Biofilm Spreading by the Adhesin-Dependent Gliding Motility of Flavobacterium johnsoniae. 1. Internal Structure of the Biofilm

The Gram-negative bacterium Flavobacterium johnsoniae employs gliding motility to move rapidly over solid surfaces. Gliding involves the movement of the adhesin SprB along the cell surface. F. johnsoniae spreads on nutrient-poor 1% agar-PY2, forming a thin film-like colony. We used electron microscopy and time-lapse fluorescence microscopy to investigate the structure of colonies formed by wild-type (WT) F. johnsoniae and by the sprB mutant (ΔsprB). In both cases, the bacteria were buried in the extracellular polymeric matrix (EPM) covering the top of the colony. In the spreading WT colonies, the EPM included a thick fiber framework and vesicles, revealing the formation of a biofilm, which is probably required for the spreading movement. Specific paths that were followed by bacterial clusters were observed at the leading edge of colonies, and abundant vesicle secretion and subsequent matrix formation were suggested. EPM-free channels were formed in upward biofilm protrusions, probably for cell migration. In the nonspreading ΔsprB colonies, cells were tightly packed in layers and the intercellular space was occupied by less matrix, indicating immature biofilm. This result suggests that SprB is not necessary for biofilm formation. We conclude that F. johnsoniae cells use gliding motility to spread and maturate biofilms.


Introduction
Flavobacterium johnsoniae is a Gram-negative rod-shaped aerobic bacterium commonly found in soil and fresh water. The cells of this bacterium move rapidly over solid surfaces by gliding motility and thus form thin spreading colonies on agar [1,2]. This characteristic motility is shared by many other members of the Bacteroidetes phylum that lack bacterial motility organelles, such as flagella or pili. Instead, their movements depend on a complex motility-specific apparatus that has been intensively studied in F. johnsoniae, which is a model system [3,4].
SprB with a mass of 669 kDa is a cell surface adhesin component of the motility machinery of F. johnsoniae and forms filaments on the cell surface [5][6][7]. SprB filaments were observed on WT cells using ammonium molybdate staining and transmission electron

WT F. johnsoniae Colonies Spread on Nutrient-Poor Agar Surfaces
To investigate F. johnsoniae colony spreading, we compared the behavior of the wildtype (WT) strain and an sprB deletion mutant strain (∆sprB) [34]. A 1 µL drop of washed F. johnsoniae UW101 WT cells was initially inoculated on nutrient-poor 1% agar PY2 medium (1% A-PY2), and the cells were incubated at room temperature (RT) (23-24 • C) for 5 days. During this period, the WT cells grew and spread radially from all edges of the inoculation spot at the same speed to form a large circular colony (Figure 1a, left). In contrast, the ∆sprB mutant CJ1922 colony did not spread (Figure 1a, right) [5,34]. The radius of the developing WT colony depended on the time after incubation ( Figure S1a).

WT F. johnsoniae Colonies Spread on Nutrient-Poor Agar Surfaces
To investigate F. johnsoniae colony spreading, we compared the behavior of the wild-type (WT) strain and an sprB deletion mutant strain (ΔsprB) [34]. A 1 μL drop of washed F. johnsoniae UW101 WT cells was initially inoculated on nutrient-poor 1% agar PY2 medium (1% A-PY2), and the cells were incubated at room temperature (RT) (23-24 °C) for 5 days. During this period, the WT cells grew and spread radially from all edges of the inoculation spot at the same speed to form a large circular colony (Figure 1a, left). In contrast, the ΔsprB mutant CJ1922 colony did not spread (Figure 1a, right) [5,34]. The radius of the developing WT colony depended on the time after incubation ( Figure S1a).  Figure S2).

