A novel estrogen receptor ligand template

Article abstract of DOI:10.1016/S0960-894X(03)00307-X

Three synthetic routes towards a novel estrogen receptor ligand template based on a rigid bicyclo-[3.3.1]-nonene core have been investigated. The prototype compound exhibits potent binding at the ERβ receptor and promising estrogen receptor subtype select

Full text of DOI:10.1016/S0960-894X(03)00307-X

Bioorganic & Medicinal Chemistry Letters 13 (2003) 1919–1922
A Novel Estrogen Receptor Ligand Template
Robert Sibley,* Holia Hatoum-Mokdad, Robert Schoenleber, Laszlo Musza,
William Stirtan, Diana Marrero, William Carley, Hong Xiao and Jacques Dumas
Bayer Research Center, Bayer Corporation, Pharmaceutical Division, 400 Morgan Lane, West Haven, CT 06516, USA
Received 13 December 2002; accepted 14 February 2003
Abstract—Three synthetic routes towards a novel estrogen receptor ligand template based on a rigid bicyclo-[3.3.1]-nonene core
have been investigated. The prototype compound exhibits potent binding at the ERb receptor and promising estrogen receptor
subtype selectivity.
# 2003 Elsevier Science Ltd. All rights reserved.
Estrogen receptors (ER) have quickly emerged as
attractive targets for therapeutic intervention in a wide
variety of diseases, including osteoporosis1 and cancer.2
The marketed drug raloxifene3 (1, Fig. 1) shows potent
binding at both ERa and ERb nuclear receptors, and
combines unique pharmacological and pharmacokinetic
properties. Recently, a large number of programs have
focused on core modifications.4 As part of a discovery
effort to find novel estrogen receptor modulator tem-
plates, the bicyclic ether 25 was identified by high
throughput screening.6
We predicted that the correct relative stereochemistry of
the phenyl and the hydroxymethylene groups should be
trans as in estradiol 4.
Subsequently, we observed that ether 2 is acid-sensitive
(presumably due to easy formation of a benzylic cation),
which makes it a poor lead template structure for an
orally administered drug. We decided to embark on the
preparation of a simplified bicyclo-[3.3.1]-nonene 3 (Fig.
2), possessing a similar rigid template.
Figure 2. Stable analogue of bicyclic ether 2.
Figure 1. Estrogen receptor modulators.
*Corresponding author. Tel.:+1-203-812-5395; e-mail: robert.sibley.b
@bayer.com
Figure 3. Proposed synthetic routes to bicyclo-[3.3.1]-nonenes.
0960-894X/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00307-X

1920
R. Sibley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1919–1922
Three synthetic routes to the bicyclic core of 3 were
successively investigated (Fig. 3).
Results and Discussion
The three routes differ by the nature of the bond formed
by trans-annular cyclization. Route A is the shortest,
and relies on a radical cyclization (Fig. 4).
Figure 4. Route A: (a) PhCH3, reflux, 38%; (b) LDA, THF, À78 ꢀC to
rt, 38%; (c) Bu3SnH, AIBN, tBuOH.
The known diene 57 was reacted with tert-butyl acrylate
to form only cyclohexene 6 and none of the undesired
regioisomer.8 Alkylation of ester 6 with E/Z 1,3-
dibromopropene using standard conditions affords 7 as
a mixture of isomers. This mixture was not separated, as
the E- and Z-vinyl radicals are known to interconvert.
Unfortunately, when we subjected 7 to typical radical
cyclization conditions, only the reduced analogue 8
could be isolated, suggesting an intramolecular hydro-
gen transfer. This route was therefore abandoned.
Figure 5. Route B: (a) Pd2(dba)3, NMP, P(Ph)3, 0 ꢀC to rt, 25%;
(b) methyl acrylate, PhCH3, reflux, 64%; (c) H2, cat Pd/C, EtOH,
47%; (d) TBAF, THF, rt, 96%; (e) MsCl, pyridine, CH2Cl2, rt, 57%;
(f) CBr4, Ph3P, CH2Cl2, rt, 80%.
Route B uses an intramolecular enolate cyclization,
resulting in the formation of a bicyclo-[3.3.1]-nonane
framework (Fig. 5). Bromoolefin 99 was coupled with
stannane 1010 using standard palladium-catalyzed reac-
tion conditions, in moderate to good yield. The use of a
trimethyl stannane instead of a tributyl stannane short-
ened the reaction time and improved the overall yield.
Diene 11 is stable at room temperature, and smoothly
reacts with methyl acrylate in refluxing toluene, to
afford cyclohexenes 12a and 12b in a circa 2:3 ratio.
This mixture, which could not be separated, was
hydrogenated, affording cyclohexane 13 as a mixture of
isomers. The latter was deprotected and then converted
to the corresponding mesylate 15a using a standard
protocol. Alternatively, bromide 15b was also obtained.
Unfortunately, intramolecular cyclization of either
mesylate 15a or bromide 15b using LDA in THF with
or without HMPA did not proceed, as only dimeric
products could be isolated.
Route C was finally investigated, despite the fact that it
is not stereoselective, because it forms the other six-
membered ring as a key step (Fig. 6).
Figure 6. Route C: (a) LDA, THF, oxirane, BF3–OEt2, À78 to 10 ꢀC,
31%; (b) PTSA, PhCH3, reflux, 20 min, quant; (c) DIBAL-H, THF,
À78 to À15 ꢀC, 43%; (d) TBDPSCl, DMF, cat DBU, imidazole, 0 ꢀC
to rt, 57%; (e) allyltriphenyl phosphonium chloride, KOtBu, THF,
À78 ꢀC to rt, 63%; (f) benzyl acrylate, o-xylene, reflux, 30 h, 75%;
(g) TBAF, THF, rt, 16 h, 62%; (h) MsCl, pyridine, CH2Cl2, 0 ꢀC to rt,
95%; (i) LDA, HMPA, THF, À78 ꢀC to rt, 4%; (j) LAH, THF, 0 ꢀC
to rt, quant; (k) BBr3, CH2Cl2, À80 ꢀC to rt, 35%.
The enolate derived from ester 17 was alkylated with
oxirane, cyclized into a lactone, and reduced with dii-
sobutyl aluminum hydride to afford lactol 18. The latter

