Dynamics of the Two Stator Systems in the Flagellar Motor of Pseudomonas aeruginosa Studied by a Bead Assay

ABSTRACT

We developed a robust bead assay for studying flagellar motor behavior of Pseudomonas aeruginosa. Using this assay, we studied the dynamics of the two stator systems in the flagellar motor. We found that the two sets of stators function differently, with MotAB stators providing higher total torque and MotCD stators ensuring more stable motor speed. The motors in wild-type cells adjust the stator compositions according to the environment, resulting in an optimal performance in environmental exploration compared to that of mutants with one set of stators. The bead assay we developed in this investigation can be further used to study P. aeruginosa chemotaxis at the level of a single cell using the motor behavior as the chemotaxis output.

IMPORTANCE Cells of Pseudomonas aeruginosa possess a single polar flagellum, driven by a rotatory motor powered by two sets of torque-generating units (stators). We developed a robust bead assay for studying the behavior of the flagellar motor in P. aeruginosa, by attaching a microsphere to shortened flagellar filament and using it as an indicator of motor rotation. Using this assay, we revealed the dynamics of the two stator systems in the flagellar motor and found that the motors in wild-type cells adjust the stator compositions according to the environment, resulting in an optimal performance in environmental exploration compared to that of mutants with one set of stators.

INTRODUCTION

Bacteria evolved many different types of motility mechanisms to adapt to complex living environments, among which the most common movement pattern is driven by flagella (1). Flagellar-mediated motility provides an efficient way for numerous bacterial species to swim in liquid and swarm across wet surfaces (2, 3). In general, the flagellum is composed of three parts: the flagellar rotary motor, which is embedded in the cell envelope; the hook, which acts as a universal joint; and the filament, which extends several micrometers from the cell surface (4). The rotation of the flagellar filaments is driven by the flagellar motor, and there are about 20 different types of structural proteins involved in the assembly of this complex nanomachine (5–7). The compositions of the flagellar motor among different bacterial species shares common features, consisting of a membrane-embedded rotor complex surrounded by various numbers of torque-generating units (stator units). The rotation direction of the flagellar motor is controlled by the cytoplasmic portion (C-ring) of the rotor, composed of the proteins FliG, FliM, and FliN. FliM and FliN form a stable base of the C-ring, and conformational changes in FliG facilitate motor switching (8, 9). Torque is generated by the stator units that form ion-conducting transmembrane channels. Ions (H⁺ or Na⁺) flow across the stator units, inducing conformational changes in the stator units that exert torque to the rotor (10–13). The mechanical functions of stators are basically conserved among different bacterial species, but the types of stator systems and driving ions are different. Many bacteria have only one flagellar system and one corresponding set of stators, such as Escherichia coli and Salmonella species, which are powered by H⁺-dependent MotAB stators (12, 14, 15). Some motile bacteria represented by Bacillus subtilis and Shewanella oneidensis possess both H⁺-powered MotAB and Na⁺-powered MotPS (for B. subtilis)/PomAB (for S. oneidensis) stators and select a suitable energy source according to the external environment (16, 17). Unlike the bacteria mentioned above, a number of species have two independent flagellar systems with corresponding stator units. For example, Aeromonas hydrophila and Vibrio parahaemolyticus rely on a single polar flagellum to swim in liquid, but its lateral flagellar system will be activated when cells swim in a viscous environment or swarm on the surface. The single polar flagellum of the two microorganisms is driven by Na⁺-dependent stator units (the A. hydrophila polar motor even has two sets of Na⁺-dependent stators), and the lateral flagellar system is driven by H⁺-powered stator units (18–21).

Pseudomonas aeruginosa is a rod-shaped, Gram-negative bacterium that exists as an opportunistic human pathogen in various natural environments such as soil and liquid. With a single polar flagellum and type IV pili, it is capable of swimming, swarming, and twitching motility (22). Unlike for the other species of bacteria mentioned above, the single polar flagellum of P. aeruginosa PAO1 relies on two sets of H⁺-powered stators, MotAB (PA4953/PA4954) and MotCD (PA1460/PA1461). Both stators can individually support cell to swim efficiently in the liquid environment, but only the MotCD stator can guarantee swimming in liquid with elevated viscosity or swarming on wet surfaces (23, 24). Both the MotAB and MotCD stators are involved in the formation of biofilms and play their respective roles in reversible/irreversible attachment process (25). The dual H⁺-powered stator systems not only generate torque for cell movement but also participate in many molecular regulation mechanisms. The MotAB stator is required for surface sensing and participates in the repression response to ethanol (26, 27). Bacterial two-hybrid experiments showed that MotC directly interacts with SadC to activate its diguanylate cyclase activity. The PilZ domain-containing protein FlgZ specifically interacts with MotC in a c-di-GMP-dependent way (28, 29).

