REVIEWS

The surprisingly diverse ways that

prokaryotes move
Ken F. Jarrell* and Mark J. McBride‡
Abstract | Prokaryotic cells move through liquids or over moist surfaces by swimming,
swarming, gliding, twitching or floating. An impressive diversity of motility mechanisms has
evolved in prokaryotes. Movement can involve surface appendages, such as flagella that spin,
pili that pull and Mycoplasma ‘legs’ that walk. Internal structures, such as the cytoskeleton
and gas vesicles, are involved in some types of motility, whereas the mechanisms of some
other types of movement remain mysterious. Regardless of the type of motility machinery
that is employed, most motile microorganisms use complex sensory systems to control their
movements in response to stimuli, which allows them to migrate to optimal environments.

Proton motive force

Prokaryotic cells have developed various mechanisms surface and acts as the propeller; and the hook, which
A special case of an to mediate movement. The use of flagella for swim- connects the basal body and the filament and acts as a
electrochemical potential. ming has been most thoroughly studied in enteric universal joint8. The basal body consists of a rod with
Proton motive force is the force bacteria, such as Escherichia coli, but there are many a series of rings. Flagella of most Gram-negative bacte-
that is created by the
variations on this theme, including different polar and ria have an L ring in the plane of the lipopolysaccharide
accumulation of protons on
one side of a cell membrane. lateral flagella systems in the same cell and the peri- in the outer membrane, a P ring in the plane of the
This concentration gradient is plasmic flagella of spirochaetes. Indeed, some swim- peptidoglycan and an MS ring (historically used to
generated using energy ming cells lack flagella and move without the aid of denote a membrane and supramembranous ring; now
sources, such as redox obvious appendages. Archaea swim by using flagella, known to be composed of one protein) that is located
potential or ATP. Once
established, the proton
but of a construction and assembly mechanism that is within and above the cytoplasmic membrane. L and P
motive force can be used to unique to their particular domain of life. Movement rings are not found in Gram-positive bacteria. An addi-
carry out work, for example, across solid surfaces includes various forms of swarm- tional ring, the C (cytoplasmic) ring, extends into the
to synthesize ATP or pump ing, gliding and twitching, which can be mediated by cytoplasm and is composed of the three switch proteins
compounds across the
flagella, type IV pili (T4P) and other nanomotors that FliG, FliM and FliN which are not only required for
membrane.
have been less well studied. In this Review, we describe rotation of the flagellum but also control the direction
the surprisingly diverse ways that prokaryotic cells have of flagellar rotation. Electron micrographs of isolated
evolved to swim, swarm, glide and float. frozen hydrated basal bodies of Salmonella enterica
serovar Typhimurium (S. typhimurium) revealed the
Bacterial flagella structure of the bacterial flagellum in extraordinary
The bacterial flagellum, the most thoroughly studied detail9. Assembly of the filament occurs by passage of
of all prokaryotic motility structures1–4 (FIG. 1), is used the flagellin subunits, via a flagellum-specific type III
for swimming in aqueous environments and also, in secretion system (T3SS), through the hollow flagellar
*Department of Microbiology some organisms, for swarming across solid surfaces 5. structure to the distal end for final assembly. Bacterial
and Immunology, Queen’s Swimming speeds can vary greatly between spe- flagella function by rotation10, which is driven by the
University, Kingston, cies: E. coli swims at a rate of 25–35 µm per second6, proton motive force, or more rarely, as observed in cer-
K7L 3N6 Ontario, Canada.
whereas Bdellovibrio bacteriovorus can reach 160 tain marine and alkaliphilic organisms, by a sodium
‡
Department of Biological
Sciences, University of µm per second7. The flagellum is an extraordinary gradient11. In many bacteria, the direction of flagellar
Wisconsin–Milwaukee, organelle in terms of its complex structure, assembly rotation is controlled by a chemotactic signal-transduc-
Milwaukee, 53201 mechanism and regulation. It is generally described as tion system that monitors environmental cues (BOX 1).
Wisconsin, USA. consisting of three substructures: the basal body, which The flagellar motor consists of five proteins: MotA
Correspondence to K.F.J.
e‑mail: jarrellk@queensu.ca
anchors the structure in the cell envelope and contains and MotB, which function as the stator and conduct the
doi:10.1038/nrmicro1900 the motor; the filament, which is approximately 20 nm passage of protons to drive flagellar rotation, and the
Published online 7 May 2008 in diameter, extends many cell lengths from the cell three switch proteins that are part of the rotor, FliG, FliM

466 | June 2008 | volume 6 www.nature.com/reviews/micro

© 2008 Nature Publishing Group
 REVIEWS

laboratory website (movie: flagella assembly); see Further

HAP2
information). A flagellum-specific T3SS that is crucial
for assembly of the components that lie outside of the
cytoplasmic membrane (rod, hook and filament) and
Assembly point consists of at least nine proteins is assembled at the
Filament base of the structure early during flagellar biogenesis2.
In addition, there are chaperones for each of the late
filament-type proteins that are exported: FliS (chap-
erone for FliC; flagellin), FliT (chaperone for FliD;
filament-capping protein), FlgN (chaperone for FlgK
CCW CW
and FlgL; hook-filament junction proteins)1 and, most
recently discovered, FliA or σ28, which is required for
late class III gene transcription (chaperone for FlgM;
anti-FliA) 16. Recently, FliJ was shown to recycle
free chaperones for the minor subunits, but not the FliC
HAP1
chaperone FliS, following subunit release at the T3SS17.
HAP3 This finding provided some of the first data on the
hierarchy of filament-type subunit export, and could
Hook explain how minor late components of the structure
H+ Basal body can be exported in a timely fashion but yet still compete
with the much more abundant flagellin subunits17. The
Outer membrane
pathways for incorporation of the early unchaperoned
subunits and the late chaperoned subunits converge at
Peptidoglycan FliI, the ATPase of the flagellar T3SS, before further
sorting18. The subunits are not distinguished by their
chaperone, but by the export and sorting signals within
their amino (N)‑terminal 100 amino acids18. Proteins
Cytoplasmic that are exported via the flagellar T3SS lack cleaved
membrane N‑terminal signal sequences and enter the growing
Motor C ring flagellum at the base of the MS ring. FliI was thought
complexes (motor switch)
to supply the energy for export via ATP hydrolysis but,
H+ Export
apparatus
recently, it has been shown that secretion of flagellar
Export proteins instead requires the proton motive force and
substrate
that ATP hydrolysis is not essential19. The severe motil-
Figure 1 | Model of the bacterial flagellum — structure ity defects observed in fliI deletion strains might be
Nature Reviews
and assembly. The bacterial flagellum | Microbiology
consists of three
due to other roles for this ATPase in flagella assembly,
parts: the filament, the hook and the complex basal body.
The locations and roles of many of its components are possibly involving the dissociation of export substrates
known. The proton gradient across the cytoplasmic from their cognate chaperones, as reported for the secre-
membrane drives rotation of the rotor and the attached tion ATPase of injectisomes. From the T3SS, the subu-
filament. CCW, counterclockwise; CW, clockwise; HAP, nits travel the length of the structure through a narrow
hook-associated protein. Modified, with permission, from hollow interior that is only 2 nm in diameter. Assembly
REF. 134  (1999) American Society for Microbiology. takes place at the distal tip of the structure, which,
for flagellins, means incorporation into the filament under
the pentameric capping protein (hook-associated protein 2)
and FliN2. MotB is physically linked to the peptidogly- many cell lengths away from the basal body1,20.
can layer via a peptidoglycan‑binding domain and thus In most bacteria, a strict and elegant regulation is
holds the stator in place while MotA interacts directly associated with flagellar assembly. Indeed, transcrip-
with the carboxyl (C) terminus of FliG and turns a ring tional, translational and post-translational regulation
of FliG molecules. Biochemical data suggest that passage have all been documented3,21,22. The enteric bacteria
of protons through the MotA and MotB channels causes E. coli and S. typhimurium have a well-studied three-
a conformational change in MotA (the power stroke), tiered transcriptional hierarchy. The class I master
which pushes FliG and turns the rotor12. MotB contains a regulators FlhC and FlhD are required for transcription
conserved aspartic acid residue (Asp32) that is essential of class II genes. These master regulators are themselves
for flagellar rotation and is thought to function directly under the control of global regulators, as well as small
in proton movement through the motor13. Bacterial molecules and quorum-sensing regulators. Class II
flagellar motors are incredibly fast, with speeds of up genes are responsible for assembly of the hook–basal
to 18,000 and 100,000 rpm reported for proton and Na+ body structure and the T3SS. The sigma factor FliA is
driven motors, respectively14,15. also encoded by a class II gene, and is required for the
Bacterial flagella are assembled in ordered stages, transcription of class III genes. Expression of class III
with the basal body assembled first, followed by genes is needed for late flagellar assembly and function.
the hook and finally the filament (Keiichi Namba’s Products of class III genes include hook-associated

