In this study, the shaking culture was performed in a water bath using a test tube of φ15 mm × 150 mm with shaking at 140 rpm at an angle of 45°. The bacterial strains and plasmids used in this study are listed in Supplementary Table 1. Standard strain S. aureus FK300, a rsbU-repaired (Giachino et al., 2001) derivative of NCTC8325-4 (Novick, 1967; Herbert et al., 2010), was used in a functional study of the role of icaB. DNA restriction system-deficient S. aureus RN4220 (Kreiswirth et al., 1983) was used as the initial recipient for manipulation of recombinant plasmids. S. aureus was routinely grown in brain heart infusion (Becton, Dickinson and Company, MD, United States) broth, tryptic soy broth (TSB; Becton, Dickinson and Company), or tryptic soy agar plates. Tetracycline (5 μg/ml) or chloramphenicol (5 μg/ml) was added to retain plasmids. The Escherichia coli strain DH5α was used for plasmid construction and maintenance. E. coli was grown in lysogeny broth (LB; 5 g yeast extract, 10 g polypeptone, and 10 g NaCl per liter; pH 7.2) or LB agar. Ampicillin (100 μg/ml) or tetracycline (12.5 μg/ml) was added to the medium when the plasmid was to be retained.

Routine DNA manipulation was performed as previously described (Sambrook et al., 1989). FK300 mutants were constructed via allelic replacement using pKFT (Kato and Sugai, 2011). Polymerase chain reaction (PCR) was performed using TaKaRa LA Taq (TaKaRa, Shiga, Japan). The thermal cycling conditions were as follows: 94°C for 2 min, followed by 30 cycles of 94°C for 15 s, 50°C, for 30 s, and 68°C for 2 min, finishing with a final extension step at 72°C for 1 min. The oligonucleotides used in this study are listed in Supplementary Table 2. Fragments were cloned into the plasmid pGEM-T Easy (Promega, Madison, WI, United States) using TA cloning and transformed into E. coli DH5α. A fragment excised from the pKS101 plasmid using a restriction enzyme was cloned into the pKFT plasmid and transformed into E. coli DH5α. Recombinant plasmids were introduced into S. aureus RN4220 via electroporation (Kraemer and Iandolo, 1990; Löfblom et al., 2007; Hisatsune et al., 2016). The modified plasmids were then electroporated into S. aureus FK300 cells for allelic replacement. Marker-less deletion mutants in tetracycline-sensitive colonies were screened using PCR. Fragments were verified using DNA sequencing with a BigDye Terminator v 3.1 Cycle Sequencing Kit (Applied Biosystems, Waltham, MA, United States). In the complementation experiments, genes were amplified using PCR with the corresponding primer pairs and then cloned into the SmaI site of pKAT (Kato, 2004). The plasmids, pKS103 and pKS104, carrying the icaR and icaB coding regions of FK300, respectively, were constructed and transformed into the S. aureus strains listed in Supplementary Table 1 using electroporation. Inserts in all plasmid constructs were verified using PCR and DNA sequencing.

A biofilm assay using polystyrene plates was performed as described previously (Heilmann et al., 1996b) with a few modifications. Briefly, overnight cultures were diluted 1:100 in TSB. Ten microliters of this diluted solution was transferred in triplicate into flat-bottom, 96-well polystyrene plates (TrueLine; Nippon Genetics Co., Ltd., Japan) containing 100 μl of TSB or TSB plus 1% glucose. Following incubation at 37°C for 20 h, the wells were gently washed thrice with 320 μl of sterile phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄·12H₂O, and 1.8 mM KH₂PO₄; pH 7.4), and the biofilm was stained with 1% crystal violet for 15 min. Unbound crystal violet was then removed by washing the plate in a container by immersing and agitating it gently 10 times in tap water and then drying it. Biofilm-bound crystal violet was solubilized in 150 μl of 33% glacial acetic acid at 25°C for 15 min. The extracts were diluted 10-fold, and then, absorbance at 590 nm was measured using an Immuno-Mini NJ-2300 spectrophotometer (Nalgene Nunc International K. K., Tokyo, Japan). Each assay was performed in triplicate and repeated three times.

Bacteria from the plate were inoculated into 3 ml of TSB in a test tube (φ15 × 150 mm), placed in a water bath, and shaken at 140 rpm at an angle of 45° for 6 h. The culture was vortexed (Vortex-Genie 2; Scientific Industries, Inc.) for 10 s. Turbidity of the culture was visually monitored using video photography. For swapping experiments, the culture supernatant of interest was passed through a 0.2-μm filter, incubated with the cells of interest, and washed several times with 10 mM PBS; thereafter, the autoaggregation assay was carried out as described above.

