Novel cationic polyene glycol phospholipids as DNA transfer reagents - Lack of a structure-activity relationship due to uncontrolled self-assembling processes

Article abstract of DOI:10.1016/j.chemphyslip.2014.04.006

Cationic glycol phospholipids were synthesized introducing chromophoric, rigid polyenoic C_20:5 and C_30:9 chains next to saturated flexible alkyl chains of variable lengths C_6-20:0. Surface properties and liposome formation

Full text of DOI:10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
Chemistry and Physics of Lipids xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Chemistry and Physics of Lipids
journal homepage: www.elsevier.com/locate/chemphyslip
Q1
Q2
1
2
3
Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to
uncontrolled self-assembling processes
Christer L. Øpstada, Muhammad Zeeshana, Asma Zaidia, Hans-Richard Sliwkaa,∗
Vassilia Partalia, David G. Nicholsona, Chinmay Surveb, Mitchell A. Izowerb,
Natalia Bilchukb, Howard H. Loub, Philip. L. Leopoldb, Helge Larsenc,
Alexandra Liberskaa,d, Nada Abdul Khaliqued, Liji Rajud, Marcella Flintermand,
Emile Jubelid, Michael D. Pungentee
4
5
6
7
8
,
a Department of Chemistry, Norwegian University of Science and Technology (NTNU), 7491 Trondheim, Norway
b Department of Chemistry, Chemical Biology & Biomedical Engineering, Stevens Institute of Technology, Hoboken, NJ 07030, United States
c Department of Physics, University of Stavanger, 4036 Stavanger, Norway
Q3
9
10
11
12
d Research Division, Weill Cornell Medical College in Qatar, P.O. Box 24144, Doha, Qatar
e Pre-Medical Unit, Weill Cornell Medical College in Qatar, P.O. Box 24144, Doha, Qatar
13 Q4
14
1350
a r t i c l e i n f o
a b s t r a c t
16
17
18
19
20
21
Article history:
Cationic glycol phospholipids were synthesized introducing chromophoric, rigid polyenoic C20:5 and C30:9
chains next to saturated flexible alkyl chains of variable lengths C6-20:0. Surface properties and liposome
formation of the amphiphilic compounds were determined, the properties of liposome/DNA complexes
(lipoplexes) were established using three formulations (no co-lipid, DOPE as a co-lipid, or cholesterol as
a co-lipid), and the microstructure of the best transfecting compounds inspected using small angle X-ray
diffraction to explore details of the partially ordered structures of the systems that constitute the series.
Transfection and cytotoxicity of the lipoplexes were evaluated by DNA delivery to Chinese hamster ovary
(CHO-K1) cells using the cationic glycerol phospholipid 1,2-dioleoyl-sn-glycero-3-ethylphosphocholine
(EPC) as a reference compound.
The uncontrollable self-association of the molecules in water resulted in aggregates and liposomes of
quite different sizes without a structure–property relationship. Likewise, adding DNA to the liposomes
gave rise to unpredictable sized lipoplexes, which, again, transfected without a structure–activity rela-
tionship. Nevertheless, one compound among the novel lipids (C30:9 chain paired with a C20:0 chain)
exhibited comparable transfection efficiency and toxicity to the control cationic lipid EPC. Thus, the pres-
ence of a rigid polyene chain in this best performing achiral glycol lipid did not have an influence on
transfection compared with the chiral glycerolipid reference ethyl phosphocholine EPC with two flexible
saturated C14 chains.
Received 3 February 2014
Received in revised form 31 March 2014
Accepted 1 April 2014
Available online xxx
22
23
24
25
26
27
28
29
Keywords:
Aggregation
Amphiphiles
Carotenoids
Cations
DNA
Lipids
© 2014 Elsevier Ireland Ltd. All rights reserved.
31
1. Introduction
32
33
34
35
36
37
Cationic lipids have been used as nucleic acid (NA) carriers to eukaryotic cells for more than 25 years (Felgner et al., 1987). Nevertheless,
gene transport with these lipids is still in a trial and error phase, as illustrated in a study wherein 1200 tested compounds revealed that only
65 (5%) delivered NA into cells as well as or better than Lipofectamine (Chu et al., 2009), a commonly applied gene delivery formulation
(Goldberg, 2008, p. 35). The various biological and chemical variables have so far prevented establishing an unambiguous structure–activity
connection (Koynova and Tenchov, 2010; Zhi et al., 2010, 2013). Even so, about 6% of all clinical gene therapy trials are based on lipoplexes
prepared with cationic phospholipids (Ginn et al., 2013). Most glycerolipid gene carriers, including EPC, contain flexible saturated or
∗
Corresponding author. Tel.: +47 73 59 5600.
E-mail address: richard.sliwka@ntnu.no (H.-R. Sliwka).
Q5
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006
0009-3084/© 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
2
Q89
Fig. 1. C20-n and C30-n cationic phosphocholine lipids composed of a “glycol scaffold” (green color), a hydrophobic polyenoic chromophore, either “C20:5” (orange color)
or “C30:9” (red color), a hydrophopic alkyl chain and a “hydrophilic head group” (blue color). (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of the article.)
38
39
40
41
42
43
44
45
46
47
48
mono-unsaturated chains. Rigidity has been introduced with steroids, diphenylethyne and triazine dendrimers (Merkel et al., 2010; Scheule
et al., 1998; Zhi et al., 2010). To our knowledge, no precedence for the interplay of polyunsaturated-stiff and saturated-flexible chains in
NA carrying phospholipids has yet been specified (Zhi et al., 2010), although this association may be important for adjusting morphology
changes along with phase state transitions (cylindrical shape → cone shape = lamellar phase → inverted hexagonal phase) (Adami et al.,
2011). Hexagonal phases are either contradictorily considered important or irrelevant for gene transfer (Koynova, 2010). Here, we report
on the synthesis of 11 new amphiphilic glycol lipids paired with polyene chains and stepwise increased saturated chains (Fig. 1). The
synthesis of the new phospholipids is based on previous connections of polyenes with phosphate groups (Foss et al., 2003, 2004, 2005a,b;
Popplewell et al., 2012; Sliwka, 1997, 1999).
A characterization of physical, chemical, and biological properties of the lipids was subsequently performed. The surface properties of
these new compounds were compared and their self-assembly into liposomes and to mixed liposomes with (R)-1,2-dioleoyl-glycero-3-
phosphoethanolamine (DOPE) and (−)-cholesterol (Fig. 2) was examined. EPC, a standard reference cationic lipid (Fig. 2) was included
Fig. 2. Structure of dioleoyl (C18:1) zwitterionic (R)-DOPE, reference dimyristoyl (C14:0) cationic (R)-EPC and (−)-cholesterol.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
3
49
50
51
52
53
54
55
for comparison with the novel polyene cationic lipids. Furthermore, lipoplex formation was assessed following combination of DNA
with the liposomes. The size and polydispersity of liposomes and lipoplexes were determined, and the ability of the lipids to deliver a
DNA cargo into cells as well as the cytotoxicity of the compounds were measured. Finally, the lipid phase for three of the lipoplexes
was resolved using small angle X-ray diffraction. The physical and biological data associated with the novel lipids were compared on
the basis of molecular structure, physical properties, DNA-transfection efficiency, and cytotoxicity. This investigation expands previous
structure–activity relationship studies varying other constitutional parts of cationic phospholipids (Floch et al., 2000; Kedika and Patri,
2011; Koynova and Tenchov, 2009; Niculescu-Duvaz et al., 2003).
56
57
2. Materials and methods
2.1. Materials
58
59
60
61
62
63
64
65
66
67
68
Ethyl ␤-apo-8ꢀ-carotenoate and retinoic acid was a generous gift from Dr. H. Ernst, BASF AG, Germany. Control cationic lipid 1,2-dioleoyl-
sn-glycero-3-ethylphosphocholine (EPC) and neutral co-lipid 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) were delivered by
Avanti Polar Lipids (Alabaster, AL). Chinese hamster ovary (CHO-K1) cells were obtained from the American Type Culture Collection (ATCC,
Manassas, VA, USA). RPMI 1640 medium was purchased from Invitrogen (Grand Island, NY, USA), FBS and antibiotic antimitotic solution
containing penicillin, streptomycin and amphotericin B were purchased from Sigma–Aldrich Chemical Co. (St. Louis, MO, USA). Plasmid DNA
containing the ␤-galactosidase gene, pCMV␤lacZnls12co was from Marker Gene Technologies, Inc. (Oregon, USA). Cell Titer 96®AQueous One
Solution Cell Proliferation Assay and Beta Glo® Assay System were purchased from Promega (Madison, WI, USA), ␤-galactosidase grade VIII
enzyme (0.5 mU/ng) from Escherichia coli was purchased from Sigma–Aldrich Chemical Co. (St. Louis, MO, USA), DNase I (2000 units/mL) was
purchased from New England Biolabs® Inc. (Ipswich, MA) and BCA Protein Quantitation Assay was purchased from Pierce Biotechnology
(Thermo Fisher Scientific, Rockford, IL, USA). 10× Tris–Borate EDTA (TBE) powder was purchased from Invitrogen (Grand Island, NY, USA).
Unless otherwise stated, all other chemicals were delivered by Sigma–Aldrich Chemical Co. (St. Louis, MO, USA).
69
2.2. Physical methods
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
NMR spectra (1H and 13C) were acquired on a Bruker Avance DPX 400 MHz and Bruker Avance 600 MHz with CDCl3 unless otherwise
stated. UV–vis spectra were recorded in CH2Cl2 using a Single Beam Thermo Spectronic, Helios ␥. Mass spectra data were acquired on
a MAT 95XL TermoQuest Finnigan mass spectrometer equipped with an electron ionization or electrospray ionization resource. Flash
column chromatography (flash-CC) was carried out with silica gel (Woelm Pharma, 60 mesh) or neutral alumina (II–III Brochmann activity,
EcoChrom, 100–150 mesh). Surface tension was determined using a Wilhelmy (Pt) plate on a Krüss Tensiometer K100. The aggregation
concentration cag was found by dissolving the compounds in the indicated solvents. H2O was added in 100 ␮l amounts and monitored VIS-
spectroscopically for aggregate formation. Inversely, the compounds were dispersed in H2O and organic solvent was added until disruption
of the aggregates.
The small angle X-ray diffraction experiments were performed at the European Synchrotron Radiation Facility (ESRF), Grenoble, France,
on the bending magnet BM29 BioSAXS beam line. BM29 is equipped with a double multilayer monochromator (energy band pass ∼ 10−2
)
and 4 mrad toroidal mirror 1.1 m long. The experimental hutch is equipped with a marble table housing the modular-length flight tube,
2D detector (Pilatus 1M) and a sample handling equipment (automated sample changer). The sample-to-detector distance was 2.8 m.
