Identification of a novel series of tetrahydrodibenzazocines as inhibitors of 17β-hydroxysteroid dehydrogenase type 3

Article abstract of DOI:10.1016/j.bmcl.2005.12.039

A novel series of 17β-hydroxysteroid dehydrogenase type 3 (17β-HSD3) inhibitors has been identified. These inhibitors, based on a dibenzazocine core, exhibited picomolar to low nanomolar inhibition of 17β-HSD3 in cell-free enzymatic as well as in cell-based transcriptional reporter assays.

Full text of DOI:10.1016/j.bmcl.2005.12.039

Bioorganic & Medicinal Chemistry Letters 16 (2006) 1532–1536
Identification of a novel series of tetrahydrodibenzazocines
as inhibitors of 17b-hydroxysteroid dehydrogenase type 3
Brian E. Fink,a,* Ashvinikumar V. Gavai,a John S. Tokarski,c Bindu Goyal,a Raj Misra,a
Hai-Yun Xiao,a S. David Kimball,a Wen-Ching Han,a Derek Norris,a Thomas E. Spires,b
Dan You,b Marco M. Gottardis,b Matthew V. Lorenzib and Gregory D. Vitea
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
Received 21 October 2005; revised 9 December 2005; accepted 9 December 2005
Available online 4 January 2006
Abstract—A novel series of 17b-hydroxysteroid dehydrogenase type 3 (17b-HSD3) inhibitors has been identified. These inhibitors,
based on a dibenzazocine core, exhibited picomolar to low nanomolar inhibition of 17b-HSD3 in cell-free enzymatic as well as in
cell-based transcriptional reporter assays.
Ó 2005 Elsevier Ltd. All rights reserved.
Regulation of androgen biosynthesis or its action on the
androgen receptor (AR) is central to the management of
prostate cancer (PCa).1 The production of androgens is
controlled at two levels: within the central nervous sys-
tem (CNS), but also locally in peripheral target tissues
of hormone action. In these non-gonadal target tissues,
the active androgens, testosterone and dihydrotestoster-
one (DHT), can be synthesized directly through the con-
version of inactive precursors. These inactive precursors,
dihydroepiandrostenedione (DHEA) and androstenedi-
one (ADT), are present at high levels in the circulation
through production by the adrenal gland.2 It has been
estimated that the levels of DHEA and ADT are in great
excess (e.g., 100- to 500-fold) relative to that of testoster-
one, representing a large reservoir of inactive precursor
for potential conversion to active hormone within target
tissues which are the site of hormone action.3 The
17b-hydroxysteroid dehydrogenases (17b-HSDs) are a
family of short-chain alcohol dehydrogenases which
have been shown to mediate the last steps in the conver-
sion of active sex steroids in peripheral target tissues
such as the prostate, ovary, and seminal vesicles.4
Studies analyzing the effects of surgical castration on
serum testosterone levels revealed a >90% decrease in
serum levels of the hormone, but levels of testosterone
within key target tissues such as the prostate were only
decreased by ꢀ50%.5 Such observations indicate that
significant locally active hormone synthesis occurs in
the absence of a gonadal source and this local tissue
synthesis of active hormone could also contribute
to the pathobiology of prostate cancer progression.
17b-HSD3 is a candidate enzyme for mediating this
non-gonadal production of active hormone and has
been characterized for its ability to mediate the conver-
sion of androstenedione to testosterone.6 17b-HSD3 is
expressed at high levels in the testis and seminal vesicles
but has also been shown to be present in prostate tissue,
suggesting its potential involvement in both gonadal and
non-gonadal testosterone biosyntheses.7,8 The role of
17b-HSD3 in testosterone biosynthesis makes this
enzyme an attractive molecular target for the identifica-
tion of small molecule inhibitors for the treatment of
prostate cancer.
Initial lead identification was accomplished through high-
throughput screening of a compound library (ꢀ200,000
compounds) using an enzymatic scintillation proximity
assay (SPA) measuring the conversion of [3H]ADT to
Keywords: 17b-HSD3; Hydroxysteroid dehydrogenase; Androstenedi-
one; Tetrahydrodibenzazocine; Prostate cancer.
