J. Med. Chem. 2004, 47, 5021-5040
5021
Design and Synthesis of Aryl Diphenolic Azoles as Potent and Selective
Estrogen Receptor-β Ligands
Michael S. Malamas,*,† Eric S. Manas,‡ Robert E. McDevitt,† Iwan Gunawan,† Zhang B. Xu,§ Michael D. Collini,‡
Chris P. Miller,| Tam Dinh,⊥ Ruth A. Henderson,O James C. Keith Jr.,# and Heather A. HarrisO
Department of Chemical and Screening Sciences, Wyeth Research, CN 8000, Princeton, New Jersey 08543-8000, Department of
Chemical and Screening Sciences, Wyeth Research, Collegeville, Pennsylvania 19426, Department of Chemical and Screening
Sciences, Wyeth Research, 200 Cambridge Park Drive, Cambridge, Massachusetts 02140, Women’s Health Research Institiute,
Wyeth Research, Collegeville, Pennsylvania 19426, and Cardiovascular and Metabolic Diseases, Wyeth Research,
Cambridge Massachusetts 02140
Received April 13, 2004
New diphenolic azoles as highly selective estrogen receptor-β agonists are reported. The more
potent and selective analogues of these series have comparable binding affinities for ERβ as
the natural ligand 17β-estradiol but are >100-fold selective over ERR. Our design strategy not
only followed a traditional SAR approach but also was supported by X-ray structures of ERβ
cocrystallized with various ligands as well as molecular modeling studies. These strategies
enabled us to take advantage of a single conservative residue substitution in the ligand-binding
pocket, ERR Met421 f ERβ Ile373, to optimize ERβ selectivity. The 7-position-substituted
benzoxazoles (Table 5) were the most selective ligands of both azole series, with ERB-041 (117)
being >200-fold selective for ERβ. The majority of ERβ selective agonists tested that were at
least ∼50-fold selective displayed a consistent in vivo profile: they were inactive in several
models of classic estrogen action (uterotrophic, osteopenia, and vasomotor instability models)
and yet were active in the HLA-B27 transgenic rat model of inflammatory bowel disease. These
data suggest that ERβ-selective agonists are devoid of classic estrogenic effects and may offer
a novel therapy to treat certain inflammatory conditions.
Introduction
Estrogens play an essential role in the growth,
development, and homeostasis of a diverse range of
tissues.1 Estrogens exert their physiological role via
estrogen receptors (ER), which function as ligandactivated transcriptional regulators.2 A number of marketed products target estrogen receptors, such as oral
contraceptives (e.g., 17R-ethynyl estradiol), hormone
therapy agents (e.g., 17β-estradiol, conjugated equine
estrogens), and breast cancer therapeutics (e.g., tamoxifen, fulvestrant).
The first discovered ER, now called ERR, was cloned
in 19863 and was believed to mediate the effects of
estrogens solely. However, in 1996, Gustafsson and coworkers discovered a second estrogen receptor during
a search for novel nuclear receptors in a rat prostate
cDNA library and named it ERβ.4 The discovery of ERβ
has caused considerable excitement within the scientific
community and has provided the motivation to identify
* To whom correspondence should be addressed. Wyeth Research,
CN 8000, Princeton, NJ 08543. Tel: 732-274-4428. Fax: 732-274-4505.
E-mail: malamam@wyeth.com.
† Department of Chemical and Screening Sciences, Wyeth Research,
CN 8000, Princeton, New Jersey 08543-8000.
‡ Department of Chemical and Screening Sciences, Wyeth Research,
Collegeville Pennsylvania 19426.
§ Department of Chemical and Screening Sciences, Wyeth Research,
200 Cambridge Park Drive, Cambridge, Massachusetts 02140.
| Present address: GlaxoSmithKline, King of Prussia, Pennsylvania
19406.
⊥ Present address: 2430 Arlington Blvd. E6, Charlottesville, Virginia 22903.
O Women’s Health Research Institiute, Wyeth Research, Collegeville,
Pennsylvania 19426.
# Cardiovascular and Metabolic Diseases, Wyeth Research, Cambridge, Massachusetts 02140.
its physiological role in mediating estrogen action.
Because the two ER isoforms exhibit overlapping but
distinct tissue distribution patterns,5 it appeared likely
that an ERβ-selective ligand would exhibit a pharmacological profile that is different from that of nonselective estrogens such as 17β-estradiol. The fact that ERβ
is widely expressed but not the dominant estrogen
receptor in the uterus or breast tissues makes it a very
attractive drug target.
Both estrogen receptors have distinct domains that
are critical to transactivation, DNA binding, and hormone binding. ERR and ERβ have modest overall
sequence identity, differing primarily in their N-terminus domains, with the sequences more conserved at the
DNA (95% identity) and ligand-binding domains (LBD)
(58% identity).6 Despite the modest sequence identity,
the overall structural differences in the ligand-binding
pocket are rather small. The X-ray structure of the
human ERβ LBD complexed to genistein6 showed that
there are two subtle amino acid differences in close
proximity to the bound ligand: ERR Leu384 is replaced
by ERβ Met336, and ERR Met421 is replaced by ERβ
Ile373. Considering this small change in the ligandbinding cavity, it is not surprising that 17β-estradiol
displays a similar affinity for both receptors.
Despite a large body of mapping studies, in vitro
characterization studies, and the creation of knockout
mice, the physiological role of ERβ has remained unclear
until recently,7 when the availability of selective ligands
helped further the investigation of the physiological
function of ERβ. ERβ-selective agonist ERB-041 (compound 117, Table 5) was used to demonstrate that this
receptor may be a useful target for inflammation,7
10.1021/jm049719y CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/08/2004
5022
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21
Malamas et al.
Figure 1.
