Discovery of atorvastatin as a tetramer stabilizer of nuclear receptor RXRα through structure-based virtual screening

Article abstract of DOI:10.1016/j.bioorg.2019.01.007

Retinoid X receptor alpha (RXRα), a central member of the nuclear receptor superfamily and a key regulator of many signal transduction pathways, has been an attractive drug target. We previously discovered that an N-terminally truncated form of RXRα can b

Full text of DOI:10.1016/j.bioorg.2019.01.007

Accepted Manuscript
Discovery of atorvastatin as a tetramer stabilizer of nuclear receptor RXRα
through structure-based virtual screening
Xin Wang, Shuyi Chong, Huiyun Lin, Zhiqiang Yan, Fengyu Huang, Zhiping
Zeng, Xiaokun Zhang, Ying Su
PII:
S0045-2068(18)31185-4
DOI:
https://doi.org/10.1016/j.bioorg.2019.01.007
YBIOO 2711
Reference:
Bioorganic Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
17 October 2018
30 December 2018
6 January 2019
Please cite this article as: X. Wang, S. Chong, H. Lin, Z. Yan, F. Huang, Z. Zeng, X. Zhang, Y. Su, Discovery of
atorvastatin as a tetramer stabilizer of nuclear receptor RXRα through structure-based virtual screening, Bioorganic
Chemistry (2019), doi: https://doi.org/10.1016/j.bioorg.2019.01.007
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Discovery of atorvastatin as a tetramer stabilizer of nuclear receptor RXRα
through structure-based virtual screening
Xin Wang a, 1, Shuyi Chong a, 1, Huiyun Lin a, Zhiqiang Yan a, Fengyu Huang a,
Zhiping Zeng a, Xiaokun Zhang a, b *, Ying Su b *
a
School of Pharmaceutical Science, Fujian Provincial Key Laboratory of
Innovative Drug Target Research, Xiamen University, Fujian, 361002, China
b
Sanford Burnham Prebys Medical Discovery Institute, 10901 N. Torrey Pines
Road, La Jolla, CA 92037, USA
1 These authors contributed equally to this work.
* To whom correspondence should be addressed:
Email: ysu@sbpdiscovery.org (Y. Su) or to xkzhang@xmu.edu.cn (X. -K.
Zhang)
Key words: RXRα, RXRα tetramer, RXRα ligands, Virtual screening,
Conformational selection
Abstract
Retinoid X receptor alpha (RXRα), a central member of the nuclear receptor
superfamily and a key regulator of many signal transduction pathways, has been
an attractive drug target. We previously discovered that an N-terminally
truncated form of RXRα can be induced by specific ligands to form
homotetramers, which, as a result of conformational selection, forms the basis
for inhibiting the nongenomic activation of RXRα. Here, we report the
identification and characterization of atorvastatin as a new RXRα tetramer
stabilizer by using structure-based virtual screening and demonstrate that virtual
library screening can be used to aid in identifying RXRα ligands that can induce
its tetramerization. In this study, docking was applied to screen the
FDA-approved small molecule drugs in the DrugBank 4.0 collection. Two
compounds were selected and purchased for testing. We showed that the

selected atorvastatin could bind to RXRα to promote RXRα-LBD
tetramerization. We also showed that atorvastatin possessed RXRα-dependent
apoptotic effects. In addition, we used a chemical approach to aid in the studies
of the binding mode of atorvastatin.
1. Introduction
Retinoid X receptor alpha (RXRα), a unique member of the nuclear
receptor superfamily, regulates a broad spectrum of physiological functions
including cell differentiation, growth, and apoptosis, and is implicated in many
diseases such as cancer, metabolic disorders and neurodegenerative diseases
[1-5]. Thus, RXRα has been an attractive drug target, especially for anticancer
therapy [1, 6, 7]. Similar to other nuclear receptors, structurally RXRα
possesses three main functional domains: a disordered N-terminal region, a
DNA-binding domain, and a ligand-binding domain (LBD). The LBD is
characterized by a canonical ligand-binding pocket (LBP), a transactivation
function domain 2, a coregulator-binding surface groove, and a dimerization
surface. A well-accepted mechanism of RXRα action as a transcription factor is
that RXRα acts as homodimers or as heterodimers partnering with many other
nuclear receptors as ligand-mediated transcription factors through binding to
specific DNA-response elements of the target genes [3, 8]. Ligand binding to
the LBP induces a conformational change that triggers a cascade of events and
lead to biological activities. Many natural and synthetic ligands have been
discovered for this canonical LBP [9]. Targretin (bexarotene), a selective
LBP ligand of RXRα, was approved for treating human cutaneous T-cell
lymphoma [10, 11].
Besides functioning as a transcriptional factor, RXRα also play important
extranuclear (nongenomic) roles through transcription-independent mechanisms
[12-16]. RXR migrates from the nucleus to the cytoplasm at different stages
during development [17] and in response to differentiation [14], survival [18,

19], apoptosis [12], and inflammation [15, 16, 18, 19]. Studies from our
laboratory show that RXR can act in the cytoplasm to cross-talk with
important biological processes such as mitochondria-dependent apoptotic
pathway [12, 20], phosphatidylinositol 3-kinase (PI3K)/AKT-mediated cell
survival pathway [19, 21], and NF-B-mediated inflammatory pathway [22].
We have previously reported that RXR is abnormally cleaved in various types
of cancer cells, producing an N-terminally-truncated RXR (tRXR) protein.
We showed that tRXR resides in cytoplasm and is oncogenic in tumor cells.
tRXR acts to promotes phosphoinositide 3-kinase (PI3K)/AKT activation and
enhance tumor cell growth via interacting with the p85 regulatory subunit of
PI3K. Thus, molecules that can bind to RXR to modulate its interaction with
p85 may have therapeutic potential. Along this line, we have identified
K-80003 (Fig. S1A), a designed analog of the non-steroidal anti-inflammatory
drug (NSAID) Sulindac, as a promising anti-cancer agent. K-80003 induces
apoptosis and inhibits the tRXR-mediated PI3K/AKT survival pathway by
binding to tRXR and disrupting the interaction between tRXR and p85
[19].F
Our recent studies of the molecular mechanism of K-80003 reveal that
K-80003 utilizes conformational selection as a mechanism to inhibit the
nongenomic function of tRXR [23]. Besides functioning in the forms of
homodimers and heterodimers, RXR can also form tetramers. We demonstrate
that K-80003 binding promotes tRXR tetramerization which results in
apoptosis via the inhibition of tRXR interaction with p85 [23]. Crystal

