Ring nitrogen-substituted non-steroidal estrogens: Pyridine and pyrimidine analogs of the phenol in deoxyhexestrol experience resonance constraints on preferred ligand conformation

Article abstract of DOI:10.1016/j.bmcl.2004.09.048

Pyridine and pyrimidine analogs of the non-steroidal estrogen deoxyhexestrol were synthesized. Their low affinity for the estrogen receptor is ascribed, in part, to resonance enforcement of a conformation unfavorable for binding. To develop compounds selective for estrogen receptor beta (ERβ), we substituted hydroxypyridine and pyrimidine heteroaryl groups for the characteristic phenol ring of non-steroidal estrogens. The unexpectedly low affinity showed by some of these compounds is ascribed, in part, to a resonance-enforced conformational constraint that prevents their optimal accommodation in the ER ligand binding pocket.

Full text of DOI:10.1016/j.bmcl.2004.09.048

Bioorganic & Medicinal Chemistry Letters 14 (2004) 5835–5839
Ring nitrogen-substituted non-steroidal estrogens: pyridine and
pyrimidine analogs of the phenol in deoxyhexestrol experience
resonance constraints on preferred ligand conformation
Meri De Angelis and John A. Katzenellenbogen*
Department of Chemistry, University of Illinois, 600 South Mathews Avenue, Urbana IL 61801, USA
Received 18 May 2004; revised 13September 2004; accepted 17 September 2004
Available online 5 October 2004
Abstract—To develop compounds selective for estrogen receptor beta (ERb), we substituted hydroxypyridine and pyrimidine
heteroaryl groups for the characteristic phenol ring of nonsteroidal estrogens. The unexpectedly low affinity showed by some of
these compounds is ascribed, in part, to a resonance-enforced conformational constraint that prevents their optimal accommodation
in the ER ligand binding pocket.
Ó 2004 Published by Elsevier Ltd.
The estrogen receptor (ER) is a member of the nuclear
hormone receptors superfamily and plays an important
role in the development and function of female repro-
ductive system and in the maintenance of bone mineral
density and cardiovascular health.1–3 Despite these pos-
itive effects, the activation of ER in some tissues such as
breast and uterus can increase the risk of cancer in these
sites.4 Consequently, the development of compounds
that can maintain the benefits of estrogen while avoiding
the risks is a major current challenge. Such compounds
are termed selective estrogen receptor modulators or
SERMs.5
a challenge for the development of ER subtype-selective
compounds. There are a number of estrogens with good
selectivity for ERa,9,10 but fewer ERb-selective com-
pounds are known. The isoflavone phytoestrogen geni-
stein (1) shows 20-fold selectivity toward ERb,11 and
similar selectivity is reported for some aryl benzothio-
phene derivatives;12 we reported a 70-fold selectivity
for compound 2 (Fig. 1).13
Recently, other more polar heterocyclic core systems—
benzothiazoles, benzimidazoles, and benzoxazoles—
have been described as ERb-selective agents;14–16
compound 3, in particular, has a remarkable 203-
fold selectivity. Notable in these ligands is a relatively
narrow structural profile, with a core system enriched
The recent discovery of a second estrogen receptor6,7
(ERb), with tissue distribution and transcriptional prop-
erties different from ERa, opens new possibilities for
developing tissue and cell-selective estrogens based on
preferential binding or activation of these two ERs.
ERa and ERb have a conserved DNA-binding domain,
and although the amino acid identity of their ligand-
binding domains is only 59%, their ligand binding pock-
ets (LBPs) are almost identical, the differences being a
somewhat smaller internal volume for ERb and the sub-
stitution of just two amino acids: Met421 in ERa corre-
sponding to Ile in ERb, and Leu384 in ERa to Met in
ERb.8 The high sequence similarity of the LBPs presents
O
O
OH
OH
OH
CN
HO
HO
1, Genistein
2, Diarylpropionitrile, DPN
HO
N
OH
F
O
Keywords: Estrogen; Ligand; Pyridine; Pyrimidine.
*
3
Corresponding author. Tel.: +1 217 333 6310; fax: +1 217 333
7325; e-mail: jkatzene@uiuc.edu
Figure 1. Some compounds selective for ERb.
0960-894X/$ - see front matter Ó 2004 Published by Elsevier Ltd.
doi:10.1016/j.bmcl.2004.09.048

5836
M. De Angelis, J. A. Katzenellenbogen / Bioorg. Med. Chem. Lett. 14 (2004) 5835–5839
Br
Br
OH
OH
a,b
H
N
N
H3CO
Br
H
7
8 (80%)
HO
HO
R'
4, Hexestrol
5, Diethyl stilbestrol (DES)
N
R'
c
HN
N
R
9
H3CO
H
R
10a R = Et, R'= H (85%)
10b R = H, R'= Et (85%)
H
9a R = Et, R' = H
9b R = H, R'= Et
HO
e
6, Deoxyhexestrol
d,e
Figure 2. Reference compounds for pyridine and pyrimidine analogs.
