Antimicrobial Resistance in Vaginal Bacteria in Inseminated Mares

Antimicrobials are added to semen extenders to inhibit the growth of bacteria that are transferred to the semen during collection. However, this non-therapeutic use of antimicrobials could contribute to the development of antimicrobial resistance. The objective of this study was to determine changes in the antibiotic susceptibility of vaginal microbiota after artificial insemination. Swabs were taken from the vagina of 26 mares immediately before artificial insemination and again 3 days later. Bacteria isolated from the vagina at both time points were subjected to antibiotic susceptibility testing and whole-genome sequencing. In total, 32 bacterial species were identified. There were increases in the resistance of Escherichia coli to trimethoprim (p = 0.0006), chloramphenicol and (p = 0.012) tetracycline (p = 0.03) between day 0 and day 3. However, there was no significant effect of exposure to antibiotics in semen extenders with respect to the resistance of Staphylococcus simulans and Streptococcus equisimilis (p > 0.05). Whole-genome sequencing indicated that most phenotypic resistance was associated with genes for resistance. These results indicate that the resistance patterns of vaginal bacteria may be affected by exposure to antibiotics; therefore, it would be prudent to minimize, or preferably, avoid using antibiotics in semen extenders.


Introduction
Antimicrobial resistance (AMR) has become a serious global concern that is accelerated by the misuse and overuse of antibiotics, as well as by inadequate infection prevention and control [1]. Although efforts are being made to reduce non-therapeutic uses of antimicrobials, one major use of antibiotics is still in animal breeding. The reproductive biotechnology artificial insemination (AI) was developed originally to prevent disease transmission when animals met for breeding, but it also facilitated faster genetic gain than could be achieved with natural mating. It is now a global phenomenon with millions of semen doses being traded around the world each year. For this international trade in semen, most countries have strict requirements regarding the health of semen donors, and they also stipulate the antibiotics that will be included in semen extenders. These antimicrobials are aimed at preventing the spread of bacterial diseases to inseminated females and the deterioration of sperm quality during storage [2].
There are several reports (reviewed recently [3]) concerning the AMR of bacteria isolated from uterine swabs, uterine lavage, vaginal swabs and clitoral swabs from the equine reproductive tract, mostly in cases of clinical disease or fertility problems. Briefly, resistance among isolates from the reproductive tract of mares has been reported in several countries including France [4], Sweden [5], India [6,7], Italy [8][9][10], Germany [11], the US [12,13], Slovakia [14] and Turkey [15]. Few detailed studies have been conducted on AMR in the reproductive microflora in healthy horses, and no studies have been conducted to investigate the effect of antimicrobial substances in semen extenders on vaginal flora. The Man, Rosa and Sharpe agars (MRS-agar) (Oxoid, Basingstoke, UK), which were incubated anaerobically at 25 ± 1 • C for 5 days. Bacterial colonies of different macromorphologies from the initial culture were re-cultured on two blood agar plates (for aerobic and anaerobic culture conditions) and incubated for 24 to 48 h at 37 ± 1 • C to obtain a pure culture. The colonies from the pure cultures were stored in cryotubes with brain heart infusion (BHI) broth (CM1135; Oxoid, Basingstoke, UK) with 15% glycerol at −70 • C for subsequent antimicrobial-resistance testing and whole-genome sequencing.
The bacteria were identified at the species level by the use of matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) (Bruker Daltonics, Billerica, MA, USA). The mass spectrum of bacterial isolates was compared with those of known bacterial strains in the database (Bruker Daltonics, Billerica, MA, USA). Score values between 2.0 and 3.0 were considered accurate at both genus and species levels, whereas score values between 1.7 and 2.0 were considered reliable only at the genus level.

Antimicrobial Susceptibility Testing
The resistance to selected antimicrobial substances was determined for most commonly isolated bacteria present in samples both before (D0) and after insemination (D3) in the same mare. The isolates were cultured on horse blood agar directly from the cryotubes and incubated for 24 h at 37 • C. They were re-cultured once to ensure a pure culture.
