In silico identification of an aryl hydrocarbon receptor antagonist with biological activity in vitro and in vivo

Article abstract of DOI:10.1124/mol.114.093369

The aryl hydrocarbon receptor (AHR) is critically involved in several physiologic processes, including cancer progression and multiple immune system activities. We, and others, have hypothesized that AHR modulators represent an important new class of targeted therapeutics. Here, ligand shape-based virtual modeling techniques were used to identify novel AHR ligands on the basis of previously identified chemotypes. Four structurally unique compounds were identified. One lead compound, 2-((2-(5-bromofuran-2-yl)-4-oxo-4H-chromen-3-yl)oxy) acetamide (CB7993113), was further tested for its ability to block three AHR-dependent biologic activities: triple-negative breast cancer cell invasion or migration in vitro and AHR ligand-induced bone marrow toxicity in vivo. CB7993113 directly bound both murine and human AHR and inhibited polycyclic aromatic hydrocarbon (PAH)- and TCDD-induced reporter activity by 75% and 90% respectively. A novel homology model, comprehensive agonist and inhibitor titration experiments, and AHR localization studies were consistent with competitive antagonism and blockade of nuclear translocation as the primary mechanism of action. CB7993113 (IC₅₀ 3.3 × 10^-7 M) effectively reduced invasion of human breast cancer cells in three-dimensional cultures and blocked tumor cell migration in two-dimensional cultures without significantly affecting cell viability or proliferation. Finally, CB7993113 effectively inhibited the bone marrow ablative effects of 7,12-dimethylbenz[a]anthracene in vivo, demonstrating drug absorption and tissue distribution leading to pharmacological efficacy. These experiments suggest that AHR antagonists such as CB7993113 may represent a new class of targeted therapeutics for immunomodulation and/or cancer therapy.

Full text of DOI:10.1124/mol.114.093369

Supplemental material to this article can be found at:
http://molpharm.aspetjournals.org/content/suppl/2014/08/26/mol.114.093369.DC1.html
1521-0111/86/5/593–608$25.00
MOLECULAR PHARMACOLOGY
http://dx.doi.org/10.1124/mol.114.093369
Mol Pharmacol 86:593–608, November 2014
Copyright ª 2014 by The American Society for Pharmacology and Experimental Therapeutics
In Silico Identification of an Aryl Hydrocarbon Receptor
Antagonist with Biological Activity In Vitro and In Vivo s
Ashley J. Parks, Michael P. Pollastri, Mark E. Hahn, Elizabeth A. Stanford, Olga Novikov,
Diana G. Franks, Sarah E. Haigh, Supraja Narasimhan, Trent D. Ashton, Timothy G. Hopper,
Dmytro Kozakov, Dimitri Beglov, Sandor Vajda, Jennifer J. Schlezinger,
and David H. Sherr
Molecular Medicine Program, Boston University School of Medicine, Boston, Massachusetts (A.J.P., E.A.S., O.N.); Department
of Environmental Health, Boston University School of Public Health, Boston, Massachusetts (A.J.P., E.A.S., O.N., S.N., J.J.S.,
DHS); Department of Chemistry and Chemical Biology, Northeastern University, Boston, Massachusetts (M.P.P., T.G.H.);
Department of Chemistry, Boston University (T.D.A.); Biology Department, Woods Hole Oceanographic Institution, Woods Hole,
Massachusetts (M.E.H., D.G.F.); Wake Forest Innovations, Wake Forest University, Winston-Salem, North Carolina (S.E.H.); and
Biomedical Engineering, Boston University, Boston, Massachusetts (D.K., D.B., S.V.)
Received April 28, 2014; accepted August 22, 2014
ABSTRACT
The aryl hydrocarbon receptor (AHR) is critically involved in 90% respectively. A novel homology model, comprehensive
several physiologic processes, including cancer progression agonist and inhibitor titration experiments, and AHR localization
and multiple immune system activities. We, and others, have studies were consistent with competitive antagonism and
hypothesized that AHR modulators represent an important blockade of nuclear translocation as the primary mechanism
new class of targeted therapeutics. Here, ligand shape–based of action. CB7993113 (IC50 3.3 Â 1027 M) effectively reduced
virtual modeling techniques were used to identify novel AHR invasion of human breast cancer cells in three-dimensional
ligands on the basis of previously identified chemotypes. Four cultures and blocked tumor cell migration in two-dimensional
structurally unique compounds were identified. One lead com- cultures without significantly affecting cell viability or pro-
pound, 2-((2-(5-bromofuran-2-yl)-4-oxo-4H-chromen-3-yl)oxy) liferation. Finally, CB7993113 effectively inhibited the bone
acetamide (CB7993113), was further tested for its ability to block marrow ablative effects of 7,12-dimethylbenz[a]anthracene in
three AHR-dependent biologic activities: triple-negative breast vivo, demonstrating drug absorption and tissue distribution
cancer cell invasion or migration in vitro and AHR ligand–induced leading to pharmacological efficacy. These experiments sug-
bone marrow toxicity in vivo. CB7993113 directly bound both gest that AHR antagonists such as CB7993113 may represent
murine and human AHR and inhibited polycyclic aromatic hydro- a new class of targeted therapeutics for immunomodulation
carbon (PAH)– and TCDD-induced reporter activity by 75% and and/or cancer therapy.
encoding a battery of cytochrome P450 enzymes that metab-
Introduction
olize at least some of those ligands into metabolic intermedi-
The aryl hydrocarbon receptor (AHR) field has undergone
ates, some of which are toxic (Hankinson et al., 1991; Nebert
a dramatic paradigm shift in the last few years. Historically,
the evolutionarily conserved AHR was studied for its ability,
upon activation by environmental ligands, to regulate genes
et al., 1991). Similarly, studies published over the last 20–
30 years demonstrate the role of the AHR in the initiation of
environmental chemical–induced cancers, to a large extent
through P450-dependent generation of mutagenic intermedi-
ates. Consequently, studies on the effects of AHR activation,
primarily by anthropogenic ligands, most frequently involved
tissue toxicity or the initiation of malignancy.
However, several landmark studies now demonstrate that
the AHR plays a critical role in several physiologic processes in
the absence of environmental chemicals. These studies show
This work was supported by the National Institutes of Health National
Institute of Environmental Health Sciences [R01-ES11624, P42-ES007381,
R01-ES006272] and National Institute of General Medical Sciences [R01-GM064700];
the Art beCAUSE Breast Cancer Foundation; The Mary Kay Foundation;
and a Boston University Ignition Award.
J.J.S and D.H.S. contributed equally to this manuscript.
dx.doi.org/10.1124/mol.114.093369.
s
This article has supplemental material available at molpharm.
aspetjournals.org.
ABBREVIATIONS: AHR, Aryl hydrocarbon receptor; AHRE, aryl hydrocarbon receptor response element; AHRR, aryl hydrocarbon receptor
repressor; b-NF, beta-naphthoflavone; CC, consensus cluster; CB7993113, 2-((2-(5-bromofuran-2-yl)-4-oxo-4H-chromen-3-yl)oxy)acetamide;
CH223191, (E)-1-methyl-N-(2-methyl-4-(o-tolyldiazenyl)phenyl)-1H-pyrazole-5-carboxamide; CMV, cytomegalovirus; DMBA, 7,12-dimethylbenz[a]
anthracene; DMSO, dimethyl sulfoxide; EGFP, enhanced green fluorescent protein; IC50, inhibitory concentration 50%; MTT, 3-(4,5-dimethylthiazol-
2-yl)-2,5-diphenyltetrazolium bromide; P450, cytochrome P450; PDB, Protein Database; PPARg, peroxisome proliferator–activated receptor
gamma; ROCS, rapid overlay of chemical structures; SR1, StemRegenin 1; TCDD, 2,3,7,8-tetrachlorodibenzo-p-dioxin; TMF, 6,29,49-
trimethoxyflavone.
593

594
Parks et al.
that the AHR, presumably activated by endogenous ligands, Lead compounds chosen from the in vitro screening assays were
plays a critical role in the manifestation or control of tissue characterized for their ability to directly bind the AHR and to
inflammation and autoimmunity through enforcement of in- block AHR nuclear translocation and transcriptional activity.
flammatory cytokine production by monocytes (Wu et al., 2011) One lead compound, CB7993113, was examined for its probable
and synoviocytes (Lahoti et al., 2013), the production of in- binding conformation to the AHR PAS-B domain. Finally,
flammatory Th17 and immunosuppressive regulatory T cells CB7993113 was tested for its ability to block three AHR-
(Veldhoen et al., 2008), and the regulation of apoptosis (Caruso dependent biologic activities, triple-negative breast cancer cell
et al., 2004, 2006). Indeed, the recent demonstration that AHR invasion and migration in vitro, and AHR ligand–induced bone
activation drives production of erythroid cells from pluripotent marrow toxicity in vivo.
stem cell precursors (Smith et al., 2013), is required for
development of gut-associated T cells (Kiss et al., 2011; Lee
et al., 2012), and influences the differentiation of human
hematopoietic stem cells (Boitano et al., 2010) suggests
a critical role of the AHR in hematopoiesis in general.
Furthermore, several studies now implicate the AHR in
cancer progression in the absence of environmental ligands.
For example, AHR activated by an endogenous tryptophan–
derived metabolite increases human glioblastoma cell survival
and migration (Opitz et al., 2011). The AHR-repressor (AHRR)
protein acts as a tumor suppressor gene in several human cancers
(Zudaire et al., 2008). AHR expression and “constitutive” (endog-
enous ligand–driven) activity in breast cancer cells correlate with
tumor aggressiveness (Schlezinger et al., 2006; Yang et al., 2008)
and control expression of genes associated with tumor invasion
(Yang et al., 2005). Surveys of cytochrome P450 protein expression
identified the AHR gene targets CYP1A1, CYP1A2, and CYP1B1 in
most types of primary human tumor samples examined, including
solid and hematologic malignancies (Maecker et al., 2003), and
found that expression of the AHR target gene CYP2S1 inversely
correlates with patient survival (Murray GI et al., 2010). Finally,
ectopic AHR expression in nonmalignant human mammary
Materials and Methods
Chemical Reagents
Commercial chemical libraries of test compounds were acquired from
ChemBridge Corporation (San Diego, CA) and Enamine (Kiev, Ukraine).
Dimethyl sulfoxide (DMSO), b-naphthoflavone (b-NF), 7,12-dimethyl-
benz[a]anthracene (DMBA), TCDD, and other chemical reagents were
obtained from Sigma-Aldrich (St. Louis, MO) unless otherwise indicated.
CB7993113 and CH223191 were synthesized as below. All compounds
submitted for biologic testing were deemed at least .98% pure by HPLC
(with UV and mass spectral detection) and 1H NMR.
Chemical Synthesis of CB7993113. 1-(2-Hydroxyphenyl)ethanone
(1.720 g, 14.29 mmol) and 5-bromofuran-2-carbaldehyde (2.5 g,
14.29 mmol) were dissolved in a round-bottom flask in 10 ml of
ethanol. A solution of 17 M NaOH (1.5 ml) in water was then added
under vigorous stirring. Precipitate formed under addition of base, and
a thick paste was formed. The mixture was stirred for 24 hours at room
temperature. Ethanol (50 ml) was added with 0.5 ml of 2.5 M NaOH.
The mixture was cooled to 15°C and hydrogen peroxide was added (35%
in water, 6.25 ml, 71.4 mmol). After 4 hours, dilute sulfuric acid was
added to neutralize to pH 7.0, and the reaction was poured into 250 ml
of water and stirred for 2 hours. The solid material was collected by
epithelial cells induces an epithelial-to-mesenchymal transition filtration and dried under vacuum to give a yellow solid (1.01 g, 23%
yield). 1 H NMR (399 MHz, DMSO-d6) d ppm 4.60 (s, 2 H) 6.98 (d, J 5
3.66 Hz, 1 H) 7.39 (br s, 1 H) 7.50 (t, J 5 7.33 Hz, 1 H) 7.67–7.78 (m, 3 H)
7.81 (d, J 5 7.33 Hz, 1 H) 8.08 (d, J 5 8.06 Hz, 1 H).) A mixture of
2-(5-bromofuran-2-yl)-3-hydroxy-4H-chromen-4-one thus obtained (910 mg,
2.96 mmol), 2-bromoacetamide (409 mg, 2.96 mmol), potassium carbonate
(1229 mg, 8.89 mmol), and dimethylformamide (30 ml) was stirred for
6 hours at 80°C. The solution was cooled and extracted with ethyl acetate.
The combined organic layers were washed with water, dried with sodium
sulfate, filtered and concentrated. Toluene was added and evaporated
repeatedly until dry crystals of crude product were formed. The resulting
and a .50% increase in cell growth rates (Brooks and Eltom,
2011). Together, these studies strongly support the hypothesis
that the AHR plays an important role in the later, more aggressive
stages of cancer, even in the absence of environmental ligands.
Given the involvement of the AHR in blood cell development
and multiple immune system phenomena, and its postulated role
in cancer progression, we and others have hypothesized that AHR
modulators, either agonists or antagonists, may represent an
important new class of targeted therapeutics (Schlezinger
et al., 2006; Zhang et al., 2009). We postulate that AHR material was purified by column chromatography, first eluting impurities
with 100% ethyl acetate, followed by 10% methanol in methylene chloride
to provide CB7993113 as a yellow solid (900 mg, 83% yield). 1H NMR
(399 MHz, DMSO-d6) d ppm 4.60 (s, 2 H) 6.98 (d, J 5 3.66 Hz, 1 H) 7.39
(br. s, 1 H) 7.50 (t, J 5 7.33 Hz, 1 H) 7.67–7.78 (m, 3 H) 7.81 (d, J 5 7.33 Hz,
1 H) 8.08 (d, J 5 8.06 Hz, 1 H). (ESI) found 363.9 [M 1 H]1.
Chemical Synthesis of CH223191 (Supplemental Fig. S1).
A solution containing 4-amino-29,3-dimethylazobenzene (602 mg, 2.67 mM),
1-methyl-1H-pyrazole-5-carboxylic acid (372 mg, 2.95 mmol,), N,N-diisopro-
pylethyl amine (1.4 ml, 8.04 mmol), and bromotripyrrolidinophosphonium
hexafluorophosphate (1.857 g, 3.98 mmol, 1.5 Eq) in 10 ml of 1,2-
antagonists in particular may be important for treatment of
high AHR expressing, triple-negative breast cancers (TNBCs),
malignancies which are particularly resistant to current chemo-
therapeutics and nonresponsive to hormone receptor–targeted
therapeutics.
The identification of novel, potent AHR modulators has been
hampered by the limited amount of data on the three-dimensional
structure of the AHR protein, and specifically the structure of its
ligand-binding domain (LBD). In its stead, researchers have
developed structural homology models on the basis of ligand- dichloroethane was heated via microwave irradiation in two equal
batches at 120°C for 22 minutes. The reaction mixture was reduced in
vacuo, taken up in ethyl acetate (100 ml), and washed using 5%
K2HPO4 (100 ml), saturated NaHCO3 (100 ml), and brine (100 ml). The
isolated organic phase was dried (Na2SO4), filtered, and concentrated
under reduced pressure to give a dark brown solid. The residue was
purified twice by rapidly stirring in the minimal amount of warm
CH2Cl2 and adding hexanes to precipitate the title compound (681 mg,
76%) as an amorphous tan solid. 1H NMR (300 MHz, CDCl3) d ppm 8.21
(d, J 5 8.7, 1 H), 7.86 (dd, J 5 8.7, 2,1, 1 H), 7.82 (br s, 1 H), 7.64 (br s, 1 H),
7.61 (br s, 1 H), 7.54 (d, J 5 2.1, 1 H), 7.39–7.32 (m, 2 H), 6.68 (d, J 5 2.1,
binding domains of familial proteins (Motto et al., 2011; Xing
et al., 2012). Although recent advancements in AHR-LBD
models have improved our understanding of the requirements
for AHR binding, the abilities of these programs to predict AHR
ligands is only beginning to be realized.
Here, we used ligand shape–based virtual screening tech-
niques to rapidly screen libraries of over 1 million commercially
available small molecule compounds for potential AHR ligands.
The focused library identified by this analysis was tested in
a high-throughput in vitro bioassay for AHR-antagonist activity. 1 H), 4.25 (s, 3 H), 2.73 (s, 3 H), 2.44 (s, 3 H). 13C NMR (100 MHz, CDCl3)

