Identification of small molecule regulators of the nuclear receptor HNF4α based on naphthofuran scaffolds

Article abstract of DOI:10.1016/j.bmc.2009.07.079

Nuclear receptors are ligand-activated transcription factors involved in all major physiological functions of complex organisms. In this respect, they are often described as drugable targets for a number of pathological states including hypercholesterolem

Full text of DOI:10.1016/j.bmc.2009.07.079

Bioorganic & Medicinal Chemistry 17 (2009) 7021–7030
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Identification of small molecule regulators of the nuclear receptor HNF4
a
based on naphthofuran scaffolds
Rémy Le Guével a, ,à, Frédérik Oger a,à, Aurélien Lecorgne b, Zuzana Dudasova c, Soizic Chevance b,
Arnaud Bondon d, Peter Barath c, Gérard Simonneaux b, Gilles Salbert a,
*
a Equipe SPARTE, Université de Rennes 1, UMR6026 CNRS, Campus de Beaulieu, Bat 13, 35042 Rennes Cedex, France
b Equipe ICMV, Université de Rennes 1, UMR6226 CNRS, Campus de Beaulieu, Bat 10C, 35042 Rennes Cedex, France
c Cancer Research Institute, Slovak Academy of Sciences, Vlarska 7, SK-833 91 Bratislava, Slovak Republic
d Equipe RMN-ILP, Université de Rennes 1, UMR6026 CNRS, Campus de Beaulieu, Bat 13, 35042 Rennes Cedex, France
a r t i c l e i n f o
a b s t r a c t
Article history:
Nuclear receptors are ligand-activated transcription factors involved in all major physiological functions
of complex organisms. In this respect, they are often described as drugable targets for a number of path-
Received 12 May 2009
Revised 24 July 2009
Accepted 26 July 2009
Available online 5 August 2009
ological states including hypercholesterolemia and atherosclerosis. HNF4
anized’ nuclear receptor which is bound in vivo by linoleic acid, although this natural ligand does not
seem to promote transcriptional activation. In mouse, HNF4 is a major regulator of liver development
and hepatic lipid metabolism and mutations in human have been linked to diabetes. Here, we have used
a yeast one-hybrid system to identify small molecule activators of HNF4 in a library of synthetic com-
pounds and found one hit bearing a methoxy group branched on a nitronaphthofuran backbone. A collec-
tion of molecules deriving from the discovered hit was generated and tested for activity toward HNF4 in
a (NR2A1) is a recently ‘deorph-
a
Keywords:
Nuclear receptors
HNF4
Naphthofuran
Yeast one-hybrid system
Linoleic acid
a
a
yeast one-hybrid system. It was found that both the nitro group and a complete naphthofuran backbone
were required for full activity of the compounds. Furthermore, adding a hydroxy group at position 7 of
the minimal backbone led to the most active compound of the collection. Accordingly, a direct interaction
of the hydroxylated compound with the ligand binding domain of HNF4a was detected by NMR and ther-
mal denaturation assays. When used in mammalian cell culture systems, these compounds proved to be
highly toxic, except when methylated on the furan ring. One such compound was able to modulate
HNF4a-driven transcription in transfected HepG2C3A cells. These data indicate that HNF4a activity
can be modulated by small molecules and suggest new routes for targeting the receptor in humans.
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
demic agents.5 Although fatty acyl-coenzyme A thioesters have
been proposed to modulate HNF4
later7 that these molecules are not good candidates for being clas-
sical ligands for HNF4 . Fatty acids have also been suggested as
endogeneous ligands of HNF4 on a structural basis. Indeed, they
were found in the ligand binding pockets of crystallized HNF4
and HNF4c 8,9
However the crystallized HNF4 ligand binding
a
activity,6 it has been reported
Orphan nuclear receptors are members of the nuclear receptor
superfamily of ligand regulated transcription factors for which
a
ligands have not been identified.1 Hepatocyte Nuclear Factor-4
a
,
a
also known as HNF4
a
, is a highly conserved member of this nucle-
a
ar receptor family that is expressed in liver, kidney, intestine, and
pancreas, activates transcription likely as a homodimer2 and was
considered as an orphan receptor until very recently. In the liver,
.
a
domain (LBD) dimer is not in the fully active conformation since
one monomer has helix 12 extended instead of the canonical
‘closed’ conformation observed in all activated receptors.10 Fatty
acid fitting into a nuclear receptor ligand binding pocket is a recur-
rent observation that has been made for several other LBDs ex-
HNF4
-hydroxylase (Cyp7a1) gene encoding a major cholesterol catab-
olizing enzyme and is itself targeted by bile acids in a complex
feed-back regulatory loop.3,4 Thus, HNF4
is an attractive target
a is known to up-regulate transcription of the cholesterol
7a
a
pressed in Escherichia coli and crystallized (ER
PPARb/d). What is unusual however, in the case of HNF4
a
, RXR
a
a
, RORb,
and
for the design and development of a new generation of hypolidi-
c
LBDs produced in E. coli, is that the bound fatty acid cannot be
exchanged.9,11 These data suggested that fatty acids may be struc-
* Corresponding author. Tel.: +33 223 23 66 25; fax: +33 223 23 67 94.
E-mail address: gilles.salbert@univ-rennes1.fr (G. Salbert).
Present address: U522 INSERM, Faculté de Médecine, Campus de Villejean, 35033
Rennes, France.
tural determinants of HNF4
mational transitions. A recent study highlighted linoleic acid as an
endogenous HNF4 ligand in cultured mammalian cells as well as
a rather than ligands regulating confor-
a
à
These two authors contributed equally to this study.
0968-0896/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2009.07.079

7022
R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
in mouse liver, but no function in the regulation of HNF4
a tran-
NO2
O
substitution of
nitro group
in position 2
scriptional activity could be attributed to the newly discovered
ligand.12 Interestingly, linoleic acid can be exchanged from the
receptor ligand binding pocket in cells.12 Furthermore, in fasted
animals, the ligand binding pocket was found to be devoid of
substitution of
methoxy group
in position 7
O
ligand, indicating that, in certain physiological situations, HNF4
exists in the apo form. Hence, synthetic or natural molecules other
than fatty acids might be used to manipulate HNF4 activity. If
a
a
Scheme 1.
such synthetic compounds were to be obtained, they would prove
to be extremely useful in precisely defining the physiological func-
summarized in Tables 2–4. The compound 7-methoxy-2-nitro-
naphtho[2,1-b]furan (2) was chosen as starting material and differ-
ent modifications of this compound have been made such as the
substitution (or modification) of methoxy group in position 7 or
the substitution of nitro group in position 2 (Table 2).
We have also prepared 2-nitronaphtho[2,3-b]furan (18), (E)-6-
methoxy-1-(2-nitrovinyl)naphthalen-2-ol (19), and 6-methoxy-1-
(2-nitroethyl)naphthalen-2-ol (20) (Table 3). Finally, we have also
decided to explore Suzuki–Miyaura cross-coupling reaction start-
ing from 7-bromo-2-nitronaphtho[2,1-b]furan (21a) to increase
the hydrophobicity of the compounds (Table 4).
tions of HNF4
ical disorders like hypercholesterolemia.
In an attempt to identify HNF4 agonists, the initial screening
a and would provide tools to intervene in physiolog-
a
of the National Center Institute (NCI, USA) diversity set of synthetic
compounds (1990 molecules) was first undertaken in a yeast one-
hybrid system. We discovered 4-methoxy-2-nitronaphtho[2,1-
b]furan (1, Table 1) that modulates HNF4a activity in yeast, at
the micromolar range. Through similarity searching of the com-
plete library (250,000 compounds), we generated a first focused
library containing 5 additional compounds (2–6) (Table 1) which
were able to stimulate HNF4a-mediated transcription, compound
We started with the classical preparation of 2-nitronaph-
tho[2,1-b]furan derivatives. The first general method for synthesiz-
ing 2-nitrobenzo[b]furans of biological interest was the
condensation between 2-hydroxybenzaldehyde and bromonitrom-
ethane reported by Royer et al.15 Since then, this method was
improved16 and extended to the 2-nitronaphthofuran series17 by
the same group. Therefore, a first synthetic route for the construc-
tion of 2-nitronaphtho[2,1-b]furan ring was adapted from this
2 leading to the highest level of transcriptional induction. Synthe-
sis of these molecules, were previously described by Royer and
Buisson.13 Because we found that these compounds also appeared
to reduce yeast growth, we chose to prepare a series of related
derivatives and analogs for further investigation. Since substitution
in position 7 seemed to give the most active compound such as 7-
methoxy-2-nitronaphtho[2,1-b]furan (2), we decided to set up a
new series with aromatic ring linked in this position. Thus, the
application of a Suzuki–Miyaura cross-coupling approach14 to a
range of phenyl C-7 substituted 2-nitronaphtho[2,1-b]furan from
a 7-bromo-2-nitronaphtho[2,1-b]furan (21a) was undertaken.
