Synthesis and structure–activity relationships of 1-benzylindane derivatives as selective agonists for estrogen receptor beta

Article abstract of DOI:10.1016/j.bmc.2016.09.047

The estrogen receptor beta (ERβ) selective agonist is considered a promising candidate for the treatment of estrogen deficiency symptoms in ERβ-expressing tissues, without the risk of breast cancer, and multiple classes of compounds have been reported as

Full text of DOI:10.1016/j.bmc.2016.09.047

PII:
S0968-0896(16)30759-3
DOI:
http://dx.doi.org/10.1016/j.bmc.2016.09.047
BMC 13300
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
12 August 2016
17 September 2016
19 September 2016
Please cite this article as: Yonekubo, S., Fushimi, N., Miyagi, T., Nakanishi, O., Katsuno, K., Ozawa, M., Handa,
C., Furuya, N., Muranaka, H., Synthesis and Structure–Activity Relationships of 1-Benzylindane Derivatives as
Selective Agonists for Estrogen Receptor Beta, Bioorganic & Medicinal Chemistry (2016), doi: http://dx.doi.org/
1
0.1016/j.bmc.2016.09.047
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 Structure–Activity Relationships of 1-Benzylindane
Derivatives as Selective Agonists for Estrogen Receptor Beta
Shigeru Yonekubo*, Nobuhiko Fushimi, Takashi Miyagi, Osamu Nakanishi, Kenji
Katsuno, Motoyasu Ozawa, Chiaki Handa, Noritaka Furuya, and Hideyuki Muranaka
Central Research Laboratory, Kissei Pharmaceutical Co., Ltd., 4365-1,
Hotakakashiwabara, Azumino, Nagano 399-8304, Japan
*
Corresponding author.
Tel: +81-263-82-8820; Fax: +81-263-82-8827.
E-mail: shigeru_yonekubo@pharm.kissei.co.jp
Keywords: Estrogen receptor, ERβ, ERβ ligand, ERβ agonist
Abstract
The estrogen receptor beta (ERβ) selective agonist is considered a promising
candidate for the treatment of estrogen deficiency symptoms in ERβ-expressing tissues,
-
Page 1 of 88 -

without the risk of breast cancer, and multiple classes of compounds have been reported
as ERβ selective agonists. Among them, 6-6 bicyclic ring-containing structures (e.g.,
isoflavone phytoestrogens) are regarded as one of the cyclized analogues of isobutestrol
5
b, and suggest that other cyclized scaffolds comprising 5-6 bicyclic rings could also
act as selective ERβ ligands.
In this study, we evaluated the selective ERβ agonistic activity of
-(4-hydroxybenzyl)indan-5-ol 7a and studied structure–activity relationship (SAR) of
1
its derivatives. Some functional groups improved the properties of 7a; introduction of a
nitrile group on the indane-1-position resulted in higher selectivity for ERβ (12a), and
further substitution with a fluoro or a methyl group to the pendant phenyl ring was also
preferable (12b, d, and e). Subsequent chiral resolution of 12a identified that R-12a has
a superior profile over S-12a. This is comparable to diarylpropionitrile (DPN) 5c, one of
the promising selective ERβ agonists and indicates that this indane-based scaffold has
the potential to provide better ERβ agonistic probes.
1
. Introduction
Estrogens play a crucial role in the development, maintenance, and functioning
of the reproductive system and nonreproductive tissues such as cardiovascular,
-
Page 2 of 88 -

1
,2
musculoskeletal, immune, and central nervous systems in both males and females.
While estrogens can be beneficial for health, the proliferative effect of estrogens may
3
,4
increase the risk of cancer in the breast and uterus. Hormone replacement therapy
HRT) reduces many symptoms such as hot flashes, sweating, and bone loss in
postmenopausal women; however, long-term use of estrogens increases the risk of
(
5
,6
breast cancer.
The multiple actions of estrogens are exerted through two estrogen receptors,
7
,8
9,10
ERα and ERβ, which have different patterns of tissue distribution and biological
1
1
regulation. The ERα, expressed at high levels in the uterus and breast is now
considered to involve a proliferative effect and can bring about malignant growth in
1
2,13
these tissues, whereas ERβ has an antiproliferative effect on breast cancer cells.
Thus, an ERβ selective ligand is considered to exert beneficial effects on
ERβ-expressing tissues, such as the prostate, colon, and brain, without the risk of breast
1
4,15
cancer.
Although ERα and ERβ share less than 60% sequence homology in the
ligand-binding domain (LBD), the ligand-binding pockets (LBPs) of the two isoforms
have a difference of only two amino acids; Leu384 and Met421 in ERα are substituted
-
Page 3 of 88 -

