Triarylethylene bisphenols with a novel cycle are ligands for the estrogen receptor

Article abstract of DOI:10.1016/S0968-0896(00)00016-X

We have prepared a series of triarylethylene and triarylethane systems, analogues of the selective antiestrogen tamoxifen, in which the alkyl substituent is tethered to the distal, rather than the proximal aryl ring by a 5-, 6-, or 7-membered carbocycle. This unusual cyclic structure rigidifies the ligand and adds bulk in a manner that is different from the more typical cyclization to the proximal aryl ring, as in the antiestrogen nafoxidine. These new systems were prepared efficiently by the addition of a benzylic sodium reagent, generated from the corresponding chloride by treatment with sodium naphthalenide, to a doubly protected 4,4'-dihydroxybenzophenone, followed by dehydration and deprotection. In all cases, formation of the exocyclic alkene was preferred. Two of the corresponding alkanes could be obtained by catalytic hydrogenation. All of these compounds have relatively high binding affinity for the estrogen receptor, and some of them demonstrate a significant level of affinity selectivity for the estrogen receptor alpha subtype. Accommodation of these newly rigidified cyclic triarylethylene systems into the ligand-binding pocket of the estrogen receptor can be visualized by molecular modeling. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0968-0896(00)00016-X

Bioorganic & Medicinal Chemistry 8 (2000) 785±793
Triarylethylene Bisphenols with a Novel Cycle are Ligands for the
Estrogen Receptor
Sung-Hoon Kim a,b and John A. Katzenellenbogen a,*
aDepartment of Chemistry, University of Illinois, 600 South Mathews Avenue, 461 Roger Adams Lab, Box 37-5, Urbana IL 61801, USA
bKorea Institute of Science and Technology, PO Box 131, Cheongryang, Seoul 13 0-650, South Korea
Received 21 July 1999; accepted 25 October 1999
AbstractÐWe have prepared a series of triarylethylene and triarylethane systems, analogues of the selective antiestrogen tamoxifen,
in which the alkyl substituent is tethered to the distal, rather than the proximal aryl ring by a 5-, 6-, or 7-membered carbocycle. This
unusual cyclic structure rigidi®es the ligand and adds bulk in a manner that is di€erent from the more typical cyclization to the
proximal aryl ring, as in the antiestrogen nafoxidine. These new systems were prepared eciently by the addition of a benzylic
sodium reagent, generated from the corresponding chloride by treatment with sodium naphthalenide, to a doubly protected 4,40-
dihydroxybenzophenone, followed by dehydration and deprotection. In all cases, formation of the exocyclic alkene was preferred.
Two of the corresponding alkanes could be obtained by catalytic hydrogenation. All of these compounds have relatively high
binding anity for the estrogen receptor, and some of them demonstrate a signi®cant level of anity selectivity for the estrogen
receptor alpha subtype. Accommodation of these newly rigidi®ed cyclic triarylethylene systems into the ligand-binding pocket of
the estrogen receptor can be visualized by molecular modeling. # 2000 Elsevier Science Ltd. All rights reserved.
Introduction
stereo- and structural speci®city, but its preference of
ligand shape and size can be rather eclectic: sometimes
closely related analogues bind with very di€erent a-
nities, whereas compounds with greatly di€ering struc-
tures may both bind very well.7 In fact, the structural
scope of non-steroidal estrogens having high anity for
ER is remarkable, being uni®ed largely by the basic
requirement for at least one p-monosubstituted or
m,p-disubstituted phenol. The recent solution of the
structure of the ER ligand binding domain by X-ray
crystallography provides a new approach for under-
standing ER structure±binding anity relationships.8,9
Estrogens have long been known to play a major role in
regulating the development and function of the princi-
pal tissues of the female reproductive systems: uterus,
ovaries, and breast. More recently, there has been a
greater appreciation of the fact that estrogens also have
important e€ects on many other organ systems: bone,
liver, cardiovascular system, brain, etc.1 Estrogen phar-
maceuticals have been developed for hormone replace-
ment in menopausal women, the most promising of
these being the recently described selective estrogen
receptor modulators (SERMs), such as raloxifene.2
Estrogen antagonists, such as the mixed antiestrogens
tamoxifen and raloxifene, and the pure antiestrogen ICI
182,780, are used for the treatment and the prevention
of breast cancer, in some cases in experimental clinical
protocols.3±5
As part of our interest in the development of ER ligands
of novel structure, we were struck by a curious fact.
Among the structural class of triphenylethylene ligands,
of which tamoxifen is a member, there are many exam-
ples of ligands in which rigidity and in some cases bulk,
has been introduced by formation of an additional
cycle. This cycle, however, has always linked carbon
atoms of the ethyl group back to the proximal phenyl
ring (Fig. 1), which in tamoxifen would prevent cis±
trans isomerism, a process that a€ects the antagonist
character of this antiestrogen.10 The additional ring
takes varied forms:11 It has been homocyclic, as in the
diphenylindenes,12 the diphenyl-dihydronaphthalene
nafoxidine,13,14 and other tetrahydro analogues,15 and
tamoxifen analogues of larger ring sizes,16 and it has
Estrogens, whether agonists or antagonists, act through
the estrogen receptor (ER). This protein is a member of
the nuclear hormone receptor gene superfamily, and it
functions as a ligand-modulated transcription factor.6
ER binds many ligands with high anity and great
*Corrresponding author. Tel.: +1-217-333-6310; fax: +1-217-333-
7325; e-mail: jkatzene@uiuc.edu
0968-0896/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(00)00016-X

786
S.-H. Kim, J. A. Katzenellenbogen / Bioorg. Med. Chem. 8 (2000) 785±793
Figure 1. Modes of ring closure from a tamoxifen-like triarylethylene antiestrogen. The common cyclization of the ethyl group to the proximal aryl
ring creates systems that resemble nafoxidine and its analogues and congeners, whereas the less common cyclization to the distal aryl ring creates
systems that resemble benzocyclofenil.
