Revisiting complexation between DNA and polyethylenimine: Does the disulfide linkage play a critical role in promoting gene delivery?

Article abstract of DOI:10.1002/mabi.201400253

Despite its cytotoxicity, polyethylenimine (PEI) is still used as a golden reference in gene transfection. Long PEI chains are more effective but also more cytotoxic. To solve this problem, an alternative strategy is to link short PEI chains into a longer PEI with disulfide bonds because they are degradable in the cell. However, how PEI promotes gene transfection is still unclear. Also, the chain length of PEI is also increased as disulfide bonds are formed. Therefore, it is important to investigate whether the increase in transfection efficiency is attributable to the disulfide linkage, chain size, or both. To distinguish between such factors, a novel method is developed here tomake longer linear PEI with disulfide bonds (lPEI_s-s) by linking the mercapto groups of short linear PEI (lPEI_i). By comparing the physiochemical properties and the transfection efficiencies of short lPEIi, long lPEI_s-s, and un-degradable long PEI, it is found that introducing disulfide bonds instead of directly using longer PEI chains has less effect on gene transfection, and it is the chain length that plays a key role in promoting gene transfection.

Full text of DOI:10.1002/mabi.201400253

Full Paper
Revisiting Complexation between DNA and
Polyethylenimine: Does the Disulfide
Linkage Play a Critical Role in
a
Promoting Gene Delivery?
Yongzheng Ma,* Baizhu Chen, Ning He, Gaojian Chen, Lianwei Li, Chi Wu
Despite its cytotoxicity, polyethylenimine (PEI) is still used as a golden reference in gene
transfection. Long PEI chains are more effective but also more cytotoxic. To solve this problem,
an alternative strategy is to link short PEI chains into a longer PEI with disulfide bonds because
they are degradable in the cell. However, how PEI promotes gene transfection is still unclear.
Also, the chain length of PEI is also increased as disulfide bonds are formed. Therefore, it is
important to investigate whether the increase in
transfection efficiency is attributable to the disulfide
linkage, chain size, or both. To distinguish between such
factors, a novel method is developed here to make longer
linear PEI with disulfide bonds (lPEI ) by linking the
s-s
mercapto groups of short linear PEI (lPEI ). By comparing
i
the physiochemical properties and the transfection
efficiencies of short lPEI , long lPEI , and un-degradable
i
s-s
long PEI, it is found that introducing disulfide bonds
instead of directly using longer PEI chains has less effect
on gene transfection, and it is the chain length that plays
a key role in promoting gene transfection.
1
. Introduction
decades, they have been developed in different forms,
[3–5]
including both viral
and non-viral vectors (cationic
[6]
[7–11]
Gene vectors have received much attention because of their
lipids and polymers
). In comparison with other
[1,2]
applications in the field of gene therapy.
In the last two
polymers, cationic polymer chains are attractive because of
their low immunogenicity, large packing size, and con-
[
12,13]
Dr. Y. Ma, B. Chen, Prof. N. He, Prof. G. Chen, Prof. C. Wu
Department of Chemistry, The Chinese University of Hong Kong,
Shatin, N. T., Hong Kong
E-mail: ericyongzheng@126.com
Dr. L. Li, Prof. C. Wu
struction flexibility.
polymers synthesized, polyethylenimine (PEI) remains one
Among thousands of cationic
[14–16]
of the most efficient and simplest non-viral vectors.
It
is known that large PEI chains (branched and linear PEI 25k)
are more effective than their small counterparts, but large
The Hefei National Laboratory of Physical Science at Microscale,
Department of Chemical Physics, University of Science and
Technology of China, Hefei, Anhui 230026, China
PEI chains exhibit high cytotoxicity, which limits the
[17–21]
application of PEI.
To solve this problem, linking short
PEI chains via disulfide bonds into a longer chain was
suggested and applied for the following reasons. It is
much better for longer PEI chains to combine, stabilize,
a
Supporting Information is available at Wiley Online Library or from
the author.
Macromol. Biosci. 2014, 14, 1807–1815
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
DOI: 10.1002/mabi.201400253 1807

Y. Ma, B. Chen, N. He, G. Chen, L. W. Li, C. Wu
www.mbs-journal.de
and protect DNA in the extracellular space. They can be
quickly degraded in the intracellular redox environment
wherein the glutathione concentration is 50–1000 times
mass similar to lPEIs-s (ꢂ 45 nm) so that we could
comparatively investigate their transfection efficiency
and cytotoxicity under identical conditions to probe the
effects of disulfide bonds.
[22–24]
higher than that in the extracellular space.
Long
PEI chains are degraded into small ones that are less
cytotoxic and bind less to DNA after their disulfide
bonds are cleaved inside the cell. Such a strategy has been
2. Experimental Section
[25]
tried by various groups. For example, Liu et al.
cross-
2
.1. Materials and Cell Lines
linked branch PEI chains using click chemistry, Peng
[
26]
et al.
thiolation of small branched ones, Breunig et al.
dithiobis(succinimidyl propionate) (DSP) as a crosslinking
prepared disulfide crosslinked PEI clusters via
ꢁ1
Linear polyethylenimine (lPEI-25k, Mw ¼ 25 kg mol ) was pur-
[27]
used
chased from Polysciences. Hydrobromic acid solution (33 wt% in
acetic acid, HBr/AcOH), branched polyethylenimine (bPEI-25k,
[28]
ꢁ1
agent, and Lee et al.
synthesized longer degradable
M
¼ 25 kg mol ), p-toluenesulfonyl chloride (p-TsCl), ethylene
w
linear PEI chains by several coupling-hydrolysis cycles.
Most of these efforts, as expected, resulted in a lower
cytotoxicity and higher gene transfection efficiency.
However, the exact role of the disulfide bonds during the
gene transfection remains unclear, though the assump-
tions made to date seem reasonable.
glycol, and potassium thioacetate from Sigma-Aldrich were used
withoutfurtherpurification.Methylp-toluenesulfonatepurchased
from Sigma-Aldrich was distilled and then stored under nitrogen.
