Mesomorphic imidazolium salts: New vectors for efficient siRNA transfection

Article abstract of DOI:10.1021/ja903028f

The preparation of chloride (1_n) and bromide (2_n) derivatives of 1-methyl-3-[3,4-bis(alkoxy)benzyl]-4H-imidazolium with n ) 6, 12, 16, 18 is described. The two series of salts possess a rich thermotropic mesomorphism, chain-length de

Full text of DOI:10.1021/ja903028f

Published on Web 08/28/2009
Mesomorphic Imidazolium Salts: New Vectors for Efficient
siRNA Transfection
William Dobbs,† Benoˆıt Heinrich,† Cyril Bourgogne,† Bertrand Donnio,†
Emmanuel Terazzi,† Marie-Elise Bonnet,‡ Fabrice Stock,‡ Patrick Erbacher,‡
Anne-Laure Bolcato-Bellemin,‡ and Laurent Douce*,†
Institut de Physique et Chimie des Mate´riaux de Strasbourg, UMR 7504, CNRS-UniVersite´ de
Strasbourg, BP 43, 23 rue du Loess, F-67034 Strasbourg Cedex 2, France, and
Polyplus-transfection, BIOPARC, BouleVard Se´bastien Brandt, BP 90018, 67401 Illkrich, France
Received April 21, 2009; E-mail: Laurent.Douce@ipcms.u-strasbg.fr
Abstract: The preparation of chloride (1n) and bromide (2n) derivatives of 1-methyl-3-[3,4-bis(alkoxy)benzyl]-
4H-imidazolium with n ) 6, 12, 16, 18 is described. The two series of salts possess a rich thermotropic
j
mesomorphism, chain-length dependent. Thus, a lamellar smectic A phase, a bicontinuous cubic Ia3d
phase, and a columnar hexagonal liquid crystalline mesophase are induced as a function of increasing
chain length. The mesomorphic properties were studied by polarizing optical microscopy, differential scanning
calorimetry, and X-ray diffraction, and with the support of dilatometry and molecular dynamics, models for
the various supramolecular arrangements of the salts are proposed. Such cationic amphiphiles were
expected to be candidate molecules to design a new delivery reagent for nucleic acid transfection, particularly
for short interfering RNA (siRNA). The use of an RNA interference mechanism, by introduction into cells
by transfection of chemically synthesized siRNAs, is a powerful method for gene silencing studies. To
exploite the potential of these amphilic imidazolium salts, these molecules were formulated with cohelper
lipids and tested for their efficacy to deliver active siRNAs. Our results show high transfection efficacy of
our formulated compounds and high silencing efficiency with more than 80% inhibition of the targeted
gene at 10 nM siRNA concentration. Taken together our results show the potency of amphiphilic imidazolium
salts as a new generation of transfection reagents for RNA interference.
DNA being one example, by Coulombic interactions.2 The
description of cationic amphiphiles active as nucleic acid
transfection agents,3 in particular, has sparked major interest
within the scientific community.4 An important application of
this procedure, defined recently by A. Z. Fire and C. C. Mello,5
concerns the fact that double-stranded RNA triggers suppression
of gene activity in a homology-dependent manner, the RNA
interference (RNAi) process. The ability of RNAi to dramatically
and selectively reduce the expression of an individual protein
in a cell makes RNAi a valuable laboratory tool, both in cell
culture and in ViVo. Furthermore, RNAi protects against RNA
virus infections, especially in plants and invertebrate animals,
1. Introduction
Uniting the properties of ionic derivativesstheir low melting
points, low volatility, nonflammability, high chemical and
radiochemical stability, tunable conductivity, and wide electro-
chemical windowsswith those of liquid crystals, with their
many forms of labile macroscopic ordering, raises fascinating
prospects. This combination could lead to a vast range of new
multifunctional materials.1 Ionic molecules have also attracted
much attention in the fields of biorelated sciences, a major reason
for this interest being the ability of appropriate cationic
compounds to form liposomes able to confine anionic species,
(2) (a) Ra¨dler, J. O.; Koltover, I.; Salditt, T.; Safinya, C. R. Science 1997,
275, 810–814. (b) Cui, L.; Zhu, L. Lamgnuir 2006, 5982–5985. (c)
Miller, A. D. Angew. Chem., Int. Ed. 1998, 37, 1768–1785. (d) Cui,
L.; Miao, J.; Zhu, L. Macromolecules 2006, 39, 2536–2545. (e) Joester,
D.; Losson, M.; Pugin, R.; Heinzelmann, H.; Walter, E.; Merkle, H. P.;
Diederich, F. Angew. Chem., Int. Ed. 2003, 42, 1486–1490.
(3) (a) Huang, Q.-D.; Chen, H.; Zhou, L.-H.; Huang, J.; Wu, J.; Yu, X.-
Q. Chem. Bio. Drug Des. 2008, 71, 224–229. (b) Ito, A.; Miyazoe,
R.; Mitoma, J.; Akao, T.; Osaki, T.; Kunitake, T. Biochem. Int. 1990,
22, 235–241. (c) van der Woude, I.; Wagenaar, A.; Meekel, A. A.;
ter Beest, M. B.; Ruiters, M. H.; Engberts, J. B.; Hoekstra, D. Proc.
Natl. Acad. Sci. U.S.A. 1997, 94, 1160–1165. (d) Zhang, S.; Zhao,
B.; Jiang, H.; Wang, B.; Ma, B. J. Controlled Release 2007, 123,
1–10.
† CNRS-Universite´ de Strasbourg.
‡ Polyplus-transfection.
(1) (a) Wasserscheid, P.; Welton, T. Ionic Liquids in Synthesis; Wiley-
VCH: Weinheim, 2003. (b) Dupont, J.; de Souza, R. F.; Suarez, P. A.
