Relevant Role of Fibronectin-Binding Proteins in Staphylococcus aureus Biofilm-Associated Foreign-Body Infections

Abstract

Staphylococcus aureus can establish chronic infections on implanted medical devices due to its capacity to form biofilms. Analysis of the factors that assemble cells into a biofilm has revealed the occurrence of strains that produce either a polysaccharide intercellular adhesin/poly-N-acetylglucosamine (PIA/PNAG) exopolysaccharide- or a protein-dependent biofilm. Examination of the influence of matrix nature on the biofilm capacities of embedded bacteria has remained elusive, because a natural strain that readily converts between a polysaccharide- and a protein-based biofilm has not been studied. Here, we have investigated the clinical methicillin (meticillin)-resistant Staphylococcus aureus strain 132, which is able to alternate between a proteinaceous and an exopolysaccharidic biofilm matrix, depending on environmental conditions. Systematic disruption of each member of the LPXTG surface protein family identified fibronectin-binding proteins (FnBPs) as components of a proteinaceous biofilm formed in Trypticase soy broth-glucose, whereas a PIA/PNAG-dependent biofilm was produced under osmotic stress conditions. The induction of FnBP levels due to a spontaneous agr deficiency present in strain 132 and the activation of a LexA-dependent SOS response or FnBP overexpression from a multicopy plasmid enhanced biofilm development, suggesting a direct relationship between the FnBP levels and the strength of the multicellular phenotype. Scanning electron microscopy revealed that cells growing in the FnBP-mediated biofilm formed highly dense aggregates without any detectable extracellular matrix, whereas cells in a PIA/PNAG-dependent biofilm were embedded in an abundant extracellular material. Finally, studies of the contribution of each type of biofilm matrix to subcutaneous catheter colonization revealed that an FnBP mutant displayed a significantly lower capacity to develop biofilm on implanted catheters than the isogenic PIA/PNAG-deficient mutant.

Staphylococcus aureus is a human pathogen that causes both nosocomial and community-acquired infections that can range from minor skin infections to life-threatening illnesses such as endocarditis, osteomyelitis, or toxic shock syndrome. The emergence of highly virulent strains, resistant to many antibiotics, such as methicillin (meticillin)-resistant S. aureus (MRSA) is of considerable concern worldwide (21, 24). In addition, patients with indwelling medical devices can easily develop staphylococcal infections, due to the ability of S. aureus to colonize the implant surface as a biofilm. Inside the biofilm, bacteria are enclosed in a self-produced, hydrated polymeric matrix that confers increased resistance to desiccation and increased tolerance to antimicrobial agents and the host immune response by mechanisms that are still somewhat unclear.

The biofilm matrix is a complex mixture of macromolecules, including exopolysaccharides, proteins, and DNA. The main exopolysaccharide of the S. aureus biofilm matrix is a polymer of poly-N-acetyl-β-(1-6)-glucosamine, termed polysaccharide intercellular adhesin (PIA) or poly-N-acetylglucosamine (PNAG), whose synthesis depends on the enzymes encoded by the icaADBC operon (10, 19, 32). The presence of PIA/PNAG exopolysaccharide is not essential for biofilm development, since several studies have uncovered the existence of S. aureus isolates able to produce alternative biofilm matrixes (6, 9, 11, 14, 28, 33, 36, 45, 46, 49). When this occurs, it appears that proteins usually take the responsibility for mediating cell-to-cell interactions and multicellular behavior. Interestingly, protein-mediated biofilm formation capacity seems to be particularly frequent among the highly virulent MRSA isolates, emphasizing the importance of this type of biofilm structure (14, 37).

The first example of a surface protein able to induce biofilm development was Bap. Bap is a large cell wall-associated protein able to mediate primary attachment to abiotic surfaces and intercellular adhesion (11). So far, the presence of Bap has only been described in mastitis-derived staphylococci (11, 50). The second example was SasG, a protein involved in adhesion to desquamated nasal epithelial cells, whose overexpression induces the formation of peritrichous fibrils of various densities on the cell wall and biofilm development (9, 27). More recently, two independent research groups have described a novel S. aureus biofilm phenotype mediated by the fibronectin-binding proteins, FnBPA and FnBPB (36, 46). Finally, we have very recently shown that increased accumulation of protein A in agr mutants, and more significantly in double agr-arlRS mutants, contributes to biofilm formation capacity (33, 49).

All genes encoding exopolysaccharidic and proteinaceous factors involved in the biofilm formation process, with the exception of bap, are widely distributed among S. aureus isolates. However, as stated above, not all of these compounds simultaneously take part in a particular biofilm, suggesting that S. aureus regulates the production of different types of matrixes depending on the environmental conditions. Thus, biofilm matrix composition might simply depend on the availability of the nutrients necessary to provide energy and biosynthetic intermediates for the synthesis of the matrix macromolecules. As a consequence of the environment encountered, every synthesized matrix would confer specific capacities to embedded bacteria. In this respect, the properties that each type of biofilm matrix bestow on the bacterial community remain unknown, because an S. aureus strain that readily converts between polysaccharidic and protein-based biofilms has not been investigated.

In this study, we have identified and analyzed a MRSA S. aureus isolate able to modulate the composition of the biofilm matrix, producing either a PIA/PNAG- or FnBP-mediated biofilm, depending on the environmental conditions. Interestingly, we have demonstrated that induction of FnBP levels due to a spontaneous agr mutation, as well as activation of the SOS response in a LexA-dependent manner, promotes FnBP biofilms. In addition, the analysis of the significance of each biofilm matrix in a murine foreign-body infection model revealed that in the case of strain 132, FnBPs played a more relevant role than PIA/PNAG during the in vivo colonization of catheters.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The most relevant bacterial strains and plasmids used in this study are listed in Table 1. The 63 ica-positive clinical isolates tested for their capacity to form biofilms are listed in Table S1 in the supplemental material. They were isolated at the Microbiology Department of the Clínica Universitaria de Navarra (Pamplona, Spain) and at the Centre National de Référence des Staphylocoques (Hôpital Edouard Herriot, Lyon, France).

TABLE 1.

