Lipid membrane adhesion and fusion driven by designed, minimally multivalent hydrogen-bonding lipids

Article abstract of DOI:10.1021/ja9072657

Cyanuric acid (CA) and melamine (M) functionalized lipids can form membranes that exhibit robust hydrogen-bond driven surface recognition in water, facilitated by multivalent surface clustering of recognition groups and variable hydration at the lipid-water interface. Here we describe a minimal lipid recognition cluster: three CA or M recognition groups are forced into proximity by covalent attachment to a single lipid headgroup. This trivalent lipid system guides recognition at the lipid-water interface using cyanurate-melamine hydrogen bonding when incorporated at 0.1-5 mol percent in fluid phospholipid membranes, inducing both vesicle-vesicle binding and membrane fusion. Fusion was accelerated when the antimicrobial peptide magainin was used to anchor trivalent recognition, or when added exogenously to a preassembled lipid vesicle complex, underscoring the importance of coupling recognition with membrane disruption in membrane fusion. Membrane apposition and fusion were studied in vesicle suspensions using light scattering, FRET assays for lipid mixing, surface plasmon resonance, and cryo-electron microscopy. Recognition was found to be highly spatially selective as judged by vesicular adhesion to surface patterned supported lipid bilayers (SLBs). Fusion to SLBs was also readily observed by fluorescence microscopy. Together, these studies indicate effective and functional recognition of trivalent phospholipids, despite low mole percentage concentration, solvent competition for hydrogen bond donor/acceptor sites, and simplicity of structure. This novel designed molecular recognition motif may be useful for directing aqueous-phase assembly and biomolecular interactions.

Full text of DOI:10.1021/ja9072657

Published on Web 10/30/2009
Lipid Membrane Adhesion and Fusion Driven by Designed,
Minimally Multivalent Hydrogen-Bonding Lipids
Mingming Ma, Yun Gong, and Dennis Bong*
Department of Chemistry, The Ohio State UniVersity, Columbus, Ohio 43210
Received August 27, 2009; E-mail: bong@chem.osu.edu
Abstract: Cyanuric acid (CA) and melamine (M) functionalized lipids can form membranes that exhibit
robust hydrogen-bond driven surface recognition in water, facilitated by multivalent surface clustering of
recognition groups and variable hydration at the lipid-water interface. Here we describe a minimal lipid
recognition cluster: three CA or M recognition groups are forced into proximity by covalent attachment to
a single lipid headgroup. This trivalent lipid system guides recognition at the lipid-water interface using
cyanurate-melamine hydrogen bonding when incorporated at 0.1-5 mol percent in fluid phospholipid
membranes, inducing both vesicle-vesicle binding and membrane fusion. Fusion was accelerated when
the antimicrobial peptide magainin was used to anchor trivalent recognition, or when added exogenously
to a preassembled lipid vesicle complex, underscoring the importance of coupling recognition with membrane
disruption in membrane fusion. Membrane apposition and fusion were studied in vesicle suspensions using
light scattering, FRET assays for lipid mixing, surface plasmon resonance, and cryo-electron microscopy.
Recognition was found to be highly spatially selective as judged by vesicular adhesion to surface patterned
supported lipid bilayers (SLBs). Fusion to SLBs was also readily observed by fluorescence microscopy.
Together, these studies indicate effective and functional recognition of trivalent phospholipids, despite low
mole percentage concentration, solvent competition for hydrogen bond donor/acceptor sites, and simplicity
of structure. This novel designed molecular recognition motif may be useful for directing aqueous-phase
assembly and biomolecular interactions.
Introduction
derivatives are well-known to assemble in low dielectric
solvents13-16 or when hydrophobically buried,17 derivative
Selective aqueous phase hydrogen bond-driven molecular
recognition faces the challenge of solvent competition. While
there are many solutions to this problem in Nature, there are
few non-native designed systems known.1-8 We have previously
reported our aqueous-phase studies on cyanurate (CA) and
melamine (M) phospholipid derivatives that drive membrane
chemistry via hydrogen-bonding interactions between the CA
and M headgroups.9 There have been numerous studies on the
CA-M system; though the parent compounds readily cocrys-
tallize in a hydrogen-bonding network in water10-12 and
assembly generally does not occur in hydrogen-bonding
solvents.12,18-20 Indeed, without the phospholipid module,
monoderivatived cyanurate (CA) and melamine (M) neither
assemble nor inhibit interactions between the lipids in water
(Figure 1).9 Surface multivalency of an assembled membrane
provides binding avidity that greatly enhances the hydrogen-
bonding interaction at the lipid-water interface.17,21-23 The
requisite density of CA/M modules for recognition was indicated
(12) Zerkowski, J. A.; Seto, C. T.; Whitesides, G. M. J. Am. Chem. Soc.
1992, 114, 5473–5475.
