Bilayer vesicles of amphiphilic cyclodextrins: Host membranes that recognize guest molecules

Article abstract of DOI:10.1002/chem.200400905

A family of amphiphilic cyclodextrins (6, 7) has been prepared through 6-S-alkylation (alkyl=n-dodecyl and n-hexadecyl) of the primary side and 2-O-PEGylation of the secondary side of α-, β-, and γ-cyclodextrins (PEG = poly(ethylene glycol)). These cyclodextrins form nonionic bilayer vesicles in aqueous solution. The bilayer vesicles were characterized by transmission electron microscopy, dynamic light scattering, dye encapsulation, and capillary electrophoresis. The molecular packing of the amphiphilic cyclodextrins was investigated by using small-angle X-ray diffraction of bilayers deposited on glass and pressure-area isotherms obtained from Langmuir monolayers on the air-water interface. The bilayer thickness is dependent on the chain length, whereas the average molecular surface area scales with the cyclodextrin ring size. The alkyl chains of the cyclodextrins in the bilayer are deeply interdigitated. Molecular recognition of a hydrophobic anion (adamantane carboxylate) by the cyclodextrin vesicles was investigated by using capillary electrophoresis, thereby exploiting the increase in electrophoretic mobility that occurs when the hydrophobic anions bind to the nonionic cyclodextrin vesicles. It was found that in spite of the presence of oligo(ethylene glycol) substituents, the β-cyclodextrin vesicles retain their characteristic affinity for adamantane carboxylate (association constant K_a = 7.1 × 10³M^-1), whereas γ-cyclodextrin vesicles have less affinity (K_a = 3.2 × 10³M^-1), and α-cyclodextrin or non-cyclodextrin, nonionic vesicles have very little affinity (K_a ≈100M ^-1). Specific binding of the adamantane carboxylate to β-cyclodextrin vesicles was also evident in competition experiments with β-cyclodextrin in solution. Hence, the cyclodextrin vesicles can function as host bilayer membranes that recognize small guest molecules by specific non-covalent interaction.

Full text of DOI:10.1002/chem.200400905

FULL PAPER
Bilayer Vesicles of Amphiphilic Cyclodextrins: Host Membranes That
Recognize Guest Molecules
Patrick Falvey,[a] Choon Woo Lim,[b] Raphael Darcy,*[a] Tobias Revermann,[c]
Uwe Karst,[c] Marcel Giesbers,[d] Antonius T. M. Marcelis,[d] Adina Lazar,[e]
Anthony W. Coleman,[e] David N. Reinhoudt,[b] and Bart Jan Ravoo*[b]
Abstract: A family of amphiphilic cy-
clodextrins (6, 7) has been prepared
through 6-S-alkylation (alkyl=n-dode-
cyl and n-hexadecyl) of the primary
side and 2-O-PEGylation of the secon-
dary side of a-, b-, and g-cyclodextrins
(PEG=poly(ethylene glycol)). These
cyclodextrins form nonionic bilayer
vesicles in aqueous solution. The bilay-
er vesicles were characterized by trans-
mission electron microscopy, dynamic
light scattering, dye encapsulation, and
capillary electrophoresis. The molecu-
lar packing of the amphiphilic cyclo-
dextrins was investigated by using
small-angle X-ray diffraction of bilay-
ers deposited on glass and pressure–
area isotherms obtained from Lang-
muir monolayers on the air–water in-
terface. The bilayer thickness is de-
pendent on the chain length, whereas
the average molecular surface area
scales with the cyclodextrin ring size.
The alkyl chains of the cyclodextrins in
the bilayer are deeply interdigitated.
Molecular recognition of a hydropho-
bic anion (adamantane carboxylate) by
the cyclodextrin vesicles was investigat-
ed by using capillary electrophoresis,
thereby exploiting the increase in elec-
trophoretic mobility that occurs when
the hydrophobic anions bind to the
nonionic cyclodextrin vesicles. It was
found that in spite of the presence of
oligo(ethylene glycol) substituents, the
b-cyclodextrin vesicles retain their
characteristic affinity for adamantane
carboxylate (association constant Ka =
7.1ꢁ103 mꢀ1), whereas g-cyclodextrin
vesicles have less affinity (Ka =3.2ꢁ
103 mꢀ1), and a-cyclodextrin or non-cy-
clodextrin, nonionic vesicles have very
little affinity (Ka ꢁ100mꢀ1). Specific
binding of the adamantane carboxylate
to b-cyclodextrin vesicles was also evi-
dent in competition experiments with
b-cyclodextrin in solution. Hence, the
cyclodextrin vesicles can function as
host bilayer membranes that recognize
small guest molecules by specific non-
covalent interaction.
Keywords: amphiphiles · cyclodex-
trins
· membranes · molecular
recognition · vesicles
Introduction
[a] Dr. P. Falvey,+ Dr. R. Darcy
Centre for Synthesis and Chemical Biology of the Conway Institute
Department of Chemistry, National University of Ireland
University College Dublin, Belfield, Dublin 4 (Ireland)
Fax : (+353)1-716-2127
Amphiphilic cyclodextrins are cyclic oligo(a-(1–4)-glucopyr-
anosides) modified with hydrophobic and hydrophilic sub-
stituents that aggregate into a variety of lyotropic phases in
water.[1–5] The hydrophobic groups drive hydrophobic aggre-
gation of the amphiphiles, while the hydrophilic groups are
required to guarantee sufficient water solubility. The type
and stability of the lyotropic phases critically depend on the
nature and number of hydrophobic and hydrophilic substitu-
ents, the balance between hydrophobic and hydrophilic
groups, and factors such as temperature, concentration, and
ionic strength. A particularly interesting example of aggre-
gation of amphiphilic cyclodextrins in water is the formation
of bilayer vesicles composed entirely of (modified) cyclodex-
trins. Cyclodextrin vesicles consist of bilayers of cyclodex-
trins, in which the hydrophobic “tails” are directed inward
and the hydrophilic macrocycle “head groups” are facing
water, thereby enclosing an aqueous interior (Figure 1A
and B). We and others have recently described vesicles com-
posed entirely of nonionic,[6] anionic,[7] and cationic[8] amphi-
E-mail: raphael.darcy@ucd.ie
[b] Dr. C. W. Lim,+ Prof. D. N. Reinhoudt, Dr. B. J. Ravoo
Supramolecular Chemistry and Technology
MESA+ Institute for Nanotechnology, University of Twente
P.O. Box 217, 7500 AE Enschede (The Netherlands)
Fax : (+31)53-489-4645
E-mail: b.j.ravoo@utwente.nl
[c] T. Revermann, Prof. U. Karst
Chemical Analysis
MESA+ Institute for Nanotechnology, University of Twente
P.O. Box 217, 7500 AE Enschede (The Netherlands)
[d] M. Giesbers, Dr. A. T. M. Marcelis
Laboratory of Organic Chemistry
Wageningen University and Research Center
Dreijenplein 8, 6703 HB Wageningen (The Netherlands)
[e] A. Lazar, Dr. A. W. Coleman
Institut de Biologie et Chimie des Protꢀines
CNRS UMR 5086, 7 Passage Vercors, 69367 Lyon 07 (France)
[+] These authors contributed equally to this work.
Chem. Eur. J. 2005, 11, 1171 – 1180
DOI: 10.1002/chem.200400905
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1171

model of recognition of substrates or ligands by receptors
on the surface of cell membranes. Molecular recognition
and specific ion binding at model membrane surfaces is a
topic of interest in supramolecular chemistry, with an in-
creasing emphasis on multivalent interactions.[10] Many rec-
ognition processes at the cell surface in nature are also am-
plified in affinity and selectivity by multivalent interac-
tions.[11]
Here we describe in detail a family of amphiphilic cyclo-
dextrins that form stable nonionic vesicles in water. Cyclo-
dextrins of various ring sizes (a-, b-, and g-cyclodextrins,
with six, seven, and eight glucose units, respectively) were
modified with hydrophobic n-dodecyl or n-hexadecyl and
hydrophilic oligo(ethylene glycol) substituents (see
Scheme 1). The properties of vesicles of these cyclodextrins
were studied, with an emphasis on the packing of the amphi-
philic macrocycles in the vesicle bilayer. We also examined
the inclusion of a small guest molecule (adamantane carbox-
ylate) in the cyclodextrin cavities at the surface of the vesi-
cles.
