Livestock-associated Methicillin-Resistant Staphylococcus aureus Sequence Type 398 in Humans, Canada

Recent emergence of infections resulting from this strain is of public health concern.

H igh prevalence of colonization with livestock-associated (LA) methicillin-resistant Staphylococcus aureus (MRSA) sequence type (ST) 398 among pigs and pig farm-ers was fi rst reported in the Netherlands (1) and has since been identifi ed in Canada (2) and the United States (3). In Canada, this LA-MRSA strain was identifi ed in pigs and pig farmers in southwestern Ontario, where prevalence of MRSA colonization was 24.9% (71/285) and 20% (5/25), respectively (2). In the United States, nasal samples from 20 production system workers and 299 swine from 2 farms in Illinois and Iowa showed that 45% (9/20) and 49% (147/299), respectively, were colonized with LA-MRSA (3). Despite such high prevalence of MRSA colonization on these tested farms, to our knowledge, no human or animal infections resulting from LA-MRSA strains have been reported in North America.
To determine whether LA-MRSA has recently emerged in the general population of Canada, we identifi ed human infections and colonizations associated with the LA-MRSA strain in Canada and molecularly characterized the isolates. We also identifi ed a novel staphylococcal cassette chromosome (SCC) mecV subtype harboring clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated genes (cas).

Materials and Methods
A convenience sample, totaling 2,358 MRSA isolates from human specimens, was submitted to the National Microbiology Laboratory (NML) for spa typing, as described (4,5). During January 2007-October 2008, the Saskatchewan Disease Control Laboratory submitted 2,008 specimens; during October 2007-August 2008, the Cadham Provincial Laboratory in Manitoba submitted 350 specimens. An additional 1,329 isolates from human specimens were spa typed by the Saskatchewan Disease Control Laboratory.
Given the client base of the Cadham Provincial Laboratory, most of these isolates would have originated from colonized and infected persons living in the community or in personal-care homes or from persons hospitalized in smaller community hospitals, whereas, for surveillance purposes, the Saskatchewan Disease Control Laboratory receives isolates from all colonized and infected persons across the province. Detailed information regarding why cultures were taken (e.g., screening admissions, outbreak investigations) and other clinical and epidemiologic data were limited. Isolates typed in this study represented ≈17% of all MRSA isolates from persons in Manitoba and ≈66% of all MRSA isolates from persons in Saskatchewan within the study period. An additional isolate was sent to the NML from Sunnybrook Health Sciences Centre in Ontario for reference purposes.
Isolates with spa types associated with ST398 were confi rmed by multilocus sequence typing; tested for Panton-Valentine leukocidin toxin, mecA, and nuc genes; and typed for SCCmec as described (6)(7)(8)(9). Pulsed-fi eld gel electrophoresis (PFGE) of SmaIor Cfr91-digested genomic DNA was conducted as described (10). Antimicrobial drug susceptibility testing was conducted by using standard broth microdilution panels according to Clinical and Laboratory Standards Institute guidelines (11). Breakpoints for fusidic acid and mupirocin resistance, which were not provided in the guidelines, were as described (12,13).
A fosmid library was constructed by cloning sheared genomic DNA from S. aureus isolate 08 BA 02176 into the pCC2FOS vector. The fosmid clones were screened by PCR to identify specifi c genes orfX, mecA, and the chromosomal region located downstream of SCCmec. Fosmid clone 1G1 was identifi ed and contained the entire SCCmec region of the 08 BA 02176 isolate. Colonies were prepared by using the CopyControl Fosmid Library Production Kit (Epicenter Biotechnologies, Madison, WI, USA) according to the manufacturer's instructions. Fosmid DNA was column purifi ed by using a QIAGEN Plasmid Mini Kit (QIA-GEN, Valencia, CA, USA).
DNA sequencing was performed on the ABI3730xl genetic analyzer (Applied Biosystems, Foster City, CA, USA). Staden (Pregap4) software was used to prepare trace data for sequence assembly (14). Sequencing reads were assembled by using the Staden Gap4 program. Gap closure was achieved by primer walking and long-range PCR. Specifi c primers were designed near the ends of neighboring contigs (contiguous sequence of DNA created by overlapping sequenced fragments of a chromosome), and PCRs were performed with chromosomal template DNA. Regions containing putative frameshifts and point mutations were resequenced to verify the fi delity of the sequence.

