Structure and membrane remodelling mechanism of the DP1/Reticulon family of ER proteins.

The endoplasmic reticulum (ER) is the largest and morphologically most striking of the organelles found inside eukaryotic cells. The ER is composed of a highly dynamic network of lipid membranes contiguous with the nuclear membrane and extending outward to the cell periphery. The ER is a hub for critical cellular functions, being the site of synthesis for membrane proteins, secreted proteins, and lipids; it is also a store for intracellular calcium, and involved in toxin inactivation. The ER close to the cell nucleus is morphologically composed mostly of sheets that provide a large surface area that is coated with protein-producing components of the cell (ribosomes). Further from the nucleus, the ER contains a large number of tubules that are rapidly and continuously remodelled, i.e. they form and disappear on the timescale of seconds. These distinct ER morphologies are conserved across eukaryotes and how these are formed and stabilised is a fundamental question in cell biology with implications for human health and disease, especially that of neurons, which have a greater reliance on highly curved ER than most other types of cells. The DP1 and reticulon families of proteins are responsible for generating and stabilising the highly curved membranes that are found in the tubular ER and the edges of ER sheets, and mutations in the human DP1 proteins called Receptor Expression Enhancing Proteins (REEPs) cause the motor neuron disease Hereditary Spastic Paraplegia (HSP). DP1 and reticulon proteins share a homologous region that is embedded in the membrane called a reticulon homology domain (RHD), however, no molecular insight into the architecture of the RHD exists, and by extension there is no understanding of how mutations in the RHD can lead to diseases like HSP. The current paradigm for RHD function is that this region of the protein contains helical hairpins that do not fully cross the membrane lipid bilayer from the cytosol to the ER, and in this way crowd the cytosolic half of the membrane to induce curvature. However, we have established an experimental system for studying a representative RHD from yeast and have found that it contains four transmembrane helices long enough to fully traversing the membrane. We have also discovered an amphipathic helix (APH) that interacts strongly with membranes, and removal of this region of the yeast RHD abolishes its ability to curve membranes. APHs are commonly found in proteins involved in membrane remodelling, lending support to the possibility that insertion of this APH into the cytosolic leaflet of the membrane bilayer is the central mechanism by which RHDs curve membranes. We will test this hypothesis by determining the complete three-dimensional structure of the RHD and evaluating its interactions with membranes. We will then introduce HSP-causing mutations and determine how they disrupt the RHD structure and function and contribute to HSP development. Using our structural findings for the DP1 RHD and APH, we will test their generality through targeted studies of the human reticulon family, which are involved in neuronal growth and repair.

DP1 proteins and the related reticulon family members are membrane remodelling proteins essential for maintaining smooth ER tubules and the edges of ER sheets. As a result, mutations in these proteins lead to motor neuron diseases such as hereditary spastic paraplegia (HSP). The current paradigm for how these proteins curve membranes involves helical hairpins within the Reticulon Homology Domain (RHD) that are proposed to only partially insert into the membrane. We have developed an experimental system for testing this model and found that the RHD of Yop1p from S. cerevisiae contains four helices long enough to fully cross the membrane. However, we have discovered an amphipathic helix (APH) in Yop1p that is highly conserved amongst DP1 family members (both in the amino acid properties and its location C-terminal to the RHD) and that is essential for Yop1p function in membrane tubule formation. Thus, while the transmembrane domain is required for RHD function, its exact role requires further investigation. The experimental system developed for Yop1p enables determination of the first 3D structure of an RHD and thus testing of structure-function relationships, which are essential for understanding the mechanism by which mutations in highly conserved amino acid positions of human DP1s such as REEP1 cause HSP. The more complete understanding of the Yop1p RHD secondary structure has also enabled a more reliable sequence alignment with the reticulons, suggesting that the RHDs have a similar transmembrane architecture, and potentially also contain an amphipathic helix C-terminal to the RHD. Thus, we will test whether the human DP1s and human reticulons contain amphipathic helices that interact with membranes, and whether this region is required for the human REEPs that are involved in HSP.

The general areas of impact are to (i) determine the molecular bases of some forms of the motor neuron disease hereditary spastic paraplegia, (ii) increase biophysical and biological understanding of the morphology and biogenesis of curved membranes (including ER tubules, the edges of ER sheets, and the nuclear membrane), (iii) provide the first structure of a unique class of eukaryotic membrane protein, and (iv) provide training in the study of structure-function correlations in membrane proteins using biophysical, biochemical and biological methods.

Potential beneficiaries for this work include: (i) Academic researchers studying membrane curvature and remodelling, organelle structure and motor neuron diseases, (ii) pharmaceutical companies wishing to exploit this knowledge to treat motor neuron diseases, (iii) health professionals (NHS) and charities (for example, the HSP Support Group) who counsel and provide information on the underlying causes of motor neuron diseases, and (iv) the public, who ultimately will benefit from the future treatments of motor neuron disease this research aims to advance.

The primary mechanism for communication of this research will be through publication in open access, peer reviewed, international journals. We will liaise at the time of publication with the University of Oxford and MRC press offices to ensure wider dissemination of results to the general public. We will also take advantage of opportunities to communicate via freely accessible media (such as the University of Oxford Department of Biochemistry website) in order to extend the impact of the findings. Our results will also be made available on the Schnell laboratory web sites (www.bioch.ox.ac.uk/aspsite/index.asp?pageid=651 and www.mpi-lab.com). Data obtained during this project, such as structural models and NMR chemical shift assignments will be deposited into the open access Protein Data Bank and BioMagResBank databases, respectively, which are routinely used in meta-analyses to advance basic understandings in structural biology.

The PDRAs employed on this grant will gain technical skills in manipulation of protein and lipid samples and their characterisation using the latest advances in NMR spectroscopy, biophysical methods, and in cell fluorescence microscopy. In addition the PDRAs will be trained in writing, IT, and presentational skills, and will benefit from working closely with expert colleagues using different but complementary techniques, thereby enhancing their future research employment prospects.

We note also that the experimental system that is developed for Yop1p is likely to be of interest to a wider community of biologists. While X-ray crystallography remains the tool of choice for high resolution structural studies, NMR spectroscopy can provide atomic level information on systems that do not readily crystallise, and can provide site-specific data on structure, dynamics, and interactions in the absence of a structural model. Polydisperse oligomerisation, such as that found in the reticulon/DP1 membrane proteins, compounds the usual difficulties in forming crystal lattice contacts for proteins solubilised in dynamic and sometimes heterogeneous surfactants. However, solution NMR can tolerate systems in which multiple conformational or oligomeric states co-exist, as we have recently demonstrated for a membrane protein oligomer with variable stoichiometry (Rodriguez et al., 2013). Thus, the methodologies developed through the study of Yop1p are likely to be applicable to a wide range of systems that are inaccessible to structural studies by other methods.