BI1071, a Novel Nur77 Modulator, Induces Apoptosis of Cancer Cells by Activating the Nur77-Bcl-2 Apoptotic Pathway

Nur77 (NR4A1; also known as NGFI-B and TR3) is perhaps the most potent apoptotic member of the nuclear receptor superfamily (1–8). The death effect of Nur77 was initially recognized during studying the apoptosis of immature thymocytes, T-cell hybridomas (9, 10). Later, we found that Nur77 mediates the death effect of the retinoid-related molecule AHPN (also called CD437) in cancer cells (11). Furthermore, we discovered a nongenomic action of Nur77, in which Nur77 migrates from the nucleus to the cytoplasm, where it targets mitochondria to trigger cytochrome c release and apoptosis in cancer cells (12–14). Further studies demonstrated in various cancer types that such a Nur77 mitochondrial apoptotic pathway is characterized by its interaction with Bcl-2 and the conversion of Bcl-2 from an antiapoptotic molecule to a pro-apoptotic molecule (6, 15). Given the pivotal role of Bcl-2 in regulating the apoptosis of cancer cells and in the resistance of cancer cells to a variety of radio- and chemotherapeutic agents, understanding how the Nur77-Bcl-2 apoptotic pathway is regulated and discovering its small-molecule modulators may offer new strategies to develop effective cancer therapeutics. However, small molecules that can activate the Nur77-Bcl-2 apoptotic pathway by binding to Nur77 to trigger Nur77 mitochondrial translocation and interaction with Bcl-2 have not been reported.

As an orphan nuclear receptor, Nur77 lacks a canonical ligand-binding pocket (LBP; refs. 16, 17), which excludes small molecules from binding to Nur77 to regulate Nur77 functions via the canonical LBP-binding mechanism. Recent advance has revealed the existence of alternate small-molecule binding regions on the surface of nuclear receptors, and compounds that bind to alternate sites other than LBP have been identified for some nuclear receptors (18, 19), including Nur77 (20–24). These developments inspire us to discover Nur77-binding compounds that can regulate the Nur77-Bcl-2 apoptotic pathway. Here, we report that a salt form of a 3,3′-diindolymethane (DIM) derivative (di(1H-indol-3-yl)(4-(trifluoromethyl)phenyl)methane; named BI1071 here) can bind to Nur77 to induce apoptosis of cancer cells through the Nur77-Bcl-2 apoptotic pathway. BI1071 binds to Nur77 at submicromolar concentration and induces apoptosis that is dependent on the expression of both Nur77 and Bcl-2. BI1071 also effectively inhibits the growth of tumor cells in animals. Moreover, BI1071 binding to Nur77 induces not only its mitochondrial targeting but also its interaction with Bcl-2. Our results therefore identify BI1071 as the first Nur77-binding small molecule that promotes the Nur77-Bcl-2 apoptotic pathway.

The following cell lines are used in our study. HCT116 colon cancer, MDA-MB-231, HS578T, BT549, MCF-7, and T47D breast cancer, breast epithelial cell line MCF-10A, HeLa ovarian cancer, mouse embryonic fibroblast (MEF) cells and HEK293T embryonic cells were cultured in DMEM, whereas ZR-75-1, HCC1937 breast cancer and SW480 colon cancer were cultured in RPIM-1640 medium containing 10% FBS. Human colonic epithelial cells (HCoEpiC) were cultured in colonic epithelial cell medium (CoEpiCM, Cat. #2951). Cell lines HCT116 (ATCC, CCL-247), SW480 (ATCC, CCL-228), and HEK293T (ATCC, CRL-11268) were obtained from the ATCC. Cell lines MCF-10A (SCSP-660), MDA-MB-231 (SCSP-5043), HeLa (TCHu187), HS578T (TCHu127), BT549 (TCHu 93), MCF-7 (SCSP-531), ZR-75-1 (TCHu126), T47D (TCHu 87), and HCC1937 (TCHu148) were obtained from Chinese Academy of Science Shanghai Cell Bank on December 09, 2016. Cell line HCoEpiC was obtained from ScienceCell (Cat. #2950) on October 05, 2018. MEF cells were isolated from embryonic day 13 wild-type (WT) and Nur77 knock out (KO) mice. The cells were grown in the cell incubator with 5% CO₂ at 37°C. Sub-confluent cells with exponential growth were used throughout the experiments. Cells plated onto cell culture dishes and kept in 10% FBS for 24 hours were treated with compounds or transfected with plasmids. Cell transfection was carried out by using Lipofectamin 2000 according to the manufacturer's instruction. The cells were tested by using Mycoplasma Hoechst Stain Assay kit (Beyotime, C0296) every 6 months. We added the Hoechest solution to stain the cells with 50% density at room temperature for 30 minutes and then used the confocal microscope to observe the cells. In the cells without Mycoplasma infection, only the blue fluorescence of the nucleus was observed. Filamentous blue fluorescence can be observed around the nucleus in Mycoplasma contaminated cell samples. The cells were prevented from Mycoplasma infection by using plasmocin (Invivogen, ant-mpt). No positive Mycoplasma tests was observed during the time of our experiments.

