In 2006 in Marseille, France, during an epidemiological survey of S. aureus in CF patients, we have detected an atypical GS-MRSA strain with a specific antibiotic susceptibility profile and a unique growth phenotype. The strain was susceptible to gentamicin and resistant to tobramycin, kanamycin, erythromycin, lincomycin, and ofloxacin. The isolate had a hetero-Glycopeptide-Intermediate phenotype of resistance (GISA) and grew with atypical intense orange pigmentation on Cepacia agar (Figure 1a). Because of the clinical importance of the spread of such strain among CF patients we sequenced the genome of one representative isolate (CF-Marseille) and compared this to the published genome sequences, to detect new genetic features responsible for pathogenicity, epidemicity or antibiotic resistance. For this purpose we used high throughput sequencing system (454 Life Science Corp., Roche) 28 coupled with microarrays and molecular genotyping methods to decipher specificities of this isolate in the CF population. We also conducted a retrospective epidemiological analysis on all S. aureus isolated from 2002 to 2007 in CF patients from our institution to understand dynamics of change and spread of the different strain phenotypes. Finally, specific sequences found in the newly sequenced genome were used to design primers to trace the strain in an epidemic setting 29.

The cell wall of CF-Marseille was significantly thicker (33.5 +/- 5.8 nm, n = 100 measurements) than MSSA strain CIP 76.25 (24.7 +/- 4.0 nm, n = 100 measurements) (p < 10^-3) (Figure 1b). Presence of phages was not visualized by electron microscopy without induction whereas phages were seen when antibiotics were added to the medium (Figure 1c). Antibiotics able to induce phages were fusidic acid, tobramycin, ciprofloxacin, cotrimoxazole, erythromycin, rifampin and imipenem whereas oxacillin, ceftazidime, vancomycin, fosfomycin, thiamphenicol, and colistin were not. Apart from its specific antibiotic susceptibility profile, CF-Marseille has a vancomycin MIC of 2–2.5 μg/ml (Etest strips with cell suspensions calibrated at 2 McFarland units) but satellite colonies grew within the ellipse of growth inhibition (Figure 1d). Profile analysis population with teicoplanin confirmed that some colonies were able to grow at 4 μg/ml (Additional file 1), thus displaying a hetero-GISA phenotype 30.

The 454 sequencing of GS-MRSA strain CF-Marseille genome gave a total of 19,064,566 nucleotides that gave a total of 131 assembled contigs with sizes ranging from 546 to 210,978 bp. The contig sizes sum up to 2,829,971 bp. The total contigs size of 2.83 Mbp provides a lower bound of the chromosome size of CF-Marseille. The 131 contigs are available in the EMBL Nucleotide Sequence Database, accession number DS:71627. We compared the CF-Marseille strain to other S. aureus strains on the basis of their protein contents. A CoDing Sequence (CDS) prediction was performed with the Glimmer software 1, which revealed the presence of 2736 hypothetical proteins in the 131 contigs. As compared to the other S. aureus proteins available in the UniProtKB/Swiss-Prot database 2 (considered strains were N315, MW2, USA300, Mu50, NCTC 8325, COL and MRSA252), 2675 out of 2736 predicted proteins showed a match of at least 90% of aminoacid identity. The strains USA300/Mu50 shows the closest protein content with 2522 matches, corresponding to more that 95% of the detected CF-Marseille proteome. The 61 proteins that did not match mostly consist of phagic proteins and very short CDS that are likely to be false positive (Open reading frame that do not correspond to CDS).

Analysis of the accessory genome of CF-Marseille revealed the presence of a new phage of 44 Kb closely related to the recently sequenced phage phiETA3 found in S. aureus strain JH1-JH9 31. Similarity search at the protein level showed that parts of this element are found in numerous other phages including phiETA3, phiROSA, phiNM4 or phage 96, whereas the function of an important number of putative ORFs remain hypothetical.

