Identification of a 4-(hydroxymethyl)diarylhydantoin as a selective androgen receptor modulator

Article abstract of DOI:10.1021/jm300281x

Structural modification performed on a 4-methyl-4-(4-hydroxyphenyl) hydantoin series is described which resulted in the development of a new series of 4-(hydroxymethyl)diarylhydantoin analogues as potent, partial agonists of the human androgen receptor. T

Full text of DOI:10.1021/jm300281x

Article
pubs.acs.org/jmc
Identification of a 4‑(Hydroxymethyl)diarylhydantoin as a Selective
Androgen Receptor Modulator
Franco
̧
is Nique, Severine Hebbe, Nicolas Triballeau, Christophe Peixoto, Jean-Michel Lefranco
̧
is,
́
Helene Jary, Luke Alvey, Murielle Manioc, Christopher Housseman, Hugo Klaassen,† Kris Van Beeck,
́
̀
Denis Guedin,‡ Florence Namour, Dominque Minet, Ellen Van der Aar, Jean Feyen, Stephen Fletcher,
́
Roland Blanque, Catherine Robin-Jagerschmidt, and Pierre Deprez*
́
GALAPAGOS, Parc Biocitech, 102 Avenue Gaston Roussel, 93230 Romainville, France
S
* Supporting Information
ABSTRACT: Structural modification performed on a 4-
methyl-4-(4-hydroxyphenyl)hydantoin series is described
which resulted in the development of a new series of 4-
(hydroxymethyl)diarylhydantoin analogues as potent, partial
agonists of the human androgen receptor. This led to the
identification of (S)-(−)-4-(4-(hydroxymethyl)-3-methyl-2,5-
dioxo-4-phenylimidazolidin-1-yl)-2-(trifluoromethyl)-
benzonitrile ((S)-(−)-18a, GLPG0492) evaluated in vivo in a
classical model of orchidectomized rat. In this model, (−)-18a
exhibited anabolic activity on muscle, strongly dissociated from the androgenic activity on prostate after oral dosing. (−)-18a has
very good pharmacokinetic properties, including bioavailability in rat (F > 50%), and is currently under evaluation in phase I
clinical trials.
INTRODUCTION
■
The androgen receptor (AR), which belongs to the nuclear
receptor superfamily, is activated by endogenous androgens,
such as testosterone (T) and its metabolite dihydrotestosterone
(DHT). Modulation of the AR results in the control of several
physiological characteristics, including differentiation and
growth of male reproductive organs, as well as effects on hair
and skin (androgenic effects). In addition, other tissues such as
muscle and bone are also a target for androgens (anabolic
effects). Various steroidal AR ligands have been developed, but
their usage has been limited because of poor oral bioavailability
and a lack of selectivity between anabolic and androgenic
effects, leading to risk of serious side effects.1 Therefore, the
concept of a nonsteroidal, selective androgen receptor
modulator (SARM) emerged as an attractive target for drug
discovery. Our goal was to identify orally available SARMs,
Figure 1. AR ligands.
potent and efficacious in anabolic effects, while avoiding
undesired androgenic activity. A number of nonsteroidal
activity on muscle. However, low oral bioavailability7 of (+)-4
(F < 9% in rat and monkey) prompted us to undertake
structure−activity relationship (SAR) studies to maintain the in
vivo efficacy and the tissue-selective profile of (+)-4 while
significantly increasing oral bioavailability. Many modifications
have been made around the hydantoin scaffold, and only a few
compounds were identified with the desired in vivo profile and
good oral bioavailability. Herein, we describe the discovery of
(−)-18a (GLPG0492), a SARM with tissue differentiation for
SARMs that are in preclinical or clinical development have
been reported to date (Figure 1).2 Chemical scaffolds, including
bicalutamide derivatives 1,3 hydantoin compounds 2,4 and
quinolinone derivatives 3,5 have revealed potent and efficacious
anabolic activity. Recently, we reported the discovery of (R)-4-
[3,4-dimethyl-2,5-dioxo-4-(4-hydroxyphenyl)imidazolidin-1-
yl]-2-(trifluoromethyl)benzonitrile ((R)-(+)-4) as a new
SARM.6
This new partial agonist compound, (+)-4, displayed tissue
selectivity following oral administration in animal models with
reduced androgenic effects on prostate as compared to anabolic
Received: February 28, 2012
Published: September 10, 2012
© 2012 American Chemical Society
8236
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
a
Scheme 1. General Synthesis of 4-Methylhydantoin Analogues ( )-9a,b
a
Reagents and conditions: (a) KCN, (NH4)2CO3·H2O, EtOH/H2O (1/1), 60 °C, 18−93 h, 16−100%; (b) 4-bromo-2-(trifluoromethyl)-
benzonitrile, Cu2O, DMAc, 130−170 °C, 16−18 h, 18−63%; (c) MeI, NaH, DMF or DMAc, rt, 3 h, 28−100%; (d) Cs2CO3, THF/MeOH, 60 °C,
30 min, 73%; (e) HCl 4 N in dioxane, MeOH, 100 °C, 24 h, 86%.
a
Scheme 2. General Synthesis of 4-Aryl- or 4-Heteroaryl-4-(hydroxymethyl)hydantoin Analogues ( )-18a−j and ( )-19a,b
a
Reagents and conditions: (a) KCN, (NH4)2CO3·H2O, EtOH/H2O (1/1), 55−80 °C, 8−90 h, 46% to quantitative yield; (b) 4-bromo-2-
(trifluoromethyl)benzonitrile, Cu2O, DMAc, 130−160 °C, 1−48 h, 36−77%; (c) MeI, K2CO3, DMF, rt, 2−18 h, 81% to quantitative yield; (d)
BF3·Me2S, CH2Cl2, rt, 2−18 h, 48−87%; (e) n-Bu3SnH, Pd(PPh3)4, AcOH, toluene, reflux, 30 min, 48%; (f) Pd(OH)2/C, cyclohexene, EtOH, 80
°C, 3−48 h, 7−71%.
muscle vs prostate and with significant improvement of oral
bioavailability (>50%) in the rat. GLPG0492 is currently in
phase I clinical trials. The biological and pharmacokinetic (PK)
evaluation of this compound will be described.
As depicted in Scheme 2, the second series of novel
templates focused on substituting the 4-methyl group with
hydroxyl, as well as replacing the phenol by a panel of different
rings. These new analogues, ( )-18a−j and ( )-19a,b, were
obtained starting from protected hydroxymethyl ketones 10a−i
and 11a,b using a synthetic route similar to that described in
Scheme 1. Appropriate 1-aryl-2-(allyloxy)ethanones9 10a−c,10
10d−i,11 and 11a,b12 employed in this study were prepared in
one, two, or three steps from suitable, commercially available
compounds. Ketones 10a−i and 11a,b were converted to cyclic
4-aryl- or 4-heteroarylhydantoins 12a−i (50% to quantitative
yield) and 13a,b (46−82%) via the Bucherer−Bergs reaction.
N-arylation with 4-bromo-2-(trifluoromethyl)benzonitrile pro-
vided 14a−i (38−71% yield) and 15a,b (36−77%), which were
N-alkylated to yield compounds 16a−i (88% to quantitative
yield) and 17a,b (81−93%). Finally, targets ( )-18a,c−i were
obtained after removal of the allyl group using boron
trifluoride−methyl sulfide complex (in 48−85% yield).
Selective deprotection of the allyl group in the presence of
the methoxyphenyl ring in compound 16b was carried out with
tri-n-butyltin hydride to furnish ( )-18b (48% yield). On the
other hand, nonselective deprotection of 16b with boron
trifluoride−methyl sulfide complex provided ( )-18j (87%
CHEMISTRY
■
The design of novel SARM compounds was performed to
explore five different chemical modifications around the
hydantoin scaffold as depicted in Schemes 1−5.
The first series of analogues 9a,b, characterized by replacing
the phenol ring with 5-indolyl and 5-benzimidazolyl rings, were
prepared as illustrated in Scheme 1. Ketones 5a,b8 were
condensed with potassium cyanide and ammonium carbonate
in a Bucherer−Bergs-type reaction to give the 4-heteroar-
ylhydantoin scaffold 6. Regioselective copper coupling with 4-
bromo-2-(trifluoromethyl)benzonitrile was effected at the N1
position to afford the requisite N-arylhydantoins 7a,b in low to
moderate yield (18−63%). Subsequent N-alkylation with
methyl iodide in the presence of sodium hydride yielded
compounds 8a,b. Finally, targets ( )-9a,b were obtained after
removal of the tosyl or (2-(trimethylsilyl)ethoxy)methyl (SEM)
protecting groups under basic or acidic conditions with 73% or
86% yield, respectively.