Plasmids pcmv-myc-Nur77, GFP-Nur77, GFP-Nur77/LBD, GST-Bcl-2, pcmv-myc-Bcl-2, Flag-cmv-Bcl-2 were constructed as described (12–14, 24). Plasmids pcmv-myc-Nur77/H372D, pcmv-myc-Nur77/H372A, pcmv-myc-Nur77/Y453L, pcmv-myc-Nur77/C566K were constructed by using the PCR and QuickChang mutagenesis kit.

Anti-Ki67 (Cat. ab15580) and anti-Hsp60 (Cat. ab46798) antibodies were purchased from Abcam (UK); anti–β-actin (Cat. 4970S), anti-Cleaved caspase-3 (Cat. 9661S), and anti-Nur77 (Cat. 3960S) antibodies were purchased from Cell Signal Technology; anti-Myc (9E10; Cat. Sc-40), anti-Nur77 (M-210; Cat. sc-5569), anti-PCNA (Santa Cruz Biotechnology sc-7907), anti–a-tubulin (Santa Cruz Biotechnology sc-8035), anti-Bcl-2 (Santa Cruz Biotechnology sc-783), anti-PARP (Santa Cruz Biotechnology sc-7150), and anti-GST (sc-138) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz Biotechnology); anti-Flag (Cat.F1804) antibody was purchased from Sigma; Mito-tracker deep red (Cat. M22426), JC-1 Probe (Cat. T3168) and mitoSOX Red Mitochondrial Superoxide Indicator (Cat. M36008) were purchased from Thermo Fisher.

Knocking out Nur77 and Bcl-2 from HeLa cells used the CRISPR/Cas9 system. gRNA targeting sequence of Nur77 (5′-ACCTTCATGGACGGCTACAC-3′) and Bcl-2 (5′-GAGAACAGGGTACGATAACC-3′) was cloned into gRNA cloning vector Px330 (Addgene, 71707) and confirmed by sequencing. The accession numbers of Nur77 and Bcl-2 are NM_-001202233 and NM_-000633.2 respectively. To screen for cells lacking Nur77 or Bcl-2, HeLa cells were transfected with control vector and gRNA expression vectors, followed by G418 selection (0.5 mg/mL). Single colonies were subjected to Western blotting using anti-Nur77 and anti–Bcl-2 antibody to select knockout cells.

Cell viability was analyzed by using colorimetric 3-(4,5-dimethylthiazol-dimethylthiazol-2-yl)-2,5-diphenyletetrazolium Bromide (MTT) assay as described previously (12–14, 24).

HEK293T cells were co-transfected with pG5 Luciferase reporter together with the plasmid encoding RXRα-LBD fused with the Gal4 DNA-binding domain and other expression plasmids as described previously (18, 25). After transfection, cells were treated with DMSO or BI1071, and assayed by using the Dual-Luciferase Reporter Assay System (Promega). Transfection efficiency was normalized to Renilla luciferase activity.

For cellular fractionation (12–14, 24), cells were lysed in cold buffer A (10 mmol/L HEPES-KOH (pH 7.9), 1.5 mmol/L MgCl₂, 10 mmol/L KCl, 0.5 mmol/L dithiothreitol) with a cocktail of proteinase inhibitors on ice for 10 minutes as described previously. Cytoplasmic fraction was collected by centrifuging at 6,000 rpm for 10 minutes. Pellets containing nuclei were resuspended in cold high-salt buffer C (20 mmol/L HEPES-KOH (pH 7.9), 25% glycerol, 420 mmol/L NaCl, 1.5 mmol/L MgCl₂, 0.2 mmol/L EDTA, 0.5 mmol/L dithiothreitol) with a cocktail of proteinase inhibitors on ice for 30 minutes.

GST or GST-Bcl-2 fusion protein (0.5 mg) was immobilized on glutathione-Sepharose beads and incubated with purified His-Nur77-LBD (0.2 mg) in the presence of different concentration of BI1071 as described previously (12–14, 24). Bound Nur77-LBD was analyzed by Western blotting.

Western blotting and co-immunoprecipitation (co-IP) were performed as described (12–14, 24).

Knocking out Nur77 and Bcl-2 from HeLa cells used the CRISPR/Cas9 system. gRNA targeting sequence of Nur77 (5′-ACCTTCATGGACGGCTACAC-3′) and Bcl-2 (5′-GAGAACAGGGTACGATAACC-3′) was cloned into gRNA cloning vector Px330 (Addgene, 71707) and confirmed by sequencing. To screen for cells lacking Nur77 or Bcl-2, HeLa cells were transfected with control vector and gRNA expression vectors, followed by G418 selection (0.5 mg/mL). Single colonies were subjected to Western blotting using anti-Nur77 and anti–Bcl-2 antibody to select knockout cells.

