Structure-activity relationship studies of symmetrical cationic bolasomes as non-viral gene vectors

Article abstract of DOI:10.1039/c6tb00870d

Compared to traditional cationic lipids, bola-type lipids have received much less attention despite their advantages including the ability to form more stable and regular-shaped liposomes. In this report, a series of novel symmetric cationic bolalipids based on lysine or cyclen headgroups were designed and synthesized. Structure-activity relationships including the effect of the hydrophobic chain length and cationic headgroup on liposome formation, DNA binding, the physical property of bolasomes, and gene transfection were systematically studied. Results reveal that an appropriate hydrophobic chain length is essential to form nano-sized bolasomes with good DNA binding and condensation ability. MTS-based cell viability assays showed low cytotoxicity of these bolasome/DNA complexes. Lys-14-10, which has a 36-atom-length hydrophobic chain, exhibited the best transfection efficiency in the two cell lines. Flow cytometry and confocal laser microscopy assays reveal that the bolaplexes formed from bolalipids with such a chain might induce the highest cellular uptake. For the cationic headgroup, lysine is more suitable than cyclen for such a bola-type vector. Although the TEs of these bolalipids are still lower than commercially used non-bola lipid lipofectamine 2000, this study may give us some clues for the design of novel bolalipids with higher TE and biocompatibility.

Full text of DOI:10.1039/c6tb00870d

Journal of
Materials Chemistry B
View Article Online
View Journal
PAPER
Structure–activity relationship studies of
symmetrical cationic bolasomes as non-viral gene
Cite this:DOI: 10.1039/c6tb00870d
vectors†
Zheng Huang, Yi-Mei Zhang, Qian Cheng, Ji Zhang,* Yan-Hong Liu, Bing Wang
and Xiao-Qi Yu*
Compared to traditional cationic lipids, bola-type lipids have received much less attention despite their
advantages including the ability to form more stable and regular-shaped liposomes. In this report, a
series of novel symmetric cationic bolalipids based on lysine or cyclen headgroups were designed and
synthesized. Structure–activity relationships including the effect of the hydrophobic chain length and
cationic headgroup on liposome formation, DNA binding, the physical property of bolasomes, and gene
transfection were systematically studied. Results reveal that an appropriate hydrophobic chain length is
essential to form nano-sized bolasomes with good DNA binding and condensation ability. MTS-based
cell viability assays showed low cytotoxicity of these bolasome/DNA complexes. Lys-14-10, which has a
3
6-atom-length hydrophobic chain, exhibited the best transfection efficiency in the two cell lines. Flow
Received 8th April 2016,
Accepted 15th July 2016
cytometry and confocal laser microscopy assays reveal that the bolaplexes formed from bolalipids with
such a chain might induce the highest cellular uptake. For the cationic headgroup, lysine is more
suitable than cyclen for such a bola-type vector. Although the TEs of these bolalipids are still lower than
commercially used non-bola lipid lipofectamine 2000, this study may give us some clues for the design
of novel bolalipids with higher TE and biocompatibility.
DOI: 10.1039/c6tb00870d
www.rsc.org/MaterialsB
1
3
tremendously restricted their applications in clinical trials.
Consequently, alternative delivery vectors have been proposed based
Introduction
1
4,15
16–19
Over the past decades, profiting from the prosperity and devel- on synthetic materials such as cationic lipids,
polymers,
2
0,21
22
opment of gene delivery vectors, gene therapy has made signifi- dendrimers,
cant progress as a potential method for treating many acquired Amongst the non-viral delivery systems reported in the litera-
diseases and genetic disorders such as AIDS, cystic fibrosis and ture, cationic lipids, as the most explored type of non-viral
and peptides.
1
–3
Parkinson’s disease. Meanwhile, these vectors can also load vectors, have shown great potential in gene delivery because of
different kinds of therapeutic cargos, which provide an alterna- their lower immunogenic nature, ability to deliver large pieces of
23
tive method to traditional chemotherapy used in treating nucleic acids and ease of handling and preparation. The vast
4
–6
cancers.
The gene vectors can effectively protect its cargos majority of traditional lipids researched so far are ‘‘monopolar’’
(
nucleic acids) from degradation by nuclease and deliver them to amphiphiles. The chemical structure of these molecules generally
7
the interior of the target cells. Broadly speaking, the vectors for comprises just one polar headgroup, which is attached to hydro-
gene delivery are mainly classified into two categories: viral and phobic tails via a linker. In comparison, bipolar amphiphiles,
8
24
non-viral (or synthetic) vectors. Although viral vectors, including which are also called bolalipids, are reported much less. In
9
,10
retroviruses and adenoviruses,
are efficient in delivering detail, bolalipids are composed of one or two hydrophobic chains
nucleic acids to numerous cell lines, they have several drawbacks that are covalently linked at both ends to hydrophilic head
2
5–27
such as immunogenicity, inflammatory reactions and problems groups.
with respect to scale-up procedures.
Bolalipids can generate monolayer lipid membranes
1
,11,12
These limitations (MLMs) in solution, which may form spherical nanoparticles
26,28
under ultrasonication.
This type of molecular architecture
mainly presents in archaebacterial membranes and may ensure
their survival in extreme environments (e.g. high temperature and
low pH) due to its exceptional stability compared to the traditional
Key Laboratory of Green Chemistry & Technology (Ministry of Education),
College of Chemistry, Sichuan University, Chengdu 610064, P. R. China.
E-mail: xqyu@scu.edu.cn, jzhang@scu.edu.cn; Fax: +86-28-85415886
24,29,30
double-layered membranes.
These amphiphilic compounds
†
Electronic supplementary information (ESI) available: Supplemental figures
and compounds characterization. See DOI: 10.1039/c6tb00870d
have been used in numerous applications including gene delivery.
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. B

View Article Online
Paper
Journal of Materials Chemistry B
It was reported that the bolasomes, which were formed from measured on a Bruker AM400 NMR spectrometer. Proton
bolalipids with the presence of a helper fusogenic lipid such as Chemical shifts of NMR spectra were given in ppm relative to
1
,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), are capable internal reference TMS (1H, 0.00 ppm). HRMS spectral data was
31,32
of delivering genes to target cells.
utilizing cationic bolalipids as non-viral gene delivery agents was
carried out by Eaton et al. Weissig and co-workers also designed scattering (DLS) experiments, transmission electron microscopy
One of the earliest studies recorded on a Bruker Daltonics Bio TOF mass spectrometer.