8237
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
a
Scheme 3. General Synthesis of α-Substituted 4-Hydroxymethyl Analogues ( )-22a−d
a
Reagents and conditions: (a) SOCl2, MeOH, rt, 48 h, 91%; (b) 4-cyano-3-(trifluoromethyl)phenyl isocyanate, Et3N, THF, 76%; (c) LiHMDS (1 M
in hexanes), THF, −78 °C, 10 min, then RCHO, −78 °C, 30 min (26−55%).
a
Scheme 4. Synthesis of N-Aryl- and N-Heteroaryl-4-hydroxymethyl Analogues ( )-23a−e
a
Reagents and conditions: (a) 1,2-dichloro-4-iodobenzene, Cu2O, DMAc, 160 °C, 3 h, 45%; (b) MeI, K2CO3, DMF, rt, 5−18 h, 13−94% for B and
C pathways; (c) BF3·Me2S, CH2Cl2, rt, 5 h, 71% (over two steps for A pathway); (d) 4-cyano-3-methoxyphenyl isocyanate, NaOH (1 N), H2O, rt,
18 h, then HCl (12 N), 120 °C, 3 h, 38%; (e) RBr, Cu2O, DMAc, microwave, 160 °C, 50−100 min, 26c (21%), 26e (34%) or RBr, K2CO3, DMF,
microwave, 150 °C, 10 min, 26d (31%); (f) HCl (4 N in dioxane), rt, 48 h, 33−77%.
yield). The O-benzyl group was removed via catalytic
hydrogenation to afford ( )-19a,b (in 7−71% yield).
outlines three different ways (A−C) to prepare new
compounds of the fourth series. Elaboration of compound
( )-23a (Scheme 4A) proceeded as outlined in Scheme 2
starting from 12a using 1,2-dichloro-4-iodobenzene instead of
4-bromo-2-(trifluoromethyl)benzonitrile (32% overall yield).
Freshly prepared 4-cyano-3-methoxyphenyl isocyanate6 was
cyclized with 2-amino-3-hydroxy-2-phenylpropanoic acid13 to
give 24 in moderate yield (38%). N-methylation provided
( )-23b (Scheme 4B, 13% yield). To introduce heteroaryl
rings, intermediate 2514 was prepared starting from 2-
hydroxyacetophenone in two steps (92% overall yield): (i)
protection of the hydroxy function with tert-butyldimethylsilyl
(TBS); (ii) cyclization under Bucherer−Bergs conditions. A
coupling reaction conducted under microwave irradiation with
a panel of bromoheteroaryls then furnished 26c−e (21−34%).
N-methylation provided 27c−e (34−94%), which were
deprotected under acidic conditions to give ( )-23c−e in
yields ranging from 33% to 77% (Scheme 4C). Both allyl and
TBS protecting groups can be used, and there is no specific
reason to use one or the other to generate compounds 23.
On the basis of compound ( )-18a, various α-substituted 4-
hydroxymethyl analogues ( )-22a−d constituting a third series
were synthesized as shown in Scheme 3 by deprotonation−
hydroxyalkylation of the carbon of the hydantoin ring. Methyl
esterification of commercially available 2-(methylamino)-2-
phenylacetic acid was carried out to give 20 in 91% yield.
Immediate cyclization with freshly prepared 4-cyano-3-
(trifluoromethyl)phenyl isocyanate provided phenylhydantoin
21 in good yield (76%). Lithiation of 21 with lithium
bis(trimethylsilyl)amide at −78 °C was followed by addition
of acetaldehyde/propionaldehyde to provide a mixture of
diastereoisomeric compounds ( )-22a,b/( )-22c,d, respec-
tively, which were separated by silica gel chromatography, but
the relative configuration of the diastereoisomers was not
determined (22a (15%), 22b (11%)/22c (26%), 22d (29%)).
With further novel templates, structure modification focused
on replacing the (trifluoromethyl)benzonitrile ring (N1
position) with a panel of aryl or heterocyclic rings. Scheme 4
8238
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
a
Scheme 5. General Synthesis of N-Alkyl- and N-Acyl-4-hydroxymethyl Analogues ( )-29a−d
a
Reagents and conditions: (a) RX, K2CO3, DMF, rt, 4−5 h 70−90%; (b) MeCOCl, TEA, CH2Cl2, rt, 18 h, 61%; (c) BF3·Me2S, CH2Cl2, rt, 4−22 h,
51−80%.
a
Scheme 6. Determination of the Absolute Configuration of Compound 18a
a
Reagents and conditions: (a) 4-cyano-3-(trifluoromethyl)phenyl isocyanate, NaOH (0.5 N), dioxane, rt, 18 h, then HCl (12 N), 120 °C, 90 min,
57% over two steps; (b) Me2SO4, K2CO3, DMAC, rt, 15 h, 72%.
A final series of analogues was synthesized to explore N3
substitution (Scheme 5). Hydantoins 28a−d were prepared in
good yields (61−90%) from common intermediate 14a by
treatment with alkyl halides or acetyl chloride in the presence of
base. A final deprotection step furnished targets ( )-29a−d
(51−80%).
Generally, the target molecules were prepared as racemates.
Selected compounds were separated into individual enan-
tiomers by chiral HPLC.15 Thus, (−)-18a and (+)-18a were
initially isolated from a racemic mixture by means of chiral
preparative HPLC in excellent yields ((−)-18a, [α]2D3 = −40.8
(c 1, EtOH), 48% yield; (+)-18a, [α]2D3 = +41.1 (c 1, EtOH),
46.5% yield). To unequivocally assign the absolute config-
urations (Scheme 6), enantiopure compound 18a was prepared
from enantiopure (S)-(−)-α-(hydroxymethyl)phenylglycine
hydrochloride (30), obtained by fractional crystallization16
using (−)-cinchonidine.
Enantiopure hydrochloride 30 reacted with 4-cyano-3-
(trifluoromethyl)phenyl isocyanate to give the urea intermedi-
ate, which cyclized in acidic conditions to give hydantoin 31 in
57% yield over two steps. Final N-alkylation provided
compound 18a in 72% yield with an enantiomeric excess of
>95% (determined by chiral HPLC). The measurement of
optical rotation gave a negative value ([α]2D3((S)-18a) = −40.4
(c 1, EtOH)), thus assigning the S configuration to the more
active enantiomer (−)-18a. This result is in agreement with the
docking studies below and also with the previously described
hydantoin ( )-4 configuration, which has been confirmed upon
cocrystallization with the ligand binding domain (LBD) of the
AR.6 The asymmetric synthesis of compound (S)-(−)-18a has
been carried out on a 100 g scale from (2R,4S)-5-oxo-2,4-
diphenyloxazolidine-3-carboxylic acid benzyl ester using
Seebach methodology, and the dedicated optimization will be
reported soon. This asymmetric synthesis gave an additional
confirmation of the absolute configuration of the active
compound 18a.
RESULTS AND DISCUSSION
■
In Vitro. The strategic goal was to identify compounds
displaying a strong dissociation between anabolic effects on
muscle and/or the bone (requiring a strong agonist activity)
versus effects on prostate (where the compound should ideally
be an antagonist or a very weak agonist). This goal translates to
the identification of a partial agonist using an in vitro
transcriptional assay (Hela cell line transfected with hAR and
an androgen responsive element (ARE) coupled with
luciferase) as described in our previous work (preceding
paper in this issue).6 A conceptual framework for classifying
activity was chosen to help guide the SAR of the series and
identify potent compounds (potency parameter). For the
SARM profile, evaluation of the partial agonism property with
the molecular efficacy parameter indicated if the compounds
prepared are suitable for further in vivo evaluation. Potency and
molecular efficacy were generated on the basis of the Monod−
Wyman−Changeux model,17 which was used heuristically to
identify in vitro most promising compounds to be tested in the
in vivo Hershberger model. Dose−response curves have been
generated, crossing a range of DHT concentrations with a
range of concentrations for each SARM compound, using a
Schild-type curve-shift paradigm approach.6 In this transcrip-
tional assay, the in vitro potency value for DHT alone is 8 nM,
and we observed modifications of the DHT dose−response
curve when different concentrations of the SARM compound
were added. We showed that this model provided an excellent
framework for evaluation of partial agonism, leading to potency
and molecular efficacy data that translate to in vivo activity and
tissue selectivity. This is exemplified by using the reference
compound ostarine (1b), which progressed in clinical develop-
8239
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
ment and is known for its tissue-selective anabolic effect. In our
assay, 1b is potent (1.5 nM) with a molecular efficacy of 60,
which indicates partial agonism in our conditions. Within the
hydantoin series, we identified compound 4,6 which exhibited
the in vitro partial agonist profile (potency 0.9 nM/molecular
efficacy 132), leading to a potent in vivo activity on levator ani
hydrophobic residues Trp741 of helix 5 and Met895, Ile899,
and Val903 of helix 12. Interestingly, Bohl et al. theorize that
interacting with residues in this region is crucial in the balance
between agonist and antagonist properties.19 The benzimida-
zole analogue ( )-9b, which is significantly more hydrophilic
than the phenol or the indole, is likely to disturb this region and
cause loss of the partial agonist property, yielding a pure
antagonist. This is in line with the model whereby antagonism
can be the result of a displacement of helix 12 over the
coactivator binding pocket as first shown for raloxifene in the
estrogen receptor.20
(LA) muscle in an orchidectomized (ORX) rat model (A50
=
0.5 mg/kg/day). This compound showed a strong dissociation
(factor of 140) between the muscle and the prostate. The SAR
indicated that the presence of the phenol substituent on the
hydantoin ring is critical for both the potency and the partial
agonism, the corresponding unsubstituted phenyl ring being an
antagonist (potency 37 nM/molecular efficacy 3). Thus, the
phenol group gives a clear advantage in terms of the overall
SARM profile; however, it is less favorable in terms of oral
exposure, especially due to the potentially direct phase II
conjugation process (sulfation, glucoronidation, ...) and
clearance. Actually, hydantoin (+)-4 displayed poor oral
bioavailability7 (less than 9% both in rat and in monkey),
which is not compatible with clinical development. The goal of
the lead optimization program was thus to keep the partial
agonism in vitro profile while improving oral bioavailability.
