This Article Abstract Full Text (PDF) An author's correction has been published Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Carballido-López, R. Search for Related Content PubMed PubMed Citation Articles by Carballido-López, R.

The Bacterial Actin-Like Cytoskeleton

Génétique Microbienne, Institut National de la Recherche Agronomique, 78352 Jouy-en-Josas Cedex, France

SUMMARY

INTRODUCTION

EUKARYOTIC ACTIN AND THE EUKARYOTIC ACTIN CYTOSKELETON

The Actin Superfamily

THE BACTERIAL ACTIN MreB

MreB Filaments Are Generated by Actin-Like Polymerization

Structure of MreB monomers and MreB protofilaments.

MreB assembly properties.

Dynamics of MreB Filaments

Ultrastructural Organization of MreB Filaments In Vivo

MreB-LIKE PROTEINS AND CELL MORPHOGENESIS

MreB Filaments Govern Cell Morphogenesis by Actively Directing Lateral Wall Biogenesis

Mbl filaments direct lateral wall synthesis in Bacillus subtilis.

PBP localization and the MreB cytoskeleton.

MreBH filaments direct lateral wall hydrolysis in Bacillus subtilis.

Is Transmission of Shape Mediated by MreB-Directed Peptidoglycan Factories?

The Essential MreBCD Complex and Lateral Wall Synthesis

MreB Proteins and Spore Lateral Wall Formation in Actinomycetes

MreB Proteins and Cell Shape Determination in Wall-Less Prokaryotes

ACTIN-LIKE PROTEINS AND DNA SEGREGATION

The Actin-Like ParM Protein and Plasmid-DNA Segregation

ParM.

Model for in vivo function of ParM filaments.

The Actin-Like MreB Protein and Chromosomal DNA Segregation

MreB-LIKE PROTEINS AND CELL POLARITY

MreB-LIKE PROTEINS AND CELL DIVISION

OTHER PROKARYOTIC ACTIN-LIKE PROTEINS

MamK

Ta0583

MreB-ASSOCIATED PROTEINS: TOWARD AN UNDERSTANDING OF THE FUNCTIONS OF THE BACTERIAL ACTIN-LIKE CYTOSKELETON

Trafficking of Proteins: Going Helical

Are There MreB-Associated Proteins That Modulate Filament Organization?

The MreB Hub: a Central Organizing Role for the Actin-Like Cytoskeleton

CONCLUSIONS

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 
The cytoskeleton is a key regulator and central organizer of many eukaryotic cellular processes, including cell shape determination (morphogenesis), division, segregation, polarity, phagocytosis, movement, and macromolecular trafficking. It is a complex and highly dynamic network of protein filaments composed of actin microfilaments, microtubules (MTs [polymers of tubulin]), and intermediate filaments (IFs). These proteins were traditionally thought to be absent in prokaryotes, and the eukaryotic origin of the cytoskeleton was a long-standing dogma of cell biology. For decades, microscopic and biochemical studies failed to detect cytoskeletal elements in bacteria. Moreover, the sequencing of an increasing number of bacterial genomes (352 to date [15 September 2006] [Genomes Online Database, release 2.0 {www.genomesonline.org}]) did not reveal any putative candidates displaying any significant primary sequence similarity to a cytoskeletal protein. However, over the last 15 years, the dogma has been overturned completely, with the identification in bacteria of structural and functional homologues of all three main eukaryotic cytoskeletal proteins: FtsZ and BtubA/B are tubulin orthologues; MreB, ParM, and the recently uncovered MamK (and the archaeal Ta0583) are actin orthologues; and crescentin is an intermediate filament protein.

The first to be identified, in the early 1990s, was the tubulin-like protein FtsZ. FtsZ is a highly conserved cytosolic GTPase (23, 136), present in virtually all eubacteria (and archaea), which forms a ring (namely the Z ring) at the future site of cytokinesis and plays an essential role in cell division (4, 6). Although FtsZ's primary amino acid sequence identity to tubulin is low (17%), their three-dimensional (3D) structures and assembly properties are remarkably similar (106, 124). Two other tubulin homologues, BtubA and BtubB, were recently identified in the bacterial genus Prosthecobacter (76). BtubA and BtubB display higher sequence identity (35%) and closer structural homology to eukaryotic tubulin than to FtsZ (BtubA/B and FtsZ sequences share only 8 to 11% identity) (76, 149). It is believed that FtsZ and tubulin diverged from a common ancestor early in evolution, whereas BtubA/B likely split from eukaryotic tubulin more recently by horizontal gene transfer (76, 149, 157).

Many attempts had been made to isolate actin-like and actinomyosin-like complexes from bacterial cells, but none of these studies was conclusive and correlated with a specific protein. The breakthrough came in 2001, when the MreB and Mbl (MreB-like) proteins of Bacillus subtilis were shown to be required for different aspects of cell morphogenesis and to assemble into helical structures that run along the length of the cell (80). Shortly after, the nature of these helical filaments was revealed when purified MreB from Thermotoga maritima was shown to undergo actin-like polymerization and to have a three-dimensional structure remarkably similar to that of actin (164). Two other actin homologues that form cytoskeletal structures in bacterial cells (ParM and MamK) and one archaeal actin (Ta0583) have since been identified.

