Motility is an important trait for some bacteria living in nature and the analyses of it can provide important information on bacterial ecology. While the swimming behavior of peritrichous bacteria such as Escherichia coli has been extensively studied, the monotrichous bacteria such as the soil inhabiting and plant growth promoting bacterium Pseudmonas fluorescens is not very well characterized. Unlike E. coli that is propelled by a left-handed flagella bundle, P. fluorescens SBW25 swims several times faster by rotating a right-handed flagellum. Its swimming pattern is the most sophisticated known so far: it swims forward (run) and backward (backup); it can swiftly ‘turn’ the run directions or ‘reorient’ at run-backup transitions; it can ‘flip’ the cell body continuously or ‘hover’ in the milieu without translocation. The bacteria swam in circles near flat surfaces with reduced velocity and increased turn frequency. The viscous drag load due to wall effect potentially accounts for the circular motion and velocity change, but not the turn frequency. The flagellation and swimming behavior of P. fluorescens SBW25 show some similarity to Caulobacter, a fresh-water inhabitant, while the complex swimming pattern might be an adaptation to the geometrically restricted rhizo- and phyllospheres.
Flagella-mediated motion is indispensable in microbial ecology, but was historically underdeveloped partially due to the fact that the physics governing microbial behavior is different from the macroscopic ecological phenomena (Berg, 1990). Motility promotes bacterial fitness through relocation to an optimal site in the ever-changing environment (Berg, 1975; Fenchel, 2002). As a consequence, different cell shapes and flagellations have been evolved in bacteria (Young, 2006). The swimming behavior of the peritrichous rods such as Escherichia coli and Salmonella enterica serovar Typhimurium has been extensively studied (Berg, 2003; Ping, 2012). Their lateral flagella form a left-handed helical bundle that rotates counterclockwise (when viewed from behind the cell) to propel the cell forward (Termed run; Darnton et al., 2007). The nearly straight runs are punctuated by erratic reorientation every 1 s, when flagella reverse rotation direction (Termed tumbles; Ping, 2012).
Unlike enteric bacteria, pseudomonads are traditionally regarded as monotrichous, i.e. they are propelled by a single flagellum at one of the cell poles (Lugtenberg et al., 2001). This is a highly diverse bacterial genus, including plant growth promoting organisms (e.g. Pseudomonas fluorescens), bioremediation agents (e.g. Pseudomonas putida), and opportunistic pathogens (e.g. Pseudomonas aeruginosa; Holloway & Morgan, 1986). The available information on their swimming behavior, though sparse in literature, clearly indicates that they adopt completely different strategies from the peritrichous ones (Ping, 2012). For example, P. aeruginosa swims much faster than E. coli (Vaituzis & Doetsch, 1969; Conrad et al., 2011); Pseudomonas citronellolis swims backward (Termed backup) rather than tumble when its flagellum reverses rotation (Taylor & Koshland, 1974); P. putida can swiftly change the direction while swim forward (Termed turn; Harwood et al., 1989; Duffy & Ford, 1997; Davis et al., 2011).
Many of the fluorescent pseudomonads are plant commensals living in rhizosphere and phyllosphere (Loper et al., 2012). Pseudmonas fluorescens SBW25 is a widely-used microorganism in ecological research due to its potential applications in agriculture (Jäderlund et al., 2008; Silby et al., 2009). It was originally isolated from the leave of sugar beet (Thompson et al., 1995). When inoculated in soil or on seeds, it colonizes the root (Maraha et al., 2004) and other plant surfaces (Thompson et al., 1995; Unge & Jansson, 2001; Humphris et al., 2005). Motility plays a key role in the dispersion and surface colonization of soil inhabiting bacteria (Barahona et al., 2010). When P. fluorescens SBW25 colonizes rhizosphere, one of the induced genes is fliF, encoding a component of bacterial flagella (Gal et al., 2003; Silby et al., 2009). Motile P. fluorescens SBW25 cells are more efficient on attaching to the plant root compared to the isogenic nonmotile cells (Turnbull et al., 2001). Hypermotile variants of P. fluorescens F113 arise spontaneously when they grow in rhizosphere (Navazo et al., 2009). However, the motility of P. fluorescens SBW25 and other strains were merely studied en masse that based on agar diffusion and capillary assays thus far (Faust & Doetsch, 1969; Turnbull et al., 2001; Barahona et al., 2010). In this communication, we report the first systematic analysis of single-cell behavior of P. fluorescens SBW25 and the comparison with what is known for the peritrichous bacteria and other Pseudomonas. The ecological implications of the sophisticated swimming pattern were also discussed.