Table 1. Phenyl Benzisoxazoles
compd
R1
R2
R3
R4
3
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
1
2
OH
H
OH
OH
H
H
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
H
H
H
H
OH
OH
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
OH
H
OH
H
OH
H
H
H
OH
H
H
H
H
Cl
CN
H
H
Br
H
F
H
Me
Me
H
propyl
H
H
propyl
OH
Me
CH2CN
H
CHdNOH
H
OH
H
OH
H
OH
H
ethyl
H
propyl
H
17β-estradiol
genistein
R5
R6
OH
OH
H
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Br
Cl
Me
propyl
propyl
ERβ
IC50
(nM)a
ERR
IC50
(nM)a
fold
selectivity
for ERβ
3.5 ( 1.3
718 ( 428
254 ( 20
54 ( 26
138 ( 75
46 ( 21
10 ( 2
12 ( 8
12 ( 1
8(2
31 ( 1
37 ( 6
20
25 ( 14
33 ( 21
96
383
1.8 ( 0.1
2 ( 0.1
5 ( 0.1
3.6
3.1
3.6 ( 1.6
10 ( 4
24 ( 8
6100
3000
815 ( 207
4068 ( 1498
819 ( 383
273 ( 125
264 ( 53
183 ( 16
161 ( 66
431 ( 30
475 ( 200
50
210 ( 14
511 ( 123
717
2390
30 ( 16
33 ( 8
65 ( 8
13
22
3.2 ( 1.0
395 ( 181
8
8
12
15
29
18
27
23
16
19
14
13
2
8
18
7
6
17
15
14
4
7
1
41
a IC
50 values are the means of at least two experiments (SD (performed in triplicate, determined from eight concentrations). Values
without SD are for a single determination only.
whereas other studies utilizing the ERR-selective agonist propylpyrazole triol (PPT) showed that many classical estrogenic effects are mediated primarily by ERR.8
Several groups have investigated a variety of nonsteroidal scaffolds mimicking either the dihydroxyl arrangement of the nonselective estradiol or the moderately ERβ-selective phytoestrogen genistein as potential
selective ERβ ligands. Although these studies have
produced excellent potent ligands for both estrogen
receptors, the generation of highly ERβ-selective ligands
has proved to be quite challenging. Naturally occurring
phytoestrogens and several modified analogues were
reported to possess modest selectivity for the ERβ
receptor (10-40-fold),9 and one series of genistein
analogues was claimed to exhibit impressive binding
selectivity.10 Diarylpropylnitriles (DPN),11,12 biphenyl
compounds,9,13 and benzothiazoles/benzoxazoles14,15 exhibited as much as ∼70-fold selectivity for ERβ, whereas
other scaffolds, for example, tetrahydrochrysenes
(THC),16-18 aryl benzothiophenes,19 isoxazoles,20,21 benzimidazoles,22 triazines,23 benzoxazines,24 and tetrahydrofluorenones,25 have been reported to be on the order
of 10-40-fold ERβ selective.
In this paper, we report the design and synthesis of
potent and selective ERβ ligands initially based on
diphenolic benzisoxazole (3) (Figure 1). Compound 3 was
identified through a competitive radioligand binding
assay screen of our in-house sample collection. It had
excellent potency for ERβ with an IC50 value of 3.5 nM,
but it was minimally selective (8-fold) for ERβ (Table
1). The development of our ligands followed a traditional
SAR approach and was supported by X-ray structures
of ERβ cocrystallized with various ligands and molecular
modeling studies to expedite the discovery of selective
ligands. The finer details of these crystallographic and
modeling studies will be reported elsewhere.26 We have
primarily concentrated our efforts on the R face of the
ERβ LBD binding pocket, utilizing only one of the amino
acid differences (ERR Met421 f ERβ Ile373) observed
when comparing the binding pocket residues of the two
ER isoforms.
Chemistry
Compounds shown in Tables 1-5 were synthesized
according to general synthetic procedures (Schemes
1-5). The benzisoxazoles in Tables 1 and 2 were
Design and Synthesis of Aryl Diphenolic Azoles
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21 5023
Table 2. Naphthyl Benzisoxazoles
compd
R1
R2
R3
R4
ERβ
IC50
(nM)a
55
56
57
58
OH
OH
OH
H
H
H
H
OH
H
OH
Br
H
OH
H
OH
OH
1.4 ( 0.3
20 ( 12
87 ( 13
112
compd
R1
R2
R3
ERβ
IC50
(nM)a
59
60
61
H
OH
H
OH
H
H
H
H
OH
17
1.5
879 ( 499
ERR
IC50
(nM)a
fold
selectivity
for ERβ
8(4
24 ( 12
360 ( 5
463
6
1
4
4
ERR
IC50
(nM)a
fold
selectivity
for ERβ
49
6
1727 ( 45
3
4
2
a IC
50 values are the means of at least two experiments (SD
(performed in triplicate, determined from eight concentrations).
Values without SD are for a single determination only.
Table 3. Naphthyl Benzoxazoles
compd
R1
R2
R3
R4
R5
R6
ERβ
IC50
(nM)a
ERR
IC50
(nM)a
fold
selectivity
for ERβ
62
63
64
65
66
67
68
69
OH
OH
OH
OH
OH
OH
H
OH
H
H
H
H
H
F
OH
H
H
H
OH
H
H
H
H
H
OH
H
H
H
Br
H
H
OH
H
OH
H
H
OH
OH
OH
H
H
H
H
H
H
H
H
Me
5(2
3 ( 0.1
1530
1050
134
46
15 ( 7
6(5
117 ( 26
35 ( 5
1900
1880
373
434
153 ( 30
163 ( 25
23
12
1
2
3
10
10
26
compd
R1
R2
ERβ
IC50
(nM)a
ERR
IC50
(nM)a
fold
selectivity
for ERβ
70
71
H
OH
OH
H
521 ( 369
533
3800 ( 3649
780
7
2
a IC
50 values are the means of at least two experiments (SD
(performed in triplicate, determined from eight concentrations).
Values without SD are for a single determination only.