Most relevant strains and plasmids used in this study

Strain or plasmid	Relevant characteristic(s)	Source
Strains
S. aureus
132	MRSA clinical strain; biofilm positive	This study
15981	Clinical strain	53
115	Clinical strain	This study
8061	Clinical strain	This study
G-6478	MRSA clinical strain	This study
10833	Clumping factor-positive variant of Newman D2C	10
ISP479c	ISP479 cured of plasmid	38
RN4220	A mutant of S. aureus strain 8325-4 that accepts foreign DNA	35
132 Δica	132 with deletion of the icaADBC operon	This study
132 ica	132 ica::tet	This study
132 Δagr	132 with deletion of the agrBDCA operon	This study
132 Δspa	132 with deletion of the spa gene	This study
132 clfA	132 clfA::pMAD	This study
132 clfB	132 clfB::pMAD	This study
132 ΔfnbA	132 with deletion of the fnbA gene	This study
132 ΔfnbB	132 with deletion of the fnbB gene	This study
132 ΔfnbAB	132 with deletion of the fnbA and fnbB genes	This study
132 Δica ΔfnbA	132 with deletion of the icaADBC and fnbA genes	This study
132 Δica ΔfnbB	132 with deletion of the icaADBC and fnbB genes	This study
132 Δica ΔfnbAB	132 with deletion of the icaADBC, fnbA, and fnbB genes	This study
132 fnbAB	132 fnbA::tet fnbB::erm	This study
132 Δica fnbAB	132 fnbA::tet fnbB::erm with deletion of the icaADBC genes	This study
132 sdrC	132 sdrC::pMAD	This study
132 ΔsdrD	132 with deletion of the sdrD gene	This study
132 ΔsdrE	132 with deletion of the sdrE gene	This study
132 sasA	132 sasA::pMAD	This study
132 ΔfmtB (ΔsasB)	132 with deletion of the fmtB gene	This study
132 ΔsasC	132 with deletion of the sasC gene	This study
132 sasD	132 sasD::pMAD	This study
132 sasE	132 sasE::pMAD	This study
132 sasF	132 sasF::pMAD	This study
132 ΔsasG	132 with deletion of the sasG gene	This study
132 sasH	132 sasH::pMAD	This study
132 sasI	132 sasI::pMAD	This study
132 sasJ	132 sasJ::pMAD	This study
132 SA0927	132 SA0927::pMAD	This study
132 SA1888	132 SA1888::pMAD	This study
132 srtA	132 srtA::erm	This study
132 lexA	132 LexA(G94E)	This study
132 ΔfnbAB pFNBA4	132 ΔfnbAB complemented with the fnbA gene	This study
132 ΔfnbAB pFNBB4	132 ΔfnbAB complemented with the fnbB gene	This study
132 ΔfnbAB pALC2073	132 ΔfnbAB transformed with the pALC2073 plasmid	This study
132 Δica ΔfnbAB pFNBA4	132 Δica ΔfnbAB complemented with the fnbA gene	This study
132 Δica ΔfnbAB pFNBB4	132 Δica ΔfnbAB complemented with the fnbB gene	This study
132 Δica ΔfnbAB pALC2073	132 Δica ΔfnbAB transformed with the pALC2073 plasmid	This study
132 pFNBA4	132 complemented with the fnbA gene	This study
132 pFNBB4	132 complemented with the fnbB gene	This study
132 pALC2073	132 transformed with the pALC2073 plasmid	This study
132 pALC2073RNAIII	132 transformed with the pALC2073RNAIII plasmid	This study
RN4220 pFNBA4	RN4220 complemented with the fnbA gene	This study
RN4220 pFNBB4	RN4220 complemented with the fnbB gene	This study
RN4220 pALC2073	RN4220 transformed with the pALC2073 plasmid	This study
G-6478 pFNBA4	G-6478 complemented with the fnbA gene	This study
G-6478 pFNBB4	G-6478 complemented with the fnbB gene	This study
G-6478 pALC2073	G-6478 transformed with the pALC2073 plasmid	This study
10833 pFNBA4	10833 complemented with the fnbA gene	This study
10833 pFNBB4	10833 complemented with the fnbB gene	This study
10833 pALC2073	10833 transformed with pALC2073 the plasmid	This study
E. coli
XL1Blue	Used for cloning assays
Plasmids
pGEM-T	Used for cloning assays	Promega
pMAD	E. coli-S. aureus shuttle vector with a thermosensitive origin of replication for gram-positive bacteria	1
pFNBA4	Vector for fnbA complementation experiments	17
pFNBB4	Vector for fnbB complementation experiments	17
pALC2073	E. coli-S. aureus shuttle vector with a tetracycline-inducible promoter.	4
pALC2073RNAIII	pALC2073 with RNAIII cloned into the EcoRI site	This study

Escherichia coli XL1-Blue cells were grown in Luria-Bertani (LB) broth or on LB agar (Pronadisa, Spain). Staphylococcal strains were cultured on Trypticase soy agar (TSA), on Congo red agar, or in Trypticase soy broth (TSB) supplemented with 0.25% glucose (TSB-gluc) or 3% NaCl (TSB-NaCl). Congo red agar was prepared as follows: 30 g/liter of Trypticase soy (Pronadisa), 15 g/liter of agar (Pronadisa), 0.8 g/liter of Congo red stain (Sigma), and 20 g/liter of sucrose. The Congo red stain and the sucrose solution were autoclaved separately (121°C for 20 min and 115°C for 15 min, respectively). For assessment of the hemolysin activity induced by RNAIII, sheep blood agar plates were used (Biomérieux).

Media were supplemented with appropriate antibiotics at the following concentrations: erythromycin (erm) at 1.5 μg ml⁻¹ or 10 μg ml⁻¹; ampicillin at 100 μg ml⁻¹; chloramphenicol at 5 μg ml⁻¹; and tetracycline (tet) at 3 μg ml⁻¹ or 125 ng ml⁻¹.

DNA manipulations.

DNA plasmids were isolated from E. coli using a Bio-Rad plasmid miniprep kit (Bio-Rad Laboratories, Inc.) according to the manufacturer's protocol. Plasmids were transformed into staphylococci by electroporation, using a previously described protocol (11).

Restriction enzymes were purchased from New England Biolabs and used according to the manufacturer's instructions. DNA polymerase was purchased from Biotools. Oligonucleotides were obtained from Thermo Electron Corporation (see Table S2 in the supplemental material).

For detection of the icaADBC operon in S. aureus strains, the icaA and icaC genes were amplified by PCR using primers icaA-1/icaA-2 and icaC-1/icaC-2, respectively, (see Table S2 in the supplemental material).

For Southern hybridization, chromosomal DNA was purified, digested, and analyzed by agarose gel electrophoresis. DNA fragments were transferred by alkaline capillary blotting onto nylon membranes (GE Healthcare) using standard methods (3). Labeling of the probe and DNA hybridization were performed according to the protocol supplied with the DIG DNA labeling and detection kit (Roche).

The substitution of glutamate for glycine at the Ala-Gly cleavage site of LexA was performed as previously described (52).

The genome of S. aureus strain 132 was sequenced using the 454 pyrosequencing strategy on a GS FLX sequencer (LifeSequencing SL, Valencia, Spain) and resulted in the generation of 66 Mb of sequences. The GS FLX reads were assembled into 44 large contigs longer than 1,000 bp (GenBank accession number ACOT01000000).

Allelic exchange of chromosomal genes.