Materials and methods
Fluorescence staining of bacterial flagella was performed with a modified procedure that was originally developed for staining the flagella of E. coli by Turner et al. (2000). Briefly, a single colony of P. fluorescens SBW25 was inoculated in LB medium and grew to the early stationary phase (OD600 = 0.4) at 23 °C with shaking at 160 rpm. Cells from 10 mL culture were washed three times with the chemotaxis medium designed for E. coli motility study (Hedblom & Adler, 1980) and re-suspended in 500 μL flagella staining buffer (Turner et al., 2000) containing 0.6 mg mL−1 Alexa Fluor 594 (Sigma) or 0.5 mg mL−1 fluorescein-5-isothiocyanate (AAT Bioquest, Inc., Sunnyvale, CA). Twenty-five microlitre of 1 M Sodium bicarbonate solution was added to adjust the pH to 7.8. The cell suspension was shaken for 1 h in dark at 23 °C. Free dye was removed by three exchanges of 10 mL chemotaxis medium after shaking for 5 min in dark. The cells were finally suspended in 1 mL staining buffer for observation.
Five microlitre cell suspension was loaded on a Superfrost Ultra Plus slide (Thermo Scientific, Gerhard Menzel GmbH, Braunschweig, Germany) and observed with an Axio Imager Z1 microscope (Carl Zeiss Microscopy GmbH, Jena, Germany) equipped with a 63 × oil objective lens. Images were taken within 5 min as described previously (Ping, 2010). Non-flagellated cells were not counted in data collection. The parameters of flagella were measured on images of stationary flagella either detached or on immobilized cells as previously described (Wang et al., 2012). The contour length (L) of flagella was calculated using formula
where N is the number of turns per filament, P is the pitch size, and D is the diameter. The sizes of cell bodies were determined on bright-field images taken on the same samples. The lower quartile of the collected data were regarded as the size of normal cells as published before (Ping et al., 2012).
Bacteria were cultivated in tryptone broth (Armstrong et al., 1967) to the early exponential phase (OD600 = 0.2) with shaking at 23 °C. Unstained and stained cells were concentrated five times in chemotaxis medium. After standing still on bench for 30 min with air exposure, 2 μL cell suspension were taken from the surface layer and filled into a sample well of a diagnostic slide (Thermo Scientific, Gerhard Menzel GmbH). Because oxygen depletion would handicap the bacterial motility, all movie clips were recorded within 5 min after putting on the coverslip. Movies were recorded at 18 frames per second using the Axio Imager Z1 microscope with 40 × oil objective lens and an AxioCam HSm high speed monochrome camera (Carl Zeiss Microscopy GmbH, Göttingen, Germany) within 5 min. Movies were analyzed with the AxioVision software provided by the manufacturer and ImageJ 1.37 (National Institutes of Health, Bethesda, MD). To calculate the full swimming speed, only runs longer than 50 μm and backups longer than 10 μm were analyzed. Cells spontaneously tethered to the slide surface by their flagella were recorded under same condition. Rotation events with an angular displacement less than π were omitted in analysis.
Influence of viscosity
Bacteria from an early exponential culture were resuspended in chemotaxis medium containing 0.1–1.0% methyl cellulose (M7140, M6385, and M0262; Sigma). It has been demonstrated that at the low concentrations, the solutions of these un-branched polymers are Newtonian (Berg & Turner, 1979). Viscosity of the media was measured at 23 °C with an AVS 350 capillary viscometer (Schott Geräte GmbH, Ludwigshafen, Germany). To study the swimming behavior, the microscope was focused to a position where uniform curvature of swimming paths was unobservable (> 100 μm away from the surfaces). Incomplete runs and backups were analyzed only if they were longer than 50 μm. When several runs and backups of the same bacteria were recorded, only the first and last runs and backups were measured to reduce bias. Those runs shorter than 10 μm and backups shorter than 5 μm were omitted. Chemotaxis medium without methylcellulose was used as control.
