Tracer diffusion through F-actin: effect of filament length and cross-linking.

We have determined diffusion coefficients for small (50- to 70-nm diameter) fluorescein-thiocarbamoyl-labeled Ficoll tracers through F-actin as a function of filament length and cross-linking. fx45 was used to regulate filament length and avidin/biotinylated actin or ABP-280 was used to prepare cross-linked actin gels. We found that tracer diffusion was generally independent of filament length in agreement with theoretical predictions for diffusion through solutions of rods. However, in some experiments diffusion was slower through short (< or = 1.0 micron) filaments, although this result was not consistently reproducible. Measured diffusion coefficients through unregulated F-actin and filaments of lengths > 1.0 micron were more rapid than predicted by theory for tracer diffusion through rigid, random networks, which was consistent with some degree of actin bundling. Avidin-induced cross-linking of biotinylated F-actin did not affect diffusion through unregulated F-actin, but in cases where diffusion was slower through short filaments this cross-linking method resulted in enhanced tracer diffusion rates indistinguishable from unregulated F-actin. This finding, in conjunction with increased turbidity of 1.0-micron filaments upon avidin cross-linking, indicated that this cross-linking method induces F-actin bundling. By contrast, ABP-280 cross-linking retarded diffusion through unregulated F-actin and decreased turbidity. Tracer diffusion under these conditions was well approximated by the diffusion theory. Both cross-linking procedures resulted in gel formation as determined by falling ball viscometry. These results demonstrate that network microscopic geometry is dependent on the cross-linking method, although both methods markedly increase F-actin macroscopic viscosity.     

The bundling of actin with polyethylene glycol 8000 in the presence and absence of gelsolin.

Actin filament and bundle formation occur in the cytosol under conditions of very high total macromolecular concentration. In this study we have utilized the inert molecule polyethylene glycol 8000 (PEG) as a means of simulating crowded conditions in vitro. Column-purified Ca-actin was polymerized in the absence and presence of gelsolin (to regulate mean filament lengths between 50 and 5000 mers) and PEG (2-8%) using various concentrations of KCl and/or 2 mM divalent cations. Bundling was characterized by the scattered light intensity and mean diffusion coefficients obtained from dynamic light scattering, as well as by fluorescence and phase-contrast microscopy. The minimum concentration of KCl required for bundling decreases both with increasing concentration of PEG at a fixed mean filament length, and with decreasing filament length at a fixed concentration of PEG. In the absence of divalent cation, bundling is reversible on dilution, as determined by intensity levels, diffusion coefficients, and microscopy. However, with either 2 mM Mg2+ or Ca2+ added, bundling is irreversible under conditions of higher PEG concentrations or longer filaments, indicating that osmotic pressure effects cannot fully explain actin bundling with PEG. Weaker divalent cation-binding sites on actin as well as disulfide bonds appear to be involved in the irreversible bundling.     

Energetics and dynamics of constrained actin filament bundling

The formation of filopodia-like bundles from a dendritic actin network has been observed to occur in vitro as a result of branching induced by Arp2/3 complex. We study both the energetics and dynamics of actin filament bundling in such a network to evaluate their relative importance in bundle formation processes. Our model considers two semiflexible actin filaments fixed at one end and free at the other, described using a normal-mode approximation. This model is studied by both Brownian dynamics and free-energy minimization methods. Remarkably, even short filaments can bundle at separations comparable to their lengths. In the dynamic simulations, we evaluate the time required for the filaments to interact and bind, and examine the dependence of this bundling time on the filament length, the distance between the filament bases, and the cross-linking energy. In most cases, bundling occurs in a second or less. Beyond a certain critical distance, we find that the bundling time increases very rapidly with increasing interfilament separation and/or decreasing filament length. For most of the cases we have studied, the energetics results for this critical distance are similar to those obtained from dynamics simulations run for 10 s, suggesting that beyond this timescale, energetics, rather than kinetic constraints, determine whether or not bundling occurs. Over a broad range of conditions, we find that the times required for bundling from a network are compatible with experimental observations.     

