Introduction

Escherichia coli is a commensal and opportunistic bacterium capable of causing extraintestinal infections with diverse disease patterns in humans, including urinary tract infections (UTIs), neonatal meningitis, and bloodstream infections (BSIs)1,2. Uropathogenic E. coli (UPEC), a subcategory of the extraintestinal pathogenic E. coli (ExPEC) causing UTIs in humans and companion animals, is well-characterised in terms of pathogenicity and antimicrobial resistance (AMR) properties. Uropathogenesis is initiated by compromised urinary function, facilitating ascending translocation to the uroepithelium for colonisation by adhesins such as type 1 fimbriae and P fimbriae. However, these adhesins can also be found in highly pathogenic strains as part of the ExPEC pathotype. Virulence genes (VGs) involved in adhesion (papA, papC, sfa/focDE, and afa/draBC), cytotoxins (hlyA, cnf, and vat), iron acquisition (fyuA and iutA), and capsule production (kpsMII) are abundant in ExPEC from various extraintestinal infections in humans, such as neonatal meningitis and UTIs3,4,5. Presence of the VGs, such as yfcV (yfcV fimbrial protein), vat (vacuolating toxin), fyuA (yersiniabactin receptor), and iut (aerobactin), contributes to effective bladder colonisation in mouse model by highly virulent E. coli strains, presumably UPEC6. The presence of these genes can be utilised as a predictive tool for UPEC classification, with the limitation that it cannot differentiate UPEC and neonatal meningitis E. coli (NMEC)6. Determination of ExPEC and/or UPEC would be beneficial in identifying highly pathogenic strains, which typically carry numerous VGs and persist opportunistically in the hosts, leading to recurrent infections.

Apart from causing UTsI in dogs and cats, E. coli is a common pathogen responsible for life-threatening or challenging-to-treat infections like pyometra, prostatitis, and BSIs in small animal medicine. These diseases are predisposed by hormonal factors that affect tissue proliferation and compromised tissue barriers, rendering them susceptible to bacterial infections.

Phylogroup B2 is one of the most prevalent lineages of endometrial pathogenic E. coli (EnPEC) associated with canine pyometra and endometritis in animals, and it carries VGs similar to those found in UPEC strains7,8. However, little is known about virulence factors and genotypes of E. coli associated with prostatitis and BSIs in pets.

In veterinary practice, fluoroquinolones (FQs) and third-generation cephalosporins (3GCs) serve as broad-spectrum antimicrobials used for the treatment of acute and systemic bacterial infections, including pyometra and BSIs9. FQs, with their capability to permeate the blood-prostate barrier, are the preferred antimicrobial drugs for treating canine bacterial prostatitis10. However, the effectiveness of antimicrobial treatment is reduced by infections caused by multidrug-resistant (MDR) bacteria, which encompass extended-spectrum β-lactamase (ESBL)- and AmpC β-lactamase-producing E. coli. MDR bacteria develop resistance mechanisms through chromosomal mutations and accumulation of antimicrobial resistance genes (ARGs) located on mobile genetic elements (MGEs) such as plasmids, via horizontal gene transfer11. Of paramount concern to public health is the global dissemination of E. coli sequence type (ST) 131 carrying blaCTX-15, which is considered a high-risk clone causing extraintestinal infection in humans11. The genome of a canine FQ-resistant EnPEC strain belonging to ST131 was sequenced, revealing a multitude of ARGs and carriage of ExPEC VGs8. Recently, high-risk E. coli clones have been categorised based on their association with AMR determinants and widespread distribution, including ST38, ST167, ST410, ST641, ST648, ST617, and ST1193. Instances of spillover of these high-risk E. coli clones causing canine and feline UTIs have been reported12. However, the clonal relationships of MDR ExPEC strains associated with reproductive tract infections and BSIs in dogs and cats remain uncertain. Therefore, this study illustrates the virulence genotypes, AMR properties, and genetic background of E. coli isolated from pyometra, prostatic abscesses (PAs), and BSIs in dogs and cats and elucidates the clonal relationships of the MDR non-urinary ExPEC strains with previously characterised high-risk E. coli clones isolated from UTIs12 using whole-genome sequencing (WGS) and in silico analyses.

Results

Phylogroups of E. coli isolated from pyometra, prostatic abscesses, and bloodstream infections

Out of 104 E. coli isolates, 101 isolates were recovered from dogs suffering from canine pyometra (61 isolates), PAs (38 isolates), and BSIs (2 isolates), and the other three isolates were recovered from cats with BSIs.

