Effect of N-acetylgalactosamine ligand valency on targeting dendrimers to hepatic cancer cells

Article abstract of DOI:10.1016/j.ijpharm.2018.04.028

The display of N-acetylgalactosamine (NAcGal) ligands has shown great potential in improving the targeting of various therapeutic molecules to hepatocellular carcinoma (HCC), a severe disease whose clinical treatment is severely hindered by limitations in delivery of therapeutic cargo. We previously used the display of NAcGal on generation 5 (G5) polyamidoamine (PAMAM) dendrimers connected through a poly(ethylene glycol) (PEG) brush (i.e. G5-cPEG-NAcGal; monoGal) to effectively target hepatic cancer cells and deliver a loaded therapeutic cargo. In this study, we were interested to see if tri-valent NAcGal ligands (i.e. NAcGal₃) displayed on G5 dendrimers (i.e. G5-cPEG-NAcGal₃; triGal) could improve their ability to target hepatic cancer cells compared to their monoGal counterparts. We therefore synthesized a library of triGal particles, with either 2, 4, 6, 8, 11, or 14 targeting branches (i.e. cPEG-NAcGal₃) attached. Conventional flow cytometry studies showed that all particle formulations can label hepatic cancer cells in a concentration-dependent manner, reaching 90–100% of cells labeled at either 285 or 570 nM G5, but interestingly, monoGal labeled more cells at lower concentrations. To elucidate the difference in internalization of monoGal versus triGal conjugates, we turned to multi-spectral imaging flow cytometry and quantified the amount of internalized (I) versus surface-bound (I⁰) conjugates to determine the ratio of internalization (I/I⁰) in all treatment groups. Results show that regardless of NAcGal valency, or the density of targeting branches, all particles achieve full internalization and diffuse localization throughout the cell (I/I⁰ ～ 3.0 for all particle compositions). This indicates that while tri-valent NAcGal is a promising technique for targeting nanoparticles to hepatic cancer cells, mono-valent NAcGal is more efficient, contrary to what is observed with small molecules.

Full text of DOI:10.1016/j.ijpharm.2018.04.028

Accepted Manuscript
Effect of N-acetylgalactosamine ligand valency on targeting dendrimers to hep-
atic cancer cells
Sibu P. Kuruvilla, Gopinath Tiruchinapally, Neha Kaushal, Mohamed E.H.
ElSayed
PII:
S0378-5173(18)30245-X
https://doi.org/10.1016/j.ijpharm.2018.04.028
IJP 17435
DOI:
Reference:
To appear in:
International Journal of Pharmaceutics
Received Date:
Revised Date:
Accepted Date:
15 May 2017
28 February 2018
13 April 2018
Please cite this article as: S.P. Kuruvilla, G. Tiruchinapally, N. Kaushal, M.E.H. ElSayed, Effect of N-
acetylgalactosamine ligand valency on targeting dendrimers to hepatic cancer cells, International Journal of
Pharmaceutics (2018), doi: https://doi.org/10.1016/j.ijpharm.2018.04.028
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Effect of N-acetylgalactosamine ligand valency on targeting dendrimers to
hepatic cancer cells
1
¶
2¶*
2
2,3
Sibu P. Kuruvilla , Gopinath Tiruchinapally , Neha Kaushal , Mohamed E.H. ElSayed
1
Department of Materials Science and Engineering, University of Michigan, Ann Arbor,
Michigan, United States of America
2
Department of Biomedical Engineering, University of Michigan, Ann Arbor, Michigan, United
States of America
3
Program of Macromolecular Science and Engineering, University of Michigan, Ann Arbor,
Michigan, United States of America
¶
These authors contributed equally to this work.
*
Corresponding Author:
Gopinath Tiruchinapally, Ph.D.
University of Michigan
Department of Biomedical Engineering
1
101 Beal Avenue, 2146 Lurie Biomedical Engineering Building
Ann Arbor, MI 48109, USA
Ph. No. 734-647-2868
E-mail: gtiruchi@umich.edu
1

Abstract
The display of N-acetylgalactosamine (NAcGal) ligands has shown great potential in improving
the targeting of various therapeutic molecules to hepatocellular carcinoma (HCC), a severe
disease whose clinical treatment is severely hindered by limitations in delivery of therapeutic
cargo. We previously used the display of NAcGal on generation 5 (G5) polyamidoamine
(
PAMAM) dendrimers connected through a poly(ethylene glycol) (PEG) brush (i.e. G5-cPEG-
NAcGal; monoGal) to effectively target hepatic cancer cells and deliver a loaded therapeutic
cargo. In this study, we were interested to see if tri-valent NAcGal ligands (i.e. NAcGal
displayed on G5 dendrimers (i.e. G5-cPEG-NAcGal ; triGal) could improve their ability to target
hepatic cancer cells compared to their monoGal counterparts. We therefore synthesized a library
of triGal particles, with either 2, 4, 6, 8, 11, or 14 targeting branches (i.e. cPEG-NAcGal
3
)
3
3
)
attached. Conventional flow cytometry studies showed that all particle formulations can label
hepatic cancer cells in a concentration-dependent manner, reaching 90-100% of cells labeled at
either 285 or 570 nM G5, but interestingly, monoGal labeled more cells at lower concentrations.
To elucidate the difference in internalization of monoGal versus triGal conjugates, we turned to
multi-spectral imaging flow cytometry and quantified the amount of internalized (I) versus
0
0
surface-bound (I ) conjugates to determine the ratio of internalization (I/I ) in all treatment
groups. Results show that regardless of NAcGal valency, or the density of targeting branches, all
0
particles achieve full internalization and diffuse localization throughout the cell (I/I ~ 3.0 for all
particle compositions). This indicates that while tri-valent NAcGal is a promising technique for
targeting nanoparticles to hepatic cancer cells, mono-valent NAcGal is more efficient, contrary
to what is observed with small molecules.
2

Keywords
N-acetylgalactosamine, drug delivery, hepatic cancer cells, poly(amidoamine) (PAMAM)
dendrimers, asialoglycoprotein receptor, targeting ligand, monoGal and triGal conjugates
Introduction
Hepatocellular carcinoma (HCC) is the fifth-most commonly occurring tumor worldwide and is
nd
the 2 highest cause for cancer-related deaths. Current treatment procedures involving the
delivery of chemotherapy and other small molecule therapies suffer from minimal efficacy and
high systemic toxicity due to the lack of targeted drug delivery. Nanotechnology has shown great
promise recently to overcome the delivery limitations to localize therapeutic molecules within
hepatic cancer tissue. Nanoparticles (NPs) such as synthetic polymers (Kallinteri et al., 2005;
Kang et al., 2011; Yoo et al., 2000), natural or metallic materials (Liu et al., 2014), and silica (Li
et al., 2010) have all been used to improve the delivery of a variety of payloads to hepatic cancer
cells both in vitro and in vivo, such as siRNA (Wang et al., 2013), imaging dyes (Cao et al.,
2
2
015), and small molecule drugs (Kuruvilla et al., 2017; Liu et al., 2014; Scott H Medina et al.,
013). The size characteristics of NPs allow them to passively target tumor tissue by exploiting
the enhanced permeation and retention (EPR) effect (Bertrand et al., 2014; Duncan et al., 2013;
Fang et al., 2011). Once inside tumor tissue, active targeting of specific molecules improves
cellular trafficking of NPs (Arias, 2011; Bertrand et al., 2014).
Targeting the asialogycoprotein receptor (ASGPR), which is specifically overexpressed on
hepatic cancer cells (Li et al., 2008; Liu et al., 2014; Trerè et al., 1999) has shown great promise
due to its high binding affinity to glycoproteins, which can be synthetically immobilized on a NP
3