Multiscale study of counterion-induced attraction and bundle formation of F-actin using an Ising-like mean-field model

An Ising-like counterion-binding model is developed and solved by a mean-field method. For G-actin, the calculated affinity constants of all the binding sites ranging from loose to tight binding match the experimental data. The model is used to calculate the interaction energy between two F-actin filaments. Within a certain counterion concentration range, a rapidly decaying attractive force between two parallel filaments is produced not only by the correlation of the counterion distributions on the two filaments, but also by the correlation of the configurations of the two filaments with fixed counterion positions, which has been ignored in previous calculations. The bundling energy depends strongly on the configuration of the filaments. Upon bundling, the tightly bound counterion site is not affected, but the medium and loosely bound ones are. The model reproduces the observed minimal divalent counterion concentration for bundling, and naturally predicts the resolubilization of bundles which is seen in recent experiments. At the optimal counterion concentration, we obtain a bundling energy of approximately -0.01 eV per monomer along the filament. The counterion valence strongly affects the optimal counterion concentration, but has only minor effects on the optimal bundling energy. We show that the attractive potential between filaments can be simplified as the sum of interactions between their monomers. This simplification makes it possible to calculate the exact free energy of a two-F-actin-filament system. We are thus able to probe the effects of filament length on F-actin bundling and obtain a critical length for bundling of 59 monomers at 1 microM monomer concentration and pH=7.2.     

Analysis of rhodamine and fluorescein-labeled F-actin diffusion in vitro by fluorescence photobleaching recovery.

Properties of filamentous acetamidofluorescein-labeled actin and acetamidotetramethylrhodamine-labeled actin (AF and ATR-actin, respectively) were examined to resolve discrepancies in the reported translational diffusion coefficients of F-actin measured in vitro by FPR and other techniques. Using falling-ball viscometry and two independent versions of fluorescence photobleaching recovery (FPR), the present data indicate that several factors are responsible for these discrepancies. Gel filtration chromatography profoundly affects the viscosity of actin solutions and filament diffusion coefficients. ATR-actin and, to a lesser degree, AF-actin show a reduction in viscosity in proportion to the fraction labeled, presumably due to filament shortening. Actin filaments containing AF-actin or ATR-actin are susceptible to photoinduced damage, including a covalent cross-linking of actin protomers within filaments and an apparent cleavage of filaments detected by a decrease of the measured viscosity and an increase in the measured filament diffusion coefficients. Quantum yields of the two photoinduced effects are quite different. Multiple cross-links are produced relative to each photobleaching event, whereas less than 1% filament cleavage occurs. Substantial differences in the filament diffusion coefficients measured by FPR are also the result of differences in illumination geometry and sampling time. However, under controlled conditions, FPR can be used as a quantitative tool for measuring the hydrodynamic properties of actin filaments. Incremented filament shortening caused by photoinduced cleavage or incremental addition of filament capping proteins produces a continuous and approximately linear increase of filament diffusion coefficients, indicating that filaments are not associated in solution. Our results indicate that actin filaments exhibit low mobilities and it is inferred that actin filaments formed in vitro by column-purified actin, under standard conditions, are much longer than has conventionally been presumed.     

Interplay between Brownian motion and cross-linking controls bundling dynamics in actin networks

Morphology changes in cross-linked actin networks are important in cell motility, division, and cargo transport. Here, we study the transition from a weakly cross-linked network of actin filaments to a heavily cross-linked network of actin bundles through microscopic Brownian dynamics simulations. We show that this transition occurs in two stages: first, a composite bundle network of small and highly aligned bundles evolves from cross-linking of individual filaments and, second, small bundles coalesce into the clustered bundle state. We demonstrate that Brownian motion speeds up the first stage of this process at a faster rate than the second. We quantify the time to reach the composite bundle state and show that it strongly increases as the mesh size increases only when the concentration of cross-links is small and that it remains roughly constant if we decrease the relative ratio of cross-linkers as we increase the actin concentration. Finally, we examine the dependence of the bundling timescale on filament length, finding that shorter filaments bundle faster because they diffuse faster.     

Cell Shape Can Mediate the Spatial Organization of the Bacterial Cytoskeleton

The bacterial cytoskeleton guides the synthesis of cell wall and thus regulates cell shape. Because spatial patterning of the bacterial cytoskeleton is critical to the proper control of cell shape, it is important to ask how the cytoskeleton spatially self-organizes in the first place. In this work, we develop a quantitative model to account for the various spatial patterns adopted by bacterial cytoskeletal proteins, especially the orientation and length of cytoskeletal filaments such as FtsZ and MreB in rod-shaped cells. We show that the combined mechanical energy of membrane bending, membrane pinning, and filament bending of a membrane-attached cytoskeletal filament can be sufficient to prescribe orientation, e.g., circumferential for FtsZ or helical for MreB, with the accuracy of orientation increasing with the length of the cytoskeletal filament. Moreover, the mechanical energy can compete with the chemical energy of cytoskeletal polymerization to regulate filament length. Notably, we predict a conformational transition with increasing polymer length from smoothly curved to end-bent polymers. Finally, the mechanical energy also results in a mutual attraction among polymers on the same membrane, which could facilitate tight polymer spacing or bundling. The predictions of the model can be verified through genetic, microscopic, and microfluidic approaches.     

