www.elsevier.com/locate/jconrel

GEN

ED

ELIV

ER

Y

Journal of Controlled Release 92 (2003) 383–394

Targeted gene delivery into HepG2 cells using complexes

containing DNA, cationized asialoorosomucoid and activated

cationic liposomes

Moganavelli Singh, Mario Ariatti*

Biochemistry, School of Biochemistry and Microbiology, Faculty of Science, University of Durban-Westville,

Private Bag X54001, Durban, South Africa

Received 14 May 2003; accepted 14 July 2003

Abstract

Unilamellar activated cationic liposomes containing 3h[N-(NV,NV-dimethylaminopropane)-carbamoyl] cholesterol, dioleoyl

phosphatidylethanolamine (DOPE) and the N-hydroxysuccinimide ester of cholesteryl hemisuccinate (4:5:1, molar ratio) have

been prepared and their DNA-binding capacity has been assessed in a gel retardation assay. Ternary complexes composed of

activated cationic liposomes, carbodiimide–cationized asialoorosomucoid (Me+AOM) and pRSVL plasmid DNA were

assembled for receptor-mediated DNA delivery into cells expressing the asialoglycoprotein receptor (ASGP-R). Binding of

complexes in which Me+AOM was replaced by fluoresceinated Me+AOM (FMe+AOM) to the human hepatocellular cell line

HepG2 at 4 jC was severely reduced by co-incubation with asialoorosomucoid (AOM). Moreover, assemblies containing

liposomes, pRSVL DNA and Me+AOM (8:1:4, w/w/w) promoted high levels of luciferase activity in this cell line (1.3� 107

relative light units/mg soluble cell protein). Assays conducted in the presence of a hundred-fold excess of the ligand AOM

afforded considerably lower levels of transfection (2.5� 105 relative light units/mg soluble cell protein). In contrast, the highest

level of luciferase activity achieved with liposome, pRSVL DNA, AOM complexes was only a quarter of the best levels obtained

with liposome, pRSVL DNA, Me+AOM assemblies. These findings strongly support the notion that complexes gain entry into

hepatocyte-derived cells by ASGP-R mediation and that they are potentially useful gene carriers to liver hepatocytes.

D 2003 Elsevier B.V. All rights reserved.

Keywords: Cationic; Liposome; Transfection; Asialoorosomucoid; HepG2

1. Introduction

The directed delivery of corrective DNA into

hepatocytes and hepatocyte-derived cells by receptor

mediation is of considerable importance in the devel-

0168-3659/$ - see front matter D 2003 Elsevier B.V. All rights reserved.

doi:10.1016/S0168-3659(03)00360-2

* Corresponding author. Tel.: +27-31-2044981; fax: +27-31-

2044942.

E-mail address: mariatti@pixie.udw.ac.za (M. Ariatti).

opment of gene therapy protocols for human diseases

and disorders of the liver. These include low-density

lipoprotein receptor deficiency, haemophilia and a1-

antitrypsin deficiency [1]. The asialoglycoprotein re-

ceptor (ASGP-R), which is abundantly and selectively

expressed in hepatocytes [2], is found predominantly

on the sinusoidal surface of parenchymal cells [3].

This lectin which has a high affinity for the galactose-

terminating triantennary N-linked heteroglycans of

GEN

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394384

asialoglycoproteins [4] is being actively investigated

for drug and gene targeting into hepatocytes. A human

hepatocellular carcinoma cell line, HepG2, which is

frequently used for in vitro assessment of gene deliv-

ery vehicles adopting this avenue of entry, displays

approximately 225000 ASGP-R per cell [5]. Thus,

Wu and Wu [6] first demonstrated ASGP-R-mediated

gene transfer into HepG2 cells and later into rat liver

in vivo [7] using a vehicle comprising the ligand

asialoorosomucoid (AOM) cross-linked to the poly-

cation poly-L-lysine onto which the pSV2CAT plas-

mid was electrostatically bound and compressed. The

specificity of the receptor for the terminal D-galactose

unit is achieved through hydrogen bonding of the

sugar 3- and 4-hydroxyl groups with receptor amide

and carboxylate side chains [3]. Therefore, the non-

immunogenic neo glycoprotein, galactosylated albu-

min [8], which has been linked to poly-L-lysine, binds

and delivers DNA into mouse parenchymal cells [9]

and HepG2 cells [10] through the ASGP-R. Simpler

delivery systems, for example those in which D-

galactose has been coupled to poly-L-lysine through

various spacers in the absence of complex carbohy-

drates and apoprotein, retain specificity for ASGP-R

in HepG2 cells [11,12], rat liver [13] and mouse liver

parenchymal cells [14,15].

However, cationic liposomes are the most com-

monly used non-viral DNA delivery agents for in

vitro applications; and although they display low

immunogenicity, several factors militate against their

more widespread use in vivo. These include the

formation of aggregates with serum proteins bearing

negative charges [3] and tissue targeting. Several

groups have nevertheless demonstrated that transfect-

ing complexes containing cationic liposomes can

indeed be targeted to parenchymal cells. Thus, Hara

et al. [16] have included palmitoylated asialofetuin in

a cationic liposome formulation which transfects

HepG2 cells by receptor mediation. In a related work,

we have reported that cationized AOM, cationic lip-

osomes and plasmid DNA afford complexes which

were capable of transfecting the same cell line by the

same mechanism [17]. The human a1-antitrypsin gene

has recently been introduced into mice using a com-

plex of asialofetuin, DOTAP containing cationic lip-

osomes and plasmids containing the hAAT gene under

the control of natural and CMV promoters [18].

