Synthesis and vesicle formation from dimeric pseudoglyceryl lipids with (CH2)m spacers: Pronounced m-value dependence of thermal properties, vesicle fusion, and cholesterol complexation

Article abstract of DOI:10.1002/(SICI)1521-3765(19990802)5:8<2335::AID-CHEM2335>3.0.CO;2-U

Eight new dimeric lipids, in which the two Me₂N⁺ ion headgroups are separated by a variable number of polymethylene units [-(CH₂)_m-], have been synthesized. The electron micrograph (TEM) and dynamic light scattering (DLS) of their aqueous dispersions confirmed the formation of vesicular-type aggregates. The vesicle sizes and morphologies were found to depend strongly on the m value, the method, and thermal history of the vesicle preparation. Information on the thermotropic properties of the resulting vesicles was obtained from microcalorimetry and temperature-dependent fluorescence anisotropy measurements. Interestingly, the T_m values for these vesicles revealed a nonlinear dependence on spacer chain length (m value). These vesicles were able to entrap riboflavin. The rates of permeation of the OH^- ion under an imposed transmembrane pH gradient were also found to depend significantly on the m value. X-Ray diffraction of the cast films of the lipid dispersions elucidated the nature and the thickness of these membrane organizations, and it was revealed that these lipids organize in three different ways depending on the m value. The EPR spin-probe method with the doxylstearic acids 5NS, 12NS, and 16NS, spin-labeled at various positions of stearic acid, was used to establish the chain-flexibility gradient and homogeneity of these bilayer assemblies. The apparent fusogenic propensities of these bipolar tetraether lipids were investigated in the presence of Na₂SO₄ with fluorescence-resonance energy-transfer fusion assay. Small unilamellar vesicles formed from 1 and three representative biscationic lipids were also studied with fluorescence anisotropy and ¹H NMR spectroscopic techniques in the absence and the presence of varying amounts of cholesterol.

Full text of DOI:10.1002/(SICI)1521-3765(19990802)5:8<2335::AID-CHEM2335>3.0.CO;2-U

FULL PAPER
Synthesis and Vesicle Formation from Dimeric Pseudoglyceryl Lipids with
(
CH ) Spacers: Pronounced m-Value Dependence of Thermal Properties,
2
m
Vesicle Fusion, and Cholesterol Complexation
Santanu Bhattacharya* and Soma De[a]
Abstract: Eight new dimeric lipids, in
a nonlinear dependence on spacer chain
length (m value). These vesicles were
able to entrap riboflavin. The rates of
with the doxylstearic acids 5NS, 12NS,
and 16NS, spin-labeled at various posi-
tions of stearic acid, was used to estab-
lish the chain-flexibility gradient and
homogeneity of these bilayer assem-
blies. The apparent fusogenic propensi-
ties of these bipolar tetraether lipids
were investigated in the presence of
Na SO with fluorescence-resonance en-
‡
which the two Me N ion headgroups
2
are separated by a variable number of
�
polymethylene units [� (CH ) �], have
permeation of the OH ion under an
2
m
been synthesized. The electron micro-
graph (TEM) and dynamic light scatter-
ing (DLS) of their aqueous dispersions
confirmed the formation of vesicular-
type aggregates. The vesicle sizes and
morphologies were found to depend
strongly on the m value, the method,
and thermal history of the vesicle prep-
aration. Information on the thermotrop-
ic properties of the resulting vesicles was
obtained from microcalorimetry and
imposed transmembrane pH gradient
were also found to depend significantly
on the m value. X-Ray diffraction of the
cast films of the lipid dispersions eluci-
dated the nature and the thickness of
these membrane organizations, and it
was revealed that these lipids organize
in three different ways depending on the
m value. The EPR spin-probe method
2
4
ergy-transfer fusion assay. Small unila-
mellar vesicles formed from 1 and three
representative biscationic lipids were
also studied with fluorescence anisotro-
1
py and H NMR spectroscopic techni-
ques in the absence and the presence of
varying amounts of cholesterol.
Keywords: cholesterol interactions
temperature-dependent
fluorescence
´
lipids ´ membranes ´ thermotropic
anisotropy measurements. Interestingly,
properties
the T values for these vesicles revealed
m
triggers fusion, as Ca2 binding induces formation of dimeric
or higher-order lipid complexes.[ Similarly, glycolipid A,
composed of two phosphate headgroups and seven hydro-
phobic chains, serves as an immunomodulator in the selective
‡
Introduction
5]
While the properties of vesicular membranes derived from a
large number of monomeric natural lipids,[1] such as phospha-
[
2]
[6]
tidylcholines as well as their synthetic analogues, have been
examined in detail, there is very little in the literature that
attempts to examine the vesicles prepared from multimeric
lipids. This is despite the fact that several multimeric lipid
systems occur naturally and are believed to possess important
biological functions. Thus cardiolipins (glycerol-bridged di-
meric phosphatidic acid), which constitute a class of complex
phospholipids that occur mainly in the heart and skeletal
recognition of toxins, bacteria, or viruses at the cell surfaces.
Cytomimetic approaches in the design of dimeric and
oligomeric amphiphiles that form micelles with unusual
morphologies have opened up new vistas in aggregate
chemistry. Indeed impressive property modifications are
observed from these types of amphiphiles and therefore they
are currently receiving a lot of attention.[ A detailed study of
the influence of dimerization on vesicle systems of well-
defined synthetic lipids seems to be the most direct approach
to understanding the role of these complex amphiphilic
molecules in living cells.
[7]
[
3]
[8]
9]
[
4]
muscles, show high metabolic activity. In addition, associa-
tion of acidic phospholipids in cell membranes with Ca2
‡
We have been interested in the development of new
amphiphiles and lipids and in the properties manifested upon
their self-organization.[ Now we turn our attention to
synthetic gemini pseudoglyceryl lipids 2. The presence of an
intervening �(CH ) � spacer at the headgroup was found to
[
a] Prof. S. Bhattacharya, Dr. S. De
Department of Organic Chemistry
Indian Institute of Science, Bangalore 560 012 (India)
Also at: The Chemical Biology Unit
Jawaharlal Nehru Centre for Advanced Scientific Research
Bangalore 560 012 (India)
10]
2
m
strongly influence the properties of these lipids at the
Fax : (‡ 91)80334-1683
E-mail: sb@orgchem.iisc.ernet.in
membrane level. Thus at low and high m values (m ˆ 3, 4,
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2335 $ 17.50+.50/0
2335

S. Bhattacharya, S. De
FULL PAPER
monomeric lipid (m ˆ 0) for the
comparison of the vesicular
properties of the dimeric lipids.
Interestingly, the herein descri-
bed dimeric lipids, especially
with longer spacer chains
(
m ˆ 20 ± 22), are also related
to the bolaamphiphilic lipids,
where the headgroups are
linked through the termini of a
hydrophobic chain.[
13]
We have synthesized alto-
gether eight new dimeric lipids
2
a ± 2h according to the plan
shown in Scheme 1. First the
CH OH group of the primary
�
2
building block, 1,2-isopropyli-
dene glycerol (3), was protected
in the form of a benzyl ether 4.
or 20 ± 22), the vesicles are nearly impermeable and show high
thermal gel ± liquid crystalline transition temperatures. The
vesicles formed from lipids with m values 5 ± 12 show little
difference in their melting temperatures, although the trans-
membrane permeation rates increase with the rise in m value.
X-Ray diffraction studies on the cast film of these membranes
confirm distinct differences in their organization and packing
arrangements as a function of m value. The evidence of
interdigitation of lipids with high m values is shown unam-
biguously from EPR measurements. The intervesicular fusion
of membranes derived from dimeric lipids also shows
discernible m-value dependence. It is also shown that
inclusion of even 30 mol% of cholesterol in dimeric lipid
membranes does not lead to the abolition of the order± dis-
order transitions of these membranes.
Results and Discussion
Synthesis: For the synthesis of this new series of biscationic
gemini lipids, we considered the pseudoglyceryl backbone as a
suitable structural anchor for such lipid construction. This
type of building block is ubiquitously present in naturally
occurring phospholipids. The structural features of the
monomeric analogue of these lipids are similar to that of
dioleyloxy-propyl-trimethylammonium chloride (DOTMA)
and related derivatives, which are widely used for gene-
Scheme 1. Conditions: a) C
HCl/MeOH, RT 84%; ii) nC16
93%; d) CBr , PPh , CH Cl , 90%; e) Me
5%; f) Br(CH Br, reflux, 60 ± 75% (exact yields depend on m);
N/EtOH, heat, pressure tube, 90%.
6
H
H
5
CH
33Br, KOH, C
NH/EtOH, heat, pressure tube,
2
Cl, KOH, C
6
H
6
, reflux, 87%; b) i) aq.
6
H
6
, reflux, 58%; c) Pd-C/H
2
,
4
3
2
2
2
9
2 m
)
g) Me
3
[
11]
transfer applications. Long hydrocarbon chains (nC H )
The isopropylidene group from 4 was then removed by
treatment with HCl in aqueous MeOH, and the diol obtained
upon preparative chromatography over silica gel was dialkyl-
ated with 2.2 equivalents of nC H Br to give 5 (58% yield).
16
33
were connected to the glycerol backbone by an ether-type
linkage to ensure hydrolytic stabilities of the resulting lipids in
both acidic and alkaline conditions even at extreme pH.
16
33
Dimeric lipids were constructed by covalent insertion of a
Upon hydrogenolysis of 5, compound 6 was formed in ꢀ93%
yield. In the subsequent step, 6 was converted to the
corresponding bromide 7 in a ꢀ90% yield by reaction with
PPh /CBr in dry CH Cl . Compound 7 was quaternized in the
‡
polymethylene chain through two Me N headgroups within a
2
gemini unit, where each unit is composed of 1,2-dihexadecyl-
[
12]
oxy-propane-3-dimethylammonium ions. In this situation,
3
4
2
2
‡
the distance between the two polar �NMe groups should be
presence of Me N in EtOH to give 1 in ꢀ80% yield. The
2
3
governed by the spacer chain length (m value) and its
conformation. We also synthesized the corresponding lipid
monomer, 1,2-dihexadecyloxypropane-3-trimethylammoni-
um bromide, (1) and decided to consider it as the reference
bisquaternary lipids 2a ± 2h were obtained from the corre-
sponding dimethylamino derivative, 8 upon refluxing with
individual alkanediyl-a,w-dibromides in 60 ± 75% yields. All
the compounds gave expected analytical and spectroscopic
2
336
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2336 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Lipids
2335±2347
data consistent with their molecular structures, see the
Experimental section.
their monomeric counterpart (m ˆ 0). However, the variation
of the hydrodynamic diameter as a function of m value was
found to be quite complex. Hence, the vesicular size was
found to be the largest for lipid 2 with m ˆ 3, then decreased
with an increase in m value up to m ˆ 12, and again increased
with m > 12. These sizes were also found to be in agreement
with the TEM data on sonicated vesicles.
