Lead evaluation of tetrahydroquinolines as nonsteroidal selective androgen receptor modulators for the treatment of osteoporosis

Article abstract of DOI:10.1002/cmdc.201300348

Tetrahydroquinoline (THQ) was deemed to be a suitable scaffold for our nonsteroidal selective androgen receptor modulator (SARM) concept. We adapted the strategy of switching the antagonist function of cyano-group-containing THQ (CN-THQ) to the agonist function and optimized CN-THQ as an orally available drug candidate with suitable pharmacological and ADME profiles. Based on binding mode analyses and synthetic accessibility, we designed and synthesized a compound that possesses a para-substituted aromatic ring attached through an amide linker. The long-tail THQ derivative 6-acetamido-N-(2-(8-cyano-3a,4,5,9b- tetrahydro-3H-cyclopenta[c]quinolin-4-yl)-2-methylpropyl)nicotinamide (1 d), which bears a para-acetamide-substituted aromatic group, showed an appropriate in vitro biological profile, as expected. We considered that the large conformational change at Trp741 of the androgen receptor (AR) and the hydrogen bond between 1 d and helix 12 of the AR could maintain the structure of the AR in its agonist form; indeed, 1 d displays strong AR agonistic activity. Furthermore, 1 d showed an appropriate in vivo profile for use as an orally available SARM, displaying clear tissue selectivity, with a separation between its desirable osteoanabolic effect on femoral bone mineral density and its undesirable virilizing effects on the uterus and clitoral gland in a female osteoporosis model. Long-tail THQ as a SARM: We describe the lead evaluation process of tetrahydroquinolines (THQs) as nonsteroidal selective androgen receptor modulators (SARMs). The strategy of switching from AR antagonist to AR agonist was successfully applied to CN-THQ, and we identified a long-tail THQ as a novel SARM. This new THQ showed a certain degree of tissue selectivity, with a separation between its desirable osteoanabolic effect and its undesirable virilizing effects in a female osteoporosis model. Copyright

Full text of DOI:10.1002/cmdc.201300348

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201300348
Lead Evaluation of Tetrahydroquinolines as Nonsteroidal
Selective Androgen Receptor Modulators for the
Treatment of Osteoporosis
Naoya Nagata,* Kazuyuki Furuya, Nao Oguro, Daisuke Nishiyama, Kentaro Kawai,
Noriko Yamamoto, Yuki Ohyabu, Masahiro Satsukawa, and Motonori Miyakawa[a]
Tetrahydroquinoline (THQ) was deemed to be a suitable scaf-
fold for our nonsteroidal selective androgen receptor modula-
tor (SARM) concept. We adapted the strategy of switching the
antagonist function of cyano-group-containing THQ (CN-THQ)
to the agonist function and optimized CN-THQ as an orally
available drug candidate with suitable pharmacological and
ADME profiles. Based on binding mode analyses and synthetic
accessibility, we designed and synthesized a compound that
possesses a para-substituted aromatic ring attached through
an amide linker. The long-tail THQ derivative 6-acetamido-N-(2-
(8-cyano-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]quinolin-4-yl)-2-
methylpropyl)nicotinamide (1d), which bears a para-acet-
amide-substituted aromatic group, showed an appropriate in
vitro biological profile, as expected. We considered that the
large conformational change at Trp741 of the androgen recep-
tor (AR) and the hydrogen bond between 1d and helix 12 of
the AR could maintain the structure of the AR in its agonist
form; indeed, 1d displays strong AR agonistic activity. Further-
more, 1d showed an appropriate in vivo profile for use as an
orally available SARM, displaying clear tissue selectivity, with
a separation between its desirable osteoanabolic effect on
femoral bone mineral density and its undesirable virilizing ef-
fects on the uterus and clitoral gland in a female osteoporosis
model.
Introduction
Osteoporosis is a common, age-related bone disease that re-
sults from an imbalance between bone formation and bone re-
sorption processes, leading to decreased bone mass and in-
creased risk of fracture.[1] Calcium, vitamin D3 supplementation,
and bone resorption inhibitors such as bisphosphonates, estro-
gens, and selective estrogen receptor modulators (SERMs) are
commonly used in the prevention and treatment of osteoporo-
sis. Bisphosphonates and estrogens increase bone mineral den-
sity (BMD) by decreasing bone resorption. SERMs also increase
BMD to the same extent as estrogen without affecting female
sexual organs, such as the breasts and uterus. In contrast, an-
drogens and parathyroid hormones (PTHs) are known to have
positive effects on BMD by increasing bone formation.[2] Al-
though one PTH (teriparatide, for example) is known as
a bone-formation enhancer, its use requires subcutaneous in-
jection for up to two years.[3]
drogens. The major biological effects of androgens are classi-
fied into two categories: androgenic effects and anabolic ef-
fects. The androgenic effects include the development and
maintenance of the male reproductive organs, secretory
glands, and so on. The anabolic effects include the growth of
muscles, bones, etc.
Steroidal AR ligands have been used therapeutically, but
their use is limited by serious side effects toward reproductive
organs (androgenic effects) and liver (hepatotoxicity). There-
fore, nonsteroidal selective androgen receptor modulators
(SARMs) are required to separate the anabolic effects from the
androgenic effects.[4] Various SARMs have been identified in
preclinical studies, but the mechanism underlying their tissue
selectivity remains unclear.[5–7] Narayanan et al. proposed four
mechanisms that may underlie the tissue selectivity of SARMs:
1) the role of 5a-reductase, 2) the tissue-specific expression of
co-regulators, 3) differences in the complexes that are formed
by the AR in anabolic and androgenic tissues, and 4) the
tissue-specific roles of intracellular signaling cascades.[7]
The endogenous androgen, testosterone (TES), and its active
metabolite, dihydrotestosterone (DHT), play important roles in
the development and maintenance of the male phenotype, in-
cluding the male reproductive system, muscles, bone, hair,
larynx, and skin. Androgens bind to the androgen receptor
(AR) and induce a conformational change at helix 12 (H12),
converting the AR to its agonist form. The activated AR translo-
cates to the nucleus and mediates the biological effects of an-
Co-crystallized ligand–protein structures have been used ef-
fectively to both understand the functional mechanisms of
members of the nuclear receptor superfamily and for struc-
ture-based drug design (SBDD).[8] Various research groups have
solved the co-crystal structures of their SARMs that possess ag-
onistic activity toward wild-type AR and have been able to use
their binding modes to design various therapeutic agents.[9–14]
For example, Nique et al. used the co-crystal structure of diaryl-
hydantoin in complex with AR for docking studies and man-
aged to improve the oral bioavailability of diarylhydantoin by
[a] Dr. N. Nagata, Dr. K. Furuya, N. Oguro, D. Nishiyama, Dr. K. Kawai,
N. Yamamoto, Y. Ohyabu, Dr. M. Satsukawa, Dr. M. Miyakawa
Central Research Laboratories, Kaken Pharmaceutical Co. Ltd.
14 Shinomiya, Minamikawara-cho, Yamashina, Kyoto 607-8042 (Japan)
E-mail: nagata_naoya@kaken.co.jp
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 197

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
introducing a hydroxymethyl group at the 4-position of the hy-
dantoin scaffold.[13] Bohl et al. analyzed multiple co-crystal
structures of diarylpropionamides with the AR and gained an
understanding of the binding modes of diarylpropionamide
derivatives.[9,10] GTx-007, a diarylpropionamide derivative, is
known to be a SARM and is currently undergoing clinical inves-
tigation (Figure 1).[15,16] The binding mode of GTx-007 is distinct
GTx-007, BMS-564929, and GLPG0492 (Figure 1).[9,12,20] We hy-
pothesized that it might be possible to convert CN-THQ from
an antagonist into an agonist while avoiding the potential risk
of mutagenicity by the para-nitro aniline portion of NO2-
THQ[21] and maintaining its tissue-selective anabolic activities.