(1) Oshovsky, G. V.; Reinhoudt, D. N.; Verboom, W. Angew. Chem.,
Int. Ed. 2007, 46, 2366–2393.
(13) Kondo, S.-I.; Hayashi, T.; Sakuno, Y.; Takezawa, Y.; Yokoyama, T.;
Unno, M.; Yano, Y. Org. Biomol. Chem. 2007, 5, 907–916.
(14) Madueno, R.; Raisanen, M. T.; Silien, C.; Buck, M. Nature 2008,
454, 618–621.
(2) Brunsveld, L.; Vekemans, J. A.; Hirschberg, J. H.; Sijbesma, R. P.;
Meijer, E. W. Proc. Natl. Acad. Sci. U S A 2002, 99, 4977–4982.
(3) Kato, Y.; Conn, M. M.; Rebek, J., Jr. Proc. Natl. Acad. Sci. U.S.A.
1995, 92, 1208–1212.
(15) Prins, L. J.; De Jong, F.; Timmerman, P.; Reinhoudt, D. N. Nature
2000, 408, 181–184.
(4) Nowick, J. S.; Cao, T.; Noronha, G. J. Am. Chem. Soc. 1994, 116,
3285–3289.
(16) Yagai, S.; Kinoshita, T.; Higashi, M.; Kishikawa, K.; Nakanishi, T.;
Karatsu, T.; Kitamura, A. J. Am. Chem. Soc. 2007, 129, 13277–13287.
(17) Kimizuka, N.; Kawasaki, T.; Hirata, K.; Kunitake, T. J. Am. Chem.
Soc. 1998, 120, 4094–4104.
(5) Nowick, J. S.; Chen, J. S. J. Am. Chem. Soc. 1992, 114, 1107–1108.
(6) Dervan, P. B.; Doss, R. M.; Marques, M. A. Curr. Med. Chem.:
Anticancer Agents 2005, 5, 373–387.
(18) ten Cate, M. G. J.; Huskens, J.; Crego-Calama, M.; Reinhoudt, D. N.
Chem.sEur. J. 2004, 10, 3632–3639.
(7) Ghadiri, M. R.; Granja, J. R.; Milligan, R. A.; McRee, D. E.;
Khazanovich, N. Nature 1993, 366, 324–327.
(19) Whitesides, G. M.; Simanek, E. E.; Mathias, J. P.; Seto, C. T.; Chin,
D.; Mammen, M.; Gordon, D. M. Acc. Chem. Res. 1995, 28, 37–44.
(20) Zerkowski, J. A.; MacDonald, J. C.; Seto, C. T.; Wierda, D. A.;
Whitesides, G. M. J. Am. Chem. Soc. 1994, 116, 2382–2391.
(21) Ariga, K.; Kunitake, T. Acc. Chem. Res. 1998, 31, 371–378.
(22) Kawasaki, T.; Tokuhiro, M.; Kimizuka, N.; Kunitake, T. J. Am. Chem.
Soc. 2001, 123, 6792–6800.
(8) Bong, D. T.; Clark, T. D.; Granja, J. R.; Ghadiri, M. R. Angew. Chem.,
Int. Ed. 2001, 40, 988–1011.
(9) Ma, M.; Paredes, A.; Bong, D. J. Am. Chem. Soc. 2008, 130, 14456–
14458.
(10) Ranganathan, A.; Pedireddi, V. R.; Rao, C. N. R. J. Am. Chem. Soc.
1999, 121, 1752–1753.
(11) Zerkowski, J. A.; MacDonald, J. C.; Seto, C. T.; Wierda, D. A.;
Whitesides, G. M. J. Am. Chem. Soc. 1994, 116, 2382–2391.
(23) Mammen, M.; Chio, S.-K.; Whitesides, G. M. Angew. Chem., Int. Ed.
1998, 37, 2755–2794.
9
10.1021/ja9072657 CCC: $40.75  2009 American Chemical Society
J. AM. CHEM. SOC. 2009, 131, 16919–16926 16919

A R T I C L E S
Ma et al.
Figure 1. Structures of compounds used in this study (CA-PE and M-PE have been previously reported). For TMM, the sequence of magainin is shown
in bold and K* is lysine(acetamidobenzamide).
by diminished interaction at CA-PE/M-PE concentrations of less
than 70% in fluid phase (ePC) membranes. Notably, when CA-
PE and M-PE were diluted in membranes with gel-phase
phospholipids (DPPC), 30% of CA/M lipid was sufficient for
vesicle binding,9,24 presumably due to clustering of CA/M
headgroups upon phase separation of CA/M-PE lipids from the
gel-phase lipids.25,26 These findings prompted our investigation
of the minimal valency required for detectable lipid-lipid
binding. This exploration was guided by the notion that each
lipid can only interact with its surrounding neighbors at any
given moment; thus, recognition must be facilitated by nearest
neighbors. Separation of hydrogen-bonding (HB) groups by just
one lipid (50 mol% concentration in ePC) abrogates lipid-lipid
binding, implying that clustering of HB groups is essential.9
We synthesized lipids in which three CA or M groups were
forced into proximity by covalent attachment to the same
phospholipid and found that this minimal design imparts robust
molecular recognition and function at the lipid-water interface
(Figure 2).
