Efficient gene delivery of primary human cells using peptide linked polyethylenimine polymer hybrid.

Abstract

Polyethylenimine (PEI) based polymers are efficient agents for cell transfection. However, their use has been hampered due to high cell death associated with transfection thereby resulting in low efficiency of gene delivery within the cells. To circumvent the problem of cellular toxicity, metal binding peptides were linked to PEI. Eight peptide-PEI derivatives were synthesized to improve cell survival and transfection efficiency. TAT linked PEI was used as a control polymer. Peptides linked with PEI amines formed nanogels as shown by electron microscopy and atomic force microscopic measurements. Polymers were characterized by spectroscopic methods and their ability to form complexes with plasmids was tested using electrophoretic studies. These modifications improved polymer biocompatibility as well as cell survival markedly, when compared to PEI alone. A subset of the modified peptide-polymers also showed significantly higher transfection efficiency in primary human cells with respect to the widely used transfection agent, lipofectamine. Study of the underlying mechanism of the observed phenomena revealed lower levels of 'reactive oxygen species' (ROS) in the presence of the peptide-polymers when compared to PEI alone. This was further corroborated with global gene expression analysis which showed upregulation of multiple genes and pathways involved in regulating intracellular oxidative stress.

Full text

Page 1

Efficient gene delivery of primary human cells using peptide linked
polyethylenimine polymer hybrid
Devaveena Deya,b,c, Mohammed Inayathullaha,b,c,d, Andrew S. Leea,b,c, Melburne C. LeMieuxd,
Xuexiang Zhangd, Yi Wua, Divya Naga,b,c, Patricia Eliza De Almeidaa,b,c, Leng Hana,b,c,
Jayakumar Rajadasd,**, Joseph C. Wua,b,c,*
aDepartment of Medicine, Stanford University School of Medicine, Stanford, CA, USA
bDepartment of Radiology, Stanford University School of Medicine, Stanford, CA, USA
cInstitute for Stem Cell Biology and Regenerative Medicine, Stanford University School of Medicine, Stanford, CA, USA
dBiomaterials and Advanced Drug Delivery Laboratory, Stanford University School of Medicine, Stanford, CA, USA
a r t i c l e i n f o
Article history:
Received 19 February 2011
Accepted 6 March 2011
Available online 8 April 2011
Keywords:
Polyethylenimine
DNA vectors
Primary human cells
Plasmids
Transfection agent
Stem cells
a b s t r a c t
Polyethylenimine (PEI) based polymers are efficient agents for cell transfection. However, their use has
been hampered due to high cell death associated with transfection thereby resulting in low efficiency of
gene delivery within the cells. To circumvent the problem of cellular toxicity, metal binding peptides were
linked to PEI. Eight peptide-PEI derivatives were synthesized to improve cell survival and transfection
efficiency. TAT linked PEI was used as a control polymer. Peptides linked with PEI amines formed nanogels
as shown by electron microscopy and atomic force microscopic measurements. Polymers were charac-
terized by spectroscopic methods and their ability to form complexes with plasmids was tested using
electrophoretic studies. These modifications improved polymer biocompatibility as well as cell survival
markedly, when compared to PEI alone. A subset of the modified peptide-polymers also showed signifi-
cantly higher transfection efficiency in primary human cells with respect to the widely used transfection
agent, lipofectamine. Study of the underlying mechanism of the observed phenomena revealed lower
levels of ‘reactive oxygen species’ (ROS) in the presence of the peptide-polymers when compared to PEI
alone. This was further corroborated with global gene expression analysis which showed upregulation of
multiple genes and pathways involved in regulating intracellular oxidative stress.
? 2011 Elsevier Ltd. All rights reserved.
1. Introduction
Stem cells are the precursors for embryonic development as
well as adult tissue regeneration. As such, stem cell therapy holds
great promise for the field of regenerative medicine. Gene therapy
involves manipulation of an entire or portion of a gene in a target
cell population in order to correct a disease condition [1]. In the
past decade, a number of studies have shown that genetic modifi-
cation of stem cells can improve their therapeutic potential in vivo
as well as allow for their monitoring via introduction of reporter
genes [2,3]. The choice of vector for targeted gene therapy into stem
cells, however, has been a topic of debate.
In order to attain optimum expression of the protein from the
vector carrying the gene of interest, multiple hurdles must be
overcome. These include (1) protection of the vector from degra-
dation before entry into the cell; (2) efficient entry of the vector
into the cell; (3) protection/prevention from nuclease degradation
within endosomes; and (4) efficient entry into the cell nucleus [4].
Viral vectors have shown high efficiency of gene delivery, over-
coming most of the hurdles listed. However, use of viral vectors has
multiple disadvantages including immunogenic responses, inser-
tional mutagenesis, and risk of tumorigenicity [5]. Hence forclinical
usage, non-viral gene therapy is the preferred method of choice.
Non-viral gene delivery methods can be broadly categorized
into utilizing polycationic polymers, liposomes, peptides, proteins,
and organic/inorganic nanoparticles [4]. Positively charged poly-
cationic polymers can efficiently complex with negatively charged
DNA thereby increasing vector stability. Use of liposomes or
micelles encapsulating the DNA enhances binding and fusion of the
* Corresponding author. Stanford University School of Medicine, 300 Pasteur
Drive, Grant S140, Stanford, CA 94305-5111, USA. Tel.: þ1650 724 1581; fax: þ1 650
736 2432.
** Corresponding author. Stanford University School of Medicine, 300 Pasteur
Drive, Grant S380, Stanford, CA 94305, USA.
E-mail addresses:
jayraja@stanford.edu
(J.C. Wu).
(J.Rajadas),joewu@stanford.edu
Contents lists available at ScienceDirect
Biomaterials
journal homepage: www.elsevier.com/locate/biomaterials
0142-9612/$ e see front matter ? 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biomaterials.2011.03.016
Biomaterials 32 (2011) 4647e4658

