Carriage of antibiotic resistant bacteria in endangered and declining Australian pinniped pups

The rapid emergence of antimicrobial resistance (AMR) is a major concern for wildlife and ecosystem health globally. Genetic determinants of AMR have become indicators of anthropogenic pollution due to their greater association with humans and rarer presence in environments less affected by humans. The objective of this study was to determine the distribution and frequency of the class 1 integron, a genetic determinant of AMR, in both the faecal microbiome and in Escherichia coli isolated from neonates of three pinniped species. Australian sea lion (Neophoca cinerea), Australian fur seal (Arctocephalus pusillus doriferus) and long-nosed fur seal (Arctocephalus forsteri) pups from eight breeding colonies along the Southern Australian coast were sampled between 2016–2019. DNA from faecal samples (n = 309) and from E. coli (n = 795) isolated from 884 faecal samples were analysed for class 1 integrons using PCRs targeting the conserved integrase gene (intI) and the gene cassette array. Class 1 integrons were detected in A. p. doriferus and N. cinerea pups sampled at seven of the eight breeding colonies investigated in 4.85% of faecal samples (n = 15) and 4.52% of E. coli isolates (n = 36). Integrons were not detected in any A. forsteri samples. DNA sequencing of the class 1 integron gene cassette array identified diverse genes conferring resistance to four antibiotic classes. The relationship between class 1 integron carriage and the concentration of five trace elements and heavy metals was also investigated, finding no significant association. The results of this study add to the growing evidence of the extent to which antimicrobial resistant bacteria are polluting the marine environment. As AMR determinants are frequently associated with bacterial pathogens, their occurrence suggests that these pinniped species are vulnerable to potential health risks. The implications for individual and population health as a consequence of AMR carriage is a critical component of ongoing health investigations.


