|Home | About | Journals | Submit | Contact Us | Français|
This paper describes a new approach for labeling intact flagella using the biarsenical dyes FlAsH and ReAsH and imaging their spatial and temporal dynamics on live Escherichia coli cells in swarming communities of bacteria by using epifluorescence microscopy. Using this approach, we observed that (i) bundles of flagella on swarmer cells remain cohesive during frequent collisions with neighboring cells, (ii) flagella on nonmotile swarmer cells at the leading edge of the colony protrude in the direction of the uncolonized agar surface and are actively rotated in a thin layer of fluid that extends outward from the colony, and (iii) flagella form transient interactions with the flagella of other swarmer cells that are in close proximity. This approach opens a window for observing the dynamics of cells in communities that are relevant to ecology, industry, and biomedicine.
Swimming cells of Escherichia coli are propelled through liquids using flagella that are arranged peritrichously (e.g., uniformly distributed). Each flagellum is rotated by a motor at a rate of ~100 Hz using the proton motive force across the cell wall. The balance of torque across the cell results in the counter rotation of the cell body at a frequency of ~10 Hz, which biases the movement of cells suspended in fluids and in close contact with surfaces (6, 19, 24, 33, 37). The biophysical details of the function and dynamics of the flagella of individual E. coli cells suspended in fluids are well understood (6). In contrast, the role and dynamics of these organelles in cells that are in multicellular communities, where the majority of bacteria arguably reside, are just beginning to emerge (8, 9, 16, 36).
Extracellular organelles including flagella, pili, and curli fibers are involved in cell motility and the attachment of cells to surfaces, critical steps in the early formation of multicellular structures (13, 22, 39, 48). In some communities, the dynamic movement of these organelles plays a central role in population-wide behavior. For example, the coordinated movement of individual bacteria in communities, referred to as “swarms,” produces cohesive motion over length scales of hundreds of micrometers and provides a mechanism for the migration of colonies across surfaces (20, 27, 30, 41, 46, 54). Swarming is a phenotype that plays a role in pathogenesis and makes it possible for bacterial colonies to transcend the confines of diffusion-limited growth. Swarming is a mechanism that cells use to replicate, expand rapidly across surfaces, and colonize niches that would be inaccessible to static multicellular structures (3, 23, 47, 52).
Swarms of E. coli cells consist of a heterogeneous population of cells with a morphology that ranges from a mononucleate, vegetative state, in which the cells are 2 to 3 μm long and have ~3 to 7 flagella, to a morphology that is multinucleate, in which the cells are 5 to 20 μm long and the density of flagella is ~2 to 3 flagella more per unit of surface area than vegetative cells (28). The most “differentiated” cells (e.g., those that are the most morphologically distinct from the vegetative state) form an organized monolayer at the migrating edge of the swarming colony that is relatively immobile. The swarmer cells located directly behind the leading edge of the community translate rapidly in small packs, or multicellular rafts, which produce the characteristic vortex-like motion that inspired the name “swarming” (27). This motion extends to the center of the colony, where cells have a morphology that is similar to that of vegetative, swimming cells forming an extended multilayer that may be approximately 100 μm tall.
We are particularly interested in the dynamics of flagella in communities of bacteria and their role in multicellular behavior (8, 51). Several observations suggest that flagella enhance the diffusion of nutrients, growth factors, secondary metabolites, and waste (17, 35) and that the bundling of flagella on adjacent cells may coordinate movement in swarming colonies (34). To better understand the role of flagella in regulating these processes in communities, we are studying the spatial and temporal dynamics of flagella on actively swarming E. coli cells.
Fluorescence microscopy is currently one of the most common techniques used to study the spatial and temporal dynamics of bacterial flagella. Many methods to fluorescently label flagella take advantage of the covalent modification of solvent-accessible thiol groups or primary amines on the side chains of cysteine and lysine residues using dyes conjugated to maleimide or succinimidyl functional groups, respectively (8, 50). Turner et al. demonstrated that Alexa Fluor dyes conjugated to a succinimidyl ester label the flagella and the cell body of E. coli (50). Our experience with these techniques is that the intense fluorescence emitted from the cell body after labeling, which may arise from the covalent modification of surface lipoproteins, masks the fluorescence of the flagella in swarming colonies of cells and makes it difficult to study the dynamics of these organelles in communities. Recently, Blair et al. substituted a cysteine residue for threonine in the FliC protein, the primary constituent of the flagellar filament, of Bacillus subtilis and labeled it specifically with an Alexa Fluor dye conjugated to a maleimide functional group (8), overcoming the issues due to the nonspecific labeling of cells with succinimidyl esters.