1
6
with Met336 and Ile373 in ERβ, respectively. In fact, 17β-estradiol (E2, an
endogenous estrogen) binds to both ERs without selectivity.
Estrogens
Stilbestrol estrogens
Phytoestrogens
OH
OH
HO
OH
O
B
2 OH
O
OH
1
1
C
D
H
C
A
B
H
H
A
H
3
7
O
OH
HO
HO
HO
HO
Estradiol (E2)
1a, Diethylstilbestrol (DES) 1b, Hexestrol
2a, Genistein
2b, S-Equol
Benzoxazoles
N
Benzopyrans
Bibenzyl-diol derivatives
Ring B seco-estradiols
OH
OH
H
OH
OH
OH
N
H
C
D
F
O
O
O
H
R
A
H
N
HO
5
HO
HO
HO
HO
3a, ERB-041
3b, WAY-292
4, SERBA-1
5a, R = H
6
5b, R = Et (Isobutestrol)
5c, R = CN (DPN)
Figure 1. Estradiol and ERβ ligands.
To date, multiple classes of compounds have been reported as ERβ selective
1
7,18
agonists.
Among them, 1,2-bis(4-hydroxyphenyl)ethane 5a is one of the common
templates for ER ligands, and the exact motif or its oxygen-containing analogues are
incorporated into many scaffolds (Fig. 1). The representative phytoestrogen genistein
1
9
20
21
2
a, daidzein enteric metabolite S-equol 2b, and synthetic ligands ERB-041 3a,
2
1
22
WAY-292 3b, and SERBA-1 4, are relatively rigid molecules comprising fused-ring
frameworks. On the other hand, ERβ ligands with more structural flexibility, such as
bibenzyl-diol derivatives 5a–c, are also known. Diarylpropionitrile (DPN) 5c is an
-
Page 4 of 88 -

2
3
attractive ligand with a 72-fold binding selectivity relative to E2, and isobutestrol 5b
with a nonpolar ethyl group instead of the cyano group in DPN 5c also exerts an 18-fold
2
4
relative binding selectivity for ERβ.
distal Phenyl ring
X
X
OH
HO
X = O, C
OH
Ethyl group
well-known scaffold
e.g., 2a,b, 4 in Fig. 1)
(
fused with
HO
5
b, Isobutestrol
proximal Phenyl ring
OH
HO
7a
Figure 2. Compounds of interest.
As illustrated in Fig. 2, the cyclized scaffold of 5b fused between the ethyl group
and the distal phenyl ring provides well-known ERβ selective ligands with a 6-6
bicyclic ring (e.g., compounds 2a, 2b, and 4). Fusing the ethyl group of 5b to the
proximal phenyl ring affords a 1-benzylindane framework, and it is also expected to be
a promising scaffold as ERβ selective ligand; however, little is known about the
estrogenic actions of such compounds. In a previous report,
1
-(4-hydroxybenzyl)indan-5-ol 7a with a 5-6 bicyclic ring may have exhibited
2
5
antigonadotropic activity and a weak estrogenic effect; however, the actions for ERs
-
Page 5 of 88 -

and ERβ selectivity remain unidentified. Therefore, we were intrigued by the possibility
of a 1-benzylindane scaffold as novel ERβ selective ligands, and investigated whether
7
a and its derivatives have selective ERβ agonistic activity.
Herein, we report that 7a is an attractive template for ERβ selective agonists and
disclose the SARs of the 1-benzylindane derivatives, some of which have excellent
agonistic activities and high selectivities for ERβ in terms of binding affinities and
transcriptional activities.
2
. Chemistry
The synthesis of racemic compounds evaluated in this article is illustrated in
Schemes 1–6. Hydroxy-substituted indane derivatives and ring-expanded analogues
bearing benzyl groups were prepared from carboxylic acid 14 or ketones 16c,d, 17, and
1
8. Compounds 7a,b were obtained by reduction of the carbonyl group with Et SiH and
3
subsequent demethylation of the methoxy groups by BBr for the benzoyl derivatives
3
1
5a,b, prepared by addition of the corresponding phenyl Grignard reagents to the
Weinreb amide obtained from 14 (Scheme 1). When using ketones as starting materials,
catalytic hydrogenation of the benzylic hydroxy groups generated by addition of benzyl
-
Page 6 of 88 -

Grignard reagent, followed by demethylation, afforded compounds 7c,d, 8, and 9
(
Scheme 2).
O
O
O
OH
O
a,b
c,d
HO
O
HO
1
4
15a (4-OMe)
7a (4-OH)
7b (3-OH)
1
5b(3-OMe)
Scheme 1. Reagents and conditions: (a) MeONHMe·HCl, EDCI·HCl, HOBt·H
2
O, Et N,
3
CH
2
Cl
2
, 0 °C to rt; (b) 4-(or 3-)MeO–PhMgBr, THF, 0 °C to rt; (c) Et
3
SiH, TFA, 0 °C
to rt; (d) BBr
3
, CH
2
Cl
2
, −78 °C to rt.
O
O
OH
a,b
c
O
(
)n
( )n
( )n
O
HO
n=1 16c(4-OMe) n=2 17 (6-OMe)
6d(6-OMe) n=3 18 (2-OMe)
n=1: 19c,19d
n=2: 20
n=1: 7c,7d
n=2: 8
1
n=3: 21
n=3: 9
Scheme 2. Reagents and conditions: (a) 4-MeO–BnMgCl, THF, Toluene, 0 °C to rt; (b)
2
Pd–C, H , MeOH, THF, rt; (c) BBr
3
, CH
2
Cl , −78 °C to rt.
2
Using the above intermediate 15a, the ethyl-substituted derivative 10 was
obtained via Wittig olefination and catalytic hydrogenation of the double bond,
followed by demethylation (Scheme 3).
-
Page 7 of 88 -

