Synthesis of a highly hydrophobic cationic lipid and structural and thermodynamic studies for interaction with DNA

Article abstract of DOI:10.1246/bcsj.20100072

Synthetic DNA-transfection reagents can overcome safety issues raised by use of viral DNA vectors. One of these candidates is a cationic lipid that can form a supramolecular complex with DNA. We have been working in a series of aromatic diamine lipids wit

Full text of DOI:10.1246/bcsj.20100072

1010 Bull. Chem. Soc. Jpn. Vol. 83, No. 9, 1010–1018 (2010)
© 2010 The Chemical Society of Japan
Synthesis of a Highly Hydrophobic Cationic Lipid and Structural
and Thermodynamic Studies for Interaction with DNA
Tomoki Nishimura,1 Takeshi Cho,1 Andrew M. Kelley,2 Magdalena E. Powell,2 John S. Fossey,3
Steven D. Bull,2 Tony D. James,2 Hiroyasu Masunaga,4 Isamu Akiba,1 and Kazuo Sakurai*1
1Department of Chemistry and Biochemistry, The University of Kitakyushu,
1-1 Hibikino, Wakamatsu-ku, Kitakyushu 808-0135
2Department of Chemistry, The University of Bath, Claverton Down, Bath, BA1 7AY, UK
3School of Chemistry, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK
4SPring-8 Japan Synchrotron Radiation Research Institute (JASRI), 323-3 Mihara, Mikazuki, Sayo-gun, Hyogo 679-5198
Received March 12, 2010; E-mail: sakurai@env.kitakyu-u.ac.jp
Synthetic DNA-transfection reagents can overcome safety issues raised by use of viral DNA vectors. One of these
candidates is a cationic lipid that can form a supramolecular complex with DNA. We have been working in a series of
aromatic diamine lipids with different tail length from C6 to C18: [N-(3,5-dialkylbenzyl)ethane-1,2-diamine, denoted
DA] as such a lipid. The present paper describes the synthesis and the fundamental properties of DA. SAXS from DA
solution showed bilayer vesicle formation, while it showed lamellar formation after the complexation with DNA. When
we measured the N/P ratio (molar ratio of the amine groups (N) in DA to phosphate groups (P) in the DNA) dependence
of SAXS, the lamellar peak (spacing = 5.0 nm) increased proportionally to the added DNA concentration at N/P ² 8. On
the other hand, in the range of N/P < 8, the spacing was increased to 5.5 nm and the area decreased as DNA increased.
These different features between N/P ² 8 and N/P < 8 suggest that the lamellar supramolecular structures differ
according to the composition.
Complexes made from DNA and cationic lipids have
attracted enormous scientific and technical interest since the
early 1980s, and complexes of liposomes have become known
as “lipoplexes.”1 Interactions between oppositely charged DNA
and cationic lipids condense DNA into a more compact form as
opposed to its highly extended conformation in solutions. This
process has been thought to take place through three steps:
(1) An electrostatic binding between the cationic headgroups
and the negatively charged phosphate groups of DNA, (2)
endothermic dehydration of the water bound to DNA, and (3) at
higher lipid-to-DNA ratios, condensation results in the for-
mation of globular structures. Considerable interest has been
generated in lipoplexes since Felgner et al.2 first found that
certain complexes formed from DNA and synthetic cationic
lipids could be efficient vehicles for DNA or RNA delivery
into a wide variety of eukaryotic cells. After almost 20 years
since the first invention, the mechanism of transfection is
still not well understood, and further investigation is necessary
to improve the transfection efficiency for the adaptation of
lipoplexes to clinical and practical use.1
into lipid bilayers (L¡C), (b) an inverted hexagonal phase with
DNA encapsulated within monolayers tubes, arranged in a
two-dimensional hexagonal lattice (HIIC), and (c) the cationic
lipids form rod-like micelles arranged in a hexagonal lattice
with DNA inserted within the interstices with a honeycomb
C
symmetry (HIC). It was shown that HII exhibited the highest
transfection and its superior efficiency was attributed to the
unstable nature of the encapsulated DNA and bound lipids.7
Recently, Marty et al.8 studied the same system as Safinya
et al.3,5 (i.e., DNA/DOTAP) with FTIR and found that there
are molecular contacts between the lipid aliphatic tails and
DNA. This result cannot be explained in the framework of any
C
structure of HIC, HIIC, and L¡
.
Independently from those studies, we reported that aromatic
amine and amidine derivatives can be used as transfection
reagents with better efficiency and less toxicity than commer-
cial products.9-11 A liposome made from an amidine derivative
is now in clinical trials for protein delivery.12 We reported that
there could be another hexagonal phase where the DNA/
cationic ion pairs are included inside of the cylinder and
the cylinder surface is covered with the remaining unbound
lipids (we denoted “hexagonally packed bilayer DNA-inclu-
It is generally believed that lipoplexes can facilitate the
ingestion of DNA in the endocytosis pathway and subsequently
transport the DNA from endosomal to cytosol vesicles. Some
studies have suggested that the lipoplex structure is the major
critical factor in determining the transfection efficiency. Small-
angle X-ray scattering (SAXS) studies of lipoplexes by Safinya
et al.3-7 demonstrated that lipoplexes take three predominant
phases: (a) A multi-lamellar phase where DNA is intercalated
C
sion cylinders,” i.e., Hb ).13 Our data showed that the aromatic
C
amine and amidine derivatives did not take HII at any
composition and HIC gave better transfection efficiency than
C
Hb . Recently, Koynova et al.14-16 synthesized more than 20
cationic phosphatidylcholine derivatives with different tail
lengths and showed that an intracellular phase transition occurs
Published on the web August 21, 2010; doi:10.1246/bcsj.20100072

