A new class of 5-fluoro-2′-deoxyuridine prodrugs conjugated with a tumor-homing cyclic peptide CNGRC by ester linkers: Synthesis, reactivity, and tumor-cell-selective cytotoxicity

Article abstract of DOI:10.1007/s11095-004-1875-x

Purpose. Tumor-targeting prodrugs of 5-fluoro-2′-deoxyuridine (5-FdUrd), which are chemical conjugations of 5-FdUrd with a tumor-homing cyclic peptide CNGRC by succinate and glutarate linkers, were synthesized to investigate the structural effects of linkers on the hydrolytic release of 5-FdUrd and the tumor-cell-selective cytotoxicity. Methods. A solid phase synthesis method was used to produce 5-FdUrd prodrugs. The kinetics and efficiency of hydrolytic 5-FdUrd release from the prodrugs were investigated in phosphate buffer (PB), fetal bovine serum (FBS), HT-1080 cell lysate, MDA-MB-231 cell lysate, and MEM containing 10% FBS. The tumor-cell-selective cytotoxicity of prodrugs was evaluated by an MTT method. Results. Two tumor-targeting prodrugs CNF1 and CNF2 bearing 5-FdUrd conjugated with a common cyclic peptide CNGRC by succinate and glutarate linkers, respectively, and their control compounds CN1 and CN2 without 5-FdUrd moiety were synthesized and identified. CNF1 underwent hydrolysis to release 5-FdUrd more rapidly and efficiently than CNF2. Both prodrugs were of lower cytotoxicity compared to 5-FdUrd, showing more selective cytotoxicity toward APN/CD13 positive cells (HT-1080) than toward APN/CD13 negative cells (HT-29, MDA-MB-231). Conclusions. A new class of tumor-targeting 5-FdUrd prodrugs CNF1 and CNF2 were successfully synthesized. These prodrugs targeted a tumor marker APN/CD13 to cause tumor-cell-selective cyctotoxicity due to 5-FdUrd release, the rate of which could be controlled by the structure of ester linker.

Full text of DOI:10.1007/s11095-004-1875-x

Pharmaceutical Research, Vol. 22, No. 3, March 2005 (© 2005)
DOI: 10.1007/s11095-004-1875-x
Research Paper
A New Class of 5-Fluoro-2؅-deoxyuridine Prodrugs Conjugated with a
Tumor-Homing Cyclic Peptide CNGRC by Ester Linkers: Synthesis,
Reactivity, and Tumor-Cell–Selective Cytotoxicity
Zhouen Zhang,1 Hiroshi Hatta,1 Kazuhito Tanabe,1 and Sei-ichi Nishimoto1,2
Received May 27, 2004; accepted November 29, 2004
Purpose. Tumor-targeting prodrugs of 5-fluoro-2Ј-deoxyuridine (5-FdUrd), which are chemical conju-
gations of 5-FdUrd with a tumor-homing cyclic peptide CNGRC by succinate and glutarate linkers, were
synthesized to investigate the structural effects of linkers on the hydrolytic release of 5-FdUrd and the
tumor-cell–selective cytotoxicity.
Methods. A solid phase synthesis method was used to produce 5-FdUrd prodrugs. The kinetics and
efficiency of hydrolytic 5-FdUrd release from the prodrugs were investigated in phosphate buffer (PB),
fetal bovine serum (FBS), HT-1080 cell lysate, MDA-MB-231 cell lysate, and MEM containing 10%
FBS. The tumor-cell–selective cytotoxicity of prodrugs was evaluated by an MTT method.
Results. Two tumor-targeting prodrugs CNF1 and CNF2 bearing 5-FdUrd conjugated with a common
cyclic peptide CNGRC by succinate and glutarate linkers, respectively, and their control compounds
CN1 and CN2 without 5-FdUrd moiety were synthesized and identified. CNF1 underwent hydrolysis to
release 5-FdUrd more rapidly and efficiently than CNF2. Both prodrugs were of lower cytotoxicity
compared to 5-FdUrd, showing more selective cytotoxicity toward APN/CD13 positive cells (HT-1080)
than toward APN/CD13 negative cells (HT-29, MDA-MB-231).
Conclusions. A new class of tumor-targeting 5-FdUrd prodrugs CNF1 and CNF2 were successfully
synthesized. These prodrugs targeted a tumor marker APN/CD13 to cause tumor-cell–selective cycto-
toxicity due to 5-FdUrd release, the rate of which could be controlled by the structure of ester linker.
KEY WORDS: anticancer prodrug; cytotoxicity; 5-fluoro-2Ј-deoxyuridine; hydrolysis; tumor-homing
peptide.
INTRODUCTION
(12). These drug delivery systems consist of nanoparticles that
can readily be transported to the well vascularized region, but
not to the ischemic region with less vascularization, of tumor
tissues (13–15). In a different drug delivery system using a
tumor-cell–targeting antibody of foreign origin, immune re-
sponse from humans/animals will cause rapid elimination of
the antibody. Furthermore, because the gene of tumor cell is
quickly mutated without interruption to lose the characteris-
tic epitope, the tumor-cell–targeting antibody will become
practically inactive (16,17).
Currently, particular attention has been given to the tu-
mor vascular targeting strategy originated from the special-
ization of tumor vasculature (18,19). The tumor vasculature is
morphologically abnormal and carries various types of tumor
molecular markers that can be used to discriminate tumor
vessels from normal vasculature: for example, integrin ␣v␤3,
␣v␤5, and aminopeptidase N (APN/CD13) (18,20). As dem-
onstrated by the in vivo phage display technology, some of
these markers could be recognized by certain peptides and
antibody fragments (20–23). Although these findings would
stimulate development of a new class of tumor-targeting an-
ticancer prodrugs (24,25), there are still few studies on the
application of such a strategy. In this view, an attempt was
herein made to develop tumor-targeting prodrugs of
5-FdUrd, employing chemical conjugation with a tumor-
Among the cytotoxic anticancer drugs, 5-fluoro-2Ј-
deoxyuridine (5-FdUrd) plays a considerable function in the
treatment of metastatic cancers (1). While exerting an appro-
priate antitumor ability via blocking thymidylate synthase to
inhibit DNA synthesis and/or incorporation of its metabolites
into DNA or RNA (2,3), 5-FdUrd produces various side ef-
fects including nonspecific toxicity toward rapidly proliferat-
ing normal tissues (1). Because of such pharmacological prop-
erties, efforts have been made to develop a variety of pro-
drugs undergoing hydrolysis to release 5-FdUrd: for example,
ester derivatives (4,5) and phosphoramidate derivatives of
5-FdUrd (6). There is also a different family of prodrugs that
are activated to release 5-FdUrd by ionizing radiation (7,8) or
UV light (9). For more sophisticated current 5-FdUrd pro-
drugs, targeting strategies are used to improve the efficiency
of drug delivery into tumor tissues: for example, packaging of
5-FdUrd or its prodrugs into the nanoparticles (10,11), and
conjugation of 5-FdUrd with tumor-cell–targeting antibodies
1 Department of Energy and Hydrocarbon Chemistry, Graduate
School of Engineering, Kyoto University, Katsura Campus, Nishikyo-
ku, Kyoto 615-8510, Japan.
2 To whom correspondence should be addressed. (e-mail: nishimot@
scl.kyoto-u.ac.jp)
381
0724-8741/05/0300-0381/0 © 2005 Springer Science+Business Media, Inc.

382
Zhang, Hatta, Tanabe, and Nishimoto
homing cyclic peptide CNGRC that should target to a specific General Methods
Proton nuclear magnetic resonance spectra (1H NMR)
and carbon nuclear magnetic resonance spectra (13C NMR)
were recorded on a JEOL JNM-EX-400 (400 MHz) spec-
trometer (JEOL, Tokyo, Japan) or a JNM-AL300 (300 MHz)
spectrometer. Fast atom bombardment mass spectrometry
(FAB-MS) was performed on a JEOL JMS-SX102A mass
spectrometer, using glycerol or 3-nitrobenzyl alcohol (NBA)
matrix. High-performance liquid chromatography was per-
formed on a Hitachi D-7000 HPLC system (Hitachi, Tokyo,
Japan), using aqueous CH3CN solution containing 0.05%
TFA as a mobile phase. An Interstil ODS-3 column (4.6␾ ×
250 mm, GL Science Inc., Tokyo, Japan) was used for ana-
lytical HPLC. Sample solution was eluted with a linear gra-
dient of aqueous CH3CN at a flow rate of 0.6 ml/min and
detected by UV absorbance at wavelength 210 nm (for com-
pounds CN1, CN2, and CN3) or 260 nm (for compounds
5-FdUrd, CNF1, and CNF2). A preparative Interstil ODS-3
column (10␾ × 250 mm, GL Science Inc.) was used for pre-
parative HPLC. Sample solution was eluted with a linear gra-
dient of aqueous CH3CN (10–26%, 20 min) at a flow rate of
3.0 ml/min and detected by UV absorbance at 210 nm.