One-hundred two isolates were identified as phylogroup B2 (65/102), B1 (16/102), A (10/102), F (5/102), D (2/102), C (1/102), E (1/102), and cryptic clade I/II (2/102), while two isolates were unidentifiable phylogroup. Table 1 presents proportions of E. coli phylogroups isolated from pyometra, PAs, and BSIs. Phylogroup B2 was predominant in all disease patterns. The proportions of phylogroup B2 isolated from pyometra and BSIs were significantly higher than in PAs, but phylogroup A isolated from PAs was significantly higher than the others. Only E. coli isolates from pyometra exhibited variations across all phylogroups.

Table 1 Virulence genes associated with extraintestinal infections in Escherichia coli classified into phylogroups and sources.
Full size table

Virulence gene carriage

A total of 104 E. coli isolates recovered from pyometra, PAs, and BSIs were screened for VG carriage. The isolates that carried ≥ 2 out of 5 VGs (papC, sfa/foc, afa, iutA, and kpsII) were categorised as ExPEC, whereas those carried ≥ 3 out of 4 VGs (yfcV, vat, fyuA, and chuA) were classified as UPEC4,6. Table 2 presents the proportions of E. coli isolated from pyometra, PAs, and BSIs, classified into UPEC and ExPEC pathotypes. A significantly higher proportion of pyometra strains was categorised as UPEC compared to PA. However, proportions of E. coli classified as ExPEC were not significantly different. A range of 3–22 VGs were detected in the E. coli isolates. VGs associated with UPEC (yfcV, fyuA, chuA), as well as irp1, hlyE, and kpsII, were significantly more prevalent in isolates from pyometra than in PAs. ExPEC VGs, including papC, iha, afa, hlyA, and cnf, were common in isolates from pyometra and PAs but were not detected in BSI isolates. VGs were associated with the phylogroups in which high numbers of VGs, including sfa, papC, yfcV, tsh, hlyA, hlyE, vat, fyuA, irp1, chuA, and kpsII, were significantly presented in phylogroup B2.

Table 2 Escherichia coli (n = 104) isolated from canine (n = 101) and feline (n = 3) extraintestinal infections classified into extraintestinal pathogenic E. coli (ExPEC) and uropathogenic E. coli (UPEC) genotypes.
Full size table

Antimicrobial resistance and genetic characteristics of 3GC- and/or FQ-resistant E. coli

Out of 104 E. coli isolates, 44 exhibited FQ resistance (FQR) and 3GC resistance (3GCR) (13/44), 3GCR only (5/44), and FQR only (26/44). These were classified into ESBL (15/44) and AmpC (7/44) phenotypes, as shown in Supplementary Table S1. Regarding ESBL and AmpC production, 15 isolates carried blaCTX-M, including blaCTX-M-15 (6/15), blaCTX-M-27 (4/15), blaCTX-M-55 (3/15), and blaCTX-M-14 (2/15). Additionally, seven AmpC-producing isolates carried CIT (5/7), FOX (2/7), and ACC (1/7). The proportions of FQR and/or 3GCR were high in BSI (5/5) and PA (20/38) isolates, but only 19 out of 61 isolates from pyometra exhibited FQR and/or 3GCR phenotypes. Overall, 39 out of 44 isolates were MDR, expressing resistance to one drug in three or more antimicrobial groups and carrying multiple ARGs. Frequencies of AMR and ARG carriage, as well as AMR patterns, are provided in Supplementary Figs. S1 and S2, respectively.

Phylogroups of the 44 E. coli isolates were identified as follows: B2 (20/44), B1 (10/44), A (9/44), E (1/44), F (3/44), and an unidentified phylogroup (1/44). DNA fingerprint analysis by repetitive extragenic palindromic PCR (REP-PCR) revealed three major clades associated with the phylogroups: clade I-B1 (8/44), clade III-A (6/44), and clade VI-B2 (10/44). The DNA fingerprint pattern of the other isolates was distinct in minor clades (Supplementary Fig. S2). An identical DNA fingerprint pattern was observed for two E. coli phylogroup B2 strains, one from pyometra (CUVET21-PYO103) and another from BSI (CUVET21-H2), in the same dog, which was presented in clade VIII.

Genome characteristics of E. coli isolates subjected for WGS analysis

The genome of 12 E. coli strains, isolated from pyometra (4 isolates), PAs (5 isolates), and BSIs (3 isolates), as indicated in Supplementary Fig. S2, were sequenced and compared to the genome of seven E. coli strains from UTIs. Chromosome sizes ranged from 4,836,704 to 5,583,986 bp and had a G + C content from 50.42 to 50.86%. Chromosome size, the number and size of plasmids, genetic features, as well as the locations of ARGs, for the 19 E. coli strains were provided in Supplementary Table S2. All strains contained at least one plasmid, except for strain CUVET21-1783 in ST648 from BSI. Sixteen strains carried MDR plasmids, which contained multiple ARGs.