T. Nishimura et al.
Bull. Chem. Soc. Jpn. Vol. 83, No. 9 (2010) 1011
OH
1
O
K2CO3
O
O
+
Br
OH
O
O
DMF, 80°C,18h
H
H
O
2
BocHNC2H4NH2, NaBH(OAc)3
1,2-Dichloroethane
H
1
N
BocN
H
O
Trifluoroacetic acid/CH2Cl2
18h
H
2
N
O
H N
2
DA
Figure 1. Synthetic scheme of the new aromatic diamine lipid for gene delivery disclosed in a preceding patent.9
when a certain lipoplex meets a cellular vesicle bilayers, the
ability to induce this phase transition relates to the superior
dry DMF (25 mL) for 18 h under an argon atmosphere. The
reaction mixture was cooled to rt, water (25 mL) was added and
the aqueous phase extracted with CH2Cl2. The product was
purified by flash chromatography over silica gel using CH2Cl2
as eluent, 1 was obtained after evaporation of the solvent
(1.83 g, 85%). 1H NMR (500 MHz, CDCl3): ¤ 9.89 (1H, s,
CHO), 6.98 (2H, s, ArH), 6.70 (1H, s, ArH), 3.98 (4H, t, J =
6.3 Hz, 2 © OCH2), 1.79 (4H, q, J = 6.9 Hz, 2 © OCH2CH2),
1.48-1.42 (4H, m, 2 © OCH2CH2), 1.34-1.22 (44H, br s,
2 © C11H22), 0.88 (6H, t, J = 6.5 Hz, 2 © CH3).
C
performance of transfection. They showed that L¡ can give
C
better transfection than others in the case that L¡ can be
transformed into cubic phase when it is mixed with cellular
membranes. Although different cationic lipids may lead to
different conclusions, it seems that the principle to relate
transfection and lipoplex structures has not been established.
Isothermal titrations had previously revealed that lipoplex
formation is a cooperative processes, mainly driven by an
entropy increase and opposed by a relatively small endothermic
enthalpy.17-19 These workers also found a large negative heat
capacity change, indicating contribution from hydrophobic
interactions between the alkyl tails and counter ion release from
DNA upon binding lipids. These thermodynamic studies can
be interpreted by a lateral interaction of the alkyl tails: the
cationic headgroups are localized at the DNA phosphates, while
hydrophobic tails would be associated with each other and lay
down on the DNA surface to exclude water.17,18 This structure
seems to contradict that from SAXS, where the alkyl tails stand
perpendicular to the DNA surface and associate through van der
Waals forces. One cause of complexity in studying lipoplexes is
that most of the cationic lipids need co-lipid to enhance their
biocompatibility or to form stable micelles in aqueous solu-
tions, most importantly, to increase transfection efficiency.
Generally, DOPE (dioleoylphosphatidylethanolamine) is added
as well as other co-lipids such as DLPC and DOPC.
We found that N-(3,5-dialkylbenzyl)ethane-1,2-diamine
(DA, Figure 1) derivatives themselves can form a stable
vesicle in water and addition of a small amount of DOPE led
to a drastic increase in transfection efficiency.10 In order to
understand the molecular mechanism for the transfection of the
DA/DOPE system, we decided to take advantage of the good
water-compatibility of DA to study interactions of DA vesicle
and DNA. This paper presents data and analysis relating to the
binding of various DAs with different tail length (C6-C18)
to DNA based on several physical measurements including
synchrotron SAXS and isothermal titration calorimetry.
Sodium triacetoxyhydroborate (1.23 g, 5.78 mmol) was
slowly added to a mixture of 1 (1.82 g, 3.85 mmol) and N-
(tert-butoxycarbonyl)ethane-1,2-diamine (0.62 g, 3.85 mmol)
in 1,2-dichloroethane (10 mL). The resultant solution was
stirred at rt for 24 h under an argon atmosphere. The reaction
was then quenched with saturated NaHCO3 solution and
stirred for 30 min, extracted with CH2Cl2, dried over MgSO4,
and purified by flash chromatography over silica gel using
CH2Cl2:methanol = 9:1 as eluant. Compound 2 was obtained
as a pale yellow oil after evaporation of solvent (1.29 g, 54%).
1H NMR (500 MHz, CDCl3): ¤ 6.46 (2H, s, ArH), 6.38 (1H, s,
ArH), 6.06 (1H, br s, NHCO2), 3.91 (4H, t, J = 6.5 Hz, 2 ©
OCH2 and 2H, s, CH2NHCO2), 3.44 (2H, s, ArCH2), 2.95 (2H,
s, NHCH2CH2NHCO2), 1.98 (4H, m, 2 © OCH2CH2), 1.45-
1.41 (13H, br s, OC(CH3)3 and 2 © OCH2CH2CH2), 1.30-1.25
(44H, br s, 2 © C11H22), 0.88 (6H, t, J = 6.9 Hz, 2 © CH3).
Compound 2 (1.29 g, 2.08 mmol) was dissolved in dry
CH2Cl2 (2 mL), then trifluoroacetic acid (18 mL) was added at
rt and then stirred for 7 h. The resultant solution was poured
into saturated NaHCO3 solution (100 mL). The aqueous phase
was extracted by CH2Cl2 and dried over MgSO4, filtered and
the solvent was removed. After purification by flash chroma-
tography over silica gel using CH2Cl2:EtOH = 15:1 as eluant,
compound 3 (N-(3,5-didodecyloxybenzyl)ethane-1,2-diamine)
was obtained as a brown solid (denoted DA12 after its dodecyl
tails). In the same manner, DA-lipids with hexyl, octyl, nonyl,
decyl, dodecyl, pentadecyl, and octadecyl tails were synthe-
sized and denoted DA6, DA8, DA9, DA10, DA12, DA15,
1
and DA18. H NMR (500 MHz, CDCl3): ¤ 6.47 (2H, s, ArH),
Experimental
6.39 (1H, s, ArH), 5.30 (1H, br s, NH), 3.88 (4H, t, J =
6.5 Hz, 2 © OCH2 and 2H, s, CH2CH2NH2), 3.14 (2H, s,
NHCH2CH2), 3.05 (2H, s, CH2NH2), 1.73 (4H, q, J = 7.2 Hz,
2 © OCH2CH2), 1.42-1.39 (4H, m, 2 © OCH2CH2CH2), 1.29-
Synthesis of DA-Lipids. A solution of 3,5-dihydroxybenz-
aldehyde (0.5 g, 4.53 mmol), 1-bromododecane (2.81 g, 11.3
mmol), and K2CO3 (3.41 g, 22.7 mmol) was stirred at 80 °C in

1012 Bull. Chem. Soc. Jpn. Vol. 83, No. 9 (2010)
Vesicle to Lamellar Transition
1.25 (44H, m, 2 © C13H27), 0.88 (6H, t, J = 7.0 Hz, 2 © CH3).
the wavelength of the X-ray - through the relation of
q ¼ 4³ sin ª=-. SAXS experiments were performed at fixed
concentrations of the DA lipid that were titrated against
increasing pDNA concentrations.
pDNA Amplification.
