Background

Equine pastern dermatitis (EPD) is one of the most frequently encountered skin disorders in equine practice. It is considered a syndrome rather than a disease entity and can result in a range of clinical signs, most typically including erythema, alopecia, scales, crusts, and thickening of the skin in the palmar and plantar regions of the pastern [1,2,3]. Its multifactorial nature not only complicates the scientific investigation of its pathogenesis, but in clinical practice it also impedes the identification of the relevant underlying cause(s) and the formulation of appropriate treatment approaches.

A critical role is often attributed to bacterial infections and, accordingly, antimicrobial treatment of this condition is regularly undertaken. Specifically, infection with Staphylococcus aureus, which is known for its opportunistic pathogenicity [4] and association with skin and soft tissue infections [5, 6], is frequently suspected in the context of EPD [1, 3, 7, 8]. S. aureus can be found in a proportion of the normal skin flora of humans, and can also colonize different animals, including horses [5]; yet, not appearing as a typical commensal on their skin [9, 10], but being found rather in the nasal cavities, if present [11, 12]. Furthermore, S. aureus is also a major opportunistic pathogen that causes a variety of infections in both humans and animals, with the nasal mucosa often serving as a reservoir for endogenous infection [13, 14]. The acquisition and expression of different virulence factors is presumed to play a role in skin and soft tissue infections, such as e.g. intercellular adhesins promoting biofilm formation [15], cytotoxins challenging many different cell types of the hematopoietic lineage [16], or superantigens excessively triggering the immune system such as the ESAT-6-like (early secreted antigen target 6 kDa) staphylococcal type VII secretion system known as ESAT-6 secretion system (ESS) [17, 18].

Furthermore, S. aureus has the ability to become resistant to antimicrobials by either the acquisition of specific genes on mobile genetic elements or by DNA mutations in target genes [19]. According to the degree of resistance, S. aureus has been categorized into methicillin-susceptible S. aureus (MSSA) and methicillin-resistant S. aureus (MRSA) [20]. MRSA are resistant to all beta-lactam antibiotics, except for anti-MRSA cephalosporins, through the acquisition of methicillin resistance genes (e.g. mecA, mecC) on the staphylococcal cassette chromosome mec (SCCmec) [21, 22]. MRSA are also frequently resistant to other classes of critical antibiotics such as aminoglycosides, macrolides and fluoroquinolones [23]. Although MSSA are generally susceptible to antimicrobials, some of them can also exhibit resistance to these classes of antimicrobials [20].

Yet, the involvement of S. aureus in EPD as a primary and/or secondary pathogenic factor has not been elucidated, prompting us to first investigate the prevalence of S. aureus colonization in affected vs. unaffected pasterns, and the nasal passages of the respective horses. We hypothesized that the frequency of colonization would be increased in pasterns affected by EPD, and that an association would exist between nasal and pastern colonization. Furthermore, we anticipated that the frequency of colonization would deviate depending on the clinical manifestation and on the pretreatment. Also, analysis of whole genome sequencing (WGS) data should give new insights into the genetic diversity, local dissemination, as well as the virulence factors and antimicrobial resistance determinants of S. aureus in horses with and without EPD.

Results

Study population

Two hundred horses were enrolled in this study, 105 affected by EPD and 95 unaffected (Table 1). Horses were of various breeds and their stables were spread widely across parts of Switzerland. The affected horses originated from 64 different stables. From 24 of these stables, unaffected control horses could be recruited (n = 58). In addition, another 37 control horses from 13 independent stables were sampled. Of the affected horses, 46 were assigned to the mild form of EPD, 32 to the exudative form, and 27 to the proliferative form. Altogether, an antimicrobial had been applied previously in 28 horses, disinfectants solutions or ointments in 46 horses, and 31 of the affected horses had not been treated at all (Table 1).