2
-Ethyl-2-oxazoline from Sigma-Aldrich was purified by vacuum
distillation overCaH . Acetonitrile from RCI Labscan was dried over
CaH and distilled under dry nitrogen. All other chemicals and
2
2
solvents were used as received without further purification.
Initial Plasmid DNA (pGL3) with an SV40 promoter and an
enhancer sequence encoding firefly luciferase was purchased from
Promega (Madison, USA). A large amount of this plasmid DNA was
prepared with a Qiagen Plasmid Maxi Kit (Qiagen, Germany). Fetal
bovine serum (FBS), penicillin-streptomycin, and Dulbecco’s
modified Eagle’s medium (DMEM) were obtained from GIBCO
A few years ago, our group developed a novel laser
light-scattering method to in situ monitor the disulfide
ꢁ
1
linking reaction of short branch chains (Mw ꢀ 2 kg mol
,
[29]
denoted as bPEI-2k). Itwas found that linking 3-4 bPEI-2k
to form the larger PEI could significantly enhance the
5
7
gene transfection by a factor of 10 –10 times, depending
on the PEI/DNA ratio. It was also found that most of the
PEI chains did not combine with DNA in the solution
mixture of polymer and DNA, especially when the N/P
(
NY, USA). 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium
bromide (MTT) was purchased from Sigma-Aldrich (Germany).
93T cells and HepG2 cells were grown at 37 8C/5% CO in DMEM
supplemented with 10% FBS, penicillin at 100 units mL , and
2
2
ꢁ1
[
30]
ratio was higher than six,
and it is those free PEI chains
ꢁ1
[31]
streptomycin at 100 mg mL , respectively.
with a size larger than ꢂ 15 nm (contour length)
that
Recently,
we also found that DNA was continuously released from
[32,33]
effectively promote the gene transfection.
2
.2. Characterization
[34]
the polyplexes prior to the disulfide bond’s cleavage.
Therefore, the disulfide bond’s effect during gene transfec-
tion should be reconsidered. It was important to
investigate whether the increase in transfection efficiency
NMR spectra were recorded on a Bruker Avance III 400 spectrome-
ter. The chemical shifts (d) were reported in ppm with the solvent
resonance as the internal standard relative to chloroform (d 7.26) or
1
1
5
7
[35]
2
D O (d 4.79) for H. All the H NMR measurements were carried out
(10 –10 times) is due to the disulfide bond cleavage
or
at room temperature. Mass spectra (ESI-MS) were obtained with a
to the effect of chain length.
HP 5989B spectrometer and determined at an ionizing voltage of
To summarize the reported studies, most of them
focused on linking small branched PEI to obtain an
effective vector. Compared to its small precursor, the
disulfide-linked branched PEI showed a higher transfec-
tion efficiency. However, comparative studies of disulfide-
linked linear PEIs are few in number. In this study, we
designed and developed a novel method to prepare
longer, structure-defined linear PEI chains with degrad-
able disulfide linkages (lPEIs-s). In comparison to branched
PEI, the chain length and structure of linear PEI were
much easier to define. We purposely chose our initial
7
0 eV; relevant data were tabulated as m/z. Matrix-assisted laser
desorption time of flight mass spectrometry (MALDI-TOF, Voyager-
DE STR, AB SCIEX) was used to characterize the molar mass of all
the PEI samples. Trans-4-hydroxy-3-methoxycinnamic acid was
used as the matrix. The solvent was acetonitrile and 3% TFA in
deionized water.
2.3. Synthesis of 1,2-Bis(p-tolylsufonyloxy)ethane
ꢁ3
To a stirred solution of ethylene glycol (0.3 g, 4.8 ꢃ 10 mol) and
TEA (2 mL) in dry methylene chloride (20 mL), a solution of p-TsCl
linear chain (lPEI
i
) with a longer chain length (ꢂ 15 nm) to
ꢁ3
(
3.6 g, 19.2 ꢃ 10 mol) in dry methylene chloride (5 mL) was added
test whether the extension of the chain length with
degradable disulfide linkages could indeed promote
gene transfection. We also synthesized a non-degradable
linearPEI (lPEI-4.2k, ꢂ44 nm)withaweight-average molar
dropwise at 0 8C. After overnight stirring, the solution was filtered
and re-crystallized in CHCl /methanol solution. The product was
3
[36] 1
dried in vacuo (yield 95%, 1.9 g).
3
H NMR (400 MHz, CDCl , d):
7.74–7.72(d, 4H, –C CH ), 7.34–7.32(d, 4H, –C H CH ), 4.18 (s, 4H,
6
H
4
3
6 4 3
Macromol. Biosci. 2014, 14, 1807–1815
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1808
www.MaterialsViews.com

Revisiting Complexation between DNA and Polyethylenimine: Does the Disulfide Linkage Play. . .
www.mbs-journal.de
–
O–CH
2
CH
2
–O–),2.45(s,3H,–C
6
H
4
CH
3
)(FigureS1intheSupporting
þ
Information). HRMS (ESI) m/z: [M þ Na] calcd for 1,2-bis(p-
tolylsufonyloxy)ethane, 393.4405; found,: 393.4408.
2
.4. Synthesis of Poly(2-ethyl-2-oxazoline)-
bis(thioacetate) (AcS-PETOZ-SAc)
ꢁ
3
A solution of 1,2-bis(p-tolylsufonyloxy)ethane (1.08 g, 2.7 ꢃ 10
ꢁ3
mol) and 2-ethyl-2-oxazoline (10 mL, 100 ꢃ 10 mol) in dry
acetonitrile (30 mL) was heated to 90 8C and stirred for 48 h under
vacuum in a Schlenk flask. The solution was then cooled down and
stored at 4 8C to synthesize polymer chains with p-toluenesulfo-
nate end groups. 24 h later, the polymer solution was precipitated
in diethyl ether. The precipitate was washed three more times with
diethyl ether and vacuum dried to obtain a yellow powder of
poly(2-ethyl-2-oxaline)-bis(tosylate) (Ts-PETOZ-Ts) with a yield of
[
37,38] 1
9
–
2
0% (8.9 g).