Plasmid DNA coding pGL3
(bp = 5256, where bp is the number of base pair) was
amplified with E. coli competent cells and purified by using
a QIAGEN HiSpeed Plasmid Maxi kit.
Isothermal Titration Calorimetry.
Heat flow during
Preparation of Cationic Assemblies Formed from the
Synthetic Lipids and Lipoplex. The cationic assemblies and
lipoplex solutions were prepared as follows: A DA-lipid
chloroform solution was poured into a 1 mL vial and dried in
vacuo for 1 h, then an aqueous solution of an appropriate
amount of NaCl was added to the vial to hydrate the lipid,
the solution was then sonicated for 10 min with a UH-50
(SMT Co., Japan). The lipoplexes made from DA-lipid/DNA
were prepared by mixing the DA lipid solution with an
appropriate amount of sonicated salmon sperm DNA (the
average number of base pair: bp was 3000) or a plasmid DNA
(pGL3, bp = 5256) solution. Most experiments were carried
out using plasmid DNA except calorimetry because it is
homogeneous in size and data were less noisy. The DA-lipid/
DNA mixing ratio was expressed in terms of the nitrogen to
phosphate molar ratio: N/P. Here, it should be noted that the
composition of the DA-lipid/DNA lipoplex is not always equal
to N/P ratio because since free DNA or/and free lipid may
exist in the mixture.
isothermal titration was measured with a VP-ITC MicroCal
microcalorimeter (Northampton, MA). Titration data were
processed with the software provided by the manufacturer.
50 ¯M bp (0.20 mg mL¹1) of the salmon DNA solution was
maintained in the sample cell (1.401 mL) at 25 °C in 50 mM
NaCl solution and the whole cell was stirred at 300 rpm. In
each experiment 30 aliquots (7 ¯L each) of a stock solution
of DA (2.0 mM) were automatically injected into a sample
cell initially filled with DNA solution. Injections were 14 s
in duration, and individual injections were programmed at
intervals of 300 s. The injection intervals as well as the volume
and the number of injections have been adjusted such that the
binding is completed toward the end of the titration. After
the peak area was integrated and the heats of dilution were
subtracted, the thermogram for the binding was obtained.
Other Characterization. The ¦-potential and the average
hydrodynamic radius (Rh) were measured with a Malvern
Zetasizer nano (Malvern, U.K.) at rt. Transmission electron
microscopy (TEM) was acquired using a JEOL TEM-3010
(accelerating voltage 200 kV). The sample was placed on 200-
mesh carbon-coated copper (Cu) grids. The TEM grid was
dried under reduced pressure for 12 h before observation.
ANS Fluorescence Studies. The fluorescence probe 8-
anilino-1-naphthalenesulfonic acid (ANS) was purchased from
TCI and used without further purification. All solutions were
prepared by diluting a stock solution (5 mM of DA, and in the
case of the lipoplex 1.0 mg mL¹1 of pDNA was added to adjust
N/P = 6.6) with distilled water containing 50 mM NaCl. The
concentration of ANS was fixed at 2.0 © 10¹5 M. The fluo-
rescence measurements were carried out with a fluorescence
spectrophotometer (Hitachi F-4500), by exciting at 385 nm and
recording the emission spectrum in the range 400-700 nm. The
Results and Discussion
N/P Dependence of ¦-Potential and Rh. It is important to
know how the net surface charge and the size are changed when
DNA is sequentially mixed with DA. The N/P dependence of
the ¦-potential and the hydrodynamic radius (Rh) for DA12 and
DA8 were investigated (Figure 2). For DA12 (triangles), when
the complex had either a negative or positive ¦-potential in the
range of N/P < 3 or N/P > 7, respectively, Rh was about
¹1
scan speed and the slit widths were 260 nm min and 5.0 nm,
respectively. From a plot of intensity vs. the lipid concen-
tration, the critical micelle (CMC) and aggregation (CAC)
concentrations were determined for the lipid and lipoplex
solutions, respectively. CAC is defined as the point at which
the polymer and surfactant start to form aggregates, in our case,
at which DNA and DA form the ion-pair complex. The Gibbs
free energy of surfactant aggregation can be calculated by the
equation ¦GCAC = ¹RT ln CAC, according to Zhu et al.,19
where R and T are the gas constant and temperature.
1500
DA8
50mM NaCl
DA12
1000
500
0
Synchrotron Small-Angle X-ray Scattering.
SAXS
measurements were carried out at BL40B2 SPring-8 with
0.71 and 0.54 m cameras equipped with a Rigaku imaging plate
(30 © 30 cm, 3000 © 3000 pixels) employing an X-ray wave-
length of 1.0 ¡. The exposure time was 5 min and the relative
X-ray intensity and the sample transmittance were determined
with ion-chambers located in front of the sample and a
photodiode behind. A bespoke chamber for SAXS measure-
ments was constructed that enables a sample to be introduced in
a vacuum and scattering measured from it as well as a vacuum-
proof cell with a 0.20 mm quartz glass plate. Combining these
two instruments, the data were then circularly averaged to
produce one-dimensional (1-D) SAXS profiles of the scattered
X-ray intensity [I(q)] versus the magnitude of the scattering
vector (q), where q is related to the scattering angle ª and
25
0
ζ
-25
-50
0
2
4
6
8
10
N/P ratio
Figure 2. The N/P dependence of the ¦-potential and the
hydrodynamic radius determined with DLS for DA8 and
DA12.