Actin filament bundling by fimbrin is important for endocytosis, cytokinesis, and polarization in fission yeast

Through the coordinated action of diverse actin-binding proteins, cells simultaneously assemble actin filaments with distinct architectures and dynamics to drive different processes. Actin filament cross-linking proteins organize filaments into higher order networks, although the requirement of cross-linking activity in cells has largely been assumed rather than directly tested. Fission yeast Schizosaccharomyces pombe assembles actin into three discrete structures: endocytic actin patches, polarizing actin cables, and the cytokinetic contractile ring. The fission yeast filament cross-linker fimbrin Fim1 primarily localizes to Arp2/3 complex-nucleated branched filaments of the actin patch and by a lesser amount to bundles of linear antiparallel filaments in the contractile ring. It is unclear whether Fim1 associates with bundles of parallel filaments in actin cables. We previously discovered that a principal role of Fim1 is to control localization of tropomyosin Cdc8, thereby facilitating cofilin-mediated filament turnover. Therefore, we hypothesized that the bundling ability of Fim1 is dispensable for actin patches but is important for the contractile ring and possibly actin cables. By directly visualizing actin filament assembly using total internal reflection fluorescence microscopy, we determined that Fim1 bundles filaments in both parallel and antiparallel orientations and efficiently bundles Arp2/3 complex-branched filaments in the absence but not the presence of actin capping protein. Examination of cells exclusively expressing a truncated version of Fim1 that can bind but not bundle actin filaments revealed that bundling activity of Fim1 is in fact important for all three actin structures. Therefore, fimbrin Fim1 has diverse roles as both a filament "gatekeeper" and as a filament cross-linker.     

Energetics and dynamics of global integrals modeling interaction between stiff filaments

The attractive and spacing interaction between pairs of filaments via cross-linkers, e.g. myosin oligomers connecting actin filaments, is modeled by global integral kernels for negative binding energies between two intersecting stiff and long rods in a (projected) two-dimensional situation, for simplicity. Whereas maxima of the global energy functional represent intersection angles of ‘minimal contact’ between the filaments, minima are approached for energy values tending to −∞, representing the two degenerate states of parallel and anti-parallel filament alignment. Standard differential equations of negative gradient flow for such energy functionals show convergence of solutions to one of these degenerate equilibria in finite time, thus called ‘super-stable’ states. By considering energy variations under virtual rotation or translation of one filament with respect to the other, integral kernels for the resulting local forces parallel and orthogonal to the filament are obtained. For the special modeling situation that these variations only activate ‘spring forces’ in direction of the cross-links, explicit formulas for total torque and translational forces are given and calculated for typical examples. Again, the two degenerate alignment states are locally ‘super-stable’ equilibria of the assumed over-damped dynamics, but also other stable states of orthogonal arrangement and different asymptotic behavior can occur. These phenomena become apparent if stochastic perturbations of the local force kernels are implemented as additive Gaussian noise induced by the cross-link binding process with appropriate scaling. Then global filament dynamics is described by a certain type of degenerate stochastic differential equations yielding asymptotic stationary processes around the alignment states, which have generalized, namely bimodal Gaussian distributions. Moreover, stochastic simulations reveal characteristic sliding behavior as it is observed for myosin-mediated interaction between actin filaments. Finally, the forgoing explicit and asymptotic analysis as well as numerical simulations are extended to the more realistic modeling situation with filaments of finite length.

---

Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.      

Structure and mobility of actin filaments as measured by quasielastic light scattering, viscometry, and electron microscopy

