Resistance to critically important antimicrobials in Australian silver gulls (Chroicocephalus novaehollandiae) and evidence of anthropogenic origins

Abstract

Objectives

Antimicrobial resistance (AMR) to critically important antimicrobials (CIAs) amongst Gram-negative bacteria can feasibly be transferred amongst wildlife, humans and domestic animals. This study investigated the ecology, epidemiology and origins of CIA-resistant Escherichia coli carried by Australian silver gulls (Chroicocephalus novaehollandiae), a gregarious avian wildlife species that is a common inhabitant of coastal areas with high levels of human contact.

Methods

Sampling locations were widely dispersed around the perimeter of the Australian continent, with sites separated by up to 3500 km. WGS was used to study the diversity and molecular characteristics of resistant isolates to ascertain their epidemiological origin.

Results

Investigation of 562 faecal samples revealed widespread occurrence of extended-spectrum cephalosporin-resistant (21.7%) and fluoroquinolone-resistant (23.8%) E. coli. Genome sequencing revealed that CIA-resistant E. coli isolates (n = 284) from gulls predominantly belonged to human-associated extra-intestinal pathogenic E. coli (ExPEC) clones, including ST131 (17%), ST10 (8%), ST1193 (6%), ST69 (5%) and ST38 (4%). Genomic analysis revealed that gulls carry pandemic ExPEC-ST131 clades (O25:H4 H30-R and H30-Rx) and globally emerging fluoroquinolone-resistant ST1193 identified among humans worldwide. Comparative analysis revealed that ST131 and ST1193 isolates from gulls overlapped extensively with human clinical isolates from Australia and overseas. The present study also detected single isolates of carbapenem-resistant E. coli (ST410-bla_OXA-48) and colistin-resistant E. coli (ST345-mcr-1).

Conclusions

The carriage of diverse CIA-resistant E. coli clones that strongly resemble pathogenic clones from humans suggests that gulls can act as ecological sponges indiscriminately accumulating and disseminating CIA-resistant bacteria over vast distances.

Introduction

Antimicrobial resistance (AMR) is regarded as one of the greatest threats to human and animal health.¹ Resistance to critically important antimicrobials (CIAs) such as extended-spectrum cephalosporins (ESCs), fluoroquinolones (FQs), colistin and carbapenems amongst Enterobacteriaceae and other Gram-negative bacteria is a major public health concern.^2,3 This is attributed to co-associated resistance to other classes of antimicrobials and limited therapeutic options to treat infections caused by CIA-resistant bacteria in both humans and animals.¹

Globally, emergence and dissemination of CIA resistance in livestock, wild birds and companion animals is a major concern due to the potential for direct or indirect transfer of such resistant bacteria to humans.^4,5 Examples of this include widespread occurrence of ESC- and FQ-resistant E. coli in livestock and companion animals,^6,7 recent reports of plasmid-mediated colistin resistance genes (mcr) in livestock E. coli⁸ and the detection of ESC-, FQ- and carbapenem-resistant E. coli in wildlife such as herring gulls, hybrid deer and wild boars.^9–13 In particular, wildlife reservoirs of these resistant elements may be involved in pathways of dissemination into livestock and humans, which have not yet been investigated.

Australia is a country that has low levels of CIA resistance among livestock.^14–17 This is attributed to Australia’s unique geography, quarantine restrictions, including bans on importation of livestock and fresh meat from 1901, and conservative regulations governing the use of CIAs including ESCs, FQs and carbapenems in food-producing animals.^14,16 However, recent studies have demonstrated that the status quo is changing, with recent reports of carbapenem-resistant Salmonella enterica serovar Typhimurium carrying bla_IMP-4 from cats¹⁸ and the carriage of carbapenem resistance in E. coli from a single, large, off-shore gull colony in New South Wales.¹⁹ Another study identified two ciprofloxacin-resistant E. coli isolates from Australian pigs,²⁰ implying that despite strict biosecurity and absence of antimicrobial selection pressure, food-producing animals can still be at risk.

The report of carbapenem resistance in sources such as gulls raises public health concern regarding the role of these and other wild birds as reservoirs and in the potential transmission of CIA-resistant bacteria in the community. In this study, we performed a cross-sectional survey of gull faecal flora from across Australia to identify the carriage of CIA-resistant E. coli. In addition, comparative genomic analysis was performed to identify the origins of these CIA-resistant E. coli clones isolated in the current study from a global perspective.

