Design and synthesis of highly potent fumagillin analogues from homology modeling for a human MetAP-2

Article abstract of DOI:10.1016/S0960-894X(99)00577-6

New fumagillin analogues were designed through structure-based molecular modeling with a human methionine aminopeptidase-2. Among the fumagillin analogues, cinnamic acid ester derivative CKD-731 showed 1000-fold more potent proliferation inhibitory activi

Full text of DOI:10.1016/S0960-894X(99)00577-6

Bioorganic & Medicinal Chemistry Letters 10 (2000) 39±43
Design and Synthesis of Highly Potent Fumagillin Analogues from
Homology Modeling for a Human MetAP-2
Cheol Kyu Han, a,c Soon Kil Ahn, a,Ã Nam Song Choi, a Ryung Kee Hong, a
Seung Kee Moon, a Hyoung Sik Chun, a Sang Joon Lee, a Jung Woo Kim, a
Chung Il Hong, a Deukjoon Kim, b Jeong Hyeok Yoon c and Kyoung Tai No c,Ã
aChong Kun Dang Research Institute 15-20, Osackdang-ri, Seonggeo-up, Cheon An, Chungcheongnamdo, 330-830, South Korea
bCollege of Pharmacy, Seoul National University, San 56-1, Shinrim-dong, Kwanak-gu, Seoul, 151-742, South Korea
cCAMD Research Center and Department of Chemistry, Soong Sil University, Seoul, 156-743, South Korea
Received 2 August 1999; accepted 14 October 1999
AbstractÐNew fumagillin analogues were designed through structure-based molecular modeling with a human methionine ami-
nopeptidase-2. Among the fumagillin analogues, cinnamic acid ester derivative CKD-731 showed 1000-fold more potent prolifera-
tion inhibitory activity on endothelial cell than TNP-470. # 1999 Elsevier Science Ltd. All rights reserved.
Introduction
drug design of fumagillin analogues from homology
modeling.9
The concept of treating cancer by inhibition of angio-
genesis (new blood vessel formation), which was pro-
posed by Folkman,1,2 is a promising strategy for cancer
therapy.3 Since Ingber et al. discovered that fumagillin
(1) from Aspergillus fumigatus inhibits new blood vessel
growth, many semisynthetic fumagillin analogues have
been synthesized from fumagillol (2), the hydrolysis
product of fumagillin.4 Among these analogues,
O-(chloroacetylcarbamoyl)fumagillol (TNP-470, AGM-
1470) is currently in phase III clinical trials for the
treatment of a variety of cancers.5
In this communication, we wish to describe our e€ort to
develop highly potent fumagillin analogues based on a
3D structure of human MetAP-2 by means of homology
modeling.
Results and Discussion
Homology models
MetAPs are metal-dependent enzymes, and the crystal
structure of E. coli MetAP-1 reveals that, two cobalt
ions are ligated by two aspartic acids, a histidine and
two glutamic acids in an active site. Five amino acids
(D, D, H, E and E; boxed letters) which bind to two
cobalts ions are conserved in both MetAP-110,11 and
MetAP-2 (Fig. 1).12±14 Based on these ®ndings, we per-
formed the homology modeling of human MetAP-2.
The sequence alignments of Pyrococcus furiosus MetAP-
2 and E. coli MetAP-1 showed a sequence identity of 35
and 28% to human MetAP-2, respectively. The ®ve
amino acids in the R1±R5 domains are also conserved in
P. furiosus MetAP-2 (Fig. 1). A three dimensional
model of human MetAP-2 was built from known X-ray
coordinates of E. coli MetAP-1 (PDB: 1MAT) and P.
furiosus MetAP-2 (PDB: 1XGS) using MODELER.15
The underlying molecular mechanism of the inhibition
of angiogenesis by these fumagillin derivatives remained
unknown until Crews and Liu independently identi®ed
a fumagillin-binding protein, methionine aminopepti-
dase type 2 (MetAP-2).6,7 Recently, Clardy and co-
workers have reported the structure of a human
MetAP-2 fumagillin complex, where the spiro-epoxide
of fumagillin forms a covalent bond with the His231 in
the active site of MetAP-2.8 These reports have promp-
ted us to disclose our study on the structure-based
Keywords: Antitumor compounds; molecular modeling/mechanics.
ÃCorresponding author. E-mail: skahn@ckd.co.kr
0960-894X/00/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(99)00577-6