In our dynamic structural analyses, we found that the combi-
nation of the cyano and terminal hydroxy groups of CN-THQ
results in the antagonistic activity of the AR.[22] Therefore, we
postulated that replacement of the hydroxy group on CN-THQ
with another functional group could be an essential step re-
quired to alter the activity of CN-THQ. Herein we describe the
lead evaluation process of our THQ derivatives as SARMs.
Results and Discussion
Docking
For the SBDD of the new THQ derivatives, we first compared
the key interactions between three different ligand–AR com-
plexes (Figure 2a–c),[10,22,23] the shapes and volumes of the
ligand binding pockets of two co-crystallized structures
(Figure 3), and the 2D substructures of GTx-007 with CN-THQ.
In the DHT–AR complex (PDB ID: 1T7T),[22] DHT is stabilized
by electrostatic interactions with the nearest neighboring polar
atom of the hydrophilic residues Asn705, Gln711, Arg752, and
Thr877 and by van der Waals (vdW) interactions with the sur-
rounding hydrophobic residues Leu701, Leu704, Met742,
Met745, Phe764, and Leu873 (Figure 2a). The 3-carbonyl group
is stabilized through Coulombic interactions with the amide
Figure 1. Nonsteroidal SARMs.
from that of other SARMs that display agonistic activity. Small
AR agonists do not interact with Trp741 in helix 4 (H4), where-
as the B ring of GTx-007 strongly interacts with the aromatic
side chain of Trp741 in H4 and is
stabilized through a p–p interac-
tion.[10] The stabilization of GTx-
007 induces the folding of H12
into its agonist form, which is
similar to that of known ago-
nists.
We recently reported nitro-
containing tetrahydroquinolines
(NO2-THQs) as novel lead com-
pounds of nonsteroidal SARMs
and that they possess tissue-se-
lective anabolic activities in both
in vitro and in vivo studies.[17–19]
These results validated THQ as
a scaffold suitable for the devel-
opment of our desired SARMs. In
our lead-finding process, we
demonstrated that THQ com-
pounds change their behavior
following subtle structural modi-
fications; NO2-THQ acted as an
agonist, while cyano-containing
THQ (CN-THQ) acted as an an-
tagonist. The strategy of switch-
ing an antagonist into an ago-
Figure 2. Ligands and their representative interaction residues in the AR. a) DHT–AR (PDB ID: 1T7T),
b) GTx-007–AR (PDB ID: 3B68), c) CN-THQ–AR (assumed binding mode), and d) long-tail THQ (1a)–AR
nist has often succeeded for
nonsteroidal SARMs, including (assumed binding mode).
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 198

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
side chain of Gln711 and the guanidinium side chain of Arg752
in the AR. The 17-hydroxy group forms two hydrogen bonds
with the amide side chain of Asn705 and the hydroxy side
chain of Thr877 in the AR. Interestingly, in the case of the
GTx-007–AR complex (PDB ID: 3B68),[10] the GTx-007 is stabi-
lized by electrostatic interactions with the four hydrophilic resi-
dues Leu704, Asn705, Gln711, and Arg752, as well as a water
molecule, and by vdW interactions with the surrounding hy-
drophobic residues Leu704, Leu707, Met742, Met745, Trp741,
Phe764, Leu873, Met895, Ile898, Ile899, and Val903 (Figure 2b).
The 4’-nitro group is stabilized by a hydrogen bond with the
amide side chain of Gln711 and by Coulombic interactions
with the guanidinium side chain of Arg752 in the AR, which is
similar to the 3-carbonyl group of DHT. The amide and hydroxy
groups form two hydrogen bonds with the main chain amide
carbonyl of Leu704 and with the amide side chain of Asn705
in the AR, respectively. A water molecule forms hydrogen
bonds with the para-acetamide group of the B ring of GTx-
007, the side chain of His874, main chain of Gln738, and main
chain of Met742 in the AR. The hydrogen bonding network of
GTx-007–water–AR is considered to mediate the binding of
GTx-007 to AR and to maintain the AR structure in its agonist
form. Thus, the hydrogen bond that surrounds the B ring of
GTx-007 is considered to play an important role in ligand bind-
ing and the agonistic activity of GTX-007. The A ring displays
an edge-to-face aromatic interaction with the aromatic side
chain of Phe764, and the B ring displays a p–p interaction with
the aromatic side chain of Trp741. In contrast, in the assumed
binding mode of the CN-THQ–AR complex,[23] CN-THQ is also
stabilized by electrostatic interactions with the five hydrophilic
residues Leu704, Asn705, Gln711, Arg752, and Thr877, and by
vdW interactions with the surrounding hydrophobic residues
Leu704, Leu707, Met742, Met745, and Leu873 (Figure 2c). The
6-cyano group is stabilized by a hydrogen bond with the
amide side chain of Gln711 and by Coulombic interactions
with the guanidinium side chain of Arg752 in the AR. The
1-NH and the terminal hydroxy groups form three hydrogen
bonds with the main chain amide carbonyl group of Leu704,
with the amide side chain of Asn705, and with the hydroxy
side chain of Thr877 in the AR. Thus, DHT, CN-THQ, and the ni-
trophenylamide moiety of GTx-007 display similar interactions
with the AR and were determined to overlap within the bind-
ing pocket using protein-based superimposition (Figure 2).
Next, the shapes and volumes of the ligand binding pockets
of DHT–AR and GTx-007–AR were estimated using SiteMap.[24]
The pocket of the DHT–AR complex is 167 ꢁ3 (Figure 3a), and
that of the GTx-007–AR complex is 199 ꢁ3 (Figure 3b). Such
a large difference between the DHT–AR complex and the
GTx-007–AR complex is caused by the dramatic conformational
change in the side chain of Trp741 that is located in the ligand
binding pocket of the GTx-007–AR complex. Thus, this large
conformational change in AR induced by GTx-007 maintains
the structure of the AR in its agonist form, and the B ring of
GTx-007 could be placed in this newly identified binding
pocket through a p–p interaction with the side chain of
Trp741 (Figure 3b).
Figure 3. Co-crystallized structures of a) DHT–AR and b) GTx-007–AR, over-
lapped with the assumed binding mode of CN-THQ (DHT: yellow, GTx-007:
blue, CN-THQ: purple, AR and ligand binding site surface in AR: green).
In addition to the results of the previously outlined analyses,
the chemical substructure of GTx-007 was compared with the
scaffold of CN-THQ. If the nitrophenylamide moiety of GTx-007
was converted into a fused ring, similar to the red dashed line
in Figure 2b, then the designed compound would be consid-
ered to be very similar to the scaffold of CN-THQ. Thus, the
THQ ring is a bioisostere of the nitrophenylamide moiety of
GTx-007, and we considered these moieties to be interchange-
able. Conversely, a bulky substituent such as a para-substituted
aromatic group could be attached in place of the hydroxy
group of CN-THQ in maintaining AR binding affinity and ago-
nistic activity in a manner similar to that of GTx-007. Docking
studies were conducted to find the most suitable linker for
connecting the THQ ring with bulky groups, like a para-substi-
tuted aromatic ring: amide, amine, ether, and sulfone were se-
lected as linkers, and para-acetamide-substituted benzyl and
phenyl groups were selected as the aromatic portions.
