Extraintestinal pathogenic E. coli (ExPEC) strains have an enhanced ability to traverse from their usual gut environment to normally sterile extraintestinal body sites and cause disease (Johnson et al., 2016). Before the 2000s most ExPEC strains were susceptible to critically important antimicrobials (CIAs) such as fluoroquinolones (FQs) and extended-spectrum cephalosporins (ESCs) (Pitout, 2012), however, multidrug resistance is increasingly common globally (Nicolas-Chanoine et al., 2014). Resistance to FQs, ESCs, and carbapenems is of special concern due these drugs’ critical role in humans for treating life-threatening infections such as urosepsis (Matsumura et al., 2017).

Emerging FQ resistance is due mostly to the expansion and spread of sequence type (ST) ST131, a clonal lineage from E. coli phylogenetic group B2, first identified in 2008 on three continents (Coque et al., 2008; Lau et al., 2008; Nicolas-Chanoine et al., 2008). Genomic epidemiological analysis using Bayesian estimation predicts that FQ-resistance-conferring point mutations in chromosomal genes gyrA and parC occurred around 1987 in North America, coinciding with the first clinical use of FQs in the USA in 1986 (Ben Zakour et al., 2016; Stoesser et al., 2016). Most FQ-resistant (FQ^R) ST131 strains derive from a single sub-ST clade, designated clade C (Petty et al., 2014), or H30R (Price et al., 2013) Clade C/H30R in turn exhibits a prominent sub-lineage, referred to as clade C2 (Petty et al., 2014) or H30Rx (Johnson et al., 2016), members of which typically exhibit ESC resistance, mediated by the CTX-M-15 extended-spectrum β-lactamase (ESBL). ST131 (and, specifically, clade C/H30R) has become the most prevalent CIA-resistant human ExPEC lineage globally (Dautzenberg et al., 2016), causing millions of antimicrobial-resistant (AMR) infections annually (Pitout and DeVinney, 2017).

Recently, ST1193, another lineage of FQ^R group B2 E. coli, has also emerged as an important multidrug-resistant (MDR) human pathogen (Johnson et al., 2019; Tchesnokova et al., 2019). ST1193 was identified first in Australia as an emerging FQ^R clonal group (Platell et al., 2012), and has now been reported worldwide, including in Asia (Kim et al., 2017; Wu et al., 2017; Xia et al., 2017), North America (Tchesnokova et al., 2019), and Europe (Jorgensen et al., 2017; Valenza et al., 2019). In a recent multi-center study (United States, 2016–2017), ST1193 represented one-quarter of all FQ^R human clinical urine isolates (Tchesnokova et al., 2019).

The cause of such widespread dissemination of AMR ExPEC lineages is unclear. Host-to-host transmission, including between animals and humans, likely contributes, in view of documented colonization of multiple members of the same household with the same ExPEC strain (Johnson et al., 2016) and the occurrence of (human-associated) pandemic lineages such as ST131 and ST1193 in dogs and cats (Johnson et al., 2009c; Pomba et al., 2009; Platell et al., 2012). The first reported case of an E. coli ST131 infection in an animal involved a dog with a UTI in Portugal. Within-household ST131 transmission between humans and pets (particularly dogs and cats) has also been documented. Additionally, in the first description of FQ^R ST1193 as a potential pandemic clone, two isolates were from dogs. As such, companion animals may be an important reservoir of these resistant clonal lineages for acquisition by humans, and humans may be a reservoir of pathogens for their pets (Timofte et al., 2014). Notably, however, in the few relevant clinical and/or ecological studies (apart from the initial report involving ST1193), both ST131 and ST1193 have been relatively rare among companion animal ExPEC isolates (Platell et al., 2012; Pitout and DeVinney, 2017), consistent with the human reservoir being primary. Additionally, both clonal lineages have been reported recently in fecal samples from wild scavenging birds (Mukerji et al., 2019; Zurfluh et al., 2019), which suggests another potential zoonotic reservoir and/or transmission pathway.

