Sulindac-derived RXRα modulators inhibit cancer cell growth by binding to a novel site

Article abstract of DOI:10.1016/j.chembiol.2014.02.017

Retinoid X receptor-alpha (RXRα), an intriguing and unique drug target, can serve as an intracellular target mediating the anticancer effects of certain nonsteroidal anti-inflammatory drugs (NSAIDs), including sulindac. We report the synthesis and characterization of two sulindac analogs, K-8008 and K-8012, which exert improved anticancer activities over sulindac in a RXRα-dependent manner. The analogs inhibit the interaction of the N-terminally truncated RXRα (tRXRα) with the p85α subunit of PI3K, leading to suppression of AKT activation and induction of apoptosis. Crystal structures of the RXRα ligand-binding domain (LBD) with K-8008 or K-8012 reveal that both compounds bind to tetrameric RXRα LBD at a site different from the classical ligand-binding pocket. Thus, these results identify K-8008 and K-8012 as tRXRα modulators and define a binding mechanism for regulating the nongenomic action of tRXRα.

Full text of DOI:10.1016/j.chembiol.2014.02.017

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Article
Sulindac-Derived RXRa Modulators Inhibit
Cancer Cell Growth by Binding
to a Novel Site
Liqun Chen,1,2,4 Zhi-Gang Wang,3,4 Alexander E. Aleshin,2,4 Fan Chen,1,2 Jiebo Chen,1,2 Fuquan Jiang,1
Gulimiran Alitongbieke,1 Zhiping Zeng,1 Yue Ma,1 Mingfeng Huang,1 Hu Zhou,1,2 Gregory Cadwell,2 Jian-Feng Zheng,3
1,2,
Pei-Qiang Huang,3 Robert C. Liddington,2 Xiao-kun Zhang,1,2, and Ying Su
*
1School of Pharmaceutical Sciences, Xiamen University, Xiamen 361102, China
*
2Sanford-Burnham Medical Research Institute, 10901 North Torrey Pines Road, La Jolla, CA 92037, USA
3Department of Chemistry and Fujian Provincial Key Laboratory of Chemical Biology, College of Chemistry and Chemical Engineering,
Xiamen University, Xiamen 361005, China
4Co-first author
*Correspondence: xzhang@sanfordburnham.org (X.-k.Z.), ysu@sanfordburnham.org (Y.S.)
http://dx.doi.org/10.1016/j.chembiol.2014.02.017
SUMMARY
domain, and a ligand-binding domain (LBD). The LBD possesses
a ligand-binding pocket (LBP) for the binding of small mole-
Retinoid X receptor-alpha (RXRa), an intriguing and cule ligands, a transactivation function domain termed AF-2
composed of helix 12 (H12) of the LBD, a coregulator binding sur-
face, and a dimerization surface (Germain et al., 2006; Szanto
unique drug target, can serve as an intracellular
target mediating the anticancer effects of certain
nonsteroidal anti-inflammatory drugs (NSAIDs),
including sulindac. We report the synthesis and char-
acterization of two sulindac analogs, K-8008 and
K-8012, which exert improved anticancer activities
over sulindac in a RXRa-dependent manner. The an-
alogs inhibit the interaction of the N-terminally trun-
et al., 2004). The ligand-dependent transcription regulation is
predominately mediated through H12 that is highly mobile.
Agonist ligand binds to the LBP and helps H12 to adopt the active
conformation that forms a surface to facilitate the binding of co-
activators and subsequent transactivation. In contrast, in the
absence of an agonist ligand or in the presence of an antagonist
ligand, H12 adopts an inactive conformation that favors the bind-
ing of corepressors to inhibit target gene transcription. Natural
RXRa ligand 9-cis-retinoic acid (9-cis-RA) and synthetic ligands
cated RXRa (tRXRa) with the p85a subunit of PI3K,
leading to suppression of AKT activation and induc-
tion of apoptosis. Crystal structures of the RXRa have been effective in preventing tumorigenesis in animals, and
RXRa has been a drug target for therapeutic applications, espe-
cially in the treatment of cancer (Bushue and Wan, 2010; Yen and
ligand-binding domain (LBD) with K-8008 or K-8012
reveal that both compounds bind to tetrameric
RXRa LBD at a site different from the classical
ligand-binding pocket. Thus, these results identify
K-8008 and K-8012 as tRXRa modulators and define
a binding mechanism for regulating the nongenomic
action of tRXRa.
Lamph, 2006). Targretin, a synthetic RXR-selective retinoid (rexi-
noid), was approved for treating cutaneous T-cell lymphoma
(Dawson and Zhang, 2002). RXRa can bind to DNA and activate
transcription of target genes either as a homodimer or a hetero-
dimer with its heterodimerization partners including retinoic acid
receptor, vitamin D receptor (VDR), thyroid hormone receptor
(TR), and peroxisome-proliferator-activated receptor (Germain
et al., 2006; Szanto et al., 2004). In addition to homodimer and
heterodimer, RXRa could also self-associate into homotetramers
INTRODUCTION
in solution, which rapidly dissociate into active dimers upon
Retinoid X receptor alpha (RXRa), a unique member of the nu- binding of a cognate ligand (Chen et al., 1998; Kersten et al.,
clear receptor superfamily, regulates a broad spectrum of phys- 1995). Tetramer formation of RXRa might serve to sequester
iological functions including cell differentiation, growth, and the receptor’s active species, dimers and monomers, into a tran-
apoptosis (Germain et al., 2006; Szanto et al., 2004). Like other scriptionally inactive tetramer complex (Gampe et al., 2000).
nuclear receptors, RXRa acts as a ligand-dependent transcrip-
Efforts ondiscoveryofsmall moleculestargeting RXRa forther-
tion factor (Germain et al., 2006; Szanto et al., 2004). Recent apeutic application have been primarily focused on the optimiza-
accumulating evidence indicates that RXRa also has extranu- tion of the molecules that bind to its classical LBP (de Lera et al.,
clear actions. RXRa resides in the cytoplasm at certain stages 2007; Germain et al., 2006; Szanto et al., 2004). However, various
during development (Dufour and Kim, 1999; Fukunaka et al., studies have recently identified small-molecule modulators of nu-
2001) and migrates from the nucleus to the cytoplasm in clear receptors that function via unknown sites and undefined
´
response to differentiation, apoptosis, and inflammation (Cao mechanisms of action (Buzon et al., 2012; Moore et al., 2010).
et al., 2004; Casas et al., 2003; Zimmerman et al., 2006). RXRa The recent report of the structure of estrogen receptor-b (ERb)
exhibits a modular organization structurally consisting of three with a second molecule of 4-hydroxytamoxifen bound in its coac-
main functional domains: an N-terminal region, a DNA-binding tivator-binding surface represents a direct example of a second
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved
1

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
Figure 1. Chemical Structures and Synthesis
(A) The structures of sulindac and its analogs
K-80003, K-8008, and K-8012.
(B) Synthesis scheme of K-8008 and K-8012.
X-ray crystallographic studies of the LBD
of RXRa in complex with K-8008 or K-
8012 revealed that both compounds
bound to the RXRa LBD in its tetrameric
form via a particular site outside of the
classical RXRa LBP, providing a different
strategy for developing RXRa-based
agents for cancer therapy.
RESULTS
K-8008 and K-8012 Are Antagonists
of RXRa
In an effort to identify improved sulindac
analogs for cancer therapy, we designed
and synthesized
a series of analogs
around K-80003 (Wang et al., 2013;
Zhou et al., 2010). The analogs shown in
Figure 1 were initially evaluated by the
reporter assay using a chloramphenicol
acetyltransferase (CAT) reporter contain-
ing TREpal that is known to bind to
RXRa homodimer (Zhang et al., 1992).
