The CA-MRSA strains were resistant to gentamicin (7%), kanamycin (89%), amikacin (86%), tobramycin (18%), tetracyclines (75%), ofloxacine (11%), ciprofloxacin (36%), erythromycin (46%), clindamycin (14%) and rifampicin (4%). All strains were susceptible to pristinamycin, vancomycin, teicoplanin, trimethoprime-sulfamethoxazole and chloramphenicol. The HA-MRSA strains were resistant to gentamicin (38%), kanamycin (90%), amikacin (90%), tobramycin (26%), tetracyclines (88%), ofloxacine (30%), ciprofloxacin (45%), erythromycin (55%), trimethoprim-sulfamethoxazole (15%), chloramphenicol (7.5%), clindamycin (18%), rifampicin (32%) and fosfomycine (10%). All strains were sensitive to pristinamycin, vancomycin and teicoplanin.

The characteristics of 41 HA-MRSA strains are summarized in Table 1. Twenty-one strains were PVL positive, while 20 strains were PVL negative. All PVL-positive strains belonged to the predicted founder group (FG, formerly called the “clonal complex”) 80 in the MLST genotype (ST80, 20 strains and ST1440, 1 strain). All strains belonged to agr group III, and four spa-types (70, 346, 435, and new) were identified among them. All PVL-positive strains carried the type IVc SCCmec element. In contrast, the PVL-negative clones were very diverse. Eight STs, three agr groups, and more than nine spa types were identified (Table 1). These strains carried SCCmec elements of type I, III, IVc, or were nontypeable (NT). Accordingly, at least eight MRSA clones (ST247-SCCmecI, ST1819-SCCmecI, ST239-SCCmecIII, ST241-SCCmecIII, ST5-SCCmecIVc, ST1-SCCmecNT, ST22-SCCmecNT and ST97-SCCmecNT) were identified in the 20 PVL-negative HA-MRSA strains, whereas there were only two MRSA clones, ST80-SCCmecIVc and ST1440-SCCmecIVc in PVL-positive HA-MRSA strains.

^aND: could not be detected.

^bNT: non typeable.

The characteristics of the 28 isolated CA-MRSA strains are summarized in Table 1. Twenty-two strains (79%) were PVL-positive and six strains (21%) were PVL-negative. All PVL-positive strains belonged to FG80 and agr group III, and carried the type IVc or NT SCCmec element similar to the cases of PVL-positive HA-MRSA strains. Three spa-types (70, 346, and new) were identified among them. The PVL-negative strains belonged to four FGs (5, 15, 45, and 80), three agr groups, I- III, and there were more than four spa types (35, 381, 1021, and new). These strains carried SCCmec elements of type IVc or NT. As a result, five MRSA clones (ST1-SCCmecNT, ST5-SCCmecI, ST5-SCCmecIVc, ST45-SCCmecNT and ST80-SCCmecIVc) were identified in six PVL-negative CA-MRSA strains.

As listed in Table 1, the SCCmec type of 59 out of 69 MRSA strains were classified by one of the extant types. All PVL-positive HA-MRSA strains and the majority of CA-MRSA strains carried type IV SCCmec of subtype c. Three PVL-positive CA-MRSA strains carried class B mec, but no ccr genes were identified so far. We expressed this as “NT-B”. SCCmecNT-B was identified in three PVL-negative strains belonging to ST45 and ST97. The SCCmec elements of the other four strains were expressed as follows: NT-1 (type 1 ccr positive, the class of the mec gene complex could not be determined), NT-A (ccr genes were not identified, but it carried the class A mec gene complex), NT-N (neither the ccr genes nor the mec gene complex could be identified), and NT-Bc (two ccr genes, types 1 and 2, were identified, and the class B mec gene complex was identified).

