Synthesis and biological evaluation of novel selective androgen receptor modulators (SARMs) Part III: Discovery of 4-(5-oxopyrrolidine-1-yl)benzonitrile derivative 2f as a clinical candidate

Article abstract of DOI:10.1016/j.bmc.2017.04.018

We previously reported that 4-(pyrrolidin-1-yl)benzonitrile derivative 1b was a selective androgen receptor modulator (SARM) that exhibited anabolic effects on organs such as muscles and the central nervous system (CNS), but neutral effects on the prostate. From further modification, we identified that 4-(5-oxopyrrolidine-1-yl)benzonitrile derivative 2a showed strong AR binding affinity with improved metabolic stabilities. Based on these results, we tried to enhance the AR agonistic activities by modifying the substituents of the 5-oxopyrrolidine ring. As a consequence, we found that 4-[(2S,3S)-2-ethyl-3-hydroxy-5-oxopyrrolidin-1-yl]-2-(trifluoromethyl)benzonitrile (2f) had ideal SARM profiles in Hershberger assay and sexual behavior induction assay. Furthermore, 2f showed good pharmacokinetic profiles in rats, dogs, monkeys, excellent nuclear selectivity and acceptable toxicological profiles. We also determined its binding mode by obtaining the co-crystal structures with AR.

Full text of DOI:10.1016/j.bmc.2017.04.018

Accepted Manuscript
Synthesis and biological evaluation of novel Selective Androgen Receptor
Modulators (SARMs) Part III: Discovery of 4-(5-oxopyrrolidine-1-yl) benzo-
nitrile derivative 2f as a clinical candidate
Katsuji Aikawa, Moriteru Asano, Koji Ono, Noriyuki Habuka, Jason Yano,
Keith Wilson, Hisashi Fujita, Hitoshi Kandori, Takahito Hara, Megumi
Morimoto, Takashi Santou, Masuo Yamaoka, Masaharu Nakayama, Atsushi
Hasuoka
PII:
S0968-0896(17)30321-8
DOI:
http://dx.doi.org/10.1016/j.bmc.2017.04.018
BMC 13688
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
20 February 2017
11 April 2017
12 April 2017
Please cite this article as: Aikawa, K., Asano, M., Ono, K., Habuka, N., Yano, J., Wilson, K., Fujita, H., Kandori,
H., Hara, T., Morimoto, M., Santou, T., Yamaoka, M., Nakayama, M., Hasuoka, A., Synthesis and biological
evaluation of novel Selective Androgen Receptor Modulators (SARMs) Part III: Discovery of 4-(5-
oxopyrrolidine-1-yl) benzonitrile derivative 2f as a clinical candidate, Bioorganic & Medicinal Chemistry (2017),
doi: http://dx.doi.org/10.1016/j.bmc.2017.04.018
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Synthesis and biological evaluation of novel Selective Androgen Receptor Modulators
SARMs) Part III: Discovery of 4-(5-oxopyrrolidine-1-yl) benzonitrile derivative 2f as a
(
clinical candidate
a*
a
a
a
b
c
a
Katsuji Aikawa, Moriteru Asano, Koji Ono, Noriyuki Habuka, Jason Yano, Keith Wilson,
a
a
a*
a
Hisashi Fujita, Hitoshi Kandori, Takahito Hara, Megumi Morimoto, Takashi Santou,
a
a
a*
Masuo Yamaoka, Masaharu Nakayama, and Atsushi Hasuoka
a
Pharmaceutical Research Division, Takeda Pharmaceutical Company Ltd., 26-1 Muraoka-higashi
2-chome, Fujisawa, Kanagawa 251-8555, Japan
b
Beryllium Discovery, 3 Preston Ct, Bedford, MA 01730
c
Structural Biology, Takeda San Diego, Inc., 10410 Science Center Drive, San Diego, California
92121
Abstract
We previously reported that 4-(pyrrolidin-1-yl)benzonitrile derivative 1b was a selective
androgen receptor modulator (SARM) that exhibited anabolic effects on organs such as muscles
and the central nervous system (CNS), but neutral effects on the prostate. From further
modification, we identified that 4-(5-oxopyrrolidine-1-yl)benzonitrile derivative 2a showed
strong AR binding affinity with improved metabolic stabilities. Based on these results, we
tried to enhance the AR agonistic activities by modifying the substituents of the
5-oxopyrrolidine
ring.
As
a
consequence,
we
found
that
4-[(2S,3S)-2-ethyl-3-hydroxy-5-oxopyrrolidin-1-yl]-2-(trifluoromethyl)benzonitrile (2f) had
1

ideal SARM profiles in Hershberger assay and sexual behavior induction assay. Furthermore,
f showed good pharmacokinetic profiles in rats, dogs, monkeys, excellent nuclear selectivity
2
and acceptable toxicological profiles. We also determined its binding mode by obtaining the
co-crystal structures with AR.
Key words: androgen receptor; AR; selective androgen receptor modulators (SARMs);
testosterone; Hershberger assay; sexual behavior;
*Corresponding author.
Tel: +81-466-32-1157
E-mail: katsuji.aikawa@takeda.com
Tel: +81-466-32-1928
E-mail: takahito.hara@takeda.com
Tel: +81-466-32-2913
E-mail: atsushi.hasuoka@takeda.com
2

