Identification of novel inverse agonists of estrogen-related receptors ERRγ and ERRβ

Article abstract of DOI:10.1016/j.bmc.2017.01.019

Estrogen-related receptors (ERRs, α, β, and γ) are orphan nuclear receptors most closely related in sequence to estrogen receptors (ERα and ERβ). Much attention has been paid recently to the functions of ERRs for their potential roles as new therapeutic t

Full text of DOI:10.1016/j.bmc.2017.01.019

Accepted Manuscript
Identification of Novel Inverse Agonists of Estrogen-Related Receptors ERRγ
and ERRβ
Donna D. Yu, Janice M. Huss, Hongzhi Li, Barry M. Forman
PII:
S0968-0896(16)31269-X
http://dx.doi.org/10.1016/j.bmc.2017.01.019
BMC 13498
DOI:
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
24 November 2016
10 January 2017
13 January 2017
Please cite this article as: Yu, D.D., Huss, J.M., Li, H., Forman, B.M., Identification of Novel Inverse Agonists of
Estrogen-Related Receptors ERRγ and ERRβ, Bioorganic & Medicinal Chemistry (2017), doi: http://dx.doi.org/
1
0.1016/j.bmc.2017.01.019
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Identification of Novel Inverse Agonists of Estrogen-Related Receptors ERR and
ERR
Donna D. Yu,* Janice M. Huss,* Hongzhi Li, and Barry M. Forman
Department of Diabetes and Metabolic Diseases Research, The Beckman Research
Institute, City of Hope National Medical Center, Duarte, California 91010, USA.
*
Corresponding Authors
D. D. Yu, Ph (626)-218-6993, E-mail: dyu@coh.org
J. M. Huss, Ph (626) 256-4673 x 65903, jhuss@coh.org
Present Addresses
Department of Diabetes and Metabolic Diseases Research, The Beckman Research
Institute at The City of Hope National Medical Center, Duarte, California 91010, USA.
Abstract
Estrogen-related receptors (ERRs, α, β, and ) are orphan nuclear receptors most closely
related in sequence to estrogen receptors (ERα and ERβ). Much attention has been paid
recently to the functions of ERRs for their potential roles as new therapeutic targets
implicated in the etiology of metabolic disorders. While no endogenous ligand has been
identified for any of the ERR isoforms to date, the potential for using synthetic small
molecules to modulate their activity has been demonstrated. In the present study, a series
of novel inverse agonists of ERR and ERRβ were synthesized using regio- and stereo-
specific direct substitution of triarylethylenes. These compounds were evaluated for their
1

ability to modulate the activities of ERRs. The rational directed substitution approach and
extensive SAR studies resulted in the discovery of compound 4a (DY40) as the most
potent ERR inverse agonist described to date with mixed ERR/ERRβ functional
activities, which potently suppressed the transcriptional functions of ERR with IC50
.01 μM in a cell-based reporter gene assay and antagonized ERR with a potency
=
0
approximately 60 times greater than its analog Z-4-OHT (Z-4-hydroxytamoxifen). In
addition, compound 3h (DY181) was identified as the most potent synthetic inverse
agonist for the ERR that exhibited excellent selectivity over ERRα/ in functional
assays. This selectivity was also supported by computational docking models that suggest
DY181 forms more extensive hydrogen bound network with ERRβ which should result
in higher binding affinity on ERRβ over ERRγ.
ABBREVIATIONS
ERRs, (Estrogen-related receptors); ER, (Estrogen receptor); NRs, (nuclear receptors);
LBD, (ligand-binding domain); SAR, (structure-activity relationship); MCAD, (medium-
chain acyl coenzyme A dehydrogenase). IC50, (half maximal inhibitory concentration).
2

Graphical Abstract:
1
. Introduction
Metabolic diseases, including obesity, diabetes, atherosclerosis and obesity-related
cancers, are the leading cause of mortality in industrialized nations. It is estimated that
over one-third of the United States population is obese, and these individuals are at
greater risk for developing diabetes, cancer and cardiovascular disease. These disorders
are responsible for over 500,000 deaths in the United States each year. The growing
incidence of this problem has led to intense interest in identifying new molecular targets
and new pharmacologic agents to treat and/or prevent these disorders.
3

