**Prevalence, Risk Factors, and Characterization of Multidrug Resistant and ESBL**/**AmpC Producing** *Escherichia coli* **in Healthy Horses in Quebec, Canada, in 2015–2016**

#### **Maud de Lagarde 1, John M. Fairbrother 1,\* and Julie Arsenault <sup>2</sup>**


Received: 20 January 2020; Accepted: 19 March 2020; Published: 20 March 2020

**Simple Summary:** Antimicrobial resistance has been recognised as a global threat by the WHO. ESBL/AmpC genes, responsible for cephalosporin resistance, are particularly worrisome. *Escherichia coli* is a ubiquitous bacterium. Most strains are commensal, although some can cause disease in humans and animals. Due to its genome plasticity, it is a perfect candidate to acquire resistance genes. We hypothesized that multidrug-resistant *E. coli* and *E. coli* resistant to cephalosporins are present in the fecal microbiota of healthy horses in Quebec. We characterised antimicrobial resistance, identified ESBL/AmpC genes and assessed potential risk factors for their presence. Fecal samples from 225 horses, distributed in 32 premises, were cultured for indicator *E. coli* (selected without enrichment) and specific *E. coli* (selected after enrichment with ceftriaxone). Of the 209 healthy horses in which *E. coli* were detected, 46.3% shed multidrug-resistant (resistant to three or more classes of antimicrobials tested) *E. coli*. Non-susceptibility was most frequently observed for ampicillin, amoxicillin/clavulanic acid or streptomycin. ESBL/AmpC genes were detected in *E. coli* from 7.3% of horses and 18.8% of premises. The number of staff and equestrian event participation within the last three months were identified as risk factors for horses shedding multidrug-resistant *E. coli* isolates. The horse intestinal microbiota is a reservoir for ESBL/AmpC genes. The presence of ESBL/AmpC in horses is both a public and equine health concern, considering the close contact between horses and owners.

**Abstract:** Although antimicrobial resistance is an increasing threat in equine medicine, molecular and epidemiological data remain limited in North America. We assessed the prevalence of, and risk factors for, shedding multidrug-resistant (MDR) and extended-spectrum β-lactamase (ESBL) and/or AmpC β-lactamase-producing *E. coli* in healthy horses in Quebec, Canada. We collected fecal samples in 225 healthy adult horses from 32 premises. A questionnaire on facility management and horse medical history was completed for each horse. Indicator (without enrichment) and specific (following enrichment with ceftriaxone) *E. coli* were isolated and tested for antimicrobial susceptibility. The presence of ESBL/AmpC genes was determined by PCR. The prevalence of isolates that were non-susceptible to antimicrobials and to antimicrobial classes were estimated at the horse and the premises level. Multivariable logistic regression was used to assess potential risk factors for MDR and ESBL/AmpC isolates. The shedding of MDR *E. coli* was detected in 46.3% of horses. Non-susceptibility was most commonly observed to ampicillin, amoxicillin/clavulanic acid or streptomycin. ESBL/AmpC producing isolates were detected in 7.3% of horses. The most commonly identified ESBL/AmpC gene was *blaCTX-M-1*, although we also identified *blaCMY-2*. The number of staff and equestrian event participation were identified as risk factors for shedding MDR isolates. The prevalence of healthy

horses harboring MDR or ESBL/AmpC genes isolates in their intestinal microbiota is noteworthy. We identified risk factors which could help to develop guidelines to preclude their spread.

**Keywords:** antimicrobial resistance; beta-lactamase; cephalosporinase; microbiota; North America; equine

#### **1. Introduction**

Antimicrobial resistance was reported by the World Health Organization (WHO) in 2014 as the largest current threat for global health [1]. Equine medicine is also involved, indeed, the first bacteria resistant to antimicrobials in horses were reported in 1971, in Canada [2]. Subsequently, the number of treatment failure reports due to antimicrobial resistance has increased [3–5]. In Europe, several studies have reported that healthy horses can carry multidrug resistant (MDR) bacteria at a relatively high prevalence (39% to 44%) [6,7] and some countries are setting up surveillance monitoring [8]. Nonetheless, molecular and epidemiological data in this species are still limited in North America. In the global approach to antimicrobial resistance recommended by the WHO, horses have been classified as companion animals, although they are also working animals and livestock and could contaminate their owner through direct contact, or even the general population via the food chain. Thus, horses have been overlooked in the general approach to antimicrobial resistance [9].

*Escherichia coli* is ubiquitous and mainly commensal in the intestinal microbiota of mammals. However, pathogenic strains have been recognized, mostly in human and in food-producing animals, and occasionally in horses [10]. Due to its ubiquity, recurrent exposure to systemic (oral, intramuscular or intravenous) antimicrobial treatment and the fast evolution of its genome, this bacterium is considered by the Canadian Integrated Program for Integrated Surveillance System (CIPARS) as an excellent indicator for antimicrobial resistance surveillance [11].

One of the main mechanisms of resistance in *E. coli* is the production of extended spectrum β-lactamases (ESBL) and/or AmpC cephalosporinases (AmpC) [12], resulting in the hydrolyzation of the β-lactam ring, which is present in penicillins, cephalosporins and monobactams. β-lactamase genes (*bla*) have spread very effectively among numerous species of Gram-negative bacteria over the last 30 years [12], both in animals and in humans. In horses, phenotypic resistance to ceftiofur, a third-generation cephalosporin, has been reported in many clinical situations [13]. *blaCTX-M-1* is the ESBL resistance gene variant most often detected [14]. However, other variants of CTX-M (i.e., *blaCTX-M-2*, *blaCTX-M-9*, *blaCTX-M-14*, *blaCTX-M-15*) have been recognized. *blaCMY* and *blaSHV-12* have also been identified [14]. All of these variants have also been found in other animal species [14] and in humans [15]. These genes spread mainly through plasmids, carrying multiple resistance genes. Thus, these plasmids convey resistance to other antimicrobial classes, promoting multidrug resistance [16]. However, the resistance gene dissemination can also be enhanced through "high-risk" clones [17]. An example of such a clone is the *E. coli* sequence-type ST410 [18], recently emerging as a public health concern in the human population.

Moreover, owning a horse has been demonstrated as a risk factor for the carriage of ESBL in people [19] in the Netherlands. Even though the author of this study nuanced these results by stating that horse owners often own other pets, and the Netherlands has a high population density which might not be representative of the situation of other countries, nevertheless, this study underlines the potential concern for human health. The colonization with ESBL-producing Enterobacteriaceae, in humans, has been associated with an increase in the length of hospitalization in ICU patients [20]. New regulations restricting the use of antimicrobials such as fluoroquinolones and cephalosporins, classified as having the highest priority by the WHO and Health Canada [21,22], to cases where the veterinarian can prove that there is no better alternative [23], came into effect in early 2019 in veterinary medicine in Canada. Nevertheless, the use of ceftiofur will likely remain common in horses due to the

lack of a better alternative, especially for neonatal sepsis and respiratory diseases in adults, possibly enhancing the dissemination of ESBL/AmpC genes.

No data are available on the presence of MDR or ESBL/AmpC-producing isolates in the healthy equine population in Quebec. Our objective was to estimate the prevalence of, and risk factors for, shedding MDR- and/or ESBL/AmpC-producing *E. coli* isolates in horses. We characterized potential ESBL/AmpC isolates for antimicrobial susceptibility and the presence of ESBL/AmpC-associated resistance genes.

#### **2. Materials and Methods**

#### *2.1. Sampling and Data Collection*

During the summer 2015, MDL sampled healthy horses from a convenience sample of premises owned by clients and located within a one-hour drive from the CHUV, a university veterinary hospital located in Saint Hyacinthe, Quebec, Canada. To increase the number of sampled horses, in April 2016, the 111 Quebec association of equine veterinary practitioners (AVEQ) members were invited to a conference introducing the project. This event took place in Saint Hyacinthe, Quebec, and was also given in a videoconference. The veterinarians were solicited to sample healthy horses in the stables they visited as part of their veterinary practice. To evaluate the number of targeted horses sampled we used the following equation

$$m = (\mathbf{Z}^2 \times \mathbf{P}(1-\mathbf{P})) \mathbf{L}^{\mathbf{2}}$$

where *n* = the number of targeted horses (*n* = 359 horses), Z = the value from the normal distribution corresponding to the 95% confidence interval (Z = 1.96), and P = the expected prevalence of MDR *E. coli* in the healthy horse population which we extrapolated from a previous article in Great Britain [6] (P = 0.37), and L the desired precision (L = 0.05). Given the large size of the total horse population in Quebec (estimated at 129,000 individuals by Equine Canada in 2010), we have not adjusted the number of horses for a finite population. This figure does not consider the potential non-independence of the status of horses in the same premises.