into synthetic lipid membranes, and their ability to guide surface
recognition and fusion was evaluated. In addition, the trichloride
headgroup (6) was coupled to the N-terminus of the membrane-
active peptide magainin;27-29 thioether formation with melamine-
thiol yielded a peptide conjugate (TMM, 3) that combines both
molecular recognition (TM) and membrane activation, in
analogy to previously reported systems24,30 (Scheme 1). We
investigated intermembrane lipid recognition in three different
contexts: (1) lipid-lipid binding (TCA-PE/TM-PE), (2) lipid-
lipid binding with membrane activation by magainin (TCA-
PE/TM-PE + mag), and (3) lipid-peptide binding (TCA-PE/
TMM). These systems were designed to explore the nature of
TCA/TM headgroup binding and the extent of membrane fusion
with and without a known disruptive element such as an
antimicrobial peptide.
Vesicle-Vesicle Binding. Lipid films were prepared that
contained TCA-PE at 0.1-5 mol% in egg phosphatidylcholine
(ePC) and TM-PE at the same concentration in 20% phosphati-
dylglycerol lipid (POPG) and ePC. TM-PE liposomes have a
propensity for self-aggregation that is completely suppressed
by the inclusion of negatively charged POPG in the preparation.
These lipid films were hydrated in phosphate buffer saline (PBS)
at pH 6.7 or 7.4 and extruded through 100 nm pore polycar-
bonate membranes to produce large unilamellar vesicles (LUVs)
that appeared to be monodisperse and nonaggregated, as judged
by dynamic light scattering (DLS) and cryo-electron microscopy
(cryo-EM). While electrostatically driven binding between
oppositely charged membranes and macromolecules in water
Results and Discussion
Design and Synthesis. Tris(hydroxymethyl)aminomethane)
was readily functionalized to symmetrically install three terminal
chlorides and one terminal carboxylate, compound 6 (Scheme
1). Amide coupling of the carboxylate to 1-palmitoyl-2-oleoyl-
sn-glycero-3-phosphoethanolamine lipid (POPE) yielded a
trichloride lipid that was easily transformed to trivalent melamine
(TM-PE, 1) or cyanurate (TCA-PE, 2) lipids via thioether
formation, with three hydrogen bonding heterocycles per lipid
headgroup (Figure 1, Scheme 1). These lipids were incorporated
(27) Oren, Z.; Shai, Y. Biopolymers 1998, 47, 451–63.
(28) Wieprecht, T.; Apostolov, O.; Seelig, J. Biophys. Chem. 2000, 85,
187–198.
(24) Gong, Y.; Ma, M.; Luo, Y.; Bong, D. J. Am. Chem. Soc. 2008, 130,
6196–205.
(29) Zasloff, M. Nature 2002, 415, 389–395.
(30) Gong, Y.; Luo, Y.; Bong, D. J. Am. Chem. Soc. 2006, 128, 14430–
14431.
(25) Stottrup, B. L.; Keller, S. L. Biophys. J. 2006, 90, 3176–3183.
(26) Veatch, S. L.; Keller, S. L. Biophys. J. 2003, 85, 3074–3083.
9
16920 J. AM. CHEM. SOC. VOL. 131, NO. 46, 2009

Lipid Membrane Adhesion and Fusion
A R T I C L E S
Figure 2. (A) Schematic illustration of how membrane anchored lipids trifunctionalized with CA or M (symbolized as red and gray wedges) might direct
lipid membrane apposition and fusion. (B) Possible modes of intermembrane hydrogen bonding, based on known CA/M assembly topologies: tape (left) and
rosette (right). The actual hydrogen-bonded structures formed have not been established.
Scheme 1
(a) TFA, H2O, CH2Cl2; (b) chloroacetic anhydride, DIEA, CH2Cl2; (c) LiOH, MeOH; (d) 4, HBTU, DIEA, DMF; (e) 95% TFA, 5% H2O and (F) DMF,
CHCl3, DIEA.
is well-established,31 we observe a phenomenon upon mixing
of TCA-PE and TM-PE membranes consistent with molecular
recognition between neutral components. Notably, despite
repulsive potentials of -13 and -23 mV for the TCA-PE and
TM-PE LUVs, respectively, mixing the vesicle populations in
a 1:1 ratio resulted in rapid doubling of size as judged by DLS
(Figure 3). At pH ≈ 7, the cyanuric acid and melamine groups
should be neutral, thus the charge on TCA-PE and TM-PE
should be -1 due to the phosphate. Replacing TCA-PE and
TM-PE with the singly charged POPG yields similar LUV ꢀ
potentials (-11 and -23 mV, respectively) suggesting that the
charge on each lipid is approximately -1, as expected.
