Occurrence and Antimicrobial Resistance Traits of Escherichia coli from Wild Birds and Rodents in Singapore

Antimicrobial resistance (AMR) in Escherichia coli (E. coli) poses a public health concern worldwide. Wild birds and rodents, due to their mobility, are potential vehicles for transmission of AMR bacteria to humans. Ninety-six wild birds’ faecal samples and 135 rodents’ droppings samples were collected and analysed in 2017. Forty-six E. coli isolates from wild birds and rodents were subjected to AMR phenotypic and genotypic characterisation. The proportion of E. coli isolates resistant to at least one of the antimicrobials tested from wild birds (80.8%) was significantly higher than that of isolates from rodents (40.0%). The proportion of E. coli isolates resistant to each antimicrobial class for wild birds was 3.8% to 73.1% and that for rodents was 5.0% to 35.0%. Six out of 26 E. coli isolates from wild birds (23.1%) and two out of 20 (10.0%) isolates from rodents were multi-drug resistant (MDR) strains. These MDR E. coli isolates were detected with various antimicrobial resistance genes such as blaTEM-1B and qnrS1 and could be considered as part of the environmental resistome. Findings in this study suggested that wild birds and rodents could play a role in disseminating antimicrobial resistant E. coli, and this underscores the necessity of environment management and close monitoring on AMR bacteria in wild birds and rodents to prevent spreading of resistant organisms to other wildlife animals and humans.


Introduction
Escherichia coli (E. coli) is a commensal bacterium found in the guts of animals [1,2]. It can be pathogenic and cause gastroenteritis, bacteraemia and urinary tract infections [2][3][4]. The bacterium is known to be susceptible to selection pressure [2] and has the high competency to pick up and transfer antibiotic resistance genes to and from other bacterial strains [2,5]. Resistance to antimicrobials using BioEdit version 7.2.6.1 software. Assembled reads were uploaded to nucleotide BLAST database (https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE_TYPE = BlastSearch) (National Centre for Biotechnology Information, United States) for E. coli identification (>90%).

Extended Spectrum Beta-Lactamases (ESBL) Testing for E. coli Isolates
Resistance to Ceftriaxone by disk diffusion was confirmed for ESBL production, as previously described [7].

Genotypic Characterisation by Whole Genome Sequencing
Based on phenotypic resistance results, eight MDR E. coli isolates (indicated in Table 1) were selected and subjected to genotypic analysis by whole genome sequencing. DNA extraction, library preparation and sequencing were performed as previously described [7]. Raw sequence data was deposited into Genbank under Bioproject accession number PRJNA625931. The raw reads were assembled using SPAdes version 3.11.0, with "-careful, -k auto and -cov-cutoff as off" parameters [14]. The genome data were analysed with reference to the ResFinder 3.1 database (https://cge.cbs.dtu.dk/services/ResFinder/) to identify antimicrobial resistance genes and chromosomal point mutations based on the following parameters: minimum length coverage of 60% and minimum identity of 90% (Centre for Genomic Epidemiology, Denmark) [15].

Occurrence of E. coli in Wild Birds and Rodents
Of the 96 wild birds' faecal samples, E. coli was detected in 26 (

Antimicrobial Resistance in E. coli Isolated from Wild Birds and Rodents
Antimicrobial susceptibility test results for all the E. coli isolates are shown in Table 1. There were 80.8% (21/26) of E. coli isolates from wild birds and 40.0% (8/20) of E. coli from rodents being resistant to at least one of the antimicrobials tested in the study ( Table 3). The proportion of E. coli isolates resistant to at least one of the antimicrobials tested from wild birds (80.8%, Z-score 2.8, p < 0.05) was significantly higher than that of isolates from rodents (40.0%) ( Table 3). The proportions of E. coli isolates resistant to each antimicrobial class for wild birds (3.8% to 73.1%) and rodents (5.0% to 35.0%) are shown in Figure 1 and Table 4.
For isolates from wild birds, the most common phenotypic resistance exhibited was against Penicillins (Ampicillin), followed by Beta-lactam/Beta-lactamase Inhibitor Combinations (Amoxicillin/Clavulanic Acid), Tetracyclines (Tetracycline), Quinolones (Nalidixic acid) and Phenicols (Chloramphenicol). Lower resistance rates (less than or equal to 15.0%) were found for the other remaining five antimicrobial classes (Table 4). No resistance was found for Aminoglycosides (Amikacin) and Carbapenems (Meropenem).   Table 4. Percentage of antimicrobial resistance in E. coli isolates from wild birds and rodents (Corresponding to Figure 1).