Table 1 Characteristics of the study population
Full size table

Frequencies of detection of S. aureus in bacterial culture

In total, 127 S. aureus isolates were obtained from EPD-affected horses, and 19 isolates from healthy control horses (Fig. 1). Based on growth on both SaSelect and CHROMagar MRSA II plates, S. aureus was significantly more prevalent in swab samples from affected pasterns than from unaffected ones (affected: n = 62 [59.0 %], control: n = 6 [6.3 %], P < 0.001). The same was observed for the nasal swab samples (affected: n = 62 [59.0 %], control: n = 8 [8.4 %], P < 0.001). The frequency of MRSA positive samples was comparatively low and did not differ significantly between affected and control horses, neither in the pasterns (affected: n = 1 [0.9 %], control: n = 1 [1.1 %], P = 0.94) nor in the nose (affected: n = 2 [1.9 %], control: n = 4 [4.2 %], P = 0.34). A modified form of a Venn diagram depicts that many horses were simultaneously positive for S. aureus and/or MRSA in their noses and/or pasterns (Fig. 1).

Fig. 1
figure 1

Overview of the investigated samples. The number of samples collected from (a) horses affected by equine pastern dermatitis (EPD) and (b) unaffected control horses is shown. Swabs were streaked onto two selective chromogenic plates for the selection of Staphylococcus aureus strains (SA+; MSSA and MRSA) and of methicillin-resistant S. aureus strains only (MRSA+). A modified form of a Venn diagram depicts that many (c) of the EPD-affected horses and (d) unaffected horses were simultaneously positive for S. aureus and/or MRSA in their noses and/or pasterns

Full size image

In affected pasterns, the association between the EPD form at time of sampling and the choice for the type of pretreatment used was not significant in our study population (P = 0.43). Furthermore, the pretreatment was not found to play a predominant role in the probability of S. aureus isolation in affected pastern samples (Padj = 0.54). However, the form of EPD showed a significant effect (Padj = 0.03), with a higher ratio of positive cultures observed in exudative (23/32 [72 %]) and proliferative lesions (19/27 [70 %]) than in mild ones (20/46 [43 %]).

Characterization and distribution of the S. aureus isolates based on WGS

In one control horse with its nasal swab being positive for both bacterial cultures (S. aureus- and MRSA-selective plate), sequencing results revealed that the respective strains were identical. This duplicate was therefore removed for the following analysis.

Altogether, 1549 genes were commonly present in all isolates and used for cgMLST analysis. In the generated cgMLST tree, 13 clusters or branches (I-XIII) of isolates were gathered, adhering to a threshold of 4 or more matching alleles in MLST profiling. There were six larger clusters of 10 to 51 isolates (clusters I, II, VI, VIII, X and XI), two smaller clusters of 2 to 5 isolates (clusters III and IX) and five singletons (branches IV, V, VII, XII and XIII). In total, 20 different ST were identified as well as 28 different spa types (Fig. 2).

Fig. 2
figure 2

Genetic relationship between the isolates. (a) Minimum spanning tree of all investigated isolated, coloured by their sequence type (ST). The size of the nodes is proportional to the number of isolates represented by the respective node, as is the length of the branches to the number of allelic differences. Thirteen clusters (I-XIII) were gathered, adhering to a threshold of 4 or more matching alleles in MLST profiling. (b) Topological representation of genetic relationships of the investigated isolates. Sequence types are coded in the same colours as in (a). The branch lengths are relative and not to scale. Isolate IDs consist of the horse ID, followed by “P” for pastern samples or “N” for nasal samples. Isolates from control horses are written in italics. Methicillin-resistant strains are marked by an asterisk (*). Closely related isolates originating from the same horse are displayed with a black background