In this study we use the biarsenical dyes FlAsH and ReAsH to specifically label the FliC protein in the flagella of swarming strains of E. coli. The small size of these dyes and the corresponding tetracysteine (TC) amino acid motif, which serves as the epitope for binding the reagent, add nominal mass to the protein, making biarsenical dyes a popular alternative to fluorescent proteins (1, 25). We demonstrate that this approach makes it possible to label flagella with fluorophores rapidly and avoid the nonspecific labeling of the cell body. In this paper we describe the application of this labeling technique to investigate the dynamics of flagella on swarming cells of E. coli located in different regions of a swarming colony. This research is beginning to shed light on the dynamics of flagella in dense populations of bacteria that move collectively across surfaces.
E. coli K-12 strain MG1655 (CGSC 8237) was the parental strain for all of the experiments described in this paper. Bacteria were grown in Luria-Bertani (LB) medium (1% [wt/vol] tryptone, 0.5% [wt/vol] yeast extract, 1% [wt/vol] NaCl) at 37°C. LB medium containing 1.5% (wt/vol) Difco agar (LB agar) was used for the growth of colonies of engineered strains of MG1655. Tryptone, yeast extract, peptone, and bacteriological agar were obtained from Becton Dickinson, Sparks, MD. Sodium chloride was obtained from Fisher Scientific, Fairlawn, NJ. Beef extract for motility medium was obtained from Remel (Lenexa, KS). Eiken agar for swarm plates was obtained from Eiken Chemical Co., Tokyo, Japan. Grease for sealing coverslips to slides was obtained from Apiezon Products M&I Materials Ltd. (Manchester, United Kingdom). Petri dishes used in these experiments were obtained from Becton Dickinson. The antibiotics ampicillin (100 μg/ml) (Fisher Scientific) and chloramphenicol (25 μg/ml) (Sigma-Aldrich, Inc., St. Louis, MO) were added to growth medium as necessary.
We synthesized the biarsenical dyes FlAsH-EDT2 and ReAsH-EDT2 on an ~50- to 300-mg scale according to a detailed protocol reported recently (2). Briefly, the biarsenical dyes FlAsH-EDT2 and ReAsH-EDT2 were prepared by making the dimercuric acetate derivatives of fluorescein and resorufin, respectively. We used a transmetallation reaction to install the biarsenical groups and chelated them to 1,2-ethanedithiol.
We created vector pGFliC by TA cloning the fliC gene into pGEM-T Easy (Promega Corporation, Madison, WI). Primers M1 (5′-AGGAGGACAGCTATGGCACAAGTCA-3′) and M2 (5′-TTAACCCTGCAGCAGAGACAGAACCT-3′) were used to amplify the fliC gene open reading frame (ORF) from strain MG1655 by PCR using Taq polymerase (Denville Scientific Inc., Metuchen, NJ). Following TA cloning, the construction of vector pGFliC was confirmed by DNA sequencing.
We used two rounds of site-directed mutagenesis to insert the tetracysteine motif around an existing proline residue at position 285 in the E. coli fliC gene by using primers M9 (5′-CTTCAGGCGGTACATGCTGCCCTGTTCAGATTG-3′) and M10 (5′-CAATCTGAACAGGGCAGCATGTACCGCCTGAAG-3′) for the first step and primers M11 (5′-CGGTACATGCTGCCCTGGATGCTGCGTTCAGATTGATAATAC-3′) and M12 (5′-GTATTATCAATCTGAACGCAGCATCCAGGGCAGCATGTACCG-3′) for the second step. The TC motif was engineered into the fliC gene by using vector pGFliC to yield vector pGD285. Site-directed mutagenesis was performed by using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA).
The fliC-D285 gene from pGD285 was originally subcloned into the high- and low-copy vectors pBAD18 and pBAD30 (26) to express the FliC-D285 protein. These vectors use arabinose-inducible promoters for gene expression and were used to rescue the motility of a ΔfliC E. coli strain. This method for the expression of fliC-D285 did not result in a suitable recovery of motility, with a control vector expressing the wild-type fliC gene (pBAD18-fliC) demonstrating a 40% reduction in motility versus a wild-type strain. Although, in principle, it would be possible to use this strain for our studies, we were concerned that it might lead to artifacts in experiments in which we are studying the dynamics of the flagella. Therefore, the fliC-D285 gene was recombined into the MG1655 chromosome to put the fliC-D285 gene under its native transcriptional regulation.