Identification of an AHR Antagonist
High-Throughput AHR Reporter Assay
595
d ppm 157.9, 150.7, 149.9, 138.1, 137.8, 137.4, 135.2, 131.3, 130.9, 129.0,
126.4, 124.8, 122.6, 122.2, 115.4, 106.6, 39.5, 17.9, 17.6. LCMS (C18):
tr(min) 5 2.01, (ESI) found 334.2 [M 1 H]1.
A previously described AHR reporter assay (Nagy et al., 2002) was
adapted for high-throughput screening. To each well of a 384-well
plate 105 H1G1.1c3 cells were added in selective medium and
incubated at 37°C for 24 hours. Culture medium was replaced with
nonselective medium consisting of MEMa (Life Technologies) supple-
mented with 10% BGS (HyClone/GE Healthcare Life Sciences, Logan,
UT) and 2 mM L-glutamine (Mediatech) prior to application of the test
compounds. b-NF was used as a positive control for induction of AHR
reporter activity. A b-NF standard curve was generated by applying
vehicle (0.1% DMSO) or final concentrations of 10210 to 1025 M b-NF
to the cultures, with each concentration applied to 24 wells. To assess
potential AHR agonist activity, vehicle or 2 ml of each of the 197
chemicals in DMSO diluted 1:10 in media (1029 to 1025 M, final
concentration) chosen from commercial libraries were applied to
triplicate wells. To screen for AHR antagonist activity, cells in 384-
well plates were treated with b-NF (1027 M) and 2 ml vehicle or test
compounds at the indicated concentrations. The plates were in-
cubated at 33°C for up to 72 hours. EGFP fluorescence was analyzed
at 24, 48, and 72 hours using a fluorometric plate reader (Spectra-
Fluor Plus; Tecan, Männedorf, Switzerland). Percent induction/
inhibition of b-NF–induced AHR activity was calculated by subtract-
ing the background fluorescence of untreated cells from all experi-
mental values and dividing the background-adjusted fluorescence in
the sample plus b-NF wells by the background-adjusted fluorescence
in the b-NF–alone wells and then multiplying by 100.
Cell Culture
H1G1.1c3 cells were generously provided by M. S. Denison (University
of California, Davis, CA) and maintained as previously described (Nagy
et al., 2002). Cultures of H1G1.1c3 cells were maintained in selective
medium consisting of DMEM (Mediatech, Manassas, VA) supplemented
with 10% bovine growth serum (HyClone, Logan, UT), 2 mM L-glutamine
(Mediatech), 5 mg/ml Plasmocin (InvivoGen, San Diego, CA), and 968 mg/l
G-418 sulfate (American Bioanalytical, Natick, MA) in a 37°C humidified
incubator in a 5% CO2 atmosphere. This murine hepatoma cell line
contains a stable enhanced green fluorescent protein (EGFP) reporter
construct regulated by AHR response elements (AHREs) derived from
the CYP1A1 promoter.
ER–, PR–, HER– BP1 cells were generously provided by Dr. J. Russo
(Fox Chase Cancer Center, Philadelphia, PA). BP1 cells were
maintained in phenol red-free DMEM-F/12 medium (Mediatech)
containing 5% equine serum (Sigma-Aldrich), 20 ng/ml of human
recombinant epidermal growth factor (Life Technologies, Grand
Island, NY), 0.5 mg/ml hydrocortisone (Sigma-Aldrich), 2 mM
L-glutamine, 100 IU penicillin per 100 mg/ml streptomycin (Mediatech),
10 mg/ml insulin (Sigma-Aldrich), and 5 mg/ml Plasmocin. SUM149
cells were graciously provided by Dr. S. Ethier of Wayne State
University (Detroit, MI), who isolated them from a primary in-
flammatory invasive ductal mammary carcinoma. Cells were
maintained in Ham’s F-12 medium (Mediatech) containing 5%
fetal bovine serum (Sigma-Aldrich), 0.5 mg/ml hydrocortisone, 2 mM
L-glutamine, 100 IU penicillin per 100 mg/ml streptomycin, 10 mg/ml
insulin, and 5 mg/ml Plasmocin. Hs578T-cell culture was described
previously (Yang et al., 2008). BP1, Hs578T, and SUM149 cells were
cultured and assayed at 37°C in a humidified incubator in a 5% CO2
atmosphere and grown as adherent monolayers at a maximum of
80% confluency.
Following the final fluorescence reading, the CellTitre-Blue cell
viability assay was used to determine toxicity as per the manufac-
turer’s instructions (Promega, Madison, WI). Toxicity was calculated
by dividing background-subtracted fluorescent readings of samples by
those of untreated cells. Compounds consistently inducing a $10%
increase in cell death, relative to baseline levels (generally 2–5%),
were excluded from further studies.
Antagonism of TCDD-Induced AHR Activity
H1G1.1c3 cells (6 Â 105) were added to each well of a 96-well plate
in selective medium and incubated at 37°C for 24 hours. Culture
medium was replaced with nonselective medium prior to application
of the test compounds. Eight culture wells were treated with 0.5%
DMSO (vehicle) or 1027 to 5 Â 10211 M TCDD with or without 0.5%
DMSO or 5 Â 1025 to 1029 M CB7993113. Plates were incubated at
33°C for 24 hours, and EGFP fluorescence was analyzed using
a Synergy 2 Multi-Mode Plate Reader (BioTek Inc., Winooski, VT).
For each plate, specific fluorescence was calculated by subtracting the
background fluorescence in untreated wells from the fluorescence in
vehicle- or compound-treated wells. Percent induction was calculated
by dividing by the specific fluorescence in the CB7993113-plus-TCDD
wells by the specific fluorescence in cultures treated with vehicle-plus-
TCDD for each given TCDD concentration and multiplied by 100. This
experiment was repeated six times.
Molecular Modeling/Predicting AHR Ligands
Commercial compound libraries were used for shape- and electrostatics-
based comparisons. Databases of 445,418 compounds from ChemBridge
Corporation (San Diego, CA) and 731,288 compounds from Enamine were
used for this scaffold-hopping approach.
The structural file containing a representative flavonoid substructure,
4-oxo-2-phenylchroman-3-yl methylcarbamate, was expanded into
a three-dimensional conformer database using OMEGA v.2.2.1 (Open-
Eye Scientific Software, Santa Fe, NM) (Hawkins and Nicholls, 2012),
allowing an energy window of 8 kcal/mol above ground state, and an
msd cutoff of 0.8 Å per the method described in Hawkins et al. (2007).
The two lowest energy conformers were selected and used for subse-
quent studies.
In a similar way, three-dimensional conformer libraries of the
Enamine and ChemBridge collections were generated. No limitation
in terms of maximum number of conformers was set. To speed this
computation, fragment libraries of each were pregenerated using the
program makefragmentlib. The sample flavonoid conformers were
Efficacy and Potency Analyses
To determine 50% inhibitory concentration (IC50) values, 5 Â 104
compared against the Enamine and ChemBridge conformer databases H1G1.1c3 cells were added to each well of a 96-well plate and cultured
using Rapid Overlay of Chemical Structures (ROCS; OpenEye). The as above. For each experiment, a TCDD standard curve (to normalize
highest scoring overlaps from ROCS were then subjected to electro- fluorescence readings between experiments) was prepared by apply-
static overlap comparison using EON (OpenEye). For hit-list ranking, ing TCDD (10210 to 1026 M) or vehicle (DMSO, 0.5%), with each
the electrostatic Tanimoto combo (ET) score was used. This is the sum concentration applied to six wells. H1G1.1c3 cells were treated with
of the shape Tanimoto and the Poisson-Boltzmann electrostatic Tanimoto.
The EON hit lists thus generated were merged into two separate 1025 M) immediately following addition of vehicle (to assay potential
structure files (Enamine and ChemBridge) using Pipeline Pilot (Accelrys, agonist activity), 1027 M 7,12-dimethylbenz[a]anthracene or 1027
San Diego, CA) and each were sorted on the basis of the electrostatic b-NF (to assay antagonist activity) (six wells/condition). Plates were
Tanimoto combo score (ET_combo). An order list was thus generated incubated at 33°C for 24 hours and EGFP fluorescence was analyzed
consisting of the top 98 hits from ChemBridge and top 99 hits from using a Synergy 2 plate reader. Percent AHR induction was calculated
Enamine. A summary of the computed properties of this library is as described above. Following the final fluorescence reading, the MTT
presented in Supplemental Table S1. cell viability assay was used to determine toxicity (Sigma-Aldrich).
vehicle (DMSO) or titered doses of CB7993113 or CH223191 (1029 to
M