Full size image

The most abundant lineage was ST15-(spa)t084 (n = 26 [18 % of all isolates]) forming the whole of cluster I. Cluster II, the largest cluster, contained 51 S. aureus isolates including the second and third most abundant lineages ST1-t127 (n = 19 [13 %]) and ST1-t1508 (n = 18 [12 %]), as well as six isolates of ST6243 (a single arcC variant of ST1). Four of the other clusters (clusters VI, VIII, IX and XI) accommodated more than one ST, also generated by single locus variants namely ST398 and ST6239 (a single pta variant of ST398) in cluster VI, ST1660, ST6238 (a single pta variant of ST1660) and ST6244 (a single glpF variant of ST1660) in cluster VIII, ST2679 and ST6242 (a single pta variant of ST2679) in cluster IX, and ST133 and ST6240 (a single aroE variant of ST133) in cluster XI. The remaining clusters contained only one ST each or consisted of branches of singletons. Among the isolates of this study, 7 novel MLST patterns were detected, assigned as ST6238 to ST6244, as were 3 novel spa types, assigned as t19544, t19545 and t19549.

The observed strains displayed a moderately high degree of diversity, with half of the isolates belonging to either of two STs (ST1 or ST15). Furthermore, the horses often harboured highly related isolates in their nasal mucosa and on their pastern skin. In 46 horses, both the nose and skin swabs were positive for S. aureus (45 horses with MSSA and 1 with MRSA). Following molecular typing, 28 of these horses carried genetically related isolates of the same ST in both locations, 27 of which also matched in the spa type. Of note, S. aureus ST1 was almost exclusively associated with nasal carriage (Fig. 2). In addition, isolates of a same genotype were also frequently shared between horses of the same stable (Fig. 3).

Fig. 3
figure 3

Distribution of sequence types. A map of Switzerland illustrating the distribution of sequence types within the visited stables. The size of the pie charts is proportional to the number of samples originating from the respective stable. It appears that cohabiting horses frequently harbour similar isolates. Map was drawn and illustrated by Sarah Kaiser-Thom

Full size image

The 8 MRSA strains were isolated from 7 different horses, and from either nose, pastern, or both. All MRSA gathered into the cluster VI with 6 isolates belonging to ST398-t011 and 2 belonging to ST6239-t1456. Notably, four of the six horses with MRSA ST398-t011 isolates lived in the same stable, just as the two horses with ST6239-t1456 isolates. All MRSA strains harboured the SCCmec type IVa.

While the S. aureus isolated from horses with EPD were represented in all clusters (I-XIII), the 18 isolates from healthy horses were distributed among 7 clusters (I, II, VI, VII, VIII, X, XI) associated with ST1-t084 (n = 1), ST15-t127 (n = 1), ST15-t398 (n = 3), ST15-t1508 (n = 2), ST398-t011 (n = 5), ST398-t14030 (n = 1), ST30-t012 (n = 1), ST1660-t3043 (n = 1), ST816-t1294 (n = 2), and ST133-t1403 (n = 1) (Fig. 2).

Identification and distribution of antimicrobial resistance and virulence genes

Antimicrobial resistance genes were more abundant in MRSA than in MSSA. All MRSA (ST398 and ST6239) exhibited the same antimicrobial resistance profile harbouring the methicillin resistance gene mecA, the β-lactamase gene blaZ, the gentamicin, tobramycin, and kanamycin resistance genes aacA-aphD, and the tetracycline resistance gene tet(M). MRSA ST6239 had additional chromosomal mutations with amino acid substitutions within the fluoroquinolone resistance determining region of GyrA (S84L) and GrlA (S80F).

The antimicrobial resistance profiles of MSSA were mostly conserved within isolates of a same cluster. MSSA isolates contained zero to two known acquired resistance genes, except one isolate which had four. The blaZ and the fosfomycin resistance gene fosB were the most frequently detected acquired resistance genes in MSSA. Other genes, such as those associated with resistance to the aminoglycosides tobramycin and kanamycin (aadD), tetracyclines (tet(K), tet(L)), macrolides and lincosamides (erm(T)), as well as a mutation in the trimethoprim binding region of the chromosomal dihydrofolate reductase DfrB (F99Y) [24], were only sporadically found in single MSSA isolates. Furthermore, three acquired genes coding for different multidrug efflux pumps conferring resistance to various antibiotics and other antimicrobial agents (lmrP, sdrM, and qacA) [25] were found among both MRSA and MSSA (Figs. 4 and 5).

