Discovery of Farnesoid X Receptor Antagonists Based on a Library of Oleanolic Acid 3-O-Esters through Diverse Substituent Design and Molecular Docking Methods

Article abstract of DOI:10.3390/molecules22050690

The pentacyclic triterpene oleanolic acid (OA, 1) with known farnesoid X receptor (FXR) modulatory activity was modified at its C-3 position to find new FXR-interacting agents. A diverse substitution library of OA derivatives was constructed in silico through a 2D fingerprint similarity cluster strategy. With further docking analysis, four top-scored OA 3-O-ester derivatives were selected for synthesis. The bioassay results indicated that all four compounds 3 inhibited chenodeoxycholic acid (CDCA)-induced FXR transactivation in a concentration-dependent mode. Among them 3b and 3d are more active than the parent compound OA. A molecular simulation study was performed to attempt to explain the structure-Activity relationship (SAR) and the antagonistic action. To the best of our knowledge, this is the first report on semi-synthetic pentacyclic triterpenoids with FXR-modulatory activities.

Full text of DOI:10.3390/molecules22050690

molecules
Article
Discovery of Farnesoid X Receptor Antagonists Based
on a Library of Oleanolic Acid 3-O-Esters through
Diverse Substituent Design and Molecular
Docking Methods
Shao-Rong Wang 1,2,†, Tingting Xu 3,4,†, Kai Deng , Chi-Wai Wong , Jinsong Liu and
Wei-Shuo Fang *
1
5
4
1
,
1
State Key Laboratory of Bioactive Substances and Functions of Natural Medicines, Institute of Materia
Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, 2A Nanwei Road,
Beijing 100050, China; wangshaorong@outlook.com (S.-R.W.); dkatalent@sina.com (K.D.)
Center for Drug Evaluation, China Food and Drug Administration, 1A Fuxing Road, Beijing 100038, China
School of Life Sciences, University of Science and Technology of China, Hefei 230026, China;
xu_tingting@gibh.ac.cn
State Key Laboratory of Respiratory Disease, Guangzhou Institutes of Biomedicine and Health, Chinese
Academy of Sciences, Guangzhou 510530, China; liu_jinsong@gibh.ac.cn
NeuMed Pharmaceuticals Limited, Unit 509, 5/F BioTech Center I, No. 9 Science Park West Avenue, Shatin,
Hong Kong, China; wongcw123456@yahoo.com
2
3
4
5
*
Correspondence: wfang@imm.ac.cn; Tel.: +86-10-6316-5229
†
These authors contributed equally to this paper.
Academic Editor: Diego Muñoz-Torrero
Received: 15 March 2017; Accepted: 21 April 2017; Published: 26 April 2017
Abstract: The pentacyclic triterpene oleanolic acid (OA, 1) with known farnesoid X receptor (FXR)
modulatory activity was modified at its C-3 position to find new FXR-interacting agents. A diverse
substitution library of OA derivatives was constructed in silico through a 2D fingerprint similarity
cluster strategy. With further docking analysis, four top-scored OA 3-O-ester derivatives were selected
for synthesis. The bioassay results indicated that all four compounds
acid (CDCA)-induced FXR transactivation in a concentration-dependent mode. Among them 3b and
are more active than the parent compound OA. A molecular simulation study was performed
3 inhibited chenodeoxycholic
3
d
to attempt to explain the structure-activity relationship (SAR) and the antagonistic action. To the
best of our knowledge, this is the first report on semi-synthetic pentacyclic triterpenoids with
FXR-modulatory activities.
Keywords: FXR antagonist; oleanolic acid; compound diversity; library design; molecular modeling
1
. Introduction
Farnesoid X receptor (FXR), initially discovered as a bile acid homeostasis regulating
transcriptional factor [
1,
2], is well known for its involvement in diverse metabolic processes, including
the control of cholesterol, lipid and glucose metabolism [
3]. This has made FXR an attractive
pharmacological target, especially for the treatment of cholestasis, atherosclerosis and diabetes, etc.
In addition, FXR has also been found to play a role beyond metabolism, e.g., in tumorigenesis [4].
Many medicinal chemistry efforts have been made to prepare FXR agonists [
transcription of various genes under the control of FXR. Some of them are very potent FXR-interacting
agents with sub- M or even nM level IC50 values, such as 6-ECDCA/INT-747, the first FXR agonist
approved by the U.S. FDA in 2016 for the treatment of primary biliary cholangitis, GW4064, feraxamine,
5,6] to activate the
µ
Molecules 2017, 22, 690; doi:10.3390/molecules22050690
www.mdpi.com/journal/molecules

Molecules 2017, 22, 690
2 of 13
MFA-1 and WAY-362450 (Figure 1), etc. To date, aside from the success of 6-ECDCA/INT-747, only a
limited number of these agonists have however been studied in humans.
Figure 1. Structures of some farnesoid X receptor (FXR) agonists.
It was reported that full agonism of FXR may lead to the activation of various downstream
genes, resulting in complex and unexpected biological responses. For example, some complete FXR
agonists were found to decrease the level of high-density lipoprotein (HDL) while increasing the
level of low-density lipoprotein (LDL) [7], which may lead to cardiovascular-related adverse effects.
In this regard, the exploration of partial agonists, antagonists, or modulators has been proposed as
an alternative.
In recent years, some FXR antagonists have displayed beneficial effects in cholestasis [
8] and
hypercholesterolemia [ ] animal models, suggesting that FXR antagonists may be worth pursuit in this
9
context. However, the available pharmacological data on FXR antagonists are still limited and some
results are contradictory, therefore, we tried to discover more antagonists to increase the structural and
also possibly functional diversity.
In 2010, one of our co-authors reported oleanolic acid (Figure 2, OA,
modulator [10]. It was found that, OA dose-dependently suppressed the activity of FXR-LBD
ligand binding domain) induced by its endogenous ligand, chenodeoxycholic acid (CDCA). At 25
concentration, it partially blocked CDCA induction of bile salt export protein (BSEP), but did not affect
the expression of organic solute transporter (OST- ). The expression of short heterodimer partner
SHP) was slightly enhanced and its induction by CDCA was not affected in the presence of OA, and
1) as a selective FXR
(
µM
β
(
the expression levels of cholesterol 7α-hydroxylase (CYP7A1) was suppressed in the presence of CDCA,
OA, and their combination. Therefore, OA functioned as a gene selective modulator of FXR in HepG2
cells. However, since both SHP and CYP7A1 are regulated by many different signaling pathways,