Cells were placed on 6-well plates with a density of 1 × 10⁶ per well. After 24 hours, the cells treated with different concentration of BI1071 for 6 hours, and then the suspended and the adherent cells were collected, stained with Annexin V-FITC for 10 minutes and with propidium iodide for 5 minutes, and analyzed immediately by cytoFLEX Flow Cytometry System (Beckman-Coulter) using FITC and PC5.5.

The determination of ΔΨm and ROS was performed as previously described elsewhere (26). JC-1 probe was used to measure mitochondrial depolarization in cells. Cells were first treated with different concentration of BI1071 for 6 hours and then followed with the addition of JC-1 staining solution (5 μg/mL) for 20 minutes at 37°C. After washing with PBS twice, mitochondrial membrane potentials were analyzed immediately by cytoFLEX Flow Cytometry System using FITC and PE. Mitochondrial depolarization was measured by a change in the ratio of green/red fluorescence intensity. ROS was monitored with the mitoSOX Red Mitochondrial Superoxide Indicator and analyzed by cytoFLEX Flow Cytometry System using PE.

Cells were fixed in 4% paraformaldehyde. For mitochondrial staining, cells were incubated with anti-Hsp60 goat immunoglobulin G (IgG; Santa Cruz Biotechnology), followed by anti-goat IgG conjugated with Cy3. The nuclei were visualized by DAPI staining. Fluorescent images were collected and analyzed by using a fluorescence microscopy or MRC-1024 MP laser-scanning confocal microscope (Bio-Rad).

The protocols for animal studies were approved by the Animal Care and Use Committee of Xiamen University, and all mice were handled in accordance with the “Guide for the Care and Use of Laboratory Animals” and the “Principles for the Utilization and Care of Vertebrate Animals.” For MMTV–PyMT mice breast cancer model, female MMTV-PyMT mice of 12-weeks-old were randomly divided into two groups (n = 7 each), treated with a daily oral dose of BI1071 (5 mg/kg) for 18 days. Standard histopathological analysis of tumor tissue was performed. BI1071 was dissolved in DMSO and diluted with normal saline containing 5.0% (V/V) Tween-80 to a final concentration 0.5 mg/mL. Normal saline with DMSO and 5.0% Tween-80 was used as the vehicle control. For xenograft nude mouse study, male BALB/c nude mice (6-weeks-old) were subcutaneously injected with log growth-phase of SW620 cells (1 × 10⁶ cells in 0.1 mL PBS). Mice were treated orally after 7 days of transplantation with BI1071 once a day. Body weight and tumor size were measured every 3 days. Tumors were measured and weighted. Tissues isolated from the nude mice were fixed with 4% paraformaldehyde. TdT-mediated dUTP nick end labeling assay was performed according to the manufacturer's instructions (In situ Cell Death Detection Kit; Roche).

Four-μm-thick sections were deparaffinized and rehydrated using xylene and a graded series of ethanol (100%, 95%, 85%, 75%, 50%), followed by washing in PBS. Antigen retrieval was performed in 10 mmol/L sodium citrate buffer (pH 6.0), which was microwaved at 100°C for 20 minutes. After rinsed twice in PBS, sections were blocked at room temperature for 1 hour by using 10% normal goat serum, followed by incubation with anti-ki67, anti-cleaved caspase 3 overnight at 4°C. Colors were developed with a DAB horseradish peroxidase color development kit.

Schrodinger's (www.schrodinger.com) Glide (27), a grid-based docking program was used for the docking study of BI1071 to the protein. The crystal structure of Nur77-LBD in complex with a cytosporone B analog (Protein Data Bank code 3V3Q) was used. Docking was performed with the implemented standard routine in Glide. The Glide GScore was used as docking score to rank the docking results. Poses were further visually investigated to check for their interactions with the protein in the docking site. Schrödinger's Maestro was used as the primary graphical user interfaces for the visualization of the crystal structure and docking results.

The binding kinetics between Nur77-LBD and compounds were performed on a BIAcore T200 instrument (GE Healthcare) at 25°C (24). Nur77-LBD were diluted to 0.05 mg/mL in 50 mmol/L NaOAc (pH 5.0) and immobilized on a CM5 sensor chip (GE Healthcare) by amine coupling at densities approximately 10,000 RU according to the manufacturer's instructions. Gradient concentrations of compounds were injected into the flow cells in running buffer (PBS, 0.4% DMSO) at a flow rate of 30 μL/min for 150 seconds of association phase followed by a 420 seconds dissociation phase and a 30 seconds regeneration phase (10 mmol/L Glycin-HCl, pH 2.5). The data were analyzed using BIAcore T200 Evaluation Software 2.0 and referenced for blank injections and reference Surface. The dissociation constant (K_d) was fitted to the standard 1:1 interaction model and calculated using the global fitting of the kinetic data from gradient concentrations.

