Structure-delivery relationships of lysine-based gemini surfactants and their lipoplexes

Article abstract of DOI:10.1039/c4sm00881b

The synthesis and properties of gemini surfactants of the type (R ¹(CO)-Lys(H)-NH)₂(CH₂)_n are reported. For a spacer length of n = 6, the hydrophobic acyl tail was varied in length (R¹ = C8, C10, C12, C14, C16, and C18) and, for R¹ = C18, the degree of unsaturation. For R¹(CO) = oleoyl (C18:1 Z) the spacer length (n = 2-8) and the stereochemistry of the lysine building block were varied; a 'half-gemini' derivative with a single oleoyl tail and head group was also prepared. The potential of the gemini surfactants to transfer polynucleotides across a cell membrane was investigated by transfection of HeLa cells with beta-galactosidase, both in the presence and absence of the helper lipid DOPE. Oleoyl was found to be by far the best hydrophobic tail for this biological activity, whereas the effect of the lysine stereochemistry was less pronounced. The effect of an optimum spacer length (n = 6) was observed only in the absence of helper lipid. The most active surfactant, i.e. the one with oleoyl chains and n = 6, formed liposomes with sizes in the range of 60-350 nm, and its lipoplex underwent a transition from a lamellar to a hexagonal morphology upon lowering the pH from 7 to 3. the Partner Organisations 2014.

Full text of DOI:10.1039/c4sm00881b

Soft Matter
View Article Online
View Journal | View Issue
PAPER
Structure–delivery relationships of lysine-based
gemini surfactants and their lipoplexes†
Cite this: Soft Matter, 2014, 10, 5702
a
a
a
Mark Damen, Edgar Cristobal-Lecina, Gloria Colom Sanmartı, Stijn F. M. van
´
`
´
a
a
b
a
Dongen, Cristina L. Garcıa Rodrıguez, Igor P. Dolbnya,‡ Roeland J. M. Nolte
´
´
and Martin C. Feiters*a
1
The synthesis and properties of gemini surfactants of the type (R (CO)-Lys(H)-NH)
2
(CH
2
)
n
are reported. For
1
a spacer length of n ¼ 6, the hydrophobic acyl tail was varied in length (R ¼ C8, C10, C12, C14, C16, and
1
1
C18) and, for R ¼ C18, the degree of unsaturation. For R (CO) ¼ oleoyl (C18:1 Z) the spacer length (n ¼ 2–8)
and the stereochemistry of the lysine building block were varied; a ‘half-gemini’ derivative with a single
oleoyl tail and head group was also prepared. The potential of the gemini surfactants to transfer
polynucleotides across a cell membrane was investigated by transfection of HeLa cells with beta-
galactosidase, both in the presence and absence of the helper lipid DOPE. Oleoyl was found to be by far
the best hydrophobic tail for this biological activity, whereas the effect of the lysine stereochemistry was
less pronounced. The effect of an optimum spacer length (n ¼ 6) was observed only in the absence of
helper lipid. The most active surfactant, i.e. the one with oleoyl chains and n ¼ 6, formed liposomes with
sizes in the range of 60–350 nm, and its lipoplex underwent a transition from a lamellar to a hexagonal
morphology upon lowering the pH from 7 to 3.
Received 23rd April 2014
Accepted 3rd June 2014
DOI: 10.1039/c4sm00881b
www.rsc.org/softmatter
molecule and, at the same time, of multiple cationic surfactant
1
. Introduction
molecules in a lipid aggregate both contribute to a strong
7
Gemini surfactants are symmetrical surfactants featuring two binding of DNA due to a multivalency effect. The mutual
sets of hydrophilic head groups and hydrophobic tails that are inuence of the multiple protonable groups in a molecule and
1
joined by a spacer. Both exible spacers and rigid spacers have within an aggregate on each other's pK
values leads to a pH-
a
2,3
been reported, and gemini surfactants generally have superior dependence of the DNA binding process, which stimulates
surfactant properties compared to their single head single tail endosomal escape when the lipid-DNA complex (or lipoplex) is
analogues. For instance, increased surface activities of several taken up by a cell.
orders of magnitude have been reported. They also typically
2 2 n
have critical micellar concentrations (CMCs) in the micromolar description (R (CO)-Lys(R )-NH) (CH ) (shorthand notation
range, much lower than those of classical surfactants. These R -Lys-R -n) and contains two moieties of the natural amino
superior attributes make gemini surfactants promising candi- acid lysine (Lys represents the natural L-isomer unless explicitly
dates in studies of biomedical applications, because concen- indicated otherwise). These amino acids are acylated on their 3-
trations can be lowered to minimize toxicity without losing amino groups, and are coupled by a diamine linker between
8
4
A promising class of gemini surfactants has the general
1
2
1
2
activity.
their carboxylic acid groups (Fig. 1). Of the series (Oleoyl-Lys(H)-
5
We have demonstrated previously that cationic gemini NH)
2
(CH
2
)
n
(abbreviated as Ol-Lys-H-n) with n ¼ 2, 4, 6, or 8, the
6
surfactants, e.g. those based on tartaric acid, can be applied in compound with n ¼ 6 (Ol-Lys-H-6) was among the most active
transfection, i.e. the process of delivering a gene to a cell. The compounds in the transfection of Chinese hamster ovary (CHO)
presence of multiple cationic groups in a single surfactant cells with luciferase, using LipofectAmine Plus® as a positive
5
control. The Arg analogues were the rst representatives of this
class of compounds, featuring antimicrobial activity, later also
found for C12-Lys-H-6.
9
a
Radboud University Nijmegen, Institute for Molecules and Materials,
Heijendaalseweg 135, 6525 AJ Nijmegen, The Netherlands. E-mail: m.feiters@
science.ru.nl; Fax: +31-24-3652929; Tel: +31-24-3652016
10,11
12
We have recently reported on transfection studies with the
so-called geminoid Ol-SPKR-Ol, a compound related to gemini
surfactants but more asymmetric, which is remarkably effective
b
DUBBLE CRG/ESRF, Netherlands Organization for Scientic Research (NWO), c/o
ESRF BP 220, F38043 Grenoble Cedex, France
†
1
‡
Electronic supplementary information (ESI) available. See DOI: without the non-cationic lysogenic helper lipid dio-
0.1039/c4sm00881b
leoylphosphatidyl ethanolamine (DOPE) usually required for
Present address: Diamond Light Source Ltd, Didcot OX11 0DE, Oxon, England,
13–15
5
transfection by lipids.
Interestingly, the preliminary results
UK.
5702 | Soft Matter, 2014, 10, 5702–5714
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Soft Matter
1
2
1
2
Fig. 1 (a) Schematic representation of a cationic gemini surfactant. (b) General structure of (R (CO)-Lys(R )-NH)
structure of the biologically active (Oleoyl-Lys(H)-NH) (CH or Ol-Lys-H-6.
2 2
(CH )
n
(or R -Lys-R -n). (c) Full
2
2 6
)
indicated that transfection by Ol-Lys-H-n also proceeds without
2.1.2. Mass spectrometry (MS). All the mass analyses were
helper lipid. We report here on the synthesis and some trans- performed by using electrospray techniques (ESI). High-Reso-
fection, with and without helper lipid, and physicochemical lution mass spectrometry measurements were performed on a
1
2
properties of various compounds (R (CO)-Lys(R )-NH)
2
(CH
2
)
n
,
JEOL AccuTOF instrument by using water, acetonitrile or
1
in which the acyl tail R (CO), the stereochemistry of Lys, and the methanol as solvents. Standard mass spectrometry measure-
spacer length n were varied.
ments were performed on a FINNIGAN LCQ Advantage Max by
using water, acetonitrile or methanol as solvents.
2
.1.3. High-performance liquid chromatography (HPLC).
2
. Materials and methods
High-performance liquid chromatography (HPLC) was carried
out on a Shimadzu LC-20AT HPLC system equipped with a UV-
2.1. General procedures
THF was puried by distillation under nitrogen from sodium/ Vis detector SPD-10AV and a fraction collector. Columns were
benzophenone, dry acetonitrile was obtained by distillation purchased from Dr Maisch GmbH. The compounds were puri-
under nitrogen from phosphorus pentoxide. Dry DMF was ed on a mg scale by using a semipreparative reversed-phase
purchased from Fluka and anhydrous pyridine from Aldrich. column. A 2 mL solution was injected in a column ReproSil
ꢀ
Other chemicals were purchased from Aldrich, Fluka or Acros 100C8, 5 m (250 ꢁ 10 mm) operating at 30 C. The detection
and used as received. Analytical thin layer chromatography wavelengths were xed at 254 and 215 nm. A gradient of water
(
TLC) was performed on Merck precoated silica gel 60 F-254 and acetonitrile both containing either 0.1% v/v HCl or TFA was
ꢂ1
plates (layer thickness 0.25 mm) and the compounds visualized used as mobile phase, with a ow rate of 4 mL min . HCl was
by ultraviolet (UV) irradiation at l ¼ 254 nm and/or l ¼ 366 nm used to ensure that the chloride was the only counterion of the
and by staining with phosphomolybdic acid reagent or KMnO
4
.
isolated compounds. In all cases the samples were prepared by
.1.1. Nuclear magnetic resonance (NMR). H NMR spectra dissolving the compound in a suitable mixture water–acetoni-
were recorded at 25 C on Varian Inova 400 or a Bruker DMX-300 trile and ltered on a nylon syringe lter (0.2 mm).