The resistance panel used for Staphylococcus spp. and Streptococcus spp. was Thermo Scientific™ Sensititre ™ STAFSTR; for Enterococcus faecalis, it was Thermo Scientific™ Sensi-titre™ EUVENC; for E. coli, it was Thermo Scientific™ Sensititre™ EUVENSEC (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Susceptibility to selected antimicrobial substances was assessed with VetMIC™ panel analysis systems: Camp EU, version 2013-10 (SVA, Uppsala, Sweden), determining the antimicrobial minimum inhibitory concentration (MIC) by broth microdilution following the standards of the Clinical and Laboratory Standards Institute [17].
A purity check and testing of concentration were performed on the culture of the final inoculum after placing the inoculum in the resistance panel on the horse blood agar, which was incubated at 37 • C for 24 h. Epidemiological cut-off (ECOFF) values for determining susceptibility were obtained from the European Committee on Antimicrobial Susceptibility Testing (EUCAST, www.eucast.org/mic_distributions_and_ecoffs, accessed on 20 February 2023). The ECOFF values classify isolates with acquired reduced susceptibility as "non-wild type" In this paper, non-wild-type isolates are called "resistant", in agreement with the Swedish Veterinary Antibiotic Resistance Monitoring report [18].

Whole-Genome Sequencing
Whole-genome sequencing (WGS) was performed to determine whether the bacterial isolates of different antimicrobial susceptibilities from the same mares sampled at both time points were likely to be of the same strain or not and if phenotypic resistance was associated with genes for resistance. Resistant bacteria were subjected to WGS, including E. coli, Streptococcus equisimilis and Staphylococcus simulans. All isolates from the individual mares with at least one isolate phenotypically resistant to any antibiotics at both points were selected for WGS. WGS was performed on 127 E. coli, 26 Streptococcus equisimilis and 4 Staphylococcus simulans isolates, which showed increased or decreased resistance to antibiotics according to antimicrobial susceptibility testing. "Increased resistance" means that the proportion of bacteria from an individual mare showing resistance was higher later in the sampling sequence, whereas "decreased resistance "means that the proportion of bacteria showing resistance was lower later in the sampling sequence.
Isolates were subcultured twice on horse blood agar plates and from a single colony prior to DNA extraction to ensure pure culture. DNA was extracted with the EZ1 DNA Tissue Kit (Qiagen, Hilden, Germany) according to the protocols for Gram-negative and Gram-positive bacteria, respectively, with the exception that 75U of mutanolysin (Sigma-Aldrich, St. Louis, MO, USA) was added, and the lysis time extended to 4 h for S. equisimilis. The extraction was Sequencing libraries were constructed with Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA, USA), and the quality was assessed by the HS DNA ScreenTape Analysis D1000 (Agilent Technologies, Inc., Santa Clara, CA, USA) on a 4150 TapeStation (Agilent Technologies, Inc.). Libraries were quantified by Qubit ds DNA HS kit (Invitrogen). WGS of the prepared libraries was conducted on the Illumina NextSeq 500 system (Illumina Inc., San Diego, CA, USA) using the Mid Output kit V2.5 with 2 × 150-bp paired-end reads (Illumina Inc).

Statistical Analysis
The difference between antimicrobial resistance results from vaginal isolates before and after contact with semen extenders was analyzed using the chi-squared test. If the criteria for the chi-squared test were not fulfilled, a Fisher's exact test was used instead. A probability level of p < 0.05 was considered statistically significant.

Bacterial Analyses
In total, 985 isolates of 32 different bacterial species were identified from the 26 mares. E. coli was the most common bacterial species isolated from 23 of 26 mares. The second and third most-isolated bacterial genera were Streptococcus spp. and Acinetobacter spp., which were identified in 24 of 26 mares and 17 of 26 mares, respectively (Table 1).

Antimicrobial Susceptibility
The resistance of Escherichia coli against trimethoprim (p = 0.0006), chloramphenicol (p = 0.012) and tetracycline (p = 0.03) increased from D0 to D3. However, E. coli resistance to other antibiotics was not different after exposure to semen extenders. Furthermore, there was no association between exposure to semen extenders and the resistance of Staphylococcus simulans and Streptococcus equisimilis (p > 0.05).
