Design, Synthesis, and in Vitro Biological Evaluation of Small Molecule Inhibitors of Estrogen Receptor α Coactivator Binding

Article abstract of DOI:10.1021/jm030404c

Nuclear receptors (NRs) complexed with agonist ligands activate transcription by recruiting coactivator protein complexes. In principle, one should be able to inhibit the transcriptional activity of the NRs by blocking this transcriptionally critical receptor-coactivator interaction directly, using an appropriately designed coactivator binding inhibitor (CBI). To guide our design of various classes of CBIs, we have used the crystal structure of an agonist-bound estrogen receptor (ER) ligand binding domain (LBD) complexed with a coactivator peptide containing the LXXLL signature motif bound to a hydrophobic groove on the surface of the LBD. One set of CBIs, based on an outside-in design approach, has various heterocyclic cores (triazenes, pyrimidines, trithianes, cyclohexanes) that mimic the tether sites of the three leucines on the peptide helix, onto which are appended leucine residue-like substituents. The other set, based on an inside-out approach, has a naphthalene core that mimics the two most deeply buried leucines, with substituents extending outward to mimic other features of the coactivator helical peptide. A fluorescence anisotropy-based coactivator competition assay was developed to measure the specific binding of these CBIs to the groove site on the ER-agonist complex with which coactivators interact; control ligand-binding assays assured that their interaction was not with the ligand binding pocket. The most effective CBIs were those from the pyrimidine family, the best binding with K_i values of ca. 30 μM. The trithiane- and cyclohexane-based CBIs appear to be poor structural mimics, because of equatorial vs axial conformational constraints, and the triazene-based CBIs are also conformationally constrained by amine-substituent-to-ring resonance overlap, which is not the case with the higher affinity alkyl-substituted pyrimidines. The pyrimidine-based CBIs appear to be the first small molecule inhibitors of NR coactivator binding.

Full text of DOI:10.1021/jm030404c

600
J . Med. Chem. 2004, 47, 600-611
Design , Syn th esis, a n d in Vitr o Biologica l Eva lu a tion of Sm a ll Molecu le
In h ibitor s of Estr ogen Recep tor r Coa ctiva tor Bin d in g
Alice L. Rodriguez, Anobel Tamrazi, Margaret L. Collins, and J ohn A. Katzenellenbogen*
Department of Chemistry, University of Illinois, 600 South Mathews Avenue, Urbana, Illinois 61801
Received August 19, 2003
Nuclear receptors (NRs) complexed with agonist ligands activate transcription by recruiting
coactivator protein complexes. In principle, one should be able to inhibit the transcriptional
activity of the NRs by blocking this transcriptionally critical receptor-coactivator interaction
directly, using an appropriately designed coactivator binding inhibitor (CBI). To guide our design
of various classes of CBIs, we have used the crystal structure of an agonist-bound estrogen
receptor (ER) ligand binding domain (LBD) complexed with a coactivator peptide containing
the LXXLL signature motif bound to a hydrophobic groove on the surface of the LBD. One set
of CBIs, based on an outside-in design approach, has various heterocyclic cores (triazenes,
pyrimidines, trithianes, cyclohexanes) that mimic the tether sites of the three leucines on the
peptide helix, onto which are appended leucine residue-like substituents. The other set, based
on an inside-out approach, has a naphthalene core that mimics the two most deeply buried
leucines, with substituents extending outward to mimic other features of the coactivator helical
peptide. A fluorescence anisotropy-based coactivator competition assay was developed to
measure the specific binding of these CBIs to the groove site on the ER-agonist complex with
which coactivators interact; control ligand-binding assays assured that their interaction was
not with the ligand binding pocket. The most effective CBIs were those from the pyrimidine
family, the best binding with Ki values of ca. 30 µM. The trithiane- and cyclohexane-based
CBIs appear to be poor structural mimics, because of equatorial vs axial conformational
constraints, and the triazene-based CBIs are also conformationally constrained by amine-
substituent-to-ring resonance overlap, which is not the case with the higher affinity alkyl-
substituted pyrimidines. The pyrimidine-based CBIs appear to be the first small molecule
inhibitors of NR coactivator binding.
In tr od u ction
thylation).1,2 X-ray crystal structures have been ob-
tained of several NR-agonist complexes with peptides
or protein fragments of p160 coactivators encompassing
one or more of the nuclear receptor interaction boxes
(NR boxes) that contain the signature LXXLL sequence
motif.3-11 These structures show that the coactivator
interacts with the NR LBD through a two-turn amphi-
pathic R-helical motif that places the first and third
leucine residues in a deep, but short, hydrophobic groove
that is made up of several residues from helices 3, 4, 5,
and 12 of the LBD. In addition, the intrinsic dipole
moment of the coactivator helix is aligned with polar
residues, a glutamate residue at its N-terminus, and a
lysine residue at its C-terminus in the case of the
estrogen receptor (ER), which together form a “charge
clamp” for the coactivator helix within the ER LBD.
Nuclear receptors (NRs) are ligand-modulated tran-
scription factors that mediate the action of steroid
hormones and various other bioactive ligands. The
members of the NR gene superfamily have distinct
domain structures, with a highly conserved DNA-
binding domain (domain C) and a moderately conserved
ligand-binding domain (LBD, domain E) being sepa-
rated by a nonconserved hinge domain (domain D) and
flanked by an N-terminal modulatory domain (A/B
domain) and a C-terminal region (F domain). Many of
the actions of NRs are initiated by hormone binding to
the LBD, a process that stabilizes the conformation of
this domain. When an agonist ligand binds to an NR
LBD, this conformational stabilization rigidifies surface
features that function as docking sites for the nuclear
receptor interaction regions of different coregulatory
protein complexes.1,2
Of the coregulator complexes implicated in NR action,
the most fully investigated is the chromatin-modifying
complex, in which a protein from the p160 class of
coactivators, also called steroid receptor coactivators
(SRCs), binds to the NR and functions as a platform
protein for the docking of other proteins (CBP/p300,
pCAF, CARM, etc.) that have enzymatic activity for
derivatizing histones (acetylation, phosphorylation, me-
The general strategy used thus far to block the action
of NR hormones has been to develop antagonists or
antihormones. Such compounds compete with agonist
ligands for binding to the ligand binding pocket, and
upon binding they induce different conformations in the
LBD.3,4,9,12-14 These antagonist conformations either
preclude the binding of the coactivators, by repositioning
the C-terminal LBD helix 12 so that it occupies the
coactivator binding groove itself, or they facilitate the
binding of corepressor proteins by disposing helix-12
away from the LBD itself.
* To whom correspondence should be addressed. Phone: 217 333
6310. Fax: 217 333 7325. E-mail: jkatzene@uiuc.edu.
10.1021/jm030404c CCC: $27.50 © 2004 American Chemical Society
Published on Web 12/30/2003

Inhibitors of ERR Coactivator Binding
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 601
F igu r e 1. Electrostatic surface rendering of (A) the crystal structure of GRIP1 peptide bound to ERR and (B) pyrimidine CBI
12a docked into ERR. The receptor is shown as a continuous surface where red coloring indicates positive charge and blue coloring
indicates negative charge. 12a is colored by atom type, whereas the coactivator peptide is depicted as a red tube.
Curiously, the effectiveness of NR antagonists can be
compromised by cellular adaptations that seem to
enable coactivators to bind to NR-antihormone com-
plexes.15 This appears to be the case with the develop-
ment of tamoxifen resistance in the treatment of breast
cancer: In tamoxifen-resistant breast tumor cells, the
ER complex with trans-4-hydroxytamoxifen, the bioac-
tive metabolite of tamoxifen, seems able to interact with
coactivators, therebysremarkablysreversing its role
and activating rather than repressing transcription.16,17
An alternative approach to inhibiting the gene-
regulating effects of NRs might be to develop compounds
capable of blocking NR-coactivator interaction itself.
A proof-of-principle demonstration of the effectiveness
of this approach is the observation that peptides having
the LXXLL sequence are able to block gene transcription
induced by estrogen agonists working through the
estrogen receptor (ER).18,19 Thus, small molecule ana-
logues that could perform the same function as these
peptidessacting as coactivator binding inhibitors (CBIs)
that block the ER-SRC interactionsmight have a net
antagonist effect that was independent of ER ligand
binding. This approach to antagonizing the action of the
ER might not be compromised by cellular adaptations,
such as increased expression of certain coactivator
proteins15 or covalent modifications of the coactivators
or the ER itself,16,17 that are able to surmount the
antagonist effect of ligand-based antiestrogens, such as
tamoxifen.
diethylstilbestrol and a 13 amino acid peptide derived
from the p160 class coactivator, GRIP1 (SRC-2),1,2
containing a single NR box (Figure 1A). The interaction
surface in the ER consists of 16 amino acid residues
from helices 3, 4, 5, and 12: L354, V355, I358, A361,
K362 (helix 3); L372 (helix 3,4 turn); F367, V368 (helix
4); Q375, V376, L379, E380 (helix 5); and D538, L539,
E542, M543 (helix 12). Three zones of interaction are
evident: L690 and L694 of the coactivator peptide are
entirely engulfed by the ER, forming a deep, strong
hydrophobic groove interaction; I689 and L693 of the
coactivator peptide interact with the receptor on one-
half of their surface, forming a weaker but apparently
significant surface hydrophobic interaction; E542 and
K362 of the ER form the charge clamp properly aligned
for interaction with the inherent dipole of the R-helix
and finally various amide functionalities on the helix
backbone, locking and holding the hydrophobic side
chains into position.
Cla ss I Design . The first approach we took to
designing small molecule inhibitors of coactivator bind-
ing was termed the “outside-in” approach. A head-on
view of the coactivator peptide shows that the positions
of the three leucine residues in the LXXLL signature
sequence motif, as registered by their CR atoms, fall
roughly into the shape of an equilateral triangle, as
shown in Figure 2. Thus, in this approach, a central core
having dimensions corresponding to this CR triangle was
first chosen, and then hydrophobic substituents were
attached in a manner that mimicssin a topologically
faithful fashionsthe positions of the three leucine
residues of the coactivator peptide. Molecular modeling
was used to refine the design of these class I inhibitors.
These CBIs are said to follow an outside-in design,
because they begin with a mimic for the helix backbone,
which is outside of the hydrophobic groove, and then
proceed inward by adding residue elements.
In this report, we describe de novo, structure-based
design approaches we have developed to prepare two
types of coactivator binding inhibitors, and we present
an in vitro assay method through which we can dem-
onstrate that some of these compounds are able to block
the interaction of ER with NR box peptides. Thus, we
provide the first proof-of-principle that small molecules
can function as effective inhibitors of the binding of
coactivator proteins to the estrogen receptor.
Four scaffolds were chosen as candidate cores for the
class I CBIs: triazene, pyrimidine, trithiane, and cy-
clohexane. Hydrophobic substituents were then placed
in alternate positions around the rings to mimic the
leucine residues, and in some cases, polar functionalities
were added to interact with the charge clamp residues
Resu lts
Str u ctu r e-Ba sed Design of Coa ctiva tor Bin d in g
In h ibitor s. We have based our design of CBIs on a
recent crystal structure of the ERR LBD complexed with