been heterocyclic, as in the benzofurans,11 benzothio-
phenes,2 benzopyrans,17 and the indoles.18 In all cases,
the fused bicyclic system that is formed by this addi-
tional ring is thought to constitute a structural mimic of
the A and B rings of estradiol. This was shown in the
the ligand that is close to the D-ring of estradiol. It has
become clear from recent crystal structures of ER±
ligand complexes that this region of the ligand-binding
pocket is rather ¯exible, so that it can readily be
reshaped by ligands of di€erent size and shape, the
SERMs in particular.8,9,21 This reshaping of the ER
complex might result in di€erent degrees of agonist and
antagonist character, and the generation of useful tissue
selectivity.1 It is also possible that such ligands might
show preferential binding anity and/or potency
towards one of the two ER subtypes, ERa or ERb.22 In
this report, we describe the synthesis of a series of
triarylethylene and triarylethane bisphenols having this
novel cyclic topology. We investigate their binding a-
nity for the ER, and model their ®t into the binding
pocket of ERa.
.
recent crystal structure of the ER raloxifene complex, in
which the benzothiophene of raloxifene was shown to
be bound in a fashion very similar to that of the A and
B rings of estradiol.8
We were curious whether alternative cycles might be
introduced into triarylethylene estrogens that would
provide high anity ligands having somewhat di€erent
shape, which in turn could reshape the ER, producing
di€ering agonist/antagonist character and tissue selec-
tivity. In particular, we wondered whether triaryl-
ethylene ER ligands in which carbon atoms of the ethyl
group are linked forward to the distal phenyl ring would
retain high ER binding anity. Although cyclic linkage
to the distal phenyl ring would not prevent ole®n iso-
merization, in the more symmetrical systems we have
examined here, this isomerism is of no consequence.
Results and Discussion
Synthesis of the cyclic bisphenols
An obvious approach to the synthesis of the new cyclic
triarylethylene target systems involves conjoining a
protected p,p0-hydoxybenzophenone with a benzo-cyclic
ketone by a reductive process, such as the McMurray
coupling with low valent titanium, exempli®ed in
Scheme 1. Such an approach has been used successfully
in the acyclic series for the synthesis of tamoxifen itself23
and with one of the cyclic systems we have investi-
gated.20 In our hands, however, McMurray coupling
between p,p0-dimethoxybenzophenone and the cyclic
The novel tetracyclic system formed by this alternative
ring closure can be considered to be an analogue of
cyclofenil, a non-steroidal bisphenol with high anity
for ER that has partial antagonist activity.19 ER is
known to be tolerant of substituents on the cyclohexyl-
idine ring of cyclofenil, and one of the analogues that
we have prepared (see below) is a benzocyclofenil deri-
vative.20 In addition, this new structural modi®cation
would introduce bulk and shape variation in a region of

S.-H. Kim, J. A. Katzenellenbogen / Bioorg. Med. Chem. 8 (2000) 785±793
787
Scheme 1.
ketones proved to be sluggish and proceeded in rela-
tively poor yield.
good yields, the reaction presumably proceeding
through the labile benzhydrol intermediate 5. Only the
isomers with exocyclic double bonds (3a±c) were
obtained at this stage. Subsequent deprotection to the
bisphenols (4a±c) with boron tribromide proceeded
smoothly with the 5- and 7-membered ring systems
(Scheme 3). The six-membered ring system, however,
gave the bisphenol as a mixture of exo- and endo-cyclic
double bond isomers, 4b and 4b0, respectively, in which
the undesired endo-cyclic isomer(4b0) predominated.
By contrast, an alternate though related approach, that
involves the addition of an organosodium reagent
derived by the corresponding benzo-cyclic chlorides
(2a±c) to p,p0-dimethoxybenzophenone (1a), worked
very well (Scheme 2). The requisite chlorides (2a±c) were
obtained from the cyclic ketones by borohydride reduc-
tion, followed by treatment with thionyl chloride.24,25
Treatment of the ketones directly with lithium metal
resulted predominantly in Wurtz coupling. However, we
found that treatment with sodium naphthalenide readily
furnished the benzylic organosodium reagents.26 Sub-
sequent treatment with 4,40-dimethoxybenzophenone
gave after workup the desired ole®ns (3a±c) directly in
It proved to be impractical to separate these regioisomers
by crystallization or chromatography. Fortunately, by
replacing the methyl ether phenol protecting group in
ketone 1a with the more labile trimethylsilyl group (1b),
we were able to e€ect the addition of the organosodium
Scheme 2.
Scheme 3.

788
S.-H. Kim, J. A. Katzenellenbogen / Bioorg. Med. Chem. 8 (2000) 785±793
reagent and the dehydration±deprotection conversions
under milder conditions that gave exclusively the exo-
cyclic double bond isomer 4b in high yield (Scheme 3).
six-membered cycle (4b and 7b) have the highest anity
(except for 4c with ERa); a moderate, but variable
degree of ER subtype anity selectivity is apparent in
the ethylene series, with the six membered analogue 4b
favoring ERb, but the seven-membered analogue 4c
favoring ERa.
