Discovery and mechanistic characterization of a novel AhR receptor ligand that induces apoptosis in cancer cells


 Background: The aryl hydrocarbon receptor (AhR) is a ligand-activated transcription factor and a member of the bHLH/PAS (basic Helix-Loop-Helix/Per-Arnt-Sim) family of proteins. The AhR was cloned and characterized for its role in mediating the toxicity of dioxins. Subsequent research has identified AhR’s role in the suppression of cancer cell growth. We hypothesized that that the AhR is a molecular target for therapeutic interventions, and that activation of the AhR by select AhR modulators in cancer cells could have anti-cancer properties including cell death. This study describes the discovery and characterization of a new class of anti-cancer agents targeting the AhR.Methods: We employed two independent small molecule screening approaches to identify novel modulators of the AhR with unique anti-cancer properties. We established the AhR selective effects of a highly selective modulator in cancer cells with or without the AhR expression and identified the mechanism of action of this anti-cancer compound.Results: We report the identification of CGS-15943, uniquely selective AhR modulator, that activates AhR signaling and induces apoptosis in an AhR-dependent manner in liver and breast cancer cell models. Investigation of the downstream signaling pathway of this newly identified modulator revealed novel upregulation of Fas-ligand, which is required for AhR-mediated apoptosis.Conclusions: We identified CGS-15943 as a novel AhR-selective modulator with anti-cancer properties using two parallel and distinct screening strategies. This compound induced AhR-dependent apoptosis in multiple mouse and human liver cancer cells. Our results provide a basis for the development of a new class of anti-cancer therapeutics targeting an underappreciated molecular target, the AhR.

Just as the range of gene targets of the AhR is broad, so too are the ligands capable of initiating these responses. These include TCDD as well as dioxin-like compounds (DLCs), halogenated polyaromatic hydrocarbons, and putative endogenous AhR activators including tryptophan metabolites 6-formylindolo [3,2-b]carbazole (FICZ) and 2-(1'H-indole-3'-carbonyl)-thiazole-4-carboxylic acid methyl ester (ITE) [15]. In recent years there has been a steady increase in the identi cation and understanding of the diverse array of AhR ligands. For example, there is signi cant interest in the AhR-mediated immunomodulatory effects of the tryptophan metabolite kynurenine [16]. An aryl-hydrocarbon receptor antagonist purine derivative StemRegenin1 was shown to promote expansion of human hematopoietic stem cells (HSCs) [17], which has recently progressed to clinical trials for expanding CD34 + HSCs umbilical cord blood [18]. Excitingly, the landscape of AhR activators and their phenotypic effects continues to expand.
In addition to the ligand-mediated actions, in vivo studies have also suggested a role for the AhR in cancer outside of exogenous ligand activation. AhR-null transgenic adenocarcinoma of the mouse prostate (TRAMP) mice develop prostate tumors faster and more aggressively that AhR-positive TRAMP mice [13]. Similarly, AhR null mice injected with diethylnitrosamine (DEN), which induces liver tumors after a single dose, exhibit increased expression of markers of proliferation and decreased expression of tumor suppressor genes compared with AhR-positive counterparts [12]. Taken together, there is a role for the AhR in regulating cancer growth [24], and activation of the anti-cancer effects of the AhR with appropriate ligands is an exciting opportunity.
Given the therapeutic potential of AhR modulation, it is necessary to tease apart AhR activation from the adverse effects of dioxins in order to realize a new class of AhR-based therapeutics. Thus, we identi ed and characterized small molecules already in use in the clinic that activate the AhR. We and others previously showed that the immunomodulatory agent le unomide, used in the treatment of rheumatoid arthritis, is an AhR ligand [19,20]. In addition, we showed that the AhR mediates some of the effects of le unomide in inhibiting melanoma cells [21]. With respect to cellular proliferation, we have shown that the tyrosine kinase inhibitor and AhR-ligand SU5416 (semaxanib) induces a p21-depedent growth inhibition in liver cancer cells. Likewise, our group has shown that the anti-androgen utamide is an AhR ligand that suppresses hepatocellular carcinoma cell proliferation through AhR-dependent induction of TGF-β1 [22]. Beyond suppression of cellular proliferation, we have also shown that the selective-estrogen receptor modulator raloxifene activates the AhR and induces apoptosis in liver and MDA-MB-231 breast cancer cells in an AhR-dependent manner [23].
In the present study, we utilized two independent small molecule screening approaches to identify AhR ligands with ability to activate the AhR without necessarily requiring canonical xenobiotic response element driven transcription. Excitingly, we identi ed the triazoloquinazoline non-xanthine adenosine A 1 and A 2A antagonist CGS-15934 as a novel AhR ligand [25]. We showed the ability of CGS-15943 to induce AhR-dependent apoptosis in mouse hepatoma and human hepatocellular carcinoma cells, as well as in triple negative breast cancer cells. The selective AhR-dependent nature of CGS-15943 in these model systems is unique. In characterizing the downstream effects of this novel AhR-ligand, we showed that CGS-15943 induces apoptosis thorough induction of FasL and subsequent activation of the extrinsic and intrinsic caspase cascades.