602 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Rodriguez et al.
F igu r e 2. Class I and class II approaches to rational design of ERR coactivator-binding inhibitors (CBIs). The central figure
depicts the GRIP1 coactivator peptide from which we based our design with NR box residues indicated by single letter amino acid
notation. The CR carbons are colored yellow. Class I design is shown to the left of the helix while class II design in shown to the
right. The atomic distances between the CR carbons of the NR box residues are indicated in each design and are mimicked by the
residue elements in the CBIs.
Sch em e 1. Synthesis of Triazene-Core CBIs
in the ER LBD. Molecules were built using a molecular
modeling platform (SYBYL), minimized, and then over-
laid onto the coactivator peptide. Adjustments were
made to the design, and the molecule was then remi-
nimized and docked into the coactivator binding groove.
A set of compounds was then chosen for synthesis based
on a combination of molecular modeling results and
relative ease of synthesis. Further structural issues
related to specific scaffolds in the outside-in design are
discussed in the Synthesis section.
Cla ss II Design . We termed our second approach to
CBIs the “inside-out” design. From the face of the
R-helix, the four hydrophobic residues of the NR box
approximate the shape of a parallelogram, as shown in
Figure 2. Accordingly, we envisioned using a hydropho-
bic small molecule to mimic both of the most deeply
buried groove residues, L690 and L694, which are
considered most crucial for stabilizing the ER-coacti-
vator interface. We hypothesized that a hydrophobic
unit of appropriate dimensions could be used to fill the
groove portion of the coactivator binding pocket, without
specifically mimicking the shape of the individual
leucine residues. From this hydrophobic small molecule
mimic of the two inside leucines, flexible linkers then
extend outward, hence the term inside-out design. These
linkers would incorporate other relevant functionalities,
including the two remaining hydrophobic residues, I689
and L693, and in some cases the polar charge-clamp
functionalities. Following the modeling protocol dis-
cussed above, compounds derived from substituted
naphthalenes were chosen for synthesis.
We expanded our study of the triazene class of
compounds to include ones containing substituents
linked directly through a carbon atom rather than
nitrogen. By obviating the resonance interaction of the
substituent nitrogen lone pair with the triazene ring,
this carbon-for-nitrogen substitution increases the tor-
sional flexibility of the substituents and effectively
eliminates the steric energy penalty encountered when
the isobutyl substituents of these triazene mimics are
overlapped with three leucines in the coactivator pep-
tide. In fact, using MacroModel, we estimate that the
torsional barrier for rotation of a nitrogen substitutent
Syn th esis. Tr ia zen es. Triaminotriazenes were syn-
thesized in a single step from commercially available
cyanuric chloride (Scheme 1). Triazene 1a was synthe-
sized by reacting cyanuric chloride with isobutylamine
to produce a compound having substituents that directly
mimic the three branched leucine amino acids. Triaz-
enes 1b -e were synthesized by reacting cyanuric
chloride with a variety of alkyl- and arylamines con-
taining simple straight-chain and cyclic alkyl and aryl
substituents.

Inhibitors of ERR Coactivator Binding
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 603
Sch em e 3. Synthesis of Pyrimidine-Core CBIs
Sch em e 2. Synthesis of Other Triazene-Core CBIs
on a triazene to be ca. 9 kcal/mol, whereas rotation of a
carbon substituent on a triazene or pyrimidine is less
than 2 kcal/mol. The synthesis of a monoalkylated
triazene began with the addition of 3-methylbutylmag-
nesium bromide to cyanuric chloride. This reaction was
run at low temperatures to minimize over addition of
the Grignard reagent. Subsequent addition of an excess
of isobutylamine afforded the differentially trisubsti-
tuted triazene 3.
Sch em e 4. Synthesis of Trithiane- and
Cyclohexane-Core CBIs
To incorporate an acid functionality for interaction
with K362 involved in the charge clamp, intermediate
2 was reacted with a single equivalent of isobutylamine
to give disubstituted triazene 4 (Scheme 2). Amines 5a ,b
were synthesized according to literature procedures (as
described in the Supporting Information) and were
reacted with the disubstituted triazenes 4 to give fully
substituted products 6a ,b. Hydrolysis of the esters
provided the desired triazenes 7a ,b. The predicted
torsional rigidity of the C-N bonds of the amine
substituents became apparent with the addition of the
final amines. Variable-temperature 1H NMR spectro-
scopy demonstrated the presence of two rotameric
species in each case (data not shown).
P yr im id in es. As discussed in the triazene section,
pyrimidine compounds were examined because they
have greater torsional flexibility than the triazene
compounds. A series of pyrimidine core CBIs was
synthesized by reacting â-keto esters 8a -c with thio-
urea to afford thiouracil derivatives 9a -c (Scheme 3).
The â-keto esters were either commercially available
(8b) or were made conveniently from the corresponding
aldehyde or ketone. Reaction of the aldehyde with ethyl
diazoacetate and SnCl2 provided the desired â-keto
esters in the case of 8c. Reaction of the methyl ketone
with sodium hydride and diethyl carbonate provided the
â-keto ester 8a .
vides uracil derivatives 10a -c and mercaptoacetic acid
as a byproduct. Phosphorus oxychloride transformed
uracils 10a -c to dichloropyrimidines 11a -c. Reaction
of the dichloropyrimidines 11a -c with the appropriate
amine afforded the desired series of pyrimidines 12a -
c. Both monoaminated products were obtained as
byproducts in most cases, contributing to decreased
yields in the final step.
Tr ith ia n e. The CBI containing a trithiane core was
synthesized from commercially available 1,3,5-trithiane
(Scheme 4). The starting material was reacted with
n-BuLi, followed by 1-bromo-3-methylbutane in three
successive additions, to yield the trisubstituted trithiane
13. According to the literature, a single conformational
isomer with all three alkyl substituents in the equatorial
positions is obtained by controlling the temperature of
addition. By performing the lithiation step at -45 °C
and the substitution at -20 °C, a single isomer was
The thioketone moiety of the thiouracils 9a -c was
converted to a ketone by reaction with chloroacetic acid.
This transformation is thought to proceed through a
spirolactone intermediate, which upon hydrolysis pro-

604 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Sch em e 5. Synthesis of Naphthalene-Core CBIs
Rodriguez et al.
F igu r e 3. Recruitment of TMR-labeled NR box peptide to the
ER as followed by fluorescence anisotropy. Estradiol-bound ER
(E2) recruits the peptide as demonstrated by an increase in
anisotropy, whereas hydroxytamoxifen-bound ER (OHT) does
not recruit the peptide.
obtained. On the basis of the symmetry evident in the
1H NMR spectrum, it is presumed to be the all-
equatorial isomer (data not shown). The conversion of
this triequatorial trialkyltrithiane to the all-axial con-
formation, as is needed to achieve good three-dimen-
sional overlap between the three substituents and the
three leucine residues in the NR box helix, undoubtedly
involves a increase in steric energy. We have estimated
this increase to be ca. 5 kcal/mol, on the basis of
molecular mechanics calculations on both conformers
(using the MMFF94 force field within SYBYL).
F igu r e 4. Displacement of TMR-labeled NR box peptide by
unlabeled NR box 2, as shown by fluorescence anisotropy.
Cycloh exa n e. CBI 14 containing a cyclohexane core
was prepared in a single step from commercially avail-
able Kemp’s triacid (KTA), as seen in Scheme 4. KTA
is a useful cyclohexane core mimic because its confor-
mational preference is just the opposite that of the 1,3,5-
trithianes. KTA is locked into a chair conformation with
the acid groups in the axial positions, as a result of
intramolecular hydrogen bonding between the acid
groups. KTA was coupled to isobutylamine under stan-
dard coupling conditions, using EDC as a coupling
reagent.
was labeled at its N-terminus with tetramethylrho-
damine (TMR), a fluorophore chosen for its short
lifetime and low environmental sensitivity.20
An ER titration was performed to validate the binding
of the labeled peptide to the ER, and the apparent Kd
of the peptide was found to be 190 nM (Figure 3).
Interestingly, the addition of the fluorophore markedly
increased the affinity of the peptide for the ER when
compared to the unlabeled octapeptide, which was ca.
5 µM. TMR is a hydrophobic molecule, which could
make favorable contacts with hydrophobic regions around
the coactivator binding groove. A similar titration was
performed by replacing the agonist ligand estradiol with
the antagonist ligand trans-4-hydroxytamoxifen (OHT)
(Figure 3). In this case, no increase in anisotropy was
observed, because the conformational change in the ER
induced by antagonist binding renders the receptor
unable to recruit coactivator proteins or peptides.
A competition was performed to validate that the
labeled peptide is binding in the coactivator binding
groove (Figure 4). A 15 amino acid peptide containing
a single NR box, specifically the second NR box in SRC-1
(SRC-1-NR box 2, CLTERHKILHRLLQE), is known to
bind to the ER in the coactivator groove with a Kd of
approximately 1 µM.22,23 In this experiment, the con-
centration of the labeled peptide was maintained at 20
nM, while an increasing amount of SRC-1-NR box 2 was
added, reaching a maximum of 100 µM. The anisotropy
significantly decreased (by 130 mA units, which is ca.
33% of the theoretically possible range),20 indicating
displacement of the labeled peptide by SRC-1-NR box
2. In this experiment, the Ki value for the SRC-1-NR
box 2 peptide was estimated to be 700 nM, similar to
previously reported values for this NR box.22,23
Na p h th a len es. Bromination of the commercially
available 2,7-dimethylnaphthlene afforded 15, which
was reacted with potassium cyanide to give bisnitrile
16 (Scheme 5). Reduction of the nitrile functionality to
the primary amine was accomplished using borane-
THF, providing the bis(aminoethyl)-substituted inter-
mediate 17. Reaction of both of these amines with acid
chlorides yielded the naphthalene-based CBIs 18a ,b.
Th e Bin d in g of Coa ctiva tor Bin d in g In h ibitor s
to th e Estr ogen Recep tor . F lu or escen ce An istr op y
Assa y. To evaluate the potential of our CBIs to bind to
the ER LBD in a manner that competes with the
binding of coactivators, we developed a simple competi-
tive binding assay, using a fluorophore-labeled coacti-
vator peptide. Anisotropy measurements are ideally
suited for studying the association of fluorescent pep-
tides to proteins,20,21 because the low anisotropy of the
rapidly tumbling free peptide (MW ca. 1500) increases
when its mobility becomes restricted as a result of its
binding to a large protein such as the ER LBD (MW ca.
60 000 for the LBD dimer). As the peptide-fluorophore
tracer, we utilized an eight amino acid peptide contain-
ing a single NR box with the sequence NH2-IL-
RKLLQE-CO2H to mimic the coactivator protein. It

