UCSF

AlfA is a filament-forming actin-like protein in Bacillus subtilis that functions to actively partition the large, low copy number plasmid by which it is encoded. Our in vitro observations of filament dynamics have revealed a set of kinetic and structural properties (namely constitutive bundling and lack of dynamic instability) that are inconsistent with previously established models for actin-like plasmid segregating proteins such as ParM. To understand the mechanism of AlfA -driven plasmid segregation, we imaged AlfA and its downstream DNA-binding protein, AlfB, interacting with plasmids in vivo and in vitro. Our live cell microscopy reveals that plasmids can move along existing AlfA structures or track the ends of growing ones, consistent with the idea that the AlfA polymer seen in vivo is actually a bundle of multiple filaments. Furthermore, these polymers can form between plasmids to push them apart, prompting us to ask how plasmids alter filament dynamics to generate this specific assembly. To address this question, we purified AlfB and found that it dramatically alters the kinetics and structure of AlfA. AlfB binds to AlfA monomers and polymers, not only increasing the critical concentration of assembly, but also preventing the otherwise very robust bundling of AlfA. The 100bp centromeric DNA region to which AlfB binds, however, rescues bundling and promotes polymerization. These observations lead us to a model of AlfA-driven plasmid segregation wherein bundles of AlfA form specifically in association with AlfB-DNA complexes. We propose that the intrinsic bundling property of the polymer, normally inhibited by a high concentration of free AlfB in the cytoplasm, functions as a capture mechanism to specifically join DNA-bound filaments to one another. Polymerization in opposite directions, driven by antiparallel bundling, would cause plasmids to be segregated from one another, ensuring their maintenance through cell division.