We used our recent crystallographic study18 of hydantoin
(+)-4 bound to the hAR LBD, which showed key interactions
between the diarylhydantoin and the binding pocket of the
receptor. Initially focusing on the interaction between the
hydroxyl group of the phenol and His874, we thought to retain
the interaction with various bioisosteres of the phenol, with the
goal to retain a H-bond interaction with His874. However, the
prepared compounds (for example, OH replaced by NH2,
CH2OH, and CH(CH3)OH) turned into full antagonists.
Similarly, benzimidazole ( )-9b displayed an antagonist profile.
Surprisingly, the indole analogue (compound ( )-9a with one
nitrogen to carbon replacement when compared to benzimi-
dazole ( )-9b) restored the desired partial agonist profile
initially observed with the phenol ( )-4, with a molecular
efficacy of 35, albeit with a 70-fold weaker in vitro potency (70
nM).
As shown in Table 1, replacement of the phenol moiety by
the indole led to a 45-fold decrease in potency (potency 70
Table 1. In Vitro Activity and Absolute Oral Bioavailability
for 4-Methyldiarylhydantoin Analogues on hAR (Hela Cells)
in vitro
potency
(nM)
in vitro
molecular
efficacy
a
chiral
center
F
compd
R
(%)
DHT
TP
8
4000
600
114
132
35
1.5
1.6
0.9
70
40
( )-4
(+)-4
( )-9a
( )-9b
p-PhOH
p-PhOH
indole
RS
R
1.5
9
RS
RS
19
100
benzimidazole
1
a
F was determined in rats with a single oral dose of 10 mg/kg in
EtOH/PEG400/H2O (1/79/20, v/v/v) vehicle and a single iv dose of
3 mg/kg in DMSO/PEG400/H2O (1/65/34, v/v/v).
nM) vs that of the phenol hydantoin ( )-4. Therefore, the
hydantoin indole was not the optimal compound. On the other
hand, oral bioavailability was significantly improved with the
indole and the benzimidazole substituents on the hydantoin
(9a, F = 19%, and 9b, F = 100%, respectively) compared to the
phenol (( )-4, F < 9%). These results confirmed that although
the phenol moiety is important for the partial agonism profile,
it also turns out to be a key issue for PK. Therefore, the
proposal was to identify active partial agonists devoid of the
phenol group, thereby potentially leading to improved PK
properties suitable for drug development.
Starting from the X-ray costructure of compound 4 within
the AR LBD,18 docking studies confirmed the interaction
between the NH of ( )-9a and His874 in the AR LBD (Figure
2), thus validating the bioisosteric replacements of the phenol
moiety. In addition, the remaining carbon atoms of the five-
membered ring were predicted to interact with nearby
Our strategy to remove the phenol moiety and overcome the
lack of bioavailability was to use and pick an additional
interaction with the AR LBD and remove the phenol
interaction with His874. Analysis of the X-ray structure of
DHT within the AR LBD indicates an H-bond interaction
between the hydroxyl group of DHT and Asn705. Starting from
the X-ray crystal of compound (+)-4, a new series was designed
around the hydantoin scaffold to target the formation of a H-
bond with the Asn705 residue. As part of this work a new series
of compounds were synthesized, including ( )-18a and
( )-18j, which bear a hydroxymethyl group on the hydantoin
scaffold to interact with Asn705.
The docking studies indicate interactions of analogues
( )-18a with Asn705 and Arg752 and ( )-18j with Asn705,
Arg752, and His874 (Figure 3). As proposed earlier,19 such
ligands are capable of stabilizing an agonist conformation of the
AR LBD by tightening helices 3, 5, 11, and 12 together through
Figure 2. Simplified view of the binding site showing how (R)-9a
(orange carbons) would interact with the AR LBD (green tube and
carbons), highlighting the two H-bonds with Arg752 and His874.
Residues interacting through a hydrophobic effect with the indole are
simply shown as sticks (Trp741 on helix 5, Met895, Ile899, and
Val903 on helix 12).
8240
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
Figure 3. Simplified view of the binding site showing how (S)-18a (left) and (S)-18j (right) would interact with the AR LBD (green tube and
carbons), highlighting the two H-bonds with Arg752 and Asn705 and the supplementary interaction of (S)-18j with His874.
both hydrogen bonding and van der Waals contacts. The
modeling studies were confirmed by the in vitro activities.
The in vitro activity and absolute oral bioavailability of these
compounds are shown in Table 2. Despite the additional
(+)-18a isomer is almost inactive. These data are in good
agreement with the docking studies. Direct binding of (S)-
(−)-18a to the AR receptor was also measured and was found
to be in a range similar to that in the transcriptional assay (21
nM vs 13 nM).
Overall, introduction of the hydroxymethyl substituent on
the hydantoin ring led to retention of potency, possibly from
the additional interaction with Asn705, and maintained a partial
agonism profile, with modulation of potency depending on the
aryl group substituent (18a−j, Tables 2 and 3). To further
Table 2. In Vitro Activity and Absolute Oral Bioavailability
for 4-(Hydroxymethyl)diarylhydantoin Analogues
( )-18a,b,j (−)-18a, and (+)-18a
Table 3. In Vitro Activity and Absolute Oral Bioavailability
for 4-(Hydroxymethyl)diarylhydantoin Analogues ( )-18c−
i and ( )-19a,b
in vitro
a
chiral
in vitro
molecular
F
compd
Ar
center potency (nM)
efficacy
(%)
( )-18a Ph
(−)-18a Ph
RS
S
12
78
160
7
63
56
13
b
(+)-18a
Ph
R
1250
30
NT
25
( )-18b p-
RS
17
a
compd
R
chiral center in vitro potency (nM)
F
(%)
PhOMe
( )-18a
( )-18c
( )-18d
( )-18e
( )-18f
( )-18g
( )-18h
( )-18i
( )-19a
( )-19b
Ph
RS
RS
RS
RS
RS
RS
RS
RS
RS
RS
12
60
5
63
( )-18j
p-PhOH
RS
0.9
60
15
b
b
o-PhCl
NT
a
F was determined in rats with a single oral dose of 10 mg/kg in
EtOH/PEG400/H2O (1/79/20, v/v/v) vehicle and a single iv dose of
3 mg/kg in DMSO/PEG400/H2O (1/65/34, v/v/v). NT = not
tested.
m-PhCl
p-PhCl
NT
99
b
8
p-PhF
11
60
150
40
150
133
87
b
b
b
b
b
m-PhCN
p-PhCN
m-PhMe
2-thienyl
3-thienyl
NT
NT
NT
NT
NT
interaction of the −CH2OH, hydantoin ( )-18j kept a high
potency similar to that of ( )-4 (potency 0.9 and 1.6 nM,
respectively) and maintained a partial agonist profile (molecular
efficacy 60 and 114, respectively). However, where the phenol
group is not present, ( )-18a also displayed a partial agonism
profile (molecular efficacy 78) with good potency (12 nM).
This was the first time we identified a partial agonist within the
hydantoin series with a compound lacking an interaction with
His874 (through the phenol), knowing that compounds
without the phenyl group usually display an antagonist profile.
This surprising finding opened new perspectives to solve the
PK issues, in particular the bioavailability, while resulting in
potent compounds with a partial agonism profile. The
(hydroxymethyl)hydantoin compound ( )-18a retained a
partial agonist profile (molecular efficacy 78), which was
clearly due to a specific effect of the hydroxymethyl substituent
as the corresponding compound bearing the methyl substituent
gave an antagonist profile (potency 37 nM, molecular efficacy
3).6 The racemic mixture ( )-18a has been resolved to its pure
enantiomers by chiral chromatography. The (S)-(−)-18a
isomer proved to be as active as the racemate, while the (R)-
a
F was determined in rats with a single oral dose of 10 mg/kg in
EtOH/PEG400/H2O (1/79/20, v/v/v) vehicle and a single iv dose of
3 mg/kg in DMSO/PEG400/H2O (1/65/34, v/v/v). NT = not
tested.
b
explore the impact of the hydroxymethyl substituent, we
synthesized substituted 4-hydroxymethyl analogues ( )-22a−d
as diastereoisomers (methyl and ethyl substituents). However,
( )-22a−d are antagonists, confirming that this region of the
molecule also plays a critical role in the agonism/antagonism
profile of the compounds. Therefore, it was decided to keep the
hydroxymethyl as a critical substituent on the hydantoin to
conserve the desired partial agonist profile.