Resistance to at least 1 of 14 antibiotics was found in 284 of 520 E. coli isolates. Twentynine isolates were multidrug-resistant (resistant to at least three classes of antibiotics). There was a significant increase in resistance to trimethoprim, chloramphenicol and tetracycline after exposure to semen extenders with antibiotics. However, resistance was also found against sulfamethoxazole, ampicillin, azithromycin and tigecycline, but differences between time points were not significant. All E. coli isolates were susceptible to cefotaxime, ceftazidime, ciprofloxacin, colistin, gentamicin, meropenem and nalidixic acid ( Table 2). No correlation was found between the resistance of Streptococcus dysgalactiae subsp. equisimilis before and after exposure to semen extenders with antibiotics. Various Streptococcus dysgalactiae subsp. equisimilis isolates were resistant to three of eight antibiotics, including tetracycline, erythromycin and nitrofurantoin. All Streptococcus dysgalactiae subsp. equisimilis isolates were susceptible to cefalothin, clindamycin, oxacillin, penicillin and trimethoprim/sulfamethoxazole. For some antibiotics, cefoxitin, enrofloxacin, fusidic acid and gentamicin, sensitivity or resistance could not be assessed since no ECOFFs were mentioned in EUCAST (Table 3).
All Enterococcus faecalis isolates were susceptible to the six tested antibiotics (ampicillin, ciprofloxacin, gentamicin, linezolid, teicoplanin and tigecycline). For some antibiotics (chloramphenicol, daptomycin, erythromycin, quinupristin/dalfopristin, tetracycline and vancomycin), sensitivity or resistance could not be assessed since no ECOFFs were given in the EUCAST (Table 4). All tested Streptococcus equi subsp. zooepidemicus isolates were susceptible to clindamycin, erythromycin, nitrofurantoin, oxacillin, penicillin, tetracycline and trimethoprim/sulfamethoxazole. For some antibiotics, cefalotin, cefoxitin, enrofloxacin, fusidic acid and gentamicin sensitivity or resistance could not be assessed since no ECOFFs were given in the EUCAST (Table 5).
Resistance to at least one of seven antibiotics was detected in different Staphylococcus simulans isolates, including oxacillin (25%), penicillin (50%), fusidic acid (50%) and trimethoprim/sulfamethoxazole (50%). One isolate was multidrug resistant. There was no association between exposure to antibiotics via artificial insemination and change in resistance. All Staphylococcus simulans isolates were susceptible to erythromycin, gentamicin and tetracycline. For some antibiotics, cefalotin, cefoxitin, clindamycin, enrofloxacin and nitrofurantoin, sensitivity or resistance could not be assessed since no ECOFFs were given in the EUCAST (Table 6).

Whole-Genome Sequencing
Whole-Genome Sequencing was used to decide whether E. coli, Streptococcus equisimilis and Staphylococcus simulans isolates of different antimicrobial susceptibility from the same mares sampled at both time points were likely to be of the same strain or not and if phenotypic resistance was associated with genes for resistance.
One hundred twenty-seven E. coli isolates from inseminated mares before (D0) and after (D3) insemination, were allocated into fourteen clusters by cgMLST analysis (Figure 1). There was, at most, a one-allele difference between isolates within each cluster, except for Cluster 2 and 13 where there was a three-allele difference. Therefore, isolates within each cluster were highly genetically related, i.e. the same strain. Nine clusters consisted of isolates from one mare (Cluster 3, 6, 7, 8, 9, 10, 11 and 14), whereas five clusters consisted of isolates from 2 to 3 mares (Cluster 1, 2, 4 5 and 13 (Figure 1)). The finding of isolates from different mares in the same clusters suggests a possible spread within the herd.
