SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Filament structures of Alps
  5. Supramolecular assemblies
  6. Conclusions
  7. Acknowledgements
  8. References

Cells rely on extensive networks of protein fibres to help maintain their proper form and function. For species ranging from bacteria to humans, this ‘cytoskeleton’ is integrally involved in diverse processes including movement, DNA segregation, cell division and transport of molecular cargoes. The most abundant cytoskeletal filament-forming protein, F-actin, is remarkably well conserved across eukaryotic species. From yeast to human – an evolutionary distance of over one billion years – only about 10% of residues in actin have changed and the filament structure has been highly conserved. Surprisingly, recent structural data show this to be not the case for filamentous bacterial actins, which exhibit highly divergent helical symmetries in conjunction with structural plasticity or polymorphism, and dynamic properties that may make them uniquely suited for the specific cellular processes in which they participate. Bacterial actin filaments often organize themselves into complex structures within the prokaryotic cell, driven by molecular crowding and cation association, to form bundles (ParM) or interwoven sheets (MreB). The formation of supramolecular structures is essential for bacterial cytoskeleton function. We discuss the underlying physical principles that lead to complex structure formation and the implications these have on the physiological functions of cytoskeletal proteins.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Filament structures of Alps
  5. Supramolecular assemblies
  6. Conclusions
  7. Acknowledgements
  8. References

Twenty years ago the cytoskeleton was thought to have only evolved in eukaryotes because relatives of actin or tubulin had not been found in bacteria. In 1992 Bork et al., using structure-based sequence alignment, observed that bacteria had genes related to actin (Bork et al., 1992). Yet it took almost another 10 years before the Löwe group demonstrated at the molecular level that the cell shape-determining protein MreB assembled into actin-like filaments (Van den Ent et al., 2001). Earlier the same group had shown that the crystal structure of bacterial cell division protein FtsZ was very similar to the atomic structure of tubulin, despite the low sequence homology (Löwe and Amos, 1998). Subsequently, many actin and microtubule related proteins have been discovered in bacteria (Becker et al., 2006; Anand et al., 2008). Like their eukaryotic counterparts, bacterial cytoskeletal filamentous proteins are key regulators and central organizers of many cellular processes including morphogenesis, cell division, DNA segregation and movement (Bugge-Jensen and Gerdes, 1999; Löwe et al., 2004).

Recently, substantial progress has been made in unravelling the structural characteristics and dynamic properties of filamentous systems formed from various bacterial actin-like proteins (Alps) (Møller-Jensen et al., 2002; Garner et al., 2004; 2007; Popp et al., 2008a; 2010a,b; Galkin et al., 2009; Polka et al., 2009). From these investigations it has become evident that nature fashioned many functional polymer designs during evolution. Additionally, it has become apparent that the bacterial cell exploits intrinsic physical properties such as molecular crowding and cation association to form polymeric suprastructures that are essential to the specific task undertaken by each cytoskeletal protein. In this review we give an overview of these diverse newly discovered polymer structures of prokaryotic actin-like cytoskeletal proteins and consider the implications for the evolution of actin filament systems. We also discuss the principles leading to complex structure formation and the ramifications that these have for the physiological roles of the cytoskeletal proteins.

Filament structures of Alps

  1. Top of page
  2. Summary
  3. Introduction
  4. Filament structures of Alps
  5. Supramolecular assemblies
  6. Conclusions
  7. Acknowledgements
  8. References

