Androstene-3,5-dienes as ER-β selective SERMs

Article abstract of DOI:10.1016/j.bmcl.2007.09.001

A series of androstene-3,5-diene derivatives were prepared. Despite lacking the C-3 hydroxyl previously believed necessary for ER activity, some of the analogs retained surprising affinity for ER-β. For example, diene 4 retained excellent selectivity and potency as an ER-β agonist and was more selective for ER-β over the androgen receptor (AR).

Full text of DOI:10.1016/j.bmcl.2007.09.001

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 17 (2007) 6295–6298
Androstene-3,5-dienes as ER-b selective SERMs
Timothy A. Blizzard,a,* Candido Gude,a Jerry D. Morgan, II,a Wanda Chan,a
Elizabeth T. Birzin,a Marina Mojena,b Consuelo Tudela,b Fang Chen,c
Kristin Knecht,c Qin Su,c Bryan Kraker,a Ralph T. Mosley,a
Mark A. Holmes,a Susan P. Rohrera and Milton L. Hammonda
aMerck Research Laboratories, PO Box 2000, Rahway, NJ 07065, USA
bMerck Research Laboratories-CIBE, Madrid, Spain
cMerck Research Laboratories, West Point, PA 19486, USA
Received 12 July 2007; revised 28 August 2007; accepted 4 September 2007
Available online 7 September 2007
Abstract—A series of androstene-3,5-diene derivatives were prepared. Despite lacking the C-3 hydroxyl previously believed neces-
sary for ER activity, some of the analogs retained surprising affinity for ER-b. For example, diene 4 retained excellent selectivity and
potency as an ER-b agonist and was more selective for ER-b over the androgen receptor (AR).
ꢀ 2007 Elsevier Ltd. All rights reserved.
The importance of the selective estrogen receptor mod-
ulators (SERMs) has prompted extensive research.1
Much effort has focused on the discovery of ER-a2
and ER-b3 subtype-selective SERMs. We have also re-
ported non-selective spiroindene SERMs.4
been reported for other steroids.5 The structure of 4
was confirmed by NMR analysis.6 Since 4 lacked the
A-ring hydroxyl group thought to be essential for ER
binding, we expected it to be at best a weak ligand for
the ERs. However, to our surprise, 4 retained excellent
selectivity and potency as an ER-b agonist (Table 1)
with reduced affinity for the androgen receptor AR
(Table 2). We prepared the original synthetic targets 2
and 3 by first deconjugating 17 by treatment with base
followed by addition of methyl Grignard to the resulting
3-keto-d-5 intermediate. Subsequent deprotection affor-
ded 2 and 3 which proved to be much weaker ER
ligands than 4 (Scheme 1).
OH
OH
20
1
17
19
10
21
3
5
HO
1
4
Molecular modeling suggests a possible explanation for
the surprising ER-b activity of diene 4 (Fig. 1).7 The ter-
minal carbon of the vinyl group at C-10 of both 1 and 4
appears to have a negative steric interaction with the
Leu384 side chain of ER-a which is consistent with the
observed decrease in ER-a activity for both series.
During the course of our medicinal chemistry program
based on the highly ER-b-selective screening hit 1,3a
we have discovered and report herein a series of andro-
stene-3,5-diene SERMs (e.g., 4) with selectivity for ER-b
ranging from 6· to 160· in a ligand binding assay.
The vinyl group also forces 1 away from Glu305 (a
ꢀ
0
The 3,5-diene analog 4 was discovered serendipitously
during our attempted synthesis of 3-methylated analogs
2 and 3. Addition of methyl Grignard to ketone 173b fol-
lowed by ammonium chloride workup and deprotection
afforded the 3,5-diene analog 4. Similar chemistry has
0.4 A shift from what is observed between estradiol
and Glu353 in 1ERE7c) and toward Phe356 and other
hydrophobic residues which line the cavity along the
a-face of the steroid. Presumably, these additional
hydrophobic interactions compensate for the weakened
Hbond interaction. When modeled into the cavity, 4 is
similarly shifted toward Phe356 and more deeply toward
His475 than is seen with 1 due to the steric bulk of the
3-methyl group along with the C10 vinyl group. It
Keywords: SERMs; Estrogen; Androstenediol; Androstenediene.
*
Corresponding author. Tel.: +1 732 594 6212; fax: +1 732 594
9556; e-mail: tim_blizzard@merck.com
0960-894X/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.09.001

