Biophysical Characterization of PEI/DNA Complexes

SIRIRAT CHOOSAKOONKRIANG,
1
BRIAN A. LOBO,
1
GARY S. KOE,
2
JANET G. KOE,
2
C. RUSSELL MIDDAUGH
1
1
The Department of Pharmaceutical Chemistry, The University of Kansas, 2095 Constant Avenue, Lawrence, Kansas 66047
2
Valentis, Inc., Burlingame, California 94010
Received 22 October 2002; revised 24 February 2003; accepted 14 March 2003
ABSTRACT: The maingoal of this study was to determine the effects of polyethylenimine
(PEI) molecular weight and structure (750 kDa, 25 kDa, 2 kDa branched, and 25 kDa
linear PEI) and the nitrogen/phosphate (N/P) molar ratio on the physical properties and
transfection efciencies of PEI/DNAcomplexes. Fourier transforminfrared spectroscopy
revealed that DNA remained in the B conformation when complexed to all PEIs. Unique
alterations inthe circular dichroismspectraof DNAwere observedinthe presence of each
PEI, whereas differential scanning calorimetry measurements showed that all PEIs
examined destabilized supercoiled DNAat N/P<3/1, but not at higher ratios. Isothermal
titration calorimetry revealed the existence of protonation changes at low ionic strength
due to possible shifts in pK
a
of the ionizable groups of PEI during complex formation.
Twenty-ve kilodalton branched and 25 kDa linear PEI complexes showed the highest
transfection efciencies at an N/P ratio of 6:1 in COS-7 and CHO-K1 cells, respectively.
These investigations have detected alterations in the physical and colloidal properties of
the complexes that were sensitive to polymer structure, molecular weight, and polymer/
DNA ratio, but these properties did not directly correlate with their transfection
efciencies. To further probe any possible relationship between these parameters and
activity, a more rened biophysical analysis of any subpopulations in these samples that
may differ in transfection activity is suggested, although the existence of such species
remains unknown. 2003 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm
Sci 92:17101722, 2003
Keywords: polyethylenimine; differential scanning calorimetry; isothermal titration
calorimetry; circular dichroism; DNA delivery; FTIR; gene delivery; physical character-
ization; plasmid DNA
INTRODUCTION
Gene therapy is a relatively new approach for the
treatment of genetic disorders using replacement
of a defective gene or introduction of new or
modied protein-based functions into cells to elicit
a therapeutic response. Development of efcient
and safe vehicles is one of the major obstacles to
the success of current gene delivery systems.
Gene transfer vectors currently available can be
broadly divided into two groups: viral and non-
viral. Although viral vectors have evolved to be
highly efcient gene delivery systems, nonviral
vectors have garnered considerable attention as
alternative vehicles because of their ease of syn-
thesis, unrestricted trans-gene size, low cost, and
low degree of immunogenicity.
1
Because naked plasmid DNA does not readily
penetrate most cellular membranes,
2
nonviral
gene delivery systems usually include agents to
increase gene delivery efciency. By complexation
to DNA, these agents typically form stable parti-
cles that are small enough to be readily trans-
ported and avoid rapid elimination by the
1710 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
Correspondence to: C. Russell Middaugh (Telephone: 785-
864-5813; Fax: 785-864-5814; E-mail: middaugh@ku.edu)
Journal of Pharmaceutical Sciences, Vol. 92, 17101722 (2003)
2003 Wiley-Liss, Inc. and the American Pharmacists Association
reticuloendothelial system. Upon cell entry, these
agents may also instill properties that promote
endosomal escape to avoid lysosomal degradation
and enhance nuclear entry.
One class of promising nonviral vector is the
polyethylenimines (PEI). PEI is a highly water-
soluble, positively charged, synthetic polymer in
which every third atom is a nitrogen that can be
protonated as well as provide a potential branch-
ing point. Approximately 20% of the nitrogens of
PEI are protonated under physiological condi-
tions.
3,4
As a result, the polymer can change its
ionization state over a broad pH range. The
cationic amines of both branched and linear PEI
also reduce the negative charge of DNA upon
complexation. This electrostatic interaction leads
to at least a partial condensation of the normally
large hydrodynamic volume of DNA.
5,6
This is seen
in 25 kDa PEI polyplexes as the formation of a
relatively homogenous population of toroidal par-
ticles of 4060 nm in diameter.
4
Branched and
linear PEIs of various molecular weights (MW) are
among the most transfection efcient nonviral
vectors in vitro
79
and in vivo.
2,1012
This high
transfection efciency is manifested in a variety of
target organs and delivery routes.
1317
The trans-
fection efciency of PEI depends on several factors
including polymer MW, the conformation of PEI
(e.g., linear vs. branched), and the target cell type.
This efciency may be further enhanced by the
ability of PEI to protect DNA from enzymatic
degradation.
15
Despite the potential use of PEI to deliver DNA,
the relationship between the properties of PEI/
DNA polyplexes and their ability to transfect
cells is poorly understood. Therefore, the purpose
of this study is to thoroughly characterize com-
plexes formed between DNA and various forms
of branched (2, 25, and 750 kDa) and linear PEI
(25 kDa) by a wide variety of biophysical methods.
We then use this information in an attempt to
correlate various physical properties with trans-
fection efciency in cell culture. To this end, the
secondary structure of DNA upon complexation
with PEI was investigated by using Fourier
transform infrared (FTIR) and circular dichroism
(CD) spectroscopy. The thermal stability of plas-
mid DNA complexed to PEI was studied using
differential scanning calorimetry (DSC). Addition-
ally, the enthalpy of binding between PEI and
DNA was explored using isothermal titration
calorimetry (ITC) and light scattering studies
were performed to assess the particle sizes and
zeta potentials of the complexes.
EXPERIMENTAL SECTION
Materials
Plasmid DNA pMB 290 (4.9 kbp), pMB 237
(9.1 kbp), and pMB 401 (encoding rey lucifer-
ase) [all >95% supercoiled (sc)] were provided by
Valentis, Inc. (Burlingame, CA). The DNA con-
centration was determined by UV absorbance
at 260 nm using an extinction coefcient of
0.02 (mg
1
cm
1
mL).
Branched PEIs (MW 750, 25, and 2 kDa) and
linear PEI (MW 25 kDa) were obtained from
Aldrich (Milwaukee, WI) and Polysciences, Inc.,
(Warrington, PA), respectively. The polymers
were used without further purication.
