BMC Microbiology BioMed Central

Background Tuberculosis (TB) is a major health problem and HIV is the major cause of the increase in TB. Sub-Saharan Africa is endemic for both TB and HIV infection. Determination of the prevalence of M. tuberculosis strains and their drug susceptibility is important for TB control. TB positive culture, BAL fluid or sputum samples from 130 patients were collected and genotyped. The spoligotypes were correlated with anti-tuberculous drug susceptibility in HIV-infected and non-HIV patients from Tanzania. Results One-third of patients were TB/HIV co-infected. Forty-seven spoligotypes were identified. Fourteen isolates (10.8%) had new and unique spoligotypes while 116 isolates (89.2%) belonged to 33 known spoligotypes. The major spoligotypes contained nine clusters: CAS1-Kili 30.0%, LAM11- ZWE 14.6%, ND 9.2%, EAI 6.2%, Beijing 5.4%, T-undefined 4.6%, CAS1-Delhi 3.8%, T1 3.8% and LAM9 3.8%. Twelve (10.8%) of the 111 phenotypically tested strains were resistant to anti-TB drugs. Eight (7.2%) were monoresistant strains: 7 to isoniazid (INH) and one to streptomycin. Four strains (3.5%) were resistant to multiple drugs: one (0.9%) was resistant to INH and streptomycin and the other three (2.7%) were MDR strains: one was resistant to INH, rifampicin and ethambutol and two were resistant to all four anti-TB drugs. Mutation in the katG gene codon 315 and the rpoB hotspot region showed a low and high sensitivity, respectively, as predictor of phenotypic drug resistance. Conclusion CAS1-Kili and LAM11-ZWE were the most common families. Strains of the Beijing family and CAS1-Kili were not or least often associated with resistance, respectively. HIV status was not associated with spoligotypes, resistance or previous TB treatment.


Background
Injection drug user (IDU) populations throughout certain areas of Europe and North America have become major risk groups associated with the epidemic spread of methicillin-resistant Staphylococcus aureus (MRSA) [1][2][3][4][5]. The transmission of MRSA clones through both communityand healthcare-associated routes is responsible for the high incidence of soft tissue infections and increases in severe infections such as endocarditis and bacteremia in IDUs [6][7][8]. Such a clonal dispersal led to MRSA becoming endemic in the Zurich IDU population, where in 2003 24% of all MRSA isolates collected at the University hospital of Zurich belonged to a single, so called "drug clone" [9]. Dissemination of this clone to IDU populations in other, geographically distinct regions of Switzerland has also been recently reported, indicating that it has a capacity for spreading and colonizing new populations [10].
Clinical detection of MRSA can be complicated due to vast strain-to-strain differences in the expression of methicillin resistance. Difficulties arise from strains expressing lowlevel but heterogenous resistance, that upon exposure to β-lactams segregate highly resistant subpopulations resulting in therapy failure [11]. Misdiagnosis of such strains with very low, phenotypically susceptible, minimum inhibitory concentrations (MICs) is a major problem necessitating the use of molecular detection methods.
Epidemiological classification of MRSA strains is important for monitoring their prevalence and spread, and relies on molecular typing of both their core genomic background and the type of staphylococcal cassette chromosome mec (SCCmec) they harbour. SCCmec is the chromosomally integrated resistance element which carries the mecA gene, encoding the alternate penicillin-binding protein PBP2a, which confers methicillin resistance. There are currently six main types of SCCmec, differentiated according to their combinations of mec complex, containing the mecA gene and various portions of its regulatory genes mecR1 and mecI, and ccr complex containing recombinases specific for the chromosomal integration and excision of the SCCmec. Further sub-typing is based on the presence of certain additional genes or resistance determinants within the J (so called junkyard) regions J1, J2 and J3 of the element [12]. A number of non mecAencoding SCC elements, sharing some common features with various SCCmecs, have also been discovered in methicillin sensitive Staphylococcus aureus (MSSA) or coagulase-negative staphylococcal strains [13][14][15][16][17][18].
Identification of the Zurich drug clone was based on a characteristic pulsed field gel SmaI restriction pattern and the presence of a unique, previously uncharacterised SCCmec element which was termed SCCmec N1 . In addition to methicillin resistance, all drug clone isolates were shown to be resistant to trimethoprim and most were resistant to sulfamethoxazole or to ciprofloxacin. MLST typing revealed that the representative isolate of this clone, MRSA CHE482, belonged to sequence type ST45, a genotype that has been associated with epidemic MSSA and low level oxacillin resistant MRSA, which seem to have high colonization and circulation capacities [19].
All the drug clone isolates have low oxacillin resistance levels, with MICs between 0.5 and 4 μg ml -1 , which can make them difficult to detect by phenotypic tests. Except for the detection of the mecA gene, genotypic tests, which rely on identifying known features of SCCmecs [20,21] or SCCmec-chromosomal junctions [22] (X. Schneider, unpublished), have also failed to identify this clone [9].
To facilitate accurate molecular identification of this clone this manuscript presents a detailed description of the novel SCCmec N1 and describes the SCCmec variability observed so far between different isolates.