Samples were prepared in 1 mL Eppendorf tubes, and loaded into the 1.8 mm diameter quartz capillary sample cell by the automated
sample changer. Data collection was performed in 16-bunch mode at an energy of 12.5 keV with an exposure time of 2 s per frame using
the dedicated beam line software BsxCuBE. The small angle X-ray diffraction curves were obtained by integrating and averaging of 20
2D images. Subsequent processing (such as background subtraction and scaling) and analysis were performed using the ATSAS-software
package (Konarev et al., 2003) and MATHEMATICA (Mathematica, 2010).
87
2.3. Biological methods
88
Preparation of lipid stock solutions. All lipids were initially dissolved in CH2Cl2 in round bottom flasks, followed by evaporation of the
organic solvent under reduced pressure resulting in thin films. Ethanol was then added to achieve stock solutions of 1 mM. Care was taken
to protect the lipids from air and ambient light. All stock solutions were stored in amber vials under an inert atmosphere (N2 or Ar) and
kept at −80 ◦C.
89
90
91
92
Liposome preparation. Liposomes from the new cationic lipids as well as EPC were obtained by hydrating thin lipid films. An overall 3:2
molar ratio of total cationic lipid to co-lipid, either DOPE or cholesterol, in ethanolic solutions were prepared separately and evaporated
under reduced pressure to generate thin films. The lipid films were hydrated with a known amount of sterile water to give 2 mM final
hydrated stock solutions, which were stored overnight at 4 ◦C. Before use, the hydrated stocks were warmed to 37 ◦C and sonicated for
30 min.
93
94
95
96
97
Preparation of lipoplexes (lipid/DNA complexes). Lipoplexes of concentrations 0.081 mM, 0.243 mM, 0.486 mM, 0.81 mM and 1.62 mM,
corresponding to the N/P (+/−) molar charge ratios of 0.5:1, 1.5:1, 3:1, 5.0:1 and 10.0:1, respectively, were prepared from the 2 mM liposome
stocks. OPTI-MEM cell culture medium (57.6 ␮L) and DNA in E-ToxateTM Water, (14.4 ␮L; 250 ng/␮L) were first combined, followed by the
addition of an equal volume of corresponding liposome (72 ␮L) to this and mixed. These lipoplex formulations were incubated at 22 ◦C for
30 min.
98
99
100
101
102
103
104
105
106
107
Liposome and lipoplex sizing. The hydrodynamic diameters dH of liposomes and lipoplexes were measured by dynamic light scattering
(DLS, Malvern Zetasizer APS, Malvern, Worcestershire, UK) at 25 ◦C with a detection angle of 90◦.
Gel retardation assays of lipoplexes. To 20 ␮L of the lipoplexes, 2 ␮L of the gel loading dye (6×) was added and mixed by pipetting.
Eighteen microliters of each sample was then loaded onto a 1% agarose gel impregnated with ethidium bromide and run at 105 V for 1 h in
1× TBE buffer. The migration of DNA complexed with the cationic lipids was impeded in the electric field. The DNA bands were observed
using an ultraviolet transilluminator.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
4
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
DNase I degradation assays of lipoplexes. The lipoplexes (20 ␮L) were incubated with DNase I (2 units) at 37 ◦C for 1 h. After incubation,
5% SDS (4 ␮L) was added and incubated for a further 30 min, followed by 2 ␮L of gel loading dye (6×). Each lipoplex sample (18 ␮L) was
then loaded onto a 1% agarose gel impregnated with ethidium bromide and run at 105 V for 1 h in 1× TBE buffer. The pDNA bands were
observed using an ultraviolet transilluminator.
Cell culture and transfection. CHO-K1 cells were grown in RPMI media supplemented with 10% fetal calf serum and 100 U/mL of peni-
cillin/streptomycin and 0.25 ␮g/mL amphotericin B. Cells were seeded 48 h before transfection onto opaque and transparent 96-well plate
at a density of 104 cells per well and incubated at 37 ◦C in presence of 5% CO2 atmosphere. Cells were grown to 80% confluence before being
washed with 1x PBS and incubated with lipoplexes containing 3.6 ␮g of plasmid DNA in a volume of 45 ␮L in triplicate for 4 h at 37 ◦C in
the presence of 5% CO2 atmosphere. Complexes were then removed and the cells washed with 1× PBS before adding 100 ␮L of complete
RPMI media. Cells were left to incubate for an additional 44 h. Following the incubation, transfection and cytotoxicity assays were carried
out according to the below mentioned protocols.
ˇ-Galactosidase assay and protein assay. Cells grown in an opaque 96-well plate were evaluated for ␤-galactosidase activity 48 h after
transfection using a Beta-Glo® Assay System (Promega) according to the manufacturer’s instructions. Luminescence was determined on a
Perkin Elmer Precisely Wallac Envision 2104 Multilabel Plate reader (Perkin Elmer, Waltham, MA). ␤-Galactosidase activity was expressed
as relative light units produced by the luminescence of luciferin, which was normalized for protein content. Total protein content was
measured using Pierce® BCA Protein Assay (Pierce Biotechnology, Rockford, IL) according to the manufacturer’s instructions. A calibration
curve obtained from a bovine serum albumin standard solution was used to determine cellular protein content per well.
Cytotoxicity assay. The cytotoxicity associated with the lipoplex formulations at N/P ratios ranging from 0.5:1 to 10:1 was evaluated
using the MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) assay. Forty-eight hours
after the application of lipoplexes, CHO-K1 cells in the transparent 96-well plates were washed with 1× PBS, 50 ␮L of DMEM (phenol
red-free media) and evaluated for cell viability using the CellTiter96® Aqueous One Solution Cell Proliferation Assay (Promega) according
to the manufacturer’s instructions. The absorbance of converted dye, which correlates with the number of viable cells, was measured at
492 nm using a Victor Envision high throughput plate reader. The percentage of viable cells was calculated as the absorbance ratio of
treated to untreated cells.
132
133
3. Synthesis (Scheme 1)
3.1. Synthesis of (2-bromoethyl)(hexyl)(2-hydroxyethyl)phosphate (4-6)
O
1´
Br
2´
2
O
P
O
HO
4´´
2´´
6´´
1
O
1´´
5´´
3´´
134
135
136
137
138
139
140
2-Chloro-1,3,2-dioxaphospholan (3) (1.25 equiv., 2.00 g, 15.8 mmol) was dissolved in dry CH2Cl2 (50 mL) and cooled on ice. Dry triethy-
lamine (1.5 equiv., 1.94 g, 19.2 mmol) and 1-hexanol (1.31 g, 12.8 mmol) were introduced drop-wise and the mixture refluxed under N2 for
16 h. After cooling to −20 ◦C Br2 was added until the solution became permanent slight yellow. Dry triethylamine (2 mL) and ethyleneg-
lycol (1) (5 mL) were added and the mixture refluxed for 12 h. Extraction of the mixture with water (3× 50 mL), drying with anhydrous
Na2SO4, and vacuum concentration gave a residue, which, after flash-CC on silica with hexane/acetone 1:1 (v/v), gave 4-6 (1.38 g, 32%).
TLC (toluene/acetone/MeOH, 6:1:1, v/v): Rf = 0.42. HRMS: C10H22BrO5P calcd. 333.0461 (M+H), found 333.0477. 1H NMR, 13C NMR and 31
P
NMR in Supplemental Information.
141
3.2. Synthesis of (2-bromoethyl)(dodecyl)(2-hydroxyethyl)phosphate (4-12)
O
1´
Br
2´
2
O
P
O
HO
6´´
4´´
10´´
2´´
8´´
12´´
1
O
1´´
5´´
3´´
7´´
9´´
11´´
142
143
144
Phosphate triester 4-12 was obtained (250 mg, 5%) from 1-dodecanol (2.39 g 12.8 mmol) as described for 4-6. TLC
(toluene/acetone/MeOH 6:1:1, v/v): Rf = 0.38. HRMS: C16H34BrO5P calcd. 417.1400 (M+H), found 417.1420. 1H NMR, 13C NMR and 31P
NMR in Supplemental Information.
145
3.3. Synthesis of (2-bromoethyl)(tetradecyl)(2-hydroxyethyl)phosphate (4-14)
O
1´
Br
2´
2
O
P
O
HO
12´´
6´´
4´´
10´´
2´´
8´´
14´´
1
O
1´´
5´´
3´´
7´´
9´´
11´´
13´´
146
147
148
Phosphate triester 4-14 was obtained (1.22 g, 21%) from 1-tetradecanol (2.74 g, 12.8 mmol) as described for 4-6. TLC
(toluene/acetone/MeOH 6:1:1, v/v): Rf = 0.45. HRMS: C18H38BrO5P calcd. 445.1713 (M+H), found 445.1729. 1H NMR, 13C NMR and 31P
NMR in Supplemental Information.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
5
Scheme 1. Synthesis of cationic polyene glycol lipids C20-n and C30-n.
149
3.4. Synthesis of 2-bromoethyl dichlorophosphate (8)
150
151
152
153
Fresh distilled 2-bromoethanol (7) (1.05 equiv., 6.08 g, 50 mmol) dissolved in dry CH2Cl2 (5 mL) was added drop-wise during 1 h to an
ice-cooled solution of fresh distilled POCl3 (6) (7.35 g, 48 mmol) dissolved in dry CH2Cl2 (10 mL). The reaction mixture was refluxed for
5 h. Vacuum-distillation (b.p. 66–69 ◦C, 0.4 mbar) gave 8 (10.4 g, 90%). HRMS: calcd. for C2H4BrCl2O2P 241.8458 (M−1); found 241.8459.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
154
3.5. Synthesis of (2-bromoethyl)(hexadecyl)(2-hydroxyethyl)phosphate (4-16)
155
156
157
158
159
160
161
2-Bromoethyl dichlorophosphate (8) (1.49 equiv., 2.97 g; 0.0122 mol) was dissolved in anhydrous CH2Cl2 (20 mL) and cooled to 0 ◦C.
Anhydrous triethylamine (1.98 equiv.; 1.642 g; 0.016 mol) was dissolved in anhydrous CH2Cl2 (10 mL) and drop wise added to the solution.
1-Hexadecanol (1 equiv., 2 g, 0.0082 mol) dissolved in anhydrous CH2Cl2 (20 mL) was drop wise added and the resulting mixture was
refluxed for 5 h. Ethylene glycol (1) (2.19 equiv., 1.11 g; 0.0082 mol) was added and the mixture refluxed for 22 h, extracted with H2O,
dried over Na2SO4 and concentrated in vacuo. Purification by flash-CC on silica with hexane:acetone 1:1 (v/v) gave 4-16 (1.51 g, 39%). TLC
(toluene/acetone/MeOH, 6:1:1, v/v): Rf = 0.47. HRMS: C20H42BrO5P calcd. 473.2024 (M+H), found 473.2024. 1H NMR, 13C NMR and 31P NMR
in Supplemental Information.
162
3.6. Synthesis of (2-bromoethyl)(octadecyl)(2-hydroxyethyl)phosphate (4-18)
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
6
163
164
165
Phosphate triester 4-18 was obtained (1.87 g, 34%) from 1-octadecanol (3.00 g, 11.0 mmol) as described for 4-16. TLC
(toluene/acetone/MeOH, 6:1:1, v/v): Rf = 0.44. HRMS: C22H46BrO5P calcd. 501.2333 (M+H), found 501.2334. 1H NMR, 13C NMR and 31P
NMR in Supplemental Information.
