The synthesis and evaluation of [2.2.1]-bicycloazahydantoins as androgen receptor antagonists

Article abstract of DOI:10.1016/j.bmcl.2004.09.049

A novel series of [2.2.1]-bicycloazahydantoins has been designed and synthesized in an enantiospecific manner. Several of these compounds were found to act as antagonists to the androgen receptor. A novel series of [2.2.1]-azahydantoins has been designed

Full text of DOI:10.1016/j.bmcl.2004.09.049

Bioorganic & Medicinal Chemistry Letters 14 (2004) 6107–6111
The synthesis and evaluation of [2.2.1]-bicycloazahydantoins
as androgen receptor antagonists
Aaron Balog,a,* Mark E. Salvati,a Weifang Shan,a Arvind Mathur,d Leslie W. Leith,d
Donna D. Wei,a Ricardo M. Attar,b Jieping Geng,b Cheryl A. Rizzo,b Chihuei Wang,b
Stanley R. Krystek,c John S. Tokarski,c John T. Hunt,b Marco Gottardisb and
Roberto Weinmannb
aDepartment of Oncology Chemistry, Bristol-Myers Squibb Pharmaceutical Research Institute,
PO Box 4000, Princeton, NJ 08543-4000, USA
bDepartment of Discovery Biology, Bristol-Myers Squibb Pharmaceutical Research Institute,
PO Box 4000, Princeton, NJ 08543-4000, USA
cDepartment of Computer Aided Drug Design, Bristol-Myers Squibb Pharmaceutical Research Institute,
PO Box 4000, Princeton, NJ 08543-4000, USA
dDepartment of Chemical Synthesis, Bristol-Myers Squibb Pharmaceutical Research Institute,
PO Box 4000, Princeton, NJ 08543-4000, USA
Received 16 August 2004; revised 16 September 2004; accepted 18 September 2004
Abstract—A novel series of [2.2.1]-azahydantoins has been designed and synthesized in an enantiospecific manner. The ability of
these compounds to act as antagonists to the androgen receptor was investigated and several were found to have potent activity
in vitro.
Ó 2004 Published by Elsevier Ltd.
Prostate cancer (CaP) is the 2nd leading cause of cancer
related death in men with an estimated 182,000 new
cases diagnosed each year in the United States.1 The
androgen receptor (AR) is a ligand binding transcrip-
tion factor in the nuclear hormone receptor super-family
and is a key molecular target in the growth and progres-
sion of prostate cancer. Binding of androgens, such as
dihydro-testosterone (DHT), to the AR provides the
mitogenic signal for growth of prostate cancer cells.
For the past 50 years, androgen ablation via castration
has been the most effective therapy for the treatment
of advanced prostate cancer in the clinic. Complete
androgen blockade is accomplished by treatment with
an antiandrogen in addition to chemical castration with
an luteinizing hormone releasing hormone (LHRH)
agonist. Casodex (1) (bicalutamide) and Eulexin (2)
(flutamide) are FDA approved AR antagonists that ini-
tially show a 80–90% response rate when used to treat
advanced CaP.2 However, when treatment is continued
for 1–2 years, tumors become androgen independent
and this therapy fails. Approximately 50% of patients
undergoing complete androgen blockade progress to
fatal androgen-independent disease within 5 years of
starting therapy.3 Clearly, there is an unmet medical
need for the treatment of advanced CaP. Thus, we
wanted to find a novel, non-steroidal, small-molecule
antagonist of the AR that is more efficacious than Caso-
dex and shows an increased time required to reach
androgen independent disease.
Our initial screening efforts led to a series of [2.2.1]-bicy-
clicsuccinimides, exemplified by compound 3, that
bound tightly to the AR and functioned as antagonists
to the wild-type (WT) and the mutant (MT) isoforms
of the AR in vitro.4,5 Closely related to the early succin-
imide leads was a series of bicyclic hydantoins repre-
sented by compound 4.6 These compounds showed
very potent binding to the WT AR in addition to an
antagonist profile in vitro.
Keywords: Androgen receptor; Hydantoin.