Data were expressed as mean ± SD. Each assay was repeated in triplicate in three independent experiments. The statistical significance of the differences among the means of several groups was determined using the Student t test.

In our effort to identify small molecules that modulate the Nur77/Bcl-2 apoptotic pathway, we evaluated an in-house compound library, which includes di(1H-indol-3-yl)(4-(trifluoromethyl)phenyl)methane (DIM-C-pPhCF₃, Fig. 1A; ref. 28). We surprisingly observed that the freshly prepared DIM-C-pPhCF₃ solution was not as active as the aged solution in apoptosis induction. In addition, the freshly made DIM-C-pPhCF₃ solution is colorless, however, it turns reddish when it is aged or exposed to air at room temperature. Thus, we presumed that the compound underwent oxidization and the oxidized DIM-C-pPhCF₃ was more active than DIM-C-pPhCF₃. To test our hypothesis, DIM-C-pPhCF₃ was oxidized in the presence of methanesulfonic acid, and the products were subsequently purified to obtain the oxidized DIM-C-pPhCF₃: di((1H-indol-3-yl)(4-trifluoromethylphenyl)methylium methanesulfonate (DIM-C-pPhCF₃⁺MeSO₃⁻; BI1071, Fig. 1A; Supplementary Methods for synthesis and purification). BI1071 was then tested in comparison with DIM-C-pPhCF₃ for growth inhibition and apoptosis induction. Figure 1B showed that BI1071 inhibited the growth of HCT116 colon cancer cells with an IC₅₀ of 0.06 μmol/L, which is about 25-fold more active than DIM-C-pPhCF₃ (IC₅₀ = 1.5 μmol/L). Treatment of MDA-MB-231 cells with 0.5 μmol/L BI1071 for 6 hours effectively induced PARP cleavage, an indication of apoptosis, while DIM-C-pPhCF₃ had no effect under the same condition (Supplementary Fig. S1A). Dose-dependent study demonstrated that BI1071 could induce PARP cleavage at submicromolar concentrations in HCT116 cells (Fig. 1C) and other cancer cell lines (Supplementary Fig. S1B).

Interestingly, BI1071 was effective in various breast cancer cell lines analyzed regardless of its hormone dependency (Supplementary Fig. S2). Furthermore, BI1071 did not display apoptotic effect in the non-transformed mammary and normal colon cells (Supplementary Fig. S3A and S3B), indicating that BI1071 selectively induces apoptosis in cancer cells. The apoptotic effect of BI1071 was also confirmed by its induction of extensive nuclear condensation and fragmentation revealed by DAPI staining in cells treated with 0.5 μmol/L BI1071 for 6 hours (Fig. 1D; Supplementary Fig. S1C) and confirmed as well by PARP and caspase 3 cleavage assays (Fig. 1E). The effect of BI1071 on cell death was further assessed using flow cytometry-based Annexin V/ Propidium iodide (PI) apoptosis assay. Dose-dependent study showed that about 31.63% of MDA-MB-231 cells were apoptotic when treated with 1 μmol/L of BI1071 for 6 hours, whereas only 1.31% of cells were apoptotic in vehicle control cells (Fig. 1F).

Loss of mitochondrial membrane potential (Δψm) represents one of the hallmarks of apoptosis. To assess whether the BI1071-induced apoptosis was related to the intrinsic mitochondrial pathway, we used JC-1, the mitochondrial-specific dye, to monitor the changes of mitochondrial membrane potential (29). MDA-MB-231 breast cancer cells treated with BI1071 were stained with JC-1. JC-1 dye accumulation in mitochondria is dependent of mitochondrial membrane potential, accompanied by a shift of JC-1 fluorescence emission from green to red. In comparison with healthy cells, apoptotic cells display an increase in the green/red fluorescence intensity ratio. Analysis of both red and green fluorescence emissions by flow cytometry revealed a dose-dependent BI1071 induction of mitochondrial membrane dysfunction. After treatment with 1 μmol/L of BI1071 for 6 hours, the green to red ratio increased from 100% to 345% (Fig. 1G). Mitochondrial dysfunction was also revealed by marked increase in intracellular mitochondrial reactive oxygen species (mito-ROS) in MDA-MB-231 cells exposed to BI1071 in a dose-dependent manner (Fig. 1H). Collectively, these data suggested that BI1071 induced mitochondrion-related apoptosis in cancer cells.

