I owe special thanks to Marnix Wieffer and Michael Krauß, who spent a lot of time helping me to get started in the lab, to overcome numerous experimental problem, I was not sure how to deal with, and frequently joined in discussions about my project with excellent advice. I am really grateful for your experimental support, Michael, and the time you invested in the project during our paper revision.Further, I would like to thank Jocelyn Laporte and his group, especially Anne-Sophie Nicot, at the IGBMC in Strasbourg for their support and a very fruitful collaboration, and Carsten Schultz and his group at the EMBL in Heidelberg for their support and constant supply with PIP/AMs. I owe special thanks to Dmytro Puchkov for the ultrastructural analysis and Eberhard Krause for mass spectrometry done within this project. I also need to acknowledge Markus Wenk and Federico Torta at the Singapore Lipidomics Incubator for joining the project at a very late stage, when we urgently needed help from lipidomics specialists. It was a pity that experiments did not work out as planned.I would further like to express my gratitude to two very skilled technicians in the AG Haucke lab: Silke Zillmann and Delia Löwe. Without your help it would have been impossible to achieve so much within the limited amount of time I had during the paper revision. by VPS34-IN1 treatment................................................... 52 2.3.11 Fluorescence microscopy................................................................................. 53 2.3.12 Flow cytometry............................................................................................ III SummaryPhosphoinositides (PIs) are a minor class of short-lived phospholipids that serve as crucial signposts of membrane identity. Thereby, PIs full fill important functions in cell signaling and membrane transport. PI 4-phosphates such as phosphatitylinositol-4-phosphate (PI(4)P) and phosphatitylinositol-4,5-bisphosphate (PI(4,5)P 2 ) are enriched at the plasma membrane (PM), on secretory organelles and lysosomes, while PI 3-phosphates, i.e.phosphatitylinositol-3-phosphate (PI(3)P), are a hallmark of the endosomal system.Directional transport between these compartments, thus, requires regulated PI conversion.However, PI conversion in exit from PI(3)P-enriched endosomes en route to the PI(4)P-and PI(4,5)P 2 -positive PM in endosomal recycling remained unknown.Here, we report that cargo exit from endosomes requires removal of PI(3)P by the PI(3)P 3-phosphatase myotubularin 1 (MTM1), and concomitant PI(4)P synthesis by PI 4-kinase type II α (PI4K2α). Loss of MTM1 causes endosomal accumulation of PI(3)P and PI(3)P effector proteins, i.e. sorting nexins, Kif16b-mediated outward traffic of PI(3)P containing endosomes and sub-plasmalemmal accumulation of exocytosis-deficient endosomes. As PI4K2α associates with MTM1 and thereby facilitates membrane recruitment of MTM1, these phenotypic changes are mimicked by loss of PI4K2α. The conversion of PI(3)P-to-PI(4)P is paralleled by a switch i...

The mechano-chemical protein dynamin is the prototype of the dynamin superfamily of large GTPases, which shape and remodel membranes in diverse cellular processes 1 .Dynamin forms predominantly tetramers in the cytosol, which oligomerize at the neck of clathrin-coated vesicles to mediate constriction and subsequent scission of the membrane 1 . Previous studies have described the architecture of dynamin dimers 2,3 , but the molecular determinants for dynamin assembly and its regulation have remained unclear. Here, we present the crystal structure of the nucleotide-free dynamin tetramer.Combining structural data with mutational studies, oligomerization measurements and molecular dynamics simulations, we suggest a mechanism of how oligomerization of dynamin is linked to the release of intramolecular auto-inhibitory interactions. We elucidate how mutations that interfere with tetramer formation and auto-inhibition can lead to the congenital diseases Charcot-Marie-Tooth neuropathy (CMT) 4 and centronuclear myopathy (CNM) 5 , respectively. Strikingly, the bent shape of the tetramer explains how dynamin assembles into a right-handed helical oligomer of defined diameter, which has direct implications for its function in membrane constriction.The three highly conserved vertebrate isoforms of dynamin contain five distinct domains (Extended Data Fig. 1a): an N-terminal GTPase (G) domain mediating nucleotide binding and hydrolysis, a bundle signaling element (BSE), a stalk, a pleckstrin homology (PH) domain involved in lipid binding, and a proline rich domain (PRD) mediating interaction with BAR-and SH3-domain containing scaffolding proteins 6 . To exert its function in clathrinmediated endocytosis (CME), dynamin assembles via the stalks into a helical array surrounding the necks of invaginating clathrin-coated pits (CCP) 7,8 . Dimerization of GTPbound G domains from neighboring helical rungs induces GTP hydrolysis 9 . The ensuing conformational changes are thought to be transmitted from the G domain via the BSE to the stalk resulting in a sliding motion of adjacent helix rungs, concomitant helix constriction 10 , and eventually membrane scission. The inherent tendency to form large assemblies at high protein concentrations has hampered crystallization of dynamin in the past. The use of non-3 oligomerizing mutants led to crystal structures of dynamin 1 2,3 . However, the postulated higher-order assembly interface was not resolved in these structures leaving the oligomerization mechanism unaddressed.We reasoned that an alternative assembly-affecting mutation, such as K361S in dynamin 3 11 , may disturb the oligomerization interface to a lesser extent than the previously used mutants. We obtained crystals of nucleotide-free dynamin 3-K361S lacking the PRD (dynamin 3(∆PRD)-K361S) that diffracted to 3.7 Å (Methods, Extended Data Fig. 1, Extended Data Table 1). Indeed, the asymmetric unit of the crystal lattice contained a dynamin tetramer that did not form the filamentous superstructures seen for dynamin 1 2,3 .The dynamin tetr...