These filopodia-inducing IM domains are found in cytoskeletal scaffolding proteins such as missing-in-metastasis and IRSp53. Sites of membrane deformation, including filopodial protrusions, are often associated with bundles of actin filaments. Previous results suggested that the IM domain contributes to filopodium formation by bundling actin, but Mattila and colleagues now dispute this result.
The authors find that the IM domain only bundles actin at unnaturally low ionic strengths, at which point the domain tends to form aggregates. The IM domain also did not colocalize with actin bundles in cells, but instead was observed at the plasma membrane surrounding the bundle.
The membrane association makes more sense, according to the authors, because the IM domain shows structural homology to another protein domain, BAR, that binds and deforms membranes.
BAR domains have a curved, banana-like structure and interact with the plasma membrane through residues on the banana's inside curved edge. IM domains have a similar although less curved structure. Here, the authors map IM's membrane-interacting residues to its outside curved edge. This opposite geometry explains why BAR proteins promote endocytotic invaginations, whereas IM domain proteins promote filopodial protrusions.
The geometry of IM's membrane binding also predicts that IM domains would bend membranes into filopodial tubules of ∼95nm diameter—closely fitting with the size of IM-induced tubules that the team observed by electron microscopy.
Filopodium formation was dependent on IM domains interacting with PI(4,5)P2-rich membrane regions. It is yet unclear, however, whether PI(4,5)P2 enrichment is a cause or coincidence of IM's membrane binding. Filopodium formation was also dependent on IM domains binding to actin. Thus, although IMs do not appear to regulate the bundling of actin, their membrane deformation activity is connected to actin dynamics to enable filopodial growth and stability.