The ethidium bromide replacement assay, dynamic light
33
and synthesized a series of bolalipids based on dequalinium (TEM), amplification and purification of plasmid DNA, cell
3
4,35
(
DQAsomes) for mitochondria-targeting gene delivery.
From culture, these procedures were performed according to our
1
6,44
then on, a variety of cell targeting bolalipids designed for gene reported literature.
3
6–41
delivery have been reported.
Most recently, Shapiro et al.
synthesized two bolalipid compounds with positively charged
headgroups from jojoba oil, and they studied the aggregate
Synthesis of the bolalipids
Preparation of compound 2. Compound 1 (1,6-hexanediol,
structures using molecular dynamics (MD) simulations and 1,8-octanediol, 1,10-decanediol or 1,12-dodecanediol, 20 mmol)
4
2
utilized them as carriers for siRNA delivery.
and imidazole (1.28 g, 20 mmol) were dissolved in dry dichloro-
Although there are some reports about bolalipid-based non- methane; a solution of tert-butyldimethylsilyl chloride (TBS-Cl,
viral gene delivery vectors, the transfection efficiency (TE) is 2.40 g, 16 mmol) in anhydrous dichloromethane (120 mL) was
not very satisfying and the structure–activity relationship of added dropwise under an ice bath. The obtained mixture was
bola-type vectors needs further study. Herein we designed and then stirred at room temperature overnight. After dilution with
synthesized a series of symmetric cationic bolalipids. The Et O (120 mL), the solution was washed with water (3 ꢀ 80 mL),
2
hydrophobic skeleton was constructed by the condensation of dried over anhydrous Na SO , filtered and evaporated under
2
4
saturated aliphatic dicarboxylic acids and a,o-diols. Then hydro- reduced pressure. The residue was purified by silica gel column
philic cationic headgroups (cyclen or lysine) were linked on the chromatography (PE : EA = 4 : 1, v/v) to give the product 2 as a
two ends by esterification to gain the final products. Bolasomes were colorless liquid.
prepared by using these molecules with DOPE. Their interaction
Preparation of compound 4. Saturated aliphatic dicarboxylic
with DNA and gene delivery abilities were systematically studied, acids (adipic acid, suberic acid, sebacic acid, dodecanedioic
and the structure–activity relationship (SAR) was discussed. This acid or tetradecanedioic acid, 4.5 mmol) were activated in the
work may afford new strategy for the construction of bola-type presence of dicyclohexylcarbodiimide (DCC, 1.86 g, 9.0 mmol)
lipids and may give some clues for further improvements of and N,N-dimethylaminopyridine (DMAP, 0.11 g, 1.8 mmol) in
bola-type gene vectors.
50 mL anhydrous CH Cl in an ice-salt-bath at 0 1C for 30 min.
2 2
Then product 2 (9.0 mmol) was added and the mixture was
stirred at room temperature overnight. The DCC condensation
mixture was then filtered, and the filtrate was evaporated under
reduced pressure, and a little amount of ethyl acetate was
added. The mixture was maintained at 0 1C for half an hour
Experimental section
Materials and methods
All of the common chemicals and reagents were obtained and then filtered. The filtrate was evaporated under reduced
commercially and were used as received. Anhydrous chloroform pressure to give the crude product which was purified by
and dichloromethane were dried and purified under nitrogen by column chromatography over silica gel (PE : EA = 20 : 1, v/v) to
using standard methods and were distilled immediately before use. yield the precursors of 4.
[
4,7,10-Tris(tert-butoxycarbonyl)-1,4,7,10-tetraaza-cyclododecan-1-yl]
Subsequently, 10 drops of BF
3
ꢁEt
2
O were added to the anhydrous
acetic acid (3Boc-cyclen acetic acid) and 2Boc-lysine were prepared dichloromethane solution of the previously obtained precursor
25,43
according to the literature.
cyclen) was purchased from Quzhou Synpartner Pharmaceutical concentrated to afford an oil, which was further purified by
1,4,7,10-Tetraazacyclododecanes and stirred at room temperature for 2 h. The mixture was
(
Technology Co., Ltd. 1,2-Dioleoyl-sn-glycero-3-phosphoethanol- column chromatography on silica gel (EA : PE = 4 : 1, v/v) to yield
amine (DOPE) was purchased from Avanti Polar Lipids, Inc. 4 as a colorless liquid or white powder.
pUC19 DNA was purchased from TIANTAI (Chengdu, China).
Preparation of compound 6. Compound 4 (1.0 mmol) was
MicroBCA protein assay kit was obtained from Pierce (Rockford, added to the anhydrous dichloromethane solution (60 mL) con-
IL, USA). Luciferase assay kit and MTS (3-(4,5-dimethylthiazol-2- taining 3Boc-cyclen acetic acid (1.12 g, 2.2 mmol) or 2Boc-lysine
yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H tetrazolium, (760.4 mg, 2.2 mmol), DCC (454.0 mg, 2.2 mmol) and DMAP
inner salt) were purchased from Promega (Madison, WI, USA). (24.3 mg, 0.2 mmol) in an ice-salt-bath at 0 1C for 1 h and then
Endotoxin-free plasmid purification kit was purchased from stirred at room temperature overnight. The DCC condensation
TIANGEN (Beijing, China). The plasmids used in the study were mixture was then filtered, and the filtrate was evaporated under
pGL-3 (Promega, Madison, WI, USA) coding for luciferase and reduced pressure, and a little amount of ethyl acetate was added.
pEGFP-N1 (Clontech, Palo Alto, CA, USA) coding for EGFP. The The mixture was maintained at 0 1C for half an hour and filtered.
Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine The filtrate was evaporated under reduced pressure to give the
s
serum (FBS) and Lipofectamine 2000 were purchased from crude products which were purified by column chromatography
1
13
Invitrogen Corp. The H NMR and C NMR spectra were over silica gel (PE : EA = 1 : 1, v/v) to yield a white solid 6.