Additionally, increasing salt concentration slightly enhances
vesicle-vesicle aggregation, consistent with a neutral rather than
electrostatic interaction.
We probed this interaction further by surface plasmon
resonance (SPR) experiments in which one of the LUV
populations was surface bound on an SPR substrate while the
complementary LUVs were flowed over the modified surface
(Figure 4). While control LUVs resulted in insignificant changes
in refractive units, complementary LUVs produced strong,
concentration-dependent signals consistent with surface deposi-
(31) Pantazatos, D. P.; Pantazatos, S. P.; MacDonald, R. C. J. Membr. Biol.
2003, 194, 129–139.
9
J. AM. CHEM. SOC. VOL. 131, NO. 46, 2009 16921

A R T I C L E S
Ma et al.
that the designed aqueous-phase interaction between TCA/TM-
PE lipids can also drive selective membrane merger, though
this occurs with contents leakage.46 Lipid mixing was taken to
represent membrane fusion and was followed by monitoring
loss of FRET between NBD-PE and Rh-DHPE lipids in an
acceptor vesicle upon fusion with an unlabeled vesicle.47 We
examined three distinct topologies for membrane reactions
mediated by TCA and TM: (1) intermembrane lipid-lipid
binding, (2) intermembrane lipid-lipid binding with membrane-
binding peptide added as a third component, and (3) intermem-
brane lipid-peptide binding. Though mixing the TCA-PE and
TM-PE LUVs resulted in aggregation (Figure 3), no lipid mixing
was observed, even when a large excess of the nonlabeled
vesicle was used (Figure S1 of the Supporting Information),
suggesting that membrane fusion is arrested in the docking stage
(Figure 5A). Addition of magainin peptide to the docked system
triggered rapid lipid mixing, indicating that disruptive peptide-
membrane binding may facilitate membrane fusion. The identi-
cal experiment was carried out with TMM (Figure 1) replacing
TM-PE and magainin; reaction of these vesicles with 5% TCA-
PE LUVs also resulted in efficient lipid mixing. Fusion with
TMM was concentration dependent, with a minimum surface
concentration of 2% required when the reacting membrane
contained TCA-PE at 2% (Figure 5B). Introduction of the TCA
or TM headgroup alone (without membrane anchor) to the TMM
fusion system resulted in significant inhibition of lipid mixing,
suggesting that the soluble trivalent headgroups were capable
of blocking surface binding sites and suppressing membrane
apposition (Figure 5C). Interestingly, inhibition with TM was
more effective than with TCA, though both elicited a decrease
in lipid mixing. While the origin of this difference is unclear,
the possibility of selective molecular recognition between
designed small molecules in aqueous milieu is intriguing; we
are investigating these interactions further. We previously
reported that LUVs of 100% M-PE and CA-PE fused efficiently,
so we compared the efficiency of fusion when the 100% M-PE
LUVs were replaced with 5% TM-PE in a POPG membrane.
Surface dehydration by hydrogen bonding has not yet been
identified as a major mechanism for native membrane fusion,
but appears to be an effective strategy that is enhanced when
combined with membrane disruption by a peptide anchor. Fusion
was observed when 5% TM-PE LUVs replaced 100% M-PE
LUVs, but both reactions were markedly accelerated by addition
of magainin peptide as catalyst, yielding similar lipid mixing
rates (Figure 5D). These findings are consistent with the notion
that extensive dehydration is necessary for fusogenic activation
via surface H-bonding; TM-PE/CA-PE binding likely results
in a smaller contact area and thus requires peptide catalyst for
productive docking. With regard to recognition, the similar
fusion rates produced by 5% TM-PE and 100% M-PE in the
presence of magainin indicate that the covalent cluster of three
Figure 3. Size change of vesicles as a function of vesicle aggregation or
fusion, measured by DLS. (Top) TCA-PE LUVs reacted with TM-PE LUVs;
(Middle) Same as top, in presence of magainin; (Bottom) TCA-PE LUVs
reacted with TMM bound to POPG LUVs. Traces represent: TCA-PE
(---); TM-PE and TMM/POPG (- -); mixed LUVs after 30 min
equilibration (s).
tion (Figure 4A); this behavior was similar regardless of which
lipid was on the surface and which was in suspension. Binding
dropped off sharply when TM-PE LUV concentrations below
1% but was still detectable at 0.1% (TCA-PE constant at 0.3%
in surface-bound LUVs, Figure 4B). Interestingly, layer-by-layer
deposition was not observed. This suggests that the TCA-PE
and TM-PE lipids become concentrated at the interface between
the apposing membranes, leaving only nonbinding phosphati-
dylcholine exposed. Furthermore, concentration changes only
affected on-rates, with no apparent desorption observed even
at the lower limit of detection, consistent with an increase in
multivalency/binding avidity32 following initial docking through
migration of recognition lipids to the LUV-LUV interface.
