One hundred and ninety three MRSA isolates collected from female and male patients (age: 1 day to 78 y with an average of 39 y) during the period of April 2010 to August 2011, were obtained from the Department of Medical Microbiology, Tshwane Academic Division, National Health Laboratory Service. The S. aureus isolates were received as MRSA positive bacteria after routine diagnosis using the Vitek2 system (bioMA(c)rieux, Mary l'Etoile, France). Briefly, the isolates were recovered from the clinical specimen using the chromogenic agar (MRSA select) (Bio-Rad Laboratories, USA), before identification with the Vitek2 system. Bacterial genomic DNA extraction from the 193 MRSA isolates was performed using the ZR Fungal/Bacterial DNA MiniPrep (Zymo Research, Thermo Scientific, USA), according to the manufacturer's instruction that was modified by the addition of Iβ-mercaptoethanol (reducing agent) to the Fungal/Bacterial DNA binding buffer to a final concentration of 0.5% (v/v).

A multiplex PCR (M-PCR) assay targeting the 16S rRNA (S. aureus genus specific detection), mecA and luk-PV genes was performed using specific primer sets Table (1) as previously described [9]. Genomic DNA from a CA-MRSA strain (ATCC CA05) served as a positive control. The M-PCR amplicons were fractionated at 100 V/cm in a 1% (m/v) MetaPhor^TM agarose gel (Lonza, Rockland, USA) containing 5 Aµl of ethidium bromide (10 mg/ml) (Promega, Madison, USA) and visualized using an Ultra Violet light box (DigiDoc, UVP product, Upland, California).

The SCCmec-typing and CA-MRSA and HA-MRSA designation were performed Using Specific Primer pairs Table (1) as previously described [10] including a PCR internal control (targeting mecA) to assess the success of the assay. The PCR amplicons were visualised as described for the 16S rRNA amplicons.

The genetic relatedness of the MRSA isolates was determined using pulsed field gel electrophoresis (PFGE) [11] and the PFGE banding patterns that were analysed [12] The MRSA isolates were spa-typed [13] using specific primers Table (1) that targeted the repetitive sequence region (Fc binding region and the X region) of the S. aureus protein A gene. Nine representative isolates (2, 71, 100, 131, 133, 134, 143, 165 and 183) were randomly selected from each UPGMA cluster for the spa-sequence type determination using specific primers Table (1) that targeted X region as previously described [14]. The amplicons sizes were verified in a 0.5 µg/ml ethidium bromide-stained agarose gel electrophoresis (1% agarose in TBE buffer; 60 V; 1 h) using a 100 bp molecular weight marker (Fermentas Life Sciences, Thermo Scientific, USA) and sequenced commercially (Inqaba biotechnical Industries, South Africa).

Multilocus sequence typing (MLST) analysis of representative MRSA isolates from the nine spa-typed MRSA isolates targeting seven housekeeping genes Table (1) was PCR-amplified using specific primers [15]. The amplicons sizes were verified by agarose gel electrophoresis; purified using the Zymoclean gel DNA recovery kit (Fermentas, Thermo Scientific, USA), and sequenced commercially. The sequences obtained for each of the housekeeping genes were analysed using the CLC main workbench version 6.0 (CLC, Denmark) and uploaded on the MLST database (http://saureus.mlst.net/sql/multiplelocus) where allelic profiles and sequence types were assigned.

The PCR assay for agr-typing consisted of a forward primer (used for all reactions) and four reverse primers (agr group I, II, III and IV) Table (1) as previously described [16]. The PCR assays were performed in two separate combinations: (i) the forward primer and two reverse primers for agr group I and II, and (ii) the forward primer and two reverse primers for agr group III and group IV. The PCR amplicons were visualised as described for the 16S rRNA amplicons.

