Dynamic Filaments of the Bacterial Cytoskeleton · 2006-04-19 · terial cytoskeleton. This revolution in the understanding of bacterial cell structure and dynamics has come about

ANRV277-BI75-21 ARI 24 February 2006 2:50

RE V I E W

S

IN

A D V A

NC

E

Dynamic Filaments of theBacterial CytoskeletonKatharine A. Michie and Jan LoweMedical Research Council Laboratory of Molecular Biology, Cambridge CB2 2QH,UK; email: [email protected], [email protected]

Annu. Rev. Biochem.2006. 75:467–92

The Annual Review ofBiochemistry is online atbiochem.annualreviews.org

doi: 10.1146/annurev.biochem.75.103004.142452

Copyright c! 2006 byAnnual Reviews. All rightsreserved

0066-4154/06/0707-0467$20.00

Key WordsFtsZ, MreB, ParM, tubulin, actin

AbstractBacterial cells contain a variety of structural filamentous proteinsnecessary for the spatial regulation of cell shape, cell division, andchromosome segregation, analogous to the eukaryotic cytoskele-tal proteins. The molecular mechanisms by which these proteinsfunction are beginning to be revealed, and these proteins show nu-merous three-dimensional structural features and biochemical prop-erties similar to those of eukaryotic actin and tubulin, revealing theirevolutionary relationship. Recent technological advances have illu-minated links between cell division and chromosome segregation,suggesting a higher complexity and organization of the bacterial cellthan was previously thought.

467

First published online as a Review in Advance on March 29, 2006

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


Cytoskeleton: theproteinaceousfilaments thatprovide intracellularorganization

Protofilament: alinear structuralprecursor assembledfrom protein that isable to assemble intoa largersuperstructure

ContentsINTRODUCTION. . . . . . . . . . . . . . . . . 468THE TUBULIN HOMOLOGUES . 468

FtsZ Protein . . . . . . . . . . . . . . . . . . . . . 468BtubA/B . . . . . . . . . . . . . . . . . . . . . . . . . 475

THE ACTIN HOMOLOGUES . . . . 476Actin . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476ParM . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477MreB . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479FtsA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482

INTERMEDIATE FILAMENTHOMOLOGUE . . . . . . . . . . . . . . . . . 483Crescentin . . . . . . . . . . . . . . . . . . . . . . . 483

WALKER A CYTOSKELETALATPASE—A NEW FAMILY OFCYTOSKELETAL PROTEINS? 483

INTRODUCTIONResearch over the past two decades has uncov-ered the existence of a well-developed bac-terial cytoskeleton. This revolution in theunderstanding of bacterial cell structure anddynamics has come about largely throughthe development of fluorescent labeling tech-niques for bacterial cells and the availabil-ity of complete genome sequences, togetherwith structural and biochemical studies. Wenow know that proteinaceous filaments en-circle and wind helically around the insideof the cell, providing intracellular organiza-tion and cytoskeletal functions reminiscent ofthose in eukaryotic cells. Although the discov-ery of a well-organized bacterial cytoskele-ton caused much excitement, another sur-prise came when sequence analysis and/or thethree-dimensional structural determinationof many of the proteins involved revealed theirhomology to eukaryotic cytoskeletal proteinsactin, tubulin, and those comprising inter-mediate filaments. The eukaryotic proteinsperform a myriad of important functions, in-cluding establishing cell shape, providing me-chanical strength, contributing to cell loco-motion, assisting in the intracellular transport

of organelles, as well as bringing about chro-mosome separation during mitosis and meio-sis. It is becoming evident that the bacterialhomologues have analogous or overlappingfunctions.

This review presents the current under-standing of the most well-characterized bac-terial cytoskeletal proteins with a focus ontheir biochemistry (for reviews on the cell bi-ology of the bacterial cytoskeleton, see Ref-erences 1–3). In particular, we focus on pro-teins whose assembly into dynamic filamentsis regulated by cycles of nucleotide bindingand hydrolysis. An intermediate filament ho-mologue is also discussed; however, interme-diate filaments are not dynamic and are notfound in all eukaryotic cells—being requiredmainly for mechanical strength.

THE TUBULIN HOMOLOGUESTubulin is an indispensable eukaryotic cy-toskeletal protein involved in many mechani-cal cellular processes (examples include chro-mosome segregation during mitosis and vesic-ular transport). Heterodimeric !"-tubulinforms protofilaments that combine to makedynamic microtubules. Microtubules form“scaffolds” that motor proteins such as dyneinand kinesin are able to track along.

FtsZ ProteinThe ftsZ gene (4) encodes a guanosinetriphosphatase (GTPase) (5, 6) that is essentialfor cell division (7, 8). FtsZ was suggested tobe a cytoskeletal protein (9) and was predictedto be a homologue of tubulin on the basis ofa short sequence motif of seven amino acids(10–12). It is a highly conserved protein thatis found in virtually all eubacteria and archaea(13), with a few exceptions (14–16). FtsZ isalso present in some chloroplasts and mito-chondria (17, 18).

FtsZ in vivo. FtsZ is the first protein knownto localize to the mid-cell position priorto septum invagination during cell division,

468 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


remaining positioned as a ring structure (theZ ring) at the leading edge of the constrict-ing division septum (9). Time-lapse images ofcells expressing FtsZ-green fluorescent pro-tein (GFP) fusions show the Z ring contract-ing while septum constriction occurs (19).Under some circumstances, FtsZ has been ob-served to form helical structures in cells (20–25), leading to the suggestion that the matureZ ring is a compressed helix (26).

FtsZ is an abundant protein [with esti-mates of between 3200 and possibly 15,000molecules per Escherichia coli cell (27, 28) and5000 for Bacillus subtilis (29)]. In vitro assem-bly of FtsZ and its homology to tubulin (dis-cussed below) has lead to the idea that FtsZassembles into linear polymers called protofil-aments. A filament formed by end-to-end as-sociation of FtsZ monomers (40 A in length)would require fewer than 500 monomers tospan the circumference of a typical bacterialcell with a 1-µm diameter, but it is likely thatthe Z ring is more complex in structure than asingle protofilament. Instead, the Z ring maycontain several of these protofilaments associ-ated (or bundled) together to form the Z-ringsuperstructure. Consistent with this, fluores-cence recovery after photobleaching (FRAP)experiments indicate that an average of 30%of the total cellular FtsZ is assembled into theZ ring, more than enough to comprise a Zring of several FtsZ filaments (30).

FtsZ displays a dynamic localization in thatit exchanges between the Z ring and the cy-toplasmic pool on a timescale of seconds (30,31). Time-lapse movies of FtsZ assemblies invivo have shown FtsZ rearranging from heli-cal structures to rings and vice versa (21, 22,26). However, as yet, there is no detailed in-formation regarding the large-scale structureof FtsZ polymers in vivo, and this significantlylimits our understanding of FtsZ action.

FtsZ self-assembles in vitro. FtsZ protofil-aments have been assembled in vitro in anucleotide-dependent manner (32, 33). As-sembly with either nonhydrolyzable GM-PCPP [guanylyl-(!,")-methylene diphos-

FRAP: fluorescencerecovery afterphotobleaching

phate] or with GDP indicates that nucleotidehydrolysis is not required for assembly (34,35). Filaments of FtsZ have been studiedby negative stain electron microscopy (EM),and a wide variety of polymer morphologieshave been observed (including tubules, sheets,asters, straight and curved protofilaments, andminirings) (32, 34, 36–40). The wide range ofconditions favoring FtsZ polymerization sug-gests that FtsZ could assemble unassisted intoa polymer in vivo; however, the diversity of su-perstructures observed for FtsZ in vitro indi-cate that some (or all) of the filament types areprobably artifactual. The question is which, ifany, are representative of the in vivo situation.FtsZ polymers show highly dynamic and flex-ible behavior in vitro (41–43), and a recentanalysis by atomic force microscopy (42) hasshown that the dynamic FtsZ filaments con-tinuously rearrange. Also, end-to-end join-ing of FtsZ filaments and depolymerizationof FtsZ from within the middle of filamentshas been observed in vitro (42).

There are a large number of in vitro FtsZfilament morphologies, but at low concentra-tions, FtsZ assembles into apparently singleprotofilaments. Coupled with the high abun-dance of FtsZ in the cell, it is thought thatthe in vivo form of FtsZ is likely to be com-posed of linear filaments of FtsZ laterally as-sociated with a defined topology. Our attemptto understand the Z-ring structure in vivois complicated by the presence of accessoryand regulatory proteins. At least eight pro-teins (FtsA, ZipA, ZapA, EzrA, Noc, SlmA,MinC, and SulA) affect FtsZ assembly eitherby direct or indirect interactions in vivo. Someof these proteins are restricted to a limitednumber of organisms (for example ZipA isonly present in the # subdivision of the gram-negative bacteria) and are not well conserved,implying their mechanisms have evolved in-dependently. These accessory proteins mayinhibit Z-ring assembly, assist in the correctpositioning of the Z ring, or directly effect thedynamics of the Z-ring structure. Very littleis known about how these proteins exert theirmolecular effect on FtsZ [with the exception

www.annualreviews.org • Bacterial Cytoskeleton 469

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


Lateral interaction:any interactionbetweenprotofilaments

of SulA, whose mode of action is to titrateaway monomeric FtsZ by binding to one ofthe polymerization interfaces (44)], and theirbiochemistry is beyond the scope of this re-view. (The reader is referred to References 45and 46 for a review of these proteins.) Interest-ingly, current structural and sequence analysisof these proteins has failed to reveal homologyto any of the eukaryotic tubulin accessory pro-teins. It is possible that the accessory proteinsevolved after the evolutionary split of prokary-otes and eukaryotes: this is consistent with thehighly extended evolutionary histories specu-lated for the large divergence in eukaryoticand bacterial actins and tubulins (47, 48).

