Over the last decade, livestock-associated methicillin-resistant Staphylococcus aureus (MRSA) belonging to clonal complex (CC) 398 has emerged worldwide in various animal species, especially in pigs as well as in people in contact with MRSA-positive animals (Petinaki and Spiliopoulou, 2012). Zoonotic transmission from pigs to humans has been substantiated by various epidemiological studies (Lewis et al., 2008; Köck et al., 2013). In Denmark the number of human infections caused by MRSA CC398 has increased steeply from 13 cases in 2009 to 240 cases in 2014, and decreased to 208 cases in 2015 (DANMAP, 2014, 2015). Presently, the CC398-associated spa types t034 and t011 account for 19% of all MRSA infections in this country (DANMAP, 2014, 2015), and the majority of cases are found in people with direct or indirect (household members) contact with pigs (Voss et al., 2005). However, a steady proportion of MRSA CC398 cases (10–15% of all cases and 37% of infections) occur in people that are not exposed to livestock (Lekkerkerk et al., 2012; DANMAP, 2014, 2015; Larsen et al., 2015), raising questions about the possible source of infection in these people.

Denmark has a population of ~5.7 million people and produces nearly 31 million pigs per year, according to data from 2015 (Statistics-Denmark, 2015a,b). Approximately 88% of the pig farms are positive for MRSA CC398¹. New studies in Denmark in 2015 have found one poultry farm, five veal calf farms and eight mink farms MRSA CC398-positive (2, 5, and 8%, respectively) (DANMAP, 2015). Other potential animal reservoirs in Denmark have not been investigated. However, studies from other countries have shown MRSA CC398 carriage in horses (0.5–11%; Van den Eede et al., 2009, 2012) as well as a high prevalence (8–15%) of this CC amongst equine clinical MRSA isolates (van Duijkeren et al., 2010; Vincze et al., 2014). It appears that a recent shift has occurred in the epidemiology of MRSA in horses, with CC398 gradually replacing CC8 as the most common CC (Loeffler et al., 2009; Van den Eede et al., 2009, 2013; van Duijkeren et al., 2010). In addition to the noticeable spread of MRSA CC398 in horses, recent studies suggest that zoonotic transmission between horses and humans may be enhanced by the presence of the phage ΦSa3 in horse isolates (Cuny et al., 2015; Jung et al., 2017). This phage carries the human immune evasion gene cluster (IEC) that provides S. aureus protection from the immune system of humans and horses (Jung et al., 2017). Data on S. aureus carriage and infection in horses are largely biased in favor of MRSA and very little is not about the prevalence and genetic diversity of methicillin-susceptible (MSSA) strains.

The objectives of this study were (i) to investigate whether horses constitute a reservoir of MRSA CC398 in Denmark and (ii) to gain information on the frequency and genetic diversity of S. aureus in this animal species, including both MRSA and MSSA.

A total of 401 horses originating from 74 farms (41 small private farms, 20 riding schools, and 13 stud farms) were included in the study. The number of horses sampled per farm ranged from 1 to 20 with a median value of four. The sampled population consisted of 211 females and 190 males with a median age of 9 years (range: 3 months to 28 years). Three hundred and ninety-five horses were sampled on the farm where they lived, and six horses were sampled upon arrival to the clinic. Of them, 393 horses did not display any clinical signs at the time of sampling. The remaining eight horses had either laminitis (n = 2), colic/intestine alterations (n = 2), Cushing's syndrome (n = 1), dermatitis (n = 1), infection of the hoof (n = 1), or lung infection (n = 1). Three hundred and fifty-five horses had not been treated with antimicrobial agents in the preceding 6 months. Among the remaining horses, antimicrobial treatment was reported in 26 horses, and the treatment status was unknown for 20 horses (Data Sheet 1). The antimicrobial drugs used included trimethoprim/sulfadiazin (n = 12), cefquinome (n = 6), benzylpenicillinprocain (n = 3), ethacillin (n = 3), ampicillin (n = 1), doxycillin (n = 1), and tobramycin (n = 1). Three of the treated horses had received unknown antimicrobial drugs, and four had been treated with more than one drug. Most farms had cats (n = 52) and/or dogs (n = 51), three had cows, two had pigs, and five farms had other animals including alpaca, deer, and mouflon (n = 1), alpaca only (n = 1), hens (n = 1), rats/mice (n = 1), or fish (n = 1).