Levels of gene expression were higher than those

achieved by vectors lacking asialofetuin. Bifunctional

cholesterol derivatives embodying a galactose unit for

targeting and an imino group for DNA binding have

been incorporated into liposomes which deliver DNA

to liver parenchymal cells by receptor mediation

[19,20]. Furthermore, transfection activity increased

as the spacer linking the cholesteryl and galacto

moieties became longer [19]. A monogalacto append-

age linked to a DNA-binding domain may not always

achieve a strong binding affinity for the ASGP-R. The

binding affinity has been shown to be very dependent

on the valency of galactose/N-acetylgalactosamine

and the spatial disposition of the galactose units

[21]. The concept of multivalency of carbohydrate

in the carbohydrate–protein interaction is therefore of

some importance in the design of ASGP-R-targeting

vehicles [22]. Recently, multivalent galactosyl com-

pounds carrying dendritic amino DNA-binding enti-

ties showed the following order of transfection

activity in HepG2 cells: Tri-Gal>Di-Gal>Mono-Gal

[22]. A similar trend was reported by Niidome et al.

[23] using multiantennary ligands containing varying

numbers of galactose units linked to a DNA-binding

cationic a-helical peptide in the human hepatoma cell

line HuH-7. In at least one case, a synthetic trianten-

nary gene delivery vehicle engaged in an interaction

of such high affinity with the ASGP-R that displace-

ment with asialofetuin was not possible [24], while a

tetraantennary-poly-L-lysine conjugate achieved lucif-

erase gene transfer into cultured hepatocytes with an

efficiency similar to that obtained with the natural

ligand, asialofetuin linked to poly-L-lysine [25]. Work

with modified low density lipoproteins (LDLs) carry-

ing cholesterol-anchored triantennary glycosides on

the hydrophilic surfaces of the lipoprotein particles

has revealed that with a 4-A spacer separating termi-

nal galactose moieties and branch points, LDLs are

targeted to the galactose/fucose receptor on Kupffer

cells [26], whereas a 20-A spacer ensures that the

LDLs are targeted to the asialoglycoprotein receptor

on parenchymal cells [27].

Given the multiplicity of factors affecting the

binding affinity of synthetic ligands for the ASGP-

R, the continued pursuit of natural ligands does not

seem inappropriate in the design of hepatocyte-spe-

cific DNA delivery systems. We report here on a

ternary vector comprising cationized AOM, N-hydrox-

ysuccinimide-activated cationic liposomes and the

Fig. 1. Schematic representation of interactions between activated

cationic liposomes, cationized asialoorosomucoid (Me+AOM) and

DNA (not to scale).

GEN

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394 385

plasmid pRSVL encoding the luciferase gene, which

assembles spontaneously through electrostatic attrac-

tion of its components but which has the capacity to

form limited stabilizing cross-links between the lipo-

some and asialoglycoprotein components during sub-

sequent maturation (Fig. 1). Complexes are relatively

non-toxic to HepG2 cells while high levels of lucif-

erase gene transfer and expression are achieved by a

mechanism that is abolished by the presence of

excess AOM.

2. Materials and methods

2.1. Chemicals and reagents

Dioleoyl-L-a-phosphatidylethanolamine (DOPE),

ethidium bromide, human orosomucoid, RPMI 1640

medium, insoluble neuraminidase from Clostidium

perfringens type IVA, cholesteryl hemisuccinate, N-

hydroxysuccinimide and NV,NV- dicyclohexylcarbodii-mide (DCC) were from Sigma (St. Louis, MO). N-

Ethyl-NV-(3-dimethylaminopropyl)carbodiimide

(ECDI), 2-[4-(2-hydroxyethyl)-1-piperazinyl] ethane-

sulfonic acid (HEPES) and fluorescein-iso-thiocya-

nate (FITC) were supplied by Merck (Darmstadt,

Germany). Agarose (ultra pure DNA grade) was from

BioRad (Richmond, CA). Trypsin–EDTA and peni-

cillin–streptomycin mixtures were purchased from

Wittaker Bioproducts (Walkerville, MD). The lucifer-

ase assay kit was from Promega (Madison, WI) and

plasmid pBR322 was from Boehringer Mannheim

(Mannheim, Germany). [3H] Methyl iodide (83 Ci/

mmol) was purchased from Amersham International

(Little Chalfont, UK).

Asialoorosomucoid (AOM) was prepared by desia-

lylation of orosomucoid according to Kawasaki and

Ashwell [28] using immobilized neuraminidase from

C. perfringens. 3h[N-(NV,NV-dimethylaminopropane)-

carbamoyl] cholesterol (Chol-T) was prepared as

described [17] by a method adapted from that for

the synthesis of DC-Chol [29]. ECDI was methylated

with methyl iodide to afford the quaternary N-ethyl-

NV-(3-trimethylpropylammonium) carbodiimide io-

dide Me+CDI by the method of Kopczynski and

Babior [30].

Ultrapure water (Milli-Q50) was used throughout

and all other reagents were of analytical grade.

2.2. Luciferase expression vector

A 6.2 k base pair expression vector containing a

Rous sarcoma virus long terminal repeat (RSV-LTR)

promoter-driven luciferase reporter gene from Photi-

nus pyralis with SV40 small T antigen and polyade-

nylation site, and ampicillin resistance gene (pRSVL)

constructed by de Wet et al. [31] was amplified in

Escherichia coli HB101 and purified using the

QIAGENR protocol.