Molecules 2017, 22, 690
3 of 13
including several nuclear receptors, and the effects on SHP or CYP7A1 cannot unambiguously be
interpreted as activity on FXR [11]. In contrast, BSEP is a more specific target gene which is almost
exclusively regulated by FXR. From this perspective, OA may better be regarded as a partial antagonist
of FXR. In addition, a low dose of OA (20 mg/kg) could protect against lithocholic acid (LCA)-induced
cholestasis in mice [12], and attenuate obstructive cholestasis in bile duct-ligated mice [13
due to its FXR antagonism.
,14], partially
Figure 2. Structure of oleanolic acid (OA).
Although OA has not been studied extensively, these preliminary findings indicate that it could
be used as a lead compound for further optimization. To the best of our knowledge, only one natural
analogue of OA, platycodin D (a 3,28-substituted saponin), has been known to exert weak activity on
FXR [15]. In addition, 3-β-(2-carboxybenzoyloxy)-oleanolic acid (NPLC441) was previously reported
to bound strongly to retinoid X receptor-LBD [16]. Based on the above information, we chose to derive
3
β-OH during our initial effort to optimize OA.
2
. Results and Discussion
In order to maximize our chance of finding better FXR-interacting compounds during our
exploration in the chemical space around the C-3 position in OA, we decided to introduce diverse
substituents, among which a limited number of representative structures were selected with the aid of
different filters, for this position.
To build up a small diverse OA 3-O-ester library consisting of various carboxyl acids from
a commercial compounds database, MDL® Available Chemicals Directory (ACD) containing a
total of 34,895 carboxylic acids was utilized. Non-compliant compounds according to Linpinski’s
“
Rule of 5” [17] were first filtered out of the ~35 K acids in the library to minimize the total number of
the building blocks by using the Discovery Studio 2.0 software (Accelrys Inc., San Diego, CA, USA).
Considering the frequently observed violation of this empirical rule by natural products, and also
with the intention of including the physiochemical properties of the parent compound
parameter settings, we modified this rule with our own customized criteria:
1 itself in the
(
(
1) For a balance between more choice of building blocks and drug-like properties, i.e., not too large a
molecular weight (MW), we set the upper MW limit of the carboxylic acids to 200. In Lipinkski’s
rule, the MW is set to below 500 for orally active compounds. Considering the MW of itself is
56, using MW 500 for 3-O-esters of OA would significantly limit the choice of carboxylic acids.
2) For other parameter settings, e.g., hydrogen bond donors (HBD) and hydrogen bond acceptors
HBA), the functional groups presenting already in OA (at C-3 and C-28) are also considered,
1
4
(
so these parameters were set as follows: HBD equal or less than 4, HBA equal or less than 8,
the number of rotatable bonds equal or less than 7, the number of rings equal or less than 2, the
number of aromatic rings equal or less than 2.

Molecules 2017, 22, 690
4 of 13
After filtering the ~35 K acids using the above criteria, only 6106 compounds were left in the
library. Next, the compounds with highly reactive or unstable groups, such as aminonitrile, thionitrile,
epoxide, aziridine, disulfide, nitrogen oxides, azo compounds,
β-lactams, etc., were also filtered
out. In addition, the molecules which seemed not to be very “drug-like”, such as long-chain fatty
acids, gem-dicarboxylic acids, compounds with boron, selenium or silicon atoms, ion exchange resins,
quaternary ammonium salt, etc., as well as inorganic or organic salts of some carboxylic acids were
also removed. After all the above filtering steps, 4454 compounds remained and were subjected to the
diversity analysis.
A 2D fingerprint [18] is a set of binary strings encoding the presence or absence of sub-structural
fragments in a certain molecule, which can be used for the measurement of inter-molecular structural
similarity and for the evaluation of structural and biological diversity of a compound library. We chose
molecular fingerprint FCFP_4 as the descriptor and clustered all of the 4454 carboxylic acids to 285
clusters using Accelrys Discovery Studio 2.0. One acid was randomly selected from each cluster to
assemble a sub-library of 285 acids. Those acids were used as the building blocks to create a 3-O-ester
library of OA, for the next step virtual screening by docking.
A total of 285 OA 3-O-esters and one positive control molecule (fexaramine, Figure 1) were
screened against FXR (PDB code 1OSH) using Autodock 4.2 (The Scripps Research Institute, La Jolla,
CA, USA). The positive control fexaramine scored the highest among those compounds with a pK of
i
10.39, and its binding mode is consistent with the experimental determination. Further examination of
the positive control and other four top-scored compounds 3a (Figure 3) using the Autodock Vina
–
d
software showed a similar scoring list and binding position in FXR compared with the docking results
from Autodock 4.2. The best compound in the triterpenoid library has a pK of 9.68 while the pKi
i
values of the other three compounds ranged from 8.00 to 9.35 (Table 1).
Figure 3. OA analogs to be synthesized.
Table 1. The predicated pK values and antagonistic activities of oleanolic analogs.
i
Prediction
1
3a
3b
3c
3d
Fexaramine
pKi
Autodock)
IC50 (µM)
8
.87
8.67
9.35
7.03
8.00
9.68
9.03
10.39
/
(
13.69
19.41
13.74
All above four triterpenoids were synthesized as below. The C-28 carboxyl group was protected
before C-3 derivatization, according to a modified procedure of literature [19], to afford benzyl
3
β-hydroxyolean-28-oate (
4) (Scheme 1) so as to improve the overall yield of the OA 3-O-esters, based
on our experience on triterpene chemistry.

