Molecular Epidemiology of Methicillin-Resistant and Methicillin-Susceptible Staphylococcus aureus in the Ovine Dairy Chain and in Farm-Related Humans

Staphylococcus aureus is a major cause of clinical infections in humans and its enterotoxins cause foodborne disease. In the present study, we tested a total of 51 isolates of S. aureus from small-ruminant dairy farms with artisan dairy facilities, all located in Latium, Italy. The farms have a known history of a high prevalence of methicillin-resistant S. aureus (MRSA). Most of the MRSA isolates (27 of 51) belonged to spa-type t127 (43.1%), followed by t2678 (3.9%), t044 (2%), t1166 (2%), and t1773 (2%). PFGE performed on mecA positive strains identified one cluster (≥ 80% of similarity), comprising 22 MRSA. Nine of twenty-two MRSA isolates were assigned human host origin, and 13 isolates did not belong to a specific host. During the characterization study, one strain isolated from bulk tank milk samples harbored the pvl gene; the strain was not enterotoxigenic with a non-specific host according to the biotyping scheme, highlighting the possible emerging risk of transmission of bacterial virulence factors by foods, the environment, and foodhandlers. These findings stress the importance of hygienic measures at all processing steps of the food production chain and underline that monitoring for the presence of MRSA throughout the food chain is essential for public health.


Introduction
S. aureus is one of the most important etiological agents of intramammary infections in small ruminants [1]. In sheep it is responsible for clinical mastitis in 5-11% of cases and for sub-clinical mastitis in 0.22-2.06% of cases [2,3]. Its presence is a major economic and public health concern in the milk and dairy product sector [4]. S. aureus contamination of dairy products can occur anywhere in the food chain, while intra-farm spread mainly occurs through milking procedures [5], where infected mammary glands are the main reservoir [6].
Previous studies have examined a limited number of strains isolated from sheep and goat livestock that showed a host specificity and close relatedness [3,7]. Despite host specificity, S. aureus can switch and adapt to different host species, leading to host restriction [8].
The exotoxin PVL gene was detected in one MRSA strain isolated from bulk tank milk that tested negative for SE genes and not-specific host type 2 (NSH2) in biotyping (Table 3). Pulsed-field gel electrophoresis (PFGE) (http://insilico.ehu.es/dice_upgma/index.php) identified 13 S. aureus pulse types. According to Mello and colleagues [33], analysis of the PFGE profiles highlighted two S. aureus clusters that simultaneously included ≥ 3 isolates with similarity ≥ 80% ( Figure 1).
Pulsed-field gel electrophoresis (PFGE) (http://insilico.ehu.es/dice_upgma/index.php) identified 13 S. aureus pulse types. According to Mello and colleagues [33], analysis of the PFGE profiles highlighted two S. aureus clusters that simultaneously included ≥ 3 isolates with similarity ≥ 80% ( Figure 1). Cluster A consisted of three strains all harboring the gene sec coding for SE type C. In terms of host specificity, all strains were related to an ovine origin. This cluster comprised nine isolates from farm 3 (Table 4), and all strains were MSSA. Cluster B comprised 22 MRSA isolated from different farms (1, 2, 3, and 4). In terms of host specificity, nine isolates were assigned to human host origin and 13 isolates did not belong to a specific host (NSH3, NSH6, and NSH4) ( Table 5).  Cluster A consisted of three strains all harboring the gene sec coding for SE type C. In terms of host specificity, all strains were related to an ovine origin. This cluster comprised nine isolates from farm 3 (Table 4), and all strains were MSSA. Cluster B comprised 22 MRSA isolated from different farms (1, 2, 3, and 4). In terms of host specificity, nine isolates were assigned to human host origin and 13 isolates did not belong to a specific host (NSH3, NSH6, and NSH4) ( Table 5).