Actin filaments of different lengths were prepared by polymerizing actin in the presence of various concentrations of gelsolin, a protein which accelerates actin polymerization by stabilizing nuclei from which filaments grow and which binds to their fast growing ends. The lengths of the actin filaments following polymerization were measured by electron microscopy and showed that the number-average filament length agreed with the predicted length if each gelsolin molecule acted as a seed for the growth of an actin filament. The distribution of lengths was independent of the actin:gelsolin ratio and was similar to that of actin filaments polymerized in the absence of gelsolin (Lw/Ln = 1.8). The mobility of these filaments in solution was studied by quasielastic light scattering and by viscometry. The translational diffusion constant determined by quasielastic light scattering was in agreement with the infinite dilution values calculated from the dimensions and the distribution of lengths determined by electron microscopy for relatively short filament lengths. Under conditions where overlap of the rotational domains of the filaments would be expected to occur, the measured diffusion rates deviated from their predicted dilute solution values and the solution viscosity increased abruptly. The dependence of the diffusion constant and the solution viscosity on the length of the actin filaments can be explained in terms of a theory that describes the restraints on diffusion of independent rigid rods in semi-dilute solution. The results suggest that the rheology of actin filaments can be accounted for by steric restraints. The length of cytoplasmic actin filaments in some cell types is such that these steric constraints are significant and could produce large changes in physical properties with small changes in filament length.     

F-actin, a model polymer for semiflexible chains in dilute, semidilute, and liquid crystalline solutions.

Single actin filaments were analyzed in solutions ranging from dilute (0.2 microgram/ml), where filaments interact only with solvent, to concentrations (4.0 mg/ml) at which F-actin forms a nematic phase. A persistence length of approximately 1.8 microns and an average length of approximately 22 microns (Kaufmann et al., 1992) identify actin as a model for studying the dynamics of semiflexible polymers. In dilute solutions the filaments exhibit thermal bending undulations in addition to diffusive motion. At higher semidilute concentrations (1.4 mg/ml) three-dimensional reconstructions of confocal images of fluorescently labeled filaments in a matrix of unlabeled F-actin reveal steric interactions between filaments, which account for the viscoelastic behavior of these solutions. The restricted undulations of these labeled chains reveal the virtual tube formed around a filament by the surrounding actin. The average tube diameter <a> scales with monomer concentration c as <a> varies; is directly proportional to c-(0.5 +/- 0.15). The diffusion of filaments in semidilute solutions (c = (0.1-2.0) mg/ml) is dominated by diffusion along the filament contour (reptation), and constraint release by remodeling of the surrounding filaments is rare. The self-diffusion coefficient D parallel along the tube decreases linearly with the chain length for semidilute solutions. For concentrations > 2.5 mg/ml a transition occurs from an isotropic entangled phase to a coexistence between isotropic and nematic domains. Analysis of the molecular motions of filaments suggests that the filaments in the aligned domains are in thermal equilibrium and that the diffusion coefficient parallel to the director D parallel is nearly independent of filament length. We also report the novel direct observation of u-shaped defects, called hairpins, in the nematic domains.     

Affinity of alpha-actinin for actin determines the structure and mechanical properties of actin filament gels.

Proteins that cross-link actin filaments can either form bundles of parallel filaments or isotropic networks of individual filaments. We have found that mixtures of actin filaments with alpha-actinin purified from either Acanthamoeba castellanii or chicken smooth muscle can form bundles or isotropic networks depending on their concentration. Low concentrations of alpha-actinin and actin filaments form networks indistinguishable in electron micrographs from gels of actin alone. Higher concentrations of alpha-actinin and actin filaments form bundles. The threshold for bundling depends on the affinity of the alpha-actinin for actin. The complex of Acanthamoeba alpha-actinin with actin filaments has a Kd of 4.7 microM and a bundling threshold of 0.1 microM; chicken smooth muscle has a Kd of 0.6 microM and a bundling threshold of 1 microM. The physical properties of isotropic networks of cross-linked actin filaments are very different from a gel of bundles: the network behaves like a solid because each actin filament is part of a single structure that encompasses all the filaments. Bundles of filaments behave more like a very viscous fluid because each bundle, while very long and stiff, can slip past other bundles. We have developed a computer model that predicts the bundling threshold based on four variables: the length of the actin filaments, the affinity of the alpha-actinin for actin, and the concentrations of actin and alpha-actinin.     