nature reviews | microbiology volume 6 | June 2008 | 467

© 2008 Nature Publishing Group
REVIEWS

Box 1 | Chemotaxis What is sometimes forgotten in discussions about

the regulation of bacterial flagellar gene expression and
Motility is rarely random112. A motile cell senses stimuli and alters the functioning of its flagellar assembly is that the three‑tiered enteric system
motility machinery to improve its chances of migrating to a better location. Swimming is not universal. For example, Pseudomonas aeruginosa
Escherichia coli cells measure the concentrations of attractants and repellents over
and Helicobacter pylori have four-tiered systems, some
time and alter the direction of rotation of their flagella. This involves a signal
spirochaetes apparently use mainly post-transcriptional
transduction system that controls the phosphorylation state of the response regulator
CheY, altering its binding affinity for the flagellar-switch protein FliM. The E. coli regulation22,31 and Caulobacter crescentus has its own
flagellum turns anticlockwise in the absence of CheY–P, and the multiple flagella form a unique control systems32.
bundle that propels the cell in a ‘run’. Binding of CheY–P to FliM results in a switch to Unusual variations also occur in the placement of
clockwise rotation, which disrupts the bundle and results in a ‘tumble’ (Ref. 113) flagella, but there are probably none stranger than the
(Howard Berg’s laboratory website (movie: bacteria swimming); see Further periplasmic flagella of spirochaetes33. In many respects,
information). Cells tumble less frequently when moving in a favourable direction and these periplasmic flagella are similar to other bacte-
thus migrate to preferred environments. Covalent modification of receptor proteins rial flagella, but the filaments do not extend from the
allows cells to adapt and reset the sensitivity of the system as they migrate up or down surface of the cell. Instead, they reside in the space
a concentration gradient. Other flagellated bacteria use modified versions of this
between the cytoplasmic and outer membranes. The
strategy and the Che (chemotaxis) proteins are conserved.
spirochaete Treponema primitia is extremely slender,
Many archaea swim and exhibit chemotaxis and phototaxis. Archaeal flagella are
unrelated to bacterial flagella, but the tactic responses are mediated by signal which allowed electron cryotomographic analysis of
transduction proteins that resemble those involved in bacterial chemotaxis112,114,115. The intact hydrated cells34,35. Such experiments revealed for
level of CheY–P might control the direction of rotation of the archaeal flagella, but the first time the structure of entire bacterial flagellar
there is no archaeal FliM and the point of interaction between the Che system and the basal bodies, including the stator components MotA
flagellar motor is unknown. and MotB. They also demonstrated unique features of
Signal-transduction systems that are related to the Che system are also found spirochaete periplasmic flagella and cell structure. For
in some bacteria that move without flagella. Myxococcus xanthus displays type IV example, whereas most Gram-negative bacteria have
pili-dependent twitching motility and adventurous gliding motility, which have a single electron-dense layer of peptidoglycan, cells
different motors but are both controlled by the frz signal-transduction system, which is
of T. primitia have two electron-dense layers in this
related to the E. coli Che system. The frz system controls the frequency of reversal of
region. The periplasmic flagella, which extend from
gliding cells116. The exact interactions between the Frz proteins and the two motors are
not yet known, but might involve the GTPase MglA and proteins such as FrzS and the poles of the cell and wrap around the cell body, lie
RomR, which migrate from pole to pole during cell reversals116–118. In addition to the frz between the two layers and are thought to push against
system, M. xanthus has seven other sets of chemotaxis-like genes. Some of these both of them as they rotate. The final outcome depends
control gene expression rather than directly altering motility, but at least one, the dif on the flexibility of the particular spirochaete. For rigid
system, is involved in chemotactic responses to lipids119. spirochaetes, such as T. primitia, the outer membrane
and cytoplasmic membrane are thought to rotate in
opposite directions, resulting in corkscrew move-
proteins, flagellin, motor proteins and chemotaxis pro- ment of the cell through viscous media. A novel cone
teins. Transcription of the late genes is held in check structure that has been observed at the poles might
until completion of the hook–basal body substructure provide a region of more stable interaction between
by the anti-sigma factor FlgM. Upon completion the membranes, which might be needed for transport
of the hook–basal body, FlgM is exported out of the and other processes. In cells of less rigid spirochaetes,
cell, remarkably with the help of FliA as chaperone, such as Borrelia burgdorferi, rotation of the periplasmic
leaving FliA behind to aid in class III gene transcrip- flagella could result in cell bending or gyrations that
tion. Completion of the hook–basal body is a crucial ultimately cause cell movement33 (Nyles Charon’s labo-
checkpoint in flagellar assembly. Further controls on ratory website (movie: Borrelia swimming); see Further
this portion of the assembly mechanism include Flk, information). In either case, when cells swim smoothly,
which inhibits premature secretion of FlgM 23, and the flagellar motors at opposite poles are thought to
FlgN, which promotes the translation of FlgM24. turn in opposite directions relative to the cell body,
With the completion of the hook–basal body, the thereby allowing the filaments to work together rather
export of FlgM and the transcription of class III genes, than in conflict. Simultaneous switching of motors
the T3SS switches from the export of rod and hook pro- at each end would result in reversal of direction, and
teins to the export of filament proteins25. Two proteins switching of flagella at only one pole would result in
that are involved in this switching are FliK and FlhB. cells stopping or bending.
FliK regulates the length of the hook, for which several Besides swimming through liquid media, bacteria
theories have been proposed as explanations26. Recent can use flagella as one of many ways to move across solid
work suggests that FliK acts as an internal molecular surfaces. Movement across a surface that is powered by
ruler27, as the size of FliK correlates with hook length numerous flagella is referred to as swarming36 (Howard
and engineered insertions within FliK result in propor- Berg’s laboratory website (movie: bacteria swarming);
tionally longer hooks28. FliK also interacts with FlhB, see Further information). Bacteria that swarm are also
a key component of the T3SS that is thought to act capable of swimming. The switch to swarming is often
as the export gate25. FlhB undergoes autocleavage and accompanied by a change in cell morphology, with
Chemotaxis
Directed movement towards
the C‑terminal domain of FlhB interacts with FliK to swarmer cells being elongated and more flagellated.
attractants or away from change the export specificity of the system from rod or Although some bacteria, such as Proteus mirabilis, have
repellents. hook to filament-type substrates29,30. a single peritrichous flagellar system that is responsible

468 | June 2008 | volume 6 www.nature.com/reviews/micro

© 2008 Nature Publishing Group
 REVIEWS

FlaB1: major flagellin

FlaB2: major flagellin
FlaB3: flagellin (hook)
FlaA: minor flagellin
N-acetyl-glucosamine
Di-N-acetyl-glucuronic acid
N-acetyl-mannuronic acid with attached threonine
Dolichol lipid carrier

S-layer

FlaJ FlaJ
AglH AglC AglA ? AglB FlaK FlaH
? ?
FlaI
Signal peptide Cytoplasmic
cleaved
membrane
ATP ADP

Figure 2 | Model of the archaeal flagellum — structure and assembly. Archaeal flagella rotate to propel cells, but
they exhibit no obvious molecular similarity to bacterial flagella50,64. Two important post-translational modifications
Nature Reviews occur
| Microbiology
to archaeal flagellins. The type IV pilin-like signal peptide that is present on the flagellin is cleaved by the archaeal prepilin
peptidase-like enzyme FlaK/PibD. In Methanococcus voltae, the attached glycan is a trisaccharide that is attached at
multiple positions in all four flagellins. AglA, AglC and AglH are glycosyltransferases that are responsible for assembly of
the glycan on a lipid carrier, whereas AglB is an oligosaccharide transferase that catalyses its final transfer to the flagellins.
The addition of archaeal flagellin subunits to the growing filament has been proposed to occur at the base of the structure
as in type IV pili, rather than at the tip as in bacterial flagella.

for both swimming and swarming, others, such as Vibrio Superficially, archaeal flagella are similar to those of
parahaemolyticus, have distinct flagella systems for each bacteria: they are rotating structures with a filament
form of motility37. V. parahaemolyticus uses a sheathed and hook and the direction of rotation is controlled by
polar flagellum that is powered by a sodium motive chemotaxis systems. However, archaeal filaments are
force for swimming. By contrast, when it encounters much thinner than their bacterial counterparts, with
a surface, it produces numerous unsheathed lateral a diameter of 10–12 nm39. Archaea have different cell
flagella, which are driven by the proton motive force, walls, which might explain why at least some archaeal
for swarming38. Various environmental signals trigger flagella, such as those of extreme halophiles and pos-
induction of lateral flagella, including increased media sibly methanococci, appear to be anchored to a discoid
viscosity and the presence of a surface. How this sens- lamellar structure that lies internal to the cytoplasmic
ing takes place and its connection to lateral flagella membrane43–45.
induction remains unknown. In spite of the superficial similarities between bacte-
rial and archaeal flagella, molecular analyses indicate that
Archaeal flagella the components of the two motility organelles are unre-
Halophile
There are far fewer studies on archaeal motility lated. For example, the main components of the archaeal
A bacterium or archaeon that compared with studies of bacterial or eukaryotic cell flagellar filaments, the flagellins, exhibit no similarity to
can grow in environments that movements, and so far only flagella-driven motility and bacterial flagellins. Whereas bacterial flagellins lack signal
contain high concentrations of buoyancy by gas vesicles have been reported. Swimming peptides and are secreted via the flagellar T3SS, archaeal
salt (at least 2 M).
motility mediated by flagella is widespread through- flagellins have signal peptides that are similar to those
Signal peptide out the major subdivisions of archaea39, and studies of of bacterial T4P46. Archaeal flagellins are processed by
A short (3–60 amino acid long) flagella in a range of archaea have been published40. the preflagellin peptidase FlaK/PibD, a homologue of the
peptide chain that directs the Archaeal swimming speed has rarely been reported: bacterial type IV prepilin peptidase47–49. Some other pro-
post-translational transport of Halobacterium spp. are reported to swim at only teins that are essential for archaeal flagellation also have
a protein. Signal peptides are
also known as targeting signals,
2–3 µm per second, which is considerably slower than homologues in the T4P system, namely an ATPase (FlaI)
signal sequences, transit E. coli, under conditions of almost saturated sodium and a conserved membrane protein (FlaJ)50, whereas
peptides or localization signals. chloride that are typical of its natural environment41,42. no homologues of bacterial genes that are involved in