surface to promote highly-efficient binding. We (Kuruvilla et al., 2017; Medina et al., 2011;
Scott H. Medina et al., 2013) and others (Lee and Lee, 1997; Rensen et al., 2004; Westerlind et
al., 2004) have shown that the display of N-acetylgalactosamine (NAcGal) ligands on a NP
surface achieves selective internalization into hepatic cancer cells. We showed that mono-valent
NAcGal ligands in the  conformation displayed on the tip of a PEG brush attached to generation
5
(G5) poly(amidoamine) (PAMAM) dendrimers (i.e. G5-PEG-NAcGal particles) achieve
controllable targeting of hepatic cancer cells (Scott H. Medina et al., 2013). The display of 12-16
moles of PEG-NAcGal branches on the G5 surface (i.e. G5-(PEG-NAcGal)12-16) enabled
efficient delivery of co-loaded drug molecules into the cytoplasm of hepatic cancer cells,
improving the therapeutic efficacy of the drug (Kuruvilla et al., 2017).
Interestingly, many studies have shown that the display of multi-valent NAcGal ligands,
particularly tri-valent NAcGal (i.e. NAcGal
in comparison to mono-valent NAcGal (Khorev et al., 2008; Lee et al., 2011; Nair et al., 2014) in
small to medium molecular weight delivery systems. Accordingly, NAcGal -targeting has
successfully been used to deliver molecules like siRNA (Nair et al., 2014) and imaging dyes
Khorev et al., 2008) to hepatic cancer cells, achieving higher intracellular concentrations either
at lower delivered concentrations of the therapeutic agent or with improved internalization
kinetics. To the best of our knowledge, the efficacy of NAcGal targeting has only been studied
on small molecules (<13 kDa) and has yet to be investigated for larger molecules such as
nanoparticles (>30 kDa). It is important to identify whether the display of NAcGal ligands on
3
), improves the ability to target and bind the ASGPR
3
(
3
3
nanoparticle surfaces can improve their distribution to hepatic cancer cells over mono-valent
targeting, similar to what is observed with small molecules.
4

In this study, we synthesized G5 dendrimers targeted by NAcGal
3
ligands attached to the surface
through a PEG brush (i.e. G5-[PEG-(NAcGal ; triGal) and measured their ability to target

)
]
3 n
hepatic cancer cells in comparison to mono-valent G5-(PEG-NAcGal)12.1 conjugates (Fig 1). We
synthesized a library of triGal conjugates with varying density of targeting branches attached,
namely with n=2, 4, 6, 8, 11, or 14 moles of PEG-(NAcGal
, T11, and T14 conjugates, respectively. We compared the internalization of these particles to
that of mono-valent G5-(PEG-NAcGal 12.1 (monoGal; M12) conjugates via conventional and
multi-spectral imaging flow cytometry methods. Results from this study are useful to understand
whether NAcGal -targeted dendrimers are a viable option to improve nanoparticle-mediated
delivery of therapeutic agents to hepatic cancer tissue.

)
3
branches, to achieve T
2
, T
4
6
, T ,
T
8

)
3
5

Materials and Methods
Materials
G5-(NH )128 dendrimers with a diaminobutane core were purchased from Andrews
2
ChemServices (Berrien Springs, MI) and purified by dialysis against deionized water using
Slide-A-Lyzer dialysis cassettes (MWCO 10 kDa, Thermo Fisher Scientific, Rockford, IL) to
remove imperfect dendrimers and debris. N-acetylgalactosamine, pyridine, trimethylphosphine
solution (1.0 M in THF), triethylamine (TEA), acetic anhydride (Ac O), 1-ethyl-3-(3-
2
dimethylaminopropyl) carbodiimide hydrochloric acid (EDC.HCl), benzotriazol-1-ol (HOBt),
trifluoroacetic acid (TFA), anhydrous dimethylsulfoxide (DMSO), anhydrous dichloromethane
(
DCM), anhydrous dimethylformamide (DMF), anhydrous tetrahydrofuran (THF), anhydrous
1
1
,4-dioxane, cis-aconitic anhydride (cis-Ac), alpha bromoacetic acid, sodium hydroxide (NaOH),
0% palladium on activated Carbon (Pd-C), fluorescein isothiocyanate (FITC) sodium
methoxide (1.0 M NaOMe solution) and bovine serum albumin (BSA) were purchased from
Sigma-Aldrich Inc. (St. Louis, MO). Trimethylsilyl trifluoromethanesulfonate (TMSOTf), N,N-
diisopropyl ethyl amine (DIPEA), camphor sulphonic acid (CSA), sodium azide (NaN ), N,N'-
3
dicyclohexylcarbodiimide (DCC), ethylacetate (EtOAc), ethanol (EtOH) were purchased from
Across Organics Chemicals (Geel, Belgium). N-hydroxysuccinimide-poly(ethylene glycol)-Boc
(
2 kDa) was purchased from JenKem Technology USA Inc (Plano, TX). 2-{2-(2-
Chloroethoxy)ethoxy}ethanol was purchased from TCI America (Portland, OR). Dialysis
cassettes (MWCO 1–10 kDa) were purchased from Thermo Fisher Scientific (Rockford, IL).
Minimum essential medium (MEM), OPTI-MEM reduced serum medium, fetal bovine serum
(
FBS), 0.25% trypsin/0.20% ethylenediaminetetraacetic acid (EDTA) solution, phosphate
6

buffered saline (PBS), penicillin/streptomycin/amphotericin solution, sodium pyruvate,
minimum non-essential amino acid (NEAA) solution, and 0.4% trypan blue solutions were
purchased from Life Technologies (Thermo Fisher Scientific, Rockford, IL).
Spectral analysis of conjugates
Complete NMR and time-of-flight matrix-assisted laser desorption/ionization (MALDI-TOF)
spectra confirming the structural identity and composition of FI
conjugates can be found in the Supporting Information. Control particles [(FITC)
mono-valent FI -G5-[PEG-NAcGal]12.1 (M12) were synthesized according to our established
protocols (Medina et al., 2011; Scott H. Medina et al., 2013). We also synthesized mono-valent
FI -G5-[PEG-NAcGal]12.2 conjugates with a lysine spacer (M12-L) between NAcGal and the
6
-G5-cPEG-(NAcGal
β
)
3
(T -T14)
2
6
-G5] and
6
6
PEG branch as a control particle to measure the effect of the spacing structure on cancer cell
uptake (Fig S40).
Synthesis of FI -G5-[cPEG-(NAcGalβ)3]y
6
We chose a similar approach to our previously published strategies to synthesize PEGylated,
(
NAcGal
β
)
3
-targeted G5 conjugates (Fig 2) (Medina et al., 2011; Scott H Medina et al., 2013).
O and pyridine to obtain D-
Briefly, D-N-acetylgalactosamine was treated with Ac
2
galactosepentaacetate (1), which was treated with TMSOTf in DCM to obtain an oxazolidine
derivative (compound 2). Commercially available 2-(2-(2-chloroethoxy)ethoxy)ethan-1-ol was
treated with NaN
3
in DMF to obtain compound 3. The oxazolidine derivative compound 2, was
7

