Epidemic HI2 Plasmids Mobilising the Carbapenemase Gene blaIMP-4 in Australian Clinical Samples Identified in Multiple Sublineages of Escherichia coli ST216 Colonising Silver Gulls

Escherichia coli ST216, including those that carry blaKPC-2, blaFOX-5, blaCTX-M-15 and mcr-1, have been linked to wild and urban-adapted birds and the colonisation of hospital environments causing recalcitrant, carbapenem-resistant human infections. Here we sequenced 22 multiple-drug resistant ST216 isolates from Australian silver gull chicks sampled from Five Islands, of which 21 carried nine or more antibiotic resistance genes including blaIMP-4 (n = 21), blaTEM-1b (n = 21), aac(3)-IId (n = 20), mph(A) (n = 20), catB3 (n = 20), sul1 (n = 20), aph(3”)-Ib (n = 18) and aph(6)-Id (n = 18) on FIB(K) (n = 20), HI2-ST1 (n = 11) and HI2-ST3 (n = 10) plasmids. We show that (i) all HI2 plasmids harbour blaIMP-4 in resistance regions containing In809 flanked by IS26 (HI2-ST1) or IS15DI (HI2-ST3) and diverse metal resistance genes; (ii) HI2-ST1 plasmids are highly related to plasmids reported in diverse Enterobacteriaceae sourced from humans, companion animals and wildlife; (iii) HI2 were a feature of the Australian gull isolates and were not observed in international ST216 isolates. Phylogenetic analyses identified close relationships between ST216 from Australian gull and clinical isolates from overseas. E. coli ST216 from Australian gulls harbour HI2 plasmids encoding resistance to clinically important antibiotics and metals. Our studies underscore the importance of adopting a one health approach to AMR and pathogen surveillance.


Introduction
Bacterial isolates resistant to antimicrobials have been isolated from wildlife [1] and there are concerns that wild animals act as critically important vectors and reservoirs for antimicrobial resistant bacteria and antibiotic resistance genes (ARGs) [2,3]. Migratory birds have the potential to spread multidrug resistant (MDR) bacteria and ARGs [4,5] posing a significant threat to biosecurity, particularly in countries that practice sound antimicrobial stewardship [1,6]. An increasing number of migratory birds have been found to host antimicrobial resistant bacteria with resistance to diverse antibiotics including those referred to as critically important (CIA) to human health such as the extended spectrum β-lactams, fluoroquinolones, carbapenems [7][8][9] and colistin [10][11][12][13].
Plasmids are vehicles that capture, assemble, maintain and spread ARGs genes, heavy metal resistance genes and virulence-associated genes (VAGs) [6] and provide flexibility to bacterial genomes by the diverse genetic cargo they carry [14]. Genes encoding resistance Microorganisms 2021, 9, 567 2 of 22 to antibiotics, heavy metals and virulence genes often coassemble on the same plasmid, mediated in part by the activity of insertion elements such as IS26 [15][16][17][18]. These factors allow the emergence of lineages that carry complex resistance regions and virulence gene profiles [14,19]. As such there is an urgent need to address the shortage of completed plasmid sequences in public databases and identify plasmids that carry virulence and antibiotic resistance genes [15,17,20] and plasmid hybrids [16,21]. Hybrid Escherichia coli carrying combinations of virulence genes from different pathovars are increasingly recognised as an emerging threat to human and animal health [22].
