Reconstitution of DNA Segregation Driven by Assembly of a Prokaryotic Actin Homolog

See allHide authors and affiliations

Science  02 Mar 2007:
Vol. 315, Issue 5816, pp. 1270-1274
DOI: 10.1126/science.1138527


Multiple unrelated polymer systems have evolved to partition DNA molecules between daughter cells at division. To better understand polymer-driven DNA segregation, we reconstituted the three-component segregation system of the R1 plasmid from purified components. We found that the ParR/parC complex can construct a simple bipolar spindle by binding the ends of ParM filaments, inhibiting dynamic instability, and acting as a ratchet permitting incorporation of new monomers and riding on the elongating filament ends. Under steady-state conditions, the dynamic instability of unattached ParM filaments provides the energy required to drive DNA segregation.

To ensure fidelity of gene transmission, DNA molecules must be evenly distributed among daughter cells before division. Eukaryotes harness the polymerization of tubulin to drive alignment and segregation of chromosomes (1); chromosome segregation in some eubacteria requires assembly of actin-like filaments (2, 3); and some extrachromosomal DNA elements have evolved their own polymer-based DNA segregation machinery (4, 5). Plasmid R1 is a large (100 kb), low–copy number plasmid with segregation machinery encoded by the par operon (6, 7). This operon is composed of three elements: parC, parR, and parM (6). parM encodes an actin-like protein that assembles into dynamically unstable filaments (8). parR encodes a protein that binds cooperatively to sequence repeats within parC, forming a complex that binds ParM filaments (9). It has been proposed that products of the par operon assemble into a bipolar spindle-like structure whose elongation drives plasmid segregation (9), but it is not yet known how these components assemble and how they convert the free energy of ParM polymerization into useful work.

By atomic structure, ParM is similar to actin (10), but by filament assembly dynamics, ParM is distinguished from actin by three important properties: (i) rapid, rather than slow, spontaneous nucleation; (ii) symmetrical, bidirectional elongation rather than polarized growth; and (iii) dynamic instability [the tendency to switch spontaneously (11) between phases of steady elongation and rapid shortening] rather than steady-state treadmilling (8). In vitro, the combination of these kinetic parameters results in a steady-state population of short (∼1.5 μm) and unstable (lifetime ∼20 s) ParM filaments (8).

To understand how ParR and parC harness filament assembly to segregate DNA, we purified all three components and tested their ability to self-assemble in vitro. We expressed and purified ParM and ParR proteins from Escherichia coli (fig. S1). We attached fluorescent (Cy3-labeled) DNA containing the parC sequence to 350-nm spherical beads. We then mixed the parC-coupled microspheres with ParR, added fluorescently labeled (Alexa 488) ParM, and induced filament assembly by adding adenosine triphosphate (ATP).

ParM filaments formed small radial arrays surrounding isolated parC beads (Fig. 1A), reminiscent of microtubule asters formed around isolated centrosomes (12). Aster formation required ATP, ParR (fig. S2A), and DNA containing the parC sequence (fig. S2B). Asters were dynamic, with filaments growing and shrinking from the surface of the bead, and grew to a maximum radius of 3 μm (Fig. 1B and movie S1), similar to the maximum lengths of individual ParM filaments in solution (8). In the presence of the nonhydrolyzable ATP analog adenylyl-imidodiphosphate (AMP-PNP), astral ParM projections grew much longer (Fig. 1C, fig. S3, and movie S2), indicating that the dynamic instability of ParM limited filament length even when one end of the filament was bound to the ParR/parC complex.

Fig. 1.

In vitro reconstitution of the R1 plasmid spindle. (A) Individual ParR/parC-coated beads with radiating fluorescently labeled ParM asters (red, Cy3-labeled DNA; green, Alexa 488–labeled ParM). (B) Left: Time-lapse sequence of an Alexa 488–labeled ParM aster. Right: Maximum intensity projection of the time-lapse sequence, illustrating fall-off in fluorescence at 3 μm. (C) Inhibition of ParM dynamic instability with AMP-PNP produces substantially larger asters. (D) Two-color images and time-lapse sequence of bipolar ParM spindles segregating ParR/parC-coated beads. (E) Individual ParM filaments run from bead to bead, as shown by EM of negatively stained R1 spindles. (F) Time-lapse series of bipolar spindle elongation. (G) Elongation of a multipolar ParM structure.

