Design and synthesis of novel prodrugs of 2′-deoxy-2′-methylidenecytidine activated by membrane dipeptidase overexpressed in tumor tissues

Article abstract of DOI:10.1016/j.bmcl.2007.01.066

DNA microarray analysis comparing human tumor tissues with normal tissues including hematopoietic progenitor cells resulted in identification of membrane dipeptidase as a prodrug activation enzyme. Novel prodrugs of 2′-deoxy-2′-methylidenecytidine (DMDC)

Full text of DOI:10.1016/j.bmcl.2007.01.066

Bioorganic & Medicinal Chemistry Letters 17 (2007) 2241–2245
Design and synthesis of novel prodrugs of 20-deoxy-20-
methylidenecytidine activated by membrane
dipeptidase overexpressed in tumor tissues
Yasunori Kohchi,a Kazuo Hattori,a Nobuhiro Oikawa,a Eisaku Mizuguchi,a
Yoshiaki Isshiki,a Kohsuke Aso,a Kiyoshi Yoshinari,a Haruyoshi Shirai,a
Masanori Miwa,b Yukiko Inagaki,b Masako Ura,b Kotaroh Ogawa,b
Hisafumi Okabe,b Hideo Ishitsukac and Nobuo Shimmaa,*
aDepartment of Chemistry Research 2, Kamakura Research Laboratories, Chugai Pharmaceutical Co., Ltd,
200 Kajiwara, Kamakura, Kanagawa 247-8530, Japan
bDepartment of Pharmaceutical Research 3, Kamakura Research Laboratories, Chugai Pharmaceutical Co., Ltd,
200 Kajiwara, Kamakura, Kanagawa 247-8530, Japan
cRoche Diagnostics Japan, 2-6-1 Shiba Minato, Tokyo 105-0014, Japan
Received 6 December 2006; revised 14 January 2007; accepted 19 January 2007
Available online 27 January 2007
Abstract—DNA microarray analysis comparing human tumor tissues with normal tissues including hematopoietic progenitor cells
resulted in identification of membrane dipeptidase as a prodrug activation enzyme. Novel prodrugs of 20-deoxy-20-methylidenecyti-
dine (DMDC) including compound 23 that are activated by membrane dipeptidase (MDP) preferentially in tumor tissue were
designed and synthesized to generate the active drug, DMDC, after hydrolysis of the dipeptide bond followed by spontaneous cycli-
zation of the promoiety.
Ó 2007 Elsevier Ltd. All rights reserved.
In recent years, several molecular targeting agents such
as signal transduction inhibitors have been developed.
These agents have better safety profiles than convention-
al cytotoxic agents, but their efficacy is limited to certain
populations of tumors. The conventional cytotoxic
agents, on the other hand, usually have a wide antitumor
spectrum, but their narrow therapeutic window due to
severe myelosuppression and intestinal toxicity often
limits the clinical efficacy. Development of tumor-acti-
vated prodrugs of cytotoxic agents is an alternative op-
tion to provide safer and more efficacious antitumor
agents. We previously developed an oral fluoropyrimi-
dine, capecitabine (Fig. 1), which is sequentially convert-
ed to 5-fluorouracil selectively in tumors1,2 by the
enzymes overexpressed in the liver and tumor tissues
but not in bone marrow. Capecitabine is now widely used
for treatment of metastatic breast and colorectal cancers.
NH2
N
NHCO2(CH2)4CH3
F
N
O
N
O
O
O
N
AcO
OAc
HO
OH
Capecitabine
1
Figure 1. Tumor targeting prodrugs.
We also reported the tumor-activated prodrug of
5-vinyluracil 1, a dihydropyrimidine dehydrogenase
inhibitor, as a potentiator of capecitabine.3
To further expand this strategy, we investigated enzymes
useful for prodrug activation that are preferentially
overexpressed in human tumors. DNA microarray anal-
ysis of human tumor tissues versus normal tissues
including hematopoietic progenitor cells yielded mem-
Keywords: Tumor targeting; Prodrug; Antitumor; Cytotoxic agent.