Nuclear receptors are ligand-dependent transcription factors that regulate gene expression
in response to small molecule ligands. Orphan members of the nuclear receptor
superfamily, which have no identified endogenous ligand, are involved in regulation of
many aspects of cellular metabolism including mitochondrial energetics as well as
1
cholesterol, bile acid and glucose metabolism. Therefore, they represent an important
class of molecular targets for the treatment or prevention of a wide array of diseases.
The estrogen-related receptor (ERR) orphan receptor subfamily comprises three subtypes,
ERRα, ERRβ, and ERR. The ERRs are first orphan nuclear receptors identified based on
1
their structural similarity with the estrogen receptor (ER). Although it was originally
believed that the developmental and physiological roles of ERRs were quite distinct from
those of the classic ERs, recent studies have shown that in some contexts ERRs share
2
target genes, coregulatory proteins, ligands, and sites of action with the ERs. Such
3
evidence has been extensively reviewed. ERRs play an important role in the
transcriptional control of metabolic genes involved in the generation and utilization of
4
cellular energy. Thus, ERRs might present a therapeutic target for the prevention of
5
obesity and type 2 diabetes. Although the ERR, ERR, and ERR isoforms all share
considerable amino acid sequence identity with ERs in both the DNA binding domain
(
DBD) and ligand binding domain (LBD), they do not respond to natural estrogens, such
6
-8
as E (17β-estradiol, Figure 1).
2
4

ERRs are primarily expressed in the heart, skeletal muscle, brain, kidney, pancreas,
placenta, and liver and are predicted to have significant differences in their ligand-
9
-10
binding preferences.
However, the mechanisms governing target gene selectivity of
the individual ERR isoforms are not well understood. ERR regulates genes involved in
mitochondrial biogenesis (Gabpa/NRF2a, Tfam), fission/fusion (Mfn1 and 2, Opa1) as
well as metabolic enzymes of -oxidation (Acadm) and ETC/oxphos (Sdha, Cytc)
pathways in conjunction with PGC-1 (peroxisome proliferator-activated receptor
1
1
gamma coactivator 1), a master regulator of lipid and glucose homeostasis. In cancer
cells ERR also regulates cellular metabolism as well as genes relevant to proliferation
1
2-13
and metastasis and is associated with poor prognosis in human breast cancer.
ERR
expression correlates HER2 status and has been shown to mediate the effects of growth
factor signaling on metabolic reprogramming that is required for the development of
1
4-15
chemoresistance.
Thus, ERRα is considered a potential drug target with potential to
1
6-18
treat metabolic disorders and cancer.
ERR is present early in the developing placenta in a subset of cells in extra-embryonic
1
9-20
ectoderm destined to make up the chorion.
ERR is likely essential for
2
1
reproduction. ERR expression is essential for the maintenance of pluripotency and
self-renewal potential in mouse embryonic stem cells and is among the core transcription
2
2-23
factors that can reprogram fibroblasts into pluripotent stem cells.
Biochemical
evidence also suggests that ERRβ may be a potential therapeutic target involved in
2
4
cancers and metabolic disorders. However, studies have been hampered by the
embryonic lethality of ERRβ-/- mouse model and the lack of small molecule modulators
5

to study ERRβ function. Selective ERRβ ligands would provide an invaluable research
tool to examine the biological function of ERRβ.
ERRγ regulates gluconeogenesis in liver and is a potential candidate drug target to
2
5-
reverse hyperglycemia and hepatic fat accumulation in the context of insulin resistance.
2
6
27
ERR/PGC-1β promotes oxidative metabolism in cancer cells, and unlike ERR,
ERR expression is a favorable biomarkers in human breast cancer. Thus, targeting the
ERR pathway may be a powerful therapeutic strategy to treat metabolic disorders and
2
8
cancer.
A critical step to advance the understanding of ERR biology is to design or identify
selective modulators (agonists and inverse agonists) that can be used to target isoform
specific processes in cells and in vivo. Notably, no endogenous ligand for any of the
ERRs has been identified to date. The synthetic selective ligands developed to target
ERRα are described as inverse agonists (antagonists of constitutive ERR activity) and
2
9-30
agonists,
and have been utilized in cell culture and in vivo studies to investigate the
physiologic activity of ERRα. However, little is known about the molecular mechanisms
or biological activities downstream of ERRβ and ERR receptor activation because of the
lack of selective ERRβ/ ligands.
Recently, a small molecule agonist of the ERR and ERR was identified that mimics the
3
1-32
protein ligand PGC-1α in activating human ERRβ and ERRγ,
and compounds
including Z-4-OHT and DES (Figure 1), have been shown to act as nonselective ERRγ
6

