Membrane transport

The making of a vesicle

Article metrics

The transport of molecules from one cellular compartment to another often requires membrane-bounded carriers. New work gives insight into how the shaping of membrane into such vesicles is linked to the selection of cargo.

If you could look inside your cells you would see several different compartments, delimited by membranes composed of lipids and proteins. You would notice that the compartments are highly dynamic and undergo a series of changes in morphology, allowing small, sack-like structures to bud off from time to time. These membrane-bounded carriers can differ in molecular composition, size and shape, and shuttle cargo from one compartment to another. One of the biggest challenges in cell biology is to unravel the molecular events that shape these vesicles emerging from a membrane, and that couple this process to the selection of cargo. On page 361 of this issue, Ford and co-workers1 propose a mechanism of action for proteins known as epsins, and suggest that they bridge the gap between cargo selection and vesicle shaping.

Among the several types of vesicle that bud off from the plasma membrane — the membrane that surrounds the cell — some can be seen to be decorated by an electron-dense meshwork when visualized at high resolution, for example by electron microscopy. This meshwork is referred to as the membrane coat and is mainly involved in shaping the vesicle. It is often composed primarily of two protein complexes: clathrin and the AP2 adaptor. A whole bunch of accessory proteins are also part of the coat and control membrane budding, through molecular connections that involve protein–protein as well as protein–lipid interactions2 (Fig. 1).

Figure 1: Bending membranes.

A vesicle is shown budding from a cell's plasma membrane into the cell. Ford et al.1 propose that epsin proteins induce the membrane curvature associated with budding, and couple it to loading of the vesicle with cargo. Epsin associates with clathrin and the AP2 adaptor complex, basic components of a budding vesicle's 'coat'. Epsin's ENTH domain also binds the phosphorylated head group of the lipid PIP2 in the nearby leaflet of a membrane bilayer (see inset). Ford et al. propose that this disturbs the bilayer's stability, contributing to membrane curvature. Through its interaction with AP2, which in turn interacts with cargo receptors, epsin might couple initial membrane invagination to cargo recruitment. Then, while the coated pit is growing, epsin might affect membrane curvature at the rim of the coat, where another of its partners, Eps15, has been detected. Note that other proteins can also induce membrane curvature, and future work should reveal their spatial and temporal roles.

Proteins of the epsin family3, which are under the spotlight in Ford and colleagues' study1, are among these accessory proteins. Epsins have at one end (the amino, or N, terminus) a structurally defined module that binds phosphoinositides — lipids bearing a head group composed of a so-called inositol ring, modified with phosphate groups (phosphorylated) at several positions. This module, called the ENTH (for 'epsin N-terminal homology') domain, is also found in other proteins involved in membrane dynamics4,5. Among these, the brain-specific protein AP180 and its ubiquitously expressed relative CALM are also implicated in the formation of clathrin-coated vesicles.

Previous structural analyses6,7 of the ENTH domains of CALM and epsin, in the presence of the lipid phosphatidylinositol-4,5-bisphosphate (PIP2), revealed a telling fact: these domains bind PIP2 differently. In fact, the domain in AP180 and CALM is now referred to as the ANTH (for 'AP180-like N-terminal homology') domain. Ford et al.1 now delve a little deeper. Their structural studies show that the PIP2-binding site in the ENTH domain of most (but not all) epsin-family members involves positively charged amino acids that are dispersed through several of the structural units (α-helices) that build up the domain. Upon PIP2 binding, the domain is reorganized into a deep basic groove in which the phosphorylated inositol head group is buried. This is radically different from the ANTH domains of AP180 and CALM, which bind PIP2 through a small, positively charged stretch of consecutive amino acids6.

Ford et al. also find that this discrepancy in the lipid-bound structures of the ENTH and ANTH domains relates to functional specificity in vesicle budding. They show that epsin is recruited to artificial PIP2-enriched liposomes (spherical structures that, like natural vesicles, are bounded by a lipid bilayer), and can bend this template into tubules. Amazingly, this also happens when the experiment is done with epsin's ENTH domain alone, hinting that the phenomenon is strictly induced by binding to PIP2. But the ANTH domain does not cause membrane deformation.

