Bacteria make tracks to the pole Aretha Fiebig and Julie A. Theriot* Departments of Biochemistry and of Microbiology and Immunology, Stanford University School of Medicine, Stanford, CA 94305-5307 arge-scale polarity, an essential property of many cells, requires active mechanisms for both estab- lishment and maintenance. In eu- Lkaryotic cells, parallel-oriented arrays of actin filaments and͞or microtubules fre- quently provide a structural framework for polarity. For example, actin cables me- diate polar growth in yeast cell division, microtubules in neuronal axons determine the directionality of long-distance vesicle transport, and directed motility in epithe- lial cells depends on polarized growth of actin filaments (reviewed in ref. 1). Polar- ity is not limited to eukaryotic cells. Many bacteria exhibit polar structures such as flagella and pili. Moreover, numerous bacterial proteins have nonuniform polar distributions (reviewed in ref. 2). In pro- Fig. 1. Cytoskeletal organization. (A) Proteins (yellow) are inherently asymmetric and can polymerize to karyotes, all previous biological evidence form asymmetric filaments (blue arrow) with slow (Ϫ) and fast (ϩ) growing ends. Structures such as for whole-cell polarity could be ascribed bacterial flagella consist of a single polymerized filament. (B) Randomly organized filaments spontane- to physical or biochemical differences be- ously align when space is limited (2 3 3). Motor proteins can reorganize random assemblies to form tween old and new poles and thus did not ordered arrays (3 3 4). Alternatively, ordered arrays can form by localized nucleation (1 3 4͞5) or selective require invoking a polarized cytoskeleton. stabilization due to bundling proteins (2 3 4͞5). (C) The microstructure of bacterial cytoskeletal arrays is However, evidence presented by Gitai et not known. Filaments in the red boxes of FtsZ tubulin-like rings and the MreB actin-like spirals are most al. (3) in this issue of PNAS suggests that likely analogous to the filaments in boxes 3 or 4. bacterial cytoskeletal components are po- larized arrays involved in generating and some segregation (8–11). CreS is involved strating that MreB provides spatial maintaining bacterial polarity. Caulobacter The ubiquity of polarized cytoskeletal in shape determination in information influencing polar protein arrays has been commonly assumed to be crescentus and may be akin to intermedi- localization and chromosome-origin segre- the reason that eukaryotic cells can be so ate filaments (12). Although we know key gation. MreB is found in a spiral pattern large, morphologically complex, and spe- cytoskeletal filaments, very few cytoskele- in Caulobacter (3, 18), as is the case in cialized. Protein structures are inherently ton-associated proteins have been identi- Escherichia coli and Bacillus subtilis (8, asymmetric, and head-to-tail assemblies of fied. Strikingly, no nucleators or motor 11). Moreover, MreB spirals are dynamic protein monomers generate long asym- proteins have been identified for any bac- throughout the cell cycle, compacting at metric filaments (4) (Fig. 1A). Eukaryotic terial cytoskeletal elements, and the only the division plane in predivisional cells, cells are able to exploit the inherent asym- known bundling protein is ZipA, which remaining there until division is complete, metry of cytoskeletal filaments to encode bundles FtsZ filaments (13). Indeed, it and then expanding to fill the cell (3, 18). positional information over long distances was thought previously that specific nucle- Analogous dynamic changes in helical and time scales and to transport cargo in ators and motor proteins may not exist for pitch have been observed for FtsZ (19) as a directed way. To exploit and amplify the bacterial cytoskeleton. FtsZ nucleation well as for bacterial flagella (20, 21). Im- filament asymmetry, eukaryotic cells use is suppressed outside the presumptive sep- portantly, when MreB is disrupted by ei- several classes of cytoskeleton-associated tation site by the Min system, suggesting ther depletion or overexpression, several proteins. For example, nucleators initiate that nucleation is spontaneous in vivo and independent cell polarity markers are mis- polymerization from specific cellular ad- may not require dedicated nucleation fac- localized or become uniformly distributed dresses and orient the resulting filaments. tors (14). Constriction of the FtsZ ring (3). These markers include chromosomal Motor proteins carry cargo along fila- during cell division and septation might origins that usually maintain extreme po- ments and can slide filaments relative to be motor-driven but might also be caused lar positions and four cell-cycle regulatory one another to sort randomly oriented by GTPase-dependent