COS-7 and CHO-K1 cells were obtained from
American Type Culture Collection (Rockville,
MD). Dulbeccos modied Eagles medium, Hams
F-12 medium, and phosphate buffered saline
(PBS) were acquired from BioWhittaker (Walk-
ersville, MD). Opti-MEM (reduced-serum modi-
cation of MEM) and trypsin-EDTA were
purchased from Gibco (Greenland, NY). Fetal
bovine serum was obtained from Atlanta Biologi-
cals (Atlanta, GA). MTT [3-(4, 5-cimethylthiazol-
2-yl)-2, 5-diphenyl tetrazolium bromide] was
obtained from Sigma (St. Louis, MO).
All buffer materials (piperazine, MES, cacodylic
acid, phosphoric acid, BES, EDA, EPPS, PIPES,
ACES, boric acid, ethanolamine, HEPES, TES,
MOPS, TEA, and Tris) were obtained from Sigma
(St. Louis, MO). Nano-puried water was used to
prepare buffer solutions.
Preparation of Complexes
Stock PEI and DNA solutions were prepared
before each experiment at various molar ratios
of PEI nitrogen (N) to DNA phosphate (P) up to
N/P=10. The pH of the stock PEI solution was
adjusted to the desired pH using HCl. Complex
formation always utilized solutions of equal
volumes with the least concentrated component
being added to the more concentrated one.
Samples were continuously stirred during addi-
tion and equilibrated at room temperature for
~20 min before measurement. Complexes were
freshly prepared before each individual measure-
ment. Complexes were formed in 10 mM Tris
buffer, pH 7.4 unless otherwise noted.
FTIR Spectroscopy
FTIRspectra were obtained with a Nicolet Magna-
IR 560 spectrometer equipped with a mercury
PEI/DNA COMPLEXES 1711
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
cadmium telluride detector (Madison, WI). Sam-
ples were measured with attenuated total reec-
tance geometry where the sample in solution was
placed directly in the well of a ZnSe plate (effective
pathlength 12 mm). Spectra were obtained at
4-cm
1
resolution under a dry air purge by
accumulation of 256 interferograms. Subtraction
of the solvent (10 mM Tris buffer) was done using
the association peak of H
2
O near 2200 cm
1
as a
reference point.
18
Additional data analysis includ-
ed baseline correction (1804 to 904 cm
1
), seven
point Satvisky-Golay smoothing (if needed) and
Omnic Peaknd software. The nal DNA concen-
tration of all samples was 1 mg/mL whereas
the PEI concentration was varied for individual
complexes.
CD
CD spectra were obtained using a 0.1-cm path-
length rectangular cell at 258C, a Jasco J-720
spectropolarimeter (Easton, MD), and were cor-
rected by subtraction of buffer spectra. Spectra
were recorded from 350 to 200 nmat a scan rate of
20 nm/min and a resolution of 0.5 nm. Three
spectra were accumulated and averaged for each
sample. The nal DNA concentration of all
samples was 50 mg/mL (1.54 10
4
M DNA
bases). The CD signal was converted to molar
ellipticity [y], deg l mole
1
cm
1
, smoothed with
a Jasco Fast Fourier transform algorithm, and
then baseline adjusted to zero at 345 nmto correct
for a small contribution by differential light
scattering.
DSC
DSC was performed with a model 5100 Nano-DSC
[Calorimetry Sciences Corporation (CSC), Amer-
ican Fork, UT]. Measurements consisted of a
single scan from 0 to 1208C at 18C/min under
3 atm of pressure. All samples were prepared in
5 mMphosphate buffer, pH7.4 and were degassed
before measurement. Buffer exchange of DNA
was performed by dialysis using a Slide-a-lyzer
10,000 MWCO dialysis cassette (Pierce, Rockford,
IL) in 2.0 L buffer at 58C overnight. Baselines
were obtained by scanning with buffer in both the
sample and reference cells. Samples were ana-
lyzed in duplicate or triplicate. CpCalc software
(CSC) was used to subtract the baseline from the
sample thermogram. The data were converted to
molar heat capacity using the MW and concentra-
tion of DNA (0.5 mg/mL). All experiments used
pMB237 except for complexes of linear 25 kDa
PEI which used pMB290.
ITC
Calorimetric titrations of PEI into DNA were
conducted at 258C with a titration program of
25 injections of 10 mL using a CSC model 4200
isothermal titration calorimeter controlled by
ITCRun software (CSC). Titrations consisted of
an equilibration time of 300 s to establish a
baseline followed by injections at 5-min intervals.
Dialysis into the required buffers and degassing
were performed on all samples before use. DNA
and PEI concentrations were 0.31 and 5.45 mM,
respectively. The titration was stopped after
DNA/PEI aggregation occurred concurrent with
a loss of binding heat. Bindworks
TM
3.0 software
was used to integrate the raw data. Blank titra-
tions of PEI into buffer were performed and the
heats obtained were subtracted from each binding
titration of PEI into DNA. The observed enthalpy
of binding of PEI to DNA (DH
obs
) was calculated
from the average of the rst ve injections of the
binding titration, divided by the molar amount of
PEI added per injection.
pH Titrations of PEI
pH measurements of HCl titrations of PEI
solutions were performed using a Mettler Toledo
MP220 pHmeter (0.01 pHunit sensitivity) and an
accuT pH electrode. A water bath temperature
controller was used to maintain a sample tem-
perature of 25.08 0.18C. Titrations consisted of
incremental additions of 0.1 N HCl into 5 mL of
PEI solution (5.0 mg/mL) prepared in water. Each
titration was performed without prior pH adjust-
ment of the PEI solution.
Particle Size and Zeta-Potential Measurement
Samples were prepared in 10 mM Tris buffer pH
7.4, which was previously ltered through 0.2-mm
polysulfone lters (Gelman Science, MI). The
DNA concentration was held constant at 100 mg/
mL whereas the N/P ratios of the PEI/DNA com-
plexes were varied. Mean hydrodynamic dia-
meters were determined by cumulant analysis
using a dynamic light scattering (DLS) instru-
ment (BT 9000AT) equipped with a 50-mW HeNe
laser with a 532-nm emission wavelength (Broo-
khaven Instruments Corp., Holtsville, NY). Scat-
tered light was monitored at 908 to the incident
1712 CHOOSAKOONKRIANG ET AL.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
beam and the mean hydrodynamic diameter was
obtained from the diffusion coefcient using the
Stokes-Einstein equation. Three continuous mea-
surements of 1-min duration were taken for each
sample and the results averaged.
