Transmission of highly virulent community-associated MRSA ST93 and livestock-associated MRSA ST398 between humans and pigs in Australia

Pigs have been recognised as a reservoir of livestock associated methicillin-resistant Staphylococcus aureus (LA-MRSA) in Europe, Asia and North America. However, little is known about the presence and distribution of MRSA in the Australian pig population and pig industry. This study describes the presence, distribution and molecular characteristics of the human adapted Australian CA-MRSA ST93 isolated from pigs, people, and the environment within a piggery. Isolates were subjected to antibiotic susceptibility testing, DNA microarray, whole genome sequencing, multi locus sequence typing, virulence and resistance gene characterization and phylogenetic analysis. MRSA were isolated from 60% (n = 52) of farm workers where 84% of isolates returned ST93 and the rest ST398. Of the thirty-one pig isolates tested further, an equal number of ST398 and ST93 (15 each) and one as ST30-V were identified. Four of six environmental isolates were identified as ST93 and two as ST398. This study has identified for the first time in Australia the occurrence of CA-MRSA ST93 and LA-MRSA ST398 amongst farm workers, pigs, and the farm environment. Comparative genome analysis indicates that ST398 is likely to have been introduced into Australia from Europe or North America. This study also reports the first linezolid resistant MRSA isolated in Australia.


Sampling.
A cross-sectional study was performed at a commercial pig enterprise in Australia to detect the presence or absence of MRSA in people, pigs and the environment. The facility was identified by the farm's jurisdictional health department due to the recurrent detection of MRSA amongst farm workers with clinical staphylococcal disease. The farm consisted of two sites (designated Farm-A and Farm-B) geographically separated by approximately 40 km. Sampling at each site occurred over a two-day period, with Farm-A visited in May and Farm-B in August 2015.
All farm workers (n = 52) were approached to participate in the study. Participation was voluntary, and signed informed consent was obtained from each participant. A sterile cotton applicator swab (Liquid Amies Elution Swab, Copan ESwab ™ ) was provided and each participant was instructed to rotate the swab in both nostrils for five seconds, before placing the swab in a covered sterile transport tube.
On both farms, pigs were housed in separate sheds based on their age group and class (i.e. dry sows, farrowing, weaners, growers, and finishers). On Farm-A, to optimise sensitivity of the testing procedure, swabs from pigs were collected from the skin caudal and adjacent to the pinnae and from the external nares 31 . With the aim of determining if MRSA was present in the pigs on Farm-A, a previously described pooled sampling method was used: to detect MRSA carriage at a prevalence of at least 2% and assuming 90% test sensitivity with 95% confidence the minimum number of swabs per pool and the minimum number of pools required was calculated to be 10 and 17 respectively, using Epitools online 32 http://epitools.ausvet.com.au. In addition, in order to obtain estimates of prevalence specific for each shed, six pools (60 swabs) were collected from each shed 33,34 . From the seven sheds on Farm-A, 420 swabs were collected, providing 42 pools. Animals were randomly selected by farm workers and animals were marked after sampling so re-sampling of the same animals could not occur.
As a large number of MRSA positive pools were found on Farm-A, the sampling procedure was refined for Farm-B. Swabs were collected from the external nares only and samples were processed individually, rather than being pooled. As the prevalence was unknown (and therefore set at 50%) and considering a precision of 5% and 95% confidence, a minimum sample size of 385 swabs was required to determine the prevalence of MRSA in the pigs on this farm. Subsequently, nasal swabs were collected from 408 pigs of different stages of production housed in 13 sheds.
