The A-CD analogue of 16β,17α-estriol is a potent and highly selective estrogen receptor β agonist

Article abstract of DOI:10.1039/c3md00194f

Selective estrogen receptor β (ERβ) agonists display neuroprotective properties in animal models and hold promise in the treatment of neurodegenerative diseases. In our quest to design, synthesize and evaluate potent and safe ERβ agonists, we focused on m

Full text of DOI:10.1039/c3md00194f

MedChemComm
View Article Online
View Journal
CONCISE ARTICLE
The A-CD analogue of 16b,17a-estriol is a potent and
highly selective estrogen receptor b agonist†
Cite this: DOI: 10.1039/c3md00194f
a
a
a
b
´
¨
´
Claire Sauvee,‡ Anja Schafer,‡ Henrik Sunden, Jian-Nong Ma,
c
b
abd
*
Anna-Lena Gustavsson, Ethan S. Burstein and Roger Olsson
Selective estrogen receptor b (ERb) agonists display neuroprotective properties in animal models and hold
promise in the treatment of neurodegenerative diseases. In our quest to design, synthesize and evaluate
potent and safe ERb agonists, we focused on making an analogue of 16b,17a-estriol (16,17-epiestriol), a
potent and one of the most ERb selective endogenous estrogens reported. Herein we disclose the
synthesis and in vitro evaluation of an analogue based on the recently introduced A-CD scaffold. A 14-
step synthesis based on the Hajos–Parrish ketone resulted in the discovery of (1S,2S,3aS,5S,7aS)-5-(4-
hydroxyphenyl)-7a-methyloctahydro-1H-indene-1,2-diol (15). This A-CD analogue of 16b,17a-estriol is a
highly selective (500-fold) ERb full agonist over ERa with a pEC50 of 7.7 at ERb. Molecular modelling
suggests that 15 turns around in the ligand-binding domain compared to estriol, thus the 7a-methyl
occupies the a-face, which might explain the high selectivity.
Received 10th July 2013
Accepted 14th August 2013
DOI: 10.1039/c3md00194f
www.rsc.org/medchemcomm
studies are ongoing in humans for MS.8,9 However, we were
more attracted by the 16b,17a-isomer of estriol (16,17-epi-
Introduction
Herein we report the design and synthesis of a new selective
ERb ligand based on one of the most ERb selective endogenous
estrogens reported. Estrogens have been shown to play a neu-
roprotective role in neurodegenerative diseases, preventing
degeneration of neurons in Parkinson's Disease (PD)2 and Alz-
heimer's Disease (AD),3 and having anti-inammatory and
remyelination properties in multiple sclerosis (MS).4 Hence,
treatment with 17b-estradiol (Fig. 1), the major endogenous
estrogen that activates both estrogen receptors a (ERa) and b
(ERb), could be a possible therapy for neurodegenerative
diseases. Unfortunately, feminizing effects and increased risk of
cancer limit its use.5 Considering that the adverse effects of
hormone therapy are associated with activation of ERa, highly
selective compounds for ERb could be a safe treatment for
neurodegenerative diseases. One of the lesser investigated
estrogens, estriol, displays a preference for ERb over ERa
(relative binding affinity (RBA) ERa: 14%; ERb: 21%).6 Estriol
shows neuroprotection in mice7 and clinical phase II and III
estriol, Fig. 1) because it had been reported to have a higher ERb
selectivity both in a functional assay (EC50: ERa 72 nM; ERb: 11
nM)10 and in binding affinity (RBA ERa: 1%; ERb: 13%).11 The
two other estriol hydroxyl isomers, 17-epiestriol and 16-epi-
estriol, both display equal affinity to ERb and ERa.11
A further indication of the ERb selectivity of 16,17-epiestriol
is uterine hypertrophy that is driven by ERa activation, and thus
aDepartment of Chemistry and Molecular Biology/Medicinal Chemistry, University of
Gothenburg, SE-412 96, Gothenburg, Sweden. E-mail: roger.olsson@med.lu.se; Fax:
+46 40 601 34 01; Tel: +46 768 87 42 17
bACADIA Pharmaceuticals, 3911 Sorrento Valley Blvd., San Diego, USA
cChemical Biology Consortium Sweden, Department of Medical Biochemistry
&
Biophysics, Karolinska Institute, SE-171 77 Stockholm, Sweden
dChemical Biology & Therapeutics, Department of Experimental Medical Science, Lund
University, SE-221 84 Lund, Sweden
† Electronic supplementary information (ESI) available: Details of chemical
preparations, NMR data and molecular modelling. See DOI: 10.1039/c3md00194f
‡ These authors contributed equally to this paper.