APN/CD13 isoform expressing within tumor tissues but to
none of the other isoforms in normal epothelia and myeloid
cells (26–27).
Previous studies demonstrated that ester derivatives of
5-FdUrd undergo hydrolysis to effectively release 5-FdUrd
under physiologic conditions and thereby produce a consid-
erable antitumor effect. It was noted that 3Ј-O-esters of
5-FdUrd showed higher antitumor effect than 3Ј-O-ester ana-
logs in vivo, due in part to improved pharmacokinetic char-
acteristics with sufficiently slow hydrolysis rate for arriving at
tumor tissues (4–6). However, a quantitative structure-
activity relationship of the prodrugs releasing 5-FdUrd by
hydrolysis, especially the influence of linker structures on the
hydrolytic activation rate, has not yet been clarified and is still
a subject of further investigations. We have therefore de-
signed and synthesized two prototypes of tumor targeting
5-FdUrd prodrugs, in which 3Ј-oxygen of 5-FdUrd was con-
jugated with a common cyclic peptide CNGRC by known
linkers based on succinate (CNF1) and glutarate (CNF2) es-
ters, respectively (28,29). As the control compounds without
5-FdUrd moiety, amide-linked derivatives of cyclic peptide
CNGRC with succinic acid (CN1) and glutaric acid (CN2)
were also synthesized. These peptide conjugates were inves-
tigated to elucidate the structural effect of ester linkers on the
hydrolytic 5-FdUrd release from 5-FdUrd–CNGRC conju-
gates, the hydrolysis reaction pathway, and the cytotoxicity
toward both APN/CD13 positive and negative cell lines by
reference to 5-FdUrd.
Synthesis of 2؅-Deoxy-5؅-O-[bis(4-methoxyphenyl)
phenylmethyl]-5-fluoro-3؅-O-(3-carboxypropanoyl)uridine
(Compound 2)
Intermediate compound 2 leading to the prodrug CNF1
was synthesized as outlined in Fig. 1a. The 5Ј-OH of 5-FdUrd
was selectively protected using 4,4Ј-dimethoxytriphenyl-
methyl chloride to give 2Ј-deoxy-5Ј-O-[bis(4-methoxy-
phenyl)phenylmethyl]-5-fluorouridine (1) (12). Succinic
anhydride was allowed to react with 1 (0.3 mmol) in dichlo-
romethane (DCM; 16 ml) in the presence of 4-dimethylami-
nopyridine (0.45 mmol). The reaction mixture was stirred for
16 h at room temperature, poured into ice-cooled aqueous
solution of 5% NaHCO3 (30 ml), acidified with 1 M HCl to
pH 4, and then extracted with DCM (50 ml). The extract was
washed by water and dried with Na2SO4. Evaporation of the
solvent and purification by column chromatography (CHCl3/
MeOH, 12:1) gave 2 (169 mg, 87%): Rf ס 0.25 (CHCl3/
MATERIALS AND METHODS
Materials
5-Fluoro-2Ј-deoxyuridine (5-FdUrd) and 4,4Ј-
dimethoxytriphenylmethyl chloride were obtained from To-
kyo Kasei (Tokyo, Japan). Fmoc-Cys(Acm)-PEG-PS-resin
and Fmoc-amino acids such as Arg(Pbf), Asn(Trt),
Cys(Acm), and Gly were purchased from Applied Biosystems
(Warrington, UK). N,N-Diisopropylethylamine (DIPEA)
and O-(7-azabenzotriazol-1-yl)-N,N,NЈ,NЈ-tetramethyl-
uronium hexafluorophosphate (HATU) were received from
Applied Biosystems. Succinic anhydride, glutaric anhydride,
trifluoroacetic acid (TFA), trimethylsilyl chloride (TMSCl),
2-propanol, culture media PRMI-1640, and Eagle’s minimal
essential medium (MEM) were purchased from Nacalai
Tesque (Kyoto, Japan). Fetal bovine serum (FBS) was ob-
tained from Thermo Trace (Melbourne, Australia). 3-(4,5-
Dimethylthiazole-2-yl)-2,5-diphenyltetrazolium bromide
(MTT) was obtained from Sigma (Steinheim, Germany).
Acetonitrile, dimethylsulfoxide (DMSO), and other chemi-
cals were purchased from Wako (Osaka, Japan). RIPA lysis
buffer was received from Upstate Biotechnology (Lake
Placid, NY, USA). All the chemicals were of either analytic
or high-performance liquid chromatography (HPLC) grade.
Cell line of HT-1080 human fibrosarcoma was supplied by the
Cell Resource Center for Biomedical Research, Tohoku Uni-
versity (Sendai, Japan). Cell lines of MDA-MB-231 human
adenocarcinoma and HT-29 human colorectal adenocarci-
noma were obtained from the American Type Culture Col-
lection (www.attc.org).
1
MeOH, 12:1). H NMR (300 MHz, [D6] acetone): ␦ ס 2.50
(m, 2 H, 2Ј-CH2), 2.63 (t, J ס 2.7 Hz, 4 H, Succ-CH2), 3.36
(dd, J ס 10.4, 3.2 Hz, 1 H, 5Ј-Ha), 3.53 (dd, J ס 10.5, 4.3 Hz,
1 H, 5Ј-Hb), 3.78 (s, 6 H, OCH3), 4.20 (dt, J ס 3.8, 3.1 Hz, 1
H, 3Ј-H), 5.47 (dt, J ס 5.9, 2.7 Hz, 1 H, 4Ј-H), 6.27 (td, J ס
7.0, 1.7 Hz, 1 H, 1Ј-H), 6.89, 7.23–7.50 (2 m, 13 H, arom H),
7.87 (d, J ס 6.7 Hz, 1H, 6-H). 13C NMR (100 MHz, [D6]
acetone): ␦ ס 29.4, 30.2 (Succ-CH2), 38.2(C-2Ј), 55.4 (2 sig-
nals, OCH3), 64.5 (C-5Ј), 75.4 (C-3Ј), 84.6(C-4Ј), 85.7 (C-1Ј),
87.6 (C-Ph3), 113.9 (2 signals, m-C in MeOPh), 124.6 (d, J ס
34.0 Hz, C-6), 127.6 (p-C in Ph), 128.6 (o-C in Ph), 128.7 (m-C
in Ph), 130.8 (o-C in MeOPh), 136.1, 136.3 (i-C in MeOPh),
141.3 (d, J ס 233.1 Hz, C-5), 145.6 (i-C in Ph), 150.3 (C-2), 157.3
(d, J ס 27.3 Hz, C-4), 159.5 (C-OMe in MeOPh), 172.4, 173.4
(Succ-CסO). FAB-MS (positive mode, NBA): m/z ס 648 [M]+,
HRMS calcd. for C34H33FN2O10 [M]+ 648.2119, found 648.2142.
Synthesis of 2؅-Deoxy-5؅-O-[bis(4-methoxyphenyl)
phenylmethyl]-5-fluoro-3؅-O-(4-carboxybutanoyl)uridine
(Compound 3)
Intermediate compound 3 leading to the prodrug CNF2
was synthesized in the same way as intermediate 2 (Fig.

New Class of 5-Fluoro-2؅-deoxyuridine Prodrugs
383
CNGRC-resin was prepared on a Pioneer peptide synthesis
system (PerSeptive Biosystems Inc., Framingham, MA,
USA), using Fmoc-Cys(Acm)-PEG-PS-resin (0.18 meq/g, 0.2
mmol scale) and Fmoc-amino acids Arg(Pbf), Asn(Trt),
Cys(Acm), and Gly, along with HATU and DIPEA as acti-
vators. Intermediate compound 2 (330 ␮mol) was manually
condensed with the protected peptide CNGRC-resin [equal
to 66 ␮mol Fmoc-Cys(Acm)-PEG-PS-resin], activated by
HATU (330 ␮mol) and DIPEA (660 ␮mol) in dimethylfor-
mamide (DMF; 3 ml) for 0.5 h at room temperature. After
filtration and washing with DMF (5 × 5 ml) and DCM (4 × 4
ml), the resulting protected 5-FdUrd-3Ј-O-Succ-CNGRC–
resin was treated with a mixture of TFA (86 ml)-TMSCl (0.72
ml)-anisole (0.99 ml) at room temperature (30). After 1 h,
DMSO (12.5 ml) was added to the reaction mixture and kept
for another 1 h at 4°C with stirring. After removal of the resin
by filtration, ice-cooled dry diethyl ether (600 ml) was added
to the filtrate. The precipitate was collected by centrifugation
and then washed with ice-cooled dry diethyl ether (4 × 20 ml).
