Phenotypic and molecular characterization of antimicrobial resistance in Enterobacter spp. isolates from companion animals in Japan

The emergence of antimicrobial resistance among Enterobacter spp., including resistance to extended-spectrum cephalosporins (ESC), is of great concern in both human and veterinary medicine. In this study, we investigated the prevalence of antimicrobial resistance among 60 isolates of Enterobacter spp., including E. cloacae (n = 44), E. aerogenes (n = 10), and E. asburiae (n = 6), from clinical specimens of dogs and cats from 15 prefectures in Japan. Furthermore, we characterized the resistance mechanisms harbored by these isolates, including extended-spectrum β-lactamases (ESBLs) and plasmid-mediated quinolone resistance (PMQR); and assessed the genetic relatedness of ESC-resistant Enterobacter spp. strains by multilocus sequence typing (MLST) and pulsed-field gel electrophoresis (PFGE). Antimicrobial susceptibility testing demonstrated the resistance rates to ampicillin (93.3%), amoxicillin-clavulanic acid (93.3%), cefmetazole (93.3%), chloramphenicol (46.7%), ciprofloxacin (43.3%), tetracycline (40.0%), ceftazidime (33.3%), cefotaxime (33.3%), trimethoprim/sulfamethoxazole (28.3%), gentamicin (23.3%), and meropenem (0%). Phenotypic testing detected ESBLs in 16 of 18 ESC-resistant E. cloacae isolates but not in the other species. The most frequent ESBL was CTX-M-15 (n = 8), followed by SHV-12 (n = 7), and CTX-M-3 (n = 1). As for AmpC β-lactamases, CMY-2 (n = 2) and DHA-1 (n = 2) were identified in ESC-resistant E. cloacae strains with or without ESBLs. All of the ESC-resistant E. cloacae strains also harbored one or two PMQRs, including qnrB (n = 15), aac(6’)-Ib-cr (n = 8), and qnrS (n = 2). Based on MLST and PFGE analysis, E. cloacae clones of ST591-SHV-12, ST171-CTX-M-15, and ST121-CTX-M-15 were detected in one or several hospitals. These results suggested intra- and inter-hospital dissemination of E. cloacae clones co-harboring ESBLs and PMQRs among companion animals. This is the first report on the large-scale monitoring of antimicrobial-resistant isolates of Enterobacter spp. from companion animals in Japan.


Introduction
Members of the genus Enterobacter, belonging to the Enterobacteriaceae, are Gram-negative bacilli that inhabit terrestrial and aquatic environments including water, sewage, and soil, as well as the intestinal tracts of mammals [1]. Enterobacter cloacae is the most medically-important species in the genus and is responsible for nosocomial infections in humans [2]. In companion animals, this bacterial species is rarely associated with urinary tract infections, wound infections, pneumonia, intravenous catheter site infections, otitis externa, peritonitis, and dermatitis [3].
The emergence of antimicrobial resistance among Enterobacter spp. is of great concern worldwide in human medicine [1,2]. It increases the risk of antimicrobial treatment failure not only in humans but also in companion animals. Similarly, the emergence of antimicrobialresistant bacteria in companion animals may have important human public health consequences if isolates are transmitted to humans from their pets [4,5]. Understanding the prevalence of antimicrobial resistance among Enterobacter spp. isolates is thus important both from veterinary medicine and public health perspectives.
Resistance to extended-spectrum cephalosporins (ESC) among Gram-negative bacteria, including Enterobacter spp., is of particular concern [6]. In Enterobacter spp., ESC resistance is most typically caused by the overproduction of AmpC β-lactamases, which is due to the derepression of a chromosomal gene or the acquisition of a transferable AmpC β-lactamase [1,6]. In addition, extended-spectrum β-lactamases (ESBLs) and carbapenemases have been identified in Enterobacter spp. [7], exacerbating the issue of ESC resistance. An even greater concern is that most ESC-resistant Enterobacteriaceae exhibit multidrug resistance, including fluoroquinolone resistance, mainly due to chromosomal mutations in the enzymes targeted by the drug, and plasmid-mediated quinolone resistance (PMQR) [8]. In recent years, these resistance mechanisms have been well documented among Enterobacter spp. isolates from companion animals across several countries, including Australia [9,10], France [11], and Germany [12]. However, the status of emerging antimicrobial resistance among Enterobacter spp. in companion animals remains unknown in many other countries, including Japan.
The aim of the present study was to investigate the prevalence of antimicrobial resistance, and provide molecular characterization of ESC resistance and PMQR among Enterobacter spp. isolates from clinical specimens taken from dogs and cats that visited veterinary hospitals throughout Japan. A further aim was to assess the epidemiological relatedness of ESC-resistant Enterobacter spp. strains.