Twenty-six Streptococcus equisimilis isolates from inseminated mares before (D0) and after (D3) insemination, were subjected to cgMLST analysis, and a minimum spanning tree (MST) was generated (Figure 2a). There were 18 allelic differences between 12 isolates from Horse H and 3 isolates from Horse J, but since the developed cgMLST scheme lacked a validated clustering distance threshold, it was not possible to determine whether the isolates belonged to 2 different clusters or 1 and the same (Figure 2a). Regardless, there were no allelic differences between isolates from each horse in this case; hence, they were of the same strain. Eleven isolates from Horse I and J clustered together with a maximum of two allelic differences, and even if the developed cgMLST scheme lacked a validated clustering distance threshold, it is likely that those isolates were genetically closely related; i.e. they were the same strain since the differences were very few (Figure 2a). The finding of genetically related isolates in two mares (Horse I and J) indicates a possible spread within the herd, as seen in E. coli isolates.
Four Staphylococcus simulans isolates from an inseminated mare (Horse K) before (D0) and after (D3) insemination, were subjected to cgMLST analysis, and a minimum spanning tree (MST) was generated (Figure 2b). There were no allelic differences between isolates P868 and PM194 and P869 and PM193, respectively; hence, isolates within each pair were genetically closely related, i.e. the same strain. Although no defined cluster distance threshold was available for this cgMLST scheme, the allelic distance between the two pairs was very large, and it is therefore unlikely that they are genetically closely related (Figure 2b).
Twenty-nine of the E. coli isolates were multiple drug resistant (MDR); twenty of these isolates belonged to Cluster 2 (Horse C and G), with one isolate each belonging to Cluster 4 (Horse B), Cluster 6 (Horse E), Cluster 7 (Horse D) and Cluster 14 (Horse B).  White fields denote the range of dilutions tested for each antibiotic and vertical bold lines indicate cut-off values used to define resistance. Grey fields denote the range outside of dilutions tested. MICs equal to or lower than the lowest concentration tested are given as the lowest tested concentration. * Significant association between antibiotics exposure and resistance results (p < 0.05).  White fields denote the range of dilutions tested for each antibiotic, and vertical bold lines indicate cut-off values used to define resistance. Grey fields denote the range outside of dilutions tested. MICs equal to or lower than the lowest concentration tested are given as the lowest tested concentration.   White fields denote the range of dilutions tested for each antibiotic, and vertical bold lines indicate cut-off values used to define resistance. Grey fields denote the range outside of dilutions tested. MICs equal to or lower than the lowest concentration tested are given as the lowest tested concentration.   White fields denote the range of dilutions tested for each antibiotic, and vertical bold lines indicate cut-off values used to define resistance. Grey fields denote the range outside of dilutions tested. MICs equal to or lower than the lowest concentration tested are given as the lowest tested concentration.   White fields denote the range of dilutions tested for each antibiotic, and vertical bold lines indicate cut-off values used to define resistance. Grey fields denote the range outside of dilutions tested. MICs equal to or lower than the lowest concentration tested are given as the lowest tested concentration.
1). There was, at most, a one-allele difference between isolates within each cluster, except for Cluster 2 and 13 where there was a three-allele difference. Therefore, isolates within each cluster were highly genetically related, i.e. the same strain. Nine clusters consisted of isolates from one mare (Cluster 3, 6, 7, 8, 9, 10, 11 and 14), whereas five clusters consisted of isolates from 2 to 3 mares (Cluster 1, 2, 4 5 and 13 (Figure 1)). The finding of isolates from different mares in the same clusters suggests a possible spread within the herd.  Eighteen of the Streptococcus equisimilis isolates were phenotypically resistant to tetracycline; eight clustered together from Horse H; ten clustered with isolates from Horse I and J. One isolate from Horse H was phenotypically resistant to erythromycin. One isolate was phenotypically resistant to nitrofurantoin, and it clustered together with isolates from Horse I and J.
Two isolates (P868 and PM194), which clustered together, were phenotypically resistant to fusidic acid. Two other isolates that clustered together (P869 and PM193) were phenotypically resistant to penicillin and trimethoprim/sulfamethoxazole. One isolate (P869) was phenotypically resistant to oxacillin.