3
3-35
inverse agonists.
inverse agonist GSK5182 and its analog (Figure 1).
isolated from the marine sponge steroids, have been reported to act as ERR
Structure-based approach was used to design the ERR selective
3
6-37
Likewise, 4-methylenesterols
3
8
antagonists. However, these new antagonists exhibit lower potency than DES in
transactivation experiments.
Despite these advances, the development of ERRβ and ERR inverse agonists with better
potency and selectivity remains a formidable challenge. As a part of our ongoing program
to explore novel classes of ERR modulators with the goal of increasing the potency and
selectivity for ERRβ and ERR subtypes which might prove to be of therapeutic value in
treating a variety of ERRβ and ERR-linked pathologies, we applied a strategy of altering
the structure of triarylethylene core which is a template of Z-4-OHT. The compound Z-4-
OHT has a very poor inverse agonist profile for ERR and ERRβ but binds to the LBD of
these receptors; therefore it was used as a scaffold to generate higher potency antagonists.
The design was based on SAR of Z-4-OHT analogs combined with analysis of the X-ray
crystal structures of Z-4-OHT bound to the ligand binding domains of ERR and ER.
Analogs bearing extension or branched alkyl groups at the C2 position of the
triarylethylene core with the basic side chain exhibited improved binding affinity and
3
9
selectivity profiles for ERR compared to Z-4-OHT. The existing X-ray crystal
structure of Z-4-OHT bound to ERR and ERα LBD provided a model for the molecular
3
6
basis of activity and selectivity. In the current study, we describe the chemical
synthesis and SAR (structure-activity relationship) for several triarylethylene derivatives
7

as well as molecular modeling of receptor binding and in vitro activity profiles of the
compounds.
Figure 1. Chemical structures of E
2
(the main physiologically relevant estrogen) and the
three previously disclosed ERR and ERR synthetic inverse agonists.
2
2
. Results and discussion
.1. Chemistry
Based on the recent structural studies we proposed the synthesis of targeted arrays of
small, triarylethylenes as ERR and ERRβ ligands. The resultant core scaffold of
triarylethylenes could be derivatized at three distinct sites: at the aliphatic side chain on
the C2 of triarylethylene position, the OH group on the 4-position of phenyl A-ring, and
8

alkylaminoethoxy-basic side chain on the 4-position of phenyl B-ring. We first examined
the biological activities of several different types of alkylaminoethoxy-basic side chain
into the phenyl B-ring and found that there was no improvement on the potency and
selectivity toward ERR and ERRβ (data not shown). Therefore, an aliphatic side chain
on the C2 of triarylethylene and 4-OH group on the phenyl A-ring were targeted to
improve the potency and selectivity toward to ERR and ERRβ.
Combining molecular basis, structure consideration, and chemical feasibility, the initial
strategy consisted of traditional medicinal chemistry and paralleled an approach
involving various substitutions on the triarylethylene core in an attempt to increase the
ERRβ and ERR specificity while performing the inverse agonistic selectivity on ERR
and ERRβ. As described in the literatures, the Z isomer of 4-OHT has the required
antiestrogenic activity, while the (E)-4-OHT has only about 5% of its affinity for the
4
0
ER. To determine if Z-4-OHT would have selectivity on ERRs, we tested the Z-4-OHT
and E-4-OHT isomers in ERRβ and ERR activity. The Z isomer of 4-OHT has the
required activity on the ERR and ERRβ, but the E-4-OHT was much less active for the
ERR targets (data not shown). Thus, analogues of triarylethylenes with a basic side chain
having a Z-form are highly desirable.
In order to establish a chemical library with various functionalities and structural
diversity, accordingly, we first focused on the extended ethyl side chains of
triarylethylene core. Retrosynthetically, we found that 1,1-bis(4-hydroxyphenyl)-2-
phenylethylene could serve as a key template for the conversion to a series of analogs
9

structurally related to the Z-4-OHT. We have previously described the synthesis of Z-4-
4
1
OHT in a regio- and stereo-selective controlled manner. Our intention was to extend
this method to versatile production of 1,1-bis(4-hydroxyphenyl)-2-phenylalkenes via
McMurry chemistrywhich is a well known established methodology for the cross-
4
2
coupling reaction (Figure 2).
The 1,1-bis(4-hydroxyphenyl)-2-phenylethylenes were synthesized by simply coupling
,4'-dihydroxybenzophenone with various propiophenone derivatives to introduce R
4
1
groups and make chemset 1 (Figure 2). To expand the number of accessible analogs, a
McMurry reductive coupling between 4,4'-dihydroxybenzophenone reacting with the
readily-available reagents such as isobutyrophenone, isovalerophenone, 4-oxo-4-phenyl-
butyric acid methyl ester, 2-methyl-4-oxo-4-phenyl-butyric acid methyl ester, 3-
benzoylpropionic acid, and 2-methyl-4-oxo-4-phenylbutyric acid in the presence of zinc
and titanium tetrachloride in dry THF under reflux conditions was performed. The
desired 1,1-bis(4-hydroxyphenyl)-2-phenylalkenes (1a-n) were obtained in good yields.
The carboxylic acid groups in 1i and 1j were reduced using BH -THF to give the
3
corresponding alcohols 1k and 1n in 75 and 74% yields, respectively. In chemset 1, 1c
and 1d are known compounds and we have re-synthesized them in excellent yields by
McMurry chemistry compared to the existing synthesis of Gust, Gauthier and
4
3-44
coworkers.
10