1. Introduction
Steroid hormones are important chemical messengers which exert various physiological
1
3
4
–
effects.
Androgen, one of steroid hormones, is a general term of C19 steroid hormones.
Androgens increase skeletal muscles, promote osteogenesis, produce erythrocyte and maintain
libido through the interaction with androgen receptor (AR). Testosterone is one of the most
5
important androgens which is synthesized mainly from cholesterol in testis and adrenal gland.
In the prostate, 5α-reductase converts testosterone to dihydrotestosterone (DHT) with the
6
,7
strongest AR agonistic activity in all naturally occurring androgens.
Figure 1. Testosterone Synthesis
In the absence of ligand, AR is maintained in an inactive conformation by HSP70 and HSP
8
9
0 in the cytoplasm. The ligand binding causes the C-terminal helix 12 to shift its position to
close the ligand binding pocket. The conformational change elicits HSPs’ dissociation, leading
to homo-dimerization of AR. This facilitates a series of further conformational changes with
helix 3 and 5 serving as the key interfaces. The following dissociation of corepressor from the
complexes leads to nuclear entry and binding to androgen responsive elements in the promoter
of androgen responsive genes. Activated AR has been shown to recruit co-regulators and
general transcription factors, which triggers enhancement or repression of target genes
1
,6
transcription.
3

9
Testicular dysfunction and hypothalamic dysregulation with aging (late on set
1
0
11
12
hypogonadism, LOH) or cachexia which was caused by cancer or COPD, induces a decline
in serum testosterone levels, which is believed to be related to various symptoms such as muscle
1
3 15
–
and body weight loss, osteoporosis, depression and decreased libido.
Several recent
clinical studies have reported that androgen replacement therapy (ART) succeeded in increasing
1
6,17
lean body mass muscle and decreasing adipose tissue.
However, use of ART is limited
because of the concerns about potential side effects, for example, exacerbation of benign
prostatic hypertrophy (BPH) or latent prostate cancer. In addition, testosterone cannot be
administered orally due to its rapid hepatic elimination. Consequently, testosterone is
administered by less convenient methods, such as intramuscular injection, surgical implantation,
1
8
or transdermal delivery using patches or gels.
administration, however, hepatotoxicity limits their use in chronic therapy.
Under these situations, nonsteroidal and bioavailable tissue selective androgen receptor
Androgenic-anabolic steroids has allowed oral
1
9,20
1
1, 21 23
–
modulators (SARMs) have received a lot of attentions as new medicines for ART.
The
2
4
concept of SARMs was brought from selective estrogen receptor modulators (SERMs), which
have been clinically used for over 2 decades to supplement the diminishing circulating estrogens
for postmenopausal females. SARMs are expected to show anabolic effects on organs such as
muscles and the central nervous system (CNS) without influencing the prostate. Non-steroidal
SARMs have potential to be orally available and safer medicines, which provide advantages
over conventional ART.
2
5,26
Several non-steroidal SARMs, such as Ostarine (GTx/Merck),
BMS564929
2
7
28
(
Bristol-Myers Squibb), LGD2941 (Ligand), VK–5211 (Viking Therapeutic, structure not
2
9
30
31-33
disclosed) , GSK–2881078 and DT-200 (Akashi Therapeutics)
have been reported to be
investigated in clinical studies. Particularly, Ostarine has been proven to increase lean body
4

mass and improve some muscle functions for cancer cachexia patients in a Phase II study,
however, further development appears to have been halted for undisclosed reasons. Ostarine is
3
1
currently under development as a treatment of AR-positive triple-negative breast cancer.
Figure 2. Known nonsteroidal SARMs.
We have previously reported that we found 4-(pyrrolidin-1-yl)benzonitrile derivative 1b
which showed good SARM profiles in vivo, by the modification of substituents on pyrrolidine
3
4
and benzonitrile of compounds 1a.
In the exploration of this series of compounds, we
identified that 4-(5-oxopyrrolidine-1-yl)benzonitrile derivative (2a,b) showed strong AR
3
5
binding affinity with improved metabolic stabilities, although their agonistic activities were
poor (Figure 3). In order to enhance the agonistic activity of
-(5-oxopyrrolidine-1-yl)benzonitrile derivatives, we analyzed the co-crystal structure of
4
3
4
compound 1a in detail (Figure 4).
The X-ray information suggested that there are two small
and hydrophobic pockets around the pyrrolidine ring. One hydrophobic pocket created with
Trp741, Met742 and Met745 exists around the 2-position of the pyrrolidine ring of 1a. On the
other hand, the other pocket is constructed by Leu701, Met780, Leu873 and positioned close to
the 4-position of the pyrrolidine ring. Furthermore, the SAR study of this series indicated that
the introduction of alkyl group at 3-position of the pyrrolidine ring affects not only agonistic
5

34
activity but also pharmacokinetic profiles, leading to the discovery of compound 1b.
We
hypothesized that the binding mode of 2a is similar to that of compound 1a, and designed more
potent agonists than compound 2a by modifying the substituents on the 5-oxopyrrolidine ring
1
5
(
R –R ).
characterizations of 4-(5-oxopyrrolidine-1-yl)benzonitrile derivatives, and the discovery of
-[(2S,3S)-2-ethyl-3-hydroxy-5-oxopyrrolidin-1-yl]-2-(trifluoromethyl)benzonitrile 2f, which
In this paper, we will describe the design, synthesis, in vitro, in vivo
4
shows ideal SARM profiles.
Figure 3. The design of 4-(5-oxopyrrolidine-1-yl)benzonitrile derivatives with more potent
AR agonistic activities.
6

Leu701
Met780
Leu873
Met742
Trp741
Met745
Figure 4. X-ray information of compound 1a with the LBD of AR.
7