Spatiotemporal Dynamics of Actomyosin Networks

Rhodamine–phalloidin-labeled actin filaments were visualized gliding over a skeletal heavy meromyosin (HMM)-coated surface. Experiments at low filament densities showed that when two filaments collided, their paths were affected in a manner that depended on collision angle. Some collisions resulted in complete alignment of the filament paths; in others, the filaments crossed over one another. Filament crossover or alignment was equally probable at ∼40° contact angle. Filaments often underwent significant bending during collision and analysis of filament shape indicated an energy requirement of ∼13 k_BT. Experiments were performed over a wide range of HMM surface density and actin filament bulk concentration. Actin filament gliding speed and path persistence plateaued above a critical HMM surface density, and at high (micromolar) actin filament concentrations, filament motion became dramatically aligned in a common direction. Spatiotemporal features of alignment behavior were determined by correlation analysis, supported by simulations. The thermal drift of individual filament tracks was suppressed as the population became more oriented. Spatial correlation analysis revealed that long-range alignment was due to incremental recruitment rather than fusion of locally ordered seed domains. The global alignment of filament movement, described by an “order parameter,” peaked at optimal actin concentrations and myosin surface densities, in contrast to previous predictions of a critical phase transition. Either hydrodynamic coupling or exchange of filaments between the surface bound and adjacent bulk phase layers might degrade order at high actin filament concentration, and high HMM surface densities might decrease alignment probability during collisions. Our results are compatible with generation of long-range order from mechanical interaction between individual actin filaments. Furthermore, we show that randomly oriented myosin motors align relatively short, submicrometer actin filaments into motile surface domains that extend over many tens of micrometers and these patterns persist for several minutes.     

Spatiotemporal Dynamics of Actomyosin Networks

Rhodamine–phalloidin-labeled actin filaments were visualized gliding over a skeletal heavy meromyosin (HMM)-coated surface. Experiments at low filament densities showed that when two filaments collided, their paths were affected in a manner that depended on collision angle. Some collisions resulted in complete alignment of the filament paths; in others, the filaments crossed over one another. Filament crossover or alignment was equally probable at ∼40° contact angle. Filaments often underwent significant bending during collision and analysis of filament shape indicated an energy requirement of ∼13 k_BT. Experiments were performed over a wide range of HMM surface density and actin filament bulk concentration. Actin filament gliding speed and path persistence plateaued above a critical HMM surface density, and at high (micromolar) actin filament concentrations, filament motion became dramatically aligned in a common direction. Spatiotemporal features of alignment behavior were determined by correlation analysis, supported by simulations. The thermal drift of individual filament tracks was suppressed as the population became more oriented. Spatial correlation analysis revealed that long-range alignment was due to incremental recruitment rather than fusion of locally ordered seed domains. The global alignment of filament movement, described by an “order parameter,” peaked at optimal actin concentrations and myosin surface densities, in contrast to previous predictions of a critical phase transition. Either hydrodynamic coupling or exchange of filaments between the surface bound and adjacent bulk phase layers might degrade order at high actin filament concentration, and high HMM surface densities might decrease alignment probability during collisions. Our results are compatible with generation of long-range order from mechanical interaction between individual actin filaments. Furthermore, we show that randomly oriented myosin motors align relatively short, submicrometer actin filaments into motile surface domains that extend over many tens of micrometers and these patterns persist for several minutes.     

Mechanistic differences in actin bundling activity of two mammalian formins, FRL1 and mDia2

Formin proteins are regulators of actin dynamics, mediating assembly of unbranched actin filaments. These multidomain proteins are defined by the presence of a Formin Homology 2 (FH2) domain. Previous work has shown that FH2 domains bind to filament barbed ends and move processively at the barbed end as the filament elongates. Here we report that two FH2 domains, from mammalian FRL1 and mDia2, also bundle filaments, whereas the FH2 domain from mDia1 cannot under similar conditions. The FH2 domain alone is sufficient for bundling. Bundled filaments made by either FRL1 or mDia2 are in both parallel and anti-parallel orientations. A novel property that might contribute to bundling is the ability of the dimeric FH2 domains from both FRL1 and mDia2 to dissociate and recombine. This property is not observed for mDia1. A difference between FRL1 and mDia2 is that FRL1-mediated bundling is competitive with barbed end binding, whereas mDia2-mediated bundling is not. Mutation of a highly conserved isoleucine residue in the FH2 domain does not inhibit bundling by either FRL1 or mDia2, but inhibits barbed end activities. However, the severity of this mutation varies between formins. For mDia1 and mDia2, the mutation strongly inhibits all effects of barbed end binding, but affects FRL1 much less strongly. Furthermore, our results suggest that the Ile mutation affects processivity. Taken together, our data suggest that the bundling activities of FRL1 and mDia2, while producing phenotypically similar bundles, differ in mechanistic detail.     