Clonal relationship of ExPEC strains from different disease patterns

Three pyometra strains and two BSI strains shared common genetic background of high-risk E. coli, including ST131-B2 (ST-phylogroup), ST1193-B2, and ST648-F, with intra-clonal variation. The core genome multilocus sequence typing (cgMLST) analysis revealed only 18 and 39 different loci between the ST1193-B2 strains from BSI and UTI, respectively, compared with the pyometra strain. ST648-F strains from pyometra and UTI were more closely related, as evidenced by a difference of only 182–189 loci, while these strains were distant from the ST648-F strain from BSI with more than 975 loci difference. Up to 686 different loci were observed between ST131-B2 strains from pyometra and UTI. Figure 1 illustrates the minimum spanning tree based on cgMLST of the 19 E. coli isolates, representing clonal relationships among and within the high-risk clones.

Figure 1
figure 1

Minimum spanning tree based on core genome multilocus sequence typing (cgMLST) of 19 Escherichia coli strains. The isolates from pyometra (4 isolates, pink circles), prostatic abscesses (5 isolates, blue circles), bloodstream infections (3 isolates, red circles) and urinary tract infections (7 isolates, yellow circles) in dogs and cats are indicated in individual circles, presenting E. coli strains, sequence types (ST), and O:H antigens. Black lines and numbers indicate genetic relatedness and allele differences, respectively.

Full size image

The single-nucleotide polymorphisms (SNP) -based phylogenetic tree, based on 2,438 core genes, clustered the E. coli strains into four clades (Fig. 2). Phylogroup B2 strains were specific to the closely related clade 1 and 2, while phylogroup F was specific to clade 3. Phylogroup A, B1 and D were separately grouped in clade 4. In addition to 2438 core genes, accessory genes, comprising 8950 cloud genes and 5000 shell genes, were found among this E. coli population. Based on the presence and absence of the accessory genes, three clusters specifically grouped the phylogroup B2, F and a group of phylogroup A, B1 and D. Additionally, a strong correlation was observed between the accessory genome and the coreSNP analysis (see Supplementary Fig. S3 online).

Figure 2
figure 2

Phylogenetic tree based on core genome single-nucleotide polymorphisms (coreSNPs) of 19 Escherichia coli strains. Colour strips indicate sources as follows: pink, pyometra (4 isolates); blue, prostatic abscesses (5 isolates); red, bloodstream infections (3 isolates); and yellow, urinary tract infections (7 isolates) in dogs and cats. Squares and triangles indicate the presence of plasmids and antimicrobial resistance genes (ARGs), respectively. Circles indicate the presence of at least one of the genes responsible for each virulence factor. Black circles and circles enclosed with red borders indicate presence of the genes on chromosome and pathogenicity islands, respectively. The same colour indicates co-localisation on the same plasmid. Genes presenting on both chromosome and plasmid were indicated by the colours and black in the same icon. Icons underlined with blue and red indicate the presence of two and three copies of the gene, respectively. ARGs comprise β-lactamases encoding genes, blaTEM-1, blaOXA-1; extended-spectrum β-lactamase encoding gene, blaCTX-M-14, blaCTX-M-15, blaCTX-M-27, and blaCTX-M-55; AmpC β-lactamase encoding genes, blaCMY-2 and blaCMY-148; plasmid-mediated quinolone resistance (PMQR) genes, aac(6′)-Ib-cr, qnrS1, and qnrB6; aminoglycoside-modifying enzyme-encoding genes, aac(3)-IIa, aac(3)-IId, aac(6′)-Iaf, aadA2, aadA5, aadA16, ant(3′′)-Ia, aph(3′)-Ia, aph(3′′)-Ib, aph(6)-Id, and rmtB; florfenicol/chloramphenicol efflux protein genes, floR and cmlA; chloramphenicol acetyltransferase gene, catA; alternative dihydropterate synthase genes, sul1, sul2, and sul3; alternative dihydrofolate reductase genes, dfrA12, dfrA14, dfr17, and dfrA27; tetracycline efflux protein genes, tet(A), tet(B), and tet(M); macrolide phosphotransferase gene, mph(A); macrolide-resistance ribosomal RNA methyltransferase gene, erm(B); lincosamide nucleotidyltransferase gene, lnu(F); and rifampin ADP-ribosyltransferase gene, arr-3.