Methods
Reagents CGS-15943 and MRS1220 were purchased from Tocris. The purity of CGS-15943 was 100% as determined by HPLC analysis. The purity of MRS 1220 was 99.1% as determined by HPLC analysis. Cycloheximide was purchased from Sigma. All other reagents were purchased from Sigma unless otherwise indicated.
Cell culture models of differential AhR expression Hepa1, TAO, C4, and vT{2} cells were used as described previously [5]. HEK293T, HepG2, and MDA-MB-468 cells were purchased from ATCC. All cells were cultured in DMEM with antibiotics (streptomycin and penicillin) from Mediatech, Inc. supplemented with 10% FBS (Tissue Culture Biologicals). Cells were grown in a humidi ed 5% CO2 atmosphere.

AhR ligand binding assays
Competitive ligand binding assays were performed with [3 H ]-3MC as described previously [44]. Differential proteolysis assays were performed by incubating whole cell Hepa1 extracts prepared as described previously [21] with subtilisin at pre-determined time intervals, quenching reactions with SDScontaining buffer, and analyzing differential AhR proteolysis (band intensity) by Western blot.
Immuno uorescence and ow cytometry Immuno uorescence experiments for nuclear translocation were performed as described previously [5,16]. DAPI staining was used to assess nuclear fragmentation as described previously [45]. Brie y, cells undergoing apoptosis were collected by centrifugation, xed, and stained overnight with DAPI (Molecular Probes, Invitrogen). Quantitation of apoptosis was determined by dividing the number of apoptotic cells (condensed/fragmented nuclear morphology) by the total number of cells counted. At least 300 cells were counted per slide, and experiments were performed in triplicate.
Flow cytometry was performed with an FC500 as described previously [20,21]. Annexin V detection was performed with the Biovision Annexin V kit, using a FITC tagged Annexin V antibody for single stain experiments, and Annexin V-PE-Cy5 for dual stain (GFP) experiments. The JO2 and CD95 antibodies for detection of Fas and FasL, respectively, were a kind gift from the laboratory of Dr. Nancy Kerkvliet.\ Reporter gene analysis XRE-based reporter gene assays were performed as described previously [20]. Brie y, Hepa1.1 cells were plated in 96 well plates at a density of 1 × 10 4 cells/well, grown overnight, and treated with compounds the following day for 24 hours [46,47]. For the GAL4/AhR screening assay, HEK293T cells were seeded in 96 well plates at a density of 1 × 10 4 cells/well in 100 µL of 10% DMEM and grown overnight. The following day, each well was transfected with 70 ng of Gal-Luciferase and 10 ng each of GAL-AhR, GAD-ARNT, and B-GAL, a plasmid expressing beta-galactosidase under a constitutive promoter that was used to normalize for transfection e ciency. Transfections were performed by combining (per well) DNA with 200 nL of PLUS reagent (Invitrogen, Carlsbad USA) in a total of 10 µL of OPTI-MEM (Invitrogen), incubating this mixture for 15 minutes, and combining it with 200 nL Lipofectamine (Invitrogen) diluted in 10 µL of OPTI-MEM. The resulting solution (20 µL/well) was incubated for an additional 15 minutes and then added directly to cells. Transfection solutions were prepared as a master mix to minimize well-towell transfection variations. Transfected cells were incubated overnight and treated with compounds as described above for the XRE-based screen. Analysis of luciferase activity from the GAL/AhR screening system was performed essentially as described above, with minor modi cations. Cells were lysed in 150 µL of cell lysis reagent, for which 100 µL of the lysate was used for measurement of luciferase activity and 10 µL was used for analysis of beta-galactosidase activity.
Gene expression analysis Semi-quantitative gene expression analysis was performed as described previously [5,17]. qPCR analysis was performed as described previously [21]. The relevant primers for qPCR analysis are described in the aforementioned references. Additional semi-quantitative PCR primer sequences were as follows: BAX FP: For targeted gene expression analysis of apoptosis associated genes, we used the PAMM-012A Apoptosis PCR Array from SA Biosciences (Qiagen) according to the manufactures recommended protocol. Data were analyzed using the online SA Biosciences analysis tools available at: http://pcrdataanalysis.sabiosciences.com/pcr/arrayanalysis.php. AhR knockdown Generation of human HepG2 cells stably expressing shRNAs to knockdown AhR was performed by lentiviral-mediated transduction as described previously [16]. For doxycycline-inducible AhR knockdown, we used the same vectors described previously [16]. Generation of HepG2 cells stably expressing shRNAs (non-inducible), we used the pLKO.1 lentivirus system. All vectors were purchased from Open Biosystems (Thermo Scienti c) and Addgene. Transient knockdown of AhR in Hepa1 cells was achieved with morpholinos targeting the AhR transcription start site. Morpholinos were linked to a polyarginine-tag to facilitate uptake after addition to cell culture media, and a FITC label to con rm uptake. Morpholinos were a kind gift from the laboratory of Dr. Nancy Kerkvliet. Transient knockdown of AhR in human HepG2 cells was performed using siRNAs from Dharmacon (Thermo Scienti c) for luciferase (control) and AhR (proprietary sequence). Transfection of siRNAs was performed with DharmaFECT-1 (Thermo Scienti c) according to the manufacture's recommended protocol. MDA-MB-468 inducible knockdowns were prepared as described previously [21].

