Methicillin-resistant Staphylococcus aureus (MRSA) is a major cause of nosocomial and community-associated infections worldwide. Most cases of community-associated MRSA (CA-MRSA) have been associated with skin and soft-tissue infections in previously healthy individuals [1, 2]. Since 2003, pigs [3–7] and other animals such as horses [8, 9], poultry [10] and calves [11] have been identified as a new reservoir for CA-MRSA. Most of the livestock related MRSA strains share the same multi locus sequence typing (MLST) type, namely ST398. Throughout Europe [9, 12–14], Canada [6] and in the United States [15] ST398 has been found in association with animal husbandry, indicating a worldwide clonal lineage. Although the clinical importance of ST398 is still controversial, there are reports indicating transmission and infections among humans [16–18]. Pulsed Field Gel Electrophoresis (PFGE) using Sma I is considered to be the gold standard for typing MRSA isolates [19]. When PFGE was performed on ST398 isolates, no banding patterns could be generated, due to methylation of the Sma I site [20]. Therefore, ST398 isolates are referred to as PFGE non-typeable (NT_{Sma I})-MRSA. Some years ago staphylococcal protein A (spa) typing was introduced as a highly discriminatory typing method to characterize S. aureus isolates [21, 22]. However, spa-typing of the ST398 isolates revealed very limited variation within this group and 80% of our ST398 isolates had either spa-type t011, t108 or t034 [23]. Recently, a multiple-locus variable number of tandem repeat analysis (MLVA) has been presented [24]. Although MLVA is significantly more discriminatory than spa-typing, it was unable to yield a better discrimination of the isolates of the ST398 lineage. The lack of a typing method that can discriminate ST398 strains has hampered studies on the origin and transmission routes of this MRSA clade.

In the Netherlands all first MRSA isolates obtained from patients with staphylococcal disease and from patients that carry the pathogen are sent to the National MRSA reference centre for typing. In 2007, 30% of all forwarded MRSA isolates were NT_{Sma I}-MRSA [23].

Recently, a neoschizomer of Sma I, designated as Cfr9 I, was shown to be insensitive for the DNA-methylation leading to NT_{Sma I}-MRSA isolates. In two studies this restriction enzyme was used for generating PFGE profiles of NT_{Sma I}-MRSA isolates [18, 25]. In the study presented here we optimized PFGE with restriction enzyme Cfr 9I and evaluated its use to characterize NT_{Sma I}-MRSA isolates.

The data will yield important information about the genetic diversity of the ST398 clonal lineage in the Netherlands and demonstrates that Cfr 9I PFGE is a powerful tool to study possible transmission and outbreaks of MRSA isolates, previously not typeable by conventional PFGE approaches.

MRSA isolates belonging to the ST398 clonal lineage are hard to discriminate based on spa-typing and/or MLST, hampering the assessment of transmission and outbreaks. Therefore, other techniques such as a modified PFGE could provide a new opportunity to differentiate ST398 isolates. The restriction enzyme Sma I does not cut the DNA of NT_{Sma I}-MRSA isolates, due to methylation of the Sma I site. However, Cfr 9I, a neoschizomer of Sma I, can be used for generating PFGE profiles of the NT_{Sma I}-MRSA isolates. When the standard Sma I protocol was used for Cfr 9I, banding patterns with smears and partial digests appeared. Other recently published articles seemed to have encountered similar problems with their Cfr 9I PFGE [18, 25]. The results indicated that lysis of ST398 isolates and digestion with restriction enzyme Cfr 9I is more cumbersome than lysis of typeable MRSA and digestion with Sma I [29]. After modifying the protocol, banding patterns of similar quality as those of typeable MRSA isolates digested with Sma I were obtained. All previously non-typeable MRSA isolates can be typed with the optimized PFGE method providing a new opportunity to differentiate the ST398 clonal lineage.

From April 2002 until January 2008, all MRSA isolates sent to the RIVM have been typed with PFGE using Sma I as restriction enzyme creating a database with more than 4000 isolates with over 700 different PFGE types. Since Cfr 9I recognizes the same restriction site as Sma I, Cfr 9I enables analysis and comparison of the patterns with other profiles in our database. No comparison was found when comparing banding patterns of NT_{Sma I}-MRSA with known PFGE patterns, suggesting that Sma I restriction modification is confined to a defined clonal lineage. Recently, ST398 isolates were typed using amplified fragment length polymorphism (AFLP). These data also suggested that ST398 is a distinct cluster recently introduced into the Dutch patient population [30].