Most of the genes differentially expressed between strains found in CF patient and isolates never found in CF patient (see Additional file 2), involved phage elements or resistance determinants, which is consistent with the CGH results. The only gene showing higher expression in the non-CF isolates compared to CF isolates was spxA, a transcription regulator involved in the biosysnthesis of thioredoxin reductase, an enzyme found down-regulated during treatment with hypochlorite 32 or in the presence of berberine chloride 33 and whose deletion yields to the accumulation of biofilm onto surface. Comparison between strains currently isolated in Marseille and CF-Marseille yields a limited number of differentially regulated genes. All these genes belong to phage elements and resistance determinants. Among these were ermC (a gene conferring resistance to erythromycin) and ORFs of the phage (Additional file 2). Most of the targets showing differential expression corresponded to up-regulated genes in one of the phages identified in strain CF-Marseille and potentially in other strains of our collection.

In the genome of CF-Marseille, resistance to beta-lactams and ofloxacin are chromosomally-encoded (Table 1). Indeed, resistance to ofloxacin can be attributed to a chromosomal mutation that results in an aminoacid change Ser84Leu in the gyrase gene gyrA (contig 00393). The SCCmec element in CF-Marseille is novel. Although sequence analysis suggested that it is a type IV cassette containing recombinases ccrA2 and ccrB2, genotyping experiments showed relatedness with N315, a HA-MRSA containing a type II cassette (accession number AM943017). A refined analysis of recombinase sequences revealed that ccrA2 of CF-Marseille was identical to ccrA2 of strain N315 whereas ccrB2 of CF-Marseille was similar to that of strain MW2. This cassette was also peculiar with a size of 29493 bp with the integration of the pUB110 plasmid carrying kanamycin and bleomycin resistance (Figure 2). Thus the SCCmec cassette of CF-Marseille is a mosaic of elements from a type II (nosocomial origin) and a type IV (community origin) cassette. Resistance to macrolides (erythromycin) and tetracycline is encoded on plasmid pWBG738 (contig 00364, 98% homology with Genbank sequence 007209) identified in CA-MRSA strain ST1-MRSA-IV similar to strain MW2. Other antibiotic encoding genes were found in the genome including a metallo-betalactamase (contig 00393 and 00379) as well as several ABC transporters.