*
Corresponding author. Tel.: +1 609 252 4632; fax: +1 609 252
6804; e-mail: aaron.balog@bms.com
0960-894X/$ - see front matter Ó 2004 Published by Elsevier Ltd.
doi:10.1016/j.bmcl.2004.09.049

6108
A. Balog et al. / Bioorg. Med. Chem. Lett. 14(2004) 6107–6111
OH
O
O
O
H
N
H
S
CF3
CN
N
CF3
NO2
O
F
Bicalutamide
Flutamide
(1, CasodexTM
)
(2, EulexinTM
)
O
O
N
O
CF3
NO2
H
N
4
N
NO2
O
3
Figure 1. Clinically used antiandrogens.
Closely related to the early succinimide leads was a ser-
ies of bicyclic hydantoins represented by compound 4.6
These series of compounds showed very potent binding
to the AR in addition to an antagonist profile in vitro
against the WT AR, comparing well to Casodex (1)
and Eulexin (2), both clinically used antiandrogens
(Fig. 1).
The promising in vitro profile seen with compounds
such as 3 and 4 led us to design a series of modified
[2.2.1]-bicyclic hydantoins. The X-ray crystal structure
of dihydrotestosterone (DHT) bound to the wild-type
(WT) AR ligand binding domain (LBD) was recently
solved in our laboratories7 and we utilized this informa-
tion for computer-aided drug design. Figure 2a shows
bicalutamide (purple) docked into the crystal structure
of DHT (green) bound to the WT AR.8 Bicalutamide
occupies a much larger space than DHT in the LBD,
thus Helix-12 (H-12) was repositioned to allow bicalut-
amide to dock into the LBD. The docking studies
suggest that the aniline portion of bicalutamide has
similar interactions with the AR as the A-ring of DHT.
Further, the hydrogen bond to the hydroxyl groupin
bicalutamide from residue N705 appears to mimic the
H-bond seen from the C-17 hydroxyl of DHT, and
appears to be essential for high affinity binding to the
AR.10 More importantly, the 4-fluorophenyl ring of
bicalutamide also appears to clash H-12, which would
force it out of the conformation found in the depicted
agonist conformation. It is commonly believed that the
orientation of H-12 is, in part, responsible for agonist/
antagonist function of nuclear hormone receptors.11
Figure 2. (a) Bicalutamide (1) (purple) and DHT (green) docked into a
model of the AR derived from the crystal structure with DHT bound.
Interaction with Helix-12 seen for bicalutamide. (b) Compound 5
docked into a model of the AR derived from the crystal structure with
DHT bound. Shows similar interactions as bicalutamide with Helix-12.
F
O
O
H
N
1
+
4
O
N
N
NO2
O
5
Figure 3. Proposed modified hydantoin 5.
Based on our proposed model of binding of bicaluta-
mide to the AR, we believe that compound 4 forms an
analogous H-bond to residue N705 as bicalutamide
through the tertiary nitrogen of the hydantoin ring sys-
tem. In contrast to bicalutamide, compound 4 lacks any
interactions with H-12, which we proposed to be a crit-
ical interaction for the generation of an antagonist phe-
notype. Based on our model of bicalutamide, we
proposed that placing an appendage off the bicyclic por-
tion of our ring system would result in a similar clash
with H-12, thus generating a new series of analogs with
improved antagonist activity against both the WT and
MT AR. Modeling suggested that positioning an
appendage at the C-5 methylene would give an optimal
overlay with our model of bicalutamide. The most flex-
ible way to accomplish this was to replace a methylene
groupwith a nitrogen in the bicycle thereby providing
a convenient point of attachment for parallel synthesis.
This idea is represented in Figure 3 and in Figure 2b,
where the proposed azahydantoin 5 is docked into the
LBD of WT AR. As is shown, the 4-fluorophenylcarba-
mate appears to invade the space occupied by H-12 in
the wild-type LBD. Based on these studies, we set out
to make a series of modified hydantoins based on com-
pound 5.
