Targeted delivery of siRNA to hepatocytes and hepatic stellate cells by bioconjugation

Article abstract of DOI:10.1021/bc100346n

Previously, we successfully conjugated galactosylated poly(ethylene glycol) (Gal-PEG) to oligonucleotides (ODNs) via an acid labile ester linker (Zhu et al., Bioconjugate Chem. 2008, 19, 290-8). In this study, sense strands of siRNA were conjugated to Gal

Full text of DOI:10.1021/bc100346n

Bioconjugate Chem. 2010, 21, 2119–2127
2119
Targeted Delivery of siRNA to Hepatocytes and Hepatic Stellate Cells by
Bioconjugation
Lin Zhu and Ram I. Mahato*
Department of Pharmaceutical Sciences, University of Tennessee Health Science Center, Memphis, Tennessee 38103, United States.
Received July 28, 2010; Revised Manuscript Received September 20, 2010
Previously, we successfully conjugated galactosylated poly(ethylene glycol) (Gal-PEG) to oligonucleotides (ODNs)
via an acid labile ester linker (Zhu et al., Bioconjugate Chem. 2008, 19, 290-8). In this study, sense strands of
siRNA were conjugated to Gal-PEG and mannose 6-phosphate poly(ethylene glycol) (M6P-PEG) for targeted
delivery of siRNAs to hepatocytes and hepatic stellate cells (HSCs), respectively. These siRNA conjugates were
purified by ion exchange chromatography and verified by gel retardation assay. To evaluate their RNAi functions,
the validated siRNA duplexes targeting firefly luciferase and transforming growth factor beta 1 (TGF-ꢀ1) mRNA
were conjugated to Gal-PEG and M6P-PEG, and their gene silencing efficiencies were determined after transfection
into HepG2 and HSC-T6 cells. The disulfide bond between PEG and siRNA was cleaved by dithiothreitol, leading
to the release of intact siRNA. Both Gal-PEG-siRNA and M6P-PEG-siRNA conjugates could silence luciferase
gene expression by about 40% without any transfection reagents, while the gene silencing effects reached more
than 98% with the help of cationic liposomes at the same dose. Conjugation of TGF-ꢀ1 siRNA with Gal-PEG
and M6P-PEG could silence endogenous TGF-ꢀ1 gene expression as well. In conclusion, these siRNA conjugates
have the potential for targeted delivery of siRNAs to hepatocytes and hepatic stellate cells for efficient gene
silencing in vivo.
cokinetic profiles. We previously reported that the conjugation
of galactose (10)-poly(ethylene glycol) (PEG) to oligonucle-
otide (ODN) using an acid-labile ester linker successfully
stabilized ODN, prolonged its circulation time, and specifically
delivered ODN to hepatocytes after systemic administration (11).
The linkage between ODN and PEG could be cleaved at pH
5.5 (endosomal pH), and free ODN could be released. In a
separate study, we have also demonstrated that mannose-6-
phosphate (M6P) as a ligand could significantly increase the
accumulation of ODN in hepatic stellate cells (HSCs) in rats.
Here, we used a similar strategy to Gal-PEG-ODN conjugate
to design two siRNA conjugates, Gal-PEG-siRNA and M6P-
PEG-siRNA, for targeted delivery of siRNAs to hepatocytes
via asialoglycoprotein receptor-mediated endocytosis and HSCs
via M6P/insulin-like growth factor-II receptor (M6P/IGF2R)
mediated endocytosis, respectively. In this study, the heterobi-
functional PEG, (ortho-pyridyl) disulfide-poly(ethylene glycol)-
N-hydroxysuccinimidyl ester (OPSS-PEG-NHS), was used as
a backbone followed by modification with either Gal or M6P
at its NHS end and with the 3′-sense strand of siRNA at its
OPSS end via disulfide bond. To make sure there is no loss in
gene silencing due to bioconjugation, Luciferase siRNA and
TGF-ꢀ1 siRNA were used as model siRNA sequences.
INTRODUCTION
Significant progress has been made in enhancing the potency
and specificity of RNA interference (RNAi) molecules, which
include double-stranded RNA (dsRNA), small interfering RNA
(siRNA) (19-23 base pairs), short hairpin RNA (shRNA), and
micro RNA (miRNA). Although they are structurally different,
all of these molecules induce sequence-specific gene silencing
by either degrading or inhibiting the target mRNA via perfect
or mismatched binding in the presence of RNA-induced
silencing complex (RISC) (1, 2). The simplest and most effective
strategy of gene silencing is to synthesize siRNAs based on
the target mRNA sequence. siRNA has been shown to have
the potential to treat numerous diseases, including cancer,
fibrosis, genetic disorders, and viral infection (3-5). However,
siRNAs have not been successfully used as clinical therapeutics
due to several obstacles, including the poor enzymatic stability,
short circulation time in the bloodstream, nonspecific tissue or
cell targeting, and insufficient intracellular transport (6).
Various cationic lipids and polymers are being used as
transfection reagents for siRNA delivery. siRNAs could be
condensed by these carriers via electrostatic interaction into
nanosized particles which are easy for cells to take up via
endocytosis, membrane fusion, or both. After complex forma-
tion, siRNAs are protected from enzymatic digestion (7).
However, cationic carriers are not suitable for in vivo delivery
of siRNAs since they may bind to negatively charged serum
proteins and undergo aggregation resulting in undesired distribu-
tion, and induction of inflammatory cytokines release and
activation of the complement system (8, 9).
MATERIALS AND METHODS
Materials. OPSS-PEG-NHS (MW: 3400 Da) was purchased
from Creative PEGWorks (Winston Salem, NC). Palladium (10
wt % on activated carbon), thiophosgene, dithiothreitol (DTT),
L-glutathione (GSH) reduced, p-aminophenyl ꢀ-D-galactopyra-
noside, ethidium bromide (EtBr), p-nitrophenyl-R-D-mannopy-
ranoside, dimethylformamide (DMF), and deuterium oxide
(D2O) were purchased from Sigma-Aldrich Chemicals (St.