Introduction
Aquatic ecosystems are being increasingly identified as a sink for antimicrobial resistance (AMR) [1,2]. Aquatic systems provide a transport medium for the global dissemination of antibiotic resistant bacteria (ARB) and associated antibiotic resistance genes (ARGs) [1][2][3].
The combination of ARB with antibiotic residues and other pollutants in aquatic environments also promote the proliferation and establishment of resistant bacterial communities [2,4]. The widespread dissemination of AMR can partly be attributed to horizontal gene transfer (HGT) which allows the transfer of ARGs and associated genetic machinery between diverse bacterial species, facilitating the acquisition of novel traits from the environment and other bacteria [5,6]. In Gram-negative bacteria in particular, the rapid evolution of resistance has been linked to HGT and mobile genetic elements [7][8][9]. Class 1 integrons, for example, are mainly found in Gram-negative bacteria [10] and are able to capture and subsequently express a multitude of ARGs [11,12], which can be transferred between bacteria via their association with transposons and plasmids. In the human context, the clinical class 1 integron is considered to be of high importance for AMR dissemination [13,14]. The class 1 integron contains a conserved 5' segment which encodes the integrase gene (intI1) and a varying number of gene cassettes that together form a gene cassette array [15]. The conserved intI1 is a useful genetic indicator of antimicrobial pollution as it is universally present, occurs in high abundance in humans and domestic animals, is highly abundant in waste streams and is rarely present in environments less affected by humans [16,17]. The recombination of gene cassettes is mediated by intI1, allowing the class 1 integron to capture, remove and express a variety of gene cassettes [15]. Variations of the class 1 integron are now emerging, with insertion sequences in the 3' segment, such as IS26, assisting in the dissemination of resistance genes in Gram-negative bacteria. These insertion sequences are associated with numerous genes that confer resistance to multiple antibiotic classes, and are able to promote and subsequently express these associated resistance genes [18,19].
Agricultural runoff, in addition to mining, municipal wastewater, and industrial and pharmaceutical waste are point sources of heavy metal pollutants frequently found in natural environments [20]. Heavy metals are considered to be co-selective agents of AMR [20]. Aquatic environments polluted by heavy metals have been associated with a greater incidence of class 1 integrons compared to non-polluted sites [21,22], through mechanisms of cross-and co-resistance [23]. The presence of heavy metals and antibiotic residues also common in the environment have the potential to exert a selective pressure that promotes the emergence and persistence of AMR in the environment [24][25][26]. As heavy metals can bioaccumulate and persist in the environment, such selective pressures are applied for extended periods of time, facilitating the development resistance traits in microbial communities [27]. However, there has been little investigation into whether there is increased acquisition of antibiotic resistant bacteria in humans and non-human animals in environments that have greater exposure to heavy metals.
Concentrations of essential trace elements and heavy metals including zinc (Zn), arsenic (As), selenium (Se), mercury (Hg) and lead (Pb) are of particular interest for wildlife health. The presence of Pb, even at low concentrations, can be associated with disease [28]. In contrast, Zn and Se are essential trace elements, but these too can have toxic effects at high concentrations [29]. Heavy metals have previously been identified in free-ranging pinnipeds [30][31][32], however, there has been no consideration of potential co-selection of ARGs in wildlife species associated with heavy metal exposure. Given the role that heavy metals play in the environmental amplification of ARB, investigating the levels of heavy metals and class 1 integron frequency in wildlife species could provide valuable insights into the factors contributing to the abundance and acquisition of ARB in free-ranging wildlife.
A diverse range of antibiotic resistant bacterial species have been detected in marine mammals [33,34], which are large-bodied, long-lived upper trophic predators considered to have a role as sentinels of marine health [35]. Escherichia coli is a Gram-negative bacterium that is commonly used as an indicator of anthropogenic pollution [24], and has been used for the investigation of class 1 integrons in many species [36][37][38][39][40]. Evidence suggests that class 1 integrons are more prevalent in E. coli isolates that are closely associated with anthropogenic pollution and human environments [7]. The presence of class 1 integrons in E. coli has been investigated in some species of free-ranging pinnipeds in the Southern Hemisphere. An absence of class 1 integrons was reported in E. coli isolated from free-ranging southern elephant seals (Mirounga leonina), Weddell seals (Leptonychotes weddellii) [41] and adult Australian sea lions (Neophoca cinerea) [40], although class 1 integrons were detected in E. coli from captive adult N. cinerea [40]. The presence of class 1 integrons in captive wildlife and comparative absence in free-ranging individuals suggest that environmental conditions and the intimate proximity to humans experienced in captivity can impact the acquisition of ARGs by wildlife species and is consistent for many wildlife species [37,38,40]. Consistent with the hypothesis that the presence of ARGs in wildlife is associated with proximity to humans, a class 1 integron was recently detected in E. coli from a single free-ranging N. cinerea pup at a colony with comparatively high anthropogenic influence compared to a more remote colony [39].
The occurrence of class 1 integrons and ARG carriage in two additional pinniped species inhabiting Australian waters, namely Australian fur seals (Arctocephalus pusillus doriferus) and long-nosed fur seals (Arctocephalus forsteri), has not been investigated. All three species, N. cinerea, A. p. doriferus, and A. forsteri, inhabit numerous offshore islands along the Australian coast from Western Australia to Tasmania [42], with the ranges of these pinniped species overlapping in South Australia. Colonies of each species experience differing levels of human interaction; those on islands remote to mainland Australia experience little to no contact with humans while others are popular, frequently visited tourist sites. The differing proximities of sympatric colonies to human habitation and exposure to anthropogenic impacts creates a naturally occurring gradient ideal for studying anthropogenic pollution in the marine environment.
The main objective of this study was to determine the prevalence of class 1 integrons and ARG carriage in both the faecal microbiota and E. coli isolates from pups of three pinniped species sampled at multiple breeding colonies throughout Southern Australia. Given the role heavy metals have as a co-selective agent for AMR, an additional aim was to determine whether there was a relationship between the concentration of trace elements and heavy metals (Zn, Se, As, Hg and Pb) and class 1 integron prevalence. It was hypothesised that class 1 integrons would be more abundant in pups at colonies in closer proximity to sources of anthropogenic pollution. This paper reports the presence of ARGs in E. coli isolates and faecal microbiota of N. cinerea and A. p. doriferus pups at multiple breeding colonies along the Australian coast. We explore the differences between species and colonies and discuss factors contributing to changes in class 1 integron prevalence across breeding seasons and colonies. We provide recommendations for future investigations to further understand the dissemination of AMR in free-ranging pinniped species.