Table 2
First series of synthetic 2-nitronaphtho[2,1-b]furan and related derivatives (modifi-
cations in position 2 and in position 7)
Here, we report the identification of new activators of HNF4
a
using a random screening method through a yeast cell-based func-
tional approach. We also report the chemical optimization study of
the initial hit which led to the identification of a series of com-
R1
O
pounds that exhibit HNF4a agonist activities.
2. Chemistry
R2
Compounds
R1
R2
Reference
The strategy for hit modification is summarized in Scheme 1. To
complete the series, starting from 2-nitronaphtho[2,1-b]furan, all
the proposed synthetic modifications classified in two series are
2
3
7
NO2
NO2
NO2
OCH3
H
OH
21
17
19
8
NO2
OCH2COOEt
17
9
NO2
OCH2COOH
17
10
11
12
13
14
15
16
17
COOEt
COOEt
COOH
COOH
CONH2
COOH
H
H
OCH3
H
OCH3
18
18
18
18
New
New
20
Table 1
Active compounds found from similarity search
NO2
O
NO2
NO2
O
H
OCH2COOH
OMe
H
O
H
20
O
O
3
2
1
Table 3
Other synthetic compounds tested on HNF4
NO2
NO2
HO
NO2
OH
NO2
NO2
O
NO2
O
O
O
O
OH
18
O
O
5
4
6
19
20

R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
7023
Table 4
Compound synthesized using Suzuki–Miyaura coupling reaction
NO2
O
R
O
3
24
28
25
29
22
23
HO
HO
F3C
27
O
26
30
O
O
8
O
32
31
33
previous method. All the compounds (7–17) are described in Table
2. Excepted compounds 14 and 15, all these synthesis have been
adapted from synthesis previously reported by Royer et al.18–20
Compound 13 was converted into compound 14 by treatment
with thionyl chloride then ammonia (73% yield). The beginning
of the synthesis of compound 15 was adapted from synthesis of
compound 9. Then it was converted into ethyl 7-(2-ethoxy-2-oxo-
ethoxy)naphtho[2,1-b]furan-2-carboxylate following a procedure
used for compound 13. Compound 15 was obtained by treatment
with sodium hydroxide in a quantitative yield.
CHO
OH
Cl2CHOMe
TiCl 4
OH
Br
Br
21c
21b
BrCH2NO 2
K2CO3
NO2
NO2
RB(OH)2
Pd(PPh3)4
O
O
To complete the series, synthetic routes to 2-nitronaphtho[2,3-
b]furan (18, Table 3), which have also been described by Royer,21
were undertaken. Synthetic routes to (E)-6-methoxy-1-(2-nitrovi-
nyl)naphthalen-2-ol (19) and 6-methoxy-1-(2-nitroethyl)naphtha-
len-2-ol (20, Table 3) have been adapted from a previously
reported method for synthesis of 2-(2-nitroethyl)-phenols.21
Finally, we decided to introduce an aromatic ring in position 7
to increase the hydrophobicity of the ligands using 7-bromo-2-
nitronaphthofuran (21a) as starting compound (Scheme 2).
Synthesis of this compound has been described by Royer and
Buisson.22 First, 6-bromo-2-naphthol (21c) was formylated in posi-
tion 1 using tetrachloride titanium and dichloromethyl methyl
ether. Second step is the condensation of bromonitromethane on
21b to give 7-bromonitronaphthofuran (21a). Having secured a
route to multigram quantities of the required intermediate, 7-bro-
mo-2-nitronaphthofuran, attention was turned out to the investiga-
tion of the Suzuki–Miyaura coupling reaction.14 Employing classical
conditions, (Pd(PPh3)4, Na2CO3, toluene, ethanol, water), the palla-
dium-catalyzed cross-coupling reaction of 7-bromo-2-nitronapht-
hofuran with a number of arylboronic acids was found to proceed
smoothly in the presence of a base to give the corresponding biaryls
in high yields (72–89%) (Table 4). Both electron rich (31) and elec-
tron-poor boronic acids (28) were found to couple in high yields.
Indeed, these reactions were carried out with commercial grade
boronic acid, using palladium complex as received.
Na2CO3
Br
R
21a
Scheme 2.
3. Results and discussion
3.1. Hit discovery
HNF4
system designed to express the full-length human nuclear receptor
HNF4 fused in C-terminal of the DNA binding domain of the bac-
a agonist activities were evaluated using a one-hybrid
a
terial LexA protein in a yeast strain containing a b-galactosidase
reporter gene controlled by a promoter harboring LexA binding
sites. Screening of the National Center Institute (NCI, USA) diversity
set of synthetic compounds with this system allowed us to dis-
cover that 4-methoxy-2-nitronaphtho[2,1-b]furan (compound 1,
see Fig. 1) enhances HNF4
ter gene activity with an EC50 close to 1.5
Suchan activationwasnotobservedwhenusingotherfusionpro-
teins including either HNF4 -related proteins such as COUP-TFII or
theestrogenreceptor (ER ), or thetranscriptionactivationdomain
a
-mediated transcriptional of the repor-
l
M.
a
a
a
of the unrelated yeast protein Gal4 (Fig. 1). Related molecules were
found by screening the complete NCI library (250,000 compounds)

7024
R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
7
6
5
4
3
A
LexA-COUP-TFII
A/B
1
C/D
E
F
465
LexA-HNF4α
Gal4
ER
1
HNF-4α1
HNF-4α7
452
130
368
HNF-4 LBDΔF
2
1
0
350
B
HNF4a7 FL
HNF4a1 FL
HNF4a1 LBDΔF
300
250
200
150
100
80
10-8
10-7
10-6
10-5
10-4
Compound 1 concentration (M)
Figure 1. 4-Methoxy-2-nitro-naphtho[2,1-b]furan selectively activates HNF4
a in
yeast. Dose–response curves obtained with 4-methoxy-2-nitro-naphtho[2,1-
b]furan in various yeast transformants. Increasing concentrations of compound
were applied to yeast for 16 h before b-galactosidase assay. Note that b-galacto-
sidase activity was enhanced only in HNF4
compound.
a expressing cells in response to the
60
40
20
with Chemfinder software (Cambridgesoft) by similarity search
using 2-nitrobenzo[2,1-b]furan as a motif. Five new compounds
were extracted to generate a first focused library (Table 1) and tested
0
10-8
10-7
10-6
10-5
10-4
Compound 2 concentration (M)
for their ability to regulate HNF4
All the compoundswere able to activateHNF4
tion, although to different extents (Table 5). The compound leading
tothehighestb-galactosidaseactivitywasfoundtobe7-methoxy-2-
nitronaphtho[2,1-b]furan (2).
a
in the yeast one-hybrid system.
-mediatedtranscrip-
a
C
5
4
3
2
1
0
HNF4a7 FL
HNF4a1 FL
HNF4a1 LBDΔF
Initial screen was run with the full-length HNF4
LexA DNA binding domain and thus active compounds may modu-
late HNF4 via interactions outside the ligand binding domain. In
a fused to the
a
order to delineate the domains required for the action of napht-
hofurans, we generated additional fusion proteins bearing various
deletions (Fig. 2A).
