Abstract
The evolutionary conserved ESCRT-III complex is a device for membrane remodelling in various cellular processes, such as the formation of intraluminal vesicles (ILVs), cytokinesis, and membrane repair. The common theme of all these processes is the abscission of membrane away from the cytosol. At its heart in Drosophila is Shrub, CHMP4 in humans, which dynamically polymerises into filaments through electrostatic interactions among the protomers. For the full activity, Shrub/CHMP4 requires physical interaction with members of the Lgd protein family. This interaction is mediated by the odd-numbered DM14 domains of Lgd, which bind to the negative interaction surface of Shrub. While only one Lgd and one Shrub exist in the genome of Drosophila, mammals have two Lgd orthologs, LGD1/CC2D1B and LGD2/CC2D1A, as well as three CHMP4s in their genomes, CHMP4A, CHMP4B, and CHMP4C. The rationale for the diversification of the ESCRT components is not understood. We here use Drosophila as a model system to analyse the activity of the human orthologs of Shrub and Lgd at an organismal level. This enabled us to use the plethora of available techniques available for Drosophila. We present evidence that CHMP4B is the true ortholog of Shrub, while CHMP4A and CHMP4C have diverging activities. Nevertheless, CHMP4A and CHMP4C can enhance the activity of CHMP4B, raising the possibility that they can form heteropolymers in vivo. Our structure-function analysis of the LGD1 and LGD2 indicates that the C2 domain of the LGD proteins has a specific function beyond protein stability and subcellular localisation. Moreover, our data specify that CHMP4B interacts more efficiently with LGD1 than with LGD2.