Bacterial motility - Wiley Online Library

38 downloads 5243 Views 368KB Size Report
Sep 7, 2005 - quorum sensing; swimming cost; flagella. FEMS Microbiol .... turbulence at a Reynolds number of 0.26, which would drive bundle dispersion. .... energy levels rather than specific chemicals in the external ..... It is hard to see how natural selection ..... standing of motility requires fixing boundaries, gradients.
MINI REVIEW

Bacterial motility: links to the environment and a driving force for microbial physics James G. Mitchell & Kazuhiro Kogure Marine Microbiology, Ocean Research Institute, University of Tokyo, Nakano-ku, Tokyo, Japan

Correspondence: James G. Mitchell, School of Biological Sciences, Flinders University, GPO 2100, Adelaide, SA 5001, Australia. Tel.: 1 618 8201 3684; fax: 1 618 8201 3015 e-mail: [email protected] Received 9 December 2004; revised 12 June 2005; accepted 23 June 2005. First published online 7 September 2005. doi:10.1111/j.1574-6941.2005.00003.x Editor: Peter Dunfield Keywords bacterial motility; chemotaxis; chemoreceptor; quorum sensing; swimming cost; flagella.

Abstract Bacterial motility was recognized 300 years ago. Throughout this history, research into motility has led to advances in microbiology and physics. Thirty years ago, this union helped to make run and tumble chemotaxis the paradigm for bacterial movement. This review highlights how this paradigm has expanded and changed, and emphasizes the following points. The absolute magnitude of swimming speed is ecologically important because it helps determine vulnerability to Brownian motion, sensitivity to gradients, the type of receptors used and the cost of moving, with some bacteria moving at 1 mm s 1. High costs for high speeds are offset by the benefit of resource translocation across submillimetre redox and other environmental gradients. Much of environmental chemotaxis appears adapted to respond to gradients of micrometres, rather than migrations of centimetres. In such gradients, control of ion pumps is particularly important. Motility, at least in the ocean, is highly intermittent and the speed is variable within a run. Subtleties in flagellar physics provide a variety of reorientation mechanisms. Finally, while careful physical analysis has contributed to our current understanding of bacterial movement, tactic bacteria are increasingly widely used as experimental and theoretical model systems in physics.

Introduction Swimming is the most immediately apparent behaviour of prokaryotes. It defined them as living entities in the early observations by van Leeuwenhoek (Ford, 1991). In the early 1800s, these observations of bacterial swimming contributed to the concept of a life force for inanimate objects. Motility and the thermal movement that became known as Brownian motion were not conceptually distinct. Einstein’s work separated motility from Brownian motion by identifying the latter as a quantifiable physical expression of temperature. However, 150 years later, motility and Brownian motion were merged again when Einstein’s quantitative description for Brownian motion was used to explain the constraints on Escherichia coli chemotaxis (Berg & Purcell, 1977; Berg, 1983). The papers reunited physics and bacterial motility and in the process laid the foundation for continuing insights in both disciplines. This review examines the recent developments in microbiology and physics that have expanded our concept of bacterial motility from a single, allFEMS Microbiol Ecol 55 (2006) 3–16

purpose strategy to a diverse repertoire of specific responses in bacterial swimming that find heuristic value beyond microbiology. Classical studies emphasized the qualitative importance of swimming for bacteria and presented them as the first microsensors of chemical gradients (Engelmann, 1883). The modern systematic study began with the work on chemotaxis by Adler (1966), Berg & Brown (1972) and Macnab & Koshland (1972). In these canonical observations, E. coli responded to an attractant gradient with 1-s swimming periods (runs) punctuated by tumbles in which cells randomly reoriented, producing a new swimming direction. Decreased frequency of tumbling in response to an attractant gradient resulted in migration up the gradient. The migration speed was a few percent of the swimming speed, which is characteristic of a biased random walk. The inefficiency results from the randomness of the tumbling and Brownian motion rotating the run path. The diffusion implicit in Brownian motion is at the heart of why bacteria move. A molecule moves across a 1-mm 2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

4

bacterial cell in 0.5 ms, but across 1 mm in 1000 s because the travel time increases as the square of the distance. Moving is advantageous when nutrients are limited as it takes cells away from competing bacteria. The function of movement, then, is to find new environments and the function of chemotaxis is to ensure those environments are high in nutrient or low in poison. Applying the run and tumble paradigm to nonenteric bacteria has met with mixed success. Rhizobia show run and tumble chemotaxis with 1- to 2-s run times and random reorientation by peritrichous flagella (G¨otz & Schmitt, 1987). However, for many bacteria that exist in environments where steep gradients are common, the run and tumble model is the starting reference for describing variations. The distinction between simple swimming and chemotaxis is significant. Bacteria can become highly motile without chemotaxis (Merrell et al., 2002) and one of the simplest variations is swimming speed. The current range of speeds for swimming in fluid is 1–1000 mm s 1 (Marwan et al., 1991; Fenchel & Thar, 2004) and both ends appear to represent physical limits. For bacteria slower than about 1 mm s 1, the value of moving may be lost as the nutrient diffuses faster than the bacterium can find it. The fastest speeds must encounter the physical limits of the maximum motor rotation rate and the length and number of flagella. Whether speed is measured as distance or body lengths per second is relevant for comparing speed among different-sized bacteria. Thus, a 25-mmlong bacterium swimming 600 mm s 1 moves 24 body lengths per second, whereas a 1-mm-long bacterium moving 450 mm s 1 moves 450 body lengths per second (Mitchell et al., 1995). The absolute value is used for calculating energy expenditure, comparing cell size and run length to diffusivity and gradient length, whereas the body-length normalized speed is more relevant for establishing and testing hypotheses on physical limits. Swimming speed and gradient length determine the minimum cell size for chemotaxis (Dusenbery, 1998; Mitchell, 2002). Bacteria have long been used as models of behaviour because of their simplicity compared with animals (Berg, 1975; Segall et al., 1982; Marques et al., 2002). The key parameters in bacterial movement are absolute speed, constancy of speed, turn angle, gradient length, cell size, receptor sensitivity and the extent to which the random walk can be biased. The values of these easily measurable parameters can be changed by the cells and the environment. This suggests considerable scope for variation in motility and chemotactic strategies. Exploring movement strategies usually requires resorting to computer modelling and laboratory experimentation. Fenchel (2002) points out that measuring bacterial swarming in the ocean is probably impossible, but that understanding it is important because all metabolically active 2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

J.G. Mitchell & K. Kogure

bacteria in the ocean seem to be motile. He goes on to point out that motility is probably even more useful in sediments and soils, where steep, complex and dynamic chemical gradients occur. The potential diversity and importance of motility in these systems is augmented by the recent theoretical and experimental demonstrations that at least some bacteria sense chemical gradients over their body lengths, in contrast to the standard paradigm that says they cannot (Berg & Brown, 1972; Dusenbery, 1998; Fenchel, 2002; Overmann & Schubert, 2002). The number of departures from the run and tumble chemotaxis paradigm is growing, including ‘helical klinotaxis’ where steering replaces tumbling, and the phototrophic consortia where epibionts are sensors and a central cell provides the flagellar propulsion (Purcell, 1977; Thar & K¨uhl, 2003). Some bacteria show emergent capabilities derived from group responses, for others new techniques needed to be developed and new chemical cues continue to be discovered. This review summarizes advances in motility grouped into methods and mechanics, individual behaviour, group behaviour and ecology. The sections are arranged to take the reader from the cell surface to the ecosystem scale. An additional section describes motility in modern physics to illustrate its use in the wider scientific environment.

