© 2006 Nature Publishing Group
Eukaryotic cells contain membrane-enclosed organelles
that communicate with each other through the exchange
of trafficking vesicles. Trafficking usually involves the
generation of a vesicle from a precursor membrane,
the transport of the vesicle to its destination and, last, the
fusion of the vesicle with the target compartment.
Despite an enormous diversity in the size and shape
of the organelles, the basic reactions — budding and
fusion — are carried out by multiprotein complexes
that consist of protein families that have been conserved
throughout eukaryotic evolution
1
.
Since the late 1980s, when SNARE (soluble
N-ethylmaleimide-sensitive factor attachment pro-
tein receptor) proteins were first characterized, rapid
progress has identified SNAREs as key elements in mem-
brane fusion. Although extracellular membrane fusion
and the fusion events of mitochondria and peroxisomes
involve unrelated proteins, SNAREs seem to mediate
membrane fusion in all of the trafficking steps of the
secretory pathway. A mechanistic molecular model
of SNARE-mediated membrane fusion has emerged
that — although not undisputed — is supported by a
steadily increasing body of evidence (see the
TIMELINE
for the key milestones in the field). According to this
model, SNARE proteins that are localized in opposing
membranes drive membrane fusion by using the free
energy that is released during the formation of a four-
helix bundle. The formation of this bundle leads to a
tight connection of the membranes that are destined to
fuse, and initiates the membrane merger. The recycling
of SNAREs is achieved through the dissociation of the
helical bundle, which is mediated by the
AAA+ protein
NSF (N-ethylmaleimide-sensitive factor).
In this article, we give an overview of the structure of
SNARE proteins, and then describe how the emerging
biophysical features of the SNARE cycle are providing
an increasingly coherent picture of SNARE-mediated
membrane fusion. For further information, the reader
is referred to recent reviews
2–4
that cover some of the
topics that are addressed here in more depth.
SNARE structure
SNARE proteins form a superfamily of small proteins with
25 members in Saccharomyces cerevisiae, 36 members
in humans and 54 members in Arabidopsis thaliana.
They have a simple domain structure, and a character-
istic of SNAREs is the SNARE motif — an evolution-
arily conserved stretch of 60–70 amino acids that are
arranged in heptad repeats. At their C-terminal ends,
most SNAREs have a single transmembrane domain
that is connected to the SNARE motif by a short linker.
Many SNAREs have independently folded domains that
are positioned N-terminal to the SNARE motif and
that vary between the subgroups of SNAREs (
FIG. 1;
reviewed in
REFS 2,3).
Although this prototypic structure applies to most
SNAREs, there are important exceptions. A subset
of SNAREs (including the evolutionarily younger
‘brevins’
5
) lacks the N-terminal domain. Another
subset lacks transmembrane domains, but most of
these SNAREs contain hydrophobic post-translational
modifications that mediate membrane anchorage. These
SNAREs include a small group that is represented by
the neuronal SNARE
SNAP-25 (25-kDa synaptosome-
associated protein), which contains two different
SNARE motifs that are joined by a flexible linker that
is
palmitoylated. In the S. cerevisiae SNARE Ykt6, the
transmembrane domain is replaced by a
CAAX box
that is
farnesylated
6
. Intriguingly, SNAREs that carry
transmembrane domains can also be palmitoylated,
which has recently been shown to protect SNAREs from
ubiquitylation and subsequent degradation
7
.
*Department of
Neurobiology, Max Planck
Institute for Biophysical
Chemistry, Am Fassberg,
37077 Göttingen, Germany.
‡
Genentech Inc., South San
Francisco, California 94080,
USA.
Correspondence to R.J.
e-mail: rjahn@gwdg.de
doi:10.1038/nrm2002
Published online
16 Aug ust 20 0 6
AAA+ proteins
(‘ATPases associated with
various cellular activities’
proteins). A superfamily of
proteins with one or two
nucleotide-binding domains,
which often form ring-like
oligomers and function as
chaperones in diverse cellular
processes. They can unfold
aggregates or tightly packed
structures.