Nuclear receptors usually possess two transcription activation
domains called AF-1 and AF-2.10 AF-1 is located in the N-terminal
part of the protein, whereas AF-2 is generated by the folding of the
ligand binding domain. HNF4
alternative promoters generating two isoforms, HNF4
a
can be synthesized in vivo from
and
7, which differ in their N-terminal domains: whereas
1 harbors an AF-1 domain, HNF4
7 does not.23–25 When
7 in yeast one-hybrid assays, we found that com-
a
1
HNF4
HNF4
a
a
10-8
10-7
10-6
10-5
10-4
a
Compound 2 concentration (M)
using HNF4
a
Figure 2. Compound 2 activates HNF4
a through the ligand binding domain. (A)
pound 2 was still active, although the intrinsic transcriptional
activity of the protein was low due to the lack of AF-1 (Fig. 2).
When using the ligand binding domain only, deleted from the F do-
main and fused to the LexA DNA binding domain, the nitronapht-
hofuran still activated transcription, indicating that its activity is
likely to rely on an interaction with the ligand binding domain of
the receptor (Fig. 2). Since the F domain, a C-terminal extension
of the protein beyond the ligand binding domain, has been shown
Schematic representation of HNF4 constructs used in this study. Numbers indicate
a
amino-acids. Letters indicate the various functional domains of the protein with A/
B: N-terminal domain harboring the transcription activation function 1 (AF-1), C/D:
DNA binding and hinge domains, E: ligand binding domain, F: C-terminal domain.
HNF4a1 and HNFa7 are naturally occurring variants generated by the use of two
alternative promoters. (B and C) Activation of the different constructs in yeast one-
hybrid assay by compound 2. Increasing concentrations of compound were applied
to yeast for 16 h before b-galactosidase assay. Note that deletion of AF-1 in HNF4
and the LBD F leads to a strong reduction in b-galactosidase activity (B) whereas
the constructs are activated to a similar extent by compound 2 (C).
a7
D
to inhibit the transcriptional activity of AF-2 in HNF4a through an
intramolecular folding mechanism inhibiting interaction with
transcriptional coactivators,24,25 our data suggest that compound
2 acts independently of this autoinhibitory process. Altogether,
these data indicate that nitronaphthofurans may stabilize the
active conformation of the nuclear receptor ligand binding domain,
likely through a direct binding into the hydrophobic pocket.
Table 5
EC50s and IC50s of selected compounds in yeast one-hybrid and MTT assays
Compound
EC50
Fold inductiona
1
2
3
4
5
6
7
14
Yeast one hybrid
lM
1.47
6.60
9.3
2.57
8.05
9.3
2.09
4.46
12.5
2.33
7.97
9.3
1.28
5.51
18.7
1.22
4.58
12.5
1.95
11.06
6.2
>100
2
100
@ (
lM)
b
MTT assay
IC50
4.41
2.24
2.51
2.28
>5
2.04
5
>5
a
Mean of fold induction in two independent experiments compared to DMSO.
Mean of IC50 from two independent experiments.
b

R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
7025
3.2. Hit derivation
We then designed and synthesized a series of new compounds
derived from the hit molecule 7-methoxy-2-nitronaphtho[2,1-
b]furan, according to Scheme 1, and tested them in yeast one-hy-
brid assay. We essentially targeted the two opposite sides of the
hit compound for modifications, namely carbon 2 and carbon 7
of the naphthofuran core. When leaving R1 intact (i.e., NO2), mod-
ification of R2 led to inactive compounds except for compound 7
which showed a twofold increase in HNF4a activation when com-
pared to the hit compound. The nitro group was then replaced by a
carboxy group in order to phenocopy fatty acids. Nonetheless,
compounds 12, 13, and 15 were all inactive. The requirement for
the nitro group for high activity of the compounds was further con-
firmed through 10, 11, 14, 16, and 17 which were also mostly inac-
tive, although compound 14 was able to activate twofold HNF4
very high concentrations (Table 5). Finally, when large hydropho-
bic groups were added in R2, activation of HNF4 was systemati-
cally lost (compounds 22–33). Naphthol derivatives (compounds
19 and 21) were also inactive. Hence, optimization of the hit com-
a at
Figure 3. (a) Expanded region of one-dimensional reference spectrum recorded at
303 K for compound 7 (200
10% DMSO-d6. Expanded region of (b) STD and (c) WaterLOGSY with NOE-ePHOGSY
spectra recorded for compound 7 in the same conditions in presence of 30
HNF4 Negative and positive signals in spectra (b) and (c) identify HNF4a
lM) in 20 mM phosphate buffer pH 7.4, 100 mM NaCl,
a
lM
a.
interacting and not interacting molecules, respectively. The positive narrow peaks
(*) originated from non-interacting molecules of the buffer.
pound generated a strong activator of HNF4a (compound 7). How-
ever, a number of hit-derived compounds lost activity, especially in
the absence of a nitro group, suggesting that the minimal backbone
whose emission properties change upon interaction with unfolded
protein. The use of environmentally sensitive dyes to monitor ther-
mal unfolding was reported in 199732 and later, this method was
developed for drug discovery to allow rapid identification of ligands
of target proteins from compound libraries.33 The plot of fluores-
cence intensity versus temperature has a hyperbolic shape for a
two-state unfolding mechanism and the melting temperature (Tm)
is defined as the midpoint of temperature of the protein-unfolding
transition. Very interestingly, thermofluor assay has been recently
used to screen the ligand binding domain of another orphan nuclear
receptor, estrogen-related receptor, and several classes of com-
pounds that stabilized the receptor were identified.34
for an HNF4a activator is the nitronaphthofuran.
3.3. Coumpound 7 directly interacts with the HNF4
binding domain
a ligand
Since nitronaphthofurans activate HNF4 through the ligand
binding domain, we examined binding of active compound 7 to
HNF4a LBD by running two kinds of NMR experiments, namely
Saturation Transfer Difference (STD) NMR and WaterLOGSY
(Water–Ligand Observation with Gradient SpectroscopY).26–31
Both techniques are known to be efficient methods to study pro-
tein–ligand recognition events and are classically used for drug
screening. Both of them do not require any labeling or immobiliza-
Purified apo-HNF4a LBD was subjected to gradually increasing
temperature and the Tm values measured in the presence of the
coactivator and/or potential ligand were compared with the Tm
values for the control protein sample without any coactiva-
tor/ligand. Subsequently, the change in unfolding temperature,
tion of either the ligand or the receptor and can be run using
concentration range in protein and excess of ligand(s).
lM
The choice of compound 7 among the active compounds was
motivated by the criterion of complete solubility of this ligand in
a minimal volume of deuterated organic solvent (10% DMSO-d6)
in order to preserve protein folding. Whereas compounds 1–5 were
not soluble in 20 mM phosphate buffer pH 7.4, 100 mM NaCl con-
taining 10% DMSO-d6, compound 7 showed a total solubility from
D
Tm, was calculated. As seen from Figure 4, the addition of SRC-1
coactivator peptide slightly increased the Tm of the HNF4 LBD.
Even though the co-crystal structure of HNF4 LBD and SRC-1 pep-
tide has been published,8 it seems that PGC-1 is the physiologically
relevant HNF4
coactivator.35 The thermal shift in the presence of
a
a
a
30 up to 500 lM. The substitution in position 7 of the 2-nitronaph-
PGC-1 peptide was indeed higher (0.84 0.28 °C) and comparable
with the one induced by compound 7 (1.06 0.48 °C). Interest-
ingly, the presence of both compound 7 and PGC-1 gave the high-
tho[2,1-b]furan by a polar group such as hydroxyl explains this
higher solubility in water. In addition, according to the results of
yeast one-hybrid assays, compound 7 is proposed to be a good can-
est average
binding of both ligand and the coactivator peptide. These positive
Tm can be coupled to an increase in structural order and a
reduced conformational flexibility. Most interestingly, the active
compound 7 was able to stabilize the protein whereas thermal sta-
bility was unchanged in the presence of the natural ligand linoleic
acid, which is inactive in mammalian cells,12 as well as in our yeast
one-hybrid system (data not shown).