Study sites and sample collection
Faecal swabs (n = 884) were collected from neonatal pups sampled at eight breeding colonies across multiple breeding seasons from 2016-2019 (Fig 1 and    forsteri pups were approximately 3-6 weeks of age and N. cinerea pups were 2-6 weeks of age at time of sampling. Samples were collected following methods described by Fulham et al. [39,43]. In brief, sterile swabs (Copan, Brescia, Italy) were inserted directly into the rectum of each pup and resulting samples were sub-sampled into sterile FecalSwab™ tubes (Copan, Brescia, Italy). FecalSwab samples were stored at 4˚C and cultured within 7-10 days of collection. Blood samples were collected from the brachial vein of pups as per methodology in Fulham et al. [39] and refrigerated at 4˚C prior to storage at -20˚C. Due to time and logistical constraints, blood collection was limited to pups sampled at Seal

E. coli culture, isolation and DNA extraction
FecalSwab™ samples were cultured following the methodology described by Fulham et al. [39,43]. In summary, a selective media, Chromocult1 coliform agar (Merck, Darmstadt, Germany), was used to isolate E. coli. After initial culture, E. coli colonies were sub-cultured and pure E. coli colonies were selected for DNA extraction based on colour and morphology. Pure E. coli colonies were inoculated into Luria-Bertani broth (5ml) and incubated at 37˚C for 24 hours in preparation for preservation and DNA extraction. DNA was extracted using a boil preparation method where the broth culture was centrifuged to pellet bacteria. Supernatant was decanted and bacterial pellet resuspended with sterile water (50μL). Samples were then heated for 5min at 95˚C followed by centrifugation and resulting bacterial lysates were stored at -30˚C.

Screening for class 1 integrons
All faecal samples and E. coli isolates positive for 16S rDNA were further screened for the presence of the class 1 integron integrase gene (intI1) using HS463a and HS464 primers (Table 2) following the methods described by Waldron and Gillings [45]. Samples containing intI1 were tested using additional PCRs to target the gene cassette array using HS458 and HS459 primers ( Table 2) and PCR conditions described by Waldron and Gillings [45]. Any samples that did not produce a band for the HS458/HS459 PCR were analysed using a secondary primer set consisting of HS458 and JL-D2, which targets the IS26 transposase, an alternate 3' terminus in integrons [18], using the conditions as described for HS458/ HS459.

Cloning, sequencing and analysis
Using the MinElute PCR Purification Kit (Qiagen, Melbourne, Australia), amplicons from the two gene cassette array PCRs (HS458/459 and HS458/JL-D2) were purified following manufacturer's instructions. Amplicons containing only a single band were sequenced directly using the purified PCR product.
Amplicons containing multiple bands, indicating the presence of more than one gene cassette, were cloned using the TOPO TA cloning kit and transformed into One Shot 1 DH5™-T1 R competent E. coli cells (Invitrogen, Carlsbad, CA, USA) as per manufacturer's protocol. Between six to twelve colonies of transformed E. coli were selected from each cloned sample and DNA from cell lysates were screened using HS458, HS459 and JL-D2 as described above. Amplicons of variable sizes were selected for sequencing.
Amplicons from HS463a/HS464 that did not amplify in HS458/HS459 or HS458/JL-D2 were purified and sequenced to confirm positive intI result.
Sequencing was performed at The Ramaciotti Centre for Genomics (University of New South Wales, Sydney, Australia) using Big Dye Terminator chemistry version 3.1 and ABI 3730/3730x1 Capillary Sequencers (Applied Biosystems, Foster City, CA, USA). Geneious Prime software (version 11.0.6; Biomatters Limited, Auckland, New Zealand) was employed to assemble and manually check sequences for quality. Assembled sequences were analysed for the presence of antibiotic resistance genes using Integrall (http:/integrall.bio.ua.pt). Class 1 integron gene cassette arrays were confirmed via detection of the 3' conserved region containing qacE. For arrays containing more than one gene cassette, the attC recombination site located between cassettes was identified using the highly conserved core sequence GTTRRRY and complementary inverse core sequence RYYAAC [47]. Representative sequences generated from this study have been lodged in GenBank under accession numbers OL960709-OL960714.

