Plasmid segregation by a moving ATPase gradient

Unlike the mitotic segregation of eukaryotic sister chromatids, DNA partitioning in bacteria is still not well understood. Bacterial high–copy-number plasmids can be stably maintained by random distribution of their copies during cell division. In contrast, the faithful transmission of low–copy-number plasmids and many chromosomes depends on an active process mediated by conserved, tripartite segregation systems (1). A central component of these machineries is a nucleoside triphosphatase driving the partitioning reaction, which can be classified as an actin-like ATPase (ParM), a tubulin-like GTPase (TubZ), or a Walker-type ATPase (ParA). Systems using an actin or tubulin homolog function by means of a filament-based pushing or pulling mechanism (2, 3). Most low–copy-number plasmids and chromosomes are, however, segregated by Walker-type ATPases. Despite extensive research, it is not yet unambiguously established how this third group of proteins harnesses the energy released during ATP hydrolysis for plasmid movement. Based on previous analyses, two competing models have been put forward. In the filament-pulling model, ParA is assumed to form polymers that move DNA by repeated polymerization/depolymerization cycles. In contrast, the diffusion-ratchet model proposes a concentration gradient of ParA dimers on the nucleoid as the driving force for DNA segregation. In PNAS, Vecchiarelli et al. (4) now provide direct evidence in support of the latter model by fully reconstituting in vitro the segregation system of the Escherichia coli F plasmid.

Plasmid segregation systems based on Walker-type ATPases position plasmid copies at regular distances over the nucleoid (5–7). In addition to the ATPase component (ParA), they comprise a centromere-like DNA sequence (parS) and an adaptor protein (ParB). Binding of ParB to parS motifs on the DNA cargo results in the formation of a so-called “partition complex” (8). This complex then dynamically interacts with ParA to drive the directed movement of the DNA cargo (Fig. 1). In the presence of ATP, ParA associates nonspecifically with DNA and exhibits a weak intrinsic ATPase activity that is stimulated synergistically by ParB and DNA (9–11). Moreover, several biochemical studies have shown that ParA can assemble into filamentous structures upon ATP binding (6, 10, 12, 13). This finding, together with analyses of ParA localization in vivo, has provided the basis for the filament-pulling model of DNA segregation (7). However, the physiological relevance of ParA polymerization is highly controversial (14, 15). In particular, recent studies investigating the E. coli P1 and F plasmid segregation systems have cast serious doubt on a role of filament formation in the partitioning process (5, 16–18). Instead, ParA was proposed to act by a diffusion-ratchet mechanism, which includes the following steps: ParA-ATP dimers bind nonspecifically to chromosomal DNA and transiently tether plasmids to the nucleoid surface through interaction with the ParB–parS partition complex (17). In the resulting quaternary complex, ParB stimulates the ATPase activity of ParA, thereby inducing its release from the DNA. Because reactivation of ParA involves a series of slow conformational changes, it is unable to immediately reassociate with the nucleoid (17). This lag creates a ParA depletion zone in the vicinity of the partition complex, which ultimately results in the detachment of the plasmid from the nucleoid surface. After its dissociation, the plasmid diffuses in a stochastically chosen direction. As it reaches the edge of the depletion zone, it encounters an increasing number of nucleoid-bound ParA dimers, which make new contacts to the plasmid partition complex. Moving along this ParA gradient, the plasmid is finally immobilized again and initiates the formation of a new depletion zone. As a central point of the model, the initial direction taken by the plasmid is reinforced by low ParA concentrations in the wake of the partition complex, thereby giving rise to robust, unidirectional movement of plasmid molecules (Fig. 1C).

Fig. 1.

(A) Partitioning (sop) locus of the E. coli F plasmid, comprising genes for an ATPase (SopA) and an adaptor protein (SopB), as well as a centromere-like region including multiple tandem repeats of the SopB binding site (sopC). (B) The sop locus is essential for the faithful segregation of F plasmids to the future daughter-cell compartments. Wild-type plasmids are regularly positioned over the nucleoid (Upper), whereas plasmids lacking the sop locus are concentrated in the polar regions of the cell (Lower). (C) Mechanism of sop-mediated plasmid movement. Nucleoid-associated SopA-ATP dimers bind to a SopB–sopC partition complex, thereby immobilizing a copy of the F plasmid on the nucleoid surface. After stimulation of its ATPase activity by SopB, SopA dissociates from the DNA, leaving behind a zone of low SopA concentration (Upper). As a consequence, the plasmid starts to diffuse and then reattaches to the nucleoid through association with neighboring SopA molecules (Lower). Iteration of these steps results in the directed movement of plasmid copies across the nucleoid surface.

