Abstract
The human oral microbiome is an assemblage of diverse bacteria that interact with one another to form communities. Bacteria in a given community are arranged in a three-dimensional matrix with many degrees of freedom. Snapshots of the community display well-defined structures, but the steps required for the construction of these structures are not understood. Here, we show that this construction is carried out, in part, by gliding bacteria that carry non-motile bacteria on their surface as public cargo. Gliding is defined as the motion of cells over a solid or semi-solid surface without the necessity of growth or the aid of pili or flagella. Genomic analysis suggests that gliding bacteria are present in human microbial communities. We focus on Capnocytophaga gingivalis which is present in abundance in the human oral microbiome. By adhering to a mobile cell-surface adhesin, SprB, cells of non-motile bacterial species attach to the surface of C. gingivalis and are propelled as cargo. The cargo cell moves along the length of a C. gingivalis cell, looping from one pole to the other. Multi-color fluorescent spectral imaging of cells of different live but non-motile bacterial species reveals their long-range transport in a swarming polymicrobial community. Some non-motile bacterial species use this public transport more efficiently than others. Tracking of fluorescently-labeled single cells and of gas bubbles carried by fluid flow shows that the swarms are layered, with cells in the upper layers moving more rapidly than those in the lower layers. Thus, cells also glide on top of one another, arranging themselves in three-dimensional space.
Significance We describe an amusing situation in which bacteria of the human microbiome are organized spatially by gliding cells, species of Capnocytophaga, that move backwards and forwards over the substratum. The mobile adhesin that pulls the cells over the substratum also attaches to cells of non-motile bacterial species that are carried up and down the motile cells as cargo. The synchronized public transport of non-motile cargo bacteria in a swarm helps shape a polymicrobial community.
Author Contributions:
A.S., F.E.D., and H.C.B. designed the experiments. A.S.,V.K.P, Y.T., and S.C.Y. performed the experiments. A.S. and H.C.B. wrote the paper
Introduction
Bacteria that attach to and colonize a surface form some of the most stable species of the human microbiome. Rapid availability of nutrition, protection from antibiotics, and long-term colonization are advantages of attaining a specific spatial niche within a microbial community. The mechanisms that drive interactions and guide the architecture of microbial communities are unclear. Active motion via the help of a molecular motor and guidance by a chemotaxis pathway enables some bacteria to sense their environment and find a spatial niche (1, 2). While the chemotaxis ability of a bacterium improves surface colonization (3, 4), motile bacteria do not dominate a microbiome. Non-motile bacteria can reach a spatial niche via diffusion or fluid flow. However, diffusion is not an efficient means of transport for an object of micron size, and fluid flows can be erratic: a sphere of radius 1 µm suspended in water at room temperature takes ∼2.5 hours to diffuse to a distance of 60 µm (5). Another way by which non-motile bacteria can be transported is by hitchhiking a ride on motile microorganisms (6, 7). Only a few examples of hitchhiking are known, and the overall significance of hitchhiking is unclear. Here, we describe a mechanism of hitchhiking where motile but non-flagellated bacteria in the human oral microbiome carry non-motile bacteria as cargo.
Collective motion and swarm behavior are observed in biological objects of all sizes, such as bacteria, insects, birds, and fish (8, 9). The transition of an isolated cell to a swarm can be broadly explained by models based on phase transitions (9). Variations in adhesion forces, the mode of motility, and a bias in the direction of motion govern the swarm behavior of a particular species. Swarms of flagellated bacteria comprise dense groups of cells that swim in a thin layer of liquid close to a surface, without adhering to that surface (10, 11). Bacteria with type 4 pili swarm as the pili attach, are retracted, and detach from the surface, performing twitching motility. Bacteria that use mobile cell-surface adhesins are called gliding bacteria. Bacterial gliding occurs when bacteria move in close contact with an external surface, continuously expending energy (12-14).