Finally, in 2003, the Caulobacter crescentus coiled-coil-rich protein crescentin was shown to assemble into filaments that play a key role in determining the curved and helical cell shapes of this bacterium and to have biochemical properties and a domain structure similar to those of IFs (3). Furthermore, in addition to homologues of eukaryotic cytoskeletal proteins (actin, tubulin, and IFs), a subclass of filament-forming Walker A ATPases (85) belonging to the large MinD/ParA superfamily was recently categorized as a new class of bacterial cytoskeletal proteins. These proteins, renamed Walker A cytoskeletal ATPases (114), form ATP-induced dynamic filaments in vivo and play important organizing roles in cell division (MinD subgroup) and plasmid/chromosome DNA segregation (ParA/Soj subgroup) in bacteria. Although Walker A cytoskeletal ATPases display no homology to known eukaryotic cytoskeletal elements, they are now considered an additional component of the prokaryotic cytoskeleton (for recent reviews, see references 114 and 142). The discovery of cytoskeletal elements in bacteria opened up new and exciting fields of research, which have evolved rapidly over the last few years. Most of the advances made have arisen from developments in imaging technology and analysis, in particular high-resolution fluorescence microscopy techniques, which previously could not be applied to organisms as small as a bacterium.

Among all prokaryotic cytoskeletal proteins, the field of bacterial actins has developed the most in recent years. The discovery of MreB has led to a continuous flow of new and important findings from several organisms, and the MreB-like proteins have become a major research focus in many laboratories. It is now clear that prokaryotic cells possess actin and that a dynamic actin-like cytoskeleton is involved in a variety of essential cellular processes in bacteria. These functions, like those of the eukaryotic actin cytoskeleton, require the targeting and accurate positioning of proteins and molecular complexes. A series of landmark papers investigating the roles of the actin-like proteins has provided tremendous insights into the mechanisms of cell wall (CW) morphogenesis, DNA segregation, and cell polarity in bacteria.

In this review, I aim to draw together what is known about the cellular, structural, and biochemical properties of MreB (and ParM and MamK) proteins (the bacterial actins). I examine their known roles before considering other possible functions for these cytoskeletal proteins. Currently, interacting proteins for MreB and its relatives remain largely unknown. The quest for them and for a few proteins known to associate with the dynamic bacterial actin-like cytoskeleton is also discussed. The bulk of these studies have been performed with the model organisms Bacillus subtilis, Escherichia coli, and Caulobacter crescentus, although some findings have emerged from other systems (e.g., Thermotoga, Rhodobacter, and Streptomyces), and they are generally thought to be conserved throughout eubacteria. Some perspectives on directions for future research in the field are also provided.

EUKARYOTIC ACTIN AND THE EUKARYOTIC ACTIN CYTOSKELETON

Top Previous Next References

 
Actin is one of the most abundant and highly conserved proteins found in all eukaryotic cells. The 43-kDa monomer of conventional actin (globular actin, or G-actin) spontaneously assembles in vitro to form long linear or branched structures (filamentous actin, or F-actin) upon the addition of salt, provided that ATP is present (86). The filaments polymerize noncovalently from both ends, with different affinities for the addition of monomers at each end. This results in an intrinsic polarity in the filament, in the form of a slow-growing end (minus end) and a fast-growing end (plus end). At steady state, the loss of subunits at the minus end and the equivalent gain at the plus end give rise to an effect known as treadmilling (17). Actin microfilaments are thin (3 to 6 nm in diameter) and flexible, and they rarely occur in isolation within the cell but, rather, in cross-linked aggregates and bundles. In vivo, they can form either stable or labile structures. Actin polymerization is a highly regulated process controlled both by nucleotide binding and hydrolysis and by the action of a number of actin-binding proteins that can cross-link, nucleate, cleave, bundle, stabilize, or destabilize the filaments (18, 86, 150). The visualization of the eukaryotic actin cytoskeleton in an unperturbed, close-to-life state was achieved only recently, with the application of cryo-electron tomography to vitrified cells of Dictyostelium discoideum (92).

The actin cytoskeleton is highly dynamic in most cells, and F-actin populations continuously assemble and disassemble, with measured half-lives on the order of a few minutes. This turnover is a consequence of the ATPase activity of actin. Irreversible hydrolysis of the bound nucleotide occurs once the monomer is fully incorporated into the filament (86), and thus, like the case for GTP hydrolysis in tubulin polymerization, it is not required to form the actin filaments. Instead, it destabilizes the polymer and promotes depolymerization from its ends since ATP monomers prefer to associate and ADP monomers prefer to disassociate (111). A difference between the two cytoskeletal polymers, i.e., F-actin and MTs, is that GTP-GDP exchange is very rapid for free tubulin (half-time of seconds) while ATP-ADP exchange is relatively slow for free actin (half-time of minutes).