The 5- and 6-membered ring triarylethylene systems (3a
and 3b) were readily hydrogenated over a platinum
oxide catalyst to give the corresponding saturated cyclic
triarylethane systems 6a and 6b (Scheme 4). The 7-
membered ring ketone (3c), curiously, was resistant to
hydrogenation under these and even more vigorous
conditions (up to 50 psi hydrogen over a platinum oxide
catalyst). Deprotection as before gave the two saturated
bisphenols, 7a and 7b. The unsaturated bisphenol 4b
could also be hydrogenated directly to the saturated
analogue 7b.
Perhaps most remarkable is the fact that the anities
measured with the pure human ERa preparations are
uniformly much higher than those measured in lamb
uterine ER preparations, which, based on results from
other species, are presumed to contain mostly ERa.22 It
is unlikely that this is the result of a species di€erence in
ER ligand binding preference, because such e€ects are
rather rare among the estrogens.21 Rather, it is more
likely a consequence of the high lipophilicity of these
compounds. This high lipophilicity is evident from
comparisons of the calculated octanol±water partition
coecients (log P values), which are also given in Table
1. Compared to estradiol, which is both the tracer and
the standard used in the competitive binding anity
assay, all of the non-steroidal compounds are predicted
to be very lipophilic, some up to 100-fold more lipo-
philic. As a result, when the assay is done in uterine
cytosol, the non-steroidal ligands would be more highly
bound to non-receptor proteins (i.e. proteins that con-
stitute the ``non-speci®c'' binding fraction) than would
estradiol. This would reduce their free concentration
and result in a lower measured competitive binding
anity. Such e€ects of reduced apparent anities of
lipophilic compounds in competitive binding assays
have been noted before.29 The puri®ed ERa and ERb
preparations should contain none of these ``non-speci®c''
proteins, so that the RBA values obtained here would
represent their inherent anities for these ERs. The log
P value for hydroxy-tamoxifen is also high, but the
uterine cytosol and pure ERa anity values are not
discordant. This compound, however, is an amine,
which at physiological pH will be protonated, so it may
Estrogen receptor binding anity of the cyclic bisphenols
The binding anity of the triarylethylenes (4a±c) and
the triarylethanes (7a±b) for the estrogen receptor
(ER) was determined in a competitive radiometric
binding assay, using tritium-labeled estradiol as tra-
cer.27,28 ER preparations used in these experiments were
lamb uterine cytosol28 or commercially obtained pur-
i®ed human ERa or ERb, expressed in a baculovirus
system.27
The binding anity of these new compounds is recor-
ded in Table 1, together with the anity of a few related
non-steroidal estrogens. Cyclization of the ethyl group
to the distal phenyl substituent produces ligands that
bind to the estrogen receptor with an anity compar-
able to that of either the acyclic triarylethylene systems
(exempli®ed by hydroxy-tamoxifen), as well as to those
in which the cycle is made with the proximal phenol
(exempli®ed by desmethyl-nafoxidine), or to the ``parent
system'' cyclofenil (see Fig. 1). There is some variation
in anity among the compounds in this series: In both
the ethylene and ethane case, the analogues with the
Scheme 4.

S.-H. Kim, J. A. Katzenellenbogen / Bioorg. Med. Chem. 8 (2000) 785±793
Table 1. Estrogen receptor binding anities
789
RBAa (%)
Compound
n
log Pb
0 ꢀC
25 ꢀC
ERa
ERb
Estradiol
4a
3.37(3.51)c
4.99
5.41
(100)
7.1
11.2
17.8
1.41
4.5
3.3
16
105
166
(100)
3.98
5.88
3.16
1.78
1.15
1.41
N/A
N/A
N/A
(100)
41.7
132
(100)
42.7
240
63.1
N/A
35.5
47.9
243
1
2
3
3
1
2
4b
4c
5.82
4.72
4.93
5.38
166
(4b0+4b)d
N/Ae
31.6
50.1
152
7a
7b
Cyclofenilf
4.57(4.81)c
5.57
5.68
Desmethylnafoxidinef
Hydroxytamoxifenf
N/A
131
N/A
62
aBinding anities are expressed relative to that of estradiol=100% (RBA=relative binding anity) and are the average of duplicate determina-
tions. The ®rst two determinations (0 and 25 ꢀC) are in lamb uterine cytosol; the last two determinations are in pure human ERa and ERb.
bThese log P values are calculated using the ChemDraw Ultra program.
cThe log P value in parentheses was measured by HPLC.
d4b0: 4b=85:15.
eN/A=not available.
fFor structures, see Figure 1.
have less protein binding than do the neutral analogues
4 and 7.
ligand onto the structure of hydroxytamoxifen (Fig. 2A)
9
.
in a recent ER tamoxifen X-ray crystal structure. Use
of the Flexidock routine within the molecular modeling
program SYBYL (Tripos) then enables this ligand to
be accommodated without van der Waals overlap (Fig.
2B), by making small, energetically allowable adjust-
ments to the side chains of the residues that line the
ligand binding pocket. Although it is not reasonable to
assign binding energies to a structure created in this
manner, the neatness of the ®nal ®t is quite appealing.
Similar structures can be generated, as well, for the 5-
and 6-membered analogues (not shown).
