Estrogen receptor ligands. Part 1: The discovery of flavanoids with subtype selectivity

Article abstract of DOI:10.1016/j.bmcl.2004.01.031

A class of flavanoids exhibiting a high degree of selectivity for ERα over ERβ has been discovered. The most active analogue 6 was found to be 66-fold ERα-selective and demonstrated uterine estradiol antagonism.

Full text of DOI:10.1016/j.bmcl.2004.01.031

Bioorganic & Medicinal Chemistry Letters 14 (2004) 1417–1421
Estrogen receptor ligands. Part 1: The discovery of flavanoids
with subtype selectivity
Helen Y. Chen,a,* Kevin D. Dykstra,a Elizabeth T. Birzin,b Katalin Frisch,b
Wanda Chan,b Yi T. Yang,b Ralph T. Mosley,a Frank DiNinno,a Susan P. Rohrer,b
James M. Schaefferb and Milton L. Hammonda
aDepartment of Medicinal Chemistry, Merck Research Laboratories, PO Box 2000, Rahway, NJ 07065, USA
bDepartment of Atherosclerosis & Endocrinology, Merck Research Laboratories, PO Box 2000,
Rahway, NJ 07065, USA
Received 13 November 2003; revised 14 January 2004; accepted 15 January 2004
Abstract—A class of flavanoids exhibiting a high degree of selectivity for ERa over ERb has been discovered. The most active
analogue 6 was found to be 66-fold ERa-selective and demonstrated uterine estradiol antagonism.
# 2004 Elsevier Ltd. All rights reserved.
Owing to a heightened awareness of the adverse effects
of hormone replacement therapy, prompted by the
Women’s Health Initiative study,1 the search for alter-
native treatments has intensified. Much of the spotlight
has rested on a class of compounds, exemplified by
tamoxifen and raloxifene, known as selective estrogen
receptor modulators (SERMs).2 These agents have the
potential ability to antagonize the proliferative effects of
estrogen on uterine and breast tissue while mimicking
estrogen’s effects on the bone and cardiovascular sys-
tem. The recent discovery of a second estrogen receptor
isoform (ERb) raises the possibility that receptor sub-
type selective ligands may offer key advantages. As part
of a medicinal chemistry program targeting selective
estrogen receptor subtype modulators (SERSMs),2c we
became interested in three naturally occurring leads:
genistein, daidzein, and coumestrol, all of which exhibit
a moderate selectivity for ERb (20X at best) and con-
tain a common benzopyran motif (Fig. 1). WS-7528
(Fig. 1), another structurally similar flavanone, isolated
from a strain of Streptomyces, was reported to have
estrogen-like characteristics.3
structure.4 Herein, we wish to describe the synthesis and
SAR of compounds shown in Table 1.
Initially, flavanones 1–8 were prepared according to lit-
erature procedures with minor modifications (Scheme 1,
Method A).5 Thus, Knoevenagel condensation of piper-
idinylethoxy–benzaldehyde with the appropriate ketones6
23 yielded a mixture of cis and trans flavanones. The
titer of the cis isomer could be increased to 50% simply
by epimerization of the 1:4 cis/trans mixture of TBS-
protected flavanones 24 with LiHMDS at À78 ꢀC.
Although the two isomers could be separated on silica
gel with great care, it became apparent early on that a
stereoselective synthesis of cis flavanones was needed.
Although many synthetic methods for the construction
of flavanones are known in the literature, only a few
stereospecific syntheses of cis flavanones exist.7
We therefore decided to pursue a stereospecific synthesis
of cis-2,3-disubstituted flavanones based on Donnelly’s
Based on these findings, we sought to identify a novel
series of SERSMs centered on the flavanone core
Keywords: SERMs; Flavanoids.
* Corresponding author. Tel.: +1-732-594-5323; fax: +1-732-594-
9556; e-mail: helen_chen@merck.com
Figure 1. Structures of naturally occurring leads.
