814
Endocytosis and vesicle trafficking
Philip R Evans* and David J Owen†
Several common themes have emerged from recent structural
and functional studies of proteins involved in the formation of
coated vesicles. For example, inositol polyphosphate lipid
headgroups are bound specifically by a variety of different
domains in ways appropriate to domain function. Another
theme is the recognition of short sequence motifs in
structureless regions of other coat components, allowing
dynamic multicomponent networks to be established.
Addresses
*MRC Laboratory of Molecular Biology, Hills Road,
Cambridge CB2 2QH, UK;
e-mail: [email protected]
†Cambridge Institute for Medical Research, University of Cambridge,
Department of Clinical Biochemistry, Wellcome Trust/MRC Building,
Hills Road, Cambridge CB2 2XY, UK;
e-mail: [email protected]
Current Opinion in Structural Biology 2002, 12:814–821
0959-440X/02/$ — see front matter
© 2002 Elsevier Science Ltd. All rights reserved.
Abbreviations
Ins(1,4)P2
InsP6
PH
PtdIns(3,4,5)P3
PtdIns3P
PtdIns(4,5)P2
PtdIns4P
inositol (1,4)-bisphosphate
inositol (1,2,3,4,5,6)-hexakisphosphate
pleckstrin homology
phosphatidylinositol (3,4,5)-trisphosphate
phosphatidylinositol 3-phosphate
phosphatidylinositol (4,5)-bisphosphate
phosphatidylinositol 4-phosphate
Introduction
Transport vesicles allow communication within the cell
between the various membrane-bound organelles and the
plasma membrane. They carry transmembrane and soluble
lumenal cargo proteins, and membrane itself between intracellular compartments. The cell contains a number of
different types of coated vesicle, which can be conveniently
classified by the composition of the protein coat surrounding
them. It is generally true that each type of vesicle carries out
one specific or several very closely related transport steps
between a particular pair of compartments.
Vesicles are formed by the deformation of part of a preexisting membrane into which cargo for transport has been
sorted, followed by scission of the vesicle from the membrane. The sorting process may well be a function of the
coat itself. Immediately after formation, the vesicle is
uncoated, removing the scaffold and adaptors, and is then
directed to its correct destination by components of the
cytoskeleton and/or by tethering proteins. Finally, the
vesicle fuses with its target membrane  a process that is
mediated in part by proteins such as SNARES, which are
incorporated into the vesicle in the early stages of its
formation. This carefully orchestrated process requires the
complex interplay of the many protein components that
form the vesicle genesis machinery. These include a
mechanical scaffold for the vesicle, adaptors that link the
scaffold to the cargo and to the membrane (together the
scaffold protein and adaptor form the ‘coat’), proteins that
deform the membrane, proteins that cause scission of the
vesicle and regulatory proteins that ensure that the various
steps occur in the correct order and in the correct location
(Figure 1; for overall reviews, see [1,2]). The principal scaffold protein is clathrin, which occurs in different types of
coated vesicle in conjunction with one of the heterotetrameric adaptor complexes AP1, AP2, AP3 and AP4, or
one the monomeric adaptors the GGAs (the involvement
of clathrin with AP3 and AP4 is unconfirmed). Coat systems that do not involve clathrin include: the COP I set of
proteins, which are distantly related to the AP complexes
and perhaps to clathrin; and the COP II system, which is
unrelated (see Update).
Many of the proteins involved in vesicle formation have
two or more binding sites for other proteins or lipids, often
on different domains, and thus cross-link other components in the assembly process (Figure 1). Protein–protein
interactions include the recognition of signalling motifs
that mark cargo for incorporation into vesicles and the
recognition of other components of the complex network
that are required for coat assembly.
Over the past few years, the structures of domains of many
of the components have been determined. These include
all the folded domains of the heterotetrameric AP2 complex and the appendage domain of γ-adaptin from AP1;
phosphoinositide-binding domains from epsin, AP180,
EEA1, Vps27 and Hrs; protein-binding domains from
GGAs (VHS), Eps15 (EH) and clathrin (propeller domain);
enzymatic domains from dynamin and synaptojanin; and
the leg domains of clathrin. When integrated with the
wealth of biochemical and cell biological data that have
been generated over the past 30 years, these structures are
yielding a detailed understanding of how transport vesicles
are generated.
Lipid-binding modules
Lipids such as phospholipids or cholesterol serve as
compartment identification markers for the membranes of
organelles. Proteins that can recognise these lipids will therefore show specific membrane and/or organelle localisation.