Bacteria of the phyla Bacteroidetes are abundant in human gut and oral microbiomes, and many members of this phyla have the ability to navigate surfaces via gliding motility. The Bacteroidetes represent about half of the bacterial population of the human gut microbiome (15, 16) and changes in their concentration correlates with diseases such as obesity, diabetes and colitis (17-19). The Type IX Protein Secretion System (T9SS) is found in many Bacteroidetes and is required for the secretion of proteins such as cell-surface adhesins, chitinase, cellulase, and proteases. T9SS is found in both motile and non-motile Bacteroidetes. The core T9SS proteins are SprA, SprE, SprT, and GldK-N. Bacteria with a cell-surface adhesin SprB, motility proteins GldA-B, GldD, GldF-J, and a functional T9SS are able to move over external surfaces (20).
The human oral microbiome is a catalog of microorganisms from lips, teeth, gingiva, tongue, cheek, palate, and contiguous extensions of the oral cavity (21). Bacteroidetes of the genus Porphyromonas, Prevotella, and Capnocytophaga are found in abundance in the human oral microbiome, and they contain the genes to encode the T9SS. Several reports show the abundance of Capnocytophaga sp. in different sites of the human oral microbiome (22-25). Some Capnocytophaga sp. isolated from the human oral microbiome form spreading colonies (26), and cells lacking the T9SS are deficient in gliding (27). All Capnocytophaga sp. isolated from the human oral microbiome possess the genes to encode the T9SS and its associated gliding machinery. This suggests that the mechanism of gliding found in the phyla Bacteroidetes is prevalent in the genus Capnocytophaga. However, no experimental data exist about the mechanism for the motion of single cells of the genus Capnocytophaga. Motile cells in a community often move as a swarm but nothing is known about the swarm behavior of any gliding Bacteroidetes. Recently, via fluorescent labeling of fixed samples of human oral biofilms, Welch et al. showed the presence of a structured and multigenus consortia within human oral microbial communities that contained bacteria of the genus Capnocytophaga (23). How a non-ordered community reaches an ordered state and arranges itself into spatial structures is not clear. Sequence and imaging data showed that bacteria of the Capnocytophaga genus are found in abundance in both human supragingival and subgingival biofilms. Why the members of a genus that represents gliding bacteria are stable and abundant in the oral microbiome of a healthy human is not known. Via spectral imaging of live cells, we analyze the evolution of a polymicrobial community that contains eight abundant bacterial species of a human subgingival biofilm. We show that swarming Capnocytophaga shape a polymicrobial community by acting as vehicles of public transport and that patterns within a microbial community arise as a result of this active motion. Over longer time scales, a combination of active motion and cell growth can shape the spatial features of a microbial community.
Results & Discussion
T9SS and its associated rotary motor are functional in gliding bacteria of the human microbiome
Some C. gingivalis cells tethered to a glass surface and rotated about fixed axes at speeds about 1.5 Hz. (Fig. 1A-B and Movie S1). Rotation of a tethered F. johnsoniae cell around a fixed axis is attributed to the presence of a rotary motor that is coupled to T9SS (28). Cells that did not tether to glass by one adhesin glide long distances. Some C. gingivalis cells that were able to glide over a glass surface moved at speeds of about 1 µm/s (Fig. 1C and Movie S2). sprA, sprE, sprT, and gldK-N genes encode the proteins that form the core T9SS. gldA-B, gldD, and gldF-J encode proteins that associate with the T9SS to enable gliding motility. The environmental gliding bacterium F. johnsoniae has all 15 genes. T9SS is found in about 62% of the sequenced species of the phyla Bacteroidetes (29). Capnocytophaga species found in the human oral microbiome have all T9SS and gliding motility genes. Of the seven abundant non-motile bacteria used in this study, P. endodontalis and P. oris have T9SS genes but they lack the gliding motility genes (Fig. 1D). These observations combined with the presence of all T9SS-gliding motility genes and mobile cell-surface adhesins demonstrate that the machinery that enables gliding motility of members of the Bacteroidetes phyla is functional in C. gingivalis.