Critically, the bioavailability of the new (hydroxymethyl)-
hydantoin series was evaluated. Conversion of the methyl to
hydroxymethyl on the hydantoin scaffold along with removal of
the hydroxyl of the phenol (i.e., (+)-4 vs ( )-18a) significantly
8241
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
improved bioavailability: ( )-18a exhibited a much better
bioavailability of 63% in rat. Comparison of the two phenol
compounds 4 and ( )-18j shows that ( )-18j has a slightly
improved oral bioavailability (9% vs 15%), confirming the
deleterious role of the phenol moiety for bioavailability.
Therefore, overall, the outcome of the (hydroxymethyl)-
hydantoin series was really unexpected and very surprising.
We observed that, despite the OH of the phenol being deleted,
the potency and the desired partial agonism profile could be
retained while resulting in a great improvement in bioavail-
ability. Thus, ( )-18a fulfilled the key criteria of potency,
partial agonism, and good oral bioavailability and was selected
for further improving the SAR of the remaining hydantoin
substituents as indicated in Tables 3 and 4.
analogues, ( )-23a−e, were synthesized and tested in the hAR
assay. Replacement of the 4-cyano-3-(trifluoromethyl)phenyl
with heterocycles still bearing electron-withdrawing groups for
the ability to make a hydrogen bond with Arg752 resulted in
analogues ( )-23c−e. These three compounds turned out to
be weakly active with an antagonist profile (molecular efficacy
<1), suggesting that heterocycles did not correctly orient to
interact with the AR LBD. On the other hand, as shown in
Table 4, the potency of the 4-cyano-3-methoxyphenyl analogue
( )-23b and more surprisingly of the 3,4-dichlorophenyl
analogue ( )-23a was reduced only 5-fold (potency 60 nM),
similarly to previous observations.6
Analogues ( )-29a−d were prepared in which the N3-
methyl group was replaced. Ethyl-substituted ( )-29a gave
similar potency (7 nM), and both isopropyl (( )-29b) and
propargyl (( )-29c) analogues resulted in a slight decrease in
potency (20 and 40 nM, respectively). Acylation as in ( )-29d
led to further loss of activity (potency 150 nM) and turned the
profile to full antagonism (molecular efficacy 2).
Table 4. In Vitro Activity and Absolute Oral Bioavailability
for 4-(Hydroxymethyl)diarylhydantoin Analogues ( )-23a,b
and ( )-29a−d
In Vivo. As several of these racemic compounds displayed
partial agonist profiles in vitro with potencies close to that of
( )-4, we evaluated the ability to stimulate muscle anabolism vs
prostate growth in the Hershberger model of ORX rats.
Castrated rats were left untreated for one week and then orally
treated for four days with test compound (10 mg/kg/day). The
weight of the LA muscle was considered a marker of anabolic
activity, whereas the weight of the ventral prostate (VP) was
regarded as a marker of androgenic activity.21 Results of these
experiments are shown in Table 5.
a
chiral
in vitro
F
compd
R1
R2
center
potency (nM)
(%)
( )-18a 3-CF3-4-
Me
RS
12
63
CNPh
b
( )-23a 3-Cl-4-ClPh
Me
Me
RS
RS
60
60
NT
NT
b
( )-23b 3-OMe-4-
CNPh
At the dose of 10 mg/kg/day, compounds ( )-18a,d,g,
( )-23a, and ( )-29a−c restored the weight of LA to near that
of sham control animals while the VP weights remained far
below those of intact animals, thereby indicating dissociation of
anabolic and androgenic effects in accordance with the in vitro
activities of a potent partial agonist. Compounds ( )-18c,h,i
displayed weaker in vivo activity on LA, in line with weaker in
vitro potency. Despite the fact that ( )-18j was the most active
compound in vitro (potency 0.9 nM), it was not the best
anabolic compound (even at 30 mg/kg/day). This might be
explained by its low oral bioavailability. In addition, compound
( )-18j did not show a great dissociation between anabolic and
androgenic effects. Finally, methoxyphenyl ( )-18b resulted in
the highest anabolic effect among all these molecules, albeit not
dissociated from androgenic activity. These data are not in line
with the in vitro profile of ( )-18b but can be explained by in
vivo demethylation of the methoxy group of ( )18b to lead to
the phenol ( )-18j, which is potent, but not very well
dissociated between muscle and prostate.
( )-29a 3-CF3-4-
Et
RS
RS
RS
RS
7
20
76
CNPh
b
b
b
( )-29b 3-CF3-4-
iPr
NT
NT
NT
CNPh
( )-29c 3-CF3-4-
propargyl
COMe
40
CNPh
( )-29d 3-CF3-4-
150
CNPh
a
F was determined in rats with a single oral dose of 10 mg/kg in
EtOH/PEG400/H2O (1/79/20, v/v/v) vehicle and a single iv dose of
3 mg/kg in DMSO/PEG400/H2O (1/65/34, v/v/v). NT = not
tested.
b
As depicted in Table 3, investigations on modifications of the
aromatic substituent close to the hydroxymethyl group were
conducted. The SAR revealed phenylhydantoin ( )-18a and
substituted phenylhydantoins ( )-18c−i are more potent than
thiophenes ( )-19a,b, which led to a significant loss in
potency.
The racemate ( )-18a and the pure enantiomer (S)-(−)-18a
completely reverted the muscle wasting in ORX rats at 10 mg/
kg/day following oral administration; no further increase was
observed with the higher dose of 30 mg/kg/day. Interestingly,
the maximum activity observed on prostate (26%) was achieved
at the dose of 10 mg/kg/day. Thus, the maximum effect on LA
muscle was reached at a dose displaying a weak activity on VP.
To further explore the anabolic potency and selectivity of
(S)-(−)-18a, and compare them to those of TP, a wide dose
range experiment was performed in the same Hershberger
model on ORX male rats (castrated rats left untreated for one
week and then orally treated for four days with (S)-(−)-18a,
Figure 4). Dose−response curves for LA and VP in response to
TP were almost superimposable, showing A50 = 0.30 and 0.20
mg/kg/day, respectively. On the contrary, (S)-(−)-18a showed
Substitution on the phenyl group indicated that introduction
of halogen atoms in the meta or para position (( )-18d−f)
induced no real change in potency (5−11 nM). The o-
chlorophenyl analogue ( )-18c resulted in a decrease of
activity (potency 60 nM). The electron-withdrawing cyano
groups in ( )-18g,h decreased potency 5−12-fold compared to
that of ( )-18a. The same observation was made with the m-
methyl group (potency 40 nM). Although the substitutions
depicted in Table 3 did not significantly improve the in vitro
profile of compound ( )-18a, the active compounds identified
from this study were evaluated in vivo.
The next study was to define the importance of the aryl ring
and substitution at the N1 position. Although it is well-known
that variations at this part of nonsteroidal androgen molecules
are very limited to retain good affinity for the AR, a number of
8242
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
Table 5. In Vivo Activity for a Panel of Hydantoin Analogues
a
compd
R1
Me
R2
R3
R4
anabolic activity(LA) (%)
99
androgenic activity(VP) (%)
73
b
TP
c
c
( )-4
p-PhOH
p-PhOH
Ph
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
3-Cl-4-ClPh
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Et
68/89
8/16
c
c
(R)-(+)-4
( )-18a
(S)-(−)-18a
( )-18b
( )-18c
( )-18d
( )-18e
( )-18g
( )-18h
( )-18i
( )-18j
( )-23a
( )-29a
( )-29b
( )-29c
Me
44/77
12/26
c
c
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
CH2OH
101/96
13/14
c
c
Ph
101/106
24/26
c
c
p-PhOMe
o-PhCl
m-PhCl
p-PhCl
m-PhCN
p-PhCN
m-PhMe
p-PhOH
Ph
130
95
26
66
35
84
12
24
7
16
0
26
11
0
c
c
82
55
96
23
59
60
43
Ph
3-CF3-4-CNPh
3-CF3-4-CNPh
3-CF3-4-CNPh
112
109
96
Ph
iPr
Ph
propargyl
a
b
All the test compounds were administered orally with a dose of 10 mg/kg/day. Testosterone propionate (TP) was administered subcutaneously at
c
1 mg/kg/day. Test compounds were administered orally with a dose of 30 mg/kg/day.
operated group. This indicates that dissociation observed in the
short-term model is maintained after long-term treatment with
10 mg/kg/day of compound (S)-(−)-18a by the oral route.