CH ), 7.16–7.14 (d, –C
.30 (m, –COCH CH and –C
H NMR (400 MHz, CDCl
CH ), 3.46 (s, –CH
CH ), 1.21–1.11 (m, –COCH
3
, d): 7.71–7.69 (d,
CH NH–), 2.45–
CH
C
H
6 4
3
6
H
4
3
2
2
2
3
6
H
4
3
2
3
)
(
Figure S2 in the Supporting Information).
ꢁ
3
A solution of Ts-PETOZ-Ts (5 g, 1.3 ꢃ 10 mol) and potassium
Scheme 1. Synthesis of degradable linear polyethylenimine (lPEIs-s
)
ꢁ3
thioacetate (0.74g, 6.5 ꢃ 10 mol) in acetonitrile (35 mL) was
stirred in a round bottom flask at 25 8C for 12h before it was
(TsO: tosylate).
diluted with CH
Cl
2 2
(65 mL) and further stirred for 1 h. The solution
2 2 2 2
D O, d): 3.65–3.53 (m, –CH CH NH–), 2.54–2.48 (m, –CH S–)
was then washed with saturated NaCl solution (3 ꢃ 100 mL) and
deionized water (100 mL). The resultant solution was dried with
anhydrous magnesium sulfate. Finally, AcS-PETOZ-SAc was har-
(Figure S5 in the Supporting Information). M ¼ 3.96, PDI ¼ 1.43.
n
Scheme 1 shows a schematic of the synthesis.
[38]
vested by precipitation in cold diethyl ether (yield 90%, 4.4 g).
1
H NMR (400 MHz, CDCl
CH ), 2.45–2.30 (m, –COCH
Figure S3 in the Supporting Information). Mn ¼ 3.74, PDI ¼ 1.37.
3
, d): 3.46 (s, –CH
2
CH
2
NH–), 3.0–2.9 (t, –SCO–
2
.7. Synthesis of Linear Polyethylenimine (lPEI-4.2k)
3
2 3 2 3
CH ), 1.21–1.11 (m, –COCH CH )
without a Degradable Disulfide Bond
(
ꢁ3
A solution of methyl p-toluenesulfonate (0.186 g, 1 ꢃ 10 mol) and
-ethyl-2-oxazoline (12 mL, 0.12 mol) in dry acetonitrile (30 mL)
2
was heated to 90 8C and stirred for 48 h under vacuum in a Schlenk
flask before it was cooled down and precipitated in cold diethyl
ether. The precipitate was washed three times with diethyl ether
and then vacuum dried, which resulted in a yellow powder of
2
.5. Synthesis of Linear Thiolated Poly(ethylenimine)
Precursor (lPEI
i
)
ꢁ3
ꢁ3
AcS-PETOZ-SAc (4 g, 1 ꢃ 10 mol) and phenol (0.045 g, 0.5 ꢃ 10
1
mol) were dissolved in 33% HBr/AcOH (20 mL) and further diluted
with deionized water (20 mL). After refluxing for 24 h, a pale brown
precipitatewasobtained.Theprecipitatewaswashedwithethanol
several times and refluxed in ethanol for 2 h. The mixture was re-
precipitated in diethyl ether and filtered to obtain linear PEI with
poly(2-ethyl-2-oxaline) (PETOZ) with a yield of 80% (9.5 g). H NMR
(400 MHz, CDCl3, d): 7.67–7.65 (d, –C H CH ), 7.16–7.14 (d,
6
4
3
–C H CH ), 3.44 (s, –CH CH NH-), 2.39–2.29 (m, –COCH CH and
6
4
3
2
2
2
3
–C H CH ), 1.11 (m, –COCH CH ). GPC (THF as eluent): M ¼ 9.8 kg
6
4
3
2
3
n
ꢁ
1
mol , PDI ¼ 1.31.
[28]
1
ꢁ3
two functional ends.
m, –CH CH NH–), 2.62–2.55 (m, –CH
Supporting Information). Mn ¼ 1.66, PDI ¼ 1.12.
2
H NMR (400 MHz, D O, d): 3.65–3.54
Then, PETOZ (5 g, 0.5 ꢃ 10 mol) was dissolved into 15% HCl
(30 mL) solution. After refluxing for 24 h, a white precipitate was
filtered and re-dissolved in 30 mL of water, whose pH was adjusted
to ꢀ9 by the addition of 5 M sodium hydroxide (NaOH). The
precipitate was washed with water until pH ꢂ 7 was reached.
(
2
2
2
SH) (Figure S4 in the
The final product was obtained by freeze-drying (yield 90%,
2
.6. Synthesis of Linear Poly(ethylenimine) with
1
1
.9 g). H NMR (400 MHz, D
2
O, d): 3.52–3.46 (m, –CH
2 2
CH NH–).
Degradable Disulfide Linkages (lPEIs-s
)
Mn ¼ 4.2 kg molꢁ (as calculated from the GPC result for PETOZ).
1
ꢁ3
lPEI
dissolved in a mixture of DMSO and deionized water (1:1 v/v,
mL). The solution was stirred at 25 8C for 7 d to ensure a sufficient
i
(0.5 g, 0.3 ꢃ 10 mol) and triethylamine (0.1 mL) were
2
2
.8. Laser Light Scattering
commercial LLS instrument (ALV5000) with
reaction. After adjusting to pH 10 with 5 M NaOH, a white
precipitate was obtained. The crude product was re-dissolved in
water(pH ¼ 4)anddialyzedagainstalargeamountofwater(for2d)
A
a vertically
polarized 22 mV He-Ne laser (632.8 nm, Uniphase) was used to
detect the degradation of lPEIs-s in real time and characterize the
hydrodynamic radius of polyplexes mediated by different PEIs. In
ꢁ1
using a dialysis bag with a cut-off molar mass ꢂ1 kg mol
1
(
MWCO). The final product was freeze-dried. H NMR (400 MHz,
Macromol. Biosci. 2014, 14, 1807–1815
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1809
www.MaterialsViews.com

Y. Ma, B. Chen, N. He, G. Chen, L. W. Li, C. Wu
www.mbs-journal.de
static LLS, the weight-average molar mass (Mw) and the z-average
where Atreated, Ablank, and Acontrol are the absorbance of the cells
treatedwithPEI,freshculturemedium,andwithoutanytreatment,
respectively. Each experiment condition was repeated in triplicate.