Methods and mechanics Tools and techniques The modern study of bacterial chemotaxis began with the microcapillary technique (Adler, 1966) and the refinements continue. For example, methods have been developed that allow enumeration by microscopy to replace plate counts (Meyer et al., 2002). The result is that direct cell counts can be used on volumes of a few microlitres. This technique is useful for finding new attractants. New sophisticated techniques are also available for studying motility. Vigeant et al. (2002) used total internal reflection aqueous fluorescence microscopy to describe motility near surfaces. They found three ‘compartments’ relevant to motility: bulk fluid, nearsurface bulk fluid and near-surface constrained fluid. The compartment size varies with ionic concentration, but in general, cells less than 10 nm from a surface become ‘constrained’ and irreversibly attached to the surface. Cells slightly above 20 nm swim parallel to and may escape the surface. Cells greater than 10 mm away do not feel the surface. Where they are common, surfaces could be a reliable method of reorienting for bacteria. Cell shape contributes to this. Spherical cells have a torque towards the surface that induces spin but does not cause hydrodynamic entrapment. Rod-shaped cells have the same surface-induced torque, but also have a balancing form-induced drag. The two torques cause hydrodynamic entrapment and result in rod-shaped FEMS Microbiol Ecol 55 (2006) 3–16

5

Bacterial motility

cells swimming parallel to the surface (Vigeant et al., 2002). The aspect ratio of the cell determines whether the torques balance, so cell shape is one determinant of how motile cells react to surfaces. Chemla et al. (1999) used a superconduction quantum interference device to study bacterial magnetotaxis. Remarkably, they created a microscope of magnetic fields of similar strength to the magnetic field of the Earth. They measured an average flagellar rotation rate of 89.2  9.6 Hz, an average cell rotation rate of 25.9  0.2 Hz, and vibrational and precessional deviations of 5.5  0.71 and 7.0  0.41 on populations of 10–100 million cells. Using nonmotile cells, they were able to measure the rotational drag coefficient, the average cell size and distribution around that average. These measurements are impossible to make simultaneously on 10–100 million cells by light microscopy. The authors claim nonmagnetotactic motility can be analysed by attaching magnetic particles to cells.

Flagellar control New techniques are only one way to advance motility research. Flagellar shape and bundle stress are known to respond to environmental change (Calladine, 1978), but new varieties of flagellar control continue to be discovered. The complex flagellar filament of Rhizobium lupini is one new variant (Scharf, 2002). This consists of three flagellin subunits connected by interflagellin bonds that lock in a right-handed helical conformation. This is critical for flagella that only rotate clockwise and that tumble because of asynchronous deceleration of the bundled filaments. The filaments are stable from pH 4 to 9, but straighten at more extreme pH and at high viscosity. Scharf (2002) proposes that this response prevents motor stalling. It is unknown if locked flagella perform direct viscotaxis or pH taxis. However, changes in these conditions make tumbling more frequent. Orientation control by speed modulation of a single conformationally dynamic flagellum was described in Rhodobacter sphaeroides (Armitage et al., 1999). Initially, R. sphaeroides was thought to turn by Brownian motion (Armitage & Macnab, 1987), but this is now known to be incorrect for this species. It may be an unlikely strategy for any species because it is useful only for cells within a few hundredths of a micrometre of 0.75 mm (Mitchell, 2002). In any case, the observation of turning with conformationally dynamic flagella and the report that R. lupini reorients between 01 and 1801 with two to three filaments and between 01 to 901 with five filaments shows reorientation methods as a fruitful area of research. Modelling suggests that flagellar bundle integrity is influenced by individual flagella creating turbulent wakes (Trachtenberg et al., 2003). This is surprising because FEMS Microbiol Ecol 55 (2006) 3–16

bacteria and their flagella are the ultimate noninertial, laminar-flow-only devices. Usually, flow is laminar for Reynolds numbers less than 1 and fully turbulent over 2000. In contrast, R. lupini flagella may achieve turbulence at their surface during their maximum rotation rate (Trachtenberg et al., 2003). The sharp-edged flagella and the Archimedean-screw-like contours are claimed to produce turbulence at a Reynolds number of 0.26, which would drive bundle dispersion. Experimental verification of this seemingly unlikely theoretical work would alter microfluidics and cause a paradigm shift in fluid mechanics. The full implications of flagella-generated turbulence are unclear; but the consequences of speed variation are clear in some species. Rhodobacter sphaeroides shows a variety of swimming behaviours within and between individual cells (Armitage et al., 1999). Speeds were constant, oscillating or intermittent. The intermittent behaviour combines the other two behaviours, where speed is high with occasional sharp, short-duration drops. Reverse intermittency was not observed, but is possible and might be useful in chasing behaviour (Barbara & Mitchell, 2003). Stopping was immediate, occurring between two movie frames taken at approximately 25 frames s 1, or as a slow deceleration. Similarly, starting could be immediate or a slow acceleration, occurring over milliseconds to seconds. A single cell may use all combinations. Flagellar conformation changes over the same timescale as speed (Armitage et al., 1999). This correlative link is supported by the mechanistic consistency of flagellar changes that are a tight, high-amplitude coil that does not propel the cell, a helix that propels the cell and a straight form that propels the cell. Transition states exist that are curled at the end and straight at the base. The transition and coiled states reorient the cell and represent distinct rotation speeds. Thus, fine motor control is important to environmental bacteria.

Chemiosmosis and motor control The flagellar motor is 45 nm and embedded in the cytoplasmic membrane. Electrochemical gradients of H 1 or Na 1 generated across the cytoplasmic membrane drive the motor (Manson et al., 1977), making it a true molecular machine that converts electrochemical energy into mechanical work (Hirota & Imae, 1983). The motors of the commonly investigated bacteria Bacillus subtilis, E. coli, Salmonella typhimurium, Streptococcus sp. and R. sphaeroides are the H 1 -driven type, whereas alkalophilic Bacillus YN-1 (Hirota & Imae, 1983) and some marine bacteria, such as Vibrio alginolyticus, Vibrio parahaemolyticus and Vibrio cholerae have the Na 1 -driven type (Yorimitsu & Homma, 2001). Limited information is available on the distribution of H 1 and Na 1 -driven motors among natural bacterial populations. About 60% of speed in some marine bacteria is 2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

6

because of Na 1 -dependent motors (Asai et al., 2003). The alkaline pH of seawater makes establishing an inwardly directed proton gradient difficult. As an Na 1 gradient is less affected than an H 1 gradient by pH, it should be a distinct advantage in saline environments. Motor components continue to be investigated. For the H 1 motors, the integral membrane proteins MotA and MotB are essential for ion determination and torque generation. They form a transmembrane proton channel anchored to a peptidoglycan layer. The Na 1 -driven motor in V. parahaemolyticus has four integral membrane proteins, PomA, PomB, MotX and MotY. PomA and PomB are homologous to MotA and MotB, respectively. Asai et al. (2003) concluded that the transmembrane domains of PomA/B are essential for Na 1 motors and that MotX/Y play the essential role of stabilizing the attachment of the extracellular domain of PomB to the cell wall. The presence of Na 1 - and H 1 -driven motors in one cell was first found in V. parahaemolyticus (Atsumi et al., 1992). The single-sheathed polar flagellum used in a liquid medium is energized by Na 1 . The numerous lateral flagella are energized by an H 1 gradient. These latter flagella are synthesized when the cells are on the surface or in viscous environments. The result is a swarming morphotype, presumably used for movement along marine animals and other surfaces. Vibrio alginolyticus has a similar system (Kawagishi et al., 1995). The marine Halomonas sp. US201 and US172 also possess dual motors, but have only one type of flagella (Kita-Tsukamoto et al., 2004). Halomonas have been isolated from diverse environments, including deepsea hydrothermal vents (Kaye & Baross, 2004), polar waters (Reddy et al., 2003) and dairy food (Maoz et al., 2003). Although more work is necessary, the dual system in Halomonas may give them functional versatility. Laser dark-field microscopy shows that E. coli flagella rotate at up to 270 r s 1 (Kudo et al., 1990) and V. alginolyticus at up to 1700 r s 1 (Magariyama et al., 1994). With one revolution per 1000 ions, the motor requires 5% of cytoplasmic Na 1 per second (Magariyama et al., 1994), making an efficient Na 1 efflux indispensable. Except for the alkalophilic Bacillus YN-1, all bacteria possessing Na 1 motors have a primary sodium pump that establishes the Na 1 electrochemical gradient as an inwardly moving sodium motive force (smf) driven directly by the respiratory chain (Tokuda & Unemoto, 1981; Tokuda & Kogure, 1989; Hase et al., 2001). We assume that smf-dependent biological functions, such as flagellar rotation and the Na 1 -cotransport system, evolved with the primary pump systems. Swimming speeds are a linear function of the flagellar rotation speed for V. alginolyticus and S. typhimurium (Magariyama et al., 1995). However, considering that ‘Ovobacter propellens’ moves at 1 mm s 1 (Fenchel & Thar, 2004), the rotation rate is well in excess of 10 000 r s 1 or 2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

J.G. Mitchell & K. Kogure

swimming speed cannot be extrapolated from rotation rate in some species.