The PK profile of (S)-(−)-18a, after intravenous (iv) and
oral (po) routes to Sprague−Dawley male rats, is shown in
Table 6. (S)-(−)-18a exhibited good exposure (area under the
Table 6. Pharmacokinetic Profile of (S)-(−)-18a
(GLPG0492) in Rats after Single Administration
CL
administration (mL/min/kg)
T1/2
(h)
Vdss
(L/kg)
AUC(0−∞)
(ng.h/mL)
F
(%)
a
b
iv
21.7
1.5
2.5
2362
c
d
d
po
2.2
4675
56
Figure 4. Comparison of the anabolic vs androgenic activities of (S)-
a
b
Two groups of three rats received the treatment. Single iv dose of 3
(−)-18a (GLPG0492) and testosterone propionate in ORX rats. A50
=
c
mg/kg in DMSO/PEG200/H2O (1/79/20, v/v/v). Single oral dose
of 10 mg/kg in EtOH/PEG200/H2O (1/79/20, v/v/v). Compound
(S)-(−)-18a was dissolved in EtOH. PEG400 and H2O were added.
The mixture was stirred and placed in an ultrasound bath to give a
AD50, dose displaying 50% activity on levator ani or ventral prostate.
Student’s t test (vs ORX): *, p < 0.05; **, p < 0.01; ***, p < 0.001.
d
limpid solution at 2 mg/mL. N = 2 rats.
a clear dissociation between anabolic and androgenic effects;
the dose displaying 50% activity on LA was 0.75 mg/kg/day,
while at the maximum 30% activity can be achieved on VP at
the highest dose tested (30 mg/kg/day). This clearly
demonstrated the tissue selectivity of LA vs VP for (S)-
(−)-18a, thereby leading to its classification as an SARM, with
a promising preclinical safety profile (muscle vs prostate) for a
potential treatment for cachexia.
Dissociation between LA muscle and VP has also been
evaluated in a longer model. Castrated rats at the age of seven
weeks were treated with (S)-(−)-18a by the oral route for three
months in a prophylactic setting at the dose of 10 mg/kg/day.
Normalizing sham-operated rat at 100% and ORX rat at 0%, we
observed in the treated group that the LA weight was fully
restored (117%), at a level similar to that of the sham-operated
group. Conversely, the increase in prostate weight in rats
treated with (S)-(−)-18a was much lower (25%), with the
prostate weight closer to that of the ORX group than the sham-
curve, AUC) after oral and iv administration, moderate
clearance (CL; around a third of hepatic blood flow), a
medium volume of distribution (Vdss), and good bioavailability7
(F = 56%). Following extensive safety profiling in vitro and in
vivo, (S)-(−)-18a was selected as a clinical candidate for the
treatment of cachexia under the code GLPG0492.
CONCLUSION
■
Starting from a hydantoin series represented by compound
( )-4, an extensive plan of structural modifications was
undertaken with the goal of maintaining the partial agonism
profile of compound ( )-4 while significantly improving the
oral bioavailability. Thus, five different modifications around the
hydantoin were synthesized and explored, leading to a series of
new hydantoins bearing a hydroxymethyl substituent. Un-
8243
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
DMF was added iodomethane. The mixture was stirred at rt for 2−18
h, evaporated to dryness, diluted with water, and extracted with ethyl
acetate. The organic layer was dried over magnesium sulfate, filtered,
and evaporated. The crude product was purified by chromatography
over silica gel to give the expected product or used in the next step
without purification.
Method E is deprotection with trifluoroborane−dimethyl sulfide
complex. To a solution of 4-((allyloxy)methyl)-3-alkyl-1-arylimidazo-
lidine-2,4-dione or 4-((allyloxy)methyl)-3-acyl-1-arylimidazolidine-2,4-
dione in dichloromethane was added trifluoroborane−dimethyl sulfide
complex in dichloromethane. The mixture was stirred at rt for 2−18 h,
poured into a saturated aqueous sodium bicarbonate solution, and
extracted with ethyl acetate. The organic layer was dried over
magnesium sulfate, filtered, and evaporated. The crude product was
purified by chromatography over silica gel or crystallized to provide
the final target.
expectedly, although the potency compared to that of the
reference ( )-4 was not improved by these modifications, the
effect on the oral bioavailability was profound, thereby opening
the way to clinical development of compounds with a partial
agonism profile. Several compounds around the
(hydroxymethyl)hydantoin series induced transcriptional activ-
ity of the hAR at nanomolar concentrations in vitro. Partial
agonism observed in vitro translated to tissue selectivity of
muscle vs prostate in vivo after oral administration in a
castrated rat model. Particularly, (S)-(−)-18a maintained
levator ani muscle weight with potency similar to that of
testosterone while causing only a weak stimulation of the
prostate. This clear dissociation between anabolic and
androgenic end points in vivo is representative of an SARM.
The discovery of this (hydroxymethyl)diarylhydantoin core
constitutes a significant advance in the search for efficient and
highly tissue-selective nonsteroidal SARMs. The promising
compound (S)-(−)-18a (GLPG0492) is now undergoing phase
I clinical trials.
4-[2,5-Dioxo-4-(hydroxymethyl)-3-methyl-4-phenylimidazo-
lidin-1-yl]-2-(trifluoromethyl)benzonitrile (( )-18a). Step 1: 1-
Phenyl-2-(2-propenyloxy)ethanone (10a). The title compound was
prepared according to method A from 2-hydroxy-1-phenylethanone
(3.66 g, 26.9 mmol), allyl bromide (17 mL), calcium sulfate (16.3 g,
120 mmol), and silver(I) oxide (10.8 g, 46.6 mmol) to give 10a as a
yellow oil (1.6 g, 33%) after purification on silica gel with 90/10
EXPERIMENTAL SECTION
1
■
heptane/ethyl acetate. H NMR (400 MHz, DMSO): δ 4.10 (d, J =
5.4 Hz, 2H), 4.88 (s, 2H), 5.20 (d, J = 10.5 Hz, 1H), 5.31 (dd, J = 1.48
and 17.3 Hz, 1H), 5.91−6.00 (m, 1H), 7.56 (t, J = 7.6 Hz, 2H), 7.68
(t, J = 7.3 Hz, 1H), 7.94 (d, J = 7.3 Hz, 2H).
Unless otherwise stated, all procedures were performed under a
nitrogen atmosphere using anhydrous solvents purchased from
commercial sources. Commercially available starting materials and
reagents were used as received without further purification. Thin-layer
chromatography was performed using precoated Merck silica gel 60
F254 plates, visualized under UV light or after staining by spraying a
solution of 20% phosphomolybdic acid in ethanol and then heating.
Preparative chromatography was conducted over Merck silica gel 60
(0.040−0.063 mm). All 1H NMR spectra were recorded using a
Bruker DPX (300 MHz) or Advance (400 MHz) spectrometer, in
solution in the deuterated solvent mentioned; chemical shifts are
expressed in parts per million as a δ value relative to the shift of the
tetramethylsilane used as a reference. LC/MS was performed on a
Waters HPLC instrument with a 2996 photodiode array detector
coupled to an LCT TOF mass spectrometer with electrospray
ionization. HPLC method was as follows: column used, Waters Xterra
RP18 (3.5 μm, 50 × 2.1 mm i.d.); solvents, (A) water + 0.01% of
formic acid and (B) CH3CN + 0.01% of formic acid; gradient method,
from 95/5 A/B at time 0 to 100% B after 4 min, then 100% B for 3
min. The purity of the described compounds was >98% according to
HPLC analysis. Melting points were determined on a Kofler block and
are uncorrected. Optical rotations ([α]2D3 were recorded on a Propol
polarimeter. Specific rotations are given as degrees per decimeter, and
the concentrations c are reported as grams per 100 mL of the specific
solvent and were recorded at 23 °C.
Step 2: 4-Phenyl-2-(2-propenyloxy)imidazolidine-2,5-dione
(12a). The title compound was prepared according to method B
from 10a (1.6 g, 9.07 mmol), potassium cyanide (1.2 g, 18.4 mmol),
and ammonium carbonate (18.4 g, 191 mmol) in 1/1 EtOH/H2O (50
mL). The mixture was heated at 55 °C for 26 h to yield 12a as a white-
yellow solid (1.9 g, 85%) which was used in the next step without
purification. 1H NMR (400 MHz, DMSO): δ 3.53 (d, J = 9.8 Hz, 1H),
3.98 (d, J = 9.8 Hz, 1H), 3.94−4.02 (m, 2H), 5.17 (d, J = 10.5 Hz,
1H), 5.27 (d, J = 17 Hz, 1H), 5.79−5.89 (m, 1H), 7.32−7.41 (m, 3H),
7.55 (d, J = 7.2 Hz, 2H), 8.62 (s, 1H), 10.74 (s, 1H).