The data is shown as the mean value ꢄ a standard deviation (ꢄ SD).
g z
root-mean square radius of gyration (<R > ) of scattering objects
canbeobtainedinasufficientlydilutesolution/dispersionfromthe
angular and concentration dependence of the excess absolute
2
2
scattering intensity (Rayleigh ratio Rvv(q)) as KC/Rvv(q) ꢂ (1 þ q R
g
/
2
2
2
4
3
)/Mw where K ¼ 4p n (dn/dc) /(N
A
l
0
) and q ¼ 4pn/l
0
sin u/2,
with dn/dc, N
A
, l
0
, n and u being the specific refractive index
2.11. Degradation Test of Long lPEIs-s
increment, the Avogadro constant, the incident wavelength in a
vacuum, the refractive index of the solvent, and the scattering
angle, respectively. Note that we have neglected the concentration
correction here.
lPEIs-s (40 mg)and1,4-dithio-D,L-threitol(DTT,1 mg)weredissolved
in PBS (2 mL). The mixture was clarified with a 0.45 mm filter to
remove dust particles. The final concentration of DTT was
ꢁ
3
ꢁ1
3
ꢃ 10 mol L . LLS was used to in situ monitor the average
In dynamic LLS, the Laplace inversion of each measured
[
2]
molar mass of lPEIs-s, i.e., its degradation, in solution by measuring
the decrease in the scattering intensity. The weight average molar
mass of lPEIs-s can be calculated using Equation (2):
intensity-intensity time correlation function (G (q,t)) in the self-
beating mode can be related to a line-width distribution G(’). For a
diffusion relaxation, ’ is further related to the translational
2
ꢀ
ꢁ ꢀ
ꢁ
diffusion coefficient D by (’/q )c!0,q!0 ¼ D. Therefore G(’) can be
Mw ¼ MlPELsꢁs ClPEIsꢁs
=
= C=
;
ð2Þ
hI0ilPEIsꢁs
hI0i
converted to into a translational diffusion coefficient distribution
G(D) or a hydrodynamic radius distribution f(Rh) using the Stokes-
where ClPEIs-s and C are the concentrations of the initial lPEIs-s and
> are the
Einstein equation, R
h
B B
¼ k T/6phD, where k , T, and h are the
the degraded lPEIs-s, respectively, and <I
0
>
lPEIs-s and <I
0
Boltzmann constant, the absolute temperature, and the solvent
viscosity, respectively. To measure the hydrodynamic radius of the
PEI/DNA polyplexes, the solution of DNA in TE was diluted to a
time average scattering intensities of the initial lPEIs-s and the
degraded lPEIs-s at the zero scattering angle, respectively.
ꢁ1
concentration of 20 mg mL with 0.03 M NaCl solution and then
clarified using a 0.45 mm filter. The small volume of dust-free PEI
solution was gently added stepwise to the DNA solution to make
PEI/DNApolyplexdispersionswithdifferentdesiredN/Pratios.The
average hydrodynamic radius of the PEI/DNA polyplexes was
detected by dynamic LLS at a low scattering angle of 208.
2
.12. Gel Retardation Assay
PEI/DNApolyplexeswithdifferentdesiredN/Pratioswereprepared
byaddinganappropriateamountofPEI(5mL)to0.4 mgofDNA(5 mL)
in PBS. The resultant polyplex dispersions were incubated at room
temperature for 5 min and loaded on a 0.8% (w/v) agarose gel
containing ethidium bromide in TAE (tris-acetate) buffer. The gel
electrophoresis was carried out at 100 V. After 1 h, the DNA bands
in the gel were visualized with a UV (254 nm) illuminator and
photographed with a Vilber Lourmat imaging system.
2
.9. Ellman’s Assay Test
To check for the presence of mercapto groups, the Ellman reagent
was prepared by dissolving 5,5’-dithiobis(2-nitrobenzoic acid) in
0
.1 M phosphate buffer (pH ¼ 8.0) with a concentration of 0.4 mg
ꢁ1 ꢁ3 ꢁ1
mL (1.0 ꢃ 10 mol mL ). 4 mg of lPEI
i
were dissolved in 2 mL of
2
.13. Zeta-Potential Measurements of PEI/DNA
phosphate buffer. 1 mL of Ellman’s solution was then added. The
reaction mixture was shaken for 15 min in the dark at room
temperature. 1 mL of the reaction mixture was diluted with
deionized water (2 mL) and analyzed using UV-vis spectrometry.
Polyplexes
Polyplexes with different N/P ratios were prepared by adding an
appropriate amount of PEI (20 mL) to 1.6 mg of DNA (20 mL) in PBS.
The resultant polyplex dispersions were further diluted with 1 mL
E
of PBS. The average mobility (m ) of the polyplexesunder an electric
field in an aqueous solution was determined from the frequency
shift in a laser Doppler spectrum using a commercial zeta-potential
spectrometer (ZetaPlus, Brookhaven) with two platinum-coated
electrodes and one He-Ne laser as the light source. Each data point
presented in the mobility measurement was averaged over 5 times
2
.10. Cytotoxicity of PEI Vectors
The MTT assay was used to test the cytotoxicity of each polymer
used. PEI solutions with different concentrations were prepared in
4
DMEM. 293T cells were seeded in 96-well culture plate at 10 cells/
well in 100 mL DMEM medium containing 10% FBS and antibiotics
at 25 8C. The zeta-potential (jpotential) can be calculated from
ꢁ1
ꢁ1
(
penicillin at 100 units mL and streptomycin at 100 mg mL ).
4 hlaterthePEIsolutionbeingstudiedwasaddedtoeachwell. The
treated cells were incubated in a humidified environment with 5%
fðkRbÞ
m ¼ 2ej
=
, where e is the permittivity of water and
E
potential
ð3hÞ
2
1
/k is the Debye screening length.