Movement of Single Bacterial Cells in the WT Colony Spread on 1% A-PY2
To visualize the movement of single bacterial cells at the colony edge, a 1:100 mixture of WT cells with and without cytoplasmic GFP expression was inoculated in the center of an agar plate, incubated for 24 h, and visualized by time-lapse fluorescence microscopy ( Figure S2a) [16,35]. This video (at 300× the speed of standard time-lapse fluorescence microscopy imaging) is shown in Figure S2a. On 1% A-PY2, the colony spread out from the inoculated spot. At the leading edge of the spreading colony, a small cell cluster at the tip of a branch moved outwards and was followed by other cells (Figure 1b, upper panel, Figure S2a). This suggested cell-to-cell connections. Furthermore, cells often moved along  Figure S2).

Movement of Single Bacterial Cells in the WT Colony Spread on 1% A-PY2
To visualize the movement of single bacterial cells at the colony edge, a 1:100 mixture of WT cells with and without cytoplasmic GFP expression was inoculated in the center of an agar plate, incubated for 24 h, and visualized by time-lapse fluorescence microscopy ( Figure S2a) [16,35]. This video (at 300× the speed of standard time-lapse fluorescence microscopy imaging) is shown in Figure S2a. On 1% A-PY2, the colony spread out from the inoculated spot. At the leading edge of the spreading colony, a small cell cluster at the tip of a branch moved outwards and was followed by other cells (Figure 1b, upper panel, Figure S2a). This suggested cell-to-cell connections. Furthermore, cells often moved along the path on the agar surface used by the preceding cell clusters ( Figure S2a), suggesting the existence of a track formed by the leading cells controlling the movement of the following cell clusters.

Static Single Bacterial Cells in the ∆sprB Mutant Colony on 1% A-PY2
The same experiment was performed for sprB deletion mutant cells. The ∆sprB colony did not spread, and curved lines of cells producing cytoplasmic GFP were apparent within the colony (Figure 1a,b and Figure S2b). The video of ∆sprB colony (at 300× the speed of standard time-lapse fluorescence microscopy imaging) is shown in Figure S2b. Because F. johnsoniae cells divide along a single axis, it is reasonable to propose that the cells in the line all originated from the same cell. The movement of individual cells in the colony was not detectable.

WT F. johnsoniae Colonies Form a Biofilm on Nutrient-Poor Agar Surfaces
To investigate the structure of a spreading colony more precisely, the WT colony was aldehyde-fixed, embedded in Epon, and vertically sectioned parallel to the direction of spreading ( Figure S1b). The approximately 70 nm thick, 1 mm wide, and 0.5 mm long sections were stained with heavy metals and inspected by TEM. At the bottom of the WT colony spreading on 1% A-PY2, i.e., close to the agar layer, cells were loosely packed and sparsely embedded in a low electron density matrix ( Figure 2). Bacterial cells were positioned in different directions. Above the bottom layer, there were fewer cells. the path on the agar surface used by the preceding cell clusters ( Figure S2a), suggesting the existence of a track formed by the leading cells controlling the movement of the following cell clusters.

Static Single Bacterial Cells in the ΔsprB Mutant Colony on 1% A-PY2
The same experiment was performed for sprB deletion mutant cells. The ΔsprB colony did not spread, and curved lines of cells producing cytoplasmic GFP were apparent within the colony (Figure 1a,b and Figure S2b). The video of ΔsprB colony (at 300× the speed of standard time-lapse fluorescence microscopy imaging) is shown in Figure S2b. Because F. johnsoniae cells divide along a single axis, it is reasonable to propose that the cells in the line all originated from the same cell. The movement of individual cells in the colony was not detectable.

WT F. johnsoniae Colonies Form a Biofilm on Nutrient-Poor Agar Surfaces
To investigate the structure of a spreading colony more precisely, the WT colony was aldehyde-fixed, embedded in Epon, and vertically sectioned parallel to the direction of spreading ( Figure S1b). The approximately 70 nm thick, 1 mm wide, and 0.5 mm long sections were stained with heavy metals and inspected by TEM. At the bottom of the WT colony spreading on 1% A-PY2, i.e., close to the agar layer, cells were loosely packed and sparsely embedded in a low electron density matrix ( Figure 2). Bacterial cells were positioned in different directions. Above the bottom layer, there were fewer cells.