Interface to the environment Motor rotation can be controlled by chemicals that reduce the proton gradient across the membrane (Minamino et al., 2003). The high pH and low organic matter concentration of seawater may contribute to the high speeds of marine bacteria (Mitchell et al., 1995; Fenchel & Thar, 2004). Alternatively, the Aer gene products sense intracellular energy levels rather than specific chemicals in the external environment (Rebbapragada et al., 1997). This allows the energy available in the cell to be maximized through movement, without having to have an extensive detection and processing system for each metabolizable type of molecule. Similarly, sensitivity is enhanced by chemoreceptors that work in teams by forming clusters at the poles of the cell (Ames et al., 2002), improving sensitivity to signals. Receptor distribution was thought to be unimportant because gradient detection across a body length was believed impossible (Berg & Purcell, 1977). The grouped receptors were the first clue that body-length detection was possible. For small, free-swimming bacteria, spatial comparison over a body length is more sensitive than temporal comparison (Dusenbery, 1998). The theoretical lower size limits are 0.29 and 0.32 mm for spatial and temporal chemical detection. The currently unanswered challenge for experimentalists is to measure the lower size limits in real bacteria for spatial and temporal chemical detection. Similarly, the minimum detectable gradient length for temporal sensing was calculated as 100 mm (Dusenbery, 1998). However, it is now known that bacteria detect gradients of as little as 10 mm (Manson et al., 1977; Blackburn et al., 1998). Whether spatial or temporal sensing is more advantageous depends on the molecule and gradient length. There are established predictions for ammonium, iron and many other compounds (Dusenbery, 1998), but experimental confirmation is needed and would provide insight into sediment bacterial distributions and migrations. Confirmation of the above predictions has begun with experiments showing that one unidentified species senses a chemical gradient over its body length (Thar & K¨uhl, 2003). These vibroid-shaped bacteria are remarkable because they progress along their short axis. The movement is propellerlike, but the rotation is not driven by a central shaft, rather by flagella at the ends of the cell. Flagellar bundle rotation speed is proportional to the ambient oxygen concentration, which causes U-shaped tracks, a few hundred micrometres long and helical klinotaxis. This is true chemotaxis, where the swimming direction and gradient correlate. This may only be possible for bacteria large enough to be negligibly affected by Brownian motion, although there is a report of FEMS Microbiol Ecol 55 (2006) 3–16

7

Bacterial motility

small marine bacteria circling moving algal cells in what could be a version of helical klinotaxis (Barbara & Mitchell, 2003). In stable gradients, temporary attachment by mucous threads is used for maintaining position. This intermittent attachment presumably reduces the expense of motility and highlights the value of motility in finding unusual species in microniches. The behaviour was described before the cells were identified or cultured, but clearly the results provide an incentive to investigate other components of its microbiology as well as looking for other bacteria that sense gradients over their body length.

Individual behaviour Ultimately, the function of cell components needs to be viewed in terms of behaviour in individual bacteria. For E. coli, the mechanics of movement and chemotaxis are sufficiently well understood to support a book (Berg, 1983), but selective use of only parts of the motility machinery may produce movement strategies that are counterintuitive. For example, it is assumed that motility must be chemotactic to be useful. However, Merrell et al. (2002) showed that hyperinfectious cholera bacteria in the intestine are highly motile, but the chemotaxis genes are repressed, which uncouples motility and chemotaxis, reducing retention time in the intestine, and increasing the likelihood of the infection spreading.

Speed changes Uncoupling motility from chemotaxis is a simple alteration. For other species, the alteration and behavioural changes are complex. Photorhabdus temperata is mutalistic with pathogenic nematodes of the Heterorhabditis genus. The bacteria have primary and secondary forms that prefer oxic and anoxic environments. The primary form is motile in both environments, but the secondary form is motile under anoxic conditions (Hodgson et al., 2003), despite an earlier report that it was not (Givaudan et al., 1995). This may reflect differences in culture conditions, strain variation and history. When the secondary form is transferred from anoxic to oxic conditions it remains motile, indicating that, once initiated, motility is maintained in a cell (Hodgson et al., 2003). Chemotaxis and motility are inhibited above 35 1C. Cells are nonmotile without sodium, but potassium and magnesium substitute for sodium with speeds reduced marginally in the primary form and marginally by potassium and by about two-thirds for magnesium in the secondary form. This is at 75 mM. Between 100 and 200 mM concentrations of potassium and magnesium, the motility decreases (Hodgson et al., 2003). Change in speed with salt concentration is consistent with ion pumps driving motility (Mitchell & Barbara, 1999). A FEMS Microbiol Ecol 55 (2006) 3–16

uniform response might be expected for organic signal molecules that are useful as sources of energy and cellular building blocks. However, E. coli and, by inference, many other bacteria, have a biphasic response to signal molecules. Leucine is an attractant in a 0–5-mM gradient, but a repellent in a 0–500-mM gradient (Khan & Trentham, 2004). It is unknown whether an intermediate concentration change exists where there is no response. The biphasic response arises from double signals with the Tar receptor, signalling attraction, and the Tsr receptor, signalling repulsion. Biphasic responses are important to consider where motility is investigated in a large number of species (Johansen et al., 2002) under identical conditions because each species is likely to respond biphasically at different attractant concentrations. Johansen et al. (2002) measured motility in 84 species and strains of marine bacteria. Swimming speeds ranged from 11 to 75 mm s 1 and accelerations from 80 to 189 mm s 2. Run times ranged from 0.11 to 0.32 s. The extremes were from unsequenced isolates, but it was unreported whether they were recent isolates. Swimming speed decreases with increasing nutrient concentration (Mitchell et al., 1995; Khan & Trentham, 2004), and Johansen et al. (2002) used fullstrength Zobell’s medium (Zobell, 1946). The implications for interpretation of the results are unclear because nutrient sensitivity was not examined and the automatic tracking system only measured a subset of the possible bacterial speeds and turning angles. The work does show the diversity of motile marine bacteria. Methods for testing the limits of bacterial motility are needed if differences and extremes in motility are to be discovered. Niche position may be useful for the assessment of motility. Schmitt (2002) emphasizes that E. coli is a specialist, whereas Sinorhizobium meliloti is a generalist. Sinorhizobium meliloti shows a more diverse chemotaxis system than E. coli. Types of flagellar filaments and their rotation differ between species. Sinorhizobium meliloti has a stiff right-handed filament for viscous swimming. Escherichia coli has monomeric flagellin subunits, whereas S. meliloti has tetrameric flagellin subunits assembled as heterodimers. The latter creates three locking, helical ribbons causing flagella to only rotate clockwise and orientational control to be by rotation rate modulation. Thus, S. meliloti, R. spheroides and many marine bacteria (Mitchell et al., 1996) change speed with chemical signal. For S. meliloti, reorientation is as a result of rotating motors at different speeds. The speed control mechanisms are unknown.

Chemoattractants and receptors The importance of behavioural repertoire is becoming clear. Vibrio fischeri, a species symbiotic with the bobtail squid, is chemotactic to N-acetylneuraminic acid (NANA), 2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

8

nucleosides and nucleotides, but not to the individual components of these compounds (DeLoney et al., 2003). Chemotaxis and respiration of thymidine in this species shows the difficulty in estimating growth by thymidine incorporation. Chemotaxis to, and metabolism of, NANA is consistent with taxis to squid, as it is produced by squid surfaces. Nucleotides and nucleosides are released during the apoptosis of symbiosis. The concomitant complexity is indicated by 40 1 chemoreceptor genes. Similarly, many species are chemotactic to many aromatic compounds (Parales & Harwood, 2002). The mechanisms are unknown. The number of chemoreceptor genes are unknown for Flavimonas oryzihabitans, but this bacterium is chemotactic to gas, oil and hexadecane, and its chemoreceptors might be as diverse as those in V. fischeri (Lanfranconi et al., 2003). There was minimal microscopic description in this study. They were observed, ‘changing directions of swimming by a sudden short backward movement’. This describes run and reverse chemotaxis, possibly a new variant. Chemotaxis was also reported towards detergents and oxygen in the negative controls, which suggests the chemotaxis was not to an alkane, but to the energy state. Energy state sensing may be common in chemotaxis.

