Investigation of novel fumagillin analogues as angiogenesis inhibitors

Article abstract of DOI:10.1016/j.bmcl.2003.10.008

Modification of fumagillin was conducted to develop MetAP-2 inhibitors with desirable pharmacological properties. Replacement of the C4 side chain by benzyloxime preserves the inhibitory activity against MetAP-2 enzyme. Fumagillin analogues containing the

Full text of DOI:10.1016/j.bmcl.2003.10.008

Bioorganic & MedicinalChemistry Letters 14 (2004) 91–94
Investigation of novel fumagillin analogues as angiogenesis inhibitors
Hyung-Jung Pyun,* Maria Fardis, James Tario, Cheng Y. Yang,
Judy Ruckman,y Dwight Henninger,{ Haolun Jin and Choung U. Kim
Gilead Sciences, 333 Lakeside Dr., Foster City, CA 94404, USA
Received 5 June 2003; revised 3 October 2003; accepted 6 October 2003
Abstract—Modification of fumagillin was conducted to develop MetAP-2 inhibitors with desirable pharmacological properties.
Replacement of the C4 side chain by benzyloxime preserves the inhibitory activity against MetAP-2 enzyme. Fumagillin analogues
containing the C4 benzyloxime moiety were found to be very sensitive to the nature of the C6 substituent on the inhibition activity
of HUVEC proliferation. This lack of correlation between MetAP-2 and HUVEC activities might be due to the cellular metabolism
of the compounds by epoxide hydrolase, which is present in the cell. Compound (E)-3d, containing ethylpiperazinyl carbamate at
C6 position, exhibited antiangiogenic effects similar to TNP-470 on matrigel plug assay and rat corneal micropocket assay.
# 2003 Elsevier Ltd. All rights reserved.
Inhibitors of angiogenesis hold promise as therapeutic
agents for treatment of a variety of diseases.1,2 A great
dealof effort has been devoted to the deveolpment of
antiangiogenic treatments particularly for cancer3 since
Folkman reported that tumor growth and metastasis
depend on blood vessel growth.4 Antiangiogenic agents
are expected to be less susceptable to mechanism of
resistance, even after protracted treatment, since the
actively dividing cancer cells are not the direct target of
the therapeutic agent. Furthermore, because angiogen-
esis is down regulated in most adult processes except
reproduction, an antiangiogenic agent may have mini-
residue of MetAP-2 by opening the spiro-epoxide.9 Sig-
nificant correlation has been reported between inhibi-
tion of MetAP-2 and the inhibition of bovine aortic
endothelial cell (BAEC) proliferation.10
In the clinical trials of TNP-470, however, its poor
pharmacokinetic behavior and dose-limiting toxicity
remain obstacles to its use as an anticancer agent.11 The
instability and toxicity of TNP-470 are likely due, at
least in part, to the presence of two epoxides and a
chloroacetyl groups. We envisioned that improvement
of TNP-470 would be attained by replacing one or more
of these functionalgroups. Here, we report ( E)-3d, an
analogue of fumagillin without the sidechain epoxide
and chloroacetylcarbamate group of TNP-470, which
showed comparable in vitro and in vivo activity to
TNP-470 (Fig. 1).
5
malside effects.
Fumagillin (1),6 a naturalproduct isoal ted from Asper-
gillus fumigatus, was found to strongly inhibit endothe-
lial cell proliferation.7 TNP-470 (2), which resulted from
the subsequent search for improved fumagillin analo-
gues and was found to have greater potency and lower
toxicity than fumagillin,7 is one of the first inhibitors of
angiogenesis to reach clinical trials. Fumagillin and
TNP-470 are proposed to inhibit angiogenesis by selec-
tive inhibition of methionine aminopeptidase type 2
(MetAP-2)8 through covalently binding to the His231
TNP-470 has three functionalgroups that are either
chemically labile or metabolically unstable in its struc-
ture, namely the two epoxides (spiro-epoxide and the
one on C4 side chain) and the chloroacetylcarbamate
moiety at C6. Reactivity of the spiro-epoxide functional
group seems crucial for the binding of fumagillin ana-
logues. Therefore, we first explored C4 sidechain modi-
fications, since crystallographic information indicated
that there was significant space within the pocket for
further optimization.9 Modification of the C4 chain was
possible by a semisynthetic approach starting from
fumagillol (4a). Reduction of the side-chain epoxide to
the diene (E)-6b was performed in a four-step sequence,
* Corresponding author. Tel.: +1-650-522-5437; fax: +1-650-522-
5899; e-mail: ppyun@gilead.com
y
Present address: CBR International, Corp., 2905 Wilderness Pl., Ste.