ꢁ1
2
CO at 37 8C for 48 h. The MTT reagent (in 20 mL PBS, 5 mg mL ) was
added to each well and further incubated at 37 8C for 4 h. The
medium was then removedand replaced with 100 mL of DMSO. The
plate was gently agitated for 30 min before the absorbance (A) at
2
.14. In Vitro Gene Transfection
The gene transfection experiments were performed in 293T and
HepG2 cells using the plasmid pGL3 as an exogenous reporter gene.
The gene transfection was conducted by using the PEI/DNA
polyplexes formed with different N/P ratios (10/1, 20/1, 30/1, 40/1,
and 50/1). The cells were seeded in a 48-well plate at an initial
4
90 nm was recorded with a microplate reader (Bio-rad, USA). The
cell viability was calculated from Equation (1):
Viability ¼ ðAtreated ꢁ AblankÞ=ðAcontrol ꢁ AblankÞ ꢃ 100%
ð1Þ
Macromol. Biosci. 2014, 14, 1807–1815
1810
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.MaterialsViews.com

Revisiting Complexation between DNA and Polyethylenimine: Does the Disulfide Linkage Play. . .
www.mbs-journal.de
4
4
density of 5 ꢃ 10 per well (293T cells) or 3 ꢃ 10 per well (HepG2
cells) for 24 h before the gene transfection experiments were
conducted. Each PEI/DNA polyplex dispersion with a desired N/P
ratio was diluted in serum-free DMEM medium (200 mL) or serum-
containing DMEM medium (200 mL) and then administered to the
cells at a final concentration of 0.4 mg DNA per well. After 4 h,
the transfection medium was removed and the complete DMEM
medium (containing FBS and antibiotics, 500 mL per well) was
added. The cells were further cultured for 48 h before the transgene
expression level was evaluated. A GloMax 96 microplate lumin-
ometer (Promega, USA) and the Bio-Rad protein assay reagent were
used to determine the transfection efficiency of the PEI/DNA
polyplexes, which is expressed as a relative luminescence unit
shown in Figure 1, where adding the Ellman reagent
led to an absorbance at ꢀ412 nm in comparison with
the untreated lPEI
i
solution. After obtaining the thiol
terminated lPEI
i
chains, we coupled them together to
form longer linear lPEI chains.
The number- and weight-average molar masses of AcS-
s-s
PETOZ-SAc, the resultant lPEI , and the coupled lPEIs-s were
i
obtained using GPC and MALDI-TOF-MS (Table S1 in the
Supporting Information). The number average molar mass
of lPEI
was ꢂ 1.59 kg mol , fairly close to that measured with
has a three
i
estimated from the GPC results for AcS-PETOZ-SAc
ꢁ
1
MALDI-TOF-MS. The resultant long lPEI
s-s
(
RLU) per cellular protein (mean ꢄ SD of triplicates).
times higher molar mass than the initial short lPEI ,
i
indicating that, on average, each lPEI chain contains
s-s
three shorter linear lPEI
i
chains. Their contour lengths (if
3
. Results and Discussion
stretched) are ꢂ 15 and ꢂ 45 nm, respectively.
Figure 2 shows the kinetics of the reduction of the
Previously, coupling small branch PEI chains via disulfide
linkage sometimes resulted in large PEI clusters with a
network-like structure and a lower transfection efficiency.
disulfide bonds inside lPEI after the addition of DTT in
s-s
terms of the decrease in weight average molar mass (M )
w
measured by LLS. Here we emphasize that the molar mass is
[28]
Recently, Lee et al.
reported a method to prepare
not the absolute value because of the polyelectrolyte’s
nature. It clearly shows that the cleavage of the disulfide
degradable linear PEI chains without network formation,
but they only checked the transfection efficiency of some
large PEIs (Mn ꢂ 10 kDa). The small precursor of these PEIs
could not be tested. In our study, we purposely linked short
s-s
bonds inside lPEI takes about 48 h in PBS, although DTT
is less effective at pH ꢅ 7. The decrease of M from
w
ꢁ
1
ꢁ1
5.6 kg mol to 2 kg mol indicates that longer linear lPEI
s-s
linear PEI chains (lPEI
i
) into a long one (lPEIs-s) so that we
chains can be nearly degraded into initial shorter lPEI
chains. Due to the redox gradient across the cell membrane,
i
were able to compare their gene transfection efficiencies
and find out whether it is the disulfide bonds or the chain
length that dominantly affects gene transfection. As
discussed before, the initiator was synthesized by the
sulfonylation of ethylene glycol and was then used to
initiate cationic ring-opening polymerization from both
ends, resulting in a linear (Ts-PETOZ-Ts) chain with two
terminal tosyl ester ends that were further substituted by
thioesters. The acid hydrolysis of such a chain led to the
s-s
itisexpectedthatlongerlPEI chainswillbedegradedafter
the polyplexes enter the intracellular space.
Figure 3 shows that, as expected, the cytotoxicity of lPEI ,
i
lPEI-4.2k, and lPEI increases with the polymer concentra-
s-s
tion, where large bPEI-25k and long lPEI-25k are used as
controls. Note that in the working concentration range of
ꢁ
3
ꢁ2
ꢁ1
gene transfection (10 –10 mg mL ), large bPEI-25k and
lPEI-25karemuchmorecytotoxicthansmalllPEI ,lPEI and
i
s-s
initial linear PEI chain (lPEI
i
) with two mercapto-ends, as
lPEI-4.2k. The cytotoxicity of large cationic polymer chains
has been studied and explained before, and is presumably
due to their disruption of anionic cell and organelle
1
1
0
0
.5
.0
.5
.0
6
4
2
Ellman reagent
4
12 nm
lPEI + Ellman reagent
i
lPEIi
0
10
20
30
40
50
2
00
300
400
500
600
t / hours
λ
/ nm
Figure 2. Time dependence of weight average molar mass of lPEIs-s
ꢁ1
Figure 1. UV-vis spectra of Ellman reagent and lPEI
i
solutions
(20 mg mL ) after the addition of D,L-dithiothreitol (DTT,
ꢁ3 ꢁ1
before and after addition of Ellman reagent.
3 ꢃ 10 mol L ).