9-cis-RA strongly induced the TREpal
reporter activity, which was inhibited by
BI-1003, a known RXRa antagonist (Lu
et al., 2009). K-8008 and K-8012 also ex-
estrogen receptor ligand-binding site and provides insight into hibited an inhibitory effect on 9-cis-RA-induced TREpal reporter
the possible pharmacological effects of the drug (Wang et al., activity in a concentration-dependent manner (Figure 2A),
2006). Compounds that bind to the surface binding sites of LBD although they did not show any agonist activity at the concentra-
have been demonstrated for other nuclear receptors, including tions used (Figure 2B). The antagonist effect of K-8008 and
´
androgen receptor (AR), VDR, and TR (Buzon et al., 2012; Moore K-8012 was much better than sulindac and comparable to
et al., 2010). However, compounds that bind to RXRa at the sites K-80003 (Figure 2A). We also used the Gal4-RXRa-LBD chimera
other than the classical LBP have not been reported.
and Gal4 reporter system to evaluate the inhibitory effect of
We recently showed that certain nonsteroidal anti-inflamma- K-8008 and K-8012 on 9-cis-RA-induced reporter activity.
tory drugs (NSAIDs), including etodolac (Kolluri et al., 2005) Cotransfection of Gal4-RXRa-LBD strongly activated the Gal4
and sulindac (Zhou et al., 2010), could bind to RXRa and modu- reporter in the presence of 9-cis-RA, which was inhibited by
late its biological activities. Interestingly, sulindac but not 9-cis- BI-1003 as well as K-8008 and K-8012 (Figure 2C). Dose-
RA could inhibit the binding of an N-terminally truncated RXRa response experiments showed that the IC50 values for K-8008
protein (tRXRa) to the p85a regulatory subunit of phosphatidyli- and K-8012 to inhibit 9-cis-RA-induced Gal4-RXRa-LBD trans-
nositol-3-OH kinase (PI3K), leading to inhibition of the tumor ne- activation were about 13.2 and 9.2 mM, respectively (Figure 2D).
crosis factor-a (TNF-a)-activated PI3K/protein kinase B (AKT) Thus, K-8008 and K-8012 are antagonists of RXRa.
pathway (Zhou et al., 2010). We also demonstrated, through a
designed sulindac analog, K-80003, the feasibility of developing K-8008 and K-8012 Induce Apoptosis and Inhibit AKT
a new generation of RXRa-specific molecules for therapeutic Activation by Preventing tRXRa from Binding to p85a
application and mechanistic studies of RXRa (Wang et al., We next evaluated K-8008 and K-8012 for their effect on the
2013; Zhou et al., 2010). These results identify sulindac and growth of cancer cells. Compared to sulindac, K-8008 and
related analogs as unique regulators of tRXRa activity through K-8012 were much more effective in inhibiting the growth of
an undefined binding mechanism. Here we report our synthesis various cancer cells, including A549 lung cancer (Figure 3A),
and characterization of K-80003-based analogs, K-8008 and PC3prostatecancer, andZR-75-1 and MB231breastcancercells
K-8012, which exhibited improved activity in inhibiting the (Figure S1 available online). One unique property of sulindac and
tRXRa-mediated PI3K/AKT signaling pathway. Moreover, our analogs is their ability to inhibit TNF-a-induced AKT activation
2
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
Figure 2. Antagonist Effect of Sulindac
Analogs
(A and B) Inhibition of RXRa transactivation.
(TREpal)2-tk-CAT and RXRa were transiently
transfected into CV-1 cells. Cells were treated with
or without 9-cis-RA (10À7 M) in the presence or
absence of the indicated concentration of sulindac
and its analogs. CAT activity was determined. BI-
1003 (1 mM) was used for comparison. 1, 2, 3, and
4
stand for sulindac, K-80003, K-8008, and
K-8012, respectively.
(C) Inhibition of 9-cis-RA-induced Gal4 reporter
activity. pBind-RXRa-LBD and pG5luc were tran-
siently transfected into HCT-116 cells. Cells were
treated with or without 9-cis-RA (10À7 M) in the
presence or absence of BI-1003 (1 mM), K-8008
(50 mM), and K-8012 (50 mM). Luciferase (LUC)
activity was determined.
(D) Dose-dependent effect of K-8008 and K-8012.
HCT-116 cells transfected with pBind-RXRa-LBD
and pG5luc were treated with the indicated con-
centrations of K-8008 and K-8012 in the presence
or absence of 9-cis-RA (10À7 M). All reporter ac-
tivity is expressed as means SD from three in-
dependent experiments.
(Zhou et al., 2010). Thus, A549 lung cancer cells were treated with the role of tRXRa, RXRa-D80, a RXRa mutant lacking its N-termi-
TNF-a intheabsenceorpresenceofK-8008orK-8012. Treatment nal 80 amino acids and mimicking tRXRa (Wang et al., 2013;
of cells with TNF-a enhanced AKT activation as revealed by Zhou et al., 2010), was transfected into HeLa cells. Transfection
western blotting (Figure 3B). However, when cells were cotreated of RXRa-D80 but not the full-length RXRa enhanced the effect of
with either K-8008 or K-8012, the TNF-a-induced AKT activation K-8008 on inducing PARP cleavage in the presence of TNF-a
was suppressed in a dose-dependent manner (Figure 3B). Similar (Figure 3F). Together, these results demonstrate that tRXRa
results were obtained in other cancer cell lines (Figure S2).
TNF-a is a multifunctional cytokine that controls diverse
plays a crucial role in mediating the biological effects of K-8008.
We next determined whether K-8008 could affect tRXRa inter-
cellular events such as cell survival and death (Balkwill, 2009; action with p85a, an interaction known to activate AKT (Zhou
Wang and Lin, 2008). Inhibition of TNF-a-induced AKT activation et al., 2010). HeLa cells were transfected with Myc-tagged
by sulindac and analogs in cancer cells led to a shift of TNF-a RXRa-D80 and Flag-tagged p85a expression vectors and treated
signaling from survival to death (Zhou et al., 2010). We therefore with or without TNF-a and/or K-8008. Coimmunoprecipitation
evaluated the effect of K-8008 and K-8012 alone or in combina- assays using anti-Myc antibody showed that Flag-p85a was
tion with TNF-a on the cleavage of poly(ADP-ribose) polymerase coimmunoprecipitated together with Myc-RXRa-D80 in cells
(PARP), an indication of apoptosis in cancer cells (Lazebnik et al., treated with TNF-a (Figure 3G). However, when cells were
1994). Treatment of A549 cells with TNF-a did not have effect on cotreated with K-8008, TNF-a-induced interaction of Myc-
PARP cleavage, whereas treatment with sulindac or analogs RXRa-D80 with Flag-p85a was almost completely inhibited. We
slightly induced PARP cleavage. Combination of sulindac or also examined the effect of K-8008 on interaction of endogenous
analogs with TNF-a, however, caused a significant induction of tRXRa with p85a in A549 cells. Cell lysates prepared from A549
PARP cleavage (Figure 3C and Figure S3). Thus, K-8008 and cells treated with TNF-a in the presence or absence of K-8008
K-8012, like sulindac, could convert TNF-a signaling from sur- were analyzed by coimmunoprecipitation using D197 anti-
vival to death in cancer cells.
RXRa antibody that recognizes both tRXRa and RXRa (Zhou
The growth inhibitory effect of K-8008 and K-8012 and the et al., 2010). Figure 3H shows that treatment of cells with TNF-a
induction of apoptosis by K-8008 occurred at low micromolar promoted the interaction of endogenous tRXRa with p85a,
concentrations, suggesting that they might exert their anticancer consistent with previous findings (Zhou et al., 2010). When cells
effects through RXRa binding. To address the issues, cancer were cotreated with K-8008, the interaction was largely inhibited.
cells were transfected with RXRa small interfering RNA (siRNA) Such an effect of K-8008 on inhibiting TNF-a-induced p85a inter-
and evaluated for the effect of K-8008 on inducing PARP cleav- action with tRXRa was also observed in other cancer cell lines,
age and inhibiting AKT activation. Knocking down RXRa expres- including PC3 and HepG2 cells (Figure S4). Together, these re-
sion by RXRa siRNA transfection significantly diminished the sults demonstrate that K-8008 can induce TNF-a-dependent
effect of K-8008 on inducing PARP cleavage (Figure 3D) and apoptosis by suppressing the tRXRa-mediated activation of
inhibiting TNF-a-induced AKT activation (Figure 3E). To address AKT through its inhibition of tRXRa interaction with p85a.
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved
3

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
Figure 3. Biological Effects of K-8008 and K-8012
(A) Growth inhibition by sulindac and its analogs. A549 lung cancer cells were treated with various concentrations of the indicated compounds. Cell viability was
measured by the MTT colorimetric assay. The values represent averages ( SD) from five independent experiments. 1, 2, 3, and 4 stand for sulindac, K-80003,
K-8008, and K-8012, respectively.
