Current Biology, Vol. 14, R242–R244, March 23, 2004, ©2004 Elsevier Ltd. All rights reserved. DOI 10.1016/j.cub.2004.02.057 Bacterial Shape: Concave Coiled Coils Dispatch Curve Caulobacter William Margolin Caulobacter crescentus is a popular model prokaryote, mainly because of its eukaryotic-like cell cycle control, different cell types and asymmetric Bacterial cells exhibit a wide variety of shapes. Recent growth [11]. A member of the prosthecate family of results indicate that the characteristic crescent shape alpha-proteobacteria, C. crescentus derives its species of Caulobacter crescentus depends upon an inter- name from the crescent shape of the cells, which have mediate filament-like protein that localizes to the a flagellum at one pole and a stalk at the opposite pole. concave side of the cell. Interestingly, after a long time in stationary growth phase, C. crescentus can elongate into long, helical cells [12], suggesting that the crescent shape of the normal Eukaryotic cells have three main types of cytoskeletal short cells is actually a truncated helix. The important element: microtubules, composed of tubulin; breakthrough came when Ausmees et al. [10] found that microfilaments, composed of actin; and intermediate a transposon insertion mutant caused C. crescentus filaments, made up of proteins from a number of cells to become straight rods. These rods still had stalks different families. A decade ago, the evolutionary origin at one pole, indicating that their characteristic polarity of these cytoskeletal systems was unknown. Now, a was intact, but the curved shape was lost. great deal of structural and biochemical evidence points The authors then identified the protein encoded by to FtsZ and MreB as being the prokaryotic ancestors of the interrupted gene, which they dubbed CreS, for cres- tubulin and actin, respectively [1,2]. This suggests, not centin. Remarkably, studies with immunofluorescence only that eukaryotic cytoskeletal proteins arose early in or with a green fluorescent protein (GFP) fusion to CreS evolution, but also that bacteria have a cytoskeleton. showed that crescentin localizes specifically to the Whereas microtubules are involved in locomotion, concave surface of the cell, near the membrane. This mitosis and some aspects of cytokinesis, FtsZ seems localization pattern is continuous and not punctate, to be limited to cytokinesis in prokaryotes and some suggesting that the protein assembles into a polymer eukaryotic organelles [3,4]. Actin is involved in many along the membrane, similar to FtsZ or MreB. This was processes, including locomotion, cell growth, and supported by the observation that CreS–GFP localizes cytokinesis, whereas MreB proteins function in to a long continuous helix in stationary phase helical controlling cell growth and shape, and have an as yet cells that matches the helical shape of the filamentous unclear role in chromosome segregation [5–7]. The cell cells, hugging the inside curve. shape function of the MreB family of proteins is strik- The expression of CreS–GFP in the absence of wild- ing. For example, inactivating a MreB homolog in nor- type CreS had a dominant-negative effect, ablating the mally cylindrical bacteria, such as Escherichia coli or usual curved cell shape. Interestingly, CreS–GFP under Bacillus subtilis, causes the cells to lose their cylindri- these conditions still localized to a helix within the cal shape and become the ‘default’ round shape [8,9]. straight cells, but this helix appeared to be detached Consistent with the view that MreB plays a role in from the membrane. This suggests that crescentin imposing a more complex cylindrical cell architecture, itself assembles into an intrinsic helical polymer and is naturally round bacteria such as streptococci do not not a decoration of an independent helical structure. have mreB homologs in their genomes. Interestingly, Because a low copy plasmid expressing creS induced rhizobia and corynebacteria do not have MreB, but are curved shape in the creS transposon insertion strain, still cylindrically shaped, possibly because they grow at the CreS filament must be able to form de novo, and is their tips like filamentous fungi [5]. Their rod shape may not dependent on continuation of existing crescentin depend on other proteins that are yet to be discovered. structures. If MreB proteins are required for maintaining a Just these localization and function results would be cylindrical shape in most bacteria, then what might of significant interest, but Ausmees et al. [10] did not confer what appear to be more complex shapes, such stop there. A search for proteins with similar sequence as helices? For example, vibrioid (‘comma’) or helical to crescentin revealed weak homology with proteins shapes are