In order to study the motile behavior of P. aeruginosa, several methods, such as swimming/swarming plates (23, 24), tether cell assay (30, 31), and single-cell tracking (32–34), were used previously. The tether cell assay, in which the cell was tethered to a surface via a single flagellum and the rotation of the cell body was monitored, allowed observation of the motile behavior in E. coli at the level of a single motor (35, 36). However, due to the polar localization of the flagellum in P. aeruginosa, the rotation trajectories were usually unstable and needed sophisticated corrections (30). A bead assay, in which a microsphere was attached to a shortened flagellar filament and used as an indicator of motor rotation, has been very useful in studies of flagellar motor in E. coli, as it allowed long-term stable observation of the motor behavior (37–40). A bead assay for P. aeruginosa flagellar motor is lacking. In this study, we developed a bead assay by using the biotin-streptavidin system to attach microspheres to shortened flagellar filaments, thereby achieving long-term stable and accurate measurement of the motor properties. Using this bead assay, we found that the two sets of stator units contribute differently to the motor output. The MotAB stators generate higher total torque to support faster movement, whereas the MotCD stators provide excellent speed stability for the motor. Different stator combinations will not change the clockwise (CW) bias of the motor. We found that the two stator mutants showed different reversal frequencies from the wild-type strain, and the wild-type strain showed the best diffusion ability. By changing the external load, we also discovered a load-dependent stator competition mechanism: the fraction of MotCD units in a wild-type motor increased with load. Thus, we suggest that the two sets of stators compete to assemble into the motor in the wild-type strain according to the external environments to achieve the optimal exploration phenotype.

RESULTS

The flagellar filament maintains left-handed chirality in motor switching.

We first determined the chirality of the flagellar filament and then mapped the motor rotation to the actual swimming behavior. To observe the dynamic behavior of filaments during the swimming process of P. aeruginosa, we introduced a cysteine mutation into the flagellin FliC, and this allowed labeling with thiol-reactive fluorescent dye. This method has been used before to label bacterial flagella or pili (41, 42). We used swimming-plate experiments to confirm that this single point mutation did not significantly affect motility and chemotaxis (see Fig. S1 in the supplemental material). E. coli is known to swim in a “run-and-tumble” mode, and the filament is a left-handed or right-handed helix when the motor rotates counterclockwise (CCW) or clockwise (CW), respectively (43). In contrast, P. aeruginosa adopts a “run-reverse” swimming mode, and each motor switching is often accompanied by a large-angle change of swimming direction (44).

We determined the flagellar chirality by analyzing the near-surface swimming trajectories of the cells (45), and a typical reversal event of the wild-type strain is shown in Fig. 1A (images taken from Movie S1). When the filament pushed the cell body forward, we observed that the trajectory of the cell was clockwise when viewing from the top (Fig. 1B). The clockwise swimming trajectory resulted from the pair of net viscous forces F₁ in the +y direction and F₂ in the −y direction acting on the cell body and filament, respectively. A left-handed filament rotating CCW (viewed from the tail of the filament) or a right-handed filament rotating CW would push the cell forward. According to the direction of the net viscous force F₂ acting on the filament, we could know that the filament rotates CCW (Fig. 1C, upper portion). Thus, the flagellum was left-handed when it pushed the cell body forward. Similarly, the counterclockwise trajectory of bacteria pulled by filament near the surface proved that the flagellum was also left-handed (Fig. 1B and C, lower portion). Therefore, the flagellar chirality was not changed when the direction of flagellar motor rotation switched. From these results, we concluded that the filament of P. aeruginosa maintain a fixed left-handed chirality throughout the whole process of swimming. When the motor rotates CCW, the filament pushes the cell forward; otherwise, the filament pulls the cell forward.

FIG 1.

(A) A typical reversal event of the wild-type strain near the surface, with image sequences taken from Movie S1. The bottom right corner is a schematic diagram of the complete cell trajectory. The cell body is indicated by a red dashed box, and the white arrows indicate the instantaneous velocity direction of the cell. (B and C) Diagrams of swimming behavior near the surface (B, top view; C, tail view). In panel B, the dashed lines with arrow indicate the direction of cell motion. “F₁” and “F₂” indicate the net viscous forces acting on the cell body and flagellum, respectively. In panel C, “F_bottom” and “F_top” represent the viscous force on the portions of the helical filament closer to and away from the surface, respectively. The upper and lower portions in panel C represent the situations when the flagellum push and pull the cell body, respectively. “CW” and “CCW” represent the rotational direction of the filament. (D) The probability distribution of angular change associated with a reversal. The upper graph counts the events changing from pushing to pulling, and the lower graph depicts the opposite situation. Abbreviation: PDF, probability density function.

We also separately counted the distributions of angular changes in swimming direction when the flagellar motor switched from CCW to CW and from CW to CCW. We found that the distributions peaked near 180° and were similar (Fig. 1D).

A bead assay for the flagellar motor in P. aeruginosa.