Bacteria at the Advancing Front of WT Colonies Secreted Many Vesicles
To observe the edge structure of WT colonies, the concentric WT colony formed on 1% A-PY2 was divided into six equal regions at the colony margin (<0.5 mm from the edge) (Figure 3a,b). Epon embedding and TEM revealed that the layered coat on the surface of cells found in all regions showed irregular undulations (Figure 3c). The intercellular space was occupied by a matrix containing thin extracellular fibers (3-7 nm in diameter) interspersed with secreted vesicles (Figure 3c). The bacteria advancing the furthest in the colonies were more densely surrounded by budding vesicles and many secreted vesicles (30-50 nm in diameter) (Figure 3c, image 1-2 and 1′). In the inner regions of the colony, each bacterial cell was surrounded by only a few smaller vesicles (Figures 3c, image

Bacteria at the Advancing Front of WT Colonies Secreted Many Vesicles
To observe the edge structure of WT colonies, the concentric WT colony formed on 1% A-PY2 was divided into six equal regions at the colony margin (<0.5 mm from the edge) (Figure 3a

WT Cells form a Queue at the Advancing Vertical Top of the Colony
Upon detailed observation of the advancing top of the WT cell layer, we found strings of cells protruding upwards away from the main colony, presumably due to the pile-up induced by the increase in cell population ( Figure 4). As shown in Figure 4a, the strings could be classified into four distinct regions, I to IV, that extend perpendicular to the culture substrate. No bacterial cells were found in the region farthest from the agar surface (region I), but fiber-like structures were rather uniformly distributed throughout the area (Figure 4bI). We suggest that these fibers were secreted from the leading edge of the cell population. Region II contained bacteria that were presumably close to the top of the colony. A thin channel without any thin extracellular fibers was observed in the bottom half of this region (Figure 4a). The bacteria in region II were densely surrounded by budding vesicles and many isolated vesicles (Figure 4bII,c). In contrast, each bacterial cell was surrounded by only a few relatively smaller vesicles in region III and by even fewer vesicles in region IV, the least distal upper region of the colony, closest to the bottom layer (Figure 4a,b). Thin extracellular fibers were viewed from top to bottom from region I through region IV and were fully distributed from the front of the cell population. The filamentous background in region II, which lies under region I, was traversed from top to bottom by a filament-less channel that contained many cells (Figure 4a). More vesicles were attached to the cells in region II than those in region III and IV (Figure 4c). of this region (Figure 4a). The bacteria in region II were densely surrounded by budding vesicles and many isolated vesicles (Figure 4bII,c). In contrast, each bacterial cell was surrounded by only a few relatively smaller vesicles in region III and by even fewer vesicles in region IV, the least distal upper region of the colony, closest to the bottom layer ( Figures  4a,b). Thin extracellular fibers were viewed from top to bottom from region I through region IV and were fully distributed from the front of the cell population. The filamentous background in region II, which lies under region I, was traversed from top to bottom by a filament-less channel that contained many cells (Figure 4a). More vesicles were attached to the cells in region II than those in region III and IV (Figure 4c).

Biofilm Maturation Depends on the Cell-Surface Adhesin SprB
Next, we investigated whether the motility adhesin SprB is required for biofilm formation in F. johnsoniae. To this end, we evaluated the spreading phenotype of an sprB deletion mutant on 1% A-PY2. Deletion of sprB prevented F. johnsoniae colonies from spreading on 1% A-PY2; the cells only grew within the small inoculation circle (Figure 1a, right). This growth pattern was in stark contrast to the behavior of the WT strain, which spread radially from all edges of the inoculation spot on the agar surface (Figure 1a, left). To investigate the role of SprB in detail, as for the WT, a ∆sprB colony was embedded in Epon, sliced, and observed by TEM ( Figure 5). Unlike the WT cells, the mutant cells were crowded and formed three distinct layers on the 1% A-PY2 medium ( Figure 5). Higher magnification images revealed that the ∆sprB cells had a smooth rod-like shape (Figure 6a-h).
In the base layer, cells were tightly packed on the agar surface (Figure 6b-d,g-h) and grouped in preferred orientations due to the high cell density; variations in the length of the similarly shaped cells within a group were relatively small (Figure 6b-d,g-h). In the middle of the base layer, the intercellular space was occupied by many large and small vesicles, extracellular filaments and a few secreted bacterial cytoskeletal-like structures, but the separation between cells was clearly small (Figure 6c,g). In some places, lysed cells that maintained gross cell shape of membranes were observed (decreased contrast in cytoplasm) (Figure 6c,g). These suggests that the biofilm formed was immature.  (Figure 4b).
Symbols on the graph represent the average density, and the error bars correspond to the SD.