Real-time PCR of ica operon in S. aureus has been previously described (Yu et al., 2017). Overnight S. aureus cultures were diluted in TSB containing 1% glucose to an initial optical density (OD) of 0.02 at 660 nm and harvested after 6 h of incubation with shaking at 37°C. Total RNA was isolated using a FastRNA Pro Blue kit (MP Biomedicals, Santa Ana, CA, United States) according to the manufacturer’s instructions. DNA was extracted by treatment with RQ1 RNase-free DNase (Promega) at 37°C for 30 min. After DNase inactivation, PCR was performed to verify the absence of contaminating DNA. RNA was then reverse-transcribed using a Transcriptor First-Strand cDNA Synthesis Kit (Roche, Mannheim, Germany). The resulting cDNA was diluted 10-fold with Tris-EDTA buffer (10 mM Tris–HCl and 1 mM EDTA; pH 8.0) and used as a template in the real-time PCR. qRT-PCR was performed using the SsoAdvanced Universal SYBR Green SuperMix (Bio-Rad, Hercules, CA, United States) and a CFX96 real-time PCR detection system (Bio-Rad). The thermal cycling conditions were as follows: 95°C for 1 min, followed by 40 cycles of 95°C for 15 s, 60°C (icaA), or 62°C (gyrB) for 15 s, and 72°C for 30 s. All PCR runs were performed in triplicate, and data were analyzed using the CFX Manager software (version 3.0; Bio-Rad) according to the manufacturer’s instructions. The housekeeping gene, gyrase subunit B (gyrB), was used as a reference gene to normalize the expression level of the target gene in each reaction. The primers used for real-time PCR are listed in Supplementary Table 2. All samples were inspected 3 times and the data were analyzed using the 2^-ΔΔCt method (Livak and Schmittgen, 2001).

Electron microscopic observations were performed using TEM, as previously reported (Yamada et al., 1996). Bacterial cells were harvested, washed twice with 0.1 M PBS, and collected using centrifugation (2,300 × g, 15 min). For TEM, cells were fixed with 2.5% glutaraldehyde and 1% OsO₄. Samples were dehydrated using an ethanol series and embedded in New Spurr (Agar Scientific Ltd., United Kingdom). Ultrathin sections were cut with an ultramicrotome (ULTRACUTS, Leica, Tokyo, Japan) and examined with a JEOL JEM-2000 EXII electron microscope (JEOL Ltd., Tokyo, Japan) at 80 or 100 kV.

Elemental analysis of autoaggregates from the culture supernatant of FK300Δ5bpΔBm was performed using a scanning electron microscope connected to an energy dispersive X-ray spectroscope (Miniscope TM-3030, Hitachi High-Technologies Corporation, Tokyo, Japan).

The culture supernatant was concentrated via centrifugation using an Amicon Ultra 15 ml filter (3 kDa cut off, Merck Millipore Corporation, Darmstadt, Germany) for 40 min at 4,000 × g, and the concentrate was treated with trichloroacetic acid (final conc. 5%) to remove proteins by centrifugation for 15 min at 6,000 × g. Subsequently, the supernatant was subjected to gel permeation chromatography using Shim-pack Diol-300 (500 × 7.9 mm) with water as the mobile phase at a flow rate 0.5 ml/min. The effluent was fractionated every 2 ml from fraction 1–21, 11 min after sample injection. Samples necessary for hydrolysis were treated at 100°C for 2 h in the presence of 2 N HCl and rehydrated with distilled water after evaporation for the colorimetric assay.

Fractions obtained by HPLC were analyzed for amino sugars using the Morgan-Elson assay (Enghofer and Kress, 1979). Samples (100 μl) were incubated with 20 μl acetone containing 1.5% acetic anhydride and 100 μl boric acid buffer (pH 9.0) at 95°C for 8 min and kept on ice. Then, the samples were mixed with 750 μl of p-(dimethylamino) benzaldehyde containing 12.5% HCl and 50 μl of 2-ethoxyethanol, and left to stand at room temperature for 15 min. OD at 545 nm was measured using a microplate reader (Varioskan LUX multimode reader, Thermo Fisher Scientific, Waltham, MA, United States).