Vesicle formation and characterization: We first examined the
vesicle-forming abilities of the above lipids in water. Re-
versed-phase evaporation (REV) and sonication methods
have been used to obtain large and small vesicles, respectively.
All the dimeric lipid vesicles prepared by either protocol
remained stable for several days, except the one with m ˆ 3.
The size and nature of the vesicle were found to depend
strongly on the lipid concentration, thermal history, and the
method of vesicle preparation. Thus while vortexing afforded
mainly open lamellae and sonication gave small, spherical,
closed microstructures, REV gave predominantly multilamel-
lar, spherical vesicles (MLVs). Because of larger sizes, we
decided to closely examine the transmission electron micro-
scopy (TEM) of the MLVs prepared from these lipids
generated by REV.
Dye entrapment: In order to examine the entrapment
�
1
abilities, vesicles from 2a ± 2g (1 mgmL ) were generated
by dispersal in a bath sonicator over a period of ꢀ30 min at
�
5
ꢀ708C in water (pH 5.8) containing riboflavin (2 Â 10 m).
Gel-filtration chromatography with a sephadex G-50 column
was performed to separate riboflavin-entrapped vesicles from
the free, unbound riboflavin. As previously observed[ some
riboflavin molecules adhered to the aggregate interfaces,
while a few got entrapped inside the vesicles. Importantly,
however, in all the cases, riboflavin-associated vesicles could
be separated from the free riboflavin dye. Relevant data on
the percentage entrapment of riboflavin for various lipids are
given in Table 1.
14]
TEM examination of the individual, air-dried, opalescent
aqueous suspensions of 2a ± 2h layered on carbon/formvar-
coated copper grids revealed the existence of closed, multi-
lamellar aggregates in all the cases. The average diameters of
2
a, 2d, 2e, and 2h were 290 Æ 20 Š, 740 Æ 25 Š, 880 Æ 50 Š,
Transmembrane permeation: Riboflavin is fluorescent in its
and 1500 Æ 25 Š, respectively. Closer scrutiny of the micro-
graphs also revealed that lamellar widths of each lipid
aggregate were in reasonable accord with the results extrapo-
lated from the XRD measurements (see below).
neutral form and upon deprotonation (pK
a
ꢀ10.2) it loses its
fluorescence. The emission due to the membrane-associated
total riboflavin was measured at 514 nm upon excitation at
374 nm at 258C. At this temperature all the vesicular
aggregates remained in their rigid gel-like states (see below).
As the pH of the vesicular solution was raised from 5.8 to 10.2,
the fluorescence intensity (514 nm) decreased initially instan-
taneously to about 63 Æ 3% for 2a, 83 Æ 4% for 2b and 2c,
and 72 Æ 4% for 2d ± 2g of the original value (recorded prior
to pH adjustment). We attribute this instantaneous loss in the
fluorescence intensity to the deprotonation of the riboflavin
molecules that were bound at the exovesicular surfaces.
However, the residual fluorescence intensity disappeared as a
function of time for the vesicular solutions of 2b ± 2 f, but
remained virtually unchanged for 2a and 2g. Since the
residual losses in the fluorescence intensities followed mono-
phasic time dependence for 2b ± 2 f, numerical fitting of the
kinetic data could be achieved by means of the monoexpo-
nential relation. The half-times of this process (t ) for all of
Dynamic light scattering (DLS): Lipid dispersions of 1, 2a ±
2
followed by three freeze ± thaw cycles afforded vesicles with
virtually unimodal distribution of particle diameters as
revealed by DLS (Table 1). Examination of DLS data shows
that vesicles from dimeric lipids were invariably larger than
h obtained by bath sonication of the hydrated lipid film
�
Table 1. Riboflavin entrapment and apparent OH permeation rates
across different bilayer membranes at 258C.[
a]
3
8
[d,e]
Lipid conc. Hydrodynamic
%
t
1/2
10 kperm 10 P
diameter [nm][
b]
[c]
� 1
� 1
[
mm]
entrapment
[min] [s
]
[cms
]
[
f]
2
2
2
2
2
2
2
a
0.73
93
52
29
29
30
37
50
2.16
2.13
4.13
1.87
1.61
2.02
2.18
±
±
b 0.72
0.71
d 0.71
21.9
10.2
4.9
0.5
1.1
2.4
6.2
0.72
0.47
0.99
4.56
17.1
±
c
1/2
these vesicles and the related data are given in Table 1.
Using the respective bilayer thicknesses as obtained from
XRD studies (see below) and assuming the time-dependent
loss of riboflavin fluorescence as a rate-limiting permeation of
e
f
g
0.67
0.63
0.62
1.9
0.7 15.6
[g]
±
[
a] The time-dependent decay of fluorescence intensity at ꢂ514 nm was
�
exovesicular OH into the vesicleꢁs interior aqueous compart-
monitored as a function of time. In each case, the extent of decay was
followed beyond 90%. The decay followed an apparent monoexponential
path. The rate constant values represent averages of three independent
experiments, and the reproducibility was Æ3% for individual samples.
�
1
ment, we can calculate the permeation constant (P in cms )
�
�
of OH towards vesicles at pH 10.2. The differences in OH
permeability provide useful information about the vesicular
packing. The rate of permeation or permeability constant of
the individual vesicular system clearly appears to depend
significantly on the length of the spacer chain (m value)
[
b] As obtained from DLS. [c] The entrapment of riboflavin given in
percentages has been corrected for an apparent absorption of marker
molecules in each case. [d] P is the permeation constant of OH toward
vesicles at pH 10.2. [e] Diameters of the vesicles and bilayer thicknesses
were obtained from DLS measurements and XRD studies, respectively.
Time-dependent loss of fluorescence intensity may be interpreted as a
permeation-limited deprotonation of riboflavin entrapped in the internal
aqueous compartment by OH . [f] No significant time-dependent changes
in fluorescence was seen at 258C. However, upon heating the vesicle
beyond T
instantaneous.
�
‡
connecting the two NMe2 groups. Thus changing the m value
from 4 to 16 leads to >31-fold increase in the transmembrane
permeation rates. When corrected for bilayer thicknesses the
corresponding ratio remains at >24. With the increase in
�
�
spacer chain length (m ꢁ 3), the rate of OH permeation
m
resulted in the release of the doped dye. [g] The decay is
increases, presumably because of the increased spacer chain
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2337 $ 17.50+.50/0
2337

S. Bhattacharya, S. De
FULL PAPER
looping and consequent loss of tightness in the molecular
packing.
Interestingly, with vesicular 2a and 2g the initial fluores-
cence intensity underwent a rapid decay upon pH adjustment
hydrocarbon chains in 1 tend to avoid water exposure and
thus self-aggregate. Although its chains pack closer, the
‡
cationic Me N headgroups tend to stay away from each other
3
within the vesicles to avoid unfavorable electrostatic repul-
(
5.8 !10.2). However, subsequent time-dependent losses of
fluorescence were not recorded with vesicular 2a and 2g at
58C as was observed with the vesicles of 2b ± 2 f. We believe
sion. Thus in the bilayers, an equilibrium distance between the
‡
Me N groups in 1 is maintained as a result of compromise
3
2
between the two opposing tendencies. However, in dimeric
‡
that the observed resistance to transmembrane permeation
with the vesicular 2a may originate from the tight nature of
the lipid packing in these bilayers at 258C (gel state). Indeed
when the riboflavin-loaded vesicles of 2a were heated above
lipid 2 with m ꢃ 4, the distance between the Me N groups
2
within a gemini unit is shorter than the equilibrium separation
‡
between the monomeric Me N groups in vesicular 1. This is
3
true even when the spacer chain in 2a or 2b adopts a fully
extended s-trans conformation. An obvious consequence of
this is the more efficient packing within vesicular 2a or 2b
compared with 1. As the headgroup separation in 2 ap-
proaches the equilibrium distance with the increase in m
value, the packing organization in the dimeric lipids start to
simulate that found in 1. The observed variations in the
vesicular properties as a function of their spacer chain lengths
with the herein described dimeric lipids are analogous to
those of the micelle-forming gemini amphiphiles described by
Zana.[ Hence, the solid-to-fluid transition temperatures of
2c ± 2e (m ˆ 5, 6, 12) are quite similar to that of vesicular 1.
On the other hand, the melting transition profiles of the
dimeric lipids, m ꢁ 16, appeared to be considerably broader
their T , time-dependent monophasic fluorescence decay
m
could be clearly seen. Interestingly, the corresponding plot of
the fluorescence emission (after pH adjustment) against T
broke at a temperature that was comparable with the
corresponding Tm value for the vesicular systems of 2a. In
the case of the corresponding gel-filtered vesicles of 2g,
however, thawing beyond Tm did not show any evidence of
riboflavin release. It appears that the decay of the emission
intensity from 2g is so fast at 258C that it merges with the
initial instantaneous decrease in fluorescence. This in turn
indicates that the lipid packing in vesicular 2g is quite loose
making it very leaky under the experimental conditions
described.
15]
(
hence less cooperative) than their counterparts with m ꢃ 12,
Thermotropic phase-transition behavior: A well-defined,
sharp, single, main phase transition (gel !liquid crystalline)
whose melting transitions were quite sharp. The increased
propensity of the spacer chain looping in m ˆ 6 ± 16 might
impair the interlipid packing in these vesicles and lead to
lower melting temperatures than that of 2a or 2b. The melting
profiles for lipids with m ˆ 20 ± 22 were, however, broader
peak (T ) was observed from practically all the dispersions
m
during the heating runs in DSC. In addition to a peak due to
the main transition for 2a, two minor peaks at ꢀ66 and 718C
were also observed. Suspensions of each of 1, 2a, 2b, and 2c
also showed one clearly detectable peak due to pre-transition,
which appeared at approximately 108C, prior to the main
transition. All lipid suspensions showed virtually reversible
transition behavior during heating as well as cooling runs. At
although their T values were quite high. This suggests that
m
the membrane organizations and packing motifs of gemini-
lipid units with high m value (ꢁ16) are quite different from
that of the lipids with low and intermediate m values. The T
m
values and cooperativity units (CU) are compared as a
function of m value in Figure 1.
different scanning rates, identical values of enthalpy, T , and
m
the height of the transition profile were obtained.
