Molecular simulations of lipid droplet biogenesis

Overview & Synopsis

Lipid droplets (LDs) are ubiquitous cellular organelles regulating the lipid storage and metabolism. LDs are generated in the endoplasmic reticulum (ER), and the mechanism of their biogenesis is not understood because extremely challenging to characterize experimentally, due to the small size and dynamic nature of nascent LDs. Malfunctioning of LD biogenesis is linked to a wide range of diseases. Understanding the formation of LD is the main objective of this thesis. I focused on three aspects: (1) the partitioning of proteins in nascent LDs, (2) the analysis of LD shape as a proxy for their tendency to bud, and (3) the mechanism of budding. Proteins play an important role in the birth and function of organelles. The respective proteome of LDs and ER are different, despite their membranes being connected, and the factors determining protein distribution are unclear. We used molecular dynamics simulations to study the partitioning of ER proteins, focusing on model proteins to mimic transmembrane and monotopic membrane proteins. We found that the chemical properties of proteins affect their distribution around nascent LDs and explain the different proteosomes of both organelle. During their formation, LDs undergo a drastic change in shape, from the lens-like shape of nascent LD embedded in a bilayer membrane to nearly spherical LDs emerging from the ER. Theoretical work allowed a mathematical description of the shape as a function of LD volume, and predicted that the change in shape, corresponding to budding, occurs when the LD grows beyond a certain threshold, estimated at a few tens of nm. However, no change in shape and no budding was observed in simulations of nascent LDs embedded in flat, periodic lipid bilayer membranes. Comparing the shape of simulated LDs with the theoretically predicted shapes, we found very good agreement with the theory, indicating that simulations can be used to gain molecular level insight into the properties and dynamics of these systems. LD budding takes place on a time and length scale beyond the reach for current microscopy techniques. Also, it is commonly believed that budding is out of reach for simulations too, due to the same limitations (in the opposite sense). Budding is therefore the least understood part of lipid droplet biogenesis. It is also possibly the most important, as the direction of budding determines the protein coverage and therefore the function of LDs in cells. Here we developed a non-equilibrium simulation method, coined POP-MD, to effectively emulate lipid synthesis and explore the driving forces and the conditions for LD budding. We also modeled the key protein in LD biogenesis, seipin, whose multiple roles are not completely clear, and inserted it tubular bilayer membranes, featuring realistic morphology, size, and chemical composition of the tubular part of the ER. POP-MD allowed us to simulate, for the first time, the budding process, and determine which conditions are actually necessary to observe it. We found that LD size is not a critical parameter to reproduce budding; instead, leaflet asymmetry is necessary and sufficient to reproduce budding, independently of all other factors. Coupling of triglyceride and phospholipid synthesis is necessary to produce correctly budded structures. Seipin scaffolds the neck of the budded LD and prevents the generation of hydrophilic defects at the LD neck, but it is not sufficient to produce budding. The complex ER lipid composition facilitates the budding mechanism as it provides softer membranes, reducing membrane defects during biogenesis. Overall, simulations provided an unprecedented molecular view of LD biogenesis, with sub-nanometer length resolution and picosecond time resolution, enabled the interpretation of numerous experimental data, and elucidated the role of the different components required for the budding process.