The fliC-D285 gene was recombined into the MG1655 chromosome according to a methodology described previously by Datsenko and Wanner (18). We engineered the fliC-D285 gene into the MG1655 chromosome without simultaneously inserting DNA scar sequences or extraneous DNA. This procedure required the construction a ΔfliC E. coli strain in the MG1655 background. We created a fliC knockout strain (MFC05) by replacing the fliC gene ORF with chloramphenicol acetyltransferase from vector pKD3 by using primers M205 (5′-TACGTAATCAACGACTTGCAATATAGGATAACGAATCATGCATATGAATATCCTCCTTAG-3′) and M206 (5′-GTTAATCAGGTTACAACGATTAACCCTGCAGCAGAGACAGGTGTAGGCTGGAGCTGCTTC-3′). The MG1655 fliC gene knockout was confirmed by colony PCR and by the loss of cell motility in swim agar (0.25% [wt/vol]).
In preparation for homologous recombination, vector pKD46 was electroporated into MFC05. We selected MFC05-pKD46 transformants on LB-ampicillin agar incubated at 30°C overnight. The fliC-D285 gene was amplified from pGD285 by using primers M207 (5′-CAATACGTAATCAACGACTTGCAATATAGGATAACGAATCATGGCACAAGTCATTAATAC-3′) and M208 (5′-TTGGCGTTGCCGTCAGTCTCAGTTAATCAGGTTACAACGATTAACCCTGCAGCAGAGACA-3′) and Pfu Ultra II fusion Hot Start DNA polymerase. We electroporated the resulting double-stranded DNA (dsDNA) into MFC05-pKD46, diluted the transformed cells into a warm solution of swim agar (0.25%, wt/vol), poured the suspension of cells into 150- by 15-mm petri dishes, gelled the agar, and isolated recombinants from motile colonies. Colonies were picked from the agar with a toothpick, streaked onto nonselective LB medium, and incubated overnight at 37°C. Single colonies from this plate were tested for chloramphenicol and ampicillin sensitivity. We verified the presence of the tetracysteine motif in MFC12 (MG1655 with fliC-D285) by colony PCR using primers specific for the 18-bp DNA sequence that codes for the 6-amino-acid TC motif.
We analyzed swarming cells of MFC12 by using a modified protocol for the labeling and imaging of swimming cells (see the supplemental material). The procedure was designed to reduce the amount of time for which swarmer cells were incubated in liquid before being transferred back onto swarm agar in order to reduce the dedifferentiation of the swarmer cells into the vegetative phenotype (23). We isolated differentiated swarmer cells by extracting a 1-cm-wide by 3-cm-long section of agar from the edge of actively swarming colonies of MFC12 or MG1655 by using a scalpel. The sections of agar were placed into a 100-mm by 15-mm petri dish. Three milliliters of MOPS (morpholinepropanesulfonic acid) buffer was gently pipetted onto the surface of the agar sections to remove the cells. The petri dish was tilted to collect ~500 to 1,000 μl of a suspension of swarmer cells, which was transferred into a tube.
We centrifuged swarmer cells at 2,000 × g for 3 min and resuspended the cell pellet in MOPS buffer containing 10 mM Tris[2-carboxyethyl]phosphine (TCEP) and 5 to 20 μM ReAsH. Cells were incubated in the dark for 10 to 15 min with gentle inversion every 5 min and then centrifuged at 2,000 × g for 3 to 5 min and washed twice with motility buffer. Cells from the final wash were resuspended in a volume of motility buffer identical to the volume initially used to collect cells.
A wild-type swarm colony was grown on swarm agar (as described above) contained within a polydimethylsiloxane (PDMS) chamber (4 cm wide by 5 cm long by 0.5 cm tall) that was bonded to a glass slide (the floor); the top of the chamber was open. To prepare these substrates, we pipetted swarm agar into the PDMS chambers to a height ~1 mm below the top of the polymer and removed excess surface liquid for 20 min at 25°C and then for 3 min in a laminar-flow hood. The swarm agar was inoculated with 2 μl of a suspension of ~9 × 104 E. coli cells/ml (strain MG1655) and the PDMS-agar device was placed into 150- by 15-mm petri dishes and incubated overnight at 30°C at 90 to 95% relative humidity. The next morning we gently pipetted 2 to 3 μl of a suspension of freshly prepared MFC12 swarmer cells labeled with biarsenical dyes onto the swarm agar ~1 to 2 mm from the edge of an advancing colony of wild-type, swarming MG1655 cells. We incubated the cells for 10 to 60 min at 30°C and then carefully placed a clean 24- by 30-mm coverslip onto the agar to avoid trapping air bubbles between the swarming cells/agar and glass coverslip. The device was placed onto the stage of an inverted microscope, and an oil immersion objective lens was gently brought into contact with the glass coverslip. We used the same protocol for imaging swarming cells labeled with biarsenical dyes as described above for swimming cells, with one minor difference. To observe unlabeled MG1655 cells in the swarm and collect data from the fluorescently labeled flagella of MFC12 cells, we simultaneously imaged swarming colonies using both phase-contrast microscopy and epifluorescence microscopy.