*
Corresponding author. Tel.: +81 467 47 2280; fax: +81 467 45
6824; e-mail: shinmanbo@chugai-pharm.co.jp
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.01.066

2242
Y. Kohchi et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2241–2245
brane dipeptidase (MDP) as a prodrug activation
enzyme. 20-Deoxy-20-methylidenecytidine (DMDC) is a
DNA polymerase inhibitor showing potent antitumor
activity against a wide variety of tumors.4 However,
because of severe hematological toxicity in human, clin-
ical development of DMDC has failed. Here, we report
the design and synthesis of novel prodrugs of DMDC
that are activated by MDP preferentially in tumor
tissues.
dipeptides. We designed prodrugs of DMDC 2 that
are activated by MDP, preferentially in tumor tissues
after hydrolysis of the dipeptide bond followed by
spontaneous cyclization of the promoiety as shown in
Figure 3.
In this design, the N4-amino group of DMDC was linked
to serine or threonine derivatives of a dipeptide. We first
confirmed the concept of this spontaneous cyclization
leading to the generation of DMDC by synthesizing
two model compounds (5 and 9) as shown in Figure 4.
Identification of membrane dipeptidase (MDP) as a pro-
drug activation enzyme. DNA microarrays were used to
identify enzymes for designing tumor-activated pro-
drugs of DMDC by comparing the mRNA levels of
genes in normal and tumor tissues (41 human colorec-
tum tumors, 30 gastric tumors, 41 non-small cell lung
carcinomas, 24 breast tumors, 15 ovarian tumors, 53
hepatocellular carcinomas, and 15 non-tumorous liver
tissue and hematopoietic progenitor cells). We found
MDP is overexpressed in human colorectal and stomach
tumor tissues but not in adjacent normal tissues, liver, or
hematopoietic progenitor cells such as colony-forming
units granulocyte-macrophage (CFU-GM) (Fig. 2).4
When the benzyl protecting groups were removed by cat-
alytic hydrogenolysis, both 5 and 9 generated cytidine
derivative 8 with simultaneous formation of cyclized
compounds (7 and 11). Release of compound 8 was
much faster in the threonine prodrug 5 (T1/2 of
<30 min) than the corresponding serine prodrug 9
(T1/2 of 48 h). The structures of 7 and 11 were confirmed
1
by LC–MS and H/13C NMR.
We then synthesized a series of dipeptide prodrugs and
examined whether or not this large side chain in dipep-
tides including DMDC is recognized by MDP. Figure 5
shows the synthetic route of representative compound
23. Compound 16 was prepared from BOC-(D)-
Thr(OBn)-OH 12 and Fmoc-cyclohexylalanine 14 in 5
steps. Compound 23 was synthesized by coupling with
activated p-nitrophenylcarbonate derivative 17 and
30,50-bis-O-TBS-DMDC 18, followed by removal of
the FMOC, TBS, and TMSE groups with TBAF.5
Design and synthesis of tumor activated DMDC prodrugs.
Membrane dipeptidase is an enzyme that hydrolyzes
Colon
Stomach
1600
(intensity)
1600
71%
% of present call
The results of the clarification of the substrate specificity
of MDP are summarized in Figure 6.
Tumor
26%
0
0
Regarding a substituent at R1, both threonine and serine
are recognized by MDP, but isopropyl serine 27 is not a
substrate. As a substituent at R2, a-methyl-(L)-serine
34 is recognized by MDP. Replacement of the terminal
carboxyl group with hydroxyl methyl 30, an amide 28,
and an ester group 29 abolished the recognition by
MDP. Therefore, we found the terminal carboxyl group
is essential for the recognition by MDP.
n=31
Stomach
1600
n=21
Liver
CFU-GM
1600
Colon
(intensity)
1600
1600
Normal
0%
4%
0%
0%
0
0
0
0
Regarding the stereochemistry of the C-terminal amino
acid, both L- and D-forms are recognized by MDP (26
and 23).
n=17
n=26
n=10
n=4
Figure 2. mRNA levels of genes of MDP in normal and tumor tissues.