structure of the RXR-LBD/K-80003 complex shows that the
RXR-LBD/K-80003 complex exists as a tetramer forming 2 large symmetric
hydrophobic cavities where 3 molecules of K-80003 are bound per cavity. The
fact that the cavity as a binding region can simultaneously accommodate 3
K-80003 molecules inspired us to ask if a molecule larger than K-80003 could
be identified to mimic the binding of the 3 K-80003 molecules while retaining
similar or better biological activities [21, 23].
Structure-based virtual screening is a powerful approach in drug discovery
where the three-dimensional structure of the protein target is available [24]. In
our previous work, we successfully discovered the first small molecule targeting
the coregulator-binding site of RXRα using this approach [25]. Thus, we
employed structure-based virtual screening to identify RXRα modulators that
can bind and stabilize RXR-LBD tetramers. As a pilot exercise we used an
FDA-approved drug collection as the first screening library. Here we report the
identification and characterization of drug atorvastatin as a non-canonical ligand
of RXRα that stabilizes the RXRα tetrameric conformation to regulate RXRα
nongenomic actions.
2. Results and discussion
2.1. Structure-based virtual screening
In this study we screened an FDA-approved drug collection of 1908
compounds downloaded from DrugBank 4.0 [26]. The collection was first
prepared by LigPrep [27] module in Maestro 10.5 and was then converted from
2D to 3D with conformation energy minimized using the OPLS3 force field.
The crystal structure of RXRα-LBD tetramer retrieved from the Protein Data
Bank (www.rcsb.org) (PDB code: 5TBP) was used for the docking study. The
protein structure was prepared using Protein Preparation Wizard module [28] in

Maestro 10.5, during which hydrogen atoms were added and crystallographic
water molecules were removed. Missing side chains and loops were built using
Prime [29] in Maestro 10.5. The binding site was defined based on the positions
of all three bound K-80003 molecules (Fig.S1B), around which the 3D grid box
was generated in a size of 20 Å per dimension for docking. Glide [30] in
Maestro 10.5 was used for generating the grids and carrying out the docking
studies. All compounds were docked using the standard precision (SP) mode for
scoring. The virtual screening process is summarized in Fig.1A.
Fig.1. Virtual Screening. (A) Structure-based virtual screening process. (B)
Chemical structure of atorvastatin and lercanidipine.
2.2. Selection of compounds
There were 1736 molecules that could dock to the binding site. The first 10
compounds (Table S1) with the highest docking scores were selected for further
evaluation using the following criteria: 1. How well the docked molecule
interacts with the protein; 2. If the docked molecule occupies a reasonable size
of the binding site; 3. If the compound is commercially available. As a result, 2
compounds, atorvastatin and lercanidipine were selected and purchased from
commercial suppliers (Fig.1B). Both purchased compounds were confirmed
with high-resolution mass spectrometry (HRMS).
2.3. Biological evaluation
2.3.1. Luciferase reporter assay evaluation

Luciferase reporter assay was first used to evaluate our docking results
(Fig.2A). Gal4-RXRα-LBD strongly activates the Gal4 reporter in the presence
of 9-cis-RA, which is inhibited by UVI-3003 [31], a known RXRα antagonist.
We expected that atorvastatin and lercanidipine would act as antagonists in this
assay if they could stabilize the tetrameric form of RXRα-LBD. Indeed, both
atorvastatin and lercanidipine could inhibit the 9-cis-RA-induced Gal4 reporter
activity (Fig. 2A) and consistently the compounds did not show any agonist
activity at the concentrations used (Fig.S2). In addition, atorvastatin, but not
lercanidipine, showed dose-dependent inhibitory effect. Thus, the reporter assay
confirmed that atorvastatin could bind to RXRα-LBD.
Fig. 2. Binding Evalution. (A) Inhibition of 9-cis-RA-induced Gal4 reporter
activity. pBind-RXRα-LBD and pG5luc were transiently transfected into 293T
cells. Cells were treated with 9-cis-RA (1 μM) in the presence of UVI-3003(1
μM), atorvastatin and lercanidipine (1 μM, 10 μM, 50 μM). Relative luciferase
(LUC) activity was determined. (B) Fluorescence quenching effect of
atorvastatin on RXRα-LBD. RXRα-LBD: 1 μM, atorvastatin: 0, 1, 3, 5 10, 15,
20, 30, 40 μM.
2.3.2. Binding evaluation by fluorescence quenching assay
To further confirm atorvastatin binds directly to RXRα, fluorescence
quenching was conducted to analyze the binding affinity of atorvastatin to
RXRα-LBD. Proteins possess intrinsic fluorescence mainly because of the
aromatic amino acid residues, such as tryptophan and phenylalanine. The
intrinsic fluorescence produced by these amino acid residues are highly