nature reviews | microbiology volume 6 | June 2008 | 469

© 2008 Nature Publishing Group
REVIEWS

flagella structure or assembly have been reported in result in a change of the helical handedness of the cell
any archaeon51,52. The interior of the archaeal flagellum and generate a moving kink68 (Joshua Shaevitz’s labora-
also shares structural homology with T4P and lacks a tory website (movie: Spiroplasma kinking); see Further
hollow interior that is large enough to allow passage of information). The process is then repeated, resulting in
flagellin monomers53,54. Indeed, the archaeal flagella fila- multiple kinks that move along the cell body. During
ment has been called a “bacterial propeller with a pilus- this process, fluid is pushed from front to back and the
like structure” (Ref. 55). Based on the lack of a central cells are propelled forward. The kinks seem to originate
channel and the similarities to T4P, archaeal flagella have from the same end of the cell, which indicates the pres-
been predicted to assemble new subunits at the base40,56 ence of a front and suggests that this might be the loca-
(FIG. 2), rather than by incorporating flagellin subunits tion of a motor that starts kink propagation, although
at the tip, as occurs for bacterial flagella. Consistent details of this putative motor remain elusive.
with this, it was recently reported that Methanococcus Some marine cyanobacteria of the genus Synechococcus
maripaludis flaB3 mutants produce hookless filaments57, also swim without flagella. These organisms are abun-
an outcome that is possible if the filament proteins are dant and actively swim at rates of 5–25 µm per second,
added before the hook components. From studies of spe- but went unnoticed until the mid-1980s69. Sodium
cies of both Methanococcus57,58 and Halobacterium59, it motive force is the energy source for their swimming70,
seems that one of the flagellins actually forms the hook but the mechanism of this mysterious form of move-
region in the archaeal flagellar filament. Hookless fila- ment is unknown. Large cell surface proteins that are
ments have not been reported in bacteria, in which the involved in motility have been identified by genetic
hook needs to be assembled before the filament. and molecular analyses71,72. Thin ‘spicules’ that extend
Another unusual feature of archaeal flagella is their from the cell and might be motility organelles have also
widespread N‑linked glycosylation. Glycosylation of bac- been observed in some strains73. A model to explain
terial flagellins is rare, and in all cases reported to date the Synechococcus swimming has been proposed that
glycan is attached via an O‑linkage 60. By contrast, involves the generation of travelling waves in the cell
the many examples of glycosylation of archaeal flagel- surface74, perhaps as a result of oar-like movements of
lins have all involved N‑linked glycans61,62. The struc- the cell surface spicules73.
tures of the attached glycans have been determined in
Halobacterium salinarum62 and Methanococcus voltae63. Movement over surfaces without flagella
In M. voltae and M. maripaludis, most of the enzymes Cells have evolved various molecular machines to ena-
responsible for the assembly and transfer of glycans to ble movement over the many types of surfaces that are
the flagella have been identified64 (D. VanDyke, J. Wu present in their natural environments. We have already
and K.F.J., unpublished observations). Interestingly, described how some bacteria use flagella to swarm over
a partial glycan of at least two sugars is required for surfaces. However, many other bacteria crawl over sur-
Methanococcus spp. flagellins to be incorporated into a faces without the aid of flagella, in processes that are
filament64 (D. VanDyke, J. Wu and K.F.J., unpublished known as twitching and gliding.
observations).
Type IV pili and twitching motility. T4P are found
Swimming without flagella in diverse bacteria and have long been suspected to
Bacterial and archaeal flagella are widespread, but be motility organelles. Their role in cell movement
many prokaryotes move without using these rotary has been unambiguously confirmed by microscopic
propellers. There is considerable diversity in the types observations and molecular analyses in the past dec-
of molecular machines that have evolved for non-flagellar ade75,76. Cell propulsion by T4P involves pilus exten-
motility. Here, we consider two examples of bacteria sion, attachment to a surface and retraction. This type
that swim without flagella. of movement is called twitching motility, as in many
Spiroplasma spp. are common insect and plant cases it results in jerky movement. In other bacteria,
pathogens that have a helical-shape, lack cell walls pilus retraction results in smooth movement that has
and maintain their helical cell morphology by internal historically been referred to as gliding motility, but in
cytoskeletal filaments. These filaments are also thought all cases the mechanism of propulsion is similar77. Cells
to be responsible for the ability of these bacteria to move by twitching at rates of 0.05–1 µm per second, and
swim. The filaments assemble into a series of three close proximity to another cell is usually required for
ribbons: two are composed of the fibril protein Fib, efficient movement. Bacteria that have T4P and display
whereas the third is probably composed of the actin-like twitching motility are phylogenetically diverse. They
protein MreB65,66. Unlike the similarly shaped spirocha- include proteobacteria, such as Neisseria gonorrhoeae,
etes, which have periplasmic flagella, Spiroplasma spp. P. aeruginosa and Myxococcus xanthus, unicellular and
swim faster in higher viscosity media independently of filamentous cyanobacteria, such as Synechocystis sp.
the gel-like nature of the media, which is suggestive PCC6803 and Nostoc punctiforme, and Gram-positive
of a novel swimming mechanism33. In contrast to the bacteria, such as Clostridium perfringens77–79.
rotary action of flagella, the swimming of Spiroplasma T4P are typically 6 nm in diameter with lengths
spp. is thought to be due to the contractile cytoskeleton that might exceed 10 µm, and they often extend from
that functions as a linear motor67. Differential length the cell poles 77. A core set of 12–15 proteins serve
changes of the filament building blocks are thought to as components of the pili or are directly involved in

470 | June 2008 | volume 6 www.nature.com/reviews/micro

© 2008 Nature Publishing Group
 REVIEWS

Direction of cell movement extension and retraction86. Pili are thought to adhere
Motor composed to surfaces primarily at their tips. This might involve
Outer the primary pilin PilA and/or additional proteins such
of Gld proteins
membrane
SprB adhesin Cytoplasmic as PilY1. Because of the geometry of PilA packing in
Peptidoglycan membrane the pilus, the region that is involved in attachment to a
surface is buried in most of the monomers, and is only
Periplasm
exposed for the PilA proteins at the tip.
Extension and retraction of pili were proposed as a
mechanism of twitching motility more than 25 years
ago based on observations of the attachment of bact­
Cytoplasm eriophages to pili during the infection of cells87. Twenty
years later, definitive proof that retraction of T4P
H+
results in cell movement was obtained by independ-
ent experiments with N. gonorrhoeae76, P. aeruginosa75
and M. xanthus88. Perhaps the most dramatic of these
demonstrations involved labelling the cell surface pro-
Substratum teins of P. aeruginosa with a fluorescent dye and subse-
quently observing the active extension and retraction
Figure 3 | Model to explain Flavobacterium johnsoniaeNature glidingReviews
motility.
| Microbiology
of pili and cell movement75 (Howard Berg’s laboratory
Flavobacterium gliding is thought to be powered by motors composed of Gld proteins in website (movies: type IV pili 1 and type IV pili 2); see
the cell envelope that propel adhesins, such as SprB, along the cell surface. Further information). The rapid assembly of pili from
newly synthesized monomers would require a tremen-
dous rate of new-protein synthesis. This requirement is
assembly and function80. Additional proteins regulate partly overcome by the cell’s ability to recycle the pri-
expression and activity. We will use the P. aeruginosa mary component of the pili, PilA. This was predicted
gene and protein designations in describing some based on theoretical calculations, and was supported by
of the core components of T4P. The pilus fibre is several observations. For example, the fluorescent pili
composed primarily of many copies of the pilin pro- mentioned above could be retracted into the cell and
tein PilA. PilA has an unusual signal peptide that is extended again while maintaining full fluorescence 75.
cleaved by the prepilin peptidase PilD, an enzyme that This is most easily explained by reuse of the labelled
is homologous to the preflagellin peptidase FlaK/PibD pilin monomers. The results of immunoelectron micro-
in archaeal flagellin processing. The N-terminal region scopy and an analysis of the relocalization of pilin that
of the processed pilin protein forms a long hydrophobic was labelled with biotin on the cell surface suggest that
α-helix, whereas the C‑terminal region folds to form disassembled pilin proteins ‘melt’ into the cytoplasmic
β-sheets. Assembly of multiple PilA monomers results membrane, where they form a pool of monomers for
in packing together of the hydrophobic helices, which assembly of the next pilus89. The structure of PilA fits
are surrounded by the β-sheet regions81. The pilus fibre well with this idea, as it is easy to envisage how the long
is helical, with approximately five pilin monomers per hydrophobic helix could aggregate with its neighbours
turn. ATP hydrolysis is thought to power extension of to form the core of the pilus and allow the individual
the pilus (addition of pilin subunits to the base) and/or monomers to melt into the cytoplasmic membrane
retraction (removal of pilin subunits from the base)82. during pilus retraction.
At least two ATPases with competing activities are T4P are used by a wide range of bacteria for move-
involved in these processes. PilB is required for pilus ment over surfaces. However, there are numerous
extension, whereas the related protein PilT is involved examples of bacteria that lack pili but still exhibit
in pilus retraction. Recent X‑ray crystallographic stud- gliding motility. Analysis of three different groups
ies of PilT suggest that binding of ATP results in large of gliding bacteria has begun to reveal the diverse
domain movements that might be responsible for pilus machineries that are involved in gliding.
retraction83. In E. coli, the cytoplasmic membrane pro-
tein BfpE (homologue of P. aeruginosa PilC) has been Flavobacterium spp. gliding: lateral movement of cell-
shown to interact with ATPase motors 84. It has been surface adhesins. Rapid gliding motility is a characteristic
proposed that ATP hydrolysis by PilB or PilT results in of many members of the Bacteroidetes phylum90. These
piston-like movements of a membrane protein (pos- bacteria typically glide at rates of 2–4 µm per second
sibly PilC) to add or remove pilin monomers during and occasionally reverse their direction of movement.
assembly and disassembly, respectively 85. In Gram- Cells often attach to a surface at one pole and rotate
negative bacteria, the secretin PilQ forms an oligomeric in place, similar to the propeller on an aeroplane, at
pore in the outer membrane to allow passage of the frequencies of approximately 2 revolutions per second.
pilus. PilQ is absent in Gram-positive bacteria with Unlike twitching motility, which is thought to be pow-
T4P79, which have no need for a secretin as they lack an ered by ATP hydrolysis, proton motive force seems to
outer membrane. Several minor pilins with sequence be the energy source for the gliding of Flavobacterium
similarity to PilA are also involved in pilus function, johnsoniae and other gliding Bacteroidetes. Genetic
probably by influencing the balance between pilus analyses have identified many genes and proteins that