Louis, MO). Sodium chloride, acetic acid, and HPLC-grade
water were purchased from Fisher Scientific (Fair Lawn, NJ).
Luciferase assay kit, TE buffer, 6× gel loading buffer, and Tris-
hydrochloride were purchased from Promega (Madison, WI).
PEGylation is known to increase the in vitro and in vivo
stability of drug molecules, as well as improve their pharma-
* Corresponding author. Ram I. Mahato, Ph.D. Professor, Department
of Pharmaceutical Sciences, University of Tennessee Health Science
Center, 19 South Manassas St., Cancer Research Building RM 226,
Memphis, TN 38103. Tel: (901) 448-6929. Fax: (901) 448-2099.
E-mail: rmahato@uthsc.edu.
10.1021/bc100346n  2010 American Chemical Society
Published on Web 10/21/2010

2120 Bioconjugate Chem., Vol. 21, No. 11, 2010
Zhu and Mahato
Human/mouse TGF-ꢀ1 ELISA kit was purchased from eBio-
science (San Diego, CA). Dialysis tubing (MWCO 1000 Da)
was purchased from Spectrum Laboratories, Inc. (Houston,
TX). Zeba Spin Desalting Columns (MWCO: 7000 Da) was
purchased from Thermo Scientific (Rockford, IL). Dulbecco’s
modified Eagle’s medium (DMEM), penicillin streptomycin
solution (100×), phosphate buffered saline (PBS), trypsin-
EDTA, agarose, and Lipofectamine 2000 were purchased
from Invitrogen Corporation (Carlsbad, CA). Bovine serum
albumin (BSA) (fraction V, purity >98%) was purchased from
Atlanta Biologicals (Lawrenceville, GA). Pyridinium cationic
liposomes containing pyridinium lipid (C16:1, amide linker,
trans)/DOPE (1:1 mol/mol) cationic liposomes were prepared
by Zhu et al. (12).
dialyzed against distilled water for 48 h with MWCO 1000 to
remove the unreacted papM6P. The product M6P-PEG-OPSS
was freeze-dried and characterized by 1H NMR after dissolving
it in D2O.
Synthesis of Gal-PEG-siRNA and M6P-PEG-siRNA. The
lyophilized 3′ thiol-siRNA was reconstituted in TE buffer of
pH 8.3-8.5 at a concentration of approximately 100 A260 units/
mL. siRNA solution was then treated with 0.1 M dithiothreitol
(DTT) aqueous solution at room temperature for 30 min. The
excess DTT was removed by gel filtration and 3′-sulfhydryl
siRNA was stored under N2 to avoid oxidative dimerization to
the disulfide before conjugation. 3′-Sulfhydryl siRNA and Gal-
PEG-OPSS (or M6P-PEG-OPSS) were dissolved in 0.5 mL PBS
(pH7.5) at a molar ratio of 1:100, and stirred under nitrogen at
room temperature for 6 h. The reaction mixture was analyzed
and purified by ion exchange high-performance liquid chroma-
tography, which was carried on a Resource Q ion exchange
column (GE Healthcare Life Sciences, Piscataway, NJ) by an
HPLC system (Waters, Milford, MA) with detection at 260 nm
using a gradient starting from 0% B to 80% B (A, 20 mM Tris-
HCl buffer; B, 20 mM Tris-HCl buffer containing 0.1 M sodium
chloride) at a flow rate of 1 mL/min at 25 °C. The purified
Gal-PEG-siRNA and M6P-PEG-siRNA were desalted using
Zeba Spin Desalting Column by centrifugation at the speed of
2000 g and freeze-dried.
Small Interfering RNAs (siRNAs). Luciferase siRNA (sense
strand: CUUACGCUGAGUACUUCGA[dT][dT], antisense
strand: UCGAAGUACUCAGCGUAAG[dT][dT]) was pur-
chased from Invitrogen Corporation. TGF-ꢀ1 siRNA (13) (sense
strand: GCUGUACAUUGACUUUAGG[dT][dT], antisense
strand: CCUAAAGUCAAUGUACAGC[dT][dT]) (Accession
no.: NM_021578) and the scramble siRNA were purchased from
Invitrogen Corporation (Carlsbad, CA). For conjugation pur-
poses, the sense strand of the siRNA was modified with a thiol
group (-(CH2)3-S-S-(CH2)3-OH-3′) at its 3′ end by Invitrogen
Corporation.
Methods. Synthesis of p-Aminophenyl ꢀ-D-Galactopyrano-
side and p-Aminophenyl-6-phospho-R-D-mannopyranoside. p-
Aminophenyl ꢀ-D-galactopyranoside was synthesized according
to our previous study (11). Briefly, p-nitrophenyl ꢀ-D-galacto-
pyranoside was reduced with 10% palladium on activated carbon
under hydrogen (1 atm) in a 1:1 (v/v) ethanol-water mixture
for 2 h. The product, p-aminophenyl ꢀ-D-galactopyranoside, was
Characterization of Gal-PEG-siRNA and M6P-PEG-siRNA.
The concentrations of siRNA and siRNA conjugate solutions
were quantitated by a UV spectrophotometer. To determine the
cleavage of the disulfide bond and siRNA dissociation from
Gal-PEG-siRNA and M6P-PEG-siRNA, 2 µg siRNA and the
conjugates containing 2 µg siRNA were incubated in 50 mM
DTT aqueous buffer for 40 min at room temperature. The DTT-
treated samples and controls (nontreated siRNA and siRNA
conjugate) were mixed with 6× gel loading buffer and applied
on the 1% agarose gel and run at 10 V/cm for 60 min at room
temperature followed by ethidium bromide (EtBr) staining and
visualization under UV light.