Percentage of Isolates Showing Resistant Phenotype (n) Wild Birds (26)
Rodents ( Table 4. Percentage of antimicrobial resistance in E. coli isolates from wild birds and rodents (Corresponding to Figure 1). For isolates from rodents, the most common phenotypic resistance exhibited was against Penicillins (Ampicillin). Lower resistance rates (less than or equal to 15.0%) were observed for other remaining seven antimicrobial classes (Table 4). No resistance was found for Aminoglycosides (Gentamicin), Third Generation Cephalosporins (Ceftriaxone), Aminoglycosides (Amikacin) and Carbapenems (Meropenem).

Percentage of Isolates Showing
Six out of 26 E. coli isolates from wild birds (23.1%) and two out of 20 (10.0%) isolates from rodents were resistant to three or more antimicrobial classes and considered as multi-drug resistant (MDR) strains. One MDR E. coli isolate (C1722) recovered from wild birds tested positive for ESBL production.

Distribution of Resistance Genes in Eight MDR E. coli Isolates from Wild Birds and Rodents
At least one antimicrobial resistance gene was detected in all eight MDR E. coli isolates from wild birds and rodents (Table 5). An isolate (C1722) from wild bird had 17 resistance genes detected, the highest among the eight MDR E. coli isolates. It is worth noting that this isolate was ESBL-producing and harboured bla CTM-X-65 gene and tet(X) gene.
Chromosomal point mutations in Quinolone resistant determining regions (QRDR) of gyrA and parC genes were observed in 37.5% (3/8) of the MDR E. coli isolates from wild birds. A gyrA mutation for amino acid substitution from Serine to Leucine at 83th position (Ser83Leu) was found in an isolate (C1722), which displayed resistance to Nalidixic Acid and Ciprofloxacin (Table 6). Two isolates (C1776 and C1797) carried double gyrA mutations responsible for amino acid change from Serine to Leucine at 83th position (Ser83Leu) and aspartic acid to asparagine at 87th position (Asp87Asn) and parC mutation for change from Serine to Isoleucine at 80th position (Ser80Ile). These two isolates were resistant to Nalidixic acid, Ciprofloxacin and Norfloxacin (Table 6).

Comparison between Phenotypic and Genotypic Characteristics of MDR E. coli Isolates
Comparison of antimicrobial phenotype and whole genome sequencing data of MDR E. coli isolates (n = 6) from wild birds and rodents (n = 2) was performed for Beta-lactams, Quinolones, Phenicols and Tetracyclines. The remaining five antimicrobial classes were not performed for comparison between antimicrobial resistance phenotype and genotype due to limitations of the antimicrobial susceptibility testing as follows: (1) Macrolides and Fosfomycin were not tested; (2) Many antimicrobial classes of Aminoglycosides (e.g., Kanamycin, Streptomycin) were excluded; (3) A combination of Sulphamethoxazole/Trimethoprim was used and there was no testing for individual antimicrobial classes of Sulphonamides and Trimethoprim.
In general, there was good agreement between phenotypic and genotypic characteristics for Phenicols and Tetracyclines. Discrepancies between phenotypic and genotypic resistance traits were detected in MDR E. coli isolates (Table 6). For instance, Quinolones resistance genes were detected in four MDR E. coli isolates (C1742, C1758, C1805_1 and 8645_0135) which were, however, phenotypically susceptible to the Quinolones included in this study. Isolate 8655_0114 was phenotypically resistant to Quinolones with no corresponding resistance gene. Three MDR E. coli (C1758 and C1797, 8655_0114) were phenotypically resistant to Penicillins and Beta-lactam/Beta-lactamase Inhibitor Combinations but there was no ESBL resistance gene detected.