Subunit structure. Although FtsZ’s primarysequence identity to tubulin is low (10% to18%), it is now generally accepted that FtsZis a true prokaryotic homologue of tubulin be-cause the three-dimensional tertiary structurerevealed that the two proteins share the samefold (Figure 1) and both proteins assembleinto remarkably similar protofilaments (49–52). FtsZ comprises two domains (52), re-flecting thermal denaturation profiles of FtsZfrom Methanococcus jannaschii and E. coli, whichboth show a clear two-step unfolding process,whereby the domains first separate from eachother and then unfold completely (53, 54). Ithas been suggested that these two domainswere derived from two separate proteins ear-lier in evolution because the N-terminal do-main has a Rossmann fold similar to thatof many ATPases, and the C-terminal do-main is homologous to the family of cho-rismate mutase-like proteins (52). The N-terminal domain contains the central helix H7and is essentially the nucleotide-binding do-main. Compared with tubulin, this domain ofFtsZ has an extra helix (called H0) that pro-trudes from the N terminus and shows highvariability across FtsZ sequences (49). TheC-terminal domain of FtsZ shows much lessconservation than the N terminus when com-pared with tubulin. The C-terminal domainis also considerably shorter. In both proteins,this domain contains some residues impor-

tant for GTP hydrolysis (51), which occursduring assembly into filaments (see below).It is interesting to note that FtsZ also has aconserved hydrophobic pocket similar to thetaxol-binding pocket in tubulin at the inter-face of the N- and C-terminal domains, im-mediately adjacent to helix H7 and the activesite. In tubulin, this binding pocket lies on theinside face of microtubules in a region thoughtto mediate lateral contacts (55). Filling thispocket with an as yet unknown natural acces-sory protein may help stabilize the filamentsin the “straight,” assembly-competent con-formation (see below). It is thought that thenatural binding partner for this pocket intubulin might be a microtubule-associatedprotein (MAP) (56), although this is contro-versial (57). Regardless of the eukaryotic sub-strate, FtsZ has a homologous pocket, andwhether proteins (such as Zap, ZipA, andFtsA) with functions analogous to MAPs thatmight stabilize FtsZ assembly bind in a similarfashion has yet to be determined.

Filament structure. Semicontinuous tubu-lin-like protofilaments of FtsZ have been suc-cessfully crystallized (52) (Figure 1, right),providing us with insight into how FtsZ mayassemble into protofilaments. Not unexpect-edly, the data suggest that FtsZ assemblesin an orientation very similar to that ob-served for polymerized tubulin, with eachFtsZ monomer maintaining head-to-tail in-teractions. These head-to-tail interactions arereferred to as longitudinal contacts and are thebasis of protofilament formation. All other in-teractions are referred to as lateral and func-tion to bring protofilaments together. Lateralinteractions may play important roles in Z-ring nucleation, assembly, regulation, and dis-assembly. Although the regions required forlateral tubulin-tubulin interactions within mi-crotubules are known, the corresponding re-gions in FtsZ are quite different, with littleconservation observed in the relatively shortloop regions of FtsZ (51), which is consistentwith the notion that FtsZ and tubulin do not

470 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


Figure 1Structures of the !/"-tubulin heterodimer (left), the BtubA/BtubB heterodimer (center) and FtsZ dimer(right), showing the position of the nucleotide at the dimer interface, the conservation of fold, and theaxis of protofilament extension (up the page). Lateral interactions between protofilaments could be formedat all or any of the interfaces perpendicular to the longitudinal axis of protofilament assembly. (left) The!/"-tubulin heterodimer observed in tubulin zinc sheets [Protein Data Bank (PDB) entry 1JFF] (172).(center) BtubA/BtubB heterodimer from Prosthecobacter dejongeii (PDB entry 2BTQ) (88). (right) FtsZdimer obtained from nucleotide-free FtsZ from Methanococcus jannaschii soaked in MgGTP (PDB entry1W5A) (52).

share similar lateral interactions or accessoryproteins.

Active site. The GTPase active site isformed by the association of two FtsZmonomers, with the catalytic T7 loop [or syn-ergy loop (58)] in the C-terminal domain ofone monomer inserting into the nucleotide-binding pocket of the N-terminal domain

of the adjacent molecule (Figure 1, right),thereby leading to association-dependent ac-tivation of the GTPase activity (51, 52, 59).Catalysis occurs by the polarization of a watermolecule hydrogen bonded to two conservedaspartate side chains (M. jannaschii residues235 and 238) within the T7 loop, promot-ing nucleophilic attack on the #-phosphateand thus hydrolysis of GTP. A further


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


Longitudinalinteraction: aninteractionresponsible for theformation ofprotofilaments

Dynamicinstability: theswitching ofbiological proteinpolymers betweenphases of steadyelongation and rapidshortening

Isodesmicassembly: allintersubunit contactsare equivalent

contribution to the polarization of the #-phosphate is provided by a magnesium ion co-ordinated by glutamine (M. jannaschii residue75), several water molecules, and the !,"phosphates. Thus, GTP hydrolysis requiresMg2+, consistent with in vitro observation (5,6, 12).

What can we learn about FtsZ from tubu-lin? Although FtsZ is similar to tubulin instructure, there are some important differ-ences between the two proteins. Microtubulesare comprised of !- and "-tubulin, whichform a tight !/" heterodimer in solution.The !/" heterodimer has a nonhydrolyzed,nonexchanging GTP bound at the dimer in-terface. In contrast, most bacteria have onlyone isoform of FtsZ (60–62), and thus eachsubunit interface is equivalent.

Tubulin assembly characteristics. Micro-tubules are assembled from !/" heterodimersjoined end-to-end so that the ! and " iso-

Protofilament axis

Isodesmic

Cooperative(allosteric)

Cooperative(lateral)

Figure 2Schematic description of isodesmic and cooperative assemblymechanisms. The isodesmic model (top) assumes all subunit additions areequivalent, and thus the likelihood of assembly is directly proportional tothe concentration of protein. Cooperative assembly (subdivided intoallosteric and lateral) occurs when subunit additions are not equivalent. Inthe case of allosteric cooperative assembly (center), the nucleation of adimer results in a conformational change within the dimer, whichincreases the affinity for the next subunit to bind. In the case of lateralcooperative assembly (bottom), assembly of more than two molecules isstabilized by a third molecule at a different interface; in the case of tubulinand FtsZ, this is known as a lateral protofilament interaction.

forms of tubulin alternate, with a second, hy-drolyzable GTP molecule between each het-erodimer subunit. After assembly, the nu-cleotide within the subunit cannot exchange,nor can subunits of tubulin within the fila-ment exchange with the cytoplasmic supply.In addition to the longitudinal interactionsbetween !/" heterodimers, extensive lateralassociations between protofilaments furtherstabilize microtubule assembly, and a com-plete microtubule is comprised of 13 parallelprotofilaments. Tubulin protofilaments havea distinct polarity because of the head-to-tailassociation of tubulin subunits, and owing tothe alternating ! and " forms of tubulin, "-tubulin is always present at one end (desig-nated the plus end or fast-growing end), and!-tubulin is present at the other end (desig-nated the minus end).

Microtubules exhibit dynamic instability,enabling them to disassemble rapidly in vivo.GTP hydrolysis drives the protofilaments to-ward a bent or curved polymeric state that isincompatible with the geometry of the mi-crotubule wall. Kinetic stability of the micro-tubule is maintained by a GTP-bound capthat restrains and stabilizes the polymer inthe straight conformation. If this cap is hy-drolyzed, the tubulin filaments can adopt thecurved or bent morphology, resulting in spon-taneous disassembly of the filament. Thus, thestate of the GTP cap determines how micro-tubules switch between states of rapid growthand rapid shrinkage (63).

Isodesmic versus cooperativity assembly.Both actin and tubulin assemble via coopera-tive mechanisms in which nucleation is a ratelimiting step (Figure 2). As a consequence,polymer assembly can be controlled by pro-viding specific nucleation sites at a definedtime and place in the cell. For an isodesmicassembly mechanism (Figure 2, top), poly-mer will rapidly assemble and disassemble atall places in the cell, and the cell must providestabilization factors (to assemble the polymerinto a defined structure) and topological in-formation (to position it correctly). Research

472 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


into FtsZ assembly has so far failed to establishconclusively the kinetics and mechanism ofFtsZ assembly. Significant factors limiting ourunderstanding are the wide variation in FtsZbehavior under different experimental condi-tions and the consequent uncertainty aboutwhich of the observed behaviors may be rele-vant in vivo.

Isodesmic assembly occurs when linearmultimers form in which each bond has anidentical contact, and nucleation is just as fa-vored as filament extension (Figure 2, top).Cooperative assembly occurs when multimersof protein only become stable after formingan unfavored but defined nucleus, whereinthe bonds between the subunits are rela-tively weak and initiation is difficult. Oncethe unfavored nucleation step occurs, the fil-ament can extend rapidly because subunits inthe larger structure are stabilized by multi-ple bonds formed with other adjoining sub-units (Figure 2, middle and bottom). Thereare three characteristics typical of coopera-tive assembly: a critical concentration for as-sembly, a lag in the assembly kinetics at lowprotein concentration, and a distribution ofsubunits at equilibrium into two distinct pop-ulations of monomers and very long polymers.

A cooperative assembly model for FtsZprotofilaments is now generally accepted forthree reasons. First, both FtsZ assembly andGTPase activity have a critical concentra-tion (64, 65). Second, FtsZ assembly showsan !1 s lag of assembly even at very highprotein concentrations, suggesting an “activa-tion” step. This might be the time required torelease GDP and bind GTP (66). Third, a sizelimit and relatively homogeneous protofil-ament length for FtsZ have been reported(67).

Results from analytical ultracentrifu-gation and scanning transmission electronmicroscopy suggest that the FtsZ protofil-ament is formed by a single chain of FtsZmonomers associated head to tail (68, 69).However, if the FtsZ molecules are rigid andif there is no communication between thetwo binding faces on a single monomer, then

linear protofilament formation cannot becooperative.

There are several models that attempt toreconcile the observed association into singleprotofilaments with the data showing coop-erative assembly. It is possible that, initially,isodesmic assembly into single protofilamentsoccurs, followed by cooperative association ofprotofilaments (such that the cooperative ki-netics overwhelm the effects of the isodesmicassembly), giving rise to an apparent overallcooperative assembly. However, this model isunlikely to be true because cooperative as-sembly is still observed at low concentrationswhen no bundling or grouping of protofila-ments is detected (64). Mingorance et al. (42)have suggested that the isodesmic assemblyof protofilaments is followed by a stabiliz-ing cyclization event that they argue wouldshow overall cooperative kinetics. They sup-port this hypothesis with images of cyclic sin-gle protofilaments formed in vitro; however,the biological relevance of the observed smallring protofilaments is unclear.

Finally, if we abandon the assumption thatFtsZ is a rigid molecule and that binding nu-cleotide or another subunit could induce aconformational change, then it is also possiblethat FtsZ exhibits cooperativity within a singleprotofilament. In this case, the initial dimer-ization process would cause a conformationalchange that increases the affinity of the nextmolecule to bind. Such a model could explainthe observed cooperative kinetics reported forsingle protofilaments of FtsZ.

Nucleotide exchange. The structural datafor the FtsZ filament strongly suggests thatthere are major differences in the solvent ac-cessibility of the nucleotide pocket of FtsZcompared with tubulin; however, this datawas obtained from crystals in which the nu-cleotide was soaked in and may not representthe bona fide nucleotide-bound state of theprotofilament (52). Within tubulin protofila-ments, the nucleotide-binding pockets are oc-cluded, and nucleotide exchange is prohibited.This characteristic is essential for the dynamic


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


instability of tubulin polymers and providesthe stabilizing GTP cap.