Molecules 2017, 22, 690
5 of 13
Scheme 1. Preparation of compound 4.
Compound 3a was prepared as shown in Scheme 2. Acid 2a was first synthesized according
to [20]. Starting from benzaldehyde ( ), consequential transformations including a [3 + 2] cycloaddition
of chlorobenzaldoxime ( ) and propargyl alcohol to afford phenyl substituted isoxazole acid 2a, which
was further converted to acyl chloride that reacted with to furnish the 3-O-acylated analogue 10.
5
7
9
4
However, the debenzylation of 10 to afford the desired product 3a proved troublesome, maybe due to
the ring-opening of the isoxazole group under hydrogenolysis conditions as reported [21]. Thus, 3a
was finally prepared in reasonable yield (63.0%) by direct acylation of 1 with 9.
Scheme 2. Synthesis of compound 3a.
For the synthesis of 3b, acid 2b was initially prepared successfully by reacting 2-aminothiazole
with bromoacetic acid (11) [22], but the aminothiazole acetic acid failed to attach to 3-OH
0
in
4
either by coupling reaction in the presence of N,N -dicyclohexylcarbodiimide (DCC) or
0
N-(3-dimethylaminopropyl)-N -ethylcarbodiimide (EDCI) in the presence of 4-dimethylaminopyridine
(
DMAP) or using the corresponding acyl chloride. Alternatively, bromoacetic acid 11 was coupled
with to afford 12 23], which was then treated with 2-aminothiazole to afford the desired product 3b
Scheme 3) in 50.6% yield over two steps.
1
[
(
Scheme 3. Synthesis of compound 3b.

Molecules 2017, 22, 690
6 of 13
Compound 3c was much easier to synthesize since acid 2c is commercially available. Ester 13 was
obtained by coupling of 2c and in the presence of DCC and DMAP, and then the benzyl group was
removed readily by catalytic hydrogenolysis to afford the final product 3c in good overall yield (89.6%)
Scheme 4).
4
(
Scheme 4. Synthesis of compound 3c.
According to the literature method [24], acid 2d was synthesized by Michael addition reaction
between 3-methylpyridine (14) and -methylacrylic acid (15) in the presence of hydroquinone in 81.0%
α
yield., After attachment of acid to 3-OH and debenzylation of 28-COOH, the final product 3d was
obtained in 41.6% yield over two steps (Scheme 5).
Scheme 5. Synthesis of compound 3d.
The biological activities of these OA analogues against FXR were evaluated using a co-transfection
assay in human embryonic kidney 293T cells. None of the compounds had any effect when
incubated with the cells alone, but they could inhibit CDCA-induced FXR transactivation in a
concentration-dependent manner. The IC50 values of compounds 3a–d were measured as 19.41,
7
.03, 13.74, and 9.03 M, respectively (Table 1). It is worth noting that the trend of measured IC50
µ
values was consistent with the predicted pK from the virtual screening, indicating the feasibility of
i
this ligand-FXR model.
A further, more exhaustive docking was carried out to study the mechanism of action for these
ligands by docking them into FXR (PDB code 1OSH) using Autodock 4.2. For 3b, it was shown that two
pairs of T-shaped π-π interactions exist between the thiazoline ring and one aromatic residue (Trp458,
H11, Figure 4). This interaction seems important for anchoring the compound in the binding site, since
other atoms of 3b contribute to interactions only by shape matching and hydrophobic interaction with

Molecules 2017, 22, 690
7 of 13
H3 (H=helix), H5, H6, and H7. It is believed that H12 plays a crucial role in the activation of FXR by
both natural or unnatural agonists. Agonists such as CDCA could activate FXR by directly stabilize
H12 through forming two important H-bonds with the protein [25] However, in some cases, not only
H12 but also the other helices, especially H3, are necessary to enable coactivator recruitment [26].
A so-called “charge clamp”, formed by a glutamic acid residue from H12 (Glu471 in FXR) and a lysine
residue from H3 (Lys307 in FXR) in the coactivator-interacting surface, exists in many NRs and is
able to stabilize their active conformations [27]. In addition, there are a significant number of contacts
between the methyl ester group in fexaramine and H3/H6 of FXR, which significantly contribute for
its potency [28]. In contrast, these interactions are absent in the model of 3b binding to FXR, although
the binding positions of these two compounds are similar. Moreover, the amide CO of fexaramine
formed two H-bonds with His298 in the middle section of H3 and with Ser336 in H5, respectively,
strengthening the stabilizing effect of the ligand. Although there is a hydrophobic interaction between
compound 3b and His298, it might not be strong enough to stabilize the overall conformation of
H3, especially hold the remote key residue Lys307 (one of the component of the “charge clamp”) in
the proper position, thus compromising the whole stabilizing effect of the compound. The binding
mode of compound 3d was very similar with 3b, so that they have similar activities. The lack of
some important interactions, such as the H-bonds with His298 and Ser336, and the hydrophobic
interaction with H3/H6, may account for their inability to exert agonist activity, so that they exhibited
only antagonistic effect.
Figure 4. Binding modes of fexaramine (yellow) and compound 3b (magenta) into FXR. Key residues
involved in the formation of “charge clamp” and the interactions with fexaramine and compound 3b
are all marked and presented as sticks. Helices discussed in the text are numbered.
In contrast, the binding poses for the other analogs are different. For OA, since the molecule is
smaller than other derivatives, there is lack of some interactions with H5 and H6 in FXR (Figure S1).
However, an existing H-bond between C-28 COOH of OA and Tyr373 at H7 might compensate some
lost interactions. For 3a (Figure S1, 3c is similar), although there is an H-bond between this compound
and Tyr373 in a subpocket, which is encompassed by H3, H11, and H12, the binding pose and position
are quite different from those of the above mentioned trierpenoids and fexaramine. Thus, the binding
affinities of compounds OA, 3a, and 3c, which are lower than the more active compounds 3b and 3d
,
made them less potent for the FXR modulation.