A total of 79 S. aureus (MRSA and MSSA) isolates were obtained from 54 horses (13.5%) originating from 30 farms (40.5%). MRSA was isolated from 17 horses (4.2%) from seven farms (9.5%) located in Funen and Zealand (Figure 1B; Table 2). The identified MRSA isolates belonged to three genotypes based on the PCR results: (i) IEC-positive MRSA CC398 t011 (10 horses from three farms), (ii) IEC-negative MRSA CC398 t034 (four horses from two farms), and (iii) IEC-negative mecC-MRSA CC130 t528 (three horses from two farms; Table 3). Each MRSA-positive farm harbored only one of these genotypes. One of the 17 MRSA carriers reported health problems (dermatitis) and unknown antimicrobial history, whereas the remaining 16 carriers were healthy and had received no antimicrobials in the previous 6 months.

Based on PCR results, 50 of the 79 S. aureus isolates were MSSA belonging to 15 spa types and carried by 41 horses. The population structure of the MSSA isolates differed notably from that of MRSA isolates. None of the MSSA isolates belonged to CC398. The two most prevalent spa types were t127 (12 horses) and t549 (4 horses), which are associated to CC1 and ST1660, respectively (Table 3). Four S. aureus isolates were confirmed by MALDI-TOF but were lost and could not be typed. Two spa types (t15815 and t15816) were identified in this study for the first time. Most S. aureus-positive horses (48 of the 54) carried only one spa type, whereas six horses were found to simultaneously carry two unrelated spa types, either two MSSA or one MSSA and one MRSA (Table 4). In addition to the MRSA CC398 t011 isolates, three MSSA CCs harbored scn: CC1 t127 (n = 2), CC5 t701 (n = 1), and CC5/CC130 t1943 (n = 1). The BURP analysis grouped the spa types into five clusters and provided presumptive ST/CC information of two spa types without previously reported multi-locus sequence type: t4735 (clustering with t2420) presumably belonging to CC133 (Gómez et al., 2015), and t15816 (clustering with t549 and t3043) presumably belonging to ST1660 (Sieber et al., 2011; Cuny et al., 2016; Table 3). None of the isolates carried PVL.

Table 4 lists the genotypic characteristics of the MRSA CC398 isolated in this study. The four equine MRSA CC398 isolates with spa type t034 carried SCCmec V(5C2&5)c, lacked IEC, and clustered with two of the three widespread porcine clones in Denmark (hereafter referred to as pig-associated MRSA CC398). The 10 MRSA t011 from horses carried SCCmec type IV(2B)a and IEC type B, and formed a separate branch in the maximum-likelihood phylogeny (hereafter referred to as horse-adapted MRSA CC398; Figure 2; Table 4). Three of the 11 human MRSA t011 IEC-positive isolates form a monophyletic branch with 100% bootstrap support (Supplementary Image 1) together with all equine isolates with spa type t011. These isolates originated from equine veterinarians, including two members of the same family. The remaining eight human isolates either grouped together with, or were nearest neighbors to the IEC-negative spa type t034 isolates from pigs and horses (Figure 2; Table 4), and originated from two pig farm workers, four family members of farm workers, one person with contact to horses, and one person without any contact to livestock. The 10 horse isolates and the three closely related human isolates carried the unique SNP called 309-2 (Table 4). We also found SNP 309-2 in three IEC-negative MRSA CC398 isolates from the worldwide collection, including two porcine spa type t011 isolates from Hungary and Italy and one equine spa type t1451 isolate from Belgium. These three IEC-negative isolates were located on the same branch as, but basal to, the IEC-positive horse-adapted isolates from Denmark (Figure 2), suggesting that IEC was acquired after diversification of the horse-adapted clone.