R3
O
R3
O
OH
R2
OH
R2
O
NH2
R1
O
O
OH
O
NH2
NH2
Spontaneous
O
OH
NH
N
HN
N
O
2
NH
OH
HO
N
N
O
N
O
O
N
HO
DMDC
O
O
HO
Cleavage enzyme
(tumor specific)
3
R3
O
R1 = alkyl, aralkyl, aryl, etc.
R2 = H, Me
OH
R2
O
R3 = H, Me
NH
O
4
Figure 3. Design of tumor-activated DMDC prodrugs.

Y. Kohchi et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2241–2245
2243
O
O
NH2
N
HN
N
O
O
O
HN
N
O
O
Fast
T1/2 < 30min
OH
Pd/C, H2
OTBS
O
O
OTBS
N
OTBS
O
N
NH2
N
O
HN
O
O
MeOH
10 min
O
O
TBSO OTBS
TBSO OTBS
O
TBSO OTBS
8
6
O
y. quant.
OH
N
5
H
O
7
O
O
NH2
HN
N
O
O
HN
N
O
O
Slow
T1/2 = 48 h
OH
Pd/C, H2
N
N
OTBS
O
OTBS
OTBS
N
NH2
O
O
N
HN
O
O
MeOH
10min
O
O
O
O
TBSO OTBS
TBSO OTBS
TBSO OTBS
O
8
O
y. quant.
OH
10
N
H
11
9
O
Figure 4. Proof of concept study using model compounds.
1) TMSethanol, DCC/DCM
2) Pd/C, H2,EtOAc
O
O
OH
O
OH
O
3) HCl/EtOAc
HN
Si
O
94%
O
NH2 HCl
12
13
OH
O
O
O
O
HOSu
DCC
OTMSE
OH
O
Et3N
p-NO2PhOCOCl, pyr.
OSu
NHFmoc
NHFmoc
HN
HN
CH2Cl2
84%
CH2Cl2
97%
dioxane
O
15
14
16
O
N+
O
TBSO
TBSO
NH2
O
O
O
N
N
O
TBSO
N
H
O
O
O
O
O
O
O
O
O
N
OH
N
N
O
O
O
TBAF
OTMSE
1
O
H
N
OTMSE
NHFmoc
O
NHFmoc
N
OH
NH2
TBSO
HN
THF
75%
THF 60ºC
88%
O
HN
HO
HN
19
23
17
Figure 5. Synthesis of compound 23.
R1 = Me (better), H (allowed)
( i-Pr: not a substrate for MDP)
R2 = H, Me (allowed)
COOH: essential
(-CH2OH, -CONH2, -CO2R
not a substrate for MDP )
R1
O
O
R2
OH
N
H
O
OH
O
N
NH2: essential
NH
N
NH2
(-NHMe, -NMe2, NHCHO,
NHCO2R: not a substrate for
MDP )
O
HO
O
Terminal substituent
Aliphatic residue: desirable
Figure 6. Summary of the substrate specificity of MDP.