View larger version (41K): [in this window] [in a new window]   FIG. 1. Properties of MreB-like proteins. (A to C) Effects of mutations of Bacillus subtilis mreB and mbl on cell shape. Phase-contrast images are shown. Bar, 5 µm. (A) Wild-type cells showing the typical rod shape. (B) Lytic phenotype of cells depleted of MreB. (Panels A and B are reprinted from reference 55 with permission from Blackwell Publishing.) (C) Aberrant morphology of mbl mutant cells. (Reprinted from reference 80 with permission from Elsevier.) (D to F) Helical filaments formed in vivo by the three MreB-like proteins of B. subtilis. (D) GFP-MreB (courtesy of A. Formstone [unpublished]); (E) GFP-Mbl; (F) GFP-MreBH. (G and H) Electron micrographs of negatively stained MreB filaments. (G) MreB forms filamentous bundles and ring-like structures. An enlarged ring structure is shown in panel H. Bars, 0.2 µm and 0.1 µm, respectively. (Reprinted from Journal of Biological Chemistry [47] with permission of the publisher.) (I and J) Comparison of the three-dimensional structures of actin and MreB. (I) Superimposition of uncomplexed actin (purple) and MreB from Thermotoga maritima (blue). (Reprinted, with permission, from the Annual Review of Biophysics and Biomolecular Structure [107] volume 33, copyright 2004 by Annual Reviews.) Despite having very low amino acid sequence identity (15%), the two proteins have essentially the same fold. The arrows point to insertions within the actin sequence responsible for binding to DNase I. The arrowhead indicates a loop proposed to make an important interstrand interaction. (J) MreB crystals contain protofilaments that are similar to one strand (protofilament) of modeled F-actin. PDB entry codes are shown in parentheses. (Reprinted from reference 164 with permission from Macmillan Publishers Ltd.)

View larger version (40K): [in this window] [in a new window]   FIG. 5. Actin-like filaments formed by plasmid segregation protein ParM. (A) ParM axial filaments in Escherichia coli visualized by combined phase-contrast and immunofluorescence microscopy. (B) Electron micrograph of negatively stained ParM filaments. Polymers have a cross-sectional diameter of 7 nm. Bar, 50 nm. (Panels A and B were reprinted from reference 120 with permission from Macmillan Publishers Ltd.) (C) Ribbon representation of the three-dimensional monomer structures of proteins of the actin superfamily, including eukaryotic actin and prokaryotic MreB, ParM, and Ta0583 (MreB-AMPPNP {adenosine 5'-[ß--imido]triphosphate}; ADP-bound form for the others) (PDB accession numbers are 1J6Z, 1JCG, 1MWM, and 2FSN, respectively). Subdomains IA, IB, IIA, and IIB correspond to subdomains 1, 2, 3, and 4 in actin. The ADP molecule bound to actin and to ParM is shown in ball-and-stick representation. (Reprinted from reference 137 with permission from Elsevier.) (D) Model for dynamic instability and in vivo function of ParM filaments during the cell cycle. ParM filaments (red) spontaneously assemble/disassemble, displaying dynamic instability in the cell. Upon plasmid (green) replication, plasmid pairs are held together by ParR-bound protein (yellow squares). Both ends of a set of ParM filaments are captured and stabilized by the ParR nucleoprotein complex. As more copies of ParM-ATP (blue) are added at the ParM-ParR interface, the filament/spindle elongates, pulling apart the plasmid copies toward either end of the cell before cell division occurs at midcell. (Reprinted from reference 57 with some modifications, with permission of the publisher. Copyright 2004 American Association for the Advancement of Science.)

Top Previous Next References

MreB depletion has been shown to induce the formation of enlarged cells with gross morphological defects and, ultimately, cell lysis in B. subtilis (Fig. 1B), E. coli, and C. crescentus (52, 80, 88). B. subtilis mbl mutants also display highly distorted morphologies, with bent, twisted, and irregularly shaped cells, a proportion of which are also affected in cell width (Fig. 1C). In light of these findings, an actively determined, MreB-dependent cell shape system was suggested to be conserved across nonspherical microorganisms.

Subcellular localization studies using immunofluorescence microscopy (IFM) and green fluorescent protein (GFP) fusions in several bacteria have shown that MreB-like proteins generally localize to helical filamentous structures that encircle the cytoplasm, just under the cell membrane (see below). In B. subtilis, all three isoforms, i.e., MreB, Mbl, and MreBH, form similar helical structures (Fig. 1D to F, respectively) (15, 26, 55, 80). Pioneer localization studies suggested that MreB, Mbl, and MreBH form distinct helical structures with different configurations (26, 80), but these studies were done with separate cell populations with different genotypes and in separate imaging experiments. Colocalization studies (i.e., simultaneous, same-cell imaging) have recently shown that all three of the B. subtilis MreBs are in fact in close proximity, in a single apparently helical structure (15). Such filament-like helices could result from the interaction of monomeric MreBs with a preexisting helical structure in the cell or, alternatively, from the noncovalent association of the monomers (i.e., polymerization) into high-order helical forms.

