PerspectiveCell Biology

No ESCRTs for Exosomes

See allHide authors and affiliations

Science  29 Feb 2008:
Vol. 319, Issue 5867, pp. 1191-1192
DOI: 10.1126/science.1155750

Exosomes are small (50 to 100 nm in diameter) membrane-bound vesicles released by a variety of cells. Originally proposed to discard excess transferrin receptor from reticulocytes during red blood cell formation (1), exosomes are now thought to play key roles in cell-to-cell communication, antigen presentation, and in the pathogenesis of retroviral infections (including HIV) and prion diseases (24). However, how exosomes are formed has not been clear. On page 1244 in this issue, Trajkovic et al. (5) provide intriguing insights into exosome formation, making these microvesicles a bit less mysterious but raising many new questions about their biogenesis.

Sorting it out.

Proteolipid protein-containing exosomes are formed by inward invagination of endosomal membranes. This requires ceramide generation on the cytosolic side by neutral sphingomyelinase 2 (nSMase2) (15). Other membrane proteins, such as the epidermal growth factor (EGF) receptor, that are sorted to intraluminal vesicles depend on ESCRT proteins instead.


An early view was that exosomes are formed by invagination of the membrane of endosomes (see the figure) to produce intraluminal vesicles, thus rendering these organelles multivesicular bodies (6, 7). Exosomes are then secreted when these multivesicular bodies fuse with the plasma membrane and release their content (6). A more recent view holds that exosomes can also form at the plasma membrane in some cell types (8). The link between exosomes and multivesicular bodies was strengthened by the discovery of the ESCRT (endosomal sorting complex required for transport) machinery (9). This highly conserved set of protein complexes recognizes membrane proteins that are modified with ubiquitin molecules and thus marked for sorting to lysosomes (either as functional components of lysosomes or as substrates for lysosomal proteolysis). ESCRT complexes sort these cargoes to specific domains of endosomes and regulate both the inward invagination of these membrane regions and the scission of invaginated membrane buds to form intraluminal vesicles. Cells that lack components of the ESCRT machinery often have fewer multivesicular bodies or fewer intraluminal vesicles in multivesicular bodies, and fail to deliver cargo to lysosomes (7). The ESCRT machinery is also required to complete the topologically related (budding of membrane vesicles away from the cytoplasm) assembly of various enveloped RNA viruses (10), including HIV, and to mediate the abscission reactions that complete mammalian cell division (11).

It was thus not unreasonable to conjecture that the ESCRT machinery would also be involved in the similar process of exosome formation. Indeed, an ESCRT-associated protein (AIP1/Alix) interacts with transferrin receptors during exosome formation in reticulocytes (12). ESCRT proteins are also recruited to proposed sites of exosome formation in lymphocytes (8) and are found in exosomes (3, 5). A regulatory role in intraluminal vesicle formation was also suggested based on the inhibition of inward budding in liposomes by the ESCRT protein Alix (13). Nevertheless, the role of the ESCRT machinery in exosome formation has remained unclear.

Trajkovic et al. used an oligodendrocyte cell line (myelinating cells of the central nervous system) to study the formation and release of exosomes containing proteolipid protein. Proteolipid protein is a major component of myelin, the lipid-rich membrane that oligodendrocytes use to enwrap and insulate axons. They find that formation of proteolipid protein-containing exosomes does not require ESCRT machinery. By contrast, sorting of the epidermal growth factor receptor to lysosomes in these cells is inhibited by depletion of ESCRT components or expression of a dominant-negative form of an ESCRT protein (Vps4).

Morphological analysis of the oligodendrocyte endosomes shows that proteolipid protein segregates into membrane domains that are distinct from domains containing cargo destined for ESCRT-mediated sorting to lysosomes. Trajkovic et al. show through mass spectrometric analysis that secreted proteolipid protein-containing exosomes purified from cell culture medium are enriched in ceramide, a lipid produced from the membrane lipid sphingomyelin by sphingomyelinases. Disrupting the expression of neutral sphingomyelinase 2 (nSMase2) by RNA interference or the use of specific inhibitors reduced secretion of proteolipid protein-containing exosomes. Moreover, when Trajkovic et al. added a bacterial sphingomyelinase to liposomes containing domains with different degrees of fluidity, budding occurred specifically from the “raft”-like lipid phase. This led them to suggest that ceramide-induced aggregation of lipid microdomains leads to domaininduced inward budding of intraluminal vesicles, perhaps promoted by the cone-shaped structure of ceramide (see the figure).

The observations of Trajkovic et al. raise several questions. Morphological experiments indicate that both proteolipid protein-containing exosomes and epidermal growth factor receptor-containing intraluminal vesicles can be formed within the same endosome. Is this the case, or are there functionally distinct populations of endosomes that generate different intraluminal vesicles (14)? Also, if both types of vesicles are present in the same multivesicular body, they must somehow be sorted to ensure that only the exosomes are secreted. It is also not clear whether all exosomes are formed through the same molecular mechanism, or if different mechanisms are used for different types of exosome cargo. The Trajkovic et al. study shows that secretion of the tetraspanin CD63, another exosome-associated membrane protein, is also blocked by a sphingomyelinase inhibitor, but not by a dominant-negative ESCRT component. If the ceramide-based process is the primary mechanism for exosome formation, it would seem that ESCRTdependent enveloped viruses have usurped the lysosomal sorting and abscission machinery for assembly, though the idea that the budding of some viruses involves raft domains could also indicate that a combination of the two processes is used.

The presence of ceramide in exosomes may imply its direct role in the lipid-phase organization of the endosomal membrane, whereby the ceramide-enriched phase ends up in the budding vesicle. This is supported by the presence of proteolipid protein—a typical membrane raft component—in exosomes. However, without knowing the lipid composition of the endosomal membrane, one cannot conclude that exosomes originate from a specific membrane domain. Also, without knowing the transbilayer organization or ceramide concentration in the endosomal membrane, the extrapolation of model membrane experiments remains problematic. Whatever the molecular mechanism by which a change in lipid composition drives vesicle budding, the process is likely to be regulated. The present work suggests that a better understanding of lipid metabolism may provide new vistas in exosome research.

References and Notes

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
View Abstract

Navigate This Article