J. Mater. Chem. B
This journal is ©The Royal Society of Chemistry 2016

View Article Online
Journal of Materials Chemistry B
Paper
Preparation of bolalipids 7. Compound 6 (0.5 mmol) was Cytotoxicity assays
suspended in anhydrous dichloromethane (5 mL), and then, a
Toxicities of bolaplexes toward HEK 293 cells and HeLa cells
were determined by using the MTS reduction assay following
solution of trifluoroacetic acid (5 mL) in anhydrous dichloro-
methane (5 mL) was added dropwise under an ice bath. And
then, the obtained mixture was stirred at room temperature for
4
literature procedures. About 1.0 ꢀ 10 cells per well were
seeded into 96-well plates. After 24 h, optimized bola/DOPE
formulations were completed with 0.2 mg of pEGFP-N1 DNA at
various N/P ratios for 30 min; 100 mL of bolaplexes were added
to the cells in the absence of serum. After 4 h of incubation,
bolaplex solutions were removed, and 100 mL of media with
6
h. After the solvent and trifluoroacetic acid were removed, the
bola-like lipids 7 were directly obtained as a colourless liquid by
treating the residues with anhydrous ethyl ether twice.
The analysis data of the above compounds are listed in
the ESI.†
10% FBS was added. After 24 h, 20 mL of MTS and 100 mL of PBS
were added to each well and the plates were incubated at 37 1C
for another 1 h. Then, the absorbance of each sample was
measured using an ELISA plate reader (model 680, BioRad) at
a wavelength of 490 nm. The cell viability (%) was obtained
according to the manufacturer’s instruction. Lipoplex prepared
from Lipofectamine 2000 was used as a control.
Formation of bolasomes and bolaplexes
Bolalipid 7 (0.0025 mmol) or its mixture with DOPE in the desired
mole ratio was dissolved in anhydrous chloroform (2.5 mL) in
autoclaved glass vials. Thin films were made by slowly rotary-
evaporating the solvent at room temperature. The last trace of
organic solvent was removed by keeping these films under
vacuum over 8 h. The dried films and 2.5 mL of Tris–HCl buffer
In vitro transfection procedure
(10 mM, pH 7.4) were preheated to 70 1C, and then the buffer
was added to the films resulting in a final bola concentration of In order to obtain about 80% confluent cultures at the time of
5
1
.0 mM. The mixtures were vortexed vigorously until the films transfection, 24-well plates were seeded with 1.0 ꢀ 10 cells per
were completely resuspended. Sonication of these suspensions well in 500 mL of antibiotic-free media 24 h before transfection.
for 5 min in a bath sonicator at 0 1C afforded the corresponding For the preparation of bolaplexes applied to cells, various
cationic bolasomes, which were then stored at 4 1C.
amounts of bolasomes and DNA were serially diluted separately
To prepare the bolaplexes, various amounts of bolasomes in antibiotic-free DMEM culture medium; then, the DNA solu-
were mixed with a constant amount of DNA by pipetting tions were added into bolasome solutions and mixed briefly by
thoroughly at various N/P ratios, and the mixture was incubated pipetting up and down several times, after which the mixtures
for 30 min at room temperature. The theoretical N/P ratio were incubated at room temperature for about 30 min to obtain
represents the charge ratio of cationic bolalipids to nucleotide bolaplexes of desired N/P ratios, the final bolaplex volume was
base (in mole) and was calculated by considering an average 100 mL, and DNA was used at a concentration of 0.8 mg per well.
nucleotide mass of 309.
After 30 min of complexation, the old cell culture medium was
removed from the wells, cells were washed once with serum-
free DMEM, and the above 100 mL of bolaplexes were added to
Gel retardation assay
To determine the formation of the bola/pDNA complex each well. The plates were then incubated for 4 h at 37 1C in a
bolaplexes), bolaplexes of various N/P ratios ranging from humidified atmosphere containing 5% CO . At the end of the
(
2
0
to 8 were prepared as described above. A constant amount incubation period, the medium was removed, and 500 mL of
of 0.125 mg of DNA was used here; 15 mL of each bolaplex fresh DMEM medium containing 10% FBS was added to each
ꢂ1
solution was electrophoresed on a 1% (W V ) agarose gel well. Plates were further incubated for a period of 24 h before
containing Gel-Red and Tris–aceate (TAE) running buffer at checking the reporter gene expression.
1
35 V for 30 min. DNA was visualized with a UV lamp using a
For fluorescent microscopy assays, cells were transfected by
complexes containing pEGFP-N1. After a 24 h incubation, the
microscopy images were obtained at a magnification of 100 and
recorded using Viewfinder Lite (1.0) software. Control transfec-
BioRad Universal Hood II.
Determination of the critical micelle concentration
The critical micelle concentration (CMC) was determined using tion was performed in each case using a commercially available
pyrene as a fluorescence probe. The block copolymer concen- transfection reagent Lipofectamine 2000t based on the standard
ꢂ6
ꢂ1
ꢂ1
tration varied from 1.0 ꢀ 10 mg mL to 1 mg mL , and the conditions specified by the manufacturer. After a 24 h transfection
ꢂ7
pyrene concentration was fixed at 6.0 ꢀ 10 M. The fluorescence of the pEGFP plasmid, cells were lysed and 100 mL of lysate was
spectra were recorded using a HITACHI F-7000 Fluorescence taken to measure the fluorescence intensity. The excitation wave-
Spectrophotometer. Both the emission and excitation slit widths length was 485 nm and the emission wavelength was 538 nm.
were 5 nm. The samples were excited at 335 nm and the emission Lipofectamine 2000t was chosen as a control.
spectra were recorded from 350 to 500 nm. The emission fluores-
For luciferase assays, cells were transfected by complexes
cence values, I373 and I383 at 373 nm and 383 nm, respectively, were containing pGL-3. For a typical assay in a 24-well plate, 24 h
used for the subsequent calculations. The CMC was determined post transfection as described above, the old medium was
from the plots of the I383/I373 ratio versus the logarithm of the removed from the wells, and the cells were washed twice with
polymer concentration using the intersection of the linear 500 mL of prechilled PBS. According to the luciferase assay kit
regression lines as the CMC value.