The docking studies were performed using Glide SP mode,
and the protein in the GTx-007–AR complex (PDB ID: 3B68)[10]
was used as a template, because of the advantage of the
aforementioned. Another advantage in using this complex is
that Dalton et al. had successfully obtained GTx-007 as a SARM
from structural modification of the existing nonsteroidal AR an-
tagonist bicalutamide.[25] Concretely, the antagonist function
was switched to the agonist function by changes in the cyano
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 199

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
the AR by electrostatic interactions with the four hy-
drophilic residues Leu704, Arg752, Thr877, and
Gln902, and by vdW interactions with the surround-
ing hydrophobic residues Leu704, Leu707, Met742,
Met745, Trp741, Phe764, Leu873, Met895, Ile898,
Ile899, and Val903. The 6-cyano group is stabilized by
Coulombic interactions with the guanidinium side
chain of Arg752 in the AR. The 1-NH group and two
of the carbonyl groups within the amide groups of
1a form three hydrogen bonds with the main chain
amide carbonyl of Leu704, with the hydroxy side
chain of Thr877, and with the amide side chain of
Gln902 in the AR, respectively. The hydrogen bond
between the para-acetamide group on the Ar portion
of THQ and the amide side chain of Gln902 on H12 is
believed to maintain the AR in its agonist form in
a manner similar to the GTx-007–AR complex. Thus,
this hydrogen bond is one of the key interactions
that we used to design our long-tail THQ derivative.
The THQ scaffold consists of an edge-to-face aromat-
ic interaction with the aromatic side chain of Phe764,
Table 1. Docking scores, DGbind values, and average number of hydrogen bonds.[a]
[b]
Compd
X
Docking
Score
DGbind
Hydrogen bond[c]
Linker
(Linker)
[kcalmolꢀ1
]
THQ
Ar
1a
2
3
4
5
6
7
8
NHCO
CONH
O
OCH2
NH
NHCH2
SO2
SO2CH2
ꢀ8.72
NA
ꢀ131.4
NA
0.7
NA
0.8
0.4
0.9
0.5
NA
NA
0.4
NA
0.0
0.0
0.0
0.2
NA
NA
1.0
NA
0.6
0.0
0.1
0.1
NA
NA
ꢀ8.98
ꢀ10.55
ꢀ9.98
ꢀ10.98
NA
ꢀ127.7
ꢀ138.1
ꢀ129.5
ꢀ129.5
NA
NA
NA
[a] NA: no appropriated docking poses were obtained. [b] Values are the means of the
calculated results of the MD trajectories using Prime MM-GBSA (1–12 ns). [c] Hydrogen
bonds are the means of the calculated results of the MD trajectories using the
g_hbond module (1–12 ns).
group and sulfonyl linker of bicalutamide to the nitro group
and ether linker. The results of the docking studies demon-
strate that the compounds containing NHCO, ether, and amine
linkers are able to dock, whereas compounds containing
CONH and SO2 linkers are unable to dock (Table 1). To select
the best linker among the four, we performed a molecular dy-
namics (MD) simulation (12 ns) using GROMACS 4.5.1,[26] and
then calculated the binding free energy (DGbind) and average
number of hydrogen bonds. DGbind was estimated using Prime
MM-GBSA, which is based on the molecular mechanics/gener-
alized Born surface area (MM/GBSA) method included in the
Schrçdinger Suite,[24] and the number of hydrogen bonds was
estimated with the g_hbond module in GROMACS.
The THQ structure was divided into three portions (THQ,
linker, and aromatic (Ar)), and the number of hydrogen bonds
contained in each portion were then counted (Table 1). The
ether linker (OCH2) compound 4 was determined to be the
best one based on the results of the DGbind calculations. How-
Figure 4. The lowest-energy binding mode of long-tail THQ 1a (long-tail
THQ: purple, AR and ligand binding site surface in AR: green; hydrogen
bond: dashed red line).
ever, we considered the amide linker (NHCO) compound 1a to
be the best compound in terms of hydrogen bonding and syn-
thetic accessibility, as well as the DGbind calculations. From the
hydrogen bond analyses, the direct hydrogen bond between
the para-acetamide group of the Ar portion of THQ and the
side chain of Gln902 on H12 were considered to be indispensa-
ble in maintaining the AR structure in its agonist form
(Figure 4). Additionally, as far as synthetic accessibility is con-
cerned, the aromatic carboxylic acid was determined to be
a useful component because amide linkages can easily be
formed by various types of condensation reactions.
and the Ar ring of 1a consists of a p–p interaction with the ar-
omatic side chain of Trp741.
Chemistry
Tricyclic THQ derivatives were obtained using Grieco three-
component condensation (Scheme 1).[27] The intermediate 13
can be synthesized in a single step by treating 4-cyanoaniline
10 with cyclopentadiene 11 and N-Boc-3-amino-2,2-dimethyl-
propionaldehyde 12 in the presence of an equimolar amount
of trifluoroacetic acid (TFA) in acetonitrile. The stereochemistry
of 13 was assigned on the basis of coupling constants and
The interaction of our designed long-tail THQ derivative 1a
with the ligand binding pocket residues of the AR is similar to
the interaction between GTx-007 and AR, as we had predicted
(Figure 2d). Based on the results of the MD simulation, 1a was
determined to be stabilized in the ligand binding pocket of
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 200

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Scheme 1. Synthesis of tetrahydroquinoline derivatives 1a–f. Reagents and conditions: a) Boc2O/THF, RT, 3 h (100%); b) WSCD·HCl, H3PO3/DMSO, RT, 30 min
(95%); c) TFA/CH3CN, RT, 1 h (65%); d) 4n HCl/dioxane, 608C, 1 h; e) Ar-COOH, WSCD·HCl, HOBt, NMM/DMF, RT, 1–17 h (42–94%, two steps).
In vitro profiles
NOESY experiments. Saturation of proton H3a in 13 gave an
NOE enhancement on H9b of 3.9% and H4 of 2.9%, confirming
the six-membered quinoline ring and cyclopentene ring are cis
fused and the protons H3a and H4 are cis (Scheme 1). Alterna-
tively, the aldehyde 12 was synthesized from the correspond-
ing amino alcohol 9 using Boc protection followed by Moffatt
oxidation. After the THQ scaffold was constructed, the Boc
group of 13 was deprotected, and the appropriate aromatic
carboxylic acids, para-substituted benzoic acids, and nicotinic
acids were coupled to THQ scaffolds via amide linkages. The
para-substituted groups were selected based on several as-
pects: the available potential for hydrogen bond formation,
the electron density of the aromatic ring, the physicochemical
properties, and so on. Acetylamino, formylamino, and trifluoro-
methoxy groups were selected as substituents for benzoic
acid, and acetylamino, ethoxy, and chloride groups were se-
lected for nicotinic acid. The desired compounds 1a–f were
obtained in moderate to high yields. Various calculated octa-
nol/water partition coefficients (clogP; 2.83–5.47) and polar
surface areas (PSA; 75.9–116.8 ꢁ2) were obtained based on the
characteristics of the aromatic rings and their substituted func-
tional groups (Table 2). The two physicochemical properties
were estimated using QikProp.[24] By NOE experiment, we iden-
tified that the stereochemistry of 1d is the same as 13. Fur-
thermore, 1a–c, 1e, and 1 f have the same stereochemistry as
To validate the results of our docking studies, we evaluated
the binding affinities of 1a–f and their agonistic activities on
the AR (Table 2). The binding affinities of 1a–f for rat AR
ranged from 9–81 nm and are in the same range as that of our
previous results.[17] Although their binding affinities were 5–40-
fold weaker than that of DHT, such binding affinities are suffi-
cient to reveal their anabolic activities in an animal model.