To generate deletion mutants, we amplified by PCR two fragments of 500 bp that flanked the left (primers A and B) and right (primers C and D) sequences of the genes targeted for deletion (see Table S2 in the supplemental material). The PCR products were purified and cloned separately in a pGEM-T Easy vector (Promega). Fragments were then fused by ligation into the shuttle plasmid pMAD, and the resulting plasmids were transformed into S. aureus by electroporation. pMAD contains a temperature-sensitive origin of replication and an erythromycin resistance gene (1). Homologous recombination experiments were performed as previously described (53). Erythromycin-sensitive white colonies, which no longer contained the pMAD plasmid, were tested by PCR using primers E and F (see Table S2 in the supplemental material). Deletion mutants of spa, sdrD, agr, and ica were constructed as previously described (33, 49, 53).

Construction of insertional mutations.

For disruption of the clfA, clfB, sdrC, sasA, sasD, sasE, sasF, sasH, sasI, sasJ, SA0927, and SA1888 genes in strain 132 (Table 1), a PCR fragment within the A regions of the different genes was amplified using the respective primers 1 and 2 (see Table S2 in the supplemental material) and cloned in a pGEM-T Easy vector (Promega). The corresponding fragments were then cloned into the shuttle vector pMAD, and the resulting plasmids were transformed into S. aureus by electroporation. The plasmids were integrated into the chromosome through homologous recombination at a nonpermissive temperature (44°C). The disruption of the respective genes was confirmed by PCR using the pMAD-1 primer and the corresponding primer 3 external to each construction (see Table S2 in the supplemental material).

The srtA::ermC mutation was transduced from S. aureus Newman SKM3 (31) into S. aureus strain 132 using phage 85α. The disruption of the gene was confirmed by PCR using primers srtA-3 and srtA-4 (see Table S2 in the supplemental material).

The fnbA::tet and fnbB::erm mutations were transduced from S. aureus DU 5883 (17) into S. aureus strain 132 and the 132 Δica mutant using phage 85α. The disruption of the respective genes was confirmed by Southern blotting (data not shown).

The ica::tet mutation was transduced into S. aureus strain 132 from S. aureus SA113 Δica, which contains a tetracycline resistance cassette that replaces ica genes (10) using phage 85α. All transductions were performed as described previously (34).

Complementation experiments.

For complementation experiments, the multicopy plasmids pFNBA4 and pFNBB4 (17) that carry, respectively, the wild-type fnbA and fnbB genes of S. aureus 8325-4 were used. The empty plasmid pALC2073 was used as a control (Table 1). For complementation of agr deficiency, RNAIII was amplified by PCR using primers RNA III-1 and -2 (see Table S2 in the supplemental material) and chromosomal DNA from strain ISP479c as template. The PCR product was purified and cloned at the EcoRI site of the pALC2073 plasmid, downstream of the tetracycline-inducible promoter. Correct orientation of the insert was confirmed by restriction mapping. Expression was induced with 125 ng ml⁻¹ of tetracycline.

Biofilm formation and detachment assays.

The biofilm formation assays in microtiter wells were performed as previously described (18). Sterile 96-well polystyrene microtiter plates from the same manufacturer (Iwaki) were used throughout the study. Each assay was performed in triplicate and repeated at least three times. Crystal violet-stained cells were quantified by solubilizing the dye with 200 μl of ethanol-acetone (80:20, vol/vol) and determining the optical density at 595 nm (OD₅₉₅).

For adherence assays in a glass tube, a single colony was transferred to 5 ml of TSB-gluc in a 15-ml tube and incubated at 37°C in an orbital shaker (250 rpm) for 12 h.

The colony morphology of S. aureus strains was analyzed using Congo red agar plates. Briefly, 5 μl of overnight cultures of S. aureus strains were spotted onto Congo red agar plates and incubated at 37°C for 24 h.

To test whether the activation of the SOS response induced biofilm formation, 0.05 μg/ml of mitomycin C (Sigma) was added to the growth media at the time of inoculation.

To study biofilm formation in the presence of RNAIII, RNAIII expression by pALC2073RNAIII was induced with 125 ng ml⁻¹ of tetracycline at the time of inoculation.

To analyze biofilm formation under flow conditions, we used 60-ml microfermenters (Pasteur Institute, Laboratory of Fermentation) with a continuous 40-ml h⁻¹ flow of medium and constant aeration with sterile compressed air (0.3 bar). Submerged Pyrex slides served as growth substratum. Approximately 10⁸ bacteria from an overnight culture of each strain grown in the appropriate medium were used to inoculate the microfermenters that were then kept at 37°C for 24 h. Biofilm development was recorded with a Nikon Coolpix 950 digital camera. To quantify the biofilm formed, bacteria adhered to the Pyrex slides were resuspended in 20 ml of TSB-gluc. The OD₆₅₀ of the suspensions was then determined.

Biofilm detachment assays with sodium metaperiodate (Scharlau) and proteinase K (Sigma) were carried out as previously described (22).

Scanning microscopy.

Scanning electron microscopy was performed on biofilms grown in microfermenters on Thermanox slides (Nalgene Nunc International) fixed on the internal removable glass slides, as described by Prigent-Combaret et al. (39) at the Laboratoire de Biologie Cellulaire et Microscopie Electronique, UFR Médecine (Tours, France).

PIA/PNAG detection.

Cell surface PIA/PNAG exopolysaccharide levels were detected as previously described (10). Overnight cultures of the strains tested were diluted 1:40 in the appropriate medium, and 2 ml of this cell suspension was used to inoculate sterile 24-well polystyrene microtiter plates (Sarstedt). After 24 h of static incubation at 37°C, the same number of cells of each strain was resuspended in 50 μl of 0.5 M EDTA (pH 8.0). Then, cells were incubated for 5 min at 100°C and centrifuged to the pellet. Each supernatant (40 μl) was incubated with 10 μl of proteinase K (20 mg/ml; Sigma) for 30 min at 37°C. After the addition of 10 μl of Tris-buffered saline (20 mM Tris-HCl, 150 mM NaCl [pH 7.4]) containing 0.01% bromophenol blue, 5 μl was spotted on a nitrocellulose filter using a Bio-Dot microfiltration apparatus (Bio-Rad), blocked overnight with 5% skimmed milk in phosphate-buffered saline (PBS) with 0.1% Tween 20, and incubated for 2 h with rabbit antibodies raised to S. aureus deacetylated PNAG conjugated to diphtheria toxoid diluted at 1:10,000 (30). Bound antibodies were detected with peroxidase-conjugated goat anti-rabbit immunoglobulin G antibodies (Jackson ImmunoResearch Laboratories, Inc., Westgrove, PA), diluted 1:10,000, and an Amersham ECL Western blotting system.

Real-time quantitative PCR.