Figure 2. Synthesis of 1,1-bis(4-hydroxyphenyl)-2-phenylethylene analogs chemset 1,
installation of the basic side chain chemset 2, and stereo- and regiospecific production of
the Z-isomer chemset 3.
11

Next, we coupled compounds 1a-h, 1k, and 1n directly with 2-(dimethylamino)ethyl
4
5
chloride hydrochloride using a known method, and obtained an equimolar mixture of
the geometric isomers of (E,Z)-4-[1-(4-dimethylaminoethoxy-phenyl)-2-phenyl-but-1-
enyl]-phenols 2, as depicted in Figure 2. The initial coupling reaction favored the
formation of geometric Z-isomer, but it was then converted to a monoalkylated geometric
mixture after a while. In this procedure, monoalkylated geometric isomers are the
majority, while the dialkylated products were detected in trace quantities. Note that Z-4-
OHT is more potent than E-4-OHT for the ERRβ/ in the cell-based cotransfection
assay. Thus, the fixed ring analogs of substituted 1,1-bis(4-hydroxyphenyl)-2-
phenylalkenes in 3 series with Z-isomer are of high interest for this project. Treatment of
the crude mixture 2 of two E/Z stereoisomers in 1:1 ratio were easily separated by
selective crystallization, and in some cases, by flash chromatography. A series of
geometric mixture E/Z in 2 series as pure regio-selective Z-isomer 3 was ultimately
effectively separated. Examples include crystallization of the white solid 2a with a 1:1
Z/E ratio from warm methanol which led to white crystals 3a. The absolute configuration
assignment of 3a was determined as Z-isomer by X-ray structure analysis ((Figure 3).
Other configurations of the stereo Z-isomer in chemset 3 were determined by NOE
1
measurements and H NMR. Alternatively, the geometric isomers of pure Z-form in
1
chemset 3 are very easily identified by general H NMR because Z-form analogs have
upfield NMR signals of the 2-dimethylaminoethoxy chain protons compared to the
corresponding E-isomer by using CD OD as a solvent. By the improved synthetic
3
methodology we were able to minimize the number of synthetic steps and develop a
purification methodology for the production of a series of new Z-isomer analogs derived
12

from Z-4-OHT, compared to the Gauthier and GSK method which requires protective
3
6
groups and pivaloate deprotection. The general synthetic route used to prepare
substituted hydroxylated 1,1,2-tribenzyl-but-1-enes of the Z-isomer are outlined in Figure
2
.
Compound 3a
Figure 3. X-ray crystallographic structure of 3a indicated that the active geometric
isomer 3a is Z-form generated in a region controlled manner.
Previously the hydroxyl group on the phenyl A-ring of Z-4-OHT was believed to be
4
6
important for the binding affinity toward to the ERα. Z-4-OHT has higher affinity for
ERα than ERR because of the similarity between ERR and ERα in the first shell of the
ligand binding pocket. Note that the hydroxyl group on the phenyl A-ring interacts with
ERR in Glu-275 and Arg-316 (ERα Glu-353 and Arg-394), and the LBDs of ERR and
13

ERRβ are almost identical. This implies that the optimization in the 4-OH in the phenyl
A-ring would be challenging for the increase of the potency on the ERR. Furthermore,
SAR from X-ray crystal structures of Z-4-OHT bound to the LBD of ERR suggested
47
that the binding pocket may tolerate small substituents other than hydroxyl group. In
order to assess if the simple substituent variations of 4-OH group on the phenyl A-ring
interaction would lead to the improved potent and selectivity for ERR and ERRβ, next,
we designed and synthesized a small focused chemical library of phenyl A-ring with
diversity at position of 4-OH to generate structure-activity relationship (SAR). To speed
the synthesis, we applied solution-phase parallel synthetic techniques to exploit variant
functionalities as R to maximize the efficient interactions in molecular levels at the
2
binding pockets of ERRγ and ERRβ.
Compounds 3a and 3h have been selected as building blocks based on their improved
inverse agonist functional profile for the ERR and ERRβ. Thus, the hydroxyl function of
active precursors 3a and 3h was converted into the chemset 4 by simple estrification and
alkylation reactions to introduce a series of ester or ether pharmacophores as diversity to
the 4-position of phenyl A-ring. The 48 commercially available chemical reagents were
selected on the basis of diversity, the criteria being lipophilicities and chemistry
compatibility to achieve more flexible substitution patterns. This combinatorial synthesis
offers the advantage of compatibility with a series of functionalities. Table 1 shows the
chemical structures of the final R
2
at the phenyl A-ring moiety by varying the
hydrophobicity and size of the system. The identity of each of the final compounds was
1
13
confirmed by H and C NMR.
14