Compound 1a displayed particularly strong binding affinity, as
predicted. We hypothesized that the predicted binding mode
of 1a would be similar to the 3D structure of GTx-007. Despite
the fact that some of the compounds do not contain a poten-
tial hydrogen bonding group in their Ar portion, the com-
pounds with similar molecular shapes also display high bind-
ing affinities. These results indicate that molecular shape is
also important for high binding affinities.
The agonistic activities of the compounds in human osteo-
blastic (TE-85) cells were evaluated using two indices: The half-
maximum effective concentration (EC50) and efficacy. The effi-
cacy is the maximal transactivation activity of the compounds
relative to that of 1000 nm DHT (100%). Compounds 1a–f all
exhibited agonistic activity (EC50) <200 nm. These results are
consistent with our predicted binding mode. In particular, the
EC50 values of 1a and 1d are 5 and 1.3 nm, respectively, and
their activities are 4–10-fold stronger than that of DHT. With re-
spect to efficacy, 1a, 1b, and 1d displayed higher efficacies
than those of the other com-
1
13, as determined by the H and 13C NMR data.
pounds. However, 1b was ex-
cluded as a candidate due to its
Table 2. In vitro biological activities of the AR and calculated physicochemical properties of the THQ deriva-
tives.
high EC50 value.
The calculated physicochemi-
cal properties (clogP and PSA)
were relevant with aqueous sol-
ubility and free compound ratio
of plasma protein binding (PPB).
In the results of in vitro ADME
analyses, the above two parame-
ters of 1d were marginally supe-
rior to that of 1a (Table 3). In
any other common parameters,
such as microsome intrinsic
Compd
X
R
clogP[a]
PSA [ꢁ2][a]
IC50 [nm][b]
Agonistic Activity[c]
EC50 [nm] Efficacy [%]
DHT
1a
1b
1c
1d
1e
1 f
2.0
9.4
33
81
38
19
5
100
88
CH
CH
CH
N
N
N
NHAc
NHCHO
OCF3
NHAc
OEt
3.75
3.01
5.47
2.83
4.49
4.22
105.2
116.8
75.9
114.6
89.2
147
10
153
59
1.3
91
38
28
14
16
54
68
Cl
80.5
[a] Calculated using QikProp. [b] Binding affinities were determined by competitive binding assays with rat AR.
[c] Agonistic activities were determined based on the transcriptional activities in human TE-85 cells.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 201

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Table 3. In vitro ADME properties of the THQ derivatives.
[c]
[d]
Compd
X
R
Solubility[a]
PPB [%free][b]
Human/Rat
CLint [mLminꢀ1 mgꢀ1
Human/Rat
]
Caco-2 Papp [10ꢀ6 cmsecꢀ1
A!B/B!A
]
Efflux Ratio[e]
[mgmLꢀ1
]
1a
1b
1c
1d
1e
1 f
CH
CH
CH
N
N
N
NHAc
NHCHO
OCF3
NHAc
OEt
2.1
2.2
0.0
4.1
0.1
0.0
1.3/2.5
4.0/51.1
0.1/0.3
2.1/2.1
2.8/5.2
0.9/4.3
0.05/0.01
0.00/0.00
0.05/0.01
0.06/0.01
0.09/0.01
0.07/0.01
6/7
29/32
2/0
18/21
27/13
15/7
1.2
1.1
0.0
1.2
0.5
0.5
Cl
[a] Measured by the weight dissolved in phosphate buffer (pH 6.8). [b] Assessed by equilibrium dialysis in plasma from the appropriate species at 378C;
free and bound concentrations were quantified by LC–MS/MS. [c] Compounds were incubated at 0.5 mgmLꢀ1 with suspended rat or human hepatocytes at
2 mgmLꢀ1, and intrinsic clearance was calculated based on the rate of disappearance of the parent compound. [d] Measured permeability (A!B/B!A)
through Caco-2 cells. [e] Papp(BꢀA)/Papp(AꢀB) in Caco-2 cells.
clearances (CLint) and Caco-2 permeability (Papp), 1a and 1d did
not clearly differ. Based on these results of such in vitro stud-
ies, we selected compound 1d as the best compound among
1a–f.
androgen-responsive tissue, whereas the CG is one of the se-
cretory glands and is an androgen-responsive tissue. Thus, in-
creases in the uterus weight could induce endometriosis or
uterine fibroids, whereas increases in the CG weight could
induce acne or oily skin.
During the next stage of our lead evaluation process, we
evaluated the receptor selectivity of 1d for five steroid hor-
mone receptors. The binding affinities (IC50) for each receptor
were determined by calculating the competitive binding activi-
ty with each endogenous radiolabeled ligand. The endogenous
ligands for each receptor were as follows: progesterone for the
progesterone receptor (PR), 17b-estradiol for the estrogen re-
ceptor (ER), dexamethasone for the glucocorticoid receptor
(GR), and aldosterone for the mineralocorticoid receptor (MR).
The IC50 of 1d for each receptor was 25.8 nm for the AR,
667 nm for the PR, >50000 nm for the GR and MR, and
>500000 nm for the ER. The IC50 value of 1d for the AR is
>25-fold higher than any other steroid hormone receptor and
is higher than the value we had determined in our previous
studies.[17] Because compound 1d displays sufficient receptor
selectivity, we chose 1d as a candidate for our in vivo oral ad-
ministration study.
In the OVX-vehicle group, femoral BMD and uterus weight
were significantly decreased, whereas the CG weight was influ-
enced far less. In the 1d oral administration group, these pa-
rameters increased in a dose-dependent manner. Femoral BMD
increased gradually and reached 99% of the sham group at
a dose of 30 mgkgꢀ1 for 1d. Uterus weight also increased
gradually. However, uterus weight reached 42% of the sham
group at a dose of 30 mgkgꢀ1 for 1d. The CG weight was in-
creased to the sham level (99%) at a dose of 3 mgkgꢀ1 and ex-
ceeded the sham level at doses of 10 and 30 mgkgꢀ1 of 1d
(124 and 147%, respectively). In the DHT administration group,
the femoral BMD was increased to the sham level (97%), and
the uterus weight was increased to 70% of the sham group.
Surprisingly, the CG weight was remarkably increased to 289%
of the sham group.
Next, the effects of 1d on these three parameters were com-
pared with DHT. Even at a dose of 3 mgkgꢀ1 of 1d, femoral
BMD reached the same level as DHT. In contrast, at doses of 3,
10, and 30 mgkgꢀ1 of 1d, the uterus weights were 39, 54, and
60% of DHT, and the CG weights were 34, 43, and 51% of
DHT. Based on these results, 1d was considered to be a SARM
with osteoanabolic effects equal to that of DHT, yet with fewer
virilizing effects. However, the uterine and CG weights were
significantly increased relative to the OVX-vehicle. These effects
should be decreased prior to clinical use, and further optimiza-
tion is required to develop an ideal SARM.
In vivo pharmacology
The tissue selectivity of the SARMs was examined in an ovari-
ectomized (OVX) rat model, because osteoporosis patients are
mainly post-menopausal elderly women.[28] Thus, our ideal
SARM should have an osteoanabolic effect and no androgenic
effects in women. OVX rats were maintained without treat-
ment for four weeks to permit bone loss. The treatment group
was treated with 1d by oral administration once daily for eight
weeks. The sham-operated control group (Sham) and the OVX-
operated control group (OVX-vehicle) were treated with the
vehicle solution, while the positive control group was treated
with DHT (10 mgkgꢀ1) by subcutaneous injection once daily
for eight weeks.
