Spatial arrangement of several flagellins within bacterial flagella improves motility in different environments

Abstract

Bacterial flagella are helical proteinaceous fibers, composed of the protein flagellin, that confer motility to many bacterial species. The genomes of about half of all flagellated species include more than one flagellin gene, for reasons mostly unknown. Here we show that two flagellins (FlaA and FlaB) are spatially arranged in the polar flagellum of Shewanella putrefaciens, with FlaA being more abundant close to the motor and FlaB in the remainder of the flagellar filament. Observations of swimming trajectories and numerical simulations demonstrate that this segmentation improves motility in a range of environmental conditions, compared to mutants with single-flagellin filaments. In particular, it facilitates screw-like motility, which enhances cellular spreading through obstructed environments. Similar mechanisms may apply to other bacterial species and may explain the maintenance of multiple flagellins to form the flagellar filament.

Fig. 1

Overview of the different types of polar flagellar filaments constructed in this study. In all strains the genes for the lateral flagellins were deleted. a–d Genetic organization of the polar flagellins and genetic modifications to obtain different filament types. In the wild-type context flaA lies upstream of flaB and transcription is controlled by a FliA-independent promoter (cp. Supplementary Fig. 2b). In contrast, transcription of flaB is controlled by a stronger, FliA-dependent promoter and mediates production of the major part of the flagellar filament. Expression of both genes relies on RpoN and FlrA. A functional filament is only produced when at least one of the flagellar genes is expressed from the flaB promoter. If no flagellin is expressed from the flaB promoter only short filament stubs are formed (see Supplementary Fig. 5). Gene deletions are marked with a cross, swapping of the flagellin gene sequences is marked with an asterisk. PflaA/B = flaA/B promoter. e–h Micrographs of cells with fluorescently labeled flagellar filaments displaying the outcome of the genetic editing of the flagellin genes. Panel e also shows the spatial distribution of the flagellins FlaA and FlaB in the wild-type filament (see also Supplementary Fig. 4). The FlaA portion is marked with an arrow. The increased fluorescence of FlaA compared to that of FlaB is likely due to a better accessibility of the cysteine residue during the fluorescent labeling process. Scale bars represent 2 µm. i–l Radial expansion of cells with different types of flagellar filaments in 0.25% soft agar. Wild-type and FlaB-only cells spread well in this structured environment while the spreading ability of FlaA-only and FlaBA cells is strongly impaired. The numbers in the center of the colonies indicate the relative spreading compared to the wild type in percent (% ± s.d.) of three individual experiments

Fig. 2

Schematic model of the sequential production and export of FlaA and FlaB. The figure summarizes our current working model of how spatial arrangement of the two flagellins is achieved in Shewanella putrefaciens, depicted in four subsequent stages of the flagellar assembly. a The current model of flagellin expression suggests that transcription is inhibited as long as the alternative sigma factor 28 (FliA) is blocked by the anti-sigma factor FlgM. In S. putrefaciens, this only applies to FlaB while FlaA production does not depend on FliA (see Supplementary Fig. 2b). Therefore, a pool of FlaA is already present while the basal body is still being assembled. b Once the hook-basal-body complex is completed the export apparatus changes its specificity from hook proteins (FlgE) to the flagellins and other late assembly proteins. At this stage also the anti-sigma factor FlgM gets exported and FliA-dependent transcription of FlaB is initiated. c The already produced FlaA monomers get exported first and assemble at the base of the flagellar filament. d The increasing production of FlaB monomers passes that of FlaA and potentially FlaA transcription and/or translation may be reduced or even completely terminated (indicated by the question marks). Thus, FlaB constitutes the majority residual part of the flagellar filament. (Pol, RNA polymerase; Rib, ribosome; σ, sigma factor)

Fig. 3

Free-swimming behavior mediated by the four different flagellar filament types. Several thousand cell tracks were obtained for each strain by holographic microscopy. The diagrams show the relative counts to account for different sample sizes. a, b Run duration histograms for wild-type (FlaAB, dark grey solid lines), FlaA-only (orange mixed dashed lines), FlaB-only (blue long dashed lines) and FlaBA (light grey short dashed lines) cells in regular medium (upper panel) and medium with increased viscosity (lower panel). P(t) indicates the probability of observing a run with a duration in the range t to t+dt. c, d Velocity distribution for the different filament types in regular medium and medium with increased viscosity, respectively. P(v) indicates the probability of observing a cell swimming with a speed in the range v to v+dv. e, f Turning angle histograms for the different filament types in regular medium and medium with increased viscosity, respectively. Low and high turning angles correspond to weak deviations from straight swimming and near-reversals, respectively. P(θ) indicates the probability that a particular re-orientation event results in a change of direction in the range θ to θ+dθ. g, h Velocity series for a single wild-type cell in regular medium and a similar cell in high viscosity medium, respectively. Under conditions of increased viscosity the cell exhibits longer periods of swimming at a slower speed (the cyan-colored areas in panel h), indicative of screw formation. i, j Stereotypical cell tracks for wild-type cells in medium with increased viscosity (left) and regular medium (right). The tracks correspond to the velocity series in panels g and h. The starting points of the tracks are marked by arrows. The color coding and the line thickness represent the cells’ swimming speed. The corresponding movie is provided as Supplementary Movie 1

Fig. 4

Experimentally observed and simulated screw formation. a The probability of screw formation of wild-type (FlaAB) and FlaB-only cells increases in high viscosity, while for FlaA-only and FlaBA screw formation was never observed. Significance was tested for all filament combinations under both conditions and for each filament between the two conditions. If no screw was observed at all, significance was not tested. All tested combinations were significant (P < 0.05, Bonferroni corrected). Error bars indicate 95% confidence intervals. About 300 cells were counted for each strain. b Observation of screw formation for varying flagellin compositions at different motor torques after a simulation time of t = 60 ms. The simulation was carried out for flagella with an increasing number of FlaA segments, starting with a flagellum completely composed of FlaB (bottom of the diagram) and successively exchanging the segments to a FlaA configuration starting from the filament’s base. Yellow squares indicate regular backward rotation, blue dots indicate screw formation. The color coding represents the z-position of the flagellum’s free end, with negative values indicating a position below the motor segment (position 0). For comparison with the actual flagellin composition in the wild type experimentally measured FlaA proportions are given on the right. Data points are displayed as individual values (gray dots) measured for 50 filaments. The box spans the central half of the data, the black bar indicates the median. A movie of the simulation with FlaA-only and FlaB-only filaments is provided as Supplementary Movie 2. c Force-torque relation extracted from the numerical simulation. The mean force F on the flagellum varies linearly with motor torque M for both forward and backward swimming. Shown are the relations for a FlaB-only flagellum (blue diamonds), a 20% admixture of FlaA (gray stars) and an 80% admixture (orange triangles). The continuous lines are linear fits to the data. The slope βeff in F = βeff M versus the FlaA content is shown in the inset: it increases rapidly up to about 20% FlaA and more slowly for larger fractions. A higher value of βeff indicates a more efficient transfer of torque into driving force