prepared via the general routes depicted in Schemes
1-3. In Scheme 1, two synthetic routes were used to
produce benzisoxazole 8b, using a common intermediate, 6. Arylbromide 4 was first treated with n-butyllithium and then with an appropriately substituted
benzaldehyde 5 to produce the addition product, which
was oxidized with either manganese dioxide or chromic
acid to afford benzophenone 6. In route a, acetone oxime
was treated with potassium tert-butoxide, and then
benzophenone 6 was added to the mixture to produce
oxime 7. The cyclization of 7 to benzisoxazole 8a was
accomplished under acidic conditions with 5% hydrochloric acid in acetonitrile. The demethylation of 8a with
either boron tribromide or a mixture of hydriodic acid,
acetic anhydride, and acetic acid afforded benzisoxazole
8b. In route b, benzisoxazole 8b was obtained via a more
direct approach, where benzophenone 6 was treated
with hydroxylamine and sodium hydride in N,N-dimethylformamide to produce 8a, which upon treatment
with boron tribromide afforded 8b. In Scheme 2, benzophenone 11a was prepared from benzoyl chloride 9
and 1,4-dimethoxybenzene 10 in the presence of aluminum chloride and 1,2-dichloroethane. The demethylation of 11a with pyridine hydrochloride at high
temperatures (200 °C) produced 11b. Benzophenone 11b
was converted to benzisoxazole 13 in two steps. First,
oxime formation (12) was accomplished with hydroxylamine in ethanol, and second, dehydration of 12 with
diethylazodicarboxylate and triphenyl phosphine furnished benzisoxazole 13. Dialkyl analogues 53 and 54
(Table 1) were prepared according to Scheme 3. The
treatment of 14a (prepared from the corresponding
phenol and chloromethyl methyl ether/NaH) first with
tert-butyllithium and second with an appropriately
substituted benzaldehyde 14b produced alcohol 15. The
dehydroxylation of 15 with triethylsilane in the presence
of trifluoroacetic acid furnished phenol 16a. The protection of phenol 16a with iodomethane in the presence of
sodium hydride afforded the corresponding anisole 16b,
which upon treatment with N-bromosuccinimide in
acetonitrile produced bromide 17. The conversion of 17
to benzisoxazole 19 was accomplished according to
Scheme 1. The benzoxazoles of Tables 3 and 4 were
prepared according to Scheme 4. Dimethoxyaniline 20
was treated with either benzoyl or naphthoyl chloride
21 in the presence of triethylamine to produce amide
22, which was converted to benzoxazole 23 upon treatment with pyridine hydrochloride at high temperatures
(200 °C). Bromo analogue 66 (Table 3) was prepared
from 63 upon treatment with bromine in acetic acid. The
7-position-substituted benzoxazoles (Table 5) were prepared according to Scheme 5. Nitrophenol 24 was first
brominated with Br2/NaOAc in acetic acid and then
reduced with H2/Ra-Ni in EtOAc to afford aniline 25b.
The coupling of 25b with appropriately substituted
benzoyl chloride 26 in the presence of pyridine produced
amide ester 27. The conversion of 27 to benzoxazole 28
was accomplished under acidic conditions (p-toluenesulfonic acid) at high temperature (150 °C). The demethylation of 28 with boron tribromide afforded the
diphenolic benzoxazole 29. The palladium-catalyzed
cross-coupling reaction27,28 of benzoxazole 29 with alkyl
stannates or aryl boronic acids produced benzoxazoles
30a and 30b. 2-Fluorovinyl analogue 31a was prepared
from 30a (R2 ) vinyl) by the initial formation of the 1,2bromofluoroethane adduct of the vinyl group with
hydrogen fluoride-pyridine and 1,3-dibromo-5,5-dimethyl hydantoin in sulfolane and subsequent hydrogen
bromide elimination with DBU.29 2-Bromovinyl analogue 31b was prepared in three steps from 28 upon
vinylation of the 7-position of the benzoxazole nucleus,
boron tribromide treatment (resulted in demethylation
of the methoxy groups and the bromination of the vinyl
group, affording1,2-dibromoethane), and hydrogen bro-
5024
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21
Malamas et al.
Table 4. 5- and 6-Hyxdroxy-2-Phenyl Benzoxazoles
compd
R1
R2
R3
R4
ERβ
IC50
(nM)a
ERR
IC50
(nM)a
fold
selectivity
for ERβ
72
73
74
75
76
77
78
79
80
H
H
H
H
H
H
Cl
H
H
OH
H
OH
H
H
H
H
H
H
H
H
OH
OH
F
Cl
H
CMe3
O-n-C4H9
OH
OH
H
OH
OH
OH
OH
OH
OH
3(1
50 ( 15
181 ( 97
105 ( 25
39 ( 10
703
157 ( 11
1600
3660
82 ( 180
902 ( 444
2353 ( 536
2410 ( 523
8430 ( 168
5000
2765 ( 7
5000
6240
26
18
13
20
22
7
18
3
2
ERR
IC50
(nM)a
fold
selectivity
for ERβ
compd
R1
R2
R3
R4
ERβ
IC50
(nM)a
81
82
83
84
85
86
87
88
89
H
H
H
H
H
Cl
Cl
Br
H
H
H
H
Cl
OH
H
H
H
OH
H
F
Cl
H
H
H
F
F
OH
OH
OH
OH
OH
OH
OH
OH
OH
H
49 ( 14
66 ( 37
239 ( 15
59 ( 62
25
16 ( 5
64 ( 11
42 ( 10
963 ( 194
1227 ( 533
1570 ( 537
5280 ( 1131
139 ( 42
190
464 ( 86
1813 ( 206
1210 ( 289
5110
25
24
22
2
8
30
29
29
5
6 ( 2.4
176 ( 76
29
90
a IC
50 values are the means of at least two experiments (SD (performed in triplicate, determined from eight concentrations). Values
without SD are for a single determination only.
mide elimination by DBU. Methoxy analogue 30c was
prepared by the displacement of the bromine of 28 with
NaOMe in the presence of CuBr. Ethynyl 32a, cyano
32b, and alkyl analogues 32c were prepared from 28
via a palladium-catalyzed cross-coupling reaction using
ethynyl(trimethyl)silane, Zn(CN)2, or alkyl zinc chlorides (Rieke reaction),30 respectively, and subsequent
demethylation with BBr3. Cyano analogue 32b was also
prepared by the displacement of the bromine of 29 with
CuCN. Metal halogen exchange of 28 with n-BuLi
followed by acetone addition and subsequent treatment
with pyridine hydrochloride at high temperature (200
°C) produced benzoxazole 32e. The reduction of 32e with
H2/Pd-C furnished isopropyl analogue 32f. 7-Lithiated
benzoxazole 28 was also treated with various electrophiles (i.e., EtI, PhMeNCHO, CN-CO2R4) to produce
analogue 32d. Compounds 100-102 (Table 5) were
successively prepared by standard synthetic protocols
from 103 upon reduction (NaBH4, MeOH), bromination
(BBr3, CH2Cl2), and nitrile formation (KCN, 18-C-6,
DMF). Bromo analogues 135 and 136 (Table 5) were
prepared from 117 upon bromination with bromine in
acetic acid, whereas bromo analogues 137 and 138 were
prepared from 91 upon treatment with N-bromosuccinimide in acetonitrile.
Results and Discussion
The primary screening assay for the program was a
competitive radioligand binding assay,31 which was used
to determine the relative binding affinity (IC50) of
compounds for the human ERR and ERβ LBD. Selected
compounds were also evaluated using mouse and rat
LBDs as well as full-length human receptors. These
data are presented in Tables 1-7. As expected in these
assays, radioinert 17β-estradiol bound equally well to
ERR and ERβ.