PNAG dot blotting was performed on each fraction obtained by fractionating the supernatant using gel filtration chromatography (Cramton et al., 1999). Alternatively, Bacteria from the plate were inoculated and incubated in 3 ml TSB medium at 37°C for 6 h with shaking. The incubated bacteria were centrifuged (8,000 g × 2 min), and the supernatant and bacteria were separated. To 1 ml of supernatant passed through a 0.2 μm sterile filter, 50 μl of Proteinase K (50 μl/ml; TaKaRa) was added and allowed to react at 55°C for 30 min, followed by incubation at 85°C for 30 min to inactivate the protease. The bacterial cells were washed once with PBS, resuspended in 100 μl of 0.5 M EDTA, and boiled at 100°C for 5 min. To 40 μl of the supernatant after centrifugation, 10 μl of Proteinase K (50 μl/ml; TaKaRa) was added and reacted at 55°C for 30 min, followed by incubation at 85°C for 30 min to inactivate the protease. The resulting bacterial surface samples and culture supernatant samples were serially diluted. The bacterial cell surface sample, supernatant sample or HPLC fraction was dropped onto a nitrocellulose membrane (Amersham Protran NC 0.45; General Electric Company, Connecticut, United States) and the membrane was immersed in a TBS-T [50 mM Tris–HCl (pH 8.0), 150 mM NaCl, and 0.05% Tween 20 (Sigma-Aldrich, Inc.)] solution containing 5% skim milk and allowed to react for 1 h to block the membranes. The membrane was then washed thrice with TBS-T for 15 min and incubated with rabbit anti- Poly-β-1,6-N-acetyl-D-glucosamine (PNAG) antiserum supplied by Dr. Gerald Pier (Skurnik et al., 2010) diluted 1:4,000 with 0.5% skimmed milk in PBS-T for 2 h at room temperature. Bound antibody was detected with peroxidase-conjugated goat anti-rabbit immunoglobulin G (IgG) antibodies (MP Biomedicals, LLC-Cappel Products, Ohio, United States; 1:6,000) and developed using Pierce enhanced chemiluminescence (ECL) western blotting substrate (Thermo Fisher Scientific, Rockford, IL, United States). Results were detected using a charge-coupled device camera-based imager (GE Healthcare Life Sciences, Piscataway, NJ, United States) or FujiRX-U film (Fujifilm, Tokyo, Japan), and developed using HI-RENDOL (Fujifilm) and HI-RENFIX (Fujifilm).

MS analysis was performed using an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific) on a direct infusion MS using a nanospray capillary.

Samples in 50% methanol containing 0.1% formic acid were ionized using electrospray ionization in positive ion mode, and the following parameters were set: spray voltage, 1.5 kV; tube lens voltage, 250 V(max); source fragmentation, ON (100 Vmax); and mass acquisition range, m/z 200–4,000. Data were processed using the Xcalibur software package provided by Thermo Fisher Scientific.

Differences between two means were evaluated by the Mann–Whitney U test. Data were analyzed by Kruskal-Wallis test of variance to compare multiple means. Differences with p-values less than 0.05 were considered statistically significant. Statistical analysis was performed using R 4. 1.3 was used, with a significance level of 5%.

The deletion of the 5-bp motif mutant from S. aureus wild type (WT) strain FK300 via allelic replacement resulted in the isolation of three independent mutants (Δ5bpBm 1, 2, and 3) showing a common but unusual phenotype (Supplementary Figure 1). Most isolates upon 5-bp deletion mutation showed normal colony morphology, but a few revealed unusual phenotypes, as indicated by colonies showing shiny flat morphology, compounded by fusion between adjacent colonies (Figure 1, colony). In broth culture, most possible 5-bp motif-deletion mutants revealed a super-biofilm-producing phenotype, where these mutants exhibited thick pellicles in test tube broth culture, as observed in a previous study (You et al., 2014; Figure 1, left, Δ5bp, before vortex). By contrast, the three isolates in our study did not show any pellicles on the wall of the test tube (Figure 1, right, Δ5bpBm 2, before vortex). Furthermore, weak mixing using a vortex mixer for a few seconds induced instant autoaggregation and subsequent sedimentation of the bacteria, resulting in a clear supernatant (Figure 1, right, Δ5bpBm 2, after vortex), whereas the genuine mutant Δ5bp remained turbid (Figure 1, left, Δ5bp, after vortex; Supplementary Video 1).

