Introduction Staphylococcus aureus is the causative agent of numerous diseases ranging from relatively benign skin conditions to fatal systemic infections. Formation of bacterial biofilms on host tissues and implanted materials contributes to chronic S. aureus infections, as biofilms are exceptionally resistant to host immune response and chemotherapies [1]. Biofilms are multicellular structures encased in a matrix of proteins, polysaccharides, extracellular DNA, and other environmental factors [1], [2]. Biomolecules that digest matrix components (e.g., proteases, DNases, and glycoside hydrolases) can disrupt established biofilms and render detached cells susceptible to antimicrobials [3], [4], [5], [6], [7]. The precise composition of the biofilm matrix varies greatly by strain, physiological state, and nutrient availability [5], [8], [9], [10], [11], [12]. In this study, we examined how growth media affects the composition of the biofilm matrix. This led to the discovery of an extracellular fibril structure in S. aureus biofilms grown in a non-standard rich media. These fibers share morphological and biophysical characteristics with functional bacterial amyloids such as curli in Escherichia coli biofilms, TasA of Bacillus subtilis, and the Fap fimbriae in Pseudomonas aeruginosa [13], [14], [15], [16]. Biochemical and genetic analysis revealed that these fibril structures are composed of small peptides called phenol soluble modulins (PSMs). Mutants incapable of producing PSMs formed biofilms that were susceptible to disassembly by enzymatic degradation and mechanical stress. Previous work has demonstrated that PSMs are surfactant-like peptides that promote biofilm disassembly [17], [18], [19], [20], [21]; exhibit antimicrobial activity against niche bacteria [22], [23], [24]; hinder host immune response by recruiting and lysing neutrophils; and are abundant virulence factors produced by community-associated MRSA strains (CA-MRSA) [18], [25], [26], [27]. The genes encoding the core family of PSM peptides are highly conserved across S. aureus strains: four are expressed from the alpha (αpsm1–4) operon, two are expressed from the beta (βpsm1&2) operon, and the delta hemolysin (hld) is encoded within the regulatory RNA, RNAIII [28], [29], [30]. The significance of the PSMs has only recently been investigated because the coding sequences of the αpsm & βpsm peptides are small enough to have eluded detection by conventional gene annotation programs, and they are still poorly annotated in public databases [29], [30]. We have found that ordered aggregation of PSM peptides into amyloid-like fibers can abrogate the biofilm disassembly activity ascribed to monomeric PSM peptides [12], [17], [18], [19], [20], [21]. Our findings suggest that PSMs can modulate biofilm disassembly using amyloid-like aggregation as a control point for their activity. This is the first report to identify and characterize extracellular fimbriae in the S. aureus biofilm, and our research could lead to new approaches in treating persistent biofilm associated infections. Results Biofilms grown in PNG media resist biofilm disassembly Biofilms that persist in the human body are often resistant to conventional antimicrobial treatment prior to dispersal. To gain insight into how the S. aureus biofilm matrix affects disassembly under different growth conditions, we grew S. aureus flow cell biofilms with various lab media. Next we used enzymes known to target primary matrix components in order to test biofilm resistance (Fig. 1A & 1B). These enzymes include proteinase K (protein), DNaseI (DNA), and dispersin B (polysaccharide). By using a variety of degradative enzymes, we expected to observe complete biofilm eradication. Biofilms grown in tryptic soy broth supplemented with glucose (TSBg) rapidly disassembled after enzymatic treatment (Fig. 1A). However, biofilms grown in peptone-NaCl-glucose (PNG) media did not disassemble after the same enzymatic treatment (Fig. 1B). 