Glucose transporter 1-mediated vascular translocation of nanomedicines enhances accumulation and efficacy in solid tumors

Article abstract of DOI:10.1016/j.jconrel.2019.02.021

Nanomedicine modification with ligands directed to receptors on tumor blood vessels has the potential for selectively enhancing nanomedicine accumulation in malignant tissues by overcoming the vascular barrier of tumors. Nevertheless, the development of broadly applicable ligand approaches capable of promoting the transvascular transport of nanomedicines in a wide spectrum of tumors has been elusive so far. By considering the indispensable and persistent glycolytic fueling of tumors, we developed glucose-installed polymeric micelles loading cisplatin (Gluc-CDDP/m) targeting the glucose transporter 1 (GLUT1), which is overexpressed in most tumors and present on vascular endothelial cells, toward improving the delivery efficiency and therapeutic efficacy. The design of the glucose ligands on Gluc-CDDP/m was engineered to control the conjugation via the carbon 6 of the glucose moieties, as well as the ligand density on the poly (ethylene glycol) (PEG) shell of the micelles. The series of micelles was then studied in vitro and in vivo against GLUT1-high human squamous cell carcinoma of the head and neck OSC-19 cells and GLUT1-low human glioblastoma-astrocytoma U87MG cells. Our results showed that precisely tuning the micelles to have glucose ligands on 25% of their PEG chains increased the efficacy against the tumors by significantly enhancing the tumor accumulation, even in GLUT1-low U87MG tumors. The enhancement of the intratumoral levels of these micelles was hindered by concomitant administration of glucose, or the GLUT1 inhibitor STF-31, confirming a GLUT1/glucose-mediated increment of the accumulation. Intravital confocal laser scanning microscopy imaging of tumor tissues further demonstrated the rapid extravasation and penetration of Gluc-CDDP/m in OSC-19 tumors compared to non-targeted CDDP/m. These findings indicate GLUT1-targeting as a promising approach for overcoming the vascular barrier and boosting the delivery of nanomedicine in tumors.

Full text of DOI:10.1016/j.jconrel.2019.02.021

Accepted Manuscript
Glucose transporter 1-mediated vascular translocation of
nanomedicines enhances accumulation and efficacy in solid
tumors
Kazumi Suzuki, Yutaka Miura, Yuki Mochida, Takuya Miyazaki,
Kazuko Toh, Yasutaka Anraku, Vinicio Melo, Xueying Liu,
Takehiko Ishii, Osamu Nagano, Hideyuki Saya, Horacio Cabral,
Kazunori Kataoka
PII:
S0168-3659(19)30096-3
DOI:
https://doi.org/10.1016/j.jconrel.2019.02.021
COREL 9660
Reference:
To appear in:
Journal of Controlled Release
Received date:
Revised date:
Accepted date:
1 November 2018
6 February 2019
17 February 2019
Please cite this article as: K. Suzuki, Y. Miura, Y. Mochida, et al., Glucose transporter
-mediated vascular translocation of nanomedicines enhances accumulation and efficacy
1
in solid tumors, Journal of Controlled Release, https://doi.org/10.1016/
j.jconrel.2019.02.021
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.

ACCEPTED MANUSCRIPT
Glucose Transporter 1-Mediated Vascular Translocation of Nanomedicines Enhances
Accumulation and Efficacy in Solid Tumors
1
, 1 , 2
1,1,3
2
1
2
Kazumi Suzuki
, Yutaka Miura
, Yuki Mochida , Takuya Miyazaki , Kazuko Toh ,
1
,2
1,2
2
1,4
3
Yasutaka Anraku , Vinicio Melo , Xueying Liu , Takehiko Ishii , Osamu Nagano , Hideyuki
3
1,*
2,4,*
Saya , Horacio Cabral horacio@bmw.t.u-tokyo.ac.jp, Kazunori Kataoka
kataoka@pari.u-
tokyo.ac.jp
1
Department of Bioengineering, Graduate School of Engineering, The University of Tokyo, 7-3-
1
Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
2
Innovation Center of Nanomedicine (iCONM), Kawasaki Institute of Industrial Promotion, 3-
2
5-14 Tonomachi, Kawasaki-ku, Kawasaki 210-0821, Japan
3
Division of Gene Regulation, Institute for Advanced Medical Research, School of Medicine,
Keio University, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan
4
Policy Alternatives Research Institute, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku,
Tokyo 113-0033, Japan
*
Corresponding authors.
1
These authors contributed equally.
2
Present affiliation: JSR Corporation 25 Miyukigaoka, Tsukuba-shi, Ibaraki 305-0841 Japan.
3
Present affiliation: Kowa Company, Ltd. 6-29, Nishiki 3-chome, Naka-ku, Nagoya, Aichi, 460-
8
625, Japan.
4
Present affiliation: Research Division, NanoCarrier Co., Ltd, 144-15 Chuo, 226-39 Wakashiba,
Kashiwa, Chiba, 277-0871, Japan.

ACCEPTED MANUSCRIPT
ABSTRACT
Nanomedicine modification with ligands directed to receptors on tumor blood vessels has
the potential for selectively enhancing nanomedicine accumulation in malignant tissues by
overcoming the vascular barrier of tumors. Nevertheless, the development of broadly applicable
ligand approaches capable of promoting the transvascular transport of nanomedicines in a wide
spectrum of tumors has been elusive so far. By considering the indispensable and persistent
glycolytic fueling of tumors, we developed glucose-installed polymeric micelles loading
cisplatin (Gluc-CDDP/m) targeting the glucose transporter 1 (GLUT1), which is overexpressed
in most tumors and present on vascular endothelial cells, toward improving the delivery
efficiency and therapeutic efficacy. The design of the glucose ligands on Gluc-CDDP/m was
engineered to control the conjugation via the carbon 6 of the glucose moieties, as well as the
ligand density on the poly(ethylene glycol) (PEG) shell of the micelles. The series of micelles
was then studied in vitro and in vivo against GLUT1-high human squamous cell carcinoma of the
head and neck OSC-19 cells and GLUT1-low human glioblastoma-astrocytoma U87MG cells.
Our results showed that precisely tuning the micelles to have glucose ligands on 25% of their
PEG chains increased the efficacy against the tumors by significantly enhancing the tumor
accumulation, even in GLUT1-low U87MG tumors. The enhancement of the intratumoral levels
of these micelles was hindered by concomitant administration of glucose, or the GLUT1
inhibitor STF-31, confirming a GLUT1/glucose-mediated increment of the accumulation.
Intravital confocal laser scanning microscopy imaging of tumor tissues further demonstrated the
rapid extravasation and penetration of Gluc-CDDP/m in OSC-19 tumors compared to non-
targeted CDDP/m. These findings indicate GLUT1-targeting as a promising approach for
overcoming the vascular barrier and boosting the delivery of nanomedicine in tumors.