We began the synthesis by treatment of commercially
available N-Boc-trans-hydroxy-L-proline methyl ester
(6)12 with tosyl chloride and pyridine to give the corre-

A. Balog et al. / Bioorg. Med. Chem. Lett. 14(2004) 6107–6111
TsO
6109
HO
TsO
H
N
H
O
Boc
CN
Bn
a
d
e
N
Bn
N
R
N
Boc
N
Boc
N
Boc
10
OMe
N
11
R= CO2Me, 7
R= CH2OH, 8
R= CHO, 9
b
c
6
f,g
H
O
H
O
H
N
Boc
Boc
+
Boc
CO2Me
N
N
h
N
N
Ar
N
Ar
N
NH
12
O
O
Ar = 4-CN-3-CF3-phenyl: 14
13
15
17
4-NO2-naphthyl:
4-CN-naphthyl:
16
18
Scheme 1. Reagents and conditions: (a) tosylchloride, pyridine, CH2Cl2, 99%; (b) LiBH4, THF, 22°C, quant.; (c) (COCl)2, DMSO, TEA, CH2Cl2,
˚
À78 to 0°C; (d) (EtO)2P(O)CN, BnNH2, 4A MS, CH2Cl2, 95%, two steps; (e) DBU, (CH2Cl)2, 80°C, 99%, 2:1 endo/exo; (f) 3N NaOH then HCl/
MeOH; (g) H2, 10% Pd/C, EtOAc, 78%, two steps; (h) arylisocyanate,16 4A MS, (CH2Cl)2, 22°C then DBU 60°C, 2h, 58%, 2:1 endo/exo (>98% ee).
˚
sponding tosylate 7, which was taken on in crude form
(Scheme 1). The ester was reduced with lithium borohy-
dride to give the primary alcohol 8 in excellent yield. The
aldehyde 9, obtained by a Swern oxidation of 8,13 was
then treated with cyanodiethylphosphonate and benzyl-
amine to give the aminonitrile cyclization precursor 10
in 95% yield as a 1:1 mixture of diastereomers.14
O
O
H
N
R
N
Boc
Boc
N
N
a or b
N
CN
CF3
N
CN
CF3
O
O
2:1 13/14
R = H; 13/14 (6:1)
R = Me; 19 (>20:1)
O
N
HO
O
N
Ar
Precedent for the desired transannular cyclization was
found in JordisÕ synthesis of diazabicyclo-[2.2.1]hep-
tanes.15 Accordingly, the amine 10 was treated with
DBU in dichloroethane at 60°C in hopes of displacing
the trans-oriented tosylate. Gratifyingly, the cyclization
occurred, giving the desired bicycle 11 in 99% yield as
a separable 2:1 mixture of diastereomers. Basic hydrol-
ysis of the mixture of diastereomeric nitriles, followed
by treatment with acidic methanol, gave the correspond-
ing methyl ester intermediate. Subsequent removal of
the N-benzyl groupby catalytic hydrogenation gave
the secondary amine 12 in 78% yield as a 2:1 mixture
of diastereomers.
N
N
Boc
Boc
OEt
H
O
21, Ar = 4-CN-3-CF3-phenyl
22, Ar = 4-NO2-naphthyl
20
Scheme 2. Reagents and conditions: (a) LDA, THF, À78°C, then aq
satd NH4Cl, 98%, 6:1 endo/exo; (b) LDA, THF, À78°C, then MeI,
92%, >20:1 endo/exo.
of the mixture with LDA, followed by quenching with
a proton source (Scheme 2). In this way, the 2:1 ratio
is transformed to 6:1 ratio favoring the endo-isomer in
96% yield. When the enolate is instead quenched with
methyl iodide, the methylated analog 19 is produced in
a >20:1 ratio by alkylation from the less hindered b-face.
Treatment of amine 12 with 3-trifluoromethyl-4-
cyanophenylisocyanate16 in the presence of 4A molecu-
˚
lar sieves at 23°C for 4h gave the corresponding urea
intermediate, which was only isolated in the initial
experiments. Once urea formation was complete, DBU
was added followed by heating to 60°C to promote cyc-
lization of the urea onto the methyl ester. After acidic
workup, the diastereomeric azahydantoins 13 (endo)
and 14 (exo) were produced in a separable 2:1 ratio,
respectively (95% yield). When the reaction was simply
concentrated in vacuo avoiding the acidic workup, and
loaded directly onto silica for purification, only the
exo-isomer was formed in 87% yield. This result suggests
that the exo-isomer 14 is the thermodynamically more
stable, and predominates when the reaction is concen-
trated without neutralizing the organic base. The endo/
exo ratio did not depend on which diastereomer of the
cyclization precursor 12 or which arylisocyanate was
used in the reaction.
