Swimming Cells with Secondary Filaments Spread Faster than Cells with Polar Flagella only.

Our previous experiments strongly indicated that CN-32 cells possessing a secondary flagellar system cover significantly larger distances in soft agar plates than cells with polar flagella only (15). To further elucidate a potential beneficial role of additional flagellar filaments in efficient spreading, we analyzed expression and production of the secondary flagellar systems within the population. To this end, we used a strain in which the flagellar motor protein FliM₂, a component only occurring in flagellar basal bodies of the lateral secondary system, was functionally fused to sfGFP. Cells producing the FliM₂-sfGFP fusion protein were placed on soft-agar plates, and after formation of a visible halo caused by radial expansion of the bacterial population, samples were taken from the lateral extension zone with increasing distance to the center (Fig. 1). Expression and localization of FliM₂-sfGFP was determined by fluorescence microscopy. Close to the center of the lateral extension zone, about 50% of sampled cells displayed green fluorescent foci at various lateral positions within the cell. This portion of FliM₂-sfGFP-producing cells was similar to that observed in planktonic cultures. In contrast, all cells isolated from the fringes of the swimming halos were found to produce FliM₂-sfGFP with three fluorescent foci on average at various lateral positions. Flagellar staining revealed that these cells possessed one or sometimes two additional filaments (Fig. S1), and the additional FliM₂-sfGFP foci likely represented incomplete secondary flagellar complexes.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Cells with synchronously functional polar and lateral flagellar systems outperform mutants with polar flagella only in both soft-agar and liquid medium. (A) Representation of CN-32 cells’ radial extension in soft agar (0.25%); 3 µL of exponentially growing cultures of the corresponding strains were allowed to spread for 16 h. Numbers in boxes mark the corresponding sampling areas (1, center; 2, intermediate; 3, rim). (B) Micrographs of CN-32 fliM₂-sfgfp cells isolated from sampling area 1 (Left) and sampling area 3 (Right). Scale bars represent 5 µm. (C) Percentage of fluorescently labeled wild-type and ∆flaAB₂-mutant cells in samples isolated from the corresponding sampling areas; 1:1 mixtures of mCherry-labeled wild type and Gfp-labeled mutant (and vice versa) were used to inoculate the plate. (D) Percentage of fluorescently labeled wild-type and ∆flaAB₂-mutant cells after traveling from reservoir 1 (R1) to reservoir 2 (R2) through a medium-filled channel. R1 was seeded with 1:1 mixtures of the wild type, and wild type and mutant, respectively, and incubated for 16 h. The error bars represent SDs from at least three samples out of two independent experiments each.

In a complementary approach using soft-agar plates, we directly compared the spreading performance of CN-32 cells with or without functional secondary flagella. To this end, we constructed a strain in which we deleted the genes encoding the flagellin subunits of the secondary flagellar system, flaA₂ and flaB₂. Flagellar staining and subsequent microscopy revealed that ΔflaAB₂ cells exclusively formed single polar flagellar filaments. Growth and the percentage of swimming cells of both strains were almost identical. To enable discrimination between wild-type and ΔflaAB₂ cells by fluorescence microscopy, both strains were fluorescently tagged by chromosomal integration of constitutively expressed gfp or mCherry. Then 1:1 mixtures of exponentially growing GFP/mCherry-producing mutant and wild-type cells were allowed to spread in soft agar for 16 h. Samples were taken at different distances from the center of the radial extension zone, and the ratio of the wild-type and mutant cells was quantified by fluorescence microscopy (Fig. 1). We observed that the larger the distance relative to the center of the lateral extension zone which was covered by the cells, the further the ratio of both strains was shifted toward the wild type. At the fringes of the swimming zones, more than 90% of the population consisted of wild-type cells. Filament staining and microscopy revealed that the vast majority of the cells in the sample from the outer rim possessed a single lateral filament in addition to the primary polar one (Fig. S1).