sensitive to their local environment, that can be used to measure the binding
ability of small molecules. In our work, there are Trp305, Phe439, Phe439 and
Phe313 in the binding site of the RXRα tetramer, giving us the opportunity to
use this method to evaluate the binding affinity of atorvastatin. Mixtures of
RXRα-LBD and atorvastatin were analyzed by scanning fluorescence emission
between 290 and 450 nm that were stimulated by an excitation wavelength of
280ꢀnm. The results showed gradual fluorescence quenching when RXRα-LBD
was exposed to different concentrations of atorvastatin (Fig.2B), suggesting that
there were interactions between atorvastatin and RXRα-LBD. To calculate the
Kd value, we used a total of 8 different compound concentrations from 1 μmol/L
to 40 μmol/L to plot the fluorescence emission at 330 nm as a function of
compound concentrations and then we performed a non-linear fitting of the
curve [32] (Fig.S3). The results showed the Kd for the RXRα-LBD/atorvastatin
complex is 10.34 ×10-6 M.
2.3.3. Atorvastatin induces RXRα-LBD tetramerization
We then asked if atorvastatin binding to RXRα can promote RXRα-LBD
tetramerization as K-80003 does [23]. In non-denaturing polyacrylamide gel
electrophoresis, purified RXRα-LBD protein in the absence of ligand existed
as two distinct bands corresponding to homodimeric and homotetrameric
RXRα-LBD respectively (Fig. 3). 9-cis-RA binding induces homodimerization
and shifts the dimers/tetramers equilibrium to dimers (Fig. 3). As expected,
atorvastatin induced RXRα-LBD tetramerization as K-80003 (Fig. 3).
Furthermore, when incubated RXRα-LBD with 9-cis-RA and atorvastatin
together, atorvastatin could competitively inhibit 9-cis-RA-induced
dimerization to promote RXRα tetramerization (Fig. 3).

Fig.3. Induction of RXRα-LBD tetramerization by atorvastatin. Equal amount
of purified RXRα-LBD was incubated with K-80003, 9-cis-RA, DMSO,
atorvastatin, 9-cis-RA/K-80003 or 9-cis-RA/atorvastatin and separated by
non-denaturing polyacrylamide gel electrophoresis followed by Coomassie
Bright Blue staining. The percentage of tetramer and dimer of RXRα-LBD was
quantitated by densitometric analysis of the corresponding blots.
2.3.4. Atorvastatin displays apoptotic effects through RXRα binding
K-80003 is a promising anticancer agent, acting by inhibiting the
RXRα-activated PI3K/AKT survival pathway and inducing apoptosis [19].
Therefore, we next determined whether atorvastatin had the same anticancer
effect and if the effect was RXRα-dependent. As shown in Fig. 4, treatment of
MCF-7 cancer cell lines with atorvastatin alone or in combination with TNF-α
effectively induced PARP cleavage, an indication of apoptosis in cancer cells.
To determine if the effect of atorvastatin on inducing PARP cleavage depended
on the expression of RXRα, MCF-7 cells were transfected with RXRα siRNA
and evaluated for the apoptotic effect of atorvastatin. The result showed that
transfection of RXRα siRNA significantly reduced the level of RXRα in MCF-7
cells and weakened the atorvastatin-induced PARP cleavage (Fig. 4). These
results demonstrated that RXRα played a crucial role in mediating the apoptotic
effect of atorvastatin in cancer cells.

Fig.4. Atorvastatin induces apoptosis of cancer cells in a RXRα-dependent
manner. RXRα siRNA transfection inhibits the apoptotic effect of atorvastatin.
MCF-7 cells transfected with control or RXRα siRNA for 48 hr were treated
with atorvastatin (50 μM), K-80003 (30 μM) and/or TNF-α (20 ng/mL) for 12
hr and analyzed by immunoblotting.
2.4. Binding mode exploration
Besides the crystal structure of the tetrameric RXRα-LBD/K-80003
complex (PDB code: 5TBP), previously we also determined another crystal
structure of RXRα-LBD tetramer in complex with a different ligand, K-8008
(PDB code: 4N8R). Although RXRα ligands K-8008 and K-80003 are
structurally similar (K-8008 was designed based on the bioisostere concept by
replacing the carboxylic acid in K-80003 with tetrazole) (Fig. S1A) and their
complexes with RXRα-LBD adopt a tetrameric structure with each
ligand-bound tetramer possessing 2 large symmetry-related interfacial cavities,
the binding mode of K-80003 is different from the binding mode of K-8008 (Fig.
S1B-C). In the complex structure of RXRα-LBD with K-80003, three molecules
of K-80003 bind to the same large hydrophobic cavity, whereas only one
molecule of K-8008 is bound in the crystal structure of RXRα-LBD/K-8008
complex (Fig. S1B-C) [21, 23]. Furthermore, we found that the orientations of
Phe439 display large difference (Fig. S1D) in the area where ligand K-8008 or
K-80003 binds. It is well known that docking results are influenced by many
factors including docking methods and orientations of side chains in the binding
site [33, 34]. Therefore, to explore the binding model, we docked atorvastatin to
both crystal structures (4N8R and 5TBP). The results proposed 2 different

binding models (Fig.5). In model A (Fig.5A), the docking score was -9.780 kcal
mol-1, and atorvastatin interacted with the protein primarily via forming
H-bonds with residues Arg316 and Ala327 and making van der Waals contacts
with Trp305, Leu309, Leu326, Phe438, Phe439 and Leu436. The carboxylic
acid group in atorvastatin appeared to be important for binding. In model B
(Fig.5B), the docking score is -9.816 kcal mol-1, and the carboxylic acid group
of atorvastatin formed a H-bond with Gln275 and the three benzene rings on
atorvastatin made hydrophobic interactions with the surrounding hydrophobic
amino acids including π-π interaction with Trp305.
Fig.5. The possible binding modes of atorvastatin. (A) The predicted binding
model A. (B) The predicted binding model B. Atorvastatin is shown in pink
sticks, and the protein is shown in gray cartoon ribbon, key residues are shown
in cyan sticks, and the potential H-bonds are shown in yellow dashes.
To help determine which model is more reasonable, four derivatives of
atorvastatin with chemical modification of the carboxylic acid group were
designed and synthesized. Methyl esterification, ethyl esterification, butyl
esterification and hexyl esterification of the carboxylic acid were carried out to
generate compounds 2a, 2b, 2c and 2d respectively using scheme 1. Model A
showed that derivatization of the -COOH group not only would disrupt the
H-bonds but also cause steric hindrance. Thus, we expected that esterification of
the carboxylic acid group would weaken the binding of atorvastatin. In model B,
there is still some unoccupied hydrophobic space around the carboxylic acid