The opposite antipode of this series (21 and 22) was
made by starting from trans-hydroxy-L-proline utilizing
the same synthetic sequence. A small number of exam-
ples were made to assess, which enantiomer had better
potency in vitro.
To evaluate the activity of this new series, we inves-
tigated the ability of compounds to bind to (Ki) and
functionally antagonize (IC50) the WT AR found in
the MDA-453 cell line as well as antagonize the MT
6
AR (T877A) found in the LNCapcell line.
The
Boc-protected analogs incorporating the 4-cyano-3-tri-
fluoromethyl aniline, 4-nitronaphthylamine, and 4-
cyanonaphthylamine (Table 1) were tested first in hopes
of narrowing the series to one diastereomer. The endo-
isomers 13 and 15 demonstrated increased binding (Ki)
to the WT AR compared to the corresponding exo-iso-
mers. In the case of compounds 17 and 18, there was
minimal difference in terms of binding. Overall, the Ki
When the endo-isomer was desired, the 2:1 endo/exo
ratio of diastereomers could be improved by enolization

6110
A. Balog et al. / Bioorg. Med. Chem. Lett. 14(2004) 6107–6111
Table 1. Biological data for initial azahydantoin analogs
O
O
H
N
H
N
R
Kia(lM)
MDA-453
b
LNCaP
b
b,c,d or e
X
Compound no.
N
N
N
Ar
N
Ar
25
IC50 (lM)
IC50 (lM)
O
13
14
21
0.040
0.189
>5.00
0.774
0.644
>5.00
4.38
>5.00
>5.00
O
23, R=Boc
24, R=H
a
X = SO2R1
Ar= 4-CN-3-CF3-phenyl
4-NO2-naphthyl
COR2
15
16
22
0.062
1.03
0.237
3.32
1.33
0.252
4.43
5.86
2.39
CONHR3
CO2R4
4-CN-naphthyl
17
18
0.248
0.197
2.54
>5.00
4.97
3.51
Scheme 3. Reagents and conditions:17 (a) TFA, CH2Cl2, 22°C, 99%;
(b) R1SO2Cl, DIEA, CH2Cl2, 22°C; (c) R2OCOCl, DIEA, 4-DMAP,
CH2Cl2, 22°C; (d) R3COCl, DIEA, 4-DMAP, CH2Cl2, 22°C; (e) R4
NCO, DIEA, CH2Cl2, 22°C.
19
4
1.40
0.78
1.30
0.012
0.17
10.4
0.001
0.005
0.065
>5.00
>5.00
0.40
3
results of our modeling exercise, we chose to pursue
additional analogs in this series from the antipode
depicted in compound 5. The methylated analog 19
showed a dropin binding and functional antagonist
activity to the MT AR as compared to 13. This result
is consistent with the methylation of hydantoins such
as 4 (data not shown).6
1
a Binding determined through direct displacement of ligand with [3H]-
DHT in the MDA-453 cell line (Ki).5
b Functional antagonist activity determined through a transiently
transfected reporter system utilizing a secreted alkaline phosphatase
(SEAP) reporter construct and a PSA AR promoter domain.5
values for these compounds compared very well to bica-
lutamide (Ki = 65nM). In terms of functional activity, it
was less clear which diastereomer was more potent over-
all. The exo-isomers 14 and 16 were slightly more potent
than the endo-isomers but neither compared well to
bicalutamide. We chose to focus our efforts on the
endo-isomers because the initial examples demonstrated
tighter binding to the AR and past results with the
hydantoin series suggested that tight binding was essen-
tial for potent functional activity.
These initial results were promising enough for us to
embark on the preparation of a small solution-phase
library of azahydantoins, focusing on the endo-isomer.
We prepared a small library of 96 compounds where
the aromatic domain and the N-substituent were varied
(Scheme 3).17 The library was made from the amine de-
picted by the general structure 24, formed by removal of
the N-Boc groupin 23 by treatment with TFA. The
amine was then converted into a series of sulfonamides,
amides, carbamates, and ureas by conventional chemis-
try as shown in Scheme 3. When the amine 24 was alk-
ylated rather than acylated, the resulting products were
found to be unstable to storage and purification.