The crude product was purified by preparative HPLC and
freeze-dried to give fluffy white powder of prodrug CNF1
[11.6 mg, 20% calculated from the Fmoc-Cys(Acm)-PEG-PS-
resin]: Analytical HPLC: tr ס 30.1 min (1–30% CH3CN, 40
min). 1H NMR (400 MHz, [D6] DMSO): ␦ ס 1.42–1.54 (m, 2
H, Arg-␥-CH2), 1.85 (m, 2 H, Arg-␤-CH2), 2.26, 2.31 (2 m, 2
H, 5-FdUrd-2Ј-CH2), 2.48–2.53 (2 m, 4 H, Succ-CH2), 2.72 (m,
2 H, Arg-␦-CH2), 2.89 (m, 2 H, Asn-␤-CH2), 3.05–3.25 (2 m,
4 H, Cys1-␤-CH2, Cys2-␤-CH2), 3.63 (br. s, 1 H, 5-FdUrd-5Ј-
CH2), 4.00 (br. s, 1 H, 5-FdUrd-3Ј-CH), 4.12 (d, J ס 7.6 Hz,
1 H, Gly-␣-CHa), 4.17 (d, J ס 7.6 Hz, 1 H, Gly-␣-CHb),
4.27–4.35 (2 m, 2 H, Arg-␣-CH, Asn-␣-CH), 4.44–4.49 (m, 2
H, Cys1-␣-CH, Cys2-␣-CH), 5.20 (br. s, 1 H, 5-FdUrd-4Ј-CH),
6.15 (t, J ס 6.8 Hz, 1 H, 5-FdUrd-1Ј-CH), 6.80 (br. s, 3 H,
Arg-␦-NH, NסCNH2), 7.29 (s, 2 H, Asn-CONH2), 7.46 (t, J
ס 4.0 Hz, 1 H, Gly-␣-NH), 7.72–7.74 (2 m, 2 H, CONH,
CסNH), 8.17–8.21 (m, 2 H, CONH, 5-FdUrd-6-H), 8.38 (d, J
ס 7.6 Hz, 1 H, CONH), 8.62 (d, J ס 7.2 Hz, 1 H, CONH),
11.85 (d, J ס 4.8 Hz, 1 H, COOH). 13C NMR(75 MHz, [D6]
DMSO): ␦ ס 24.9 (Arg-␥-C), 28.4 (Arg-␤-C), 29.0 (Succ-
CסO), 29.7 (Succ-CסO), 35.8 (Asn-␤-C), 36.9 (5-FdUrd-C-
2Ј), 40.1 (Cys-␤-C), 40.3 (Cys-␤-C), 41.8 (Arg-␦-C), 42.7 (Gly-
␣-C), 49.7 (Asn-␣-C), 51.2 (Arg-␣-C), 52.0 (Cys-␣-C), 52.5
(Cys-␣-C), 61.2 (5-FdUrd-C-5Ј), 74.8 (5-FdUrd-C-3Ј), 84.5,
84.9 (5-FdUrd-C-4Ј, 5-FdUrd-C-1Ј), 124.2/124.7 (5-FdUrd-C-
6), 138.6/141.6 (5-FdUrd-C-5), 149.0 (5-FdUrd-C-2), 156.7
(CסNH), 156.8/157.2 (5-FdUrd-C-4), 169.4, 170.5, 170.6,
171.5, 171.6, 172.0 (8C, CסO). FAB-MS (positive mode, glyc-
Fig. 1. (a) Syntheses of intermediate compounds 2 and 3. (b) Syn-
theses of 5-FdUrd–CNGRC conjugates CNF1 and CNF2 with succi-
nate and glutarate linkers, respectively.
1a). Reaction of glutaric anhydride (1.8 mmol) with com-
pound 1 (0.3 mmol) and purification by column chromatog-
raphy (CHCl3/MeOH, 12:1) gave 3 (170 mg, 85%): Rf ס 0.27
(CHCl3/MeOH, 12:1). 1H NMR (300 MHz, [D6] acetone):
␦ ס 1.88, (m, 2 H, Glut-CH2), 2.35–2.52 (m, 6 H, Glut-
COCH2, 2Ј-H), 3.36 (dd, J ס 10.4, 3.2 Hz, 1 H, 5Ј-Ha), 3.54
(dd, J ס 10.4, 4.2 Hz, 1 H, 5Ј-Hb), 3.78 (s, 6 H, OCH3), 4.12,
4.18 (dd*2, J ס 3.3, 3.2 Hz, total 1 H, 3Ј-H), 5.38, 5.48 (dt*2,
J ס 5.9, 2.7 Hz, total 1 H, 4Ј-H), 6.27 (td, J ס 7.9, 1.7 Hz, 1
H, 1Ј-H), 6.87, 7.20–7.50 (2 m, 13 H, Ph), 7.87 (d, J ס 6.7 Hz,
1H, 6-H). 13C NMR (75 MHz, [D6] acetone): ␦ ס 20.8 (Glut-
CH2CH2), 33.1, 33.6 (Glut-CH2CסO), 38.2(C-2Ј), 55.4 (2 sig-
nals, OCH3), 64.5 (C-5Ј), 75.3 (C-3Ј), 84.8 (C-4Ј), 85.82 (C-1Ј),
87.7 (C-Ph3), 114.0 (2 signals, m-C in MeOPh), 124.7 (d, J ס
34.2 Hz, C-6), 127.7 (p-C in Ph), 128.7 (o-C in Ph), 128.9 (m-C
in Ph), 130.9 (o-C in MeOPh), 136.3, 136.5 (i-C in MeOPh),
141.5 (d, J ס 233.2 Hz, C-5), 145.8 (i-C in Ph), 149.7 (C-2),
157.4 (d, J ס 26.8 Hz, C-4), 159.7 (C-OMe in MeOPh), 172.9,
174.1 (Glut-CסO). FAB-MS (positive mode, NBA): m/z ס
662 [M]+, HRMS calcd. for C35H35FN2O10 [M]+ 662.2276,
found 662.2255.
e r o l ) : m / z
ס
8 7 8 [ M + H ] + , H R M S c a l c d . f o r
C31H45FN11O14S2 [M+H]+ 878.2573, found 878.2573.
Synthesis of 5-FdUrd-3؅-O-Glut-CNGRC
Conjugate (CNF2)
Prodrug CNF2, 5-FdUrd-3Ј-O-Glut-CNGRC conjugate,
was synthesized in the same way as prodrug CNF1 (Fig. 1b).