group, and alkylation modification of the carboxyl group would not necessarily
lead to the weakening of the binding. Therefore, ester compounds would aid in
assessing if model A or model B is the possible binding model for atorvastatin.
Scheme 1. Reagents and conditions: (a) ROH, H2SO4, r. t., 2.5h; (b) ROH,
CH2Cl2, H2SO4, r. t., 2.5h.
Luciferase reporter assay was used to evaluate the binding of these four
derivatives (Fig.6). Results showed that all 4 ester compounds (2a, 2b, 2c and
2d) inhibited the 9-cis-RA-induced Gal4 reporter activity more strongly than
atorvastatin and compounds (2b and 2c) worked better than compounds (2a and
2d). These results implied that the aforementioned binding model B of
atorvastatin was more reasonable. This is because the binding model A
anticipated that derivatives 2a-d would display weaker inhibition of the
9-cis-RA-induced Gal4 reporter activity, whereas the binding model B
supported that derivatives 2a-d could interact more strongly with the protein
and display stronger inhibition of the 9-cis-RA-induced Gal4 reporter activity
compared to atorvastatin.
Fig. 6. Inhibition of the 9-cis-RA-induced Gal4 reporter activity.
pBind-RXRα-LBD and pG5luc were transiently transfected into 293T cells.

Cells were treated with 9-cis-RA (1 μM) in the presence of UVI-3003 (1 μM),
atorvastatin, 2a, 2b, 2c and 2d (20 μM). Relative luciferase (LUC) activity was
determined.
Superimposition of the binding mode of atorvastatin in model B with the
K-80003-bound RXRα-LBD structure showed that atorvastatin occupied the
space taken by two of the three K-80003 molecules (Fig. 7A). Visual
examination revealed that the unoccupied hydrophobic region near Gln275 (Fig.
7B) could accommodate the alkyl group in compounds 2a, 2b, 2c and 2d, which
would support the results that 2a, 2b, 2c and 2d displayed stronger inhibition of
the 9-cis-RA-induced Gal4 reporter activity than atorvastatin. Consistently
docking results showed that the alkoxy moiety in 2a, 2b, 2c or 2d bound to the
unoccupied hydrophobic region next to Gln275 (Fig.7B), resulting in tighter
binding. In summary, the data suggested that atorvastatin could bind to the
RXRα tetramer via the binding model B shown in Fig. 5B.
Fig. 7. Binding mode exploration. (A) The orthogonal view of RXRα tetramer

showing one of the hydrophobic symmetric voids (semitransparent grey) at the
interface between two dimers. 3 bound K-80003 molecules are shown as blue
balls. Docked atorvastatin is shown as red balls. (B) Docking results of 2a, 2b,
2c and 2d. Atorvastatin is shown in red sticks, 2a is shown in wheat sticks, 2b is
shown in yellow sticks, 2c is shown in green sticks and 2d is shown in purple
sticks. Binding cavity is shown in semitransparent grey surface.
3. Conclusion
Atorvastatin, also known as Lipitor, is a synthetic HMG-CoA reductase
inhibitor which has been widely used for lowering cholesterol. Here, we report
that atorvastatin could target the binding site of the tetrameric RXRα-LBD via
structure-based virtual screening. Further experiments confirmed that
atorvastatin could induce RXRα-LBD tetramerization and had potential
anticancer effects. In addition, we explored the binding model of atorvastatin
through docking studies, chemical synthesis and biological evaluation. Our
results demonstrated that molecules such as atorvastatin, larger than K-80003,
can mimic 2 or more copies of K-80003 and occupy the large cavity in the
RXR-LBD tetramer to stabilize RXR-LBD tetramer. Virtual screening is a
feasible approach for discovering such molecules as novel RXR ligands.
Furthermore, out results imply that atorvastatin could be optimized as a lead
compound targeting RXR for anticancer therapy.
4. Experimental section
4.1. Virtual screening
Schrodinger suite (Schrödinger, LLC, New York, NY, 2016) was employed
in our structure-based virtual screening. The 2D structures of the compounds
were downloaded from DrugBank (www.drugbank.ca) and transformed to 3D
by LigPrep module in Maestro 10.5. The crystal structure of RXRα tetramer
was retrieved from the Protein Data Bank (www.rcsb.org) (PDB code: 5TBP)

and was prepared by Protein Preparation Wizard module in Maestro 10.5.
Molecular docking was performed by Glide module in Maestro 10.5. SP
(standard precision) scoring was used. 5 poses per ligand were performed to
post-docking minimization and the best of these 5 poses was collected. After
docking, the top 10 compounds with the highest docking scores were selected
for further evaluation using the following criteria: 1. How well the docked
molecule interacts with the protein; 2. If the docked molecule occupies a
reasonable size of the binding site; 3. If the compound is commercially
available. Based on these 3 criteria, 2 compounds were selected and purchased
for biological testing.
4.2. Biology
4.2.1. Cell Culture and Transfection
Human breast cancer cell line MCF-7 were cultured in MEM medium
supplemented with 10% fetal bovine serum. These cells were maintained at 5%
CO2 at 37 ℃. Subconfluent cells with exponential growth were used throughout
the experiments. Cell transfections were carried out by using Lipofectamine
2000 (Invitrogen) according to the instructions of the manufacturer.
4.2.2. Dual-Luciferase Reporter Assay
The experiments were performed as described previously [21]. Briefly, cells
were seeded at a density of 3×103 cells/well in a 48-well plate. Cells were
transfected with the corresponding plasmids for 24 hours and then treated with
compounds for 18 hours. Cells were lysed and luciferase relative activity was
tested by the Dual-Luciferase Reporter Assay System according to the
manufacturer’s instructions. Transfection efficiency was normalized to Renilla
luciferase activity.
4.2.3. Western Blotting
The experiments were performed as described previously [21]. Briefly, cells