Statistical analyses
All statistical analyses were conducted using RStudio software (V 1.2.5042, Boston, Massachusetts, USA). The Shapiro-Wilk's test was used to test for normality of data and any variables with significant (<0.05) and non-normal distribution were log transformed which normalised the data set and allowed for parametric statistical analysis. Significance was determined when p < 0.05.
The statistical analysis of class 1 integron distribution was conducted using Fisher's exact test to test for differences in class 1 integron occurrence between species. Pearson's chisquared test was used to test for differences in class 1 integron occurrence within species across sampling sites and breeding seasons.
Welch's two sample t-test was used to test for significance in the relationship between integrons and trace element and heavy metal concentrations in A. p. doriferus pups (n = 52) sampled at Seal Rocks 2018. Of these 52 individuals, 14 were integron positive.

Detection of class 1 integrons
Escherichia coli was isolated from a total of 795 faecal samples (89.9%) with PCR screening for intI1 revealing 36 isolates (4.52%, n = 795) that tested positive based on the presence of the expected 473bp product. Seven of the positive E. coli isolates were from N. cinerea and 29 from A. p. doriferus. Screening of faecal DNA detected intI1 in 15 faecal samples (4.85%, n = 309), with four from N. cinerea and 11 from A. p. doriferus. All faecal samples and E. coli isolates from A. forsteri were negative for intI1 (Fig 2).

Gene cassette array diversity
DNA sequencing identified five different gene cassette arrays from the 51 positive samples. The majority of the samples (n = 40) contained gene cassette arrays void of ARGs. Of the samples containing integrons with ARGs, seven contained a single gene cassette and the remaining four arrays each had two gene cassettes (Fig 3). Class 1 integrons identified in samples from A. p. doriferus were the most diverse, encoding seven different ARGs, while only two types of ARGs were detected in N. cinerea (Fig 3). The most common cassette array was dfrA7 (n = 4), identified in E. coli isolate DNA from both N. cinerea and A. p. doriferus pups.
The vast majority (49 of 51) of gene cassette arrays detected in this study contained the typical 3' conserved segment (qacEΔ), whereas, in the remaining two gene cassette arrays, qacEΔ was replaced with an IS26 transposase (Fig 3).