The phenotypic resistance was compared between isolates from individual mares before and after the exposure to antibiotics within the same cluster, i.e. isolates of the same strain (Tables 7 and 8). For E. coli, increased resistance to sulfamethoxazole (Cluster 1), trimethoprim (Cluster 2 and 10), chloramphenicol (Cluster 2, 6 and 14) and tetracycline (Cluster 6) was found. ning tree (MST) was generated (Figure 2b). There were no allelic differences between isolates P868 and PM194 and P869 and PM193, respectively; hence, isolates within each pair were genetically closely related, i.e. the same strain. Although no defined cluster distance threshold was available for this cgMLST scheme, the allelic distance between the two pairs was very large, and it is therefore unlikely that they are genetically closely related ( Figure  2b). (b) Minimum spanning tree (MST) created for four Staphylococcus simulans from mares visualizing core genome multi-locus sequence typing (cgMLST) results. Nodes corresponding to sequenced isolates are colored according to individual mares. D0 and D3 are referring to isolates before and after insemination, respectively. The numbers between nodes represent allelic differences. Line lengths are not proportional to the number of differences. (b) Minimum spanning tree (MST) created for four Staphylococcus simulans from mares visualizing core genome multi-locus sequence typing (cgMLST) results. Nodes corresponding to sequenced isolates are colored according to individual mares. D0 and D3 are referring to isolates before and after insemination, respectively. The numbers between nodes represent allelic differences. Line lengths are not proportional to the number of differences. Table 7. Distribution of resistant E. coli isolates and corresponding cluster and horse from which samples were originating. All sequenced Escherichia coli isolates were resistant to sulfamethoxazole both phenotypically and genotypically; 6 isolates had the sul1 gene, and 26 isolates had the sul2 gene, which are responsible for sulfamethoxazole resistance. However, no resistance genes were found in 95 of the isolates. Thirty-two E. coli isolates were phenotypically resistant to trimethoprim; twenty-six of these isolates had dfrA1; seventeen isolates had dfrA14; no resistance genes were found in two isolates. Ten E. coli isolates were phenotypically resistant to tetracycline; six of these isolates had tet(A); one isolate had marR_S3N; no resistance genes were found in the other three isolates. All 32 E. coli isolates that were phenotypically resistant to chloramphenicol had mdf(A). One E. coli isolate phenotypically resistant to tigecycline had mdf(A). Of the seven E. coli isolates phenotypically resistant to ampicillin, six isolates had blaEC-5/blaTEM-1 and blaTEM-1B; one isolate had blaEC. The gene marR_S3N was found in twenty-seven E. coli isolates susceptible to ciprofloxacin, chloramphenicol and tigecycline and twenty-six isolates susceptible to tetracycline. The gene blaEC was found in 119 isolates susceptible to meropenem, cefotaxime and ceftazidime and 118 isolates susceptible to ampicillin. The gene blaEC-5/blaTEM-1 was found in six isolates susceptible to meropenem, cefotaxime and ceftazidime.
The gene catB3 was found in six E. coli isolates susceptible to chloramphenicol. The gene tet(A) was found in six isolates susceptible to tigecycline. The gene aadA5 was found in six isolates susceptible to gentamicin. The gene aph(6)-ld was found in 26 isolates susceptible to gentamicin. In addition, the gene mdf(A), conferring resistance to broad-spectrum drugs via an efflux pump, was found in 92 isolates.
Five Streptococcus equisimilis isolates had lsaC as a resistance gene; one of these isolates had phenotypical resistance to erythromycin. Eighteen isolates phenotypically resistant to tetracycline were not shown to have resistance genes. Furthermore, one isolate, which was phenotypically resistant to nitrofurantoin, did not have resistance genes.
Two Staphylococcus simulans isolates had the blaZ resistance gene responsible for phenotypical resistance to penicillin and oxacillin. However, two isolates phenotypically resistant to fusidic acid and two isolates phenotypically resistant to trimethoprim/sulfamethoxazole did not have any resistance genes.