The estrogen receptor is known to accommodate var-
ious non-steroidal ligands of di€erent shape and size,
especially when these variations are at some distance
from the portion of the ligand that mimics the A-ring of
estradiol. Thus, the good anity of these ligands is not
surprising. A picture of how one of these ligands, the
7-membered analogue 4c, might be accommodated in
the binding pocket of the estrogen receptor is illustrated
in Figure 2. This ®gure was created by ®rst docking the
Panel A
Panel B
Figure 2. Overlay of distal cyclic analogue 4c and hydroxytamoxifen (Panel A) and accommodation of analogue 4c into the ligand-binding pocket of
estrogen receptor a (Panel B). The overlay (Panel A) was done using a minimized conformation of compound 4c and the structure of hydroxy-
tamoxifen from its crystal structure with the estrogen receptor,9 and the ligand pocket ®t (Panel B) was done using Flexidock routine within the
modeling program SYBYL (Tripos).

790
S.-H. Kim, J. A. Katzenellenbogen / Bioorg. Med. Chem. 8 (2000) 785±793
Conclusions
Microanalytical Service Laboratory at the University of
Illinois.
To probe the tolerance of the estrogen receptor (ER)
binding to de®ned structural changes in certain non-
steroidal ligands, we have developed an ecient and
generally useful synthesis of triarylethylene ligands for
the estrogen receptor that have a novel ring topology,
created by forming an additional cycle between the ethyl
group and the distal aryl ring. This additional con-
formational constraint alters the shape of these ligands
in a region that corresponds to the D ring of estradiol, a
region where steric changes are known to be well toler-
ated by ER. The compounds could cause a change in
the shape of the ER, which might result in increased
selectivity, or the generation of di€erential agonist/
antagonist character or ERa versus ERb subtype a-
nity and/or potency selectivity. The sensitive benzylic
organosodium reagents that are required by our
approach can be formed conveniently from the corre-
sponding benzylic chlorides using sodium naphthale-
nide, and protecting group strategies were developed so
that only the exocyclic double bond isomers are
obtained. Saturated analogues of two of these cyclic
alkenes were also produced by hydrogenation.
All reactions using water- and air-sensitive reagents
were conducted under a nitrogen atmosphere with dry
solvent. Solvents were distilled as follows: CH2Cl2 from
CaH2 and tetrahydrofuran (THF) from sodium benzo-
phenone ketyl. 1-Chlorobenzocycloalkanes were pre-
pared according to known procedures.24,25 All other
reagents were purchased from commercial suppliers and
were used without further puri®cation.
Biological methods
Binding anity measurements. Competitive radiometric
binding assays were performed with lamb uterine cyto-
sol preparations or puri®ed ERa and ERb (PanVera),
as reported previously.27 The receptor preparation was
diluted to approximately 1.5 nM of receptor and was
incubated with several concentrations of unlabeled
competitor, together with 10 nM [3H]estradiol, for 18±
24 h. Unlabeled competitors were dissolved and diluted
in 1:1 DMF bu€er to ensure solubility; the ®nal DMF
concentration was 7%. The receptor±ligand complexes
were separated from free tracer by adsorption of the
free ligand onto charcoal-dextran when using lamb
uterine cytosol preparations, or by adsorption of the
bound complex onto hydroxylapatite when using pur-
i®ed ERa and ERb preparations.
The binding anities of these new triarylethylene
ligands for lamb uterine cytosol ER or human ERa and
ERb are comparable to those of related acyclic and
cyclic triarylethylene or cyclofenil parental ligands, and
some moderate anity selectivity between the two
human ER subtypes is apparent. Much higher anities
are obtained in assays done with puri®ed receptor pre-
parations than with a uterine cytosol preparation. These
®ndings indicate that the estrogen receptor is remark-
ably shape and size tolerant in the region remote from
the A-ring binding subpocket, a feature that can be
visualized by molecular modeling.
Octanol±water partition coecient measurements. Octa-
nol±water partition coecients, a measure of lipo-
philicity, were estimated from the long k0w values
determined by reversed phase HPLC methods, as pre-
viously reported.31,32
Chemical synthesis
1-[Bis(4-methoxyphenyl)methylidenyl]indane (3a). Sodium
naphthalenide complex was prepared by adding sodium
(138 mg, 6.0 mmol) to a solution containing naphtha-
lene (1.2 g, 10 mmol) in dry THF (40 mL) and stirring
until dissolution of sodium was complete. A mixture of
4,40-dimethoxybenzophenone (1a, 605 mg, 2.5 mmol)
and 1-chloroindane (2a, 458 mg, 3.0 mmol) in THF (25
mL) was added dropwise to this deep-green solution at
0±5 ꢀC under nitrogen atmosphere, over a 1-h period.
After the reaction was complete, the excess of sodium
naphthalenide was destroyed by the addition of few
drops of methanol. Water (50 mL) and ether (100 mL)
were added to the reaction mixture, and the pH of the
water was adjusted to 7 with 3 N HCl solution. The
organic layer was separated and was washed with brine,
water (100 mLÂ1) , and dried over MgSO4. Removal of
the solvent, followed by silica gel ¯ash column chroma-
tography of the residue with hexane (300 mL) removed
the non-polar naphthalene derivatives. Subsequent elu-
tion with CH2Cl2 (150 mL) gave a title compound (3a)
as a colorless solid (555 mg, 65%). Sample for micro-
analysis was prepared by recrystalization from ether;
Experimental
General methods
Melting points (uncorrected) were recorded on Thomas-
Hoover Electrothermal apparatus. Analytical thin-layer
chromatography (TLC) was performed on Merck silica
gel F-254 glass-baked plates, with visualization by
short-wave (254 nm) UV light. Flash column chroma-
tography was performed according to Still30 using
Woelm 32±63 mm silica gel.