reacted with an alcohol, 2-(2-(2-azidoethoxy)ethoxy)ethan-1-ol (compound 3) in the presence of
o
D-10-CSA in DMSO at 40 C to yield compound 4 having an azide group at the terminal end.
The azide functional group of compound 4 was reduced to an amine with Me P in THF to obtain
3
compound 5. Commercially available N6-carbobenzyloxy-L-Lysine was treated with α-
º
bromoacetic acid and NaOH in water at 50 C to obtain compound 6. The peptide coupling
between triacid 6 and the D-galactosamine amine (5) was facilitated by DCC, HOBt, and DIPEA
in DCM:DMF to obtain a N6-carbobenzyloxy-L-Lysine-(NAcGal
β
)
3
derivative (7) having
on the other end. The carbobenzyloxy
Cbz) group was deprotected by hydrogenolysis under 10% Pd on activated carbon in
-L-Lysine-(NAcGal (8), which was reacted
with a hetero-functional PEG derivative, (BocNH-PEG-COONHS) in the presence of EDC.HCl,
HOBt, and DIPEA in DMF at room temperature to obtain (NAcGal -L-Lysine-6-NH-PEG-
NHBoc (9). Acid hydrolysis of compound 9 with TFA:DCM created a Boc-deprotected material
NAcGal -L-Lysine-6-NH-PEG-NH (10), which after reaction with cis-aconitic anhydride in
O:1,4 dioxane mixture gave a corresponding acid ((NAcGal -L-Lysine-6-NH-PEG-NHc-
(
(
NAcGal
β
)
3
group at one end and Cbz-protected NH
2
EtOH/EtOAc at room temperature to obtain 6-NH
2
)
β 3
)
β 3
(
β
)
3
2
H
2
)
β 3
acid; 11). These acid functional groups were created to help in coupling them to G5-amine
dendrimers. We fluorescently-labeled the G5 dendrimer with fluorescein isothiocyanate (FITC)
by treating commercially available G5-(NH
obtain compound 12 (Fig 3). Compound 12 was reacted with different equivalents of
(NAcGal -L-Lysine-6-NH-PEG-NHc-acid (11) to obtain a library of conjugates with different
2
)
128 dendrimers with FITC in H
2
O:1,4-Dioxane to
(
)
β 3
targeting ligand concentration on the G5 dendrimer (compounds 13-18). These coupling
reactions were carried out under EDC and HOBt reagents in 6.0 pH phosphate buffer solution.
The conjugates were individually reacted with Ac
2
O in pyridine to convert the G5 amines into
8

N-acetyl amines. The materials were then purified by dialysis (10kDa MWCO) and lyophilized
to obtain acetylated G5 particles which were further treated with NaOMe in methanol to
deprotect the O-acetate groups from galactosamine moieties. The reaction mixture was purified
by dialysis against deionized water (10kDa MWCO) for 2 days and lyophilized to obtain pure
T
2
: (FITC)
NAcGal
NH-Lysine-(NAcGal
FITC) -G5-[cPEG-6-NH-Lysine-(NAcGal
loading to the nearest whole number, and therefore refer to the particles as having either 2, 4, 6,
, 11, or 14 cPEG-(NAcGal branches attached.
6
-G5-[cPEG-6-NH-Lysine-(NAcGal
: (FITC) -G5-[cPEG-6-NH-Lysine-(NAcGal
-G5-[cPEG-6-NH-Lysine-(NAcGal
14.2. For clarity purposes, we rounded branch
β
)
3
]
2
; T
4
: (FITC)
6
-G5-[cPEG-6-NH-Lysine-
(
β
)
3
]
3.6; T
6
6
β
)
3
]
5.8; T
8
: (FITC) -G5-[cPEG-6-
6
β
)
3
]
8.1; T11: (FITC)
6
β
3
) ]10.6 and T14:
(
6
β
3
) ]
8
 3
)
Characterization of triGal conjugates
We measured the particle size of the nanoparticle formulations by dynamic light scattering
(
DLS) using a 90Plus particle size analyzer (Brookhaven Instruments, Holtsville, NY). The
nanoparticles (0.2 mg) were dissolved in 1 mL DI H 0 and tip sonicated using Q500 sonicator
for 30 sec. T -T14 conjugate solutions were then sterile-filtered through syringe filters with a pore
2
2
size of 0.45 µm and warmed to ꢀꢁbefore measurements. Raw distribution data was plotted in
Graphpad Prism software and fit using a Gaussian curve, with the mean being taken as the
particle size for that replicate. The average of three separate replicates was taken to find the
mean particle size ± standard error of the mean (SEM). We also determined the zeta potential of
the conjugates using a 90Plus Zeta Potential Analyzer (Brookhaven Instruments, Holtsville, NY).
9

Particle formulations were dissolved in DI water at 1:20 v/v and warmed to ꢀꢁbefore analysis.
The average of three separate replicates was taken to find the mean zeta potential ± SEM.
Cell culture
HepG2, Hep3B, and SK-Hep1 cells were cultured in T-75 flasks using MEM supplemented with
1
0% FBS, 1% antibiotic-antimycotic, 1% sodium pyruvate, 1% non-essential amino acids, and 1
mL gentamicin. HepG2, Hep3B, and SK-Hep1 cells were maintained at 37ꢂꢁ, 5% CO , and 95%
2
relative humidity and medium was changed every 48 hours. The cells were passaged at 80-90%
confluency using a 0.25% trypsin/0.20% EDTA solution.
Uptake of triGal vs. monoGal conjugates into hepatic cancer cells
The internalization of triGal and monoGal conjugates into HepG2, Hep3B, and SK-Hep1 cells
was measured as a function of particle composition and concentration via flow cytometry.
Briefly, 250,000 HepG2, Hep3B, or SK-Hep1 cells were seeded in 24-well plates and allowed to
adhere overnight. Treatment solutions of M12, M12-L, or T
concentration) were prepared in OPTI-MEM and then incubated with the cells for 24 hours at
7ꢂꢁ. After removing the treatment medium and washing the cells with warmed PBS twice, the
2
-T14 conjugates (142-570 nM G5
3
adherent cells were removed from the plates using a 0.25% trypsin/0.20% EDTA solution and
then suspended in fresh culture medium. The cells were then transferred to flow cytometry tubes,
centrifuged at 1000 RPM for 5 minutes at 4ꢁ, kept on ice, and resuspended immediately before
analysis. Samples were analyzed by flow cytometry using the intrinsic fluorescence of FITC (λex
:
1
0