DTm (1.43 0.39) (Fig. 4), suggesting cooperative
didate as a ligand. The apo-ligand binding domain of HNF4
a was
prepared from E. coli through a purification/renaturation protocol
and therefore was devoid of fatty acids. As shown in Figure 3,
STD and WaterLOGSY experiments gave identical results and sug-
D
gested that compound 7 is an exchangeable ligand of HNF4a. In-
deed, in STD experiments, signal from free ligand can only be
observed if there is direct interaction within the protein binding
site and dissociation of the bound ligand. In WaterLOGSY experi-
ments, the negative sign of the free ligand signals is relevant of
3.5. Modeling compound interactions with the HNF4
binding pocket
a ligand
compound 7 binding to HNF4
ously interacts with the ligand binding domain of HNF4
a
. Hence, compound 7 unambigu-
a
in vitro.
The ligand binding pocket of HNF4a
has been described.11 It is
3.4. Coumpound 7 induces conformational changes in HNF4
a
largely hydrophic in nature with only three polar residues (Ser181,
Ser256 and Arg226). The charged residue (Arg226) is involved in a
salt bridge with the carboxy group of fatty acids. We run docking
experiments with compounds 1, 2, 5, and 7 and the crystallized
ligand binding domain
The fluorescence-based thermal stability assay is used to identify
ligands that stabilize a protein against thermal denaturation. This
method measures fluorescence from a hydrophobic fluoroprobe
HNF4a ligand binding domain in the ‘closed’ conformation, after

7026
R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
could form respectively one or two bonds with Arg226. Binding
of a nitro group to an arginine residue (Arg288) has already been
described for the ligand binding pocket of the nuclear receptor
PPAR
being on two different helices, PPAR
c
occupied by nitrated linoleic acid.36 Interestingly, despite
c
Arg288 and HNF4a Arg226
positions are close in the ligand binding pocket (see additional
Fig. S2). Since nitrated linoleic acid is a much stronger activator
than linoleic acid,37 it is possible that binding of nitro group at this
position leads to productive conformational changes and could be
viewed as a determinant of transcriptional activity. However, in
view of the high activity of compound 1 in yeast-based assay, it
is likely that hydrophobic interactions are crucial for binding to
the protein and to achieve the structural rearrangements required
for transcriptional activation. Alternatively, nitronaphthofurans
may bind to another cryptic site to regulate the transcriptional
Figure 4. Thermal shift assay results for renatured HNF4a LBD in combination with
its coactivators and potential ligands. The HNF4a LBD was aliquoted into wells of a
activity of HNF4a.
96-well plate at 0.4 mg/mL in the presence of SYPRO orange. Peptides and/or
potential ligands were added in 3x molar excess and the plate was heated from
3.6. Modulation of HNF4a transcriptional activity in
mammalian cells
20 °C to 90 °C. Changes in the unfolding transition temperature (DTm) are means
from three to six independent experiments. Error bars represent the standard errors
to the means.
Results obtained in yeast suggested that nitronaphthofurans
presented either growth inhibiting or toxic effects on cells. Hence,
we first undertook toxicity assays in human HepG2C3A hepatoma
cells. Cells were treated with increasing concentrations of selected
compounds during 24 h before being subjected to the classical MTT
procedure. At the tested concentrations, all compounds were found
to be highly toxic to human cells, except compound 5, whose IC50
could not be determined within the range of concentrations tested
(Table 5). Accordingly, only this compound was used for subse-
quent analysis of nitronaphthofuran impact on the transcriptional
protein minimization. As depicted in Figure 5, data suggested that
all compounds could fit in the hydrophobic pocket of the nuclear
receptor while generating few close contacts, especially with
Ala223 and Leu236 (minimal distance between two atoms:
1.806 Å). Whereas compounds 2, 5, and 7 were predicted to gener-
ate bonds and hydrophobic interactions with the protein, com-
pound 1 could only be engaged in hydrophobic interactions due
to its quite different structure and position in the ligand binding
pocket (additional Fig. S1). Interestingly, the most active com-
pound (compound 7) was predicted to generate two bonds with
Arg226 and one bond with Ser256, whereas compounds 2 and 5
activity of HNF4
promoters (ApoB and ApoCIII) linked to the luciferase reporter gene
were transfected, together with a HNF4 expression vector, in
HepG2C3A cells which were then treated with compound 5 as well
as with linoleic acid for 24 h. Expression of HNF4 strongly acti-
a in cancer cells. Two known HNF4a responsive
a
a
vated both reporter genes and linoleic acid showed no effect, as
described earlier in similar reporter assays.12 When compound 5
was added, both basal and HNF4
were partially reduced in a dose-dependent manner, but in differ-
ent proportions. Accordingly, the ability of HNF4 to activate the
a-induced transcriptional levels
a
reporter genes, as expressed in ‘fold induction’, increased with
increasing concentrations of compound 5 (Fig. 6). This enhance-
ment of HNF4
ApoCIII reporter than with ApoB (Fig. 6A). At high concentrations,
linoleic acid was also found to reduce both basal and HNF4 -in-
a-driven activity was more pronounced with the
a
duced activity of the reporter genes, but this did not result in
any variation in the fold induction, except for the highest concen-
tration, which led to a significant reduction in the level of HNF4a-
driven transcriptional activity (Fig. 6B). Consistent with data
obtained in yeast, these results indicate that compound 5, despite
a certain toxicity, is able to modulate HNF4a activity in human
cells.
4. Conclusion
Screening of a chemical compound library led to the discovery
of a family of small molecules activating the nuclear receptor
HNF4a in a yeast cell-based assay. We further demonstrated that
Figure 5. Modeling of compound 1, 2, 5, and 7 in the ligand binding pocket of
HNF4a. (A) Model of compound 5 bound in the ligand binding pocket of HNF4a
(PDB 1m7w) generated by the docking software Gold 3.0. Images where generated
with Chimera software. Polar residues have been highlighted in white (Arg226 and
these compounds can bind directly to the ligand binding domain
of the nuclear receptor, most likely in the hydrophobic pocket
known to be occupied in vivo by linoleic acid.12 Whereas the nat-
ural ligand (i.e., linoleic acid) does not seem to be able to regulate
Ser256). The ligand binding pocket is shown as
representation (50% transparency). (B) Model of compound 7 bound in the ligand
binding pocket of HNF4 . (C) Model of compound 2 in the ligand binding pocket of
HNF4 . (D) Model of compound 1 in the ligand binding pocket of HNF4 . Note that
a grey cloud by ‘surface’
a
a
a
HNF4
naphthofurans can enhance HNF4
replacement of the nitro group by a carboxy group mimicking fatty
a
transactivation function (our data and those from12), nitro-
the presence of a methoxy group at position 4 shifts the nitro group away from
Arg226. For A, B and C, color patches indicate close contacts with Leu236 and
Ala223.
a
-driven activity. Interestingly,

R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
7027
5.1.1.1. Procedure for compound 14. To 157 mg of 12
(0.74 mmol) under argon and at 0 °C was added 81 L (1.11 mmol)
l
of thionyl chloride. The mixture was heated under reflux (65 °C) for
20 min. After cooling, the mixture was dried under vacuum to give
a solid which was washed two times with 5 mL of dichlorometh-
ane. This solid was dissolved in 10 mL of THF at 0 °C and 10 mL
of aqueous ammonia are slowly added. The mixture was stirred
for 100 min at 0 °C then dried under vacuum. The residue was
washed in 15 mL of water, filtered and the solid was washed two
times with 10 mL of water to give 125 mg of a brown solid
(73%); mp 186 °C, 1H NMR: (200 MHz, CDCl3), d (ppm): 5.94 (sl,
1H, CONH2); 6.59 (sl, 1H, CONH2); 7.48–7.78 (m, 3H, 3ArH);
7.81–8.09 (m, 3H, 3ArH); 8.19 (d, 1H, J = 8.1 Hz) MS: M+Å calcd for
C13H9O2, 211.0633. Found: 211.0623.
5.1.1.2. Compound 15. Mp 212 °C, 1H NMR: (300 MHz, DMSO-d6), d
(ppm): 4.82 (s, 2H, OCH2); 7.36 (d, 1H, J = 8.85 Hz, ArH); 7.50 (s, 1H,
ArH); 7.75–7.95 (m, 2H, 2ArH); 8.30 (s, 1H, ArH); 8.36 (d, 1H,
Å
J = 8.90 Hz, ArH) MS: M+ calcd for C14H9NO6, 286.04774. Found:
286.0485.