R. Sibley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1919–1922
Table 1. In vitro pharmacology of compounds 1–4
1921
nonene skeleton, using an 11-step route. While the pre-
paration of our first analogue is not stereoselective and
remains impractical on large scale, the existence of this
new series provides an important starting point to
future studies focusing on estrogen receptor subtype
selectivity, and its impact on pharmacology.
Acknowledgements
We would like to thank J. Brice for HPLC purification
of bicyclic analogue 3, E. Medvedeff for the preparation of
a mouse monoclonal ERb antibody, as well as D. Brittelli,
B. Dixon, and Prof. D. Curran for helpful discussions.
1
2
3
4
References and Notes
ERa (IC50, nM)
ERb (IC50, nM)
ERa/ERb
pS2 expression in MCF-7 (EC50, nM)
pS2 expression in MCF-7
(% agonist compared to E2)
7
470
0.015
ND
0
3200
850
3.8
20
149
550
75
7.3
<10 0.01
139 100
2
2
1
1. Gowen, M.; Emery, J. G.; Kumar, S. Emerg. Drugs 2000, 5,
1.
2. Dhingra, K. Invest. New Drugs 1999, 17, 285.
3. Hagmeyer, K. O.; Meyer, T. K. J. Pharm. Technol. 1999,
15, 37.
4. Smith, R. A.; Chen, J.; Mader, M. M.; Muegge, I.; Moeh-
ler, U.; Katti, S.; Marrero, D.; Stirtan, W. G.; Weaver, D. R.;
Xiao, H.; Carley, W. Bioorg. Med. Chem. Lett. 2002, 12, 2875.
For a recent review on SERM templates, see: Fink, B. E.;
Mortensen, D. S.; Stauffer, S. R.; Aron, Z. D.; Katzen-
ellenbogen, J. A. Chem. Biol. 1999, 6, 205.
5. 1H NMR analysis of bicyclic ether 2 shows that this sample
consists of a 9:1 mixture of two isomers. Stereochemical con-
siderations suggest that the major isomer is also the active
ingredient.
did not react with allyl-triphenyl phosphonium chloride
under Wittig conditions, and was therefore opened
and protected as the silyl ether 19, which in turn was
converted into diene 20. When diene 20 was sub-
jected to benzyl acrylate in refluxing benzene, a very
complex mixture of isomers was formed, and carried
on in the next three steps (desilylation, mesylation,
and intramolecular cyclization). No attempt was
made to separate these mixtures. We expected to
obtain a mixture of isomeric bicyclo-[3.3.1]-nonanes
and bicyclo-[4.3.0]-nonanes from the key cyclization
step. This, however, did not turn out to be the case,
as the desired cyclization product 24 was the only
monomeric product observed among a mixture of
dimerization products. Furthermore, product 24 could
easily be isolated by flash chromatography, albeit in
very low yield. It can be hypothesized that the relative
success of this cyclization is due to a favorable chair-
like transition state. Ester 24 was converted to our
target molecule 3 by reduction with lithium aluminum
hydride followed by demethylation with boron tri-
bromide. The relative stereochemistry of racemic alco-
hol 3 was established by NMR.11 Bicyclic analogue 3
exhibits a very intriguing in vitro pharmacological
profile (Table 1).
6. Estrogen receptors (ER) a and b were purchased from
PanVera Corporation (Madison, WI, USA). [2,4,6,7,16,17-3H]
Estradiol (157 Ci/mmol) and antimouse antibody binding
scintillation proximity assay (SPA) beads were purchased from
Amersham Pharmacia Biotech (Piscataway, NJ, USA). Mouse
monoclonal ER antibody was purchased from Santa Cruz
Biotechnology Inc. (Santa Cruz, CA, USA) and mouse
monoclonal ERb antibody was prepared in-house. SPA
microtitre plates were purchased from Wallac (Gaithersburg,
MD, USA). Using an antibody binding SPA format, reactions
were conducted in 100 mL of assay buffer (50 mM HEPES pH
7.4, 0.5% BSA, 0.2% Tween 20, 1.25% glycerol, 150 mM
NaCl) in the presence or absence of test compounds. Test
compounds were prepared in 5% DMSO, delivered to the
assay plate at final concentrations ranging from 1 to 10,000
nM in 0.25% DMSO (final concentration). For ERa binding
assays the reaction mixture contained ERa (0.02 mg/well),
mouse monoclonal ER antibody (0.2 mg/well), and
[2,4,6,7,16,17-3H6] estradiol (2 nM). ERb binding assays con-
tained ERb (0.03 mg/well), mouse monoclonal ERb antibody
(0.03 mg/well), and [2,4,6,7,16,17-3H6] estradiol (4.8 nM).
Antimouse antibody SPA binding beads were then added to
each well (0.25 mg/well), the reaction mixture was equilibrated
(3 h), the plate centrifuged (2000 rpm, 10 min), and counted in
a Microbeta scintillation counter (Wallac, Inc.). Raloxifene 1
was used as a positive control, unlabeled estradiol 4 was used
as a non-specific binding control.
According to our data,12 bicyclic alcohol 3 shows
improved binding to ERb, when compared to our ori-
ginal analogue 2. This may be due to increased
lipophilicity, or possibly an unfavorable interaction of
the methyl group in 2 with the receptor. In addition,
there is an apparent reversal of ERa/ERb selectivity by
3 compared to raloxifene 1.13 Unlike raloxifene, both
analogues 2 and 3 show full agonistic activity in MCF-7
cells14 (pS215 gene induction with bDNA detection).
In conclusion, we have prepared a new estrogen recep-
tor ligand prototype, featuring a novel bicyclo-[3.3.1]-