Biofilm Maturation Depends on the Cell-Surface Adhesin SprB
Next, we investigated whether the motility adhesin SprB is required for biofilm formation in F. johnsoniae. To this end, we evaluated the spreading phenotype of an sprB deletion mutant on 1% A-PY2. Deletion of sprB prevented F. johnsoniae colonies from spreading on 1% A-PY2; the cells only grew within the small inoculation circle (Figure 1a, right). This growth pattern was in stark contrast to the behavior of the WT strain, which spread radially from all edges of the inoculation spot on the agar surface (Figure 1a, left). To investigate the role of SprB in detail, as for the WT, a ΔsprB colony was embedded in Epon, sliced, and observed by TEM ( Figure 5). Unlike the WT cells, the mutant cells were crowded and formed three distinct layers on the 1% A-PY2 medium ( Figure 5). Higher magnification images revealed that the ΔsprB cells had a smooth rod-like shape (Figure 6a-h). In the base layer, cells were tightly packed on the agar surface (Figure 6b-d,g-h) and grouped in preferred orientations due to the high cell density; variations in the length of the similarly shaped cells within a group were relatively small (Figure 6b-d,g-h). In the middle of the base layer, the intercellular space was occupied by many large and small vesicles, extracellular filaments and a few secreted bacterial cytoskeletal-like structures, but the separation between cells was clearly small (Figure 6c,g). In some places, lysed cells that maintained gross cell shape of membranes were observed (decreased contrast in cytoplasm) (Figure 6c,g). These suggests that the biofilm formed was immature. A large cluster of cells was observed above the base layer, the region we refer to as the 2nd layer ( Figures 5 and 6a,b,e,f). In contrast to the tightly packed base layer, curved and straight rod-shaped cells of various lengths were adjacent to one another, resulting in less dense communities and loss of directionality ( Figure 5, 2nd layer, Figure 6e,f). Some cells spread out from the cluster, forming a less populated 3rd layer further towards the top ( Figure 5, 3rd layer, Figure 6a,e). Similar to the WT cells, the space between the ΔsprB cells in the 2nd and 3rd layers was occupied by a substance containing thin extracellular fibers and vesicles (Figure 6a,b,e,f). Within the 2nd layer, each ΔsprB cell possessed a thick surface coat structure (Figure 7, lower), similar to WT (Figure 7, upper). A large cluster of cells was observed above the base layer, the region we refer to as the 2nd layer ( Figures 5 and 6a,b,e,f). In contrast to the tightly packed base layer, curved and straight rod-shaped cells of various lengths were adjacent to one another, resulting in less dense communities and loss of directionality ( Figure 5, 2nd layer, Figure 6e,f). Some cells spread out from the cluster, forming a less populated 3rd layer further towards the top ( Figure 5, 3rd layer, Figure 6a,e). Similar to the WT cells, the space between the ∆sprB cells in the 2nd and 3rd layers was occupied by a substance containing thin extracellular fibers and vesicles (Figure 6a,b,e,f). Within the 2nd layer, each ∆sprB cell possessed a thick surface coat structure (Figure 7, lower), similar to WT (Figure 7, upper).  Figure 1a. Each cell in the bacterial layer on the agar surface had a thick layered coat structure (arrows). The coat structure had regular undulations along the long and short axes, and was often in contact with the coat of the neighboring cells (arrowheads).