T. Nishimura et al.
Bull. Chem. Soc. Jpn. Vol. 83, No. 9 (2010) 1013
102
101
100
10-1
100 nm or less. When we added pDNA at the isoelectric point
around N/P = 5, Rh increased drastically, indicating that the
complex underwent secondary aggregation. The isoelectric
point of N/P = 5.0 indicates that about 40 mol % of DA12 are
positively charged due to protonation of the diamine head-
group. Since the pKa value of ethylenediamine in aqueous
solutions is around 1020, the relationship pH = pKa +
log([base]/[acid]) predicts that the DA12 head diamine is
completely protonated at pH 7.0, which is in contradiction with
the observed results. This discrepancy could be explained in a
similar way to the Manning effect21,22 in that clustered ion
groups tend to be more neutralized by counter ion condensation
to reduce electrostatic repulsion than free ions. The same
phenomena was reported in the case of spermine and a
spermine lipid (dioctadecylaminoglycyl spermine),23 where the
titration ended at [spermine]/[DNA] = 0.3 for spermine itself,
while more spermine lipids were needed to reach the isoelectric
point: [spermine lipid]/[DNA] = 0.5. When we used DA in
DNA delivery, N/P was adjusted to 6.6 to obtain the highest
transfection efficiency.10 Therefore, we fixed N/P = 6.6 here-
inafter unless otherwise noted. N/P dependences on the ¦-
potential for DA18 were almost the same as DA12 (see the
Supporting Information). The circular marks in Figure 2 show
the N/P dependence of the ¦-potential and Rh for DA8.
Comparing with those of DA12, the isoelectric point was
similar with the DA12 system, around N/P = 5.0; however,
there were relatively large particles in the positive potential
side. Similar behavior was also observed for DA18, which may
be related to the fact that DA18 showed less propensity to form
complexes as opposed to DA12, as presented below.
DA8
CAC
DA12
CAC
CMC
CMC
10-2
10-3
CL / mM
10-3
10-1
10-2
10-1
10-1
10-2
10-3
CMC
CAC
5
10
15
Nc
Figure 3. The DA concentration dependence of the ANS
fluorescence intensity in DA8 (top left) and DA12 (top
right), lipid (circles) and lipoplex (squares), and the DA
alkyl tail length (NC) dependence of the CMC and CAC
(bottom) determined from intersections of the ANS
fluorescence plots.
Critical Micelle or Aggregation Concentration and
Thermodynamics.
The fluorescence intensity of ANS
sensitively reflects the polarity of its environment and its
intensity can be measured in dilute solutions such as less
than 10¹5 M of ANS. Therefore ANS is a useful probe for
determination of CMC. The upper two panels of Figure 3
show examples of the ANS fluorescence intensity versus the
lipid concentration (CL) for DA8 and DA12 as well as their
lipoplexes with the pDNA at N/P = 6.6. The obtained CMC
and CAC values are summarized in Table S1 and plotted
against NC (alkyl tail length) in Figure 3. As presented in
the lower panel in Figure 3, the CAC is around 10¹2 mM for
all NC. The CMC is larger than CAC at NC = 5 and 8 and
decreases with an increase of NC and seems to merge with the
CAC at NC > 10. CAC and CMC are observed in mixtures of
polyelectrolytes and the oppositely charged ionic surfactants,
and generally CAC is lower than CMC,24-27 this relationship is
also supported by theoretical considerations.28 Although not
all of the data are presented, the observed results, including
Figure 3, are consistent with these general features for CAC
and CMC. CMC of CTAB (hexadecyltrimethylammonium)
and other cationic lipids such as CPC (hexadecylpyridinium
chloride) and BAK12 (benzyldodecyldimethylammonium
chloride) were determined to be 120 ¯M,17 60 ¯M (150 mM
NaCl in phosphate buffered saline (PBS)),19 and 1 mM
(150 mM NaCl in PBS),19 respectively. DOTAP ((2,3-dioleoyl-
propyl)trimethylammonium chloride), one of the most studied
lipids in terms of lipoplex formation, has a CMC at 70 ¯M.29
When a multi-amine group is attached as the headgroup such as
dioctadecylamidoglycyl spermine, the headgroup increases
affinity with water and thus increases CMC. As presented in
Table S1, CMC values for DA series are much lower than those
of other compounds. This lower CMC can be attributed to the
presence of the highly hydrophobic aromatic moiety of DA,
which may form more attractive ³-³ stacking than the van der
Waals interactions between alkyl chains in the vesicle. There
are a few systematic studies of the CAC for cationic lipid/DNA
systems. Zhu et al.19 reported CAC for four compounds; 30 ¯M
for CPC, 550 ¯M for BAK12, 46 ¯M for BAK14, and 550 ¯M
for CTAC. From the isothermal titration curve for spermine
lipid/DNA lipoplex, CAC can be estimated to be around
100 ¯M.23 Comparing these values with those of DA-lipo-
plexes, it is clear that DA-lipoplexes show a considerably low
CAC, ranging from about 10 ¯M. The thermograms of the
titrations of the DA solutions into the DNA solutions are
shown in the upper panel of Figure S1. The corresponding heat
enthalpy change per mole of DA normalized with the DA
concentration is presented in the lower panel of Figure S1. At
50 mM, a simple endothermic reaction was observed, indicat-
ing that the present system is driven by entropy gain. The
thermogram showed drastic decrease with increasing DA
concentration at N/P < 3, and finally the decrease seemed
to finish at around N/P = 5.0 completely. This is in good
agreement with the results obtained in the ¦-potential experi-

1014 Bull. Chem. Soc. Jpn. Vol. 83, No. 9 (2010)
Vesicle to Lamellar Transition
¹¡
ment. The isothermal curve can be fit by a single site model and
the obtained thermodynamic parameters are listed in Table S2.