nature reviews | microbiology volume 6 | June 2008 | 471

© 2008 Nature Publishing Group
REVIEWS

Polysaccharide although cells of sprB mutants fail to glide on agar,

a
they exhibit some limited movement in wet mounts
on glass. Second, the F. johnsoniae genome is predicted
to encode several proteins with similarity to SprB, and
a double mutant that lacks both SprB and one of its
paralogues was far less motile in wet mounts than
b Cell movement strains with single mutations in either gene. These data
suggests that SprB and related proteins might be cell
surface adhesins that are rapidly propelled by the Gld
Rotation
motor (FIG. 3). The presence of various semi-redundant
cell surface adhesins might explain how F. johnsoniae
Motors maintain fixed positions with respect to attaches to and glides over such diverse surfaces as agar,
the substratum and push the cell body forward glass, polystyrene and Teflon.
Gliding motility of Flavobacterium spp. has long
Figure 4 | Models to explain Myxococcus xanthus
adventurous motility. a | Polysaccharide secretion model. been described as rapid movement over surfaces
Nature Reviews | Microbiology
Polar secretion and hydration of polysaccharide propels without the presence of obvious motility organelles.
the cell forward. b | Focal adhesion model. A proposed However, recent cryo-electron microscopic obser-
motor in the cytoplasm, or cytoplasmic membrane, vations of cell surface appendages suggest that this
interacts with the cytoskeleton and with cell surface definition requires modification93. Wild-type cells had
adhesins that are attached to the substratum. The motor 5 nm wide filaments that extended from their outer
remains fixed with respect to the adhesins and the membranes, whereas filaments were absent from cells
substratum and propels the cell forward. Modified, with of a non-motile gldF mutant. The molecular nature of
permission, from REF. 100  (2007) American Association these filaments is unknown, but they might be the cell
for the Advancement of Science.
surface adhesins that are involved in motility and SprB
is an obvious candidate component.
are involved in the gliding of Flavobacterium spp. Many
of these are unique to members of the Bacteroidetes M. xanthus adventurous gliding motility: focal adhesion
phylum, suggesting that gliding of Flavobacterium spp. model and the role of polysaccharides. Cells of
and their relatives have evolved independently from M. xanthus migrate slowly (2–4 µm per minute) over
other forms of bacterial locomotion91. The required surfaces as they attack and digest other microorganisms.
components of the machinery include three proteins When the cells run out of nutrients, they aggregate to
that are thought to constitute a transmembrane ATP- form multicellular fruiting bodies that are filled with
binding-cassette transporter (GldA, GldF and GldG), dormant spores. M. xanthus exhibits two independent
five lipoproteins of unknown function (GldB, GldD, forms of surface motility during its complicated life
GldH, GldJ and GldK), a cytoplasmic membrane pro- cycle: T4P-mediated social motility (S-motility; a type
tein (GldL) and two periplasmic proteins (GldM and of twitching motility94) and pilus-independent adven-
GldN). Some of these proteins probably form the glid- turous motility (A-motility). Two different models have
ing motor, as disruption of any of the genes that encode recently been proposed to explain A‑motility (FIG. 4). The
these proteins results in complete loss of motility. first model suggests that polysaccharide extrusion and
Cell surface components of the motility machin- hydration propels cells95. This is based on the observa-
ery seem to have redundant functions and thus have tion that cells secrete polysaccharide slime as they glide,
eluded detection until recently. Analysis of mutants often forming a trail behind them. Slime secretion has
with partial defects in gliding resulted in the identifi- also been suggested as a model for the gliding of some
cation of several enormous proteins that are exposed cyanobacteria96. Mutations in several M. xanthus genes
on the cell surface. One of these, the highly repetitive that are probably involved in polysaccharide synthesis
669 kDa SprB, is required for movement over agar result in partial defects in A‑motility97. In addition, two
and appears to be a mobile component of the motil- homologues each of E. coli TolB, TolQ and TolR are
ity apparatus92. Antibodies against SprB inhibit cell required for M. xanthus A‑motility98,99. The E. coli Tol
movement, which suggests that SprB has an important proteins are involved in active transport of molecules
role in motility. Protein-G-coated latex spheres that across the outer membrane. The M. xanthus homologues
carry antibody against SprB bound specifically to might be involved in cell propulsion by polysaccharide
cells that were expressing SprB92 and were propelled secretion or might have other functions that are essential
along the cell surface at approximately 2 µm per sec- to gliding. Recent results from the Zusman laboratory100
ond (Supplementary information S1 (movie)). These suggest another possible mechanism for A‑motility:
spheres often moved continuously along the length of they observed that a fluorescently tagged version of
the cell, around the pole and back down the opposite the cytoplasmic AglZ protein, which is required for
side of the cell. Multiple spheres that were attached to A‑motility, remained stationary with respect to the
a cell could follow the same or different paths. Protein- substratum as the cells glided. They propose that AglZ
G-coated spheres without antibody failed to bind to might be part of a motor complex that pushes against
cells. Several lines of evidence suggest that SprB is a the helical cytoskeleton and against cell surface adhes-
redundant component of the motility machinery. First, ins. This focal adhesion complex model has much in