(B) Inhibition of TNF-a-induced AKT activation. A549 cells were pretreated with sulindac or analogs for 1 hr before being exposed to TNF-a (10 ng/ml) for 30 min.
Phosphorylated AKT and total AKT were analyzed by immunoblotting.
(C) Induction of apoptosis by sulindac and analogs in the presence of TNF-a. Cells cultured in medium with 1% FBS were treated with TNF-a (10 ng/ml) and/or
compound (40 mM) for 4 hr and analyzed for PARP cleavage by immunoblotting.
(D) RXRa siRNA transfection inhibits the apoptotic effect of K-8008. HeLa cells transfected with control or RXRa siRNA for 48 hr were treated with K-8008 (40 mM)
and/or TNF-a (10 ng/ml) for 6 hr and analyzed by immunoblotting.
(E) RXRa siRNA transfection antagonizes the inhibitory effect of K-8008 on AKT activation. HeLa cells transfected with control or RXRa siRNA for 48 hr were
treated with K-8008 (40 mM) for 1.5 hr before being exposed to TNF-a (10 ng/ml) for 30 min and analyzed by immunoblotting.
(F) Myc-RXRa-D80 transfection enhances the apoptotic effect of K-8008. HeLa cells transfected with Myc-RXRa or Myc-RXRa-D80 were treated with K-8008
(20 mM) and/or TNF-a for 12 hr. PARP cleavage and transfected RXRa expression were analyzed by immunoblotting.
(legend continued on next page)
4
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
Figure 4. Inhibition of HepG2 Tumor Growth
in Animals
(A–C) Nude mice with HepG2 heptoma xenografts
were intraperitoneally injected daily with vehicle, K-
8008 (20 mg/kg), or K-80003 (20 mg/kg) for
12 days. Tumors were removed and measured.
Tumor sizes and weights in control and K-80003-
and K-8008-treated mice were compared. In (B),
one of three similar experiments is shown. Error
bars represent SEM.
(D) Lysates prepared from three tumors treated
with vehicle or K-8008 were analyzed by western
blotting assay for p-AKT expression.
(E) H&E staining and TUNEL assay. Tumor sections
were stained for H&E or TUNEL by immunohisto-
chemistry. Increased apoptotic tumor cells were
observed in tumor from K-8008-treated mice.
(F) K-8008 does not exhibit apparent toxicity. Body
weight was measured every 3 days. Each point
represents the mean standard deviation of six
mice. The differences between the compound-
treated group and control group are not significant
(p > 5%). One of three similar experiments is
shown. Error bars represent SEM.
and K-8012 might bind to the canonical
binding site, the LBP of RXRa, acting as
conventional antagonists. Thus, we evalu-
ated their binding to the LBP of RXRa
using the classical radioligand competi-
tion binding assay (Zhou et al., 2010).
Unlike 9-cis-RA and K-80003 that com-
To further evaluate the anticancer effect of K-8008, mice with peted well with [3H]9-cis-RA for binding to the LBP of RXRa,
HepG2 tumor xenografts were treated with 20 mg/kg K-8008 or K-8008 and K-8012 failed to replace [3H]9-cis-RA for its binding
K-80003. Administration of K-8008 inhibited the growth of to the RXRa LBP (Figure 5A).
HepG2 tumor in a time-dependent manner (Figure 4A), resulting
Results of the [3H]9-cis-RA binding competition assay demon-
in a 61.23% reduction of tumor weight after a 12-day treatment strated that K-8008 and K-8012 did not bind to the canonical
(Figures 4B and 4C), which was comparable with the inhibitory binding site, suggesting a different binding mechanism. Other
effect of K-80003 (54.84% reduction). Consistent with our than the classical LBP, recent structural and functional studies
in vitro observation, examination of three tumors treated with have revealed the existence of distinct small molecule binding
´
or without K-8008 showed reduction of AKT activation by sites on the surface of the LBD of nuclear receptors (Buzon
K-8008 (Figure 4D). Moreover, TUNEL staining revealed induc- et al., 2012; Moore et al., 2010). One of the potential alternative
tion of apoptosis by K-8008 (Figure 4E). Significantly, administra- sites in which small molecules could bind and function as antag-
tion of either K-80003 or K-8008 did not show any apparent toxic onists is the coregulator-binding site in the LBD. Antagonistic
effects such as loss of body weight (Figure 4F).
properties displayed by ligands through binding to the coregula-
tor site have been reported for several members of the nuclear
receptor family. For example, small molecules mimicking the
structure of the coactivator LXXLL a-helical motif have been
K-8008 and K-8012 Do Not Bind to the Classical LBP
of RXRa
Although sulindac and analogs can bind tRXRa and induce developed for modulating the activities of AR, ER, and VDR
tRXRa-dependent apoptosis of cancer cells, how they bind to (Mita et al., 2010; Rodriguez et al., 2004; Arnold et al., 2005;
tRXRa to regulate tRXRa functions remains undefined. Accord- Moore et al., 2010). We therefore tested whether K-8008 and
ing to current understanding of the mechanism by which ligands K-8012 could bind to an alternative surface binding site by
regulate the transcriptional activity of nuclear receptors, K-8008 using the time-resolved fluorescence resonance energy transfer
(G) K-8008 inhibits the interaction of transfected tRXRa and p85a. HeLa cells transfected with Myc-RXRa-D80 and/or Flag-p85a for 24 hr were treated with
vehicle or 20 mM K-8008 in the presence of absence of 10 ng/ml TNF-a for 1 hr. Cell lysates were immunoprecipitated (IP) using anti-Myc antibody and analyzed
by western blotting (WB) analysis using the indicated antibody.
(H) K-8008 inhibits the interaction of endogenous tRXRa and p85a. A549 cells were pretreated with vehicle or 40 mM K-8008 for 1 hr before being exposed to
10 ng/ml TNF-a for 30 min. Cell lysates were immunoprecipitated with DN197 anti-RXRa antibody and analyzed by western blotting.
See also Figure S1–S4.
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved
5

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
Figure 5. Unique Binding of K-8008 and
K-8012 to RXRa
(A) K-8008 and K-8012 fail to compete with the
binding of 9-cis-RA to RXRa. RXRa-LBD protein
was incubated with [3H]9-cis-RA in the presence or
absence of sulindac analogs K-80003, K-8008,
K-8012, or unlabeled 9-cis-RA. Bound [3H]9-cis-
RA was quantitated by liquid scintillation counting.
One of three similar experiments is shown. Error
bars represent SEM. 1, 2, 3, and 4 stand for su-
lindac, K-80003, K-8008, and K-8012, respectively.
(B) K-8008 and K-8012 reduce 9-cis-RA-induced
FRET signal. GST-RXRa-LBD was incubated with
K-8008 or K-8012 in the presence or absence of
9-cis-RA (10À8 M). BI-1003 (1 mM) was used as a
control. The values represent averages ( SD) from
three independent experiments.
(C) Dose-dependent effect of K-8008 and K-8012
on 9-cis-RA-induced FRET signal. GST-RXRa-
LBD was incubated with K-8008 or K-8012 in the
presence of 9-cis-RA (10À8 M). The values repre-
sent averages ( SD) from three independent
experiments.
and Figure S5A). Superposition of our
crystal structure with the published
apo structure (PDB code 1G1U) shows
that the corresponding monomers have
almost identical folds with small shifts
found in the orientation of H12 in the
(TR-FRET) RXRa coactivator peptide competition assay. Our monomer where a K-8008 molecule is bound (Figure 6B). N-ter-
results showed that both compounds could inhibit 9-cis-RA- minal residues, from 223 to 260, were found to be disordered and
induced interaction of RXRa LBD with its coactivator peptide undetermined in the complex structures, though residues 231–
(Figure 5B). The inhibitory effect of K-8008 and K-8012 was 260 were defined in the apo structure. In a tetramer, two modu-
much stronger than sulindac, with IC50 values of 16.8 and lator molecules were found to bind to one homotetramer, with a
14.5 mM, respectively (Figure 5C), which correlated well with their binding stoichiometric ratio of 1:2 between ligand and protein,
inhibition of 9-cis-RA-induced RXRa transactivation (Figure 2D). because one ligand molecule binds only to one monomer within
Taken together, K-8008 and K-8012 might act as RXRa antago- a dimer (Figure 6A). Observation of the same stoichiometric ratio
nists by binding to a novel RXRa surface site, leading to inhibition between ligand and protein has been reported for another
of coactivator binding.