Fig. 4
figure 4

Distribution of detected antimicrobial resistance and virulence genes. Isolate IDs consist of the horse ID, followed by “P” for pastern samples or “N” for nasal samples. Key for column ‘group’: “A” = horse affected by EPD, “C” = control horse. In the column ‘pretreatment’, the type of pretreatment of the affected pasterns is disclosed: “no” = no antibacterial pretreatment, “dis” = disinfectant, “ab” = antibiotic. Further details on the clinical data can be found in the supplementary material. For a detailed key of the depicted genes please refer to the caption of Fig. 6

Full size image
Fig. 5
figure 5

Distribution of detected antimicrobial resistance and virulence genes. Isolate IDs consist of the horse ID, followed by “P” for pastern samples or “N” for nasal samples. Key for column ‘group’: “A” = horse affected by EPD, “C” = control horse. In the column ‘pretreatment’, the type of pretreatment of the affected pasterns is disclosed: “no” = no antibacterial pretreatment, “dis” = disinfectant, “ab” = antibiotic. Further details on the clinical data can be found in the supplementary material. For a detailed key of the depicted genes please refer to the caption of Fig. 6

Full size image

The isolate with the second highest frequency of AMR genes was an MSSA belonging to ST1, obtained from an EPD-affected pastern. The corresponding horse was one of two horses having the highest assigned EPD score in the clinical assessment (score: 17/21), and its medical history revealed prolonged treatment of EPD, including various antimicrobial ointments. Otherwise, no association between antimicrobial treatment of EPD and resistance profile could be made, since S. aureus exhibiting the same resistance pattern were found in both treated and non-treated horses. Further details on the clinical data as well as a breakdown of all antimicrobial resistance and virulence-associated genes can be found in the supplementary material (Additional file 1).

Overall, isolates from the nasal mucosa showed slightly higher frequencies of antimicrobial resistance genes than those from pastern samples, independently of being from healthy horses or horses affected by EPD (Fig. 6).

Fig. 6
figure 6

Heatmaps of detected antimicrobial resistance and virulence genes. Frequencies of selected antimicrobial resistance genes (a) and further virulence factor genes (b) in nose and pastern samples from horses affected by equine pastern dermatitis and unaffected control horses are depicted. Antimicrobial resistance genes and functions: MDT, lmrP, sdrM and qacA, multidrug transporters (MDT); dfrBmut, mutated chromosomal dihydrofolate reductase gene leading to amino acid substitution (F99Y) and trimethoprim (TMP) resistance; gyrAmut (S84L) and grlAmut, chromosomal mutations with amino acid substitutions within the fluoroquinolone (FQ) resistance determining region of DNA GyrA (S84L) and topoisomerase GrlA (S80F); fosB, fosfomycin (FOS) thioltransferase gene; erm(T), macrolides, lincosamides and streptogramins B (MLSB) 23 S rRNA methylase gene; mecA, methicillin-resistance gene encoding PBP2a for resistance to all β-lactam-antibiotics; blaZ, β-lactamase gene; tet(K), tet(L), tetracycline (TET) efflux genes; tet(M), ribosome protection tetracycline resistance gene; aacA-aphD, aminoglycoside (AMG) acetyltransferase and phosphotransferase tandem genes for resistance to gentamicin, tobramycin and kanamycin; aadD, ant(4’)-Ia, amikacin and tobramycin nucleotidyltransferase gene. Virulence genes and their functions: seh, seu, and other staphylococcal enterotoxin genes (sea, seb, sec, seg, sei, sel, sell, sem, sen, seo, sep); tsst-1, toxic shock syndrome toxin-1 gene; hlb and other hl (hld, hlgA, hlgB, hlgC, hlIII, hly/hla); ica-family, genes associated with intercellular adhesion; luk-family, leucotoxin genes; set-family, genes for staphylococcal exotoxin-like proteins; “ESS complete” = all genes components of the ESAT-6 secretion system (ESS) present (esaA, essA, essB, essC, esaB, esaG, esxA, esxB, esxC, esxD, and esaD); “ESS incomplete” = five genes components of the ESS (namely essC, esxB, esxC, esxD, and esaD) missing; clf-family, fibrinogen-binding clumping factor genes; fnbA, fibronectin-binding protein gene; map, gene for immunomodulatory protein binding to extracellular matrix (ECM) components; sak, staphylokinase gene; ica-family, genes for intercellular adhesion; cap-family, genes for S. aureus capsular polysaccharides; ssl-family, genes for staphylococcal superantigen-like proteins; sspB, staphylococcal cysteine proteinase staphopain B gene