At 5 mM, the thermal response suggests that DA/DNA binding
proceeds in a cooperative manner, the exothermal initial
reaction leads to the subsequent endothermic reaction. Zhu
and Evans19 and Matulis et al.17 observed a similar result and
interpreted that there is no interaction below the CAC, but once
DNA and lipid concentrations reach it, binding takes place. In
our ITC experiments, we injected the 7 ¯L DA12 solution with
CL = 2.0 mM to a 1.401 mL DNA solution. Since the CMC for
DA12 is 8 ¯M, the DA12 molecules in the ITC syringe had
already formed the vesicle, and even at the first injection CL
in the sample cell should be beyond the CMC and CAC.
Therefore, the thermodynamic parameters obtained here should
be related to the interaction between the vesicle and DNA.
With increasing salt concentration, the first exothermic binding
mode disappeared and the second endothermic mode became
dominate. This confirms that the first binding mode is due to
electrostatic interactions between the anionic phosphate groups
and cationic amine groups and the second one is due to the
hydrophobic interactions that mainly cause the complexation.
We analyzed the isothermal curve in terms of the two step
sequential model.30 Table S1 shows the ¦GCAC values deter-
mined from ¦GCAC = ¹RT ln CAC.19 The electrostatic con-
q seemed to follow the I(q)-q relationship. Here, ¡ is about
2.6 at DA6, it decreased with an increase of NC and reached
¡ = 2.0 at DA15, except for DA18. DA18 showed an ¡ = 2.5,
which may be a result of secondary aggregation. The presence
of the aggregation was also supported by DLS for DA18 and
thus we decided not to take account of the low q data for DA18.
There was characteristic amplitude: Two minima at q = 0.6-1.0
and 3-5 nm and one maximum around q = 1-2 nm¹1, and
their positions shifted toward lower q with an increase of NC.
For DA12 and DA15, a second maximum at q = 3.5 nm
becomes appreciable, DA18 showed even a third maximum
at q = 6 nm¹1. The presence of the higher maxima suggested
that the local structure of these cationic assemblies are well
defined and have a rather narrow distribution in size. One
¹1
¹1
¹1
might suppose that we should be able to observe the I(q)-q
relationship at low q, as a characteristic relationship for the
form factor of bilayer vesicles. This relationship is generally
observed with neutron scattering in the range of q µ 10¹3 nm
¹1
or less.32 To confirm vesicle formation, TEM observation for
freeze-dried DA12 solution was performed, a typical image is
presented in Figure 4B. The images seem to be torus shaped
objects. We presume that these objects represent the vesicle
made from DA12. To analyze SAXS from DA-lipid solutions,
we supposed a bilayer vesicle model as presented in Figure 4C,
where the bilayer consists of two shells made from the
headgroups with a thickness of th and the core made of the
alkyl tails with a thickness of tc and the diameter of the vesicle
is too large to give characteristic q dependence. The bilayer
form factor for such a model can be expressed as below.32-34
tribution of monoamines is approximately estimated to be
¹1 17
¹2.6 kJ mol
,
which is almost equal to the observed first
interaction. Referencing the ITC measurements for a series of
alkylamines with long aliphatic chains, reveals the entropy gain
for the aggregate formation due to the hydrophobic interaction
between the alkyl chains is T¦S = 5-10 kJ mol¹1. ITC studies
for DNA binding with CPC, BAK12,19 and alkyl amines17
showed that T¦S = 5-20 kJ mol¹1. These values are much
smaller than those of the DA12 system, which reflect the
fact that DA molecules consist of a highly hydrophobic
aromatic moiety and the alkyl chains, although the nature of
the interaction is completely different in the CAC and ITC
experiments. At CAC, the DA molecules dissolved in water
start to assemble triggered with the ion-pair formation with
DNA, while ITC measured the DA-vesicle/DNA interactions.
Nevertheless, the obtained free energies almost coincide with
each other (see Table S1 and Table S2). This agreement
suggests that the molecular origin of the entropy gain in both
cases may be identical. It is difficult to uncover the origin of
the large entropy gain upon complex formation. It is widely
accepted that the major contribution in ITC comes from
counter ion release, according to Wagner et al.31 Additionally,
dehydration may have also contributed to the positive change
in entropy. With all these factors, one might suggest that the
cationic group of DA lipid binds phosphate sites, while the
DA hydrophobic group may have to lay down on the DNA.
Therefore, the observed T¦S is seemingly much larger than
other cases. The large entropy gain and the agreement of ¦G
from CAC and ITC suggest that the alkyl chain and aromatic
moiety may also be involved in the complex formation.
ꢀ
ꢁ
ꢂ
ꢁ
ꢂꢃ
2
4³
q4
tc
2th þ tc
IðqÞ ¼
ðµc ꢀ µhÞ sin q
þ ðµh ꢀ µsÞ sin q
2
2
ð1Þ
Here, µc, µs, and µh are the electron densities for the
hydrophobic core, solvent, and hydrophilic shell, respectively.
By fitting the data with eq 1, we evaluated the best combination
of µh, th, and tc, values so as to reproduce the data points under
¹3
the condition of µs = 334 e nm and µc = 277 e nm¹3. Here,
µs was calculated from the NaCl concentrations and the
electron density of water and µc was assumed to take a value
that had been reported for the alkyl domains for vesicle.35 Since
all DA molecules, regardless of the alkyl tail length, should
have the same shell structure due to the identical headgroup, we
decided to use the same values for th and µh for all of the fitting.
The most reliable values of µs, th, and tc were evaluated, the
resultant values are summarized in Table 1. The best-fit curves
are compared with the data as a solid line in Figure 4A.