conformational proteins that dynamically localize to spe- filaments into polarized arrays. Bundling changes and filament depolymerization cific poles at different times in the cell proteins hold neighboring filaments to- (15). Moreover, rapid turnover of FtsZ in cycle. Ϸ gether to strengthen and stabilize ordered the assembled Z ring [half-life 30 sec The most remarkable result presented Ϸ arrays (Fig. 1B). (16)] and Mbl in spirals [half-life 8min in this article is that when MreB is re- A major advance in prokaryotic cell (17)] suggests that these arrays are made stored to depleted cells, polar protein biology in the last decade has been the up of small, overlapping filaments. Until localization is reestablished rapidly. More- discovery of the prokaryotic cytoskeleton. now no evidence has indicated whether FtsZ, homologous to eukaryotic tubulin, these arrays consist of parallel or mixed- forms rings at bacterial cell-division sites polarity filaments. See companion article on page 8643. (5–7). MreB and Mbl, prokaryotic coun- Gitai et al. (3) provide evidence to sug- *To whom correspondence should be addressed. E-mail: terparts to actin, have well established gest that the MreB spiral in Caulobacter [email protected]. roles in shape determination and chromo- consists of parallel filaments by demon- © 2004 by The National Academy of Sciences of the USA 8510–8511 ͉ PNAS ͉ June 8, 2004 ͉ vol. 101 ͉ no. 23 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0402881101 Downloaded by guest on September 25, 2021 COMMENTARY over, PleC and DivJ, which are normally However, there still must be another fac- exclusively at a single pole, are randomly tor (perhaps a discriminating bundler) to localized to either pole, thus half the cells ensure that all the filaments in the de regain polarity with reversed orientation novo assembled array are oriented parallel (Fig. 2A). The reestablishment of protein to each other. localization after restoration of MreB Next, assuming that MreB spirals are suggests that MreB is used to localize pro- polarized arrays, how does the cell use the teins to the poles. Furthermore, the ran- spiral to deliver polarity markers to the dom polar localization of PleC and DivJ appropriate ends of the cell? MreB is after MreB restoration implies that MreB required for chromosome segregation in self-assembles in a way that provides posi- Caulobacter (3), E. coli (11), and B. subtilis tional information spanning the entire cell (10). What links chromosome movement to MreB? Do motor proteins travel in an length. The simplest explanation for this Fig. 2. Polarity switching. (A) In wild-type cells, result is that MreB spirals are ordered orientation-specific manner to deliver PleC (blue dot) is localized to the swarmer pole, proteins and chromosomes to their appro- parallel arrays and when MreB filaments whereas DivJ (yellow dot) is localized to the stalk are formed de novo, the orientation of the pole. When MreB is depleted, localization of both priate locations? Alternatively, do end- array is random. One critical question not proteins is lost. After recovery of the MreB fila- binding proteins analogous to EB1 for addressed by the Gitai et al. (3) article is, ment, polar localization of both PleC and DivJ is microtubules or formin for actin filaments, are PleC and DivJ mislocalized in the restored, but the orientation is now random. It is which remain associated with the fast- same cells? important to note that PleC and DivJ were ob- growing ends, function to deliver cargo to This intriguing result opens the door for served separately, and it is not known whether the particular cellular locations? localization pattern is reversed for both proteins in In the future, it will be critical to con- many exciting questions and avenues of the same cells (3). (B) Compaction of MreB spirals to research. First, if the Caulobacter MreB duct time-lapse experiments to track fur- the midcell during division may provide a means to ther the recovered cells that have spiral is a polarized array, how does it reset array orientation after each cell cycle to en- switched polarity. Several questions re- form and how is it maintained? What are sure the same orientation in both daughter cells. main about these reversed cells. Are these the dynamics and pattern of de novo array cells able to grow and divide, or are they formation? The random orientation of orientation switching may be related to stuck in a developmental dead end? If polarity argues against a prelocalized nu- they are able to proceed, what happens cleator initiating spiral regrowth from a the cell-cycle-dependent compaction and regrowth of MreB spirals. Old filaments during division and in the next genera- single fixed point. Instead, polymerization tion? Does MreB collapse to the