Molecules 2017, 22, 690
8 of 13
3
. Materials and Methods
3
.1. General Information
Oleanolic acid (purity > 98%) was purchased from National Institute for the Control of
Pharmaceutical and Biological Products (Beijing, China). All chemical reagents were of standard
quality and used without further purification. Column chromatography was carried out with silica gel
1
13
as the stationary phase. H- and C-NMR spectra were measured on Mercury-500 or Mercury-300
spectrometers (Varian, Palo Alto, CA, USA). The mass spectra (MS) were measured on an 1100 LC/MSD
high performance ion trap mass spectrometer and an ESI mass spectrometer (LCQ) (Agilent, Santa
Clara, CA, USA).
3
.2. Synthesis of Compound 10
To a stirred solution of
4
(20 mg, 0.036 mmol) in dry dichloromethane (DCM) (0.5 mL) was
(0.072 mmol) in dry DCM (0.5 mL) under Ar.
The mixture was stirred for 3 h at room temperature, then DCM (10 mL) was added, and the organic
layer was washed with 2 M HCl (aq.) (5 mL 3), sat. NaHCO (aq.) (5 mL 3) and sat. NaCl (aq.)
5 mL 3) sequentially, dried over Na SO and concentrated. The residue was purified through a
added Et N (15.6 µL, 0.108 mmol) and a solution of 9
3
×
×
3
(
×
2
4
silica chromatography column (petroleum ether
−
EtOAc 20:1) to afford the product 10 as a white solid
1
(
16 mg, 61.5%). H-NMR (300 MHz, CDCl ):
δ
7.84 (2H, m, H-2, 6 of Ph), 7.48 (3H, m, H-3, 4, 5 of Ph),
3
7
.35 (5H, m, Bn), 7.20 (1H, s, H of isoxazole), 5.30 (1H, br s, H-12), 5.08 (2H, ABq, J = 12.6 Hz, CH of
2
Bn), 4.79 (1H, t, J = 8.1 Hz, H-3α), 2.92 (1H, d, J = 9.6 Hz, H-18), 1.14 (3H, s, CH ), 1.00 (3H, s, CH ),
3 3
+
0
.96 (6H, s, 2CH ), 0.93 (3H, s, CH ), 0.91 (3H, s, CH ). ESI-MS: m/z [M + H] 718.7.
3
3
3
3
.3. Synthesis of Compound 3a
To a stirred solution of 1a (25 mg, 0.055 mmol) in dry DCM (0.5 mL) was added Et N (23.8
µL,
3
0
.165 mmol) and a solution of 9 (0.11 mmol) in dry DCM (0.5 mL) under Ar. The mixture was stirred
for 10 h at room temperature, then DCM (10 mL) was added, and the organic layer was washed with
M HCl (aq.) (5 mL 3), sat. NaHCO (aq.) (5 mL 3) and sat. NaCl (aq.) (5 mL 3) sequentially,
dried over Na SO and concentrated. The residue was purified through a silica chromatography
2
×
×
×
3
2
4
column (petroleum ether
−
EtOAc
−
HOAc 15:1:0.1) to afford the product (3a) as a white solid (19 mg,
7.84 (2H, m, H-2, 6 of Ph), 7.48 (3H, m, H-3, 4, 5 of Ph), 7.21
), 2.83 (1H, d, J = 12 Hz, H-18), 1.16 (3H, s,
CH ), 1.01 (3H, s, CH ), 1.00 (3H, s, CH ), 0.97 (3H, s, CH ), 0.94 (3H, s, CH ), 0.92 (3H, s, CH ), 0.79
1
63.0%). H-NMR (300 MHz, CDCl ):
δ
3
(
1H, s, H of isoxazole), 5.30 (1H, brs, H-12), 4.80 (1H, t, H-3α
3
3
3
3
3
3
13
(
3H, s, CH3). C-NMR (125 MHz, CDCl3):
δ
183.8, 163.0, 161.3, 156.7, 143.6, 130.5, 129.1 (2C), 128.1,
1
2
26.9 (2C), 122.5, 107.1, 83.6, 55.3, 47.6, 46.5, 45.8, 41.6, 40.9, 39.3, 38.1, 37.0, 33.8, 33.1, 32.5, 32.4, 30.7,
8.2, 27.7, 25.9, 23.6, 23.5, 23.4, 22.9, 18.2, 17.2, 16.8, 15.4. ESI-MS: m/z [M − H] 626.4.
−
3
.4. Synthesis of Compound 12
To a stirred solution of 1a (30 mg, 0.066 mmol) in DCM (1 mL) was added bromoacetic acid
(
18 mg, 0.13 mmol), DCC (54 mg, 0.26 mmol) and DMAP (0.8 mg, 0.0066 mmol). Then, the mixture was
stirred at room temperature for 1 h. Then the precipitate was filtered and the filtrate was concentrated.
The residue was purified through a silica chromatography column (petroleum ether
−
EtOAc 20:1) to
5.27(1H, brs,
1
afford the product (12) as a white solid (35 mg, 92.0%). H-NMR (300 MHz, CDCl ):
δ
3
H-12), 4.55 (1H, t, J = 8.1 Hz, H-3α), 3.82 (2H, ABq, J = 12.3 Hz, CH ), 2.81 (1H, dd, H-18), 1.18 (3H, s,
2
CH ), 0.94 (3H, s, CH ), 0.92(3H, s, CH ), 0.90 (6H, br s, 2CH ), 0.88 (3H, s, CH ), 0.79 (3H, s, CH ).
3
3
3
3
3
3
−
ESI-MS: m/z [M + H] 577.4.