The prevalence of each variable tested was determined using percentages and the MRSA clonal diversity (pulsotypes and spa clustering) was determined as previously described [1]. A 70% similarity coefficient was used as a cut-off point for cluster definition with a tolerance of 1.1% and band optimisation of 1.41% for spa-clustering analysis. The spa types were determined [16] using the Ridom Staph Type software (Ridom GmbH, WA1/4rzburg, Germany). Z-test independent proportion groups (https://www.mccallum-layton.co.uk/tools/statistic-calculators) was used to determine the level of significant difference of parametric variables.

Table (2) shows the sources and recovery of MRSA isolates investigated in this study. Blood cultures yielded the most isolates [39% (75/193)] followed by pus swabs [23% (44/193)] and central venous pressure (CVP) tips [13% (25/193)]. The contribution of MRSA isolates from other specimen types ranged from 0.5-8%. All the MRSA isolates [100% (193/193)] were positive for the 16S rRNA (756 bp) and mecA (310 bp) genes. One [0.5% (1/193)] CA-MRSA isolate was positive for the PVL (433 bp) gene (Fig. S1A).

Table (S1) shows a comprehensive data obtained in this investigation. Assessment of the data for possible associations between the variables evaluated showed that HA-MRSA was significantly (p<0.05) more prevalent in adult female [71% (60/84)] than in adult male [31% (32/103)] patients Table. (S1). On the contrary, in paediatric patients (ages 1d to 8y), HA-MRSA was more prevalent in male [72% (23/32)] than in female paediatrics [(26% (16/60)]. Similarly, the CA-MRSA isolates were more prevalent in the adult female [46% (39/84)] than in the adult male patients [28% (29/103)]; however, not statistically significant (p>0.05). Moreover, there was no significant difference in the number of CA-MRSA isolates [7% (2/29)] in the male paediatric patients compared to that in the female paediatrics [13% (5/39)].

It was also oberved that 2.1% (4/193) of the isolates were found to be of SCCmec type II + IV occuring only in the CA-MRSA genotype, -of agr III type, -of pulsotype J isolated from blood cultures and in female patients alone (Table S1).

The PFGE fingerprints of two (1%) of the isolates were un-typeable with the SmaI restriction enzyme. The clustering of the remaining 191 typeable MRSA isolates revealed eleven clonal patterns Tables (S1, 2, Fig. S1B) and 58% (110/191) of the clinical MRSA isolates showed similar finger-prints and were designated pulsotype-A (which included subtypes A1 to A6). The remaining isolates were designated pulsotypes B to K which differed from pulsotype-A by more than seven bands (Fig. S1B, Table 2).

Twenty-two per cent (43/193) of the MRSA isolates were un-typeable by the SCCmec-typing method, but were positive for the mecA [(internal control) gene (147 bp)]. Five per cent (9/193) of the MRSA isolates showed double bands for SCCmec type II (398 bp) and SCCmec subtype IVc (200 bp). The remaining 73% (142/193) typeable MRSA isolates were characterised as HA-MRSA [65% (92/142)] and CA-MRSA [35.2% (50/142)] Tables (2, S1). None of the MRSA isolates were positive for SCCmec subtype IVc (200 bp) or SCCmec type V (325 bp).

A total of 193 MRSA isolates were typeable by the agr-typing assay. The isolates were grouped into agr I to III including a hybrid (agr group I and III) Tables (2, S1). No agr group IV isolates were detected.

Out of the 193 PFGE typeable isolates, 3% (6/193) were un-typeable by spa-typing and were excluded in the final spa-clustering analysis. Twenty distinct spa-clusters designated A to T and seven outliers Fig. (S2) were recovered. Majority of the isolates belonged to cluster D [25.6% (46/187)] and A [20.9% (39/187)]. Four spa types and spa-sequence types were identified and included those with worldwide distribution while isolates 134 and 165 were un-typeable by the spa-typing method Table (2). The allele number for the yqiL gene of isolate 2 was not found on the MLST website as well as the allelic numbers for arcC, aroE and pta gene in isolates 134 and 165 Table (3). The Vitek2 system antibiogram data of the spa- sequenced/MLST-typed representative MRSA isolates (n =10) is presented in Table (4).