FasL knockdown
Knockdown of FasL in mouse Hepatoma (Hepa1 cells) was performed with pGIPZ vectors (shScramble or shFasL) purchased from Open Biosystems. Stable-expressing cells were generated by lentiviralmediated transduction as described previously [21]. Cell sorting (Mo-Flow XP) was used to obtain a population of stably infected cells using the co-expressed GFP reporter contained on the GIPZ vector. Knockdowns were con rmed by qPCR. Viability and proliferation assays Viability and proliferation assays were performed as described previously [21]. Brie y, cell viability was determined using the Promega GLO-titer reagent and analyzed with a luminometer. Plates were allowed to equilibrate to room temperature prior to addition of the GLO-Titer reagent, and incubated for a minimum of 30 minutes before analysis with a luminometer. Real time analysis of cellular proliferation was performed using a DP-RTCA assay as described previously [21]. Colony forming assays were performed as described previously [45].

Caspase Activity Assays
Analysis of caspase 3/7, caspase 8, and caspase 9 activation was performed using kits from Promega (Caspase-GLO assays) according to the manufacturer's recommended protocol.

Western blotting
Western blotting was performed as described previously [21]. Antibodies used were as follows: AhR Data analysis Data were analyzed by ANOVA as appropriate, and multiple comparisons were performed with Tukey's post-hoc analysis. All Caspase, Annexin V, Real-time proliferation, and viability assays are representative of at least three similar experiments. Flow cytometry data was analyzed with FlowJo (Tree Star Inc). Values of p < 0.05 were considered statistically signi cant.