Conclusions
The lead evaluation process of our THQ derivatives as SARMs
was described. THQ is a validated scaffold that possesses os-
teoanabolic tissue selectivity, so we adapted the strategy of
switching from the antagonist activity of THQ to the agonist
function, and optimized THQ for use as an orally available
SARM. From the viewpoints of the computational study and
synthetic accessibility, we designed and synthesized several
CN-THQ compounds that possess para-substituted aromatic
Desired and/or undesired activities were evaluated using
three indices. The BMD of the femur was used as an index of
desirable osteoanabolic activity, while the weights of the
uterus and clitoral gland (CG) were used as indices of undesira-
ble side effects (Figure 5).[29] The uterus, which is composed of
an endometrium and a myometrium, is both an estrogen- and
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 202

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 5. Comparison of the desired osteoanabolic effect on the femoral BMD and the undesired side effects on the uterine and CG of 1d and DHT in OVX
rats (##p<0.01 vs. Sham, **p<0.01 vs. OVX-vehicle); BW: body weight.
MD simulations: MD simulations were performed using the GRO-
MACS 4.5.1 package[26] with AMBER ff03 force field parameters[30]
and periodic boundary conditions. Protein–ligand complexes were
solvated in 91ꢂ84ꢂ91 ꢁ3 boxes at a density of 1 gcmꢀ3, with the
TIP3P water model.[31] An appropriate number of counterions (Clꢀ)
were added in the box to neutralize the systems. The temperature
control was set using the Berendsen thermostat at 300 K, and pres-
sure coupling was implemented using the Berendsen barostat with
a constant pressure of 1 atm.[32] Non-bonded interactions were cut
off at 10 ꢁ with updates at every five steps, and the particle mesh
Ewald (PME) method was used for the long-range electrostatic in-
teractions.[33,34] The LINCS procedure for covalent bond constraints
was used.[35] MD simulations were performed using a constant-NPT
ensemble. The energy of these complexes was minimized using
the steepest descent approach and heating the system for 20 ps
using the NVT ensemble. An equilibration step of 1 ns was followed
by a production run of 11 ns, with a time step of 1.0 fs, and trajec-
tories were collected every 10 ps. The total simulation time was
12 ns for each complex for the binding mode selection. Each analy-
sis was performed using the GROMACS modules and MD trajecto-
ries (1–12 ns).[26] Binding free energies (DGbind) were calculated
using Prime MM-GBSA, which is included in the Schrçdinger
Suite[24] and MD trajectories (1–12 ns).
rings connected via amide linkers. The long-tail THQ derivative
1d was determined to be a promising orally available com-
pound based on the results of in vitro studies. The in vivo
tissue selectivity for bone was revealed based on the fact that
1d clearly dissociates from DHT in OVX rats following oral ad-
ministration. Thus, as predicted, this compound was deter-
mined to have an attractive SARM profile. However, the in-
creased uterus and CG weights induced by 1d must be de-
creased to the levels observed in the OVX-vehicle. Therefore,
further optimization processes will be required to decrease vir-
ilizing effects to obtain ideal SARMs.
Experimental Section
Computational study
Ligand and AR preparation: Hydrogen atoms and charges were
added to the 2D structures of the ligands, and then converted into
their corresponding 3D structures using LigPrep.[24] The protein
structure of the AR (PDB ID: 3B68) was obtained from the Protein
Data Bank.[10] All of the residues were protonated, and the protein
structures were refined to relieve steric clashes with a restrained
minimization of the OPLS2005 force field until a final RMSD of
0.030 ꢁ was reached with respect to the input protein coordinates
using the Protein Preparation Wizard, which is included in the
Schrçdinger Suite 2012.[24]
Chemistry
General: All starting materials and reagents were purchased from
commercial suppliers and were used without further purification.
1H and 13C NMR spectra were recorded using a JEOL JNM-ECA
400 MHz spectrometer in solution in the deuterated solvent
(CDCl3); chemical shifts (d) are expressed in parts per million rela-
tive to tetramethylsilane, which was used as a reference. High-reso-
lution mass spectrometry (HRMS) analyses were performed on
a Thermo Scientific Exactive Mass Spectrometer in positive or neg-
ative electrospray ionization mode operating at 25000 resolution.
Docking: The docking studies were performed using Glide.[24] The
energy grid was built from the protein structures and was pre-
pared as described above. An atomic van der Waals radius scaling
factor, a partial atomic charge, and the sizes of the enclosing and
bounding boxes were used as the default values. The SP docking
protocol (ligands were docked flexibly, the sampling of ring confor-
mations was included, and nonplanar amide conformations were
penalized) was performed to be energy minimized on the OPLS-AA
non-bonded interaction with the grid, and each term (the Coulom-
bic term (DEcoul) and the vdW term (DEvdW)) was decomposed for
each residue.
N-(tert-Butyloxycarbonyl)-3-amino-2,2-dimethylpropionaldehyde
(12): Boc2O (59 g, 314 mmol) was added to a solution of 3-amino-
2,2-dimethylpropanol
9
(25 g, 242 mmol) in anhydrous THF
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 203

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
(120 mL) at 08C. The mixture was stirred at room temperature for
3 h. The mixture was evaporated, and the residue was diluted
using EtOAc (100 mL) and H2O (30 mL). The organic layer was sepa-
rated and washed with brine (30 mL), dried over Na2SO4, and
evaporated. The desired N-Boc-protected amino propanol was ob-
tained as an intermediate (49.6 g, >100%). A portion of the N-Boc-
protected aminopropanol (5.0 g, 24.6 mmol) and phosphoric acid
(0.65 mL, 12.5 mmol) were dissolved in DMSO (50 mL). Water-solu-
ble carbodiimide hydrochloride (WSCD·HCl; 14 g, 74.0 mmol) was
added and stirred at room temperature for 30 min. The mixture
was poured into cold water (50 mL) and extracted with EtOAc
(100 mLꢂ3). The combined organic layers were washed with H2O
(50 mL) and brine (50 mL), dried over Na2SO4, and evaporated to
generate the desired product (4.68 g, 95%). 1H NMR (400 MHz,
CDCl3): d=1.09 (s, 6H), 1.42 (s, 9H), 2.61 (s, 2H), 4.83 (brs, 1H),
9.45 ppm (s, 1H).
132.0, 133.1, 137.4, 147.5, 153.8, 165.9, 169.3 ppm; HRMS-ESI m/z
[M+H]+ calcd for C25H27N5O2: 429.2165, found: 430.2236.
4-Acetamido-N-(2-(8-cyano-3a,4,5,9b-tetrahydro-3H-cyclopen-
ta[c]quinolin-4-yl)-2-methylpropyl)benzamide (1a): Compound
1a was prepared by following the same method used to generate
compound 1d. 4-Acetamidoenzoic acid was used as an aromatic
1
carboxylic acid (22 mg, 94%). H NMR (400 MHz, CDCl3): d=1.03 (s,
3H), 1.13 (s, 3H), 2.19 (s, 3H), 2.29 (dd, J=8.24, 14.20 Hz, 1H),
2.48–2.55 (m, 1H), 2.88 (q, J=8.24 Hz, 1H), 3.17 (dd, J=5.95,
14.65 Hz, 1H), 3.30 (d, J=1.83 Hz, 1H), 3.79 (dd, J=7.33, 14.65 Hz,
1H), 3.89 (d, J=8.24 Hz, 1H), 4.66 (brs, 1H), 5.76 (d, J=5.04 Hz,
1H), 5.86–5.91 (m, 1H), 6.44 (t, J=6.41 Hz, 1H), 6.65 (d, J=8.70 Hz,
1H), 7.15–7.20 (m, 2H), 7.55 (d, J=8.70 Hz, 2H), 7.64 (s, 1H),
7.65 ppm (d, J=8.70 Hz, 2H); 13C NMR (100 MHz, CDCl3): d=22.4,
24.5, 31.2, 32.6, 38.3, 40.9, 47.4, 48.6, 58.4, 77.3, 77.5, 77.6, 116.5,
119.5, 120.7, 126.3, 128.2, 130.6, 132.0, 133.1, 133.5, 141.3, 141.9,
150.4, 167.5, 180.0 ppm; HRMS-ESI m/z [M+H]+ calcd for
C26H28N4O2: 428.2212, found: 429.2283.