Full size image

Pathogenicity islands (PAIs) and plasmids

PAI(s) were identified on the chromosome of 16 E. coli isolates (Fig. 3), and they were absent in three isolates, including ST44-A and ST744-A from PAs, and ST648-F from BSI. The phylogroup B2 isolates contained up to five PAIs that carried several VGs (Fig. 3 and Supplementary Table S3). PAIs that co-harbored iss and ompT were specifically found in phylogroup A and B1. Common yersiniabactin-associated PAIs, classified as high PAIs (HPI), were present across multiple phylogroups, except for phylogroup A. However, a variant of the yersiniabactin-associated PAIs containing the type IV secretion system (T4SS) was detected in CUVET16-242 of phylogroup A from UTI. Moreover, a PAI containing sfa/foc genes was unusually inserted following pheU, also found in strain CUVET20-PYO5 (phylogroup B2), which carried the most numerous VGs. The aerobactin-encoding gene cluster was abundantly found on multi-replicon plasmids that carried ARGs, considered hybrid plasmids. The numbers and replicon types of resistance and virulence plasmids are illustrated in Fig. 2. Details of total plasmids and ARG and VG localisation are provided in Supplementary Table S2.

Figure 3
figure 3

Pathogenicity islands (PAIs) identified in 19 Escherichia coli strains. PAIs containing virulence genes inserted at the insertion sites of 19 Escherichia coli strains isolated from canine pyometra (4 isolates) and prostatic abscesses (5 isolates), and bloodstream infections (3 isolates) and urinary tract infections (7 isolates) in dogs and cats, classified into phylogroups and sequence types (ST). Red squares indicate the presence of virulence genes in each PAI.

Full size image

Virulome and resistome

Among the 19 sequenced E. coli isolates, 139 VGs encoding 27 virulence factors, and 41 ARGs mediating resistance to 10 antimicrobial classes were identified (Fig. 2 and Supplementary Fig. S4). Overall, high numbers of VGs were observed in phylogroup B2 (78–97 gene, with an average of 85.3 genes), followed by phylogroup F (63–95 genes, with an average of 76.4 genes). PAI-encoded ExPEC toxin and capsular genes, including cnf1, hlyA, and kps, were specifically present in phylogroup B2 and F isolates. In contrast, VG carriage was fewer in the strains in phylogroup A (35–60 genes, with an average of 51.2 genes) and phylogroup B1 (38–56 VGs with an average of 47 genes). Strain CUVET21-PYO5, belonging to ST998-B2, carried 97 VGs encoding 22 virulence factors but harbored only two ARGs, including blaCTX-M-14 and mph(A) on the chromosome. Strain CUVET19-1426, belonging to ST13037-A isolated from PA, had the lowest VG carriage, containing only 35 genes, but carried the highest number of ARGs on plasmids.

Regarding ARG carriage, phylogroup A isolates carried the highest number (11–17 genes with an average of 13.2 genes), while phylogroup B2 isolates had the lowest number (2–16 genes with an average 8.3 genes). Localisation of ARGs on the chromosome and plasmids in each isolate is indicated in Fig. 2. The majority of ARGs were located on plasmids in all plasmid-carrying isolates, except for ST12-B2. Larger than 120 kb hybrid virulence/MDR multi-replicon plasmids, carrying aerobactin genes, were more common in the phylogroup A and F isolates (Fig. 2 and Supplementary Table S2). Among the 14 blaCTX-M-positive isolates, six strains harbored blaCTX-M on the chromosome, four of which were phylogroup B2. The remaining eight blaCTX-M-positive isolates contained the gene on IncF plasmids, except for one strain carrying on an IncR/N plasmid. Only within ST648, chromosomal blaCMY-2 and plasmidic blaCMY-148 were detected. Among the 19 E. coli strains, mutation at gyrA:pS83L was detected in 18 strains, but 17 expressed FQR. Three mutations of quinolone resistance-determining regions (QRDR), including gyrA:pS83L, gyrA:pD87N, and parC:pS80I, were observed in 15 FQR strains. Of these, six strains harbored plasmid-mediated quinolone resistance (PMQR) genes, distributed on the chromosome and plasmids. Despite the absence of mutation of QRDR, one strain in phylogroup B2 contained two PMQR genes mediating FQR, including aac(6')-Ib-cr on chromosome and qnrB6 on a plasmid.