So the authors propose that the burying of PIP2 deep within epsin's ENTH domain encourages deformation of a membrane bilayer (Fig. 1). They suggest that this burial allows the ENTH domain to insert its N-terminal α-helix into the outer leaflet of the bilayer, pushing the lipid head groups apart. This would provide sufficient mechanical force to bend a flat membrane into an emerging bud. As further work showed that epsin binds to the AP2 adaptor and triggers clathrin-coat assembly, and that the combination of epsin and clathrin induces invagination of lipid monolayers, the authors propose a crucial role for epsin in vesicle budding. According to their model, epsin links its membrane-bending effect to cargo recruitment (through AP2, which binds the tails of transmembrane cargo receptors) and to coat assembly.

Epsin is not the only protein that can induce membrane curvature. At least three other proteins that are involved in clathrin-coated vesicle formation can trigger membrane tubulation independently in vitro. These proteins are dynamin, amphiphysin and endophilin8,9,10. Why are so many actors playing what appears to be the same part?

It seems likely that these proteins have subtly different roles in vivo, which are difficult to detect in in vitro tests. For instance, for a long time clathrin was envisaged as being sufficient to drive membrane budding, given its in vitro ability to self-assemble spontaneously into a polyhedral structure resembling the one surrounding a coated vesicle. Nowadays, that view is being revised, on the basis of experiments and theoretical calculations11, and it seems clear that several proteins must work together to bend a biological membrane. A challenge now is to determine whether these proteins induce membrane curvature in vivo as well as in vitro by binding to phospholipids.

It will also be important to find out when the proteins work — some might function during the initiation of budding, some later, and some at multiple stages. Although in vitro data show that all these proteins can initiate deformation from a flat membrane, suggesting that they take part in early events, this may result only from artificial capping of too many of the target lipids, because artificial templates are more enriched in these lipids than are biological membranes. This might be reinforced if the proteins insert significantly into the outer leaflet of a bilayer, inducing membrane bending by increasing the surface of the leaflet12.

Of all the proteins that initiate membrane curvature in vitro, epsin might be the best candidate for this effect in vivo, given that its bending action may go hand in hand with cargo recruitment — a process that begins when the membrane starts to curve. Nevertheless, it is plausible that some of these proteins induce membrane curvature in several different steps of budding. In vivo evidence indicates that endophilin might do so12. Such could be the case for epsin, too. As well as working in the earliest steps of membrane curvature, epsin might also play a later part, through its binding to Eps15 — a protein that has been detected at the rim of budding coated vesicles13.

Understanding all the molecular events that govern the making of a vesicle will require the development of ever more sophisticated approaches, using in vitro systems combined with morphological analysis at the ultrastructural level, as exemplified by Ford and colleagues' work1. This will be complemented by the use of biophysical techniques, some based on high-resolution imaging, that allow us to study the process in atomic detail14.


  1. 1

    Ford, M. G. J. et al. Nature 419, 361–366 (2002).

  2. 2

    Slepnev, V. I. & De Camilli, P. Nature Rev. Neurosci. 1, 161–172 (2000).

  3. 3

    Chen, H. et al. Nature 394, 793–797 (1998).

  4. 4

    Hyman, J. et al. J. Cell Biol. 149, 537–546 (2000).

  5. 5

    De Camilli, P. et al. FEBS Lett. 513, 11–18 (2002).

  6. 6

    Ford, M. G. et al. Science 291, 1051–1055 (2001).

  7. 7

    Itoh, T. et al. Science 291, 1047–1051 (2001).

  8. 8

    Sweitzer, S. M. & Hinshaw, J. E. Cell 93, 1021–1029 (1998).

  9. 9

    Takei, K. et al. Nature Cell Biol. 1, 33–39 (1999).

  10. 10

    Farsad, K. et al. J. Cell Biol. 155, 193–200 (2001).

  11. 11

    Nossal, R. Traffic 2, 138–147 (2001).

  12. 12

    Huttner, W. B. & Schmidt, A. A. Trends Cell Biol. 12, 155–158 (2002).

  13. 13

    Tebar, F. et al. J. Biol. Chem. 271, 28727–28730 (1996).

  14. 14

    Higgins, M. K. & McMahon, H. T. Trends Biochem. Sci. 27, 257–263 (2002).

Download references

Author information

Correspondence to Anne A. Schmidt.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Schmidt, A. The making of a vesicle. Nature 419, 347–349 (2002) doi:10.1038/419347a

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.