Explore the vast range of titles available.

Compartmentalization of cellular material in droplet-like structures is a hallmark of liquid–liquid phase separation 1 , 2 , but the mechanisms of droplet removal are poorly understood. Evidence suggests that droplets can be degraded by autophagy 3 , 4 , a highly conserved degradation system in which membrane sheets bend to isolate portions of the cytoplasm within double-membrane autophagosomes 5 – 7 . Here we examine how autophagosomes sequester droplets that contain the protein p62 (also known as SQSTM1) in living cells, and demonstrate that double-membrane, autophagosome-like vesicles form at the surface of protein-free droplets in vitro through partial wetting. A minimal physical model shows that droplet surface tension supports the formation of membrane sheets. The model also predicts that bending sheets either divide droplets for piecemeal sequestration or sequester entire droplets. We find that autophagosomal sequestration is robust to variations in the droplet-sheet adhesion strength. However, the two sides of partially wetted sheets are exposed to different environments, which can determine the bending direction of autophagosomal sheets. Our discovery of this interplay between the material properties of droplets and membrane sheets enables us to elucidate the mechanisms that underpin droplet autophagy, or ‘fluidophagy’. Furthermore, we uncover a switching mechanism that allows droplets to act as liquid assembly platforms for cytosol-degrading autophagosomes 8 or as specific autophagy substrates 9 – 11 . We propose that droplet-mediated autophagy represents a previously undescribed class of processes that are driven by elastocapillarity, highlighting the importance of wetting in cytosolic organization. A theoretical model, in vitro reconstitution and in vivo experimentation show that competition between droplet surface tension and membrane sheet instability dictates the form and function of autophagosomal membranes.

The ESCRT machinery mediates reverse membrane scission. By quantitative fluorescence lattice light-sheet microscopy, we have shown that ESCRT-III subunits polymerize rapidly on yeast endosomes, together with the recruitment of at least two Vps4 hexamers. During their 3–45 s lifetimes, the ESCRT-III assemblies accumulated 75–200 Snf7 and 15–50 Vps24 molecules. Productive budding events required at least two additional Vps4 hexamers. Membrane budding was associated with continuous, stochastic exchange of Vps4 and ESCRT-III components, rather than steady growth of fixed assemblies, and depended on Vps4 ATPase activity. An all-or-none step led to final release of ESCRT-III and Vps4. Tomographic electron microscopy demonstrated that acute disruption of Vps4 recruitment stalled membrane budding. We propose a model in which multiple Vps4 hexamers (four or more) draw together several ESCRT-III filaments. This process induces cargo crowding and inward membrane buckling, followed by constriction of the nascent bud neck and ultimately ILV generation by vesicle fission.

Compartmentalization of cellular material in droplet-like structures is a hallmark of liquid-liquid phase separation1,2, but the mechanisms of droplet removal are poorly understood. Evidence suggests that droplets can be degraded by autophagy3,4, a highly conserved degradation system in which membrane sheets bend to isolate portions of the cytoplasm within double-membrane autophagosomes5-7. Here we examine how autophagosomes sequester droplets that contain the protein p62 (also known as SQSTM1) in living cells, and demonstrate that double-membrane, autophagosome-like vesicles form at the surface of protein-free droplets in vitro through partial wetting. A minimal physical model shows that droplet surface tension supports the formation of membrane sheets. The model also predicts that bending sheets either divide droplets for piecemeal sequestration or sequester entire droplets. We find that autophagosomal sequestration is robust to variations in the droplet-sheet adhesion strength. However, the two sides of partially wetted sheets are exposed to different environments, which can determine the bending direction of autophagosomal sheets. Our discovery of this interplay between the material properties of droplets and membrane sheets enables us to elucidate the mechanisms that underpin droplet autophagy, or 'fluidophagy'. Furthermore, we uncover a switching mechanism that allows droplets to act as liquid assembly platforms for cytosol-degrading autophagosomes8 or as specific autophagy substrates9-11. We propose that droplet-mediated autophagy represents a previously undescribed class of processes that are driven by elastocapillarity, highlighting the importance of wetting in cytosolic organization.

We, the authors, have withdrawn this manuscript. During our work on the revision of this manuscript for eLife, it became clear that several key results of the manuscript were not reproducible by my lab. The rules for ESCRT-III assembly and the model that we have proposed in this manuscript are no longer supported by the data. Currently we are performing additional experiments to test some of our conclusions further. Therefore, we do not wish this work to be cited. We sincerely apologize to the scientific community for any inconvenience resulting from this manuscript. If you have any questions, please contact the corresponding author. Competing Interest Statement The authors have declared no competing interest.