(promega) manufacture, 100 mL of 1ꢀ cell lysis buffer diluted
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. B

View Article Online
Paper
Journal of Materials Chemistry B
with PBS was then added to each well, and the cells were lysed into the well for another 30 min incubation at 37 1C. Then
for 30 min in a horizontal rocker at room temperature. The cell cells were rinsed 3 times with PBS (pH 7.4), fixed with 4%
lysate was transferred completely to Eppendorf tubes and paraformaldehyde (dissolved with PBS buffer) for 10 min,
centrifuged (4000 rpm, RT) for 2 min; the supernatant was lysosome staining was done with LysoTracker green. The CLSM
transferred to Eppendorf tubes and stored in ice. For the assay, observation was performed using confocal laser scanning
20 mL of this supernatant and 100 mL of the luciferase assay microscopy (CLSM, ZEISS LSM 780) at excitation wavelengths
substrate (Promega) were used. The lysate and the substrate of 504 nm for LysoTracker green (green) and 633 nm for Cy5
were both thawed to RT before performing the assay. The (red), respectively.
substrate was added to the lysate, and the luciferase activity was
measured in a luminometer (Turner designs, 20/20, Promega,
USA) in standard single-luminescence mode. The integration time Results and discussion
of measurement was 10 000 ms. A delay of 2 s was given before
Synthesis of the bolalipids
each measurement. The protein concentration in the cell lysate
A series of novel symmetrical cationic bolalipids containing a
supernatant was estimated in each case with the Lowry protein
protonated cyclen or lysine headgroup and various hydrophobic
assay kit (PIERCE, Rockford, IL, USA). A comparison of the
cores were synthesized. The length of the hydrophobic skeleton
was adjusted by using dicarboxylic acids and a,o-diols with
various chain lengths. The use of cyclen as the hydrophilic head
transfection efficiencies of the individual bolas was made based
on the data for luciferase expressed as relative light units (RLU)
per mg of protein. All the experiments were done in triplicates,
group is based on its unique structural characteristics. Cyclen
and the results presented are the average of at least two such
has four amine groups with different chemical properties such
independent experiments done on the same days.
as pK values. Such diversity makes cyclen a potential candidate
a
Flow cytometry assay
for both DNA condensation and pH buffering. Meanwhile, the
cyclic backbone that is hard to self-fold can also retain its high
DNA binding ability. Our previous studies demonstrated that
cyclen-based cationic lipids may act as promising materials
The cellular uptake of the bolasome/fluorescein labelled-DNA
complexes was analyzed by flow cytometry. The label IT Cy5
Labeling Kit was used to label pGL-3 with Cy5 according to the
manufacturer’s protocol. Briefly, HEK 293 cells were seeded
1
4,15,43
for gene delivery.
Lysine has good biocompatibility and
5
excellent DNA binding ability for its two primary amines, and it
was also widely used for the cationic headgroup of lipidic
onto 12-well plates (2.0 ꢀ 10 cells per well) and allowed
to attach and grew for 24 h. Before transfection, the medium
was replaced with a serum-free culture medium. Cells were
incubated with Cy5 labelled DNA nanoparticles (1.6 mg of DNA
per well, optimal N/P ratio of each sample) in media for 4 h at
2
4,28,45
vectors.
As shown in Scheme 1, one hydroxyl of a,o-diol
was first protected by tert-butyldimethylsilyl chloride (TBSCl),
then the other hydroxyl was reacted with saturate aliphatic
dicarboxylic acid 3 in the presence of dicyclohexylcarbodiimide
37 1C. Subsequently, the cells were washed with 1ꢀ PBS and
(
DCC) and 4-dimethylaminopyridine (DMAP) to form two ester
bonds. Subsequent TBS deprotection by BF gave the hydro-
phobic skeleton diol 4 (Diol-(m+2)-n). Compound 4 was coupled with
Boc-cyclen acetic acid or 2Boc-lysine (5) to give the precursor 6.
harvested with 0.25% trypsin/EDTA and resuspended in 1ꢀ PBS.
Cy5-labelled plasmid DNA uptake was measured in the FL4
channel using the red diode laser (633 nm). Data from 10000
events were gated using forward and side scatter parameters
to exclude cell debris. The flow cytometer (BD Accurit C6) was
calibrated for each run to obtain a background level of B1% for
control samples (i.e., untreated cells).
3
3
Then target bolalipids 7 (represented as Cyc-(m+2)-n or Lys-(m+2)-n
based on the headgroup) were obtained by removing the Boc
groups with trifluoroacetic acid in anhydrous CH Cl . The hydro-
2 2
phobic chain length of the bolalipids can be easily calculated
as (m + 2) + 2n + 2, which represent carbon atoms of diacid,
carbon atoms of diol, and two oxygens in the middle ester bonds,
respectively. For example, Lys-14-10 has a 36-atom chain between
the two cationic headgroups. All novel compounds in each step
Confocal laser scanning microscopy (CLSM)
5
HEK 293 cells were seeded at a density of 2.5 ꢀ 10 cells per
well in a 35 mm confocal dish (F = 15 mm), 24 h prior to
transfection. For transfection in the absence of serum, the
medium was exchanged with serum-free medium (for transfection
with serum, the medium was exchanged with serum-containing
medium). Complexes of bolasomes and Cy5-labelled pGL-3 at a
given concentration were added to each well. After incubation at
1
were purified and their structures were confirmed by H NMR,
13
C NMR and HRMS. The critical micelle concentrations (CMCs)
of two typical bolalipids Lys-14-10 and Cyc-14-10 were measured
to study their natural aggregation behaviors via fluorescence
46
spectrometry using a pyrene probe. Their CMCs were found to
37 1C for 4 h, cells were rinsed 3 times with PBS (pH 7.4), fixed
be 47.4 and 24.5 mM, respectively (Fig. S1, ESI†).
with 4% paraformaldehyde (dissolved with PBS buffer) for 10 min
and nuclear staining was done with DAPI. The CLSM observation
was performed using confocal laser scanning microscopy (CLSM,
Formation of bolasomes and their interaction with DNA
ZEISS LSM 780) at excitation wavelengths of 405 nm for DAPI In general, cationic lipids usually form liposomes with the presence
blue) and 633 nm for Cy5 (red), respectively. For the endosome of a helper fusogenic lipid such as DOPE due to its especial
(
32
escape experiment, after incubation at 37 1C for different times membrane fusion ability. The bolasomes were prepared by using
24
(4 h, 6 h, 8 h and 12 h), LysoTracker green (1 : 750) was added the same strategy with DOPE by the thin film hydration method,
J. Mater. Chem. B
This journal is ©The Royal Society of Chemistry 2016

View Article Online
Journal of Materials Chemistry B
Paper
Scheme 1 Synthetic routes of target bolalipids.
and the mole ratio of lipid/DOPE was 1 : 1 according to the best retardation ability than the lysine-contained ones, and full
behavior in the transfection experiments (Fig. S2, ESI†). For the retardation was observed at N/P of 2. This might be attributed
twelve synthetic bolalipids, it was found that Cyc-6-6, Cyc-8-6, to the more positively charged amino groups in the structure.