Results and discussion
Flagellation and cell size
Bacteria swim in the low Reynolds number regime of fluid dynamics (Berg, 1975; Ping, 2012). The cell dimension and flagellation are key parameters that determine output. The cells of P. fluorescens SBW25 are monotrichous as expected, but not in sensu stricto. More than 60% flagellated cells are with one flagellum (Fig. 1a inset). Among the 254 fluorescence-labeled cells examined, c. 38% carry more than two flagella. The average number of flagella per cell is 1.5 ± 1.1, lower than those of the other reported Pseudomonas species. The average number of flagella on Pseudomonas syringae pv. tabaci cells is 2.7 (Sigee & El-Masry, 1989; Kanda et al., 2011). Pseudomonas putida PRS2000 generally have five to seven flagella (Harwood et al., 1989). The number of flagella on P. fluorescens SBW25 cells is probably similar to that of P. aeruginosa, which often carry a single flagellum (E.P. Greenberg and R. Ramphal, pers. commun.).
The flagella of P. fluorescens SBW25 are right-handed helices with averagely 2.5 turns per filament. The flagellar contour length is 8.4 ± 1.3 μm. The pitch size and diameter are much larger than those reported for P. aeruginosa strain PAK, P. syringae pv glycinea, and P. syringae pv tabaci 6605 (Table 1). At physiological pH, the normal flagellar form of P. aeruginosa and P. syringae have been reported as left-handed (Fujii et al., 2008; Taguchi et al., 2008). However the left-handed curly flagella on E. coli mutants cannot generate enough swimming force (Wang et al., 2012). Whether the previously studied P. aeruginosa and P. syringae pathovars are motile needs further verification. On the other hand, the flagella of the swarmer cells of Caulobacter crescentus, another fresh water monotrichous bacterium, are right-handed (Koyasu & Shirakihara, 1984). The Caulobacter flagellum has even smaller pitch and diameter, comparing to P. fluorescens SBW25 (Table 1).
Table 1. Flagellar characters of Pseudmonas fluorescens SBW25 and other monotrichous bacteria
The cell body of P. fluorescens SBW25 is larger than that of E. coli. The average length of cell bodies of P. fluorescens SBW25 is 3.1 ± 0.8 μm and the diameter is 0.9 ± 0.1 μm. These results were in good agreement with the previous observations on other pseudomonads (Harwood et al., 1989; Sigee & El-Masry, 1989). The flagellar filaments of P. putida PRS2000 usually have two to three wavelength and are about two body lengths long (Harwood et al., 1989). Similar observation was obtained with transmission electron microscopy on P. syringae pv. tabaci strain Wolf and Foster 1917 (Sigee & El-Masry, 1989).
Sophisticated free-swimming behavior
Pseudmonas fluorescens SBW25 exhibited a sophisticated swimming behavior unknown to any other bacteria. Figure 1a shows the trajectory of a free-swimming cell. It can swiftly adjust the direction when swimming forward (runs) without retardation on speed. This behavior has been called ‘turns’ by Harwood et al. (1989). When they swim backward (backups), they sometimes follow the same path of the immediately preceding run (Between segment II and III), but often change the direction to an acute angle at transitions; with the next run deviated again from the backup path, a zig-zag trajectory was hence generated. Turn has only been observed in P. putida previously (Harwood et al., 1989; Duffy & Ford, 1997; Davis et al., 2011), and the zig-zag trajectory in P. citronellolis (Taylor & Koshland, 1974). In previous computer-aided analyses on P. putida swim, a bimodal distribution of the changing angle was observed, with frequency peaking at about 40° and 160°, respectively (Duffy & Ford, 1997; Davis et al., 2011). The obtuse angle would be resulted from turns, and the acute angle from reorientation at run-backup transitions.
Pseudmonas fluorescens SBW25 is one of the fastest swimmers in Pseudomonas. The average run speed of P. fluorescens SBW25 was observed as 77.6 μm s−1 and the maximum speed 102.0 μm s−1; the average backup speed is 18.0 μm s−1 and the maximum 22.4 μm s−1 (Fig. 1b). Backups are often short. Cells that run fast tend to backup fast (Fig. 1c). Pseudomonas syringae pv. tabaci 6605 was observed to swim at c. 50 μm s−1 (Kanda et al., 2011) or 83 μm s−1 (Taguchi et al., 2008). Pseudomonas aeruginosa ATCC 15692 swim at c. 60 μm s−1 (Conrad et al., 2011). The average swimming speed of P. putida was reported to be 44 μm s−1 with maximum at 75 μm s−1 (Harwood et al., 1989). Pseudomonas putida strain KT2440 swims even slower, with an averaged c. 20.9 μm s−1 and a maximum at 51.2 μm s−1 (Davis et al., 2011). None of the backup speed of other Pseudomonas species has been reported.