2.3. Liposome preparation and electron microscopy

The N-hydroxysuccinimide ester of cholesteryl

hemisuccinate (0.4 Amol), Chol-T (1.6 Amol) and

DOPE (2 Amol) were dissolved in CHCl3 (1 ml).

The solutes were deposited as a thin film on the inner

wall of a test tube by evaporation of solvent in vacuo

at 20 jC (Buchii Rotavapor-R). The film was re-

hydrated overnight in sterile HEPES (20 mM, pH 7.5,

1 ml) containing NaCl (150 mM). The suspension was

vortexed briefly and sonicated in a bath sonicator

GEN

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394386

(Elma Transsonic T460/H) at 20 jC for 5 min to

afford liposome suspension, which was stored at 4 jC.Samples for transmission electron microscopy

(TEM) were prepared by mixing the liposome sus-

pension (50 Al) with 5% (w/v) bovine serum albumin

(100 Al), diluting with Tris–HCl (0.1 M, pH 7.2, 100

Al) followed by the addition of 25% (v/v) glutaralde-

hyde (50 Al). After 20 min, the resulting gel was diced

and transferred into vials containing OsO4. Samples

were stored in the dark for 24 h before stepwise

dehydration in increasing concentrations of ethanol

(70–100%). Gels were then placed successively in

propylene oxide (20 min), propylene oxide: Spurr’s

resin (1:1, v/v, 20 min) and Spurr’s resin (45 min).

Samples were embedded in beem capsules in vacuo

(60 jC, 48 h). The resultant blocks were sectioned

(Reichert–Jung ultracut microtome) and sections col-

lected on G-200 copper grids. Grids were stained with

uranyl acetate and lead citrate. Stained sections were

viewed in a Philips 301 electron microscope at 60 kV.

2.4. AOM modifications

2.4.1. Radiolabeling of AOM

To a solution of AOM (0.2 mg, 0.005 Amol) in

borate buffer pH 9.8 (200 Al) at 4 jC was added [3H]

methyl iodide (2 mCi, 0.024 Amol) in toluene (200

Al). The mixture was incubated in the dark at 4 jC for

96 h. Thereafter, the organic layer was removed and

the tritiated AOM was desalted on a Sephadex G-25

column (1� 20 cm). Fractions (1 ml) were collected

and samples (50 Al) were taken for liquid scintillation.

The fraction reflecting the highest specific activity

was reserved for liposome binding studies. Thereafter,

it was brought to a specific activity of 10000 dpm/Agby dilution with non-radioactive AOM.

2.4.2. Fluoresceinated and cationized AOM

(FMe+AOM)

AOM was cationized by treatment with Me+CDI

by a method adapted from the procedure reported by

Timkovich [32] for the modification of cytochrome C

with [14C] labeled ECDI and is described elsewhere

[17]. Fluoresceination of the derivatized AOM

(Me+AOM) was carried out by a standard procedure

[33] with some modification. Briefly, a solution of 50

mM NaHCO3 buffer (1 ml, pH 9.0) containing

Me+AOM (1 mg) and FITC (23 Ag) was incubated

overnight followed by exhaustive dialysis against

ultrapure water for 24 h.

2.5. Liposome–AOM interaction

Reaction mixtures (100 Al) containing tritiated

AOM (10 Ag, 0.25 nmol) and liposomes (0–100 Ag)in 20 mM HEPES, 150 mM NaCl (pH 7.5 or 8.5)

were incubated at 20 jC for 30 min whereupon they

were overlaid on a cushion of incubation buffer (3 ml)

and centrifuged at 100000� g for 30 min in a Beck-

man L-80 ultracentrifuge using the SW 50.1 rotor.

Liposome pellets were resuspended in incubation

buffer (200 Al) and assayed for radioactivity. Results

were processed to reflect the number of AOM mole-

cules associated with 350 nm vesicles.

2.6. Gel retardation assays

2.6.1. Liposome–DNA interactions

pBR322 plasmid DNA (0.5 Ag) was incubated withincreasing amounts of the liposome preparation (2–6

Ag) for 20 min in 20 mM HEPES, 150 mM NaCl (10

Al, pH 7.5) at 20 jC. Samples were then subjected to

electrophoresis on 1% agarose (40 V) in a buffer

containing 36 mM Tris–HCl, 30 mM sodium phos-

phate and 10 mM EDTA (pH 7.5) for 1 h. After

staining with ethidium bromide (1 Ag/ml), gels were

viewed in a CAMAG transilluminator at 300 nm.

2.6.2. Me+AOM and FMe+AOM interactions with

DNA

Me+AOM (2–8 Ag) and FMe+AOM (4–10 Ag)were separately incubated with pBR322 DNA (0.5

Ag) for 20 min in 20 mM HEPES, 150 mM NaCl (10

Al, pH 7.5). Samples were then subjected to agarose

gel electrophoresis and results viewed as described in

Section 2.6.1.

2.7. Maintenance and growth of HepG2 cells

HepG2 cells were propagated in 25 cm2 screw cap

flasks in RPMI 1640 medium (5 ml) supplemented

with penicillin G (100 U/ml), streptomycin (100 Ag/ml), insulin (1 Ag/ml), 20 mM HEPES (pH 7.5) and

heat-inactivated foetal calf serum (10% v/v). Cultures

were routinely trypsinized (0.25% w/v trypsin, 0.1%

w/v EDTA) and split 1:3 after 5 days.

