Intrinsic disorder and amphitropism in the mitochondrial fission mechanoenzyme Drp1

The Dynamin superfamily is a class of large GTPases that perform essential membrane remodeling events via self-assembly stimulated GTP hydrolysis in a manner that is not well understood. Malfunction in these enzymes is the basis of a number of diseases. Dynamin superfamily enzymes hydrolyze GTP via a G domain, self-assemble via a stalk domain, and may couple assembly and hydrolysis via a “bundle signaling element” domain. An additional “variable domain” (VD) is present in many superfamily members and may enable specific targeting to sites of action, but the nature of VD function is unclear in most dynamins. While VDs are not conserved in amino acid sequence, they may have similarities in function, namely allosteric regulation and membrane interaction. We identify two modes of membrane interaction in the dynamin superfamily: (1) tethering to the membrane via a transmembrane helix or a lipophilic anchor and (2) untethered, reversible interaction with membranes (amphitropism). We propose a subclassification of the dynamin superfamily into these two categories. We employ the mitochondrial dynamin, Drp1 as a model of the amphitropic dynamins because of its relative simplicity and relevance to human health. Specifically, we seek to identify the role of the Drp1 VD, or B domain, anticipating that it will also inform understanding of VDs in other amphitropic dynamins. We find that removal of the B domain from Drp1 results in enhanced assembly and GTP hydrolysis, suggesting that the B domain has an auto-inhibitory role. We find that the B domain is intrinsically disordered (ID) and surprisingly, undergoes phase separation or coacervation under conditions that induce other IDPs to fold. Finally, we show that the B domain binds lipid membranes with a specificity for cardiolipin, and that this interaction is enhanced under conditions that favor the coacervated state. Based on these findings we suggest the possibility that the B domain acts as an entropic bristle in order to inhibit Drp1 assembly, and that this auto-inhibition is relieved upon interaction of the B domain with a lipid membrane, perhaps involving the process of coacervation. We suspect that this model may also be applicable to other amphitropic dynamins