In general, the DH and DS values for the main transition of
Thermotropic properties of vesicular 1 and 2a ± 2h at much
lower concentrations (1.0 ± 2.5mm) were also examined by
microcalorimetry. Vesicular specimens of the lipids with m ˆ
lipids with varying m values are linearly correlated with T
m
values except for 2a (m ˆ 3). The presence of multiple
transitions makes the situation complex for 2a. However,
when the enthalpic contributions of other transitions are
added to that of the main transition, one finds reasonable
correspondence for the total DHcalcd value with the entire
melting transition of 2a. It should be noted that dimers 2a ±
2h are mixtures of two diastereoisomers. Therefore the
complex melting behavior manifested with 2a may originate
from the formation of small phase-separated domains of the
different diastereomeric components. It is possible that such
phase segregation may be significant only when the spacer
chain length is small (m ˆ 3), and with higher m values the
corresponding lipid diastereomers could somehow accommo-
0
, 3, 4, and 5 reproduced the main Tm values observed at
15mm, and also showed the presence of distinct pre-
1
transition peaks, a phenomenon often observed with the
melting of natural phospholipids. Additional minor peaks at
66.1 and 71.68C for 2a and 58.08C for 2g were also observed.
HPLC analyses of the DSC samples of either 2a or 2g
confirmed that additional peaks were not due to the presence
of any impurities or caused by lipid decomposition.
Since DSC and microcalorimetry results at high and low
lipid concentrations, respectively, were essentially the same,
we have confined our discussions to microcalorimetry only.
The solid-to-fluid melting transition temperatures (T ) of
date each other and mix well giving rise to a single T value.
m
m
various members of 2 depend strongly on the m value of the
spacer chain [� (CH ) �]. Relative to 1 (m ˆ 0), there is a
Fluorescence anisotropy measurements: All the lipid disper-
sions showed reasonably cooperative main phase transitions
2
m
substantial increase in the T values of the membranes
m
derived from dimeric 2a. In addition, unlike the membranes
that are assembled from 1, which show a sharp endothermic
main transition at ꢀ458C, those made from 2a exhibit a more
complex melting pattern. Upon dispersion in water, the
(T ) in the plots of r versus T (Figure 2). The breaks related to
m
the T from these plots were generally similar to the T values
m
m
obtained from microcalorimetry, except for the lipids with
m ˆ 20 and 22, which showed broader and apparently lower
2
338
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2338 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Lipids
2335±2347
consequence of perturbation
‡
within the Me N headgroups
2
and/or direct interaction of the
spacer chain with the nonpolar
hydrocarbon chain segments.
Taken together these results
point towards the differences
in organizational features of the
membranes with lipids of m
>
16 to their low m value coun-
terparts.
X-Ray diffraction (XRD): In
order to gain information on
the lamellar thicknesses of
these aggregates, self-support-
ing cast films of aqueous dis-
�
1
persions (10 mgmL ) of 1 and
a ± 2h were examined by
2
Figure 1. Dependence of thermotropic phase-transition temperature (T
different amphiphiles 1, 2a ± 2h on m values.
m
) and cooperativity unit (CU) of
X-ray diffraction. A series of
higher order reflections for all
the lipid dispersions were ob-
served in the diffraction pat-
T values (ꢀ62 and ꢀ588C, respectively) in this experiment.
terns. Table 2 summarizes the long spacings obtained from the
cast films of each of these lipids, and these results clearly
suggest the formation of different kinds of aggregates from
lipids of various m values.
m
The observed broadness in the r versus T transition profiles
are consistent with lower CU values seen in calorimetric
studies. The apparent orders of the lipid hydrocarbon chains
at both above and below T were also found to be influenced
m
by their m values. Figure 2 (inset) shows the variations of r
values above and below Tm with m value. Thus with the
Table 2. Unit layer thicknesses obtained from XRD studies of self-
supporting cast films of 1, 2a ± 2h and their proposed packing.
Lipid
m value
Unit bilayer thickness [Š]
Proposed packing[c]
[
a]
[b]
observed
calculated
0
3
4
5
6
2
6
0
2
49.58
48.07
49.81
49.12
49.35
31.95
29.59
29.80
30.30
49
49
49
49
49
C
C
C
C
C
16 bilayer
16 bilayer
16 bilayer
16 bilayer
16 bilayer
[
d]
[e]
[e]
1
1
2
2
49, 16.8, 31
9, 21.9, 31
49, 27
49, 29
Tilted interdigitated bilayer
Tilted interdigitated bilayer
Monolayer ± bilayer
[
d]
[
d]
[
d]
Monolayer ± bilayer
[
a] As obtained from reflection XRD of cast films. [b] Lengths of two
molecular layers of lipids as obtained from energy-minimized CPK models
INSIGHT). [c] See Figure 6. [d] Length of unit monolayer± bilayer
arrangment of lipids as obtained from energy-minimized CPK models.
e] Length of unit tilted, interdigitated bilayer arrangment of lipids as
(
[
obtained from energy-minimized CPK models.
The diffraction patterns of 1 and 2a ± 2d showed membrane
thicknesses (d) of 49.6, 48.1, 49.8, 49.1, and 49.3 Š, respec-
tively, on the basis of the higher order reflections (n ˆ 3, 4, 5,
6). Assuming the long hydrocarbon chains in these lipids to be
primarily in an s-trans conformation, the widths of these
bilayers were estimated to be ꢀ49 Æ 1 Š from the CPK model
of the two molecules oriented parallel to the bilayer normal
(see below). This suggests the formation of a untilted bilayer
arrangement (Figure 3a) with lipid 1 and the gemini lipids of
m values 3, 4, 5, and 6. These cast films, however, did not show
any evidence of polymorphism as other types of reflections
were not observed. Cast films of dispersions of 2e (m ˆ 12)
Figure 2. Fluorescence anisotropy (r) versus T [8C] plots for vesicular 1
(
*), 2a (*), 2c (
variations of r values with m value at above (
&
), 2 f (
&
), 2g (~), and 2h (~) samples. The inset shows the
&
) and below (*) T
m
.
increase in m value (ꢁ16), the apparent order in the gel state
of these bilayers was found to be considerably less than their
low m value counterparts. At the same time the lipids with
m ˆ 3, 4, and 20 ± 22 became less fluid above T relative to
m
their intermediate m value counterparts. This could be a
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2339 $ 17.50+.50/0
2339

S. Bhattacharya, S. De
FULL PAPER
bilayer hybrid organization as shown in Figure 3f. Since these
lipids exhibit high Tm values, the possibility of formation of
interdigitated, tilted, bilayers (Figure 3d) may be ruled out.
EPR spectra in the gel phase: The EPR technique has been
frequently used to understand the nature of phases, for
example, mixed or fully interdigitated, that occur in the gel
phase.[ Here, we have investigated the hydrocarbon chain
packing in conjunction with the possible effects of spacer
chain looping using a series of fatty acid spin labels with the
nitroxide group located at specific positions along the chain in
the gel state. This study is based on the relative changes in
rotational motion and order parameters of the fatty acid spin
labels embedded in the vesicular dispersions.
16]
Figure 3. Schematic representation of possible bilayer forming motifs with
gemini lipids as a function of m value.
The first harmonic absorption EPR spectra of different
fatty acid spin-label positional isomers incorporated in
vesicular dispersions are shown in Figure 4. At 258C, the
spectra for 1, 2a, and 2e display a lesser degree of
immobilization on proceeding towards the terminal methyl
end of the chain from the polar headgroup region as they
exhibit extensive motional narrowing of the hyperfine aniso-
and 2 f (m ˆ 16) were found to consist of considerably thinner
intercalated layers with thicknesses of 32 and 29.6 Š, respec-
tively. Such widths of unit layers cannot be explained by tilting
alone (Figure 3c). Probably these lipid molecules interdigitate
at the midplane of the bilayers
in addition to tilting with re-
spect to the bilayer normal
(Figure 3d). Since the spacer
chain is quite long for these
lipids, looping is increasingly
favored. So, the propensities of
these additional modes of lipid
packing increase, this also helps
to fill the voids that might
otherwise be created inside the
lipid assemblies. An alternative
mode shown in Figure 3e is not
feasible, as in this situation the
hydrocarbon chains have to
protrude into the polar head-
group regions.
Cast films of dispersions of
2
g and 2h exhibited one series
of higher order reflections cor-
responding to layer thicknesses
of 29.8 and 30.3 Š, respectively.
These values compare favor-
ably with the stacked lamellae
thickness of ꢀ28 Š found for
another model archaebacterial
membrane lipid developed by
[
13e]
Thompson et al.
To explain
their reflection XRD data, we
suggest a kind of monolayer±
Figure 4. EPR spectra of 5-, 12-, and 16-doxylstearic acids at 258C incorporated in solid gel phases of membranes
of a) 1, b) 2a, c) 2e, and d) 2g.
2
340
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2340 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Lipids
2335±2347
tropy on the conventional EPR time scale. However, there is a
substantial difference in the chain mobility as a function of
lipid m value. For lipids with m ˆ 0 (1) and 3 (2a), the
immobilization of the hydrophobic tail appears to increase
from position-5 to position-12, thereafter it decreases again at
position-16. This is obvious when one compares both the
labeled much closer to the polar headgroup region (5-NS).
This is because the terminal methyls of the fatty acid chains of
lipids in 2g are located near the bilayer-aqueous interface,
presumably owing to the formation of a interdigitated
monolayer± bilayer kind of lipid organization. This type of
behavior is reported in the literature[ for the interdigitated
lipid phases in phosphatidylcholine (PC) and phosphatidyl-
glycerol (PG) induced by glycerol and polymyxin, respective-
ly. Hence these results can be attributed to the formation of
fully interdigitated lipid bilayers by 2g compared with the
noninterdigitated gel-phase lipid bilayers by 1, 2a, and 2e.
These results and conclusions are in accord with the results
obtained from the other studies.
17]
splittings and the widths of the outer hyperfine lines (T )
k
of the spectra (Figures 4 and 5, Table 3). The T value
k
Fusion assay: Since the primary function of membranes is to
maintain the integrity of cells within their environment and to
maintain the organization of subcellular compartments inside
the cell membrane, events like fusion have to be adequately
regulated in space and time. Information about the tendencies
of 1 and 2a ± 2h to fuse under given conditions is useful, since
the molecular packing in the membranes is delicately bal-
anced depending on the m value of the lipid dimers. We also
wanted to know whether a temporal separation of the
membrane leaflets of the interacting vesicles would be
allowed with these dimeric lipids. This is important, since
the formation of a hemifusion intermediate requires a
temporal separation. These considerations prompted us to
explore the relative fusion efficiencies of individual vesicular
systems, especially those composed of membrane-spanning
lipids. Here we present the relative fusogenic properties of
k
Figure 5. Outer hyperfine splitting T (G) as a function of the label
position, in the fatty acid spin labels in dispersions of 1 (*), 2a (
and 2g (!).