Page 2

DNA-lipid complex to the cell surface, thereby enhancing vector
entry into the cell. However, despite intensive research being
conducted on non-viral gene delivery methods and existence of
more than 50 commercially available kits for transfection, the
majority of these methods suffer from two major drawbacks: (1)
cell toxicity and (2) low transfection efficiency.
Polyethyeleneimine (PEI) is a promising candidate among poly-
cationic polymers used for transfection [6,7]. This is due to its
relatively high and efficient capacity to complex with DNA. The
ability of PEI to complex with DNA so well is attributed to its large
numberof protonable amino nitrogen atoms, which results in a high
cationic charge densityat physiological pH. This structure makes the
polymer an effective ‘proton pump’ under virtually any pH, enabling
osmotic swelling and rupture of endosomes and enhancing the
release of DNA from the endosomal complex within cells.Due to this
path, endosomes containing PEI avoid trafficking to degradative
lysosomes. In addition, PEI complexes with DNA and RNA in nano-
meter range nanosize complexes, thereby enhancing the delivery of
these nucleic acids into cells. Despite these encouraging results, PEI
continues to be plagued by persistent problems such as low in vitro
and in vivo transfection efficiency and strong cytotoxicity [8].
In this work, we have designed histidine-based peptide linked
PEI polymers for transfection studies [9]. The pH buffering effects
were achieved by addition of the co-polymers composed of histi-
dine and lysine, which are also highly efficient carriers of plasmids.
Peptide-based polymers are more advantageous than PEI-based
polymers because they are easily metabolized and non-toxic to
cells, making different combinations possible. In order to test the
efficacy of these polymers as transfection agents, multiple cell lines
and primary human cells were used. The primary human cell types
used in this study were adipose stromal cells (ASCs), dermal
fibroblasts, and cardiac progenitor cells (CPCs), all of which are
promising candidatesforautologous
Because of their human origin, these three cell types have high
clinical relevance. ASCs and fibroblasts in particular are plentiful
and easy to obtain from patients. Both these cells types have
recently been targeted for deliveryof pluripotent transgenes for the
generation of induced pluripotent stem cells [10,11]. Cardiac
progenitor cells, on the other hand, comprise of an adult stem cell
population that can give rise to all the cell types that recapitulate
the heart [12] and therefore have a strong potential to be used for
cardiac repair and regeneration following myocardial infarction. In
the present study, we have tested our hypothesis that PEI-based
polymer hybrids can transfect primary cells at higher efficiencies
than other existing non-viral methods. We have also examined ROS
production, microarray based gene expression profiling and
mechanistic properties that might be associated with cell survival
and the transfection efficiency of these modified polymers. We
report and discuss the results of these studies here, which we
believe to yield significant advancements to gene delivery and in
vivo stem cell therapy for clinical applications.
tissuetransplantation.
2. Materials and Methods
2.1. Chemicals
Polyethylenimine 25 kDa (PEI) was purchased from SigmaeAldrich (St Louis,
MO), DNase I from Qiagen (Valencia, CA), and fetal calf serum (FCS) from Invitrogen
(Carlsbad, CA). All other chemicals were of reagent grade.
2.2. DNA vectors used
Two vectors were used for transfection studies. The first was a 12 kb vector
containing firefly luciferase (Fluc) gene driven by CMV promoter. Transfection of this
vector into cells was monitored by luciferase assay. The second vector was a 3.5 kb
pMAX-GFP vector expressing green fluorescent protein under the CMV promoter.
Transfectionwith this vector into cells was monitored by flowcytometry. Both of the
vectors were amplified in the DH5a strain of E. coli and purified by using a Qiagen
Maxi kit as per the manufacturer’s instructions.
2.3. Polymer synthesis
The polymers were synthesized in theBioADD facilityat StanfordUniversity. The
sequences of polymers were as follows: (i) PEI-Arg; (ii) PEI-His; (iii) GHK-PEI; (iv) (t-
Boc)2-GHK-PEI; (v) L-Carnosine-PEI; (vi) t-Boc-Carnosine-PEI; (vii) b-Ala-His-GHK-
PEI, which is a combined peptide of L-Carnosine and GHK; and (viii) t-Boc-TAT-PEI,
whereby TAT (GRKKRRQRRRPQK) is a cell permeable peptide. Details of polymer/
peptide synthesis, purification and characterization are available in Supplementary
Methods.
2.4. Transmission electron microscopy (TEM)
Peptide-PEI and peptide-PEI/DNA samples were adsorbed onto a Formvar-
coated, carbon-stabilized copper grid (400 mesh). The grid was then rinsed briefly
with water, negatively stained with 2% aqueous uranyl acetate, air-dried, and
examined with a JEOL JEM 1230 transmission electron microscope at an accelerating
voltage of 80 kV.
2.5. DNase protection assay
The integrityof theplasmid DNA releasedfrom the polymer/DNAcomplexes after
exposure to DNase I was assessed by agarose gel electrophoresis as previously
described[13].Briefly,thecomplexwaspreparedbyadding2mlDNA(500ng/ml)to6ml
of HBS (HEPES 10 mM and NaCl 150 mM; pH 7.3) followed by the addition of 50 ml
polymer and incubated at room temperature for 20 min. Then 6 ml of a solution con-
taining MgCl2(35 mM) and MnCl2(35 mM) was added to the complex. Naked DNA
samples were also prepared by replacing polymer with the same volume of water.
Afterwards, 20 ml of complex or naked DNA sample was incubated with 2.2 ml DNase I
(2.5U/ml)orwaterat37?Cfor20min.Degradationwasstopped byadding2.1ml EDTA
(250mM)andplacingthesamplesimmediatelyinanicebath.ToreleaseDNAfromthe
complex for agarose gel electrophoresis, 2.2 ml heparin (200 mg/ml) was added and
incubatedfor10min.Postincubation,30mlofthecomplexornakedDNAwasloadedin
a1.2%agarose gelpreparedin1XTAEbuffer(pH8).EthidiumBromide (0.5mg/ml)was
used for detection of DNA in the gel. The gel was run at 5 V/cm.
2.6. Cell culture conditions
The transformed human embryonic kidney cell line HEK 293FT (Invitrogen) and
the immortal cervical cancer cell line HeLa (American Type Culture Collection) were
used for the standardization of optimum polymer concentration. The primary
human cells used were adipose stromal cells (ASCs), human skin fibroblasts, and
cardiac progenitor cells (CPCs) as described in Supplemental Methods. HEK 293FT
cells, ASCs, fibroblasts, and HeLa were cultured in DMEM (Sigma Aldrich) supple-
mented with 10% fetal bovine serum (Invitrogen) and antibiotics (Invitrogen) under
standard tissue culture conditions of 37oC and 5% CO2. All the studies were carried
out in 24-well plates (Greiner Bio-One). Human CPCs were cultured in a specialized
media, EGM-2, which was prepared as per a previous report [14], and supplemented
with serum (Gibco) and defined factors (Lonza Biosciences). In short, the cells were
thawed, resuspended in the specialized media and seeded on a 24-well plate coated
with 0.1% gelatin. Media was changed every 3e4 days. Transfections were carried
out 24 h post seeding.
2.7. Transfection assays
Equal numbers of cells were seeded per well in 24-well plates 24 h prior to
transfection. Briefly, for each well, 0.8 mg of DNA encoding for firefly luciferase was
added to 50 ml of DMEM þ 10% FBS in a 1.5 ml tube. The calculated volume of
polymer stock (resulting in 6 mg, 20 mg, 50 mg, or 100 mg final quantity of polymer)
was added to the DNA-media mix. The cocktail was incubated at room temperature
for 8 min after which 450 ml of DMEM þ 10% FBS was added. The media present in
the well was removed and the DNA-polymer cocktail was added drop wise to the
cells. The plate was incubated for 2 h at 37oC and 5% CO2. After 2 h, the cells were
washed with PBS followed by addition of fresh media (DMEM þ 10% FBS). Analyses
of cell survival and transfection efficiency were done at 24 and 48 h post trans-
fection. The control lipofectamine transfection was carried out exactly as per the
manufacturer’s protocol (Invitrogen). Cell survival was studied using trypan blue
dye exclusion analysis. For 6-well and 96-well plates, the quantity of DNA and
polymers were adjusted accordingly.
2.8. Bioluminescence studies
Bioluminescence imaging of the 24-well plates containing HEK 293FT cells
transfected with Fluc reporter gene was carried out on the Xenogen IVIS System
(Alameda, CA) with 30 s of exposure and medium binning. The signal was quantified
using the ROI analysis tools of Living Image software. Analysis of luciferase activity
was done 48 h post transfection using the Promega Luciferase Assay kit (E1910) as
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4648