The zeta potentials of the PEI/DNA complexes
were determined by phase analysis light scatting
(PALS) at a scattering angle of 158 at 258C with a
Zeta PALS instrument (Brookhaven Instruments
Corp.). An electric eld strength between 14 and
16 V/cmwas used. Data were collected with 1015
cycles of the electric eld for each experiment and
averaged. The zeta potential was calculated from
the measured electrophoretic mobility using the
Smoluchowski approximation. The same samples
used for DLS measurements were used for zeta-
potential analysis.
Transfection Studies
Preparation of complexes was conducted as
described above, with the exception that a solu-
tion of plasmid pMB 401, encoding rey lucifer-
ase at 50 mg/mL was used. The resulting
complexes were diluted 10-fold into Opti-MEM
just before application to cells. All cells were
maintained in 75-cm
2
asks at 378C and 5% CO
2
with COS-7 cells grown in Dulbeccos modied
Eagles medium with L-glutamine and 4.5 g/L
glucose and CHO-K1 cells in Hams F-12 media
with L-glutamine, each supplemented with 10%
fetal bovine serum. Cells were subcultured every
4 days using standard procedures with trypsin/
EDTA for cell lifting. Before seeding, the cells
were trypsinized, counted, and diluted to a
concentration of approximately 80,000 cells/mL.
Then 0.1 mL of this dilution was added to each
well of a 96-well plate and the cells were
incubated in a humid 5% CO
2
incubator at 378C
for 1820 h. Immediately before transfection, the
cells were washed once with PBS and the complex
(250 ng of DNA) was added to each well. Cells
were incubated with the complexes for 5 h. The
transfection agent was then removed and 100 mL
of culture medium was added followed by a
further incubation of 48 h. A luciferase expression
assay was performed using the Luciferase Assay
System
TM
from Promega (Madison, WI) following
the manufacturers recommended protocol. The
cells were washed with PBS and 20 mL of 1X lysis
buffer was added per well. The cells were allowed
to stand at roomtemperature for 30 min. The plate
was then placed into a Fluostar
TM
Galaxy micro-
titer plate reader (BMG, Germany) equipped with
an injector. A volume of 100 mL of luciferase assay
reagent was added to each well immediately
before measurement. A luciferase standard cali-
bration curve was obtained and used to convert
light units to nanograms of luciferase. The data
are reported as the meanstandard error for a
minimum of three to ve samples per data point.
Assessment of Cytotoxicity (MTT Assay)
Both COS-7 and CHO-K1 cells were grown as
described in the transfection experiments. Cells
were treated with the PEI/DNA complexes (using
the same N/P ratio used in the transfection
studies) or PEI alone (using the same concentra-
tion of PEI present in the complexes) for 5 h. The
complexes or PEI were then removed and 100 mL
of culture medium was added as needed for each
cell type. The cells were then incubated for an
additional 40 h at 378C. At this point, 11 mL of
MTT was added to each well and the cells were
incubated at 378C with 5% CO
2
for 3 h. An aliquot
of 110 mL of MTTSS (10% Triton X-100 0.1 N
HCl in 125 mL of isopropanol) was added to each
well and the cells were incubated at 378C over-
night. The absorbance at 570 nm was measured
using a Fluorostar
TM
microtiter plate reader and
the percent cell viability was calculated using the
following equation:
% Cell viability
=
[Absorbance of the test sample + 100[
[Absorbance of control (cells alone)[
(1)
RESULTS
Infrared Spectral Properties of PEI/DNA Complexes
In Figure 1A, representative infrared spectra of
DNA and its complexes with 750 kDa PEI are
shown stacked in order of increasing N/P ratio,
with uncomplexed DNA at the bottom and
PEI alone at the top. In the absence of polymer,
DNA is in the B conformation as indicated by
the presence of the guanine/thymidine (G/T)
carbonyl stretching band at 1714 cm
1
(represen-
tative of interstrand base-pairing), an asymmetric
phosphate stretching vibration at 1224 cm
1
,
and a sugar-phosphate coupled vibration at
970 cm
1
.
19,20
Additional bands near 1328, 1281,
and 897 cm
1
(the latter not shown) support
this assignment of the B conformation.
21,22
DNA
vibrational bands arising from base carbonyls
PEI/DNA COMPLEXES 1713
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
(1715 cm
1
), the imidazole nitrogen of guanine
(1492 cm
1
), and the phosphate backbone
(1224 cm
1
) have been found to be particularly
sensitive to changes in cationic lipid/DNA com-
plexes.
19,23
Interference from PEI CH
2
scissoring
(1471 cm
1
) and NH bending (1515 cm
1
)
vibrations, however, make it difcult to consis-
tently resolve the 1492 cm
1
guanine imidazole
nitrogen mode of DNA. Therefore, only the base
carbonyl and antisymmetric phosphate stretching
vibrations were monitored in this study.
Discrete changes in these two vibrational
frequencies upon addition of PEI to DNA are
shown in Figure 1BI. The frequency of the base
in-plane C

Odouble-bond vibration is plotted as a

function of N/P ratio in panels BE. Increases in
the peak frequency of this vibration are seen in
complexes formed at increasing N/P molar ratios
containing the 2, 25, and 750 kDa branched PEIs,
with the 25 kDa branched polymer producing
the largest increase. When linear PEI is used, an
abrupt increase in peak position is seen only at a
low N/P ratio of 0.5. The peak position of the DNA
asymmetric phosphate vibration is plotted with
respect to the N/P ratio of the complex in panels
FI of Figure 1. Decreases in the value of
this vibrational frequency are observed in com-
plexes containing each MW of PEI, although the
trends differ between the individual polymers. In
contrast, no change inpeak positionis observed for
the symmetric phosphate vibration (~1089 cm
1
)
(data not shown), which has generally been found
to be independent of DNA geometry.
19
CD of PEI/DNA Complexes
CD spectroscopy was also used to monitor the
secondary structure of DNA in PEI/DNA com-
plexes. PEI has no signicant CD signal within
the UV region monitored. Therefore, the observed
Figure 1. FTIR absorbance spectra of PEI/DNA complexes in solution. The spectra
were collected and processed as described in the text. The ratio of 750 kDa branched PEI
nitrogen/DNAphosphate is indicated onthe left withthe exceptionof DNAandPEI alone
which are also indicated on the left. The spectra are all representative of complexes at
1 mg/mLDNAconcentration (A). Effect of the molar ratio of PEI/DNAcomplexes onDNA
vibrational modes in solution. The peak positions of the DNA base carbonyl (BE) and
DNA antisymmetric phosphate stretch (FI) are plotted against the N/P ratio of PEI/
DNA. The different MWs of PEI are indicated in the individual panels.