These proteins can in turn recruit further proteins or modify
membrane components in a membrane- or organellespecific manner. The most extensively studied organellespecific lipids are the phosphatidylinositides that mark the
plasma membrane [PtdIns(4,5)P2 and PtdIns(3,4,5)P3], the
lysosome/vacuole and endosome (PtdIns3P), and the Golgi
(PtdIns4P). Some of these molecules also serve as second
messengers in signal transduction pathways, a role closely
linked to their functions in vesicle formation and as
Endocytosis and vesicle trafficking Evans and Owen
815
Figure 1
Molecular interactions in clathrin-coated
vesicles. Bottom right: schematic diagram of a
vesicle, showing cargo embedded in the inner
membrane layer, an outer clathrin scaffold and
a middle layer of adaptors and accessory
proteins. Top left: some of the most important
proteins involved in forming the coat. Lines
indicate protein–protein and protein–lipid
interactions.
Transmembrane cargo
Membrane
AP2 adaptor
Dab2
β
Synaptojanin
α
Epsin
Endophilin
Amphiphysin
AAK
Dynamin
Eps15
Clathrin
Auxilin
AP180
GAK
Adaptors
Cargo
Membrane
Accessory
proteins
Clathrin coat
Current Opinion in Structural Biology
compartment markers. As might be expected given the
importance and variety of the roles of phosphoinositides,
they are recognised by numerous distinct protein modules.
The structures of many different protein modules that recognise phosphoinositides have been determined (see Figure 2
for the structures discussed here). These include lipid-binding modules, as well as enzymes such as kinases and
phosphatases that modify lipid headgroups. A recent addition to the enzymes is the inositol polyphosphate
5-phosphatase domain of Schizosaccharomyces pombe synaptojanin [3•], with and without the reaction product [Ins(1,4)P2]
and a Ca2+ ion. This enzyme is similar in structure and mechanism to DNases such as DNase I, exonuclease III and the
endonuclease APE1, for which a detailed reaction mechanism is known from a series of crystal structures. The
stereochemistry of product binding shows clearly why the
phosphatase activity is specific for the 5-position.
Inositide kinases and phosphatases enclose their substrates in deep clefts in order to recognise them and carry
out the reactions. Some of the binding modules, such as
the well-studied pleckstrin homology (PH) domains, also
bury their ligands in clefts, but in other ‘binding-only’
modules the binding site is located on the surface. The
most dramatically exposed of these is that of the N-terminal
domain of proteins of the AP180 family [4,5•], which has
now been termed the ANTH (AP180 N-terminal homology)
domain [6••] to distinguish it from the structurally related
ENTH (epsin N-terminal homology) domain from the
epsin family (see below). The ANTH domain is a member
of a large family of proteins containing a helical superhelix
or solenoid. The binding site is a positively charged cluster
of mostly lysine residues that binds a pair of phosphates
attached to adjacent carbon atoms [such as in PtdIns(4,5)P2]
on the tips of the lysines, with no interactions with the rest
of the inositol ring [5•].
The α-adaptin subunit of AP2 is also a helical solenoid ([7••];
see below for a description of the AP2 core structure) and
also has a surface site that binds phosphoinositides
[PtdIns(4,5)P2 in the cell; InsP6 in the crystal structure], but
the structures and binding sites of α-adaptin and ANTH are
different and cannot be superimposed. Another subunit of
AP2, µ2-adaptin, also binds PtdIns(4,5)P2 on the tips of a
cluster of lysines [8] (Figure 2), although in this case the
lysines are displayed on an all-β structure. In contrast, the
ENTH domain of epsin, despite its close similarity to the
ANTH domain, binds the lipid headgroup Ins(1,4,5)P3 in a
completely different manner [6••], in a deep, positively
charged cleft formed by folding the N terminus into an additional helix that is not present in the unligated structure [9].
Ligand binding involves residues that were implicated by
NMR and binding data [10], including those in the new
helix. Folding of the new helix creates an exposed hydrophobic ridge. This ridge may insert into the hydrophobic layer of
816
Proteins
Figure 2
896
(AP180)
281
575
158
N
Epsin
(a)
(b)
CALM
(AP180)
(c)
α-adaptin
(AP2 core)
N
(d)
µ2-adaptin (AP2 core)
(e)
C
N
Phosphatidylinositide-binding modules. The
proteins are shown in a schematic
representation, with dashed lines indicating
where parts of the full-length molecule were
not included in the crystal structure.
Phosphoinositides are shown as stick models.
In all cases, the orientation relative to the
membrane is unknown. The ENTH domain of
epsin, the ANTH domain of CALM (and
AP180) and the N terminus of α-adaptin
(from the AP2 core structure) all recognise
PtdIns(4,5)P2 on the plasma membrane.
(a) Binding of PtdIns(4,5)P2 to the epsin
ENTH domain induces the N terminus to fold
into a new helix with a hydrophobic ridge that
might insert into the membrane [6••]. Note
that only the headgroup is bound in the crystal
structure and the acyl chains shown here are
modelled. (b) The ANTH domain of CALM
binds PtdIns(4,5)P2 in an exposed surface
site [5•]. This view is equivalent to that of the
ENTH domain in (a). (c) The superhelical
solenoid of α-adaptin binds PtdIns(4,5)P2
also in a surface site, but its structure is not
the same as AP180 [7••]. Note that InsP6
from the crystal structure is shown here.