2244
Y. Kohchi et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2241–2245
Table 1. Results of cytotoxicity and plasma stability of prodrugs
Cytotoxicitya IC50 (lM)
MDP selectivity
Plasma stability
residual % after
2 h
R3
R2
O
O
O
OH
N
H
NH
O
R5
R1
N
N
R4
O
HO
Compound R1
R2 R3
R4
—
R5
HCT116 MDP(ꢀ) HCT116 MDP(+) Relative IC50 ratio Mouse Human
DMDC
20
21
—
—
H
H
H
H
H
H
H
H
H
H
H
H
H
H
—
—
Phenyl
Phenyl
0.3–0.7
8.7
0.3–0.7
106.7 102.2
(S)Me
(S)Me
H
(S)COOH
(S)COOH
(R)COOH
(R)COOH
(R)COOH
(R)COOH
(S)COOH
(R)COOH
(R)NH2
(S)NH2
(S)NH2
(S)NH2
(S)NH2
(S)NH2
(S)NH2
(S)NH2
15.7
0.6
0.8
6.1
6.1
4.4
3.2
5.5
1.0
1.1
1.0
1.0
1.0
1.3
NDb
ND
52.3
100.5
86.7
97.7
ND
88.5
95.1
0
84.3
90.0
87.4
97.8
86.0
85.5
55.6
89.3
103.9
87.2
85.4
85.4
86.8
ND
4.1
Cyclohexyl 17.0
Cyclohexyl 18.8
5.1
2.8
22
23
(S)Me
(S)Me
(S)Me
(S)Me
(S)i-Pr
(S)Me
(S)Me
(S)Me
(S)Me
(S)Me
(S)Me
H
3.1
24
25
Biphenyl
Naphtyl
i-Propyl
67.2
26.3
4.4
15.4
8.3
26
27
0.8
Cyclohexyl 20.3
20.3
7.3
28
29
(R)CONH2 (S)NH2
(R)COO-i-Pr (S)NH2
(R)CH2OH (S)NH2
Cyclohexyl
Cyclohexyl
8.1
7.4
8.2
30
31
Cyclohexyl 10.8
Cyclohexyl 47.8
Cyclohexyl 56.6
10.5
48.9
44.9
>50.0
5.9
ND
91.4
101.3
ND
19.4
101.5
(R)COOH
(R)COOH
(R)COOH
(S)NHMe
(S)NMe2
32
33
(S)NHCHO Cyclohexyl >50.0
(S)NH2
(S)NH2
—
34
35
Me (S)COOH
Me (R)COOH
Cyclohexyl 72.3
Cyclohexyl 51.9
12.3
2.4
95.9
89.1
H
21.8
a Assay condition; 24 h pulse treatment and 72 h incubation.
b ND, not done
The terminal amino group is essential for recognition
by MDP. Mono-methylamino 31, di-methylamino 32,
N-formyl 33, and the carbamate derivatives (not listed)
are not recognized by MDP. For the N-terminal substi-
tuent, lipophilic and aliphatic residue is preferable for
enzyme recognition (22, 23, 24, 25, 26, and 34).
23 with (D)-Thr-(L)-cyclohexylalanine (L-Cha) showed
selectively inhibited the growth of the MDP-positive
HCT116 (IC50 = 3.1 lM vs IC50 = 18.8 lM in
MDP(negative) HCT116). In an attempt to improve
the selectivity to other peptidases, we synthesized (L)-
a-methylserine-(L)-Cha derivative 34 and found further
improvement of selectivity to MDP-positive HCT116
(IC50 = 5.9 lM vs IC50 = 72.3 lM in MDP-negative
HCT116). Regarding plasma stability, both compound
23 and compound 34 were stable in human plasma,
but the latter was rather unstable in mouse plasma (only
19.4% remained after 2 h incubation).
We also examined the ability of the prodrugs to inhibit
the growth of human colon cancer HCT116 cells that
does not overexpress MDP (MDP-negative HCT116)
and those transfected with MDP cDNA (MDP-positive
HCT116) for proof-of-concept in vitro and stability
under different pH conditions and in plasma (mice and
human). The results are summarized in Table 1.
The highly selective conversion of compound 23 to
DMDC in the cell free enzyme assay with MDP-positive
HCT116 versus MDP-negative HCT 116 and hemato-
poietic progenitor cells was further confirmed by direct
measurement of the prodrug and DMDC levels by
LC–MS (Fig. 7). In the MDP-negative cell line and
Among the (L)-threonine derivatives, 20 and 21 having
L- and D-Phe, respectively, as a trigger moiety showed
no selectivity in cytotoxic activity in MDP-positive
and MDP-negative HCT116 cells, whereas compound
hematopoietic
progenitor cells
HCT116
MDP-negative
HCT116
MDP-positive
120
100
80
60
40
20
0
120
100
80
60
40
20
0
100
80
60
40
20
0
0
5
10
0
2
4
6
8
0
2
4
6
8
(hr)
(hr)
(hr)
DMDC, compound 23
Figure 7. The in vitro conversion of compound 23 to DMDC in MDP-negative HCT116, MDP-positive HCT116, and hematopoietic progenitor
cells.