Response times Aer and Tsr receptors are independent but redundant and probably universal in containing a PAS domain (Stock, 1997). This emphasizes the importance of aerotaxis. There is a trade-off between sampling and sensitivity (Johansen et al., 2002). The limits for each are unclear, but experiments suggest that most bacteria are far from those limits. Tumble rates reach 5 s 1 (Barbara & Mitchell, 2003; Fenchel & Thar, 2004) and delayed responses in rotation rate of up to 100 s are reported (Chernova et al., 2003). This indicates longterm response dynamics and represents a new research area. Short-distance, rapid-motility responses are open for investigation since the finding that bacteria track swimming microalgae (Barbara & Mitchell, 2003). This is unusual in four ways: the bacteria remain within 10 mm of the alga; they move ahead of the microalga; they turn in the direction that the microalga is travelling; and the bacteria accelerates and changes direction with the microalga, sometimes using a circular orbit pattern without distinct turns. Directionspecific reversing and circular orbits are new behaviours and indicate new control mechanisms. There appear to be at least seven distinct motility patterns. In addition to the classical run and tumble chemotaxis, bacteria are capable of run and reverse, steering, localizing, tracking, orbiting, and run and stop (Fig. 1). Steering and localization are the only nonmigratory behaviours. Reorientation is by tumbling, reversing or steering. 2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

J.G. Mitchell & K. Kogure

f

a

b c

g

d e

h

Fig. 1. Movement patterns for motile bacteria. (a) Run and tumble random walk with a bias towards the right. (b) Run and stop random walk with a bias towards the right. The dotted line indicates that this method, while previously attributed to Rhodobacter spheroides, is currently not reported for any bacterial species. (c and d) Run and reverse chemotaxes with biases to the right for marine bacteria (Mitchell et al., 1995) and (e) freshwater bacteria (Mitchell et al., 1991). (f) Run and arc, steering or, more formally, helical klinotaxis, with the ideal gradient concentration a line/plane across the centre. (g) Run and reverse used for localizing around a point source. (h) Run and reverse tracking followed by orbiting of bacteria around a single algal cell. Increasing dark grey shading indicates increasing attractant concentration.

The energetic cost of individual movement patterns can be calculated. Combining size, run length and the minimum chemotactic cost for each movement pattern shows where those patterns are most cost effective (Fig. 2) (Mitchell, 2002). The model indicates that all moving organisms follow one universal law (Fig. 3). Confirmation of this model requires the experimental determination of the energetic cost of bacterial movement.

Collective behaviour Group behaviour, in the broadest sense of bacteria moving together, has been recognized since Engelmann (1883) and Adler (1966). Recent work on group behaviour reveals fundamental insights and diversity in bacterial interactions. Some effects are simple, with 50 mM calcium salts disrupting Pseudomonas putida and Pseudomonas flourescens swarming, but KCl and NaCl having little effect up to 500 mM (Sakai et al., 2003). However, group behaviour appears most complex in quorum systems. Spontaneous hyperswimmer mutants of V. fischeri were poor at initiating symbiosis in bobtail squid (Millikan & Ruby, 2002), despite their faster swimming, probably because of hyperflagellation. Hs strains had up to 16 flagella per cell whereas the wild type had two, the minimum number for dissociative bundle reorientation. The hyperswimmer strains are believed to be defective (Millikan & Ruby, 2002), but the two distinct populations in the squid’s symbiosis sacks might mean that the FEMS Microbiol Ecol 55 (2006) 3–16

9

Bacterial motility

1 E+6 1 E+05

energy use (cal/g/km)

1 E+5

run length (µm)

1 E+4 1 E+3

a

1 E+2

s

1 E+1

tumble t

reverse

1 E+0

y = 5x 1 E+03

1 E+01

1 E-1 1 E-01 1 E-16

1 E-2 0

1

2

3

4

1 E-12

5

1 E-08

1 E-04

1 E+00

1 E+04

mass (grams)

cell radius (µm) Fig. 2. Envelopes for possible run lengths. The main envelopes are shaded and bounded by the maximum and minimum for the different reorientation methods. The letters are a for arc, s for stop and t for track. The tumble and reverse envelopes were calculated at the lower and upper limits. The a, s and t envelopes are estimated placements and shapes, and may overlap or extend beyond other envelopes. The thick black line at the bottom of the envelope is the point at which the run length equals the cell radius. Values below this line imply chemosensing over the body length, a possibility in some situations. After Mitchell (2002).

hyperswimmers are a third population (Visick & McFallNgai, 2000). There are two reasons why hyperswimmer bacteria may be distinct strains. First, the spontaneous hyperswimmer forms did not revert to wild type. Second, at about 100 mm s 1, the hyperswimmer was 150% faster than the wild type and all three spontaneous mutant types were within 2% of each other in speed (Millikan & Ruby, 2002). Further testing on the hyperswimmer strain is necessary. Although the three strains were identical in their movement, they did show differences in the formation of mucoid colonies, the inability to hemagglutinate red blood cells and the inability to produce light. The speed of the hyperswimmer strains came at a cost: they colonized slowly, initially losing to the wild type, but catching them in 48 h. The hyperswimmer strains were apparently unable to migrate through the squid’s pores. Group aggregation is necessary for pore entry. The hyperswimmer aggregations were rare and small, about 15 cells, whereas the wild-type aggregations were hundreds of cells. This is consistent with the work that is discussed below (Park et al., 2003a), showing that groups have a mechanism for collapsing into a pore that is not available to individuals or small clusters. The strains are distinct ecotypes, probably caused by a single change, such as flagellin overexpression. The overexpression of flagellin consumes energy so that it is outgrown by the wild type, producing a natural Lenski-type competition experiment (Lenski et al., 1991). The work on V. fischeri has yielded insight into the dynamics of symbiosis and is complemented by predictive FEMS Microbiol Ecol 55 (2006) 3–16

Fig. 3. Prokaryotes indicating a possible universal law for movement of organisms. The relationship between the mass of an organism and its energy use for movement. The solid circles and lines in the upper left are calculated costs for prokaryotes based on the literature. The dashed line is conjectural and the open circle is conjectural for the cost of the largest known motile bacteria. The shaded area is the transition to turbulence that Schmidt–Nielsen predicted would cause different allometric energy equations to exist for microscopic and macroscopic organisms. The one equation for both is shown in the centre. The four points for fish are from empirical measurements. There are no empirical determinations for energy use in prokaryotes, but they are clearly needed. After Mitchell (2002).

models. Mazzag et al. (2003) used 12 measured parameters in Azospirillum brasilense movement to construct a model using 16 variables. This model produced bands of similar position and width to their experiments, despite using a constant velocity of 40 mm s 1 in the model, while the bacteria in the experiment swam 18 mm s 1 outside of the band and 49 mm s 1 inside the band. If constant velocity results in band formation, the variable velocity may indicate an energy saving mechanism. If the goal was to stay in the band, then a chemokinetic response, where bacteria slow down in the band and speed up outside of the band would be expected. The observed behaviour is best adapted for traversing the band quickly. It is hard to see how natural selection would favour spending the least time at the ‘optimal’ concentration. Instead it may select for the steepest gradient. Mazzag et al. (2003) show that an oxygen gradient has six distinct regions in 1 mm. These are areas where: (1) the oxygen is high and there are no bacteria; (2) the oxygen is decreasing, but is high enough so that bacteria are reversing; (3) the oxygen is ‘optimal’ and bacteria are at maximum speed; (4) the oxygen is low and most bacteria are reversing; (5) the oxygen is undetectable with background concentrations of bacteria and (6) the oxygen is zero and bacterial concentrations are background. The presence of particles, multiple species and multiple signal compounds may produce many more regions and more complex chemical and species distributions over millimetres, but this remains 2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

