From December 2010 to April 2015, antimicrobial susceptibility test (AST) data of all coagulase-positive staphylococci isolated from horses, cats, and dogs (Supplementary Table S1) were collected through the Resapath network. The Resapath is a network of 78 veterinary laboratories that are transmitting on a voluntary basis all their AST data as well as the strains of interest to the ANSES (French agency for Food, Environmental and Occupational Health Safety). Only clinical isolates were considered in the frame of this study since the Resapath network does not encompass bacteria colonizing animals. All cefoxitin-intermediate or cefoxitin-resistant isolates, as tested by laboratories member of Resapath using the disk diffusion method (see below), were sent to the ANSES laboratory in Lyon for further characterization. After confirmation of the phenotype by disk diffusion, two PCRs were systematically performed: a triplex PCR targeting the 16S rRNA, mecA and nuc genes to detect the presence of MRSA (Maes et al., 2002), and a simplex PCR using mecC-specific primers to detect non-mecA MRSA (Garcia-Alvarez et al., 2011). Species identification was then further assessed using the PCR-RFLP designed by Blaiotta et al. (2010) to differentiate S. aureus from S. pseudintermedius, which is the most frequently encountered coagulase-positive Staphylococcus in dogs.

Antimicrobial susceptibility test was confirmed by the ANSES laboratory using the disk diffusion method and interpreted according to the guidelines of the Antibiogram Committee of the French Society for Microbiology³. S. aureus ATCC 25923 was used as quality control. In addition, 16 antibiotics of veterinary and/or human interest were tested: penicillin G, cefoxitin, cefovecine, kanamycin, gentamicin, tobramycin, tetracycline, erythromycin, spiramycin, lincomycin, chloramphenicol, florfenicol, fusidic acid, enrofloxacin, vancomycin, teicoplanin (Mast Diagnostics, Amiens, France).

The presence of the mecA/mecC gene was systematically confirmed by PCR (Maes et al., 2002; Garcia-Alvarez et al., 2011). Specific detection of the CC398 by PCR was also systematically performed (Stegger et al., 2011). This PCR is based on the detection of a specific sau1-hsdS1 gene (responsible for the restriction modification specificity), which is conserved among the CC398 isolates but differs from other MRSA clones. The spa-types were determined and assigned as previously described⁴ (Koreen et al., 2004).

All mecA/mecC-positive strains were characterized using microarray-based assay (S. aureus Genotyping, Identibac – Alere) allowing detection of virulence and resistance genes, SCCmec types, and assignment to clones and/or MLST clonal complexes. Protocols, target genes for virulence and resistance, databases, typing information, and interpretation management were available at the following internet address: https://alere-technologies.com/fileadmin/Media/Downloads/op/10620/Manuals/05_16_04_0001_V05_Manual_S.aureus_Genotyping_Kit_2_0.pdf.

A total of 130 consecutive non-duplicate clinical MRSA isolates were collected from 43 different French districts: 68 isolates from horses, 34 from cats, and 28 from dogs (Supplementary Table S1). Since 2012, the total number of MRSA collected has significantly decreased in all three animal species (Figure 1), despite the fact that the total number of coagulase-positive isolates collected (susceptible and resistant to methicillin) followed either a steady or increasing trend.

None of the MRSA infected animals presented any epidemiological links since they all originated from different cities, departments or regions, and none of the cats or dogs came from a kennel. MRSA were mostly isolated from skin and soft tissue infections (SSTI) [horse, n = 32 (32/68, 47.1%); dog, n = 11 (11/28, 39.3%); cat, n = 9 (9/34, 26.5%)], while MRSA from reproductive tract infections were also highly prevalent in horses (22/68, 32.4%) (Supplementary Table S2). Isolates globally presented high levels of resistance to several antibiotics, particularly to aminoglycosides (50.0% to gentamicin, 70.0% to kanamycin, 67.7% to tobramycin), tetracyclines (60.0%) and enrofloxacin (48.50%) (Table 1). Major differences were however observed between the CC398 and human-related isolates, with resistances to tetracyclines, phenicols, and aminoglycosides being more prevalent in CC398 isolates, while resistances to macrolides-lincosamides, fusidic acid, and enrofloxacin were more prevalent in human-related isolates (Table 1).

