The details of the bacterial strains used herein are shown in Table 1 (Simon and Puhler, 1983; Rhodes et al., 2011a). Flavobacterium collinsii GiFuPREF103 was grown in modified Cytophaga (MCP) medium (Wakabayashi and Egusa, 1974) at 20°C. To culture erythromycin-resistant F. collinsii GiFuPREF103 strains, erythromycin (100 μg/mL) was added to the medium. Flavobacterium johnsoniae strains were grown in casitone yeast extract medium at 25°C. For the selection and maintenance of antibiotic-resistant F. johnsoniae strains, antibiotics were added to the medium at the following concentrations: streptomycin, 100 μg/mL and erythromycin, 100 μg/mL. To observe colony spreading, F. collinsii GiFuPREF103 or F. johnsoniae cells were grown in MCP medium at 20°C or 25°C with shaking (175 rpm) overnight, respectively. The cells were centrifuged at 800 × g for 10 min at 20°C. The pellet was resuspended in the washing buffer (10 mM Tris–HCl pH 7.4) by vortexing, and the suspension was centrifuged at 800 × g for 10 min at 20°C. The process was repeated twice. The cells were spotted onto MCP agar medium (agar; BD BACTO Agar, Becton, Dickinson and Co., Franklin Lakes, NJ, USA) in a 3 or 9 cm dish.

Bacterial strains that formed yellow colonies on MCP agar medium and had the ability to disperse were isolated from Plecoglossus altivelis with BCWD. Each strain cell was used as a template and PCR was performed using primer pair 27F & 1500R to amplify the 16S rRNA region. The PCR products were sequenced using primers 27F, 1500R, 800F, and 800R.

MCP broth was inoculated with F. collinsii GiFuPREF103 and incubated under aerobic conditions. Genomic DNA was extracted using a Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions. The library was prepared using MGIEasy FS DNA Library Preparation Set, MGIEasy Circularization Kit, and DNBSEQ-G400 High-throughput Sequencing Set (MGI Tech Co., Shenzhen, China), according to the manufacturer’s instructions. The sample was sequenced using the DNBSEQ-G400 system (MGI Tech Co.) with 2 × 150-bp reads. Subsequent de novo assembly using the Spades (ver. 3.13.2) protocol yielded 189 contigs. Each contig sequence was annotated using DFAST and deposited as F. collinsii GiFuPREF103 in the DDBJ/EMBL/GenBank database under the accession number BOVI01000001–BOVI01000189 ( Table S1 ).

The 16S rRNA phylogenetic tree was constructed using the neighbor-joining method with the MEGA tool (ver. 11). DNA sequences of 16S rRNA of Flavobacterium spp. were obtained from the National Center for Biotechnology Information database ( Table S2 ).

For the construction of the ermF/ITR delivery vector in Flavobacterium spp., the pMI07 plasmid was digested using SalI and SphI and inserted into the SalI-SphI region of pRR51 to yield pFG001. The plasmid pFG001 was introduced into F. collinsii GiFuPREF103 through conjugation with Escherichia coli S17-1λpir, as previously described (Rhodes et al., 2011a). An erythromycin-resistant transconjugant was obtained by plating cells on MCP agar containing erythromycin. Transformants were screened for non-spreading colony formation. The insertion sites of ermF/ITR in non-spreading colony formation mutants were identified through nested arbitrarily primed polymerase chain reaction (AP-PCR) as previously described (Ichimura et al., 2014). The genomes of mutants were used as templates for AP-PCR. First round of PCR was performed using a random primer (AR8) and mariner-A. Second round PCR was performed using the primer AR2 and mariner-B. The product was subsequently purified and sequenced using the sequence primer mariner-S.

The details of plasmids used herein are shown in Table 1 (Chen et al., 2007; Rhodes et al., 2011a; Ichimura et al., 2014; Imamura et al., 2018).

F. johnsoniae gene deletion mutants were constructed as follows: DNA regions upstream and downstream of a gene were PCR-amplified from the chromosomal DNA of F. johnsoniae. The primers used herein were gene-UF-BamHI plus gene-UR-SalI and gene-DF-SalI plus gene-DR-SphI, respectively, where ‘U’ indicates upstream, ‘F’ indicates forward, ‘D’ indicates downstream, and ‘R’ indicates reverse. The primers used herein are listed in Table S3 . The amplified DNA was cloned into the pGEM-T Easy vector (Promega). The upstream region was digested using BamHI and SalI. The downstream DNA was digested with SalI and SphI. Both the digested products were ligated using pRR51 that was digested with BamHI and SphI. The plasmid was introduced into F. johnsoniae CJ1827 via triparental conjugation and deletion mutants were isolated as previously described (Rhodes et al., 2011a).