We first biotinylated the filament (with cysteine point mutation FliC^T394C) with maleimide- polyethylene glycol 2 (PEG₂)-biotin, which took advantage of the specific and efficient binding between the maleimide group and sulfhydryl group at pH 6.5 to 7.5. The cells were immobilized on a coverslip modified with poly-l-lysine. Streptavidin-modified beads were then attached to biotinylated flagellar filaments, and the rotation of individual beads was measured (Fig. 2A). Because streptavidin-biotin binding is one of the strongest noncovalent interactions in nature and the sizes of different beads showed a coefficient of variation of less than 5% (46), it is possible to observe the properties of individual flagellar motors for a long time under well-defined load conditions. Cells were sheared to truncate the flagellar filaments so that the rotation radius of the microspheres was small, and thus, stability of the rotation was ensured. A typical video of rotating microsphere is shown in Movie S2.

FIG 2.

(A) Schematic diagram of bead assay. Flagellin containing a cysteine point mutation was biotinylated by using sulfhydryl-maleimide conjugation. Bacterial cells were immobilized onto a poly-l-lysine coated coverslip, and a streptavidin-coated bead was attached to the biotinylated filament stub. (B) A two-dimensional XY signal plot, showing the trajectory of the bead image.

By imaging through phase-contrast microscopy and recording the images with a high-speed scientific complementary metal oxide semiconductor (sCMOS) camera, we can obtain real-time position information of the microspheres with millisecond temporal resolution and nanometer spatial resolution. We monitored the motor rotation in motility buffer (MB) at room temperature, and Fig. 2B shows a sample trajectory of rotation when a 1-μm bead was attached to shortened flagellar filament. Compared with the tether cell assay, due to the better stability of the marker, the original trajectory could be directly used for further analysis to obtain motor dynamics without excessive postprocessing.

MotAB stators provide higher total torque than MotCD stators.

In order to describe the dynamics of the motor more accurately, approximately 150 motors were measured for each of the wild-type, ΔmotAB, and ΔmotCD strains. The typical time traces of the rotation speeds of the wild type and two stator mutants are shown in Fig. 3A. The motors of wild-type strain PAO1 rotated CCW and CW at 51.75 ± 9.06 Hz and 48.51 ± 8.95 Hz, respectively. The motor speed of the ΔmotAB strain (CCW at 33.35 ± 8.45 Hz and CW at 34.56 ± 8.92 Hz) was nearly 40% lower than that of the wild type, whereas the ΔmotCD strain (CCW at 42.51 ± 11.36 Hz and CW at 41.30 ± 10.11 Hz) showed a motor speed slightly lower than that of the wild type (shown in Fig. 3B). The one-way analysis of variation (ANOVA) and paired-sample t tests showed significant statistical differences between the motor speeds of the ΔmotAB strain and ΔmotCD strain (P value < 10⁻⁴). These results are consistent with the previous measurements of the swimming speed of P. aeruginosa: the swimming speed of the ΔmotCD strain was shown to be only slightly lower than that of the wild type (23). Therefore, MotAB stators provide higher total torque than the MotCD stators. In this study, we further verified that the rotation speeds of the motor were almost the same in the CCW and CW directions at high load.

FIG 3.

(A) Typical time traces of rotation velocities for the wild-type strain and the two stator mutants (from top to bottom are the wild-type, ΔmotAB, and ΔmotCD strains). A hyphen before “50” indicates CCW direction. (B) CW and CCW rotation speeds of the wild-type, ΔmotAB, and ΔmotCD strains. The speed of the ΔmotAB strain is nearly 40% lower than that of the wild type, and the ΔmotCD strain shows a speed comparable to that of the wild type. (C) The speed increases per stator for MotAB and MotCD are almost the same. (D) The number of stators in the ΔmotCD strain is more than in the ΔmotAB strain. In the box-whisker plot, the box represents the middle 50% of the data. The median value is shown as a line in the box, and the whiskers denote the data range of the 5th and 95th percentiles.

We tried to further explore whether the speed difference between the two stator mutants is caused by difference in the torque generated per stator unit or the number of stators bound. Long-term tracking of individual motors allowed us to observe the real-time speed change caused by the stochastic binding and unbinding of single stator unit and then determine the increase in speed per stator unit and the number of stators in the motor under steady-state conditions. Eighteen ΔmotAB ΔcheY cells and 10 ΔmotCD ΔcheY cells were analyzed, and a single trace contained at least 3 speed jumps (see Fig. 5 below for sample traces). The speed increase per MotAB stator was calculated to be 5.81 ± 1.84 Hz and the speed increase per MotCD stator was 5.41 ± 1.12 Hz (Fig. 3C). The speed contributions for the stator units were almost the same, demonstrating that the torque generated per stator unit was the same for the two sets of stators. We use the steady-state speed of the motor and the speed increase per stator to calculate the steady-state number of stators in the motor. ΔmotAB cells (MotCD stators only) and ΔmotCD cells (MotAB stators only) have, on average, 6 ± 2 and 9 ± 3 stator units, respectively (shown in Fig. 3D). These findings indicated that the reason why the ΔmotCD strain showed a faster speed (than the ΔmotAB strain) was because of greater number of stators bound in the motor.

FIG 5.