ACCEPTED MANUSCRIPT
Keywords. Polymeric micelles, glucose, GLUT1, cisplatin, tumor vasculature, EPR effect.
INTRODUCTION
Nanomedicines have shown selective accumulation in tumors based on the enhanced
permeability and retention (EPR) effect by exploiting the permeable vasculature and limited
lymphatic drainage in cancerous tissues [1,2]. However, a limited fraction of the total
administered dose reaches the tumor site by this route [3]. Moreover, in clinical tumors, the level
of leaky vessels and extent of the EPR effect appear to be highly variable, which can limit the
success of nanomedicines [4,5]. An attractive strategy for improving accumulation is
circumventing the vascular barrier of tumors by promoting the translocation of nanomedicines
through ligands directed to receptors on tumor blood vessels readily exposed to the bloodstream
[6,7]. Nevertheless, despite this approach critically augmented nanomedicine levels in certain
malignancies, such as cyclic RGD peptide-installed nanomedicines targeting αVβ3- and αVβ5-
expressing vasculature in brain tumors [8,9] or iRGD peptide facilitated transport in tumors
expressing neuropilin-1 (or neuropilin-2) on their endothelium [10], the development of ligand
strategies promoting the transvascular transport of nanomedicines in a wide range of tumors
remains to be satisfied, as the discovery of ubiquitous tumor vascular markers accessible in vivo
has been exceptionally difficult.
Targeting receptors on tumor blood vessels that are involved in indispensable and
persistent tumoral processes could provide an effective modality for broadly improving the
delivery efficiency of nanomedicines. In this regard, glucose transporters, notably glucose
transporter 1 (GLUT1), are crucial for the hallmark glycolytic fueling of tumors [11], as glucose
cannot pass through cellular membranes by simple diffusion. Indeed, markedly increased
accumulation of glucose has been reported in many human tumor types [12,13], with GLUT1

ACCEPTED MANUSCRIPT
expression closely correlating with tumor malignancy and resistance to therapies [14-16]. Thus,
several studies have used glucose as ligands [17], including glucose-installed nanomedicines
[18,19], to target the glucose transporters on tumor cells. On the other hand, since the transport
of glucose into tissues is mediated by GLUT1 on vascular endothelial cells [20,21], it is
reasonable to assume the presence of GLUT1 in the tumor vasculature to satisfy the high glucose
demand of cancer cells, thereby, representing a significant target for strategies aimed at
enhancing transvascular transport. Nevertheless, such approach remains unexplored, as all
previous studies on glucose-installed nanomedicines were aimed to target the tumor cell itself
without focusing on the targeting of GLUT1 on the endothelial cells.
Here, we studied GLUT1-targeted nanomedicines as a new strategy directed to overcome
vascular barrier, enhancing delivery and efficacy in solid tumors. These nanomedicines were
prepared by controlled installation of glucose on polymeric micelles incorporating the antitumor
agent cisplatin, which is used for the treatment of a wide range of tumors (Figure 1). The
cisplatin-loaded micelles (CDDP/m) serving as the platform for this study have shown potent
antitumor effects and reduced side effects in humans, and are being evaluated in Phase III
clinical trials against pancreatic cancer [7,22]. Before evaluating the effects of GLUT1-targeting
on these micelles, the expression of GLUT1 on the endothelial cells of tumor blood vessels was
validated in xenografts and clinical tumor biopsies. Then, the ability of glucose-installed
CDDP/m (Gluc-CDDP/m) to improve delivery efficiency through the glucose/GLUT1 system
was compared in vitro, as well as in vivo against two tumor models, i.e. the high-GLUT1-
expressing human squamous cell carcinoma of the head and neck (OSC-19), which harbor a
large subpopulation of cisplatin-resistant cells with cancer stem-like cell characteristics [23], and
the low-GLUT1-expressing human glioblastoma-astrocytoma (U87MG) [24]. Our results

ACCEPTED MANUSCRIPT
demonstrated that precisely tuning the micelles to have 25% of their PEG chains with glucose
conjugated at the carbon 6 via ether linkage promoted the selective accumulation in the tumors
through GLUT1-associated vascular translocation. The intratumoral levels of these micelles in
both GLUT1-high OSC-19 and GLUT1-low U87-MG tumors were 2-fold higher than those of
the control CDDP/m in these tumors, which are solely based on the EPR effect, allowing Gluc-
CDDP/m to improve the antitumor activity. These findings indicate our approach of exploiting
the GLUT1/glucose interaction by glucose-installed nanomedicines as a promising new strategy
for boosting delivery and efficacy of nanomedicines against solid tumors.
Figure 1. Construction of glucose-installed cisplatin-loaded micelles (Gluc-CDDP/m) from
poly(ethylene glycol)-poly(L-glutamic acid) (PEG-P(Glu); MWPEG = 12 kDa; polymerization
degree of P(Glu) = 40) and α-glucopyranos-6-O-yl-poly(ethylene glycol)-poly(L-glutamic acid)
(Gluc-PEG-P(Glu); MWPEG = 12 kDa; polymerization degree of P(Glu) = 40) block copolymers.
Glucose was installed through carbon 6 into the α-end of the poly(ethylene glycol) (PEG) block
via ether linkage. The micelles were self-assembled by the polymer-metal complex formation
between the carboxylate group in the P(Glu) segment and cisplatin. The density of glucose on the

ACCEPTED MANUSCRIPT
surface of the micelles was controlled through the mixing ratio of Gluc-PEG-P(Glu) and CH3O-
PEG-P(Glu).
MATERIALS AND METHODS
Materials
1
,2-O-isopropylidene--D-glucofuranose
toluenesulfonic acid monohydrate were purchased from Tokyo Chemical Industries, Ltd.,
Tokyo, Japan). Dimethyl sulfoxide (>99%), triethylamine (TEA, >99%), pyridine (>99%),
(MIG),
2,2-dimethoxypropane,
p-
(
tetrahydrofuran (THF, >99%), and methanesulfonyl chloride (MsCl, >99%) were purchased
from Wako Pure Chemical Industries, Ltd., (Tokyo, Japan), purified by distillation over drying
agents, e.g., CaH or P O , and were transferred under argon. Ethylene oxide (Sumitomo Seika
2
2
5
Chemical, Osaka, Japan) was dried over CaH2 and distilled under an argon atmosphere.
Potassium naphthalene was prepared as a THF solution according to a previous paper [25]. α-
Methoxy-ω-amino-poly(ethylene glycol) (MeO-PEG-NH2, 12,000 g/mol, NOF Corporation,
Tokyo, Japan) was purified using an ion exchange CM Sephadex C-50 column (GE Healthcare
Ltd., UK). Pivaloyl chloride (Tokyo Chemical Industries, Ltd., Tokyo, Japan), ammonia solution
(ca. 28%, Nakarai Tesque, Kyoto, Japan), cis-dichlorodiamineplatinum (II) (cisplatin, Aldrich
Chemical Co., Inc. Milwaukee, WI), N-carboxy anhydride of -benzyl L-glutamate (Chuo
Kaseihin Co., Inc., Tokyo, Japan), Alexa 674-succinimidyl ester (Life Technologies Corporation,
Tokyo, Japan), Alexa 555-succinimidyl ester (Life Technologies Corporation, Tokyo, Japan),
and other reagents were used without further purification. Dulbecco’s modified eagle’s medium
(DMEM)
and
STF-31
(4-[[[[4-(1,1-Dimethylethyl)phenyl]sulfonyl]amino]methyl]-N-3-