5.1.2. General procedure used for Suzuki–Miyaura cross-
coupling (22–33)
5.1.2.1. 7-(4-Methylphenyl)-2-nitronaphtho[2,1-b]furan (24). To a
solution of 150 mg of 21a (0.51 mmol) in 20 mL of a mixture (50/50)
in toluene and ethanol under argon was added 105 mg (0.77 mmol)
of 4-methoxybenzeneboronic acid, 82 mg (0.77 mmol) of sodium car-
bonate in 1.5 mL of water and 30 mg (0.03 mmol) of tetrakistriphe-
nylphosphine palladium. The mixture was heated under reflux
(90 °C) for 2 h. After cooling, the mixture was dried under vacuum
to give a solid. The residue was also washed with water (30 mL), ex-
tracted three times with dichloromethane (30 mL) and the combined
organic extracts were dried on sulfate magnesium. Dichloromethane
was removed and the residue was purified by silica gel column chro-
matography and eluted with dichloromethane to give 118 mg (76%)
of a yellow solid, mp 226 °C, 1H NMR: (300 MHz, CDCl3) d 2.46 (s, 3H,
Me); 7.35 (d, 2H, J = 7.93 Hz, 2ArH); 7.65 (d, 2H, J = 8.08 Hz, 2ArH);
7.72 (d, 1H, J = 9.11 Hz, ArH); 7.95 (d, 1H, J = 8.44 Hz, ArH); 8.07 (d,
2H, J = 9.16 Hz, ArH); 8.18 (s, 2H, 2ArH); 8.25 (d, 1H, J = 8.51 Hz, ArH)
MS: M+Å calcd for C19H13NO3, 303.0895. Found: 303.0890.
Figure 6. HNF4
human cancer cells. HepG2C3A cells were transfected with ApoB and ApoCIII
reporter genes together with a HNF4 expression vector. Cells were then treated for
24 h with compound 5 (A) or linoleic acid (B). Data are from one representative
experiment run in triplicates and correspond to the ratio of HNF4 -induced versus
a transactivation function is regulated by compound 5 in cultured
a
a
basal luciferase expression levels (fold induction) for each concentration of ligand.
acids led to transcriptionally inactive compounds although they
may bind to HNF4a (our NMR data, not shown). Hence, natural
or synthetic ligand anchoring by a carboxy group is probably
unable to induce conformational transitions which lead to a trans-
criptionally active structure. Further investigations of the nitro-
specific structural remodeling of HNF4a ligand binding domain
will certainly shed light on the structural determinants required
for transcription activation by this particular nuclear receptor.
5.1.2.2. 2-Nitro-7-phenylnaphtho[2,1-b]furan (22). Yellow solid
(80%), mp 206 °C, 1H NMR: (500 MHz, CDCl3) d 7.45 (t, 1H,
J = 7.53 Hz, ArH); 7.55 (t, 2H, J = 7.53 Hz, 2ArH); 7.77 (m, 3H,
5. Experimental
5.1. Chemical synthesis
3ArH); 8.00 (d, 2H, J = 6.71 Hz, ArH); 8.11 (d, 2H, J = 9.13 Hz,
Å
ArH); 8.22 (m, 2H, 2ArH); 8.27 (d, 1H, J = 8.51 Hz, ArH) MS: M+
5.1.1. General remarks
calcd for C18H11NO3, 289.0739. Found: 289.0770.
All reactions were performed under argon. Solvents were dis-
tilled from appropriate drying agent prior to use: Et2O and THF
from sodium and benzophenone, toluene from sodium, CH2Cl2
from CaH2, CHCl3 from P2O5 and all other solvents were HPLC
grade. Commercially available reagents were used without further
purification unless otherwise stated. All reactions were monitored
by TLC with Merck pre-coated aluminum foil sheets (Silica Gel 60
with fluorescent indicator UV254). Compounds were visualized
with UV light at 254 nm and 365 nm. Column chromatography
was carried out using silica gel from Merck (0.063–0.200 mm).
1H NMR was recorded using Bruker (Advance 300dpx and
200dpx spectrometers) at 300 MHz or 200 MHz. High-resolution
mass spectra were recorded on a Varian MAT 311 Micromass spec-
trometer in EI mode at the CRMPO (Centre Régional de Mesures
Physiques de l’Ouest, University of Rennes).
5.1.2.3. 7-(4-Methoxyphenyl)-2-nitronaphtho[2,1-b]furan
(23). Yellow solid (74%), mp 214 °C, 1H NMR: (200 MHz, CDCl3), d
3.91 (s, 3H, OMe); 7.07 (d, 2H, J = 8.7 Hz, 2ArH); 7.71 (m, 3H,
Å
3ArH); 7.90–8.30 (m, 5H, 5ArH) MS: M+ calcd for C19H13NO4,
319.0845. Found: 319.0838.
5.1.2.4. 7-(4-Butylphenyl)-2-nitronaphtho[2,1-b]furan (25). Yel-
low solid (88%), mp 152 °C, 1H NMR: (300 MHz, CDCl3), d 0.98 (t,
3H, J = 7.32 Hz, CH2CH3); 1.43 (m, 2H, CH2CH3); 1.70 (m, 2H,
ArCH2CH2); 2.71 (t, 2H, J = 7.55 Hz, ArCH2CH2); 7.35 (d, 2H,
J = 8.07 Hz, 2ArH); 7.66 (d, 2H, J = 8.11 Hz, 2ArH); 7.72 (d, 1H,
J = 9.14 Hz, ArH); 7.96 (d, 1H, J = 8.48 Hz, ArH); 8.07 (d, 2H,
J = 9.17 Hz, ArH); 8.18 (s, 2H, 2ArH); 8.23 (d, 1H, J = 8.46 Hz, ArH)
Å
MS: M+ calcd for C22H19NO3, 345.1365. Found: 345.1366.
All the syntheses of the compounds belonging to the first series
were adapted from a method previously reported excepted synthe-
sis of compounds 14 and 15 and compounds resulting from Suzu-
ki–Miyaura cross-coupling (22–33) which are described below.
5.1.2.5. 7-(6-Methoxynaphthalen-2-yl)-2-nitronaphtho[2,1-
b]furan (26). Orange solid (87%), mp 258 °C, 1H NMR: (300 MHz,
CDCl3), d 1.57 (s, 3H, OMe); 7.25–7.35 (m, 2H, 2ArH); 7.75 (d, 1H,

7028
R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
J = 9.14 Hz, ArH); 7.82–7.95 (m, 3H, 3ArH); 8.07–8.18 (m, 3H,
obtained from the National Cancer Institute (USA). For screening of
compound activity we used the DupLex-ATM yeast two-hybrid sys-
tem from OriGene Technologies, Inc. (Rockville MD, USA). The
Å
3ArH); 8.22 (s, 1H, ArH); 8.25–8.35 (m, 2H, 2ArH) MS: M+ calcd
for C23H15NO4, 369.1001. Found: 369.0984.
HNF4
rat HNF4
man HNF4
a
1 expression vector as well as the cDNA for human and
were gifts from Dr Talianidis (Athens). Full-length hu-
1 and human HNF4 7 were transferred in frame with
5.1.2.6. 4-(2-Nitronaphtho[2,1-b]furan-7-yl)phenol (27). Red so-
lid (72%), mp 251 °C, 1H NMR: (300 MHz, DMSO-d6), d 6.92 (d, 2H,
J = 8.53 Hz, 2ArH); 7.69 (d, 2H, J = 8.51 Hz, 2ArH); 7.95 (d, 1H,
J = 9.14 Hz, ArH); 8.03 (d, 1H, J = 8.58 Hz, ArH); 8.24 (d, 1H,
a
a
a
the LexA DNA binding domain (DBD) in the one-hybrid vector
pEG202 (OriGene Technologies, Inc.), after digestion with BamHI/
J = 9.20 Hz, ArH); 8.34 (s, 1H, ArH); 8.54 (d, 1H, J = 8.55 Hz, ArH);
