Materials and Methods

Animals—The study was performed on samples collected from 85 dogs and 19 cats. These companion animals had urogenital infections and were patients at the Veterinary Medical Teaching Hospital of Nippon Veterinary and Life Science University and several neighboring private veterinary clinics in Japan between 2003 and 2010.

Samples and bacterial isolates—A swab was used to collect a sample of urine or a sample from the genital tract of each dog and cat (1 sample/animal). A total of 104 E coli isolates were derived from dogs (85 strains [1/dog]) and cats (19 strains [1/cat]) with urogenital infection. Of the 85 E coli isolates derived from dogs, 66 were obtained from urine and 19 were obtained from genital tract samples; of the 19 E coli isolates derived from cats, 15 were obtained from urinary samples and 4 were obtained from genital tract samples. No information was available regarding any previous antimicrobial treatments for urinary and genital tract infections that the dogs and cats had at the time of the study.

Swab samples of urine or from the genital tract were plated on desoxycholate-hydrogen sulfide-lactose agar and incubated overnight at 37°C for isolation and identification of E coli. Subsequently, lactose-fermenting colonies were selected (1 colony/swab) and tested by use of standard protocols (Gram staining and indole, methyl red, Voges-Proskauer, and Simmons citrate [IM-ViC] tests) or a commercial kit^a for identification. All confirmed E coli isolates were stored in 10% skim milk at −80°C until analyzed.

Antimicrobial susceptibility testing—Susceptibility to each of 9 antimicrobials was determined by use of the agar dilution method according to the guidelines of the Clinical and Laboratory Standards Institute.²² The antimicrobials used were AMP, CFZ, CEF, DHS, KAN, OTC, CHL, ENR, and TMS. The breakpoints established by the Clinical and Laboratory Standards Institute²² were used for AMP, CFZ, KAN, CHL, ENR, and TMS; breakpoints for CEF, DHS, and OTC were each set as the midpoint between peaks when the minimum inhibitory concentrations were bimodally distributed. An American Type Culture Collection organism (E coli ATCC 25922) was used as a quality-control strain. The median (aggregate) score of AMR in each group was defined as follows: the ([n + 1]/2)th item (for an odd number of samples) or the arithmetic mean of the 2 middlemost terms (for an even number of samples) in the ordered list of number of AMRs in each group.

Virulence genotyping and phylogenetic analysis—The detection of VF genes, including the pyelonephritis-associated pili (pap), S fimbriae (sfa), afimbrial adhesion I (afa), hemolysin (hly), cytotoxic necrotizing factor I (cnf), and aerobactin (aer) genes, was performed by use of multiplex PCR assay as reported previously.²³ Additionally, all isolates were assigned to 1 of the 4 main phylogenetic groups of E coli (A, B1, B2, or D) by use of multiplex PCR assay, as described by Clermont et al.²⁴ All PCR testing was done in conjunction with relevant positive and negative controls. Dubious PCR results were clarified by repeating the PCR assays. The median (aggregate) score of VF in each group was defined as follows: the ([n + 1]/2)th item (for an odd number of samples) or the arithmetic mean of the 2 middlemost terms (for an even number of samples) in the ordered list of number of VF in each group.

Detection of clone O25b:H4-ST131 that produces CTX-M-15—The clone O25b:H4-ST131, which produces CTX-M-15, has resistance to penicillins and cephalosporins and belongs to phylogenetic group B2.²⁵ Thus, CFZ-resistant (minimum inhibitory concentrations ≥ 32 μg/mL) phylogroup B2 isolates were screened for this clone by use of triplex PCR assay, as described by Blanco et al.²⁵

RAPD analysis—All isolates were typed by RAPD analysis as described previously.²⁶ Briefly, all PCR assays consisted of a denaturation cycle of 94°C for 5 minutes, followed by 10 cycles of 94°C for 1 minute, 40°C for 1 minute, and 72°C for 1 minute; 22 cycles of 94°C for 1 minute, 55°C for 1 minute, and 72°C for 1 minute; and 1 cycle of 72°C for 4 minutes. Primer PB1 (5′-GC-GCTGGCTCAG-3′) was used in the study. An in-house apparatus was used to digitize DNA fingerprints, images were saves as TIFF files for subsequent analysis by use of commercial software.^b Cluster analysis was performed by the unweighted pair group method with arithmetic means.