Figure 1. Cyclodextrin vesicles consist of bilayers of cyclodextrins (in
which the hydrophobic ꢃtailsꢃ are directed inwards and hydrophilic
macrocycle ꢃhead groupsꢃ are facing water) enclosing an aqueous interior.
A) Schematic representation of a unilamellar vesicle. B) Illustration of an
extended, all-trans packing of the alkyl chains in a cyclodextrin bilayer.
C) Illustration of an interdigitated packing of the alkyl chains in a cyclo-
dextrin bilayer.
philic cyclodextrins. From these studies it is evident that a
combination of hydrophobic alkyl substituents on one face
of the cyclodextrin ring and hydrophilic (poly(ethylene
glycol), sulfonate, ammonium, etc.)[6–8] substituents on the
other side of the cyclodextrin ring is essential to obtain
water-soluble amphiphiles. Several other cyclodextrin deriv-
atives can be admixed with liposomes, but are not able to
form stable vesicles by themselves.[9] Most of the studies
cited above have been limited to b-cyclodextrins.
Cyclodextrin vesicles com-
Results and Discussion
Synthesis: The preparation of the amphiphilic cyclodextrins
is outlined in Scheme 1. The synthesis of 6-chloro-cyclodex-
trins 2a–c and 6-bromo-cyclodextrins 3a–c was carried out
bine the properties of lipo-
somes and macrocyclic host
molecules in their potential to
encapsulate water-soluble mole-
cules in the aqueous interior, to
absorb hydrophobic molecules
into the bilayer membrane, and
finally to recognize and bind
specific types of guest mole-
cules through inclusion in the
cyclodextrin cavities at the sur-
face of the vesicle. In a recent
communication,[6c] we demon-
strated that vesicles of nonionic
b-cyclodextrin derivatives (6b)
bind tert-butylbenzoate and
adamantane carboxylate, and
they have a high affinity for
“guest polymers” (polyelectro-
lytes modified with adamantyl
and tert-butylbenzyl groups)
due to multivalent interactions
between the cyclodextrin hosts
in the bilayer and the hydro-
phobic guest substituents on the
polymer. The recognition of
small guest molecules by cyclo-
dextrin hosts assembled in a bi-
Scheme 1. Synthesis of amphiphilic cyclodextrins. a) CH3SO2Cl, DMF, 658C, two days (to give chlorides 2a–
2c) or NBS, Ph3P, DMF, 608C, 4 h (to give bromides 3a–3c); b) tBuOK or NaH, RSH, DMF, 808C, 3–4 days;
c) ethylene carbonate, K2CO3, N,N,N’,N’-tetramethylurea, 1508C, 4 h. DMF=N,N-dimethylformamide, NBS=
layer membrane is
a useful N-bromosuccinimide.
1172
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1171 – 1180

Bilayer Vesicles of Amphiphilic Cyclodextrins
FULL PAPER
according to literature procedures.[12,13] Either of these pre-
cursors can be used to obtain the alkyl thioethers 4a–c and
5a–c by nucleophilic substitution, although compounds 2a–c
are less reactive than 3a–c and usually require longer reac-
tion times to afford complete substitution in a good yield.
While b-cyclodextrins 4b and 5b have been described,[14–16]
a- and g-cyclodextrins 4a, 4c, 5a, and 5c are new. The most
effective purification of these alkyl thioethers is a Soxhlet
extraction with hexane to remove excess alkyl thiols. Am-
phiphilic cyclodextrins 6a–c and 7a–c were obtained in
good yields by graft polymerization of ethylene carbonate in
the presence of potassium carbonate. This reaction has been
described in detail for the preparation of 6b and 7b.[17] Cy-
clodextrins 6a, 6b, 7a, and 7c are new derivatives, and their
preparation (analogous to that of 6b and 7b)[17] and full
characterization is given in the Experimental Section. We
note that 6a–c and 7a–c are polydisperse products with an
average degree of substitution of 2–3 units of ethylene oxide
per glucose based on MALDI-TOF MS. Substitution occurs
exclusively at C-2, not C-3.[17] Although the degree of poly-
merization tends to vary from batch to batch (and is rather
high, with approximately 4 units of ethylene oxide per glu-
cose for 7c), we did not observe any significant differences
in reactivity for the a-, b-, and g-cyclodextrins.
Cyclodextrin vesicles: According to dynamic light scattering,
cyclodextrin vesicles prepared by extrusion through a 0.1 mm
polycarbonate membrane invariably have an average diame-
ter of 140–160 nm, irrespective of the type of cyclodextrin
(a-, b-, or g-cyclodextrin) and the alkyl chain length (n-do-
decyl or n-hexadecyl).[6c] Cyclodextrin vesicles prepared by
sonication are usually smaller (80–100 nm).[6a,b,8] These ob-
servations were confirmed by using transmission electron
microscopy.[6,8] Representative micrographs are shown in
Figure 2. Vesicles of this size are expected to be mostly uni-
lamellar, not multilamellar. The cyclodextrin vesicles are
stable in aqueous solution for several days, although they
tend to precipitate if kept longer or in the presence of salt
(>0.1m). The larger vesicles prepared by extrusion precipi-
tate faster than the smaller vesicles prepared by sonication.
Vesicles of n-dodecyl tri(ethylene glycol) ether (C12EO3)
were prepared to serve as reference vesicles with similar sur-
face potential but without cyclodextrin host cavities. As re-
ported in the literature,[18] the reference vesicles of C12EO3
are generally somewhat larger (180–200 nm) than the cyclo-
dextrin vesicles and tend to precipitate after several hours.
Further evidence that cyclodextrins 6a–c and 7a–c form
closed bilayer vesicles was obtained by encapsulation of the
hydrophilic fluorescent dye carboxyfluorescein in the aque-
ous interior of the vesicles. Cyclodextrin vesicles were pre-
pared in a solution of carboxyfluorescein at self-quenching
concentration (see the Experimental Section for details).
Encapsulated carboxyfluorescein was separated from the
free dye by gel filtration on a column of Sephadex G25. The
vesicles eluted at 5–8 mL, whereas the free dye eluted
beyond 12 mL. The results are presented as the ratio of fluo-
rescence intensity after (FTX) and before (Finit) addition of
Figure 2. Transmission electron microscopy of a-cyclodextrin vesicles
stained with uranyl acetate. a) Vesicles of 6a. b) Vesicles of 7a. Scale
bar: 500 nm.
0.1% of Triton TX-100 (which solubilizes the cyclodextrin
vesicles and causes release of their contents); this can be
taken as a measure of the presence of entrapped carboxy-
fluorescein (Figure 3). Coincidence of entrapped dye with
the elution of the vesicles confirms the existence of an aque-
ous interior. As anticipated, the amount of entrapped car-
boxyfluorescein in, for example, vesicles of 7b correlates
with the concentration of 7b (Figure 3b).[6a] Encapsulation
of carboxyfluorescein in vesicles of 6a–c was generally less
efficient, which may be due to the shorter alkyl chains in
6a–c compared to those in 7a–c. The amount of entrapped
carboxyfluorescein is rather small considering the diameter
of vesicles. Typically, the entrapment efficiency is in the
order of 1 mmol carboxyfluorescein per mmol cyclodextrin,
or 1:1000. For liposomes, the entrapment is usually about
1:100.[19]
Bilayer packing of cyclodextrin amphiphiles: A particularly
important question is the packing of the amphiphilic cyclo-
dextrin molecules in the bilayer. The molecular surface area
of the a-, b-, or g-cyclodextrin molecule (170 ꢄ2 for 1a,
190 ꢄ2 for 1b, and 240 ꢄ2 for 1c) is significantly larger than
Chem. Eur. J. 2005, 11, 1171 – 1180
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1173

R. Darcy, B. J. Ravoo et al.
Table 1. First-order reflection angle (2q), d spacing, and molecular sur-
face area (A0 at zero compression and AC at maximum compression) of
amphiphilic cyclodextrins.