2. Chemistry
Synthesis of TBS protected pyrrolidine-5-one derivatives 4a is depicted in Scheme 1.
3
4
Previously reported (2S,3S)-3-hydroxy-2-methylpyrrolidine-5-one (3) was treated with TBSCl
and imidazole to afford 4a.
a
Scheme 1. Synthesis of compound 4a
a
Reagents and conditions: (a) TBSCl, imidazole, DMF, 74%
2
,3-Dimethylpyrrolidin-5-one analog 4b was synthesized from optically pure
3
4,36
(
2S,3S)-2,3-Dimethylpyrrolidin-3-ol 1/2 oxalate (5).
Boc protection of pyrrolidine 5 was
with cat.
afford 7 in 34% yield, and the successive de-protection gave 4b in 94%
carried out to give 6 in 72% yield. The oxidation of pyrrolidine ring by active RuO
RuO •H O and NaIO
yield (Scheme 2).
4
2
2
4
a
Scheme 2. Synthesis of 2,3-dimethylpyrrolidin-5-one analog 4b
a
2 2 2 4
Reagents and conditions: (a) Boc O, 1 M NaOH aq., THF, 72%; (b) cat. RuO /H O, NaIO ,
EtOAc/H
2
O, 34%; (c) 4 M HCl in EtOAc, 94%.
The preparation of 4c, 4d is shown in Scheme 3. N-Cbz-protected (S)-2-butanoic acid (8)
8

was activated with CDI, and the subsequent condensation with lithium enolate gave 9 in 62%
yield. Diastereo mixtures 10 were prepared by the ketone selective reduction of compound 9
followed by the protection of hydroxyl group using TBSOTf and 2,6-lutidine in 66% yield.
N-Cbz group was cleaved by hydrogenation with palladium on carbon under hydrogen
atmosphere.
purification by silica gel provided (2S,3S)-5-oxopyrrolidine derivative 4c in 13% and
2S,3R)-5-oxopyrrolidine derivative 4d in 59% yield, respectively (Scheme 3). The relative
Successive cyclization by NaOMe and the following chromatographic
(
3
7
configurations of compounds 4c, d were determined by X-ray and NOE analysis using 2d and
3
8
2
e, respectively (Figure 5).
a
Scheme 3. Synthesis of ethyl analog 4c,d
a
Reagents and conditions: (a) (1) CDI, THF, (2) n-BuLi, i-Pr
b) (1) NaBH , MeOH, (2) TBSOTf, 2,6-lutidine, THF, 66% for 2 steps; (c) (1) Pd/C, H
MeOH, (2) NaOMe, MeOH, (3) Chromatographic purification by silica gel, 13% (4c) and 59%
4d) for 3 steps.
A coupling reaction of pyrrolidine-5-ones 4a–d with 4’-halobenzonitriles 11 was
accomplished by Buchwald reaction using Pd (dba) , Xantphos, cesium carbonate, and the
subsequent de-protection of TBS group in the case of 4a, c, gave
2
NH, EtOAc, THF, 62% for 2 steps;
(
4
2
,
(
2
3
d
4’-(5-oxopyrrolidine-1-yl)benzonitrile derivatives 2a–h in 6–87% yield (Scheme 4).
a
Scheme 4. Synthesis of compounds 2a–h by Buchwald reaction
9

a
Reagents and conditions: (a) Pd
2
(dba)
3
, Xantphos, Cs
2
3
CO , 1,4-dioxane; (b) HCl aq., EtOH,
THF, 33–86%.
3
7
38
X-ray analysis of Compound 2d
NOE analysis of Comound 2e
Figure 5. X-ray and NOE analysis of compound 2d and 2e to determine the relative structure.
4
The preparation of 4-alkyl (R ) pyrrolidine-5-one analog 2i–l was performed by dilithiation
of 2a and 2d followed by the selective alkylation in 5–42% yield. The stereoselectivity was
caused by steric hindrance of hydroxyl group. The relative configuration of 2i was determined
3
8
by NOE analysis (Figure 6).
4
a
Scheme 5. Synthesis of 4-alkyl (R ) pyrrolidine-5-one analog 2g–i
10

a
4
4
Reagents and conditions: (a) n-BuLi, i-Pr NH, THF then R Br or R I, 5–42%
2
3
8
Figure 6. NOE analysis of compound 2i to determine the relative structure.
4
,4-Dimethyl pyrrolidine-5-one analog 2m was synthesized according to scheme 6.
Activation of Cbz-protected amino acids 12 with CDI followed by the condensation with the
lithium enolate afforded 13 in 46% yield. Subsequent dimethylation gave 14 in 85% yield, and
the following deprotection induced cyclization to produce 15 in 87% yield. Coupling reaction
of 16 with 2-chloro-4-bromobenzonitrile under Buchwald reaction conditions gave 18 in 53%
yield. Cis-selective reduction of 16 with L-Selectride® and the following chiral HPLC
purification provided optical pure 2m in 32% yield (99.9% ee) and 2m’ in 34% yield (99.5%
3
8
ee). The relative configuration of 2m was determined by NOE analysis, and the absolute
3
9
configuration was assigned based on their binding and agonistic activities.
a
Scheme 6. Synthesis of 4-dimethyl analog 2m
a
2
Reagents and conditions: (a) (1) CDI, THF, (2) n-BuLi, i-Pr NH, EtOAc, THF, 46%; (b) MeI,
11