The most prevalent MRSA clone in horses was the CC398 MRSA clone presenting a SCCmec type IV (hereafter named CC398-IV), represented in forty-nine MRSA isolates (49/68, 72.1%) (Figures 1, 2 and Supplementary Table S2). The spa-type t011 was identified in the majority of the isolates (n = 42/49, 85.7%), while t899 (n = 2), t779 (n = 2), t108 (n = 1), t1451 (n = 1) and one non-typeable isolate were also detected. Conversely to cats and dogs, only two isolates displayed enterotoxin genes [i.e., sek/seq (n = 1) and sea (n = 1)]. The IEC was present in 12.2% (n = 6, including 5 type B and 1 type D) of the 49 CC398 isolates. Tetracycline resistance, which has been strongly associated with CC398 strains from livestock so far (Price et al., 2012), was systematically detected and the tet(M) gene identified in 48 of the 49 isolates. The aacA-aphD gene conferring resistance to gentamicin, kanamycin/amikacin, and tobramycin was also frequently identified (46/49, 93.9%) and correlated with phenotypic resistance (Supplementary Table S2). The other resistance genes were scarce, with the sporadic occurrence of genes conferring resistances to macrolides/lincosamides/streptogramins B [erm(A), n = 3], streptogramin A [vga(A), n = 2], trimethoprim (dfrS1, n = 2), fosfomycin (fosB, n = 1), chloramphenicol (cat, n = 1), tetracyclines [tet(K), n = 2], or quaternary ammonium compounds (qacC, n = 1).

Twelve MRSA (17.6%) isolates clustered with the CC8-IV USA500 clone [spa-type t394 (n = 8) and t1952 (n = 1)]. All isolates presented the seb, sek, and seq enterotoxin genes, and the tet(M), aacA/aphD, cat, dfr(S1), and fos(B) resistance genes were detected in all t394 isolates (Supplementary Table S2). None of them presented the IEC cluster that is classically related to human origin/adaptation.

Among the seven isolates belonging neither to the CC398-IV nor the CC8-IV USA500 clone, three were assigned to the CC8-IV EMRSA-14 (n = 2) or to the CC8 Hannover EMRSA (n = 1) clones. This CC8 Hannover clone belonged to the ST249-t009 and presented an atypical SCCmec element comparable to those described by Monecke et al. (2011). All three strains presented multiple resistance genes, including to tetracyclines (Supplementary Table S2). The last four isolates harbored the mecC gene and have already been reported in the frame of another study (Haenni et al., 2015). They belonged to the CC130-XI-t6220, CC130-XI-t1736, CC130-XI-t11050, and CC49-XI-t208 clones (Supplementary Table S2). These four mecC-positive isolates were the only equine isolates that were not resistant to tetracyclines and presented no other resistance genes.

Most isolates belong to the prototypic CC8-IV Lyon clone (13/34, 38.2%) or its sea-negative variant (n = 1). The majority have spa-type t008 (n = 8) (Figure 1 and Supplementary Table S2). In addition to these 14 isolates, another 10 isolates belong to human-related clones: CC5-IV Pediatric (n = 3) and New Pediatric (n = 2) clones, CC5-II (n = 3), CC22-IV Barnim (n = 1), CC5-I Geraldine (n = 1) clones. Among the latter, the IEC (marker of human adaptation) was detected in a vast majority of the isolates (n = 17; 17/24, 70.8%). The IEC type A (sea, sak, scn, chp) was identified in one isolate, the IEC type B (lacking sea) in four isolates, and the IEC type D (lacking chp) in 12/13 isolates from the Lyon clone (van Wamel et al., 2006). Among the human-related clones, only one isolate was resistant to tetracyclines without tet(M) gene.

On the contrary, the tet(M) gene was present in all nine isolates belonging to the CC398 MRSA clone, among which 6 harbored SCCmec type IV (CC398-IV; including four isolates with spa type t011) and three of the SCCmec type V (CC398-V). The IEC cluster was only found in three isolates [type B (n = 1) or type D (n = 2)]. All CC398-V clones presented at least the tet(M) and fexA resistance genes, whereas the CC398-IV clone carried the tet(M) and aacA-aphD genes. The co-occurrence of the tet(M) and tet(K) genes was identified in two spa-type t034 CC398-V. This association, recently shown to increase the fitness of CC398-V strains in the presence of tetracyclines (Larsen et al., 2016), was also identified here in a t1451 CC398-IV isolate originating from a horse (Supplementary Table S2). Finally, one isolate, belonging to the CC130-XI t843 clone, harbored the mecC gene with no associated resistance or enterotoxin genes.