Table 1. Combinatorial synthesis of a library of precursors 3a, 3h with diversity at
position 4-OH of the pendant phenyl A-ring was synthesized using parallel solution-
phase synthetic techniques to induce R functions.
2
15

_
_______________________________________________________________________
16

To achieve different types of esters and an alternative access to substituted ethers in
chemset 4, a total of 48 commercially available acyl chlorides and alkyl halides were
reacted in hot DMF containing potassium or cesium carbonates as the base. The optimal
o
temperature for the reaction was 50 C. The construction of a focused small library by
parallel solution-phase synthesis was carried out in a Mettler-Toledo Bohdan RAM
organic synthesizer with an automated liquid handling robot. A workstation of Mettler-
Toledo system was used to tare each reactant vial in an array of vials. After the added
reagent was weighed, it was diluted to the desired concentration with organic solvents
and vortexed. Various combinations of reactants were then transferred to each of reaction
tubes for the screw synthesizer. The solution-phase in the library synthesis always posed
a challenge in purification procedures. To get quick and efficient purification in higher
purity in sufficient quantity for cell-based biological testing, the Biotage Quad 3
automated medium pressure flash chromatography was used for parallel purification. The
crude mixtures were then directly purified by automated unit system which performed 12
simultaneous flash chromatographs using prepacked silica gel columns. Twenty fractions
could be easily obtained from each run in under 20 minutes. The average yields of the
targeted compounds were > 85% and purity > 82%. The resulting fractions were
1
13
analyzed by flow injection NMR to confirm the chemical structures by H and C NMR.
By this protocol, a total of 96 corresponding esters and ethers of chemset 4 were obtained
in fair yields.
17

2
.2. Biological evaluations
To assess the functional transcriptional activity of these compounds, cell-based assays
using chimeric receptor Gal4 DNA-binding domain (DBD)-NR ligand binding domain
cotransfection assay (LBDs of ERRα, ERRγ, and ERRβ) were performed. As an initial
screen to determine the ERR and ERRβ inhibitory activity and selectivity, all
compounds were evaluated for in vitro potency and selectivity by transfection testing,
using CV-1 cells transfected with human and mouse ERRs. MH100 x 4 is a luciferase
reporter with four copies of a GAL4 UASG response element, where GAL-I refers to the
ligand-binding domain of the indicated receptor fused to the C-terminus of the GAL4
DNA-binding domain. Assays were performed using a Biomek automated workstation in
which the genes for the nuclear receptor, as well as a plasmid containing a response
element upstream of a luciferase cDNA, are transfected into CV-1 cells. Compounds
were evaluated for their ability to inhibit human ERRs at 0.03, 0.1, 0.5, 1, 2.5, 3, and 10
µM test concentrations. Those compounds that showed good ERRβ and ERR potency
and selectivity were evaluated at additional concentrations. Z-4-OHT used as a known
ERRγ and ERRβ inverse agonist to compare the constitutive activity of unliganded ERRγ
and ERRβ. A series of synthetic compounds of this focused library has been identified
and showed similar activities at the 1 μM concentration as compared with Z-4-OHT.
2
.3. Structure−Activity Relationships (SAR)
43
In 2002, Gust and coworkers reported a series of C2-alkyl substituted of triarylethylene
derivatives without basic side chains. They came to the conclusion that those compounds
they synthesized without a basic side chain in the B-ring possessed the same antagonistic
18