1922
R. Sibley et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1919–1922
7. Crenshaw, R. R.; Luke, G. M.; Jenks, T. A.; Partyka, R. A.;
Bialy, G.; Bierwagen, M. E. J. Med. Chem. 1973, 16, 813.
8. When using benzyl or methyl acrylate as the dienophile, the
ratio of the desired regioisomer 6 to the undesired was 4:1 in
each case.
stereochemical assignment of bicyclic analogue 3 is the obser-
vation of cross-peaks between the olefinic proton H8 and the
aromatic proton H11 in the ROESY spectrum, indicating
spatial proximity in agreement with distance calculations. The
presence of a strong cross-peak between protons H2 and H9a
provides further evidence for the assigned stereochemistry.
12. Leatherbarrow, R. J. Grafit, Version 4.0; Erithacus Soft-
ware: Staines, UK, 1998. Calculated IC50 values showed an
error of ꢁ15%.
9. Rappoport, Z.; Gal, A. J. Chem. Soc., Perkin Trans. 2
1973, 301.
10. Stannane 10 is prepared from known 2-bromo-pent-1-ene-
5-ol in two steps (tBuPh2SiCl, DMF, imidazole, 75 ꢀC, 2 h,
then t-BuLi, THF, Me3SnCl, À78 ꢀC to rt).
13. While our data indicate that raloxifene is approximately
70-fold selective for ERa over ERb, the number has histori-
cally been 10- to 20-fold: see: Miller, C. P. Curr. Pharm. Des.
2002, 8, 2089. Miller, C. P.; Komm, B. S. Ann. Rep. Med.
Chem. 2001, 36, 149, and references cited therein for informa-
tion on the binding of raloxifene and other SERMs to ERa
and ERb.
11. NMR spectra of 3 were acquired on a Bruker DMX-500
spectrometer equipped with a three-axis-gradient inverse triple
resonance probe, with the exception of the 13C spectrum,
which was obtained using a heteronuclear broadband probe.
The sample was dissolved in DMSO-d6, and the chemical
shifts were referenced to the solvent signals. A gradient
enhanced COSY spectrum was employed to elucidate the net-
14. MCF-7 cells were obtained from the ATCC and grown in
Growth Medium (GM, see below). The cells are grown to
80% confluency and trypsinized and washed to remove phenol
red, then plated at 10,000 cells/well in Starve Medium (SM, see
below). After 2 days, the medium is removed from the cells
and either fresh SM or SM plus test compound is added (100
mL total volume). The next day the bDNA assay is started by
adding Capture Hybridization Buffer to each capture well of
the bDNA plate (all solutions supplied with QuantiGene kit).
Subsequently, medium is removed from the 96-well plate con-
taining cells and then the solution of lysis buffer and oligomers
(designed by ProbeDesigner 1.0 for the gene of interest) is
added to the 96-well plate (100 mL/well). The cell and lysis/
oligomer mixture is incubated for 15 min at 53 ꢀC and then
vortexed for 1 min. The lysate is then transferred to the bDNA
capture plate and mixed. The plate is then sealed and incu-