(d) The second InsP6 on the AP2 core is
bound to a positively charged patch of lysines
on the surface of the β sheet of the
µ2-adaptin signal-binding domain [7••].
(e) FYVE domains are specific for PtdIns3P
and the dimer of EEA1 presents two binding
sites to the membrane [15••]. The tip of the
loop on the outside may also dip into the
membrane. (f) The inositol polyphosphate
5-phosphatase domain presents several
positive residues to aid its binding to
membranes [12].
(f)
EEA1 (FYVE domain)
S. pombe synaptojanin
(Ins 5-phosphatase domain)
Current Opinion in Structural Biology
the membrane and cause curvature of the lipid bilayer,
because mutation of its residues prevents the domain from
forming lipid tubules, while not preventing its ability to bind
to phosphoinositide headgroups or membranes [6••]. The
ANTH and ENTH domains have similar structures and
related sequences, and bind to the same membrane headgroup, but do so in completely different ways, reflecting their
different functions in vesicle formation.
Most of the phosphoinositide-binding modules seem to
come in different versions with a wide range of specificities
for different phosphorylated inositides. So far, all FYVE
domains, which occur in a number of signalling and trafficking proteins, have been found to be specific for PtdIns3P.
Three crystal structures and an NMR structure together
show how this module recognises its ligand (for a review, see
[11]). The first crystal structure was of the Vps27p FYVE
domain [12] and the binding site was predicted from an
anionic crystal contact and residue conservation. The structure of the tandem VHS and FYVE domains of Hrs [13]
showed a citrate ion trapped in a dimer interface; this was
proposed as the PtdIns3P-binding site. The NMR structure
of the monomeric FYVE from the endosomal protein EEA1
[14] roughly mapped the true binding site, which was confirmed in detail by the crystal structure of the dimeric form
of the EEA1 FYVE domain, including part of the coiled-coil
that precedes it [15••] (Figure 2). Both binding sites in the
dimer can be simultaneously presented to the membrane,
enhancing its affinity for membranes, whereas dimerisation
does not much alter binding to soluble headgroups.
Endocytosis and vesicle trafficking Evans and Owen
Protein–protein interactions
Many of the protein–protein interactions in vesicle formation consist of a folded domain binding to a short region of
4–10 amino acids of another protein, which may have no
ordered structure before it is bound. Such interactions can
be mimicked with short peptides and have a KD in the
micromolar range, resulting in the fast on and off rates that
are required for multiple interactions in a dynamic process.
This is in contrast to the tight association between folded
subunits in a stable complex, which are mediated by large
contact areas. Some of these interactions have been studied
as complexes with peptides, by both crystallography and
NMR (Figure 2). There are many different ways in which
binding can occur, as might be expected, because peptides
bind to proteins with the same variety of interactions that
occur within protein folds themselves.
VHS domains (reviewed in [16]) were identified from
sequence comparisons of signal transduction proteins. The
first two structures, VHS domains from Tom1 [17] and Hrs
[13], showed the domain to consist of an eight α-helical solenoid, rather like the ENTH domain. The function of the
VHS domains from these proteins is unknown. More recently,
the VHS domains of the monomeric adaptors GGAs have
been shown to bind cargo proteins marked with sorting signals of the acidic cluster dileucine family. The structures of
VHS domains from GGA1 and GGA3 have been determined in complex with signal peptides [18,19••,20•]. The
peptides bind in a shallow groove between helices 6 and 8,
with six or seven residues visible. The main determinants of
binding are an aspartic acid and two leucine residues
arranged in the following motif: xDxxLLxx. For efficient
binding, this motif needs to be near the C terminus (numbering the aspartic acid residue as 0, the C terminus is
optimally at D+5 or D+6). Binding affinity is also modulated
by the charge or phosphorylation state at the D–1 position,
but this is not an on–off switch: the affinities (KD) for a
series of peptides with charges at D–1 of +1 (arginine),
0 (serine) and –1 (phosphoserine) are, respectively, 29, 7.9
and 2.3 µM for GGA1, and 56, 10.9 and 3.5 for GGA3 [20•].
The residue at D–1 is close to a positively charged patch on
the protein surface, explaining the preference for a negative
charge. These affinities also show that the specificities of
GGA1 and GGA3 are not very different.
A better-known sorting signal is the YxxΦ motif (where Φ
is a hydrophobic residue), which is recognised by the
µ2-adaptin subunit of the AP2 complex. This peptide was
previously shown to bind as an extra β strand to the all-β
C-terminal domain, with the Y and Φ sidechains tucked
into pockets on either side of the sheet [21]. Studies with
other peptides have now shown that residues at the Y–3
position can bind into a third pocket (Figure 3) and that this
can contribute significantly to the binding affinity [22,23].