Agreement between the data and calculation is considerably
good so as to conclude that DA forms a bilayer vesicle with
the obtained structural parameters. From molecular modeling
(MOPAC), a full stretch length of the alkyl chain of DA (Lt)
was estimated and each value is listed in Table 1. At NC = 6,
2Lt < tc, At NC = 8, 2Lt µ tc, and NC = 9-18, 2Lt > tc. This
fact indicates that when the tail is short, the methyl tail-tips do
not enter the opposite side tail layer, yet with increasing tail
length, they become intercalated with the opposite side or pack
in a tilted tail-to-tail fashion.
Small-Angle X-ray Scattering from DA-Lipids. Figure 4
presents SAXS profiles from DA-lipids in 50 mM NaCl
aqueous solution. We did not observe any Guinier region,
suggesting that the scattering objects were too large to provide
the overall size in the q range investigated. The intensity at low
Small-Angle X-ray Scattering from DA-Lipoplexes.
Figure 5 presents the SAXS profiles when DNA was added

T. Nishimura et al.
Bull. Chem. Soc. Jpn. Vol. 83, No. 9 (2010) 1015
1010
A
B
109
108
107
106
105
104
103
102
DA6
DA8
DA9
50nm
DA10
DA12
C
th
tc
DA15
DA18
ρh
101
100
ρ
ρcs
2
3 4 56789
2
3 4 5678
1
Distance
q / nm-1
Figure 4. SAXS profiles from DA lipid (A), a typical TEM image for DA12 lipid solution after freeze-drying, depicting vesicles
(B), and a schematic of the vesicle model used in the fitting (C).
Table 1. Structural Parameters Obtained from SAXS for the
DA Vesicles and Lipoplexes
1010
Lamellar spacing (d)/nm
Sample code tc/nma),c) Lt/nmb)
at N/P = 3.0c)
109
DA6
DA8
DA9
DA10
DA12
DA15
DA18
1.80
1.82
1.92
2.13
2.35
2.63
2.91
0.75
1.01
1.14
1.25
1.51
1.89
2.28
4.82
4.87
5.06
5.16
5.44
6.17
6.41
DA6
DA8
108
107
106
105
104
103
102
101
100
DA9
a) When we analyzed the data with bilayer vesicle model,
we found that the combination of th = 0.30 nm and µh =
DA10
¹3
575 e nm can give the best fitted curves for all DA samples.
¹3
¹3
We fixed µc = 277 e nm and µs = 334 e nm based on the
chemical compositions and densities. Therefore, only tc is the
adjustable parameter for the fitting. b) Full length of the alkyl
tail calculated with a MOPAC program in Chem 3D. c) The
experimental error range are tc = «0.01 nm, lamellar spacing
(d) = «0.01 nm.
DA12
DA15
DA18
to DA (for DA6-DA18) at N/P = 3.0. All of the samples
showed lamellar diffraction patterns, the lamellar distances
(d) were determined from the peak positions, summarized in
Table 1 and plotted against NC in Figure 6, showing that the
lamellar distance was increased with an increase of NC. When
the alkyl chain of the DA lipid is longer, attenuation of the
second-order peak and broadening of the first-order peak
indicate reduced regularity. Figures 7A and 7B present the N/P
2.53.0 3.5
0.5 1.0 1.5 2.0
q / nm-1
Figure 5. SAXS profiles from lipoplex solutions measured
at SPring-8 BL40B2 with a 0.54 or 0.71 m camera at
an X-ray wavelength of 1.0 ¡: For the DA lipid/pDNA
lipoplexes, N/P = 3.0.

1016 Bull. Chem. Soc. Jpn. Vol. 83, No. 9 (2010)
Vesicle to Lamellar Transition
dependence of SAXS for DA12. When we added DNA at
N/P = 500 (in the molar ratio, DA12:base = 250:1), the form
factor from the DA12 vesicle became completely invisible,
and converted into diffraction from a lamella structure with
the spacing d = 5.02 nm. Further addition of DNA increased
the peak intensity in the range of N/P > 8. When DNA was
added at N/P < 8, the spacing slightly increased and the peak
intensity decreased upon addition of further DNA. Figure 8A
plots d against N/P and Figure 8B plots the peak area against
DNA concentration (where we chose DNA concentration
instead of N/P ratio, because the peak intensity is proportional
to the DNA concentration at N/P < 8 owing to the fixed lipid
concentration.). The peak area increased proportionally with
DNA concentration at N/P ² 8 (0.4 mM) and decreased at
N/P < 8, except for N/P = 4. In Figure 7A, there was an
upturn at low q region at N/P = 2-4. This upturn can be
ascribed to aggregation. At N/P = 4, there was a large amount
of precipitation in the cell because of the isoelectric point
(Figure 2) and thus the concentration at the position of X-ray
irradiation will be decreased, which explains the low peak
intensity and thus deviation from the proportional line in
Figure 8B. The N/P ratio dependences of d and the peak
intensity in Figure 8 indicate that different mechanisms govern
the formation of the lamellar at N/P > 8 and N/P < 8.
Additionally, we found that d did not depend on the ionic
strength (5-200 mM NaCl) and the difference between
d = 5.01 and 5.44 nm at the two regions was confirmed to be
beyond experimental error (the data are not presented). When
we convert the ratio of N/P = 500 to the number of pDNA to
DA, we find one pDNA to 1.31 © 106 DA molecules. Suppose
the size of the vesicle is about 100 nm, one vesicle would
consist of about 104,5 DA molecules, which leads a ratio of one
pDNA to approximately ten vesicles at N/P = 500. There are
several other novel features observed in Figures 7 and 8 that
have gone unnoticed in previous studies. The original vesicle
scattering almost disappeared at N/P = 500, indicating that the
addition of a very small amount of DNA drastically deformed
the bilayer structure of the DA lipid. The size of pDNA with
bp = 5256 is about 200-500 nm, which is about 2-5 times
larger than one vesicle. The comparison in the ratio and size
suggest that one pDNA can interact with all the vesicles at an
8
d (Lamella)
6
tc+2th+DNA
4
tc+2th
2
0
tc
th
20
5
10
15
Nc
Figure 6. The alkyl tail length (NC) dependence of the inner
layer thickness (th), the total bilayer thickness (2th + tc),
the sum of 2th + tc and the DNA diameter (2.0 nm), and the
lamellar distance (d) at N/P = 3.0.