166
3.7. Synthesis of (2-bromoethyl)(icosyl)(2-hydroxyethyl)phosphate (4-20)
167
168
Phosphate triester 4-20 was obtained (44 g, 21%) from 1-icosanol (3.82 g, 12.819 mmol) as described for 4-6. TLC (toluene/acetone/MeOH
6:1:1, v/v): Rf = 0.47. HRMS: C24H50BrO5P calcd. 528.2579, found 528.260. 1H NMR, 13C NMR and 31P NMR in Supplemental Information.
169
3.8. Synthesis of 2-((2-bromoethoxy)(hexyloxy)phosphoryl)ethoxy-retinoate (10-6)
17
16
O
O
P
1´
19
2
18
Br
4´´
11
7
b
9
13
12
O
O
6
2´
1
15
O
8
10
14
2´´
a
6´´
O
5
3
1´´
3´´
5´´
20
4
170
171
172
173
174
175
176
Retinoic acid (9) (200 mg, 0.666 mmol), 4-6 (266 mg, 0.799 mol), chlorotripyrrolidinophosphonium hexafluorophosphate (PyCloP,
1.25 equiv., 349 mg, 0.868 mmol), N-ethyl diisopropylamine (DIEA, 0.65 equiv., 58 mg, 0.451 mmol), and DMAP (1.25 equiv., 106 mg,
0.868 mmol) were dissolved in dry CH2Cl2 (50 mL) and the mixture was refluxed under N2 for 24 h. Extraction of the mixture with water
(2× 50 mL), aqueous HBr (0.1 M, 2× 50 mL), and water (2× 50 mL), drying over anhydrous Na2SO4 and concentration gave a residue, which,
after flash-CC on silica with a toluene-acetone gradient, gave 10-6 (377 mg, 92%). TLC (toluene/acetone/MeOH 6/1/1, v/v): Rf = 0.68. UV/vis
(CH2Cl2): ꢀmax = 370 nm. HRMS: C30H48BrO6P calcd. 615.2445 (M+H), found 615.2467. 1H NMR, 13C NMR and 31P NMR in Supplemental
Information.
177
3.9. Synthesis of 2-((choline)(hexyloxy)phosphoryloxy)ethoxy-retinoate (C20-6)
4´
5´
O
O
P
Br
1´
N
b
O
O
6´
2´
O
4´´
2´´
a
6´´
O
1´´
3´´
5´´
178
179
180
181
10-6 (358 mg, 0.582 mmol), was dissolved in CHCl3/iPrOH/DMF (3/5/5, v/v, 50 mL), NMe3 (45% in water, 10 mL) was added and the
mixture stirred at room temperature under N2 for 4 days. Purification by flash-CC gave C20-6 (288 mg, 74%). TLC (CHCl3/MeOH/H2O
40:50:10, v/v): Rf = 0.28. UV/vis (CH2Cl2): ꢀmax = 368 nm. HRMS: C33H57NO6P calcd. 594.3924 (M+), found 594.3926. 1H NMR, 13C NMR and
31P NMR in Supplemental Information.
182
3.10. Synthesis of 2-((2-bromoethoxy)(tetradecyloxy)phosphoryloxy)ethoxy-retionoate (10-14)
O
O
P
1´
Br
4´´
b
O
O
2´
O
8´´
10´´
12´´
6´´
2´´
14´´
a
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
183
184
185
Retinoic acid (9) (200 mg, 0.666 mmol) and 4-14 (356 mg, 0.799 mmol) were reacted as described for 10-6 giving 10-14 (423 mg, 87%). TLC
(toluene/acetone/MeOH 6:1:1, v/v): Rf = 0.80. UV/vis (CH2Cl2): ꢀmax = 369 nm. HRMS: C38H64BrO6P calcd. 727.3697 (M+H), found 727.3730.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
186
3.11. Synthesis of 2-((choline)(tetradecyloxy)phosphoryloxy)ethoxy-retinoate (C20-14)
4´
5´
O
O
P
1´
Br
6´´
N
b
6´
O
O
2´
O
4´´
8´´
10´´
12´´
2´´
14´´
a
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
187
188
189
10-14 (432 mg, 0.594 mmol) was reacted as described for C20-6. Purification by flash-CC gave C20-14 (256 mg, 55%). TLC
(CHCl3/MeOH/H2O 40:50:10, v/v): Rf = 0.38. UV/vis (CH2Cl2): ꢀmax = 371 nm. HRMS: C41H73NO6P calcd. 706.5176 (M+), found 706.5187. 1
H
NMR, 13C NMR and 31P NMR in Supplemental Information.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
7
190
3.12. Synthesis of 2-((2-bromoethoxy)(hexadecyloxy)phosphoryloxy)ethoxy-retionoate (10-16)
O
O
P
1´
Br
4´´
b
O
O
2´
O
8´´
10´´
12´´
6´´
2´´
14´´
15´´
a
16´´
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
191
192
193
Retinoic acid (9) (515 mg, 1.7 mmol) and 4-16 (1.24 equiv., 1 g, 2.10 mmol) were reacted as described for 10-6 giving 10-16 (990 mg,
69%). TLC (toluene/acetone/MeOH 6/1/1, v/v): Rf = 0.61. UV/vis (CH2Cl2): ꢀmax = 355 nm. HRMS: C40H68BrO6P calcd. 754.3920 (M+H), found
754.3921. 1H NMR, 13C NMR and 31P NMR in Supplemental Information.
194
3.13. Synthesis of 2-((choline)(hexadecyloxy)phosphoryloxy)ethoxy-retinoate (C20-16)
4´
5´
6´
O
O
P
1´
Br
6´´
N
b
O
O
2´
O
4´´
8´´
10´´
12´´
2´´
14´´
15´´
a
16´´
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
195
196
197
10-16 (400 mg, 0.52 mmol) was reacted as described for C20-6. Purification by flash-CC gave C20-16 (450 mg, 99%). TLC
(CHCl3/MeOH/H2O 70:30:3, v/v): Rf = 0.81. UV/vis (CH2Cl2): ꢀmax = 356 nm. HRMS: C43H77NO6P calcd. 734.5476 (M+), found 734.5473.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
198
3.14. Synthesis of 2-((2-bromoethoxy)(octadecyloxy)phosphoryloxy)ethoxy-retionoate (10-18)
4´
5´
6´
O
O
P
1´
Br
6´´
N
b
O
O
2´
O
4´´
8´´
10´´
12´´
2´´
18´´
14´´
15´´
a
16´´
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
17´´
199
200
201
Retinoic acid (9) (500 mg, 1.6 mmol) and 4-18 (1.24 equiv., 1 g, 2.10 mmol) were reacted as described for 10-6 giving 10-18 (949 mg,
75%). TLC (toluene/acetone//MeOH 6/1/1, v/v): Rf = 0.65. UV/vis (CH2Cl2): ꢀmax = 355 nm. HRMS: C42H72BrO6P calcd. 782.4245 (M+), found
782.4245. 1H NMR, 13C NMR and 31P NMR in Supplemental Information.
202
3.15. Synthesis of 2-((choline)(octadecyloxy)phosphoryloxy)ethoxy-retinoate (C20-18)
4´
5´
6´
O
O
P
1´
Br
6´´
N
b
O
O
2´
O
4´´
8´´
10´´
12´´
2´´
18´´
14´´
15´´
a
16´´
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
17´´
203
204
205
10-18 (350 mg, 0.44 mmol) was reacted as described for C20-6. Purification by flash-CC gave C20-18 (300 mg, 81%). TLC
(CHCl3/MeOH/H2O 70:30:3, v/v): Rf = 0.84. UV/vis (CH2Cl2): ꢀmax = 355 nm. HRMS: C45H81NO6P calcd. 762.5781 (M+), found 762.5779.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
206
3.16. Synthesis of 2-((2-bromoethoxy)(icosyloxy)phosphoryloxy)ethoxy-retinoate (10-20)
4´
5´
6´
O
O
P
1´
Br
6´´
N
b
O
O
2´
O
4´´
8´´
10´´
12´´
2´´
14´´
15´´
a
16´´
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
207
208
209
Retinoic acid (9) (200 mg, 0.666 mmol) and 4-20 (423 mg, 0.799 mmol) were reacted as described for 10-6 giving 10-20 (478 mg,
88%). TLC: (toluene/acetone/MeOH 6:1:1, v/v): Rf = 0.85. UV/vis (CH2Cl2): ꢀmax = 370 nm. HRMS: C44H76BrO6P calcd.811.4636 (M+H), found
811.4668. 1H NMR, 13C NMR and 31P NMR in Supplemental Information.
210
3.17. Synthesis of 2-((choline)(icosyloxy)phosphoryloxy)ethoxy-retinoate (C20-20)
4´
5´
6´
O
O
P
1´
Br
6´´
N
b
O
O
2´
O
4´´
8´´
10´´
12´´
2´´
18´´
14´´
15´´
a
16´´
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
17´´
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
8
211
212
213
10-20 (460 mg, 0.566 mmol), was reacted as described for C20-6. Purification by flash-CC gave C20-20 (362 mg, 74%). TLC
(CHCl3/MeOH/H2O 40:50:10, v/v): Rf = 0.42 UV/vis (CH2Cl2): ꢀmax = 369 nm. HRMS: C47H85NO6P calcd.790.6115 (M+), found 790.6125.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
214
3.18. Synthesis of ˇ-apo-8ꢀ-carotenoic acid (2)
4´
5´
6´
O
O
P
1´
Br
6´´
N
b
O
O
2´
O
4´´
8´´
10´´
12´´
2´´
18´´
14´´
15´´
a
16´´
O
7´´
3´´
5´´
1´´
9´´
11´´
13´´
17´´
215
216
Hydrolysis of ␤-apo-8ꢀ-carotenoate gave 2 in 90% yield. 1H NMR: in accordance with Ref. Larsen et al. (1998).
3.19. Synthesis of 2-((2-bromoethoxy)(hexyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (5-6)
O
O
P
1´
Br
4´´
2
O
O
2´
C29
O
1
6´´
2´´
O
3´´
1´´
5´´
217
218
219
220
221
222
223
␤-Apo-8ꢀ-carotenoic acid (2) (300 mg, 0.694 mmol), 4-6 (1.2 equiv., 257 mg, 0.773 mmol), chlorotripyrrolidinophosphonium hexaflu-
orophosphate (PyCloP, 1.25 equiv., 349 mg, 0.868 mmol), N-ethyl diisopropylamine (DIEA, 0.65 equiv., 58 mg, 0.451 mmol), and DMAP
(1.25 equiv., 106 mg, 0.868 mmol) were dissolved in dry CH2Cl2 (50 mL) and the mixture was refluxed under N2 for 24 h. Extraction of the
mixture with water (2× 50 mL), aqueous HBr (0.1 M, 2× 50 mL), and water (2× 50 mL), drying over anhydrous Na2SO4 and concentration
gave a residue, which, after flash-CC on silica with a toluene–acetone gradient, gave 5-6 (322.3 mg, 62%). TLC (toluene/acetone/MeOH
6:1:1, v/v): Rf = 0.68. UV/vis (CH2Cl2): ꢀmax = 453 nm. MS: 746.85 + 748.85 (1:1, M+); HRMS: C40H60BrO6P calcd. 748.32993 (M +), found
748.33023. 1H NMR, 13C NMR and 31P NMR in Supplemental Information.