The Vitek2 automated antibiogram (bioMA(c)rieux, Mary l'Etoile, France) of the representative MRSA isolates that were selected from distinct PFGE clusters revealed that the CA-MRSA isolates were resistant to frontline antibiotics including glycopeptides (vancomycin (MIC ≥ 32), and teicoplanin (MIC ≥ 32),), ciprofloxacin (MIC ≥ 8), rifampicin (MIC ≥ 32), and fusidic acid (MIC % 32) Table (4).

A number of genetic distribution events were observed among the isolates within the clinical setting of this study Table (S1). Summarily, the distribution of MRSA strains among the different clinical conditions was rarely dependent on the genetic backbone or genotype. The genotypes with the highest frequency of occurrence were Spa Cluster B-SCCmec I-agr I [19.2%; (37/193)]; Spa Cluster A-SCCmercury-agr I [14.5%; (28/193)] and Spa Cluster E-SCCmercury-agr I [7.8%; (15/193)] (Table 5).

Although 193 (100%) isolates were positive for the mecA and Staphylococcus genus specific 16S rRNA genes, there were variations in the number of isolates that could be typed using other molecular algorithms including PFGE [99% (191/193)], SCCmec [74% (142/193)], spa [31% (6/193)], agr [100% (193/193)] and MLST of representative isolates from PFGE clustering [100% (9/9)] algorithms. Despite these variations, majority of the isolates clustered into one major pulsotype (58%), as shown by PFGE typing Table (2).

An in-depth characterization of clinical MRSA isolates revealed significant flux in the epidemiological genetics and dynamics of these pathogens over time when the current data (spanning year 2010-11) was compared to that in the report of Makgotlho and colleague [1], (data spanning year 2007-8). Molecular and phenotypic characterisation of the MRSA isolates showed evidence of increased CA-MRSA; consistently low PVL carriage among MRSA isolates; high HA-MRSA prevalence in females, and CA-MRSA clones resistant to frontline antibiotics. The recovery of MRSA isolates from blood, pus and CVP tips suggest the ability of S. aureus to remain viable in diverse environment within and outside the host.

Significant changes in the genetic background of S. aureus isolates between the 2009 study (1) and the current study included:

Community-associated MRSA is reported to be more virulent, spread rapidly and able to cause different clinical syndromes [2]. Therefore, whether the observed increase in CA-MRSA/HA-MRSA ratio is suggestive of a gradual replacement of the HA-MRSA genotype in S. aureus-associated disease or, a means towards the establishment of a balance in the population of both genotypes in order to ensure persistence in the human host, is yet to be unravelled.

Although only the spa-typed MRSA isolates were randomly selected for the MLST typing, three pandemic clonal complexes could be identified among the MRSA isolates. While this is epidemiologically relevant, the authors acknowledge that the derived MLST classification may not necessarily be representative of all other isolates in the same Spa cluster. The PFGE was anlaysed using the method described by Tenova and colleagues [12], while this method was able to differentiate the clinical S. aureus isolates based on their genetic content, it did not allow the assessment of the genetic relationship between the genotypes. A limitation in this study is the lack of in-depth clinical data of the patients from which the samples were obtained. This would have allowed for correlation between the molecular and clinical data. In addition, the definition of CA-MRSA and HA-MRSA was based on molecular principles alone [28] and did not include the clinical criteria. Despite the small sample size, this report forms the basis for similar studies in other African nations.

In conclusion, this study reveals an increase in the CA-MRSA/HA-MRSA ratio (within a 5 year period) despite the continuous dominance of the HA-MRSA genotype. More importantly, the increased diversity in MRSA genetic background was associated with resistance to frontline antibiotics. A similar trend regarding the increase in CA-MRSA genotype and MRSA diversity has been previously reviewed however, with no such data from the African continent [2]. An in-depth phenotypic, genetic and epidemiological correlation as described in this report is a necessary step towards understanding the diversity and epidemio-ecological dynamics of clinical S. aureus isolates in African hospitals.