All horse and human isolates within the horse-specific clade carried aacA-aphD, the gentamicin and kanamycin resistance determinant (Data Sheet 2). The gene was otherwise only found in two livestock-associated MRSA isolates from the worldwide collection, including a porcine spa type t011 isolate from Italy and a bovine spa type t567 isolate from Belgium, which is the closest neighbor to the horse-adapted clone. None of the isolates within the horse-specific clade carried the genes encoding LukPQ.

Occurrence of total MRSA, MRSA CC398, or pig-adapted MRSA CC398 t034 showed no significant associations at the individual-level analysis with age, health status, and antimicrobial use in the preceding 6 months, or at the farm-level analysis with farm type, presence of other animals on the farm or density of pigs or pig farms in the same geographical location based on postal code (Supplementary Image 2).

We detected MRSA in 4.2% of the horses in the study population, which mainly included healthy horses without recent admission to clinics or antimicrobial treatment. This prevalence is higher than those reported by previous studies in healthy horses from Canada, Belgium, Ireland, and the UK (0–1.7%; Burton et al., 2008; Abbott et al., 2010; Maddox et al., 2012; Van den Eede et al., 2012). A previous Danish study conducted in 2005 failed to detect MRSA amongst 100 horses, including both healthy and diseased horses (Bagcigil et al., 2007). Although, the study populations are not fully comparable, these data suggest an increase in the prevalence of MRSA in the horse population in Denmark over the last decade. The recent spread of MRSA in horses potentially poses a health risk to equine veterinary practitioners, horse owners, and other people in contact with positive animals. It also poses risks to animal health as evidenced by the increasing prevalence of MRSA infections in horses worldwide (Cuny and Witte, 2016).

Our study shows that horses in Denmark are a reservoir for a variety of MRSA genotypes, including both horse-adapted and pig-adapted MRSA CC398 as well as MRSA CC130 harboring mecC. IEC-positive MRSA CC398 t011 harboring SNP 309-2, which is a marker of a previously identified horse-adapted clone (Abdelbary et al., 2014), constituted 59% of the equine MRSA isolates. This clone was recently shown to be the most common MRSA type in clinical samples from horses in central Europe (Abdelbary et al., 2014). The IEC-positive MRSA CC398 t011 isolates shared other characteristics with horse isolates from Germany and the Netherlands in that they carried SCCmec type IV(2B)a and the gentamicin and kanamycin resistance determinant, aacA-aphD (Cuny et al., 2008; van Duijkeren et al., 2010). The latter genotypic feature may reflect the widespread usage of gentamicin in horses.

Whole-genome phylogenetic analysis showed that three of the human isolates (one from healthy carriage and two from infections) were highly similar to those isolated in this study from horses (Figure 2; Table 4). Interestingly, all these isolates originated from equine veterinarians, indicating an occupational risk for MRSA CC398 carriage and infection in veterinary staff. The horse-adapted clone accounted for 0.2% (2/834) of all MRSA CC398 infections during 2004–2015, according to the national MRSA database. In a previous study by Abdelbary et al. (2014), the horse-adapted clone was detected in Austrian veterinary personnel but not in other humans. Danish horse veterinarians have previously been associated to a higher risk of MRSA carriage compared to people not exposed to horses (Moodley et al., 2008). The current MRSA control policy in Denmark includes screening and decolonization of people at risk of MRSA carriage upon hospital admission, e.g., pig farmers. Whether or not horse veterinarians should be included as another risk group requires further epidemiological evidence through studies that assess the prevalence of MRSA carriage and infection in people in contact with horses, or investigate the exposure to horses in MRSA-positive patients.