To study the transcription of icaC and fnbB in TSB-gluc and TSB-NaCl media, an overnight culture of strain 132 was diluted (1:100) in the appropriate medium and grown to exponential phase (OD₆₀₀ = 0.6). Total S. aureus RNA was isolated using a Fast RNA-Blue kit (MP Biomedicals, LLC) according to the manufacturer's instructions. In order to minimize variability, nine independent RNAs were purified from each strain and divided into three pools. RNAs were then subjected to DNase (Ambion) treatment for 20 min at 37°C. After inactivation of the enzyme, and to verify the absence of genomic DNA, the RNAs were reverse transcribed in the presence and absence of reverse transcriptase (RT) (SuperScript III RT; Invitrogen). All preparations were purified using Qiagen columns (QIAquick PCR purification kit). To measure expression, icaC transcripts were amplified using primers icaC2-SG-5 and icaC2-SG-3 and SYBR GreenER SuperMix (Invitrogen), and fnbB transcripts were amplified using primers fnbB-FW and fnbB-RV (see Table S2 in the supplemental material).

An ABI PRISM model 9700HT (Applied Biosystems) instrument was used to monitor the real-time quantitative PCR. The gyrB transcripts that were constitutively expressed were amplified as endogenous controls by using primers gyrB-SG-5 and gyrB-SG-3 (see Table S2 in the supplemental material). To monitor the specificity, final PCR products were analyzed by melting curves and electrophoresis. Only samples with no amplification of the minus RT aliquots were considered in the study. The amount of icaC and fnbB transcripts was expressed as the n-fold difference relative to the control gene (2^−ΔCt, where ΔCt represents the difference in threshold cycles between the target and control genes).

Western ligand affinity blotting.

For detection of the level of FnBPs after induction of the SOS response, an overnight culture of strain 132 was diluted 1:50 in TSB-gluc medium supplemented with increasing mitomycin C concentrations (0.005, 0.01, 0.02, 0.04, and 0.06 μg/ml) and grown for 4 h, at 37°C, without shaking. The same number of cells of each strain was collected, and surface-associated protein extracts were prepared as previously described (2). For detection of the level of FnBPs after RNAIII induction, an overnight culture of strain 132 carrying pALC2073RNAIII was diluted 1:50 in TSB-gluc medium supplemented with 125 ng ml⁻¹ of tetracycline, and 2 ml of this cell suspension was used to inoculate sterile 24-well polystyrene microtiter plates (Sarstedt). After 24 h of static incubation at 37°C, the same number of cells of each strain was collected, and surface-associated protein extracts were prepared as previously described (2).

Electrophoretic separation of cell wall-associated protein preparations was carried out by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The acrylamide concentration was 8% in the resolving gel and 4% in the stacking gel. FnBPs were detected by Western ligand affinity blotting by incubation for 1 h with biotinylated human fibronectin (50 μg/ml) in PBS containing 0.1% Tween 20 (PBST). An EZ-Link sulfo-N-hydroxysuccinimide-LC biotinylation kit (Pierce) was used to biotinylate human fibronectin (Sigma). After washing with PBST, membranes were incubated for 1 h with streptavidin-peroxidase conjugate (Roche; 1:3,000 dilution). Finally, membranes were developed with the ECL Western blotting system. The strain 132 ΔfnbAB double mutant was used as an internal negative control.

Coinfection experiments.

Prior to coinfection assays, a competition experiment was performed. The same number of bacteria coming from an overnight culture of each strain was used to inoculate 5 ml of TSB-gluc. After having grown together overnight, 100 μl of an optimal dilution were plated on TSA and TSA tet, in triplicate, in order to estimate plate counts corresponding to each strain.

To examine the in vivo interaction of fnbAB mutants with abiotic surfaces, the murine model of catheter-associated biofilm formation described by Beenken et al. was used with some modifications (6). Briefly, a total of 66 male Swiss mice were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg). Polyurethane catheters (15 mm; B. Braun Medical, Bethlehem, PA) were inserted and coinfected by injection of 200 μl of a 10⁸ CFU/ml mixture containing the same quantity of either strain 132 and 132 fnbAB, strain 132 and 132 Δica fnbAB, or strain 132 and 132 ica. The skin of each mouse was cleansed with alcohol prior to surgery, and as a control of the procedure, six mice were infected with sterile PBS. Mice were euthanized at 3 or 7 days postinfection, and groups of 10 mice were used for each comparison. The catheters were removed aseptically, washed, and placed in 1 ml of sterile PBS. Catheters were then washed internally with the aid of a sterile syringe and sonicated to remove adherent bacteria. The number and identity of recovered bacteria was determined by plate count on TSA supplemented with and without tetracycline at 3 μg ml⁻¹. Competitive indexes were calculated by dividing the number of tet-sensitive CFU by the number of tet-resistant CFU.

Statistical analysis.

Statistical analyses were performed with the GraphPad Prism program. P values assessing the significance of the difference from 1 of the competition indexes were calculated using the Wilcoxon's test. Data corresponding to gene expression were compared using the Mann-Whitney test. Differences were considered statistically significant when P was <0.05. In the biofilm quantification assays, microtiter plate data represent the means of at least nine measures from three independent experiments.

Nucleotide sequence accession number.

The 44 large contigs sequenced in this study have been deposited in the GenBank database under accession number ACOT01000000.

RESULTS

Selection of an S. aureus isolate able to develop PIA/PNAG-dependent and -independent biofilms.

With the aim of selecting a natural S. aureus isolate able to switch between the production of a polysaccharidic and a proteinaceous biofilm matrix depending on the environmental conditions, we first analyzed both the capacity to form a biofilm on polystyrene microtiter plates and the capacity to produce PIA/PNAG in TSB-gluc in a collection of 63 ica-positive S. aureus clinical isolates (see Table S1 in the supplemental material). The strains were classified into four groups: (i) biofilm positive-PIA/PNAG positive (16%); (ii) biofilm negative-PIA/PNAG positive (36%); (iii) biofilm positive-PIA/PNAG negative (19%); and (iv) biofilm negative-PIA/PNAG negative (29%). Figure 1A shows the biofilm formation and PIA/PNAG production of one representative strain from each group. Then, we focused on group III, which contained strains that are capable of forming a biofilm independently of PIA/PNAG synthesis under TSB-gluc growing conditions but are perhaps able to synthesize PIA/PNAG under conditions known to induce its synthesis (3% NaCl and subinhibitory concentrations of antibiotics) (37, 41, 42). Using this approach, we selected one isolate (MRSA strain 132) that formed a biofilm and produced detectable levels of PIA/PNAG when NaCl (3%) was added to the medium (Fig. 1B). To verify that PIA/PNAG was actually required for biofilm formation under TSB-NaCl and not TSB-gluc conditions, we generated a strain 132 ΔicaADBC deletion mutant and tested its biofilm formation capacity. The results confirmed that ica deletion had no impact on the capacity of MRSA strain 132 to produce a biofilm in TSB-gluc, whereas it was required for biofilm development in TSB-NaCl medium (Fig. 2A).