472 | June 2008 | volume 6 www.nature.com/reviews/micro

© 2008 Nature Publishing Group
 REVIEWS

a b source for gliding is ATP, a finding that was most

convincingly demonstrated by an elegant experiment
that involved solubilization of the M. mobile mem-
brane with detergent followed by digestion of nucleic
acids106 (Supplementary information S2 (movie)). The
resulting ‘ghosts’ were not viable, but they regained
motility when ATP was added. The gene that encodes
P42 is co-transcribed with the gli (gliding) genes
ATP
and encodes a novel ATPase that might be the glid-
ADP + Pi ing motor107. P42 is an unusual ATPase, as it lacks a
Walker A box or any other features that would indicate
that it has ATPase activity.
Mycoplasma pneumoniae exhibits slow gliding motil-
ity that is invariably in the direction of the terminal
organelle, a narrow constriction at the front of the cell
that has a complex cytoskeleton. Analysis of mutants
indicates that large cell surface proteins located at the
Figure 5 | Two models to explain the gliding of different mycoplasmas. a | Centipede front of the cell and cytoskeletal proteins of the termi-
model for Mycoplasma mobile gliding. Large cell surface GliNature
proteins constitute
Reviews the ‘legs’
| Microbiology nal organelle are involved in movement. One mutant
(pink), which are localized in the ‘neck’ region of the cell. The legs attach to the lacks the protein P41, which is required to anchor the
substratum. ATP hydrolysis by motor components in the cytoplasm or cytoplasmic terminal organelle to the cell body. When cells of this
membrane drive conformational changes of the legs that result in cell movement. mutant crawl over a surface, the powerful gliding motor
b | Inchworm model for Mycoplasma pneumoniae gliding. Adhesins on the surface of the sometimes rips the terminal organelle away from the
terminal organelle at the front of the cell (red) attach to the substratum. Repeated cell body. Amazingly, the orphan terminal organelles,
extension and contraction of the terminal organelle cytoskeleton (pink), which could which are only approximately 100 x 200 nm in size,
involve bending of the entire structure or sliding of some components against others,
continue to glide for more than 30 minutes, which
and coordinated release of adhesins from the substratum during the cycle, results in cell
movement. Panel a modified, with permission, from Ref. 102  (2008) Elsevier Science.
convincingly demonstrates that the motility apparatus
resides in this tiny region of the cell103 (Supplementary
information S3, S4 (movies)). The exact mechanism
of M. pneumoniae movement is unknown, but recent
common with models for the gliding of protists such electron microscopy cryotomographic data suggest
as Plasmodium spp.101, although the individual com- that the cytoskeleton in the terminal organelle forms
ponents involved do not seem to be genetically related. a multi-subunit dynamic motor108,109. Inchworm-like
Polysaccharides might play an important part by conformational changes of the cytoskeleton are thought
coating the substratum and interacting with cell surface to cause the terminal organelle to extend and contract,
adhesins. Numerous other genes and proteins involved pulling the cell along in the process (FIG. 5). The models
in A‑motility have been identified99, and further studies proposed for M. mobile and M. pneumoniae gliding are
of these could determine whether either or both models distinct, but further experiments might reveal a com-
are correct. mon underlying mechanism in spite of the absence of
sequence similarities among the known components
Mycoplasma gliding: centipede or inchworm? of their motility machineries.
Mycoplasma species are tiny bacteria (usually less than
0.3 µm across) that lack cell walls and have minimal Passive and parasitic movements
genomes which typically contain 500–700 protein- In addition to the active forms of motility discussed
encoding genes. Despite their small size and simple above, some prokaryotes take a more passive approach
genomes, they have complex cytoskeletal features to movement. For example, gas-vesicle mediated verti-
and some exhibit active gliding motility 102. Cells are cal movements in the water column are common in bac-
often asymmetrical and motile cells invariably move teria and archaea (BOX 2). Other bacteria lack obvious
in the direction of the ‘head’. Mycoplasma mobile motors but form spreading colonies by using the expan-
moves continuously at a speed of 2–5 µm per sec- sive forces that result from cell growth. The modified
ond, whereas most other gliding mycoplasmas move cell surfaces and surfactants produced by these bacteria,
sporadically and approximately 10 times more slowly. which include members of the genus Mycobacterium
Surprisingly, molecular, genetic and genomic stud- among others, allow groups of cells to slide passively
ies suggest that the machineries responsible for cell away from the colony centre, sometimes resulting in the
movement in the fast gliders and the slow gliders formation of elaborate branching colonies110.
might not be closely related 103 (FIG. 5). In M. mobile, Another novel form of cell movement that occurs
large cell surface Gli proteins that are localized to the only within a eukaryotic host cell is used by intracellular
‘neck’ region are required for gliding 102,104,105. These parasites such as Listeria monocytogenes and Shigella
are probably connected to the cytoskeleton, and are flexneri. These bacteria migrate within and between
thought to function as tiny legs, with movement of the host cells by polar polymerization of host cell actin111.
legs resulting in centipede-like motion102. The energy Specific proteins produced by the bacteria localize

nature reviews | microbiology volume 6 | June 2008 | 473

© 2008 Nature Publishing Group
REVIEWS

Box 2 | Gas vesicles

Some aquatic bacteria and archaea, including some with flagella, use gas vesicles to provide buoyancy and allow cells to
move within the water column. Phototrophs might use gas vesicles to reach areas with appropriate light intensity,
whereas aerobic bacteria or archaea might use them to float to oxygenated surface waters120. Gas vesicles are hollow,
gas-filled cylindrical structures with conical caps, are composed entirely from protein and are generally conserved in
morphology throughout prokaryotes, although they vary in width121. The volume of gas vesicles changes little over a wide
range of pressures, but they collapse irreversibly beyond a critical pressure. Narrower gas vesicles withstand greater
pressure and are found in organisms that live at greater depths.
Gas-vesicle research has focused on cyanobacteria and halophilic archaea for which, typically, 10–14 gvp genes are
involved in their formation122–125. Gas-vesicle protein composition has been a difficult issue to resolve owing to the
extreme resistance of the structure to solubilization. The coils of the structure are composed of 5 nm ribs of GvpA126,127,
a small hydrophobic protein that bestows on the gas vesicle its permeability to gases and impermeability to water. The
sequence of the highly conserved GvpA influences both the width and the threshold pressure for collapse120. The gas-
vesicle structure is strengthened by GvpC, a protein with repeats of a 33 amino acid motif that adheres to the outside of
the ribs, possibly by binding several ribs like a clamp128. Several different models for the interactions of GvpA and GvpC
have been proposed121,129,130. Matrix-assisted laser desorption/ionization–time of flight analysis of purified gas vesicles of
the cyanobacterium Anabaena flos-aquae failed to detect additional proteins129, but five additional components of the
gas vesicles of Halobacterium salinarum (GvpF, GvpG, GvpJ, GvpL and GvpM) were identified by immunoblotting131. Some
of these might be nucleation proteins that are involved in the initiation of vesicle formation.
The mechanism of buoyancy regulation in cyanobacteria involves the control of gas-vesicle formation or collapse and
the synthesis or breakdown of dense cell components, especially carbohydrates121. Buoyancy at high irradiance is lost
owing to an increase in the amount of carbohydrates that accumulate by photosynthesis and a decrease in the total gas-
vesicle volume, whereas increased buoyancy at low irradiance is the combined effect of metabolism of carbohydrate to
less dense components and an increase in gas-vesicle volume132,133.

to the cell pole and facilitate actin polymerization. discovered and characterized in bacteria. Archaea are
Disruption of the genes needed for actin polymeriza- only known to have one type of active motility organelle,
tion compromises the ability of these pathogens to the archaeal flagellum. This might reflect the lack of sci-
cause disease and spread in the host. entific focus on archaeal motility, however, rather than a
lack of initiative among the archaea. The huge diversity
Conclusions of prokaryotic motility mechanisms is now clear, but the
Twenty years ago, a review of prokaryotic motility would detailed workings of the individual machines are poorly
probably have discussed bacterial flagellar motility and, understood. Because some of these machines are used
briefly, gliding motility. At the time, a single mechanism not only for cell movement, but also seem to have roles
for gliding was plausible and archaeal flagella were in such diverse processes as macromolecule transport,
assumed to be similar to bacterial flagella. We now adhesion and cell division, continued study of prokaryo-
know that numerous novel motors have evolved to tic motility will illuminate many areas of prokaryotic cell
propel prokaryotic cells. All but one of these have been physiology and molecular biology.

1. Macnab, R. M. Type III flagellar protein export and 11. McCarter, L. L. Dual flagellar systems enable motility aid the early formation of the hook-filament
flagellar assembly. Biochim. Biophys. Acta 1694, under different circumstances. J. Mol. Microbiol. junction and the capping structure.
207–217 (2004). Biotechnol. 7, 18–29 (2004). 18. Stafford, G. P. et al. Sorting of early and late flagellar
2. Macnab, R. M. How bacteria assemble flagella. Annu. 12. Kojima, S. & Blair, D. F. Conformational change in the subunits after docking at the membrane ATPase of the
Rev. Microbiol. 57, 77–100 (2003). stator of the bacterial flagellar motor. Biochemistry type III export pathway. J. Mol. Biol. 374, 877–882
3. Aldridge, P. & Hughes, K. T. Regulation of flagellar 40, 13041–13050 (2001). (2007).
assembly. Curr. Opin. Microbiol. 5, 160–165 (2002). 13. Blair, D. F. Flagellar movement driven by 19. Paul, K., Erhardt, M., Hirano, T., Blair, D. F. & Hughes,
4. Apel, D. & Surette, M. G. Bringing order to a complex proton translocation. FEBS Lett. 545, 86–95 K. T. Energy source of flagellar type III secretion.
molecular machine: the assembly of the bacterial (2003). Nature 451, 489–492 (2008).
flagella. Biochim. Biophys. Acta 24 July 2007 14. Chen, X. & Berg, H. C. Torque-speed relationship of The energy source for the transport of flagellin
(doi: 10.1016/j.bbamem.2007.07.005). the flagellar rotary motor of Escherichia coli. Biophys. subunits through the T3SS, originally thought to be
5. Kaiser, D. Bacterial swarming: a re-examination of cell- J. 78, 1036–1041 (2000). supplied by the hydrolysis of ATP by FliI, was shown
movement patterns. Curr. Biol. 17, R561–R570 (2007). 15. Magariyama, Y. et al. Very fast flagellar rotation. instead to be dependent on the proton motive force,
6. Lowe, G., Meister, M. & Berg, H. C. Rapid rotation of Nature 371, 752 (1994). with ATP hydrolysis not being essential for secretion.
flagellar bundles in swimming bacteria. Nature 325, 16. Aldridge, P. D. et al. The flagellar-specific transcription 20. Yonekura, K. et al. The bacterial flagellar cap as the
637–640 (1987). factor, σ28, is the type III secretion chaperone for the rotary promoter of flagellin self-assembly. Science
7. Lambert, C. et al. Characterizing the flagellar filament flagellar-specific anti-σ28 factor FlgM. Genes Dev. 20, 290, 2148–2152 (2000).
and the role of motility in bacterial prey-penetration 2315–2326 (2006). 21. Lee, H. J. & Hughes, K. T. Posttranscriptional control
by Bdellovibrio bacteriovorus. Mol. Microbiol. 60, 17. Evans, L. D., Stafford, G. P., Ahmed, S., Fraser, G. M. of the Salmonella enterica flagellar hook protein
274–286 (2006). & Hughes, C. An escort mechanism for cycling of FlgE. J. Bacteriol. 188, 3308–3316 (2006).
8. Minamino, T. & Namba, K. Self-assembly and type III export chaperones during flagellum assembly. 22. McCarter, L. L. Regulation of flagella. Curr. Opin.
protein export of the bacterial flagellum. J. Mol. Proc. Natl Acad. Sci. USA 103, 17474–17479 Microbiol. 9, 180–186 (2006).
Microbiol. Biotechnol. 7, 5–17 (2004). (2006). 23. Aldridge, P., Karlinsey, J. E., Becker, E., Chevance, F. F.
9. Thomas, D. R., Francis, N. R., Xu, C. & DeRosier, D. J. Showed that FliJ has a novel chaperone-escort & Hughes, K. T. Flk prevents premature secretion of
The three-dimensional structure of the flagellar rotor function in flagellum assembly by specifically the anti-σ factor FlgM into the periplasm. Mol.
from a clockwise-locked mutant of Salmonella enterica recruiting unladen chaperones for the minor Microbiol. 60, 630–643 (2006).
serovar Typhimurium. J. Bacteriol. 188, 7039–7048 filament-type subunits (hook-associated proteins) 24. Aldridge, P., Karlinsey, J. & Hughes, K. T. The type III
(2006). and transferring them to their cognate subunits. secretion chaperone FlgN regulates flagellar assembly
10. Berg, H. C. & Anderson, R. A. Bacteria swim by Because FliJ does not escort the chaperone for via a negative feedback loop containing its chaperone
rotating their flagellar filaments. Nature 245, flagellin, these data suggest a mechanism that substrates FlgK and FlgL. Mol. Microbiol. 49,
380–382 (1973). would favour the export of the minor subunits to 1333–1345 (2003).