1
2
ꢀ
1
3
machines operating at 400 and 300 MHz, respectively. C NMR
spectra were recorded on a Bruker DMX-300 machine operating
1
2.2. Synthesis
at 75 MHz. H NMR chemical shis (d) are reported in parts per
million (ppm) relative to a residual proton peak of the solvent, d Details of the preparation of the target compound with n ¼ 6 are
4.79 for D OD, d ¼ 7.26 for CDCl , and d ¼ given here; the experimental details of the preparations of its
O, d ¼ 3.31 for CD
.50 for DMSO-d . Multiplicities are reported as: s (singlet), d enantiomer, as well as of a classical (non-gemini) analogue, the
¼
2
3
3
2
6
(
(
doublet), t (triplet), q (quartet), dd (doublet of doublets), ddd compounds with n ¼ 2, 3, 4, 5, and 8, with acyl tails varying in
doublet of doublet of doublets) or m (multiplet). Broad peaks length (C8, C10, C12, C14, C16, and C18) and unsaturation
are indicated by b. Coupling constants are reported as a J value (C18:1(Z), C18:1(E), C18:2(Z)), which were carried out analo-
in Hertz (Hz). The number of protons (n) for a given resonance gously to that of the compound with n ¼ 6, are given in the ESI.†
is indicated as nH, and is based on spectral integration values.
2 2 6
2.2.1. Synthesis of (Z-Lys(Boc)-NH) (CH ) . Z-Lys(Boc)-
1
3
C NMR chemical shis (d) are reported in ppm relative to a OH$DCHA (dicyclohexylammonium) salt (3.0016 g, 5.343
carbon peak of the solvent, d ¼ 49.0 for CD OD, d ¼ 77 for CDCl mmol) was suspended in ethyl acetate (50 mL) and a few mL of a
and d ¼ 40 for DMSO-d ; when the solvent was D O, sodium-3- 10% KHSO solution were added. The mixture was stirred for 15
trimethylsilylpropionate-2,2,3,3-d was added as internal stan- minutes until all starting material had dissolved. Subsequently,
dard, d ¼ 0.38. The signals of the protons on hydroxyl, amino extractions were carried out with H O and brine until a neutral
and carboxylic groups could not be observed due to the fast pH was reached. The organic layer was dried with Na SO and
3
3
6
2
4
4
2
2
4
exchange with acids present in traces in the deuterated solvents. ltered. The solvent was removed and the remaining oil was
1
3
The resolution of C spectra was increased when necessary by dried in vacuo, yielding the free amino acid as a white foamy
performing an exponential apodization of the FID.
substance (1.9159 g, 93.9%). This material was then dissolved in
This journal is © The Royal Society of Chemistry 2014
Soft Matter, 2014, 10, 5702–5714 | 5703

View Article Online
Soft Matter
Paper
1
ethyl acetate saturated with water, and HOBt$H
6
2
O (941.2 mg, ¼ 1101.67, 47.12%) R
f
¼ 0.61 (CHCl
3
–MeOH ¼ 9 : 1). H NMR
.146 mmol), 1,6-diaminohexane (270.0 mg, 2.324 mmol), and (300 MHz, CDCl ) d: 6.93 (br t, 2H, diamine: NH), 6.77 (br d, 2H,
3
EDC$HCl (1.154 g, 6.020 mmol) were added. A white precipitate lysine: a-NH), 5.34 (m, 4H, oleoyl: CH]CH), 4.79 (t, 2H, lysine:
was formed overnight. The suspension was extracted two times 3-NH), 4.47 (q, 2H, lysine: a-CH*), 3.35 (m, 2H, diamine: 1,6-
with a 10% aqueous Na
with an aqueous 10% KHSO
times with an aqueous 10% Na
brine, respectively. The precipitate was isolated by ltration and b-CHH); (m, 2H, lysine, b-CHH); (m, 4H, lysine: d-CH
dried in vacuo in a desiccator. The organic layer of the extraction oleoyl: b-CH ), 1.48 (t, 4H, diamine: 2,5-CH ), 1.43 (s, 18H, Boc:
was evaporated and checked with TLC using CHCl –MeOH ¼ CH ), 1.41–1.20 (m, 4H, lysine: g-CH ); (m, 40H, oleoyl: CH );
2
CO
3
solution, once with H
solution, once with H
CO solution, and once with 2.10 (br q, 8H, oleoyl: CH
2
O, two times CHH), 3.10 (m, 2H, diamine: 1,6-CHH); (m, 4H, lysine: 3-CH
2
),
),
4
2
O, two 2.31 (t, 0.65H, oleic acid: a-CH ), 2.17 (dt, 4H, oleoyl: a-CH
2
2
2
3
2
–CH]CH), 1.88–1.55 (m, 2H, lysine,
); (m, 4H,
2
2
2
3
3
2
2
5
: 1 (v/v) as an eluent; this fraction was discarded as traces of (m, 4H, diamine 3,4-CH ), 0.88 (t, 6H, oleoyl: CH ). LCQ (ESI)
2 3
+
both Z-Lys(Boc)-OH (ninhydrine positive, R ¼ 0.34) and HOBt calculated (C H N O ): 1100.9, found: 1101.5 (M + H) ,
f
64 120
6
8
+
(
UV positive, R
NH) (CH
was obtained as a white solid. Yield: 1.864 g (MW ¼
41.04, 95.4%), R
MHz, CDCl ) d: 7.33 (s, 10H, Z Ar), 6.51 (br s, 2H, diamine NH), mmol) was dissolved in 4 mL of CHCl
.83 (br s, 2H, lysine a-NH), 5.04 (q, 4H, Z CH ), 4.69 (br s, 2H, 10 mL of 2 M HCl in ethyl acetate. The reaction mixture was
f
¼ 0.18) could be detected. (Z-Lys(Boc)- 1123.7 (M + Na) .
2
2
)
6
2.2.3. Synthesis
of
(Oleoyl-Lys(H)–NH)
(CH (115.3 mg, 0.105
and added dropwise to
2 2 6
(CH ) $2HCl.
1
8
f
¼ 0.63 (CHCl
3
–MeOH ¼ 5 : 1). H NMR (300 Compound (Oleoyl-Lys(Boc)-NH)
2
2 6
)
3
3
5
2
lysine 3-NH), 4.15 (m, lysine a-CH*), 3.29 (m, 2H, diamine 1,6- stirred for 45 min and was regularly checked by TLC. Aer the
CHH), 3.21 (m, 2H, diamine 1,6-CHH), 3.07 (m, 4H, lysine 3- reaction was nished, the solvent was evaporated partly under
CH ), 1.79 (m, 2H, lysine b-CHH), 1.64 (s, 2H, lysine b-CHH); reduced pressure and a white precipitate appeared. Subse-
2
1
(
(
.53 (br s, 4H, lysine d-CH
m, 8H, diamine 2,3,4,5-CH
300 MHz, CDCl ) d: 172.2, 156.2, 136.2, 128.5, 128.0, 79,1, 66.9, (Oleoyl-Lys(H)–NH)
5.0, 39.9, 38.2, 32.1, 29.5, 28.4, 24.9, 22.6. LCQ (ESI) calculated in 98% yield; this was subjected to one more column chroma-
2
), 1.45 (s, 18H, Boc CH
3
), 1.29–1.35 quently, t-butanol was added and the mixture was evaporated
1
3
2
); (m, 4H, lysine g-CH
2
). C NMR (two times) to remove the excess of HCl. The initial product
(CH $2HCl was obtained as a white solid
3
2
2 6
)
5
+
+
(C
44
H
68
N
6
O
10): 840.5, found: 841.6 (M + H) , 863.5 (M + Na) , tography step to remove any residual oleic acid, yield: 238 mg
+
1
1703.9 (2M + Na) .