Materials and methods

Sample collection

Swabs (n = 562) were taken from freshly voided faeces from silver gulls (December 2015–April 2017) and transported in Amies charcoal medium (Copan). Sampling occurred in all Australian states (excluding the Northern Territory and Australian Capital Territory) and consisted of sites where silver gulls congregated on beaches and foreshores. The sampling locations included Geraldton (August 2016), Perth (Bibra Lake and Cottesloe; November 2016), Bunbury (November 2016) and Busselton (November 2016) in Western Australia; Adelaide (February 2017) in South Australia; Hobart (October 2016) in Tasmania; Wollongong (November 2015), Sydney (August 2016), Batemans Bay (December 2016), Ulladalla (December 2016) and Kiama (December 2016) in New South Wales (December 2016); Sandgate (March 2017), Gold Coast (March 2017) and Manley (March 2017) in Queensland; and Melbourne (April 2016 and April 2017) and Geelong (April 2016) in Victoria.

Bacterial isolation

Swabs were incubated in buffered peptone water (3 mL; ThermoFisher) for 4 h followed by aerobic culturing (37°C; 16–20 h) on three selective agar plates to identify FQ-resistant (MacConkey agar infused with 1 mg/L ciprofloxacin, ThermoFisher), ESC-resistant (Brilliance ESBL, ThermoFisher) and carbapenem-resistant (Brilliance CRE, ThermoFisher) E. coli. Presumptive E. coli (one colony per plate) were subcultured and species identity was confirmed by MALDI-TOF MS (Bruker). If the first isolate per plate was not identified as E. coli, another presumptive E. coli (if present) was subcultured and entered into the study.

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing was performed on 284 E. coli isolates using disc diffusion susceptibility testing as per CLSI guidelines and recommended breakpoints for interpreting phenotypic antimicrobial resistance were applied.²¹ The 12 antimicrobials used were: ampicillin, ciprofloxacin, imipenem, meropenem, trimethoprim/sulfamethoxazole, tetracycline, gentamicin, ceftriaxone, streptomycin, cefoxitin, amoxicillin/clavulanate and chloramphenicol. E. coli ATCC 25922 was used as a control as per CLSI guidelines during antimicrobial susceptibility testing.²¹

Isolates in which the transmissible colistin resistance gene mcr-1 was detected were subjected to MIC determination using the broth microdilution method as per ISO 20776-1:2006 CLSI using 96-well microtitre plates (Thermo Fisher, Australia).²¹

WGS

Genome sequencing was undertaken on 284 E. coli isolates using Illumina Chemistry (Nextera XT) as previously described.²² Generated sequence reads were deposited in the NCBI Sequence Read Archive database under accession number PRJNA486855.

Data analysis

Raw reads were de novo assembled using SPAdes²³ and antimicrobial resistance genes, serotypes, multilocus STs and virulence factors inferred with the aid of the Center for Genomic Epidemiology (http://www.genomicepidemiology.org/). SNPs within QRDRs were identified using Snippy v3.2.²³ SNP analysis for all data sets except ST131 was conducted using the Nullarbor platform²³ and parsed through Gubbins v2.3.1²⁴ to remove recombination. The phylogenetic analysis was performed using PhyML v3.0.²⁵ For analysis regarding ST131, sequence reads from international H27 and H30 isolates (n = 58)²⁶ and all 40 available Australian ST131 isolates (38 human isolates, one companion animal and one unknown source) identified in EnteroBase (http://enterobase.warwick.ac.uk) with accessible raw sequence data on 19 March 2017 were included for SNP detection using NASP,²⁷ as previously described.²⁶ Selected isolates were removed due to low sequence depth to increase core genome size prior to purging using Gubbins v2.3.1. Phylogenetic analysis was performed using PhyML v3.0 with Smart Model Selection,²⁵ rooted with the fimH 41 (H41) cluster and visualized using iTol.²⁸

Results

CIA-resistant E. coli carriage among gulls

Of the total number of samples collected (n = 562) from across eastern (n = 384), western (n = 144) and southern (n = 34) states of Australia, 24% (n = 135) were positive for FQ-resistant E. coli, while 22% (n = 125) were positive for ESC-resistant E. coli. One isolate was confirmed as carbapenem resistant by phenotypic and WGS methods despite 5% of swabs yielding growth on Brilliance CRE plates. The statewide frequency of FQ- and ESC-resistant E. coli isolates per swab is presented in Figure 1.