LA-MRSA Characterization
A total of 3,687 MRSA isolates were examined; 5 contained ST398-associated spa types (4 t034 and 1 t1250). The additional isolate submitted to NML by Sunnybrook Health Sciences Centre in Ontario, isolate T40929, also contained a t034 spa type. Further molecular characterization of these 6 isolates determined that they were all ST398, SCCmecV, and negative for the Panton-Valentine leukocidin-encoding genes ( were resistant to tetracycline, but all were susceptible to the other 12 antimicrobial drugs tested ( Table 2). From the surveillance in Manitoba and Saskatchewan, patient information was limited and showed no geographic links (all 5 persons resided in different health regions but were all within the southeastern portion of Saskatchewan) ( Figure 1). Of the 5 isolates, 4 were obtained from infected persons (average age 67.8 years, range 51-79 years) ( Table  1). The earliest identifi ed LA-MRSA isolate (08 BA 2176) associated with an infection was obtained from a postoperative surgical site. Further follow-up was not possible because of the patient's health problems. This patient is unlikely to have had any recent direct contact with livestock because she had been confi ned to her home with limited mobility for several years before her hospitalization. Additional nasal swabs from this patient remained positive for this strain for at least 7 months. Additional clinical and epidemiologic information for the remaining 3 patients with skin and soft tissue or wound infections were limited ( Table 1).
The isolate submitted to the NML by Sunnybrook Health Sciences Centre, outside the surveillance program, was from a 59-year-old man from Ontario. He had been hospitalized in December 2007 for treatment of metastatic squamous cell carcinoma of the larynx. In the previous year, he had undergone a total laryngectomy, neck node dissection, and tracheostomy. A MRSA isolate was recovered from screening specimens from his nose and the tracheostomy site that had no indication of infection. He was unaware of any animal contact and had no history of exposure to pigs or pig farms. A review of the medical records and standard epidemiologic investigations determined that this was not a nosocomial or healthcare-associated isolate.
The 6 LA-MRSA isolates were nontypeable by PFGE using SmaI. However, PFGE using the neoschizomer Cfr91 showed that the 6 LA-MRSA isolates were closely related ( Figure 2, panel A). Control MRSA strains digested indi-vidually with either Cfr91 or SmaI showed no differences in fi ngerprint banding patterns when the 2 enzymes were compared (data not shown), which enabled comparisons   Table 3). This element carried a class C2 mec complex, which putatively contained a non- Class C2 mec ccrC2   (Table 4) and illustrated as solid lines. Shown below the schematic is the DNA sequence of the confi rmed 1,107-bp CRISPR array in the J1 region, which provides the 36-bp direct repeat consensus (boldface) and the variable 15 spacer sequences.
functional IS431 transposase and a type 5 ccr gene complex (ccrC2). Other than mecA, no additional antimicrobial drug resistance genes were identifi ed within this element. The fi rst unique feature of this SCCmecV element included 3 ORFs in the J3 region sharing high sequence identity with ORFs from S. epidermidis RP62A (Gen-Bank accession no. CP000029), which included an ADPribosylglycohydrolase, a permease for cytosine/purines, and a ribokinase (Table 3). A second unique feature was a CRISPR array, identifi ed by using CRISPRFinder (16), in the J1 region, which appears to have replaced the type 1 restriction modifi cation system (hsdR, hsdS, hsdM) through recombination. The CRISPR array (1,107 bp) contained a 36-bp direct repeat consensus and 15 spacers of variable sequence and length (33-38 bp) (Figure 3). Downstream of this CRISPR array was a combination of putative CRIS-PR-associated (cas) genes, sharing sequence identity with those previously identifi ed in S. epidermidis RP62A. This array was followed by a second questionable CRISPR array (183 bp) containing a 38-bp direct repeat consensus and 2 spacers of variable sequence (Figure 3; Table 3).
Design of primers spanning the entire SCCmec element was based on the DNA sequence obtained from 08 BA 02176 (Figure 3; Table 4). PCR of these select regions produced amplicons of expected size for 3 additional LA-MRSA isolates (07 BA 06477, 08 BA 13895, 08 BA 22334) but were negative for some of the J1 and J3 regions in 08 BA 08100 and T40929 (Table 4).