Cell Cultures and Transfections. HepG2 cells were main-
tained in a complete growth medium containing DMEM medium
supplemented with 10% FBS, 100 IU/mL penicillin, and 100
µg/mL streptomycin at 37 °C in a 5% CO2 humidified
atmosphere. HSC-T6 cells, kindly provided by Dr. Scott
Friedman (Mount Sinai School of Medicine, New York), were
maintained in the same condition as HepG2 cells. Before
transfection, the cells were seeded in 24-well plates and
maintained in 0.5 mL of complete growth medium at 37 °C for
24 h. The cell number was about 4 × 104 cells per well.
1
concentrated by solvent evaporation and characterized by H
NMR and mass spectrometry after dissolving in D2O.
p-Aminophenyl-6-phospho-R-D-mannopyranoside was syn-
thesized according to our previous study (14) by dissolving
p-nitrophenyl-R-D-mannopyranoside (pnpM) (0.3 g, 1 mmol)
in the mixture of acetonitrile, pyridine, and water (1/0.4/0.04,
v/v/v). Phosphorus oxide chloride (0.4 mL, 4.4 mmol) was added
and stirred for 1 h on an ice-water bath. The reaction mixture
was poured onto 20 g of ice. After the ice melted, pH was
adjusted to 7.0 by slowly adding 2.5 M NaOH, and the solution
was evaporated to dryness. The solid material was dissolved in
water and crystallized at 4 °C overnight. The crystals were
filtered and washed with 5 mL of absolute ethanol. The
compound was recrystallized in a mixture of 1 mL of water
and 10 mL of ethanol, redissolved in water, and lyophilized to
give p-nitrophenyl-6-phospho-R-D-mannopyranoside (pnpM6P).
PnpM6P was reduced with 100 mg of 10% palladium on
activated carbon under H2 (1 atm) in a 4:1 (v/v) methanol-water
mixture for 2 h. After filtration, the residue was redissolved in
water and lyophilized to give p-aminophenyl-6-phospho-R-D-
mannopyranoside (papM6P). The product was characterized
Luciferase expression plasmid (pDNA3-Luc) was first trans-
fected with cells by pyridinium cationic liposomes (12). Briefly,
0.2 µg/well luciferase plasmid and pyridinium cationic lipo-
somes were diluted to 25 µL by Opti-MEM I medium,
respectively, and incubated for 5 min at room temperature.
Lipoplexes were prepared by mixing plasmid and liposomes
in equal volume and incubated for 20 min at room temper-
ature to allow complex formation. Immediately before
transfection, the medium was removed, and cells were washed
gently with sterile phosphate buffered saline. Lipoplexes were
added to each well and diluted to 0.5 mL by Opti-MEM I
medium, and the plate was incubated under 5% CO2 at 37
°C for 6 h. Following transfection, the culture medium was
replaced with complete growth medium and maintained
overnight before further treatment.
1
using H NMR and mass spectrometry.
Synthesis of Gal-PEG-OPSS and M6P-PEG-OPSS. OPSS-
PEG-NHS (50 mg) and p-aminophenyl ꢀ-D-galactopyranoside
(50 mg) were dissolved separately in 0.4 mL of dimethylfor-
mamide (DMF), mixed together, and stirred for 6 h at room
temperature under N2 protection in the dark. The reaction
mixture was dialyzed against distilled water for 48 h with
MWCO 1000 to remove the unreacted p-aminophenyl ꢀ-D-
galactopyranoside. The product Gal-PEG-OPSS was freeze-
dried and characterized by 1H NMR after dissolving it in D2O.
OPSS-PEG-NHS (50 mg) were dissolved in 0.4 mL of DMF,
and papM6P (50 mg) was dissolved in 0.4 mL of distilled water.
Then, they were mixed and stirred for 6 h at room temperature
under N2 protection in the dark. The reaction mixture was
To transfect luciferase siRNA and its conjugate, two methods
were used. In the first method, siRNA and its conjugates were
added directly into cell cultures without the assistance of cationic
liposomes. In the second method, siRNA and its conjugates were
mixed with cationic liposomes to form the lipoplexes before

Targeted Delivery of siRNA by Bioconjugation
Bioconjugate Chem., Vol. 21, No. 11, 2010 2121
Figure 1. Synthesis scheme of Gal-PEG-siRNA and M6P-PEG-siRNA.
adding into cell cultures. After 6 h of transfection, the medium
was replaced with the complete growth medium and incubated
for an additional 42 h.
flexible long chain of PEG and eluted between PEG and siRNA.
The chromatograms of Gal-PEG-OPSS and Gal-PEG-siRNA
were not shown, since they are similar to those of M6P-PEG-
OPSS and M6P-PEG-siRNA. Gal-PEG-siRNA and M6P-PEG-
siRNA conjugates were also identified by 1% agarose gel
electrophoresis and visualized under UV light (Figure 5).
Dissociation of siRNAs from Gal-PEG-siRNA and M6P-
PEG-siRNA Conjugates. To determine whether the disulfide
bond between Gal-PEG or M6P-PEG and siRNA is cleavable
and free siRNA can be released, siRNA and its conjugates were
treated with DTT aqueous buffer for 40 min and analyzed by
gel retardation assay. Gal-PEG-siRNA and M6P-PEG-siRNA
conjugates migrated more slowly than free siRNA. However,
after treatment with DTT for 40 min, the disulfide bond was
completely cleaved and free siRNA was released and migrated
at the same speed as the siRNA control (Figure 5).
