Clinically relevant antibiotic resistance in Escherichia coli from black kites in southwestern Siberia: a genetic and phenotypic investigation

ABSTRACT Wild birds including raptors can act as vectors of clinically relevant bacteria with antibiotic resistance. The aim of this study was to investigate the occurrence of antibiotic-resistant Escherichia coli in black kites (Milvus migrans) inhabiting localities in proximity to human-influenced environments in southwestern Siberia and investigate their virulence and plasmid contents. A total of 51 E. coli isolates mostly with multidrug resistance (MDR) profiles were obtained from cloacal swabs of 35 (64%, n = 55) kites. Genomic analyses of 36 whole genome sequenced E. coli isolates showed: (i) high prevalence and diversity of their antibiotic resistance genes (ARGs) and common association with ESBL/AmpC production (27/36, 75%), (ii) carriage of mcr-1 for colistin resistance on IncI2 plasmids in kites residing in proximity of two large cities, (iii) frequent association with class one integrase (IntI1, 22/36, 61%), and (iv) presence of sequence types (STs) linked to avian-pathogenic (APEC) and extra-intestinal pathogenic E. coli (ExPEC). Notably, numerous isolates had significant virulence content. One E. coli with APEC-associated ST354 carried qnrE1 encoding fluoroquinolone resistance on IncHI2-ST3 plasmid, the first detection of such a gene in E. coli from wildlife. Our results implicate black kites in southwestern Siberia as reservoirs for antibiotic-resistant E. coli. It also highlights the existing link between proximity of wildlife to human activities and their carriage of MDR bacteria including pathogenic STs with significant and clinically relevant antibiotic resistance determinants. IMPORTANCE Migratory birds have the potential to acquire and disperse clinically relevant antibiotic-resistant bacteria (ARB) and their associated antibiotic resistance genes (ARGs) through vast geographical regions. The opportunistic feeding behavior associated with some raptors including black kites and the growing anthropogenic influence on their natural habitats increase the transmission risk of multidrug resistance (MDR) and pathogenic bacteria from human and agricultural sources into the environment and wildlife. Thus, monitoring studies investigating antibiotic resistance in raptors may provide essential data that facilitate understanding the fate and evolution of ARB and ARGs in the environment and possible health risks for humans and animals associated with the acquisition of these resistance determinants by wildlife.


Sampling of black kites
Cloacal samples (n = 55) from free-living nestlings of black kites were collected in their natural habitat through two consecutive years in 2018 (n = 16) and 2019 (n = 39) using swabs with culture medium (Amies transport medium with activated charcoal, Czech Republic), then transported to the laboratory and stored at 4°C till the initiation of the enrichment protocol. Sampling was performed in three localities in Russia's southwest ern Siberia (Fig. 1). Sampling localities included urban and agricultural areas around the cities of Biysk (n = 16, July 2018, Altai Krai) and Kyzyl (n = 33, June 2019, Republic of Tuva) and a small rural community in Kokorya (n = 6, June to August 2019, Altai Republic). (Refer to Table S1 for details on the sampling geolocation and collection year for each sample.)

Selection of E. coli isolates
E. coli isolates were selected by cultivation of primary cloacal samples enriched overnight in peptone buffer (37°C with shacking at 140 RPM) on MacConkey agar with cefotaxime (2 mg/L), ciprofloxacin (0.05 mg/L), or meropenem (0.125 mg/L). In addition, SuperPolymyxin medium (23) was utilized for the selection of isolates with resistance to colistin. One presumptive E. coli isolates from each plate was taken and identified to species level by matrix-assisted laser desorption ionization time-of-flight mass spectrom etry (24). Isolates identified as E. coli (n = 51) were subjected to further testing.