Condensation of protected CNGRC-resin [equal to 66 ␮mol
Fmoc-Cys(Acm)-PEG-PS-resin] with intermediate com-
pound 3 and purification by preparative HPLC gave fluffy
white powder of prodrug CNF2 [10.6 mg, 18% calculated
from the Fmoc-Cys(Acm)-PEG-PS-resin]: Analytical HPLC:
tr ס 30.8 min (1–30% CH3CN, 40 min). 1H NMR (400 MHz,
[D6] DMSO): ␦ ס 1.42–1.56 (m, 2 H, Arg-␥-CH2), 1.75 (m, 2
H, Glut-␤-CH2), 1.83 (m, 2 H, Arg-␤-CH2), 2.12–2.37 (3
Synthesis of 5-FdUrd-3؅-O-Succ-CNGRC
Conjugate (CNF1)
Prodrug CNF1, 5-FdUrd-3Ј-O-Succ-CNGRC conjugate,
was synthesized as shown in Fig. 1b. Protected peptide

384
Zhang, Hatta, Tanabe, and Nishimoto
m, 6 H, Glut-␣-CH2, Glut-␥-CH2, 5-FdUrd-2Ј-CH2), 2.70 (m, Synthesis of N␣-Hemiglutarimide-CNGRC (CN2)
2 H, Arg-␦-CH2), 2.86 (m, 2 H, Asn-␤-CH2), 3.00–3.2 (2 m, 4
CN2 was synthesized in the same way as CN1. Conden-
H, Cys1-␤-CH2, Cys2-␤-CH2), 3.62 (br. s, 2 H, 5-FdUrd-5Ј-
sation of protected CNGRC-resin [equal to 20 ␮mol Fmoc-
CH2), 3.99 (m, 1 H, 5-FdUrd-3Ј-CH), 4.12 (d, J ס 7.6 Hz, 1 H,
Cys(Acm)-PEG-PS-resin] with glutaric anhydride (80 ␮mol)
Gly-␣-CHa), 4.17 (d, J ס 7.6 Hz, 1 H, Gly-␣-CHb), 4.27 (m, 1
and purification by preparative HPLC gave fluffy white pow-
H, Arg-␣-CH), 4.36 (m, 1 H, Asn-␣-CH), 4.42–4.49 (m, 2 H,
der of CN2 [6.0 mg, 45% calculated from the Fmoc-
Cys1-␣-CH, Cys2-␣-CH), 5.21 (m, 1 H, 5-FdUrd-4Ј-CH), 6.14
Cys(Acm)-PEG-PS-resin]: Analytical HPLC: tr ס 24.5 min
(t, J ס 7.6 Hz, 1 H, 5-FdUrd-1Ј-CH), 6.79 (br. s, 3 H, Arg-␦-
(1–30% CH3CN, 40 min). 1H NMR (400 MHz, [D6] DMSO):
NH, NסCNH2), 7.32 (s, 2 H, Asn-CONH2), 7.55 (dd, J ס 7.6,
␦ ס 1.43–1.56 (m, 2 H, Arg-␥-CH2), 1.70 (m, 2 H, Glut-␤-
7.6 Hz, 1 H, Gly-␣-NH), 7.74 (d, J ס 8.4 Hz, 1 H, CONH),
CH2), 1.85 (m, 2 H, Arg-␤-CH2), 2.12–2.33 (2 m, 4 H, Glut-
7.80 (m, 1 H, CסNH), 8.18–8.22 (m, 2 H, CONH, 5-FdUrd-
␣-CH2, Glut-␥-CH2), 2.69 (m, 2 H, Arg-␦-CH2), 2.87 (m, 2 H,
6-CH), 8.27 (d, J ס 8.0 Hz, 1 H, CONH), 8.64 (d, J ס 8.0 Hz,
Asn-␤-CH2), 3.00–3.23 (2 m, 4 H, Cys1-␤-CH2, Cys2-␤-CH2),
1 H, CONH), 11.86 (d, J ס 4.8 Hz, 1 H, COOH). 13C
4.14 (d, J ס 8.0 Hz, 1 H, Gly-␣-CHa), 4.18 (d, J ס 8.0 Hz, 1
NMR(100 MHz, [D6] DMSO): ␦ ס 20.4 (Glut-␤-C), 25.0
(Arg-␥-C), 28.5 (Arg-␤-C), 32.8, 34.1 (Glut-␣-CסO, Glut-␥-
CסO), 35.8 (Asn-␤-C), 36.9 (5-FdUrd-C-2Ј), 40.1 (Cys-␤-C),
40.4 (Cys-␤-C), 42.0 (Arg-␦-C),42.7 (Gly-␣-C), 49.6 (Asn-␣-
C), 51.1 (Arg-␣-C), 52.0 (Cys-␣-C), 52.3 (Cys-␣-C), 61.2 (5-
FdUrd-C-5Ј), 74.6 (5-FdUrd-C-3Ј), 84.4, 84.9 (5-FdUrd-C-4Ј,
5-FdUrd-C-1Ј), 124.2/124.6 (5-FdUrd-C-6), 138.8/141.6 (5-
FdUrd-C-5), 148.9 (5-FdUrd-C-2), 156.6 (CסNH), 156.7/
156.9 (5-FdUrd-C-4), 169.2 (CסO), 170.4 (CסO), 170.5
(CסO), 171.3 (CסO), 171.4 (CסO), 171.5 (CסO), 172.1 (2
H, Gly-␣-CHb), 4.29 (td, J ס 8.4, 4.8 Hz, 1 H, Arg-␣-CH),
4.36 (m, 1 H, Asn-␣-CH), 4.41–4.49 (m, 2 H, Cys1-␣-CH,
Cys2-␣-CH), 6.80 (br. s, 3 H, Arg-␦-NH, NסCNH2), 7.31 (s,
2 H, Asn-CONH2), 7.49 (t, J ס 5.6 Hz, 1 H, Gly-␣-NH), 7.72
(d, J ס 8.4 Hz, 1 H, CONH), 7.80 (m, 1 H, CסNH), 8.18 (d,
J ס 8.4, 1 H, CONH), 8.26 (d, J ס 8.0 Hz, 1 H, CONH), 8.64
(d, J ס 8.0 Hz, 1 H, CONH). 13C NMR (100 MHz, D2O):
␦ ס 21.0 (Glut-␤-C), 25.0 (Arg-␥-C), 28.4 (Arg-␤-C), 33.6,
35.1 (Glut-␣-CסO, Glut-␥-CסO), 36.3 (Asn-␤-C), 41.1 (Cys-
␤-C), 41.2 (Cys-␤-C), 42.0 (Arg-␦-C), 43.4 (Gly-␣-C), 51.3
signal, 2 C, CסO). FAB-MS (positive mode, NBA): m/z ס (Asn-␣-C), 53.5 (Arg-␣-C), 54.1 (Cys-␣-C), 54.3 (Cys-␣-C),
892 [M+H]+, HRMS calcd. for C32H47FN11O14S2 [M+H]+
157.3 (CסNH), 171.7 (CסO), 172.6 (CסO), 173.9 (CסO),
174.5 (CסO), 175.3 (CסO), 176.6 (CסO), 178.4 (CסO).
FAB-MS (positive mode, glycerol): m/z ס 664 [M+H]+,
HRMS calcd. for C23H38N9O10S2 [M+H]+ 664.2183, found
664.2186.
892.2729, found 892.2744.
Synthesis of N␣-Hemisuccinimide-CNGRC (CN1)
Hydrolysis of 5-FdUrd–CNGRC Conjugates CNF1
and CNF2
Protected CNGRC-resin [equal to 20 ␮mol Fmoc-
Cys(Acm)-PEG-PS-resin] was condensed with succinic anhy-
dride (80 ␮mol) in pyridine (3 ml) for 1 h at room tempera-
ture. After filtration and washing with DMF (5 × 5 ml) and
DCM (4 × 5 ml), the resulting protected N␣-hemisuccinimide-
CNGRC-resin was treated with a mixture of TFA (26 ml)-
TMSCl (0.22 ml)-anisole (0.3 ml)-DMSO (3.9 ml) in the same
manner as in synthesis of CNF1. Precipitation and prepara-
tive HPLC gave fluffy white powder of CN1 [5.6 mg, 43%
CNF1 or CNF2 (0.05 ␮mol) was dissolved in 10 mmol/L
phosphate buffer (PB, 0.5 ml, pH 7.4), fetal bovine serum
(FBS) / H2O (v/v ס 1:1, 0.5 ml), MEM containing 10% FBS
(0.5 ml), HT-1080 cell lysate (0.5 ml) or MDA-MB-231 cell
lysate (0.5 ml) and then incubated at 37°C. At various time
intervals, an aliquot (10 ␮l) was sampled for analytical HPLC.
The samples in PB and FBS were eluted with a linear gradient
calculated from the Fmoc-Cys(Acm)-PEG-PS-resin]: Ana- of aqueous CH3CN (1–30%, 40 min), while those in MEM-
lytical HPLC: tr ס 25.0 min (1–30% CH3CN, 40 min). 1H 10%FBS and HT-1080 cell lysate were eluted with another
NMR (400 MHz, [D6] DMSO): ␦ ס 1.46–1.52 (m, 2 H, Arg- linear gradient of aqueous CH3CN (5.9–30.4%, 30 min). The
␥-CH2), 1.86 (m, 2 H, Arg-␤-CH2), 2.28–2.30 (m, 4 H, Succ- HPLC elution peaks of starting prodrug and hydrolysates
␣-CH2, Succ-␤-CH2), 2.72 (m, 2 H, Arg-␦-CH2), 2.90 (m, 2 H, were fractionated and identified by FAB-MS.