Chem. ReV. 2002, 102, 3667–3692. (c) Song, C. E. Chem. Commun.
2004, 1033–1043. (d) Ohno, H. Electrochemical Aspects of Ionic
Liquids; Wiley-Interscience: 2005. (e) Crosthwaite, J. M.; Aki,
S. N. V. K.; Maginn, E. J.; Brennecke, J. F. J. Phys. Chem. B 2004,
108, 5113–5119. (f) Yoshizawa, M.; Xu, W.; Angell, C. A. J. Am.
Chem. Soc. 2003, 125, 15411–15419. (g) Fry, A. J. J. Electroanal.
Chem. 2003, 546, 35–39. (h) Ionic Liquids IIIA: Fundamental,
Progress, Challenges and Opportunities; Rogers, R. B., Seddon, K. R.,
Eds.; ACS Symposium Series 901; American Chemical Society:
Washington DC, 2005. (i) Ionic Liquids IIIB: Fundamental, Progress,
Challenges and Opportunities; Rogers, R. B., Seddon, K. R., Eds.;
ACS Symposium Series 902; American Chemical Society: Washington
DC, 2005.
(4) (a) Aigner, A. Curr. Opin. Mol. Ther. 2007, 9, 345–352. (b) Li, W.;
Szoka, F. C. Pharm. Res. 2007, 24, 438–49.
(5) (a) Fire, A. Z. Angew. Chem., Int. Ed. 2007, 46, 37, 6966–6984. (b)
Mello, C. C. Angew. Chem., Int. Ed. 2007, 46, 37, 6985–6994.
9
13338 J. AM. CHEM. SOC. 2009, 131, 13338–13346
10.1021/ja903028f CCC: $40.75  2009 American Chemical Society

Mesomorphic Imidazolium Salts for siRNA Transfection
A R T I C L E S
Scheme 1. Synthetic Procedurea
Figure 1. Phase and temperature transitions as a function of the alkyl chain
length (Cr ) Crystal, SmA ) smectic A phase, Cub ) cubic phase, Colh
) columnar hexagonal phase, Dec ) decomposition temperature). The
transition temperatures are assigned from the onset temperatures during
heating cycles.
a (i) DMF, K2CO3, BrCnH2n+1, 70 °C; (ii) THF, LiAlH4; (iii) CH2Cl2,
SOBr2, or SOCl2 in THF or toluene, methylimidazole, N2.
and ensures genome stability by keeping mobile elements silent.
Despite the current interest in cationic compounds for RNAi
delivery, it seems that only a few examples using an imidazo-
lium salt backbone as a polar headgroup on a cationic lipid
formulation have been investigated.6 The present article concerns
a synthetic route to amphiphilic imidazolium salts, which not
only exhibit thermotropic liquid crystal behavior but also are
capable of forming lipoplexes with nucleic acids. In association
with a colipid (dioleoyl phosphatidylethanolamine, DOPE), these
amphiphilic imidazolium salts are able to deliver siRNAs into
cells, leading then to efficient and specific endogenous gene
silencing. The ability of our amphiphilic imidazolium-based
formulations to complex chemically synthesized siRNA was
evaluated. Gene silencing efficiency of imidazolium/siRNA
complexes was examined on a human lung carcinoma cell line
stably expressing the luciferase (A549Luc) gene. Moreover the
particle size and zeta potential of the complex were measured.
Taken together our data showed the great potency of our
imidazolium salts to deliver active siRNAs into cells and point
out a new generation of efficient nonviral reagents for siRNA
delivery.
chromatography or crystallization and characterized by 1H NMR,
13C NMR{1H}, FT-IR spectroscopy, and elemental analysis.
Distinctive signals assigned to the CH group at the 2-position
1
of the imidazolium ring appear in the H NMR spectra of, for
example, 112 and 212 at δ ) 10.51 and 10.83 ppm and in the
13C NMR{1H} spectra at δ ) 138.07 and 137.47 ppm,
respectively. Synthesis of the siRNA was performed by Euro-
gentec (Seraing, Belgium) and is described in the Supporting
Information.
2.2. Investigation of the Liquid Crystalline Behavior. As
expected, all the ionic compounds 1n and 2n show liquid crystal
properties.8 The mesomorphic properties and the phase transition
temperatures were preliminarily studied by differential scanning
calorimetry (DSC) and polarizing microscopy (POM). These
data are summarized in Figure 1.
The melting temperatures of all the compounds with n > 6
show a low dependence upon chain length and are quasi
counteranion invariant (between 80-100 °C); the first homo-
logues of the series are nearly (16) or room-temperature liquid
crystals (26). Decomposition was however systematically de-
tected for all the other compounds above 200 °C, that is in the
isotropic liquid for 16 and 26 (clearing temperatures 180 and
165 °C, respectively, for the SmA-I phase transformation
temperatures) and still in the mesophase for the other members
of the series. The mesophases’ nature could be quite easily
assigned on the basis of their optical textures. Thus, a smectic
A (SmA) lamellar arrangement was induced for the short alkyl
chain-length homologues (16 and 26): on cooling from the
isotropic liquid, the samples appeared black (uniform homeo-
tropic texture) when viewed under cross polarizers alongside
alignment-induced birefringent defects near trapped air bubbles.