The PFGE patterns of the two most prevalent spa-types (t011 and t108) within the NT_{Sma I}-MRSA isolates showed more variation than spa-typing or MLST. The genetic diversity within the ST398 clonal lineage of MRSA sharing the same spa-type creates an opportunity for improved investigation of outbreak and potential transmission events. Spa-typing, which is currently used as a MRSA typing standard, cannot differentiate these isolates further. Using Cfr 9I PFGE, spa-type t011 seemed to be more diverse than t108. Although the minimal similarity of the t108 isolates was 50%, this was mainly caused by a single isolate with a very distinct PFGE pattern (pattern H). Without this isolate the minimal similarity of the t108 isolates was 80%. The t011 isolates showed a minimal similarity of 64% (data not shown). SCCmec typing showed an almost equal distribution between SCCmec type IV (n = 14) and V (n = 16) for t011 isolates, whereas all t108 isolates carried SCCmec type V or a SCCmec type V variant. Huijsdens and colleagues performed SCCmec typing on 300 NT_{Sma I}-MRSA isolates and they showed similar results [23]. This variation in SCCmec types may also indicates a higher diversity among t011 MRSA isolates compared to t108 isolates.

The minimal similarity of the Cfr 9I PFGE patterns among ST398 isolates was 35% and showed variation within spa-types, but the diversity within this lineage is still limited. Furthermore, one isolate with spa-type t108 yielded a very distinct PFGE pattern which causes the similarity to be 35% (figure 1). When excluding this isolate from the dendrogram the minimal similarity was 62%. Comparing the PFGE results using the criteria by Tenover et al. and when a similarity cut-off of 80% was applied, most NT_{Sma I}-MRSA isolates should be classified as one PFGE cluster [31, 32]. However, the Cfr 9I PFGE is still better in discriminating possible differences between NT_{Sma I}-MRSA isolates.

No geographical relation could be found in either spa-type. However, most NT_{Sma I}-MRSA isolates are found in areas with the highest pig density. This could be explained by the frequent movement of pigs between farms in the Netherlands. This facilitates the dissemination of ST398 MRSA on a national scale. A similar situation took place during the foot- and -mouth epidemic in England of 2001 [33].

To provide additional resolution on the molecular evolution and dissemination of MRSA lineages, several typing techniques such as PFGE, SCCmec- and spa-typing have been developed. Since PFGE with Sma I does not digest the DNA of ST398 isolates, spa-typing has been the method of choice for characterizing NT_{Sma I}-MRSA isolates. However, given the low diversity in spa-types it is hard to ascertain health care-associated transmission if two or more different spa-types are present in the same institution. Fanoy et al. described an outbreak in a residential care facility where two spa-types (t2383 and t011) were prevalent [18]. After re-examination of the same isolates the PFGE profiles using Cfr 9I were indistinguishable, indicating isogenicity. Moreover, the discriminatory ability of spa-typing of NT_{Sma I}-MRSA is compromised by the fact that more than 80% of the NT_{Sma I}-MRSA in the Netherlands belong either to spa-type t011 or t108 [23]. With the modified Cfr 9I PFGE a better tool for epidemiological investigation has become available.

The results obtained by Cfr 9I PFGE of isolates from veterinarians and their close family members showed possible transmission of ST398. Five out of eight pairs had identical profiles. The family members had themselves no contact with animals and were presumably infected by the occupationally exposed veterinarian. Two pairs of PFGE patterns among family members were not identical. Their isolates also had different spa-types. Family members may have been colonized by one MRSA through the veterinarian and subsequently the veterinarian may have been re-colonized by another MRSA after occupational exposure. One pair differed only in a single PFGE band probably as a consequence of micro-evolution.

A study on nine different farms revealed that the PFGE patterns of isolates from seven farms were related, but PFGE patterns varied within and between the farms. For example, farm 7, yielded only 2 very closely related PFGE patterns (D14, D21; similarity 95%), while other farms, like farm 8, showed 5 different PFGE patterns (B1, D1, D3, D4 and K) and had a similarity of only 66%. Different batches of animals entering the farm, carrying different NT_{Sma I}-MRSA, could have caused variation within farms. Further study is needed to confirm that farms with a fast turnover of pigs indeed show a higher diversity of PFGE patterns of NT_{Sma I}-MRSA.

In conclusion, the modified PFGE protocol for Cfr 9I provided highly informative banding patterns and showed good reproducibility. The PFGE results showed diversity within and between the two most prevalent spa-types among NT_{Sma I}-MRSA. PFGE confirmed transmission of the ST398 clonal lineage within families and in a residential care facility. The modified PFGE approach can be used as a method for selecting important and distinct ST398 isolates for further research. The adjustments in the PFGE protocol using Cfr 9I are easy to implement in laboratories which already have a PFGE facility, creating a powerful tool to study the ST398 clonal lineage.