Molecular characterization and antimicrobial resistance of potentially human‐pathogenic Escherichia coli strains isolated from riding horses

Background Transmission of antimicrobial resistant and virulent Escherichia coli (E. coli) from animal to human has been considered as a public health concern. This study aimed to determine the phylogenetic background and prevalence of diarrheagenic E. coli and antimicrobial resistance in healthy riding-horses in Iran. In this research, the genes related to six main pathotypes of E. coli were screened. Also, genotypic and phenotypic antimicrobial resistance against commonly used antibiotics were studied, then phylo-grouping was performed on all the isolates. Results Out of 65 analyzed isolates, 29.23 % (n = 19) were determined as STEC and 6.15 % (n = 4) as potential EPEC. The most prevalent antimicrobial resistance phenotypes were against amoxicillin/clavulanic acid (46.2 %) and ceftriaxone (38.5 %). blaTEM was the most detected resistance gene (98.4 %) among the isolates and 26.15 % of the E. coli isolates were determined as multi-drug resistant (MDR). Three phylo-types including B1 (76.92 %), A (13.85 %) and D (3.08 %) were detected among the isolates. Conclusions Due to the close interaction of horses and humans, these findings would place emphasis on the pathogenic and zoonotic potential of the equine strains and may help to design antimicrobial resistance stewardship programs to control the dissemination of virulent and multi-drug resistant E. coli strains in the community.


Background
Over time, horses have been bred by humans for meat, leisure and sport. Today, horses are known as important companion animals and have become popular for horseback riding. Horse-riding clubs are a significant platform for human-horse interaction due to close contact between horses, horse handlers, horse riders and spectators [1]. Horses as companion animals could be considered as a potential reservoir of microbial agents which cause infections and complications in various hosts such as human. Among these microorganisms, some strains of Escherichia coli (E. coli) possess antimicrobial resistance (AMR) and virulence determinants which could be transmitted by direct or indirect contact [2].
Many strains of E. coli have intrinsic or/and acquired antimicrobial resistance which should be addressed as a significant threat to public health. Resistant E. coli could be selected among gut microbiota due to use of antimicrobial agents. Epidemiologically, antimicrobial resistant E. coli strains and their AMR determinants may be transferred from animal to human by horizontal transmission of AMR genes or clonal transfer of resistant strains via direct contact, indirect contact and consumption of fecal contaminated food.
Based on phylogenetic assessments, E. coli have been classified into eight phylo-groups including A, B1, B2, C, D, E, F and Escherichia cryptic clade I, by a PCR method [5]. The gut commensal E. coli strains predominantly belong to group A or B1, however, antimicrobial susceptible strains usually belong to the B1 rather than groups A and D [6,7]. E. coli phylogenetic background is of importance for understanding the relationship between strains, antimicrobial resistance and disease [8].
To the best of our knowledge, there are no comprehensive study on virulence, antimicrobial resistance and phylogenetic analysis of equine E. coli isolates in Middle East. Hence, this study was designed to assess these variables in healthy riding horses in Iran to assess the potential risks of these animals for public health. The results could help to understand the public health risks associated with horse-riding clubs and transmission of antimicrobial resistance and virulence factors of E. coli from horses to humans.

Sampling and E. coli isolation
In this study, 65 rectal swabs were collected from healthy horses (n = 65) from five riding horse clubs during July to September, 2018, in Kerman province, southeast of Iran. Sterile saline moistened swabs were inserted 10 cm into the rectum and carried to veterinary microbiology laboratory in individual tubes containing Amies transport Medium (Merck, Germany) within 12 h. For E. coli isolation, all samples were cultured on MacConkey agar (Merck, Germany) and incubated at 37°C for 24 h. One presumptive E. coli colony was selected from each sample and confirmed via biochemical tests. One confirmed E. coli isolate from each sample was subjected to phenotypic resistance analysis and genetic assessments.

Phenotypic antimicrobial resistance assessment
In this study we evaluated antimicrobial resistance of isolates to seven antimicrobial agents which are the drugs of choice in the treatment of equine bacterial infections by disk diffusion method; including amoxicillin/ clavulanic acid (20/10 µg), cefazolin (30 µg), ceftriaxone (30 µg), amikacin (30 µg), streptomycin (10 µg), gentamicin (10 µg) and trimethoprim/sulphamethoxazole (1.25/23.7 µg). The horse-specific/human breakpoints have been used to evaluate the results of disk diffusion method; the diameter of growth inhibition zones have been measured and the E. coli isolates were determined as resistant, intermediate and susceptible groups. According to CLSI VET08, human breakpoints are considered to provide zones of inhibition when there are no veterinary breakpoints available for some antimicrobial agents for all animal species [9,10]. E. coli ATCC 25922 was used for quality control of the test (Table 1).
Phylogenetic background of all isolates was determined using the PCR-based method explained by Clermont et al. (2013). In this scheme an E. coli strain could be classified into one of the phylo-types A, B1, B2, C, D, E, F and cryptic clades I to V [5] (Table 2).
All PCR methods were carried out as simplex in 25 µl final reaction volume including 3 µl DNA extract, 0.3 µM of each primer, 12.5 µl 2× Taq DNA Polymerase Master Mix RED (Ampliqon, Denmark) and sterile distilled water up to 25 µl. PCR products were loaded on an 1.5 % electrophoresis agarose gel containing Green Viewer stain (ParsTous, Iran) and imaged by a GelDoc 1000 (Vilber Lourmat, France).