Bacterial isolates
A total of 60 clinical isolates of Enterobacter spp., consisting of E. cloacae (n = 44), E. aerogenes (n = 10), and E. asburiae (n = 6), were collected from dogs (n = 44) and cats (n = 16) kept by different owners that visited veterinary hospitals between 2003 and 2015. These hospitals were located at the following 15 prefectures in Japan: Hokkaido, Fukui, Gunma, Ibaraki, Saitama, Tokyo, Chiba, Kanagawa, Nagano, Aichi, Osaka, Hyogo, Tottori, Yamaguchi, and Fukuoka prefectures. The specimens were isolated from various anatomical sites, assessed as being sites of bacterial infection by clinical veterinarians, including the urinary tract (n = 27), pus from unspecified locations (n = 10), nasal cavity (n = 5), ear (n = 4), skin (n = 2), eye (n = 2), ascites (n = 2), and the other sites (n = 8). The details of Enterobacter spp. isolates used in this study are shown in S1 Table. No information was available regarding previous antimicrobial treatment of the dogs and cats. Ethical approval was not needed according to the ethical guidelines for epidemiological research by the Japanese government because this study focused on bacterial aspects. Bacterial identification was conducted by assessing the growth status on CHRO-Magar orientation medium [13], using the API 20E kit (SYSMEX bioMérieux Co., Ltd., Tokyo, Japan), and MALDI-TOF MS with the Bruker MALDI Biotyper system (Bruker Daltonics, Bremen, Germany) [14]. All confirmed Enterobacter spp. isolates were stored at −80˚C in 10% skim milk.

Phenotypic analysis of ESC-resistant Enterobacter spp. strains
ESC-resistant [i.e. minimum inhibitory concentration (MIC) for CTX or CAZ of ! 2 or 8 μg/ mL, respectively] strains were screened for ESBLs by the double-disc synergy test using CTX, CAZ, cefepime, and ACV disks on Mueller-Hinton agar plates without or with 200 μg/mL cloxacillin [18]. In addition, ESC-resistant isolates without synergism with clavulanate and with inhibition zones augmented upon cloxacillin were classified as organisms overexpressing AmpC β-lactamase [19].
Isolates with an MIC for MPM of ! 0.25 μg/mL, the recommended cut-off value for carbapenemase-producing Enterobacteriaceae by the European Committee on Antimicrobial Susceptibility Testing [20], were screened for the production of Klebsiella pneumoniae carbapenemase (KPC) and metallo-β-lactamase, and porin loss, using a D70C carbapenemase detection set (MASTDISCS™ID, UK).