Whole genome sequencing
Whole genome DNA was extracted from 36 E. coli isolates using NucleosSpin Micro bial DNA kit (Macherey-Nagel, Germany). Preparation of DNA libraries was performed using Nextera XT DNA library preparation kit followed by sequencing on the NovaSeq platform (Illumina, San Diego, California, USA). Trimming of short reads for quality (Q ≤ 20) and adaptor residues was carried out using Trimmomatic v0.36 (35). Short reads were assembled using SPAdes v3.12.0 (36). Complete and closed plasmids of one E. coli isolates (DR162-CEF) belonging to APEC-linked ST354 were obtained using long-read sequencing. Whole genome DNA was extracted using a QIAGEN midi kit (Qiagen, Hilden, Germany), and library preparation was performed using microbial multiplexing based on the manufacturers' recommendation. DNA was sheared using g-tubes (Covaris, Massachusetts, USA), but size selection was not performed for library preparation. Sequel 1 platform (Pacific Biosciences, California, USA) was used for sequencing followed by assembly using a Microbial Assembly pipeline in SMRT LNK v9.0 software (Pacific Biosciences, California, USA) with a minimum seed coverage of 30X. Quality control of obtained short-and long-read sequences was performed using FASTQC (https:// www.bioinformatics.babraham.ac.uk/projects/fastqc/).

Phylogenetic analysis
Phylogenetic analysis based on single nucleotide polymorphisms (SNPs) was performed on all sequenced E. coli isolates from kites where E. coli K12-MG1655 was used as a reference genome. CSI Phylogeny 1.4 (available at https://cge.cbs.dtu.dk/services/) (40) was employed for SNPs analysis, and its results were visualized using iTOL (v.6.4) (41).

E. coli metadata
Short read sequences of E. coli isolates from black kites were deposited on GenBank (BioProject ID PRJNA702622) and on EnteroBase in the Escherichia/Shigella database (Table S1 for accession and barcode numbers, respectively).
Phylogenetically, isolates from black kites had high heterogeneity (0-46714 SNP variants) with no observed clustering (Fig. 2). Two E. coli ST162 isolates with MDR and ESBL phenotypes from kites in Biysk region (DR164-CEF and DR167-CEF) were closely related (10 SNPs difference) and shared identical fimH type (32), serotype (O88:H10), and ARGs content. Similarly, two isolates originating from kites in proximity to Kyzyl city had two clonal mcr-1-positive isolates (DR356a-COL, and DR358b-COL with 0 SNPs difference) belonging to ST93 with fimH53 and serotype O21:H16. Both of the latter isolates were ESBL producers and shared identical ARGs and VAGs content and a highly similar plasmid profile. Two isolates from kites in proximity to Biysk and Kyzyl cities (DR161-CEF ST2197 and DR370-CEF ST12666, respectively) were closely aligned (103 SNPs difference) and had identical fimH type (23) and serotype (O128:H26) as well as virulence profile. However, their ARGs and plasmid content were distinct with the exception of sharing tet(B) and IncI1-I plasmid.

High diversity of plasmid replicons and spread of intI1 and IS26 in E. coli from black kites
In total, 18 different plasmid replicons were identified with F and various Col plasmids present in most isolates (28/36, 77% and 27/36, 75%, respectively). Sequenced isolates carried between zero and nine plasmids with an average of three plasmids (Table S1).

Resistance to colistin associated with mcr-1-IncI2 plasmids
The mcr-1 gene in all six isolates from kites in Biysk and Kyzyl was carried on IncI2 plasmids with highly homologous structures (Fig. 3). Isolates harboring mcr-1 were MDR, belonged to diverse STs (48, 93, 1,642, 2,197, and 2,280), phylogroups A (4/6) and B1 (2/6), and three of them were ESBL producers. Comparative analysis of IncI2 plasmids from black kites (Fig. 3) showed high identity and coverage with IncI2 plasmids originating from clinical samples in China (hosted by Salmonella enterica), Argentina, and Russia (hosted by E. coli) and a sewage sample in China (harbored in E. coli).