2. Results and Discussion
2.1. Synthesis. Chloride (1n) and bromide (2n) derivatives of
1-(3,4-bis(alkyloxy)benzyl)-3-methyl-1H-imidazo-3-ium with n
) 6, 12, 16, 18 have been synthesized in good yield, on the
gram scale. These compounds were obtained in three steps,
following slightly modified literature procedures.7 Methyl 3,4-
dihydroxybenzoate was etherified with various 1-bromoalkanes
(CnH2n+1Br, n ) 6, 12, 16, 18) in the presence of K2CO3 in
DMF, and the resulting amphipathic esters reduced by LiAlH4
into the corresponding benzyl alcohols. The different resulting
alcohols were converted into 3,4-bis(alkoxy)benzyl chlorides
or bromides with either thionyl chloride or thionyl bromide in
quantitative yields. The final compounds 1n and 2n were obtained
by quaternalization of 1-methylimidazole with the chlorides and
bromides under refluxing toluene or THF in inert atmosphere
(Scheme 1). The final compounds were purified by flash
(8) (a) Binnemans, K. Chem. ReV. 2005, 105, 4148–4204. (b) Kato, T.;
Misoshita, N.; Kishimoto, K. Angew. Chem., Int. Ed. 2006, 45, 38–
68. (c) Yoshio, M.; Mukai, T.; Ohno, H.; Kato, T. J. Am. Chem. Soc.
2004, 126, 994–995. (d) Bowlas, C. J.; Bruce, D. W.; Seddon, K. R.
Chem. Commun. 1996, 1625–1626. (e) Gordon, C. M.; Holbrey, J. D.;
Kennedy, A. R.; Seddon, K. R. J. Mater. Chem. 1998, 8, 2627–2636.
(f) Bradley, A. E.; Hardacre, C.; Holbrey, J. D.; Johnston, S.; McMath,
S. E. J.; Nieuwenhuyzen, M. Chem. Mater. 2002, 14, 629–635. (g)
De Roche, J.; Gordon, C. M.; Imrie, C. T.; Ingram, M. D.; Ingram,
A. R.; Kennedy, A. R.; LoCelso, F.; Triolo, A. Chem. Mater. 2003,
15, 2003. (h) Lee, K.-M.; Lee, Y.-T.; Lin, J. B. J. Mater. Chem. 2003,
13, 1079–1084. (i) Chiou, J. Y. Z.; Chen, J. N.; Lei, J. S.; Lin, I. J. B.
J. Mater. Chem. 2006, 16, 2972–2977. (j) Yazaki, S.; Kamikawa, Y.;
Yoshio, M.; Hamasaki, A.; Mukai, T.; Ohno, H.; Kato, T. Chem. Lett.
2008, 37,5, 528–539.
(6) (a) Me´vel, M.; Breuzard, G.; Yaouanc, J.-J.; Cle´ment, J.-C.; Lehn,
P.; Pichon, C.; Jaffre´s, P.-A.; Midoux, P. ChemBioChem 2008, 9,
1462–1471. (b) Midoux, P.; Pichon, C.; Yaouanc, J.-J.; Jaffre´s, P. Br. J.
Pharmacol. 2009, 157, 166–178.
(7) Dobbs, W.; Douce, L.; Allouche, L.; Louati, A.; Malbocs, F.; Welter,
R. New J. Chem. 2006, 30, 528–532.
9
J. AM. CHEM. SOC. VOL. 131, NO. 37, 2009 13339

A R T I C L E S
Dobbs et al.
Figure 2. POM textures of 212 on cooling (a) in the columnar hexagonal
phase at T ) 130 °C and (b) at the transition to the cubic phase (formation
of black polygonal areas).
j
Figure 4. Diffraction small-angle X-ray pattern of the Ia3d cubic phase
of 212 recorded at T ) 90 °C.
The cubic symmetry for the intermediate mesophase of 112
and 212 was assigned unambiguously by SAXS measurements
j
as belonging to the Ia3d space group. Up to six sharp reflections
were detected in the low angle range for which the reciprocal
d-spacings were in the ratios ꢀ3:ꢀ4:ꢀ7:ꢀ8:ꢀ10:ꢀ11. Since the
ratio ꢀ7 is not compatible with a cubic symmetry [(h2+k2+l2)1/2
) a/dhkl)], the sequence of numbers must be doubled as ꢀ6:
ꢀ8:ꢀ14:ꢀ16:ꢀ20:ꢀ22 (Figure 4, Table 1). Complete space
group determination of mobile thermotropic cubic phases is
difficult, although the large initial number of theoretical pos-
sibilities (36) can be reduced by logical analysis of the data, as
shown below. In our standard diffraction experiment, all
noncentrosymmetric groups (groups with Laue classes 23, 432,
or 43m) can be disregarded on account of Friedel’s law,9 which
leaves only 17 cubic space groups to consider. The sequence
of the ratios is not compatible with a face-centered-cubic
network (F) due to the presence of ratio ꢀ14 and is thus
excluded. Similarly, primitive Bravais lattices (P) are highly
improbable due to the unexplained absence of numerous
authorized reflections. The reflections were therefore success-
fully indexed as (211), (220), (321), (400), (420), and (332) of
a body-centered cubic network (I) with the cubic lattice
Figure 3. Temperature dependence of the lamellar SmA periodicities of
16 and 26.
For all the other homologues with a chain length greater than
12, a columnar phase was recognized by the observation of
large, birefringent developing domains coalescing into a mosaic-
like texture. For the intermediate compounds 112 and 212, on
cooling (rapid first, and then slow) from 200 °C, the temperature
at which decomposition starts to occur, the growth of a
homogeneous mosaic-like texture was clearly observed (Figure
2a). On further cooling, the materials undergo a transition into
a cubic phase, identified by the complete loss of birefringence
and the formation of large black, optically extinct, highly
viscous, polygonal areas (Figure 2b).
parameter, a, in the 9 nm range for both 112 and 212 (Table 1).
9
j
The extinction rule of the Ia3d space group is theoretically
The identification of all the mesophases was supported by
small-angle X-ray diffraction. SAXS patterns were collected for
all the salts as a function of temperature. Typically, in all cases,
a series of sharp and intense peaks were observed in the low-
angle part of the diffractogram, characterizing the mesophase
symmetry, while, in the large angle region, i.e., at ca. 4.5-4.6
Å, a very intense and diffuse scattering halo was observed,
corresponding to the molten alkyl chain in liquid-like order.