Discussion

In addition to UTIs, E. coli phylogroup B2 was associated with pyometra, PAs, and BSIs and contained the most abundant VGs. The majority of E. coli isolated from pyometra, PAs, and BSIs in this study belonged to phylogroup B2, as those isolated from extraintestinal infections in pets and humans in previous studies8,13,14,15,16. However, the characteristics of BSI isolates should be interpreted with consideration regarding the small number of isolates, which might lead to bias in statistical analysis. The virulence genotyping revealed that 72.1% and 80.3% of the E. coli isolated from pyometra contained VGs of ExPEC and UPEC pathotypes5,6, respectively, which were likely associated with the phylogroup rather than the disease patterns. Due to the lower numbers of phylogroup B2 in PA strains, ExPEC and UPEC genotypes were infrequently found in E. coli isolates from PAs. Non-B2 phylogroups isolated at lower frequencies in all diseases and contained less complete VGs required for the steps of ExPEC pathogenesis. In other extraintestinal infections such as UTIs, non-B2 E. coli phylogroups contained fewer VGs; however, genes encoding adhesins and siderophores were detected in most of the isolates4. The binding of type 1 fimbriae encoded by fimA to uroepithelium and uteroepithelium is considered a critical step in adhesion during uropathogenesis and uteropathogenesis, respectively17,18. The presence of specific ExPEC adhesin genes, such as sfa encoding S-fimbriae, can support successful extraintestinal colonisation. S-fimbriae specifically bind α-sialyl-2-3-β-galactose (NeuAc-α 2,3-Gal) on the animal cell surface and can enhance adhesion of UPEC17. The expression of fyuA and irp1 (yersiniabactin) in alkaline conditions, like urinary bladder and uterus, assists in iron acquisition for extraintestinal survival6,14,19,20. Likewise, hemolysin A and capsular genes are more specific to phylogroup B2 in UPEC and NMEC strains13. Therefore, the extraintestinal infections frequently caused by E. coli phylogroup B2 could be enhanced by all VGs corresponding to adhesion, toxin production, iron acquisition, and immune evasion. However, adhesins and iron acquisition systems were likely required in the extraintestinal pathogenesis of any E. coli phylogroups.

According to PAI and virulome analyses, PAIs served as the key genetic determinants for the virulence of the ExPEC isolates. High numbers of VGs, such as capsular, hemolysin A, and pyelonephritis-associated pili genes, were consistently detected on the PAIs found in phylogroup B2 across various disease patterns. Essentially, HPI containing yersiniabactin gene cluster, were non-specifically widespread in all phylogroups, supporting the fundamental requirement of an iron-acquiring system to survive outside the intestine21. Evidence supports the notion that pathogenic attenuation of E. coli strains containing the HPI can be caused by irp2 inactivation22. In phylogroup B2, the pheV-PAI containing iron-regulated gene homologue adhesin (iha), secreted autotransporter toxin (sat), aerobactin (iuc and iut), and capsular (kps) genes, were conserved in isolates from any infections. Additionally, the pheU-PAI containing the pap gene cluster, hemolysin A (hlyA), cytotoxin necrotising factor (cnf), and contact-dependent growth inhibitor (cdi), were more specific to ST131-B2. These PAIs bearing VGs associated with uropathogenesis are detected in phylogroup F and D from urinary sources23, but not in these phylogroups isolated from PAs, and BSIs in our study. Sharing of ExPEC VGs among the E. coli clones could be described by the presence of S fimbriae-associated PAI (pheU-PAI) in ST998-B2 and ST617-A. Additionally, the PAI-encoded aerobactin genes (iutA and iuc gene cluster) in B2 strains were carried on plasmids in non-B2 strains. The acquisition of MGEs encoding ExPEC VGs as part of the accessory genome, such as PAI-encoded sfa/foc and plasmid-encoded iut/iuc, could be considered an evolutionary process leading to ExPEC, as assessed by the ExPEC criteria6.

The genomic background of MDR ExPEC from pyometra and BSIs using multilocus sequence typing (MLST) analysis showed a clonal relationship to high-risk ExPEC strains causing UTIs, including ST131-B2, ST1193-B2, and ST648-F, which are important for global monitoring, but PA isolates appeared to be distinct. Canine uterosepsis caused by E. coli ST1193-B2 was presented in a dog concurrently suffering from pyometra and BSI. Due to the identical genetic characteristics of strains from the uterus and blood, only the strain isolated from blood (CUVET21-H2) was subjected to WGS, revealing significant VG carriage such as the capsule production gene kpsII for immune evasion. E. coli ST1193 contains VGs similar to those found in ST131 but commonly carries the senB gene on IncF plasmids and is associated with infections in humans in the community rather than nosocomial infections24.

