Diversity of MRSA SCC mec elements in Pretoria region of South Africa : The problem of variation in the assigned SCC mec types by different multiplex-polymerase chain reaction ( PCR ) methods and a call for an African consensus

The SCCmec element is one of the recommended targets for MRSA characterization and several multiplex-PCR SCCmec typing methods have been developed over the past years. However, there are no data on the consistency of the SCCmec types in clinical isolates as detected by these methods. Using different previously published, commonly used M-PCR methods, this report describes the diversity of SCCmec elements in MRSA isolates in the Pretoria region of South Africa and the discrepancies observed in the assigned SCCmec types. Different SCCmec types were assigned to the same clinical MRSA isolates. The discrepancies included the assignment of composite SCCmec types [(SCCmec II and SCCmecury) 20.7% (40/193)] and [(SCCmec type II+IVc) 22.3% (43/193)] to some of the clinical MRSA isolates. Summarily, the combination of the result of the M-PCR methods showed that the MRSA genotypes circulating in the healthcare facility studied potentially carried SCCmec types I, II, IV (subtypes IVa, IVb and IVd) and V. No SCCmec types III or VIII was detected among the isolates. At least 25.91% of SCCmec type IV was detected in this study, thus corroborating previous findings of the global encroachment of MRSA strains into the hospital settings. The associated epidemiological significance of these observations is discussed and we also call for an African consensus SCCmec typing method in order to allow effective epidemiological data comparison across the countries.