a,b
O
OH
c
1
5a
O
HO
2
2
10
+
−
Scheme 3. Reagents and conditions: (a) EtP Ph
3
Br , n-BuLi, THF, 0 °C to rt; (b) Pd–C,
H
2
, MeOH, THF, rt; (c) BBr
3
, CH
2
Cl , −78 °C to rt.
2
The synthesis of 1-alkylated-1-benzylindane derivatives 11a–d was achieved by
alkylation of cyanide 23 with alkyl iodides in the presence of NaH and subsequent
conversion of the nitrile groups into 4-methoxybenzyl groups before demethylation. The
nitrile groups of the alkylated intermediates were reduced into aldehydes 24a–d using
DIBAL-H and then reacted with lithiated anisole, followed by removal of the resulting
hydroxy group as described above (Scheme 4).
1
R
1
R
1
R
O
O
OH
N
O
a,b
O
c,d
e
H
O
HO
1
1
1
2
3
24a (R =Me)
25a (R =Me)
11a(R =Me)
1
1
1
2
2
2
4b(R =Et)
25b(R =Et)
11b(R =Et)
1
1
1
4c(R =n-Pr)
25c(R =n-Pr)
11c(R =n-Pr)
1
1
1
4d(R =n-Bu)
25d(R =n-Bu)
11d(R =n-Bu)
1
Scheme 4. Reagents and conditions: (a) NaH, R -I, DMF, 0 °C to rt; (b) DIBAL-H,
-
Page 8 of 88 -

Toluene, −78 °C, then 2 M HCl aq., rt; (c) 4-Br-anisole, n-BuLi, THF, −78 °C to rt; (d)
2
Pd–C, H , MeOH, THF, rt; (e) BBr
3
, CH
2
Cl , −78 °C to rt.
2
The 1-cyano derivatives 12a–g were more easily prepared by introduction of
various benzyl groups onto the nitrile-substituted carbon under modified conditions of
the above alkylation and subsequent demethylation (Scheme 5).
N
N
2
R
2
a
O
b
OH
R 2
2
3
3
2
2
R
3
O
HO
Cl(or Br)
2
7a-g
2
2
O
12a (R =H)
12e (R =3-F)
2
2
2
2
12b(R =2-F) 12f (R =3-Cl)
2
2
2
2
6a (R =H)
26e (R =3-F)
2
2
2
2
12c(R =2-Cl) 12g(R =3-Me)
6b(R =2-F) 26f (R =3-Cl)
2
2
2
12d(R =2-Me)
6c(R =2-Cl) 26g(R =3-Me)
2
6d(R =2-Me)
Scheme 5. Reagents and conditions: (a) 26a–g, LDA, THF, −78 °C to rt; (b) BBr ,
3
CH
2
Cl , −78 °C to rt.
2
Homologation of the aldehyde obtained by reduction of the above intermediate
7a was achieved by acidic hydrolysis of the methyl enol ether prepared by the Wittig
reaction, and conversion of the formyl group into a nitrile group by heating with
2
-
Page 9 of 88 -

hydroxylamine hydrochloride, followed by demethylation of the 1-cyanomethyl product
2
9, afforded compound 13 (Scheme 6).
O
N
N
a,b,c
d
O
e
OH
H
O
2
7a
O
HO
O
2
8
29
13
Scheme 6. Reagents and conditions: (a) DIBAL-H, Toluene, −78 °C, then 2 M HCl aq.,
+
−
rt; (b) MeOCH
2
P Ph Cl , n-BuLi, THF, 0 °C to rt; (c) HCl, THF, 75 °C; (d)
3
HO–NH ·HCl, NMP, 130 °C; (e) Pyridine·HCl, 220 °C.
2
Preparations of each enantiomer of the key compounds 7a and 12a are described
in Schemes 7 and 8. In the synthesis of the 1-unsubstituted-1-benzylindane enantiomers
S-7a and R-7a, the corresponding chiral benzoyl intermediates R-15a and S-15a were
prepared, respectively. Chiral resolution was achieved by flash column chromatography
on silica gel of the diastereomixture of chiral oxazolidinone derivatives R,R-30 and
R,S-30, obtained by acylation of the oxazolidinone nitrogen with the acid chloride of 14
using n-BuLi as a base. In each separated diastereomer, the other isomer was not
detected by NMR spectroscopy. After removal of the chiral auxiliary using lithium
hydrogen peroxide, S-7a and R-7a were synthesized in a manner similar to that of
-
Page 10 of 88 -