Imidazolium salts 16 and 26 exhibit a layered structure over
the whole mesomorphic temperature range, with two sharp and
intense small-angle reflections, in the ratio 1:2, systematically
observed on the diffractograms. In both systems (1 and 2), the
smectic periodicity decreases with increasing temperature (see
Tables 2 and 3 in Supporting Information), reflecting the higher
thermal mobility of the alkyl chains, concomitant to the smectic
layers’ compression; this trend is perfectly reproducible, and
the periodicity values can be quasi superimposed on both heating
and cooling (Figure 3). Interestingly also, the periodicity of the
chloride salts is always greater than that of the bromides, in
concordance with the slightly larger ionic radius of the former
(rCl- ) 99.4 pm < rBr- ) 114.5 pm), and thus the larger cross-
sectional area (vide infra).
compatible with this set of observed reflections; moreover, the
strongest hkl reflections, (211) and (220), are the first reflections
10
j
permitted by the Ia3d cubic space group.
Finally, the XRD patterns recorded for the salts 112, 116, 118,
212, 216, and 218 are characterized by the presence of two up to
four sharp small-angle X-ray reflections, depending on the
sample, with reciprocal d-spacings in the ratio 1:ꢀ3:2:ꢀ7. These
three features are most readily assigned as the (10), (11), (20),
and (21) reflections of a columnar phase with a hexagonal lattice.
The variation of the cross-columnar section area S (S ) 1/2a2ꢀ3)
was followed as a function of temperature for each of the
derivatives (Figure 5 and Supporting Information). S is found
(9) [0kl:k,l ) 2n, hhl:2h + l ) 4n, h00:h ) 4n, where n is an integer, and
h, k, and l are permutable] International Tables for Crystallography,
4th ed.; Hahn, T., Ed.; The International Union of Crystallography;
Kluwer Academic Publishers: Dordrecht, The Netherlands, Boston,
London, 1995; Vol. A.
(10) (a) Diele, S. Curr. Opin. Colloid Interface Sci. 2002, 7, 333. (b)
Kutsumizu, S. Curr. Opin. Solid State Mater. Sci. 2002, 6, 537. (c)
Impe´ror-Clerc, M. Curr. Opin. Colloid Interface Sci. 2005, 9, 370.
(d) Lynch, M. L.; Spicer, P. T. Bicontinuous Liquid Crystals; Taylor
& Francis: 2005; p 127.
9
13340 J. AM. CHEM. SOC. VOL. 131, NO. 37, 2009

Mesomorphic Imidazolium Salts for siRNA Transfection
A R T I C L E S
Table 1. Indexation of the Reflections Detected in the Cubic Liquid-Crystalline Phase by SAXS for Compounds 112 and 212 at a Given
Temperature
112 at T ) 100 °C
dcalcd/Åc
212 at T ) 90 °C
dcalcd/Åc
hkla
I/a.u.b
dmeas/Åc
parametersd
dmeas/Åc
parametersd
211
220
321
400
420
332
VS (sh)
S (sh)
M (sh)
W (sh)
VW (sh)
VW (sh)
37.12
32.15
24.39
22.76
20.36
19.38
37.16
32.18
24.33
22.76
20.35
19.41
37.03
32.10
24.38
22.46
20.28
19.31
37.00
32.05
24.22
22.66
20.26
19.32
a ) 91.02 Å
V ) 754068 Å3
N ) 732
a ) 90.64 Å
V ) 744663 Å3
N ) 730
a Miller indices. b I corresponds to the intensity of the reflections (VS: very strong, S: strong; M: medium; W: weak; VW: very weak; br and sh stand
c
for broad and sharp). dmeas and dcalcd are the measured and calculated diffraction spacing. d Phase parameters: a, lattice parameter of the cubic phase; V,
volume of the cubic cell (Å3), and N, number of molecules within the cubic cell (N ) Å3/Vmol).
Figure 6. Molecular volume variation of 212 as function of temperature,
determined by dilatometry.
Figure 5. Variation of the columnar cross sections (Colh phases) with
temperature, chain length, and counterion.
liquid crystals.11 In detail, a slight change in the slope could be
detected at the cubic-to-columnar phase transformation (at ca.
118 °C; see Figure 6), although this change remains meaning-
less, and in the following, only the linear variation over the
entire temperature domain will be considered. The slope is in
fairly good agreement with the volume variation in the meth-
ylene groups in liquid alkanes,12 an indication that the aromatic
component of the volume (Var) is almost unchanged with
temperature in the mesophase (and will be considered thereafter
as a fixed value over the mesomorphic temperature range) and
that the molecular volume is dominated by that of the aliphatic
chains. Therefore, the molecular volume Vmol can be written as
a sum of elementary volumes associated with the different
segments of the molecule according to Vmol(T) ) Var + Vch(T),
where Var is the volume of the rigid part, and Vch(T) ) 2nVCH2(T)
) 2n(26.5616 + 0.02023T). Note that the contribution of the
counterion to the molecular volume is negligible within the same
series but also between series 1 and 2, with a difference of less
than 10 Å3.
to decrease with T in agreement with the increasing chain
mobility and disorder. It can be observed that this decrease is
strongly dependent on both the chain length and the counter-
anion. For instance, the slope increases very steeply as the chain
length is augmented from 12 to 18 methylene units per chain,
which suggests strong chain coiling and/or interdigitation for
the longer chains’ homologues. Moreover, while the increase
in the cross section is quite regular for the chloride derivatives,
we observe a settling as far as the bromides are concerned. This
point will be discussed in the following.