FIG. 1.

Selection of the S. aureus MRSA strain 132. (A) Summary of the classification of 63 ica-positive S. aureus isolates according to their capacities to form biofilm and synthesize PIA/PNAG in TSB-gluc medium. Biofilm formation capacity on polystyrene microtiter plates after 24 h hours of growth at 37°C in TSB-gluc medium, quantification of the biofilm formed, and dot blot analysis of the PIA/PNAG produced by one representative S. aureus isolate of each group are shown (15981, 115, 132, and 8061). Data of OD₅₉₅ represent the means of at least nine counts from three independent experiments. Error bars show standard deviations. (B) Biofilm formation capacity on polystyrene microtiter plates, quantification of the biofilm formed, and dot blot analysis of the PIA/PNAG accumulated by S. aureus MRSA strain 132 after 24 h hours of growth at 37°C in TSB-gluc and TSB-NaCl media. For dot blot analysis, a 1:25 and a 1:50 dilution of the samples were spotted onto nitrocellulose membranes, and PIA/PNAG production was detected with an anti-PIA/PNAG antiserum. OD₅₉₅ data represent the means of at least nine counts from three independent experiments. Error bars show standard deviations.

FIG. 2.

Effect of ΔicaADBC deletion on the biofilm-forming capacity of MRSA strain 132 and phenotypic characterization of the PIA/PNAG-independent biofilm. (A) Biofilm formation of S. aureus strain 132 and its Δica mutant after having grown on microtiter plates for 24 h under static conditions at 37°C in TSB-gluc and TSB-NaCl media. To measure the biofilm formed, the amount of crystal violet-stained cells was quantified by solubilizing the dye in ethanol/acetone and determining the absorbance at 595 nm. Data represent the means of at least nine counts from three independent experiments. Error bars show standard deviations. (B) Detachment experiments. Biofilms formed by S. aureus strain 132 after having grown in TSB-gluc for 24 h were treated with sodium metaperiodate at 10 mM for 2 h at 37°C or with proteinase K at 100 μg/ml for 2 h at 37°C. The quantification of the nondetached biofilm was performed as described for panel A.

We next investigated the composition of the PIA/PNAG-independent biofilm matrix produced in TSB-gluc by means of dispersion experiments with proteinase K and sodium metaperiodate. The results showed that the glucose-induced biofilm was resistant to removal by sodium metaperiodate but sensitive to detachment by proteinase K, suggesting that a proteinaceous compound was responsible for the glucose-induced biofilm development (Fig. 2B). Taken together, these results allowed the selection of an S. aureus strain (132) that was able to form two types of biofilms: one that depended on PIA/PNAG production and was formed under TSB-NaCl (3%) growth conditions, and another one that was PIA/PNAG independent while requiring a proteinaceous compound and was produced under TSB-gluc incubating conditions.

Identification of the protein(s) mediating glucose-induced biofilm development.

Based on previous results showing that different LPXTG proteins, such as Bap, SasG, and protein A, are able to mediate biofilm development in S. aureus, we first explored the possibility that a member of the LPXTG family of proteins was responsible for biofilm development in TSB-gluc. To test this hypothesis, we mutated the sortase A gene, required for the covalent anchoring of the LPXTG proteins to the cell wall, in strain 132. The resulting mutant completely lost the capacity to form a biofilm in TSB-gluc medium but retained the ability to form a biofilm in TSB-NaCl (Fig. 3A). These results suggested that at least one LPXTG protein was involved in the biofilm-forming capacity of strain 132 in TSB-gluc, whereas it had no role in PIA/PNAG-mediated biofilm development.

FIG. 3.

Analyses of the contributions of the LPXTG proteins to the production of the proteinaceous biofilm formed by S. aureus strain 132. (A) Biofilm formation capacities of S. aureus strain 132 and its srtA mutant on microtiter polystyrene plates after 24 h of static incubation in TSB-gluc and TSB-NaCl media. (B) Biofilm formation capacities of S. aureus strain 132 and each LPXTG mutant on microtiter polystyrene plates after 24 h of static incubation in TSB-gluc medium. Representative pictures of at least three independent assays are shown. To measure the biofilm formed, the amount of crystal violet-stained cells was quantified by solubilizing the dye in ethanol/acetone and determining the absorbance at 595 nm. Data represent the means of at least nine counts from three independent experiments. Error bars show standard deviations.

Analyses of the different S. aureus sequenced genomes showed that each isolate contains a unique repertoire of LPXTG proteins. Additionally, some specific proteins, such as Bap and Pls, have only been described in a few strains. In order to determine the set of LPXTG proteins present in MRSA strain 132, its corresponding genome was sequenced using the 454 technology. After that, we carried out the systematic mutation of each of the 20 LPXTG proteins encoding the identified genes. Thereby, bap, cna, pls, and sasK were not considered in the mutagenesis protocol, since they are not present in the genome of MRSA strain 132, and two putative LPXTG proteins (SA0927 and SA1888) were included in the assay (Table 1). Biofilm formation experiments of the resulting mutant strains in TSB-gluc medium revealed that only deletion of fnbB diminished the capacity of strain 132 to form a biofilm (Fig. 3B). Considering the sequence homology between fnbB and fnbA, we hypothesized that both proteins might be involved in the process and that compensatory effects might occur in single mutants. As a consequence, effects of the mutation of fnbA on biofilm formation would be veiled by the activity of FnBPB. To explore this hypothesis, a double ΔfnbAB mutant was constructed. Analysis of its biofilm formation capacity revealed that strain 132 ΔfnbAB completely lost the capacity to form a biofilm in TSB-gluc, indicating that in this media, both FnBPB and FnBPA cooperate to produce a biofilm (Fig. 4A). To exclude any role of PIA/PNAG exopolysaccharide in the production of biofilms by single fnb mutants, double Δica ΔfnbA and Δica ΔfnbB mutants as well as a triple Δica ΔfnbAB deletion mutant were constructed, and their biofilm-forming capacities were analyzed. The results demonstrated no role for PIA/PNAG production in the formation of the FnBP-mediated biofilm (Fig. 4A). We next investigated whether the specific requirement of FnBPs for biofilm development in TSB-gluc and of PIA/PNAG for biofilm development in TSB-NaCl correlated with higher fnbB/ica expression levels. For that, fnbB and icaC transcription levels in strain 132 grown in TSB-gluc and TSB-NaCl were determined by RT-PCR. In agreement with the observed phenotypes, the fnbB mRNA levels were significantly higher in cells grown in TSB-gluc than in cells grown in TSB-NaCl (Fig. 4D). In relation to icaC, despite the fact that strain 132 accumulates significantly higher levels of PIA/PNAG exopolysaccharide in TSB-NaCl (Fig. 1B), RT-PCR experiments did not reveal significant differences in icaC transcriptional levels when cells incubated in both media were analyzed (Fig. 4D), suggesting the existence of posttranscriptional regulatory mechanisms controlling PIA/PNAG synthesis.