Cyclodextrin
2q [8]
d [ꢄ]
A0
AC
[ꢄ2 molꢀ1
]
[ꢄ2 molꢀ1
]
6a
6b
6c
7a
7b
7c
2.0143
2.1089
2.0758
1.9463
1.9304
1.9749
43.8
41.9
42.5
45.4
45.7
44.7
340
375
420
285
410
424
142
162
240
174
160
128
Figure 3. Encapsulation of carboxyfluorescein in cyclodextrin vesicles.
Elution profiles of vesicles of cyclodextrins 6a–6c (a) and cyclodextrins
7a–7b (b) loaded with carboxyfluorescein on a 18ꢁ1 cm Sephadex G25
column. The relative fluorescence intensity scale FTX/Finit indicates encap-
&
sulated carboxyfluorescein only, as explained in the text. a) : 6a at
5 mgmLꢀ1
5 mgmLꢀ1
;
;
:
6b at 5 mgmLꢀ1
;
:
6c at 5 mgmLꢀ1
.
b)
7a at
~
*
^
:
ꢀ1
ꢀ1
ꢀ1
~
&
*
: 7b at 5 mgmL
;
: 7b at 8 mgmL
;
: 7b at 20 mgmL
.
the combined area of six, seven, or eight all-trans alkyl
chains, respectively (6ꢁ20 ꢄ2 =120 ꢄ2, 7ꢁ20 ꢄ2 =140 ꢄ2,
and 8ꢁ20 ꢄ2 =160 ꢄ2). Hence, in order to fill the ꢃvoidꢃ
under the macrocycle to obtain a nonleaky bilayer mem-
brane, either the alkyl chains should interdigitate, tilt, or
fold back, or the macrocycle should collapse. The first sce-
nario is well known for many amphiphilic molecules in bi-
layers and monolayers and can be verified by measuring the
thickness of the bilayers and the average orientation of the
alkyl chains. The second scenario is known for extensively
substituted cyclodextrins.[20] Collapse of the cavity would
have important consequences for the inclusion properties of
these cyclodextrin molecules and can be verified by measur-
ing the molecular surface area of the molecules in a mono-
layer on the air–water interface.
Figure 4. Small-angle X-ray diffraction pattern for cyclodextrin vesicles
deposited and air-dried on glass cover slides. a) Diffraction pattern for n-
dodecyl cyclodextrins 6a–6c. b) Diffraction pattern for n-hexadecyl cy-
clodextrins 7a–7c.
proximately 42 ꢄ for cyclodextrins 6a–c and approximately
45 ꢄ for cyclodextrins 7a–c at 258C (Table 1 and Figure 4).
Several higher order reflections are observed (in particular
for 6a–c and 7c, but not for 7a and 7b),[21] which indicates a
strong tendency for layering of these amphiphiles. The d
spacing can be taken as a measure of the average bilayer
Small-angle X-ray diffraction of an air-dried multilayer
deposited on a glass cover slide shows a d spacing of ap-
1174
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1171 – 1180

Bilayer Vesicles of Amphiphilic Cyclodextrins
FULL PAPER
thickness. As expected, the d spacing depends on the length
of the alkyl substituent (n-dodecyl versus n-hexadecyl) but
is independent of the cyclodextrin ring size. From Corey–
Pauling–Koltun (CPK) models we estimate a molecular
length of approximately 31 ꢄ for 6a–c and approximately
37 ꢄ for 7a–c (assuming all-trans alkyl substituents at C6
and di(ethylene glycol) at C-2 of the cyclodextrin), which
would suggest a bilayer thickness of 2ꢁ31=62 ꢄ and 2ꢁ
37=74 ꢄ, respectively. Since the experimental bilayer thick-
ness is much smaller than the predicted thickness for ex-
tended cyclodextrin molecules, the alkyl chains of the mole-
cules must be deeply interdigitated, tilted, or back-folded,
or exist in a combination of these three possibilities.
somewhat obscures the clear trend of molecular area with
cyclodextrin ring size and may explain the differences ob-
served between cyclodextrins with n-dodecyl and n-hexadec-
yl substituents. A notable exception to the observed trend is
the small value of AC for 7c, which may arise either from
collapse of the cavity of the more flexible g-cyclodextrin or
from extrusion of the amphiphile into the subphase as vesi-
cles upon compression of the monolayer. Cyclodextrin 7c
has a higher degree of hydrophilic substitution than any of
the other cyclodextrins studied.
In summary, according to the pressure–area isotherms, the
molecular areas of 6a–c and 7a–c are significantly increased
due to the presence of the ethylene glycol substituents.
Since we find larger, not smaller, molecular areas as a result
of ethylene glycol substitution, there is no reason to assume
that the cyclodextrin cavities have collapsed. Since the diam-
eter of an alkyl chain is approx-
Brewster angle microscopy indicates that cyclodextrins
6a–c and 7a–c form homogeneous monolayers on the air–
water interface (Figure 5). Characteristic pressure–area iso-
imately 20 ꢄ2, the alkyl groups
in 6a and 7a occupy 6ꢁ20 ꢄ2 =
120 ꢄ2 when they are in an all-
trans conformation (hence 7ꢁ
20 ꢄ2 =140 ꢄ2 for 6b and 7b
and 8ꢁ20 ꢄ2 =160 ꢄ2 for 6c
and 7c). So, the ethylene glycol
substituted cyclodextrin “head
group” has a molecular surface
area A0 that is almost twice the
area of the seven alkyl chains
together, and to ensure optimal
packing with minimal voids the
alkyl chains must either interdi-
gitate, tilt, or back-fold, as dis-
cussed above. By using CPK
models it can be estimated that
for fully interdigitated alkyl
chains (that is, the terminal
Figure 5. Pressure–area isotherm for a Langmuir monolayer of b-cyclodextrin 7b spread on the air–water inter-
face, including Brewster angle micrographs.
methyl groups of one bilayer
leaflet reside near the first
methylene of the opposite bi-
layer leaflet) the thickness of
therms for each cyclodextrin were obtained upon gradual
compression of these monolayers. The collapse of the mono-
layers occurs at 40–50 mNmꢀ1, which is a normal value for
monolayers of cyclodextrins with long (>C10) hydrophobic
groups at the primary face.[14–16] The area per molecule at
zero compression (A0) and also at maximum compression
(AC) scales with the cyclodextrin ring size (Table 1), but it is
similar for cyclodextrins with n-dodecyl and n-hexadecyl
substituents.[22] The area per molecule at zero compression
(A0) is larger than the area for the cyclodextrins without eth-
ylene glycol substitution. For the analogues of 4a–c and 5a–
c described in the literature,[14–16] the molecular area is
hardly different from unmodified cyclodextrin (170 ꢄ2 for
1a, 210 ꢄ2 for 1b, and 240 ꢄ2 for 1c). However, the pres-
ence of ethylene glycol residues clearly increases the molec-
ular surface area. In particular, the A0 value is sensitive to
the degree of substitution with poly(ethylene glycol), which
the bilayer would be approximately 44 ꢄ for 6a–c and ap-
proximately 50 ꢄ for 7a–c. These values are close to the ex-
perimental values from X-ray diffraction. Given a 1:2 “mis-
match” of head-group area and alkyl-chain volume, such a
bilayer thickness can also be achieved by a chain tilt of 608,
which seems an unrealistically high tilt angle.[23] Alternative-
ly, one could imagine that many alkyl chains fold back in-
stead of extending all-trans, but this is unlikely due to the
high entropy penalty. Hence, the most likely molecular
packing is deep interdigitation of extended alkyl chains
(with possibly a modest chain tilt and some back-folding), as
observed for asymmetric phospholipids[24] and also for am-
phiphilic calixarenes.[25] This interdigitated mode of bilayer
packing obviously restricts the mobility of the alkyl chains
in the bilayer and would nicely explain the lower enthalpy
for the cyclodextrin amphiphile compared to a phospholipid
in an Lb–La phase transition.[6a,24] The result is a relatively
Chem. Eur. J. 2005, 11, 1171 – 1180
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1175

R. Darcy, B. J. Ravoo et al.
thin but dense bilayer, with hydrophobic alkyl chains ex-
tending inwards and hydrophilic cyclodextrins decorated
with oligo(ethylene glycol) pointing outwards (Figure 1C).
anions at the interface between the PEGylated cyclodextrin
vesicles and bulk aqueous solution.[30] For an electrophoretic
mobility of ꢀ8ꢁ10ꢀ9 m2 Vꢀ1 sꢀ1 and a vesicle diameter of
160 nm, the Smoluchovski equation predicts a Zeta potential
of ꢀ11 mV. In the presence of excess adamantane carboxy-
late, vesicles of 6b and 6c have an electrophoretic mobility
of approximately ꢀ30ꢁ10ꢀ9 m2 Vꢀ1 sꢀ1, which would imply a
Zeta potential of approximately ꢀ42 mV.