4
88 nm; λem: 525 nm) on a Beckman Coulter Cyan ADP instrument provided by the Flow
Cytometry Core at the University of Michigan (Ann Arbor, MI). Data is presented as the mean ꢃ
SEM for n=4 replicates, and we used untreated cells in blank OPTI-MEM as our negative
control. Two-way ANOVA was used to determine the statistical difference between M12 and
each triGal conjugate at the same concentration (#) and between different concentrations of the
same treatment (*) and is denoted by ## or ** for p<0.01 and ### or *** for p<0.001.
Multi-spectral imaging flow cytometry of triGal vs. monoGal
conjugates in hepatic cancer cells
The internalized versus surface-bound ratio of triGal and monoGal conjugates in HepG2 and
6
Hep3B cells was measured using multi-spectral imaging flow cytometry. First, 1x10 HepG2 or
Hep3B cells were seeded in 24-well plates and allowed to adhere overnight. Treatment solutions
of M12 or T -T14 conjugates at 285 nM G5 concentration were prepared in OPTI-MEM and then
2
incubated with the cells for 24 hours at 37ꢁ. After removing the treatment solution and washing
the cells twice with PBS, the adherent cells were removed from the plates using a 0.25%
trypsin/0.20% EDTA solution and then suspended in fresh culture medium. The cells were spun
down at 1000 RPM at 4ꢁ, the supernatant aspirated, and then resuspended in PBS with 2% FBS
7
at 10 cells/mL in microcentrifuge tubes. The cells were then kept on ice and resuspended
immediately before analysis. On an Amnis ImagestreamX multi-spectral imaging flow cytometer
provided by the Flow Cytometry Core, singular cells in focus were measured for their FITC
signal. IDEAS software (EMD Millipore, Billerica, MA) was used to generate two populations
of FITC-positive cells based on an internalization ratio determined by the software as a
comparison between FITC intensity inside the cell versus the entire cell. We divided the high
1
1

0
internalization group (high internalization ratio, I ) by the surface-bound group (low
internalization ratio, I) in order to quantitatively assess the extent of nanoparticle internalization
0
0
(
I /I) between M12 and T
2
-T14 conjugates. I /I values are represented as the mean  SEM of three
replicates. We used a one-way ANOVA with Tukey’s multiple comparisons test to determine the
0
significance between I /I values for each group, with significance being denoted by * for p<0.05.
1
2

Results and Discussion
 3
We synthesized G5 dendrimers functionalized with a varying density (n) of cPEG-(NAcGal )
branches by modifying our previous synthetic strategies to achieve FITC-labeled G5-[cPEG-
NAcGal conjugates (Fig 2) (Kuruvilla et al., 2017; Scott H. Medina et al., 2013). We used
the same FITC-labeled G5-NH precursor for monoGal (M12) and the library of triGal conjugates
-T14) to ensure equivalent fluorescence activity (6 moles of FITC) between each conjugate.
We used N6-Cbz-lysine, a known starting material for synthesizing the triGal spacer. (Lee et al.,
011) N- alkylation of N6-Cbz-lysine was conducted as described by Du Roure et al. (Du Roure
(
 3 n
) ]
2
(
T
2
2
et al., 2003) N6-Cbz-lysine was treated with bromoacetic acid yielded an N6-carbobenzylaxy-L-
lysine triacid (6), which was coupled to NAcGal-amine (5) to obtain compound 7. The Cbz was
deprotected by hydrogenolysis (8) and then coupled to a heterofunctional, 2kDa Boc-NH-PEG-
NHS ester to obtain compound 9. After de-protecting the Boc group to establish compound 10,
reaction with cis-aconitic anhydride yielded the cPEG-(NAcGal
coupled compound 11 with the FITC-labeled G5-NH dendrimer (12) via peptide coupling at
varying molar ratios to achieve 13-18 with different ratios of cPEG-(NAcGal branches
attached (Fig 3). Finally, the remaining primary G5 amines on these conjugates were acetylated
 3
) targeting arms (11). We
2
)
 3
and the O-acetyl groups from NAcGal ligands was de-protected to achieve conjugates T -T14.
2
1
The conjugates were characterized by H-NMR and MALDI for their ligand concentration and
molecular weights, which can be found in Table 1. We also synthesized monoGal conjugates
with the lysine spacer (M12-L) to measure the effect of the spacer on cancer cell uptake in
comparison to the original M12 conjugates (Fig S40).
1
3

The variation in loading of cPEG-(NAcGal
and 11.1 mole% PEGylation of T , T , T , T
28 functional primary amines on G5 dendrimer. Given that PEG chains adopt a “mushroom”
conformation at low PEG densities (<5 mol%) and switch to a “brush” regime at higher
densities, it is expected that T -T14 conjugates should have differing PEG conformations based

)
3
branches corresponded to 1.6, 2.8, 4.5, 6.3, 8.3,
2
4
6
8
, T11, and T14 conjugates, respectively based on
1
2
on their varying PEG density. PEG chains attached to spherical nanoparticles in the brush
conformation typically impart higher hydrodynamic diameters (HD) to the nanoparticles, due to
the thin, bristle-like extension of the PEG away from the nanoparticle surface. Conversely,
nanoparticles with PEG in the mushroom conformation typically have smaller HDs due to the
coiling of the PEG chains. We performed dynamic light scattering (DLS) to identify the HD of
triGal conjugates and found that all conjugates exhibit an HD of approximately 7-8 nm, with no
statistical significance between them (Table 1). This suggests that the differences in PEGylation
between T -T14 conjugates that confers different PEG conformations does not impart significant
2
impacts on the HD of NPs at the nanometer scale.
All conjugates exhibited a size profile that confers the ability to surpass renal filtration from the
blood (HD < 5nm (Choi et al., 2009; Longmire et al., 2008)), thereby extending their retention
time within the bloodstream. MALDI analysis confirmed that the molecular weight (MW) of
 3
triGal particles increased with increasing density of cPEG-(NAcGal ) branches (Table 1). It is
important to note that both T and T conjugates (34,725 and 39,789 Da, respectively) fall under
2
4
the MW cut-off (40,000 Da) estimated to enable nanoparticles to exploit the EPR effect (Duncan
et al., 2013; Maeda et al., 2009; Seki et al., 2009). Studies in tumor-bearing mice will help
identify whether these conjugates are retained within the bloodstream and cleared through the
1
4