Recently, the monomeric and filamentous structures of Alps from various bacteria have been intensively studied. These include MreB, AlfA, R1-ParM and pSK41-ParM (Popp et al., 2008a; 2010a,b; Galkin et al., 2009; Polka et al., 2009). All Alps belong to the actin/Hsp70 superfamily of proteins that are structurally composed of two domains, termed domains I and II, which are further subdivided into subdomains Ia, Ib and IIa, IIb (Kabsch et al., 1990) (Fig. 1). Characteristic of the actin/Hsp70 superfamily is the presence of two similar and conserved structural motifs in subdomains Ia and IIa, which comprise a four- or five-stranded β-sheet surrounded by three α-helices. It has been postulated that this ‘actin fold’, which is vital for ATP or GTP binding and hydrolysis (Fig. 1), evolved through gene duplication of this α/β motif. Acquisition of additional domains served to modulate protein function allowing members of this superfamily to play many diverse roles from filament formation, glycolysis and chaperone activity to DNA segregation (Kabsch and Holmes, 1995). Although the atomic structures of members of the actin family are very similar the overall sequence identity within the entire actin family is in the range of only 20% (Fig. 2) (Kabsch and Holmes, 1995; Löwe et al., 2004; Popp et al., 2010b). In contrast, the most abundant filament-forming protein F-actin is remarkably well conserved across eukaryotic species. Indeed, actin can be found in all branches of eukaryotes sharing at least 70% amino acid identity with human β-actin (Fig. 2). From yeast to human, an evolutionary distance of over one billion years the sequence identity has been 90% conserved and the filament structure has been highly conserved. Particularly, muscle α-actin has been completely conserved in sequence and structure, since the emergence of early animals some 600 million years ago.

image

Figure 1. The atomic structures of different Alp monomers are similar despite the low sequence homology. The green circle highlights the nucleotide-binding site. A. Comparison between psK41-ParM and Ta0583 an actin from Archaea. B. Comparison between psK41-ParM and R1-ParM.

Download figure to PowerPoint

image

Figure 2. Phylogram depicting the relatedness of actin (red) from species from each branch of the eukaryote tree with the bacterial Alps (blue) discussed in this review. Several potential filament-forming bacterial Alps (Alp10, Alp14 and Alp21, blue) and a viral Alp (Alp6, green) have also been included in the analysis. The eukaryotic actins share > 70% identity whereas the Alps share ∼20% identity. Multiple Alps can be identified in the same bacterium, whereas some bacteria have no identifiable Alp and several Alps appear to be specific to a particular bacterial species. ParM and MreB show more widespread distributions, although they are rarely found together in the same bacterium.

Download figure to PowerPoint

Therefore, it came as a surprise that a high amount of structural variability is observed among bacterial actin filament systems (Popp et al., 2008a; 2010a,b; Galkin et al., 2009; Polka et al., 2009). F-actin, as well as most Alp filaments studied in detail to date, consists of two strands of protomers twisting around each other to form a double helix. The only exception being the cell shape-determining protein MreB that forms both single-stranded helical and linear filaments depending upon the nucleotide state (Van den Ent et al., 2001; Popp et al., 2010c) (Fig. 3). Despite this preference for helical geometry, architectural constructions are diverse.

image

Figure 3. Schematic diagram illustrating the variations in filament organization and helical parameters. ATP and GTP indicate the fuel type used to originally generate the filament. After hydrolysis and phosphate release, aged filaments will have ADP or GDP bound nucleotides.

Download figure to PowerPoint

Simplistically, the symmetry of a helical structure can be defined in terms of two parameters, the subunit repeat (h) and the repeat of the helix (C). X-ray diffraction patterns from oriented gels showed F-actin to be a right handed 13/6 helix (13 subunits in 6 turns) with h = 27.5 Å and C ∼ 360 Å (Holmes et al., 1990) (Fig. 3). In electron micrographs, the DNA segregation protein AlfA from Bacillus subtilis mainly formed 7/3 helices with h ∼ 25 Å and C ∼ 170 Å (Popp et al., 2010a). DNA segregation protein pSK41-ParM from Staphylococcus aureus formed filaments with a 10/4 helical symmetry with h ∼ 25 Å and C ∼ 250 Å (Fig. 3) (Popp et al., 2010b). Escherichia coli R1-ParM was observed to have a similar symmetry as F-actin, yet interestingly the helix was left-handed opposed to the right-handed F-actin helix (Popp et al., 2008a; Galkin et al., 2009) (Fig. 3). AlfA was also identified to form a left-handed helix (Polka et al., 2009) (Fig. 3). Left-handedness of bacterial segregation proteins seems to be important for the interaction with the adaptor protein, whose geometry is dictated by the fixed handedness of DNA (Møller-Jensen et al., 2007). F-actins right-handedness may have been implemented by the emergence of processive myosin motors, which preferentially ‘walk’ on this geometry of actin helix (Walker et al., 2000) or by actin sequestering proteins and capping proteins like gelsolin (Burtnick et al., 2004).