*
Corresponding author. E-mail: brian.fink@bms.com
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.12.039

B. E. Fink et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1532–1536
1533
[3H]testosterone. Cellular effects were evaluated using a
secreted alkaline phosphatase reporter assay (SEAP), as
previously described.8 The dibenzothiazocine 19 repre-
sented an attractive starting point for medicinal chemis-
try, based on its potency in the biochemical and
cell-based assays. A brief SAR generated from the
high-throughput screen surrounding this chemotype is
described in Table 1. While both sulfur and carbon were
tolerated at the 12-position (1 and 2), substitution with
oxygen or nitrogen led to a substantially reduced potency
in the in vitro assays.
component was prepared from commercially available
2-aminobenzothiazoles 7. Treatment with KOH in
iPrOH at reflux afforded the dithianes 8 in quantitative
yield. Reduction of 8 with dithiothreitol (DTT) in
DMF and subsequent reaction with the benzyl bromide
6 afforded the thioether 9 in one pot with fair to excel-
lent yields. The ester 9 was reduced to the corresponding
alcohol 10 with lithium borohydride. Treatment of the
alcohol with triphenylphosphine dibromide generated
the corresponding bromide that underwent an intramo-
lecular cyclization to provide the dibenzoazocines 11 in
moderate yield. N-acylation with acetic anhydride in
refluxing toluene yielded the desired dibenzothiazocines
12 in good yield.
Our initial focus, therefore, centered around the
dibenzothiazocine (DBT, 1) and tetrahydrodibenzazo-
cine (THB, 2) series. Compounds in the DBT and
THB series were synthesized by procedures outlined in
Schemes 1 and 2, respectively.
The synthesis of the dibenzazocines followed the route
of Jung and is shown in Scheme 2.10 Readily available
indanones 13 were condensed with arylhydrazines under
acid catalysis to form the corresponding hydrazones.
The intermediate hydrazones underwent indole cycliza-
tion in refluxing toluene to afford 14 in moderate yield.
Treatment of 14 with an oxidant such as mCPBA or
ozone led to ring opening and resulted in keto-amides
The C-ring portion of the dibenzothiazocines 12 was
prepared from appropriately substituted 2-methyl
benzoates, which were subjected to benzylic bromina-
tion by treatment with NBS in dichloromethane to
afford the corresponding bromides 6. The A-ring
15 with
a central 8-membered ring. Concurrent
reduction of the amide and ketone functionalities with
BH3ÆSMe2 afforded a C-12 benzyl alcohol intermediate
which readily eliminated water under acidic conditions
to provide the dihydrodibenzazocines 16. The desired
N-acyl dihydrodibenzazocines (DHB) 17 were prepared
by treatment of 16 with the corresponding acid
chlorides. The olefin could be reduced under standard
Pd/C conditions; however, 5% Rh/C was preferred as
the hydrogenation catalyst since it selectively reduced
the olefin in the presence of other sensitive functional
groups, such as a halogen.
Table 1. Novel 17b-HSD3 inhibitors identified from high-throughput
screena
Me
O
N
A
1
B
C
X
10
b
Compound
X
S
IC50 (lM)
Enzyme
Cellular
1
2
3
4
0.40
0.09
2.97
4.74
0.32
0.22
NA
>10
CH2
O
Structure–activity relationships in the DBT and THB
series are presented in Table 2. Within the DBT series,
incorporation of a halogen at either the 2- or 3-position
(12a–12d) resulted in a 4- to 10-fold improvement in
potency in the biochemical as well as cell-based assays.
NH
a IC50 values are representative of multiple determinations (N = 2–3).
b See Ref. 8 for assay conditions.