To assess the apoptotic effect of BI1071 in animals, SW620 colon cancer cells were inoculated subcutaneously in the right and left hind-side flank of nude mice. Administration of tumor-bearing nude mice with BI1071 inhibited the growth of SW620 xenograft tumor in a dose- and time-dependent manner (Fig. 2A–C). TUNEL assay revealed extensive apoptosis in BI1071-treated tumor specimens as compared with control tumor (Fig. 2D). MMTV-PyMT-transgenic mouse model of breast cancer was also used to evaluate the anticancer effect of BI1071. Administration of the MMTV-PyMT mice with BI1071 (5 mg/kg) potently inhibited the growth of PyMT mammary tumor (Fig. 2E and F). Western blotting of tumor tissues prepared from treated and non-treated mice revealed that the expression levels of two proliferation markers, PCNA and Ki67, were markedly reduced by BI1071 (Fig. 2G). Immunostaining also showed a reduced expression of Ki67 and enhanced expression of cleaved caspase 3 in tumor tissue specimens prepared from mice treated with BI1071 (Fig. 2H). There was not significant difference in the body weight (without tumor weight) between the control mice and the BI1071-treated mice in both animal models (Supplementary Fig. S4A and S4B). These data demonstrated that BI1071 potently inhibited the growth of tumor cells in animals through its induction of apoptosis.

We next determined whether BI1071-induced apoptosis was Nur77-dependent by examining its apoptotic effect in mouse embryonic fibroblast (MEF) and MEF lacking Nur77 (Nur77^−/−MEF). BI1071 dose-dependently inhibited the growth of MEFs, but such an inhibitory effect was significantly diminished in Nur77^−/−MEFs (Fig. 3A). Induction of PARP cleavage in MEFs by BI1071 was also attenuated in Nur77^−/−MEFs (Fig. 3B). The death effect of BI1071 was also evaluated in Nur77 genome knockout HeLa cells generated by CRISPR/Cas9 technology. Induction of PARP cleavage and caspase 3 activation by BI1071 were strongly suppressed in Nur77^−/−HeLa cells (Fig. 3C). Annexin V/PI staining revealed a reduced apoptotic effect of BI1071 in Nur77^−/−HeLa cells than in the parental HeLa cells (from 35.55% to 3.25%; Fig. 3D). Furthermore, unlike the parental HeLa cells, Nur77^−/−HeLa cells did not display BI1071-induced mitochondrial membrane potential loss measured by JC-1 staining (Fig. 3E) or BI1071-induced release of mitochondrial ROS (Fig. 3F). To further address the role of Nur77, we transfected the ligand-binding domain (LBD) of Nur77, Nur77-LBD, into HEK293T cells and asked whether the overexpression of Nur77-LBD could influence the effect of BI1071. Indeed, transfection of Nur77-LBD enhanced the killing effect of BI1071, with 36% of the transfected HEK293T cells undergoing apoptosis, while 4.5% of the non-transfected cells were apoptotic (Supplementary Fig. S5). Together, these results demonstrated that BI1071 targets Nur77 to induce cancer cell apoptosis.

Our observation that BI1071 induced mitochondria-dependent apoptosis prompted us to determine whether BI1071 exerted its Nur77-dependent apoptosis by promoting Nur77 mitochondrial targeting. Immunostaining showed that Nur77 was mainly localized in the nucleus of HCT116 cells. However, it was predominantly cytoplasmic when cells were treated with 0.5 μmol/L of BI1071 for 2 hours (Fig. 3G). To confirm the effect of BI1071 on Nur77 cytoplasmic localization, HEK293T cells were transfected with GFP-Nur77 and subsequently treated with 0.5 μmol/L BI1071. Transfected GFP-Nur77 resided in the nucleus, however it was diffusely distributed in both the cytoplasm and nucleus upon BI1071 treatment (Fig. 3H). Cells transfected with GFP-Nur77-LBD also responded well to BI1071. Only for cells treated with BI1071, GFP-Nur77-LBD colocalized extensively with the mitochondria-specific Hsp60 protein revealed by confocal microscopy (Fig. 3I) and co-accumulated with the Hsp60 protein in the heavy membrane fraction shown by cellular fractionation (Fig. 3J). The effect of BI1071 on inducing Nur77 mitochondrial targeting was further illustrated by cellular fractionation experiments showing that a significant amount of transfected Myc-Nur77 accumulated in the mitochondria-enriched heavy membrane (HM) fraction when cells were treated with BI1071 (Fig. 3K). Taken together, these data demonstrated that BI1071 exerted its Nur77-dependent apoptotic effect by promoting Nur77 mitochondrial targeting.