For construction of an F. johnsoniae strain expressing Fjoh_0352-green fluorescent protein (GFP) and Fjoh_0353-GFP, the F. johnsoniae chromosomal genes encoding Fjoh_0352 and Fjoh_0353 were PCR-amplified using the primer pair Fjoh0352-coF-BamHI & Fjoh0352-GR-NotI and Fjoh0353-coF-BamHI & Fjoh0353-GR-NotI, respectively. The amplified DNA was digested with BamHI and NotI and inserted into the corresponding region of pNS1, resulting in the plasmid pNS1 containing Fjoh_0352-gfp (pFG002) and Fjoh_0353-gfp (pFG003). Plasmids pFJ29, pFG002, and pFG003 were introduced into F. johnsoniae CJ1827, FLG001, and FLG002, respectively, through triparental conjugation, and GFP-expressing cells were isolated from each strain, as previously reported (Rhodes et al., 2011a).

Cells were examined using microscopy to identify Pep25 and Lbp26 on the cell membrane. F. johnsoniae expressing Δfjoh_0352/fjoh_0352-GFP or Δfjoh_0353/fjoh_0353-GFP (200 μL) were fixed using 1% formaldehyde for 3 min at 25°C on a glass slide. After washing twice with phosphate-buffered saline (PBS), DNA and cell membranes were detected by incubating with 1/500 dilution of 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen, Waltham, MA, USA) and FM4-64 (Invitrogen) 30 min. After incubation, F. johnsoniae cells were subsequently washed twice with PBS and examined under an Olympus IX81 microscope (Olympus, Tokyo, Japan). Images were visualized with a phase contrast objective LUCPlanFLN 100× (Olympus) and captured with a monochrome CoolSNAPHQ digital camera (Photometrics, Tucson, AZ, USA) using MetaMorph software version 6.1 (Molecular Devices, San Jose, CA, USA).

Bacterial cells from a 200-ml culture were collected, suspended in 10 mM Tris-HCl pH 7.5 and vortexed. After centrifugation at 20,000 x g for 10 min at 4°C, the supernatant, containing cell surface materials fraction, was collected. The pellet of cells was resuspended in 20 ml of PBS containing 0.1 mM Na-p-tosyl-L-lysine chloromethyl ketone (TLCK), 0.1 mM leupeptin, 25 µg/ml DNase I and 25 µg/ml RNase A disrupted using a French pressure cell with two passes at 100 MPa. The remaining intact bacterial cells were removed by centrifugation at 2,400 x g for 10 min, and the supernatant was subjected to ultracentrifugation at 100,000 x g for 60 min. The supernatant, containing the cytoplasm and periplasm fraction, was retained. The pellets, containing the membrane fraction, was retained. These sample were subjected to sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE).

The polyclonal antiserum against SprB of F. johnsoniae was prepared by immunizing rabbits (Eve Bioscience, Wakayama, Japan) with a peptide derived from the amino acid sequence C³⁶²⁷NGGSNGTIKVTLGAGNTD³⁶⁴⁵. In the peptide, a cysteine residue was synthesized at the N-terminus of the peptide and conjugated to keyhole limpet hemocyanin (Sigma Genosys, The Woodlands, TX, USA).

Proteins separated on SDS-PAGE (3%–10%, gradient gel [ATTO]) gels were electroblotted onto a PVDF membrane (Sequi-Blot PVDF Membrane, Bio-Rad). The blots were incubated in blocking buffer containing TBS, 0.5% Tween-20, and 5% skim milk (Becton, Dickinson and Company). Subsequently, the blot was incubated with anti-SprB (diluted in blocking buffer) at 4°C overnight. Proteins were probed with the HRP-conjugated anti-rabbit immunoglobulins (Dako) (diluted in blocking buffer), visualized using the Luminata Forte Western HRP substrate (Merck Millipore), and finally detected using enhanced chemiluminescence (GE Healthcare).

Samples were subjected to SDS-PAGE and stained with Coomassie Brilliant Blue. Glycoprotein bands were visualized using Pro-Q Emerald 300 Glycoprotein Gel and Blot Stain Kit per manufacturer’s instruction (Invitrogen, Thermo Fisher Scientific).