Results
A dual screening approach with XRE-reporter and Heterologous AhR reporter genes identi es CGS-15943 as a novel AhR activator.
Our study began with an attempt to identify novel ligands of the AhR and study their functional effects in cells, with the goal of isolating molecules with anti-cancer potential. XRE reporter genes are based on the canonical XRE binding sequence such as those found in the promoter of CYP1A1, and are e ciently activated by AhR ligands [26]. However, because it is these same ligands that are often associated with dioxin-like toxicity, we considered that it would be advantageous to employ a second screening approach that would allow us to identify ligands capable of bringing the AhR into a transcriptionally activate state without the strict requirement of canonical XRE binding per se. Thus, we employed a heterologous yeast GAL4/AhR fusion reporter in which the AhR DNA-binding domain (DBD) was replaced with the GAL4 yeast DBD coupled with a GAL-activating domain fused with ARNT. Together, these constructs were capable of activating a luciferase reporter gene driven by a GAL4-response element. A schematic of the screening constructs used in this study are shown in Fig. 1A. We used two independent screenings in parallel, Hepa1 cells expressing the pGudLuc1.1 reporter gene for the XRE-based screen, and heterologous GAL4-reporter based AhR activation screen. Using these two systems, we screened the library of pharmacologically active compounds (LOPAC1280) and 50,000 Chembridge pharmacophore library. The screening results with LOPAC library are shown in Fig. 1B. There were unique hits in each reporter system, but among the top 10 hits for each screen, only 4 compounds were shared in common ( Fig. 1D). We focused on those compounds highly active in the non-canonical XRE system but minimal activity in the XRE screen.
Con rmatory screening of CGS-15943 showed that it activated the GAL4/AhR reporter system in 239T cells in a dose-dependent manner, and was also capable of activating the reporter system to a much greater extent than the relative maximal concentration of TCDD (1 nM) (Fig. 1D). To our surprise, con rmatory screening in the XRE-system performed at a shorter timepoint (4 hrs vs ~ 24 for the screen) showed that CGS-15943 was also capable of dose-dependently activating the XRE-reporter system (Fig. 1E). We did note a high degree of cell death in the Hepa1 cells treated with CGS-15943 for prolonged treatment times. Thus, CGS-15943 does activate AhR transcription, but on our initial screen was falsely negative in the XRE-based screen due to a high proportion of cell death.

CGS-15943 is a ligand of the AhR
Having identi ed CGS-15943 as an AhR activator, we next performed in vitro ligand-displacement assays to con rm speci c AhR binding. CGS-15943 exhibited an EC 50 for binding of 96 nM, while TCDD had an EC 50 of 3.2 nM (Fig. 1G), consistent with previous observations for TCDD [15]. To provide a separate line of evidence for direct binding of CGS-15943 to the AhR, we next showed that CGS-15943 alters the proteolysis of AhR in limited proteolysis experiments, which has been demonstrated for TCDD [27] (Fig. 1H). We also con rmed that CGS-15943 induces nuclear translocation of the AhR using immunohistochemical studies (Fig. 1I). Because we observed XRE-activation in a shorter-term reporter assay, we next tested whether CGS-15943 could activate several known classic AhR-target genes.
Consistent with this possibility, CGS-15943 activated the AhR target genes CYP1A1 and NADPH: quinone oxidoreductase (NQO1) in mouse Hep1c1c7 cells (Fig. 1K), and CYP1A1 in the HepG2 human hepatoma cell line (Fig. 1J), in both cases to a similar degree as 1 nM TCDD. Taken together, these results showed that our screening strategy was effective in identifying novel AhR ligands, including CGS-15943.

CGS-15943 induces apoptosis
In con rmatory screening, we observed that CGS-15943 potently induces cell death in cultures of Hepa1 cells. Given its ability to active the AhR, we were intrigued by the possibility the apoptotic effects of CGS-15943's may be mediated by the AhR. To address this possibility, we evaluated the effects of CGS-15943 on Hepa1 cells with time lapse video microscope (Video S1). We noted signi cant rounding of cells and membrane blebbing in cells treated with CGS-15943 ( Fig. 2A), consistent with apoptosis. We quantitatively evaluated the effect of CGS-15943 on Hepa1 cells using an xCelligence assay, which we have used previously to study the phenotypic effects of AhR-ligands [23]. Addition of CGS-15943 to proliferating Hepa1 cells decreased the normalized cell index. Concentrations of CGS-15943 from 20 µM to 2 µM led to a dramatic decline in cell abundance, approaching that of the initial cell index at plating by 48 hours (Fig. 2B). Quantitative analysis of the normalized cell index values at 24 hours showed a statistically signi cant decrease in normalized cell index at concentrations of CGS-15943 down to 500 nM, and by 48 hours a signi cant effect of 100 nM CGS-15943 was also observed (Fig. 2B). Using DNA staining and microscopy, we visually con rmed that CGS-15943 induces nuclear fragmentation in Hepa1 cells, consistent with apoptosis (Fig. 2C). We also con rmed that CGS-15943 induces apoptosis using Annexin V binding (Fig. 2D), and activation of Caspases 3/7 (Fig. 2E). In all of our studies, equivalent AhR-activating concentrations of TCDD (1 nM) had no effect on Hepa1 cell with respect to induction of apoptosis.

