ESBL-producing Escherichia coli in Swedish gulls—A case of environmental pollution from humans?

ESBL-producing bacteria are present in wildlife and the environment might serve as a resistance reservoir. Wild gulls have been described as frequent carriers of ESBL-producing E. coli strains with genotypic characteristics similar to strains found in humans. Therefore, potential dissemination of antibiotic resistance genes and bacteria between the human population and wildlife need to be further investigated. Occurrence and characterization of ESBL-producing E. coli in Swedish wild gulls were assessed and compared to isolates from humans, livestock and surface water collected in the same country and similar time-period. Occurrence of ESBL-producing E. coli in Swedish gulls is about three times higher in gulls compared to Swedish community carriers (17% versus 5%) and the genetic characteristics of the ESBL-producing E. coli population in Swedish wild gulls and Swedish human are similar. ESBL-plasmids IncF- and IncI1-type carrying ESBL-genes blaCTX-M-15 or blaCTX-M-14 were most common in isolates from both gulls and humans, but there was limited evidence of clonal transmission. Isolates from Swedish surface water harbored similar genetic characteristics, which highlights surface waters as potential dissemination routes between wildlife and the human population. Even in a low-prevalence country such as Sweden, the occurrence of ESBL producing E. coli in wild gulls and the human population appears to be connected and the occurrence of ESBL-producing E. coli in Swedish gulls is likely a case of environmental pollution.


Isolation and verification of ESBL-producing Escherichia coli
Fecal samples were enriched in Brain Heart Infusion broth with 16 mg/L Vancomycin and subsequently cultivated on chromID R ESBL bacterial plates (Biomerieux). Species identity of presumptive E. coli was assessed using MALDI-TOF (Biotyper, Bruker Corporation, The Netherlands). To confirm ESBL phenotype, isolates were prepared and spread on Muller-Hinton agar (Linnaeus University in Kalmar) according to EUCAST disc diffusion method for antimicrobial susceptibility testing and five antibiotic discs-amoxicillin/clavulanic acid 30/ 1 μg, cefotaxime 5 μg, ceftazidime 10 μg, cefepime 30 μg, and cefoxitin 30 μg-were placed on the plate. Specific inhibition of bacterial growth around the antibiotic discs was used to identify ESBL phenotypes [13,14].

Characterization of ESBL-producing Escherichia coli
ESBL-genes were identified using the Check-MDR microarray system (CT-101 or -103) (Checkpoint BV, Wageningen, the Netherlands) and sequenced using Sanger sequencing [15]. All isolates were subjected to multilocus sequence typing (MLST) according to protocols at the University of Warwick web site (http://mlst.warwick.ac.uk/mlst/dbs/Ecoli). Alleles and STs were determined using BioNumerics 7.0 MLST Plug-in (Applied Maths, Gent, Belgium). All isolates were tested for susceptibility to additional 13 antibiotics by disc diffusion according to EUCAST disc diffusion method for antimicrobial susceptibility testing and/or using VetMIC GN-mo panels according to the manufacturer (SVA, Uppsala) (Table 1), with the exception of fosfomycin were susceptibility was determined by Etest (bioMérieux, Sweden). Epidemiological cut-off values for resistance according to EUCAST (www.eucast.org) were used. Isolates with decreased susceptibility to ! 3 antibiotic classes in including beta-lactam antibiotics were classified as multiresistant.
Transformation and characterization of plasmids carrying ESBLproducing E. coli-genes On a subset of randomly selected isolates, transfer of plasmids carrying ESBL-producing E. coli-genes was assessed by electroporation to ElectroMax DH10B cells (Life Technologies, Carlsbad, CA, USA) and transformation was confirmed by detection of the same genes as previously described [16]. All transformants positive for an ESBL gene were tested for plasmid replicon types using the Diatheva PBRT-kit (Fano, Italy). For transformants positive for incompatibility group incI1, the plasmid was subjected to plasmid MLST (pMLST) according to the Plasmid MLST database (http://pubmlst.org/plasmid/).
Transformants and corresponding donor cells were also tested for susceptibility to 14 antibiotics by microdilution according to CLSI standard [16] using VetMIC GN-mo panels (SVA, Uppsala) ( Table 1).