Typical traces of motor speed in a steady state for motors of the ΔmotAB (upper) and ΔmotCD (lower) strains. The blue lines are the average speeds found by the step-finding algorithm.

Motor switching properties with different stator compositions.

Various types of swimming bacteria achieve chemotaxis through different motility mechanisms. The best-known model bacterium, E. coli, changes the up-gradient and down-gradient run durations by biasing the probability of the flagellar motor rotation direction (47). P. aeruginosa has an unbiased flagellar motor (CW bias of 0.5), and the CW bias does not change with the external chemoattractant concentration (44). In light of the fact that the two sets of stators are not functionally equivalent, and different combinations of stators may be used in actual movement according to specific environmental conditions, we sought to test whether different combinations of the stators would change the CW bias. We measured the CW bias distributions of the two deletion mutants and the wild type, and the average CW bias is close to 0.5 for all of the three strains (Fig. 4A). These findings showed that for P. aeruginosa flagellar motors, different stator combinations had the same unbiased properties of motor rotation direction.

FIG 4.

(A) CW biases of the wild-type, ΔmotAB, and ΔmotCD strains are all close to 0.5. (B) CW and CCW intervals of the wild-type, ΔmotAB, and ΔmotCD strains. Compared to the case with the wild-type strains, both the CW and CCW intervals lengthened for the ΔmotAB strain, whereas they shortened for the ΔmotCD strain. (C) Switching rates of the wild-type, ΔmotAB, and ΔmotCD strains.

To elucidate the influence of the two sets of stators on motor switching, we carried out a large quantity of measurements on individual motors with the two stator mutants and the wild type using the bead assay. Two hundred ninety-four motors of the wild-type strain, 301 motors of the ΔmotAB strain, and 139 motors of the ΔmotCD strain were measured, each about 3 min long, extracting more than 8000 CW/CCW intervals for each strain. Figure 4B and C show the average CW and CCW intervals for the three strains, together with the switching rates for transitions between the CW and CCW states. Compared to the wild-type strains (with switching rate [k_switch] of 0.42 ± 0.14 s⁻¹), both CW and CCW intervals lengthened appreciably for the ΔmotAB strain (with decreased k_switch [0.30 ± 0.14 s⁻¹]), whereas they shortened for the ΔmotCD strain (with increased k_switch [0.68 ± 0.26 s⁻¹]).

MotCD stators maintain higher stability in the motor speed than MotAB stators.

Previous studies showed that the stator system of the bacterial flagellar motor in E. coli is dynamic. In E. coli, when stator proteins were produced from an inducible promoter in a stator-defective mutant, the motor speed showed stepwise increases (38, 48, 49). Fluorescently labeled MotB proteins were observed to exchange between the flagellar motor and the membrane pool containing about 200 MotB molecules (50, 51). Under a high-load steady-state condition, the flagellar motor speed showed stochastic stepwise jumps, indicating the random binding and unbinding of the stators to the motor (52). These studies provided convincing evidence for the dynamic turnover of the stators in motors of E. coli. We sought to test whether the stators also dynamically turn over in P. aeruginosa. We conducted long-time measurements on individual motors of two P. aeruginosa stator mutants (along with cheY deletion), and typical traces of motor speeds are shown in Fig. 5 (along with the average speeds found by our step-finding algorithm). Both showed stepwise jumps of the motor speeds, indicating the dynamic turnover of the stators. We extracted more than 100 speed segments for each strain. The duration of each segment was extracted as the stator dwell time. For ΔmotAB and ΔmotCD strains, the average dwell times were 111.36 ± 13.99 s and 26.28 ± 2.74 s, respectively. We further calculated the average dwell time for the “off” segments, which are defined as the segments that are followed by a speed decrease, and the unbinding rate constant was calculated as the inverse of it. We found that the unbinding rate constant of the ΔmotAB strain (0.033 ± 0.005 s⁻¹) is much smaller than that of the ΔmotCD strain (0.149 ± 0.015 s⁻¹), indicating a much higher stability for the MotCD stators.

By adding Ficoll to the motility buffer to increase the viscosity, we measured the relationship between the dwell time and the external load of the motor. For ΔmotAB cells in motility buffer containing 5% Ficoll, the dwell time increased to 155.81 ± 18.84 s. In contrast, for ΔmotCD cells in motility buffer containing 2.5% Ficoll, the dwell time decreased to 15.95 ± 2.36 s. This implied that increasing the motor external load will further improve the stability of the MotCD stator and also increase the turnover rate of the MotAB stator. Therefore, the MotCD stator units make an important contribution to the stable speed output of the motor. The reason for the high turnover rate of MotAB stator units is unclear.

Evidence of load-dependent stator assembly in a wild-type motor.

The MotCD stator units are believed to power movement on wet agar surfaces (23), and the ΔmotCD strain completely loses its swimming ability in a relatively high-viscosity liquid environment (24). A recent tether cell experiment found that tethered ΔmotCD bacteria rarely rotated, while ΔmotAB cells were observed to rotate for a long time (31). The above-described experiments all indicate that the MotCD stator units are essential for the rotation of the motor under a high-load condition, which makes us wonder whether the fraction of MotCD stator units in the wild-type motor is load dependent.