Dissemination of emergent and dominant multidrug resistant bacterial clades is a major driving force behind the global spread of antibiotic-resistant bacteria [14,23]. E. coli ST216 is known to carry genes encoding resistance to a broad range of antibiotics including those of clinical relevance [24][25][26]. However, little is known about ST216 virulence and antibiotic resistance gene cargo that it carries and the hosts it occupies. E. coli ST216 belongs to 'commensal' phylogroup A. With the exception of ST10 and STs belonging to clonal complex 10, phylogroup A has not been widely recognised as pathogenic in humans or non-human animals [27]. However E. coli belonging to the commensal phylogroups A and B1 are able to acquire virulence genes [17] and ARGs [28,29] and cause disease [17]. While ST216 is not one of the 20 most frequently reported E. coli STs on a global scale [27], MDR E. coli isolates of ST216 are recovered from humans with clinical infections [25,26,30,31]. Notably, Klebsiella pneumonia carabapenemase (KPC)-producing E. coli ST216 was linked with a large (125 isolates) and recalcitrant outbreak in Central Manchester University Hospital in the United Kingdom in 2015 [24]. During that episode, WGS showed that IncHI2 plasmids carrying bla KPC spread from ST216 to other E. coli STs as well as other Enterobacteriaceae. A notable feature of the UK outbreak was the recalcitrant nature of the contamination and the extraordinary measures taken to eliminate bla KPC + ST216 from several cardiac wards by replacing plumbing infrastructure, a measure that only partially alleviated subsequent episodes of infection. The bla KPC-2 gene in the UK outbreak was a component of a Tn4401a transposon [24], known for its enhanced KPC expression [32] and this study, while notable, represents the only one retrieved by a PubMed search using "E. coli ST216" as a search parameter at the time of writing (26/12/2020). However, bla KPC-2 was shown to be associated with a Tn4401g transposon located on an N plasmid in a clinical isolate of E. coli ST216 in Israel [25] and hospital-acquired E. coli ST216 carrying bla FOX-5 (serine β-lactamase with a substrate specificity for cephalosporins) linked to an IncC plasmid were recovered from a senior patient at an intensive care unit in the University of Maryland Medical Center in the United States [26]. In a possible episode of patient to family transmission, two bla CTX-M-15 -positive E. coli ST216 isolates were identified in a family member of a patient at Tel-Aviv Sourasky Medical Center in Israel [30] and colistin resistance was reported in an E. coli ST216 harbouring mcr-1 in a urine sample from a patient in Italy [31]. In the environment, E. coli ST216 carrying bla KPC-2 have been found on an R plasmid from a river ecosystem in Barcelona, Spain [33]. These episodes point to ST216 E. coli being proficient in the capture and dissemination of genes encoding β-lactamases and carbapenemases and suggest aquatic environments and both domestic and industrial wash basins and sinks may be an ideal niche. While speculative, the ability of ST216 to form recalcitrant biofilms in hospital wastewater pipes [24] is consistent with this view.
The silver gull typically breeds in large colonies on offshore islands and frequents urban environments including garbage dumps, shopping centres, railway stations, municipal parks and promenades along large river embankments. The Australian silver gull is known to harbour multiple drug-resistant E. coli [34,35] and Salmonella enterica spp. [36] highlighting hotspots where drug-resistant bacteria accumulate in urban environments [37,38]. In this study, we conducted whole genome sequencing of 22 MDR E. coli ST216 isolates from silver gulls (Chroicocephalus novaehollandiae) nesting on Big Island, 60 km south from Sydney in Australia, one of Australia's largest silver gull breeding sites. We investigated their phylogeny and serotype composition and determined the antibiotic and virulence gene cargo they carry. Long read sequencing of plasmid DNA enabled the determination of the complete sequence of several plasmids that carry genes encoding resistance to CIA. Phylogenetic analysis of 123 MDR E. coli ST216 isolates and their plasmid content enabled an assessment of the spread of these isolates and their associated mobile elements and ARGs both within Australia and globally, and shed light on the potential risks they present to human and animal health.

E. coli ST216 Collection from Gulls
Bacterial isolates were obtained from our previous study investigating silver gulls (Chroicocephalus novaehollandiae) as carriers of antibiotic resistant bacteria in New South Wales, Australia [34]. In that study, cloacal samples (n = 504) from gull chicks were collected at three locations (Five Islands, White Bay in Sydney and Montague Island) and 27 IMPproducing E. coli ST216 were obtained, all originating from silver gulls at Five Islands. Based on PFGE genomic and plasmid profiles, six representative isolates were selected for further analysis. An additional 16 E. coli ST216 isolates were obtained by cultivation of primary cloacal samples enriched overnight in buffered peptone on MacConkey agar with cefotaxim (2 mg/L) or ciprofloxacin (0.05 mg/L). Their sequence types were determined following WGS of all E. coli isolates (n = 448) obtained from silver gulls at the three sampling locations (our unpublished data). A total of 22 E. coli ST216 isolates, all originating from gulls in Five Islands, were obtained and subjected to WGS (Supplementary Table S1).