1714 CHOOSAKOONKRIANG ET AL.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
signals arise entirely from the DNA molecules.
The spectra of PEI/DNA complexes containing
each of the four different PEI polymers are shown
in Figure 2AD. All PEI/DNA complexes pre-
pared at N/P ratios between 2 and 4 formed
microaggregates (see below, Fig. 3A) giving
artifactual spectra due to precipitation and/or
differential light scattering.
24
The CD spectrum
of uncomplexed plasmid DNA is typical of the
B conformation consisting of a positive peak at
274 nmand a negative minimumnear 246 nm.
25,26
Upon PEI complexation, both regions of the DNA
CD spectrum are altered. Generally, the spectra
show a decrease in the value of the molar ellip-
ticity of the positive band, concomitant with a
red shift as the PEI fraction of the complexes is
increased. In most cases, the negative band also
shifts in position to higher wavelengths and
decreasing molar ellipticity values with increas-
ing N/P ratio. Complexes of DNA containing the
lowest MW PEI display a rapid decrease in both
bands with their values near their minimum at an
N/P ratio of 1 (Fig. 2A). The decrease that is seen
in the intensity of these bands with 750 kDa PEI/
DNA complexes is more gradual with increasing
N/P ratios and is similar to that observed with
cationic lipid/DNA complexes.
24,25,27,28
The linear
PEI/DNA complexes show a dramatic decrease in
Figure 2. Effect of PEI MW and N/P ratio on the CD spectra of DNA (pMB 290) and
PEI/DNA complexes. (A) PEI (2 kDa)/DNA complexes; (B) PEI (750 kDa)/DNA; (C) PEI
(25kDa)/DNAcomplexes; and(D) PEI (25kDalinear)/DNA. Legend: (&) =plasmidDNA;
(*) =N/P 0.5; (~) =N/P 1; () =N/P 2; (*) =N/P 4; (~) =N/P 6; and (&) =N/P 10.
Figure 3. Hydrodynamic size (A) and zeta potential
(B) of PEI/DNA complexes as a function of increasing
N/P molar ratio. The data represent the mean and
standard error of at least three separate measurements.
PEI/DNA COMPLEXES 1715
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
the intensity of the positive peak at N/P=0.5 but
undergo no further spectral shifts at N/P ratios
>4. However, the branched 25 kDa PEI /DNA
complexes show a gradual decrease in intensity
of the positive peak at 276 nm with a shift to
higher wavelength (~282 nm) culminating at N/P
ratios >6.
DSC of PEI/DNA Complexes
The thermal stability of DNA in PEI/DNA com-
plexes was investigated using DSC (Fig. 4). The
thermal disruption of the sc form of the plasmid
is detected as a broad transition above 908C
whereas the linear and nicked open circular (lin/
oc) species produce a series of small transitions
within the 608908C range.
29
The 2 and 750 kDa
forms of PEI stabilized the lin/oc DNA species at
N/P ratios below 2:1 (750 kDa) and 1:1 (2 kDa).
The sc DNA in these complexes, however, was
destabilized and its transition broadened. Colloi-
dal instability of 2 kDa PEI/DNA complexes (due
to the high concentration required for DSC)
precluded their investigation at N/P ratios >1.
The 750 kDa PEI/DNA complexes display a
shoulder near 908C at N/P ratios above 2, possibly
resulting from the stabilization of lin/oc DNA.
Complexes containing 25 kDa PEI and DNA are
distinct from all other MW PEIs tested in that
both the sc and lin/oc forms of DNA are destabi-
lized at N/P _2 (Fig. 4C). Stabilization of all forms
of DNA, however, was observed in the presence of
a charge excess of this polymer. Linear 25 kDa
PEI in contrast, destabilized only the sc forms of
DNA at N/P ratios <2 (Fig. 4D). Complexes con-
taining the linear PEI at N/P ratios >2 were not
colloidally stable and manifested large exother-
mic (negative) peaks upon sc DNA disruption (not
shown).
ITC
To explore any possible differences in the ener-
getics of complex formation between these PEIs
and DNA, calorimetric titrations of each PEI into
DNA were performed at 258C in the presence of
buffers of varying ionization enthalpy, at low
ionic strength (<0.01) and within the pH range
6.09.0. An afnity constant for the binding of
PEI to DNA could not be obtained because the
Figure 4. Effect of PEI MW and N/P ratio on the thermal stability of plasmid DNA in
PEI/DNAcomplexes. DSCthermograms of DNA(pMB237) and complexes at various N/P
molar ratio. (A) PEI (2 kDa)/DNA complexes (from bottom to top: DNA, PEI/DNA at N/
P=0.25, 0.5, 0.75, and1, respectively); (B) PEI (750kDa)/DNAcomplexes (frombottomto
top: DNA, PEI/DNA at N/P=0.5, 1, 2, 4, and 8, respectively); (C) PEI (25 kDa)/DNA
complexes (from bottom to top: DNA, PEI/DNA at N/P=0.5, 1, 2, 4, 8, and 10,
respectively); and (D) PEI (25 kDa linear)/DNA(pMB290) complexes (frombottomto top:
DNA, PEI/DNA at N/P=0.25, 0.5, 0.75, 1, and 2, respectively).
1716 CHOOSAKOONKRIANG ET AL.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
complexation process was limited by aggregation
of the complexes (not shown). The observed
binding enthalpies of PEI (25 kDa) to DNA as a
function of pH and buffer ionization enthalpy are
shown in Figure 5A. The binding enthalpy of this
PEI to DNA is dependent on the pH and the
ionization enthalpy of the buffer, which suggests
that protonation changes occur during complex
formation. Other PEIs also displayed similar
variations in binding enthalpy under these con-
ditions (not shown). Theoretically, the observed
binding enthalpy (DH
obs
) is linearly dependent on
the buffer ionization enthalpy (DH
ioniz
). The y-
intercept of such a relationship corresponds to the
buffer-independent binding enthalpy (DH
o
) and
the slope to the degree of protonation of the ligand
(n), as dened by the following equation
30
:
DH
obs
= DH
o
n DH
ioniz
(2)
Positive n values indicate that PEI is protonated
upon binding to DNA, whereas the negative y-
intercepts correspond to favorable (exothermic)
buffer-corrected enthalpies of binding. These
changes in protonation presumably result from
an increase in pK
a
of the primary, secondary, and
tertiary amine groups of PEI upon complex
formation.