Statistical analysis—The prevalences of AMR and the VF genes among isolates were compared by use of a Fisher exact test (for comparisons between 2 groups) or the Ryan method (for comparisons between 3 or more groups). Virulence factors and AMR data for canine and feline isolates were compared by use of the Mann-Whitney U test. Values of P < 0.05 were considered significant. Associations between AMR phenotypes and VF genes were considered significant at values of P < 0.05; for significant associations, odds ratios and 95% CIs were calculated.

Results

Distribution of AMR, VF genes, and phylogenetic groups—Resistance to 1 or more of the antimicrobials tested was detected in 65 of the 104 (62.5%) E coli isolates. Among the 9 antimicrobials tested, resistance to AMP was the most prevalent (55 [52.9%] isolates), followed by resistance to ENR (48 [46.2%]), OTC (43 [41.3%]), DHS (39 [37.5%]), CFZ (33 [31.7%]), and CEF (28 [26.9%]; Table 1). With regard to multidrug-resistance profiles, 45 of the 104 (43.3%) isolates were resistant to > 2 of the 9 antimicrobials tested, 36 (34.6%) were resistant to > 4 antimicrobials, and 19 (18.3%) were resistant to > 6 antimicrobials (Table 2). Among the 57 multidrug-resistant isolates, 51 (89.5%) and 46 (80.7%) had resistance to AMP and ENR, respectively.

The distribution of VF genes among the 104 E coli isolates was determined (Table 3). Of the 6 VF genes evaluated, the sfa gene was the most prevalent (detected in 57 of 104 [54.8%] isolates), followed by aer (detected in 54 [51.9%] isolates), cnf (detected in 53 [51.0%] isolates), pap (detected in 36 [34.6%] isolates), and hly (detected in 29 [27.9%] isolates; Table 1). Twenty-seven isolates had only 1 VF gene. The most common gene identified singly was aer (22 isolates); this was also the most common gene profile among the 104 isolates. Twenty combinations of the 6 VF genes were identified in the E coli isolates. Of these 20 gene combinations, the most common combination profiles were pap, sfa, hly, and cnf (16 isolates); sfa, aer, and cnf (11 isolates); sfa and aer (6 isolates); and pap, sfa, aer, and cnf (5 isolates).

Results of multiplex PCR-based phylotyping indicated that 64 of the 104 (61.5%) E coli isolates belonged to the phylogenetic group B2. Twenty-two (21.2%) isolates belonged to group D, 13 (12.5%) isolates belonged to group B1, and 5 (4.8%) isolates belonged to group A (Table 1). Comparisons of canine and feline isolates revealed no significant differences in prevalences of AMR or VFs genes, except that hly genes were detected in a greater number of isolates derived from cats. The AMR and VF scores for canine and feline isolates differed (median score, 2 vs 4 and 2 vs 3, respectively). In addition, no significant difference was observed in prevalence of phylogenetic origins of isolates.

Screening of clone O25b:H4-ST131 that produces CTX-M-15—Among the 64 group B2 isolates, 12 were resistant to CFZ. These isolates were screened by use of a triplex PCR assay for clone O25b:H4-ST131 producing CTX-M-15. The clone was not detected among the 12 isolates.

Prevalence of AMR and VF genes among phylogenetic groups—Group B2 isolates were significantly (P < 0.05) less resistant to 3 antimicrobials (AMP, OTC, and ENR), compared with group A isolates. Also, group B2 isolates were significantly (P < 0.05) less resistant to all tested antimicrobials, compared with group D isolates (Table 4). The aggregate AMR score was lower for group B2 isolates than scores for the other groups of isolates (median score, 0 vs 4, 5, or 6). With respect to VF genes, the hly gene and pap, hly, and cnf genes were significantly (P < 0.05) more prevalent in group B2 isolates than in group B1 and D isolates, respectively. The aer gene was significantly (P < 0.05) more prevalent in group D isolates than in group B2 isolates.