202, Boulder, CO 80301, USA.
Present address: OSI Pharmaceuticals, 2860 Wilderness Place,
Boulder, CO 80301, USA.
{
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.10.008

92
H.-J. Pyun et al. / Bioorg. Med. Chem. Lett. 14 (2004) 91–94
Table 1. Activities of selected fumagillin analogues in enzymatic and
HUVEC-based assays
Compd
MetAP-2a IC50 (nM)
HUVECb EC50 (nM)
1
2
4b
0.63c
0.43ꢄ0.13d
0.51c
0.82ꢄ0.25
0.52ꢄ0.03e
3.0ꢄ1.1
0.65ꢄ0.10
ND
490ꢄ100
450ꢄ180
ND
4c
0.39c
Figure 1.
(E)-6b
(E)-6c
(E)-3b
(Z)-3b
(E)-3c
(Z)-3c
(E)-3d
8.9ꢄ2.2
1.7ꢄ0.5
1.2ꢄ0.32
25ꢄ2
with some improvement in yields over the procedure of
Liu et al.12,13 as shown in Scheme 1. Opening of the
spiro-epoxide before the reduction of the side-chain
epoxide to olefin14 is primarily responsible for the
improvement. The diene (E)-6b was converted by ozo-
nolysis to the methylketone 7b, a common precursor for
the C4 modifications.
1.2ꢄ0.17f
57ꢄ22
88ꢄ55
ND
0.90ꢄ0.18
0.8ꢄ0.08
a Assays were duplicated, unless otherwise indicated.
b EC50’s were measured using the sulforhodamine B (SRB) dye binding
assay.15 HUVEC proliferation assays were triplicated, unless other-
wise indicated.
c n=1.
According to the data reported previously by Liu et al.,
compounds with a rotatable single bond between C10
and C20 carbons on the C4 side chain show reduced
inhibitory activity against MetAP-2 and BAEC pro-
liferation, compared to analogues bearing more rigid
epoxide and diene side chains.13 We hypothesized that
planarity of the C10 and C20 atoms of the C4 side chain
was essentialfor optimalactivity. Compounds with the
diene moiety at the C4 chain (6b and 6c) exhibited sig-
nificant but somewhat weaker enzyme activities toward
MetAP-2 (Table 1). Among severalmoieties investi-
gated for fumagillin C4 side-chain replacement, we
found benzyloxime ether to be the best bioisostere.16
The IC50 values of compound (E)-3b and (E)-3c toward
d n=27.
e n=10.
f n=5.
MetAP-2 were approximately 1 nM, and comparable to
epoxides 4b and 4c, respectively. Comparing the two
benzyloxime isomers, the each E-isomer was more
active than the Z-isomer in the enzyme assay and was
consistently the major product in the reaction.
With the benzyloxime moiety estabilshed at the C4
position, optimization of the C6 substituent was per-
formed. X-ray crystal structures of fumagillin-MetAP-2
complex showed that the C6 side chain is placed outside
of the MetAP-2 binding pocket.9 From this X-ray
structure we expected that the C6 sidechain would not
have a significant effect on enzyme inhibition. Optimi-
zation of the C6 substituent of fumagillin has been
studied extensively.7,17ꢀ19 Previous reports clearly
showed though that C6 substituents significantly affect
the activity of these analogues.7,17ꢀ19 Most benzyloxime
bearing analogues exhibited excellent activities (ꢁ1 nM)
in the MetAP-2 assay, whereas inhibition of the human
umbilical vein endothelial cell (HUVEC) proliferation
varied significantly depending on the C6 substituent.