Inhibitors of ERR Coactivator Binding
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 605
pocket can tolerate hydrophobic substituent function-
alities that differ from the direct branched leucine
mimics.
CBIs 13 and 14 containing trithiane and cyclohexane
cores, respectively, showed virtually no affinity for the
coactivator binding groove (Table 1). When compound
14 was designed, it was hoped that the prepositioned
axial stereochemistry of the side chains would force the
leucine mimics into the same direction and into the
coactivator groove, thereby producing an effective CBI.
Molecular modeling shows the Kemp’s triacid core 14
to be somewhat smaller than the coactivator peptide
when the two are overlaid. Thus, it is possible that the
hydrophobic side chains are too close together and thus
unable to span the groove in a productive binding mode.
F igu r e 5. Displacement of TMR-labeled peptide by pyrimi-
dine CBIs.
Ta ble 1. Summary of Binding Affinity Data for Triazene,
Pyrimidine, Trithiane, Cyclohexane and Naphthalene CBIs
Class II CBIs containing a naphthalene core displayed
weak binding to the receptor, as shown by fluorescence
anisotropy (Table 1). Compounds 18a ,b, which contain
functionalities mimicking all four hydrophobic portions
of the coativator, showed weak binding at high concen-
trations. It did not make a difference whether the
flexible side chains contained leucine or isoleucine
mimics. These initial studies lead us to believe that the
naphthalene class of compounds does not show promise
as CBIs. It may be that the naphthalene unit itself is
too planar or too thin to sufficiently fill the deep
hydrophobic groove, or that the class II “inside-out”
design concept is not practical when applied.
CBI
Ki (µM)
CBI
Ki (µM)
CBI
Ki (µM)
1a
1b
1c
1d
1e
590
290
240
290
290
3
7a
7b
12a
12b
650
790
410
29
12c
13
14
18a
18b
49
>1000
>1000
>1000
>1000
32
Biologica l Eva lu a tion . The Class I compounds
containing a triazene core showed weak inhibition of
coactivator peptide binding (Table 1). The most potent
triazene was 7b, which showed a minimal decrease in
anisotropy. Addition of polar functionalities improved
the binding slightly in the case of 7b, but did not
improve binding in any other case. In fact, for the
symmetric triaminotriazenes there appears to be virtu-
ally no improvement in binding as the hydrophobic
functionality of the substituents is varied. The weak
inhibitory activity of this class of compounds might be
due to the inherent rigidity of the triaminotriazene core,
as discussed previously. The partial double bond char-
acter of the C-N bond directly attached to the triazene
core may prevent the compounds from adopting a
conformation optimal for binding to the ER.
La ck of CBI Bin d in g in t h e Liga n d Bin d in g
P ock et. In assaying compounds for inhibition of coac-
tivator binding, it is essential to establish that this
inhibition arises from a direct competition between the
CBI and the coactivator peptide at the coactivator
binding groove, rather than by an indirect competition
at the ligand binding pocket of the LBD. The latter,
indirect process would involve small molecule binding
to the ligand binding pocket in the same manner as an
antagonist, inducing a conformation of helix 12 that
occludes the coactivator binding groove and thus indi-
rectly precludes coactivator binding. To ensure that the
inhibition of coactivator binding we observe is not due
to the indirect binding mechanism, we have measured
the affinity of the CBIs for the ligand binding pocket of
ERR, using a competitive radiometric binding assay
with [3H]estradiol (data not shown). Our most effective
CBI, pyrimidine 12a , has an affinity of 0.01% that of
estradiol (estradiol is set at 100%), and the rest of our
CBIs have an affinity equal to or lower than 12a .
Because the concentration of estradiol used in the CBI
assay is 10 µM, we can estimate that the IC50 for
competition by an indirect mechanism would be 100
mM. Because we measure an IC50 of 32 µM for this
compound, we can conclude that it is competing for
coactivator binding by a direct mechanism.
By contrast, the class I compounds containing a
pyrimidine core showed the most promise as CBIs
(Figure 5). Compound 12a , which contains branched
alkyl substituents that directly mimic L690, L693, and
L694, displaces the TMR-labeled coactivator peptide
with a Ki of 29 µM. By directly comparing the results of
pyrimidine 12a with those of triazene 1a , the conclusion
can be drawn that the increased affinity of 12a results
from the increased torsional flexibility of its two ad-
ditional methylene units.
After we obtained the initial results with the lead
pyrimidine 12a , we prepared additional pyrimidines
that contained varying hydrophobic subtituents, includ-
ing a straight-chain alkyl group and an ethylphenyl
substituent, to probe the capacity of the coativator
binding groove to accommodate different hydrophobic
units. The binding behavior of these additional com-
pounds in the pyrimidine series varied. Pyrimidine 12c
with terminal phenyl substituents inhibited recruitment
of coactivator peptide with a Ki of 49 µM, being
somewhat less effective than the direct leucine mimic
12a . Pyrimidine 12b with straight-chain alkyl substit-
uents inhibited recruitment of the labeled peptide with
a Ki of 32 µM, very similar to the lead pyrimidine 12a .
These results demonstrate that the coactivator binding
Molecu la r Mod elin g. We have also used molecular
modeling to investigate the possible binding orientation
of lead pyrimidine 12a in ERR. Using the FlexiDock
routine in the molecular modeling platform SYBYL, we
docked pyrimidine 12a into the receptor. Initially, the
pyrimidine was overlayed onto the coactivator peptide
by aligning each CH group in the branched side chains
with those in the side chains of L690, L693, and L694.
The result of this docking-minimization study is shown
in Figure 1B, adjacent to a rendering of the ERR-