GEN

ED

ELIV

ER

Y

Contr

2.8. Interaction of FMe+AOM-containing ternary

assemblies with HepG2 cells

Cells (1.2� 104) were seeded into sterile Tract

vials containing circular glass slides and complete

medium (0.5 ml). After 24 h at 37 jC, the medium

was removed, and after washing with phosphate

buffered saline (PBS), cells were bathed in PBS

(0.5 ml). Ternary assemblies were generated by

allowing a mixture of FMe+AOM (2 Ag) with lip-

osomes (5 Ag) to incubate at 20 jC for 30 min

followed by the addition of pRSVL plasmid DNA

(0.5 Ag). Complexes were matured for 1 h at 20 jC.After chilling to 4 jC, complexes were added to

cells at the same temperature. After 1 h, cells were

viewed on a Zeiss Axiophotk fluorescence micro-

scope and photographed. Some incubations were

conducted in the presence of a hundred-fold excess

AOM.

2.9. Assay for cell growth inhibition by ternary

assemblies

The toxicity of liposome-Me+AOM-DNA ternary

assemblies to the human hepatoblastoma cell line

HepG2 was determined in a growth inhibition assay

adapted from Schellekens and Stitz [34]. Briefly,

cells were seeded into a 24 well plate at a density

of 3� 104 cells per well in complete medium (0.5

ml). After a 24-h incubation at 37 jC, the medium

was replaced with serum-free medium (RPMI

1640 + insulin + antibiotics, 0.5 ml). Ternary com-

plexes were assembled by addition of a fixed quantity

of Me+AOM (2 Ag) to varying amounts of liposomes

(1–6 Ag) in 20 mM HEPES, 150 mM NaCl (pH 7.5)

in a final volume of 10 Al. After incubation for 30

min at 20 jC, pRSVL plasmid vector (0.5 Ag) wasintroduced into each mixture and solutions were

diluted to 20 Al with 20 mM HEPES, 150 mM NaCl

(pH 7.5). After a further hour, the final assemblies

were introduced to the cells. After 4 h at 37 jC, themedium was replaced with complete medium. Cells

were incubated for a further 48 h whereupon wells

were drained and cells washed twice with PBS and

stained with crystal violet as described [34]. Stain

was extracted into 2-methoxyethanol (0.5 ml/well)

over a 36-h period and intensities were measured at

550 nm.

M. Singh, M. Ariatti / Journal of

2.10. Gene transfer experiments

Transfecting complexes containing pRSVL vector

(0.5 Ag), Me+AOM (3 Ag) and varying amounts of

liposomes (0–4 Ag) were prepared in 20 mM

HEPES, 150 mM NaCl (pH 7.5, 25 Al) as described

in Section 2.9. Cells were seeded at a density of

4.5� 104 cells per well in a 24 well plate and

incubated in complete medium overnight. Cells were

washed (PBS) and bathed in serum-free medium (0.5

ml) and complexes were added. After 4 h at 37 jC,the medium was replaced by complete medium and

cells were incubated for a further 36 h. After washing

(PBS, 2� 0.5 ml), the cells were processed for

luciferase determination.

2.11. Evaluation of gene expression

Cells from transfection assays were treated with

lysis buffer (5 mM Tris–phosphate, pH 7.8; 0.4 mM

DTT; 0.4 mM 1,2-diamiocyclohexane-N,N,NV,NV-tet-raacetic acid; 2.5% glycerol and 0.2% Triton X-100)

and processed for luciferase activity using the Prom-

ega Luciferase Assay system according to the man-

ufacturer’s instructions. Luminescence was measured

on a Lumac Biocounter 1500 as relative light units

(RLU) emitted for 10 s. The protein content of

supernatants was determined by the Bradford method

[35].

olled Release 92 (2003) 383–394 387

3. Results and discussion

3.1. Liposomes and liposome–DNA complexes

Stable cationic liposomes were obtained by soni-

cation of a film containing Chol-T, DOPE and the N-

hydroxysuccinimide ester of cholesteryl hemisucci-

nate (4:5:1, molar ratio) in 20 mM HEPES, 150 mM

NaCl (pH 7.5). On examination of the preparation by

TEM (Fig. 2), vesicles were found to be unilamellar

and in the size range 200–500 nm. A liposome

preparation from equimolar amounts of Chol-T and

DOPE was shown to have similar characteristics [17].

The formation of complexes between pBR322 plas-

mid DNA and the cationic liposomes was demon-

strated in a gel retardation assay. Thus, liposome-

associated DNA migration in an electric field is

Fig. 2. Transmission electron micrograph of cationic liposomes

containing Chol-T, DOPE and the N-hydroxysuccinimide ester of

cholesteryl hemisuccinate (4:5:1, molar ratio). Bar = 200 nm.

GEN

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394388

retarded and very large, and electroneutral complexes

fail to enter the gel matrix. It is clear from Fig. 3a that

all the DNA is liposome-associated at a plasmid/

liposome ratio 1:10 (w/w). Assuming a mass per

charge ratio of 325 for DNA [36] and assuming that

Chol-T is fully protonated at pH 7.5, the lipoplex has

a positive/negative charge ratio of 2:1. Although this

would suggest that when the DNA is fully associated

with the cationic liposomes there is an excess of

liposome positive charges over DNA negative

charges, it has been suggested that this finding may

be attributed to the bulky nature of cationic liposomes

and superhelical DNA which may prevent close

contact of the two species [37]. Moreover, about half

of the positive charges which are on the interior

surface of the 4 nm thick bilayer will have limited

interaction with the DNA. A similar position has been

taken by Cao et al. [38] who state that interactions

between oligonucleotides and cationic liposomes in-

volve the binding of positive charges on the exterior

surface of liposomes with the negative charges of the

oligonucleotides.

