Flattening out: A new ESCRT structure in cell adhesions

Endosomal Sorting Complexes in Retrograde Transport (ESCRTs) were discovered over the last 20 years (1, 2). Among the four known complexes (ESCRT-0, -I, -II, -III), ESCRT-III has gained the most attention, as it was discovered to be involved in essentially all membrane functions where the membrane is fissioned with proteins acting from within the neck. ESCRT-III also functions during membrane repair of both the plasma membrane and of the lysosomal membrane, making it the most ubiquitous membrane remodeling machinery and one of the most important for cell survival. The best described molecular function of ESCRT-III is the formation of intra-lumenal vesicles (ILVs), where it both deforms and fissions the membrane to bud tiny vesicles inward into the endosome lumen (3). Two characteristic features of this process are its coupling with sorting of transmembrane cargoes within the bud and the exclusion of ESCRT molecules themselves from the bud.

Here, Stempels et al. (4) report the formation of unusually large, flat ESCRT-III structures at the surface of the plasma membrane in macrophages and dendritic cells. These structures are formed from large round-shaped protein assemblies, sometimes tightly bound together, sometimes more spread apart, and contain the ESCRT-III subunit IST1. By far, these structures are the largest, the most stable, and the flattest structures of ESCRTs found in cells to date. They seem to be specific to cells that can infiltrate into dense tissues, in particular antigen gathering or antigen presenting immune cells. A question remains how such large structures have been missed so far. Other ESCRT proteins, including ALIX, CHMP1, CHMP4, and VPS4, fully or partially colocalize with IST1 in these structures. Overall, even though the scale is different, they look similar to the flat carpets of ESCRT-III spirals observed in vivo (5, 6) and in vitro (7).

The most striking pattern associated with these ESCRT-III assemblies are the transmembrane cargoes: tetraspanins, integrins, and several other cargoes including ubiquitinated proteins, localize in highly dense foci at the center of the spiral-like structures. These transmembrane proteins are then left behind together with ESCRT-III spirals during cell migration, which is surprising since integrins are normally recycled during cell migration through clathrin-dependent and independent pathways (8). Interestingly, these ESCRT-III-linked adhesion sites are not connected to the actin cytoskeleton. Recent studies have identified other integrin adhesions that are also actin independent, namely reticular adhesions (9). These contain specific integrins, flat AP2-rich clathrin lattices, and are connected to specific ECM proteins (10). Whereas the molecular composition of canonical, actin-associated focal adhesion is well-studied (11), the molecular composition and ultrastructure of reticular adhesions and novel ESCRT-III-positive integrin adhesions remain largely a mystery. Furthermore, whether these ESCRT-III-positive integrin adhesions are derived from focal adhesions or reticular adhesions remains unknown. It is also unclear if IST1 spirals appear only within specific cell types, for a specific set of integrins, or in contact with specific ECM molecules. Nevertheless, Stempels et al. show that worm-like ESCRT-III-structures surround non-recycled adhesion sites. As discussed later, ESCRT-III spirals could be an indicator of membrane damage at the integrin adhesion sites.

While a sequence of subunit recruitment probably drives membrane deformation, the link to cargo sorting remains obscure (12). The early “lasso” model of sorting suggested that the spirals would initially form a large corral around the cargo molecules, which would get condensed by spiral growth (13). The striking localization of cargoes in the center of the large ESCRT structures supports this idea. However, the authors also show that the ESCRT-III structures form rapidly after induction, while the cargo proteins in the center of the spiral condense several tens of minutes after ESCRT-III assembly.

Another classic idea about the ESCRT-III machinery is that the cargoes are sorted within the bud, but the ESCRT machinery remains in the cytoplasm to be recycled. Stempels et al. show that the adhering structures, and ESCRT spirals along them, can be “lost” by cells when the adhering protrusion they belong to is cut off from the cell body. Would this mean that the ESCRT-III machinery, at least in some functions, could escape recycling and get destroyed during the process it catalyzes? In agreement, Alix, a nucleator of ESCRT-III, is found in exosomes, which come from ILVs liberated by back-fusion of endosomes with the plasma membrane (3). This suggests that some ESCRT subunits are present in ILVs and thus degraded.

The most important question remains: what is the function of these large worm-like structures? The authors indicate that they are part of adhesive structures, and that reducing actin polymerization increases their size, while promoting actin-driven motility reduces their extent. These structures seemed to be promoted by plasma membrane damaging agents, which could mean that they are involved in membrane repair during integrin adhesion release from the cell body. However, they could also hold a completely new function, specific to immune cells, as these gigantic ESCRT-III spirals have not been observed previously despite numerous studies using other cell types. It is not difficult to think of the immune synapse when seeing these ESCRT structures as they are large adhesive structures in between cells capable of concentrating cell–cell adhesion molecules and signaling transmembrane proteins (14). But which immune function could these structures participate in? Could these structures participate in sensing cells or damaged membrane while immune cells are crawling through the tissue?

Overall, the discovery of these large ESCRT-III structures opens a new area of research and provide a new perspective on ESCRTs. In vitro, most of the ESCRT structures observed so far were flat or tubular, while in vivo, most of the structures observed were conical, although this is not an absolute separation. But the fact that large assemblies of ESCRT-III molecules are found in vivo in some cell types, most likely with a specific function, strengthens the point that flat ESCRT assemblies that are often seen in vitro could have some physiological relevance. Moreover, beside the beauty of the images obtained by Stempels et al. (4), they are large and stable enough so that many of the usual microscopy assays can provide clear results. Fascinatingly, they could also be an optimal target for in situ cryoET studies. The abscission ESCRT structure, such as found in nuclear envelope reformation or at the endosomal membrane, are too small and too short-lived to gain enough information on possible differences of localization and dynamics of different subunits. This has limited the progress of understanding the mechanisms by which ESCRT-III molecules perform a large variety of their functions in cells. With these novel integrin-associated structures, the hope is that patterns of colocalization of different subunits and ESCRT-III partners, but also dynamics of recruitment and disassembly, will help the understanding of ESCRT functions. Finally, this study will open new directions for integrin recycling and more broadly cell migration studies that will shed light on these key cellular functions.