Macromol. Biosci. 2014, 14, 1807–1815
1811
www.MaterialsViews.com
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Y. Ma, B. Chen, N. He, G. Chen, L. W. Li, C. Wu
www.mbs-journal.de
1
20
lPEI
i
lPEIs-s
lPEI-25k
bPEI-25k
7
3
0
0
8
4
0
0
0
lPEIi
lPEIs-s
lPEI-4.2k
lPEI-25k
bPEI-25k
-10
-50
0
3
6
9
0
5
10
15
20
N/P
C / (
μg/mL)
Figure 3. Relative cell viabilities of lPEI
i
, lPEIs-s, lPEI-4.2k, lPEI-25k,
Figure 5. N/P ratio dependence of zeta-potential of PEI/DNA
complexes in PBS. (N/P ¼ 1, 2, 3, 5, 7, 10, 20.)
and bPEI-25k in 293T cells at different working concentrations.
[
17]
membranes so the cell undergoes necrosis.
It should
chains
atN/P ꢂ 3;furthermore, thezetapotentialreachesaplateau
at N/P ꢂ 5. This clearly indicates that the complexes
becomes positively charged at N/P ꢂ 3 and are nearly
covered by cationic PEI chains at N/P ꢂ 5. At higher N/P
ratios, most of the PEI chains added must be free in the
also be noted that linking a few short linear lPEI
i
together to form a longer lPEIs-s chain slightly increases
its cytotoxicity. However, lPEIs-s has a similar cytotoxicity
with non-degradable lPEI-4.2k, indicating that it is the
chain length rather than the degradable –S–S– linkage that
mainly determines the cytotoxicity of cationic polymer
chains.
[32,33]
solution mixture.
Our previous results have shown
that it is these free chains that promote gene transfection.
Figure 6 shows that most of the PEI/DNA polyplexes
prepared at different N/P ratios are smaller than 400 nm
Figure 4 shows that short linear lPEI
i
can completely
condense DNA at N/P ꢂ 5, reflected in the disappearance of
free DNA strips, while longer linear lPEIs-s starts to retard
DNA at N/P ꢂ 3 and also completely condenses DNA at
N/P ꢂ 5. This indicates that using the disulfide bonds to
link short lPEI chains together to form a longer lPEI chain
only slightly helps the formation of PEI/DNA polyplexes.
It is worth noting that lPEIs-s has similar condensation
ability to large bPEI-25k and lPEI-25k (results not shown)
in condensing anionic DNA to form stable polyplexes,
suggesting that the condensation ability remains once the
chain reaches a certain length.
and the average hydrodynamic radius (<R
maximum at N/P ꢂ 5. Further addition of PEI apparently
leads to a smaller <R >. It is worth noting that in the past
h
>) reaches its
h
suchadecreaseof<R >hasoftenbeenattributedtofurther
h
condensation of the polyplexes. However, those free PEI
chains present in the complex solution have a smaller size,
whichleadstoadecreaseintheaveragehydrodynamicsize,
especially at the higher N/P ratio. In our cases here, most of
the added PEI chains are free in the solution mixture after
i
s-s
h
N/P > 5. Therefore, a decrease in <R > is expected at
higher N/P ratios.
Figure 5 shows that the zeta potential of the resultant
PEI/DNA complexes changes from –30 mV to 20 mV when
morecationicPEIchainsareaddedintothesolutionmixture
of PEI and DNA, i.e., as the N/P ratio increases. In accordance
with the gel retardation results, the addition of cationic PEI
chains inverses the zeta-potential from negative to positive
Figure 7 and Figure 8 reveal that, in the absence or
presence of serum, all of the PEI vectors (short lPEI , long
i
lPEI , lPEI-4.2k, and large but more cytotoxic bPEI-25k and
s-s
600
lPEIi
lPEIs-s
lPEI-25k
bPEI-25k
4
2
00
00
0
1
3
5
7
10
20
N/P
Figure 6. Average hydrodynamic radius of PEI/DNA complexes
prepared with different N/P ratios (N/P ¼ 1, 3, 5, 7, 10, 20) in
Figure 4. Gel retardation assay of lPEI
complexes (N/P ¼ 1, 2, 3, 5, 7, 10, 20), where naked DNA
N/P ¼ 0) is used for comparison (as a control).
i
/DNA and lPEIs-s/DNA
0.03 M NaCl, where <R
scattering at 208.
h
> was measured by dynamic laser light
(
Macromol. Biosci. 2014, 14, 1807–1815
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1812
www.MaterialsViews.com

Revisiting Complexation between DNA and Polyethylenimine: Does the Disulfide Linkage Play. . .
www.mbs-journal.de
DNA
Blank
in 293T
(A)
medium and at N/P ꢂ 30 in serum-containing medium. We
also tested the transfection efficiencies in CHO and COS-7
cells and obtained similar results (Figure S6 and Figure S7 in
the Supporting Information). It should also be noted
lPEIi
lPEI-25k
11
1
0
lPEIs-s
lPEI-4.2k
bPEI-25k
that lPEI showed similar cytotoxicity to non-degradable
s-s
7
1
0
lPEI-4.2k. This indicates that disulfide bonds might not play
an important role in increasing the transfection efficiency
and reducing the cytotoxicity. The real role of disulfide
bonds should be re-considered.
3
0
1
0
DNA
Blank
in HepG2 (B)
1
1
0
lPEIi
lPEI-25k
As shown in Figure 7 and Figure 8, the transfection
lPEIs-s
lPEI-4.2k
bPEI-25k
efficiencies of lPEI , lPEIs-s, and lPEI-4.2k are comparable,
i
which is very different to our previous study, in which bPEI-
5
7
7
k with 2–3 disulfide linkages was 10 –10 times more
6
2
[29]
1
1
0
0
effective than its precursor, small branched bPEI-2k.
difference between these two studies is that the chain
The
length of branched bPEI-2k is much shorter than that of
linear lPEI
i
though they have a similar weight average
molar mass. The question then is why did the chain coupled
via the degradable disulfide bonds have a huge effect in a
previous study but nearly no effect here? Our answer is that
it is the chain length of PEI rather than the disulfide bonds
0
10 20 30 40 50
N/P
Figure 7. In vitro gene transfection efficiency of PEI/DNA
complexes prepared with different PEIs at various N/P ratios
that plays an important role in gene delivery.