Benzisoxazole Analogues. High-throughput screening-hit benzisoxazole 3 (Figure 1) was successfully
cocrystallized with ERβ (Figure 2a). As shown in Figure
2a and discussed above, there are only two conservative
amino acid differences among the residues closest to the
ligand, ERβ Met336 f ERR Leu384 and ERβ Ile373 f ERR
Met421. Benzisoxazole 3 occupies the ligand-binding
cavity in an orientation where the hydroxyl group of the
benzisoxazole nucleus interacts with the receptor via a
hydrogen-bonding network involving the side chains of
Glu305 and Arg346 and a buried water molecule, whereas
the 4′-hydroxyl group of the resorcinol nucleus extends
to the distal end of the cavity making a hydrogen-bond
interaction with His475. Both of these hydroxyl groups
are important to the binding affinity of the compound
because the elimination of either hydroxyl group (examples 34, 35; Table 1) proved to be detrimental to the
compound’s potency. However, the benzisoxazole hydroxyl group appears to be more important to the ligand
binding than the 4′-hydroxyl group of the resorcinol,
given that 34 was 3× less potent than 35. This is
consistent with the placement of 3 in the electron
Design and Synthesis of Aryl Diphenolic Azoles
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21 5025
Table 5. 7-Substituted 2-Phenyl Benzoxazoles
compd
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117 (ERB-041)
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
R1
R2
R3
R4
R5
ERβ
IC50
(nM)a
OMe
Br
Br
Br
Br
Br
CN
CN
CN
CH2Br
CH2CN
CH2OH
CHO
CO2Me
CO2Et
CONH2
CO2H
ethyl
propyl
isopropyl
butyl
ethynyl
allyl
allyl
allyl
vinyl
vinyl
2-F-vinyl
2-Me-vinyl
2-Br-vinyl
2-Br-vinyl
3-Me-vinyl
vinyl
vinyl
vinyl
vinyl
vinyl
vinyl
phenyl
2-furyl
2-furyl
2-thienyl
2-thiazole
cyclopentane
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
F
H
H
H
H
H
H
H
H
H
F
CF3
H
H
H
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
H
H
F
F
H
H
H
H
H
F
H
H
F
H
H
H
F
H
H
H
H
H
H
H
F
CH3
H
H
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
F
H
H
H
H
H
F
F
F
F
F
F
H
CH3
H
H
H
H
H
H
59 ( 19
2(1
3(1
166
1.4 ( 0.4
1.4 (1.0
6(2
26 ( 11
2.4 ( 0
40 ( 29
1040 ( 1408
1340
59 ( 22
356
190 ( 88
95 ( 40
>5000
13 ( 6
11 ( 6
82
79
15 ( 3
13 ( 5
14 ( 10
3(1
3.5 ( 1.7
5.0 ( 4
3.2 ( 2.2
142 ( 92
45
16
23 ( 6
1.9 ( 0.6
3.7 ( 0.6
2.2 ( 1.1
66 ( 23
201 ( 125
27 ( 30
235
135
313
97
366
102
compd
R1
R2
R3
R4
R5
ERβ
IC50
(nM)a
135
136
137
138
vinyl
vinyl
OMe
OMe
H
Br
H
Br
Br
Br
Br
Br
F
F
F
F
H
H
H
H
25 ( 13
155
52 ( 30
64
ERR
IC50
(nM)a
fold
selectivity
for ERβ
2557 ( 1618
155 ( 47
260 ( 93
1870
47 ( 12
44 ( 12
411 ( 131
1435 ( 92
138 ( 5
2975 ( 2298
>5000
b
2638 ( 519
>5000
7827 ( 3427
9620
>5000
537 ( 81
390 ( 25
1200
498
481 ( 133
727 ( 356
1100 ( 544
98 ( 38
447 ( 226
1216 ( 688
376 ( 201
775 ( 208
462
196
539 ( 179
227 ( 107
474 ( 211
249 ( 76
4040
>10 000
1116 ( 1556
1300
809
1980
1030
1340
1010
43
68
81
11
32
32
72
56
57
74
>5
45
>14
41
101
40
37
15
6
33
55
78
39
129
226
116
6
10
12
23
123
127
115
62
>50
41
6
6
6
11
4
10
ERR
IC50
(nM)a
fold
selectivity
for ERβ
1036 ( 488
803
2668 ( 1172
559
41
5
52
9
a IC
50 values are the means of at least two experiments (SD (performed in triplicate, determined from eight concentrations). Values
without SD are for a single determination only. b not tested.
density shown in Figure 2b, with the benzisoxazole
hydroxyl mimicking the A-ring hydroxyl of 17β-estradiol32 as described above. The elimination of the 2′hydroxyl of the resorcinol also affected the potency of 3
but to a lesser extent (3 vs 36). This hydroxyl appears
to participate in an intramolecular hydrogen-bonding
interaction with the nitrogen of the benzisoxazole
nucleus, rendering planarity to the molecule. This
intramolecular hydrogen bond is thus likely to act in a
manner similar to that of genistein, improving the
5026
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21
Scheme 1a
Malamas et al.
Scheme 4a
a
a Reagents: (a) n-BuLi, THF; (b) MnO or CrO , H SO ; (c)
2
3
2
4
acetone oxime, tert-BuOK; (d) HCl, MeCN; (e) BBr3 or HI, AcOH,
Ac2O; (f) NaH, NH2OH‚HCl.
Scheme 2a
a Reagents: (a) AlCl , ClCH CH Cl; (b) pyridine‚HCl; (c)
3
2
2
NH2OH‚HCl, EtOH; (d) diethylazodicarboxylate, PPh3, THF.
Scheme 3a
a Reagents: (a) MOM-Cl, Et N; (b) t-BuLi, R CHO; (c) Et SiH,
3
2
3
TFA; (d) NaH, MeI; (e) N-bromosuccimide, CH2Cl2; (f) n-BuLi, 2-F,
4-OMe-benzaldehyde, THF; (g) MnO2, CHCl3; (h) acetone oxime,
tert-BuOK; (i) HCl, MeCN; (j) BBr3, CH2Cl2.
potency by increasing the effective lipophilicity. Regioisomeric 5-hydroxy analogues 37 and 38 were more
selective for ERβ but less potent.
The examination of the ERβ complex with 3 suggested
that the introduction of groups at positions 2′ and 3′of
the resorcinol nucleus can be directed toward the R face
of the ERβ LBD binding pocket, exploiting the ERR
Reagents: (a) Et3N, CH2Cl2; (b) pyridine-HCl, 200 °C.
Met421 f ERβ Ile373 residue substitution to achieve
greater ERβ selectivity. To confirm this hypothesis, we
prepared 3′-chloro analogue 39 and 2′-cyano analogue
40 (their fit to the pocket was confirmed by docking
calculations), and both were found to be more selective
than 3 (27- and 23-fold selective for ERβ, respectively).