CO2Me
R
CO2Me
a
R
Br
6
5
NH2
NH2
CO2Me
R
N
R
c
b
R
R
NH2
S
)
S
S
2
7
8
9
d
Me
H
O
OH
R
N
NH2
S
f
e
N
R
R
R
R
R
S
S
12
11
10
Scheme 1. Reagents and conditions: (a) NBS, (C6H5CO)O2, CH2Cl2, 54–75%; (b) KOH, reflux, 100%; (c) (1) DTT, DMF; (2) 6, DMF, rt, 20–100%;
(d) LiBH4, THF, reflux, 60-80%; (e) PPh3ÆBr2, imidazole, CH2Cl2, 50 °C, 33–85%; (f) Ac2O, DMAP, toluene, reflux, 80–100%.

1534
B. E. Fink et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1532–1536
O
H
H
N
O
N
a
b
R2
R1
R1
R2
R2
O
15
13
14
c
O
R3
O
R3
H
N
e
d
N
N
R2
R1
R2
R2
R1
R1
16
18
17
Scheme 2. Reagents and conditions: (a) (1) arylhydrazine, H2SO4, EtOH, reflux; (2) AcOH, H2SO4, reflux, 65–80%; (b) mCPBA, CH2Cl2, 50–70%;
(c) (1) BH3ÆMe2S; (2) HCl, EtOH, reflux, 60–85%; (d) R3C(O)Cl, CH2Cl2, 90–100%; (e) 5% Rh/C, H2, MeOH, 50–80%.
Table 2. A-ring SARa
activity was observed in the DBT series, the aryl sulfur
atom in these compounds represents a potential meta-
bolic ‘soft spot’. Therefore, the carbocyclic THB series
was pursued for further development of SAR.
Me
O
4
N
3
R
2
X
Y
While halogen substitution of the A-ring in the carbo-
cyclic systems did not improve activity, bromine substi-
tution on the C-ring at either the 8- (18c) or 9-position
(18d) dramatically improved both biochemical and
cellular activity (Table 3). The acetamide functionality
at the 5-position of the THB represented an optimal
substitution for 17b-HSD3-inhibitory potency. Reduc-
tion of the carbonyl (18f) or replacement of the methyl
with an amino group (18e) resulted in a precipitous drop
in enzymatic activity. Even a small increase in size (18g)
was not tolerated. Other substitutions at the 8- or
9-positions were also explored. A primary amine (18h)
or a primary carboxamide (18i) resulted in a loss of
activity; however, activity could be recovered by
incorporating more hydrophobic character. For exam-
ple, benzylamine 18j retained activity relative to 18c.
Similarly, the more lipophilic phenyl amide (18k) and
1
b
Compound
X–Y
R
IC50 (lM)
Enzyme Cellular
12a
12b
12c
12d
12e
12f
12g
12h
12i
S-CH2
S-CH2
S-CH2
S-CH2
S-CH2
S-CH2
S-CH2
S-CH2
S-CH2
S-CH2
S-CH2
S-CH2
2-Cl
3-Cl
0.10
0.03
0.09
0.25
0.01
11.5
0.76
2.81
2.06
4.86
>10
1.0
0.16
0.17
0.29
0.12
0.13
NA
1.3
2-F
2-Br
2,3-diCl
4-Cl
1,2-diCl
3-CF3
2-OMe
2-OH
2-SO2Me
2-Me
2-Cl
NA
>10
NA
NA
0.87
0.50
0.63
0.13
12j
12k
12l
17a
18a
18b
CH@CH
CH2-CH2
CH2-CH2
0.24
0.07
0.05
2-Cl
3-Cl
a IC50 values are representative of multiple determinations (N = 2–3).
b See Ref. 8 for assay conditions.