As a first step toward verifying the diffusion-ratchet model, the P1 and F plasmid segregation systems have recently been reconstituted in a cell-free set-up (16, 18). To mimic the nucleoid surface, the bottom of a 25-µm-thick microfluidic flow cell was densely coated with nonspecific DNA fragments. The flow cell was then loaded with a mixture of ParA-GFP, ParB, and a fluorescently labeled parS-bearing plasmid prepared in an ATP-containing buffer. Subsequently, the dynamics of the components were visualized by total internal reflection microscopy (TIRFM), a technique that specifically resolves surface-associated processes. The two systems exhibited similar dynamics, suggesting a comparable segregation mechanism. Initially, plasmids were stably tethered to a layer of ParA-GFP associated with the DNA carpet. After some time, they started to move around their anchor point in a circular Brownian motion, leading to the displacement of ParA-GFP in their surroundings. Eventually, the plasmids dissociated from the surface and diffused away, leaving behind a zone depleted of ParA-GFP molecules. These observations strikingly mirrored the in vivo dynamics of the P1 plasmid (5) and were consistent with key aspects of the diffusion-ratchet model described above. Importantly, ParA molecules were evenly distributed over the DNA carpet and exchanged rapidly with molecules in solution (16, 18), supporting the idea that the active, nucleoid-bound species of ParA consists of dimeric complexes rather than filaments. The visualized dynamics were dependent on the presence of both ParA and ParB, as well as on the ATPase activity of ParA. When either of the Par proteins was omitted from the reaction, plasmids did not associate with the flow cell surface. In contrast, in the absence of ATP hydrolysis, the plasmids remained irreversibly attached to the surface, highlighting the importance of transient ParA–DNA interactions. A shortcoming of the in vitro system used in these studies was that it could only recapitulate the initial stages of the diffusion-ratchet model because the large size of the flow cells allowed plasmids to diffuse away from the surface after detachment. As a consequence, it was not possible to observe the iterative diffusion/association cycles that were proposed to drive directional plasmid movement. In the bacterial cell, the diffusion of plasmids is constrained by the narrow space between the nucleoid surface and the cytoplasmic membrane, a situation that greatly facilitates repeated interactions of ParB–parS partition complexes with nucleoid-bound ParA molecules. Therefore, in line with the experimental results, spatial confinement was proposed to be another key requirement for ParA-mediated plasmid

The study by Vecchiarelli et al. significantly advances our understanding of bacterial DNA segregation.

segregation by a diffusion-ratchet mechanism (16, 18).

In the present study, Vecchiarelli et al. (4) have verified this hypothesis by reinvestigating the dynamics of F plasmid segregation using an advanced version of their cell-free system. The plasmids used in the previous studies were replaced by magnetic beads coated with fluorescently labeled sopC-containing DNA. Spatial confinement was then simulated by application of a magnetic force perpendicular to the DNA-coated flow cell surface. This experimental set-up not only reproduced the previously observed P1 and F plasmid dynamics, but also made it possible to follow the cargo during its mobile phase. Strikingly, beads migrated in a directed manner over distances of several microns, driven by repeated cycles of ParA depletion and surface detachment. In doing so, they tracked along a moving SopA gradient, leaving behind a region of low SopA concentration that was slowly refilled with SopA molecules from solution. It still needs to be determined if, analogous to P1 ParA, a time-delay between ATP binding and reassociation with DNA is involved in the emergence of the SopA gradient. Interestingly, not all beads exhibited directional movement. Whereas directed beads consistently stayed in close contact with the surface, others diffused freely along the surface and “bounced” in and out of the TIRFM illumination area, ruling out a significant contribution of the magnetic force to the straight motion of directed beads.

Although the characteristics of a magnetic bead differ from those of a DNA molecule, the study by Vecchiarelli et al. (4) significantly advances our understanding of bacterial DNA segregation. The work identifies the partitioning reaction driven by (at least a subset of) Walker-type ATPases as a diffusion-regulated process, and thus adds to the growing body of evidence suggesting that the physiological relevance of filament formation by these proteins may have been overestimated. It will be interesting to perform the same kind of analysis on other ParA-dependent segregation systems that have been proposed to use a polymerization-based mechanism (19). Moreover, modeling studies will be required to understand how the movement of plasmids along ParA gradients finally leads to the faithful distribution of sister copies to the two daughter cells.