MreB assembly properties. Despite their high structural homology, MreB and actin display significantly different assembly properties and nucleotide-binding specificities. Light-scattering and EM studies have been used to explore the basic assembly and mechanical properties of MreB from T. maritima (47, 48). As with F-actin, MreB assembly is triggered by ATP in vitro, and the filament ultrastructure and polymerization are temperature and cation dependent (47, 164). Furthermore, MreB catalyzes ATP hydrolysis and releases phosphate (P_i) at a similar rate to that of F-actin (47). However, GTP (but not ADP or GDP) can mediate MreB assembly as effectively as ATP (whereas eukaryotic actin assembly is favored in the presence of ATP over GTP), indicating that MreB is an equally effective ATPase and GTPase (48, 164). MreB polymerizes much more rapidly than actin, without nucleation (or nucleation is highly favorable and fast) and with little or no contribution from filament end-to-end annealing (i.e., joining of filaments through the direct association of filament ends [annealing], which contributes significantly to the assembly of actin filaments) (47). MreB exhibits a critical concentration of 3 nM, which is 100-fold lower than that of actin. Finally, without the need for accessory proteins, MreB was shown to form predominantly filamentous bundles that display different morphologies and have the ability to spontaneously form ring-like structures (Fig. 1G and H) (47). The presence of both straight and curved filaments was suggested to depend upon the state of nucleotide hydrolysis within the filament (48), a phenomenon that has also been observed in filamentous proteins such as microtubules (125) and FtsZ (108). Using quantitative rheometry, Esue et al. (48) recently showed that MreB filaments possess significant elasticity and mechanical stiffness, also like MTs, and are much less labile than actin filaments in networks. It should be noted that another difference between MreB and actin applies at the filament level, as MreB assembles into single straight protofilaments (164), not into double-helical protofilaments that twist around each other like the case for F-actin (and ParM) (165; see below). Since the polypeptide chain of actin (375 amino acids) is longer than that of MreB (336 amino acids), there are a number of insertions that occur within the actin sequence, and these might account for the differences in the properties of the two proteins. For example, one insertion (arrowhead in Fig. 1I) forms a loop that has been proposed to make an important interstrand interaction that holds the actin filament together, whereas two other insertions (arrows in Fig. 1I) are responsible for the binding of actin to DNase I (41).

View larger version (45K): [in this window] [in a new window]   FIG. 2. In vivo dynamics of MreB filaments. (A) Time-lapse microscopy of GFP-Mbl dynamics in Bacillus subtilis. A 4-h time course was monitored every 30 min. The crosses indicate fragmentation of the helical structures at about the time of division. Bar, 2 µm. (Reprinted from reference 13 with permission from Elsevier.) (B to D) MreB forms a contracting and expanding spiral in Caulobacter. (B) Helical pattern of GFP-MreB in a stalked cell. The panels show, from top to bottom, a single deconvolved optical section, a volume projection of 15 optical slices, and a cartoon interpretation of the spiral pattern overlaid on the image from the middle. (C) Hollow ring organization of division-plane GFP-MreB localization. The ring is visualized as a continuous band in one optical section (top) and as two separate points in an optical section 300 nm deeper into the cell (middle). The bottom panel shows a cartoon representation of the ring and cell outline overlaid on the image from the panel above. Bars, 1 µm. (D) Schematic representation of the dynamic behavior of GFP-MreB during the Caulobacter cell cycle. (Reprinted from reference 61 with permission of the publisher. Copyright 2004 National Academy of Sciences, U.S.A.) (E) FRAP analysis of GFP-Mbl in live B. subtilis cells. The first panel shows the helical cables of Mbl 0.5 min before photobleaching. The next frame shows the laser photobleaching of a rectangular region covering the lateral half of a cell (t0). The following image sequence monitors fluorescence recovery in the bleached Mbl cables at the indicated times (in min). The recovery of fluorescence occurred at the expense of fluorescence in the unbleached cables (indicating an exchange of subunits), with a half-time of 8 min. Bar, 1 µm. (Reprinted from reference 13 with permission from Elsevier.)

MreB-LIKE PROTEINS AND CELL MORPHOGENESIS

Top Previous Next References

 
Prokaryotic cells display a wide diversity of shapes. In all eukaryotic cells, shape is determined primarily by cytoskeletal structures, particularly actin filaments. Since such structures were traditionally thought to be absent from bacteria, the tough external peptidoglycan (PG) cell wall was assumed to be the primary determinant of bacterial cell shape. Indeed, isolated PG sacculi comprise a single huge molecule that retains the shape associated with their original cell (69), and conversely, removal of the wall results in spherical protoplasts (i.e., loss of shape). Moreover, most genes identified to be involved in cell shape (morphogenes) were associated with CW biosynthesis. For decades, research on cell shape in bacteria focused on cell wall synthesis and structure (for a recent review, see reference 133). However, the discovery of a morphogenetic actin-like cytoskeleton brought about a radical change in the context in which bacterial cell shape is studied.

Mbl filaments direct lateral wall synthesis in Bacillus subtilis. A range of studies with rod-shaped bacteria suggested that PG insertion occurs at the nascent septum and randomly all over the surface of the lateral wall during growth (10, 27, 30, 113, 117). However, the sensitivity and resolution of the methods used in these studies were limited: they did not provide adequate spatial resolution to determine the underlying patterns of PG incorporation. A more sensitive, high-resolution probe for nascent PG insertion in nonfixed cells was recently developed by Daniel and Errington (22). These authors used a fluorescent derivative of the antibiotic vancomycin (a cell wall synthesis inhibitor that binds specifically to PG intermediates) to label nascent PG in gram-positive bacteria by fluorescence microscopy (note that gram-negative bacteria do not stain because their outer membrane presents a permeability barrier to vancomycin) (169). This novel staining method revealed that, at least in B. subtilis, synthesis of the wall occurs in a helical pattern over the cylindrical part of the cell and also specifically at the septum in dividing cells (Fig. 3A). The lateral helical pattern of fluorescein-labeled vancomycin (Van-FL) was reminiscent of the helical localization of MreB and Mbl, both of which are required for cell shape determination in B. subtilis (1, 80). Strikingly, the helical staining was abolished in a strain lacking Mbl (Fig. 3C) and not in a strain lacking MreB (Fig. 3D) (22, 55). The septal insertion was dependent on cell division (FtsZ), as expected (Fig. 3E) (22). Recently, in a similar independent study that used fluorescent derivatives of vancomycin and ramoplanin (another PG-binding antibiotic), insertion of nascent PG along the lateral wall of B. subtilis was confirmed to be helical (160). However, in this study, a sidewall helical staining pattern qualitatively similar to that observed in wild-type cells (although less regular) was observed in cells lacking Mbl (160); the authors of that study concluded that Mbl plays an indirect role in directing PG synthesis but that it is not essential for the incorporation of sidewall PG.