Full size image

A total of 125 genes associated with bacterial virulence were identified in the genomes of the investigated S. aureus strains. Selected genes associated with toxins, superantigens, colonization, and immune evasion are illustrated in Figs. 4 and 5. The virulence factor gene profiles were highly conserved among S. aureus strains of a same cluster. Slightly different virulence profiles were observed within clusters II (ST15), VI (ST398), and XI (ST133), which further distinguished between strains displaying a different spa type within the same ST. The isolates possessed genes coding for proteins associated with intercellular adhesion (ica-family genes), capsule synthesis (cap), leukotoxins (luk), aureolysin (aur), and hemolysins (hl other than hlb). The hlb gene was found more frequently in pastern samples, as were staphylococcal enterotoxin genes (se), especially those that were not seh, and genes essential for the functionality of the ESS including those coding for the membrane proteins (esaA, essA, essB and essC), the cytosolic proteins (esaB and esaG), and the secreted substrates (esxA, esxB, esxC, esxD, esaD). In the respective isolates, the ESS was thus completed, while the system remained incomplete in the rest of the isolates, where essC, esxBCD, and esaD were missing. Genes of the ESS were lacking in isolates of the cluster I (ST1), VI (ST398) and IX (ST2679 and ST6242) which were almost exclusively composed of nasal isolates. On the other hand, genes coding for clumping factors (clf), staphylococcal superantigen-like proteins (ssl), staphopain B (sspB), and staphylococcal exotoxin-like proteins (set) were slightly more abundant in the nasal isolates (Fig. 6). Frequencies of AMR genes and virulence genes in EPD-affected horses and unaffected horses could not be statistically compared, as the respective groups of isolates differed considerably in size.

Discussion

EPD is a multifactorial syndrome rather than a single disease entity and its complex pathogenesis is still not fully understood. Regarding bacterial infectious agents, staphylococcal species, particularly S. aureus, have frequently been implicated as primary or secondary pathogenic factors in the development of the disease [1, 3, 7, 8, 26,27,28]. The first objective of our study was, therefore, to investigate the prevalence of S. aureus colonization in pasterns of EPD-affected and unaffected horses, and to compare it to the frequency of respective isolates in the nasal mucosa, as the nasal passages represent a known reservoir for staphylococci [13, 29, 30]. Confirming our hypothesis, we found that the proportion of positive pastern samples in EPD-affected horses (59.0 %) was markedly higher than in unaffected horses (6.3 %). Opportunistic staphylococcal colonization of lesional sites has been described in various dermatological disorders and during wound healing in horses [27, 31]. However, it is important to note that S. aureus is not a typical commensal of the normal equine skin, unlike for example S. sciuri and S. epidermidis [9, 32]. The relatively low frequency of detection of S. aureus in unaffected horses is in accordance with these previous reports. Recently, our group investigated the skin microbiota in horse pasterns, where some of the here enrolled EPD-affected horses were included [33]. We found that, although Staphylococcaceae (microbiota composition was only resolved to the family level) were also detected in the unaffected pasterns, there was a substantial increase in relative abundance in the affected pasterns, particularly in exudative and proliferative lesions, essentially showing the same proportional pattern as we observed in the present study. As stated by Chiers et al., the isolation of pathogenic S. aureus from equine skin lesions, considering their relatively rare occurrence on healthy equine skin, suggests a role in the aetiology and progression of the respective lesions [27]. Our results thus indicate that S. aureus may also play a key role in EPD.