Gene Silencing Study. Firefly luciferase expression plasmid
was transfected into both HepG2 and HSC-T6 cells using
previously optimized pyridinium cationic liposomes at the
charge ratio of 3:1 (N/P) (12). Gene silencing efficiencies of
siRNA conjugates were determined after transfection of lu-
ciferase expression plasmid. In the absence of cationic lipo-
somes, Gal-PEG-siRNA conjugate showed luciferase gene
silencing up to about 45% at the dose of 400 nM compared to
the naked siRNA group and the control group (plasmid group)
without siRNA treatment (Figure 6A). Addition of an excessive
amount of Gal-PEG-OPSS into HepG2 cells before transfection
of siRNA conjugate abolished the gene silencing ability of Gal-
PEG-siRNA conjugate. Similarly, pretreatment of HSC-T6 cells
with an excess amount of M6P-PEG-OPSS before the trans-
fection of M6P-PEG-siRNA conjugate abolished the gene
silencing ability of this conjugate. Target cells were pretreated
with an excess amount of Gal-PEG-OPSS or M6P-PEG-OPSS
to saturate the surface receptors and inhibit the receptor-mediated
cellular uptake of siRNA conjugates. Since the molecular
weights of galactose or mannose 6-phosphate are too small and
may not be effective in saturating the target receptors, the
conjugates containing the ligands are commonly used (11). To
keep data comparable, we used 400 nM siRNA conjugates in
Gal-PEG+ conjugate and M6P-PEG+conjugate groups.
Figure 6B showed the luciferase gene silencing efficiency of
siRNA and Gal-PEG-siRNA. After complex formation with
pyridinium cationic liposomes, the gene silencing of Gal-PEG-
siRNA significantly increased and reached up to 90% even at
the dose of 100 nM, while the gene silencing effect of luciferase
siRNA was just about 60% at the same dose. Figure 6C showed
the gene silencing ability of M6P-PEG-siRNA in HSC-T6 cells
without pyridinium cationic liposomes. M6P-PEG-siRNA could
To determine luciferase gene expression, the cells were
collected and lysed. The cell lysate was measured for luciferase
gene expression by Luciferase Assay Kit (Promega, Madison,
WI). The total protein concentration was measured by bicin-
choninic acid (BCA) protein assay reagent (Pierce, Rockford,
IL). Luciferase activity was normalized by the total protein
content and expressed as relative light unit (RLU) per microgram
of protein (RLU/µg protein).
To determine their gene silencing effects, TGF-ꢀ1 siRNA
and its conjugates were mixed with the media and added into
HepG2 cells and HSC cells. After 6 h of transfection, the
medium was replaced with the complete growth medium and
incubated for additional 42 h. To determine TGF-ꢀ1 gene
expresstion, the supernatant was collected and measured by
Human/Mouse TGF-ꢀ1 ELISA kit.
RESULTS
Synthesis of Gal-PEG-siRNA and M6P-PEG-siRNA. The
synthesis scheme of Gal-PEG-siRNA and M6P-PEG-siRNA was
shown in Figure 1. Before conjugation, p-nitrophenyl ꢀ-D-
galactopyranoside was reduced to p-aminophenyl ꢀ-D-galacto-
1
pyranoside. ESI-MS and H NMR spectra of this intermediate
product were not shown but could be found in our previous
paper (11). p-Nitrophenyl-R-D-mannopyranoside (pnpM) was
phosphorylated to p-nitrophenyl-6-phospho-R-D-mannopyra-
noside (pnpM6P) and pnpM6P was reduced to p-aminophe-
nyl-6-phospho-R-D-mannopyranoside (papM6P). ESI-MS and
1H NMR results of this intermediate product are shown in
Figure 2.
Gal-PEG-OPSS and M6P-PEG-OPSS were synthesized from
heterofunctional PEG derivative, NHS-PEG-OPSS and identified
by 1H NMR. After reaction with Gal or M6P, the two peaks (a
and b positions) of the benzene ring were found in the NMR
spectra. The data indicated that Gal and M6P reacted with NHS
group and linked with PEG (Figure 3). Gal-PEG-OPSS and
M6P-PEG-OPSS were then conjugated to 3′-sulfhydryl siRNA
by disulfide exchange reaction between sulfhydryl and OPSS
groups. Ion exchange HPLC was used to monitor the conjuga-
tion reaction and purify the final products. The ion exchange
HPLC chromatograms of siRNA, M6P-PEG-OPSS, and M6P-
PEG-siRNA were shown in Figure 4. M6P-PEG-OPSS eluted
first, since they have no charge and cannot bind to anion
exchange resin; free siRNA is a negatively charged molecule
and bound to anion exchange resin and eluted last. After
conjugation, the charge of siRNA was partially blocked by the

2122 Bioconjugate Chem., Vol. 21, No. 11, 2010
Zhu and Mahato
Figure 2. Synthesis and characterization of p-aminophenyl-6-phospho-R-D-mannopyranoside with ESI-MS and 1H NMR. (A) Synthesis scheme of
p-aminophenyl-6-phospho-R-D-mannopyranoside. (B) Electron spray ionization mass spectra were obtained after dissolving the product in a 4:1
1
(v/v) methanol-water mixture on an ESQUIRE-LC ion trap LC/MS sy stem in negative mode. (C) For H NMR, samples were dissolved in D2O,
and the spectra were recorded on a Bruker ARX-500 MHz NMR spectrometer at 25 °C.
silence luciferase gene expression at a similar level to Gal-PEG-
siRNA. Luciferase gene silencing was dependent on Gal-PEG-
siRNA and M6P-PEG-siRNA dose in both the absence and
presence of pyridinium cationic liposomes.