were seeded at a density of 5×104 cells/well in a 12-well plate. Cells were
transfected with siRNA control or siRXR for 48 hours and then treated with
compounds for 12 hours. After treatment, cell lysates were prepared using
NP-40 buffer. Cell lysates (determined by the Bradford protein assay) were
boiled in sodium dodecyl sulfate (SDS) sample loading buffer, 20 µg of protein
extracts from compounds-treated MCF-7 cells resolved by 10%
SDS–polyacrylamide gel electrophoresis (SDS–PAGE) and transferred to
nitrocellulose. The membranes were blocked in 5% milk in Tris-buffered saline
and Tween 20 (TBST; 10 mM Tris–HCl [pH 8.0], 150 mM NaCl, 0.05% Tween
20) for 1 h at room temperature. After washing twice with TBST, the
membranes were incubated with appropriate primary antibodies in TBST for 1 h
and then washed thrice at room temperature, probed with horseradish
peroxide-linked anti-immunoglobulin. After three washes with TBST,
immunoreactive products were visualized using enhanced chemiluminescence
reagents and autoradiography.
4.2.4. Antibody and reagents
PARP (Cat. 9542) was purchased from Cell Signal Technology (Beverly, MA,
USA). RXRα (D-20) (Cat. sc- 553) and GAPDH (Cat. 47724) were purchased
from Santa Cruz Biotechnology (Santa Cruz, CA, USA). SiRNA control and
SiRXR were from Sigma.
4.2.5. Protein expression and purification
The human RXRα-LBD (223-462) was cloned as an N-terminal
histidine-tagged fusion protein in pET15b expression vector and overproduced
in Escherichia coli BL21 DE3 strain. Briefly, cells were harvested and sonicated,
and the extract was incubated with the His60 Ni Superflow resin. The
protein-resin complexes were washed and eluted with imidazole. The eluent was
collected and concentrated to 5 mg/mL for subsequent trials.
4.2.6. Non-denaturing gel electrophoresis

Purified RXRα-LBD protein (0.2 mg mL-1) was incubated with DMSO,
9-cis-RA (0.5 mM), and/or K-80003, atorvastatin (20 mM) for 12 h at 4 ℃ in a
total volume of 20 mL, and proteins were separated by 8% non-denaturing
PAGE followed by Coomassie Bright Blue staining.
4.2.7. Fluorescence measurements
Fluorescence measurements were performed on Agilent Cary Eclipse
Fluorescence Spectrophotometer using 10 mm quartz cuvette. The protein
concentration of 1 mM was used for RXRα-LBD. Working solutions of
compounds were made in 1 mM DMSO, and used in a concentration range of 1
to 40 μM. Protein was excited at 280 nm and the emission spectra were
recorded between 290 and 450 nm at 25 ℃ using slits with a 5 nm band pass for
excitation and emission, respectively. Fluorescence data were fitted to binding
curves using the methods reported for dissociation constant (Kd) calculation
[32]. All experiments were performed in triplicates and data were processed
using the software Origin 2016.
4.3. Chemistry
4.3.1. General information
All commercially available starting materials and solvents were reagent
grade and were purchased from Energy-chemical and used without further
purification. Atorvastatin sodium and Lercanidipine hydrochloride were
purchased from Hubei Jusheng Technology Co. Ltd. and its purity was 98%.
13
1H and C NMR were recorded on a Bruker Spectrospin DPX 600 MHz
and Bruker Spectrospin DPX 151 MHz spectrometer, respectively using CDCl3
as a solvent and trimethylsilane (TMS) as the internal standard. Splitting
patterns are designated as follows; s = singlet; d = doublet; t = triplet; m =
multiplet; br = broad; J = coupling constant in hertz (Hz). Chemical shift values
are given in ppm. HRMS were recorded by ESI-MS (Thermo Scientific Q

Exactive).
4.3.2. General procedure for synthesis of 2a-2b
To a stirred solution of atorvastatin sodium (58 mg, 0.1 mmol) in ROH (2
mL) was added concentrated sulfuric acid (30 μL) in dropwise at 0 °C over 5
min. The resulting reaction mixture was stirred at room temperature for 2.5 h.
Then 10% NaHCO3 aq. (5 mL) was added to quench the reaction. The resulting
mixture was extracted with ethyl acetate. The organic phase was washed with
saturated brine. The combined organic phase was dried over anhydrous MgSO4,
and the solvent was removed under vacuum. The residue was purified by flash
chromatography using gradient 12–66% ethyl acetate in hexanes to afford
compound.
4.3.2.1.
Methyl
(3R,5R)-7-(2-(4-fluorophenyl)-5-isopropyl-3-phenyl-4-(phenylcarbamoyl)-1H-
pyrrol-1-yl)-3,5-dihydroxyheptanoate (2a).
1
White solid, yield 88%. H NMR (Fig. S4, 600 MHz, CDCl3) δ 7.13 - 7.21
(m, 9H), 7.06 (d, J = 7.89 Hz, 2H), 6.97 - 7.02 (m, 3H), 6.86 (s, 1H), 4.08 - 4.19
(m, 2H), 3.91 - 3.97 (m, 1H), 3.73 - 3.77 (m, 1H), 3.71 (s, 3H), 3.69 (br. s., 1H),
3.63 (br. s., 1H), 3.57 (td, J = 7.13 and 14.17 Hz, 1H), 2.41 (d, J = 6.05 Hz, 2H),
1.61 - 1.71 (m, 2H), 1.54 (d, J = 6.97 Hz, 6H), 1.43 - 1.50 (m, 1H), 1.24 - 1.29
13
1
(m, 1H). C NMR (Fig. S4, 151 MHz, CDCl3) δ 173.2, 165.0, 162.4 (d, JC–F
=
3
247.6 Hz), 141.6, 138.5, 134.7, 133.3(d, JC–F = 7.7 Hz), 130.6, 128.8, 128.8,
2
128.5, 128.4, 126.7, 123.7, 121.9, 119.7, 115.5(d, JC–F = 22 Hz), 115.4, 69.8,
69.1, 52.1, 41.8, 41.4, 41.2, 39.2, 26.3, 21.9, 21.8. HRMS (ESI): m/z [M + H]+
calcd for C34H38FN2O5, 573.2759; found, 573.2756.
4.3.2.2.
Ethyl
(3R,5R)-7-(2-(4-fluorophenyl)-5-isopropyl-3-phenyl-4-(phenylcarbamoyl)-1H-
pyrrol-1-yl)-3,5-dihydroxyheptanoate (2b).
1
White solid, yield 68%. H NMR (Fig. S5, 600 MHz, CDCl3) δ 7.14 - 7.21