Of note, the affected horses concurrently exhibited a significantly higher number of S. aureus-positive nasal samples (59 %) as compared to the healthy control group (8.4 %), closely reflecting the numbers observed in the skin samples. The high rates of S. aureus isolation from both pasterns and noses of affected horses, coupled with the fact that the respective isolates were frequently also genetically highly related may indicate endogenous infection [13, 34]. Nevertheless, the cumulative occurrence of similar strains within stables also suggests horizontal transmission between horses, which could well extend to humans in contact with these horses, as has been reported before [12, 35]. Across all isolates, we observed only a moderately high diversity of STs and spa types, most of them already associated with S. aureus from horses (ST1-t127, ST1660-t549/t3043/t2484, ST15-t084, ST133-t1166, ST398-t011) [36,37,38,39], as well as from humans (ST15-t084, ST1-t127, ST1660-t549, ST1660-t3043, ST672-t003) [38,39,40,41]. Among them, ST15-t084, ST1-t127 and ST1-t1508 were the most predominant lineages. The lineage ST1-t127 seems to have a broad dissemination potential as it was found in diverse hosts, as for example in healthy pigs [42], healthy wild boars [43], nasal colonization and infections of humans [44], in humans in contact with pigs [45, 46], in mastitis of dairy cows [47], and in a purulent wound of a racehorse [48]. The reported strains frequently consisted of MRSA, contrarily to those of this study which were MSSA. The MRSA isolates in our study belonged either to the lineage ST398-t011 or to the novel, but closely related ST6239-t1456, and were all of SCCmec type IVa. MRSA of the clonal complex 398 represents the predominant lineage in livestock as well as in equine environments, being associated with both colonization and infections in horses [38, 49,50,51]. In our study, MRSA ST398 was more frequently detected in the nasal cavities of healthy horses than in the pastern of horses with EPD, suggesting that it may not have the virulence factors necessary to develop into EPD. For instance, they do not contain a complete ESS, like the MSSA ST15 which were most exclusively found in nasal samples. These S. aureus isolates lack the secreted substrate genes esxB, esxC, esxD, esaD and the membrane protein gene essC, which encodes for a membrane-anchored ATPase that putatively assembles into a complex with the other membrane components (essA, essB and esaA) and is necessary to release the aforementioned substrates [17, 18]. However, a markedly greater proportion of pastern isolates possessed both the substrate genes and all the genes of the ESS necessary for the secretion of these virulence factors. The substrates of this secretion system have previously been associated with the modulation of apoptosis in staphylococcal intracellular infection [52], as well as the ability of abscess formation [53], and might therefore be of importance in the pathogenesis of EPD. The pastern isolates also exhibited greater abundances for the hlb gene and for a larger variety of staphylococcal enterotoxins. The hlb gene encodes for a hemolytic toxin with sphingomyelinase activity and has been shown to play a crucial role in the ability to colonize the skin [54]. Amongst other properties of the enterotoxins, these heat-stable exotoxins can induce cytokine production as well as T-lymphocyte proliferation [55, 56], and may, consequently, induce or exacerbate inflammation. Moreover, a full range of additional virulence factor genes was encountered, providing the isolates competitive advantage as well as pathogenic potential. The isolates contained cap and ica genes, assuring capsule synthesis and formation of biofilms [15, 57], as well as leucocidin genes luk, aureolysin gene aur, and an array of hemolysin genes hl-, additionally paving the way for a more mobile and invasive behaviour [58,59,60].