596
Parks et al.
Briefly, after a 48-hour incubation, 10 ml MTT (5 mg/ml) was added determined by reactions containing an empty vector [unprogrammed
to each well and incubated at 37°C for 4 hours. Formazan crystals were
solubilized by the addition of 100 ml/well DMSO and incubation at 37°C
for 2 hours. Absorbance at 570 nm was quantified using a Synergy 2 plate
reader. Toxicity was calculated by subtracting the average absorbance in
the untreated wells from the average absorbance in the experimental
wells.
lysate].
Human AHR–Driven Reporter Assay
BP1 cells (2 Â 104/well) were plated in 24-well plates and allowed to
adhere overnight. Cells were cotransfected with 0.1 mg of the pGudluc
reporter plasmid, 0.1 mg CMV green, and 0.5 mg of either pcDNA or an
Ahrr expression plasmid using Lipofectamine 2000 (Life Technologies),
as we previously described (Yang et al., 2008). Cultures were incubated
for 3 hours and then dosed with vehicle (DMSO, 0.1%), CH223191
(0.01–1 mM), or CB7993113 (1–20 mM). After 1 hour, the transfection
medium was replaced, the cultures were redosed, and then incubated
for 24 hours. Cells were harvested in Glo Lysis Buffer (Promega) and
luciferase activity was determined with the Bright-Glo Luciferase
System per the manufacturer’s instructions (Promega). Luminescence
and fluorescence values were determined using a Synergy 2 plate
reader. To calculate “fold change from naïve” the luminescence value
was divided by the fluorescence value for each sample.
Homology Modeling/Theoretical AHR Binding
The X-ray structure of HIF-2a PAS-B domain cocrystallized with
N-(3-chloro-5-fluorophenyl)-4-nitro-2,1,3-benzoxadiazol-5-amine was cho-
sen as the template for model-building (Scheuermann et al., 2009). The
structure was downloaded from the Protein Data Bank (PDB) (Berman
et al., 2002). The PDB identifier (PDB ID) of the structure is 4GHI.
MODELER (Fiser and Sali, 2003) was used to model the human AHR
PAS-B domain. Only nonidentical residues and regions around the gaps
were optimized. To assure that the binding site would not collapse in
the process of model building, CB7993113 was aligned inside the homology
model, on the basis of the position of the double ring of the ligand in the
template and considering the polarity/hydrophobicity of the environ-
ment around it. The ligand-protein complex was then minimized using
the CHARMM potential to avoid possible steric clashes (Brooks et al.,
1983). The computational solvent mapping algorithm FTMap was used
to predict the most probable CB7993113-AHR PAS-B domain–binding
position. This method places small molecular probes of various sizes
and shapes on a dense grid around the protein, finds favorable positions
using empirical energy functions, clusters the conformations, and ranks
the clusters on the basis of the average energy. All ligands and
crystallographic water molecules were removed prior to mapping, and
the probes were initially distributed over the entire protein surface
(including all cavities) without any assumptions about the binding site.
The regions that bind multiple low energy probe clusters [consensus
cluster (CC) sites], identified the most important ligand-binding sites.
The hot spots were ranked in terms of the number of overlapping probe
clusters contained. The consensus cluster with the highest number of
probe clusters was ranked first as CC1 and nearby consensus clusters
within 7 Å were also joined with CC1 to form the predicted ligand-
binding site.
Human Peroxisome Proliferator–Activated Receptor
Gamma– and Cytomegalovirus-Driven Reporter Assays
Cos-7 cells were transiently transfected with vectors containing human
peroxisome proliferator–activated receptor gamma (PPARg, PPARG1)
(kindly provided by V. K. Chatterjee, University of Cambridge,
Cambridge, UK) (Gurnell et al., 2000) and human RXRA (plasmid 8882;
Addgene, Cambridge, MA) with PPRE Â3-TK-luc (plasmid 1015;
Addgene) and cytomegalovirus (CMV)–eGFP-reporter constructs using
Lipofectamine 2000 (Life Technologies). Transfected cultures were
incubated for 3 hours. The medium was replaced with antibiotic-free
DMEM with 5% fetal bovine serum and the cultures were incubated
overnight. Cultures received no treatment (naïve) or were treated with
vehicle (Vh; DMSO, 0.1%) or with rosiglitazone (1 mM) and Vh,
CH223191 (10 mM), or CB7993113 (10 mM), and incubated for 24 hours.
Cells were lysed in Glo Lysis Buffer (Promega). Lysates were transferred
to a 96-well plate to which Bright Glo Reagent (Promega) was added.
Luminescence and fluorescence were determined using a Synergy 2
plate reader. PPARg-specific luminescence was normalized to the GFP
fluorescence in the same well. For the CMV-driven reporter assay,
human mammary tumor cells (BP1) were transfected with CMV-driven–
eGFP-reporter plasmid (.80% transfection efficiency) and treated with
CH223191 or CB7993113 as described above. GFP fluorescence was
assayed 24 hours later. Constitutive CMV-reporter activity was
calculated relative to levels seen in naïve cultures.
Only nonidentical residues and regions around the gaps were
optimized and the predicted ligand-protein complex was then mini-
mized using the CHARMM potential (Brooks et al., 1983) to avoid
potential steric clashes. The FTMap algorithm (http://ftmap.bu.edu)
(Brenke et al., 2009) was used to predict the AHR PAS-B domain-
binding hot spots.
[A model of the human AHR PAS-B domain in complex with
CB7993113 is provided in Supplemental Data file CID1.pdb. A model of
the human AHR PAS-B domain in complex with CB7993113, together
with computational solvent mapping (http://ftmap.bu.edu) of the PAS-B
domain, is provided in Supplemental Data file CID.pdb.]
Mouse AHR Immunoblotting
H1G1.1c3 cells were plated at 2 Â 106 cells in T75 flasks and allowed
to adhere overnight. Cells were treated with CB7993113 (10 mM),
vehicle (DMSO, 0.1%), or left untreated for 1 hour, followed by DMBA
(1027 M) treatment for 30 minutes. Nuclear and cytoplasmic cell
extracts were prepared using the Nuclear Extract Kit (Active Motif,
Carlsbad, CA) per the manufacturer’s instructions. Protein concentra-
tion was quantified using Protein Assay Reagent (Bio-Rad, Hercules,
CA). Protein (30 mg) was resolved on 12% SDS-polyacrylamide gels and
transferred to 0.2-mm nitrocellulose membranes (Bio-Rad). Membranes
were probed with the following primary antibodies: mouse anti-AHR
(Pierce, Rockford, IL), rabbit anti-Lamin-A/C (Cell Signaling
Technologies, Danvers, MA), and mouse anti-a-tubulin (EMD
Millipore, Billerica, MA). Immunoreactive bands were detected
using horseradish peroxidase (HRP)–conjugated secondary anti-
bodies [goat anti-rabbit (Bio-Rad), goat anti-mouse (Pierce)], and
ECL substrate.
In Vitro Protein Synthesis and Competitive-Binding Assay
Murine and human AHR proteins were synthesized from AHR
expression constructs (pSportMAHR or pSportAHR2, respectively, gifts
of Dr. C. Bradfield, University of Wisconsin, Madison, WI) (Burbach
et al., 1992; Dolwick et al., 1993) using a TnT Quick Coupled reticulocyte
lysate system (Promega). The ability of CB7993113 or CH223191 to
compete with [3H]TCDD (35 Ci/mmol; Chemsyn Science Laboratories,
Lenexa, KS) for binding to human or mouse AHR was measured by
velocity sedimentation on sucrose gradients in a vertical tube rotor as
described earlier (Karchner et al., 2006). Briefly, single TnT reactions
(50 ml) were diluted 1:1 with MEEDMG buffer (25 mM MOPS, 1 mM
EDTA, 5 mM EGTA, 0.02% NaN3, 1 mM DTT, 20 mM molybdate,
10% (v:v) glycerol, pH 7.5) and incubated overnight at 4°C with [3H]TCDD
(2 nM) 6 DMSO or competitor (10 mM final concentration, dissolved in
DMSO). Incubations were applied to 10- to 30%-sucrose gradients and
Human CYP1B1 mRNA Expression
BP1 cells (3 Â 106) were plated in T225 flasks and allowed to adhere
analyzed as described (Karchner et al., 2006). Nonspecific binding was overnight. Cultures were dosed with CB7950998 (10 mM) or vehicle