The swimming behavior of P. fluorescens SBW25 is much more sophisticated than the peritrichous rods. Besides the above mentioned runs, backups, turns, and run-backup reorientations, they can make a series of continuous rocking (Fig. 1d), or standstill for a while. In both cases, there is no net displacement. These two kinds of behavior will be referred as ‘flip’ and ‘hover’ respectively for convenience. The boxed area in Supporting Information, Movie S1 highlights a prolonged flip, while normal flips of P. fluorescens SBW25 were shorter and simpler (Fig. 1d). The bacteria must control their flagellar rotation very precisely to perform this kind of movement. When hovering still, the cells still rotated. Movie S2 showed a congenitally curved cell. The cell body rotated counterclockwise at 6.7 Hz. The flagellar propulsion force and the translational viscous drag load must be precisely balanced in this process.
Flagella and cell body rotation
The flagellum and cell body must rotate against each other to generate propulsion force (Berg, 2003). A right-handed flagellum pushes the cell forward when it rotates clockwise, and pull the cell backward when rotates counterclockwise. When fluorescence-labeled cells swam into the inspection field relatively slowly, this could be clearly observed. We also examined bacteria that spontaneously tethered to the glass slide surface (Fig. 2). The cells rotate alternatively to both directions equally well with a slight preference to clockwise direction on speed (P = 0.38, two-tailed paired Student's t-test) and duration (P = 0.09), confirming that flagella rotate clockwise during the runs. It has been reported that the cell body of P. citronellolis rotates chiefly in counterclockwise direction, and change to a clockwise rotation periodically in an isotropic medium (Taylor & Koshland, 1974). Our observation on P. fluorescens SBW25 is more similar to the observation on E. coli. When E. coli cells were tethered by a single flagellum, the cells rotate alternatively to both directions randomly at similar angular speed (Block et al., 1989).
In our experiments, the rotation speed of tethered P. fluorescens SBW25 cells towards clockwise directions was c. 2.4 Hz. The tethered polyhook mutant of E. coli rotate at 2–9 Hz (Silverman & Simon, 1974), and a fully energized E. coli cell spins at 10 Hz (Berg & Turner, 1993). Unlike E. coli, whose flagellar motors sit on lateral surface, the polar filaments of Pseudomonas must bend when the cell bodies rotate in a horizontal plane. The disalignment of the cell bodies with the flagellar axis would account for, at least partially, the observed slow speed. The peaks appearing immediately after direction switching testified this assumption: that likely results from a sudden release of the accumulated tension (Fig. 2a). Flagellum(a) tethered P. aeruginosa ATCC 15692 cells also spin towards both directions equally well, but with a lower angular velocity of c. 0.8 Hz (5 rad s−1; Conrad et al., 2011), probably due to additional thrust/hindrance from surplus flagella.
Circular motion near flat surface
When living in rhizo- and phyllosphere, bacteria unavoidably encounter liquid-solid interface, of which a flat solid surface is the simplest form (Ping, 2012). The presence of a solid surface in solution increases the viscosity of nearby fluid. This is termed wall effect (Ramia et al., 1993; Lauga et al., 2006; Ping, 2012). It is negligible in macroscopic system, but significant at the microscopic scale. Pseudmonas fluorescens SBW25 cells performed circular motion in runs when they swam near the slide surface (< 50 μm above the surface). When viewed from above and the solid surface is below, they appear to be rolling to the left (Fig. 3a). They swam in right-hand direction in backups. Near the surface, their average run speed decreases to c. 65 μm s−1 in circular motion (Fig. 3b). The swarmer cells of C. crescentus perform similar circular motion. They swim clockwise near the bottom surface in run when observed from below (Li et al., 2008; Ping, 2012). It has been reported that P. citronellolis swims in clockwise circles near bottom when viewed from above like E. coli (Taylor & Koshland, 1974; Berg, 2003), and hereby different from of P. fluorescens SBW25. Pseudomonas putida PRS2000 swim in a counterclockwise direction near a bottom surface when viewed from below, but whether it was in run or in backup was not clarified (Harwood et al., 1989). The circular trajectories of P. aeruginosa ATCC 15692 was also observed, however, the direction was not determined (Conrad et al., 2011).