Discussion
To our knowledge, this is the first report on the occurrence and antimicrobial resistant phenotype and genotype in E. coli isolates from wild birds and rodents in Singapore.
This study revealed that the occurrence of E. coli in wild birds (27.1%) in Singapore was relatively lower than that reported in other countries such as Switzerland (53.7%) and Saudi Arabia (93.0%) [16,17]. Similarly, the occurrence of E. coli in rodents (14.8%) was relatively lower than it is reported in other countries such as Trinidad and Tobago (83.8%) and Canada (62.7%) [18,19]. One possible reason for relatively lower occurrences on both wild birds and rodents could be due to differences in sampling and laboratory methods used in respective studies (e.g., the convenient collection of samples used in this study) that rendered comparison of occurrence data between studies challenging. Furthermore, the occurrence data could be affected by storage conditions of the collected samples. Another limitation of our study was that we were unable to identify the rodent species despite it being known that common rodent species found in Singapore include Black Rat (Rattus norvegicus), Brown Rat (Rattus rattus) and House Mouse (Mus musculus) [20]. The occurrence of E. coli could hypothetically originate from these pools of common rodent species. Our study provided an insight into the occurrence of E. coli isolates from wild birds and rodents in Singapore, which enhances our understanding of the local epidemiology of E. coli and could guide future epidemiological studies.
The proportion of E. coli isolates resistant to at least one of the antimicrobials tested from wild birds was significantly higher than that of isolates from rodents. Although different in behavior, diet and migration potential with regard to species, wild birds generally have a higher movement pattern than rodents and could have higher exposure to antimicrobial resistance determinants in the ecological niches. This could lead to a higher probability for wild birds in disseminating AMR determinants to humans or other wildlife animals [21]. Our results differ from a previous study in Singapore which indicated there was no phenotypic antimicrobial resistance detected in Salmonella isolates recovered from wild birds [22]. The difference in phenotypic resistance observed for E. coli and Salmonella isolates could be due to the fact that E. coli has a greater ability to acquire resistance than Salmonella, making E. coli more susceptible to antimicrobial selection pressure than Salmonella for the tested antimicrobials [23,24]. This implies the importance of using multiple bacteria organisms (both commensal and pathogenic) as AMR indicators in surveillances for better understanding of the distribution of resistant organisms or resistance determinants in the environment.
The antimicrobial resistance rate among E. coli isolates could be related to the usage of antimicrobials. E. coli isolates from both wild birds and rodents were phenotypically resistant to Penicillins, whereas isolates recovered from wild birds also displayed resistance to Beta-lactam/ Beta-lactamase Inhibitor Combinations and Tetracyclines. These antimicrobials are commonly used in clinical and agricultural sectors [25][26][27][28][29][30]. This is of public health concern as these antimicrobials are the first-line drugs of choice for empirical treatment of infections caused by E. coli. The widespread resistance to these antimicrobials will render these antimicrobials ineffective for the treatment of infections and will increase the need for an alternative antimicrobial therapy option.
The lower percentage (less than or equal to 15.0%) of resistance observed in E. coli isolates from wild birds to antimicrobial classes of Aminoglycosides and Third Generation Cephalosporins suggested that there could be a relatively lower selection pressure of these antimicrobial classes as compared to other commonly used antimicrobials in wild birds. As these antimicrobial classes are not commonly used in food animal production [4,31], nor in clinical settings as a first-line drug for E. coli infection but as a drug of choice for invasive/resistant infections [4,32,33], the level of antimicrobial residual pollution in the environment is expected to be lower, and consequently lower exposure and selection pressure for wild birds [34]. Nevertheless, further monitoring of efficacy for these antimicrobials remains necessary.