606 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Rodriguez et al.
coactivator peptide crystal structure for comparison. The
receptor is shown as a continuous electrostatic surface,
where red coloring indicates positive charge and blue
coloring indicates negative charge. The CBI 12a (Figure
1B) is colored by atom type, with hydrogens removed
for clarity, whereas the coactivator peptide (Figure 1A)
is depicted as a red tube, with extraneous side chains
removed for clarity.
effective in blocking the interaction of a coactivator NR
box peptide with the ER, and the best of these were in
the pyrimidine series. The trithiane-based and Kemp
triacid-based CBIs are probably poor structural mimics
of the bound coactivator peptide, the former being too
wide (triequatorial) and the latter too narrow (triaxial).
The class I CBIs in the triazene series are less effective
than the pyrimidines, probably because they are also
more conformationally constrained from reaching a
geometry that effectively mimics the bound coactivator
peptide, the result of amine-substituent-to-ring reso-
nance overlap. Although the 1,3,5-substituent display
of the triazenes and the pyrimidines are essentially
identical, we made a conscious decision to minimize this
torsional constraint by investigating the alkyl-substi-
tuted triazenes and pyrimidines.
This final minimized model shows two alkyl substi-
tutuents of CBI 12a projecting downward into the deep
hydrophobic groove of the receptor, filling the area
normally occupied by L690 and L694. The third alkyl
substituent sits comfortably on the hydrophobic shelf
occupied by I689 in the ERR crystal structure. The
interaction between pyrimidine 12a and the receptor
appears to be entirely hydrophobic in nature, as indi-
cated by the lack of hydrogen bonding or electrostatic
interactions in this model. In fact, one should recall that
attempts to engage the charge clamp interactions in the
CBIs of triazene design did not improve binding affinity.
Significantly, in the model of pyrimidine 12a binding
to ER, the charge clamp residue K362 is repositioned:
In the crystal structure of the ER-DES/GRIP1 NR-2 box
peptide, the lysine residue extends outward from the
receptor and is involved in a polar interaction with the
coactivator peptide; however, because pyrimidine 12a
contains no hydrogen bond or polar partners for K362,
this residue no longer projects outward from the recep-
tor. Instead, it repositions itself in such a manner that
it hydrogen bonds with Q375 and pinches off one end
of the coactivator binding groove. This creates a smaller
hydrophobic pocket that better matches the size of the
smaller pyrimidine when compared to the coactivator
peptide.
It is worth noting that, through our various docking
studies, we have found the coactivator binding groove
to be a fairly flexible surface. In fact, it can expand or
contract to accommodate molecules both larger and
smaller than the coactivator peptide from which we
based our design. The charge clamp residues, which in
part determine the “length” of the hydrophobic groove,
are also able to bend and stretch to reach an optimal
position when docking small molecules, as we found in
our docking of pyrimidine 12a . The implications of this
apparent flexibility of the coactivator binding groove in
terms of the development of small molecule CBIs by
design will only become clear as more and more com-
pounds are explored.
The most extensive work on the interruption of
nuclear receptor coactivator interaction has been done
by McDonnell and co-workers, using short peptides
generated by phage display. Most of the peptides
identified by screening contained a canonical LXXLL
NR box, and some were able to discriminate between
ERR and ERâ, and among ERs complexed with ligands
of different structure, including some antagonists.24-27
These findings highlight the fact that the topographical
features of the coactivator binding groove reflect not just
nuclear receptor subtype but also the detailed structure
of the ligand bound to the receptor, that sequences
flanking the NR box may also play a role in nuclear
receptor subtype selectivity,28 and that interaction sites
persist in the ER complexes with antagonists, despite
the apparent occlusion of the coactivator groove by the
antagonist-induced displacement of helix 12 shown by
X-ray structures.3,4,12,13 In addition, some of these pep-
tides were found to disrupt ER-mediated transcription
induced by estrogen agonists in target cells; certain
peptides even blocked the partial agonism shown in
some cells by the antiestrogen tamoxifen, suggesting
that a blockade of ER signaling at the level of NR-
coactivator interaction might prove effective in overcom-
ing tamoxifen resistance that often develops in the
treatment of breast cancer.18,19
In the area of nonnatural peptide-based CBIs for
nuclear receptors, Guy and co-workers demonstrated
that the binding affinity of an NR box peptide could be
increased by up to 15-fold by introducing chemical
modifications in the residues (e.g., macrolactamization)
that stabilize the R-helical conformation.29 In libraries
of similarly constrained peptides containing nonnatural
R-amino acids in place of the leucines in the LXXLL
motif, this group found that members differing at only
one site of substitution showed a 600-fold preference for
binding to ERR vs ERâ or TRâ.30 This large degree of
selectivity among members of the nuclear receptor
family provides encouragement that similar selectivity
might eventually be achieved with small, nonpeptide-
based CBIs. More recently, Leduc and co-workers31
described other amide and disulfide constrained LXXLL
peptides as CBIs. The latter analogues, in particular,
bound to the ERR with high affinity and good selectivity;
although these analogues were found not to be helical
when free in solution, they were shown by X-ray
analysis to adopt a helical structure when bound to the
ER.
Discu ssion
The development of compounds that can block the
interaction of agonist-liganded nuclear receptors with
coactivator proteins could provide unique pharmacologi-
cal tools for interrupting the signal transduction cascade
of these transcription regulators. In this paper, we have
described an approach to developing small molecule
coactivator binding inhibitors (CBI) by a de novo,
structure-inspired approach, and we have evaluated a
set of candidate CBIs for their activity in blocking the
binding of a model nuclear receptor interaction box (NR
box) peptide. Thus far, of the two design approaches we
have taken (class I, outside-in; class II, inside out) and
the various types of CBIs we have investigated, only
compounds from the class I series were found to be

Inhibitors of ERR Coactivator Binding
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 607
tetramethylsilane or by reference to proton resonances result-
ing from incomplete deuteration of the NMR solvent. Low- and
high-resolution electron impact (EI) and chemical ionization
(CI) mass spectra were performed in the Mass Spectrometry
Lab of the University of Illinois. Elemental analyses were
performed by the Microanalytical Service Laboratory of the
University of Illinois. Melting points were determined on a
Thomas-Hoover melting point apparatus and are uncorrected.
Radiolabeled estradiol ([3H]E2) ([6,7-3H]estra-1,3,5,(10)-triene-
3,17-â-diol), 54 Ci/mmol, was obtained from Amersham Bio-
sciences (Piscataway, NJ ). Isopropyl â-D-thiogalactopyranoside,
imidazole, â-mercaptoethanol, estradiol, and trans-4-hydroxy-
tamoxifen were obtained from Sigma Chemical Co. (St. Louis,
MO). The NR box 2 peptide of SRC-1 (SRC-1-NR box 2 residues
683-696, CLTERHKILHRLLQE) was synthesized as de-
scribed previously.22
Relative binding affinity (RBA) analyses were performed
according to literature methods.45,46 Fluorescence experiments
used a Perkin-Elmer Life Science Wallac Victor2 V 1420
multilabel HTS counter with Wallac 1420 workstation soft-
ware (PerkinElmer, Boston, MA). All data were analyzed using
Prism 3.00 (GraphPad Software, San Diego, CA). ER prepara-
tions used in these experiments were lamb uterine cytosol,
purified full-length human ERR and ERâ (PanVera Inc.), or
purified His6-tagged ERR LBD (304-554).46 Chemical synthe-
ses of all intermediate compounds are available in the Sup-
porting Information.
The nuclear receptor interaction box 2 peptide (NR-2),
corresponding to the NR-2 box sequence of SRC-1, and the
octapeptide for fluorophore labeling (see the text) were syn-
thesized by the University of Illinois Biotechnology Center,
utilizing FMoc solid-phase strategy on a multiple peptide
synthesizer and were purified by C18 reversed phase HPLC.
The octapeptide was labeled on the N-terminus with rhodamine
using the succinimidyl ester. Peptide quality was determined
by analytical HPLC and mass spectroscopy (University of
Illinois Biotechnology Center).
Ch em ica l Syn th esis. Gen er a l P r oced u r e for Tr ia m i-
n otr ia zen es. Isobutylamine (1.78 mL, 17.9 mmol) and diiso-
propylamine (2.35 mL, 17.9 mmol) were dissolved in DMF (10
mL). Cyanuric chloride (1.0 g, 5.4 mmol) was added. The
reaction bubbled vigorously and a precipitate formed which
dissolved after 30 min of heating. The reaction mixture was
refluxed for 16 h. Upon cooling, a precipitate formed which
was filtered. The filtrate was extracted with CH2Cl2, and the
extracts were dried over NaSO4 and concentrated under
reduced pressure.
Inhibition of protein-protein interactions with small
molecules is a relatively young field with a number of
unique challenges.32,33 Often, protein-protein binding
occurs over a relatively large surface area, which makes
it difficult to focus on essential, high-affinity interactions
that might be blocked with low molecular weight organic
molecules. Sometimes, the binding surface between two
proteins is relatively “smooth”, lacking grooves or
pockets that make good sites for small molecule binding.
Despite these challenges, there are examples of small
molecules acting as potent and specific inhibitors of
various protein-protein interactions, including the
helix-groove interactions of the type we have been
investigating. Hamilton and co-workers developed a
terphenyl scaffold to mimic extended regions of an
R-helix, using these helix analogues to inhibit gp41
assembly and viral fusion and to block the binding of
Bcl2 to Bcl-xL, which regulates apoptosis.34-36 These are
longer helix-groove interactions that involve three
turns of the R-helix (vs two turns for the NR coactivator
interactions), and the substituents on the terphenyl core
nicely mimic the i, i + 4, i + 7 sites of a three-turn
R-helix, as do substituents on a related oligoamide
foldamer.37 Using computer-based screening strategy,
Huang and co-workers discovered small molecules that
inhibited Bcl-2 function by competing for its binding to
Bcl-xL.38-41 More recently, Asada and co-workers screened
a library of tryptophan derivatives to find an inhibitor
of the interaction of a helical element of the nuclear
protein ESX with the Sur-2/DRIP130 transcription
factor.42 The best inhibitor, a derivative of pindolol,
termed adamanolol, had a Ki of 8 µM, which is similar
to that of our best pyrimidine 12a .
Our discovery that certain small molecules, particu-
larly the 2,4,6-trialkyl-substituted pyrimidines, are able
to inhibit coactivator binding to the estrogen receptor,
provides a proof-of-principle that effective small mol-
ecule CBIs can be developed. The highest affinity CBI
that we have identified so far, pyrimidine 12a , however,
has a Ki of 29 µM, which is insufficient for it to be
studied effectively in cell culture or in vivo models of
estrogen action (Katzenellenbogen et al., unpublished).
Nevertheless, these pyrimidine CBIs appear to be the
first small molecule inhibitors of nuclear receptor co-
activator binding that have been reported. We are
pursuing further refinements in the design of CBI to
improve their binding affinity.
N,N′,N′′-Tr iisobu tylm ela m in e (1a ). Triazene 1a was
prepared according to the general procedure for the synthesis
of triaminotriazenes. Flash chromatography (75:25 hexane:
EtOAc) afforded 1 as a white waxy residue (405 mg, 43%): 1H
NMR (500 MHz, CDCl3) δ 1.19 (d, 6H, J ) 6.76 Hz, CH3), 1.81
(bs, 1H, CH), 3.17 (bs, 2H, CH2), 4.70 (bs, 1 H, NH); 13C NMR
(125 MHz, CDCl3) δ 20.2, 28.6, 48.2, 116.5; LRMS (EI, 70 eV)
m/z 294 (M+); HRMS (EI) calcd for C15H30N6 294.2532, found
294.2529.
Exp er im en ta l Section
N,N′,N′′-Tr ip r op ylm ela m in e (1b). Triazene 1b was pre-
pared according to the general procedure for the synthesis of
triaminotriazenes. Flash chromatography (1:1 EtOAc:hexane)
provided the product as a yellow oil (368 mg, 27%): 1H NMR
(CDCl3, 400 mHz) δ 0.89 (t, 9H, J ) 7.40 Hz, CH3), 1.51 (sextet,
6H, J ) 6.93 Hz, CH2CH3), 3.27 (bs, 6H, CH2NH), 5.19 (bs,
3H, NH); 13C NMR (CDCl3, 125 MHz) δ 11.3, 22.9, 42.3, 165.9;
LRMS (EI, 70 eV) m/z 252.2 (M+), 237.2 (M+ - CH3), 223.2
(M+ - CH2CH3); HRMS calcd for C12H24N6 252.2062, found
252.2069.
Gen er a l Meth od s. All reagents and solvents were obtained
from Aldrich or Fisher. Tetrahydrofuran, diethyl ether, and
dichloromethane were dried by the solvent delivery system
(SDS) (neutral alumina columns) designed by Meyer.43 Tri-
ethylamine, diisopropylamine, and diethyldiisopropylamine
were distilled from and stored over potassium hydroxide
pellets. Butyllithium was titrated using N-pivaloyl-o-toluidine,
according to a literature method.44 All reactions were per-
formed under a dry (Drierite) nitrogen atmosphere unless
otherwise stated.
Reaction progress was monitored using analytical thin-layer
chromatography (TLC) on 0.25 mm Merck F-254 silica gel
glass plates. Visualization was achieved by either UV light
(254 nm) or potassium permanganate indicator. Flash chro-
matography was performed with Woelm silica gel (0.040-0.063
N,N′,N′′-Tr iben zylm ela m in e (1c).47 Triazene 1c was pre-
pared according to the general procedure for the synthesis of
triaminotriazenes. Recrystallization from EtOH followed by
flash chromatography (75:25 hexane:EtOAc) provided the
product as an off white solid (491 mg, 23%): mp 90-93 °C;
1H NMR (CDCl3, 400 mHz) δ 4.55 (bs, 2H, CH2), 5.25 (bd, 1H,
NH), 7.18-7.36 (m, 5H, ArH); 13C NMR (CDCl3, 125 MHz) δ
44.6, 127.1, 127.6, 128.5, 139.2, 160.9; LRMS (EI, 70 eV) m/ z
1
mm) packing. H and 13C spectra were recorded on a U400 or
U500 Varian FT-NMR spectrometer. Chemical shifts (δ) are
reported in parts per million (ppm) downfield from internal