Although Nur77 lacks a LBP and no endogenous ligands have yet been identified (16, 17), recent crystallographic studies have identified several regions on the surface of the Nur77 protein as small-molecule binding regions (20, 22, 30). We therefore determined whether BI1071 binds directly to Nur77 to induce its mitochondrial targeting and apoptosis. Surface plasmon resonance (SPR) analyses revealed that DIM-C-pPhCF₃ bound to Nur77-LBD with a K_d of 3.0 μmol/L (Fig. 4A) and that BI1071 bound to Nur77-LBD protein with a K_d of 0.17 μmol/L (Fig. 4B), which demonstrated that BI1071 binding to Nur77-LBD was 18-fold stronger than DIM-C-pPhCF₃. We also evaluated the effect of BI1071 on the transcriptional activity of Nur77/RXRα-LBD heterodimer. Co-transfection of pBind-RXRα-LBD and Nur77 strongly activate the reporter transcriptional activity when cells were treated with 9-cis-RA, a RXRα ligand (Fig. 4C). BI1071 further dose-dependently induced the reporter activity (Fig. 4C), likely due to its binding to Nur77. To exclude the possibility that BI1071 acted on RXRα, Glu453 and Glu456 in the activation function 2 (AF2) region of RXRα (31) were substituted with Ala and the resulting mutant, RXRα-LBD/E453,6A, was used to repeat the reporter assay. As expected, 9-cis-RA failed to induce the reporter activity in cells transfected with Nur77 and pBind-RXRα-LBD-E453,6A. However, BI1071 could still activate the reporter gene transcription (Fig. 4C), demonstrating that BI1071 induced reporter gene transcription through Nur77 binding but not RXRα. We also excluded the possibility of BI1071 binding to other nuclear receptors using reporter assays (Supplementary Fig. S6A). We next employed molecular docking approach to study how BI1071 bound to Nur77-LBD. Our docking studies showed that BI1071 docked better to a binding region formed by helices H1, H5, H7 and H8, and loops H1-H2, H5-B1 and H7-H8. The docked mode also suggested that the indole ring of BI1071 made key interaction with the side chains of H372 and Y453 located in H1 and H5, respectively (Fig. 4D). For comparison, we also docked DIM-C-pPhCF₃ to the same region. As shown in Fig. 4D, BI1071 fit better to the binding groove with the bis-indolyl rings embedded deeper in the groove. The bis-indolyl rings of BI1071 also was positioned to form π-π interaction with Y453 and to make stronger interaction with H372. To test this binding mode, H372 was mutated into either Ala or Asp. When tested in the Gal4 reporter assay for its response to BI1071, Nur77/H372D (Fig. 4E) or Nur77/H372A (Supplementary Fig. S6B) could not induce the reporter gene transcription in response to BI1071 treatment, revealing a critical role of H372 in BI1071 binding. Nur77/H372A also failed to respond to BI1071 to accumulate in the heavy membrane fraction in the cellular fractionation assay (Fig. 4F). To further characterize the binding of BI1071 and its apoptotic effect, we made 2 more mutants: mutant Nur77/Y453L, another key residue suggested by the docking studies, and mutant Nur77/C566K, a residue located in another reported ligand-binding region (20–24). BI1071 failed to induce PARP cleavage in cells transfected with Nur77/H372D or Nur77/Y453L, whereas it strongly induced PARP cleavage in cells transfected with Nur77 or Nur77/C566K (Fig. 4G). Similarly, in cells transfected with Nur77/H372D or Nur77/Y453L, the effect of BI1071 on inducing Mito-ROS generation (Fig. 4H), on apoptosis analyzed using dual staining with fluorescent Annexin V and PI (Fig. 4I), or on loss of mitochondrial membrane potential (Fig. 4J) was much attenuated when compared with cells transfected with Nur77 or Nur77/C566K. Taken together, these data demonstrated that BI1071 exerted its Nur77-dependent apoptotic effect by a direct Nur77-binding mechanism.

We previously showed that the Nur77 mitochondria-dependent apoptotic pathway involved Nur77 interaction with Bcl-2 (14). We therefore asked whether Bcl-2 plays a role in BI1071-induced apoptosis. Thus, the apoptotic effect of BI1071 was evaluated in MEFs and MEFs lacking Bcl-2 (Bcl-2^−/−MEF). BI1071 at 0.5 μmol/L effectively induced PARP cleavage in MEFs, whereas it had no apparent effect on PARP cleavage in Bcl-2^−/−MEFs (Fig. 5A). This was confirmed by DAPI staining showing that Bcl-2^−/−MEF cells were much more resistant than MEFs to the apoptotic effect of BI1071 (Fig. 5B). In addition, the impaired effect of BI1071 in Bcl-2^−/−MEFs could be rescued by re-expression of Bcl-2 (Supplementary Fig. S7). In response to 0.5 μmol/L of BI1071, 40% of MEFs displayed chromatin condensation and nuclear fragmentation, whereas only 14% of Bcl-2^−/−MEF cells exhibited similar apoptotic features. We also used the CRISPR/Cas9 technology to generate Bcl-2 knockout HeLa cells and showed that the effect of BI1071 on inducing PARP cleavage was almost completely suppressed in Bcl-2^−/−HeLa cells (Fig. 5C). The role of Bcl-2 in mediating the death effect of BI1071 was also illustrated by Annexin V/PI staining showing a reduced apoptotic effect of BI1071 in Bcl-2^−/−HeLa cells compared with its effect in the parental HeLa cells (from 24.59% to 7.95%), and in Bcl-2^−/−MEF cells compared with the parental MEF cells (from 76.65% to 10.27%; Fig. 5D). Induction of the mitochondrial membrane potential loss by BI1071 was also suppressed in Bcl-2^−/−MEFs and Bcl-2^−/−HeLa cells (Fig. 5E). Furthermore, BI1071-induced release of mitochondrial ROS was compromised by loss of Bcl-2 (Fig. 5F). These results revealed a crucial role of Bcl-2 in mediating the Nur77-dependent apoptotic effect of B1071, demonstrating that the compound acts through the Nur77-Bcl-2 apoptotic pathway.