Molecules 2017, 22, 690
9 of 13
3
.5. Synthesis of Compound 3b
To a stirred solution of 12 (35 mg, 0.06 mmol) in acetone (1.5 mL) was added 2-aminothiazole
◦
(
(
18 mg, 0.18 mmol). The mixture was stirred at 50 C for 20 h, then DCM (20 mL) and NaHCO
3
aq.) (10 mL) were added. And the mixture was stirred for another 30 min at room temperature.
3). The organic layer was collected,
dried over Na SO and concentrated. The residue was purified through a silica chromatography
Then the aqueous layer was extracted with DCM (5 mL
×
2
4
column (CHCl3 MeOH
−
Et N 25:1:0.2) to afford the product (3b) as a white solid (20 mg, 55.0%).
−
3
1
H-NMR (500 MHz, C D N):
δ
6.25 (1H, d, J = 5.0 Hz, H of thiazolimine), 5.96 (1H, d, J = 5.0 Hz, H of
),
.33 (1H, d, J = 10.5 Hz, H-18), 1.24 (3H, s, CH ), 0.99 (3H, s, CH ), 0.94 (9H, br s, 3CH ), 0.85 (3H, s,
5
5
thiazolimine), 5.44 (1H, br s, H-12), 4.83 (2H, q, J = 10.2 Hz, CH ), 4.75 (1H, dd, J = 5.0, 12.0 Hz, H-3
α
2
3
3
3
3
13
CH ), 0.77 (3H, s, CH ). C-NMR (125 MHz, C D N):
δ 180.3, 168.9, 163.0, 144.9, 128.3, 122.3, 97.4,
3
3
5
5
8
2
2.1, 55.4, 47.8, 47.3, 46.7, 46.5, 42.1, 42.0, 39.7, 38.1, 38.0, 37.1, 34.2, 33.3, 33.2, 33.0, 31.0, 28.3, 28.2, 26.2,
3.8, 23.7, 23.7, 18.4, 17.4, 16.9, 15.3. ESI-MS: m/z [M − H] 595.5.
−
3
.6. Synthesis of Compound 13
To a stirred solution of
4
(40 mg, 0.07 mmol) in DCM (1 mL) was added 2c (55 mg, 0.29 mmol),
◦
EDCI (70 mg, 0.37 mmol) and DMAP (5 mg, 0.04 mmol). The mixture was stirred at 50 C for 4 h. Then
the reaction solution was concentrated. The residue was purified through a silica chromatography
column (petroleum ether
H-NMR (300 MHz, CDCl ):
−
EtOAc 20:1) to afford the product (13) as white solid (41 mg, 78.0%).
7.78 (2H, d, J = 7.5 Hz, H-2, 6 of Ph) 7.42 (3H, m, H-3, 4, 5 of Ph), 7.34
1
δ
3
(
5
1
5H, s, Bn), 7.10 (1H, d, J = 1.8 Hz, H of furan), 6.73 (1H, d, J = 1.8 Hz, H of furan), 5.30 (1H, br s, H-12),
.08 (2H, ABq, J = 12.0 Hz, CH of Bn), 4.72 (1H, t, J = 8.7 Hz, H-3 ), 2.91 (1H, d, J = 12.9 Hz, H-18),
.14 (3H, s, CH ), 0.99 (3H, s, CH ), 0.95 (6H, br s, 2CH ), 0.92 (3H, s, CH ), 0.90 (3H, s, CH ), 0.63 (3H,
α
2
3
3
3
3
3
+
s, CH ). ESI-MS: m/z [M + Na] 739.6.
3
3
.7. Synthesis of Compound 3c
To a stirred solution of 13 (41 mg, 0.057 mmol) in a mixed solvent of MeOH (0.4 mL) and
DCM (0.1 mL) was added 10% Pd/C (4 mg) and H . The mixture was stirred for 20 h at room
2
temperature, then filtered and the filtrate was concentrated. The residue was purified through a
silica chromatography column (petroleum ether
solid (29 mg, 89.6% based on 4 mg recovery of substrate). H-NMR (300 MHz, CDCl ):
−
acetone 15:1) to afford the product (3c) as a white
1
δ 7.78 (2H,
3
d, J = 7.5 Hz, H-2, 6 of Ph), 7.41 (2H, t, H-3, 5 of Ph), 7.33 (1H, t, H-4 of Ph),7.20 (1H, d, J = 3.3 Hz,
H of furan), 6.73 (1H, d, J = 3.6 Hz, H of furan), 5.29 (1H, br s, H-12), 4.74 (1H, t, H-3 ), 2.83 (1H, d,
J = 10.5 Hz, H-18), 1.16 (3H, s, CH ), 1.00 (6H, br s, 2CH ), 0.96 (3H, s, CH ), 0.94 (3H, s, CH ), 0.92
α
3
3
3
3
13
(
3H, s, CH ), 0.78 (3H, s, CH ). C-NMR (75 MHz, CDCl3): δ 184.1, 158.7, 157.4, 144.1, 143.6, 129.6,
3 3
1
3
28.8 (3C), 124.8 (2C), 122.6, 119.4, 106.7, 81.5, 55.3, 47.5, 46.5, 45.8, 41.6, 40.9, 39.3, 38.1, 37.0, 33.8, 33.1,
−
2.5, 32.4, 30.7, 28.2, 27.7, 25.9, 23.6, 23.4, 22.9, 18.2, 17.2, 16.8, 15.4. ESI-MS: m/z [M − H] 625.5.
.8. Synthesis of Compound 16
3
To a stirred solution of
4 (10 mg, 0.018 mmol) in DCM (0.4 mL) was added 2d (6 mg, 0.036 mmol),
DCC (14.8 mg, 0.072 mmol) and DMAP (0.2 mg, 0.0018 mmol). The mixture was stirred for 2 h at
room temperature. Then the precipitate was filtered and the filtrate was concentrated. The residue
was purified through a silica chromatography column (petroleum ether
−
EtOAc 20:1) to afford the
1
product (16) as a white solid (8 mg, 63.0%). H-NMR (300 MHz, CDCl ):
δ 7.29 (5H, s, Bn), 7.24 (1H, s,
3
H-4 of pyrazole), 5.96 (1H, s, H-5 of pyrazole), 5.28 (1H, br s, H-12), 5.12–5.02 (2H, ABq, J = 12.6 Hz,
CH of Bn), 4.48 (1H, t, J = 8.1 Hz, H-3 ), 4.36 (1H, dd, -N-CH2-CH-, J = 6.9, 13.5 Hz), 4.03 (1H,
-COOH), 2.90 (1H, d, J = 9.6 Hz, H-18), 2.25 (3H, s, CH of
α
2
m, -N-CH2-CH-), 3.08 (1H, m, -CH -C
H
2
3
pyrazole), 1.16 (3H, d, CH3-CH-COOH), 1.12 (3H, s, CH ), 0.91 (3H, s, CH ), 0.89 (6H, br s, 2CH ), 0.82
3
3
3
+
(
6H, br s, 2CH ), 0.60 (3H, s, CH ). ESI-MS: m/z [M + H] 697.8.
3 3