INTRODUCTION
Staphylococcus aureus is a virulent bacterial pathogen which is responsible for infections seen in healthcare and *Corresponding author.E-mail: marthie.ehlers@up.ac.za.
Author(s) agree that this article remains permanently open access under the terms of the Creative Commons Attribution License 4.0 International License community settings (Kim, 2009).Infections caused by MRSA were previously associated with healthcare settings [Healthcare-associated MRSA (HA-MRSA)] but the emergence of community-associated MRSA (CA-MRSA) worsened the health challenges associated with MRSA (Moussa et al., 2012).Epidemiological history shows that, the CA-MRSA differed from the HA-MRSA in various ways: (i) the lack of traditional risk factors associated with MRSA among patients, (ii) a susceptibility pattern with resistance to few antimicrobial agents and (iii) the inclusion of specific virulence factors such as the Panton Valentine leucocidin (PVL) genes (Weber, 2005;Lo et al., 2011).In addition, a previous study have shown that CA-MRSA and HA-MRSA are demographically, clinically, and microbiologically different (Naimi et al., 2003).However, recent reports now show that the clinical definition of CA-MRSA and HA-MRSA (based on disease on-set, risk-factors and possession of PVL gene) are becoming blurred (David et al., 2010;Prosperi et al., 2013).A study by Peterson et al. (2012) showed that demographics including the disease on-set and the associated risk-factors are not consistent with the genotypic classification of CA-MRSA and HA-MRSA.The S. aureus genome includes a mobile genetic element [staphylococcal cassette chromosome mec elements (SCCmec)] that carries the determinant for beta-lactam resistance encoded by mecA (IWG-SCC, 2009) and mecC (Paterson et al., 2013).Earlier reports indicated that the HA-MRSA strains harbor primarily SCCmec type I, II, III or VI (Naimi et al., 2003), while CA-MRSA carries the SCCmec type IV, V, VII, or VIII and are resistant to only β-lactam antibiotics and sensitive to nonβ-lactam antibiotics (Daum et al., 2002).The possibility of transfer of the antimicrobial resistance determinant (the SCCmec) between CA-MRSA and HA-MRSA isolates in healthcare and community settings necessitates accurate and reliable methods for the detection and identification of these strains (Song et al., 2011).Moreover, the lack of healthcare associated risk factors for the definition of CA-MRSA as prescribed by the Centers for Disease Control and Prevention (CDC) (Morrison et al., 2006) was not sufficient in defining the emergence of CA-MRSA-and HA-MRSA-associated infection in the community and the association of CA-MRSA strains with healthcareassociated infections (O"Brien et al., 1999;Saiman et al., 2003).This led to the use of molecular typing tools (based on SCCmec element) for the classification of MRSA (Daum et al., 2002;Naimi et al., 2003).Several methods for SCCmec typing have been developed and have been previously validated and characterized using MRSA strains with known SCCmec elements.These methods were designed in response to new epidemiological and genomic information.For an extensive review of structure of the SCCmec element in S. aureus, refer to the work of Shore et al. (2013).An in-depth description of the molecular basis for the SCCmec typing and other typing methods have been previously reviewed (Stefani et al., 2012).Consequently, a brief description of the regions targeted by the primer sets/SCCmec typing methods investigated in this report and their limitations in detecting SCCmec types are thus presented: the primer sets of Oliveira and de Lencastre (2002) targets the upstream and downstream of mecA complex incorporating the cassette chromosome recombinase (ccr) allele AB.The Oliviera and de Lencastre ( 2002) method described SCCmec type V as type III and did not consider the differentiation between type IV subtypes.An updated version of the Oliviera and de Lencastre (2002) method focused on the detection of SCCmec type IV (Milheiriço et al., 2007) by amplifying regions within the ccrAB allotypes, five polymorphic J1 regions and a new J1 region that was detected in EMRSA-15 clone.The multiplex PCR (M-PCR) primer sets designed by Zhang et al. (2005) focused on the identification of types 1-4 using the mec and ccr elements and the subtypes designation are based on the junkyard region.Five isolates were not typable by this method however, Oliveira and de Lencastre (2002) method designated those isolates as SCCmec type III while four isolates identified as types I or II were designated as type II by the Oliveira and de Lencastre (2002) method.
It is noteworthy that these two methods showed a 100% agreement in typing control strains.An updated version (Zhang et al., 2012) of the previous Zhang and colleagues " (2005) method was later reported.This improvement addressed the following: (a) detection of SCCmec type II strains that lack the kdpE gene; (b) SCCmec type III lacking the J1 region; (c) detection of subtype IVc by targeting the J1 region and subtype IVe by targeting the J3 region; (d) differentiation of the SCCmec VIII and II.Despite this updated method, 4.5% (24/533) of the isolates were not typable.Boye et al. (2007) developed a method to differentiate between HAMRSA from CA-MRSA carrying SCCmec types IV and V thereby, preventing the mistyping of SCCmec type V as type III.Six (1.92%) of the isolates tested by this method were not typable.However, four of the six isolates were typable using the Oliveira and de Lancaster (2002) method.The McClure et al. (2010) method focused on the detection of SCCmec type VIII by amplifying regions within the class A mec gene and type IV ccr gene complexes using five PCR targets.
In an effort to incorporate more variable regions within the S. aureus genomic make up, Kondo et al. (2007) described a method that included the ccr genes, mec class A-C, open reading frame of J1 region, transposons Tn554 and ᵠTn554 in the J2 regions and plasmids PT181 and Pub110 in the J3 regions.Despite the extensive coverage of the variable regions 93/99 MRSA control strains could be assigned by this method while the ccr genes of six mecA positive strains could not be defined by this method.The authors reported that the M-PCR reported did not conflict with previous methods by Oliveira and de Lancaster (2002) and Zhang et al. (2005).
While the complete review of the various SCCmec typing methods is not the primary scope of this report, a snap shot of the adoption frequency of the methods (discussed in this report) by different laboratories indicates that there is no uniform method or criteria for the use of a particular method (Table 1).
To circumvent the inherent limitations of individual SCCmec typing methods, five published M-PCR based SCCmec typing were combined in order to determine the diversity of SCCmec elements in the Pretoria region of South Africa and to observe the differences in the SCCmec types assigned by methods that detect the same range of known SCCmec types.

MRSA sample source and total bacterial DNA purification
One hundred and ninety three (193) MRSA isolates were obtained from the Diagnostic Laboratory, Department of Medical Microbiology, University of Pretoria Tshwane Academic Division, National Health Laboratory Service.The MRSA isolates were subcultured on Blood agar plates (Oxoid, England) at 37°C for 18 to 24 h to obtain single colonies for Gram-staining in order to confirm the purity.Genomic DNA was purified from the 193 MRSA isolates using the ZR Fungal/Bacterial DNA MiniPrep (Zymo Research, Thermo Scientific, USA), according to the manufacturer"s instructions.Ethical approval for this study was obtained from the Research Ethics Committee of the University of Pretoria (protocol number S189/2010 and S175/2011).