Surface Connecting Structure of F. johnsoniae Cells in Biofilm
Within the biofilm, each WT cell was oriented in various directions and contained a thick layered surface coat structure (Figure 7). The layered coat structure had regular undulations along the long and short axes of the cells and contacted the coat of neighboring cells (Figure 7). The coat structure was not observed when the bacterial cells isolated from the colony grown on 1% A-PY2 were washed with washing buffer (10 mM Tris-HCl pH 7.5) prior to aldehyde fixation (Figure 8). The remaining outer membrane surrounding the WT cells was undulated (Figure 8, left), as observed in Figure 7. Such undulation of outer membrane was also observed for the cells from ΔsprB colony (Figure 8a, right).  Figure 1a. Each cell in the bacterial layer on the agar surface had a thick layered coat structure (arrows). The coat structure had regular undulations along the long and short axes, and was often in contact with the coat of the neighboring cells (arrowheads).

Surface Connecting Structure of F. johnsoniae Cells in Biofilm
Within the biofilm, each WT cell was oriented in various directions and contained a thick layered surface coat structure (Figure 7). The layered coat structure had regular undulations along the long and short axes of the cells and contacted the coat of neighboring cells (Figure 7). The coat structure was not observed when the bacterial cells isolated from the colony grown on 1% A-PY2 were washed with washing buffer (10 mM Tris-HCl pH 7.5) prior to aldehyde fixation (Figure 8). The remaining outer membrane surrounding the WT cells was undulated (Figure 8, left), as observed in Figure 7. Such undulation of outer membrane was also observed for the cells from ∆sprB colony (Figure 8a, right).

Discussion
The attachment of colonies to a solid surface is a complex process mediated by cellcell interactions. For F. johnsoniae, this process was affected by the cell surface components and environmental factors, such as the nutrient supply and moisture (Figures 1-5) [16]. Biofilms of many bacteria, such as P. aeruginosa and E. coli, are surface-attached microbial communities composed of cells embedded in an extracellular polymeric matrix (EPM) [36,37]. These matrices are composed of polysaccharides, extracellular DNA, and protein structures such as curli, fimbriae and pili [38,39,40,41]. The biofilm formed by P. aeruginosa, which exhibits flagellum-mediated swimming motility and surface-associated swarming and twitching motilities, is established in five main stages: (i) attachment, involving adhesion of bacteria to the substrate; (ii) cell-cell adhesion; (iii) early development of biofilm architecture; (iv) maturation of biofilm architecture; and (v) dispersion of single dissociated cells from the biofilm [42,43,44].
As reported earlier [2], WT F. johnsoniae cells grew and formed a characteristic large circular colony on 1% A-PY2 (Figure 1a). It is not clear how the cells were distributed and connected three-dimensionally in the spreading colony. Here, we studied the precise structure of the SprB-dependent spreading and the ΔsprB nonspreading colonies using electron microscopy (EM) (Figures 2-6). This showed that cells were embedded in a selfproduced EPM, forming a biofilm, while they divided and glided on the agar medium. Consequently, the biofilm of the colony spread from the central region outwards.
TEM of colony sections of WT F. johnsoniae revealed that the space between the cells was occupied by an EPM containing fibers and vesicles (Figures 3 and 4). These fibers, 3-7 nm in diameter, were somewhat thicker than those observed in biofilms formed by Staphylococcus aureus [45] and P. acnes [46]. This might reflect the tough structure required for the mobile nature of the F. johnsoniae biofilm, in contrast to the static biofilms