Among the compounds tested, the ethylpiperazinyl car-
bamate, (E)-3d, was found to be as active as fumagillin,
TNP-470, and AGM-1883 (4c) in the HUVEC-based
proliferation assay.20
Although a reasonable correlation between MetAP-2
inhibition and the inhibition of BAEC proliferation has
been reported,10 we discovered an apparent lack of
correlation between MetAP-2 activity and HUVEC
proliferation. Table 2 shows that most compounds of
this series exhibited similar stability in cell culture med-
ium (67–75% intact after 20 h). In cell lysate, however,
the stability of fumagillin analogues varied con-
siderably. For example, less than 3% of compound
(E)-3c and (E)-6c, compared to ꢂ76% of 4c, remained
after 20 h in HUVEC lysate. Interestingly, compound
(E)-3d was reasonably stable in the cell lysate.
Scheme 1. Reagents and conditions: (a) LiCl, AcOH, THF, rt, 96%;
(b) Ac2O, DMAP, CH2Cl2, 0 ꢃC, 86%; (c) WCl6, n-BuLi, THF,
ꢀ78 ꢃC, 92%; (d) t-BuOK, THF, 0 ꢃC, 77%; (e) K2CO3, MeOH, rt,
100%; (f) ClCOOPh, DMAP, CH2Cl2, 0 ꢃC, 100%; (g) NH4OH,
EtOH, 0 ꢃC, 85%; (h) O3, CH2Cl2, MeOH, ꢀ78 ꢃC, then Me2S, 74%;
(i) H2NOBn, p-TsOH, 4 A sieves, THF, 50 ꢃC, 40% (E-isomer), 17%
(Z-isomer); (j) HN(CH2CH2)2NEt, CH2Cl2, rt, 78%.
We speculated that the decomposition of fumagillin ana-
logues might be due to epoxide hydrolase (EH) present in

H.-J. Pyun et al. / Bioorg. Med. Chem. Lett. 14 (2004) 91–94
93
Table 2. Stability of fumagillin analogues in HUVEC culture med-
ium and in HUVEC lysatea
enzymatic activities of this class of compounds are all
quite good.
Compd
% of the compound remainingb
Cell culture medium Cell lysate
4 h 20 h
Compound (E)-3d was found to dose-dependently (all
data not included) inhibit angiogenesis induced by basic
fibroblast growth factor (bFGF) in a mouse matrigel
plug assay (Table 4), as measured by hemoglobin con-
tent of the subcutaneous plug. Compound (E)-3d was
administered at 2.5 mg/kg/day for 7 days by an osmotic
pump, and exhibited an anti-angiogenic effect similar to
TNP-470 (76% vs 71% inhibition). In the bFGF-
induced rat cornealmicropocket assay, compound
(E)-3d also showed a similar dose-dependent (all data
not included) anti-angiogenic effect (86%) to TNP-470
(90%) as shown in Table 5. In both in vivo models,
(E)-3d caused weight changes similar to those observed
with TNP-470. Thus, compound (E)-3d was as effective
as TNP-470 in inhibition of MetAP-2, HUVEC pro-
liferation, and angiogenesis.
0 h
0 h
4 h
20 h
4d
100
100
100
100
95
76
90
96
75
68
69
67
100
100
100
100
91
36
20
97
76
1
3
68
(E)-6c
(E)-3c
(E)-3d
a Error of this measurements is estimated to be 10–15%.
b Each compound (1 mM) was incubated in HUVEC lysate at 37 ꢃC for
up to 20 h.
HUVEC since HUVEC has been reported to express
21
microsomalepoxide hydroalse gene products.
Micro-
somal epoxide hydrolase is known to hydrolyze the
spiro-epoxide of metabolite IV (4c) during the metabo-
lism of TNP-470.22 When compound 4c was treated
with EH,23 it was readily hydrolyzed as expected (Table
3). On the other hand, hydrolysis of 4c was inhibited
completely when cyclohexene oxide, a specific EH inhi-
bitor,24 was added to the incubation mixture. This sug-
gested that the hydrolysis of 4c was mainly due to the
enzyme, EH. Interestingly, compound (E)-6c seemed
much less stable toward EH than 4c since it was hydro-
lyzed even in the presence of cyclohexene oxide. Thus, it
seems that the C4 side-chain epoxide impacts on the
stability of the spiro-epoxide in fumagillin analogues.
Compound (E)-3c appears more labile than 4c but more
stable than (E)-6c based on its rate of hydrolysis in the
presence of the inhibitor.
Modification of C4 side chain of fumagillin scaffold
was accomplished by
a
semisynthetic approach.