To determine the effect of bioconjugation on TGF-ꢀ1 gene
silencing in HepG2 and HSC-T6 cells, the sense strand of a
potent TGF-ꢀ1 siRNA duplex was conjugated with either Gal-
PEG-OPSS or M6P-PEG-OPSS to form TGF-ꢀ1 siRNA con-
jugates. After transfection, the level of TGF-ꢀ1 protein in the
cell medium was measured by ELISA kit. The gene silencing
effects of siRNA conjugates were shown in Figure 7. Without
transfection reagents, conjugation of TGF-ꢀ1 siRNA with Gal-
PEG-OPSS or M6P-PEG-OPSS significantly enhanced TGF-
ꢀ1 gene silencing effects up to 40% of the nontreatment group
compared to TGF-ꢀ1 siRNA.
DISCUSSION
Effective delivery of siRNA to specific cell types in vivo is
critical for the successful therapeutic use of siRNA. The current
studies of siRNA delivery systems are focused on the electro-
static interaction with cationic liposomes and polymers, since
negatively charged siRNAs are poorly taken up by the cells.
Although the use of these cationic carriers is effective in vitro,
an ideal siRNA carrier should not bear any positive charge, since

Targeted Delivery of siRNA by Bioconjugation
Bioconjugate Chem., Vol. 21, No. 11, 2010 2123
1
Figure 3. Characterization of Gal-PEG-OPSS (A) and M6P-PEG-OPSS (B) with H NMR. The samples were dissolved in D2O, and the spectra
were recorded on a Bruker ARX-500 MHz NMR spectrometer at 25 °C.
it may cause an immune response and unexpected distribution
due to nonspecific ionic interaction with plasma proteins. To
avoid the use of polycations, Rajue et al. directly conjugated
asialoglycoprotein to ODNs via disulfide bonds to target
hepatocytes (15). Nakagawa et al. directly conjugated the small-
molecule ligand, anisamide, to ODNs, to target the σ receptor
on the surface of human lung cancer cells (16). The lipophilic
molecule, R-tocopherol, was also covalently conjugated to the
antisense strand of siRNA to target the liver (17). Although
direct conjugation of ligands to ODN or siRNA increased their
targeting efficiencies, it might not significantly improve the
stability or dissociation of loaded ODN and siRNA. PEGylation
is known to significantly enhance the stability of ODNs and
siRNAagainst exonucleases and reduce renal clearance com-
pared to unmodified ones (18-20), which improves their
bioactivity by increasing plasma residence time and decreasing
nuclease degradation. The flexible PEG chains also shield the
inherent negative charge of these ODNs and siRNA, thereby
possibly facilitating cellular uptake of the conjugated ODNs and
siRNA (21).
Oligonucleotides and siRNA are cleared rapidly from the
blood circulation and accumulated in the peripheral tissues,
especially in the liver and kidney (22-24). Their in vivo activity
depends on their pharmacokinetic profiles, especially the rate
of excretion and degradation. In a previous study, we success-
fully conjugated ODN to Gal-PEG via an acid-labile linkage
and demonstrated that systemic administration of Gal-PEG-
ODN significantly increased accumulation in hepatocytes via
galactose/asialoglycoprotein receptor-mediated endocytosis (11).
In a separate study, we synthesized M6P-BSA and conjugated
this to ODN. Systemic administration of M6P-BSA-ODN
conjugate increased the accumulation of ODNs in HSCs via
M6P/IGF2R mediated endocytosis compared to other types of
liver cells in rats (14).

2124 Bioconjugate Chem., Vol. 21, No. 11, 2010
Zhu and Mahato
Figure 4. Purification of M6P-PEG-siRNA conjugate by ion-exchange chromatography. The reaction mixture was carried on a Resource Q ion
exchange column (GE Healthcare Life Sciences, Piscataway, NJ) by an HPLC system (Waters, Milford, MA) with detection at 260 nm using a
gradient starting from 0% B to 80% B (A, 20 mM Tris-HCl buffer; B, 20 mM Tris-HCl buffer containing 0.1 M sodium chloride) at a flow rate
of 1 mL/min at 25 °C.
we used cleavable disulfide bonds as the linkage between siRNA
and PEG to ensure the release of intact siRNA molecules from
the conjugates in the reducing environment in the cytoplasm
where RNAi takes action after endocytosis. This is expected to
be better than the use of PEGylated cationic liposomes and PEI,
where the PEG chain may influence the endosomal escape, as
the PEG chain often cannot be released from the carriers.
Chronic liver injury and inflammation of hepatocytes or other
hepatic cells cause liver fibrosis, resulting in excessive ac-
cumulation of extracellular matrix (ECM) proteins and activation
and transformation of HSCs to proliferative myofibroblast-like
cells. Many inflammatory cytokines were up-regulated during
liver fibrosis. Among them, TGF-ꢀ1, which is mainly expressed
by HSCs, plays a key role in initiation and maintenance of liver
fibrosis (36). TGF-ꢀ1 gene silencing has been shown to inhibit
the ECM production as well as accelerate its degradation
resulting in resolution of fibrogenesis (37). In our previous study,
we demonstrated siRNA sequence and dose-dependent TGF-
ꢀ1 gene silencing and consequent inhibition of collagen
synthesis (13). Therefore, hepatocytes and HSCs are often
Figure 5. Identification of siRNA conjugates by gel retardation assay.
chosen as targets to deliver the therapeutics to treat liver fibrosis.