potency for ERs as compared with Z-4-OHT which contains a basic side chain that was
4
6
believed exhibiting antagonistic activity of 4-OHT on the ERs. However, it is important
to note that all of the tested compounds by Gust and co-workers did not cover extensive
studies of the structure-function relationship for the activity on the ERRs. Thus, we
explored if compounds in chemset 1, which is in the absence of basic side chain
functionality, would also exert some biological effects by modulating the activities of
ERRs. The compounds in chemset 1 have been evaluated by cell-based screening for
inverse agonistic activity against ERRs. As compared to the Z-4-OHT, most of the new
analogs in chemset 1 did not possess ERRs activity, but possess ERs activity (data not
shown). We came to the conclusion that in the class of 1,1-bis(4-hydroxyphenyl)-2-
phenylalkenes, having a basic side chain is not a prerequisite for exhibiting antagonistic
effect on the ER receptor. The antiestrogenic properties comprising estrogen receptor
binding depend only on the length of the C2-alkyl chain. These results are consistent with
previous findings that removal of the dimethylaminoethoxy side chain of Z-4-OHT did
4
5
not decrease the antagonistic effects on the MCF-7-2a cell line.
However, as shown in Figure 4, except those the most of the new analogs in chemset 1,
specifically, compounds1g and 1h were found to be active and have selectivity for ERR
with IC50 values of 0.6 and 0.3 M, respectively. This is a surprising finding because
compounds 1g and 1h exhibited excellent selectivity and inverse agonist efficacy on the
ERR and they were not active for the ERRα, also were not active for the ERR receptor
at the higher dose. In contrast, the analogs 1g and 1h, which possessed the corresponding
extended ethyl side chains bearing methyl esters in the C2-alkyl substituted system and
19

did not bear any basic side chain on the phenyl B-ring, displayed improved selectivity
profiles toward the ERRβ (Figure 4). This demonstrated that the presence of the extended
methyl ester side chain on the C2-alkyl position is crucial for the ERRβ to gain an
antagonistic activity.
Figure 4. Compound 1g and 1h in chemset 1 have been identified as selective inverse
agonists for the ERR.
20

Having a basic side chain is not a prerequisite for exhibiting high binding affinity and
antagonistic effects on the ER. However, as predictable, having a basic side chain of 1,1-
bis(4-hydroxyphenyl)-2-phenylalkenes is a prerequisite for exhibiting high binding
affinity and inverse agonistic effects on the ERR. Wherein the triarylethylene system, it
can be seen that a dimethylaminoethoxy basic side chain on one of the phenolic groups
3
6
(
phenyl B-ring) increase ERR inverse agonistic activity and selectivity. In our
experiments variation of a side chain on the C2 of triarylethylene core with basic side
chain appeared to be beneficial for ERR, and for the ERRβ inhibitory activity. All of
the resulting compounds in 3 series were found to possess the inverse agonistic potency
on the ERR and ERRβ. The data are shown in Table 2 with a select set only. This study
confirmed that the introduction of basic side chain on the para-position of the phenyl B-
ring requires to providing generally improved potencies on the ERR and ERRβ.
Interestingly, as can be seen in Table 2, replacing an ethyl group of Z-4-OHT with an
isopropyl group substituent (3a) led to an increase in inhibitory activity. The potency of
3
a was increased on the ERR and ERR (IC50 = 0.29/0.21M) by 3-fold as compared to
the Z-4-OHT. However, when an isopropyl was replaced by an isobutyl group, the
resulted compound (3g) has slightly less in potency on ERR and ERRβ (IC50 = 1.0/0.75
µM) compared to Z-4-OHT, as well as a 3-fold ERR/ potency less than 3a, as shown in
the Table 2. Particularly less inhibitory activities or a loss of inverse agonistic activity on
the ERR were observed with those compounds additionally bearing longer and bulkier
groups at the C-2 position. This result demonstrated that the ethyl group at the C2-
extension position could be replaced with small alkyls without substantial loss of activity.
21

The isopropyl group at the C2-position (3a) seems beneficial in terms of potency
increasing on the ERR and ERR.
However, in contrast, when the isopropyl group at the C2-position (3a) was changed to
the methyl esters located at the extended ethyl side chain, the corresponding compounds
3
g and 3h (DY181) exhibited significant ERR binding affinity and selectivity which has
no effect on ER, ER, ERRα, and ERR transcriptional activity up to 3 M. As shown
in Figure 5, the most potent ERR inverse agonist 3h with an extended methyl ester
group at the C2-position has been identified with excellent selectivity over ERRα/.
Compound 3h (DY181) displayed the highest ERR affinity (IC50 = 50 nM). These
results are consistent with previous findings from chemset1 (1h and 1g) wherein the
extension of C-2 ethyl side chain system having corresponding methyl esters appeared to
be beneficial for ERRβ activity. Apparently, a combination of extension of C-2 ethyl
side chain system having corresponding methyl ester and a basic side chain contributes
high potency and inverse agonistic effect on the ERRβ receptor. The correct mechanism
for this observed contribution to the selectivity is yet to be established.
22

