Whole genome sequence analysis of ESBL-producing Escherichia coli recovered from New Zealand freshwater sites

Extended-spectrum beta lactamase (ESBL)-producing Escherichia coli are often isolated from humans with urinary tract infections and may display a multidrug-resistant phenotype. These pathogens represent a target for a One Health surveillance approach to investigate transmission between humans, animals and the environment. This study examines the multidrug-resistant phenotype and whole genome sequence data of four ESBL-producing E . coli isolated from freshwater in New Zealand. All four isolates were obtained from a catchment with a mixed urban and pastoral farming land-use. Three isolates were sequence type (ST) 131 (CTX-M-27-positive) and the other ST69 (CTX-M-15-positive); a phylogenetic comparison with other locally isolated strains demonstrated a close relationship with New Zealand clinical isolates. Genes associated with resistance to antifolates, tetracyclines, aminoglycosides and macrolides were identified in all four isolates, together with fluoroquinolone resistance in two isolates. The ST69 isolate harboured the bla CTX-M-15 gene on a IncHI2A plasmid, and two of the three ST131 isolates harboured the bla CTX-M-27 genes on IncF plasmids. The last ST131 isolate harboured bla CTX-M-27 on the chromosome in a unique site between gspC and gspD. These data highlight a probable human origin of the isolates with subsequent transmission from urban centres through wastewater to the wider environment.


INTRODUCTION
Antimicrobial resistance (AMR) is becoming an increasing problem in the treatment of community-acquired infections [1,2]. In New Zealand, extended spectrum beta lactamase (ESBL)-producing Escherichia coli are commonly associated with multidrugresistant urinary tract infections (UTIs) [3].
A key step in reducing the dissemination of AMR is through understanding where transmission occurs. An important pathway for the community spread of antimicrobially resistant bacteria is through person-to-person transmission [4], but other transmission pathways may also be relevant, including contact with animals, ingestion of food products or indirectly through contaminated waterways [5][6][7][8]. Recreational swimming has been identified as a risk factor for ESBL-producing E. coli-associated community infections [9]. Waterways including rivers, lakes and some beach waters have been identified as a vector for ESBL-producing OPEN ACCESS E. coli [10][11][12][13][14][15][16], with the main origin of these bacteria being human faeces rather than emergence of AMR within the waterway itself [17]. It has been found that the concentration of ESBL-producing E. coli can be high (1×10 2 -1×10 5 c.f.u. l -1 ) in treated sewage being discharged into waterways [16]. Genetically similar isolates have also been identified in the few studies that have been undertaken comparing water with clinical ESBL-producing isolates [10,18].
In New Zealand, E. coli levels in waterways are measured and faecal source tracking has identified the main sources of faecal pollution in these waterways [19,20]. However, little is known about the source of antimicrobially resistant E. coli bacteria. Here we characterize four ESBL-producing E. coli isolated from a New Zealand stream and explore their genetic relationship with New Zealand human clinical isolates.

Sample sites and collection
Three freshwater sites with separate catchments were visited on 11 occasions over a 13 month period from March 2020 to March 2021. All three sites are in the general vicinity of Dannevirke in the Tararua region of the lower North Island, New Zealand (Fig. 1 We describe the phenotypic and genetic basis for multidrug-resistant Escherichia coli isolated from freshwater in New Zealand. Comparative phylogenetic analysis with clinical isolates provides evidence of a human source and transmission from wastewater to the wider environment, highlighting the need for a One Health approach to investigate antibiotic-resistant bacterial transmission. On each sampling occasion, water was stored in a chilly bin with frozen ice blocks at approximately 4 °C and processed within 4 h at the Hopkirk Research Institute (Palmerston North).