Identification of an AHR Antagonist
597
(DMSO, 0.1%) and incubated for 24 hours. mRNA was extracted
normalization. The Cq value from untreated animals was used as the
using RNeasy Plus Mini Kit (Qiagen, Valencia, CA). cDNA was prepared reference point.
from total RNA using the GoScript Reverse Transcription System
For flow cytometry analyses, RBC-depleted bone marrow samples
(Promega), with a 1:1 mixture of random and oligo (dT)15 primers. All were treated with Fc-blocking solution (BD Biosciences, San Jose, CA)
real-time qPCR reactions were performed using the GoTaq RT-qPCR and suspended in phosphate-buffered saline (Life Technologies) supple-
Master Mix System (Promega). Validated primers were purchased
mented with 2% fetal bovine serum (Gemini, West Sacramento, CA) for
from Qiagen: human CYP1B1–QT00209496 and human RRN18S– 15 minutes at 4°C. Samples were subsequently surface-stained with
QT00199367. Real-time qPCR reactions were performed using a 7900HT
Fast Real-Time PCR instrument (Applied Biosystems, Carlsbad, CA):
Hot-Start activation at 95°C for 2 minutes, 40 cycles of denaturation
(95°C for 15 seconds), and annealing/extension (55°C for 60 seconds).
Relative gene expression was determined using the Pfaffl method
(Pfaffl, 2001) with the threshold value for RRN18S for normalization.
The Cq value from untreated cultures was used as the reference point.
a cocktail of IgM-, CD24-, CD43-, and B220-specific antibodies, a cocktail
of Gr-1-, CD3-, NK1.1-, CD11b-, and B220-specific antibodies, or their
appropriate isotype-matched antibodies for 30 minutes at 4°C in the
above staining buffer. All flow cytometry antibodies were obtained from
BD Biosciences or eBiosciences (San Diego, CA). Samples were analyzed
on a BD LSR II instrument. Post-acquisition analysis was performed
using FlowJo analysis software (TreeStar).
Invasion and Migration Assays
Statistical Analysis
For Matrigel assays, BP1 cells (3 Â 105 cells/well) were plated in six-
well plates and allowed to adhere overnight. Cells were pretreated with
vehicle (DMSO), CH223191 (10 mM), CB7993113 (5 mM), or left untreated
for 24 hours. Cells were harvested and prepared as a single cell suspension
for addition to the Matrigel branching assay. Matrigel basement
membrane matrix (BD Biosciences, Bedford, MA) was diluted to 6.3 mg/ml.
Matrigel solution (200 ml) was added to a 24-well plate and solidified at
37°C for 30–45 minutes to form a base layer. Single-cell suspensions
containing 2 Â 104 pretreated BP1 cells in 10 ml serum-free media were
mixed with 190 ml of Matrigel containing 0.1% vehicle or 5–10 mM
CB7993113. Vehicle or CB7993113 was also added to the Matrigel top
layer and the layer allowed to solidify at 37°C for 30–45 minutes.
Complete medium (0.5 ml) containing vehicle (DMSO, 0.1%) or
CB7993113 (5 or 10 mM) was added on top of solidified Matrigel and
was replaced with fresh dosing solution every other day. Colony
morphology was captured using a Zeiss Axiovert 200 M microscope.
Images were captured using a Nikon Coolpix4300 digital camera.
For “scratch-wound” assays (Li et al., 2013), Hs578T or SUM149
cells (500,000 cells/well) were plated in six-well plates and allowed to
grow until 100% confluent. Cultures were scratched with a pipette tip,
left untreated, or treated with vehicle, 10 mM CH223191, or 10 mM
CB7993113, and allowed to regrow for 48 hours. Cultures were
photographed using a Nikon CoolPix 4300 attached to a Zeiss
Telaval31 inverted microscope at 0, 24, and 48 hours.
Statistical analyses were performed with GraphPad Prism (La Jolla,
CA). Data are presented as means 6 S.E. where applicable. One-tailed
Student’s t tests and one-way analysis of variance with Dunnett,
Tukey-Kramer, or Newman-Keuls post-hoc were used to determine
significance.
Results
Virtual Screening and Generation of Focused Li-
braries. Development of modulators that can alter AHR
function is of growing importance given the established role
that the AHR plays in mediating toxic effects of environmental
chemicals and the emerging role of the AHR in inflammatory
diseases, autoimmunity, hematopoiesis, and cancer. To this end,
known AHR-ligand structures and shape-based modeling were
used to predict new AHR ligands and to generate targeted libraries
of small molecules with substructures similar to those of known
natural and anthropogenic AHR ligands. Noting the prevalence
of flavonoid-like molecules in known AHR ligands, a generic
substructure, 4-oxo-2-phenylchroman-3-yl methylcarbamate
(Supplemental Fig. S2), was selected as a basis to search for
shape- and electrostatic-based similarities in commercial sets
of compounds from two vendors, Enamine and ChemBridge, Inc.
Databases consisting of 445,418 and 731,288 small molecules
from ChemBridge and Enamine, respectively, were scrutinized
for three-dimensional conformers similar to the flavonoid sub-
In Vivo Studies
Male, 6–8 week old C57BL/6J mice were purchased from The
Jackson Laboratory (Bar Harbor, ME). All animal studies were
reviewed and approved by the Institutional Animal Care and Use structure using the OMEGA algorithm. The three-dimensional
Committee at Boston University. For pharmacokinetic studies,
13-week-old mice (four/group) were treated with 50 mg/kg CB7993113
by oral gavage or intraperitoneal injection. Serum was collected 4, 8,
and 16 hours after treatment and from untreated mice. Serum samples
were analyzed by liquid chromatography–tandem mass spectrometry
for CB7993113 at Apredica/Cyprotex (Watertown, MA). For DMBA-
induced bone marrow toxicity studies, mice (six per group) were
injected intraperitoneally with vehicle (100 ml vegetable oil), CH223191
(50 mg/kg), or CB7993113 (50 mg/kg) 24 hours and 1 hour before dosing
with 200 ml sesame oil or 50 mg/kg DMBA by oral gavage. Mice were
euthanized 48 hours after DMBA treatment. Liver was snap frozen
for mRNA analysis. Bone marrow was flushed from femurs and
tibiae and red blood cells (RBCs) were removed using ACK Lysing Buffer
(BioWhittaker, Lonza, Allendale, NJ).
conformer set generated from this analysis was then compared
with generic flavonoid substructures by ROCS shape comparison.
Subsequently, compounds that ranked highest in structural
overlap from the ROCS comparison were subjected to an
electrostatic analysis using EON electrostatic comparison
and were reranked according to their electrostatic Tanimoto (ET)
score. Focused libraries of the top 197 small-molecule hits from the
Enamine and ChemBridge databases were generated (Supple-
mental Table S1) and tested for AHR modulating activity in vitro.
In Vitro Identification of Novel AHR Antagonists. A mu-
rine reporter–based bioassay (Nagy et al., 2002) was adapted
for semiautomated, high-throughput screening of the aforemen-
tioned focused library for AHR agonists and antagonists. This
assay utilizes the murine hepatoma cell line H1G1.1c3 that
is stably transfected with an AHR-responsive EGFP reporter
to measure AHR transcriptional activity. To assess AHR
agonist activity, each experimental compound at 1 mM or 5 mM
concentration was added alone to H1G1.1c3 cells. For AHR-
antagonist screening, each experimental compound was added
For gene expression analyses, mRNA was extracted from liver tissues
using the RNeasy Plus Mini Kit (Qiagen) and concentrated, if required,
using the RNeasy MinElute Cleanup Kit (Qiagen). cDNA was prepared
and real-time qPCR was carried out, as described above. Primer sequences
for murine Cyp1a1 and murine Gapdh were previously described (Xu and
Miller, 2004) and were synthesized by Integrated DNA Technologies, Inc.
(Coralville, IA). Relative gene expression was determined using the Pfaffl
method (Pfaffl, 2001), with the threshold value for Gapdh used for immediately prior to addition of 1027 M b-NF, a well-described

598
Parks et al.
AHR agonist. EGFP expression was assayed 24, 48, and 72 hours 2003; Dvorak et al., 2008). Partial AHR agonists may appear to
later. inhibit the AHR in bioassays since they compete with high-
Of the 197 compounds screened, 31 compounds exhibited efficacy ligands for AHR binding but cannot fully activate the
agonist activity, as defined by a $25% increase in reporter activity receptor. None of the four putative AHR antagonists discovered
at 1 and 5 mM and at all three time points when added alone to the in this screen induced AHR-driven reporter activity in the high-
reporter assay (data not shown). Four compounds demonstrated throughput assay, even at the highest concentration tested
significant antagonist activity as defined by a $25% reduction (10 mM) (data not shown and Figs. 3 and 4), indicating that
in b-NF responses at 1 and 5 mM and at all three time points they are not partial agonists.
(Fig. 1, arrows). The structures of the putative antagonist
The potency and efficacy of the four discovered AHR
compounds and, for comparison, those of other previously described antagonists were compared with CH223191 using the high-
AHR antagonists, CH223191 (Kim et al., 2006), StemRegenin throughput H1G1.1c3 assay (Fig. 3). At maximal concentrations,
(SR1) (Boitano et al., 2010), GNF351 (Smith et al., 2011), and 6, the four discovered AHR antagonists inhibited between 40 and
29,49trimethoxyflavone (Zhao et al., 2010) are provided in Fig. 2. 70% of the b-NF–induced EGFP signal, with CB7993113
Note the relative structural dissimilarity of CB7993113 when exhibiting the greatest efficacy (Table 1). IC50 values ranged
compared with most of the other “pure” AHR antagonists.
from 0.023 to 1.85 mM (Table 1). None of the compounds were
Several compounds (e.g., a-naphthoflavone, galangin) exhibit toxic even at the highest dose tested, as defined by a $10%
partial AHR agonism (Santostefano et al., 1993; Zhang et al., increase in dead cells after 48 hours of exposure. All of the
Fig. 1. An in vitro AHR bioassay identifies novel AHR antagonists. H1G1.1c3 cells were left untreated or treated in triplicate wells with vehicle (0.1%
DMSO), 1 or 5 mM concentrations of 197 compounds from the ChemBridge and Enamine focused libraries immediately prior to addition of 1027 M b-NF.
Cells were cultured at 33°C for 72 hours. EGFP fluorescence was analyzed at 24, 48, and 72 hours. Data are presented as the average of three wells, and
are calculated as the percent induction, with 100% set at levels seen with 1027 M b-NF alone (dashed line). Data points representing AHR antagonist
hits, defined as those compounds with an average induction of #75% of b-NF treated cells (green dotted line) for all measured endpoints, are indicated
with arrows. Data are representative of a single set of high-throughput screen experiments.

Identification of an AHR Antagonist
599
Fig. 2. Structures of antagonist hit compounds. Structures of antagonist hits are presented along with their chemical identification numbers. The
structure of the previously described AHR antagonists CH223191, SR1, GNF351, and trimethoxyflavone are presented for comparison.
compounds were within Lipinski’s guidelines for drug-likeness in H1G1.1C3 cells (IC50s 2.1 mM versus 0.7 mM, respectively;
(Lipinski et al., 2001). Interestingly, the low IC50 (i.e., high Fig. 4A). However, CB7993113 was more effective than CH223191
potency) exhibited by T0515-7358 (0.02 mM) was not reflected at reducing DMBA-induced reporter activity under these con-
in a significantly higher efficacy (68% inhibition) than the other ditions (46 versus 28% respectively). Again, neither CB7993113
discovered compounds. This is not an unusual finding since the nor CH223191 acted as a partial agonist at doses ranging from
potency, in part a function of ligand-receptor affinity, does not 1 nM to 10 mM (Fig. 4B) and neither compound exhibited
necessarily correlate directly with efficacy, a function of the cytotoxicity as determined in an MTT assay (Fig. 4C). Further-
biologic response to the ligand. This result further highlights more, neither compound affected reporter activity mediated by an
differences between AHR antagonists.
unrelated nuclear receptor, PPARg, or reporter activity driven by
Compounds CB7993113 and T0515 exhibited the greatest a constitutive CMV promoter (Supplemental Fig. 3). Collectively,
efficacy (maximum percent inhibition). As anticipated on the these data indicate that at least one predicted new compound,
basis of calculated lipophilicity (c Log P) values, CB7993113 was CB7993113, specifically antagonizes murine AHR transcrip-
qualitatively observed to be more soluble in phosphate-buffered tional activity, does not exhibit partial AHR agonist activity,
saline than the other compounds and therefore was prioritized for and is not cytotoxic, even at relatively high concentrations.
further study.
To begin to assess whether CB7993113 is probably a com-
In a separate series of experiments, head-to-head comparisons petitive or allosteric inhibitor, titered concentrations of b-NF
were made between CH223191 and lead compound CB7993113, (1028 to 1025 M) or TCDD (10213 to 1029 M) and CB7993113
using a prototypic, toxic AHR ligand, DMBA (Teague et al., 2010), (1027 to 1025 M) were added to H1G1.1c3 cells and the percent
which also was used in in vivo bone marrow toxicity assays (see induction of TCDD-induced, AHR-driven reporter activity
below). CB7993113 was slightly less potent than CH223191 as assayed 24 hours later. If CB7993113 is an allosteric inhibitor,
measured by reduction of DMBA-induced AHR reporter activity then little or no change in the potency (EC50) of an AHR agonist
TABLE 1
AHR antagonist hit compounds
Vendor catalog numbers, chemical names, molecular weights, and Log P (partition coefficient) values are presented. The c Log P values were
determined using Pipeline Pilot (Accelrys). Maximum percent inhibition values were determined from data presented in Fig. 3 using 50 mM
concentrations of the respective compounds and assaying on day 3. IC50 values were determined from the dose-response curves generated in Fig. 3
using GraphPad Prism.
Chemical ID
Chemical name
MW
Log P Max % inhibition IC50
mM
CB7993113 2-{[2-(5-Bromo-2-furyl)-4-oxo-4H-chromen-3-yl]oxy}acetamide
T0515-7358 2-Phenyl-5,6,7,8-tetrahydro-4H-chromene-4-thione
364.1 1.03
242.3 3.88
339.4 3.91
371.4 4.60
70.3
67.7
61.6
40.3
0.33
0.02
1.85
0.04
T5448133
T6047668
2-(Benzyloxy)-N-(5-ethyl-1,3,4-thiadiazol-2-yl)benzamide
2-((4-Fluorobenzyl)oxy)-N-(5-isopropyl-1,3,4-thiadiazol-2-yl)
benzamide
CH223191
(E)-1-Methyl-N-(2-methyl-4-(o-tolyldiazenyl)phenyl)-1H-pyrazole-5-
carboxamide
333.4 4.85
69.6
0.24