Our study detected E. coli isolates from wild birds and rodents that were resistant to Quinolones (Nalidixic Acid), which is an indicator of reduced susceptibility for Fluoroquinolones (e.g., Ciprofloxacin and Norfloxacin). Indeed, Nalidixic Acid resistant E. coli isolates from wild birds (n = 5) also showed resistance to Ciprofloxacin (60.0%, 3/5) and Norfloxacin (40.0%, 2/5). A similar observation was found in rodents where Nalidixic Acid-resistant E. coli isolates (n = 2) were resistant to Ciprofloxacin (50.0%, 1/2) and Norfloxacin (50.0%, 1/2). These findings indicated the possible increasing trend for Fluroquinolones-resistant organisms present in the environment. It is important to implement continuous monitoring of Quinolone and Fluroquinolone resistance in E. coli from wild birds and rodents in order to prevent the spread of the resistance determinants into environmental niches.
All MDR isolates in our study were detected with MLS resistance gene mdf(A). A study showed that this resistance gene is expressed constitutively in E. coli [35] and it encodes for a multidrug efflux pump [36]. Another study has shown that the expression of mdf(A) in E. coli would confer multidrug resistance [37]. Isolate C1776 (from wild bird) harboured another MLS resistance gene mph(A), which encodes for enzymes capable of inactivating Erythromycin (a type of Macrolides) [38].
Our observation of tet(A) being the most common tetracycline resistance gene in all MDR isolates was in agreement with other studies in birds and rodents [26,39,40]. Isolate C1797 from wild bird was detected with the tet(B) gene-another gene which codes for efflux pump to transport Tetracycline out of bacterial cell [7]. Isolate C1722 from wild bird harboured tet(X), which is the first described tetracycline resistance gene that encodes for an enzyme which inactivates Tetracycline. The tet(X) gene was previously identified in environmental bacteria from soil, sewage plants and human clinical samples [41].
Aminoglycosides resistance genes encode for enzymes such as acetyltransferases (such as aac (3)-IV detected in our study), nucleotidyltransferases (such as aadA1 and aadA2 found in this study) or phosphotransferases (such as aph(3 )-Ia and aph (6)-Id identified in our study) which inactivate Aminoglycosides. The most common Aminoglycoside genes detected were aadA1, aadA2 and aph (6)-Id. Genes aadA1 and aadA2 confer resistance to Aminoglycosides (e.g., Spectinomycin and Streptomycin) [42] and they were also found in a study on birds from Australia [43]. The gene aph (6)-Id was reported as the most commonly detected gene in E. coli clinical isolates from Egypt [44].
Both Phenicol resistance genes floR and cml1a that encode for efflux transporter [45] were detected in our study. The floR gene was reported in Klebsiella pneumoniae from clinical isolates [46] and E. coli from cattle [47], while cml1a was found in E. coli from poultry [48]. Another gene catA1 was found in isolate C1797, which encodes for an enzyme that inactivates Phenicols [45] and it was previously also reported in birds [43].
Two types of ESBL resistance genes (bla TEM and bla CTX-M ) were detected in MDR E. coli isolates and both resistance genes encode for Amber Class A Beta-lactamases [45]. The most commonly detected ESBL resistance gene bla TEM-1B , encodes for TEM-1 Beta-lactamase that hydrolyses Penicillins and First Generation Cephalosporins [7], and was reported in another study on birds [49]. The gene bla TEM-176 , was reported in birds such as gull and rook [9]. On the other hand, bla CTX-M-65 which encodes for CTX-M Beta-lactamase that hydrolyses Penicillins and First to Third Generation Cephalosporins [50], was reported in clinical isolates [51] and raw retail chicken [52].
Qnrs1 encodes for a protein which binds to and protects both DNA gyrase and topoisomerase IV from inhibition by Quinolones [53]. Qnrs1 represented the most commonly detected PMQR gene among MDR isolates in this study and this is in agreement with other studies on E. coli and Enterobacteriaceae from birds [54,55]. Another PMQR gene oqxB that was detected in this study, was shown to encode for multidrug efflux pump [56]. This gene was found in Enterobacteriaceae from rooks throughout the European continent [55] and human clinical isolates from Korea [56]. Sulphonamide resistance genes sul2 and sul3 were identified, which is in line with other reports on birds [43,49] and rodents [40].