K
2
CO
Xantphos, Cs
2% for 2m and 34% for 2m’, respectively..
3
, acetone, 85%; (c) Pd/C, H
2
2 3
, MeOH, 87%; (d) 4-bromo-2-chlorobenzonitrile, Pd (dba) ,
®
2
CO
3
, 1,4-dioxane, 53%; (e) (1) 1.0 M L-Selectride (2) Chiral HPLC purification,
3
3
8
Figure 7. NOE analysis of compound 2m to determine the relative structure.
,4-Difluoro-pyrrolidin-5-one analog 2n was synthesized via Reformatsky reaction as a
key reaction (Scheme 7). 3-Chloro-1-iodobenzene was condensed with
S)-2-aminopropan-1-ol (17) by Ullmann-Goldberg reaction with CuI as a catalyst to afford 18,
4
(
which was converted to TBS protected 19 using TBSCl and imidazole. Boc protection of
nitrogen atom followed by the selective TBS deprotection with TBAF gave 20 in 64% yield.
Following Swern oxidation of primary alcohol 20 provided aldehyde 21 in 96% yield.
Reformatsky reaction was carried out for 21 and ethyl 2-bromo-2,2-difluoroacetate with Zn
successfully afforded secondary alcohol 22a and 22b as diastereo mixtures. Deprotection of
Boc group followed by the intramolecular lactamization gave carbamate product 23 and lactam
24 in 38% and 21% yield, respectively. Hydrolysis of desired isomer 23 with aqueous sodium
hydroxide was performed to provide carboxylic acid 25, which was quickly cyclized to give
lactam 28 in 97% yield in 2 steps. Iodination of 26 with NIS under acidic condition followed
by cyanation using CuCN successfully afforded 4,4-difluoro-5-oxopyrrolidine analog 2n.
4
0
Relative configuration of 2n was determined by X-ray crystal analysis, and high optical purity
4
1,42
of 2n was confirmed by chiral HPLC analysis (99.8% ee).
12

a
Scheme 7. Synthesis of 4,4-difluoro-pyrrolidin-5-one analog 2n
a
Reagents and conditions: (a) 1-Chloro-3-iodobenzene, CuI, (CH
2
OH)
7%; (b) TBSCl, imidazole, DMF, 50 °C, 97%; (c) (1) n-BuLi, THF, −78 °C; (2) Boc
8 °C to room temperature; (3) TBAF, THF, 64% for 3steps; (d) (COCl) , DMSO, Et N, THF, −
0 °C to −20 °C, 96%; (e) Br(F) CCO Et, Zn, THF, reflux; (f) (1) 4 M HCl in EtOAc; (2)
NEt, THF, reflux, 38% for 3 steps (21% for 27); (g) 8 M NaOH aq., MeOH, reflux; (h)
MeOH, reflux, 97% for 2 steps; (i) (1) NIS, concentrated H SO , AcOH; (2) CuCN, NMP,
50 °C; (3) recrystallization from EtOAc/n-hexane, 41% for 3 steps (99.8% ee).
2
, K
3
PO
4
, i-PrOH, 80 °C,
9
7
6
2
O, −
2
3
2
2
i-Pr
2
2
4
1
4
0
Figure 8. X-ray analysis of compound 2n.
13

1
Further modifications of R at 2-position of pyrrolidine-5-one were performed as shown in
Scheme 8. Condensation of amino acids 27a, b with 2-chloro-4-fluorobenzonitrile 28 afforded
29a, b in 72 and 100% yield, respectively. Treatment of 29a, b with Meldrum's acid, and the
successive cyclization by refluxing in EtOAc afforded compounds 30a, b in 32 and44%,
respectively. 4-(5-Oxopyrrolidine-1-yl)benzonitrile derivatives 2o, p were synthesized by
cis-selective reduction using NaBH
4
and AcOH in 59 and 62% yield, respectively. The
relative configuration of 2o was confirmed by X-ray analysis, and that of 2p was determined
4
3
based on the configuration of 2o (Figure 9).
1
a
Scheme 8. Synthesis of various 3-alkyl (R ) pyrrolidine-5-one analog 2o, p
a
Reagents and conditions: (a) Cs
2
3
CO , DMF, 72–100%; (b) (1) Meldrum's acid, CDI, DMAP,
THF, (2) EtOAc, reflux, 32–44% for 2 steps; (c) NaBH
4
, AcOH, CH CN, 59–62%.
3
4
3
Figure 9. X-ray analysis of compound 2o to determine the relative structure.
Stereo selective synthesis of (2S,3S)-5-oxopyrrolidine derivative 2f was performed for in
14

vivo study, because the separation of the compound 2c and 2d was quite difficult by silica gel
column chromatography. Activation of N-Cbz-protected (S)-2-butanoic acid (31) with
triphosgene followed by the cyclization gave 32 in 58% yield. Highly reactive compound 32
was condensed with Medrum's acid and subsequent cyclization afforded compound 33 in 83%
yield. Cis-selective reduction of 33 was performed with sodium tetrahydroborate-AcOH.
Following protection of hydroxyl group with TBS and the subsequent cleavage of Cbz provided
4c in 19% yield (Scheme 9). Finally, bulk synthesis of compound 2f was completed according
to a similar synthetic procedure with Scheme 4. The optical purity of the compound 2f was
4
4,45
confirmed by HPLC analysis.
a
Scheme 9. Cis-selective synthesis of ethyl analog 2f
a
Reagents and conditions: (a) triphosgene, THF, 50 °C, 89%; (b) benzyl chloroformate,
N-methylmorpholine, 0 °C, 65%; (c) (1) Meldrum's acid, Et
steps; (d) (1) NaBH , AcOH, quant.; (2) TBSCl, imidazole, DMF, (3) Pd/C, H
for 3 steps.
3
N, THF, (2) EtOAc, reflux, 83% for
2
4
2
, MeOH, 45%
15

3. Results and discussion
3.1. AR binding and reporter gene assay
3
AR binding affinities were evaluated by competitive displacement of a radiolabeled [ H]
mibolerone from AR, and the data were reported as IC50 values. Functional activities were
determined by luciferase activities and described as EC50 values.
Initially, we evaluated the effects of the substituents at 4-position of the 5-oxopyrrolidine
4
ring (Table 1). Amazingly, just the introduction of methyl group (2i) at trans-position (R )
enhanced the binding affinity by 10 times (IC50 = 1.3 nM), and agonistic activities by 800 times
(EC50 = 1.0 nM) compared with 2a. Ethyl group (2j) also enhanced the activity, however,
benzyl group (2k) did not contribute to increasing the potency. 4,4-Dimethyl (2m, EC50 = 30
nM) and 4,4-difluoro derivatives (2n, EC50 = 12 nM) showed medium agonistic activities.
These results indicates that there is a small and lipophilic pocket around the 4-position,
especially trans, of the 5-oxopyrrolidine ring. Contrary to our expectations based on the SAR
obtained in the series of compound 1b, installation of methyl group at the 3-position of the
2
34
5
-oxopyrrolidine ring (R ) was not effective to improve agonistic activity (2c, EC50 = 150 nM).
Table 1. Binding inhibitory and agonistic activities of 4-(5-oxopyrrolidine-1-yl)benzonitrile
derivatives.
5
2
R R
R4
4
OH
3
O
N
Cl
CN
4
5
2
a
a
Compd.
R
R
R
AR binding
AR reporter
16