608 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Rodriguez et al.
396.2 (M+), 305.2 (M+ - C7H7); HRMS calcd for C24H24N6
35.20, 43.85, 43.96, 45.31, 45.37, 47.18, 47.29, 47.47, 47.69,
47.92, 48.06, 48.08, 48.15, 52.18, 52.25, 155.82, 156.35, 156.98,
161.75, 161.9, 162.29, 169.35, 173.75, 173.95; LRMS (EI, 70
eV) m/z 336.2 (M-1), 320.0 (M+ - OH), 308.2 (M+ - C2H5);
HRMS calcd for C17H30N5O2 336.2399, found 336.2397.
396.2062, found 396.3062.
N,N′,N′′-Tr is(cycloh exylm eth yl)m ela m in e (1d ). Triaz-
ene 1d was prepared according to the general procedure for
the synthesis of triaminotriazenes. Flash chromatography (75:
25 hexane:EtOAc) provided the product as an off-white oil (53
mg, 2.3%): 1H NMR (CDCl3, 400 MHz) δ 0.94 (m, 2H, CH2),
1.19 (m, 3H, aliphH), 1.50 (m, 1H, CH), 1.71 (m, 5H, aliphH),
3.15 (m, 2H, NHCH2), 4.92 (bd, 1H, NH); 13C NMR (CDCl3,
125 MHz) δ 25.9, 26.5, 30.9, 38.2, 46.9, 166.0; LRMS (EI, 70
eV) m/z 414.4 (M+), 346.4 (M+ - C6H10), 318.3 (M+ - C7H12);
HRMS calcd for C24H42N6 414.3471, found 414.3467.
N,N′,N′′-Tr is(1-n a p h th ylm eth yl)m ela m in e (1e). Triaz-
ene 1e was prepared according to the general procedure for
the synthesis of triaminotriazenes. Flash chromatography (1:1
EtOAc:hexane) provided the product as a solid which was
recrystallized from EtOH (256 mg, 49%): mp 115-117 °C; 1H
NMR (CDCl3, 500 MHz) δ 4.95 (d, 2H, J ) 5.62 Hz, CH2), 5.74
(bs, 1H, NH), 7.50 (m, 3H, ArH), 7.84 (dd, 1H, J ) 7.32, 2.20
Hz, ArH), 7.89 (m, 1H, ArH), 7.89 (m, 1H, ArH); 13C NMR
(CDCl3, 125 MHz) δ 40.2, 123.3, 124.9, 125.3, 126.1, 126.7,
126.8, 128.7, 131.2, 132.8, 133.8, 160.7.
Gen er a l P r oced u r e for 2-Th ioxo-2,3-1H-p yr im id in -4-
on es. Thiourea (1 g, 13.0 mmol) was dissolved in 1.6 mL of
water at 70 °C. â-Keto ester (3.4 g, 19.7 mmol) was added
followed by K2CO3 (2.73 g, 19.37 mmol), upon which the
solution became cloudy. It was heated to 105 °C, where the
precipitate dissolved. The reaction was heated at this tem-
perature open to air for 1 h, boiling off any remaining ethanol
and leaving a light yellow solid. Heat was removed and the
reaction allowed to cool to room temperature over a period of
2 h. Water (6.6 mL) was added followed by 6 mL of concen-
trated HCl, causing the reaction to bubble vigorously. The
white precipitate was collected and washed with water and 1
M HCl.
Gen er a l P r oced u r e for Con ver sion of Th iou r a cil to
Ur a cil Der iva tives. Chloroacetic acid (1.34 g, 0.0142 mol)
was dissolved in H2O (5 mL). Crude 5 (1.4 g, 0.0071mol) was
dissolved in H2O/THF and added to the stirring solution. The
reaction mixture was refluxed for 6 h. Concentrated HCl (0.36
mL) was added carefully, and the reaction returned to reflux
for 12 h. The reaction mixture was extracted with ether/THF,
washed with H2O, dried over NaSO4, and concentrated at
reduced pressure.
N,N′-Diisob u t yl-6-(3-m et h ylb u t yl)[1,3,5]t r ia zen e-2,4-
d ia m in e (3). Crude monoaminodichlorotriazene 2 (0.47 mL,
2.1 mmol) was dissolved in THF (10 mL). Isobutylamine (0.52
mL, 5.3 mmol) and diisopropylamine (0.7 mL, 5.3 mmol) were
added. The reaction mixture was refluxed for 16 h. After
cooling, the reaction mixture was poured over water. The
residue was extracted with EtOAc, and the extracts were
washed with H2O and brine, dried over NaSO4, and concen-
trated at reduced pressure. Flash chromatography (9:1 hexane:
EtOAc) afforded the product as a light yellow solid (0.34 g,
Gen er a l P r oced u r e for Con ver sion of Ur a cil Der iva -
tives to Dich lor op yr im id in es. n-Butyluracil 10a (1 g, 5.9
mmol) was suspended in 9 mL of POCl3 and heated to reflux.
After 20 min of heating, the reaction became homogeneous.
After 3 h, the reaction was cooled to room temperature and
concentrated at reduced pressure. The residue was diluted
with 30 mL of ice water. After 1 h, the residue was partitioned
between EtOAc and H2O, and the extracts were dried over
NaSO4 and concentrated at reduced pressure.
1
55%): mp 89-90 °C; H NMR (CDCl3, 500 MHz) δ 0.92 (d, 6
H, J ) 6.3 Hz, CH(CH3)2), 0.94 (d, 12 H, J ) 6.92 Hz, NHCH2-
CH(CH3)2), 1.59 (m, 3 H, CH), 1.84 (bs, 2 H, CHCH2CH2), 2.40
(bs, 2 H, CHCH2CH2), 3.32 (s, 4 H, CH2NH), 5.01 (bs, 1 H,
NH), 5.19 (bs, 1 H, NH); 13C NMR (CDCl3, 125 MHz) δ 20.2,
22.4, 28.3, 28.5, 37.0, 48.1, 133, 166.1; LRMS (EI, 70 eV) m/z
294.2 (M+); HRMS calcd for C16H31N5 293.2579, found 293.2567.
Gen er a l P r oced u r e for Su bstitu tion of Dich lor op y-
d r im id in es. Pyrimidine 11a (152 mg, 0.685 mmol) was
dissolved in diisopropylamine (0.90 mL, 6.85 mmol) and
isobutylamine (0.68 mL, 6.85 mmol). The reaction was refluxed
for 12 h. After cooling it was extracted with CH2Cl2, and the
extracts were dried over NaSO4 and concentrated under
vacuum.
N2,N4-Diisob u t yl-6-(3-m et h ylb u t yl)p yr im id in e-2,4-d i-
a m in e (12a ). Pyrimidine 12a was synthesized according to
the general procedure for the substitution of dichloropydrim-
idines. Flash chromatography afforded the product (175 mg,
60%) as a yellow solid: 1H NMR (CDCl3, 500 MHz) δ 0.89 (d,
6 H, J ) 6.68 Hz, CH3), 0.93 (d, 6 H, J ) 6.58 Hz, NHCH2-
CH(CH3)2), 0.93 (d, 6 H, J ) 6.43 Hz, NHCH2CH(CH3)2), 1.49
(q, 2 H, J ) 7.66 Hz, CHCH2CH2), 1.58 (nonet, 1 H, J ) 6.49,
CH), 1.83 (nonet, 2 H, J ) 6.75 Hz, CHCH2NH), 2.38 (t, 2 H,
J ) 8.06 Hz, CHCH2CH2), 3.05 (bs, 1 H, NH), 3.15 (t, 2 H, J
) 6.23 Hz, CHCH2NH), 4.78 (bs, 1 H, NH), 5.50 (s, 1 H, Cd
CH); 13C NMR (CDCl3, 125 MHz) δ 20.2/20.3, 22.4, 27.9, 28.4/
28.5, 35.5, 37.8, 48.9, 91.0, 162.1, 163.8; LRMS (EI, 70 eV) m/z
293.3 (M+) 249.2 (M+ - C3H8); HRMS calcd for C17H33N4
293.2705, found 293.2704.
{Isob u t yl[4-isob u t yla m in o-6-(3-m e t h ylb u t yl)[1,3,5]-
tr ia zin -2-yl]a m in o}a cetic Acid (7a ). Triazene 6a (57 mg,
0.15 mmol) was dissolved in 3 mL of methanol at room
temperature. Then 0.25 mL of 4 M NaOH was added dropwise.
After stirring for 18 h, the reaction was partitioned between
EtOAc and brine, and the extracts were dried over NaSO4 and
concentrated to a white solid (43.6 mg, 83%). No further
purification was necessary: mp 175-181 °C; 1H NMR (CDCl3,
400 MHz) major rotamer δ 0.94 (m, 18H, 6 × CH3), 1.60 (m,
3H, CH2CH), 1.92 (m, 2H, 2 × CH), 2.66 (t, 2H, J ) 7.32 Hz,
ArCH2), 3.03 (t, 2H, J ) 6.34 Hz, NHCH2CH), 3.56 (d, 2H, J
) 7.32 Hz, NCH2CH), 4.18 (s, 2H, CH2CO); minor rotamer δ
0.85 (m, 18H, 6 × CH3), 1.60 (m, 3H, CH2CH), 1.92 (m, 2H, 2
× CH), 2.57 (t, 2H, J ) 7.32 Hz, ArCH2), 3.21 (t, 2H, J ) 6.22
Hz, NHCH2CH), 3.44 (d, 2H, J ) 7.32 Hz, NCH2CH), 4.33 (s,
2H, CH2CO); 13C NMR (CDCl3, 125 MHz) both rotamers δ 19.9,
20.0, 20.1, 20.2, 22.0, 22.1, 27.1, 27.38, 27.45, 27.47, 27.95,
28.04, 32.5, 32.6, 34.6, 34.7, 48.0, 48.2, 50.5, 50.7, 56.5, 56.7,
155.1, 155.5, 162.7, 162.9, 168.9, 169.1, 170.7, 171.2; LRMS
-
(EI, 70 eV) m/z 351.3 (M+), 336.3 (M+ - CH3), 308.3 (M+
C3H7); HRMS calcd for C18H33N5O2 351.2634, found 351.2639.
6-Bu tyl-2,4-(p r op yla m in o)p yr im id in e (12b). Pyrimidine
12b was synthesized according to the general procedure for
the substitution of dichloropydrimidines. Flash chromatogra-
phy (1:1 EtOAc:hexane) provided the product as an oil (12 mg,
9.4%) along with both monoaminated products accounting for
the remainder of the mass balance: 1H NMR (CDCl3, 400
MHz) δ 0.91 (t, 3H, J ) 7.32 Hz, CH3), 0.95 (t, 3H, J ) 6.84
Hz, CH3), 0.96 (t, 3H, J ) 6.84 Hz, CH3), 1.36 (sextet, 2H, J )
7.42 Hz, CH2CH3), 1.59 (m, 6H, 3 × CH2), 2.39 (dd, 2H, J )
8.06 Hz, 7.57, CH2Ar), 3.20 (q, 2H, J ) 6.43 Hz, CH2NH), 4.63
(bs, 1H, NH), 4.86 (bs, 1H, NH), 5.53 (s, 1H, CH); 13C NMR
(CDCl3, 125 MHz) δ 11.46 (2), 11.51, 13.9, 22.5, 22.7, 23.0,
109.7, 162.2, 163.8 (2); LRMS (EI, 70 eV) m/z 250.2 (M+), 235.2
(M+ - CH3), 221.2 (M+ - C2H3); HRMS calcd for C14H26N4
250.2157, found 250.2153.
{[4-Isobu tyla m in o-6-(3-m eth ylbu tyl)[1,3,5]tr ia zin -2-yl]-
isop r op yla m in o}a cetic Acid (7b). Triazene 6b (130 mg, 0.36
mmol) was dissolved in 3 mL of methanol. Then 0.44 mL of 4
M NaOH was added slowly. After stirring for 18 h, the reaction
was partitioned between EtOAc and brine, and the extracts
were dried over NaSO4 and concentrated to a white residue
(59.3 mg, 49%). No further purification was necessary: 1H
NMR (CDCl3, 400 MHz) rotameric mixture δ 0.86 (m, 12H,
CH3), 1.20 (m, 6H, (CH3)2CHN), 1.57 (m, 3H, CH2CH), 1.85
(m, 1H, CH), 2.53 (m, 2H, CH2), 3.15 (m, 2H, CH2CO2H), 4.85-
5.20 (m, 1H, NCH), 9.09 (bs, 1H, OH); 13C NMR (CDCl3, 125
MHz) rotameric mixture δ 19.6, 19.7, 20.0, 20.05, 20.06, 20.09,
20.16, 20.17, 22.08, 22.13, 27.33, 27.39, 27.6, 27.64, 27.98,
28.05, 28.14, 28.16, 32.64, 32.8, 33.05, 34.82, 34.87, 35.05,