N
N
N
11 (70%)
N
HO
HO
in heteroatoms, characteristics that appear favor
ERb-selective binding (though do not appear to be
essential).17 Recently, deoxyhexestrol (Fig. 2), a com-
pound we first examined long before the discovery of
ERb, was tested again for its binding to both ERs; it
had good affinity, especially for ERb, being in this
regard more preferential than its congeners hexestrol
and diethylstilbestrol (DES) (Table 1).
12 (80%)
Scheme 1. Reagents and conditions: (a) BuLi, B(OiPr)3, H2O2, ether
À78°C; (b) NaH, MeI, DMF, rt; (c) Pd2(dba)3 (2mol%), NaOtBu,
( )-Binap (4mol%), toluene 70°C (for 10a) Pd2(dba)3 (1.5mol%),
NaOtBu, (À)Ppfa (4.5mol%), toluene, 70°C (for 10b); (d) NaH, EtI,
DMF, reflux; (e) NaH, EtSH, DMF, reflux.
To develop other ERb-selective ligands of novel struc-
ture, we considered analogs of deoxyhexestrol in which
the phenolic ring would be replaced with either 5-
hydroxypyridine or 5-hydroxypyrimidine units (see 12
and 17, both of which also contain a 2-amino group).
In this manner, we would be substituting heterocyclic
analogs for the phenol, the most characteristic function-
ality of estrogens, a group that is present in almost all
nonsteroidal estrogens and is known to be a dominant
feature in the binding of estrogens to the ER. By basing
the design of these new ligands on deoxyhexestrol, we
ensured that there would not be another phenol in the
molecule. Thus, we presumed that the hydroxypyridine
or hydroxypyrimidine unit would, of necessity, replace
the phenol in binding to the ERs. Also, because these
two heteroaryl moieties have somewhat lower hydro-
phobicities than phenols, yet (particularly with the 2-
amino group) are not excessively basic (see below), we
thought they might be well accommodated within the
ER LBPs, perhaps preferentially within what appears
to be the more polar-tolerant ERb LBP. Some aryl sys-
tems containing one or more nitrogen atoms have been
reported as ER ligands,18,19 and our selection of the
pyridine and pyrimidine systems was encouraged by
these precedents, as well as ease of synthesis considera-
tions. Herein, we present the synthesis of six new pyri-
dine- and pyrimidine-based nonsteroidal ER ligands
(compounds 11, 12, 16, and 17, 21a, 21b) and an analysis
of their ER binding affinities.
The efficient synthesis of pyrimidine derivatives 11 and
12 was accomplished by the procedure shown in Scheme
1. It was possible to synthesize 2-bromopyridinyl-5-
boronic ester from 2,5-dibromopyridine (7) through a
selective C-5 lithiation,20 followed by quenching with
triisopropyl borate. Oxidation of the resulting boronic
ester with hydrogen peroxide21 was followed by protec-
tion of a hydroxyl group as the methyl ether, giving
compound 8. Palladium-mediated coupling of this
bromopyridine with the substituted benzyl amines 9a,b
gave aminopyridine products 10a,b, but the best yields
were obtained with different catalytic systems: Pd2(dba)3
and ( )Binap for 10a and Pd2(dba)3 and (À)Ppfa for
10b.
Table 1. ERa and ERb binding affinity and selectivity, and ClogP of
compounds in this study and related compounds for comparison
Compound
Relative binding
affinity (RBA)a
b/a Ratio
ClogP
ERa
ERb
Hexestrol (4a)
Isobutestrol (4b)
280
2.2
430
18
700
40
2.5
18
5.11
4.18
4.95
DES (5)
Deoxyhexestrol (6)
280
734
0.6
Compound 10a was then alkylated and subsequently
deprotected to give derivative 12. Compound 11 was ob-
tained from derivative 10b after a facile deprotection.
Derivatives 10a,b were both deprotected under basic
conditions, instead of the usual acid ones, because com-
pounds 11 and 12 are not stable in acid.
5.78
11
12
0.019
0.026.94 1.33
0.030
0.007
0.007
0.081
0.010
0.077
0.004
2.5
0.5
2.0
8.4
4.78
3.51
4.35
5.23
16
17
0.014
0.684
0.0232.3
21a
21b
5.23
a RBA values are determined by a competitive binding assay, using
[3H]estradiol as tracer, according to a published method.22 We have
found that this assay consistently gives RBA values with CVs of
60.3.