Molecular recognition of adamantane carboxylate by cyclo-
dextrin vesicles: Given that all amphiphilic cyclodextrins de-
scribed above form bilayer vesicles in water, is it possible to
bind small guest molecules in the cyclodextrin cavities at the
surface of these vesicles? Adamantane carboxylate—like all
adamantanes—is known to be a good guest for inclusion
into b-cyclodextrin 1b (association constant Ka =3.2ꢁ104 mꢀ1
at pH 7.2 and 258C), while it has much weaker interaction
with g-cyclodextrin 1c (Ka =5.0ꢁ103 mꢀ1) and even less affin-
ity for a-cyclodextrin 1a (Ka =2.3ꢁ102 mꢀ1).[26] The inclusion
interaction of this typical guest with the cyclodextrin host
vesicles was investigated by using capillary electrophoresis.
This technique exploits the difference in electrophoretic mo-
bility between the free host and the host–guest complex (or
the free guest and the host–guest complex) to quantify host–
guest interactions.[27] In our hands, capillary electrophoresis
has proven particularly useful for quantifying the interaction
between b-cyclodextrin and anionic guests,[28] as well as be-
tween b-cyclodextrin vesicles and anionic guests[6c] in dilute
aqueous solution. Here, the electrophoretic mobility of the
host vesicles of 6a–c was measured in the presence of an in-
creasing concentration of adamantane carboxylate in the
background electrolyte (Figure 6). There is some precedent
The increase of electrophoretic mobility of the host vesi-
cles in the presence of an increasing concentration of the
guest was analyzed in terms of the formation of a 1:1 inclu-
sion complex of 6a–c and adamantane carboxylate, charac-
terized by the Ka value. The results are summarized in
Table 2. As anticipated, vesicles of the b-cyclodextrin amphi-
Table 2. Binding constants, Ka, of adamantane carboxylate to vesicles in
10 mm phosphate buffer (pH 7.5) at 258C.
Vesicle
Ka [mꢀ1
]
6a
6b
6c
C12EO3
96ꢂ40
7.1ꢂ0.6ꢁ103
3.2ꢂ0.3ꢁ103
3.1ꢂ0.6ꢁ102
phile 6b have the highest binding constant with Ka =7.1ꢁ
103 mꢀ1. This is significantly lower than for b-cyclodextrin 1b.
The difference might be attributed to some hindrance of in-
clusion into the cavity of 6b due to the presence of oligo-
(ethylene glycol) residues or a degree of anticooperativity
due to the increasing presence of anionic guests on the vesi-
cle surface. However, the Scatchard plot has a linear slope
and an abscissa intercept very close to 1.0, a fact indicating
the presence of identical and independent binding sites on
the vesicle surface. We can therefore ascribe the inferior
binding constant for 6b (as compared to 1b) to some steric
hindrance and some reduction in the hydrophobicity of the
host by the oligo(ethylene glycol) residues. This is consistent
with the observation that nonamphiphilic PEGylated cyclo-
dextrins are also poorer hosts than native cyclodextrins.[31]
In comparison to vesicles of 6b, vesicles of 6c have con-
siderably less affinity for adamantane carboxylate (Ka =3.0ꢁ
103 mꢀ1), a result reflecting the lower tendency of a g-cyclo-
dextrin cavity relative to that of a b-cyclodextrin to form in-
clusion complexes with this guest. Also, one expects steric
hindrance and reduction in the hydrophobicity of the host
by the oligo(ethylene glycol) residues. For host 6c we do not
exclude the formation of 1:2 host–guest complexes, particu-
larly at high guest concentrations. However, the Scatchard
plot again has a linear slope and an abscissa intercept very
close to 1.0.
Figure 6. Electrophoretic mobility (mep) of vesicles in the presence of the
!
*
adamantane carboxylate guest. : a-Cyclodextrin vesicles (6a); : b-cy-
!
*
clodextrin vesicles (6b); : g-cyclodextrin vesicles (6c); : C12EO3 refer-
ence vesicles.
for the investigation of liposomes by using capillary electro-
phoresis.[29]
The association constant of vesicles of 6a with adaman-
tane carboxylate (Ka =96mꢀ1) is very small compared to that
of 6b and 6c; this reflects the fact that the cavity of a-cyclo-
dextrin is too narrow to be a good host for adamantane
guests. In fact, the affinity of 6a is comparable to that of the
reference vesicles of C12EO3, which lack any specific host
cavities. The increase in electrophoretic mobility of vesicles
All cyclodextrin vesicles, as well as the C12EO3 reference
vesicles, invariably have significant negative electrophoretic
mobility (mep ꢁꢀ8ꢁ10ꢀ9 m2 Vꢀ1 sꢀ1) at neutral pH values in
dilute buffer solution. The negative electrophoretic mobility
most likely results from a preferential absorption of hydroxy
1176
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1171 – 1180

Bilayer Vesicles of Amphiphilic Cyclodextrins
FULL PAPER
of 6a and C12EO3 in the presence of high concentrations of
adamantane carboxylate (>1 mm) most likely results from
partitioning of the hydrophobic anion from aqueous solution
into the hydrophobic bilayer.
To confirm the specific and reversible binding of adaman-
tane carboxylate to vesicles of 6b, competition experiments
were carried out in the presence of b-cyclodextrin 1b. To
this end, the electrophoretic mobility of vesicles of 6b was
determined in the presence of a given concentration (0.5,
1.0, and 5.0 mm) of 1b and various concentrations of ada-
mantane carboxylate in the capillary (Figure 7). The associa-
Conclusion
Bilayer vesicles formed by nonionic amphiphilic cyclodex-
trins function as host membranes that bind suitable guest
molecules by hydrophobic inclusion at their surface. Capilla-
ry electrophoresis provides quantitatively reliable informa-
tion about these dynamic interactions at the membrane sur-
face. The cyclodextrin cavities function as independent host
sites and their characteristic affinity and selectivity for a
given guest molecule is not affected when they are confined
to a hydrophobic bilayer membrane. The recognition of
small guest molecules by cyclodextrin hosts assembled in a
bilayer membrane is a useful model of recognition of sub-
strates and ligands by receptors on the surface of cell mem-
branes. At present we are investigating the multivalent inter-
action of oligomeric guest molecules with host molecules at
the vesicle surface. Critical parameters will be the density
and mobility of the host in the membrane, the number and
flexibility of binding moieties on the guest, and the presence
of monovalent competitors. We aim to exploit these specific
interactions to bind molecules to vesicles, vesicles to vesi-
cles, and vesicles to surfaces.
Experimental Section
Synthesis: All commercial reagents were used without further purifica-
tion. Cyclodextrins 1a–c were dried in vacuum at 808C for at least 5 h. n-
Dodecyl tri(ethylene glycol) ether (C12EO3; containing some n-tetradecyl
tri(ethylene glycol) ether (C14EO3)) was kindly donated by Servo Sasol
(Delden, The Netherlands; the trade name is Serdox NES3). Chlorocy-
clodextrins 2a–c and bromocyclodextrins 3a–c were prepared according
to literature procedures.[12,13] b-Cyclodextrins 4b, 5b, 6b, and 7b have
been described previously.[15,17]
Figure 7. Electrophoretic mobility (mep) of b-cyclodextrin vesicles (6b) in
the presence of the adamantane carboxylate guest and competing b-cy-
!