Y. Kohchi et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2241–2245
Monkey
2245
Mouse
brain
heart
lung
liver
pancreas
spleen
kidney
stomach
duodenum
jejunum
illeum
colon
rectum
S.I.
L.I.
bone marrow
pasmal
Colo205
0
1
2
3
4
5
1.0
0.8
0.6
0.4
0.2
0
( nmol (DMDC) / mg protein / min )
( nmol (DMDC) / mg protein / min )
Figure 8. The results of level of DMDC by crude enzyme extracts from mouse and monkey tissues.
hematopoietic progenitor cells, compound 23 was quite
stable. Thus, in vitro proof-of-concept of the tumor-ac-
tivated prodrugs was achieved.
vs normal tissue including hematopoietic progenitor
cells) performed. A general concern with the prodrug
approach is species difference in prodrug activation since
the absolute enzyme levels in all human tissues are
unknown. In the case of compound 23, some species
difference was observed between mice and monkeys.
Further investigation is needed to assure product safety
and efficacy for clinical development of this type of
prodrug.
The tumor selective conversion of compound 23
(50 lM) to DMDC was further examined with extracts
of COLO 205 human colon cancer xenograft that ex-
pressed MDP and was compared with extracts of vari-
ous normal tissues from mice and monkeys (Fig. 8).
Compound 23 was efficiently converted to DMDC in
COLO 205, whereas its conversion was very low in other
normal tissues from mice. On the other hand, among the
monkey tissues examined, the level of DMDC was much
higher in kidney than, and nearly equal in jejunum and
ileum to, that in COLO 205.
References and notes
1. Shimma, N.; Umeda, I.; Arasaki, M.; Murasaki, C.;
Masubuchi, K.; Kohchi, Y.; Miwa, M.; Ura, M.; Sawada,
N.; Tahara, H.; Kuruma, I.; Horii, I.; Ishitsuka, H. Bioorg.
Med. Chem. Lett. 2000, 8, 1697.
2. Ishitsuka, H.; Shimma, N.; Horii, I. Yakugaku Zasshi 1999,
119, 881.
3. Hattori, K.; Kohchi, Y.; Oikawa, Nobuhiro; Suda, H.;
Ura, M.; Ishikawa, T.; Miwa, M.; Endoh, M.; Eda,
H.; Tanimura, H.; Kawashima, A.; Horii, I.; Ishit-
suka, H.; Shimma, N. Bioorg. Med. Chem. Lett. 2003,
13, 867.
4. (a) Takenuki, K.; Matsuda, A.; Ueda, T.; Sasaki, T.; Fujii,
A.; Yamagami, K. J. Med. Chem. 1988, 31, 1063; (b) Ueda,
T.; Matsuda, A.; Yoshimura, Y.; Takenuki, K. Nucleosides
Nucleotides 1989, 8, 743; (c) Matsuda, A.; Takenuki, K.;
Tanaka, M.; Sasaki, T.; Ueda, T. J. Med. Chem. 1991, 34,
812; (d) Yamagami, K.; Fujii, A.; Arita, M.; Okumoto, T.;
Sakata, S.; Matsuda, A.; Ueda, T.; Sasaki, T. Cancer Res.
1991, 51, 2319.
Species difference in tissue distribution of MDP is of con-
cern for development of oral compound 23 for human.
Further investigation is required to identify prodrugs
with less species difference in enzymatic activation.
In conclusion, this study demonstrates the gene expres-
sion profiling of tumor and normal tissues including
hematopoietic progenitor cells by DNA microarray is
an effective approach in identifying enzymes for the
design of tumor-activated prodrugs of cytotoxic agents.
We designed compound 23 which was selectively con-
verted to an active drug, DMDC, by MDP-positive
tumor tissues. The desired tumor selective conversion
was clearly demonstrated in vitro from the cytotoxicity
assays (tumor cells with high vs low MDP), cell free
enzyme assays, and tissue conversion assays (tumor
5. Ishitsuka, H.; Okabe, H.; Shimma, N.; Tsukuda, T.;
Umeda, I., PATENT WO 2003043631 A2, 2003.