Table 3. B and C-ring SARa
R2
N
7
8
R1
Chlorine was the optimum substitution, as activity
decreased with both fluorine and bromine. The 2,3-
dichloro analog 12e was the most potent in the enzymat-
ic assay; however, this did not result in a commensurate
increase in potency in the cell-based assay. Halogen
substitution was not tolerated at the 1- or 4-position
(12f–12g), suggesting that the enzyme makes close con-
tacts with these regions of the DBT nucleus. Indeed,
bulkier substitutions on the A-ring (e.g., 12h–12l) were
not well tolerated. The drop in activity of the 2-hydroxyl
(12j) and the 2-methyl (12l) analogs highlights the
electronic requirements for A-ring substitution. Similar
halogen substitution within the DHB (17a) and THB
(18a–18b) series, while tolerated, did not improve
biochemical or cellular activity. Overall, analogs
prepared within the DHB series were several fold less
active than their THB counterparts. Although good
9
10
Compound R1
R2
IC50 (lM)
b
Enzyme Cellular
18c
18d
18e
18f
18g
18h
18i
8-Br
9-Br
C(O)Me
C(O)Me
C(O)NH2
Et
0.009
0.007
0.28
0.032
0.036
0.93
9-Br
9-Br
>1.0
2.2
1.75
9-Br
C(O)c-Prop 0.32
9-NH2
9-C(O)NH2
9-NHBn
C(O)Me
C(O)Me
C(O)Me
0.49
>1.0
0.49
2.98
18j
0.008
0.16
0.004
0.26
18k
18l
9-NHC(O)Ph C(O)Me
C(O)Me
C(O)Me
C(O)NHnBu
8-Ph
0.035
0.00002
0.052
0.004
18m
a IC50 values are representative of multiple determinations (N = 2–3).
b See Ref. 8 for assay conditions.

B. E. Fink et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1532–1536
1535
Table 4. SAR surrounding C-Ring 8-positiona
butyl amide (18l) exhibited an improved potency versus
the primary amide. Taken together, these data suggest
that this region of the molecule extends into a hydro-
phobic pocket. This hypothesis was further substantiat-
ed by the observation that the introduction of a phenyl
group at C-8 (18m) resulted in a significant improvement
in potency in the biochemical as well as the cell-based
assays.
Me
O
R
N
Cl
b
Compound
R
IC50 (lM)
Enzyme
Cellular
18o
22a
22b
22c
H
o-OMe
0.004
0.002
0.008
0.003
0.005
0.003
0.003
0.002
0.14
0.018
0.005
0.053
0.048
0.062
0.015
0.019
0.009
0.19
Further exploration of SAR at the 8- and 9-positions
was facilitated by the use of the Suzuki reaction
for the rapid preparation of analogs as described in
Scheme 3.
m-OMe
p-OMe
22d
22e
o-CN
m-CN
p-CN
22f
Either bromo-substituted DHB (17) or THB (18) could
be directly coupled with an appropriately substituted
arylboronic acid to afford the corresponding coupled
products 21 or 22, respectively. Alternatively, given the
greater variety of aryl halides than boronic acids, 17
and 18 could be converted to the corresponding aryl
boronate ester 19 or 20 and then subjected to Suzuki
coupling to afford 21 or 22. At any point, conversion
of the DHB to the corresponding THB analog could
be accomplished by hydrogenation of the B-ring olefin
catalyzed by 5% Rh/C.
22g
22h
22i
o-NHSO2Me
m-NHSO2Me
m-CH2NH2
30-pyridyl
40-pyridyl
o-CO2Me
m-CO2Me
o-C(O)Me
o-CH2OH
o-CH2NH2
o-C(O)NHMe
o-CO2Me
0.29
0.04
0.18
0.02
22j
22k
22l
0.04
0.04
0.00002
0.0014
0.00002
0.0009
0.239
0.025
0.0007
0.0005
0.002
0.0005
0.002
0.182
0.025
0.001
22m
22n
22o
22p
22q
21a (DHB)
a IC50 values are representative of multiple determinations (N = 2–3).
The SAR developed for the 8-aryl-substituted THB
series is described in Table 4. Introduction of electron-
donating groups around the aryl ring was well tolerated
in both the enzyme and cellular assays. A preference was
observed for ortho-substitution, particularly in the cellu-
lar assay (compare 22a with 22b and 22c). Electron-
withdrawing groups were also well tolerated in the
enzyme assay (22d–22f) although there was an overall
decrease in cellular activity compared to electron-
donating substituents. The same preference for substitu-
tion at the ortho-position continued for sulfonamides
22g and 22h, with 22g being 100-fold more potent.