A plot of the experimentally determined values
of n for each PEI versus pH (Fig. 5B) permits a
comparison of the pH-dependent changes in pro-
tonation during complexation of each polymer to
DNA. The 2 and 750 kDa PEIs clearly display two
maxima in this plot, which indicates that these
polymers undergo two distinct shifts in pK
a
in this
pH range upon complexation with DNA. Because
the branched PEIs have three different ionizable
groups (18, 28, and 38 amines), the detection of only
two ionization events suggests that two of the
three protonatable groups onthe PEI polymer may
be cooperatively shifting inpK
a
, producing asingle
ionization event. These two groups are probably
the primary and secondary amines of PEI. Ter-
tiary amines would experience a greater degree
of electrostatic suppression of protonation by the
adjacent primary and secondary amines and
would therefore be the most acidic species.
Acid Titration of PEIs
Investigations to quantify the buffering capacity
and the pK
a
s of the amines of PEI in the free state
used acid titrations of each polymer in unbuffered
solution at 258C (Fig. 5C). The buffer capacity of
each polymer was obtained from the reciprocal
Figure 5. (A) A representative plot of the variations
in the observed binding enthalpy of PEI to DNA
measured using ITCversus buffer ionization enthalpies
at 25.08C and I <0.01. The pH of each data set is
indicated in the gure. The buffers used and their
corresponding ionization enthalpies are described in
reference.
49
(B) The pH dependency of the degree of
PEI protonation (n) obtained from ITC titrations of
different MWs of PEI into DNA. Legend: (~) =25 kDa
linear PEI; (~) =2 kDa PEI; (^) =25 kDa PEI;
(&) =750 kDa PEI. (C) Buffering capacity (d[HCl]/
dpH) versus pH from acid titrations of PEI at 25.08C.
Inset: rawdata of a representative acid titrationcurve of
PEI with HCl.
PEI/DNA COMPLEXES 1717
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
slope (d[HCl]/dpH)
31
of the raw pH titration data
(inset in Fig. 5C) and the pK
a
s were then
estimated from the maxima of these buffer capa-
city curves. The largest buffering capacity of all
polymers was above pH 8.0. The pK
a
of PEI in this
pH range decreased with increasing MW of
branched PEI: pK
a
~9 for 2 kDa PEI, 8.5 for
branched and linear 25 kDa PEI, and 8.3 for
750 kDa PEI. In the pH range 6.510.0, a similar
buffering capacity was observed between the
linear and branched 25 kDa PEIs. In the pH
range 4.06.0, however, a small maximum was
observed with branched but not linear PEI. This
maximum probably represents the average pK
a
of primary amines present in branched but not
linear PEI. Therefore, the large buffering capacity
above pH 7 would appear to be due to the secon-
dary amines that are present in all PEIs.
DLS and Zeta-Potential Measurement
of PEI/DNA Complexes
The mean hydrodynamic diameter and zeta poten-
tial of PEI/DNA complexes were assessed by DLS
and PALS, respectively. The average size of all
PEI/DNA complexes was approximately 100 nm
at the highest and lowest N/P ratios tested and
independent of PEI MW. At intermediate N/P
ratios (24), an approximately two-fold increase
in size was observed with several of the polymers
(Fig. 3A). In all cases, the polydispersity of the
complexes complex was <0.2, indicating a fairly
narrow size distribution. In only one case, 2 kDa
PEI at N/P=2, did the complexes extensively
aggregate (d>1 mm). The zeta potentials of these
complexes are shown in Figure 3B. As expected,
zeta potentials are negative for complexes at low
N/P (<2) ratios and are positive for complexes at
higher values (N/P>4).
Transfection and Cytotoxicity
The transfection efciency and toxicity of PEI/
DNA complexes were examined in COS-7 and
CHO-K1 cells. With the exception of 2 kDa PEI,
all PEIs showed a maximum in their dependency
of transfection efciency on N/P molar ratio
(Fig. 6). A statistical analysis of these data using
analysis of variance found these trends in gene
expression with varying N/P ratio to be statisti-
cally signicant inall cases, except for the 750 kDa
PEI in CHO-K1 cells and for the 2 kDa PEI in both
cell lines (Table 1). Although the 25 kDa branched
PEI produced the highest transfection efciency
in COS-7 cells, the 25 kDa linear form showed
maximum effectiveness in CHO-K1 cells. Overall,
complexes containing the 2 kDa PEI were much
less effective than any other polymer tested. The
cytotoxicity of PEI/DNA complexes and individual
PEI polymers were evaluated as a function of N/P
ratio and polymer concentration by an MTT assay
(data not illustrated). Cell viability decreased
from 100 to 8090% with increasing N/P ratio in
both CHO-K1 and COS-7 cells. Thus, PEI/DNA
complexes were not highly toxic under these
conditions. Complexes were approximately 10%
less toxic than PEI alone.
DISCUSSION
PEI is a commonly used nonviral vector for in vitro
gene delivery. Little is known, however, about
Figure 6. Transfection of COS-7 (A) and CHO-K1
cells (B) using different MWs of PEI as a function of N/P
molar ratio. Luciferase activity was measured 48 h post-
transfection. The data represent nanograms of
expressed luciferase determined as described in Experi-
mental Section. Data represent the average of 46
replicates and the error bars represent the standard
error of the mean.
1718 CHOOSAKOONKRIANG ET AL.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
any possible relationship between the biophysical
properties of PEI/DNA complexes and their
transfection efciencies. Therefore, the main goal
of this study was to measure a wide variety of
physical characteristics of PEI/DNA complexes
and to determine if any of these properties
correlate with in vitro transfection efciency.
As described in previous reports, FTIR has
proven to be a practical tool for characterizing the
secondary structure of DNA in cationic lipid/DNA
complexes.
19,20,23
These reports have clearly
shown that DNA remains in the Bform in cationic
lipid/DNA complexes. In the present study, we
show that DNA is also maintained in the B form
when complexed with different amounts, MWs,
and forms of PEI. We therefore conclude that the
interaction between PEI and DNA causes very
little if any alteration to the overall helical form of
plasmid DNA. An interaction between PEI and
DNA is clearly indicated, however, by the changes
in position of the DNA carbonyl and asymmetric
phosphate vibrations. The reduced frequency of
the asymmetric phosphate stretching vibration
can be directly attributed to electrostatic interac-
tions between PEI and DNA. An earlier study of
cationic lipid/DNA complexes found a 3-cm
1
shift
to higher frequencies. This difference is probably
due to greater dehydration at lipid/DNA rather
than PEI/DNA interfaces. A study of PEI/DNA
interactions in multilayer lms has also indicated
that interactions occur between DNA phosphates
and the amino groups of PEI through both electro-
static and hydrogen-bonding interactions.