474 | June 2008 | volume 6 www.nature.com/reviews/micro

© 2008 Nature Publishing Group
 REVIEWS

25. Ferris, H. U. & Minamino, T. Flipping the switch: 51. Faguy, D. M. & Jarrell, K. F. A twisted tale: the origin 72. Brahamsha, B. An abundant cell-surface polypeptide
bringing order to flagellar assembly. Trends Microbiol. and evolution of motility and chemotaxis in is required for swimming by the nonflagellated marine
14, 519–526 (2006). prokaryotes. Microbiology 145, 279–281 (1999). cyanobacterium Synechococcus. Proc. Natl Acad. Sci.
26. Waters, R. C., O’Toole, P. W. & Ryan, K. A. The FliK 52. Nutsch, T., Oesterhelt, D., Gilles, E. D. & Marwan, W. USA 93, 6504–6509 (1996).
protein and flagellar hook-length control. Protein Sci. A quantitative model of the switch cycle of an archaeal 73. Samuel, A. D., Petersen, J. D. & Reese, T. S. Envelope
16, 769–780 (2007). flagellar motor and its sensory control. Biophys. J. 89, structure of Synechococcus sp. WH8113, a
27. Journet, L., Agrain, C., Broz, P. & Cornelis, G. R. The 2307–2323 (2005). nonflagellated swimming cyanobacterium. BMC
needle length of bacterial injectisomes is determined by 53. Cohen-Krausz, S. & Trachtenberg, S. The structure of Microbiol. 1, 4 (2001).
a molecular ruler. Science 302, 1757–1760 (2003). the archeabacterial flagellar filament of the extreme 74. Ehlers, K. M., Samuel, A. D., Berg, H. C. &
28. Shibata, S. et al. FliK regulates flagellar hook length halophile Halobacterium salinarum R1M1 and its Montgomery, R. Do cyanobacteria swim using
as an internal ruler. Mol. Microbiol. 64, 1404–1415 relation to eubacterial flagellar filaments and type IV traveling surface waves? Proc. Natl Acad. Sci. USA 93,
(2007). pili. J. Mol. Biol. 321, 383–395 (2002). 8340–8343 (1996).
29. Ferris, H. U. et al. FlhB regulates ordered export of 54. Cohen-Krausz, S. & Trachtenberg, S. The flagellar 75. Skerker, J. M. & Berg, H. C. Direct observation of
flagellar components via autocleavage mechanism. filament structure of the extreme acidothermophile extension and retraction of type IV pili. Proc. Natl
J. Biol. Chem. 280, 41236–41242 (2005). Sulfolobus shibatae B12 suggests that Acad. Sci. USA 98, 6901–6904 (2001).
30. Minamino, T., Ferris, H. U., Moriya, N., Kihara, M. & archaeabacterial flagella have a unique and common Reported the direct observation of extension and
Namba, K. Two parts of the T3S4 domain of the hook- symmetry and design. J. Mol. Biol. 375, 1113–1124 retraction of fluorescently labelled T4P, which are
length control protein FliK are essential for the (2008). associated with twitching motility.
substrate specificity switching of the flagellar type III Analysed two phylogenetically distant archaea to 76. Merz, A. J., So, M. & Sheetz, M. P. Pilus retraction
export apparatus. J. Mol. Biol. 362, 1148–1158 show that their archaeal flagellar filament powers bacterial twitching motility. Nature 407,
(2006). structures were unique compared with bacterial 98–102 (2000).
31. Charon, N. W. & Goldstein, S. F. Genetics of motility filaments. Also confirmed the lack of a central 77. Mattick, J. S. Type IV pili and twitching motility. Annu.
and chemotaxis of a fascinating group of bacteria: channel in both flagella filaments, a feature that is Rev. Microbiol. 56, 289–314 (2002).
the spirochetes. Annu. Rev. Genet. 36, 47–73 crucial for assembly models. 78. Duggan, P. S., Gottardello, P. & Adams, D. G.
(2002). 55. Trachtenberg, S. & Cohen-Krausz, S. The Molecular analysis of genes in Nostoc punctiforme
32. England, J. C. & Gober, J. W. Cell cycle control of cell archaeabacterial flagellar filament: a bacterial involved in pilus biogenesis and plant infection.
morphogenesis in Caulobacter. Curr. Opin. Microbiol. propeller with a pilus-like structure. J. Mol. Microbiol. J. Bacteriol. 189, 4547–4551 (2007).
4, 674–680 (2001). Biotechnol. 11, 208–220 (2006). 79. Varga, J. J. et al. Type IV pili-dependent gliding
33. Wolgemuth, C. W., Charon, N. W., Goldstein, S. F. & 56. Jarrell, K. F., Bayley, D. P. & Kostyukova, A. S. The motility in the Gram-positive pathogen Clostridium
Goldstein, R. E. The flagellar cytoskeleton of the archaeal flagellum: a unique motility structure. perfringens and other Clostridia. Mol. Microbiol. 62,
spirochetes. J. Mol. Microbiol. Biotechnol. 11, J. Bacteriol. 178, 5057–5064 (1996). 680–694 (2006).
221–227 (2006). 57. Chaban, B. et al. Systematic deletion analyses of the 80. Carbonnelle, E., Helaine, S., Nassif, X. & Pelicic, V.
34. Murphy, G. E., Matson, E. G., Leadbetter, J. R., Berg, fla genes in the flagella operon identify several genes A systematic genetic analysis in Neisseria meningitidis
H. C. & Jensen, G. J. Novel ultrastructures of essential for proper assembly and function of flagella defines the Pil proteins required for assembly,
Treponema primitia and their implications for motility. in the archaeon, Methanococcus maripaludis. Mol. functionality, stabilization and export of type IV pili.
Mol. Microbiol. 67, 1184–1195 (2008). Microbiol. 66, 596–609 (2007). Mol. Microbiol. 61, 1510–1522 (2006).
35. Murphy, G. E., Leadbetter, J. R. & Jensen, G. J. In situ 58. Bardy, S. L., Mori, T., Komoriya, K., Aizawa, S. & 81. Craig, L., Pique, M. E. & Tainer, J. A. Type IV pilus
structure of the complete Treponema primitia flagellar Jarrell, K. F. Identification and localization of flagellins structure and bacterial pathogenicity. Nature Rev.
motor. Nature 442, 1062–1064 (2006). FlaA and FlaB3 within flagella of Methanococcus Microbiol. 2, 363–378 (2004).
36. Harshey, R. M. Bacterial motility on a surface: many voltae. J. Bacteriol. 184, 5223–5233 (2002). 82. Burrows, L. L. Weapons of mass retraction. Mol.
ways to a common goal. Annu. Rev. Microbiol. 57, 59. Beznosov, S. N., Pyatibratov, M. G. & Fedorov, O. V. Microbiol. 57, 878–888 (2005).
249–273 (2003). On the multicomponent nature of Halobacterium 83. Satyshur, K. A. et al. Crystal structures of the pilus
37. Merino, S., Shaw, J. G. & Tomas, J. M. Bacterial lateral salinarum flagella. Microbiology (Russ.) 76, 435–441 retraction motor PilT suggest large domain
flagella: an inducible flagella system. FEMS Microbiol. (2007). movements and subunit cooperation drive motility.
Lett. 263, 127–135 (2006). 60. Logan, S. M. Flagellar glycosylation — a new Structure 15, 363–376 (2007).