racemate 7a. The optical purity of each enantiomer was confirmed to be >98% ee by
high performance liquid chromatography (HPLC) analysis (Scheme 7).
O
O
O
O
a
1
4
+
N
N
O
O
O
O
R,R-30
R,S-30
O
OH
O
e,f
b,c,d
R,R-30
R,S-30)
(
HO
O
R-15a
S-15a)
S-7a
(R-7a)
(
Scheme 7. Reagents and conditions: (a) (i) (COCl)
2
, CH
2
Cl , rt; (ii)
2
(
R)-4-Ph-oxazolidin-2-one, n-BuLi, THF, −78 °C to rt; (b) LiOH·H
2
O, H
2
O
2
, THF, H O,
2
0
4
°C; (c) MeONHMe·HCl, EDCI·HCl, HOBt·H
2
O, Et
3
N, CH
2
Cl
2
, 0 °C to rt; (d)
, CH Cl , −78 °C to
-MeO–PhMgBr, THF, 0 °C; (e) Et SiH, TFA, 0 °C to rt; (f) BBr
3
3
2
2
rt.
Chiral 1-cyano derivatives R-12a and S-12a were also synthesized from 14. For
resolution, (R)-oxazolidinone chiral auxiliary was similarly introduced into
1
1
-benzylindane-1-carboxylic acid 31 as described above, in which 31 was prepared by
-benzylation of the methyl ester of 14 and subsequent hydrolysis. Although the two
-
Page 11 of 88 -

diastereomers R,R-32 and R,S-32 were not separable by flash column chromatography,
pure R,R-32 was fortunately obtained as a single diastereomer by recrystallization from
ethyl acetate/n-hexane. After removal of the chiral auxiliary, the resulting chiral
carboxylic acid was converted into cyanide via carboxamide in a general manner,
followed by demethylation to provide R-12a. In the synthesis of S-12a,
(
S)-oxazolidinone was used instead of the (R)-isomer, as a chiral auxiliary. In each
enantiomer obtained by these methods, the other enantiomer was not detected by HPLC
analysis (Scheme 8).
O
N
HO
O
N
O
d
e,f,g,h
OH
a,b,c
O
O
1
4
O
HO
O
3
1
O
R,R-32
S,S-32)
R-12a
(S-12a)
(
Scheme 8. Reagents and conditions: (a) H
2
SO , MeOH, 70 °C; (b) 4-MeO–BnCl, LDA,
4
THF, −78 °C to rt; (c) LiOH·H
2
O, THF, MeOH, H
2
O, rt; (d) (i) (COCl)
2
, CH
2
Cl , rt; (ii)
2
(
R) (or (S))-4-Ph-oxazolin-2-one, n-BuLi, THF, −78 °C to rt; (iii) recrystallization; (e)
LiOH·H
2
O, H
2
2
O , THF, H
2
O, 0 °C; (f) CDI, THF, rt, then NH
3
aq., rt; (g) TFAA, Et N,
3
CH
2
Cl
2
, 0 °C to rt; (h) BBr
3
, CH
2
Cl , −78 °C to rt.
2
-
Page 12 of 88 -

The absolute configurations of R-7a, S-7a, R-12a, and S-12a were determined
2
6
based on their corresponding intermediates.
3
. Results and Discussion
The relative binding affinities (RBAs) of test compounds for ERs were
determined by a fluorescence polarization displacement assay using commercial
estrogen receptor competitor assay kits. The correlations of the RBA values and
selectivities of the reference compounds 5a–c listed in Table 1 were comparable to the
2
7
literature values measured in competitive radiometric binding assays. To further study
the agonistic properties, the transcriptional activities were evaluated by a cell-based
reporter gene assay in HEK293T cells transiently transfected with either ERα or ERβ
expression plasmid, the estrogen responsive element (ERE)-driven luciferase plasmid,
and the internal control plasmid. These results are presented in Tables 1–3.
To prove the hypothesis that 7a might be an ERβ agonist, we evaluated the
binding affinities with ERs. Expectedly, 7a showed high RBA for ERβ. The RBA value
of 67.8% on ERβ observed in 7a was higher than DPN 5c, the selectivity for ERβ
(
6.8-fold) was also higher than isobutestrol 5b, and subsequent assessment of the
transcriptional activities revealed that 7a was an ERβ selective agonist (Table 1).
-
Page 13 of 88 -

Table 1. Binding affinities and transcriptional activities for ERs and ERβ selectivities of
1
-benzylindane derivative 7a and related compounds.
6
OH
Y
5
3
X
4
4
(
)n
HO
7
a-d
8 (n=1), 9 (n=2)
a
d
RBA (%)
Transcriptional acitivity EC50 (nM)
c
Ligand
X
Y
Log P
b
e
hERα
hERβ
β/α
hERα
hERβ
Rel.β/α
E2
-
-
-
-
-
-
-
-
100
100
1
3.75
3.90
4.63
3.18
0.35 ± 0.03 0.04 ± 0.01
1
f
5a
0.9 ± 0.2
27.2 ± 11.6
1.6 ± 0.9
11.7 ± 1.6
100 ± 14
37.0 ± 11.1
13
3.7
23
47%
7.12 ± 0.53
0.23 ± 0.01
-
5b
185 ± 17
101
372
5c
2,390 ± 89 0.82 ± 0.07
1
3
1
2
,170 ±
65
7
a
4-OH 5-OH
3-OH 5-OH
4-OH 4-OH
4-OH 6-OH
9.9 ± 3.6
3.1 ± 1.5
5.6 ± 2.0
4.1 ± 1.2
67.8 ± 9.4
11.6 ± 0.6
17.3 ± 5.9
21.3 ± 4.3
6.8
3.7
3.1
5.2
4.24
4.24
4.24
4.24
1.14 ± 0.21
32.5 ± 7.2
72.4 ± 29.7
25.9 ± 7.1
131
12
3
2
,090 ±
35
7b
5
,830 ±
01
7
c
1
,110 ±
98
7d
10
4
8
-
-
-
-
35.4 ± 20.7
62.6 ± 28.4
91.6 ± 38.6 2.6
4.68
5.13
102 ± 30
123 ± 45
0.18 ± 0.03
0.15 ± 0.05
71
9
98.9 ± 9.8
1.6
103
-
Page 14 of 88 -