Fig. 3. Band shift assay of DNA binding interactions. (a) Incubation

mixtures contained pBR322 plasmid DNA (0.5 Ag) and increasing

amounts of activated cationic liposomes (lanes 1–4; 2, 4, 5, 6 Ag,respectively). (b) Incubation mixtures contained pBR322 DNA (0.5

Ag) and increasing amounts of cationized AOM (lanes 1–4; 0, 2, 4,

8 Ag Me+AOM) or fluoresceinated and cationized AOM (lanes 5–

8; 4, 6, 8, 10 Ag FMe+AOM).

Fig. 4. Binding of fluoresceinated complexes containing liposomes,

pRSVL DNA and FMe+AOM: (10:1:4 weight ratio) to HepG2 cells

at 4 jC. Cells on glass slides and under coverslips were

photographed with 2-min exposures. Green images were digitized

and converted to grey scale. (a) Without and (b) with a 100-fold

excess of AOM.

GEN

ED

ELIV

ER

Y

Controlled Release 92 (2003) 383–394 389

3.2. Me+AOM– and FMe+AOM–DNA complexes

AOM was cationized using the water-soluble car-

bodiimide Me+CDI at a glycoprotein/carbodiimide

ratio of 1:1 800 (molar ratio) under conditions which

disfavour conjugative protein–protein coupling [39].

Under these conditions, the number of cationic N-

acylurea moieties per glycoprotein molecule was

shown to be 10 [40]. The formation of Me+AOM–

DNA complexes was demonstrated in a gel retarda-

tion assay. It is seen in Fig. 3b that large complexes

which barely enter the gel are formed at a glycopro-

tein/DNA ratio of 8:1 (w/w). At a ratio of 16:1,

however, large insoluble complexes are formed which

fail to enter the gel and float out of wells during

staining. Electroneutral complexes are therefore

achieved at a ratio higher than 8:1 or, greater than

550 molecules of Me+AOM per plasmid molecule. In

turn, this represents at least 15 positive charges per

Me+AOM molecule interacting with a corresponding

number of phosphodiester negative charges on the

plasmid DNA backbone. Fluoresceination of the cat-

ionized AOM was carried out under conditions that

permit the addition of two fluorescein groups per

apoprotein moiety. Derivatization takes place at amino

functions thereby reducing the number of positive

charges on the labeled glycoprotein. The small loss

of positive charge marginally reduced the DNA-bind-

ing capacity of FMe+AOM (Fig. 3b, lanes 3 and 6).

3.3. Ternary assemblies

In generating transfecting complexes, ratios of

constituent cationic liposomes, plasmid DNA and

cationized AOM were chosen to permit interaction of

both cationic species with the DNA. Hence, 0.5 Ag of

plasmid DNA was combined with 2–4 Ag liposomes

and 2 Ag Me+AOM. At these ratios, liposomes or

Me+AOM alone would only partially neutralize the

plasmid negative charges. Assuming (i) an average

cross-sectional area (a) of 0.6 nm2 for the lipid mol-

ecules (0.55 nm2 for DOPE, 0.4 for cholesterol and 0.7

for cationic lipids [41]), (ii) an average Mr of 630 for

the three component lipids (weighted according to

mole percentage composition) and (iii) an average

liposome diameter of 350 nm (Fig. 2), the number of

molecules (N) in a liposome may be calculated to be

1.3� 106 (8kr2/a). Moreover, the number of vesicles

M. Singh, M. Ariatti / Journal of

per Ag lipid (7.4� 108) may be obtained from the

equation: ((1�106/630)� 6� 1023)/(8kr2/a) [42].

Employing the Avogadro number once more, we

may calculate that the 6.2 k base pair pRSVL plasmid

contains 1.5� 1011 molecules/Ag and cationic

Me+AOM (Mr 40 000) 1.5� 1013 molecules/Ag.Therefore, a ternary assembly with a liposome/DNA/

Me+AOM composition of 8:1:4 (w/w/w) corresponds

to 25 molecules of plasmid DNA and 10000 mole-

cules of Me+AOM per liposome.

It has been previously shown [43] that the incor-

poration of the N-hydroxysuccinimide ester of choles-

teryl hemisuccinate permits the cross-linking of

liposomes to amino group-containing molecules (pu-

romycin) and proteins (peroxidase). It was therefore

GEN

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394390

reasoned that inclusion of this carboxyl group acti-

vated cholesterol derivative into the cationic liposome

bilayer would give added stability to ternary com-

plexes of cationic liposomes, plasmid DNA and

cationized AOM. The formation of such complexes

is initially driven by electrostatic attractions between

the two polycationic species and the polyanionic

nucleic acid. In a subsequent maturation process, a

number of cross-links may be formed between lip-

osomes and some abutting Me+AOM molecules al-

though the degree of cross-linking has not been

determined.

3.4. Binding of fluoresceinated tertiary assemblies to

HepG2 cells

When incubated at 4 jC with HepG2 cells,

complexes containing liposomes/pRSVL DNA/

FMe+AOM (10:1:4, w/w/w) concentrated on the exte-

rior layer of the membrane (Fig. 4a). The binding was,

however, severely reduced when incubation mix-

tures included an excess of AOM (Fig. 4b). This

finding strongly suggests that complexes are mem-

brane-associated through an interaction with the

ASGP-R. At 4 jC, ligand-receptor complexes will

Fig. 5. Growth inhibition of HepG2 cells by ternary complexes. Cells (3

complexes containing pRSVL DNA (0.5 Ag), Me+AOM (2 Ag) and varyin

was replaced by complete medium after 4 h and assays were conducted

(n= 4).

form without internalization [42]. Cationization and

fluoresceination of AOM under conditions described

in Section 2.4.2 followed by association with DNA

and cationic liposomes do not eliminate the capac-

ity of the asialoglycoprotein to bind to its cognate

receptor.