From May 2001 to December 2006, 108 adults (age ≤ 18 years) and 98 children or adolescents (age < 18 years) were followed in the two CF centres. Overall, 127 patients (61.7%) were found to be colonized at least once by S. aureus (82 children, 83.7% and 45 adults, 41.7%) and represented 699 isolates encompassing 270 non redundant strains (182 in children, 67.4% and 88 in adults, 32.6%). Among these 270 strains, 80 were resistant to methicillin (29.6%) with 56 (70%) in children and 24 (30%) in adults. Among these 80 MRSA isolates, 18/56 (32.1%) and 7/24 (29.2%) showed the CF-Marseille phenotype in children and adults, respectively. The CF-Marseille phenotype was firstly detected in 2002 and since then newly detected isolates have increased to reach a total of 25 patients (Figure 3). Presence of CF-Marseille was significantly associated with P. aeruginosa colonization (22/25 patients) when compared to patients colonized with MSSA (56/190) (p < 0.05). In addition, when compared to acquisition of MSSA, the risk of acquisition of CF-Marseille was twice higher in patients with P. aeruginosa infection and colistin treatment (Risk ratio = 2.00 [1.02–3.99]; p = 0.047). Phylogenetic relationship between the 29 additional strains isolated in the same hospital from CF patients and from 10 non-CF patients using multi-locus variable number of tandem repeat assay (MLVA) is given in Figure 4. Patient information and typing of the strains are given in Additional file 3. By combining these results with those of oxacillin/methicillin sensitivity, it can be seen that one large group is formed exclusively of MRSA isolated both from cystic fibrosis and non-CF patients (group A). This cluster is composed of SCCmec IV isolates with a type 1 agr locus. The second group is composed of both MRSA and MSSA strains, isolated from cystic fibrosis and non-CF patients (group B). CF-Marseille belongs to Group B (arrow on Figure 4). In this subgroup, strain CF-Marseille co-clustered with several isolates harbouring SCCmec IV element and type 2 agr loci but also with strains harbouring SCCmec II elements. Based on the structure of the MLVA tree, this cluster segregated distantly from control community acquired isolates such as MW2 or USA300. Among the sequenced isolates used as controls for the MLVA assay, strain CF-Marseille appears highly related to N315, a Japanese isolate harboring a type II SCCmec element showed to evolve to the GISA phenotype. A third smaller group (Group C) is exclusively formed of MSSA strains isolated from cystic fibrosis patients and it is separate from Group A and Group B in the dendrogram analysis. This small cluster of isolates harbours agr 3 locus and the TSST-1 gene. Finally, 14 out of 21 MRSA strains from CF patients and 3 out of 10 MRSA strains from non-CF patients contained the phage found in CF-Marseille as determined using specific primers and probe. Complete genome hybridization shows that CF-Marseille appears clearly related to N315/Mu50 in terms of genomic content, as shown by microarray CGH experiments, covering 100% and 95% of N315 and Mu50 ORFs, respectively as well as 96% of ORFs detected in 6 other sequenced strains (Additional file 4). In terms of genetic content, the genome of CF-Marseille displays some alterations compared to other strains currently isolated in Marseille. Plasmids and other mobile elements encoding for specific resistance determinants were clearly visible during this comparison. Parts of sequenced phages were also present in CF-Marseille and absent in the other strains (in particular segments of a phage characterized in USA300). CF-Marseille was also the only strain to harbor enterotoxin m and the sdrD genes. In addition, some important genes such as part of the agr locus (particularly the conserved gene agrB) presented some divergence between the 2 populations. List of genes absent in CF-Marseille but present in the other isolates contained only mobile genetic elements such as resistance determinants or integrases originating probably from different phages harbored in the 2 strain populations.

All CF patients followed in the two CF reference centres at Marseille, France, were included in this study from May 2001 to December 2006 for the epidemiological analysis. Statistical analyses were performed using Epi Info 6.0 Software. S. aureus strains isolated from sputum samples were collected from January 2006 to December 2006 using routine laboratory culture methods and standard identification methods 5051. Criteria for selecting epidemic MRSA strains were susceptibility to gentamicin combined to resistance to oxacillin, tobramycin, and kanamycin. Twenty one of these isolates as well as 9 MSSA from CF patients and ten MRSA isolated in the same hospital during the same period in non-CF patients were further analyzed using genotypic markers and microarray experiments by using previously described procedures 525354. This study has been approved by our local ethic committee. No written consent was needed for this work in accordance with the "LOI n° 2004-800 relative à la bioéthique" published in the "Journal Officiel de la République Française" the 6 August 2004 since no additional sample was taken for the study.

MRSA strain CF-Marseille, the prototype of GS-MRSA recovered in CF patient, was isolated in January 2006 from the sputum of a 14-year old CF girl. MIC values of antimicrobials were determined according to the Committee for Antimicrobial Testing of the French Society for Microbiology using a Vitek2* system (bioMérieux, Marcy l'Etoile, France) with Gram positive susceptibility test cards. MIC against vancomycin was tested using Etest strip (AB Biodisk, Solna, Sweden) performed at 0.5 and 2.0 McFarland inocula on BHIA as previously described 55. Plates were incubated at 37°C and read after 48 h. CF-Marseille was also tested for glycopeptide-intermediate susceptibility by population analysis 3755. Finally, one hundred microliters of a bacterial suspension adjusted to McFarland standard 2.0 was spread on brain heart infusion agar (Becton Dickinson, Le Pont de Claix, France) plates with 6 mg/l of vancomycin (Merck, Lyon, France) as described previously 3755. Plates were incubated and growth observed after 48 h. MRSA strain CF-Marseille and vancomycin susceptible S. aureus (VSSA, strain CIP 7625) were examined with a transmission electron microscope Philips -Morgagni 368D (Fei Company, Eindhoven, The Netherlands). The strains were grown on trypticase soya agar at 37°C and were then stained with ruthenium red as described by Luft 56 prior to processing for electron microscopic examination. The cell wall thickness was observed using a Mega View II camera and measured using Analysis 3.2 and Soft Imaging System software. Statistical analysis was done using Epi Info version 6.0 software (CDC, Atlanta, Ga.); p values < 0.05 were considered statistically significant.