Fig. 1 17b-estradiol, the ERb preferring agonist genistein, estriol, 16b,17a-
estriol, the 17b-estradiol A-CD analogue from Asim1 and our compound 15.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun.

View Article Online
MedChemComm
Concise Article
is a good way to assess ERb/ERa selectivity in vivo. Takahashi
et al.12 analyzed different estrogens for uterus growth. While
estriol, 17-epiestriol and 16-epiestriol all showed considerable
uterus growth comparable with 17b-estradiol, 16,17-epiestriol
showed an effect comparable with the vehicle-treated controls.12
Recently, it has been shown1,13 that in A-CD estrogens (Fig. 1)
the active isomer is always the one with the 5(S)-stereochemistry
and that they generally have a higher ERb selectivity compared
to the steroid-based ones. The reason for this may be the cis
stereochemistry between the CD rings in the A-CD estrogens, in
comparison to the trans-relation between the CD rings in the
steroids. We speculated that the stereochemistry and change of
orientation in the binding mode would potentially accentuate
the selectivity of an A-CD 16,17-epiestriol analogue. Previously
we have established structure–activity relationships and per-
formed pharmacological evaluations of non-steroidal ERb
agonists in models of neuropathic pain, PD, and AD.14–17 Based
on the available literature, we set out to invest in the synthesis
of the 16b,17a-estriol A-CD analogue. The estrone- and 17b-
estradiol A-CD analogues disclosed previously were synthesized
for comparing the ERa and ERb activity in a functional whole
cell assay.
Scheme 2 Reagents and conditions for the A-CD estrone analogue synthesis. (k)
Dess–Martin periodinane, DCM, 140 ꢁC, 1 h, MW; (l) TBAF, THF, rt, 3 h.
periodinane (Scheme 2).20 The resulting mixture of diastereo-
mers 10a and 10b was separated by ash chromatography at
this stage of the synthesis and the isomers were then desilylated
with TBAF to yield 11a and 11b in 99 and 73% yields, respec-
tively. 11a is the faster eluting isomer and was assigned to the
natural stereochemistry (5S-hydroxyphenyl) according to
Wright et al.13
Finally, the synthesis of 16b,17a-estriol analogue 15 started
with acetylation of the ketone 11a (Scheme 3).21 A subsequent
epoxidation of compound 12 with m-CPBA gave compound 13.22
Treating 13 with LiAlH4 gave compound 14 via an epoxide
opening and deprotection of the secondary alcohol,21 but not of
the phenol. Therefore, a mild deprotection method with
ammonium acetate in aqueous medium23 was used to form
16b,17a-estriol analogue 15. The (S)-stereochemistry of the
hydroxyl groups at positions 1 and 2 was assigned by COSY and
NOESY spectra. For details see ESI.†
Synthesis
Hajos–Parrish ketone 4 was synthesized following the literature
procedure (see ESI†),18,19 reduced to 5 with H2 Pd/C 10% and
used as the starting material for further synthesis. The
following routes were used in the synthesis of the estrogen
analogues: for the estradiol analogue, the method of Katze-
nellenbogen and co-workers was followed (Scheme 1).1,13
Compound 5 was reacted with tert-butyldimethylsilyl- (TBDMS)
protected bromophenol and n-butyl lithium leading to the
formation of compound 6. Elimination of the tertiary alcohol
using a catalytic amount of p-TsOH gave the desired product 7
as an olenic mixture in 19% yield. Compound 7 was then
subjected to Pd/C 10% to reduce the double bond and form
compound 8. Deprotection of the phenol with TBAF delivered
compound 9 in 83% yield as a mixture of diastereomers.
The estrone analogues 10a and 10b were obtained by
oxidation of compound 8 to the ketone using Dess–Martin
Pharmacological testing and molecular modelling
Compounds 9, 11a, 11b and 15 were tested for activation of
estrogen receptors in a whole cell reporter gene assay as
reported previously (Table 1).14 The results were normalized to
estradiol, the endogenous ligand, and compared with estriol
and genistein, both compounds that prefer ERb. All compounds
selectively activated ERb over ERa. The most potent compound,
estradiol analogue 9, was about 6-fold less potent than the
natural ligand estradiol. The most selective compound was the
16,17-epiestriol analogue 15 with a 500-fold selectivity for ERb
Scheme 1 Reagents and conditions for the A-CD estradiol analogue synthesis.