&
), 2e (~),
vesicular 1, 2b, 2d, 2e, and 2g induced by the inorganic anion
SO42
�
.
Table 3. ESR spectral parameters of fatty acid spin labels in 1, and 2a ± 2h
at 258C.[
a]
In this study we employed the fluorescence resonance
energy transfer (FRET) assay[ that relies upon the overlap
between the emission and excitation spectrum of two,
nonexchangeable, fluorophoric phospholipid analogues, N-
NBD-PE (donor) and N-Rh-PE (acceptor). Since the effi-
ciency of FRET between the two given fluorophores is
dependent upon their spatial separation, this technique
provides the details of how lipid mixing occurs during
membrane fusion. The fluorophores have been incorporated
into dimeric lipid membranes at concentrations that do not
18]
 10� 9 [s]
21.17
±
3.53
10.4
±
Fatty acid spin label
T
k
[G]
t
0
S
1
2
2
2
5-NS
29.98
36.92
17.16
29.07
36.05
18.94
28.98
17.51
17.51
44.31
42.82
0.454
±
0.297
0.524
±
0.363
0.463
0.335
0.326
0.556
1.0
1
1
2-NS
6-NS
a
e
g
5-NS
1
1
2-NS
6-NS
5.4
5-NS
10.07
10.27
1.78
8.44
2.23
1
1
2-NS
6-NS
5-NS
6-NS
1
[
a] See Experimental Section for the details.
(
ꢀ30 G) obtained for these bilayers is in good agreement
with that of 1,2-dipalmmitoyl-sn-glycero-3-phosphocholine
(
5
DPPC; ꢀ30.8 G) in HEPES buffer (pH 7.4) as sensed by
[
16b]
-NS, incorporated in the vesicular DPPC.
It is to be noted
that for 2e (m ˆ 12), there is little difference in spectral
anisotropy with the label at the 12-position compared with
that at the terminal methyl end.
In contrast, the maximum hyperfine splittings, T , of the
k
spectrum of a fatty acid, spin-labeled near the chain terminal,
1
6-NS for 2g (m ˆ 20) are significantly larger compared with
those for 1, 2a, and 2e. Strikingly, this value remains almost
unchanged when one compares this with that of a segment
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2341 $ 17.50+.50/0
2341

S. Bhattacharya, S. De
FULL PAPER
perturb the parent membrane structure. In this situation the
efficiency of energy transfer should be proportionally related
to their surface density. Upon fusion of fluorescently labeled
vesicles with nonlabeled vesicles (probe dilution method),[
bilayer mixing reduces the surface density of the fluorophores
and results in a decrease of FRET efficiency; this can be
followed continuously either as an increase in donor NBD
fluorescence or a decrease in the fluorescence of the acceptor
19]
[
20]
rhodamine.
When cationic vesicles containing both N-NBD-PE
0.6 mol%) and N-Rh-PE (0.6 mol%) were incubated with
(
an equal amount of nonlabeled vesicles, no change in the
rhodamine fluorescence could be detected. This indicates that
the probes did not spontaneously transfer by means of a
mechanism with the involvement of diffusion of monomers
through the aqueous phase between labeled and nonlabeled
membranes. In the presence of Na SO (10mm), however, a
2
4
slow decrease in the rhodamine fluorescence was observed
Figure 6) in all the cases, except that with 2g. This suggests
(
Figure 7. Percentage change in fluorescence with time, 1 (*), 2b ( ), 2d
~), 2e (!) and 2g (^).
&
(
a result of the onset of aggregation between the vesicles of 2g,
and this is generally not observed in the probe dilution
method.[ Very little is known about the fusogenic properties
of bolaform lipids. It may be possible that larger energy
barriers and greater geometrical constraints are involved in
the fusion of membranes composed of this lipid (m ˆ 20);
these resemble membranes formed from bolaamphiphiles.
Indeed a similar phenomenon is reported for the lipids
extracted from the thermophilic archaeon Sulfolobous sol-
fataricus irrespective of the fusogen employed,[ while the
large unilamellar liposomes composed of bipolar tetraether
lipids extracted from the thermophilic archaeon Sulfolobous
acidocaldarius were found to undergo a nonleaky fusion
process in the presence of calcium phosphate, the rate being
dependent on the pH and the concentration of the salt.[
19]
20]
Figure 6. Decay of rhodamine fluorescence with time: a) 1, b) 2b, c) 2d,
d) 2e and e) 2g.
21]
that Na SO4 probably induces fusion and intermixing of
2
membrane lipids among the liposomes of 1, 2b, 2d, and 2e.
Importantly, however, the rate of this fluorescence decay is
different for membranes with different m values. For compar-
ison, the percentage change in fluorescence (% DF/DFmin) for
each lipid vesicle was plotted against time (Figure 7). While
the percentage change in fluorescence for 1 was 60.7% after
Interaction of dimeric lipids with cholesterol: Cholesterol is
an integral part of mammalian-cell membranes[ and it is
known to noncovalently associate with different lipids within
the cell.[ Such interactions lead to an increase in the
orientational order of lipid hydrophobic chains in mem-
branes.[ This in turn abolishes the solid !fluid transition
temperature of the resulting bilayers.[ Does cholesterol
interact with dimeric lipids in the same way as it does with
monomeric lipids? Are such interactions dependent on the m
value of the lipid dimers? To examine the effects of
cholesterol inclusion in dimeric lipid membranes, we have
employed two widely used methods, that is, fluorescence
22]
23]
24]
25]
30 min, for 2b it was only 5.6% after the same period. This is
probably because of the rigidness of the bilayer packing in the
case of 2b (m ˆ 4) compared with that of monomeric 1 (m ˆ
0). However, the percentage change in fluorescence intensity
was enhanced with the increase in spacer chain length up to a
certain m value (17.0% for 2d and 56.7% for 2e after 30 min
of the injection of salt solution to the vesicular dispersions).
This could indicate possible fusion of these membranes
through a process that involves transient separation of two
bilayer leaflets and the formation of a hemifusion intermedi-
ate structure.
In contrast, the vesicles of 2g showed no decrease in the
rhodamine-fluorescence intensity with time. Instead it showed
a marginal increase in the fluorescence intensity. It is not
apparent why this happens. One possibility is that this may be
1
anisotropy and H NMR spectroscopy.
We measured the fluorescence anisotropy (r) of 1,6-
diphenyl-1,3,5-hexatriene (DPH) solubilized in different ve-
sicular assemblies in the presence or absence of cholesterol.
For this study we selected four representative lipids 1 (m ˆ 0),
2a (m ˆ 3), 2e ( m ˆ 12), and 2g (m ˆ 20), which were shown
to form three different types of membrane organizations.
Figure 8 (a ± d) shows that the effect of cholesterol incorpo-
ration on the thermal behavior of a given lipid depends on the
2
342
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2342 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Lipids
2335±2347
dimeric lipids. Remarkably, with dimeric lipids upon choles-
terol incorporation, the orders in their fluid states were hardly
affected, while the resulting mixtures in their gel phases were
rendered disordered. We believe that the presence of a
(
CH ) -spacer chain in these lipids influences the location of
2 m
cholesterol in the resulting membranes, and it is the temper-
ature-induced conformational variation of the spacer chain
that predominantly regulates the solid !fluid transition even
when cholesterol is present.
To further ascertain the nature of the dimeric-lipid ± cho-
lesterol interactions in their fluid states, various cholesterol-
free and cholesterol-laced vesicular preparations were ana-
1
lyzed by H NMR spectroscopy above their T s. Studies were
m
carried out at 608C for 1, 658C for 2a, and 708C for 2g.
Compound 2e was not studied by this method, since the order
in its fluid state remained almost unaffected upon incorpo-
ration of cholesterol as observed from its r versus T plot.
1
Figure 9 shows the H NMR spectra of chain polymethylene
(
CH ) and terminal methyl groups (CH ) for different lipid
2
n
3
systems in the absence and presence of cholesterol. Addition
of cholesterol in increments (0 ± 30 mol%) led to progressive
broadening of the downfield (CH2)n proton signals in 1,
indicating that inclusion of cholesterol significantly restricted
Figure 8. Fluorescence anisotropy (r) versus temperature plots for lipid
the mobility of the (CH ) units; however, it had much less
2 n
vesicles: a) 1, b) 2a, c) 2e and d) 2g with 0 mol% (*), 10 mol% (
0 mol% (~) and 30 mol% (*) cholesterol.
&
),
effect on the terminal CH groups. Similar results were also
3
2
reported for cholesterol doped phosphatidylcholine vesi-
cles.[
23b, 26]
Strikingly, the virtually opposite observation was
and CH
3
lipid structure, in particular on the m value. With vesicular 1,
the breaks in the r versus T plot signifying the melting
transition became gradually blurred with the increase in
cholesterol percentage. This type of behavior is reminiscent of
recorded with 2a. Both the resonances due to (CH
)
2
n
got sharpened with the increasing percentage of cholesterol.
Under comparable conditions 2g, however, showed a com-
plex behavior. Addition of cholesterol (10 mol%) initially
[
26]
that reported for cholesterol-phosphatidylcholine and cho-
broadened both the (CH ) and terminal CH signals. With
2 n 3
[
10f]
lesterol ± cationic-lipid
mixtures. While the order in the gel
20 mol% cholesterol, these signals became considerably
sharper. However, with 30 mol% of cholesterol these signals
appeared to broaden again.
The observed differences in the interactions of cholesterol
with different types of cationic lipids could be explained on
phase was little influenced by the inclusion of cholesterol in 1,
it was substantially enhanced in its fluid state. In contrast,
breaks related to solid !fluid melting were not abolished
even after inclusion of 30 mol% cholesterol in the case of the
1
Figure 9. The H NMR spectra of chain polymethylene (CH
2
)
n
and terminal methyl (CH
3
) groups for different lipid systems in the absence and presence of
different mol% of cholesterol above T
m
, a) 1 (at 608C), b) 2a (at 658C), and c) 2g (at 708C).