b
b
IC50 (nM)
EC50 (nM)
2
2
2
2
2
2
2
a
H
H
H
H
H
H
H
H
7.8 (5.7-11)
1.3 (1.2-1.4)
3.2 (2.6-4.1)
6.0 (5.1-7.0)
14 (11-17)
810 (500-1300)
1.0 (0.70-1.5)
3.6 (2.2-5.9)
220 (160-290)
30 (19-47)
i
Me
Et
Bn
Me
F
H
j
H
k
m
n
c
H
Me
F
1.1 (0.90-1.3)
12 (7.0-19)
H
H
Me 15 (12-18)
150 (59-370)
a
b
Human AR was used. IC50 and EC50 values are presented as means of duplicate experiments,
with 95% confidence intervals (95%CI) in parentheses.
1
We moved on to the modification of R at the 2-position of the 5-oxopyrrolidine ring
(Table 2). Surprisingly, replacement of a methyl group (2a) with an ethyl group (2d) led to a
130-fold increase in the agonistic activity (EC50 = 6.0 nM). On the other hand, an isobutyl
group was not effective to enhance agonistic activity (2o, EC50 = 170 nM). Introduction of a
polar cyanomethyl group (2p) resulted in significant loss of agonistic activity. These findings
suggest that the space around the 2-position is small and hydrophobic. Based on these results,
we performed additional modification of compound 2d. As expected, compound 2l with
trans-4-methyl group showed greatly strong agonistic activity (EC50 = 0.39 nM). (R)-OH
group also increased the agonistic activity (2e, EC50 = 8.0 nM), which was a similar trend shown
4
6
in the series of naphthonitrile compounds.
Table 2. Binding inhibitory and agonistic activities of 4-(5-oxopyrrolidine-1-yl)benzonitrile
derivatives.
17

a
a
AR binding
AR reporter
1
4
2
3
Compd.
R
R
R
R
b
b
IC50 (nM)
EC50 (nM)
2
2
2
2
2
2
a
d
o
p
l
Me
Et
H
H
OH
OH
OH
OH
OH
H
7.8 (5.7-11)
3.9 (2.7-5.6)
5.5 (4.7-6.5)
14 (10-20)
810 (500-1300)
6.0 (4.3-8.4)
170 (69-410)
>10000
H
H
i-Bu
H
H
CH
2
CN
H
H
Et
Me
H
H
1.8 (1.6-2.1)
3.4 (2.6-4.5)
0.39 (0.29-0.52)
8.0 (5.3-12)
e
Et
OH
a
b
Human AR was used. IC50 and EC50 values are presented as means of duplicate experiments,
with 95% confidence intervals (95%CI) in parentheses.
Then, we examined the effects of substituents on the benzonitrile (R). 2’-Trifluomethyl
derivative 2f showed highly potent agonistic activity (EC50 = 4.7 nM), comparable with 2d.
2’-Methoxy (2g, EC50 = 24 nM) and 2’-chloro-3’-methyl (2h, EC50 = 23 nM) groups resulted in
decrease of agonistic activity.
Table 3. Binding inhibitory and agonistic activities of 4-(5-oxopyrrolidine-1-yl)benzonitrile
derivatives.
18

a
a
AR binding
AR reporter
Compd.
R
b
b
IC50 (nM)
EC50 (nM)
2
2
2
2
d
2’-Cl
3.9 (2.7-5.6)
3.6 (2.5-5.0)
9.2 (6.8-13)
9.0 (6.9-12)
6.0 (4.3-8.4)
4.7 (3.3-6.8)
24 (18-31)
23 (16-34)
f
2’-CF
3
g
h
2’-OMe
2’-Cl-3’-Me
a
b
Human AR was used. IC50 and EC50 values are presented as means of duplicate experiments,
with 95% confidence intervals (95%CI) in parentheses.
3.2. Tissue selectivity in in vivo study
4
7
Tissue selectivity was investigated using the Hershberger assay.
In this assay, the
weight of the levator ani muscle was used as an indicator of anabolic effects, and the weights of
the prostate was used as an indicator of androgenic activity. The compounds (2.5 mg/kg, bid,
po) were administered to immature castrated male SD rats (4 weeks old) for 4 days. To
extrapolate this rat assay to a human study, testosterone propionate (TP, 0.5 mg/kg/day, sc, qd)
was administered along with the test compounds to complement adrenal testosterone, which is
not produced in rats. In this type of study, TP was reported to have equivalent potency and
efficacy for the prostate and levator ani muscle when dosed subcutaneously.
Compound 2g with weak agonistic activity (EC50 = 24 nM) did not show any effects in this
assay. On the other hand, another weak agonist 2c (EC50 = 150 nM) reduced the weight of
19