Figure 5. Compounds 3h (DY181) and 3g have been identified as selective inverse
agonists for the ERR.
Combinatorial construction of a focused library by carrying out solution phase-assisted
synthesis led to the 96 analogs by removal of the hydroxyl group (4-OH) or replacing it
with small acetyl and alkyl substituents. Analogs which loaded onto the 4-position of
hydroxyl on the A-phenyl ring in chemset 4 as R
2
were generated by incorporating esters
or ethers to potentially increase lipophilic character which would be beneficial for ligands
23

to interact with the hydrophobic LBD of ERRs. As expected, the ligands with small
acetoxy improved potency and selectivity on the ERR, whereas the other ligands with
bulkier substituents as R showed less potency and sterically not well tolerated by ERRβ
2
and ERR. Several interesting compounds derived from 3a having acetoxy have been
identified in 4 series in repressing ERR and ERR. The acetate analog, exemplified by
DY40 (4a {R2(1)}) derived from 3a showed the greatest functional potency with the IC50
of 10 nM as shown in Figure 6. The improved potency and inverse agonist activity of the
acetoxy-substituted analog 4a {R2(1)} demonstrated that an acetoxy group on the A-ring
instead of hydroxyl seems to efficiently diversify the potency of ERR and ERR.
In fact, in chemset 4, small substituents R
bulky R substituents further attenuated activity of ERRβ and ERR. To rationalize these
results we proposed that the bulkier R substituents are not suitable for ERR and ERRβ
2
led to a slight increase inhibitory activity, and
2
2
activity, and probably the LBD of the receptors may not be regulated and tolerated by
bigger molecules. Thus, the size and shape of the ligand are key factors for the protein
modulation. However, analogs which loaded onto the 4-position of hydroxyl on the A-
phenyl ring in chemset 4 as R from the precursor 3h did not improve the potency as
2
compared with DY40 (4a) (data not shown). To gain further insight into the selectivity on
the ERRs, the same reaction expanded to a 0.5 g scale of compounds 4a{R2(1)} resulted
in an 88% yields. Additional experiments included the compounds 3k and 3n which were
also converted into the corresponding analogs 4k{R2(1)}, and 4n{R2(1)} in a 0.5 g scale
production resulted in 82%, and 79 yields, respectively.
24

Figure 6. Compound 4a (DY40) was evaluated in potency and selectivity by cell-based
transactivation assays using CV-1 cells transfected with human and mouse ERRs. CV-1
cells were transfected with reporter constructs and expression vectors and the fold
activation of the reporter construct was determined at several concentrations.
2
.4. In vitro Functional Data for Selected Compounds at ERRs Receptors
A high-throughput binding assay using cell-based assay at different concentrations was
carried out with the LBD of ERRs. From initial screening against ERRs, over thirty
active compounds have been identified as having modest ERR and ERRβ affinity, and
data are highly reproducible. Data were re-evaluated for 12 members of library (2
members in chemset 1, 6 in chemset 3, and 3 members in chemset 4). A summary of the
IC50 values that were derived from in vitro transactivation assays against the human
25

ERR, ERR and mouse ERR subtypes for compounds (1g, 1h, 3a, 3f, 3h, 3g, 3k, 3n,
4
4
a, 4k, and 4n) with Z-4-OHT is outlined in Table 2. As shown in Table 2, compound
a (DY40) as an inverse agonist of ERRβ and ERR displayed the greatest functional
potency ((IC50 = 10 nM) at about 60-fold greater than Z-4-OHT. The acetylating of
compounds 4k and 4n efficiently increased the affinity for ERRβ and ERR almost 3 to
7
-fold as compared with 3k and 3n. These data indicated that the acetoxy group on
phenyl A-ring is critical for the ERRβ and ERR in terms of potency and selectivity. The
selective members of three chemsets exhibited a wide range of inverse agonistic activity
on ERRβ and ERR and clearly indicated that these new compounds all were inactive on
ERR.
Interestingly, among them, 1g, 1h, 3g, and 3h were identified as ERRβ inverse agonists.
3
a, 3f, 3g, 3n, 4a, 4k, and 4n were identified as ERRβ and ERR inverse agonists. These
most promising inverse agonists indicated that the ERRβ and ERR inhibitory potency
and selectivity could be extremely sensitive to minor changes in chemical structure
within a modified series. However, based on the selectivity evaluation, except for
compounds 1g, 1h, 3g, and 3h, the number of compounds we synthesized in this project
cannot fully separate ERR activity from ER activity. For example, 4a (DY40) remains
an antagonist potency of ERα. It is worth noting that the compounds we synthesized that
are selective over ER remain a key challenge.
26