NotI for HNF4a1 and EcoRI/NotI for HNF4a7. The rat HNF4a LBD
Å
8.91 (s, 1H, ArH); 9.65 (s, 1H, OH) MS: M+ calcd for C18H11NO4,
(aa 130–368) cDNA was transfered in frame from pGEX-6P-1 (GE
Healthcare) to pEG202 through EcoRI digestion. The human
COUP-TFII cDNA encoding aa 52–414 was amplified by PCR (for-
ward primer: 50-ACGCCATGGCAGACGGC-30, reverse primer: 50-
CTGTTTCACTCGAGCTTCTT ATTTTA-30) and was inserted in frame
with the LexA DBD in pEG202 after restriction digest with NcoI
and XhoI. The YEphER and the control Gal4 yeast expression vec-
tors were as previously described.38 The ApoCIII and ApoB-Lucifer-
ase reporter plasmids were gifts from Dr. Sladek (Riverside,
California) and have been described.39,40
305.0688. Found: 305.0697.
5.1.2.7. 2-Nitro-7-(4-(trifluoromethyl)phenyl)naphtho[2,1-b]
furan (28). Yellow solid (75%), mp 212 °C, 1H NMR: (300 MHz,
CDCl3), d 7.75–7.90 (m, 5H, 5ArH); 7.99 (d, 1H, J = 8.48 Hz, ArH);
8.12 (d, 1H, J = 9.14 Hz, ArH); 8.21 (s, 1H, ArH); 8.23 (s, 1H, ArH);
Å
8.30 (d, 1H, J = 8.50 Hz, ArH) MS: M+ calcd for C19H10NO3F3,
357.0613. Found: 357.0601.
5.1.2.8. (4-(2-Nitronaphtho[2,1-b]furan-7-yl)phenyl)methanol
(29). Orange solid (84%), mp 247 °C, 1H NMR: (300 MHz, DMSO-d6),
d 4.58 (d, 2H, J = 5.44, CH2OH); 5.27 (t, 1H, J = 5.60 Hz, CH2OH), 7.47
(d, 2H, J = 8.09 Hz, 2ArH); 7.83 (d, 2H, J = 8.11 Hz, 2ArH); 7.98 (d, 1H,
J = 9.18 Hz, ArH); 8.11 (d, 1H, J = 8.63 Hz, ArH); 8.30 (d, 1H,
5.2.2. One-hybrid assays and library screening
Compounds from the NCI library were tested in a 96-well-plate
format assay as described,41 using the LexADBD-HNF4
a1 one-hy-
brid system. In this assay, the LexADBD-HNF4a1 fusion protein
J = 9.23 Hz, ArH); 8.46 (s, 1H, ArH); 8.60 (d, 1H, J = 8.56 Hz, ArH);
binds to the promoter of the b-galactosidase gene and activates
transcription. Thus, the production of b-galactosidase in the cell re-
flects the transcriptional activity of the fusion protein. Compounds
were given to cells in DMSO (1% highest final concentration). Con-
trol experiments run with DMSO alone were used to correct for the
solvent effect. After cell lysis, b-galactosidase activity was revealed
Å
8.95 (s, 1H, ArH) MS: M+ calcd for C19H13NO4, 319.0845. Found:
319.0869.
5.1.2.9. 2-Nitro-7-(4-nonylphenyl)naphtho[2,1-b]furan (30).
Yellow solid (89%), mp 126 °C, 1H NMR: (300 MHz, CDCl3), d
0.91 (t, 3H, J = 6.06 Hz, CH2CH3); 1.20–1.45 (m, 12H, 6CH2);
1.70 (m, 2H, ArCH2CH2); 2.70 (t, 2H, J = 7.54 Hz, ArCH2CH2);
7.35 (d, 2H, J = 8.03 Hz, 2ArH); 7.66 (d, 2H, J = 8.05 Hz, 2ArH);
7.73 (d, 1H, J = 9.15 Hz, ArH); 7.98 (d, 1H, J = 8.49 Hz, ArH);
8.08 (d, 2H, J = 9.17 Hz, ArH); 8.19 (s, 2H, 2ArH); 8.24 (d, 1H,
J = 8.48 Hz, ArH) MS: [MꢀCH3]+ calcd for C26H26NO3, 400.1913.
Found: 400.1902.
with the fluorescent substrate 4-Methylumbelliferyl b-
D-galacto-
pyranoside (4-MUG, Sigma–Aldrich) at a final concentration of
40 lg/mL. In the primary screen, each compound was assessed at
three concentrations in triplicates (1.6 ꢁ 10ꢀ6 M, 8 ꢁ 10ꢀ6 M, and
4 ꢁ 10ꢀ5 M). Compounds that induced at least a twofold increase
in HNF4a-mediated transcriptional activity were selected for a sec-
ondary screen with concentrations ranging from 4 ꢁ 10ꢀ8 M to
10ꢀ4 M. Only those compounds showing a robust and consistent
5.1.2.10. 7-(Benzo[d][1,3]dioxol-5-yl)-2-nitronaphtho[2,1-b]
furan (31). Orange solid (88%), mp 280 °C, 1H NMR: (300 MHz,
DMSO-d6), d 5.76 (s, 2H, OCH2O); 7.07 (d, 1H, J = 8.07 Hz, ArH);
7.36 (d, 1H, J = 8.14 Hz, ArH); 7.47 (s, 1H, ArH); 7.98 (d, 1H,
J = 9.15 Hz, ArH); 8.06 (d, 1H, J = 8.62 Hz, ArH); 8.26 (d, 1H,
activation of HNF4a were selected for further analysis. EC50s were
deduced from activation curves obtained in yeast one-hybrid
assays.
5.2.3. Toxicity assay
J = 9.16 Hz, ArH); 8.40 (s, 1H, ArH); 8.57 (d, 1H, J = 8.56 Hz,
One hundred microliters of HepG2C3A cells (ATCC, CRL-10741)
Å
ArH); 8.94 (s, 1H, ArH) MS: M+ calcd for C19H11NO5, 333.0637.
at a concentration of 300 cells/
96-well plate and cultured for 24 h. Confluent cells were treated
with various concentrations of compounds from 0.002 M to
5 lM to flank EC50 determined in yeast one-hybrid experiment,
lL were inoculated to each of the
Found: 333.0611.
l
5.1.2.11. 7-(4-tert-Butylphenyl)-2-nitronaphtho[2,1-b]furan (32).
Yellow solid (86%), mp 202 °C, 1H NMR: (300 MHz, CDCl3), d 1.42
(s, 9H, Me); 7.56 (d, 2H, J = 8.37 Hz, 2ArH); 7.65–7.75 (m, 3H,
while DMSO (0.05%) was used as the negative control. Each condi-
tion was performed in quadruplicate. After treatment for another
3ArH); 8.00 (d, 1H, J = 8.46 Hz, ArH); 8.09 (d, 2H, J = 9.16 Hz, ArH);
48 h, 10 lL MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltet-
razolium bromide; Thiazolyl blue, 5 mg/mL) were added to each
Å
8.19 (s, 2H, 2ArH); 8.25 (d, 1H, J = 8.46 Hz, ArH) MS: M+ calcd for
C22H19NO3, 345.1365. Found: 345.1331.
well and cultured for 2 h. Then, after removing the medium, forma-
zan crystals were dissolved in 100 lL DMS0. Absorbance was mea-
5.1.2.12. 7-(Benzofuran-2-yl)-2-nitronaphtho[2,1-b]furan (33).
Orange solid (88%), mp 266 °C, 1H NMR: (300 MHz, DMSO-d6), d
7.25–7.40 (m, 2H, 2ArH); 7.65–7.75 (m, 3H, 3ArH); 8.03 (d, 1H,
sured at 570 nm and background at 630 nm using microplate
reader (Power Wave X5; Biotek). Data were acquired using KC4 ver-
sion 3.4 software (Biotek instruments Inc.). Results (% of cell viabil-
J = 9.14 Hz, ArH); 8.30–8.40 (m, 2H, 2ArH); 8.59–8.72 (m, 2H,
ity) were calculated as follows: [(A570–A630) compound/(A570–A630)
DMSO]X100. Results were from two independent experiments.
Å
2ArH); 8.96 (s, 1H, ArH) MS: M+ calcd for C20H11NO4, 329.0688.
Found: 329.0663.