To fluorescently label flagella, we used the biarsenical fluorophores FlAsH and ReAsH (excitation wavelength [λex] = 508 nm and emission wavelength [λem] = 530 nm for FlAsH; λex = 597 nm and λem = 608 nm for ReAsH) (1) (Fig. (Fig.1B).1B). We engineered fluorescence into the flagella by inserting the six-amino-acid TC motif CCPGCC (in single-letter amino acid code) into the E. coli FliC protein. Specifically, the CCPGCC motif was incorporated around an existing proline residue at amino acid position 285 (Fig. (Fig.1A);1A); hereafter, we refer to this insertion site as D285, in reference to the D-loop domain of the FliC protein. The insertion site for the TC motif was chosen to maximize the accessibility of the motif to the biarsenical dye and minimize any detrimental effects associated with the proper folding of the FliC protein. We chose the location for the insertion of the epitope based on a homology model of the E. coli FliC protein described in the supplemental material.
We created a motile strain of E. coli expressing a fliC transcript with the TC motif at D285 by homologous recombination into the chromosome according to the methodology described previously by Datsenko and Wanner (18). Successful recombinants were selected by identifying motile colonies on swim agar (0.25% [wt/vol] agar). A strain expressing the gene containing the TC motif at D285 (MFC12) was confirmed by PCR analysis. A complete list of strains, plasmids, and DNA oligonucleotides used in this study can be found in Tables Tables11 and and22.
The expression of the FliC-D285 protein in strain MFC12 was confirmed by Western blots using whole cells and an anti-FliC antibody (Fig. (Fig.2A).2A). We investigated the binding and specificity of the biarsenical dyes FlAsH and ReAsH to FliC-D285 using SDS-PAGE “FlAsH” gels, where the biarsenical fluorophore was mixed into a whole-cell lysate and separated on a polyacrylamide gel (Fig. (Fig.2B).2B). The gel was subsequently scanned on a fluorescence imager, and a band at the appropriate molecular weight for the FliC-D285 protein was observed only for strain MFC12. These results demonstrate that TC-encoded FliC is expressed and specifically binds the biarsenical fluorophores FlAsH and ReAsH.
We determined the effect of the TC motif in FliC on motility by assaying the swimming and swarming capabilities of MFC12 versus the parental strain MG1655. The two classes of motility were analyzed by measuring the radius of swim and swarm colonies using 0.25% (wt/vol) and 0.45% (wt/vol) agar plates over 14.5 and 16 h, respectively. The collective swimming motility of MFC12 was decreased ~12% compared to that of its parent wild-type strain MG1655 (Table (Table3).3). Single-cell velocity measurements for MFC12 demonstrated an ~9% decrease in swimming motility compared to that of MG1655 (Table (Table3).3). These data suggest that the introduction of the TC motif into FliC has an affect on motility. The cause of this discrepancy is still unknown. As these motility experiments were carried out in the absence of a reducing agent, disulfide bond formation within the flagella may play a role in the reduction in motility.
We labeled wild-type and MFC12 cells suspended in motility buffer with biarsenical dyes (Fig. 3A to C and D to F, respectively) and imaged the flagella using epifluorescence microscopy (see Materials and Methods and the supplemental material). The biarsenical dyes specifically labeled the flagella containing the FliC-D285 protein; no fluorescence was detected on the cell bodies. Furthermore, the labeled cells retained their motility following the labeling procedure and attachment of fluorophores to the flagellar filament (see Movies S1 and S2 in the supplemental material). Wild-type MG1655 cells did not display an affinity for the fluorophore, further illustrating the specificity of these dyes for the flagella on the engineered D285 strain.