urine before they can concentrate into tumor tissue. T -T14, however, have MWs that should
6
enable their easy exploitation of the EPR effect. Finally, we measured the zeta potential of T -T14
2
conjugates and confirmed that they are neutral (Table 1), which should prevent non-specific
charge-charge interactions (Sadekar and Ghandehari, 2013) and protein opsonization (Alexis et
al., 2008) while circulating in the bloodstream.
 x n
Table 1: Physicochemical Properties of G5-[cPEG-(NAcGal ) ]
#
of cPEG- Total
Molecular
Weight
Zeta
Potential
(mV)
Particle
Type
Graphical
Depiction
(NAcGal)x NAcGal
Particle
Size (nm)
Chemical Structure
branches
n)
loading
per G5
(
Da)
(
(
[
FITC)
cPEG-
6
-G5-
monoGal
12
12.1
59,171
34,275
39,789
47,230
55,256
63,863
76,221
7.4 ± 0.30
8.3 ± 1.4
7.7 ± 0.2
7.5 ± 1.0
6.8 ± 1.0
7.5 ± 0.97
8.6 ± 0.30
-0.30 ± 0.21
-4.6 ± 0.28
0.0 ± 0.0
-1.5 ± 1.6
0.0 ± 0.0
0.0 ± 0.0
0.0 ± 0.0
(
x=1)
M12
T2
NAcGal
]
 12
(
[
(
FITC)
cPEG-
NAcGal
6
-G5-
2
5.2
β
)
3
]
2
(
[
(
FITC) -G5-
cPEG-
NAcGal
6
4
11.0
18.0
23.4
25.5
41.4
T4
β
)
3
]
4
(
[
(
FITC)
cPEG-
NAcGal
6
-G5-
6
T6
β
)
3
]
6
triGal
x=3)
(
(
[
(
FITC) -G5-
cPEG-
NAcGal
6
8
T8
β
)
3
]
8
(
[
(
FITC) -G5-
cPEG-
NAcGal
6
11
14
T1
1
β
)
3
]
11
14
(
[
(
FITC)
cPEG-
NAcGal
6
-G5-
T1
4
β
) ]
3
1
5

Uptake of triGal vs. monoGal in hepatic cancer cells
We used flow cytometry to establish whether the library of triGal conjugates could be recognized
and internalized by hepatic cancer cells, and to identify how this internalization compared to that
of monoGal conjugates. Briefly, we incubated M12 and T -T14 conjugates at 142, 285, and 570
2
nM G5 concentration with either HepG2, Hep3B, or SK-Hep1 cells for 24 hours. Data for
Hep3B and SK-Hep1 cells can be found in supplementary Fig S57 and S58, respectively. We
previously established that the internalization of G5-based conjugates targeted by monoGal
ligands are internalized by hepatic cancer cells at a NAcGal
nM (Kuruvilla et al., 2017; Medina et al., 2011; Scott H. Medina et al., 2013). With 12-16 moles
of cPEG-NAcGal attached to the dendrimers, these concentrations corresponded 7-285 nM of

concentration range of 100-4000

G5 dendrimers. At 142 and 285 nM G5, we achieve 100% uptake into HepG2 and Hep3B cells
with monoGal-targeted particles (M12). Correspondingly we chose a concentration range of 142-
5
70 nM in this study to compare the internalization of monoGal versus triGal conjugates into
HepG2 and Hep3B cells. Results show that T -T14 conjugates label HepG2 in a concentration-
2
dependent manner, labeling only 2-4% of HepG2 cells at 142 nM G5 but reaching 100% cell
labeling at 570 nM (Fig 4). Interestingly, at a low concentration of 142 nM, M12 labels 8- to 15-
fold more cells than any of the triGal conjugates, reaching strong statistical significance between
it and all triGal conjugates (p<0.0001, M12 vs. T
by T -T14 conjugates increases significantly (64-89% cells labeled). Of particular note, T
and T14 conjugates achieve higher cell labeling (89, 89 and 85%, respectively) than T , T
(64, 72, 72%, respectively) with statistical significance (p<0.01 for all comparisons).
Importantly, M12 labels virtually all cells (99%) with statistically significant differences from all
triGal conjugates (p<0.001). At this concentration of G5, the NAcGal present on M12 conjugates
2
-T14). At 285 nM, the labeling of HepG2 cells
2
8
, T11,
2
4
, and
T
6

1
6

(
3420 nM) falls in between that displayed by T11 (3135 nM) and T14 (3990 nM) conjugates.
Regardless, M12 achieves statistically higher labeling of cells above both of T11 and T14 (89 and
8
1
5% of cells, respectively). At the highest concentration of 570 nM G5, all conjugates label
00% of cells. We conducted similar uptake studies in another ASGPR-positive hepatic cancer
cell line, Hep3B, and found similar results (Fig S57). Further, we investigated the uptake of
triGal and monoGal conjugates in SK-Hep1 cells, an ASGPR-deficient HCC cell line (Saxena et
al., 2002; Tai et al., 2012), to ensure that the uptake was mediated and dependent on this receptor
(
Fig S58). For all particles, there was less than <10% internalization into SK-Hep1 cells at all
concentrations after 24 hours, verifying that internalization for both M12 and T -T14 conjugates is
ASGPR-mediated. Additionally, to ensure the difference in uptake between M12 and T
conjugates is not due to the lysine spacer included in the T -T14 conjugates, we synthesized a
2
2
-T14
2
lysine-based monoGal, M12-L, and showed that it exhibits no difference in uptake into HepG2
cells compared to M12 conjugates in this concentration range (Fig S49). We therefore conclude
that the presence of the spacer does not affect hepatic cancer cell uptake of NAcGal-targeted
dendrimers and is not the reason for differences between M12 and T -T14 labeling.
2
Taken together, our results show that triGal conjugates are ASGPR-specific and able to label
00% HepG2 cells at the highest concentrations, while at lower concentrations they achieve cell
labeling but to a lower extent. However, M12 conjugates label 1.5-8 folds more HepG2 cells at
1
lower G5 concentrations (i.e 142-285 nM), suggesting that mono-valent NAcGal

-targeting of
G5 dendrimers is more efficient at being recognized by hepatic cancer cells.
1
7

Surface versus internal localization of triGal vs monoGal in HepG2
cells
Given our initial flow cytometry results, we sought to understand why triGal conjugates, while
being able to label HepG2 cells, cannot do so as efficiently as their monoGal counterparts despite

similar concentrations of NAcGal . We hypothesized that the decrease in cell labeling may come
from one of a few theories related to valency-dependent cell-binding. In particular, many
investigators have described receptor cross-linking using coiled-coil networks, which describes
molecules targeting a receptor at the cell surface that are connected to a larger polymer network
(
i.e. coiled-coil), creating a crosslinking of receptors through this extracellular network and
rendering them deficient or even sometimes leading to apoptosis induction (Wu et al., 2010;
Zacco et al., 2015). We hypothesized that triGal conjugates could be inducing a similar “receptor
 3
crosslinking” phenomenon. This would be possible if the same cPEG-(NAcGal ) branch was
bound to multiple ASGPRs, where both would “pull” on the same NP for endocytosis, but face
competition by an equal and opposite force from another engaged receptor. In this way, the G5
dendrimer would be the “network” causing the receptor crosslinking, and while triGal conjugates
would be bound to the cell surface they may not be internalized fully. This would help explain
why other investigators (Khorev et al., 2008; Lee et al., 2011; Nair et al., 2014) have observed
that triGal-targeted small molecules exhibit improved distribution to hepatic cancer cells, but
when triGal is attached to larger molecules (e.g. nanoparticles), the geometry complicates the
binding and internalization potential.
1
8