224
3.20. Synthesis of 2-((choline)(hexyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (C30-6)
4´
5´
6´
O
O
P
Br
1´
N
2
O
O
2´
C29
O
1
4´´
6´´
2´´
O
3´´
1´´
5´´
225
226
227
228
Carotenoate 5-6 (322 mg, 0.43 mmol) was dissolved in CHCl3/iPrOH/DMF (3/5/5, v/v, 50 mL), NMe3 (45% in water, 10 mL) was added and
the mixture stirred at room temperature under N2 for 4 days. Flash-CC on neutral Al2O3 gave C30-6 (264 mg, 77%). TLC (CHCl3/MeOH/H2O
70:30:3, v/v): Rf = 0.08. UV/vis (CH2Cl2): ꢀmax = 458 nm. HRMS: C43H69NO6P calcd. 726.4803 (M+), found 726.4869. 1H NMR, 13C NMR and
31P NMR in Supplemental Information.
229
3.21. Synthesis of 2-((2-bromoethoxy)(dodecyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (5-12)
O
O
P
1´
Br
4´´
2
O
O
2´
C29
O
1
6´´
8´´
10´´
2´´
12´´
O
3´´
1´´
5´´
11´´
9´´
7´´
230
231
232
␤-Apo-8ꢀ-carotenoic acid (2) (200 mg, 0.459 mmol) and 4-12 (1.2 equiv., 230 mg, 0.550 mmol) were reacted as described for 5-6 giving
5-12 (200.7 mg, 53%). TLC (hexane/acetone 8:2, v/v): R = 0.30. UV/vis (CH2Cl2): ꢀmax = 457 nm. MS (EI): 830.1 + 832.2 (1:1, M+). HRMS:
C46H72BrO6P calcd. 830.42500 (M+), found 830.42748. 1fH NMR, 13C NMR and 31P NMR in Supplemental Information.
233
3.22. Synthesis of 2-((choline)(dodecyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (C30-12)
4´
5´
6´
O
O
P
Br
1´
N
2
O
O
2´
C29
O
1
4´´
6´´
8´´
10´´
2´´
12´´
O
3´´
1´´
5´´
11´´
9´´
7´´
234
235
236
5-12 (200.7 mg, 0.241 mmol) was reacted as described for C30-6. Purification by flash-CC gave C30-12 (130 mg, 61%). TLC
(CHCl3/MeOH/H2O 70/30/3, v/v): Rf = 0.47. UV/vis (CH2Cl2): ꢀmax = 458 nm. HRMS: C49H81NO6P calcd. (M+) 810.5802 (M+), found 810.5793.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
9
237
3.23. Synthesis of 2-((2-bromoethoxy)(tetradecyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (5-14)
O
O
P
1´
Br
4´´
2
O
O
2´
C29
O
1
6´´
8´´
10´´
14´´
2´´
12´´
O
13´´
3´´
1´´
5´´
11´´
9´´
7´´
238
239
240
␤-Apo-8ꢀcarotenoic acid (2) (300 mg, 0.694 mmol) and 4-14 (1.2 equiv., 344 mg, 0.773 mmol) were reacted as described for 5-6 to 5-14
(425 mg, 71%). TLC (hexane/acetone 8:2), v/v: Rf = 0.27. UV/vis (CH2Cl2): ꢀmax = 457 nm. 1H NMR, 13C NMR and 31P NMR in Supplemental
Information.
241
3.24. Synthesis of 2-((choline)(tetradecyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (C30-14)
4´
5´
6´
O
O
P
Br
1´
N
2
O
O
2´
C29
O
1
4´´
6´´
8´´
10´´
14´´
2´´
12´´
O
13´´
3´´
1´´
5´´
11´´
9´´
7´´
242
243
244
5-14 (400 mg, 0.465 mmol) was reacted as described for C30-6. Purification by flash-CC gave C30-14 (242 mg, 57%).TLC
(CHCl3/MeOH/H2O 70:30:3, v/v): Rf = 0.68. UV/vis (CH2Cl2): ꢀmax = 461 nm. HRMS: C51H85NO6P calcd. 838.6115 (M+), found 838.6121.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
245
3.25. Synthesis of 2-((2-bromoethoxy)(hexadecyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (5-16)
O
O
P
1´
Br
4´´
2
O
O
2´
C29
O
1
6´´
8´´
10´´
14´´
16´´
2´´
12´´
O
13´´
3´´
1´´
5´´
11´´
15´´
9´´
7´´
246
247
248
␤-Apo-8ꢀ-carotenoic acid (2) (450 mg, 1.04 mmol) and 4-16 (2.39 equiv., 1.174 mg, 2.48 mmol) were reacted as described for 5-6 giving
5-16 (631 mg, 68%). TLC (toluene/acetone/MeOH 6:1:1, v/v): Rf = 0.80. UV/vis (CH2Cl2): ꢀmax = 455 nm. HRMS: C50H80BrO6P calcd. 886.4869
(M+H), found 886.4869. 1H NMR, 13C NMR and 31P NMR in Supplemental Information.
249
3.26. Synthesis of 2-((choline)(hexadecyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (C30-16)
4´
5´
6´
O
O
P
Br
1´
N
2
O
O
2´
C29
O
1
4´´
6´´
8´´
10´´
14´´
16´´
2´´
12´´
O
13´´
3´´
1´´
5´´
11´´
15´´
9´´
7´´
250
251
252
5-16 (200 mg, 0.225 mmol) was reacted as described for C30-6. Purification by flash-CC gave C30-16 (155 mg, 72%). TLC
(CHCl3/MeOH/H2O 70:30:3, v/v): Rf = 0.62. UV/vis (CH2Cl2): ꢀmax = 456 nm. HRMS: C53H89NO6P calcd. 866.6422 (M+), found 866.6423.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
253
3.27. Synthesis of 2-((2-bromoethoxy)(octadecyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (5-18)
O
O
P
1´
Br
4´´
2
O
O
2´
C29
O
1
6´´
8´´
10´´
14´´
16´´
2´´
18´´
12´´
O
13´´
3´´
17´´
1´´
5´´
11´´
15´´
9´´
7´´
254
255
256
␤-Apo-8ꢀcarotenoic acid (2) (670 mg, 1.54 mmol), 4-18 (2.39 equiv., 1856 mg, 3.70 mmol) were reacted as described for 5-6 giving 5-18
(858 mg, 61%). TLC (toluene/acetone/MeOH 6:1:1, v/v): Rf = 0.75. UV/vis (CH2Cl2): ꢀmax = 454 nm. HRMS: C52H84BrO6P calcd. 914.5196 (M+),
found 914.5189. 1H NMR, 13C NMR and 31P NMR in Supplemental Information.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
10
257
3.28. Synthesis of 2-((choline)(octadecyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (C30-18)
4´
5´
6´
O
O
P
Br
1´
N
2
O
O
2´
C29
O
5´
1
4´´
6´´
8´´
10´´
14´´
16´´
2´´
18´´
12´´
O
13´´
3´´
17´´
1´´
5´´
11´´
15´´
9´´
7´´
258
259
260
5-18 (240 mg, 0.261 mmol) was reacted as described for C30-6. Purification by flash-CC gave C30-18 (200 mg, 78%). TLC
(CHCl3/MeOH/H2O 70:30:3, v/v): Rf = 0.82. UV/vis (CH2Cl2): ꢀmax = 455 nm. HRMS: C55H93NO6P calcd. 894.6729 (M+), found 894.6728.
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
261
3.29. Synthesis of 2-((2-bromoethoxy)(icosyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (5-20)
O
O
P
1´
Br
4´´
2
O
O
2´
C29
O
1
6´´
8´´
10´´
14´´
16´´
2´´
18´´
12´´
O
20´´
13´´
3´´
17´´
1´´
5´´
11´´
15´´
19´´
9´´
7´´
262
263
264
␤-Apo-8ꢀ-carotenoic acid (2) (397 mg, 0.919 mmol) and 4-20 (340 mg, 0.584 mmol) were reacted as described for 5-6 giving 5-20
(333 mg, 62%). TLC (hexane/acetone 8:2, v/v): Rf = 0.28. UV/vis (CH2Cl2): ꢀmax = 458 nm. HRMS: C54H88BrO6P calcd. 945.5568, found
945.5547 (M+H). 1H NMR, 13C NMR and 31P NMR in Supplemental Information.
265
3.30. Synthesis of 2-((choline)(icosyloxy)phosphoryloxy)ethoxy-ˇ-apo-8ꢀ-carotenoate (C30-20)
4´
5´
6´
O
O
P
Br
1´
N
2
O
O
2´
C29
O
1
4´´
6´´
8´´
10´´
14´´
16´´
2´´
18´´
12´´
O
20´´
13´´
3´´
17´´
1´´
5´´
11´´
15´´
19´´
9´´
7´´
266
267
268
5-20 (300 mg, 0.317 mmol) was reacted as described for C30-6. Purification by flash-CC gave C30-20 (303 mg, 95%). TLC
(CHCl3/MeOH/H2O 70:30:3, v/v): Rf = 0.78. UV/vis (CH2Cl2): ꢀmax = 459 nm. HRMS: C57H97NO6P calcd. 922.7054, found 922.7048 (M+).
1H NMR, 13C NMR and 31P NMR in Supplemental Information.
269
270
4. Results and discussion
4.1. Synthesis
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
Cationic lipids are often identified with phospholipids and phospholipids are habitually recognized as glycerophospholipids) (Foss et al.,
2003, 2005a). Yet, the glycerol scaffold complicates structure–activity studies by possible formation of mono- and di-glycerol isomers and
enantiomers, and by inter- and intra-molecular acyl migration (Chang et al., 1997; Mangroo and Gerber, 1988). We, therefore, replaced
the glycerol (propanetriol n = 3) backbone with glycol (1) (ethanediol n = 2; glycolphospholipids occur in minor amounts in natural lipids)
(Bergelson, 1970). We further exchanged one of the flexible saturated acyl chains characteristically located in lipid gene carriers (Koynova
and Tenchov, 2009) with ␤-apo-8ꢀ-carotenoic acid (C30:9) (2) (C30-acid, food color E160f) and retinoic acid (C20:5) (9) (Scheme 1). Keeping
the glycol unit, the polyene chains and the charged head group constant, the length of the phosphate ester alkyl chain was varied (C6,
C12, C14, C16, C18, C20); we acquired two C20-n and C30-n series (Fig. 1 and Scheme 1). Cationic phospholipids bearing different chains or
chains with conjugated unsaturation are rarely examined (Adami et al., 2011; Koynova et al., 2009; Zhi et al., 2010). The dependence of
transfection efficiency on phosphate ester alkyl chains has earlier been accentuated (Rosenzweig et al., 2000).