Whole Genome Sequencing
Genomic DNA of 22 gull ST216 was isolated using NucleoSpin ® Tissue kit (Macherey-Nagel GmbH & Co, Duren, Germany). DNA libraries were prepared using Nextera XT DNA library preparation kit and sequenced on a NovaSeq (Illumina, San Diego, CA, USA) platform. Assembly of obtained short reads was performed using Shovill v0.9.0 software [39].
Genomic DNA from E. coli CE1537 was selected for long read sequencing to obtain a complete E. coli ST216 reference sequence. Whole-genome DNA was extracted using NucleoSpin ® Tissue kit (Macherey-Nagel GmbH & Co, Duren, Germany) and library preparation was performed using microbial multiplexing based on the manufacturer's recommendation. The DNA was sheared using g-tubes (Covaris, Woburn, MA, USA) but size selection was not performed for library preparation. The Sequel 1 platform (Pacific Biosciences, Menlo Park, CA, USA) was used for long-read sequencing. Sequence assembly was carried with HGAP v4.0 software [40] and resulted in 11 contigs with an average 203-fold coverage. The incomplete sequence of an HI2-ST1 plasmid was identified on a separate contig, therefore, long-read sequencing using plasmid DNA of CE1537 was performed as described below.
Long-read sequencing of plasmid DNA (pDNA) extracted from isolates CE1537 and CE1681 was carried out to generate complete plasmid sequences. Plasmid DNA was extracted using a QIAGEN ® midi kit (Qiagen, Hilden, Germany) and library preparation using a microbial multiplexing protocol was performed as described above. The Sequel 1 platform (Pacific Bioscinces, Menlo Park, CA, USA) was used for pDNA sequencing followed by assembly of obtained reads with SMRT LNK v8.0 software (Pacific Biosciences, Menlo Park, CA, USA). This resulted in the assembly of six circular contigs from isolate CE1537 (pCE1537-A to pCE1537-F) with an average 317-fold coverage and six contigs from isolate CE1681 (pCE1681-A to pCE1681-F) with an average 326-fold coverage (Supplementary Table S2).

E. coli ST216 Metadata
All E. coli ST216 short reads were deposited on EnteroBase in the Escherichia/Shigella database and on Genbank (see Table 1 for barcode numbers and accession numbers, respectively) [41]. Short read sequences of ST216 isolate CE1537 were discarded due to contamination.
Long read sequence of isolate CE1537 was deposited on GenBank (accession number (AN For phylogenetic analysis genomes of E. coli ST216 (n = 99) from the EnteroBase Escherichia/Shigella database with one allele mismatch were selected and their assembly barcode used in the constructed phylogenetic tree (Figure 1). E. coli ST216 sequences with no sample collection date were excluded. In addition, one E. coli ST216 from Genbank with assembly and accession numbers GCA_002263825.1 and NNAL00000000 respectively, was used in ST216 phylogenetic analysis.

Phylogenetic Analysis
Single nucleotide polymorphisms (SNPs) (Figure 1 and Supplementary Figure S1) were identified and used for phylogenetic analysis of ST216 using CSI Phylogeny 1.4 [53]. A second SNP analysis (Supplementary Figure S2) including only gull E. coli ST216 isolates sequenced in this study was also performed. Phylogenetic trees were visualised with iTOL v4 [54]. Comparison and alignment of long-and short-read sequences with E. coli reference genomes from K12-MG1655 and E. coli ATCC8739 (GenBank Accession no.: U00096.3 and NZ_CP022959.1, respectively) were performed using progressiveMauve [55].

Phylogenetic Analysis of Complete and Closed Plasmids
To investigate the distribution of closed contigs (pCE1537-A, pCE1537-B, pCE1681-A, pCE1681-B, pCE1681-D) within the genome sequences of E. coli ST216 sourced from gulls, heat maps were generated and visualised using the chooklord pipeline (accessed on 4 March 2021 and available at https://github.com/maxlcummins/chooklord [56]) with closed plasmids as reference ( Figure 5A,B and Supplementary Figure S3A,E). A BLASTn search was performed and complete plasmid sequences with an identity threshold of ≥99% and a query coverage threshold of ≥91% were selected for further analysis ( Table 2). A different coverage threshold (86%) was set for investigating plasmid pCE1681-E because pEc1677, which showed the highest detected coverage (86%) with pCE1681-E, was selected from a BLASTn analysis. BRIG v0.95 software [57] and SnapGene ® v5.0.6 software (GSL Biotech LLC, Chicago, IL, USA) were used to perform comparisons of plasmid sequences. IS26-associated regions of plasmids pCE1537-A were subjected to further analysis using Easyfig v2.2.3 [58] for comparison and visualisation.