Cargo-transporter relationship exists between non-motile and gliding bacteria of the human microbiome
Genes that encode the bacterial T9SS and a cell surface adhesin SprB are present in Capnocytophaga gingivalis, which is abundant in human oral biofilms. C. gingivalis belongs to the phylum Bacteroidetes and is related to the environmental model gliding bacterium Flavobacterium johnsoniae. Addition of an antibody raised against the SprB of F. johnsoniae to a culture of C. gingivalis caused cell aggregation (Fig. S1). To test the hypothesis that the adhesin SprB of a gliding bacterium could bind to the cell-surface of non-motile bacteria, seven abundant and non-motile bacterial species found in human subgingival biofilms were co-cultured with gliding C. gingivalis. The non-motile bacteria were Porphyromonas endodontalis, Prevotella oris, Parvimonas micra, Actinomyces sp. oral taxon-169, Fusobacterium nucleatum, Streptococcus sanguinis, and Veilonella parvula. Cells of the non-motile bacterial species attached to the cell-surface of C. gingivalis and were transported as cargo along the length of a C. gingivalis cell from one pole to another (Fig. 2A and Movie S3-S9). The adhesin SprB of F. johnsoniae moves in a spiral fashion on the cell-surface (12). The trajectory of the path travelled by a non-motile bacterium on the surface of a C. gingivalis is similar to the trajectory of the adhesin SprB (Fig. 2B). During their natural lifestyle, bacteria of the human microbiome encounter surfaces that are composed of tissue of varying stiffness. Such surfaces are often layered with mucus and are colonized by a polymicrobial community. The adhesin SprB enables binding of a bacterial cell to polysaccharides. Bacterial cell surfaces are abundant in polysaccharides and the SprB adhesin of C. gingivalis might bind to the cell-surface polysaccharides of non-motile bacteria.
Gliding bacteria from the human microbiome display synchronized swarm behavior
C. gingivalis cells displayed swarm behavior over an agar surface. Such swarms moved in a synchronized manner and circular motion of cells was observed near the edge of the swarm. Their motion was tracked with micron-sized gas bubbles with surfactant walls. Such bubbles floated on the top layer of the swarm. (Fig. 3A-B and Movie S10). The gas bubbles near the leading edge of a swarm moved in circular trajectories predominantly in the counter-clockwise direction (Fig. 3C). The speed of bubbles, which is also the speed of the top layer of the swarm, was around 18 µm/min. (Fig. 3D).
Speeds of layers of a swarm are additive
The swarms of C. gingivalis seemed to form layers which were superimposed on top of one other. If layers move in similar directions and cells in the bottom layer move with a speed v, then cells in the second layer would move with a speed 2v. To test this hypothesis, a small percentage of fluorescently labeled C. gingivalis cells were mixed with unlabeled C. gingivalis cells. Fluorescently labeled cells were tracked in a swarm containing both labeled and unlabeled cells (Fig. 4 inset and Movie S11). The speed distribution of cells was bimodal with peaks around both 9 and 18 µm/min (Fig. 4A). Small groups of cells or ‘slugs’ that appeared near the edge of a swarm were tracked and their speed was around 9 µm/min (Fig. 4B-C and Movie S12). This suggests that the speed of a swarming cell and its synchronization correlates with its location in a three-dimensional spatial domain.
Counter-clockwise motion of a swarm can be explained from the right-handed motion of single cells
Single gliding cells move in a manner similar to a right handed screw. If the front end of a flexible gliding cell moving as a right handed screw sticks to an external surface, the cell body bends resulting in the front end of the cell pointing towards the left (Fig 4D and S2). Models that explain group behavior suggest that in a group, the direction of motion of an individual cell depends on the direction of motion of its neighbors and perturbations in the noise of the system result in varying patterns of a swarm (9). Cells that are in the vicinity of a bent cell move along the direction of the bend, and this direction of motion gets propagated in a swarm, thus resulting in its counter-clockwise motion (Fig. 4D).