Media with different Ficoll concentrations (0, 2.5, 5, 7.5, and 10%) were used to adjust the external load of the motor with the bead assay. Approximately 30 wild-type and ΔmotAB cells were measured under each load condition, each for about 3 min. We found that as the load increased, the wild-type and ΔmotAB strains became more and more similar in switching rate and rotation speed, whereas motors of the ΔmotCD cells stopped rotation within 3 min in motility buffer containing 5% Ficoll or above (Fig. 6). Nevertheless, as seen from the velocity traces (shown in Fig. S2), the speed stability of the wild-type strain in motility buffer containing 10% Ficoll is still a bit weaker than that of the ΔmotAB mutant, which implies that the MotAB stator units may still have a subtle effect on the speed stability in the wild-type strain even at high loads.

FIG 6.

Behavior of the flagellar motor for the wild-type, ΔmotAB, and ΔmotCD strains under different load conditions. (A) The switching rates of the three strains under different load conditions. (B and C) The CCW (B) and CW (C) rotation speeds of the three strains under different load conditions. (Inset) Viscosities of the motility buffer with different Ficoll concentrations at room temperature.

We further measured the average number of stators for the two stator mutants when the viscosity was changed using Ficoll. For ΔmotAB cells in the motility buffer containing 5% Ficoll, the stator number increased to 8 ± 2. In contrast, for ΔmotCD cells in the motility buffer containing 2.5% Ficoll, the stator number decreased to 7 ± 2 (compared to 0% Ficoll). These findings indicated that the greater the external load, the higher the proportion of the MotCD stator units in the motor, which made the wild-type strain exhibit a phenotype similar to that of the ΔmotAB mutant at high loads.

The wild-type strain performs better than the mutant strains with one set of stators in environmental exploration.

The measurements above showed that the sets of stators contribute differently to behavior of the motor. The MotAB stators generate higher total torque to support faster movement, whereas the MotCD stators provide excellent speed stability for the motor. The two sets of stators induce different switching rates of the motor. The wild-type motor adjusts its stator composition according to the external conditions. We envisioned that the wild-type strain includes both sets of stators to achieve optimal performance in environmental exploration compared to the mutant strains with one set of stators.

To test that, we compared the swimming behavior of the wild-type, ΔmotAB, and ΔmotCD strains by using swimming plates (0.3% agar) and single-cell tracking assay. The sizes of the expansion rates on a swimming plate are different for the wild-type and the stator mutant strains, with the expansion ring of the wild-type always larger than that of the mutants after the same duration of incubation. We measured the expansion rates of the three strains, each measured for 10 swimming plates, and obtained average expansion radii of 2.62 ± 0.17, 2.18 ± 0.11, and 2.05 ± 0.12 cm after 18 h of incubation for the wild-type, ΔmotAB, and ΔmotCD strains, respectively (Fig. 7A). This is consistent with previous experiments on MotAB and MotCD stators, in which the wild-type cells showed a larger expansion ring than motAB or motCD mutants on swimming plates (23).

FIG 7.

(A) The expansion of the three strains on a swimming plate. (Upper portion) Scatterplot of the expansion radius for the wild-type, ΔmotAB, and ΔmotCD strains. (Lower portion) Photo of a swimming plate with three strains. The plate was photographed at 18 h after inoculation. (B) The relationship between mean squared displacement (MSD) and the time lag of the wild-type, ΔmotAB, and ΔmotCD strains (upper portion, linear coordinates; lower portion, double logarithmic coordinates), all exhibiting superdiffusive behavior, while the wild type is most diffusive.

We also analyzed the swimming trajectories of the wild-type and two stator mutants in bulk liquid (100 trajectories for each strain), and the relationship between mean squared displacement (MSD) and the time lag of the three strains are shown in Fig. 7B. A power law form, MSD = 4D × t^α, was used to fit this relationship and then determined the diffusion coefficient (D) and power law exponent (α). The diffusion coefficient for the wild type (522.2 μm²/s) is larger than that for the ΔmotAB (66.0 μm²/s) and ΔmotCD strains (260.4 μm²/s), whereas the power law exponents for the three strains are similar (1.86 ± 0.20, 1.60 ± 0.43, and 1.68 ± 0.33, respectively). These indicated that each of them exhibited superdiffusive behavior, while the wild type was the most diffusive, with the highest diffusion coefficient. Therefore, the combination of the two sets of stators in the wild-type strain achieves a better performance in environmental exploration than the mutants with one set of stators.