Discussion
In this study, a PVL positive strain (lukS-lukF+) was highlighted during a study for the screening of MRSA in six ovine farms that tested positive in a previous study. PVL is one of the most important virulence factors produced by S. aureus, contributing to the pathogenicity of this microorganism. This toxin is associated with different diseases in humans, such as pneumonia and necrotizing dermatitis [34]. The Panton-Valentin leukocydin gene (pvl) gene has also been identified in S. aureus strains isolated from cases of mastitis in animals [35][36][37]. In Europe, the most common clone is the European ST80 [38]; however, this is the first report of S. aureus isolated from ovine raw milk. The strain tested positive for the mecA gene and negative for the enterotoxin genes tested (SEs profile number 2); it belonged to spa-type t044 and showed a non-host specificity (NSH2) in biotyping. The PFGE profile of the strain and the phylogenetic study showed a high genomic distance (strain 5, Figure 1) from the population studied, and it belonged to ST80, which, according to the spa-ridom server, is widely disseminated in Europe.
This study revealed the presence of multiple MRSA strains among animals at the farms with possible human contamination. In fact, 11 of the mecA positive strains confirmed by PCR (27 strains in total) indicated a human host origin in biotyping. The five strains isolated from the operators (nasal swabs, oral swab, and skin swab) from two farms (3 and 4) were considered the common infectious agent for the individual milk samples (3 isolates), bulk tank milk (2 samples), and ovine swab skin (1 isolate). Although the isolates showed the same spa-type (t127), they highlighted a different PFGE profile using SmaI between the strains isolated from two operators (spa-type t127 and PFGE profile 6) and the strains isolated from the individual milk samples of two animals (t127 and PFGE profile 7), even though the strains are genetically correlated and belong to the same cluster ( Figure 1). Surprisingly, the strain isolated from the operator from farm 4 (not correlated with farm 3) had the same genetic profiles (t127 and PFGE profile 7) as the two strains isolated from the individual milk samples of animals from farm 3. Furthermore, when all the isolates with different biotyping results but belonging to cluster B were compared, the MRSA strains were present at all four farms (1, 2, 3, and 4) and were isolated from different samples, including the cheeses ready for commercialization (farms 1 and 4). Interestingly, four isolates (ID 2, ID8, ID11, and ID15) had spa-type t1773 and one carried the mecA gene (negative for the mecA LGA251 ), contrary to previous studies [39][40][41][42] that reported spa-type t1773 isolates positive for mecA LGA251 . In those studies, the strains were isolated from cats (skin) [41,42] and humans [39,40]. In the present study, the ID2 MRSA isolate (mecA positive) was isolated from bulk tank milk (Table S1) and had a human origin according to the biotyping scheme. The other three strains were isolated from an individual milk sample (ID8), mammary pustules (ID11), and skin swab (ID15) from the same animal and had ovine origin according to the biotyping scheme.
S. aureus usually shows limited host specificity, though transfer between different host species may occur [43]. The transmission of milk-associated S. aureus strains between cows and humans was suggested by Lee [44] who found MRSA in milk samples with comparable antibiotypes as those in humans, but the transfer to humans was not demonstrated as, for example, the transmission from ovines to humans and vice versa. While molecular epidemiological typing did not reveal the spread of MRSA from animals to humans, it did reveal the spread of MRSA strains among ovines on two farms.
It should be emphasized that the study of S. aureus populations isolated from small ruminants holds importance since information is scarce. Recent detection of new spa types [45] and the pvl strain in this study suggests that the characterization of S. aureus isolates from ovine and caprine livestock should be encouraged.
Detection of MRSA in sheep herds is unexpected, given that the farm management practices differ from those for cattle. In fact, in small ruminant herds, breeders often prefer culling to antimicrobial treatments, resulting in a less selective pressure on S. aureus populations.
The presence of MRSA in farm animals is a serious health hazard for the community and could pose a difficult challenge for treating infections in humans. In Italy, Alba and colleagues observed that the MRSA strains isolated from cattle have a very high genetic relatedness with human isolates [20]. Therefore, it might be important to genetically correlate strains isolated from sheep farms to human isolates in the same area, investigating genomic features, including the arginine catabolic mobile element (ACME), which was identified in a specific clone of MRSA, called USA300 [46]. Lastly, the investigation of these genes can be useful for control the spread of possible new MRSA clones, studying their genetic evolution and the horizontal acquisition of novel genetic elements of strains isolated in different host species.

Conclusions
Characterization of SE genes in strains isolated in dairy farms is helpful for the indirect monitoring of eventual enterotoxigenic strains present in the food chain. The enterotoxin G and H genes exhibited the highest prevalence in the MRSA isolates in this study. Since there are no methods available for the identification of these enterotoxins in foodstuffs, an underestimation of the risks for humans can be assumed [47][48][49]. In addition, the potential risk for antimicrobial resistance in the food chain has to be taken into account [47][48][49]. On the other hand, the reported enterotoxin type C gene found in the MSSA, but not in the MRSA isolates, confirmed its diffusion in ovine isolates, as previously reported [28,50,51]. The description of an isolate harboring the gene for pvl poses a risk for the dissemination of this toxin in sheep herds. There are no directives for mandatory screening of this virulence factor in food sources. The results of this study highlight the need for more detailed investigations to improve consumer protection and public health safety.