AhR-dependent induction of apoptosis by CGS-15943
Given that the induction of apoptosis by CGS-15943 occurred at concentrations similar to those required to activate the AhR, we next asked if the effects were AhR-dependent. To address this possibility we rst utilized transient knockdown of the AhR using morpholino oligonucleotides targeting the AhR and conjugated with a cell-penetrating peptide [28]. We con rmed successful knockdown of AhR using a morpholino targeting the AhR AUG start codon (Fig. 3A). Using these knockdown and control cells, we quantitatively assessed apoptosis by counting the number of cells with nuclear fragmentation. We found that AhR knockdown signi cantly decreased the percentage of cells with nuclear fragmentation in response to treatment with CGS-15943 compared to control cells (Fig. 3B). Visually, we also saw a noticeable difference in morphology between AhR knockdown cells treated with CGS-15943 compared to treated control cells, with decreased rounding of cells (Fig. 3C). These results indicated that the induction of apoptosis by CGS-15943 are mediated through AhR activation.
We have previously used Hepa1 cell culture models of high and low expression of AhR to investigate the AhR-dependent phenotypic effects of various ligands [28]. In the present study, we rst turned to C12 and C12 + AhR cells, which are derivative cell lines of the Hepa1 cell line that have low and stably-restored expression of AhR, respectively (Fig. 3D). Using these cells, we found that CGS-15943 increased Annexin V staining only in the presence of AhR (Fig. 3E). Consistent with the results of the transient knockdown experiments, there was a signi cant decrease in the degree of nuclear fragmentation in the absence of AhR compared to AhR-expressing cells (Fig. 3F). Absence of AhR (C12 cells) was associated with a signi cant increase in the number of viable cells compared to C12 + AhR cells (Fig. 3G).
To provide a third independent line of evidence for the requirement for AhR in mediating the apoptotic effects of CGS-15943, we utilized Hepa1 and TAO cells, which we and others have also used to characterize the effects of AhR ligands [28], as TAO cells express very low levels of AhR (Fig. 3H). Using these cells, we rst performed real-time analysis of cellular proliferation as described above. TAO cells were signi cantly resistant to apoptosis induced by CGS-15943 compared to WT Hepa1 cells, but did exhibit some changes presumably due to low-baseline AhR levels (Fig. 3I). We also performed time-lapse microscopy in Hepa1 and TAO cells cultured side-by-side but physically separated in a small culture ask to observe the effects of the compound on the two cell types simultaneously. Consistent with the results of the xCelligence assay (Fig. 3I), TAO cells remained viable in the presence of CGS-15943 while Hepa1 cells underwent rapid and e cient apoptosis ( Figure S2 and Video S2). We con rmed these results with cellular viability assays, in which there was a clear demarcation in the number of viable cells according to AhR expression (Fig. 3J). Interestingly, while CGS-15943 in uenced TAO cell proliferation, there was absolutely no activation of caspase 3/7 in these cells at concentrations of CGS-15943 up to 20 µM (Fig. 3K). In contrast, Hepa1 cells exhibited an approximately 20-fold increase in caspase 3/7 activity that appeared to saturate at a concentration of 5 µM. In colony forming assays, pulsing Hepa1 or TAO cells with CGS-15943 for 24 hours with CGS-15943 at 10 µM or 5 µM (Fig. 3L) led to a complete loss of Hepa1 cell colonies, whereas TAO cells were virtually unaffected. Further, to provide additional evidence for a lack of activation of apoptosis in TAO cells, we performed Western blot analysis for PARP and caspase 3. As shown in Fig. 3M, TAO cells were completely resistant to the formation of cleaved PARP and caspase 3 in response to treatment with CGS-15943. In addition, we found that CGS-15943 decreased levels of AhR in treated Hepa1 cells, consistent with its ability to bind the AhR and induce the phenomenon of receptor recycling [28].
ARNT is required for CGS-15943 mediated apoptosis ARNT is the obligate heterodimer of the AhR, and is required for the transcriptional activity of the AhR [28]. Thus, it was logical to ask whether ARNT is also required for CGS-15943 mediated induction of apoptosis. As we have done previously to study the ARNT-dependent effects of SU5416, Le unomide, and Raloxifene [28], we used a cell culture model consisting of C4 cells, which express a mutant and transcriptionally inactive ARNT, and vT{2} cells, which stably re-express a WT ARNT protein. We rst con rmed that ARNT is required for transcriptional activation of the AhR gene target CYP1A1 by CGS-15943 using semi-quantitative PCR (Fig. 4A). Next, we performed real-time growth pro ling of C4 and vT{2} cells using the xCelligence assay. C4 cells were completely resistant to the effects of CGS-15943, appearing no different than cells treated with vehicle or 1 nM TCDD (Fig. 4B). Conversely, the normalized cell index values of vT{2} cells treated with CGS-15943 declined precipitously. Time lapse video microscopy using C4 and vT{2} cells in the parallel culture assay con rmed these ndings ( Figure S3 and Video S3). C4 cells were largely resistant to the effects of CGS-15943 in colony forming assays during continuous exposure to CGS-15943 (Fig. 4C). Nuclear fragmentation was also almost completely abrogated in C4 cells treated with CGS-15943 compared to vT{2} cells (data not shown). Likewise, only vT{2} cells treated with CGS-15943 exhibited increased staining with Annexin V (Fig. 4D) or activation of caspase 3/7 (Fig. 4E). Together, these results clearly showed that ARNT is required for the induction of apoptosis by CGS15943. Furthermore, the absence of any signi cant effect of CGS-15943 on C4 cells argue for an AhR-dependent growth-inhibitory pathway activated by CGS-15943.
Identi cation of Fas ligand as a putative target of CGS-15943 mediated AhR activation Upregulation of the pro-apoptotic molecule Bax has been described in literature as an AhR-target gene responsible for oocyte destruction mediated by the PAH 7,12-dimethylbenz[a]anthracene (DMBA) [29].
Thus, we asked if the potent induction of apoptosis by CGS-15943 led us to investigate the potential involvement of BAX. CGS-15943 did not increase the expression of BAX at the transcriptional level ( Figure  S4) or at the level of protein (data not shown). However, stable knockdown of BAX in Hepa1 cells partially blocked the induction of apoptosis by CGS-15943 ( Figure S4). Based on these data, we hypothesized that the CGS-15943 mediated apoptosis may involve the intrinsic cell death pathway.
We next used a focused RT-qPCR-gene array to identify mediators of CGS-15943 induced apoptosis downstream of AhR activation in Hepa1 cells. To narrow the potential list of gene targets, we performed the screen in parallel with cells treated with 1 nM TCDD, reasoning that any genes also upregulated by TCDD would be unlikely candidates for mediators of CGS-15943 induced apoptosis (Fig. 5A). Several putative gene targets were potently upregulated by CGS-15943 but not TCDD, the most notable of which was Fas ligand (FasL, CD95L). FasL is involved in immune cell regulation, inducing death of target cells expressing the Fas receptor (CD95). To ensure that induction of FasL could be a regulator of apoptosis in Hepa1 cells, presumably in an autocrine fashion, we con rmed that Hepa1 cells express Fas receptor (Fig. 5B). We next determined if induction of FasL by CGS-15943 correlated with AhR expression. FasL was signi cantly upregulated by CGS15943 in Hepa1 cells and vT{2} cells, but not the reciprocal AhR and ARNT null cell lines, respectively (Figs. 5C-D). Similarly, surface expression of FasL as determined by ow cytometry was increased to a greater extent in Hepa1 cells than Tao cells in response to CGS-15943 (Fig. 5E). To investigate whether FasL is a direct or indirect target of the AhR, we performed experiments in which Hepa1 cells were co-treated with CGS-15943 and the protein synthesis inhibitor cycloheximide (CHX). Addition of cycloheximide to Hepa1 cells treated with TCDD results in a super-induction of CYP1A1 [28], while blockade of protein synthesis would be expected to inhibit the transcription of any secondary gene targets. We found that CHX co-treatment with CGS-15943 led to a superinduction of CYP1A1, but blocked the increase in FasL mRNA as determined by qPCR ( Figure S5).