The C-terminal appendage or ‘ear’ domains of the large
subunits of AP2 (α- and β2-adaptin) recruit accessory
proteins (e.g. eps15, epsin, AP180, amphiphysin, auxilin,
817
numb, Dab2, GAK and AAK) to the budding vesicle by
binding to short sequence motifs in these proteins. The
binding site was previously mapped to a site on the ‘platform’ subdomain by mutagenesis guided by the structures
[24–26], but the details have now been shown by a series
of crystal structures of the α-adaptin appendage bound to
peptides [27••] (Figure 3). Two distinct modes of binding
to this site were observed. The first is seen for peptides
containing the motif DPF or DPW; these peptides bind in
a type I β turn held together by a hydrogen bond from the
peptide amino group of phenylalanine or tryptophan to
the sidechain of the peptide aspartic acid, with the phenylalanine or tryptophan inserted into the deep hydrophobic
pocket noted above. The second mode of binding occurs
with peptides of the form FxDxF, which bind with the
first phenylalanine in the hydrophobic pocket and extend
over a larger region of the surface. This second motif is
found in amphiphysin. Comparisons with the related
appendage domain of β2-adaptin [26] predict that the
appendage domain would bind only the DPF/DPW motif,
explaining why β2-adaptin does not bind amphiphysin. A
second distinct binding site on the N-terminal ‘sandwich’
subdomain was also observed with the DPW peptide
in one crystal form, but it is not clear whether this site is
biologically significant.
The equivalent appendage domain of γ-adaptin from the
AP1 complex lacks the platform subdomain, but the binding site for its partners, γ-synergin, Eps15 and rabaptin5,
has been mapped by mutagenesis by two groups who
solved the crystal structure of this domain [28,29]. Taken
together, the γ-adaptin mutants delineate a binding site
stretching from a hydrophobic pocket between the two
sheets of the sandwich across the face of one sheet. This is
the opposite sheet to the one on which the second site in
α-adaptin is located (Figure 3). A similar domain is found
from sequence alignments at the C terminus of the
monomeric GGA adaptor proteins.
Other structures and domains
The heterotetrameric adaptor proteins, exemplified by AP2,
consist of two large subunits, α (110 kDa) and β2 (110 kDa);
a medium subunit, µ2 (50kDa); and a small subunit,
σ2 (17 kDa). The large subunits can be subdivided into a
trunk domain of ~70 kDa joined to the ~30 kDa
appendage domain by a flexible linker. The structure of
the core complex consisting of the trunk domains of α and
β2 with the complete µ2 and σ2 subunits has now been
solved [7••]. The two large subunits form helical solenoids
that wrap round the small domains (σ2 and the N-terminal
domain of µ2) (Figure 4). This pair of heterodimers (α/σ2
and β2/N-µ2) forms a shallow dish carrying the bananashaped all-β, C-terminal domain of µ2, which contains a
binding site for YxxΦ peptides. Comparison with the structures of signal peptides bound to the isolated C-terminal
domain of µ2 [21–23] shows that the binding site is
blocked over much of its length in the intact core structure,
so this represents an inactive conformation. Biochemical
818
Proteins
Figure 3
(c) µ2-adaptin signal-binding site
(a) GGA3 VHS domain
Ser(P)–1
Asp0
Y
Y+3
Leu3
Leu4
Y–3
(d) Clathrin propeller domain
(b) GGA1 VHS domain
Asp0
Leu3
Leu4
(e) α-adaptin appendage
peptide-binding site
(f) α-adaptin appendage
(g) γ-adaptin appendage
Region of binding site
Peptide-binding sites. (a,b) GGA1 and GGA3
VHS domains bind xDxxLL signal peptides in
a shallow groove between two helices
[18,19••]. Phosphoserine at the –1 position
increases binding by interacting with positive
charge on the protein [20•]. Left, schematic
view: right, surface coloured electrostatic
potential. (c) ΦxxYxxΦ signal peptides bind to
µ2-adaptin as a three-pin plug into three
pockets along the edge of the β sheet:
purple indicates the P-selectin peptide [22]
and yellow indicates the CTLA4 peptide
TTGVYVKMPP [23]. (d) ‘Clathrin box’
peptides DTNLIEFE and AVSLLDLDA bound
to the N-terminal clathrin propeller domain
[37]. (e) The platform subdomain of the
α-adaptin appendage domain binds
DPF/DPW and FxDxF peptides in distinct
modes, but in both the deep hydrophobic
pocket is used to bind an aromatic sidechain
[27••]. View is looking down from the top of
the structure shown in (f). (f) Ribbon diagram
of the α-adaptin appendage domain showing
the main peptide site (top) and the subsidiary
(possibly nonphysiological) site behind the
lower sandwich subdomain [27••]. (g) The
γ-adaptin appendage domain lacks the
platform domain and the peptide-binding site
maps to a region including a hydrophobic
pocket between the β sheets [28,29]. View is
equivalent to that shown in (f). Electrostatic
surfaces were calculated in GRASP [38] and
are coloured from red –10 kT to blue +10 kT.