prostate. Compound 2c showed extremely good pharmacokinetic profiles, which would be the
4
8
reason for the potent efficacy in this assay.
This result indicates that weak agonists can show
antagonistic effects on prostate if the enough amount of compound was delivered. Compounds
with strong agonistic activities, 2n (EC50 = 12 nM), 2d (EC50 = 6.0 nM), 2e (EC50 = 8.0 nM) and
2f (EC50 = 4.7 nM), demonstrated tissue-selective in vivo pharmacological activity. These
compounds increased the weight of the levator ani muscle up to about 130% relative to that in
TP-treated vehicle rats. On the other hand, these 4 compounds did not influence the weight of
the prostate. In contrast, the highly strong agonist 2l (EC50 = 0.39 nM) increased levator ani
muscle and prostate, indicating that this compound works as just an agonist to the both tissues..
Through this in vivo study we confirmed that compound 2n,d,e,f were SARM having tissue
selective pharmacological activity to levator ani and prostate.
The mechanism for the tissue selectivity was not clearly elucidated, however, there are
some publications that shows the ability to recruit cofactors will be different between SARMs
4
9,50
and pure agonist, such as testosterone.
Table 4. Tissue selectivity in vivo (3-week old immature rats, 2,5 mg/kg, bid, po, 4-days
treatment).
a
a
c
Compd.
AR binding
AR reporter
Tissue Selectivity (%)
b
b
IC50 (nM)
EC50 (nM)
Levator ani
114
Prostate
90
2g
2c
2n
2d
2e
2f
9.2 (6.8-13)
15 (12-18)
24 (18-31)
150 (59-370)
12 (7.0-19)
6.0 (4.3-8.4)
8.0 (5.3-12)
4.7 (3.3-6.8)
116
76*
85
1.1 (0.90-1.3)
3.9 (2.7-5.6)
3.4 (2.6-4.5)
3.6 (2.5-5.0)
139*
134*
94
124*
84
129*
106
20

2l
1.8 (1.6-2.1)
0.39 (0.29-0.52) 161*
149*
a
b
Human AR was used. IC50 and EC50 values are presented as means of duplicate experiments,
c
with 95% confidence intervals (95%CI) in parentheses. The values shown are the values
*
compared with the control group (TP, 0.5 mg/kg/day, sc, qd). p ≤ 0.05 (Student’s t-test) when
compared with the TP (0.5 mg/kg)-treated control group (n=5).
We selected 2f for the confirmation of dose dependency in vivo. Compound 2f (0.15, 0.50, 1.5,
5.0, 15 mg/kg/day, qd, po) was administered to 8-weeks-old castrated male SD rats for 7 days
along with TP (0.4 mg/kg/day, qd, sc). Compound 2f exerted significant effects on levator ani
muscle at the minimal dose (0.15 mg/kg/day), and the efficacy increased in a dose dependent
manner, without influencing the weight of prostate even at the highest dose (15 mg/kg/day).
Based on this study, we chose compound 2f for further development.
Figure 10. Dose dependency for tissue selectivity in vivo (8-weeks-old SD rats, 7 days
treatment).
*
p ≤ 0.05 (Dunnett test) when compared with the TP (0.4 mg/kg)-treated control group (n = 5).
#
p ≤ 0.05 (T-test) when compared with the TP (0.4 mg/kg)-treated control group
21

3.3. Sexual behavior induction assay
To confirm the agonistic activity on the CNS, a sexual behavior induction assay was
performed (Table 5). Selected fertile male rats were castrated to eliminate sexual behavior.
After treatment with 2f for 3 weeks, sexual behavior induction was confirmed by the
pseudopregnancy rate in female rats. Compounds 2f induced sexual behavior at the minimal
dose (0.5 mg/kg/day, qd, po). These results proved that 2f could act as androgen agonist on
the CNS.
Table 5. Sexual behavior induction assay
Dose (mg/kg, qd, for 21 days)
Compound
a
a
a
b
0
0
/
0.5
1.5
5.0
3.0
/
2
f
100%*
/
100%*
/
86%*
/
TP (3 mg/kg, sc)
100%*
a
b
*
po sc induction rate of pseudopregnancy (% of total number), p ≤ 0.05 (Fisher's exact
test-test) when compared with the vehicle-treated control group (n = 7).
3.4. PK profiles
Pharmacokinetic profiles in rats, dogs and monkeys were evaluated, and the results are
shown in Table 6. Compound 2f displayed good oral bioavailability (B.A.) in all species [rats
74.6%), dogs (76.6%) and monkeys (48.1%)]. In addition, 2f was found to show good
metabolic stabilities with little species difference between rat, dog, mouse, monkey and human
Table 7). These data suggest that PK profiles of 2f correlate with its metabolic stabilities in
(
(
vitro, which indicates that compound 2f is expected to show good PK profiles also in human.
a
Table 6. Pharmacokinetic parameters of 2f
Animal
iv (0.1 mg/kg)
22

C
5min (ng/mL)
AUC0-24h (ng•h/mL)
MRT (h)
2.32
Vd(ss) (mL/kg)
2167
CLtotal (mL/h/kg)
b
rat
70.8
108.5
150.5
118.3
934
681
853
c
dog
106.2
4.84
3297
d
monkey
72.9
1.72
1469
po (1 mg/kg)
Animal
C
max (ng/mL)
218.1
T
max (h)
0.50
0.75
1.67
AUC0-24h (ng·h/mL)
809.4
MRT (h)
3.06
B.A. (%)
74.6
b
rat
c
dog
264.9
1159.5
4.34
76.6
d
monkey
71.2
557.8
5.21
48.1
a
b
c
Data are expressed as the mean ± S.D. (n = 3). Crl:CD(SD) rat (male). Beagle dog (male).
d
Cynomolgus monkey (male).
Table 7. Metabolic stabilities of compound 2f.
Metabolic stability
Animal
(
µL/min/mg)
rat
21
4
dog
mouse
monkey
human
3
5
1
3.5. Nuclear receptor selectivity
Nuclear receptor selectivity against MR (Mineralocorticoid Receptor), GR (Glucocorticoid
Receptor), ER (Estrogen Receptor), PR (Progesterone Receptor) was examined, and the results
are shown in Table 8. These data indicate that 2f is highly specific for androgen receptor.
Table 8. Nuclear receptor selectivity of 2f
a
b
a
b
Binding (IC50 , nM)
MR GR
>10000 >10000 >10000
360 540 7300
Reporter (EC50 , nM)
Compound
AR
6.3
ER
AR
7.6
3.0
PR
2
f
>10000
8200
Dihidrotestosterone
0.67
a
b
Human AR was used (n = 2). IC50 and EC50 values shown are the mean values of duplicate
23