DISCUSSION

In this study, we developed a robust bead assay for motors of P. aeruginosa. We used the biotin-streptavidin system to reliably attach microbeads to the shortened flagellar filaments of P. aeruginosa, thus achieving long-term stable measurement of the dynamic properties of individual flagellar motors. Using this bead assay, we studied motor dynamics for the wild-type, ΔmotAB, and ΔmotCD strains. The motor speed for the ΔmotCD strain is only slightly lower than that of the wild type, whereas that for the ΔmotAB strain is 40% smaller. This was because of the greater number of stators bound in motors of the ΔmotCD strain compared to that for the ΔmotAB strain, as the torques generated per stator unit were similar for the MotAB and MotCD stators. In addition, the CW and CCW rotation speeds were similar at high loads. We found that the CW bias of the two stator mutants was close to 0.5, similar to that of the wild type. This means that the unbiased characteristic of the motor is maintained even if the composition ratio of the two stator units is changed according to the environmental conditions.

Previous research showed that P. aeruginosa performs chemotaxis by regulating the motor switching rate (44). We found that the motor switching rates of the two stator mutants were significantly different under the same condition, whereas that of the wild-type strain lay somewhere in between. We suspect that the wild-type strain could also regulate the motor switching rate by adjusting the stator composition ratios under different environmental conditions, thereby achieving an additional level of chemotaxis capability. The mechanism for the difference in switching rates for the two sets of stators is unclear. A possible mechanism might involve c-di-GMP. Previous studies showed that the loss of any stator units will reduce the intracellular c-di-GMP level, as MotC could interact with the diguanylate cyclase SadC to stimulate c-di-GMP production (29). The level of c-di-GMP could affect the motor switching rate in multiple ways. One possible way is through the interaction of c-di-GMP-binding effector protein FlgZ with MotCD stators (28). The other is through the c-di-GMP-binding adaptor protein MapZ, which is located at the cell pole, by its interaction with the methyltransferase CheR1, and when bound with c-di-GMP, it affects the methylation level of methyl-accepting chemoreceptors and ultimately affects motor switching (53, 54).

Long-term observation of individual motors showed different stabilities of the two sets of stators, and by changing the external load, we observed the competitive binding dynamics for the MotAB and MotCD stators. The different characteristics of the flagellar motor induced by the two sets of stators suggested that the joint participation of the MotAB and MotCD stators makes wild-type P. aeruginosa achieve better performance in environmental exploration than the mutants with one set of stators, which we demonstrated by swimming-plate assay and single-cell tracking. The flagellar motor systems of B. subtilis and S. oneidensis also have two sets of stator systems, but each stator system depends on different coupling ions (H⁺ and Na⁺). In B. subtilis, it was found that the active number of MotPS stators increased significantly with the increase of external viscosity, whereas the opposite is true for MotAB stators. Increasing Na⁺ concentration in the solution will increase the number of Na⁺-powered stators MotPS (for B. subtilis) or PomAB (for S. oneidensis). The above-mentioned facts indicated that the mechanism of stator competitive assembly may be widely present in dual-stator motors, which provided a possible avenue for such bacteria to better respond to environmental changes.

In E. coli, the stators alternate between the flagellar motor and a membrane pool, with the rates of stator binding/unbinding to the motor depending on the load. This results in load-dependent stator remodeling dynamics, and the number of stator units increases with load (55–59). In this study, we observed in P. aeruginosa that the number of MotCD stator units increased with load, whereas the number of MotAB stator units decreased with load. This indicated that the assembly dynamics of the MotCD stator may be similar to that of the E. coli stator, whereas that of MotAB stator is different.

We sought to compare the amino acid sequences of P. aeruginosa MotAB/MotCD and E. coli MotAB, to extract possible information on the origin of the similarity and difference between the assembly dynamics of these stators. The degrees of identity and similarity between the amino acid sequences of the two sets of P. aeruginosa stators have been described previously (24). We further analyzed the charged residues in MotA/MotC that interact with FliG, as well as the peptidoglycan-binding (PGB) motif in MotB/MotD (as generated by the software Clustal Omega [60]). Several functionally important charge residues (Arg-90 and Glu-98 in MotA) were identified previously that were essential for the interaction between the rotor and stator in E. coli (61). These residues are conserved in P. aeruginosa MotA and MotC (Fig. S3A). This implies that the stator-rotor electrostatic interaction mechanism in P. aeruginosa is similar to that in E. coli. The PGB motif is widespread in many membrane proteins, and mutations near the PGB motif of MotB in E. coli appear to misalign the stator and the rotor (62, 63). We found this conserved domain in MotB and MotD of P. aeruginosa (Fig. S3B), which indicates that MotAB and MotCD stators rely on similar PGB motifs to target and stably anchor to the peptidoglycan layer. Therefore, the different phenotypes of the two stator mutants might be due to difference in regions other than the critical charged residues or the PGB motif. In the future, we look forward to using the bead assay we developed in this investigation to further study this dual-stator motor in order to unveil its mystery and to study P. aeruginosa chemotaxis at the level of the single cell.

MATERIALS AND METHODS

Strains and cell culture.