Bacterial culture and antibiotic susceptibility testing
Water (500 ml) was filtered through a 0.45 µm nitrocellulose filter using a bench-top negative pressure system, and incubated in 10 ml EC broth (Fort Richard) for 18 h at 35 °C. After enrichment, 10 µl of broth was inoculated onto CHROMagar ESBL (Fort Richard), and streaked for individual colonies. Agar plates were incubated for 18 h at 35 °C and examined for pink colonies indicative of ESBL-producing E. coli. In parallel with the recovery of ESBL-producing E. coli, 500 ml of water was filtered through a separate 0.45 µm filter, and incubated in 20 ml Bolton's broth (Fort Richard) for enrichment of Campylobacter species as part of a parallel study. The broth was incubated at 42 °C in a microaerobic atmosphere (85 % N 2 , 10 % CO 2 , 5 % O 2 ) for 48 h before an aliquot was plated onto modified charcoal-cefoperazone-deoxycholate agar (mCCDA) agar (Fort Richard) with a sterile cotton swab. mCCDA plates were examined after 48 h of incubation at 42 °C in a microaerobic atmosphere.
Putative ESBL-producing E. coli strains were sub-cultured onto CHROMagar ECC and MacConkey agar plates, then purified on sheep blood agar plates and identified using MALDI-TOF MS using the full 'in tube formic acid extraction' method (Bruker) [21]. Primary evaluation for AMR was undertaken against cefotaxime (30 µg), cefoxitin (30 µg), ceftazidime (30 µg) or cefpodoxime (10 µg), tetracycline (30 µg), streptomycin (10 µg) and ciprofloxacin (5 µg), and interpreted according to CLSI guidelines using Kirby-Bauer disc diffusion tests [22]. The AmpC and ESBL AMR phenotype was confirmed for isolates resistant to either cefoxitin and cefotaxime and/or cefpodoxime according to EUCAST guidelines using either a three-disc (D69C AmpC disc test; Mast Group) or double-disc comparison assay (D62C cefotaxime and D64C ceftazidime ESBL disc tests; Mast Group), respectively [23]. The AmpC-producing E. coli NZRM4402 and the ESBL-producing Klebsiella pneumoniae NZRM3681 were used as positive controls in the AmpC and ESBL confirmatory disc assays, respectively, and the susceptible E. coli NZRM916 was used as a negative control.

Whole genome sequencing and assembly
Genomic DNA was extracted from the ESBL-producing isolates grown in lysogeny broth, using the Promega wizard genomic DNA purification kit as previously described [24]. Sequencing was performed using both short-and long-read technologies. For short-read sequencing, libraries were prepared using the Nextera XT DNA library preparation kit (Illumina) and sequencing was performed using an Illumina HiSeq X with 2×125 bp paired-end reads (Novagene). Raw sequence reads were trimmed and assembled using the Nullarbor bioinformatics pipeline (v.2.0.20181010) with default settings [25]. The trimming of reads in this pipeline was carried out using trimmomatic (v.0.39) [26]. Long-read sequencing using Oxford Nanopore Technologies (ONT) was then carried out as previously described [24] followed by base-calling using Guppy (v.5.0.7). The ONT reads were demultiplexed using qcat (v.1.1.0), trimmed using porechop (v.0.2.4) and filtered using Filtlong (v.0.2.0) in which 95 % of the reads were kept with a minimum length of 1000 bp and target number of bases of 500 Mb (100× depth). A hybrid assembly using both the illumina and ONT reads was carried out using unicycler (v.0.4.9b).

Summary of sample sites
Three separate freshwater sites were each sampled on 11 occasions over a 13 month period. No ESBL-producing E. coli were identified from Site 2 or Site 3. However, four ESBL-producing E. coli (AGR4587, AGR5151, AGR6128 and AGR6137) were isolated from Site 1, Tapuata Stream (Dannevirke, New Zealand), and all four strains were isolated using mCCDA agar plates. Both Bolton's broth and mCCDA agar have previously been problematic for isolating Campylobacter, with the increasing isolation of ESBL-producing Enterobacteriaceae [50,51]. No ESBL-producing E. coli were isolated using enrichment in EC broth followed by plating onto CHROMagar ESBL, although subsequent subculture confirmed that the four ESBL-producing E. coli were able to grow on CHROMagar ESBL. This highlights the difficulties with using various culture-based methods for determining the prevalence of ESBL-producing E. coli in different sources [52].
Although much of the Tapuata Stream catchment is situated to the north-west of Dannevirke, a small tributary, the Mangapurupuru Stream, runs through and beneath the town of Dannevirke, flowing into the Tapuata Stream about 200 m upstream of Site 1. Studies have found ESBL-producing E. coli and other antibiotic-resistant bacteria in waterways where treated sewage outlets are located [53]. Although our sampling site was not located downstream of a treated sewage outlet, there may have been leakage from ageing wastewater infrastructure or raw sewage overflow. During high rainfall events raw sewage may overflow into storm water drains. However, in this study all the ESBL-producing E. coli were isolated during the summer or early autumn when rainfall was low. Previous studies have found a higher prevalence of third-generation cephalosporin-resistant E. coli in autumn [18,54].