measurements (n = 2).
3.6. Toxicological study
Two-week oral gavage toxicity studies were conducted in rats, dogs and monkeys. All the
findings are related to the expected pharmacology except for convulsion and myocardial
necrosis in monkeys at the plasma exposure more than 200 times higher than that at efficacy
dose level. No effects on electrocardiography, heart rate and mean blood pressure were noted
in anesthetized beagle dogs after the single intravenous dose at supra-pharmacologic exposures.
5
1,52
The true CV risks will be evaluated in the clinical study.
3.7. Co-crystal structure of 2f with LBD of androgen receptor.
To determine the binding mode, we solved the X-ray co-crystal structure of compound 2f
with AR (Figure 11). The co-crystal structure of 2f with AR ligand binding domain, residues
71–919, was obtained at 1.4 Å resolutions. This data supports our hypothesis that there are
6
small and hydrophobic pockets around the 2- and 4-position of the 5-oxopyrrolidine ring.
Trp741, Met742, Met745 constructs a small and lipophilic pocket around 2-position, and
Leu701, Met780, Leu873 constituted the pocket around 4-positon. Especially, the pocket
around 4-position exists in trans direction, which supports the high potency of trans-4-methyl
compounds 2i,l. The binding mode of 2f is almost the same as that of previously reported
3
4
compound 1a and 1b.
hydrogen bonding on both ends of the molecule. It is stabilized by water mediated HBs
hydrogen bonds) to the side chain of Arg752 and the backbone carbonyl of Met745. The
Binding of the compound 2f binding with AR is stabilized by
(
other end of the molecule forms HBs between the hydroxyl group on the 5-oxopyrrolidine ring
and the side chain oxygen atom of Asn705. Compound 2f does not form a direct HB to
24

Thr877, even though it potentially could with an alternate rotamer.
Leu7 01 1
Asn705
Leu701
Thr877
Met780
Leu873
Trp741
Met742
Met745
Arg752
Figure 11. Co-crystal structure of 2f with AR.
25

4. Conclusion
With the aim of developing SARMs, we modified the substituents on the 5-oxopyrrolidine
ring of compound 2a considering the co-crystal structure of compound 1a with AR. We
enhanced the agonistic activities more than 100 times by modification of 2a. As a result of
further investigation, we found three types of compounds with different tissue selectivity: (1)
Antagonistic effect on prostate (2b), (2) Agonistic effect on muscle and no effect on prostate
(2n,d,e,f), (3) Agonistic effects on both muscle and prostate (2l). Of those compounds, the
compound 2f showed good tissue selectivity in a dose dependent manner In addition,
compound 2f induced sexual behavior in castrated rats, which suggests that the compound could
also act as an agonist on the CNS. Compound 2f showed good pharmacokinetic profiles in
vivo as well as in vitro metabolic stabilities with a little species difference, which indicates that
compound 2f is expected to show good PK profiles in human. Compound 2f also has highly
specific nuclear receptor selectivity and acceptable toxicological profiles. Finally, our
hypothesis to improve the agonistic activity was confirmed by co-crystal structure of 2f with
LBD of AR. Further exploratory studies using 2f are under way, and the results will be
reported in due course.
26

5
. Experimental section
.1. Chemistry
5
Melting points were determined on a Büchi melting point apparatus and were not corrected.
1
Proton nuclear magnetic resonance ( H NMR) spectra were recorded on a Varian Gemini-200
(200 MHz), a Varian Mercury-300 (300 MHz), a Bruker DPX300 (300 MHz) or a Bruker
Avance II+ 600 (600 MHz) instrument. Chemical shifts are reported as δ values (ppm)
downfield from internal tetramethylsilane of the indicated solution. Peak multiplicities are
expressed as follows: s, singlet; d, doublet; t, triplet; dd, doublet of doublet; dt, doublet of
triplet; br s, broad singlet; m, multiplet. Coupling constants (J values) are given in hertz (Hz).
Elemental analyses were performed by Takeda Analytical Laboratories. Reaction progress was
determined by thin layer chromatography (TLC) analysis on silica gel 60 F254 plates (Merck) or
NH TLC plates (Fuji Silysia Chemical Ltd.). Chromatographic purification was performed on
silica gel columns 60 (0.063–0.200 mm or 0.040–0.063 mm, Merck), basic silica gel
(
ChromatorexNH, 100–200 mesh, Fuji Silysia Chemical Ltd.), or Purif-Pack (SI 60 µM or NH
0 µM, Fuji Silysia, Ltd.). Commercial reagents and solvents were used without additional
purification. Compounds 2a–p were confirmed to be pure by elemental analysis. The yields
were not optimized. The following abbreviations are used. Boc O = di-tert-butyl dicarbonate,
CDI N,N'-carbonyldiimidazole, DIEA diisopropylethylamine, DMAP
-dimethylaminopyridine, DMF = N,N-dimethylformamide, DMSO = dimethyl sulfoxide, Et
6
2
=
=
=
4
3
N
=
triethylamine, Et
2
O = diethyl ether, EtOAc = ethyl acetate, EtOH = ethanol, IPE = diisopropyl
®
ether,
L-Selectride
=
lithium
tri-s-butylhydroborate,
Meldrum's
acid
=
2
,2-dimethyl-1,3-dioxane-4,6-dione, MeOH = methanol, NIS = N-iodosuccinimide, NMP =
N-methylpyrrolidone, Pd (dba) tris(dibenzylideneacetone)dipalladium(0), i-PrOH =
2
3
=
27