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2343 $ 17.50+.50/0
2343

S. Bhattacharya, S. De
FULL PAPER
the basis of variations in the molecular packing of these lipids
in the vesicular matrices. The interaction between 1 and
cholesterol should originate mainly from the hydrophobic
contacts, unlike lipids containing ester or amide groups in
which hydrogen bonding takes place between the lipid
monomer and 3b-OH of cholesterol in addition to the
hydrophobic interactions. Thus 1 should be able to accom-
modate cholesterol in the membrane depending on the
miscibility of the latter. Since it is believed that above the
Tm value of the lipid, the presence of cholesterol induces
lipid-molecular level. These findings emphasize the continu-
ing need for new designs of synthetic lipid structures to
expand our understanding of their behavior upon membrane
formation.
Experimental Section
Methods and materials: Descriptions of the general procedures and the
analytical instruments used for various spectroscopic characterizations
[
10b]
[
27]
have been described previously.
n-Hexadecyl bromide, 1,2-isopropyli-
additional phases; membrane fluidity also changes accord-
ingly. Current findings indicate that, in general, the mem-
branes composed of dimeric lipids can withstand insertion of
more cholesterol compared with the monomeric counterparts.
This is a significant observation. Generally inclusion of
cholesterol stiffens the resulting membranes. It appears that
since the herein described lipid dimers have four hydrocarbon
chains, the cholesterol inclusion into such membranes does
not lead to a significant increase in the orientational order of
the resulting aggregates.
dene glycerol, 1,3-dibromopropane, 1,4-dibromobutane, 1,5-dibromopen-
tane, 1,6-dibromohexane, 1,12-dibromododecane, carbon tetrabromide,
triphenylphosphine, 1,6-diphenyl-1,3,5-hexatriene (DPH), and riboflavin
were purchased from Aldrich. 1,16-Dibromohexadecane, 1,20-dibromoei-
cosane, and 1,22-dibromodocosane were synthesized from tetradecane-
1
,14-dioic acid, eicosanedioic acid, and docosanedioic acid, respectively, by
the reduction of their respective ethyl esters with lithium aluminium
hydride in dry THF followed by bromination with HBr/H SO at 808C for
24 h. They were separated by chromatography over a silica gel (60 ±
20 mesh) column with hexane as the eluent.
2
4
1
Synthesis: Synthesis of monomeric lipids and different precursors of the
dimeric lipids are synthesized as follows. First, 1,2-isopropylidene glycerol
(
3) was protected with benzyl chloride in dry benzene under refluxing
conditions to obtain the corresponding benzyl ether 4, and the isopropyl-
Conclusion
idiene group was then removed with aqueous HCl/MeOH at room
temperature to obtain dl-a-O-benzylglycerol.
[
28]
The series of experiments described above, reveal the effects
of lipid dimerization. The introduction of a polymethylene
spacer chain at the headgroup brings about a dramatic effect
on the aggregation behavior, membrane organization, and
lipid packing. Dimeric lipids with low m values (3, 4) and high
m values (20 ± 22) showed exceptional thermal, lipid-packing,
and cholesterol-association properties. Possibly because of an
increase in headgroup bulk and spacer chain looping with the
rise in m value up to m ꢃ16, the membranes from the
corresponding lipids show a poor hydrocarbon chain packing.
This is also manifested in the increase in permeability with the
rise in m value. This probably also enables these vesicles to
undergo facile fusion relative to their low (m ˆ 3, 4) or high
dl-a,b-Di-O-hexadecyl-O-benzylglycerol (5): A mixture of dl-a-O-benzyl-
glycerol (5.0 g, 27.4 mmol), 1-bromohexadecane (18.5 g, 60.4 mmol), and
powdered potassium hydroxide (5.9 g, 104.2 mmol) in dry benzene (50 mL)
was stirred and refluxed for 24 h. The cooled mixture was diluted with
chloroform (50 mL) and washed with water, and the organic layer was
4
dried over anhydrous MgSO . The solvent was removed under vacuo and
the crude product was purified by column chromatography over a silica gel
column with a mixture of hexane and ethyl acetate to afford a viscous oil
1
(
5
10.0 g, 58%). H NMR (CDCl
3
, 400 MHz) d ˆ 0.90 (t, 6H, CH
), 3.40 ± 3.75 (m, 9H, OCH , OCH), 4.55
(s, 2H, CH Ph), 7.31 (s, 5H, aromatic Hs).
3
), 1.10 (brs,
2H, CH
2
), 1.50 ± 1.80 (m, 4H, CH
2
2
2
dl-a,b-Di-O-hexadecylglycerol (6): A solution of 5 (10 g, 15.8 mol) in
methanol (150 mL) was hydrogenated over 5% Pd/C (0.2 g) for 24 h at
4
6 psi pressure during which almost all the starting materials were
converted to the alcohol. Methanol was removed from the reaction
mixture, and the solid mass obtained was heated with EtOAc. To remove
the catalyst, this solution was filtered and kept at room temperature for
crystallization. Upon repeated crystallizations, a flaky white solid was
(
m ˆ 20 ± 22) m value counterparts. The membranes derived
from the lipids with m values 20 ± 22 are thermally more
resistant to melting. One point that deserves special mention
is the evidence of strong interdigitation with high m values as
revealed by EPR studies. The membranes from the same set
of lipids have also been shown by X-ray diffraction studies to
adopt interdigitated monolayer± bilayer hybrid structures.
This may be one way by which the bolaphilic lipids that occur
in thermophilic bacteria allow the resulting organisms to
thrive at elevated temperatures. Finally, a novel feature of
cholesterol interaction with dimeric lipids has been identified,
that is, even after its inclusion up to physiologically relevant
concentrations (30 mol%) present in mammalian-cell mem-
branes, the gel ± liquid crystalline melting transition is not
abolished. Since eukaryotic cells are rich in cholesterol, such
an observation may suggest that at the cellular level this may
be a plausible mechanism by which solid-to-fluid transitions
could still be maintained.
obtained (8 g, 92%), which was found to be pure by TLC (R
f
ꢀ 0.3, CHCl
, 400 MHz) d ˆ 0.89 (t, 6H,
), 1.50 ± 1.80 (m, 4H, CH ), 3.40 ± 3.75 (m,
3
).
[
28]
1
M.p. 478C (lit. 48 ± 498C); H NMR (CDCl
CH ), 1.15 ± 1.40 (brm, 52H, CH
H, OCH , OCH).
dl-a,b-Di-O-hexadecylglycerol bromohydrin (7):
(1.44 g, 4.32 mmol) in dry CH Cl2 was added dropwise over a period of
3
3
2
2
9
2
A
3
solution of Ph P
2
30 min to a well-stirred solution of 6 (2.0 g, 3.62 mmol) and CBr
.3 mmol) in dry CH Cl at 08C. The reaction mixture was stirred at room
temperature for another 8 h till TLC showed the disappearance of the
starting material. CH Cl was removed under vacuum, and the residual
4
(2.42 g,
7
2
2
2
2
mass was repeatedly extracted with hexane. The combined extracts were
evaporated to dryness. It was purified by column chromatography over a
silica gel column with 100:1 hexane/EtOAc. The fraction which was
collected using 50:1 hexane/EtOAc gave a viscous liquid on removal of
organic solvent. It solidified on standing (1.97 g, 90%). TLC single spot
1
(
R
f
ꢀ 0.3, hexane); H NMR (CDCl
3
, 200 MHz) d ˆ 0.89 (t, 6H, CH
3
),
1.15 ± 1.40 (brm, 52H, CH ), 1.54 ± 1.80 (m, 4H, CH ), 3.33 ± 3.50 (m, 9H,
2
2
OCH
dl-[1,2-Di-O-hexadecyl-3-(dimethylamino)]propane (8): Compound 7(1 g,
.65 mmol) was dissolved in a 40% methanolic solution of Me NH (20 mL,
02 mmol) in a screw-top pressure tube. The tube was sealed, and its
contents were heated with stirring at ꢀ908C for 36 h. The solvent was
removed to afford a crude residue, which was then treated with NaHCO
(5%, 25 Â 2 mL) and the resulting product was extracted several times with
2 2
, OCH & CH Br).
1
1
2
Clearly the covalent connection of two monomeric lipids by
a simple �(CH ) � chain brings about effects at the mem-
2
m
brane level that are much more pronounced than one would
expect from seemingly trivial structural modifications at the
3
2
344
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2344 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Lipids
2335±2347
‡
‡
CHCl
3
(25 Â 4 mL). The organic layer was separated, washed with brine
CO . The solvent was removed
N CH
(apparent d, 4H, OCH
Br
12.23, N 1.70.
Eicosanediyl-a,w-bis[dl-N-(1,2-di-O-hexadecyl-3-propyl)-N,N-dimethyl-
ammonium bromide] (2g): M.p. 55 ± 608C; 1H NMR (CDCl
, 400 MHz)
2
CH
2
), 3.38 ± 3.50 (m, 20H, N (CH
3
)
2
, alkyl chain OCH
2
), 3.58
‡
(
10 mL), and passed through anhydrous K
2
3
2
), 3.60 ± 4.30 (several m, 10H, CH
2
(1520.29): calcd C 71.10, H 12.33, N 1.84; found C 71.37, H
2
NCH
2
, OCH);
to give a waxy semi-solid (0.9 g, 95%) which was found to be pure by TLC
C
90
H
186
N
2
O
4
1
(
R
f
ꢀ 0.8, CHCl
H, CH
m, 2H, CH
dl-N-(1,2-Di-O-hexadecyl-3-propyl)-N,N,N-trimethylammonium bromide
1): Compound 7 (0.3 g, 0.5 mmol) was dissolved in a 30% ethanolic
solution of NMe (20 mL, 102 mmol)in a screw-top pressure tube. The tube
3
/MeOH, 10:1). H NMR (CDCl
3
, 200 MHz) d ˆ 0.89 (t,
), 2.32 (s, 6H, N(CH ), 2.42 ± 2.60
N), 3.45 ± 3.75 (m, 7H, OCH , OCH).
6
(
3
), 1.15 ± 1.60 (brm, 52H, CH
2
3 2
)
2
2
3
d ˆ 0.88 (t, 12H, CH
CH ), 1.50 ± 1.80 (m, 12H, alkyl chain CH
3.36 ± 3.50 (m, 20H, N (CH
OCH
3
), 1.25 ± 1.40 (s and brm, 136H, CH
2
, spacer chain
‡
(
2
2
, spacer chain N CH
2
CH
2
),
‡
3
3
)
2
, alkyl chain OCH
), 3.62 ± 4.10 (several m, 10H, CH
2
), 3.52 ± 3.61 (m, 4H,
Br
2
‡
was sealed, and its contents were heated with stirring at ꢀ908C for 36 h.