Most of the S. aureus isolates did not contain multiple antimicrobial resistance genes, with the exception of the few MRSA ST398-t011 and ST6239-t1456, and one MSSA ST1-t127. Overall, the β-lactamase gene blaZ and the fosfomycin resistance gene fosB were the most frequently detected genes. Yet, penicillins and fosfomycin do not seem to be common active ingredients used in topical treatment of EPD, for which aminoglycosides were predominantly used. Metadata investigation of the participating affected horses indicated that the antimicrobial ointments applied in the 13 antibiotically pretreated pasterns each contained an aminoglycoside, either neomycin (12/13) or gentamicin (1/13), most frequently in combination with thiostrepton (8/13) and once in combination with gramicidin. However, aminoglycoside resistance genes were only detected in 8 isolates which consist of the MRSA and the multidrug-resistant MSSA ST1-t127. With regard to these, at least in our cohort, commonly utilized topical antimicrobials, the current molecular-based resistance status of the investigated staphylococci can be considered noncritical. However, it may be possible that some of the isolates exhibit some resistance phenotypes for which no known acquired mechanisms has yet been reported and absent in the current databases. Therefore antimicrobial susceptibility testing should be performed to insure targeted treatment.

Conclusions

Our study highlights the potential importance of S. aureus in the development and pathogenesis of EPD, revealing a substantial higher abundance of S. aureus in affected horses, as well as specific genetic features of the S. aureus population. To our knowledge, a comparably extensive whole genome sequencing approach of S. aureus strains has not yet been performed in the equine setting. Whole genome sequencing gave new insights into the genetic features of the investigated S. aureus strains, particularly resistance and virulence factors, and framed their genetic relationships. Our observations suggest that dissemination of strains takes place between different sites within the same horses, and that spreading also occurs between horses living in the same stable. Some S. aureus lineages containing specific virulence factors such as ESS, hemolysin and enterotoxins were more present in affected pastern than in nasal cavities. It remains open, whether S. aureus plays a rather primary or secondary role in the development of EPD, and what the clinical significance of these toxins is. Further controlled and longitudinal studies on the bacterial impact in the pathogenesis of EPD, as well as investigations on colonization with S. aureus strains in affected horses will be needed to identify the key virulence factors contributing to the pathogenesis of EPD. However, our study underlines that the presence of S. aureus in horses should not be neglected in the diagnostic, prevention and treatment of EPD following antimicrobial susceptibility testing.

Methods

Study design

This study was part of a superordinate project to investigate the role of dermal bacteria in EPD. Some of the EPD-affected horses were included in a previous study by our group regarding the skin microbiota in EPD [33]. Participants were recruited through announcements on social media platforms. Sampling started in April 2019 and was completed in August 2020. Whenever possible, healthy control horses were recruited from the same stables as the affected horses. To achieve comparable group sizes, further healthy horses from unrelated stables were also included.

All horses underwent a general physical examination, followed by thorough inspection of all four pasterns. Diagnosis of EPD was based on evidence of typical clinical signs and lesion severity was scored using a standardized scoring system [33]. The score accounts for skin pathologies that are commonly associated with EPD, including scales, crusts, ulceration and formation of skin folds [1,2,3]; the cumulative value of score for each pastern can range between 0 (not affected) and 21 (severely affected) [33]. Affected pasterns were also assigned one of three EPD forms (mild, exudative or proliferative) as described by Yu [1]. Horses were designated as controls, if they had not shown signs of EPD in the preceding two years. In EPD-affected horses, lesional pastern areas were included irrespective of local disinfecting or antimicrobial pretreatment, except that no topical treatment which could impair bacterial growth was applied at the day of sampling. However, no horses treated with systemic antimicrobials or hospitalized in the preceding six months were included in this study.