Statistical analysis
For descriptive statistical analysis, all data about presence or absence of studied factors in each strain were imported into SPSS (SPSS 19; IBM) program as binomial variables; prevalence, 95 % confidence level and P value were calculated.

Results
In the present study, sixty five E. coli strains isolated from 65 horses have been evaluated. The most prevalent antimicrobial resistance phenotype was against the two β-lactam antibiotics, amoxicillin/clavulanic acid (46.2 %) and ceftriaxone (38.5 %), and the other antimicrobial resistance (AR) phenotypes were observed in less than 25 % of the E. coli isolates. In agreement with the phenotypic findings, the gene related to β-lactam resistance (bla TEM ) was the most detected gene (98.4 %) among the isolates (Table 3).
Twenty-three phenotypic AR patterns were recognized in this study. No significant difference was observed in prevalence of AR patterns (p-value > 0.05); the frequency of the patterns was in the range of 1 to 4. According to studied antimicrobial agents, 26.15 % (n = 17) of the isolates were determined as multi-drug resistant; resistant strains to at least three antimicrobial agents from three different antimicrobial classes are defined as MDR [17]. Most of MDR isolates belonged to B1 phylo-group (12/ 17; 70.58 %), followed by A (3/17), Unknown (1/17) and D (1/17) phylo-types (Tables 3 and 4).
Among the resistant VG-positive isolates (n = 14), five isolates were phenotypically multidrug resistant. All of the VG-positives harbored at least one of screened AR genes excluding one; three different AR gene profiles were identified including bla TEM , bla TEM /sulII, bla TEM / bla CTX−M in which bla TEM was the most prevalent significantly (p-value < 0.0001). All VG + /AR-gene + isolates belonged to B1 except one belonging to A phylo-type (Table 6).
In this study, sixty one E. coli isolates have been distributed among three phylo-types including B1 (76.92 %), A (13.85 %) and D (3.08 %). Four (6.15 %) E. coli isolates could not be classified into the phylo-groups according to Clermont scheme and named as unknown (U). Similarly, B1 was the most prevalent phylo-type in virulent/non-virulent and resistant/non-resistant isolates; no significant difference (p-value > 0.05) has been observed in phylogenic distribution patterns of virulent and resistant isolates (Tables 3 and 4).