10

J.G. Mitchell & K. Kogure

untested. This work allows for prediction of the distance between the meniscus and band (Mazzag et al., 2003), and so could show how gradients group bacteria. Brenner et al. (1998) investigated motility-driven pattern formation by looking for long-range interactions in earlier work (Budrene & Berg, 1995). The results provide technical and conceptual insights. For example, the swarm-ring velocity decreases as the attractant concentration increases (Brenner et al., 1998). This is consistent with previous work (Mitchell et al., 1995). Four pathways leading to aggregation were found: (1) stable continuous expansion; (2) stable to a concentration threshold; (3) spontaneously unstable and (4) increasingly unstable. The pathway followed depends on motility parameters, the primary attractant concentration and the presence of secondary attractants. Aggregates act as if they have zero, one, two or three dimensions. The significance of this for environmental microbiology is that there are two possible origins for an aggregation: (1) is de novo creation in response to a gradient (Mazzag et al., 2003) or (2) it is the remnant of a past interaction aggregation (Brenner et al., 1998). Specifically, bacterial aggregations are inherently three-dimensional at the start and collapse sequentially into lower dimensions (Brenner et al., 1998). Only the lower dimensions can be accurately modelled. These are two new mechanisms (Brenner et al., 1998; Mazzag et al., 2003) for the band migration described and modelled in the 1960s and 1970s (Fig. 4a). One model emphasizes environmental spatial structure (Mazzag et al., 2003), whereas the other emphasizes chemotaxis (Brenner et al., 1998). Park et al. (2003a) combine spatial structure and chemotactic behaviour to describe a system analogous to quorum sensing in E. coli that is based on glycine and could be a general system for all chemotactic bacteria. Escherichia coli shows complex cluster dynamics in a random maze of 100-m m-long walls. The maze was a photolithographic simulation of complex topologies, such as sediments. The accumulation

a

b

2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

is a two-step process. First, ‘travelling waves’ or, in the parlance of microbiology, chemotactic bands form (Fig. 4a). The behaviour of the group is described with wave physics. The second step occurs in complex topologies, where the wave collapses into a confining topology (Fig. 4). The collapse is not because of a chemoattractant, but is the result of nutrient depletion and subsequent excretion of glycine (Park et al., 2003a). This suggests that bacteria under stress seek each other and confinement. Bacteria can cluster into local populations on agar plates at high succinate concentrations (Budrene & Berg, 1995). Park et al. (2003a) found a similar clustering or collapse as described above. The tsr gene controls this behaviour as shown by the loss of clustering on tsr deletion, but its retention upon tar or aer deletions. Additionally, 1-mM L-serine eliminated clustering, but 1-mM L-asparatate did not, again indicating tsr control. Chloramphenicol also stopped clustering, showing the importance of gene expression and growth. Stationary phase cells had no glycine excretion and no collective behaviour. During clustering, the walls act as wave guides, or in this case collective behaviour guides (Park et al., 2003a). The microbiological significance is that individual bacteria are not specifically seeking out small spaces, but only as a group can and does this behaviour emerge. A significant implication of the paper is that traditional homoserine lactone quorum sensing systems are special cases of a general phenomenon of communication with amino acids. This highlights the importance of motility for discovery in microbiology. The glycine clustering is a nonlinear, positive feedback system. So, aggregation occurs when other models predict dispersal. However, the phenomenon is transient with a 22-min time constant (Park et al., 2003a). The smaller the ratio of the opening area to the volume, the lower the critical bacterial concentration for collapse. Thus, the pore geometry influences and might control which bacteria enter the chamber.

c

d

Fig. 4. Examples of collective behaviour. (a) A band of bacteria creating a gradient by consuming attractant (grey). (b) Bacteria undergoing a quorum sensing collapse through excretion and uptake of attractant. (c) Bacterial group collapse into a dead end. (d) Heterogeneity within a group of bacteria collapsed into a quorum attractant.

FEMS Microbiol Ecol 55 (2006) 3–16

11

Bacterial motility

The band dynamics in a maze do not end with the collapse, as these have internal dynamic structure (Park et al., 2003b). This structure has not been studied in detail, but clusters of 10 mm indicate the bacteria are not operating at the limit of their clustering ability when they invade spaces such as the light organs of squid. This might explain why there are distinct populations at distinct locations within the light organs. This is an area ripe for investigation.

Ecology of movement Individual and group swimming make for neat experiments. However, the prevalence and range of behaviour in the environment is still uncertain. Environmental nutrient concentrations are invariably low and sporadic. Bacteria can be modelled ‘almost perfectly’ for motility, but this is based largely on cultures (Murray & Jumars, 2002). Motility is a foraging strategy for overcoming crowding in sediment, but diffusion means crowding occurs when bacteria cover 0.1% of a surface. In sediments, attached bacteria can overcome nutrient limitation by excreting proteolytic enzymes (Vetter et al., 1998). However, group behaviour at densities that cause nutrient limitation helps in finding nutrient sources. Models of group behaviour provide insight into optimization problems involving online-distributed and cooperative-computing challenges (Liu & Passino, 2002). Banding produces social foraging, where multiple individuals respond to gradients with a low signal-to-noise ratio. Bacteria are the simplest of social foragers and, with local clonal populations not being competitors, the selective advantage is high. Sediment grain size alters the efficiency of migration, decreasing it 20-fold as grain size is reduced from 800 to 80 m m (Barton & Ford, 1995). Experiments that mimic the smallest gaps relevant to motility find that E. coli in a 6-m m-wide capillary swim in one direction, unable to turn because of their flagella (Liu & Papadopoulos, 1995). In a 3mm-wide capillary, geometric restriction precludes passing as well as turning (Liu et al., 1996). The result is two- and three-cell clusters that form and are maintained. The clusters moved 15 instead of 18 mm s 1, but gained the advantage of direction reversal in the capillary. A pair of cells is the smallest possible group and direction reversal is among the simplest of behaviours, but pairing for reversing could be a necessary and widespread method for exploring restricted environments. Escherichia coli was used (Liu et al., 1996), so the prevalence of paired-cell behaviour as a strategy compared with turning by altering rotation rate is unknown. In confined spaces, for chamber diameters below 6 mm, movement patterns switch from a diffusive to a propagative wave (Chen et al., 1998). The length of the flagella and cell determines the cut-off size. This suggests that porosity selects for flagella length, flagella number and cell size. FEMS Microbiol Ecol 55 (2006) 3–16

Clean walls make glass microcapillaries an unrealistic confined environment. Modelling motility in a capillary with sticky walls indicates that bacterial distributions are controlled by the ratio of the square of the channel radius to the motility-driven diffusivity (Bonilla & Cushman, 2002). Four underlying mechanisms influence the bacterial distributions in these circumstances: bacteria–wall interactions at tens of nanometres, hydrodynamic interactions up to a few bacterial radii, and convection and motility above tens of micrometres. The dynamics become nonlinear with sticky walls and thus difficult or impossible to predict. The unpredictability makes empirical studies important at the micro- and macro-scales (Mitchell et al., 1991; Barbara & Mitchell, 1996). Walls constrain movement, but stabilize gradients, damp directional and turbulent flow, and so enhance the effectiveness of chemotaxis. Where turbulence dominates movement and gradients are transient, the research focuses on the detection and response times of cells. Kirboe & Jackson (2001) model run and tumble vs. run and reverse search strategies for bacteria chasing 0.02–1.5-cm marine snow particles. These particles are rare, at a few per cubic metre, but are important for nutrient generation and habitat heterogeneity. How long a bacterium can stay in the chemical plume behind a particle depends on physical constraints, search pattern, chemical sensitivity, signal integration time, swimming speed and initial position relative to the particle (Kirboe & Jackson, 2001). The plume was assumed to have a mean total amino acid concentration of 30 nM. Zero was assumed for the background concentration. The swimming speeds chosen were 10 and 100 mm s 1. The shortest run length with the highest sensitivity produced the greatest rate of particle colonization. However, only bacteria moving 100 mm s 1 enhanced their plume residence time. The resulting nutrient enhancement of two- to three-fold was considered modest (Kirboe & Jackson, 2001). However, a 1% growth increase capacity gives a significant competitive advantage (Lenski et al., 1991), so even a factor of 2 probably has substantial selective and ecological significance. The kind of search strategy made no difference to the rate of colonization or nutrient enhancement. This is at odds with other models and experimental results (Manson et al., 1977; Luchsinger et al., 1999), but is probably because of differences in plume size and the detailed implementation of the models and experiments. The uptake enhancement depended inversely on the marine snow size and was a 20-fold enhancement for a 0.02-cm-radius particle, but a two-fold enhancement for a 1.5-cm-radius particle (Kirboe & Jackson, 2001), whereas previous work considered particles that were 10-fold smaller (Manson et al., 1977; Luchsinger et al., 1999). The conclusion is that chemotaxis potentially increases bacterial growth rates from 0.1 to 0.5 day 1 to 1 to 10 day 1 (Kirboe & Jackson, 2001). 2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