Every participating veterinarian receive d a sampling kit, including 100 rectal swabs (BBLTM CultureSwabTM Plus, Becton Dickinson, France) and the material to ship the samples to the Ecl laboratory at 4 ◦C, within 48 h of collection. Protocols were explained in detail in the kit. Each veterinarian could sample up to 10 horses per premise up to a maximum of 10 premises, until the overall target sample size was reached. Only horses over two years old and considered healthy by their owner were eligible for the study. We focused our study on adult horses because breeding does not take a huge place in equestrian activity in Quebec (around 1% of horse riders are interested in breeding in Quebec according to the Cheval Quebec activity report in 2016 (https://cheval.quebec/Rapport-annuel)), therefore we expect that most contacts between people and horses during these activities are with adult horses. The sampled horses were not necessarily part of the veterinarian clientele. Each owner agreed to participate on a voluntary basis. The protocol was approved by the Université de Montréal Ethic Committee for use of animals (15-Rech-1800).

For each sampled horse, the owner and the recruiting veterinarian each completed a questionnaire online, through the Surveymonkey web platform (https://www.surveymonkey.com). They were available in both French and English and are found in the Supplementary Data of this article (Supplementary Material Figures S1 and S2). Questions were based on previously reported and suspected risk factors in the horse [24] and were related to the facility management and horse medical history. Each premise was geocoded based on its 6-digit postal code, performed in GeoPinpoint suite version 6.4 (DMTI Spatial).

#### *2.2. Indicator Collection: Non-enriched Culture, Antimicrobial Susceptibility Testing, ESBL*/*AmpC Gene Identification, and Prevalence Estimation*

On reception at the Ecl Laboratory, rectal swabs were held in Luria-Bertani (LB) broth for a maximum of 15 min at room temperature, then 100 μL of LB broth was transferred on MacConkey agar and incubated at 37 ◦C overnight. All lactose-positive colonies, up to a maximum of three, were selected for each rectal sample and cultured in LB broth then plated on MacConkey agar to ensure purity. Isolates were confirmed as *E. coli* by the presence of the *uidA* gene [25], as detected by PCR. Each sample and isolate were stored in 15% glycerol at −80 ◦C.

Isolates were tested for susceptibility to the 14 antimicrobial agents examined in the CIPARS using the disk-diffusion assay. We used the same disks and techniques as described for the indicator collection of our previous work [7].

When isolates were non-susceptible (intermediate or resistant) to 3rd generation cephalosporins, we looked for 5 β-lactamase resistance genes (*blaSHV*, *blaTEM*, *blaCMY-2*, *blaOXA*, *blaCTX-M*) by multiplex PCR. We used the same DNA extraction, PCR protocols and CTX-M-variant identification protocols as described for the indicator collection of our previous work [7].

We estimated the prevalence and 95% confidence intervals of (1) horses shedding non-susceptible (i.e., resistant or intermediate) isolate(s) for each antimicrobial, and (2) horses shedding isolate(s) non-susceptible to ≥ 1, 3, 5, 7 and 9 classes of antimicrobials. An isolate was considered MDR if non-susceptible to at least one agent in three or more antimicrobial classes [26]. We used the same method of calculation (with adjustment for sampling weights and clustering within premises) and the same software as described for in the indicator collection of our previous work [7]. We also estimated these prevalences and 95% confidence intervals at the premises level, as previously described [7]; for each outcome, a positive status was attributed when the premises housed at least one positive horse.

#### *2.3. Potential ESBL*/*AmpC Producing E. coli Collection: Culture, Antimicrobial Susceptibility Testing, ESBL*/*AmpC and Virulence Gene Identification and Descriptive Statistics*

ESBL/AmpC-producing bacteria may be shed in small quantities in healthy individuals [27]. To improve detection sensitivity and allow for a more accurate estimation of the proportion of positive horses, we carried out enrichment with ceftriaxone [8,27]. For each rectal swab suspension in LB broth, 1 mL was inoculated in 9 mL of MacConkey broth containing 1 mg/L of ceftriaxone and incubated overnight at 37 ◦C. When bacterial growth was positive, 100 μL of MacConkey broth was inoculated on MacConkey agar containing 1 mg/L of ceftriaxone and incubated at 37 ◦C overnight. All isolates up to a maximum of five lactose-positive isolates per sample were selected. All isolates of this collection were confirmed as *E. coli* by the presence of the *uidA* gene [25] as detected by PCR and were tested for susceptibility to 14 antimicrobials, as described above. All isolates in this collection were systematically examined for the presence of five β-lactamase resistance genes (*blaSHV*, *blaTEM*, *blaCMY-2*, *blaOXA*, *blaCTX-M*) by multiplex PCR (PCR and gene identification protocols are described above).

Descriptive statistics were used to present the non-susceptibility patterns of isolates from this collection. We estimated prevalence with 95% confidence intervals of horses shedding ESBL/AmpC isolates and of the premises housing these horses, using the same calculation method described above.

#### *2.4. MDR and ESBL*/*AmpC: Risk Factors*

For the risk factor analyses, two outcome variables were investigated: MDR and ESBL/AmpC status for each horse. A positive MDR or ESBL/AmpC status was defined as the detection of at least one MDR or ESBL/AmpC isolate, respectively, for that horse. All potential risk factors from the questionnaire were categorized. Putative risk factors with *p* < 0.20 (Wald test) in univariable multilevel (facilities, horses) logistic regressions were selected for inclusion in a full multivariable multilevel model for each outcome. Pairwise associations between these selected variables were assessed by χ<sup>2</sup> test; in the presence of significant association (*p* < 0.05), only one of two correlated variables was kept based on the biological relevance with the outcome. The final multivariable model was refined by sequentially omitting variables with *p* > 0.05 (Wald test). Analyses were performed in MLwiN version 2.36 using 2nd order penalized quasi-likelihood estimation, with no extrabinomial variation permitted. The fit of the final model was evaluated by visual assessment of standardized residuals at the premise level against normal scores and against fixed part prediction.

#### **3. Results**

In 2015, MDL sampled 67 horses distributed in 10 premises. Following the conference, in April 2016, 14 equine practitioners agreed to participate in the study. Although samples were collected one year apart, the results are presented together as the sampling was similar and there was no modification in the equine practice in Quebec from 2015 to 2016.

A total of 225 horses were sampled, distributed in 32 premises, as illustrated in Figure 1. Between two and 12 horses were sampled in each premise, with a mean of seven horses sampled.

**Figure 1.** Geographical distribution of the sampled premises (based on center point of their 6-digit postal code) over the administrative regions of the province of Quebec in a cross-sectional study of 209 healthy adult horses in 32 premises performed in 2015 and 2016. Two premises in Capitale-Nationale and two premises in Monteregie were very close, and therefore are overlapping on the map. Mapping was performed in ArcGIS version 10.6, using reference maps from Statistics Canada (2016 census).

Among the sampled horses, 48% were geldings, 49% were female and 3% were stallions. The mean age was 12 years old with a range from 2 to 30 years old.

#### *3.1. Indicator Collection*

*E. coli* isolates were detected in 209 (93%) of the 225 rectal swabs. A total of 609 *E. coli* isolates were selected from 209 samples, originating from the 32 premises.

The prevalence estimates of horses shedding non-susceptible isolates per antimicrobial and of the premises housing those horses are shown in Figure 2. Over 40% of horses shed isolates that were non-susceptible to ampicillin, streptomycin or amoxicillin + clavulanic acid. Over 60% of premises housed horses that shed isolates non-susceptible to streptomycin, nalidixic acid, folate pathway

inhibitors (trimethoprim-sulfamethoxazole and sulfisoxazole), ampicillin, amoxicillin + clavulanic acid or tetracycline.

**Figure 2.** Prevalence estimates (%) of non-susceptibility (yellow and red) for each antimicrobial, at the horse level (**A**), and at the premises level (**B**), in a cross-sectional study of 209 healthy adult horses, in 32 premises, performed in 2015 and 2016, in Quebec. Bars represent 95% confidence intervals for prevalence of non-susceptible isolates. The proportion of resistant isolates for each antimicrobial is presented in red. A total of 609 isolates were tested. Abbreviations: AZY = azithromycin, STR = streptomycin, GEN = gentamicin, CRO = ceftriaxone, XNL = ceftiofur, CEF = cefoxitin, CIP = ciprofloxacin, NAL = nalidixic acid, SSS = sulfisoxazole, TMS = trimethoprim–sulfamethoxazole, AMP = ampicillin, AMC = amoxicillin/clavulanic acid, CHL = chloramphenicol, TET = tetracycline.

As illustrated in Figure 2, in this collection, non-susceptibility to third generation cephalosporins (ceftiofur and ceftriaxone), was observed in 12.8% of horses and 46.8% of premises. We did not identify any *bla* genes as tested by PCR in these isolates.

The prevalence estimate of non-susceptibility to nalidixic acid, a first-generation quinolone, was high (24.7% of horses and 59.4% of premises). In contrast, the estimated prevalence of non-susceptibility to ciprofloxacin, a fluoroquinolone, was relatively low (1.0% of horses and 6.3% of premises).

Prevalence estimates (%) of horses shedding isolates non-susceptible to ≥ 1, 3, 5, 7 or 9 classes of antimicrobials and premises housing these horses, are summarized in Table 1. The prevalence of horses shedding isolates non-susceptible to at least one antimicrobial and MDR isolates were high (80.0% and 46.3%, respectively). Of the 32 premises, 81.3% housed at least one horse shedding MDR isolates. In addition, 1.4% of horses shed isolates non-susceptible to nine classes of antimicrobials, and therefore had a potential for extensive resistance [26].