This study detected three MDR E. coli isolates with chromosomal point mutations in QRDR. Mechanisms of resistance to Quinolones include target gene mutations, active efflux pumps, decreased permeability for outer membrane and acquisition of resistance genes such as qnrS1 [30,57]. Target gene mutations include alteration of QRDR in DNA gyrase subunit A (gyrA) and topoisomerase IV subunit C (parC) [58]. In this study, detection of gyrA and parC mutations was found in 37.5% of the MDR E. coli isolates. These gyrA and parC mutations were identical as reported previously for Quinolone-resistant E. coli strains [57]. One ESBL-producing and MDR E. coli isolate (C1722) which displayed resistance to Nalidixic Acid and Ciprofloxacin was found to have a single gyrA mutation. Our finding differed from the report which indicated that a single gyrA mutation would result in resistance to Nalidixic Acid and an additional mutation in gyrA or parC would be required for resistance in Ciprofloxacin [57]. The different phenotypic resistance observed in the C1722 isolate could be due to the resistance mechanism for co-existence of the ESBL resistance gene (such as bla CTM-X-65 ) and single gyrA mutation, which was reported in a clinical study from China [51].
Our data, supported by other studies, suggest that the antimicrobial resistance genotype does not always correspond well with phenotypic expression and vice versa [59,60]. It is known that there are multiple complex mechanisms that can lead to bacteria becoming resistant to antimicrobials [59,60]. In the absence of corresponding resistance genes that encode for proteins responsible for enzymatic degradation of antimicrobials and alteration of bacterial proteins targeted by antimicrobials, isolates can exhibit resistance due to other mechanisms such as porin loss and efflux pumps [61]. For example, the detection of mdf(A) in MDR E. coli isolate from rodent (8655_0114), which encodes for a multidrug efflux pump, could possibly explain the observation that this isolate was conferred resistance to many antimicrobials but had no detectable corresponding resistance genes.
In our study, PMQR genes were detected in four MDR E. coli isolates (C1742, C1758, C1805_1 and 8645_0135) which were, however, phenotypically susceptible to the Quinolones tested. This is congruent with a study on E. coli from environmental samples in pig farms, which reported that PMQR genes alone are insufficient to confer resistance to Quinolones [62]. As discussed in the previous paragraph, further mechanisms such as mutations in QRDR regions would confer resistance to Quinolones. Despite the absence of phenotypic resistance traits, isolates harbouring antimicrobial resistance genes may be subjected to a transfer of resistance determinants to other bacterial isolates/species via horizontal gene transfer [63]. Thus, risks posed by susceptible isolates carrying resistance determinants should not be underestimated. Our findings suggested that the bacterial isolates should be characterised for both phenotypic and genotypic resistance traits, for a holistic interpretation and more thorough risk assessment. To have a more comprehensive antimicrobial resistance gene portfolio of the E. coli isolates obtained in this study, further genotypic screening could be carried out for the remaining antimicrobial resistant E. coli isolates.
Our study reports antimicrobial resistant E. coli isolates from wild birds and rodents in Singapore. In addition, genotypic characterisation by whole genome sequencing revealed the diversity of resistance genes in eight MDR E. coli isolates, which demonstrated the value of whole genome sequencing as an epidemiological tool for further understanding of antimicrobial resistance gene profiles in bacterial isolates. Once wild birds and rodents acquire antimicrobial resistant bacteria, these bacteria could continue to colonise and infect the hosts [64]. Therefore, wild birds and rodents could play a role for the dissemination of antimicrobial resistant bacteria and/or genes across different wildlife species and environmental sectors, perhaps via their faecal materials, as supported by other E. coli studies on wild birds and rodents [10,16,21,40,65]. An increasing number of cities today are undergoing urban rewilding, which transforms dense urban areas into green cities with nature assimilated. While the increased biodiversity in cities brings many benefits, it could also facilitate the crossovers of antimicrobial resistance pathogens or genes between urban and wildlife ecosystems. Hence, a close monitoring programme on the antimicrobial resistant bacteria in wildlife, especially in those animals that are in close proximity to human habitats, is recommended to complement surveillance systems in food animals, food and humans.

Conclusions
This study provides baseline data of occurrence and antimicrobial resistance characteristics of E. coli in wild birds and rodents representing a part of the environment in Singapore. Wild birds and rodents could play a contributing role to further spread antimicrobial resistance to other wildlife and environmental sectors through faecal contamination. The findings of our study highlight the