12

For chasing plumes to be advantageous they must be encountered frequently enough to have an ecological impact or result in colonization. Predictions of colonization rates of marine aggregates based on bacterial-motility parameters and search patterns show that motile bacteria rapidly colonize aggregates whereas nonmotile bacteria do not (Kirboe et al., 2002). Flow enhances colonization. Chemotactic strains preferentially colonize organic-rich aggregates. The work indicates that bacteria spend about 3 h on an aggregate, making ‘visit’ a more accurate word than ‘colonization’. The encounter rate is less than once per day, but in the model this is sufficient to explain all of their food requirements. The field observations were markedly different from the models, and indicated that competition and predators may be more important than colonization for influencing marine bacterial population dynamics. A confounding factor in bacterial motility present in the environment, but absent in cultures initially is nutrient limitation. For bacteria in the environment, motility is energetically expensive (Mitchell, 2002) and as such should be used sparingly. Grossart et al. (2001) addressed this issue by carrying out a 10-month study of the percent of motile bacteria at the Scripps pier. The range of motility was 5–25% in autumn and winter, and 40–70% in spring and summer. Bacteria did not swim continuously or at a constant speed. Many bacteria swam less than 20% of the time. Motility also showed a distinct diel pattern, apparently only correlated with variations in the concentration of particulate organic carbon. The fraction of motile bacteria increased sharply at the end of blooms, and increased motility was associated with increased colonization of living and dead surfaces. Removing the particulate matter decreased the percent of motile cells. Small cells (o 0.4 mm) had the lowest motile fraction. Adding nutrients caused already motile bacteria to swim for longer periods, but did not increase the motile fraction. Comparative experiments showed that 10% of cells were positive for the respiration indicator 5-cyano-2,3-ditolyl tetrazolium chloride (CTC), whereas 60% were motile. This suggests that CTC is a poor indicator of cell viability. The fraction that were motile was also higher than the fraction of cells with a high DNA content, providing support that at least some cells with low DNA concentrations are alive and functioning. Percent flagellation could not be assessed because the NanoOrange method (Molecular Probes, Eugene, OR) showed no flagellated bacteria in any of the samples, despite extensive motility. This indicates that the method does not work for all environments and supports the idea that nonantibody methods of flagella detection are unreliable. Overall, the study indicates that most marine bacteria are active or capable of being active. The results demonstrate that, beyond behaviour, motility is useful as a check on CTC, tritiated thymidine incorporation, high2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

J.G. Mitchell & K. Kogure

DNA cells measured by flow cytometry and ATP as indicators of bacterial metabolic state. Seasonal and bulk changes in motility and activity, such as those examined by Grossart et al. (2001) are easily predicted, modelled and the causes understood. The utility of short-term changes can be surprising and it can be a challenge to assess their importance in microbial ecosystems. At what is probably near the extreme for short-term interactions, Blackburn et al. (1998) looked at the role of chemotaxis in microscale patches of nutrients. Their motivation was in defining some of the largely unknown natural targets of bacterial chemotaxis. They found that bacteria swarmed around a variety of locations, including around the material expelled from a ciliate vacuole after the ciliate had swum away, creating a patch with no source at its centre. Swarms and patches were transient, both lasting about 10 min; however, the nutrient gain for chemotactic bacteria was a factor of 2 over nonmotile bacteria at a total nutrient content for a patch of 1 pmol. Thus, microscale patches are a source of nutrients. What is still not known is the size distribution of patches in most environments and the critical size and concentration limits above and below which chemotaxis is no longer advantageous in uptake and becomes a liability. The ecological significance of the interaction is that patches are transient resources that should stimulate competitive foraging and increase nutrient transfer rates.

Motility-driven physics For microbiologists, the focus of research is often the response of a cell, group behaviour or use of flagella and receptors. From a more distant view, bacterial chemotaxis is wondrous in its compactness, effectiveness and diversity, and has become a standard measure and model of complex, moving systems in physics. Physicists have linked bacterial movement to nonliving systems ranging from robots to galaxies. Symmetry is a fundamental concept of modern physics. An object is symmetric if an operation, e.g. rotating a square 901, leaves it unchanged in appearance. A rotation of say 661 breaks the symmetry. Much of the interest in chemotaxis by physicists has arisen since Berg (1996) described symmetry breaking in bacterial motility. The establishment and breaking of symmetry occurs on three levels in bacteria, the need for cyclic, propagating wave motion and not reciprocal motion, the generation of rotary motion to produce the waves that propagate down the flagella and pattern formation in group behaviour, such as a ring on a Petri dish. This last symmetry breaks when strong attractants lead to Budrene patterns (Budrene & Berg, 1995). Behaviour is often considered too complex to describe by reductionist rules or by applying physical laws. However, bacterial FEMS Microbiol Ecol 55 (2006) 3–16

13

Bacterial motility

behaviour permits prediction of not only simple linear responses, such as migration, but more complex pattern formations as well. Symmetry breaking was implicit in Luchsinger et al. (1999) on the rotation of motile bacteria in a turbulent ocean near a microalgal surface, in which passing bacteria rotated slightly, breaking symmetry giving run and reverse chemotaxis a 120-fold uptake advantage over a run and tumble strategy. Bacteria-flow interaction is relevant away from surfaces. Bearon (2003) showed that chemotaxis minimizes dispersion in the direction of taxis. However, the chemotactic crossing of streamlines increases dispersion perpendicular to the signal. This may help explain why aquatic bacteria often exhibit run-and-reverse chemotaxis, whereas intestine and soil bacteria often exhibit run-and-tumble chemotaxis. Some physicists use chemotaxis data to test ideas, such as how to control externally a spatially extensive, nonlinear system (Wisdom, 2003). Controlling the transient dynamics of the external environment gives an experimental solution to a mathematical optimization problem, which may be useful in drug development. Bacterial chemotactic strategies are used in a robot that finds volatile chemical sources (Marques et al., 2002). With this strategy, the robot took 324 s to produce a 65% probability of finding an ethanol source. A dual sensor, silkworm algorithm took 97 s to produce an 80% probability of finding the ethanol source. The bacteria-based algorithm was inferior with the run-andtumble strategy, but indicates the influence of bacterial chemotaxis beyond microbiology. Thus, for some physicists, bacterial chemotaxis has become a standard system for extracting principles and insights. Some uses of motility seem ridiculous. Wisdom (2003) used low Reynolds number arguments to show movement on the curved space–time manifold of Einstein’s relativity. Bacteria were an example, being the self-propelled objects with the least inertia. The paper weakly links relativity and motility. The forward motion is 10 20 times the body length (Sire & Chavanis, 2002), but the point is that motility is used to test the limits of cosmic effects. Similarly, chemotaxis is linked to string theory (Chavanis, 2003), but the intricate mathematics makes description and interpretation beyond the scope of this review. Physicists also link chemotaxis to cosmic dimensions, pointing out that chemotaxis is a long-range interaction and so dynamically similar to gravitational attraction (Chavanis, 2003). Classical diffusion fails both systems, as the stable equilibrium is not the most mixed state. Stellar dynamics shows that chemotaxis stabilizes far from equilibrium and is particularly appropriate for environmental chemotaxis, where signals, diffusion and motility are transient and heterogeneous (Chavanis, 2003). However, precise understanding of motility requires fixing boundaries, gradients FEMS Microbiol Ecol 55 (2006) 3–16

and fluxes. These are well defined for stellar systems, but are only beginning to be described in microbial ecosystems.

Conclusions The diversity and subtlety in the structures and behaviours associated with motility are increasing. Fenchel (2002) argued that microbial responses in the environment are not stereotypical. This is true when the stereotype is E. coli runand-tumble chemotaxis, but it was described first and there were no alternative observations. Repertorial diversity is now found in individual species and across groups. Even E. coli shows nonstereotypical behaviour (Park et al., 2003a). The response of a bacterium depends on environmental conditions, metabolic state and the history of the cell. The full behavioural diversity has yet to be fully elucidated. Motility is useful on two scales: locally, environmental information is accumulated over a single run; on the scale of entire gradients, the local process is iterated and combined for exploration. These two scales are well documented in the laboratory, but there is little in situ observation. Opportunities abound in the laboratory and the environment for studying attractant behaviour near the cell surface and studying individual and collective behaviour across long distances and times. Motility provided one of the first clues to the dynamic and complex nature of the microscopic world. It inspired the ideas of life force and, in the self-corrective process of scientific progress, led to quantitative descriptions of Brownian motion and chemotaxis (Berg, 1983). The movement of bacteria continues to inform our understanding of the intricacies of the microscopic world and act as a foil and challenge for physics.