40
C. K. Han et al. / Bioorg. Med. Chem. Lett. 10 (2000) 39±43
Structure optimization of fumagillin derivatives using
structure-based design
Synthesis of fumagillin derivatives
Fumagillol (2) was acylated with NaH and an acid
chloride to provide compound 3.5 Treatment of the
known phenoxycabonyl fumagillol with amine pro-
duced the carbamate 4. Preparation of carbonate 5 was
achieved by coupling of fumagillol (2) with benzyl
chloroformate. Xanthate derivative 6 was prepared
from 2 using CS2 and benzyl bromide (Scheme 1).
With the human MetAP-2 model, a series of fumagillin
derivatives 1±7 were designed (Table 1). In designing the
inhibitors, the distance between the methylene carbon of
the spiro-epoxide moiety of the fumagillin and the NH
group of His231 was constrained to 3 A because they
form a covalent bond in the complex. The docked
structure of human MetAP-2 and CKD-731, the most
potent inhibitor of the series is illustrated in Figure 2. In
the complex, a water molecule is bound to two cobalt
ions. The distances between the cobalt ions and the
water oxygen are 1.95 A and 1.97 A. The same water
forms a hydrogen bond with the oxygen of the spiro-
epoxide of CKD-731, which might facilitate the opening
of the epoxide ring. The structure also reveals another
water molecule which forms hydrogen bonds with the
oxygens of the methoxy group and the epoxide of the
alkyl side chain of CKD-731. This water also plays a
role in stabilizing His231. A pocket formed with His339,
Ile338, Phe219 and Tyr444 is almost fully occupied by
the terminal isopropylidene group. Since there are a
well-de®ned hydrophobic valley formed with Leu328
and Leu447, and a large pocket which is connected to
this valley and surrounded by Asn327, Val374, Asp376
and His375, we designed fumagillin analogues with
functional groups which can bind eciently to this val-
ley and pocket. To optimize hydrophobic interactions
with the valley and van der Waals contacts at the
pocket, we introduced several functional groups as
described in Table 1. We have designed fumagillin deri-
vatives by introducing functional groups with an aro-
matic ring which can interact hydrophobically with the
Leu447 of human MetAP-2. Cinnamic acid esters 3b±
3e, phenylalkanoic acid esters 3f±3g, benzyl carbamates
4a±4d and benzyl carbonates 5a±5c have been designed.
For comparison purposes, alkyl acid ester 3a and xan-
thate 6 were also prepared.
Biological assays
Antiproliferating activities of fumagillin derivatives
were evaluated against calf pulmonary artery (SPAE,
ATTIC HRL 209) endothelial cells, lymphoma EL-4
cells and murine leukemia P388D1. IC50 values were
colorimetrically measured by SRB (CPAE cell) or MTT
(EL-4, P388D) methods. The biological data for com-
pounds 1±7 are shown in Table 1.
Structure±activity relationships
Among the designed compounds, the cinnamic acid
esters 3b±3d and benzyl carbamates 4a±4d showed more
potent activity than TNP-470, while the phenylalkanoic
acid esters 3f±3g and benzyl carbonates 5a±5c were less
active. It seems that the aromatic ring should be posi-
tioned to contact with Leu447 for maximizing hydro-
phobic interaction. The trans-cinnamic acid esters have
an optimum ®xed geometry for the hydrophobic inter-
action with Leu447. The activity tends to decrease as the
bonds between the carbonyl group and the aromatic
ring in the group R are more freely rotatable. In the
docking model of the phenylalkanoic acid esters 3f±3g
and benzyl carbonate derivatives 5a±5c, the aromatic
rings point away from the Leu447. To support this
assumption, we also prepared cis-cinnamic acid ester
derivative 3e which has an aromatic ring but cannot
interact with Leu447.16 The cell proliferation inhibitory
Figure 1. Sequence alignment of the R1±R5 domains of MetAP-1 and MetAP-2.