N-(tert-Butyloxycarbonyl)-2-(8-cyano-3a,4,5,9b-tetrahydro-3H-cy-
clopenta[c]quinolin-4-yl)-2-methyl propylamine (13): N-Boc-pro-
tected aldehyde 12 (2.0 g, 10.0 mmol) was added to a mixture of
4-cyanoaniline 10 (1.2 g, 10.0 mmol), freshly prepared cyclopenta-
diene 11 (1.7 mL, 20.0 mmol), and TFA (0.88 mL, 11.0 mmol) in
CH3CN (5 mL) at 08C. The mixture was stirred at room temperature
for 1 h. The precipitate was filtered, washed with Et2O, dried under
reduced pressure, and the desired product was then obtained
N-(2-(8-Cyano-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]quinolin-4-
yl)-2-methylpropyl)-4-formamidebenzamide (1b): Compound 1b
was prepared by following the same method used to generate
compound 1d. 4-Formamidoenzoic acid was used as an aromatic
1
carboxylic acid (1.20 g, 59%). H NMR (400 MHz, CDCl3): d=1.08 (s,
3H), 1.13 (s, 3H), 2.30 (dd, J=8.24, 14.20 Hz, 1H), 2.46–2.56 (m,
1H), 2.83–2.93 (m, 1H), 3.17 (dd, J=5.50, 14.20 Hz, 1H), 3.29 (d, J=
1.83 Hz, 1H), 3.80 (dd, J=7.79, 14.20 Hz, 1H), 3.89 (d, J=7.79 Hz,
1H), 4.66 (s, 1H), 5.77 (d, J=5.50 Hz, 1H), 5.86–5.91 (m, 1H), 6.40
(t, J=6.87 Hz, 1H), 6.65 (d, J=8.24 Hz, 1H), 7.16–7.21 (m, 2H), 7.55
(s, 1H), 7.56 (d, J=8.70 Hz, 2H), 7.66 ppm (d, J=8.70 Hz, 2H);
13C NMR (100 MHz, CDCl3): d=22.4, 24.5, 32.6, 38.3, 40.9, 47.4, 48.6,
58.3, 77.0, 77.3, 77.6, 100.2, 116.5, 119.5, 120.9, 126.3, 128.2, 129.6,
130.6, 132.1, 133.1, 141.5, 150.5, 167.9, 169.3 ppm; HRMS-ESI m/z
[M+H]+ calcd for C25H26N4O2: 414.2056, found: 415.2102.
1
(2.4 g, 65%). H NMR (400 MHz, CDCl3): d=0.99 (s, 3H), 1.02 (s, 3H),
1.34 (s, 9H), 2.26 (dd, J=8.24, 14.65 Hz, 1H), 2.47–2.54 (m, 1H),
2.82–2.90 (m, 1H), 3.27 (s, 1H), 3.39 (dd, J=8.24, 14.20 Hz, 1H),
3.93 (d, J=8.24 Hz, 1H), 4.51 (brs, 1H), 4.70 (brs, 1H), 5.74–5.76
(m, 1H), 5.86–5.88 (m, 1H), 6.60 (d, J=8.24 Hz, 1H), 7.17–7.22 ppm
(m, 1H); HRMS-ESI m/z [M+H]+ calcd for C22H29N3O2: 367.2260,
found: 368.2331.
General amide formation procedure using aromatic carboxylic
acids: Preparation of 6-acetamido-N-(2-(8-cyano-3a,4,5,9b-tetra-
hydro-3H-cyclopenta[c]quinolin-4-yl)-2-methylpropyl)nicotin-
amide (1d): A solution of HCl in dioxane (4n, 136 mL) was added
to a stirred solution of N-Boc-protected tetrahydroquinoline 13
(40 g, 109 mmol) in anhydrous THF (100 mL) at room temperature.
The resulting mixture was then stirred at 608C for 1 h. The suspen-
sion was diluted with EtOAc (50 mL) and poured into 5n NaCl(aq)
(50 mL). The organic layer was separated and extracted with EtOAc
(100 mLꢂ3). The combined organic layers were washed with brine
(50 mL), dried over Na2SO4, and evaporated. The desired amine
was obtained as an intermediate. The N-Boc-deprotected tetrahy-
droquinoline in DMF (100 mL) was added to the mixture of 6-acet-
amidonicotinic acid (29 g, 163 mmol), WSCD·HCl (31 g, 163 mmol),
1-hydroxybenzotriazole (HOBt; 22 g, 163 mmol), and N-methylmor-
pholine (NMM; 29 mL, 163 mmol) in DMF (200 mL) at 08C. The
mixture was stirred at room temperature for 3 h. The mixture was
quenched with H2O (50 mL), extracted with EtOAc (100 mLꢂ3),
washed with brine (50 mL), dried over Na2SO4, and evaporated.
The residue was purified by column chromatography (hexane/
EtOAc 1:4) followed by charcoal treatment. The title compound
was obtained as a white solid (27.4 g, 59%). 1H NMR (400 MHz,
CDCl3): d=1.09 (s, 3H), 1.14 (s, 3H), 2.23 (s, 3H), 2.30 (dd, J=8.24,
14.20 Hz, 1H), 2.46–2.56 (m, 1H), 2.84–2.94 (m, 1H), 3.22 (dd, J=
5.95, 14.20 Hz, 1H), 3.32 (d, J=1.37 Hz, 1H), 3.78 (dd, J=7.33,
14.20 Hz, 1H), 3.90 (d, J=8.24 Hz, 1H), 4.57 (s, 1H), 5.77 (d, J=
5.04 Hz, 1H), 5.87–5.92 (m, 1H), 6.45 (t, J=6.41 Hz, 1H), 6.65 (d, J=
8.24 Hz, 1H), 7.17–7.22 (m, 2H), 8.01 (dd, J=2.29, 8.70 Hz, 2H),
8.21 (s, 1H), 8.24 (d, J=8.70 Hz, 1H), 8.64 ppm (d, J=2.29 Hz, 1H);
13C NMR (100 MHz, CDCl3): d=22.5, 25.1, 32.5, 38.3, 40.9, 47.4, 48.5,
58.4, 67.3, 77.0, 77.3, 77.6, 100.6, 113.2, 116.4, 120.7, 125.8, 130.6,
N-(2-(8-Cyano-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]quinolin-4-
yl)-2-methylpropyl)-4-(trifruoromethoxy)benzamide (1c): Com-
pound 1c was prepared by following the same method used to
generate compound 1d. 4-(Trifluoromethoxy)benzoic acid was
1
used as an aromatic carboxylic acid (31.0 g, 82%). H NMR (CDCl3):
d=1.09 (s, 3H), 1.14 (s, 3H), 2.31 (dd, J=8.24, 14.20 Hz, 1H), 2.47–
2.51 (m, 1H), 2.89 (q, J=8.24 Hz, 1H), 3.21 (dd, J=5.50, 14.20 Hz,
1H), 3.30 (d, J=1.83 Hz, 1H), 3.79 (dd, J=7.33, 14.20 Hz, 1H), 3.90
(d, J=8.24 Hz, 1H), 4.57 (s, 1H), 5.77 (d, J=5.04 Hz, 1H), 5.87–5.92
(m, 1H), 6.36 (t, J=6.41 Hz, 1H), 6.66 (d, J=8.24 Hz, 1H), 7.19 (s,
1H), 7.21 (dd, J=1.83, 8.24 Hz, 1H), 7.24–7.28 (m, 2H), 7.72–
7.78 ppm (m, 2H); 13C NMR (100 MHz, CDCl3): d=22.5, 24.4, 32.5,
38.3, 40.9, 47.4, 48.6, 58.4, 77.0, 77.3, 77.6, 100.7, 116.5, 120.7,
121.0, 126.3, 129.1, 130.6, 132.0, 132.8, 133.1, 133.5, 150.3, 151.9,
166.9 ppm; HRMS-ESI m/z [M+H]+ calcd for C25H24F3N3O2:
455.1821, found: 456.1891.