FasL is required for CGS-15943 induced apoptosis
Having shown that CGS-15943 induces apoptosis in Hepa1 cells in a strongly AhR and ARNT dependent manner, and identi ed FasL and BAX as putative downstream regulators, we next asked if FasL is required for CGS-15943 mediated apoptosis in Hepa1 cells. To address this possibility, we generated stable Hepa1 cells lines expressing shRNAs directed against FasL. We con rmed basal knockdown of FasL in these cells by qPCR, and also showed that the knockdown blocked induction of FasL in response to treatment with CGS-15943 (Fig. 5F). Knockdown of FasL signi cantly decreased caspase 3/7 activation compared with WT Hepa1 cells, indicating that FasL is required for CGS-15943 mediated cell death (Fig. 5H).
FasL-mediated apoptosis can occur through type I signaling, involving only the extrinsic cell death pathway, or type II signaling, which involves cross-talk with the intrinsic mitochondrial-dependent cell death pathway [30,31]. Our results with Bax knockdown ( Figure S4) suggested mitochondrial involvement. Thus, to determine if CGS-15943 induces cell death occurs through the intrinsic cell death pathway, or involves the mitochondria, we asked if CGS-15943 could increase the activity of caspases 8 and 9. Speci cally, activation of both caspases would suggest a mechanism by which FasL signaling activates caspase 8 to cleave the pro-apoptotic protein BID to induce mitochondrial-mediated apoptosis and subsequent activation of caspase 9 [30,31]. We observed increased activity of both caspases 8 and 9 in an AhR-dependent manner (Fig. 5H), suggesting that apoptosis induced by CGS-15943 involves both the intrinsic and extrinsic cell death pathways through a Type II signaling mechanism.
MRS1220 is a CGS-15943 derivative with unique properties Given the unique ability of CGS-15943 to induce AhR-dependent apoptosis in such a selective manner, we were next interested in identifying similar related molecules. We investigated a series of derivatives of CGS-15493, including those with additions of various moieties at the free amine group, for their ability to induce apoptosis in Hepa1 cells. From this effort, we identi ed one compound, MRS1220, henceforth named CGS B , with properties similar to CGS-15943 (Fig. 6A). Interestingly, CGS B was largely inactive in short-duration XRE-reporter assays (Fig. 6B), and did not appear to signi cantly increase expression of CYPA1A in semi-quantitative RT-PCR experiments (Fig. 6C). On the other hand, while not as potent as the parent molecule, CGS B retained the ability to induce apoptosis in Hepa1 cells according to xCelligence (Fig. 6D) and caspase 3/7 activation (Fig. 6E). Thus, while MRS1220 does not appear to exhibit increased potency with respect to the induction of apoptosis compared with CGS-15943, it may selectively regulate AhR transcription and is thus an attractive target for future studies.