1H and 13C NMR were recorded on Varian U400 and
Varian U500 spectrometers. Chemical shifts (d) were
recorded in ppm down®eld from TMS and were refer-
enced to either TMS internal standard or the residual
proton peak in CDCl3, acetone-d6, or methanol-d4 sol-
vent peak. Coupling constants are reported in Hz.
Electron ionization (EI) mass spectra were obtained
using Finnigan-MAT-CH5 spectrometer at 70 eV. Fast
atom bombardment (FAB) mass spectra were recorded
on a VG ZAB-SE spectrometer, whereas a VG VSE-B
instrument provided chemical ionization (CI) mass
spectra. Elemental analysis was performed by the
mp 139±140 ꢀC; H NMR (CDCl3, 500 MHz) d 2.93 (s,
1
4H), 3.81 (s, 3H), 3.84 (s, 3H), 6.49 (d, J=7.5 Hz, 1H),
6.83±6.88 (m, 5H), 7.06 (t, J=7.5 Hz, 1H), 7.13±7.23

S.-H. Kim, J. A. Katzenellenbogen / Bioorg. Med. Chem. 8 (2000) 785±793
791
(m, 5H); 13C NMR (CDCl3, 125 Hz) d 30.71, 34.47,
55.22, 113.23, 114.06, 116.53, 124.88, 125.02, 125.57,
127.07, 130.44, 131.12, 134.22, 135.39, 139.45, 141.68,
147.58, 158.15, 158.64; EI±MS m/z (%) 342 (M+, 100),
221 (9), 121 (9). Anal. calcd for C24H22O2: C, 84.18; H
6.48. Found C, 84.30; H, 6.54.
compound 4a as a colorless solid (170 mg, 82%). Fur-
ther puri®cation for microanalysis and binding anity
assay was carried out by recrystallization from metha-
nol: mp 235±236 ꢀC; H NMR (methanol-d4, 400 MHz)
1
d 2.70 (t, J=7.2 Hz, 2H), 2.90 (t, J=7.2 Hz, 2H), 6.50±
7.10 (m, 12H); 13C NMR (methanol-d4, 100 Hz) d 28.02,
30.50, 113.74, 114.32, 125.35, 125.47, 125.82, 126.31,
128.98, 130.96, 131.74, 134.08, 134.15, 135.40, 136.14,
137.32, 154.94, 155.43; FABMS m/z (%) 314 (M+, 78),
199 (17), 155 (56), 135 (44), 119 (100). Anal. calcd for
C22H18O2: C, 84.05; H 5.77. Found C, 83.82; H, 5.82.
1-[Bis(4-methoxyphenyl)methylidenyl]-1,2,3,4-tetrahydro-
naphthalene (3b). The reaction of 1-chloro-1,2,3,4-tetra-
hydronaphthalene (2b, 500 mg, 3.0 mmol) according to
the procedure described for the preparation of com-
pound 3a, followed by chromatography as described
previously, gave a mixture of 1-[bis(4-methoxyphenyl)-
hydroxymethyl]-1,2,3,4-tetrahydronaphthalene and 3b.
However, after evaporation of the CH2Cl2, only com-
pound (3b) was detected, as a colorless solid (657 mg,
1-[Bis(4-hydroxyphenyl)methylidenyl]-1,2,3,4-tetrahydro-
naphthalene (4b). A. The reaction of 4,40-bis(trimethyl-
silyloxy)benzophenone (1b) with 1-chloro-1,2,3,4-tetra-
hydronaphthalene (2b). 4,40-Di(trimethylsilyloxy)benzo-
phenone (1b, 890 mg, 2.94 mmol) was prepared by the
reaction of 4,40-dihydroxybenzophenone (642 mg, 3.0
mmol) with trimethylsilyl chloride (658 mg, 6.5 mmol)
in THF (30 mL). This material was allowed to react
with 1-chloro-1,2,3,4-tetrahydronaphthalene (2b, 599
mg, 3.6 mmol) and sodium naphthalenide, as described
in the preparation of 3a. Puri®cation, as described, gave
compound 4b as a colorless solid (675 mg, 70%). 4b:
74%). 5: mp 103±104 ꢀC; H NMR (CDCl3, 400 MHz)
1
d 1.40±1.52 (m, 1H), 1.71±1.83 (m, 2H), 1.91±2.02 (m,
1H), 2.03 (s, 1H, OH), 2.60±2.78 (m, 2H), 3.77 (s, 6H,
2ÂOCH3), 4.09 (t, J=6.4 Hz, 1H), 6.54±7.48 (m, 12H);
13C NMR (CDCl3, 125 Hz) d 21.51, 26.33, 30.10, 46.03,
55.18, 55.21, 81.16, 113.34, 113.37, 125.04, 126.38,
127.11, 127.23, 129.18, 130.43, 135.11, 138.71, 140.53,
141.27, 157.86, 158.07; EI±MS m/z (%) 373 (M+ 1, 2),
357 (M+ OH, 59), 356 (39), 267 (49), 243 (100), 153
(21.7). Anal. calcd for C25H26O3: C, 80.18; H 7.00.