These improvements in selectivity are the result of an
∼11-fold decrease in affinity for ERR versus an ∼3-fold
decrease in affinity for ERβ in both cases. In comparison, various other groups (entries 41-49) exhibited
similar or reduced ERβ potency and, in most cases,
smaller improvements in selectivity compared to 3. The
introduction of similar groups at the 5′-position of the
resorcinol nucleus in compounds 50-52 resulted in
compounds similar to 3 in potency but only about 2-fold
more selective. Disubstituted alkyl analogues 53 and
54 were similar to 3.
Because the A-C ring hydroxyl-hydroxyl distance of
3 is only 10.6 Å compared to the 12-Å distance for
genistein, we attempted to extend this distance by
replacing the phenyl with a naphthalene to gain more
optimal hydrogen-bonding interactions at both ends of
the cavity. In addition, because the majority of the ERβ
ligand-binding pocket is hydrophobic in nature, by
increasing the ligand size and lipophilicity it was quite
reasonable to expect an enhancement in ligand affinity.
In fact, bulkier naphthyl analogue 55 (Table 2) had an
IC50 value of 1.4 nM for ERβ. However, this compound
was only 8-fold selective for ERβ. 5′-Hydroxyl analogue
56 was 15-fold weaker than 55, whereas bromo analogue
57 and 5-hydroxyl benzisoxazole 58 were almost 60-fold
weaker than 55. Regioisomeric 1′-naphthalene benzoxazole 59 (Table 2) was 10× less potent than 58, whereas
5′-hydroxyl analogue 60 was similar to 55. 7′-Hydroxyl
analogue 61 was about 600-fold weaker than 60. Even
though both the phenyl and naphthyl benzisoxazoles
have produced ligands with excellent affinity for the
ERβ receptor, the selectivity against the ERR receptor
was only modest (10-30-fold).
Our intention was that extension of the phenyl to a
naphthalene would preserve the benzisoxazole as the
A-B ring, allowing us to use naphthalene substituents
to modulate selectivity. Although docking calculations
suggested that this was a reasonable outcome, the larger
size of 55 relative to that of genistein made it difficult
to rank the order of the potential binding modes. Thus,
to test the above hypothesis and to help validate our
docking calculations, 55 was cocrystallized with ERβ.
As can be seen in Figure 3, the naphthalene now acts
as the A-B ring, and the only substitutable position in
Design and Synthesis of Aryl Diphenolic Azoles
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21 5027
Scheme 5a
a Reagents: (a) Br , NaOAC, ACOH; (b) H , Ra-Ni, THF; (c) aroyl-chloride (26), pyridine, CH Cl ; (d) p-toluenesulfonic acid, p-xylene,
2
2
2
2
150 °C; (e) BBr3, CH2Cl2; (f) R2-stannyltributyl, [P(o-tolyl)3]2PdCl2, diethoxyethane or R2-B(OH)2, Pd(PPh3)4, Na2CO3, toluene; (g) MeONa,
CuBr, DMF; (h) pyridine‚HF, 1,3-dibromo-5,5-dimethyl hydantoin, sulfolane; (i) DBU, CH3CN; (j) ethynyl(trimethyl)silane, Pd(PPh3)4,
CuI, Et3N or Zn(CN)2, Pd(PPh3)4, DMF or R3-ZnCl, P(o-tolyl)3]2PdCl2, THF; (k) n-BuLi, acetone or CNCO2-R4 or EtI or PhMeNCHO; (l)
pyridine‚HCl; (m) H2, 10% Pd-C.
Table 6. Binding Affinity (IC50) of Selected Compounds for Rat and Mouse ERβ and ERR LBD
rat
a
mouse
compd
ERβ
IC50
(nM)a
ERR
IC50
(nM)a
fold
selectivity
for ERβ
ERβ
IC50
(nM)a
ERR
IC50
(nM)a
fold
selectivity
for ERβ
36
73
81
92
93
97
103
117 (ERB-041)
17β-estradiol
36 ( 12
20 ( 4.9
22 ( 6
1.2 ( 0.8
2 ( 0.8
4 ( 0.4
29 ( 7.5
3.14 ( 2.1
1.7 ( 0.5
716 ( 76
821 ( 150
1295 ( 189
132 ( 17
171 ( 17
362 ( 35
2252 ( 27
618 ( 72
1.9 ( 0.4
19
40
57
109
85
89
76
197
1
13 ( 2.1
9.5 ( 2
9.9 ( 2
1.7 ( 1
2.7 ( 1.7
4.3 ( 0.7
16.5 ( 5
3.7 ( 4.0
2.3 ( 0.7
1089 ( 603
1335 ( 550
1629 ( 521
132 ( 6.9
199 ( 71
423 ( 29
2601 ( 565
746 ( 303
2.2 ( 0.5
80
140
164
78
74
98
158
200
1
IC50 values are the means of at least two experiments (SD (performed in triplicate, determined from eight concentrations).
Table 7. Binding Affinity (IC50) of Selected Compounds for
Human Full-Length ERR and ERβ
compd
ERβ
IC50
(nM)a
ERR
IC50
(nM)a
fold
selectivity
for ERβ
91
92
93
97
98
99
103
106
113
117 (ERB-041)
118
124
137
17β-estradiol
52 ( 16
2 ( 0.43
4.4 ( 2.2
5.3 ( 0.7
18 ( 7
2.9 ( 0.9
60 ( 3
93 ( 45
15 ( 7.3
2.8 ( 1.5
6.3 ( 1.4
3.1 ( 1.4
82 ( 9
2.6 ( 0.8
4124 ( 423
187 ( 63
315 ( 54
671 ( 258
1970 ( 459
253 ( 52
3314 ( 892
>10 000
1455 ( 170
2572 ( 646
690 ( 182
412 ( 88
4480 ( 1131
3.5 ( 1.2
79
94
71
127
109
87
55
>107
99
920
110
133
55
1
a IC
50 values are the means of at least two experiments (SD
(performed in triplicate, determined from eight concentrations).
proximity to ERR Met421/ERβ Ile373 is the 7-position of
the benzisoxazole. Unfortunately, it appeared to us that
substituents introduced at this position to enhance ERβ
selectivity would also tend to have unfavorable interac-
tions with His475, thereby lowering ERβ affinity. Given
this observation, we decided to turn our attention to
more promising strategies. The observation that benzisoxazole can occupy either end of the cavity when
comparing 3 (Figure 2) with 55 (Figure 3) is similar to
what we report below for the benzoxazole series, and
thus we defer an explanation for this behavior to the
following section.