View larger version (28K): [in this window] [in a new window]   FIG. 3. Control of lateral wall synthesis by MreB-like proteins. (A) Van-FL staining of nascent cell wall synthesis in wild-type cells of Bacillus subtilis. A compilation of images of typical cells during the cell cycle progression is shown from left to right. (B) Model for separate systems for septal and cylindrical PG synthesis in B. subtilis. Gray lines and ovals show the sites of PG insertion during elongation and division, respectively. Arrows show the directions of elongation or division driven by cell wall biosynthesis. (C to E) Effects of mbl null mutation (C), mreB null mutation (D), and FtsZ depletion (E) on Van-FL staining. (Panels A, B, C, and E were reprinted from reference 22 with permission from Elsevier. Panel D was reprinted from reference 55 with permission from Blackwell Publishing.) (F) Schematic representation of interactions within the MreBCD complex in E. coli and the possible interaction of MreBCD with periplasmic proteins involved in peptidoglycan synthesis (PBP2 and RodA). (Reprinted from reference 88 with permission from Blackwell Publishing.)

Taken together, these findings have several important implications. First, insertion of PG along the cylindrical walls of B. subtilis (and possibly E. coli) cells occurs in a helical pattern. Second, there are two spatially specialized systems for PG synthesis in B. subtilis (and probably in all rod-shaped bacteria with an MreB system): dispersed helical insertion of PG throughout the lateral wall during growth results in rapid cylindrical extension (elongation), and cell division-directed PG synthesis allows septum formation (division) (Fig. 3B). Third, Mbl may be the MreB homologue required mainly, or exclusively, for lateral (helical) wall biosynthesis. This has raised interesting questions about the role of MreB in B. subtilis morphogenesis. MreB is essential under normal growth conditions and has an important role in the control of cell width (55, 80). Consistent with a role in CW integrity, the lethal mreB mutant phenotype could be ameliorated by high concentrations of magnesium (Mg²⁺) (55), like the phenotypes of several B. subtilis mutants thought to be required for different aspects of PG synthesis (94, 95, 121, 138, 139; see below). The mechanism by which Mg²⁺ is able to rescue the phenotypes of these mutants is currently unknown. It also remains unclear how MreB controls cell width, but it might influence the synthesis or structure of the cylindrical and/or septal CW. Van-FL staining showed that MreB was not primarily required for lateral PG synthesis (22, 55; see above), although a redundant role in this process cannot be excluded. Alternatively, MreB could be required for insertion of teichoic acids or of autolysins into the CW, but all of these hypotheses remain to be tested. It has been suggested that MreB might act either continuously to restrain the diameter during elongation or discontinuously to reset the correct diameter when the cell divides (55). Finally, it is exciting to mention that a new, distinct role in cell wall morphogenesis has recently been uncovered for MreBH, the third MreB isologue in B. subtilis (15; see below).

PBP localization and the MreB cytoskeleton. The Van-FL findings provided strong support for the view that at least the cables of Mbl direct the synthesis of lateral PG in a spatially controlled manner. Conceivably, this might involve the localization of PG-synthesizing enzymes, named penicillin-binding proteins (PBPs), that incorporate the PG precursors into the growing CW sacculus. Two factors (or rather a combination of them) are thought to be critical for PBP localization, and they are protein-protein interactions and substrate recognition (147). The PG-synthesizing septal machinery (i.e., septal PBPs) has been shown to be recruited by the FtsZ ring (44, 109). However, no factors for targeting of PBPs to the lateral wall have yet been identified. Hence, the helical MreB scaffolding structures could direct PBP localization either by providing a substrate(s) that can be recognized by elongation-specific PBPs or by providing a track for protein-protein interactions that target PBPs to their site(s) of action (see below).