Characterization of β-lactamase genes and PMQR in ESC-resistant Enterobacter spp. isolates
Genomic DNA from each of the isolates was prepared by suspending several colonies in 0.5 mL of water and boiling for 10 min. These samples were used as templates for further genetic analyses. All of the ESC-resistant strains were screened for class A β-lactamase genes (i.e. bla-TEM and blaSHV), which were identified using PCR and DNA sequencing, as previously reported [21]. Class D β-lactamase genes (blaOXA) were detected using multiplex PCR [22], and were amplified and bi-directionally sequenced using specific primers [23]. In addition, AmpC β-lactamase genes (i.e. the ACC, FOX, MOX, DHA, CIT, and EBC groups) were screened by multiplex PCR [24], and were amplified and then bi-directionally sequenced using specific primers [25,26]. In ESBL-positive strains, the CTX-M-type β-lactamase genes were detected using multiplex PCR [27]; for the positive isolates, the genes were amplified and sequenced to identify CTX-M subtypes using group-specific PCR primers [21].
Pulsed-field gel electrophoresis (PFGE) was performed on ESC-resistant E. cloacae strains, as previously described [30]. DNA embedded in agarose was digested with XbaI (Takara Bio, Inc., Tokyo, Japan) and then electrophoresed using CHEF DRIII (Bio-Rad, Hercules, CA, USA). PFGE profiles were digitized for analysis using BioNumerics software (version 5.10; Applied Maths, TX, USA). All fragment sizes within the gel were normalized using the molecular weight method. A similarity matrix was calculated using the Dice coefficient, and cluster analysis was performed using the UPGMA algorithm. A cluster was defined based on a similarity cut-off of 80% with 1.0% optimization and 1.0% band tolerance.

MLST typing of ESC-resistant Enterobacter spp. strains
As shown in Table 1

PFGE analysis of ESC-resistant Enterobacter spp. strains
In PFGE analysis, ESC-resistant E. cloacae strains formed three distinct clusters (Fig 3). Clusters I and III consisted of three ST121-CTX-M-15 strains and seven ST591-SHV-12 strains, respectively, obtained from the same veterinary hospital. In addition, cluster II contained two ST171-CTX-M-15 strains obtained from the same hospital.