To further determine whether the observed advantage of cells with a secondary flagellar system is not restricted to conditions occurring in soft-agar plates, we conducted a similar spreading competition experiment using chambers that consist of two reservoirs connected by a channel. One reservoir was inoculated with a 1:1 mixture of GFP/mCherry-producing wild-type and ΔflaAB₂ cells. After 12 h of incubation, samples were taken from the second reservoir, and fluorescence microscopy and flagellar staining were applied to dissect the population of cells that had traveled through the channels. Similar to the observations in soft-agar plates, the resulting population was significantly enriched in wild-type cells with a single additional lateral filament (Fig. 1).

The Presence of Lateral Flagella Affects the Directional Changes.

The results of the competition experiments demonstrated that the presence of one or two additional secondary flagellar filaments enhances spreading of CN-32 cells. Preliminary experiments have demonstrated that the presence of the secondary system does not provide an increase in swimming speed (15). To determine whether this is also the case for cells that have covered the greatest distance in soft agar, we determined the swimming speeds of ΔflaAB₂ and wild-type cells that were isolated from the fringes of the swimming halos formed on soft-agar plates (Fig. S2). The population of wild-type cells with additional lateral filaments had a velocity of 46.95 ± 14.52 µm·s⁻¹ and were significantly slower than ΔflaAB₂ cells which exhibited swimming speeds of 57.30 ± 17.18 µm·s⁻¹. Thus, an increase in velocity could be excluded as the reason for the beneficial role of the secondary flagellar system in swimming motility. However, while recording cell trajectories for measuring swimming speeds, we noticed that cells with lateral flagellar filaments exhibited marked differences in their movement patterns compared with cells with polar filaments only. Cells with single polar flagella periodically (about 10 s) switched from forward to backward swimming. Under the conditions tested, the time interval for backtracking was short (0.3 s), in which time the cells covered less than 5 µm distance. Upon resuming forward movement, quick cellular realignments occurred at a range of angles that centered at 90° (Fig. 2). Thus, swimming of monopolarly flagellated S. putrefaciens CN-32 cells apparently follows the “forward-reverse-flick” pattern that has recently been described for Vibrio species (8, 9). A similar forward-backward movement with respect to time and distance intervals was observed for cells with secondary lateral filaments. However, the directional changes upon resuming forward movement occurred at a much wider array of angles, with an average turning angle below 90° (Fig. 2). In addition, the average period between directional switching events increased to about 20 s. We proposed that these two factors, smaller average turning angle and longer runs, may benefit spreading of wild-type cells by increasing directional persistence of swimming, i.e., correlation in the swimming direction over time.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

The presence of a lateral filament affects the trajectories of swimming cells. A representative trajectory is displayed for the wild-type (A) and the ∆flaAB₂ mutant (B), demonstrating a typical forward run (red triangles), reversal (blue square), short backward run (blue triangles), flick (purple circle), and forward run movement. The time between each trajectory point equals 0.07 s. (C) Turning angle distribution of wild-type (black) and ∆flaAB₂-mutant (gray) cells.

Main Propulsion and Directional Switches Are Mediated by the Primary Flagellar System of S. putrefaciens CN-32.

We further conducted a complementary set of experiments in which we determined the potential interaction of the chemotaxis system with the two different flagellar motors. According to the genome data, S. putrefaciens CN-32 has a single chemotaxis system with a broad sensory repertoire represented by 37 putative methyl-accepting chemotaxis sensor proteins. To dissect the chemotaxis pathway in CN-32, we determined the effect of defined mutants in CheY on the two potential receiving motor systems. To this end, we constructed a constitutively active version of CheY (cheY^D12K;Y105W; CheY-GOF “gain of function”) and a nonactive version (cheY^D56N; CheY-LOF; loss of function) (16, 17). The mutated cheY versions were introduced into CN-32 wild type and the mutant backgrounds ΔflaAB₁ and ΔflaAB₂. In addition, a cheY deletion (ΔcheY) was constructed in all three strains. ΔcheY and CheY-LOF mutations in the wild-type background resulted in straight forward-swimming cells, and almost no directional changes were observed in planktonic cultures. In contrast, in cells bearing a CheY-GOF mutation, the average period between forward and backward movements was drastically shortened (<4 s compared with ∼20 s for wild-type cells). All strains were characterized for their ability to navigate in soft-agar plates (Fig. 3).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

CheY interacts with the polar flagellar system. Displayed is the radial extension of CN-32 wild-type and flagellar/chemotaxis mutants in 0.25% soft agar. The strains are labeled accordingly. The nonmotile ΔfliF_1/2 mutant served as a negative control. Strains indicated with “cheY GOF” harbor the gain-of-function variant of CheY; those indicated with “cheY LOF” harbor the corresponding loss-of-function variant. Note that the arrangement of the strains on the plate has been shifted for clarity. An image of the original plate can be found in Fig. S3.