To image the flagella on swarming cells, we removed swarmers of MFC12 from the edge of a swarming colony on agar and labeled the cells with biarsenical dyes (see Materials and Methods) (Fig. 3G to I). We transplanted a suspension of labeled MFC12 swarmers on agar ~1 to 2 mm from the edge of expanding colonies of swarming MG1655 cells and incubated the substrates; the small amount of excess liquid introduced with the labeled cells was rapidly absorbed by the gel. As the swarming colony expanded across the agar surface, it incorporated labeled MFC12 cells into the swarming community. To observe the dynamics of MFC12 cells, we used a high-magnification oil immersion objective with a high numerical aperture (NA). To accommodate this objective, we carefully placed a glass coverslip on top of a migrating swarming colony. The addition of a coverslip (untreated or coated with bovine serum albumin) did not perturb the collective dynamics and characteristic motion of swarming communities (see Movies S3 and S4 in the supplemental material). For some samples, the addition of a glass coverslip did not change the characteristic dynamics, but it did increase the rate of migration of the leading edge of the swarm (see Table S1 in the supplemental material). The presence of a coverslip extends the approximately two-dimensional layer of liquid in which the cells are suspended beyond the typical swarming colony boundary, which may explain the increased rate of migration of swarming colonies covered with a glass coverslip (see Fig. S1 in the supplemental material).
We imaged the spatial and temporal dynamics of flagella on swarming cells at different locations in the swarming colony using this technique and discovered several previously unobserved phenomena: (i) the bundles of flagella on swarmer E. coli cells are remarkably flexible and bend during cell movement due to frequent collisions with adjacent cells; (ii) the flagella on nonmotile cells at the leading edge of the colony are actuated, although these cells are temporarily “pinned”; and (iii) the flagella on adjacent cells form transient interactions and bundles. We describe these phenomena in more detail below.
The bundling of flagella on cells arises from their simultaneous counterclockwise (CCW) rotation (as viewed from behind the cell) and produces a “run,” while the clockwise (CW) rotation of the motors leads to the breakup of the bundle, referred to as a “tumble” (6). In swarming colonies, in which cell densities may be as high as ~106 cells/mm2, the dynamics of the flagella appear to be remarkably different from those of cell motility in freely suspended fluids. We find that bundles of flagella on swarmer cells frequently stray from a geometry that is parallel to the long axis of the cell body. The bundles may bend at sharp angles as a result of collisions with neighboring, motile, swarming cells (Fig. (Fig.44 and see Movies S5 and S6 in the supplemental material). Figure Figure44 (2.127 s) shows part of a bundle of flagella bending ~90° relative to the other flagella that remain close to the cell body. The bundle remains intact during bending and continues to propel cells clustered together in multicellular rafts.
For swimming cells suspended in bulk fluids, the reversal of the flagellar motors causes a dramatic change in cell motility. Tumbling leads to the loss of the net linear displacement of the cell and the reorientation of the cell body due to Brownian motion. The reorientation of the cell plays a key role in the biased random walk that swimming cells use to navigate along gradients of chemoattractants and chemorepellants (29). Swarming cells also switch between states in which their flagella are bundled and splayed, as a result of CCW and CW flagella motor rotation, respectively, and it was previously reported that the reversal of the direction of motors is required for the rapid colonization of surfaces by swarming colonies of bacteria (40, 53). The relative contribution of these states to community-wide swarming behavior is unclear.
We observed that bundles of flagella on swarmer cells periodically splay apart, leading to the reorientation of the cell. During a reorientation event, swarmer cells may “drift” due to the motion and transfer of momentum from adjacent cells in the community. Swarmer cells spend a considerable amount of their time in the community in this passive state responding to the motion of adjacent cells (see Movies S5 and S6 in the supplemental material). Although the role of motor reversals in swarming motility remains unclear, it is plausible that the coordinated movement of cells at a high density requires them to oscillate between active and passive motion and thus periods of CCW and CW flagellar motor rotation.
To investigate the role of motor reversals in swimming and swarming cells, we analyzed the trajectory of MFC12 cells suspended in nutrient broth and of green fluorescent protein (GFP)-expressing MFC12 cells mixed into wild-type populations of actively swarming cells on nutrient broth agar, respectively. We used a particle-tracking script to measure the track length of each cell, that is, the total distance that the cell moved over a time interval. The displacement of the cell was defined as the shortest distance between its starting and ending points. By calculating the ratio of the track length to the net displacement for both swimmer and swarmer cells, we indirectly measured the degree of reorientation of the cells. A cell with a ratio of ~1 deviates very little from its displacement (e.g., a straight line that connects the start to the end point of a track) and is characteristic of a “run” by a swimming cell. In contrast, a cell with a ratio greater than 1 follows a trajectory that deviates from the straight line between the start and end positions of its track (see Fig. S2 in the supplemental material). In this case, the cell has reoriented or “tumbled” during its displacement, which causes it to deviate from a linear trajectory. The ratio for swarming cells was ~1.25-fold higher than that for swimming cells, suggesting that cell reorientation occurs more frequently when the cells are at a high cell density (e.g., swarming populations) (see Table S2 in the supplemental material). We discuss the significance of these observations in Discussion.