We determined the nucleotide sequence of a PVL phage lysogenized in a PVL-positive CA-MRSA strain, JCSC7401, isolated in 2006. The strain belonged to ST80 and carried nontypeable SCCmec (NT-B). This phi7401PVL was 45,334 bp in length from the rightmost phage attachment site (attP-R) to the leftmost site (attP-L), in which 44 predicted ORFs larger than 99 bp were identified. The core sequences of 29 nucleotides were located at both ends of phi7401PVL. The G+C content of phi7401PVL was 33.2%, and was comparable to other staphylococcal phages. The overall organization of phi7401PVL was the same as that of previously-reported PVL phages, which consisted of five regions relating to 1) lysogeny, 2) DNA replication/transcriptional regulation, 3) structural modules (the packaging/head and tail), 4) the lysis module, and 5) lukS-PV and lukF-PV (Figure 1a). The phage was highly homologous to phiSa2mw, which belongs to group 2 of sfi21-like Siphoviridae (Figure 1a and 1b). The entire genome of the phage showed nucleotide identity of more than 95% to that of phiSa2mw. Forty-two of the 44 ORFs were highly homologous to those of phiSa2mw, with the nucleotide identities ranging from 91-100% (Additional file 1: Table S1). The int gene was truncated, although it was highly homologous to extant PVL phages. Two ORFs, TUP03 encoding Na/K ATPase and TUP16 encoding dUTPase, were less homologous to phiSa2mw.

In order to determine whether the Tunisian PVL positive strains also carried the same PVL phage as phi7401PVL, we conducted two PCR studies to identify the regions in common with two PVL phages (phi7401PVL and phiSA2mw): a long range PCR study identifying gene linkage lukS and the tail gene that can identify PVL phages of the elongated head type and another PCR study identifying the region related to lysogeny (Additional file 1: Table S1 and Figure 1a). In our experiments, all the PVL positive strains were positive in both PCR studies.

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 27 . 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.

The phi7401 carried by a ST80 Tunisian MRSA was highly homologous to phiSa2mw carried by ST1-SCCmecIVa MRSA. Only two ORFs, TUP03 and TUP16, showed a lower identity to those of phiSa2mw. Interestingly, TUP03 was identical to ORFs in phi12, phi13, and the bacteriophage in MRSA strains JH1 and JH9, and TUP16 was highly homologous to dUTPase in phiSLT and phi108PVL, with nucleotide identities of 97%. These data suggest that the components of phages were chimeric. Numerous lysogenized phages were induced from the cells of four strains, including JCSC7401 by mitomycin C induction. However, a hybridization experiment with a PVL probe showed that no plaque of the PVL phase was observed. This might have been due to the carriage of a truncated int. It seems that lysogenization of the phage occurred early to thus cause a mutation in the phage genome or that the ST80 strains might have an ability to cause a mutation in the int to keep the inserted phage genome in the chromosome in a stable form.

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 28 . 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 30 , the majority of them carried a type IVa SCCmec element or their SCCmec subtype was not determined. While two CA-MRSA isolates from Belgium 29 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 27 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 34 . Concerning the HA-MRSA, the agr group I was predominant, as reported previously in Tunisian MRSA 27 . 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 36 . Three isolates belonged to the clone ST241-SCCmecIII. Two belonged to the ST247-SCCmecI (Iberian) clone, which is one of predominant clones in Poland 37 . Two other isolates belonged to ST239-SCCmecIII (Hungarian) clone, which is predominant in Turkey 38 .

One hundred and fifty-four non-replicated HA-MRSA strains were isolated from 1999 through 2008 at Charles Nicolle Hospital of Tunis. Among them, 41 strains isolated from 2004 through 2008 were chosen based on their resistance profiles. HA-MRSA strains were isolated from mucous pus and blood cultures, puncture fluids, urine, and biomaterials of inpatients.

A total of 28 non-replicated CA-MRSA strains were isolated from January 2004 through June 2008 in two Tunisian hospitals (Charles Nicolle Hospital and Habib Bourguiba Hospital). CA-MRSA strains were isolated from the specimens of the patients with MRSA infections who had not been recently (¬within the past year) hospitalized or undergone a medical procedure (such as dialysis, surgery, catheterization). The CA-MRSA strains were generally recovered from mucous pus, puncture fluids, urine and biomaterials from outpatients. Some MRSA strains isolated from patients within 48 h of hospitalization, e.g., after surgery, in the intensive care unit, in the departments of nephrology, otorhinolaryngology and gynecology, were also included.