By contrast, the FtsZ nucleotide-bindingpocket is partially exposed, potentially allow-ing nucleotide exchange, which could makenucleotide hydrolysis the rate-limiting step infilament disassembly (52, 70). This potentialfor nucleotide exchange may have importantimplications for FtsZ assembly and dynamics.If significant nucleotide exchange can occur,FtsZ filaments would not experience dynamicinstability because the high ratio of GTP toGDP in the cell (71) would ensure that everymolecule in the filament is in the GTP-boundstate.

Data suggesting that FtsZ is unable toexchange nucleotides and has a GTP caphave been reported (41, 72), consistent withthe tubulin paradigm of dynamic instability.However, other strong data suggest that nu-cleotide exchange does occur, thus excludingthe possibility of dynamic instability (70, 73).Some attempts to determine the nucleotidestate within filaments have suggested that themajority of FtsZ in protofilaments is boundto GTP (70, 73), although other data suggestthe contrary (74). The conflicting data mostprobably arise from the different conditionsused in the various assays, such as Ca2+ con-centrations (discussed below) and bundlingstates of the filaments that may sterically in-hibit nucleotide exchange. In particular, theinitial form of the soluble protein may be en-tirely monomeric in some investigations butinclude dimers and larger oligomers in othercases. Some mutants with reduced GTPaseactivity are able to support cell division, al-though with much slower dynamics (30, 75,76), suggesting that either GTPase activity ismodulated in vivo, or GTPase activity is muchhigher than required to support division. Theobserved rates of FtsZ assembly appear to dif-fer greatly between different organisms (77),but these differences may also arise from vary-ing experimental conditions.

The presence of Ca2+ reduces FtsZ’s GT-Pase activity in vitro (40, 43, 78) and in-creases bundling of the protofilaments (40,

43, 78, 79). Ca2+ has also been reported toreduce the exchange of nucleotide (73). Thiseffect on bundling observed with Ca2+ maysimply be an indirect effect of a decrease inGTPase activity. The decreased GTPase ac-tivity would lead to an increased stability offilaments. Long filaments should be presentfor a longer period of time, and this shouldresult in conditions promoting lateral asso-ciation. It is also possible that bundling maysterically inhibit nucleotide hydrolysis. Alter-natively, increased bundling may be due toa more specific mechanism involving a cur-rently unknown Ca2+-binding site (40, 80).

Conformational change. It is thought thatnucleotide hydrolysis brings about the dis-assembly of microtubules by a mechanismlinked to a conformational change (81–83).In this hypothesis, the GDP-bound form oftubulin adopts a bent or curved form [ob-served experimentally (81)] that destabilizeslateral bonds in the microtubule, resulting inthe peeling back of curved filaments from theend of a microtubule and the eventual disas-sembly of the filaments by heterodimer dis-sociation. Bending occurs at all interfaces inthe tubulin protofilament (83), even betweenthe two subunits of each heterodimer wherethe GTP is never hydrolyzed; thus bendingneed not be absolutely linked to nucleotidehydrolysis (see below).

Early work observed that FtsZ protofila-ment disassembly was concomitant with rapidnucleotide hydrolysis (33). FtsZ in the GTP-bound state predominantly forms straightprotofilaments (32, 36), whereas GDP-boundFtsZ forms curved protofilaments (37, 65, 69,84, 85). This observation has led to the sug-gestion that nucleotide hydrolysis causes aconformational change in the FtsZ filamentthat may be involved in converting the chem-ical energy of nucleotide hydrolysis into me-chanical energy for constricting the Z ring(35). It has been predicted, on the basis ofin silico modeling, that the T3 loop in thenucleotide-binding site will undergo a large

474 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


conformational change upon the GDP-to-GTP transition (86); however, no large con-formational changes have been observed inthe various cocrystal structures of FtsZ withGDP or GTP (52). It is important to notethat FtsZ filaments formed in vitro do not al-ways show consistent morphologies with thetype of bound nucleotide (42). It has been sug-gested that the hydrolysis of GTP to GDPsimply produces a repulsive electrostatic ef-fect with the loss of the # phosphate and thatthis chemical repulsion is the mechanism be-hind disassembly (52, 68).

FtsZ function and disassembly in vivo maybe regulated by the combination of a confor-mational change and an electrostatic effect.However, the oversimplified “spring” modelin analogy to the GDP-depolymerization oftubulin is unlikely to reflect the real mech-anism of FtsZ function in vivo. The fila-ment energy formed by large superstructures(where the potential binding surfaces of thesubunits are large) is very significant and canovercome almost all other molecular effects.For example, although FtsZ may prefer toadopt a bent conformation when bound toGDP in a free protofilament, the binding en-ergy that accompanies favorable lateral asso-ciation of protofilaments in a certain filamen-tous form of FtsZ might be large enough torestrain the FtsZ protofilament so that it isstraight. Again, only elucidation of the in vivosuperstructure of FtsZ will allow us to deter-mine which of the behaviors observed in vitrois biologically relevant.

BtubA/BTwo tubulin homologues (BtubA and BtubB)have been identified recently in the Prosthe-cobacter bacterial genus, and both show a closerrelationship to eukaryotic tubulin than to FtsZ(87). BtubA is 31% to 35% identical andBtubB is 34% to 37% identical to !- and"-tubulin, respectively but only 8% to 11%identical to FtsZ. These proteins do not ex-ist in most bacterial species, and their lowdivergence from eukaryotic tubulin suggests

that they are a product of a distant horizon-tal gene transfer (88). The cellular functionof these two proteins is unknown; however,they assemble in vitro, and it has been spec-ulated that BtubA/BtubB may contribute tothe elongated spindle shape of Prosthecobacter(89).

The structure of BtubA bound to GTPclosely resembles the structure of tubulin(Figure 1, middle), including the long loopsresponsible for lateral interactions in micro-tubules and the large helix-loop-helix domain(often referred to as the third C-terminal do-main) that forms the outer surface of micro-tubules (88). The third C-terminal domain isabsent from FtsZ (49). As in both tubulin andFtsZ, the N-terminal domain (which providesloops T1–T6 for nucleotide binding) is sepa-rated from the second domain by the centralhelix T7. Similarly, the second domain pro-vides the T7 loop (which deviates substantiallyfrom both tubulin and FtsZ) that activates nu-cleotide hydrolysis in the protofilament wheninserted into the active site of the adjacentsubunit.

BtubA and BtubB form a weak het-erodimer in vitro (88). A point of interestis that both BtubA and BtubB are able torefold in vitro without the help of chaper-ones, which is similar to some FtsZs andunlike normal tubulins (88). The crystal struc-ture of BtubA/BtubB shows a tubulin-like het-erodimer, with both intra- and interdimerbends evident. It is not possible to describeBtubA or BtubB as being analogous to ei-ther !- or "-tubulin because both BtubA andBtubB have mixed characteristics of the twotubulin forms (88).

In vitro self-assembly. BtubA does not self-assemble in vitro (regardless of nucleotidepresence), whereas a His-tagged BtubB (withany guanine nucleotide supplied) assemblesinto rings that appear to be one subunit thick(89). Interestingly, mixtures of BtubB andBtubA assemble into bundled linear protofil-aments. Sontag et al. (89) observed bundles ofBtubA/BtubB comprising 4–7 protofilaments


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


that were similar to double protofilaments aswell as bundles of straight double protofila-ments that twist, as shown by Schlieper et al.(88). It was not possible to determine the par-allel/antiparallel arrangement of BtubA andBtubB protofilaments within these structures;however, the monomer repeat is 41.6 A, avalue that lies between the monomer spacingsfor FtsZ and tubulin, suggesting the filamentsare similar (88). Filament bundles with hollowtubular profiles 40 nm in diameter (thickerthan the 25 nm of microtubules) have alsobeen observed (89). Analysis of BtubA/BtubBprotofilaments indicates an equimolar ratio ofeach protein within the filaments, further sup-porting the idea that these proteins assemblein a fashion similar to !"-tubulin wherebythe BtubA/BtubB subunits alternate withinthe polymer (88, 89).

Little is known about the kinetics ofBtubA/BtubB assembly, but critical concen-trations for both GTPase activity and assem-bly have been observed, suggesting a coopera-tive assembly mechanism (89). Protofilamentassembly is reversible; the filaments assemblerelatively quickly and, over time, disassemblebecause of GTP consumption (88).

The role of nucleotide hydrolysis. BtubAand BtubB are able to bind one molecule ofguanine nucleotide each (89); however, therates of GTPase activity differ significantlybetween the two proteins (0.40 mol GTP permin per mol for BtubB and 0.13 mol GTPper min per mol for BtubA). When mixed to-gether in equimolar amounts, the GTPase ac-tivity of the combined proteins is higher thanthat of either protein alone, suggesting the di-rect interaction of BtubA and BtubB (89).

THE ACTIN HOMOLOGUESBesides tubulin, actin is the other essentialand ubiqutious eukaryotic cytoskeletal pro-tein. Actin forms double-helical thin filamentscomposed of two strands. Actin filamentsform the “tracks” that myosin (a motor pro-tein) is able to move along.

ActinActin is the prototypical member of a super-family of ATPases that includes hexokinaseand Hsp70. The actin family is very diversein sequence and in function, showing a con-served fold related to ATPase activity. In alandmark paper in 1992, it was reported thatthe actin family also includes three bacterialproteins: ParM (StbA), MreB (and relatives),and FtsA. These proteins were identified asactin homologues (showing higher similarityto actin and Hsp70 than to the hexokinases)because they contain five conserved sequencemotifs related to nucleotide binding and hy-drolysis (90).

The actin fold is comprised of two largedomains (named I and II). These two domainscan be divided into two subdomains (A andB), and the larger subdomains (designated IAand IIA) share a common fold consisting ofa five-strand ß-sheet surrounded by three !-helices. The two smaller subdomains (IB andIIB) show a wider variability in size and struc-ture across the actin family, bestowing someof the properties unique to each protein. Thetwo major domains of actin can rotate withrespect to one another, and between the twodomains lies a highly conserved ATP-bindingpocket. Proteins of the actin family bind ATP,normally in association with Mg2+ or Ca2+,and coordinating Asp residues are importantfor nucleotide hydrolysis. Unlike Hsp70 andhexokinase, actin assembles in vivo into a dy-namic helical polymer (called F-actin) withcooperative assembly characteristics. Withinthe filaments, actin assembles in a head-to-tailarrangement; thus, similar to tubulin, it hasa distinct asymmetry. Actin shows structuralchanges upon polymerization (91) and ex-hibits the characteristic termed treadmilling.Treadmilling occurs when the two ends of afilament have different affinities for polymer-ization. New subunits assemble at the pre-ferred end, and after nucleotide hydrolysis andphosphate release, subunits dissociate fromthe nonpreferred end thus leading to a fluxof subunits though the filament. When the

476 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


rates of assembly and disassembly are equiva-lent, the filament maintains a constant length.Along with the rate of treadmilling, filamentgrowth and shrinkage are controlled by therates of monomer addition and dissociation.Actin dynamics are regulated by a range ofaccessory proteins that affect the assembly,disassembly, and rearrangement of actin fil-aments in vivo.

