Regulation of Human Atlastin Fusion Activity

The endoplasmic reticulum (ER)’s network is comprised of sheets and tubules fused together generating a lattice-like network. Critical for three-way junctions and maintenance of the ER structure are the atlastin (ATL) GTPases that catalyze homotypic tubule fusion. Mammals possess three ATL paralogs (ATL1/2/3), which are each differentially expressed. Recently our lab successfully reconstituted human ATL1/2 fusion activity in vitro revealing that both are autoinhibited by their non-conserved C-termini, albeit to differing degrees. This finding was transformative in our understanding of ATLs, but also raised many questions regarding regulation of the proteins, including if autoinhibition must be alleviated for every round of fusion, and if so how. Here I reconstitute the fusion activity of the third previously understudied paralog, ATL3 revealing that uniquely it is a constitutively active fusogen lacking autoinhibition by its C-termini. Despite prior uncertainty regarding ATL3’s capacity as a fusogen, I demonstrate both that ATL3 can catalyze fusion as robustly as the maximally active ATL2 when at higher concentrations, and that ATL3 is sufficient to maintain the ER network alone under overexpression conditions. From these results we pursued investigations into the evolutionary relationship between paralogs as well as alternate splice isoforms. We learn that autoinhibition was a recent evolutionary innovation independently gained by paralogs. We also demonstrate that both autoinhibited paralogs possess alternatively spliced isoforms that differ in their Cterminal extensions and have varying levels of activity. Finally, through probing mechanisms for relief of autoinhibition, I demonstrate that the C-terminal autoinhibitory domain (CAD) can interact intermolecularly, allowing for activation when autoinhibited ATL is present along with a disinhibited ATL. Together my investigations establish ATL3 as a constitutive fusogen potentially capable of maintaining the ER network in the absence of autoinhibition relief, and identify additional layers of ATL regulation including splice variants with differing activity and intermolecular CAD interactions.