DISCUSSION
Clinical and agricultural use and misuse of antibiotics are the main drivers for the emergence of antibiotic resistance that can contaminate the ecosystem including the Research Article mSphere environment and wildlife (45,46). The existence of E. coli in wildlife with multidrug resistance (MDR) to clinically important antibiotics represents a potential hazard to human and animal health (47,48), where wildlife can act as a reservoir and spreader of ARB in the human-environment-animal interphase (48,49). Black kites in Russian western Siberia reside in localities with access to human food refuse (i.e., landfills) (50) and habitat intersection with agricultural elements (i.e., fertilized agricultural lands (51) and food animals), all of which are known to be common sources of ARB (52)(53)(54). These factors support the hypothesis of a potential anthropogenic spillover of ARB and antibiotic-resistant determinants to black kites in the Altai Krai region which warranted this study. Through this investigation, we identified prominent occurrences of E. coli with antibiotic resistance including MDR and ESBL profiles in black kites residing in south western Siberia in proximity to human-influenced environments. The high occurrence of antibiotic-resistant E. coli in samples near major urban areas such as Biysk and Kyzyl supports the finding of previous studies (55)(56)(57) associating wildlife in environ ments influenced by anthropogenic activity with high carriage of ARB. The occurrence of antibiotic-resistant E. coli in Kokorya might implicate human populations in rural communities as carriers and possible transmitters of ARB that was observed in studies from rural regions in India (58) and Peru (59). Another possible source for the acquisition and carriage of antibiotic-resistant E. coli observed in black kites could be their wintering grounds that are located in the densely populated areas in India and Pakistan where they feed mainly on municipal waste from landfills (13). Considering that sampled kites were in their nestling stage, it is possible that their parents carried the antibiotic-resist ant isolates from their wintering habitat in the Indian subcontinent and transmitted observed antibiotic-resistant E. coli via feeding to their offspring. Although, it is well known that bird nestlings can get inoculated by different bacteria from their environ ment particularly food provided by their parents which is mixed with the parents' saliva (60), reports investigating the carriage duration of bacteria in avian wildlife are still scarce (61,62).
Our Black kites in the Biysk region utilize available human food refuse from nearby landfills as their main food source (50). We predict a similar behavior for kites in Kyzyl where a landfill is located in the proximity to the city (Fig. 1). Based on these feeding behaviors, the observed ARB and their resistance profiles in these birds might be associated with anthropogenic spillover that was captured by black kites. This postula tion is supported by results from a phylogenetic analysis that identified two clonal mcr-1-positive E. coli isolates (DR356a-COL and DR358b-COL) originating from two black kites in Kyzyl with zero SNPs difference indicating a recent common source of the two isolates. Notably, these isolates belonged to ST93 which is frequently reported in APEC and ExPEC (66,67). Similarly, several isolates from kites in Biysk and Kyzyl belonged to STs associated with APEC (ST624 (68)), ExPEC (ST38 (69)), and overlapping APEC-ExPEC (ST10, ST23, ST354, and ST1011 (66,69,70)). Besides one isolate of ST162 that is commonly detected in livestock (71,72), companion animals (73,74), and wildlife (75,76), black kites in Kokrya had no isolates with known STs linked to pathogenic E. coli. The association of avian wildlife foraging mainly on human food refuse with a significant prevalence of E. coli encoding resistance to clinically important antibiotics was well documented in studies involving wild bird species such as gulls (Laridae) (57,77,78), white storks (Ciconia ciconia) (79), and bald eagles (Haliaeetus leucocephalus) (78). These studies reported high diversity of ARGs encoding resistance to multiple classes of antibiotics including ESBL, which was also observed in black kites from Biysk and Kyzyl. The spread of ESBL/AmpC production, chromosomal quinolone resistance and considerable virulence content in sequenced isolates is concerning as it primes them as potentially pathogenic lineages.
Class one integrase gene intI1 was suggested as a proxy for anthropogenic pollution and as a marker for antibiotic resistance (80). Based on that, the high occurrence of intI1 in the sequenced isolates from kites in southwestern Siberia (22/36, 61.1%) may indicate a prominent anthropogenic influence on these populations. IS26 has emerged as a critical element in the mobilization of ARGs in Gram-negative bacteria where it is usually found in complex resistance regions harboring resistance genes to multiple classes of antibiotics (81,82). The presence of IS26 in isolates originating from kites in Biysk and Kyzyl might be one of the factors contributing to the high diversity of their ARGs.