a
Relative binding affinity (RBA) values were calculated by IC50[E2]/IC50[test
compound] × 100 (RBA of E2 = 100), shown as mean ± SD of three or four
independent experiments, performed in duplicate.
b
β/α values = ERβ-RBA/ERα-RBA (ERβ selectivity of binding affinity).
c
Log P values were calculated using Marvin 6. 2. 0 (ChemAxon).
d
EC50 values are shown as Mean ± SD of three independent experiments, performed in
duplicate.
e
Rel.β/α
values
=
{
ERβ-EC50[E2]/ERβ-EC50[compound]}/{ERα-EC50[E2]/ERα-EC50[compound]} (ERβ
selectivity of agonistic activity).
f
The value is the reporter activation at 10 µM, normalized by the maximum response of
E2, which is set as 100%.
Because two hydroxy groups at both ends of many ER ligands would play an
important role in achieving high binding affinities and transcriptional activities by
interacting with the Glu-Arg-water triad (Glu353/Arg394 in ERα, Glu305/Arg346 in
1
7
ERβ) and His (His524 in ERα, His475 in ERβ), we next evaluated the optimal position
of each hydroxy group of 7a. By moving the hydroxy group on the pendant phenyl ring
-
Page 15 of 88 -

of 7a from the 4-position to the 3-position (7a vs 7b), a 6-fold decrease in RBA for ERβ
was observed. On the indane ring, 4-hydroxy and 6-hydroxy analogues also showed
4
5
-fold (7c) and 3-fold (7d) diminished RBAs for ERβ, compared to substitution in the
-position, respectively. Although reduced RBAs for ERα were also observed in
compounds 7b–d, the resulting selectivities for ERβ were decreased. Additionally,
remarkable losses in agonistic activities and ERβ selectivities were observed in the
reporter gene assay. Thus, each hydroxy group in the original position of 7a was
considered to be optimal, indicating that these groups are involved in forming favorable
hydrogen bonds with the receptor, like those of other ER agonists. Among the two
2
8,29
hydrogen bonds observed in the complex structures of E2 with ERs,
the interaction
of the estradiol-3-hydroxy group with the Glu-Arg-water triad is considered to be
effective for maintaining the affinity of E2. More profound effects on the affinities
observed in 4-hydroxy group on the pendant phenyl ring suggested that this group faced
toward Glu-Arg-water triad in ERs. Further, the estimated distance between two
3
0
16
hydroxy groups of 7a (12.0 Å) was comparable to that of genistein 2a (12.1 Å),
suggesting that the hydroxy group on the indane ring is able to form a hydrogen bond
with the His475, supported by the alignment of 2a in the X-ray crystal structure
1
6
complexed with ERβ.
-
Page 16 of 88 -

To determine the optimum size of the bicyclic ring, we next evaluated
ring-expanded analogues with 6- and 7-membered aliphatic rings (8, 9). The
improvement of the RBA value on ERβ was observed in both analogues, but loss of
selectivity was observed because of the higher effect on ERα. Since the volume of the
3
1
LBP of ERα is proposed to be larger than that of ERβ, these ring-expanded analogues
were assumed to be more efficiently filling the larger cavity of ERα.
Thus, 7a was determined to be an optimal template for exploring substituent
effects. We next attempted to introduce functional groups onto the linker carbons
between the two phenyl rings. The results of the ring-expanded analogues described
above suggested that modifications at the 2- and 3-positions of the indane ring might
increase ERα affinities; therefore, substitution at these positions was avoided. By
introduction of an ethyl group into the methylene carbon of the 1-benzyl group, one of
the highest RBA values of 122% on ERβ in this study was observed with compound 10;
however, the selectivity was almost abrogated by about a 10-fold increase in the RBA
for ERα (Table 2). A similar SAR is known in the case of stilbene compounds.
Diethylstilbestrol (DES) 1a has two ethyl groups filling in two regions of ERα-LBP not
3
2
occupied by E2, located at the 7α and 11β positions of E2, and removal of one of the
ethyl groups results in reduction of binding affinities, whereas its ERβ selectivity is
-
Page 17 of 88 -