0960-894X/$ - see front matter # 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.01.031

1418
H. Y. Chen et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1417–1421
Table 1. Flavanones and their derivatives
Compd
Typec
X
R1
R2
R3
R4
R5
Methodb
1
2
3
4
5
6
7
8
I
I
I
—
—
—
H
H
H
H
OH
H
—
—
—
Me
iPr
—
—
—
Pipa
Pip—
Pip
H
Pip—
Pip—
Pip
Pip
Pip
Pip
Pip
Pip
Pip
Pip
Pip
Pip
—
—
—
—
—
—
—
—
—
A
A
A
A
4-Hydroxyphenyl
Me
iPr
4-Hydroxyphenyl
4-Hydroxyphenyl
Me
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
III
III
III
OH
OH
OH
OH
OH
OH
H
OH
OH
OH
OH
OH
OH
OH
OH
OH
—
—
A
A
—
—
—
—
A or B
Ad
—
—
e,f
9
Phenyl
B
10
11
12
13
14
15
16
17
18
19
20
21
22
H
4-Hydroxyphenyl
4-Hydroxyphenyl
4-Hydroxyphenyl
4-Hydroxyphenyl
4-Hydroxyphenyl
4-Hydroxyphenyl
4-Hydroxyphenyl
2-Methoxy-4-hydroxyphenyl
3-Methoxy-4-hydroxyphenyl
3,5-Dimethyl-4-hydroxyphenyl
—
—
—
—
—
—
—
—
—
—
—
H
—
—
—
—
—
—
—
—
—
—
OH
H
B
B
B
B
B
Bf
Bf
B
B
B
Me
Et
Pr
Pentyl
F
Cl
H
H
H
—
—
—
—
—
—
—
—
—
—
—
—
H
—
=NOH
a Pip, piperidinoethyl.
b Scheme 1 and 2.
c Racemic, abs. configuration unknown.
d Used 4-hydroxybenzaldehyde in stepa.
e Used Ni2B as the reductant in steph.
f Used mono-MOM-protected 25 in stepa.
Knoevenagel condensation of ketones9 25 with the
TBDPS-protected benzaldehyde according to Scheme 2
(Method B). Stereospecific introduction of the aryl
10
groupat C-3
was performed by addition of organo-
lead reagents prepared according to literature proce-
dures.11 Subsequent TBS-deprotection and elaboration
to the basic side chain was accomplished with chlor-
oethyl–piperidine and cesium carbonate in refluxing
acetone. Treatment with 2 N HCl successfully removed
the MOM groups to give the penultimate substrate 27 in
good yields. At this time, we explored the reductive
removal of the SPh groupunder a variety of conditions.
Following the literature precedent,8 Ni2B afforded a
mixture of the isomers varying from 2:1 cis/trans to 10:1
cis/trans depending on the freshness of the catalyst.
Interestingly, with SmI2, a reverse stereochemical out-
come of 2:1 trans/cis ratio was obtained. Success was
finally achieved using excess RaNi as the reductant to
give 49–88% yield of the desired cis isomer. In the case
where X=Cl, F (16 and 15), some dehalogenation was
observed (66% and 10%, respectively).
Scheme 1. Method A, reagents and conditions: (a) piperidine, toluene,
120 ꢀC, overnight; (b) NaOAc, MeOH, 80 ꢀC, 3 h, 41–95% from 23;
(c) 2 N HCl, EtOH, rt, 66–100%; (d) TBSCl, Et3N, CH2Cl2, 78–100%;
(e) LiHMDS, THF, À78 ꢀC, 39–42%; (f) TBAF, AcOH, THF, 0 ꢀC,
62–87%.
work,8 wherein 3-phenylthio-chroman-4-ones would not
only allow for the requisite arylation reaction but the
phenylthio group would also serve as our stereocontrol
element upon reductive removal. To that end, the
flavanone intermediate 26 was prepared via the
With the cis flavanone 6 in hand, compounds of Type
III were readily accessible. Following modified literature
procedures,5b reduction of the ketone using lithium

H. Y. Chen et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1417–1421
Table 2. Binding affinities13 and in vivo data14
1419
Compd
Human
ERa
IC50 (nM) IC50 (nM)
Human
ERb
Selectivity
[b]/[a]
Uterine
Wt. assay
(@1 mpk, sca)
1
2
3
4
5
6
7
8
9
10
11
12
1415
109
49
3563
89
>10,000
930
7
9
40
2
13
66
1
2
19
3
39
2
NDb
15% Agonism
17% Inhibition
ND
0% Inhibition
50% Inhibition
1947
5645
1200
2049
598
31
490
551
531
179
14
ND
ND
ND
ND
1200
>10,000
510
546
152
46% Inhibition
ND
68
13
14
652
71
207
454
0.3
6
ND
ND
15
16
17
18
19
20
105
58
149
413
1050
929
5390
1270
1360
>10,000
>10,000
7000
51
22
9
24
10
8
17% Agonism
ND
ND
ND
ND
ND
Scheme 2. Method B, reagents and conditions: (a) piperidine, toluene,
120 ꢀC, overnight; (b) NaOAc, MeOH, 80 ꢀC, 3 h, 30–85% from 25;
(c) MOMCl, DMF, Hunig’s base, 47–88%; (d) R2Pb(OAc)3, pyridine,
CHCl3, 40 ꢀC, 52–81%; (e) TBAF, AcOH, THF, 0 ꢀC, 60–96%; (f)
Cs2CO3, 1-(2-chloroethyl)piperidine monohydrochloride, acetone,
60 ꢀC, 79–100%; (g) 2 N HCl, MeOH, 80 ꢀC, 56–100%; (h) RaNi,
EtOH, 49–88%.