Association of AMR phenotypes with VF genes—Of the 6 VF genes tested, only the aer gene was directly associated with resistance of isolates to 8 of the 9 antimicrobials tested (AMP, CFZ, CEF, DHS, KAN, OTC, ENR, and TMS; Table 5). The pap, sfa, hly, and cnf genes were inversely associated with resistance to 4, 2, 5, and 2 antimicrobials, respectively. There was no significant association between the afa gene and resistance of isolates to any of the antimicrobials tested.

Assessment of clustering via RAPD analysis—Randomly amplified polymorphic DNA analysis revealed 3 major clusters: I, II, and III (Figure 1). Cluster I included 63 isolates and was composed mainly of group B2 isolates (61 [96.8%] isolates). Cluster I was further differentiated into 2 subclusters: Ia (46 isolates) and Ib (12 isolates). When the 2 subclusters were compared, there were significant (P < 0.05) differences in the AMR and the VF scores. Compared with subcluster Ib isolates, subcluster Ia isolates had a higher median VF score (3 [95% CI, 3 to 4] vs 1 [95% CI, 1 to 2]) and lower median AMR score (0 [95% CI, 0 to 0] vs 3 [95% CI, 2 to 7]). Cluster II included 19 isolates and was composed mainly of group B1 isolates (13 [68.4%] isolates). Cluster III included 22 isolates and was composed mainly of group D isolates (19 [86.4%] isolates).

View Full Size

Dendrogram based on RAPD profiles with detected VF genes and AMR status among 104 ExPEC isolates from dogs and cats. The horizontal scale at the top left indicates the percentage of similarity. The gray horizontal strips represent RAPD profiles.

Citation: American Journal of Veterinary Research 73, 3; 10.2460/ajvr.73.3.409

• Download Figure
• Download figure as PowerPoint slide

Discussion

To our knowledge, this is the first study to investigate the distributions of AMR phenotypes, VF genes, and phylogenetic origins in urogenital E coli isolates obtained from dogs and cats in Japan. We were also able to examine the associations between AMR and VF genes or phylogenetic origins. The study findings indicated an alarmingly high prevalence of resistance of the urogenital E coli isolates to many antimicrobials commonly used in Japan. Multidrug resistance was often detected in canine and feline isolates, which has important implications for antimicrobial treatment of companion animals infected with ExPEC. Approximately half of the isolates had resistance to β-lactams and fluoroquinolones, which was observed in most of the multidrug-resistant isolates. In contrast, researchers from other countries have reported lower frequencies of AMR. Among 26 hemolytic E coli isolates obtained from the urogenital tracts of Danish dogs, resistance to AMP was detected in 4 (15.4%) and resistance to CEF was detected in 1 (3.9%).¹⁵ Among 80 isolates from Swedish dogs with pyometra, resistance to AMP was detected in 8 (10%) and resistance to ENR was detected in 3 (3.8%).¹⁴ In a study¹³ of 674 nonenteric E coli isolates obtained from dogs in the United States, it was reported that 280 (42.0%) and 50 (7.7%) were resistant to AMP and ENR, respectively. In Japan, β-lactams and fluoroquinolones are routinely used in small animal veterinary practice, and this may lead to an increase in the prevalence of AMR. The data obtained in the present study have highlighted the need to pay greater attention to the emergence of AMR, especially against β-lactams and fluoroquinolone agents, among ExPEC. Currently, in Japan, there are few legal regulations regarding antimicrobial use in companion animals, in contrast to the laws associated with food-producing animals. It is hoped that veterinarians treating companion animals use relevant antimicrobial drugs responsibly and prudently.

In recent years, the worldwide dissemination of AMR E coli clones of specific sequence types has become a great issue in human medicine.²⁷ Ewers et al¹⁸ reported that the O25:H4-ST131 CTX-M-15 β-lactamase-producing clone, one of the most notable E coli clones in human medicine, was present among companion animals in European countries; however, this clone was not detected among isolates in the present study. Additionally, other clones of sequence types such as ST69, ST73, ST14, and ST95 are also implicated in human infection.²⁸ Further investigation would help to comprehend the prevalence of these sequence type clones among companion animals.