Transferability of HI2 Plasmids
The MDR profile of sequenced E. coli ST216 isolates (CE1537 and CE1681) prevented the performance of a direct conjugation assay with a suitable recipient (employing a unique selection marker from the donor cells). We performed a multiphase test to circumvent this obstacle and check the conjugative properties of the resolved HI2 plasmids. HI2 plasmids were selected for transferability testing due to their carriage of bla IMP-4 which was a primary focus for this study. Plasmid DNA was extracted from donor cells using the QIAGEN ® midi kit (Qiagen, Hilden, Germany) and then transferred via electroporation to plasmid-free E. coli Top10 cells. Transformants positive for HI2 plasmid were selected on LB agar supplemented with cefotaxime (2 mg/L) and incubated overnight at 37 • C. The presence of bla IMP and the HI2 plasmid in transformants was confirmed by PCR [34] and replicon typing [59], respectively. Total cellular DNA from transformants was digested with S1 nuclease and subjected to pulsed-field gel electrophoresis to confirm plasmid carriage. Conjugative transfer of HI2 positive transformants to a plasmid-free, rifampicin and sodium-azide resistant E. coli MT102 recipient strain [60] was performed using filtermating method with incubation for 4 h at 28, 30 and 37 • C followed by selection of transconjugants on LB agar plates supplied with cefotaxime (2 mg/L), rifampicin (25 mg/L) and sodium azide (100 mg/L) and incubation overnight at 37 • C. The presence of bla IMP and the HI2 plasmid in four transconjugants was confirmed by PCR [34] and replicon typing [59], respectively.  Table S1). All 22 E. coli ST216 isolates that were sequenced are phylogroup A. In silico O:H typing identified two serotypes: O45:H4 (n = 1) and O154:H4 (36%, n = 8). Thirteen isolates were O-non-typable with flagella type H4 (9%, n = 2), O-non-typable with H-non-typable (36%, n = 8) and O45 with H-non-typable (14%, n = 3). fimH profiling revealed three fimH types including fimH23 (36%, n = 8), fimH69 (45.5%, n = 10), and fimH1248 (18%, n = 4) (Supplementary Table S1).

Phylogenetic Analysis of E. coli ST216
A SNP-based phylogenetic tree was constructed with genomes of 122 ST216 including the 22 E. coli ST216 isolates from silver gulls and 20 isolates with STs that carried a variant in a single multilocus sequence allele from ST216 ( Figure 1). All isolates, including those from silver gulls, segregated into two main clades with significant diversity between them (3000-9000 SNP variants). Silver gull isolates also divided into two main SNP cluster groups (SCG) and six phylogenetic subgroups interspersed throughout the SNP-tree ( Figure 1). Isolates from silver gulls (n = 22) were phylogenetically diverse showing a minimum of six and maximum of 8505 SNPs differences. The first SNP cluster (SCG1) which included eight E. coli ST216 isolates that differed only by 11-52 SNPs was located on a distinct branch (clade I in Figure 1) that included a handful of clinical isolates from the USA and a single clinical isolate from Norway. The gull cluster in clade I was most closely aligned (maximum of 283 SNPs difference between SCG1 and ESC_QA4689AA_AS) to the clinical isolate from Norway (ESC_QA4689AA_AS) (Figure 1). ST216 gull isolates in the second cluster (SCG2, n = 14) were more distinct (6-2347 SNPs difference) and included isolates from the clinic, environment, and domestic animals (clade II in Figure 1). Gull isolate 1720H in clade II was related (163-259 SNPs difference) to domestic animal isolates (ESC_TA2295AA_AS and ESC_GA6917_AS) from Australia and the USA, clinical isolates (ESC_ZA4597AA-AS and ESC_AA8402AA_AS) from Kenya and the USA and to an environmental isolate (ESC_GB5355AA_AS) of unknown origin.  