Step 3: 4-[2,5-Dioxo-4-phenyl-4-[(2-propenyloxy)methyl]-
imidazolidin-1-yl]-2-(trifluoromethyl)benzonitrile (14a). The title
compound was prepared according to method C from 12a (1.85 g,
7.51 mmol), copper(I) oxide (0.644 g, 4.5 mmol), and 4-bromo-2-
(trifluoromethyl)benzonitrile (1.9 g, 7.51 mmol) in DMAc (4 mL).
The mixture was heated at 160 °C for 3 h, and the crude product was
purified by chromatography over silica gel with 80/20 heptane/ethyl
1
acetate to provide 14a as a white amorphous solid (1.9 g, 61%). H
NMR (400 MHz, DMSO): δ 3.74 (d, J = 9.8 Hz, 1H), 4.08−4.09 (m,
2H), 4.19 (d, J = 9.8 Hz, 1H), 5.16 (d, J = 11.1 Hz, 1H), 5.24 (d, J =
17.2 Hz, 1H), 5.83−5.93 (m, 1H), 7.41−7.50 (m, 3H), 7.66 (d, J = 7.4
Hz, 2H), 8.00 (d, J = 8.4 Hz, 1H), 8.11 (s, 1H), 8.33 (d, J = 8.4 Hz,
1H), 9.62 (s, 1H). LC/MS (tR = 3.02 min): m/z 414 (M − H)−.
Step 4: 4-[2,5-Dioxo-3-methyl-4-phenyl-4-[(2-propenyloxy)-
methyl]imidazolidin-1-yl]-2-(trifluoromethyl)benzonitrile (16a).
The title compound was prepared according to method D from 14a
(1.9 g, 4.58 mmol), iodomethane (0.97 mL, 15.5 mmol), and
potassium carbonate (0.75 g, 5.5 mmol) in DMF (4 mL). The mixture
was stirred at rt for 4 h to give 16a as a yellow oil (1.9 g, 97%) which
Method A is protection with allyl bromide following the protocol
described in the literature.10
Method B is a Bucherer−Bergs reaction. To a solution of 1-
heteroarylethanone, 1-aryl-2-(allyloxy)ethanone, or 1-heteroaryl-2-
(benzyloxy)ethanone in ethanol/water (50/50) were added potassium
cyanide and ammonium carbonate. The mixture was heated at 55−80
°C for 8−93 h. The reaction mixture was diluted with water and
extracted with ethyl acetate. The organic solution was washed with a
saturated aqueous sodium chloride solution, then dried over sodium
sulfate, and evaporated to yield the expected product, which was used
in the next step without purification.
Method C is a copper coupling reaction. To a solution of
imidazolidine-2,4-dione in dimethylacetamide were added copper(I)
oxide and 4-bromo-2-(trifluoromethyl)benzonitrile. The mixture was
heated at 130−170 °C for 1−48 h. At rt, the mixture was diluted with
a 50% aqueous solution of ammonia and extracted with ethyl acetate.
The organic layer was dried over sodium sulfate, filtered, and
evaporated. The crude product was purified by chromatography over
silica gel to provide the expected product.
1
was used in the next step without purification. H NMR (400 MHz,
DMSO): δ 2.90 (s, 3H), 4.14−4.16 (m, 2H), 4.23 (d, J = 10.1 Hz,
1H), 4.45 (d, J = 10.1 Hz, 1H), 5.19 (d, J = 10.4 Hz, 1H), 5.25 (dd, J =
1.6 and 17.2 Hz, 1H), 5.88−5.98 (m, 1H), 7.48−7.55 (m, 5H), 8.04
(d, J = 8.6 Hz, 1H), 8.17 (s, 1H), 8.37 (d, J = 8.5 Hz, 1H). LC/MS (tR
= 3.08 min): m/z 430 (M + H)+.
Step 5: 4-[2,5-Dioxo-4-(hydroxymethyl)-3-methyl-4-phenylimida-
zolidin-1-yl]-2-(trifluoromethyl)benzonitrile (( )-18a). The title
compound was prepared according to method E from 16a (1.9 g,
4.42 mmol) and trifluoroborane−dimethyl sulfide complex (1.5 mL,
14.3 mmol) in dichloromethane (5 mL). The mixture was stirred at rt
for 4 h, and the crude product was purified by chromatography over
silica gel with 95/5 dichloromethane/ethyl acetate to give the final
Method D is an N-alkylation reaction. To a solution of N-
arylhydantoin or N-heteroarylhydantoin and potassium carbonate in
1
target 18a as a white crystalline solid (1.5 g, 83%). Mp = 160 °C. H
8244
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
NMR (400 MHz, DMSO): δ 2.88 (s, 3H), 4.09 (dd, J = 2.8 and 5.3
Hz, 1H), 4.44 (dd, J = 5.8 and 11.6 Hz, 1H), 5.80 (t, J = 5.5 Hz, 1H),
7.45−7.47 (m, 5H), 8.07 (d, J = 8.5 Hz, 1H), 8.20 (s, 1H), 8.35 (d, J =
8.4 Hz, 1H). 13C NMR (100 MHz, DMSO): δ 26.3, 61.3, 72.2, 107.4,
115.6, 118.6, 121.2, 123.9, 124.0, 126.7, 127.2, 129.4, 129.5, 129.9,
131.3, 131.6, 132.0, 132.3, 133.8, 136.8, 137.0, 154.7, 172.4. LCMS (tR
= 2.82 min): m/z 358 (M − CH2OH)−.
Separation of ( )-18a by Chiral HPLC. The two enantiomers of
( )-18a were separated from a racemic mixture (1.5 g) by
chromatography on Chiralcel OD (LC50 Prochrom column), eluting
with a 75/25 heptane/2-propanol mixture.15
separating, in the expression of molecular efficacy, the contribution of
the tissue from that of the compound: The molecular efficacy KA/KR
expresses the way the compound does or does not discriminate
between the two states, whereas L0 determines the contribution of the
tissue. Efficacies reported in this paper are the ratio KA/KR related to
the compound contribution. Applying the law of mass action to each
of the two states also leads to the potency equation 1/Kd = L0/(1 +
L0)(1/KR) + 1/(1 + L0)(1/KA), expressing that the apparent affinity is
simply the harmonic mean of the two molecular affinities, weighed by
the initial position of the equilibrium.
We therefore used a Schild-type curve-shift paradigm approach by
crossing a range of concentrations of our compound and a range of
concentrations of DHT as the reference agonist. We analyzed both the
rightward shift of the DHT curve induced by the antagonist activity of
our compound and the upward shift of its lower plateau induced by
the agonist activity of our compound. To reach the two molecular
affinities KA and KR of our SARM compounds, we built the
mathematical theoretical behavior of our system with parametrized
equations where KA and KR are the parameters identified by fitting the
bundle of curves from the curve-shift paradigm as a whole. From KA
and KR, potency and molecular efficacy were calculated as indicated.
In Vivo Pharmacology. The selective modulating activity of the
compounds was tested in an adapted model of castrated immature
young rats which is widely recognized for evaluating the anabolic and
androgenic effects of androgens on muscles and on genitalia.21b
Castrated rats were left untreated for one week and then treated for
four days with test compound. Castration resulted in a rapid atrophy
of the VP as well as of the anus-lifting LA muscle. These effects were
completely compensated by an exogenous administration of
androgens. Stringency of the therapeutic setting is greater than that
of the prophylactic setting and allows detection of more potent
myotrophic/anabolic compounds. After four days of treatment, the
activity of the compound was expressed as a percentage and calculated
using the following formula: (W/BW(treated) − W/BW(ORX))/(W/
BW(sham) − W/BW(ORX)) × 100, where W is weight and BW is
body weight. The dissociated activity of the test compound was
expressed as the ratio between the dose displaying 50% activity (A50)
on VP and that displaying 50% activity on LA.
(S)-(−)-4-[2,5-Dioxo-4-(hydroxymethyl)-3-methyl-4-phenylimida-
zolidin-1-yl]-2-(trifluoromethyl)benzonitrile ((S)-(−)-18a). The S
enantiomer was eluted first. By evaporating the solvent, (S)-(−)-18a
was obtained as an amorphous white solid (720 mg of 99.0% pure
compound, 48% yield). [α]2D3 = −40.8 (c 1, EtOH). HPLC (Chiralcel
OD, column 250 × 4.6 mm, heptane/2-propanol (75/25), flow rate 1
mL/min): tR = 9.01 min.
(R)-(+)-4-[2,5-Dioxo-4-(hydroxymethyl)-3-methyl-4-phenylimida-
zolidin-1-yl]-2-(trifluoromethyl)benzonitrile ((R)-(+)-18a). The R
enantiomer was eluted second. Further purification in the same
conditions followed by evaporation of the solvent provided (R)-
(+)-18a (700 mg of 98.1% pure compound, 46.5% yield) as an
amorphous white solid. [α]2D3 = +41.1 (c 1, EtOH). HPLC (Chiralcel
OD, column 250 × 4.6 mm, heptane/2-propanol (75/25), flow rate 1
mL/min): tR = 13.24 min.