Palmitoylation
A post-translational
modification of proteins in
which a palmitate fatty acyl
chain is covalently attached
to a cysteine side chain by a
thioester bond.
SNAREs — engines for membrane
fusion
Reinhard Jahn* and Richard H. Scheller
‡
Abstract | Since the discovery of SNARE proteins in the late 1980s, SNAREs have been
recognized as key components of protein complexes that drive membrane fusion. Despite
considerable sequence divergence among SNARE proteins, their mechanism seems to be
conserved and is adaptable for fusion reactions as diverse as those involved in cell growth,
membrane repair, cytokinesis and synaptic transmission. A fascinating picture of these robust
nanomachines is emerging.
REVIEWS
NATURE REVIEWS
|
MOLECULAR CELL BIOLOGY VOLUME 7
|
SEPTEMBER 2006
|
631
© 2006 Nature Publishing Group
CAAX box
A C-terminal motif of four
amino acids — cysteine (C),
two aliphatic amino acids (AA)
and then any amino acid (X) —
that is recognized as a
substrate by
farnesyltransferase and
geranylgeranyltransferase I.
Farnesylation
A post-translational
modification of proteins in
which a 15-carbon farnesyl
residue is covalently attached
to the cysteine of a CAAX-box
motif by a thioester bond.
SNARE motifs. Key to understanding the function of
SNAREs in membrane fusion was the discovery that dif-
ferent sets of SNAREs that are present in two opposing
membranes associate into complexes that are sub sequently
disassembled by NSF. Originally, it was assumed that there
was a strict separation between SNAREs on the ‘donor’
compartment and the ‘acceptor’ compartment, which led
to their functional classification as v-SNAREs (vesicle-
membrane SNAREs) or t-SNAREs (target-membrane
SNAREs)
8
. However, this terminology is not useful in
describing homotypic fusion events, and certain SNAREs
function in several transport steps with varying partners.
For example, the S. cerevisiae SNARE
Sec22 functions
in both anterograde and retrograde traffic between the
endoplasmic reticulum (ER) and the Golgi apparatus.
In anterograde traffic, Sec22 is colocalized with Bos1 and
Bet1 on the transport vesicle, but only Bet1 was classified
as a v-SNARE, whereas Bos1 and Sec22 were classified as
t-SNAREs
9
. In retrograde transport, Sec22 is thought to
be the sole functional SNARE on the transport vesicle
10–12
.
A more rigorous and invariant classification comes from
understanding SNARE complexes.
Complex formation is mediated by the SNARE motifs,
and is associated with conformational and free-energy
changes. When SNAREs are monomeric, SNARE motifs
are unstructured. However, when appropriate sets of
SNAREs are combined, the SNARE motifs spontaneously
associate to form helical core complexes of extraordinary
stability (reviewed in
REF. 3). The crystal structures of
two, only distantly related, SNARE core complexes have
revealed a remarkable degree of conservation
13,14
(FIG. 2).
Core complexes are represented by elongated coiled coils
of four intertwined, parallel α-helices, with each helix
being provided by a different SNARE motif. The centre of
the bundle contains 16 stacked layers of interacting side
chains
(FIG. 2). These layers are largely hydrophobic, except
for a central ‘0’ layer that contains three highly conserved
glutamine (Q) residues and one highly conserved arginine
(R) residue. Accordingly, the contributing SNARE motifs
are classified into Qa-, Qb-, Qc- and R-SNAREs
15,16
(FIG. 1a). Functional SNARE complexes that drive mem-
brane fusion are hetero-oligomeric, parallel four-helix
bundles, and each bundle is invariant, requiring one of
each of the Qa-, Qb-, Qc- and R-SNAREs. Indeed, a phy-
logenetic analysis of SNARE sequences from S. cerevisiae,
A. thaliana and mammals showed that these four SNARE
subfamilies are highly conserved and diverged early in
eukaryotic evolution
2,16
.