Results
Occurrence of ESBL-producing E. coli in wild gulls ESBL-producing E. coli were detected in 13 out of 96 sample collected from Malmö and 16 out of 74 samples collected from Gothenburg, yielding occurrence frequencies of 14% and 22% respectively and 17% combined. None of the 29 isolates were resistant to carbapenems.

Comparison to other sectors
The characteristics of the isolates from the current study (Table 1) were compared to isolates from two other studies conducted in the same year in Sweden; a large Swedish national study reported by Börjesson et al [11], investigating the dissemination of ESBL-producing in community carriers, bloodstream infection and livestock as well as a study by Egervärn et al [12] investigating ESBL-producing E. coli in Swedish surface water. Comparisons were made at the level of ESBL gene, plasmid replicon type, MLST and the combinations of all mentioned.
ESBL-producing E. coli genes. In Fig 1, the frequency of ESBL-producing E. coli genes in different sectors is visualized. The vast majority, 89%, of identified genes in gulls were of CTX-M-type and 11% were of SHV-type. The dominating gene in Swedish gulls, bla CTX-M-15 , was detected in 45% of the isolates and bla CTX-M-14 was the second most common gene, detected in 17% of isolates. bla CTX-M-15 and bla CTX-M-14 were the most prevalent genes isolated from community carriers and bloodstream infections [11] and bla CTX-M-15 was the most common gene in Swedish surface water [12]. ESBL-producing E. coli isolates from poultry were exclusively of bla CTX-M-1 type and isolates from pigs/calves carried genes similar to the human and environmental sectors.
Plasmids and combinations of plasmids and genes. In Fig 2, the frequency of plasmid replicon type connected with plasmid carrying ESBL-producing E. coli genes in different sectors is visualized. ESBL genes identified in E. coli isolates from gulls were mainly connected to plasmids belonging to different IncF replicon types and IncI1. The distribution of plasmid types in gulls was similar to the distribution of plasmid types in community carriers and bloodstream infections. A high percentage of plasmids from gulls (50%), community carriers (36%), bloodstream infections (42%) and pigs/calves (50%) could not be transferred.
Eight isolates from gulls were successfully assigned a replicon type and genotype, out of which six isolates displayed combinations of plasmid replicon types and genes that were found in healthy human carriers and  (Table 2). Three isolates from gulls had identical plasmid/gene combinations to isolates found in Swedish surface water. Only one isolate from pigs/calves had a combination of plasmid/gene that was identical to one isolate found in gulls, IncI1-bla CTX-M-15 . Isolates from poultry displayed no plasmid/gene overlap when compared to isolates from gulls. To further characterize a selected proportion of isolates, plasmids of IncI1 type (n = 2) were subjected to pMLST. IncI1-pST3 carried the bla SHV-12 gene and IncI1-pST175 carried the bla CTX-M-15 gene. These two combinations of pMLST, plasmid replicon type and gene were not found among isolates from other sectors.

ST-types and combinations of ST-types, plasmids and genes.
Nine out of twelve STtypes identified from gulls were detected in Swedish community carriers and/or bloodstream infections in the same year (Fig 3). Four out of the overlapping nine ST-types were also found in Swedish surface water in the same year [12]. ST10 and ST58 were found in Swedish calves and ST155 in Swedish poultry sampled in the same year [11]. The combination ST131bla CTX-M-15 was found in one gull in this study. This combination was very common in bloodstream infections where ST131-bla CTX-M-15 was found in 31% of all isolates. When comparing clonal distributions, defined as identical ST, plasmid, and gene, of E. coli isolates from the different sectors, no overlapping combinations could be identified.