Hydrophobic surfaces in (c–e) were
calculated as in [24]; dark green indicates a
favourable hydrophobic interaction.
Current Opinion in Structural Biology
evidence [30] suggests that a conformational change that
allows binding of the YxxΦ motif is induced by a combination of factors, including phosphorylation of the
disordered linker region between the two domains of µ2
and the presence of PtdIns(4,5)P2 in the planar plasma
membrane. The structure of the core complex also shows
two binding sites for phosphoinositides (see above), which
serve to localise the adaptor to the plasma membrane [7••].
Dynamins are large multidomain GTPases involved in the
scission of vesicles from the membrane. The structure of
the GTPase domain [31] of dynamin A from Dictyostelium
is, as expected, related to other G proteins, such as Ras,
but with an insertion that adds two extra β-sheet strands
and a helix to the core fold. The structure was determined
with and without GDP, with little conformational change
on GDP binding. Full-length dynamin contains several
other domains, including an effector domain (GED) and a
membrane-binding PH domain of known structure (which
may not be present in Dictyostelium dynamin A). The GTPbinding protein hGBP1 [32] may provide a model for the
regulatory domains, although its GTPase domain is not
particularly similar to the G-protein domain of dynamin A.
Dynamin forms helical assemblies around lipid tubules
and a helical reconstruction from cryo-electron micrographs at 20 Å resolution shows the arrangement of dimers,
with the PH domains on the inside and the GTPase
domains on the outside [33]. Crystal structures of these
domains can be fitted into the density, leaving a central
stalk region to account for regulatory domains. These
reconstructions were generated from the best-ordered
specimens, constricted tubes formed from dynamin lacking
the C-terminal proline-rich domain in the presence of the
GTP analogue GMP-PCP.
Endocytosis and vesicle trafficking Evans and Owen
819
Figure 4
Loosely connected domains. (a) A
representation of full-length epsin, showing
the folded ENTH domain at the bottom and an
unfolded C-terminal region containing multiple
binding sites for ubiquitin, adaptors (AP2),
clathrin and Eps15. (b) The entire
heterotetrameric AP2 complex, compiled from
crystal structures of the core (trunk) [7••] and
the two ‘appendage’ domains [24,26]. The
loose linkers (hinges) attaching the
appendages are schematic. The superhelical
trunk domains of α- and β2-adaptin wrap
round the small domains of σ2-adaptin and
the N-terminal domain of µ2-adaptin. The
signal-binding site on the C-terminal domain
of µ2-adaptin is blocked by β2-adaptin. The
two InsP6 sites correspond to the single site
between molecules in the crystal.
(a) Epsin
575
(b) AP2
Eps15
(NPF)
α-adaptin appendage
Clathrin
β2-adaptin appendage
InsP6
Adaptors
(DPF/DPW)
α-adaptin
β2-adaptin
Clathrin
σ2-adaptin
Signalbinding site
Ubiquitin
(UIMs)
µ2-adaptin
InsP6
158
Current Opinion in Structural Biology
Comparison of different helical states induced by GTP or
GDP by electron microscopy, together with crystal structures of the domains, is beginning to address the various
controversies over the mechanochemical basis of vesicle
scission, but there is still some way to go. Electron
microscopy is also allowing other large complexes to be
studied at lower resolution (~20 Å); visualisation of part of
the COP II coat in the cytosol [34] has followed earlier
studies of dynamin, clathrin and Eps15.
Domains, strings and hooks: protein fishing
It has long been recognised that most proteins involved in
vesicle formation are constructed from multiple domains,
each serving to bind another protein (or lipid). The usual
models of such multidomain proteins have been the tight
association between different domains (as seen here with
the AP2 core); a looser more flexible arrangement allowing
regulatory conformational change (e.g. in dynamin); or
‘beads-on-a-string’, functional folded domains separated
by loose connections that are not themselves functional. It
is becoming increasingly clear that unfolded flexible
regions of proteins do have a major function in
protein–protein interactions, in that they contain short
sequence motifs (up to about 10 amino acids) that can bind
to a folded domain. For example, the appendage domains
of AP2 are attached to the core by loose linkers or hinges,
and the principal clathrin-binding region is the β2-adaptin
hinge. These ‘links’ can display multiple copies of different motifs along their lengths. This organisation permits
the establishment of complex networks of interactions,
with the relative affinity between members being determined by the number of motifs that each one possesses. In
addition, this organization also allows regulation by phosphorylation, because a phosphorylated group can have a
major effect on the affinity of a small interacting site. This
has most recently been proposed to occur in the recognition
of an internal phosphorylated dileucine motif in the hinge
of GGA by its own N-terminal VHS domain [35].