Asn-␤-CH2), 3.06–3.20 (2 m, 4 H, Cys1-␤-CH2, Cys2-␤-CH2),
The concentrations of prodrugs and hydrolysates during
4.13 (d, J ס 8.0 Hz, 1 H, Gly-␣-CHa), 4.16 (d, J ס 8.0 Hz, 1 hydrolysis were quantified by analytical HPLC, using the re-
H, Gly-␣-CHb), 4.26–4.34 (m, 2 H, Arg-␣-CH, Asn-␣-CH), spective calibration curves prepared by reference to the au-
4.42 (dt, J ס 6.8, 6.8 Hz, 1 H, Cys1-␣-CH), 4.50 (dt, J ס 9.2, thentic samples except for less stable intermediate CN3 pro-
3.2 Hz, 1 H, Cys2-␣-CH), 6.80 (br. s, 3 H, Arg-␦-NH, duced in the hydrolysis of CNF1 (Fig. 3). For convenience, the
NסCNH2), 7.29 (s, 2 H, Asn-CONH2), 7.48 (t, J ס 5.4 Hz, 1 concentration of CN3 was estimated by HPLC analysis 1 h
H, Gly-␣-NH), 7.70 (d, J ס 8.4 Hz, 1 H, CONH), 7.75 (m, 1 after the hydrolysis of CNF1, assuming that the theoretical
H, CסNH), 8.17 (d, J ס 8.0, 1 H, CONH), 8.38 (d, J ס 8.0 material balance of Ct(CN3) ס Ctס0(CNF1) – Ct(CNF1) –
Hz, 1 H, CONH), 8.60 (d, J ס 7.2 Hz, 1 H, CONH). 13C Ct(CN1) is satisfied in the initial stage of hydrolysis, where
NMR(100 MHz, [D6] DMSO): ␦ ס 25.0 (Arg-␥-C), 28.4 Ctס0 and Ct are the concentrations of each product at the
(Arg-␤-C), 29.0, 30.0 (Succ-␣-CסO, Succ-␤-CסO), 35.7 initial time t ס 0 and a given time t, respectively. Evidently,
(Asn-␤-C), 40.1 (Cys-␤-C), 40.4 (Cys-␤-C), 42.7 (Arg-␦-C), this assumption became invalid as the hydrolysis proceeded,
45.7 (Gly-␣-C), 49.8 (Asn-␣-C), 51.2 (Arg-␣-C), 52.0 (Cys-␣- probably due to the side reactions of CNF1 that occurred
C), 52.6 (Cys-␣-C), 156.5 (CסNH), 169.2, 170.4, 170.5, 171.2, without 5-FdUrd release.
171.4, 171.5, 171.8, 173.7 (8 C, CסO). FAB-MS (positive
mode, glycerol): m/z ס 650 [M+H]+, HRMS calcd. for CNF2 in all the media (PB, FBS, MEM-10% FBS, and HT-
C22H36N9O10S2 [M+H]+ 650.2027, found 650.2022.
1080 and MDA-MB-231 cell lysates) were well represented in
Because the hydrolytic decompositions of CNF1 and

New Class of 5-Fluoro-2؅-deoxyuridine Prodrugs
385
terms of the first-order kinetics (for CNF1 shown in Fig. 2d), plates were incubated for 72 h at 37°C under an atmosphere
the hydrolysis rate constants (k1) were evaluated from Eq. (1):
Ct = Ct=0 и exp −k t͒ (1)
of humidified 5% CO2, and then 10 ␮l of 5 mg/ml MTT in
PBS was added to each well. After MTT cleavage for 4 h at
37°C, 100 ␮l of 0.04 N HCl in 2-propanol was added to each
well to dissolve the dark blue crystals. The absorbance at a
test wavelength of 570 nm was measured on a Model-550
Microplate Reader (Bio-Rad Laboratories, Hercules, CA,
USA). The concentration of drug to reduce the cell survival
to half of the control value (IC50) was calculated from a re-
gression analysis of cell survival vs drug concentration.
͑
1
Accordingly, the half-life periods (t(1/2)) of CNF1 and
CNF2 were also evaluated as the measures of their stability
from Eq. (2):
t 1 = ln2
ր
k1
(2)
ր
2
͒
͑
In Vitro Cytotoxicity Assays
Preparation of Cell Lysates
HT-1080 and MDA-MB-231 cells were cultured in MEM
containing 1% nonessential amino acids and 10% FBS, while
HT-29 cells were in RPMI-1640 medium containing 10%
FBS. All the cells were grown in an atmosphere of 5% CO2
and 90% relative humidity at 37°C.
Inhibition of the cellular growth was evaluated using a
method of MTT assay described by Mosmann (31). Suspen-
sions of 5000 cells (50 ␮l) were seeded in each well of a
96-well cell culture microplate and treated with either 50 ␮l
fresh drug–medium solution at various concentrations from
10−10 to 10−3 mol/L or 50 ␮l control medium solution. These
HT-1080 or MDA-MB-231 cells (about 4 × 107 cells)
were collected from 8 dishes (␾100 mm dish/75 cm2), washed
twice with ice-cold PBS, followed by centrifugation at 1200
rpm in a table-top centrifuge for 5 min to pelletize the cells.
The resulting cell pellet was added to ice-cold RIPA lysis
buffer (4 ml, pH 7.4) and shaken for 15 min to lyse cells. The
lysate was centrifuged at 14,000 rpm for 15 min at 4°C, and
then the supernatant was transferred to seven fresh tubes for
hydrolysis studies of peptide conjugates CNF1 and CNF2.
Fig. 2. (a) Concentration vs. time plots for (छ) CNF1, (ᮀ) 5-FdUrd, (᭿) CN3, and (᭡) CN1 in the hydrolysis of 0.1 mM CNF1 in 10 mM
phosphate buffer (PB, pH 7.4) at 37°C. (b) Concentration vs. time plots for (छ) CNF1, (ᮀ) 5-FdUrd, (᭿) CN3, and (᭡) CN1 in the hydrolysis
of 0.1 mM CNF1 in fetal bovine serum (FBS) at 37°C. (c) Concentration vs. time plots for (᭺) CNF2, (ᮀ) 5-FdUrd, and (᭿) CN2 in the
hydrolysis of 0.1 mM CNF2 in 10 mM PB (pH 7.4) at 37°C. (d) First-order kinetic plots for the hydrolysis of CNF1 in (᭹) PB (pH 7.4), (᭺)
HT-1080 cell lysate, (ᮀ) MDA-MB-231 cell lysate, (⌬)FBS, and (᭡) MEM + 10% FBS.

386
Zhang, Hatta, Tanabe, and Nishimoto
Hydrolysis of 5-FdUrd–CNGRC Conjugates
Extracellular and Intracellular Concentrations of 5-FdUrd
and Its Prodrugs CNF1 and CNF2
Releasing 5-FdUrd
Ester compounds have been well-known to undergo both
nonenzymatic and enzymatic hydrolysis in physiologic envi-
ronment (28,29). We therefore studied the hydrolysis of
5-FdUrd–peptide conjugates, CNF1 with succinate linker, and
CNF2 with glutarate linker in phosphate buffer (PB) at pH
7.4, fetal bovine serum (FBS), MEM containing 10% FBS,
and HT-1080 and MDA-MB-231 cell lysates.
For analyzing the cellular concentrations of 5-FdUrd and
its prodrugs (CNF1 and CNF2), we used a similar method as
previously reported in the analysis of 5-FdUrd within EMT6
cells (32). 5-FdUrd, CNF1 or CNF2 (14 ␮mol) was added to
70 ml of HT-1080 (or MDA-MB-231) cell suspension (107
cells/ml in MEM containing 10% FBS medium) in a spinner
flask rotated at 100 rpm at room temperature. At appropri-
ated time intervals, aliquots of the cell suspensions (6 ml)
were centrifuged at 1200 rpm for 5 min. The supernatant (0.1
ml) was transferred to a tube and 0.1 ml of 1% aqueous
CH3CN containing HCl (pH 0.5) was added to stop the hy-
drolysis. The resulting solution was sampled for analyzing
extracellular concentration of 5-FdUrd or prodrug by analyti-
cal HPLC with a linear gradient of aqueous CH3CN (5.9–
30.4%, 30 min). After washing by PBS, the cell pellet was
treated by 0.1 ml of 1% aqueous CH3CN containing HCl (pH
0.5) and sonicated for 5 min to dissolve 5-FdUrd or prodrug.
The suspension was transferred to a centrifugal filter Micro-
con YM-3 (Millipore Co., Bedford, MA, USA) and centrifu-
gated at 14,000 rpm for 20 min at 4°C. The filtrate thus ob-
tained was sampled for analyzing intracellular concentration
of 5-FdUrd or prodrug by analytical HPLC with a linear gra-
dient of aqueous CH3CN (5.9–30.4%, 30 min).