Inhibitors of ERR Coactivator Binding
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 609
6-(2-P h en yleth yl)-2,4-(ben zyla m in o)p yr im id in e (12c).
Pyrimidine 12c was synthesized according to the general
procedure for the substitution of dichloropydrimidines. Flash
chromatography (1:1 EtOAc:hexane) provided the product (16
mg, 51%) as a yellow oil: 1H NMR (CDCl3, 400 MHz) δ 2.70
(ABq, 2H, νa ) 2.73, νb ) 2.67, J ab ) 5.60 Hz, CH2), 2.83 (ABq,
2H, νa ) 2.98, νb ) 2.71, J ab ) 7.15 Hz, CH2), 4.48 (bd, 2H, J
) 4.39 Hz, CH2NH), 4.58 (d, 2H, J ) 5.86 Hz, CH2NH), 5.55
(s, 1H, CH), 7.25 (m, 15H, ArH); 13C NMR (CDCl3, 125 MHz)
δ 34.1, 44.9, 45.1, 93.7, 126.2, 127.0, 127.5, 127.55, 127.59,
128.4, 128.7, 138.0, 139.1, 140.6, 163.3; LRMS (EI, 70 eV) 394.2
(M+), 317.2 (M+ - C6H5), 303.2 (M+ - C7H7); HRMS calcd for
2,7-Bis[2-((3-m et h ylp en t a n oyl)a m in o)et h yl]n a p h t h a -
len e (18b). Naphthalene 17 (30 mg, 0.14 mmol) was combined
with 2-methylbutyl acid choride (0.042 mL, 0.31 mmol) and
triethylamine (0.043 mL, 0.31 mmol) in 2 mL of CH2Cl2. The
reaction was refluxed for 6 h. After cooling, the reaction was
partitioned between EtOAc and water/brine, and the extracts
were dried over NaSO
4 and concentrated under vacuum. Flash
chromatography (95:5 CH2Cl2:MeOH) provided the product
(41.8 mg, 73%) as a white powder which was recrystallized
from hexane:EtOH. Mp 179-181 °C. Material for binding
assay was further purified by reversed phase HPLC (hexane:
ethanol 85:15): 1H NMR (CDCl3, 400 MHz) δ 0.87 (m, 12H,
CH3), 1.15 (m, 2H, CH3CH2), 1.33 (m, 2H, CH3CH2), 2.05 (m,
6H, CHCH2CO), 2.97 (t, 4H, J ) 6.96 Hz ArCH2), 3.61 (q, 4H,
J ) 6.47 Hz, CH2NH), 5.46 (bs, 2H, NH), 7.31 (dd, 2H, J )
8.30 Hz, 1.46, H3,6), 7.57 (bs, 2H, H1,8), 7.77 (d, 2H, J ) 8.30
Hz, H4,5); 13C NMR (CDCl3, 125 MHz) δ 11.3, 19.1, 29.3, 32.3,
35.9, 40.4, 44.2, 126.7, 126.8, 128.1, 130.1, 133.7, 136.8, 172.7;
LRMS (EI, 70 eV) m/z 410.4 (M+), 367.4 (M+ - C3H7), 341.4
(M+ - C5H9); HRMS calcd for C26H38N2O2 410.2933, found
410.2927.
C
26H26N4 394.2157, found 394.2154.
2,4,6-Tr is(3-m et h ylb u t yl)[1,3,5]t r it h ia n e (13). 1,3,5-
Trithiane (250 mg, 1.81 mmol) was suspended in THF (10 mL)
and cooled to -35 °C. n-Butyllithium (1.25 mL, 1.99 mmol)
was added dropwise. The reaction was allowed to warm to
room temperature and stir for 45 min. The reaction was then
cooled to -20 °C and 3-methylbutyl bromide (0.32 mL, 2.71
mmol) was added dropwise. The reaction was warmed to room
temperature and stirred for 1.5 h. The procedure was repeated
two additional times, allowing the reaction to stir at room
temperature for 12 h after the final addition of the bromide.
The reaction was quenched with H2O (5 mL) and extracted
with chloroform, and the extracts were dried over NaSO4 and
concentrated at reduced pressure. Flash chromatography (98:2
hexane:EtOAc) afforded a light yellow solid which was recrys-
tallized from 2-propanol to give white crystals (340 mg, 54%
F lu or escen ce An isotr op y Assa ys. Purified ERR ligand
binding domain and tetramethylrhodamine-labeled NR box
peptide (*ILRKLLQE) were used for the fluorescence aniso-
tropy assays. The His6-tagged ERR-LBD (304-554) was ex-
pressed from a pET-15b vector in BL21(DE3)pLysS Escheri-
chia coli and purified as described previously.46 A stock
solution (9 µL) of ERR-LBD (444 nM), estradiol (22 µM), and
ovalbumin (0.3 mg/mL) in Tris-glycerol buffer (50 mM Tris,
10% glycerol, pH 8.0) was placed in separate wells of a black
96-well Molecular Devices HE high efficiency microplate
(Molecular Devices, Inc., Sunnyvale, CA) and incubated at
room temperature for 1 h in the dark. In a second 96-well Nunc
polypropylene plate (Nalge Nunc International, Rochester,
NY), a solution of the coactivator binding inhibitor (556 µM)
was serially diluted in a 1:10 fashion into ovalbumin (0.3 mg/
mL) in Tris-glycerol (pH 8.0) buffer. Each concentration of CBI
(9 µL) or vehicle (buffer) was transferred into the 96-well
Molecular Devices plate, and the background fluorescence
signal was measured with a 544/15 nm excitation and 590/10
nm emission filter pair to correct for any autofluorescence
artifacts from ligand or CBI in our experiments. A 2 µL
solution of the TMR-peptide (200 nM) and ovalbumin (0.3 mg/
mL) in Tris-glycerol (pH 8.0) buffer was added to each well of
the 96-well Molecular Devices plate. The final concentrations
of the reagents were as follows: ERR (200 nM), estradiol (10
µM), CBI (0-250 µM), TMR-peptide (20 nM). The plate was
incubated at room temperature for 1.5 h in the dark and the
fluorescence anisotropy measured with a 544/15 nm excitation
and 590/10 nm emission filter pair. Correction was made for
any change in fluorescence intensity of TMR-peptide between
its bound vs free state, according to the method recommended
by PanVera (PanVera Fluorescence Polarization Applications
Guide; PanVera Corp., Madison, WI).
1
yield): mp 60-62 °C; H NMR (CDCl3, 500 MHz) δ 0.89 (d,
6H, J ) 6.51 Hz, CH3), 1.46 (m, 2H, (CH3)2CHCH2), 1.56
(nonet, 1H, J ) 6.54 Hz, (CH3)2CH), 1.87 (m, 2H, CH2CHS),
4.10 (t, 1H, J ) 6.45 Hz, CHS); 13C NMR (CDCl3, 125 MHz) δ
22.3, 27.9, 33.4, 35.3, 54.1; LRMS (EI, 70 eV) m/z 348.1 (M+);
HRMS calcd for C16H36S3 348.1979, found 348.1984.
1,3,5-Tr im et h ylcycloh exa n e-1,3,5-t r ica r b oxylic Acid
Tr is(isobu tyl a m id e) (14). Kemp’s triacid (200 mg, 0.77
mmol), isobutylamine (0.39 mL, 3.87 mmol), and HOBt (522
mg, 3.87 mmol) were added to 10 mL of DMF and cooled to 0
°C. Triethylamine (1.08 mL) was added, and the reaction
stirred for 5 min. EDC (742 mg, 3.87 mmol) was added and
the reaction allowed to warm to room temperature. The
reaction was refluxed for 12 h after which the DMF was
distilled off under house vacuum. The residue was taken up
in CH2Cl2 and washed with brine and NaHCO3, and the
extracts were dried over NaSO4 and concentrated under
vacuum. Flash chromatography (75:25 hexane:EtOAc) pro-
vided the product (289 mg, 88%) as a yellow oil: 1H NMR
(CDCl3, 400 MHz) δ 0.88 (d, 18H, J ) 6.59 Hz, (CH3)2CH),
1.05 (d, 3H, J ) 15.63 Hz, CH(ring)), 1.22 (s, 9H, CH3), 1.75
(septuplet, 3H, CH), 2.90 (m, 9H, CH(ring) and CH2NH), 7.51
(bt, 3H, J ) 5.37 Hz, NH); 13C NMR (CDCl3, 125 MHz) δ 20.1,
25.9, 27 LRMS (EI, 70 eV) m/z 252.2 (M+).1, 28.4, 31.9, 40.2,
42.1, 43.6, 44.8, 46.4, 47.0, 173.4, 176.8; LRMS (EI, 70 eV) m/z
423.1 (M+), 351.3 (M+ - C4H10N); HRMS calcd for C24H45N3O
423.3461, found 423.3458.
Molecu la r Mod elin g. The X-ray crystal structure of the
estrogen receptor bound to diethylstilbestrol and the GRIP1
peptide (ERD in the protein data bank) was used for modeling
purposes. The starting conformations for all ligands were
generated using the MMFF94 force field in SYBYL, by
minimizing to a gradient of 0.05 kcal/mol. The minimized CBI
was then prepositioned in the coactivator binding groove by
overlaying it onto the GRIP1 peptide using a least-squares
multifitting of select atoms with the ligand. Once positioned,
GRIP1 was deleted. Rotatable bonds in the CBI were chosen
and the CBI docked into the receptor using the FlexiDock
routine within the BioPolymer module of SYBYL beginning
on a random seed, allowing 25 000 interations. The lowest
energy solution was then extracted from the receptor and the
docking process repeated a second time, this time allowing both
the ligand and side chains of the receptor within 5 Å of the
ligand to have rotatable bonds. The lowest energy solution was
then put through a three-step minimization protocol as previ-
ously described.48
2,7-Bis[2-(4-m et h ylp en t a n oyla m in o)et h yl]n a p h t h a -
len e (18a ). Naphthalene 17 (30 mg, 0.14 mmol) was combined
with 3-methylbutyl acid choride (0.043 mL. 0.31 mmol) and
triethylamine (0.043 mL, 0.31 mmol) in 2 mL of CH2Cl2. The
reaction was refluxed for 4 h. After cooling, the reaction was
partitioned between EtOAc and water/brine, and the extracts
were dried over NaSO4 and concentrated under vacuum. The
crude solid was recrystallized from hexane/CH2Cl2 (25 mg,
45%). Mp 179-180 °C. Material for binding assay was further
purified by reversed phase HPLC (hexane:ethanol 85:15): 1H
NMR (CDCl3, 400 MHz) δ 0.84 (d, 12H, J ) 6.10 Hz, CH3),
1.49 (m, 6H, CHCH2), 2.10 (t, 4H, J ) 7.69 Hz, CH2CO), 2.96
(t, 4H, J ) 6.84 Hz, CH2Ar), 3.59 (q, 4H, J ) 6.59 Hz, CH2-
NH), 5.56 (bs, 2H, NH), 7.30 (d, 2H, J ) 8.06 Hz, H3,6), 7.56
(s, 2H, H1,8), 7.76 (d, 2H, H4,5); 13C NMR (CDCl3, 100 MHz) δ
22.3, 27.7, 34.6, 35.9, 40.4, 126.7, 126.8, 128.0, 131.0, 133.6,
136.8, 173.3; LRMS (EI. 70 eV) m/z 410.4 (M+); HRMS calcd
for C26H38N2O2 410.2933, found 410.2936.