Filamentous Biopolymers on Surfaces: Atomic Force Microscopy Images Compared with Brownian Dynamics Simulation of Filament Deposition

Nanomechanical properties of filamentous biopolymers, such as the persistence length, may be determined from two-dimensional images of molecules immobilized on surfaces. For a single filament in solution, two principal adsorption scenarios are possible. Both scenarios depend primarly on the interaction strength between the filament and the support: i) For interactions in the range of the thermal energy, the filament can freely equilibrate on the surface during adsorption; ii) For interactions much stronger than the thermal energy, the filament will be captured by the surface without having equilibrated. Such a ‘trapping’ mechanism leads to more condensed filament images and hence to a smaller value for the apparent persistence length. To understand the capture mechanism in more detail we have performed Brownian dynamics simulations of relatively short filaments by taking the two extreme scenarios into account. We then compared these ‘ideal’ adsorption scenarios with observed images of immobilized vimentin intermediate filaments on different surfaces. We found a good agreement between the contours of the deposited vimentin filaments on mica (‘ideal’ trapping) and on glass (‘ideal’ equilibrated) with our simulations. Based on these data, we have developed a strategy to reliably extract the persistence length of short worm-like chain fragments or network forming filaments with unknown polymer-surface interactions.     

A nucleation--elongation mechanism for the self-assembly of side polar sheets of smooth muscle myosin.

Self-assembled filaments of smooth muscle myosin were observed by low dose electron microscopy to be flat side-polar sheets, in which the component molecules appeared straight and close-packed. Fraying experiments released small oligomers, in which molecules were staggered in parallel by about +/- 14 nm relative to two immediate neighbours, and were bound also to an antiparallel partner via a approximately 14 nm overlap at the very tip of the tail. We suggest a filament model which preserves these packing relationships. Adding stoichiometric amounts of MgATP to the filaments caused them to disassemble completely by progressive loss of material from their ends, at a limiting rate equivalent to about 2 monomers per second per end in physiological saline. The rate of the competing association reaction varied linearly with the monomer concentration, as determined in pressure-jump experiments. This suggests that myosin monomers, rather than dimers or higher oligomers, are the building blocks of these filaments. Shearing and annealing of assembled filaments appeared negligible on a time scale of a few hours. In consequence, filament number and filament length were dependent on the rate at which monomers were supplied to the assembly reaction, and on the number of filaments already present at the start of the assembly reaction.     

Assembly of Acanthamoeba actin in the presence of Acanthamoeba profilin measured by fluorescence photobleaching recovery

Assembly of Acanthamoeba actin, of which trace quantities had been labeled with 5-(iodoacetamido)-fluorescein, was quantified using the modulation detection method of fluorescence photobleaching recovery (FPR). This technique permits explicit determination of the fraction of labeled actin incorporated into filaments and the translational diffusion coefficients of the filaments, from which filament length can be calculated. Addition of Acanthamoeba profilin in molar ratios to actin of about 1.1:1 and 2.3:1 retarded the initial kinetics of assembly (induced by addition of 2mM Mg+2) and reduced the fraction of actin incorporated into filaments. The diffusion coefficients of filaments formed were greatly changed by the presence of profilin at short times, but the differences became increasingly smaller at longer times. After 26 hr. the filaments formed in 1.1:1 profilin were about 12% shorter and in 2.3:1 profilin were about 20% shorter than filaments formed by actin alone under the same conditions.     

Molecular mechanism of bundle formation by the bacterial actin ParM

The actin homolog ParM plays a microtubule-like role in segregating DNA prior to bacterial cell division. Fluorescence and cryo-electron microscopy have shown that ParM forms filament bundles between separating DNA plasmids in vivo. Given the lack of ParM bundling proteins it remains unknown how ParM bundles form at the molecular level. Here we show using time-lapse TIRF microscopy, under in vitro molecular crowding conditions, that ParM-bundle formation consists of two distinct phases. At the onset of polymerization bundle thickness and shape are determined in the form of nuclei of short helically disordered filaments arranged in a liquid-like lattice. These nuclei then undergo an elongation phase whereby they rapidly increase in length. At steady state, ParM bundles fuse into one single large aggregate. This behavior had been predicted by theory but has not been observed for any other cytomotive biopolymer, including F-actin. We employed electron micrographs of ParM rafts, which are 2-D analogs of 3-D bundles, to identify the main molecular interfilament contacts within these suprastructures. The interface between filaments is similar for both parallel and anti-parallel orientations and the distribution of filament polarity is random within a bundle. We suggest that the interfilament interactions are not due to the interactions of specific residues but rather to long-range, counter ion mediated, electrostatic attractive forces. A randomly oriented bundle ensures that the assembly is rigid and that DNA may be captured with equal efficiency at both ends of the bundle via the ParR binding protein.