2.3. Dilatometric Studies. Prior to the analysis of the mo-
lecular packing of the various compounds in their mesophases,
the molecular volume (Vmol) of one salt, 212, was measured
experimentally by dilatometry and its variation followed as a
function of temperature. The experiment was carried out with
approximately 1 g of compound, between room temperature and
160 °C, including the Cr-to-Cub and the Cub-to-Colh phase
transformations (Figure 6). The molecular volume was found
to augment with temperature in the stability range of each one
of the phases observed. Upon cooling, the volume decreases
quite reversibly in the whole range explored, and the cubic phase
supercools, as usual, slightly below the melting temperature.
From the slope of the molecular variation of 212 as a function
of temperature, with a linear fit approximation, the volume is
found to be equal to 935.7 + 0.95T, with T in degree Celsius.
Thus, the volume of 2i or 1i can be written as Vmol(2i/1i) ) 935.7
+ 0.95T + 2 × (i - 12) × (26.5616 + 0.02023T).
At ∼80-81 °C in agreement with the results from optical
microscopy and differential scanning calorimetry, the crystal-
to-cubic phase transition was detected by an abrupt jump in
the molecular volume, corresponding to the melting of the
aliphatic chains. Then, above this transition, its temperature
variation increases quasi-linearly despite the change from one
mesophase to another (Cub-to-Colh), in common with other
2.4. Supramolecular Organization in the Mesophases.
2.4.1. Smectic Phase. In smectic systems, it is practical to
consider the molecular cross-sectional area, AM. This parameter
(11) Mathevet, F.; Masson, P.; Nicoud, J.-F.; Skoulios, A. Chem.sEur. J.
2002, 8, 2248.
(12) Doolittle, A. K. J. Appl. Phys. 1951, 22, 1471.
9
J. AM. CHEM. SOC. VOL. 131, NO. 37, 2009 13341

A R T I C L E S
Dobbs et al.
Figure 7. Schematic drawing of the possible arrangement of the imidazolium species and anions in an ionic double layer.
close to 52.5 Å2 (estimated by the linear fits as a function of
temperature of the AM temperature variations), at T ) 180 and
T ) 165 °C, respectively (Figure 8), indicative that the process
of layer destruction is the same for both compounds and is solely
governed by the molecular cross-sectional area.
2.4.2. Columnar Phase. Due to the curvature of the aliphatic-
ionic interface intervening at a longer chain length (n g 12),
i.e., when the contribution of the aliphatic volume fraction is
becoming the dominant factor, a finite number of molecules
self-aggregate together in the head-to-head fashion (with the
polar parts in the columns interior or spine) to generate the
columns (or discs).8b The lateral dimension of the columns is
significantly larger than twice the length of the molecules as
estimated from molecular modeling (for example, 112 corre-
sponds to 24.5 Å). These measurements strongly suggest that
the molecules therefore are not stacked on top of each other in
single rows but densely aggregated around the columnar axis.
From a geometrical point of view, the columnar struc-
tures are characterized mainly by two parameters: the columnar
cross section (S) and the stacking periodicity along the columnar
axis (h).13 Knowledge of these two structural parameters permits
interpretation of the molecular packing inside the columns.14
S, h, and Vmol are linked analytically through the relation given
by the equation hS ) NVmol, in which N is the number of
molecules (or molecular equivalents) within a columnar stratum
of thickness h and section S. It was found, for h fixed to 4.5
Å,14 that N varies significantly as the function of the molecular
structure (counterion and chain length) and with temperature
(Figure 9). In all cases, the number of molecular equivalents
per columnar stratum 4.5 Å thick decreasees with temperature,
in concordance with the reduction of the columnar cross section.
This result can be interpreted as the shrinkage and stretching
of the columnar cores with increasing temperature with columnar
undulations becoming less and less important. This effect is also
Figure 8. Molecular area variation in the smectic phase as a function of
temperature and counterion (16, 26).
permits evaluation of the lateral packing of the molecules within
the smectic layers and, moreover, discrimination between mono-
and bilayer smectic arrangements. Considering a cylindrical
conformation for the elemental molecular assembly(ies), AM is
obtained by the ratio NVmol/d, in which d is the layer periodicity
measured by XRD, Vmol the molecular volume measured by
dilatometry, and N the number of molecules (N ) 1 or 2). Thus,
after calculation, molecular areas were found to be compatible
with a bilayer structure (N ) 2), with two molecules coupled
head-to-head through their ionic headgroups, and with the chains
homogeneously distributed on either side of the ionic polar plane
formed by the polar heads and the counterions (Figure 7). The
area of derivative 2 is slightly larger than that of 1, in good
agreement with the slightly larger radius of the bromide with
respect to the chloride ion. As the temperature is raised, the
molecular area increases linearly (identical behavior for both
compounds, same slope, Figure 8) concomitantly with the
increasing disordering of the molten chains, until chain fluctua-
tions become too strong and disorganize the layers’ arrange-
ments completely. An interesting geometrical relationship has
been deduced between both compounds 16 and 26. Indeed, they
both are clear when their molecular areas reach the same value
(13) Guillon, D. Struct. Bonding (Berlin) 1999, 95, 41.
(14) (a) Morale, F.; Date, R. W.; Guillon, D.; Bruce, D. W.; Finn, R. L.;
Wilson, C.; Blake, A. J.; Schro¨der, M.; Donnio, B. Chem.sEur. J.
2003, 9, 2484. (b) Donnio, B.; Heinrich, B.; Allouchi, H.; Kain, J.;
Diele, S.; Guillon, D.; Bruce, D. W. J. Am. Chem. Soc. 2004, 126,
15258.