Figure 1.

Frequency of detection of CIA-resistant E. coli from each swab collected from gulls based on growth selection plate and confirmed by disc diffusion. WA, Western Australia; SA, South Australia; QLD, Queensland; VIC, Victoria; NSW, New South Wales; TAS, Tasmania; FQ_R, FQ resistant; CRO_R, ceftriaxone resistant; CRE_R, carbapenem resistant.

Phenotypic and genotypic characteristics of CIA-resistant E. coli

A total of 284 E. coli isolates recovered from selective agars were subjected to antimicrobial susceptibility testing. The E. coli isolates exhibited high levels of resistance to ampicillin (86%), trimethoprim/sulfamethoxazole (56%), tetracycline (51%) and streptomycin (48%). Resistance was high to the CIAs ciprofloxacin (64%) and ceftriaxone (62%). Only one isolate was resistant to imipenem, but no resistance was demonstrated towards meropenem. Low levels of resistance were demonstrated to amoxicillin/clavulanate (21%), cefoxitin (18%), gentamicin (18%) and chloramphenicol (12%). The statewide and overall frequency of CIA-resistant E. coli from gulls is presented in Table 1.

Table 1.

Number of E. coli isolates demonstrating phenotypic resistance to CIAs in disc diffusion antimicrobial susceptibility testing across different states

	Western Australia	South Australia	Queensland	Victoria	New South Wales	Tasmania	Overall
Swabs, na	144	34	76	114	153	41	562
Positive swabs, n (%)b	57 (40)	15 (44)	28 (37)	77 (68)	55 (36)	26 (63)	258 (46)
Isolates, n (%)c	57 (40)	23 (68)	26 (34)	85 (75)	65 (42)	28 (68)	284 (51)
AmpicillinR, n (%)	46 (81)	20 (87)	23 (88)	75 (88)	57 (88)	23 (82)	244 (86)
CiprofloxacinR, n (%)	37 (65)	13 (57)	17 (65)	55 (65)	45 (69)	17 (61)	184 (65)
ImipenemR, n (%)	0	0	0	1 (1)	0	0	1 (0.4)
MeropenemR, n (%)	0	0	0	0	0	0	0
CeftriaxoneR, n (%)	27 (47)	16 (70)	12 (46)	52 (61)	47 (72)	21 (75)	175 (62)
CefoxitinR, n (%)	6 (11)	3 (13)	5 (19)	8 (9)	23 (35)	5 (18)	50 (18)
Amoxicillin/clavulanateR, n (%)	6 (11)	6 (26)	9 (35)	8 (9)	21 (32)	9 (32)	59 (21)

R

, resistant.

a

Total number of swabs collected.

b

Total number of swabs positive for E. coli.

c

Total number of individual isolates picked from three different selective agars.

Molecular characterization

WGS of 284 E. coli isolates revealed a high frequency of carriage of resistance genes bla_TEM and bla_CTX-M-15 (41% and 29%, respectively) in the E. coli isolates. The other resistance genes detected were bla_CMY-2 (13%), bla_CTX-M-27 (12%) and bla_CTX-M-14 (6%). Plasmid-mediated quinolone resistance determinants qnrB4, qnrB6 and qnrS1 were detected in 16% of the E. coli isolates (Table 2).

Table 2.

Numbers and percentages of selectively recovered E. coli isolates from gull faecal samples across different states of Australia containing genes encoding CIA resistance