ACCEPTED MANUSCRIPT
pyridinyl-benzamide) were obtained from Sigma-Aldrich Co. (ST. Louis, MO) and Thermo
Fisher Scientific Inc. (Waltham, MA). Fetal bovine serum (FBS) was purchased from Dainippon
Sumitomo Pharma Co., Ltd. (Osaka, Japan). Phosphate-buffered saline (PBS) was purchased
from Wako Pure Chemical Industries Ltd. (Osaka, Japan). Anti-CD31 monoclonal antibody,
anti-mouse IgG AlexaFluor488 and anti-mouse IgG AlexaFluor647 were purchased from Abcam
(Cambridge, UK), while anti-GLUT1 antibody was bought from Cosmo Bio Co., Ltd (Tokyo,
Japan). Tissue arrays of human head and neck cancer, and brain tumors were purchased from
Biomax (Rockville, MD). AlexaFluor555- and AlexaFluor647-NHS esters were purchased from
Thermo Fischer Scientific, Waltham, MA.
Cell lines and animals.
Human glioblastoma-astrocytoma (U87MG) cells, human colorectal carcinoma (HT29),
and triple negative breast cancer cells (MDA-MB-231) were purchased from ATCC (Manassas,
VA). Human squamous cell carcinoma cells (OSC-19) were obtained from Kanazawa University.
These cell lines were grown in Dulbecco’s Modified Eagle’s Medium (DMEM) plus 10 % FBS
o
in a humidified atmosphere containing 5 % CO2 at 37 C. BALB/C nu/nu female mice (female;
body weight, 18-20 g; age, 6 weeks) were obtained from Oriental Yeast Co., Ltd. and were
allowed to acclimatize for 7 days before inoculation of tumors. The animal experiments were
performed in accordance with the Guidelines for the Care and Use of Laboratory Animals as
stated by The University of Tokyo, as well as by the Animal Care and Use Committee of the
Innovation Center of Nanomedicine.

ACCEPTED MANUSCRIPT
Tumor models
6
The tumors were prepared by subcutaneous inoculation of OSC-19 (1x10 cells/mouse)
6
3
and U87MG (1x10 cells/mouse). The tumors were allowed to grow to approximately 50 mm
3
for the antitumor activity experiments, and 100 mm for the evaluation of the tumor
accumulation of free CDDP and the series of micelles. For the histological analysis, not only
OSC-19 and U87MG tumors were prepared, but also subcutaneous tumors HT-29 and MDA-
6
MB-231 after inoculating 1x10 cells/mouse. The tumors for the histology evaluation were
3
grown until they reached 100 mm .
Assessment of GLUT1 and CD31 colocalization by immunofluorescence
To study the presence of GLUT1 on the tumor vasculature, the tumors were collected and
fixed with 4% formaldehyde, incubated in 30% sucrose and embedded in Tissue-Tek OCT
Compound (Sakura-finetek Japan). Sections from each tumor were cut by a Cryostat (Leica
CM1950), and stained for GLUT1 by using anti-GLUT1 mouse antibody, which also cross-
reacts with human GLUT1, and endothelial cells of tumors by using anti-CD31 antibody, with
fluorescent secondary antibodies. Moreover, GLUT1 and CD31 were also stained in the tissue
arrays of human head and neck cancer, and brain tumors after retrieving the antigens. The
samples were imaged by using a confocal laser-scanning microscope (LSM780 Meta, Zeiss;
Germany). The colocalization of CD31 with GLUT1 was determined by using ImageJ in 3
samples of the murine xenografts, and 6 samples in the human tissue arrays.

ACCEPTED MANUSCRIPT
Measurements
Nuclear magnetic resonance spectra were recorded using JEOL JNM-ECS400 (400 MHz
1
13
for H and 100 MHz for C) instruments with CDCl3, d-DMSO, and D2O containing 0.05%
o
tetramethylsilane or trimethylsilyl propanoic acid at 22 C. Size exclusion chromatography
(SEC), which is a chromatographic technique that allows the separation of molecules in solution
based on their size, was used to determine the polydispersity index (PDI; Mw/Mn) of the
molecular weight of the block copolymers. SEC was performed at 40 °C using a Tosoh HLC-
8
220 GPC system equipped with a TSKgel G3000HHR column (linear, 7.8 mm × 300 mm; pore
4
size, 7.5 nm; bead size, 5 m; exclusion limit, 6 × 10 g/mol), a TSKgel G4000HHR column
5
(
linear, 7.8 mm × 300 mm; pore size, 20 nm; bead size, 5 m; exclusion limit, 4 × 10 g/mol),
and a TSKgel guard column HHR-L in DMF containing lithium bromide (10 mM) at a flow rate
of 0.5 mL/min. The number-averaged molecular weight (Mn) and PDI (Mw/Mn) of PEG
derivatives were calculated on the basis of the PEG calibration. The sizes of the polymeric
micelles were measured using zetasizers (Malvern Zetasizer Nano ZS90) equipped with a 4.0
mW He–Ne laser operating at 633 nm with 90° collecting optics. The zeta potential values of the
micelles were determined by laser Doppler electrophoresis in 10 mM phosphate buffer (pH 7.4)
by using Zetasizer. Data were analyzed using Malvern Dispersion Technology 4.20. Platinum
concentration of the micelles was determined by inductively coupled plasma-mass spectrometry
(ICP-MS) using a 7700x ICP-MS (Agilent Technologies, Santa Clara, CA; RF power, 1550 W;
sampling depth, 8.0 mm; plasma gas current, 16 L/min; carrier gas flow rate, 1.02 L/min;
peristaltic pump, 0.10 rps; monitoring mass, 195 (Pt); integration interval, 0.1 s; sampling period,
0
.31 s). Microscopic images of the micelles were obtained by transmission electron microscopy
(TEM; JEOL, Japan) at an accelerating voltage of 75 kV. The sample solutions (2 μL) were