(MW ¼ 974.36, 93%) R ¼ 0.19 (CHCl –MeOH ¼ 5 : 1). H NMR
f
3
2
.2.2. Synthesis of (Oleoyl-Lys(Boc)-NH) (CH ) . (Z-Lys- (300 MHz, DMSO-d ) d: 7.78–7.94 (m, 10H, NH), 5.35 (m, 4H,
2 2 6 6
(
Boc)-NH) (CH ) (0.5014 g, 0.5962 mmol) was dissolved in 50 oleoyl CH]CH), 4.20 (q, 2H, lysine a-CH*), 3.05 (q, 4H, diamine
2 2 6
mL of DMF. The solution was ltered, collected in a round- 1,6-CH
bottomed ask and the palladium carbon catalyst (0.1 g) was 2.01 (m, 8H, oleoyl CH
added. Aer removal of all air from the round bottom ask, a 2,5-CH ); (m, 4H, lysine d-CH
hydrogen atmosphere was established using a balloon. The lysine b-CH ), 1.28–1.31 (m, 40H, oleoyl: CH
reaction was followed on TLC using BAW as an eluent and a 3,4-CH ); (m, 4H, lysine g-CH ), 0.89 (t, 6H, oleoyl CH
3
2
), 2.76 (q, 4H, lysine 3-CH
–CH]CH), 1.45–1.70 (m, 4H, diamine
); (m, 4H, oleoyl b-CH ); (m, 4H,
); (m, 4H, diamine
). LCQ
2 2
), 2.15 (t, 4H, oleoyl a-CH ),
2
2
2
2
2
2
2
2
+
ninhydrin solution as a staining agent. The starting material, (ESI) calculated (C H N O ): 900.8, found: 451.3 (M + 2H ),
5
1
100 6 4
+
the intermediate product with one Z protecting group le, and 901.6 (M + H ).
the nal product could be distinguished very clearly with Rf
values of 0.94, 0.74 and 0.46, respectively. When the hydroge-
nation was complete (typically aer 1 or 2 hours) the DMF
solution was ltered under suction over a P3 glass lter that was
2
.3. Gene transfection experiments
2.3.1. Cell culturing. HeLa T-rex cells were grown in T75
+
protected by Kieselguhr. To the DMF solution oleic acid (0.46 bottles in EMEM (Eagles minimal essential medium contain-
mL, 1.45 mmol), HOBt$H
2
O (0.736 g, 4.81 mmol), and EDC$HCl ing 0.5% (vol) PenStrep and 10% (vol) Fetal Calf Serum (FCS)
ꢀ
(0.690 g, 3.61 mmol) were added. The mixture was le stirring and 2 mM Glutamax-L) at 37 C under a 5% CO atmosphere
2
overnight and the progress of the reaction was monitored by and passed twice a week. Prior to transfection experiments, cells
TLC using BAW as an eluent and a ninhydrin solution for were washed, detached using trypsin and diluted 7.5-fold using
+
staining. The product and a sideproduct with only one oleoyl EMEM . Each well of a 6-well plate was then seeded with 2 mL of
tail appeared at R
only product remained the DMF was evaporated and dried erate for one day before transfection experiments.
under high vacuum. The residue was dissolved in chloroform 2.3.2. Preparation of the aggregates of the cationic lipids.
and extracted two times with a 10% aqueous Na CO solution, For the positive control, 50 mL of DDAB and 50 mL of DOPE
f
values of 0.94 and 0.67, respectively. When the cell suspension (ꢃ200 000 cells per well) and le to prolif-
2
3
two times with an aqueous 10% KHSO solution, two times with solutions in ethanol (6.6 mM and 13.2 mM, respectively) were
4
an aqueous 10% Na CO solution, and once with brine, mixed. Subsequently, this 100 mL ethanol solution was injected
2
3
respectively. Aer drying the organic layer over anhydrous into 1.9 mL of Milli-Q water and mixed by vortexing for 20 s at
sodium sulphate, ltering, solvent evaporation, and drying at ambient temperature. The aggregates of the gemini surfactants
an oil pump, the product was puried by column chromatog- used for transfection with the helper lipid DOPE were prepared
raphy (eluent: 10% methanol in chloroform), followed again by in exactly the same way as the positive control, replacing DDAB
solvent evaporation and drying at an oil pump. (Oleoyl-Lys(Boc)- with the gemini of interest. For studies without DOPE, the
2 2 6
NH) (CH ) was obtained as a white solid. Yield: 0.3095 g (MW helper lipid was simply le out.
5704 | Soft Matter, 2014, 10, 5702–5714
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Soft Matter
2
.3.3. Preparation of the transfection complexes. As could be tested exactly as they were used in the transfection
reporters for the transfection, plasmids (p-DNA) coding for experiments. To remove dust and other large particles, however,
either EGFP or b-galactosidase (b-gal) under a constitutive the transfection solutions had to be ltered prior to the DLS
promotor were employed. For each well of a six-well plate 0.5 mg measurement; the pore size of the ltration Millipore frits was
CMV b-gal plasmid and 0.5 mg EGFP plasmid were dissolved in 200 nm. The measurements were performed on a Malvern
ꢂ
2
00 mL EMEM (Eagles minimal essential medium containing Instruments Zetasizer Nano-S (ZEN 1600), equipped with a He–
0
.5% (vol) PenStrep and 2 mM Glutamax-L but no Fetal Calf Ne laser (633 nm, 4 mW) and an Avalanche photodiode detector
Serum). Transfection mixes as described above were then added at an angle of 173 . The data were processed with Dispersion
ꢀ
as described in the legends of Fig. 3–6, and the solution was Technology Soware (Malvern Instruments).
allowed to incubate for half an hour.
.3.4. Transfection using transfection complexes of dispersed in Milli-Q water, or phosphate buffer (pH 7), or
2.4.3. Small angle X-ray scattering. A surfactant was
2
cationic surfactants. The medium of the cells in the six-well phosphate/citrate buffer (pH 3) to a concentration in the 5 mM
ꢂ
plates was changed to 1 mL EMEM . Subsequently, the trans- range. Lipoplexes were prepared by adding the appropriate
fection complexes were added and the mixtures were then le to amount of salmon sperm DNA solution to get N/P ratios
ꢀ
incubate for 6 h at 37 C under a 5% CO atmosphere. Aer this between 1 and 7. SAXS experiments were conducted on the
2
+
period, the medium was removed and replaced by 2 mL EMEM . SAXS/WAXS station (BM26B) at the Dutch-Flemish beamlines
The cells were then allowed to proliferate for two days.
.3.5. Qualitative EGFP assay. To qualitatively assess Grenoble, France.
(DUBBLE) at the European Synchrotron Radiation Facility in
16,17
2
During the experiments the ESRF was
transfection efficiency, a vector containing an EGFP gene under running at currents between 200 and 140 mA, and the mono-
˚
a constitutive promoter was transfected. This enabled simple chromator was set at 1.24 A (10 keV). The typical photon ux at

uorescence microscopy as a quick identier of successful the sample position with a beam size of 0.3 mm ꢁ 0.4 mm was
9
transfections, giving preliminary results based on uorescence 10 photons per s. SAXS data of liposomes and lipoplexes in
brightness.
brass cells with mica windows were recorded with the gas
.3.6. Quantitative b-gal assay. Cells transfected with a multiwire 2-dimensional detector at a sample to detector
18
2
constitutively expressed b-gal-coding gene had their medium distance of 1.5 m. The background scattering due to solvent and
removed and were subsequently lysed by incubation with 200 mL empty cell was subtracted, and the spatial distortion and small
reporter lysis mix (RLM: 25 mM bicine, pH 7.5, 0.05% Tween-20 inhomogeneities of the gas-lled detector (at-eld correction)
and 0.05% Tween-80 in water) for 15 minutes under continuous were corrected for. Spatial calibration was performed with silver
19
shaking. The lysates were then scraped and 20 mL was trans- behenate with an estimated error margin of ꢄ0.5% in the
ferred to a perspex tube. Then 200 mL b-gal-mix (1% galacton, 5 observed periodicities.
mM MgSO , 100 mM phosphate buffer pH 8.0) was added to
4
each tube, and the mixtures were le to incubate for 20
3
. Results and discussion
minutes. Aer this, 300 mL Accelerator II mix was added and the
mixtures were incubated before the lysates were measured in a
luminometer for 10 seconds per tube.
3.1. Synthesis of the compounds (Oleoyl-Lys(H)–NH)
2(CH2)n
The synthesis of the class of compounds Ol-Lys-H-n with varying
n is represented here by the compound with n ¼ 6 as an
example. The exploratory pathway for the preparation of
2
.4. Physicochemical characterization
1
2.4.1. Electron microscopy.
Transmission electron (R (CO)-Lys(H)-NH) (CH ) (dashed arrows in Scheme 1), which
2
2 n
5
microscopy (TEM) was carried out on dispersions prepared by was used to prepare the compounds in our earlier report,
ꢀ
sonication of the pure lipids (without DOPE) in water at 55 C, involved acylation of the a-amino-group of lysine of which the
which were allowed to dry on a carbon–formvar grid and stained 3-amino-group was Boc-protected, followed by activation of the
with uranyl acetate. Because of the possible impact of the drying lysine carboxylic acid group with ethyl chloroformate, coupling
step in the sample preparation procedure for TEM, cryo-SEM to the diamine, and subsequent deprotection of the Boc group.