Proteins of the epsin and AP180 families are perhaps
extreme examples of this. They contain a folded domain
(ENTH or ANTH) at their N terminus, but the rest of the
protein (about 425 residues for epsin1 and about 600 for
AP180) seems to have no detectable structure, according to
a series of biophysical studies [36••] (e.g. hydrodynamic
tests and circular dichroism), even though these regions
are heat stable. The C-terminal regions of both epsin and
AP180 contain multiple binding sites for clathrin (such as
DLL), for AP2 adaptors (DPF/DPW motifs for the
α-adaptin appendage domain) and, in the case of epsin, for
ubiquitin (ubiquitin-interacting motifs) and for the EH
domains (NPF motifs) of Eps15 (Figure 4). These proteins
are perhaps best visualised as long fishing lines anchored
to the membrane by the ENTH/ANTH domain, with
multiple hooks acting as bait that can cross-link their various
partners during the assembly process [36••].
Perspectives
The assembly of vesicle coats involves an elaborate series of
protein–protein and protein–lipid interactions. A combination
820
Proteins
of structural, biochemical and cell biology is beginning to
elucidate the details of the molecular mechanisms. The
principles of both phosphoinositide recognition and the
recognition of short sequence motifs in unfolded regions of
one protein by a folded domain of another are themes also
relevant to many other processes in the cell. Our views of
proteins have been coloured by the successes of structure
determination, which show us the folded and ordered
domains, but unfolded parts of proteins may be of critical
importance.
[10]. This creates an exposed hydrophobic ridge on the helix. Mutations in
this hydrophobic ridge show that it is essential both for the formation of lipid
tubules by the ENTH domain and for the epsin-stimulated assembly of
clathrin on lipid monolayers.
7.
••
Collins BM, McCoy AJ, Kent HM, Evans PR, Owen DJ: Molecular
architecture and functional model of the endocytic AP2 complex.
Cell 2002, 109:523-535.
The authors report the crystal structure of the 200 kDa tetrameric core of the
AP2 complex in an inactive conformation with the signal-binding site
blocked. InsP6 binds between two molecules in the crystal, identifying two
physiological binding sites on the α- and µ2-adaptin subunits.
8.
Rohde G, Wenzel D, Haucke V: A phosphatidylinositol
(4,5)-bisphosphate binding site within µ2-adaptin regulates
clathrin-mediated endocytosis. J Cell Biol 2002, 158:209-214.
Update
9.
The structure of the Sec23/Sec24 heterodimer, the major
components of the COP II coat, has been determined
[39••]. The structures also show the details of the interaction of Sec23 with the small G protein Sar1 (related to
ARF), which is involved in recruiting Sec23/24 to the
membrane. The elongated bow-tie-shaped complex has a
curved surface matching the curvature of a vesicle. The
architecture is completely unrelated to the clathrin/AP2
system discussed above: Nature has more than one way of
forming a vesicle coat.
Hyman J, Chen H, Di Fiore PP, De Camilli P, Brunger AT: Epsin 1
undergoes nucleocytosolic shuttling and its Eps15 interactor
NH2-terminal homology (ENTH) domain, structurally similar to
armadillo and HEAT repeats, interacts with the transcription factor
promyelocytic leukemia Zn2+ finger protein (PLZF). J Cell Biol
2000, 149:537-546.
10. Itoh T, Koshiba S, Kigawa T, Kikuchi A, Yokoyama S, Takenawa T: Role
of the ENTH domain in phosphatidylinositol-4,5-bisphosphate
binding and endocytosis. Science 2001, 291:1047-1051.
11. Misra S, Miller GJ, Hurley JH: Recognizing phosphatidylinositol
3-phosphate. Cell 2001, 107:559-562.
12. Misra S, Hurley JH: Crystal structure of a phosphatidylinositol
3-phosphate-specific membrane-targeting motif, the FYVE
domain of Vps27p. Cell 1999, 97:657-666.
We thank JP Luzio and HM McMahon for helpful comments.
13. Mao YX, Nickitenko A, Duan XQ, Lloyd TE, Wu MN, Bellen H,
Quiocho FA: Crystal structure of the VHS and FYVE tandem
domains of Hrs, a protein involved in membrane trafficking and
signal transduction. Cell 2000, 100:447-456.
References and recommended reading
14. Kutateladze T, Overduin M: Structural mechanism of endosome
docking by the FYVE domain. Science 2001, 291:1793-1796.
Acknowledgements
Papers of particular interest, published within the annual period of review,
have been highlighted as:
• of special interest
•• of outstanding interest
1.