Using established protocols for dust sampling, environmental samples were collected and pooled (five swabs in each pool) from all sheds and effluent collection ponds on both farms 35 . Samples were collected from inside each shed and included walk ways, pens floors, feeders, fences, and walls. Swabs were also collected from the human environment (offices, showers, toilets, and kitchens/sitting rooms). Laboratory analysis. All swabs were transported to the microbiology laboratory on ice within 12 hours of collection and immediately refrigerated upon arrival. Swabs were processed within a week of arriving in the laboratory using the isolation method recommended by the European Union Reference Laboratory for AMR (EURL-AMR) for the testing of MRSA in food-producing animals and food samples 36, 37 . MRSA isolation. Swabs were pre-enriched in Mueller-Hinton broth (BD ™ Difco ™ ) containing 6.5% sodium chloride for 24 hours at 37 °C. Post incubation selective enrichment was performed by transferring one ml of pre-enriched broth into Tryptone Soy Broth (CM0129, Oxoid ™ ) containing 3.5 mg/L cefoxitin and 75 mg/L aztreonam and incubating overnight at 37 °C. Following incubation, a loopful of the selective enriched cultured was inoculated onto chromogenic MRSA agar (CHROMagar ™ MRSA) and incubated for 24 hours. As per the manufacturer's instructions, rose to mauve coloured round colonies were preliminarily identified as MRSA. A single presumed MRSA colony from chromogenic agar was plated onto 5% sheep blood agar and incubated for 24 hours at 37 °C. Cultures were subsequently identified as S. aureus based on the Gram stain, catalase production and the S. aureus Protein-A latex agglutination test (staphylase, Oxoid ™ ). S. aureus ATCC 29213 was used as the control strain.
Linezolid non-susceptibility was confirmed by minimum inhibitory concentration (MIC) testing using Etest ® strips according to the manufacturer's recommendations (bio Mérieux).

Molecular characterisation.
Molecular testing was performed on 68 MRSA isolates which included the 31 MRSA isolated from humans, and a subset of the pig (n = 31) and environmental (n = 6) MRSA isolates. Pig and environmental MRSA isolates were selected based on their phenotypic antimicrobial susceptibility profiles to include as much phenotypic diversity as possible. The pig isolates were selected to include at least two isolates from each age group on both farms. nuc and mecA characterization. S. aureus species and methicillin resistance were confirmed by the detection of nuc (thermostable extracellular nuclease) and mecA (methicillin resistance) genes respectively using multiplex PCR 39 . DNA microarray. Isolates were initially characterised using a S. aureus DNA microarray (Alere Technologies, Jena Germany). Arrays and reagents were obtained from Alere Technologies, Jena Germany. The principle of the assay, related procedures, and a list of targets have been described previously 40 . The microarray was used to detect the presence of virulence and antimicrobial resistance genes. Probes for mecA, ugpQ, xylR, and two probes for mecR were used for SCCmec typing. The last two probes allowed detection and discrimination for untruncated mecR and mecR, respectively. Probes for the recombinase genes ccrA1, ccrB1, ccrA2, ccrB2, ccrA3, ccrB3, ccrA4, ccrB4 and ccrC1; the fusidic acid resistance marker Q6GD50; and the J region proteins, dcs, plsSCC and the kdp-operon also were included. Ambiguous array results were considered negative.
Genomic DNA libraries were prepared using the Nextera XT kit (Illumina) and sequenced on the 300-bp pair-ended chemistry. Raw sequences reads were denovo assembled using CLC genomics workbench (8.5.1).

Multilocus sequence typing (MLST).
DNA sequences collated at http://saureus.mlst.net/ belonging to 3,044 sequence types (ST) were downloaded in FASTA format and used as the database in the SRST2 pipeline to match the corresponding MLST profiles to the Illumina sequence data 41 . In-silico virulence and resistance gene characterization. The presence of virulence and resistance genes were confirmed using VirulenceFinder, ResFinder tools from Centre for Genomic Epidemiology (CGE) website (www.genomicepidemiology.org) from the FASTA files generated using Illumina sequencing. Additional virulence factors, resistance gene determinants, beyond those detected by VirulenceFinder and ResFinder were determined using CLC Genomics Workbench.
Comparative Genomics. The MRSA identified in this study were compared to previously published MRSA strains. For ST93 the reference genome JKD6159 (NC_017338) was used as a template and aligned against all contigs to determine SNP locations 42 . The comparative genomic analysis was performed using a comprehensive Australian collection of MRSA ST93 using the 519 SNPs previously reported 43 . The comparative genome analysis of MRSA ST398 was performed by aligning the ST398 genomes sequenced in this study against the reference genome HO 5096 0412 (HE681097) 44 . The reference genome was modified to exclude all phage and intergenic regions. An international collection of MRSA ST398 from a previous study was utilised to perform comparative genomics 23 . SNPs were obtained using Panseq 45 , Maximum parsimony phylogenetic trees, based on the SNPs, were generated for both strains using MEGA (V6.06) 46 .