(
Scanning Electron Microscopy) was chosen for the investiga- A similar approach was used in the synthesis of C12-Lys-H-6 by
10
tion of the lipoplexes. Higher (approx. factor 40) lipid concen- others. Although this 3-step procedure gave the HCl salts of
trations had to be used in order to be able to nd good the compounds Ol-Lys-H-n (n ¼ 2, 4, 6, 8) in 10–25% overall
populations of aggregates instead of one or some aggregates. A yield, it had a number of drawbacks with regard to the diaste-
lipid solution of 6.70 mM in ethanol was prepared, of which 1 reomeric purity of the products. The rst is that a technical
mL was evaporated in a rotary evaporator, resulting in a lipid grade of oleic acid was used, which contains 15–20% of the

lm on the wall of the glass vessel. This was dispersed in 1 mL trans-isomer (elaidic acid) besides the cis-isomer (oleic acid). In
sterile MilliQ water under vortexing for 20 s while heating till the preparation of gemini surfactants, the presence of such
ꢀ
5
0–60 C.
isomers can lead to a mixture of 3 stereoisomers, viz. the
2.4.2. Dynamic light scattering. Dynamic light scattering product with both tails cis, the product with both tails trans, and
(DLS) is a much faster and less elaborate technique for the product with 1 cis and 1 trans tail. Another problem is the
screening the particle sizes than TEM or cryo-SEM, with no need use of the activated amino acid derivatives in combination with
for concentrated samples, so that the transfection solutions the basic conditions applied to prepare and couple them. Under
This journal is © The Royal Society of Chemistry 2014
Soft Matter, 2014, 10, 5702–5714 | 5705

View Article Online
Soft Matter
Paper
1
4
Scheme 1 General scheme for the preparation of (R (CO)-Lys(H)-NH)
2
(CH
2
)
n
ꢀ
. Dashed arrows refer to the published exploratory preparation,
1
ꢀ
the solid arrows refer to the present work. (i) 0.1 M NaOH in water/dioxane, 0 C; (b) R (CO)-succinimide; (ii) (a) EtOCOCl, Et
H N(CH ) NH ; (iii) 4 M HCl in EtOAc; (iv) EDC$HCl, HOBt/EtOAc, H O; (v) H , (Pd/C)/DMF; (vi) R -COOH, EDC$HCl, HOBt/DMF.
2 2 n 2 2 2
3
N, THF, ꢂ20 C; (b)
1
basic circumstances the amino acids can racemize via the according to the procedure depicted in Scheme 1, using the
ketene intermediate, or, in the case of the N-acyl amino acid, via appropriate carboxylic acid in step (vi).
20,21
the cyclic oxazolone intermediate,
in which the acidity of the
hydrogen atom at the chiral center is enhanced. This leads to a
mixture of products in which the two Lys residues either have
both the natural L-conguration, or the D-conguration, or a
3
.2. Transfection
3.2.1. Design of a protocol. The transfection abilities of the
product in which one lysine has the L-conguration and the synthesized cationic gemini surfactants were investigated by
other the D-conguration. This last product is a meso compound studying the expression of exogenous genes in cultured HeLa T-
23
if the molecule is otherwise (i.e. with respect to the acyl Rex cells in a protocol adapted from the literature. The
1
substituent R ) symmetrical.
In order to overcome these drawbacks a new synthetic route protein (EGFP) gave a quick qualitative indication of the
expression of the highly uorescent enhanced green uorescent
24
25
was designed in the present work (Scheme 1, solid arrows) in transfection activity (Fig. 2), that of beta-galactosidase (b-gal),
which the original problems were avoided. In this route, the which could be analyzed with an assay for the enzymatic
order of reactions was changed. As the lysine was coupled rst activity, gave a more accurate estimate of the transfection effi-
to the diamine via its carboxylic acid group (step iv), both its a- ciencies. Such results are at best semi-quantitative, in view of
and 3-amines had to be protected with orthogonal groups, so Z- the variation from experiment to experiment, and a positive
2
6,27
28
Lys(Boc)-OH was selected as the starting material. In the second control – DDAP/DOPE
step the Z group on the a-amine was selectively removed via later experiments – was always included.
hydrogenation (v) without affecting the Boc protection of the 3-
3.2.2. Optimization of the transfection parameters. The
in most experiments, Fugene in our
22
5
amine. Aer acylation of the unprotected a-amine (vi), the Boc most active transfection agent in our exploratory study, the
protected 3-amine could be deprotected under acidic conditions n ¼ 6 gemini surfactant Ol-Lys-H-6, was used to optimize a
(
iii). The large increase in the optical rotation measured for Ol- number of variables for the transfection of HeLa T-rex cells with
Lys-H-6, i.e. ꢂ14.0 (c ¼ 1.0, MeOH) compared to ꢂ7.8 (c ¼ 0.92, EGFP and b-gal vectors, viz. the ratio of positive charge in the
MeOH) for the product of the exploratory synthesis ((ref. 5), lipid and negative charge in the nucleotide, the so-called
Scheme 1, top/dashed arrows) is a strong indication that epi-
merization had been largely prevented in the present case. A
more detailed discussion of the synthetic approach is given in
the Materials and methods section and the ESI.†
In the same way as described above for Ol-Lys-H-6, the
analogous compounds with spacer lengths 2, 3, 4, 5, 7, and 8
instead of 6 were also prepared, starting from protected enan-
tiopure L-amino acids, and using the appropriate diamine in
1
step (iv). The analogues with varying acyl tails (R (CO)-Lys(H)-
1
NH)
2
(CH
2
)
6
(with R ¼ C18, stearoyl; C16, palmitoyl; C14,
myristoyl; C12, lauroyl; C10, caprinoyl; C8, capryloyl; C18:1 (E),
elaidoyl; and C18:2 (Z,Z), linoleoyl) were also prepared Fig. 2 (a) Fluorescence and (b) corresponding light microscope image
of the transfection of HeLa T-Rex cells by a vector coding for EGFP.
5706 | Soft Matter, 2014, 10, 5702–5714
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Soft Matter
ammonium/phosphate or N/P ratio, and the cationic lipid to Repeated experiments conrmed the subtle trend that the Ol-
helper lipid’ or L/H ratio. The effect of the N/P ratio, which Lys-H-n compounds with n ¼ 2 and in particular 3 are less
‘
has a large inuence on eventual toxicity of the transfection active, while those with n ¼ 7 and in particular 8 are more active
29
surfactant, was investigated with the same L/H ratio as that in than average. This result is an apparent contrast with that of the
5
the positive control, DDAB/DOPE, viz. 1/2. In Fig. 3a there is a exploratory study, which indicated a pronounced transfection
clear optimum corresponding to a N/P ratio of 1.27, implying optimum for n ¼ 6 compared to the compounds with n ¼ 4 and
that, assuming that all amines are protonated, there is a surplus n ¼ 8. It should be noted, however, that those results were
of positive charge in the lipid–DNA complex (lipoplex). This obtained on a different transfection system (luciferase in more
would mean that aer neutralizing all negatively charged easily transfectable CHO cells) without the helper lipid DOPE.
phosphates from the pDNA, some positive charge would remain We therefore investigated the transfection activity of the Ol-Lys-
to make the lipoplex positively charged and stick better to the H-n surfactants without DOPE, with Ol-Lys-H-6/DOPE as a
negatively charged surface of the cell. This N/P ratio was positive control and reference. The results in Fig. 4b show that,
subsequently used in the experiments in which the optimum L/ alone among this series, Ol-Lys-H-6 showed a remarkable
H ratio was explored (Fig. 3b), which was found to be at 1/2. The transfection activity without DOPE, in line with the exploratory
transfection efficiency using the L/H ratio of 1/1 in this experi- results on CHO cells. Compared to the transfections with DOPE,
ment was very low, almost as low as the transfection efficiency the reproducibility of the transfections by Ol-Lys-H-6 without
that was obtained without any transfection solution.
In a separate experiment (not shown) we investigated the aggregates, but this was not further investigated.
whether the same transfection efficiencies could be achieved 3.2.4. The effect of the chirality on the transfection activity.
DOPE was poor (not shown), possibly due to a lack of stability of
with less (viz. 0.5 or 0.25 mg) than 1 mg pDNA per well for a six- Ol-Lys-H-6 was synthesized starting not only from natural L-
well plate, but this turned out not be the case. Improved lysine, but also from its enantiomer D-lysine. The resulting Ol-D-
transfection efficiencies could be obtained (not shown) if the Lys-H-6 appeared to be a somewhat better transfection agent
transfection mix was not removed aer 6 h, but instead retained (Fig. 5a) than its L-enantiomer, in particular at an N/P ratio of
+
in the HeLa cells by just adding the EMEM medium. We did 1.27. This difference in biological activity between the enan-
not adapt our protocol in this direction, in order not to stimu- tiomers must be due to a difference in diastereoisomeric
late the transfection process beyond the count rate limitations interaction with the DNA, and with other chiral biomolecules
of the b-gal assay. Only in the case when Fugene was used as a such as membrane proteins and lipids. For another set of
reference and positive control, this procedure of retaining the gemini surfactants, it has been shown that variations in the
2
8
transfection mix was applied because the Fugene protocol
stereochemistry of chiral centres result in different interactions
30
demanded so. All other experiments in the transfection series, with DNA, and different transfection activities and toxicities.
i.e. those with DDAB/DOPE as a positive control, were per- Because we prefer the biodegradation products of our
formed using our standard protocol. We also found that
transfection in a phosphate buffered solution tended to be
somewhat higher than in sterile Milli-Q water, but we found it
more convenient to work with Milli-Q because of the better
stability of the transfection solutions in the refrigerator.
3.2.3. Transfection efficiencies for gemini surfactants with
different spacer lengths. With the transfection protocol estab-
lished and optimized for the Ol-Lys-H-6/DOPE transfection
mixture, the series of different gemini surfactants could be
tested for the inuence of the spacer length on the transfection
Fig. 4 Transfection efficiency of the series Ol-Lys-H-n (2 # n# 8) with
efficiencies. Fig. 4a shows that there is little variation in the 1 mg pDNA (50% b-gal) and 12 mL of transfection solution (N/P ¼ 1.27)
transfection efficiency as a function of the spacer length. per well. (a) with DOPE; PC, Fugene. (b) without DOPE; PC, Ol-Lys-H-
6/DOPE ¼ 1/2.