600
Parks et al.
Fig. 3. Hit compound characterization led to selection of lead compound, CB7993113. H1G1.1c3 cells were treated with vehicle (0.1% DMSO) or 1029 to
1025 M AHR antagonists immediately prior to stimulation with 1027 M b-NF. Fluorescence (AHR-dependent reporter activity) was assayed 24, 48, and
72 hours later as in Fig. 1. Dose response curves were generated utilizing a three-parameter dose-response curve model in GraphPad Prism with
a standard Hill slope of –1. Data are presented as the average of three wells and are calculated as the percent induction with 100% set at levels seen with
1027 M b-NF alone.
should be seen after addition of titered concentrations of the AHR-binding characteristics. Therefore, the ability of CB7993113
inhibitor. In contrast, if CB7993113 is a competitive inhibitor, to bind human AHR also was assessed using in vitro translated
then agonist EC50 should increase as increasing concentrations human AHR. The ability of CB7993113 to inhibit the binding of
of inhibitor are added. Indeed, the b-NF and TCDD titration [3H]TCDD to human AHR was similar to that seen with murine
curves shifted to the right (increasing EC50) as the concentra- AHR, whereas CH223191 was slightly more effective as an
tion of CB7993113 was increased (Supplemental Fig. 4). These inhibitor of [3H]TCDD binding to human AHR compared
data are consistent with competitive inhibition.
with mouse AHR (Fig. 5B).
A competitive AHR antagonist would be expected to bind
As would be expected from a competitive antagonist, 10 mM
directly to the AHR and, at a minimum, block agonist-induced CB7993113 completely prevented DMBA-induced AHR nu-
nuclear translocation. Therefore, the capacity of CB7993113 to clear translocation (Fig. 5C). Partial inhibition of AHR nuclear
directly bind to in vitro translated murine AHR protein was translocation was seen in two experiments with as little as
determined in a [3H]TCDD competitive binding assay. CH223191 1 mM CB7993113 (data not shown). Consistent with our findings
was included as a positive control. As predicted from the cal- with the AHR-driven reporter assay, 10 mM CB7993113 alone
culated IC50s (Figs. 3 and 4, Table 1), 10 mM CB7993113 and did not induce nuclear translocation of the AHR protein and thus
CH223191 were comparable in their ability to block [3H]TCDD- was not acting as a partial AHR agonist (Fig. 5C). These data
AHR binding (Fig. 5A).
collectively indicate that CB7993113 is probably a competitive
Several AHR ligands, including at least one AHR competitive AHR antagonist and that it blocks AHR nuclear translocation
antagonist (Boitano et al., 2010), exhibit species-specific and, thereby, transcriptional activity.
Fig. 4. CB7993113 blocks DMBA-induced AHR-dependent reporter activity. (A) H1G1.1c3 cells were treated with vehicle (0.1% DMSO) or 1029 to 1025 M
CB7993113 or CH223191 immediately prior to stimulation with 1027 M DMBA and culture at 33°C. EGFP fluorescence (AHR reporter activity) was analyzed
24 hours later. (B) AHR agonist activity was measured by treating H1G1.1c3 cells with vehicle or 1029 to 1025 M CB7993113 or CH223191. Cells were cultured
at 33°C and EGFP fluorescence was analyzed after 24 hours. Dose-response curves were generated utilizing a three-parameter dose-response curve model in
GraphPad Prism with a standard Hill slope of –1 (A and B). (C) Cellular toxicity after the above treatment was measured by the MTT reduction assay 48 hours
after addition of compounds. Data are presented as means 6 S.E. of three independent experiments. (One-way analysis of variance with Dunnett’s post-test,
***P , 0.001 for CB7993113 compared with vehicle, ++P , 0.01 or +++P , 0.001 for CH223191 compared with vehicle.)

Identification of an AHR Antagonist
601
Homology Modeling of CB7993113-AHR Binding. To Institute) (Trott and Olson, 2010). The 10 lowest energy binding
begin to visualize the orientation of this putative competitive poses were retained for each ligand. Second, for the selection of
inhibitor in the binding pocket of the human AHR PAS-B the most probable pose, the AutoDock Vina energy score and the
domain, the X-ray structure of the HIF-2a PAS-B domain, atom densities calculated from the mapping results were
cocrystallized with the small antagonist N-(3-chloro-5- included in the calculus. Probe density was defined at each
fluorophenyl)-4-nitro-2,1,3-benzoxadiazol-5-amine (4GH1) point binding site as the total number of probe atoms within
(Scheuermann et al., 2013) was selected as the template for a 1.25-Å radius. We considered each retained ligand pose and
model building. The recent crystal structure of the mouse PAS1 summed the atomic densities for all heavy atoms, resulting in
(PAS-A) domain (PDB ID 4M4X) (Wu et al., 2013) indicates a measure of overlap between the pose and the probe density.
that the human AHR PAS-B domain begins after residue Q273, The poses with low energy scores were ranked on the basis of
and hence AHR residues 287–390 were aligned with HIF-2a this overlap measure, and the pose with the best overlap was
residues 244–348 (Supplemental Fig. 5). The two sequences selected (Kozakov et al., 2011).
have 27% identity and 50% similarity in this range. Although
Figure 6A shows the template, the PAS-B domain of HIF-2a,
this level of sequence identity is relatively low for homology with the bound antagonist. The backbone of the homology
modeling, the identical residues distribute almost equidistantly model of the human AHR is very similar (Fig. 6B). There is no
along the domain with only a two-percent (three amino acids) template for the AHR loop 371–376 and thus it was constructed
gap. This type of distribution makes it very probable that the by MODELER. (Note that this loop is far from the binding site
backbones of the two proteins are very similar. Indeed, structures and its exact structure is not important for this model.)
are available for a number of PAS domains with similar level of Figure 6B also shows the probe clusters obtained by the
sequence identity and good three-dimensional overlap.
mapping program FTMap and indicating the energetically
The MODELER mapping results were used in two different most important regions of the binding site. We note that the
ways to approximate CB7993113 docking. First, a box with 4-Å sequence differences between HIF-2a and AHR results in
padding was created around the predicted binding site. The a larger binding cavity and greater hydrophobicity of the AHR.
docking was carried out restricting consideration to this box and Consequently, the site can accommodate CB7993113 in a num-
using the standard settings of AutoDock Vina 1.1.0 (Scripps ber of conformations and the 10 lowest energy docked poses of
Fig. 5. CB7993113 directly binds murine AHR protein and blocks AHR nuclear translocation. In vitro-expressed murine AHR (mAHR) (A) or human
AHR (hAHR) (B) protein was incubated with 2 nM [3H]TCDD in the presence of vehicle (DMSO), 10 mM CH223191, or 10 mM CB7993113 for 12 hours.
Quantification of [3H]TCDD-AHR binding was analyzed by velocity sedimentation on sucrose gradients. Data are representative of two independent
experiments. (C) H1G1.1c3 cells were treated with vehicle (0.1% DMSO) or 10 mM CB7993113 for 1 hour, followed by treatment with 1027 M DMBA for
30 minutes. Cytoplasmic and nuclear extracts were isolated and analyzed for AHR, a-tubulin, and lamin A/C content by immunoblotting. Data are
representative of four independent experiments.

602
Parks et al.
CB7993113 show substantial variation. Figure 6C shows one of the tended to decrease endogenous CYP1B1 levels in this short-
lowest energy structures (pose 2 from AutoDock Vina) that also term experiment, although statistical significance was only
overlaps well with the binding hot spots. The hydrophobic part of reached at a concentration of 20 mM (Fig. 7B).
the chromen-3 moiety of the ligand is surrounded by the side
Our laboratory (Trombino et al., 2000; Schlezinger et al.,
chains of F295, M340, and A367, whereas H291 and Q383 donate 2006) and others (Brooks and Eltom, 2011; Opitz et al., 2011)
hydrogen bonds to the more polar regions of CB7993113. For have postulated that AHR hyperexpression and activity
better visibility, Fig. 6D shows only a few of the side chains that are facilitate malignant transformation. It would then be predicted
within 5 Å of the ligand. Interestingly, the differences in the amino that AHR antagonists would reverse at least some component
acids shown in the AHR yield less bulky side chains, and thus of the malignant phenotype. The growth of immortalized cells
increase the size of the ligand-binding cavity relative to HIF-2a.
in irregular, branching colonies in Matrigel is widely seen as
CB7993113-Mediated Inhibition of AHR-Dependent a marker for invasiveness (Hughes et al., 2008). Indeed, we have
Biologic Responses. AHR expression and activation in the demonstrated that inhibition of AHR activity with AHRR or
absence of environmental ligands is associated with several AHR knockdown with AHR-specific small-interfering RNA
human cancers, including breast cancer (Brooks and Eltom, blocks the formation of branching colonies in human breast
2011; Opitz et al., 2011; Schlezinger et al., 2006). It is presumed cancer cell lines (S. Narasimhan et al., manuscript in prepara-
that this basal activity is mediated by endogenous AHR ligands. tion). Therefore, the ability of CB7993113 to alter the morphol-
Therefore, it was predicted that CB7993113, like AHR repressor ogy of BP1 cell colonies in Matrigel was assessed. Untreated or
protein (Yang et al., 2008), would lower baseline levels of AHR- vehicle-treated BP1 cell colonies exhibited the branched,
driven reporter activity and endogenous target gene expression irregular morphology typical of invasive cells as early as day 3
in human breast cancer cells. Malignant triple-negative BP1 of culture (Fig. 7C). However, treatment with either 10 mM
cells were used for these studies since we had previously CH223191 or 5 mM CB7993113 clearly reduced the size of the
established relatively high baseline levels of AHR (reporter) colonies and their degree of branching. Higher magnification
activity and AHR-dependent enforcement of high CYP1B1 images revealed that CB7993113 tended to decrease forma-
levels in this invasive cell line (Yang et al., 2008). BP1 cells tion of invasive cell processes (Fig. 7D).
were transiently transfected with the pGudLuc AHR-driven
It is possible that the morphologic changes seen in the
reporter plasmid. As previously reported (Yang et al., 2008), presence of CB7993113 are, in part, a function of altered cell
cotransfection with a plasmid carrying the AHR repressor gene growth or toxicity. However, [3H]thymidine incorporation ex-
(Ahrr) significantly reduced baseline pGudLuc reporter activity periments in two-dimensional BP1 cultures failed to demon-
(Fig. 7A). Similarly, treatment of pGudLuc-transfected cells strate a significant change in cell growth or viability (more than
with CB7993113 significantly decreased baseline AHR activity five experiments, data not shown). Furthermore, the number of
at concentrations as low as 5 mM (Fig. 7A). CB7993113 also viable cells recovered from 7- to 10-day-old three-dimensional
Fig. 6. Homology modeling of CB7993113
binding to human AHR. Homology mod-
eling on the basis of the crystal structure
of ligand-bound HIF-2a was employed as
described in the text. (A) Structure of ligand-
bound HIF-2a. (B) A proposed model of
CB7993113 docking to human AHR, in-
cluding the probe clusters, obtained by the
mapping program FTMap, and indicating
the energetically most important regions of
the binding site. (C) Hydrophobic part of the
chromen-3 moiety of the ligand that is
surrounded by the side chains of F295,
M340, and A367. (D) A few side chains
that are within 5 Å of the ligand. Residue
numbers in parentheses represent the
corresponding residues of HIF-2a (see
Supplemental Fig. 5).

Identification of an AHR Antagonist
603
Fig. 7. CB7993113 and CH223191 reduce the invasive phenotype of human breast cancer cell colonies in 3D Matrigel assays. (A) BP1 cells were
cotransfected with AHRE-driven firefly luciferase reporter (pGudluc) and control CMV green (GFP) vectors and incubated for 3 hours. In some groups,
Ahrr plasmid was cotransfected as a positive control. Cultures were left untreated or were treated with vehicle (0.1% DMSO) or 1–20 mM CB7993113.
Cells were harvested 24 hours later and luciferase activity assayed. Luciferase (AHR-reporter) activity was first normalized to the GFP signal to control
for transfection efficiency and then normalized to the luciferase signal obtained from untreated cells. Data are expressed as means 6 S.E. obtained from
three independent experiments. (***P , 0.001 compared with vehicle groups; one-way analysis of variance (ANOVA) with Dunnett’s post-test.) (B) BP1
cells were treated with vehicle (DMSO) or 1–20 mM CB7993113 for 24 hours. mRNA was extracted and analyzed by qPCR for CYP1B1 expression
normalized to 18sRNA expression. qPCR was performed in duplicate. Data are expressed as means 6 S.E. from three to five independent experiments.
(*P , 0.05, compared with vehicle groups, ANOVA with Dunnett’s post-test.) (C) BP1 cells were treated with vehicle (0.1% DMSO), 10 mM CH223191,
5 mM CB7993113, or left untreated for 24 hours. Cells were harvested, counted, and plated in Matrigel. Representative images from three independent
experiments were taken on days 3 and 5. (D) BP1 cells were treated as in (C) with vehicle or CB7993113 (10 mM). High magnification images were
captured on day 5 to illustrate cell morphology. CB7993113-treated cells extracted from Matrigels were ∼95% viable. No significant differences in the
numbers of cells in cultures after extraction from three-dimensional cultures were seen.
Matrigel cultures, as determined by Trypan blue exclusion and SUM149 (inflammatory breast cancer–derived) cells were
visual inspection or by propidium iodide exclusion and flow “wounded” in a scratch assay (Li et al., 2013), treated with
cytometry, was not affected by CB7993113 treatment (three vehicle (DMSO), 10 mM CH223191, or 10 mM CB7993113,
experiments, data not shown). Therefore, the changes in and photographed 24 and 48 hours later. Naïve or vehicle-
morphology observed with CB7993113 were not attributable to treated Hs578T and SUM149 cultures closed the scratch
decreased cell growth or increased toxicity.
wound completely by approximately 56 and 72 hours, re-
To assess whether CB7993113 inhibits another marker of spectively (not shown). Whereas vehicle had no effect on the
tumor invasiveness, i.e., migration across a “scratch wound,” rate of wound closure, treatment with either CH223191 or
and to extend studies to two other cell lines, confluent cultures CB7993113 consistently inhibited wound closure in both cell
of AHRhigh (Yang et al., 2008) Hs578T (triple-negative) and lines at 24 and 48 hours (Fig. 8).