RXRa-LBD/antagonist (Gampe et al., 2000). K-8008 binds to a
region that is close to the dimer-dimer interface, making interac-
tion primarily with one monomer of the dimer and some interac-
tion with one monomer of the other dimer. The structure of RXRa
K-8008 and K-8012 Bind to a Tetrameric Structure of the
RXRa LBD
To gain direct and structural understanding of the binding of LBD in complex with K-8012 is very similar to that of RXRa LBD
K-8008 or K-8012 to RXRa, we performed crystallographic in complex with K-8008 (Figure S5B). Therefore we describe and
studies of these ligands bound to the RXRa LBD. Crystals of pro- discuss only the protein/K-8008 complex structure here.
tein/ligand complexes were obtained using a cocrystallization Both K-8008 and K-8012 bind to a hydrophobic region of LBD
method. The structures of RXRa LBD in complex with K-8008 near the entry and the edge of the cognate LBP. This region does
˚
and K-8012 were determined to the resolution of 2.0 and 2.2 A, not overlap with the binding region of 9-cis-RA (Figure 6B), which
respectively. Both protein/ligand complexes crystallized as explains why both compounds failed to compete with the
tetrameric oligomers in the space group of P21 with similar unit binding of 9-cis-RA (Figure 5A). To get a sense of how far
cell parameters, and the molecular replacement method was the K-8008 binding region is away from the LBP, we calculated
used to obtain the initial phasing by using the published RXRa the distance between the centroid of the bound 9-cis-RA
structure, Protein Data Bank (PDB) code 1G1U. Statistics of and the centroid of the bound K-8008 molecule using the tools
structure refinement and data collection are summarized in available in Maestro. In the superimposed structures of the
Table S1.
K-8008-bound LBD structure with the 9-cis-RA-bound LBD
˚
The crystal structure of the RXRa LBD in complex with the structure, the two centroids are about 7 A apart (Figure S6A).
K-8008 exists as noncrystallographic homotetramer similar to The K-8008 hydrophobic binding region is made of side chains
the reported apo homotetramer (Gampe et al., 2000), in which primarily from one monomer: Ala271 and Ala272 from H3;
two homodimers pack in a bottom-to-bottom manner (Figure 6A Trp305 and Leu309 from H5; Leu326 and Leu330 from the
6
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
Figure 6. Crystal Structure of the RXRa LBD
in Complex with K-8008
(A) The tetramer structure of RXRa LBD in complex
with K-8008. The two bound K-8008 molecules are
shown as magenta sticks surrounded by an elec-
tron density mesh (see Figure 6D legend for
details). The two biological dimers (A1-B1 and A2-
B2) are shown as green/cyan and yellow/brown
pairs, respectively. The N and C termini of four
subunits are marked by the corresponding residue
numbers.
(B) Superposition of the RXRa LBD monomers
from the K-8008-binding structure (brown rib-
bons) and the apo protein structure (purple rib-
bons, from PDB entry 1G1U). K-8008 is shown
as sticks (carbon/nitrogen atoms are in magenta/
blue). The classic ligand binding site is also
marked by a VDW ball model (in cyan/red) of
9-cis-RA taken from a superimposed PDB entry
1FBY.
(C) The hydrophobicity of the K-8008 binding site
presented as a surface fragment on top of the
RXRa LBD monomer (brown ribbon). The hy-
drophobic side chains that contribute to the
region are shown in green, and K-8008 is shown
in the same fashion as in Figure 6B. For clarity,
residues contributing to this region are not
labeled.
(D) The protein side chains (in sticks) that make
VDW interaction with K-8008. The displayed
region is an enlargement of the black box in
Figure 6A. The view is slightly rotated, and
fragments of the green subunit that do not
interact with K-8008 are removed. The protein
surface is shown as semitransparent envelope.
The Fo-Fc electron density is shown around the
ligands as a black mesh. It was calculated at a
3-s level with omitted ligand atoms. The posi-
tive end of the H11 helix dipole is highlighted in
orange.
(E) Side chains around K-8008 that make signifi-
cant changes in comparison with the apo protein
(PDB code 1G1U). Side chains of the apo protein
are shown in green sticks, and the protein/K-8008
complex is shown in orange sticks. The complex
structure is shown in brown ribbon, and the apo
structure is in purple. K-8008 is presented in the
same fashion as in Figure 6B.
See also Figures S5 and S6.
beta-turn; Leu433 from H10; Leu436 from L10-11; Phe437, The overall structure of the K-8008-bound RXRa LBD is identical
Phe438, Ile442, and Gly443 from H11 of chain B2; and Leu436 to the apo structure of RXRa LBD tetramer. Binding of K-8008
from L10-11 of chain A1 (Figure 6C). With respect to the mono- does not induce much significant change in the surrounding
mer of RXRa LBD, this region is located on the surface of the side chains except for the side chains of Phe439 and Leu309.
RXRa monomer molecule. However, in the tetramer structure, The side chain of Phe439 swings out to make room for the ligand
this region is buried. K-8008 makes both hydrophobic interaction to bind, and the side chain of Leu309 rearranges to make better
and polar interaction with the protein. The negatively charged VDW contact with the ligand (Figure 6E). Compared to the struc-
tetrazole of the ligand sits on the top of the N-terminal end of ture of the RXRa LBD bound to 9-cis-RA (Egea et al., 2000),
˚
H11 with a distance of 3 A from the backbone N of residue K-8008 binding does not result in change in the shape of the
Phe438 (Figure S6B), making charge-diploe interaction with LBP, whereas 9-cis-RA binding induces a substantial change
˚
H11 (Figure 6B). At 4.0 A cutoff, the lipophilic part of the ligand in the shape of the LBP that includes the movement of H3 and
makes van der Waals (VDW) contacts with side chains of H11 and the reorientation of H12 (Figure 7A). 9-cis-RA binding
Ile268, Ala271, Trp305, Leu436, Phe438, Phe439, and Ile442 results in the formation of the coactivator-binding site; how-
from chain B2 and Leu436 from chain A1 (Figures 6A and 6D). ever, K-8008 binding does not promote the formation of the
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved
7

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
Figure 7. Structural Comparison and Muta-
genesis Analysis of the K-8008 Binding Site
(A) Structural superposition of the protein/K-8008
complex and the protein/9-cis-RA complex. The
9-cis-RA-bound structure (PDB code 1FBY) is
in pink cartoon, and 9-cis-RA is in cyan (C atoms)
and red (O atoms) sticks. The K-8008-bound
structure is in light orange cartoon, and K-8008
is in gray (C atoms) and blue (N atoms) sticks.
Side chain Arg316 is displayed for distance
comparison between distances to –COOH and to
tetrazole.
(B–D) Mutational analysis of the K-8008 binding
site. The LBD of RXRa or mutants cloned into
pBind vector and pG5luc were transiently co-
transfected into HCT-116 cells. Cells were
treated with or without 9-cis-RA (10À7 M) in the
presence or absence of BI-1003 (1 mM) or K-
8008 (50 mM). Luciferase (LUC) activity was
determined.
See also Figure S6. The reporter activity is ex-
pressed as mean SD from three independent
experiments.
coactivator-binding site. Instead it leaves the receptor protein in transactivation could be strongly induced by 9-cis-RA, but
an autorepression state. Thus, K-8008 acts as an antagonist. K-8008 failed to inhibit the induced transcriptional activity (Fig-
Similar to 9-cis-RA ligand, K-8008 is made of largely a lipophilic ure 7D). Together, the mutagenesis studies confirm the K-8008
body linked to a charged group. The carboxylate group of 9-cis- binding site identified by the crystal structure.