Reacting chlorodioxaphospholane (3) with glycol (1) and subsequent addition of the appropriate alcohol Cn-OH and Br2 gave inter-
mediate phosphate 4-n (Byun and Bittman, 1996), which with C30-acid 2 and the coupling reagents chlorotripyrrolidinophosphonium
hexafluorophosphate (PyCloP) (Coste et al., 1994), diisopropylethylamine (DIEA) and 4-dimethylaminopyridine (DMAP) formed bro-
mophosphotriester 5-n. Aminolysis of 5-n gave C30-6, C30-12, C30-14 and C30-20 (Scheme 1). Phosphorous diesters were retrieved as
byproducts and bromine reacted promiscuously with phosphorous and the exo- and endocyclic methylene groups next to the P O-bonds
in phospholane 3 (Predvoditilev et al., 2001). Otherwise, trichlorophosphate (6) was reacted with bromoethanol (7) to bromophosphate
8, which with glycol (1) and alcohol Cn-OH gave again intermediate 4-n. By these sequences C30-16 and C30-18 were obtained. Similarly,
phosphate 4-n reacted with retinoic acid 9 to the C20-n compounds.
289
4.2. Molecular structure
290
291
292
293
The combination of rigidity and flexibility in the new lipids is exemplary demonstrated for C30-14 (Fig. 3). Diverse conformational
arrangements of acyl chains have been previously noticed in a zwitterionic phospholipid with a hexatriene chain (Ryhänen, 2006, p. 55).
Since molecules with flexible chains are prone to variation in length, molecular volume figures as a more appropriate expression to account
for chain extension (Fig. 4).
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
11
Fig. 3. Hypothetical structures of rigid polyene chains and flexible saturated alkyl chains in C30-14.
294
4.3. Surface properties
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
The surface tension ꢁ of the synthesized amphiphilic phospholipid surfactants was determined with a tensiometer (Pt-plate). Calculation
of the tensiometric data assessed the critical aggregation concentration cM, the area per molecule at the filled monolayer am and other
associated data such as ꢁc , surface concentration ꢂ , free energy of aggregation ꢃGa0g and adsorption ꢃGa0d, surfactant performance
M
indicator AMER (Skrylev et al., 2000), equilibrium constants for aggregation and adsorption kag and kad (Tables 1 and SI 1) (Foss et al.,
2005b,c). The surface area am was established assuming no dissociation of the molecules in water (one species) or complete dissociation
(two species). The values for am, supposing complete dissociation, were too high and in variance with molecular calculations indicating
that the cationic molecules aggregate with tightly attached counter ions. The importance of tensiometric data for controlling gene transfer
has been previously demonstrated by investigating Langmuir–Blodgett films (Antipina et al., 2009).
C20-16 is most surface active, decreasing the surface tension ꢁ of water to 35 mN/m, whereas C20-20 and C30-14 are not effective
surfactants with ꢁ = 62 and 63 mN/m, respectively (Table 1 and Fig. 5). C20-16 aggregates in water at low concentration (cM = 10.6 ␮M),
C20-20 is much more soluble with an aggregation concentration of 75 ␮M. The observation in prior studies that cM of most surfactants
decreases linearly with the number of carbon atoms in a chain is not reproduced with the C20-n and C30-n compounds (Shinoda, 1963;
Yoshimura et al., 2012). The surface area am is rather similar, except for C20-14 and C30-14 with unexpected large am values (Fig. 6).
The hydrophilic chains in the C30-n series can adopt a “closed V” or a “stretched V”-shaped-conformation upon rotation about the
oxygen carbon bond of the phosphate ester acyl side chain (Fig. 7). The reasons why C20-14 and C30-14 orient differently are not evident.
Analogous large surface areas have been detected with polyene bolaamphiphiles (Breukers et al., 2009). The comparison of the calculated
molecular volumes, which are independent of chain orientation, revealed that ECP is equal to
3
˚
Fig. 4. Molecular volume (A ) of Cn-n compounds and EPC. C30-12 and EPC have comparable molecular volumes (spartan08, semi empirical PM3).
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
12
Table 1
Q10
Surface property data of C20-n and C30-n.
2
˚
cM (␮M)
31.7
50
10.6
13
ꢁc (mN/m)
ꢂ (␮mol/m2)
am (A )
C20-6
46
57
35
48
63
2.05
1.2
2.32
2.63
2.6
81
139
72
63
63
C20-14
C20-16
C20-18
C20-20
75
C30-2
C30-6
23.3
19
46.6
41.7
46
49
53
53
62
49
42
47
2.6
64
56
86
191
56
64
2.97
1.93
0.87
2.95
2.61
2.1
C30-12
C30-14
C30-16
C30-18
C30-20
16
18.9
79
aRef. Øpstad et al. (2013).
312
4.4. Aggregation
313
314
315
316
317
318
319
320
321
322
323
324
Aggregation of the C20-n and C30-n phospholipids started at irregular concentrations cM (Table 1) and the morphology of the formed
aggregates was hardly foreseeable, e.g. resulting for C30-2 in rods, spheroids and cones (Øpstad et al., 2013). It is assumed that the other C20-
n and C30-n lipids similarly self-assemble in water to polymorphic particles. Independent of their morphology, aggregates are characterized
by their absorption spectrum reflecting the molecular arrangement in the aggregation unit (Foss et al., 2005a), e.g. the aqueous aggregate
dispersion of C30-14 absorbed at lower wavelengths than a molecular solution in organic solvents, suggesting an exitonic card-pack
association of the molecules characterized as H-aggregates (C30-14 in EtOH ꢀmax = 444 nm, C30-14 in H2O (H-aggregate) ꢀmax = 400 nm
(Fig. 8) (Horn and Rieger, 2001; Wang et al., 2012). Aggregation depends on subtle conditions and amphiphilic molecules can also adopt
head-to tail arrangements leading to higher absorbing J-aggregates. Such details in molecule arrangements are not easily distinguished in
saturated lipids or in lipids with non conjugated double bonds (Sliwka et al., 2010).
Adding water to concentrated solvent solutions of the C30-n lipids allowed detecting the aggregation concentration cag; correspond-
ingly, adding organic solvents to aqueous C30-n dispersions assessed the concentration of aggregate disruption. The monomer–aggregate
equilibrium for C30-14 was, for instance, found at 31% EtOH and 69% H2O (Fig. 8).
325
4.5. Liposomes
326
327
328
Thin films of pure cationic lipids or mixtures of cationic lipids with co-lipids, DOPE or cholesterol, were hydrated initiating the formation
of liposomes. Cholesterol and zwitterionic DOPE (thought to have a role in the protection of nucleic acids and promotion of transfection,
Fig. 2) (Dabkowska et al., 2012; Hirsch-Lerner et al., 2005; Pozzi et al., 2012) were paired with the C20-n and C30-n compounds at a constant
Fig. 5. Maximum and minimum reduction of water surface tension (ꢁ = 73 mN/m) with surface active C20-n and C30-n compounds.
Fig. 6. Molecule area am at the water surface of C30-16 (left, representing roughly the average of most C20-n and C30-n molecules), of C20-14 (center) and C30-14 (right).
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
13
Fig. 7. Hypothetically oriented C30-12 (closed V) and C30-14 (stretched V) at the water surface, the black line — indicates the diameter of the surface area am at the water
surface
(semi-empirical, AM1 in Spartan 08, Wavefunction, Irvine, California, USA).
2,0
1,6
1,2
A
0,8
0,4
0,0
250
300
350
400
450
/nm
500
550
600
Fig. 8. Monomolecular solution of C30-14 in EtOH
. Adding H2O initiate formation of H-aggregates
. Aggregate dispersion of C30-14 in H2O
. Adding EtOH
disrupts the aggregates forming a monomolecular solution
. The aggregation concentration (% H2O in EtOH) cag = 69 (in MeCN cag = 74). (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of the article.)
329
330
331
332
333
334
335
336
337
338
339
molar ratio of 3:2 (lipid/co-lipid). The hydrodynamic radius and polydispersity index (PdI) of the liposomes was ascertained by dynamic
light scattering (DLS). C30-16 appeared as the smallest liposome (dH = 70 nm) and liposome C20-18 as the largest (dH = 1640 nm, Table 2).
Within the cholesterol series C20-18/Chol formed the smallest and C30-20/Chol the largest liposomes (dH = 170 nm and dH = 1400 nm,
respectively). The high variance in the polydispersity index of the pure liposomes (PdI = 0.2–1) was reduced in the liposomes with DOPE to
a PdI range of 0.3–0.6. The EPC-containing liposomes had low polydispersity and small sizes, as reported elsewhere (Øpstad et al., 2013),
suggesting that the variability in liposome size and polydispersity among C30-n and C30-n liposomes was inherent to their self-association
properties. Thus, the polyene cationic lipids, regardless of formulation, yielded a range of liposome with no obvious structure–property
relationship. Our results contrast with the outcome of identical sized liposomes independent of chemical structure (Ivanova et al., 2013)
Having completed a thorough analysis of physical and chemical characteristics of the novel lipids as well as an analysis of liposome
formation, the data comparing structure and property can be viewed graphically. The dataset manifests no structure–property relationship
of the new lipids (Fig. 9).
340
4.6. Lipoplexes
341
342
343
344
345
346
Negatively charged DNA was subsequently added to cationic liposomes; lipoplexes were obtained, mediated by electrostatic interactions
and hydrophobic effects. Lipoplex formulations are described in terms of defined amine:phosphate (N/P) or molar charge ratios (+/−). In this
study, N/P ratios of 0.5:1, 1.5:1, 3.0:1, 5.0:1 and, in some cases, 10:1 were evaluated. Dynamic light scattering was employed to characterize
lipoplexes. Measurements indicated a wide range of hydrodynamic diameters, dH (Table SI 2). C20-n lipoplexes were smaller on average
when the saturated chain was shorter (i.e. C20-14), but as this chain increased in length, formulations with cholesterol produced reduced
lipoplexes, i.e. C20-18/Chol and C20-20/Chol. C30-n lipoplexes were usually smaller without co-lipid and largest with cholesterol. The
Table 2
Hydrodynamic diameter dH (nm) of liposomes and polydispersity index (PdI) measured by DLS. C30-6 was omitted for instability reasons.