In addition to dynamic asters, ParM filaments also formed long, stable bundles connecting pairs of parC beads (Fig. 1, D to G), similar to bipolar structures previously observed in vivo (7). The bundles and attached beads moved as a single unit with fluid flow, indicating that ParM filaments were tightly attached to the ParR/parC complex. By electron microscopy (EM) we observed parC beads connected by ParM bundles of varying thickness. We traced individual filaments from one parC bead to the other (Fig. 1E), indicating that both ends of the ParM filament interacted with the ParR/parC complex. In time-lapse movies, the bivalently attached ParM bundles elongated at a constant rate, pushing parC beads in opposite directions (Fig. 1, D and F, and movies S3 and S4) over long distances (>120 μm). Thus ParM, ParR, and parC are sufficient to form a bipolar, DNA-segregating spindle. In addition, multiple parC beads often interacted to form multipolar linear chains and polygons of ParM bundles (Fig. 1G and movies S5 and S6), structures that also expanded equilaterally as each bundle elongated at the same rate.

The fact that we observed long, stable ParM filaments only between pairs of parC beads argues that both ends of each filament are stabilized against catastrophic disassembly by interaction with the ParR/parC complex. This is supported by the observation of a “search-and-capture” (13) process of spindle formation: When two unconnected ParM asters come into proximity, they stabilize a filament bundle whose elongation pushes the asters apart (Fig. 2A, fig. S4, and movies S7 and S8). We tested whether bipolar attachment was required for stabilization by using laser irradiation to cut through ParM spindles. After cutting, both severed ends of the bundle depolymerized back to the parC beads (Fig. 2B, fig. S5, and movies S9 and S10). This requirement for bipolar stabilization of ParM spindles provides an explanation for the apparent plasmid counting observed in vivo, where ParM spindles occur only in cells containing two or more plasmids (9).

Fig. 2.

ParM filaments are stabilized when bound at each end by the ParR/parC complex. (A) ParM spindles form by “search and capture.” Sequence shows isolated ParM asters forming bipolar spindles that subsequently push the beads apart. Red arrows indicate capture events. (B) ParM filament bundles are stabilized a both ends by interaction with ParR/parC beads. Cutting a ParM spindle by laser irradiation (red box) results in depolymerization of severed ends. (C) The R1 spindle finds the long axis of a channel. Time-lapse sequence shows R1 spindles confined within a microfabricated channel.

To determine whether the R1 spindle was sufficient to find the long axis of a bacterium-sized space, we assembled spindles in microfabricated channels of various shapes. Spindles always aligned with the long axis of the channel and pushed parC beads in opposite directions (Fig. 2C and movie S11), demonstrating that orientation of R1 spindles can occur without cellular landmarks. Spindles elongated freely until they encountered resistance. Elongation stalled at the ends of the channels and slowed at bends. Thus, the R1 spindle could find the long axis of a rod-shaped cell by a simple Brownian ratchet-type mechanism.

We next used photobleaching and speckle microscopy to determine where new ParM monomers incorporated into the R1 spindle. We first used low-intensity laser irradiation to photobleach pairs of reference marks onto elongating spindles (Fig. 3A, fig. S6, and movie S12). In all cases (n = 6), the distance between the two bleached marks did not change with time, whereas the distance between each mark and the nearest parC bead increased at a constant rate (Fig. 3B). We never observed recovery of fluorescence within a bleached zone, which suggests that no polymerization took place in the middle of the ParM bundle and that individual filaments did not slide past each other. Speckle microscopy (14) of elongating spindles indicated that monomers incorporated exclusively at the surface of parC beads (Fig. 3C and movies S13 and S14).

Fig. 3.

R1 spindles elongate by insertion of ParM monomers at the ParR/parC complex. (A) Photobleaching (indicated by red arrows) of ParM spindles reveals symmetrical, bipolar elongation at the bead surface. (B) ParM spindles elongate at each end at the same rate as free filament ends. The distance between the two photobleached stripes or distance between the stripe and the bead in (A) was measured on a per-frame basis, and the change in distance was plotted against time. (C) By speckle microscopy, all new ParM monomers add to the spindle at the ParR/parC interface. Green, Alexa 488–ParM; red, rhodamine-ParM doped at 1:1000 to produce fluorescent speckles. Left: Time-lapse sequence of an elongating spindle end (80 s per frame). Right: Kymograph of the time-lapse sequence.

Interaction with the ParR/parC complex did not affect the rate of ParM filament growth. From the photobleaching and speckling experiments, we determined that filaments within a spindle elongate at a rate of 11.6 ± 1.7 monomers s–1 (n = 5 ends). From the elongation rate and steady-state monomer concentration of 2.3 μM (15), we calculate a rate constant for elongation of bead-attached ParM filaments of 5 μM–1 s–1, identical to the rate constant for elongation of free ParM filaments (8).