Acknowledgements We wish to thank B. Kranz for suggesting the review and critical reading of the manuscript. This work was supported by the University of Tokyo, Flinders University and the Australian Research Council.

References Adler J (1966) Chemoreceptors in bacteria. Science 166: 1588–1597. Ames P, Studdert CA, Reiser RH & Parkinson JS (2002) Collaborative signaling by mixed chemoreceptor teams in Escherichia coli. Proc Natl Acad Sci USA 99: 7060–7065. Armitage JP & Macnab RM (1987) Unidirectional, intermittent rotation of the flagellum of Rhodobacter sphaeroides. J Bacteriol 169: 514–518.

2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

14

Armitage JP, Pitta TP, Vigeant MA-S, Packer HL & Ford RM (1999) Transformations in flagellar structure of Rhodobacter sphaeroides and possible relationship to changes in swimming speed. J Bacteriol 181: 4825–4833. Asai Y, Yakushi T, Kawagishi I & Homma M (2003) Ion-coupling determinants of Na 1 -driven and H 1 -driven flagellar motors. J Mol Biol 327: 453–463. Atsumi T, McCarter L & Imae Y (1992) Polar and lateral flagellar motors of marine Vibrio are driven by different ion-motive forces. Nature 355: 182–184. Barbara G & Mitchell JG (1996) Microlaminations of motile bacteria in microbial mats. Appl Environ Microbiol 62: 3985–3990. Barbara G & Mitchell JG (2003) Bacterial tracking of motile algae. FEMS Microbiol Ecol 44: 79–87. Barton JW & Ford RM (1995) Determination of effective transport coefficients for bacterial migration in sand columns. Appl Environ Microbiol 61: 3329–3335. Bearon RN (2003) An extension of generalized Taylor dispersion in unbounded homogeneous shear flows to run-and-tumble chemotactic bacteria. Phys Fluids 15: 1552–1563. Berg HC (1975) Bacterial behaviour. Nature 254: 389–392. Berg HC (1983) Random Walks in Biology. Princeton University Press, Princeton, NJ. Berg HC (1996) Symmetries in bacterial motility. Proc Natl Acad Sci USA 93: 14 225–14 228. Berg HC & Brown DA (1972) Chemotaxis in Escherichia coli analysed by three-dimensional tracking. Nature 239: 500–504. Berg HC & Purcell EM (1977) Physics of chemoreception. Biophys J 20: 193–219. Blackburn N, Fenchel T & Mitchell J (1998) Microscale nutrient patches in planktonic habitats shown by chemotactic bacteria. Science 282: 2254–2256. Bonilla FA & Cushman JH (2002) Effect of a-stable sorptive waiting times on microbial transport in microflow cells. Phys Rev E 66: 031915. Brenner MP, Levitov LS & Budrene EO (1998) Physical mechanisms for chemotactic pattern formation by bacteria. Biophys J 74: 1677–1693. Budrene EO & Berg H (1995) Dynamics of formation of symmetrical patterns by chemotactic bacteria. Nature 376: 49–53. Calladine CR (1978) Change in waveform in bacterial flagella: the role of mechanics at the molecular level. J Mol Biol 118: 457–479. Chavanis P-H (2003) Generalized thermodynamics and Fokker–Planck equations: applications to stellar dynamics and two-dimensional turbulence. Phys Rev E 68: 036108. Chemla YR, Grossman HL, Lee TS, Clarke J, Adamkiewicz M & Buchanan BB (1999) A new study of bacterial motion: superconducting quantum interference device microscopy of magnetotactic bacteria. Biophys J 76: 3323–3330. Chen KC, Ford RM & Cummings PT (1998) Mathematical models for motile bacterial transport in cylindrical tubes. J Theor Biol 195: 481–504.

2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

J.G. Mitchell & K. Kogure

Chernova AA, Armitage JP, Packer HL & Maini PK (2003) Response kinetics of tethered bacteria to stepwise changes in nutrient concentration. Biosystems 71: 51–59. DeLoney CR, Wolfe AJ & Visick KL (2003) Chemoattraction of Vibrio fischeri to serine, nucleosides, and N-acetylneuraminic acid, a component of squid light-organ mucus. Appl Environ Microbiol 69: 7527–7530. Dusenbery DB (1998) Spatial sensing of stimulus gradients can be superior to temporal sensing for free-swimming bacteria. Biophys J 74: 2272–2277. Engelmann TW (1883) Bakterium photometricum. Ein Beitrag zur vergleichenden Physiologie des Lict- und Farbensinnes. Pfleugers Arch Gesamte Physiol Menschen Tiere 42: 183–186. Fenchel T (2002) Microbial behavior in a heterogeneous world. Science 296: 1068–1071. Fenchel T & Thar R (2004) ‘‘Candidatus Ovobacter propellens’’: a large conspicuous prokaryote with an unusual motility behaviour. FEMS Microbiol Ecol 48: 231–238. Ford BJ (1991) The Leeuwenhoek Legacy. Biopress, Bristol, and Farrand Press, London. G¨otz R & Schmitt R (1987) Rhizobium meliloti swims by unidirectional, intermittent rotation of right-handed flagellar helices. J Bacteriol 169: 3146–3150. Givaudan A, Baghdiguian S, Lanois A & Boemare N (1995) Swarming and swimming changes concomitant with phase variation in Xenorhabdus nematophilus. Appl Environ Microbiol 61: 1408–1413. Grossart H-P, Riemann L & Azam F (2001) Bacterial motility in the sea and its ecological implications. Aquat Microb Ecol 25: 247–258. Hase CC, Fedorova ND, Galperin MY & Dibrov PA (2001) Sodium ion cycle in bacterial pathogens: evidence from crossgenome comparisons. Microbiol Mol Biol Rev 65: 353–70. Hirota N & Imae Y (1983) Na 1 -driven flagellar motors of an alkalophilic Bacillus strain YN-1. J Biol Chem 258: 10 577–10 581. Hodgson MM, Day B, White DJ & Tisa L (2003) Effects of growth conditions on the motility of Photorhabdus temperata. Arch Microbiol 180: 17–24. Johansen JE, Pinhassi J, Blackburn N, Zweifel UL & Hagtr¨om A (2002) Variability in motility characteristics among marine bacteria. Aquat Microb Ecol 28: 229–237. Kawagishi I, Maekawa Y, Atsumi T, Homma M & Imae Y (1995) Isolation of the polar and lateral flagellum-defective mutants in Vibrio alginolyticus and identification of their flagellar driving energy sources. J Bacteriol 177: 158–160. Kaye JZ & Baross JA (2004) Synchronous effects of temperature, hydrostatic pressure, and salinity on growth, phospholipid profiles, and protein patterns of four Halomonas species isolated from deep-sea hydrothermal-vent and sea surface environments. Appl Environ Microbiol 70: 6220–6229. Khan S & Trentham DR (2004) Biphasic excitation by leucine in Escherichia coli chemotaxis. J Bacteriol 186: 588–592.