Page 3

per the instructions of the manufacturer. Briefly, cells were washed in PBS 3 times,
followed bycell lysis using the ‘PassiveLysisBuffer’ (Promega).10mlof the lysatewas
used to measure firefly luciferase activity. The same lysate was used to estimate the
total protein in each sample using the standard BCA protein assay. Bovine serum
albumin (BSA) was used for the protein standard plot. Luciferase activity of each
sample was normalized to the total protein content per sample. This normalized
value was plotted in Excel.
2.9. Fluorescence measurements
Transfection with pMAX-GFP was done in a 6-well tissue culture plate, and
immunoflorescence measurement was done using flow cytometry. 48 h post
transfection, cells in the 6-well plate were dissociated with trypsin, washed with
PBS, and the pellet was resuspended in 300 ml of PBS. GFP signal intensity was
measured on the FACSVantage (BD Biosciences) on FL1 channel (Excitation: 488 nm;
N
N
H
NH 2
PEI (Branched, MW 25,000)
HN
O
N
H
NN
N
N
H
NH 2
N
N
H
NH
x
y
O
O CH 3
H 3 C CH 3
O
N
N
H
NH 2
N
N
H
NH
x
y
NN
H
O
N N
H
O
OOCH3
CH3
H 3 C CH 3
O
HN
N
O
NH
O
O
H 3 C
CH 3
H 3 C
BGP
N
NH
BCP
N
NN
H
N
N N
H
NNNN
HN
O
NH 2
O
N
NH 2
NH
x
y
O
N
H
NH 2
N
H
NH
x
y
N
H
O
NH 2
N N
HN
N
O
H 2 N
GP
NH
LCP
N
NN
H
N
N
NN
H
N
NN
H
HN
O
NH 2
O
HN
N
N
O
HN HN
NH 2
N
H
NH
x
y
NH 2
NH
x
y
GRKKRRQRRRPQK
t-Boc
t-Boc
t-Boc
t-Boc
N
H
O
NH 2
N
NH
O
B AHGP
TP
Polyethylenimine PEIBoc-L-Carnosine-PEI BCPL-Carnosine-PEI LCP
Boc-2-GHK-PEI BGPTAT-PEI TP β -Ala-His-GHK-PEI BAHGPGHK-PEI GP
Fig.1. Schematic diagram of the structures of the basic polymer PEI (top) and the peptide-PEI based polymers described in the current study. The peptide groups were linked to the
primary amine group of PEI.
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4649