38. Atsumi, T., McCarter, L. & Imae, Y. Polar and lateral component of the motility repertoire? Microbiology 84. Crowther, L. J., Anantha, R. P. & Donnenberg, M. S.
flagellar motors of marine Vibrio are driven by different 152, 1249–1262 (2006). The inner membrane subassembly of the
ion-motive forces. Nature 355, 182–184 (1992). 61. Lechner, J. & Wieland, F. Structure and biosynthesis of enteropathogenic Escherichia coli bundle-forming
39. Thomas, N. A., Bardy, S. L. & Jarrell, K. F. The prokaryotic glycoproteins. Annu. Rev. Biochem. 58, pilus machine. Mol. Microbiol. 52, 67–79 (2004).
archaeal flagellum: a different kind of prokaryotic 173–194 (1989). 85. Craig, L. et al. Type IV pilus structure by cryo-electron
motility structure. FEMS Microbiol. Rev. 25, 147–174 62. Sumper, M. Halobacterial glycoprotein biosynthesis. microscopy and crystallography: implications for pilus
(2001). Biochim. Biophys. Acta 906, 69–79 (1987). assembly and functions. Mol. Cell 23, 651–662
40. Jarrell, K. F., Ng, S. Y. & Chaban, B. in Archaea: 63. Voisin, S. et al. Identification and characterization of (2006).
Molecular and Cellular Biology (ed. Cavicchioli, R.) the unique N‑linked glycan common to the flagellins 86. Winther-Larsen, H. C. et al. A conserved set of pilin-
385–410 (ASM, Washington DC, 2007). and S‑layer glycoprotein of Methanococcus voltae. like molecules controls type IV pilus dynamics and
41. Alam, M., Claviez, M., Oesterhelt, D. & Kessel, M. J. Biol. Chem. 280, 16586–16593 (2005). organelle-associated functions in Neisseria
Flagella and motility behaviour of square bacteria. 64. Chaban, B., Voisin, S., Kelly, J., Logan, S. M. & Jarrell, gonorrhoeae. Mol. Microbiol. 56, 903–917 (2005).
EMBO J. 3, 2899–2903 (1984). K. F. Identification of genes involved in the biosynthesis 87. Bradley, D. E. A function of Pseudomonas aeruginosa
42. Marwan, W., Alam, M. & Oesterhelt, D. Rotation and and attachment of Methanococcus voltae N‑linked PAO polar pili: twitching motility. Can. J. Microbiol.
switching of the flagellar motor assembly in glycans: insight into N‑linked glycosylation pathways in 26, 146–154 (1980).
Halobacterium halobium. J. Bacteriol. 173, Archaea. Mol. Microbiol. 61, 259–268 (2006). 88. Sun, H., Zusman, D. R. & Shi, W. Type IV pilus of
1971–1977 (1991). Identified genes that are involved in N‑linked Myxococcus xanthus is a motility apparatus
43. Kupper, J. et al. The flagellar bundle of Halobacterium glycosylation in archaea. Also showed that controlled by the frz chemosensory system. Curr. Biol.
salinarium is inserted into a distinct polar cap underglycosylated flagellins are either not 10, 1143–1146 (2000).
structure. J. Bacteriol. 176, 5184–5187 (1994). incorporated or incorporated only poorly into 89. Morand, P. C. et al. Type IV pilus retraction in
44. Metlina, A. L. Bacterial and archaeal flagella as filaments, thereby demonstrating an essential role pathogenic Neisseria is regulated by the PilC proteins.
prokaryotic motility organelles. Biochemistry (Mosc.) for glycosylation in the assembly of archaeal EMBO J. 23, 2009–2017 (2004).
69, 1203–1212 (2004). flagella. 90. McBride, M. J. Bacterial gliding motility: multiple
45. Jarrell, K. F. & Koval, S. F. Ultrastructure and 65. Wolgemuth, C. W. & Charon, N. W. The kinky mechanisms for cell movement over surfaces. Annu.
biochemistry of Methanococcus voltae. Crit. Rev. propulsion of Spiroplasma. Cell 122, 827–828 Rev. Microbiol. 55, 49–75 (2001).
Microbiol. 17, 53–87 (1989). (2005). 91. Braun, T. F., Khubbar, M. K., Saffarini, D. A. &
46. Faguy, D. M., Jarrell, K. F., Kuzio, J. & Kalmokoff, 66. Kurner, J., Frangakis, A. S. & Baumeister, W. Cryo- McBride, M. J. Flavobacterium johnsoniae gliding
M. L. Molecular analysis of archael flagellins: similarity electron tomography reveals the cytoskeletal structure motility genes identified by mariner mutagenesis.
to the type IV pilin-transport superfamily widespread of Spiroplasma melliferum. Science 307, 436–438 J. Bacteriol. 187, 6943–6952 (2005).
in bacteria. Can. J. Microbiol. 40, 67–71 (1994). (2005). 92. Nelson, S. S., Bollampalli, S. & McBride, M. J. SprB is
47. Bardy, S. L. & Jarrell, K. F. Cleavage of preflagellins by 67. Trachtenberg, S. The cytoskeleton of Spiroplasma: a a cell-surface component of the Flavobacterium
an aspartic acid signal peptidase is essential for complex linear motor. J. Mol. Microbiol. Biotechnol. johnsoniae gliding motility machinery. J. Bacteriol.
flagellation in the archaeon Methanococcus voltae. 11, 265–283 (2006). 190, 2851–2857 (2008).
Mol. Microbiol. 50, 1339–1347 (2003). 68. Shaevitz, J. W., Lee, J. Y. & Fletcher, D. A. Spiroplasma Presented evidence that the 669 kDa protein SprB
48. Bardy, S. L. & Jarrell, K. F. FlaK of the archaeon swim by a processive change in body helicity. Cell 122, is a moving cell surface component of the
Methanococcus maripaludis possesses preflagellin 941–945 (2005). F. johnsoniae gliding motility machinery.
peptidase activity. FEMS Microbiol. Lett. 208, 53–59 69. Waterbury, J. B., Willey, J. M., Franks, D. G., Valois, 93. Liu, J., McBride, M. J. & Subramaniam, S. Cell
(2002). F. W. & Watson, S. W. A cyanobacterium capable of surface filaments of the gliding bacterium
49. Albers, S. V., Szabo, Z. & Driessen, A. J. Archaeal swimming motility. Science 230, 74–76 (1985). Flavobacterium johnsoniae revealed by cryo-
homolog of bacterial type IV prepilin signal peptidases 70. Willey, J. M., Waterbury, J. B. & Greenberg, E. P. electron tomography. J. Bacteriol. 189,
with broad substrate specificity. J. Bacteriol. 185, Sodium-coupled motility in a swimming cyanobacterium. 7503–7506 (2007).
3918–3925 (2003). J. Bacteriol. 169, 3429–3434 (1987). 94. Wall, D. & Kaiser, D. Type IV pili and cell motility. Mol.
50. Ng, S. Y., Chaban, B. & Jarrell, K. F. Archaeal flagella, 71. McCarren, J. & Brahamsha, B. SwmB, a 1.12- Microbiol. 32, 1–10 (1999).
bacterial flagella and type IV pili: a comparison of megadalton protein that is required for nonflagellar 95. Wolgemuth, C., Hoiczyk, E., Kaiser, D. & Oster, G.
genes and posttranslational modifications. J. Mol. swimming motility in Synechococcus. J. Bacteriol. How myxobacteria glide. Curr. Biol. 12, 369–377
Microbiol. Biotechnol. 11, 167–191 (2006). 189, 1158–1162 (2007). (2002).