Statistical Analysis. All analyses were performed using the R statistical package 47 . Data on inhibition zone size for each drug were converted to dichotomous classification of resistant or susceptible based on inhibition zone diameters recommended in the CLSI documents M100-S24 38 and VET01-S2 48 . All intermediate resistance isolates were considered as susceptible. Confidence intervals (CI) were calculated at the 95% level of significance. The 31 human MRSA were identified as either ST93-IV (26 isolates) or ST398-V (5 isolates). Of the 31 pig isolates typed further, 15 were identified as ST398-V, 15 as ST93 (14 harbouring SCCmec-type IV and one with SCCmec-type V), and one as ST30-V. ST93-IV (4 isolates) and ST398-V (2 isolates) were isolated in the environment.

Isolation
On Farm-A, all ten MRSA (100%) isolated from the human participants were characterised as ST93. The pig isolates were characterised as ST93 (n = 5) and ST398 (n = 3). ST93 and ST398 were also isolated in the environment.
On Farm-B, 16 of 21 MRSA (76%) isolated from the human participants were characterised as ST93-IV. The remaining five isolates were characterised as ST398-V (24%). Amongst the pig isolates four different MRSAs were characterised: 12 ST398-V, nine ST93-IV, and single isolates of ST93-V and ST30-V. MRSA ST398-V and ST93-IV were also identified in the environment.
Antimicrobial resistance phenotypes and genotypes. The antimicrobial resistance phenotype and clonal association are described in Table 1. In addition to the β-lactams, in ST93 and ST398 chloramphenicol, clindamycin, erythromycin, neomycin, and tetracycline, resistance was detected. Quinupristin-dalfopristin and linezolid resistance were also detected in ST398. The ST30 isolate was only resistant to the β-lactams and tetracycline.
A large proportion of ST398 were co-resistant to tetracycline, erythromycin, and clindamycin ( Table 2). Resistance to erythromycin (n = 18) and clindamycin (n = 19) was conferred by ermC (19/22). Tetracycline resistance (n = 22) was conferred by the carriage of the tetM, tetK, and tetL genes. Ten isolates carried the vgaA gene which encodes resistance to lincosamides, pleuromutilins and streptogramin A. The lincosamides resistant isolates also carried the lnuB gene. Chloramphenicol resistance (10 isolates) was conferred by the fexA gene and aminoglycoside resistance (11 isolates) by the aadD (2 isolates) and aadE (9 isolates) genes. Three ST398 MRSA harboured the cfr gene cassette; two from the pig samples and one from a human participant. Only the human isolate was phenotypically linezolid resistant (MIC 32 mg/L). Although the dfrG gene was identified in nine isolates, none were phenotypically co-trimoxazole (trimethoprim/sulfamethoxazole) resistant.
The ST30-V isolate was found to carry tetK, tetL and vgaA antimicrobial resistance genes.
Virulence gene data. The carriage of virulence genes was influenced by the MRSA type ( Table 2).
The ST30-V isolate harboured the egc complex. The isolate was negative for the PVL and IEC associated genes.

ST93 phylogeny.
Between the ST93 MRSA isolates very few SNP mutations were detected. The isolates were closely related to the reference genome JKD6159 (NC_017338) and shared nine common SNP markers. Based on the SNPs the isolates could be classified into three clusters (Fig. 1). There was no relation between the source of the isolates and the cluster.
ST398 phylogeny. The maximum parsimony tree constructed from ST398 isolates showed three clusters suggesting either multiple introductions of ST398 or an ongoing evolution of the strain. When compared to the international collection of ST398 MRSA genomes, the Australian ST398 isolates from the farmworkers and pigs were similar to the ST398 isolates from Europe and North America (Fig. 2).

Discussion
This study has investigated the carriage and the molecular characteristics of MRSA isolated from an Australian pig farm spanning two sites (Farm A and B) that had on-going clinical MRSA infections amongst its farm workers. The study has shown concurrent carriage of MRSA in humans and pigs on both farms with two predominant strains; the Australian CA-MRSA ST93-IV and the international LA-MRSA ST398-V. Furthermore, whole genome sequencing suggests not only has anthropozoonotic and zoonotic transmission occurred, the movement of ST93-IV from human to pig and then back into humans in the presence of the multi-resistant ST398-V has resulted in the acquisition of multiple antimicrobial resistance determinants, such as those encoding for resistance to tetracycline, clindamycin and chloramphenicol, that are not normally found in ST93-IV.