9
13342 J. AM. CHEM. SOC. VOL. 131, NO. 37, 2009

Mesomorphic Imidazolium Salts for siRNA Transfection
A R T I C L E S
Figure 9. Variation of the number of molecular equivalents, N, as a function
of temperature, chain length (n ) 12, 16, 18) and couterion (1 vs 2).
attenuated with the length of the chains: the undulations are
less and less perceptible as the chains increase, and even
essentially absent at long chain length (for n g 16). As expected,
the number of molecular equivalents per elementary columnar
slice increases with increasing n, but this increase is strongly
counterion dependent (Figure 9).
To further understand the self-organization inside the column,
we have conducted molecular dynamic (MD) simulations in
which a column model was built from the experimental X-ray
data. In a first step the cation moiety of 212 was optimized with
the AM1 semiempirical method (MOPAC 2007) to obtain
reliable atomic electronic densities for the MD calculations.
Then a hexagonal cell incorporating 12 molecules of 212 was
built with a and b parameters set according to the X-ray lattice
parameters and with a thickness calculated (h ≈ 9 Å) so that
the average density of the material is the same as that measured
experimentally by dilatometry. After a few optimization steps,
the cell was equilibrated with an isothermal 200 ps MD
simulation in periodic boundary conditions. The resulting
molecular arrangement showed a columnar backbone made from
the bromide and polar part of the imidazolium moiety sur-
rounded by molten alkyl chains, and the columnar cross section
is fully filled (Figure 10). The columnar cross section does not
Figure 10. Snapshots showing (a) the molecular self-assembly of 212 into
columns and (b) their packing in the hexagonal 2D network (polar central
imidazolium with the nitrogen atoms in blue sheating with alkyloxy chains
(oxygen atoms in red) and bromide counterion shown in pale red).
appear circular but is averaged along the columnar axes and
from one column to another, giving rise to the appropriate 6-fold
symmetry of the columnar phase.
2.4.3. Cubic Phase. While commonly induced in various
binary lyotropic systems,15 and being well described mathemati-
cally by films of constant thickness separating polar and
nonpolar regions (the so-called infinite periodic minimal surfaces
(15) (a) Fontell, K. AdV. Coll. Interface 1992, 41, 127. (b) Seddon, J. M.;
templer, R. H. Phil. Trans. R. Soc. Lond. A 1993, 344, 377.
j
concept), the cubic mesophase with the Ia3d space group is
(16) (a) Etherington, G.; Leadbetter, A. J.; Wang, X. J.; Gray, G. W.;
Tajbakhsh, A. Liq. Cryst. 1986, 1, 209. (b) Kutsumizu, S.; Yamada,
M.; Yano, S. Liq. Cryst. 1994, 3, 155. (c) Fischer, S.; Fischer, H.;
Diele, S.; Pelzl, G.; Jankowski, K.; Schmidt, R. R.; Vill, V. Liq. Cryst.
1994, 17, 855. (d) Bruce, D. W.; Donnio, B.; Hudson, S. A.; Levelut,
A. M.; Megtert, S.; Petermann, D.; Ve´ber, M. J. Phys. II (France)
1995, 5, 289. (e) Donnio, B.; Heinrich, B.; Gulik-Krzywicki, T.;
Delacroix, H.; Guillon, D.; Bruce, D. W. Chem. Mater. 1997, 9, 2951.
(f) Tsiourvas, D.; Kardassi, D.; Paleos, C. M.; Gehant, S.; Skoulios,
A. Liq. Cryst. 1997, 23, 269. (g) Paleos, C. M.; Kardassi, D.; Tsiourvas,
D.; Skoulios, A. Liq. Cryst. 1998, 25, 267. (h) Go¨ring, P.; Diele, S.;
Fischer, S.; Wiegeleben, A.; Pelzl, G. Liq. Cryst. 1998, 25, 467. (i)
Tsiourvas, D.; Paleos, C. M.; Skoulios, A. Macromolecules 1999, 32,
8059. (j) Kutsumizu, S.; Morita, K.; Ichikawa, T.; Nojima, S.;
Yamaguchi, T. Liq. Cryst. 2002, 29, 1447. (k) Tsiourvas, D.;
Stathopoulou, K.; Sideratou, Z.; Paleos, C. M. Macromolecules 2002,
35, 1746. (l) Neve, F.; Impe´ror-Clerc, M. Liq. Cryst. 2004, 31, 907.
(m) Massiot, P.; Impe´ror-Clerc, M.; Veber, M.; Deschenaux, R. Chem.
Mater. 2005, 17, 1946. (n) Kutsumizu, S.; Saito, K.; Nojima, S.; Sorai,
M.; Galyametdinov, Y. G.; Galyametdinova, I.; Eidenschink, R.;
Haase, W. Liq. Cryst. 2006, 33, 75. (o) Martin, J. D.; Keary, C. L.;
Thornton, T. A.; Novotnak, M. P.; Knuston, J. W.; Folmer, J. C. W.
Nat. Mater. 2006, 5, 271. (p) Kutsumizu, S.; Mori, H.; Fukatami, M.;
Naito, S.; Sakajiri, K.; Saito, K. Chem. Mater. 2008, 20, 3675.
rarely observed in thermotropic systems.16 Moreover, its su-
pramolecular organization is not yet completely understood,
particularly on space filling geometrical grounds: indeed, the
available volume must be completely filled by the mesogens,
because it is not compensated by another constituent, such as
water, as in the counterpart lyotropic phase. The structure of
j
the thermotropic bicontinuous Ia3d cubic phase can be under-
stood by the double-network structural model and consists of
two intertwined, unconnected 3D networks of 24 short column
segments, coplanarly linked 3 × 3 at each node of the network,
and separated by the Gyro¨ıd infinite periodic minimal surface.