The solvent was removed to afford a crude white mass, which was
recrystallized several times from acetone/EtOAc (20:1) (0.3 g, 90%). It was
2
2
NCH
2
, OCH); C94
H
194
N
2
O
4
(1576.4): calcd C 71.62, H 12.40, N 1.78; found C 71.76, H 12.42, N 1.52.
Docosanediyl-a,w-bis[dl-N-(1,2-di-O-hexadecyl-3-propyl)-N,N-dimethyl-
ammonium bromide] (2 h): M.p. 52 ± 558C; H NMR (CDCl
d ˆ 0.88 (t, 12H, CH
CH ), 1.50 ± 1.80 (m, 12H, alkyl chain CH
.40 ± 3.44 (m, 20H, N (CH
OCH
found to be pure by TLC (R
f
ꢀ 0.5, CHCl
lit. 115 ± 1168C); 1H NMR (CDCl
, 400 MHz) d ˆ 0.89 (t, 6H, CH
.20 ± 1.40 (brm, 52H, CH ), 1.60 ± 1.80 (m, 4H, CH ), 3.40 ± 3.75 (s and m,
3
/MeOH, 10:1). M.p. 113 ± 1148C
1
3
, 400 MHz)
[
29]
(
1
3
3
),
3
), 1.25 ± 1.40 (s and brm, 140H, CH
2
, spacer chain
2
2
, spacer chain N CH
‡
CH
),
2
2
2
2
‡
‡
1
6H, N(CH
662.97): calcd C 68.85, H 12.16, N 2.11; found C 68.63, H 11.97, N 1.87.
General procedure for the synthesis of bis(quaternary ammonium) lipids
3
)
3
, OCH
2
, OCH), 4.10 ± 4.20 (m, 2H, N CH
2
); C38H
80NO
2
Br
‡
3
3
)
2
, alkyl chain OCH
), 3.62 ± 4.10 (several m, 10H, CH
2
), 3.57 ± 3.58 (m, 4H,
Br
2
(
‡
2
2
NCH
2
, OCH); C96
H
198
N
2
O
4
(
1604.45): calcd C 71.87, H 12.44, N 1.75; found C 71.97, H 12.36, N 1.52.
(
2a ± 2h): A solution of dl-[1,2-O-hexadecyl-3-(dimethylamino)]propane 8
Vesicle preparation
(
1.0 mmol) and an appropriate a,w-dibromoalkane (0.35 mmol) in dry
Method A: Dry films of a given lipid (2.5 mmol) film were obtained on the
walls of Wheaton vials upon removal of chloroform from the solutions of
the lipid in the same solvent. The requisite amount of water was added to
this, and the resulting mixture was first kept at 48C for approximately 12 h
and then vortexed to allow optimal hydration of the lipid film. It was
vortexed for a few minutes and thawed to 70 ± 758C for 10 min and then
EtOH was refluxed over a period of 48 ± 72 h. It was cooled and the solvent
was evaporated to give a crude solid. It was repeatedly washed with dry
acetone to remove any traces of unreacted starting materials and finally
subjected to repeated crystallizations from a mixture of acetone and
methanol (ꢀ9:1). This gave a white solid and the yields ranged from 60 ±
75%. The purities of these lipids could be ascertained from TLC, the R
f
2
frozen with liquid N for 15 min. These cycles were repeated three times to
ranged from ꢀ0.25 to ꢀ0.3 in 10:1 CHCl
3
/MeOH. All compounds 2a ± 2h,
ensure optimal hydration. However, films of 2g and 2h did not form stable
suspensions even after 5 ± 7 freeze ± thaw cycles. For effective dispersal
these were sonicated with a bath sonicator (Sidilu Ultrasonic Bath) for
0 min at 808C and then subjected to freeze ± thaw cycles. All the lipid
suspensions including that of 2g and 2h were found to be stable for several
1
gave satisfactory H NMR and IR spectra, and C,H,N analysis. Pertinent
spectroscopic and analytical data are given below.
Propanediyl-a,w-bis[dl-N-(1,2-di-O-hexadecyl-3-propyl)-N,N-dimethyl-
ammonium bromide] (2a): M.p. 175 ± 1788C; H NMR (CDCl
2
1
3
, 400 MHz)
d ˆ 0.85 (t, 12H, CH
3
), 1.25 ± 1.50 (brm, 104H, CH
2
), 1.55 ± 1.80 (brm, 8H,
days.
‡
‡
alkyl chain CH
2
), 2.80 ± 2.90 (brm, 2H, N CH
2
CH
2
CH
2
N ), 3.30 ± 4.20 (m,
3
Method B: Solid lipid (2.5 mmol) was dissolved in CHCl (5 mL) and to this
an equal volume of pure water was added. The resulting two-phase mixture
was briefly dispersed in a bath-sonicator (35 kHz) for 15 min at ambient
‡
‡
34H, N (CH
3
)
2
, OCH
2
, OCH, CH
2
N CH
2
); C77
H
160
N
2
O
4
Br
2
(1337.94):
calcd C 69.12, H 12.05, N 2.09; found C 68.96, H 12.09, N 1.83.
Butanediyl-a,w-bis[dl-N-(1,2-di-O-hexadecyl-3-propyl)-N,N-dimethylam-
monium bromide] (2b): M.p. 175 ± 1828C; H NMR (CDCl
3
temperature to produce an emulsion. From this, CHCl was slowly removed
1
3
, 400 MHz)
under reduced pressure with a rotary evaporator. The resulting preparation
was then subjected to 5 ± 6 freeze ± thaw cycles followed by a brief
sonication for 5 min above 708C. This afforded stable translucent
suspensions. Vesicles prepared by this protocol remained practically stable
for several days as judged by the examination of the onset of turbidity of
these suspensions as a function of time.
d ˆ 0.85 (t, 12H, CH
3
), 1.25 ± 1.50 (brm, 104H, CH
), 2.19 ± 2.30 (m, 4H, spacer chain CH
2
 52), 1.55 ± 1.80 (m,
8
H, alkyl chain CH
2
2
), 3.20 (s, 12H,
‡
‡
N (CH
3
)
2
), 3.30 ± 3.52 (m, 8H, OCH
, OCH); C78
N 2.05; found C 69.58, H 12.36, N 1.99.
2
), 3.60 ± 4.20 (m, 14H, CH
2
N CH
2
,
alkyl chain OCH
2
H
162
N
2
O
4
Br
2
(1351.97): calcd C 69.3, H 12.07,
Pentanediyl-a,w-bis[dl-N-(1,2-di-O-hexadecyl-3-propyl)-N,N-dimethyl-
ammonium bromide] (2c): M.p. 808C (soften), 888C (clear melt); H NMR
Transmission electron microscopy (TEM): This experiment used vesi-
cles(2.5mm) in water. Uranyl acetate (0.5% w/v) was used as the staining
1
(
CDCl
3
, 400 MHz) d ˆ 0.85 (t, 12H, CH
3
), 1.25 ± 1.40 (brm, 106H, CH
2
,
agent. Vesicles were prepared by the method B described above. One drop
‡
spacer chain N CH
2
2
CH
2
CH
2
), 1.55 ± 1.80 (m, 8H, alkyl chain CH
2
), 2.15 ±
TM
of a given vesicular solution was placed on a 400 mesh carbon ± formvar
-
‡
‡
2 2 3
.19 (m, 4H, spacer chain N CH CH ), 3.30 ± 3.50 (m, 20H, N (CH )
2
,
coated copper grid. A JEOL-TEM 200CX electron microscope was used as
alkyl chain OCH
2
), 3.58 ± 3.62 (m, 4H, OCH
2
), 3.65 ± 3.67 (m, 2H, OCH),
Br (1366.0): calcd C 69.46, H
[10k]
described.
and 96000.
The micrographs were recorded at magnifications of 70000
‡
3.70 ± 4.10 (m, 8H, CH
2
N CH
2
); C79
H
164
N
2
O
4
2
12.1, N 2.05; found C 69.68, H 12.17, N 1.85.
Dye entrapment and permeability: The vesicles from each of the above
Hexanediyl-a,w-bis[dl-N-(1,2-di-O-hexadecyl-3-propyl)-N,N-dimethylam-
� 1
lipids (1 mg1mL ) were prepared by the method A in aqueous media
1
monium bromide] (2 d): M.p. 788C (soften), 848C (clear melt); H NMR
containing riboflavin (20mm) at pH 5.8. At ambient temperature 3 mL of a
vesicle solution was loaded into a pre-equilibrated column (26 cm  1 cm)
of sephadex G-50M (Pharmacia). Upon elution with water (pH ꢀ5.8), gel-
filtered fractions containing dye-loaded vesicles were pooled. An aliquot of
a gel-filtered vesicle (1 mL) was placed into a cuvette and the pH was
adjusted from 5.8 to ꢀ10.2 by the addition of KOH solution(66 mL, 0.03m).
Immediately upon mixing, time-dependent changes in the fluorescence
emission at 514 nm (upon excitation at 374 nm) were measured at 25 Æ
0.58C with a Hitachi Model F-4500 Spectrofluorimeter. The time-courses
followed an apparent monoexponential decay, from which the rate
constants were obtained by analyzing the data by means of a pseudo-
first-order equation.
(
CDCl
3
, 400 MHz) d ˆ 0.85 (t, 12H, CH
3
), 1.25 ± 1.40 (brm, 108H, CH
2
,
spacer chain CH
2
), 1.55 ± 1.70 (m, 8H, alkyl chain CH
2
‡
), 2.10 ± 2.19 (brm,
‡
4
H, spacer chain N CH
2
CH
2
), 3.30 ± 3.50 (m, 20H, N (CH
OCH), 3.60 ± 4.05 (m, 8H,
(1380.02): calcd C 69.63, H 12.12, N 2.03;
found C 69.72, H 12.19, N 1.86.
3 2
) , alkyl chain
OCH
2
‡
), 3.55 ± 3.57 (brm, 6H, OCH ,
2
2 2 166 2 4 2
CH N CH ); C80H N O Br
Dodecanediyl-a,w-bis[dl-N-(1,2-di-O-hexadecyl-3-propyl)-N,N-dimethyl-
1
ammonium bromide] (2e): M.p. 808C (soften), 958C (clear melt); H NMR
(
CDCl
3
, 400 MHz) d ˆ 0.85 (t, 12H, CH
3
), 1.25 ± 1.50 (brm, 120H, CH
2
,
spacer chain CH
2
), 1.55 ± 1.70 (m, 8H, alkyl chain CH
2
‡
), 1.78 ± 1.84 (m, 4H,
‡
spacer chain N CH
OCH ), 3.58 ± 3.59 (brm, 4H, OCH
OCH); C86 Br
0.53, H 12.48, N 1.75.