FIG. 4.

FnBPA and FnBPB cooperate in the production of the biofilm formed in TSB-gluc. (A) Biofilm formation on microtiter polystyrene plates after 24 h of static incubation at 37°C, in TSB-gluc medium, of single and double mutants in FnBPs in strain 132 and a strain 132 Δica background. (B) Complementation experiments. Mutants 132 ΔfnbAB and 132 Δica ΔfnbAB were complemented with plasmid pFNBA4, which overexpresses the fnbA gene, plasmid pFNBB4, which overexpresses the fnbB gene, or plasmid pALC2073, as a control. (C) Overexpression experiments. Strains 132, RN4220, G-6478, and 10833 were transformed with plasmid pFNBA4, which overexpresses the fnbA gene, plasmid pFNBB4, which overexpresses the fnbB gene, or plasmid pALC2073, as a control. Biofilm formation capacity in microtiter plates was analyzed using TSB-gluc as a growth medium. Representative pictures of at least three independent assays are shown. wt, wild type. (D) RT-PCR analysis of fnbB (left side) and icaC (right side) mRNA levels in S. aureus strain 132 grown in TSB-gluc or TSB-NaCl at 37°C. The gyrB transcript was used as an endogenous control, and the results were expressed as the n-fold difference relative to the control gene (2^−ΔCt, where ΔCt represents the difference in threshold cycle between the target and control genes). Significant differences were detected when fnbB transcription in TSB-gluc and TSB-NaCl was compared (*, P < 0.05; n = 4; Mann-Whitney test).

Sequence analysis of the intergenic region between the fnbA and fnbB genes revealed the existence of a small open reading frame (ORF) (SACOL2510) encoding a protein of 40 amino acids of unknown function that is only present in some of the previously sequenced strains. To exclude the possibility that the deletion of this ORF during fnbAB allelic exchange could influence the biofilm-deficient phenotype displayed by the ΔfnbAB mutant, we performed a nonpolar deletion of this ORF. The resulting mutant 132 ΔSACOL2510 retained the capacity to form a biofilm in both TSB-gluc and TSB-NaCl (data not shown).

We further investigated the functionality of both FnBPA and FnBPB in biofilm development by complementation experiments. First, complementation of the double ΔfnbAB and triple Δica ΔfnbAB mutants with either fnbA or fnbB led to the production of a very thick and robust biofilm, indicating that both FnBPA and FnBPB are sufficient to promote biofilm development when individually overexpressed in the cells (Fig. 4B). Second, overexpression of either FnBPA or FnBPB in the wild-type strain 132 and other genetically unrelated S. aureus strains such as RN4220, G-6478, and 10833 also resulted in a significant induction/enhancement of the biofilm-forming capacity of these strains, demonstrating that promotion of biofilm development by FnBPs occurs irrespective of the genetic background (Fig. 4C). Overall, these results identified FnBPs, among all LPXTG proteins, as essential components of the biofilm formed by S. aureus strain 132 under TSB-gluc conditions.

FnBPs induce biofilm formation under continuous-flow conditions.

We next investigated whether S. aureus strain 132 retained the capacity to switch between proteinaceous and exopolysaccharidic biofilm matrixes when grown under continuous-flow culture conditions in microfermenters. In agreement with previous results, visualization of both macroscopic biofilm development in microfermenters and biofilm cell density on removable glass slides showed that strain 132 ΔfnbAB lost the capacity to form a glucose-induced biofilm but retained the ability to produce a biofilm after 24 h of growth in TSB-NaCl. Conversely, a 132 Δica mutant lost only the capacity to form a NaCl-induced biofilm (Fig. 5A). Together, these results indicate that FnBPs are also able to promote cell-to-cell interactions and biofilm formation under continuous-flow conditions in TSB-gluc.

FIG. 5.

Biofilm formation under continuous-flow conditions. (A) Biofilm development of wild-type S. aureus strain 132 and its ΔfnbAB and Δica mutants in microfermenters under continuous-flow conditions in TSB-gluc or TSB-NaCl media after 24 h at 37°C. The microfermenters contained the glass slides where bacteria formed the biofilms. The results of a representative experiment are shown. (B) Scanning electron microscopy analysis of the biofilms developed in microfermenters. Scanning electron micrographs of the S. aureus strain 132 Δica mutant grown in TSB-gluc medium (a, c) and the strain 132 ΔfnbAB mutant grown in TSB-NaCl medium (b, d) at ×10,000 (a, b) and ×30,000 (c, d) magnifications. Scale bars: a and b, 2 μm; c and d, 1 μm.

Next, we compared the architectures of the FnBP- and PIA/PNAG-dependent biofilm matrixes by scanning electron microscopy. For that, biofilms developed in microfermenters by strain 132 Δica in TSB-gluc medium and strain 132 ΔfnbAB in TSB-NaCl medium were observed. The scanning electron micrographs revealed that cells growing in a FnBP-mediated biofilm form highly dense bacterial aggregates where a close contact between cells exists. The presence of extracellular matrix surrounding the bacteria or occupying the little void space between bacteria was undetectable (Fig. 5B). In the case of the PIA/PNAG-dependent biofilm, cells were embedded in an abundant extracellular matrix mesh that interconnected the bacteria (Fig. 5B).

Role of agr in the S. aureus strain 132 biofilm formation process.

Sequence analysis of the S. aureus strain 132 genome revealed the existence of a single nucleotide deletion on agrC that causes a frameshift, which results in the premature inactivation of the protein. Since it has been shown that transcription of fnb genes is negatively regulated by agr (44, 56, 57), we decided to investigate the impact of such agr deficiency on the biofilm formation process of strain 132. For that, we complemented strain 132 with the pALC2073RNAIII plasmid that contains RNAIII under the control of a tetracycline-inducible promoter. Expression of RNAIII was confirmed by hemolytic activity on sheep blood agar plates (data not shown). We also generated a complete deletion of agrBDCA to exclude any effect of agrA on FnBP expression (40).

As shown in Fig. 6, expression of RNAIII in strain 132 reduced the FnBP levels and consequently its biofilm formation capacity in TSB-gluc, whereas it did not affect PIA/PNAG expression and biofilm formation in TSB-NaCl. Deletion of agrBDCA neither affected FnBP levels nor biofilm formation capacity compared to strain 132.

FIG. 6.