The equine IEC-negative MRSA CC398 t034 isolates were closely related to two of the three clones that are widespread among pigs in Denmark (Figure 2), suggesting a recent transmission between these two animal hosts. The origin of pig-adapted MRSA CC398 in the horses is unknown, since pigs were not present in any of the positive horse farms. The lack of contact with pigs suggests that horses could be a reservoir for the pig-adapted MRSA CC398 t034 in addition to pigs. Moreover, the presence of SNP 309-2 in the two porcine t011 isolates from Hungary and Italy suggests a possible jump of MRSA CC398 from pigs to horses, where it subsequently acquired IEC through acquisition of the ΦSa3 phage.

The mecC-positive MRSA CC130 t528 was detected for the first time in horses in Denmark. mecC-positive MRSA CC130 and CC49 belonging to other spa types have been isolated from horses in France and Germany (Haenni et al., 2015; Cuny et al., 2016). MRSA CC130 is known to have a broad host spectrum, mainly ruminants among domestic animals (Paterson et al., 2012; Loncaric et al., 2013; Monecke et al., 2013; Gomez et al., 2014). In Denmark, mecC-positive MRSA CC130 t528 has been recognized as a cause of human infections (Petersen et al., 2013) and as a commensal in the nasal cavity of cattle (García-Álvarez et al., 2011), small ruminants (Eriksson et al., 2013), and wild animal species at the Copenhagen Zoo (Espinosa-Gongora et al., 2015). Further insight into the phylogenetic relationship between these isolates could shed light on the possible role of horses in the evolution and dissemination of this MRSA clone.

The overall S. aureus carriage rate in the study population was 13.5%, which is slightly higher than the MSSA prevalence reported in healthy horses in Canada (7.9%; Burton et al., 2008). The most common MSSA lineage, CC1 t127, was isolated from 12 horses at 11 farms. This is a ubiquitous spa type that has previously been isolated from horse MRSA infections (Cuny et al., 2008), and in Denmark it is the most common spa type isolated from human cases of S. aureus bacteremia (DANMAP, 2015). The apparent lack of host-specificity of t127 and thereby potential zoonotic relevance requires confirmation by high-resolution genotyping methods such as the WGS approach used in this study for CC398 isolates. The second most common spa type, t549 (ST1660), is considered a horse-adapted S. aureus clone (Cuny et al., 2016).

No significant associations were found between MRSA carriage and the variables covered by the questionnaire (age, health status, recent antimicrobial therapy, farm type, presence of other animals on the farm, and density of pigs or pig farms in the area of the farm). Our data suggest that antimicrobial use is not a major risk factor for MRSA carriage in horses. The factors contributing to the recent spread of MRSA CC398 in the horse population in Europe remain unknown. Further epidemiological studies are needed to assess whether MRSA transmission may be enhanced by movement of horses and horse owners in relation to sport and breeding activities.

To conclude, horses can be healthy carriers of distinct MSSA and MRSA clones that are implicated in human infections. In addition to horse-adapted MRSA CC398, which is a rare cause of human infection, horses in Denmark are exposed to distinct MRSA clones that are likely to originate from other animal species (CC130 t528 and pig-adapted CC398 t034) as well as to a MSSA clone that is prevalent in human cases of S. aureus bacteremia in Denmark (CC1 t127). Further research is needed to determine the stability of MRSA and MSSA carriage in horses, to identify risk factors and elucidate mechanisms of transmission between equine farms, and to quantify the risk of infection in people exposed to horses.

Design of the work: LG, AM, PD, JL; Acquisition, analysis, or interpretation of data: MI, CE, RNS, RM, LH; Drafting the work or revising it critically for important intellectual content: MI, CE, PD, RNS, AM, RS, JL, LG; Final approval of the version to be published (all authors).

The study was funded by the Danish Pig Levy Fund (Svineafgiftsfonden), and the Knowledge Centre for Agriculture/Danish Pig Research Centre (SEGES). SEGES Horses and the Danish Equestrian Federation funded whole-genome sequencing of MRSA isolates obtained from horses.