isopropanol, TBAF = tetrabutylammonium fluoride, TBSCl = tert-butyldimethylsilyl chloride,
THF = tetrahydrofuran, Xantphos = 4,5-bis(diphenylphosphino)-9,9-dimethylxanthene.
(2S,3S)-3-{[tert-Butyl(dimethyl)silyl]oxy}-2-methylpyrrolidin-5-one (4a). To a solution of
(2S,3S)-3-hydroxy-2-methylpyrrolidin-5-one (3) (14.8 g, 129 mmol) in THF (50 mL), imidazole
(10.5 g, 154 mmol) and tert-butyldimethylsilyl chloride (21.3 g, 141 mmol) was added at 0 °C
and the mixture was stirred at room temperature for 18 h. Then the mixture was poured into
water and extracted with EtOAc. The organic layer was washed with water, brine, dried over
4
MgSO and concentrated in vacuo. The residue was purified using silica gel column
chromatography (n-hexane:EtOAc = 2:1 to 0:1) to give 4a (21.9 g, 95.5 mmol, 74%) as a
1
colorless solid. H-NMR (300 MHz, CDCl
3
) δ 0.07 (6H, s), 0.90 (9H, s), 1.17 (3H, d, J = 6.3
Hz), 2.27 (1H, dd, J = 16.5, 4.2 Hz), 2.52 (1H, dd, J = 16.5, 9.3 Hz), 3.70-3.80 (1H, m),
.37-4.45 (1H, m), 6.00 (1H, br s).
4
(2S,3S)-tert-Butyl 3-hydroxy-2,3-dimethylpyrrolidine-1-carboxylate (6). To a solution of
(2S,3S)-2,3-dimethylpyrrolidin-3-ol 1/2 oxalate (5) (3.31 g, 20.7 mmol) in THF (50 mL) were
2
added 1 M aqueous NaOH solution (40 mL, 40 mmol) and a solution of Boc O (5.42 g, 24.8
mmol) in THF (10 mL) at 0 °C. After stirring at room temperature for 20 h, water (50 mL)
was added to the reaction mixture at room temperature. The resulting mixture was extracted
with EtOAc twice. The combined organic layers were washed with water and brine, dried over
4
MgSO , and concentrated in vacuo. The residue was purified using silica gel column
chromatography (n-hexane: EtOAc = 9:1 to 3:2) to give 6 (3.21 g, 14.9 mmol, 72%) as a white
powder. Furthermore, the recrystallizaion from EtOAc/n-hexane gave 6 (2.43 g, 11.4 mmol,
1
5
5%) as colorless prisms. Mp 109–112 °C. H NMR (300 MHz, CDCl
3
) δ 1.21 (3H, d, J =
28

6.4 Hz), 1.33 (3H, s), 1.46 (9H, s), 1.66 (1H, s), 1.73–1.85 (1H, m), 1.86–1.98 (1H, m), 3.40
(2H, dd, J = 7.2, 6.7 Hz), 3.46–3.57 (1H, m). IR (KBr) 3430, 2965, 2936, 2878, 1688, 1472,
-
1
1
402, 1338, 1252, 1216, 1170, 1091, 1019, 988, 934, 919 cm . Anal. Calcd for C11
3
H21NO : C,
61.37; H, 9.83; N, 6.51. Found: C, 61.46; H, 9.76; N, 6.46.
(
2S,3S)-tert-Butyl 3-hydroxy-2,3-dimethyl-5-oxopyrrolidine-1-carboxylate (7).
solution of 6 (3.05 g, 14.2 mmol) in EtOAc (45 mL) and H O (68 mL) were added RuO
566 mg, 4.25 mmol) and NaIO (4.57 g, 21.3 mmol) at room temperature. After stirring at
To a
2
2
·H
2
O
(
4
room temperature for 1.5 days, the organic layer was separated, and the aqueous layer was
extracted with EtOAc. To the combined organic layers was added i-PrOH (0.7 mL). The
4
resulting mixture was dried over MgSO , and concentrated in vacuo. The residue was purified
using silica gel column chromatography (n-hexane: EtOAc = 9:1 to 3:2) to give 7 (1.11 g, 4.83
mmol, 34%) as a white powder. Furthermore, the recrystallizaion from EtOAc/n-hexane gave
1
7
as colorless prisms. Mp 122–125 °C. H NMR (300 MHz, CDCl
3
) δ 1.34 (3H, d, J = 6.4
Hz), 1.46 (3H, s), 1.54 (9H, s), 1.72 (1H, s), 2.47 (1H, d, J = 17.0 Hz), 2.73 (1H, d, J = 17.0 H),
3.91 (1H, q, J = 6.4 Hz); IR (KBr) 3465, 2970, 2935, 2874, 1773, 1680, 1539, 1473, 1457, 1370,
-1
1
339, 1287, 1257, 1161, 1083, 1053, 1035, 1024, 1005, 966, 935 cm . Anal. Calcd for
11 4
C H19NO : C, 57.62; H, 8.35; N, 6.11. Found: C, 57.61; H, 8.33; N, 5.93.
(2S,3S)-3-Hydroxy-2,3-dimethylpyrrolidin-5-one (4b). A mixture of 7 (1.00 g, 4.36 mmol)
and 4 M HCl solution in EtOAc (10 mL, 40 mmol) was stirred at room temperature for 14 h.
Then, the mixture was concentrated in vacuo. The residue was purified by recrystallization
form i-PrOH/n-hexane to give 4b (309 mg, 2.40 mmol, 55%) as pale red prisms. The filtrate
was concentrated in vacuo. The residue was purified using silica gel column chromatography
29