Complementation with RNAIII reduces FnBP-mediated biofilm formation. (A) Biofilm formation on microtiter plates and FnBP levels of S. aureus strain 132, its Δagr mutant, and strain 132 transformed with pALC2073RNAIII or pALC2073 empty plasmid after 24 h of static incubation at 37°C in TSB-gluc medium. Tetracycline was added at 125 ng/ml to plasmid-containing strains to facilitate induction (upper panel). Surface-associated protein extracts were analyzed by SDS-PAGE and detected by Western ligand affinity blotting using biotinylated human fibronectin as a ligand (bottom panel). (B) Biofilm formation on microtiter plates and PIA/PNAG levels of S. aureus strain 132, its Δagr mutant, and strain 132 transformed with pALC2073RNAIII or pALC2073 empty plasmid after 24 h of static incubation at 37°C in TSB-gluc medium. Tetracycline was added at 125 ng/ml to plasmid-containing strains to facilitate induction (upper panel). For dot blot analysis, a 1:25 and a 1:50 dilution of the samples were spotted onto nitrocellulose membranes and PIA/PNAG production was detected with an anti-PIA/PNAG antiserum (bottom panel).

SOS response induces FnBP expression and biofilm formation.

It has been described previously that LexA specifically binds to the fnbB promoter region, repressing gene transcription and thus protein expression until high or persistent DNA damage occurs (7). We wondered whether activation of the SOS response could enhance FnBP-mediated biofilm formation. To test this hypothesis, biofilm formation assays were performed on polystyrene plates in the presence of mitomycin C as a SOS response inductor. The biofilm formation capacities of the wild-type strain and ΔfnbA and Δica mutants increased when mitomycin C (0.05 μg/ml) was added to the growth medium (Fig. 7A). As expected, such induction was observed neither in the ΔfnbB nor in the ΔfnbAB mutant. To prove that the addition of mitomycin C correlated with increased FnBP expression, a Western ligand affinity blotting was performed in the presence of increasing concentrations of mitomycin C. As expected, mitomycin C induced accumulation of FnBPs in a concentration-dependent manner (Fig. 7B).

FIG. 7.

The activation of a LexA-dependent SOS response induces FnBPB expression, resulting in a reinforcement of biofilm development. (A) Biofilm formation on microtiter plates of S. aureus strain 132 and its ΔfnbA, ΔfnbB, ΔfnbAB, and Δica mutants after 24 h of static incubation at 37°C in TSB-gluc medium supplemented with or without 0.05 μg/ml of mitomycin C. A representative picture of at least three independent assays is shown. (B) Detection of FnBP levels in the presence of increasing mitomycin C concentrations. Strain 132 surface-associated protein extracts prepared from cells grown for 4 h, in TSB-gluc medium supplemented with increasing mitomycin C concentrations, were analyzed by SDS-PAGE and detected by Western ligand affinity blotting using biotinylated human fibronectin as a ligand. (C) Biofilm formation on microtiter plates of S. aureus strain 132 and its lexA(G94E) mutant after 24 h of static incubation at 37°C in TSB-gluc medium supplemented with or without 0.05 μg/ml of mitomycin C. A representative picture of at least three independent assays is shown.

It is known that substitution of glutamate for glycine at the Ala-Gly LexA cleavage site results in a noncleavable repressor [LexA(G94E)] that is less sensitive to SOS induction (29). To confirm the role of LexA in the mitomycin C-mediated FnBP biofilm induction, we replaced the lexA coding sequence with the noncleavable lexA allele in the wild-type strain 132 and found that mitomycin C was not able to induce FnBP expression and biofilm formation in the resulting mutant (Fig. 7C).

Together, these results demonstrate that an increase of FnBPB expression by activation of a LexA-dependent SOS response promotes the development of a FnBP-dependent biofilm.

FnBPs contribute to subcutaneous catheter infection.

Considering the advantage that strain 132 is able to produce at least two types of biofilms that differ in their matrix nature, we investigated which one, the PIA/PNAG- or the FnBP-mediated biofilm, plays a more relevant role in vivo. To address this question, a murine subcutaneous catheter infection model was used. A total of 60 mice divided in groups of 10 were used in the experiment. Catheters were aseptically implanted in the subscapular spaces of the mice and coinfected with equal numbers of strain 132 and its corresponding fnbAB mutant, with strain 132 and its corresponding ica mutant, and with strain 132 and its double Δica fnbAB mutant. Bacteria were recovered from implanted catheters at 3 or 7 days postinfection, and the number of bacteria was determined by a plate count as described in Materials and Methods. To evaluate and quantify the contribution of each strain tested to the infection, we used the competition index method. Using this technique, a competition index equal to 1 indicates similar virulence for both strains, a competition index significantly greater than 1 indicates a defect in virulence of the mutant strain, and a competition index significantly lower than 1 indicates the opposite. Prior to the start of the coinfection assays, a competition experiment was performed with all strains tested to confirm that strain growth was not affected by coincubation with others.

The results showed that after 3 days, both strain 132 fnbAB and strain 132 Δica fnbAB mutants showed a significant defect in catheter invasion in relation to the wild-type strain (Fig. 8A). This defect became very significant after 7 days postinfection (Fig. 8B). No significant differences were found when strain 132 ica and the wild-type strain were compared, at either 3 days or 7 days postinfection (Fig. 8).

FIG. 8.

Comparison of the contributions of a PIA/PNAG- and a FnBP-dependent biofilm to S. aureus virulence. Competition index of fnbAB, Δica fnbAB, and ica mutants tested against the wild-type (wt) 132 strain. Catheters were coinfected with equal numbers of the strains tested, and bacteria were recovered from implanted catheters and counted after 3 days (A) or 7 days (B) postinfection. Competition indexes were calculated and statistical differences were determined with the Wilcoxon's test. Horizontal bars represent the means of the competition indexes. Error bars represent standard errors of the means. Asterisks indicate that competition indexes are significantly greater than one (*, P < 0.05 [significant]; **, P < 0.01 [very significant]). ns, nonsignificant differences.

Together, these results strongly suggest that in strain 132, FnBPs play a more relevant role than PIA/PNAG exopolysaccharide in the colonization of implanted devices.

DISCUSSION

One of the most unforeseen results revealed by the intense study of the S. aureus biofilm formation process during the last few years is the existence of surface proteins able to induce biofilm development in the absence of exopolysaccharides. The existence of alternative mechanisms to induce biofilm development implies that each S. aureus isolate might have specialized in the development of a particular type of biofilm matrix that would better suit the environment where it grows. Alternatively, different mechanisms might coexist in the same isolate, and thus, bacteria would make use of a particular one depending on specific environmental conditions. In support of the last hypothesis, a spontaneous switch to a proteinaceous biofilm matrix in a Staphylococcus epidermidis ica::IS256 insertion mutant has recently been reported (20). In our study, we have identified and analyzed a clinical MRSA isolate able to switch between a polysaccharidic (PIA/PNAG) and a proteinaceous (FnBP) biofilm matrix depending on environmental conditions. The importance of this strain lies in the possibility of studying, on one hand, the specific properties that each biofilm matrix bestows on bacteria and, on the other hand, the signal transduction mechanisms involved in the modulation of the biofilm matrix composition according to specific environmental cues.