The canonical helices of AlfA, R1-ParM, pSK41-ParM and F-actin showed some variations depending on nucleotide association, ionic strength or pH (Egelman et al., 1982; Oda et al., 1998; Galkin et al., 2009; Popp et al., 2010a,b). Thus, for many of these polymers there is no single structure. Rather, these polymers may use a multiplicity of states to explicate various functions at different stages of a cell's life cycle (Egelman, 2008). The molecular intra-strand and inter-strand filament contacts have been determined for a number of these filaments (Holmes et al., 1990; Popp et al., 2008a; 2010a,b; Galkin et al., 2009; Polka et al., 2009). Interestingly, the dominant intra-strand contacts of the different Alps and F-actin proved to be rather similar, whereas the inter-strand contacts differ. For R1-ParM, two salt bridges were found to be the key inter-filament contacts that stabilize the left-handed double helix (Popp et al., 2008a). Point mutations of these residues completely abrogated R1-ParM helical filament formation (Popp et al., 2009). Yet similar mutagenesis studies on pSK41-ParM a ParM homologue from S. aureus revealed that the inter-strand filament contacts were entirely different to those of R1-ParM (Popp et al., 2010b). From these observations we can draw several conclusions concerning the evolution of actins. Two construction principles seem to have survived that are essential for the function of actin-like polymers. First, the actin fold ensures nucleotide binding and subsequent hydrolysis, providing an energy source and a conformational switch (Kabsch and Holmes, 1995). The switch can be triggered by either nucleotide binding for Alps or ions binding with F-actin. Upon nucleotide binding, domains I and II close the interdomain cleft of ParM-R1 by ∼25° leading to rapid filament formation (Van den Ent et al., 2001; Popp et al., 2008a). Addition of monovalent or divalent ions triggers the polymerization of G-actin, through a rotation of domains I and II by ∼20°, into stable F-actin filaments (Oda et al., 2009). One other fundamental difference should be noted: Alp monomers are stable without bound nucleotide, whereas G-actin readily denatures with loss of nucleotide.

Second, the intra-strand filament contacts are conserved indicating that filaments may have been originally built as linear polymerizing machines. The inter-strand filament contacts are not preserved, emphasizing that the Alp polymerizing machineries have diverged to produce a variety of filament geometries with diverse properties that are tailored to specific biological processes, for example, AlfA in the complicated process of chromosome segregation during bacterial sporolation (Becker et al., 2006). Using the analogy of cars, as long as the essential parts, such as the engine and wheels, are provided the car is functional, yet the overall design of the car can vary largely due to fashion or purpose. One of the earliest polymers working as a linear polymerizing machine may have been bacterial cell shape-determining protein MreB, as it operates as a single-stranded helical or linear protofilament (Van den Ent et al., 2001; Daniel and Errington, 2003; Popp et al., 2010c). Double helical filaments posses a more sophisticated design and may have evolved later. Here, two protofilament strands gently wind around each other giving the filament more mechanical resilience. In most bacteria a wide variety of different filament systems coexist that perform different tasks (Carballido-Lopez and Errington, 2003). Polymers with inherently different symmetries will avoid confusion between the discrete assemblies, suggesting that a variety of filament designs may have coevolved.

Another surprising observation was that, although all studied bacterial Alps contain the actin fold typical for an ATP-binding protein (Kabsch and Holmes, 1995), these proteins also bind GTP with almost the same or even higher efficiency than to ATP, leading to the induction of polymerization with similar or higher rates (Bean and Amann, 2008; Popp et al., 2008a; 2010a,b,c; Polka et al., 2009). The polymerization rates of Alps were found to be up to a hundred times faster than F-actin and the kinetic schemes needed for the interpretation of the polymerization curves did not necessarily need any nucleation step as in the case of F-actin (Popp et al., 2008a; 2010a). In the bacterial cell the ATP : GTP ratio is about 3:1 mM (Jewett et al., 2009) and it is highly likely that filaments consisting of both ATP and GTP bound subunits coexist. This may indicate that in the early stages of actin evolution the selectivity for the ‘fuel type’ was not pronounced.