A putative candidate for MreB-directed targeting is the product of pbpA, PBP2, a high-molecular-weight PBP that displays transpeptidase activity (i.e., catalyzes PG cross-linking) and has classically been associated with sidewall synthesis during elongation (158). In E. coli, a functional GFP-PBP2 fusion localized preferentially in a spot-like pattern over the cylindrical part of the envelope, and also at midcell during cell division (28). The localization of GFP-PBP2 over the lateral wall was suggestive of a helical pattern strikingly similar to that of MreB. Consistent with this, PBP2 formed a banding pattern reminiscent of that formed by MreB filaments in C. crescentus cells, as shown by both IFM (35, 52) and the use of a GFP-PBP2 fusion (40). Since the distinct banding pattern of PBP2 was lost (although PBP2 foci were still present) in C. crescentus cells that had been depleted of MreB for 10 h, it was originally suggested that PBP2 localization was dependent on MreB (52). Under such conditions, the aberrant localization pattern observed could indeed be attributed to the lack of MreB or, alternatively, to a secondary effect resulting from the severe shape defects resulting from the long-term absence of MreB. The latter hypothesis was supported by a recent IFM study, where rapid disruption of the MreB filaments by treatment with A22 did not affect the helical pattern of PBP2 (35). A22 is a small molecule that specifically and rapidly (<1 min) delocalizes MreB in Caulobacter cells, allowing the effects of the absence of MreB helices to be assessed prior to observable deformations in shape (62; see below for more details). However, in an independent study, GFP-PBP2 was reported to mislocalize to the division plane upon A22 treatment (40). Interestingly, only newly synthesized GFP-PBP2 seemed to relocalize to the septum, and thus it was suggested that (MreB-dependent) helical localization of PBP2 was regulated at the level of insertion and that established helical PBP2 patterns are stable in the absence of MreB (35). Although this finding may explain the differences between the two reports, whether PBP2 localization depends on MreB in C. crescentus remains to be unambiguously demonstrated. (Indeed, in the GFP-PBP2 study [40], cells were treated with 10 µg/ml A22 for 6 h, whereas in the IFM study [35] cells were treated with 50 µg/ml A22 for 2 h. After 2 h of A22 treatment at 50 µg/ml, growth is arrested but shape defects are not yet visible [62], and thus the previously inserted PBP2 patterns could be stable. On the other hand, C. crescentus cells grown in 10 µg/ml A22 for 6 h keep growing [albeit slowly] but already display significant cell shape deformations [62], which could indirectly affect PBP2 localization.)

IFM studies of R. sphaeroides cells showed that MreB colocalized with PBP2 in a cell cycle-dependent manner (155). PBP2 localized in partial rings (which presumably represent unresolved helices) at the middle of elongating cells and at the one-quarter and three-quarter positions in septating cells. MreB colocalized with PBP2 during elongation only; during septation, MreB remained at the septation site, whereas PBP2 relocalized to the midcell sites of the forming daughter cells. It was concluded that MreB and PBP2 interact during elongation to synthesize PG at or near midcell but are involved in different cellular roles during septation (155). The possible dependence of PBP2 localization on MreB in R. sphaeroides (and in E. coli) remains to be tested, and this point needs to be resolved for C. crescentus (see above). Work with B. subtilis showed that some PBPs fused to GFP also localize specifically to the sidewall in distinct foci and bands around the cell periphery, again reminiscent of the helical distribution of MreB and Mbl (146). This further suggested that PG synthesis occurs at distinct regions of the lateral wall and that new PG is inserted in a helical manner during the elongation-specific phase. In this study, however, the patterns of GFP-PBP localization were not detectably altered in the absence of either MreB or Mbl (146). Nevertheless, in light of the recent finding that MreB, Mbl, and MreBH colocalize in a single helical structure in B. subtilis (15), it remains plausible that PBP localization depends on more than one MreB isoform in this organism. The link between the MreB cytoskeleton and the cell wall synthesis machinery undoubtedly remains a major challenge for future work.

MreBH filaments direct lateral wall hydrolysis in Bacillus subtilis. It was recently shown that MreBH also has an important role in B. subtilis cell morphogenesis and that this function is effected, at least in part, by controlling the autolytic activity over the lateral wall by direct interaction with LytE, a cell wall hydrolase (15). Depletion of MreBH led to a mild cell shape defect, in contrast to the case for MreB- and Mbl-depleted cells (24). Similarly, under normal growth conditions, mreBH mutant cells displayed a mild phenotype; they were slightly affected in length and width and were bent at points that appeared correlated with abnormal thickening of the CW (15). However, all aspects of the mreBH mutant phenotype were strongly affected by the Mg²⁺ concentration, and mreBH mutants required higher levels of Mg²⁺ for viability than did the wild type (15). A similar dependence on Mg²⁺ concentration has been observed for mreB (55; see above), mbl (15), and mreCD (95; see below) mutant cells and for several other B. subtilis mutants thought to be affected in different aspects of CW synthesis and structure, as mentioned before (94, 121, 138, 139). Strikingly, MreBH was found to specifically interact with LytE (previously thought to be a cell septum-specific autolysin [72, 127]) in a genome-wide two-hybrid screen, and the Mg²⁺ dependence and shape defetcts of lytE mutants appeared remarkably similar to those of mreBH mutants (15). MreBH, like Mbl and MreB, forms dynamic helical filaments (see above and Fig. 1F) (15, 26). A functional LytE-GFP fusion localized to ongoing division sites and to cell separation sites, as previously observed by IFM (178), but surprisingly, it also localized as punctate fluorescence over the external lateral wall (15). Targeting of LytE to the sidewall of the cell was dependent on MreBH (and not on MreB or Mbl), while targeting of LytE to the sites of division was dependent on early (FtsZ) and late (PBP 2B) division proteins (15).

On the basis of the similar phenotypes, the direct protein-protein interaction in vivo and in vitro, and the MreBH-dependent localization of LytE along the lateral wall, it was concluded that MreBH influences lateral wall maturation by directing the localization of LytE (15). In light of these findings, together with the colocalization of all three MreB isoforms, an MreBH-dependent helical mode of cell wall hydrolysis that is coordinated with an Mbl-dependent helical mode of cell wall insertion has been suggested to control elongation of the rod-shaped B. subtilis cells during growth (15).