Due to their amphiphilic nature, SNARE motifs
can also associate in other combinations that result
in helical bundles that are less stable than core com-
plexes. Particularly noteworthy are the complexes that
are formed by the neuronal SNAREs. These include
a Qaaaa complex (an antiparallel four-helix bundle
17
), a
Qabab complex (a parallel four-helix bundle
18
), a Qaabc
complex (a parallel four-helix bundle with some dis-
ordered regions
19,20
) and, surprisingly, an antiparallel
QabcR complex
21,22
. These complexes might not have
the correct membrane topology or they might not con-
tribute sufficient energy to drive membrane fusion. They
therefore probably represent ‘off-pathway’ reactions
(see later; see also
REF. 3).
N-terminal domains. Unlike the conserved SNARE
motifs, there are different types of independently folded
N-terminal domain
23,24
. Qa-SNAREs, and some Qb- and
Qc-SNAREs, have N-terminal antiparallel three-helix
bundles
(FIG. 1a,b). These bundles can vary in length and
are connected to the SNARE motif by a flexible linker.
By contrast, the N-terminal domains of many R-SNAREs
have profilin-like folds
23–26
(FIG. 1c), which are some-
times referred to as longin domains and are also found
Timeline | The discovery of SNAREs and the role of SNARE cycling in membrane fusion
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998
(1988–1990). Mammalian homologues of Sec18 and Sec17, termed
NSF and SNAPs, are identified as soluble factors that are required to
support vesicular transport in a mammalian cell-free system
141,142
.
(1980–1988). Screening for temperature-
sensitive trafficking mutants in
Saccharomyces cerevisiae and
characterizing novel genes identifies Sec17,
Sec18, Sec20 and Sec22
(REFS 139, 140).
(1988–1992). VAMP (also known as
synaptobrevin) and syntaxin are
identified as important constituents of
membranes that participate in synaptic
exocytosis and are proposed to be
receptors for α-SNAP and NSF
143–145
.
(1992–1993). VAMP/synaptobrevin, SNAP-25
and syntaxin are shown to be targets of the
botulinum and tetanus neurotoxins
149–151
.
NSF and α-SNAP are shown to be involved in vesicle
priming, but not in vesicle docking and fusion
38
.
SNAREs are shown to be sufficient
to induce the fusion of artificial
membranes
129
.
(1991–1993). Saccharomyces cerevisiae
genes that are involved in membrane
traffic are found to be homologous to
VAMP/synaptobrevin and syntaxin
146 –148
.
The ATP-driven disassembly of
the synaptic SNARE complex by
NSF and SNAPs is discovered
8
.
The parallel alignment of SNAREs in
opposing membranes is discovered,
which provides the foundation for the
‘zipper’ model of SNARE function
39,40,72
.
NSF, N-ethylmaleimide-sensitive factor; SNAP, soluble NSF attachment protein; SNAP-25, 25-kDa synaptosome-associated protein; SNARE, soluble NSF attachment
protein receptor; t, target membrane; v, vesicle membrane; VAMP, vesicle-associated membrane protein.
VAMP/synaptobrevin, a resident of synaptic vesicles, and
SNAP-25 and syntaxin, both residents of the neuronal plasma
membrane, are identified as membrane receptors for NSF and
SNAPs. From this time, VAMP/synaptobrevin is termed a
v-SNARE, and SNAP-25 and syntaxin are termed t-SNAREs
152
.
The crystal structure of the first
SNARE complex is determined,
which results in a structure-
based reclassification of SNAREs
into Q- and R-SNAREs
13,15
.