Mapping of SCCmec N1
The size of SCCmec N1 in CHE482 was estimated to be 45.7-kb, based on a series of overlapping long range PCR products amplified with primers shown in Figure 1 and listed in Table 1. This is larger than the community-associated SCCmec type IV (21-25-kb), type V (27.6-kb) and type VI (approximately 22-kb) elements, falling within the range of the classical hospital associated SCCmec types I-III which range in size from 34-67 kb [23]. Loci of interest within SCCmec N1 were then further mapped and sequenced.
mec complex typing SCCmec typing [20] results suggested that the mec complex did not contain mecI and PCR using primers spanning IS1272 and ΔmecR1 and sequencing over the gene junction then confirmed the presence of a class B mec complex (mecA-ΔmecR1-IS1272).

ccr typing
No ccr complex could be detected using the multiplex PCR for ccr types 1 to 3 as described by Ito et al. [21], however with ccr type 4-specific primers C1 and C2, a weak amplificate was produced [9]. Further sequence analysis revealed that this SCCmec contained two complete ccrAB loci which are both similar to ccrAB4 from strain HDE288, a pediatric clone isolated in Portugal carrying a type VI SCCmec element [24,25]. Therefore specific primers to identify the drug clone ccrAB4-1/-2 genes were designed (primers 27 and 28). One of the loci, ccrAB4-1, was located at the usual ccrAB position downstream of the mec complex at the border of the J1 region. The other recombinase complex, ccrAB4-2 was located within the J3 region ( Figure 1). Sequence alignments of ccrAB4-1 CHE482 , ccrAB4-2 CHE482 , ccrAB4 HDE288 and ccrAB4 ATCC12228 genes showed that all four loci were different, with ccrA4 genes sharing between 85.2% and 89.4% similarity with each other and ccrB4 sequences sharing between 94.3% and 92.9% similarity. Nucleotide sequence similarities of these four ccrA4 genes and ccrA genes from complex types 1-3, and of the four ccrB4 genes with ccrB genes from complexes 1-3, are shown by phylogenetic tree (Figure 2). For these alignments the sequence of ccrB4 HDE288 was adjusted because the database sequence is truncated as the result of an adenine deletion at nt position 1325; leaving it 99-aa shorter than ccrB4-1 CHE482 and 100-aa shorter than ccrB4-2 CHE482 .
By adding back this adenine we could compare the whole length sequences.
The phylogenetic trees show that ccrAB4-1 CHE482 , ccrAB4-2 CHE482 , ccrAB4 HDE288 and ccrAB4 ATCC12228 (S. epidermidis) form a distinct ccrAB4 cluster. The presence of two complete ccrAB4 loci in the CHE482 SCCmec indicated that SCCmec N1 had been composed from at least two different complete or partial SCC elements. Other such mosaic or composite SCC elements have been described previously [13,26,27], however this is the first SCCmec found to contain two copies of the same ccr complex.
Due to the presence of both a class B mec complex and ccrAB4, the CHE482 SCCmec would be most closely related to SCCmec type VI. However, due to a number of unique features, including the presence of a second ccr locus and additional antibiotic resistance determinants, it appears to be a distinct subtype of this group that we are provisionally calling SCCmec N1 .
The fusB1 gene, found within the J1 region of SCCmec N1 , was identical to the hypothetical fusidic acid resistance gene SAS0043 from the methicillin-susceptible strain MSSA476 [14], located on the 22.8-kb SCC-like element The fusB1 gene in CHE482 conferred only low level fusidic acid resistance of 6 μg ml -1 .

SCCmec boundaries
The boundaries of the SCCmec element were sequenced using the primers 2, 3, 13 and 14 and compared to reference sequences of SCC 476 from MSSA476 and SCCmec type II from N315 ( Figure 3). SCCmec N1 had integrated at the same position within the attB SCC sequence at the 3' end of orfX as all previously described SCCmec and SCC elements. The ends of SCCmec N1 contained the characteristic direct and degenerate-inverted repeats found at the ends of SCCmec types I-IV and SCC 476 . Integration site sequences (ISS) with the consensus sequence 5'-(GANGC-NTATCATAANTN)-3 [23] were present at both boundaries. A third ISS sequence (ISS*) was also identified about 6.4-kb upstream from the right end junction.