Figure 2a illustrates a time-course profile observed for
the hydrolysis of CNF1 to release 5-FdUrd in phosphate
buffer at pH 7.4. The 5-FdUrd release accompanied the for-
mation of N␣-succinimide-CNGRC (CN3: positive FAB-MS
(glycerol) m/z ס 632 [M+H]+; HRMS found 632.1872, calcd.
for C22H34N9O9S2 [M+H]+ 632.1921; analytical HPLC reten-
tion time tr ס 29.5 min) as an intermediate. The formations of
similar intermediates were previously reported in the hydro-
lysis of 3Ј-azido-3Ј-deoxythymidine-5Ј-O-Succ-peptide conju-
gates (28). The intermediate peptide CN3 was subsequently
hydrolyzed and converted to the final product CN1. The main
hydrolysis mechanism of CNF1 is outlined in Fig. 3, while the
side reactions leading to unknown by-products occurred si-
multaneously with the main hydrolysis, as confirmed by the
analytical HPLC. Similar characteristics involving the inter-
mediate CN3 were also observed in the hydrolysis of CNF1 in
FBS (Fig. 2b), MEM-10% FBS and cell lysates (data not
shown). As plotted in Fig. 2d, the hydrolysis of CNF1 in each
medium was of a first-order reaction kinetics with respective
rate constant (Table I). Obviously, the hydrolysis rates of
CNF1 in FBS-containing media and cell lysates are greater
than that in PB. This acceleration effect suggests that some
RESULTS AND DISCUSSION
Synthesis of 5-FdUrd–CNGRC Conjugates with
Ester Linkers
The 5Ј-OH–protected 5-FdUrd derivative 1, which was
synthesized according to the procedure of Goerlach (12), was
condensed with succinic anhydride or glutaric anhydride to
obtain two types of intermediate compounds 2 and 3 leading
to the corresponding 5-FdUrd prodrugs CNF1 and CNF2,
respectively (Fig. 1a). Protected peptide CNGRC-resin pre-
pared by the Fmoc solid phase synthesis method was con-
densed with the intermediate compounds 2 and 3, followed by
elimination of protecting groups and disulfide cyclization with
the TFA-TMSCl-anisole-DMSO system in one-pot procedure
to produce CNF1 and CNF2, respectively, as shown in Fig. 1b
(30). The overall yields of prodrugs CNF1 and CNF2 were
20% and 18%, respectively. The lower yields than those of
CN1 and CN2 (see below), are attributable to partial decom-
positions of 5-FdUrd-3Ј-O-Succ-CNGRC-resin and 5-FdUrd-
3Ј-O-Glut-CNGRC-resin at their 3Ј-O-ester linkages during
the treatment with TFA-TMSCl-anisole-DMSO.
In order to identify the molecular mechanism of CNF1
and CNF2 as the potential antitumor prodrugs, the corre-
sponding control compounds N␣-hemisuccinimide-CNGRC
(CN1) and N␣-hemiglutarimide-CNGRC (CN2) were also
synthesized. The protected peptide CNGRC-resin was simi-
larly condensed with succinic anhydride or glutaric anhydride,
followed by elimination of protecting groups and disulfide
cyclization with the TFA-TMSCl-anisole-DMSO system to
yield CN1 (43%) or CN2 (45%). The purity of each peptide
conjugate was more than 97%, as confirmed by analytical
HPLC.
Fig. 3. Reaction pathway for the hydrolysis to release 5-FdUrd of
5-FdUrd–CNGRC conjugate CNF1 with succinate linker.

New Class of 5-Fluoro-2؅-deoxyuridine Prodrugs
Table I. Hydrolyses of 5-FdUrd-CNGRC Conjugates in Various Media at 37°C
387
Yield of 5-FdUrd (%)
after hydrolysis for a
given time (h) shown
in the parentheses
k1 (h−1
(mean SD,
)
t1/2 (h)
(mean SD,
n ס 3)
Selectivity of 5-FdUrd
release (%)
Prodrug
Medium
PB (pH 7.4)
n ס 3)
CNF1
CNF1
CNF1
CNF1
CNF1
CNF2
CNF2
CNF2
CNF2
CNF2
0.185 0.004
0.410 0.003
0.580 0.010
0.255 0.011
0.263 0.01
0.0066 0.0001
0.0098 0.0007
0.0414 0.0019
0.043 0.004
0.041 0.001
3.74 0.08
1.69 0.01
1.19 0.04
2.72 0.11
2.63 0.08
105.0 1.6
70.9 5.3
16.7 0.8
16.2 1.4
16.9 0.4
80 (24)
90 (6)
83 (6)
70 (6)
71 (6)
82
93
85
92
87
63
66
66
48
49
FBS
MEM + 10% FBS
HT-1080 cell lysate
MDA-MB-231 cell lysate
PB (pH 7.4)
13 (24), 35 (120)
13 (24), 41 (120)
40 (24), 57 (48)
32 (24)
FBS
MEM + 10% FBS
HT-1080 cell lysate
MDA-MB-231 cell lysate
31 (24)
enzymatic catalysis may be involved in the hydrolysis of diate hydrolysate CN3 bearing succinimide moiety (see Figs.
CNF1. 2 and 3).
In contrast to the behavior of CNF1, hydrolysis of CNF2 In another aspect, by reference to similar conjugates con-
in all media (PB, FBS, cell culture medium, or cell lysates) led sisting of ester linkers derived from primary alcohols such as
directly to final products 5-FdUrd and CN2 in almost 3Ј-azido-3Ј-deoxythymidine-5Ј-O-Succ-peptide conjugates
equimolar amounts (see Fig. 2c for a representative time- (unstable in FBS) (28), doxorubicin-14-O-Glut-CNGRC con-
course in PB) without formation of intermediate peptide like jugate (t(1/2) ס 53 min in human serum; t(1/2) ס 76 min in cell
CN3. The hydrolysis of CNF2 in each medium was also rep- culture medium DMEM containing 10% FCS) (33), and an-
resented by a first-order reaction kinetics with its own rate other doxorubicin-14-O-Glut-peptide conjugate AN-152
constant (Table I). The hydrolysis rate constants also show (t(1/2) ס 19.5 min in mouse serum) (29); the greater stability
the presence of enzymatic catalysis accelerating the hydrolysis of CNF1 and CNF2 should be attributed to their ester linkers
of CNF2 in FBS-containing media and cell lysates.
derived from the secondary alcohol 3Ј-OH of 5-FdUrd. Thus,
the ester linker structure is one of the significant factors that
control the pharmacokinetic properties of CNGRC-
conjugated anticancer prodrugs.
Structural Effect of Ester Linkers on the Stability and
5-FdUrd Release Efficiency of 5-FdUrd–
CNGRC Conjugates
Tumor-Cell–Selective Cytotoxicity
The 5-FdUrd–CNGRC conjugates, CNF1 with succinate
and CNF2 with glutarate linkers, showed their stability and
In order to get insight into the anticancer effect and the
5-FdUrd releasing efficiency, depending on their linker struc- specificity to tumor marker APN/CD13 of 5-FdUrd–CNGRC
tures (Table I). The hydrolysis rate constants of CNF1 are conjugates as the prodrugs, in vitro cytotoxicity was evaluated
about 6∼42 times greater than the corresponding values of by an MTT method. Three kinds of cells were used for the in
CNF2. The hydrolysis of CNF1 in each medium also gave vitro cytotoxicity experiments, including APN/CD13 positive
higher 5-FdUrd yield than that of CNF2. For example, the cell line of HT-1080 as well as APN/CD13 negative cell lines
hydrolysis of CNF1 in MEM-10% FBS for 6 h gave 83% yield of HT-29 and MAD-MB-231 (33–35). As summarized in
of 5-FdUrd, while the hydrolysis of CNF2 even for up to 48 h Table II, neither CN1 nor CN2 affected the cell survival frac-
only 57% yield. CNF1 showed much higher selectivity of hy- tion in the concentration range from 10 nM to 1 mM. While
drolysis to release 5-FdUrd than CNF2: the hydrolysis of CNF1 and CNF2 had lower cytotoxicity in comparison with
CNF1 and CNF2 resulted in 82∼93% and 48∼66% selectivity, the typical anticancer agent of 5-FdUrd, their apparent cyto-
respectively (Table I). The higher hydrolysis reactivity of toxicity was related to their hydrolysis rate and 5-FdUrd yield:
CNF1 relative to CNF2 for releasing antitumor agent CNF1 with higher hydrolysis rate and yield of toxic 5-FdUrd
5-FdUrd in both nonenzymatic and enzymatic hydroyses, can (Table I) showed about 6–200 times higher cytotoxicity than
be interpreted in terms of intramolecular catalysis effect of CNF2. It should be more striking that CNF1 showed almost
succinate linker, as is evident from the formation of interme- the same level (only 1.6 times lower) of cytotoxicity as
Table II. Cytotoxicity of 5-FdUrd and Its Prodrug 5-FdUrd-CNGRC Conjugates CNF1 and CNF2
Toward Various Tumor Cells
IC50 (M, mean SD, n ס 5)
Cell lines
HT-1080
HT-29
MDA-MB-231
CNF1
CNF2
5-FdUrd
CN1
CN2
(4.0 1.6) × 10−8
(5.0 0.4) × 10−6
(8.0 0.7) × 10−6
(9.0 3.6) × 10−6
(1.0 0.2) × 10−4
(5.0 0.8) × 10−5
(2.5 0.2) × 10−8
(1.0 0.2) × 10−8
(7.0 2.0) × 10−9
Nontoxic
Nontoxic
—
Nontoxic
Nontoxic
—

388
Zhang, Hatta, Tanabe, and Nishimoto
5-FdUrd toward APN/CD13 positive cell lines of HT-1080, obviously higher than that within MDA-MB-231 cells, and
while being 500 to almost 1000 times lower cytotoxicity than even higher than the corresponding extracellular concentra-
5-FdUrd toward APN/CD13 negative cell lines HT-29 and tion. Similar results were also observed for CNF2. After ex-
MAD-MB-231, respectively. Similar tumor-selectivity was posure to 0.2 mM CNF2, the intracellular concentration of
also observed for CNF2, though its selectivity was obviously CNF2 within HT-1080 cells was up to about 0.1 mM within 1
lower than that of CNF1. CNF2 caused about 360 times lower h and maintaining on this level for more than 12 h, while
cytotoxicity toward HT-1080, while about 7100 to 10,000 being only on a level of 0.04∼0.05 mM within MDA-MB-231
times lower cytotoxicity toward HT-29 and MDA-MB-231 cells during 12 h. After exposure for 12 h, the intracellular
cells, relative to 5-FdUrd.