610 J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3
Rodriguez et al.
(19) Connor, C. E.; Norris, J . D.; Broadwater, G.; Willson, T. M.;
Gottardis, M. M.; et al. Circumventing tamoxifen resistance in
breast cancers using antiestrogens that induce unique confor-
mational changes in the estrogen receptor. Cancer Res 2001, 61,
2917-2922.
(20) Pope, A. J .; Haupt, U. M.; Moore, K. J . Homogeneous fluores-
cence readouts for miniaturized high-throughput screening:
Theory and practice. Drug Discov. Today 1999, 4, 350-362.
(21) J ameson, D. M.; Seifried, S. E. Quantification of protein-protein
interactions using fluorescence polarization. Methods 1999, 19,
222-233.
(22) Gee, A. C.; Carlson, K. E.; Martini, P. G. V.; Katzenellenbogen,
B. S.; Katzenellenbogen, J . A. Coactivator peptides have a
differential stabilizing effect on the binding of estrogens and
antiestrogens with the estrogen receptor. Mol. Endocrinol. 1999,
13, 1912-1923.
(23) Tamrazi, A.; Carlson, K. E.; Daniels, J . R.; Hurth, K. M.;
Katzenellenbogen, J . A. Estrogen receptor dimerization: Ligand
binding regulates dimer affinity and dimer dissociation rate.
Mol. Endocrinol. 2002, 16, 2706-2719.
(24) Paige, L. A.; Christensen, D. J .; Gron, H.; Norris, J . D.; Gottlin,
E. B.; et al. Estrogen receptor (ER) modulators each induce
distinct conformational changes in ER alpha and ER beta. Proc.
Natl. Acad. Sci. U.S.A. 1999, 96, 3999-4004.
(25) Chang, C.; Norris, J . D.; Gron, H.; Paige, L. A.; Hamilton, P. T.;
et al. Dissection of the LXXLL nuclear receptor-coactivator
interaction motif using combinatorial peptide libraries: Discov-
ery of peptide antagonists of estrogen receptors alpha and beta.
Mol. Cell Biol. 1999, 19, 8226-8239.
(26) Hall, J . M.; Chang, C. Y.; McDonnell, D. P. Development of
peptide antagonists that target estrogen receptor beta-coacti-
vator interactions. Mol Endocrinol 2000, 14, 2010-2023.
(27) Huang, H.-J .; Norris, J . D.; McDonnell, D. P. Identification of a
negative regulatory surface within estrogen receptor alpha
provides evidence in support of a role for corepressors in
regulating cellular responses to agonists and antagonists. Mol.
Endocrinol. 2002, 16, 1778-1792.
(28) McInerney, E. M.; Rose, D. W.; Flynn, S. E.; Westin, S.; Mullen,
T. M.; et al. Determinants of coactivator LXXLL motif specificity
in nuclear receptor transcriptional activation. Genes Dev. 1998,
12, 3357-3368.
Ack n ow led gm en t. We are grateful for support of
this research through a grant from the U.S. Army
Breast Cancer Research Program (DAMD17-00-1-0293).
NMR spectra were obtained in the Varian Oxford
Instrument Center for Excellence in NMR Laboratory.
Funding for the instrumentation was provided in part
by the W.M. Keck Foundation, the National Institutes
of Health (PHS 1 S10 RR104444-01), and the National
Science Foundation (NSF CHE 96-10502). Mass spectra
were obtained on instruments supported by grants form
the National Institute of General Medical Sciences (GM
27029), the National Instituted of Health (RR 01575),
and the National Science Foundation (PCM 8121494).
Su p p or tin g In for m a tion Ava ila ble: A description of the
synthesis of compounds 2, 4, 6a -b, 9a -c, 10a -c, 11a -c, 15,
16, and 17. This information is available free of charge via
the Internet at http://pubs.acs.org.
Refer en ces
(1) McKenna, N. J .; O’Malley, B. W. Minireview: Nuclear receptor
coactivatorssan update. Endocrinology 2002, 143, 2461-2465.
(2) McKenna, N. J .; O’Malley, B. W. Combinatorial control of gene
expression by nuclear receptors and coregulators. Cell 2002, 108,
465-474.
(3) Shiau, A. K.; Barstad, D.; Loria, P. M.; Cheng, L.; Kushner, P.
J .; et al. The structural basis of estrogen receptor/coactivator
recognition and the antagonism of this interaction by tamoxifen.
Cell 1998, 95, 927-937.
(4) Shiau, A. K.; Barstad, D.; Radek, J . T.; Meyers, M. J .; Nettles,
K. W.; et al. Structural characterization of a subtype-selective
ligand reveals a novel mode of estrogen receptor antagonism.
Nat. Struct. Biol. 2002, 9, 359-364.
(5) Darimont, B. D.; Wagner, R. L.; Apriletti, J . W.; Stallcup, M.
R.; Kushner, P. J .; et al. Structure and specificity of nuclear
receptor-coactivator interactions. Genes Dev. 1998, 12, 3343-
3356.
(6) Gampe, R. T.; Montana, V. G.; Lambert, M. H.; Miller, A. B.;
Bledsoe, R. K.; et al. Asymmetry in the PPARgamma/RXRalpha
crystal structure reveals the molecular basis of heterodimeriza-
tion among nuclear receptors. Mol. Cell 2000, 5, 545-555.
(7) Egea, P. F.; Mitschler, A.; Moras, D. Molecular recognition of
agonist ligands by RXRs. Mol. Endo. 2002, 16, 987-997.
(8) Warnmark, A.; Treuter, E.; Gustafsson, J .; Hubbard, R. E.;
Brzozowski, A. M.; et al. Interaction of transcriptional interme-
diary factor 2 nuclear receptor box peptides with the coactivator
binding site of estrogen receptor alpha. J . Biol. Chem. 2002, 277,
21862-21868.
(29) Geistlinger, T. R.; Guy, R. K. An inhibitor of the interaction of
thyroid hormone receptor beta and glucocorticoid interacting
protein 1. J . Am. Chem. Soc. 2001, 123, 1525-1526.
(30) Geistlinger, T. R.; Guy, R. K. Novel selective inhibitors of the
interaction of individual nuclear hormone receptors with
a
mutually shared steroid receptor coactivator 2. J . Am. Chem.
Soc. 2003, 125, 6852-6853.
(31) Leduc, A. M.; Trent, J . O.; Wittliff, J . L.; Bramlett, K. S.; Briggs,
S. L.; et al. Helix-stabilized cyclic peptides as selective inhibitors
of steroid receptor-coactivator interactions. Proc. Natl. Acad.
Sci. U.S.A. 2003, 100, 11273-11278.
(32) Toogood, P. T. Inhibition of protein-protein association by small
molecules: Approaches and progress. J . Med. Chem. 2002, 45,
1543-1558.
(9) Xu, H. E.; Stanley, T. B.; Montana, V. G.; Lambert, M. H.;
Shearer, B. G.; et al. Structural basis for antagonist-mediated
recruitment of nuclear co-repressors by PPAR alpha. Nature
2002, 415, 813-817.
(10) Mi, L. Z.; Devarakonda, S.; Harp, J . M.; Han, Q.; Pellicciari, R.;
et al. Structural Basis for Bile Acid Binding and Activation of
the Nuclear Receptor FXR. Mol. Cell 2003, 11, 1093-1100.
(11) Downes, M.; Verdecia, M. A.; Roecker, A. J .; Hughes, R.;
Hogenesch, J . B.; et al. A Chemical, Genetic, and Structural
Analysis of the Nuclear Bile Acid Receptor FXR. Mol. Cell 2003,
11, 1079-1092.
(33) Peczuh, M. W.; Hamilton, A. D. Peptide and protein recognition
by designed molecules. Chem. Rev. 2000, 100, 2479-2494.
(34) Ernst, J . T.; Kutzki, O.; Debnath, A. K.; J iang, S.; Lu, H.; et al.
Design of a protein surface antagonist based on alpha-helix
mimicry: Inhibition of gp41 assembly and viral fusion. Angew.
Chem. Int. Ed. Engl. 2002, 41, 278-281.
(35) Kutzki, O.; Park, H. S.; Ernst, J . T.; Orner, B. P.; Yin, H.; et al.
Development of a potent Bcl-x(L) antagonist based on alpha-
helix mimicry. J . Am. Chem. Soc. 2002, 124, 11838-11839.
(36) Orner, B. P.; Ernst, J . T.; Hamilton, A. D. Toward proteomi-
metics: Terphenyl derivatives as structural and functional
mimics of extended regions of an alpha-helix. J . Am. Chem. Soc.
2001, 123, 5382-5383.
(12) Brzozowski, A. M.; Pike, A. C.; Dauter, Z.; Hubbard, R. E.; Bonn,
T.; et al. Molecular basis of agonism and antagonism in the
oestrogen receptor. Nature 1997, 389, 753-758.
(13) Pike, A. C.; Brzozowski, A. M.; Hubbard, R. E.; Bonn, T.;
Thorsell, A. G.; et al. Structure of the ligand-binding domain of
oestrogen receptor beta in the presence of a partial agonist and
a full antagonist. EMBO J . 1999, 18, 4608-4618.
(14) Pike, A. C.; Brzozowski, A. M.; Walton, J .; Hubbard, R. E.;
Thorsell, A. G.; et al. Structural insights into the mode of action
of a pure antiestrogen. Structure (Camb) 2001, 9, 145-153.
(15) Shang, Y.; Brown, M. Molecular determinants for the tissue
specificity of SERMs. Science 2002, 295, 2465-2468.
(16) Schiff, R.; Massarweh, S.; Shou, J .; Osborne, C. K. Breast cancer
endocrine resistance: How growth factor signaling and estrogen
receptor coregulators modulate response. Clin. Cancer Res. 2003,
9, 447S-454S.
(17) Osborne, C. K.; Bardou, V.; Hopp, T. A.; Chamness, G. C.;
Hilsenbeck, S. G.; et al. Role of the estrogen receptor coactivator
AIB1 (SRC-3) and HER-2/neu in tamoxifen resistance in breast
cancer. J . Natl. Cancer Inst. 2003, 95, 353-361.
(18) Norris, J . D.; Paige, L. A.; Christensen, D. J .; Chang, C. Y.;
Huacani, M. R.; et al. Peptide antagonists of the human estrogen
receptor. Science 1999, 285, 744-746.
(37) Ernst, J . T.; Becerril, J .; Park, H. S.; Yin, H.; Hamilton, A. D.
Design and application of an alpha-helix-mimetic scaffold based
on an oligoamide-foldamer strategy: Antagonism of the Bak
BH3/Bcl-xL complex. Angew. Chem. Int. Ed. Engl. 2003, 42,
535-539.
(38) Wang, J . L.; Liu, D.; Zhang, Z. J .; Shan, S.; Han, X.; et al.
Structure-based discovery of an organic compound that binds
Bcl-2 protein and induces apoptosis of tumor cells. Proc. Natl.
Acad. Sci. U.S.A. 2000, 97, 7124-7129.
(39) Liu, D.; Huang, Z. Synthetic peptides and nonpeptidic molecules
as probes of structure and function of Bcl-2 family proteins and
modulators of apoptosis. Apoptosis 2001, 6, 453-462.
(40) Huang, Z. The chemical biology of apoptosis. Exploring protein-
protein interactions and the life and death of cells with small
molecules. Chem. Biol. 2002, 9, 1059-1072.

