An allosteric network within dynamin

Dynamins are large enzymes that catalyse the hydrolysis of GTP (GTPase activity). They assemble through oligomerisation into helical polymers during endocytosis to facilitate membrane scission and vesicle release in all known eukaryotic cell types. Dynamin oligomerisation is associated with an increase in GTPase activity. This drives constriction of the dynamin polymer to induce membrane fission at the neck of endocytic vesicles, releasing the new vesicles into the cytoplasm. This requires GTP hydrolysis at multiple G domains within the helix to be coordinated. The underlying mechanisms that regulate such coordination remain poorly understood. It is reasonable to predict that these enzymes possess a highly developed allosteric network that is responsible for this coordination. There are three classical dynamin (dynI-III) genes in the mammalian genome, each of which is alternately spliced. DynI is a neuronal isoform required for synaptic vesicle recovery mechanisms within the presynapse of neurons. There are 5 dynamin domains: the catalytic G domain; the bundle signalling element (BSE); the dynamin middle domain; the pleckstrin homology (PH) domain; and the C-terminal proline rich domain (PRD). Dyn I and II share a homologous splice site within the middle domain, which is the primary domain mediating assembly. This splice site produces two variants ‘a’ and ‘b’ with the alternate exon encoding the complete second α-helix of the middle domain (Mα2). This helix is part of a predicted regulatory interface in dynamin that allosterically regulates the G domain via the BSE The overall aim was to understand dynamin allostery by focusing on the predicted regulatory interface between the middle domain α2 helix (Mα2) and the BSE of dynamin, which is known as interface 5. The work utilised natural splice variation of the Mα2 as a key tool to explore the properties of this interface and to uncover the role of alternative splicing at this site. Firstly, it was hypothesised that alternate splicing of Mα2 in dynI would produce variants with unique assembly and activity properties. This was addressed by characterising these properties using recombinantly expressed and purified dynamin splice variants. A combination of activity assays and biophysical techniques were used to compare the two dynI isoforms. The data suggested fundamental differences in the activity and assembly of these isoforms. Cell-based endocytosis assays confirmed they«br /» have different endocytic capacities. Quantitation of the level of each isoform in the developing rat brain revealed a remarkable isoform-specific regulation of protein expression, with the dynIa isoform greatly increasing in abundance with the formation of synaptic terminals while the dynIb isoform changed little during development. This supports our hypothesis that dynamin middle domain splice variation produces isoforms that exhibit modified enzyme properties and further suggests that these represent adaptations to specific cellular functions. Second, the relationship between the Mα2 helix and the G domain was explored. This analysis began with direct comparison of previously published crystal structures of each isoform, and a distinct rearrangement of key catalytic loops within the G domain was identified. This predicted changes in the GTP binding affinity of the isoforms, which was experimentally validated using substrate kinetics under a variety of assembly conditions. The data confirmed the predictions made from structural comparison. It also revealed for the first time the presence of allosteric coupling of substrate binding between G domains, which we call 'G domain coupling'. This represents a novel finding for any enzyme within the dynamin family. G domain coupling was responsive to assembly state, a phenomenon we called the 'dynI allosteric switch'. This causes the coupling of G domains to be activated or inactivated according to assembly state. The conditions that activate G domain coupling were different for the dynI isoforms, with coupling activated in the basal state for dynIb, and the helical state for dynIa. This represents the biggest functional difference between these two isoforms. Third, a combination of computational and mutagenesis studies enabled identification of residues that may contribute to allosteric communication within dynamin. Allostery was mediated by an extensive allosteric network that functionally couples the G domain to the PH domain, PRD and middle domains. By analysing the hydrolysis activity of a variety of dynamin isoforms and mutants, the Mα2 helix was identified as a key component of this network via the intermolecular interface 5, which exists between Mα2 and a neighbouring BSE domain within the tetramer building block. The data revealed the existence of an overall assembly signal propagation pathway within both unassembled dynamin tetramers and within oligomers that differentially regulates GTP hydrolysis according to the assembly state, via interface 5. The overall work represents a fundamentally new understanding of the catalytic mechanism of dynamin enzymes. The study is the first to systematically reveal and characterise an allosteric network operating within the classical dynamins. This network regulates a previously uncharacterised form of allostery between dynamin G domains. It therefore has important implications for the coordination of GTP hydrolysis within the helix and therefore, the vesicle scission mechanism of endocytosis. Elements of this network are conserved within the classical dynamins, and this may represent conservation within the broader dynamin family