Intra-clonal variation was observed in the high-risk STs by coreSNP analysis, cgMLST, variations in O and H antigens, as well as differences in the type and location of accessory genes associated with virulence and resistance. E. coli ST131-B2 strains from both pyometra and UTI exhibited the greatest genetic diversity within the same ST. The chromosomally encoded blaCTX-M-15 E. coli ST131 strain associated with UTI belonged to clade C2/H30Rx, which is related to the predominant bacteremic E. coli ST131 SEA-C2 clone in Southeast Asia and disseminated worldwide25. Furthermore, heterogeneity was observed among ST1193 strains in terms of 3GCR development, illustrating acquisition of blaCTX-M, which was found in one strain from UTI. The emergence of MDR E. coli ST1193 has been increasingly reported, representing a successful global high-risk clone following the footsteps of E. coli ST13126. In Thailand, ST131, ST648, and ST1193 are the most common STs associated with high-risk ESBL-producing E. coli carried by hospitalised patients27. High-risk MDR E. coli clones dominate the population of 3GCR and FQR E. coli. Selecting 3GCR and FQR isolates in AMR monitoring of ExPEC in animals could support the findings of high-risk clones of global concern.

Although a low number of 3GCR was observed, the carriage of blaCTX-M was found to be common in 3GCR ExPEC associated with high-risk E. coli clones from UTIs, and pyometra, as well as in non-high-risk clones from PAs. The detected blaCTX-M variants in this study, including blaCTX-M-15, blaCTX-M-14, blaCTX-M-27, and blaCTX-M-55, are the most common variants in E. coli across humans, livestock, and the environment28, suggesting the wide dissemination within the E. coli population. The predominant blaCTX-M variants were the same as those found in canine and feline UPEC in Thailand12. The presence of blaCTX-M genes on plasmids is more common; however, chromosomal integration of blaCTX-M and AmpC, mediated by ISEcp1, supported the stabilisation of the gene in the genome. This could facilitate clonal dissemination, as found in ST13129,30. Mutations in the QRDR region in gyrA and/or parC are the major mechanism of FQR in bacteria and a primary factor for the successful clonal spread of ExPEC ST131 and ST1193, contributing to their emergence worldwide26. PMQR genes are an alternative mechanism in enterobacteria, developed through the acquisition of MGEs. In this study, most of the ARGs were carried by multi-replicon IncF plasmids, which plays a crucial role in bacterial evolution and primarily influences resistance evolution in these ExPEC strains31. The multiple recombination of the IncF plasmids, generating large multi-replicon plasmids containing numerous ARGs, VGs, and transfer modules, results in greater stability within bacterial hosts and more efficient transfer32,33. The evolution from IncF plasmid carriage and plasmid recombination might potentially promote E. coli to persist outside the gastrointestinal tract through iron acquisition in infections and enhance resistance to antimicrobials during the treatment.

Pan-genome analysis demonstrates the concurrent evolution of core genes by SNPs and accessory genes by gene acquisition in each lineage, indicating co-evolution in both core and accessory genomes. There is a high correlation between the core genome and accessory genome of E. coli ST117, reflecting diverse evolution in this clone by acquisition of PAIs and ARGs34. Overall, both core and accessory genome of the ExPEC strains are associated with evolution within the lineages and AMR. However, limitations of the study were addressed by the low numbers of WGS-sequenced isolates that were selected based on the highest numbers of virulence and resistance genes among the isolates from different disease patterns, as well as the far fewer isolates from BSIs.

This study highlighted the virulence traits of ExPEC causing canine pyometra and PAs, and BSIs in dogs and cats, most of them belong to phylogroup B2 containing numerous ExPEC VGs similar to those found in UPEC. However, non-B2 phylogroups may at least require adhesion and an iron-acquiring system. The 3GCR and FQR E. coli containing high numbers of ARGs and VGs from pyometra and BSIs are related to three high-risk E. coli lineages: B2-ST131, B2-ST1193, and F-ST648. Genetic features of E. coli from PAs are distinct from those of other extraintestinal infections. Virulence in the lineages is presented by PAIs in the chromosome, and IncF plasmids, especially multi-replicon plasmids, also play a crucial role in MDR evolution. Thus, pets act as a reservoir of high-risk E. coli, posing a risk of transmission to humans in the community. Effective diagnostic antimicrobial susceptibility testing to improve antimicrobial uses should be encouraged for the treatment of diseases caused by MDR E. coli. Animal neutering could reduce the risk of canine prostatitis and pyometra which may lead to life-threatening BSIs. Additionally, appropriate hygiene management is considered a preventive intervention to improve pet health.