Preparation of (S)-(−)-18a Using a Resolution Procedure. (S)-
(−)-4-[2,5-Dioxo-4-(hydroxymethyl)-4-phenylimidazolidin-1-yl]-2-
(trifluoromethyl)benzonitrile ((S)-(−)-31). To a solution of (S)-
(−)-α-(hydroxymethyl)phenylglycine hydrochloride (30)16 (0.54 g,
2.5 mmol) in NaOH (0.5 N, 16 mL, 3.2 mmol) was added a solution
of freshly prepared isocyanate (3 × 0.65 g, 9 mmol total) in dioxane (3
× 6 mL). The mixture was stirred at rt for 4 h, treated with HCl (12 N,
15 mL), and heated at 120 °C for 90 min. The mixture was
evaporated, extracted with ethyl acetate, dried over magnesium sulfate,
filtrated, evaporated, and purified on silica gel with 50/50 heptane/
ethyl acetate to give 31 (0.54 g, 57%). 1H NMR (400 MHz, CDCl3): δ
4.05 (d, J = 11.5 Hz, 1H), 4.43 (d, J = 11 Hz, 1H), 6.35 (s, 1H), 7.48−
7.53 (m, 3H), 7.64−7.66 (m, 2H), 7.93−8.00 (m, 2H), 8.12 (s, 1H).
[α]2D3 = −15.2 (c 1.13, MeOH).
Molecular Modeling. All molecular modeling studies were
performed with version 2.0 of the Discovery Studio (DS) platform,
available from Accelrys (San Diego, CA). Compounds were sketched,
prepared, and converted into 3D by use of the Ligand Preparation
protocol. In the objective of docking the prepared compounds, the
resolved crystal structure of AR LBD/(+)-4 described in our previous
work18 was protonated at pH 7.4, and hydrogens were reoptimized
with the HBUILT routine of CHARMm (CHARMm Momany and
Rone force field).22 Prior to removal, (+)-4 was used to define the
binding site volume using the cavity detection algorithm available
within the interface.23 Since the AR LBD exhibits a well-defined and
buried binding site, no particular edition of the resulting shape was
necessary. Docking was performed with LigandFit, resorting to an
energy grid calculated with Dreiding24 to interpolate the protein
energy fields. Poses were optimized as rigid bodies by 15 iterations of
steepest descent and the 10 best reoptimized by 150 iterations of the
Broyden−Fletcher−Goldfarb−Shanno (BFGS) minimization routine.
Other parameters remained the default values within DS. Lastly, poses
were flexibly minimized (DS “Smart minimizer” with a final 1000
iterations of conjugate gradient minimization) and rank ordered with
the scoring function LigScore2.25
(S)-(−)-4-[2,5-Dioxo-4-(hydroxymethyl)-3-methyl-4-phenylimida-
zolidin-1-yl]-2-(trifluoromethyl)benzonitrile ((S)-(−)-18a). To a
solution of (S)-(−)- 31 (188 mg, 0.5 mmol) and potassium carbonate
(193 mg, 1.4 mmol) in DMAc (5 mL) was added dimethyl sulfate
(160 μL, 1.7 mmol). The mixture was stirred at rt for 15 h, evaporated
to dryness, diluted with water, and extracted with ethyl acetate. The
organic layer was dried over magnesium sulfate, filtered, and
evaporated. The crude was purified by chromatography over silica
gel with 50/50 heptane/ethyl acetate to give the S enantiomer as an
amorphous white solid (140 mg, 72). [α]2D3 = −40.4 (c 1, EtOH).
In Vitro Transactivation Assay. The in vitro molecular efficacy
has been assessed using a host stable cell line derived from Hela cells
expressing hAR and a reporter gene placed under the transcriptional
control of the probasin AREs. This transactivation assay allows the
identification of pure agonists (such as testosterone and DHT), partial
agonists, and antagonists.
In the Monod−Wyman−Changeux model,19 the receptor is
constantly in interconversion equilibrium between two states (active,
A, and inactive, R), and any ligand of the receptor actually has two
molecular affinities for the receptor: one for the active state, A
(denoted KA), and one for the inactive state, R (denoted KR). The
ratio of the affinities of the compound for the two states, KA/KR, is
called the molecular efficacy of the compound. The position of the
equilibrium in the absence of ligand is defined by the ratio [R]/[A]
and is equal to a constant classically denoted L0, L[X] being the value of
this ratio in the presence of the ligand at a concentration [X]. Applying
the law of mass action to each of the states, [RX] = [R][X]/KR and
[AX] = [A][X]/KR, leads to the key expression L[X] = L0(1 + [X]/
KR)/(1 + [X]/KA). At saturating concentrations of the ligand, a limit
value is obtained: L∞ = L0KA/KR. This is precisely the equation
ASSOCIATED CONTENT
■
S
* Supporting Information
Synthetic procedures for the final targets ( )-9a,b, ( )-18b−j,
( )-19a,b, ( )-22a−d, ( )-23a−e, and ( )-29a−d. This
material is available free of charge via the Internet at http://
pubs.acs.org.
8245
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
(5) Long, Y. O.; Higushi, R. I.; Caferro, T. R.; Lau, T. L. S.; Wu, M.;
Cummings, M. L.; Martinborough, E. A.; Marshke, K. B.; Chang, W.
Y.; Lopez, F. J.; Karanewsky, D. S.; Zhi, L. Selective Androgen
Receptor Modulators Based on a Series of 7H-[1,4]Oxazino[3,2-
g]quinolin-7-ones with Improved in Vivo Activity. Bioorg. Med. Chem.
Lett. 2008, 18 (9), 2967−2971.
AUTHOR INFORMATION
Corresponding Author
*Phone: +33 149424687. Fax: +33 149424658. E-mail: pierre.
deprez@glpg.com.
■
Present Addresses
†Cistim, Minderbroedersstraat 12 Gebouw N, 3000 Leuven,
Belgium.
(6) Nique, F.; Hebbe, S.; Peixoto, C.; Annoot, D.; Lefranco
Duval, E.; Michoux, L.; Triballeau, N.; Lemoullec, J.-M.; Mollat, P.;
Thauvin, M.; Prange, T.; Minet, D.; Clement-Lacroix, P.; Robin-
Jagerschmidt, C.; Fleury, D.; Guedin, D.; Deprez, P. Discovery of
̧ is, J.-M.;
‡Institut de Recherche Servier, 125 chemin de ronde, 78290
Croissy-sur-Seine, France.
́
́
́
Diarylhydantoins as New Selective Androgen Receptor Modulators
(SARMs). J. Med. Chem. 2012, DOI: 10.1021/jm300249m.
(7) Mandagere, A. K.; Thompson, T. N.; Hwang, K. Graphical Model
for Estimating Oral Bioavailability of Drugs in Humans and Other
Species from Their Caco-2 Permeability and in Vitro Liver Enzyme
Metabolic Stability Rates. J. Med. Chem. 2002, 45, 304−311.
(8) N-Tosylindole-5-methyl ketone 5a and N-((2-(trimethylsilyl)
ethoxy)methyl)benzimidazole-5-methyl ketone 5b were prepared in
three or four steps from indole-5-carboxaldehyde or methyl 1H-
benzimidazole-5-carboxylate, and the syntheses are reported in the
Supporting Information.
(9) The synthesis of compounds 10a−i and 11a,b is described in the
Supporting Information. 1-Aryl-2-hydroxyethanones were directly
protected with allyl bromide to give 10a−c (in 33−63% yield).10 In
a two-step sequence, 1-aryl-2-bromoethanones provided 10d−h (in
23−53% yield over two steps). Finally, 2-(allyloxy)-1-m-tolylethanone
(10i) was synthesized in three steps starting from 3′-methylacetophe-
none (in 25% yield over three steps).11 1-Heteroaryl-2-(benzyloxy)
ethanones 11a,b were prepared in moderate yield (23−36%) from
(benzyloxy)acetyl chloride and heteroarylboronic acid by a palladium-
catalyzed coupling reaction.12
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The chiral chromatographic separation was efficiently per-
formed by Serge Droux and his team (Kyralia SAS, Parc
Biocitech, Romainville, France). We thank Kara Bortone for her
careful review of the manuscript and her advice.
ABBREVIATIONS USED
■
SARM, selective androgen receptor modulator; AR, androgen
receptor; ORX, orchidectomized; T, testosterone; TP,
testosterone propionate; DHT, 5α-dihydrotestosterone; SAR,
structure−activity relationship; PK, pharmacokinetic; TBS, tert-
butyldimethylsilyl; HPLC, high-pressure liquid chromatogra-
phy; AR LBD, ligand binding domain of the androgen receptor;
ARE, androgen responsive element; LA, levator ani muscle; VP,
ventral prostate
(10) Molander, G. A.; McKie, J. A. Stereochemical Investigations of
Samarium(II) Iodide-Promoted 5-Exo and 6-Exo Ketyl−Olefin
Radical Cyclization Reactions. J. Org. Chem. 1995, 60, 872−882.