As for cats, nine isolates belonged to the CC8-IV Lyon clone (32.1%) or to other sporadically occurring human-related clones (n = 10) (Figure 1 and Supplementary Table S2). A total of 16 out of the 28 isolates displayed the IEC cluster [type B, n = 6; type D, n = 10 (including the nine Lyon clone isolates)]. Only one isolate did not present any associated enterotoxin gene. Resistance phenotypes to macrolides (10/28, 35.7%), aminoglycosides (15/28, 53.6%), and fosfomycin (18/28, 64.3%) were frequently encountered, whereas only one isolate was resistant to tetracyclines in the absence of the tet(M) gene. The CC398-IV clone was identified in six tetracycline-resistant isolates (21.4%), among which two presented the IEC cluster (type B and type D). The last three isolates belonged to the CC-spa types CC30-t012, CC5-t450, and CC8-t2054 but could not be assigned to any specific clone by the DNA array. Finally, no mecC gene was detected in dogs.

The molecular characterization of the horse collected isolates highlights that MRSA identified in these animals are independent to the human MRSA epidemiology. Two major animal-associated clones were identified in France: the CC8-IV USA500 and the CC398-IV-t011 clones. The CC8-IV USA500, which is the predominant clone infecting horses in North America, also recently described as an environmental contaminant in an equine veterinary center (Weese et al., 2005a; van Balen et al., 2014), was detected in France between 2011 and 2013, with a last occurrence in early 2014. This suggests that this clone, uncommon in the human population in France and worldwide, is prone to colonize and/or infect horses worldwide. Yet CC8-IV USA500 seems to have been totally replaced by the horse-specific CC398-IV belonging to the spa-type t011 in France (Figure 1), confirming its international spread and high fitness/adaptation to horses. Between 2013 and 2015, only 7/39 (17.9%) isolates did not belong to the CC398 clone. Of note, four out of these seven isolates displayed the mecC gene. Whether horses may be prone to disseminate mecC genes or not remains questionable. Both in terms of management by veterinarians and with regards to the global burden antimicrobial resistance, the replacement of CC8-IV USA500 by the CC398-IV-t011 clone may be considered good news since CC398 were co-resistant to aminoglycosides and tetracyclines only, whereas USA500 additionally displayed resistances to macrolides-lincosamides, chloramphenicol, sulfonamides, and fosfomycin. Even though these antibiotic molecules are not all licensed for horse treating, any decrease in antibiotic resistance burden is to be noticed since it may also decrease the risk of co-selection of resistant isolates in other animal species or in humans. This is of course only on a resistance perspective, and the differences in virulence capacities (such as biofilm formation or the presence of adhesion factors) will also have to be considered.

Considering the nosocomial and zoonotic potential of MRSA isolated from horses, equine veterinarians should pay specific attention to both antibiotic treatments and hygiene measures, to limit MRSA selection and transmission. The rate of MRSA carriage in healthy horses, as well as the potential human-to-animal or animal-to-human transmission in both veterinary clinics and in equestrian centers was studied on certain occasions and certainly deserve further investigation (van Duijkeren et al., 2010; Cuny et al., 2016; Koop, 2016).

Our data also underline that MRSA isolated from cats and dogs mirror the epidemiology of hospital-acquired human MRSA (Loeffler and Lloyd, 2010). The CC8-IV Lyon clone was the most frequently identified MRSA clone in both animal species, followed by the occurrence of numerous sporadic lineages. This is in accordance with a similar study previously conducted in France on a smaller number of isolates (Haenni et al., 2012). The ST22 Barnim clone is again rare in France despite its high prevalence in animals in surrounding countries (Strommenger et al., 2006; Grundmann et al., 2010; Harrison et al., 2014; Vincze et al., 2014; Couto et al., 2016; Wipf and Perreten, 2016). Its rare occurrence is most likely related to its low prevalence in French hospitals and in the community. The CC398 clone was detected in 21.4% of all dogs’ and 26.5% of all cats’ isolates, which is more frequent that what has been reported in Germany and much more than what has been recently observed in France (2/23, 8.7%) or in other European countries (Haenni et al., 2012; Vincze et al., 2014; Wipf and Perreten, 2016). Since no links could be inferred between animals sampled and pig farming or contact with livestock, the hypothesis of a progressive replacement of existing human MRSA clones in cats and dogs by the CC398 cannot be excluded, as observed in horses,. Data collected are too limited to confirm such an hypothesis and this will deserve to be followed-up, yet CC398 clones disseminated among humans have recently been exemplified in Denmark, not all to be attributed to contact with livestock (Larsen et al., 2015). Interestingly, all CC398 isolates presented the tet(M) gene, whereas all non-CC398 were devoid of the gene. Other tet genes being rare (Supplementary Table S2), tetracycline susceptibility in MRSA from cats and dogs may be used as a marker for human-related isolates, mirroring the fact that tetracycline-resistance [or the tet(M) presence] can be used as a marker for MRSA CC398 in humans living in high pig-farming areas (Lozano et al., 2012).