Liposome
Liposome/DOPE
Liposome/Chol
Cationic lipid
dH (nm)
PdI
dH (nm)
PdI
dH (nm)
PdI
EPC
120
180
480
1645
815
270
375
75
0.6
0.4
0.8
0.8
0.5
1
0.5
0.5
0.8
0.2
490
1015
350
510
670
275
382
293
300
880
0.6
0.6
0.3
0.5
0.6
0.5
0.6
0.4
0.4
0.6
300
550
365
170
260
640
397
265
260
1400
0.4
0.7
0.8
0.4
0.4
0.4
0.8
0.4
0.4
0.8
C20-14
C20-16
C20-18
C20-20
C30-12
C30-14
C30-16
C30-18
C30-20
230
205
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
14
Fig. 9. Comparison of compound structure, surface data, and liposome size. The linear increase in molecular volume is not reflected in the physical and structural data
3
˚
obtained during characterization of the lipids. (A) Calculated molecular volume (A ). (B) Aggregation concentration cM (␮M). (C) Surface tension ꢁc (mN/m). (D) Surface area
2
˚
am (A ). (E) Hydrodynamic radius rH (nm) of liposomes composed of cationic lipids. (F) Hydrodynamic radius rH (nm) of liposomes composed of cationic lipids with DOPE
as a co-lipid. (G) Hydrodynamic radius rH (nm) of liposomes composed of cationic lipids with cholesterol as a co-lipid. The open bars represent hydrodynamic radii derived
from samples with a high polydispersity index (≥0.7) and thus, may not include a single population size.
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
polydispersity index (PdI) of lipoplexes in the C20-n and C30-n series ranged from of 0.2 to 1 (Table SI 2). PdI values greater than 0.7
indicated that multiple species or a broad distribution of sizes were present.
Lipoplexes formed by the reference lipid EPC were typically smaller without co-lipid and larger with either DOPE or cholesterol (Chol).
EPC lipoplexes were mono-disperse for all three formulations. The lipoplexes formed by EPC were relatively small (dH < 600 nm) at low or
high N/P ratios, while some formulations, particularly involving molar ratios between 1.5 and 5, included large structures exceeding 2 ␮
(Table SI 2).
It is easy to anticipate a regular relationship between lipoplex size and liposome size or between lipoplex size and the ratio of cationic
lipids to DNA. However, our observations with lipoplexes containing cationic C20-n or C30-n lipids resist the application of a simple
structure–property relationship. Some lipoplexes were smaller than the liposomes (C20-14 liposome dh = 180 nm, C20-14 lipoplex N/P 0.5
dh = 170 nm, C30-18 liposome dh = 230 nm, C30-18 lipoplexes N/P 10 dh = 120 nm. Other lipoplexes were significantly larger (e.g. C20-18
liposome, C20-18/DOPE N/P 1.5 lipoplex, Table SI 2). The large lipoplexes may arise from liposome fusion in the buffer solution, from
electrostatic adhesion of neighboring liposomes or from lipoplex aggregation (Balbino et al., 2012; Kedika and Patri, 2011). Small lipoplex
structures could result from the compaction of the plasmid DNA by the cationic lipids associated with the liposomes through electrostatic
and hydrophobic forces. The electrostatic adhesion of several small neighboring lipoplexes gives rise to large lipoplex aggregates. Since a
structure–property relationship for liposomes was missing (Table 2), a correlation between lipoplex activity (i.e., DNA delivery) and their
structural properties was, similarly, not expected (Fig. 9).
363
4.7. Gel retardation assays of lipoplexes
364
365
366
A gel retardation assay was employed to study the binding interaction in the lipoplexes between the C20-n and C30-n liposomes and
DNA. The assay revealed that, in general, DNA binding improved with increasing lipid ratio in the lipoplex formulations; typically, near
complete retention was achieved at an N/P ratio of approximately 5.0:1 or 10:1 (Fig. SI 1).
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
15
Fig. 10. Transfection efficiency to CHO-K1 cells for lipoplexes composed of EPC (blue bars, left), C20-n (red bars) or C30-n (green bars) at N/P ratios ranging from 0.5 to 5. Data
are expressed in terms of the activity of the enzyme that was expressed by the transfected DNA (beta galactosidase) measured in relative light units (RLU) and normalized to
the protein concentration of the transfected cell lysate (␮g protein) collected 48 h after transfection. Negative controls included untreated cells (“C,” blue bar, right) or cells
treated with only DNA (“D,” blue bar, right). (A) Transfection using lipoplexes formed from liposomes containing only cationic lipid C20-n or C30-n. (B) Transfection using
lipoplexes formed from liposomes containing the cationic lipids plus DOPE as a co-lipid. (C) Transfection using lipoplexes formed from liposomes containing the cationic
lipids plus Chol as a co-lipid. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
367
4.8. DNase I degradation assays of lipoplexes
368
369
370
371
372
A DNase I degradation assay verified the accessibility of the DNA in the lipoplexes to nucleases. DNA was partially protected from nuclease
degradation in all lipoplex formulations at all charge ratios (Fig. SI 2). C20-18 and C20-20 based lipoplexes gave greater DNA protection,
correlating with the better DNA binding observed in the gel retardation assays (Fig. SI 2). C30-12 and C30-20 lipoplexes appeared most
protective. The C30-n/DOPE lipoplexes were equally protective for the DNA plasmid, whereas C30-12/Chol and C30-20/Chol were best in
the C30-n/Chol formulations.
373
4.9. Transfection
374
375
376
377
378
379
380
381
Each liposome was formulated into lipoplexes by combination with DNA plasmids. The plasmid used for lipoplex formation contained
a gene expression cassette including a gene for ␤-galactosidase to enable the detection of successful gene delivery to the nucleus of target
cells resulting in the production of active enzyme. The enzyme activity was quantified as microgram of protein in the cell lysate of CHO-K1
cells. The values obtained with the C20-n and C30-n lipoplexes were generally significantly higher than those obtained from DOPE and
Chol formulated lipoplexes. Remarkably, the lipid with the longest chains C30-20 at N/P 3 and 5 functioned slightly better than EPC; the
lipoplex of C30-18 at N/P 1.5 was likewise outperforming EPC (Fig. 10). Lipoplexes formulated with DOPE and Chol showed significant
transgene activity, though always lower than EPC. Gene expression was generally lower with DOPE than with Chol, especially in the case of
several formulations of C20-14, C20-16, and C20-18. No trend appeared concerning performance for the C20-n and C30-n lipoplexes with
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
16
Fig. 11. Cell viability following transfection of CHO-K1 cells with lipoplexes composed of EPC (blue bars, left), C20-n (red bars) or C30-n (green bars) with N/P ratios ranging
from 0.5 to 5. Data are expressed as percentage viability for CHO-K1 cells after 48 h compared to untreated cells (“C,” blue bar, right) or cells treated with DNA alone (“D,” blue
bar, right). (A) Transfection using lipoplexes formed from liposomes containing only cationic lipid C20-n or C30-n. (B) Transfection using lipoplexes formed from liposomes
containing cationic lipid plus DOPE as a co-lipid. (C) Transfection using lipoplexes formed from liposomes containing cationic lipid plus Chol as a co-lipid. (For interpretation
of the references to color in this figure legend, the reader is referred to the web version of the article.)
382
383
respect to the variation of saturated chain length (Fig. 10). Recent studies contradictorily found improved gene delivery through increasing
or decreasing unsaturation and reducing or elongating chain lengths (Jones et al., 2013; Liberska et al., 2009; Loizeau et al., 2013).
384
4.10. Transfection and lipoplex size
385
386
387
388
389
390
391
392
393
The importance of the lipoplex size for gene transfer has continuously been stressed. The most suitable lipoplex size has been attained
either between 10 and 200 nm or 300 and 2000 nm (Almofti et al., 2003; Li and Szoka, 2007; Saha et al., 2012). Greater cellular association
and uptake improved with larger lipoplexes in in vitro assays; endocytis with CHO-K1 cells was enhanced by increasing particle size to
2200 nm (Ross and Hui, 1999). Our data showed that lipoplex C30-20 at N/P 5 (dH = 600 nm) performed slightly better than small EPC
lipoplex at N/P 5 (dH = 280 nm, Fig. 10 and Table SI 2). Large C30-20/3 (dH = 2080 nm) was likewise superior to small EPC/3 (dH = 380 nm).
C30-18/1.5 (dH = 1770 nm) was more than double as effective in gene delivering than large EPC/1.5 lipoplex. Large C20-20/1.5 and C20-
18/DOPE/1.5 transfected better than large EPC/1.5. The smallest lipoplexes, i.e. C20-14/0.5 (dH = 170 nm) and EPC/0.5 (dH = 150 nm) were
both inactive in gene transfer. Thus lipoplex size is not a significant determinant for CHO-K1 cell transfection when using C20-n and C30-n
based lipoplexes.
394
4.11. Cellular viability
395
396
397
398
399
Lipoplexes were generally well tolerated by CHO-K1 and relatively high transfection levels could be obtained, although enhanced
transgene expression was accompanied by higher cytotoxicity (Fig. 11). Cell viability was, as expected, high with lipoplexes of low N/P
ratios; increasing the N/P ratio decreased cell viability. Of interest, C30-20 stood out among the new compounds in that it showed highest
transfection efficiency, higher in fact than EPC in any formulation tested. At the same time, C30-20 in the same formulations that provided
high gene expression (N/P of 3 and 5 without co-lipid) lacked the characteristically high toxicity of those N/P ratios.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
17
Fig. 12. SAXRD profile for C30-20 and C20-20 lipid/DNA lipoplex formulations without colipid (A and D), and with co-lipids DOPE (B and E) and cholesterol (C and F) an N/P
molar charge ratio 1.5:1. Data are plotted as signal intensity (arbitrary units) as a function of the scattering vector.
400
401
402
Having completed a thorough analysis of physical and chemical characteristics of the novel lipids without obtaining a structure–property
relationship, the further analysis of the bioactivity of lipoplexes did, similarly, not provide evidence of a structure–activity relationship
(Figs. 10 and 11).
403
4.12. Small angle X-ray diffraction (SAXRD)
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
The intense X-rays provided by synchrotron sources have given rise to powerful tools for studying self-assembled nanostructures.
One such technique is small angle X-ray scattering, which is useful for obtaining information on single particles. Additional information is
available for those macromolecular systems whose structures are ordered, or partially ordered, because they can yield diffraction peaks that
are superimposed on the SAXS spectrum. These peaks can be used to extract structural information. This is the situation for the lipid–DNA
systems studied here and the data are logically designated as small angle X-ray diffraction (SAXRD). The scattering that constitutes SAXRD
stems from single particle and inter-particle interference. Although the measurements were carried out using the SAXS experimental setup,
it is the SAXRD part of the SAXS/SAXRD data that we make use of and which is analyzed here.