Fig. 3 Transfection efficiencies in a b-gal assay with Ol-Lys-H-6 and 1 Fig. 5 Transfection efficiencies for various amounts (N/P) of trans-
mg of pDNA (50% b-gal); PC, DDAB/DOPE positive control. (a) N/P ratio fection mixtures using 1 mg of pDNA (50% b-gal), L/H ratio 1/2, PC,
(
amount of transfection solution per well) varied at L/H ¼ 1/2. (b) L/H Fugene. (a) Effect of chirality: Ol-L-Lys-H-6 (left) vs. Ol-D-Lys-H-6
ratio varied at N/P ¼ 1.27 with naked DNA without transfection mix (right) in PBS. (b) Effect of gemini structure: Ol-Lys-H-6 (left) vs. half-
DNA) as a negative control. gemini Ol-Lys-NHEt (right).
(
This journal is © The Royal Society of Chemistry 2014
Soft Matter, 2014, 10, 5702–5714 | 5707

View Article Online
Soft Matter
Paper
transfection agent to be as natural as possible we continued to
work with L-lysine.
3.2.5. Gemini surfactant vs. half-gemini monomeric
surfactant. Although most cationic lipids used in transfection
5
studies discussed here so far have 2 alkyl tails, there are
occasional examples where lipids with a single alkyl chain gave
3
1–33
34
35
efficient transfections,
these lipids resembles our gemini surfactant because it
contains the amino acid ornithine (1 CH group less than
with or without DOPE. One of
Fig. 6 Transfection efficiency (b-gal assay) with L/H ¼ 1/2, 12 L
transfection solution per well (N/P ¼ 1.27), and 1 g of pDNA. (a) Effect
of length of acyl tail. PC, Ol-Lys-H-6/DOPE. (b) Effect of degree of
unsaturation of acyl tail. PC, DDAB/DOPE.
2
33
lysine) linked to oleyl alcohol by an ester bond. It should be
noted that the active single chain surfactants all have multiply
charged headgroups. It is remarkable that there are only few
cases where the efficiency of a cationic gemini surfactant has
1
shown). It has to be concluded that our (R (CO)-Lys(H)-
5,36
actually been compared to its monomeric counterpart.
We
2 2 6
NH) (CH ) gemini surfactants with saturated acyl groups
therefore also tested the monomeric surfactant analogue of Ol-
Lys-H-6 (lacking the spacer), the so-called half-gemini Ol-Lys-
NHEt, for its transfection efficiency. As can be seen from Fig. 5b,
Ol-Lys-H-6 (le) gave a nice transfection compared to the posi-
tive control at all amounts of transfection solutions, whereas
the half-gemini Ol-Lys-NHEt (right, monomeric surfactant
analogue) gave no transfection efficiency at all; omitting DOPE
and increasing the amount of Ol-Lys-NHEt (not shown) did not
improve this result. It can be concluded that the advantages of
1
R (CO) are not active in transfection.
5,36,50
Many biologically active gemini
and non-gemini
51
surfactants have one double bond in their hydrophobic C18
tails. It should be noted that, although oen not explicitly
mentioned, many synthetic procedures for the preparation of
C18:1 containing surfactants start with a technical grade of oleic
52
acid that contains 15–20% elaidic acid. With few exceptions,
the compound with the double bond in the cis-conguration
(C18:1 Z, ole(o)yl) gave more efficient transfection when it was
5
the gemini structure, as highlighted in the Introduction
section, also apply to Ol-Lys-H-6, and are essential for trans-
fection activity.
directly compared to the trans (elaidoyl) and the saturated
53,5
54,55
8
(
stearoyl) isomer. For the so-called SAINT lipids,
geminis,
12
and geminoids, there are strong indications that the presence
of cis-double bonds increases the uidity of the lipids in the
lipoplex, which in turn is an important factor in the efficiency of
the DNA release from the lipoplexes upon encounter with (intra)
cellular membranes. Because the phase transition temperature
of the surfactants and the uidity of the bilayer depend on the
unsaturation in the alkyl tails, it was considered to be of interest
to investigate what the effect of C18 tails with varying number
3.2.6. Effect of length and degree of unsaturation of the
acyl tail. The effects of variation in the hydrophobic domain of
cationic lipids on their physical properties have been thor-
31
oughly explored. The structure of the hydrophobic domain
determines the phase transition temperature and the stability
of the formed liposomes at room temperature. Moreover, lipid
membranes of high uidity tend to promote fusion of bilayers,
37,38
and hence increase the transfection efficiency.
Furthermore,
1
and geometry of the double bonds would be in the series R -Lys-
the hydrophobic domain also contributes to the protection of
the complexed DNA against nucleases and may contribute to
the endosomal escape. Cationic surfactants with branched
H-6. As can be seen in Fig. 6b, the gemini surfactants containing
oleoyl tails were by far superior to all other C18 tail containing
gemini surfactants, when formulated with DOPE. For C18-Lys-
H-6 no real optimum for the transfection efficiencies could be
detected (data not shown), mainly because the transfection
efficiencies were too low. A transfection experiment using the
elaidoyl containing gemini surfactants without the helper lipid
DOPE did not show any signicant transfection efficiency for
various concentrations (data not shown). For the linoleoyl tail
containing gemini surfactants no transfection efficiency was
39
35,40
tails or acetylenic tails
polyethylene glycol (PEG)
as well as with special sections like
41,42
43,44
or uorocarbon
regions in their
hydrophobic chains have been reported in the literature. While
the best hydrophobic tail for gemini surfactants is found to be
5
the ole(o)yl tail, C18:1 (Z), there are several reports on non-
gemini surfactants that give higher transfection efficiencies
45–48
with shorter saturated alkyl tails.
been shown to exhibit antimicrobial and haemolytic activities,
and to form interesting condensed structures with DNA in a
Moreover, C12-Lys-H-6 has
10
56,57
found at all, in spite of the fact that there are some reports
on transfection by cationic surfactants with more than one
double bond in an alkyl tail. The most important result of this
series of experiments is that it shows that for our gemini
surfactant, oleoyl tails are superior to elaidoyl tails, in line with
49
combined cryo-TEM and SAXS study. These literature results
prompted us to synthesize a series of gemini surfactants with
saturated alkyl chains of different lengths and to test their
transfection efficiencies using our general formulation and
transfection protocol. As can be seen from Fig. 6a, the trans-
fection efficiencies for the gemini surfactants with the saturated
tails were lower by 2 orders of magnitude at optimal concen-
trations, compared to Ol-Lys-H-6 as a reference and positive
control. Increasing the N/P ratio by adding more transfection
5
what was reported before for geminis in general, and with the
aforementioned effect of the cis double bond on the membrane

uidity.
3.3. Physical properties of the surfactants
mixture at equal amounts of pDNA per well did not improve the In order to gain more insight into the relation between trans-
transfection efficiency, nor did varying the L/H ratio (data not fection efficiency, as described in the previous section, on the
5708 | Soft Matter, 2014, 10, 5702–5714
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Soft Matter
one hand, and the molecular structure and aggregate formation
on the other, the gemini surfactants were studied by a number
of physical techniques, viz. by imaging (electron microscopy),
scattering (dynamic light), and diffraction (with synchrotron X-
rays).
3
.3.1. Electron microscopy. Interesting morphological
1
differences were observed with TEM within the R -Lys-H-6
family of compounds (Fig. 7). Flat structures, typical for bila-
6
yers, are found for the compounds with saturated tails (C18, Fig. 8 Cryo-SEM pictures of Ol-Lys-H-6, L/H ratio 1/2, at various
magnifications.
here represented as stearoyl or St, Fig. 7a; C14, Fig. 7b) and
mono-(E)-unsaturation (elaidoyl or El, Fig. 7c) where the bila-
yers have a tendency to roll up. Presumably, the alkyl tails of
these compounds can easily adopt a linear conformation, which
allows them to be packed tightly in a bilayer with little curva-
ture. In contrast, the compound with mono-(Z)-unsaturated
tails (oleoyl or Ol) formed vesicles (Fig. 7d), implying the
formation of more strongly curved bilayers.
ꢀ
Table 1 DLS results (diameter in nm, 25 C) for the series Ol-Lys-H-n
(6.7 mM in Milli-Q) with 2 molar equivalents of the helper lipid DOPE
With DOPE, L/H ¼ 1/2
Spacer length (n)
Z-average (nm)
PDI
Attenuator
For cryo-SEM, lipid dispersions of concentrations 40 times as
high as those used in the transfection experiments were
prepared. As can be seen from the micrographs of Ol-Lys-H-6
with DOPE in Fig. 8, large amounts of aggregates were formed.