FEMS Microbiol Ecol 55 (2006) 3–16

15

Bacterial motility

Kirboe T, Grossart H-P, Ploug H & Tang K (2002) Mechanisms and rates of bacterial colonization of sinking aggregates. Appl Environ Microbiol 68: 3996–4006. Kirboe T & Jackson GA (2001) Marine snow, organic solute plumes, and optimal chemosensory behavior of bacteria. Limnol Oceanogr 46: 1309–1318. Kita-Tsukamoto K, Wada M, Yao K, Nishino T & Kogure K (2004) Flagellar motors of marine bacteria Halomonas are driven by both protons and sodium ions. Can J Microbiol 50: 369–374. Kudo S, Magariyama Y & Aizawa S (1990) Abrupt changes in flagellar rotation observed by laser dark-field microscopy. Nature 346: 677–680. Lanfranconi MP, Alvarex HM & Studdert CA (2003) A strain isolated from gas oil-contaminated soil displays chemotaxis towards gas oil and hexadecane. Environ Microbiol 5: 1002–1008. Lenski R, Rose MR, Simpson SC & Tadler SC (1991) Long-term experimental evolution in Escherichia coli. I. Adaptation and divergence during 2,000 generations. Am Nat 138: 1315–1341. Liu Z, Chen W & Papadopoulos KD (1996) Bacterial motility, collisions, and aggregation in a 3-mm-diamter capillary. Biotechnol Bioeng 53: 238–241. Liu Z & Papadopoulos KD (1995) Unidirectional motility of Escherichia coli in restrictive capillaries. Appl Environ Microbiol 61: 3567–3572. Liu Y & Passino KM (2002) Biomimicry of social foraging bacteria for distributed optimization: models, principles, and emergent behaviors. J Optim Theor Appl 115: 603–628. Luchsinger R, Bergersen B & Mitchell JG (1999) Bacterial motility in the turbulent ocean. Biophys J 77: 2377–2386. Macnab RM & Koshland DE (1972) The gradient-sensing mechanism in bacterial chemotaxis. Proc Natl Acad Sci USA 69: 2509–2512. Magariyama Y, Sugiyama L, Muramoto K, Maekawa Y, Kawagishi I, Imae Y & Kudo S (1994) Very fast flagellar rotation. Nature 371: 752. Magariyama Y, Sugiyama S, Muramoto K, Kawagishi I, Imae Y & Kudo S (1995) Simultaneous measurement of bacterial flagellar rotation rate and swimming speed. Biophys J 69: 2154–2162. Manson MD, Tedesco P, Berg HC, Harold FM & Drift CVD (1977) A protonmotive force drives bacterial flagella. Proc Natl Acad Sci USA 74: 3060–3064. Maoz A, Mayr R & Scherer S (2003) Temporal stability and biodiversity of two complex antilisterial cheese-ripening microbial consortia. Appl Environ Microbiol 69: 4012–4018. Marques L, Nunes U & de Almeida AT (2002) Olfaction-based mobile robot navigation. Thin Solid Films 418: 51–58. Marwan W, Alam M & Oesterhelt D (1991) Rotation and switching of the flagellar motor assembly in Halobacterium halobium. J Bacteriol 173: 1971–1977. Mazzag BC, Zhulin IB & Mogilner A (2003) Model of bacterial band formation in aerotaxis. Biophys J 85: 3558–3574.

FEMS Microbiol Ecol 55 (2006) 3–16

Merrell DS, Butler SM, Qadri F, Dolganov NA, Alam A, Cohen MB, Calderwood SB & Schoolnik GK, et al. (2002) Hostinduced epidemic spread of the cholera bacterium. Nature 417: 642–645. Meyer G, Schneider-Merck T, B¨ohme S & Sand W (2002) A simple method for investigations on the chemotaxis of Acidithiobacillus ferrooxidans and Desulfovibrio vulgaris. Acta Biotechnol 22: 391–399. Millikan DS & Ruby EG (2002) Alterations in Vibrio fischeri motility correlate with a delay in symbiosis initiation and are associated with additional symbiotic colonization defects. Appl Environ Microbiol 68: 2519–2528. Minamino T, Imae Y, Oosawa F, Kobayashi Y & Oosawa K (2003) Effect of intracellular pH on rotation speed of bacterial flagellar motors. J Bacteriol 185: 1190–1194. Mitchell JG (2002) The energetics and scaling of search strategies in bacteria. Am Nat 160: 727–740. Mitchell JG & Barbara G (1999) The effect of ionic uncouplers on the speed of marine bacterial communities. Aqua Microb Ecol 18: 227–233. Mitchell JG, Martinez-Alonso M-R, Lalucat J, Esteve I & Brown S (1991) Velocity changes, long runs and reversals in the Chromatium minus swimming response. J Bacteriol 173: 997–1003. Mitchell JG, Pearson L, Bonazinga A, Dillon S, Khouri H & Paxinos R (1995) Long lag times and high velocities in the motility of natural assemblages of marine bacteria. Appl Environ Microbiol 61: 877–882. Mitchell JG, Pearson L & Dillon S (1996) The behavior of motile marine bacterioplankton communities clustering in narrow bands. Appl Environ Microbiol 62: 3716–3721. Murray JLS & Jumars PA (2002) Clonal fitness of attached bacteria predicted by analog modeling. Bioscience 52: 343–355. Overmann J & Schubert K (2002) Phototrophic consortia: models systems for symbiotic interrelations between prokaryotes. Arch Microbiol 177: 201–208. Parales RE & Harwood CS (2002) Bacteria chemotaxis to pollutants and plant-derived aromatic molecules. Curr Opin Microbiol 5: 266–273. Park S, Wolanin PM, Yuzbashyan EA, Lin H, Darnton NC, Stock JB, Silberzan P & Austin R (2003a) Influence of topology on bacterial social interaction. Proc Natl Acad Sci USA 100: 13 910–13 915. Park S, Wolanin PM, Yuzbashyan EA, Silberzan P, Stock JB & Austin RH (2003b) Motion to form a quorum. Science 301: 188. Purcell E (1977) Life at low Reynolds number. Am J Phys 45: 3–10. Rebbapragada A, Johnson MS, Harding GP, Zuccarelli AJ, Flectcher HM, Zhulin IB & Taylor BL (1997) The Aer protein and the serine hemoreceptor Tsr independently sense intracellular energy levels and transduce oxygen, redox, and energy signals for Escherichia coli behavior. Proc Natl Acad Sci USA 94: 10 541–10 546. Reddy GSN, Raghavan PUM, Sarita NB, Prakash JSS, Nagesh N, Delille D & Shivaji S (2003) Halomonas glaciei sp nov isolated

2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

16

from fast ice of Adelie Land, Antarctica. Extremophiles 7: 55–61. Sakai M, Futamata H & Kanazawa S (2003) Effects of high concentration of inorganic salts on swarming ability in fluorescent Pseudomonas strains. Biosci Biotechnol Biochem 67: 1479–1484. Scharf B (2002) Real-time imaging of fluorescent flagellar filaments of Rhizobium lupine H13-3: flagellar rotation and pH-induced polymorphic transistions. J Bacteriol 184: 5979–5986. Schmitt R (2002) Sinorhizobial chemotaxis: a departure from the enterobacterial paradigm. Microbiology 148: 627–631. Segall JE, Manson MD & Berg HC (1982) Signal processing times in bacterial chemotaxis. Nature 296: 855–857. Sire C & Chavanis P-H (2002) Thermodynamics and collapse of self-gravitating Brownian particles in D dimensions. Phys Rev E 66: 046133. Stock AM (1997) Energy sensors for aerotaxis in Escherichia coli: something old, something new. Proc Natl Acad Sci 94: 10 487–10 489. Thar R & K¨uhl M (2003) Bacteria are not too small for spatial sensing of chemical gradients: an experimental evidence. Proc Natl Acad Sci USA 100: 5748–5753. Tokuda H & Kogure K (1989) Generalized distribution and common properties of Na 1 -dependent NADH: quinone

2005 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved

c

J.G. Mitchell & K. Kogure

oxidoreductases in Gram-negative marine bacteria. J Gen Microbiol 135: 703–709. Tokuda H & Unemoto T (1981) A respiration-dependent primary sodium extrusion system functioning at alkaline pH in the marine bacterium Vibrio alginolyticus. Biochem Biophys Res Commun 102: 265–271. Trachtenberg S, Fishelov D & Ben-Artzi M (2003) Bacterial flagellar microhydrodynamics: laminar flow over complex flagellar filaments, analog Archimedean screws and cylinders and its perturbations. Biophys J 85: 1345–1357. Vetter Y-A, Deming JW, Jumars PW & Krieger-Brockett B (1998) A predictive model of bacterial foraging with freely-released extracellular enzymes. Microb Ecol 36: 75–92. Vigeant MA-S, Ford RM, Wagner M & Tamm LK (2002) Reversible and irreversible adhesion of motile Escherichia coli cells analyzed by total internal reflection aqueous fluorescence microscopy. Appl Environ Microbiol 68: 2794–2801. Visick KL & McFall-Ngai MJ (2000) An exclusive contract: specificity in the Vibrio fischeri–Euprymna scolopes partnership. J Bacteriol 182: 1779–1787. Wisdom J (2003) Swimming in spacetime: motion by cyclic changes in body shape. Science 299: 1865–1869. Yorimitsu T & Homma M (2001) Na 1 -driven flagellar motor of Vibrio. Biochim Biophys Acta 1505: 82–93. Zobell CE (1946) Marine Microbiology: A Monograph on Hydrobacteriology. Cronica Botanica, Waltham, MA.

FEMS Microbiol Ecol 55 (2006) 3–16