The three major findings arise from this study are as follow: First, we have identified for the first time in Australia the dissemination of CA-MRSA ST93-IV and LA-MRSA ST398-V amongst farm workers, pigs and the farm environment; Second, we have found, using comparative genome analysis, that ST398-V was likely to have been introduced into Australia from Europe or North America and the ST93-IV pig MRSA isolates came from humans with subsequent pig adaptation; and third, to the best of our knowledge we report the first linezolid resistant MRSA isolated in Australia.
ST93-IV, colloquially known as "Queensland CA-MRSA" was first described in the early 2000s in a population of Caucasian humans in Queensland 49 . Typically, ST93-IV has been associated with a range of skin and soft tissue infections in humans only. A singleton by MLST eBURST analysis, ST93-IV has been shown to be highly virulent    51 . Consequently, the acquisition of ST93-IV in pigs is likely to have been anthropozoonotic (i.e. human to animal). Although comparative genome analysis showed the porcine and human ST93-IV isolates in our study were closely related to ST93-IV previously characterised in the region ( Fig. 1 Supplementary data), the isolates had a lower than expected carriage of the bacteriophage encoded PVL-associated luk-F PV and lukS-PV genes. In previous studies upto 100% of ST93-IV have been identified as PVL positive 10,52 . In contrast, one-third of ST93-IV isolates in our study were PVL negative; 39% of farmworker isolates and 20% of pig isolates. Similarly, a lower than expected carriage of the фSa3 prophage-mediated type B human invasion gene cluster (IEC; sak/scn/chp) was observed. Unlike previous studies, which have shown 100% of ST93-IV harbor a type B IEC, the majority of the pig (80%) and the farmworker (58%) ST93-IV isolates lacked an IEC. The loss of the human associated virulence lukF-PV and lukS-PV and sak/scn/chp genes in a large proportion of the ST93-IV isolates suggests the strain has successfully adapted to the porcine host. A similar adaption for S. aureus moving from the human to porcine host has been reported in CC398 23 .
Contrary to the loss of virulence factors, the ST93-IV identified in our study were phenotypically multi-resistant (including chloramphenicol, clindamycin, erythromycin, neomycin and tetracycline) and harboured multiple resistance genes. Typically, ST93-IV is only beta-lactam resistant with only up to 20% and 2% of isolates resistant to erythromycin and ciprofloxacin respectively [52][53][54] . It is possible that the observed elevated frequency of resistance to non-beta lactam antimicrobials among the ST93-IV isolates in our study is due to selection within the pig environment, linked to antimicrobial use on the pig farm. The anthropozoonotic transmission of the highly successful CA-MRSA ST93-IV in pigs and the acquisition of additional antimicrobial resistance genes and loss of human associated virulence genes indicate the strain has successfully evolved in the porcine host. This successful evolution may lead to the co-selection, long term maintenance, and further adaptation of ST93-IV in pigs and in the piggery environment 55 .
Prior to our study, ST398-V has only been reported in Australia on two occasions 29,30 . In our study, 22 isolates were identified as ST398-V (representing 23% of MRSA isolates from farmworkers and also isolated from pigs and the environment). Previous studies have distinguished the livestock-associated S. aureus ST398 clade from the human clade by the presence of the tetM gene and the absence of IEC 56 . As the ST398-V isolated in our study lacked an IEC and a high proportion carried the tetM gene it can be assumed the isolates belong to the livestock clade 56 . Comparative phylogenetic analysis suggests the ST398-V was introduced into Australia from Europe or North America 56 . The identification of three ST398-V clusters suggests either multiple introductions into Australia have occurred or the strain has diverged within the farm environment. However, as one of the Scientific RepoRts | 7: 5273 | DOI:10.1038/s41598-017-04789-0 clusters (cluster 3 versus clusters 1 and 2) differed by more than 34 SNPs over a short period it is more likely that both events have occurred (Fig. 2 Supplementary data).
The importation of CC398 into a country previously free of CC398 and the subsequent transmission into the country's pig population followed by zoonotic spread is not unique 57 . In Europe, the spread of LA-MRSA between countries is often mediated by animal trading of piglets sold by specialised producers 58 . However, in Australia, the importation of live pigs ceased over 40 years ago. Consequently, it is possible that the introduction of ST398-V into Australia has occurred from pig farmers or pig veterinarians who have had contact with European or North American pig farms, or through the entry of other animal species, such as horses, where the same movement restrictions are not present.