CF-Marseille was grown for 2 hours in Trypticase soya broth (TSB, BioMérieux, Marcy l'Etoile, France) at 37°C. Mitomycin C (SIGMA-ALDRICH, Saint-Quentin Fallavier, France) was used in phage mobilization as described previously 31. Briefly, 1 μg/ml of mitomycin C was added to the TSB culture and after 3 hours of incubation with shaking at 30°C, the cell lysate was passed through 0.22 μm filters. Plaque assay was performed to verify phage induction using S. aureus strain CIP 76.25. The effects of sub-inhibitory concentrations of other antibiotics on phage induction of CF-Marseille were also analyzed. The following antibiotics were used as described above: oxacillin (PANPHARMA, France) (8 μg/ml), ceftazidime (GlaxoSmithKline, Marly-Le-Roi, France) (8 μg/ml), imipenem (Merck Sharp & Dohme-Chibret, Paris, France) (10 μg/ml), tobramycin (MERCK, Lyon, France) (8 μg/ml), ciprofloxacin (MERCK, Lyon, France) (8 μg/ml), gentamicin (PANPHARMA, France) (0.5 μg/ml), rifampicin (Sanofi aventis, Paris, France) (0.5 μg/ml), vancomycin (MERCK génériques, Lyon, France) (1 μg/ml), fusidic acid (LEO, St; Quentin Yvelines, France) (0.5 μg/ml), fosfomycin (ERN, S.A., Barelona, Espain) (8 μg/ml), thiamphenicol (Sanofi aventis, Paris, France) (10 μg/ml), sulfamethoxazole-trimethoprim (Roche) (10 μg/ml), erythromycin (AMDIPHARMA, Dublin, Irlande) (8 μg/ml), metronidazole (Fresenius Kabi, Sèvres, France) (10 μg/ml), and colimycin (Sanofi Aventis, Paris, France) (300 IU/ml). Presence of induced phages was examined with a transmission electron microscope Philips -Morgagni 368D as described above.

The microarray was manufactured by in situ synthesis of 10'807 long oligonucleotide probes (Agilent, Palo Alto, CA, USA), selected as previously described 62. It covers >98% of all ORFs annotated in strains N315 and Mu50 12, MW2 11 and COL 15, NCTC8325, USA300 14, MRSA252 an MSSA476 13 including their respective plasmids. Genomic DNA (gDNA) was prepared from isolated colonies grown overnight on Mueller Hinton (MH) agar at 37°C. Briefly: 10⁹ cells were lyzed in 100 μLTris EDTA buffer (10 mM Tris-1 mM EDTA, pH = 8) containing 50 μg/ml lysostaphin (Ambicin from Applied Microbiology, Tarrytown, NY) for 10 min at 37°C. DNA was then isolated and purified using DNeasy^® kit (Qiagen, Hilden, Germany) according to manufacturer's instructions, including RNAse treatment. DNA quantification and protein contaminations were assessed by using a NanoDrop^® ND-1000 Spectrophotometer (NanoDrop Technologies, Inc. Rockland, DE).