604
Parks et al.
Fig. 8. CB7993113 blocks mammary tumor cell migration
in a scratch-wound assay. Confluent monolayers of Hs578T
(triple-negative breast cancer) and SUM149 (inflammatory
breast cancer) cells, scratched with a pipette tip, were treated
with vehicle, 10 mM CH223191, or 10 mM CB7993113, and
allowed to regrow for 48 hours. Cultures were photographed 24
and 48 hours after wounding. Images taken at 24 and 48 hours
are representative of results obtained in six experiments.
Results in both the Matrigel and scratch-wound assays are sufficient concentrations in situ to block this AHR-dependent
consistent with a role for the AHR in breast cancer cell inva- bone marrow toxicity, C57Bl/6 mice were treated with vehicle,
siveness and suggest the use of AHR antagonists to pharmaco- 50 mg/kg CB7993113, or, as a positive control, 50 mg/kg CH223191,
logically alter constitutive AHR activity in and potential by intraperitoneal injection 24 hours and 1 hour before oral
invasiveness of cancer cells.
gavage with 50 mg/kg DMBA. Mice were euthanized 48 hours
In Vivo Efficacy of the AHR Antagonist CB7993113. later, bone marrow cells were collected, and hematopoietic
Drugs to be used as therapeutics must exhibit several properties cells were phenotyped by flow cytometry. As expected from in
in vivo, including sufficient solubility, absorption, stability, and vitro studies in which CB7993113 failed to exhibit toxicity, it
accumulation in target organs at biologically relevant concen- also failed to affect the number of bone marrow cells recovered
trations. Although in silico computation predicted that CB7993113 48 hours after intraperitoneal injection (Fig. 9B). In contrast,
would satisfy these criteria, in vivo studies were required to a significant reduction in the total number of bone marrow
confirm this prediction. In the absence of an animal model system cells was observed after DMBA treatment. This acute bone
that faithfully and consistently recapitulates breast cancer cell marrow cell ablation was inhibited by treatment with either
invasion in humans, we chose to evaluate CB7993113 in vivo CB7993113 or CH223191 (Fig. 9B).
efficacy using surrogate endpoints of bioactivity, i.e., the ability
Phenotypes of bone marrow subpopulations were analyzed
to block acute AHR-mediated CYP1A1 induction in liver and to determine which hematopoietic cell subsets were affected
AHR-regulated bone marrow toxicity (Teague et al., 2010; N’jai and to determine if CB7993113 could protect all subsets
et al., 2011).
from DMBA-induced toxicity. No significant changes were
In pharmacokinetics experiments, 50 mg/kg CB7993113 was seen in the resident bone marrow T cell (CD31) or natural
readily absorbed in vivo following either intraperitoneal or oral killer cell (NK1.11) populations following treatment with
administration. Thus, 823 6 263 nM and 395 6 162 nM DMBA, CH223191, CB7993113, or combinations of DMBA
CB7993113 were detected in sera 1 hour after intraperitoneal with either antagonist (data not shown). In contrast, DMBA
injection or oral gavage, respectively. The compound exhibited treatment significantly reduced the number of pre/pro-B
a serum half-life of 4.0 hours (data not shown). Therefore, cells (IgM–/B2201/CD431/HSA–), pro-B cells (IgM–/B2201/
a dose of 50 mg/kg was used for further in vivo studies.
CD431/HSA1) (Hardy and Hayakawa, 2001), and neutro-
As expected from in vitro studies, intraperitoneal injection of phils (CD11bhi/GR-1hi) (Sukhumavasi et al., 2007) (Fig. 9,
50 mg/kg CB7993113 did not induce CYP1A1 mRNA expres- C–E). Notably, pretreatment of mice with either antagonist
sion in liver as assessed by real-time qPCR (Fig. 9A), indicating significantly inhibited DMBA-induced toxicity in all three
that neither this compound nor its metabolites are partial AHR cell populations. These data confirm the ability of a pro-
agonists in vivo. In contrast, an equal concentration of DMBA totypic AHR ligand to adversely affect bone marrow cells
induced a significant, 200- to 400-fold, increase in CYP1A1 at destined to contribute to the adaptive immune response,
this 48-hour time point. DMBA-induced CYP1A1 induction was i.e., pre/pro- and pro-B cells, and to the innate immune
inhibited by both CH223191 and CB7993113, although inhibition response, i.e., neutrophils. Importantly, CB7993113 pre-
of DMBA-induced CYP1A1 expression was less consistent with vented a significant DMBA-induced loss of these three bone
CH223191 than with CB7993113.
marrow cell subsets. These data demonstrate the achieve-
In previous studies, we and others demonstrated that DMBA ment of physiologically relevant doses of CB7993113 in
exposure induces a dramatic loss of bone marrow pro- and pre- vivo, and suggest that this nontoxic antagonist could be
B cells and other hematopoietic cell types, probably through used to block AHR activity either induced acutely by
induction of apoptosis (Mann et al., 1999; Teague et al., 2010; environmental ligands or chronically during a variety of
N’jai A et al., 2011). To determine if CB7993113 could reach pathologic conditions.

Identification of an AHR Antagonist
605
Fig. 9. CB7993113 prevents DMBA-induced toxicity in vivo. Eight-week-old C57BL/6J mice (six per group) were treated by intraperitoneal injection
with vehicle (vegetable oil), 50 mg/kg CH223191, or 50 mg/kg CB7993113 24 hours and 1 hour before DMBA treatment. Vehicle (sesame oil) or 50 mg/kg
DMBA was then administered by oral gavage. Mice were sacrificed 48 hours after DMBA treatment. (A) Liver mRNA was extracted and analyzed by
qPCR for CYP1A1 expression normalized to GAPDH expression. qPCR was performed in duplicate. Data bars represent the mean values from six mice.
(B) Bone marrow was harvested from the right tibia and femur of each mouse and viable cells were counted by Trypan blue exclusion. Data bars
represent the mean values from six mice. (C–E) Bone marrow was harvested and analyzed by flow cytometry. (C) Pre/Pro-B cells are defined as
IgM–/B220+/CD43+/HSA–. (D) Pro-B cells are defined IgM–/B220+/CD43+/HSA+. (E) Neutrophils are defined as CD11bhi/GR-1hi. Data bars represent the
mean values from six mice. Data were analyzed by one-way analysis of variance with Tukey-Kramer post-test (A and B) or Newman-Keuls post-test
(C–E). (*P , 0.05, **P , 0.01, ***P , 0.001 compared with vehicle; +P , 0.05, ++P , 0.01, or +++P , 0.001 compared with DMBA treatment.)
in models of multiple sclerosis and type 1 diabetes (Quintana
et al., 2008; Kerkvliet et al., 2009). Perhaps most strikingly,
AHR modulators have a promising application in stem cell
biology as demonstrated by the expansion of human CD341
progenitor cells from cord blood by culture with the purine-
derived AHR antagonist SR1 (Boitano et al., 2010) and the
development of bipotential hematopoietic stem cells, mega-
karyocytes, and erythroid cells from AHR-activated, induced
pluripotent stem cells (Smith et al., 2013).
Here, we present a combined in silico and high-throughput in
vitro screening platform for the identification of AHR modulators,
both agonists and antagonists. Several bioflavonoids, among
other phytochemicals, have been identified as AHR ligands.
The activity of most of these previously identified com-
pounds ranges from weakly agonistic to weakly antagonistic
(Lu et al., 1996; Zhang et al., 2003). However, select compounds
induce significant AHR-dependent biologic effects, including
several flavones, such as b-NF, used here as a prototypic AHR
agonist. Therefore, we chose a generic substituted flavone
backbone as our pharmacophore upon which to perform shape
and electrostatic comparisons. In a novel approach to generating
a template molecule that could be used to screen over a million
small molecules in silico, the flavone-based model pharmacophore
was expanded into a three-dimensional conformer database and
the molecular shapes were compared with a similar database
containing conformers of 1,176,756 commercially available, drug-
like molecules. A focused library of 197 compounds was selected on
the basis of three-dimensional electrostatic similarities.
Discussion
The AHR has been recognized for many years as a key
regulator of environmental chemical toxicity and carcinogenic-
ity. Indeed, it mediates the biologic effects of some of the most
potent toxicants known, including TCDD. Of equal concern,
from an environmental point of view, is the ability of the AHR
to bind and respond to a variety of structurally disparate
chemicals, including planar polychlorinated biphenyls, dioxins,
polycyclic aromatic hydrocarbons, plant-derived flavonoids and
tryptophan-derived metabolites. Whereas this receptor’s pro-
miscuity and the resulting pathologic consequences are reason
enough to study AHR signaling pathways, in the end, the
AHR’s role in regulating normal and pathologic biologic
activities in the absence of environmental ligands may prove
to be the more compelling story. In this vein, the AHR promotes
the production of IL-17–secreting T cells critical to inflammation-
based diseases and some forms of autoimmunity (Quintana et al.,
2012), and affects production of regulatory T cells that oppose
inflammatory and autoimmune responses (Apetoh et al., 2010;
Gandhi et al., 2010). AHR signaling also plays a key role in
development of gut-associated leukocytes that mediate both the
innate and adaptive immune responses required to prevent
microbial infiltration and the resulting inflammatory colitis (Kiss
et al., 2011; Li et al., 2011). In various model systems, the AHR
also contributes to hematopoietic stem cell development (Boitano
et al., 2010; Casado et al., 2010; Smith et al., 2013). Con-
sequently, the ability to regulate AHR activity may be im-
portant for treatment of hematologic diseases. Accordingly,
The validity of this approach was confirmed by a relatively
studies have shown that AHR modulation attenuates disease high “hit” rate. Of the 197 compounds assayed in the bioassay,