What triggered the selected use of ATP in F-actin assembly remains unclear. One might speculate this may be a consequence of emerging actin-signalling proteins, which are mostly GTPases and started to dominate assembly and disassembly of the actin cytoskeleton in eukaryotic cells thus avoiding competition for nucleotide. One problem with this argument is that deep branching eukaryotes like protozoa utilize ATP for F-actin polymerization and dynamics, yet mostly lack the GTPases used in signalling.

In eukaryotic systems, microtubules and F-actin filaments are known to exhibit unique dynamics. F-actin filaments treadmill (Neuhaus et al., 1983), whereas, microtubule filaments (which are GTPases) show dynamic instability that causes the filament to suddenly collapse (Mitchison and Kirschner, 1984). Of the filament bacterial systems studied so far, only R1-ParM displays microtubule-like dynamic instability (Popp et al., 2008a), which was linked to its operation as a GTP-dependant molecular switch similar to a G-protein. In contrast, pSK41-ParM, AlfA and MreB appear to treadmill (Kim et al., 2006; Polka et al., 2009; Popp et al., 2010a,b,c). These dynamics are likely tailored to the specific task of each filament system. Recently, a wide variety of new Alp's have been identified in various bacteria from sequence analysis (Derman et al., 2009) (Fig. 2) and their characterization is expected to yield a new wealth of filament geometries and dynamic properties within the bacterial actin family.

Supramolecular assemblies

  1. Top of page
  2. Summary
  3. Introduction
  4. Filament structures of Alps
  5. Supramolecular assemblies
  6. Conclusions
  7. Acknowledgements
  8. References

Light microscopy observations of fluorescently labelled cytoskeletal proteins within bacterial cells had indicated that their structures were not single filaments but rather consisted of more complex arrangement of multiple filaments (Carballido-Lopez and Errington, 2003). Bundles formed by DNA segregation protein R1-ParM (Gerdes et al., 2010) and cable-like structures of bacterial cell shape protein MreB were observed (Jones et al., 2001). The bacterial cytoskeleton is difficult to visualize at high resolution by electron tomography in vivo (Li et al., 2007), as a result of the small number of filamentous structures that are easily lost in the crowded granular cytoplasm of the cell. Therefore, important questions remained unanswered: What are the structures of the filamentous assemblies? How do they arise? And how may they function? In attempting to provide insight to these questions we should take a look what may influence the structures, functions and interactions between filamentous macromolecules within the cell.

First, ions and water molecules screen charged molecules from each other within a cell. However, the biological polymers that constitute the cytoskeleton associate with counterions that can result in strong electrostatic interactions between the like-charged filaments (Fig. 4) (Wong and Pollack, 2010). In addition, the interior of a cell is about 40% comprised of various macromolecules, termed molecular crowding (Minton, 2000; 2005). Molecular crowding has been known for almost half a century to have profound effects on the thermodynamics and kinetics of cellular processes and can enhance the native state of proteins (Minton, 2000; 2005). Because of the excluded volume effect, molecular crowding leads to short range forces between cytoskeletal proteins and shifts the equilibrium from single filaments to more complex supramolecular structures (Popp et al., 2008b) (Fig. 4).

image

Figure 4. Molecular crowding and cations have profound effects on filamentous proteins. A. In aqueous solutions filaments (light blue) are randomly oriented. B. In a crowded environment, represented by various coloured symbols, filaments will organize into suprastructures (here a bundle) because of osmotic pressure (red arrows). C, D. Filaments (blue) are highly negatively charged (red symbol). C. In low ionic strength solutions, this charge keeps the filaments apart. D. Cations (shown are light blue donuts of Mg2+ ions) can associate tightly to the filaments, leading to an attractive force.

