Membrane trafficking: ESCRTs Cut Some Slack

In this issue of Nature Cell Biology, Mercier et al show that acute changes in membrane tension may be a physiological trigger for ESCRT assembly, which drives membrane scission, luminal vesicle budding, and a wide array of other membrane remodeling events throughout the cell.

The ESCRT apparatus is involved in myriad membrane remodeling events that require membrane deformation and resealing¹. Initially discovered for their role in the formation of intralumenal vesicles (ILVs) that bud into the interior of early sorting endosomes, ESCRTs perform similar tasks in budding of viruses and extracellular vesicles from the cell surface and repair of ruptured membranes at the nucleus, lysosome, and plasma membrane¹. The biochemistry that all of these events share is mediated by part of the ESCRT apparatus, ESCRT-III, an ancient protein polymer conserved throughout eukaryotes and also found in^{1, 2}. ESCRT-III is an assemblage of various small subunits that can polymerize on the surface of membranes and mediate scission underneath them without themselves being consumed in the ensuing sealing or vesicle budding process². ESCRT-III is coopted into its different functions by adaptors such as ESCRT-II or ALIX that position the nucleation of the polymer at events such as ubiquitinated cargo sorting, viral particle formation, and cytokinesis. ESCRT-II and ALIX in turn are targeted to these processes by upstream factors such as ESCRT-0 for ubiquitinated proteins, viral gag proteins for viral budding, and CEP55 at the midbody to complete cytokinesis¹. Depolymerization of ESCRT-III is performed by the AAA-ATPases Vps4 and Spastin, that can dismantle the entire polymer but likely also edit the polymer as it grows to ensure it has the proper subunit combinations and shape transformations². The ESCRT-III proteins use a basic-charged surface together with a membrane insertion motif to adhere to the membrane surface as subunits interlock to form their larger polymer. These surfaces are obscured when individual subunits reside in the cytosol in a closed conformation until their open conformation can be stabilized by binding to a nucleator or the growing end of the polymer. What Mercier et al. show in this issue of Nature Cell Biology, is that cells might utilize membrane tension as a biological mechanism to drive ESCRT assembly and function³.

A number of in vitro and in vivo experiments have shown that ESCRT-III polymers comprised of different subunits can distort the membrane in a variety of ways that produce tubules, negative curvature, or positive curvature, providing the capacity of ESCRT-III to foster different membrane modeling events depending on where it is deployed^{1, 2}. Just as ESCRT polymers help bend membranes, bent membranes have been shown to promote ESCRT-III polymerization as would be expected from a membrane curvature ‘sensor’^4–6. The question that arises from the connection between polymerization and membrane deformation is whether membrane deformation itself could be a physiological driver for ESCRT assembly and function in vivo. And if so, could that property be somehow regulated to exert finer control over when and where ESCRT-III polymers are employed? Mercier et al provide intriguing data to suggest the answers are Yes. Here, the authors use the simple yet elegant treatment of cells with hypertonic solutions to transiently deflate internal membranes of water and observe massive accumulation of ESCRTs onto endosomal membranes. That slack in endosomal membranes is actually reduced was confirmed with measuring endosomal volume as well as with a probe, FliptR⁷], that responds to changes in membrane tension with changes in its fluorescence lifetime. The specificity of assembly onto endosomal membranes versus those of other organelles or the wrinkled plasma membrane is explained by the observation that ALIX was required for ESCRT-III accumulation, consistent with ALIX being a nucleator for assembly and occupying endosomal surfaces to begin with¹. Interestingly ALIX as well as VPS4 were co-recruited to deflated endosomes as well, implying that either ALIX is first recruited as the rate-limiting nucleator by honing in a particular property of flaccid endosomes, or follows massive ESCRT-III recruitment to perform additional functions with VPS4 that might help remodel the forming polymer deployed⁸. Importantly, the assembly of ESCRT-III onto deflated endosomes is not likely due solely to preceding recruitment of ALIX because the authors confirm with in vitro polymerization assays that the CHMP4b (Snf7) ESCRT-III subunit polymerizes onto the surface Giant Unilamellar Vesicle membranes much faster when membrane tension is reduced by hypertonicity or physical manipulation of the GUV. What is especially satisfying regarding the effect of hypertonicity on endosomal membranes is that the hyper-recruitment of ESCRT-III correlates with increased incorporation of EGFR into intralumenal vesicles showing that accelerated ESCRT assembly is productive and relevant to ILV formation.