In ParM mutants defective in ATP hydrolysis, the critical concentration for assembly is 0.6 μM (8). This is the free monomer concentration above which ATP ParM filaments elongate and below which they shorten. The critical concentration of adenosine diphosphate (ADP) ParM filaments is greater by a factor of more than 200 (>120 μM), and this nucleotide-dependent increase is the basis for dynamic instability of ParM filaments. At steady state, in the presence of ATP, the rate of filament nucleation is balanced by the rate of filament catastrophe, and the combination of growing and shrinking filaments produces a steadystate monomer concentration of 2.3 μM. Beneath 2.3 μM, no filaments are detectable by fluorescence microscopy, fluorescence resonance energy transfer (FRET), or high-speed pelleting (8).

To determine how binding to the ParR/parC complex affects the critical concentration of ParM filaments, we investigated the concentration dependence of R1 spindle formation. At concentrations above 2.3 μM, ParM formed numerous dense and stable spindles connecting pairs of parC beads (Fig. 4A). Below 2.3 μM, ParM formed spindles between beads, but their frequency and lifetime decreased with decreasing ParM concentration. Below 0.6 μM, ParM failed to form any detectable spindles, even between parC beads in close contact. Thus, interaction with the ParR/parC complex stabilized ParM filaments down to, but not beneath, the ATP critical concentration of 0.6 μM, presumably because the ParR/parC complex simply inhibits dynamic instability.

Fig. 4.

Dynamic instability of free ParM filaments provides the energy for R1 spindle elongation. (A) The ParR/parC complex stabilizes ParM filaments down to the ParM-ATP critical concentration. Spindle assembly reactions were performed at the indicated concentrations of Alexa 488–labeled ParM. (B) By abolishing the difference in critical concentration between free and ParR/parC-bound ParM filaments, AMP-PNP eliminates sustained polymerization on ParR/parC beads. (C) Small amounts of ATP added into AMP-PNP restore sustained polymerization on ParR/parC beads. (D) Hydrolysis-dependent ParM tail elongation occurs specifically at filament ends bound to ParR/parC complexes and not at free filament ends. Assembly reactions as in (C) were initiated with Alexa 488–ParM (green); at indicated times, the reactions were spiked with Cy3-ParM (red). Results were visualized after 50 min. Two examples of each time point are shown. (E) Rates of ParM tail elongation scale with the fraction of hydrolyzable nucleotide. Reactions as in (C) were performed at various ATP/AMP-PNP ratios. The rate of tail elongation was plotted against the percent of ATP within the nucleotide mixture.

One consequence of our results is that the energy to segregate parC beads must be supplied by dynamic instability of unattached ParM filaments. For polymerization to perform useful work, the monomer-polymer balance at the load must be kept away from equilibrium (16). We found that ParM filaments in the spindle are stabilized to the ATP critical concentration (0.6 μM) (Fig. 4A) yet elongate at a rate determined by the steady-state monomer concentration of 2.3 μM (Fig. 3B), a concentration that is maintained by dynamic instability of unattached ParM filaments (8).

To demonstrate this directly, we varied the energy difference between attached and unattached filaments by adding different ratios of ATP and AMP-PNP to our reconstituted DNA segregation system. In 100% AMP-PNP, the ParR/parC-attached and unattached filaments had the same critical concentration (0.6 μM), and the system quickly reached equilibrium (∼2 min) with parC beads surrounded by a small, nondynamic shell of ParM filaments (Fig. 4B). Reestablishing an energy difference by adding small amounts of ATP produced sustained ParM polymerization at the bead surface. In 5% ATP and 95% AMP-PNP, slow-growing, stable, monopolar tails assembled on the surface of large parC beads, propelling them through the medium at a constant rate for up to 2 hours (Fig. 4C and movies S15 and S16). By using two colors of labeled ParM, added at different times (Fig. 4D), we found that filament elongation occurred exclusively on ParR/parC associated ends and not on free ends. Hence, the unattached filament ends are at steady state, whereas ParR/parC-associated ends elongate in a hydrolysis-dependent manner. The elongation rate of stabilized tails increased with the proportion of hydrolyzable ATP (Fig. 4E), demonstrating that the energy difference driving bead motility was proportional to the amount of hydrolysis-capable monomer that could be stabilized by the ParR/parC complex.

In response to evolutionary pressure, the R1 plasmid has evolved a self-contained, three-component system to ensure its segregation. All key functions of the R1 spindle require dynamic instability of ParM filaments. Dynamic instability enables unbound filaments to turn over without disassembly factors (8). Filaments stabilized at one end by ParR/parC can search for additional plasmids (fig. S8 and movie S17); filaments bound at both ends are stabilized against dynamic instability, forming a productive spindle. Finally, the dynamic instability of unbound filaments provides the excess monomer to drive elongation of the stabilized filaments within the spindle.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S8

Movies S1 to S17

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article