Wild-type cells bearing ΔcheY, CheY-LOF, or CheY-GOF mutations exhibited a drastically reduced radial expansion. Notably, the level of lateral extension in all three mutants was almost identical to that of a ΔflaAB₁ mutant. Furthermore, cells lacking the ability to form the primary polar filament (ΔflaAB₁) were not further affected in swimming motility by additional mutations in or loss of CheY. In contrast, the same CheY mutations introduced into strain background ΔflaAB₂, which lacks the secondary system, resulted in cells that were no longer (ΔflaAB₂ ΔcheY; ΔflaAB₂ CheY-LOF) or just barely (ΔflaAB₂ CheY-GOF) capable of navigating through soft agar. Thus, the observed radial expansion of the cheY mutants on soft agar plates was mainly or exclusively conferred by the secondary lateral system.

In addition, we used light microscopy on ΔflaAB₁ mutants to identify potential differences in swimming behavior due to loss or mutation of CheY. All actively swimming cells were observed to move in irregular patterns, likely due to the lateral position of the flagellar filament, and never switched from forward to backward movement. Cells that were tethered to the glass surface by the lateral flagellar filament displayed constant CCW rotation, and we did not observe directional switches in any of the strains tested (Movie S1). Based on these results, we concluded that CheY predominantly or even exclusively interacts with the primary polar motor. Thus, main propulsion and chemotaxis-induced forward-backward movements are mediated by the primary polar flagellar system. On the other hand, the lateral system has a role in confining the cellular reorientation to smaller angles.

Computational Model of Spreading of Shewanella Wild-Type and Mutant Cells.

To determine whether the observed differences in swimming behavior are sufficient to explain the observed advantage in spreading, we performed a mathematical analysis and computer simulations of motility and chemotaxis of wild-type vs. mutant cells. The movement of cells in a uniform environment without gradients can be described analytically as a 2D correlated random walk (18, 19). In this approximation, the mean square displacement (MSD) of the population after time t, R(t), can be obtained from the autocorrelation function of the velocity:<v(t)v(0)> =exp(−(λ+2Dr)t) v2⁡exp(λγt)= v2exp(−[λ(1−γ) +2Dr]t),[1]where λ is the turning rate, i.e., reciprocal of the mean run duration, D_r is the coefficient of rotational diffusion, v is the speed of swimming, and γ is the persistence factor of the movement, i.e., the mean of the cosine of the turning angles, γ = <cos(Θ)>. Here we assumed an exponential distribution of run durations and neglected the short backtracking movement of S. putrefaciens CN-32 following runs. Double integration in time on Eq. 1 then gives the value of the MSD:R(t)=2 v2{exp[−t(2Dr+λ(1− γ))]+2Drt−1}+2λ v2(1−γ)t(2Dr+λ(1−γ))2.[2]Since the exponential term in the numerator goes to zero on the relevant timescale of the experiments (hours), Eq. 2 simplifies to:R(t)=2 v2t 2Dr+λ(1− γ)− 2 v2 (2Dr+ λ(1− γ))2.[3]Eq. 3 shows that the value of R(t) increases at higher values of the persistence factor γ, as proposed already in a previous theoretical study describing insect movement (18). The lower average turning angle observed for the wild-type cell movement can thus yield higher persistence and lead to faster spreading. This conclusion was confirmed by calculating the root mean square distance (RMSD) for the mutant and wild-type strains using the full expression of Eq. 2 and the experimentally determined parameter values (Fig. 4A and Fig. S4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Modeling and simulations of spreading and chemotaxis for wild-type and ∆flaAB₂ cells. (A) Analytical solution for nondirectional spreading of bacterial cells (RMSD) in absence of chemotactic gradient, using Eq. 2 with experimentally measured parameter values (v_mutant = 57 μm/s, λ_mutant = 0.1 s⁻¹, γ_mutant = 0.058 and v_WT = 47 μm/s, λ_WT = 0.05s⁻¹, γ_WT = 0.214). The coefficient of rotational diffusion was set to D_r = 0.023 rad²·s⁻¹. (B) Numerical simulations of the population spreading with same parameters as in A and using the experimentally measured turning angle distribution, also including backtracking. The lines show the mean of 10 independent simulations, with RMSD of 200 cells determined in each simulation. Dashed lines show minimal and maximal values. (C) Percentage of wild-type and ∆flaAB₂ cells from B after 16 h, at different (radial) distances from the center. The ratio of wild-type to mutant cells rises with increasing distance from the center. The error bars are based on 10 independent simulations. (D) Simulations of chemotaxis in gradients, using the experimentally measured turning angle distributions for wild-type and ∆flaAB₂ cells. The scaling factor ε (see SI Materials and Methods) was set to ε = 0.1. Results are from five independent simulations, each including 1,000 cells. Gradients of indicated steepness (dc/dx) are linearly increasing along the x axis, with c = 0 at x = −100. At the onset of simulation, cells are placed in random orientations at x = 0. The mean position, <x>, of the cell population along the x axis indicates chemotactic drift along the gradient. The units of distance are millimeters, whereas the unit for concentration c is arbitrary. The basal turning rate was set to λ₀ = 0.2 s⁻¹. For simulations with other values of λ₀ and ε, see Fig. S4.

