NDKB and Atlastin Structure Endoplasmic Reticulum Membranes
Of the membrane bound organelles in eukaryotic cells, the endoplasmic reticulum (ER) may be the most complex. It is the largest both in terms of surface area and volume. It includes several subdomains: the nuclear envelope (NE) as well as an extensive network of both highly inter-connected fenestrated tubular membranes and flat cisternal sheets (1, 2). While the structure and organization of the ER is thought to be important for the execution of a myriad of essential cellular functions including protein and lipid synthesis and export as well as calcium sequestration and drug metabolism (3), how this membrane system is generated and maintained despite continuous turnover is only just beginning to be unraveled (4). The microtubule cytoskeleton and molecular motors are clearly important for the extension of tubules from the existing network so that they may fuse with nearby ER tubules to generate new three-way junctions (5-8). However, an ER-like network can be generated in vitro in the absence of microtubules (9), which suggests the existence of additional mechanisms for the extension and scaffolding of the ER network. Once the tubular ER does extend out from the existing network the next critical step is fusion. Soluble NSF attachment-protein receptors (SNAREs) were an obvious candidate for this role (10), but ER homotypic fusion events have not yet been found to depend on SNAREs. Recently, a member of the dynamin super-family of large GTPases, atlastin, was implicated in ER homotypic fusion. An in vitro fusion assay (11) and knockdown experiments (12) in conjunction with the crystal structures of the soluble domains of atlastin (13, 14) have led to a possible mechanism of ER fusion, but this model remains to be tested. In my thesis I will describe two projects. One focuses on the role and mechanism of nucleoside diphosphate kinase B (NDKB) in ER network extension and stabilization. The other focuses on the role and mechanism of atlastin in fusing ER membranes.
NDKB was initially implicated as a stimulator of ER export in permeabilized cells (15). Subsequent work suggested that its effect on ER export might be through an effect on ER network morphology. Through in vitro assays, we found that NDKB not only stabilizes the ER network but also actively promotes ER network extension. In order to perform this function we hypothesized that it might interact directly with ER membranes. Indeed, we found a pair of positively charged residues that mediated direct binding of NDKB to anionic phospholipids. When these residues ware mutated to negatively charged residues, NDKB no longer bound anionic phospholipids and failed to mediate ER extension in our semi-intact cell assay. In order to gain insight into the mechanism for how NDKB might be performing its ER network extension function we took another in vitro approach. Anionic synthetic liposomes were incubated with NDKB and we found that NDKB was able to arrange these liposomes into large arrays that resembled the ER network. Together these results implicate NDKB and anionic phospholipids in a role for ER network morphology, in particular as a means to stabilize and extend the ER network.
We initially became interested in the atlastin GTPase as a result of an ER overexpression phenotype observed by the Blackstone lab indicating a potential role in ER morphogenesis (16). In strong support for a required role for atlastin in ER structuring, we found that siRNA depletion of atlastin from HeLa cells resulted in a reduction in network density which could be rescued by the addition of an siRNA immune atlastin transgene. This established a structure function assay we could use to dissect the functional domains of atlastin. Concurrent with our identification of key residues required for atlastin function, it was observed by another lab that atlastin could fuse synthetic liposomes (11), suggesting that the ER structuring role we had observed for atlastin might correspond to the membrane fusion step. Simultaneously, structure determinations for the soluble domain of atlastin were reported (13, 14). Together, the collective data suggested the following model for atlastin: GTP dependent dimerization of atlastin leads to tethering and subsequent GTP hydrolysis leads to a large conformational change that drives membrane fusion (13, 14). To test the model, we exploited our identification of a required salt bridge central to the large conformational change proposed to convert GTP-bound tethered atlastin dimers into a postfusion state. We established that although blocking the salt bridge had no effect on GTP binding and hydrolysis, it abolished stable atlastin dimer formation. Then, through a series of crosslinking assays probing the conformational state of the atlastin soluble domain, we showed that atlastin adopts the postfusion conformation in the GTP bound state, without the need for GTP hydrolysis. As a result of our studies, we have modified the current model for atlastin’s function. In our revised model, GTP binding is required for atlastin’s initial dimerization and begins a cascade of conformational changes that results in the large rearrangement thought to drive membrane fusion. We speculate based on our work that hydrolysis may be necessary to complete the fusion cycle and/or function to disassemble the postfusion complex for multiple rounds of fusion.
In summary, these studies provide both an initial analysis of a protein with an ER network extension role and important insights into the mechanism of ER membrane fusion. It is hoped that this work will add to our understanding of the biogenesis and maintenance of the ER network.