(f) n-BuLi, THF, ꢀ78 ꢁC, 1 h; (g) p-TsOH, DCM, rt, 16 h; (h) Pd/C 10%, H2, MeOH, rt;
(i) TBAF, THF, rt, 3 h.
Scheme 3 Reagents and conditions for the A-CD 16,17-epiestriol analogue
synthesis. (m) Isopropenyl acetate, H2SO4 conc., 70 ꢁC, 16 h; (n) m-CPBA, toluene,
rt, 20 h; (o) LiAlH4, THF, rt, 2 h; (p) NH4OAc, MeOH : H2O 4 : 1, rt, 2 h.
Med. Chem. Commun.
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Concise Article
MedChemComm
Table 1 Estrogenic activities of ER ligands 9, 7a, 11a, 11b and 15a
Selectivity
ERb/ERa
Compound
pEC50 ERa
Efficacy [%]
pEC50 ERb
Efficacy [%]
RBA ERad
RBA ERbd
Estradiolb
Estriolb
Genisteinc
9
11a
11b
10.0
100
10.4
100
ꢂ2
16
25
79
>40
63
100d
—
100d
—
8.7 ꢃ 0.1
6.3 ꢃ 0.1
7.5 ꢃ 0
<5
84 ꢃ 19
91 ꢃ 9
82 ꢃ 2
67 ꢃ 3
62 ꢃ 2
56 ꢃ 1
9.9 ꢃ 0.2
7.7 ꢃ 0.1
9.4 ꢃ 0.2
6.6 ꢃ 0.1
6.9 ꢃ 0.3
7.7 ꢃ 0.2
79 ꢃ 18
115 ꢃ 23
127 ꢃ 33
148 ꢃ 8
144 ꢃ 1
108 ꢃ 2
—
—
1.47 ꢃ 0.26d
21.5 ꢃ 4.6d
0.008 ꢃ 0.001d
0.112 ꢃ 0.01d
5.1 ꢃ 0.7
—
—
—
—
15
5.0 ꢃ 0
501
a
Values are mean ꢃ SD of at least 2 independent experiments; the functional data are reported as [%] efficacy and potency (pEC50 ¼ ꢀlog [EC50]).
b
Natural ligand, for comparison. c Corroborates with ref. 14. d From ref. 13 for comparison; RBA ¼ relative binding affinity.
over ERa (Table 1). Both estrone analogues, 11a and 11b, dis-
played similar activation of ERb in the whole cell assay (Table 1,
pEC50 ERb: 6.6 and 6.9, respectively). Comparison with the
previously reported results of 9 and 11a is difficult, as functional
data were not reported; nevertheless, the selectivity prole
correlates between the two assays (Table 1).
To investigate the origin of the high selectivity of compound
15, docking studies were performed. Compound 15 was docked
in the co-crystal structure of 17-epiestriol with ERb (2J7Y.pdb)24
using the Schrodinger soware,25 with the previously described
¨
protocol.26 17-Epiestriol reveals an interesting binding mode
when co-crystallized with ERb.24 Compared to estriol, it turns
around in the ligand-binding pocket and the methyl occupies
the a-face instead of the b-face. According to the proposed
binding mode of compound 15 shown in Fig. 2a, it adopts the
same orientation as 17-epiestriol with the methyl group posi-
tioned down into the a-face of the ligand binding pocket. The
co-crystal structure of 17-epiestriol with ERb is the only
estrogen/ER structure of an agonist deposited in the pdb that
shows this “upside down” orientation of the ligand. There is
one structure of an antiestrogen that also shows the ipped
binding mode, that of the antiestrogen ICI-164.384
(1HJ1.pdb).27
Fig. 2 (a) ERb LBD co-crystallized with 17-epiestriol (2J7Y.pdb): docking of
compound 15 compared to 17-epiestriol. (b) ERa LBD co-crystallized with estriol
(3Q95.pdb): superimposed with the docked compound 15 compared to estriol.
Compound 15: grey, estriol: light blue, 17-epiestriol: green, amino acid residues
ERb: grey, and amino acid residues from ERa: light blue. Numbering of amino
acids is from ERa, if not indicated otherwise.