RA acts as an anchor via forming charge-charge interaction with
Arg316; however, the bioisostere tetrazole of the carboxylate in DISCUSSION
K-8008 makes no interaction with Arg316. The tetrazole group in
˚
the RXRa LBD/K-8008 structure is about 15 A away from Arg316. We report here our identification of two sulindac analogs,
Both 9-cis-RA and K-8008 make extensive hydrophobic interac- K-8008 and K-8012, which showed potent tRXR inhibitory
tions with the protein. Some of the residues contributing to the effects through a unique binding mechanism. Our results
hydrophobic interactions between ligand (9-cis-RA or K-8008) demonstrated that K-8008 and K-8012 were more effective
and the protein are shared, such as Ile268, Ala271, and than sulindac in inhibiting RXRa transactivation (Figure 2A).
Leu326, but because of the LBP shape change induced by the Sulindac binds to RXRa with an IC50 of 82.9 mM based on the
binding of 9-cis-RA, these shared residues are located in classical ligand competition assays (Zhou et al., 2010). K-8008
different regions of the structures of RXRa LBD/K-8008 and and K-8012 could antagonize 9-cis-RA-induced transactivation
RXRa LBD/9-cis-RA (Figure S6C).
and inhibit coactivator peptide binding to RXRa with IC50 value
To further support and cross-validate the identification of this of around 10 mM. Consistently, K-8008 and K-8012 showed
binding site on RXRa, we designed a couple of mutants that improved activity than sulindac in inhibiting AKT activation and
could potentially impact the binding of K-8008 but not 9-cis- inducing apoptosis. About 100 mM of sulindac is normally used
RA. Comparison of the binding nature of the K-8008 molecule to achieve its anticancer effects (Weggen et al., 2001; Yamamoto
to that of the 9-cis-RA reveals that Leu433 is close to the tetra- et al., 1999; Zhang et al., 2000), whereas 10–50 mM of K-8008
zole group and that replacing Leu433 with Glu could weaken and K-8012 were able to inhibit AKT activation and induce
the binding of K-8008 because of the repulsive interaction be- apoptosis of cancer cells. Furthermore, K-8008 showed potent
tween the deprotonated Glu and the negatively charged tetra- inhibitory effect on the growth of tumor cells in animals without
zole. However, Leu433Glu is not likely to impact the binding of apparent toxicity. Similar to sulindac, inhibition of AKT activation
9-cis-RA as in the 9-cis-RA-bound RXRa structure (PDB code and induction of apoptosis by K-8008 and K-8012 were RXRa
1FBY); Leu433 is not close to the ligand and is also partially sol- dependent, likely because of their inhibition of the interaction be-
vent-exposed. This prediction is supported by our reporter tween tRXRa and p85a.
assays showing that 9-cis-RA could similarly activate the wild-
K-8008 and K-8012 were synthesized as part of the structure-
type RXRa (Figure 7B) and the Leu433Glu RXRa mutant, activity relationship studies of sulindac, by replacing the carbox-
RXRa-L433E (Figure 7C). By contrast, K-8008 inhibited 9-cis- ylate group in K-80003 with its bioisotere tetrazole (Herr, 2002).
RA-induced transactivation of RXRa but not RXRa-L433E. Based on the principle of bioisosteric replacement (Matta
Structural comparison also suggests that mutating both et al., 2010), we anticipated that tetrazole group acted like the
Phe438 and Phe439 into Ala could affect the binding of K-8008 carboxylate group, a common motif found in most of the cognate
but not 9-cis-RA. Indeed, simultaneous substitution of Phe438 RXR ligands, which interacts with Arg316 in the LBP, and there-
and Phe439 with Ala resulted in a mutant, RXRa-F438,9A, whose fore both K-8008 and K-8012 would compete 9-cis-RA for
8
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
binding. To our surprise, both compounds, unlike sulindac and 2008; Mizwicki et al., 2004; Moore et al., 2010; Sun et al.,
K-80003, failed to compete with 9-cis-RA for binding to the 2011). Furthermore, in the case of sulindac and analogs, weak
LBP, demonstrating that they exert their antagonist effect antagonism may be clinically relevant because conventional
through a different binding mechanism from sulindac and administration of sulindac can result in about 10-15 mM sulindac
K-80003. Our structure analysis confirmed that the tetrazole in the serum of patients (Davies and Watson, 1997; Yamamoto
group of K-8008 and K-8012 binds to a region away from et al., 1999). Up to approximately 50 mM of sulindac could be
Arg316, and it anchors to the RXRa protein by sitting atop the detected in the plasma of humans depending on dose and
N terminus of helix 11, forming the charge-helix dipole interac- schedule (Davies and Watson, 1997), and sulindac can be
tion. This charge-dipole interaction may also function to stabilize concentrated in epithelial cells at concentrations that are at least
the orientation and conformation of H11 because in most of the 20-fold higher than those seen in the serum (Duggan et al., 1980;
cases ligand binding to the cognate LBP induces the conforma- Yamamoto et al., 1999). Consistently, we observed that admin-
tion change and reorientation of H11(Egea et al., 2000; Sato istration of K-8008 at the dose that effectively inhibited the
et al., 2010). Tetrazole is a nonclassical isotere of –COOH moiety growth of tumor cells did not show any apparent toxicity to ani-
(Herr, 2002; Matta et al., 2010). It has a similar PKa to –COOH mals (Figure 4). Thus, although showing a relatively weak binding
but has a different steric characteristics and different number to RXRa, these compounds could be clinically relevant.
of atoms from –COOH. Spatially tetrazole moiety is bulkier
than –COOH, and it seems that some spatial clash could prevent
the tetrazole group of K-8008 from making an anchoring charge-
charge interaction with Arg316. As a result, K-8008 and K-8012
bind to a novel region. The structural results show that the com-
pounds bind to a region that does not overlap with the 9-cis-RA
binding space, offering a structural explanation for the inability of
K-8008 and K-8012 to compete with 9-cis-RA for RXRa binding.
Furthermore, our mutagenesis data (Figures 7B–7D) support the
existence of this K-8008 binding site.
SIGNIFICANCE
RXRa represents an intriguing target for pharmacologic
intervention. Unfortunately, the development of RXRa-
based drugs has been hampered by the side effects associ-
ated with targeting its cognate LBP. Thus, discovery of
new strategies for targeting RXRa is significant. We report
here our identification of two sulindac-derived analogs
that exert their anticancer effects by binding to a site of
RXRa, which is different from the classical LBP. These mol-
ecules could serve as chemical probes for studying the role
of tRXRa in cancer. Furthermore, our results provide an
opportunity to specifically target this surface site and thus
may circumvent side effects associated with binding to the
classical RXRa LBP. The development of tRXRa-selective
inhibitors targeting a novel binding site may support a de-
parture from the traditional approach of targeting the LBP
and lead to a new paradigm targeting a functionally impor-
tant surface site, which may lead to more effective and spe-
cific therapeutics.
Our crystal structures revealed that K-8008 and K-8012 bind
to a RXRa LBD tetramer structure through a novel hydrophobic
region that is located on the surface of a monomer and near
the dimer-dimer interface in the tetramer. Unlike the binding of
other published ligands, the binding of K-8008 does not change
the shape of the apo RXRa LBP. In addition, K-8008 interacts
with monomers of each dimer in the tetramer, contributing to
the dimer-dimer interaction. Taken together, K-8008 or K-8012
binding may help to stabilize the tetramer. Stabilizing the tetra-
meric state of RXRa through ligand binding may have important
implications for the regulation of the nongenomic biological
activities of RXRa. Previous studies have demonstrated that
tetramer formation of RXRa serves as a mechanism to suppress
its transactivation (Gampe et al., 2000; Kersten et al., 1998). The
fact that inhibition of tRXRa interaction with p85a by K-8008 and
K-8012 was associated with their binding to the RXRa tetramer
raises an intriguing question that the formation of RXRa homote-
tramer may also represent a mechanism to suppress its interac-
tion with cytoplasmic p85a and activation of tRXRa-dependent
PI3K/AKT signaling.
EXPERIMENTAL PROCEDURES
Compound Synthesis
K-8008 and K-8012 were synthesized using scheme of Figure 1B. See Supple-
mental Information for details.
Cell Culture and Transfection
PC3 prostate cancer, ZR-75-1 breast cancer, and HeLa cervical cancer cells
were grown in RPMI 1640. CV-1 African green monkey kidney, HCT-116 colon
cancer, and A549 lung cancer cells were cultured in DMEM containing 10%
fetal bovine serum. The cells were maintained at 5% CO2 at 37ꢀC. Subconflu-
ent 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. Myc-RXRa-r80 and Flag-
p85a expression vectors as well as RXRa siRNA were described (Zhou
et al., 2010). RXRa-L433E and RXRa-F438,9A mutants were constructed by
standard procedure, and their LBDs were cloned into pBind expression vector
(Wang et al., 2013).