Page 4

Emission: 500e520 nm). 20,000 and 50,000 events were recorded for analysis. Data
analysis was performed on FlowJo software (BD Biosciences) and the ‘Mean Flour-
escence Intensity’ (MFI) value as obtained from the FL1 histograms was plotted on
excel.
2.10. Measurement of ROS levels
ROS levels within cells were measured 24 h post transfection with the hybrid
polymers, using the fluorescence indicator dye, carboxy-DCFDA (Invitrogen). Cells
were incubated with 10 mM DCFDA at 37oC for 30 min. This was followed by two
washes with phosphate buffered saline. ROS level was measured in a fluorescence
plate reader by recording the emission at 510 nm.
2.11. Microarray hybridization and data analysis
Total RNA samples isolated using Qiagen mRNAeasy kit were hybridized to
Affymetrix GeneChip Human Gene 1.0 STArray, and then normalized and annotated
by the Affymetrix? Expression Console? software. The Pearson Correlation Coef-
ficient was calculated for each pair of samples using the expression level of the top
5000 genes showing greatest variance. For hierarchical clustering, Pearson corre-
lation for average linkage clustering was used. Genes with expression fold change
greater than 2 were selected, and the significant pathways analysis was done by
Ingenuity Pathway Analysis software (Ingenuity Systems; www.ingenuity.com). The
enriched pathways were selected based on the p-value cutoff of <0.05.
2.12. Statistical analysis
For the firefly luciferase and GFP assays post transfection and cell survival assay,
paired one-tailed t-test was used to compare the statistical significance of the
differences between two transfection groups at a time. Graph Pad Prism software
was used for analysis. * represents p values < 0.05; ** (p < 0.005); *** (p < 0.0005).
3. Results
3.1. Characterization of the primary structure of peptide-PEI
polymers
To synthesize peptide-PEI hybrid polymers, peptide moieties
were linked with the primary amine groups of PEI (Fig. 1). To
confirm the incorporation of the peptides in the PEI backbone,
the tertiary butyl groups of the peptides were monitored using
1H-NMR spectra of the polymers (Suppl Fig. 1). Protons for Boc
group are generally observed as a singlet between 1.3 and
1.5 ppm region and the PEI backbone is observed as multiplet
between 2.3 and 2.8 ppm. From these results, the number of
peptide chains grafted onto a PEI backbone was calculated based
on the relative integration values between the single CH3- proton
peak in Boc (s, w1.35 ppm) and the multiple-CH2eproton peaks
in PEI (m, w2.49 ppm). According to the1H NMR results, the
average number of conjugated peptide was 0.1 parts (w10%) per
PEI backbone.
3.2. Structure and function characterization of the peptide-PEI-DNA
complex
Transmission electron microscopy (TEM) was used to reveal the
microstructure of the peptide-PEI hybrid polymers, and the effect
of DNA conjugation on this microstructure. TEM micro graphs
indicate that the PEI peptide preparations contain distinct, sepa-
rated aggregates with a mean size (diameter) of roughly 10e20 nm
and PEI peptide e DNA complex nanogels size range varied from 10
to 80 nm, out of which w5e15% of the nanogels appeared to be
40 nm or more (Fig. 2). In addition, AFM imaging indicated that
peptide-PEI samples form smaller aggregates compared to DNA-
conjugated samples (Supp Fig. 2). However, fibril formation was
also seen in these samples, which can be clearly visualized in the
phase images (Supp Fig. 2A and C; right panel). This fibril formation
was more prevalent in Carnosine-PEI compared to the Boc2-
GHK-PEI (Supp Fig. 2A and C). For the DNA-conjugated samples, the
microstructure of the particles was less spherical and more plate-
like (Supp Fig. 2B and D). Moreover, when compared to samples
without DNA, less particle aggregation was observed in the DNA
tethered samples, with absolutely no fibril formation. Such
dispersion effect on polymers due to DNA binding has been
observed earlier [15].
Fig. 2. Transmission electron microscopy (TEM) images of polymer alone (left panel)
and polymer-DNA complex (right panel) for all the polymers. Scale bar represents
500 nm.
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4650

Page 5

We next studied the functional significance of the peptide-
polymer hybrid binding to DNA by a DNase protection assay. In our
pilot study, the DNA samples mixed with polymers could not be
detected by ethidium bromide staining followed by agarose gel
electrophoresis, whereas naked DNA could be visualized (data not
shown). This was seen in the absence of heparin. However, addition
of heparin at the final concentration of 1.7% (w/v) to the polymer-
DNA samples effectively released the DNA from the polymers Boc2-
GHK-PEI, Boc-L-Carnosine-PEI, and PEI, and a comparable intensity
in electrophoretic bands was observed when compared to naked
DNA (Fig. 3A). In order to evaluate the potential protective effects of
the polymers, naked DNA and DNA-polymer complexes were
incubated with DNase I for 20 min. This treatment degraded naked
DNA almost completely, as indicated by the lack of band in the
agarose gel (Fig. 3B, lane 1). In contrast, complexing DNA with PEI
served as ‘protection’ from DNase digestion (Fig. 3B, lane 7). Similar
protection was observed in the complexes formed by TAT-PEI and
carnosine-PEI. Other polymers provided relatively less potent but
discernable protection (Fig. 3B).
3.3. Determination of optimum concentration for activity of
polymers
To test the efficacy of the polymers for transfection, synthesized
DNA complexes were first used to transfect the human embryonic
kidney cell line (HEK 293FT), a standard cell type regularly used for
transfection studies. A plasmid DNA containing the firefly luciferase
reporter gene was used. Transfection of cells with the plas-
midepolymercomplex was carried out as described in the Materials
and Methods section. In order to determine the optimum concen-
tration to be used for the polymers, four different concentrations
were tested for all polymers using a constant quantity of plasmid
DNA (0.8 mg). A commercially available transfection reagent, Lip-
ofectamine,wasusedasapositivecontrol.Bioluminescenceimaging
was used to assess transfection efficiency. A significant number of
polymers showed positive bioluminescence signal at one or more
concentrations (Fig. 4A). PEI-Histidine and PEI-Arginine failed to
show positive signals at any concentrations.
Cell viability was used as a second parameter to determine the
optimum concentration to be used for each polymer. Trypan blue
based analysis of cell death indicated that all the modified hybrid
polymers demonstrated high cell survival when compared to the
parent polymer, PEI (Fig. 4B). The cell survival in presence of the
modified polymers was similar to the positive control. This study
also indicated that cell survival was minimal at 100 mg (the highest
concentration) for almost all polymers (Fig. 4B). Based on the
bioluminescence and cell survival data, six polymers were selected
(GHK-PEI,Boc2-GHK-PEI,L-Carnosine-PEI,
b-Ala-His-GHK-PEI, TAT-PEI) at the concentration of minimum cell
death and maximum transfection efficiency. Based on these exper-
iments, the concentrations of polymers to be used for 0.8 mg of DNA
which was finalized for all further experiments are as follows: GHK-
PEI (20 mg); Boc2-GHK-PEI (6 mg); L-Carnosine-PEI (50 mg); Boc-L-
Carnosine-PEI (6 mg); b-Ala-His-PEI (6 mg), and TAT-PEI (50 mg).
Boc-L-Carnosine-PEI,
3.4. Transfection of primary human adipose stromal cells (ASCs)
using screened polymers
The selected polymers were used at their optimum concentra-
tions to transfect primary ASCs. Primary cells have high clinical
relevance for both cell and gene therapy. However, these cells are
highly resistant to transfection when compared to transformed cell
lines such as HEK 293FT [16]. An additional problem with primary
cells is high cell death associated with transfection. The same
plasmid used to transfect the HEK 293FTcells was used to transfect
the ASCs. Lipofectamine was used as the positive control. 24 h post
transfectionwith the polymers, cell morphologyand anyassociated
cell death were studied under a phase contrast microscope (Supp
Fig. 3A). Luciferase assay of the cells conducted 24 h post trans-
fection demonstrated higher transfection efficiency in cells incu-
bated with the polymers as compared with lipofectamine (Fig. 5A).
Trypan blue analysis of cell survival was carried out at 48 h post
Fig. 3. Polymer protection of DNA from degradation. (A) Addition of heparin in the samples effectively released DNA from polymers thus enabling DNA to be visualized by Ethidium
Bromide staining. (B) In the presence of DNase I, all polymers showed some degree of protection of DNA from degradation, while the naked DNA was completely digested (lane 1).
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4651