2
7
improved. A similar relationship has been reported between hexestrol 1b and
2
7
isobutestrol 5b. Compound 10 is a cyclic analogue of hexestrol 1b formed by
connecting one ethyl group to its proximal phenyl ring. The 5-membered aliphatic ring
of indane might partially play a role as one ethyl group of hexestrol 1b; thus, 10
exhibited the highest RBA for both ERs with a remarkable loss in selectivity.
Table 2. Binding affinities and transcriptional activities for ERs and ERβ selectivities of
1
-benzylindane derivatives with substitutions on linker carbons and the pendant phenyl
group.
1
R
2
R
OH
2
OH
3
HO
HO
1
0
11a-d, 12a-g, 13
a
d
RBA (%)
Transcriptional acitivity EC50 (nM)
1
2
c
Ligand
R
R
Log P
b
e
hERα
hERβ
β/α
hERα
hERβ
Rel.β/α
1
0
-
-
88.9 ± 39.7
17.0 ± 3.2
122 ± 21
1.4
4.97
4.54
4.98
5.43
6.33 ± 0.80 0.10 ± 0.01
8
11a
Me
Et
H
H
H
90.2 ± 38.7 5.3
481 ± 101
116 ± 12
169 ± 6
0.43 ± 0.17
0.36 ± 0.05
0.53 ± 0.06
144
41
11b
32.9 ± 15.9 82.6 ± 11.7
2.5
2.0
11c
n-Pr
50.9 ± 22.9
100 ± 19
40
-
Page 18 of 88 -

f
1
1
1
1d
2a
2b
n-Bu
CN
CN
CN
CN
CN
CN
CN
H
H
13.0 ± 7.7
1.9 ± 1.1
12.6 ± 5.6
72.4 ± 35.4
4.8 ± 1.9
3.5 ± 1.5
1.9 ± 0.6
0.7 ± 0.2
4.4 ± 2.5
42.0 ± 10.3 3.2
5.87
3.79
3.93
4.39
4.30
3.93
4.39
4.30
3.72
42%
6.59 ± 1.39
-
36.3 ± 16.0
88.9 ± 4.5
130 ± 29
19
7.1
1.8
13
2,790 ± 152 1.65 ± 0.29
216
217
106
253
256
-
2-F
2-Cl
2-Me
3-F
3-Cl
3-Me
H
256 ± 29
44.4 ± 3.8
950 ± 52
0.15 ± 0.03
0.05 ± 0.02
0.48 ± 0.23
1
2c
12d
64.7 ± 15.1
32.1 ± 5.3
31.7 ± 11.6
11.2 ± 1.2
16.3 ± 4.8
1
2e
9.3
17
1,820 ± 433 0.91 ± 0.24
f
1
2f
7%
11.6 ± 1.8
10.7 ± 3.9
3.05 ± 0.17
f
1
2g
16
2%
-
1
3
2
CH CN
3.7
635 ± 57
27
a
Relative binding affinity (RBA) values were calculated by IC50[E2]/IC50[test
compound] × 100 (RBA of E2 = 100), shown as mean ± SD of three or four
independent experiments, performed in duplicate.
b
β/α values = ERβ-RBA/ERα-RBA (ERβ selectivity of binding affinity).
c
Log P values were calculated using Marvin 6. 2. 0 (ChemAxon).
d
EC50 values are shown as Mean ± SD of three independent experiments, performed in
duplicate.
e
Rel.β/α
values
=
{
ERβ-EC50[E2]/ERβ-EC50[compound]}/{ERα-EC50[E2]/ERα-EC50[compound]} (ERβ
-
Page 19 of 88 -

selectivity of agonistic activity).
f
The value is the reporter activation at 3 µM, normalized by the maximum response of
E2, which is set as 100%
In contrast, introduction of a methyl group on the indane-1-position enhanced
the binding affinity for ERβ with a slight loss of selectivity (7a vs 11a), and this result
prompted us to examine the effects of the alkyl chain length. Increasing the size of the
alkyl substituent at the indane-1-position from methyl to ethyl (11b) and propyl (11c)
maintained a high RBA for ERβ however at the expense of selectivity. Further
extension to butyl (11d) caused reduction of the RBA values on both ERs. This
difference in substituent effects on ERs was probably due to the slight difference in the
cavity size around this position. In addition to the spatial acceptance of alkyl
substitution at this position, the pharmacophore model constructed from indazole-type
3
3
ligands suggested that analogues substituted with polar or polarizable groups instead
of alkyl groups could result in a more selective ERβ ligand, and therefore we next
examined 1-cyano derivative 12a and its homologue 13.
As expected, introduction of a nitrile group resulted in an improvement in ERβ
selectivity due to the marked loss of ERα affinity. Cyanide 12a showed improved
-
Page 20 of 88 -