Mp 216±217 ꢀC; H NMR (CDCl3, 400 MHz) d 1.77
1
(quintet, J=6.4 Hz, 2H), 2.46 (t, J=6.4 Hz, 2H), 2.78
(t, J=6.4 Hz, 2H), 2.82 (brs, 2H, 2ÂOH), 6.52±7.24 (m,
12H); 13C NMR (methanol-d4, 100 Hz) d 23.95, 29.38,
30.39, 114.43, 114.48, 124.26, 125.81, 127.69, 130.36,
131.51, 132.18, 133.50, 135.65, 135.69, 137.22, 137.92,
138.97, 154.79, 155.08; FABMS m/z (%) 328 (M+, 86),
307 (34), 154 (100), 136 (72). Anal. calcd for C23H20O2:
C, 84.12; H 6.14. Found C, 83.73; H, 6.09.
Found C, 80.28; H, 7.01. 3b: Mp 135±136 ꢀC; H NMR
1
(CDCl3, 400 MHz) d 1.84 (quintet, 2H, J=6.4 Hz), 2.52
(t, 2H, J=6.4 Hz), 2.80 (t, 2H, J=6.4 Hz), 3.75 (s, 3H),
3.83 (s, 3H), 6.66±7.26 (m, 12H): 13C NMR (CDCl3, 125
Hz) d 24.11, 29.56, 30.60, 55.10, 55.22, 113.28, 124.58,
126.23, 127.97, 130.56, 131.61, 132.28, 134.35, 136.58,
136.61, 136.99, 137.90, 139.27, 158.08, 158.31. Anal.
calcd for C25H24O2: C, 84.24; H 6.79. Found C, 84.28;
H, 6.85.
B. The treatment of 3b with boron tribromide. The reac-
tion of 3b (182 mg, 0.5 mmol) with boron tribromide
(1.5 mL, 1 M CH2Cl2 solution) as described in the pre-
paration of compound 4a gave a reaction mixture (175
mg) composed of 4-[bis(4-hydroxyphenyl)methyl]-1,2-
dihydronaphthalene (4b0) and 4b in the ratio of 85:15, as
determined by 1H NMR analysis. Separation of this
mixture of regioisomers by routine methods, using ¯ash
column chromatography and fractional crystallization,
failed. 4b0: Mp 107±110 ꢀC; 1H NMR (CDCl3, 400
MHz) d 2.25 (m, 2H, CH2-vinyl), 2.77 (t, J=7.6 Hz
2H), 4.81(brs , 2H, 2ÂOH), 5.30 (s, 1H, -CH(Ph)2), 6.72
(d, 4H), 7.00±7.24 (m, 8H); FABMS m/z (%) 328 (M+,
80), 307 (37), 154 (100), 136 (70).
5-[Bis(4-methoxyphenyl)methylidenyl]-benzocycloheptene
(3c). The reaction of 5-chloro-5H-benzocycloheptene
(2c, 542 mg, 3.0 mmol) according to the procedure
described for the preparation of compound 3a, followed
by chromatography as described previously, gave 3c as
a colorless solid from hexane; mp 103±104 ꢀC; yield
82%; 1H NMR (CDCl3, 400 MHz) d 1.55±1.86 (m, 6H),
2.86±2.97 (m, 2H), 3.68 (s, 3H), 3.82 (s, 3H), 6.49±7.23
(m, 12H); 13C NMR (CDCl3, 125 Hz) d 27.44, 30.40,
33.51, 36.27, 54.98, 55.21, 112.67, 113.67, 126.02,
126.16, 128.59, 129.60, 130.84, 131.01, 135.58, 135.65,
137.72, 140.72, 141.12, 144.45, 157.45, 158.11; EI±MS
m/z (%) 370 (M+, 100), 227 (55), 121 (14). Anal. calcd
for C26H26O2: C, 84.29; H 7.07. Found C, 84.08; H, 6.98.
5-[Bis(4-hydroxyphenyl)methylidenyl]-benzocycloheptene
(4c). The reaction of 3c (335 mg, 0.9 mmol) with boron
tribromide (3 mL, 1 M CH2Cl2 solution), as described
in the preparation of 4a, followed by ¯ash column
chromatography on silica gel with ether as an eluent,
gave 4c (302 mg, 97%) as a colorless solid. Recrystali-
zation from chloroform was carried out; mp 168±
1-[Bis(4-hydroxyphenyl)methylidenyl]indan (4a). To the
solution of 3a (227 mg, 0.66 mmol) in dichloromethane
(5 mL) was added dropwise boron tribromide (2.5 mL,
1 M CH2Cl2 solution) by syringe at 78 ꢀC under a
nitrogen atmosphere. When the addition was complete,
the reaction mixture was allowed to warm to room
temperature overnight with stirring, and then was
hydrolyzed with few drops of water. Chloroform (15 mL)
and water (5 mL) were added, and the organic phase
was collected, washed with aqueous 5% NaHCO3,
brine, and water, dried over MgSO4, and collected by
®ltration. Evaporation under reduced pressure gave
169.5 ꢀC; H NMR (CDCl3, 400 MHz) d 1.55±1.89 (m,
1
6H), 2.86±2.97 (m, 2H), 4.52 (s, 1H, OH), 4.76 (s, 1H,
OH), 6.49±7.23 (m, 12H); 13C NMR (CDCl3, 100 Hz) d
27.29, 30.26, 33.13, 36.14, 114.04, 115.01, 125.82,
126.04, 128.43, 129.40, 130.87, 131.03, 135.58, 135.61,
137.36, 140.76, 140.95, 144.17, 153.16, 153.84; FABMS
m/z (%) 342 (M+, 58), 307 (30), 289 (18), 154 (100), 136

792
S.-H. Kim, J. A. Katzenellenbogen / Bioorg. Med. Chem. 8 (2000) 785±793
(69). Anal. calcd for C24H22O2: C, 84.18; H 6.48. Found
C, 83.84; H, 6.36.