To distinguish between surface-bound and internalized nanoparticles, we turned to multi-spectral
imaging flow cytometry, which adds the ability to microscopically image cells being sorted by
flow cytometry in order to identify the cellular localization of the fluorescence, and in this case,
the nanoparticles (Phanse et al., 2012). We incubated either M12 or T -T14 conjugates at 285 nM
2
for 24 hours with HepG2 (Fig 5) and for Hep3B (Fig S59) cells and used the data collection
software to measure the internalization ratio of each FITC-labeled cell and separate them into
cells with surface-bound nanoparticles or cells with mostly internalized ones. It is important to
note that these two populations are presented as a percentage of all FITC-labeled cells, so
differences in cell labeling mentioned above are accounted for in this analysis and are not
relevant. Results indicate that the localization of nanoparticles does not differ between triGal and
monoGal conjugates, and also does not differ between the triGal conjugates themselves. Images
collected for all treatments during flow cytometry show both punctate fluorescence at cell
membranes, indicating surface-bound nanoparticles, and diffuse fluorescence within the cell
indicating nanoparticle diffusion throughout the cell body.
We also quantitatively evaluated the extent of internalization for all nanoparticle treatments by
0
0
determining I /I, a metric comparing the extent of nanoparticle internalization (I ) versus surface
0
localization (I) HepG2 (Fig 6) and for Hep3B (Fig S60). I /I > 1 would indicate higher
0
internalization than surface-bound localization, while I /I < 1 would indicate higher surface
0
localization than internalization. Results show that I /I ranged from 3.2 for T
2
conjugates up to
4
.1 for T conjugates, with no statistical significance between any of the treatment groups. This
8
indicates that there was significantly higher internalization than surface localization for all
0
treatments. Further, the lack of difference in I /I values between triGal and monoGal conjugates
1
9

indicates that there are in fact no higher surface-localized triGal conjugates than there are
monoGal conjugates. Since these populations contain an equivalent number of cells that are
labeled by the NPs, the equivalence in internalized versus surface-bound particles between
monoGal and triGal treatments indicates that the internalization kinetics are similar between the
two.
Our earlier results showing that monoGal labels more cells than triGal suggests that the
differences in valency contribute to the kinetics of receptor binding, but not of internalization
once the particles are bound to the ASGPR. We believe therefore that receptor-crosslinking may
not be occurring with triGal conjugates. It is possible that the geometric spacing between
NAcGal
ASGPR. Khorev et al. showed through molecular modeling that the length of the flexible spacer
attaching NAcGal ligands to the backbone as well as the space between each NAcGal ligand
  3
ligands at the tip of cPEG-(NAcGal ) branches affects their kinetics of binding to the


specifically impacted their binding affinity to ASGPR (Khorev et al., 2008). Zacco et al. built off
of this work to create a glycopeptide library studying various combinations of spacer lengths and
 
distances between NAcGal ligands, identifying that NAcGal ligands spaced 7 amino acids from
each other on a peptide backbone and at the tip of an 18 angstrom spacer achieved the best
targeting of the ASGPR (Zacco et al., 2015). While these conditions were all established by
testing triGal ligands either incubated alone or attached to small molecules, we believe the
specific geometric requirements will be even more important when triGal ligands are attached to
nanoparticles. In fact, with multiple triGal branches attached to the same nanoparticle, as is the
case in the present study, further complications may arise from steric hindrance, repulsion, and
competitive binding. Hence the discrepancy we are observing between triGal targeting of
2
0

nanoparticles versus previously established targeting of small molecules. We are currently
undertaking molecular modeling and geometric measurements of our triGal library to help
elucidate these differences. Nevertheless, this study suggests that the display of triGal branches
on G5 dendrimers does not improve internalization kinetics compared to the display of monoGal
ligands, as was the phenomenon observed for small molecules. However, the triGal conjugates
presented here still provide a library of potential drug delivery vehicles that can be used for
hepatic cancer therapy, depending on the constraints of fabrication, therapeutic loading, and
biodistribution.
Conclusions
This study focuses on the synthesis and validation of a library of G5 dendrimers displaying a
 3
varying density (n=2, 4, 6, 8, 11, or 14) of tri-valent (NAcGal ) ligands as potential drug
delivery vehicles for hepatic cancer therapy. Our results indicate that triGal conjugates achieve
concentration-dependent internalization into hepatic cancer cells that is comparable to our-

previously established mono-valent, NAcGal -targeted dendrimers. Interestingly, triGal
conjugates do not exhibit improved internalization over monoGal conjugates, as was observed
previously with triGal-targeting of small molecules. In fact, monoGal conjugates more
efficiently label hepatic cancer cells than triGal conjugates at lower concentrations. Further,
multi-spectral imaging flow cytometry confirmed that the localization of triGal conjugates is
both intracellular and at the surface, similar to their monoGal counterparts. Taken together, it is
evident that binding of G5 dendrimers to the ASGPR is affected by NAcGal valency, but the
process of how they are internalized is less susceptible to the difference. To the best of our
2
1


knowledge, this is the first study to describe the effect of multi-valent NAcGal ligands on
nanoparticle targeting of hepatic cancer cells. The synthetic strategies, biological findings, and
library of conjugates established in this work offers information crucial to improving drug
delivery strategies for the treatment of hepatic cancer.
Acknowledgments
S.P.K. and G.T. contributed equally to this work. The authors would like to thank Dr. Jinsang
Kim for providing access to their fluorescence spectrophotometer instrument. Sibu P. Kuruvilla
recognizes the support of the NSF Graduate Research Fellowship (GRFP) award and the
University of Michigan Rackham Merit Fellowship (RMF).
References
Alexis, F., Pridgen, E., Molnar, L.K., Farokhzad, O.C., 2008. Factors Affecting the Clearance
and Biodistribution of Polymeric Nanoparticles. Mol. Pharm. 5, 505–515.
Arias, J.L., 2011. Drug targeting strategies in cancer treatment: an overview. Mini Rev. Med.
Chem. 11, 1–17. doi:10.2174/138955711793564024
Bertrand, N., Wu, J., Xu, X., Kamaly, N., Farokhzad, O.C., 2014. Cancer nanotechnology: The
impact of passive and active targeting in the era of modern cancer biology. Adv. Drug
Deliv. Rev. 66, 2–25. doi:10.1016/j.addr.2013.11.009
Cao, Y., He, Y., Liu, H., Luo, Y., Shen, M., Xia, J., Shi, X., 2015. Targeted CT imaging of
human hepatocellular carcinoma using low-generation dendrimer-entrapped gold
nanoparticles modified with lactobionic acid. J. Mater. Chem. B 3, 286–295.
doi:10.1039/C4TB01542H
Choi, H.S., Liu, W., Misra, P., Tanaka, E., Zimmer, J.P., Itty, B., Bawendi, M.G., Frangioni, J.
V, 2009. Renal Clearance of Nanoparticles. Nat. Biotechnol. 25, 1165–1170.
doi:10.1038/nbt1340.Renal
2
2