Excessive ECM causes the loss of sinusoidal fenestrae (38),
suggesting that liposomal and other particulate delivery systems
may not be suitable for siRNA delivery to hepatic cells. In
comparison to particulate delivery systems, the water-soluble
delivery systems, Gal-PEG-siRNA and M6P-PEG-siRNA, are
much more attractive for siRNA delivery to hepatocytes and
HSCs in fibrotic liver. In this study, the 3′-thiol modifier was
first reduced to a 3′-sulfhydryl group, which was active and
could form a disulfide bond with an (ortho-pyridyl)-disulfide
(OPSS) group by disulfide exchange reaction in PBS. The
disulfide bond was cleaved, and free siRNAs were released after
treatment with a reducing agent DTT. The gel retardation assay
confirmed that, after conjugation with PEG, the migration rate
of siRNA conjugates was significantly slower than that of free
siRNA, but after DTT treatment, the linker between siRNA and
PEG was cleaved, and the released siRNA showed a similar
migration rate to unconjugated siRNA (Figure 5). We did not
use mass spectrophotometry to identify the siRNA conjugates,
because siRNA may get degraded during the MALDI-TOF MS
test, and it requires a large amount of siRNA to get the
signal (39, 40).
Two microgram aliquots of siRNAs, Gal-PEG-siRNA, and M6P-PEG-
siRNA were incubated with 50 mM DTT at room temperature for 40
min. The treated samples and nontreated controls were applied on 1%
agarose gel and run at 10 V/cm for 60 min followed by EtBr staining,
then visualized under UV light.
siRNAs can be covalently linked with functional molecules
via either noncleavable or cleavable linkages, such as acid-labile
ester (11, 25) and reducible disulfide bond (19, 26). For single-
stranded ODNs, both 3′- and 5′-termini of these molecules are
reported to be used as functional sites for the conjugation
reaction (11, 27). For the siRNA duplex, however, the antisense
strand of siRNA plays a key role and works as the template in
RISC to perform RNAi. Improper modification of antisense
strands may cause the loss of RNAi function (28). It was further
shown that the 5′-terminus of the antisense strand was more
important than the 3′-terminus and determined RNAi activity
(29-31). In comparison to the antisense strand, manipulation
of the sense strand of siRNA is relatively safe and effective.
Both 3′- and 5′-ends of the sense strand can be used for
conjugation (32).
In the present study, we investigated the application of a PEG-
based macromolecular delivery system using two different
ligands for siRNA delivery. Since the disulfide bond has been
proven to be cleaved by reducing agents in the cytoplasm and
was extensively used in drug delivery systems (19, 33-35),
Luciferase gene silencing of siRNA conjugates indicated that
siRNAs did not lose their RNAi function during conjugation
and after dissociation from siRNA conjugates (Figure 6). siRNA

Targeted Delivery of siRNA by Bioconjugation
Bioconjugate Chem., Vol. 21, No. 11, 2010 2125
Figure 6. Luciferase gene silencing effects of Gal-PEG-siRNA and M6P-PEG-siRNA. Before the gene silencing study, luciferase plasmid was
transfected into both HepG2 and HSC-T6 cells using lipoplexes formed between luciferase plasmid and pyridinium cationic liposomes at the charge
ratio of 3/1 ((). The dose of plasmid was 0.2 µg per well in a 24-well plate. To study the gene silencing ability of Gal-PEG-siRNA, luciferase
siRNA and its Gal-PEG-siRNA conjugate were added into HepG2 cells without (A) and with (B) the formation of siRNA/liposome complexes
using pyridinium cationic liposomes. For M6P-PEG-siRNA, luciferas siRNA and its M6P-PEG-siRNA conjugate were added directly into HSC-T6
cells without cationic liposomes (C). After 6 h of transfection, the medium was replaced with the complete growth medium and incubated for
additional 42 h. The cell number was about 4 × 104 cells per well in a 24-well plate.
conjugates had about 40% down-regulation of luciferase gene
at the dose of 400 nM in both HepG2 and HSC-T6 cells in the
absence of cationic liposomes (Figure 6A,C), while the lu-
ciferase gene was almost completely silenced in the presence
of cationic liposomes at the same condition (Figure 6B). Without
condensation with cationic liposomes, the flexible hydrophilic
PEG chain surrounds and shields siRNA or the ligand on it
from accessing the cell membrane, thereby decreasing both
specific and nonspecific endocytosis. After complex formation
with cationic liposomes, the lipoplexes were slightly positively
charged and easy for cells to take up (nonspecific) and had high
gene silencing effects. Furthermore, the combination of
ligand-receptor-mediated specific endocytosis, siRNA conju-
gates showed higher gene silencing effect compared to naked

2126 Bioconjugate Chem., Vol. 21, No. 11, 2010
Zhu and Mahato
specifically accumulated in hepatocytes or HSCs and minimize
the nonspecific distribution and immunostimulatory effect in
vivo.
In conclusion, we successfully conjugated siRNA to Gal-
PEG-OPSS and M6P-PEG-OPSS via a disulfide bond. The gene
silencing effects of these siRNA conjugates were observed in
HepG2 and HSC-T6 cells. These siRNA conjugates have the
potential to be used for targeted delivery of siRNAs to
hepatocytes and hepatic stellate cells in vivo.
ACKNOWLEDGMENT
This study was supported by a grant EB003922 from the
National Institute of Health.
LITERATURE CITED
(1) Montgomery, M. K., Xu, S., and Fire, A. (1998) RNA as a
target of double-stranded RNA-mediated genetic interference in
Caenorhabditis elegans. Proc. Natl. Acad. Sci. U.S.A. 95, 15502–
7.
(2) Fire, A., Xu, S., Montgomery, M. K., Kostas, S. A., Driver,
S. E., and Mello, C. C. (1998) Potent and specific genetic
interference by double-stranded RNA in Caenorhabditis elegans.
Nature 391, 806–11.
(3) Song, E., Lee, S. K., Wang, J., Ince, N., Ouyang, N., Min, J.,
Chen, J., Shankar, P., and Lieberman, J. (2003) RNA interference
targeting Fas protects mice from fulminant hepatitis. Nat. Med.
9, 347–51.