Undetectable off-rates could also arise from noncovalent reac-
tion. We examined the possibility of membrane fusion catalyzed
by low concentrations of TCA/TM recognition using standard
FRET dilution assays for lipid mixing.47
Vesicle-Vesicle Fusion. Synthetic vesicle fusion systems
driven by molecular recognition between coiled-coil peptides33,34
and proteins derived from the SNARE synaptic vesicle fusion
machinery,35-38 nucleic acid,39-41 and small molecules9,24,30
have been previously reported, as well as systems driven by
metal complexation42-44 and electrostatics.31,45 We have found
(32) Nieba, L.; Krebber, A.; Plu¨ckthun, A. Anal. Biochem. 1996, 234, 155–
165.
(33) Robson Marsden, H.; Elbers, N. A.; Bomans, P. H.; Sommerdijk, N. A.;
Kros, A. Angew. Chem., Int. Ed. 2009, 48, 2330–2333.
(34) Kashiwada, A.; Matsuda, K.; Mizuno, T.; Tanaka, T. Chem.sEur. J.
2008, 14, 7343–7350.
(35) Rothman, J. E.; Sollner, T. H. Science 1997, 276, 1212–1213.
(36) Parlati, F.; Weber, T.; McNew, J. A.; Westermann, B.; Sollner, T. H.;
Rothman, J. E. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 12565–12570.
(37) Fix, M.; Melia, T. J.; Jaiswal, J. K.; Rappoport, J. Z.; You, D.; Sollner,
T. H.; Rothman, J. E.; Simon, S. M. Proc. Natl. Acad. Sci. U. S. A.
2004, 101, 7311–7316.
(42) Lim, C. W.; Crespo-Biel, O.; Stuart, M. C. A.; Reinhoudt, D. N.;
Huskens, H.; Ravoo, B. J. Proc. Natl. Acad. Sci. U. S. A. 2007, 104,
6986–6991.
(43) Waggoner, T. A.; Last, J. A.; Kotula, P. G.; Sasaki, D. Y. J. Am.
Chem. Soc. 2001, 123, 496–497.
(38) Wang, T.; Smith, E. A.; Chapman, E. R.; Weisshaar, J. C. Biophys. J.
2009, 96, 4122–4131.
(44) Mart, R. J.; Liem, K. P.; Wang, X.; Webb, S. J. J. Am. Chem. Soc.
2006, 128, 14462–14463.
(39) Chan, Y.-H. M.; van Lengerich, B.; Boxer, S. G. Biointerphases 2008,
3, 17–21.
(45) Marchi-Artzner, V.; Gulik-Krzywicki, T.; Guedeau-Boudeville, M.-
A.; Gosse, C.; Sanderson, J. M.; Dedieu, J.-C.; Lehn, J.-M. ChemP-
hysChem 2001, 2, 367–376.
(40) Chan, Y. H.; van Lengerich, B.; Boxer, S. G. Proc. Natl. Acad. Sci.
U.S.A. 2009, 106, 979–84.
(46) Ellens, H.; Bentz, J.; Szoka, F. C. Biochemistry 1984, 23, 1532–1538.
(47) Struck, D. K.; Hoekstra, D.; Pagano, R. E. Biochemistry 1981, 20,
4093–4099.
(41) Stengel, G.; Zahn, R.; Hook, F.J. Am. Chem. Soc. 2007, 129,
9584–5.
9
16922 J. AM. CHEM. SOC. VOL. 131, NO. 46, 2009

Lipid Membrane Adhesion and Fusion
A R T I C L E S
Figure 4. Concentration dependent SPR traces of complementary LUV interactions. (A) Total lipid concentration is shown; trivalent lipids are at 5% of
total (ePC and POPG) in both surface bound and suspension LUVs. TM-PE LUVs flowed over chip with TCA-PE LUVs bound. (B) Total lipid concentration
in suspension was maintained at 60 µM with mol% of TM-PE varied as labeled. Surface bound LUVs had 0.3% TCA-PE for each run.