!
*
*
clodextrin host (1b). : 0 mm 1b; : 0.5 mm 1b; : 1 mm 1b; : 5 mm 1b.
General procedure for the preparation of cyclodextrin alkyl thioethers
4a–c and 5a–c:[17] n-Alkylthiol (3 equiv for each halogen of cyclodextrin)
was dissolved in DMF, and NaH or tBuOK (3 equiv) was added. The
mixture was stirred at room temperature for 1 h. The appropriate cyclo-
dextrin 2a–c or 3a–c was added and the mixture was stirred at 60–808C
for 3–5 days. The reaction was monitored by TLC (EtOAc/iPrOH/
NH4OH/H2O 7:7:5:2, Rf =0.8 for 2a–c and 3a–c, Rf =0 for 4a–c and 5a–
c). The reaction mixture was then cooled to room temperature before
being added to water. The resulting white precipitate was filtered off and
washed with water, methanol, and hexane. The solid was stirred in boil-
ing hexane for 1 h to remove excess thiol, filtered, and dried under high
vacuum at 608C for 5 h. In some cases, residual thiol was removed by
Soxhlet extraction in hexane.
tion constant of adamantane carboxylate to cyclodextrin 6b
was calculated by using the concentration of free adaman-
tane carboxylate calculated from the total concentration
after subtraction of adamantane carboxylate complexed
with 1b, as calculated from the binding constant (Ka =3.2ꢁ
104 mꢀ1). As can be readily deduced from Figure 7 and
Table 3, a similar binding constant was obtained from each
Table 3. Binding constants, Ka, of adamantane carboxylate to b-cyclodex-
trin vesicles (6b) obtained from competition experiments in the presence
of b-cyclodextrin (1b).
Hexakis(6-dodecylthio)-a-cyclodextrin 4a: Compound 2a (3.0 g) was
treated with dodecanethiol to yield 4a (4.5 g, 80%). Alternatively, 3a
b-Cyclodextrin [mm]
Ka [mꢀ1
]
1
0
7.1ꢂ0.6ꢁ103
7.7ꢂ0.5ꢁ103
3.0ꢂ0.2ꢁ103
7.4ꢂ0.9ꢁ103
(6.0 g) was treated with dodecanethiol to yield 4a (6.1 g, 64%). H NMR
(300 MHz, [D6]DMSO): d=5.90 (s, 6H, OH-2); 5.84 (s, 6H, OH-3); 4.89
(d, 6H, J1,2 =3.1 Hz, H-1); 3.77 (m, 6H, H-3); 3.6 (m, 6H, H-5); 3.05–
3.32 (m, 12H, H-2, H-4); 2.7–2.85 (m, 12H, H-6); 2.57 (t, SCH2); 1.53
(m, SCH2CH2); 1.27 (brm, 108H, CH2); 0.87 (d, 21H, CH3) ppm;
13C NMR (75 MHz, [D6]DMSO): d=105.9 (C-1), 88.7 (C-4), 76.6, 76.2,
75.5 (C-3, C-2, C-5), 37.4–22.0 (dodecyl), 26.0 (C-6), 14.0 (CH3) ppm; ele-
mental analysis calcd (%) for C108H204O24S6 (2077.3): C 62.39, H 9.89, S
9.25; found: C 60.55, H 9.55, S 8.99; MALDI MS: m/z: 2099 [M+Na]+,
2115 [M+K]+.
0.5
1.0
5.0
competition experiment. These results demonstrate that ca-
pillary electrophoresis provides reliable quantitative infor-
mation about these dynamic equilibria.
Octakis(6-dodecylthio)-g-cyclodextrin 4c: Compound 2c (2.0 g) was
treated with dodecanethiol to yield 4c (3.5 g, 91%). Alternatively, 3c
(0.75 g) was treated with dodecanethiol to yield 4c (1.2 g, 75%).
1H NMR (300 MHz, [D6]DMSO): d=5.90 (s, 8H, OH-2); 5.84 (s, 8H,
Chem. Eur. J. 2005, 11, 1171 – 1180
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1177

R. Darcy, B. J. Ravoo et al.
OH-3); 4.89 (d, 8H, J1,2 =3.1 Hz, H-1); 3.77 (m, 8H, H-3); 3.6 (m, 8H,
H-5); 3.05–3.32 (m, 16H, H-2, H-4); 2.7–2.85 (m, 16, H-6); 2.57 (t, 16H,
SCH2); 1.53 (m, 16H, SCH2CH2); 1.27 (brm, 144H, CH2); 0.87 (d, 24H,
CH3) ppm; 13C NMR (75 MHz, [D6]DMSO): d=105.9 (C-1), 88.7 (C-4),
76.6, 76.2, 75.5 (C-3, C-2, C-5), 37.4–22.0 (dodecyl), 26.0 (C-6), 14.0
(CH3) ppm; elemental analysis calcd (%) for C144H272O32S8 (2769.7): C
62.39, H 9.89, S 9.25; found: C 63.51, H 9.70, S 9.18; MALDI MS: m/z:
2792 [M+Na]+, 2808 [M+K]+.
(%): 3764 (39) [M22EO+Na]+, 3720 (62) [M21EO+Na]+, 3676 (87)
[M20EO+Na]+, 3632 (100) [M19EO+Na]+, 3588 (92) [M18EO+Na]+, 3544
(67) [M17EO+Na]+, 3500 (29) [M16EO+Na]+.
Hexakis [6-hexadecylthio-2-oligo(ethylene oxide)]-a-cyclodextrin 7a:
Compound 5a (4.0 g) was treated with ethylene carbonate to yield 7a
(4.3 g, 89%). 1H NMR (300 MHz, [D]CHCl3): d=5.05 (brs, 6H, H-1);
4.0–3.3 (m, H-3, H-5, H-2, H-4, OCH2CH2O); 3.00 (m, 12H, H-6); 2.60
(m, 12H, SCH2); 1.57 (m, 12H, CH2); 1.30 (brs, 156H, CH2); 0.88 (t,
18H, CH3) ppm; 13C NMR (75 MHz, [D]CHCl3): d=100.9 (C-1), 81.2 (C-
2, C-4), 71.0–72.5 (C-3, C-5, CH2O), 61.5 (CH2OH), 34.1 (C-6), 33.7
(CH2S), 32.0 (CH2), 29.8 (CH2)n, 29.7 (CH2), 29.5 (CH2), 29.4 (CH2), 29.2
(CH2), 22.7 (CH2), 14.1 (CH3) ppm; elemental analysis calcd (%) for
C156H300O36S6 (2941.4): C 63.64, H 10.27, S 6.53; found: C 61.22, H 9.75, S
6.10; MALDI MS: m/z (%): 3070 (24) [M15EO]+, 3027 (55) [M14EO]+, 2983
(86) [M13EO]+, 2940 (100) [M12EO]+, 2896 (90) [M11EO]+, 2853 (55)
[M10EO]+, 2808 (26) [M9EO]+.
Hexakis(6-hexadecylthio)-a-cyclodextrin 5a: Compound 2a (1.0 g) was
treated with hexadecanethiol to yield 5a (1.8 g, 74%). Alternatively, 3a
(6.5 g) was treated with hexadecanethiol to yield 5a (6.8 g, 58%).
1H NMR (300 MHz, [D6]DMSO): d=5.90 (s, 6H, OH-2); 5.84 (s, 6H,
OH-3); 4.89 (d, 6H, J1,2 =3.1 Hz, H-1); 3.77 (m, 6H, H-3); 3.6 (m, 6H,
H-5); 3.05–3.32 (m, 12H, H-2, H-4); 2.7–2.85 (m, 14H, H-6); 2.57 (t,
12H, SCH2); 1.53 (m, 12H, SCH2CH2); 1.27 (brs, 156H, 13ꢁCH2); 0.87
(d, 18H, CH3) ppm; 13C NMR (75 MHz, [D6]DMSO): d=102.9 (C-1),
86.7 (C-4), 74.8, 74.1, 73.5 (C-3, C-2, C-5), 60.2 (C-6), 37.4–22.0 (hexa-
decyl), 26.0 (C-6), 13.9 (CH3) ppm; elemental analysis calcd (%) for
C132H252O24S6 (2413.7): C 65.63, H 10.51, S 7.96; found: C 63.83, H 9.70, S
7.48; MALDI MS: m/z: 2437 [M+Na]+, 2451 [M+K]+.