Relevant to the human-companion animal interface, previous studies have documented the emergence of FQ^R ST131 in Australian companion animals (Platell et al., 2010, 2011a; Guo et al., 2015), as both an extraintestinal pathogen and a gastrointestinal colonizer of hospitalized dogs. Whereas Platell et al. and Guo et al. investigated companion animal-source FQ^R E. coli from only one region in Australia, Saputra et al. (2017) later surveyed companion animal-source E. coli on an Australia-wide level; they found that 9.3% of 514 dog isolates and 5% of 341 cat isolates were FQ^R. Based on these findings, the present study aimed to determine the prevalence of ST131 and ST1193 among both FQ^R (both ST131 and ST1193) and FQ-susceptible (FQ^S) (ST131 only) members of this Australia-wide E. coli isolate collection. An additional aim was to compare any identified companion animal ST131 isolates with members of a multi-national collection of human and animal ST131 isolates using whole-genome sequencing (WGS).

This study, which sought to characterize genetically a nationwide collection of clinical FQ^R E. coli isolates from extraintestinal infections in Australian dogs and cats (January 2013–January 2014) – including a WGS-based phylogenetic comparison across host species and geographical regions – led to four main conclusions. First, most (58%) FQ^R dog and cat isolates represented commensal phylogenetic groups A (24%), B1 (31%), and C (3%), rather than the group B2-derived pandemic lineages (ST131 and ST1193) that predominate among human FQ^R isolates. Second, the prevalence of ST131 as a cause of FQ^R E. coli infections in Australian companion animals was similarly low here among comparatively recent isolates (January 2013–January 2014) (6/59, 10%) as that noted previously by Platell et al. among historical isolates (2007–2009) (9/125, 7.2%), suggesting stability over time. Third, in a SNP-based phylogeny, the ST131 isolates from companion animals and humans were intermingled throughout the tree, rather than forming host-specific clades. This, together with the comparatively low prevalence of ST131 among isolates from companion animals, supports the hypothesis that companion animals in Australia are spillover hosts for ST131. Fourth, the absence of ST1193 among the present companion animal-source FQ^R E. coli isolates and its rarity in other companion animal-source isolate collections suggests that the same spillover hypothesis likely applies also to this clonal group, which is emerging among humans but not in animals.

Regarding our first main conclusion, according to most studies ExPEC derive mainly from phylogroups B2, D, and F, and non-pathogenic commensal E. coli from phylogroups A and B1 (Johnson and Stell, 2000; Vangchhia et al., 2016). Today, resistance to FQs and ESCs among human-source clinical E. coli is due mainly to the group B2-derived lineages ST131 and ST1193, most members of which qualify as ExPEC (Johnson et al., 2019; Tchesnokova et al., 2019). By contrast, prior to these two clonal groups’ recent emergence and expansion, most FQ^R human-source clinical isolates belonged to groups A and B1, and typically contained fewer VGs than did FQ^S isolates (Clermont et al., 2000).

The present study confirms that this historic pattern of phylogroup distribution persists among recent dog and cat-source FQ^R clinical E. coli isolates from Australia. Whilst the overall prevalence of FQ resistance in the Australia-wide collection of companion animal E. coli is low (< 10%) by international standards (Saputra et al., 2017), the FQ^R isolates belonged predominantly to groups A and B1 and possessed fewer VGs than group B2 and D isolates, with only seven of 53 (13%) FQ^R non-ST131 isolates qualifying molecularly as ExPEC. These findings echo a comparable study conducted over 10 years ago on E. coli isolates from North American dogs with UTI, in which FQ^R differed significantly in VG content (fewer) and phylogenetic background (mostly non-group B2) from FQS isolates (more VGs, mostly B2) (Johnson et al., 2009a).