3.5. Growth inhibition assay

Complexes in which the Me+AOM/pRSVL DNA

ratio was fixed at 2 Ag:0.5 Ag and in which the

liposome content varied between 1 and 6 Ag were

shown to be relatively non-toxic to HepG2 cells

under transfection conditions (Fig. 5). In particular,

complexes and compositions chosen for transfection

studies inhibited cell growth in the range 17–33%.

A similar toxicity profile was reported for tertiary

complexes containing liposomes formulated with

Chol-T and DOPE alone [17].

3.6. Transfection of HepG2 cells

Gene transfer experiments were conducted with

assemblies containing 0–4 Ag liposomes, 0.5 AgpRSVL DNA and 2 Ag Me+AOM. Results presented

� 104) in 0.5 ml serum-free RPMI 1640 medium were exposed to

g amounts of activated liposomes (as indicated) at 37 jC. Medium

after a further 48-h incubation. Data are presented as meansF S.D.

GEN

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394 391

in Fig. 6a show a linear increase in transfection

activity measured in relative light units (RLU) per

mg soluble cell protein as the liposome content of

complexes increases from 2 to 4 Ag. The inclusion of

a 100-fold excess of AOM in incubation mixtures

markedly reduced transfection activities. At a lipo-

some/DNA/Me+AOM ratio of 4:0.5:2 Ag which

afforded the highest activity (1.3� 107 RLU/mg pro-

tein), inclusion of competing AOM reduced activity

50-fold (2.5� 105 RLU/mg protein). Complexes not

containing liposomes exhibited luciferase activity ap-

Fig. 6. Transfection of HepG2 cells. (a) Cells (4.5� 104/well) were incuba

Me+AOM (2 Ag) and increasing amounts of activated cationic liposomes

presence of 100-fold excess of AOM (shaded bars). (b) HepG2 cells (5�(0.5 Ag), AOM (3 Ag) and liposomes as shown (4–8 Ag). Data are presen

proximately 3 orders of magnitude lower than that

obtained with ternary complexes. This supports the

notion that transfecting complexes containing

Me+AOM gain entry into the HepG2 cells by

ASGP-R mediation.

The need to utilize cationized AOM for successful

gene transfer is illustrated by results presented in Fig.

6b. Thus, transfections carried out with mixtures

containing 4–8 Ag liposomes, 0.5 Ag pRSVL DNA

and 3 Ag AOM resulted in levels no greater than 25%

of the best values achieved with Me+AOM-containing

ted with transfecting assemblies containing pRSVL DNA (0.5 Ag),as indicated (0–4 Ag). Competition assays were conducted in the

104/well) were incubated with assemblies containing pRSVL DNA

ted as meansF S.D. (n= 4).

Fig. 7. Binding of AOM to activated cationic liposomes. Aliquots of tritiated AOM (10 Ag, 10000 dpm) were incubated with increasing

amounts of liposomes. Liposome-associated DNAwas determined by ultracentrifugation (100000� g). Incubations were carried out at pH 7.5

(.) and pH 8.5 (n).

GEN

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394392

ternary complexes. Since AOM shows no detectable

binding to plasmid DNA at levels up to 4 Ag AOM

per 0.5 Ag plasmid DNA [44], AOM present in the

latter complex is presumed to be largely liposome

associated. In an experiment carried out to determine

the degree of association between cationic liposomes

and AOM at pH 7.5, a maximum level of 60 AOM

molecules per vesicle was achieved (Fig. 7). As

expected, at a pH of 8.5, somewhat closer to the pkaof the head group of the cationic cholesterol derivative

(pka of trimethylamine 9.8 [45]), AOM binding to

liposomes was considerably reduced. The highest

ratio of liposome/AOM in transfecting complexes

(Fig. 6b) would therefore be expected to result in

approximately 30 molecules of AOM per 350 nm

vesicle, a value considerably lower than the estimated

number of Me+AOM molecules per liposome in

liposome: DNA/Me+AOM (8:1:4, w/w/w) complexes

(1�104).

4. Conclusions

Hepatocytes, which account for about 50% of the

cells in the liver and in which most plasma proteins

are synthesized, are important targets for gene therapy

approaches to the correction of several diseases and

disorders. Interaction of DNA-bearing vectors with

these epithelial cells should be facilitated by the large

number of fenestrated sinusoids in the liver. These

vectors may be tagged with ligands to promote

targeting to hepatocytes through specific interaction

with cognate receptors. We have constructed a hepa-

tocyte-specific vector directed to the plasma mem-

brane-located ASGP-R utilizing Me+AOM, a

cationized derivative of a natural ligand AOM and

activated cationic liposomes. Ternary non-covalent

complexes formed between these components and

plasmid DNA, which have the capacity to stabilize

further through formation of cross-links between

glycoproteins and liposomes, are well tolerated by

the hepatocyte-derived human cell line HepG2 which

is transfected demonstrably through ASGP-R media-

tion. Results obtained suggest that this system may be

a promising candidate for further investigation in

vivo.

Acknowledgements

The authors would like to thank the University of

Durban-Westville for the financial assistance, Dr. Y.

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394 393

Naidoo for assistance with the electron microscopy,

Mr. A. Rajh for his guidance in the use of the

fluorescence microscope and Mr. P. Govender and Mr.