The synthesis of pyrimidine analogs 16,17 shown in
Scheme 2, was accomplished in good yield, in 2–3steps.
In a sealed tube, 2-choloro-5-bromo-pyrimidine (13) and
benzyl amines 14a,b were heated at 150°C for 40min, in

M. De Angelis, J. A. Katzenellenbogen / Bioorg. Med. Chem. Lett. 14 (2004) 5835–5839
5837
a competitive radiometric assay; affinities are expressed
relative to that of estradiol (100%) to give relative bind-
ing affinity (RBA) values (Table 1).22 The binding affin-
ity of all of the nitrogen-substituted analogs (11, 12, 16,
17, 21a,b) was much lower than that of the reference
compounds (4a,b, 5, 6). The erythro derivative 21a is
the highest binder and most ERb-selective compound
of the new compounds, the other compounds binding
considerably less well than 21a.
N
Cl
R'
N
N
a
N
Br
13
N
R
Br
+
15a R=Et, R'=H (85%)
15b R=H, R'=Et (85%)
R'
HN
c
b,c
R
14a R=Et, R'=H
14b R=H, R'=Et
N
N
N
N
It is of note that the aminopyridine and aminopyrimi-
dine systems (11, 12 and 16, 17) are not very basic (cal-
culated pKa values are 6.7 and 3.9 for the two series;
ACD Labs), nor do any of the ligands studied have an
overall lipophilicity (estimated from ClogP) that places
them far out of the range of other high affinity ER lig-
ands; yet, they are still not effective analogs of the carbo-
cyclic systems, such as deoxyhexestrol (6). Nevertheless,
some trends in the binding data should be noted: (1)
RBA values increase when the central atoms that link
the two rings are both alkylated (compare 11 vs 12
and 16 vs 17, Table 1), consistent with the higher affinity
of the doubly alkylated all-carbon analogs, hexestrol
(4a) versus isobutestrol (4b).23 From X-ray crystal struc-
tures, it is known that these two ethyl groups nicely fill
preformed pockets in the ER that are above position
C-11b and below position C-7a of estradiol.24 (2) Of
the two series, the pyridine derivative 21a binds 2–10
times better than compound 12, and derivatives 11, 12
bind 2–5 times better than the pyrimidine analogs 16,
17, consistent with the higher ClogP values of the for-
mer system (Table 1). (3) The erythro isomer 21a binds
better than threo isomer 21b, consistent with the behav-
ior reported for their phenol analogs.23
N
N
HO
HO
16 (33%)
17 (28%)
Scheme 2. Reagents and conditions: (a) iPrOH, diisopropylethylam-
ine, 150°C, sealed tube; (b) NaH, EtI, DMF, rt; (c) BuLi, B(OiPr)3,
H2O2, À78°C.
the presence of iPr2NEt, to give the aminopyrimidines
15a,b. Derivative 15a was then alkylated, and the
bromine substituent converted to the OH group as
done previously for compound 8, giving compound 17.
Compound 16 was obtained from derivative 15b in a
similar manner.
To better compare the activity of the previous deriva-
tives with deoxyhexestrol, we also planned to synthesize
pyridine analogs in which the benzylic nitrogen is substi-
tuted with a carbon, and compounds in which a p-OH-
aniline replaces the phenol. The synthesis of the pyridine
derivatives (Scheme 3) was accomplished treating pico-
line 18, with LDA in the presence of 1-bromo-propyl-
benzene 19. The compound obtained was then
alkylated and deprotected to give diasteroisomers 21a,
21b. Unfortunately, we were unable to prepare the cor-
responding aniline analogs, because these derivatives
proved to be very unstable after deprotection.
One possible reason for the low affinity shown by deriv-
atives 11, 12, 16, and 17 is that they lack a second hydr-
oxyl group, but comparing the RBA data of these
derivatives with that of the reference monophenol 6, it
seems that they are in some other general ways poorly
suited for the ER LBPs. One possible factor could arise
from resonance overlap between the heterocycle and the
2-amino substituent: the need for p overlap is expected
to induce sp2 hybridization in the benzylic nitrogen
whose substituents should consequently be constrained
to be in the same plane as the heterocycle (angle h in
Fig. 3). This constraint could also have a secondary ef-
fect on the conformation of the amine substituents, with
the result that the overall conformation of the heterocy-
clic analog of deoxyhexestrol might be such that it can-
not be optimally accommodated within the ER LBP. To
investigate this possibility, we used quantum mechanical
calculations (6-31G*/B3LYP) to compare minimum en-
ergy conformations of hexestrol, DES, and derivatives
12, 17 (Fig. 3).
