In the studies published in Nature Communications and Journal of Cell Science respectively, HU Junjie's group at the Institute of Biophysics of the Chinese Academy of Sciences systematically analyzed the molecular mechanism by which endoplasmic reticulum (ER) "tubule-forming proteins" induce membrane curvatures, as well as the structural basis for the regulation of cisternal thickness by ER luminal "bridging proteins."

The ER is an important organelle in eukaryotic cells responsible for important processes such as protein synthesis, lipid synthesis, and calcium storage. Morphologically, it is a continuous membrane system composed of a network of tubules and sheets.

The formation of tubular ER relies on the integral membrane proteins of the reticulon (RTN) and receptor expression-enhancing protein (REEP) families, which induce membrane curvature at the cross-section of tubules and the edges of sheets. The thickness of sheets is primarily determined by Climp63. The shaping of the ER is directly related to its physiological functions. However, these shaping proteins tend to be multimeric and heterogeneous, making it difficult to understand the molecular mechanism.

Using recently developed deep learning-based structure prediction methods, researchers obtained the conformation of the yeast REEP Yop1p, and found that elements such as the transmembrane regions are inter-locked in relatively stable positions within the molecule. They deciphered the combination of multiple dimerization interfaces of Yop1p through chemical cross-linking and three-dimensional printing model assembly, and discovered that Yop1p multimers have a curved shape that matches ER tubules. They also found that REEP1 is associated with hereditary spastic paraplegia, and its multiple mutations are directly linked to the disruption of REEP1 polymerization, demonstrating the importance of the oligomeric assembly of tubule-forming proteins.

Then, researchers applied a similar strategy to predict the structure of Climp63, and found that the luminal domain of Climp63 is primarily helical, with differences in multiple prediction models showing a relatively weak assembly of the helical bundles. They discovered that in one of the conformations, when forming an end-to-end dimer, it has a length consistent with the ER sheet thickness observed in cells. Cross-linking experiments confirmed the existence of this dimerization interface. However, the surface of the intraluminal structural domain of Climp63 is enriched with charged clusters, which may allow for various modes of same-side or opposite-side polymerization. Additionally, the regulation of same-side polymerization of Climp63 was found to be coupled with opposite-side polymerization.

Overall, these works suggested that the shaping of biological membranes typically relies on the regulation of the oligomerization of shaping proteins, which often integrate multiple weak homotypic interactions. This regulatory mode makes the shaping mechanism highly adjustable and flexible. The same theory also applies to the establishment of membrane contact sites in organelle interactions.