E. coli ST216 isolates from silver gulls had a higher content of plasmids and ARGs (mean of 4.5 plasmids and 15 ARGs in silver gulls) compared to international ST216 isolates (mean of 1.7 plasmids and 1 ARG; Supplementary Figure S1), an observation that is consistent with selection on media containing antibiotics. In contrast to SCG1 and SCG2, HI2 plasmids were absent from almost all of the 100 international isolates. FIB(K) plasmids were the most common plasmids in international isolates (31%, n = 31) and were dominant in SCG1 and SCG2 (91%, n = 20). R plasmids were present in international isolates (11%, n = 11) and SCG2 (50%, n = 7) but absent from SCG1. Similarly, X5 plasmids were dominant in SCG2 (71%, n = 10) and not detected in SCG1. Resistance genes bla IMP-4 , bla SHV-12 and dfrA19 were only identified in SCG1 and SCG2 and were not detected in any international isolate (Supplementary Figure S1) indicating that HI2 plasmids harbour genes encoding these CIA. Other ARGs including qnrS1, aac(6')-Ib-cr and dfrA14 were identified both in SCG2 (71%, 93%, 64%, n = 10, n = 13, n = 9, respectively) and in international isolates (6%, 10% and 4%; n = 6, n = 10, n = 4, respectively) (Supplementary Figure S1). VAGs in SCG1, SCG2 and international isolates were common to E. coli species (Supplementary Figure S1). However, several VAGS were only present in SCG2 and international isolates including putative type III secreted effector espX1, fimbrial associated genes fimC, fimD, fimE and fimI and general secretion pathway genes gspC, gspD, gspE, gspF, gspI and gspK (Supplementary Figure S1).

Virulence Associated Genes (VAGs) of Gull Isolates
Carriage of virulence genes by the 22 ST216 isolates was unremarkable. Between 18 and 33 VAGs previously described among E. coli were identified in ST216 sequences. VAGs included enterobactins and elements of type II/III secretory systems, type I fimbriae regulators, ferric enterobactin transport system and general secretion pathway proteins (Supplementary Table S1). In isolate CE1537, a flagellin gene fliC was located on an FIA(HI1) plasmid (pCE1537-B) with a fliC repressor gene residing between two IS elements (Supplementary Figure S4-B and Supplementary Table S2). In isolate CE1681, a colicin E7 operon that includes the colicin E7 protein, colicin E7 immunity protein and a colicin E7 lysis protein is located on a Col156 plasmid (pCE1681-C) (Figure 2). In the same isolate (CE1681), a haemolysis expression-modulating protein (Hha) is located on X5 plasmid pCE1681-E ( Figure S4-C).
We identified a total of 30 ARGs in the sequenced population of gull E. coli ST216 isolates and the strains carry between 7 and 20 ARGs each (Supplementary Table S1 Table S1). The quinolone resistance genes qnrA1 (36%, n = 8), qnrS1 (45.5%, n = 10) and aac(6')-Ib-cr (45.5%, n = 10) were less frequent. All ST216 isolates carried a class 1 integrase intI1 and IS26 (Supplementary Table S1). The resistance phenotype of most ST216 isolates correlated with their genotype except for isolate CE1681 and 1720H which exhibited phenotypic resistance to trimethoprim and chloramphenicol, respectively, with the absence of corresponding ARGs (Supplementary Table S1).
regulators, ferric enterobactin transport system and general secretion pathway proteins (Supplementary Table S1). In isolate CE1537, a flagellin gene fliC was located on an FIA(HI1) plasmid (pCE1537-B) with a fliC repressor gene residing between two IS elements (Supplementary Figure S4- B and Supplementary Table S2). In isolate CE1681, a colicin E7 operon that includes the colicin E7 protein, colicin E7 immunity protein and a colicin E7 lysis protein is located on a Col156 plasmid (pCE1681-C) (Figure 2). In the same isolate (CE1681), a haemolysis expression-modulating protein (Hha) is located on X5 plasmid pCE1681-E ( Figure S4-C).