Compounds 21 and 22 were tested to determine which
antipode would have superior binding and functional
activity. Compound 21 is completely devoid of any
AR activity but, compound 22 does have potent binding
to the AR and reasonable functional activity. Modeling
was performed in an attempt to explain the significant
difference in activity seen between compounds 21 and
22. Docking of each compound into the AR model
(not shown) demonstrated that compound 22 was better
able to form the critical H-bond with N705 as compared
to compound 21. Based on initial SAR as well as the
Table 2 contains the biological data for the library com-
pounds with the highest affinity for WT AR. Com-
pounds not presented either had Ki > 1.5lM or had
MDA-453 IC50 > 1.0lM. The urea and sulfonamide
series had no active members by these criteria and are
not presented. The most active carbamate, 26, contains
the same two aromatic domains as bicalutamide, and
Table 2. Results from solution library synthesis with compound 25
Compound no.
X
Ara
Kib (lM)
MDA-453
c
IC50 (lM)
LNCaP
c
IC50 (lM)
26
27
28
A
B
C
0.018
0.011
0.041
0.200
0.349
0.588
3.92
30.0
1.87
F
O
O
O
29
30
A
B
0.124
0.012
0.300
>5.000
2.60
2.54
O
31
32
33
A
B
C
0.693
0.073
0.106
17.9
0.060
0.580
33.7
4.75
6.80
Bu
a A = 4-Cyano-3-trifluoromethylphenyl; B = 4-nitronaphthyl; C = 4-cyanonaphthyl.
b Binding determined through direct displacement of ligand with [3H]-DHT in the MDA-453 cell line (Ki).5
c Functional antagonist activity determined through a transiently transfected reporter system utilizing a secreted alkaline phosphatase (SEAP)
reporter construct and a PSA AR promoter domain.5

A. Balog et al. / Bioorg. Med. Chem. Lett. 14(2004) 6107–6111
6111
Roberge, J.; Corte, J. R.; Spergel, S. H.; Rampulla, R. A.;
Misra, R.; Xiao, H.-Y. World Patent Application WO
2003062241, 2003.
performed very well in binding and functional antago-
nism to WT AR. When the aniline portion was changed
to the 4-nitronaphthyl or 4-cyanonaphthyl, 27, and 28,
respectively, the binding improved but the functional
activity decreased.
5. New analogs were assayed for effect on AR function in
MDA-453, a cell line expressing WT AR and LNCaP, a
cell line expressing MT AR.
Transactivation assays were done in whole cells and IC50Õs
were calculated. Both cell lines were obtained from
American Type Culture Collection (Rockville, MD).
Binding constants (Ki) were determined in whole cells
measuring displacement by tritiated DHT in the MDA-
453 cell line. See Ref. 6 below for experimental details on
assays.
The benzamide analogs (29–33) also showed tight bind-
ing and moderate antagonist activity against the AR.
Both examples contain electronically neutral aromatic
portions, found to be essential for potent antagonist
activity in the amide series. The benzamides 29 and 30
were very different in terms of binding and function,
suggesting a different conformation of the LBD for each
ligand. Compound 30 bound very tightly but had no
antagonist function under our assay conditions.
Whereas compound 29 demonstrated reduced binding
but reasonable antagonist activity against MDA-453.
The butyl-substituted benzamides 31–33 also showed
promising activity. Of the library compounds, 32
showed the most potent antagonist activity against the
WT cell line. However, against the MT cell line, it did
not compare well to bicalutamide. Interestingly, alkyl
amides and carbamates were not as potent as their aryl
analogs, suggesting a common interaction in the LBD,
which will be investigated further. Overall, compounds
26 and 32 were the most promising, and related com-
pounds are under investigation.
6. Salvati, M.; Balog, A.; Shan, W.; Geise, S. U.S. Patent
885,798, 2003.
7. Sack, J. S.; Kish, K. F.; Wang, C.; Attar, R. M.; Kiefer, S.
E.; An, Y.; Wu, G. Y.; Scheffler, J. E.; Salvati, M. E.;
Krystek, S. R., Jr.; Weinmann, R.; Einspahr, H. M. Proc.
Natl. Acad. Sci. 2001, 98, 4904.