bated at 53 ꢀC for 16 h. The next day, the plate is cooled for 10
min at room temperature and the Amplifier Reagent mixture
is prepared. The wells are then washed twice with 200 mL of
Wash Solution A. Amplifier Reagent is added (50 mL/well)
and the plate is sealed again and incubated at 53 ꢀC for 30
min. The plate is then cooled for 10 min at room temperature
(rt) and the Label Probe Reagent is prepared. The plate is
washed twice with 200 mL of Wash Solution A and 50 mL/well
of Label Probe Reagent is added. The plate is sealed and
incubated at 53 ꢀC for 15 min. The plate is cooled for 10 min
at rt and the Substrate Mixture is prepared. The plate is again
washed twice with 200 mL Wash Solution A and then twice
with Wash Solution D. The Substrate Mixture is then added
(50 mL) and the plate is sealed. The plate is incubated at 37 ꢀC
for 30 min and then read on the Quantiplex luminometer.
Raloxifene is used as a positive control in this assay, and
behaves as a full antagonist of estrogen-induced pS2 expres-
sion.
Growth Medium (GM): MEM (GIBCO) with 10% HyClone
heat inactivated fetal bovine serum (FCS), 1 mM sodium pyr-
uvate solution (GIBCO), 0.1 mM non-essential amino acid
solution (GIBCO), 2 mM l-Glutamine (GIBCO), 100 U/mL
penicillin G sodium, 100 mg/mL streptomycin sulfate and 0.25
mg/mL amphotericin B (GIBCO).
Starve Medium (SM): MEM (Without Phenol Red,
GIBCO) with 5% HyClone charcoal/Dextran Treated Fetal
bovine serum (the remainder of the medium is identical to GM
above).
15. pS2 is an estrogen-responsive gene identified in
breast cancer cell lines. Rio, M. C.; Chambon, P. Can-
cer Cells 1990, 2, 269. Davidson, N. E.; Bronzert, D. A.;
Chambon, P.; Gelmann, E. P.; Lippman, M. E. Cancer Res.
1986, 46, 1904.
1
work of coupled protons. One-bond H–13C correlations were
obtained from a phase-sensitive HMQC experiment. Long-
range 1H–13C couplings from a gradient enhanced HMBC
experiment were utilized to confirm the assignments. For a
qualitative estimate of interproton distances, a transverse
NOE experiment was collected in phase sensitive mode using
time proportional phase incrementation.
1H (ppm)
13C (ppm)
1
2
3a
3b
4a
4b
1.88 m
2.57 m
1.83 m
1.59 m
1.97 m
1.25 m
48.6
51.8
33.3
35.3
5
6a
6b
7
8
9a
9b
10
11
12
13
14a
14b
13 OH
14 OH
43.7
22.0
1.94 m
1.86 m
5.62 m
5.36 m
1.68 m
1.33 m
126.1
128.1
29.0
133.8
127.9
115.0
155.5
65.6
7.00 d (J=8.5 Hz)
6.66 d (J=8.5 Hz)
3.27 dd (J=5.5 and 10.5 Hz)
3.20 dd (J=5.5 and 10.5 Hz)
9.13 s
4.56 t (J=5.5 Hz)
Elucidation of relative stereochemistry was based on NOE
data in comparison with molecular models. Molecular
mechanics calculations were performed on a Silicon Graphics
Indy workstation using Tripos force fields and the Gasteiger
and Huckel method for charges as implemented in SYBYL
version 6.6 (Tripos, St. Louis, MO, USA). The basis for the