The E. coli cells at the edge of a swarming colony are immobile for reasons that are not entirely clear. One possible explanation is that they are pinned at an air-liquid interface. The absence of techniques available for measuring the volume of the fluid layer on hydrogels makes it difficult to test this hypothesis. Although it is difficult to image the height of the liquid layer on the gel, there is a region of agar that extends ~20 to 100 μm beyond the front edge of a colony that has a different refractive index than the rest of the agar surface and can be observed by phase-contrast microscopy (5) (see Fig. S1 in the supplemental material; see also supplemental Movie S7 at http://www.biochem.wisc.edu/faculty/weibel/lab/gallery/default.aspx). Dworkin and Shapiro were the first to describe the presence of this zone on a swarm plate (21). This optical effect is probably due to the secretion of a surface-active agent by the cells that changes the refractive index of agar in this region and is involved in the separation of a two-dimensional layer of liquid from the gel through which the cells move.
We observed that the flagella on nonmotile swarmer cells at the edge of a swarming colony are actuated and are frequently oriented toward the uncolonized region of the agar surface. The flagella on these cells form bundles and rotate within a thin layer of fluid that presumably extends outward from the advancing swarming colony (see Movies S8 and S9 in the supplemental material). Three cells with labeled flagella are located at the leading edge of a swarming colony in Movie S9. The flagella on the cell at the right in this image are initially rotating rapidly and extend outward (in the 7- to 8-o'clock position) toward the uncolonized agar. As the swarm front progresses, moving from the right of the field of view to the left, one filament in this bundle of flagella changes its orientation to the 11-o'clock position (at 3.830 s) and begins to rotate in the direction of the uncolonized agar. At 4.125 s, a flagellum on the leftmost cell dislodges from the surface and begins to rotate. This movie demonstrates that as the swarm colony migrates outward, the flagella on the nonmotile cells at the leading edge of the colony begin rotating rapidly.
The actuation of the flagella suggests that these organelles are suspended in fluid, but the lack of cell motility may be because the height of the fluid layer is insufficient for the movement of the cell body. The presence of a glass coverslip on top of the swarm, which makes it possible to use a higher-numerical-aperture objective to increase the collection of photons, may extend the layer of fluid out beyond the colony, but it does not make it possible for these leading-edge cells to move outward individually on the surface. Cells at the edge of the colony lose their motility as they are forced onto the agar surface by the transfer of momentum from other cells. These cells are eventually integrated into the colony as it expands past them, and they move rapidly, and in some cases laterally, among their neighbors (see Movie S9 in the supplemental material) before migrating toward the center of the swarm (Fig. (Fig.55 and Movie S8).
Swarming colonies display several characteristics that support the hypothesis that flagellum-flagellum interactions may promote the physical coordination of motility between adjacent cells: (i) the cells are packed together at a high density, which brings them into close proximity; (ii) there are frequent cell-cell interactions; (iii) swarmer cells have many flagella, and in communities, these organelles are in close contact with adjacent cells and even deform around them; and (iv) the flagella are frequently unbundled and rebundled by motor reversals. Jones et al. demonstrated the intercellular bundling of flagella on adjacent swarming cells of Proteus mirabilis in colonies that were chemically fixed and imaged using scanning electron microscopy (34). We wished to test this observation with live swarming colonies of E. coli cells.
We found that flagella on fluorescently labeled cells make transient contact with flagella on adjacent cells, defined by a merged fluorescent signal between the labeled flagella of two cells. Figure Figure66 and Movie S10 in the supplemental material depict an example in which the flagella between two swarming cells form a transient interaction that persists for ~0.3 s before the flagella separate. We detected these interactions infrequently, as we conducted experiments using a very low density of cells with fluorescently labeled flagella that were admixed into wild-type swarming colonies. The interaction between flagella on adjacent cells is an intriguing observation that deserves further attention and quantitative analysis.