Benzoxazole Analogues. A logical progression in
the SAR study was to explore the regioisomeric benzoxazoles rather than the benzisoxazoles. In addition,
docking studies suggested that 2-phenyl- and 2-naphthyl-benzoxazoles would provide a greater opportunity
to access ERR Met421/ERβ Ile373. Regioisomeric 6-hydroxylnaphthyl-benzoxazole 62 (Table 3) of benzisoxazole 60 showed a 3-fold decrease of ERβ potency but
was found to be 6× more selective for ERβ. 6′-Hydroxyl
analogue 63 was slightly more potent in ERβ but with
weaker selectivity (12-fold) against ERR. 4′-Hydroxyl
analogues 64 were 500-fold weaker than 62. Halogen
analogues 60 and 67 were also weaker, as was 5′hydroxyl naphthyl-benzoxazole 68.
Docking studies with 62, later confirmed by cocrystallization with ERβ (Figure 4), suggested that substitu-
5028
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21
Malamas et al.
Figure 2. (a) Schematic representation of 3 cocrystallized with ERβ, showing key interactions as well as opportunities to improve
ERβ selectivity. (b) Unbiased 2fo-fc map contoured at σ, showing the electron density for 3 complexed with ERβ.
Figure 3. Unbiased 2fo-fc map contoured at σ, showing the
electron density for 55 complexed with ERβ.
Figure 4. Unbiased 2fo-fc map contoured at σ, showing the
electron density for 62 complexed with ERβ.
tion at the 7′-position of the naphthalene would provide
the best access to ERR Met421/ERβ Ile373 to enhance ERβ
selectivity. This binding mode shown in Figure 4 is in
contrast to the binding mode of naphthyl benzisoxazole
55, where the naphthalene acts as the A-B ring and
occupies the opposite end of the cavity. (See Figure 3
and Discussion below.) Even so, the naphthalene moiety
of 62 already fills the pocket near ERR Met421/ERβ Ile373
so well that there is minimal room left to explore the
remainder of the pocket with functional groups. Unfortunately, although the more lipophilic 7′-methyl analogue 69 retained the ERβ potency of 62, the ERβ
selectivity did not improve significantly. Regioisomeric
2′-naphthalene benzoxazole analogues 70 and 71 were
only weakly active (IC50 about 500 nM). Similar to the
naphthyl-benzisoxazoles, the naphthyl-benzoxazole ligand
affinities were also strongly dependent on the dihydroxyl substitution pattern. Small regiosomeric hydroxyl modifications resulted in wide range of potencies,
consistent with the critical hydroxyl-mediated anchoring
of dihydroxyl ligands to the ER’s binding cavity shown
in Figure 4.
Regioisomeric 6-hydroxyl benzoxazole 72 (Table 4), an
analogue of benzisoxazole 3, showed similar ERβ potency and increased ERβ selectivity (26-fold). The
elimination of the 2′-hydroxyl of the resorcinol nucleus
(73) resulted in the 15-fold decrease of ERβ potency,
similar to that of benzisoxazole 36. Several other
hydroxyl regioisomers (74 and 75) as well as halogensubstituted analogues (76-78) were also found to have
a weaker affinity for ERβ. Bulky substituents (79 and
80) next to the phenolic hydroxyl were detrimental to
the potency. As one might expect, the regioisomeric
5-hydroxyl benzoxazoles (81-89) exhibited a similar
SAR pattern to that of the 6-hydroxyl benzoxazoles.
The docking of benzoxazole 81 to the X-ray crystal
structure of benzofuran 9026 (Table 4) bound to ERβ
revealed a nearly perfect superimposition of these two
structures (Figure 5). Interestingly, the phenol of 81 is
predicted to act as the A ring, in contrast to the
orientation of 2-naphthylbenzoxazole 62 (Figure 4),
where the benzoxazole acts as the A ring. Similar
changes in the binding mode are observed when comparing 81 to benzisoxazole 3 (Figure 2) and, as pointed
out above, when comparing benzoxazole 62 to benzisoxazole 55 (Figure 3) and benzisoxazoles 3 and 55 with
each other. Clearly, this is a somewhat general phenomenon, which deserves a brief explanation. These
“flipping” effects appear to result primarily from interactions between the hydrophobic scaffold and the binding pocket as well as the relative geometric orientation
of the hydroxyl groups, both of which in turn affect the
way these hydroxyl groups are presented to key hydrogen-bonding residues Glu305, Arg346, and His475. This is
consistent with the fact that docking calculations, in
conjunction with a molecular mechanics evaluation of
the potential binding modes, are generally quite predic-
Design and Synthesis of Aryl Diphenolic Azoles
Figure 5. Compound 81 docked to the binding pocket of ERβ
complexed with 90.26 Only key residues as well as both ligands
are shown. Compound 90 is colored white. All other atoms are
colored by atom type.
tive of the X-ray binding modes of these ERβ ligands.
The docking calculations on ER tend to be less predictive
mainly with larger ligands such as 55, where the
flexibility of the protein is more likely to play an
important role in determining the true binding mode.
The examination of docked benzoxazole 81 and the
X-ray binding mode of 90 revealed that substitutions
at the 7-position offered an opportunity to improve the
ERβ selectivity of the phenyl-benzoxazole scaffold by
targeting the ERR Met421 f ERβ Ile373 residue substitution. In addition, there appeared to be more unoccupied
space in the pocket compared to that of the naphthyl
benzoxazoles. Therefore, we felt that the 7-position of
the phenyl-benzoxazole scaffold would be ideal to
explore with diverse functional groups to enhance the
ERβ potency and selectivity.