In contrast, in the co-crystal of ERa with estriol (3Q95.pdb),28
the methyl group of estriol points up into the b-face of the
receptor pocket (Fig. 2b), whereas the docking with 15 in this
structure still shows the “upside down” orientation, maybe due
to steric clashes with Met421. In the case of 17-epiestriol in ERb,
the methyl group reaches down into the a-face towards the hydrogen bond to His524/475 using the 16a-hydroxyl group as a
amino acid residue that differs between the receptors (Met421 surrogate for the 17b-hydroxyl group in estriol, which is not
in ERa and Ile373 in ERb), an area which has previously been possible in the classic orientation where none of the hydroxyls
reported to be responsible for selectivity.29 The selectivity seen will be able to hydrogen bond to His524/475. For 17-epiestriol,
for compound 15 may be due to this proposed binding mode. an additional stabilization for the ipped binding mode is
However, 17-epiestriol shows the same binding mode but is not provided by a second hydrogen bonding interaction between
ERb selective. Hence, one reason for the selectivity of 15 may be the 17a-hydroxyl group of the ligand to the Gly427 (not shown)
that the cis stereochemistry between the CD rings of compound backbone carbonyl group in the receptor.
15 leads to a more puckered conformation of the D ring. This
This may in part also apply to compound 15. However, by
makes the D ring less at and it can protrude further into the b- simultaneously inverting the stereochemistry at both hydroxyls,
face. However, if the compound experienced steric clashes in corresponding to 16 and 17 in estriol, the hydroxyl groups adopt
the LBD in the ipped mode, it could simply evade these by a conformation that makes it possible to form the same
adopting the classic binding mode. Therefore, there must be an hydrogen bonds in an “upside down” binding mode as estriol
additional reason to why the ipped binding mode is preferred. does in the classical binding mode. Thus a similar bifurcated
17-Epiestriol rotates 180ꢁ in the LBD mainly to be able to bond with His524/475 as seen in the estriol co-crystallization is
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun.

View Article Online
MedChemComm
Concise Article
made possible. Alternatively, in the A-CD 16,17-epiestriol 15 it 11 B. T. Zhu, G. Z. Han, J. Y. Shim, Y. Wen and X. R. Jiang,
appears that the distance between the oxygen from the 17a- Endocrinology, 2006, 147, 4132–4150.
hydroxyl group and the keto group in the backbone is more 12 M. Takahashi, T. Shimomoto, K. Miyajima, M. Yoshida,
˚
optimal for hydrogen bonding in the “upside down” (ꢂ3.1 A)
S. Katashima, F. Uematsu, A. Maekawa and D. Nakae,
˚
than in the classic (ꢂ4.5 A) binding mode. In summary, in
Cancer Lett., 2004, 211, 1–9.
either case the second hydrogen bond would favor the “upside 13 J. S. Wright, H. Shadnia, J. M. Anderson, T. Durst, M. Asim,
down” mode over the classical binding mode.
M. El-Salti, C. Choueiri, M. A. C. Pratt, S. C. Ruddy, R. Lau,
K. E. Carlson, J. A. Katzenellenbogen, P. J. O'Brien and
L. Wan, J. Med. Chem., 2010, 54, 433–448.
Conclusions
14 F. Piu, C. Cheevers, L. Hyldto, L. R. Gardell, A. L. Del
Tredici, C. B. Andersen, L. C. Fairbairn, B. W. Lund,
M. Gustafsson, H. H. Schiffer, J. E. Donello, R. Olsson,
D. W. Gil and M. R. Brann, Eur. J. Pharmacol., 2008, 590,
423–429.
15 R. Olsson, L. Hyldto, M. Gustafsson and B. Wintherlund, in
WO2009055734 (A1), W. I. P. Organization, 2009, p. 79.
16 H. Sunden and R. Olsson, Org. Biomol. Chem., 2010, 8, 4831–
4833.
In summary, we have synthesized and evaluated a novel and
potent ERb agonist, an AC-D 16,17-epiestriol analogue (15), with
500-fold selectivity between ERa and ERb. In addition, the
molecular modelling study suggests that longer alkyl chains
instead of the methyl group in the 7a position could protrude
further into the a-face to improve ERb selectivity, as demon-
strated with the butyl residue in the ERb selective
tetrahydrouorenones.30
17 K. McFarland, D. Price, C. N. Davis, J.-N. Ma, D. W. Bonhaus,
E. S. Burstein and R. Olsson, ACS Chem. Neurosci., 2013, DOI:
10.1021/cn400132u, accepted.
18 Z. G. Hajos and D. R. Parrish, Org. Synth. Coll., 1990, 7, 368.
19 D. E. Ward, C. K. Rhee and W. M. Zoghaib, Tetrahedron Lett.,
1988, 29, 517–520.
20 D. B. Dess and J. C. Martin, J. Org. Chem., 1983, 48, 4155–
4156.