5.2.4. Transfection assays
5.2. Biology
HepG2C3A cells were plated at a density of 6 ꢁ 104 per well in
24-wells plates in Dulbecco’s Modified Eagle Medium (DMEM,
Invitrogen), 24 h before transfection. Cells were transfected using
JetPEI reagent (polyplus-transfection) according to manufacturer’s
instructions. Briefly, cells were transfected using 800 ng of total
5.2.1. Materials and cloning
All yeast media were from Sigma–Aldrich. The synthetic com-
pound library as well as individual compounds from Table 1 were

R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
7029
plasmid DNA [250 ng of reporter plasmid (ApoB-Luc or ApoCIII-
Luc), 250 ng of plasmid coding b-galactosidase (pCH-110), 50 ng
frequency, can be found in the literature.29–31 A reference 1D spec-
trum of the ligand alone and a 1D WaterLOGSY spectrum in the
presence of the protein were recorded. The first water selective
180° pulse, which does not require high selectivity, was applied
during 5 ms. The first two Pulsed Field Gradients (PFGs) had a typ-
ical duration of 2 ms. This strength is sufficient to destroy the un-
wanted magnetization and, at the same time, it avoids signal losses
due to diffusion occurring between the first two PFGs. A weak rect-
angular PFG is applied during the entire length of the mixing time
(1 s). A short gradient recovery time of 1 ms is applied at the end of
the mixing time before the detection pulse. The water suppression
in both experiments was achieved with the excitation sculpting se-
quence (Hwang and Shaka, 1995). The data were collected with a
sweep width of 14 ppm, an acquisition time of 0.584 s, and a relax-
ation delay of 2 s. Prior to Fourier transformation the data were
multiplied with an exponential function with a line-broadening
of 0.5 Hz.
of plasmid coding full-length HNF4
pCR3.1 as ballast to complete to 800 ng] during 24 h in 500
DMEM. Then, media were removed and cells were treated using
increasing concentrations of compound 5 (from 1 M to 20 M),
linoleic acid (from 1 M to 500 M) or DMSO 0.5% as control, dur-
ing 24 h. At the end of treatment, media were removed and 100
a
(HNF4
a-pCR3.1) and plasmid
l
L of
l
l
l
l
lL
of Reporter Lysis Buffer (Promega) were directly added into each
well of plates which were frozen (ꢀ80 °C) during 24 h. Cells were
then harvested, transferred into microtubes and centrifuged
(14,000 rpm, 1 min, 4 °C). Supernatants were collected and 30 lL
were used to quantify luciferase activity using Luciferase Assay
System (Promega) following supplier’s recommendations. To
determine transfection efficiency, b-galactosidase activity was
measured using 4-methylumbelliferyl b-
MUG) fluorescent substrate (Sigma Aldrich). Briefly, 40
concentrated 4-MUG solution (80 g/mL) were added to 40
supernatant and fluorescence was read (excitation 365 nm, emis-
sion 450 nm) in microplate-reading fluorimeter (Fluorolite
D
-galactopyranoside (4-
L of 2X
L of
l
l
l
5.4. Thermofluor assays
a
1000, Dynatech Laboratories). All experiments were performed at
least three times in independent manner in triplicates and fold
induction activities were calculated as follows: [(mean of lucifer-
ase activity)[compound]/(mean of b-gal activity)[compound]]/[(mean of
luciferase activity)DMSO/(mean of b-gal activity)DMSO]. Standard
deviations were calculated using triplicate values in each indepen-
dent experiment.
Ligand binding domain of human HNF4a (amino-acids 133–
368) has been cloned into expression vector pETM11 (EMBL, Hei-
delberg) containing 6xHis tag and proteolytic cleavage site for
TEV protease. Recombinant protein was bacterially expressed and
purified under denaturing conditions. Purification, renaturation
and tag cleavage of HNF4a LBD will be published elsewhere
(Dudasova et al., manuscript in preparation). In thermofluor assay,
the coactivator peptides bearing the following primary sequence
were used: SRC-1 (NH2-RHKILHRLLQEGSPS-COOH); PGC-1 (NH2-
EEPSLLKKLLLAPANT-COOH).
5.3. NMR
5.3.1. NMR samples
A real-time PCR device (Roche) was used to monitor protein-
unfolding by the increase in the fluorescence of the fluorophor SY-
PRO Orange (Sigma Aldrich S5692). Protein sample (final concen-
Ligand stock solution of 10 mM was prepared in DMSO-d6.
10
HNF4
containing 100 mM NaCl. 40
ligand concentration was 200
l
L of this stock solution was diluted in 450
LBD solution prepared in 20 mM phosphate buffer pH 7.4
L of DMSO-d6 were added. The final
M. A reference tube containing only
lL of 30 lM apo
a
tration of 0.3 mg/mL or 11
containing 100 mM NaCl and the 3ꢁ molar excess of the test com-
pound (33 M final concentration) in a reaction volume of 20
lM) in 20 mM Tris–Cl buffer (pH 8.0)
l
l
l
lL
the ligand (without protein) was prepared with identical
conditions.
were incubated in 96-well microplates covered by optical foil.
Water was added instead of test compound in the control samples.
SYPRO Orange was diluted 100ꢁ to yield a 20ꢁ working concentra-
tion. The thermofluor assay was performed using Light Cycler 480
Instrument II (Roche) by increasing the temperature at 1 °C/min
ranging from 20 to 90 °C and data points were collected in 1 °C
intervals. The wavelengths for excitation and emission were 498
and 580 nm, respectively. Melting temperatures for individual
wells were established by analysis with provided software. In the
absence of ligands (control), Tm = T0, and the ligand-dependent
5.3.2. Acquisition of NMR spectra
All NMR spectra were recorded at 303 K with a spectral width of
14 ppm on Bruker Avance 500 MHz spectrometer, equipped with a
5 mm TXI inverse triple-resonance cryoprobe head. (PRISM plat-
form, University of Rennes 1, France).
5.3.3. STD (Saturation Transfer Difference) experiments
Selective saturation of the protein was achieved by a train of
Gauss-shaped pulses of 50 ms length each, truncated at 1%, and
separated by a 1 ms delay. 40 selective pulses were applied, lead-
ing to a total length of the saturation train of 2 s. The on-resonance
irradiation of the protein was performed at a chemical shift of
0.7 ppm. Off-resonance irradiation was set at 40 ppm, where no
protein signals are present. Unwanted magnetization water was
suppressed using a WATERGATE 3-9-19 pulse sequence. A 30 ms
spin lock pulse was applied to remove protein background signals.
Total scan number in the STD experiments was 128. The spectra
on- and off-resonance were subtracted after multiplication by an
exponential line-broadening function of 0.5 Hz prior to Fourier
transformation. Spectra processing was performed using TOPSPIN
1.3 software (Bruker).
changes in midpoint temperature,
for individual samples.
D
Tm = Tm ꢀ T0, were calculated
5.5. Methods for computer-simulated ligand binding (docking)
5.5.1. Protein input file preparation
A clean HNF4a input file was generated by using the protein
preparation wizard of Maestro software (Maestro 8.5, academic
campaign, http://www.schrodinger.com). The protein was cleaned
by removing water molecules, ligands and subunits b, c and d from
the 1m7w original pdb file. The bond order was assigned and
Hydrogen atoms were added. The cleaned protein was minimized
using the CHARMm22 molecular mechanic force field imple-
mented in Hyperchem 8.0 software (Hypercube, Inc.) using the
Fletcher-Reeves conjugate gradient algorithm. The resulting recep-
5.3.4. WaterLOGSY (Water–Ligand Observation with Gradient
SpectroscopY) experiments
tor (HNF4a a subunit) was saved to a PDB file.
The details of the pulse sequence version used for the Water-
LOGSY experiment reported here, using ePHOGSY-NOE selective
excitation with a 180 shaped pulse at H2O position or at another
5.5.2. Ligand input file preparation
The ligand input structure was generated and 3d optimized
with MarvinSketch Academic Package (MarvinSketch 5.1.4, 2008,

7030
R. Le Guével et al. / Bioorg. Med. Chem. 17 (2009) 7021–7030
9. Wisely, G. B.; Miller, A. B.; Davis, R. G.; Thornquest, A. D., Jr.; Johnson, R.;
Spitzer, T.; Sefler, A.; Shearer, B.; Moore, J. T.; Miller, A. B.; Wilson, T. M.;
Williams, S. P. Structure 2002, 10, 1225.