Because nonsteroidal bisphenols are generally of higher
affinity ligands than monophenols, we also attempted to
prepare a series of derivatives in which the phenyl group
was replaced by a 4-hydroxyphenyl ring. However, these
compounds also proved to be unstable after deprotec-
tion, presumably experiencing a facile 1,6-elimination
of the benzylic amine via
intermediate.
a
quinonemethide
The binding affinity of the six compounds prepared
above, for human ERa and ER was determined using
a,b
Br
+
N
N
It is of note that the minimum energy conformation of
DES matches quite closely the conformation it was
shown to adopt in the crystal structure with ERa: the
angle h shown in Figure 3 is close to 90°, as in the experi-
mental structure.24 The computed structure of hexestrol
is similar to that of DES, with h = 86° and the ethyl sub-
stituent equally extended. By contrast, because of the
TBDMSO
RO
18
19
20a-erythro R= OTBDMS (25%)
20b-threo R= OTBDMS (15%)
21a-erythro R= H (quant.)
21b-threo R=H (quant.)
c
Scheme 3. Reagents and conditions: (a) LDA, À78°C, THF; (b) LDA,
EtBr, À78°C, THF; (c) TBAF, 0°C, THF.

5838
M. De Angelis, J. A. Katzenellenbogen / Bioorg. Med. Chem. Lett. 14 (2004) 5835–5839
Figure 3. Dihedral angle h is defined by atoms 1–4.
stronger resonance-induced planarity, the hetero-
arenes are constrained to much lower dihedral angles,
between 13° and 23° with respect to the central bond
(Fig. 3).
In summary, we have investigated the substitution of the
phenolic ring, a characteristic structural motif in estro-
gen receptor ligands, with a 5-hydroxypyridine or a 5-
hydroxypyrimidine, to investigate the extent of hetero-
atom tolerance of the ER subtypes. Although derivative
21a is more selective then deoxyhexestrol (6) toward
ERb, with an RBA close to 1% for this receptor, all
other compounds prepared, however, showed surpris-
ingly weak affinity for both receptors. Nevertheless,
the low binding affinity of these last pyridine and pyrim-
idine non-steroidal estrogens (12 and 17) is instructive
because it shows that introduction of nitrogen hetero-
atoms within the flexible structure of a high affinity
all-carbon ligand, deoxyhexestrol, can dramatically
reduce binding affinity, even without altering the overall
hydrophobicity of the ligand, most likely because of
resonance-induced conformational constraints. This
information will prove useful in directing our continued
efforts to prepare novel ER ligands with high affinity
and subtype selectivity.
DES (and by analogy, hexestrol) fits very well to the ER:
one phenol fits in the narrow A-ring binding pocket just
does as the A-ring of estradiol, and because of the 86°
ring/3-hexene backbone torsional angle, the two ethyl
groups can nicely fill the major 7a, 11b subpockets.24
By contrast, because of the more pronounced amine-
heterocycle resonance, the pyridine and pyrimidine ana-
logs of deoxyhexestrol, 12 and 17, appear to be forced to
adopt a conformation in which the backbone is almost
coplanar with the hydroxy-containing heteroarene (pre-
sumed to be the A-ring phenol mimic), with the result
that the two ethyl groups are not well disposed to fill
the 7a, 11b subpockets. Therefore, we believe that the
low ER binding affinity of the 2-aminopyridine and
pyrimidine analogs for ER arises—in large part—from
a nonoptimal conformation, strongly enforced by reso-
nance, which is poorly suited for binding to the receptor,
although other factors (unsatisfied polar sites on hetero-
aryl groups or alterations in hydrogen bonding) might
also be playing a role. Although this argument alone
cannot account for the low affinity of the pyridine ana-
log 21a (which has a calculated h value of 79°, similar to
hexestrol), it is of note that this system is considerably
more basic (calculated pKa of 6.0) and its affinity is con-
siderably higher than that of the corresponding 2-amino
analog 12 on both ERs.
Acknowledgements
This research was supported by an NIH grant (PHS
5R37 DK15556). Funding for NMR instrumentation
was provided in part by the W. M. Keck Foundation,
NIH (PHS 1 S10 RR104444-01) and NSF (NSF CHE
96-10502). Funding for mass spectrometers was pro-
vided by NIH (GM 27029 and RR 01575), and NSF
(PCM 8121494). We thank Kathryn Carlson for the

M. De Angelis, J. A. Katzenellenbogen / Bioorg. Med. Chem. Lett. 14 (2004) 5835–5839
5839
12. Schopfer, U.; Schoeffter, P.; Bischoff, S. F.; Nozulak, J.;
Feuerbach, D.; Floersheim, P. J. Med. Chem. 2002, 45,
1399.
binding assays, and Dr. Alessandro Troisi for quantum
mechanical calculations.