2
CH
2
), 3.30 ± 3.50 (m, 20H, N (CH
3
)
2
, alkyl chain
‡
2
2
), 3.60 ± 4.08 (m, 10H, CH
2
NCH
2
,
Calorimetry: Thermograms were recorded at two different lipid concen-
trations with a Perkin ± Elmer DSC-4 instrument and an MC-2 ultra-
sensitive differential scanning calorimeter (Microcal) interfaced to a IBM
compatible computer. Lipid dispersions at ꢀ120 mmol concentration were
prepared by vortexing followed by freeze ± thaw treatments. These were
transferred into a steel pan (sample) which was sealed and kept at ꢀ908C
inside the calorimeter to ensure optimal hydration. DSC heating and
H
178
N
2
O
4
2
(1464.18): calcd C 70.55, H 12.25, N 1.91; found C
7
Hexadecanediyl-a,w-bis[dl-N-(1,2-di-O-hexadecyl-3-propyl)-N,N-di-
1
methylammonium bromide] (2 f): M.p. 65 ± 708C; H NMR (CDCl
3
2
,
,
4
00 MHz) d ˆ 0.88 (t, 12H, CH
3
), 1.25 ± 1.50 (s and brm, 128H, CH
), 1.50 ± 1.80 (m, 12H, alkyl chain CH , spacer chain
2
spacer chain CH
2
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2345 $ 17.50+.50/0
2345

S. Bhattacharya, S. De
FULL PAPER
cooling runs were recorded with the sample against a reference pan
containing an identical volume of pure water at a scanning rate of 5 Kmin
12-(4,4-dimethyl-3-oxazolidinyl)oxyoctadecanoic acid, 12NS, and 16-(4,4-
dimethyl-3-oxazolidinyl)oxyoctadecanoic acid, 16NS were purchased from
Sigma. Lipid dispersions were prepared by mixing the lipids with 1 mol%
�
1
in the Perkin ± Elmer DSC-4 instrument. Samples (2.5mm) for micro-
calorimetry made by the hydration of dry lipid film followed by sonication
were degassed under vacuum prior to loading into the sample cell. The scan
2 2
of the spin label probe in CH Cl solution, evaporating off the solvent with
a stream of dry nitrogen and then placing them under vacuum overnight.
The dried lipids were then dispersed in water to a final concentration of
�
1
rate was 908Ch . A baseline recorded with pure water in both the sample
�
1
� 1
and reference cells and normalized to mcaldeg , was subtracted from each
heat capacity to eliminate any variations due to mismatch between the two
cells. Excess heat capacity data for the transitions were filtered digitally and
the normalized curves were established by connecting the baseline before
and after the transition and then performing a point-by-point subtraction
from this straight line. Transition enthalpies were calculated by numerical
integration of the resulting heat capacity values as a function of temper-
4 mg0.4 mL by hydrating for approximately 30 min followed by vortex
mixing for 10 min at a temperature above the gel-to-fluid phase transition.
The suspensions were transferred to 1 mm (i.d.), 100 mL glass capillaries,
sealed at one end. The other end was sealed with parafilm. The sealed
sample capillary tube was accommodated in a standard 4 mm quartz EPR
tube.
EPR measurements: EPR spectra were measured on a Varian E-series
spectrometer operating at a X-band frequency of 9.05 GHz. Conventional
absorption EPR spectra were recorded at a modulation frequency of
100 kHz and a modulation amplitude of 4 G peak-to-peak, at a microwave
�
1
� 1
ature, [Eq. (1)]. Main transition entropies DS, in units of calmol
estimated based on the assumption that at the T the two phases are at
equilibrium, in which case Equation (2) applies, where DHcal is expressed in
K , were
m
�
1
calmol and T
m
in Kelvin. The vanꢂt Hoff enthalpy DHvH was obtained
power of 2 mW. The motional parameter, t , can be derived from spectral
0
[
30]
from the Equation (3) in which DT1/2 is the width at the half-maximum
excess specific heat. These enthalpies were used to determine the
cooperative unit (CU) of the transition, that is, the number of monomers
undergoing the phase transition from the relationship expressed in
Equation (4).
parameters as described, and is given by Equation (6), in which K ˆ
� 10
6.5 Â 10 s and is fixed arbitrarily at its limiting value in the case of rapid
isotropic tumbling, W is the width of the center line, and h and h are the
0
0
� 1
heights of the center and high-field first-derivative lines respectively.
t
0
ˆ KW
0
[(h
0
/h� 1)1/2 � 1]
(6)
„
DHcal ˆ CpdT
(1)
(2)
(3)
(4)
The order parameter S gives a measure of the amplitude of motion of the
long molecular axis about which the average orientation of the hydro-
carbon chains occur in the lipid bilayers. It is obtained from the anisotropic
outer and inner hyperfine splittings, T
Equation (7), where Tzz and Txx are the principal N-hyperfine coupling
tensors for a rigid lattice obtained from the single crystal spectra: Tzz
3
complete disorder results in a value of S ˆ 0.
DS ˆ DHcal/T
m
k
and T
?
, respectively, by using the
DHvH ˆ 6.9T
m
2/DT1/2
14
ˆ
2.9 G, Txx ˆ 5.9 G. The maximum value of S is 1.0 for perfect order, while
CU ˆ DHvH/DHcal
Fluorescence anisotropy studies: Upon excitation of 1,6-diphenyl-1,3,5-
hexatriene (DPH) at 360 nm embedded in various vesicular solutions
S ˆ (T � T )/(T � Txx)
(7)
k
?
zz
(
2.5mm), the fluorescence anisotropies (r) were measured with a Hitachi
Model F4500 spectrofluorimeter. At each temperature, the emission
spectra (390 ± 480 nm) were recorded by adjusting the polarizers in four
different positions. The anisotropy (r) at different temperatures for each
vesicular solution was calculated by employing Perrinꢂs equation, [Eq. (5)],
Fusion assay
Liposome preparation: 1,2-Dioleoyl-sn-glycero-3-phosphoethanolamine-
N-(7-nitro-1,3-benzoxadiazol-4-yl) (N-NBD-PE) and l-a-phosphatidyl-
ethanolamine-N-(lissamine rhodamine B sulfonyl) (Egg; N-Rh-PE) were
purchased from Avanti Polar Lipids. Lipid films were dried under high
rˆ (I
k
� I
?
G)/(I
k
‡2I
?
G)
(5)
vacuum and suspended at a concentration of 10 mgmL� 1 in HEPES buffer
(
10mm, pH 6.0). Lipids were dispersed by sonication with a bath sonicator
for 5 min at ꢀ608C followed by one freeze ± thaw treatment.
in which I
k
and I
?
are the observed intensities measured with polarizers
parallel and perpendicular to the vertically polarized exciting beam,
respectively. G is a factor used to correct for the inability of the instrument
to transmit differently polarized light equally. For each aggregate, r versus
T plots gave the information about the gel-to-liquid-crystalline phase
change as a function of temperature. The values reported herein for r are
the average values of three independent measurements. The systemic gel-
to-liquid-crystalline phase transition temperatures were calculated from
the midpoints of the breaks related to the temperature-dependent
anisotropy values.
Fusion measurements: Vesicle fusion was monitored by resonance energy
[
18]
transfer (RET) fusion assay as described by Struck et al. Vesicles were
formed both with and without 0.6 mol% each of N-NBD-PE and N-Rh-PE
�
1
and suspended at the concentration of 10 mgmL in buffer. Labeled and
unlabeled vesicles were mixed at a 1:1 ratio and diluted to a final lipid
concentration of 50mm in the above buffer. Fluorescence measurements
were performed at 288C with a Hitachi F-4500 spectrofluorimeter at an
excitation and emission wavelength of 475 and 580 nm, respectively. Fusion
was initiated by the addition of 10 mL Na SO solution (final concentration
2
4
in the cuvette was 10mm). The extent of fusion was determined by
subjecting a one-to-one mixture of labeled and unlabeled vesicles to three
[
20]
cycles of freeze ± thawing. This treatment has been shown to be adequate
for obtaining complete lipid mixing. Fused vesicles were diluted to 50mm,
and the fluorescence level of this suspension was set to 100% (Fmin). The
time course for each lipid mixing was normalized by subtracting the initial
fluorescence (F ) and dividing by the fluorescence achieved by infinite
0
probe dilution (Fmin). The percentage change in fluorescence was calculated
Dynamic light scattering (DLS): Light scattering measurements were
measured at 258C on filtered vesicular solutions (Millex units, Pore size:
0.25 mm). Vesicles (1mm) were prepared under identical conditions for all
the lipids [bath sonication (Julabo USR3) for 10 min, at 708C and at
‡
35 KHz]. A submicron particle-size analyzer employed an incident Kr
laser beam (647 nm) keeping the detector angle at 908. An interfaced
autocorrelator was used to generate the full autocorrelation of the
scattered intensity. The time-correlated function was analyzed by a method
of cumulants and the calculations imposed virtually unimodal distribution
of particle size populations.
as shown in Equation (8).
%DF/DFmin ˆ 100(F � F)/(F � F
0
0
min
(8)
X-Ray diffraction and molecular modeling studies: X-ray diffraction and
molecular modeling studies were carried out as described previously.
1H NMR spectrsocopy
[
10a]
3
Vesicle preparation: CHCl solutions of biscationic lipids with or without
Electron paramagnetic resonance studies
Sample preparation: The spin probes, doxylstearic acids with the nitroxide
group, that is, 5-(4,4-dimethyl-3-oxazolidinyl)oxyoctadecanoic acid, 5NS;
cholesterol were evaporated to dryness. The resulting lipid films were then
hydrated in D
O for ꢀ1 h at room temperature, vortexed thoroughly, and
subjected to several freeze ± thaw cycles (above the T value of the
2
m
2
346
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2346 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 8

Lipids
2335±2347
corresponding lipid) keeping the vial protected from extraneous moisture
during the preparation. The stable MLVs thus produced were rapidly
transferred into NMR tubes and immediately closed under dry N .