6296
T. A. Blizzard et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6295–6298
Table 1. ER binding and transactivation data
OH
OH
OH
R17
OH
OH
R17
17
17
17
17
17
R10
R10
R10
R10
R10
1
1
1
1
1
3
3
5
5
3
5
3
5
3
5
HO
R3
R3
HO
R3
O
1-2, 10-12
3
4-9, 15
13-14
16
#
R3
R10
R17
ER binding (IC50, nM)8
ER transactivation (EC50, nM)9
hER-a
hER-b
a/b
ER-a (% Ag)
ER-b (% Ag)
a/b
1
H
CH@CH2
CH@CH2
CH@CH2
CH@CH2
CH@CH2
CH@CH2
CH@CH2
CH@CH2
CH@CH2
CH@CH2
CH@CH2
CH3
H
H
—
H
H
H
H
2240
5300
>10000
1440
3110
>10000
5740
4380
1850
1050
150
11
204
21
980 (35)
4.1 (90)
240 (57)
>1000 (31)
69 (92)
240
>4
2
CH3
CH3
CH3
255
2490
9
>1000 (28)
> 1000 (27)
>1000 (4)
> 1000 (3)
>1000 (0)
820 (38)
3
>4
160
27
1
4
>14
>1.8
1
5
CH2CH3
C6H5
Cl
115
1760
64
565 (54)
>1000 (0)
170 (84)
480 (52)
31 (80)
6
>5.6
90
83
7
4.8
>2
8
CH3
CH3
H
CH3
CH ꢀ CH
53
72
>1000 (4)
490 (15)
9
26
42
15
10
11
12
13
14
15
16
CH3
CH ꢀ CH
25
14
>1000 (13)
130 (82)
26 (91)
190 (65)
24 (78)
>5.2
5.4
4.6
1.9
> 4.8
2.4
>2.6
0.36
H
11
18
H
—
H
180
9.9
120
34
5.6 (77)
230 (59)
210 (67)
220 (59)
380 (78)
2.1 (100)
CH@CH2
CH3
CH3
—
—
H
>10000
640
110
>83
19
430 (49)
>1000 (29)
540 (49)
—
CH3
—
9.4
780
1.2
12
CH@CH2
—
—
—
>10000
1.4
>13
1.2
>1000 (12)
0.75 (100)
Estradiol
—
Table 2. Comparison of hER-b and AR binding data
OTBS
OH
#
hER-b IC50 (nM)8 AR IC50 (nM)10 AR/ER-b
1
11
33
3
i, ii
2
255
2490
9
842
2610
560
500
3160
520
1615
170
290
230
330
40
3.3
1
O
3
17
4 (56% overall)
4
62
5
115
1760
53
4.3
1.8
9.8
22
6
OH
OH
8
9
10
72
25
iii, i, ii
+
6.8
21
HO
11
12
14
HO
9.9
120
34
23
2 (14% overall)
3 (12% overall)
13
14
2.8
1.2
240
0.01
22
Scheme 1. Reagents: (i) MeMgBr, NH4Cl workup; (ii) n-Bu4NBr,
THF; (iii) KOtBu.
15
16
9.4
780
1.2
2250
7.9
Estradiol
26
2.7
17 with ethyl Grignard afforded initially the exo-ethyl-
ene analog 18 which was isomerized to the desired 3-
ethyl derivative 5 (Scheme 2).
Testosterone >10,000
<0.0002
Similarly, treatment of 17 with phenyl Grignard gave
the desired 3-phenyl analog 6 (Scheme 3).
may be that the proximity of the planar diene of 4 to
Phe356 and increased ability to Hbond with His475 is
responsible for maintaining its potency despite the lack
of the A-ring hydroxyl group.
The 3-chloro analog 7 was prepared by reaction of ke-
tone 163a with acetic anhydride (to protect C-17 OH)
followed by phosphorus oxychloride (introduce C-3
Cl) and deprotection (Scheme 4).
The surprising activity of 4 prompted us to prepare
additional androstenediene analogs. Reaction of ketone