**Table 1.** Prevalence estimates (%) with 95% confidence intervals (95% CI) of healthy adult horses shedding *E. coli* isolates that are non-susceptible to more than 1, 3, 5, 7 or 9 classes of antimicrobials and premises housing these horses based on the indicator collection results in a cross-sectional study of 209 horses in 32 premises in Quebec in 2015 and 2016. Abbreviations: CI = confidence interval, MDR = multidrug-resistant.


#### *3.2. ESBL*/*AmpC Collection*

A total of 7.3% [95% CI 0–17.6] of the 209 horses shed ESBL/AmpC isolates non-susceptible to ceftriaxone, therefore belonging to the ESBL/AmpC collection, and 18.8% [95% CI 4.5–33] of the 32 premises housed these horses.

Non-susceptibility pattern of the 74 isolates of this collection originating from the 17 positive horses, found in six premises, is shown in Figure 3. All isolates were non-susceptible to ampicillin and ceftriaxone, although three isolates presented susceptibility to ceftiofur and 60 isolates presented susceptibility to cefoxitin, a cephamycin, also considered as a second-generation cephalosporin [28].

**Figure 3.** Susceptibility profiles of *E. coli* isolates in the ESBL/AmpC collection, in a cross-sectional study performed on healthy adult horses, in Quebec in 2015 and 2016 (n = 74 isolates distributed in 17 horses among 6 premises). Abbreviations: AZY = azithromycin, STR = streptomycin, GEN = gentamicin, CRO = ceftriaxone, XNL = ceftiofur, CEF = cefoxitin, CIP = ciprofloxacin, NAL = nalidixic acid, SSS = sulfisoxazole, TMS = trimethoprim–sulfamethoxazole, AMP = ampicillin, AMC = amoxicillin/clavulanic acid, CHL = chloramphenicol, TET = tetracycline.

Non-susceptibility to aminoglycosides (gentamicin and streptomycin), tetracycline, folate inhibitors (trimethoprim-sulfonamides, sulfizoxasole) and chloramphenicol were present in over 60% of the isolates.

A total of 54.1% of isolates were non-susceptible to a first-generation quinolone (nalidixic acid) and 20.3% of isolates were non-susceptible to ciprofloxacin, a fluoroquinolone, in this collection. These isolates were therefore non-susceptible to two families of antimicrobial classified as having the highest priority in human medicine by both Canadian Health and the WHO [21,22].

The main ESBL genes identified were *blaCTX-M-1* (43/74 tested isolates) and *blaSHV* (15/74), four isolates carried a combination of *blaCTX-M-1* and *blaSHV*. Nine isolates carried the AmpC gene *blaCMY-2*. In four isolates we could not detect tested ESBL/AmpC genes by PCR.

#### *3.3. Risk Factors*

A total of 13 potential risk factors were derived from the questionnaire (Table 2). Eleven were considered at the individual level and two were considered at the premise level.

**Table 2.** Descriptive statistics and *p*-value (Wald test) from univariable multilevel logistic regression analyses of potential risk factors for MDR status in horses in a cross-sectional study performed on healthy adult horses, in Quebec, in 2015 and 2016. In bold are the factors that were retained for the multivariable analysis.



**Table 2.** *Cont.*

<sup>1</sup> Defined as a horse that stays on pasture at night and has a shelter in the pasture. <sup>2</sup> Categorization was done *a posteriori*, based on the mean of the number of horses in the premises we sampled. <sup>3</sup> This variable was already categorized in the questionnaire.

Data with missing values, representing almost half of the dataset, were excluded from modeling. A total of five risk factors were selected for multivariate modeling (all *p* < 0.20 in univariable logistic regressions) for the MDR outcome. The variable "The horse presented an infection" was then excluded as it was associated with "The horse has been medically treated within the last 3 months". The variable "Transportation within the last 3 months" was excluded because it was associated with "Participating in an equestrian event within the last 3 months".

According to the multivariate model, the odds of being an MDR horse were 3.5 times higher (*p* = 0.03) among the horses that had participated to an equestrian event within the last three months and 3.4 times higher (*p* = 0.01) if the horse was in a premise where the staff were composed of more than five persons (Table 3). Visual assessment of residuals at the premise level suggested that our model fitted the data.

**Table 3.** Parameter estimates and odds ratios from a multivariable regression modeling MDR positive status at the horse level, based on the results of a cross-sectional study performed on 32 premises and 209 healthy adult horses, sampled in Quebec, in summers 2015 and 2016. The estimated variance at the premises level was 0.171 (standard error of 0.316).


Interaction between the two variables of the final multivariable model were checked but were not significant (*p* = 0.41, Wald test) and thus not kept in the model.

For the ESBL/AmpC outcome, considering the high percentage of missing data and low number of positive horses, no statistical modelling was performed.

#### **4. Discussion**

The present study illustrates that the fecal microbiota of healthy horses in Quebec, Canada, harbor MDR and ESBL/AmpC *E. coli* isolates. The prevalence of horses shedding ESBL/AmpC *E. coli* isolates (7.3%) is comparable to that which was detected phenotypically in the United Kingdom in 2012 (6.3%) [6]. Nevertheless, at the premise level, it seems that the prevalence in Quebec (18.8%) is inferior to the prevalence reported in France (29.0%) [7]. However, these regional differences in apparent prevalence might be related to a higher sensitivity in the detection of positive premises in the study in France, considering that in France we tested more horses per premise (between six and 36 horses per premises) and ESBL/AmpC isolates were detected by two enrichment methods. The prevalence we found in horses in Quebec contrasts with the 1% prevalence observed in Ontario among 188 healthy dogs in 2009 [29], although, even if the calculation methods are not the same, this is still lower than the 26.5% of fecal carriage of ESC-resistant Enterobacteriaceae in healthy dogs in Ontario in 2018 [30]. The prevalence of horses shedding ESBL/AmpC *E. coli* isolates in Quebec is higher than that reported in Sable Island horses, where 1/508 horses shedding an ESBL gene [31] (*blaCTX-M-1*) was found. This is not surprising because our horse population is in contact with the populations of other species in which ESBL/AmpC genes have been detected, such as pigs [32], poultry [33], cattle [34] and humans [35], underlining the importance of the one health approach [1,36] to address the problem.

Our study reported the presence of isolates that are non-susceptible to nine classes of antimicrobial in an indicator collection of *E. coli* from horses for the first time to our knowledge, which is worrisome. Although these isolates may be commensals, it is possible that putative resistance genes are carried by mobile genetic elements, such as plasmids, and are therefore transmissible to potential pathogenic or zoonotic strains. The dissemination of extensive resistance to pathogenic strains could lead to an increased risk of complications in the treatment of infections caused by these strains.

Enrofloxacin, a fluoroquinolone, is classified as having a very high importance in human medicine [22] and is approved for veterinary use in equine medicine. Resistance to quinolones is known to be acquired and is mostly due to the apparition of chromosomal mutations, although resistance genes carried by plasmids have also been reported [37]. Often, the chromosomal mutations appear consecutively and are localized on the genes *gyrA* and *parC* (coding for gyrase and topoisomerase, respectively, both involved in the DNA synthesis). The number of mutations is proportionate to the minimal inhibitory concentration (the more the higher). Hence, non-susceptibility to nalidixic acid is generally precursory for fluoroquinolone treatment failure [38]. In the indicator collection, we detected 24.7% of horses and 59.4% of premises presenting a non-susceptibility to nalidixic acid, suggesting that enrofloxacin should be used with caution, to maintain its efficacy in horses in Quebec. In the ESBL/AmpC collection, we found 14 isolates presenting a non-susceptibility to both 3rd generation cephalosporin and fluoroquinolones. Even though these isolates are unlikely to be pathogenic, they still represent a risk of dissemination due to their high capacity to resist antimicrobial pressure. They could acquire virulence genes through the transfer of plasmids thus becoming a threat for public and/or equine health.

In our study, the predominant ESBL gene found was *blaCTX-M-1*. ESBL of the CTX-M family have become a public health concern in the last two decades, their incidence and diversity having increased dramatically during this time and have overridden other ESBL variants such as *blaTEM* and *blaSHV* in gram negative bacteria [15]. The *blaCTX-M* encoded ESBL family is characterized by the ability to inhibit 3rd and 4th generation cephalosporins and monobactams, but not cephamycins and carbapenems. These ESBLs are also known to be susceptible to β-lactam inhibitors. However, no cephamycin or penicillin/β-lactam inhibitor combinations are approved or used off-label (to the authors' knowledge) to treat horses. The predominance of *blaCTX-M-1* suggests a global dissemination of this gene in the equine population both in Europe and in North America. The absence of other variants of *blaCTX-M* in

the Quebec horse population contrasts with the high diversity of *blaCTX-M* found in the healthy equine population in France and throughout Europe [7]. This suggests that the presence of this family of genes may have occurred later in North America than in Europe, and that the genes may not yet have had the time to diversify.