Fumagillol was converted to the diene (E)-6b in 58%
yield by a four-step sequence, improving the reported
procedure of Liu et al.13,16,20 The E-benzyloxime
moiety, lacking the potentially labile epoxide group,
was found to be the best substitute for the C4 side
chain of fumagillin. Fumagillin analogues containing
Table 4. Antiangiogenic activity in mouse matrigelassay a
Compd
% Inhibition
% Body weight change
2
(E)-3d
71
76
ꢀ11
ꢀ13
The introduction of an ethylpiperazinyl carbamate
group at the C6 position clearly increased the stability
of the spiro-epoxide of (E)-3d against EH activity.
When compound (E)-3d was treated with EH in the
presence of cyclohexene epoxide, its hydrolysis was
almost completely inhibited as was observed in the
case of 4c. This suggested that introduction of an
ethylpiperazinyl carbamate group at the C6 position
makes the compound significantly less susceptible to
EH. The stability of fumagillin analogues against EH
activity correlates with the HUVEC activity while the
a Dose, 2.5 mg/kg/d for 7 days, s.c. n=8 mice/group.
Table 5. Antiangiogenic activity in rat cornealmicropocket assay a
Compound
% Inhibition
% Body weight changeb
2
(E)-3d
90
86
+7
+10
a Dose, 2.5 mg/kg/d for 5 days, s.c.
b % Body weight change of the controlgroup was +28% regardelss
with or without bFGF. n=8 eyes/group.
Table 3. Stability of fumagillin analogues in the presence of epoxide hydrolase (EH)a
% of the compound remaining or producedb
c
ControlWith EH
With EH and an inhibitor
3 min
0.5 h
2 h
3 min
0.5 h
2 h
3 min
0.5 h
2 h
4c
Substrate
Hydrolysate
Substrate
Hydrolysate
Substrate
Hydrolysate
Substrate
Hydrolysate
100
ND
100
ND
100
ND
100
ND
103
ND
92
ND
103
ND
103
ND
106
ND
90
ND
100
ND
102
ND
84
22
18
77
10
93
96
6
ND
103
ND
79
2
95
ND
100
ND
100
1
100
ND
100
98
ND
108
ND
106
ND
110
ND
109
ND
75
21
95
15
101
ND
107
ND
5
87
60
60
106
ND
(E)-6c
(E)-3c
(E)-3d
ND
96
a Error of this measurements is estimated to be 10–15%.
b Each compound (2 mM) was incubated with or without epoxide hydrolase at 37 ꢃC. The protein concentration was 1 mg/mL in the finalincubation
mixture.
c Cyclohexene oxide (50 mM) was used as an inhibitor of epoxide hydrolase.

94
H.-J. Pyun et al. / Bioorg. Med. Chem. Lett. 14 (2004) 91–94
E-benzyloxime moiety at C4 sidechain generally showed
good MetAP-2 inhibition activity whereas HUVEC-
based activities varied significantly depending on the C6
substituent. The poor correlation between enzyme
activity and HUVEC activity might be due to the
metabolic stability of the analogues in the HUVEC
assay. The C6 substituents of fumagillin analogues were
found to influence the susceptibility of the analogues to
epoxide hydrolase activity, which might deactivate
fumagillin analogues in HUVEC. Among the com-
pounds tested, compound (E)-3d exhibited the best
activity in the HUVEC-based assay, and is roughly
equipotent to TNP-470. Compound (E)-3d also showed
activities similar to TNP-470 in the matrigel assay and
the rat cornealmicropocket assay.
W. G.; Crews, C. M. Proc. Natl. Acad. Sci. U.S.A. 1997,
94, 6099.
9. Liu, S.; Widom, J.; Kemp, C. W.; Crews, C. M.; Clardy,
J. Science 1998, 282, 1324.
10. Griffith, E. C.; Su, Z.; Turk, B. E.; Chen, S.; Chang, Y.-H.;
Wu, Z.; Biemann, K.; Liu, J. O. Chem. Biol. 1997, 4, 461.
11. P. Bhargava, et al., In Angiogenesis in Health and Disease;
Rubanyi, G. M., Ed.; MarcelDekker: New York, 2000;
Chapter 26.
12. Griffith, E. C.; Su, Z.; Niwayama, S.; Ramsay, C. A.;
Chang, Y.-H.; Liu, J. O. Proc. Natl. Acad. Sci. U.S.A.