32
The
increase in the base carbonyl vibrational frequen-
cy can be interpreted as an alteration in the
hydrogen bonding of this group either through
direct PEI-base interactions or throughchanges in
the hydration state of DNA.
The CD spectra of PEI/DNA complexes suggest
some type of change in the structure or rearrange-
ment of DNA upon complexation, as shown by
discrete changes in DNA CD peak positions and
intensities. The spectra can be placed into two
categories based on spectral similarities: those
with an N/Pmolar ratio <4 (an excess of DNA) and
those with an N/P ratio >4 (an excess of PEI). It is
possible that these categories reect an actual
rearrangement of the complex structure above and
below charge neutrality. Previous studies suggest
that these CD changes cannot be explained by
differential scattering or absorption attening
artifacts.
24,33
A previous hypothesis that such
alterations in CD spectra arise from conversion
of the DNA to C form is not supported by the FTIR
results.
25,28
Collapse of the DNA into supramolec-
ular chiral structures whose optical activity op-
poses that of B-form DNA seems possible but has
not been supported by detailed CD, FTIR, Raman,
and molecular dynamic studies of cationic lipid/
DNA complexes
24,33
which display changes in CD
spectra similar to those described here for PEI.
With respect to cationic lipid/DNA complexes, it
seems that these CDchanges are best explainedby
limited local changes in base/base interactions. If
these interactions between the bases decrease in
the complex, ellipticity maxima would be expected
to decrease as seen here. Previous work by Fish
et al.
34
and Chen et al.
3537
with model systems
supports such an explanation. Thus, we believe
such local structural perturbations due perhaps to
a direct interaction between PEI and the DNA
bases (as reected in the change seen in the carbo-
Table 1. Statistical Analysis of PEI Transfection by ANOVA (Analysis of Variance)
Polymer/Cell Line
Degrees of Freedom
(Effect, Error) F p Mse
COS-7
25 kDa linear (3,12) 9.843 0.0015 3.235
25 kDa branched (3,16) 6.266 0.0051 28.97
750 kDa branched (3,8) 23.26 0.0003 0.4551
2 kDa branched (3,16) 1.361 0.29
a
4.89 e-7
CHO-K1
25 kDa linear (3,16) 4.522 0.018 4.522
25 kDa branched (3,16) 8.362 0.0014 0.031
750 kDa branched (3,16) 2.342 0.11
a
0.1358
2 kDa branched (3,16) 2.994 0.062
a
3.66 e-7
a
Differences between expression levels of N/P ratios of 410 are not statistically signicant
(p>0.05, F~1).
PEI/DNA COMPLEXES 1719
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
nyl stretching region with FTIR) provide the most
parsimonious explanationfor the observedresults.
The N/P ratios at which FTIR and CD changes
reach their maximal effect appear to coincide with
transfectionefciency. It seems unlikely, however,
that the observed spectral changes in the DNA
component are directly related to transfection. In
fact, the spectral values generally reach their
maximal intensity or position shift at an N/P of 6
or 8 and remain constant at higher ratios, whereas
transfection efciency decreases above an N/P
ratio of 6. This discrepancy at highN/Pratios could
be due to excess unbound PEI inhibiting transfec-
tion by competing for binding sites on cell surface
proteoglycans.
38
Differences in transfection of
complexes containing different MW PEIs also do
not correlate with the observed spectral changes.
Thus, the overall spectral properties of complexes
provide an indication of the interaction between
PEI and DNA but only indirectly relate with
transfection efciency.
DSC studies of PEI/DNA complexes suggest an
effect of complexation on the thermal stability of
the sc DNA component. A destabilization of sc
DNA was observed in complexes at low N/P ratios
withall forms of PEI. The decrease instabilityseen
at lowN/Pratios probably reects changes inDNA
tertiary structure, consistent with a loss of nega-
tive supercoiling of the plasmid and perhaps the
decreases in base/base interactions suggested by
the CDresults. The lin/oc forms of DNA(present as
impurities) are, in contrast, generally stabilized at
low N/P ratios. These DNA forms are not topolo-
gically constrained and reveal that neutralization
of the DNA increases its thermal stability. The
contrasting effect of PEI on the thermal stabilities
of sc andlinear/oc DNAhas also beendetectedwith
PLLbut not polyarginine.
29
Once again, there is no
apparent direct correlation between these stabili-
zation (destabilization) effects and transfection
among the various forms of PEI.
The ITC titrations were able to directly demon-
strate the protonationof PEI uponbindingto DNA.
The degree of protonation of PEI, however, is low
compared with that of the helper lipid DOPE in
DOTAP/DOPEliposomes.
39
No apparent relation-
ship exists between the different sizes of PEI and
the degree of the protonation of PEI upon com-
plexation with DNA in the pH range below 7.5.
It has been proposed that the buffering of the
endosome interior by PEI induces osmotic swelling
and subsequent endosomal disruption, releasing
DNA for transport to the nucleus.
7
For all PEIs,
the region of highest buffering capacity, typically
between 8 and 9.5, lies above the physiological pH
range. Based on these physical characteristics of
PEI/DNA complexes, this hypothesis is not sup-
ported. Godbey et al.
40
also showed that there was
no difference in the endosomal pH in the presence
of PEI compared with the control cells. Moreover,
Suh et al.
3
demonstrated that unfavorable electro-
static inductive effects on amine protonation occur
with increasing ionic strength under physiological
conditions.
It is generally thought that the size and surface
charge of gene delivery complexes are important
factors in modulating their cellular uptake.
41
No
major differences were observed in the size or zeta
potential of complexes prepared with different
MW PEIs. Although all complexes show a similar
patternof zetapotential (inverting insignbetween
an N/P range of 24), maximum transfection
efciency seems to occur near an N/P ratio of 6,
arguing against any apparent correlations be-
tween these parameters and activity.
We do nd that the ratio of cationic polymer to
DNA and the MW of PEI are factors that strongly
inuence transfectionefciency. This is consistent
with previous results.
2,11,17,4244
Two kilodalton
PEI forms complexes that do not transfect either
COS-7 or CHO-K1cells. This size of PEI has shown
efcient transfection in vitro only in combination
with replication-defective adenoviruses.
45
In this
study, only positively charged complexes (>4 N/P)
successfully transfected cells. It is probable that
these complexes require a net positive charge to
facilitate interaction with cell surface proteogly-
cans, an event that seems necessary for efcient
transfection.