Statistical analysis were done using Epi Info Software version 6.0 http://www.cdc.org for Chi-square – Mantel-Haenszel two-tailed test and risk ratio.

The 131 contigs of the genome of CF-Marseille are available in the EMBL Nucleotide Sequence Database under accession number DS:71627. The SCCmec element in CF-Marseille is available in the EMBL Nucleotide Sequence database under accession number AM943017.

Eric Bapteste, Dalhousie University, Canada

Rolain et al. submitted a solid and well-written scientific work. They certainly discovered that an atypical MRSA strain thriving in Marseille harbours an antibiotic inducible bacteriophage. Although this is definitely worth to be published, I am not sure however that this paper belongs to Biology Direct. The analyses presented here seem more interesting from a medical point of view than from an evolutionary perspective, or than from a perspective of comparative genomics. It is thus unclear to me why these authors did not submit their manuscript to the Lancet or another such review. In Biology Direct readers would likely expect deeper (comparative) genomic analyses of this newly sequenced strain of Staphylococcus aureus. For instance, in their conclusion, the authors comment about lateral gene transfer, the genomic plasticity and the conservation of core genes in their strain, but they do not present any phylogenomic analyses to back up their claims. As a result, their discussion seems a bit vague, while a deeper analysis could prove of further interest to biologists as well as to the medical community. In terms of genomics, lots of interesting discoveries remain to be done with their data.

Finally, the MLVA tree – and the methods behind it- should be better explained; a list of the abbreviations used should be introduced in the beginning of the paper to help the reader.

Pierre Pontarotti, Université d'Aix-Marseille, France

Rolain et al. report the complete sequence of a representative isolate of a Methicillin-Resistant Staphylococcus aureus from a cystic fibrosis patient (name strain CF-Marseille) and have conducted a comparative genome analysis with the published sequences. This comparative analysis shows that the CF Marseille strain displays a similar core genome with other sequenced MRSA, with some apomorphism due to mobile elements and a new phage closely related to Phi ETA3. The bioinformatics and microbiologic analysis are robust and the paper is highly informative. I would, however, advise the author to write a short introduction on the origin of multi- antibiotic resistance and artificial selection (see for example Palumbi: Humans as the World's Greatest Evolutionary Force Science 7 September 2001: Vol. 293. no. 5536, pp. 1786 – 1790 science.293.5536.1786). This will be helpful for the BD readers that are not specialists in the domain. An explanation of the role of virus (that do not encode for resistance antibiotic genes) in the emergence of antibiotic resistance is also necessary, as this role is not well explained for a non specialist.

Igor Zhulin, University of Tennessee, USA

The paper by J.-M. Rolain et al. describes a genomic analysis of a multidrug-resistant strain of S. aureus. I am curious why the authors decided to send their study to Biology Direct: this work would be more appropriate for a journal dedicated to infectious diseases (e.g. Journal of Clinical Microbiology). I am not sure that clinicians who need to know about this would necessarily read Biology Direct. Nonetheless, this paper presents a very interesting case: a strain, which is resistant to multiple antibiotics, has emerged from cystic fibrosis patients and the mechanism for this emergence is phage induction by administered antibiotics. The phage activity is likely responsible for promoting the spread

This is a straightforward study from the genome analysis point of you. It does show the power of array-based pyrosequencing in obtaining critical biological data in a very short time. Clearly, in this particular case, there was no need to spend extra time and money to chase the remaining 4% of DNA and close the genome. It became obvious upon analysis of the genomic data that sequences unique to this strain (in comparison with a number of available genomes of various S. aureus isolates) comprise a novel phage and a modified mobile genetic element. The rest of the study links this genomic data to the biology of the pathogen.

Despite having a degree in microbiology, I am not quite qualified to comment on the epidemiological aspects of this work. However, I read the paper with a great interest and did not notice any obvious flaws in experimental design and interpretation.

Overall, this finding appears to be important from the medical perspective and will contribute to better understanding of infectious diseases.