CGS-15943 induces AhR-dependent apoptosis in human liver and breast cancer cells
Having characterized the effects of CGS-15943 in a mouse liver cancer cell line, we were next interested in expanding the repertoire of cell types in which CGS-15943 may have e cacy as an anti-cancer agent. We rst investigated whether CGS-15943 has activity against human HepG2 liver cancer cells. We found that CGS-15943 decreased HepG2 cell viability, and that transient knockdown of AhR with an siRNA rescued HepG2 cells from CGS-15943 induced cell death (Figs. 7A; Figure S6 and Video 4). CGS-15943 increased caspase 3/7 activation in HepG2 cells (Fig. 7B), and increased the expression of FasL mRNA (Fig. 7C). Thus, CGS-15943 induces apoptosis in a human liver cancer cell line, also consistent with our previous observations for the AhR-ligand raloxifene [28]. In subsequent evaluation of additional cell lines susceptible to CGS-15943, we identi ed the triple-negative (ER-/PR/HER2-) MDA-MB-468 cell line. To demonstrate that apoptosis induced by CGS-15943 in these cells required the AhR, we generated a stable cell line expressing a doxycycline inducible shRNA against the AhR and co-expressed with an RFPreporter (Fig. 7D). Partial knockdown of AhR in these cells (Fig. 7D) led to signi cantly decreased activation of Caspase 3/7 in response to treatment with CGS-15943 (Fig. 7E). In addition, the AhR knockdown in MDA-MB-468 cells signi cantly rescued the decreased cell viability seen with CGS-15943 (Fig. 7F).