ParMParM (previously StbA) is one of three com-ponents required for the correct partitioningof R1 low-copy number plasmids in E. coli.The two other components of the par systemare parC and ParR. parC is a centromere-likesequence of DNA that contains the R1 parpromoter sequence. ParR is a repressor pro-

tein that binds to the parC locus. ParM inter-acts with the ParR-parC complex (92) and ap-parently functions as a primitive mitosis-likespindle to move newly replicated plasmids toopposite poles of the cell (93).

The crystal structure of ParM confirmedthe original sequence-based assignment of anactin fold (94), although ParM does show sig-nificant differences in loop, helix, and sheetarrangement within domains IB, IA, and IIB(Figure 3). The subdomain IB lacks a helixpresent in both MreB (see below) and actin.This subdomain also shows a longer loop(quite different from the equivalent Dnase I-binding loop of actin) as well as an unusualinsertion of a strand from domain IA that isnot observed for any other member of theactin family. Other differences include the re-placement of a "-sheet with a helix-loop-helix

Figure 3(left) Structures of F-actin filaments (PDB entry 1YAG) (91), (second from the left) MreB filaments fromThermotoga maritima (PDB entry 1JCE) (112), (center) ParM in the “open” and “closed” conformations,(second from the right) ParM filament (94), and (right) FtsA, showing the position of the nucleotide withinthe interdomain cleft, the conservation of fold, and the axis of the protofilament extension (arrow). (center,top) Note the closed conformation (PDB entry 1MWM) and (bottom) the open conformation of the apoand ADP-bound forms of ParM, both from E. coli plasmid R1 (94). The conformational change ispredicted for all actin homologues. (right) AMPPNP-bound FtsA from T. maritima (PDB entry 1E4G)(136).


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


motif at the top of subunit IIB and the absenceof a helix in subdomain IIA. These structuraldifferences occur in the equivalent regionsof actin that are involved in protofilamentcontacts. The three-dimensional structures ofADP-bound and apo forms of ParM revealed a! 25" conformational difference between do-mains I and II that closes the interdomain cleft(Figure 3). This conformational change is fa-cilitated by the connecting helix H5 that actsas a mechanical hinge. Like actin, ParM is ableto assemble into filaments, and the change inconformation is thought to be linked to theassembly characteristics of ParM filaments.

ParM filaments. In vitro, ParM self-assembles in the presence of ATP, ATP#S,AMPPNP, or ADP (93, 95), forming longdouble-helical filaments that gently twistwith a crossover of 300 A (shorter than thevariable actin crossover that averages 360 A).The helical nature of these filaments isthought to increase the overall strength ofthe filament and reduce the propensity toform stable lateral interactions, making thesefilaments less likely to bundle. ModelingParM filaments using the crystal structureand three-dimensional reconstruction ofParM filaments from negative stain EM,coupled with our understanding of MreB andactin filaments, suggests that ParM assemblesin a head-to-tail orientation (Figure 3) (94).

Kinetic analysis of ParM assembly invitro also shows clear differences from actin.ParM filaments are believed to nucleate via anucleation-condensation mechanism involv-ing three monomers, as expected for a two-stranded helical polymer. ParM shows a 300-fold faster rate of nucleation to that of F-actin(95). Actin has a significant kinetic challengeto spontaneously nucleate, requiring specificnucleation factors that assist filament assem-bly in vivo. No such factors are apparently re-quired for ParM. Although the in vitro rate ofassembly (5.3 ± 1.3 µM#1s#1) of ParM fila-ments is similar to actin, ParM shows someinteresting characteristics. ParM filamentsdisplay symmetrical, bidirectional polymer-

ization, but actin assembles unidirectionally.The rate of assembly of ParM filaments inboth directions is equal, whereas the disassem-bly of ParM filaments is unidirectional andcatastrophic (shortening results in rapid, com-plete filament disassembly with a disassemblyrate of 64 ± 20 s#1). This dynamic instabil-ity is also a feature of tubulin assembly but notactin. Interestingly, in vitro filaments of ParMassemble to a fairly constant length of 1.5 µm(95).

In vitro ParM shows cooperative ATPaseactivity (92). Disassembly of ParM filamentsrequires nucleotide hydrolysis (ADP-ParMfilaments are extremely unstable with a crit-ical concentration of !100 µM compared to!2 µM for ATP-ParM), and as a consequenceof this, mutants deficient in ATPase activ-ity show hyperstability, both in vitro and invivo. The dynamic instability of ParM fil-aments is an integral part of ParM func-tion, and ATPase mutants form stable fila-ments in vivo that are not dynamic and donot support plasmid partitioning (92, 93). Itis assumed that ParM filament disassemblyoccurs via a mechanism similar to that postu-lated for F-actin: The energy released by nu-cleotide hydrolysis might invoke a large con-formational change, which in turn weakensthe intramolecular bonds between subunits,promoting filament disassembly. The confor-mational change seen in the two crystal struc-tures of ParM containing ADP but without anucleotide might represent such a mechanism(Figure 3) (94).

Finally, the spontaneous disassembly ofADP-ParM filaments differs from ADP-actin filaments, which requires severing fac-tors, such as cofilin, to promote disassembly.Monomeric ADP-ParM subunits dissociatefrom filaments at a rate !100 times faster thanADP-actin, and actin requires the nucleotideexchange factor profilin to achieve the samerate of ADP dissociation from monomers asParM achieves alone. Recent work by Garneret al. (95) indicates that like actin, ParM fil-aments are stabilized by a cap of ATP-boundmonomers.

478 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


In vivo filaments—how does ParM reallywork? Immunofluorescence microscopy re-vealed that ParM assembles into dynamicpole-to-pole axial filaments that are essentialfor plasmid partitioning (93). The intracel-lular expression of ParM produces !15,000–18,000 molecules per cell (93). When assem-bled into filaments, this should be enoughParM to form a filament 15–20 times the celllength. So, it is certainly possible that theParM filament is comprised of several parallelfilaments.

Most par systems of plasmid partitioningonly comprise three components. ParM isunable to assemble without the other compo-nents ParR and parC (93), and plasmid repli-cation is required for ParM filament forma-tion (96). The binding of ParM to ParR-parCis ATP dependent, and this stimulates ParM’sATPase activity. Møller-Jensen et al. (96) haveobserved plasmids attached to each end of theParM filament. As the ParM filament extendsin length, so does the distance between theplasmids. These data imply that ParM mayprovide the mechanical force required to pushthe plasmids to the poles of the cell, analogousto the mitotic apparatus in eukaryotic cells. Itwas proposed that ParR-parC complex func-tions as a nucleation point for ParM polymer-ization (96).

Garner et al. (95) propose that, at cellu-lar concentrations of ParM, spontaneous nu-cleation and filament elongation would occurthroughout the cell. Thus, rather than nucle-ation being a regulatory point in the mech-anism, only filaments that manage to locateand bind to plasmid DNA (the ParR-parC in-teraction is known to stabilize ParM filamentsbelow the steady-state critical concentration)would be protected from dynamic instability.Only when both ends are stabilized by bindingto the ParR/plasmid complex, would segrega-tion occur. It has been proposed that bidirec-tional elongation of ParM filaments stabilizedby the interaction with the ParR-parC com-plex drives plasmid segregation.

Questions still remain unanswered. How isthe ParM filament extended in vivo if the plas-

mids are bound to each end? It has been sug-gested that the DNA might be propelled bya treadmilling mechanism of ParM filaments,and the addition of new subunits to the endsof filaments may move the plasmid forward. Ithas also been speculated that in an unidenti-fied mechanism ParM filaments might serve asa track on which motor proteins carry the plas-mids to their destination. No potential motorproteins have been identified as yet. Anotherinteresting issue is raised by the attachmentof plasmids to each end of the ParM filament.As described above, ParM assembles into fila-ments with an inherent polarity. A mechanismwhereby asymmertical depolymerization oc-curs from a symmetrically assembled filamentis difficult to reconcile. Either the two ends ofthe filament are chemically different, and theplasmid attachment sites are nonidentical, orthe in vivo filament of ParM consists of fila-ments aligned in an antiparallel manner thusforming equivalent ends.

Finally, an axial filament that crosses themid-cell site must be disassembled prior tocell division or somehow severed. In cells overexpressing ParM with a deficient ATPase, hy-perstable filaments form, and cell division isblocked. How is disassembly regulated? It ispossible that once the plasmids meet the polesof the cell, they attach to the membrane orsome kind of target, releasing the ParR-parCinteraction and causing ParM filament to be-come unstable and disassemble.

MreBMreB is encoded in a cluster of genes in-volved in determining cell shape formationalthough its precise role(s) are still unknown(97–100). Some bacteria contain several re-lated MreB-like genes. For example B. sub-tilis encodes MreB, Mbl, and MreBH, witheach one showing a similar degree of sequenceidentity (101, 102). In B. subtilis, MreB appearsto be required for the control of cell diame-ter (102, 103), and Formstone & Errington(98) suggest that MreB specifically functionsto restrain cell diameter. Mbl in B. subtilis


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


is specifically involved in cell elongation andis required for the helical insertion of thepeptidoglycan necessary for growth of somerod-shaped cells (104). In Caulobacter crescen-tus, MreB depletion causes abnormal lemon-shaped cell morphology (instead of the nor-mal rod-shaped cell) with defects in cell wallintegrity (105). Analysis of the conservationof MreB is complicated because the phylo-genic assignment of these proteins is diffi-cult owing to their similarities. It is inter-esting to note that those organisms that donot have MreB often have Mbl homologues,which may eventually turn out to be MreB(or vice versa).

Because MreB shows homology with actinand ParM, it was expected that MreB wouldform some kind of filament capable of me-chanical work. On the basis of such anidea, hypothesis-driven research has impli-cated MreB in chromosome segregation forseveral organisms, including B. subtilis (101,103, 106), E. coli (107), and C. crescentus (108),and data from the latter organism are the mostconvincing. MreB’s involvement in chromo-some segregation has been investigated withthe aid of a small-molecule inhibitor calledA22 (whose specific target is MreB) (108,109). Treatment of C. crescentus cells with A22causes a specific, rapid, and reversible disrup-tion of MreB function. Studies involving theadministration of A22 at specific times in thecell cycle revealed that MreB played an im-portant role in the segregation of the origin-proximal loci of the C. crescentus chromosome.It appears that segregation requires at leasttwo separate mechanisms, the first being anMreB-dependent separation, whereby a re-gion near the origin is initially segregated,followed by a second mechanism that is in-dependent of MreB, whereby the rest of thechromosome follows the origin (108, 110). Inaddition to a role in cell shape determinationand chromosome segregation, MreB is alsothought to mediate cell polarity in C. crescen-tus (111). Sequence analysis of MreB initiallyrevealed similarities to FtsA (97), and MreBwas predicted to have an ATPase fold similar

to actin and hsp70 in the paper by Bork et al.(90).