In summary, it is currently believed that the MreB helical structures are spatial regulators of cell wall biogenesis. How MreB-like proteins direct the insertion and maturation of CW material remains to be elucidated, but the current model is that they target/position cell wall enzymes (synthases and hydrolases) and/or membrane-associated and extracellular proteins, such as MreC and MreD (see below), that are involved in morphogenesis. It is interesting that the same genomic arrangement of the mre gene cluster (mreBCD) upstream of the mrd gene cluster (pbpA and rodA, both of which are associated with lateral wall elongation), transcribed colinearly with the direction of replication, is found in C. crescentus (52), R. sphaeroides (156), S. coelicolor (9), Magnetospirillum magneticum AMB-1, and Magnetospirillum magnetotacticum MS-1 (NCBI and GenBank accession no. AP007255 and AAAP00000000 [unfinished sequence], respectively), suggesting a functional interaction between the encoded proteins and further supporting the concerted action of the MreB cytoskeleton and PG synthesis. A complex consisting of MreB, MreC, MreD, PBP2, and RodA has been proposed to function in the extension of the lateral CW (Fig. 3F and see below).

Although biochemical evidence that PBPs are present in a complex is more and more compelling (2, 52, 148, 168) and a complex between a PG polymerase and a PG hydrolase was isolated in the presence of a scaffolding protein in E. coli (168), the model of Höltje has yet to be demonstrated conclusively. Nevertheless, it is extremely attractive to suppose that such multienzyme complexes exist and that they function as organized PG-synthesizing factories that are spatially controlled by the MreB helical scaffold to generate a functional CW architecture, and thereby the final three-dimensional structure of the cell. Several PBPs have been shown to display a putative helical distribution over the lateral cell wall, but the determinants of their localization remain unclear (28, 35, 146). However, the lateral (presumably helical) localization of one PG synthase (PBP2) and one PG hydrolase (LytE), in C. crescentus (40) and B. subtilis (15), respectively, was recently shown to be directly dependent upon MreB-like helices. These are the first (and almost certainly not the last) specific effectors of cell wall morphogenesis shown to be controlled by an MreB homologue in bacteria.

Thus, the MreC and MreD proteins (with as yet unknown biochemical functions) were excellent candidates for proteins that interact with MreB (and/or Mbl). In a model in which the membrane-associated MreBCD complex directs lateral cell wall synthesis, in a process essential to maintain cylindrical elongation of rod-shaped cells, MreC/MreD could anchor the cytoplasmic MreB filaments to the membrane and couple them to the extracellular cell wall synthetic machinery (14, 88, 95). Using a bacterial two-hybrid system, it was found that E. coli MreC interacted with both MreB and MreD and with itself. MreB also self-interacted, consistent with its polymerization into filaments (88). These results suggested that the mre-encoded proteins might form a multiprotein higher-order complex in which MreC interacts with both MreB (and/or Mbl in B. subtilis [95]) and MreD (88). Further supporting this suggestion, MreB localization is perturbed in cells depleted for either MreC or MreD in both E. coli (88) and B. subtilis (25). In light of these and other findings, a model of the MreBCD complex and how it communicates with enzymes in the periplasm to direct lateral CW synthesis was proposed and is shown in Fig. 3F (88).

In C. crescentus, MreB helices and rings were still observed in MreC-depleted cells (40), and the helical or banded patterns adopted by MreC along the cell length were not affected by rapid disruption of MreB upon A22 treatment (35, 40). The mechanism by which MreC localizes helically remains unclear. MreC did not colocalize with MreB, suggesting that they form independent helical structures in Caulobacter (40). However, they anticolocalized, i.e., when both proteins formed helices, they interdigitated, but when MreB was a ring, MreC was absent from this site, and this requires there to be some communication between the two (40). PBP2 has been shown to partially colocalize with MreC (40), and biochemical evidence has been provided for a direct interaction between the two (35). In light of these findings, it was proposed that MreC promotes helical PBP2 localization along the lateral wall, i.e., lateral wall synthesis (40). Consistently, IFM studies of R. sphaeroides showed that MreC colocalized with PBP2 throughout the cell cycle (while MreB colocalized with PBP2 only at certain stages of the cell cycle [see above]), again suggesting that MreC and PBP2 function in concert in PG synthesis during elongation (155). Interestingly, several PBPs (including PBP2 and PBP1) and outer membrane proteins were isolated from C. crescentus cell extracts by affinity chromatography using Sepharose-bound MreC (35). Undoubtedly, elucidation of the biochemical functions of the MreC and MreD proteins remains a major question for future research.