The docking calculations were used to assist in the
selection of functional groups. In addition, we hypothesized that electron-rich and/or chemically hard groups
(nitrogen (e.g., nitrile), oxygen (e.g., carbonyl), halogens
lighter than iodine, and alkenes) would have a greater
likelihood of differentiating between ERR Met421 and
ERβ Ile373 given the electronegative and polarizable
nature of the methionine sulfur atom. A detailed
justification of this hypothesis, as well as the optimization of the related phenyl-benzofuran scaffold 7-position, is beyond the scope of this paper and will be
elaborated elsewhere.26
The introduction of a methoxy group (91, Table 5) at
the 7-position of the benzoxazole nucleus resulted in a
small improvement of the selectivity (43-fold) without
affecting the ERβ potency. However, the introduction
of a bromo substituent (92) resulted in a marked
increase in ERβ potency (IC50 ) 2 nM) and a significant
improvement in ERβ selectivity (68-fold). The small
fluorine group ortho to the A-ring hydroxyl (entry 93)
did not alter the potency or selectivity of the 7-bromo
analogue, whereas the bulkier trifluoromethoxy group
(entry 94) noticeably decreased ligand binding. This
finding is not surprising because this hydroxyl participates in key interactions with the Glu305 and Arg346
residues, which may have been adversely affected by
the bulkier trifluoromethoxy group. Substitution at the
meta position of phenol (F, CH3) resulted in a small
increase in potency and about a 2-fold loss of selectivity
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21 5029
(entries 95 and 96). 7-Cyano analogues 97 and 99
exhibited similar potency and selectivity to that of
7-bromo analogues 92 and 95, whereas ortho-substituted fluoro analogue 98 maintained its selectivity but lost
about 8-fold in ERβ potency (98 vs 93). Methyl bromide
100 showed similar selectivity to that of 92 but had a
20-fold reduction in potency, whereas acetonitrile 101
was weakly active in ERβ. The carbonyl class of substituents (103-106) proved to be much weaker for both
ERs but displayed moderate to good levels of ERβ
selectivity. Amide analogue 106 was about 100-fold
selective for ERβ. The ethyl and propyl substituents
(108 and 109) exhibited small decreases in potency and
selectivity relative to 92, whereas the bulkier isopropyl
and butyl groups (110, 111) had noticeably weaker
binding affinity for ERβ. The ethynyl and allyl groups
(112, 113) were similar to the ethyl and propyl groups
with respect to potency and selectivity. However, the
incorporation of a vinyl group (116) increased the
selectivity to >100-fold in favor of ERβ. Introducing
fluorine ortho to the hydroxyl group produced 117 (ERB041), which showed a somewhat greater selectivity (226fold). 2-Fluorovinyl analogue 118 was equipotent to 117
but 2-fold less selective, whereas the bulkier 2-bromovinyl and 2-methylvinyl analogues (119-122) were substantially less potent and selective for ERβ. It is likely
that unfavorable steric interactions between these
bulkier groups and Ile373 are responsible for the decreased ligand affinity.
Some interesting findings were also observed with
various fluorine analogues of 117. Fluoro and difluro
analogues 123-125 were about 2-3× more potent and
2× less selective than 117, whereas difluoro analogues
126 and 127 were considerably less potent than 117.
Considering that fluoro substituents do not significantly
alter the size of the molecule, electrostatic repulsion
involving one of the fluorine groups is the likely reason
for the loss of potency. Supporting evidence for this
hypothesis is the fact that 2,6-difluoro analogue 127,
most likely to experience repulsion with the carboxylic
acid of Glu305, was the least potent analogue among
them. For 2,5-difluoro analogue 126, it is likely that the
5-fluoro analogue experiences somewhat weaker electrostatic repulsion with the carbonyl of Leu346.
A methyl group meta to the A-ring hydroxyl group
(128) caused a 9-fold decrease in ERβ potency and 2-fold
reduction in selectivity. Various aromatic and carbocyclic groups (129-134) were found to be 50-100-fold less
potent and selective than 116, most likely because of
unfavorable steric interactions with the pocket. The
introduction of a bromine group at the 4-postion of the
benzoxazole nucleus of compounds 117 (7-vinyl) and 91
(7-methoxy) produced monobromo analogues 135 and
137, respectively, causing a reduction in potency. Dibromovinyl analogue 136 exhibited an additional loss
of ERβ potency, although analogous methoxy analogue
138 maintained its ERβ potency. Both dibromo analogues were 5-10× less ERβ selective than the monobromo parent compounds.
To confirm that functional groups at the 7-position
were targeting the ERR Met421/ERβ Ile373 pocket, we
cocrystallized compound 117 (ERB-041) with human
ERβ (Figure 6).26 The binding mode of compound 117
is similar to what we predicted for parent compound
5030
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21
Malamas et al.
Table 9. Effect of Compounds on Rat Uterine Weight
uterine weight
(mean mg ( SEM)
vehicle
17R-ethynyl-17β-estradiol (EE; 0.06 µg/rat)
92 (2 mg/rat)
92 + EE
93 (2 mg/rat)
93 + EE
30.5 ( 3.2
104.7 ( 5.4a
39.2 ( 0.7
95.9 ( 5.5a
38.3 ( 1.7
93.9 ( 5.9a
vehicle
17R-ethynyl-17β-estradiol (EE; 0.06 µg/rat)
97 (2 mg/rat)
97 + EE
117 (ERB-041) (2 mg/rat)
117 (ERB-041) + EE
21.4 ( 1.6
85.5 ( 3.1a
30.3 ( 1.5
76.6 ( 3.0a
14.2 ( 1.1
80.7 ( 5.3a
a
Figure 6. ERβ complexed with 117 (ERB-041) (colored by
atom type). A Connolly surface is used to represent the shape
of the binding site. Dihedral angles determining the bound
conformation are indicated. As intended, the vinyl group sits
in a groove consisting of Ile373, Phe377, and Ile376, confirming
that we have indeed targeted the ERR Met421 f ERβ Ile373
residue substitution
Table 10. Effect of Compounds on Mouse Uterine Weight
uterine weight
(mean mg ( SEM)
Table 8. Activity of Compounds in a Cell-Based
Transcriptional Assay
regulation of IGFBP-4 MRNA in SAOS-2 cells
compd (1uM)
91
93
95
98
99
105
108
109
112
114
116
117 (ERB-041)
118
122
123
124
125
% activity relative to
10 nM 17β-estradiol
130
100
117
122
100
83
86
100
117
158
117
120
120
131
120
117
81
a
81, where the phenolic hydroxyl of 117 interacts with
the Glu305-Arg346-water triad through a hydrogenbonding network, whereas the hydroxyl group of the
benzoxazole nucleus extends to the distal end of the
cavity making a hydrogen-bond interaction with His475.
The 2-phenol and 7-vinyl groups exhibit dihedral angles
of 23 and 38°, respectively, relative to the benzoxazole
plane. The 7-vinyl group extends into the ERR Met421/
ERβ Ile373 pocket as intended and sits in a groove
formed by Ile373, Ile376, and Phe377. The vinyl CH acts
as a “hinge” that directs the ethylene moiety into this
relatively narrow groove and forces it to be in close
proximity to ERR Met421/ERβ Ile373. We hypothesize that
the substitution of ERβ Ile373 with ERR Met421 within
this groove would lead to a combination of electrostatic
and steric repulsion associated with the methionine side
chain, leading to enhanced ERβ selectivity.26 The crystallography studies also confirmed that helix 12 of ERβ
maintains an agonist-like conformation when 117 is
bound to the receptor, allowing for the binding of a
nuclear receptor box coactivator peptide, consistent with
the fact that 117 behaves as a full agonist on ERβ and
ERR (see below).