606
Parks et al.
31 consistently induced AHR reporter activity with 27 of those after either oral or intraperitoneal administration were consis-
exhibiting an EC50 of ,10 mM. Four compounds exhibited AHR tent with this conclusion. Most importantly, CB7993113 effec-
antagonist activity, all with IC50s ,5 mM. This represents a hit tively blocked acute, 50 mg/kg DMBA–induced hepatic CYP1A1
rate for probable AHR ligands of 17.8%. Cell-free AHR binding induction and bone marrow toxicity in vivo, demonstrating not
studies performed on one antagonist (CB7993113, Fig. 5) and only sufficient drug absorption but adequate tissue distribution
one agonist (data not shown) confirmed that each molecule was and persistence for inhibiting a strong AHR-dependent (Mann
in fact an AHR ligand. Since both AHR agonists and antagonists et al., 1999; N’jai A et al., 2011) biologic signal.
may be useful in various therapeutic settings, this new
approach to rapidly screening large libraries in silico for AHR as cancer therapeutics. Earlier studies implicated the AHR in
ligands has great utility. the control of important cell functions dysregulated during
Finally, pure AHR antagonists like CB7993113 may be useful
Our AHR ligand screen identified a disproportionate number malignant transformation, including cell growth (Barhoover
of agonists. This result is consistent with the literature in which et al., 2010) and cell migration (Dietrich and Kaina, 2010).
identification of AHR agonists is far more common than the A more recent, high-profile study demonstrated that AHR, consti-
identification of pure antagonists, i.e., competitive AHR antag- tutively activated by an endogenous ligand, drives human
onists that do not exhibit partial agonism at higher doses. To glioblastoma invasion (Opitz et al., 2011). Similarly, accumu-
date, relatively few “pure” AHR antagonists have been discov- lating evidence suggests that hyper-expressed, “constitutively
ered. These include CH223191 (Kim et al., 2006), 6,29,49- active” AHR plays a role in the malignant transformation of
trimethoxyflavone (TMF) (Murray IA et al., 2009), SR1 (Boitano breast epithelial cells (Hall et al., 2010; Brooks and Eltom,
et al., 2010), and GNF351 (Smith et al., 2011). Identification of 2011; Goode et al., 2013), in the expression of epithelial-to-
additional AHR antagonists, such as CB7993113, is important mesenchymal transition markers (Schlezinger et al., 2006;
because each antagonist appears to exhibit unique properties in Shin et al., 2006), and, most recently, in breast cancer stem cell
terms of affinity for AHRs from different species, “ligand homeostasis (Dubrovska et al., 2012; Zhao et al., 2012). As
selectivity,” or ability to block AHR response element– shown here, CB7993113 significantly reduces the invasive
dependent signaling (Smith et al., 2011). For example, SR1 phenotype of ER–/PR–/HER2– breast cancer cells in vitro
is a potent antagonist of the human AHR but has little or no (Figs. 7 and 8). Although further in vivo experiments are
effect on ligand binding to murine AHR. CH223191 exhibits required, these experiments collectively suggest that this,
ligand-selective inhibition, e.g., a propensity to block haloge- or similar AHR antagonists may represent new targeted
nated hydrocarbon–induced but not flavone-induced AHR therapeutics for some kinds of cancers, including triple-
activation (Zhao et al., 2010). In contrast, CB7993113 appears negative breast cancers.
to be a more generally applicable AHR antagonist in that it
inhibits both human and murine AHR, blocks DMBA-, TCDD-,
Acknowledgments
The authors thank Jessalyn Ubellacker and Faye Andrews for their
outstanding technical contributions. MPP acknowledges a free aca-
demic license to the OpenEye Suite of Software.
flavone (b-NF)–, and tryptophan metabolite (e.g., FICZ; data
not shown)–induced AHR activity and reduces baseline AHR
activity (e.g., baseline pGudLuc activity). It shows no agonist
activity either in vitro at doses at least as high as 20 mM or in
vivo at doses at least up to 50 mg/kg. Given the number and
diversity of possible therapeutic applications for AHR antag-
Authorship Contributions
Participated in research design: Parks, Sherr, Schlezinger, Pollastri,
Hahn, Haigh, Vajda, Kozakov, Beglov.
onists, including but probably not limited to human hemato-
poietic stem cell expansion (Boitano et al., 2010), T-cell
expansion (Carlin et al., 2013), inhibition of inflammatory
Th17 development (Veldhoen et al., 2009), reduction in regula-
tory T-cell development (Apetoh et al., 2010), expansion of
erythroid and megakaryocyte lineage cells (Smith et al., 2013),
and inhibition of breast cancer cell invasion (Figs. 7 and 8),
identification of new, nontoxic AHR antagonists is of consider-
able import. Furthermore, expanding the database of AHR
antagonists, which tend to be dissimilar in structure, will
facilitate the definition of critical structural characteristics that
confer the ability to inhibit, as opposed to induce, AHR activity.
Again, this knowledge is critical to the rational design of more
potent, nontoxic, pure AHR antagonists for therapeutic use.
Paramount to the identification of a novel antagonist is the
confirmation of its drug-likeness and lack of toxicity. In our in
vitro screen, compounds exhibiting any level of toxicity in murine
H1G1 hepatoma cells were discarded. CB7993113 exhibited no
toxicity when added at concentrations at least up to 20 mM to
several human cells including HepG2 hepatoma cells, BP1, D3,
Hs578T, or MDA-MB-231 breast cancer cells, and primary
human-induced pluripotent stem cells (data not shown). Further-
more, CB7993113 is predicted to conform to Lipinski’s rules and
thereby is expected to demonstrate oral bioavailability. Pharma-
cokinetics studies demonstrating significant serum bioavailability
Conducted experiments: Parks, Stanford, Novikov, Franks, Hahn,
Narasimhan, Kozakov, Beglov, Schlezinger.
Contributed new reagents or analytic tools: Ashton, Hopper,
Pollastri.
Performed data analysis: Parks, Sherr, Schlezinger, Pollastri,
Hahn, Vajda, Kozakov, Beglov.
Wrote or contributed to the writing of the manuscript: Parks,
Pollastri, Hahn, Kozakov, Beglov, Vajda, Schlezinger, Sherr.
References
Apetoh L, Quintana FJ, Pot C, Joller N, Xiao S, Kumar D, Burns EJ, Sherr DH,
Weiner HL, and Kuchroo VK (2010) The aryl hydrocarbon receptor interacts with
c-Maf to promote the differentiation of type 1 regulatory T cells induced by IL-27.
Nat Immunol 11:854–861.
Barhoover MA, Hall JM, Greenlee WF, and Thomas RS (2010) Aryl hydrocarbon
receptor regulates cell cycle progression in human breast cancer cells via a func-
tional interaction with cyclin-dependent kinase 4. Mol Pharmacol 77:195–201.
Berman HM, Battistuz T, Bhat TN, Bluhm WF, Bourne PE, Burkhardt K, Feng Z,
Gilliland GL, Iype L, Jain S et al. (2002) The Protein Data Bank. Acta Crystallogr
D Biol Crystallogr 58:899–907.
Boitano AE, Wang J, Romeo R, Bouchez LC, Parker AE, Sutton SE, Walker JR,
Flaveny CA, Perdew GH, and Denison MS et al. (2010) Aryl hydrocarbon receptor
antagonists promote the expansion of human hematopoietic stem cells. Science
329:1345–1348.
Brenke R, Kozakov D, Chuang GY, Beglov D, Hall D, Landon MR, Mattos C,
and Vajda S (2009) Fragment-based identification of druggable ‘hot spots’ of pro-
teins using Fourier domain correlation techniques. Bioinformatics 25:621–627.
Brooks B, Bruccoleri R, Olafson B, States D, Swaminathan S, and Karplus M (1983)
CHARMM: a program for macromolecular energy, minimization, and dynamics
calculations. J Comput Chem 4:187–217.
Brooks J and Eltom SE (2011) Malignant transformation of mammary epithelial cells
by ectopic overexpression of the aryl hydrocarbon receptor. Curr Cancer Drug
Targets 11:654–669.