View larger version (53K): [in this window] [in a new window]   FIG. 4. Cytoskeleton structures organized by actin-like proteins in Spiroplasma and Magnetospirillum cells. (A to E) Cryo-electron tomography reveals the cytoskeleton structure of Spiroplasma melliferum. (A) Volume-rendering representation of a whole cell of S. melliferum with a tomogram slice cutting through it longitudinally. (B) 3D simulation of an S. melliferum cell indicating the localization of the three parallel ribbons of the cytoskeleton (red, purple, and green), which are anchored to the cell membrane (blue) and span the length of the cell. (Panels A and B were reprinted from reference 92 with permission from Elsevier.) (C to E) Superimposed slices of a tomogram (z sections are indicated by a bounding box) (C) and two corresponding 3D visualizations (D and E) of part of an S. melliferum cell showing the arrangement of the three ribbons. (C) The cytoskeleton comprises two outer ribbons of thicker filaments and a region of thinner filaments sandwiched between them. (D) Simplified 3D representation of the filament ribbons (green, purple, and red) that wind in parallel helically around the cell just underneath the cell membrane (blue). (E) Idealized visualization of the filaments, with a smooth transition to the original data (yellow). (Reprinted from reference 91 with permission of the publisher. Copyright 2005 American Association for the Advancement of Science.) (F to I) MamK, a homologue of MreB, organizes the magnetosome chain and forms filaments in vivo in Magnetospirillum magneticum sp. strain AMB-1. (F and G) 3D reconstructions of a wild-type AMB-1 cell (F) and a mamK mutant cell (G). The cell membrane (gray), magnetosome membrane (yellow), magnetite (orange), and magnetosome-associated filaments (green) are rendered. (H) MamK fused to GFP (green) forms filaments that localize to the inner curvature of the cell (cell membrane is stained red with FM4-64). (I) Phylogenetic relationship between MamK and other bacterial actins, demonstrated by an unrooted tree. These proteins separate into three distinct groups, i.e., MamK (green), ParM/StbA (red), and MreB (blue). (Reprinted from reference 84 with permission of the publisher. Copyright 2006 American Association for the Advancement of Science.)

Top Previous Next References

ParM. Immunofluorescence microscopy revealed that ParM formed highly dynamic filaments along the longitudinal axis of the cell (Fig. 5A), which were essential for the DNA partitioning process and displayed the properties expected for a force-generating or -directing cytoskeletal element (120). ParM (also called StbA) was one of the four bacterial proteins (together with Hsp70, FtsA, and MreB) reported to belong to the actin superfamily by Bork et al. (7; see above). Like the case for MreB, structural homology was uncovered behind the functional homology, and ParM was shown to have an atomic structure closely related to that of eukaryotic actin (and MreB) (Fig. 5C) and to undergo ATP-dependent polymerization/depolymerization into double-helical filaments similar to actin filaments (Fig. 5B) (helical repeat sizes are as follows: for F-actin, 55 Å; for MreB, 51 Å; and for ParM, 49 Å) (165). Importantly, the structure of ParM was solved in two states, i.e., in the absence of nucleotide and with a bound nucleotide (ADP) (165). The transition between these two states (unbound and ADP bound, called closed and open conformations, respectively) was shown to involve a conformational change in which the two major domains (I and II) undergo a rigid twist of 25° with respect to each other upon nucleotide binding. This finding was extremely significant because actin (and any actin-like molecule) subunits are expected to undergo a similar domain rotation upon nucleotide binding and/or hydrolysis, but so far only the closed state (relative to the ParM conformations) of G-actin (and MreB) has been trapped in crystals and observed at a high resolution.

Model for in vivo function of ParM filaments. Cytological and biochemical studies revealed that segregating plasmids localize to the ends of the dynamic ParM filaments and that ParM interacts specifically with the plasmid DNA-bound protein ParR in an ATP-dependent manner (119). It was proposed that ParM nucleates via the interaction with the parC-ParR complex at midcell and that the subsequent filament polymerization provides the mechanical force to propel the associated molecules apart before cell division occurs (120). No other factors are known to be required for the partitioning process, making the par system the simplest and best-understood mechanism for DNA segregation in bacteria. Recently, Garner et al. (57) used total internal reflection fluorescence microscopy and fluorescence resonance energy transfer to provide spectacular evidence that the actin-like ParM filaments display dynamic instability and bidirectional polymerization (rather than polarized polymerization, like in F-actin [131, 132] and MreB [82; see above]). Dynamic instability (i.e., catastrophic decay) has usually been associated with microtubules (33) rather than with actin filaments and consists of periods of steady polymerization (elongation) followed by rapid disassembly (catastrophe) which are regulated by nucleotide hydrolysis. Interestingly, regulation of ParM filaments by dynamic instability appears to be an important component of the plasmid-DNA segregation process. These striking findings led to the current model, which is shown in Fig. 5D. According to this model, ParM spontaneously nucleates and polymerizes in the cell, and the filaments spontaneously depolymerize unless they are stabilized by interaction with ParR-parC-paired plasmids. Only ParM filaments with plasmid bound at both ends are stabilized against catastrophic disassembly, and bidirectional polymerization at the ParM-ParR interface drives plasmid segregation (57). Such an insertional polymerization merchanism has been proposed for elongating MT ends attached to kinetochores (8, 71, 116) and for various examples of actin-based motility (21, 34, 90, 128). Also, as mentioned above, MTs, but not F-actin, display dynamic instability. Thus, the function of the par system actually seems to be the direct equivalent of mitosis in eukaryotes. Do the bacterial actin-like ParM filaments use a force-generating strategy similar to that of the eukaryotic microtubule-based mitotic spindle? This apparent inversion of function is also extended to FtsZ, the tubulin homologue, which drives cytokinesis in bacteria, in a reversal of the actin-based contractile ring in eukaryotic cells. Is there a general inversion of actin and tubulin functions in the prokaryotic and eukaryotic lineages (59)? These findings are raising important questions regarding the evolution of the cytoskeleton. Furthermore, the identification of the ParM-dependent plasmid segregation mechanism raised the intriguing possibility of a similar, MreB-dependent, mitotic-like chromosome segregation mechanism in bacteria.