E. coli flagella are complex structures that require more than 50 genes for their biosynthesis, assembly, and function (15). The tightly regulated assembly of the flagella, the structure of the FliC protein, and the dimensions of the channel/pore in the filament make the fusion of a fluorescent protein to the N- or C-terminal region of FliC an incompatible approach for engineering fluorescent labels into the flagella (42). In contrast, fluorescent small molecules that bind specifically to peptide epitopes are a useful strategy for labeling macromolecular, homopolymeric structures, such as flagella, as they transcend issues associated with fusions to large fluorescent proteins.
The use of biarsenical fluorophores for the labeling of extracellular organelles on bacterial cells is interesting to us for several reasons: (i) they bind tightly to divicinal TC motifs with dissociation constants of ~10−11 M (1), (ii) the quantum yield increases significantly (~0.5) upon binding to the TC motif (1), (iii) they are synthesized using straightforward chemistry (2), (iv) the label can be removed using millimolar concentrations of 1,2-ethanedithiol (EDT) or 2,3-dimercaptopropanol (1), (v) TC-labeled proteins are more likely to retain their native function during investigations via fluorescent microscopy than fluorescent proteins (31), and (vi) a variety of biarsenical fluorophores that have unique physical and optical properties are available (7, 11, 12, 45, 49). Here we demonstrate the application of the biarsenical dyes FlAsH and ReAsH to specifically label the primary flagellar filament protein of E. coli, FliC (Fig. (Fig.11 to to3).3). This method adds an additional tool to the repertoire of technologies currently available for observing flagella, or other extracellular organelles, on live, motile bacterial cells (8, 38, 50). Importantly, this approach is orthogonal to other labeling strategies (e.g., succinimidyl esters or maleimides conjugated to Alexa Fluor dyes), making biarsenical dyes an attractive option for labeling and observing multiple extracellular structures on bacterial cells simultaneously.
The use of FlAsH-EDT2 and ReAsH-EDT2, and not their membrane-impermeable derivatives (1), to label the E. coli flagellar filament is, to the best of our knowledge, the first example of the use of biarsenical dyes to visualize extracellular structures on bacterial cells. We demonstrate that biarsenical dyes may be used to label proteins in an oxidizing environment provided that the thiol groups on cysteine side chains are reduced prior to binding the fluorescent dye. FlAsH and ReAsH bind the TC motif reversibly and can be removed with ethanedithiol, which makes it possible to exchange the fluorophores and carry out multicolor imaging experiments. The option of reversibly labeling structures on and/or in bacteria with biarsenical dyes may also alleviate issues of photobleaching by exchanging fluorophores in long-term imaging experiments.
This paper describes the first steps toward exploring the role and dynamics of flagella on cells in actively migrating swarming colonies. The study demonstrates several new insights into the dynamics of flagella in swarming colonies, including findings that (i) bundles of flagella on swarmer cells remain cohesive during frequent collisions with neighboring cells that may bend the bundle sharply off axis; (ii) flagella on swarmer cells at the leading edge of the colony protrude in the direction of the uncolonized agar and are actively rotated in a layer of liquid, even though the cell body may be pinned at the surface; and (iii) flagella form transient interactions with the flagella of other swarmer cells that are in close proximity.
The structure and dynamics of flagella on bacteria are, not surprisingly, context dependent. Previous studies of the spatial and temporal dynamics of flagella have focused on the dilute limit of log-phase cell cultures, characterized by a mean cell-cell distance that is approximately 25 μm. In these samples, cells very rarely make physical contact with other cells, and motility involves the intracellular bundling of the flagella along an axis that is parallel to the long axis of the cell body. We have found that in contrast to swimmer cells, the bundles of flagella on swarmers may bend normally to the long axis of the cell body in response to collisions with neighboring swarmer cells. The apparent flexibility of the flagellar bundle makes it possible for swarmer cells to retain a mechanism of self-propulsion in a crowded environment.
The chemotaxis system is responsible for regulating the direction of motor rotation during bacterial cell motility. During swimming motility a combination of thermal noise and the rotational bias of the flagellar motor directs the net migration of cells toward attractants and away from repellents. The role of chemotaxis in swarming is currently unclear. For E. coli and Salmonella enterica serovar Typhimurium there is no evidence that swarmer cells move in response to chemical gradients (10, 27). Instead, the chemotaxis system plays a subtle role that is connected to regulating the direction of rotation of flagellar motors and its puzzling effect on community-wide motility. The change in the direction of rotation of flagellar motors on swarmer cells is important for lubricating the agar surface and the expansion of swarming colonies. However, the connection between surface wetness and the direction of rotation of the motors is not understood (40). Our observation that swarmer cells have a larger track length/displacement ratio than swimming cells suggests that swarmer cells reorient frequently, which may coincide with changes in periods of CCW and CW flagellar motor rotation. Frequent motor switching may play a role in the movement of cells in high-density communities by cycling cells through phases of active and passive motility, which may be important for the coordination of colony-wide motility. It may also provide a strategy for flagella to contact and bundle with flagella on adjacent cells. Our observations corroborate previous evidence that chemotaxis plays a mechanical role in swarming motility but do not explain why the ability to switch between CW and CCW directions of motor rotation is crucial for swarming or how this ability promotes surface wetness (40).