Drug clone variability
Analysis of PFGE profiles of all drug clone isolates characterised by Qi et al. in 2003 [9] revealed that there were small variations in size in the 208-kb SmaI band containing mecA. Therefore the SCCmec of CHE482 and a selection of three other drug clone strains (ZH4, ZH43 and ZH81, Table 2) were cured using plasmid pSR3-1. The SmaI band carrying SCCmec was slightly larger in CHE482 and ZH81 than in ZH4 and ZH43 before curing, whereas after curing the resulting patterns were identical in all four strains ( Figure 4A). This indicated that there was variability, presumably within the SCCmec.

SCCmec N1 variation
Using the long range overlapping PCR products, variations in the SCCmec elements of strains CHE482, ZH4, ZH43 and ZH81, were compared. Fragment sizes between primers 6 and 4 varied by 100 to 200 bp, and between primers 7 and 9 from 1000 to 1500 bp. The variation in the hypervariable region between the mec complex and

B
Sequencing in from the right junctions showed that the ends of CHE482/ZH81 were identical to each other with no significant similarity to any database sequences (data not shown) but they were different to those of ZH4/ZH43 (Figure 3). The latter sequences of ZH4/ZH43 were identical to the end of SCC 476 . In contrast, the left end chromosome-SCCmec junction sequences were identical in all drug clones analysed.

Antibiotic resistance variability
SCCmec variability also appeared to correlate with other strain differences. Strains ZH4 and ZH43 which had identical SCCmecs, were also both ciprofloxacin resistant; meanwhile CHE482 and ZH81, which share identical SCCmecs, were ciprofloxacin susceptible ( Table 2). There was also variation in fusidic acid resistance levels. Most strains had relatively low fusidic acid resistance. ZH43, however, was highly resistant and resistance was not lost upon curing (Table 3), therefore resistance in this strain was probably additionally caused by a mutation in the chromosomal elongation factor G, EF-G (fusA) [33]. Since both ciprofloxacin and sulfamethoxazole resistance are chromosomal, the SCCmec variants found in the Zurich drug clones are very likely associated with different, closely related genetic backgrounds.
ccr activity CHE482 was cured using either pME21 (ccrAB4-1) or pME22 (ccrAB4-2). Resulting isolates were screened for methicillin, trimethoprim and fusidic acid susceptibility and for amplification of a PCR product spanning the former SCCmec-chromosomal junctions (primers 1 and 14, Figure 1). Both ccrAB4-1 and ccrAB4-2 were functional and able to excise SCCmec even though their ccrA and ccrB Phylogenetic relatedness of selected ccrA and ccrB nucleotide sequences amino acid sequences differed by 11.3% and 4.6%, respectively. This is consistent with the finding that several different ccrAB loci from type IV SCCmecs were all shown to be active, despite varying up to 3.7% in their amino acid sequences [34].

Excision variants
CHE482 was cured using pSR3-1 [35], the resulting strain ME21 was sensitive to oxacillin, fusidic acid and trimethoprim ( Figure 1, Table 3). During curing experiments with pME21 and pME22 we discovered that there were also many strains that had not been completely cured. One set of cured CHE482 variants maintained a fragment of 6.4kb, and sequencing confirmed that this fragment was the portion between the ISS at the right junction and ISS*. This indicated that excision of the main SCCmec fragment containing all three resistance determinants had occurred through recombination between the ISS at the left junction and ISS*.
ccrAB4-1 and its predicted promoter were also cloned into the E. coli-S. aureus shuttle vector pAW17 to produce the recombinant plasmid pME15. Attempts to cure CHE482 of its SCCmec element using pME15 resulted in another partially cured set of variants which had maintained fusidic acid resistance but lost oxacillin and trimethoprim resistance (CHE482Δ, Figure 1B). Analysis of these strains by PFGE, showed that the SCCmec-containing band had become smaller but not to the same extent as the completely cured strain ME21 ( Figure 4B). These results indicated that only a portion of the SCCmec, containing mecA and dfrA, had been lost. PCR mapping identified the loca-tion of the missing portion and sequencing over the excision sites revealed that excision was likely to be mediated by homologous recombination across regions of high nucleotide sequence similarity surrounding the two ccr loci, as no additional ISS sequences were found. It appeared that recombination between the two ccr regions resulted in the deletion of a 22-kb fragment containing ccrAB4-1, the class B mec complex and ΔTn4003. This recombination left an SCC-like element of 23.7-kb, which contained one ccrAB complex (ccrAB4-2) and the fusidic acid resistance determinant ( Figure 1B). This truncated SCC is similar in size to the MSSA476 SCC 476 which also contains fusB1 and a ccrAB locus, although in SCC 476 the ccr genes are most similar to ccrAB type 1. Therefore we have identified three possible excision variants, two resulting from the presence of three ISS, as has been seen in SCCmec type IV strains [26], and the third variant caused by recombination between regions of high sequence similarity.