	Western Australia	South Australia	Queensland	Victoria	New South Wales	Tasmania
Swabs, n	144	34	76	114	153	41
Isolates, n	57	23	26	85	65	28
qnr	7 (12)	2 (9)	8 (31)	15 (18)	4 (6)	7 (25)
blaCMY-2	6 (11)	1 (4)	3 (12)	2 (2)	19 (29)	5 (18)
blaCMY-13	0 (0)	0 (0)	0 (0)	0 (0)	1 (2)	0 (0)
blaCMY-42	0 (0)	0 (0)	0 (0)	2 (2)	0 (0)	0 (0)
blaCMY-60	0 (0)	0 (0)	2 (8)	0	0 (0)	0 (0)
blaCTX-M-11	0 (0)	0 (0)	0 (0)	1 (1)	0 (0)	0 (0)
blaCTX-M-14	4 (7)	1 (4)	1 (4)	4 (5)	6 (9)	1 (4)
blaCTX-M-15	10 (18)	8 (35)	5 (19)	35 (41)	13 (20)	11 (39)
blaCTX-M-24	0 (0)	0 (0)	0 (0)	1 (1)	0 (0)	0 (0)
blaCTX-M-27	5 (9)	4 (17)	4 (15)	8 (9)	6 (9)	7 (25)
blaCTX-M-3	1 (2)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
blaCTX-M-55	1 (2)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
blaOXA-1	2 (4)	3 (13)	2 (8)	17 (20)	2 (3)	3 (11)
blaOXA-48	0 (0)	0 (0)	0 (0)	1 (1)	0 (0)	0 (0)
blaSHV	0 (0)	0 (0)	0 (0)	2 (2)	0 (0)	0 (0)
blaTEM	28 (49)	10 (43)	16 (62)	32 (38)	26 (40)	5 (18)
mcr-1	1 (2)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)

One E. coli ST410 isolate from Victoria recovered from Brilliance CRE was carbapenem resistant and positive for bla_OXA-48. Another E. coli ST345 isolate from Western Australia (recovered from MacConkey agar infused with 1 mg/L ciprofloxacin) harboured mcr-1 and was phenotypically resistant to colistin (MIC = 12.5 mg/L).

Phylogenetic analysis

MLST of 284 CIA-resistant isolates revealed a heterogeneous population of E. coli clones represented by 72 STs (Table S1, available as Supplementary data at JAC Online). The major STs were ST131 (17% of isolates), ST10 (8%), ST1193 (6%), ST69 (5%), ST38 (4%), ST2179 (4%), ST744 (4%) and ST963 (4%) (Tables 3 and 4). MLST was indeterminate for six isolates due to low sequence coverage.

Table 3.

E. coli STs of public health significance; total numbers and n (%) isolated from different selective media

ST	Total number of E. coli isolates	MacConkey+ciprofloxacin plates, n (%)	Brilliance ESBL plates, n (%)	Brilliance CRE plates, n (%)
ST131	47	33 (70)	9 (19)	5 (11)
ST10	23	8 (35)	15 (65)	0
ST1193	16	14 (88)	1 (6)	1 (6)
ST69	15	2 (13)	13 (87)	0
ST38	11	3 (27)	8 (73)	0
ST2179	10	4 (40)	0	6 (60)
ST744	10	9 (90)	1 (10)	0
ST963	10	0	9 (90)	1 (10)
ST405	7	4 (57)	3 (43)	0
ST450	7	7 (100)	0	0
ST773	4	4 (100)	0	0

Table 4.

Distribution in numbers and percentages of dominant E. coli STs across different states

	Western Australia	South Australia	Queensland	Victoria	New South Wales	Tasmania	Overall
Swabs, n	144	34	76	114	153	41	562
Isolates, n	57	23	26	85	65	28	284
ST, n (%)
ST131	11 (19)	8 (35)	1 (4)	11 (13)	9 (14)	7 (25)	47 (17)
ST10	4 (7)	1 (4)	3 (12)	9 (11)	6 (9)	0	23 (8)
ST1193	7 (12)	1 (4)	0	3 (4)	4 (6)	1 (4)	16 (6)
ST69	1 (2)	0	2 (8)	10 (12)	0	2 (7)	15 (5)
ST38	4 (7)	0	2 (8)	1 (1)	2 (3)	2 (7)	11 (4)
ST2179	0	0	0	0	10 (15)	0	10 (4)
ST744	1 (2)	0	0	5 (6)	4 (6)	0	10 (4)
ST963	2 (4)	0	0	1 (1)	4 (6)	3 (11)	10 (4)
ST405	0	0	0	2 (2)	5 (8)	0	7 (2)
ST450	4 (7)	0	0	2 (2)	1 (2)	0	7 (2)
ST773	0	0	0	4 (5)	0	0	4 (1)