Fig. 8B reveals that this particular sample was very polydisperse,
with particle sizes ranging from small sized 60–70 nm particles
to very large vesicles of some 350 nm. In this picture it can be
seen that some of the large aggregates have exploded in the
vacuum of the cryo-SEM, thereby revealing the water containing
lumen inside. This suggests that these large aggregates are big
unilamellar vesicles, and not small multilamellar vesicles, in
2
3
4
5
6
7
8
56.36
74.95
58.75
89.77
76.78
83.06
114.3
364
257
264
231
231
248
228
9
8
9
7
8
8
7
autocorrelation function, and the attenuator setting is a
measure for the signal intensity (range: 1–11 in kcounts). The
PDI values for the measurements on the mixtures with DOPE
indicate that all samples are very polydisperse, but are still
acceptable for a reliable measurement; the attenuator settings
for these results are also acceptable.
58
line with literature results.
.3.2. Dynamic light scattering (DLS). The results of the
3
DLS experiments of the Ol-Lys-H-n family are summarized in
Table 1. The Z-average is the hydrodynamic radius (diameter in
nm) of monodisperse spherical particles that correspond to the
measured scattering; the PDI is the polydispersity index of the
particles in the measured sample as calculated from the
The particle sizes measured for the mixtures of the surfac-
tants with the helper lipid DOPE (ratio 1/2) correlate very
roughly with the spacer lengths, increasing from 56 nm for Ol-
Lys-H-2 to 114 nm for Ol-Lys-H-8. Although it is not possible to
relate the transfection efficiency (Fig. 4a) to the particle size of
each surfactant, it is interesting to note that there is an
underlying trend in which it increases with spacer length and
particle size. Although the size measured by DLS is strictly a
hydrodynamic radius, the results for n ¼ 6 correlate very well
with those of the cryo-SEM experiments (Fig. 8), if we take into
account that the larger aggregates in the DLS experiments are
probably removed by the ltration step. The average particle
size of the Ol-Lys-H-6 aggregates increased with increasing
amounts of DOPE from 67 nm for a L/H ratio of 1/1 via 77 for L/
H 1/2 (Table 1) to 91 and 97 nm for L/H 1/3 and L/H 1/4,
respectively.
The DLS measurements on surfactants with different spacer
lengths without DOPE, for which the transfection results were
very poor, except for n ¼ 6 (Fig. 4b), suffered from a lack of
signal, as reected in the high values (11) for the attenuator
setting; the PDI values (0.6–1.0) are also too high, which indi-
cates that the values for the Z-average are based on a very poor
Fig. 7 TEM pictures with negative staining by uranyl acetate (a) St-Lys- t. The problems may be explained in two ways. Either the
H-6 (St ¼ stearoyl). (b) C14-Lys-H-6. (c) El-Lys-H-6 (El ¼ elaidoyl). (d) particles are very small at the concentrations used, like in the
Ol-Lys-H-6.
This journal is © The Royal Society of Chemistry 2014
Soft Matter, 2014, 10, 5702–5714 | 5709

View Article Online
Soft Matter
Paper
case of small micelles, and not visible, or the particles are very
large (>200 nm) and were removed by the ltration step. In any
case, the conclusion can be drawn that if no aggregates of a
detectable size are formed, no or little transfection activity is
measured, except for n ¼ 6.
3.3.3. Small angle X-ray scattering (SAXS). Small angle X-
ray scattering (SAXS) experiments were carried out on disper-
sions of the free surfactants and on lipoplexes with N/P ratios of
3/1 and 1/1. In a number of cases, the pH and temperature
dependences were also investigated. The results concerning the
spacer and the tail dependence are given in Table 2 and shown
in Fig. 9 and 10.
For the new set of diastereomerically pure surfactants with n
¼
6, the position of the diffraction peak for elaidoyl surfactants
4
˚
at 45 A is comparable to that found earlier for a set of surfac-
˚
tants with natural oleic acid incorporated (45 A), whereas that
Fig. 9 SAXS patterns for (a) liposomes and (b) lipoplexes (N/P 3)
showing tail and spacer dependences at 25 C. Numbers 2, 4, etc.
represent spacer length in Ol-Lys-H-n; 1/2 ¼ ‘half-gemini’ Ol-Lys-
˚
for (all-cis)-oleoyl surfactants is increased (50 A, Table 2, le); in
ꢀ
neither case were sharp diffraction patterns obtained (Fig. 9a).
The diffraction peaks are tentatively assigned to stacks of NHEt; El ¼ elaidoyl; St ¼ stearoyl.
˚
intercalated lipid bilayers (estimated molecular length 30 A) of
the liposomes of the free surfactants; the strong diffraction
peak at low q is taken to represent the rst order reection, q001
,
˚
(
n ¼ 2) and long (n ¼ 8, both 57 A) spacers. In most lipoplexes,
of a lamellar system with a periodicity d, or lamellar spacing,
corresponding to ¼ 2p/q001. For n ¼ 6 at neutral pH, the bilayer
the bilayer thickness increased compared to those of the lipo-
somes, which indicates that the lipoplex consists of (alter-
nating) lipid and DNA layers. For very short spacer lengths (n ¼ 2,
n ¼ 4) a hexagonal phase was found at neutral pH; this is
recognized by the occurrence of (110) and (200) reections in
addition to the (100) reection, from which the hexagonal or
columnar spacing a can be derived as 4p/O3q100. Also for n ¼ 6 a
hexagonal phase was found at pH 3 (Fig. 10b). These hexagonal
phases could consist of cylindrical micelles of lipid with
extended DNA molecules between them, or of an inverted
structure with DNA at the center of a cylinder of lipid molecules.
The amino groups of the lysine-based gemini surfactants
have to be protonated in order to give positively charged groups
that can have an electrostatic interaction with the negatively
thicknesses measured for the stearoyl and elaidoyl analogues
˚
(
39 and 45 A, respectively) are much lower than the bilayer
˚
thickness of the oleoyl surfactant (50 A). This points to a much
lower degree of surfactant intercalation in the bilayer structure
in the latter case, which correlates nicely with the less tight
packing of this surfactant as derived from the results of the
electron microscopy study (Fig. 7). For the oleoyl surfactants,
˚
59
the bilayer thicknesses for n ¼ 4 and n ¼ 5 (48 and 47 A,
8
respectively) are relatively small, which points to a large degree
of intercalation of the surfactant tails comparable to that of the
˚
elaidoyl surfactant with n ¼ 6 (45 A). Relatively large bilayer
thicknesses were found for the oleoyl surfactants with short
1
Table 2 SAXS results for liposomes and lipoplexes of R (CO)-Lys-H-n (d ¼ 2p/q001, lamellar spacing; a ¼ 4p/O3q100, hexagonal or columnar
ꢀ
spacing) at 25 C. Ol, C18:1(Z) tails; El, C18:1(E) tails; St, saturated C18 tails
a
b
Free surfactant
Lipoplex N/P 3
Lipoplex N/P 1
1
ꢂ1
ꢂ1
ꢂ1
˚
q (A
˚
d (A)
˚
q (A
˚
d (A)
˚
a (A)
˚
q (A
˚
d (A)
˚
a (A)
R
n
pH
)
)
)
Ol
Ol
Ol
Ol
Ol
Ol
Ol
2
4
5
6
6
6
½
8
6
6
6
6
7
7
7
9
7
3
7
7
7
9
7
3
0.110, 0.236
132
135
127
126
126
106
110
139
57
48
47
50
50
50
59
57
45
41
39
50
0.106, 0.187, 0.215
0.112, 0.196, 0.227
0.098, 0.112
118
118
0.112, 0.138
105
111
115
121
68
65
106
59
57
57
0.112, 0.197, 0.227
111
No peak
111
65
65
64
53
53
56
60
57
55
52
54
57
c
d
e
0.112, 0.196, 0.227
104
114
105
117
105
115
60
55
60
54
60
55
Olf
El
St
154
169
127
g
St
117
110
h
St
a
b
c
d
e
f
g
h
Fig. 9a. Fig. 9b. Fig. 10a. Fig. 10b. n ¼ 1/2 ‘half-gemini’ Ol-Lys-NHEt. Fig. 10c. Fig. 10d. Fig. 10e.
5710 | Soft Matter, 2014, 10, 5702–5714
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Soft Matter
ꢀ
Fig. 10 SAXS patterns for various N/P ratios including free surfactants (N/P N) at 25 C (a–e). (a) and (b) Ol-Lys-H-6 at pH 7 and 3, respectively;
c) and (d) C18-Lys-H-6 at pH 7 and 3, respectively; (e) El-Lys-H-6 at pH 7; (f) temperature dependence of C18-Lys-H-6 at pH 3.