ACCEPTED MANUSCRIPT
dropped onto a 400 mesh copper grids, and stained by deposition of a drop of an aqueous
solution of 50% ethanol and 2-wt% uranyl acetate. The size distribution of the micelles was
analyzed from the TEM images by using ImageJ.
Synthesis of 1,2-O-Isopropylidene-6-O-pivaloyl--D-glucofuranose (6-Piv-MIG)
MIG (25.0 g, 113.5 mmol) was dissolved in dry pyridine (150 mL) and pivaloyl chloride
(13.7 g, 113.6 mmol) was slowly added. Then, the mixture was stirred for 5 h at room
temperature. The solvent was evaporated to dryness and the residue was washed with water, and
filtered to obtain the crude product as solid. The crude product was dried under vacuum, and then
purified by recrystallization from ethyl acetate. Yield: 28.4 g (82.3 %); Rf = 0.25
o
1
(
dichloromethane:methanol = 9 :1 (v/v)); mp 145-147 C; H NMR (400 MHz, CDCl3):  5.96
(d, J =3.2 Hz, 1H), 4.55 (d, J = 3.6 Hz, 1H), 4.44 (dd, J = 5.6 Hz, J = 13.6 Hz 1H), 4.36 (d, J =
2
1
3
.0 Hz, 1H), 4.26-4.21 (m, 2H), 4.08 (dd, J = 2.8 Hz, J = 6.0 Hz 1H), 1.48 (s, 3H), 1.32 (s, 3H),
1
3
.23 (s, 9H); C NMR (100 MHz, CDCl3):  179.4, 111.8, 104.9, 85.1, 79.2, 75.4, 69.1, 66.1,
8.9, 27.1, 26.8, 26.2; The other characteristic data were matched with the literature data.
Synthesis of 1,2:3,5-Di-O-isopropylidene-6-O-pivaloyl--D-glucofuranose (6-Piv-DIG(6))
A mixture of 6-Piv-MIG (20.0 g, 65.7 mmol), p-toluenesulfonic acid monohydrate (500
mg, 2.90 mmol) and 2,2-dimethoxypropane (40 mL) was refluxed for 1 h. After 2,2-
dimethoxypropane was evaporated to dryness, the residue was dissolved to dichloromethane, and
was washed twice with saturated NaHCO3 and three times with brine. The dichloromethane layer

ACCEPTED MANUSCRIPT
was dried with anhydrous Na2SO4. After the solvent was removed by evaporator, the crude
product was purified by column chromatography on silica gel with hexane/ethyl acetate gradient
system (from 1/0 to 7/3 (v/v)) to give 6-Piv DIG(6) as a white solid. Yield: 10.3 g (45.5 %); Rf =
1
0
.81 (hexane : ethyl acetate = 7:3 (v/v)); H NMR (400 MHz, CDCl3):  6.00 (d, J = 3.6 Hz, 1H),
4
4
.58 (d, J = 4.0 Hz, 1H), 4.31 (dd, J = 3.2 Hz, J = 12.0 Hz, 2H), 4.21 (d, J = 3.6 Hz, 1H), 4.21-
.14 (m, 1H), 3.78 (m, 1H), 1.61 (s, 1H), 1.45 (s, 3H), 1.35 (s, 1H), 1.34 (d, J = 3.2 Hz, 2H), 1.21
1
3
(
s, 9H); C NMR (100 MHz, CDCl3):  178.1, 112.2, 106.4, 101.0, 83.9, 79.5, 75.0, 70.3, 64.1,
3
8.8, 27.2, 26.6, 24.0, 23.8; The other characteristic data were matched with the literature data.
Synthesis of 1,2:3,5-Di-O-isopropylidene--D-glucofuranose (DIG(6)) (Figure S1A)
A solution of 6-Piv DIG(6) (5.0 g, 14.5 mmol) in methanol (100 mL) including 1 M
sodium methoxide was stirred during 24 h. The reaction mixture was neutralized by addition of
dry ice. The mixture was extracted three times with dichloromethane, dried with anhydrous
Na2SO4, filtered, and evaporated. The oily residue was purified by flash chromatography
1
(
hexane/ethyl acetate, 3:2 v/v, Rf = 0.29) to afford DIG(6). Yield: 2.80 g (74.1 %). H NMR (400
MHz, CDCl ):  6.00 (d, J = 4.0 Hz, 1H), 4.59 (d, J = 3.6 Hz, 1H), 4.38 (dd, J = 3.6 Hz, J = 7.2
3
Hz, 1H), 4.19 (d, J = 3.6 Hz, 1H), 3.87-3.82 (m, 1H), 3.72-3.63 (m, 1H), 2.03 (t, J = 6.4 Hz, 1H),
1
3
1
7
.49 (s, 3H), 1.37 (s, 6H), 1.33 (s, 3H); C NMR (100 MHz, CDCl3):  112.2, 106.4, 101.0, 84.0,
9.0, 75.0, 63.4, 27.1, 26.5, 24.1, 24.0; The other characteristic data were matched with the
literature data.

ACCEPTED MANUSCRIPT
Polymerization of ethylene oxide (Figure S1B)
A typical polymerization of ethylene oxide (EO) was carried out under an argon
atmosphere according to a previously reported method with a minor modification [26]. Briefly,
the pre-dried glucose-based initiator, DIG(6) (390 mg, 1.50 mmol), potassium naphthalene (1.50
mmol), and dry THF (100 mL) were added into a reaction glassware equipped with a three-way
stopcock. After stirring for 15 min, EO (18.9 g, 428 mmol) was added to the solution via a
o
cooled cannula. The polymerization was carried out at 25 C. The polymerization mixture was
poured into a large amount of diethyl ether. The precipitate was purified by reprecipitation with
dichloromethane/diethyl ether and then dried in vacuo to give -DIG(6)-ω-hydroxy-PEG (17.0 g,
yield = 98 %) in a 99 % conversion. The Mn and Mw/Mn values were 12800 g/mol and 1.03,
1
respectively. H NMR (400 MHz, CDCl3):  5.99 (d, J = 3.6 Hz, 1H), 4.57 (d, J = 3.6 Hz, 1H),
4
3
.32 (dd, J = 3.6 Hz, J = 7.6 Hz, 1H), 4.19 (d, J = 3.6 Hz, 1H), 3.90-3.34 (m, 1086H), 1.49 (s,
H), 1.37 (s, 3H), 1.36 (s, 3H), 1.32 (s, 3H).
Preparation of α-DIG(6)-ω-amino-poly(ethylene glycol) (DIG(6)-PEG-NH2) (Figure S1B)
α-DIG(6)-ω-amino-poly(ethylene glycol) was synthesized as described previously [26].
Briefly, a solution of MsCl (0.32 mL, 14.1 mmol) in THF (200 mL) was dropwisely added into a
mixture of TEA (0.89 mL, 6.39 mmol) and DIG(6)-PEG-OH (19.0 g, 1.48 mmol) in THF (120
mL). The reaction mixture was stirred in a dry argon atmosphere at 0 C for 30 min and at room
temperature for overnight. The solution was concentrated and poured into a large amount of cold
diethyl ether. The precipitate was purified by reprecipitation with dichloromethane/diethyl ether,
and dried under reduced pressure to yield α-DIG(6)-ω-methanesulfonyl-terminated poly(ethylene