Kathryn E. Carlson for performing the receptor binding
studies and Marvin J. Meyers for assistance with the
preparation of Figure 2. NMR spectra were obtained in
the Varian Oxford Instrument Center for Excellence in
NMR Laboratory. Funding for this instrumentation
was provided in part from the W. M. Keck Foundation
and the National Science Foundation (NSF CHE 96-
10502). Mass spectra were obtained on instruments
supported by grants from the National Institute of
General Medical Sciences (GM 27029), the National
Institute of Health (RR 01575), and the National Sci-
ence Foundation (PCM 8121494).
1-[Bis(4-hydroxyphenyl)methyl]-1,2,3,4-tetrahydronaph-
thalene (7b). A solution of 4b (20 mg, 0.06 mmol) and a
catalytic amount of platinum oxide in methanol (10 mL)
was stirred overnight under hydrogen atmosphere at
room temperature. After the reaction was complete, the
catalyst was removed by ®ltration. Evaporation of the
solvent gave compound 7b (20 mg, 99%) as a colorless
solid: mp 103±105 ꢀC; H NMR (acetone-d6, 400 MHz)
1
d 1.54±1.66 (m, 2H), 1.78±1.94 (m, 2H), 2.69±2.77 (m,
1H), 2.83±2.90 (m, 1H), 3.53±3.58 (m, 1H), 3.96 (d, 1H,
J=10.8 Hz), 6.42 (d, 1H, J=8.4 Hz), 6.61±6.74 (m, 2H),
6.78 (d, 2H, J=8.4 Hz), 6.93±7.01 (m, 4H), 7.25 (d, 2H,
J=8.4 Hz), 8.06 (brs, 2H, 2ÂOH); 13C NMR (acetone-
d6, 100 Hz) d 17.89, 26.00, 28.19, 42.32, 55.37, 114.63,
115.14, 123.94, 125.48, 128.58, 129.32, 129.48, 130.20,
135.37, 135.96, 136.84, 139.75, 155.40, 155.54; MS(CI)
m/z (%) 331 (M++1, 1.5), 329 ((M+ H2)+1, 4.1), 328
(3.8), 199 ((4,40-dihydroxyphenylmethyl)+, 100), 131
(1,2,3,4-tetrahydronaphtyl+, 12.1); CI-HRMS calcd/
found (C23H23O2 (M++1)) 331.16979/331.16981.
References
1. Katzenellenbogen, J. A.; O'Malley, B. W.; Katzenellenbogen,
B. S. Mol. Endocrinol. 1996, 10, 119.
2. Grese, T. A.; Dodge, J. A. Curr. Pharm. Des. 1998, 4, 71.
3. Wakeling, A. E. Breast Cancer Res. Treat. 1993, 25, 1.
4. Fisher, B.; Costantino, J. P.; Wickerham, D. L.; Redmond,
C. K.; Kavanah, M.; Cronin, W. M.; Vogel, V.; Robidoux, A.;
Dimitrov, N.; Atkins, J.; Daly, M.; Wieand, S.; Tanchiu, E.;
Ford, L.; Wolmark, N. J. Natl. Cancer Inst. 1998, 90, 1371.
5. Jordan, V. C. J. Natl. Cancer Inst. 1998, 90, 967.
6. Katzenellenbogen, J. A.; Katzenellenbogen, B. S. Chem.
Biol. 1996, 3, 529.
1-[Bis(4-Methoxyphenyl)methyl]-indane (6a). This com-
pound was obtained in 90% yield from the reduction of
3a by the same method described in the preparation of
7. Gao, H.; Katzenellenbogen, J. A.; Garg, R.; Hansch, C.
Chem. Rev. 1999, 99, 723.
7b: mp 119.5±120.5 ꢀC; H NMR (CDCl3, 400 MHz) d
1
1.74±1.83 (m, 1H), 2.06±2.15 (m, 1H), 2.75±2.78 (m,
2H), 3.77 (s, 3H), 3.78 (s, 3H), 3.86 (d, J=10.4 Hz, 1H),
3.97±4.03 (m, 1H), 6.43 (d, J=7.6 Hz, 1H), 6.79±6.85
(m, 4H), 6.90 (t, J=7.6 Hz, 1H), 7.10 (t, J=7.6 Hz,
1H), 7.16±7.21 (m, 5H); 13C NMR (CDCl3, 125 Hz) d
30.88, 31.73, 48.99, 55.21, 55.22, 55.26, 113.76, 113.80,
124.31, 125.37, 125.52, 126.45, 128.96, 129.30, 137.05,
137.23, 144.66, 145.98, 157.82, 157.98; FABMS m/z
(%) 345 (M++1, 3.5), 343 ((M+ H2)+1, 7.3), 227
((4,40-dimethoxyphenylmethyl)+, 100), 117 (indanyl+,
6.8); FAB±HRMS calcd/found (C24H25O2 (M++1))
345.18544/345.18545.
8. Brzozowski, A. M.; Pike, A. C.; Dauter, Z.; Hubbard,
R. E.; Bonn, T.; Engstrom, O.; Ohman, L.; Greene, G. L.;
Gustafsson, J.-A.; Carlquist, M. Nature 1997, 389, 753.