E. coli ST131

There are three main clades within ST131, i.e. O16:H5 fimH41 (clade A), O25:h3 fimH22 (clade B) and O25:H4 fimH30 (clade C). Within clade C there are two dominant clades: H30-R (C1) and H30-Rx (C2).²⁹ The ST131 isolates (n = 47) from Australian gulls predominantly belong to clade C (n = 37, 79% of total ST131 isolates). Among the clade C isolates, 23 of the isolates belong to H30-R (C1) and 14 belong to H30-Rx (C2, positive for bla_CTX-M-15). In addition, nine H30-R isolates were positive for bla_CTX-M-27, demonstrating acquisition of ESC resistance. A small proportion of ST131 isolates were clade A, which contained five O16:H5 fimH41 isolates, two ST131-O16:H5 isolates positive for a fimH141 allele and three ST131-O25:H4 isolates with a fimH41 allele (n = 10, 21.3% of total ST131 isolates).

FQ resistance in E. coli ST131 isolates was attributed to point mutations in the gyrA, gyrB, parC and parE subunits within the QRDRs. Mutations in the gyrA subunit at nucleotide positions C248T and G259A resulting in Ser-83→Leu and Asp-87→Asn alterations along with changes in nucleotide position G239T (Ser-80→Ile) of the parC subunit were observed in 34 (72%) ST131 isolates. The main uropathogenic E. coli (UPEC) virulence-associated genes identified in E. coli ST131 isolates included toxin-encoding genes cnf1, sat and astA, the siderophore-related gene iroN, the adhesin-encoding gene iha and the protectin/invasin-encoding gene iss.³⁰

Phylogenetic analysis of E. coli ST131 in gulls and publicly available ST131 genomes from humans including Australian isolates (Figure 2) revealed distinct clustering in relation to overall fimH types [H41, H22 and H31 (data not shown)]. Similar to previous findings, an FQ-resistant clade (ST131 H30-R) and an FQ-resistant and ESBL-producing H30-Rx sister clade were observed, that combined 95% of the CIA-resistant gull ST131 isolates (Figure 2). Significant overlap was evident between the Australian gull and human isolates within each of these groups.

Figure 2.

Radial phylogenetic tree of all non-H41 E. coli ST131 from gulls and publicly available ST131 genomes from humans including Australian isolates showing distinct clustering of fimH types. Two sister groups were evident in the H30 cluster, an FQ-resistant clade (ST131 H30-R) and an FQ-resistant and ESBL-producing H30-R nested clade (H30-Rx). Intermingling of international and Australian isolates is observed, as is the overlap between isolates of human and gull origin.

E. coli ST10

The second most frequently detected E. coli ST (n = 23, 8%) was the broad host range lineage ST10 reported to cause severe infection in humans, livestock and wild birds.¹⁵ Of the total number of isolates identified, 6 isolates (26%) were ciprofloxacin resistant, 12 isolates (52%) were ceftriaxone resistant and 5 isolates (22%) demonstrated resistance to both ciprofloxacin and ceftriaxone. O9:H9 fimH54 (6 isolates; 26%) was the dominant serotype among the ST10 isolates. The ESC resistance-encoding genes bla_CTX-M-15, bla_CTX-M-27, bla_CTX-M-14 and bla_CTX-M-11 were identified in 35%, 13%, 7% and 4% of the isolates, respectively. Sequence analysis detected QRDR FQ resistance-associated mutations in gyrA, gyrB, parC and parE in all isolates. The main virulence-associated genes identified were iss, astA, pic, iha and iroN. The phylogenetic analysis revealed multiple clades of E. coli ST10 (Figure S1) with no distinct geographical clustering.

E. coli ST1193

Another major E. coli ST identified belonged to ST1193, an emerging virulent and resistant lineage³¹ (Figure 3). All isolates (n = 16) were resistant to ciprofloxacin and susceptible to ceftriaxone, except two isolates, which were positive for bla_CTX-M-15 and demonstrated resistance to ceftriaxone. The FQ resistance of all isolates was attributed to the mutations in the QRDRs. The majority of the isolates (n = 15) belonged to serotype O75:H5 and the fimH64 variant. The virulence factors iha, sat and vat were most frequently identified. Phylogenetic analysis of these isolates demonstrated heterogeneous make up of ST1193 clones that belong to three different clusters distributed evenly across all states except Queensland.

Figure 3.

A maximum likelihood phylogeny obtained using SNPs comparing E. coli ST1193 from Australian silver gulls with globally collected ST1193, indicating the wide geographical dispersal of E. coli ST1193. The inner circle with coloured dots represents the continent of origin. The E. coli ST1193 isolates detected in gulls from Australia are scattered throughout the phylogeny.