The strains and plasmids used in this study are listed in Table 1. The plasmid fliC^T394C-pJN105 was electroporated into a filament-defective strain for the bead assay. A single-colony isolate was grown in 3 ml of LB broth (1% Bacto tryptone, 0.5% yeast extract, and 1% NaCl) overnight to saturation on a rotary shaker (250 rpm) at 37°C. An aliquot was diluted 1:100 into 10 ml of LB broth and grown to exponential phase (the growth curves of wild-type and gene knockout mutants are shown in Fig. S4). To prevent plasmid loss, 30 μg/ml of gentamicin was added to media for cultivation of the strains containing pJN105 derivative vectors. Arabinose (0.01%) was added to induce protein expression. A 5-ml volume of cells was harvested by centrifugation at 4,000 × g for 2 min, washed twice in an equal volume of motility buffer (MB) (50 mM potassium phosphate, 15 μM EDTA, 0.15 M NaCl, 5 mM Mg²⁺, and 10 mM lactic acid [pH 7.0]) (44, 64), and resuspended in 2 ml of MB. The Escherichia coli TOP10 strain was used for standard genetic manipulations.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Genotype, phenotype, or description	Source
Strains
P. aeruginosa
PAO1 mutant	Wild-type strain	J. D. Shrout
ΔfliC mutant	Nonpolar fliC deletion in PAO1	This study
ΔfliC ΔmotAB mutant	Nonpolar motAB deletion in ΔfliC mutant	This study
ΔfliC ΔmotCD mutant	Nonpolar motCD deletion in ΔfliC mutant	This study
ΔfliC ΔmotAB ΔcheY mutant	Nonpolar cheY deletion in ΔfliC ΔmotAB mutant	This study
ΔfliC ΔmotCD ΔcheY mutant	Nonpolar cheY deletion in ΔfliC ΔmotCD mutant	This study
E. coli
Top10	F− mcrA Δ(mrr-hsRMS-mcrBC) ϕ80lacZΔM15 ΔlacX74 recA1 ara D139 Δ(araleu) 7697 galU galK rpsL (Nalr) endA1 nupG	Invitrogen
Plasmids
pex18gm	oriT+ sacB+; gene replacement vector with MCS from pUC18; Gmr	Fan Jin Group
fliC-pex18gm	In-frame deletion of fliC cloned into pex18gm; Gmr	This study
motAB-pex18gm	In-frame deletion of motAB cloned into pex18gm; Gmr	This study
motCD-pex18gm	In-frame deletion of motCD cloned into pex18gm; Gmr	This study
cheY-pex18gm	In-frame deletion of cheY cloned into pex18gm; Gmr	This study
fliC-pJN105	fliC overexpression vector in pJN105; fliC expression is controlled by PBAD promoter; Gmr	This study
fliCT394C-pJN105	fliCT394C overexpression vector in pJN105; fliCT394C expression is controlled by PBAD promoter; Gmr	This study

Selection of residues for cysteine substitution.

The mutation residues were selected by following the protocol described previously (65). Briefly, serine or threonine residues in FliC sequence were selected as potential mutation targets. Then the software NetSurfP was used to calculate relative surface accessibility (RSA) of individual amino acid residues. The residues with too-high or too-low RSA scores were excluded. We selected approximately 10 sites (including T394, T248, S290, etc.) for mutation and then used maleimide dye to label the mutant strains (details in “Labeling of flagella and fluorescence imaging” below). The strain with the mutation fliC^T394C showed the highest labeling efficiency; thus, it was used in this study.

Construction of gene deletion mutants in P. aeruginosa.

PCR was used to generate ∼1,000-bp DNA fragments upstream (Up) or downstream (Dn) from the fliC, motA-motB, motC-motD, and cheY genes. The Up and Dn DNA fragments for each gene to be deleted and the linearized pex18gm vector were connected via Gibson assembly (66). The recombinant vectors were electroporated into P. aeruginosa, and the gene deletion mutants were obtained by double selection on LB plates supplemented with gentamicin (30 μg/ml) and NaCl-free LB plates containing 15% sucrose at 37°C (67).

Swimming-plate assay.

Swim agar (0.3% Bacto agar, 1% Bacto tryptone, 0.5% yeast extract, and 1% NaCl, supplemented with 0.01% arabinose and antibiotics if necessary) was prepared. Polystyrene petri plates (150-mm diameter) were filled with 30 ml of swim agar and the plates were inoculated with a 5-μl drop of a fresh overnight culture of bacteria. The swimming plates were incubated at 37°C for ∼18 h, and then the radii of the swimming zones were measured.

Bead assay.

Cells were sheared to truncate filament by passing 1 ml of the washed-cell suspension 200 times between two syringes equipped with 23-gauge needles and connected by a 7-cm length of polyethylene tubing (0.58-mm inside diameter [i.d.], no. 427411; Becton, Dickinson) and were condensed into 500 μl of MB. A 5-μl solution of maleimide-PEG₂-biotin (Thermo Scientific; 10 mg/ml in dimethyl sulfoxide [DMSO]) was added, and biotinylation of the filaments was allowed to proceed for 20 min at room temperature, with gyrorotation at 200 rpm.