Discussion
The purpose of this study was to determine if the vaginal microbiota of mares was affected by exposure to antibiotics in semen extenders and also whether this exposure affected the AMR of the bacteria isolated from the vagina.

Bacterial Analyses
In the present study, 300 Gram-positive and 685 Gram-negative bacterial isolates were identified from the vagina. This result is in agreement with other studies where Gram-negative bacteria dominated the microbiota in the vagina [7,26]. The isolates most frequently found in our study included E. coli (52.8%), S. equisimilis (14.0%) and S. zooepidemicus (3.6%). A study in Korea revealed that the most frequently isolated species of bacteria were Escherichia coli (19.8%), Staphylococcus aureus (14.9%) and Proteus mirabillis (14.9%); the samples were collected from vaginal mucosa and clitoral fossa [26]. In contrast, in an Indian report, the dominating bacteria in the vagina were E. coli (21.7%), Enterobacter agglomerans (16.7%), Enterococcus faecalis (15.6%) and Enterococcus faecium (15.6%) [7]. However, bacteria in these other studies were isolated using various types of agar plates and cultured under different conditions compared to our study, and they were identified at the species level using different methods compared to ours. Although other methods of analysis were used in different studies, E. coli seems to be the most isolated bacterial species, most likely because it is one of the most common bacteria in the reproductive tract and the surroundings.
Lactobacillus spp. were not detected in our study, although the selective culture medium, MRS agar, was used. This observation is in contrast to a study performed in Uruguay, in which Lactobacillus spp. were isolated from all vaginal samples [27]. Time and temperature might affect the viability of Lactobacilli; in the study by Fraga et al. [27], the swabs were kept at 4 • C and transported to the laboratory within 3 h. In our study, the stud was at a considerable distance (> 860 km) from the laboratory, necessitating overnight transport of the swabs, which took place at ambient temperatures.

Antimicrobial Susceptibility
The vaginal microbiota of mares in our study was susceptible to most of the tested antibiotics. In fact, all Enterococcus faecalis isolates were susceptible to all 12 tested antibiotics in our study. In other studies, most isolates were obtained from infertile mares, which may have been treated with antibiotics. In other studies, Enterococcus spp. isolates have been described as resistant to all antibiotics tested [7] and resistant to some antibiotics except penicillin, cefquinome, florfenicol, amoxicillin/clavulanic acid, amoxicillin, gentamicin and colistin [15] or resistant to some antibiotics except for gentamicin, vancomycin, linezolid, penicillin and tetracycline [9].
Interestingly, all Streptococcus equi subsp. zooepidemicus isolates of the exposed mares were susceptible to all tested antibiotics. This result is in line with monitoring antibiotic resistance in veterinary medicine in Sweden, where Streptococcus zooepidemicus isolated from horses have remained uniformly susceptible over the years [18]. In contrast, studies of this bacterium from other countries reported resistance to most antibiotics [10,12,28].
Resistance to seven different antibiotics was found in different E. coli isolates in our study, including sulfamethoxazole (90%), trimethoprim (10.7%), chloramphenicol (10.6%), and tetracycline (2%) ampicillin (2.5%), azithromycin (0.5%) and tigecycline (0.5%). Trimethoprim, chloramphenicol and tetracycline showed increased resistance three days after AI. However, these antibiotics were not included in the semen extender. Genes conferring resistance to antibiotics may not be specific for these antibiotics but 'merely' allow survival in a hostile environment, with resistance to antibiotics being coincidental. Therefore, bacteria do not necessarily need to be exposed to an antibiotic to develop resistance to it. Alternatively, exposure to another antibiotic could cause a transfer of resistance between bacteria, e.g. via plasmids [29]. Other factors might influence the association, such as previous treatment or exposure to environmental bacteria on personnel. The resistance of E. coli isolates to some tested antibiotics in our study was similar to other studies, where trimethoprim, chloramphenicol, tetracycline, ampicillin and azithromycin resistance was found [11,13,15]. To our knowledge, there have not been any previous reports of resistance to tigecycline in E. coli isolates from the vagina of mares. Resistance to 4 of 14 antibiotics could be detected in different Staphylococcus simulans isolates in our study, including oxacillin (25%), penicillin (50%), fusidic acid (50%) and trimethoprim/sulfamethoxazole (50%). Other studies showed the resistance of Staphylococcus spp. to most antibiotics tested, including amoxicillin/clavulanic acid, cefotaxime, cefquinome, colistin, enrofloxacin, erythromycin, gentamicin, kanamycin, marbofloxacin, penicillin, ri-fampicin, sulfisoxazole, tetracycline and trimethoprim/sulfamethoxazole [8,14,15]. As in our study, penicillin and trimethoprim/sulfamethoxazole resistance were found in these studies.