38
Alternatively, a recent report has
found that the majority of PEI molecules (~86%)
exist in the uncomplexed state in complexes
prepared at N/P ratios active in transfection
(presumably at N/P~4).
46
This nding implies
that in vivo, if any free PEI were able to diffuse
away from the complex, the remaining particle
(if initially formed at an N/P ratio of 6) would be
approximately charge neutral. This hypothesis
may explain the enhanced distribution and ef-
ciency of PEI complexes formed at this N/P ratio
in vivo.
10
Under these conditions, the toxicity of all
PEIs and their DNAcomplexes were fairly lowand
therefore do not explain the loss of transfection
efciency at high N/P ratio.
Insummary, this studyhas investigatedseveral
spectral and calorimetric properties of PEI/DNA
complexes, which do not correlate with their levels
of in vitro transfection. One limitation (although
currently untested) may be the intrinsic hetero-
1720 CHOOSAKOONKRIANG ET AL.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
geneity of these complexes in which one or more
subpopulations may be primarily responsible for
most of the transfection activity. Subpopulations
of cationic lipid/DNA complexes possessing high
activity and low toxicity have been isolated using
density gradient centrifugation, suggesting that
unique compositional or structural features may
be present with these complexes.
47,48
Thus, the
identication and isolation of such a population
and a biophysical comparison to the less efcient
ones may yield more correlative results than those
presented here. If this effort also fails to reveal
correlations to transfection, it seems likely that
more complex relationships between physical
properties of complexes and transfectionefciency
may be present. Nevertheless, these biophysical
studies much better dene nonviral gene delivery
vehicles froman analytical perspective, a result of
signicant pharmaceutical utility.
REFERENCES
1. Huang L, Viroonchatapan E, editors. 1999. Non-
viral vectors for gene therapy. San Diego: Academic
Press.
2. Godbey WT, Wu KK, Mikos AG. 1999. Poly(ethyle-
nimine) and its role in gene delivery. J Control
Release 60(23):149160.
3. Suh J, Paik H-J, Hwang BK. 1994. Ionization of
poly(ethylenimine) and poly(allylamine) at various
pHs. Bioorg Chem 22:318327.
4. Tang MX, Szoka FC. 1997. The inuence of polymer
structure on the interactions of cationic polymers
with DNA and morphology of the resulting com-
plexes. Gene Ther 4(8):823832.
5. Pelta J, Livolant F, Sikorav JL. 1996. DNA
aggregation induced by polyamines and cobalthex-
amine. J Biol Chem 271(10):56565662.
6. Vijayanathan V, Thomas T, Shirahata A, Thomas
TJ. 2001. DNAcondensation by polyamines: Alaser
light scattering study of structural effects. Bio-
chemistry 40(45):1364413651.
7. Boussif O, Lezoualch F, Zanta MA, Mergny MD,
Scherman D, Demeneix B, Behr JP. 1995. A
versatile vector for gene and oligonucleotide trans-
fer into cells in culture and in vivo: Polyethyleni-
mine. Proc Natl Acad Sci USA 92(16):72977301.
8. Boussif O, Zanta MA, Behr JP. 1996. Optimized
galenics improve in vitro gene transfer with
cationic molecules up to 1000-fold. Gene Ther
3(12):10741080.
9. Kircheis R, Kichler A, Wallner G, Kursa M, Ogris M,
Felzmann T, Buchberger M, Wagner E. 1997.
Coupling of cell-binding ligands to polyethylenimine
for targeted gene delivery. Gene Ther 4(5):409418.
10. Abdallah B, Hassan A, Benoist C, Goula D, Behr
JP, Demeneix BA. 1996. A powerful nonviral vector
for in vivo gene transfer into the adult mammalian
brain: Polyethylenimine. Hum Gene Ther 7(16):
19471954.
11. Goula D, Benoist C, Mantero S, Merlo G, Levi G,
Demeneix BA. 1998. Polyethylenimine-based intra-
venous delivery of transgenes to mouse lung. Gene
Ther 5(9):12911295.
12. Goula D, Remy JS, Erbacher P, Wasowicz M, Levi
G, Abdallah B, Demeneix BA. 1998. Size, diffusi-
bility, and transfection performance of linear PEI/
DNA complexes in the mouse central nervous
system. Gene Ther 5(5):712717.
13. Coll JL, Chollet P, Brambilla E, Desplanques D,
Behr JP, Favrot M. 1999. In vivo delivery to tumors
of DNA complexed with linear polyethylenimine.
Hum Gene Ther 10(10):16591666.
14. Boletta A, Benigni A, Lutz J, Remuzzi G, Soria MR,
Monaco L. 1997. Nonviral gene delivery to the rat
kidney with polyethylenimine. Hum Gene Ther
8(10):12431251.
15. Ferrari S, Pettenazzo A, Garbati N, Zacchello F,
Behr JP, Scarpa M. 1999. Polyethylenimine shows
properties of interest for cystic brosis gene
therapy. Biochim Biophys Acta 1447(23):219
225.
16. Gautam A, Densmore CL, Waldrep JC. 2000.
Inhibition of experimental lung metastasis by
aerosol delivery of PEI-p53 complexes. Mol Ther
2(4):318323.
17. Lemkine GF, Demeneix BA. 2001. Polyethyleni-
mines for in vivo gene delivery. Curr Opin Mol Ther
3(2):178182.
18. Alex S, Dupuis P. 1989. FT-IR and Raman
investigation of cadmium binding by DNA. Inorga-
nica Chimica Acta 157:271281.
19. Choosakoonkriang S, Wiethoff CM, Anchordoquy
TJ, Koe GS, Smith JG, Middaugh CR. 2001.
Infrared spectroscopic characterization of the inter-
action of cationic lipids with plasmid DNA. J Biol
Chem 276(11):80378043.
20. Choosakoonkriang S, Wiethoff CM, Kueltzo LA,
Middaugh CR. 2001. Characterization of synthetic
gene delivery vectors by infrared spectroscopy. In:
Findeis M, editor. Methods in molecular medicine.
Totowa, NJ: Humana Press Inc., pp 285317.
21. Taillandier E, Liquier J. 1992. Infrared spectro-
scopy of DNA. Methods Enzymol 211:307335.
22. Premilat S, Albiser G. 1986. DNA models for A, B,
C, and D conformations related to ber X-ray,
infrared, and NMR measurements. J Biomol Struct
Dyn 3(5):10331043.