Octakis [6-hexadecylthio-2-oligo(ethylene oxide)]-g -cyclodextrin 7c:
Compound 5c (0.75 g) was treated with ethylene carbonate to yield 7c
(0.75 g, 85%). 1H NMR (300 MHz, [D]CHCl3): d=5.05 (brs, 8H, H-1);
4.0–3.3 (m, H-3, H-5, H-2, H-4, OCH2CH2O); 3.00 (m, 16H, H-6); 2.60
(m, 16H, SCH2); 1.57 (m, 14H, CH2); 1.30 (brm, 208H, CH2); 0.88 (t,
24H, CH3) ppm; 13C NMR (75 MHz, [D]CHCl3): d=100.9 (C-1), 81.2 (C-
2, C-4), 71.0–72.5 (C-3, C-5, CH2O), 61.5 (CH2OH), 34.1 (C-6), 33.7
(CH2S), 32.0 (CH2), 29.8 (CH2)n, 29.7 (CH2), 29.5 (CH2), 29.4 (CH2), 29.2
(CH2), 22.7 (CH2), 14.1 (CH3) ppm; elemental analysis calcd (%) for
Octakis(6-hexadecylthio)-g-cyclodextrin 5c: Compound 2c (0.3 g) was
treated with hexadecanethiol to yield 5c (0.28 g, 43%). Alternatively, 3c
(0.75 g) was treated with hexadecanethiol to yield 5c (1.0 g, 76%).
1H NMR (300 MHz, [D6]DMSO): d=5.90 (s, 8H, OH-2); 5.84 (s, 8H,
OH-3); 4.89 (d, 8H, J1,2 =3.1 Hz, H-1); 3.77 (m, 8H, H-3); 3.6 (m, 8H,
H-5); 3.05–3.32 (m, 16H, H-2, H-4); 2.7–2.85 (m, 16H, H-6); 2.57 (t,
16H, SCH2); 1.53 (m, 16H, SCH2CH2); 1.27 (brm, 208H, CH2); 0.87 (d,
24H, CH3) ppm; 13C NMR (75 MHz, [D6]DMSO): d=105.9 (C-1), 88.7
(C-4), 76.6, 76.2, 75.5 (C-3, C-2, C-5), 37.4–22.0 (hexadecyl), 26.0 (C-6),
14.0 (CH3) ppm; elemental analysis calcd (%) for C176H336O32S8 (3218.2):
C 65.63, H 10.51, S 7.96; found: C 64.81, H 10.21, S 8.09; MALDI MS:
m/z: 3255 [M+K]+.
C
S
208H400O48S8 (3922.7): C 63.64, H 10.27, S 6.53; found: C 63.82, H 10.01,
6.48; MALDI MS: m/z (%): 4756 (67) [M35EO+Na]+, 4741 (75)
[M34EO+Na]+, 4696 (92) [M33EO+Na]+, 4652 (92) [M32EO+Na]+, 4607 (82)
[M31EO+Na]+, 4568 (77) [M30EO+Na]+, 4524 (65) [M29EO+Na]+.
Vesicle preparation: Vesicles of 6a–c and 7a–c were prepared by sonica-
tion or extrusion. Typically, amphiphilic cyclodextrin (several mg) in
chloroform (approximately 1 mL) was dried by rotary evaporation to
yield a thin film in a glass vial. Residual solvent was removed under high
vacuum. Water or buffer (1–5 mL, 10 mm phosphate or 2-[4-(2-hydroxy-
ethyl)-1-piperazinyl]ethanesulfonic acid (HEPES), pH 7.2) was added
and the sample solution was kept for 1 h at room temperature (6a–c) or
at 508C (7a–c). The resulting suspension was sonicated in a Bran-
sonic 1510 sonication bath for 1 h to give small unilamellar vesicles. Al-
General procedure for the preparation of amphiphilic cyclodextrins 6a–c
and 7a–c:[17] The appropriate alkyl thioether 4a–c or 5a–c (0.5–5.0 g),
K2CO3 (10% by weight of cyclodextrin), and ethylene carbonate
(50 equiv) were mixed in N,N,N’,N’-tetramethylurea (5–15 mL). The mix-
ture was stirred at 1508C for 4 h. The reaction was monitored by TLC
(CHCl3/MeOH/H2O 50:10:1, Rf =0 for 4a–c and 5a–c, Rf =0.6 for 6a–c
and 7a–c). The solvent was evaporated under high vacuum at 708C in a
bulb-to-bulb destillation unit. In large scale reactions (>1 g), the crude
product was stirred overnight in a solution of NaOMe (0.1m) in metha-
nol.[17] The crude product was purified by size-exclusion chromatography
(Sephadex LH-20, methanol). For further purification, 6c was crystallized
in methanol and 7c was eluted over a silica-gel column with a mixture of
chloroform and methanol (9:1).
ternatively, the suspension was repeatedly passed through
a poly-
carbonate membrane with 0.1 mm pore size in a LiposoFast extruder.
Vesicles of 6a–c were sonicated or extruded at room temperature, where-
as vesicles of 7a–c (with a melting temperature, Tm, estimated to be
around 488C)[6a] were sonicated or extruded at 508C. Reference vesicles
were prepared from C12EO3 (10 mg) that was dissolved in chloroform
(1 mL) and then dried by rotary evaporation for 3 h to give a thin film.
Water (2.5 mL) was added and the sample was sonicated below 208C for
30 min.
Hexakis [6-dodecylthio-2-oligo(ethylene oxide)]-a-cyclodextrin 6a: Com-
pound 4a (5.0 g) was treated with ethylene carbonate to yield 6a (4.9 g,
1
Transmission electron microscopy (TEM): Samples for TEM were pre-
pared on 200 mesh formvar-carbon-coated copper grids. A drop of cyclo-
dextrin solution (approximately 0.1 mgmLꢀ1) was left on the grid for
2 min then gently blotted with filter paper. The samples were stained
with a drop of 2% (w/w) uranyl acetate, left for 5 min, and blotted again.
The samples were investigated in a JEOL 2000 transmission electron mi-
croscope operating at 80 kV.
93%). H NMR (300 MHz, [D]CHCl3): d=5.05 (brs, 6, H-1); 4.0–3.4 (m,
H-3, H-5, H-2, H-4, OCH2CH2O); 3.00 (m, 12, H-6); 2.60 (m, 12H,
SCH2); 1.60 (m, 12H, CH2); 1.27 (brs, 108H, CH2); 0.89 (t, 18H,
CH3) ppm; 13C NMR (75 MHz, [D]CHCl3): d=100.7 (C-1), 81.0 (C-2, C-
4), 71.0–72.0 (C-3, C-5, CH2O), 61.2 (CH2OH), 33.4 (C-6), 33.4 (CH2S),
31.7 (CH2), 29.5 (CH2)n, 29.2 (CH2), 28.8 (CH2), 22.4 (CH2), 13.9
(CH3) ppm; elemental analysis calcd (%) for C132H254O36S6 (2607.6): C
60.75, H 9.81, S 7.37; found: C 59.45, H 9.75, S 7.42; MALDI MS: m/z
(%): 2851 (24) [M17EO+Na]+, 2807 (47) [M16EO+Na]+, 2762 (73)
[M15EO+Na]+, 2719 (100) [M14EO+Na]+, 2674 (100) [M13EO+Na]+, 2630
(71) [M12EO+Na]+, 2586 (36) [M11EO+Na]+.
Dynamic light scattering: Dynamic light-scattering measurements were
carried out at room temperature by using Malvern instrumentation. The
amphiphile concentration was approximately 0.2 mgmLꢀ1. The solutions
were filtered through 0.45 mm Gelman Acrodisk syringe filters prior to
light-scattering measurements. Size distributions were obtained from a
CONTIN analysis of the scattering data.