Download figure to PowerPoint

F-actin and other actin-like filaments in aqueous buffers are highly negatively charged with a linear charge density of about 4–5 e nm−1 (Tang et al., 1997; Popp et al., 2010d). This strong overall negative surface charge may be expected to make the interaction between filaments repulsive. Yet, it has long been known that positively charged counterions can be trapped in the vicinity of a negatively charged filament. Oosawa first proposed that counterion fluctuations lead to the formation of supramolecular association of long charged biopolymers (Oosawa, 1971) and the theory was later expanded by Manning (Manning and Ray, 1998) (Fig. 4). Therefore, we expect two physical principles, molecular crowding and cation condensation, lead to the formation of filamentous suprastructures. Lateral interactions as a result of the osmotic pressure generated from molecular crowding will be predominant, and this pressure increases with ionic strength (Tang et al., 1997) [∼350 mM KCl in the bacterial cytoplasm (Cayley et al., 1991)]. Molecular crowding induced association but will be modulated by counterion fluctuations that are favoured under low-salt conditions.

The complex eukaryotic cytoskeleton is tightly regulated by a large number of actin- or microtubule-associated proteins. In contrast, bacterial actins and microtubules have few associated regulatory proteins. This simplifies the quest for observing the basic physical principles that govern the formation of filamentous suprastructures.

Suprastructures of the bacterial actin cytoskeleton have been recently directly visualized by high resolution in vitro electron microscopy. We review two examples:

R1-ParM

In cryo-sectioned E. coli cells, R1-ParM filaments were recently observed to consist of small filament bundles rather than individual filaments (Salje et al., 2009), although the precise molecular organization could not be determined. Instead, 2-D and 3-D bundles of R1-ParM could be formed in vitro using crowding agents that mimic the physiological conditions within bacterial cells. Electron micrographs of in vitro R1-ParM rafts, which are 2-D analogues of 3-D bundles, were used to identify the main molecular inter-filament contacts within these suprastructures (Popp et al., 2010d). Surprisingly, the interfaces between filaments were similar for both parallel and anti-parallel orientations suggesting that the distribution of filament polarity is random within a bundle consistent with previous dual-colour TIRF observations (Popp et al., 2007) (Fig. 5). Furthermore, the inter-filament interactions were not due to the interactions of specific residues but rather to long-range, counter ion-mediated, electrostatic attractive forces between large patches of residues (Popp et al., 2010d). This bundle design has two advantages when bidirectionally segregating large DNA in the prokaryotic cell. The randomly oriented nature of the bundle allows DNA to be captured with equal efficiency at both ends of the bundle. Second the bundling of filaments greatly increases stiffness allowing the system to maneuver relatively large payloads, DNA plasmids (Fig. 6).

image

Figure 5. Schematic model of typical randomly oriented R1-ParM bundles. Three filaments with their pointed ends (p) up are shown in yellow one filament with the barbed end (b) up is shown in blue. Parallel filaments within the bundle (filaments 1,2) share similar large areas of molecular interaction (illustrated as red and green patches) as filaments arranged anti-parallel (filaments 2,3 or 3,4).

Download figure to PowerPoint

image

Figure 6. Stiffness and elongation of R1-ParM bundles. A. The force needed to bend a beam is proportion to the cube of the radius. B. A small hexagonal bundle consisting of seven filaments will have three times the radius but will be ∼80 times stiffer than a single filament. C. The bundle nucleus forms at the onset of polymerization and consists of short filaments that rapidly elongate.

Download figure to PowerPoint

The molecular mechanism of how these bundles form has been revealed by real-time TIRF microscopy (Popp et al., 2010d). Bundle formation of R1-ParM was mainly a result of molecular crowding. At the onset of polymerization, bundle thickness and shape were determined in the form of nuclei of short filaments arranged in a liquid-like lattice (Fig. 6C). These nuclei then underwent an elongation phase whereby they rapidly increased in length. Interestingly, R1-ParM filaments in newly formed bundles, within several minutes after initiating polymerization, were not well ordered as observed by electron microscopy (Popp et al., 2010d), whereas filaments in matured bundles adopted a defined canonical helical structure (Fig. 7). Young filaments appear to have less stable inter-strand contacts compared with aged filaments (Fig. 7). This contradicts the general wisdom learned from kinetic studies (Popp et al., 2008a) that suggests that old filaments with bound ADP or GDP subject to dynamic instability should show more structural variability than newly formed ATP- or GTP-bound filaments.