(4) de Fougerolles, A., Vornlocher, H. P., Maraganore, J., and
Lieberman, J. (2007) Interfering with disease: a progress report on
siRNA-based therapeutics. Nat. ReV. Drug DiscoVery 6, 443–53.
(5) Oh, Y. K., and Park, T. G. (2009) siRNA delivery systems for
cancer treatment. AdV. Drug DeliVery ReV. 61, 850–62.
(6) De Paula, D., Bentley, M. V., and Mahato, R. I. (2007)
Hydrophobization and bioconjugation for enhanced siRNA
delivery and targeting. RNA 13, 431–56.
(7) Herringson, T. P., and Altin, J. G. (2009) Convenient targeting
of stealth siRNA-lipoplexes to cells with chelator lipid-anchored
molecules. J. Controlled Release 139, 229–38.
(8) Hartikka, J., Bozoukova, V., Ferrari, M., Sukhu, L., Enas, J.,
Sawdey, M., Wloch, M. K., Tonsky, K., Norman, J., Manthorpe,
M., and Wheeler, C. J. (2001) Vaxfectin enhances the humoral
immune response to plasmid DNA-encoded antigens. Vaccine
19, 1911–23.
(9) Moghimi, S. M., and Patel, H. M. (2002) Modulation of murine
liver macrophage clearance of liposomes by diethylstilbestrol.
The effect of vesicle surface charge and a role for the complement
receptor Mac-1 (CD11b/CD18) of newly recruited macrophages
in liposome recognition. J. Controlled Release 78, 55–65.
(10) Hermanson, G., Whitlow, V., Parker, S., Tonsky, K., Rusalov,
D., Ferrari, M., Lalor, P., Komai, M., Mere, R., Bell, M.,
Brenneman, K., Mateczun, A., Evans, T., Kaslow, D., Galloway,
D., and Hobart, P. (2004) A cationic lipid-formulated plasmid
DNA vaccine confers sustained antibody-mediated protection
against aerosolized anthrax spores. Proc. Natl. Acad. Sci. U.S.A.
101, 13601–6.
(11) Zhu, L., Ye, Z., Cheng, K., Miller, D. D., and Mahato, R. I.
(2008) Site-specific delivery of oligonucleotides to hepatocytes
after systemic administration. Bioconjugate Chem. 19, 290–8.
(12) Zhu, L., Lu, Y., Miller, D. D., and Mahato, R. I. (2008)
Structural and formulation factors influencing pyridinium lipid-
based gene transfer. Bioconjugate Chem. 19, 2499–512.
(13) Cheng, K., Yang, N., and Mahato, R. I. (2009) TGF-beta1
gene silencing for treating liver fibrosis. Mol. Pharm. 6, 772–9.
(14) Ye, Z., Cheng, K., Guntaka, R. V., and Mahato, R. I. (2005)
Targeted delivery of a triplex-forming oligonucleotide to hepatic
stellate cells. Biochemistry 44, 4466–76.
Figure 7. TGF-ꢀ1 gene silencing effects of Gal-PEG-siRNA and M6P-
PEG-siRNA. (A) TGF-ꢀ1 siRNA and Gal-PEG-siRNA were added
directly into HepG2 cells without cationic liposomes. (B) TGF-ꢀ1
siRNA and M6P-PEG-siRNA were added directly into HSC-T6 cells
without cationic liposomes. After 6 h of transfection, the medium was
replaced with the complete growth medium and incubated for an
additional 42 h. There were about 4 × 104 cells per well in a 24-well
plate.
siRNAs (Figure 6B). However, in the in vivo situation,
PEGylated siRNAs will be retained in blood circulation and
cleared very slowly compared to naked siRNAs. Our previous
studies showed that pegylated ODN could significantly prolong
the elimination half-life (t1/2ꢀ) of ODN up to 118.61 ( 22.06
min (11) compared to only 34.60 ( 4.32 min of naked ODN
after i.v. injection in rats (14). The longer residence will provide
sufficient time for siRNA conjugates to access their target cells
and initiate ligand-receptor-mediated endocytosis. In Figure
6A,C, preaddition of an excessive amount of Gal-PEG or M6P-
PEG significantly inhibited the gene silencing effects of
corresponding siRNA conjugates, which indicated that the
cellular uptake of siRNA conjugates was ligand-receptor-
mediated endocytosis, not nonspecific endocytosis.
To verify the function of PEG-based siRNA conjugate, we
used TGF-ꢀ1 siRNA as the model sequence. The similar gene
silencing effects were observed when TGF-ꢀ1 siRNA was
conjugated to Gal-PEG-OPSS and M6P-PEG-OPSS (Figure
7A,B). On the basis of these results, we can predict that cation-
free Gal-PEG-siRNA and M6P-PEG-siRNA conjugates will be
(15) Rajur, S. B., Roth, C. M., Morgan, J. R., and Yarmush, M. L.
(1997) Covalent protein-oligonucleotide conjugates for efficient

Targeted Delivery of siRNA by Bioconjugation
Bioconjugate Chem., Vol. 21, No. 11, 2010 2127
delivery of antisense molecules. Bioconjugate Chem. 8, 935–
40.
(16) Nakagawa, O., Ming, X., Huang, L., and Juliano, R. L.
Targeted intracellular delivery of antisense oligonucleotides via
conjugation with small-molecule ligands. J. Am. Chem. Soc. 132,
8848–9.
(26) Muratovska, A., and Eccles, M. R. (2004) Conjugate for
efficient delivery of short interfering RNA (siRNA) into mam-
malian cells. FEBS Lett. 558, 63–8.
(27) Harrison, J. G., and Balasubramanian, S. (1998) Synthesis and
hybridization analysis of a small library of peptide-oligonucle-
otide conjugates. Nucleic Acids Res. 26, 3136–45.