Page 6

transfection. This demonstrated that in presence of five polymers,
namely GHK-PEI, Boc2-GHK-PEI, L-Carnosine-PEI, Boc-L-Carnosine-
PEI and b-Ala-His-GHK-PEI, cell survival was comparable to the
positive control (lipofectamine), and there was no significant
difference in cell survival among the different groups except TAT-
PEI (Fig. 5B).
3.5. Transfection of human dermal fibroblasts with the screened
polymers
In order to determine the efficiency of the polymers in a wide
range of cell types, we next tested the same set of six polymers
upon primary human dermal fibroblasts isolated from patients as
described in the Supplemental Methods section. Photomicrographs
of the transfected cells 24 h post transfection is shown in Supp
Fig. 3B. Similar to ASCs, the polymers showed significantly higher
transfection efficiency in fibroblasts when compared to lipofect-
amine as assessed by luciferase assay of the cells 24 h post trans-
fection (Fig. 6A). Cell survival in presence of the polymers was also
similar to that of the positive control lipofectamine (Fig. 6B), except
for L-Carnosine-PEI, which demonstrated lower cell survival when
compared to the other polymers for this specific cell type.
3.6. Transfection of human cardiac progenitor cells (CPCs) with
screened polymers
Human CPCs were used as a third human cell type to validate the
efficiency of the PEI-derived polymers. Human CPCs give rise to all
the functional cell types of the heart and therefore have very high
clinical relevance. One of the paramount goals in clinical cardiology
is to successfully perform gene therapy of cardiac cells in order to
correct genetic mutations that underlie multiple intractable cardiac
Fig. 4. Determination of the optimum concentration of the polymers to be used for transfection. (A) Transformed human embryonic kidney (HEK 293FT) cells were transfected with
a luciferase-encoding vector in the presence of four different concentrations of the polymers such that luminescence intensity was used as a readout of transfection. Lipofectamine
was used as the positive control. (B) Cell death was assessed for each concentration for all the polymers using trypan blue assay and cell survival was plotted as a percentage. Each
group was compared to the blank to analyze the statistical differences observed in the relative luminescence and/or cell survival. Error bars represent triplicates with standard
deviation (*p < 0.05; **p < 0.005; ***p < 0.0005).
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4652

Page 7

diseases such as dilated and hypertrophic cardiomyopathies [17].
Polymer-based transfections using the firefly luciferase reporter
gene demonstrated significantly higher transfection efficiency in
these cells when compared to positive control (lipofectamine) as
assessed by a quantitative luciferase assay (Fig. 7A). Importantly, cell
survival in presence of the polymers was similar to lipofectamine
(Fig. 7B).
3.7. Transfection of primary human ASCs and CPCs with GFP-
expressing vector using L-Carnosine and Boc-L-Carnosine
Based upon the transfection efficiencies of all 6 polymers tested
for human ASCs, dermal fibroblasts, and CPCs, the two polymers
withthe highestoverall transfection efficiencieswere determined to
beL-Carnosine-PEIand its Boc derivative.To confirmthepotential of
these two polymers to transfect primary stem cells at high effi-
ciencies, we applied L-Carnosine-PEI (50 mg for 0.8 mg DNA) and its
Boc derivative (6 mg for 0.8 mg DNA) upon human ASCs and CPCs
usinga different vectorencodingfor GFPonly. Flowcytometrybased
fluorescence intensity measurements were used as a readout of
transfection efficiency using this vector. Fluorescence intensity
values of the samples confirmed the high transfection efficiencies
observed in the luciferase assay measurements conducted previ-
ously (Supp Fig. 4B and E). In both cell types, GFP signal was
significantly higher for L-Carnosine-PEI and Boc-L-Carnosine-PEI
than for the commercially available agent lipofectamine. These
results validate our earlier observation that the modified-PEI poly-
mers can serve as highly efficient gene delivery agents for primary
human cell types, which are otherwise difficult to transfect. Finally,
we compared transfection efficiency of the GFP reporter gene in the
HeLa cell line using L-Carnosine-PEI, Boc-L-Carnosine-PEI, and lip-
ofectamine. Similarly, we found that L-Carnosine-PEI and Boc-L-
Carnosine-PEI were characterized by higher transfection efficiencies
as compared to lipofectamine (Supp Fig. 5B).
3.8. Analysis of oxidative stress in cells transfected with the
modified polymers
One of the major problems with PEI-based transfections has
been the high cell toxicity resulting in cell death and low efficiency
Fig. 5. Testing the transfection efficiency of selected polymers at their optimum concentration on human adipose stromal cells (ASCs). (A) Transfection efficiency was assessed by
quantitative luciferase assay, with the signal intensity normalized to the total protein (*p < 0.05; **p < 0.005; ***p < 0.0005). (B) Cell death was assessed by trypan blue staining
and represented as % cell survival. Each polymer group was individually compared to the blank to analyze the statistical differences observed in the relative luminescence as well as
cell survival. Error bars represent triplicates with standard deviation.
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4653