Cyc-10-6, Cyc-12-6, Lys-10-6 and Lys-12-8 with a relatively
To further evaluate the DNA binding ability of the bola-
shorter chain length could not fuse with DOPE during the somes, the ethidium bromide (EB) exclusion assay was carried
4
7
hydration process, and a white precipitate was formed. However, out. EB can produce high fluorescence when intercalating
the last three bolalipids (Cyc-12-6, Lys-10-6, Lys-12-8) mentioned into DNA base pairs. The addition of another agent with higher
above could yield a clear solution after 10 min of sonication DNA affinity may expel EB from its intercalation site, leading to
(Table 1). On the other hand, bolalipids with a longer chain fluorescence quenching. The results in Fig. 2 show that the
length could form bolasomes well with DOPE, indicating that a fluorescent intensities caused by EB dramatically decreased in
special length is necessary for their self-assembly with DOPE the N/P range of 0–2, indicating the strong binding between the
(Fig. 1). We chose five typical bolalipids (Lys-10-6, Lys-12-8, bolasomes and DNA. A further increase of N/P led to only slight
Lys-14-10, Lys-14-12 and Cyc-14-10) in the following studies. Gel fluorescence weakening. More significantly, the fluorescence
electrophoresis results reveal that the ability to form bolasomes quenching levels of the five bolasomes were much different.
is essential to DNA binding and condensation. As shown in The bolasome formed from Lys-10-6, which has the shortest
Fig. 1, most of the materials can effectively bind plasmid DNA hydrophobic chain, showed the narrowest fluorescence quenching
and can completely retard DNA migration at the N/P ratio of 4. extent, suggesting its weaker DNA binding ability than others.
However, Lys-10-6, which has a weaker ability to form a bolasome, This is consistent with the results obtained in gel electrophoresis.
showed much lower binding ability, and could not retard DNA With the increase of the hydrophobic chain, the DNA binding
even at a higher N/P of 8. For the cationic headgroup, cyclen- ability of the bolasomes got stronger. Compared to the chain
derived liposome Cyc-14-10 seems to have stronger DNA length, the cationic headgroup has less effect on the fluorescence
quenching ability, and bolasomes containing lysine or cyclen
gave a similar quenching extent (Fig. 2).
The physicochemical characteristics including particle size
and zeta-potentials of the prepared bolasomes were measured
by dynamic light scattering (DLS). The effect of the bolalipid/
DOPE ratio on the properties of bolasomes was firstly studied. It
was found that the particle sizes and zeta-potentials dramatically
decreased with the participation of DOPE, while a further
increase of the DOPE amount didn’t cause an obvious change
Table 1 The solubility of the bolalipid and DOPE mixture in preparation of
a bolasome (ꢀ: insoluble; O: soluble)
Bolalipids
Hydration
Sonication
Cyc-6-6
Cyc-8-6
ꢀ
ꢀ
ꢀ
ꢀ
O
O
O
O
ꢀ
ꢀ
O
O
ꢀ
ꢀ
ꢀ
O
O
O
O
O
O
O
O
O
Cyc-10-6
Cyc-12-6
Cyc-12-8
Cyc-14-6
Cyc-14-8
Cyc-14-10
Lys-10-6
Lys-12-8
Lys-14-10
Lys-14-12
(Fig. S3A and B, ESI†). Thus subsequent experiments were
performed with the DOPE/bolalipid ratio of 1 : 1. The average
particle sizes of these bolasomes varied from 52 nm to 235 nm.
It’s clearly shown that the mean size decreased along with the
increase of the hydrophobic chain length, also indicating that
a special length is important for the formation of nano-sized
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. B

View Article Online
Paper
Journal of Materials Chemistry B
Fig. 1 Left: Schematic illustration for the formation of a bolasome with DOPE; right: electrophoretic gel retardation assays at different N/P ratios. The
molar ratio of bolalipid/DOPE was 1 : 1.
hydrophobic chains resulted in higher potentials (Fig. S3D, ESI†).
This might also be caused by the smaller liposome particles,
which led to increased surface charge density. Although lysine has
less amino groups than cyclen, liposomes with a lysine headgroup
gave slightly higher potentials than that with cyclen, suggesting
that lysine is better protonated in aqueous solution. Then, these
physical properties of the bolasome–DNA complexes (bolaplexes)
formed from bolalipids 7 were studied. As shown in Fig. 3A, after
binding with plasmid DNA, the size of the bolaplex particles
increased evidently. Under a lower N/P ratio (r4), the particle
sizes were irregular due to the incomplete DNA condensation.
From a N/P ratio of 6, all bolasomes were able to condense DNA
into steady nanosized particles (B300 nm) except Lys-10-6, which
could not form bolasomes well. Zeta-potentials of the bolaplexes
Fig. 2 Fluorescent quenching assay of EB/DNA with the addition of rose along with the increase of the N/P ratio, and the structure–
bolasomes. The molar ratio of bolalipid/DOPE was 1 : 1.
property relationship was similar to the empty bolasomes
(Fig. S3D, ESI†).
Transmission electron microscopy (TEM) was used to visually
particles (Fig. S3C, ESI†). In other words, since it was known study the morphological characteristics of the bolasomes and
that liposomes with diameters below 100 nm would have better bolaplexes. The images in Fig. 4 show that lysine-based bolasomes
4
8
DNA binding and delivery ability, the chain lengths of Lys-10-6 have spherical distribution with the diameters ranging from
and Lys-12-8 are not long enough to form nanoparticles with the 40 nm to 100 nm. Interestingly, cyclen-contained bolasome
proper size, while Lys-14-10, Lys-14-12 and Cyc-14-10 could form Cyc-14-10 gave a more homogeneous distribution and much
bolasomes with B55 nm diameters. The zeta potentials of the smaller particles (diameter B15 nm), which was supposed to
bolasomes were found in the range of +40–70 mV, and longer be spherical micelles but not bolasomes. The polyamine ring in
Fig. 3 Mean particle sizes (A) and zeta-potentials (B) of the bolaplexes formed from bolasome and pUC19 DNA under various N/P ratios (DLS at room
temperature). Data represent mean ꢃ SD (n = 3).