Discussion
The high prevalence of LA-MRSA colonization of pigs and pig farmers in Canada (2) and the United States (3) and this report of human infections suggest that this LA-MRSA strain from Canada poses potential public and occupational health concern in North America. This strain has been associated with various types of infections in pigs (17,18) and humans (19,20) and is transmissible from animal patients to veterinary workers (21), healthcare workers (22), and family members (1). Evidence also suggests that this strain might be spreading from animals to the environment, which may facilitate the colonization or infection of persons who are not involved in animal husbandry (23). Whereas in 2006 in the Netherlands LA-MRSA accounted for >20% of all MRSA isolated (24), carriage of this strain in the general population of 2 provinces in Canada (Manitoba and Saskatchewan) appears rare (0.14%). This difference could be attributed to the substantially higher density of pigs in the Netherlands (1,244 pigs/km 2 ) than in Manitoba (55 pigs/km 2 ), Saskatchewan (6 pigs/km 2 ), and Ontario (91 pigs/km 2 ) (www.agriculture.gov.sk.ca/ Pig_Densities). It is also plausible that the much lower proportions of LA-MRSA in Canada, relative to a country with low MRSA endemicity such as the Netherlands, is attributable to competition with other highly successful human epidemic MRSA clones circulating in Canada, including CMRSA2 (USA200/800), CMRSA7 (USA400), and CMRSA10 (USA300) (25,26). The tested LA-MRSA isolates were highly susceptible to most classes of antimicrobial drugs, except β-lactams and tetracyclines, the latter of which has been attributed to its high usage in animal husbandry (27). The complete sequence of the SCCmec region showed a novel SCC-mecV subtype sharing sequence identity in its J1 and J3 regions with chromosomal genes in the S. epidermidis RP62A chromosome (GenBank accession no. CP000029), including a CRISPR system. CRISPRs and associated cas genes are present in many other bacterial (≈40%) and archaeal (≈90%) genomes (28,29) and have been shown to be involved in sequence-directed immunity against phages (30,31) and plasmids (32). The resistance to plasmids and phages encoded by this system could explain why many of these ST398-MRSA-V strains contain fewer antimicrobial drug resistance genes and phage-encoded virulence factors than do other epidemic MRSA strains (33,34). The origin of this CRISPR system is unknown, but the propagation of CRISPR loci throughout prokaryote genomes has been proposed to occur through horizontal gene transfer by conjugation of megaplasmids >40 kb (35). Because the CRISPR system identifi ed in this study is encoded within a putative mobile genetic element, we propose that an additional mechanism of mobilization to other methicillin-susceptible Staphylococcus spp. is plausible.
This novel subtype of SCCmecV was found in only 4 of the 6 LA-MRSA isolates identifi ed in this study. One isolate not containing this novel SCCmec subtype (08 BA 08100) could also be distinguished by a different but closely related spa type (t1250) ( Table 1)    isolates, which suggests that at least 2 epidemiologically different strains of LA-MRSA circulate in Saskatchewan. The other LA-MRSA isolate that did not contain this novel SCCmec element was obtained in Ontario. However, this isolate was the same spa type (t034) and was closely related, according to PFGE, to the LA-MRSA isolates identifi ed in Saskatchewan. Therefore, in addition to PFGE and spa typing, SCCmec subtyping could provide a useful epidemiologic tool for surveillance, outbreak investigations, or traceability studies of this emerging strain. For detection of this SCCmecV subtype (tentatively designated V.2.1.2; Vb), we propose using primer set 1 (spanning orfX into Sk02 in the J3 region) and primer set 7 (spanning Sk20 into cas1 in the J1 region) ( Table 4). Visual comparison of PFGE fi ngerprints from this study with those reported from patients from the Dominican Republic and the United States (northern Manhattan, New York, NY) (36), showed substantial variations in fi ngerprint patterns, as well as related but different spa types. These variations suggest further molecular and geographic diversity of these LA-MRSA strains on a global scale.
Because cases of LA-MRSA infections have only recently been identifi ed in Canada, additional surveillance efforts are required to monitor the emergence and clinical relevance of this MRSA strain in Canada, including communities, the environment, livestock, farmers, and production facility workers. Whether these strains pose a major threat to human health in light of the low livestock density and continued spread of epidemic hospital and community strains of MRSA in Canada remains unknown.