(
a
charged phosphate groups in DNA; because the pK values of upon lowering the pH from neutral to 3 (Fig. 10c and d).
two groups within a molecule are different and both are shied Surprisingly, the increased charge on the gemini headgroup
to lower pH when the lipid molecules aggregate, the interaction due to the protonation of both amines results in a decreased
with DNA can be expected to be pH dependent in a pH range intercalation of the surfactant tails. The bilayer initially
that is relevant for the effectivity of the transfection, in partic- becomes even thicker when the temperature is increased to
ꢀ
˚
ular when this occurs by a mechanism of DNA condensation 37 C (from 50 to 57 A) but at higher temperatures decreases to
˚
ꢀ
followed by endocytosis. It has been reported earlier that the pH values characteristic of strong intercalation (37 A at 55 C,
˚
ꢀ
dependence of the complex of DNA and cationic lipids can Fig. 10f; 39 A at 70 C). The structures of the corresponding
result in transitions between lamellar and hexagonal packings liposomes were hardly affected by the temperature variation.
of the resulting lipoplex which can trigger the endosomal The temperature dependence of a number of lipoplexes with
8
escape, and the tendency for a surfactant to form the hexagonal sharp diffraction peaks was also investigated. The columnar
phase has been taken as an explanation for the efficiency of spacing of the Ol-Lys-H-2 lipoplex (N/P 3) varies subtly but
˚
ꢀ
some gemini-like surfactants to give good transfection without signicantly with temperature, starting at 68.1 A at 25 C to a
9
,60
˚
ꢀ
˚
helper lipids. In the present series of results, the aforemen- maximum value of 69.1 A at 37 C and then back to 67.1 A at
ꢀ
˚
ꢀ
tioned ndings of hexagonal phases for the lipoplexes of some 55 C and 66.6 A at 70 C; a similar observation was made for the
of the Ol-Lys-H-n gemini surfactants do not appear to correlate Ol-Lys-H-4 lipoplex (N/P 3; a values are 64.7, 65.3, 64.8, and
˚
well with the transfection activities in the absence of helper 62.5 A, respectively).
lipid; they are found for n ¼ 2, n ¼ 4, and n ¼ 6, of which only
n ¼ 6 is active without helper lipid (Fig. 5a). The results do make
4
. Conclusions
sense in the context of the proposed model for endosomal
escape when the pH is considered: the hexagonal phase found
for n ¼ 6 at pH 3 may be a good explanation for the efficiency of
this surfactant without helper lipid (Fig. 5b), whereas the
hexagonal phase found for n ¼ 2 and 4 at neutral pH apparently
is not an advantage for an efficient transfection.
The exploratory synthesis of lysine-based gemini surfactants of
the type (R (CO)-Lys(H)-NH) (CH ) , on which we reported
2 2 n
1
4
earlier, led to products that may have contained up to 10
stereoisomers, including two meso compounds and two pairs of
enantiomers, due to the use of technical oleic acid, combined
with a racemization-prone synthetic protocol. A new strategy
was therefore followed in the present study, starting from pure
It is of interest to note (Table 2) that pH changes do not affect
the Ol-Lys-H-6 liposomes (Fig. 10a and b) but that the bilayer
thickness of St-Lys-H-6 liposomes increases from 39 to 50 A˚
1
oleic acid. Lysine-based gemini surfactants of the type (R (CO)-
This journal is © The Royal Society of Chemistry 2014
Soft Matter, 2014, 10, 5702–5714 | 5711

View Article Online
Soft Matter
Paper
1
Lys(H)-NH)
2
(CH
2
)
n
with R ¼ oleoyl and n ¼ 2–8 were success-
Ol
Oleoyl
fully synthesized. Transfection studies showed that the gemini
surfactant with n ¼ 6 was the biologically most active
compound. Therefore, the corresponding enantiomeric
PC
PBS
Positive control
Phosphate buffered saline
pDNA plasmid DNA
compound starting from D-lysine was also prepared, as well as RLM
Reporter lysis mix
L-lysine derived analogues with n ¼ 6 containing hydrophobic RLU
Relative luminescence units
Small angle X-ray scattering
Scanning electron microscopy
Stearoyl
1
tails varying in length (R (CO) ¼ C8, C10, C12, C14, C16, and SAXS
C18) and degree of unsaturation (C18:1 E and Z, C18:2 (Z,Z)). SEM
The ‘half gemini’ (monomeric surfactant analogue) was St
prepared by incorporating aminoethane.
TFA
TLC
Z
Triuoroacetic acid
Thin layer chromatography
Carboxybenzyl
The potential of the gemini surfactants to transfer poly-
nucleotides across cell membranes was investigated by trans-
fection of HeLa cells with beta-galactosidase with and without
the presence of the helper lipid DOPE. Oleoyl (C18:1(Z)) was by
far the best hydrophobic tail for obtaining high biological
Acknowledgements
activity, whereas the effect of the lysine stereochemistry was less The authors thank the ENGEMS network for supporting the
pronounced. Only in the absence of helper lipid an optimum initial phase of this work, and the Dutch Research Council
spacer length (n ¼ 6) was observed. The gemini structure was (NWO) for facilitating the SAXS investigations at the DUBBLE
essential; the ‘half-gemini’ (monomeric surfactant analogue) beam line. The Nijmegen Molecular Biology department (Siebe
was less active by 2 orders of magnitude. The most active van Genesen, Prof. N. H. Lubsen) are acknowledged for trans-
surfactant (Oleoyl-Lys(H)-NH)
sizes in the range 60–350 nm. Its lipoplex with N/P ratio 1 funded by the Dutch Technology Foundation STW (M.D., grant
underwent transition from lamellar to hexagonal stw.06565). R.J.M.N acknowledges support from the Ministry of
2 2 6
(CH ) , formed liposomes with fection facilities and valuable discussions. The research was
a
a
morphology upon lowering the pH from 7 to 3 in the absence of Education, Culture and Science (Gravity program 024.001.035).
helper lipid, a morphological change which is consistent with
endosomal escape triggered by lowering of the pH.
Notes and references
1
F. M. Menger and C. A. Littau, J. Am. Chem. Soc., 1991, 113,
451–1452.
R. Zana, J. Colloid Interface Sci., 2002, 248, 203–220.
Abbreviations
1
2
BAW
b-gal
Boc
CMC
CMV
Butanol : acetic acid : water ¼ 4 : 1 : 1 (v/v/v)
3 A. Bajaj, P. Kondaiah and S. Bhattacharya, J. Med. Chem.,
008, 51, 2533–2540.
b-Galactosidase
2
t-Butyloxycarbonyl
Critical micelle concentration
Cytomegalovirus
4
5
M. J. Rosen, CHEMTECH, 1993, 23, 30–33.
A. J. Kirby, P. Camilleri, J. B. F. N. Engberts, M. C. Feiters,
R. J. M. Nolte, O. S ¨o derman, M. Bergsma, P. C. Bell,
M. L. Fielden, C. L. Garc ´ı a Rodr ´ı guez, P. Gu ´e dat,
A. Kremer, C. McGregor, C. Perrin, G. Ronsin and
M. C. P. van Eijk, Angew. Chem., Int. Ed., 2003, 42, 1448–1457.
P. J. J. A. Buynsters, C. L. Garc ´ı a Rodr ´ı guez, E. L. Willighagen,
N. A. J. M. Sommerdijk, A. Kremer, P. Camilleri,
M. C. Feiters, R. J. M. Nolte and B. Zwanenburg, Eur. J. Org.
Chem., 2002, 1397–1406.
DCHA Dicyclohexyl-ammonium
DDAB Didodecyl ammonium bromide
DMF
DLS
DNA
N,N-Dimethylformamide
Dynamic light scattering
Desoxyribonucleic acid
6
DOPE Dioleoylphosphatidyl ethanolamine
EDC
EGFP
El
Ethylene dimethylamino-propyl carbodiimide
Enhanced green uorescent protein
Elaidoyl
7
8
M. Mammen, S.-K. Choi and G. M. Whitesides, Angew.
Chem., Int. Ed., 1998, 37, 2754–2794.
P. C. Bell, M. Bergsma, I. P. Dolbnya, W. Bras, M. C. A. Stuart,
A. E. Rowan, M. C. Feiters and J. B. F. N. Engberts, J. Am.
Chem. Soc., 2003, 125, 1551–1558.
EM
Electron microscopy
EMEM Eagle's minimum essential medium
EMEM EMEM with FCS
+
FCS
FID
HOBt
LCQ-
ESI
Fetal calf serum
Free Induction Decay
1-Hydroxybenzotriazole
(Liquid chromatography) quadrupole electrospray
ionisation mass spectrometry
Molar lipid/helper lipid (DOPE) ratio
Molar lipid/phosphate ratio
9
L. P ´e rez, J. L. Torres, A. Manresa, C. Solans and M. R. Infante,
Langmuir, 1996, 12, 5296–5301.
10 A. Colomer, A. Pinazo, M. A. Manresa, M. P. Vinardell,
M. Mitjans, M. R. Infante and L. P ´e rez, J. Med. Chem.,
2011, 54, 989–1002.
11 A. Colomer, A. Pinazo, M. T. Garcia, M. Mitjans,
M. P. Vinardell, M. R. Infante, V. Mart ´ı nez and L. Per ´e z,
Langmuir, 2012, 28, 5900–5912.
L/H
L/P
mRNA messenger RNA
NMR
N/P
Nuclear magnetic resonance
Molar positively charged nitrogen/negatively charged
phosphate ratio
12 M. Damen, J. Aarbiou, S. F. M. van Dongen, R. M. Buijs-
Offerman, P. P. Spijkers, M. van den Heuvel, K. Kvashnina,
5712 | Soft Matter, 2014, 10, 5702–5714
This journal is © The Royal Society of Chemistry 2014

View Article Online
Paper
Soft Matter
R. J. M. Nolte, B. J. Scholte and M. C. Feiters, J. Controlled
Release, 2010, 145, 33–39.
Biochem. Biophys. Res. Commun., 2001, 281, 536–
543.