Du Roure, O., Debiemme-Chouvy, C., Malthête, J., Silberzan, P., 2003. Functionalizing surfaces
with nickel ions for the grafting of proteins. Langmuir 19, 4138–4143.
doi:10.1021/la020636z
Duncan, R., Sat-Klopsch, Y.-N., Burger, A.M., Bibby, M.C., Fiebig, H.H., Sausville, E. a, 2013.
Validation of tumour models for use in anticancer nanomedicine evaluation: the EPR effect
and cathepsin B-mediated drug release rate. Cancer Chemother. Pharmacol. 72, 417–427.
doi:10.1007/s00280-013-2209-7
Fang, J., Nakamura, H., Maeda, H., 2011. EPR effect: the unique characteristics of tumor blood
vessels for drug delivery, factors involved, its limitation and augmentation. Adv. Drug
Deliv. Rev. 63, 136–151.
Kallinteri, P., Higgins, S., Hutcheon, G. a, St Pourçain, C.B., Garnett, M.C., 2005. Novel
functionalized biodegradable polymers for nanoparticle drug delivery systems.
Biomacromolecules 6, 1885–94. doi:10.1021/bm049200j
Kang, Y.M., Kim, G.H., Kim, J. Il, Kim, D.Y., Lee, B.N., Yoon, S.M., Kim, J.H., Kim, M.S.,
2
011. In vivo efficacy of an intratumorally injected in situ-forming
doxorubicin/poly(ethylene glycol)-b-polycaprolactone diblock copolymer. Biomaterials 32,
556–4564. doi:10.1016/j.biomaterials.2011.03.007
4
Khorev, O., Stokmaier, D., Schwardt, O., Cutting, B., Ernst, B., 2008. Trivalent, Gal/GalNAc-
containing ligands designed for the asialoglycoprotein receptor. Bioorg. Med. Chem. 16,
5
216–31. doi:10.1016/j.bmc.2008.03.017
Kuruvilla, S.P., Tiruchinapally, G., ElAzzouny, M., ElSayed, M.E.H., 2017. N-
Acetylgalactosamine-Targeted Delivery of Dendrimer-Doxorubicin Conjugates Influences
Doxorubicin Cytotoxicity and Metabolic Profile in Hepatic Cancer Cells. Adv. Healthc.
Mater. 6, 1601046. doi:10.1002/adhm.201601046
Lee, R.T., Lee, Y.C., 1997. Facile Synthesis of a High-Affinity Ligand for Mammalian Hepatic
Lectin Containing Three Terminal N-Acetylgalactosamine Residues. Bioconjug. Chem. 8,
7
62–765. doi:10.1021/bc9700796
Lee, R.T., Wang, M.-H., Lin, W.-J., Lee, Y.C., 2011. New and more efficient multivalent glyco-
ligands for asialoglycoprotein receptor of mammalian hepatocytes. Bioorg. Med. Chem. 19,
2
494–500. doi:10.1016/j.bmc.2011.03.027
Li, L., Tang, F., Liu, H., Liu, T., Hao, N., Chen, D., Teng, X., He, J., 2010. In vivo delivery of
silica nanorattle encapsulated docetaxel for liver cancer therapy with low toxicity and high
efficacy. ACS Nano 4, 6874–6882.
Li, Y., Huang, G., Diakur, J., Wiebe, L.I., 2008. Targeted Delivery of Macromolecular Drugs:
Asialoglycoprotein Receptor (ASGPR) Expression by Selected Hepatoma Cell Lines used
in Antiviral Drug Development. Curr. Drug Deliv. 5, 299–302.
doi:10.2174/156720108785915069
Liu, H., Wang, H., Xu, Y., Guo, R., Wen, S., Huang, Y., Liu, W., Shen, M., Zhao, J., Zhang, G.,
Shi, X., 2014. Lactobionic acid-modified dendrimer-entrapped gold nanoparticles for
targeted computed tomography imaging of human hepatocellular carcinoma. ACS Appl.
Mater. Interfaces 6, 6944–53. doi:10.1021/am500761x
2
3

Longmire, M., Choyke, P.L., Kobayashi, H., 2008. Clearance Properties of Nano-sized Particles
and Molecules as Imaging Agents: Considerations and Caveats. Nanomedicine 3, 703–717.
doi:10.2217/17435889.3.5.703.Clearance
Maeda, H., Bharate, G.Y., Daruwalla, J., 2009. Polymeric drugs for efficient tumor-targeted drug
delivery based on EPR-effect. Eur. J. Pharm. Biopharm. 71, 409–419.
Medina, S.H., Chevliakov, M. V, Tiruchinapally, G., Durmaz, Y.Y., Kuruvilla, S.P., Elsayed,
M.E.H., 2013. Enzyme-activated nanoconjugates for tunable release of doxorubicin in
hepatic cancer cells. Biomaterials 34, 4655–4666. doi:10.1016/j.biomaterials.2013.02.070
Medina, S.H., Tekumalla, V., Chevliakov, M. V, Shewach, D.S., Ensminger, W.D., El-Sayed,
M.E.H., 2011. N-acetylgalactosamine-functionalized dendrimers as hepatic cancer cell-
targeted carriers. Biomaterials 32, 4118–4129. doi:10.1016/j.biomaterials.2010.11.068
Medina, S.H., Tiruchinapally, G., Chevliakov, M. V., Durmaz, Y.Y., Stender, R.N., Ensminger,
W.D., Shewach, D.S., ElSayed, M.E.H., 2013. Targeting Hepatic Cancer Cells with
PEGylated Dendrimers Displaying N‐ Acetylgalactosamine and SP94 Peptide Ligands.
Adv. Healthc. Mater. 2, 1337–1350. doi:10.1002/adhm.201200406
Nair, J.K., Willoughby, J.L.S., Chan, A., Charisse, K., Alam, M.R., Wang, Q., Hoekstra, M.,
Kandasamy, P., Kelin, A. V., Milstein, S., Taneja, N., Oshea, J., Shaikh, S., Zhang, L., Van
Der Sluis, R.J., Jung, M.E., Akinc, A., Hutabarat, R., Kuchimanchi, S., Fitzgerald, K.,
Zimmermann, T., Van Berkel, T.J.C., Maier, M.A., Rajeev, K.G., Manoharan, M., 2014.
Multivalent N -acetylgalactosamine-conjugated siRNA localizes in hepatocytes and elicits
robust RNAi-mediated gene silencing. J. Am. Chem. Soc. 136, 16958–16961.
doi:10.1021/ja505986a
Phanse, Y., Ramer-Tait, A.E., Friend, S.L., Carrillo-Conde, B., Lueth, P., Oster, C.J., Phillips,
G.J., Narasimhan, B., Wannemuehler, M.J., Bellaire, B.H., 2012. Analyzing cellular
internalization of nanoparticles and bacteria by multi-spectral imaging flow cytometry. J.
Vis. Exp. e3884. doi:10.3791/3884
Rensen, P.C.N., van Leeuwen, S.H., Sliedregt, L.A.J.M., van Berkel, T.J.C., Biessen, E.A.L.,
2
004. Design and Synthesis of Novel N-Acetylgalactosamine-Terminated Glycolipids for
Targeting of Lipoproteins to the Hepatic Asialoglycoprotein Receptor. J. Med. Chem. 47,
798–5808. doi:10.1021/jm049481d
5
Sadekar, S., Ghandehari, H., 2013. Transepithelial Transport and Toxicity of PAMAM
Dendrimers: Implications for Oral Drug Delivery. Adv. Drug Deliv. Rev. 64, 571–588.
doi:10.1016/j.addr.2011.09.010.TRANSEPITHELIAL
Saxena, A., Yik, J.H.N., Weigel, P.H., 2002. H2, the minor subunit of the human
asialoglycoprotein receptor, trafficks intracellularly and forms homo-oligomers, but does
not bind asialo-orosomucoid. J. Biol. Chem. 277, 35297–35304.
doi:10.1074/jbc.M205653200
Seki, T., Fang, J., Maeda, H., 2009. Tumor-targeted macromolecular drug delivery based on the
enhanced permeability and retention effect in solid tumor, in: Lu, Y., Mahato, R. (Eds.),
Pharmaceutical Perspectives of Cancer Therapeutics. Springer Science & Business Media,
p. 100.
Tai, W.-T., Cheng, A.-L., Shiau, C.-W., Liu, C.-Y., Ko, C.-H., Lin, M.-W., Chen, P.-J., Chen,
2
4