Interestingly, P. fluorescens SBW25 cells turn much more frequently in circular motion than in free-swim. Kinks were observed on the circular trajectories (Fig. 3a). However, between two kinks the curvatures were always constant for a given bacterium. The median of the radius of these curvatures was 12.5 μm and the average 14.6 μm (Fig. 3c). The radius of the circular path of P. aeruginosa ATCC 15692 was reported to be c. 12 μm, in good agreement with our data (Conrad et al., 2011). The heterogeneity of viscous drag load originated from wall effect that experienced by the rotating cell body and flagellar helix pushes the swimming cell constantly off track and slows it down (Lauga et al., 2006). The radius of the circular path is determined by the physical dimension of the organism, the swimming velocity and the out-of-plane rotation rate (Ramia et al., 1993; Lauga et al., 2006). When these parameters are fixed, the curvature should remain unchanged.
Influence of isotropic viscosity
To test whether the high turn frequency in circular motion is triggered by the high viscous drag load on cells, chemotaxis media with different viscosities were used (Table 2). The backup frequency of P. fluorescens SBW25 was significantly reduced as viscosity increased (Table 2). At 19 centipoises (cP), backup was seldom observable. The run speed of P. fluorescens SBW25 did not show significant difference at 2 cP compared to the controls (Table 2), but the backup speed was slightly increased. On the other hand, the path-length of runs and backups at 2 cP were significantly longer than those at 1 cP (Table 2). As viscosity was further increased, the path-lengths of runs and backups decreased linearly until their minima were reached at 7 and 4 cP, respectively. The speed of runs and backups decreased linearly as well. At 19 cP, some runs and backups showed intermittent slowdown. We attribute this to the suppressed turns and backups.
Table 2. Influence of viscosity on the free-swimming behavior of Pseudmonas fluorescens SBW25
Relative amount of run (%)
Length of backup (μm)
Velocity (μm s−1)
Duration of run (s)
> 50 μm
15.9 ± 11.1
45.9 ± 12.2
18.5 ± 7.0
1.2 ± 0.6
18.3 ± 10.7
44.3 ± 12.1
19.3 ± 6.6
1.2 ± 0.5
11.1 ± 6.8
27.6 ± 9.7
8.3 ± 4.9
1.6 ± 0.9
13.4 ± 6.4
27.9 ± 9.3
9.6 ± 4.4
1.6 ± 0.7
17.8 ± 11.7
17.1 ± 7.6
7.5 ± 2.1
2.4 ± 1.3
Low viscosity is known to bestead bacterial swim (Keller, 1974). Pseudomonas aeruginosa and E. coli swim fastest at 2 cP (Schneider & Doetsch, 1974; Atsumi et al., 1996), but the speed of P. fluorescens SBW25 was unchanged at this viscosity. It is worth noting that E. coli cells are multiply flagellated, while P. fluorescens SBW25 is, dominantly, monoflagellated. That of P. aeruginosa is uncertain (E.P. Greeberg, pers. commun.). A mutant of Vibrio alginolyticus that only produces a single polar flagellum swim fastest at 1 cP, while the isogenic strain that produce multiple lateral flagella swim at maximum speed at 5 cP (Atsumi et al., 1996). Above 2 cP, the swimming speeds of P. aeruginosa and E. coli correlate with medium viscosity inversely (Greenberg & Canale-Parola, 1977; Atsumi et al., 1996). Pseudmonas fluorescens SBW25 behaved the same. Because the path-length and the velocity decreased proportionally, the durations of each run and backup actually remained unchanged.
The sophisticated swimming behavior of P. fluorescens SBW25 has not been observed in any other well-studied bacteria so far. It is known that V. alginolyticus flicks their flagellum upon resuming runs to change body orientation (Xie et al., 2011). If the turns, flips, and run-backup reorientations of Pseudomonas were operated with similar mechanism, Pseudomonas would have a very precise control on flagella movements. Furthermore, the flagella of bacteria like Vibrio can sense the surrounding viscosity to initiate lateral flagella biosynthesis (McCarter et al., 1988). Whether the flagella of Pseudomonas can also serve as a dynamometer is unclear. However, the flagellar motors of Vibrio are sodium-driven (McCarter et al., 1988; Xie et al., 2011), while the Pseudomonas motors are proton-driven (Kanda et al., 2011). Vibrio swim faster in backups than in runs (Magariyama et al., 2001). They might employ very different mechanisms.