10.1371/journal.ppat.1002744.g001 Figure 1 Growth media influences biofilm disassembly. Confocal micrographs of S. aureus SH1000 biofilms grown in TSBg media (A) for 30 hours readily disassemble upon exposure to biofilm matrix degrading enzymes proteinase K, dispersin B, and DNaseI at 0.2 µg/mL each. S. aureus biofilms grown in PNG media (B) for 30 hours fail to disassemble upon exposure to matrix-degrading enzymes. Images are representative of three separate experiments and each side of a grid square represents 20 µm. (C) Biofilms at the air-liquid interface of test tube cultures withstand 1% SDS exposure when grown in PNG media but disassemble when grown in TSBg. Top images show stained test tube biofilms; graph below is quantification of biofilm biomass. * P 90% pure by HPLC. Synthetic peptides were prepared and assayed as previously described [15], [34] to eliminate large aggregates from lyophilization prior to assay. Each dry peptide stock was dissolved to a concentration of 0.5 mg/mL in a 1∶1 mixture of trifluoroacetic acid (TFA) and hexafluoroisopropanol (HFIP). Peptides were then sonicated for 10 minutes and incubated at room temperature for 1 h. Solvent TFA/HFIP was removed by speedvac at room temperature. Dried peptide stocks were stored at −80°C. All assays were performed with equal stoichiometric ratios of 0.1 mg/mL peptide unless otherwise noted. All polymerization assays were performed in 96-well black opaque, polystyrene, TC-treated plates (Corning). Prior to assay, treated peptides were thawed and dissolved in dimethyl sulfoxide (DMSO) to a concentration of 10 mg/mL immediately prior to assay. Freshly dissolved peptides were diluted into sterile ddH2O containing 0.2 mM thioflavin T (ThT) and assayed at room temperature. Fluorescence was measured every 10 minutes after shaking by a Tecan Infinite M200 plate reader at 438 nm excitation and 495 nm emission. ThT fluorescence during polymerization was corrected by subtracting the background intensity of an identical sample without ThT. Additionally, ThT fluorescence and Congo red (CR) absorbance scans were performed on polymerized peptides that were allowed to polymerize for 48 h in ddH2O. Samples were incubated in 0.2 mM ThT or 0.001% (w/v) CR in ddH2O for 15 minutes prior to assay on the Tecan plate reader. CR and ThT scans were corrected by subtracting the background intensity of an identical sample without dye. Circular dichroism spectroscopy Treated peptide stocks were thawed and dissolved in hexafluoroisopropanol (HFIP) to a concentration of 10 mg/mL immediately prior to assay. Triplicate samples consisting of 0.1 mg/mL of each freshly dissolved peptide diluted together in 500 µL sterile ddH2O were incubated with shaking at room temperature for 48 h. Samples were then pelleted at 15,000 RPM for 30 minutes to isolate any aggregated species. The supernatant was carefully removed from the pellet by aspiration and transferred to a clean, sterile eppendorf tube. The remaining pellet was resuspended in 200 µL ddH2O by bath sonication for 10 minutes. The supernatant and pellets of each sample were assayed separately. Far UV circular dichroism (CD) measurements were performed with a Jasco-J715 spectropolarimeter using quartz cells with 0.1 cm path length. CD spectra between 190 and 250 nm were recorded in millidegrees and converted to molar ellipticity using an average MRW of 113 for αPSM1–4, βPSM1&2, and δ-toxin. The average of five scans was recorded at 25°C using a 2 nm bandwidth with a 20 nm min−1 scanning speed. All triplicate samples showed similar ellipticity patterns. Biofilm dispersal assay Synthetic PSM peptides were allowed to polymerize overnight, and fibril formation was verified by TEM imaging. Equivalent concentrations of either polymerized or freshly diluted peptides were added to 24 hours SH1000 biofilms grown in 66% TSB+0.2% glucose and incubated at 37°C for 6 hours. Biofilms were washed to remove non-adherent cells then stained with 0.1% crystal violet, dried, and solublized with acidified ethanol and spectroscopically quantitated at A595. Statistics were performed using a 1-way analysis of variance (ANOVA). Results are expressed as mean ± standard deviation.