J. Mater. Chem. B
This journal is ©The Royal Society of Chemistry 2016

View Article Online
Journal of Materials Chemistry B
Paper
Fig. 4 TEM images of selected bolasomes and bolaplexes. Scale bar: 50 nm.
cyclen has a bigger cross-sectional area than the lysine group,
therefore, compared to the bolasomes formed from lysine-derived
bolalipids, Cyc-14-10 might form a micelle structure by bending Fig. 6 Fluorescent microscopy images of HEK 293 cells and HeLa cells
2
6,42
and folding of the hydrophobic skeleton.
Furthermore, the transfected by bolasomes at the optimal N/P ratio (HEK 293: 12, 10, 6, 6 and
0; HeLa: 12, 6, 4, 4 and 10, respectively). The molar ratio of bola/DOPE was
: 1 and lipofectamine 2000 was used as the control. The cells were observed
by fluorescence microscopy after a 24 h transfection. Scale bar: 100 mm.
1
1
last two images show that DNA could also be compacted into
spherical nanoparticles with a diameter of 40–60 nm by two
lysine-derived bolasomes. The particle sizes determined by
TEM were not identical with those obtained by DLS. This might
be due to the different experimental conditions: DLS data provide synthesized materials than the positive control, especially at
a hydrodynamic size of the particles, which could further aggre- the N/P ratio used in subsequent transfection assays (o16).
gate to form larger particles, while in TEM measurements, the
In vitro gene transfection
particles are stabilized by adsorption on the surface of carbon-
coated copper meshes.
4
9
The TE of bolaplexes was preliminarily studied by using the
green fluorescent protein (GFP) reporter gene. Transfection was
mediated by the bolaplexes at various N/P ratios in two cell
Cytotoxicity
One of the criteria for an ideal non-viral gene delivery vector is lines (HEK 293 and HeLa). Fig. 6 shows the eGFP expression in
to possess a low cytotoxicity profile. Bolasomes have been two cell lines at optimal N/P ratios observed by an inverted
reported with relatively lower toxicity than traditional cationic fluorescent microscope. Although the density of the green
5
0
liposomes. MTS-based cell viability assays were carried out fluorescence induced by the bolaplexes was lower than that
in HEK 293 and HeLa cells to investigate the cytotoxicity of by lipofectamine 2000, we may find that the TE relied heavily on
the bolaplexes at various N/P ratios, and lipid lipofectamine the chemical structure of the bolalipids. The TE increased with
2000 was used as a positive control. Results shown in Fig. 5 the extension of the hydrophobic chain, and reached the best
reveal that both cell lines showed higher tolerance against the transfection at Lys-14-10. Further elongation of the chain led to
Fig. 5 In vitro cytotoxicity of the bolaplexes at various N/P ratios in HEK 293 (A) and HeLa cells (B) for a 24 h incubation. Data represent mean ꢃ SD (n = 3).
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. B

View Article Online
Paper
Journal of Materials Chemistry B
Fig. 7 Luciferase expression in HEK 293 (A) and HeLa (B) cells transfected by bolaplexes at various N/P ratios. Data represent mean ꢃ SD (n = 3).
a decreased TE (Lys-14-12). As for the cationic headgroup,
bolaplexes derived from cyclen based bolalipid (Cyc-14-10) gave
much lower TE than its lysine analog. Moreover, the luciferase
reporter gene was also used to quantitatively study the TE of the
bolaplexes. As shown in Fig. 7, the results were found to be
similar to the GFP assays, and Lys-14-10 gave the best TE, which
was about 50% of lipofectamine 2000. We speculate that Lys-14-10
has the most proper chain length (36 atoms including 2 oxygens)
to form bolasomes with the thickness similar to a cell membrane,
leading to better membrane fusion and cellular uptake. Its analog,
Cyc-14-10, gave comparable TE in HEK 293 cells and much lower
TE in HeLa cells. Such results might be attributed to its stronger
DNA binding ability (Fig. 1), which would result in more difficult
DNA release in the cell. Furthermore, the effect of serum on the
Fig. 8 Cellular uptake (columns, percentage of Cy5-positive cells) and
mean fluorescence intensity (dots and lines) of bolaplexes at the optimal
TE was also investigated by using three representative bolaplexes
(Fig. S4, ESI†). Unfortunately, compared to lipofectamine 2000,
N/P ratio in HEK 293 cells quantified by flow cytometry. Lipofectamine
000 was used as the control. Data represent mean ꢃ SD (n = 3).
these bolalipids did not exhibit better serum tolerance, and the TE
gave a measurable decrease. However, Lys-14-10 showed distinctly
higher serum resistibility than Cyc-14-10, and this might guide us
to design novel bolalipids with higher TE and biocompatibility.
2
difficult release of DNA and lower transfection. Subsequently,
The fluorescence-activated cell sorting (FACS) technique was the cellular uptake assays were also carried out using Lys-14-10
applied for further study of the transfection mechanism without or with 10% serum in HeLa cells to study the effect
mediated by these bolasomes in HEK 293 cells. After a 4 h cell of serum. Results in Fig. S5 (ESI†) show that the uptake cell
incubation with the bolaplexes at each optimized N/P ratio, the percentage and fluorescence intensity were seldom influenced
percentage of Cy5-positive cells and the mean fluorescence by 10% serum, indicating a comparable serum tolerance to
intensity of Cy5 were calculated and shown in Fig. 8. It lipofectamine 2000, especially in HeLa cells.
was confirmed that a hydrophobic skeleton with the length
Confocal laser scanning microscopy (CLSM) was also
of ‘‘14-10’’ has superiority for cellular uptake. Lys-14-10 and applied to visually examine the internalization and intracellular
Cyc-14-10 gave significantly higher values in both uptake cell location of Cy5-labelled DNA transferred by bolaplexes at the
percentage and fluorescence intensity than other bolalipids. optimal N/P ratio in HEK 293 cells. DNA was labelled by Cy5
0
However, the values were still lower than the non-bola lipid (red), while the cell nuclei were stained with 4 ,6-diamidino-2-
lipofectamine 2000, and this might contribute to their lower TE in phenylindole (DAPI, blue). As shown in Fig. 9, after 4 h of
the transfection assays. On the other hand, it was reported that incubation with bolaplexes derived from Lys-14-10 and Cyc-14-10,
51
materials with secondary amines showed higher cellular uptake,
a considerable amount of the fluorescent labelled DNA accumu-
thus the result that Cyc-14-10 had a higher cellular uptake than lated in the perinuclear region, while very weak signals were
Lys-14-10 might be reasonable. Nevertheless, Cyc-14-10 gave lower observed for the transfection by the other three bolaplexes.