2.
Brodsky FM, Chen C-Y, Knuehl C, Towler MC, Wakeham DE:
Biological basket weaving: formation and function of
clathrin-coated vesicles. Annu Rev Cell Dev Biol 2001,
17:517-568.
Kirchhausen T: Three ways to make a vesicle. Nat Rev Mol Cell Biol
2000, 1:187-198.
3.
•
Tsujishita Y, Guo SL, Stolz LE, York JD, Hurley JH: Specificity
determinants in phosphoinositide dephosphorylation: crystal
structure of an archetypal inositol polyphosphate 5-phosphatase.
Cell 2001, 105:379-389.
The lipid phosphatase domain of S. pombe synaptojanin is structurally related
to nucleases. The complex with the headgroup of the reaction, Ins(1,4)P2,
explains the specificity for cleavage at the 5-position.
4.
Mao YX, Chen J, Maynard JA, Zhang B, Quiocho FA: A novel all helix
fold of the AP180 amino-terminal domain for phosphoinositide
binding and clathrin assembly in synaptic vesicle endocytosis.
Cell 2001, 104:433-440.
5.
•
Ford MGJ, Pearse BMF, Higgins MK, Vallis Y, Owen DJ, Gibson A,
Hopkins CR, Evans PR, McMahon HT: Simultaneous binding of
PtdIns(4,5)P2 and clathrin by AP180 in the nucleation of clathrin
lattices on membranes. Science 2001, 291:1051-1055.
Two papers [4,5•] describe the structures of the ANTH domain of two
proteins of the AP180 family, LAP from Drosophila and rat CALM. Ford
et al. [5•] describe complexes with lipid headgroups, showing that the two
adjacent phosphates are bound on the tips of lysine sidechains in a most
unusual surface binding site. Full-length AP180 stimulates clathrin assembly and forms clathrin buds on lipid monolayers (visualised by electron
microscopy), which become more invaginated in the presence of the AP2
adaptor complex.
6.
••
Ford MG, Mills IG, Peter BJ, Vallis Y, Praefke GJK, Evans PR,
McMahon HT: Curvature of clathrin-coated pits driven by epsin.
Nature 2002, 419:361-366.
The ENTH domain of epsin co-crystallised with the lipid headgroup
Ins(1,4,5)P3 shows that the ligand binds in a deep cleft formed by the folding of N-terminal residues that are disordered in the unligated structure [9]
into a new α helix, including residues previously identified to bind the ligand
15. Dumas JJ, Merithew E, Sudharshan E, Rajamani D, Corvera S,
•• Lambright DG: Multivalent endosome targeting by homodimeric
EEA1. Mol Cell 2001, 8:947-958.
The structure of the C-terminal FYVE domain of EEA1 with part of the dimerisation coiled-coil in complex with Ins(1,3)P2 finally identifies the true binding
site for the lipid headgroup and the reason for its specificity.
16. Lohi O, Poussu A, Mao YX, Quiocho F, Lehto VP: VHS domain – a
longshoreman of vesicle lines. FEBS Lett 2002, 513:19-23.
17.
Misra S, Beach BM, Hurley JH: Structure of the VHS domain of
human Tom1 (target of myb1): Insights into interactions with
proteins and membranes. Biochemistry 2000, 39:11282-11290.
18. Shiba T, Takatsu H, Nogi T, Matsugaki N, Kawasaki M, Igarashi N,
Suzuki M, Kato R, Earnest T, Nakayama K et al.: Structural basis for
recognition of acidic-cluster dileucine sequence by GGA1. Nature
2002, 415:937-941.
19. Misra S, Puertollano R, Kato Y, Bonifacino JS, Hurley JH: Structural
•• basis for acidic-cluster-dileucine sorting-signal recognition by
VHS domains. Nature 2002, 415:933-937.
Two papers [18,19••] show details of the binding and recognition of xDxxLL
signal peptides to the VHS domains of GGA1 and GGA3.
20. Kato Y, Misra S, Puertollano R, Hurley JH, Bonifacino JS:
•
Phosphoregulation of sorting signal-VHS domain interactions by
a direct electrostatic mechanism. Nat Struct Biol 2002, 9:532-536.
Signal binding to GGA VHS domains is modulated but not completely
determined by phosphorylation of a serine residue before the aspartic acid
in the recognition motif S(P)DxxLL, which binds to a positively charged
patch on the protein.
21. Owen DJ, Evans PR: A structural explanation for the recognition of
tyrosine-based endocytotic signals. Science 1998,
282:1327-1332.