In the present study, the distribution of VFs in urogenital E coli isolates obtained from dogs and cats was also investigated. Findings indicated that there was a high prevalence of sfa, aer, and cnf genes, with positive results in approximately half of the isolates. In a previous study of E coli strains isolated from dog and cat urine in Japan by Yuri et al,⁷ the pap and hly genes, in addition to the sfa and cnf genes, were detected with frequencies (52% to 64%) similar to those determined in the present study; however, the frequency of the aer gene was lower (4% and 20% in dogs and cats, respectively). In E coli isolates from dogs with urinary tract infection and pyometra in Brazil, the sfa gene was the most prevalent (43.1% and 46.1%, respectively), followed by cnf (21.6% and 40.4%, respectively), hly (33.3% and 34.6%, respectively), and pap (23.5% and 36.5%, respectively) genes.⁹ In E coli isolates obtained from dogs with pyometra in Australia, the sfa gene was detected in 16 of 23 (70%) isolates, whereas the pap, hly, and cnf genes were commonly detected in 52%.² In E coli isolates obtained from dogs with urinary tract infections in the United States, the prevalences of the sfa, cnf, and hly genes were relatively low at 26%, 18%, and 21%, respectively.⁸ These findings imply that the prevalences of the main VFs in ExPEC isolates obtained from companion animals are influenced to some extent by the disease, geographic location, and year of investigation. This bias should be taken into account when ExPEC is monitored. The present study revealed that the afa gene, which encodes afimbrial adhesin I, was extremely rare in urogenital isolates from dogs and cats. This result was in agreement with findings of previous studies in Japan⁷ and other countries.^2,6,8,9 Thus, the afa gene is not likely to be associated with pathogenesis of urinary tract infection or pyometra in companion animals.

The phylogenetic origins of pathogenic E coli have been studied in several animal species. Wang et al^19,20 found that avian and porcine isolates mainly belonged to group A, whereas other researchers found that human isolates mainly belonged to groups B2 and D.^10,12,29 In the present study, the E coli isolates from dogs mainly belonged to phylogenetic groups B2 and D, in agreement with findings by Johnson et al.⁸ Isolates from cats had similar phylogenetic shifts as isolates from dogs, which is a novel finding. In terms of phylogenetic origin, these results indicated that ExPEC isolates from dogs and cats are more closely related to human isolates than they are to isolates from other animal species, such as poultry and pigs.

Previously, significant relationships between phylogenetic groups and AMR phenotypes in ExPEC have been reported.^8,10,21 Johnson et al^8,10 reported that the prevalence of fluoroquinolone resistance in group B2 isolates was less than that in other group isolates in dogs and humans. Likewise, Moreno et al²¹ reported less prevalence of resistance to quinolone, fluoroquinolone, and TMS in group B2 isolates from the urine of humans, compared with other group isolates. In the present study, we performed susceptibility testing with other antimicrobials and found that resistances to each of the antimicrobials tested was less prevalent (with lower aggregate AMR scores) in group B2 isolates, compared with the other group isolates. This suggested that group B2 E coli strains have a comparatively inverse association with resistance to a variety of antimicrobials in addition to those evaluated previously.

Unlike AMR phenotypes, the prevalence of VFs among phylogenetic origins of ExPEC from companion animals has not been comprehensively studied. The present study revealed that the pap, hly, and cnf genes were more prevalent in group B2 isolates, compared with findings for other group isolates. Similar findings were confirmed in E coli isolated from human urine samples by Piatti et al,¹¹ who reported a higher prevalence of those 3 VFs in addition to fimA, which encodes type 1 fimbriae in group B2 strains. It is known that the pap, hly, and cnf genes are physically linked in a specific pathogenicity island,^11,30 and this gene combination was identified in the present study and has been identified in other investigations.^2,6,31 Results of the present study and previous studies suggest that group B2 strains may frequently harbor such pathogenicity islands, compared with other group strains. Interestingly, in the present study, the combination of those 3 VF genes was detected more frequently, albeit not significantly (P = 0.059), in group B2 strains from cats (8/13 isolates) than in those from dogs (16/51 isolates), suggesting that animal species may also be involved with prevalence of the 3 genes among group B2 isolates. To fully comprehend the distribution of those 3 VFs among group B2 isolates, further investigation would be required. In contrast, the aer gene was less prevalent in group B2 isolates in the present study, compared with prevalence in group D isolates, although the reason for this is not clear. Overall, it appears that phylogenetic origin can affect the prevalence of both VFs and AMR phenotypes.