TE than Lys-14-10, also suggesting that good cellular uptake is a Similarly, the red fluorescence induced by Cyc-14-10 was
necessary but not sufficient condition for high TE. As mentioned slightly stronger than that by Lys-14-10, further suggesting a
above, the stronger DNA binding ability of Cyc-14-10 may lead to better cellular uptake, which is consistent with the flow
J. Mater. Chem. B
This journal is ©The Royal Society of Chemistry 2016

View Article Online
Journal of Materials Chemistry B
Paper
Fig. 9 CLSM images of HEK 293 cells transfected with Cy5-labelled DNA by the bolaplexes at the optimal N/P ratio. For each row, left: cell nuclei stained
by DAPI (blue); middle: Cy5-labelled pDNA (red); right: merged image. Scale bar: 10 mm.
cytometry results. In addition, the intracellular trafficking of (No. 21472131, 21232005). We thank the comprehensive train-
cargo DNA mediated by Cyc-14-10 and Lys-14-10 was also ing platform of specialized laboratory, College of Chemistry,
studied. Resulting images in Fig. S6 (ESI†) demonstrate that Sichuan University for sample analysis.
DNA (red) could escape well from the endosome/lysosome
(green), which is necessary for efficient transfection.
References
Conclusions
1
2
I. M. Verma and N. Somia, Nature, 1997, 389, 239–242.
S. Bhattacharya and A. Bajaj, Chem. Commun., 2009,
In summary, a series of novel symmetric cationic bolalipids with
various hydrophobic chain lengths were designed and synthesized.
Lysine and cyclen were chosen as the cationic headgroup. Structure–
activity relationships of these bolalipids in liposome formation, DNA
binding, the physical property of the bolasome, and gene transfec-
tion were systematically studied. Results reveal that an appropriate
hydrophobic chain length is essential to form bolasomes with good
DNA binding and condensation ability. Bolalipids with short chains
could not condense DNA well, leading to a poor gene transfection
efficiency (TE). Lys-14-10, which has a 36-atom-length hydrophobic
chain, exhibited the best TE in the two cell lines. Flow cytometry and
confocal laser microscopy assays reveal that the bolaplexes formed
from bolalipids with such a chain might induce the highest cellular
uptake. We speculate that the thickness of such bolasomes is close
to a cell membrane, thus facilitating the membrane fusion and
subsequent endocytosis. Although the TEs of these bolalipids are
still lower than commercially used non-bola lipid lipofectamine
4
632–4656, DOI: 10.1039/B900666B.
M. A. Mintzer and E. E. Simanek, Chem. Rev., 2009, 109,
59–302.
S. S. Kelkar and T. M. Reineke, Bioconjugate Chem., 2011, 22,
879–1903.
3
4
5
6
7
2
1
M. Khan, Z. Y. Ong, N. Wiradharma, A. B. E. Attia and Y.-Y.
Yang, Adv. Healthcare Mater., 2012, 1, 373–392.
H. Wang, Y. Chen, X.-Y. Li and Y. Liu, Mol. Pharmaceutics,
2
007, 4, 189–198.
Z. U. Rehman, I. S. Zuhorn and D. Hoekstra, J. Controlled
Release, 2013, 166, 46–56.
X. Guo and L. Huang, Acc. Chem. Res., 2012, 45, 971–979.
Y. Yue and C. Wu, Biomater. Sci., 2013, 1, 152–170.
8
9
1
1
0 U. L ¨a chelt and E. Wagner, Chem. Rev., 2015, 115, 11043–11078.
1 N. Cartier, S. Hacein-Bey-Abina, C. C. Bartholomae,
G. Veres, M. Schmidt, I. Kutschera, M. Vidaud, U. Abel,
L. Dal-Cortivo, L. Caccavelli, N. Mahlaoui, V. Kiermer,
D. Mittelstaedt, C. Bellesme, N. Lahlou, F. Lefr `e re,
S. Blanche, M. Audit, E. Payen, P. Leboulch, B. l’Homme,
P. Bougn `e res, C. Von Kalle, A. Fischer, M. Cavazzana-Calvo
and P. Aubourg, Science, 2009, 326, 818–823.
2 K. Braeckmans, K. Buyens, B. Naeye, D. Vercauteren,
H. Deschout, K. Raemdonck, K. Remaut, N. N. Sanders,
J. Demeester and S. C. De Smedt, J. Controlled Release, 2010,
148, 69–74.
2000, this study may give us some clues for the design of novel
bolalipids with a higher TE and biocompatibility.
Acknowledgements
1
This work was financially supported by the National Program
on Key Basic Research Project of China (973 Program,
2012CB720603) and the National Science Foundation of China
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. B

View Article Online
Paper
Journal of Materials Chemistry B
1
1
1
1
1
1
1
2
2
3 S. Kang, K. Lu, J. Leelawattanachai, X. Hu, S. Park, T. Park,
I. M. Min and M. M. Jin, Gene Ther., 2013, 20, 1042–1052.
J. J. B. Perry, R. J. Taylor, A. Turner and A. N. Weir, Angew.
Chem., Int. Ed., 2000, 112, 4229–4233.
4 Y.-M. Zhang, Y.-H. Liu, J. Zhang, Q. Liu, Z. Huang and 34 G. G. M. D’Souza, R. Rammohan, S.-M. Cheng, V. P. Torchilin
X.-Q. Yu, RSC Adv., 2014, 4, 44261–44268. and V. Weissig, J. Controlled Release, 2003, 92, 189–197.
5 Z. Huang, Y.-H. Liu, Y.-M. Zhang, J. Zhang, Q. Liu and 35 G. G. M. D’Souza, S. V. Boddapati and V. Weissig, Mitochondrion,
X.-Q. Yu, Org. Biomol. Chem., 2015, 13, 620–630. 2005, 5, 352–358.
6 Q.-F. Zhang, W.-J. Yi, B. Wang, J. Zhang, L. Ren, Q.-M. Chen, 36 J. Gaucheron, C. Santaella and P. Vierling, Bioconjugate
L. Guo and X.-Q. Yu, Biomaterials, 2013, 34, 5391–5401. Chem., 2001, 12, 569–575.
7 W.-J. Yi, X.-C. Yu, B. Wang, J. Zhang, Q.-Y. Yu, X.-D. Zhou 37 K. Fabio, J. Gaucheron, C. Di Giorgio and P. Vierling,
and X.-Q. Yu, Chem. Commun., 2014, 50, 6454–6457. Bioconjugate Chem., 2003, 14, 358–367.