To identify cell-surface localized SprB, F. johnsoniae wild-type and mutant cells were examined using immunofluorescence microscopy. Cells were incubated overnight in MCP medium at 25°C. Ten microliters of cells were diluted in 140 μL of MCP and fixed with 1% formaldehyde for 15 min. Cells were washed three times with PBS (200 μL) and blocked with 0.1% BSA in PBS for 30 min. After incubation, cells were re-incubated with of purified anti-SprB (200 μL, 1:200 dilution of 0.1% BSA in PBS) for 90 min. Cells were washed five times with PBS (200 μL) and incubated with secondary antibody conjugated to Alexa 488 (Invitrogen) and DAPI (200 μL, 1:5,000 dilution of 0.1% BSA in PBS). Cells were incubated for 3 h in the dark. After incubation, cells were washed five times with PBS and observed using a BZ-X800 microscope (KEYENCE).

Wild-type and mutant cells were examined for movement using phase contrast microscopy. Cells were cultured overnight in MCP medium at 20°C on a shaker, as described previously. Tunnel slides were prepared using double sided tape, glass microscope slides, and glass coverslips. Cells in growth medium were introduced into the tunnel slide, incubated for 3 min, and motility was observed with a phase contrast microscope. Images were recorded using a CoolSNAPHQ camera (Photometrics) and analyzed using MetaMorph software version 6.1 (Molecular Devices) and ImageJ version 1.53k. Rainbow traces of cell movements were made using ImageJ and the Color FootPrint macro. The average velocity for each strain was calculated from any 100 individuals. Data were analyzed for statistical significance using Student’s unpaired t-test and p-value of <0.05 was used as threshold for significance.

Time-lapse videos were used to observe the colony margins, as previously reported (Sato et al., 2021b). The plate was inverted on the sample stage and observed from underneath using an All-in-One Fluorescence Microscope BZ-X800 (KEYENCE). Images were visualized with a phase contrast objective LUCPlanFLN 20× (Olympus). The phase contrast microscope images were taken every 30 s. The images were analyzed, adjusted, and cropped using a BZ-X800 analyzer software (KEYENCE).

Time-lapse videos were made around the colony spreading of GFP-expressing F. johnsoniae strains using a BX50F microscope (Olympus). The culture plate was placed on a sample stage, and a cover glass was carefully placed over the colony. After the top surface of the dendrites formed was imaged using phase contrast microscopy, fluorescence signals in the same area were observed to track the cells using confocal laser scanning fluorescent microscopy (CLSM) with an objective lens of UPlanFl 100× (Olympus) and ANDOR iXon EMCCD camera (Oxford Instruments, Abingdon, UK) in combination with Andor iQ3 software (Oxford Instruments). Exposure time of each image was typically 100 ms (excitation light: 490–510 nm, emission light: 520–550 nm). Fluorescence images were taken every 3 s for 10 min, and movies were produced.

Bacterial cells were grown overnight in MCP broth at 20°C and then adjusted to an OD₅₉₅ of 0.6. The medium was allowed to stand for 10 min to allow the bacteria to settle. The following formula was used to quantify the auto-aggregating property: 100 – (mean OD₅₉₅ of supernatant/0.6) × 100% (Sorroche et al., 2012). Data were analyzed for statistical significance using Student’s unpaired t-test and p-value of <0.05 was used as threshold for significance. Significant difference tests were performed using a student-t test based on the results of three independent experiments per sample.

Hydrophobicity was assessed as previously described (Nakao et al., 2012). Bacterial cells were grown overnight in MCP broth and standardized at OD₅₉₅ = 1 in PBS. They were then placed into a polyethylene tube, hexadecane was added, and the tube was vigorously vortexed and subsequently incubated for 10 min at RT to allow for phase separation before the OD₅₉₅ of the lower aqueous phase was measured. The percent hydrophobicity was calculated using the following formula: % hydrophobicity = [1 − (OD₅₉₅ after vortexing/OD₅₉₅ before vortexing)] ×100. Significant difference tests were performed using a student-t test based on the results of four independent experiments per sample.