Multiplex-PCR assays for the designation SCCmec types
Five commonly described SCCmec typing methods in the literature were investigated.Multiplex PCR (M-PCR) reactions using specific primers were employed as previously described for each of the methods tested (Table 1) and the genomic DNA from a CA-MRSA strain (ATCC CA05) served as a positive control in the M-PCR assays.The samples were reconfirmed using the S. aureus specific primers (McClure et al., 2006).The M-PCR amplicons were electrophoretically separated at 100 V/cm in a 1% MetaPhor TM agarose gel (Lonza, Rockland, USA) containing 5 µl of ethidium bromide (10 mg/ml) (Promega, Madison, USA) and visualized using an Ultra Violet light box (DigiDoc, UVP product, Upland, California).The assignments of SCCmec types were performed as previously described for individual methods (Table 2).

RESULTS AND DISCUSSION
All the 193 previously determined MRSA samples were reconfirmed using the S. aureus specific primers (McClure et al., 2006).The 16S rRNA and the mecA gene were detected in all the samples tested.However, variations were observed in the proportion of samples designated as a specific SCCmec type or untypable by each SCCmec typing method assessed (Table 3).The electrophoretic pattern of the M-PCR amplicons used for the assignment of the SCCmec types is shown in supplementary material (Figure S1-S7).Table 3 shows that methods1 and 3 were able to designate equal number and same set of MRSA isolates as SCCmec I (3.1%) and SCCmec II (9.33%).The number of isolates assigned as SCCmercury by methods 1, 2 and 3 were different, with method 3 designating 61.14% (118/193) (51/193) assigned by method 2 and 3 respectively.As SCCmercury was detected by methods 1, 2 and 3 and also in composite SCCmec type detected by method 1, it is possible that the SCCmercury is carried in separate plasmid within the bacterial cell.No SCCmec type III or type VIII was detected by the methods 4 and 5, respectively.The proportion of untypable MRSA isolates was 4.7% [(9/193); (Method 2)] and 37.82% {(73/193); (Method 4)}.
Table 4 shows the number of clinical MRSA isolates that were designated the same SCCmec type by different M-PCR methods.Methods 2 and 4 assigned SCCmec type II to 30.57% (59/193) of the same set of MRSA isolates.This was the highest number of isolates designated the same SCCmec type by the different methods investigated.Moreover, about 26% (50/193) of the same isolates were assigned SCCmec type IV by methods 2 and 3, while 15.54% (30/193) of the MRSA isolates were designated as SCCmecury by methods 1, 2 and 3.
These observations indicated that the assessed M-PCR methods were able to assign a specific SCCmec type to the same MRSA isolates, most of the remaining isolates were designated different SCCmec types by the methods investigated.In a separate experiment, an attempt to categorize the SCCmec types defined by each SCCmec typing method in this study revealed that there was no specific distribution pattern of SCCmec type(s) among the pulsed field gel electrophoresis derived pulsotypes (data not shown) suggesting that there was no specific association between the chromosomal DNA content of the MRSA isolates and the SCCmec type assigned by the methods evaluated.A spectacular instance of misassigned ST398-SCCmec III MRSA isolates that took about two years to be reassigned as SCCmec type V has been previously reported (van Loo et al., 2007;Jansen et al., 2009) Such incidence would include a redesignation of the isolates from SCCmec III to SCCmec V based on the molecular typing criteria (Ito et al., 2001(Ito et al., , 2004)).This report showed that at least 25.91% of the MRSA isolates was of SCCmec type VI and correlates with a number of recent reports which have indicated an increase in the number of infections associated with SCCmec type IV, V, VII or VIII in the hospital setting (Magilner et al., 2008;David et al., 2010) including the presence of the different genotypes in specific environments (Marchese et al., 2009).Although, the overall epidemiological picture presented in these reports may still be biologically relevant based on the general pattern observed across the different countries involved, the estimated statistics may be misrepresented due to the lack of a unified standard method for SCCmec classification.
The need for standardization of SCCmec typing and genotype designation is evident by a number of reports including: (i) the continuous blurring of the clinical and genetic distinctions between CA-MRSA and HA-MRSA (David et al., 2010;Prosperi et al., 2013) (ii) the probability that CA-MRSA isolates might displace HA-MRSA in future and become the most prevalent strains in clinical settings (Popovich et al., 2008) and (iii) the likelihood for the eventual co-existence of the two MRSA genotypes based on epidemiological modeling (Kouyos et al., 2013).Therefore, the lack of a consensus typing method will make it difficult to predict the actual genetic changes and evolution of the SCCmec elements in S. aureus.A standard and consensus typing method will  2009).The multiplex PCR method described by Kondo et al. (2007) attempts to improve the accuracy of detection by an initial PCR identification of the mec and cassette chromosome recombinases (ccr) types followed by identifying the genes in the "joining regions" (J-regions).Accordingly, sequence variations in the joining regions are then used to classify SCCmec I-V.There is an ongoing effort to test the performance of this method on clinical MRSA isolates from a number of African countries, in our laboratory.Despite this continuous improvement, consensus criteria for choosing a typing method for SCCmec typing is required.
Based on the variations observed in the designation of SCCmec types by various methods targeting different sites and genes within the SCCmec elements, it is obvious that the designation of SCCmec types across different laboratories around the world may not be in synchrony.This is epitomized in the fact that laboratories across the globe adopt different SCCmec typing methods (Table 1).
A recent review indicated that SCCmec typing was recommended as one of the methods for the monitoring of the molecular epidemiology of MRSA at national and international levels (IWG-SCC, 2009).The current study presents one of the challenges in the practicality of such endeavor.A more detailed study primarily designed to compare all published SCCmec typing methods on MRSA strains with known SCCmec sequence information would be required to make informed decision on a consensus M-PCR characterization of the SCCmec element.
While the SCCmec elements described to date include types I-XI, this study focused on the SCCmec types I-V and VIII based on our laboratory dataset on the prevalence of the SCCmec types in Pretoria, South Africa.This work attempts to paint a practical picture of the difficulties encountered in low income laboratories that are still using M-PCR for MRSA genotyping and hence focuses on the mostly reported M-PCR methods as presented in Table 2. Therefore, not all reported M-PCR methods could be covered for an in-depth comparative study.
In conclusion, this report shows the differences in the assigned SCCmec types by the different M-PCR methods as observed in our laboratory.The fact that in spite of the extensive coverage of the variable regions as observed for each method, some clinical isolates could not be SCCmec-typed in the original reports by the authors of these methods was also highlighted.
The M-PCR detection of composite SCCmec types in clinical MRSA isolates (SCCmec II + SCCmecury and SCCmec type II+IVc)) was also reported.A plan is underway to investigate the whole genome sequence of these isolates in order to confirm this finding.From the above discussion, a number of questions thus arise: Is the inability to type clinical strains by SCCmec method attributed to different structural types or rearrangement and/or recombination of known SCCmec elements?Is there still a clinical-epidemiological relevance of HA-MRSA and CA-MRSA differentiation using SCCmec element, considering the reported blurring of the distinction (Peterson et al., 2012) between these two categories?If yes, do we have a consensus algorithm for making this distinction?Is SCCmec element still a reliable tool for typing MRSA isolates as previously suggested (IWG-SCC, 2009) taking into consideration the variations in the nucleic acid content of this element and the associated discrepancies in identification?While it is obvious that diagnostic microarray, sequencing of SCCmec elements and whole genome sequencing are among the modern methods of choice that may resolve this problem, majority of the laboratories in low income countries of Africa are still not able to afford the routine use of these methods.As a way forward, the adoption of a consensus method in South Africa and Africa in general is recommended, in order to allow effective epidemiological data comparison.