T. A. Blizzard et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6295–6298
6297
OH
OH
O
OH
R17
i - iii
i
ii or iii
4
O
Cl
16
7
19
8 R17 = CH3
9 R17 = CCH
Scheme 5. Reagents: (i) TPAP, NMO, 53%; (ii) MeMgBr, 59%; (iii)
LiC ꢀ CTMS then MeOH, NaOH, 49%.
The exo-methylene analog 13 was prepared from ketone
163a by Wittig olefination5a (Scheme 6).
For comparison with the 10-vinyl dienes 4 and 13, the
corresponding 10-methyl analogs5a 15 and 14 were pre-
pared from testosterone 20 (Scheme 7). Wittig olefin-
ation of 20 afforded the exo-methylene analog 14
which was readily isomerized to the 3,5-diene 15.
Figure 1. Superposition of crystallographically determined 1 (orange)
with 4 (cyan) in the context of hER-a (green) and hER-b (purple)
complexed with compound 1. Unless otherwise indicated, residue
numbering is that of hER-b.
The novel steroids were evaluated in estrogen receptor
(ER-a and ER-b)8 binding assay and in a cell-based
ER-b transactivation assay9 to measure estrogen agon-
ism in HEK293 cells (Table 1). New compounds were
also evaluated in an androgen receptor (AR)10 ligand
binding assay (Table 2). Diene 4 was comparable to
the lead compound 1 as an ER-b ligand (IC50 = 9 nM
for 4 vs 11 nM for 1), despite the lack of a C-3 hydroxyl
group, and retained excellent selectivity for ER-b over
ER-a although 4 (160·) is a bit less selective than 1
(204·). The excellent binding affinity of 4 was also re-
flected in the ER-b transactivation assay wherein 4
had a higher EC50 (69 nM for 4 vs 4 nM for 1) but com-
parable maximum agonism (92% of estradiol for 4 vs
90% for 1). Interestingly, 4 was much more selective
for ER-b over AR (62· for 4 vs 3· for 1). Similar
SAR was observed in the 10-methyl series (compare 12
and 15); diene 15 is comparable to alcohol 12 as an
OH
OH
i, ii
iii
17
18
5
Scheme 2. Reagents: (i) EtMgBr, NH4Cl workup; (ii) n-Bu4NBr, THF,
94% from 17; (iii) HCl, EtOH, 61%.
OH
i, ii
17
6
OH
OH
Scheme 3. Reagents: (i) PhMgBr, NH4Cl workup; (ii) n-Bu4NF, THF,
50% overall.
i
O
16
13
t
Scheme 6. Reagents: (i) Ph3PCH3Br, BuLi, 56%.
The 17-methyl (8) and 17-ethynyl (9) analogs of 4 were
readily prepared by oxidation of the 17-hydroxyl to
afford ketone 19, followed by reaction with either
methyl Grignard or lithium TMS-acetylide (Scheme 5).
We have previously reported the corresponding 3-hy-
droxy analogs (10 and 11).3a
OH
OH
i
O
OH
OH
20
14
OH
i - iii
ii
O
Cl
16
7
15
Scheme 4. Reagents and condition: (i) Ac2O, pyridine, DMAP, 100%;
(ii) POCl3, AcOH, rt, 50%; (iii) KOH, MeOH, 40%.
Scheme 7. Reagents: (i) Ph3PCH3Br, tBuLi, 26%; (ii) HCl, EtOH, 99%.