Discussion
The attachment of colonies to a solid surface is a complex process mediated by cell-cell interactions. For F. johnsoniae, this process was affected by the cell surface components and environmental factors, such as the nutrient supply and moisture (Figures 1-5) [16]. Biofilms of many bacteria, such as P. aeruginosa and E. coli, are surface-attached microbial communities composed of cells embedded in an extracellular polymeric matrix (EPM) [36,37]. These matrices are composed of polysaccharides, extracellular DNA and protein structures such as curli, fimbriae and pili [38][39][40][41]. The biofilm formed by P. aeruginosa, which exhibits flagellum-mediated swimming motility and surface-associated swarming and twitching motilities, is established in five main stages: (i) attachment, involving adhesion of bacteria to the substrate; (ii) cell-cell adhesion; (iii) early development of biofilm architecture; (iv) maturation of biofilm architecture; and (v) dispersion of single dissociated cells from the biofilm [42][43][44].
As reported earlier [2], WT F. johnsoniae cells grew and formed a characteristic large circular colony on 1% A-PY2 (Figure 1a). It is not clear how the cells were distributed and connected three-dimensionally in the spreading colony. Here, we studied the precise structure of the SprB-dependent spreading and the ∆sprB nonspreading colonies using electron microscopy (EM) (Figures 2-6). This showed that cells were embedded in a selfproduced EPM, forming a biofilm, while they divided and glided on the agar medium. Consequently, the biofilm of the colony spread from the central region outwards.
TEM of colony sections of WT F. johnsoniae revealed that the space between the cells was occupied by an EPM containing fibers and vesicles (Figures 3 and 4). These fibers, 3-7 nm in diameter, were somewhat thicker than those observed in biofilms formed by Staphylococcus aureus [45] and P. acnes [46]. This might reflect the tough structure required for the mobile nature of the F. johnsoniae biofilm, in contrast to the static biofilms of S. aureus and P. acnes. In the WT F. johnsoniae colony spreading on 1% A-PY2, the internal space was sparsely populated by cells. At its marginal top, extracellular fibers were distributed from the leading edge of the cell population (Figure 3b). In the matrix, a cell trajectory was clearly identified by the presence of a filament-less channel, through which following cells could move (Figure 4a, upper left). These observations suggest that secretion of EPM preceded the migration of F. johnsoniae cells. In total, colony expansion seems to reflect the morphological, behavioral and physiological characteristics of the cells in the biofilm colony ( Figure 4).
Attachment to a solid surface is an important first step in biofilm formation [47][48][49]. Because gliding motility is a movement that allows bacteria to stay in contact with the solid surface, it appears that F. johnsoniae cells, partly with the assistance of adhesins such as SprB, can glide on the agar surface covered with secreted EPM while secreting vesicles, thus extending the biofilm. The bacteria in the advancing fronts toward horizontal (Figure 3a, region 1-2) or upward direction (Figure 4a, region II) were more densely surrounded by budding vesicles and many secreted vesicles in the colony (Figure 3d, image 1-2 and Figure 4c, region II). The abundant vesicle secretions might support EPM formation because vesicles could include or attach to proteins, including chaperons, contributing to EPM formation.
WT cells grew and formed a large circular colony on 1% A-PY2, whereas the ∆sprB mutant formed nonspreading colonies. Although the ∆sprB cells were densely packed, the intercellular space was occupied by a substance containing thin extracellular fibers and vesicles (Figures 5 and 6). Consequently, the cell surface adhesin SprB has a role in the expansion of F. johnsoniae biofilm, although SprB is not required for biofilm formation. This is in good accordance with a previous study showing that the P. aeruginosa biofilm architecture formed by a flagella-and type IV-pili-double mutant was different from that formed by WT [49]; flagella-and twitching-motility have roles in shaping the biofilm architecture, although they are not necessary for biofilm formation.
The volume of extracellular matrix of the ∆sprB mutant colony was significantly less than that of the WT colony ( Figure 6). Our data suggest that gliding movement is associated with the volume of extracellular matrix and thus affects the maturation of biofilm architecture. Recently, we investigated the SprB-independent colony spreading of F. johnsoniae on soft agar containing glucose [16]. The extracellular matrix of the colony contained a network of extracellular fibers and many secreted vesicles, which is similar with that on 1% A-PY2 [16,50].
Death of cells was observed in the tightly packed region of the F. johnsoniae ∆sprB mutant colony that grew on 1% A-PY2 ( Figures 5 and 6, the base layers). Such death might reflect social apoptosis, including activities forming cell-free channels, to allow external nutrition to diffuse into the biofilm. Consequently, the thick fiber-containing EPM might include double-stranded DNA filaments and extracellular polysaccharides, as observed in the biofilms of S. aureus and P. acnes immersed in aqueous liquid by atmospheric scanning electron microscopy (ASEM) [45,46]. Further experiments are required to understand the water-rich architecture of F. johnsoniae biofilms, for example, using ASEM that enabled observations of various water-rich phenomena of organic and inorganic substances at high resolution [16,51,52]. The undulations observed for the layered surface coats of F. johnsoniae cells in biofilm has a possibility to be artefacts, due to the dehydration pretreatment of the Epon-embedded TEM. It should be also addressed using the liquid-phase ASEM.
The results reported here shed light into the liquid-phase biofilm structures formed at the interfaces between air and water, and also between water and solid substrate. The various electron microscopy and optical microscopy techniques used in this studies will accelerate study of such biofilms.