concentration of 5-FdUrd within HT-1080 and MDA-MB-231
In order to clarify the tumor-selective cytotoxicity, we cells arrived at 0.055 mM and 0.035 mM, respectively. These
further investigated the cellular concentrations of 5-FdUrd are obviously lower than those observed for CNF1, in accord
and prodrugs (CNF1 and CNF2) after incubation of 0.2 mM with the lower hydrolysis reactivity and cytotoxicity of CNF2
5-FdUrd or prodrugs, respectively, in HT-1080 (or MDA- comparing to CNF1. The above results indicate that 5-FdUrd-
MB-231) cell suspension at room temperature. As shown in CNGRC conjugates would target APN/CD13 positive cells
Fig. 4, the intracellular concentration of 5-FdUrd within HT- and enhance the intracellular concentration of their hydroly-
1080 cells exposed to 0.2 mM CNF1 quickly reached the same sate 5-FdUrd, thereby showing the tumor-selective cytotoxic-
level as in the exposure to 0.2 mM 5-FdUrd, while the con- ity. Such a tumor-selective cytotoxicity of CNF1 and CNF2 is
centration within MDA-MB-231 cells exposed to 0.2 mM attributable to the conjugation of 5-FdUrd with a tumor hom-
CNF1 was considerably lower level. It is also noted that the ing peptide CNGRC that has an ability of binding selectively
intracellular concentration of CNF1 within HT-1080 cells was to a specific protein APN/CD13 overexpressed on the tumor
tissue. In a separate experiment, we have also confirmed that
a fluorescent probe molecule bearing the tumor homing pep-
tide CNGRC could penetrate into APN/CD13 positive HT-
1080 cells, but not into APN/CD13 negative MDA-MB-231
cells (data reported elsewhere). In these contexts, the previ-
ous study demonstrated that CNGRC enhances the selective
accumulation of its conjugate with TNF (tumor necrosis fac-
tor alpha) in tumor tissues, thereby resulting in higher immu-
notherapeutic effect in vivo (27,36). On the other hand, in
view of lower cytotoxicity toward APN/CD13 negative cells,
the 5-FdUrd–CNGRC conjugates CNF1 and CNF2 may be
tumor vascular targeting anticancer prodrugs that show lower
side effect on normal cells. A similar result was previously
observed where the doxorubicin–CNGRC conjugate exhib-
ited lower side effect than free doxorubicin (24). The strategy
of tumor vascular targeting prodrug should be further pro-
moted by the investigation on the mapping of human vascu-
lature (37), and the tumor-homing peptide could also be a
useful device for modifying nanoparticles that can deliver
packaged agents to tumor tissues (38).
CONCLUSIONS
A new class of 5-FdUrd prodrugs CNF1 and CNF2 con-
jugated with a tumor-homing cyclic peptide CNGRC by suc-
cinate and glutarate linkers have been designed and synthe-
sized. The linker structure affected the hydrolysis mechanism
and efficiency of 5-FdUrd–CNGRC conjugate to release
5-FdUrd, thus being a useful tool to control the pharmacoki-
netic property. Although both CNF1 and CNF2 had higher
selectivity of cytotoxicity toward APN/CD13 positive tumor
cell line of HT-1080 than toward APN/CD13 negative cell
Fig. 4. (a) Extracellular concentration (◊) and intracellular concen-
tration (ࡗ) of 5-FdUrd after incubation of 0.2 mM 5-FdUrd in
HT-1080 cell suspension; extracellular concentrations of (᭹) CNF1
and (᭡) 5-FdUrd and intracellular concentrations of (᭺) CNF1 and
(⌬) 5-FdUrd after incubation of 0.2 mM CNF1 in HT-1080 cell sus-
pension. (b) Extracellular concentration (◊) and intracellular concen-
tration (ࡗ) of 5-FdUrd after incubation of 0.2 mM 5-FdUrd in MDA-
MB-231 cells suspension; extracellular concentrations of (᭹) CNF1
and (᭡) 5-FdUrd and intracellular concentrations of (᭺) CNF1 and
(⌬) 5-FdUrd after incubation of 0.2 mM CNF1 in MDA-MB-231 cell
suspension.
lines of HT-29 and MDA-MB-231, CNF1 is evidently more
promising as tumor-targeting prodrug of 5-FdUrd than CNF2.
ACKNOWLEDGMENTS
The authors thank Prof. Isao Morishima for kindly sup-
plying the Peptide Synthesis System. The authors are grateful
to Prof. Side Yao, Prof. Yuta Shibamoto, Dr. Satoshi Taka-
hashi, and Dr. Tetsunari Kimura for their valuable discus-
sions.

New Class of 5-Fluoro-2؅-deoxyuridine Prodrugs
389
22. K. Porkka, P. Laakkonen, J. A. Hoffman, M. Bernasconi, and E.
Ruoslahti. A fragment of the HMGN2 protein homes to the nu-
clei of tumor cells and tumor endothelial cells in vivo. Proc. Natl.
Acad. Sci. USA 99:7444–7449 (2002).
23. L. A. Landon and S. L. Deutscher. Combinatorial discovery of
tumor targeting peptides using phage display. J. Cell. Biochem.
90:509–517 (2003).
24. W. Arap, R. Pasqualini, and E. Ruoslahti. Cancer treatment by
targeted drug delivery to tumor vasculature in a mouse model.
Science 279:377–380 (1998).
25. M. Trepel, W. Arap, and R. Pasqualini. In vivo phage display and
vascular heterogeneity: implications for targeted medicine. Curr.
Opin. Chem. Biol. 6:399–404 (2002).
REFERENCES
1. E. Chu, A. Mota, and M. C. Fogarasi. Pharmacology of cancer
chemotherapy. In V. T. DeVita, S. Hellman and S. A. Rosenberg
(eds.), Cancer Principles & Practice of Oncology, 6th ed., Vol. 1,
Lippincott Williams & Wilkins: Philadelphia, 2001, pp. 388–415.
2. C. E. Myers. The pharmacology of the fluoropyrimidines. Phar-
macol. Rev. 33:1–15 (1981).
3. J. A. van Laar, Y. M. Rustum, S. P. Ackland, C. J. van Groenin-
gen, and G. J. Peters. Comparison of 5-fluoro-2Ј-deoxyuridine
with 5-fluorouracil and their role in the treatment of colorectal
cancer. Eur. J. Cancer 34:296–306 (1998).
4. S. Fukushima, T. Kawaguchi, M. Nishida, K. Juni, Y. Yamshita,
M. Takahashi, and M. Nakano. Selective anticancer effects of
3Ј,5Ј-dioctanoyl-5-fluoro-2Ј-deoxyuridine, a lipophilic prodrug of
5-fluoro-2Ј-deoxyuridine, dissolved in an oily lymphographic
agent on hepatic cancer of rabbits bearing VX-2 tumor. Cancer
Res. 47:1930–1934 (1987).
26. R. Pasqualini, E. Koivunen, R. Kain, J. Lahdenranta, M. Saka-
moto, A. Stryhn, R. A. Ashmun, L. H. Shapiro, W. Arap, and E.
Ruoslahti. Aminopeptidase N is a receptor for tumor-homing
peptides and a target for inhibiting angiogenesis. Cancer Res.
60:722–727 (2000).