Genetic relatedness
Whole genome sequence analysis determined that three of the four ESBL-producing E. coli strains were ST131 and one strain was ST69 (Table 1), with the difference in the number of SNPs ranging from 92 to 1066 between the three ST131 strains (Table S3). ST131 has previously been shown to be the predominant sequence type (41-54 %) amongst New Zealand clinical E. coli isolates, and ST69 has a prevalence of approximately 3 % [3,55].
A phylogenetic comparison with other previously sequenced New Zealand ST131 and ST69 strains demonstrated that the four water strains were closely related to New Zealand clinical strains (Figs 2 and 3). ST131 is the predominant lineage associated with ESBL-associated human urinary tract and blood infections globally [56,57], although other lineages such as ST69 are on the rise [56,58]. Both ST131 and ST69 strains have been isolated from other animals, particularly poultry and dogs, worldwide [27,[59][60][61].
The three ST131 strains belonged to clade A and all harboured a bla CTX-M-27 gene. Two (AGR4587 and AGR6128) of the three ST131 water strains had a fimH41-type allele whereas AGR5151 had a fimH42 allele. ST131 clade A strains frequently carry a fimH41 allele [40,62]. The close clustering (<10 SNPs) of these three water strains with clinical strains previously isolated in New Zealand suggests that the origin of these strains is humans. Additionally, there was evidence of clonal spread of CTX-M-27-producing E. coli within New Zealand over at least the past 6 years. The ST69 strain AGR6387 harboured the bla CTX-M-15 gene and clustered (43-

Distribution of AMR and virulence genes
The four water strains all displayed an ESBL-producing phenotype in agreement with their genotype. A multidrug resistance phenotype and genotype was also observed in the four water isolates (Table 1, Fig. 2b,c) with the genes associated with resistance comprising the four antibiotic classes: beta-lactams, antifolates (dfrA17, sul1 and sul2), tetracyclines [tet(A)), aminoglycosides (aac(3)-IId, aadA5, aph(6)-Id, aph(3″)-Ib), and macrolides [mph(A)]. Additionally, the ST131 strain AGR6128 harboured mutations in both the gyrA and parC genes, in concordance with its fluoroquinolone-resistance phenotype ( Table 2). The acquisition of fluoroquinolone resistance in clade A ST131 strains is reported to be rare, with fluoroquinolone resistance being more commonly associated with the C2 H30 subclade [64]. However, a recent study found ST131 clade A strains isolated from wastewater harboured a higher prevalence of fluoroquinolone resistance compared with clade C strains [65]. In our study, the ST69 strain AGR6137 carried a qnrB1 gene, which has previously been found to be carried by bla CTX-M-15 plasmids [66][67][68] but to our knowledge rarely in bla CTX-M-27 plasmids.
A multidrug resistance genotype (ranging from three to ten genes) was observed in most of the New Zealand ESBL-producing strains isolated from humans: 55 of the 64 (86 %) clade A ST131 strains and 20 of the 25 (80 %) ST69 strains. Multidrug-resistant ESBL-producing E. coli have been isolated from waterways worldwide [66,69].
All four strains carried multiple virulence genes associated with extraintestinal pathogenicity [70,71], such as papA, papC, sfaC, afaC, kpsM and iutA (Table 3). In addition, they carried the toxin genes sat (carried by strains AGR4587, AGR5151 and AGR6128) and vat (carried by AGR6137) as well as the siderophore gene chuA (carried by strains AGR4587, AGR5151 and AGR6128), which are genes typical of uropathogenic E. coli. This supports the notion that that these water strains originated from humans. No Shigatoxin-associated genes were detected.