Unlike classical nuclear receptor antagonists that bind to
the RXRa LBP with high affinity, weak antagonism is commonly
observed for antagonists that target the receptor surface binding
sites (Caboni et al., 2012; Gunther et al., 2008; Hwang et al.,
2009; Rodriguez et al., 2004). Similarly, K-8008 and K-8012
bind to a surface hydrophobic site and display weak antagonist
effects. However, the therapeutic relevance of targeting the
RXRa through this binding site is evidenced by our observation
that both K-8008 and K-8012 could inhibit tRXRa activities in
cancer cells in vitro and tumor growth in animals (Figure 4). In
fact, the existence and therapeutic relevance of the novel sites
other than the LBP have been described for several nuclear
receptors, including estrogen receptor, androgen receptor,
and vitamin D receptor (Caboni et al., 2012; Kojetin et al.,
CAT Assay
(TREpal)2-tk-CAT (100 ng), b-galactosidase (100 ng), and RXRa (20 ng) were
transiently transfected into CV-1 cells (Zhang et al., 1992). Cells were then
treated with or without 9-cis-RA (10À7 M) in the presence or absence of
increasing concentrations of compounds for an additional 24 hr. Cells were
harvested and assayed for CAT and b-galactosidase (b-gal) activity. To
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved
9

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
normalize for transfection efficiency, CAT activities were corrected to b-gal
cutaneously with 100 ml of HepG2 cells (2 3 10À6). For drug treatment, mice
(n = 6) were treated intraperitoneally after 7 days of transplantation with
K-8008 (20 mg/kg), K-80003 (20 mg/kg), or vehicle (Tween 80) once a day.
Body weight and tumor size were measured every 3 days. Mice were
scarified after 12-day drug treatment, and the tumors were removed for
various assessments. All experimentations and animal usage were per-
formed and approved by the Animal Care and Use Committee of Xiamen
University.
activities.
Mammalian One Hybrid
HCT-116 cells seeded in 24-well plates were transiently transfected with 50 ng
of pG5luc, 25 ng of pBind-RXRa-LBD, or mutant. Twenty-four hours after
transfection, the medium was replaced by medium containing the sulindac
analogs and/or 9-cis-RA. Cells were washed, lysed, and assayed by using
the Dual-Luciferase Reporter Assay System (Promega). Transfection effi-
ciency was normalized to Renilla luciferase activity.
Histology and Apoptosis Analysis
Paraffin wax-embedded tumors were cut into 5-mm-thick sections. These
sections were deparaffinized and stained with hematoxylin and eosin (H&E)
according to the standard protocol. Tumor sections of HepG2 xenografts
were also stained with TUNEL for assessing spontaneous apoptosis
according to the manufacturer’s instructions (In situCell Death Detection
Kit; Roche). The images were taken under a fluorescent microscope (Carl
Zeiss).
Protein Expression and Purification
The human RXRa LBD (residues Thr223 to Thr462) was cloned as an N-terminal
histidine-tagged fusion protein in pET15b expression vector and overproduced
in Escherichia coli BL21 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. The eluent was collected and concen-
trated to 5 mg/ml for subsequent trails (Bourguet et al., 1995; Peet et al.,
1998). For crystallization experiment, the His tag was cleaved by bovine
thrombin (Sigma) and removed on the Resource-Q column (GE Healthcare),
using 0.1–1.0 M NaCl gradient and the Tris-Cl buffer (pH 8.0). The additional pu-
rification was done by the gel filtration on a Superdex 200 2660 column (GE
Healthcare)pre-equilibratedwiththe75mMNaCl, 20mMTris-Cl buffer(pH 8.0).
Crystallization and Structure Solution of the RXR LBD/Ligand
Complexes
The initial crystallization conditions were determined using the sitting-drop
vapor-diffusion method and the crystallization screens Index and PEG-Ion
(Hampton research). Other crystallization chemicals were from Hampton
Research and Sigma. The data were collected from crystals grown in sitting
drops of the 96-well Intelli-Plates (ARI) by the vapor-diffusion method. 0.2 ml
of the protein/ligand complex containing 0.37 mM of RXR LBD, 0.5-0.7 mM
of a ligand, 100 mM NaCl, and 20 mM Tris-Cl buffer (pH 8.0) were mixed
with 0.2 ml of the well solution (20% PEG3330 and 0.2 M magnesium formate
for the K-8008 complex or 0.2 M sodium acetate for the K-8012 complex) and
incubated at 20ꢀC. The first crystals appeared in 5–10 days and grew within
same amount of time into 0.2 3 0.2 3 0.05 mm plates. The crystals were
flash-frozen against the well solution containing 20% PEG400 as a cryoprotec-
tant. The diffraction data were collected from the cryo-cooled crystals (at
100 K) at Beamline BL11-2 of SSRL and processed using the program suites
XDS (Kabsch, 2010) and ccp4i (Collaborative Computational Project, Number
4, 1994).
Ligand-Binding Competition Assay
The His-tagged human RXRa-LBD(223–462) was incubated in tubes with un-
labeled 9-cis-RA or different concentrations of compounds in 200 ml of binding
buffer (0.15 M KCl, 10 mM Tris,HCl [pH 7.4], 8% glycerol, and 0.5% CHAPS
detergent) at 4ꢀC for 1 hr. [3H]9-cis-RA was added to the tubes to final concen-
tration of 7.5 nM and final volume of 300 ml and incubated overnight at 4ꢀC. The
RXRa-LBD was captured by nickel-coated beads. Bound [3H]9-cis-RA was
quantitated by liquid scintillation counting.
TR-FRET Retinoic X Receptor a Coactivator Assay
Invitrogen’s LanthsScreen TR-FRET RXRa Coactivator Assay was conducted
according to the manufacturer’s protocol. The TR-FRET ratio was calculated
by dividing the emission signal at 520 nm by the emission signal at 495 nm.
The structures were solved by the molecular replacement program Phaser
(McCoy et al., 2007) using PDB entry 1G1U as an initial model. The model
rebuilding and refinement were done with Coot (Emsley and Cowtan, 2004)
and the program suite Phenix (Adams et al., 2010). The initial models and
parameter files for the ligands were prepared by eLBOW of Phenix. The data
collection and refinement statistics are presented in Table S1.
MTT Assay
Confluent cells cultured in 96-well dishes were treated with various concentra-
tions of compounds for 48 hr. The cells were then incubated with 2 mg/ml
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) for 4 hr
at 37ꢀC. MTT solution was then aspirated, and formazan in cells was instantly
dissolved by addition of 150 ml of DMSO each well. Absorbance was measured
at 570 nm.
Data Analyses
Data are expressed as means SD from three or more experiments. Statistical
analysis was performed using Student’s t test. Differences were considered
statistically significant with p < 0.05.
Western Blotting
Cells were lysed, and equal amounts of the lysates were electrophoresed on
10% SDS-PAGE gels and transferred onto polyvinylidene difluoride mem-
branes (Millipore). The membranes were blocked with 5% skimmed milk in
TBST (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, and 0.1% Tween 20) for 1 hr
and then incubated with primary antibodies and secondary antibodies and de-
tected using the ECL system (Thermo). The dilutions of the primary antibodies
were anti-RXRa (rN197; Santa Cruz) in 1:1,000, anti-PARP (H-250; Santa Cruz)
in 1:3,000, anti-p85a (Millipore) in 1:1,000, anti-p-AKT (D9E; Cell Signaling
Technology) in 1:1,000, anti-AKT1/2/3 (H-136; Santa Cruz) in 1:1,000, anti-
b-actin (Sigma) in 1:5,000, anti-c-myc (9E10; Santa Cruz), and anti-Flag
(F1804; Sigma).
ACCESSION NUMBERS
The coordinates for the crystal structures of RXRa LBD in complex with K-8008
or K-8012 were deposited with the Protein Data Bank under accession codes
4N8R and 4N5G, respectively.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
six figures, and one table and can be found with this article online at http://
dx.doi.org/10.1016/j.chembiol.2014.02.017.