Figure 5. Lipid mixing assay for membrane fusion.47 NBD-PE and Rh-HDPE were incorporated at 1.5% in the TCA-PE containing LUV, TM, and TCA
represent soluble headgroup only, without phospholipid. (A) 3 systems: 5% TMM/POPG/ePC LUVs reacted with 5%TCA-PE/ePC LUVs (b), 5% TM-
PE/POPG/ePC LUVs reacted with 5% TCA-PE/ePC and 5% magainin (2), 5% TM-PE/POPG/ePC reacted with 5% TCA-PE/ePC ([); (B) concentration
dependence: POPG/ePc reacted with 2% TCA-PE/ePC LUVs at 1% ([), 2% (1), 3% (2), 4% (9), and 5% (b) TMM; (C) inhibition: 3% TMM/POPG/ePC
LUVs reacted with 3% TCA-PE/ePC LUVs (b), inhibited with 5 equiv. TCA (- -), inhibited with 5 equivalents TM (---), POPG/ePC LUVs reacted with
3% TCA-PE/ePC LUVs with 3% magainin ([); (D) trivalent/monovalent comparison: 100% M-PE LUVs and 97% CA-PE LUVs with 5% magainin (4)
and same, without magainin (-4-), 5% TM-PE in POPG/ePC LUVs reacted with 97% CA-PE LUVs with 5% magainin (b) and same, without magainin
(-b-).
melamine rings in TM-PE can produce similar docking effects
as an entire surface of M-PE.
closely approximate a lamellar phase membrane. We used cryo-
TEM to examine 5% TM-PE and 5% TCA-PE LUVs for signs
of hexagonal phase formation and vesicular adhesion before and
after mixing. The reactant LUVs alone appeared to be nonag-
gregating, monodisperse populations of LUVs (Figure S2 of
the Supporting Information), while the images of the LUV
mixture consistently indicated lamellar phase LUVs (Figure 6).
Cryo-Electron Microscopy. Hydrogen bonding lipids are
known to induce lamellar to hexagonal phase transitions,48,49
as we found with CA/M-PE LUVs.9 Given the low mole
percentage of trivalent lipids in the vesicles studied, we
anticipated that the behavior of the membrane should more
(48) Epand, R. M. Methods Mol. Biol. 2007, 400, 15–26.
(49) Boggs, J. M. Biochim. Biophys. Acta 1987, 906, 353–404.
9
J. AM. CHEM. SOC. VOL. 131, NO. 46, 2009 16923

A R T I C L E S
Ma et al.
Figure 6. Cryo-TEM of 5% TM-PE/ePC LUVs reacted with 5% TCA-
PE/ePC LUVs; (left) adherent vesicles, (right) adherent and possibly fused
vesicles. Scale bar ) 100 nm.
Vesicular aggregates in the mixture were observed as expected,
as well as many structures that appeared to be captured
immediately postfusion, with elongated form approximately
twice the diameter of a single LUV and a constriction around
the center, possibly from the expansion of a fusion stalk. This
observation runs counter to the lack of vesicular fusion detected
by fluorescence (Figure 5) and suggested that the aggregated
vesicles may be metastable with regard to fusion. We postulated
that the sharp and extreme temperature decrease inflicted on
the sample by plunging into liquid ethane slush during freezing
for cryo-EM analysis triggered rapid fusion of the vesicles
docked by TCA/TM-PE recognition, resulting in the postfusion
vesicle structures. To address this possibility, we formed docked
vesicles as previously, with fluorescent lipids to follow lipid
mixing, and immersed this sample in liquid nitrogen until frozen.
Upon thawing, we found the fluorescence signature of fusion,
which was absent in the reactant and unfunctionalized vesicles
(Figure S3 of the Supporting Information); multiple freeze-thaw
cycles induced fusion in all LUVs, as expected given the cell-
membrane rupturing effect of freeze-thaw procedures.
Though this experiment does not duplicate the conditions of
cryo sample preparation, the partial fusion resulting from one
cycle of freeze-thaw suggests heightened membrane instability
in the docked state and supports the notion that cryo-TEM
imaging has captured some assemblies shortly after fusion.
Vesicle Adhesion with Supported Lipid Bilayers. Supported
lipid bilayers (SLBs) are a powerful tool for studying lipid
membrane interactions.50 We constructed SLBs with a 0-5%
concentration gradient of TCA-PE across a surface that was
patterned with 50 µm2 fibronectin grids that act as diffusion
corrals.51 These SLBs were used to examine the ability of TCA/
TM-PE recognition to direct spatially selective LUV-SLB
deposition on the micrometer scale. Fluorescence microscopy
imaging revealed that TM-PE LUV deposition reproduced the
surface pattern of TCA-PE with high fidelity, to the extent that
adhesion to the unfunctionalized SLB was undetectable (Figure
7). We also monitored deposition of TM-PE LUVs with TCA-
PE absent from the bilayer, as well as deposition of simple
POPG LUVs onto the TCA-PE SLB; in both cases, no vesicle
adhesion was detectable. In fact, adhesion of TM-PE LUVs on
the TCA-PE SLB following treatment with POPG LUVs
produced an identical result to TM-PE LUV deposition without
prior POPG deposition. These data strongly support the notion
of spatially selective membrane apposition driven by binding
Figure 7. Fluorescence microscopy of a supported lipid bilayer (SLB),
formed on a fibronectin grid (scale bar ) 50 µm) with sharp gradients of
Texas Red lipid from left to right and TCA-PE from right to left. (Top)
Both sides of the SLB were treated with POPG LUVs containing Oregon
Green lipid (OG-PE); (Middle) both sides were treated with TM-PE/POPG
LUVs containing OG-PE; (Bottom) OG fluorescence across the SLB for
top image (black) and middle image (green).