8 Y.-H. Zhang, Y. Chen, Y.-M. Zhang, Y. Yang, J.-T. Chen and 38 G. Zuber, L. Zammut-Italiano, E. Dauty and J.-P. Behr,
Y. Liu, Sci. Rep., 2014, 4, 7471. Angew. Chem., Int. Ed., 2003, 42, 2666–2669.
9 Y.-H. Zhang, Y.-M. Zhang, Q.-H. Zhao and Y. Liu, Sci. Rep., 39 C. Lain ´e , E. Mornet, L. Lemi `e gre, T. Montier, S. Cammas-
2
016, 6, 22654.
0 H. Liu, H. Wang, W. Yang and Y. Cheng, J. Am. Chem. Soc.,
012, 134, 17680–17687.
1 M. Wang, H. Liu, L. Li and Y. Cheng, Nat. Commun., 2014, 5,
053–3060.
Marion, C. Neveu, N. Carmoy, P. Lehn and T. Benvegnu,
Chem. – Eur. J., 2008, 14, 8330–8340.
2
40 J.-X. Chen, X.-D. Xu, S. Yang, J. Yang, R.-X. Zhuo and
X.-Z. Zhang, Macromol. Biosci., 2013, 13, 84–92.
41 D. Y. Cho, H. Cho, K. Kwon, M. Yu, E. Lee, K. M. Huh,
D. H. Lee and H. C. Kang, Adv. Funct. Mater., 2015, 25,
5479–5491.
42 K. Gupta, K. A. Afonin, M. Viard, V. Herrero, W. Kasprzak,
I. Kagiampakis, T. Kim, A. Y. Koyfman, A. Puri, M. Stepler,
A. Sappe, V. N. KewalRamani, S. Grinberg, C. Linder,
E. Heldman, R. Blumenthal and B. A. Shapiro, J. Controlled
Release, 2015, 213, 142–151.
3
2
2
2 V. Gopal, J. Controlled Release, 2013, 167, 323–332.
3 B. Wang, W.-J. Yi, J. Zhang, Q.-F. Zhang, M.-M. Xun and
X.-Q. Yu, Bioorg. Med. Chem. Lett., 2014, 24, 1771–1775.
4 N. Jain, Y. Arntz, V. Goldschmidt, G. Duportail, Y. M ´e ly and
A. S. Klymchenko, Bioconjugate Chem., 2010, 21, 2110–2118.
5 M. Brunelle, A. Polidori, S. Denoyelle, A.-S. Fabiano,
P. Y. Vuillaume, S. Laurent-Lewandowski and B. Pucci,
C. R. Chim., 2009, 12, 188–208.
2
2
43 Q. Liu, Q.-Q. Jiang, W.-J. Yi, J. Zhang, X.-C. Zhang, M.-B. Wu,
Y.-M. Zhang, W. Zhu and X.-Q. Yu, Bioorg. Med. Chem., 2013,
21, 3105–3113.
2
2
6 J.-H. Fuhrhop and T. Wang, Chem. Rev., 2004, 104, 2901–2938.
7 L. Latxague, M. A. Ramin, A. Appavoo, P. Berto, M. Maisani,
C. Ehret, O. Chassande and P. Barth ´e l ´e my, Angew. Chem., 44 W.-J. Yi, Q.-F. Zhang, J. Zhang, Q. Liu, L. Ren, Q.-M. Chen,
Int. Ed., 2015, 54, 4517–4521. L. Guo and X.-Q. Yu, Acta Biomater., 2014, 10, 1412–1422.
8 N. Jain, V. Goldschmidt, S. Oncul, Y. Arntz, G. Duportail, 45 H. Zeng, M. E. Johnson, N. J. Oldenhuis, T. N. Tiambeng
2
2
3
3
Y. M ´e ly and A. S. Klymchenko, Int. J. Pharm., 2012, 423,
92–400.
9 S. Grinberg, V. Kolot, C. Linder, E. Shaubi, V. Kas’yanov,
and Z. Guan, ACS Cent. Sci., 2015, 1, 303–312.
46 W.-X. Wu, N. Wang, B.-Y. Liu, Q.-F. Deng and X.-Q. Yu,
Soft Matter, 2014, 10, 1199–1213.
3
R. J. Deckelbaum and E. Heldman, Chem. Phys. Lipids, 2008, 47 D. Lleres, E. Dauty, J.-P. Behr, Y. M ´e ly and G. Duportail,
53, 85–97. Chem. Phys. Lipids, 2001, 111, 59–71.
0 G. R. Dakwar, I. A. Hammad, M. Popov, C. Linder, S. Grinberg, 48 Y. Liu and T. M. Reineke, J. Am. Chem. Soc., 2005, 127,
1
E. Heldman and D. Stepensky, J. Controlled Release, 2012, 160,
15–321.
1 S. Denoyelle, A. Polidori, M. Brunelle, P. Y. Vuillaume,
3004–3015.
3
49 M. Ma, F. Li, Z.-F. Yuan and R.-X. Zhuo, Acta Biomater., 2010,
6, 2658–2665.
S. Laurent, Y. ElAzhary and B. Pucci, New J. Chem., 2006, 50 T. Kim, K. A. Afonin, M. Viard, A. Y. Koyfman, S. Sparks,
3
0, 315–321.
2 I. Koltover, T. Salditt, J. O. R ¨a dler and C. R. Safinya, Science,
998, 281, 78–81.
E. Heldman, S. Grinberg, C. Linder, R. P. Blumenthal and
B. A. Shapiro, Mol. Ther.–Nucleic Acids, 2013, 2, e80.
51 Q.-F. Zhang, Q.-Y. Yu, Y. Geng, J. Zhang, W.-X. Wu, G. Wang,
Z. Gu and X.-Q. Yu, ACS Appl. Mater. Interfaces, 2014, 6,
15733–15742.
3
3
1
3 M. A. W. Eaton, T. S. Baker, C. F. Catterall, K. Crook,
G. S. Macaulay, B. Mason, T. J. Norman, D. Parker,
J. Mater. Chem. B
This journal is ©The Royal Society of Chemistry 2016