Table 2. A select set of Z-4-OHT analogs on ERR, β, and  binding affinities.
Compound R1
R2
R3
ERRα
ERRβ
ERR
IC50(uM) IC50(uM) IC50(uM)
Z-4-OHT CH
2
CH
3
H
(CH
2
)
2
N(CH
3
)
2
NA
0.65
0.6
1
1
3
h
g
a
CH
CH
CH(CH
CH CH(CH
2
CH(CH
3
)COOCH
3
H
H
H
H
H
H
NA
NA
NA
NA
0.3
0.6
5.5
6.5
0.29
1
2
CH COOCH
2
3
3
)
2
(CH
2
)
)
2
2
N(CH
3
)
)
2
2
0.21
0.75
3
3
f
h
2
3
)
2
(CH
2
N(CH
3
(
DY181)
CH
CH
CH
2
CH(CH
CH COOCH
CH
3
)COOCH
3
H
H
H
(CH
(CH
(CH
2
)
)
)
2
2
2
N(CH
N(CH
N(CH
3
)
)
)
2
2
2
NA
NA
NA
0.05
0.35
3.5
2.78
3.5
3
3
g
2
2
3
2
3
k
2
2
CH
2
OH
2
3
2.7
(
Z-GSK5182)
3
4
n
CH
2
CH(CH
3
)CH
2
OH
H
(CH
2
)
)
2
2
N(CH
3
)
)
2
2
NA
NA
2.25
0.01
2.5
a (DY40) CH(CH
3
)
2
CH
3
CO
(CH
2
N(CH
3
0.01
4
4
k
n
CH
2
CH
2
CH
2
OH
)CH
CH
3
CO
CO
(CH
2
)
)
2
2
N(CH
3
)
)
2
2
NA
NA
1.1
0.3
CH
2
CH(CH
3
2
OH
CH
3
(CH
2
N(CH
3
0.55
0.35
NA = not active at 10 µM.
IC50 is the concentration of test compound that gave 50% maximum efficacy.
CV-1 cells were co-transfected with appropriate reporter constructs and expression
vectors. The fold activation of the reporter construct by ligand was determined at several
concentrations of selected compounds. No induction of ERR dependent transcription
49
was observed for any of the compounds in tested.
27

2
.5. Validation of DY40 effects on endogenous ERR target gene expression and
metabolism in cell based models.
We next examined whether the newly synthesized DY40 could repress endogenous
expression of an ERRγ target gene. Previous studies showed that ERRγ is a constitutive
activator of the small heterodimer partner (SHP) gene, which is selectively regulated by
ERRγ via a direct binding of the receptor to a site located in the proximal promoter
5
0-51
region.
Consistent with these findings, ERRγ over-expression in 293HEK
upregulated the SHP transcript by 2.5-fold and treatment with 100nM DY40 completely
blocked ERRγ mediated induction (Figure 7A). We also evaluated the effects of the
compound in C2C12 myotubes that have significant endogenous ERRγ expression.
Consistent with the observed effects in 293HEK cells, DY40 repressed endogenous SHP
transcript levels by ~60% in C2C12 myotubes (Figure 7B).
Previous studies have shown that ERRγ deficient myocytes have reduced capacity
5
2
oxidize long-chain fatty acids. Therefore, we investigated the effects of treatment with
DY40 on -oxidation rates in C2C12 myotubes. Treatment of C2C12 myotubes with
DY40 for 48 hours resulted in a significant reduction in palmitate oxidation rates (-42%)
in myocytes compared to vehicle-treated controls (Figure 7C). Notably, knockdown of
ERRγ expression in C2C12 decreased oxidation by the same magnitude (data not shown).
Collectively, these data demonstrate robust and selective effects of DY40 on ERR
dependent gene regulation and metabolism in cells.
28

Figure 7. Compound 4a (DY40) inhibits expression of a specific ERRγ target gene and
reduces mitochondrial -oxidation rates. A. 293 HEK cells expressing either GFP or
ERR via adenoviral transduction were treated with either vehicle (DMSO) or 100 nM
DY40 for 48 hours. Expression of endogenous SHP (NR0B2) was quantitated by real-
time PCR. Values are normalized to -actin transcript and reported relative to GFP
untreated levels (=1.0). Data represent mean ± S.E.M. for 3 independent trials. B. C2C12
myotubes were treated with either vehicle or 100 nM DY40 for 48 hours. Expression of
endogenous SHP was measured by quantitative real-time PCR. Values are normalized to
3
6b4 (Rplp0) transcript and reported relative to expression in DMSO treated cells (=1.0).
Data represent mean ± S.E.M. for 3 independent trials. *, P < 0.05 DMSO versus DY40
treated values. C. C2C12 myotubes were treated with either vehicle or 100 nM DY40 for
3
4
8 hours. Oxidation of H-palmitate-BSA was measured as described in Methods.
Values are normalized to total cellular protein and reported relative to expression in
DMSO treated cells (=1.0). Data represent mean ± S.E.M. for 4 independent trials. * P <
0
.001 DMSO versus DY40 treatment.
29