The isolates were identified by the conventional methods (Gram-positive cocci, catalase positive, mannitol fermenting and DNase-positive) and were confirmed to be S. aureus by their ability to coagulate rabbit plasma (bioMérieux, Marcy l’Etoile, France) and to produce clumping factor (Staphyslide test, bioMérieux). The biotypes were determined using Api20 Staph (bioMérieux, Marcy l’Etoile, France).

Strains were purified twice, first on MRSA BD select agar (from Baird Dickinson) and then on the Difco Tryptic Soy Agar (TSA) supplemented with an antimicrobial agent (ceftizoxime sodium) from Astellas Pharma Inc, Tokyo, Japan. The detection of the mecA gene by PCR was conducted as described previously 39 .

The MIC of oxacilline was determined by the agar dilution method according to the Clinical and Laboratory Standards Institute guidelines (CLSI) 40 . S. aureus ATCC 29213 was used as a reference strain.

The antibiotic susceptibility of the isolates was assessed using the disk diffusion method according to the CLSI guidelines, except for pristinamycine, which was used according to the CA-SFM guidelines. The following antimicrobial disks were tested: penicillin G (10UI), oxacillin (1 μg), ampicillin (10 μg), amoxicillin + clavulanic acid (20/10 μg), cephalotin (30 μg), cefoxitin (30 μg), kanamycin (30 μg), gentamicin (10 μg), tobramycin (10 μg), tetracyclines (30 μg), chloramphenicol (30 μg), ofloxacin (5 μg), ciprofloxacin (5 μg), trimethoprim + sulfamethoxazole (1.25/23.75 μg), erythromycin (15 μg), clindamycin (2 μg), vancomycin (10 μg), teicoplanin (30 μg), rifampicin (5 μg), fosfomycin (5 μg) and pristinamycine (15 μg).

The strains were grown on TSB culture at 37°C overnight with shaking. Genomic DNA used as a target for PCR assays was extracted by using a Qiagene kit (QIAamp DNA Mini Kit (250) QUIAGEN. Sciences - US) according to the manufacturer’s instructions.

The SCCmec elements were typed using two multiplex PCR strategies (M-PCR1 and M-PCR2) which are used for SCCmec typing assignment, M-PCR3 was used for the J1 region difference-based subtyping, as described previously 41 . The reference strains used were as follows: NCTC10442(type I), N315(type II), 85/2082(type III), CA05(type IVa), 8/6-3P(type IVb), and 81/108(type IVc).

The carriage of lukF-PV and lukS-PV genes encoding PVL was examined by PCR as described previously 42 .

The presence of the accessory gene regulator, agr, was determined by multiplex PCR amplifying the hypervariable domain of the agr locus, as described previously 43 . PCR amplification was performed in a 50 μl reaction mixture composed of 2U of Ex Taq (Takara Shuzo Co., Ltd., Kyoto, Japan), 10 pmol of each primer, 0.2 mM deoxynucleoside triphosphate mixture, 10 ng of chromosomal DNA, 1X reaction buffer (Takara Shuzo Co., Ltd.) and H₂O. Thermal cycling was set at 30 cycles (30s for denaturation at 95°C, 1 min for annealing at 55°C, and 2 min elongation at 72°C) and was performed with a Gene Amp PCR system 9600 (Perkin-Elmer, Wellesley, Massachusetts).

The genotypes were determined by Multilocus Sequence Typing (MLST) according to the procedure used by Enright et al 44 . The alleles of each locus were compared, and sequence types (STs) were assigned based on the S. aureus MLST database (http://saureus.mlst.net/).

The typing of the polymorphic region of the protein A gene (spa) was performed as described previously 45 . Purified spa PCR products were sequenced, and short sequence repeats (SSRs) were assigned using the spa database website (http://www.tools.egenomics.com/).

Genomic DNA of strain JCSC7401 was extracted with phenol/chloroform and the nucleotide sequences were determined using a 454 genetic analyzer. PCR studies were conducted to amplify the DNA fragment covering the gap of the contigs obtained by the 454 genetic analyzer. The nucleotide sequence of PCR products amplified by long-range PCRs with primer’s pairs listed in Additional file 2 were determined using an ABI sequencer. The nucleotide sequence of phi7401PVL was deposited to the DDBJ/EMBL/GenBank databases under accession no. AP012341.