In this case, the most plausible hypothesis is to consider that
the aromatic parts form two separate continuous interwoven
networks of channels immersed in the aliphatic matrix, therefore
adopting an inverse type structure (Figure 11).10,15 From
dilatometry, ∼730 molecules are needed to completely fill the
available volume of the cubic phase, i.e., 365 molecules per
individual network. Moreover from dilatometry, approximately
9
J. AM. CHEM. SOC. VOL. 131, NO. 37, 2009 13343

A R T I C L E S
Dobbs et al.
Figure 11. Schematic representation of the rod model associated to the
j
Ia3d cubic symmetry: two interwoven unconnected networks linked 3 × 3
in a coplanar fashion (24 columns per unit cell). In the model, the rods are
filled with the polar aromatic parts whereas the void is filled by the molten
aliphatic chains (see text).
two-thirds of the cubic lattice is filled by molten aliphatic chains,
whereas the networks correspond to the remaining third (the
polar part).
Figure 12. Electrophoresis on 1.2% agarose gel of complexes formed
between siRNAs and 212 (a), 112 (b), 118 (c) compounds formulated with
DOPE. 700 ng of siRNAs were complexed with increasing amount of
formulated compound (1, 2, 3, 5, 10, and 20 µL, i.e., 1, 2, 3, 50, 10, and
20 nmol of imidazolium).
A detailed description of the cubic phase structure is
complicated because of the estimation of the volume (and shape)
at the junctions of three different minicylinders. Indeed, the
column segments are not perfectly joined end-to-end at these
connections but separated by a short gap which depends on the
diameter of the column segment, φ: therefore their length is
smaller than in the case of columns joined without discontinuity
(φ ) 0, i.e., mathematical lines). The distance between two
ternary junctions forming the continuous network frame (el-
emental rod length), l0, is directly related to the cubic lattice
parameter l0 ) a/ꢀ8 (φ ) 0). However, when φ * 0, this
distance is reduced to lφ ) a/ꢀ8 s ꢀ(φ). Neither the evaluation
of ꢀ nor therefore the repartitioning of molecules within the
cubic structure is straightforward. At these crystallographic
positions, the number of molecules increases necessarily with
respect to the monotonous distribution in the cylinders. In the
present case, one can assume that the overall column integrity
is kept during the Colh-to-Cub phase transformation (identical
cross sections, SCub ) SColh ) S, where Scub corresponds to the
cross-columnar section of the minicolumn within the cubic
phase), a reasonable hypothesis due to great molecular flexibility,
and that, as the temperature is reduced, the columns formed by
the piled polar aromatic cores in the Colh phase evolve into
the rod networks, which are surrounded by the molten aliphatic
chain continuum. This transformation may occur through regular
column undulations and interconnections, which increase sub-
stantially with reducing temperature, to permit the formation
of the two interlaced networks, similarly to the supramolecular
model described earlier for the bistilbazole silver complexes.16e
2.5. Transfection Activities. Delivery of siRNA into cells is
one of the major concerns for therapeutic applications of siRNA.
Three stages are required for efficient siRNA delivery: (1)
protection of nucleic acid, (2) its penetration into cells, and (3)
its transfer inside the cytosol. Based on their structure, our
molecules can have the capacity to achieve the three steps. The
cationic part is able to bind the siRNA, whereas the liphophilic
part enables the molecule to cross the lipids bilayers of cell
membrane and go into cells. Compounds 212, 112, and 118 were
evaluated for their ability to form a complex with synthetic
siRNA using a gel retardation assay. Each compound was
formulated at 1 mM with and without 2 mM of 1,2-dioleoyl-
sn-glycero-3-phosphoethanolamine (DOPE) in water. A colipid
was added to the cationic liposomal formulation not only to
increase lipoplex fusogenicity but also to decrease their toxicity.
DOPE was chosen as it can form the H11 phase structure known
to induce supramolecular arrangement leading to the fusion of
a lipid bilayer at a temperature greater than 5-10 °C. Its
incorporation is also supposed to destabilize the endosomal
membrane.17 Constant amounts of siRNAs were mixed with
increasing amounts of imidazolium-based formulations and
loaded on agarose gel.
These experiments (Figure 12) show the naked siRNA and
the complexes, retained in the well, formed between the siRNAs,
and 212-, 112-, and 118-based formulations. The results showed
that siRNA binding to the compound tested is not affected by
the counterion (Cl- or Br-), as the same volume of formulated
212, 112 is necessary to complex all the siRNAs. However, the
length of the alkyl chain influenced the siRNA binding efficacy,
as a higher amount of 118 is required to complex all the siRNAs.
These data show the ability of our compounds to interact with
siRNA and form complexes, the first property required for
(17) (a) Gruner, S. M.; Tate, M. W.; Kirk, G. L.; So, P. T.; Turner, D. C.;
Keane, D. T.; et al. Biochemistry 1988, 27, 2853–2866. (b) Farhood,
H.; Serbina, N.; Huang, L. Biochim. Biophys. Acta 1995, 1235, 289–
295.
9
13344 J. AM. CHEM. SOC. VOL. 131, NO. 37, 2009

Mesomorphic Imidazolium Salts for siRNA Transfection
A R T I C L E S
Table 2. Particle Size and Zeta Potential of siRNA (700 ng)
Complexed with 4 µL of Formulated 112 and 118
formulation
size (nm)a
zeta potential (mV)
112/DOPE/siRNA
118/DOPE/siRNA
68.7 ( 25.7
593 ( 25.7b
+61.1
-0.28
a Each value represents the mean
(
SD of 15 measurements.
b Polydisperse particle sizes have been observed.
efficient siRNA delivery. Then, to bind to the cell surface and
enter into cells, positively charged complexes are required. The
physical properties such as surface charge and size of the
particles were determined using Dynamic Light Scattering
(DLS) measurements. These parameters influence cellular
interactions and nanoparticle distribution. The size (mean around
70 nm) and zeta potential (highly positive) (Table 2) of the
particles (complexes) formed between siRNA and the compound
112 are compatible with the first step based on the sticking of
positively charged nanoparticles onto the plasma membrane,
followed by endocytosis.