K.-F., 2012. Dovitinib induces apoptosis and overcomes sorafenib resistance in
hepatocellular carcinoma through SHP-1-mediated inhibition of STAT3. Mol. Cancer Ther.
1
1, 452–63. doi:10.1158/1535-7163.MCT-11-0412
Trerè, D., Fiume, L., De Giorgi, L.B., Di Stefano, G., Migaldi, M., Derenzini, M., 1999. The
asialoglycoprotein receptor in human hepatocellular carcinomas: its expression on
proliferating cells. Br. J. Cancer 81, 404–8. doi:10.1038/sj.bjc.6690708
Wang, H.-X., Xiong, M.-H., Wang, Y.-C., Zhu, J., Wang, J., 2013. N-acetylgalactosamine
functionalized mixed micellar nanoparticles for targeted delivery of siRNA to liver. J.
Control. Release 166, 106–14. doi:10.1016/j.jconrel.2012.12.017
Westerlind, U., Westman, J., Törnquist, E., Smith, C.I., Oscarson, S., Lahmann, M., Norberg, T.,
2
004. Ligands of the asialoglycoprotein receptor for targeted gene delivery, part 1: synthesis
of and binding studies with biotinylated cluster glycosides containing N-
acetylgalactosamine. Glycoconjugate J. 21, 227–241.
Wu, K., Liu, J., Johnson, R.N., Yang, J., Kopecek, J., 2010. Drug-Free Macromolecular
Therapeutics: Induction of Apoptosis by Coiled-Coil-Mediated Cross-Linking of Antigens
on the Cell Surface. Angew. Chemie 49, 1451–1455.
Yoo, H.S., Lee, K.H., Oh, J.E., Park, T.G., 2000. In vitro and in vivo anti-tumor activities of
nanoparticles based on doxorubicin-PLGA conjugates. J. Control. Release 68, 419–431.
doi:10.1016/S0168-3659(00)00280-7
Zacco, E., Hütter, J., Heier, J.L., Mortier, J., Seeberger, P.H., Lepenies, B., Koksch, B., 2015.
Tailored Presentation of Carbohydrates on a Coiled Coil-Based Scaffold for
Asialoglycoprotein Receptor Targeting. ACS Chem. Biol. 10, 2065–2072.
doi:10.1021/acschembio.5b00435
2
5

Fig 1: Strategy for multi-valent targeting of hepatic cancer cells. We compared the effect of
tri-valent NAcGal display on targeting G5 dendrimers to hepatic cancer cells in comparison to
mono-valent NAcGal. We attached NAcGal
brush and a cis-aconitic spacer (c) to achieve G5-(cPEG-NAcGal
3
ligands attached to the G5 surface through a PEG
(i.e. triGal) conjugates and
)
3 n
compared their distribution to hepatic cancer cells to mono-valent G5-(cPEG-NAcGal)12.1 (i.e.
monoGal) conjugates. We synthesized a library of triGal conjugates with varying density of
targeting branches attached, namely with n=2, 4, 6, 8, 11, or 14 moles of PEG-NAcGal
branches, to achieve T , T , T , T , T11, and T14 conjugates, respectively. We compared the
3
2
4
6
8
internalization of monoGal and triGal conjugates via conventional and multi-spectral imaging
flow cytometry methods.
Fig 2: Synthesis of (NAcGal
β
)
3
-Lysine-6-NH-PEG-NH (10)
2
Fig 3: Synthesis of T -T14 conjugates
2
Fig 4: Uptake of triGal vs. monoGal into HepG2 cells. We measured the uptake of monoGal
and triGal conjugates into HepG2 cells via flow cytometry. M12, T , T , T , T
conjugates were incubated at G5 concentrations of 142, 285, and 570 nM, which corresponded to
various NAcGal concentrations based on the loading density, as seen in the table. Results show
that both monoGal and triGal label HepG2 cells with increasing concentration, with monoGal
achieving higher labeling at lower conjugate concentrations. T , T , and T exhibited lower
internalization than T , T11, and T14 conjugates, likely due to their loaded NAcGal differences.
2
4
6
8
, T11, and T14

2
4
6
8

Results are presented as the mean of four replicates  SEM. Two-way ANOVA was used to
determine the statistical difference M12 and each triGal conjugate at the same concentration (#)
and between different concentrations of the same treatment (*), and is denoted by # or * for
p<0.05, ## or ** for p<0.01, and ### or *** for p<0.001.
Fig 5: Surface versus internalized localization of monoGal and triGal conjugates in HepG2
cells. We used multi-spectral imaging flow cytometry to visualize surface-bound conjugates and
internalized conjugates, as assessed by an IDEAS software-based internalization algorithm.
Results show that monoGal and triGal conjugates achieve both surface-localization and
internalization after 24 hour incubation at 285 G5 nM.
0
Fig 6: Internalized versus surface bound ratio (I /I) of T
2
-T14 and M12 conjugates in HepG2
cells. We quantitatively assessed the ratio of FITC-labeled HepG2 cells with high internalization
0
0
(
I ) versus low internalization (I) by determining I /I. Results show that for all treatment groups,
the conjugates were internalized to a much greater extent than they were maintained at the
0
0
surface, as indicated by I /I values > 1. Further, the I /I values between monoGal and triGal
2
6

conjugates are not statistically different, indicating that they achieve the same ratio of
internalized particles when they label cells. Results are presented as the mean of three replicates

SEM. A one-way ANOVA test was used to determine differences between each conjugate
group.
2
7

2
8

2
9