The bacterial strains used in this study are summarised in Table 1. Laboratory strain 8325-4 and its ΔfnbA/B derivative DU5883 were used as a control for fnbA/B 9. Clinical isolates were characterised for the presence of the fnbA, fnbB, agr1-4 and mecA genes by PCR as previously described 1011, and MLST was performed as described by Enright et al. to identify their genetic background 12.

The antibiotics used in this study were as follows: oxacillin, gentamicin, clindamycin, rifampicin and vancomycin purchased from Sigma-Aldrich (L'Isle d'Abeau, France); linezolid provided by Pfizer (Amboise, France); and moxifloxacin provided by Bayer (Wuppertal, Germany). Minimal inhibitory concentrations were determined by broth microdilution assay as recommended by the Clinical Laboratory Standards Institute (CLSI) standards 13.

The strains were cultured on trypticase blood agar plates and incubated overnight at 37°C. Isolated colonies were resuspended in 5 ml brain heart infusion (BHI) in glass tubes (AES Chemunex France) and adjusted to 0.5 McFarland turbidity, corresponding to 10⁸ CFU/ml, as confirmed by bacterial count. Bacterial suspensions were cultivated at 37°C with 300 rpm gyratory shaking. After 1 h, antibiotics were added to the culture medium at a concentration of half the MIC, and the incubation was continued for 2 additional hours to reach the mid-exponential phase. McFarland turbidity was measured at the end of the incubation step to determine the impact of antibiotics treatment on bacterial density. Aliquots were then taken, and cellular pellets were prepared as described below for total RNA extraction, the microplate adhesion assay, and the whole cell adhesion and invasion assay.

Aliquots of 1 mL of the S. aureus 8325-4 cultures were centrifuged at 13,000 g, and the pellets were washed with 1 mL of 10 mM Tris buffer and adjusted to an optical density at 600 nm (OD600) of 1, corresponding to approximately 1 × 10⁹ S. aureus cells/mL. One mL of adjusted and washed bacterial suspension was centrifuged at 13,000 g, and the pellets were treated with lysostaphin (Sigma-Aldrich) at a final concentration of 200 mg/L. The total RNA of the pellets was then purified using the RNeasy Plus Mini Kit (Qiagen) according to the manufacturer's instructions. The RNA yield was assessed by UV absorbance, and 1 microgram of total RNA was reverse transcribed using the Reverse Transcription System (Promega) with random primers, as recommended by the provider. The resulting cDNA was used as the template for real-time amplification of gyrB, fnbA and fnbB using specific primers (Table 2). The relative amounts of the fnbA and fnbB amplicons were determined by quantitative PCR relative to a gyrB internal standard, as described elsewhere 14. The calibrators in our study were the transcripts from the S. aureus 8325-4 strain grown without antibiotics, normalised with respect to gyrB transcription level. gyrB expression was not modified by sub-inhibitory antibiotics, thus allowing its use as an internal control. The relative fold changes in the fnbA and fnbB expression levels were calculated using the 2^-ΔΔCt method using the RealQuant software (Roche Diagnostics).

Flat-bottom 96-well plates (Immuno Nunc) were coated with 50 μg/mL plasma human fibronectin (Sigma-Aldrich) and incubated overnight at 4°C. Coated plates were inoculated with 200 μL per well of bovine serum albumin BSA and incubated for 20 min at 37°C, and then each well was washed 3 times.

Aliquots of the bacterial cultures described above were centrifuged at 13,000 g for 10 min, and the cellular pellets were washed and resuspended in PBS (Dulbecco's Phosphate Buffered Saline, Sigma-Aldrich). Bacterial suspensions were adjusted to an OD600 of 1, corresponding to approximately 1 × 10⁹ S. aureus cells/mL. One hundred μL of each bacterial suspension was incubated in 3 different wells of the fibronectin-coated plate for 45 min at 37°C with mild shaking. Each well was washed 3 times with PBS to remove non-adherent bacteria. Adherent bacteria were fixed with glutaraldehyde (2.5% v/v in 0.1 mol/L PBS) for 2 h at 4°C and then stained with crystal violet (0.1% m/v) for 30 min at room temperature. Excess stain was rinsed off with Triton X100 solution (0.2% v/v, H₂O), and the plates were dried at room temperature. Bacterial adhesion to fibronectin was assessed spectrophotometrically (Spectrophotometer MR5000, Dynatec) by determining the optical density at 570 nm (OD570). The results were expressed as the mean ± standard deviation based on triplicates. To assess the potential confounding role of antibiotics-induced reduction of bacterial density in our model, we also searched for a correlation between n-fold changes in bacterial densities and fibronectin binding levels in antibiotics-treated strain 8325-4, as compared to the untreated control.

All cell culture reagents were purchased from GIBCO (Paisley, UK). The human osteoblastic cell line MG-63 (LGC Standards, Teddington, UK) was grown in Dulbecco's modified Eagle medium (DMEM) containing 2 mM L-glutamine and 25 mM HEPES, 10% foetal bovine serum (FBS) and 100 U/mL penicillin and streptomycin (culture medium) at 37°C and 5% CO₂. Cells were subcultured twice a week and used up to passage 10 after thawing.