Overnight cultures of F. collinsii GiFuPREF103 strains in MCP broth were centrifuged, washed with PBS, and suspended in PBS at an OD ₅₄₀ = 0.5. The bacterial suspensions were then diluted in a 2-fold series with PBS. A 100-µl aliquot of each suspension was mixed with an equal volume of defibrinated lacked chicken erythrocyte suspension (1% in PBS) and incubated in a round-bottom microtiter plate at room temperature for 3 h.

Three hundred microliters of 100-fold diluted overnight culture was added to a 24-well assay plate and incubated at 20°C for 24 h. Prior to fluorescence microscopy, the supernatant of the biofilm containing suspended cells was removed and 2 mL of fresh medium was added. DAPI was used to visualize cells in the biofilm. For this purpose, cells were stained with DAPI solution in 2 mL of fresh medium, incubated at room temperature for at least 60 min, and then washed twice with 2 mL of fresh medium. Images were acquired at an excitation wavelength of 360 nm and emission wavelength of 460 nm. Fluorescently labeled lectins were used to visualize extracellular polymeric matrix (EPM) in the biofilm. Prior to addition to the biofilm, fluorescein-bound concanavalin A (Sigma-Aldrich), which binds to α-mannopyranosyl and α-glucopyranosyl residues, was added for a final concentration of 50 µg/mL; the excitation and emission wavelengths of fluorescein-conjugated ConA were 494 and 518 nm, respectively. After incubation, the biofilm was washed with fresh medium to remove excess label, and images were taken at an excitation wavelength of 470 nm and emission wavelength of 525 nm. Images were recorded on an All-in-One Fluorescence Microscope BZ-X800 (KEYENCE) and the image data were processed using a BZ-X800 Analyzer software (KEYENCE).

A gel plug containing proteins was subjected to the following procedures: washing with 50% (vol/vol) acetonitrile, washing with 100% acetonitrile, reduction with 10 mM DTT, alkylation with 55 mM iodoacetamide, washing/dehydration with 50% (vol/vol) acetonitrile, and digestion for 10 h with 10 μg/mL trypsin. The resulting peptides were extracted from the gel plug with 0.1% (vol/vol) trifluoroacetic acid/50% (vol/vol) acetonitrile and concentrated using C-18 ZipTips (Merck Millipore). Digests were spotted on a MALDI target using α–cyano-4-hydroxycinnamic acid as a matrix. Spectra were acquired on an autoflex max TOF/TOF system (Bruker). MS/MS spectra were acquired automatically. Proteins were identified using the Mascot search engine (Matrix Science, London, UK).

Outbreaks of BCWD in ayu Plecoglossus altivelis have occurred frequently in rivers or aquaculture farm of Gifu prefecture Japan, since the first detection of F. psychrophilum in 2002. Plecoglossus altivelis with BCWD displayed ulcers on the lower jaw, ulcerative lesions on the body surface, and anemia. Bacterial strains that form yellow colonies on MCP agar medium and have colony-spreading ability were isolated from Plecoglossus altivelis with BCWD. The 16S rRNA sequence analysis revealed that these strains were either Flavobacterium sp. including F. psychrophilum, or Chryseobacterium. We focused on GiFuPREF103, a strain isolated from the gills of Plecoglossus altivelis, that allowed transposon insertion mutants to be created. A genomic homology search using DFAST (https://dfast.ddbj.nig.ac.jp/) revealed that GiFuPREF103 was 98.00% homologous to Flavobacterium collinsii CECT7796. Therefore, strain GiFUPREF103 was designated as F. collinsii GiFuPREF103. F. collinsii GiFuPREF103 was a strain phylogenetically closer to the soil bacterium F. johnsoniae than to F. psychrophilum, known fish disease bacteria ( Figure 1A ). The genome information revealed that not only the genes encoding the T9SS components (porK, porL, porM, porN/O, sprA, sprE, and sprT), which has been studied in F. johnsoniae and P. gingivalis, but also the genes related to gliding motility (gldA, gldB, gldD, gldF, gldG, gldH, gldI, and gldJ) and motility adhesin (sprB) were conserved in F. collinsii GiFuPREF103 ( Figure 1B and Table S4 ). The genes related to protein secretion in P. gingivalis (porU, porV, porQ, and porZ) were also identified in F. collinsii GiFuPREF103. We also identified 38 carboxy-terminal domain (CTD) proteins, which are predicted secreted proteins of T9SS, and 15 porP-like genes including sprF in F. collinsii GiFuPREF103 ( Table S5 ) (McBride, 2019).