The precise movement control of P. fluorescens SBW25 certainly has ecological advantages in the restricted geometry of soil and plant cavities. Intuitively, backup is helpful to prevent jam and turn, flip and hover to avoid collision. It has been observed that the turn frequency is significantly increased when P. aeruginosa swim in glass-bead column (Chen & Jin, 2011). The cell body spinning has been proposed to facilitate the release of tethered cells (Conrad et al., 2011). Other soil bacteria such as Sinorhizobium meliloti are known to adopt different strategies for the same goal: they use a combination of peritrichous and lophotrichous flagella to swim (Götz & Schmitt, 1987; Armitage & Schmitt, 1997). Although the right-handed flagellar bundle only rotates clockwise, they can slowdown, turn, or stop through motor control.
Monotrichous flagellaion, rather than peritrichous flagellaion is believed to be an adaption to the nutrient-scarce aquatic environment for fast swim with low energy cost (Ping, 2012). On several aspects, P. fluorescens SBW25 resembles more to the fresh-water bacterium C. crescentus than the enteric bacteria (Li et al., 2008; Ping, 2012). For free-swimming bacteria, translation at low Reynolds number is torque free and equals the drag load on the cell body given by Stoke's law:
here v is swimming velocity and r is the radius of a spherical cell. A typical E. coli cell is 2.5 μm long and 0.8 μm in diameter (Berg, 2003). The radius of a sphere with equal volume is c. 0.62 μm. Escherichia coli generally swims at 25 μm s−1 (Berg, 2003). The P. fluorescens SBW25 cells are slightly larger and the radius of the equivalent sphere is c. 0.75 μm. If we take the viscosity of the medium η as 1 cP like water, the propulsion force generated by a typical E. coli swimming cell and a P. fluorescens SBW25 cell can be calculated as c. 0.29 and 1.11 pN, respectively.
In the circular motion of E. coli, frequent reorientation was not observed, except that they might leave the surface after tumble (Frymier et al., 1995). Since the turn frequency of P. fluorescens SBW25 was not influenced by the viscosity of bulk media, it is possible that the symmetry breaking of viscous drag load due to wall effect is responsible for the phenomenon. Nevertheless, the isotropic viscosity has its own significance in rhizo- and phyllospheres, because plant secretion and decomposed organic matters would increase viscosity locally. Speeding up at low cP values and slowing down at high cP values as well as the suppression of turn and backup at high viscosity might all influence nutrient uptake and chemotaxis efficiency. In water-saturated rhizo- and phyllosphere, submerged surface is very common. The circular motion with low swimming speed might also enhance chemotaxis sensitivity and nutrient gain as proposed for Caulobacter (Li et al., 2008). At mean time, it mimics the behavior of foraging animals, though operated passively in bacteria, and might enhance the host seeking and invasion.
In accordance with the sophisticated swimming behavior, the chemotactic sensors in Pseudomonas are highly diverse. There are 26 receptor genes belonging to the methyl-accepting-chemotaxis proteins (MCPs) family on the genome of P. aeruginosa PAO1, while only four in E. coli (Kato et al., 2008). On the genome of P. fluorescens SBW25 (Accession no. NC_012660), 52 putative MCPs have been annotated. Loper et al. (2012) compared this genome with nine other genomes and found a tremendous ecological and physiological diversity within the P. fluorescens group. When comparative genomic analysis is combined with single-cell motility study in the future the ecological importance of these detector-propeller networks will begin to emerge.
This research is supported by the Max-Planck Society. We thank Christian Kost of MPI-CE Jena for providing the bacterial strain, who obtained the strain from Paul Rainey at New Zealand Institute for Advanced Study. We also thank Ewald Grosse-Wilde of MPI-CE Jena for assistance on microscopy, Marcus Franke and Yanze Ren of FSU Jena for assistance on viscosity measurements. We also thank E. Peter Greenberg of University of Washington and Reuben Ramphal of University of Florida for sharing their opinions about the flagellation of P. aeruginosa.