Coimmunoprecipitation Assay
Cells were harvested and lysed in buffer containing 50 mM HEPES-NaOH
(pH 7.5), 2.5 mM EDTA, 100 mM NaCl, 0.5% NP40, and 10% glycerol, with
1 mM dithiothreitol and proteinase inhibitor cocktail. Immunoprecipitation
was performed as described (Zhou et al., 2010).
AUTHOR CONTRIBUTIONS
X.-K.Z., Y.S., P.Q.H., and R.C.L. supervised the project; L.C, Z.G.W., A.E.A,
P.Q.H., X.-K.Z., and Y.S. designed the study and experiments; L.C., F.C.,
J.C., F.J., G.A., Y.M., and M.H. conducted the biological and animal experi-
ments with input from H.Z. and Z.P.Z.; A.E.A. carried out the crystallization,
diffraction data collection, and structure determination with the help of G.C.
on protein purification. Z.G.W. and J.F.Z. synthesized the compounds; L.C.,
HepG2 Xenografts
Nude mice (BALB/c, SPF grade, 16–18 g, 4–5 weeks old) were housed
at 28ꢀC in a laminar flow under sterilized conditions. Mice were injected sub-
10 Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
Z.G.W., A.E.A., P.Q.H., X.-K.Z., and Y.S. analyzed the data; and X.-K.Z. and
Y.S. wrote the manuscript with input from all authors.
de Lera, A.R., Bourguet, W., Altucci, L., and Gronemeyer, H. (2007). Design of
selective nuclear receptor modulators: RAR and RXR as a case study. Nat.
Rev. Drug Discov. 6, 811–820.
Dufour, J.M., and Kim, K.H. (1999). Cellular and subcellular localization of six
retinoid receptors in rat testis during postnatal development: identification of
potential heterodimeric receptors. Biol. Reprod. 61, 1300–1308.
ACKNOWLEDGMENTS
This work was supported by grants from the U.S. Army Medical Research and
Material Command (W81XWH-11-1-0677), the National Institutes of Health
(CA140980, GM089927, and CA179379), the National Natural Science Foun-
dation of China (NSFC-91129302 and NSFC-21332007), the Ministry of Edu-
cation of China (IRT1037), the Science and Technology Bureau of Xiamen
(3502Z20133008), and the National Basic Research Program (973 Program)
of the Ministry of Science and Technology, China (2010CB833200).
Duggan, D.E., Hooke, K.F., and Hwang, S.S. (1980). Kinetics of the tissue dis-
tributions of sulindac and metabolites. Relevance to sites and rates of bio-
activation. Drug Metab. Dispos. 8, 241–246.
Egea, P.F., Mitschler, A., Rochel, N., Ruff, M., Chambon, P., and Moras, D.
(2000). Crystal structure of the human RXRalpha ligand-binding domain bound
to its natural ligand: 9-cis retinoic acid. EMBO J. 19, 2592–2601.
Emsley, P., and Cowtan, K. (2004). Coot: model-building tools for molecular
Received: October 21, 2013
Revised: January 22, 2014
Accepted: February 19, 2014
Published: April 3, 2014
graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132.
Fukunaka, K., Saito, T., Wataba, K., Ashihara, K., Ito, E., and Kudo, R. (2001).
Changes in expression and subcellular localization of nuclear retinoic acid
receptors in human endometrial epithelium during the menstrual cycle. Mol.
Hum. Reprod. 7, 437–446.
REFERENCES
Gampe, R.T., Jr., Montana, V.G., Lambert, M.H., Wisely, G.B., Milburn, M.V.,
and Xu, H.E. (2000). Structural basis for autorepression of retinoid X receptor
by tetramer formation and the AF-2 helix. Genes Dev. 14, 2229–2241.
´
Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N.,
Headd, J.J., Hung, L.W., Kapral, G.J., Grosse-Kunstleve, R.W., et al. (2010).
PHENIX: a comprehensive Python-based system for macromolecular struc-
ture solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221.
Germain, P., Chambon, P., Eichele, G., Evans, R.M., Lazar, M.A., Leid, M., De
Lera, A.R., Lotan, R., Mangelsdorf, D.J., and Gronemeyer, H. (2006).
International Union of Pharmacology. LXIII. Retinoid X receptors. Pharmacol.
Rev. 58, 760–772.
Arnold, L.A., Este´ banez-Perpin˜ a´ , E., Togashi, M., Jouravel, N., Shelat, A.,
McReynolds, A.C., Mar, E., Nguyen, P., Baxter, J.D., Fletterick, R.J., et al.
(2005). Discovery of small molecule inhibitors of the interaction of the thyroid
hormone receptor with transcriptional coregulators. J. Biol. Chem. 280,
43048–43055.
Gunther, J.R., Moore, T.W., Collins, M.L., and Katzenellenbogen, J.A. (2008).
Amphipathic benzenes are designed inhibitors of the estrogen receptor a/ste-
roid receptor coactivator interaction. ACS Chem. Biol. 3, 282–286.
Herr, R.J. (2002). 5-Substituted-1H-tetrazoles as carboxylic acid isosteres:
medicinal chemistry and synthetic methods. Bioorg. Med. Chem. 10, 3379–
3393.
Balkwill, F. (2009). Tumour necrosis factor and cancer. Nat. Rev. Cancer 9,
361–371.
Bourguet, W., Ruff, M., Bonnier, D., Granger, F., Boeglin, M., Chambon, P.,
Moras, D., and Gronemeyer, H. (1995). Purification, functional characteriza-
tion, and crystallization of the ligand binding domain of the retinoid X receptor.
Protein Expr. Purif. 6, 604–608.
Hwang, J.Y., Arnold, L.A., Zhu, F., Kosinski, A., Mangano, T.J., Setola, V.,
Roth, B.L., and Guy, R.K. (2009). Improvement of pharmacological properties
of irreversible thyroid receptor coactivator binding inhibitors. J. Med. Chem.
52, 3892–3901.
Bushue, N., and Wan, Y.-J.Y. (2010). Retinoid pathway and cancer therapeu-
Kabsch, W. (2010). Xds. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132.
tics. Adv. Drug Deliv. Rev. 62, 1285–1298.
Kersten, S., Kelleher, D., Chambon, P., Gronemeyer, H., and Noy, N. (1995).
Retinoid X receptor alpha forms tetramers in solution. Proc. Natl. Acad. Sci.
USA 92, 8645–8649.
Buzo´ n, V., Carbo´ , L.R., Estruch, S.B., Fletterick, R.J., and Este´ banez-Perpin˜ a´ ,
E. (2012). A conserved surface on the ligand binding domain of nuclear recep-
tors for allosteric control. Mol. Cell. Endocrinol. 348, 394–402.
Kersten, S., Dong, D., Lee, W., Reczek, P.R., and Noy, N. (1998). Auto-
Caboni, L., Kinsella, G.K., Blanco, F., Fayne, D., Jagoe, W.N., Carr, M.,
Williams, D.C., Meegan, M.J., and Lloyd, D.G. (2012). ‘‘True’’ antiandro-
gens-selective non-ligand-binding pocket disruptors of androgen receptor-
coactivator interactions: novel tools for prostate cancer. J. Med. Chem. 55,
1635–1644.
silencing by the retinoid X receptor. J. Mol. Biol. 284, 21–32.
Kojetin, D.J., Burris, T.P., Jensen, E.V., and Khan, S.A. (2008). Implications of
the binding of tamoxifen to the coactivator recognition site of the estrogen
receptor. Endocr. Relat. Cancer 15, 851–870.
Kolluri, S.K., Corr, M., James, S.Y., Bernasconi, M., Lu, D., Liu, W., Cottam,
H.B., Leoni, L.M., Carson, D.A., and Zhang, X.K. (2005). The R-enantiomer
of the nonsteroidal antiinflammatory drug etodolac binds retinoid X receptor
and induces tumor-selective apoptosis. Proc. Natl. Acad. Sci. USA 102,
2525–2530.
Cao, X., Liu, W., Lin, F., Li, H., Kolluri, S.K., Lin, B., Han, Y.-H., Dawson, M.I.,
and Zhang, X.K. (2004). Retinoid X receptor regulates Nur77/TR3-dependent
apoptosis [corrected] by modulating its nuclear export and mitochondrial tar-
geting. Mol. Cell. Biol. 24, 9705–9725.