Discussion
This study identified class 1 integrons in both E. coli and faecal DNA from free-ranging N. cinerea and A. p. doriferus pups at seven breeding colonies in Southern Australia, representing the first time ARGs have been detected in A. p. doriferus, and in the faecal microbiota from N. cinerea pups. The occurrence of class 1 integrons in A. p. doriferus pups is of particular interest given the higher carriage of intI1 in comparison with the other pinniped species studied.
There was similar class 1 integron abundance and gene cassette diversity across all four A. p. doriferus colonies sampled in 2017. These colonies differ in terms of size, topography, pup production and population density [48] and cover a wide geographical area. The similar intI1 abundances indicates that these colonies are exposed to similar sources and levels of anthropogenic pollution, however, the number of intI1 genes detected at Seal Rocks showed considerable change over sampling years (2017-2019), with a significant increase observed in 2018. This increase was not sustained over multiple breeding seasons and further investigation is needed to determine if this increase is due to a gradient of anthropogenic pollution or whether it is an aberrant finding at this colony. Seal Rocks was the only A. p. doriferus colony sampled over multiple breeding seasons limiting our ability to fully assess and compare trends in intI1 abundance across other A. p. doriferus colonies over time.
The highest number of class 1 integrons was detected in A. p. doriferus pups at Seal Rocks. Class 1 integrons are generally more prevalent in locations closer to urbanised environments that are exposed to higher levels of anthropogenic pollution [16,49]. Seal Rocks is located 1.8km from Phillip Island and approximately 7km from the Mornington Peninsula, both of which are densely populated over summer months when pups are born and sampled. Seal Rocks is also located within 150km of Melbourne, Australia's second largest city (~5 million people), which could result in continuous exposure of pinnipeds to a higher number of anthropogenic sources of pollution compared to A. p. doriferus pups at the other colonies sampled (for example, Deen Maar Island and Cape Bridgewater are~250km and 300km respectively from Melbourne), thereby facilitating the greater acquisition of class 1 integrons.
The abundance of intI1 is presumed to change rapidly based on environmental factors [16]. Rapid changes in environments caused by extreme weather events can also introduce higher levels of runoff and industrial waste into the marine environment and distribute pollutants across wider geographical ranges [50]. Environmental conditions at pinniped breeding colonies were not considered as part of this study, however, given the differences in intI1 abundance observed, the influence of stochastic events on the presence of class 1 integrons would need to be considered when attempting to understand the trends observed in A. p. doriferus pups.
In addition to the higher abundance of intI1 genes detection in A. p. doriferus pups at Seal Rocks in 2018, was the presence of two IS26 class 1 integron variants in E. coli isolates. There Gene symbols are as follows: dfrA genes encode dihydrofolate reductases that confer resistance to trimethoprim; aacA genes encode aminoglycoside (6') acetyltransferases (aacA) that confer resistance to aminoglycoside antibiotics; aadA genes encode aminoglycoside (3") adenylyltransferases that confer resistance to streptomycin and spectinomycin; arr3 genes encode ADP-ribosyl transferases that confer resistance to rifampin; qac genes encode efflux pumps that confer resistance to quaternary ammonium compounds; qacEΔ and IS26 represent the 3' terminus of some the gene cassette arrays depicted.
https://doi.org/10.1371/journal.pone.0258978.g003 are multiple reports of IS26 class 1 integron variants in E. coli from animals in Australia, including cows and pigs [18,51,52] and Grey-headed flying foxes (Pteropus poliocephalus) [53], as well as in human clinical cases [54]. The presence of class 1 integrons with IS26 variants in E. coli isolates from free-ranging A. p. doriferus pups provides further evidence to suggest that this colony is exposed to a marine environment that is contaminated by various sources of anthropogenic pollution.
The finding of the arr3 gene cassette in multiple A. p. doriferus pups at Seal Rocks in 2018 was unexpected. This gene cassette, arr3, encodes ADP-ribosyl transferases which potentially confers resistance to rifampicin and has been identified in numerous Gram-negative pathogens (such as Proteus spp.) [55,56] and integrons in bacterial isolates from hospital patients [57]. The aacA4-arr3-qacEΔ cassette array detected in A. p. doriferus is not widely reported in the literature [57], however, the closest matches on GenBank were predominately detected in Proteus spp. and Klebsiella spp. (e.g. CP053614, LC549808) isolates from hospital patients and zoo and production animals in China. In this study, the array was detected in faecal DNA and the bacterial carrier remains unknown. As the arr3 cassette is associated with human and domestic animal pathogens, the presence of this gene cassette in free-ranging pinnipeds provides further evidence to suggest that the Seal Rocks colony is exposed to anthropogenic microbial pollution.
In contrast to A. p. doriferus and N. cinerea, class 1 integrons were not detected in A. forsteri pups (n = 150) sampled over multiple breeding seasons. Over 5000 pups are born at the Cape Gantheaume colony each breeding season [58] and there is a much higher population density compared to Seal Slide, a small N. cinerea colony <10km away. Despite the similarity in location and higher population density between these two pinniped species a class 1 integron was detected in one of four N. cinerea sampled pups during the 2018 breeding season. The difference in intI1 abundance between species is likely multifactorial and further investigation is necessary to better understand the factors that contribute to the acquisition of class 1 integrons in free-ranging pinniped species.