SAXRD data were measured for C20-20 and C30-20 liposomes and lipoplexes and those formulated with DOPE and cholesterol, all at an
N/P ratio of 1.5. The presence of at least a single sharp peak expresses periodicity within the lipoplex membrane. The unimodal SAXS profile
of the liposomes is associated with the absence of ordered arrangements. Similarly, the broad peaks of C20-20 and C30-20 lipoplexes are
consistent with reduced bilayer order (Fig. 12), and, therefore, a precise value for the interlayer distance (ı) is difficult to obtain, although an
approximate value may be extracted. The SAXRD profile of C20-20/DOPE and C20-20/Chol with their broad primary diffraction peaks and
barely emergent secondary peaks are consistent with the evolution of periodicity. The secondary diffraction peak of lipoplex C30-20/DOPE
(Fig. 12) is somewhat better resolved, reflecting a move to a more intervallic structure. The two narrow diffraction peaks of C30-20/Chol
stem from the presence of ordered domains, most likely lamellar structures. The small shoulder near the intense SAXS-peak of C30-20/Chol
is compatible with a diffraction signature from the DNA-strands in the lamellar◦segments, indicating an in-plane distance between the
˚
strands of dDNA ≈ 59 A. The well-defined lamellar ordering of C30-20/Chol at 20 C is still maintained when the temperature is raised to
55 ◦C; the small DNA peak became slightly more visible at higher temperature (Fig. SI 3).
The SAXRD profiles of C20-20 and C30-20 lipoplexes in Fig. 12 are most likely consistent with lamellar ordering with repetition distances
given in Table 3. Comparison of the lipoplex profiles of these two rigid chain lipids with that of the lipoplex of EPC with its flexible
chains (Fig. SI 4) is consistent with the bilayer being disrupted by the polyene chains. The effect of a disrupted bilayer on transfection
efficiency is not clear. At an N/P ratio of 1.5, C20-20 lipoplexes were least organized but were more effective than either C20-20/DOPE or
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
18
Table 3
Microstructure and interlayer distance ı of the lipoplexes derived from SAXRD diffraction data (the corresponding liposomes did not exhibit diffraction).
Lipoplex (N/P = 1.5:1)
Profile
Lipid formulation
˚
Distance ı (A)
C30-20
C30-20/DOPE
C30-20/Chol
Lamellar (most likely)
Lamellar
Lamellar
∼59
63
70
C20-20
C20-20/DOPE
C20-20/Chol
Lamellar (most likely)
Lamellar
Lamellar
∼58
∼61
∼60
426
427
428
429
430
C20-20/Chol in DNA delivery (Fig. 10). In contrast, the DNA transfection efficiency of C30-20 lipoplexes was low and similar to the more
highly organized C30-20/DOPE or C30-20/Chol at N/P 1.5. But at a slightly higher N/P ratio of either 3 or 5, the transfection efficiency of
C30-20 was noticeably improved with all three formulations (Fig. 10). Thus, the adverse effect of the rigid chains on ordering does not
predict transfection efficiency using the formulations tested here. Ultimately, destabilization of bilayers by rigid polyenes may prove to
influence fusion and unpackaging of lipoplexes inside of cells.
431
4.13. Chain rigidity-flexibility, configuration, and transfection
432
433
434
435
436
Alkyl chain flexibility in phospholipid membranes accounts for molecular conformer changes accompanied by phase transitions from
gel to fluid liquid crystalline (Mineva et al., 2013). Aggregation distorts as well the polyene chain of phospholipids, even though in a much
more restricted manner (Foss et al., 2005a). The polyene chains in the C20-n and C30-n lipids are, therefore, considered rigid. At N/P 3 and
5 the achiral glycol lipid C30-20 with an unsaturated rigid chain transfected cells to a comparable level as the glycerolipid enantiomer
(R)-EPC with two flexible saturated chains.
437
5. Conclusion
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
A new class of cationic glycol phospholipids has been synthesized with chromophoric, rigid polyene chains and flexible alkyl chains.
Essential property data such as surface tension ꢁ, aggregate concentration cM and the molecular area am could not be interconnected with
the structure of the C20-n and C30-n compounds. When the homologous C20-n and C30-n amphiphiles came in contact with water, a
heterologous behavior was observed caused by unpredictable assembling of the molecules to aggregates and liposomes. Neither liposome
nor lipoplex dimensions were defined by chain lengths. The combination of a C30:9 chain with a C20:0 alkyl chain showed the best in vitro
DNA transfection. However, this result could not be linked to other parameters. Formulations employing co-lipids DOPE or cholesterol
revealed no trend in transfection with the lengths of saturated and polyunsaturated chains.
Lipoplexes of achiral C30-20 with a rigid polyene chain and a flexible saturated chain behaved in transfecting CHO-K1 cells comparable
to reference (R)-EPC with two flexible saturated chains. Consequently, avoiding chirality and introducing a rigid polyene chain had no
obvious effect within our reference frame.
The periodicity in the bilayers of EPC with saturated chains is disturbed in compounds with a rigid polyene chain as established by
SAXRD. However, the presence or absence of an ordered microstructure had no influence on transfections when comparing C30-20 with
EPC.
Although selected members of the new lipids with low cytotoxicity offer promising gene delivery performance providing a basis for
further investigating a structure–activity connection could not be established.
As stated previously, the magnitude of biological and chemical variables have so far prevented establishing an unambiguous struc-
ture–activity connection (Koynova and Tenchov, 2010; Zhi et al., 2010, 2013). This report has demonstrated that the reason for the lack of
regularities between structural and physical data, and transfection of the polyene gene carriers is caused by their uncontrollable molecular
associations. This finding supports a complementing conclusion articulated in a recent investigation on lipid packing and transfection
(Moghaddam et al., 2011).
A straightforward structure–activity correlation may only be established when the outcome of the self-assembly processes, i.e. size,
fine structure, morphology, DNA compacting, could be controlled, e.g. by relying on liposomes and lipoplexes of predefined shape and size
(Svenson, 2004). The preparation of liposomes with defined size and shape will be the topic of a forthcoming communication.
461
462
Conflict of Interest
The authors declare that there are no conflicts of interest.
463
Acknowledgements
464
465
466
467
We thank Dr. Hansgeorg Ernst, BASF SE, Ludwigshafen (Germany) for a generous gift of C30-ester, and Dr. Susana V. Gonzales (NTNU)
and Dr. Zhihua Yang (Stevens) for performing mass spectrometric determinations. Synchrotron beam time was granted by the European
Synchrotron Radiation Facility, Grenoble, France, using the facilities of the BIOSAXS beamline (BM29). We are grateful to Adam Round and
Christoph Mueller-Dieckmann for technical assistance in setting up the experiments for the SAXS/SAXRD data collection. AZ thanks the
46Q8 6 Higher Education Commission of Pakistan for a scholarship. This work was made possible by a grant from the Qatar National Research Fund
469
under the National Priorities Research Program, award NPRP08-705-3-144 (LPI: M. Pungente). Its contents are solely the responsibility of
the authors and do not necessarily represent the official views of the Qatar National Research Fund.
470
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
19
471
Appendix A. Supplementary data
472
473
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemphyslip.2014.04.
006.
474
References
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
Adami, R.C., Seth, S., Harvie, P., Johns, R., Fam, R., Fosnaugh, K., Zhu, T.Y., Farber, K., McCutcheon, M., Goodman, T.T., Liu, Y., Chen, Y., Kwang, E., Templin, M.V., Severson, G.,
Brown, T., Vaish, N., Chen, F., Charmley, P., Polisky, B., Houston, M.E., 2011. An amino acid-based amphoteric liposomal delivery system for systemic administration of
siRNA. Mol. Ther. 19, 1141–1151.
Almofti, M.R., Harashima, H., Shinohara, Y., Almofti, A., Li, W.H., Kiwada, H., 2003. Lipoplex size determines lipofection efficiency with or without serum. Mol. Membr. Biol.
20, 35–43.
Antipina, M.N., Schulze, I., Heinze, M., Dobner, B., Langner, A., Brezesinski, G., 2009. Physical–chemical properties and transfection activity of cationic lipid/DNA complexes.
ChemPhysChem 10, 2471–2479.
Balbino, T.A., Gasperini, A.A.M., Oliveira, C.L.P., Azzoni, A.R., Cavalcanti, L.P., de La Torre, L.G., 2012. Correlation of the physicochemical and structural properties of pDNA/cationic
liposome complexes with their in vitro transfection. Langmuir 28, 11535–11545.
Bergelson, L.D., 1970. Diol lipids. Prog. Chem. Fats Lipids 10, 241–286.
Breukers, S., Øpstad, C.L., Sliwka, H.R., Partali, V., 2009. Hydrophilic carotenoids: surface properties and aggregation behavior of the potassium salt of the highly unsaturated
diacid norbixin. Helv. Chim. Acta 92, 1741–1747.
Byun, H.S., Bittman, R., 1996. Efficient stereospecific synthesis of diamide analogs of phosphatidylcholine starting from 1-(4ꢀ-methoxyphenyl)-sn-glycerol. J. Org. Chem. 61,
8706–8708.
Chang, C.D., Siegel, C., Lee, E., Harris, D.J., 1997. Intermolecular acyl group exchange between cationic lipid and co-lipid of cationic lipid-based gene transfer agents. Abstr.
Pap. 214 Nat. Meeting, Am. Chem. Soc., 19-ANYL.
Chu, Y., Masoud, M., Gebeyehu, G., 2009. US Patent 7,479,573. Transfection Reagents, Invitrogen, USA.
Coste, J., Frérot, E., Jouin, P., 1994. Coupling N-methylated amino-acids using pybrop and pyclop halogenophosphonium salts – mechanism and fields of application. J. Org.
Chem. 59, 2437–2446.
Dabkowska, A.P., Barlow, D.J., Hughes, A.V., Campbell, R.A., Quinn, P.J., Lawrence, M.J., 2012. The effect of neutral helper lipids on the structure of cationic lipid monolayers. J.
R. Soc. Interface 9, 548–561.
Felgner, P.L., Gadek, T.R., Holm, M., Roman, R., Chan, H.W., Wenz, M., Northrop, J.P., Ringold, G.M., Danielsen, M., 1987. Lipofection – a highly efficient, lipid-mediated
DNA-transfection procedure. Proc. Natl. Acad. Sci. U.S.A. 84, 7413–7417.
Floch, V., Loisel, S., Guenin, E., Hervé, A.C., Clément, J.C., Yaouanc, J.J., des Abbayes, H., Férec, C., 2000. Cation substitution in cationic phosphonolipids: a new concept to
improve transfection activity and decrease cellular toxicity. J. Med. Chem. 43, 4617–4628.
Foss, B.J., Nalum Naess, S., Sliwka, H.R., Partali, V., 2003. Stable and highly water-dispersible, highly unsaturated carotenoid phospholipids – surface properties and aggregate
size. Angew. Chem. Int. Ed. 42, 5237–5240.
Foss, B.J., Sliwka, H.R., Partali, V., Cardounel, A.J., Zweier, J.L., Lockwood, S.F., 2004. Direct superoxide anion scavenging by a highly water-dispersible carotenoid phospholipid
evaluated by electron paramagnetic resonance (EPR) spectroscopy. Bioorg. Med. Chem. Lett. 14, 2807–2812.
Foss, B.J., Sliwka, H.R., Partali, V., Köpsel, C., Mayer, B., Martin, H.D., Zsila, F., Bikadi, Z., Simonyi, M., 2005a. Optically active oligomer units in aggregates of a highly unsaturated,
optically inactive carotenoid phospholipid. Chem. Eur. J. 11, 4103–4108.