MreB structure. The MreB crystal struc-ture shows that MreB has a conserved actinfold comprising the two domains (I and II)with a nucleotide-binding site in the inter-domain cleft between them (Figure 3) (112).Typical of other members of the actin family,the smaller domains, IB and IIB, show morediversity when compared to those of actin.These smaller domains, however, show thesame topology as those of actin, suggestinga closer relationship of MreB to actin thanHsp70, FtsA, or hexokinase because the topol-ogy of these domains differs. Significant dif-ferences between MreB and actin are evidentwithin the helix H8 loop. In actin, this regioncontains specific sequence insertions requiredfor subunit-to-subunit interactions. These se-quences are absent in MreB (112, 113).

MreB filament. All biochemical studies havebeen performed on MreB from Thermotogamaritima (a hyperthermophilic eubacterium)because MreB from most mesophilic organ-isms is difficult to handle. An advantage ofthis is that the data obtained about MreB aredirectly comparable. MreB assembles in vitrointo straight and curved protofilaments in thepresence of ATP (112, 114). Ring-like struc-tures and filament bundling have also beenreported; however, the biological implicationsof these structures have yet to be determined.Filamentous bundles of MreB formed in vitrohave an increased rigidity, increasing the over-all strength of the filaments, which may bevery important to their function particularlyif they are part of a mechanical apparatus(114).

No high-resolution data of the in vivoMreB filament are currently available, al-though crystals containing protofilaments ofMreB have provided us with an atomic res-olution insight into the self-association ofMreB monomers (Figure 3). MreB assem-bles into filaments similar to that of F-actin,in that the subunit repeat, structure, and

480 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


subunit orientation are approximately thesame (91, 112). However, F-actin shows ax-ial rotation (or twist) within the filaments (F-actin filaments can be described as two twistedprotofilaments), but only a small number ofMreB filaments show a slight axial rotation(112).

The critical concentration for assemblyof MreB (!3 nM) is much lower than thecritical concentration of F-actin (!0.25 µM)(114, 115), implying that MreB has a higheraffinity for other MreB monomers and MreBfilaments than actin does for other actinmonomers and filaments. This also suggeststhat MreB nucleation is a much more fa-vorable process than actin nucleation, and ithas been proposed that MreB polymerizationoccurs either without a nucleation phase orwith an extremely short-lived nucleation step(114). Rapid nucleation is also suggested bythe almost instantaneous assembly observedin the presence of ATP even at very low MreBconcentration (114). Although actin requiresaccessory proteins to assist with in vivo assem-bly, the ease of MreB nucleation suggests thatMreB may be more kinetically tuned to haveless need for regulatory or accessory proteinsto assist in filament assembly. Consistent withthis, MreB shows faster polymerization ratesthan actin.

The role of nucleotide hydrolysis withrespect to MreB assembly dynamics ispoorly understood. Comparison of the MreBnucleotide-binding site with that of the ATP-bound actin indicates that most of the active-site residues are in the same position, with theexception of some of the residues that bind the#-phosphate. Free phosphate is released in so-lutions of self-assembling MreB after ATP ad-dition. A time lag between phosphate releaseand MreB polymerization is the basis for thesuggestion that ATP hydrolysis might occurafter MreB monomers incorporate into fila-ments (114). Currently, there is no data aboutthe stability of MreB filaments in vitro, andthe study of the disassembly of MreB filamentswith respect to nucleotide hydrolysis shouldbe very interesting.

MreB in vivo. In a variety of organisms,MreBs have been observed assembling intovaried structures, i.e., rings in Rhodobactersphaeroides (116), helical structures in B. sub-tilis (102, 117) and E. coli (107, 118), as wellas bands and helical structures in C. crescentus(105). Time-lapse images and FRAP experi-ments have revealed that MreB and Mbl fil-aments are dynamic in vivo (101, 111, 117),although the biological significance for this isunknown.

Molecular function. How might MreB pro-teins function in vivo and what might be theirroles? It has been speculated that the helicalfilaments of MreB-like proteins might providepositional information for the localization ofthe wall-synthesizing enzymes, the penicillin-binding proteins (PBPs), thereby controllingcell wall morphogenesis and thus cell shape(104, 105, 119). In the 1970s, experimentalobservations of B. subtilis mutants with heli-cal cell morphology led to the suggestion thatcell wall elongation occurs by helical peptido-glycan insertion (120). The suggestions thatMreB-like proteins may distribute factors thataffect the organization and mechanics of thecell wall and that the filament structures them-selves may directly contribute to the mechan-ics of the cell wall (102, 104) provide a possiblemolecular mechanism for helical peptiogly-can growth. Results from C. crescentus revealedMreB localizes in an FtsZ-dependent man-ner to the mid-cell position during cell di-vision, leading to the proposal that MreBdirects the switch from cell wall elonga-tion to septum extension during division(105).

MreB’s involvement in chromosome seg-regation may arise from an indirect func-tion of MreB, or MreB might attach to an(as yet) unknown bacterial centromere (108,119). Finally, it is possible that the functionsof MreB proteins provide physical markersin the cell to which other essential processesare linked, thus giving rise to the appar-ent involvement in chromosome segrega-tion, polarity determination, and cell shape


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


WACA: Walker Acytoskeletal ATPase

definition. Recently, it has been shown thatMreC and MreD are linked to cell wall syn-thesis, and it has been proposed that the Mreproteins (MreB, Mbl, MreC and MreD) pro-vide the link between intracellular organiza-tion and the extracellular cell wall syntheticmachinery (121, 122). It may not be generallyappreciated that filaments of MreB, MreC,MreD, and the peptidoglycan-synthesizingPBPs comprise a system vaguely analogous tothe machinary used to position cellulose syn-thase in plants. This plant machinary utilizescytosolic helical microtubules to provide in-formation about where cellulose (rather thanpeptidoglycan in the case of MreB) is he-lically inserted on the outside of the cell(123).

FtsAFtsA was one of the first cell division proteinsto be identified (124), and sequence analy-sis indicated that FtsA might belong to theactin superfamily (90). This caused some ex-citement because FtsA interacts directly withFtsZ (125–131). In vivo FtsA localizes to theZ ring and also to FtsZ helical structures (23)in an FtsZ-dependent manner (20, 23, 29). ItsC terminus (containing a conserved amphi-pathic helix) has a role in interacting with FtsZand in interacting with the membrane, sug-gesting that FtsA tethers the Z ring to the cellmembrane (132, 133). Consistent with this, adefined FtsA/FtsZ ratio is required for normalcell division to occur (134, 135).

Structure determination of FtsA revealedan actin fold (less conserved than ParM orMreB) showing some differences particularlyin the topology of the two small subdomains(136), with the small subdomain located onthe opposite side of domain I when com-pared to actin (Figure 3) (136). This arrange-ment shows no homology with any knownstructure.

The actin superfamily is a diverse familyunited by a common ATPase domain, andself-assembly is an exception rather than therule. FtsA’s conserved ATPase fold, includ-

ing a conserved catalytic ATP-binding pocketand its preferential ability to bind ATP sug-gest that an intrinsic part of FtsA’s functionis to hydrolyze ATP. However, large differ-ences in its enzymatic activity have been ob-served. Although ATPase activity has beenreported for B. subtilis FtsA (29), no activityhas been detected for FtsA from Streptococcuspneumoniae (127). It is possible that ATP hy-drolysis by FtsA is regulated in vivo by a con-formational change evoked by some form ofprotein-protein interaction.

The observation that FtsA localizes to theZ ring in vivo, with its similarities to actin,indicates that FtsA itself might self-assembleinto a polymer. Generally, attempts to studythe self-assembly of FtsA from a variety oforganisms has yielded largely negative re-sults; however, very recently S. pneumoniaeFtsA was observed to polymerize in vitro intobent and bundled long corkscrew-like helices,composed of paired protofilaments (127). Thefilaments were highly stable, requiring bothadenosine nucleotide and magnesium for ini-tial assembly, and showed no dynamic behav-ior (127). Because FtsA is unable to assembleinto a detectable superstructure in vivo in theabsence of FtsZ, it is possible that FtsA mayrequire in vivo accessory proteins to assemble,explaining the different self-assembly poten-tials observed. The high stability of the FtsAfilaments reported for S. pneumoniae FtsAhints at a more complicated in vivo scenario.Filaments that assemble spontaneously andthat are extremely stable would need to be verycarefully controlled by the cell. It is possiblethat with an external activator, FtsA filamen-tation or self-interaction, becomes reversible,as in the case of the Walker A cytoskeletalATPase (WACA) family of proteins (see be-low). Very small changes in vivo may be sig-nificant in transforming a monomeric pro-tein into a multimeric assembly. Alternatively,FtsA does not self-assemble but plays a rolein tethering the Z ring to the membrane andin stabilizing the ring structure. This pos-sibility is supported by the recent observa-tion that a conserved amphipathic helix in the

482 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


C terminus of FtsA is essential for targetingFtsA to the membrane and to the Z ring(132).

INTERMEDIATE FILAMENTHOMOLOGUEIntermediate filaments are a class of cytoskele-tal elements in eukaryotes that are oftenexpressed tissue-specifically. They are com-prised of five different filament structuresformed from various forms of keratins, lamins,and other specialized proteins. Examples in-clude filensin (which is found in the lens of theeye), the keratins that are expressed in epithe-lial cells, and the lamins (which are requiredfor nuclear envelope integrity).

CrescentinCrescentin has been postulated to be a bacte-rial homologue of intermediate filament pro-teins (IFs) (137). Its amino acid sequence has adistinct seven-residue repeat that is predictedto form coiled-coil structures. Because of thedominating coiled-coil repeat, sequence com-parisons are unreliable, but crescentin sharessome important overall features with eukary-otic IF proteins. Analysis has revealed that thedomain organization of crescentin is similarto animal IF proteins, suggesting that cres-centin probably is a prokaryotic homologue ofIFs.

Crescentin is required for determiningthe vibrioid or helical shape of C. crescen-tus cells. In vivo immunofluorescence mi-croscopy and deconvolution analysis revealedthat crescentin localizes as a continuous pole-to-pole helical filament along one side of thecell (137). In vitro purified crescentin is able toassemble into filaments with a width of about10 nm. Remarkably, these filaments assemblespontaneously (in the absence of any energysource or cofactor) similar to IFs.