Sample collection

Two swab samples were obtained from every participating horse, one from the pastern skin and one from the nasal mucosa. In the affected horses, the most severely affected pastern was chosen for sampling based on the cumulative lesion score. As hind legs are generally more susceptible to EPD, for comparison, the pastern sample from the healthy horses was obtained by stroking the pastern regions of both hind legs consecutively with the same swab. The nasal swab was obtained through the right nostril in all horses, in order to facilitate sampling by head-shy horses. Samples were collected using flocked swabs (Puritan Opti-Swab, Puritan Medical Products, Gulford, ME, US). For sampling of the skin, swabs were slightly moistened with sterile 0.9 % saline solution. Swabs were then transported in 1 mL of liquid Amies medium and processed on the same day.

Cultivation and identification of strains

Swabs were placed into Mueller-Hinton broth containing 6.5% NaCl for overnight enrichment at 37 °C with shaking. A loopful of the cultures was then streaked onto chromogenic plates SaSelect (Bio-Rad, Hercules, CA, US) for the selection of S. aureus, and CHROMagar MRSA II (Becton and Dickinson Company, Franklin Lakes, NJ, US) for the selection of MRSA. The plates were incubated for 24 h at 37 °C under aerobic conditions. Colonies were identified by matrix-assisted laser desorption/ionisation time-of-flight mass spectrometry (MALDI-TOF MS) (Bruker Daltonics GmbH, Bremen, Germany), and sub-cultivated onto trypton soy agar plates containing 5 % sheep blood (TSA-S, Becton and Dickinson Company, Franklin Lakes, NJ, US). The S. aureus isolates were preserved by freezing in glycerol stocks at -80 °C.

A chi-square test of independence was used to compare the number of samples with cultures positive for S. aureus and/or MRSA between EPD-affected and healthy control horses, and within the affected horses, to compare the number of positive cultures between the three EPD forms, as well as the type of previous treatment. When indicated, a Bonferroni correction for multiple comparisons was performed. A P-value of <0.05 was considered as statistically significant.

Whole genome sequencing

DNA for library preparation was extracted directly from bacterial colonies grown on TSA-S using enzymatic lysis by Proteinase–K in combination with mechanical disruption by glass beads (PowerBead, Qiagen, Hilden, Germany). Extracts were purified using the AMPure XP paramagnetic bead-based chemistry (Beckman Coulter, Brea, CA, US). Libraries were prepared with the Nextera DNA Flex Library Prep Kit (Illumina Inc., San Diego, CA, US) following the manufacturer’s instructions. Whole genome sequencing (2 × 150 bp paired-end) was performed on an Illumina MiSeq platform (Illumina Inc., San Diego, CA, US) at the Next Generation Sequencing Platform, Institute of Genetics, University of Bern.

Analyses of WGS data

The WGS raw data was imported into the commercially available software SeqSphere+ (version 7, Ridom GmbH, Münster, Germany). The preprocessing tool Trimmomatic [61] was used for quality-based filtering and downsampling. A core genome multilocus sequence typing (cgMLST) scheme was created with 1,861 queried target genes, as previously described [62]. All genes which were present in all isolates were defined as the core genome and included for further analysis. A minimum spanning tree was constructed and visualized with the help of GrapeTree [63] and iTOL [64]. The software SeqSphere+ was also used for multilocus sequence typing (MLST), staphylococcal protein A (spa) typing, as well as for the prediction of antimicrobial resistances using NCBI’s AMRFinder, and of virulence factors using the Virulence Factor Database (VFDB). Screening for antimicrobial resistance chromosomal mutations using ResFinder 4.1, and staphylococcal cassette chromosome mec (SCCmec) typing using SCCmecFinder 1.2 were performed using tools and default settings of the Center for Genomic Epidemiology (http://www.genomicepidemiology.org/). A map of Switzerland was created using the free software GIMP (https://www.gimp.org) to illustrate the spread of sequence types (STs) within the visited stables. The open-source software R [65] was used to construct heatmaps of the frequencies of antimicrobial resistance (AMR) genes and further virulence factor genes.