Materials and methods

Ethics approval

This study was conducted according to the faculty regulations and approved by the Institutional Biosafety Committee of Faculty of Veterinary Science, Chulalongkorn University (CU-VET-IBC) (Protocol code IBC 2131030) on 13 December 2020. The authors confirm their adherence to the ethical policies outlined in the journal’s author guidelines. All uterine samples utilised for bacterial isolation in this study were obtained from canine patients who had undergone ovariohysterectomy for therapeutic purposes, under the care of licensed veterinarians in small animal hospitals, with the consent of the owners. The acquisition of samples from animal patients was conducted in compliance with the approval granted by the Institutional Animal Care and Use Committee of the Faculty of Veterinary Science, Chulalongkorn University (Protocol code 2131054) on January 26, 2021.

Bacterial isolates

A total of 104 E. coli from non-urinary extraintestinal infections, including canine pyometra (n = 61), canine PAs (n = 38), and canine and feline BSIs (n = 5) collected from 2016 to 2021, were included for E. coli phylogrouping, ExPEC VG detection, and screening of FQR and 3GCR. The 61 E. coli strains from pyometra were isolated from the endometrium of 48 out of 100 uteri of female dogs that underwent ovariohysterectomy for surgical treatment of canine pyometra at the Obstetrics, Gynecology, and Reproduction Unit, Small Animal Teaching Hospital, Chulalongkorn University, Thailand. From 12 dogs, two distinct E. coli isolates were recovered from the uterus, each presenting different colony characteristics. Additionally, two separate E. coli isolates were recovered: one from the endometrium and another from the blood of the same dog suffering from pyometra, which had progressed to sepsis. The 38 isolates from canine PAs were obtained from PA exudate collected by ultrasound-guided aspiration, and the 5 isolates (2 from dogs and 3 from cats) causing BSI were isolated from positive blood culture bottles using the BD BACTEC™ FX blood culture system (Becton-Dickenson, USA).

E. coli phylogrouping and ExPEC virulence gene detection

Phylogroups and ExPEC VG carriage of the 104 E. coli were identified using PCR-based methods. Genomic DNA was extracted using Nucleospin® Tissue DNA extraction kits (Machery-Nagel, Germany). Clermont E. coli phylo-typing, which consists of a quadruplex PCR panel and two simplex PCRs, was performed to differentiate E. coli phylogroup A, B1, B2, C, D, E, F and E. coli cryptic clades35. VGs associated with extraintestinal pathogenesis of E. coli, including adhesion (afa, crl, fimA, iha, papC, sfa/foc, yfcV, and tsh), cytotoxicity (hlyA, hlyE, cnf, sat, and vat), iron acquisition (iroN, fyuA, irp1, chuA, iutA, and iucD), and bacterial cell protection (bssS, hmsP, iss, and kpsII) were detected by PCR (see Supplementary Table S4 online). All PCRs were conducted in a 25 µL total volume containing 5X Firepol Master Mix (Solis Biodyne, Estonia), and the concentration of each primer was 0.2 µM.

Isolates that tested positive for ≥ 2 out of 5 VGs (papC, sfa/foc, afa, iutA, and kpsII) were presumably classified as ExPEC, whereas isolates demonstrating ≥ 3 out of 4 VGs (yfcV, vat, fyuA, and chuA) were presumably classified as UPEC4,6.

Screening of FQ- and/or 3GC-resistant E. coli and antimicrobial susceptibility testing

FQR and 3GCR were examined in all 104 E. coli isolates using ciprofloxacin, cefotaxime, and ceftazidime disk diffusion methods36. Isolates displaying resistance to cefotaxime and/or ceftazidime were included for detection of ESBL and AmpC β-lactamase production by the combination disk test and cefoxitin disk diffusion test, respectively. The AMR phenotypes of FQR and/or 3GCR E. coli were determined using broth microdilution with a customised Sensititre™ plate model COMPGN1F (Thermo Scientific, UK). The plate contained amikacin, amoxicillin/clavulanic acid, ampicillin, cefazolin, cefovecin, cefpodoxime, ceftazidime, cephalexin, chloramphenicol, doxycycline, enrofloxacin, gentamicin, imipenem, marbofloxacin, orbifloxacin, piperacillin/tazobactam, pradofloxacin, tetracycline, and sulfamethoxazole/trimethoprim. Resistance to the drugs was interpreted according to the minimum inhibitory concentration (MIC) breakpoints for veterinary isolates from the Clinical and Laboratory Standards Institute37.