Antibiotic Resistance Phenotypes and Genes of Gull Isolates
The collection of 22 sequenced isolates present variable MDR phenotypes with resistance ranging from 4 to 15 antibiotics (Supplementary Table S1). Most isolates (95%, 21/22) are carbapenemase producers (Supplementary Table S1). All isolates are resistant to streptomycin and sulphonamides (100%, 22/22) with 95% (21/22) of the isolates resistant to ampicillin, trimethoprim/sulfamethoxazole, cefalotin, ceftazidime and amoxicillin/clavulanic acid. Resistance to chloramphenicol and gentamicin were both observed in Most E. coli ST216 (95.5%) have at least one plasmid-mediated quinolone resistance gene (21/22) with only nine isolates expressing phenotypic resistance to ciprofloxacin and nalidixic acid. Of the 22 ST216 isolates, six are resistant only to nalidixic acid, one isolate is resistant only to ciprofloxacin while six isolates are susceptible to ciprofloxacin and nalidixic acid (Supplementary Table S1). Resistance to ertapenem, meropenem and imipenem was detected in 15/21 bla IMP-4 -positive E. coli ST216 isolates (Supplementary Table S1). Known chromosomal gyrA mutation S83L that confers resistance to nalidixic acid and ciprofloxacin was identified in isolates 1556m1, 1548R1 and 1605m2 (Supplementary Table S1) [61]. Moreover, several chromosomal mutations with unknown effect were detected in 16S_rrsB, 16S_rrsC, 16S_rrsH, 23S, pmrB, parC and gyrA genes for isolates CE1537 and CE1586. Isolate CE1586 had an additional chromosomal mutation in pmrA gene.
In all bla IMP-4 -positive E. coli ST216 isolates, bla IMP-4 was found on HI2-ST1 plasmids as a component of In809 that was flanked by IS26 ( Figure 3 Figure 5B). and CE1586. Isolate CE1586 had an additional chromosomal mutation in pmrA gene.
In all blaIMP-4-positive E. coli ST216 isolates, blaIMP-4 was found on HI2-ST1 plasmids as a component of In809 that was flanked by IS26 ( Figure 3
pCE1681-B shared high sequence coverage and identity with five plasmids ( Table 2). Four of these pCFSAN061762, pFZ11, pCFSAN061763 and pCFSAN061768 were from E. coli and CP020344.1 was from Shigella flexneri. These five plasmids carried qnrS1, bla TEM-1b and dfrA14 ( Figure 6). Plasmids CP020344.1 and pFZ11 had a clinical source in China while plasmids pCFSAN061762, pCFSAN061763 and pCFSAN061768 originated from raw milk in Egypt ( Table 2).
Analysis of plasmid pCE1681-C showed 100% coverage and ≥ 99% identity with two E. coli plasmids (Table 2), all carrying a colicin E7 operon (Figure 2). Plasmids ColE7-K317 and pECAZ146_5 originated from an unidentified source in Pakistan and from a clinical source in Italy, respectively.
Plasmid pCE1681-E showed 86% coverage and 100% identity to a bla IMP-4 -positive X5 plasmid pEc1677 (Supplementary Figure S4-C) originating from an E. coli isolate from a silver gull in Sydney [69]. The two plasmids had a similar backbone structure but differed in their variable region with the absence of ARGs in pCE1681-E while pEc1677 harboured bla IMP-4 and other ARGs. In plasmid pEc1677 bla IMP-4 was part of a class 1 integron that included antibiotic resistance genes aac(6 )-Ib3, catB3 and sul1 and quaternary ammonium resistance genes qacG and qacE. The organisation of these genes was like that observed in pCE1537-A (HI2-ST1) (Figure 4). Both plasmids, pEc1677 and pCE537-A shared an IS26 element upstream of bla IMP-4 while they had IS elements IS4321 and IS91-like bordering sul1.

Discussion
Wild and urban-adapted birds carry, cycle, and transmit mobile elements carrying ARGs and VAGS between humans, animals and the environment [3,35,36,70,71]. Carriage of MDR bacteria by wild and urban bird populations remind us of the need to remove anthropogenic pollutants, particularly antibiotic resistant bacteria, antibiotic residues, disinfectants and metals from the environment.