image

Figure 7. Structural plasticity. A. Juvenile filaments of R1-ParM are helically disordered indicating, that inter-strand contacts were labile (yellow wave). Green arrows indicate the subunit variability. B. Matured filaments develop the canonical helical parameters, suggesting the inter-strand contacts are now strong (yellow rectangle). In the Alp filaments only the intra-strand contacts (red rectangles) are preserved, suggesting filaments were originally designed as linear polymerizing motors.

Download figure to PowerPoint

R1-ParM segregates DNA within a bacterial cell of several µm in length in less than 1 min (Bugge-Jensen and Gerdes, 1999; Popp et al., 2007). Hence, during DNA segregation the longitudinal intra-strand contacts will dominate over inter-strand contacts, which take a substantially longer time to develop in a defined way. This would argue that a well-defined canonical helical symmetry of R1-ParM is not necessary for the task of DNA segregation, rather that its function as a linear motor is essential. A similar phenomenon was previously reported for F-actin, where juvenile filaments exhibit substantial disorder, whereas, aged filaments showed the typical features of the classical 13/6 canonical helical symmetry (Kueh and Mitchison, 2009). In the case of actin (and also for microtubules) this led to the hypothesis that multiple structural states of the filaments exist that are not necessarily coupled to a defined chemical state of bound nucleotide. This concept was termed structural plasticity and was seen a new important feature for influencing and regulating the dynamics of the eukaryotic cytoskeleton at the leading edge of the cell or for segregating chromosomes (Kueh and Mitchison, 2009). Bacteria had earlier evolved and tested structural plasticity for its Alp polymers. Dynamic microtubule-like instability has also been probed in bacteria, where R1-ParM is a good example. After segregating the DNA, filaments will switch from polymerization to depolymerization mode, leading to complete disintegration of filaments and equal distribution of monomers before cell division.

Other bacterial actin-like DNA segregation proteins, AlfA and pSK41-ParM, spontaneously form bundles, sheets or nets even in the absence of molecular crowding (Popp et al., 2010a,b). Single filaments were virtually non-existent. The formation of these suprastructures can be attributed to counterions, especially the tight condensation of Mg2+ and the abundant polycation, spermine, to these highly negatively charged filaments (Tabor and Tabor, 1976; 1985). Unlike R1-ParM, these filament systems did not display any obvious sign of structural plasticity, yet polymorphism in the form of variations of helical parameters was observed (Popp et al., 2010a,b).

MreB

In vivo fluorescence microscopy studies have shown that the bacterial shape-determining protein and actin homologue, MreB, forms cable-like structures that spiral around the periphery of the cell (Jones et al., 2001). Until recently the molecular structures of these cables have not been extensively studied. Surprisingly, MreB from Thermatoga maritima appeared in vitro to consist of complex, several µm long multilayered sheets of interwoven filaments in the presence of either ATP or GTP (Popp et al., 2010c) (Fig. 8A). The crystalline order was highest in the presence of Mg2+ and GTP. This architecture, in agreement with recent rheological measurements on MreB cables (Esue et al., 2006), has superior mechanical properties compared with a single filament or a sheet with filaments aligned in parallel and could be an important feature for maintaining bacterial cell shape. Protofilaments formed in the presence of ATP or GTP were helical 3/1 filaments with a repeat of ∼200 Å (Fig. 8B). Surprisingly, in the presence of ADP or GDP linear filaments were formed that were similar to those seen in MreB crystals, an environment that is dominated by the crystal packing (Van den Ent et al., 2001) (Fig. 8C). This may indicate that the binding of GTP or ATP to MreB monomers stabilizes a different conformation to that of ADP or GDP, which in turn leads to the preferential formation of either interwoven helical filaments or sheets consisting of linear protofilaments. It is unknown whether bacterial cells utilize ATP, GTP or both nucleotides to assemble MreB in vivo and whether the cell exploits different nucleotides for specific MreB-based functions. In nitrogen-fixing bacteria like T. maritima the rate of ATP regeneration is considered to be growth rate limiting because of the high ATP requirement for N2 fixation and the ATP/ADP ratio is estimated to be between 1.4 and 2.0 (Upchurch and Mortenson, 1980). In E. coli, during exponential growth, the ATP and GTP concentrations are reported to be relatively constant at about 3 mM and 0.9 mM respectively (Jewett et al., 2009). Assuming similar concentrations in T. maritima, the nucleotide concentrations can be estimated to be about 2 mM ATP, 1 mM ADP and 1 mM GTP. Should such nucleotide concentrations in T. maritima lead to interwoven or linear sheet structures or coexistence of both forms, depending on various stages in the bacterial cell cycle remains an open question.