21 N. S. Leeds, D. K. Fukushima and T. F. Gallagher, J. Am.
Chem. Soc., 1954, 76, 2943–2948.
22 Shagua, R. Singh and G. Panda, Indian Chem. Eng., Sect. B,
2009, 48, 989–995.
23 C. Ramesh, G. Mahender, N. Ravindranath and B. Das,
Tetrahedron, 2003, 59, 1049–1054.
24 A. C. W. Pike, A. M. Brzozowski, R. E. Hubbard, J. Walton,
T. Bonn, A.-G. Thorsell, O. Engstrom, J. Ljunggren,
J.-A. Gustaffson and M. Carlquist, DOI: 10.2210/pdb2j7y/
pdb, to be published.
Acknowledgements
We thank the Michael J. Fox Foundation for Parkinson's
Research for nancial support and one of the reviewers of the
manuscript for the helpful and insightful comments regarding
the molecular modelling.
Notes and references
1 M. Asim, M. El-Salti, Y. Qian, C. Choueiri, S. Salari,
J. Cheng, H. Shadnia, M. Bal, M. A. Christine Pratt,
K. E. Carlson, J. A. Katzenellenbogen, J. S. Wright and
T. Durst, Bioorg. Med. Chem. Lett., 2009, 19, 1250–1253.
2 C. Leranth, R. H. Roth, J. D. Elsworth, F. Naolin,
T. L. Horvath and D. E. Redmond, J. Neurosci., 2000, 20,
8604–8609.
3 C. Behl, T. Skutella, F. Lezoualc'h, A. Post, M. Widmann,
C. J. Newton and F. Holsboer, Mol. Pharmacol., 1997, 51,
535–541.
¨
25 Schrodinger Suite 2012 the Maestro, version 9.3,
¨
Schrodinger, LLC, New York, 2012.
4 L. B. Morales, K. K. Loo, H. B. Liu, C. Peterson, S. Tiwari-
Woodruff and R. R. Voskuhl, J. Neurosci., 2006, 26, 6823–
6833.
5 L. Bergkvist, H. O. Adami, I. Persson, R. Hoover and
C. Schairer, N. Engl. J. Med., 1989, 321, 293–297.
´
26 H. Sunden, J.-N. Ma, L. K. Hansen, A.-L. Gustavsson,
E. S. Burstein and R. Olsson, ChemMedChem, 2013, 8,
1283–1294.
27 A. C. W. Pike, A. M. Brzozowski, J. Walton, R. E. Hubbard,
˚
A.-G. Thorsell, Y.-L. Li, J.-A. Gustafsson and M. Carlquist,
6 G. G. J. M. Kuiper, B. Carlsson, K. Grandien, E. Enmark,
˚
Structure, 2001, 9, 145–153.
¨
J. Haggblad, S. Nilsson and J.-A. Gustafsson, Endocrinology,
28 S. S. Rajan, Y. Kim, K. Vanek, A. Joachimiak and
G. L. Greene, DOI: 10.2210/pdb3q95/pdb, to be published.
29 S. Nilsson, K. F. Koehler and J. A. Gustafsson, Nat. Rev. Drug
Discovery, 2011, 10, 778–792.
30 R. R. Wilkening, R. W. Ratcliffe, E. C. Tynebor,
K. J. Wildonger, A. K. Fried, M. L. Hammond, R. T. Mosley,
P. M. D. Fitzgerald, N. Sharma, B. M. McKeever, S. Nilsson,
M. Carlquist, A. Thorsell, L. Locco, R. Katz, K. Frisch,
E. T. Birzin, H. A. Wilkinson, S. Mitra, S. Cai, E. C. Hayes,
J. M. Schaeffer and S. P. Rohrer, Bioorg. Med. Chem. Lett.,
2006, 16, 3489–3494.
1997, 138, 863–870.
7 S. M. Gold and R. R. Voskuhl, J. Neurol. Sci., 2009, 286, 99–
103.
8 N. L. Sicotte, S. M. Liva, R. Klutch, P. Pfeiffer, S. Bouvier,
S. Odesa, T. C. Wu and R. R. Voskuhl, Ann. Neurol., 2002,
52, 421–428.
9 ClinicalTrials.gov
NCT01466114.
10 T. Barkhem, B. Carlsson, Y. Nilsson, E. Enmark,
J. Gustafsson and S. Nilsson, Mol. Pharmacol., 1998, 54,
105–112.
identiers:
NCT00451204
and
Med. Chem. Commun.
This journal is ª The Royal Society of Chemistry 2013