Casas, F., Daury, L., Grandemange, S., Busson, M., Seyer, P., Hatier, R.,
Carazo, A., Cabello, G., and Wrutniak-Cabello, C. (2003). Endocrine regulation
of mitochondrial activity: involvement of truncated RXRalpha and c-Erb
Aalpha1 proteins. FASEB J. 17, 426–436.
Lazebnik, Y.A., Kaufmann, S.H., Desnoyers, S., Poirier, G.G., and Earnshaw,
W.C. (1994). Cleavage of poly(ADP-ribose) polymerase by a proteinase with
properties like ICE. Nature 371, 346–347.
Lu, J., Dawson, M.I., Hu, Q.Y., Xia, Z., Dambacher, J.D., Ye, M., Zhang, X.-K.,
and Li, E. (2009). The effect of antagonists on the conformational exchange of
the retinoid X receptor alpha ligand-binding domain. Magn. Reson. Chem. 47,
1071–1080.
Chen, Z.-P., Iyer, J., Bourguet, W., Held, P., Mioskowski, C., Lebeau, L., Noy,
N., Chambon, P., and Gronemeyer, H. (1998). Ligand- and DNA-induced
dissociation of RXR tetramers. J. Mol. Biol. 275, 55–65.
Collaborative Computational Project, Number 4 (1994). The CCP4 suite: pro-
grams for protein crystallography. Acta Crystallogr. D Biol. Crystallogr. 50,
760–763.
Matta, C.F., Arabi, A.A., and Weaver, D.F. (2010). The bioisosteric similarity of
the tetrazole and carboxylate anions: clues from the topologies of the electro-
static potential and of the electron density. Eur. J. Med. Chem. 45, 1868–1872.
Davies, N.M., and Watson, M.S. (1997). Clinical pharmacokinetics of sulindac.
McCoy, A.J., Grosse-Kunstleve, R.W., Adams, P.D., Winn, M.D., Storoni, L.C.,
and Read, R.J. (2007). Phaser crystallographic software. J. Appl. Crystallogr.
40, 658–674.
A dynamic old drug. Clin. Pharmacokinet. 32, 437–459.
Dawson, M.I., and Zhang, X.K. (2002). Discovery and design of retinoic acid re-
ceptor and retinoid X receptor class- and subtype-selective synthetic analogs
of all-trans-retinoic acid and 9-cis-retinoic acid. Curr. Med. Chem. 9, 623–637.
Mita, Y., Dodo, K., Noguchi-Yachide, T., Miyachi, H., Makishima, M.,
Hashimoto, Y., and Ishikawa, M. (2010). LXXLL peptide mimetics as inhibitors
Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved 11

Please cite this article in press as: Chen et al., Sulindac-Derived RXRa Modulators Inhibit Cancer Cell Growth by Binding to a Novel Site, Chemistry &
Biology (2014), http://dx.doi.org/10.1016/j.chembiol.2014.02.017
Chemistry & Biology
Sulindac Analogs Bind to an RXRa Site
of the interaction of vitamin D receptor with coactivators. Bioorg. Med. Chem.
Wang, Y., Chirgadze, N.Y., Briggs, S.L., Khan, S., Jensen, E.V., and Burris, T.P.
(2006). A second binding site for hydroxytamoxifen within the coactivator-
binding groove of estrogen receptor b. Proc. Natl. Acad. Sci. USA 103,
9908–9911.
Lett. 20, 1712–1717.
Mizwicki, M.T., Keidel, D., Bula, C.M., Bishop, J.E., Zanello, L.P., Wurtz, J.M.,
Moras, D., and Norman, A.W. (2004). Identification of an alternative ligand-
binding pocket in the nuclear vitamin D receptor and its functional importance
in 1alpha,25(OH)2-vitamin D3 signaling. Proc. Natl. Acad. Sci. USA 101,
12876–12881.
Wang, Z.-G., Chen, L., Chen, J., Zheng, J.-F., Gao, W., Zeng, Z., Zhou, H.,
Zhang, X.K., Huang, P.-Q., and Su, Y. (2013). Synthesis and SAR study of
modulators inhibiting tRXRa-dependent AKT activation. Eur. J. Med. Chem.
62, 632–648.
Moore, T.W., Mayne, C.G., and Katzenellenbogen, J.A. (2010). Minireview: Not
picking pockets: nuclear receptor alternate-site modulators (NRAMs). Mol.
Endocrinol. 24, 683–695.
Weggen, S., Eriksen, J.L., Das, P., Sagi, S.A., Wang, R., Pietrzik, C.U., Findlay,
K.A., Smith, T.E., Murphy, M.P., Bulter, T., et al. (2001). A subset of NSAIDs
lower amyloidogenic Abeta42 independently of cyclooxygenase activity.
Nature 414, 212–216.
Peet, D.J., Doyle, D.F., Corey, D.R., and Mangelsdorf, D.J. (1998). Engineering
novel specificities for ligand-activated transcription in the nuclear hormone
receptor RXR. Chem. Biol. 5, 13–21.
Yamamoto, Y., Yin, M.-J., Lin, K.-M., and Gaynor, R.B. (1999). Sulindac in-
Rodriguez, A.L., Tamrazi, A., Collins, M.L., and Katzenellenbogen, J.A. (2004).
Design, synthesis, and in vitro biological evaluation of small molecule inhibitors
of estrogen receptor a coactivator binding. J. Med. Chem. 47, 600–611.
hibits activation of the NF-kappaB pathway. J. Biol. Chem. 274, 27307–27314.
Yen, W.C., and Lamph, W.W. (2006). A selective retinoid X receptor agonist
bexarotene (LGD1069, Targretin) prevents and overcomes multidrug resis-
tance in advanced prostate cancer. Prostate 66, 305–316.
Sato, Y., Ramalanjaona, N., Huet, T., Potier, N., Osz, J., Antony, P., Peluso-
´
Iltis, C., Poussin-Courmontagne, P., Ennifar, E., Mely, Y., et al. (2010). The
Zhang, X.K., Lehmann, J., Hoffmann, B., Dawson, M.I., Cameron, J.,
Graupner, G., Hermann, T., Tran, P., and Pfahl, M. (1992). Homodimer forma-
tion of retinoid X receptor induced by 9-cis retinoic acid. Nature 358, 587–591.
‘‘Phantom Effect’’ of the rexinoid LG100754: structural and functional insights.
PLoS One 5, e15119.
Sun, A., Moore, T.W., Gunther, J.R., Kim, M.S., Rhoden, E., Du, Y., Fu, H.,
Snyder, J.P., and Katzenellenbogen, J.A. (2011). Discovering small-molecule
estrogen receptor a/coactivator binding inhibitors: high-throughput screening,
ligand development, and models for enhanced potency. ChemMedChem 6,
654–666.
Zhang, L., Yu, J., Park, B.H., Kinzler, K.W., and Vogelstein, B. (2000). Role of
BAX in the apoptotic response to anticancer agents. Science 290, 989–992.
Zhou, H., Liu, W., Su, Y., Wei, Z., Liu, J., Kolluri, S.K., Wu, H., Cao, Y., Chen, J.,
Wu, Y., et al. (2010). NSAID sulindac and its analog bind RXRalpha and inhibit
RXRalpha-dependent AKT signaling. Cancer Cell 17, 560–573.
Szanto, A., Narkar, V., Shen, Q., Uray, I.P., Davies, P.J.A., and Nagy, L. (2004).
Retinoid X receptors: X-ploring their (patho)physiological functions. Cell Death
Differ. 11 (Suppl 2), S126–S143.
Zimmerman, T.L., Thevananther, S., Ghose, R., Burns, A.R., and Karpen, S.J.
(2006). Nuclear export of retinoid X receptor alpha in response to interleukin-
1beta-mediated cell signaling: roles for JNK and SER260. J. Biol. Chem.
281, 15434–15440.
Wang, X., and Lin, Y. (2008). Tumor necrosis factor and cancer, buddies or
foes? Acta Pharmacol. Sin. 29, 1275–1288.
12 Chemistry & Biology 21, 1–12, May 22, 2014 ª2014 Elsevier Ltd All rights reserved