Introduction of basic functionality as in benzyl amine
22i or basic heterocycles such as pyridines 22j or 22k
generally resulted in reduced activity. Interestingly,
when a carbonyl group was incorporated at the ortho-
position, a dramatic increase in potency was observed.
b See Ref. 8 for assay conditions.
Methyl ester 22l showed picomolar activity, more than
100-fold greater than that of the corresponding
meta-substituted analog (22m). The methyl ketone 22n
was equipotent, as were trifluoromethyl and aldehyde
analogs (data not shown); however, the benzyl alcohol
22o was less active by an order of magnitude, indicating
a possible requirement for a strong hydrogen bond
acceptor at this position. This was further substantiated
by the loss in activity upon incorporation of an amino-
methyl group at this position (22p). Replacement of the
ester with a methyl amide isostere resulted in a 1000-fold
loss in activity, indicating a subtle stereoelectronic
requirement (22q).
In keeping with the increase in activity seen within the
THB series when incorporating the ester group, the
DHB series also showed marked improvement in both
enzymatic and cellular activity. While 21a lost an order
of magnitude relative to 22l, activity remained in the low
or sub-nanomolar range. This unsaturated series
historically showed improved metabolic stability relative
to the THB series. For example, incubation of 22l and
21a in human liver microsomes at 10 lM for 10 min
provided metabolic rates of 0.9 and 0.4 nmol/min/mg,
respectively. The improvement in metabolic stability
bodes well for improved drug exposure in vivo.
O
Me
O
Me
N
N
O
O
b
R1
Br
B
R1
17 CH=CH
18 CH2CH2
19 CH=CH
20 CH2CH2
a
Me
N
c
O
a
d
R2
R1
21 CH=CH
22 CH2CH2
In order to rationalize the favorable impact of the
carbonyl group of 22l on the biochemical potency, a
homology model of human 17b-hydroxysteroid-dehy-
drogenase type 3 (17b-HSD-3) was built from the crystal
structure of 17b-HSD-1 complexed with estradiol and
Scheme 3. Reagents and conditions: (a) 5% Rh/C, H2, MeOH, 80%;
(b) PdCl2(dppf), dppf, KOAc, bis(pinacolato)borane, 80%; (c) arylbo-
ronic acid, Pd(PPh3)4, Na2CO3, 50–100%; (d) arylbromide, Pd(PPh3)4,
Na2CO3, 30–70%.
NADP+11 using Schrodinger’s Prime software.12
A

1536
B. E. Fink et al. / Bioorg. Med. Chem. Lett. 16 (2006) 1532–1536
model will have to await future crystallographic and
kinetic data.
The search for novel non-steroidal inhibitors of
17b-HSD3 has led to the identification of a potent series
of compounds based on the dibenzazocine scaffold.
These inhibitors demonstrated low nanomolar to pico-
molar activity in enzymatic as well as cellular assays.
Compounds targeting 17b-HSD314 may provide useful
tools for examining the role of the enzyme in both endo-
crine-sensitive and -resistant prostate tumor models and
may provide an alternative approach for the disruption
of testosterone biosynthesis in the treatment of prostate
cancer.
Figure 1. Model showing D4-dione (carbon atoms in yellow), 22l
(carbon atoms in green), and NADP+ (carbon atoms in cyan) in the
homology model of 17b-HSD-3 (blue ribbon representation). Potential
hydrogen bonds of D4-dione to S185 and Y198 are shown with dotted
lines.
References and notes
1. (a) Scher, H. I.; Buchanan, G.; Gerald, W.; Butler, L. M.;
Tilley, W. D. Endocr. Relat. Cancer 2004, 11, 459; (b)
Roach, M. J. Urol. 2003, 170, S35.
2. (a) Labrie, F.; Luu-The, V.; Lin, S. X.; Simard, J.; Labrie,
C. Trends Endocrinol. Metab. 2000, 11, 421; (b) Labrie, F.;
Luu-The, V.; Labrie, C.; Simard, J. Front Neuroendocrinol.