The neutral zeta potential of the particles formed with the
118-based formulation confirms its low ability to complex
siRNAs and consequently is a sign of a probably low capacity
to deliver synthetic genetic material into cells. Moreover, the
size of the complex formed with compound 118 (ca. 600 nm)
which is not suitable for cellular uptake indicated an aggregation
tendency, suggesting a colloidal instability behavior not favor-
able for transfection. Then, the ability of each formulation to
deliver siRNA into cells, in the presence of serum, was
determined using fluorescently labeled siRNAs. 50 nM siRNAs
labeled with Rhodamine and complexed with a constant amount
of each formulation (i.e., 212, 112, and 118) were delivered into
a human lung carcinoma cell line stably expressing the luciferase
(A459Luc cells). The fluorescence was observed 4 and 24 h
after transfection. As shown in Figure 13, siRNAs delivered
with compounds 212 and 112 are visible, associated to the cells
4 h after transfection. Fluorescence into the cells is more uniform
24 h after transfection with decreasing intensity, demonstrating
intracellular release of the siRNAs from the endosomes into
the cytoplasm (data not shown). With compound 118 complex
aggregates were seen outside and not associated to the cells.
Moreover, the number of cells transfected with the siRNAs is
less important.
Figure 13. A549Luc cells were transfected with 50 nM Rhodamine labeled
siRNAs complexed with 6 µL of 212 (a), 112 (b), 118 (c) compounds
formulated with DOPE. Fluorescence was observed 4 h after transfection
using a microscope.
The presence of aggregates outside the cells confirms the size
value obtained by DLS (Table 2). We show that compounds
212 and 118 have the ability to (1) bind to the cell and (2) transfer
siRNA into the cytosol. However, functionality of the delivered
siRNA has to be demonstrated. Then, A549Luc cells, stably
expressing the luciferase gene, were used as a model system.
Luciferase specific and mismatch siRNAs (10 and 1 nM) were
delivered after complexation with a constant amount of formu-
lated 212-, 112-, and 118-based formulations. The luciferase
expression was determined 48 h after transfection. As a positive
control for transfection, Lipofectamine2000 was used in this
experiment. Figure 14 shows that the luciferase level, expressed
relative to the level obtained with the mismatch siRNAs, was
reduced by 70-80% when siRNAs (10 nM) were delivered
using 212 and 112 formulated with DOPE. No significant
luciferase inhibition (<5%) was observed by using the mismatch
siRNA (data not shown).
Figure 14. A549Luc cells, stably expressing the luciferase gene, were
transfected with luciferase specific siRNA complexed with 112, 212, and 118
formulated with DOPE. As a positive control, Lipofectamine2000 was
included in the experiment. Luciferase gene expression was measured 48 h
after transfection.
to 4 mM). At 10 nM, the highest inhibition was observed with
the formulation 1:2 (1 mM imidazolium:2 mM DOPE). A strong
toxicity was observed with the ratio 1:1 (data not shown). These
data confirm that the use of a colipid is necessary to increase
transfection efficiency. It is also shown that, in our case, the
best ratio between the cationic liposomal formulation and
the colipid is 1:2. As expected because of its low siRNA binding
efficacy, the inhibition obtained with the 118-based formulation
was lower (<30% at 10 nM of siRNA) confirming that the length
of the alkyl chain is important for the delivery of active siRNA.
These two formulations (212 and 112) are still significantly
efficient, even at a low siRNA concentration of 1 nM. To
confirm the necessity of a cohelper lipid, compound 112 was
also formulated with different amounts of DOPE (from 0 mM
9
J. AM. CHEM. SOC. VOL. 131, NO. 37, 2009 13345

A R T I C L E S
Dobbs et al.
Taken together, our results show that successful gene silencing
can be achieved with imidazolium-based formulations (exam-
plified with 212 and 112 compounds) when formulated with a
colipid DOPE, a well-known cohelper phospholipid for trans-
fection. We showed also that the length of the alkyl chain of
the imidazolium compound has to be optimized for an efficient
binding of siRNA, as well as for efficient gene silencing in cells.
As transfection is not restricted to siRNA binding and plasma
membrane sticking, this suggest also that the imidazolium
moiety exhibits the required properties to efficiently achieve
the intracellular transfection steps leading to the final release
of siRNA inside the cytoplasm.
promising amphiphilic molecules for siRNA delivery when
formulated with colipids. It is also noteworthy that only the
compounds bearing dodecyl tails (112 and 212) exhibit both cubic
and columnar phases as well as show greater ability to
encapsulate siRNA and a higher transfection yield. We are now
attempting to investigate the lyotropic behavior of these
compounds.
Acknowledgment. We are especially grateful to Dr. J.
Harrowfield for critical evaluation of the manuscript. This work
was supported by the CNRS and University of Strasbourg.
Supporting Information Available: Experimental procedures,
products characterization (spectroscopic and analytical data),
biological experimental section, including the synthesis of
siRNA and X-ray experimental data of the mesophases. This
material is available free of charge via the Internet at http://
pubs.acs.org.
3. Summary
In conclusion, we have prepared new dicatenar imidazolium
compounds exhibiting a smectic A mesophase for short alkyl
chains and a columnar hexagonal phase constituted of aggregates
for long tails. We have also characterized a rare example of an
j
Ia3d bicontinuous cubic phase with this imidazolium head.
JA903028F
These salts not only show thermotropic properties but also are
9
13346 J. AM. CHEM. SOC. VOL. 131, NO. 37, 2009