6298
T. A. Blizzard et al. / Bioorg. Med. Chem. Lett. 17 (2007) 6295–6298
4. Blizzard, T. A.; Morgan, J. D.; Mosley, R. T.; Birzin, E.
T.; Frisch, K.; Rohrer, S. P.; Hammond, M. L. Bioorg.
Med. Chem. Lett. 2003, 13, 479.
5. (a) Sondheimer, F.; Mechoulam, R. J. Am. Chem. Soc.
1957, 79, 5029; (b) Weintraub, P. L.; Tiernan, H. D.;
Benson, H. D.; Grunwell, J. F.; Johnston, J. O.; Petrow,
V. J. Med. Chem. 1976, 19, 1395.
ER ligand (both are less selective for ER-b than the cor-
responding 10-vinyl analogs 4 and 1) but is more selec-
tive for ER-b over AR. Diene 4 and lead compound 1
were both considerably more active and selective than
alcohols 2 and 3. In the diene series (compounds 4–7),
increasing the size of the C-3 substituent is clearly detri-
mental to ER-b binding and selectivity; with the largest
compound, the 3-phenyl analog 6, being inactive in the
ER transactivation assay. In the diene series, addition
of a substituent at C-17 (compounds 8 and 9) results
in a slight decrease in ER-b binding and selectivity rela-
tive to the unsubstituted analog 4. A similar result was
observed in the 3-hydroxy series (compare compounds
10 and 11–1). The internal diene 4 was clearly a better
ER-b ligand than the external diene 13 and was also
more selective for ER-b over AR. A similar trend was
observed in the 10-methyl series (compare compounds
15 and 14). The external dienes 13 and 14 are better
AR ligands than the corresponding internal dienes 4
and 15 but are weaker ligands than the corresponding
ketones 163a and testosterone. Diene 15 had the best
ER-b/AR ratio of all the tested compounds due to its
poor AR binding. However, 15 was not very potent in
the ER transactivation assay.
6. All new compounds were characterized by LC–MS and 500 or
600 MHz H NMR. HMBC was useful in confirming the
1
carbon skeleton of 4. Key HMBC correlations observed for 4
were H-21 ! C-2, C-3, and C-4; H-4 ! C-2 and C-6; and H-
6 ! C-4, C-10, and C-7. In addition, NOE correlations were
1
observed for H-4 ! H-21 and H-6 ! H-7 of 4. Selected H
NMR data (600 MHz, CDCl3, d) for 4: 5.75 (dd, J = 17,
10 Hz, 1H, H-19), 5.73 (br s, 1H, H-4), 5.49 (t, J = 3 Hz, 1H,
H-6), 5.18 (dd, J = 10, 2 Hz, 1H, H-20), 4.84 (dd, J = 17, 2 Hz,
1H, H-20), 3.63 (t, J = 9 Hz, 1H, H-17), 1.67 (s, 3H, H-21),
0.69 (s, 3H, H-18). Selected 13C NMR data for 4: (125 MHz,
CDCl3, d): 140.1 (C-19), 137.7 (C-5), 133.6 (C-3), 124.5 (C-4),
122.6 (C-6), 117.4 (C-20), 81.8 (C-17), 11.0 (C-18).
0
7. (a) Models were built using the 1.8 A resolution crystal-
ꢀ
lographic coordinates of compound 1 as cocrystallized
with hER-b (Fitzgerald et al., in preparation). Energy
minimization for all of the models within context of the
hER-b receptor (1 cocrystallized) was accomplished by
rigidly fix0ing all residues except for side chains which fell
ꢀ
within 5 A of the modeled ligand which were allowed to
minimize in conjunction with the ligand. All minimiza-
tions were conducted using the MMFFs force field7b with
a distance dependent dielectric model of 2r. Multiple
binding orientations of 2 were considered, including those
that had the D-ring flipped up to interact with Glu305.
However, the most energetically favored binding orienta-
tion was found to be the same as estradiol (Fig. 1); (b)
Halgren, T. A. J. Comp. Chem. 1999, 20, 730; (c)
1ERE = pdb code for crystal structure of estradiol in
ER-a.
In conclusion, androstene-3,5-diene 4 exhibits excellent
binding affinity and selectivity for ER-b over ER-a
and AR and is a potent ER-b agonist despite lacking
the traditional hydroxyl substitution at C-3.
References and notes
1. (a) Jordan, V. C. J. Med. Chem. 2003, 46, 883; (b) Jordan,
V. C. J. Med. Chem. 2003, 46, 1081; (c) Veeneman, G. H.
Curr. Med. Chem. 2005, 12, 1077.
2. Blizzard, T. A.; DiNinno, F.; Chen, H. Y.; Kim, S.; Wu, J.
Y.; Chan, W.; Birzin, E. T.; Yang, Y.; Pai, L.; Hayes, E.
C.; DaSilva, C. A.; Rohrer, S. P.; Schaeffer, J. M.;
Hammond, M. L. Bioorg. Med. Chem. Lett. 2005, 15,
3912.
3. (a) Blizzard, T. A.; Gude, C.; Morgan, J. D., II; Chan, W.;
Birzin, E. T.; Mojena, M.; Tudela, C.; Chen, F.; Knecht,
K.; Su, Q.; Kraker, B.; Mosley, R. T.; Holmes, M. A.;
Sharma, N.; Fitzgerald, P. M. D.; Rohrer, S. P.;
Hammond, M. L. Bioorg. Med. Chem. Lett. 2006, 16,
834; (b) Blizzard, T. A.; Gude, C.; Chan, W.; Birzin, E. T.;
Mojena, M.; Tudela, C.; Chen, F.; Knecht, K.; Su, Q.;
Kraker, B.; Holmes, M. A.; Rohrer, S. P.; Hammond, M.
L. Bioorg. Med. Chem. Lett. 2007, 17, 2944.
8. The IC50 values were generated in a scintillation proximity
estrogen receptor ligand binding assay conducted in NEN
Basic Flashplates using tritiated estradiol and full length
recombinant human ER-a or ER-b proteins. Compounds
were evaluated in duplicate in a single assay. This assay
provides IC50 values that are reproducible to within a factor of
2–3.
9. This assay was run in agonist mode at Merck-CIBE in
Spain using the procedure described by Barkhem, T.;
Carlsson, B.; Nilsson, Y.; Enmark, E.; Gustafsson, J.;
Nilsson, S. Mol. Pharmacol. 1998, 54, 105, Compounds
were tested in triplicate; results were generally reproduc-
ible to within a factor of 2–3.
10. Chen, F.; Knecht, K.; Leu, C.; Rutledge, S. J.; Scafonas,
A.; Gambone, C.; Vogel, R.; Zhang, H.; Kasparcova, V.;
Bai, C.; Harada, S.; Schmidt, A.; Reszka, A.; Freedman,
L. J. Ster. Biochem. Mol. Biol. 2004, 91, 247.