8. Molecular modeling was performed using ICM software
(Molsoft LLC, San Diego, CA).9 The docking of the
ligands into WT AR LBD was carried out using the ICM
docking procedure which is a two step process. Initial
docking of ligands were carried out using grid potential
representation of the receptor and flexible ligand. Five grid
potentials describe the shape, hydrophobicity, electrostat-
ics, and hydrogen-bonding potential of the receptor. The
conformations from the grid were then optimized with a
full atom representation of the receptor and flexible ligand,
by an ICM stochastic global optimization algorithm as
implemented in version 2.7 the MOLSOFT ICM program9.
9. (a) Molsoft ICM 2.7. Program Manual, 1998 (Molsoft,
San Diego, CA), (b) Totrov, M.; Abagyan, R. Proteins
Suppl. 1997, 1, 215–220.
10. (a) Poujol, N.; Wurtz, J.-M.; Tahiri, B.; Lumbroso, S.;
Nicolas, J.-C.; Moras, D.; Sultan, C. J. Biol. Chem. 2000,
275, 24022; (b) Marhefka, C. A.; Moore, B. M.; Bishop, T.
C.; Kirkovsky, L.; Mukherjee, A.; Dalton, J. T.; Miller, D.
D. J. Med. Chem. 2001, 44, 1729.
11. Wurtz, J.; Bourguet, W.; Renaud, J.; Vivat, V.; Chambon,
P.; Moras, D.; Gronemeyer, H. Nat. Struct. Funct. 1991, 3, 87.
12. The Aldrich Chemical Company, Inc.
In summary, we have developed a versatile, enantiospe-
cific synthesis of a novel series of androgen receptor
antagonists. These modified [2.2.1]-bicyclic hydantoins
were proposed based on the biological activity of the
parental ring-system and molecular modeling utilizing
the crystal structure of DHT bound to the WT AR.
Several of the new compounds demonstrated potent
binding to the WT AR but fell short of the desired func-
tional antagonist activity in the cell lines expressing the
WT and especially the MT AR. However, several of
these compounds compared favorably to bicalutamide.
We are currently investigating modifications of this
series based on the most active compounds and will
report these results in due course.
13. Mancuso, A. J.; Huang, S.-L.; Swern, D. J. Org. Chem.
1978, 43, 2480.
14. Kim, Y. J.; Kido, M.; Bando, M.; Kitahara, T. Tetrahe-
dron 1997, 53, 7501.
15. (a) Jordis, U.; Sauter, F.; Siddiqi, S. M.; Kuenburg, B.;
Bhattacharaya, K. Synthesis 1990, 925; (b) Rosen, T.;
Chu, D. T. W.; Lico, I. M.; Fernandes, P. B.; Marxh, K.;
Shen, L.; Cepa, V. G.; Pernet, A. G. J. Med. Chem. 1988,
31, 1598.
16. Isocyanate was generated by treatment of the correspond-
ing aniline with phosgene and NaHCO3 in CH2Cl2. The
isocyanate was isolated and used in crude form directly.
17. The solution phase library was run in 96-well format with
amine 24 (1.0equiv) and an acylating reagent (3.0equiv) in
CH2Cl2 utilizing polymer-bound HunigÕs base (5.0equiv)
for each series. Workupincluded tris-(2-aminoethyl)amine
polystyrene (5.0equiv) to scavenge excess acylating rea-
gents. Eighty percent of the products generated did not
require purification and isolated yields ranged from 35%
to 99%.
References and notes
1. Jemal, A.; Thomas, H.; Murray, T.; Thun, M. CA Cancer
J. Clin. 2002, 52, 23.
2. (a) Blackledge, G. Cancer 1993, 1, 3830; (b) Maucher, A.;
von Angerer, E. J. Cancer Res. Clin. Onc. 1993, 119, 669;
(c) Sufrin, G.; Coffey, D. S. Invest. Urol. 1976, 13, 429; (d)
Narayana, A. S.; Loening, S. A.; Culp, D. A. Brit. J. Urol.
1981, 53, 152.
3. (a) Huggins, C.; Stevens, R.; Hodges, C. V. Arch. Surg.
1947, 43, 209; (b) Westin, P.; Stattin, P.; Damber, J.-E.;
Bergh, A. Am. J. Pathol. 1995, 146, 1368.
4. Salvati, M. E.; Balog, A.; Pickering, D. A.; Giese, S.;
Fura, A.; Li, W.; Patel, R.; Hanson, R. L.; Mitt, T.;