We made use of scanning electron microscopy to compare the structures of the FnBP- and PIA/PNAG exopolysaccharide-dependent biofilm matrixes. The results revealed that cells growing in a FnBP-mediated biofilm were closely attached without detectable extracellular matrix surrounding the bacteria. In contrast, PIA/PNAG-dependent biofilms showed abundant extracellular material in which bacteria were embedded. Thus, the FnBP-dependent multicellular behavior does not fit, sensu stricto, into the conventional biofilm definition of “a microbially derived sessile community characterized by cells that are irreversibly attached to a substratum or interface or to each other and embedded in a matrix of extracellular polymeric substances that they have produced” (12). Taking this into account, and considering that scanning microscopy analysis of a Bap-mediated biofilm showed similar tightly attached bacteria (our unpublished results), we propose that a protein-mediated biofilm should be referred to as “multicellular behavior.”

The versatility of the elements responsible for S. aureus multicellular behavior is not restricted to the simple option between a polysaccharidic or a proteinaceous component. S. aureus cells can opt for at least four different surface proteins, Bap, SasG, protein A, and FnBPs, to achieve protein-mediated multicellular behavior (9, 11, 27, 33, 36, 46). Thus, the question arises of whether the properties that the community lifestyle confers to bacteria differ depending on the protein that promotes cell-to-cell interactions. In the case of strain 132, we performed the systematic disruption of the 20 genes encoding LPXTG proteins present in its genome to find out the protein(s) involved in the production of a protein-mediated biofilm formed under TSB-gluc conditions. Analysis of every mutant revealed that only the absence of FnBPs inhibited the capacity of strain 132 to produce multicellular communities in that medium. Thus, in the case of this particular strain and under TSB-gluc environmental conditions, neither SasG nor protein A seems to play any role in multicellular behavior. This situation might simply be due to compensatory effects between other proteins that would mask the effect of the single mutants. Alternatively, it is likely that the laboratory conditions used in our study might not be appropriate for inducing SasG or protein A expression to the levels required to induce a macroscopically noticeable multicellular behavior. In support of this assumption, our results confirmed a direct relationship between the FnBP expression levels and the strength of the multicellular phenotype. First, overexpression of FnBPs in wild-type strains induced multicellular behavior. Second, the enhanced FnBP accumulation due to the spontaneous agr mutation promoted multicellular behavior. Third, induction of FnBPB expression through activation of the SOS response significantly increased the number of bacteria taking part of the multicellular community. In this regard, regulation of FnBPB expression by the SOS response has been previously determined (7). In their study, Bisognano et al. showed that LexA is a repressor of fnbB expression, and consequently, SOS response induction by commonly used fluoroquinolones resulted in the increased expression of the FnBPB protein on the bacterial surface (7). As FnBPs have been shown to play an important role during invasion of mammalian cells (47, 48) and in mediating adherence to different host extracellular matrix components, such as fibronectin, fibrinogen, elastin, or tropoelastin (23, 43, 55), an adverse consequence of SOS induction by antibiotics would be that it might facilitate colonization of host cells and medical implants (13, 17). Accordingly, our results suggest that an increase in FnBPB expression through activation of the SOS response is sufficient to induce a noticeable increase in the capacity of the strain to produce multicellular communities on abiotic surfaces in vitro.

It has recently been shown that spontaneous agr mutants frequently arise in vivo during the course of the infection (51). Although the benefits that agr genotypic instability might provide to S. aureus in human infection remain unknown, different studies have shown that agr mutants have a tendency to form biofilms in vitro. Enhanced biofilm formation by agr mutants has been attributed to (i) decreased production of RNAIII-encoded δ-toxin, which has surfactant properties (54), (ii) decreased production of extracellular proteases, which can digest the proteinaceous biofilm matrix (8), and (iii) increased accumulation of protein A that promotes aggregation and cell-to-cell interactions (33). In addition to these mechanisms, our results indicate that the increased accumulation of FnBPs accounts for another mechanism through which agr mutation can contribute to biofilm development. A previous study investigating the role of agr in clinical S. aureus MRSA isolates reported that inactivation of agr increased ica-independent biofilm development in 5 out of the 13 isolates investigated (37). However, it is likely that this study underestimated the implication of agr since some of these isolates might already be natural agr mutants and consequently complementation with RNAIII would be needed to determine the effects of agr on biofilm development. Afterward, the same group demonstrated that FnBPs were responsible for promoting biofilm development in five genetically unrelated MRSA isolates (36). These results suggest that MRSA isolates, like the one used in this study, with the capacity to switch between a FnBP- and a PIA/PNAG-based biofilm might frequently occur in clinical settings.

To address the biological significance of PIA/PNAG and FnBPs during colonization of implanted devices by strain 132, we compared the capacities of the wild-type strain and the isogenic fnbAB, ica, and Δica fnbAB mutants to colonize an implanted catheter by using a murine foreign-body infection model. We found that the absence of FnBPs caused a significant reduction in the capacity of the strain to colonize catheters in vivo after 3 and 7 days postinfection. In contrast, the absence of the icaADBC operon did not lead to any significant reduction. It is important to note that in our experimental model, attachment of S. aureus to native implants occurred because catheters were immediately infected after implantation without making use of a precoating period with plasma proteins such as fibronectin. The role of PIA/PNAG exopolysaccharide during the infection process has been addressed in various studies with contradictory outcomes. Thus, PIA/PNAG production has been described as essential for virulence in murine models of systemic infection while deletion of the ica locus failed to alter virulence in device-related infection models (15, 16, 25, 26). Our results are in agreement with a previous study focused on the analysis of the role of SarA in biofilm development in vivo. In this study, Beenken et al. found that an S. aureus UAMS-1 sarA mutant presented a reduced capacity to colonize catheters in vivo whereas the UAMS-1 ica mutant colonized the catheters as efficiently as the wild-type strain using a murine model in which catheters were infected 1 h after implantation (6). Interestingly, biofilm formation in the UAMS-1 isolate was enhanced by coating the substrate with plasma proteins and was not dependent on PIA/PNAG production (5, 6). Based on these data, and taking into account that expression of FnBPs is completely abolished in the absence of sarA (36, 46), it is tempting to speculate that the UAMS-1 strain might produce a FnBP-dependent biofilm whose expression would be completely abolished in the sarA mutant strain. Thus, a reduced recovery of the sarA mutant in coinfection experiments with implanted catheters might be due to FnBP deficiency (6).

An association between methicillin resistance and protein-dependent multicellular behavior in S. aureus has been postulated (14, 36, 37). If such an association exists, elucidation of the properties that FnBPs confer to the bacterial community during the colonization of implanted medical biomaterials may contribute to our understanding of the factors that play a role in the persistence and transmission of MRSA isolates across the hospitals of developed countries.