22. Owen DJ, Setiadi H, Evans PR, McEver PP, Green SA: A third
specificity-determining site in µ2 adaptin for sequences upstream
Φ sorting motifs. Traffic 2001, 2:105-110.
of YxxΦ
23. Follows ER, McPheat JC, Minshull C, Moore NC, Pauptit RA,
Rowsell S, Stacey CL, Stanway JJ, Taylor IWF, Abbott WM: Study of
the interaction of the medium chain µ2 subunit of the
clathrin-associated adapter protein complex 2 with cytotoxic
T-lymphocyte antigen 4 and CD28. Biochem J 2001, 359:427-434.
Endocytosis and vesicle trafficking Evans and Owen
24. Owen DJ, Vallis Y, Noble ME, Hunter JB, Dafforn TR, Evans PR,
McMahon HT: A structural explanation for the binding of multiple
ligands by the α-adaptin appendage domain. Cell 1999,
97:805-815.
25. Traub LM, Downs MA, Westrich JL, Fremont DH: Crystal structure of
the alpha appendage of AP-2 reveals a recruitment platform for
clathrin-coat assembly. Proc Natl Acad Sci USA 1999,
96:8907-8912.
26. Owen DJ, Vallis Y, Pearse BMF, McMahon HT, Evans PR: The
structure and function of the β2-adaptin appendage domain.
EMBO J 2000, 19:4216-4227.
27.
••
Brett TJ, Traub LM, Fremont DH: Accessory protein recruitment
motifs in clathrin-mediated endocytosis. Structure 2002,
10:797-809.
Crystal structures of peptide complexes of the ‘appendage’ domain of
α-adaptin show two modes of binding into the previously identified
hydrophobic site (see [24,25]).
28. Nogi T, Shiba Y, Kawasaki M, Shiba T, Matsugaki N, Igarashi N,
Suzuki M, Kato R, Takatsu H, Nakayama K et al.: Structural basis for
the accessory protein recruitment by the γ-adaptin ear domain.
Nat Struct Biol 2002, 9:527-531.
29. Kent HM, McMahon HT, Evans PR, Benmerah A, Owen DJ: γ-Adaptin
appendage domain: structure and binding site for Eps15 and
γ-synergin. Structure 2002, 10:1-20.
30. Ricotta D, Conner S, Schmid S, von Figura K, Honing S:
Phosphorylation of the AP2 µ subunit by AAK1 mediates high
affinity binding to membrane protein sorting signals. J Cell Biol
2002, 156:791-795.
31. Niemann HH, Knetsch MLW, Scherer A, Manstein DJ, Kull FJ: Crystal
structure of a dynamin GTPase domain in both nucleotide-free
and GDP-bound forms. EMBO J 2001, 20:5813-5821.
32. Prakash B, Praefcke GJK, Renault L, Wittinghofer A, Herrmann C:
Structure of human guanylate-binding protein 1 representing
821
a unique class of GTP-binding proteins. Nature 2000,
403:567-571.
33. Zhang PJ, Hinshaw JE: Three-dimensional reconstruction of
dynamin in the constricted state. Nat Cell Biol 2001, 3:922-926.
34. Lederkremer GZ, Cheng YF, Petre BM, Vogan E, Springer S,
Schekman R, Walz T, Kirchhausen T: Structure of the Sec23p/24p
and Sec13p/31p complexes of COPII. Proc Natl Acad Sci USA
2001, 98:10704-10709.
35. Doray B, Bruns K, Ghosh P, Kornfeld SA: Autoinhibition of the
ligand-binding site of GGA1/3 VHS domains by an internal acidic
cluster-dileucine motif. Proc Natl Acad Sci USA 2002,
99:8072-8077.
36. Kalthoff C, Alves J, Urbanke C, Knorr R, Ungewickell EJ: Unusual
structural organization of the endocytic proteins AP180 and
epsin 1. J Biol Chem 2002, 277:8209-8216.
For both epsin and AP180, no region of the protein apart from the N-terminal
ENTH/ANTH domain shows evidence of any folded structure, by hydrodynamics or circular dichroism. This extended region (~400 residues for epsin
and ~600 for AP180) contains a succession of binding sites for other
endocytosis components, notably clathrin and AP2.
37.
ter Haar E, Harrison SC, Kirchhausen T: Peptide–in-groove
interactions link target proteins to the β-propeller of clathrin.
Proc Natl Acad Sci USA 2000, 97:1096-1100.
38. Nicholls A, Sharp KA, Honig B: Protein folding and association:
insights from the interfacial and thermodynamic properties of
hydrocarbons. Proteins 1991, 11:281-296.
39. Bi XP, Corpina RA, Goldberg J: Structure of the Sec23/24-Sar1
•• pre-budding complex of the COPII vesicle coat. Nature 2002,
419:271-277.
The crystal structures of the Sec23/Sec24 complex and the Sec23/Sar1
complex (with the GTP analogue GMPPNP) provide a detailed model of the
Sec23/24/Sar1/GTP complex. Sec23 is an activating protein (GAP) for
Sar1 and inserts an arginine sidechain into its active site.