Long-range cargo transport shapes a microbial community
The spatial localization of bacteria was studied in a polymicrobial community comprising C. gingivalis and seven non-motile bacterial species. The non-motile bacteria used in the study were: Porphyromonas endodontalis, Prevotella oris, Parvimonas micra, Actinomyces sp. oral taxon-169, Fusobacterium nucleatum, Streptococcus sanguinis, and Veilonella parvula. All eight bacteria of this polymicrobial community are found in abundance in a human subgingival biofilm. Single cells of the non-motile bacteria establish a cargo-transporter relationship with single C. gingivalis cells (Fig.1A). To explore if C. gingivalis swarms had the capability to transport the non-motile bacteria over long distances, the cell surfaces of the non-motile bacteria were labeled using seven different Succinimidyl Ester Alexa Flour fluorescent dyes. Physiological features, such as the motility of labeled C. gingivalis remained unaltered, thus providing a powerful tool for live-cell imaging. A polymicrobial community containing seven fluorescently labeled non-motile bacteria and gliding C. gingivalis was imaged, and spectral imaging was used to separate the fluorescent signals. Long-range transport of non-motile bacteria by swarming C. gingivalis was observed in a polymicrobial community (Fig. 5A and Movie S13). Within 20 minutes, the polymicrobial community evolved in such a way that many non-motile bacteria were arranged as islands surrounded by swarms C. gingivalis and some non-motile bacteria (Fig. 5A). The non-motile bacteria were actively transported via the swarms as cargo (Fig. 5B). Polymicrobial aggregates of non-motile bacteria were transported by C. gingivalis. Such aggregates changed their structures while moving with a swarm (Fig. 5C). Tracking of individual bacterial species revealed that some non-motile bacteria moved distances of upto 30 µm in a swarm. Some non-motile bacteria, such as Fusobacterium nucleatum used this mode of transport more efficiently than others (Fig. 6A-E). It is possible that differences in adhesive forces result in preferential transport. F. nucleatum, which is transported the longest distance, has several cell-surface adhesins that might enhance its binding to C. gingivalis. Alternatively, differences in the composition of LPS or cell-surface proteins of the non-motile bacteria could have an impact on their ability to be transported as cargo. Based on our results, we propose a model where a microbial community containing gliding bacteria can use surface motility and cargo transport to find a specific niche. Once a niche is established, the bacteria can arrange in well-defined spatial structures (Fig. 7). Evidently, cargo transport plays a role in forming the spatial features of polymicrobial biofilms that contain Capnocytophaga sp. In the human body, such biofilms are widespread in the human gingival regions (22) and the tongue. It is known that cooperation in the form of mutualism amongst bacterial species occurs via the sharing of public goods (30). We demonstrate that public transport enables bacterial cooperation and shapes communities of bacteria found in the human body.
Methods
Strains and growth
Cells of Capnocytophaga gingivalis ATCC 33624 were used for gliding motility assays. The strain was originally isolated from the periodontal lesion of a human subject (26). C. gingivalis cells were motile when grown on a medium containing trypticase soy broth (TSB) 30 g/L, yeast extract 3 g/L, and 1.5% agar in a CO2-rich anaerobic environment with high relative humidity. This was achieved by placing the bacterial cultures in an AnaeroPack System jar (Mitsubishi Gas Chemical Co.) containing 3 candles and an ignited sheet of Kimtech paper. The cells were incubated at 37°C for 48 hours. Fusobacterium nucleatum, Actinomyces sp. oral taxon-169, Parvimonas micra, Streptococcus sanguinis, and Veilonella parvula were grown on a medium containing tryptic soy agar 20 g/L, brain heart infusion agar 26 g/L, yeast extract 10 g/L, hemin 5 mg/L, 5% sheep blood, and agar 4 g/L at 37°C in an anaerobic chamber. Prevotella oris was grown on a medium containing the above ingredients and 1,4 dihydroxy-2-naptholic acid. Porphyromonas endodontalis was grown on a medium containing TSB 37 g/L, hemin 5 mg/L, NaHCO3 1 g/L, yeast extract 1 g/L, 5% sheep blood, and agar 15 g/L at 37°C in an anaerobic chamber.