Limitation of the study
This report was based on empirical observations of realworld scenarios in the laboratory and therefore was not designed to effectively compare and contrast the individual methods mentioned.Such experiments will include the use of well characterized ATCC strains of MRSA and all published SCCmec typing algorithms.However, the results are useful as a basis for an agreement on a consensus SCCmec typing method in Africa.

Table 1 .
List of primers for six SCCmec typing methods including S. aureus confirmation.

Table 2 .
Number of citations of some SCCmec typing methods as observed in Google scholar database.

Table 3 .
Boye et al. (2007))ortion of SCCmec types detected among the 193 clinical MRSA isolate tested by the M-PCR typing methods assessed in this study.Not all the primer sets described in the original methods were tested in this study.Therefore, data shown are those of the primers tested for each method.See Table2for details of the primers tested in each method.Methods 4 and 5 shown in Table2did not detect any SCCmec type III and VIII in this study.ND:Not detected.This indicates the non-detection of expected SCCmec types detectable by the primers used in this study.#Thenumber of isolates that were not typeaple by theZhang et al. (2012)method was derived by deducting the number of isolates that were already assigned other SCCmec types not covered by this method.±Resultobtainedusing a single-target PCR method byBoye et al. (2007)assigned a composite SCCmec type (SCCmec type II+IVc) to 26.42% (51/193). *