21
22
6.7
1040
1.8
1.3
92
8.9
8610
12
1
8
7
1
0.04
0.1
0.2
ND
ND
Raloxifene
b-Estradiol
Genistein
Daidzein
Coumestrol
96% Inhibitionc
100% Agonismd
63% Agonism
ND
1.1
4
303
2
triethyl borohydride in THF at 0 ꢀC afforded 20 in 76%
yield. Compound 20 could be further reduced upon
careful treatment with TFA and triethylsilane in meth-
ylene chloride to give flavan 21 in 68% yield along with
its corresponding chromene. As reported by Donnelly,12
oximation of flavanone 3 or 6 with hydroxylamine HCl
and piperidine in pyridine cleanly gave the cis oxime 22
in 61% yield.
2160
11
ND
a sc=subcutaneous.
b ND=not determined.
c @ 0.6 mpk.
d @ 2 mg/kg.
no antagonist activity in an in vitro coactivation
assay.15 However, the presence of the basic side-chain
was not the only requirement for antagonism as evi-
denced by compound 5 (R2=i-Pr) which was totally
devoid of in vivo activity, and thus pointed toward the
need for the hydroxyphenyl substituent as well. Both
hydroxyls were required for optimal binding as shown
by the loss of binding to the receptor upon removal of
either hydroxyl at R1 or R2 (9, 10). Similarly, replace-
ment of the carbonyl by an oxime (22), diminished the
binding as well as the ERa selectivity. Reduction of
the carbonyl to the alcohol 20 was likewise detrimental
to the binding activity; however, further reduction of 20,
afforded a potent non-selective ligand 21. In general,
substitutions X (12–16), whether an extended aliphatic
chain or a halogen substituent, resulted in a loss of
receptor affinity and subtype selectivity, as compared
with 6. One exception to this trend occurred when
X=Me (11). Although not as selective as 6, 11 bound
with greater affinity to the alpha receptor and demon-
strated an equal level of antagonism in the uterine
weight assay. Further addition of substituents on the 4-
hydroxyphenyl group (17–19) offered no improvements.
Compounds 1–22 were tested for potency and selectivity
in an ER competitive binding assay with tritiated 17-b
estradiol.13 Agonist and antagonist activities of select
compounds were evaluated in vivo using an immature
rat uterine weight assay.14 The results are shown in
Table 2.
When we derivatized racemic, synthetic WS-7528 to
generate flavanone 8, we observed a slight shift towards
ERa selectivity rather than the anticipated selectivity for
ERb. As the size of the substituent at C-3 was increased
to isopropyl (5), not only was binding to the ERa receptor
enhanced, but selectivity over ERb was also improved.
We found optimal binding and almost 70-fold selectivity
with the introduction of a 4-hydroxyphenyl group (6).
Although the binding data would suggest that the trans
isomer differed only slightly from the cis isomer, further
evaluation in the immature rat uterine assay clearly
identified the cis isomer as the more pharmacodynamically
active antagonist. For instance, at 1 mpk sc, the trans
flavanone 3 exhibited a 17% inhibition of the estradiol
effect in contrast to 50% inhibition with its cis isomer 6.
It was thus established early in the project that the cis
stereochemistry was critical for the development of a
SERM.
As depicted in Figure 2, molecular modeling of 6, in
white against raloxifene in purple, docked in the ER
receptors, showed that although structurally quite dif-
ferent, 6 mapped fairly well with raloxifene and main-
tained the well established interactions in the ligand
binding domain.16,17
As with raloxifene, the basic side chain proved to be
crucial for in vivo antagonist activity, since 7 displayed

1420
H. Y. Chen et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1417–1421
groupof the flavanone is crucial for maintaining sub-
type selectivity. This work has led to the development of
more potent SERSMs2c or in this instance, selective
estrogen receptor alpha modulators (SERAMs) which
will be reported in future communications from this
laboratory.
Acknowledgements
We gratefully acknowledge Professor Barry M. Trost
and Professor David Evans for their helpful discussions
during their scientific consultations.