Discussion
In the current study, molecular pathotyping of equine E. coli isolates showed that more than one-third of them belonged to one of the diarrheagenic E. coli pathotypes including STEC and EPEC; the most prevalent pathotype was STEC (more than one-fourth of the isolates) followed by potential EPEC (less than 5 %). A few studies have reported the pathotypes in horses which mostly revealed low prevalence of them; Kennedy et al. (2018) in Ireland showed that none of the equine E. coli isolates obtained from 83 fecal samples were STEC or EPEC [18]. In the USA, a very low STEC prevalence, one from 242 equine E. coli isolates, has been reported [19]. Also Pichner et al. (2005) found only one STEC isolate among the 400 screened horse fecal samples in Germany [20].  16.7 % and 11.7 % frequencies, respectively; the sample size in the two latter studies were small (less than 20 horses) [21,22]. Chandran et al. (2013) have screened E. coli isolates from 11 different host sources and revealed that EPEC had the highest prevalence in horses (50 %) [23]. Despite the low STEC prevalence in horses, exposure to horse feces has been reported as a significant risk factor for clinical cases of STEC infections in humans [24,25]. STEC pathotypes are usually detected from healthy horses and no clinical features have been observed in these cases. In a research in USA, positive horses for STEC were kept on farms containing ruminants [19]. Generally, ruminants are considered as the primary reservoir of the intestinal pathotypes; equids are not the main source of STEC and EPEC and they are known as spillover hosts, the secondary species host exposed to the STEC through close contact with ruminants or feeding materials contaminated with ruminant manure [21].   Interestingly, all detected pathotypes of this study belonged to the horses with no history of ruminant direct contact. Most STECs of this research were only positive for stx1 gene and some possessed stx1 and stx2 simultaneously. It is believed that stx1 and stx2 positives could cause more severe cases of STEC infections in human [26,27]. The STEC strains which are positive for eae gene could potentially induce the attaching and effacing lesions in human. It has to be considered that the detection of the stx and eae genes is not enough to determine the pathogenicity and virulence of the E. coli strains recovered from animals in the human. Therefore, further phenotypic evaluations in laboratory animal models and intestinal cell lines are considered essential.
The occurrence of antimicrobial resistant E. coli in companion animals has drawn attention as a public health issue in the last decade [28]. Resistance to highly prescribed antimicrobial agents such as betalactames, aminoglycosides and sulphonamides has been studied in equine E. coli strains [29].
Our phenotypic results revealed that the prevalence of all the resistance phenotypes were less than 50 %. The highest prevalence of antimicrobial resistance has been observed against amoxicillin/clavulanic acid followed by ceftriaxone. Penicillins, sulfonamides and aminoglycosides are amongst the most commonly used antimicrobial classes in equine medicine for various conditions such as respiratory, digestive and pyogenic infections [30]. In a study in South Africa, resistance rate to ceftriaxone and amikacin were similar to the ones found in this work, while the prevalence of resistant isolates to amoxicillin/clavulanic, trimethoprim/sulphamethoxazole and gentamicin were higher than the current study [31]. Fortunately, our resistance rate to trimethoprim/sulphamethoxazole was notably low while other studies worldwide reported higher frequencies. A wide range of resistance against streptomycin and gentamicin in E. coli equine isolates has been reported from various studies, explaining that MDR in this study is rather low (less than 15 %). Variation in prevalence of antimicrobial resistance may be due to evaluation method, sample size, season, antibiotic prescription patterns, microbial population type of gastrointestinal microflora and exposure to antimicrobial resistance determinants [32]. Multidrug resistant bacteria can lead to complicated infections in the susceptible hosts and are addressed as a major public health issue [33]. According to MDR definition, a multidrug resistant bacterium is non-susceptible to at least one antimicrobial agent from three or more different antibiotic categories [17]. In the current study, more than one-fourth of the E. coli isolates and a considerable number of VG-positives were resistant against multiple antibiotics which are highly prescribed in human and equine medicine. This is less than the reported MDR prevalence in Kennedy et al. (2018) and de Lagarde et al.
(2020) studies [18,34]. Dissemination of MDR bacteria may cause the spread of nosocomial and communityacquired infections which could lead to rising antibiotic use, healthcare costs, morbidity and mortality [35]. Two main mechanisms have been proved for MDR; accumulation of several resistance genes by the bacteria and increased expression of resistance genes [36]. The dissemination of antimicrobial resistance is mainly associated with AR genes which are mostly located on mobile genetic elements. Thus, detection of resistance genes in bacteria may help to understand the resistance transmission and improvement of antibiotic-therapy strategies. In this study bla TEM was the most frequent AR gene, identified significantly higher (P < 0.  [18,37,38]. The next most prevalent genes were associated with resistance against sulfonamides including sulII (10.7 %) and sulI (1.5 %) which were considerably lower than the 57 % prevalence reported by Kennedy et al. (2018) in Ireland [18]. For the remaining genes, our results are comparable with the study in Jordan; the prevalence of our screened genes were between 0 and 6 % while the frequencies in the Gharaibeh et al. (2020) study were more than 10 % [38]. The diversity in prevalence rates may be due to the use of different methods, genotypic variety of E. coli populations and antibiotic exposure in horse in different countries.
In the current study, phylogenetic assessment of the equine isolates showed no relationship among virulence, resistance and phylogenetic background; resistant/nonresistant and virulent/non-virulent E. coli strains frequently belonged to B1 phylo-group. In agreement with our results, many studies around the world have reported B1 as the predominant phylo-type in equine E. coli isolates [34,39,40]. Conversely, Sukmawinata et al. (2019), reported B2 as the most common phylogenetic group among extended-spectrum β-lactamase-producing E. coli isolates from healthy thoroughbred race horses in Japan [41]. Phylo-typing of E. coli strains help to determine the evolutionary relationships among the microorganisms and is a fundamental issue in microbial studies. High dissemination of B1 in all isolate types could be due to various reasons such as nutrition, host species, sex, age, body mass, climate, geographic location and the combination of gut microflora [42].

Conclusions
In this study, a significant number of equine isolates belonged to one of the diarrheagenic E. coli pathotypes including STEC and potential EPEC, in which the most prevalent pathotype was STEC. Equids are known as spill-over hosts when exposed to the STEC through close contact with ruminants or feeding materials contaminated with ruminant manure. All detected pathotypes of this study belonged to the horses with no history of ruminant contact, which indicates that horses may take a role as a potential reservoir in spreading virulent pathotypes. Moreover, some VG-positive isolates were recognized as MDR. All the sampled animals in this study were riding horses with close contact to human including horse riders, club personnel and spectators. This represents the pathogenic and zoonotic potential of the equine strains in human medicine and would place emphasis on the design of antimicrobial resistance stewardship programs to control the dissemination of virulent and multi-drug resistant E. coli strains in the community.