Page 8

of gene delivery. Modification of PEI by peptide conjugation might
reduce the oxidative damage within the cells, thereby resulting in
higher cell survival and high transfection efficiency. In order to
address this, we compared the level of ‘reactive oxygen species’
(ROS) within cells transfected with PEI and the hybrid polymers.
DCFDA based fluorescence measurements indicated that ROS levels
were significantly lower in cells transfected with the hybrid poly-
mers as compared to PEI (Fig. 8A and B). This was confirmed
independently in two primary cell types, adipose stromal cells and
cardiac progenitor cells.
We next carried out microarray based global gene expression
profiling of cardiac progenitor cells transfected with L-Carnosine-
PEI and Boc-L-carnosine-PEI to study the genes and pathways
involved in regulating oxidative stress under these conditions
when compared tountreated CPCs (‘Blank’) and the positive control
(‘Lipofectamine’). A set of genes known to be involved in mito-
chondrial functions, glucose and fatty acid metabolism, and those
regulating oxidative stress were shortlisted and their expression
profile was analyzed by hierarchical clustering of the gene sets
(Fig. 8C). Genes encoding enzymes which are typically involved in
reducing oxidative stress like superoxide dismutase and metal-
lothioneins were found to be upregulated in cells transfected with
L-Carnosine-PEI and Boc-L-carnosine-PEI when compared to the
blank, shown in boxes in the cluster diagram (Fig. 8C). Some of the
interleukins, which are released by cells in response to external
stimuli, were also found to be upregulated. In addition, some of the
pathways activated to enhance survival under stress were found to
be upregulated in the two transfected groups as determined by
Ingenuity Pathway Analysis (Supplementary Tables 1 and 2).
4. Discussion
The present study focused on the development of a set of
transfection agents for efficient gene delivery into cells, which is
necessary for successful genetic modification of stem cells to take
place. The polymers described here have been derived by chemical
Fig. 6. Testing the transfection efficiency of selected polymers at their optimum concentration on human dermal fibroblasts. (A) Transfection efficiency was assessed by quantitative
luciferase assay, with the signal intensity normalized to the total protein (*p < 0.05; **p < 0.005; ***p < 0.0005). (B) Cell death was assessed by trypan blue staining and rep-
resented as % cell survival. Each polymer group was individually compared to the blank to analyze the statistical differences observed in the relative luminescence as well as cell
survival. Error bars represent triplicates with standard deviation.
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4654

Page 9

modification of the 25 kDa basic cationic molecule, poly-
ethylenimine (PEI), with distinct side groups, namely L-Carnosine,
GHK, Boc, b-Ala-His and TAT. Transfection of the human embryonic
kidney cell line (HEK 293FT) with the modified polymers demon-
strated high transfection efficiencyas well as enhanced cell survival
when compared to PEI. When further tested on three human
primarycell types (fibroblasts, ASCs, and CPCs), which areknownto
be highly resistant to transfection, the modified PEI polymers could
transfect these cells with high efficiency. These polymers include
GHK-PEI, Boc2-GHK-PEI,L-Carnosine-PEI,
b-Ala-His-GHK-PEI, and TAT-PEI. An additional advantage of using
these polymers is that all transfections are carried out in the same
medium in which cells are grown in the presence of serum and
antibiotics. This eliminates the need to change medium before,
during, or post transfection and therefore highly increases the ease
of transfection.
Transfection with the basic molecule, PEI, resulted in very high
cell death as assessed by trypan blue based cell viability assays
Boc-L-Carnosine-PEI,
(Fig. 4B) even at the lowest concentrations tested. These results
support findings by other studies that show high cytotoxicity is one
of the biggest drawbacks of PEI-based transfections [7]. In contrast,
most of the modified PEI compounds demonstrated very low cell
toxicity (Fig. 4B). For some of the polymers such as Boc-L-Carno-
sine-PEI, PEI, and PEI-Arg, higher levels of cell death was observed
with increasing concentration levels, whereas for others such as
GHK-PEI, L-Carnosine-PEI, and b-Ala-His-GHK-PEI, differences in
cell death at different concentrations were minimal. Overall, trypan
blue based cell viability analysis clearly indicated that chemical
modification of PEI significantly reduces cellular toxicity.
In our study, a subset of the eight polymers was chosen (in
addition to PEI as control) as especially promising for clinical
applications based on two parameters: (1) transfection efficiency
and (2) cell survival in the presence of these polymers. Based on
these two criteria, two of the polymers (PEI-Histidine and PEI-
Arginine) were excluded for further analysis as they demonstrated
negligible luciferaseactivity atall concentrations.The six
Fig. 7. Testing the transfection efficiency of selected polymers at their optimum concentration on human cardiac progenitor cells (CPCs). (A) Transfection efficiency was assessed by
quantitative luciferase assay, with the signal intensity normalized to the total protein (*p < 0.05; **p < 0.005; ***p < 0.0005). (B) Cell death was assessed by trypan blue staining
and represented as % cell survival. Each polymer group was individually compared to the blank to analyze the statistical differences observed in the relative luminescence as well as
cell survival. Error bars represent triplicates with standard deviation.
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4655

Page 10

remaining polymers demonstrated significantly strong luciferase
activity at one or more concentrations. The control PEI was
excluded due to its high cell toxicity even when used at low
concentrations (Fig. 4B). Overall, our findings are concordant to the
report of Hashemi et al. [18] who elegantly demonstrated the
effectiveness of KH peptides-linked PEI complexes as gene carriers
and the significant enhancement of gene delivery by these modi-
fied polymers when compared to PEI alone.
The ultimate goal of all gene and cell therapies is application in
a clinical setting. We therefore tested the efficiency of these poly-
mers to deliver plasmid DNA into primary human cells, which are
known to be significantly more difficult to transfect than immor-
talized cell lines. Lipofectamine, a widely used transfection agent
was used as the positive control. To this end, we selected three cell
types relevant to regenerative medicine and clinical research;
namely adipose stromal cells [19], fibroblasts [20], and cardiac
progenitorcells [21]. Because these cells are known to be difficult to
transfect, a sensitive luciferase assay with normalization to total
protein content was used as readout of transfection efficiency. Cell
death was measured by trypan blue staining. Similar to the results
seen in the HEK 293FT cells, the six shortlisted polymers (GHK-PEI,
Boc2-GHK-PEI, L-Carnosine-PEI, Boc-L-Carnosine-PEI, b-Ala-His-
GHK-PEI and TAT-PEI) demonstrated robust cell survival, which was
comparable to the positive control (lipofectamine). Morphologi-
cally, most of the hybrid polymer-transfected groups were similar
to the blank control (Supp Fig. 3). Exceptions were TAT-PEI and L-
Carnosine-PEI, where clustering and shrinkage of cells was seen.
One of the reasons for this could be the high concentration of the
two polymers that was used for transfection (50 mg for 0.8 mg DNA).
Analysis of luciferase activity showed significantly high trans-
fection efficiency in the hybrid polymer-transfected primary cells
as compared to lipofectamine. Different polymers demonstrated
selective propensities to transfect certain cell types. For example,
TAT-PEI showed higher transfection efficiency in ASCs compared to
fibroblasts, whereas b-Ala-His-GHK-PEI showed higher trans-
fection in fibroblasts compared to ASCs. This may reflect the
differences in cell surface receptors on different cell types (which is
exploited by the TAT-PEI system) or differences in intracellular
Fig. 8. Quantification of intracellular ROS levels by DCFDA staining 24 h post transfection with PEI and the modified, hybrid-PEI polymers. ROS level was assessed for (A) adipose
stromal cells (ASCs) and (B) cardiac progenitor cells (CPCs). Lipofectamine was used as the positive control. Boxed region indicates the comparison of ROS levels in PEI versus the
modified polymers. Error bars represent triplicates with standard deviation (*p < 0.05). (C) Hierarchical clustering of microarray based gene expression data of cardiac progenitor
cells transfected with L-Carnosine-PEI and Boc-L-Carnosine-PEI. Genes showing differential regulation with respect to the control (‘Blank’) have been highlighted in boxes.
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4656