2
Measurements: 1H NMR spectra for each of these suspensions were
recorded at 500 MHz on a Bruker DRX-500 Fourier transform Spectrom-
eter equipped with a silicon graphics work station. The chemical shift due
to the HDO peak was taken as the internal standard. The typical spectral
conditions were as follows: pulse width 6.0 ms (flip angle 908), sweep width
[10] a) S. Bhattacharya, S. De, M. Subramanian, J. Org. Chem. 1998, 63,
7640; b) S. Bhattacharya, K. Snehalatha, S. K. George, J. Org. Chem.
1998, 63, 27; c) S. Bhattacharya, K. Snehalatha, Langmuir 1997, 13,
378; d) S. Bhattacharya, K. Snehalatha, J. Org. Chem. 1997, 62, 2198;
e) S. Bhattacharya, S. S. Mandal, Biochim. Biophys. Acta 1997, 1323,
2
2
9; f) S. Bhattacharya, S. Haldar, Biochim. Biophys. Acta 1996, 1283,
1; g) S. Bhattacharya, S. De, Chem. Commun. 1996, 1283; h) S.
Ghosh, K. R. K. Easwaran, S. Bhattacharya, Tetrahedron Lett. 1996,
7, 5769; i) S. Bhattacharya, K. Snehalatha, Langmuir 1995, 11, 4653;
3
6
0
009.62 Hz, data points 16 K, acquisition time 1.363 s, FID resolution of
.3668 Hz/point, and the number of scans 270. H NMR experiments were
1
j) S. Bhattacharya, S. Haldar, Langmuir 1995, 11, 4748; k) S.
Bhattacharya, S. De, J. Chem. Soc. Chem. Commun. 1995, 651.
carried out with vesicular mixtures of 1, 2a, and 2g containing 10, 20, and
1
[11] a) S. Bhattacharya, S. S. Mandal, Biochemistry 1998, 37, 7764; b) J.-P.
Vigneron, N. Oudrhiri, M. Fauquet, L. Vergely, J.-C. Bradley, M.
Basseville, P. Lehn, J.-M. Lehn, Proc. Natl. Acad. Sci. USA 1996, 93,
3
0 mol% cholesterol. For comparison, H NMR spectra were also recorded
in the absence of cholesterol for each of the lipid vesicles. For each
experiment, the lipid concentration was maintained at a constant 10mm.
NMR spectra were recorded above the T of the vesicular systems.
m
9
682.
[
12] In a preliminary communication, we described some properties of
these dimeric lipids, see: S. Bhattacharya, S. De, S. K. George, Chem.
Commun. 1997, 2287.
Acknowledgments
[13] a) J. M. Delfino, S. L. Schreiber, F. M. Richards, J. Am. Chem. Soc.
1
993, 115, 3458. b) D. H. Thompson, K. F. Wong, R. Humphry-Baker,
This work was supported by the Swarnajayanti Fellowship grant of the
Department of Science and Technology. We are grateful to SERC for giving
access to the computational facilities.
J. J. Wheeler, J.-M. Kim, S. B. Rananavare, J. Am. Chem. Soc. 1992,
114, 9035; c) R. A. Moss, J.-M. Li, J. Am. Chem. Soc. 1992, 114, 9227;
d) E. A. Runquist, G. M. Helmkamp, Jr., Biochim. Biophys. Acta
1
988, 940, 10; e) J.-M. Kim, D. H. Thompson, Langmuir 1992, 8, 637.
[
[
[
14] a) T. Kunitake, Y. Okahata, S. Yasunami, Chem. Lett. 1981, 1397; b) H.
Fukuda, K. Kawata, H. Okuda, S. L. Regen, J. Am. Chem. Soc. 1990,
[
1] a) K. M. Merz, Jr., B. Roux, Biological Membranes. A Molecular
Perspective from Computation and Experiment, Birkhäuser, Boston,
1
12, 1635.
15] a) S. Karaborni, K. Esselink, P. A. J. Hilbers, B. Smit, J. Karthauser,
N. M. van Os, R. Zana, Science 1994, 266, 254; b) D. Damino, Y.
Talmon, R. Zana, Langmuir 1995, 11, 1448.
16] a) R. Bartucci, T. Pali, D. Marsh, Biochemistry 1993, 32, 274; b) J. M.
Boggs, J. T. Mason, Biochim. Biophys. Acta 1986, 863, 231.
17] J. M. Boggs, G. Rangaraj, Biochim. Biophys. Acta 1985, 816, 221.
18] D. K. Struck, D. Hoekstra, R. E. Pagano, Biochemistry 1981, 20, 4093.
19] N. Düzgüne sË , T. M. Allen, J. Fedor, D. Papahadjopoulos, Biochemistry
1
996; b) J. L. Slater, C.-H. Huang, in The Structure of Biological
Membranes, (Ed.: P. Yeagle), CRC, Boca Raton, FL, 1992, pp. 175 ±
10; c) D. Marsh, CRC Handbook of Lipid Bilayers, CRC, Boca
2
Raton, FL, 1990.
[
2] a) J. B. F. N. Engberts, D. Hoekstra, Biochim. Biophys. Acta 1995,
[
[
[
1
241, 323; b) O. Murillo, S. Watanabe, A. Nakano, G. W. Gokel,
Biochim. Biophys. Acta 1995, 117, 7665; c) T. Kunitake, Angew. Chem.
992, 104, 692; Angew. Chem. Int. Ed. Engl. 1992, 31, 709; d) H.
1
1
987, 26, 8435.
Ringsdorf, B. Schlarb, J. Venzmer, Angew. Chem. 1988, 100, 117;
Angew. Chem. Int. Ed. Engl. 1988, 27, 113; e) J. H. Fendler, Science
[
[
[
20] A. Relini, D. Cassinadri, Q. Fan, A. Gulik, Z. Mirghani, M. De Rosa,
A. Gliozzi, Biophys. J. 1996, 71, 1789.
21] M. G. L. Elferink, J. van Breemen, W. N. Konings, A. J. M. Driessen, J.
Wilschut, Chem. Phys. Lipids 1997, 88, 37.
22] K. Block, in Biochemistry of Lipids and Membranes, (Eds.: D. E.
Vance, J. E. Vance), The Benzamin/Cummins, Menlo Park, CA, 1985,
pp. 1 ± 24.
23] a) P. L. Yeagle, Biochim. Biophys. Acta 1985, 822, 267; b) W. K.
Subczynski, A. Wisniewska, J.-J. Yin, J. S. Hyde, A. Kusumi, Bio-
chemistry 1994, 33, 7670.
1984, 223, 888.
[
3] a) N. A. J. M. Sommerdijk, T. H. L. Hoeks, M. Synak, M. C. Feiters,
R. J. M. Nolte, B. Zwanenburg, J. Am. Chem. Soc. 1997, 119, 4338;
b) S. J. Vigmond, T. Dewa, S. L. Regen, J. Am. Chem. Soc. 1995, 117,
7838; c) P. Klotz, L. A. Cuccia, N. Mohamed, G. Just, R. B. Lennox, J.
Chem. Soc. Chem. Commun. 1994, 2043; d) J. A. Killian, M. C.
Koorengevel, J. A. Bouwstra, G. Gooris, W. Dowhan, B. de Kruijff,
Biochim. Biophys. Acta 1994, 1189, 225.
[
[
[
[
[
4] a) W. Hübner, H. H. Mantsch, M. Kates, Biochim. Biophys. Acta
24] a) R. B. Gennis, in Biochemistry: Molecular Structures and Functions,
Springer, New York, 1989, chapter 8; b) M. B. Sankaran, T. E.
Thompson, Biochemistry 1990, 29, 10676.
1
1
991, 1066, 166; b) S. M. Grunner, M. K. Jain, Biochim. Biophys. Acta
985, 818, 352; Other examples of dimeric lipids are also known, see
for example: M. DeRosa, A. Gambacorta, Prog. Lipid Res. 1988, 27,
53.
25] M. C. Block, L. L. M. van Deenen, J. de Gier, Biochim. Biophys. Acta
1
1
977, 464, 509.
5] a) Molecular Mechanisms of Membrane Fusion, (Eds.: S. Okhi, D.
[
[
26] N. Chatterjee, H. Brockerhoff, Biochim. Biophys. Acta 1978, 511, 116.
27] a) P. Almeida, W. Vaz, T. E. Thompson, Biochemistry 1992, 31, 6739;
b) M. B. Sankaran, T. E. Thompson, Proc. Natl. Acad. Sci. USA 1991,
Doyle, T. D. Flanagan, S. W. Hui, E. Mayhew), Plenum, New York,
1988; b) T. Ohnishi, T. Ito, Biochemistry 1974, 13, 881.
[
[
[
6] O. Lockhoff, Angew. Chem. 1991, 103, 1639; Angew. Chem. Int. Ed.
8
8, 8686; c) C. Mateo, A. Reyes, A. U. Acuna, J.-C. Bronchon,
Engl. 1991, 30, 1611, and references therein.
Biophys. J. 1995, 68, 978; d) S. Scarlata, Biophys. Chem. 1997, 69, 9.
28] M. Kates, T. H. Chan, N. Z. Stanacev, Biochemistry 1963, 2, 394.
29] R. A. Moss, S. Ganguli, Y. Okumura, T. Fujita, J. Am. Chem. Soc.
7] a) F. M. Menger, C. A. Littau, J. Am. Chem. Soc. 1991, 113, 1451; b) R.
Zana, Y. Talmon, Nature 1993, 362, 228.
8] a) S. De, V. K. Aswal, P. S. Goyal, S. Bhattacharya, J. Phys. Chem.
[
[
1
990, 112, 6391.
1996, 100, 11664; b) S. De, V. K. Aswal, P. S. Goyal, S. Bhattacharya, J.
[
30] a) J. M. Boggs, M. A. Moscarello, J. Membrane Biol. 1978, 39, 75; b) S.
Eletr, A. D. Keith, Proc. Natl. Acad. Sci. USA 1972, 69, 1353; c) J.
Seelig, J. Am. Chem. Soc. 1970, 92, 3881.
Phys. Chem. B 1997, 101, 5639; c) S. De, V. K. Aswal, P. S. Goyal, S.
Bhattacharya, J. Phys. Chem. B 1998, 102, 6152.
[
9] a) R. Oda, I. Huc, S. J. Candau, Chem. Commun. 1997, 2105; b) F. L.
Duivenvoorde, M. C. Feiters, S. J. van der Gaast, J. B. F. N. Engberts,
Langmuir 1997, 13, 3737; c) R. Zana, Curr. Opin. Colloid Interface Sci.
1996, 1, 566; d) M. J. Rosen, CHEMTECH 1993, 30.
Received: November 3, 1998 [F1424]
Chem. Eur. J. 1999, 5, No. 8
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0508-2347 $ 17.50+.50/0
2347