REVIEWS
632
|
SEPTEMBER 2006
|
VOLUME 7 www.nature.com/reviews/molcellbio
© 2006 Nature Publishing Group
Qa-SNARE
Qb-SNARE
Qc-SNARE
Qbc-SNARE
R-SNARE
N-terminal
domains
SNARE
motif
Transmembrane
domain
N-terminal domain
of syntaxin-1
PX domain
Longin domain‘Closed’ conformation
of syntaxin-1
abc
Phox-homology (PX)
domain
A conserved domain of ~120
residues that is found in many
proteins. PX domains
preferably bind to
phosphatidylinositol-3,4,5-
trisphosphate, a polyphospho-
inositide lipid that is enriched
in endosomes and vacuoles.
in proteins that are unrelated to SNAREs
5
. R-SNAREs
that have longin domains are conserved among all
eukaryotes, whereas the evolutionarily younger brevin
R-SNAREs lack a folded N-terminal domain and have
only a few amino acids beyond their SNARE motif
5
.
The S. cerevisiae Qc-SNARE
Vam7 is unique in that it
has a
Phox-homology (PX) domain
27
that is responsible
for membrane binding
(FIG. 1c). The presence of further
folds in the N-terminal domains of Qb- and Qc-SNAREs
cannot be excluded, because some of these domains are
divergent in sequence and length.
What is the function of the N-terminal domains?
Some N-terminal domains of the three-helix-bundle
type reversibly associate with the SNARE motif of the
same SNARE to form a ‘closed’ conformation, which
prevents the SNARE motif from forming a SNARE
complex
28,29
(FIG. 1b). However, others cannot assume
closed conformations (reviewed in
REFS 5,24), which
indicates that this conformation is not essential for
SNARE function. The N-terminal domains might
function as recruitment platforms for the binding of
other proteins such as SM (Sec1/Munc18-related)
proteins. SM proteins are a small family of soluble pro-
teins that have a conserved structure and are essential
for fusion
30
. Surprisingly, the crystal structures of two
Qa-SNARE–SM complexes have shown that there are
at least two different binding modes. In the first mode,
the arch-shaped SM protein encloses and stabilizes the
closed SNARE conformation (this conformation was
studied in the Munc18–syntaxin-1 complex; see
FIG. 1b
for the closed conformation of syntaxin-1). In the sec-
ond mode, the interaction is confined to a surface inter-
action with the N-terminal end of the SNARE
30
. It has
recently been proposed that the longin domain of the
S. cerevisiae SNARE Ykt6 catalyses the palmitoylation
of
Vac8, a protein that is involved in vacuole fusion
31
.
However, the profilin fold is shared by proteins that are
unlikely to function as palmitoyltransferases, and there
is no sequence similarity between longin domains and
the known S. cerevisiae palmitoyltransferase families
32
.
Are the N-terminal domains needed for fusion? For
some SNAREs, such as the S. cerevisiae Qa-SNARE Sso1,
the N-terminal domains are essential for cell viability
33
.
However, when the N-terminal domain of Sso1 was
replaced with the Qbc motifs of Sso9, a fused ‘tandem
SNARE’ was created and function was restored
34
. These
findings showed that, at least in this case, the N-terminal
domain is not needed for the recruitment of essential fac-
tors, which is in line with work showing that N-terminal
domains seem to be dispensable for fusion
35
.
SNARE cycling in membrane fusion
Shortly after the discovery that SNAREs are targets of
NSF, it was proposed that fusion is mediated by the action
of NSF on pre-assembled SNARE docking complexes
8
.
As an alternative, some proposed that SNARE assembly
might lead directly to fusion
36,37
. However, only after it
was realized that NSF is not involved in fusion itself
38
,
Figure 1 | The structures of SNAREs. a | The domain structure of the SNARE (soluble N-ethylmaleimide-sensitive factor
attachment protein receptor) subfamilies. Dashed domain borders highlight domains that are missing in some subfamily
members. Qa-SNAREs have N-terminal antiparallel three-helix bundles. The various N-terminal domains of Qb-, Qc- and
R-SNAREs are represented by a basic oval shape. Qbc-SNAREs represent a small subfamily of SNAREs — the SNAP-25
(25-kDa synaptosome-associated protein) subfamily — that contain one Qb-SNARE motif and one Qc-SNARE motif. These
motifs are connected by a linker that is frequently palmitoylated (zig-zag lines in the figure), and most of the members of
this subfamily function in constitutive or regulated exocytosis. In this, and the following figures, the same colour scheme
is used for the SNARE subfamilies (Qa-SNARE, red; Qb-SNARE, light green; Qc-SNARE, dark green; and R-SNARE, blue).