We detected the AmpC resistance gene *blaCMY-2* in several horses. This gene has been frequently found in poultry and pigs, including in Quebec [32,39]. Although this gene has previously been identified in one healthy horse in France, the fact that we identified it in several healthy horses in Quebec suggests the possibility of AmpC gene spread between animal species. Indeed, horses can be in contact with other animal species, including dogs, cats, poultry among others, in the premises.

Even though we detected 12.8% of horses carrying isolates non-susceptible to 3rd generation cephalosporins in the indicator collection, none of these isolates carried the tested ESBL/AmpC genes, similar to what had been found in the indicator collection of our previous work [7]. These findings suggest that other mechanisms of resistance to cephalosporins (for example, alteration of the protein binding protein) may be present in the population. These alternative mechanisms are less likely to spread through plasmids but could impact cephalosporin efficacy, and therefore could affect equine welfare. We also found four isolates of the ESBL/AmpC collection in which we could not identify a *bla* gene. This could indicate that other, less common, *bla* genes are present in the horse population.

Among the risk factors model selected for modeling, the correlation between the variable "The horse presented an infection" and "The horse has been medically treated within the last 3 months" was to be expected, because a horse with an infection is often treated for this infection. The medical treatment of the horse was considered more biologically relevant to influence the shedding of MDR *E. coli* rather than the infection itself. However, this variable was not retained in the final model, perhaps because of the absence of specific information about the type of treatment, which could include treatments other than antimicrobials.

A correlation between "Transportation within the last 3 months" and "Participating in an equestrian event within the last 3 months" was also observed, which was not surprising, as horses which participate in an equestrian event are often transported to the equestrian event. We chose to consider participation in the equestrian event because of the possibility of transmission of antimicrobial resistance genes inter- and intraspecies during the event.

To our knowledge, we demonstrated for the first time that participation in an equestrian event was a risk factor for shedding MDR isolates at the horse level. Considering the correlation between the horse participation in an equestrian event and transportation, this effect could also be driven by contacts occurring during transportation. Based on this association, we could suggest isolating horses that are participating in equestrian events or at least the implementation of appropriate biosecurity measures. As an example, limiting contact between these horses and horses that stay at home or handling horses that stayed at home before horses that travelled might be beneficial to limit antimicrobial gene dissemination. However, more longitudinal studies are needed to establish the duration of shedding, and therefore be more accurate in these recommendations.

Our results suggest that a higher number of persons taking care of horses daily increases the risk of detecting MDR isolates in the horse's intestinal microbiota. We previously documented that this factor was associated with a higher risk of detecting ESBL/AmpC isolates in the healthy equine population in France [7]. The fact that this variable was found to be significant in both studies is noteworthy. Indeed, such information is easily obtained, and therefore could be helpful for elaborating guidelines to improve equine health. It could help equine veterinarians in defining "at-risk" equine populations and encourage the use of antimicrobial susceptibility testing in these populations.

The absence of a probability sampling method in our study might affect the representativeness of our prevalence estimate. The extrapolation of such estimate to the general equine population should be made cautiously, as the horses selected for our study are more likely representative of a subpopulation of horses under regular veterinary follow-up examination. Another limitation of our study is the recruitment of a smaller sample size of horses than planned, combined with a high percentage of missing data for the questionnaire among recruited horses (almost 50%), thus reducing the precision of the prevalence estimates and statistical power of the risk factor analyses. The low participation rate could be due to a lack of awareness of the importance of antimicrobial resistance in the equine industry. A higher proportion of missing values were present in horses shedding MDR isolates. This could be due to some regional differences and/or owner characteristics influencing both the risk of MDR and interest to participate in our study. The validity of our results depends on the absence of association between response rate and exposure to identified risk factors. Such association seems unlikely considering that the MDR status and associated risk factors were unknown for both horse owners and veterinarians at the time of data collection.

A valuable follow up to this study would be to sample the veterinarians and owners of these horses and see if there is a correlation between horses and horse handlers for the carriage of ESBL/AmpC producing *E. coli*. Another interesting follow up would be to repeat the study a few years after the regulations (see introduction) have been set up and see if these have made a difference.

#### **5. Conclusions**

In conclusion, we found a noteworthy prevalence of ESBL/AmpC genes and MDR isolates in the fecal microbiota of healthy horses in Quebec. Surveillance of ESBL/AmpC gene dissemination and the quantification of MDR isolates would be beneficial to characterize the nature and the extent of the risk they represent, with the aim of limiting their transmission between horses, but also to other species including humans and to the environment. The detection of risk factors for MDR shedding could be used to help equine veterinarians in managing at-risk populations.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-2615/10/3/523/s1, Figure S1: Questionnaire in French, Figure S2: Questionnaire in English.

**Author Contributions:** Conceptualization, M.d.L., J.A., J.M.F.; methodology, M.d.L., J.A., J.M.F.; software, M.d.L., J.A.; validation, J.A.; formal analysis, M.d.L., J.A.; investigation, M.d.L.; resources, M.d.L., J.M.F.; data curation, M.d.L., J.A., J.M.F.; writing—original draft preparation, M.d.L.; writing—review and editing, J.A., J.M.F.; visualization, M.d.L., J.A., J.M.F.; supervision, J.A., J.M.F.; project administration, M.d.L.; funding acquisition, M.d.L., J.M.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Ecl Laboratory, the AVEQ and the MAPAQ.

**Acknowledgments:** We thank all the equine veterinarians from Quebec that participated to our study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Multidrug-Resistant ESBL**/**AmpC-Producing** *Klebsiella pneumoniae* **Isolated from Healthy Thoroughbred Racehorses in Japan**

#### **Eddy Sukmawinata 1, Ryoko Uemura 2,3,\*, Wataru Sato 2, Myo Thu Htun <sup>2</sup> and Masuo Sueyoshi 1,2,3**


Received: 28 January 2020; Accepted: 20 February 2020; Published: 25 February 2020

**Simple Summary:** Extended-spectrum β-lactamases (ESBLs) and AmpC β-lactamases (AmpCs) have been recognized as an emerging global problem in humans and animals. These enzymes provide a mechanism of resistance by inactivating β-lactam antibiotics and are mostly encoded on plasmids, which can be easily transmitted to other bacteria in humans, animals, and the environment. Several clinical diseases caused by *Klebsiella* spp. infection have been confirmed in the horse community. The emergence of antimicrobial resistance in *Klebsiella* spp. increases the risk of treatment failure in infected horses. In this study, we investigated the presence of ESBL/AmpC-producing *Klebsiella* spp. isolated from healthy Thoroughbred racehorses in Japan. The results showed that ESBL/AmpC-producing*Klebsiella pneumoniae*(ESBL/AmpC-KP) isolated from horses have co-resistance to other β-lactam antibiotics as multidrug-resistant (MDR) bacteria. Genetic relatedness analysis suggested that plasmid-mediated AmpC-KP clones may spread between horses. This is the first study to show *K. pneumoniae* carrying MDR plasmid-mediated AmpC isolated from racehorses. Continuous monitoring antimicrobial resistance to this species is required in order to control the spread of MDR ESBL/AmpC-KP in the racehorse community.

**Abstract:** Extended-spectrum β-lactamase (ESBL)- and AmpC β-lactamase (AmpC)-producing *Klebsiella* spp. have become amajor health problem, leading to treatment failurein humans and animals. This study aimed to evaluate the presence of ESBL/AmpC-producing *Klebsiella* spp. isolated from racehorses in Japan. Feces samples from 212 healthy Thoroughbred racehorses were collected from the Japan Racing Association Training Centers between March 2017 and August 2018. ESBL/AmpC-producing *Klebsiella* spp. were isolated using selective medium containing 1 μg/mL cefotaxime. All isolates were subjected to bacterial species identification (MALDI-TOF MS), antimicrobial susceptibility test (disk diffusion test), characterization of resistance genes (PCR), conjugation assay, and genetic relatedness (multilocus sequence typing/MLST). Twelve ESBL/AmpC-producing *Klebsiella pneumoniae* (ESBL/AmpC-KP) were isolated from 3.3% of horse samples. Antimicrobial resistance profiling for 17 antimicrobials showed all ESBL/AmpC-KP were multidrug-resistant (MDR). Only 1 isolate was confirmed as an ESBL producer (*bla*CTX-M-2-positive), whereas the other 11 isolates were plasmid-mediated AmpC (pAmpC) producers (*bla*CMY positive). On the basis of MLST analysis, the ESBL-KP isolate was identified as sequence type (ST)-133 and four different STs among AmpC-KP isolates, ST-145, ST-4830, ST-4831, and ST-4832, were found to share six of the seven loci constituting a single-locus variant. This is the first study to show *K. pneumoniae* carrying MDR pAmpC isolated from a racehorse.