E. coli with variable MDR profiles including carbapenem resistance encoded by bla IMP-4 carried on HI2-ST1 and HI2-ST3 plasmids have been recovered from cloacal samples of silver gulls in Australia [34]. Based on this report, we utilised WGS to detect and investigate E. coli that carry genes encoding resistance to CIA. After ST457 [35], E. coli ST216 represented the second most common E. coli ST accounting for 8.4% of all E. coli isolates (n = 262) obtained from cloacal samples at Big Island and 5% of the 448 E. coli recovered from the three coastal sampling sites (Five Islands, Sydney and Montague Island) in New South Wales, Australia. SNP-based phylogenetic analysis of international ST216 isolates ( Figure  1) divided ST216 into two clades. Clade I is small and includes isolates of clinical and domestic animal origin in addition to isolates sequenced here from silver gulls (SCG1). ST216 genomes in SCG1 show a clonal-like distribution and many strains carry HI2-ST1 plasmids that are highly similar to HI2-ST1 plasmids observed in Australian clinical and companion animal sources ( Figure 4). Documented variability in the carriage of genetic cargo residing in complex resistance regions of HI2-ST1 plasmids in SCG1 ( Figure 5A) suggest that multiple plasmid acquisition events may occur or that the resistance regions respond rapidly to selection pressures where ST216 persists.
The phylogeny of ST216 genomes in SCG2 (Figure 1) indicates that E. coli ST216 is globally distributed and occupies a diverse host range, including isolates from environmental sources as well as from humans, domestic animals and wildlife (Supplementary Figure  S1). The distribution of SCG2 into several subphylogenetic groups ( Figure 1) coupled with the diverse plasmid content compared with SCG1 ( Figure 1 and Table 1) suggest multiple introduction events of ST216 in silver gulls.
The risk of dissemination of E. coli ST216 carrying bla IMP-4 and genes encoding resistance to other CIA to humans and domestic animals is a cause for concern. Silver gulls in the Sydney-Wollongong region share common spaces that include facilities with high human contact (rail and bus stations and municipal parks) and ST216 has been isolated from silver gulls in different geographic regions of Australia [72]. Plasmids carrying multidrug resistance genes with different incompatability markers have been recovered from multiple E. coli and Salmonella enterica lineages from gulls in Australia [34][35][36]72]. E. coli ST216 show a clear propensity to acquire plasmids that carry diverse resistance gene cargo and are linked to aquatic environments where they have been responsible for recalcitrant, carbapenem-resistant infections in hospital drainage waste systems [24]. ST216 belongs to commensal phylogroup A. Apart from the outbreak cluster in the UK, most reports of ST216 in humans are sporadic cases [25,30] in hospital. There is a clear association of E. coli ST216 with wildlife and reports to date all describe carriage of genes encoding resistance to CIA [33,72] [This study]. Given its comparatively low reporting in humans and its repeated links with wildlife and aquatic environments it is tempting to speculate that water may be a natural reservoir for E. coli ST216 and aquatic wildlife hosts, particularly birds, for its distribution. The pressing question is what role wildlife will play, if any, in the continued evolution of E. coli ST216 given the propensity for it to acquire self-replicating mobile genetic elements.
The ARG content in our ST216 population (Supplementary Table S1) is largely due to the carriage of HI2 (ST1 and ST3) and FIB(K) plasmids (Supplementary Table S2). Isolation of Enterobacterales from gulls using antibiotic selection likely created a bias in what lineages were observed and explains the high carriage of plasmids and ARGs we found in the silver gulls. Comparisons to strains not isolated under an antibiotic selective pressure will be necessary to reveal the distribution of the now identified AMR genes and their transfer mechanisms. Insertion sequences IS26 and IS15DI play a major role in capturing and mobilising antibiotic resistance genes [62,73,74] and are often found in close association with class 1 integrons [64,[75][76][77]. The presence of these insertion elements serves as a hotspot for capture of ARGs flanked by IS26 [73]. IS26 also plays an important role in altering the structure of class 1 integrons by truncating the 3'-CS and the 3 end of intI1 [77][78][79]. IS26 has played a role in shaping the resistance regions in HI2 ST1 and ST3 and F plasmids that are carried by E. coli ST216 from silver gulls described in this study. IS26 elements can facilitate hybrid plasmid formation [16,80] and may have a role in promoting plasmid stability [81]. Comparative analysis of HI2-ST1 plasmid pCE1537-A provides evidence that plasmids circulate closely within different bacterial species and hosts (humans, companion animals and wildlife) in Australia. It is also concerning that IS26 has been implicated in the mobilisation of virulence genes [82].