Many isolates (91.1%) of Streptococcus equisimilis were resistant to tetracycline but showed low resistance to erythromycin (1.1%) and nitrofurantoin (1.1%). Furthermore, all Streptococcus equisimilis isolates in this study were susceptible to cefalotin, clindamycin, oxacillin, penicillin and trimethoprim/sulfamethoxazole, which differed significantly from some other studies where resistance to all tested antibiotics was reported [7,8].

Whole-Genome Sequencing
Following antimicrobial resistance testing results, sequenced E. coli isolates from inseminated mares at both time points were separated into 14 MST clusters; 9 clusters consisted of isolates from 1 mare, whereas 5 clusters consisted of isolates from 2 to 3 mares. The finding of genetically related isolates in more than one mare indicates a possible spread within the herd. Moreover, for Streptococcus equisimilis, there was an indication of spread within the herd, since the same isolate was potentially detected in isolates from two horses (Horse I and J). Twenty-nine of the sequenced E. coli isolates (22.8%) were MDR, and they were found in five clusters; one cluster consisted of isolates from two mares. The possibility of spreading within the herd might occur also for MDR isolates.
Increased or decreased resistance was observed from isolates from the same horse within the same cluster. Following antimicrobial resistance testing results, the proportions of resistant isolates before and after insemination were compared. Increased resistance to sulfamethoxazole, trimethoprim, chloramphenicol and tetracycline and decreased resistance to azithromycin and tigecycline were found in E. coli isolates. Streptococcus equisimilis showed increased resistance to tetracycline and decreased resistance to erythromycin. An increase in phenotypic resistance might be caused by different mechanisms after experiencing an unfavorable environment; for example, it was reported previously that low-level exposure to antibiotics could lead to antimicrobial resistance [30]. The vaginal bacteria of the mares in this study were exposed to the antibiotics in the semen extender used to prepare the semen doses. However, since single isolates were selected for WGS based on microdilution results in our study, it is not possible to be certain that the same bacterial strain was chosen on all sampling occasions.
Two E. coli isolates and one Streptococcus equisimilis isolate, which had phenotypic resistance, did not contain resistance genes. Furthermore, there were no resistance genes in two Staphylococcus simulans isolates showing phenotypic resistance to fusidic acid and trimethoprim/sulfamethoxazole. Natural or acquired resistance usually results from resistance genes. A study showed that many resistance genes occurring in environmental bacteria might potentially be transferred to bacteria in the mares' reproductive system [31]. Gene transmission in nature occurs via three mechanisms: conjugation, transduction and transformation [32]. Even dead bacteria can pass the resistance gene to other bacteria by transformation [33]. In the present study, possible explanations for the difference between phenotypic and genotypic resistance could be that the genes or the mechanisms conferring resistance have not been detected yet, or there may be a discrepancy between genotypic and phenotypic resistance.
In conclusion, most isolates from the cranial vagina of mares were Gram-negative bacteria. The exposure of vaginal bacteria to antibiotics in the semen extender was associated with a change in the resistance of Escherichia coli against trimethoprim, chloramphenicol and tetracycline. The antibiotic resistance pattern of the genetically related isolates in the same horse increased or decreased in some Escherichia coli and Streptococcus equisimilis isolates after exposure to the semen extender. These results are interesting and warrant further study.