**Keywords:** extended-spectrum β-lactamase; AmpC β-lactamase; *Klebsiella pneumoniae*; horse; multidrug resistance

#### **1. Introduction**

*Klebsiella* spp. is a normal intestinal bacteria in horses [1] and is ubiquitous in the environment [2]. However, some studies have reported *Klebsiella* spp. as a causal agent for infections in horses, such as mares with metritis and cervicitis, foals with septicemia and pneumonia [3], and pneumonia in adult horses [2], and disease severity depends on the pathogenicity of the strains [1]. In the horse industry, about 25%–60% of economic losses are caused by endometritis, and *Klebsiella pneumoniae* was reported as one of the causal infections that can be transmitted through the venereal route [4,5]. First-, second-, and third-generation cephalosporin has been used for treatment of bacterial infection in equine medicine for several years. Ceftiofur, which belongs to third-generation cephalosporin, is approved for used in horses and effective in treatment of *Klebsiella* infection. In special cases, such as septicemia in foals and respiratory tract disease in horses, cefquinome, which is a fourth-generation cephalosporin, is accepted for use in the United Kingdom [6]. However, the occurrence of antimicrobial resistance (AMR) in *Klebsiella* spp. has increased the risk of treatment failure [7].

Extended-spectrum β-lactamases (ESBLs) and AmpC β-lactamases (AmpCs) have emerged globally in humans and animals [8]. These enzymes can hydrolyze extended-spectrum cephalosporin [8], whereas AmpCs have a broader resistance spectrum to cephalosporins, including cephamycins (cefoxitin and cefotetan) [9]. ESBL and AmpC genes are mainly located on mobile genetic elements such as plasmids, which can be transferred to other bacteria in humans, animals, or the environment [10]. Nonetheless, AmpC is less frequently reported than ESBL [11,12]. β-Lactamase inhibitors such as clavulanic acid, sulbactam, and tazobactam have the effect of inhibiting the production of ESBL [9,13], but these have much less effect on AmpC β-lactamase [12].

Extended-spectrum β-lactamase-producing *Enterobacteriaceae* have gained special attention on AMR in horses due to their presence as a potentially zoonotic bacteria [14]. The CTX-M family of ESBL have been reported as the predominant type of ESBL after the TEM and SHV types [15], and more than 200 CTX-M variants have been identified worldwide [16]. On the other hand, some species of *Enterobacteriaceae* (such as *Enterobacter cloacae*, *Enterobacter aerogenes*, *Aeromonas* sp., *Citrobacter freundii*, *Providencia* sp., *Serratia marcescens*, *Hafnia alvei*, *Morganella morganii*, and *Pseudomonas aeruginosa*) have resistance to extended-spectrum cephalosporin, which may be caused by inducible chromosomal AmpC. Furthermore, plasmid-mediated AmpC (pAmpC) were identified from *Enterobacteriaceae* such as, *Klebsiella* spp., *Escherichia coli*, *Salmonella* spp., and *Proteus mirabilis* [17]. The distribution of pAmpC seems to be more frequent in animals than in humans [16]. Although ESBL/AmpC-producing *Klebsiella* spp. (ESBL/AmpC-K) are considered a major global concern, information is still lacking for AMR in horses [1]. Moreover, information on ESBL/AmpC-K in horses is unavailable in Japan. This study aimed to evaluate the presence of ESBL/AmpC-K isolated from healthy Thoroughbred racehorses in Japan. In addition, although carbapenems are rarely used in pet animals, these antimicrobials are frequently considered as the last option of treatment for ESBL/AmpC-producing bacteria infection [18,19]. In this work, all ESBL/AmpC-positive isolates were also tested for carbapenemase production.

#### **2. Materials and Methods**

#### *2.1. Isolation of ESBL*/*AmpC-K*

Feces samples from 212 healthy Thoroughbred racehorses were collected by veterinarians at the Japan Racing Association (JRA) between March 2017 and August 2018. Sampling locations were the Miho Training Center (103 samples) and Ritto Training Center (109 samples). No samples were from horses under treatment with antibiotics. Fresh feces samples from each individual horse were collected and stored in sterile plastic bags. Samples were sent immediately to our laboratory in a cooling box. ESBL/AmpC-K was screened on the basis of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guideline by using MacConkey agar (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) supplemented with 1 μg/mL cefotaxime (CTX; Duchefa Biochemie B.V. Haarlem, North Holland, the Netherlands) [20]. One to three colonies with pink, mucoid, and lactose fermented appearance were selected for species identification by using MALDI-TOF MS (Bruker, Billerica, MA, USA). All presumptive ESBL/AmpC-K isolates were stored frozen in trypticase soy broth (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) with 20% glycerol at −80 ◦C for further analysis. *Klebsiella pneumoniae* ATCC 700603 and *E. coli* ATCC 25922 were used as positive and negative control type strains, respectively.

All presumptive isolates were confirmed for ESBL and AmpC production by using the AmpC and ESβL Detection Set (D68C). All ESBL/AmpC positive isolates were further tested for carbapenemase production by Mastdiscs Combi Carba Plus (D73C), and the results were interpreted based on manufacturer guidelines (Mast Diagnostics, Merseyside, United Kingdom).

#### *2.2. Antimicrobial Susceptibility Test*

The antimicrobial susceptibility testing of all isolates were performed by disk diffusion assay to 17 antimicrobial agents belonging to 8 classes of antimicrobial, β-lactam (ampicillin 10 μg (ABPC), cefuroxime 30 μg (CXM), cefotaxime 30 μg (CTX), ceftazidime 30 μg (CAZ)), aminoglycoside (gentamicin 10 μg (GM), kanamycin 30 μg (KM), streptomycin 10 μg (SM), tetracycline (tetracycline 30 μg (TC), oxytetracycline 30 μg (OTC), doxycycline 30 μg (DOXY)), amphenicol (chloramphenicol 30 μg (CP)), polypeptide (colistin 10 μg (CL)), quinolone (nalidixic acid 30 μg (NA), norfloxacin 10 μg (NFLX), marbofloxacin 5 μg (MAR)), fosfomycin 200 μg (FOM), and folate antagonist-sulfonamide (trimethoprim/sulfamethoxazole 1.25/23.75 μg (STX)). Minimum inhibition zones were interpreted using the Clinical Laboratory Standard Institute (CLSI) criteria [21]. Multidrug-resistant (MDR) bacteria were termed to isolates that had resistance to at least three or more classes of antimicrobials [22]. *E. coli* ATCC 25922 strain was used for quality control.

#### *2.3. Molecular Characterization of ESBL*/*AmpC-K*

DNA from ESBL/AmpC-K isolates was extracted on the basis of the previously described method [23]. All ESBL/AmpC-positive isolates, the CTX-M-type β-lactamase and pAmpC genes were detected by multiplex PCR [24,25]. The *bla*TEM and *bla*SHV genes were identified by PCR and directly sequenced to confirm the type of β-lactamase [24]. Chromosomal AmpC, *bla*CMY, *strA*, *strB*, *aphA1*, *tetA*, *tetB*, *cat*, and *floR* genes were identified by PCR [26–28], then one positive sample for each gene was selected for DNA sequencing to confirm the expected size, which was used as a positive control for other samples [24]. The results were analyzed with MEGA 7.0 (https://www.megasoftware.net/) and were examined with the National Center for Biotechnology Information, Basic Local Alignment Search Tool (NCBI BLAST) program (http://www.ncbi.nlm.nih.gov/blast/). The sequence types (STs) of *K. pneumoniae* were identified by multilocus sequence typing (MLST) on the basis of a previous report [29]. Novel STs were submitted to *Klebsiella pneumoniae* PubMLST and were termed as new STs (https://bigsdb.pasteur.fr/klebsiella/klebsiella.html).

#### *2.4. Conjugation Assay*

Transfer of antibiotic resistance was studied using conjugation for all ESBL/AmpC-K isolates. A plasmid-free and nalidixic acid-resistant (F<sup>−</sup>, Nar ) of *E. coli* DH5α (Takara Bio Inc., Shiga, Japan) was used as a recipient strain, whereas all ESBL/AmpC-K resistant to NA served as donors. Conjugation was performed on the basis of our previous study [24].

#### *2.5. Statistical Analysis*

The antimicrobial susceptibility profile and the efficiency of conjugation were analyzed by descriptive statistics using Excel 2017 (version 15.40; Microsoft, Redmond, WA, USA).

#### **3. Results**

#### *3.1. Resistance Phenotype*

In this study, 12 ESBL/AmpC-producing *K. pneumoniae* (ESBL/AmpC-KP) were isolated from 7 (3.3%; 7/212) healthy Thoroughbred racehorse feces samples. Phenotypically, 11 isolates were confirmed as AmpC producers from 6 samples (2.8%; 6/212) that came from the Ritto Training Center, and only 1 sample from the Miho Training Center was confirmed as an ESBL producer. All ESBL/AmpC isolates were not identified as carbapenemase producers. All samples were resistant to ABPC, CXM, CTX, TC, OTC, DOXY, and FOM, followed by CAZ (83.3%; 10/12), GM (75.0%; 9/12), KM (66.7%; 8/12), SM (8.3%; 1/12), and CP (8.3%; 1/12). All isolates (100%; 12/12) were defined as MDR, meaning that they were resistant to at least three classes of antimicrobial. ESBL, AmpC, and resistance phenotype patterns are shown in Figure 1.