D.B. Jagganath for the pRSVL vector.

GEN

References

[1] J. Wu, G.Y. Wu, M.A. Zern, The prospects of hepatic drug

delivery and gene therapy, Expert Opin. Investig. Drugs 7

(1998) 1795–1817.

[2] T. Kawasaki, G. Ashwell, Chemical and physical properties of

an hepatic membrane protein that specifically binds asialogly-

coproteins, J. Biol. Chem. 251 (1976) 1296–1302.

[3] J. Wu, M.H. Nantz, M.A. Zern, Targeting hepatocytes for drug

and gene delivery: emerging novel approaches and applica-

tions, Front. Biosci. 7 (2002) 717–725.

[4] M.C. Bilder, R. Cescato, P. Neno, M. Spiess, High affinity

ligand to subunit H1 of the asialoglycoprotein receptor in the

absence of subunit H2, Eur. J. Biochem. 230 (1995) 207–212.

[5] A.L. Schwartz, H.J. Genze, H.F. Lodish, Recycling of the

asialoglycoprotein receptor: biochemical and immunocyto-

chemical evidence, Philos. Trans. R. Soc. Lond., B Biol. Sci.

300 (1982) 229–235.

[6] G.Y. Wu, C.H. Wu, Receptor-mediated in vitro gene trans-

formation by a soluble DNA carrier system, J. Biol. Chem.

262 (1987) 4429–4432.

[7] G.Y. Wu, C.H. Wu, Receptor-mediated gene delivery and ex-

pression in vivo, J. Biol. Chem. 263 (1988) 14621–14624.

[8] L. Fiume, A. Mattioli, C. Busi, G. Spinosa, T. Wieland, Con-

jugates of 9-alpha-D-arabinofuranoside monophosphate (ara-

AMP) with lactosaminated homologous albumin are not im-

munogenic in the mouse, Experientia 38 (1982) 1087–1089.

[9] M. Nishikawa, Y. Ohtsubo, J. Ohno, T. Fujita, Y. Koyama, F.

Yamashita, M. Hashida, H. Sezaki, Pharmacokinetics of re-

ceptor-mediated hepatic uptake of glycosylated albumin in

mice, Int. J. Pharm. 85 (1992) 75–85.

[10] J. Han, M. Lim, Y.I. Yeom, Receptor-mediated gene transfer

to cells of hepatic origin by galactosylated albumin–polyly-

sine complexes, Biol. Pharm. Bull. 22 (1999) 836–840.

[11] P. Midoux, C. Mendes, A. Legrand, J. Raimond, R. Mayer, M.

Monsigny, A.C. Roche, Specific gene transfer mediated by

lactosylated poly-L-lysine into hepatoma cells, Nucleic Acids

Res. 21 (1993) 871–878.

[12] J. Han, Y.I. Yeom, Specific gene transfer mediated by galac-

tosylated poly-L-lysine into hepatoma cells, Int. J. Pharm. 202

(2002) 151–160.

[13] J.C. Perales, T. Ferkol, H. Bergen, O.D. Ratnoff, R.H. Hanson,

Gene transfer in vivo: sustained expression and regulation of

genes introduced into the liver by receptor-mediated uptake,

Proc. Natl. Acad. Sci. 91 (1994) 4086–4090.

[14] M. Hashida, S. Takemura, M. Nishikawa, Y. Takahura, Tar-

geted delivery of plasmid DNA complexed with galactosy-

lated poly (L-lysine), J. Control. Release 53 (1998) 301–310.

[15] M. Nishikawa, S. Takemura, Y. Takakura, M. Hashida, Tar-

geted delivery of plasmid DNA to hepatocytes in vivo: Op-

timization of the pharmacokinetics of plasmid DNA/

galactosylated poly (L-lysine) complexes by controlling their

physicochemical properties, J. Pharmacol. Exp. Ther. 287

(1998) 408–415.

[16] T. Hara, Y. Aramaki, S. Takada, K. Koike, S. Tsuchiya, Re-

ceptor-mediated transfer of pSV2CAT DNA to a human hep-

atoblastoma cell line HepG2 using asialofetuin-labeled

cationic liposomes, Gene 159 (1995) 167–174.

[17] M. Singh, N. Kisoon, M. Ariatti, Receptor-mediated gene

delivery to HepG2 cells by ternary assemblies containing cat-

ionic liposomes and cationized asialoorosomucoid, Drug De-

liv. 8 (2001) 29–34.

[18] F. Dasi, M. Benet, J. Crespo, A. Crespo, S.F. Alino, Asialo-

fetuin liposome-mediated human a1-antitrypsin gene transfer

in vivo results in stationary long-term gene expression, J. Mol.

Med. 79 (2001) 205–212.

[19] S. Kawakami, F. Yamashita, M. Nishikawa, Y. Takakura, M.

Hashida, Asialoglycoprotein receptor-mediated gene transfer

using novel galactosylated cationic liposomes, Biochem. Bio-

phys. Res. Commun. 252 (1998) 78–83.

[20] S. Kawakami, S. Fumoto, M. Nishikawa, F. Yamashita, M.

Hashida, In vivo gene delivery to the liver using novel

galactosylated cationic liposomes, Pharm. Res. 17 (2000)

306–313.

[21] Y.C. Lee, Biochemistry of carbohydrate–protein interaction,

FASEB 6 (1992) 3193–3200.

[22] T. Ren, G. Zhang, D. Liu, Synthesis of galactosyl compounds

for targeted gene delivery, Bioorg. Med. Chem. 9 (2001)

2969–2978.