The correlation that crescentin is involvedin determining cell shape and that it formslong filaments within the cell suggest thatcrescentin filaments assemble and somehow

IF: intermediatefilament

(directly or indirectly) associate with the cy-toplasmic membrane specifically on one sideof the cell. Furthermore, if the shape and he-licity of the filament was somehow applied tothe cell, a vibrioid or helical cell shape mightbe formed. In support of this, in stationary-phase cultures of C. crescentus, the cellsbecome filamentous but also helical (138).Vibrioid cells are shorter than the helical pitchof the filament and are simply curved. In cellstreated with cephalexin (which disrupts nor-mal peptidoglycan synthesis), the intracellu-lar localization of the crescentin filament wasgradually disrupted. So, the function of cres-centin is somehow linked to the biosynthesisof peptidoglycan, and it seems likely that itsfunction must also be coordinated with thecell cycle. This poses an interesting question.Because crescentin does not require cofactorsor nucleotide, and apparently assembles in-dependently, how is this filament regulated?When a cell is dividing, how is the crescentinfilament disassembled to allow daughter cellseparation? It seems most likely that cofactorsmust be present to control crescentin assem-bly and disassembly.

WALKER A CYTOSKELETALATPASE—A NEW FAMILY OFCYTOSKELETAL PROTEINS?From the discussion above it is clear thatprokaryotes possess a cytoskeleton composedof classical actin, tubulin and possibly inter-mediate filament-like proteins. However, thespatial organization of bacterial cells also re-lies on a further group of proteins that hasno known direct counterpart in the cyto-plasm of eukaryotes. We propose a new sub-class of proteins, called the WACA proteins,which are required for the spatial regulationof chromosome partitioning and cell divi-sion. The WACA proteins belong to a largeand functionally diverse family of ATPasesthat have a conserved deviant Walker A mo-tif and dimerize in an ATP-dependent man-ner (139). Although these deviant Walker Aproteins are structurally homologous, their


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


functions differ significantly, and it has beenproposed that these proteins may be molecu-lar switches (140). Recently, nitrogenase (anarchetypal deviant Walker A ATPase) wasshown to undergo conformational changesthought to control electron transfer processes.It was suggested that such conformationalchanges might be used to achieve directedmotion, thus bestowing possible mechani-cal roles on the deviant Walker A ATPases(141). The WACA proteins are a specific sub-set of the deviant Walker A ATPases thathave evolved the specialized function of form-ing ATP-induced, surface-dependent poly-mers (140), which might be considered asan additional component of the prokaryoticcytoskeleton.

The WACA family of proteins is com-prised of MinD and the ParA/Soj plasmid andchromosome partitioning proteins, includingSopA and ParF [reviewed by Hiraga (142)].These proteins share extensive sequence ho-mology and a similar three-dimensional struc-ture (140, 143) (Figure 4). MinD is in-volved in the processes of Z-ring position-ing during cell division, and ParA and Soj

have roles in the processes of chromosomesegregation, transcription, and organizationof plasmids and chromosomes. All deviantWalker A proteins form dimers and are ableto bind and hydrolyze ATP. The ATPase ac-tivity and dynamic behavior of the WACAproteins are regulated by the interaction ofthe WACA with an activation protein (144–146). MinD’s ATPase activity is modulatedby an interaction with MinE, ParB contains asmall, N-terminal peptide known to activateParA, and similarly Spo0J has a 20-amino acidN-terminal tail that activates Soj’s ATPase(140, 147).

Typically the WACA proteins show dy-namic behavior in vivo. They all have time-dependent localization patterns in the cell,with ParA alternating between nucleoids(148), MinD (from E. coli) oscillating fromcell pole to cell pole (149), and Soj movingfrom pole to pole or nucleoid to nucleoid(150, 151). A notable difference between theWACA proteins is in the period of their os-cillation. Although the Min system shows aregular fast oscillation of roughly a minute(152), both ParA and Soj show irregular,

Figure 4Structures of the Walker A cytoskeletal ATPase (WACA proteins). (left) Crystal structure of theMgATP-induced Soj (D44A mutant) dimer from Thermus thermophilus (PDB entry 2BEK) (140). (right)Structure of MinD (binding AMPPCP, a slowly hydrolyzable ATP analogue) from Pyrococcus furiosus(PDB entry code 1G3R) (143).

484 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


erratic “jumping” with periodicities of min-utes, sometimes extending up to an hour (148,150, 151).

The mechanism behind MinD oscillationis probably best understood. Discovered sev-eral years ago in E. coli (153), the MinCDEproteins help place the septum in the middleof the cell by inhibiting cell division at the cellpoles. ATP-bound MinD is tethered to themembrane (154) by a C-terminal amphipathichelix that binds to the phospholipid bilayer(155, 156), similar to the amphipathic he-lix required for FtsA’s membrane interaction(see below) (132). ATP-bound MinD inter-acts with MinC [the cell division inhibitor thatdirectly inhibits FtsZ-ring formation (157)],recruiting MinC to the membrane. MinE isalso able to bind to the ATP-bound MinDvia a short activating peptide (at its N ter-minus). This interaction enhances MinD’sATPase turnover (158), and the ADP-boundMinD is released from the membrane. Thus,a self-organizing oscillating system is gener-ated because MinD is more likely to rebindthe membrane where there is no MinE andwhere there is already some MinD bound (onthe other side of the cell). Thus, MinD os-cillation is closely linked to ATP hydrolysis,and consistent with this, MinD mutants defi-cient in ATPase activity show a reduced oscil-lation rate (158). Several mathematical modelsof this system have been produced and faith-fully reproduce properties observed in livingcells (159–162). It is possible that related os-cillatory mechanisms may be used by ParA(163) and Soj (164) to position plasmids andchromosome origins, respectively. It is inter-esting to note that MinD in B. subtilis hasnot been observed to oscillate but appearsto be tethered to the cell poles by anotherprotein called DivIVA. It is surprising thathomologous proteins performing identicalfunctions have such different mechanisms ofaction.

The WACA proteins show other interest-ing similarities consistent with cytoskeletal el-ements. Both MinD (118) and ParA (163) havebeen shown to assemble into dynamic helical

structures in vivo. In the case of MinD, thesehelices were similar to but different in helicalpitch and general spread along the length ofthe cell from those found for MreB (102), andit is thought that the MinD helices are inde-pendent of MreB filaments (118). The helicalstructures of ParA were also observed in theabsence of MinD, indicating that ParA fila-ments are not simply interacting with MinDassemblies. The potential relationship of ParAwith MreB has not been explored, but it is pos-sible that the ParA structures are independentof other known helical filaments.

In vitro filaments of MinD have been ob-served; however, their formation is stronglydependent on the presence of both phospho-lipid and ATP (154, 165). Similarily, Soj as-sembly also has a dependence on ATP andDNA, and in their presence, Soj forms nu-cleoprotein filaments in a cooperative man-ner (140). Filaments have also been reportedfor ParF (a ParA homologue) with a require-ment for ATP (166). The in vitro filamentsdescribed for Soj (140), MinD (165), and ParF(166) are not dynamic by themselves, but thissimply reflects the fact that in contrast to FtsZ,the WACA subclass of proteins have special-ized activator proteins that are required forATP hydrolysis.

WACA proteins show an unusual charac-teristic whereby they bind to the entire sur-face area of their substrate. In vitro MinDcoats phospholipid vesicles with high density;something that could be explained by surface-assisted polymerization (154), and Soj binds toDNA coating it completely (167). The processof surface-assisted polymerization can explainSoj/Spo0J oscillation along the nucleoid, sim-ilar to membrane binding of MinD and subse-quent displacement by MinE, leading to pole-to-pole oscillation (146, 168).

An important unresolved question ishow these proteins travel through the cell.Simple diffusion is one possibility; how-ever, polymerization-depolymerization dy-namics is another. Because there is notenough MinD in the cell to cover the wholemembrane and helical structures have been


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


detected in vivo for the proteins, the sec-ond alternative seems more likely to us at themoment.

Dynamic behavior (fuelled by nucleotideturnover) seems to be generated by a set oftwo proteins (the WACA subclass and theirATPase-activating counterparts) in these po-sitioning systems. In contrast, dynamic behav-ior is maintained by a single component in

the case of FtsZ. However, it is interestingto note that it has been speculated that FtsZprotein arises from the fusion of a GTP-binding protein with its activator protein (52).Thus, in order to evolve, a dynamic system bi-ology required two components: a nucleotide-binding protein and an activation domain;however, the two components need not beseparate.

SUMMARY POINTS

1. Bacterial cells contain a variety of dynamic filamentous proteins that bring aboutspatial and temporal organization analogous to the eukaryotic cytoskeleton.

2. Many of these proteins are related to actin and tubulin by an extended evolutionaryhistory. These proteins retain the actin and tubulin protein fold, and they are able toform filaments in vitro and in vivo.

3. The tubulin homologues include FtsZ and a pair of cotranscribed proteins calledBtubA and BtubB.

4. The actin homologues include ParM (a plasmid partitioning protein), MreB, andFtsA.

5. A single protein that forming filaments homologous to intermediate filaments hasbeen identified in C. crescentus and is thought to regulate cell shape. Surprisingly, ithas some sequence similarity and domain arrangement that are analogous to IFs butalso assembles in the absence of a nucleotide or cofactors.

6. We propose that a subclass of the deviant Walker A ATPases [named Walker A cy-toskeletal ATPases (WACAs)] have important, dynamic roles in organizing bacte-rial cells during cell division and plasmid/chromosome partitioning. These proteinsshould be categorized as a new class of bacterial cytoskeletal proteins.

7. Great leaps forward in our understanding of bacterial cellular organization have beenfacilitated by modern technologies (such as in vivo fluorescent tagging of proteins),and the discovery of the existence of cytoskeletal scaffolds has provided the meansto speculate on the molecular mechanisms by which cell division, cell shape, andchromsome segregation are executed within bacterial cells. It seems likely that manyof the mechanisms that have evolved in bacteria share similarities to the molecularprocesses that regulate cell shape, cell division, chromosome segregation, and possiblyendo- and exocytosis in eukaryotes.

8. Currently, our understanding of the molecular processes behind cell division, cellshape determination, and plasmid/chromosome segregation is restricted by our in-ability to determine the in vivo forms of the superstructures, formed by the bacterialcytoskeletal proteins. It is expected that in time many accessory proteins with rolesin assisting the assembly, disassembly, and regulation of such superstructures will beidentified.

486 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


FUTURE ISSUE TO BE RESOLVED

1. The first reports of dynamic filaments observed in bacteria caused quite a stir; how-ever, with hindsight, the requirement for such in vivo assemblies is not surprising. Ithas always been perplexing that bacteria maintain their many and varied shapes in theabsence of cytoskeletal elements, and the mechanisms involved in chromosome seg-regation and cytokinesis seemed hard to fathom without the existence of some formof scaffold or cytoskeletal organization. Various models have been proposed that canexplain both chromosome segregation and cytokinesis in the absence of cytoskeletalelements (169, 170). Although bacteria clearly possess proteins that form cytoskeletalelements, the molecular mechanisms by which these cytoskeletal proteins may affectcellular processes are unknown, and much biochemical information is required toresolve this.