E. coli ST69

Susceptibility to ciprofloxacin was demonstrated by 80% of the 15 E. coli ST69 isolates, a strain that is considered to be a highly virulent and globally successful UPEC.³² The serotypes identified included O17/O77:H18 (seven isolates, 47%), O17/O44:H8 (four isolates, 27%), O25:H18 (two isolates, 13%), O15:H1 (a single isolate) and O15:H6 (a single isolate). All isolates were fimH27 except one isolate from Tasmania with serotype O17/O77:H18 and H297. Resistance to ceftriaxone was demonstrated by 93% of the isolates carrying one of the resistance-conferring genes bla_CTX-M-15, bla_CTX-M-27 or bla_CTX-M-14. The common virulence-associated genes were iha, iss, sat and iroN. Multiple clades of ST69 were evident in the phylogenetic tree. The majority of the isolates from Victoria grouped together in the phylogeny, while the isolates from the other states were intermingled (Figure S2).

E. coli ST38

The different serotypes identified among the 11 E. coli ST38 isolates detected included serotype O86:H18 with H5 and O1:H15 with H65. Resistance to both ciprofloxacin and ceftriaxone was observed in six isolates. One isolate demonstrated ciprofloxacin resistance, while four isolates demonstrated resistance towards ceftriaxone only. The genes conferring ESC resistance were bla_CTX-M-15, bla_CTX-M-27, bla_CTX-M-14 and bla_CMY-2. The major virulence-encoding genes identified were astA, iss, iha and sat. Phylogenetic analysis revealed an even distribution of ST38 clones across all states of Australia excluding South Australia (Figure S3).

Discussion

This study aimed to identify the carriage of CIA-resistant E. coli isolated from Australian gulls and describe the underlying resistance determinants, virulence genes and phylogenetic relationships of these resistant isolates. A high level of CIA-resistant E. coli carriage was observed among Australian gulls, along with detection of carbapenem and colistin resistance at low frequencies. The rate of CIA-resistant E. coli carriage in gulls was surprisingly high in a country where low levels of FQ- and ESC-resistant E. coli were observed in a national surveillance programme albeit using non-selective culturing in human sepsis cases (7.3%–11.2% and 6.4%–10.8%, respectively)³³ and in livestock (0%–1% and 0%–5%, respectively).^15,17,20

A large proportion of the CIA-resistant E. coli STs recovered were identified in this study as human-associated clones, suggesting bidirectional transmission between humans and gulls or the existence of an unidentified reservoir with ecological links to both. Additionally, the CIA-resistant E. coli among gulls was widely dispersed around the perimeter of the Australian continent. Most notably, E. coli ST131 was found to be common. This is a globally disseminated, highly virulent strain in humans that is often associated with FQ resistance and CTX-M-15-type ESBL production.³⁴ The E. coli ST131 isolated from the gull faeces also belonged to the O25:H4, H30, CTX-M-15-positive group (Figure 2), which is the most predominant lineage associated with urinary tract infections in humans.³⁵

ST10, ST69 and ST405 have previously been identified as UPEC,³⁶ and the ST405 lineage has been described as a global disseminator of CTX-M-type ESBL, in a similar manner to E. coli ST131.³⁷ Another noteworthy observation is the identification of FQ-resistant ST1193, which has been described as a rapidly emerging extra-intestinal pathogenic E. coli (‘ExPEC’) clone with widespread occurrence in humans in the USA.³¹ Also detected in this study was E. coli ST744, which has been identified as a human pathogen with the potential to develop resistance in the presence of selection pressure.³⁸ Table 3 lists the various STs identified from gull faecal samples.