Foss, B.J., Sliwka, H.R., Partali, V., Naess, S.N., Elgsaeter, A., Melø, T.B., Naqvi, K.R., 2005b. Hydrophilic carotenoids: surface properties and aggregation behavior of a highly
unsaturated carotenoid lysophospholipid. Chem. Phys. Lipids 134, 85–96.
Foss, B.J., Sliwka, H.R., Partali, V., Naess, S.N., Elgsaeter, A., Melø, T.B., Naqvi, K.R., O‘Malley, S., Lockwood, S.F., 2005c. Hydrophilic carotenoids: surface properties and aqueous
aggregation of a rigid, long-chain, highly unsaturated dianionic bolaamphiphile with a carotenoid spacer. Chem. Phys. Lipids 135, 157–167.
Ginn, S.L., Alexander, I.E., Edelstein, M.L., Abedi, M.R., Wixon, J., 2013. Gene therapy clinical trials worldwide to 2012 – an update. J. Gene Med. 15, 65–77.
Goldberg, M.S., (Ph.D. Thesis) 2008. Screening, Synthesis, and Applications of “Lipidoids”, A Novel Class of Molecules Developed for the Delivery of RANi Therapeutics,
Chemistry. Massachusets Institute of Technology, Boston, pp. 35.
Hirsch-Lerner, D., Zhang, M., Eliyahu, H., Ferrari, M.E., Wheeler, C.J., Barenholz, Y., 2005. Effect of “helper lipid” on lipoplex electrostatics. Biochim. Biophys. Acta 1714, 71–84.
Horn, D., Rieger, J., 2001. Organic nanoparticles in the aqueous phase – theory, experiment, and use. Angew. Chem. Int. Ed. 40, 4331–4361.
Ivanova, E.A., Maslow, M.A., Kabilova, T.O., Puchkov, P.A., Alekseeva, A.S., Boldyrev, I.A., Vlassov, V.V., Serebrennikova, G.A., Morozova, N.G., Zenkova, M.A., 2013. Struc-
ture–transfection acitivity relationships in a series of novel cationic lipids with heterocycic head-groups. Org. Biomol. Chem. 11, 7164–7178.
Jones, C.H., Chen, C.K., Ravikrishnan, A., Rane, S., Pfeifer, B.A., 2013. Overcoming nonviral gene delivery barriers: perspective and future. Mol. Pharm. 10, 4082–4098.
Kedika, B., Patri, S.V., 2011. Design, synthesis, and in vitro transfection biology of novel tocopherol based monocationic lipids: a structure–activity investigation. J. Med. Chem.
54, 548–561.
Konarev, P.V., Volkov, V.V., Sokolova, A.V., Koch, M.H.J., Svergun, D.I., 2003. PRIMUS: a Windows PC-based system for small-angle scattering data analysis. J. Appl. Crystallogr.
36, 1277–1282.
Koynova, R., 2010. Analysis of lipoplex structure and lipid phase changes. Methods Mol. Biol. 606, 399–423.
Koynova, R., Tenchov, B., 2009. Cationic phospholipids: structure–transfection activity relationships. Soft Matter 5, 3187–3200.
Koynova, R., Tenchov, B., 2010. Cationic lipids: molecular structure/transfection activity relationships and interactions with biomembranes. Top. Curr. Chem. 296, 51–93.
Koynova, R., Tenchov, B., Wang, L., MacDonald, R.C., 2009. Hydrophobic moiety of cationic lipids strongly modulates their transfection activity. Mol. Pharm. 6, 951–958.
Larsen, E., Abendroth, J., Partali, V., Schulz, B., Sliwka, H.R., Quartey, E.G.K., 1998. Combination of vitamin E with a carotenoid: alpha-tocopherol and trolox linked to beta-
apo-8ꢀ-carotenoic acid. Chem. Eur. J. 4, 113–117.
Li, W.J., Szoka, F.C., 2007. Lipid-based nanoparticles for nucleic acid delivery. Pharm. Res. 24, 438–449.
Liberska, A., Unciti-Broceta, A., Bradley, M., 2009. Very long-chain fatty tails for enhanced transfection. Org. Biomol. Chem. 7, 61–68.
Loizeau, D., Le Gall, T., Mahfoudhi, S., Berchel, M., Maroto, A., Yaouanc, J.J., Jaffres, P.A., Lehn, P., Deschamps, L., Montier, T., Giamarchi, P., 2013. Physicochemical properties of
cationic lipophosphoramidates with an arsonium head group and various lipid chains: a structure–activity approach. Biophys. Chem. 171, 46–53.
Mangroo, D., Gerber, G.E., 1988. Phospholipid synthesis – effects of solvents and catalysts on acylation. Chem. Phys. Lipids 48, 99–108.
Mathematica, 2010. Mathematica Version 8.0. Wolfram Research Inc., University of Illinois Press, Champaign, IL.
Merkel, O.M., Mintzer, M.A., Librizzi, D., Samsonova, O., Dicke, T., Sproat, B., Garn, H., Barth, P.J., Simanek, E.E., Kissel, T., 2010. Triazine dendrimers as nonviral vectors for
in vitro and in vivo RNAi: the effects of peripheral groups and core structure on biological activity. Mol. Pharm. 7, 969–983.
Mineva, T., Krishnamurty, S., Salahub, D.R., Goursot, A., 2013. Temperature dependence of the molecular conformations of dilauroyl phosphatidylcholine: a density functional
study. Int. J. Quantum Chem. 113, 631–636.
Moghaddam, B., McNeil, S.E., Zheng, Q., Mohammed, A.R., Perie, Y., 2011. Exploring the correlation between lipid packing in lipoplexes and their transfection efficacy.
Pharmaceutics 3, 846–848.
Niculescu-Duvaz, D., Heyes, J., Springer, C.J., 2003. Structure–activity relationship in cationic lipid mediated gene transfection. Curr. Med. Chem. 10, 1233–1261.
Øpstad, C.L., Sliwka, H.R., Partali, V., Elgsaeter, A., Leopold, P.L., Jubeli, E., Khalique, N.A., Raju, L., Pungente, M.D., 2013. Synthesis, self-assembling and gene delivery potential
of a novel highly unsaturated, conjugated cationic phospholipid. Chem. Phys. Lipids 170–171, 65–73.
Popplewell, L.J., Abu-Dayya, A., Khanna, T., Flinterman, M., Khalique, N.A., Raju, L., Øpstad, C.L., Sliwka, H.R., Partali, V., Dickson, G., Pungente, M.D., 2012. Novel cationic
carotenoid lipids as delivery vectors of antisense oligonucleotides for exon skipping in Duchenne muscular dystrophy. Molecules 17, 1138–1148.
Pozzi, D., Marchini, C., Cardarelli, F., Amenitsch, H., Garulli, C., Bifone, A., Caracciolo, G., 2012. Transfection efficiency boost of cholesterol-containing lipoplexes. Biochim.
Biophys. Acta 1818, 2335–2343.
Predvoditilev, D.A., Suvorkin, S.V., Nifant’ev, E.E., 2001. New approach to the synthesis of phosphamide models of cationic phosphatidyl cholines. Russ. J. Gen. Chem. 71,
873–880.
Rosenzweig, H.S., Rakhmanova, V.A., McIntosh, T.J., MacDonald, R.C., 2000. O-alkyl dioleoylphosphatidylcholinium compounds: the effect of varying alkyl chain length on
their physical properties and in vitro DNA transfection activity. Bioconjug. Chem. 11, 306–313.
Ross, P.C., Hui, S.W., 1999. Lipoplex size is a major determinant of in vitro lipofection efficiency. Gene Ther. 6, 651–659.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006

G Model
CPL42881–20
ARTICLE IN PRESS
C.L. Øpstad et al. / Chemistry and Physics of Lipids xxx (2014) xxx–xxx
20
552
Ryhänen, S., 2006. Biophysical Studies on Cationic Liposomes. University of Helsinki, Finland, pp. 55.
55Q3 7 Saha, K., Kim, S.T., Yan, B., Miranda, O.R., Alfonso, F.S., Shlosman, D., Rotello, V.M., 2012. Surface functionality of nanoparticles determines the cellular uptake mechanisms in
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
mammalian cells. Small.
Scheule, R.K., Bagley, R.G., Eastman, S.J., Cheng, S.H., Marshall, J., Yew, N.S., Harris, D.J., Lee, E.R., Siegel, C.S., Chang, C.D., Hubbard, C.S., 1998. Cationic Amphiphile/DNA
Complexes. Genzyme Corporation, USA.
Shinoda, K., 1963. Colloidal Surfactants. Some Physial Chemical Properties. Academic Press, New York.
Skrylev, L.D., Streltsova, E.A., Skryleva, T.L., 2000. Adsorption-micellar energy ratio as criterion for predicting surfactant performance. Russ. J. Appl. Chem. 73, 1364–1367.
Sliwka, H.R., 1997. Selenium carotenoids 3: first synthesis of optically active carotenoid phosphates. Acta Chem. Scand. 51, 345–347.
Sliwka, H.R., 1999. Conformation and circular dichroism of beta,beta-carotene derivatives with nitrogen-, sulfur-, and selenium-containing substituents. Helv. Chim. Acta 82,
161–169.
Sliwka, H.R., Partali, V., Lockwood, S.F., 2010. Hydrophilic carotenoids: carotenoid aggregates. In: Landrum, J.T. (Ed.), Carotenoids – Physical, Chemical and Biological Funtions
and Properties. CRC Press, Boca Raton, p. 47.
Svenson, S., 2004. Controlling surfactant self-assembly. Curr. Opin. Colloid Interface Sci. 9, 201–212.
Wang, C., Berg, C.J., Hsu, C.C., Merrill, B.A., Tauber, M.J., 2012. Characterization of carotenoid aggregates by steady-state optical spectroscopy. J. Phys. Chem. B 116, 10617–10630.
Yoshimura, T., Chiba, N., Matsuoka, K., 2012. Supra-long chain surfactants with double or triple quaternary ammonium headgroups. J. Colloid Interface Sci. 374, 157–163.
Zhi, D., Zhang, S., Cui, S., Zhao, Y., Wang, Y., Zhao, D., 2013. Headgroup evolution of cationic lipids for gene delivery. Bioconjug. Chem. 24, 487–519.
Zhi, D.F., Zhang, S.B., Wang, B., Zhao, Y.N., Yang, B.L., Yu, S.J., 2010. Transfection efficiency of cationic lipids with different hydrophobic domains in gene delivery. Bioconjug.
Chem. 21, 563–577.
Please cite this article in press as: Øpstad, C.L., et al., Novel cationic polyene glycol phospholipids as DNA transfer
reagents—Lack of a structure–activity relationship due to uncontrolled self-assembling processes. Chem. Phys. Lipids (2014),
http://dx.doi.org/10.1016/j.chemphyslip.2014.04.006