Detection of antimicrobial resistance genes

Common ARGs in E. coli encompassing β-lactamases (blaOXA, blaSHV, and blaTEM), ESBL (blaCTX-M groups), cephalosporinases (ACC, CIT, DHA, EBC, FOX, and MOX groups), PMQR [aac(6′)-Ib-cr, qepA, qnrA, qnrB, qnrC, qnrD, and qnrS], aminoglycoside-modifying enzymes [aac(3)-IIa, aac(6′)-Ib, aac(6′)-Ib-cr, ant(2′′)-Ia, aph(3′′)-Ib, and aph(6)-Id], tetracycline efflux proteins [tet(A), tet(B), and tet(C)], alternative dihydropterate synthase (sul1 and sul2), alternative dihydrofolate reductase (dfrA1, dfrA5, dfrA7, dfrA12, dfrA17, and dfrB), florfenicol/chloramphenicol efflux protein (floR), chloramphenicol efflux protein (cmlA), and chloramphenicol acetyltransferase (catA) were detected in 44 E. coli exhibiting resistance to 3GCs and/or FQs by using PCR-based methods (see Supplementary Table S5 online).

Strain selection for whole genome sequencing

The clonal relatedness of 44 FQR and/or 3GCR E. coli from pyometra (19 isolates), PAs (20 isolates), and BSIs (5 isolates) was assessed by analysing REP-PCR DNA band patterns with more than 70% similarity38. Genetically related E. coli strains from different diseases and non-genetically related E. coli strains from the same disease, which contained the highest numbers of VGs and ARGs, were selected for WGS. A total of 12 E. coli strains, including 4 from pyometra, 5 from PAs, and 3 from BSIs, were chosen. The genomes of seven characterised 3GCR E. coli strains associated with global high-risk clones from canine and feline UTIs12, including ST38, ST131, ST617, ST641, ST648, and ST1193, were additionally sequenced for comparison. Genomic DNA extraction was performed using the G-spin™ Total DNA Extraction Mini Kit (intron Bio, South Korea) for short-read Illumina sequencing and the Nucleospin® Tissue DNA extraction kits (Machery-Nagel, Germany) for long-read Oxford Nanopore Technologies (ONT) sequencing. DNA libraries were prepared using the NEBNext® Ultra™ DNA Library Prep Kit (New England Biolabs, USA) and were subsequently loaded into an Illumina NextSeq system (Illumina, USA) to obtain 150 bp pair-end reads by a private service company (Celemic, South Korea). For long-read sequencing, the Native Barcoding Kit (SQK-NBD112.24) was utilised for multiplex library preparation from the genomic DNA, followed by loading into an R10.4 flowcell (FLO-MIN112) (ONT, UK). Trimmomatic v.0.39 was employed to trim short reads and remove adaptors to obtain high-quality reads39. Guppy v.6.3.8 was used for base-calling and de-multiplexing of long reads. Unicycler pipeline v.0.4.8 was employed for the assembly of short and long reads, resulting in generation of circular chromosomes and plasmids40.

Pan-genome, phylogenetic and bioinformatic analyses

The pan-genome of 19 sequenced strains was analysed to identify core genes and accessory genes using Roary version 3.11.241. The coreSNPs were extracted from core genome alignment using SNP sites for constructing a phylogenetic tree using RAxML version 8 with 1000 bootstraps42,43. The coreSNP tree was visualised using Interactive Tree of Life (iTOL) (https://itol.embl.de/itol.cgi). STs and complex types were determined based on Achtman’s MLST and cgMLST, respectively, by submitting short reads to EnteroBase (enterobase.warwick.ac.uk)44. A minimum spanning tree based on cgMLST of 2513 core genes was generated using GrapeTree (available at enterobase.warwick.ac.uk) to illustrate genomic relationships among the whole-genome-sequenced strains45.

Acquired ARGs and point mutations of QRDR associated with FQR were detected using ResFinder v.4.1 and the NCBI AMR Finder tool46,47. VGs were identified using the Virulence Factor Database (VFDB)48. MGEs, including plasmids and PAIs, were identified using PlasmidFinder and IslandViewer 4, respectively49,50. The complete genome of UPEC strain CFT073 was utilised as the reference genome for PAI analysis. FimHTyper 1.0 and SerotypeFinder 2.0 were employed to determine E. coli FimH types and serotypes, respectively51,52. The sequenced contigs were submitted to the NCBI Prokaryotic Genome Annotation Pipeline for ORF prediction and gene annotation.

Statistical analysis

Fisher’s exact test was used to ascertain the association between various phylogroups and/or disease patterns and the presence of the VGs, and ExPEC and/or UPEC genotypes and disease patterns. The investigation was assessed by using IBM® SPSS® Statistics version 26. Statistical significance was set at p < 0.05.