**Figure 1.** Characteristics of extended-spectrum β-lactamase/AmpC β-lactamase-producing *Klebsiella pneumoniae* (ESBL/AmpC-KP) isolated from Thoroughbred racehorses in Japan. <sup>1</sup> Miho, Miho Training Center, Japan Racing Association, Ibaraki; Ritto, Ritto Training Center, Japan Racing Association, Shiga. <sup>2</sup> ABPC, ampicillin; CXM, cefuroxime; CTX, cefotaxime; CAZ, ceftazidime; GM, gentamicin; KM, kanamycin; SM, streptomycin; TC, tetracycline; OTC, oxytetracycline; DOXY, doxycycline; CP, chloramphenicol; CL, colistin; NA, nalidixic acid; NFLX, norfloxacin; MAR, marbofloxacin; FOM, fosfomycin; STX, trimethoprim/sulfamethoxazole; red, resistance; yellow, intermediate; green, susceptible.

#### *3.2. Molecular Characteristic of ESBL*/*AmpC-KP*

The presence of SHV-1 β-lactamase genes (non-ESBL) were detected in all (100%; 12/12) ESBL/AmpC-KP isolates, and CTX-M-2 was only detected from one isolate that showed an ESBL phenotype. None of the isolates were positive for chromosomal AmpC genes, whereas CMY, which belonged to the CIT family of pAmpC, was detected in all AmpC phenotype isolates. The *strA*, *strB*, and *tetA*, genes were detected in nearly all (91.7%; 11/12) ESBL/AmpC-KP isolates, followed by *floR* (75.0%; 9/12) and *tetB*, which was only detected in one isolate (8.3%; 1/12). All isolates were subjected to MLST analysis. As a result, ESBL-KP isolate was identified as ST-133, and four different STs among AmpC-KP isolates, ST-145 (54.5%; 6/11), ST-4830 (27.3% 3/11), ST-4831 (9.1%; 1/11), and ST-4832 (9.1%; 1/11), shared six of the seven loci constituting a single-locus variant (SLV). In these results, AmpC-KP ST-4830, ST-4831, and ST-4832 were termed as new STs. Characteristic ESBL/AmpC-KP and other resistance genes are summarized in Figure 1.

#### *3.3. Conjugation Assay*

Conjugation assay was only successful in ESBL-KP ST-133. Horizontal transmission was confirmed by detection of *bla*CTX-M-2 in the transconjugant strain with the frequency of transfer 2 <sup>×</sup> 10−<sup>4</sup> per donor cell.

#### **4. Discussion**

In this study, 3.3% of samples from racehorse feces were confirmed as having ESBL/AmpC-KP. Interestingly, 91.7% of total isolates were AmpC producers, which were only isolated from the Ritto Training Center. One isolate (8.3%) was identified as ESBL-KP, derived from the Miho Training Center. ESBL-KP isolated from horses was reported at 0.2% (3/1347) in the Netherlands [13]. In Germany and other European countries, 3.1% (5/160) of ESBL-KP was reported among clinical horse samples [30]. Another study showed that ESBL-KP was isolated from 1.8% (1/55) of foals on admission to hospital, and the shedding rate increased during hospitalization in Israel [31]. The selection of ESBL producers among *Enterobacteriaceae* is expected as the impact of cephalosporin antibiotics used for medical treatment in horses [24].

In our results, all ESBL/AmpC-KP isolates were detected as carrying *bla*SHV-1, which is resistant to penicillin and early generation cephalosporin but not resistant to third-generation cephalosporin. SHV-1 is mainly reported in *K. pneumoniae* and may be due to the gene encoded SHV-1, which was located on the chromosome of this species. SHV-1 β-lactamase has also been reported for up to 20% of plasmid-mediated ampicillin *K. pneumoniae* [32]. Our study also confirmed that ESBL-KP isolate was carried the *bla*CTX-M-2 gene. CTX-M-2-producing *E. coli* were also detected from the same horse feces sample (data not shown), as reported in our previous study [33]. Conjugation assay showed that *bla*CTX-M-2 was transferred with the frequency of transfer 2 <sup>×</sup> 10−<sup>4</sup> per donor cell. This finding suggests that horizontal transmission among bacterial species in horse intestine occurred. In Japan, CTX-M-2-producing *K. pneumoniae* have been confirmed in dogs [7], humans [34,35], and broiler chickens [36]. In addition, conjugative plasmids carrying *bla*CTX-M-2 have been reported in *K. pneumoniae* isolated from dairy cows with clinical mastitis [37]. In contrast to ESBL-KP, the presence of AmpC-KP in horses is less well documented, but our study identified them as a dominant β-lactamase producer.

The screening test for detection of AmpC-producing bacteria can be performed by the same protocol for ESBL screening test, and multiplex PCR has been developed to identify pAmpC [17]. All AmpC phenotype isolates in our study contained *bla*CMY belonging to the *bla*CIT type of the pAmpC gene. CMY-2 is prevalent among AmpC enzymes in the animal sector [38]. None of the pAmpC-KP isolates were conjugative under our experimental conditions. To our knowledge, no previous studies have been published describing the rate of *K. pneumoniae* carrying pAmpC isolated from horses. Plasmid-mediated AmpC has been reported worldwide from enterobacteria not predicted to produce AmpC β-lactamases [12]. In equine medicine, previous studies have shown that pAmpC genes belonging to *bla*CMY-2 were detected from extended-spectrum cephalosporin-resistant (ESCR) *E. coli* isolated from diseased horses in the Netherlands (0.1%; 1/1347) and the United Kingdom (3.8%; 2/52) [13,39]. The *bla*CMY was also identified from *Salmonella* spp. isolated from horses in the United States and Ireland [8]. The *bla*EBC (5.8%; 3/52) identified from ESCR *E. coli* has been reported in the United Kingdom [39]. Plasmid-mediated AmpC-KP has been isolated from dogs and/or cats in South Korea [10], China [40,41], Japan [7], Switzerland [42], and Italy [3], and most of these belong to the CMY and DHA groups. In this work, no AmpC-KP isolates were also confirmed as ESBL producers, and vice versa. This might be related to the antimicrobials used in the treatment of animals [43]. In a previous study, CTX-M-2- and CMY-2-producing *E. coli* were reported in broiler chickens in Japan [44]. In addition, the susceptibility to carbapenems could be decreased by combination of AmpC production and porin deficiency [18]. Nevertheless, no ESBL/AmpC-KP showed activity as carbapenemase producers in this study.

ESBL and pAmpC-producing bacteria mostly have co-resistance with other antimicrobials [3,38]. The ESBL/AmpC genes are frequently located on an MDR plasmid, which plays a key role in their dissemination [45]. Our results showed the occurrence of MDR ESBL/AmpC-KP isolated from horses (3.3%; 7/212) was lower than from dogs and cats (30.1%; 31/103) in Japan [7]. Most MDR ESBL/AmpC-KP isolates showed co-resistance with aminoglycoside (*strA*- and *strB*-positive), tetracycline (*tetA*- and/or *tetB*-positive), and FOM. Only ESBL-KP isolates showed resistance to CP, but the *floR* gene, which is

responsible for CP resistance, was detected in most CP non-susceptible AmpC-KP isolates. Similar to our results, MDR ESBL/AmpC-KP against aminoglycosides, tetracyclines, and amphenicol-mediated *strA*/*B*, *tet*, and *cat* genes have also been confirmed from dogs and cats in Italy [3]. Co-selection, when using antimicrobials other than ESCs for therapy, may maintain the existence of MDR ESBL/AmpC-producing bacteria in animals [38]. Treatment options for MDR ESBL/AmpC-KP infection might be limited when considering that several clinical cases have been reported from this species in horses.

MLST analysis showed that *K. pneumoniae* ST-133 was identified as an ESBL producer in this study. Previously, ESBL-KP ST-133 has been reported in humans in Japan [46]. Four different STs of AmpC-KP (ST-145, ST-4830, ST-4831, and ST-4832) in this study have not been reported between humans and animals in Japan. AmpC-KP ST-145 and three new STs, which are SLV of ST-145, were only distributed at the JRA Ritto Training Center. Further investigation is needed to confirm whether the dissemination of ESBL/AmpC-KP occurred inside or outside the training center.

#### **5. Conclusions**

In conclusion, this is the first study that has shown *K. pneumoniae* carrying MDR pAmpC isolated from racehorses. Interestingly, our results showed that the percentage of pAmpC-KP is higher than ESBL-KP, as compared with other previous reports. Dissemination of MDR ESBL/AmpC-KP through fecal material in the training centers requires special attention among the racehorse community, as indirect transmission may occur in the environment. Risk of infection by MDR ESBL/AmpC-KP may occur in people who work in close contact with racehorses (e.g., veterinarians, caretakers, and owners).