(m, 9H), 7.06 (d, J = 7.89 Hz, 2H), 6.96 - 7.02 (m, 3H), 6.86 (s, 1H), 4.09 - 4.19
(m, 4H), 3.91 - 3.97 (m, 1H), 3.72 - 3.77 (m, 1H), 3.70 (br. s., 1H), 3.62 (s, 1H),
3.58 (td, J = 7.15 and 14.31 Hz, 1H), 2.40 (d, J = 6.05 Hz, 2H), 1.63 - 1.69 (m,
13
2H), 1.54 (d, J = 7.15 Hz, 6H), 1.44 - 1.50 (m, 1H), 1.25 - 1.28 (m, 4H). C
1
NMR (Fig. S5, 151 MHz, CDCl3) δ 172.8, 165.0, 162.4(d, JC–F = 247.6 Hz),
3
141.7, 138.5, 134.8, 133.3(d, JC–F = 8.8 Hz), 130.6, 128.9, 128.8, 128.5, 128.5,
126.7, 123.6, 122.0, 119.7, 115.5(d, 2JC–F = 20.9 Hz), 69.8, 69.1, 61.1, 41.9, 41.4,
39.2, 26.3, 21.9, 21.8, 14.3. HRMS (ESI): m/z [M + H]+ calcd for C35H40FN2O5,
587.2916; found, 587.2915.
4.3.3. General procedure for synthesis of 2c-2d
To a stirred solution of atorvastatin sodium (58 mg, 0.1 mmol) in
dichloromethane (2 mL) and ROH (1 mL) was added concentrated sulfuric acid
(40 μL) in dropwise at 0 °C over 5 min. The resulting reaction mixture was
stirred at room temperature for 2.5 h. Then 10% NaHCO3 aq. (5 mL) was added
to quench the reaction. The resulting mixture was extracted with ethyl acetate.
The organic phase was washed with saturated brine. The combined organic
phase was dried over anhydrous MgSO4, and the solvent was removed under
vacuum. The residue was purified by flash chromatography using gradient
12–66% ethyl acetate in hexanes to afford compound.
4.3.3.1.
Butyl
(3R,5R)-7-(2-(4-fluorophenyl)-5-isopropyl-3-phenyl-4-(phenylcarbamoyl)-1H-
pyrrol-1-yl)-3,5-dihydroxyheptanoate (2c).
White solid, yield 59%. 1H NMR (Fig. S6, 600 MHz, CDCl3)  7.13 - 7.23
(m, 9H), 7.07 (d, J = 7.89 Hz, 2H), 6.96 - 7.03 (m, 3H), 6.86 (s, 1H), 4.08 - 4.18
(m, 4H), 3.90 - 3.98 (m, 1H), 3.72 - 3.77 (m, 1H), 3.68 (br. s., 1H), 3.60 (br. s.,
1H), 3.54 - 3.59 (m, 1H), 2.40 (d, J = 6.05 Hz, 2H), 1.66 - 1.71 (m, 1H), 1.59 -
1.64 (m, 3H), 1.54 (d, J = 7.15 Hz, 6H), 1.43 - 1.51 (m, 1H), 1.33 - 1.41 (m, 2H),
13
1.25 - 1.28 (m, 1H), 0.93 (t, J = 7.43 Hz, 3H). C NMR (Fig. S6, 151 MHz,

1
CDCl3)  172.9, 164.9, 162.4 (d, JC–F = 248.7 Hz), 141.7, 138.5, 134.8, 133.3
3
(d, JC–F = 7.7 Hz), 130.6, 128.9, 128.8, 128.5, 126.7, 123.6, 122.0, 119.7, 115.5
2
(d, JC–F = 22 Hz), 69.8, 69.2, 65.0, 41.9, 41.4, 41.4, 39.2, 30.6, 26.3, 21.9, 21.8,
19.2, 13.8. HRMS (ESI): m/z [M + H]+ calcd for C37H44FN2O5, 615.3229; found,
615.3227.
4.3.3.2.
Hexyl
(3R,5R)-7-(2-(4-fluorophenyl)-5-isopropyl-3-phenyl-4-(phenylcarbamoyl)-1H-
pyrrol-1-yl)-3,5-dihydroxyheptanoate (2d).
Colorless oil, yield 70%. 1H NMR (Fig. S7, 600 MHz, CDCl3)  7.13 - 7.21 (m,
9H), 7.06 (d, J = 7.70 Hz, 2H), 6.96 - 7.02 (m, 3H), 6.86 (s, 1H), 4.08 - 4.18 (m,
4H), 3.91 - 3.97 (m, 1H), 3.72 - 3.77 (m, 1H), 3.69 (br. s., 1H), 3.61 (br. s., 1H),
3.55 - 3.59 (m, 1H), 2.40 (d, J = 6.05 Hz, 2H), 1.67 - 1.71 (m, 1H), 1.60 - 1.65
(m, 3H), 1.54 (d, J = 7.15 Hz, 6H), 1.44 - 1.51 (m, 1H), 1.26 - 1.35 (m, 7H),
13
0.89 (t, J = 6.69 Hz, 3H). C NMR (Fig. S7, 151 MHz, CDCl3)  172.9, 164.9,
162.4 (d, 1JC–F = 247.6 Hz), 141.7, 138.5, 134.8, 133.3 (d, 3JC–F = 8.8 Hz), 130.6,
128.9, 128.8, 128.5, 126.7, 123.6, 122.0, 119.7, 115.5 (d, 2JC–F = 20.9 Hz), 69.8,
69.1, 65.3, 41.9, 41.4, 41.4, 39.2, 31.5, 28.6, 26.3, 25.6, 22.6, 21.9, 21.8, 14.1.
HRMS (ESI): m/z [M + H]+ calcd for C39H48FN2O5, 643.3542; found, 643.3539.
Acknowledgements
This work was supported by the grants from the National Nature Science Fund
of China (NSFC-31271453, NSFC-31471318, NSFC-91129302 and
NSFC-81301888), the Fundamental Research Funds for the Central Universities
(2013121038, 20720180052), the Natural Science Foundation of Fujian
Province of China (No. 2018J01133), and the US National Institutes of Health
(CA198982)

