Antibiotic resistance to agents other than β-lactams
The majority of the HA-MRSA isolates were resistant to kanamycin, amikacin and tetracycline. Although the ratio was slightly low (25~55%), these strains were also frequently resistant to tobramycin, gentamicin, erythromycin, quinolones and rifampicin. Recently, it has been reported that rifampicin resistance is related to glycopeptides resistance [25, 26]. Since the ratio of rifampicin resistant strains was relatively high, there is a possibility that there might be glycopeptides related to low resistance strains, e.g., hetero-VISA strains. However, glycopeptide resistance is beyond the focus of this study, so we did not examine the details for these findings.
Similar to HA-MRSA isolates, the majority of CA-MRSA isolates were resistant to kanamycin, amikacin and tetracycline, but were susceptible to other antibiotics, except for erythromycin and ciprofloxacin. These data suggest that Tunisian CA-MRSA strains were more resistant to kanamycin, tetracycline and erythromycin than U.S. and Oceanian isolates . In our study, only four CA-MRSA strains were resistant to clindamycin, thus suggesting that clindamycin can be used for the treatment of CA-MRSA infections in Tunisia.
Characteristics of Tunisian MRSA
We determined MLST genotype of all 69 MRSA strains. Our data clearly indicated that all PVL-positive MRSA strains belonged to predicted founder group (FG) 80, which was previously indicated as clonal complex (CC) 80 at the MLST website. In contrast, the PVL-negative MRSA strains belonged to diverse FGs. In this study, we used the FG, which is used at present in the eBurst system on the MLST website. However, by using the old CC system, we can distinguish some lineages more clearly, e.g., ST239 that carries type III SCCmec as CC8 and ST5 that carries type II SCCmec as CC5, both of which belonged to FG5. Therefore, we listed both the present and former grouping systems in Table 1.
The agr types were well correlated with the MLST genotypes; group I, STs 45, 97, 239, 241, 247, and 1819; group II, STs 5 and 22; group III, STs 1, 80, 153, 1440, and new. There was only one exceptional case of a ST80 strain belonging to the agr group II. Further experiments including nucleotide sequence determination will be needed to clarify this discrepancy.
The SCCmec types of the strains were further determined by multiplex PCR studies, leaving 10 strains still nontypeable. The type IVc SCCmec was the most representative one in Tunisia. It was identified both in CA-MRSA (79%) and HA-MRSA (56%). PVL-positive MRSA strains carried SCCmec IVc and NT-B, which was supposed to be a novel SCCmec type.
The characteristics of Tunisian MRSA strains were also reported by Ben Nejma et al . It has also been reported that the CC80 CA-MRSA strains were predominant clones in Tunisia, similar to many Europeans countries like France, Belgium, and Switzerland [27, 29]. The predominance of the type IVc SCCmec stain was also reported.
The majority of our CA-MRSA (79%) and HA-MRSA (51%) isolates were pvl-positive and belonged to FG80. Our study suggested that the PVL-positive MRSA strains disseminated in Tunisia might be unique to Tunisia or the surrounding countries. Although CC80 PVL positive MRSA strains have been identified in European countries , the majority of them carried a type IVa SCCmec element or their SCCmec subtype was not determined. While two CA-MRSA isolates from Belgium  were reported to belonged to ST153-MRSA-IV, the report did not show its subtype.
According to previous studies, PVL-positive MRSA isolates were reported to be associated with an agr group III background [27, 28, 31]. Among our CA-MRSA isolates, the most predominant agr group was group III, followed by group II, then group I.
The PVL-positive MRSA clones disseminated in other countries belonged to ST1, ST8, ST22, ST30 and ST59, and carried distinct SCCmec elements. Recently, ST30 has been associated with CA-MRSA strains in the United States and in Ireland [27, 31] and the ST93 and ST772 strains have been reported in Australia and India, respectively [32, 33]. These data suggest that the possibility of simultaneous co-evolution of CA-MRSA organisms in different locations  is higher than the possibility of dissemination of a single CA-MRSA clone all over the world. PVL positive strains might therefore have emerged elsewhere and spread in the community and at hospitals.
It is interesting that the PVL-negative MRSA clones were the same MRSA strains isolated in other countries. Two other CA-MRSA isolates belonged to ST5-MRSA-IV which is one of predominant clones in the Netherlands . Concerning the HA-MRSA, the agr group I was predominant, as reported previously in Tunisian MRSA . The predominance of a group I background was also reported in United States and in Korea [35, 36]. Similar results were obtained in European countries such as Germany and Belgium . Three isolates belonged to the clone ST241-SCCmecIII. Two belonged to the ST247-SCCmecI (Iberian) clone, which is one of predominant clones in Poland . Two other isolates belonged to ST239-SCCmecIII (Hungarian) clone, which is predominant in Turkey .