Figure 8. Fluorescence microscopy of a vesicle-bound SLB on a 50 µm
feature fibronectin grid. The SLB was labeled with OG-PE and TCA-PE;
vesicles were labeled with TR-PE and TM-PE. Surface was photobleached
and FRAP monitored in both the SLB (green) and vesicle (red) layers. (Left)
OG-PE fluorescence channel immediately after photobleaching; (Left,
Center) 4 min after; (Right, Center) TR-PE channel immediately after
photobleaching; (Right) 4 min after.
of TCA-PE and TM-PE. The adhered LUVs (labeled with TR-
DHPE) exhibited markedly less lipid mobility than the SLB.
While the SLB fluorophore exhibited significant fluorescence
recovery after photobleaching (FRAP), the adsorbed layer did
not recover after extended monitoring (Figure 8). This dif-
ferential mobility suggests the absence of SLB-LUV fusion, as
expected from vesicle-vesicle fusion experiments (Figure 5)
though it is possible that the FRAP experiment is less sensitive
to this process. Vesicle fusion with the SLB should lead to
similar FRAP in both channels; as it is not observed with
vesicle-bound TR-DHPE, then it is likely that TR-DHPE is part
of an unfused, multivalently tethered vesicle. The numerous
multivalent TCA/TM-PE interactions clustered at the SLB/LUV
interface present a significant diffusion barrier, unlike DNA-
tethered vesicle systems,52 possibly due to higher surface
concentration. The requirement for FRAP in the adhered LUV
layer (assuming no membrane fusion) is essentially desorption/
resorption, and our SPR data indicate that this process is
extremely slow (Figure 4). Thus, it is likely that the hydrogen
(50) Chan, Y. H.; Boxer, S. G. Curr. Opin. Chem. Biol. 2007, 11, 581–
587.
(52) Yoshina-Ishii, C.; Chan, Y.-H. M.; Johnson, J. M.; Kung, L. A.; Lenz,
(51) Kam, L.; Boxer, S. G. J. Am. Chem. Soc. 2000, 122, 12901–12902.
P.; Boxer, S. G. Langmuir 2006, 22, 5682–5689.
9
16924 J. AM. CHEM. SOC. VOL. 131, NO. 46, 2009

Lipid Membrane Adhesion and Fusion
A R T I C L E S
Table 1. Conditions for LUV-SLB Reaction
SLB (in ePC)
LUV (in ePC)
lipid mixing
A
B
C
D
E
F
G
H
I
5% POPG
25% POPG, 1% TMM
25% POPG, 1% TMM
25% POPG, 1% Mag
20% POPG, 1% Mag, 5% TM-PE
25% POPG
20% POPG, 5% TM-PE
1% TMM alone
1% Mag alone
(-)
(+)
(-)
(+)
(-)
(+)
(-)
(-)
(-)
(-)
5% TCA-PE
5% TCA-PE
5% TCA-PE
5% POPG
5% TCA-PE
5% TCA-PE
5% TCA-PE
5% POPG
20% POPG, 1% Mag, 5% TM-PE
25% POPG, 1% Mag
J
5% POPG
bonding lipids involved in the surface contact on both the LUV
and SLB are immobile while the fluorescently labeled lipids
remain mobile in the continuous SLB.
Vesicle Fusion with Supported Lipid Bilayers. LUV-SLB
fusion38 was studied using the same lipid mixing FRET assay
as in suspension.47 Treatment with TM-PE or TMM function-
alized LUVs should result in vesicle docking and fusion,
depending on conditions (Figure 5). Lipid mixing would dilute
the surface bound NBD C6-HPC/Rh-DHPE FRET pair and
dequench the FRET donor (NBD); we monitored both NBD
and Rh fluorescence channels and evaluated fusogenic condi-
tions using ten experiments (A-J, Table 1). SLB imaging
generally tracked with LUV-LUV suspension results. Both
TCA/TM recognition groups had to be present to produce a
lipid mixing signal while control experiments (A, C, E, G-J,
Table 1) yielded minimal signal changes (Figure 9). Replace-
ment of recognition groups with simple charges did not result
in detectable reaction, again ruling out electrostatic interactions
as the main driver of binding. One notable deviation from
LUV-LUV and LUV-SLB experiments (Figures 5 and 7) was
the positive fusion signal observed with TM-PE and TCA-PE
interactions. Our previous experiments indicated that this system
exhibited surface adhesion without fusion and vesicle fusion
experiments did not even suggest partial lipid mixing that could
be attributed to hemifusion. Under LUV-SLB fusion conditions
(Figures 9 and 10F), strong lipid mixing was observed. This
again points toward the metastability of the docked system with
regard to fusion or hemifusion (these experiments do not
distinguish the two), as suggested by cryo-TEM data (Figure
6). Variation in stoichiometry, lipid composition, and fluoro-
phore system between LUV-SLB fusion, LUV-LUV fusion,
and LUV-SLB adhesion experiments could cause a range of
membrane reactivity and detectable response.24,40,53 Overall,
these data indicate a lipid recognition system in which there is
only a slender mechanistic separation of surface recognition and
membrane activation and merger, thus minor perturbations in
Figure 10. (Top) Schematic illustration of vesicle-SLB fusion, followed
by dilution of NBD/Rh FRET in the SLB. (Below) NBD fluorescence
microscopy of SLBs containing 2% NBD-C6 HPC, 2%Rh-DHPE in ePC.