References
[1] L. Altucci, M.D. Leibowitz, K.M. Ogilvie, A.R. de Lera, H. Gronemeyer,
RAR and RXR modulation in cancer and metabolic disease, Nat Rev
Drug Discov 6(10) (2007) 793-810.
[2] R.M. Evans, D.J. Mangelsdorf, Nuclear Receptors, RXR, and the Big
Bang, Cell 157(1) (2014) 255-266.
[3] P. Lefebvre, Y. Benomar, B. Staels, Retinoid X receptors: common
heterodimerization partners with distinct functions, Trends Endocrin Met
21(11) (2010) 676-683.
[4] A. Szanto, V. Narkar, Q. Shen, I.P. Uray, P.J.A. Davies, L. Nagy,
Retinoid X receptors: X-ploring their (patho)physiological functions, Cell
Death Differ 11 (2004) S126-S143.
[5] P.E. Cramer, J.R. Cirrito, D.W. Wesson, C.Y.D. Lee, J.C. Karlo, A.E.
Zinn, B.T. Casali, J.L. Restivo, W.D. Goebel, M.J. James, K.R. Brunden,
D.A. Wilson, G.E. Landreth, ApoE-Directed Therapeutics Rapidly Clear
beta-Amyloid and Reverse Deficits in AD Mouse Models, Science
335(6075) (2012) 1503-1506.
[6] Y. Su, Z.P. Zeng, Z.W. Chen, D. Xu, W.D. Zhang, X.K. Zhang, Recent
Progress in the Design and Discovery of RXR Modulators Targeting
Alternate Binding Sites of the Receptor, Curr Top Med Chem 17(6) (2017)
663-675.

[7] X.K. Zhang, H. Zhou, Y. Su, Targeting truncated RXR alpha for
cancer therapy, Acta Bioch. Bioph. Sin. 48(1) (2016) 49-59.
[8] P.F. Egea, A. Mitschler, N. Rochel, M. Ruff, P. Chambon, D. Moras,
Crystal structure of the human RXR alpha ligand-binding domain bound
to its natural ligand: 9-cis retinoic acid, Embo J 19(11) (2000) 2592-2601.
[9] M. Dominguez, S. Alvarez, A.R. de Lera, Natural and Structure-based
RXR Ligand Scaffolds and Their Functions, Curr Top Med Chem 17(6)
(2017) 631-662.
[10] M.I. Dawson, X.K. Zhang, Discovery and design of retinoic acid
receptor and retinoid X receptor class- and subtype-selective synthetic
analogs of all-trans-retinoic acid and 9-cis-retinoic acid, Curr Med Chem
9(6) (2002) 623-637.
[11] C. Querfeldt, L.V. Nagelli, S.T. Rosen, T.M. Kuzel, J. Guitart,
Bexarotene in the treatment of cutaneous T-cell lymphoma, Expert Opin
Pharmaco 7(7) (2006) 907-915.
[12] X. Cao, W. Liu, F. Lin, H. Li, S.K. Kolluri, B. Lin, Y.-H. Han, M.I.
Dawson, X.-k. Zhang, Retinoid X receptor regulates
Nur77/TR3-dependent apoptosis by modulating its nuclear export and
mitochondrial targeting., Mol Cell Biol 24(22) (2004) 9705-9725.
[13] F. Casas, L. Daury, S. Grandemange, M. Busson, P. Seyer, R.
Hatier, A. Carazo, G. Cabello, C. Wrutniak-Cabello, Endocrine regulation

of mitochondrial activity: involvement of truncated RXRalpha and c-Erb
Aalpha1 proteins, FASEB J 17(3) (2003) 426-36.
[14] Y. Katagiri, K. Takeda, Z.X. Yu, V.J. Ferrans, K. Ozato, G. Guroff,
Modulation of retinoid signalling through NGF-induced nuclear export of
NGFI-B, Nat Cell Biol 2(7) (2000) 435-40.
[15] R. Ghose, T.L. Zimmerman, S. Thevananther, S.J. Karpen,
Endotoxin leads to rapid subcellular re-localization of hepatic RXRalpha:
A novel mechanism for reduced hepatic gene expression in inflammation,
Nucl Recept 2(1) (2004) 4.
[16] T.L. Zimmerman, S. Thevananther, R. Ghose, A.R. Burns, S.J.
Karpen, Nuclear export of retinoid X receptor alpha in response to
interleukin-1beta-mediated cell signaling: roles for JNK and SER260, J
Biol Chem 281(22) (2006) 15434-40.
[17] K. Fukunaka, T. Saito, K. Wataba, K. Ashihara, E. Ito, R. Kudo,
Changes in expression and subcellular localization of nuclear retinoic
acid receptors in human endometrial epithelium during the menstrual
cycle, Mol Hum Reprod 7(5) (2001) 437-46.
[18] G.H. Wang, F.Q. Jiang, Y.H. Duan, Z.P. Zeng, F. Chen, Y. Dai, J.B.
Chen, J.X. Liu, J. Liu, H. Zhou, H.F. Chen, J.Z. Zeng, Y. Su, X.S. Yao,
X.K. Zhang, Targeting truncated retinoid X receptor-alpha by CF31
induces TNF-alpha-dependent apoptosis, Cancer research 73(1) (2013)