Using transposon mutagenesis in F. collinsii GiFuPREF103, we isolated two independent mutants with altered colony morphology on MCP agar medium. An AP-PCR was performed to analyze the insertion site of ermF/ITR in each mutant (Ichimura et al., 2014). Sequencing of the PCR products revealed that the insertion sites were located 286 bp and 1812 bp downstream of the first nucleotide residues of the initiation codons of Flavo103_17390 and Flavo103_17400 in strains FTN26 and FTN25, respectively ( Figure 2A ). The gene cluster containing Flavo103_17390 and Flavo103_17400 was also conserved in F. johnsoniae ( Figure S1 ). Flavo103_17400 showed 85% amino acid sequence similarity with Fjoh_0353 (polysaccharide export protein) of F. johnsoniae UW101. Thus, we named Flavo103_17400 as pep25. Flavo103_17390 showed 70% amino acid sequence similarity with Fjoh_0352 (lipopolysaccharide biosynthesis protein) of F. johnsoniae UW101 and has a Wzz motif (PF02706.18) in the N-terminal region. Thus, we named Flavo103_17390 as lbp26. FTN25 and FTN26 mutants showed less colony spreading ability than wild-type ( Figure 2B ).

To elucidate the roles of Pep25 and Lbp26 proteins, we analyzed the profiles of glycosylated material of wild-type, FTN25, and FTN26 strains. SDS-PAGE profiles of the bacterial surface proteins revealed broad bands with molecular weights ranging from 270 kDa to >770 kDa in wild-type cells, whereas these were absent in the FTN25 and FTN26 strains ( Figure 2C left). We further examined if the materials observed in wild-type cells were glycosylated. The results of Pro-Q Emerald glycan staining suggested that two high-molecular-weight molecules of approximately 250 kDa were glycosylated ( Figure 2C right). Flavo103_21710, type A CTD protein, and Flavo103_03160 were identified by mass spectrometry from cell surface proteins of FTN25 and FTN26, However, bands of >250 kDa macromolecules could not be identified by mass spectrometry ( Figure S2 ).

As a substitute for F. collinsii GiFuPREF103, we used F. johnsoniae, a closely related species capable of genetic recombination, because genetic recombination methods other than transposon-introduced mutagenesis have not been established for F. collinsii GiFuPREF103. To understand the role of Pep25 and Lbp26, F. johnsoniae Δfjoh_0352 and Δfjoh_0353 deletion mutants were constructed. fjoh_0352 and Δfjoh_0353 are orthologs of pep25 and lbp26, respectively. The colony spreading ability of F. johnsoniae ΔFjoh_0352 and ΔFjoh_0353 was reduced compared to that of wild-type strain same as FTN25 and FTN26, respectively ( Figures 2B , S3 ). To observe the localization, Fjoh_0352-GFP and Fjoh_0353-GFP fusion proteins were expressed in F. johnsoniae cells. Fluorescence microscopy observations indicated the presence of green fluorescence at the periphery of cells, suggesting the presence of Fjoh_0352 and Fjoh_0353 on the cell membrane ( Figure S4 ). FM4-64 (red) and DAPI (blue) were used to detect cell membranes and DNA, respectively.

SprB is the first protein identified to be involved in gliding motility of the F. johnsoniae family and is one of the cell-surface adhesin proteins that allow the bacterial cells to adhere and glide on solid surfaces. SprB-deficient strains formed non-spreading colonies on agar. SprB of F. johnsoniae is a large protein with a repeat motif and molecular weight of 669 kDa (6497 amino acid residues). Although the number of motif repeats varies, other gliding Bacteroidetes bacteria also harbor the SprB protein. To investigate the effect of Fjoh_0352 and Fjoh_0353 on SprB expression, bacterial surface proteins of F. johnsoniae were extracted and examined through SDS-PAGE ( Figure S5A ). Immunoblot analysis revealed bands in the lanes of F. johnsoniae wild-type, Δfjoh_0352, and Δfjoh_0353, which were not observed in ΔsprB ( Figure S5B ). Anti-SprB immuno-staining and fluorescence microscopy revealed that SprB was located at the surface of F. johnsoniae wild-type, Δfjoh_0352, and Δfjoh_0353 cells ( Figure S5C ). These results suggest that Pep25 and Lbp26 do not affect SprB expression.