Regarding our second main conclusion, comparison of the present findings with those of the regionally based study of Platell et al. (2010) confirmed that the prevalence of ST131 as a cause of FQ^R clinical infections in Australian companion animals did not increase significantly over the 5 year between-study interval. Two recent studies documented ST131 as a cause of extraintestinal infections in companion animals in Europe, but identified different proportions of C1/H30R1 vs. C2/H30Rx among cat and dog-source isolates (Belas et al., 2018; Melo et al., 2019). However, neither included a detailed genomic comparison with human ST131 isolates. Specifically, among urine E. coli isolates from cats and dogs in Portugal (1999–2015), 15% (25/172) of group B2 isolates were ST131, of which seven (28%) represented C/H30. Three (43%) of the C/H30 isolates represented the C2/H30Rx subclade, and two of these possessed bla_CTX–M–15 (Belas et al., 2018). Likewise, among more recent ESC-resistant E. coli isolates from cats and dogs in France (2010–16), most ST131 isolates represented C/H30 (50/56, 89%), 32 (64%) of which represented the C2/H30Rx subclade, and several possessed bla_CTX–M–27 (Melo et al., 2019). By contrast, 36% (4/11) of the present cat and dog ST131 isolates (one of which was identified as carrying the bla_CTX–M–27 gene), and only 11% (1/9) of those studied by Platell et al. (2010), represented the C2/H30Rx subclade. Taken together, these four studies suggest that the C2/H30Rx clade is becoming more common as a cause of extraintestinal infection in companion animals. A larger multicenter study involving genomic comparison of more recent companion animal ST131 isolates is needed to confirm this hypothesis.

Regarding human-companion animal ST131 clonal commonality, our comparison of 20 Australian cat and dog-source ST131 genomes with 173 human-source ST131 genomes (Ben Zakour et al., 2016) confirmed that the animal-source isolates were widely distributed throughout the ST131 phylogeny, in many cases placed closely to human ST131 strains. The absence of a separate clade containing only companion animal ST131 strains suggests a spillover model for ST131 occurrence in companion animals, as hypothesized previously (Bourne et al., 2019). Indeed, some animal ST131 strains shared very high identity with human strains. For example, MS10899, an Australian cat urine isolate (2007–2009), differed from P189EC, a Spanish human fecal isolate (2011), by only 66 non-recombinant core-genome SNPs (Figure 2). Although we cannot exclude that an individual from Spain came into contact with a cat in Australia in 2005, a more plausible explanation is the existence of an as-yet undefined reservoir or transmission network of circulating ST131 strains that links Australia and Spain (and, likely, much of the rest of the world).

Regarding our third main conclusion, human-animal clonal commonality implies interspecies transmission potential (of indeterminate direction), whereas the comparative prevalence of ST131 among human vs. animal clinical isolates implicates humans rather than companion animals as the main reservoir. Combined with the fact that the relative prevalence of ST131 among clinical isolates from dogs in Australia was fairly stable between 2007 and 2009 (Platell et al., 2010) and 2013 (this study), this adds further support for the hypothesis that ST131 causes infections predominantly in humans, with cats and dogs representing occasional spillover hosts (Melo et al., 2019). It is plausible, however, that pets may carry significant numbers of ST131 strains in the gut and could facilitate interspecies transfer to humans, who may simply be more susceptible than pets to developing extraintestinal infection due to ST131. By contrast, pets may be predisposed to developing infection due to other E. coli STs such as ST372, which is comparatively rarer among humans (LeCuyer et al., 2018; Kidsley et al., 2020; Valat et al., 2020).

Despite the overall genetic diversity of the Australian cat and dog-source ST131 isolates, some sub-clustering was apparent within the reference strain ST131 phylogeny. Based on the geographical, temporal, and host species metadata and minimal SNP diversity, it is likely that a single epidemiological source (per cluster) gave rise to each of the three clusters of closely related companion animal isolates within ST131 clades B, C1/H30R1, and C2/H30Rx. Given the distribution of the 34 animal ST131 isolates within the SNP-based phylogeny and the spillover hypothesis, further work will be required to characterize the reservoirs and transmission pathways that facilitate the spread of ST131 beyond the human population (Platell et al., 2011b).