Page 11

metabolism of chemical moieties (which might affect the release of
DNA from the b-Ala-His-GHK-PEI complex within different cells).
Comparison of the six polymers chosen for further analysis
revealed two polymers to be the most promising as demonstrated
by consistently high transfection efficiencies in all cell types:
L-Carnosine-PEI and Boc-L-Carnosine-PEI. L-Carnosine is a dipep-
tide of b-Alanine and histidine, and is an endogenous peptide
present in excitable tissues like the brain and muscles where it has
been shown to reduce endoplasmic reticulum (ER) stress [22].
L-carnosine acts as an antioxidant by scavenging hydroxyl radicals,
binding divalent metals, upregulating enzymes that metabolize
free radicals, and by promoting antiglycation [23,24]. As a result,
carnosine has been used as a protective agent in stroke models [25],
liver injuries [26], and ophthalmologic disorders [27].
We hypothesized that one of the major reasons of the toxicity
of PEI as a transfection agent could be due to PEI-mediated
oxidation of the Cu2þand Fe2þmoieties of intracellular enzymes,
which results in production of hydroxyl radicals, consequently
leading to enzymatic inactivation. Analysis of oxidative stress in
transfected cells by DCFDA staining demonstrated reduced levels
of ROS in the presence of hybrid polymers when compared to PEI
(Fig. 8A and B). Global gene expression profiling studies indicated
the upregulation of multiple enzymes involved in regulating ROS
and reducing oxidative stress, like superoxide dismutase and
metallothioneins (Fig. 8C). Downregulation of Acyl Co-A oxidase in
cells transfected with L-Carnosine-PEI and Boc-L-Carnosine-PEI
might also contribute to the lowering of oxidative stress since this
enzyme produces H2O2as one of the end products. In addition,
several pathways involved in mitochondrial functions and glucose
and fatty acid metabolism were also found to be activated
(Supplementary Tables 1 and 2). Activation of the PPAR pathway
especially indicated activity of peroxisomes, the cellular organelles
involved in scavenging ROS and reducing intracellular oxidative
stress. Taken together, these mechanistic studies suggest that L-
Carnosine-PEI and Boc-L-Carnosine-PEI are able to reduce oxida-
tive damage in cells when compared to PEI. They do so by upre-
gulationofmultipleenzymes
scavenging free radicals and ROS and downregulating those which
generate ROS and free radicals.
The altered structure of PEI might underlie the reduction in ROS
levels. Introduction of the H-b-Ala-His (L-Carnosine) as a side chain
in the PEI backbone might reduce the toxic effect of the large
number of free available eNH2groups in PEI. Interestingly, ROS
levels reduce even further in the presence of Boc-L-Carnosine-PEI.
This might be explained by the introduction of the tri-methyl
containing ‘Boc’ group of Boc-L-Carnosine-PEI, which protects all
the reactive amine groups of L-Carnosine-PEI. This further reduces
the toxicity of PEI by blocking some and making the other eNH2
groups inaccessible for reaction with intracellular components.
Protection of L-Carnosine-PEI with the Boc group significantly
increased the effectivenessof this polymer in twoways. First, Boc-L-
Carnosine-PEI potentiated transfection efficiency such that 8-fold
less concentration of the polymer was required to induce maximal
plasmid expression as compared to L-Carnosine-PEI alone (50 mg of
L-Carnosine-PEI vs 6 mg of Boc-L-Carnosine-PEI for the same
quantity of DNA). Second, cell death was much lower at this
concentration of polymer. The combination of higher transfection
efficiency with lower cell death results in a much higher number of
live, positively transfected cells from Boc-L-Carnosine-PEI trans-
fection compared to other agents.
In summary, our studies demonstratethat L-Carnosine (H-b-Ala-
His) derivatives of PEI significantly enhance transfection efficiency
of primary human cells, which are otherwise known to be highly
resistant to transfection. Furthermore, modification of PEI with
carnosine appears to reduce the cytotoxic effect of PEI by lowering
and pathwaysinvolvedin
the formation of ROS within cells by modulating the activity of
multiple enzymes and pathways involved in metabolism of free
radicals and reactive oxygen species. Protection of L-Carnosine-PEI
with the ‘Boc’ group further enhances the antioxidant property of
the hybrid polymer. For future studies, the concentration of the
present formulation of L-Carnosine-PEI needs to be optimized so
that this polymer can achieve the same efficiency at lower
concentrations which will result in even higher cell survival.
5. Conclusions
In this study, a panel of polymers has been described which can
improve gene delivery to primary human stem cells. We have
demonstrated that peptide-linked PEI complexes can efficiently
condense DNA into nano-aggregates. These polyplexes significantly
reduce the toxic effects of the 25 kDa PEI and demonstrate marked
improvement in cell survival and transfection efficiency when
compared to the polyplexes of linear PEI polymers. PEI linked with
the dipeptide, L-Carnosine, and its derivative, Boc-L-Carnosine,
appear tobe the most promising genedeliveryagentsamong all the
complexes based on their reduced production of reactive oxygen
species and markedly high transfection efficiency. Future research
should focus on other important applications of these polymers
such as potential for in vivo gene delivery in diseased patients, with
the goal of a safe and effective clinical translation in the near future.
Acknowledgements
This research was supported by the Bio-X Program at Stanford
(ASL), NIH HL095571, and EB009689 (JCW). The authors would like
to thank the Imaging Facility Center and the Flow Cytometry
Facility at Stanford.
Appendix. Supplementary data
Supplementary data related with this article can be found at doi:
10.1016/j.biomaterials.2011.03.016.
References
[1] Persons DA. Gene therapy: targeting beta-thalassaemia. Nature 2010;467:
277e8.
[2] Mangi AA, Noiseux N, Kong D, He H, Rezvani M, Ingwall JS, et al. Mesenchymal
stem cells modified with Akt prevent remodeling and restore performance of
infarcted hearts. Nat Med 2003;9:1195e201.
[3] Hiona A, Wu JC. Noninvasive radionuclide imaging of cardiac gene therapy:
progress and potential. Nat Clin Pract Cardiovasc Med 2008;5:S87e95.
[4] van der Woude I, Wagenaar A, Meekel AA, ter Beest MB, Ruiters MH,
Engberts JB, et al. Novel pyridinium surfactants for efficient, nontoxic in vitro
gene delivery. Proc Natl Acad Sci U S A 1997;94:1160e5.
[5] Pfeifer A, Verma IM. Gene therapy: promises and problems. Annu Rev
Genomics Hum Genet 2001;2:177e211.
[6] Boussif O, Lezoualc’h F, Zanta MA, Mergny MD, Scherman D, Demeneix B, et al.
A versatile vector for gene and oligonucleotide transfer into cells in culture
and in vivo: polyethylenimine. Proc Natl Acad Sci U S A 1995;92:7297e301.
[7] Schafer J, Hobel S, Bakowsky U, Aigner A. Liposome-polyethylenimine
complexes for enhanced DNA and siRNA delivery. Biomaterials 2010;31:
6892e900.
[8] Brownlie A, Uchegbu IF, Schatzlein AG. PEI-based vesicle-polymer hybrid gene
delivery system with improved biocompatibility. Int J Pharm 2004;274:
41e52.
[9] Beretta G, Artali R, Regazzoni L, Panigati M, Facino RM. Glycyl-histidyl-lysine
(GHK) is a quencher of alpha, beta-4-hydroxy-trans-2-nonenal: a comparison
with carnosine. insights into the mechanism of reaction by electrospray
ionization mass spectrometry, 1H NMR, and computational techniques. Chem
Res Toxicol 2007;20:1309e14.
[10] Takahashi K, Yamanaka S. Induction of pluripotent stem cells from mouse
embryonic and adult fibroblast cultures by defined factors. Cell 2006;126:
663e76.
[11] Sun N, Panetta NJ, Gupta DM, Wilson KD, Lee A, Jia F, et al. Feeder-free
derivation of induced pluripotent stem cells from adult human adipose stem
cells. Proc Natl Acad Sci U S A 2009;106:15720e5.
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4657