23. Choosakoonkriang S, Wiethoff CM, Anchordoquy
TJ, Koe GS, Smith JG, Middaugh CR. 2003. An
infrared spectroscopic study of the effect of hydra-
tion on cationic lipid/DNA complexes. J Pharm Sci
92:115130.
PEI/DNA COMPLEXES 1721
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003
24. Braun CS, Jas GS, Choosakoonkriang S, Koe GS,
Smith JG, Middaugh CR. 2003. The structure of
DNA within cationic lipid/DNA complexes. Biophys
J 84(2):11141123.
25. Akao T, Fukumoto T, Ihara H, Ito A. 1996.
Conformational change in DNA induced by cationic
bilayer membranes. FEBS Lett 391(12):215218.
26. Gray DM, Ratliff RL, Vaughan MR. 1992. Circular
dichroism spectroscopy of DNA. Methods Enzymol
211:389406.
27. Simberg D, Danino D, Talmon Y, Minsky A, Ferrari
ME, Wheeler CJ, Barenholz Y. 2001. Phase
behavior, DNA ordering, and size instability of
cationic lipoplexes. Relevance to optimal transfec-
tion activity. J Biol Chem 276(50):4745347459.
28. Zuidam NJ, Barenholz Y, Minsky A. 1999. Chiral
DNA packaging in DNA-cationic liposome assem-
blies. FEBS Lett 457(3):419422.
29. Lobo BA, Rogers SA, Choosakoonkriang S,
Smith JG, Koe G, Middaugh CR. 2002. Differential
scanning calorimetric studies of the thermal stabi-
lity of plasmid DNA complexed with cationic lipids
and polymers. J Pharm Sci 91(2):454466.
30. Baker BM, Murphy KP. 1996. Evaluation of linked
protonation effects in protein binding reactions
using isothermal titration calorimetry. Biophys J
71(4):20492055.
31. von Harpe A, Petersen H, Li Y, Kissel T. 2000.
Characterization of commercially available and
synthesized polyethylenimines for gene delivery.
J Control Release 69(2):309322.
32. Sukhorukov GB, Montrel MM, Petrov AI, Shabar-
china LI, Sukhorukov BI. 1996. Multilayer lms
containing immobilized nucleic acids: Their struc-
ture and possibilities in biosensor applications.
Biosens Bioelectron 11(9):913922.
33. Braun CS, Kueltzo LA, Middaugh CR. 2001.
Ultraviolet absorption and circular dichroism spec-
troscopy of nonviral gene delivery complexes. In:
Findeis M, editor. Methods in molecular medicine.
Totowa, NJ: Humana Press Inc., pp 253284.
34. Fish SR, Chen CY, Thomas GJ Jr, Hanlon S. 1983.
Conformational characteristics of deoxyribonucleic
acid-butylamine complexes with C-type circular
dichroismspectra. II. ARaman spectroscopic study.
Biochemistry 22(20):47514756.
35. Chen C, Ringquist S, Hanlon S. 1987. Covalent
attachment of an alkylamine prevents the B to Z
transition in poly(dG-dC). Biochemistry 26(25):
82138221.
36. Chen CY, Pheiffer BH, Zimmerman SB, Hanlon S.
1983. Conformational characteristics of deoxyribo-
nucleic acid-butylamine complexes with C-type
circular dichroism spectra. I. An X-ray ber
diffraction study. Biochemistry 22(20):47464751.
37. Chen C, Kilkuskie R, Hanlon S. 1981. Circular
dichroism spectral properties of covalent complexes
of deoxyribonucleic acid and n-butylamine. Bio-
chemistry 20(17):49874995.
38. Wiethoff CM, Smith JG, Koe GS, Middaugh CR.
2001. The potential role of proteoglycans in cationic
lipid-mediated gene delivery. Studies of the inter-
action of cationic lipid-DNA complexes with model
glycosaminoglycans. J Biol Chem 276(35):32806
32813.
39. Lobo BA, Smith JG, Koe G, Middaugh CR.
Thermodynamic analysis of binding and protona-
tion of DOTAP/DOPE (1:1) DNA complexes using
isothermal titration calorimetry. In press.
40. Godbey WT, Barry MA, Saggau P, Wu KK, Mikos
AG. 2000. Poly(ethylenimine)-mediated transfec-
tion: A new paradigm for gene delivery. J Biomed
Mater Res 51(3):321328.
41. Zanta MA, Boussif O, Adib A, Behr JP. 1997.
In vitro gene delivery to hepatocytes with galacto-
sylated polyethylenimine. Bioconjug Chem 8(6):
839844.
42. Gebhart CL, Kabanov AV. 2001. Evaluation of
polyplexes as gene transfer agents. J Control
Release 73(23):401416.
43. Fischer D, Bieber T, Li Y, Elsasser HP, Kissel T.
1999. A novel non-viral vector for DNA delivery
based on low molecular weight, branched polyethy-
lenimine: Effect of molecular weight on transfec-
tion efciency and cytotoxicity. Pharm Res 16(8):
12731279.
44. Godbey WT, Wu KK, Mikos AG. 1999. Size matters:
Molecular weight affects the efciency of poly-
(ethylenimine) as a gene delivery vehicle. J Biomed
Mater Res 45(3):268275.
45. Baker A, Saltik M, Lehrmann H, Killisch I,
Mautner V, Lamm G, Christofori G, Cotten M.
1997. Polyethylenimine (PEI) is a simple, inexpen-
sive and effective reagent for condensing and
linking plasmid DNA to adenovirus for gene
delivery. Gene Ther 4(8):773782.
46. Clamme JP, Azoulay J, Mely Y. 2003. Monitoring
of the formation and dissociation of polyethyleni-
mine/DNA complexes by two photon uorescence
correlation spectroscopy. Biophys J 84(3):1960
1968.
47. Xu Y, Hui S, Frederik P, Szoka FJ. 1999. Physico-
chemical characterization and purication of catio-
nic lipoplexes. Biophys J 77(1):341353.
48. Smith J, Wedeking T, Vernachio J, Way H, Niven
R. 1998. Characterization and in vivo testing
of a heterogeneous cationic lipid-DNA formulation.
Pharm Res 15(9):13561363.
49. Cooper A, Johnson CM. 1994. Introduction to
microcalorimetry and biomolecular energetics. In:
Jones C, Mulloy B, Thomas AH, editors. Micro-
scopy, optical spectroscopy, and macroscopic tech-
niques, Vol. 22. Totowa, NJ: Humana Press Inc.,
pp 109124.
1722 CHOOSAKOONKRIANG ET AL.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 8, AUGUST 2003