Heavy metals are environmental pollutants that can act as co-selecting agents for antibiotic resistance [20]. Heavy metals do not degrade and therefore persist in the environment for long periods of time, maintaining selective pressure for extended periods [22]. Furthermore, the abundance of class 1 integrons has been found to correlate with heavy metal concentrations, with previous studies focusing on the relationship between heavy metal pollutants and intI1 abundance in environmental samples including water and soil [21,59]. It has been well established that concentrations of heavy metals are present in free-ranging marine mammals [31,32] and can bioaccumulate in upper trophic predators [30,60]. Despite the association between heavy metal concentrations and class 1 integron abundance in environmental samples, this relationship has not previously been investigated in free-ranging wildlife. A significant relationship between the concentrations of ARB carriage and two trace elements and three heavy metals was not seen, although the small sample size for this specific comparison could have limited this analysis. In addition, the current comparison was limited to concentrations of trace and heavy metals in whole blood which reflect recent exposure. The relationship between ARB carriage and heavy metal concentrations in other tissues (for example, liver) could be explored, given the bioaccumulation potential of this tissue when compared to whole blood [61].
Free-ranging wildlife are exposed to differing environmental conditions, selective pressures and exposure to ARB that likely drive the acquisition of ARGs [62]. Pinniped pups sampled as part of this study were less than two months of age and confined to the breeding colonies. As such, the only potential source of exposure to bacteria and ARGs for pups are those present in the breeding colony environment, which could be contaminated by wastewater run-off, faecal contamination from other wildlife species (including sea birds) and juvenile and adult pinnipeds that inhabit these colonies [36,63]. Evidence suggests that the prevalence of ARB in freeranging wildlife is influenced by exposure to anthropogenic pollution and environmental contamination [62], with the latter varying depending on habitat occupation and behaviours exhibited by wildlife species [62]. For example, foraging is a behaviour that can increase the likelihood of wildlife species being exposed to ARB [3]. The pinniped species studied herein have differing foraging strategies. While N. cinerea and A. p. doriferus are benthic foragers [64][65][66][67], A. forsteri are pelagic foragers [68,69]. In some aquatic species, differences in toxicant accumulation between benthic and pelagic feeders has been observed [61], thus differences in foraging behaviours could lead to exposure of higher levels of pollutants and increased ARB acquisition in benthic feeding pinniped species.
Another factor to consider is the proximity to wastewater treatment plants and the release of effluent. The environment created within wastewater treatment plants has been found to promote the proliferation of ARB through the exposure of bacteria to sub-inhibitory concentrations of antibiotics [70], disinfectants, and heavy metals [16]. Effluent released from wastewater treatment plants is being increasingly identified as an important source of ARB, with some studies finding higher frequencies of integron-positive E. coli downstream of such effluent [71]. Colonies located in closer proximity to wastewater treatment plants could therefore be exposed to higher levels of ARB, which could influence the acquisition of ARB and ARGs in free-ranging pinniped populations.
Interactions with other species in ecosystems is another aspect to consider when attempting to understand the transfer of AMR. The presence of other species at pinniped breeding colonies such as sea birds, known carriers of numerous ARGs [63], could also influence the carriage of AMR in free-ranging pinnipeds. Sampling both environmental substrates and other wildlife species present within study areas is required to gain a greater understanding of the source of AMR and the transfer of ARGs in free-ranging wildlife species.

Conclusion
This study detected bacteria carrying diverse gene cassettes encoding resistance to multiple classes of antibiotics in two species of free-ranging pinniped pups in Australia. The detection of class 1 integrons, mobile genetic elements that have been identified as useful indicators of antimicrobial pollution, suggests these populations are exposed to anthropogenic pollution. Furthermore, the detection of E. coli carrying IS26 class 1 integron variants in free-ranging pinniped pups indicates that these isolates originated from domestic animals and/or humans.
Further investigation to better understand how antibiotic resistant bacteria are being acquired by free-ranging pinniped pups is critical and could be used as an additional mechanism to monitor anthropogenic pollution in marine ecosystems. Ongoing monitoring of antibiotic resistant bacteria in these species will also assist in understanding the role of increasing anthropogenic pollution on the long-term survival of these marine sentinel species. marine charters. We thank Robert McQuilty, Royal Prince Alfred Hospital NSW, for his assistance in trace and heavy metal analysis.