Discussion
In this study, we found that the genetic characteristics of ESBL-producing E. coli in Swedish wild gulls mirrors those found in humans living in Sweden. The two most common ESBL genes identified in E. coli isolates from Swedish gulls were bla CTX-M-15 and bla CTX-M-14 , which are the same genes most frequently observed in isolates from community carriers and bloodstream infections in Sweden. Furthermore, isolates from Swedish gulls and humans carried ESBL-plasmids belonging to IncF-and IncI1-types, and in both populations, a large proportion of genes belonging to bla CTX-M-15 could not be transferred. The unsuccessful transfer of bla CTX-M-15 genes could be due either to a non-transferrable plasmid, or that the gene was chromosomally located, the latter have previously been described to be relatively common in ESBL-producing E. coli harbouring bla CTX-M-15 [17]. The genes bla CTX-M-15 /bla CTX-M-14 and plasmids IncF/IncI1 were also common in isolates from surface water and healthy pigs/calves, previously recognized as groups carrying ESBL-producing E. coli strains with characteristics similar to strains carried by humans [11,12]. The genetic characteristics of ESBL-producing E. coli in gulls and humans were generally different to the characteristics of poultry isolates. In Swedish poultry, isolated ESBL-producing E. coli carried bla CTX-M-1 genes. However, the clear majority of cephalosporin resistant E. coli in Swedish poultry are of pAmpC-type and harbour bla CMY-2 genes [18]. No bla CMY-2 genes were detected in Swedish gulls in this study, implying that genetic transfer between the wild bird population and the poultry population is limited. It should nevertheless be noted that the current study used chromID ESBL agar plates, which are primarily selective for ESBL-producing isolates. It is therefore likely that we underestimated the occurrence of bla CMY-2 and other pAmpCs. Though if strains carrying bla CMY-2 had been common in Swedish gulls it is highly likely that at least a fraction would have been detected using our methodology. In concordance with our findings, most previous studies have also reported a dominance of bla CTX-M-type genes in gulls [6, 19- . While in European farm animals, bla CTX-M-1 appears to be the most common ESBL gene in cattle, pigs and poultry, and in poultry the pAmpC gene bla CMY-2 also appears to be widespread [29]. There was a high heterogeneity of STs found in gulls in this study. All of the STs have previously been detected in isolates from humans (MLST Database of Warwick, 2017-04-01), and a majority of them were detected in isolates from Swedish humans in the same year [30]. The similarities in genetic characteristics in isolates from gulls and humans indicate transmission of ESBL-producing E. coli between the two groups. However, despite the overlapping sequence types, ESBL-genes and ESBL-carrying plasmids in gulls and humans, there was no evidence of clonal spread due to the absence of identical combinations of sequence type/gene/plasmid. Although one should be aware that the lack of direct overlap between the different sector could also be due to the small sample size in the current study. The fact that only one isolate was collected per sample in each of the studies, likely also influenced the outcome. However, the absent evidence of clonal spread and heterogeneity of ST-types could also suggest dissemination of genes encoding ESBLs rather than clonal transmission of resistant bacteria. ESBL-genes are transferred easily between bacteria through mobile genetic elements, primarily via horizontal transfer of plasmids [31]. A high diversity of ESBL-producing E. coli in communities is a common finding in previous studies, especially pronounced in CTX-M-type ESBL-producing E. coli [32]. On the contrary, two findings from our study could be consistent with clonal transfer-firstly, one of the gull isolates from Gothenburg might be derived from the successful subclone of bla CTX-M-15 -producing Escherichia coli ST131. This sub-clone has spread as a pandemic among humans, often identified as the pathogen in severe bloodstream infections [33]. Secondly, ST617, ST38 and ST10, found in one, two and seven gulls respectively, are also sequence types with clinical importance in human medicine [34][35][36][37]. A recent study also described high relatedness of clinical isolates of ST38 from UK and one isolate of ST38 isolated from a Mongolian wild bird [36]. In that study it was also shown that the ST38 isolates carried the genes bla CTX-M-14 and bla CTX-M-15 chromosomally, and in the current study these same two genes in ST38 were non-transferable with transformation (Table 1).