characteristic of a relatively small number having a particular organization of coiled-coil domains, of bacterial species. Is there a gene to keep a helical one example of which is nuclear lamin A, a member of shape from reverting to a straight cylinder, analogous the intermediate filament protein family. The domain to the need for mreB in cylindrical shape? Or is helical organization of crescentin is also similar to that of shape the default, with specific genes being required intermediate filament proteins, so the authors decided for straight cylinders? Ausmees et al. [10] have to test whether purified crescentin might polymerize recently addressed this question, and arrived at an under conditions used to assemble intermediate fila- astonishing answer. ments. They found that crescentin can self-assemble into 10 nm wide filaments after solubilization in 6 M Department of Microbiology and Molecular Genetics, guanidinium and subsequent removal of the denatu- University of Texas Medical School, 6431 Fannin, Houston, rant by dialysis, precisely the conditions used to Texas 77030, USA. E-mail: [email protected] assemble other intermediate filaments. Current Biology R243 Figure 1. A model for crescentin action. E. coli C. crescentus C. crescentus lacking crescentin Crescentin localizes as an intermediate fil- ament-like polymer bundle along the concave side of C. crescentus cells and is required for the characteristic curved shape of this species. It is proposed that the crescentin filament (red) interacts with the actin-like MreB coil (green) to inhibit cell wall biosynthesis at specific sites (yellow), resulting in left-right asymmetry during growth. Cells lacking crescentin, such as E. coli or crescentin-deficient C. crescentus mutants, are straight cylinders because of the action of MreB. Even without crescentin, C. crescentus displays one axis of cellular asymmetry, with a fla- gellum at one pole (top) and a stalk at the opposite pole (bottom). No obvious asymmetry North-south asymmetry North-south asymmetry East-west asymmetry Current Biology Intermediate filaments are present in many eukaryotic structure or expression of bacterial actin homologs cells, but are not required in every cell type. Moreover, sometimes results in dramatically curved shapes in nor- obvious homologs have not been found in plants and mally straight E. coli and B. subtilis cells [8,18], indicat- fungi, although intermediate filaments may still be ing that crescentin may mimic some of these abnormal present in these organisms. The roles of intermediate fil- interactions. Whether crescentin, like lamins, can aments are not completely understood, but they seem induce shape changes in heterologous bacterial hosts to provide resistance to mechanical stresses, and they is yet to be determined, but might be expected if the can withstand deforming forces without breaking much interacting partners are conserved. better than can actin filaments or microtubules [13]. How widespread are intermediate filament-like The nuclear lamins are thought to be the evolution- shape-determining proteins in bacteria? Standard ary precursors of the general intermediate filament sequence similarity searches are not too helpful, as they superfamily [14]: they function as tensegrity elements, yield other proteins with coiled-coil domains, such as resisting deformation and conferring a specific shape myosins. Ausmees et al. [10] did, however, find several on the nucleus [15]. Interestingly, a specific lamin is candidate proteins in Helicobacter pylori, another necessary to maintain hook-shaped nuclei in mouse curved bacterium, by searching for proteins with similar spermatocytes and sufficient to induce hook-shaped domain organization. It will be useful to learn whether nuclei in somatic cells [16]; this is intriguingly reminis- vibrios, curved relatives of E. coli, have crescentin cent of curved C. crescentus cells, although the mech- homologs. Interestingly, treponemes, members of the anism is likely to be different. Lamin mutations result in spirochete family, already have a well-known shape- shape abnormalities, even massive deformation, of the determining cytoskeleton — a thick, helical cytoplasmic nuclear envelope. Lamins have also been implicated in filament that is necessary for the distinctive helical transcriptional regulation [15]. So it is not far-fetched shape of these organisms [19]. The filament itself is to think that crescentin-type proteins might be able to composed of only one protein, CfpA, which has no regulate cell shape in multiple ways, and even control homologs in the database, but is in a complex that is genes and proteins involved in cell wall biosynthesis. anchored to the membrane and has other components How might crescentin actually