Inhibitors of ERR Coactivator Binding
J ournal of Medicinal Chemistry, 2004, Vol. 47, No. 3 611
(41) Huang, Z. Structural chemistry and therapeutic intervention of
protein-protein interactions in immune response, human im-
munodeficiency virus entry, and apoptosis. Pharmacol. Ther.
2000, 86, 201-215.
(46) Carlson, K. E.; Choi, I.; Gee, A.; Katzenellenbogen, B. S.;
Katzenellenbogen, J . A. Altered ligand binding properties and
enhanced stability of a constitutively active estrogen receptor:
Evidence that an open pocket conformation is required for ligand
interaction. Biochemistry 1997, 36, 14897-14905.
(47) Shigehiro, O.; Tsumomu, S. N,N′,N′′-Tribenzylmelamine (Patent
written in J apanese). J pn. Tokkyo Koho 1972, Patent No.
J P47021109; Kind B4; Date 19720614; Application No. J P1969-
45379; Application Date 19690611.
(48) Stauffer, S. R.; Coletta, C. J .; Tedesco, R.; Sun, J .; Katzenellen-
bogen, B. S.; et al. Pyrazole Ligands: Structure-Affinity/Activity
Relationships of Estrogen Receptor-a Selective Agonists. J . Med.
Chem. 2000, 43, 4934-4947.
(42) Asada, S.; Choi, Y.; Uesugi, M. A Gene-Expression Inhibitor that
Targets an alpha-Helix-Mediated Protein Interaction. J . Am.
Chem. Soc. 2003, 125, 4992-4993.
(43) Pangborn, A. B.; Giardello, M. A.; Grubbs, R. H.; Rosen, R. K.;
Timmers, F. J . Safe and convenient procedure for solvent
purification. Organometallics 1996, 15, 1518-1520.
(44) Suffert, J . Simple direct titration of organolithium reagents using
N-pivaloyl-o-toluidine and/or N-pivalolyl-o-benzylaniline. J . Org.
Chem. 1989, 54, 509-510.
(45) Katzenellenbogen, J . A.; J ohnson, H. J ., J r.; Myers, H. N.
Photoaffinity labels for estrogen binding proteins of rat uterus.
Biochemistry 1973, 12, 4085-4092.
J M030404C