Images A-F correspond to conditions described in Table 1.
the docked system may spontaneously trigger disruption and
lipid mixing.
Conclusions
Our examination of designed lipid recognition at the
bilayer-water interface has yielded findings along two funda-
mental lines: molecular recognition and membrane merger.
While selective membrane fusion requires both recognition and
disruption, these functions need not be performed by the same
molecule. Trivalent lipid-lipid binding can induce membrane
apposition, but in the vesicle-vesicle context, no fusion or lipid
mixing. Addition of a membrane-disrupting peptide as a third
component (TCA-PE + TM-PE + Mag) results in lipid mixing
with high efficiency (Figures 5 and 10). Similarly, fusion was
observed when one of the components was membrane anchored
with the same peptide (TCA-PE + TMM). Given the fusoge-
nicity of the three-component system, it appears that membrane
activation need not be precisely at the site of molecular
recognition. Though the surface concentration of H-bonding
trivalent lipids is relatively low, the docked vesicular aggregate
is unstable with regard to fusion, possibly due to headgroup
H-bonding which facilitates the formation of nonbilayer fusion
intermediates.48,49 Indeed, the metastability of TCA/TM lipid
docked membranes may explain why fusion is not observed in
Figure 9. Total integrated NBD fluorescence change on SLB surface after
20 min of reaction and surface washing. Experimental conditions are shown
in Table 1. All conditions have 1.5% each NBD-PE and Rh-PE in the SLB.
Reactions and measurements were performed in triplicate.
(53) Malinin, V. S.; Haque, M. E.; Lentz, B. R. Biochemistry 2001, 40,
8292–8299.
9
J. AM. CHEM. SOC. VOL. 131, NO. 46, 2009 16925

A R T I C L E S
Ma et al.
LUVs, but observed in LUV-SLB binding and possibly during
cryo-TEM sample preparation. Although mechanistic questions
remain, these experiments further illustrate the general principles
of membrane merger and demonstrate a new designed functional
recognition motif.
hydration repulsion forces54-56 between membranes to induce
both spatially selective surface adhesion and membrane merger.
The biophysics of this binding interaction in water, as well as
the range of aqueous-phase assembly chemistry that may be
mediated by cyanurate and melamine groups, remains an
intriguing topic with potential relevance to the development of
new functional materials.
These studies have revealed that low membrane concentra-
tions of the trivalent CA/M lipids or peptides retain the robust
molecular recognition properties found with surfaces composed
entirely of monovalent CA/M recognition lipids, though mem-
brane activation upon binding is diminished. The minimal nature
of headgroup design and synthetic simplicity stands in contrast
to the effectiveness of recognition function. Recognition appears
to occur between neutral components: though it is conceivable
that a pKA shift induced at the lipid-water interface could
generate a negatively ionized cyanuric acid and a positively
charged melamine, we find that the contribution of electrostatic
interactions is minor to nonexistent in our membrane-anchored
TCA/TM system. Thus, we conclude that the main driving force
for recognition is not charge complementation, but rather, a more
subtle hydrogen-bond donor-acceptor pattern recognition be-
tween largely neutral species in competing aqueous solvent.
Moreover, this interaction is sufficiently strong to overcome
Acknowledgment. This work was supported in part by grants
NSF-0927778, NSF-0747194 (CAREER), and the Institute for
Materials Research at the Ohio State University. We thank the Ohio
Bioproducts Innovation Center for instrumentation support.
Supporting Information Available: Detailed synthetic and
analytical procedures, compound characterization, additional
cryo-EM, and fluorescence data. This material is available free
of charge via the Internet at http://pubs.acs.org.
JA9072657
(54) McIntosh, T. J. Chem. Phys. Lipids 1996, 81, 117–131.
(55) Cho, J. K.; Kim, D.-W.; Namgung, J.; Lee, Y.-S. Tetrahedron Lett.
2001, 42, 7443–7445.
(56) Klayman, D. L.; Woods, T. S. J. Org. Chem. 1974, 39, 1819–1823.
9
16926 J. AM. CHEM. SOC. VOL. 131, NO. 46, 2009