Could reduced membrane tension itself be a driver for ESCRT assembly? One characteristic of Hela cells is that stimulation of EGFR with EGF not only leads to EGFR incorporation into ILVs but that the number of ILVs produced is dramatically increased⁹. The study by Mercier et al. show with the FliptR membrane-tension probe that the endosomal membrane indeed loosens up during EGF-stimulated ILV production³. The converse has also been observed in previous studies that generated swollen sucrose-filled endosomes (‘sucrosomes’) that retain high osmotic pressure, which were unable to form of intralumenal vesicles¹⁰. This leads to the critical question of how this physical mechanism might be exploited by cells to regulate proper ESCRT-III function. The present study provides a nice example using a peptide that perforates lysososomes. Previous studies have shown that ESCRTs are among the first responders to repair lysosome punctures¹¹. Mercier et al observe the same pattern of ESCRT-III assembly and ALIX and VPS4 recruitment onto ruptured lysosomes that demonstrably lost their membrane tension that was also observed on hypertonically deflated endosomes. This provides a simple mechanism whereby the punctured membrane itself provides the physical basis for ESCRT-III to assemble and repair. Aside from this dramatic example of organelle rupture which might be a rare event for cells, regulated volume/tension changes might be harnessed for more constitutive purposes such as intralumenal vesicle formation at endosomes. One of the defining characteristics of early sorting endosomes where the process of ILV formation begins is that they are large, electron lucent, and for the most part quite round - inflated due to osmotic pressure maintained in part by proper ion balance in¹². The V-ATPase, which would be a the central source of energy to power such gradients is required to maintain inflated endosomes¹³. Moreover, perturbing the V-ATPase and other ions can inhibit ESCRT-dependent ILV formation in vitro¹⁴. Certainly the decreasing pH of endosomes as they mature into multivesicular late endosomes by forming more ILVs, which relies on both V-ATPase activity and other ion currents, is one of the defining features of the endocytic system. A prediction from this idea is that an endosome-localized ion transport system should be functionally linked in ILV formation. That prediction is borne out by the Na/H+ exchanger whose role in ILV formation and sorting into MVBs has been noted for years since it was found to share many of the same phenotypes with ESCRT mutants. Moreover, proper sorting of MVB cargos into ILVs lacking the Na/H+ exchanger can be corrected by manipulating the ionic environment of endosomes¹⁵. One of the intriguing aspects of the possibility of regulating endosome volume and function with ions and in particular the Na/H+ exchanger and V-ATPase is that their activities can be acutely regulated and synchronized with the overall flux through the endocytic pathway. The results of Mercier et al. might now provide a rationale for how ion gradients and the osmotic pressure and membrane tension they may cause can provide important process-driving physical properties on endosomes.

Figure 1:

ESCRT-III drives many membrane modeling events throughout the cell, including the completion and scission of intralumenal vesicles that bud from the limiting membrane of early endosomes, pictured here. Initially, cargo proteins such as ubiquitinated membrane proteins destined for lysosomal degradation, are clustered in subregions on the endosome surface. This sorting is facilitated by the action of early ESCRTs (ESCRT-0/I/II or ALIX/Bro1/HD-PTP) and organized by a clathrin network. Whereas early sorting endosomes are inflated to high membrane tension, local and transient reduction in membrane tension could provide a membrane that is easier to deform and conducive for productive polymerization of ESCRT-III subunits for the final stages of intralumenal vesicle formation. Potential mechanisms for reducing membrane tension could be altering the osmotic pressure within endosomes by regulating the activity of ion transporters such as the V-ATPase, the Na+/H+ exchangers, and chloride channels. Alternatively, membrane shape may be constricted by clathrin-coated endosomal subdomains that may re-arrange or disperse to reduce membrane tension nearby.