nature reviews | microbiology volume 6 | June 2008 | 475

© 2008 Nature Publishing Group
REVIEWS

96. Hoiczyk, E. & Baumeister, W. The junctional pore 112. Szurmant, H. & Ordal, G. W. Diversity in chemotaxis 132. Walsby, A. E., Ng, G., Dunn, C. & David, P. A.
complex, a prokaryotic secretion organelle, is the mechanisms among the bacteria and archaea. Comparison of the depth where Planktothrix
molecular motor underlying gliding motility in Microbiol. Mol. Biol. Rev. 68, 301–319 (2004). rubescens stratifies and the depth where the daily
cyanobacteria. Curr. Biol. 8, 1161–1168 (1998). 113. Turner, L., Ryu, W. S. & Berg, H. C. Real-time imaging insolation supports its neutral bouyancy. New Phytol.
97. Yu, R. & Kaiser, D. Gliding motility and polarized of fluorescent flagellar filaments. J. Bacteriol. 182, 162, 133–145 (2004).
slime secretion. Mol. Microbiol. 63, 454–467 2793–2801 (2000). 133. Oliver, R. L. & Walsby, A. E. Direct evidence for the
(2007). 114. Rudolph, J. & Oesterhelt, D. Deletion analysis of the role of light-mediated gas vesicle collapse in the
98. White, D. J. & Hartzell, P. L. AglU, a protein required che operon in the archaeon Halobacterium salinarium. bouyancy regulation of Anabaena flos-aquae
for gliding motility and spore maturation of J. Mol. Biol. 258, 548–554 (1996). (cyanobacteria). Limnol. Oceanogr. 29, 879–886
Myxococcus xanthus, is related to WD‑repeat 115. Rudolph, J., Tolliday, N., Schmitt, C., Schuster, S. C. & (1984).
proteins. Mol. Microbiol. 36, 662–678 (2000). Oesterhelt, D. Phosphorylation in halobacterial signal 134. Macnab, R. M. The bacterial flagellum: reversible
99. Youderian, P., Burke, N., White, D. J. & Hartzell, P. L. transduction. EMBO J. 14, 4249–4257 (1995). rotary propellor and type III export apparatus.
Identification of genes required for adventurous 116. Zusman, D. R., Scott, A. E., Yang, Z. & Kirby, J. R. J. Bacteriol. 181, 7149–7153 (1999).
gliding motility in Myxococcus xanthus with the Chemosensory pathways, motility and development in
transposable element mariner. Mol. Microbiol. 49, Myxococcus xanthus. Nature Rev. Microbiol. 5, Acknowledgements
555–570 (2003). 862–872 (2007). The authors thank the members of their laboratories for
100. Mignot, T., Shaevitz, J. W., Hartzell, P. L. & Zusman, 117. Mignot, T., Merlie, J. P. Jr & Zusman, D. R. Regulated helpful discussions and the researchers who generously sup-
D. R. Evidence that focal adhesion complexes power pole‑to‑pole oscillations of a bacterial gliding motility plied videos. Research in the authors laboratories is sup-
bacterial gliding motility. Science 315, 853–856 protein. Science 310, 855–857 (2005). ported by grants from the National Science Foundation
(2007). 118. Leonardy, S., Freymark, G., Hebener, S., Ellehauge, E. (MCB‑0641366) to M.J.M. and from the Natural Sciences
Showed that the M. xanthus adventurous gliding & Sogaard-Andersen, L. Coupling of protein and Engineering Research Council of Canada to K.F.J.
motility protein AglZ remained fixed relative to the localization and cell movements by a dynamically
substratum as cells glided. These data suggest a localized response regulator in Myxococcus xanthus.
model of cell movement that involves intracellular EMBO J. 26, 4433–4444 (2007).
motors that exert force on the cytoskeleton and on 119. Bonner, P. J. & Shimkets, L. J. Phospholipid directed DATABASES
cell surface adhesion complexes. motility of surface-motile bacteria. Mol. Microbiol. 61, Entrez Genome Project: http://www.ncbi.nlm.nih.gov/
101. Heintzelman, M. B. Cellular and molecular mechanics 1101–1109 (2006). entrez/query.fcgi?db=genomeprj
of gliding locomotion in eukaryotes. Int. Rev. Cytol. 120. Beard, S. J., Hayes, P. K., Pfeifer, F. & Walsby, A. E. The Bdellovibrio bacteriovorus | Borrelia burgdorferi | Caulobacter
251, 79–129 (2006). sequence of the major gas vesicle protein, GvpA, crescentus | Clostridium perfringens | Escherichia coli |
102. Miyata, M. Centipede and inchworm models to explain influences the width and strength of halobacterial gas Flavobacterium johnsoniae | Halobacterium salinarum |
Mycoplasma gliding. Trends Microbiol. 16, 6–12 vesicles. FEMS Microbiol. Lett. 213, 149–157 (2002). Helicobacter pylori | Listeria monocytogenes | Methanococcus
(2008). 121. Walsby, A. E. Gas vesicles. Microbiol. Rev. 58, maripaludis | Methanococcus voltae | Mycoplasma mobile |
103. Hasselbring, B. M. & Krause, D. C. Cytoskeletal 94–144 (1994). Mycoplasma pneumoniae | Myxococcus xanthus | Neisseria
protein P41 is required to anchor the terminal 122. Mlouka, A., Comte, K., Castets, A. M., Bouchier, C. & gonorrhoeae | Nostoc punctiforme | Proteus mirabilis |
organelle of the wall-less prokaryote Mycoplasma Tandeau de Marsac, N. The gas vesicle gene cluster Pseudomonas aeruginosa | Salmonella typhimurium | Shigella
pneumoniae. Mol. Microbiol. 63, 44–53 (2007). from Microcystis aeruginosa and DNA flexneri | Vibrio parahaemolyticus
104. Uenoyama, A. & Miyata, M. Identification of a 123- rearrangements that lead to loss of cell buoyancy. Entrez Protein: http://www.ncbi.nlm.nih.gov/entrez/query.
kilodalton protein (Gli123) involved in machinery for J. Bacteriol. 186, 2355–2365 (2004). fcgi?db=protein
gliding motility of Mycoplasma mobile. J. Bacteriol. 123. Offner, S., Hofacker, A., Wanner, G. & Pfeifer, F. Eight BfpE | FlgN | FliA | FliG | FliI | FliJ | FliM | FliN | FliS | FliT | MotA |
187, 5578–5584 (2005). of fourteen gvp genes are sufficient for formation of MotB | P42 | PilA | PilB | PilD | TolB | TolQ | TolR
105. Seto, S., Uenoyama, A. & Miyata, M. Identification of gas vesicles in halophilic archaea. J. Bacteriol. 182,
a 521-kilodalton protein (Gli521) involved in force 4328–4336 (2000). FURTHER INFORMATION
generation or force transmission for Mycoplasma 124. DasSarma, S., Arora, P., Lin, F., Molinari, E. & Yin, Ken F. Jarrell’s homepage: http://microimm.queensu.ca/
mobile gliding. J. Bacteriol. 187, 3502–3510 (2005). L. R. Wild-type gas vesicle formation requires at least people/faculty/Jarrell%20Lab/default.htm
106. Uenoyama, A. & Miyata, M. Gliding ghosts of ten genes in the gvp gene cluster of Halobacterium Mark J. McBride’s homepage: http://www.uwm.edu/Dept/
Mycoplasma mobile. Proc. Natl Acad. Sci. USA 102, halobium plasmid pNRC100. J. Bacteriol. 176, Biology/Docs/Faculty/mcbride.html
12754–12758 (2005). 7646–7652 (1994). Howard Berg’s laboratory website (movies: bacteria
Presented evidence that the energy source for 125. Scheuch, S. & Pfeifer, F. GvpD-induced breakdown of swarming; type IV pili 1 and type IV pili 2): http://www.
M. mobile gliding is ATP. Cells were permeabilized the transcriptional activator GvpE of halophilic archaea rowland.harvard.edu/labs/bacteria/movies/fluo_fil_leave.
with Triton X‑100 to form non-motile and non-viable requires a functional p‑loop and an arginine-rich region movhttp://www.rowland.harvard.edu/labs/bacteria/
ghosts that regained motility when ATP was added of GvpD. Microbiology 153, 947–958 (2007). showmovie.php?mov=smarc_p_swarm_1; http://www.
exogenously. 126. Offner, S., Ziese, U., Wanner, G., Typke, D. & Pfeifer, F. rowland.harvard.edu/labs/bacteria/showmovie.
107. Ohtani, N. & Miyata, M. Identification of a novel Structural characteristics of halobacterial gas vesicles. php?mov=paeru_f_move_1;
nucleoside triphosphatase from Mycoplasma mobile: a Microbiology 144, 1331–1342 (1998). http://www.rowland.harvard.edu/labs/bacteria/showmovie.
prime candidate motor for gliding motility. Biochem. J. 127. Walsby, A. E. & Dunton, P. G. Gas vesicles in php?mov=paeru_f_move_2
403, 71–77 (2007). actinomycetes? Trends Microbiol. 14, 99–100 (2006). Joshua Shaevitz’s laboratory website (movie: Spiroplasma
108. Seybert, A., Herrmann, R. & Frangakis, A. S. 128. Pfeifer, F. et al. Gas vesicle formation in halophilic kinking): http://genomics.princeton.edu/shaevitzlab/
Structural analysis of Mycoplasma pneumoniae by Archaea. Arch. Microbiol. 167, 259–268 (1997). Movies/Spiroplasma%20Swimming.mov
cryo-electron tomography. J. Struct. Biol. 156, 129. Dunton, P. G., Mawby, W. J., Shaw, V. A. & Walsby, Keiichi Namba’s laboratory website (movie: flagella
342–354 (2006). A. E. Analysis of tryptic digests indicates regions of assembly):
109. Henderson, G. P. & Jensen, G. J. Three-dimensional GvpC that bind to gas vesicles of Anabaena http://www.fbs.osaka-u.ac.jp/labs/namba/npn/movie5.html
structure of Mycoplasma pneumoniae’s attachment flos-aquae. Microbiology 152, 1661–1669 (2006). Nyles Charon’s laboratory website (movie: Borrelia
organelle and a model for its role in gliding motility. 130. Buchholz, B. E., Hayes, P. K. & Walsby, A. E. The swimming): http://www.hsc.wvu.edu/som/micro/faculty/
Mol. Microbiol. 60, 376–385 (2006). distribution of the outer gas vesicle protein, GvpC, on Charon/charon_spirochete.asp#Barrel_swim
110. Martinez, A., Torello, S. & Kolter, R. Sliding motility the Anabaena gas vesicle, and its ratio to
in mycobacteria. J. Bacteriol. 181, 7331–7338 GvpA. J. Gen. Microbiol. 139, 2353–2363 (1993). SUPPLEMENTARY INFORMATION
(1999). 131. Shukla, H. D. & DasSarma, S. Complexity of gas See online article: S1 (movie) | S2 (movie) | S3 (movie) |
111. Stevens, J. M., Galyov, E. E. & Stevens, M. P. Actin- vesicle biogenesis in Halobacterium sp. strain NRC‑1: S4 (movie)
dependent movement of bacterial pathogens. Nature identification of five new proteins. J. Bacteriol. 186, All links are active in the online pdf
Rev. Microbiol. 4, 91–101 (2006). 3182–3186 (2004).

476 | June 2008 | volume 6 www.nature.com/reviews/micro

© 2008 Nature Publishing Group