In addition to this analytical calculation, we performed numerical simulations taking into account backward runs. The values of turning angles were generated by assigning discrete probability values to the experimentally measured angles and binning randomly generated numbers by the probability intervals. Durations of individual backward and forward runs were generated using Monte Carlo simulations as in Gillespie’s algorithm (20), assuming exponential probability distributions with the experimentally determined means. Numerical simulations confirmed that the higher persistence factor of the wild-type cells’ movement yields more efficient spreading, if the other parameters have identical values (Fig. 4B and Fig. S4). Although the experimental values of the run duration and swimming speed are different for the wild-type compared with mutant cells, the effects of their longer run periods and lower speed are mutually compensatory, as can be calculated from Eq. 2 (Fig. 4A and Fig. S4) and confirmed by numerical simulations (Fig. 4B and Fig. S4). Therefore, ∼90% of the difference in the RMSD results from the higher persistence of the wild-type movement. While the calculated difference in the RMSD is rather small, it yields a consistent increase in the ratio of wild-type to mutant cells at the edge of the simulated spreading population (Fig. 4C), similar to that observed experimentally (Fig. 1C).

Such enhancement of cell spreading in uniform environments might thus alone explain the benefit conferred by the lateral flagella. Nevertheless, higher persistence of movement has also been proposed to have a positive effect on the chemotactic movement of bacteria in gradients (21⇓–23). We thus simulated the effect of the difference in the measured turning angle distribution on the chemotactic movement using the phenomenological model of chemotaxis described in Locsei (22). This model does not require knowledge of detailed biochemical parameters and assumes that the pathway response to weak stimuli (in shallow gradients) can be described as a convolution integral of the stimulus history with the impulse response (24, 25). Although the impulse response function was measured for E. coli (see details in SI Materials and Methods), it is believed to be generally required for bacterial chemotaxis (26) and should thus be applicable for S. putrefaciens CN-32. We further varied the time window for sensing and the gradients’ steepness, to investigate the effect of persistence under different conditions.

Wild-type cells indeed showed faster chemotactic movement in shallow attractant gradients (Fig. 4D), suggesting that under these conditions, the observed higher persistence of movement is sufficient to enhance chemotaxis. This difference became negligible in steeper gradients (Fig. 4D and Fig. S4), presumably because already short directional runs in steep gradient enable cells to experience strong chemotactic stimulation. Moreover, the model used here is likely to become imprecise in steep gradients. The positive effect of persistence on the chemotactic efficiency is also diminished by the increase in the run time (Fig. S4), because during longer runs rotational diffusion results in the loss of directional correlation. The exact relation between the run time and the benefit of persistence depends on the value of the coefficient of rotational diffusion, which is not known exactly for S. putrefaciens CN-32.