In our study, ESBL-producing E. coli occurred in 14% and 22% of gulls sampled in the Swedish cities Malmö and Gothenburg, respectively, 17% combined. This is slightly higher than the reported occurrence in gulls sampled in Stockholm in 2010, 9% [38], but at a similar level as gulls sampled in the Swedish city Hudiksvall in 2009, 21% [5]. Thus, it appears the occurrence frequency in Swedish gulls is relatively stable the last few years, although one should be careful in making direct comparisons due to small sample sizes (200-300) and different sampling locations and to some extent species of gulls. Interestingly enough the occurrence of ESBL-producing E. coli in wild gulls was therefore more than three times higher than that described in Swedish human carriers [30]. This finding is also in concordance with studies performed in other European countries, but with the difference that the ratio of ESBL-producing E. coli in gulls versus humans is often even higher [5]. This difference in carriages might be due to a higher degree of exposure and possible accumulation of resistant bacteria of the gulls; e.g. proximity to human activity has been identified as a risk factor for occurrence of antibiotic resistant bacteria in wild animals in several studies [39][40][41][42]. One such human activity could be wastewater treatment plants (WWTPs) which have been recognized as hotspots for the accumulation, selection, and spread of antibiotic resistance to the environment [43]. This as gulls are often observed in connection to open WWTP basins containing untreated faecal material from entire cities. After sewage treatment in WWTPs, bacterial load is reduced by approximately 99%, but ESBL-producing E. coli are still frequently detected in rivers downstream of WWTPs and exemplifies a direct influx of antibiotic resistance from humans to the environment [44]. In this study, the isolates from gulls and humans showed high similarities to those originating from Swedish surface waters, suggesting WWTPs as one possible source. Comparable a recent Norwegian study described that ST131, ST10 and ST38 carrying the genes bla CTX-M-1 , bla CTX-M-14 and bla CTX-M-15 were frequently identified from surface waters, wastewaters and human UTIs [37]. The same gene and MLST combinations were also frequent in Swedish gulls, humans and surface waters (Table 1, Fig 1, Fig 3).
There is evidence supporting a human-to-wild bird directed transmission of antibiotic resistant bacteria/antibiotic resistant genetic elements, but an important question is whether birds can also transmit antibiotic resistance to the human population? As pointed out by Guenther et al [1], in contrast to the human population, there is no sewage system for bird feces and droppings are shed directly into the environment, potentially exposing human and animal populations to resistant bacteria. Interestingly, wild birds have been found to transmit Campylobacter to crops of peas which led to an outbreak of Campylobacteriosis in humans in Alaska [45] and Campylobacter from wild birds have been isolated from children´s playground in New Zealand [46]. Furthermore, a case of bovine salmonellosis in a Japanese dairy farm was associated with Salmonella enterica Typhimurium in wild sparrows habituating near the farm [47]. As pointed out earlier, the ESBL-producing E. coli isolates from Swedish gulls resemble isolates from Swedish surface water used as drinking water sources. Wild gulls are waterfowls that feed and defecate in the water, and these settings could potentially serve as a dissemination route of ESBL-producing E. coli from wild birds back to humans. One can also not rule out other settings such as faeces-contaminated agricultural land fields and playgrounds for small children as transmission routes to humans.

Conclusions
The genetic similarities to human isolates and high occurrence frequency of ESBL-producing E. coli in gulls suggest that ESBL-producing E. coli are a form of environmental pollution. The results of the study support the hypothesis that gulls can function as environmental reservoirs and as indicators of environmental contamination of ESBL genes. In addition, the results suggest that gulls could be used as indicators of what types of antibiotic resistance are circulating in a human population. Our findings highlight the need for efforts to minimize the risk of exposing wildlife to human waste and sewage to prevent further contamination and dissemination of antibiotic resistance.