(
N/P ¼ 0, 10, 20, 30, 40, 50) in serum-free medium.
[32,33]
Firstly, according to our previous studies,
there are
two kinds of PEI chains in the solution mixture – those
bound to DNA and those free in the solution mixture.
The bound chains condense DNA to form polyplexes,
providing charge neutrality. The free chains can signifi-
cantly increase the uptake rate constant of the polyplexes,
depending on the chain length. In our previous experi-
DNA
Blank
lPEI-25k
bPEI-25k
in 293T
(A)
lPEIi
1
0
6
1
0
lPEIs-s
l
PEI-4.2k
[29]
ments,
the initial branched PEI-2k chains were too
1
0
short to increase the uptake rate and protect DNA from
degradation while disulfide linked bPEI-2k, i.e., bPEI-7k, was
much longer so could work well (probably helping DNA to
escape from the endosome-lysosome pathway). This
explains why simple chain extension results in a huge
increase in the gene transfection efficiency. In the current
2
0
1
1
0
DNA
lPEIi
lPEIs-s
Blank
lPEI-25k
bPEI-25k
in HepG2
(B)
1
0
lPEI-4.2k
study, the initial linear lPEI chains are already long enough
i
because of their linear structure. Further chain extension
with disulfide bonds has little effect on gene transfection
efficiency, but leads to a higher cytotoxicity. Secondly,
our recent research indicated that DNA continuously
6
2
1
1
0
0
[
34]
releases from the polyplexes after intracellular uptake.
Some polyplexes can escape from the endosome-lysosome
pathway; others, as well as the disulfide bonds, will be
destroyed in the lysosome. There is no strong correlation
between the cleavage of the disulfide bonds and the release
of DNA from the polyplexes. Meanwhile, our current study
proved that degradable lPEIs-s does not show a higher
transfection efficiency than non-degradable lPEI-4.2k. A
combination of all these studies clearly shows that it is the
chain length rather than the disulfide bond that plays a
dominant role in promoting gene transfection. Finally, it
should be also noted that long PEI chains also induce higher
cytotoxicity. It is therefore important to control the chain
0
10 20 30 40 50
N/P
Figure 8. In vitro gene transfection efficiency of PEI/DNA
complexes prepared with different PEIs at various N/P ratios
(
N/P ¼ 0, 10, 20, 30, 40, 50) in serum-containing medium.
lPEI-25k chains) have similar gene transfection efficiency.
Unexpectedly, couplingafewshortlPEI chainsintoalonger
i
lPEI does not make it better than short lPEI and non-
s-s
i
degradable lPEI-4.2k chains with a similar length. The
maximum efficiency occurs at N/P ꢂ 20 in serum-free
Macromol. Biosci. 2014, 14, 1807–1815
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1813
www.MaterialsViews.com

Y. Ma, B. Chen, N. He, G. Chen, L. W. Li, C. Wu
www.mbs-journal.de
Acknowledgements: Financial support from the National
Natural Scientific Foundation Projects (20934005 and
5
1173177), the Ministry of Science and Technology Key Project
(
(
2012CB933802), and the Hong Kong SAR Earmarked Projects
CUHK4036/11P, 2130281; 2060431 and CUHK7/CRF/12G) is
gratefully acknowledged.
Received: May 24, 2014; Revised: August 17, 2014; Published
online: September 29, 2014; DOI: 10.1002/mabi.201400253
Keywords: degradation; disulfide bonds; gene delivery;
polyethylenimine
[
1] D. W. Pack, A. S. Hoffman, S. Pun, P. S. Stayton, Nat. Rev. Drug
Discov. 2005, 4, 581.
Scheme 2. Schematic of the effect of chain length on intracellular
trafficking. Long PEI chains can effectively help DNA to escape
from the endosome-lysosome pathway, but short PEI cannot.
[
[
2] A. Rolland, Adv. Drug Delivery Rev. 2005, 57, 669.
3] S. A. Rosenberg, P. Aebersold, K. Cornetta, A. Kasid, R. A.
Morgan, R. Moen, E. M. Karson, M. T. Lotze, J. C. Yang, S. L.
Topalian, M. J. Merino, K. Culver, A. D. Miller, R. M. Blaese,
W. F. Anderson, N. Engl. J. Med. 1990, 323, 570.
4] R. G. Crystal, Science 1995, 270, 404.
length to balance the transfection efficiency and cytotoxic-
ity. Scheme 2 summarizes the effect of chain length on
intracellular trafficking.
[
[5] E. Poeschla, P. Corbeau, F. WongStaal, Proc. Natl. Acad. Sci. USA
1996, 93, 11395.
[
6] P. L. Felgner, T. R. Gadek, M. Holm, R. Roman, H. W. Chan, M.
Wenz, J. P. Northrop, G. M. Ringold, M. Danielsen, Proc. Natl.
Acad. Sci. USA 1987, 84, 7413.
4
. Conclusion
[
7] Y. Z. Ma, S. Hou, B. Ji, Y. Yao, X. Z. Feng, Macromol. Biosci. 2010,
10, 202.
By purposely synthesizing an initial linear polyethyleni-
[8] Y. Yao, D. F. Feng, Y. P. Wu, Q. J. Ye, L. Liu, X. X. Li, S. Hou, Y. L.
ꢁ1
mine (lPEI
i
, Mw ꢂ 2 kg mol ) that is long enough and
Yang, C. Wang, L. Li, X. Z. Feng, J. Mater. Chem. 2011, 21, 4538.
[
9] H. Q. Bao, Y. Ping, Y. Z. Pan, L. Li, J. Li, L. H. Gan, Soft Matter
012, 8, 2518.
10] B. Y. Shi, H. Zhang, Z. Y. Shen, J. X. Bi, S. Dai, Polym. Chem. 2013,
, 840.
coupling a few of them together to form a longer lPEIs-s
chain via degradable disulfide bonds, we conducted a
comparative study of their cytotoxicity and in vitro gene
transfection efficiency in different cell lines. Our results
revealed that the extension of the chain length makes
2
[
[
4
11] G. Wang, H. Yin, J. C. Y. Ng, L. P. Cai, J. Li, B. Z. Tang, B. Liu,
Polym. Chem. 2013, 4, 5297.