Page 12

[12] Wu SM, Chien KR, Mummery C. Origins and fates of cardiovascular progenitor
cells. Cell 2008;132:537e43.
[13] Moret I, Esteban Peris J, Guillem VM, Benet M, Revert F, Dasi F, et al. Stability
of PEI-DNA and DOTAP-DNA complexes: effect of alkaline pH, heparin and
serum. J Control Release 2001;76:169e81.
[14] Smits AM, van Vliet P, Metz CH, Korfage T, Sluijter JP, Doevendans PA, et al.
Human cardiomyocyte progenitor cells differentiate into functional mature
cardiomyocytes: an in vitro model for studying human cardiac physiology and
pathophysiology. Nat Protoc 2009;4:232e43.
[15] Zheng M, Jagota A, Semke ED, Diner BA, McLean RS, Lustig SR, et al. DNA-
assisted dispersion and separation of carbon nanotubes. Nat Mater 2003;2:
338e42.
[16] Partridge KA, Oreffo ROC. Gene delivery in bone tissue engineering: progress
and prospects using viral and nonviral strategies. Tissue Eng 2004;10:
295e307.
[17] Barry SP, Townsend PA. What causes a broken heartemolecular insights into
heart failure. Int Rev Cell Mol Biol 2010;284:113e79.
[18] Hashemi M, Parhiz BH, Hatefi A, Ramezani M. Modified polyethyleneimine
with histidine-lysine short peptides as gene carrier. Cancer Gene Ther 2011;
18:12e9.
[19] Kim WS, Park BS, Sung JH. The wound-healing and antioxidant effects of
adipose-derived stem cells. Expert Opin Biol Ther 2009;9:879e87.
[20] Mathiasen AB, Haack-Sorensen M, Kastrup J. Mesenchymal stromal cells for
cardiovascular repair: current status and future challenges. Future Cardiol
2009;5:605e17.
[21] Gersh BJ, Simari RD, Behfar A, Terzic CM, Terzic A. Cardiac cell repair therapy:
a clinical perspective. Mayo Clin Proc 2009;84:876e92.
[22] Oh YM, Jang EH, Ko JH, Kang JH, Park CS, Han SB, et al. Inhibition of 6-
hydroxydopamine-induced endoplasmic reticulum stress by l-carnosine in
SH-SY5Y cells. Neurosci Lett 2009;459:7e10.
[23] Gariballa SE, Sinclair AJ. Carnosine: physiological properties and therapeutic
potential. Age Ageing 2000;29:207e10.
[24] Aydin AF, Kusku-Kiraz Z, Dogru-Abbasoglu S, Uysal M. Effect of carnosine
treatment on oxidative stress in serum, apoB-containing lipoproteins fraction
and erythrocytes of aged rats. Pharmacol Rep 2010;62:733e9.
[25] Cheng J, Wang F, Yu DF, Wu PF, Chen JG. The cytotoxic mechanism of
malondialdehyde and protective effect of carnosine via protein cross-linking/
mitochondrialdysfunction/reactive
neurons. Eur J Pharmacol; 2010.
[26] Artun BC, Kusku-Kiraz Z, Gulluoglu M, Cevikbas U, Kocak-Toker N, Uysal M.
The effect of carnosine pretreatment on oxidative stress and hepatotoxicity in
binge ethanol administered rats. Hum Exp Toxicol 2010;29:659e65.
[27] Babizhayev MA, Yegorov YE. Telomere attrition in lens epithelial cells -
a target for N-acetylcarnosine therapy. Front Biosci 2010;15:934e56.
oxygenspecies/MAPK pathwayin
D. Dey et al. / Biomaterials 32 (2011) 4647e4658
4658