Regarding our fourth main conclusion, ST1193 – another emerging clonal lineage of FQ^R E. coli within phylogroup B2 (Johnson et al., 2019; Tchesnokova et al., 2019; Valenza et al., 2019) — accounted for none of the 53 present FQ^R cat and dog isolates, despite having accounted for two FQ^R dog clinical isolates from Australia in the earlier Platell et al. study (Platell et al., 2012). To our knowledge only two other studies have recovered ST1193 from companion animals. Maeyama et al. (2018) identified five ESC resistant ST1193 isolates among 381 clinical E. coli isolates from companion animals in Japan in 2016 (Maeyama et al., 2018); while Zhang et al. (2018) identified one (n = 24) CTX-M-positive ST1193 of clinical origin in Canada (Zhang et al., 2018). ST1193’s rarity of isolation from companion animals suggests that ST1193, like ST131, is likely is predominantly a human-source clone, for which dogs are mainly spillover hosts. Regarding possible alternate transmission pathways for ST1193, in a recent Australian study a high proportion of silver gulls carried ESC-resistant and FQ^R E. coli in their feces. Whilst the most commonly identified ST was ST131 (17%), a substantial proportion of isolates represented ST1193 (6%), and these were closely related to human isolates (Mukerji et al., 2019).

This study has relevant limitations. First, selection bias may have influenced the observed distribution of FQ^R STs, since the study isolates came from veterinary diagnostic laboratories, which preferentially receive specimens from complicated cases and hosts with prior antimicrobial therapy (unpublished data). Second, detailed clinical data were lacking. Third, the clinical impact of the observed in vitro and genotypically predicted antimicrobial resistance is unknown. The study also has strengths: these include the large sample size (n = 855, making it the largest such study to date), the “snapshot” 1-year study period, the broad geographical sampling across Australia, and the genomic comparison with a large multi-national collection of (predominantly human-source) reference ST131 genomes.

In conclusion, we found that despite ST131 being the most prevalent ExPEC strain reported globally among humans, it occurred only at comparatively low frequency among FQ^R and FQ^S E. coli clinical isolates from Australian cats and dogs over a 1-year period in 2013–2014 (10%), with little change since 2007–2009 (7.2%). Phylogenomic comparisons of ST131 isolates from humans, cats, and dogs in Australia showed that despite some sub-clustering, the animal isolates were widely distributed throughout the ST131 phylogeny, often closely resembling human isolates. By contrast with ST131, ST1193, although identified first as an emerging clonal group among Australian FQ^R human and animal-source E. coli from 2007–2009 (Platell et al., 2012), was absent from the present FQ^R dog and cat isolate collection.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: Illumina sequence data for all 20 isolates have been deposited to the SRA under the accession numbers SRR11608154–SRR11608173 (BioProject PRJNA627752).

AK performed the experiments and data analysis, and drafted and prepared the manuscript. RW performed the whole genome sequencing and built and analyzed the phylogenetic tree. SS performed the initial antimicrobial susceptibility testing. JM provided the additional ST131 isolates. SB, MS, MO’D, DG, JJ, SA, and DT were involved in the experimental design development and manuscript preparation. All authors contributed to the article and approved the submitted version.

DT has received research funding and undertaken consultancies for Bayer, Zoetis, Boehringer Ingelheim, Virbac, Luoda Pharma, Neoculi, and IRiccorgpharm. SA has received research funding from Zoetis and Neoculi. JJ has received research support from and/or has undertaken consultancies for Achaogen, Allergan, Crucell/Janssen, Melinta, Merck, Shionogi, Syntiron, and Tetraphase. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.