The techniques described in this paper make it possible to study the dynamics of flagella on cells in different regions of a community. For example, at the leading edge of a swarming colony, the cells do not move, but their flagella are actively rotating, implying suspension in a layer of fluid. It is unclear why these cells at the edge of the colony are not motile. The generally accepted hypothesis is that the cells are stuck at the air-liquid interface and are not freely suspended in the approximately two-dimensional layer of liquid within which other cells in the colony move (4, 14, 43). In the experiments described in this paper, we use a glass coverslip as the ceiling of the swarming chamber to accommodate a high-numerical-aperture oil immersion objective for imaging. The coverslip may increase the rate of surface colonization, but it does not change the characteristic motion or behavior of swarming colonies. Interestingly, the coverslip does appear to extend the two-dimensional layer of liquid beyond the swarming colony by creating a thin film (see Fig. S1 in the supplemental material), which we anticipate may remove the constraint of an air-liquid interface directly adjacent to the edge of the colony. Experimental techniques for directly visualizing or measuring the thickness of layers of fluid on the surface of gels with excellent spatial resolution would be useful for testing the structure of fluid in these systems.
The formation of multicellular rafts of cells is commonly observed during the swarming motility of various species of Eubacteria (16, 27, 30). It is possible that flagellum-flagellum interactions may coordinate the formation and/or the maintenance of raft motility. Our results demonstrate that flagella between adjacent swarming bacteria come into close proximity. The diffraction limit of light (~1.6 λ/NA) unfortunately makes it impossible to determine exactly how close the structures are by using optical microscopy. Current superresolution microscopy techniques provide an attractive opportunity for improving these experiments due to their excellent spatial resolution, but they provide only modest temporal resolution (32). The analysis and quantification of these interactions will require innovative approaches for optical microscopy.
The general requirements for swarming motility in the Eubacteria include (i) cell motility using flagella, (ii) cell contact with surfaces, and (iii) physical contact between cells. Specific mechanisms for cell migration across surfaces (e.g., the secretion of surfactants and morphology of swarmer cells) and the classes of surfaces on which cells swarm are dramatically different for different genera of bacteria. This paper focuses on swarming in E. coli K-12 strains, which swarm only on a low concentration of agar (0.45%, wt/vol). It is possible that the dynamics of flagella that we describe in this paper are unique to the swarming of E. coli and other Eubacteria that swarm only on relatively low concentrations of agar, including S. Typhimurium, Serratia marcescens, and Bacillus subtilis. It would be interesting to carry out parallel experiments on organisms that swarm on much higher concentrations of agar, including Vibrio parahaemolyticus and Proteus mirabilis, to determine if the flagellar dynamics that we observed are conserved across the Eubacteria.
Insight into the function and dynamics of flagella on individual swimming cells in bulk fluids has made tremendous progress over the last several decades, but our understanding of these structures in dynamic colonies remains relatively limited. Since most bacteria dwell in communities, the development of tools and techniques for the study of cells, flagella, and other organelles will shed light on the organization, physiology, and behavior of bacteria in multicellular structures. The approach that we describe in this paper opens new windows through which one can observe the dynamics and structure of communities of bacteria. We anticipate that imaging flagella with biarsenical dyes will prove useful for other studies of E. coli motility, including biophysical studies of cell motility, the dynamics of molecular motors, and mechanisms of cell infection by phages.
Funding for this work was provided by DARPA, the USDA (grant WIS01366), a Searle Scholar Award (D.B.W.), and a 3M nontenured faculty award (D.B.W.). M.F.C. is funded by an NIH biotechnology training grant (grant 5T32GM08349). H.H.T. is supported by an NIH National Research Service award (award T32 GM07215).
We thank Nicholas Frankel and Thierry Emonet for assistance with Matlab and particle tracking. We are grateful for the gift of the anti-FliC antibody from Howard Berg.
Published ahead of print on 18 December 2009.
†Supplemental material for this article may be found at http://aem.asm.org/.