To examine the gliding motility of F. collinsii GiFuPREF103, bacterial cells were introduced into tunnel slides and observed with a phase contrast microscope. FTN25 and FTN26 mutant cells showed gliding in the reverse direction, resulting in reduced net progress compared to the wild-type strain ( Figure 3 , see Movies S1–S3 ). On comparing the average gliding velocities of 100 bacteria, the wild-type strain had a gliding velocity of 1.24 µm/s, while FTN25 and FTN26 had velocities of 0.86 and 0.82 µm/s, respectively. FTN25 and FTN26 also showed an increase in the frequency of inversions ( Figure 3A ). These findings suggest that FTN25 and FTN26 may be partially defective in motility.

FTN25 and FTN26 with partial defects in gliding motility also had diminished colony spreading capacity compared to that in the wild-type ( Figure 2B ). When observing the edge of wild-type F. collinsii GiFuPREF103 colonies under high magnification, small cell clusters were found to break away from the main colony and glide ( Figure 4 , see Movie S4 ). In contrast, in FTN25 and FTN26, such small gliding clusters away from the main wild-type colony were not observed, and the colonies spread gradually by extending dendritic protrusions at their edges ( Figure 4 , see Movies S5, S6 ). To examine this in detail, bacterial cells at the edges of mutant colonies were observed using fluorescence microscopy. For these experiments, F. johnsoniae was used as a surrogate for F. collinsii GiFuPREF103, as a strain expressing a fluorescent protein was required for observation. Observation of the edges of wild-type F. johnsoniae colony revealed clusters of actively gliding bacterial cells ( Movie S7 ). In contrast, at the edges of F. johnsoniae Δfjoh_0352 and Δfjoh_0353 clusters, most bacterial cells were either still or moved back and forth, while several cells kept moving forward but were sometimes interrupted by the immobile cells (see Movies S8, S9 ). At the edge of the spreading colony, the wild-type strains exhibited fast population movement, from densely crowded areas toward less crowded areas, whereas for Δfjoh_0352 and Δfjoh_0353, individual bacterial cells exhibited gliding motility, but no collective behavior was observed.

Hydrophobicity, biophysical characteristics of the cell surface, affect both cell-cell and cell-surface interactions and are involved in biofilm formation in hydrophilic environments. In F. johnsoniae Δfjoh_0352, reports have suggested that the cells are more hydrophobic and have stronger cell–cell interactions and tight connections in the liquid medium (Li et al., 2021). FTN25 and FTN26 showed stronger auto-aggregation than the wild-type ( Figures 5A, B ). Furthermore, the hydrophobicity of the bacterial surface was examined using a hexadecane assay. FTN25 and FTN26 had higher cell surface hydrophobicity than the parent strain ( Figure 5C ). These results suggest that pep25 and lbp26 contribute to the hydrophilicity of the bacterial cell surface.

The hemagglutinating properties of F. collinsii GiFuPREF103 were further investigated. The lpb26 mutant FTN26 showed less hemagglutination compared to the wild-type ( Figure 6 ). The pep25 mutant FTN25 showed almost the same hemagglutination as that of the wild-type strain ( Figure 6 ). These results demonstrate that the lpb26 gene contributed to cell-induced hemagglutination in F. collinsii GiFuPREF103. Similar results were obtained in hemagglutination tests with rabbit erythrocyte.

Fish pathogen F. columnare (Cai et al., 2013) forms mature biofilms containing EPM and waterways when grown on glass slides soaked in liquid medium (Levipan and Avendano-Herrera, 2017). To determine whether F. collinsii GiFuPREF103 wild-type, FTN25, and FTN26 differ in biofilm formation, the strains were cultured for 24 h in glass bottom dishes and the biofilms formed were analyzed using fluorescence microscope. DAPI was used to visualize the cells. The results revealed that FTN25 and FTN26 formed dense, multilayered biofilms. In the wild-type strain, the DAPI signal was relatively uniform compared to that in the mutant strains ( Figure 7 , center panel, DAPI). To visualize the production of EPM in the biofilms, the biofilms were stained with the fluorescence (FITC)-labeled lectin ConA. Strong ConA signals were observed in the biofilms formed by FTN25 and FTN26 ( Figure 7 , left panel, ConA-FITC) and the signals were closely co-localized with the DAPI-stained cells ( Figure 7 , right panel, overlay). The other lectins such as phytohemagglutinin-L, soybean agglutinin, and wheat germ agglutinin did not interact with the wild-type or with mutants, suggesting the presence of glucose and/or mannose residues in the biofilms of the mutants.