ACCEPTED MANUSCRIPT
1
glycol) (DIG(6)-PEG-Ms; Yield = 93 %, H NMR (400 MHz, CDCl3):  5.99 (d, J = 3.6 Hz, 1H),
4
.57 (d, J = 3.6 Hz, 1H), 4.39-4.37 (m, 2H), 4.32 (d, J = 3.6 Hz, J = 7.2 Hz, 1H), 4.20 (d, J = 3.6
Hz, 1H), 3.89-3.40 (m, 1090H), 3.09 (s, 2H), 1.49 (s, 3H), 1.37 (s, 3H), 1.36 (s, 3H), 1.32 (s,
H)). Then, DIG(6)-PEG-Ms (15.0 g, 1.17 mmol) was dissolved in 28% NH3 aq. (150 mL) and
3
stirred for 3 days at room temperature. The solution was concentrated by evaporation and then
dialyzed against distilled water, followed by lyophilization to yield -DIG(6)--amino-
poly(ethylene glycol) (DIG(6)-PEG-NH2). The obtained DIG(6)-PEG-NH2 was further purified
with an ion-exchange CM Sephadex C-50 column (GE Healthcare UK Ltd., Buckinghamshire,
1
UK) before use. Yield = 89 %, Conv.-NH2 = 97 %, Mn = 12,900, Mw/Mn = 1.04. H NMR (400
MHz, CDCl3):  5.99 (d, J = 3.6 Hz, 1H), 4.57 (d, J = 3.6 Hz, 1H), 4.32 (dd, J = 4.0 Hz, J = 7.6
Hz, 1H), 4.20 (d, J = 3.6 Hz, 1H), 3.92 (t, J = 4.8 Hz, 2H), 3.81-3.45 (m, 1090H), 3.17 (t, J = 4.8
Hz, 2H), 1.49 (s, 3H), 1.37 (s, 3H), 1.36 (s, 3H), 1.33 (s, 3H).
Preparation of α-glucopyranos-6-O-yl-poly(ethylene glycol)-poly(L-glutamic acid) block
copolymer (Gluc-PEG-P(Glu)) (Figure S1B)
A typical procedure for the polymerization of N-carboxy anhydride of -benzyl L-
glutamate is as follows: A solution of N-carboxy anhydride of -benzyl L-glutamate (0.471 g,
1
.79 mmol) in DMSO (15.0 mL) and a solution of DIG(6)-PEG-NH2 (0.494 g, 0.0388 mmol) in

DMSO (40.0 mL) were combined and stirred in a dry argon atmosphere at 25 C. After 72 h, the
solution was poured into a large amount of cold diethyl ether. The precipitate was purified by
reprecipitation with dichloromethane/diethyl ether, and dried under reduced pressure to yield -
DIG(6)-poly(ethylene glycol)-poly(-benzyl L-glutamate) (DIG(6)-PEG-PBLG; yield = 89%, Mn

ACCEPTED MANUSCRIPT
=
21,700; Mw/Mn = 1.07; NMR (400 MHz, d-DMSO):  7.39-7.10 (m, 190H), 5.93 (d, J = 3.6 Hz,
1
3
3
H), 5.20-4.80 (m, 76H), 4.54 (d, J = 3.6 Hz, 1H), 4.40-4.20 (m, 39H), 4.13 (d, J = 3.6 Hz, 1H),
.90 (br, 2H), 3.80-3.30 (m, 1090H), 2.70-1.70 (m, 154H*), 1.39 (S, 3H), 1.31 (s, 3H), 1.28 (s,
H), 1.26 (s, 3H) (*Due to peak overlapping with DMSO, the proton number was calculated by
NMR result performed with CDCl3). To deprotect benzyl groups on polymer backbone, the
obtained block copolymer was stirred in 0.5 N NaOH aq. (31 mL) for overnight at room
temperature. The polymer was purified in water using a dialysis membrane (Spectra/Pro 6
Membrane: MWCO, 3500), followed by lyophilization to yield DIG(6)-PEG-P(Glu). To
deprotect isopropylidene acetal groups, the DIG(6)-PEG-P(Glu) was stirred in TFA aq.
(water/TFA = 1/4, vol/vol, 10 mL) for 30 min at room temperature. The polymer was purified in
water using a dialysis membrane (Spectra/Pro 6 Membrane: MWCO, 3500), followed by
lyophilization to yield α-glucopyranos-6-O-yl-PEG-P(Glu) (Gluc-PEG-P(Glu): Yield = 89%, Mn
1
=
18,100, Mw/Mn = 1.09. H NMR (400 MHz, D2O with TMS):  5.26 (m, 1H), 4.50-4.24 (m,
3
8H), 4.25-4.08 (m, 4H), 4.05-3.30 (m, 1094H), 3.17 (m, 1H), 2.60-1.70 (m, 156H).
Preparation of glucose-installed cisplatin-loaded micelles (Gluc-CDDP/m)
Cisplatin-loaded polymeric micelles were prepared according to the previously described
synthetic method with a slight modification [27,28]. The glucose densities on the micelles were
tuned to 0, 25 and 50% by controlling the ratio between MeO-PEG-P(Glu), and Gluc-PEG-
P(Glu) block copolymers, resulting in micelles with 0, 25 and 50% of the number of their PEG
chains modified with glucose moieties. For example, the typical procedure for the preparation of
α-glucopyranos-6-O-yl-conjugated micelles (Gluc-CDDP/m) containing 25 mol% glucose

ACCEPTED MANUSCRIPT
residues on their surface was as follows: a solution of CDDP (16.5 mg, 0.055 mmol) in water
(10.2 mL) was stirred for 1 h at 50 °C. The obtained solution (10 mL) was filtered using a
membrane (pore size, 0.22 m) and added to a reaction vial containing MeO-PEG-P(Glu) and
Gluc-PEG-P(Glu) ([Glu] = 5 mmol/L, [CDDP]/[Glu] = 1.0 (mol/mol), [MeO-PEG-
P(Glu)]/[Gluc-PEG-P(Glu)] = 0.75/0.25 (mol/mol)). The reaction mixture was stirred for 120 h
at 37 C, and then purified by dialysis using the cellophane tube (Spectra/Pro 6 Membrane:
MWCO, 3500). The micelle stability and drug release were evaluated by DLS, and by the
dialysis method in 10 mM PBS plus 150 mM NaCl at 37 °C, respectively.
In vitro cytotoxicity assay
OSC-19 or U87MG spheroid arrays were prepared by culturing these cells in DMEM
containing 10% FBS on a 96-well micro-patterned multiplate designed for spheroid preparation,
4
Cell-able (Toyo Gousei Co., Ltd., Tokyo, Japan, 2.0 × 10 cells/well). Stable spheroid arrays are
confirmed to form after 4 days of culture, and subsequently, they were incubated with free
CDDP or the series of polymeric micelles at different CDDP-based concentrations ranging from
1
00- to 0.001-μM. After 6 h incubation, the spheroids were washed with fresh medium, and post-
incubated for 60 h. The 50% growth inhibitory concentrations (IC50) were evaluated by a
CellTiter-Glo 3D Cell Viability Assay kit (Promega Corporation, Madison, USA), following the
manufacturer's protocol. In addition, the GLUT1 in OSC-19 cells was knocked down by using
Dharmacon’s lentiviral shRNA kit (Horizon; Cambridge, UK) by following the specifications of
the provider for producing cells without GLUT1 as a negative control. These GLUT1(-) OSC-19
cells were used to prepare spheroids, and the cytotoxicity of free CDDP or the series of