Sample Collection
A total of 51 S. aureus strains were collected from six dairy sheep farms, including two farms with artisan dairy facilities, all located in Latium, Italy, between 2013 and 2015. The farms had a history of MRSA in bulk tank milk, as documented in a previous survey carried out between 2012 and 2015. Five farms operated a semi-extensive breeding system, two of which also had an artisan cheese factory, while the sixth farm operated an intensive system. The circulation of MRSA and of MSSA was investigated in four of the six positive farms. The flock size ranged from 83 to 556 lactating ewes. The ewes were milked twice daily using a milking machine. The workers wore dedicated coveralls and boots, but not gloves, when handling the animals. Teat-washing before milking and treatment with Toxins 2017, 9, 161 7 of 11 antibiotics at dry-off were not performed. The dairies made raw milk cheeses, and the farm owners were usually involved in milking and cheese making.
Individual milk samples were collected from all lactating ewes in all four herds. At the end of milking, bulk tank milk was collected. Nasal and skin samples were taken using cotton-tipped swabs (Amies Agar Gel with Charcoal, Laboindustria s.p.a., Italy) and Sodibox ® wipes (Sodibox, Névez, France), respectively, from the farm owners and workers. None of the subjects reported recent hospitalization or the presence of a healthcare worker in the household. Two raw cheese samples were taken from each dairy plant. All samples were immediately transported to the laboratory in ice-cooled containers and analyzed within 24 h.
Overall, the isolates were obtained from udder milk (n = 17), mammary pustules (n = 2), bulk tank milk (n = 11), nasal swabs of workers (n = 8), skin swabs of workers (n = 4), skin swabs of ewes (n = 4), environmental swabs (n = 3), and raw cheese (n = 2). Isolates were confirmed as S. aureus by modified species-specific PCR, using primers targeting the femA gene (132 bp) [52] and screened for methicillin resistance using the disk diffusion method according to Clinical Laboratory Standards Institute (CLSI) criteria. All human samples were obtained voluntarily, and the farm owner consented to animal sampling (in Italy, MRSA infection in animals is not a notifiable disease). All procedures followed were in accordance with ethical standards of the relevant national and institutional committees on experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Farm workers gave oral informed consent to participate in the study.

Typing of the Bacterial Isolates
All isolates were analyzed by phenotypic tests for growth on Baird Parker rabbit plasma fibrinogen agar (Merck, Darmstadt, Germany). DNA was extracted using the InstaGene™ Matrix (Bio-Rad, Milan, Italy). Staphylococcal enterotoxin genes (SEs) were detected using two multiplex PCR (mPCR) assays according to the protocols of the European Union Reference Laboratory for Coagulase-Positive Staphylococci, including Staphylococcus aureus (EU-RL CPS) (ANSES-Mesons Alfort, France) [53]. The first mPCR assay detects genes encoding SEs A to E and R; the second detects genes encoding SEs G to J and P. Reference strains of S. aureus provided by the EU-RL CPS, FRI S6 (sea, seb), FRI 137 (seg, seh, sei), FRI 326 (see), FRI 361 (sec, sed, ser), and HMPL 280 (seg, sei, selj, selp) were used as positive controls. The biotyping scheme described by Devriese [54] was utilized to assign specific ecological variants (ecovars) associated with different host species; the method includes the production of β-hemolysin, bovine plasma coagulation, staphylokinase, and crystal violet growth reaction. The possible results are: bovine, human, ovine, and six different non-specific hosts (NSH1 to NSH6) [8].
All S. aureus isolates were analyzed for the presence of mecA, the new mecA homologue gene (mecC, formerly named mecA LGA251 ), and Panton-Valentine Leukocidin (PVL) (lukS-PV/lukF-PV) genes by PCR, as described previously [39]. PCR assays were performed on a GeneAmp System 9700 thermal cycler (Applied Biosystems, Foster City, CA, USA). All isolates were typed by DNA macrorestriction digestion analysis using a SmaI restriction enzyme and pulsed-field gel electrophoresis (PFGE) according to the EU-RL CPS protocol [55] modified as follows. Briefly, electrophoretic separation of DNA fragments was performed on 1% agarose gel; electrophoresis was performed using the CHEF MAPPER System (Bio-Rad Laboratories, Hercules, CA, USA); the gel was stained with GelRed (Biotium, Hayward, CA, USA) 1X solution in ultrapure water for 30 min. The restriction profiles were analyzed with BioNumerics software ver. 7.1 (Applied Maths, Sint-Martens-Latem, Belgium) using the Dice correlation coefficient and then visualized as a dendrogram by the unweighted pair-group method using arithmetic averages (UPGMA), with 1% tolerance and 1% optimization settings. Similarity cutoff values of 90%, 75%, and 70% were used to define a cluster. The polymorphic X region of the protein A gene (spa) was sequenced to assign the spa-type [56] using BioNumerics software ver. 7.1.

Supplementary Materials:
The following are available online at www.mdpi.com/2072-6651/9/5/161/s1, Table S1: Type of sample and number of the farm where the strain was isolated. Author Contributions: G.M. conceived the study, designed the experiments, performed the analysis, analyzed the data and wrote the paper. G. Giacinti analyzed the data, coordinated the sampling and wrote the paper. A.B. drafted the experimental section and analyzed the data. S.G. and D.M.B. assisted with data analysis. D.S., N.M. and G. Giangolini assisted with sampling and performed the microbiology analysis. L.D. and S.A. coordinated the investigation and critically reviewed the manuscript. All the authors have read and approved the final manuscript.

Conflicts of Interest:
The authors declare no conflict of interest.