(28) Chiu, Y. L., and Rana, T. M. (2003) siRNA function in RNAi:
a chemical modification analysis. RNA 9, 1034–48.
(29) Schwarz, D. S., Hutvagner, G., Du, T., Xu, Z., Aronin, N.,
and Zamore, P. D. (2003) Asymmetry in the assembly of the
RNAi enzyme complex. Cell 115, 199–208.
(30) Khvorova, A., Reynolds, A., and Jayasena, S. D. (2003)
Functional siRNAs and miRNAs exhibit strand bias. Cell 115,
209–16.
(31) Schwarz, D. S., Hutvagner, G., Haley, B., and Zamore, P. D.
(2002) Evidence that siRNAs function as guides, not primers, in
the Drosophila and human RNAi pathways. Mol. Cell 10, 537–
48.
(17) Nishina, K., Unno, T., Uno, Y., Kubodera, T., Kanouchi, T.,
Mizusawa, H., and Yokota, T. (2008) Efficient in vivo delivery
of siRNA to the liver by conjugation of alpha-tocopherol. Mol.
Ther. 16, 734–40.
(18) Healy, J. M., Lewis, S. D., Kurz, M., Boomer, R. M.,
Thompson, K. M., Wilson, C., and McCauley, T. G. (2004)
Pharmacokinetics and biodistribution of novel aptamer composi-
tions. Pharm. Res. 21, 2234–46.
(19) Kim, S. H., Jeong, J. H., Lee, S. H., Kim, S. W., and Park,
T. G. (2006) PEG conjugated VEGF siRNA for anti-angiogenic
gene therapy. J. Controlled Release 116, 123–9.
(20) Lee, S. H., Kim, S. H., and Park, T. G. (2007) Intracellular
siRNA delivery system using polyelectrolyte complex micelles
prepared from VEGF siRNA-PEG conjugate and cationic
fusogenic peptide. Biochem. Biophys. Res. Commun. 357,
511–6.
(21) Kim, S. H., Jeong, J. H., Lee, S. H., Kim, S. W., and Park,
T. G. (2008) Local and systemic delivery of VEGF siRNA using
polyelectrolyte complex micelles for effective treatment of
cancer. J. Controlled Release 129, 107–16.
(22) Mahato, R. I., Rolland, A., and Tomlinson, E. (1997) Cationic
lipid-based gene delivery systems: pharmaceutical perspectives.
Pharm. Res. 14, 853–9.
(23) Graham, M. J., Crooke, S. T., Monteith, D. K., Cooper, S. R.,
Lemonidis, K. M., Stecker, K. K., Martin, M. J., and Crooke,
R. M. (1998) In vivo distribution and metabolism of a phos-
phorothioate oligonucleotide within rat liver after intravenous
administration. J. Pharmacol. Exp. Ther. 286, 447–58.
(24) Santel, A., Aleku, M., Keil, O., Endruschat, J., Esche, V.,
Fisch, G., Dames, S., Loffler, K., Fechtner, M., Arnold, W.,
Giese, K., Klippel, A., and Kaufmann, J. (2006) A novel siRNA-
lipoplex technology for RNA interference in the mouse vascular
endothelium. Gene Ther. 13, 1222–34.
(25) Oishi, M., Nagasaki, Y., Itaka, K., Nishiyama, N., and
Kataoka, K. (2005) Lactosylated poly(ethylene glycol)-siRNA
conjugate through acid-labile beta-thiopropionate linkage to
construct pH-sensitive polyion complex micelles achieving
enhanced gene silencing in hepatoma cells. J. Am. Chem. Soc.
127, 1624–5.
(32) Jeong, J. H., Mok, H., Oh, Y. K., and Park, T. G. (2009) siRNA
conjugate delivery systems. Bioconjugate Chem. 20, 5–14.
(33) Rozema, D. B., Lewis, D. L., Wakefield, D. H., Wong, S. C.,
Klein, J. J., Roesch, P. L., Bertin, S. L., Reppen, T. W., Chu,
Q., Blokhin, A. V., Hagstrom, J. E., and Wolff, J. A. (2007)
Dynamic PolyConjugates for targeted in vivo delivery of siRNA
to hepatocytes. Proc. Natl. Acad. Sci. U.S.A. 104, 12982–7.
(34) Al-Abd, A. M., Lee, S. H., Kim, S. H., Cha, J. H., Park, T. G.,
Lee, S. J., and Kuh, H. J. (2009) Penetration and efficacy of
VEGF siRNA using polyelectrolyte complex micelles in a human
solid tumor model in-vitro. J. Controlled Release 137, 130–5.
(35) Derfus, A. M., Chen, A. A., Min, D. H., Ruoslahti, E., and
Bhatia, S. N. (2007) Targeted quantum dot conjugates for siRNA
delivery. Bioconjugate Chem. 18, 1391–6.
(36) Bauer, M., and Schuppan, D. (2001) TGFbeta1 in liver fibrosis:
time to change paradigms. FEBS Lett. 502, 1–3.
(37) Friedman, S. L. (2003) Liver fibrosis -- from bench to bedside.
J. Hepatol. 38 (Suppl 1), S38–53.
(38) Bataller, R., and Brenner, D. A. (2005) Liver fibrosis. J. Clin.
InVest. 115, 209–18.
(39) Turner, J. J., Jones, S. W., Moschos, S. A., Lindsay, M. A.,
and Gait, M. J. (2007) MALDI-TOF mass spectral analysis of
siRNA degradation in serum confirms an RNAse A-like activity.
Mol. Biosyst. 3, 43–50.
(40) Bahr, U., Aygun, H., and Karas, M. (2008) Detection and
relative quantification of siRNA double strands by MALDI mass
spectrometry. Anal. Chem. 80, 6280–5.
BC100346N