ACCEPTED MANUSCRIPT
polymeric micelles was studied as described above. The experiments were repeated 4 times, and
the average IC50 values were calculated.
Antitumor activity
To evaluate the antitumor efficacy, BALB/c nude mice (Oriental Yeast Co., Ltd., Tokyo,
Japan) bearing OSC-19 or U87MG subcutaneous tumor (n = 6) were treated 4 times
intravenously every four days with 1 mg/kg (on a CDDP basis) of polymeric micelles. The tumor
2
size (V) was estimated using the following equation: V = a × b /2, where a and b are the major
and minor axes of the tumor measured by using a caliper. Student’s t-test was used to analyze the
significance in the tumor volume between the experimental and control groups. P < 0.05 was
considered statistically significant in each analysis. In addition, the body weight of mice was
followed during the experiment.
Plasma clearance and biodistribution of free CDDP and polymeric micelles
To evaluate the plasma clearance and biodistribution, BALB/c nude mice bearing OCS19
tumors (n = 6) were intravenously injected with a series of polymeric micelles or free CDDP at
1
00 µg/mouse on a CDDP basis. Tumors were removed at 0.2, 1, 4, 8 and 24 h after injection,
washed with PBS, and decomposed with concentrated nitric acid at 200 C. The dried samples
were dissolved in 1 vol.% nitric acid (1.0 mL), and the platinum concentrations were measured
by ICP-MS. For analysis of plasma clearance, blood was collected from the inferior vena cava,
heparinized, and centrifuged to obtain the plasma. These plasma samples were treated with nitric
acid as described above and analyzed by ICP-MS. Moreover, the inhibition of the accumulation
of the Gluc-25%-CDDP/m was done by intraperitoneally injecting a 50 wt% glucose solution
(200 µL) in OSC-19 tumor-bearing mice before and 30 min after the micelle administration, or

ACCEPTED MANUSCRIPT
by intraperitoneally injecting the GLUT1 inhibitor STF-31 (15 mg/kg; 50 µL) at 48 h, 24 h and 5
min before the micelle administration. The tumors were collected at 1 and 4 h after the micelle
injection, and the drug concentration in the tissues was analyzed as described above.
Intravital microscopy
For this experiment, we prepared fluorescent labeled micelles. Thus, Alexa 647-NHS or
Alexa 555-NHS were conjugated to the primary amino group at the end of the poly(amino acid)
segment of α-CH3O-PEG-P(Glu) block copolymer. The fluorescent labeled polymers were
purified by using disposable PD10 columns. Then, CDDP/m were prepared by using Alexa 647-
labeled α-CH3O-PEG-P(Glu) block copolymers, while Gluc-25%-CDDP/m were prepared by
using 75% of Alexa 555-labeled α-CH3O-PEG-P(Glu) and 25% of Gluc-PEG-P(Glu) block
copolymers. The intravital CLSM was performed according to the previously reported method
3
[
29]. OSC-19 tumors were grown subcutaneously until they reached 100 mm . The tumors were
then exposed by making a 4-cm surgical incision without damaging the vessels in the
surroundings. Then, the skin flap covering the tumor was folded and the tumor was set under the
objective lenses of a Nikon A1R CLSM system connected to an upright ECLIPSE FN1 (Nikon,
Japan). Ten minutes before imaging, the mice received an intravenous injection of 100 μL
Hoechst (1 mg/mL) to stain the nuclei of the cells in the tumors. Then, Alexa 647-labeled
CDDP/m and Alexa 555-labeled Gluc-25%-CDDP/m at a dose of 10 mg/kg on a CDDP basis
were intravenously co-injected via the tail vein. The Alexa 647-labeled CDDP/m and the Alexa
5
55-labeled Gluc-25%-CDDP/m were detected by using 633/670 nm and 560/620 nm

ACCEPTED MANUSCRIPT
excitation/emission filters, respectively. Time-lapse snapshots of the tumor were taken every 15
min for 3 h. The images were then analyzed by using ImageJ.
RESULTS
Expression of GLUT1 on tumor vasculature
The expression of GLUT1 on the endothelial cells of the blood vessels of tumors was
first studied to validate the availability of the target in vivo. Thus, we performed
immunofluorescence staining of GLUT1 (Figure 2A-C; green) and CD31 (Figure 2A-C; red)
in tissue samples from both OSC-19 and U87MG tumors, as well as from tissue microarrays of
human head and neck tumors, and brain tumors, by using anti-GLUT1 and anti-CD31 antibodies,
and fluorescent-labeled secondary antibodies. The colocalization of GLUT1 and CD31 is shown
in yellow color in the microscopic images (Figure 2A-C). The histology analysis demonstrated
high colocalization of GLUT1 and CD31 in the tumor models (Figure 2A and 2B), with a
colocalization ratio of approximately 90% (Figure 2D). It is also worth noting the contrasting
GLUT1 expression in OSC-19 and U87MG tumors, as shown by the GLUT1 staining of the
interstitium of the tumors (Figure 2A and 2B; green). In the human tissue array samples, a high
level of colocalization between GLUT1 and CD31 was detected, which corresponded with the
observations in the murine tumor models (Figure 2C and 2D). Interestingly, benign head and
neck lesions (Schwannomas) showed less GLUT1 expression both in the blood vessels and in the
interstitium (Figure 2C and 2D). In addition, the colocalization of GLUT1 and CD31 was
detected in xenografts of colon (HT-29) and breast (MDA-MD-231) cancers (Supplementary
Figure S2). Thus, these results underline the occurrence of GLUT1 in the vessels of tumors, and