[
[
[
12] S. D. Li, L. Huang, J. Controlled Release 2007, 123, 181.
13] M. A. Mintzer, E. E. Simanek, Chem. Rev. 2009, 109, 259.
14] O. Boussif, F. Lezoualch, M. A. Zanta, M. D. Mergny, D. Scherman,
B. Demeneix, J. P. Behr, Proc. Natl. Acad. Sci. USA 1995, 92, 7297.
longer lPEI chains slightly more cytotoxic but has nearly
s-s
no effect on gene transfection, which is different from our
previous study where the gene transfection efficiency
increased by ꢂ 5–7 orders of magnitude when three to four
[15] W. T. Godbey, K. K. Wu, A. G. Mikos, Proc. Natl. Acad. Sci. USA
1999, 96, 5177.
ꢁ1
small branched PEI chains (Mw ꢀ 2 kg mol ) were coupled
together to form a larger one (>15 nm). Such a drastic
difference enables us to hypothesize that free cationic PEI
chainswithasufficientlengthmayincreasetheuptakerate
and protect DNA from degradation. In the current study,
[
[
[
[
[
16] L. Wightman, R. Kircheis, V. Rossler, S. Carotta, R. Ruzicka, M.
Kursa, E. Wagner, J. Gene Med. 2001, 3, 362.
17] D. Fischer, T. Bieber, Y. X. Li, H. P. Elsasser, T. Kissel, Pharm. Res.
1999, 16, 1273.
18] W. T. Godbey, M. A. Barry, P. Saggau, K. K. Wu, A. G. Mikos, J.
Biomed. Mater. Res. 2000, 51, 321.
19] K. Kunath, A. von Harpe, D. Fischer, H. Peterson, U. Bickel, K.
Voigt, T. Kissel, J. Controlled Release 2003, 89, 113.
20] H. Petersen, K. Kunath, T. Kissel, A. L. Martin, S. Stolnik, C. J.
Roberts, M. C. Davies, J. Controlled Release 2003, 87, 286.
initial linear lPEI
i
chains were already long enough, so
further chain extension via disulfide bonds had nearly no
effect on gene transfection, while in our previous study, the
initialbranchedbPEI-2kchainswithasimilarmolarmassof
lPEI
i
were much smaller, so chain extension greatly affected
[21] A. M. Funhoff, C. F. van Nostrum, G. A. Koning, N. M. E.
Schuurmans-Nieuwenbroek, D. J. A. Crommelin, W. E.
Hennink, Biomacromolecules 2004, 5, 32.
[
[
gene transfection. A combination of our previous and
current results clearly shows that it is the chain length that
plays the dominant role in promoting gene transfection
rather than the degradable disulfide linkage. Our results
also revealed that in the development of non-viral vectors it
is important to control the chain length to balance the gene
transfection efficiency and cytotoxicity.
22] A. Meister, M. E. Anderson, Annu. Rev. Biochem. 1983, 52, 711.
23] G. Bellomo, M. Vairetti, L. Stivala, F. Mirabelli, P. Richelmi, S.
Orrenius, Proc. Natl. Acad. Sci. USA 1992, 89, 4412.
[24] D. S. Manickam, D. Oupicky, J. Drug Target. 2006, 14, 519.
[25] J. Liu, X. L. Jiang, L. Xu, X. M. Wang, W. E. Hennink, R. X. Zhuo,
Bioconjugate Chem. 2010, 21, 1827.
Macromol. Biosci. 2014, 14, 1807–1815
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1814
www.MaterialsViews.com

Revisiting Complexation between DNA and Polyethylenimine: Does the Disulfide Linkage Play. . .
www.mbs-journal.de
[
[
26] Q. Peng, Z. L. Zhong, R. X. Zhuo, Bioconjugate Chem. 2008, 19, 499.
27] M. Breunig, U. Lungwitz, R. Liebl, A. Goepferich, Proc. Natl.
Acad. Sci. USA 2007, 104, 14454.
[33] Y. A. Yue, F. Jin, R. Deng, J. G. Cai, Z. J. Dai, M. C. M. Lin, H. F.
Kung, M. A. Mattebjerg, T. L. Andresen, C. Wu, J. Controlled
Release 2011, 152, 143.
[
[
28] Y. Lee, H. Mo, H. Koo, J. Y. Park, M. Y. Cho, G. W. Jin, J. S. Park,
Bioconjugate Chem. 2007, 18, 13.
29] R. Deng, Y. Yue, F. Jin, Y. C. Chen, H. F. Kung, M. C. M. Lin, C. Wu,
J. Controlled Release 2009, 140, 40.
30] Z. J. Dai, C. Wu, Macromolecules 2012, 45, 4346.
31] I. Teraoka, Polymer Solutions: An introduction to physical
properties, Wiley, New York, USA 2002.
[34] Y. Ma, C. Wu, J. Mater. Chem. B 2014, 2, 3282.
[35] S. Bauhuber, C. Hozsa, M. Breunig, A. Gopferich, Adv. Mater.
2009, 21, 3286.
[36] D. L. Mohler, G. Shen, Org. Biomol. Chem. 2006, 4, 2082.
[37] Z. Y. Zhong, J. Feijen, M. C. Lok, W. E. Hennink, L. V. Christensen,
J. W. Yockman, Y. H. Kim, S. W. Kim, Biomacromolecules 2005,
6, 3440.
[
[
[
32] Y. A. Yue, F. Jin, R. Deng, J. G. Cai, Y. C. Chen, M. C. M. Lin, H. F.
Kung, C. Wu, J. Controlled Release 2011, 155, 67.
[38] S. Bauhuber, R. Liebl, L. Tomasetti, R. Rachel, A. Goepferich, M.
Breunig, J. Controlled Release 2012, 162, 446.
Macromol. Biosci. 2014, 14, 1807–1815
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1815
www.MaterialsViews.com