems, and is commonly found on a 51 kb IncX3 plasmid [14], as observed in this study. The bla OXA-181 gene was initially identified in Enterobacter cloacae and Klebsiella pneumoniae isolates in India in 2007 [15]. Enterobacteriaceae isolates producing bla OXA-181 are globally distributed, although in most cases the patients report recent travel to the Indian subcontinent [16,17]. The presence of bla OXA-181 on a self-transmissible IncX3 plasmid is of significance, as IncX3 plasmids have been found as a common vehicle mediating the dissemination of other carbapenemases, and have the potential to disseminate widely [15]. Due to the co-location of qnrS1 on the same plasmid, treatment with ciprofloxacin may indirectly select for resistance to the carbapenems, as well as the fluroquinolones.
A previous study in the UK showed that E. coli ST38 was the ST most commonly associated with bla OXA-48 -like genes, and 18 % (62/351) had bla OXA-181 [17]. The authors suggested that the accumulation of bla OXA-48 -like carbapenemases within the UK is due to repeated importations, coupled with both the spread of successful clones (e.g. E. coli ST38) and, in particular, the dissemination of successful plasmids [17].
Finally, an assembly utilizing only the ONT reads was generated with Unicycler (v0.4.2) with default parameters. To estimate the accuracy of the derived assembly, Illumina reads were mapped using BWA v0.7.13 and Samtools v1.1  which the key AMR determinants are located, and identifies the combination of different AMR determinants co-located on the same mobile element, but may lack the accuracy required to identify AMR associated with mutations in chromosomal genes [19,20]. A combination of short-and long-read WGS data facilitates the detailed analysis of AMR in E. coli, contributing to a better understanding of the transmission of co-located AMR determinants in MDR E. coli causing gastrointestinal and extra-intestinal infections.