**Author Contributions:** Conceptualization, E.S. and R.U.; methodology, E.S., W.S., and M.T.H.; software, E.S.; validation, E.S., R.U., and M.S.; formal analysis, E.S. and R.U.; investigation, E.S.; resources, E.S. and R.U.; data curation, E.S.; writing—original draft preparation, E.S.; writing—review and editing, E.S., R.U., and M.S.; visualization, E.S.; supervision, R.U. and M.S.; project administration, R.U.; funding acquisition, R.U. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors gratefully thank the Japan Ricing Association for providing the fecal samples for use in the present study. We are grateful to the team of curators of the Institut Pasteur MLST and whole genome MLST databases for curating the data and making them publicly available at http://bigsdb.pasteur.fr.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


isolates from companion animals in Japan: Clonal dissemination of multidrug-resistant extended-spectrum β-lactamase-producing *Klebsiella pneumoniae*. *Front. Microbiol.* **2016**, *7*, 1021. [CrossRef]


extended-spectrum-β-lactamase- and AmpC-producing clinical isolates derived from companion animals and horses. *J. Antimicrob. Chemother.* **2012**, *67*, 1368–1374. [CrossRef] [PubMed]


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Broad-Spectrum Cephalosporin-Resistant** *Klebsiella* **spp. Isolated from Diseased Horses in Austria**

#### **Igor Loncaric 1,\*, Adriana Cabal Rosel 2, Michael P. Szostak 1, Theresia Licka 3, Franz Allerberger 2, Werner Ruppitsch <sup>2</sup> and Joachim Spergser <sup>1</sup>**


Received: 30 January 2020; Accepted: 15 February 2020; Published: 20 February 2020

**Simple Summary:** Broad-spectrum cephalosporin-resistant *Klebsiella pneumoniae* is considered as a serious problem for public human health. To date, only a few broad-spectrum cephalosporin-resistant *Klebsiella* have been isolated from horses. Considering the zoonotic potential of the *Klebsiella* spp., and the close relationship between man and horse, this study intended to generate data on the genetic background of broad-spectrum cephalosporin-resistant *Klebsiella* spp. isolated from horses in Austria. Overall, samples isolated between 2012 and 2019 from 1541 horses underwent bacteriological testing, resulting in 51 specimens tested positive for *Klebsiella* ssp. Antimicrobial susceptibility tests revealed that seven *Klebsiella* ssp. isolates were not only cefotaxime-resistant but also showed resistance against other classes of antibiotics so that they were considered to be multidrug-resistant. Data from whole genome sequencing and mating experiments strongly suggest that the majority of antibiotic resistance genes is encoded on plasmids in these seven multidrug-resistant *Klebsiella* ssp. Considering the potential threat when commensal *Klebsiella* inhabiting a healthy human gut acquire new antibiotic resistances due to the exchange of plasmids with multidrug-resistant *Klebsiella* ssp. from horses, further monitoring of horses and other domestic animals for the presence of broad-spectrum cephalosporin-resistant *Klebsiella*, not only in Austria but worldwide is therefore advisable.

**Abstract:** The aim of the present study was to investigate the diversity of broad-spectrum cephalosporin-resistant *Klebsiella* spp. isolated from horses in Austria that originated from diseased horses. A total of seven non-repetitive cefotaxime-resistant *Klebsiella* sp. isolates were obtained during diagnostic activities from autumn 2012 to October 2019. Antimicrobial susceptibility testing was performed. The isolates were genotyped by whole-genome sequencing (WGS). Four out of seven *Klebsiella* isolates were identified as *K. pneumoniae*, two as *K. michiganensis* and one as *K. oxytoca*. All isolates displayed a multi-drug resistant phenotype. The detection of resistance genes reflected well the phenotypic resistance profiles of the respective isolates. All but one isolate displayed the extended-spectrum β-lactamases (ESBL) phenotype and carried CTX-M cefotaximases, whereas one isolate displayed an ESBL and AmpC phenotype and carried cephamycinase (CMY)-2 and sulfhydryl variable (SHV)-type b and Temoniera (TEM) β-lactamases. Among *Klebsiella pneumoniae* isolates, for different sequence types (ST) could be detected (ST147, ST307, ST1228, and a new ST4848). Besides resistance genes, a variety of virulence genes, including genes coding for yersiniabactin were detected. Considering the high proximity between horses and humans, our results undoubtedly identified a public health issue. This deserves to be also monitored in the years to come.

**Keywords:** AmpC; ESBL; *Klebsiella pneumoniae*; antibiotic-resistance; β-lactamases; horses

#### **1. Introduction**

Among the member of the genus *Klebsiella*, broad-spectrum cephalosporin-resistant *Klebsiella* (*K.*) *pneumoniae* is frequently associated with severe nosocomial infections in humans, and due to its antibiotic-resistant traits, infections leave limited therapeutic options [1,2]. In early 2017, the World Health Organization (WHO) listed carbapenem-resistant and 3rd generation cephalosporin-resistant *Enterobacteriales* (including, e.g., *K. pneumoniae*, *Escherichia coli*, *Enterobacter* spp., *Serratia* spp., *Proteus* spp., *Providencia* spp., *Morganella* spp.) "Priority 1: Critical group" bacterial pathogens. These bacteria are in focus on the discovery and development of new antibiotics [2,3].

Today, broad-spectrum cephalosporin-resistant *K. pneumoniae* is recognized as a serious public health problem in human medicine [4,5]. Contrarily, there is still a scarcity of information on broad-spectrum cephalosporin-resistant *K. pneumoniae* and members of the genus *Klebsiella* isolated from horses and other domestic animals. To date, only a few equine broad-spectrum cephalosporin-resistant *Klebsiella* have been isolated and characterized [4,6–10]. Recent studies reported that some of the characterized resistant *K. pneumoniae* isolates of equine origin were human-associated multidrug-resistant (MDR) *K. pneumoniae* [4,8]. At present, there are no published data on the genetic background of broad-spectrum cephalosporin-resistant *Klebsiella* spp. isolated from horses in Austria. Therefore, there is a need to generate such data to understand the molecular epidemiology of these particular pathogens.

In the present study, we have characterized a collection of equine broad-spectrum cephalosporin-resistant *Klebsiella* sp. from clinical samples by multiphasic approach, including whole-genome sequencing (WGS).

#### **2. Materials and Methods**

At the Institute of Microbiology, University of Veterinary Medicine, Vienna, approximately 350 susceptibility tests are performed on clinical isolates from horses each year. During the study period (2012 until October 2019), samples of 1541 horses underwent bacteriological testing. In 51 specimens, *Klebsiella* sp. was detected, wherefrom a total of seven non-repetitive cefotaxime-resistant isolates, which were identified to the species level by matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF) mass spectrometry (Bruker Daltonik, Heidelberg, Germany), and were further analyzed. They originated from lavage (isolates 1505 and 2826), wound (isolates 2668 and 2742), fistula (isolate 1635), trachea (isolate 2341b), and feces (isolate 4545). All isolates were stored in glycerol stocks at −80 ◦C. All samples originated from non-food producing horses. All these clinical samples were received from third parties and, therefore, not subject to reporting obligations of the Ethics and Animal Welfare Commission of the University of Veterinary Medicine in Vienna.

Antimicrobial susceptibility testing was performed by agar disk-diffusion according to standards of the Clinical and Laboratory Standards Institute (CLSI) [11]. *Escherichia coli* ATCC® 25922 served as quality control strains. The following antimicrobials were used: cefotaxime, ceftazidime, aztreonam, imipenem, meropenem, gentamicin, amikacin, tobramycin, ciprofloxacin, trimethoprim-sulfamethoxazole, tetracycline, chloramphenicol, and fosfomycin (Becton Dickinson, Heidelberg, Germany). In addition, isolates were checked for extended-spectrum β-lactamase (ESBL) production by ESBL-test via agar disk diffusion [11]. Furthermore, cefoxitin (30 μg) was added to this test to detect AmpC phenotypes.

Whole-genome sequencing (WGS) was performed by isolating and sequencing bacterial DNA, as previously described [12]. *De novo* assembly of raw reads, whole genome sequencingt (WGS) data analysis, including multi-locus sequence typing (MLST) and core genome multi-locus sequence-based typing (cgMLST), were performed, as previously described [13,14].

Species identification was conducted with the JSpecies workspace using the ANIb (average nucleotide identity via Basic Local Alignment Search Tool (BLAST) analysis tool [15]. The identification of acquired resistance genes and chromosomal mutations was performed using the Comprehensive Antibiotic Resistance Database (CARD; https://card.mcmaster.ca/home) [16], as well as ResFinder 3.2 (https://cge.cbs.dtu.dk/services/ResFinder/) [17] were used. eBURST (Based Upon Related Sequence Types) analysis (a plugin at https://bigsdb.pasteur.fr/) was conducted to identify clonal complexes (CCs), defined as groups of two or more independent isolates sharing identical alleles at six or more loci.

The presence of plasmids was determined using PlasmidFinder 1.3 available from the Center for Genomic Epidemiology web server (http://www.genomicepidemiology.org/) [18]. Probability Prediction of the location of a given antibiotic resistance gene was achieved by applying mlplasmids trained on *K. pneumonia* [19]. Posterior probability scores >0.7 and a minimum contig length of 1000 bp indicate that a given contig is plasmid-derived.

Mating experiments were conducted by conjugation as well as transformation, as previously described [20]. Variable regions of class 1 and class 2 integrons were determined by PCR [20]. The quinolone resistance-determining regions (QRDR) of *gyrA* and *parC* in ciprofloxacin-resistant isolates were amplified by PCR and sequenced [21].

The presence of virulence genes was examined by using the virulence allele library from the Institute Pasteur BIGSdb database for *K. pneumoniae* (http://bigsdb.pasteur.fr/klebsiella).