Octakis [6-dodecylthio-2-oligo(ethylene oxide)]-g-cyclodextrin 6c: Com-
pound 4c (100 mg) was treated with ethylene carbonate to yield 6c
(90 mg, 69%). 1H NMR (300 MHz, [D]CHCl3): d=5.05 (brs, 8H, H-1);
4.0–3.4 (m, H-3, H-5, H-2, H-4, OCH2CH2O); 3.00 (m, 16H, H-6); 2.60
(m, 16H, SCH2); 1.60 (m, 16H, CH2); 1.27 (brs, 144H, CH2); 0.89 (t,
24H, CH3) ppm; 13C NMR (75 MHz, [D]CHCl3): d=100.7 (C-1), 81.0 (C-
2, C-4), 71.0–72.0 (C-3, C-5, CH2O), 61.2 (CH2OH), 33.4 (C-6), 33.4
(CH2S), 31.7 (CH2), 29.5 (CH2)n, 29.2 (CH2), 28.8 (CH2), 22.4 (CH2), 13.9
(CH3) ppm; elemental analysis calcd (%) for C176H336O48S8 (3474.2): C
60.80, H 9.74, S 7.38; found: C 61.74, H 9.68, S 8.11; MALDI MS: m/z
Dye encapsulation: A 10 mm solution of carboxyfluorescein was prepared
in 10 mm HEPES buffer solution (pH 7.2). Cyclodextrin vesicles were
prepared by dissolving the appropriate cyclodextrin 6a–c or 7a–c in
chloroform and then evaporating the solvent to form a thin film. The car-
boxyfluorescein solution (1 mL) was added and the cyclodextrin film was
hydrated for 1 h. Next, the sample was shaken vigorously and sonicated
for 1 h at 508C. Cyclodextrin concentrations ranging from 0.5–
20 mgmLꢀ1 were evaluated, although little entrapment of dye molecules
1178
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1171 – 1180

Bilayer Vesicles of Amphiphilic Cyclodextrins
FULL PAPER
was obtained below concentrations of 1.5 mgmLꢀ1. Above concentrations
of 1 mgmLꢀ1, 7c and carboxyfluorescein precipitated and no encapsula-
tion could be measured. The solution of encapsulated guest molecules
(250 mL) was loaded onto a Sephadex G-25 size-exclusion column (18ꢁ
1 cm, void volume=4 mL) with HEPES (10 mm) solution as the eluent.
Fractions of 1 mL were collected. These samples were made up to 2.5 mL
adamantane carboxylate concentration, with the assumption that the con-
centration of complexed guest is always small relative to the total guest
concentration.[28]
in transparent perspex cuvettes.
A Perkin–Elmer LB50 fluorescence
spectrometer was used to measure the fluorescence of the fractions (lex
=
[1] R. C. Petter, J. S. Salek, C. T. Sikorski, G. Kumaravel, F. T. Lin, J.
Am. Chem. Soc. 1990, 112, 3860–3868.
485 nm and lem =520 nm). Vesicles were detected by light scattering at
lex =lem =400 nm. The fluorescence of each fraction was measured imme-
diately and after storage for 2 days at room temperature, over which time
there was less than 10% difference. Finally, the fluorescence of each frac-
tion was reexamined upon addition of the detergent Triton X-100, which
lyses the vesicles.
[2] a) H. Parrot-Lopez, C.-C. Ling, P. Zhang, A. Baszkin, G. Albrecht,
C. de Rango, A. W. Coleman, J. Am. Chem. Soc. 1992, 114, 5479–
5480; b) A. Schalchli, J. J. Benattar, P. Tschoreloff, P. Zhang, A. W.
Coleman, Langmuir 1993, 9, 1968–1970; c) P. C. Tschoreloff, M. M.
Boissonnade, A. W. Coleman, A. Baszkin, Langmuir 1995, 11, 191–
196.
[3] a) M. Skiba, D. DuchÞne, F. Puisieux, D. Wouessidjewe, Int. J.
Pharm. 1996, 129, 113–121; b) A. Gulik, H. Delacroix, D. Wouessid-
jewe, M. Skiba, Langmuir 1998, 14, 1050–1057.
[4] a) R. Auzꢀly-Velty, F. Djedaïni-Pilard, S. Dꢀsert, B. Perly, T. Zemb,
Langmuir 2000, 16, 3727–3734; b) R. Auzꢀly-Velty, C. Pꢀan, F. Dje-
daïni-Pilard, T. Zemb, B. Perly, Langmuir 2001, 17, 504–510.
[5] a) A. Mazzaglia, B. J. Ravoo, R. Darcy, P. Gambadauro, F. Malla-
mace, Langmuir 2002, 18, 1945–1948; b) D. Lombardo, A. Longo,
R. Darcy, A. Mazzaglia, Langmuir 2004, 20, 1057–1062.
[6] a) B. J. Ravoo, R. Darcy, Angew. Chem. 2000, 112, 4494–4496;
Angew. Chem. Int. Ed. 2000, 39, 4324–4326; b) D. Nolan, R. Darcy,
B. J. Ravoo, Langmuir, 2003, 19, 4469–4472; c) B. J. Ravoo, J. C. Jac-
quier, G. Wenz, Angew. Chem. 2003, 115, 2112–2116; Angew. Chem.
Int. Ed. 2003, 42, 2066–2070; d) A. Mazzaglia, D. Forde, D. Garoz-
zo, P. Malvagna, B. J. Ravoo, R. Darcy, Org. Biomol. Chem. 2004, 2,
957–960.
X-ray diffraction: The X-ray diffraction measurements were performed
on a Panalytical XꢃPert Pro diffractometer (Panalytical, Almelo, The
Netherlands) by using nickel-filtered CuKa radiation (tube operating at
40 kV and 40 mA). The data were collected by using an automatic diver-
gence slit (5 mm irradiated length) and a 0.2 mm receiving slit. The data
were collected from 1–108 (2q) with a step size of 0.0058 (2q) and a
counting time of 1.0 s per step. The samples were prepared by depositing
a droplet of cyclodextrin vesicle solution (approximately 2 mgmLꢀ1) on a
microscope cover slide and letting this dry in a flow of warm air until a
thin film of cyclodextrins remained on the cover slide. In the X-ray dif-
fractogram, first-, second-, and third-order Bragg peaks were observed,
except with cyclodextrins 7a and 7b, for which only first-order peaks
were observed. From the positions of the first-order peaks, the bilayer
thickness was calculated by using Braggꢃs law.
Langmuir monolayers: Spreading solutions were prepared by dissolving a
known quantity (approximately 5 mg) of the cyclodextrin in chloroform
(5 mL). Pressure–area measurements were carried out in a Teflon trough
of 400 mL. Cyclodextrin solutions were deposited in an appropriate
volume (approximately 12 mL) with a micropipette at the air–water inter-
face. 30 min was allowed for solvent evaporation and equilibration. Pres-
sure–area isotherms were measured at 208C on a Langmuir type balance
(Nima Technology). Compressions were performed continuously at a rate
of 20 cm2 minꢀ1 from 510–50 cm2. Each sample was run at least twice to
ensure reproducibility of results. Brewster angle microscopy was carried
out by using an NFT Mini Brewster angle microscope on the Nima film
balance. Compression rates were the same as for pressure–area measure-
ments. The image size is 4ꢁ6 mm and resolution is <20 mm.
[7] T. Sukegawa, T. Furuike, K. Niikura, A. Yamagishi, K. Monde, S.-I.
Nishimura, Chem. Commun. 2002, 430–431.
[8] R. Donohue, A. Mazzaglia, B. J. Ravoo, R. Darcy, Chem. Commun.
2002, 2864–2865.
[9] a) M. J. Pregel, L. Jullien, J.-M. Lehn, Angew. Chem. 1992, 104,
1695–1699; Angew. Chem. Int. Ed. Engl. 1992, 31, 1637–1640; b) L.
Jullien, T. Lazrak, J. Canceill, L. Lacombe, J.-M. Lehn, J. Chem.
Soc. Perkin Trans. 2 1993, 1011–1020; c) J. Lin, C. Creminon, B.