Although S. aureus does not cause much illness in pigs, CC398 MRSA does cause a wide spectrum of infections in humans, ranging from relatively minor or localised infections to more serious or invasive infections 59 . Furthermore, nosocomial infections and outbreaks have occurred associated with this MRSA. However, similar to other S. aureus, transmission of CC398 MRSA is primarily mediated by physical contact and overall relatively few cases of CC398 MRSA have occurred in people who are not directly involved in livestock production. The low prevalence of CC398 MRSA in people not associated with pig farming is probably due to the low transmissibility of the organism 60 . Consequently, the clinical significance of CC398 MRSA within the community is thought to be low. However, the emergence of ST398 in the Australian pig industry is a public health concern for those working within the industry. Although ST398 MRSA is less virulent and less transmissible compared to other S. aureus, the organism exhibits co-resistance to many non β-lactam antimicrobials used in medical and veterinary practice including macrolides, tetracycline, trimethoprim, gentamicin, ciprofloxacin, trimethoprim-sulfamethoxazole, lincosamides and streptogramin B 57,59 . Although ST398 remains susceptible to the glycopeptides, daptomycin, tigecyclines, fusidic acid and rifampicin, isolated cases of linezolid resistance have been reported 57 . In ST398 MRSA, linezolid resistance is due to the acquisition of a plasmid harbouring the cfr gene which in addition to linezolid resistance mediates resistance to lincosamides, fenicols and pleuromutilins 61 . The selective pressure in favour of the spread of the cfr containing plasmid may be influenced by the use of linezolid in human medicine and florfenicol and tiamulin in veterinary medicine. The linezolid resistance identified in a ST398 MRSA isolated from a human in this study is the first instance of linezolid resistance in MRSA to be reported in Australia.
Although multidrug resistance may compromise the treatment of ST398 MRSA infections, the reservoir of resistance genes harboured by ST398 MRSA is more concerning from a clinical standpoint. This concern was recently highlighted by Brennan et al. following the introduction of CC398 into the Republic of Ireland's pig industry who suggested that the reservoir of resistance genes harboured by the less virulent and less transmissible CC398 MRSA could potentially spread to other animal and human strains 62 . Unfortunately, this may have Figure 2. Phylogenetic tree constructed by core genome SNPs of MRSA ST398 isolated from humans, pigs, and piggery environment from this study compared to isolates from different parts of the world. Each circle represents individual MRSAs. Isolates from pigs, humans, turkey meat, cattle, environment and horses identified by green, red, purple, orange, blue and dark green colours respectively. Isolates are coded by the country of origin and year in which they were isolated. Isolates from this study is coded as AUS (2015). occurred in Australia. It appears from our study the movement of the highly virulent and highly transmissible ST93-IV CA-MRSA from humans to pigs and the importation of ST398-V LA-MRSA into the Australian pig farms has resulted in the emergence of a multi-resistant PVL-positive ST93-IV strain which continues to harbour a type B IEC. However, a comprehensive molecular comparison of the antimicrobial resistance genes and the associated mobile genetic elements found in the ST93-IV and ST398-V is required. Furthermore, the fitness cost to ST93-IV following the acquisition of the additional genetic material will need to be determined in order to assess the human health risk, if any, to those animals and humans not exposed to the particular pig herds investigated in this study.

Conclusion
In conclusion, our study has demonstrated for the first time the widespread occurrence of the highly virulent Australian ST93-IV CA-MRSA and the international ST398-V LA-MRSA strains in an Australian piggery. Furthermore, the evidence suggests that anthropozoonotic and zoonotic transmission may have occurred. The presence of multiple antimicrobial resistance determinants that are not usually found in ST93-IV could be a consequence of the transmission of ST93-IV from humans to pigs and then back into humans in the presence of the multi-resistant ST398-V. The outcomes of this study indicate that in Australia, and probably elsewhere, greater scrutiny of staphylococcal infections in humans and animals is warranted in the form of coordinated investigation of outbreaks to identify the factors predisposing each host species to infection and emergence of virulent pathogens with novel resistance traits.