b | The upper panel shows the three-dimensional structure of the isolated N-terminal domain of syntaxin-1
(REF. 137). This
structure is an N-terminal three-helix bundle that is typical of Qa-SNAREs, as well as of some Qb- and Qc-SNAREs. The
lower panel shows the ‘closed’ conformation of syntaxin-1, in which the N-terminal domain of syntaxin-1 (red, as in the
upper panel) is associated with part of its own SNARE motif (beige structure; absent in the upper panel). This structure was
solved as part of the structure of the Munc18–syntaxin-1 complex
138
. c | Three-dimensional structures of the N-terminal
domains of other SNAREs, which exemplify the structural diversity that exists. The upper panel shows the Phox-homology
(PX) domain of the Qc-SNARE Vam7
(REF. 27), which seems to be unique for this particular SNARE. The lower panel shows
the profilin or longin domain of the R-SNARE Ykt6
(REF. 26). The authors are indebted to F. Gräter (Max Planck Institute for
Biophysical Chemistry, Göttingen, Germany) for help in preparing the figure.
REVIEWS
NATURE REVIEWS
|
MOLECULAR CELL BIOLOGY VOLUME 7
|
SEPTEMBER 2006
|
633
© 2006 Nature Publishing Group
a
b
c
–7 –6 –5 –4 –3 –2 –1 0 +1 +2 +3 +4 +5 +6 +7 +8
COPII
(Coatomer protein complex-II).
COPII assembles at the exit
sites of the endoplasmic
reticulum, which results in the
formation of COPII-coated
transport vesicles that are
destined for the cis face of the
Golgi apparatus or for an
intermediate compartment.
and that the SNAREs syntaxin and VAMP (vesicle-
associated membrane protein; also known as synapto-
brevin) are aligned in parallel with their transmembrane
domains next to each other, did it become apparent that
SNARE assembly — rather than disassembly — might
be the driving force behind fusion
39,40
(see TIMELINE).
According to this concept, the ‘zippering’ of the SNARE
motifs from their N-terminal ends towards their
C-terminal membrane anchors clamps the membranes
together and initiates fusion — that is, SNAREs directly
function as fusion catalysts. Although it is still contro-
versial, this model
(FIG. 3) has gained wide acceptance
and will be discussed in detail below.
The status of ‘free’ SNAREs in membranes. The activity
of NSF guarantees that free — that is, uncomplexed
— SNAREs are constantly regenerated. However, the
fact that they are free does not necessarily exclude the
possibility that SNAREs interact with other proteins.
Indeed, numerous diverse proteins have been shown to
bind to specific SNAREs.
What might be the function of these binding pro-
teins? First, they can be involved in SNARE sorting and
recycling by associating with SNAREs during transport
— for example, the transmembrane protein synapto-
physin binds to the neuronal R-SNARE VAMP/synap-
tobrevin and can assist in its sorting
41
. Second, they can
have a role in the recruitment of SNAREs into trafficking
vesicles. For example, the ER–Golgi SNAREs Bet1, Sed5
and Sec22 were reported to bind to the Sec23–Sec24
subcomplex of the coatomer protein complex-II
(COPII)
coat during export from the ER
42
, and VAMP4 interacts
with the
adaptor protein-1 (AP1) complex at the trans-Golgi
network
43
. Third, they can be involved in the formation
of docking complexes. Examples include the early endo-
somal tethering factor EEA1 (early endosomal antigen-1),
which binds to the endosomal SNAREs syntaxin-6 and
syntaxin-13, and the S. cerevisiae multisubunit dock-
ing complexes HOPS/VpsC (homotypic fusion and
vacuole protein sorting/class C vacuolar protein sorting
(Vps) protein complex) and VFT (Vps fifty three), which
bind to the Qc-SNAREs
Vam3 and Tlg1 (t-SNARE
affecting a late Golgi compartment-1), respectively
44,45
.
Last, they can have a role in regulating the capability of
SNARE motifs to enter SNARE complexes. For example,
membrane-anchored VAMP/synaptobrevin was reported
to be unable to form SNARE complexes due to the par-
tial membrane insertion of the membrane-proximal
part of its SNARE motif
46,47
. Activation by other pro-
teins might therefore be required. Candidates for this
role include the Rho GTPase Cdc42
(REF. 48) and the
exocytic Ca
2+
receptor synaptotagmin
47
. By contrast,
the Q-SNAREs syntaxin-1 and SNAP-25 readily form
core complexes with exogenous R-SNAREs in the
plasma membrane, which indicates that these SNAREs
are constitutively active
49
.
It must be stated, however, that for some of these
proteins, and for many other proteins, the evidence that
SNARE binding is specific and functionally relevant is
not compelling. Free SNARE motifs are conformation-
ally adaptable, which increases the chances of adsorp-
tive and non-specific associations with other proteins
in vitro. Furthermore, SNAREs are highly abundant
— for example, syntaxin-1 and SNAP-25 each consti-
tute ~1% of the total brain protein
50
, which is probably
greater than the sum of all of the ion channels and
receptors that are in the neuronal plasma membrane. It
is therefore unsurprising that almost every ion channel
or receptor that has been studied has been reported to
bind syntaxin-1 in a ‘specific’ manner. For the synaptic
SNAREs alone, the number of reported binding proteins
Figure 2 | SNARE core complexes. a | A crystal structure of the neuronal SNARE
(soluble N-ethylmaleimide-sensitive factor attachment protein receptor) core complex.
This complex contains the SNARE motifs of syntaxin-1 (Qa; red), SNAP-25 (25-kDa
synaptosome-associated protein; Qb and Qc; both green), and VAMP (vesicle-
associated membrane protein)/synaptobrevin (R; blue). The C-terminal ends of the
helices, which all point towards the membrane, are orientated to the right. Modified
with permission from
REF. 13 © Macmillan Magazines Ltd. b | A skeleton diagram that
indicates the position of the central layers of interacting side chains (numbered) in the
neuronal SNARE core complex. Cα traces are shown in grey, the helical axes are
highlighted by lines that are the same colour as the helices in part a, and the
superhelical axis is highlighted by a black line. The ‘0’ layer is coloured red and all other
layers are coloured black. Modified with permission from
REF. 13 © Macmillan
Magazines Ltd. c | Overlays of individual layers, which are each shown contained in a
shaded circle, from the neuronal SNARE core complex (grey) and the endosomal
SNARE core complex (coloured). The endosomal SNARE core complex contains
syntaxin-7 (Qa; red), VTI1b (Vps ten interacting-1b; Qb; light green), syntaxin-8 (Qc;
dark green) and VAMP8 (R; blue). The three upper panels exemplify highly asymmetric
layers that include the polar 0 layer and the –3 and +6 layers (the –3 and +6 layers
contain conserved phenylalanines). The lower two panels show the hydrophobic layers
that surround the 0 layer and are also highly conserved. Modified with permission from
REF. 14 © Macmillan Magazines Ltd.
REVIEWS
634
|
SEPTEMBER 2006
|
VOLUME 7 www.nature.com/reviews/molcellbio
© 2006 Nature Publishing Group
Vesicle
Loose trans-SNARE complexes Tight trans-SNARE complexes
Cis-SNARE complexe
s
Acceptor complexes
Free SNARE clusters Disassembly complex
Trans → cis
(fusion)
ADP ATP
Late regulatory proteins
(for example, complexins and synaptotagmin)
Binding proteins?