selectivity for ERβ compared to unsubstituted indane 7a, even though its ERβ affinity
was lowered. High ERβ selectivities of some nitrile-containing ligands are explained by
a slight difference in interactions of the nitrile group with two non-conserved residues in
the LBP of ERs, Leu384 (ERα)/Met336 (ERβ) and Met421 (ERα)/Ile373 (ERβ). The
nitrile in WAY-292 3b is thought to cause a repulsive interaction with Met421 (ERα)
and show a greater increase in binding affinity for ERβ by dispersive and inductive
3
4
interactions with Leu373 (ERβ). Alternatively, a modeling study of DPN 5c shows
that a productive interaction of nitrile with Met336 (ERβ) contributes to a more elevated
3
5
affinity for ERβ. In contrast to the improving effects of nitrile group in these ligands
on ERβ affinities, the lower RBA of 12a on ERβ compared to 7a suggests that the role
of the nitrile group of 12a is different from the interactions with the ERs proposed
above. In its homologue 13, a significant decrease in the RBA for ERβ compared to 7a
was observed, suggesting that the cyanomethyl group might cause steric hindrance,
similar to the butyl group of 11d.
Of note, cyanide 12a exhibits attractive ERβ selectivity and has a lower
lipophilicity (calculated value: Log P = 3.79, approximately equal to that of endogenous
ligand E2) than 7a and other derivatives. Hence, 12a and its analogues are likely a
better probe to explore the in vivo effects of ERβ agonist.
-
Page 21 of 88 -

Further SAR studies were conducted using 12a as a second template and we
examined substituent effects on the pendant phenyl group. Every 2-substituted
derivatives 12b–d exhibited higher RBAs to both ERs compared to 12a, and a relative
increase in ERα affinity resulted in a decrease in ERβ selectivity. In particular, the
2
-chloro derivative 12c showed superior affinity for ERβ compared to E2, but the
subordinate increase in ERα affinity resulted in a higher loss of selectivity. These
substituent effects on the pendant phenyl of 12a are similar to those on A-ring of ring B
3
6
seco-estradiol 6, in which the corresponding structure of the B-ring of E2 is absent,
suggesting that the pendant phenyl of 12a may mimic the A-ring of E2. Computational
docking studies of ring B seco-estradiols with ERα indicate that 5-chloro substitution of
6
could restore the hydrophobic interactions with neighboring residues of the B-ring of
3
6
E2, thus, the chlorine atom of 12c likely contributes to the high RBA for ERα in the
same manner described above.
On the other hand, substitution at the 3-position did not improve the ERβ affinity,
but compounds 12e–g showed higher selectivities for ERβ compared to the
corresponding 2-substituted derivatives.
In this series of substituted phenols, compounds 12b, 12d, and 12e exerted
comparable or improved functional activities and selectivities for ERβ compared to 12a,
-
Page 22 of 88 -

unfortunately the agonistic activity of 12f was weak. Because cell-based reporter gene
assays are affected by various transcriptional co-regulators or cellular metabolism,
unexpected factors might be involved in the differences between the two assay systems.
Except for 12f, the transcriptional activities of compounds listed in Table 2 were mostly
correlated to their binding affinities.
Table 3. Binding affinities and transcriptional activities for ERs and ERβ selectivities of
the enantiomers of 7a and 12a.
a
c
RBA (%)
Transcriptional acitivity EC50 (nM)
Ligand
b
d
hERα
hERβ
β/α
hERα
hERβ
Rel.β/α
rac-7a
R-7a
9.9 ± 3.6
9.2 ± 2.5
8.2 ± 2.8
1.9 ± 1.1
1.9 ± 0.3
1.5 ± 0.4
67.8 ± 9.4
44.8 ± 7.5
59.7 ± 6.3
36.3 ± 16.0
44.0 ± 21.9
23.8 ± 6.2
6.8
4.9
7.3
19
1,170 ± 265
1,080 ± 89
1,160 ± 131
2,790 ± 152
2,410 ± 56
3,020 ± 184
1.14 ± 0.21
1.40 ± 0.62
1.28 ± 0.46
1.65 ± 0.29
0.78 ± 0.12
3.17 ± 0.88
131
99
S-7a
115
216
394
122
rac-12a
R-12a
S-12a
24
16
a
Relative binding affinity (RBA) values were calculated by IC50[E2]/IC50[test
compound] × 100 (RBA of E2 = 100), shown as mean ± SD of three or four
-
Page 23 of 88 -

independent experiments, performed in duplicate.
b
β/α values = ERβ-RBA/ERα-RBA (ERβ selectivity of binding affinity).
c
EC50 values are shown as Mean ± SD of three independent experiments, performed in
duplicate.
d
Rel.β/α values =
{
ERβ-EC50[E2]/ERβ-EC50[compound]}/{ERα-EC50[E2]/ERα-EC50[compound]} (ERβ
selectivity of agonistic activity).
The above SAR studies, employing readily available racemic mixtures, clarified
the potential of the 1-benzylindane scaffold to afford highly potent and selective ERβ
agonists comparable to isobutestrol 5b and DPN 5c. As such, we confirmed the
pharmacological differences of the optical isomers. We prepared each enantiomer of the
two template compounds 7a and 12a, and evaluated their properties on ERs; however,
no improvement was observed with unsubstituted R-7a and S-7a compared to rac-7a
(
Table 3). The 1-cyano derivative R-12a showed higher binding affinity and agonistic
activity for ERβ compared to rac-12a and S-12a, and its ERβ selectivities were also
improved in both assays. The enantiomers of DPN 5c show the same degree of action
3
7
for ERs, likely due to the high flexibility of the molecule. In contrast, SERBA-1 4
-
Page 24 of 88 -