[23] T. Niidome, M. Urakawa, H. Sato, Y. Takahara, T. Anai, T.

Hatakayama, A. Wada, T. Hirayama, H. Aoyagi, Gene transfer

into hepatoma cells mediated by galactose-modified a-helical

peptides, Biomaterials 21 (2000) 1811–1819.

[24] J.-S. Remy, A. Kichler, V. Mordvinov, F. Schuber, J.-P. Behr,

Targeted gene transfer into hepatoma cells with lipopoly-

amine-condensed DNA particles presenting galactose ligands:

a stage towards artificial viruses, Proc. Natl. Acad. Sci. 92

(1995) 1744–1748.

[25] C. Plank, K. Zatloukal, M. Cotten, K. Mechtler, E. Wagner,

Gene transfer into hepatocytes using asialoglycoprotein recep-

tor mediated endocytosis of DNA complexed with an artificial

tetra-antennary galactose ligand, Bioconjug. Chem. 3 (1992)

533–539.

[26] T.J.C. van Berkel, J.K. Kruijt, H.H. Spanjer, J.F. Nagelkerke,

L. Harkes, H.-J.M. Kempen, The effect of a water-soluble

tris–galactoside-terminated cholesterol derivative on the fate

of low density lipoproteins and liposomes, J. Biol. Chem. 260

(1985) 2694–2699.

[27] E.A.L. Biessen, H. Veitsch, T.J.C. van Berkel, Cholesterol

derivative of a new triantennary cluster galactoside directs

low- and high-density lipoproteins to the parenchymal liver

cell, Biochem. J. 302 (1994) 283–289.

[28] T. Kawasaki, G. Ashwell, Isolation and characterization of an

avian hepatic binding protein specific for N-acetylglucos-

amine-terminated glycoproteins, J. Biol. Chem. 252 (1997)

6536–6543.

[29] X. Gao, L. Huang, A novel cationic liposome reagent for

GEN

ED

ELIV

ER

Y

M. Singh, M. Ariatti / Journal of Controlled Release 92 (2003) 383–394394

efficient transfection of mammalian cells, Biochem. Biophys.

Res. Commun. 179 (1991) 280–285.

[30] M.G. Kopczynski, B.M. Babior, Mechanism of action of etha-

nolamine ammonialyase, an adenosylcobalamin-dependent

enzyme: evidence for a carboxyl at the active site, J. Biol.

Chem. 259 (1984) 7652–7654.

[31] J.R. de Wet, K.V. Wood, M. DeLuca, D.R. Helinski, S. Sub-

ramani, Firefly luciferase gene: structure and expression in

mammalian cells, Mol. Cell. Biol. 7 (1987) 725–737.

[32] R. Timkovich, Detection of the stable addition of carbodii-

mide to proteins, Anal. Biochem. 79 (1977) 135–143.

[33] R.D. Nargessi, J. Landon, Indirect quenching fluoro immuno-

assay, Methods Enzymol. 74PtC (1981) 60–79.

[34] H. Schellekens, L.W. Stitz, Simple method for measuring

growth inhibition by interferon of cells in monolayer, J. Virol.

Methods 1 (1980) 197–200.

[35] M.M. Bradford, A rapid and sensitive method for the quanti-

tation of microgram quantities of protein utilizing the principle

of protein-dye binding, Anal. Biochem. 72 (1976) 248–254.

[36] L. Dekie, V. Toncheva, P. Dubruel, E.H. Schacht, L. Barrett,

W. Seymour, Poly-L-glutamic acid derivatives as vectors for

gene therapy, J. Control. Release 65 (2000) 187–202.

[37] H. Farhood, R. Bottega, R.M. Epand, L. Huang, Effect of

cationic cholesterol derivatives on gene transfer and protein

kinase C activity, Biochim. Biophys. Acta 1111 (1992)

239–246.

[38] A. Cao, D. Braine, R. Coudert, J. Vassy, N. Lievre, E. Olsman,

E. Tamboise, J.L. Salzmann, J.P. Rigaut, E. Taillandier, Deliv-

ery and pathway in MCF7 cells of DNAvectorized by cationic

liposomes derived from cholesterol, Antisense Nucleic Acid

Drug Dev. 10 (2000) 369–380.

[39] K.W. Carraway, D.E. Koshland, Carbodiimide modification of

proteins, Methods Enzymol. 25 (1972) 616–623.

[40] P. Govender, M. Ariatti, Interaction of N-acylurea modified

asialoorosomucoid with plasmid DNA and the transfection of

HepG2 cells, Med. Sci. Res. 27 (1999) 831–833.

[41] D.D. Lasic, Liposomes in gene delivery, CRC Press, Boca

Raton, FL, 1997.

[42] H. Schreier, P. Moran, I.W. Caras, Targeting of liposomes to

cells expressing CD4 using glycosylphosphatidylinositol-

anchored gp120, J. Biol. Chem. 269 (1994) 9090–9098.

[43] G.M. Radford, M. Ariatti, A.O. Hawtrey, Cholesterylsuccinyl-

N-hydroxysuccinimide as a cross-linking agent for the attach-

ment of protein to liposomes, Biochem. Pharmacol. 41 (1990)

307–309.

[44] N.N. Moodley, An investigation on the preparation and inter-

action of glycoprotein–DNA conjugates with eukaryotic cell

receptors, MSc Dissertation, University of Durban-Westville,

1986.

[45] P. Sykes, A guidebook to mechanism in organic chemistry,

3rd ed., Longman Group, London, 1970.