Identification of an AHR Antagonist
607
Burbach KM, Poland A, and Bradfield CA (1992) Cloning of the Ah-receptor cDNA
reveals a distinctive ligand-activated transcription factor. Proc Natl Acad Sci USA
89:8185–8189.
Carlin SM, Ma DD, and Moore JJ (2013) T-cell potential of human adult and cord
blood hemopoietic stem cells expanded with the use of aryl hydrocarbon receptor
antagonists. Cytotherapy 15:224–230.
Lipinski CA, Lombardo F, Dominy BW, and Feeney PJ (2001) Experimental and
computational approaches to estimate solubility and permeability in drug discov-
ery and development settings. Adv Drug Deliv Rev 46:3–26.
Lu YF, Santostefano M, Cunningham BD, Threadgill MD, and Safe S (1996)
Substituted flavones as aryl hydrocarbon (Ah) receptor agonists and antagonists.
Biochem Pharmacol 51:1077–1087.
Caruso JA, Mathieu PA, Joiakim A, Leeson B, Kessel D, Sloane BF, and Reiners JJ,
Jr (2004) Differential susceptibilities of murine hepatoma 1c1c7 and Tao cells to
the lysosomal photosensitizer NPe6: influence of aryl hydrocarbon receptor on ly-
sosomal fragility and protease contents. Mol Pharmacol 65:1016–1028.
Caruso JA, Mathieu PA, Joiakim A, Zhang H, and Reiners JJ, Jr (2006) Aryl hydro-
carbon receptor modulation of TNFalpha -induced apoptosis and lysosomal disruption
in a hepatoma model that is caspase-8 independent. J Biol Chem 281:10954–10967.
Casado FL, Singh KP, and Gasiewicz TA (2010) The aryl hydrocarbon receptor:
regulation of hematopoiesis and involvement in the progression of blood diseases.
Blood Cells Mol Dis 44:199–206.
Dietrich C and Kaina B (2010) The aryl hydrocarbon receptor (AhR) in the regulation
of cell-cell contact and tumor growth. Carcinogenesis 31:1319–1328.
Dolwick KM, Schmidt JV, Carver LA, Swanson HI, and Bradfield CA (1993) Cloning
and expression of a human Ah receptor cDNA. Mol Pharmacol 44:911–917.
Dubrovska A, Hartung A, Bouchez LC, Walker JR, Reddy VA, Cho CY, and Schultz
Maecker B, Sherr DH, Vonderheide RH, von Bergwelt-Baildon MS, Hirano N,
Anderson KS, Xia Z, Butler MO, Wucherpfennig KW, and O’Hara C et al. (2003)
The shared tumor-associated antigen cytochrome P450 1B1 is recognized by spe-
cific cytotoxic T cells. Blood 102:3287–3294.
Mann KK, Matulka RA, Hahn ME, Trombino AF, Lawrence BP, Kerkvliet NI,
and Sherr DH (1999) The role of polycyclic aromatic hydrocarbon metabolism in
dimethylbenz[a]anthracene-induced pre-B lymphocyte apoptosis. Toxicol Appl
Pharmacol 161:10–22.
Motto I, Bordogna A, Soshilov AA, Denison MS, and Bonati L (2011) New aryl hy-
drocarbon receptor homology model targeted to improve docking reliability.
J Chem Inf Model 51:2868–2881.
Murray GI, Patimalla S, Stewart KN, Miller ID, and Heys SD (2010) Profiling the
expression of cytochrome P450 in breast cancer. Histopathology 57:202–211.
Murray IA, Flaveny CA, DiNatale BC, Chairo CR, Schroeder JC, Kusnadi A,
and Perdew GH (2009) Antagonism of aryl hydrocarbon receptor signaling by
6,29,49-trimethoxyflavone. J Pharmacol Exp Ther 332:135–144.
PG (2012) CXCR4 activation maintains
a stem cell population in tamoxifen-
resistant breast cancer cells through AhR signalling. Br J Cancer 107:43–52.
Dvorak Z, Vrzal R, Henklova P, Jancova P, Anzenbacherova E, Maurel P, Svecova L,
Pavek P, Ehrmann J, and Havlik R et al. (2008) JNK inhibitor SP600125 is
a partial agonist of human aryl hydrocarbon receptor and induces CYP1A1 and
CYP1A2 genes in primary human hepatocytes. Biochem Pharmacol 75:580–588.
Fiser A and Sali A (2003) Modeller: generation and refinement of homology-based
protein structure models. Methods Enzymol 374:461–491.
Gandhi R, Kumar D, Burns EJ, Nadeau M, Dake B, Laroni A, Kozoriz D, Weiner HL,
and Quintana FJ (2010) Activation of the aryl hydrocarbon receptor induces hu-
man type 1 regulatory T cell-like and Foxp3(1) regulatory T cells. Nat Immunol 11:
846–853.
N’jai AU, Larsen MC, Bushkofsky JR, Czuprynski CJ, and Jefcoate CR (2011) Acute
disruption of bone marrow hematopoiesis by benzo(a)pyrene is selectively reversed
by aryl hydrocarbon receptor-mediated processes. Mol Pharmacol 79:724–734.
Nagy SR, Sanborn JR, Hammock BD, and Denison MS (2002) Development of a green
fluorescent protein-based cell bioassay for the rapid and inexpensive detection and
characterization of ah receptor agonists. Toxicol Sci 65:200–210.
Nebert DW, Petersen DD, and Puga A (1991) Human AH locus polymorphism and
cancer: inducibility of CYP1A1 and other genes by combustion products and dioxin.
Pharmacogenetics 1:68–78.
Opitz CA, Litzenburger UM, Sahm F, Ott M, Tritschler I, Trump S, Schumacher
T, Jestaedt L, Schrenk D, and Weller M et al. (2011) An endogenous tumour-
promoting ligand of the human aryl hydrocarbon receptor. Nature 478:
197–203.
Pfaffl MW (2001) A new mathematical model for relative quantification in real-time
RT-PCR. Nucleic Acids Res 29:e45.
Goode GD, Ballard BR, Manning HC, Freeman ML, Kang Y, and Eltom SE (2013)
Knockdown of aberrantly upregulated aryl hydrocarbon receptor reduces tumor
growth and metastasis of MDA-MB-231 human breast cancer cell line. Int J Cancer
133:2769–2780.
Gurnell M, Wentworth JM, Agostini M, Adams M, Collingwood TN, Provenzano C,
Browne PO, Rajanayagam O, Burris TP, and Schwabe JW et al. (2000) A dominant-
negative peroxisome proliferator-activated receptor gamma (PPARgamma) mutant is
a constitutive repressor and inhibits PPARgamma-mediated adipogenesis. J Biol
Chem 275:5754–5759.
Hall JM, Barhoover MA, Kazmin D, McDonnell DP, Greenlee WF, and Thomas RS
(2010) Activation of the aryl-hydrocarbon receptor inhibits invasive and metastatic
features of human breast cancer cells and promotes breast cancer cell differenti-
ation. Mol Endocrinol 24:359–369.
Quintana FJ, Basso AS, Iglesias AH, Korn T, Farez MF, Bettelli E, Caccamo M,
Oukka M, and Weiner HL (2008) Control of T(reg) and T(H)17 cell differentiation
by the aryl hydrocarbon receptor. Nature 453:65–71.
Quintana FJ, Jin H, Burns EJ, Nadeau M, Yeste A, Kumar D, Rangachari M, Zhu C,
Xiao S, and Seavitt J et al. (2012) Aiolos promotes TH17 differentiation by directly
silencing Il2 expression. Nat Immunol 13:770–777.
Santostefano M, Merchant M, Arellano L, Morrison V, Denison MS, and Safe S (1993)
alpha-Naphthoflavone-induced CYP1A1 gene expression and cytosolic aryl hy-
drocarbon receptor transformation. Mol Pharmacol 43:200–206.
Hankinson O, Brooks BA, Weir-Brown KI, Hoffman EC, Johnson BS, Nanthur J,
Reyes H, and Watson AJ (1991) Genetic and molecular analysis of the Ah receptor
and of Cyp1a1 gene expression. Biochimie 73:61–66.
Hardy RR and Hayakawa K (2001) B cell development pathways. Annu Rev Immunol
19:595–621.
Hawkins PC and Nicholls A (2012) Conformer generation with OMEGA: learning
from the data set and the analysis of failures. J Chem Inf Model 52:2919–2936.
Hawkins PCD, Skillman AG, and Nicholls A (2007) Comparison of shape-matching
and docking as virtual screening tools. J Med Chem 50:74–82.
Scheuermann TH, Tomchick DR, Machius M, Guo Y, Bruick RK, and Gardner KH
(2009) Artificial ligand binding within the HIF2alpha PAS-B domain of the HIF2
transcription factor. Proc Natl Acad Sci USA 106:450–455.
Scheuermann TH, Li Q, Ma HW, Key J, Zhang L, Chen R, Garcia JA, Naidoo J,
Longgood J, and Frantz DE et al. (2013) Allosteric inhibition of hypoxia inducible
factor-2 with small molecules. Nat Chem Biol 9:271–276.
Schlezinger JJ, Liu D, Farago M, Seldin DC, Belguise K, Sonenshein GE, and Sherr
DH (2006) A role for the aryl hydrocarbon receptor in mammary gland tumori-
genesis. Biol Chem 387:1175–1187.
Hughes L, Malone C, Chumsri S, Burger AM, and McDonnell S (2008) Character-
isation of breast cancer cell lines and establishment of a novel isogenic subclone to
study migration, invasion and tumourigenicity. Clin Exp Metastasis 25:549–557.
Karchner SI, Franks DG, Kennedy SW, and Hahn ME (2006) The molecular basis for
differential dioxin sensitivity in birds: role of the aryl hydrocarbon receptor. Proc
Natl Acad Sci USA 103:6252–6257.
Kerkvliet NI, Steppan LB, Vorachek W, Oda S, Farrer D, Wong CP, Pham D,
and Mourich DV (2009) Activation of aryl hydrocarbon receptor by TCDD prevents
diabetes in NOD mice and increases Foxp31 T cells in pancreatic lymph nodes.
Immunotherapy 1:539–547.
Kim SH, Henry EC, Kim DK, Kim YH, Shin KJ, Han MS, Lee TG, Kang JK,
Gasiewicz TA, and Ryu SH et al. (2006) Novel compound 2-methyl-2H-pyrazole-3-
carboxylic acid (2-methyl-4-o-tolylazo-phenyl)-amide (CH-223191) prevents 2,3,7,8-
TCDD-induced toxicity by antagonizing the aryl hydrocarbon receptor. Mol Pharmacol
69:1871–1878.
Kiss EA, Vonarbourg C, Kopfmann S, Hobeika E, Finke D, Esser C, and Diefenbach A
(2011) Natural aryl hydrocarbon receptor ligands control organogenesis of in-
testinal lymphoid follicles. Science 334:1561–1565.
Shin SR, Sánchez-Velar N, Sherr DH, and Sonenshein GE (2006) 7,12-dimethylbenz
(a)anthracene treatment of a c-rel mouse mammary tumor cell line induces epi-
thelial to mesenchymal transition via activation of nuclear factor-kappaB. Cancer
Res 66:2570–2575.
Smith BW, Rozelle SS, Leung A, Ubellacker J, Parks A, Nah SK, French D, Gadue P,
Monti S, and Chui DH et al. (2013) The aryl hydrocarbon receptor directs hema-
topoietic progenitor cell expansion and differentiation. Blood 122:376–385.
Smith KJ, Murray IA, Tanos R, Tellew J, Boitano AE, Bisson WH, Kolluri SK, Cooke
MP, and Perdew GH (2011) Identification of a high-affinity ligand that exhibits
complete aryl hydrocarbon receptor antagonism.
318–327.
J Pharmacol Exp Ther 338:
Sukhumavasi W, Egan CE, and Denkers EY (2007) Mouse neutrophils require JNK2
MAPK for Toxoplasma gondii-induced IL-12p40 and CCL2/MCP-1 release.
J
Immunol 179:3570–3577.
Teague JE, Ryu H-Y, Kirber M, Sherr DH, and Schlezinger JJ (2010) Proximal events
in 7,12-dimethylbenz[a]anthracene-induced, stromal cell-dependent bone marrow
B cell apoptosis: stromal cell-B cell communication and apoptosis signaling. J
Immunol 185:3369–3378.
Kozakov D, Hall DR, Chuang GY, Cencic R, Brenke R, Grove LE, Beglov D, Pelletier
J, Whitty A, and Vajda S (2011) Structural conservation of druggable hot spots in
protein-protein interfaces. Proc Natl Acad Sci USA 108:13528–13533.
Lahoti TS, John K, Hughes JM, Kusnadi A, Murray IA, Krishnegowda G, Amin S,
and Perdew GH (2013) Aryl hydrocarbon receptor antagonism mitigates cytokine-
mediated inflammatory signalling in primary human fibroblast-like synoviocytes.
Ann Rheum Dis 72:1708–1716.
Trombino AF, Near RI, Matulka RA, Yang S, Hafer LJ, Toselli PA, Kim DW, Rogers
AE, Sonenshein GE, and Sherr DH (2000) Expression of the aryl hydrocarbon
receptor/transcription factor (AhR) and AhR-regulated CYP1 gene transcripts in
a rat model of mammary tumorigenesis. Breast Cancer Res Treat 63:117–131.
Trott O and Olson AJ (2010) AutoDock Vina: improving the speed and accuracy of
docking with a new scoring function, efficient optimization, and multithreading.
J Comput Chem 31:455–461.
Lee JS, Cella M, McDonald KG, Garlanda C, Kennedy GD, Nukaya M, Mantovani A,
Kopan R, Bradfield CA, and Newberry RD et al. (2012) AHR drives the de-
velopment of gut ILC22 cells and postnatal lymphoid tissues via pathways de-
pendent on and independent of Notch. Nat Immunol 13:144–151.
Li NY, Weber CE, Wai PY, Cuevas BD, Zhang J, Kuo PC, and Mi Z (2013) An MAPK-
dependent pathway induces epithelial-mesenchymal transition via Twist activa-
tion in human breast cancer cell lines. Surgery 154:404–410.
Li Y, Innocentin S, Withers DR, Roberts NA, Gallagher AR, Grigorieva EF, Wilhelm
C, and Veldhoen M (2011) Exogenous stimuli maintain intraepithelial lymphocytes
via aryl hydrocarbon receptor activation. Cell 147:629–640.
Veldhoen M, Hirota K, Westendorf AM, Buer J, Dumoutier L, Renauld JC,
and Stockinger B (2008) The aryl hydrocarbon receptor links TH17-cell-mediated
autoimmunity to environmental toxins. Nature 453:106–109.
Veldhoen M, Hirota K, Christensen J, O’Garra A, and Stockinger B (2009) Natural
agonists for aryl hydrocarbon receptor in culture medium are essential for optimal
differentiation of Th17 T cells. J Exp Med 206:43–49.
Wu D, Nishimura N, Kuo V, Fiehn O, Shahbaz S, Van Winkle L, Matsumura F,
and Vogel CF (2011) Activation of aryl hydrocarbon receptor induces vascular in-
flammation and promotes atherosclerosis in apolipoprotein E-/- mice. Arterioscler
Thromb Vasc Biol 31:1260–1267.

608
Parks et al.
Wu D, Potluri N, Kim Y, and Rastinejad F (2013) Structure and dimerization prop-
erties of the aryl hydrocarbon receptor PAS-A domain. Mol Cell Biol 33:4346–4356.
Xing Y, Nukaya M, Satyshur KA, Jiang L, Stanevich V, Korkmaz EN, Burdette L,
Kennedy GD, Cui Q, and Bradfield CA (2012) Identification of the Ah-receptor
structural determinants for ligand preferences. Toxicol Sci 129:86–97.
Xu M and Miller MS (2004) Determination of murine fetal Cyp1a1 and 1b1 expres-
sion by real-time fluorescence reverse transcription-polymerase chain reaction.
Toxicol Appl Pharmacol 201:295–302.
Yang X, Liu D, Murray TJ, Mitchell GC, Hesterman EV, Karchner SI, Merson RR, Hahn
ME, and Sherr DH (2005) The aryl hydrocarbon receptor constitutively represses c-myc
transcription in human mammary tumor cells. Oncogene 24:7869–7881.
Yang X, Solomon S, Fraser LR, Trombino AF, Liu D, Sonenshein GE, Hestermann
EV, and Sherr DH (2008) Constitutive regulation of CYP1B1 by the aryl hydro-
carbon receptor (AhR) in pre-malignant and malignant mammary tissue. J Cell
Biochem 104:402–417.
Zhang S, Qin C, and Safe SH (2003) Flavonoids as aryl hydrocarbon receptor
agonists/antagonists: effects of structure and cell context. Environ Health Perspect
111:1877–1882.
Zhao B, Degroot DE, Hayashi A, He G, and Denison MS (2010) CH223191 is a ligand-
selective antagonist of the Ah (Dioxin) receptor. Toxicol Sci 117:393–403.
Zhao S, Kanno Y, Nakayama M, Makimura M, Ohara S, and Inouye Y (2012) Acti-
vation of the aryl hydrocarbon receptor represses mammosphere formation in
MCF-7 cells. Cancer Lett 317:192–198.
Zudaire E, Cuesta N, Murty V, Woodson K, Adams L, Gonzalez N, Martínez A,
Narayan G, Kirsch I, and Franklin W et al. (2008) The aryl hydrocarbon receptor
repressor is a putative tumor suppressor gene in multiple human cancers. J Clin
Invest 118:640–650.
Address correspondence to: Dr. David H. Sherr, Department of Environ-
mental Health, Boston University School of Public Health, 72 East Concord
Street (R-408), Boston, MA 02118. E-mail: dsherr@bu.edu
Zhang S, Lei P, Liu X, Li X, Walker K, Kotha L, Rowlands C, and Safe S (2009) The
aryl hydrocarbon receptor as a target for estrogen receptor-negative breast cancer
chemotherapy. Endocr Relat Cancer 16:835–844.