ACCEPTED MANUSCRIPT
suggest the potential of GLUT1 as a target for glucose-installed nanomedicines directed to
overcome the vascular barrier.
Figure 2. Colocalization of blood vessels (red) and GLUT1 (green) in murine tumor models and
tissue microarrays of human tumors, determined by fluorescence immunohistochemistry. A.
Murine model of high-GLUT1-expressing human squamous cell carcinoma of the head and neck
(OSC-19). B. Murine model of glioblastoma-astrocytoma (U87MG). Scale bar = 100 μm. C.
Representative human tissues of head and neck tumor and brain tumors. Scale bar = 500 μm. D.
Quantification of colocalization of CD31 and GLUT1 in the mouse tumor models. Data are
shown as the mean ± S.D. (n = 3). E. Quantification of colocalization of CD31 and GLUT1 in

ACCEPTED MANUSCRIPT
the human samples. Data are shown as the mean ± S.D. (n = 6). The cell nuclei were stained by
using Hoechst (blue).
Construction of glucose-installed cisplatin-loaded micelles
Motivated by the presence of GLUT1 in the lumen of the tumor vessels, we proceeded to
develop CDDP/m having glucose moieties on their surface. It is important to note that the design
of the glucose ligands to enhance cell uptake and in vivo delivery should consider the structural
features of GLUTs for avoiding modifications on the hydroxyls of glucose implicated in
hydrogen bonding interactions with the transporters [17]. Previous reports indicate that the
hydroxyl groups at carbon 1 and carbon 3, as well as the pyran oxygen 5 in the closed
conformation of glucose, are involved in stabilizing hydrogen bonds with the amino acid
residues in GLUT1 [30,31], while the hydroxyl groups of glucose at carbon 2, carbon 4 and
carbon 6 are less related to the recognition of glucose by GLUT1. Thus, in this study, we
prepared acetalized glucose-installed PEG (Mw: 12,000) via the carbon 6 position (DIG(6)-PEG)
through ring-opening polymerization of ethylene oxide by using a glucose-based initiator
(potassium salt of DIG(6)) (Figure 1 and Supplementary Figure S1). Hydroxy group at the ω-
end of DIG(6)-PEG was subsequently functionalized to primary amino group to initiate the ring
opening polymerization of N-carboxy anhydride of γ-benzyl L-glutamate for preparing DIG(6)-
PEG-poly(γ-benzyl L-glutamate) block copolymer (DIG(6)-PEG-PBLG). Acetal groups in
DIG(6) as well as benzyl groups in the side chain of DIG(6)-PEG-PBLG were then deprotected
to obtain Gluc-PEG-poly(glutamic acid) block copolymer (Gluc-PEG-P(Glu)). Degree of

ACCEPTED MANUSCRIPT
polymerization (DP) of Glu units in the block copolymers was adjusted in the range of 36~38.
The characteristics of the Gluc-PEG-P(Glu) are summarized in Supplementary Table 1.
A series of Gluc-CDDP/m with varying degree of glucose density on their surface (0, 25
and 50 mol%) were prepared by adjusting the ratio of Gluc-PEG-P(Glu) and α-CH3O-PEG-
P(Glu) block copolymers, and mixing with cisplatin in aqueous solution (Figure 1). The micelles
were self-assembled in aqueous milieu after the formation of coordination bonds between the
carboxylate groups in the block copolymers and the Pt atom in cisplatin (Figure 1) [27,28]. The
diameter and surface charge of the Gluc-CDDP/m were found to be comparable to those of non-
functionalized CDDP/m (CDDP/m), i.e. approximately 30 nm with similar size distribution
(Supplementary Figure S3), and close to neutral, respectively (Table 1). Moreover, TEM
analysis of the micelles was done after negative staining with 2-wt% uranyl acetate. The TEM
images showed that CDDP/m and Gluc-CDDP/m present comparable spherical morphology and
narrow size distribution with an average diameter of approximately 14 nm (Supplementary
Figure S4). It is worth noting that the PEG shell is not visible after uranyl acetate staining by
TEM, suggesting that the size distribution obtained from the TEM images corresponds to that to
the core of the micelles. In addition, all the micelles showed a comparable drug loading of
approximately 30% in weight (Table 1).
Table 1. Structural features of the series of polymeric micelles
a
a
b
Micelle
Glucose density (% of
total PEG chains)
Size
PDI
Z-potential (mV)
Drug loading
(%weight)
c
(nm)
CDDP/m
0
31
29
0.13
0.13
-3.1
-3.2
29
31
Gluc- CDDP/m
25

ACCEPTED MANUSCRIPT
50
28
0.13
-1.4
29
a
Determined by dynamic light scattering.
Determined by laser Doppler electrophoresis in 10 mM phosphate buffer(pH 7.4)
Determined by inductively coupled plasma-mass spectrometry.
b
c
The stability of the micelles was evaluated in physiological saline (10 mM PBS plus 150
mM NaCl) at 37 °C by following the changes in the size, PDI and scattering light intensities. The
size and polydispersity index of all the micelles was maintained throughout the experiment
(Figure 3A and 3B), correlating with our recent observation of an erosion-like dissociation of
CDDP/m in physiological conditions [28]. On the other hand, the scattering light intensities of
Gluc-CDDP/m decreased to approximately 10% of the initial intensity after 96 h, and the decay
profiles were comparable to that of CDDP/m (Figure 3C). In addition, the drug release rate of
the series of micelles was evaluated by using the dialysis method under the same conditions of
the stability experiments, i.e. 10 mM PBS plus 150 mM NaCl at 37 °C. The results showed that
all the micelles have comparable drug release rate (Figure 3D). Therefore, glucose installation
on CDDP/m did not affect the characteristics and behavior of the micelles in physiological
conditions, allowing us to compare the different formulations in biological settings.

ACCEPTED MANUSCRIPT
Figure 3. Stability and drug release rate of CDDP/m and Gluc-CDDP/m in 10 mM PBS plus 150
mM NaCl. A. Z-average diameter of the micelles. B. Polydispersity index (PDI). C. Relative
light scattering intensity. D. Cumulative drug release. Data are shown as the mean ± S.D. (n = 3).
In vitro activity of glucose-installed cisplatin-loaded micelles
The biological activity of glucose-conjugated polymeric micelles was evaluated in vitro
against three-dimensional multicellular spheroids, as spheroids offer significant advantages over
monolayer cell culture for predicting in vivo efficacy, including the replication of several
microenvironmental features of tumors, such as nutrient and oxygen supplies, drug gradients,
three-dimensional multilayered cellular network, and intratumoral pH [32,33]. High-GLUT1-

ACCEPTED MANUSCRIPT
expressing OSC-19 and low-GLUT1-expressing U87MG were selected for this study as cell
lines with contrasting expression of GLUT1. The cells were incubated in spheroid microarray
plates to form spheroids with 100-μm diameter, which were then incubated with free cisplatin
and the series of polymeric micelles for 6 h, washed with fresh medium and post-incubated for
6
0 h. The results indicated that the Gluc-CDDP/m has a trend of higher cytotoxicity compared to
CDDP/m against OCS-19 spheroids (Figure 4; black bars). These results indicate the
importance of controlling the density of the glucose-ligand and the conjugation scheme for
enhancing the activity of the micelles.
Knocking down GLUT1 on OCS19 cells (GLUT1(-) OCS19 cells) decreased the activity
of the Gluc-CDDP/m against the OCS19 spheroids (Figure 4; grey bars). Moreover, in U87MG
spheroids, the IC50 values for all the micelle systems were comparable regardless of the glucose
functionalization and conjugation scheme (Figure 4; white bars). Thus, these results strongly
suggest that the cytotoxicity of the Gluc-CDDP/m was mediated by GLUT1-targeting on the
cancer cells. In addition, it is worth noting that even though the cell culture media used in these
experiments contains high glucose concentration (4 g/L), the Gluc-CDDP/m showed enhanced
cytotoxicity against OSC-19 spheroids.