MG-63 cells were seeded at 50,000 cells/well in 24-well plates and incubated at 37°C with 5% CO₂ for 48 h in culture medium. S. aureus strain 8325-4 was treated with sub-inhibitory concentrations of oxacillin, linezolid or rifampicin as described above and then washed and resuspended in antibiotic-free culture medium. The untreated S. aureus strain DU5883 (isogenic mutant of strain 8325-4 deleted for the genes fnbA/B) was used as a negative control. MG-63 cells were washed twice with DMEM and were infected with bacterial suspensions at a multiplicity of infection of approximately 50:1, as confirmed by bacterial count. Cells were incubated for 2 h at 37°C to allow for adhesion and internalisation of the bacteria and then washed twice with DMEM to remove unbound bacteria. For the adhesion assay, cells were analysed using osmotic shock in pure water and extensively pipetted to achieve full release of cell-associated bacteria. For the invasion assay, infected cells were further incubated for 1 h in culture medium containing 200 mg/L gentamicin to kill extracellular bacteria but not internalised bacteria. Cells were then washed twice in DMEM and analysed as described above to release internalised bacteria. For both the adhesion and invasion assays, viable bacteria in cell lysates were enumerated by plate counting on agar. The number of adherent bacteria was calculated by subtracting the number of internalised bacteria from the number of total cell-associated bacteria. The results were expressed as the mean ± standard deviation of the percentage of recovered internalised or adherent bacteria with respect to inoculated bacteria, derived from four independent experiments performed in duplicate.

The statistical analyses were based on the use of one-way ANOVA followed by the a posteriori Dunnett's test. Correlation analysis was performed using Spearman's rank correlation coefficient. The level of statistical significance was set at 0.05. The tests were carried out with SPSS for Windows version 12.0 software.

We first examined the influence of the experimental conditions on the MIC values of tested strains. The oxacillin, gentamicin, vancomycin, clindamycin, linezolid, moxifloxacin and rifampin MICs were determined using CLSI recommendations and compared to those obtained with BHI inoculated with 5 × 10⁵ CFU/mL (Table 3). MICs in BHI were of the same magnitude as those obtained in Mueller-Hinton, therefore we used BHI medium for the rest of this study.

We determined the influence of antibiotics on the adhesion of S. aureus to fibronectin using a fibronectin-coated microplate adhesion assay. We tested standardised bacterial suspensions of 6 strains previously cultured without antibiotics or with 1/2 MIC oxacillin, gentamicin, vancomycin, clindamycin, linezolid, rifampicin or moxifloxacin. As shown in Figure 1, the adhesion to fibronectin was differentially modulated by the antibiotics. Oxacillin-, moxifloxacin-, clindamycin- and linezolid-treated bacteria displayed increased binding to fibronectin. This effect was observed for all strains tested except fnbA/B-negative DU5883. The increase in amplitude of fibronectin binding was strain-dependent. Oxacillin treatment increased fibronectin binding from 1.8- to 2.7-fold relative to the untreated control; moxifloxacin treatment increased binding from 1.4- to 2.3-fold; clindamycin treatment increased binding from 1.5- to 1.8-fold; and linezolid treatment increased binding from 1.6- to 2.3-fold, depending on the tested strain. By contrast, fibronectin binding was significantly reduced after rifampicin treatment. The decrease was strain-dependent and ranged from 1.5- to 3.5-fold compared to the untreated control. Vancomycin and gentamicin had no effect on bacterial adhesion to fibronectin-coated plates (data not shown). Antibiotics-induced reduction in bacterial density had no significant confounding effect on fibronectin binding in our model, as demonstrated by the absence of correlation between n-fold changes in bacterial density and fibronectin binding in antibiotics-treated strain 8325-4 (Additional File 1). The DU5883 strain, defective for fnbA and fnbB genes 9, did not adhere to fibronectin-coated plates in any condition (with or without antibiotics). Clindamycin could not be tested with the DU5883 strain as it harbours the ermB gene and therefore is resistant to clindamycin (Table 3).

We explored the effect of antibiotics on mRNA expression levels of the fnbA and fnbB genes which encode FnBPA/B. The fnbA and fnbB mRNA levels in exponential phase cultures of S. aureus 8325-4, grown with 1/2 MIC oxacillin, gentamicin, vancomycin, clindamycin, linezolid, rifampicin or moxifloxacin, were compared to those in an untreated culture and expressed as n-fold variation. A change in mRNA level was interpreted as significant if there was greater than 2-fold variation. As shown in Figure 2, oxacillin induced a 5.5-fold increase in the fnbA mRNA level and 8.5-fold increase in the fnbB mRNA level; moxifloxacin induced a 2.7-fold increase in the fnbA mRNA level and 4.5-fold increase in the fnbB mRNA level; and linezolid induced a 3.8-fold increase in the fnbA mRNA level and 6.5-fold increase in the fnbB mRNA level. No significant changes in fibronectin binding gene expression were observed for gentamicin, vancomycin, clindamycin or rifampicin.

We investigated whether antibiotic-mediated modulation of the expression of fnbA and fnbB induced changes in S. aureus adhesion to and invasion of host cells in an ex vivo model. We infected osteoblastic MG-63 cells with the following: (i) S. aureus 8325-4, either untreated or treated with 1/2 MIC linezolid, oxacillin or rifampicin and (ii) invasion-deficient strain DU5883. We then compared the amounts of adherent and internalised bacteria recovered after 2 h. As shown in Figure 3, oxacillin-treated S. aureus exhibited significantly increased adhesion (682 ± 374%) compared to untreated S. aureus (256 ± 128%), whereas the adhesion of bacteria treated with linezolid or rifampicin (279 ± 141% and 306 ± 190%, respectively) did not differ significantly from the untreated control. Strain DU5883 showed a tendency towards impaired adhesion (151 ± 40%) compared to its parental strain 8325-4. With respect to bacterial invasion, bacteria treated with linezolid, oxacillin or rifampicin (6.7 ± 4.9%, 9.2 ± 4.1% and 10.4 ± 7.8%, respectively) did not exhibit significant differences compared to the untreated control (6.0 ± 5.1%), while host cell invasion was abolished in strain DU5883 lacking fnbA and fnbB (0.0 ± 0.0%).