ChemAxon http://www.chemaxon.com). The ligand structure was
saved as a mol2 file.
10. Gronemeyer, H.; Gustafsson, J. A.; Laudet, V. Nat. Rev. Drug Disc. 2004, 3, 950.
11. Dhe-Paganon, S.; Duda, K.; Iwamoto, M.; Chi, Y. I.; Shoelson, S. E. J. Biol. Chem.
2002, 277, 37973.
12. Yuan, X.; Ta, T. C.; Lin, M.; Evans, J. R.; Dong, Y.; Bolotin, E.; Sherman, M. A.;
Forman, B. M.; Sladek, F. M. Plos One 2009, 4, e5609.
13. Royer, R.; Buisson, J. P. Eur. J. Med. Chem. 1983, 18, 79.
14. Miyaura, M.; Suzuki, A. Synth. Commun. 1981, 11, 513.
15. Royer, R.; Demerseman, P.; René, L. Bull. Soc. Chim. Fr. 1970, 10, 3740.
16. Tromelin, A.; Demerseman, P.; Royer, R. Synthesis 1985, 1074.
17. Einhorn, J.; Lamotte, G.; Buisson, J. P.; Demerseman, P.; Royer, R. Eur. J. Med.
Chem. 1984, 19, 143.
18. Cavier, R.; Buisson, J. P.; Lemoine, J.; Royer, R. Eur. J. Med. Chem. 1981, 16, 73.
19. Royer, R.; Buisson, J. P.; Rene, L. Bull. Soc. Chim. Fr. 1972, 11, 4158.
20. Dixit, M.; Sharon, A.; Maulik, P. R.; Goel, A. Synlett 2006, 1497.
21. Dauzonne, D.; Royer, R. Synthesis 1984, 1054.
5.5.3. Gold 3.0 docking protocol
For the study, the binding pocket of the receptor was defined
from the crystallographic coordinates of Met182. Dockings were
performed under standard default settings of GOLD software
(CCDC Software Ltd, Cambridge): population size of 100, number
of islands of 5, number of operations 100,000, niche size of 2 and
a selection pressure of 1.1. Images were generated with Chimera
software (UCSF).
Acknowledgments
22. Royer, R.; Buisson, J. P. Eur. J. Med. Chem. 1980, 15, 275.
23. Thomas, H.; Jaschkowitz, K.; Bulman, M.; Frayling, T. M.; Mitchell, S. M.;
Roosen, S.; Lingott-Frieg, A.; Tack, C. J.; Ellard, S.; Ryffel, G. U.; Hattersley, A. T.
Hum. Mol. Genet. 2001, 10, 2089.
24. Eeckhoute, J.; Moerman, E.; Bouckenooghe, T.; Lukoviak, B.; Pattou, F.;
Formstecher, P.; Kerr-Conte, J.; Vandewalle, B.; Laine, B. Endocrinology 2003,
144, 1686.
25. Ruse, M. D., Jr.; Privalsky, M. L.; Sladek, F. M. Mol. Cell. Biol. 2002, 22, 1626.
26. Mayer, M.; Meyer, B. Angew. Chem., Int. Ed. 1999, 38, 1784.
27. Piotto, M.; Saudek, V.; Sklenar, V. J. Biomol. NMR 1992, 2, 661.
28. Sklenar, V.; Piotto, M.; Leppik, R.; Saudek, V. J. Magn. Reson., Ser A 1993, 102,
241.
We thank the National Cancer Institute (USA) for providing
compounds. We also thank Maud Bizot, Gaëlle Palierne and Cath-
erine Martin for technical support. We are grateful to Dr. Eeckho-
ute for critical reading of the manuscript. This work was funded
by the European Community (QLG1-CT-2001-01513 and LSHM-
CT-2006-037498), the Région Bretagne (SIE 211-B3-11), the Centre
National pour la Recherche Scientifique and the Ministère de l’ens-
eignement supérieur et de la recherche.
29. Dalvit, C. J. Magn. Reson., Ser B 1996, 112, 282.
Supplementary data
30. Dalvit, C.; Ramage, P.; Homme, U. J. Magn. Reson. 1998, 131, 148.
31. Dalvit, C.; Pevarello, P.; Tato, M.; Veronesi, M.; Vulpetti, A.; Sundstrom, M. J.
Biomol. NMR 2000, 18, 65.
32. Poklar, N.; Lah, J.; Salobir, M.; Macek, P.; Vesnaver, G. Biochemistry 1997, 36,
14345.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmc.2009.07.079.
33. Pantoliano, M. W.; Petrella, E. C.; Kwasnoski, J. D.; Lobanov, V. S.; Myslik, J.;
Graf, E.; Caryer, T.; Asel, E.; Springer, B. A.; Lane, P.; Salemme, F. R. J. Biomol.
Screen. 2001, 6, 429.
References and notes
34. Abad, M. C.; Askari, H.; O’Neill, J.; Klinger, A. L.; Milligan, C.; Lewandowski, F.;
Springer, B.; Spurlino, J.; Rentzeperis, D. J. Steroid Biochem. Mol. Biol. 2008, 108, 44.
35. Yoon, J. C.; Puigserver, P.; Chen, G.; Donovan, J.; Wu, Z.; Rhee, J.; Adelmant, G.;
Stafford, J.; Kahn, C. R.; Granner, D. K.; Newgard, C. B.; Spiegelman, B. M. Nature
2001, 413, 131.
36. Li, Y.; Zhang, J.; Schopfer, F. J.; Martynowski, D.; Garcia-Barro, M. T.; Kovach, A.;
Suino-Powell, K.; Baker, P. R.; Freeman, B. A.; Chen, Y. E.; Xu, E. Nat. Struct. Mol.
Biol. 2008, 15, 865.
37. Schopfer, F. J.; Lin, Y.; Baker, P. R.; Cui, T.; Garcia-Barro, M.; Zhang, J.; Chen, K.;
Chen, Y. E.; Freeman, B. A. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 2340–2345.
38. Métivier, R.; Petit, F. G.; Valotaire, Y.; Pakdel, F. Mol. Endocrinol. 2000, 14, 1849.
39. Sladek, F. M.; Dallas-Yang, Q.; Nepomuceno, L. Diabetes 1998, 47, 985.
40. Dallas-Yang, Q.; Jiang, G.; Sladek, F. M. BioTechniques 1998, 24, 580.
41. Le Guével, R.; Pakdel, F. Biotechniques 2001, 30, 1000.
1. Akiyama, T. E.; Gonzales, F. J. Biochim. Biophys. Acta 2003, 1619, 223.
2. Bogan, A. A.; Dallas-Yang, Q.; Ruse, M. D., Jr.; Maeda, Y.; Jiang, G.; Nepomuceno,
L.; Scanlan, T. S.; Cohen, F. E.; Sladek, F. M. J. Mol. Biol. 2000, 302, 831.
3. De Fabiani, E.; Mitro, N.; Anzulovich, A. C.; Pinelli, A.; Galli, G.; Crestani, M. J.
Biol. Chem. 2001, 276, 30708.
4. Inoue, Y.; Yu, A. M.; Yim, S. H.; Ma, X.; Krausz, K. W.; Inoue, J.; Xiang, C. C.;
Brownstein, M. J.; Eggerstein, G.; Björkhem, I.; Gonzales, F. J. J. Lipid Res. 2006,
47, 215.
5. De Fabiani, E.; Mitro, N.; Gilardi, F.; Caruso, D.; Galli, G.; Crestani, M. J. Biol.
Chem. 2003, 278, 39124.
6. Hertz, R.; Magenheim, J.; Berman, I.; Bar-Tana, J. Nature 1998, 392, 512.
7. Bogan, A. A.; Dallas-Yang, Q.; Ruse, M. D., Jr; Maeda, Y.; Jiang, G.; Nepomuceno,
L.; Scanlan, T. S.; Cohen, F. E.; Sladek, F. M. J. Mol. Biol. 2000, 302, 831.
8. Duda, K.; Chi, Y. I.; Shoelson, S. E. J. Biol. Chem. 2004, 279, 23311.