C. K. Han et al. / Bioorg. Med. Chem. Lett. 10 (2000) 39±43
41
Table 1. In vitro cell proliferation inhibitory activity of fumagillin derivatives against lymphoma EL-4 cell, calf pulmonary artery endothelial
(CPAE) cell and murine leukemia P388D1a
Subst. R
EL-4
0.42
CPAE
0.37
P388D1
>1
Subst. R
EL-4
0.01
CPAE
0.07
P388D1
>1
1
2
4a
4b
5.16
5.05
15
12
>10
>1
0.042
0.014
0.044
>1
>1
3a
4c
0.0054
3b
3c
0.05
0.046
>1
>1
4d
5a
0.042
0.044
>1
0.00019
0.00006
42
18
ND
3d
0.00015
0.00003
>1
5b
2.47
0.11
>1
b
3e
3f
12
36
24
44
NDb
5c
6
153
14
2.3
42
ND
ND
ND
ND
3g
83
125
7
0.03
0.04
>10
aIC50 values in ng/mL.
bND: not determined.
activity of the cis-cinnamic acid ester derivative 3e was
105 fold less than that of the trans-cinnamic acid ester
derivatives 3b±3d. Moreover the xanthate 6 was less
active than the carbonyl ester derivatives 3b±3d.
In summary, a human MetAP-2 model was built
through homology modeling and several potent inhibi-
tors were designed based on this model structure.
Although the cell-level assay results do not always

42
C. K. Han et al. / Bioorg. Med. Chem. Lett. 10 (2000) 39±43
Figure 2. The docked structure of CKD-731 complex with MetAP-2.
Scheme 1.
re¯ect the molecular-level interactions between enzyme
and ligands, this study does provide some clues to
designing fumagillin analogues as antiangiogenic antic-
ancer agents.
thanks the Basic Science Research Institute Program of
the Ministry of Education (BSRI-96-3448) for research
grants.
References and Notes
Acknowledgements
1. Folkman, J.; Shing, Y. J. Biol. Chem. 1992, 267, 10931.
2. Folkman, J. Adv. Cancer Res. 1985, 43, 175.
3. Kohn, E. C.; Liotta, L. A. Cancer Res. 1995, 55, 1856.
We acknowledge ®nancial support from the Ministry of
Health and Welfare (HMP-98-D-4-0027). Dr. K. T. No

C. K. Han et al. / Bioorg. Med. Chem. Lett. 10 (2000) 39±43
43
4. Ingber, D.; Fujita, T.; Kishimoto, S.; Sudo, K.; Kanamaru,
T.; Brem, H.; Folkman, J. Nature 1990, 348, 555.
5. Marui, S.; Itoh, F.; Kozai, Y.; Sudo, K.; Kishimoto, S.
Chem. Pharm. Bull. 1992, 40, 96.
6. Sin, N.; Meng, L.; Wang, M. Q. W.; Wen, J. J.; Bornmann,
W. G.; Crews, C. M. Proc. Natl. Acad. Sci. USA 1997, 94, 6099.
7. Grith, E. C.; Su, Z.; Turk, B. E.; Chen, S.; Chang, Y.-H.;
Wu, Z.; Biemann, K.; Liu, J. O. Chem. Biol. 1997, 4, 461.
8. Liu, S.; Widom, J.; Kemp, C. W.; Crews, C. M.; Clardy, J.
Science 1998, 282, 1324.
N.; Sato, S.; Tabata, S.; Ishikawa, K. I.; Kawarabayasi, Y.;
Kotani, H.; Nomura, N. DNA RES 1995, 2, 37.
12. Wu, S.; Gupta, S.; Chatterjee, N.; Hileman, R. E.; Kinzy,
T. G.; Denslow, N. D.; Merrick, W. C.; Chakrabarti, D.;
Osterman, J. C.; Gupta, N. K. J. Biol. Chem. 1993, 268, 10796.
13. Ar®n, S. M.; Kendall, R. L.; Hall, L.; Weaver, L. H.;
Stewart, A. E.; Matthews, B. W.; Bradshow, R. A. Proc. Natl.
Acad. Sci. USA 1995, 92, 7714.
14. Tsunasawa, S.; Izu, Y.; Miyagi, M.; Kato, I. J. Biochem.
1997, 122, 843.
9. Korean Patent Application 98-17636 and PCT/KR99/00229.
10. Roderick, S. L.; Matthews, B. W. Biochemistry 1993, 32,
3907.
15. Sali, A.; Blundell, T. L. Mol. Biol. 1993, 234, 779.
16. All the calculations were carried out with DISCOVER/
INSIGHT(MSI) modeling software and for energy calculation
CVFF force ®eld was used.
11. Nagase, T.; Miyajima, N.; Tanaka, A.; Sazuka, T.; Seki,