9. Shiau, A. K.; Barstad, D.; Loria, P. M.; Cheng, L.; Kushner,
P. J.; Agard, D. A.; Greene, G. L. Cell 1998, 95, 927.
10. Katzenellenbogen, J. A.; Carlson, K. E.; Katzenellenbogen,
B. S. J. Steroid Biochem. 1985, 22, 589.
11. Magarian, R. A.; Overacre, L. B.; Singh, S.; Meyer, K. L.
Curr. Med. Chem. 1994, 1, 61.
12. Anstead, G. M.; Altenbach, R. J.; Wilson, S. R.;
Katzenellenbogen, J. A. J. Med. Chem. 1988, 31, 1316.
13. Bindal, R. D.; Katzenellenbogen, J. A. J. Steroid Biochem.
1985, 23, 929.
14. Bindal, R. D.; Katzenellenbogen, J. A. Photochem. Photo-
biol. 1986, 43, 121.
15. Tatee, T.; Carlson, K. E.; Katzenellenbogen, J. A.;
Robertson, D. W.; Katzenellenbogen, B. S. J. Med. Chem.
1979, 22, 1509.
1-[Bis(4-hydroxyphenyl)methyl]indane (7a). The treat-
ment of 6a with boron tribromide as described in the
preparation of 4a, followed by column chromatography
on silica gel with ether as an eluent ꢀa€orded compond
1
16. McCague, R.; Leclercq, G.; Jordan, V. C. J. Med. Chem.
1988, 31, 1285.
7a as a colorless solid: mp 170±171 C; yield 57%; H
NMR (acetone-d6, 400 MHz) d 1.68±1.76 (m, 1H), 2.01±
2.08 (m, 1H), 2.66±2.84 (m, 2H), 3.77 (d, J=10.8 Hz,
1H), 3.99±4.05 (m, 1H), 6.42 (d, J=7.2 Hz, 1H), 6.81±
6.77 (m, 4H), 6.83 (t, J=7.2 Hz, 1H), 7.03 (t, J=7.2 Hz,
1H), 7.11±7.22 (m, 5H); 13C NMR (CDCl3, 100 Hz) d
30.81, 31.64, 49.01, 55.25, 115.05, 115.11, 116.53, 124.20,
125.39, 126.31, 128.98, 129.34, 136.38, 136.54, 144.65,
146.09, 154.57, 154.74; EI±MS m/z (%) 316 (M+, 1.5),
314 (M+ H2, 2.2), 199 ((4,40-dihydroxyphenylmethyl)+,
100); EI-HRMS calcd/found (C22H20O2) 316.14633/
316.14633.
17. Tremblay, A.; Tremblay, G. B.; Labrie, C.; Labrie, F.;
Giguere, V. Endocrinology 1998, 139, 111.
18. Biberger, C.; Vonangerer, E. J. Steroid Biochem. Mol. Biol.
1998, 64, 277.
19. Einer-Jensen, N. Acta Pharmacol. Toxicol. 1965, 23, 365.
20. Shoji, H.; Ando, S. Triphenylalkenes for Treatment of
Osteoporosis; Shoji, H.; Ando, S., Eds.; Japan, 1993; Vol. JP
05,112,510.
21. Anstead, G. M.; Carlson, K. E.; Katzenellenbogen, J. A.
Steroids 1997, 62, 268.
22. Kuiper, G. G. J. M.; Carlsson, B.; Grandien, K.; Enmark,
E.; Haggblad, J.; Nilsson, S.; Gustafsson, J.-A. Endocrinology
1997, 138, 863.
23. Coe, P. L.; Scriven, E. C. J. Chem. Soc. Perkin Trans. 1
1986, 3, 475.
Acknowledgements
24. Youssef, E.-A.; Elisabeth, L.; Philippe, M.; Akraham, T.;
Jean, F. T.; Raymond, D.; Gilles, T. O. R. J. Med. Chem.
1997, 40, 952.
We are grateful for support of this research through a
grant from the National Institutes of Health. We thank

S.-H. Kim, J. A. Katzenellenbogen / Bioorg. Med. Chem. 8 (2000) 785±793
793
25. Youssef, E.-A.; Elisabeth, L.; Philippe, M.; Akraham, T.;
Gilles, T. O. R. Heterocycles 1997, 45, 723.
26. Pearce, P. J.; Richards, D. H.; Scilly, N. F. J. Chem. Soc.
Perkin Trans. 1 1972, 1655.
the Design of Estrogen Radiopharmaceuticals; Eckelman, W. C.
Ed.; CRC Press: Boca Raton, 1982; pp 93±126.
30. Still, W. C.; Kahn, M.; Mitra, A. J. Org. Chem. 1978, 43,
2923.
27. Carlson, K. E.; Choi, I.; Gee, A.; Katzenellenbogen, B. S.;
Katzenellenbogen, J. A. Biochemistry 1997, 36, 14897.
28. Katzenellenbogen, J. A.; Johonson, J. H. J.; Myers, H. N.
Biochemistry 1973, 12, 4085.
29. Katzenellenbogen, J. A.; Heiman, D. F.; Carlson, K. E.;
Lloyd, J. E. In Vivo and In Vitro Steroid Receptor Assays in
31. Minick, D. J.; Frenz, J. H.; Patrick, M. A.; Brent, D. A. J.
Med. Chem. 1988, 31, 1923.
32. Pomper, M. G.; VanBrocklin, H.; Thieme, A. M.; Thomas,
R. D.; Kiesewetter, D. O.; Carlson, K. E.; Mathias, C. J.;
Welch, M. J.; Katzenellenbogen, J. A. J. Med. Chem. 1990, 33,
3143.