2. Why are dynamic assemblies a feature of the cytoskeleton? First, the dynamic natureof these filaments allows for their rapid reorganization. These dynamic structures haveinherent flexibility, perfect for bringing about mechanical work and also for adaptingto changing cell shape and size. Furthermore, dynamic instability greatly assists inthe regulation of filament nucleation. Any filaments that might form without definedinitiation directed by the cell will disassemble (171).

3. Biology has typically supplied accessory proteins to modulate the behavior of poly-merizing proteins inside eukaryotic cells. In bacteria, the cytoskeletal proteins areapparently self-assembling proteins, and for most, no accessory proteins have beenidentified, with the exception of FtsZ. In the future, it is likely that many more acces-sory proteins will be identified that have roles in the specific nucleation, localization,disassembly, or stabilization of the various filaments. So far, all the bacterial proteinslikely to be involved in accessory roles for the filamenting cell division, cell shape, andplasmid/chromosome partitioning proteins show no similarities to accessory proteinsfrom eukaryotes.

4. Because many of these filamentous proteins are likely to assemble in vivo into struc-tures with a thickness greater than a single molecule, the formation of lateral interac-tions between protofilaments and their regulation have important consequences forunderstanding filament assembly and function. Currently, little is known about any ofthe lateral interactions of any of the bacterial cytoskeletal proteins, and this limitationseverely impedes our understanding the molecular mechanisms behind the processesof cell shape determination, cell division, and possibly chromosome segregation.

5. Identification of the mechanisms behind filament nucleation and assembly, coupledwith an understanding of the in vivo filament form, should help us answer fundamentalquestions, such as how the asymmetry of the crescentin filament is established. Manyof these cytoskeletal proteins assemble into helical structures in vivo; however, in vitrothey form straight filaments. How is the helicity established?

6. Both actin and tubulin filaments show clear polarity that controls the directionof motor proteins and enables spatial organization in the eukaryotic cell. So far,there is little data indicating that any of these cytoskeletal homologues provide po-larity in bacteria (except for MreB in C. crescentus). Although the proteins assemble in a


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


head-to-tail fashion and could provide polarity, little is known about any in vivo conse-quence of this. No motor proteins that might interact with the bacterial cytoskeletonhave been identified.

7. Many of these cytoplasmic bacterial cytoskeletal proteins apparently convey structuraldirection to the peptidoglycan, thus adaptor proteins that traverse the membraneshould exist to pass information from the cytoplasm to the peptioglycan synthesizingmachinery in the periplasmic space. Recent data implicates MreC and MreD withthese roles in the case of cell shape determination (122). It seems likely that similarmolecular links will assist in the correct positioning and assembly of the septumsynthesizing apparatus during cell division and also in the coupling of cell divisionwith chromosome/plasmid segregation. These molecular links should reveal muchabout how the processes of cell shape, cytokinesis, and chromosome segregation arebrought about.

ACKNOWLEDGMENTSK.A. Michie acknowledges support from UNESCO-L’Oreal and ESF/MRC EuroDYNA.

LITERATURE CITED1. Carballido-Lopez R, Errington J. 2003. Trends Cell Biol. 13:577–832. Errington J, Daniel RA, Scheffers DJ. 2003. Microbiol. Mol. Biol. Rev. 67:52–653. Møller-Jensen J, Lowe J. 2005. Curr. Opin. Cell Biol. 17:75–814. Lutkenhaus JF, Wolf-Watz H, Donachie WD. 1980. J. Bacteriol. 142:615–205. de Boer P, Crossley R, Rothfield L. 1992. Nature 359:254–566. RayChaudhuri D, Park JT. 1992. Nature 359:251–547. Beall B, Lutkenhaus J. 1991. Genes Dev. 5:447–558. Dai K, Lutkenhaus J. 1991. J. Bacteriol. 173:3500–69. Bi EF, Lutkenhaus J. 1991. Nature 354:161–64

10. Bermudes D, Hinkle G, Margulis L. 1994. Microbiol. Rev. 58:387–40011. Erickson HP. 1995. Cell 80:367–7012. Mukherjee A, Dai K, Lutkenhaus J. 1993. Proc. Natl. Acad. Sci. USA 90:1053–5713. Margolin W, Wang R, Kumar M. 1996. J. Bacteriol. 178:1320–2714. Glass JI, Lefkowitz EJ, Glass JS, Heiner CR, Chen EY, Cassell GH. 2000. Nature

407:757–6215. Stephens RS, Kalman S, Lammel C, Fan J, Marathe R, et al. 1998. Science 282:754–5916. Kawarabayasi Y, Hino Y, Horikawa H, Yamazaki S, Haikawa Y, et al. 1999. DNA Res.

6:83–10117. Beech PL, Nheu T, Schultz T, Herbert S, Lithgow T, et al. 2000. Science 287:1276–7918. Osteryoung KW, Vierling E. 1995. Nature 376:473–7419. Sun Q, Margolin W. 1998. J. Bacteriol. 180:2050–5620. Addinall SG, Bi E, Lutkenhaus J. 1996. J. Bacteriol. 178:3877–8421. Ben-Yehuda S, Losick R. 2002. Cell 109:257–6622. Grantcharova N, Lustig U, Flardh K. 2005. J. Bacteriol. 187:3227–3723. Ma X, Ehrhardt DW, Margolin W. 1996. Proc. Natl. Acad. Sci. USA 93:12998–3003

488 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


24. Mileykovskaya E, Sun Q, Margolin W, Dowhan W. 1998. J. Bacteriol. 180:4252–5725. Stricker J, Erickson HP. 2003. J. Bacteriol. 185:4796–80526. Thanedar S, Margolin W. 2004. Curr. Biol. 14:1167–7327. Rueda S, Vicente M, Mingorance J. 2003. J. Bacteriol. 185:3344–5128. Lu C, Stricker J, Erickson HP. 1998. Cell Motil. Cytoskelet. 40:71–8629. Feucht A, Lucet I, Yudkin MD, Errington J. 2001. Mol. Microbiol. 40:115–2530. Stricker J, Maddox P, Salmon ED, Erickson HP. 2002. Proc. Natl. Acad. Sci. USA 99:3171–

7531. Anderson DE, Gueiros-Filho FJ, Erickson HP. 2004. J. Bacteriol. 186:5775–8132. Mukherjee A, Lutkenhaus J. 1994. J. Bacteriol. 176:2754–5833. Mukherjee A, Lutkenhaus J. 1998. EMBO J. 17:462–6934. Lowe J, Amos LA. 2000. Biol. Chem. 381:993–9935. Lu C, Reedy M, Erickson HP. 2000. J. Bacteriol. 182:164–7036. Bramhill D, Thompson CM. 1994. Proc. Natl. Acad. Sci. USA 91:5813–1737. Erickson HP, Taylor DW, Taylor KA, Bramhill D. 1996. Proc. Natl. Acad. Sci. USA

93:519–2338. Lowe J, Amos LA. 1999. EMBO J. 18:2364–7139. Oliva MA, Huecas S, Palacios JM, Martin-Benito J, Valpuesta JM, Andreu JM. 2003. J.

Biol. Chem. 278:33562–7040. Yu XC, Margolin W. 1997. EMBO J. 16:5455–6341. Chen Y, Erickson HP. 2005. J. Biol. Chem. 280:22549–5442. Mingorance J, Tadros M, Vicente M, Gonzalez JM, Rivas G, Velez M. 2005. J. Biol. Chem.

280:20909–1443. Mukherjee A, Lutkenhaus J. 1999. J. Bacteriol. 181:823–3244. Cordell SC, Robinson EJ, Lowe J. 2003. Proc. Natl. Acad. Sci. USA 100:7889–9445. Lowe J, van den Ent F, Amos LA. 2004. Annu. Rev. Biophys. Biomol. Struct. 33:177–9846. Romberg L, Levin PA. 2003. Annu. Rev. Microbiol. 57:125–5447. Doolittle RF. 1995. Philos. Trans. R. Soc. London Ser. B 349:235–4048. Doolittle RF, York AL. 2002. BioEssays 24:293–9649. Lowe J, Amos LA. 1998. Nature 391:203–650. Nogales E, Wolf SG, Downing KH. 1998. Nature 391:199–20351. Nogales E, Downing KH, Amos LA, Lowe J. 1998. Nat. Struct. Biol. 5:451–5852. Oliva MA, Cordell SC, Lowe J. 2004. Nat. Struct. Mol. Biol. 11:1243–5053. Andreu JM, Oliva MA, Monasterio O. 2002. J. Biol. Chem. 277:43262–7054. Santra MK, Panda D. 2003. J. Biol. Chem. 278:21336–4355. Nogales E, Whittaker M, Milligan RA, Downing KH. 1999. Cell 96:79–8856. Kar S, Fan J, Smith MJ, Goedert M, Amos LA. 2003. EMBO J. 22:70–7757. Santarella RA, Skiniotis G, Goldie KN, Tittmann P, Gross H, et al. 2004. J. Mol. Biol.

339:539–5358. Erickson HP. 1998. Trends Cell Biol 8:133–3759. Scheffers DJ, de Wit JG, den Blaauwen T, Driessen AJ. 2002. Biochemistry 41:521–2960. Faguy DM, Doolittle WF. 1998. Curr. Biol. 8:R338–4161. Gilson PR, Beech PL. 2001. Res. Microbiol. 152:3–1062. Margolin W, Long SR. 1994. J. Bacteriol. 176:2033–4363. Desai A, Mitchison TJ. 1997. Annu. Rev. Cell Dev. Biol. 13:83–11764. Caplan MR, Erickson HP. 2003. J. Biol. Chem. 278:13784–8865. Huecas S, Andreu JM. 2004. FEBS Lett. 569:43–4866. Chen Y, Bjornson K, Redick SD, Erickson HP. 2005. Biophys. J. 88:505–14


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


67. Gonzalez JM, Velez M, Jimenez M, Alfonso C, Schuck P, et al. 2005. Proc. Natl. Acad.Sci. USA 102:1895–900

68. Rivas G, Lopez A, Mingorance J, Ferrandiz MJ, Zorrilla S, et al. 2000. J. Biol. Chem.275:11740–49

69. Romberg L, Simon M, Erickson HP. 2001. J. Biol. Chem. 276:11743–5370. Romberg L, Mitchison TJ. 2004. Biochemistry 43:282–8871. Neuhard J, Nygaard P. 1987. Escherichia coli and Salmonella Typhimurium: Cellular and

Molecular Biology. Washington, DC: Am. Soc. Microbiol. Press72. Scheffers DJ, den Blaauwen T, Driessen AJ. 2000. Mol. Microbiol. 35:1211–1973. Mingorance J, Rueda S, Gomez-Puertas P, Valencia A, Vicente M. 2001. Mol. Microbiol.

41:83–9174. Scheffers DJ, Driessen AJ. 2002. Mol. Microbiol. 43:1517–2175. Lu C, Stricker J, Erickson HP. 2001. BMC Microbiol. 1:776. Phoenix P, Drapeau GR. 1988. J. Bacteriol. 170:4338–4277. White EL, Ross LJ, Reynolds RC, Seitz LE, Moore GD, Borhani DW. 2000. J. Bacteriol.