Discussion
There have been few reports on the prevalence of antimicrobial resistance in overall populations of Enterobacter spp. isolates from companion animals worldwide. This study demonstrated that almost all Enterobacter spp. isolates in our collection exhibited resistance to β-lactams excluding AMP, ACV, and CMZ, possibly due to chromosomal AmpC β-lactamases [1,2]. The rates of resistance to CTX and CIP in our collection (33.3% and 43.3%, respectively) were similar to those in K. pneumoniae isolates (39.3% and 41.6%, respectively) [31], but were much higher than those in Proteus mirabilis isolates (1.9% and 5.8%, respectively) [32] from companion animals in Japan. Compared with Enterobacter spp. isolates from human in Japan [33], the rate of CIP resistance in our collection was extremely high (4.4% vs. 43.3%). Similarly, a higher rate of CIP resistance than human isolates has been reported in E. coli [34] and K.  pneumoniae [31] isolates from companion animals. These findings implied the heavy use of fluoroquinolone drugs in veterinary medicine. To evaluate inter-country diversity of the prevalence of antimicrobial-resistant Enterobacter spp. from companion animals, systematic surveillance would be needed in many countries. We found higher prevalence of ESBLs in Enterobacter isolates (16/60, 26.7%), compared with that reported from dogs (11/314, 3.5%) and cats (11/108, 10.2%) in France [11] and human isolates in Japan (22/364, 6.0%) [35]. In addition, the carriage rate of ESBLs in our collection was slightly lower that in K. pneumoniae isolates (31/89, 34.8%) [31], but was much higher than that in P. mirabilis isolates (0/103, 0%) [32] from companion animals in Japan. These data suggest that the risk of ESBL carriage is relatively high in Enterobacter spp. isolates from companion animals in Japan. In this study, the major types of ESBLs were CTX-M-15 and SHV-12. Similarly, these ESBLs have been frequently detected in companion animals in France and Australia [9,11]. As for ESBLs detected in human strains in Japan, CTX-M-3, which was an uncommon ESBL in this study (n = 1), was predominant, whereas CTX-M-15 was not detected [35]. These data suggest that the major types of ESBLs differ between Enterobacter spp. isolates from companion animals and human. Therefore, the risk of zoonotic transmission of ESBL-producing Enterobacter spp. is not likely to be high, although attention should be paid to Enterobacter spp. isolates from companion animals as reservoirs of ESBLs.
Besides ESBLs, we identified several AmpC β-lactamase genes (ACT, CMY, and DHA-type genes) in ESC-resistant strains. Notably, ACT-type enzymes are known to be representative chromosomal AmpC β-lactamases of Enterobacter spp. [36]. As for ACT-type genes, two E. asburiae strains were positive, but all of the E. cloacae strains were negative, which might imply the lack of AmpC genes or sequence variability among chromosomal AmpC genes, as previously reported [37]. The remaining two types of plasmid-mediated AmpC β-lactamases have rarely been identified in Enterobacter spp. isolated from humans [38,39]. We also found that several non-ESBL-producing ESC-resistant strains overexpressed chromosomal AmpC β-lactamase. Our findings suggest that plasmid-mediated AmpC β-lactamases, in addition to chromosomal AmpC β-lactamases, partly contribute to ESC resistance in Enterobacter spp. isolates from companion animals. On the other hand, all ESC-resistant Enterobacter spp. strains were negative for carbapenemases. This finding indicates that carbapenemases are uncommon among Enterobacter spp. isolates from companion animals in Japan, as well as among other Gram-negative bacteria [31,32,40,41], although they have been reported among animal isolates in other countries [12] and among human isolates in Japan [42].
All of the ESC-resistant E. cloacae strains possessed one or more PMQR genes, which is likely to contribute to the extremely high rate of CIP resistance among these strains (18/20, 90.0%). We found the high prevalence of qnrB and aac(6')-Ib-cr among ESC-resistant Enterobacter strains, and similar findings have been confirmed among Enterobacter isolates from companion animals in Australia [10]. Notably, most aac(6')-Ib-cr-positive strains also harbored OXA-1 and/or CTX-M-15 β-lactamases, which may indicate a strong association between these genes [43,44]. In addition, we detected the qnrS gene, which is the most prevalent PMQR gene among ESBL-producing Enterobacter spp. isolates from humans in Japan [35] but was not previously detected among isolates from companion animals in Australia [10]. These findings may suggest that qnrS gene is locally spread among companion animals and humans in Japan.
Recently, Izdebski et al. [19] found that ST66, ST78, ST108, and ST114 strains were widely spread as high-risk international clones of ESC-resistant E. cloacae. Of these STs, ST114 was identified in our collection and was associated with CTX-M-15. This is the first report of the ST114-CTX-M-15 clone in companion animals in any country other than France [11]. Furthermore, we found eight STs that had not been identified in the study by Haenni et al. [11].
This implies that ESC-resistant E. cloacae isolated from different countries generally belong to different lineages. Unfortunately, the prevalence of STs has not been reported for human ESCresistant E. cloacae isolates in Japan, and thus further studies to compare STs between animal and human isolates are needed.
Based on the PFGE analysis, we found that ESBL-producing E. cloacae clones were disseminated among different patients in the same hospital. This result strongly suggests nosocomial infections of ESBL-producing E. cloacae clones, and similar findings have previously been reported [11]. Surprisingly, several clones were repeatedly identified at an interval of many months, implying that ESBL-producing E. cloacae clones can survive inside or outside of hospitals for a long period [45]. Therefore, the dissemination of ESBL-producing E. cloacae clones among companion animals may occur not only via direct spread from animal to animal, but also via indirect transmission from potential reservoirs and sources in the environment. Our data emphasize the need for infection control in hospitals and in the community to prevent dissemination of ESBL-producing E. cloacae clones among companion animals.

Conclusion
Our data demonstrate the high prevalence of ESBLs and PMQR genes among ESC-resistant E. cloacae strains isolated from companion animals in Japan. Epidemiological data suggest that E. cloacae clones co-harboring ESBLs and PMQR genes are disseminated via intra-hospital and inter-hospital transmission.
Supporting information S1