Bacterial Strain and Biofilm Cultivation
F. johnsoniae strains were grown in casitone-yeast extract (CYE) medium at 24 • C (Becton, Dickinson and Co., Franklin Lakes, NJ, USA). The details of the bacterial strains and plasmids used are shown in Table 1. For selection and maintenance of antibioticresistant F. johnsoniae strains, antibiotics were added to the medium at the following concentrations: streptomycin, 100 µg/mL; erythromycin, 100 µg/mL. Ap r Em r , Expression vector carrying with ompA promoter and gfp [35] To observe colony spreading, F. johnsoniae WT and sprB deletion mutant CJ1922 (∆sprB) cells were grown in CYE medium at 27 • C with shaking (175 rpm) overnight. The cells were collected as a pellet by centrifugation at 800× g for 10 min, at 22 • C. The pellet was resuspended in the same volume of washing buffer (10 mM Tris-HCl pH 7.4) by vortexing, and the suspension was centrifuged at 800× g for 10 min at 22 • C. These steps were repeated twice. The cells were spotted onto peptone yeast (PY2) agar medium (peptone and yeast extract, Becton, Dickinson and Co. and agar, Ina Food Industry Co., Ltd., Nagano, Japan) in a dish 9 cm in diameter and incubated at 24 • C [15]. Construction of an F. johnsoniae strain expressing GFP was carried out as follows: After the mating of E. coli S17-1 λpir carrying pFj29 with F. johnsoniae WT (CJ1827) and CJ1922, an Em r transconjugant was obtained [34,53].

Fixation
For Epon embedding and TEM, spreading colonies on agar medium and cultured bacterial cells were fixed with 1% paraformaldehyde and 3.5% glutaraldehyde in 0.1 M phosphate buffer (PB) (pH 7.4) at room temperature (RT) for 3 h and further with 1% osmic acid in the same buffer at 4 • C for 1 h.

Epon Embedding and Sectioning
Fixed colonies were dehydrated through a gradient series of alcohol at RT and embedded in Epon 812. Ultrathin sections (70 or 400 nm thick) were cut parallel to the colony spreading direction and perpendicular to the agar medium surface. This allowed both spreading across the surface of the agar medium and any penetration into the agar to be monitored. A Leica Ultracut UCT microtome (Leica, Wetzlar, Germany) was employed. A series of ultrathin sections was cut at RT and collected on EM grids.

TEM Imaging
Epon sections mounted on grids were stained with uranyl acetate (UA) and lead citrate (LC) and observed with an H7600 TEM (Hitachi, Tokyo, Japan) at 80 kV.

Quantification of Vesicles
The number of vesicles attached to the outer circumference of bacteria was manually counted and divided by the outer circumference to get vesicle density (vesicles/µm). The circumference of the cells was measured using ImageJ (National institute of Health). Thirty-three cells were analyzed for each region.