N-(2-(8-Cyano-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]quinolin-4-
yl)-2-methylpropyl)-6-ethoxynicotinamide (1e): Compound 1e
was prepared by following the same method used to generate
compound 1d. 6-Ethoxynicotinic acid was used as an aromatic car-
boxylic acid (6.46 g, 71%). 1H NMR (400 MHz, CDCl3): d=1.09 (s,
3H), 1.14 (s, 3H), 2.31 (dd, J=8.24, 14.20 Hz, 1H), 2.46–2.56 (m,
1H), 2.89 (q, J=8.24 Hz, 1H), 3.22 (dd, J=5.50, 14.20 Hz, 1H), 3.31
(d, J=1.89 Hz, 1H), 3.78 (dd, J=7.33, 14.20 Hz, 1H), 3.90 (d, J=
8.24 Hz, 1H), 4.57 (s, 1H), 5.74–5.80 (m, 1H), 5.87–5.93 (m, 1H),
6.41 (t, J=6.40 Hz, 1H), 6.65 (d, J=8.24 Hz, 1H), 7.17–7.23 (m, 2H),
8.00 (dd, J=2.29, 8.70 Hz, 1H), 8.24 (d, J=8.70 Hz, 1H), 8.64 ppm
(d, J=2.29 Hz, 1H); 13C NMR (100 MHz, CDCl3): d=14.9, 22.5, 24.3,
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 204

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
32.5, 38.2, 40.9, 47.3, 48.4, 58.3, 77.0, 77.3, 77.6, 100.8, 112.7, 116.5,
120.1, 126.3, 130.6, 132.0, 133.0, 133.6, 135.6, 141.3, 150.2,
164.6 ppm; HRMS-ESI m/z [M+H]+ calcd for C25H28N4O2: 416.2212,
found: 417.2282.
Keywords: androgen receptors · docking · selective androgen
receptor
modulators
·
osteoanabolic
activity
·
tetrahydroquinolines
6-Chloro-N-(2-(8-cyano-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]qui-
nolin-4-yl)-2-methylpropyl)nicotinamide (1 f): Compound 1 f was
prepared by following the same method used to generate com-
pound 1d. 6-Chloronicotinic acid was used as an aromatic carbox-
[1] Osteoporosis Prevention, Diagnosis, and Therapy, NIH Consensus State-
ment 2000, 17, 1–36.
[2] A. Schmidt, S. Harada, G. A. Rodan, Principles of Bone Biology (Eds.: J. P.
Bilezikian, L. G. Raisz, G. R. Rodan), Academic Press, San Diego, 1999,
Ch. 81, pp. 1125–1134.
1
ylic acid (292 mg, 88%). H NMR (400 MHz, CDCl3): d=1.07 (s, 3H),
1.12 (s, 3H), 1.36 (t, J=7.33 Hz, 3H), 2.24–2.35 (m, 1H), 2.44–2.57
(m, 1H), 2.82–2.92 (m, 1H), 3.17 (dd, J=5.95, 14.20 Hz, 1H), 3.30
(m, 1H), 3.74 (dd, J=7.79, 14.20 Hz, 1H), 4.01 (q, J=7.33 Hz, 2H),
4.58 (s, 1H), 5.75–5.82 (m, 1H), 5.87–5.94 (m, 1H), 6.32 (t, J=
6.41 Hz, 1H), 6.51 (d, J=9.16 Hz, 1H), 6.64 (d, J=8.24 Hz, 1H),
7.16–7.22 (m, 1H), 7.18 (s, 1H), 7.52 (dd, J=2.75, 9.62 Hz, 2H),
8.07 ppm (d, J=2.75 Hz, 1H); 13C NMR (100 MHz, CDCl3): d=22.6,
24.3, 32.5, 38.3, 40.9, 47.4, 48.5, 58.5, 77.0, 77.3, 77.6, 100.8, 116.5,
120.6, 124.8, 129.0, 130.6, 132.0, 133.6, 138.3, 148.0, 150.1,
165.3 ppm; HRMS-ESI m/z [M+H]+ calcd for C23H23ClN4O2:
406.1560, found: 407.1632.
[3] R. M. Neer, C. D. Arnaud, J. R. Zanchetta, R. Prince, G. A. Gaich, J. Y. Re-
ginster, A. B. Hodsman, E. F. Eriksen, S. Ish-Shalom, H. K. Genant, O.
Wang, B. H. Mitlak, N. Engl. J. Med. 2001, 344, 1434–1441.
[4] J. Rosen, A. Negro-Vilar, J. Musculoskeletal Neuronal Interact. 2002, 2,
222–224.
[5] M. L. Mohler, C. E. Bohl, A. Jones, C. C. Coss, R. Narayanan, Y. He, D. J.
Hwang, J. T. Dalton, D. D. Miller, J. Med. Chem. 2009, 52, 3597–3617.
[6] W. Gao, C. E. Bohl, J. T. Dalton, Chem. Rev. 2005, 105, 3352–3370.
[7] R. Narayanan, M. L. Mohler, C. E. Bohl, D. D. Miller, J. T. Dalton, Nucl.
Recept. Signaling 2008, 6, e010.
[8] R. V. Weatherman, R. J. Fletterick, T. S. Scanlan, Annu. Rev. Biochem.
1999, 68, 559–581.
[9] C. E. Bohl, D. D. Miller, J. Chen, C. E. Bell, J. T. Dalton, J. Biol. Chem. 2005,
280, 37747–37754.
[10] C. E. Bohl, Z. Wu, J. Chen, M. L. Mohler, J. Yang, D. J. Hwang, S. Mustafa,
D. D. Miller, C. E. Bell, J. T. Dalton, Bioorg. Med. Chem. Lett. 2008, 18,
5567–5570.
[11] F. Wang, X. Q. Liu, H. Li, K. N. Liang, J. N. Miner, M. Hong, E. A. Kallel, A.
van Oeveren, L. Zhi, T. Jiang, Acta. Crystallogr. Sect. F 2006, 62, 1067–
1071.
[12] A. A. Nirschl, Y. Zou, S. R. Krystek, Jr., J. C. Sutton, L. M. Simpkins, J. A. Lu-
pisella, J. E. Kuhns, R. Seethala, R. Golla, P. G. Sleph, B. C. Beehler, G. J.
Grover, D. Egan, A. Fura, V. P. Vyas, Y. X. Li, J. S. Sack, K. F. Kish, Y. An, J. A.
Bryson, J. Z. Gougoutas, J. DiMarco, R. Zahler, J. Ostrowski, L. G.
Hamann, J. Med. Chem. 2009, 52, 2794–2798.
[13] F. Nique, S. Hebbe, N. Triballeau, C. Peixoto, J. M. LefranÅois, H. Jary, L.