1998, 95, 15183.
13. (a) Dai, X. B.; Griffith, E. C.; Liu, J. O. Abstracts of
Papers, 219th NationalMeeting of American Chemical
Society, San Francisco, CA, March 26–30, 2000; Amer-
ican ChemicalSociety: Washington, DC, 2000;
MEDI061. (b) Dai, X. B. Personalcommunication.
14. Sharpless, K. B.; Umbreit, M. A.; Nieh, M. T.; Flood,
T. C. J. Am. Chem. Soc. 1972, 94, 6538.
15. Skehan, P.; Storeng, R.; Scudiero, D.; Monks, A.; McMa-
hon, J.; Vistica, D.; Warren, J. T.; Bokesch, H.; Kenney,
S.; Boyd, M. R. J. Natl. Cancer Inst. 1990, 82, 1107.
16. Tario, J. D.; Pyun, H.-J.; Fardis, M.; Jin, H.; Kim, C. U.
Manuscript in preparation.
17. (a) Marui, S.; Itoh, F.; Kozai, Y.; Sudo, K.; Kishimoto, S.
Chem. Pharm. Bull. 1992, 40, 96. (b) Folkman, M. J.;
Ingber, D.; Takeshi, F. US Patent 6017954, 2000.
18. (a) Marui, S.; Kishimoto, S. Chem. Pharm. Bull. 1992, 40,
575. (b) Kishimoto, S.; Marui, S.; Takeshi, F. Eur. Patent
0386667, 1990.
Acknowledgements
We are deeply indebted to Dr. X. Chen for his
modeling and helpful discussions. We also thank Mr.
Y. Lin, Mr. L. Green and Dr. B. Hicke for their
help on providing MetAP-2 and HUVEC based assay
data, as well as Ms. W. Gillette for in vivo assay
data. Dr. X. Dai is also acknowledged for his helpful
suggestions.
19. Han, C. K.; Ahn, S. K.; Choi, N. S.; Hong, R. K.; Moon,
S. K.; Chun, H. K.; Lee, S. J.; Kim, J. W.; Hong, C. I.;
Kim, D.; Yoon, J. H.; No, K. T. Bioorg. Med. Chem.
Lett. 2000, 10, 39.
20. Fardis, M.; Pyun, H-J.; Tario, J. D.; Jin, H.; Kim, C. U.;
Ruckman, J.; Lin, Y.; Green, L.; Hicke, B. Bioorg. Med.
Chem., in press.
References and notes
1. Folkman, J.; Shing, Y. J. Biol. Chem. 1992, 267, 10931.
2. Rubanyi, G. M., Ed. Angiogenesis in Health and Disease;
MarcelDekker: New York, 2000.
3. (a) Zhang, H.-T.; Harris, A. L. Exp. Opin. Invest. Drugs
1998, 7, 1629. (b) Cherrington, J. M.; Strawn, L. M.;
Shawver, L. K. In Adv. Cancer Res.; Academic, 2000; p 1.
4. Folkman, J. N. Engl. J. Med. 1971, 285, 1182.
5. Folkman, J. Breast Cancer Res. Treat. 1995, 36, 109.
6. Hanson, T. E. Antibiot. Chemother. (Washington D.C.)
1951, 1, 54.
21. Farin, F. M.; Pohlman, T. H.; Omiecinski, C. J. Toxicol.
Appl. Pharmacol. 1994, 124, 1.
22. (a) Placidi, L.; Cretton-Scott, E.; de Sousa, G.; Rahmani,
R.; Placidi, M.; Sommadossi, J.-P. Cancer Res. 1995, 55,
3036. (b) Figg, W. D.; Pluda, J. M.; Lush, R. M.; Saville,
M. W.; Wyvill, K.; Reed, E.; Yarchoan, R. Pharmaco-
therapy 1997, 17, 91.
23. Epoxide hydrolase from Rhodococcus rhodochrous was
purchased from Sigma-Aldrich Company.
24. Guenthner, T. M. Biochem. Pharmacol. 1986, 35, 839.
7. Ingber, D.; Fujita, T.; Kishmoto, S.; Sudo, K.; Kana-
maru, T.; Brem, H.; Folkman, J. Nature (London) 1990,
348, 555.
8. Sin, N.; Meng, L.; Wang, M. Q.; Wen, J. J.; Bornmann,