Significantly >vehicle, p < 0.05
vehicle
17β-estradiol
92 (50 mg/kg)
81 (50 mg/kg)
13.7 ( 0.8
40.5 ( 5.8a
13.1 ( 0.8
13.7 ( 0.8
vehicle
17β-estradiol
114 (50 mg/kg)
9.6 ( 0.5
40 ( 2a
10.3 ( 0.7
vehicle
17β-estradiol
117 (ERB-041) (50 mg/kg)
117 (ERB-041) (100 mg/kg)
11.7 ( 0.5
41.9 ( 2.9a
10.7 ( 0.9
10.6 ( 0.3
vehicle
17β-estradiol
124 (50 mg/kg)
106 (50 mg/kg)
118 (50 mg/kg)
9.8 ( 1.2
42.9 ( 4.9a
9.0 ( 0.3
9.5 ( 0.6
9.8 ( 0.7
vehicle
17β-estradiol
123 (50 mg/kg)
10.3 ( 0.8
45.3 ( 1.9a
10.3 ( 0.4
vehicle
17β-estradiol
113 (50 mg/kg)
125 (50 mg/kg)
9.5 ( 0.3
46.7 ( 2.5a
10.0 ( 0.6
10.0 ( 0.9
Significantly >vehicle, p < 0.05.
Selected compounds were evaluated in rat and mouse
ERR/β LBD binding assays as well as human full-length
ERR/β binding assays. In both rat and mouse LBD
binding assays, the majority of the tested compounds
(Table 6) exhibited similar potency and selectivity
relative to the human LBD assays, with the exception
of compounds 73, 81, and 103 that were about 4-7×
more potent in the mouse LBD assay. In the human fulllength binding assays, all tested compounds (Table 7)
were similar to the LBD assays with respect to potency
and selectivity, with the exception of compounds 97, 98,
113, and 117, which demonstrated somewhat higher
selectivity relative to the human LBD assays for ERβ.
Biological Evaluation. Cell-Based Transcriptional Activity. Two assays were used during the
program to determine whether compounds were ERβ
agonists. Both assays used the human osteosarcoma cell
line, SAOS-2, and these cells were engineered to overexpress ERβ via adenovirus infection. One assay measured increases in metallothionein-II mRNA,33 and the
other measured increases in insulinlike growth factor
binding protein-4 (IGFBP-4) mRNA.7 All compounds
Design and Synthesis of Aryl Diphenolic Azoles
Journal of Medicinal Chemistry, 2004, Vol. 47, No. 21 5031
Table 11. Evaluation of Bone Mineral Density in the Ovariectomized Rat
total bone mineral density
(mean mg/cm3 ( SEM)
trabecular bone mineral density
(mean mg/cm3 ( SEM)
vehicle
17β-estradiol (2 µg/rat)
93 (10 mg/kg)
92 (10 mg/kg)
92 (10 mg/kg) + 17β-estradiol
(2 µg/rat)
543.49 ( 14.24
639.49 ( 14.47a
501.40 ( 11.97
525.51 ( 7.93
682.41 ( 24.01a
353.96 ( 13.46
453.28 ( 24.93a
312.34 ( 19.73
287.56 ( 17.56
491.43 ( 36.43a
sham operated
685.28 ( 15.68a
510.96 ( 16.99a
compd
a
Significantly >vehicle, p < 0.05.
tested were essentially full agonists. (Representative
examples from the IGFBP-4 assay are shown in Table
8.)
In Vivo Evaluation. We had two goals during the
in vivo evaluation of our compounds. The first was to
assess selectivity and/or classic estrogenic activity, and
the second was to evaluate efficacy in a model of
inflammation.
Rodent Uterotrophic Assays. The sexually immature rodent uterus is a classic estrogen target tissue
and is used as a sensitive estrogenic bioassay. Nonselective estrogens, such as 17β-estradiol and 17R-ethynyl17β-estradiol, increase organ weight in both rats and
mice approximately 4-fold, and an ERR-selective ligand
(propylpyrazole triol (PPT)) is as efficacious as these
reference estrogens.8 These data suggest that ERR
activation is sufficient to elicit a full estrogenic response
in the uterus (as measured by organ weight increase).
For the rat assay, ERβ-selective compounds were administered for 3 days at a dose of 2 mg/rat/day, which
is equivalent to 36-53 mg/kg when the typical growth
of the animals is taken into account. For the mouse
assay, ERβ-selective compounds were dosed for 4 days
at 50 mg/kg (based on the initial weight of the mice).
As shown in Tables 9 and 10, all ERβ-selective compounds tested were nonuterotrophic. Compounds 92, 93,
97, 117 (ERB-041), and 124 were not antagonistic when
tested in combination with 17R-ethynyl-17β-estradiol.
Taken together, these data show that these compounds
are functionally selective for ERβ in vivo and do not
impact ERR activity. The in vivo selectivity of these
compounds is striking in that even at very high doses
no activation of ERR is seen. This finding, likely, cannot
be explained by binding selectivity alone. It is well
recognized that the receptor-ligand interaction is but
the first step in receptor activation, and it is possible
that although the compounds interact weakly with ERR
they do not elicit the conformational changes required
for dimerization and coactivator recruitment.
Rat Model of Osteopenia. After ovariectomy, rats
lose bone mineral density (mass), which can be prevented by the administration of nonselective estrogens
(e.g., 17β-estradiol), selective estrogen-receptor modulators (e.g., TSE-424),34 or an ERR-selective ligand (PPT).8
However, 92 and 93 had no effect on either total or
trabecular bone mineral density (Table 11). Moreover,
when 92 was combined with 17β-estradiol, no antagonistic effect was seen. Previously, we showed that 117
(ERB-041) was also inactive in this assay.7 These data
are consistent with the results from the uterotrophic
assay above, suggesting that the estrogenic response in
this model is mediated via ERR and that these ERβselective compounds do not impact ERR activity in vivo.
Table 12. Effect of Compounds on Rat Vasomotor Instability
tail skin temperature
change 15 min after
naloxone injection
(mean ( SEM)
vehicle
17R-ethynyl-17β-estradiol (0.3 mg/kg)
propylpyrazole triol (PPT; 15 mg/kg)
92 (15 mg/kg)
92+ PPT
97(15 mg/kg)
97 + PPT
4.6 ( 0.8
2.1 ( 1.1a
2.0 ( 0.8a
5.3 ( 0.7
1.9 ( 0.8a
5.2 ( 0.7
2.7 ( 1.1b
a Significantly
Purchase answer to see full
attachment