References and notes
Figure 2. Molecular modeling of 6 (white) against Raloxifene (purple).
HERa is depicted in purple and hERb in green. Residue numbering is
hERa unless otherwise indicated.
1. Writing Groupfor the WHI Investigators, JAMA, 2002,
288, 321.
2. (a) Recent reviews on SERMs: Jordan, V. C. J. Med.
Chem. 2003, 46, 883. (b) Jordan, V. C. J. Med. Chem.
2003, 46, 1081. (c) Meegan, M. J.; Lloyd, D. G. Curr.
Med. Chem. 2003, 10, 181. (d) Miller, C. P. Curr. Pharm.
Design 2002, 8, 2089. (e) Bryant, H. U. Endocrine and
Metabolism Disorders 2002, 3, 231.
3. Nakayama, O.; Yagi, M.; Tanaka, M.; Kiyoto, S.;
Uchida, I.; Hashimoto, M.; Okuhara, M.; Kohsaka, M. J.
Antibiotics XLIII 1990, 1394.
4. (a) For recent examples of synthetic flavanoid ligands, see:
Miller, C. P.; Collini, M. D.; Harris, Heather, A. Bioorg.
Med. Chem. Lett. 2003, 13, 2399. (b) Kim, Y.-W.;
Mobley, J. A.; Brueggmeier, R. W. Bioorg. Med. Chem.
Lett. 2003, 13, 1475. (c) Pouget, C.; Lauthier, F.; Simon,
A.; Fagnere, C.; Basly, J.-P.; Delage, C.; Chulia, A.-J.
Bioorg. Med. Chem. Lett. 2001, 11, 3095.
5. (a) Saeed, A.; Sharma, A.; Durani, N.; Jain, R.; Durani,
S.; Kapil, R. S. J. Med. Chem. 1990, 33, 3210. (b) Saeed,
A.; Sharma, A.; Durani, S.; Kapil, R. S. J. Med. Chem.
1990, 33, 3222.
It is postulated that the crucial difference responsible for
the alpha-selectivity of 6 lies in the interaction of its
carbonyl with the two discriminating residues lining the
receptor pocket (Fig. 3). In ERb, there is both a steric
and electronic repulsion between the carbonyl oxygen
atom of the ligand with the Met 354 residue, which
would be expected to be absent in ERa, where the
corresponding residue is Leu 384.
In summary, we have created a series of cis flavanones
which exhibit a greater affinity for ERa over ERb and in
the process, have developed a stereoselective synthesis
to these compounds. With compound 6, we have gener-
ated an almost 70-fold ER alpha selective ligand with
demonstrated in vivo estradiol antagonism on the
uterus. We have determined from our SAR that a cis
relationshipis preferred, and both hydroxyls at R and
1
6. (a) The ketones 23 in Scheme 1, where commercially
unavailable, were generally prepared by Friedel–Crafts
acylation of the appropriate acid chloride or carboxylic
acid with resorcinol according to literature procedures:
Wahala, K.; Hase, T. J. Chem. Soc., Perkin Trans. 1 1991,
3005. (b) Chiba, K.; Sonoyama, J.; Tada, M. J. Chem.
Soc., Perkin Trans 1 1996, 1435. (c) Chandra, H.; Zilliken,
F.; Offermann, W.; Breitmaier, E. Can. J. Chem. 1981, 59,
2266. (d) Selective protection with THP was accomplished
using established methodology: Kapil, R. S. J. Med.
Chem. 1990, 33, 3222.
R2 are required for optimal binding to the alpha recep-
tor. The basic side chain, as demonstrated in other
SERM platforms, is required for in vivo antagonism
and is optimal with a 4-hydroxyphenyl group at the C-3
position of the isoflavanone. Finally, the carbonyl
7. (a) Donnelly, D. M. X.; Keenan, A. K.; Leahy, T.;
Philbin, E. M. Tetrahedron 1972, 28, 2545. (b) Fujise, S.;
Fujise, Y.; Hishida, S. Nippon Kagaku Zasshi 1963, 84, 78.
(c) Chem. Abstr. 1964, 60, 5444c. (d) Bognar, R.; Rakosi,
M.; Litkei, Gy. Acta Chim. Engl. 1962, 34, 253. (e) Clark-
Lewis, J. W.; Spotswood, T. M.; Williams, L. R. Aust. J.
Chem. 1963, 16, 107. (f) McKervey, A. M.; Ye, T. J.
Chem. Soc., Chem. Commun. 1992, 823.
