Plasmids (green) are pushed (left to right) around a bacterium by polymerizing filaments of ParM (red).
The prokaryotic actin look-alike is ParM, which forms a filament that, like microtubules, is dynamically unstable. ParM is encoded by plasmid operons that also contain centromeric sequences and the gene for a DNA-binding protein that hooks plasmids to ParM. To watch ParM in action, the authors imaged bacteria containing a low-copy plasmid that is segregated to daughter cells. The videos unveiled a sloppy, dynamic segregation machinery.
When ParM protein was present, plasmids were pushed around much faster than by diffusion. In cells that had two plasmid copies, these erratic movements occasionally brought plasmids close enough together for a bundle of ParM filaments to link the two. Filaments then elongated, thereby pushing the plasmids to opposite ends of the bacterium.
Once plasmids reached the poles, the filaments rapidly collapsed, perhaps triggered by the force of the plasmids' contact on the cell membrane. The cycle then repeated: plasmids were again nudged along by new ParM filaments, found each other, and were pushed apart. This cyclic behavior continued until the cell divided, usually landing one plasmid in each daughter.
The group studied plasmid separation because it's easy to spy on in vivo, but plasmids might have pilfered the system from bacterial chromosomes. Mullins says that this DNA separation system is good enough for bacteria, whose large numbers can withstand occasional errors. It's 100-fold more efficient than no system and probably requires less energy than do high fidelity eukaryotic segregation systems. And the need for only two proteins, as opposed to the dozens eukaryotes use, helps keep the genome compact. 
Reference:
