-
How curvature-generating proteins build scaffolds on
membranenanotubes
Simunovic, M., Evergren, E., Golushko, I., Prévost, C., Renard,
H-F., Johannes, L., McMahon, H. T., Lorman, V.,Voth, G. A., &
Bassereau, P. (2016). How curvature-generating proteins build
scaffolds on membranenanotubes. Proceedings of the National Academy
of Sciences, 113(40),
11226-11231.https://doi.org/10.1073/pnas.1606943113
Published in:Proceedings of the National Academy of Sciences
Document Version:Peer reviewed version
Queen's University Belfast - Research Portal:Link to publication
record in Queen's University Belfast Research Portal
Publisher rightsCopyright © 2016 PNAS.This work is made
available online in accordance with the publisher’s policies.
General rightsCopyright for the publications made accessible via
the Queen's University Belfast Research Portal is retained by the
author(s) and / or othercopyright owners and it is a condition of
accessing these publications that users recognise and abide by the
legal requirements associatedwith these rights.
Take down policyThe Research Portal is Queen's institutional
repository that provides access to Queen's research output. Every
effort has been made toensure that content in the Research Portal
does not infringe any person's rights, or applicable UK laws. If
you discover content in theResearch Portal that you believe
breaches copyright or violates any law, please contact
[email protected].
Download date:01. Jul. 2021
https://doi.org/10.1073/pnas.1606943113https://pure.qub.ac.uk/en/publications/how-curvaturegenerating-proteins-build-scaffolds-on-membrane-nanotubes(ef57a30e-43bd-4dce-a1bb-f0c0325d5ca6).html
-
How curvature-generating proteins build sca�olds onmembrane
nanotubesMijo Simunovic 1,2 †, Emma Evergren 3‡, Ivan Golushko 4 ,
Coline Pr évost 1 , Henri-Fran çois Renard 5¶, Ludger Johannes 5
,Harvey T. McMahon 3 , Vladimir Lorman 4 , Gregory A. Voth 2 , and
Patricia Bassereau 1,6
1 Laboratoire Physico Chimie Curie, Institut Curie, PSL Research
University, CNRS UMR168, 75005, Paris, France 2 Department of
Chemistry, Institute forBiophysical Dynamics, James Franck
Institute and Computation Institute, The University of Chicago,
5735 S Ellis Avenue, Chicago, IL 60637, USA 3 MedicalResearch
Council Laboratory of Molecular Biology, Francis Crick Avenue,
Cambridge CB2 0QH, UK 4 Laboratoire Charles Coulomb, UMR 5221
CNRSUniversit é de Montpellier, F-34095, Montpellier, France 5
Institut Curie, PSL Research University, Chemical Biology of
Membranes and Therapeutic Deliveryunit, CNRS UMR3666, INSERM U1143,
F-75248 Paris, France 6 Sorbonne Universit és, UPMC Univ Paris 06,
75005, Paris, France
Submitted to Proceedings of the National Academy of Sciences of
the United States of America
Bin/Amphiphysin/Rvs (BAR) domain proteins control the
curvatureo� ipid membranes in endocytosis, traf�cking, cell
motility, theformation of complex sub-cellular structures, and many
other cel-lular phenomena. They form three-dimensional assemblies,
whichact as molecular sca�olds to reshape the membrane and alter
itsmechanical properties. It is unknown, however, how a
proteinsca�old forms and how BAR domains interact in these
assembliesat protein densities relevant for a cell. In this work,
we em-ploy various experimental, theoretical and simulation
approachesto explore how BAR proteins organize to form a sca�old
ona membrane nanotube. By combining quantitative microscopywith
analytical modeling, we demonstrate that a highly curvingBAR
protein endophilin nucleates its sca�olds at the ends of amembrane
tube, contrary to a weaker curving protein centaurin,which binds
evenly along the tube’s length. Our work implies thatthe nature o�
ocal protein-membrane interactions can a�ect thespeci�c
localization of proteins on membrane-remodeling sites.Furthermore,
we show that amphipathic helices are dispensable informing protein
sca�olds. Finally, we explore a possible molecularstructure of a
BAR-domain sca�old using coarse-grained moleculardynamics
simulations. Together with �uorescence microscopy, thesimulations
show that proteins need only to cover 30–40% ofa tube ’s surface to
form a rigid assembly. Our work providesmechanical and structural
insights into the way BAR proteins maysculpt the membrane as a
high-order cooperative assembly inimportant biological
processes.
protein sca�old | BAR proteins | coarse-grained simulations
IntroductionCurvature o� ipid membranes plays important roles in
the cell.It allows dynamic cellular phenomena, such as tra�cking
orcell division, and it can also mediate the interactions amongmany
membrane-bound proteins (1, 2). Proteins containing
aBin/Amphiphysin/Rvs (BAR) domain participate in
numerousmembrane-curving processes, such as endocytosis,
tra�cking,motility, the formation of T-tubules, cytokinesis, etc.
(3, 4). BARdomains are characterized by a crescent shape whose
curvature,length, and binding a�nity to the membrane are distinct
amongdi�erent members (4-6). Many BAR proteins also contain
am-phipathic helices that shallowly insert into the bilayer.
BAR proteins generate curvature as a combination of (a),adhesive
electrostatic interactions via their BAR domain and (b),the
insertion of amphipathic helices. Additionally, BAR proteinscan
associate into highly ordered assemblies on the membranethus
collectively altering its shape andmechanics (7-10). Preciselyhow
they assemble and a�ect the membrane is argued to dependon the
surface density of proteins, membrane tension, and mem-brane shape
(11). On a �at membrane at a low surface density,BAR proteins can
form strings and a mesh-like network, whichcan give rise to budding
and tubulation (12-16). At a su�cientlyhigh protein density, they
impact themechanical properties of themembrane and stabilize
membrane nanotubes (7, 10, 17-20).
An assembly of BAR proteins on cylindrical membranes hasso far
only been visualized using electron microscopy (EM), e.g.(8, 9,
21). While these studies provide important and detailed
as-sessments of how BAR domainsmay interact with one another
oncurved membranes as a packed protein arrangement, membranetubules
in those experiments were generated typically from highlycharged
liposomes exposed to very high protein concentrations.In the cell,
especially in the context of endocytosis, protein con-centration is
not high enough to induce appreciable spontaneoustubulation, nor
would such a mechanism be bene�cial to the cell.Importantly, a
tightly packed assembly of BAR proteins wouldpreclude the
recruitment of many other proteins required inendocytosis and
tra�cking.
To achieve close packing, protein-protein interactions
wereimplicated to be important, namely the lateral interactions
be-tween neighboring BAR domains in F-BAR proteins (8) or be-tween
N-terminal amphipathic helices in N-BAR proteins (9). Itis unclear
whether BAR proteins in endocytosis and tra�ckingcooperatively
shape the membrane by virtue of speci�c protein-protein
interactions or if they assemble as a result of a more gen-eral
membrane-mediated mechanism. Moreover, it is importantto understand
how BAR proteins assemble at much lower proteinsurface densities
and onmembrane compositions thatmuchmorelikely resemble those found
within the cell.
We hypothesize that BAR proteins can oligomerize on amembrane
nanotube at densities much lower than close packing
Signi�cance
Lipid membranes are dynamic assemblies, changing shape onnano-
to micron-sized scales. Some proteins can sculpt mem-branes by
organizing into a molecular sca�old, dictating themembrane ’s shape
and properties. We combine microscopy,mathematical modeling, and
simulations to explore how BARproteins assemble to form sca�olds on
nanotubes. We showthat the way protein locally deforms the membrane
a�ectswhere it will nucleate before making a sca�old. In this
process,the protein ’s amphipathic helices —which shallowly insert
intothe membrane —appear dispensable. Surprisingly, the
sca�oldforms at low protein density on the nanotube. We simulatea
structure of protein sca�olds at molecular resolution, shed-ding
light on how these proteins may sculpt the membrane tofacilitate
important dynamic events in cells.
-
Fig. 1. Sca�olding by endophilin A2. (A) Endophilin A2 N-BAR
domain (aa1–247) binds to the tube ’s base and forms a sca�old that
continuously growsalong the tube (note the progressive constriction
in the tube radius fromthe GUV toward the OT). White circle = OT.
(B) A kymogram of sca�oldgrowth from the GUV to the bead
(�uorescence dims near the end as thetube buckles in and out o�
ocus). Lipid and protein channels are overlaid.The plot shows
tube-retraction force, f , as a function of time, t . The x-axisof
the kymogram coincides with the x-axis of the plot. (C) Time lapse
of astriated pattern induced by endophilin A2 N-BAR domain. In all:
scale bar,2 μm; GUV, giant unilamellar vesicle; OT, optical trap;
endo, endophilin A2N-BAR domain; t = 0 marks the time when protein
was detected on the tube.
Fig. 2. Sca�olding by N-BAR versus BAR domains. (A) β2 centaurin
BARdomain (aa 1-384) binds evenly along the tube (red: lipid;
green: protein) andcauses a decrease in tube-retraction force, f ,
just like endophilin. Scale bar, 2μm. (B) Dilation of a narrow tube
induced by a sca�old of β2 centaurin BARdomain (overlaid are
�uorescence intensity of the protein on the tube, Itub ,and the
tube radius, r , deduced from lipid �uorescence). (C) The
mechanicsof the reference membrane ( N = 45) and after the
formation of a sca�oldby endophilin A2 WT (endo WT, N = 7) and β2
centaurin (centa, N = 5). Tubeforce, f, measured from the optical
trap; tube radius, r , measured from lipid�uorescence.
Table 1. Radius ( r ) of sca�olded tubes measured from
lipid�uorescence. Mean ±SD (N measurements). Endo WT =
wild-typeendophilin A2 (data from the full length protein and the
N-BARdomain is pooled); endo ΔH0 = endophilin A2 with
truncatedN-terminal helices; endo mut = endophilin A2 N-BAR
domainE37K, D41K.
endo WT endo ΔH0 endo mut centa
r (nm) 9.8 ±2.8 (10) 21.4 ±11.6 (7) 19.9 ±3.0 (7) 42.5 ±7.0
(5)
Fig. 3. Amphipathic helices do not determine the sca�old
initiation site.Shown are force plots (white) overlaid on kymograms
o� ipid �uorescenceof a membrane nanotube (red marker) during
binding and sca�olding byendophilin mutants. As before, the
formation of a sca�old is evident fromtube constriction. Endo ΔH0 =
endophilin A2 with truncated N-terminalhelices; endo mut =
endophilin A2 N-BAR domain E37K, D41K.
Fig. 4. Strongly-curving proteins nucleate at the base of a
pinned and�uctuating tube. Mathematical model: strain energy
variation pro�le, E, as afunction of the axial position on the
tube, z (in percentage of total length),plotted using = 0.25%
(orange) and 0.05% (blue), = 50 kBT , = 100.
Fig. 5. Simulation of N-BAR domains on nanotubes. Shown are
�nalsnapshots of CG MD simulations of membrane tubes coated with
N-BARproteins at the indicated protein surface densities. Scale
bar, 20 nm.
owing to membrane-mediated attractions. We refer to this
struc-ture as a protein sca�old. It is to be noted that the term
sca�old isoften used to describe a single BAR domain, imprecisely
termedthe sca�olding domain. Here, a sca�old represents a
three-dimensional rigid assembly of multiple proteins that
adheresto the membrane and a�ects the shape and properties of
themembrane.
In this work, we combine in vitro reconstitution,
�uorescentmicroscopy, mechanical measurements, and analytical
modeling
-
to describe the mechanism by which BAR proteins assemble
onmembrane nanotubes to form a sca�old. We also demonstratethat
rigid protein sca�olds form at much lower surface densitiesthan
full packing. We simulate the protein sca�old at
molecularresolution using coarse-grained (CG) molecular dynamics
(MD).
Finally, as the relative contribution of BAR domain
versusamphipathic helices in inducing curvature is still highly
debated,we explore how these domains contribute to the sca�old
forma-tion. To this end, we tested three proteins with
well-distinguishedstructural features: endophilin A2 (an N-BAR
protein containingfour amphipathic helices), endophilin A2 mutants,
β2-centaurin(a classical BAR domain with no amphipathic helices),
and epsin1 (a protein that binds membranes via an amphipathic helix
in itsepsin N-terminal homology domain).
ResultsEndophilin sca�old initiates at the base of a tube. To
study theinteractions of BAR proteins with a cylindrical membrane,
weused a previously developed micromanipulation setup (7). In
theexperiment, we pull a nanotube from a giant unilamellar
vesicle(GUV) using optical tweezers. A nanotube connected to
thebase membrane is a typical con�guration characteristic of
someendocytic processes, such as in a clathrin-independent
endocyticmechanism mediated by endophilin (22, 23). The vesicle is
heldby a micropipette whose aspiration pressure sets the
membranetension, implicitly tube radius, in the absence of proteins
(24, 25)(see SI Text). Thus, we have a direct control of the
initial radiusof curvature, which in our case ranges from � 10 nm
to � 100 nm(7). With another micropipette, we inject the protein
near thetube, starting from low vesicle tension. The N-BAR domain
ofthe wild-type endophilin A2 and β2 centaurin (BAR +
pleckstrinhomology domain) were �uorescently labeled so that we
coulddirectly observe their binding to the membrane with
confocalmicroscopy. By measuring the lipid and the protein
�uorescence,we can calculate the tube radius and the protein’s
surface density,respectively (7) (see Fig. S1 and SI Text).
Therefore, at the sametime, we observe how proteins a�ect the shape
of the membrane,while controlling membrane tension and membrane
curvature.
We prepared GUVs using a total lipid brain extract,
supple-mented with 5% PI(4,5)P 2. As such a natural composition
hasnot yet been used for quantitative mechanical measurements
(26,27), we con�rmed that the membrane curvature scales with
GUVtension as theoretically expected for �uid membranes (25)
andthat these vesicles are not undergoing phase separation (28)
(seeSI Text, Figs. S2 and S3).
First, we studied how the N-BAR of endophilin A2 (29, 30)(Fig.
S4) forms a sca�old on a membrane tube, by injecting theprotein at
0.5–2 5 µM (dimeric concentration in the pipette). Notethat due to
di�usion, the concentration of the protein near theGUV is
approximately half that in the pipette (31). Endophilinshowed a
remarkable speci�city for the base of a pulled nanotube,binding
�rst either at the interface with the vesicle or with thetrapped
bead (Fig. 1A). Note that the two interfaces are mor-phologically
equivalent, having the same saddle-like membranegeometry. Out of 59
experiments, endophilin �rst bound to theGUV-tube interface in 53
of them, while also simultaneouslybinding to the interface with the
bead in 27 experiments. In fourcases, endophilin appeared to bind
homogeneously along the tubewhere, possibly, the initial binding
was not recorded su�cientlyfast. Only in the two remaining cases
considered as negative, theprotein �rst bound to a region other
than the interface.
Shortly after binding, the region covered by endophilin
con-tinuously grew along the tube eventually partially or fully
coveringit (Fig. 1 A and B; see SI Text for additional statistics).
In mostcases, the growth of the endophilin sca�old was linear and
itranged from � 20 nm.s−1 to � 300 nm.s−1 (Fig. 1 B , see also
Fig.S5 and Movie S1).
The marked reduction of the lipid �uorescence
intensityunderneath the protein (Fig. 1 A , lipid channel)
indicates that en-dophilin changes the tube radius independently of
GUV tension.Hence, it forms a stable three-dimensional structure
that dictatesthe membrane curvature. Tube constriction has
previously beenobserved with other members of the BAR family (7,
22, 32), al-though the dynamics of sca�old formation has not been
captured.Binding and constriction under the sca�old are concomitant
withthe progressive drop in force required to hold the nanotube
(Fig.1B ). A fully covered tube at low GUV tension imposes no
forceon the optical trap and undergoes buckling (see the
deformationof the tube in the bottom panel of Fig. 1A , also see
Movie S1). Ofnote, in the experiments, the proteins are also bound
to the GUV(see e.g. Fig. 1 A).
We observed no di�erence in the tube-binding behavior be-tween
the full-length endophilin A2 (N-BAR + SH3 domain) andonly its
N-BAR domain, indicating that the location of sca�oldinitiation is
not determined by the protein ’s SH3 domain (Fig. S5).
Interestingly, sometimes at higher injected concentrations(>1
5 µM in the injection pipette), endophilin initially formed
astriated pattern on the nanotube, marked by a brief (few
seconds)beading instability (Fig. 1 C , observed in six out of 31
experi-ments). The striation rapidly coarsened leading to a growth
of thesca�old fromboth bases of the tube. To some extent, this
behavioris reminiscent of the way dynamin binds to membrane
tubes.Dynamin binds in a striated pattern and a�ects the
membraneforce. In the case of dynamin, however, the membrane
forcechanges only after the entire tube is covered with the protein
(33,34), contrary to endophilin, in which case a decrease in the
forceis seen immediately upon binding.
Role of protein subdomains in sca�olding. We then aimed
toexamine how changing the intrinsic curvature and the presence
ofamphipathic helices a�ect the sca�olding dynamics. β2
centaurinprovides a good testing ground, as it is one o� ew BAR
pro-teins without an N-terminal amphipathic helix (35).
Additionally,the BAR domain of centaurin is much shallower than
that ofendophilin, as judged by their atomic models (see SI Text,
Fig.S4). Contrary to endophilin, centaurin bound homogeneouslyalong
the nanotube, with no detectable preference to the neck(Fig. 2 A).
Nevertheless, there was a reduction in the membraneforce during
binding, leading to a buckling instability at lowtension (Fig. 2
A). Importantly, binding of the protein changedthe curvature of the
tube, even though the aspiration pressureremained the same. Figure
2 B shows an example where bindingof β2 centaurin dilates a 30-nm
tube by � 20 nm. Furthermore,once the sca�old forms, either by
centaurin or endophilin, thetube radius remains constant; its
magnitude is characteristic ofthe protein, but independent of GUV
tension (Fig. 2 C ). Namely,the tube sca�olded by centaurin is
approximately four timeswiderthan the one sca�olded by endophilin
(42.5 nm compared to 10nm, see Table 1). This observation is in
line with the di�erence inintrinsic curvatures of their BAR domains
(Fig. S4).
The formation of a sca�old by either endophilin or centaurinalso
drastically changes the mechanics of the membrane, evidentfrom the
systematic reduction in the equilibrium tube force for
alltestedmembrane tensions (Fig. 2C ). Based on previous
analyticalmodeling, the force of a sca�olded tube—characterized by
aconstant radius—is expected to linearly depend on GUV
tension,whereas a bare membrane is expected to have a
square-rootdependence (7, 25). Indeed, membrane force of
protein-coveredtubes in experiments shown in Fig. 2C display a
linear dependenceon tension (Fig. S6), thus con�rming the formation
of a sca�oldby a measurement independent of tube radius.
These experiments demonstrate that both BAR domains thatcontain
membrane inserting amphipathic helices (endophilin)and those that
do not (β2 centaurin) are capable o� orming arigid structure that
controls the curvature of the membrane. They
-
also show that proteins from the same family may bind to
themembrane at di�erent locations (we explore this point in the
nextsection).
To further investigate the role of amphipathic helices versusthe
BAR domain in sca�olding, we constructed two endophilinmutants. In
the �rst, we truncated the N-terminal amphipathichelix of the
full-length endophilin A2 (endo △H0). In the sec-ond, we mutated
one glutamate and one aspartate from themembrane-binding region of
endophilin A2 N-BAR domain intolysines (E37K, D41K) (endo mut),
which enhances the bindingstrength of the BAR domain to the
membrane. Both variantsconstricted the tube starting from an
interface (Fig. 3, red �uo-rescence) and decreased the force (Fig.
3, white plot) and tuberadius (Table 1), in the same manner as the
WT. This observa-tion con�rms that the N-terminal amphipathic
helices are notnecessary for the formation of the sca�old or,
interestingly, forthe preferential binding to the tube’s base in
these experiments,although the sca�olding rate appears slower (Fig.
3).
Finally, we tested the full-length epsin 1, another impor-tant
endocytic protein, which participates in the initial stagesof
clathrin-mediated endocytosis (36). Epsin does not containa BAR
domain; instead, it binds and bends the membrane viaan amphipathic
helix. There was a clear mechanical e�ect uponthe injection of
epsin 1, characterized by a systematic reductionin both the
equilibrium tube force and the tube radius for awide range of
membrane tensions, indicating that the proteininduces positive
spontaneous curvature (7) (Fig. S7). Similarly tocentaurin, the
constriction did not start from the base; rather itappeared
homogenous along the tube length. Unlike endophilinand centaurin,
the force never decreased to zero and so we neverobserved buckling.
The square-root scaling of the force withmembrane tension (Fig. S6)
indicates that no sca�old forms, evenat very high protein
concentration (ten-fold higher than minimalendo WT concentration
that makes a sca�old). In summary,amphipathic helices alone may
remodel the membrane, as in thecase of epsin. However, the
anisotropic BAR domain is criticalfor forming a rigid sca�old.
Pinning a �uctuating tube determines the protein ’s bindingsite.
So far, we demonstrated that BAR proteins lacking am-phipathic
helices may form sca�olds just as N-BAR proteins,however it is
still unclear what determines the nucleation site ofthe protein.
Our experiments cannot provide a general mecha-nism to answer this
question and so we developed a mathemat-ical model of BAR proteins
interacting with a membrane tube.Several models have already been
proposed for an equivalentsystem (7, 37), but those models did not
capture the location ofprotein nucleation. We extend these models
in two ways. First,we generalize the protein-membrane interactions
by assumingthat the proteins induce a local perturbation, expressed
in termsof a tension or a pressure variation. Second, instead of
takingperiodic boundary conditions, wemodel amembrane tube pinnedat
its ends assuming that the radial displacement of the bilayer
isstrongly limited at the one end by the optical trap and on the
otherby the vesicle.
As we show in the SI Text in detail, we decompose the freeenergy
into the costs of (a) bending and (b) stretching the mem-brane,
supplemented by (c) a term accounting for membrane-protein
interactions, and (d) the energy associated with a pointforce
keeping the membrane tubular (Eq. S14) (25, 37, 38).Solving the
equation in the limit o� ow protein concentration, weobtain the
mechanical strain energy variation (Eq. S17) inducedby
membrane-protein interactions, whose minima essentially in-dicate
the binding sites of the protein. Importantly, the shapeof this
function strongly depends on the protein-induced localtension (or
curvature) perturbation. When taking a local tensionvariation of
0.25%, the energy pro�le has a minimum at eachof the tube’s ends
separated by a very high energy barrier at the
tube’s center (Fig. 4). Reducing the local perturbation
�ve-foldto 0.05% lowers the barrier to
-
aggregation previously predicted for N-BAR proteins and, to
aweaker degree, spherical particles (12-14, 41). Under con�ne-ment
(on a �at or spherical surface), the proteins pack into amesh(12),
however it appears that a tubular surface directs the proteinsinto
a helix, with 7–8 N-BAR domains making a full helical turn(Fig.
5).
We note that in CG MD simulations the helix contiguouslywraps
the tubule at 30–40% protein coverage, in excellent agree-ment with
the experimentally measured sca�old density. Onceattaining this
density, the proteins cease to exchange neighborsand the helix
becomes quasi-static (Fig. 5, see SI Text, Fig. S9).
DiscussionTwo related curvature-generating proteins can initiate
a sca�oldat di�erent membrane locations, as shown by our in vitro
recon-stituted system. Namely, an N-BAR protein endophilin
nucleatesat the tube’s ends, whereas a BAR protein centaurin binds
evenlyalong it. Our mathematical modeling predicts that speci�c
bind-ing to the saddle-shaped neck of a pinned and �uctuating
mem-brane tube is a consequence of strong local membrane
pertur-bations. An important conclusion from these observations is
thatthe nature o� ocal protein-membrane interactions can a�ect
thespeci�c initial localization of proteins on curved membranes
and,thus, the dynamics of their assembly on
membrane-remodelingsites.
Although the complexity ofmulti-protein interactionsmay di-vert
the nucleation preference of BAR proteins in a cell, previousin
vivo studies of endocytosis seem to very well agree with
our�ndings. Immunoelectron microscopy of endophilin on
clathrin-coated pits in cells at endogenous protein concentrations
showedthat endophilin indeed sits at the base of the clathrin coat
(42).In the same study, in cells treated with a non-hydrolysable
GTP,which form long dynamin-covered tubes, endophilin was againonly
found at the base of the coat (42). By contrast, dynamin wasfound
all along the tubule’s length.
Endophilin interacts with other proteins in a dynamic
way.Namely, the tubulation e�ciency and the amount of
dynaminrecruited to GUVs or lipid tubules are signi�cantly
increasedby endophilin, and vice versa (42, 43). Furthermore,
acutelyperturbing endophilin using antibodies against the SH3- or
theBAR-domain stalled the formation of clathrin-coated pits
beforethe sculpting of a narrow neck and the saddle (44, 45).
Hence,endophilin could potentially play important roles in
directingother endocytic proteins to their binding site.
Concerning protein ’s subdomains, BAR domain appears cru-cial
for the formation of a rigid sca�old. As previously demon-strated
on a �at membrane, local membrane deformations me-diate the
interactions among BAR proteins and induce their as-sembly. The
anisotropic shape of the BAR domain likely furtherfacilitates an
ordered packing and the formation of a sca�old.Therefore, a BAR
domain is indeed a sca�olding domain, al-though not because a
single protein imprints its shape on themembrane, but owing to a
collective e�ect imposed by an orderedmembrane-mediated helical
assembly. Moreover, amphipathichelices appear dispensable in
sca�olding; however, their role isstill important in facilitating
protein recruitment to the mem-brane (22) and in increasing the
membrane ’s spontaneous cur-vature (Table 1). They may also have a
role at the molecular levelto help properly orient the BAR domains
into a rigid sca�old,evidenced by the wide distribution of tubular
radii when they aretruncated (Table 1) (22), agreeing with previous
work (9).
Importantly, our results show that a sca�old can form
atmuchlower surface densities than full packing. Dense protein
packingwould be problematic for endocytosis. According to
previoussimulations, the shape of a basic unit of a BAR-domain
lattice onthe membrane a�ects the radius of the sca�old (18).
Therefore,the radius of the tubule sca�olded by the same protein
would
be variable, depending on the way it formed the lattice,
whichseems unfavorable for endocytosis and tra�cking that requirea
tight curvature control. Indeed, tubule radii from di�erent invitro
studies were infrequently di�erent for the same protein.
Forexample, tubule radii formed and sca�olded by amphiphysin 1
invitro (measured between themembranemidplanes) were found tobe 21
nm (35) and � 11 nm (46), both based on EM imaging, com-pared to 7
nmmeasured by �uorescencemicroscopy (7). Based onour combined
experimental and simulation data, under proteinconcentrations much
lower than used in EM imaging in vitro,BAR proteins do not build
lattices on pre-formed tubes. Instead,they only cover 30–45%of the
surface (depending on the protein),forming a stable and a rigid
sca�old with constant curvature, inresemblance to in vivo EM images
in which membrane tubuleswere created in the cell by endogenous
protein concentrations(42, 47). In turn, this assembly provides
structural integrity forendocytosis and leaves su�cient membrane
area for the bindingof accessory proteins crucial in the process
(42, 46, 48).
Based on our work, we can propose di�erent biologically
rel-evant purposes for the N-BAR domain sca�olds. First, in
endo-cytosis, they constrict the membrane tube between the
endocyticvesicle and the underlying membrane, thus reducing the
energybarrier for scission by dynamin (33) or by elongation forces
(22).Second, highly curving proteins like endophilin are
speci�callyrecruited to the neck and so in clathrin-dependent
endocytosis,where endophilin recruits dynamin to the tube (43), the
scissionsite will be highly localized to the base of the coat.
Third, sca�oldsprovide a powerful control of membrane curvature
that maybe used in forming complex cellular architectures, such as
inthe formation of T-tubules or the maintenance of
mitochondrialshape, which require N-BAR proteins amphiphysin 2 (49)
andendophilin B1 (50), respectively. The subtle di�erences in
struc-tures of these proteins give rise to a complexity in
intracellulararchitectures and the highly dynamic behavior of the
membrane.These di�erences are also likely the key way of
modulatingthe function and localization of BAR proteins. We also
expectthat in the near future, the higher-order organization of
BARproteins will be shown crucial in additional important
membrane-remodeling phenomena.
MethodsPulling nanotubes and making protein sca�olds. GUVs (95%
total lipidbrain extract (26), 5% PI(4,5)P 2 , supplemented with
0.1% di-stearoyl phos-phatidyl ethanolamine-PEG(2000)-biotin and 1%
BODIPY TR ceramide) wereprepared by electroformation on Pt-wires
over night at 4 °C in a salt-containing bu�er (51). To pull a tube,
the GUV was aspirated in a mi-cropipette, brought in contact with a
streptavidin-coated optically trappedbead then gently pulled away.
Proteins were injected near the tube withanother micropipette. The
aspiration pressure sets the membrane tension
and the tube radius, r , in the absence of proteins, as ,
whereis membrane sti�ness and is membrane tension (7, 24, 52-54).
The tubeforce, f , was measured by video-microscopy as , where
isthe trap sti�ness and and are the current and the equilibrium
beadpositions, respectively. The r (in the presence or absence of
proteins) wasmeasured from lipid �uorescence as , where and arethe
�uorescence intensities o� ipids in the tube and in the GUV,
respectively,and = 200±50 nm is a previously measured calibration
constant (7, 32).
CGMD simulations. We used a solvent-free three-site CG
lipidmodel (55)and a 26-site elastic network model of an N-BAR
domain dimer of endophilinA1 (9), with protein-membrane
interactions modeled using a Lennard-Jonespotential as described
previously (12). We simulated N-BARs on a lipid bilayertube (150 nm
in length and 20 nm in diameter interacting with its periodicimages
in the tube direction) at 5%, 10%, 30%, and 40% surface
coverage.The simulations were carried at constant number of
molecules, box volumeand temperature ( NVT ) for � 30 million time
steps at a time step of 12 fsusing LAMMPS (56).
Acknowledgements.We thank A. Callan-Jones for insightful
discussions. M.S. and G.A.V.
were supported by the National Institutes of Health (grant
R01-GM063796)and the National Science Foundation (Xsede
computational resources grantTG-MCA94P017, supercomputer Stampede),
P.B. and L.J. by the AgenceNationale pour la Recherche
(ANR-11BSV201403 to P.B. and ANR-09BLAN283to H.F.R. and L.J.), L.J.
of the European Research Council advanced grant
-
(project 340485), E.E. and H.T.M by the Medical Research Council
UK (grantU105178795). M.S. was funded in part by the Chateaubriand
fellowship,France and Chicago Collaborating in the Sciences grant,
and the UniversityParis Diderot. The P.B. group belongs to the CNRS
consortium CellTiss, P.B.
and L.J. groups to the Labex CelTisPhyBio (ANR-11-LABX0038), and
to ParisSciences et Lettres (ANR-10-IDEX-0001-02). The V.L. group
belongs to theLabex NUMEV.
1. McMahon HT & Gallop JL (2005) Membrane curvature and
mechanisms of dynamic cellmembrane remodelling. Nature
438(7068):590-596.
2. Phillips R, Ursell T, Wiggins P, & Sens P (2009) Emerging
roles for lipids in shapingmembrane-protein function. Nature
459(7245):379-385.
3. Mim C &Unger VM (2012) Membrane curvature and its
generation by BAR proteins. TrendsBiochem Sci 37(12):526-533.
4. Qualmann B, Koch D, & Kessels MM (2011) Let's go bananas:
revisiting the endocytic BARcode. EMBO J 30(17):3501-3515.
5. Suetsugu S, Toyooka K, & Senju Y (2010) Subcellular
membrane curvature mediated by theBAR domain superfamily proteins.
Semin Cell Dev Biol 21(4):340-349.
6. Rao Y & Haucke V (2011) Membrane shaping by the
Bin/amphiphysin/Rvs (BAR) domainprotein superfamily. Cell Mol Life
Sci 68(24):3983-3993.
7. Sorre B , et al. (2012) Nature of curvature coupling of
amphiphysin with membranes dependson its bound density. Proc Natl
Acad Sci U S A 109(1):173-178.
8. Frost A , et al. (2008) Structural basis of membrane
invagination by F-BAR domains. Cell132(5):807-817.
9. Mim C , et al. (2012) Structural basis of membrane bending by
the N-BAR protein endophilin.Cell 149(1):137-145.
10. Shi Z & Baumgart T (2015) Membrane tension and
peripheral protein density mediatemembrane shape transitions. Nat
Commun 6:5974.
11. Simunovic M, Voth GA, Callan-Jones A, & Bassereau P
(2015) When Physics Takes Over:BAR Proteins and Membrane Curvature.
Trends Cell Biol 25(12):780-792.
12. Simunovic M, Srivastava A, & Voth GA (2013) Linear
aggregation of proteins on themembrane as a prelude to membrane
remodeling. Proc Natl Acad Sci U S A 110(51):20396-20401.
13. Simunovic M & Voth GA (2015) Membrane tension controls
the assembly of curvature-generating proteins. Nat Commun
6:7219.
14. Noguchi H (2016) Membrane tubule formation by banana-shaped
proteins with or withouttransient network structure. Sci Rep
6:20935.
15. Traub LM (2015) F-BAR/EFC Domain Proteins: Some Assembly
Required. Dev Cell35(6):664-666.
16. McDonald NA, Vander Kooi CW, Ohi MD, & Gould KL (2015)
Oligomerization but NotMembrane Bending Underlies the Function of
Certain F-BAR Proteins in Cell Motility andCytokinesis. Dev Cell
35(6):725-736.
17. Zhu C, Das SL, & Baumgart T (2012) Nonlinear sorting,
curvature generation, and crowdingof endophilin N-BAR on tubular
membranes. Biophys J 102(8):1837-1845.
18. Yu H & Schulten K (2013) Membrane sculpting by F-BAR
domains studied by moleculardynamics simulations. PLoS Comput Biol
9(1):e1002892.
19. Cui H , et al. (2013) Understanding the role of amphipathic
helices in N-BAR domain drivenmembrane remodeling. Biophys J
104(2):404-411.
20. Ramesh P , et al. (2013) FBAR syndapin 1 recognizes and
stabilizes highly curved tubularmembranes in a concentration
dependent manner. Sci Rep 3:1565.
21. Pang X , et al. (2014) A PH domain in ACAP1 possesses key
features of the BAR domain inpromoting membrane curvature. Dev Cell
31(1):73-86.
22. Renard HF , et al. (2015) Endophilin-A2 functions in
membrane scission in clathrin-independent endocytosis.Nature
517(7535):493-496.
23. Boucrot E , et al. (2015) Endophilin marks and controls a
clathrin-independent endocyticpathway.Nature 517(7535):460-465.
24. Kwok R & Evans E (1981) Thermoelasticity o� arge
lecithin bilayer vesicles. Biophys J35(3):637-652.
25. Derenyi I, Julicher F, & Prost J (2002) Formation and
interaction of membrane tubes. PhysRev Lett 88(23):238101.
26. Yu S , et al. (2006) Identi�cation of phospholipid molecular
species in porcine brain extractsusing high mass accuracy of 4.7
Tesla Fourier transform ion cyclotron resonance massspectrometry. B
Kor Chem Soc 27(5):793-796.
27. Rawicz W, Olbrich KC, McIntosh T, Needham D, & Evans E
(2000) E�ect of chain lengthand unsaturation on elasticity o� ipid
bilayers. Biophys J 79(1):328-339.
28. Sorre B , et al. (2009) Curvature-driven lipid sorting needs
proximity to a demixing point andis aided by proteins. Proc Natl
Acad Sci U S A 106(14):5622-5626.
29. Gallop JL , et al. (2006) Mechanism of endophilin N-BAR
domain-mediated membranecurvature. EMBO J 25(12):2898-2910.
30. Capraro BR , et al. (2013) Kinetics of endophilin N-BAR
domain dimerization andmembraneinteractions. J Biol Chem
288(18):12533-12543.
31. Simunovic M, Lee KY, & Bassereau P (2015) Celebrating
Soft Matter's 10th anniversary:screening of the calcium-induced
spontaneous curvature o� ipid membranes. Soft
Matter11(25):5030-5036.
32. Prevost C , et al. (2015) IRSp53 senses negative membrane
curvature and phase separatesalong membrane tubules. Nat Commun
6:8529.
33. Morlot S , et al. (2012) Membrane shape at the edge of the
dynamin helix sets location andduration of the �ssion reaction.
Cell 151(3):619-629.
34. Roux A , et al. (2010) Membrane curvature controls dynamin
polymerization. Proc Natl AcadSci U S A 107(9):4141-4146.
35. Peter BJ , et al. (2004) BAR domains as sensors of membrane
curvature: the amphiphysinBAR structure. Science
303(5657):495-499.
36. Chen H , et al. (1998) Epsin is an EH-domain-binding protein
implicated in clathrin-mediatedendocytosis. Nature
394(6695):793-797.
37. Monnier S, Rochal SB, Parmeggiani A, & Lorman VL (2010)
Long-range protein cou-pling mediated by critical low-energy modes
of tubular lipid membranes. Phys Rev Lett105(2):028102.
38. Golushko IY, Rochal SB, & Lorman VL (2015) Complex
instability of axially compressedtubular lipid membrane with
controlled spontaneous curvature. Eur Phys J E Soft
Matter38(10):112.
39. Shlomovitz R, Gov NS, & Roux A (2011) Membrane-mediated
interactions and the dynamicsof dynamin oligomers on membrane
tubes. New J Phys 13.
40. Simunovic M , et al. (2013) Protein-mediated transformation
o� ipid vesicles into tubularnetworks. Biophys J
105(3):711-719.
41. Saric A & Cacciuto A (2012) Fluid membranes can drive
linear aggregation of adsorbedspherical nanoparticles. Phys Rev
Lett 108(11):118101.
42. Sundborger A , et al. (2011) An endophilin-dynamin complex
promotes budding of clathrin-coated vesicles during synaptic
vesicle recycling. J Cell Sci 124(Pt 1):133-143.
43. Meinecke M , et al. (2013) Cooperative recruitment of
dynamin and BIN/amphiphysin/Rvs(BAR) domain-containing proteins
leads to GTP-dependent membrane scission. J Biol
Chem288(9):6651-6661.
44. Andersson F, Low P, & Brodin L (2010) Selective
perturbation of the BAR domain ofendophilin impairs synaptic
vesicle endocytosis. Synapse64(7):556-560.
45. Ringstad N , et al. (1999) Endophilin/SH3p4 is required for
the transition from early to latestages in clathrin-mediated
synaptic vesicle endocytosis. Neuron 24(1):143-154.
46. Takei K, Slepnev VI, Haucke V, & De Camilli P (1999)
Functional partnership betweenamphiphysin and dynamin in
clathrin-mediated endocytosis. Nat Cell Biol 1(1):33-39.
47. Ferguson SM , et al. (2009) Coordinated actions of actin and
BAR proteins upstream ofdynamin at endocytic clathrin-coated pits.
Dev Cell 17(6):811-822.
48. Daumke O, Roux A, & Haucke V (2014) BAR domain sca�olds
in dynamin-mediatedmembrane �ssion. Cell 156(5):882-892.
49. Lee E , et al. (2002) Amphiphysin 2 (Bin1) and T-tubule
biogenesis in muscle. Science297(5584):1193-1196.
50. Karbowski M, Jeong SY, & Youle RJ (2004) Endophilin B1
is required for the maintenanceof mitochondrial morphology. J Cell
Biol 166(7):1027-1039.
51. Montes LR, Alonso A, Goni FM, & Bagatolli LA (2007)
Giant unilamellar vesicles electro-formed from native membranes and
organic lipid mixtures under physiological conditions.Biophys J
93(10):3548-3554.
52. Cuvelier D, Derenyi I, Bassereau P, & Nassoy P (2005)
Coalescence of membrane tethers:experiments, theory, and
applications. Biophys J 88(4):2714-2726.
53. Helfrich W (1973) Elastic properties o� ipid bilayers:
theory and possible experiments. ZNaturforsch C 28(11):693-703.
54. Evans EA (1983) Bending elastic modulus of red blood cell
membrane derived from bucklinginstability in micropipet aspiration
tests. Biophys J 43(1):27-30.
55. Srivastava A & Voth GA (2013) A Hybrid Approach for
Highly Coarse-grained Lipid BilayerModels. J Chem Theory Comput
9(1):750-765.
56. Plimpton S (1995) Fast Parallel Algorithms for Short-Range
Molecular-Dynamics. J ComputPhys 117(1):1-19.