image

Figure 8. Schematic diagram of the molecular arrangement of interwoven MreB filaments within cables. A. In the presence of ATP or GTP. B. MreB filaments within ATP or GTP cables are not linear but posses a helical 3/1 symmetry. C. Cables in the presence of ADP or GDP consist of linear protofilaments arranged side by side.

Download figure to PowerPoint

Conclusions

  1. Top of page
  2. Summary
  3. Introduction
  4. Filament structures of Alps
  5. Supramolecular assemblies
  6. Conclusions
  7. Acknowledgements
  8. References

Linear polymerizing motors in the form of dynamic biopolymers originally evolved in early single cellular organisms to perform various crucial cellular functions are evident in modern day bacteria. Structural plasticity, polymorphism, treadmilling and dynamic instability of cytoskeletal filaments were probed in the early stages of life as tools to explicate various functions and dynamics. Eukaryotes later developed a multitude of actin-interacting proteins to more precisely control the spatial and temporal formation of filament structures in adapting actin to a wider set of tasks. This compelled the central polymerizing motor to remain relatively unchanged in order to maintain this large number of interactions. In the absence of such conservative evolutionary pressure, bacterial Alp polymerizing proteins were free to probe a greater range of amino acid sequence space, resulting in a broad array of functional designs, often duplicating filament systems to accomplish the full array of bacterial cell functions. Interestingly helical actin filaments were found to be left- or right-handed, most likely depending on their interacting partners. For instance, the left-handed helical design of bacterial DNA segregation proteins is probably dictated by the fixed handedness of DNA. Whereas, the right-handed F-actin helix is likely to have been fixed to this specific geometry by interaction partners, such as motor, capping and nucleating proteins.

The fuel necessary for Alp filament elongation in bacteria can come from ATP or GTP, whereas eukaryotic actin uses only ATP. What caused the eukaryotic actin filament to become restricted to be an ATPase remains an unsolved question. Bacterial actins can show either microtubule-like dynamic instability or F-actin-like treadmilling, implying that these important mechanisms must have been developed early in evolution. Interestingly, actin-like and tubulin-like proteins seem to have exchanged roles during evolution because cytokinesis depends upon the constriction of the tubulin-like FtsZ ring in bacteria but by an actin ring in eukaryotic cells. Conversely, chromosomes are separated, in part, by microtubules in eukaryotes but can be moved apart by Alps in bacteria.

Molecular crowding and counterions are fundamental elements exploited by bacterial cells to obtain various 3D structures. These suprastructures often have superior mechanical properties compared with a single filament, which helps in moving large cargos and keeps cell walls from deforming. The lateral interactions within these suprastructures are in general not specific, giving these systems a great degree of freedom in its operations. Hence, bacterial filaments use principles of molecular crowding and counterions to achieve complex 3-D structures that in eukaryotes require complex arrays of associated proteins.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Filament structures of Alps
  5. Supramolecular assemblies
  6. Conclusions
  7. Acknowledgements
  8. References

The authors would like to thank the Biomedical Research Council of A*STAR, Singapore and DP the ERATO ‘Maeda actin filament dynamics project’, JST, Japan for support.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Filament structures of Alps
  5. Supramolecular assemblies
  6. Conclusions
  7. Acknowledgements
  8. References