Single cell motility assay
C. gingivalis cells suspended in TSY broth were injected in a tunnel slide and were incubated for 5 minutes. Then, 0.25% methyl-cellulose (Methocel 90 HG Sigma-Aldrich catalog number 64680; viscosity of a 2% solution in water at 20°C about 4000 cP) was added to the tunnel slide. After another 5 minute incubation, the gliding cells were imaged using a Nikon Plan 40x BM NA 0.65 objective (Nikon, Melville, NY) and a ThorLabs DCC1545M-GL camera (ThorLabs, Newton, NJ).
Images were analyzed using a custom MATLAB (The MathWorks, Natick, MA) code. To image and analyze tethered and rotating cells, a suspension of C. gingivalis cells was injected into a tunnel slide and washed with TSY broth. Some of the cells tethered to glass and displayed rotation. Images of such cells were captured and analyzed using the method described above.
Generation of microbubbles and tracking of fluid flow patterns
About 70 μL water-insoluble surfactant Span 83 (Sorbitan sesquioleate, S3386; Sigma-Aldrich) (31) was mixed with 100 mL sterile and deionized water in a glass bottle with two inlets. Air was injected into the mixture for 1 hour after which the solution turned milky white in color. To prepare an agar pad, about 700 µl of liquid TSY agar was applied to a glass slide (#48300-026, VMR, 25X75 mm, 1.0 mm thick). A 1 μL sample of Span 83 suspension was placed on top of the agar pad and placed in a high relative-humidity chamber for about 40 minutes. 1 μL of C. gingivalis cells were added at the edge of the surfactant containing area. After incubation for about 20 minutes, most of the water of the 1 µL drop of cell suspension dried out. As the water dried out, the Span 83 droplet exploded and later contracted into micron sized air bubbles with surfactant walls. As a swarm of gliding C. gingivalis approached, the buoyant bubbles moved along with it. Microbubbles formed by the surfactant were used as tracers to image the patterns of motion of the top layer of a swarm. Microbubbles were imaged using the Nikon microscope and ThorLabs camera system described above and their motion was tracked using ImageJ and a custom MATLAB script.
Cargo transport and polymicrobial interactions
In order to study cargo transport by single cells, an equal amount of C. gingivalis was mixed with individual non-motile bacterial species from the human oral microbiome. The mixture of cell pairs was incubated for 10 minutes after which they were injected in a tunnel slide, incubated for 5 minutes, washed with 0.25 % methyl-cellulose (defined above) and imaged using the Nikon microscope and ThorLabs camera system described above. Motion of cargo cells on the transporter cells was tracked using ImageJ and the motion of the transporter cell was tracked using a custom MATLAB script. To obtain the trajectory of a moving cargo cell on a stationary transporter cell, the motion of the transporter cell was subtracted from the motion of the cargo cell using a custom MATLAB script. To test if the adhesin SprB is present on the surface of a C. gingivalis cell, 5 µL of affinity purified antibody raised against the F. johnsonaie SprB was added to 100 µL of C. gingivalis cells suspended in Phosphate buffered saline (PBS). After incubation for 10 minutes at 25°C, the preparation and control with 5 µL PBS were injected into tunnel slides and were imaged using the Nikon microscope and ThorLabs camera system described above.