ACCEPTED MANUSCRIPT
Figure 4. In vitro cytotoxicity of free CDDP, CDDP/m and Gluc-CDDP/m against multicellular
spheroids of OSC-19 and U87MG cells. Data are shown as the mean ± S.D. (n = 4). *p < 0.05
determined by Student’s t-test.
In vivo activity of glucose-installed cisplatin-loaded micelles
The activity of free cisplatin and the series of CDDP/m was studied in vivo after
systemically injecting into mice bearing subcutaneous OSC-19 tumors at 1 mg/kg on a cisplatin
basis on days 0, 4, 8, and 11 (Figure 5 A). At day 36 after the start of the experiment, the volume
of the tumors treated with Gluc-25%-CDDP/m was more than 3-fold smaller than that of the
groups treated with free cisplatin and CDDP/m. Moreover, the Gluc-25%-CDDP/m was more
effective than Gluc-50%-CDDP/m groups, indicating that the glucose conjugation scheme and
the glucose density are critical for improving the efficacy. The significant antitumor activity of
Gluc-25%-CDDP/m against OSC-19 tumors is also worthy of note, since these tumors are highly

ACCEPTED MANUSCRIPT
resistant to CDDP, and present a large population of recalcitrant cancer cells with cancer stem-
like cell properties [23]. The efficacy of the micelles was further studied in subcutaneous
U87MG tumors, by following the same dosing and schedule as in the mice with OSC-19 tumors.
Interestingly, despite the low GLUT1 expression of U87MG tumors (Figure 2B) [24], Gluc-
2
5%-CDDP/m also inhibited the tumor growth more effectively than free cisplatin, CDDP/m,
and Gluc-50%-CDDP/m (Figure 5B). Moreover, during the experiments, the body weight of
mice remained unchanged, which indicated the safety of the treatments (Figure 5C).

ACCEPTED MANUSCRIPT
Figure 5. Antitumor activity of free CDDP and the series of micelles. A. OSC-19 tumors. B.
U87MG tumors. C. Body weight changes during the antitumor activity experiment against OSC-
1
9 tumors. The drugs were administered intravenously at 1 mg/kg on a cisplatin basis on days 0,

ACCEPTED MANUSCRIPT
4
, 8, and 11. Data are shown as the mean ± S.E.M (n = 6). *p < 0.05 determined by Student’s t-
test.
The activity enhancement of Gluc-25%-CDDP/m against these tumors suggests that it
may present enhanced accumulation in the tumors compared to the other formulations. Thus, to
confirm the contribution of the glucose ligands on the drug delivery efficiency, we evaluated the
plasma clearance of free CDDP and the series of micelles, as well as their distribution in major
organs and tumors (Figure 6). The amount of the injected dose present in the plasma and tissues
was normalized to the total injected dose, i.e. 100 μg, and expressed as the % of injected dose per
mL of plasma and g of tissue, respectively. The profiles of plasma clearance of CDDP/m and
Gluc-25%-CDDP/m were comparable (Figure 6A) with approximately 88% dose/mL in plasma
at 1 h after administration and 10% dose/mL at 24 h post-injection. The area under the plasma
clearance curve (AUCPlasma/0-24 h) for Gluc-25%-CDDP/m was on part with that of CDDP/m
(Table 2). Moreover, these tendencies are also in agreement with the plasma clearance of
CDDP/m from our previous reports [27,28], indicating that the 25% modification of the PEG on
the micelles with Gluc did not affect their blood circulation. Moreover, the comparable
concentration of CDDP/m and Gluc-25%-CDDP/m in plasma suggests that Gluc-25%-CDDP/m
did not undergo strong binding with GLUT1 on erythrocytes. On the other hand, increasing the
density of Gluc to 50% decreased the level of the micelles in the bloodstream (Figure 6A).
These results were also evident from the AUCPlasma/0-24 h (Table 2), and indicate that the density
of glucose on the micelles has significant effects on the clearance and bioavailability of the Gluc-
CDDP/m.

ACCEPTED MANUSCRIPT
The distribution of free CDDP and the micelles was also assessed in normal tissues,
including liver, kidney, spleen, muscle, heart and brain, which express a wide range of glucose
transporters [34]. In liver, Gluc-50%-CDDP/m displayed higher accumulation levels than
CDDP/m at all the examined time points, whereas the accumulation of Gluc-25%-CDDP/m was
comparable to that of CDDP/m (Figure 6B). The drop in the blood concentration of the micelles
with 50% glucose correlated with their accumulation in liver, which was higher than that of the
micelles having 0% and 25% glucose, suggesting that at 50% glucose installation, the micelles
cannot overcome the liver barrier, leading to the drop in blood concentration. These observations
are further supported by the area under the liver accumulation curves (AUCLiver/0-24 h) determined
by trapezoidal rule (Table 2). These results indicate that appropriately tuning the glucose density
on the micelles is key for overcoming the liver barrier, probably by reducing the interactions of
Gluc-CDDP/m with the various GLUTs present in the intravascular compartment of liver [35],
including GLUT1 in the sinusoidal membrane of hepatocytes, endothelial cells and Kupffer cells,
GLUT2 in hepatocytes, GLUT3 in hepatocytes and GLUT4 in stellate cells. Moreover, the
modification of the surface of the micelles with glucose did not affect their accumulation in
kidney, spleen, muscle, heart and brain (Figures 6C, 6D, 6E, 6F and 6G, respectively), as all
the micelles showed comparable low levels and AUCs (Table 2). The low accumulation of Gluc-
CDDP/m to the brain, which was comparable to the minimal values observed for muscle, is
particularly notable given the high expression of GLUT1 on the endothelium of the blood–brain
barrier [36], and excludes any concern on off-target brain toxicities.
The accumulation of the micelles was also evaluated in both OSC-19 and U87MG
xenografts. Accordingly, Gluc-25%-CDDP/m showed rapid accumulation in both tumors within
1
-4 h after administration, and achieved intratumoral levels that doubled those of CDDP/m and