SM
proteins
α-SNAP
NSF
Qa-SNARE Qb-SNARE
Qc-SNARE R-SNARE NSF
α-SNAP
Acceptor membrane
Adaptor protein-1 (AP1)
complex
The AP1 complex, which is
one of four structurally
related protein complexes,
forms a bridge between the
clathrin coat and membrane
components (cargo) during
the formation of clathrin-
coated vesicles at the trans-
Golgi network.
is greater than 100. Most of the evidence is confined
to qualitative approaches such as pull-downs and co-
immunoprecipitations, which are notorious for yielding
false positives with ‘sticky’ proteins. Quantitative and/or
structural data about these presumed SNARE–target-
protein complexes are therefore largely missing. To vali-
date these interactions, detailed structural information
and a rigorous assessment of their in vivo relevance are
required.
Plasma-membrane SNAREs are not uniformly distrib-
uted in the membrane, but are clustered in nano domains,
the stability of which depends on cholesterol
51–53
. The
homomeric association of SNARE transmembrane
domains has been reported, and this might contribute
to cluster formation
54
. Secretory vesicles selectively dock
and fuse at such clusters
51
. It remains to be seen whether
cluster formation is a hallmark of all SNAREs or is a
speciality of plasma membranes and other membranes
that are rich in steroid lipids. Clustering achieves high
local SNARE concentrations that might result in more
efficient fusion.
Acceptor complexes: intermediates in the fusion pathway?
How does the assembly of four unstructured SNAREs
into a SNARE complex proceed? Detailed studies on
exocytic S. cerevisiae and neuronal SNARE complexes
in vitro have shown that assembly proceeds through a
defined and partially helical Qabc intermediate
55–57
, the
formation of which is rate limiting. Although it is not yet
known whether other SNAREs form such intermediate
acceptor complexes, it is probable that they represent a
key step in the fusion pathway of all SNAREs — that
is, it is likely that assembly is an ordered, sequential
reaction rather than a random collision of four differ-
ent SNARE motifs. Only when an acceptor scaffold is
available in which the N-terminal ends of the SNARE
motifs are structured is the final SNARE able to bind
with biologically relevant kinetics and nucleate the
zippering reaction.
Acceptor complexes are highly reactive and are there-
fore difficult to characterize. For example, in vitro, the
neuronal acceptor complex readily recruits a second
Qa-SNARE, which results in a ‘dead-end’ Qaabc complex.
Figure 3 | The SNARE conformational cycle during vesicle docking and fusion. As an example, we consider three
Q-SNAREs (Q-soluble N-ethylmaleimide-sensitive factor attachment protein receptors) on an acceptor membrane and
an R-SNARE on a vesicle. Q-SNAREs, which are organized in clusters (top left), assemble into acceptor complexes, and this
assembly process might require SM (Sec1/Munc18-related) proteins. Acceptor complexes interact with the vesicular
R-SNAREs through the N-terminal end of the SNARE motifs, and this nucleates the formation of a four-helical trans-
complex. Trans-complexes proceed from a loose state (in which only the N-terminal portion of the SNARE motifs are
‘zipped up’) to a tight state (in which the zippering process is mostly completed), and this is followed by the opening of the
fusion pore. In regulated exocytosis, these transition states are controlled by late regulatory proteins that include
complexins (small proteins that bind to the surface of SNARE complexes) and synaptotagmin (which is activated by an
influx of calcium). During fusion, the strained trans-complex relaxes into a cis-configuration. Cis-complexes are
disassembled by the AAA+ (ATPases associated with various cellular activities) protein NSF (N-ethylmaleimide-sensitive
factor) together with SNAPs (soluble NSF attachment proteins) that function as cofactors. The R- and Q-SNAREs are then
separated by sorting (for example, by endocytosis).
REVIEWS
NATURE REVIEWS
|
MOLECULAR CELL BIOLOGY VOLUME 7
|
SEPTEMBER 2006
|
635