(11) (a) Shue, H.-J.; Chen, X.; Schwerdt, J. H.; Paliwal, S.; Blythin, D.
J.; Lin, L.; Gu, D.; Wang, C; Reichard, G. A.; Wang, H.; Piwinski, J. J.;
Duffy, R. A.; Lachowicz, J. E.; Coffin, V. L.; Nomeir, A. A.; Morgan, C.
A.; Varty, G. B.; Shih, N.-Y. Cyclic Urea Derivatives as Potent NK1
Selective Antagonists. Part II: Effects of Fluoro and Benzylic Methyl
Substitutions. Bioorg. Med. Chem. Lett. 2006, 16, 1065−1069.
(b) Utsukihara, T.; Nakamura, H.; Watanabe, M.; Akira Horiuchi, C.
Microwave-Assisted Synthesis of α-Hydroxy Ketone and α-Diketone
and Pyrazine Derivatives from α-Halo and α,α′-Dibromo Ketone.
Tetrahedron Lett. 2006, 47, 9359−9364.
(12) Fujii, N.; Mallari, J. P.; Hansell, E. J.; Mackey, Z.; Doyle, P.;
Zhou, Y. M.; Gut, J.; Rosenthal, P. J.; McKerrow, J. H.; Kiplin Guy, R.
Discovery of Potent Thiosemicarbazone Inhibitors of Rhodesain and
Cruzain. Bioorg. Med. Chem. Lett. 2005, 15, 121−123.
(13) Commercially available from Apollo Scientific (CAS 6321-86-4)
or synthesis reported in the literature in three steps from 2-
oxophenylethyl acetate: Henze, H.; Winfred, C. The Interaction of
Certain α-(4-Morpholinyl)alkyl Aryl Ketones with Potassium Cyanide
and Ammonium Carbonate. J. Org. Chem. 1945, 10, 2−9.
(14) The synthesis of compound 25 is described in the Supporting
Information.
REFERENCES
■
(1) (a) Nieschlag, E. Testosterone Treatment Comes of Age: New
Options for Hypogonadal Men. Clin. Endocrinol. (Oxford, U. K.) 2006,
65, 275−281. (b) Brueggemier, R. W.; Miller, D. D.; Witiak, D. T. In
Principles of Medicinal Chemistry, 4th ed.; Foye, W. O., Lemke, T. L.,
Williams, D. A., Eds.; Williams and Wilkins: Baltimore, MD, 1995;
Chapter 23, pp 486−493. (c) Basaria, S.; Dobs, A. S. Risks versus
Benefits of Testosterone Therapy in Elderly Men. Drugs Aging 1999,
15, 131−142. (d) Rhoden, E. L.; Morgentaler, A. Risks of
Testosterone-Replacement Therapy and Recommendations for
Monitoring. N. Eng. J. Med. 2004, 350, 482−492. (e) Morgentaler,
A. Testosterone and Prostate Cancer: An Historical Perspective on a
Modern Myth. Eur. Urol. 2006, 50, 935−939. (f) Kicman, A. T.
Pharmacology of Anabolic Steroids. Br. J. Pharmacol. 2008, 154, 502−
521.
(2) (a) Narayanan, R.; Mohler, M. L.; Bohl, C. E.; Miller, D. D.;
Dalton, J. T. Selective Androgen Receptor Modulators in Preclinical
and Clinical Development. Nucl. Recept. Signaling 2008, 6 (1), e010.
(b) Jones, J. O. Improving Selective Androgen Receptor Modulator
Discovery and Preclinical Evaluation. Expert Opin. Drug Discovery
2009, 4 (9), 981−993.
(3) (a) Marhefka, C. A.; Gao, W.; Chung, K.; Kim, J.; He, Y.; Yin, D.;
Bohl, C.; Dalton, J. T.; Miller, D. D. Design, Synthesis, and Biological
Characterization of Metabolically Stable Selective Androgen Receptor
Modulators. J. Med. Chem. 2004, 47 (4), 993−998. (b) Kim, J.; Wu,
D.; Hwang, D. J.; Miller, D. D.; Dalton, J. T. The para Substituent of
S-3-(Phenoxy)-2-hydroxy-2-methyl-N-(4-nitro-3-trifluoromethyl-phe-
nyl)-propionamides Is a Major Structural Determinant of in Vivo
Disposition and Activity of Selective Androgen Receptor Modulators.
J. Pharmacol. Exp. Ther. 2005, 315 (1), 230−239.
(4) Ostowski, J.; Kuhns, J. E.; Lupisella, J. A.; Manfredi, M. C.;
Beehler, B. C.; Krystek, S. R.; Jr, Bi, Y.; Sun, C.; Seethala, R.; Golla, R.;
Sleph, P. G.; Fura, A.; An, Y.; Kish, K. F.; Sack, J. S.; Mookhtiar, K. A.;
Grover, G. J.; Hamann, L. G. Pharmacological and X-ray Structural
Characterization of a Novel Selective Androgen Receptor Modulator:
Potent Hyperanabolic Stimulation of Skeletal Muscle with Hypo-
stimulation of Prostate in Rats. Endocrinology 2007, 148 (1), 4−12.
(15) HPLC: Chiralcel OD, column 250 × 4.6 mm, heptane/2-
propanol (75/25), flow rate 1 mL/min.
(16) Olma, A. α-Hydroxymethylphenylglycine and α-Hydroxyme-
thylphenylalanine: Synthesis, Resolution and Absolute Configuration.
Pol. J. Chem. 1996, 70, 1442−1447.
(17) Garcia, H. G.; Kondev, J.; Orme, N.; Theriot, J. A.; Phillips, R.
Thermodynamics of Biological Processes. Methods Enzymol. 2011,
492, 27−59.
(18) Thauvin, M.; Robin-Jagerschmidt, C.; Nique, F.; Mollat, P.;
́
Fleury, D.; Prange, T. Crystallization and Preliminary X-ray Analysis of
the Human Androgen Receptor Ligand-Binding Domain with a
Coactivator-like Peptide and Selective Androgen Receptor Modu-
lators. Acta Crystallogr. 2008, F64, 1159−1162.
8246
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247

Journal of Medicinal Chemistry
Article
(19) Bohl, C. E.; Miller, D. D.; Chen, J.; Bell, C. E.; Dalton, J. T.
Structural Basis for Accommodation of Nonsteroidal Ligands in the
Androgen Receptor. J. Biol. Chem. 2005, 280 (45), 37747−37754.
(20) Brzozowski, A. M.; Pike, A. C.; Dauter, Z.; Hubbard, R. E.;
Bonn, T.; Engstrom, O.; Ohman, L.; Greene, G. L.; Gustafsson, J. A.;
̈
Carlquist, M. Molecular Basis of Agonism and Antagonism in the
Oestrogen Receptor. Nature 1997, Oct 16, 389 (6652), 753−758.
(21) (a) Eisenberg, E.; Gordan, G. S. The Levator Ani of the Rat as
an Index of Myotrophic Activity of Steroidal Hormones. J. Pharmacol.
Exp. Ther. 1950, 99 (1), 38−44. (b) Hershberger, L. G.; Shipley, E. G.;
Meyer, R. K. Myotrophic Activity of 19-Nortestosterone and Other
Steroids Determined by Modified Levator Ani Muscle Method. Proc.
Soc. Exp. Biol. Med. 1953, 83 (1), 175−180. (c) Kicman, A. T.
Pharmacology of Anabolic Steroids. Br. J. Pharmacol. 2008, 154 (3),
502−521.
(22) Momany, F. A.; Rone, R. J. Validation of the General Purpose
QUANTA 3.2/CHARMm Force Field. J. Comput. Chem. 1992, 13,
888−900.
(23) Venkatachalam, C. M.; Jiang, X.; Oldfield, T.; Waldman, M.
LigandFit: A Novel Method for the Shape-Directed Rapid Docking of
Ligands to Protein Active Sites. J. Mol. Graphics Modell. 2003, 21,
289−307.
(24) Mayo, S. L.; Olafson, B. D.; Goddard, W. A. DREIDING: A
Generic Force Field for Molecular Simulations. J. Phys. Chem. 1990,
94, 8897−8909.
(25) Krammer, A.; Kirchhoff, P. D.; Jiang, X.; Venkatachalam, C. M.;
Waldman, M. LigScore: A Novel Scoring Function for Predicting
Binding Affinities. J. Mol. Graphics Modell. 2005, 23, 395−407.
8247
dx.doi.org/10.1021/jm300281x | J. Med. Chem. 2012, 55, 8236−8247