182:4028–3478. Marrington R, Small E, Rodger A, Dafforn TR, Addinall SG. 2004. J. Biol. Chem.

279:48821–2979. Esue O, Tseng Y, Wirtz D. 2005. Biochem. Biophys. Res. Commun. 333:508–1680. Santra MK, Beuria TK, Banerjee A, Panda D. 2004. J. Biol. Chem. 279:25959–6581. Chretien D, Fuller S, Karsenti E. 1995. J. Cell Biol. 129:1311–2882. Ravelli RB, Gigant B, Curmi PA, Jourdain I, Lachkar S, et al. 2004. Nature 428:198–20283. Wang HW, Nogales E. 2005. Nature 435:911–1584. Lu C, Erickson HP. 1999. Cell. Struct. Funct. 24:285–9085. Huecas S, Andreu JM. 2003. J. Biol. Chem. 278:46146–5486. Diaz JF, Kralicek A, Mingorance J, Palacios JM, Vicente M, Andreu JM. 2001. J. Biol.

Chem. 276:17307–1587. Jenkins C, Samudrala R, Anderson I, Hedlund BP, Petroni G, et al. 2002. Proc. Natl.

Acad. Sci. USA 99:17049–5488. Schlieper D, Oliva MA, Andreu JM, Lowe J. 2005. Proc. Natl. Acad. Sci. USA 102:9170–7589. Sontag CA, Staley JT, Erickson HP. 2005. J. Cell Biol. 169:233–3890. Bork P, Sander C, Valencia A. 1992. Proc. Natl. Acad. Sci. USA 89:7290–9491. Holmes KC, Popp D, Gebhard W, Kabsch W. 1990. Nature 347:44–4992. Jensen RB, Gerdes K. 1997. J. Mol. Biol. 269:505–1393. Møller-Jensen J, Jensen RB, Lowe J, Gerdes K. 2002. EMBO J. 21:3119–2794. van den Ent F, Møller-Jensen J, Amos LA, Gerdes K, Lowe J. 2002. EMBO J. 21:6935–4395. Garner EC, Campbell CS, Mullins RD. 2004. Science 306:1021–2596. Møller-Jensen J, Borch J, Dam M, Jensen RB, Roepstorff P, Gerdes K. 2003. Mol. Cell

12:1477–8797. Doi M, Wachi M, Ishino F, Tomioka S, Ito M, et al. 1988. J. Bacteriol. 170:4619–2498. Formstone A, Errington J. 2005. Mol. Microbiol. 55:1646–5799. Levin PA, Margolis PS, Setlow P, Losick R, Sun D. 1992. J. Bacteriol. 174:6717–28

100. Varley AW, Stewart GC. 1992. J. Bacteriol. 174:6729–42101. Soufo HJD, Graumann PL. 2004. EMBO Rep. 5:789–94102. Jones LJ, Carballido-Lopez R, Errington J. 2001. Cell 104:913–22103. Soufo HJD, Graumann PL. 2003. Curr. Biol. 13:1916–20104. Daniel RA, Errington J. 2003. Cell 113:767–76105. Figge RM, Divakaruni AV, Gober JW. 2004. Mol. Microbiol. 51:1321–32

490 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


106. Soufo HJD, Graumann PL. 2005. BMC Cell Biol. 6:10107. Kruse T, Møller-Jensen J, Lobner-Olesen A, Gerdes K. 2003. EMBO J. 22:5283–92108. Gitai Z, Dye NA, Reisenauer A, Wachi M, Shapiro L. 2005. Cell 120:329–41109. Iwai N, Nagai K, Wachi M. 2002. Biosci. Biotechnol. Biochem. 66:2658–62110. Bates D, Kleckner N. 2005. Cell 121:899–911111. Gitai Z, Dye N, Shapiro L. 2004. Proc. Natl. Acad. Sci. USA 101:8643–48112. van den Ent F, Amos LA, Lowe J. 2001. Nature 413:39–44113. Galkin VE, VanLoock MS, Orlova A, Egelman EH. 2002. Curr. Biol. 12:570–75114. Esue O, Cordero M, Wirtz D, Tseng Y. 2005. J. Biol. Chem. 280:2628–35115. Nishida E, Sakai H. 1983. J. Biochem. 93:1011–20116. Slovak PM, Wadhams GH, Armitage JP. 2005. J. Bacteriol. 187:54–64117. Carballido-Lopez R, Errington J. 2003. Dev. Cell 4:19–28118. Shih YL, Le T, Rothfield L. 2003. Proc. Natl. Acad. Sci. USA 100:7865–70119. Kruse T, Gerdes K. 2005. Trends Cell Biol. 15:343–45120. Mendelson NH. 1976. Proc. Natl. Acad. Sci. USA 73:1740–44121. Kruse T, Bork-Jensen J, Gerdes K. 2005. Mol. Microbiol. 55:78–89122. Leaver M, Errington J. 2005. Mol. Microbiol. 57:1196–209123. Burk DH, Ye ZH. 2002. Plant Cell 14:2145–60124. Donachie WD, Begg KJ, Lutkenhaus JF, Salmond GP, Martinez-Salas E, Vincente M.

1979. J. Bacteriol. 140:388–94125. Descoteaux A, Drapeau GR. 1987. J. Bacteriol. 169:1938–42126. Din N, Quardokus EM, Sackett MJ, Brun YV. 1998. Mol. Microbiol. 27:1051–63127. Lara B, Rico AI, Petruzzelli S, Santona A, Dumas J, et al. 2005. Mol. Microbiol. 55:699–711128. Ma X, Sun Q, Wang R, Singh G, Jonietz EL, Margolin W. 1997. J. Bacteriol. 179:6788–97129. Ma X, Margolin W. 1999. J. Bacteriol. 181:7531–44130. Wang X, Huang J, Mukherjee A, Cao C, Lutkenhaus J. 1997. J. Bacteriol. 179:5551–59131. Yan K, Pearce KH, Payne DJ. 2000. Biochem. Biophys. Res. Commun. 270:387–92132. Pichoff S, Lutkenhaus J. 2005. Mol. Microbiol. 55:1722–34133. Pla J, Dopazo A, Vicente M. 1990. J. Bacteriol. 172:5097–102134. Dai K, Lutkenhaus J. 1992. J. Bacteriol. 174:6145–51135. Dewar SJ, Begg KJ, Donachie WD. 1992. J. Bacteriol. 174:6314–16136. van den Ent F, Lowe J. 2000. EMBO J. 19:5300–7137. Ausmees N, Kuhn JR, Jacobs-Wagner C. 2003. Cell 115:705–13138. Wortinger MA, Quardokus EM, Brun YV. 1998. Mol. Microbiol. 29:963–73139. Koonin EV. 1993. J. Mol. Biol. 229:1165–74140. Leonard TA, Butler PJ, Lowe J. 2005. EMBO J. 24:270–82141. Tezcan FA, Kaiser JT, Mustafi D, Walton MY, Howard JB, Rees DC. 2005. Science

309:1377–80142. Hiraga S. 1992. Annu. Rev. Biochem. 61:283–306143. Cordell SC, Lowe J. 2001. FEBS Lett. 492:160–65144. Ma L, King GF, Rothfield L. 2004. Mol. Microbiol. 54:99–108145. Zhou H, Schulze R, Cox S, Saez C, Hu Z, Lutkenhaus J. 2005. J. Bacteriol. 187:629–38146. Leonard TA, Møller-Jensen J, Lowe J. 2005. Philos. Trans. R. Soc. London Ser. B 360:523–35147. Radnedge L, Youngren B, Davis M, Austin S. 1998. EMBO J. 17:6076–85148. Ebersbach G, Gerdes K. 2001. Proc. Natl. Acad. Sci. USA 98:15078–83149. Raskin DM, de Boer PA. 1999. J. Bacteriol. 181:6419–24150. Marston AL, Errington J. 1999. Mol. Cell 4:673–82151. Quisel JD, Lin DC, Grossman AD. 1999. Mol. Cell 4:665–72


Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.


152. Raskin DM, de Boer PA. 1999. Proc. Natl. Acad. Sci. USA 96:4971–76153. de Boer PA, Crossley RE, Rothfield LI. 1989. Cell 56:641–49154. Hu Z, Gogol EP, Lutkenhaus J. 2002. Proc. Natl. Acad. Sci. USA 99:6761–66155. Szeto TH, Rowland SL, Rothfield LI, King GF. 2002. Proc. Natl. Acad. Sci. USA

99:15693–98156. Zhou H, Lutkenhaus J. 2003. J. Bacteriol. 185:4326–35157. Hu Z, Mukherjee A, Pichoff S, Lutkenhaus J. 1999. Proc. Natl. Acad. Sci. USA 96:14819–24158. Hu Z, Lutkenhaus J. 2001. Mol. Cell 7:1337–43159. Howard M, Rutenberg AD, de Vet S. 2001. Phys. Rev. Lett. 87:278102160. Kruse K. 2002. Biophys. J. 82:618–27161. Meinhardt H, de Boer PA. 2001. Proc. Natl. Acad. Sci. USA 98:14202–7162. Drew DA, Osborn MJ, Rothfield LI. 2005. Proc. Natl. Acad. Sci. USA 102:6114–18163. Ebersbach G, Gerdes K. 2004. Mol. Microbiol. 52:385–98164. Marsh JW, Taylor RK. 1999. J. Bacteriol. 181:1110–17165. Suefuji K, Valluzzi R, RayChaudhuri D. 2002. Proc. Natl. Acad. Sci. USA 99:16776–81166. Barilla D, Rosenberg MF, Nobbmann U, Hayes F. 2005. EMBO J. 24:1453–64167. Leonard TA, Butler PJ, Lowe J. 2004. Mol. Microbiol. 53:419–32168. Doubrovinski K, Howard M. 2005. Proc. Natl. Acad. Sci. USA 102:9808–13169. Lemon KP, Grossman AD. 2000. Mol. Cell 6:1321–30170. Woldringh CL, Mulder E, Huls PG, Vischer NOE. 1991. Res. Microbiol. 142:309–20171. Mitchison T, Kirschner M. 1984. Nature 312:237–42172. Lowe J, Li H, Downing KH, Nogales E. 2001. J. Mol. Biol. 313:1045–57

492 Michie · Lowe

Ann

u. R

ev. B

ioch

em. 2

006.

75. D

ownl

oade

d fro

m a

rjour

nals.

annu

alre

view

s.org

by C

AM

BRID

GE

UN

IVER

SITY

on

04/1

9/06

. For

per

sona

l use

onl

y.

Documents

Dynamic Filaments of the Bacterial Cytoskeleton · 2006-04-19 · terial cytoskeleton. This revolution in the understanding of bacterial cell structure and dynamics has come about