307-18.
[19] H. Zhou, W. Liu, Y. Su, Z. Wei, J. Liu, S.K. Kolluri, H. Wu, Y. Cao, J.
Chen, Y. Wu, T. Yan, X. Cao, W. Gao, A. Molotkov, F. Jiang, W.G. Li, B.
Lin, H.P. Zhang, J. Yu, S.P. Luo, J.Z. Zeng, G. Duester, P.Q. Huang, X.K.
Zhang, NSAID sulindac and its analog bind RXR alpha and inhibit RXR
alpha-dependent AKT signaling, Cancer Cell 17(6) (2010) 560-73.
[20] B. Lin, Kolluri, S., Lin, F., Liu, W., Han, Y.H., Cao, X., Dawson, M.I.,
Reed, J.C., and Zhang, X.K., Conversion of Bcl-2 from Protector to Killer
by Interaction with Nuclear Orphan Receptor Nur77/TR3, Cell 116 (2004)
527-540.
[21] L. Chen, Z.G. Wang, A.E. Aleshin, F. Chen, J. Chen, F. Jiang, G.
Alitongbieke, Z. Zeng, Y. Ma, M. Huang, H. Zhou, G. Cadwell, J.F. Zheng,
P.Q. Huang, R.C. Liddington, X.K. Zhang, Y. Su, Sulindac-derived
RXRalpha modulators inhibit cancer cell growth by binding to a novel site,
Chem Biol 21(5) (2014) 596-607.
[22] Z. Zeng, Z. Sun, M. Huang, W. Zhang, J. Liu, L. Chen, F. Chen, Y.
Zhou, J. Lin, F. Huang, L. Xu, Z. Zhuang, S. Guo, G. Alitongbieke, G. Xie,
Y. Xu, B. Lin, X. Cao, Y. Su, X.-k. Zhang, H. Zhou, Nitrostyrene
Derivatives Act as RXR alpha Ligands to Inhibit TNF alpha Activation of
NF-kappa B, Cancer research 75(10) (2015) 2049-2060.
[23] L. Chen, A.E. Aleshin, G. Alitongbieke, Y. Zhou, X. Zhang, X. Ye, M.

Hu, G. Ren, Z. Chen, Y. Ma, D. Zhang, S. Liu, W. Gao, L. Cai, L. Wu, Z.
Zeng, F. Jiang, J. Liu, H. Zhou, G. Cadwell, R.C. Liddington, Y. Su, X.K.
Zhang, Modulation of nongenomic activation of PI3K signalling by
tetramerization of N-terminally-cleaved RXRalpha, Nat Commun 8 (2017)
16066.
[24] E. Lionta, G. Spyrou, D.K. Vassilatis, Z. Cournia, Structure-Based
Virtual Screening for Drug Discovery: Principles, Applications and Recent
Advances, Curr Top Med Chem 14(16) (2014) 1923-1938.
[25] F. Chen, J. Liu, M. Huang, M. Hu, Y. Su, X.K. Zhang, Identification of
a New RXRalpha Antagonist Targeting the Coregulator-Binding Site,
ACS Med Chem Lett 5(7) (2014) 736-741.
[26] V. Law, C. Knox, Y. Djoumbou, T. Jewison, A.C. Guo, Y. Liu, A.
Maciejewski, D. Arndt, M. Wilson, V. Neveu, A. Tang, G. Gabriel, C. Ly, S.
Adamjee, Z.T. Dame, B. Han, Y. Zhou, D.S. Wishart, DrugBank 4.0:
shedding new light on drug metabolism, Nucleic Acids Res 42(Database
issue) (2014) D1091-7.
[27] Schrödinger, Release 2016-1: LigPrep, Version 3.7, Schrödinger,
LLC, New York, NY, 2016.
[28] G.M. Sastry, M. Adzhigirey, T. Day, R. Annabhimoju, W. Sherman,
Protein and ligand preparation: parameters, protocols, and influence on
virtual screening enrichments, J Comput Aided Mol Des 27(3) (2013)

221-34.
[29] Schrödinger, Release 2016-1: Prime, Version 4.3, Schrödinger, LLC,
New York, NY, 2016.
[30] Schrödinger, Release 2016-1: Glide, Version 7.0, Schrödinger, LLC,
New York, NY, 2016.
[31] V. Nahoum, E. Perez, P. Germain, F. Rodriguez-Barrios, F. Manzo,
S. Kammerer, G. Lemaire, O. Hirsch, C.A. Royer, H. Gronemeyer, A.R.
de Lera, W. Bourguet, Modulators of the structural dynamics of the
retinoid X receptor to reveal receptor function, P Natl Acad Sci USA
104(44) (2007) 17323-17328.
[32] Z.X. Wang, N.R. Kumar, D.K. Srivastava, A Novel Spectroscopic
Titration Method for Determining the Dissociation-Constant and
Stoichiometry of Protein Ligand Complex, Anal Biochem 206(2) (1992)
376-381.
[33] H. Zhao, A. Caflisch, Molecular dynamics in drug design, Eur J Med
Chem 91 (2015) 4-14.
[34] M. Conda-Sheridan, E.J. Park, D.E. Beck, P.V. Reddy, T.X. Nguyen,
B. Hu, L. Chen, J.J. White, R.B. van Breemen, J.M. Pezzuto, M.
Cushman, Design, synthesis, and biological evaluation of
indenoisoquinoline rexinoids with chemopreventive potential, Journal of
medicinal chemistry 56(6) (2013) 2581-605.

Highlights:
· A pilot virtual screening was performed to identify RXRα-LBD tetramer
stablizers.
· Atorvastatin was identified to bind to RXRα and promote the RXRα-LBD
tetramer formation
· Atorvastatin induces PARP cleavage in RXRα-dependent manner.