Alvey, M. Manioc, C. Housseman, H. Klaassen, K. Van Beeck, D. Guꢃdin, F.
Namour, D. Minet, E. Van der Aar, J. Feyen, S. Fletcher, R. Blanquꢃ, C.
Robin-Jagerschmidt, P. Deprez, J. Med. Chem. 2012, 55, 8236–8247.
[14] J. Ostrowski, J. E. Kuhns, J. A. Lupisella, M. C. Manfredi, B. C. Beehler,
S. R. Krystek, Jr., Y. Bi, C. Sun, R. Seethala, R. Golla, P. G. Sleph, A. Fura, Y.
An, K. F. Kish, J. S. Sack, K. A. Mookhtiar, G. J. Grover, L. G. Hamann, En-
docrinology 2007, 148, 4–12.
Biology
In vitro assays: The in vitro binding affinity for the AR was mea-
sured by competitive binding assay with the crude rat AR protein
fraction, which was prepared from the prostates of 11-week-old
male rats that had been castrated three days prior. Specific binding
was defined as the difference between the binding of radiolabeled
ligands in the presence (nonspecific binding) and absence (total
binding) of excess unlabeled ligands. Half-maximal inhibitory con-
centration (IC50) values were determined from a nonlinear regres-
sion analysis of the competitive binding curve. The in vitro agonis-
tic activity was measured using an AR transactivation assay with
the human TE-85 osteosarcoma cell line that had been transfected
with human AR and a luciferase reporter gene. The luciferase activ-
ity induced by 1000 nm DHT was set at 100% and used as a refer-
ence. Half-maximal effective concentration (EC50) values were de-
termined from a nonlinear regression analysis of the dose–re-
sponse curve.
[15] J. D. Kearbey, W. Gao, R. Narayanan, S. J. Fisher, D. Wu, D. D. Miller, J. T.
Dalton, Pharm. Res. 2007, 24, 328–335.
[16] J. N. Miner, W. Chang, M. C. Chapman, P. D. Finn, M. H. Hong, F. J. Lꢄpez,
K. B. Marschke, J. Rosen, W. Schrader, R. Turner, A. van Oeveren, H. Vi-
veros, L. Zhi, A. Negro-Vilar, Endocrinology 2007, 148, 363–373.
[17] N. Nagata, M. Miyakawa, S. Amano, K. Furuya, N. Yamamoto, K. Inogu-
chi, Bioorg. Med. Chem. Lett. 2011, 21, 1744–1747.
[18] N. Nagata, M. Miyakawa, S. Amano, K. Furuya, N. Yamamoto, H. Nejishi-
ma, K. Inoguchi, Bioorg. Med. Chem. Lett. 2011, 21, 6310–6313.
[19] K. Hanada, K. Furuya, N. Yamamoto, H. Nejishima, K. Ichikawa, T. Naka-
mura, M. Miyakawa, S. Amano, Y. Sumita, N. Oguro, Biol. Pharm. Bull.
2003, 26, 1563–1569.
[20] F. Nique, S. Hebbe, C. Peixoto, D. Annoot, J. M. LefranÅois, E. Duval, L.
Michoux, N. Triballeau, J. M. Lemoullec, P. Mollat, M. Thauvin, T. Prangꢃ,
D. Minet, P. Clꢃment-Lacroix, C. Robin-Jagerschmidt, D. Fleury, D.
Guꢃdin, P. Deprez, J. Med. Chem. 2012, 55, 8225–8235.
[21] J. Kazius, R. McGuire, R. Bursi, J. Med. Chem. 2005, 48, 312–320.
[22] N. Nagata, K. Kawai, I. Nakanishi, J. Chem. Inf. Model. 2012, 52, 2257–
2264.
In vivo pharmacology: The tissue-selective activity of the com-
pounds was examined using a post-menopausal rat osteoporosis
model. Twelve-week-old SD rats were bilaterally OVX or sham-oper-
ated. All animals were maintained without treatment for four
weeks to permit bone loss and were then treated with the com-
pounds once daily for eight weeks. Compound 1d was orally ad-
ministered once daily in a volume of 5 mLkgꢀ1 body weight (BW).
DHT was injected subcutaneously once daily in a volume of
1 mLkgꢀ1 BW. After eight weeks of treatment, the uterus and CG
of each mouse were excised and weighed immediately. The femo-
ral BMD was measured using dual-energy X-ray absorptiometry
with a bone mineral analyzer (DCS-600EX-IIIR; Aloka, Tokyo, Japan).
The experimental protocol was approved by the Institutional
Animal Care and Use Committee at Kaken Pharmaceutical Co. Ltd.
[23] E. Hur, S. J. Pfaff, E. S. Payne, H. Grøn, B. M. Buehrer, R. J. Fletterick, PLoS
Biol. 2004, 2, 1303–1312.
Acknowledgements
[24] Schrçdinger Suite 2012, Schrçdinger LLC, New York, NY (USA), 2012:
Maestro v. 9.3, SiteMap v. 2.6, Glide v. 5.8, Prime MM-GBSA v. 2.0, Qik-
Prop v. 3.5, LigPrep v. 2.5.
[25] a) J. T. Dalton, A. Mukherjee, Z. Zhu, L. Kirkovsky, D. D. Miller, Biochem.
Biophys. Res. Commun. 1998, 244, 1–4; b) C. A. Marhefka, W. Gao, K.
Chung, J. Kim, Y. He, D. Yin, C. E. Bohl, J. T. Dalton, D. D. Miller, J. Med.
The authors thank Dr. Hiroaki Nejishima, Sayuri Kataoka, Dr. Hir-
ohide Ishige, and Dr. Kiyoshi Inoguchi (Kaken Pharmaceutical) for
their support.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 205

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Chem. 2004, 47, 993–998; c) J. Kim, D. Wu, D. J. Hwang, D. D. Miller, J. T.
Dalton, J. Pharmacol. Exp. Ther. 2005, 315, 230–239.
[32] H. J. C. Berendsen, P. J. M. Postma, W. F. van Gunsteren, A. DiNola, J. R.
Haak, J. Chem. Phys. 1984, 81, 3684–3690.
[26] a) H. J. C. Berendsen, D. van der Spoel, R. van Drunen, Comput. Phys.
Commun. 1995, 91, 43–57; b) B. Hess, C. Kutzner, D. van der Spoel, E.
Lindahl, J. Chem. Theory Comput. 2008, 4, 435–447.
[33] T. Darden, D. York, L. Pedersen, J. Chem. Phys. 1993, 98, 10089–10092.
[34] U. Essmann, L. Perera, M. L. Berkowitz, T. Darden, H. Lee, L. G. Pedersen,
J. Chem. Phys. 1995, 103, 8577–8593.
[27] P. A. Grieco, A. Bahsas, Tetrahedron Lett. 1988, 29, 5855–5858.
[28] X. Zhang, Z. Sui, Expert Opin. Drug Discovery 2013, 8, 191–218.
[29] K. Furuya, N. Yamamoto, Y. Ohyabu, A. Makino, T. Morikyu, H. Ishige, K.
Kuzutani, Y. Endo, Biol. Pharm. Bull. 2012, 35, 1096–1104.
[30] J. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman, D. A. Case, J. Comput.
Chem. 2004, 25, 1157–1174.
[35] B. Hess, H. Bekker, H. J. C. Berendsen, J. G. E. M. Fraajie, J. Comput.
Chem. 1997, 18, 1463–1472.
Received: August 29, 2013
Revised: October 24, 2013
[31] W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, M. L. Klein,
J. Chem. Phys. 1983, 79, 926–935.
Published online on November 22, 2013
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 197 – 206 206