1010
108
10
1010
A
B
N/P =500
350
N/P=10
108
1
1
2
3
4
q / nm-1
8
DA
N/P =500
350
106
104
102
5
106
200
100
75
50
40
4
104
3
30
102
100
20
2
15
12.5
1
10
8
100
1.0
2.0
3.0
4.0
1.0 2.0 3.0 4.0
q / nm-1
Figure 7. The N/P dependence (from 1 to 10) of the SAXS profiles (A) and the N/P dependence (from 8 to 500) for DA12/pDNA
lipoplexes at a fixed DA12 concentration of 3.2 mM (B).

T. Nishimura et al.
Bull. Chem. Soc. Jpn. Vol. 83, No. 9 (2010) 1017
4000
3000
2000
1000
N/P = 8
6.0
5.5
5.0
4.5
A
B
( )
N/P=4
Isoelectric composition
0
1
10
N/P ratio
100
0
1
2
3
4
DNA concentration / mM (in base pair)
Figure 8. The N/P dependence of the lamellar distance (A) and the DNA concentration dependence of the first lamellar peak area
(B). The area at N/P = 4 was reduced because of a decreased solution concentration due to precipitation at the isoelectric point.
N/P = 500. Once pDNA interacts with one vesicle, this event
induces the entire vesicle to change its cross sectional structure.
Furthermore, the first peak position did not shift upon further
addition of DNA. If the number of the lamellar layers was
less than 5, the peak position would shift to lower q (for
example, two, three, and five layers satisfies the relation of
q2/q¨ = 0.917*, q3/q¨ = 0.971*, and q5/q¨ = 0.998*,
respectively. qn is the first peak position of the n layer
lamellar). Therefore, the absence of shift in the peak position
indicates that a multi-layer lamellar with at least 5 layers had
been already formed at N/P = 500. Second, the peak intensity
increasing proportionally to the DNA concentration implies
that the formation of the lamellar proceeds in a stoichiometric
manner and stoichiometric balance is achieved at one phos-
phate anion to four DA12 molecules (because the intensity
reached the maximum at N/P = 8.0). This composition is
different from the isoelectric point.
negative side in Figure 2), all of the added DA molecules
should be interacting with DNA. The only suitable model to
explain the lamellar formation in this case is the sandwich
model. This can be confirmed when tc + 2th + the diameter of
DNA (2.0 nm) is plotted as a function of NC as shown in
Figure 5. The values of tc + 2th + DNA are smaller than that of
d at N/P = 3.0, the difference between them is approximately
0.6 nm at NC = 6 and increases to approximately 1.1 nm at
NC = 18. The model for the region of N/P > 8 seems to be
rather difficult to establish. When we added a small amount
of DNA to DA-vesicle solutions, we also observed lamellar
formation. This can be explained by either the sandwich or
inclusion model, while it is difficult to draw a definitive
conclusion only from the present SAXS data.
Conclusion
We synthesized a series of aromatic diamine lipids (DA)
with alkyl tail lengths from C6 to C18 and explored the DA/
DNA complexes as a function of N/P with DLS, SAXS, and
ITC. The following facts are clarified: (1) The isoelectric point
was achieved at N/P = 5.0, (2) the CMC and CAC were
determined to be around 10 ¯M, which is very small compared
with other cationic lipids, (3) SAXS profiles from the DA lipids
were reproduced with a monolayer vesicle model, (4) SAXS
profiles from the lipoplexes showed that lamellar were formed
upon complexation, (5) When the N/P dependence of SAXS
were measured, characteristic vesicle scattering disappeared
even at N/P = 500 and the lamellar peak area reached a
maximum at N/P = 8, (6) the lamellar spacing was about
5.5 nm at N/P < 8, while it was about 5.0 nm at N/P > 8,
(7) ITC showed the DA/DNA complexation was driven by
entropy. These results at N/P < 8 can be rationalized by
assuming the sandwich model. However, two possible models
[sandwich or inclusion model] can be considered for the range
of N/P > 8.
Vesicle to Lamellar Transition. Lamellar formation upon
DNA/lipid interactions is well reported3-5,36-42 and has
attracted enormous interest since this phenomenon is related
to DNA condensation and thus important in understanding
the transfection efficiency of lipoplexes. Safinya et al. were the
first to propose a structural model of the lamellar and their
model has been widely used to analyze the lipoplex lamellar.4
According to them, DNA molecules are confined to the two-
dimensional cationic surface of the lamellar and sandwiched
C
between the lipid bilayers (denoted L¡ or DNA-sandwich
model). On the other hand, Matulis et al.17 and Marty et al.8
showed that the presence of hydrophobic interaction via lipid
aliphatic tails and hydrophobic region in DNA. According to
their model, once DNA is coated with hydrophobic groups, the
DNA/lipid complex could be transported into hydrophobic
domain (denoted DNA-inclusion model). As presented in
Figure 8, the lamellar peak (spacing = 5.01 nm) intensity
increased proportionally to the DNA concentration at N/P ²
8 (0.4 mM). In the range of N/P < 8, the spacing increased
to 5.5 nm and the intensity decreased with increasing DNA
concentration. These different features for N/P ² 8 and
N/P < 8 suggest that the lamellar supramolecular structures
differ according to the composition. At N/P < 8 (including the
This work is finically supported by the JST CREST program.
All SAXS measurements were carried out at SPring-8 BL40B2
(No. 2008B0012). The Royal Society, U.K. (International Joint
Project) are thanked for support.

1018 Bull. Chem. Soc. Jpn. Vol. 83, No. 9 (2010)
Supporting Information
Vesicle to Lamellar Transition
18 C. H. Spink, J. B. Chaires, J. Am. Chem. Soc. 1997, 119,
10920.