Mobile colistin resistance gene mcr-1 is increasingly observed in avian wildlife including aquatic and migratory species (3,83). Suggested ability of wild birds for prolonged carriage and transmission of ARB (61) is worrisome especially in the case of migratory birds as black kites harboring isolates with resistance to last-line antibiotics. The spread of identical mcr-1-positive IncI2 plasmids in isolates from black kites of different STs and locations might indicate mobilization of the plasmid through horizon tal gene transfer or a gradual acquisition of mcr-1-positive isolates by black kites on multiple occasions from unknown sources. Identification of highly similar IncI2 plasmids originating from black kites and clinical samples in Russia, China, and Argentina and a sewage sample from China (Fig. 3) supports the reports of global spread of mcr-1-posi tive IncI2 plasmids (84).
The association of DR162-CEF with ExPEC-APEC and its carriage of MDR and virulent plasmids is concerning as E. coli ST354 was linked to serious clinical infections including urinary tract, prostate, and bloodstream infections (69,84,85).
To our best knowledge, we present the first report of qnrE1 in E. coli from a wildlife source. Recently, quinolone resistance gene qnrE1 was identified in a Salmonella enterica isolate from retail meat in China (86) and in Enterobacter asburiae from a clinical sample in Thailand (87). Most reports on qnrE1 originate from South America where it was found in clinical (commonly on IncM plasmid) and domestic animal samples, hosted by Klebsiella pneumoniae, Salmonella Typhimurium, and E. coli (88)(89)(90)(91)(92)(93). A recent study indicated a diseased parrot Amazona aestiva in Brazil as a carrier of qnrE1-positive IncM1 plasmid (94). However, the parrot was a companion animal [confirmed through written confirmation with the corresponding author (94)] which implies a possible human-com panion animal transmission. The qnrE1 gene was also identified in a MDR K. pneumoniae isolate from a native Amazonian fish in Brazil where it was harbored in a hybrid IncFIB/ IncHI1B plasmid (95). qnrE1 originated from the chromosome of Enterobacter spp. with a suggested role for ISEcp1 in its mobilization to K. pneumoniae (26). The association of qnrE1 with ISEcp1 and its insertion in an MDR IncHI2-ST3 plasmid (pDR162-CEF-A) in E. coli ST354 may indicate an ongoing interspecies dissemination of qnrE1, facilitated by ISEcp1 and conjugative plasmids. In addition, the association of MDR IncHI2-ST3 plasmids with metal resistance and biocins might facilitate their persistence and spread in the absence of antibiotic selection pressure (96).
There are several limitations in our study including the uneven distribution of samples between black kites in the three sampled locations and the temporal difference in sampling of kites from Biysk (sampled in 2018) and those from Kyzyl and Kokorya (both sampled in 2019). The use of selective cultivation method results in targeting resistant bacterial isolates, thus hindering a better understanding of the avian micro flora including common and persistent bacterial lineages. In addition, the unavailabil ity of national clinical and environmental monitoring data on antibiotic resistance in southwestern Siberia prevented a comparative analysis that would potentially support determining possible sources and transmission routes of identified ARB and their associated ARGs.
Our results imply a human factor behind some of the observed diversity in resistance and plasmids profiles, it also demonstrates that even in environments with minimal human activity such as Altai rural communities, E. coli with considerable resistance and virulence determinants can be detected. The risk of carriage and transmission of these isolates on domestic animals and humans is not fully understood. We advise further investigation that adopts a One Health approach in investigating ARB in black kites and other avian species to determine the origin, spread, transmission routes, and persistence of these isolates in the human-domestic animal-environment-wildlife interphase. This in turn can help in determining their reservoirs and provide an informative concession of their health risk.

DATA AVAILABILITY STATEMENT
The sequencing data presented in this study are openly deposited in GenBank within Bioproject PRJNA702622.

ETHICS APPROVAL
All procedures and techniques used for sampling and handling of black kites in this study were conducted by trained personal in accordance to Russian federal law No. 52-FZ "On wildlife". Sampling of black kites was conducted in a manner to ensure the conservation of the bird species and their habitats and minimize the disturbance to black kite population. Precautions were implemented to minimize any potential harm or distress to the birds during handling. Black kites were released back into their natural habitat following sampling.

ADDITIONAL FILES
The following material is available online.