To image collective motion in a polymicrobial community, the cell-surface of live bacterial cells was stained with an Alexa Fluor NHS Ester (Succinimidyl Ester) dye (32). The following bacteria-dye combinations were used: Porphyromonas endodontalis: Alexa 514, Prevotella oris: Alexa 555, Parvimonas micra: Alexa 568, Actinomyces sp. oral taxon-169: Alexa 594, Fusobacterium nucleatum: Alexa 633, Streptococcus sanguinis: Alexa 647, Veilonella parvula: Alexa 660. Labeling was performed by mixing 1 µL of the dye with 10 µL of the respective bacterial cell suspension, incubated in the dark for 1 hour and washed twice to remove cell-free dye. The seven fluorescently-labeled bacteria were mixed with an equal concentration of C. gingivalis. 1 µL of the eight species polymicrobial mixture was added to an agar pad and incubated for 20 minutes at high relative humidity. After the incubation, swarming was observed. To maintain an anaerobic environment, the agar pad was sealed with a cover glass and paraffin wax. Cells were imaged with a Zeiss LSM 880 Airyscan Confocal microscope in a chamber set at 37°C. Bacteria labeled with the dyes described above were used for calibration, and spectral imaging was used to separate the 7 fluorescent signals. The images were analyzed using ImageJ and superimposition of the fluorescent signal onto a Differential Interference Contrast (DIC) image provided the location of C. gingivalis. The motion of individual bacteria was tracked using ImageJ and was analyzed using a custom MATLAB script.
Imaging of fluorescent cells in a swarm
1 μL of C. gingivalis cells were added to an agar pad and incubated for 20 minutes in a chamber at high relative-humidity. The edge of a swarm had smaller motile groups of cells (slugs) which were imaged using the Nikon microscope and ThorLabs camera system described above. Motion of the slugs was tracked using a custom MATLAB script. The cell-surface of live C. gingivalis cells was labeled with an Alexa Flour NHS Ester (Succinimidyl Ester) dye (ThermoFisher) using a method described above. The labeled cells were mixed with unlabeled cells such that the final amount of labeled cells in the mixture was 1%. The mixture was placed on an agar pad and fluorescently-labeled cells were imaged using a Zeiss LSM 880 Airyscan Confocal microscope. Motion of the cells was tracked using a custom MATLAB script.
Supplementary Movies
Movie S1. C. gingivalis cell tethered to a glass surface and rotating about a fixed axis.
Movie S2. Gliding of single C. gingivalis cells over a glass surface.
Movie S3. P. oris transported as cargo along the length of a single C. gingivalis cell.
Movie S4. F. nucleatum transported as cargo along the length of a single C. gingivalis cell.
Movie S5. P. micra transported as cargo along the length of a single C. gingivalis cell.
Movie S6. S. sanguinis transported as cargo along the length of a single C. gingivalis cell.
Movie S7. Actinomyces sp. 169 transported as cargo along the length of a single C. gingivalis cell.
Movie S8. V. parvula transported as cargo along the length of a single C. gingivalis cell.
Movie S9. P. endodontalis transported as cargo along the length of a single C. gingivalis cell.
Movie S10. A swarm of C. gingivalis moving in a circular fashion with layers on top of one another. Gas bubbles (bright white spheres) move along the top layer of the swarm. A less dense region with small mobile aggregates is seen near the center of the swarm.
Movie S11. Fluorescently-labeled C. gingivalis cells in a swarm containing both labeled and unlabeled C. gingivalis cells.
Movie S12. Motion of slugs found near the edge of a swarm.
Movie S13. A polymicrobial community with seven non-motile fluorescent bacteria being moved by a C. gingivalis (grey) resolved via spectral imaging. See Fig. 5A for a color key.
Movie S14. Time-lapse images of a small region from Movie S13 shows that a polymicrobial aggregate of non-motile bacteria moves long distances in a swarm. For better contrast, C. gingivalis cells are represented in black while non-motile bacteria are represented by the color key shown in Fig. 5.
Acknowledgements
This research was supported by an NIH-NIDCR K99/R00 Pathway to Independence Award DE026826 to A.S. and an NIH-NIAID R01 AI016478 grant to H.C.B. We thank the Harvard Center for Biological Imaging for infrastructure and support. We thank Karen A. Fahrner for comments on the manuscript.
Footnotes
The authors declare no conflict of interest
Author Contributions:
A.S., F.E.D., and H.C.B. designed the experiments. A.S.,V.K.P, Y.T., and S.C.Y. performed the experiments. A.S. and H.C.B. wrote the paper