Whole-genome sequencing of the three mutants established that their 5-bp motif was defective, indicating that the deletion of 5-bp had been successful. In addition, we identified a point mutation in the coding sequence of icaB in each strain (Supplementary Figure 2). For Δ5bpBm 3, a frameshift mutation occurred due to the insertion of T, resulting in the creation of a termination codon 67 bases downstream, whereas Δ5bpBm 2 and Δ5bpBm 1 demonstrated single and double missense mutations, respectively. Thereafter, FK300Δ5bpBm 2 was used as the representative strain in the experiment. When FK300Δ5bpBm 2 mutant was complemented with icaB, it formed adherent biofilms after culture and shown pellicle formation in vitro (Figure 2A). After voltexing, the formed biofilm was not completely dissolved as that of Δ5bp probably due to overproduction of adherent biofilm and remained detached was suspended as clumps. Biofilm assays using plastic plates substantiated that FK300Δ5bpBm 2 had completely lost the ability to form adherent biofilms, whereas complementation with icaB restored strong biofilm production (Figure 2B). These results suggested that icaB was involved in the unusual phenomenon displayed by the three FK300Δ5bpBm 2 mutant. Therefore, we generated FK300Δ5bpΔicaB and verified that the mutant showed the same phenotype as FK300Δ5bpBm 2. This phenotype was also successfully complemented by icaB. To establish that this phenomenon was dependent on the ica operon, we overexpressed icaR in FK300Δ5bpΔicaB to regulate the production of PNAG. FK300Δ5bpBm 2 picaR completely lost the autoaggregating phenotype observed in FK300Δ5bpΔicaB, indicating its dependence on the ica operon (Figure 2).

Deletion of the 5-bp motif is known to increase biofilm formation (Yu et al., 2017). However, other regulatory repressor regullators (rob and icaR) are also involved in biofilm formation, and the deletion of these factors has been shown to increase biofilm formation (Jefferson et al., 2004; Yu et al., 2017). We therefore created icaB deletion mutants and double deletion mutants combining the repressor and ΔicaB to see if it would develop a self-aggregating phenotype as observed with Δ5bpΔicaB. Although all three deletion mutants lost the adherent biofilm formation ability (Figure 3B), only FK300Δ5bpΔicaB showed an autoaggregating phenotype (Figure 3A). This result was reproduced when the culture was shaken even 250 rpm (Unpublished data). Tiny autoaggregates were observed in the broth of FK300ΔrobΔicaB and FK300ΔicaRΔicaB, but there were no pellicles and the supernatant remained turbid even after mixing using a vortex mixer. This may likely have been due to the differences between the magnitudes of activation of the ica operon in the deletion mutants (Δrob, ΔicaR, and Δ5bp), and we, therefore, measured mRNA levels of ica operon in the respective mutants. The results showed that expression of ica operon by Δ5bp was much stronger than that by Δrob or ΔicaR (Figure 4). The ica operon expression levels of each defective mutant compared to WT were about 31.46-fold for Δrob, 173.90-fold for ΔicaR, and 2486.17-fold for Δ5bp. This is consistent with the report of Yu et al. (2017). Based on these observations, we analyzed fractions of the culture supernatant and cell surface of each strain using anti-PNAG antiserum dot blot in order to investigate the actual production of PNAG. No signal was observed on the cell surfaces of FK300Δ5bpΔicaB, FK300ΔicaRΔicaB, or FK300ΔrobΔicaB. However, in the culture supernatant, a clear strong signal demonstrating a reaction with the anti-PNAG antiserum was seen for FK300Δ5bpΔicaB and a weak one for FK300ΔicaRΔicaB, but not for FK300ΔrobΔicaB (Figures 5A,B). On the other hand, strong signals were detected in both the culture supernatant and the cell surface in FK300Δrob, FK300ΔicaR, and FK300Δ5bp, respectively as expected. For FK300, a very weak signal was observed only at the cell surface, and no signal was detected in the culture supernatant. For FK300ΔicaB, no signal was observed in both culture supernatant and cell surface.