13. Meyers, M. J.; Sun, J.; Carlson, K. E.; Marriner, G. A.;
Katzenellenbogen, B. S.; Katzenellenbogen, J. A. J. Med.
Chem. 2001, 44, 4230.
References and notes
1. Davidson, N. E. N. Engl. J. Med. 1995, 352, 1638.
2. Barrett-Conner, E.; Bush, T. L. J. Am. Med. Assoc. 1991,
265, 1861.
14. AstraZeneca A. B. WO Patent 02/051821-A1, 2002.
15. AstraZeneca A. B. WO Patent 02/46168-A1, 2002.
16. Wyeth, WO Patent 03/050095-A1, 2003.
3. Yaffe, K.; Sawaya, G.; Lieberbung, I.; Grady, D. J. Am.
Med. Assoc. 1998, 279, 688.
17. Edsall, R. J.; Harris, H. A.; Manas, E. R.; Mewshaw, R.
E. Bioorg. Med. Chem. 2003, 11, 3457.
4. Beresford, S. A.; Weiss, N. S.; Voigt, L. F.; McKnight, B.
Lancet 1997, 349, 458.
5. Dutertre, M.; Smith, C. L. J. Pharm. Exp. Ther. 2000, 295,
431.
18. Henke, B. R.; Drewry, D. H.; Jones, S. A.; Stewart, E. L.;
Weaver, S. L.; Wiethe, R. W. Bioorg. Med. Chem. Lett.
2001, 11, 1939.
19. Henke, B. R.; Consler, T. G.; Go, N.; Hale, R. L.;
Hohman, D. R.; Jones, S. A.; Lu, A. T.; Moore, L. B.;
Moore, J. T.; Orband-Miller, L. A.; Robinett, R. G.;
Shearin, J.; Spearing, P. K.; Stewart, E. L.; Turnbull, P. S.;
Weaver, S. L.; Williams, S. P.; Wisley, G. B.; Lambert, M.
H. J. Med. Chem. 2002, 45, 5492.
6. Mosselman, S.; Polman, J.; Dijekema, R. FEBS Lett.
1996, 392, 49.
7. Kuiper, G. G. J. M.; Enmark, E.; Pelto-Huikko, M.;
˚
Nilsson, S.; Gustafsson, J.-A. PNAS 1996, 93, 5925.
8. Pinke, A. C. W.; Brzozowski, A. M.; Hubbard, R. E.;
Bonn, T.; Thorsell, A.-G.; Engstro¨m, O.; Ljunggren, J.;
20. Bouillon, A.; Lancelot, J.-C.; Collot, V.; Bovy, P. R.;
Rault, S. Tetrahedron 2002, 58, 2885.
˚
Gustafsson, J.-A.; Carlquist, M. EMBO J. 1999, 18, 4608.
9. Sun, J.; Meyers, M. J.; Fink, B. E.; Rajendran, R.;
Katzenellenbogen, J. A.; Katzenellenbogen, B. S. Endocr-
inology 1999, 140, 800.
10. Mortensen, D. S.; Rodriguez, A. L.; Carlson, K. E.; Sun,
J.; Katzenellenbogen, B. S.; Katzenellenbogen, J. A. J.
Med. Chem. 2001, 44, 3838.
21. Krow, G. R.; Xiao, Y.; Cannon, K.; Swan, S. A.; Nickel,
A. Synth. Commun. 2000, 30(22), 4093.
22. Carlson, K. E.; Choi, I.; Gee, A.; Katzenellenbogen, B. S.;
Katzenellenbogen, J. A. Biochemistry 1997, 14897.
23. Kilbourn, M. R.; Arduengo, A. J.; Park, J. T.; Katzenel-
lenbogen, J. A. Mol. Pharmacol. 1981, 19, 388.
24. Shiau, A. K.; Barstad, D.; Loria, P. M.; Cheng, L.;
Kushner, P. J.; Agard, D. A.; Greene, G. L. Cell 1998, 95,
927.
11. Kuiper, G. G. J. M.; Carlsson, B.; Grandien, J.; Enmark,
˚
E.; Ha¨ggblad, J.; Nilsson, S.; Gustafsson, J.-A. Endocri-
nology 1997, 138, 863.