Perly, F. Djedaïni-Pilard, J. Chem. Soc. Perkin Trans. 2 1998, 2639–
2646; d) R. Auzꢀly-Velty, B. Perly, O. Tachꢀ, T. Zemb, P. Jꢀhan, P.
Guenot, J. P. Dalbiez, F. Djedaïni-Pilard, Carbohydr. Res. 1999, 318,
82–90.
[10] a) D. Y. Sasaki, T. A. Waggoner, J. A. Last, T. M. Alam, Langmuir
2002, 18, 3714–3721; b) E. L. Doyle, C. A. Hunter, H. C. Phillips,
S. J. Webb, N. H. Williams, J. Am. Chem. Soc. 2003, 125, 4593–4599;
c) B. Bondurant, J. A. Last, T. A. Waggoner, A. Slade, D. Y. Sasaki,
Langmuir 2003, 19, 1829–1837; d) J. Huskens, A. Mulder, T. Aulet-
ta, C. A. Nijhuis, M. J. W. Ludden, D. N. Reinhoudt, J. Am. Chem.
Soc. 2004, 126, 6784–6797; e) V. Marchi-Artzner, M.-J. Brienne, T.
Gulik-Krzywicki, J. C. Dedieu, J.-M. Lehn, Chem. Eur. J. 2004, 10,
2342–2350.
[11] a) M. Mammen, S.-K. Choi, G. M. Whitesides, Angew. Chem. 1998,
110, 2908–2953; Angew. Chem. Int. Ed. 1998, 37, 2754–2794;
b) L. L. Kiessling, J. E. Gestwicki, L. E. Strong, Curr. Opin. Chem.
Biol. 2000, 4, 696–703.
[12] F. Guillo, B. Hamelin, L. Jullien, J. Canceill, J.-M. Lehn, L. de Ro-
bertis, H. Driguez, Bull. Soc. Chim. Fr. 1995, 132, 857–866.
[13] A. Gadelle, J. Defaye, Angew. Chem. 1991, 103, 94–95; Angew.
Chem. Int. Ed. Engl. 1991, 30, 78–79.
Capillary electrophoresis (CE): Capillary electrophoresis was carried out
as described previously.[6c,28] Measurements were carried out on a 57 cm
(48.5 cm from inlet to detector) fused silica capillary (75 mm internal di-
ameter; Polymicro Technologies) with a separation voltage of 25 kV, by
using an Agilent HP 3D CE system. The capillary was conditioned with
1n NaOH (5 min), water (1 min), and 10 mm phosphate buffer (1 min)
before each series of measurements and running buffer (1 min) before
each measurement. The running buffer was prepared with a varying con-
centration of adamantane carboxylate in 10 mm phosphate buffer adjust-
ed to pH 7.5. The analyte sample (0.2 mgmLꢀ1 in 5 mm phosphate buffer)
was introduced with 34.5 mbar injection for 5 s and detected with a diode
array detector at 200 nm. Measurements were repeated two or three
times for each concentration. For competition experiments, the elution
time of the vesicles was determined in the presence of a known concen-
tration (0.5, 1.0, or 5.0 mm) of b-cyclodextrin 1b and various concentra-
tions of adamantane carboxylate in the capillary. The electrophoretic mo-
bility, mep, of the vesicles was determined from the elution time according
to Equation (1), where l and L denote the effective length (in m) of the
capillary from injector to detector and the total length (in m), respective-
ly, V is the voltage (in V), and teof and t represent the elution times (in s)
of the electroosmotic flow (detected by a negative peak) and the sample,
respectively.
[14] Y. Kawabata, M. Matsumoto, M. Tanaka, H. Takahashi, Y. Irinatsu,
S. Tamura, Y. Tagaki, H. Nakahara, K. Fukuda, Chem. Lett. 1986,
1933–1934.
[15] C.-C. Ling, R. Darcy, W. Risse, J. Chem. Soc. Chem. Commun. 1993,
438–440.
[16] K. Kobayashi, K. Kajikawa, H. Sasabe, W. Knoll, Thin Solid Films
1999, 349, 244–249.
mep ¼ l L Vꢀ1ð1=tꢀ1=teof
Þ
ð1Þ
Binding contants, Ka, were calculated from a nonlinear regression of the
change of electrophoretic mobility of the vesicles as a function of the
[17] A. Mazzaglia, R. Donohue, B. J. Ravoo, R. Darcy, Eur. J. Org.
Chem. 2001, 1715–1721.
Chem. Eur. J. 2005, 11, 1171 – 1180
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1179

R. Darcy, B. J. Ravoo et al.
[18] D. A. van Hal, J. A. Bouwstra, A. van Rensen, E. Jeremiasse, T.
de Vringer, H. E. Junginger, J. Colloid Interface Sci. 1996, 178, 263–
273.
[26] a) W. C. Cromwell, K. Bystrçm, M. R. Eftink, J. Phys. Chem. 1985,
89, 326–332; b) M. Weickenmeier, G. Wenz, Macromol. Rapid
Commun. 1996, 17, 731–736.
[19] F. Szoka, D. Papahadjopoulos, Proc. Natl. Acad. Sci. USA 1978, 75,
4194–4198.
[27] I. J. Colton, J. D. Carbeck, J. Rao, G. M. Whitesides, Electrophoresis
1998, 19, 367–382.
[20] L. Jullien, J. Canceill, L. Lacombe, J.-M. Lehn, J. Chem. Soc. Perkin
Trans. 2 1994, 989–1002.
[28] B. J. Ravoo, J.-C. Jacquier, Macromolecules 2002, 35, 6412–6416.
[29] a) M. A. Roberts, L. Locascio-Brown, W. A. MacCrehan, R. A.
Durst, Anal. Chem. 1996, 68, 3434–3440; b) K. Kawakami, Y. Nishi-
hara, K. Hirano, Langmuir 1999, 15, 1893–1895; c) S. K. Wiedmer,
J. Hautala, J. M. Holopainen, P. K. J. Kinnunen, M. L. Riekkola,
Electrophoresis 2001, 22, 1305–1313; d) C. F. Duffy, S. Gafoor, D. P.
Richards, H. Admadzadeh, R. OꢃKennedy, E. A. Arriaga, Anal.
Chem. 2001, 73, 1855–1861; e) A. N. Phayre, H. M. V. Farfano,
M. A. Hayes, Langmuir 2002, 18, 6499–6503; f) J. McKeon, M. G.
Khaledi, J. Chromatogr. A 2003, 1004, 39–46.
[30] a) Y. H. M. Chan, R. Schweiss, C. Werner, M. Grunze, Langmuir
2003, 19, 7380–7385; b) M. Johnsson, A. Wagenaar, J. B. F. N. Eng-
berts, J. Am. Chem. Soc. 2003, 125, 757–760.
[31] I. N. Topchieva, P. Mischnick, G. Kꢅhn, V. A. Polyakov, S. V. Elez-
kaya, G. I. Bystryzky, K. I. Karezin, Bioconjugate Chem. 1998, 9,
676–682.
[21] The X-ray diffraction spectra for 7a and 7b show no higher order
reflections. Instead, a broad peak is found at 14.5 ꢄ. This could be
the distance between the cyclodextrin rings.
[22] Given a molecular surface area of 325 ꢄ2, approximately 50000 mol-
ecules of cyclodextrin would be required to make a 160 nm diameter
unilamellar vesicle.
[23] From Fourier transform infrared spectroscopy of the multilayers
used for X-ray diffraction, such a dramatic chain tilt was ruled out:
the polarization dependence of the intensity of CH3 and CH2 vibra-
tions was evident only in the parallel (Brewster angle) configuration,
not in the perpendicular (transmission) configuration, a fact indicat-
ing that the chains are mostly oriented perpendicular to the bilayer
surface.
[24] J. M. Seddon, G. Cevc in Phospholipids Handbook (Ed.: G. Cevc),
Marcel–Dekker, New York, 1993, pp 403–454.
[25] P. Shahgaldian, M. Cesario, P. Goreloff, A. W. Coleman, Chem.
Commun. 2002, 326–327.
Received: September 3, 2004
Published online: December 27, 2004
1180
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1171 – 1180