This whole-genome shotgun project has been deposited in DDBJ/EMBL/GenBank under the project number PRJNA600879. Raw sequence data for each strain were deposited under Sequence Read Archive (SRA) accession numbers SRR10899218 to SRR10899224.

#### **3. Results**

Four out of seven cefotaxime-resistant *Klebsiella* isolates were identified as *K. pneumoniae*, two as *K. michiganensis*, and one as *K. oxytoca* (Table 1). All but one isolate displayed the ESBL phenotype, whereas one isolate displayed an ESBL and AmpC phenotype. Besides cefotaxime, all *K. pneumoniae* isolates were resistant to ceftazidime, and one isolate additionally to aztreonam. All examined isolates were resistant to gentamicin and tobramycin. None of the analyzed isolates was resistant to carbapenems and amikacin. Five isolates were resistant to tetracycline, doxycycline, and chloramphenicol, whereas six were resistant to trimethoprim-sulfamethoxazole. All *K. pneumoniae* isolates were resistant to ciprofloxacin, and one isolate to fosfomycin (Table 1). Hence, all examined isolates were considered to be multidrug-resistant [22]. The detection of resistance genes reflected well the phenotypic resistance profiles of the examined isolates (Table 1). In two ciprofloxacin-resistant *K. pneumoniae* isolates, beside fluoroquinolone resistance genes *oqxA*, *oqxB*, *qrnB1*, and *aac(6 )-Ib-cr*, mutations in the quinolone resistance-determining regions (QRDRs) of the genes *gyrA* and *parC* were observed (Table 1). Three isolates, both *K. michiganensis* isolates and the *K. oxytoca* isolate contained a class 1 integron with a variable part of ca. 1.7 kb in size, which harbored an *aadA5* and a *dfr17* cassette.

In total, ten different plasmids IncFIA(HI1), IncFIB(K), IncFIB(pHCM2), IncHI1A, IncHI1B(R27), IncI1, IncN, IncQ1, IncR, and Col440l were identified (Table 2). They shared between 92.11 and 100% DNA similarity with corresponding reference sequences. A *K. michiganensis* isolate and two *K. pneumoniae* isolates carried IncFIA(HI1), IncFIB(pHCM2), IncHI1A, IncHI1B(R27), and IncQ1. The *K. oxytoca* isolate carried IncI1 and IncN, whereas a *K. pneumoniae* carried IncN and IncR and another *K. michiganensis* IncFIB(K). According to mlplasmids analyses, the majority of resistance genes might be located on plasmids, especially all *bla*CTX, *bla*TEM, and *bla*OXA genes as well as all detected genes for resistance against aminoglycosides, trimethoprim/sulfamethoxazole, or chloramphenicol (Table 1).

Among virulence factors, *K. pneumoniae* type 3 fimbriae encoded by *mrk* operon genes as well *iutA* (aerobactin siderophores receptor) were detected in all *K. pneumoniae* isolates, whereas genes coding for yersiniabactin (*ybt*) were detected in only one isolate (Table 3).


Characteristicsofsevenexaminedcefotaxime-resistant*Klebsiella*isolates.


**Table 1.** *Cont.*

mutation in *gyrA* and *parC* of quinolone

resistance-determining

 region (QRDR).


**Table 2.** Identified plasmids in *Klebsiella* isolates.

**Table 3.** Identified virulence factors in four *K. pneumoniae* isolates. Numbers correspond to the exact alleles detected.


All four *K. pneumoniae* isolates belonged to different sequence types (ST) and cgMLST complex type (CT) (ST147-CT1202, ST307-CT4645, ST1228-CT4644 and a new ST4848-4643). These ST belonged to 4 clonal complexes: CC147 (ST147), CC37 (ST1228), CC307 (ST307) and CC702 (new ST4848). The minimum number of allelic differences between the isolates was 3686.

#### **4. Discussion**

The present study demonstrates that broad-spectrum cephalosporin-resistant members of the genus *Klebsiella* are present in the Austrian horse population, although the prevalence in clinical samples seems to be low. These findings are in accordance with previous studies describing the presence of these particular bacteria in horse populations of other countries [4,6–10]. Moreover, a previous study carried out in 2018 on clinical samples from Austrian patients reported 8.4% of *K. pneumoniae* isolates as resistant to third-generation cephalosporins [23].

In the present study, the most prevalent cefotaximase type was CTX-M-1 carried by all three *Klebsiella* species identified; this β-lactamase is commonly associated with *Enterobacteriales* from livestock [24]. CTX-M-15, the dominating cefotaximase, is considered the most common ESBL in *K. pneumoniae* from humans and animals worldwide [4]. To the best of our knowledge, the present study describes for the first time CTX-M-1 producing *K. michiganensis*, a close relative of *K. oxytoca*. One *K. pneumoniae* isolate that displayed both the AmpC and ESBL phenotype carried three different β-lactamases, *bla*CMY2, *bla*SHV11, and *bla*TEM-1. *bla*CMY2 was carried by an IncI1 conjugative plasmid. *K. pneumoniae* carrying plasmid-borne AmpC cephalosporinases (pAmpC) is a rare observation [6,25].

Another important observation is the co-existence of an arsenal of virulence factors and antibiotic resistance characters in one*K. pneumoniae*isolate (ST1228-CT4644). This isolate carried the yersiniabactin locus. Yersiniabactin is a siderophore, which is strongly associated with invasive clinical manifestations in humans [26]. Another siderophore, aerobactin, as well as type 3 fimbriae, which were detected in all *K. pneumoniae* isolates, may enhance colonization and adherence to host cells, invasiveness, and biofilm formation [27].

Among *K. pneumoniae* isolates examined, four different sequence types belonging to four different clonal complexes were identified. Two of these STs, ST147, and ST37, have been recognized as high-risk epidemic multiresistant human-associated clonal lineages [5]. ST147-CC147 is a human-related clone notorious for its multi-drug resistant character and harboring different β-lactamases, including carbapenemases [5]. Recently, this particular clone has emerged in companion animals [5,28]. In contrast, ST1228 has only one entry in the Institut Pasteur MLST database (http://bigsdb.pasteur.fr) and to our knowledge, had never been associated with horses. ST1228 belongs to CC37 whose predicted founder is ST37. *K. pneumoniae* ST37 isolates have been associated with different resistance properties, including carbapenem and colistin resistance, and were isolated from humans and animals [5,29]. One fecal isolate analyzed in the present study belonged to ST307-CC307. ST307 is a relatively new but highly successful pandemic clone, which was previously recovered from human patients, and recent data suggest a multi-drug resistant character of this clone [30]. β-lactamase producing *K. pneumoniae* ST307 has also been detected among different animals [31]. In the present study, a new sequence type, ST4848, belonging to the clonal complex CC702 (predicted founder ST702), has been identified by eBURST analysis. CC702 is a rare clone that has never been associated with broad-spectrum cephalosporin-resistant *K. pneumoniae* of equine origin. Data generated in this study (mating experiments, PlasmidFinder analysis, posterior probability plasmid analysis) strongly suggest that the majority of resistance genes are plasmid-borne. All identified replicons (IncFIA(HI1), IncFIB(K), IncFIB(pHCM2), IncHI1A, IncHI1B(R27), IncI1, IncN, IncQ1, IncR) are considered as vehicles of *bla*CTX-M-15 and *bla*CTX-M-1 dissemination in humans and animals [5,32].

#### **5. Conclusions**

Even though the overall prevalence of broad-spectrum cephalosporin-resistant *Klebsiella* sp. among specimens of equine origin in Austria appears to be low, the proportion of broad-spectrum cephalosporin-resistant *Klebsiella* spp. vs. non-resistant *Klebsiella* spp. is worth mentioning, since commensal *Klebsiella* spp. can acquire antimicrobial resistance. As such, the broad-spectrum

cephalosporin-resistant *Klebsiella* spp. especially in combination with other resistance properties, are of special clinical importance because of dramatically narrowing the possibility of antibiotic treatment. Due to the regular contact and proximity between horses and humans monitoring horses for the presence of cephalosporin-resistant *Klebsiella* spp. is advisable in order to prevent further spread of these zoonotic agents.

**Author Contributions:** Conceptualization, I.L., T.L. and J.S.; methodology, I.L., M.P.S. and W.R.; software, M.P.S.; validation, I.L. and M.P.S.; formal analysis, I.L., A.C.R. and W.R.; investigation, I.L., A.C.R., M.P.S. and W.R.; resources, F.A. and J.S.; data curation, I.L., A.C.R. and T.L.; writing—original draft preparation, I.L.; writing—review and editing, I.L., A.C.R., M.P.S., T.L., F.A., W.R. and J.S.; project administration, I.L., F.A., W.R., J.S.; funding acquisition, W.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** Part of the sequencing-work was funded by a grant awarded under the "MedVetKlebs" Horizon 2020 Framework Programme H2020-SFS-2016-2017 (H2020-SFS-2017-1). Open Access Funding by the University of Veterinary Medicine Vienna.

**Acknowledgments:** We thank the team of curators of the Institut Pasteur MLST and whole genome MLST databases for curating the data and making them publicly available at http://bigsdb.pasteur.fr. We would also like to express our thanks to Michael Steinbrecher, Anna Stöger, and Barbara Tischler for technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