Mobile genetic elements analysis
The ST69 strain harboured an IncHI2A plasmid encoding a bla CTX-M-15 gene as well as an IncF plasmid ( Table 4). The three ST131 strains harboured IncF plasmids (Table 4), with two strains encoding the bla CTX-M-27 gene on the plasmid. The third ST131 strain, AGR4587, harboured bla CTX-M-27 on the chromosome, which was flanked by the insertion sequence elements ISEcp1 (belonging to the IS1380 family of insertion sequences) and IS903B. The insertion site of the bla CTX-M-27 gene was between the gspC and gspD genes, a possible unique site that to our knowledge has not been previously described [72,73]. ISEcp1 insertion sequence elements have been commonly associated with the bla CTX-M genes [3,74,75], and studies suggest it was also involved with the original mobilization of the bla CTX-M gene from the chromosome of Kluyvera species to a plasmid [76]. The bla CTX-M genes have been reported to be chromosomally encoded (particularly the bla CTX-M-15 variant) [64,[72][73][74]77], but to our knowledge this is the first report of a chromosomally encoded bla CTX-M-27 . All five plasmids contained the genes required for conjugation, including traD (encoding a coupling protein) and traI (encoding a relaxome). A core gene analysis of the IncF plasmids with their closest relatives (Table S1) was carried out (Fig. 4a). ESBL-producing ST131 strains have previously been found to predominantly carry IncF plasmids with multiple plasmid replicons including the IncFII, IncFIA and/or IncFIB types [68]. IncF-carrying bla CTX-M plasmids containing the multiple plasmid replicons IncFII, IncFI and Col156 representing sequence type [F1:A2:B20] have been associated with STs other than ST131, including ST38 [62]. The IncF plasmids from the three ST131 strains contained the same backbone and similar resistance genes but were distinct from the ST69 IncF plasmid (Fig. 4a,b). However, differences arose in their type and number of transposases. For example, in AGR6128 bla CTX-M-27 is flanked by an IS6 family transposase IS26 on either side, whereas in AGR5151 bla CTX-M-27 is flanked by an IS6 family transposase IS15 and an IS5 family transposase IS903 (Fig. 4c).
The closest relatives of the pAGR6137a plasmid were IncHI2 plasmids previously isolated from Enterobacter spp. (Table S2). This IncHI2 plasmid has been isolated from a variety of Enterobacteriaceae species including the Enterobacter cloacae complex, Salmonella enterica, Klebsiella pneumoniae, Citrobacter freundii and E. coli [78][79][80][81]. It has been suggested that the reason for the successful spread of this IncHI2 plasmid is because it harbours numerous metal resistance genes [80]. Studies suggest that the ter operon, encoding tellurite resistance, is found on all IncHI2 plasmids [82].
In conclusion, the isolation and whole genome sequencing of the four ESBL-producing E. coli isolates collected from a local freshwater site shows the importance of taking a One Health approach in understanding the sources and transmission pathways of AMR bacteria. The use of long-read sequencing enabled the genetic context of the bla CTX-M-27 and bla CTX-M-15 genes to be elucidated. Additionally, differences in the order and content of resistance gene cassettes between strains highlight the malleability of resistance genes and their associated mobile elements.

Funding information
Funding for the Food Integrity Programme (PRJ0126322) was through the AgResearch Strategic Science Investment Funding originating from the New Zealand Ministry of Business, Innovation and Employment. A.C. also received funding for the long-read sequencing undertaken in this work from DairyNZ in support of MBIE project C10X1908 (Novel discriminatory tests for E. coli to improve water quality assessments), but DairyNZ staff did not play a role in the study or in the preparation of the article or decision to publish.