2001, 22, 185; (c) Ebeling, P.; Koivisto, V. A. Lancet 1994,
343, 1479.
3. Adams, J. B. Mol. Cell. Endocrinol. 1985, 41, 1.
4. Adamski, J.; Jakob, F. J. Mol. Cell. Endocrinol. 2001, 171, 1.
5. (a) Labrie, F.; Dupont, A.; Belanger, A. In Important
Advances in Oncology 1985, 193; (b) Nishiyama, T.;
Hashimoto, Y.; Takahashi, K. Clin. Cancer Res. 2004,
10, 7121.
model of 4-androstene-3,17-dione (D4-dione) was manu-
ally placed in the 17b-HSD-3 model in the correspond-
ing binding pocket of estradiol. The ligand and
residues of 17b-HSD-3 were allowed to relax during
1000 cycles of conjugate gradient energy minimization
with the CFF force field.13 It was observed in the model
that the carbonyl oxygen at C-17 of D4-dione is within
hydrogen bonding of the conserved residues S185 and
Y198, interactions that most likely assist in the reduc-
tion of the carbonyl. In addition, the C-17 carbon is
˚
within 4 A of the carbon bearing the transferring
hydride of the co-factor NADP+. A low-energy
conformation of 22l could be manually docked into
the homology model in a number of ways guided by
the shape similarity between the inhibitor and D4-dione,
mainly the overlap of the corresponding hydrocarbon
skeletons. Figure 1 shows one possible binding mode
in which the amide methyl group occupies the pocket
normally bound by the C-19 methyl group of D4-dione.
Support for this orientation is derived from the very
narrow SAR at this position, as only a specifically
positioned methyl group was tolerated. The electroneg-
ative Cl atom occupies the same area as does the C-3
carbonyl oxygen of D4-dione. The methyl benzoate
group of 22l is positioned near the nicotinamide ring
suggesting an aromatic ring–ring type of interaction.
In a second orientation (not pictured), rotation of the
methyl benzoate group relative to the core reveals
another possible binding mode which would place the
ester carbonyl within hydrogen bonding distance of
S185 and Y198 and the ester methyl group in the
D4-dione C-18 pocket. Further refinements of this
6. (a) Geissler, W. M.; Davis, D. L.; Wu, L.; Bradshaw, K.
D.; Patel, S.; Mendonca, B. B.; Elliston, K. O.; Wilson, J.
D.; Russell, D. W.; Andersson, S. Nat. Genet. 1994, 7, 34;
(b) Luu-The, V.; Zhang, Y.; Poirier, D.; Labrie, F. J.
Steroid Biochem. Mol. Biol. 1995, 55, 581.
7. (a) Sha, J. A.; Dudley, K.; Rajapaksha, W. R.; O’Shaugh-
nessy, P. J. J. Steroid Biochem. Mol. Biol. 1997, 60, 19; (b)
Koh, E.; Noda, T.; Kanaya, J.; Namiki, M. Prostate 2002,
53, 154.
8. Spires, T. E.; Fink, B. E.; Kick, E.; You, D.; Rizzo, C. A.;
Takenaka, I.; Salvati, M. E.; Vite, G. D.; Weinmann, R.;
Attar, R. M.; Gottardis, M. M.; Lorenzi, M. V. Prostate
2005, 65, 159.
9. Yale, H. L.; Sowinski, F.; Spitzmiller, E. R. J. Heterocycl.
Chem. 1972, 9, 899.
10. Jung, E.; Vossius, V.; Coldewey, G.W. DE Patent 1952019.;
11. Breton, R.; Housset, D.; Mazza, C.; Fontecilla-Camps, J.
Structure 1996, 4, 905.
12. Prime, version 1.2, Schrodinger, LLC New York, NY, 2005.
13. Hagler, A. T.; Ewig, C. S. Comp. Phys. Comm. 1994, 84,
131.
14. Maltais, R.; Luu-The, V.; Poirier, D. J. Med. Chem. 2002,
45, 640.