Conclusion
The general structure of SCCmec N1 (ccrAB4-2, dfrA, class B mec complex, ccrAB4-1, fusB1) was distinctly different from already published SCCmec types. Several regions of variability were found between different clinical drug clone isolates, especially in the right-end region where the presence or absence of a DNA fragment framed by ISS sequences was detected. Nevertheless this clone can now be identified by its resistance profile and its combination of class B mec complex and ccrAB4 complex sequences, thus allowing easier epidemiological identification.

Bacterial strains and growth conditions
Bacterial strains and plasmids are listed in Table 2. The four clinical MRSA isolates CHE482, ZH4, ZH43 and ZH81 were clones associated with intravenous drug users in the Zurich area. Apart from the type strain CHE482, strains were selected from the epidemiological study in 2003 based on their PFGE patterns and resistance profiles ( Table 2) [9]. Growth was at 37°C in Luria Bertani broth (Difco Laboratories, Detroit, MI, USA). Strains harbouring the temperature-sensitive plasmids pME21 or pME22 were propagated at 30°C in the presence of 10 μg ml -1 tetracycline and those with plasmid pME15 were grown at 37°C in the presence of 50 μg ml -1 kanamycin.

Cloning of the ccrAB genes of CHE482
Each of the two ccrAB complexes identified in strain CHE482, including their own promoter, were cloned into the BamHI site of the temperature-sensitive plasmid pYT3, using primers 22 and 24 for ccrAB4-1 CHE482 and primers 25 and 26 for ccrAB4-2 CHE482 . The resulting plasmids pME21 and pME22, respectively, were electroporated into RN4220 and then transduced by Φ80α into the MRSA clinical isolates to be cured of SCCmec. The ccrAB4-1 CHE482 was also cloned into the E. coli-S. aureus shuttle vector pAW17, using the primers 22 and 23 ( Table 1). The resulting plasmid pME15 was then electroporated into RN4220 and transduced into the MRSA to be cured of SCCmec.

Curing of SCCmec
Curing of SCCmec was done by the method of Katayama [35] using either the temperature-sensitive plasmid pSR3-1 containing ccrAB type 2 recombinase genes, or plasmids pME21 (ccrAB4-1) or pME22 (ccrAB4-2). Curing with plasmid pME15 (ccrAB4-1) was done by transducing the plasmid into the clinical MRSA isolates, selecting for kanamycin resistant transductants at 37°C, which were then pooled and plated on LB agar containing kanamycin, grown overnight, and screened by replica plating for loss of oxacillin resistance on 1 μg ml -1 oxacillin.

Sequence analysis
Sequencing was performed with an ABI PRISM BigDye Terminator Cycle sequencing reaction kit (US Biochemicals, Cleveland, Ohio) and an ABI Prism 310 genetic analyzer (Applied Biosystems, Foster City, California).

A B
Sequence assembly was accomplished using the DNAStar sequence analysis package (Lasergene 6.0).
Sequencing of the SCCmec-chromosome junctions of four drug clones was done by direct chromosomal sequencing [41] of the original and cured clones using primer 2. This nucleotide sequence provided the template for the design of primers 14, 13 and 3 (Table 1) over the chromosome-SCCmec junctions.

Mapping of SCCmec
To estimate the size of the CHE482 SCCmec, long range PCR amplification of six main fragments was performed using the polymerase TaKaRa Ex Taq (TAKARA BIO INC., Shiga, Japan). PCR was done as described by the manufacturer's recommendation. Primer pairs utilised were: 1 and 5; 4 and 6; 4 and 8; 7 and 9; 10 and 11 and 12 and 14 (Table 1). Amplified PCR fragments were run against molecular weight markers (Marker II, Fermentas International, Ontario, CA; 1 kb+, Invitrogen, Carlsbad, CA) on a 0.5% agarose gel.