Molecules 2017, 22, 690
10 of 13
3
.9. Synthesis of Compound 3d
To a stirred solution of 16 (8 mg, 0.011 mmol) in MeOH (0.5 mL) was added 10% Pd/C (1 mg) and
H . The mixture was stirred for 2 h at room temperature. After another portion of 10% Pd/C (4 mg)
2
was added, the reaction was maintained for another 12 h, then filtered and the filtrate was concentrated.
The residue was purified through a silica chromatography column (petroleum ether
to afford the product (3d) as a white solid (4 mg, 66.0%). H-NMR (300 MHz, CDCl ):
−
acetone 15:1)
7.24 (1H,
), 4.36 (1H,
-COOH), 2.82 (1H, d, J = 10.2 Hz,
H-18), 2.25 (3H, s, CH of pyrazole), 1.16 (3H, d, CH3-CH-COOH), 1.13 (3H, s, CH ), 0.93 (6H, s,
1
δ
3
s, H-4 of pyrazole), 5.95 (1H, s, H-5 of pyrazole), 5.27 (1H, br s, H-12), 4.48 (1H, m, H-3
m, -N-CH -C-), 4.04 (1H, m, -N-CH -C-), 3.06 (1H, m, -CH -C
α
H
2
2
2
3
3
13
2
CH ), 0.90 (3H, s, CH ), 0.84 (3H, s, CH ), 0.82 (3H, s, CH ), 0.74 (3H, s, CH ). C-NMR (125 MHz,
3 3 3 3 3
CDCl3):
δ 182.9, 174.2, 148.8, 128.9, 126.8, 122.5, 104.8, 81.2, 55.3, 47.5, 46.5, 45.8, 41.6, 41.0, 39.3, 38.0,
3
[
7.0, 33.8, 33.0, 32.5, 32.4, 30.7, 28.0, 27.8, 27.7, 25.9, 23.6, 23.4, 22.9, 18.1, 17.1, 15.3, 15.1. ESI-MS: m/z
M − H]− 605.8.
3
.10. Assembling the Database of Carboxylic Acids
®
A database of all the commercial available carboxylic acids was extracted from the MDL ACD
package as an .sdf file. The file was input into Accelrys Discovery Studio 2.0 (Accelrys Inc., San Diego,
CA, USA) to calculate the molecular properties, such as Molecular Weight, H-bond Acceptor, H-bond
Donor, etc. The output file was opened in Discovery Studio Visualizer, and the undesired molecules
were filtered out manually from the file according to the criterions, which were demonstrated in
the context.
3
.11. Evaluating the Diversity of the Carboxylic Acids Database
The database file generated in the last step was input into Accelrys Discovery Studio 2.0 (Accelrys
Inc., San Diego, CA, USA) to be clustered into 285 classes using a 2D fingerprint descriptor FCFP_4.
One molecule was randomly selected from each cluster and put into an .sdf file to consist of a
sub-library consisting of 285 members.
3
.12. Docking
Autodock 4.2 was employed to screen a library of OA-containing structurally diverse
-O-esters [29]. The crystal structure of FXR in complex with its agonist fexaramine (PDB code
3
1OSH) was used as the receptor. As a positive control, fexaramine was first docked against the
FXR structure to validate the docking method. Docking was performed with a 60 60 60 grid
×
×
covering the entire ligand binding site and 10 runs per ligand were done using the Lamarckian Genetic
Algorithm (LGA) method. Other parameters were set as defaults. Top five compounds (including
positive control) from the above screening were selected and further examined using Autodock Vina
with default parameters [30]. To obtain more precise docking poses, a second run of Autodock was
performed on these selected compounds and the parameter of LGA runs was set to 100. Docking
figures were prepared using the PyMOL program (DeLano Scientific LLC: Palo Alto, CA, USA).
3
.13. Cell Culture
93T cells were maintained in MEM media with 10% fetal bovine serum (FBS), 100 U/mL
2
penicillin and streptomycin and 1 mM non-essential amino acid under humidified air containing 5%
◦
CO at 37 C.
2
3
.14. Transient Transfection and Luciferase Reporter Assay
93T cells were plated in 96-well plates at 2.5
pCMV-GAL4-DBD-hFXR-LBD expression vector, 25 ng/well pFRLuciferase and 5 ng/well
Renilla-Luciferase reporter plasmids were transiently transfected into cells using 0.25 L/well
2
×
104 per well.
Five ng/well
µ