5. B. S. Vig, P. J. Lorenzi, S. Mittal, C. P. Landowski, H. Shin, H.
Mosberg, J. M. Hilfinger, and G. L. Amidon. Amino acid ester
prodrugs of fluorodeoxyuridine: synthesis and effects of struc-
ture, stereochemistry, and site of esterification on rate of hydro-
lysis. Pharm. Res. 20:1381–1388 (2003).
6. Z. Xia, L. I. Wiebe, G. G. Miller, and E. E. Knaus. Synthesis and
biological evaluation of butanoate, retinoate, and bis(2,2,2-
trichloroethyl)phosphate derivatives of 5-fluoro-2Ј-deoxyuridine
and 2Ј,5-difluoro-2Ј-deoxyuridine as potential dual action anti-
cancer prodrugs. Arch. Pharm. (Weiheim) 332:286–294 (1999).
7. K. Tanabe, Y. Mimasu, A. Eto, Y. Tachi, S. Sakakibara, M. Mori,
H. Hatta, and S. Nishimoto. One-electron reduction characteris-
tics of N(3)-substituted 5-fluorodeoxyuridines synthesized as ra-
diation-activated prodrugs. Bioorg. Med. Chem. 11:4551–4556
(2003).
8. Y. Shibamoto, Y. Tachi, K. Tanabe, H. Hatta, and S. Nishimoto.
In vitro and in vivo evaluation of novel antitumor prodrugs of
5-fluoro-2Ј-deoxyuridine activated by hypoxic irradiation. Int. J.
Radiat. Oncol. Biol. Phys. 58:397–402 (2004).
9. Y. Wei, Y. Yan, D. Pei, and B. Gong. A photoactivated prodrug.
Bioorg. Med. Chem. Lett. 8:2419–2422 (1998).
27. F. Curnis, G. Arrigoni, A. Sacchi, L. Fischetti, W. Arap, R.
Pasqualini, and A. Corti. Differential binding of drugs containing
the NGR motif to CD13 isoforms in tumor vessels, epithelia, and
myeloid cells. Cancer Res. 62:867–874 (2002).
28. H. Tamamura, A. Omagari, K. Hiramatsu, T. Kanamoto, K.
Gotoh, K. Kanbara, N. Yamamoto, H. Nakashima, A. Otaka, and
N. Fuji. Synthesis and evaluation of biofunctional anti-HIV
agents based on specific CXCR4 antagonist-AZT conjugation.
Bioorg. Med. Chem. 9:2179–2187 (2001).
29. A. Nagy, A. Polnowski, and A. V. Schally. Stability of cytotoxic
luteinizing hormone-releasing hormone conjugate (AN-152) con-
taining doxorubicin 14-O-hemiglutarate in mouse and human se-
rum in vitro: implications for the design of preclinical studies.
Proc. Natl. Acad. Sci. USA 97:829–834 (2000).
30. H. Tamamura, T. Ishihara, H. Oyake, M. Imai, A. Otaka, T.
Ibuka, R. Arakaki, H. Nakashima, T. Murakami, M. Waki, A.
Matsumoto, N. Yamamoto, and N. Fujii. Convenient one-pot
synthesis of cystine-containing peptides using the trimethylsilyl
chloride-dimethyl sulfoxide/trifluoroacetic acid system and its ap-
plication to the synthesis of bifunctional anti-HIV compounds. J.
Chem. Soc., Perkin Trans. 1 1:495–500 (1998).
10. G. A. Koning, J. A. Kamps, and G. L. Scherphof. Efficient in-
tracellular delivery of 5-fluorodeoxyuridine into colon cancer
cells by targeted immunoliposomes. Cancer Detect. Prev. 26:299–
307 (2002).
11. G. A. Koning, A. Gorter, G. L. Scherphof, and J. A. Kamps.
Antiproliferative effect of immunoliposomes containing 5-fluo-
rodeoxyuridine-dipalmitate on colon cancer cells. Br. J. Cancer
80:1718–1725 (1999).
12. A. Goerlach, K. G. Krauer, I. F. McKenzie, and G. A. Pietersz. In
vitro antitumor activity of 2Ј-deoxy-5-fluorouridine-monoclonal
antibody conjugates. Bioconjug. Chem. 2:96–101 (1991).
13. D. C. Drummond, O. Meyer, K. Hong, D. B. Kirpotin, and D.
Papahadjopoulos. Optimizing liposomes for delivery of chemo-
therapeutic agents to solid tumors. Pharmaco. Rev. 51:691–743
(1999).
31. T. Mosmann. Rapid colorimetric assay for cellular growth and
survival: application to proliferation and cytotoxicity assays. J.
Immunol. Methods 65:55–63 (1983).
32. Y. Shibamoto, Y. Mimasu, Y. Tachi, H. Hatta, and S. Nishimoto.
Comparison of 5-fluorouracil and 5-fluoro-2Ј-deoxyuridine as an
effector in radiation-activated prodrugs. J. Chemother. 14:390–
396 (2002).
33. Y. van Hensbergen, H. J. Broxterman, Y. W. Elderkamp, J.
Lankelma, J. C. Beers, M. Heijn, E. Boven, K. Hoekman, and
H. M. Pinedo. A doxorubicin-CNGRC-peptide conjugate with
prodrug properties. Biochem. Pharmacol. 63:897–908 (2002).
34. J. S. Shim, J. H. Kim, H. Y. Cho, Y. N. Yum, S. H. Kim, H. J.
Park, B. S. Shim, S. H. Choi, and H. J. Kwon. Irreversible inhi-
bition of CD13/aminopeptidase N by the antiangiogenic agent
curcumin. Chem. Biol. 10:695–704 (2003).
14. S. K. Hobbs, W. L. Monsky, F. Yuan, W. G. Roberts, L. Griffith,
V. P. Torchilin, and R. K. Jain. Regulation of transport pathways
in tumor vessels: role of tumor type and microenvironment. Proc.
Natl. Acad. Sci. USA 95:4607–4612 (1998).
15. I. Brigger, C. Dubernet, and P. Couvreur. Nanoparticles in cancer
therapy and diagnosis. Adv. Drug Deliv. Rev. 54:631–651 (2002).
16. P. D. Senter and C. J. Springer. Selective activation of anticancer
prodrugs by monoclonal antibody-enzyme conjugates. Adv. Drug
Deliv. Rev. 53:247–264 (2001).
17. E. C. Ko, X. Wang, and S. Ferrone. Immunotherapy of malignant
diseases. Challenges and strategies. Int. Arch. Allergy Immunol.
132:294–309 (2003).
18. E. Ruoslahti. Specialization of tumor vasculature. Nat. Rev. Can-
cer 2:83–90 (2002).
35. D. Riemann, A. Kehlen, K. Thiele, M. Löhn, and J. Langner.
Induction of aminopeptidase N/CD13 on human lymphocytes af-
ter adhesion to fibroblast-like synoviocytes, endothelial cells, ep-
ithelial cells, and monocytes/macrophages. J. Immunol. 158:3425–
3432 (1997).
36. F. Curnis, A. Sacchi, L. Borgna, F. Magni, A. Gasparri, and A.
Corti. Enhancement of tumor necrosis factor alpha antitumor
immunotherapeutic properties by targeted delivery to aminopep-
tidase N (CD13). Nat. Biotechnol. 18:1185–1190 (2000).
37. W. Arap, M. G. Kolonin, M. Trepel, J. Lahdenranta, M. Cardo-
Vila, R. J. Giordano, P. J. Mintz, P. U. Ardelt, V. J. Yao, C. I.
Vidal, L. Chen, A. Flamm, H. Valtanen, L. M. Weavind, M. E.
Hicks, R. E. Pollock, G. H. Botz, C. D. Bucana, E. Koivunen, D.
Cahill, P. Troncoso, K. A. Baggerly, R. D. Pentz, K. A. Do, C. J.
Logothetis, and R. Pasqualini. Steps toward mapping the human
vasculature by phage display. Nat. Med. 8:121–127 (2002).
38. F. Pastorino, C. Brignole, D. Marimpietri, M. Cilli, C. Gambini,
D. Ribatti, R. Longhi, T. M. Allen, A. Corti, and M. Ponzoni.
Vascular damage and anti-angiogenic effects of tumor vessel-
targeted liposomal chemotherapy. Cancer Res. 63:7400–7409
(2003).
19. P. Alessi, C. Ebbinghaus, and D. Neri. Molecular targeting of
angiogenesis. Biochim. Biophys. Acta 1654:39–49 (2004).
20. E. Ruoslahti and D. Rajotte. An address system in the vascula-
ture of normal tissues and tumors. Annu. Rev. Immunol. 18:813–
827 (2000).
21. R. Pasqualini and E. Ruoslahti. Organ targeting in vivo using
phage display peptide libraries. Nature 380:364–366 (1996).