In this study, we have showed that BI1071 can bind to Nur77 to induce its migration from the nucleus to mitochondria, where it interacts with Bcl-2 and triggers Bcl-2–dependent apoptosis. However, if the binding of BI1071 to Nur77 promotes the interaction between Nur77 and Bcl-2 is not clear. Therefore, we investigated whether BI1071 binding to Nur77 enhanced the Nur77 interaction with Bcl-2. In vitro GST-pull down assays showed that Nur77-LBD was pulled down by GST-Bcl-2 in a BI1071 concentration-dependent manner (Fig. 6A). Cell-based Co-IP showed that Nur77 (Fig. 6B) or Nur77-LBD (Fig. 6C) transfected in HEK293T cells interacted with Bcl-2 when cells were treated with BI1071. Endogenous Nur77 could be specifically immunoprecipitated together with endogenous Bcl-2 by anti-Bcl-2 antibody only when cells were treated with BI1071 (Fig. 6D). Moreover, confocal microscopy analysis revealed that BI1071 promoted extensive mitochondrial colocalization of transfected GFP-Nur77 (Fig. 6E; Supplementary Fig. S8A) or GFP-Nur77-LBD (Fig. 6F; Supplementary Fig. S8B) with Bcl-2 in cells. To further address the role of BI1071 on inducing Nur77 interaction with Bcl-2, the aforementioned Nur77 mutants were analyzed. Fig. 6G showed that Nur77/C566K like the wild-type Nur77 interacted strongly with Bcl-2 in a BI1071-dependent manner. In contrast, Nur77/H372D and Nur77/Y453L failed to interact with Bcl-2 in the presence of BI1071. The importance of the binding of BI1071 on inducing Nur77 interaction with Bcl-2 was also illustrated by immunostaining showing extensive colocalization of transfected Flag Bcl-2 with the wild-type Nur77 LBD, but not with Nur77 LBD H372A (Supplementary Fig. S8C). Thus, binding of BI1071 to Nur77 promotes Nur77 interaction with Bcl-2 and mitochondrial localization.

BI1071 is a salt form of DIM-C-pPhCF₃ (Fig. 1A) previously reported to induce Nur77-dependent apoptosis (30, 32). However, relative high concentrations (around 10 μmol/L) of DIM-C-pPhCF₃ are required for its induction of apoptosis and activation of Nur77. To our surprise, we observed that aged DIM-C-pPhCF_3. was generally more active than the freshly prepared one in apoptosis induction. This led to our synthesis of the oxidized product of DIM-C-pPhCF₃, methanesulfonate salt of DIM-C-pPhCF₃ (BI1071). Our evaluation of BI1071 revealed its superior death effect in cancer cells. BI1071 was also very effective in other cancer cell lines and in animal tumor models. The apoptotic effect of several DIM derivatives has been shown to be Nur77-dependent and various pathways were proposed to account for their apoptotic effect (2). However, the mechanism by which Nur77 mediates their death effect remains elusive, which is conceivably due to the activation of multiple pathways by the high compound concentration used in the studies. For instance, both transcriptional agonist such as DIM-C-pPhOCH₃ and transcriptional antagonist such as DIM-C-pPhOH were shown to induce Nur77-dependent apoptosis (28). Our finding that oxidization of DIM-C-pPhCF₃ could augment its death effect offered an opportunity to delineate the mechanism by which Nur77 mediates the apoptotic effect of DIM-related small molecules. To this end, our studies showed that the potent apoptotic effect of BI1071 was Nur-77 dependent (Fig. 3) and was a result of its induction of Nur77 mitochondrial targeting via a direct Nur77-binding mechanism. Furthermore, our results revealed that the death effect of BI1071 was also dependent of Bcl-2 expression and that BI1071 could induce Nur77 interaction with Bcl-2 leading to Nur77 colocalization with Bcl-2 at mitochondria and apoptosis.