Molecules 2017, 22, 690
11 of 13
lipofectamine (Invitrogen, Carlsbad, CA, USA). Eighteen hours after transfection, cells were
treated with CDCA at 25 M in fresh MEM containing 0.5% charcoal-stripped FBS. Chemical
µ
derivatives dissolved in dimethyl sulfoxide (DMSO) were additionally added in concentrations
indicated. Luciferase activities were measured after an additional 24 h using Dual-Luciferase
Reporter Assay System (Promega, Madison, WI, USA). Relative activity was defined as pFRLuciferase
activity/Renilla-Luciferase activity. The relative activity of FXR in the presence of 25
was set as 100%.
µM of CDCA
4
. Conclusions
In summary, a virtual screening strategy by combining 2D similarity search and molecular docking
has been applied to the discovery of new pentacyclic triterpenoid OA-based FXR antagonists. Four
OA analogs were synthesized, and two 3-O-ester compounds showed the improvement on FXR
antagonism. These results proved the feasibility of our research strategy and the current ligand-FXR
model in the further pursuit of more potent FXR interacting agents.
Supplementary Materials: Supplementary materials are available online: experimental details for the synthesis
1
13
of the known compounds 2a
, 2d, 4, 6-9. Copies of H-NMR and C-NMR spectra of the novel compounds 3a–d,
1
3, 16. Binding modes of OA (green) and 3a (yellow) into FXR.
Acknowledgments: W.-S.F. thanks the National Natural Science Foundation of China (Grant No. 81473118)
and the Hong Kong, Macao and Taiwan Science & Technology Cooperation Program of China (Grant No.
2
012DFH30030) for financial support. J.L. acknowledges the financial supported by Bureau of Science and
Information Technology of Guangzhou Municipality, China (Grant No. 2013J4500008), and the support from the
Guangzhou Branch of the Supercomputing Center of CAS.
Author Contributions: All the authors together conceived and planned the research, discussed the results and
wrote the manuscript. S.-R.W. and T.X. contributed equally to this paper.
Conflicts of Interest: The authors declare no conflict of interest.
References
1
.
Makishima, M.; Okamoto, A.Y.; Repa, J.J.; Tu, H.; Learned, R.M.; Luk, A.; Hull, M.V.; Lustig, K.D.;
Mangelsdorf, D.J.; Shan, B. Identification of a nuclear receptor for bile acids. Science 1999, 284, 1362–1365.
[CrossRef] [PubMed]
2
.
Parks, D.J.; Blanchard, S.G.; Bledsoe, R.K.; Chandra, G.; Consler, T.G.; Kliewer, S.A.; Stimmel, J.B.;
Willson, T.M.; Zavacki, A.M.; Moore, D.D.; et al. Bile acids: Natural ligands for an orphan nuclear receptor.
Science 1999, 284, 1365–1368. [CrossRef] [PubMed]
3.
4.
5.
6.
7.
Fiorucci, S.; Rizzo, G.; Donini, A.; Distrutti, E.; Santucci, L. Targeting farnesoid X receptor for liver and
metabolic disorders. Trends Mol. Med. 2007, 13, 298–309. [CrossRef] [PubMed]
Modica, S.; Murzilli, S.; Salvatore, L.; Schmidt, D.R.; Moschetta, A. Nuclear bile acid receptor FXR protects
against intestinal tumorigenesis. Cancer Res. 2008, 68, 9589–9594. [CrossRef] [PubMed]
Fiorucci, S.; Mencarelli, A.; Distrutti, E.; Palladino, G.; Cipriani, S. Targetting farnesoid-X-receptor: From
medicinal chemistry to disease treatment. Curr. Med. Chem. 2010, 17, 139–159. [CrossRef] [PubMed]
Merk, D.; Steinhilber, D.; Schubert-Zsilavecz, M. Medicinal chemistry of farnesoid X receptor ligands: From
agonists and antagonists to modulators. Future Med. Chem. 2012, 4, 1015–1036. [CrossRef] [PubMed]
Neuschwander-Tetri, B.A.; Loomba, R.; Sanyal, A.J.; Lavine, J.E.; Van Natta, M.L.; Abdelmalek, M.F.;
Chalasani, N.; Dasarathy, S.; Diehl, A.M.; Hameed, B.; et al. Farnesoid X nuclear receptor ligand obeticholic
acid for non-cirrhotic, non-alcoholic steatohepatitis (FLINT): A multicentre, randomised, placebo-controlled
trial. Lancet 2015, 385, 956–965. [CrossRef]
8
.
.
Fiorucci, S.; Distrutti, E.; Ricci, P.; Giuliano, V.; Donini, A.; Baldelli, F. Targeting FXR in cholestasis: Hype or
hope. Expert Opin. Ther. Targets 2014, 18, 1449–1459. [PubMed]
Huang, H.; Yu, Y.; Gao, Z.; Zhang, Y.; Li, C.; Xu, X.; Jin, H.; Yan, W.; Ma, R.; Zhu, J.; et al. Discovery and
optimization of 1,3,4-trisubstituted-pyrazolone derivatives as novel, potent, and nonsteroidal farnesoid X
receptor (FXR) selective antagonists. J. Med. Chem. 2012, 55, 7037–7053. [CrossRef] [PubMed]
9