Discussion
In the present study, we described the pro-apoptotic functions of a novel AhR ligand CGS-15943 in mouse (Figs. 2-4) and human liver cancer cells (Fig. 7), and MDA-MB-468 triple negative breast cancer cells (Fig. 7). In each of these cell lines, we found that the effects of CGS-15943 required expression of the AhR. We also identi ed a derivative of CGS-15943, MRS1220 (CGS B ), that induces apoptosis in an AhRdependent manner, but does not signi cantly induce the AhR target gene CYP1A1 or an XRE-reporter gene (Fig. 6).
Mutants of the AhR expressed in AhR-null cells support the ability of the AhR in inducing apoptosis. A constitutively active AhR mutant, termed as CA-AhR, lacking part of the ligand binding domain, induced apoptosis in Jurkat T lymphoma cells [36]. We previously reported that Raloxifene induces apoptosis in Hepa1 cells and MDA-MB-231 breast cancer cells [23]. The results of the current study indicate that CGS-15943 anti-cancer effects can be attributed completely to AhR and ARNT activation (Figs. 3-4). Indeed, C4 cells containing an inactive ARNT protein are completely resistant to the effects of CGS-15943. Thus, CGS-15943 represents a novel mediator of AhR-dependent apoptosis. AhR-selective ligands such as CGS-15943 may exert their anti-cancer effects by activating AhR tumor suppressive signaling. We observed no such effects with well-studied AhR ligand TCDD (Figs. 2D, 3G, and 4B) strongly supporting ligandselective effects of the AhR.
Pro-apoptotic protein Bax is not induced, but is required for AhR-mediated apoptosis upon treatment with CGS-15943. We identi ed for the rst time induction of FasL as a pathway that mediates AhR-regulated apoptosis in cancer cells. Interestingly, involvement of Fas-signaling and the AhR has been reported previously in various contexts [39,40]. We observed that induction of FasL by CGS-15943 was completely blocked by CHX co-treatment, as was cell death ( Figure S5). These results may suggest that fasL is an indirect gene target of the AhR, and that activation of an intermediate product is required for CGS-15943 mediated apoptosis upstream of fasL. Consistent with this possibility, only Fas, but not FasL, contain XRE sequences in their promoter [41]. Alternatively, these results may be explained by a fast-decaying factor induced by the CGS-15943 through the AhR that is required to stabilize increased abundance of FasL mRNA. Both of these possibilities will be investigated in the future.
CGS 15943 is a triazoloquinazoline non-xanthine adenosine antagonist with nanomolar a nity for adenosine receptors A1 and A2A [25]. Interestingly, both adenosine and CGS-21680 have been reported to induce apoptosis in Human HepG2 cells via mitochondrial mediated cytochrome c release and subsequent activation of caspases 3 and 9, an effect that could be abrogated by knockdown of the A2A adenosine receptor [42]. The results reported by Tamura et al. were obtained with adenosine receptor agonists, whereas CGS-15943 is an A2A antagonist, and because an A2A agonist blocked the effects of CGS-21860 [42], these results appear to be separate from those observed in our study. Conversely, another study showed that adenosine induces apoptosis in HepG2 cells, but that the effect is independent of mitochondrial-mediated cell death and is not blocked by A2A antagonists [43]. With respect to CGS-15943, the differences in concentrations associated with its adenosine receptor antagonist functions [25] vs its AhR-dependent effects (nanomolar and low-micromolar, respectively), as well as the signi cant degree to which the effects of CGS-15943 require both expression of AhR and Arnt (Figs. 2-4) as well as FasL (Fig. 5), argue against involvement of adenosine receptor activity in its biological effects.

Conclusions
In conclusion, we have identi ed CGS-15943 as an novel AhR-selective ligand with anti-cancer properties using two parallel and distinct screening strategies. CGS-15943 induced AhR-dependent apoptosis in mouse and human liver cancer cells. Discovery and development of CGS-15943 as an AhR-selective anticancer ligand to investigate AhR mediated anti-cancer effects has been an important area of research in our laboratory for the past several years. Both CGS-15943 and its analog MRS1220 represent AhRtargeted therapeutics for the treatment of liver and breast cancer. Our results argue for the use of phenotypic AhR-dependent screening strategies for identifying novel AhR ligands with anti-cancer actions and we have successfully implemented this approach to discover a novel AhR signaling with utility for cancer treatment.

Declarations
Author Contributions EFO contributed to study design, execution of experiments, data analysis, and wrote the manuscript. HSJ performed experiments and contributed to data analysis. DCK performed experiments, contributed to data analysis, and provided input on study design. NIK provided input on study design and provided material support. SKK designed and supervised the study, and helped prepare and revise the manuscript.