19 D.-M. Zhu, R. K. Evans, Langmuir 2006, 22, 3735.
20 V. S. Bryantsev, M. S. Diallo, W. A. Goddard, III, J. Phys.
Chem. A 2007, 111, 4422.
ITC for DA12, DLS and ¦-potential for DA18, Tables for
CMC, CAC data. This material is available free of charge on
the web at http://www.csj.jp/journals/bcsj/.
21 G. S. Manning, J. Chem. Phys. 1969, 51, 924.
22 G. S. Manning, J. Phys. Chem. B 2007, 111, 8554.
23 M. M. Patel, T. J. Anchordoquy, Biophys. J. 2005, 88,
2089.
24 D. Y. Chu, J. K. Thomas, J. Am. Chem. Soc. 1986, 108,
6270.
References
1
L. Huang, H. Mien-Chie, E. Wanger, Nonviral Vectors for
Gene Therapy, Academic Press, San Diego, 1999.
P. L. Felgner, T. R. Gadek, M. Holm, R. Roman, H. W.
2
Chan, M. Wenz, J. P. Northrop, G. M. Ringold, M. Danielsen,
Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 7413.
25 P. Hansson, M. Almgren, J. Phys. Chem. 1995, 99, 16694.
26 P. Hansson, M. Almgren, J. Phys. Chem. 1996, 100, 9038.
27 M. Nichifor, M. Bastos, S. Lopes, A. Lopes, J. Phys. Chem.
B 2008, 112, 15554.
3
I. Koltover, T. Salditt, J. O. Rädler, C. R. Safinya, Science
1998, 281, 78.
4
J. O. Rädler, I. Koltover, T. Salditt, C. R. Safinya, Science
28 H. Diamant, D. Andelman, Macromolecules 2000, 33,
8050.
1997, 275, 810.
5
6
C. R. Safinya, Curr. Opin. Struct. Biol. 2001, 11, 440.
A. Zidovska, H. M. Evans, K. K. Ewert, J. Quispe, B.
29 J. Dalvi-Malhotra, L. Chen, J. Phys. Chem. B 2005, 109,
3873.
Carragher, C. S. Potter, C. R. Safinya, J. Phys. Chem. B 2009, 113,
3694.
30 T. Wiseman, S. Williston, J. F. Brandts, L.-N. Lin, Anal.
Biochem. 1989, 179, 131.
7
K. Ewert, N. L. Slack, A. Ahmad, H. M. Evans, A. J. Lin,
C. E. Samuel, C. R. Safinya, Curr. Med. Chem. 2004, 11, 133.
R. Marty, C. N. N’Soukpoé-Kossi, D. Charbonneau, C. M.
31 K. Wagner, D. Harries, S. May, V. Kahl, J. O. Rädler, A.
Ben-Shaul, Langmuir 2000, 16, 303.
32 S. Ristori, J. Oberdisse, I. Grillo, A. Donati, O. Spalla,
Biophys. J. 2005, 88, 535.
33 L. Cantù, M. Corti, E. Del Favero, M. Dubois, T. N. Zemb,
J. Phys. Chem. B 1998, 102, 5737.
8
Weinert, L. Kreplak, H.-A. Tajmir-Riahi, Nucleic Acids Res. 2009,
37, 849.
9
K. Koiwai, K. Tokuhisa, R. Karinaga, Y. Kudo, S. Kusuki,
Y. Takeda, K. Sakurai, Bioconjugate Chem. 2005, 16, 1349.
10 K. Sakurai, S. Kusuki, E. Hamada, Y. Takeda, Jpn. Kokai
Tokkyo Koho 2008-143844, 2008.
34 A. Salvati, S. Ristori, J. Oberdisse, O. Spalla, G. Ricciardi,
D. Pietrangeli, M. Giustini, G. Martini, J. Phys. Chem. B 2007,
111, 10357.
11 K. Shimizu, M. Isozaki, K. Koiwai, U.S. Patent 6228391,
2001.
35 T. Kawaguchi, T. Hamanaka, Y. Kito, H. Machida, J. Phys.
Chem. 1991, 95, 3837.
12 K. Kawahara, A. Sekiguchi, E. Kiyoki, T. Ueda, K.
Shimamura, Y. Kurosaki, S. Miyaoka, H. Okabe, M. Miyajima,
J. Kimura, Chem. Pharm. Bull. 2003, 51, 336.
13 M. Sakuragi, S. Kusuki, E. Hamada, H. Masunaga, H.
Ogawa, I. Akiba, K. Sakurai, J. Phys.: Conf. Ser. 2009, 184,
012008.
36 K. K. Ewert, H. M. Evans, A. Zidovska, N. F. Bouxsein, A.
Ahmad, C. R. Safinya, J. Am. Chem. Soc. 2006, 128, 3998.
37 T. Kawashima, A. Sasaki, S. Sasaki, Biomacromolecules
2006, 7, 1942.
38 I. Koltover, T. Salditt, C. R. Safinya, Biophys. J. 1999, 77,
915.
14 R. Koynova, R. C. MacDonald, Biochim. Biophys. Acta
2007, 1768, 2373.
39 I. Koltover, K. Wagner, C. R. Safinya, Proc. Natl. Acad.
Sci. U.S.A. 2000, 97, 14046.
15 R. Koynova, Y. S. Tarahovsky, L. Wang, R. C. MacDonald,
Biochim. Biophys. Acta 2007, 1768, 375.
40 R. Koynova, L. Wang, R. C. MacDonald, Proc. Natl. Acad.
Sci. U.S.A. 2006, 103, 14373.
16 R. Koynova, B. Tenchov, L. Wang, R. C. Macdonald, Mol.
Pharmacol. 2009, 6, 951.
41 C. R. Safinya, E. B. Sirota, R. J. Plano, Phys. Rev. Lett.
1991, 66, 1986.
17 D. Matulis, I. Rouzina, V. A. Bloomfield, J. Am. Chem.
Soc. 2002, 124, 7331.
42 S. Zhou, D. Liang, C. Burger, F. Yeh, B. Chu, Biomacro-
molecules 2004, 5, 1256.