Molecules 2017, 22, 690
12 of 13
1
1
0. Liu, W.; Wong, C. Oleanolic acid is a selective farnesoid X receptor modulator. Phytother. Res. 2010, 24,
69–373. [CrossRef] [PubMed]
1. Schmidt, J.; Klingler, F.M.; Proschak, E.; Steinhilber, D.; Schubert-Zsilavecz, M.; Merk, D. NSAIDs ibuprofen,
3
indometacin, and diclofenac do not interact with farnesoid X receptor. Sci. Rep. 2015, 5, 14782. [CrossRef]
[PubMed]
1
2. Chen, P.; Zeng, H.; Wang, Y.; Fan, X.; Xu, C.; Deng, R.; Zhou, X.; Bi, H.; Huang, M. Low dose of
oleanolic acid protects against lithocholic acid–induced cholestasis in mice: Potential involvement of
nuclear factor-E2-related factor 2-mediated upregulation of multidrug resistance-associated proteins.
Drug Metab. Dispos. 2014, 42, 844–852. [CrossRef] [PubMed]
1
3. Chen, P.; Li, J.; Fan, X.; Zeng, H.; Deng, R.; Li, D.; Huang, M.; Bi, H. Oleanolic acid attenuates
obstructive cholestasis in bile duct-ligated mice, possibly via activation of NRF2-MRPs and FXR antagonism.
Eur. J. Pharmacol. 2015, 765, 131–139. [CrossRef] [PubMed]
1
4. Chai, J.; Du, X.; Chen, S.; Feng, X.; Cheng, Y.; Zhang, L.; Gao, Y.; Li, S.; He, X.; Wang, R.; et al. Oral
administration of oleanolic acid, isolated from Swertia mussotii Franch, attenuates liver injury, inflammation,
and cholestasis in bile duct-ligated rats. Int. J. Clin. Exp. Med. 2015, 8, 1691–1702. [PubMed]
5. Zhao, H.L.; Cho, K.H.; Ha, Y.W.; Jeong, T.S.; Lee, W.S.; Kim, Y.S. Cholesterol-lowering effect of platycodin D
in hypercholesterolemic ICR mice. Eur. J. Pharmacol. 2006, 537, 166–173. [CrossRef] [PubMed]
6. Lin, Z.H.; Zhang, Y.; Zhang, Y.N.; Shen, H.; Hu, L.H.; Jiang, H.L.; Shen, X. Oleanolic acid derivative NPLC441
potently stimulates glucose transport in 3T3-L1 adipocytes via a multi-target mechanism. Biochem. Pharmacol.
1
1
2
008, 76, 1251–1262. [CrossRef] [PubMed]
7. Lipinski, C.A.; Lombardo, F.; Dominy, B.W.; Feeney, P.J. Experimental and computational approaches to
estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 1997
3, 3–25. [CrossRef]
8. Willett, P. Similarity-based virtual screening using 2D fingerprints. Drug Discov. Today 2006, 11, 1046–1053.
CrossRef] [PubMed]
9. Cheng, M.S.; Yan, M.C.; Liu, Y.; Zheng, L.G.; Liu, J. Synthesis of
Triterpenoid saponins bearing a unique cytotoxicity-inducing disaccharide moiety. Carbohydr. Res. 2005, 341,
0–67. [CrossRef] [PubMed]
1
,
2
1
1
[
β
-hederin and Hederacolchiside A1:
6
2
2
0. Wang, X.Z.; Jia, J.; Zhang, Y.; Xu, W.R.; Liu, W.; Shi, F.N.; Wang, J.W. Concise Synthesis and Antimicrobial
Activities of New Substituted 5-Isoxazolpenicillins. J. Chin. Chem. Soc. 2007, 54, 643–652. [CrossRef]
1. Saxena, R.; Singh, V.; Batra, S. Studies on the catalytic hydrogenation of Baylis–Hillman derivatives
of substituted isoxazolecarbaldehydes. Unusual retention of isoxazole ring during Pd–C-promoted
hydrogenation of Baylis–Hillman adducts. Tetrahedron 2004, 60, 10311–10320. [CrossRef]
2
2. Bachir, L.; D’Amour, K.; Casida, J.E. Novel and potent 6-chloro-3-pyridinyl ligands for the α4β2 neuronal
nicotinic acetylcholine receptor. J. Med. Chem. 1999, 42, 2227–2234.
2
3. Fu, J.; Liu, L.; Huang, Z.; Lai, Y.; Ji, H.; Peng, S.; Tian, J.; Zhang, Y. Hybrid molecule from
O -(2,4-dinitrophenyl)diazeniumdiolate and oleanolic acid: A glutathione S-transferase -activated nitric
π
2
oxide prodrug with selective anti-human hepatocellular carcinoma activity and improved stability.
J. Med. Chem. 2013, 56, 4641–4655. [CrossRef] [PubMed]
2
2
2
2
4. Attaran, O.S.; Baltayan, A.O.; Matsoyan, S.G. Synthesis of
-methyl- -(5-methylpyrazol-1-yl)propionic acids and their esterification with vinyl acetate. Russ. J.
Gen. Chem. 2007, 77, 297–300. [CrossRef]
α-methyl-β-(3-methylpyrazol-1-yl)-and
α
β
5. Mi, L.Z.; Devarakonda, S.; Harp, J.M.; Han, Q.; Pellicciari, R.; Willson, T.M.; Khorasanizadeh, S.; Rastinejad, F.
Structural basis for bile acid binding and activation of the nuclear receptor FXR. Mol. Cell 2003, 11, 1093–1100.
[CrossRef]
6. Soisson, S.M.; Parthasarathy, G.; Adams, A.D.; Sahoo, S.; Sitlani, A.; Sparrow, C.; Cui, J.; Becker, J.W.
Identification of a potent synthetic FXR agonist with an unexpected mode of binding and activation.
Proc. Natl. Acad. Sci. USA 2008, 105, 5337–5342. [CrossRef] [PubMed]
7. Nolte, R.T.; Wisely, G.B.; Westin, S.; Cobb, J.E.; Lambert, M.H.; Kurokawa, R.; Rosenfeld, M.G.; Willson, T.M.;
Glass, C.K.; Milburn, M.V. Ligand binding and co-activator assembly of the peroxisome proliferator-activated
receptor-γ. Nature 1998, 395, 137–143. [PubMed]

Molecules 2017, 22, 690
13 of 13
2
2
3
8. Downes, M.; Verdecia, M.A.; Roecker, A.J.; Hughes, R.; Hogenesch, J.B.; Kast-Woelbern, H.R.; Bowman, M.E.;
Ferrer, J.L.; Anisfeld, A.M.; Edwards, P.A.; et al. A chemical, genetic, and structural analysis of the nuclear
bile acid receptor FXR. Mol. Cell 2003, 11, 1079–1092. [CrossRef]
9. Morris, G.M.; Huey, R.; Lindstrom, W.; Sanner, M.F.; Belew, R.K.; Goodsell, D.S.; Olson, A.J. Autodock4 and
AutoDockTools4: Automated docking with selective receptor flexiblity. J. Comput. Chem. 2009, 16, 2785–2791.
[CrossRef] [PubMed]
0. Trott, O.; Olson, A.J. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring
function, efficient optimization and multithreading. J. Comput. Chem. 2010, 31, 455–461. [CrossRef] [PubMed]
Sample Availability: Samples of the compounds are not available from the authors.
©
2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).