8. Donnelly, D. M. X.; Fitzpatrick, B. M.; O’Reilly, B. A.;
Finet, J.-P. Tetrahedron 1993, 49, 7967.
9. The ketones 25 in Scheme 2 were generally prepared by
Friedel–Crafts acylation of 3-substituted resorcinols with
commercially available (phenylthio) acetyl chloride using
SnCl4 as the Lewis acid. In the case where X=H, the desired
ketone was obtained by displacement of the Houben–
Hoesch reaction product of resorcinol and chloroaceto-
nitrile with NaSPh. A representative procedure for the
Figure 3. Modeling of 6 in the ERa and ERb receptors. Arcs represent
proposed clash between carbonyl O and Met 354 (hERb) sidechain.

H. Y. Chen et al. / Bioorg. Med. Chem. Lett. 14 (2004) 1417–1421
1421
Friedel–Crafts acylation is given below for X=F: To a
mixture of 5-fluoro-1,3-dimethoxybenzene (0.5 g, 3.2
mmol) in CH2Cl2 (5 mL) was added (phenylthio) acetyl
chloride (0.48 mL, 3.2 mmol) at 0 ꢀC under a nitrogen
atmosphere, followed by dropwise addition of a 1 M
solution of SnCl4 in CH2Cl2 (4.2 mL, 4.2 mmol). After
stirring for 30 min, the reaction was allowed to warm to
ambient temperature and stirred for another 1 h. The
reaction was partitioned between ethyl acetate and 2 N HCl.
The organic layer was washed with brine, dried over sodium
sulfate, filtered, and concentrated in vacuo. The crude
material contained a 2:1 mixture of the desired product
and its other isomer. Purification by silica gel chromato-
graphy with 20% ethyl acetate in hexane as the eluant
afforded the desired product in 32% yield. Standard
deprotection of the methoxy groups was accomplished
using BBr3 to give the desired ketone 25 in 40% yield.
10. An X-ray crystal structure determination of a compound
related to 27, where X is H, R2 is 4-hydroxyphenyl,
the MOM group is replaced by a methyl group, and the
TBDPS groupis replaced by a hydrogen atom revealed
that the phenyl groups at C-2 and C-3 were trans to one
another.
11. Kozyrod, R. P.; Morgan, J.; Pinhey, J. T. Aust. J. Chem.
1985, 38, 1147.
12. Donnelly, D. M. X.; Keenan, A. K.; Leahy, T.; Philbin,
E. M. Tetrahedron 1972, 28, 2545.
13. The IC50 values were generated in an estrogen receptor
ligand binding assay. This scintillation proximity assay
was conducted in NEN Basic Flashplates using tritiated
estradiol and full length recombinant human ER-alpha
and ER-beta proteins, with incubation times of 3–23 h. In
our experience, this assay provides IC50 values that are
reproducible to within a factor of 2–3. Most compounds
are single point determinations. The binding results for 6
reflect an average of three determinants with incubation
times of 3 and 15 h. For estradiol, the binding data
reflects an average of over 100 determinants at 3 h of
incubation.
14. 20-day old intact female Sprague–Dawley rats were trea-
ted (sc) with test compounds for 3 days at 1 mpk. The
uteri wet weights were determined on day 4 and dry
weights were determined after air-drying the tissue sam-
ples for 3 days. The anti-estrogenic activity of compounds
was determined by co-administration of the compound
with a subcutaneous injection of 17-beta-estradiol and
reported as % inhibition. The estrogenic activity (partial
agonism) of the compounds was determined by adminis-
tering the test compound without estradiol and reported
as % control.
15. Mitra, S. W.; Cai, S.; Wilkinson, H.; Salzmann, G.; Zhou,
G.; Hayes, E.; Dashkevicz, M.; Cummings, R.; Smith, R.;
Schaeffer, J. M. Abstracts, 80th Annual Meeting of the
Endocrine Society, New Orleans, June, 1998, P1-562,
p235.
16. (a) Shiau, A. K.; Barstad, D.; Loria, P. M.; Cheng, L.;
Kushner, P. J.; Agard, D. A.; Greene, G. L. Cell 1998, 95,
927. (b) Brzozowski, A. M.; Pike, A. C. W.; Dauter, Z.;
Hubbard, R. E.; Bonn, T.; Engstrom, O.; Ohman, L.;
Greene, G. L.; Gustafsson, J. A.; Carlquist, M. Nature
1997, 389, 753.
17. Our docking and energy minimization approach, which
will be described elsewhere by Ralph Mosley in due
course, identified a binding mode of compound 6 with the
absolute configuration of [2S,3R] as being the most likely
to explain the SAR.