In this study, one bla_OXA-48-positive E. coli isolate was detected from Melbourne, Victoria. Since its initial identification in Turkey,³⁹ OXA-48 has emerged as the most prevalent carbapenemase worldwide.⁴⁰ The presence of OXA-48 carbapenemase-producing Enterobacteriaceae has extended beyond the hospital setting and entered the community, livestock, companion animals, wildlife and the environment.^18,41,42 There are reports of detection of NDM-1-producing Salmonella in wild birds from Germany,¹⁰ VIM-1-producing E. coli in yellow legged gulls from France¹¹ and IMP-4 Salmonella in silver gulls from Australia.¹⁹ Thus far, there are no published reports of detection of OXA-48 in wild birds. Moreover the presence of the bla_OXA-48 gene in E. coli ST410 is concerning, as recently this clone has been identified as one of the internationally emerging ‘high-risk’ clones.⁴³ In contrast to the previous study, no IMP-4-producing E. coli was detected in gulls in the present study, although sampling was performed at the same locations of New South Wales, Australia (Five Islands, Wollongong). The detection of such human-associated CIA resistance genes in gulls is suggestive of possible interspecies transmission and indicates risk of zoonotic transfer to humans and other animals including livestock.

This study reports, to the best of our knowledge, the first detection of the colistin resistance-encoding gene mcr-1 in E. coli isolates obtained from Australian animals. The isolate identified as E. coli ST345 was obtained from a single silver gull from Cottesloe, Western Australia. Following the first detection of mcr-1 in Europe in 2016 from a gull (Larus argentatus),⁴⁴ studies from other continents (South America and Asia) have identified migratory wild birds carrying mcr-1-positive E. coli.^8,45

The heterogeneous populations of E. coli ST131 and other major CIA-resistant E. coli clones detected in this study exhibited extensive overlap with human clinical isolates. The source of emergence of CIA-resistant bacteria in wildlife can be attributed to multiple factors. Human waste, wastewater treatment facilities and livestock waste are all considered to be the major hotspots for CIA-resistant E. coli as well as for horizontal gene transfer of resistance genes.⁴⁶ Furthermore, the close proximity of gulls and humans is considered to be yet another factor resulting in transmission of CIA-resistant bacteria. Although gulls are mostly known to be opportunistic marine feeders, they also readily feed on leftover human food and garbage, which provides an opportunity to acquire bacteria harbouring resistance genes. The gull samples collected in this study were from areas with dense human populations, which may have resulted in gulls acquiring the CIA-resistant E. coli at higher frequency than in more remote regions.

The migratory habits of gull species have been cited as the major factor responsible for dissemination of CIA-resistant bacteria.⁹ The Australian silver gull’s movement pattern is thought to be confined to within 80–1600 km from their hatchery—as such they are designated non-migratory.⁴⁷ Nevertheless, the occurrence of silver gulls in human and livestock environments is very common due to the species’ successful adaptation as a scavenger. The extent of gull–human co-mingling is high, particularly in coastal urban areas of Australia, thus providing an easy route for exchange of faecal flora. Moreover, gulls are often observed to share roosting locations with other bird species, such as the little tern (Sternula albifrons), which has a worldwide distribution and migration pathways extending between Asia and Australia.⁴⁸ Additionally, they also inhabit tidal lagoons and wetlands with ducks and wading birds, which may then travel further inland to farming regions. This increases the possibility of transferring these resistant clones to food-producing animals, either via direct contamination of feed and water, or via a ‘leapfrog’ effect through other aquatic birds, which are not as coastal centric. The potential therefore exists for gulls to be a key element in a web of wildlife capable of acquiring and disseminating CIA-resistant E. coli not only between human populations over vast distances but also in livestock and agricultural produce.²²

Further studies would necessitate identification of CIA-resistant E. coli from gull populations in remote locations with sparse human populations to evaluate the possible impact of human activity on transmission of CIA-resistant bacteria and the source of acquisition of these bacteria by gulls.

Conclusions

This study establishes that Australian silver gulls are carriers of virulent and CIA-resistant human-associated E. coli clones. The detection of transferable colistin- and carbapenem-resistant E. coli clones in an Australian animal also raises questions regarding the sources of these resistant bacteria. The relatedness of isolates among the various STs across states indicates that gulls can act as ecological sponges, indiscriminately accumulating both pathogenic and non-pathogenic CIA-resistant bacteria and potentially disseminating them in a cycle encompassing humans and the broader environment.

Acknowledgements

We would like to thank Mrs Bertha Rusdi for her technical assistance in the laboratory.

Funding

This study was supported by a small research grant from the School of Veterinary and Life Sciences (Murdoch University, Murdoch, Western Australia, Australia). A PhD scholarship and operating fund were awarded to S. M. by The University of Adelaide.

Transparency declarations

None to declare.

Author notes

Mark O’Dea and Sam Abraham authors contributed equally.