2
2
having a rigid fused-ring, is superior to the opposite enantiomer. These findings
suggest that structural flexibility of the ligand molecule might influence
stereo-recognition by ERs. Therefore, the nitrile group in the indane-1-position could be
considered to have an effect on increasing the rigidness of the molecule.
4
. Modeling Study
To rationalize the high ERβ selectivity of R-12a, we conducted molecular
modeling of this compound based on the X-ray crystal structure complex of ERB-041
3
4
3
a with ERβ. Fig. 3 shows the most suitable conformation of R-12a within the
ERβ-LBP docked by FRED program (Open-Eye).
-
Page 25 of 88 -

Figure 3. The most likely binding conformation of R-12a (stick, colored by element)
docked into the ligand binding cavity of ERβ (red line ribbons) complexed with
ERB-041 3a (green stick). Only key residues are displayed for simplicity (ball and stick,
colored by element). Potential hydrogen bonds to key residues are shown as blue lines.
Consistent with our expectations, the two hydroxy groups are positioned at the
two important hydrogen-bonding sites, and the hydroxy group on pendant phenyl ring is
facing toward the Glu-Arg-water triad. The nitrile of R-12a is located in close proximity
to the side chain of Leu298 (not shown in Fig. 3). In addition, it is projected onto the
35
α-face of the cavity opposite to that of DPN 5c, and the position of this group is also
-
Page 26 of 88 -

3
4
different from that of WAY-292 3b. This observation is consistent with our SAR
studies in which the influence of nitrile group on ERβ activity was different from DPN
5
c and WAY-292 3b. Although the role of the nitrile of R-12a on an improvement in
ERβ selectivity is not well-defined, the linear geometry of the nitrile functionality might
contribute to hold a preferred binding position for ERβ selectivity.
Interestingly, the ethylene moiety comprising the indane 2- and 3-positions was
expected to overlap with the vinyl group of ERB-041 3a that interacts with Met421
(
ERα)/Ile373 (ERβ), one of two non-conserved residues in LBP of ERs. Thus, we
speculated that the 5-membered ring of indane caused an unfavorable interaction with
Met421 (ERα) that contributed to the high ERβ selectivities of 7a and 12a. Furthermore,
the indane was likely immobilized by the introduction of the 1-nitrile substituent and,
consequently, the extremely high ERβ selectivity of R-12a was achieved.
5
. Conclusions
In this study, we initially identified that 1-benzylindane derivative 7a was an
ERβ agonist with the possibility of a novel ERβ selective scaffold. We next carried out
SAR studies using 7a to confirm the optimal positions of two hydroxy groups and the
suitable size of the bicyclic ring, and investigated substituent effects on the linker
-
Page 27 of 88 -

between the two phenyl rings, leading to the more selective cyanide 12a. Subsequent
modifications of the pendant phenyl ring of 12a also improved the properties and
afforded more promising ERβ selective agonists 12b, 12d, and 12e. We also prepared
each enantiomer of the representative templates 7a and 12a, and the (R)-isomer of 12a
showed an improved profile compared to the racemic mixture, comparable to
structurally related DPN 5c. These findings suggested that this indane-based scaffold
has comparable potential to provide various attractive ERβ agonistic probes with other
multiple nonsteroidal ERβ ligands.
Further derivatization of this series of compounds and investigation of their in
vivo pharmacology will appear in due course.
6
. Experimental
.1. Chemistry
All reagents and solvents were purchased from commercial sources and used
6
without further purification. All moisture and air sensitive reactions were carried out
under an argon atmosphere. Reactions were monitored by thin layer chromatography
(
TLC) using Merck Kieselgel 60 F254 plates or Fuji Silysia Chemical Ltd. Chromatorex
NH-TLC plates. Flash column chromatography was performed on Biotage prepacked
-
Page 28 of 88 -

columns using an automated flash chromatography system (Biotage, Isolera One).
Melting points were determined on a Yanaco micro melting point apparatus (MP-J3).
1
H NMR spectra were recorded on a Bruker AV400M spectrometer at 400.1 MHz.
Chemical shifts (δ) are reported in parts per million (ppm) units using tetramethylsilane
as an internal standard. Data are presented as follows; chemical shift, multiplicity (s,
singlet; d, doublet; t, triplet; m, multiplet; br, broad; br s, broad singlet), coupling
1
3
constant, and integration. C NMR spectra were recorded on a Bruker AV400M
spectrometer at 100.6 MHz with complete proton decoupling. Chemical shifts (δ) are
reported in parts per million (ppm) with the solvent as the internal reference (δ 77.16 in
3
8
CDCl
3
or δ 39.52 in DMSO-d ). High resolution mass spectra (HRMS) were recorded
6
on an Agilent Technologies 6520 Accurate-Mass Q-TOF instrument using electrospray
ionization (ESI, positive or negative ion mode). The chemical purity and optical purity
of tested compounds were determined by HPLC–UV analysis, which was performed on
a Shimadzu LC-VP series instrument (conditions and chromatograms are shown in
supplemental data). Optical rotations were recorded on a HORIBA SEPA-300 using a
3
.5 mm × 50 mm cell. Data are presented as follows; specific rotation, concentration (c
g/100 mL), and solvent.
=
-
Page 29 of 88 -