F plasmids carrying ARGs and VAGs are widespread and are commonly associated with Enterobacterales from clinical sources and food-animals [83,84]. Isolates from silver gulls carry various VAGs, metal transport systems and the colicin E7 operon on several plasmids. The acquisition of these genetic elements can enhance the survival, colonisation and persistence characteristics of bacteria that inhabit different niches [65,[85][86][87][88]. Moreover, the presence of both metal resistance genes and ARGs leads to coselection and promotion of ARGs in bacterial populations in the absence of antibiotics [89]. These data suggest that the evolution of ST216 and their mobilome are influenced by anthropogenic pollution. Feeding and flight behaviours [90] are likely to have a profound influence on the silver gull resistome.
At the time of writing, a total of 137 ST216 isolates were deposited in the Enterobase database. Of these 121 (80%) were deposited after 2010 with a global distribution across clinical, environmental, and animal hosts. Only a few of the ST216 isolates were collected from clinical samples associated with diarrhoea (ESC_FB7867AA and ESC_FB7867AA) in China and septicaemia (ESC_AA2218AA) in Germany. The increased frequency of reports of E. coli ST216 is concerning, particularly in light of their ability to acquire multiple diverse plasmids carrying ARGs, VAGS, biocins and metal resistance genes and their ability to colonise the gastrointestinal tracts of wild and urban birds. These characteristics are known to be important in the evolution of successful MDR bacterial clades [91]. However, unlike many dominant bacterial clades that show high genetic conservation within a geographical region [14], ST216 isolates appear to be phylogenetically interspersed and distanced from each other even within a single geographical location, suggesting that flight behaviour is an important attribute in understanding how wildlife, particularly birds, acquire drug-resistant flora. Based on these observations E. coli ST216 warrants further monitoring in bird populations in Australia and internationally. Furthermore, studies of the enterobacterial populations in urban-adapted and wild bird species using nonselective approaches is needed to improve understanding of lineages that colonise and persist in the avian gut. We also advise adopting a one health approach and investigating ST216 populations in silver gulls, humans and their surrounding environment to understand the transmission pathways and other features that influence pathogen evolution.

Conclusions
ST216 is a broad host range phylogroup A E. coli. Here we report the carriage of MDR E. coli ST216 by silver gulls on Five Islands near Wollongong, Australia harbouring diverse plasmids that carry multiple ARGs, VAGs and metal resistance genes. Most of the ST216 isolates were carbapenemase producers and carried bla IMP-4 on HI2-ST1 and HI2-ST3 plasmids. ARGs within HI2 plasmids were assembled in complex resistance regions together with metal resistance genes and multiple copies of IS elements including IS26 and IS26 derivative IS15DI. We report the spread of highly related IncHI2-ST1 plasmids between various bacterial hosts from different sources that include humans, domestic animals and wildlife in Australia. The recent increase of global ST216 reports isolated from different sources, expressing ARGs for critically important antibiotics and causing long lasting clinical outbreaks are concerning. To understand the transmission cycle of ARGs and MDR bacteria and its associated human risk, it is essential to adopt a one health approach that take into consideration all aspects of the ecological system with a focus on intermediate hosts (as gulls) that can act as vectors and sentinels for the spread of ARGs. Another consideration is the importance of interactions between different bacterial species mediated by mobile genetic elements such as HI2 plasmids and IS26 and its effect on the evolution and pathogenicity of these organisms.

Supplementary Materials:
The following are available online at https://www.mdpi.com/2076-2 607/9/3/567/s1, Figure S1: SNP-tree showing clonal relationship and genetic characteristics of Escherichia coli ST216 isolates from silver gulls at five islands and international related isolates. Figure  S2: SNP-tree showing clonal relationship and genetic characteristics of Escherichia coli ST216 isolates from silver gulls at Five Islands. Figure S3: Heat maps showing the distribution of reference plasmids within sequenced short reads of Escherichia coli ST216 isolates from silver gulls at Five Islands. Figure  S4-A: Schematic diagram of plasmid pCE1681-A (IncHI2-ST3). Figure S4-B: Schematic diagram of pCE1537-B (IncFIA) plasmid. Figure S4-C: BRIG comparison of IncX5 plasmid pCE681-E with similar plasmid sequence retrieved from GenBank. Figure S4-D: Schematic diagram of pCE1681-D (IncR) plasmid. Table S1: Characteristics of sequenced Escherichia coli ST216 isolates from silver gulls at Five Islands, Table S2: Characteristics of sequenced closed and complete plasmids in Escherichia coli ST216 isolates CE1537 and CE1681 from silver gulls at Five Islands.