To further understand the prevalence of AMR in canine and feline E coli isolates, we investigated the association between AMR phenotypes and VFs in the present study. The findings indicated all VF genes, except aer, have an inverse association with the prevalence of AMR. A similar tendency has been previously reported for fluoroquinolone resistance in canine⁸ and human^12,21 isolates. Furthermore, the present study revealed that VFs were highly prevalent among E coli isolates susceptible to antimicrobials other than fluoroquinolones; thus, use of these antimicrobials for companion animals is unlikely to select for highly pathogenic E coli.

Only the aer gene was directly associated with resistance to all of the tested antimicrobials, except for CHL. Direct associations of the aer gene with resistances to 5 antimicrobials (AMP, CFZ, CEF, ENR, and TMS) were also evident within the group B2 isolates (data not shown); however, in other groups, the statistical correlation of any association could not be examined because of the small numbers of isolates included in the present study. Johnson et al⁸ reported that of the many VF genes, only iutA, which is regarded as one of the aerobactin receptor genes, was directly associated with fluoroquinolone resistance in E coli isolates from dogs. A similar finding in human isolates has been confirmed.^10–12 Aerobactin is the most effective chelation system used by enteric bacteria for iron acquisition and confers a growth advantage under conditions of low iron concentration.^5,32 This fact implies that aerobactin-producing strains have a greater chance of acquiring AMR, compared with strains that do not produce aerobactin. Aerobactin determinants are occasionally found on plasmids carrying AMR genes,^33–35 although the presence of such plasmids in companion animals has not been extensively studied. These traits of the aerobactin system and its determinants may explain the results of the present study and previous findings. We believe that aerobactin-positive strains should be cautiously monitored, together with AMR phenotypes.

The biased distribution of VFs among phylogenetic groups in the present study may possibly explain the difference in AMR prevalence among these groups. A lower prevalence of AMR in group B was likely attributable to a lower prevalence of aer that is directly associated with AMR and a higher prevalence of pap, hly, and cnf that are inversely associated with AMR. Many previous studies^8,10,12,21 have revealed a difference in prevalence of AMR among phylogenetic groups; however, most of these studies did not consider the relationship among VFs and phylogenetic origins. Results of the present study have indicated that the AMR phenotype is closely related with phylogenetic origins and VFs in ExPEC from companion animals; thus, both traits should be simultaneously taken into account in the monitoring of AMR in ExPEC.

Genetic diversity among ExPEC isolates from dogs and cats has not yet been evaluated, to our knowledge. Based on results of RAPD analysis, all isolates in the present study were classified in the 3 major clusters, which were highly correlated with phylogenetic origins rather than animal species. It is noteworthy that cluster I, which consisted of mainly group B2 isolates, was further divided into 2 subclusters composed of isolates that differed in VFs and AMR status. Notably, subcluster Ib (which consisted of a minor population) had lower and higher aggregate scores of VFs and AMRs, respectively, compared with subcluster Ia (which consisted of a major population within cluster I). This finding suggested that genetic diversity may affect acquisition of VF and AMR by group B2 strains. Further genotyping would help further understanding of the prevalence of these traits in ExPEC.

In the present study, characterization of the AMR status, VFs, and phylogenetic origins of urogenital E coli isolates obtained from in dogs and cats from Japan revealed a high prevalence of multidrug resistance, including resistance to fluoroquinolones and β-lactams and a distribution of phylogenetic origins similar to that of human-origin isolates. These results underscore the need to consider dogs and cats as a possible reservoir of antimicrobial-resistant ExPEC, which could be transmitted to humans. We confirmed that AMR, VFs, and phylogenetic origins (and, to some degree, genetic diversity) were closely associated with each other. Overall, these findings provide valuable knowledge regarding the comprehensive assessment of AMR in ExPEC obtained from companion animals.