To measure motor dynamic properties (such as rotation speed, CW bias, switching rate, etc.) for the wild type and two stator mutants, the acquisition time of a single motor is 3 min. A 50-μl volume of cells was added to the chamber (constructed with two pieces of double-stick tape, a glass slide, and a poly-l-lysine-coated coverslip), incubated for 3 min, and rinsed with 100 μl of MB. Streptavidin-coated beads (1 μm; Invitrogen-Molecular Probes) were washed in phosphate-buffered saline (PBS), resuspended in MB (supplemented with 2.5%, 5%, 7.5%, or 10% Ficoll if necessary), and then injected into the chamber to spontaneously attach to the biotinylated filaments (59). The chamber was sealed with grease, and each sealed chamber was used for at most 15 min to ensure that the external environment (including oxygen concentration, chemical composition, etc.) had not changed appreciably.

In order to quantify the stability of the motor stator, long-time measurement of the motor speed trace was performed. Cells were placed in a hydrodynamic flow chamber (constructed by using a ring-shape double-sided sticky tape as a spacer between a glass slide with two 0.9-mm-diameter holes and a glass coverslip coated with poly-l-lysine, sealed with grease) to ensure that the external oxygen concentration is stable. A solution of 1.0-μm-diameter streptavidin-coated beads was injected into the flow chamber to spontaneously attach to the sheared flagellar stubs. The flow chamber was kept under a constant flow of fresh MB (50 μl/min) using a syringe pump (pump 22; Harvard Apparatus). Each motor was usually measured for more than 10 min. The rotation of the beads was observed using a Nikon Ti-E phase-contrast microscope with a 40× objective and recorded with an sCMOS camera (C11440; Hamamatsu) at 500 frames per second (fps).

Labeling of flagella and fluorescence imaging.

Cells were labeled following the protocol described previously (68). Cells (1 ml of culture with cells in exponential phase) were harvested by centrifugation at 2,000 × g for 10 min and washed twice in 1 ml of MB. The final pellet was adjusted to a volume of ∼100 μl that concentrated the bacteria 10-fold. Four microliters of Alexa Fluor 568 maleimide (Invitrogen-Molecular Probes; 5 mg/ml in DMSO) was added, and labeling was allowed to proceed for 30 min at room temperature, with gyrorotation at 80 rpm. Then unused dye was removed by washing cells with MB three times.

To judge the flagellar chirality, 100 μl of the cell suspension was added to a three-dimensional sample chamber, which was made from a coverslip supported by two strips of 1-mm-thick double-sided sticky tape on a glass slide. The boundary of the chamber was then sealed with Apiezon vacuum grease. The chamber was put on a Nikon Ti-E inverted fluorescence microscope with a filter set for fluorescein, a 100× oil-immersion objective, and an sCMOS camera (Primer95B; Photometrics). The swimming cells near the lower coverslip were observed at 80 fps, using a 12.5-ms exposure time.

Cell tracking.

To track the cells and calculate the MSD of cell swimming in the bulk liquid environment, the three-dimensional sample chamber describe above was used. Cell motile behavior was observed by using a Nikon Ti-E phase-contrast microscope with a 20× dry objective and an sCMOS camera (C11440; Hamamatsu) at a frame rate of 30 fps. To eliminate surface effect on swimming of the cells, the focal plane was set to be about 300 μm above the lower glass surface.

Data analysis.

Data analysis was carried out with custom scripts in MATLAB (Mathworks). The motor rotation speeds were calculated as described previously (69). Briefly, the x and y positions were converted to polar coordinates, and the angle was differentiated with respect to time to yield the rotation velocity (CW positive and CCW negative). Speed histogram was plotted for each measurement and the steady-state velocities during CW and CCW rotation were extracted. The span between the two steady-state velocities was equally divided into three sections, and the transition between the top section and the bottom section was defined as a motor switching. The motor velocity traces were smoothed with a 10-point running average, resulting in a time resolution of about 0.02 s. ImageJ software was used for tracking the trajectory of cell. Trajectories longer than 6 s were used to calculate MSD.

The speed traces of long-time motor speed measurements showed stepwise jumps of the motor speed. The length and position of each segment of the speed step were determined using a step-finding algorithm described previously (57, 70). After determining the speed change for each speed jump, we sought to determine the speed change per stator unit, as some of the speed changes may correspond to the addition of two stator units within a short time span such that the step-finding algorithm was unable to distinguish them (55). The speed changes for each motor speed trace were defined as an independent data set, and the average value of the speed changes for the data set was called Δv. We defined α = 1.5 × Δv as the threshold, and speed changes larger than α were removed from the data set. Then Δv and α for the updated data set were calculated, and the data set was updated again by removing speed changes larger than the new α. The procedure was iterated until Δv stabilized. The resulting Δv is the speed change per stator unit for this motor. The number of stator units for each motor was estimated as the average speed of the speed trace divided by Δv.

Data availability.

All data analysis scripts were uploaded online (http://github.com/dyht/bac_analysis).