Binding studies showed that BI1071 could bind to Nur77 better than DIM-C-pPhCF₃. This is likely due to the difference in the structural conformations between the oxidized and the unoxidized forms of DIM-C-pPhCF3, perhaps resulted from different atomic orbital hybridization of the central C atom. In the oxidized form the central C is sp2-hybridized, positively charged and bonded to 3 atoms with a co-planner arrangement, whereas in the unoxidized form, C is sp3-hybridized and bonded to 4 atoms with a tetrahedral arrangement. Differences in structural conformation and charge distributions can affect how molecules bind to proteins. Our docking results suggested that BI1071 could interact more strongly with Nur77-LBD than DIM-C-pPhCF₃ due to the different conformations adopted by the compounds. Mutagenesis studies confirmed that H372 and Y453 were 2 key residues involving in the binding of BI1071 as suggested by the docking studies. Previously, we located the critical region in Nur77 responsible for its interaction with Bcl-2 and identified a peptide NuBCP-9 as Bcl-2-converting peptide, capable of inducing apoptosis of cancer cells in vitro and in animals (12). NuBCP-9 is located at the C-terminal portion of H7. Interestingly, our docking studies suggested that H7 was part of the BI1071-binding region and BI1071 could potentially interact directly with amino acids D499 and A450, structurally flanking the residues from which NuBCP-9 is derived. Therefore, it is conceivable that the BI1071-bound Nur77 offers a more suitable Bcl-2–interacting interface that promotes the formation of Nur77/Bcl-2 complex, and thus augments the biological effect of BI1071.

A critical step in the Nur77 mitochondrial apoptotic pathway is the interaction of Nur77 with Bcl-2, which induces a conformation change in Bcl-2 and converts Bcl-2 from a pro-survival to a killer (12, 14). Members of the Bcl-2 family are critical regulators of apoptosis. As the funding member of the Bcl-2 family, Bcl-2 acts as a survival molecule to protect cells from programmed cell death. Bcl-2 overexpression is often observed in cancer cells and is associated with cancer treatment resistance and poor prognosis (33–36). Thus, Bcl-2 has been an important drug target (37, 38). Two strategies are commonly used to develop therapeutic agents targeting Bcl-2: The first relies on making use of antisense oligonucleotides to block Bcl-2 expression and the second relies on designing and optimizing BH3 small-molecule or peptide mimetics that bind the Bcl-2 BH3-binding cleft, antagonizing its antiapoptotic activity (37, 38). Like Bcl-2, Nur77 is overexpressed in a variety of cancer cells and plays a dual role in mediating apoptosis and survival of cancer cells (39, 40). Although the growth promoting effect of Nur77 appears to be dependent on its nuclear action, the death effect of Nur77 involves its translocation from nucleus to cytoplasm (41–43). Our results showed that cancer cells are sensitive to the treatment of BI1071 as compared with normal cells, consisting with the fact that the level of Nur77 is elevated in cancer cells. Thus, targeting Nur77 by BI1071 will have less effect on normal cells, and therefore likely offer a high therapeutic index. The ability of Nur77 to interact with Bcl-2 to not only suppress its antiapoptotic function but also convert Bcl-2 into a pro-apoptotic molecule (12, 14) provides a promising strategy to target both Nur77 and Bcl-2 for cancer therapy. Agents that can bind directly to Nur77 to promote Nur77 translocation and interaction with Bcl-2 are unique in that they can simultaneously target both Nur77 and Bcl-2. The reported Nur77-derived peptide with 9 amino acids (NuBCP-9) and its enantiomer as Bcl-2–converting peptides has demonstrated such a potential (12, 44, 45). In this regard, our identification of small molecules that can directly bind Nur77 to activate the Nur77-Bcl-2 apoptotic pathway is significant and BI1071 represents the first lead of this class of small molecules, which warrants further evaluation.

No potential conflicts of interest were disclosed.

Conception and design: X. Chen, X. Cao, G. Alitongbieke, Z. Xia, Y. Su, X.-K. Zhang

Development of methodology: X. Chen, X. Cao, X. Tu, G. Alitongbieke, M. Yin, D. Xu, S. Guo, Y. Su

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Chen, X. Cao, X. Tu, G. Alitongbieke, X. Li, D. Xu, S. Guo, Z. Li, L. Chen, X. Zhang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Chen, X. Cao, X. Tu, G. Alitongbieke, Z. Xia, X. Li, Z. Chen, D. Xu, S. Guo, Z. Li, H. Zhou, Y. Su, X.-K. Zhang

Writing, review, and/or revision of the manuscript: X. Chen, X. Cao, M. Gao, Y. Su, X.-K. Zhang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Chen, X. Cao, X. Tu, G. Alitongbieke, Z. Xia, X. Li, M. Yin, D. Xu, S. Guo, Z. Li, L. Chen, D. Xu, J. Liu, Z. Zeng, H. Zhou, Y. Su

Study supervision: X. Cao, X. Li, H. Zhou, Y. Su, X.-K. Zhang

The authors thank Dr. Marcia Dawson for her contributions in chemistry (including conception and design) to this work, and to her memory this article is dedicated. We also thank Dr. Lin Li for her critical reading of this article. This study was supported in part by grants from the Natural Science Foundation of China (U1405229, 81672749, 31271453, 31471318; to X.-k. Zhang), Regional Demonstration of Marine Economy Innovative Development Project (16PYY007SF1; to X.-k. Zhang), the Fujian Provincial Science and Technology Department (2017YZ0002-1; to X.-k. Zhang), and the National Institutes of Health (R01 CA198982; to X.-k. Zhang).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.