We conducted a swapping experiment to identify the factor(s) involved in autoaggregation on the cell surface or in the culture supernatant of FK300Δ5bpΔicaB. The culture medium of each bacteria was separated into cells and culture supernatant, and a vortex experiment was performed by multiplying the selected supernatant with the selected cells to see which sample was important for aggregation. The strains used in the experiment were FK300, FK300Δ5bp, or FK300Δ5bpΔicaB. The autoaggregating factor(s) was present in the culture supernatant of FK300Δ5bpΔicaB, but not in the culture supernatant of FK300Δ5bp or on the cell surface of FK300Δ5bpΔicaB (Figure 6). The results suggest that the important factor for aggregation is in the culture supernatant of FK300Δ5bpΔicaB. We examined the ultrastructure of the autoaggregates using transmission electron microscopy (TEM). The autoaggregate preparation obtained from sedimented FK300Δ5bpBm 2 was compared with a cell pellet of WT FK300. There was no significant difference between the ultra-structures of the cell wall peptidoglycan layers of the strains (Figure 7A). However, string-like substances surrounding FK300Δ5bpBm 2 were observed, implying the presence of string-like substances in the culture supernatant of FK300Δ5bpBm 2. We further analyzed the elements of the autoaggregates obtained from the culture supernatant of FK300Δ5bpΔicaB using a TM-3030 Hitachi microscope connected to energy dispersive X-ray spectroscope. Dried autoaggregate was spotted using a miniscope, and the spotted site was analyzed. The results suggested that the ratios of carbon, oxygen, and nitrogen were similar to those of N-acetylglucosamine (Figure 7B). To gain further insights into the biochemical structure of the string-like substances surrounding the mutant, we prepared a culture supernatant of FK300Δ5bpΔicaB, which was then vigorously shaken to determine whether any autoaggregation occurred without bacterial cells and also whether any tiny but distinct whitish autoaggregates would appear after shaking (Supplementary Figure 3). Such substances appeared in FK300Δ5bpΔicaB culture supernatants but not in FK300 and FK300Δ5bp culture supernatants. We studied in vitro autoaggregation in the culture supernatant of FKΔ5bpΔicaB, to analyze the inhibitory effect of several additives, including NaCl, ethylenediaminetetraacetic acid (EDTA), dispersin B, proteinase K, and DNase. Only dispersin B, a β-hexosaminidase that specifically hydrolyzes β-1,6-glycosidic bonds in acetylglucosamine polymers (Kaplan et al., 2003; Ramasubbu et al., 2005; Itoh et al., 2005), inhibited the appearance of aggregates in vitro. This finding strongly suggested that acetylglucosamine polymers were the major factor(s) involved in the autoaggregation of FK300Δ5bpΔicaB.

To further analyze the biochemistry of these polymers, the FK300Δ5bpΔicaB culture supernatant was fractionated using gel filtration chromatography, and each fraction with or without hydrolysis was analyzed for amino sugar concentration. A broad peak appeared in the void-volume-column (fractions 3–8) of the HCl-hydrolyzed fractions, indicating that the polymers in the supernatant of FK300Δ5bpΔicaB contained amino sugars (Figure 8A). Furthermore, dot blot analysis with antisera against PNAG clearly demonstrated that the corresponding fractions were positive for PNAG (Figure 8B). Collectively, these results strongly suggested that the factor(s) involved in the autoaggregation of FK300Δ5bpΔicaB in the supernatant was an N-acetylglucosamine polymer.

N-acetylglucosamine polymers secreted from the cytosol of S. aureus undergo partial deacetylation by IcaB, which functions as a deacetylase (Vuong et al., 2004). The resulting extracellular positively charged deacetylated polymers interact with the negatively charged cell surface to form a biofilm matrix around the cells. FK300Δ5bpΔicaB lacks IcaB deacetylase and is therefore unable to catalyze the deacetylation of N-acetylglucosamine polymers, which releases the polymer into the culture supernatant. Therefore, we investigated whether the polymers involved in the autoaggregation of FK300Δ5bpΔicaB were deacetylated, by performing mass spectrometry (MS) analysis of the polymers. PNAG on the surface layer of the biofilm-producing strain, FK300Δrob, was extracted and used as a control. MS analysis of PNAG gathered from the surface of FK300Δrob cells showed MS signals with a regular interval of m/z = 203, which corresponded to the N-acetylglucosamine (GlcNAc) molecule. In addition, several signals corresponding to partially deacetylated polymers, such as (GlcN)₁-(GlcNAc)₆, (GlcN)₂-(GlcNAc)₅, and (GlcN)₃-(GlcNAc)₄ between (GlcNAc)₆ and (GlcNAc)₇, were observed (Figure 9A). By contrast, polymers recovered from the culture supernatant of FK300Δ5bpΔicaB showed signals corresponding to (GlcNAc)_n, but not those corresponding to deacetylated fragments (Figure 9B). These results clearly showed that the polymers in the supernatant of FK300Δ5bpΔicaB were polymers of N-acetylglucosamine without deacetylation.

Genomic DNA extraction and whole-genome sequencing were performed as described previously27. The raw data reads of the 5-bp (icaR-icaA intergenic region) deletion mutants from S. aureus FK300, FK300Δ5bpBm 1, 2, and 3, have been deposited in the DDBJ/Sequence Read Archive under the accession numbers DRA013849.