1
1
1
3 I. Koltover, T. Salditt, J. O. R ¨a dler and C. R. Sanya, Science, 36 C. McGregor, C. Perrin, M. Monck, P. Camilleri and
998, 281, 78–81. A. J. Kirby, J. Am. Chem. Soc., 2001, 123, 6215–6220.
4 G. Caracciolo, D. Pozzi, H. Amenitsch and R. Caminiti, 37 J. O. R ¨a dler, I. Koltover, T. Salditt and C. R. Sanya, Science,
Langmuir, 2005, 21, 11582–11587. 1997, 275, 810–814.
5 M. Mu n˜ oz-Ubeda, A. Rodrıguez-Pulido, A. Nogales, 38 R. Zantl, L. Baicu, F. Artzner, I. Sprenger, G. Rapp and
1
´
´
A. Mart ´ı n-Molina, E. Aicart and E. Junquera,
Biomacromolecules, 2010, 11, 3332–3340.
6 W. Bras, An SAXS/WAXS beamline at the ESRF and future
J. O. R ¨a dler, J. Phys. Chem. B, 1999, 103, 10300–10310.
39 M. E. Ferrari, D. Rusalov, J. Enas and C. J. Wheeler, Nucleic
Acids Res., 2002, 30, 1808–1816.
1
experiments, J. Macromol. Sci., Part B: Phys., 1998, 37, 557– 40 S. Fletcher, A. Ahmad, E. Perouzel, A. Heron, A. D. Miller and
65. M. R. Jorgensen, J. Med. Chem., 2006, 49, 349–357.
7 M. Borsboom, W. Bras, I. Cerjak, D. Detollenaere, 41 S. Bhattacharya and P. V. Dileep, Tetrahedron Lett., 1999, 40,
D. G. van Loon, P. Goedtkindt, M. Konijnenburg, 8167–8170.
P. Lassing, Y. K. Levine, B. Munneke, M. Oversluizen, 42 S. Bhattacharya and P. V. Dileep, Bioconjugate Chem., 2004,
5
1
R. van Tol and E. Vlieg, J. Synchrotron Radiat., 1998, 5,
18–520.
8 A. Gabriel and F. Dauvergne, Nucl. Instrum. Methods, 1982,
01, 223–230.
9 T. C. Huang, H. Toraya, T. N. Blanton and Y. Wu, J. Appl.
Crystallogr., 1993, 26, 180–184.
15, 508–519.
5
43 J. Gaucheron, C. Santaella and P. Vierling, Bioconjugate
Chem., 2001, 12, 114–128.
44 K. Fabio, C. Di Giorgio and P. Vierling, Biochim. Biophys.
Acta, 2005, 1724, 203–214.
45 R. P. Balasubramaniam, M. J. Bennett, A. M. Aberle,
J. G. Malone, M. H. Nantz and R. W. Malone, Gene Ther.,
1996, 3, 163–172.
1
1
2
2
0 S. Doonan and E. W. Abel, Peptides and proteins, Royal
Society of Chemistry, Great Britain, 2002, p. 37, ISBN
0
854046925, 9780854046928.
1 J. H. Jones and M. J. Witty, J. Chem. Soc., Chem. Commun.,
977, 125, 281–282.
2 M. Bergmann and L. Zervas, Ber. Dtsch. Chem. Ges., 1932, 65,
192–1201.
46 V. Floch, M. P. Audrezet, C. Guillaume, E. Gobin, J. G. Le
Bolch, J. C. Clement, J. J. Yaouanc, H. Des Abbayes,
B. Mercier, J. Leroy, J. F. Abgrall and C. Ferec, Biochim.
Biophys. Acta, 1998, 1371, 53–70.
47 H. S. Kim, J. Moon, K. S. Kim, M. M. Choi, J. E. Lee, Y. Heo,
D. H. Cho, D. O. Jang and Y. S. Park, Bioconjugate Chem.,
2004, 15, 1095–1101.
48 J. A. Heyes, D. Niculescu-Duvaz, R. G. Cooper and
C. J. Springer, J. Med. Chem., 2002, 45, 99–114.
49 A. L. Barr ´a n-Berd ´o n, M. Mu n˜ oz-Ubeda, C. Aicart-Ramos,
L. P ´e rez, M.-R. Infante, P. Castro-Hartmann, A. Mart ´ı n-
Molina, E. Aicarta and E. Junquera, So Matter, 2012, 8,
7368–7380.
2
2
2
1
1
3 L. Doerwald, S. T. van Genesen, C. Onnekink, L. Marin-
Vinader, F. de Lange, W. W. de Jong and N. H. Lubsen,
Cell. Mol. Life Sci., 2006, 63, 735–743.
4 G. H. Zhang, V. Gurtu and S. R. Kain, Biochem. Biophys. Res.
Commun., 1996, 227, 707–711.
5 J. Alam and J. L. Cook, Anal. Biochem., 1990, 188, 245–254.
6 M. J. Campbell, BioTechniques, 1995, 18, 1027–1032.
7 C. R. Dass, T. Walker and M. A. Burton, Drug Delivery, 2002,
2
2
2
2
9
, 11–18.
50 M. L. Fielden, C. Perrin, A. Kremer, M. Bergsma, M. C. Stuart,
P. Camilleri and J. B. F. N. Engberts, Eur. J. Biochem., 2001,
268, 1269–1279.
2
2
8 http://labs.mmg.pitt.edu/gjoerup/Fugene6.pdf.
9 M. Mu n˜ oz-Ubeda, S. K. Misra, A. L. Barr ´a n-Berd ´o n, C. Aicart-
Ramos, M. B. Sierra, J. Biswas, P. Kondaiah, E. Junquera, 51 D. Liu, T. Ren and X. Gao, Curr. Med. Chem., 2003, 10, 1005–
S. Bhattacharya and E. Aicart, J. Am. Chem. Soc., 2011, 133,
8014–18017.
0 S. Aleandri, M. G. Bonicelli, F. Bordi, S. Casciardi,
M. Diociaiuti, L. Giansanti, F. Leonelli, G. Mancini,
1013.
1
52 I. van der Woude, A. Wagenaar, A. A. P. Meekel, M. B. A. ter
Beest, M. H. J. Ruiters, J. B. F. N. Engberts and D. Hoekstra,
Proc. Natl. Acad. Sci. U. S. A., 1997, 94, 1160–1165.
3
G. Perrone and S. Sennato, So Matter, 2012, 8, 5904– 53 S. Obika, W. Yu, A. Shimoyame, T. Uneda, K. Miyashita,
5
915.
T. Doi and T. Imanishi, Bioorg. Med. Chem., 2001, 9, 245–
3
3
3
1 D.-F. Zhi, S.-B. Zhang, B. Wang, Y.-N. Zhao, B.-L. Yang and
S.-J. Yu, Bioconjugate Chem., 2010, 21, 563–577.
2 P. Pinnaduwage, L. Schmitt and L. Huang, Biochim. Biophys.
Acta, 1989, 985, 33–37.
254.
ˇ
´
54 J. Smisterova, A. Wagenaar, M. C. A. Stuart, E. Polushkin,
G. Ten Brinke, R. Hulst, J. B. F. N. Engberts and
D. Hoekstra, J. Biol. Chem., 2001, 276, 47615–47622.
3 F. H. Cameron, M. J. Moghaddam, V. J. Bender, 55 I. S. Zuhorn, V. Oberle, W. H. Visser, J. B. F. N. Engberts,
R. G. Whittaker, M. Mott and T. Lockett, Biochim. Biophys.
Acta, 1999, 1417, 37–50.
U. Bakowsky, E. Polushkin and D. Hoekstra, Biophys. J.,
2002, 83, 2096–2108.
3
3
4 F. Tang and J. A. Hughes, J. Controlled Release, 1999, 62, 345– 56 J. Heyes, L. Palmer, K. Bremner and I. MacLachlan,
58. J. Controlled Release, 2005, 107, 276–287.
5 M. Patel, E. Vivien, E. Hauchecorne, N. Oudrhiri, 57 B. Yu, S.-H. Hsu, C. Zhou, X. Wang, M. C. Terp, Y. Wu,
3
R. Ramasawmy, J.-P. Vigneron, P. Lehn and J.-M. Lehn,
L. Teng, Y. Mao, F. Wang, W. Xue, S. T. Jacob,
This journal is © The Royal Society of Chemistry 2014
Soft Matter, 2014, 10, 5702–5714 | 5713

View Article Online
Soft Matter
Paper
K. Ghoshal, R. J. Lee and L. J. Lee, Biomaterials, 2012, 33, 59 L. Wasungu, M. C. A. Stuart, M. Scarzello, J. B. F. N. Engberts
5
924–5934.
and D. Hoekstra, Biochim. Biophys. Acta, 2006, 1758, 1677–
5
8 M. Mu n˜ oz-Ubeda, A. Rodr ´ı guez-Pulido, A. Nogales,
1684.
O. Llorca, M. Quesada-P ´e rez, A. Mart ´ı n-Molina, E. Aicart 60 R. Koynova and B. Tenchov, So Matter, 2009, 5, 3187–
and E. Junquera, So Matter, 2011, 7, 5991–6004. 3200.
5714 | Soft Matter, 2014, 10, 5702–5714
This journal is © The Royal Society of Chemistry 2014