Repression of Noxa by Bmi1 contributes to deguelin‐induced apoptosis in non‐small cell lung cancer cells

Abstract Deguelin, a natural rotenoid isolated from several plants, has been reported to exert anti‐tumour effects in various cancers. However, the molecular mechanism of this regulation remains to be fully elucidated. Here, we found that deguelin inhibited the growth of non‐small cell lung cancer (NSCLC) cells both in vitro and in vivo by downregulation of Bmi1 expression. Our data showed that Bmi1 is highly expressed in human NSCLC tissues and cell lines. Knockdown of Bmi1 significantly suppressed NSCLC cell proliferation and colony formation. Deguelin treatment attenuated the binding activity of Bmi1 to the Noxa promoter, thus resulting in Noxa transcription and apoptosis activation. Knockdown of Bmi1 promoted Noxa expression and enhanced deguelin‐induced apoptosis, whereas overexpression of Bmi1 down‐regulated Noxa protein level and deguelin‐induced apoptosis. Overall, our study demonstrated a novel apoptotic mechanism for deguelin to exert its anti‐tumour activity in NSCLC cells.

B cell-specific Moloney murine leukaemia virus integration site 1 (Bmi1) is thought to be a critical component of the polycomb repressive complex 1 (PRC1), including a number of proteins (Bmi1, Ring1, HPH1, HPC1, and HPC2). Bmi1 is an epigenetic regulator for the stable maintenance of gene repression 3 that involved in the regulation of development, stem cell self-renewal, cell cycle, senescence, and tumourigenesis. [4][5][6][7] Studies have found that Bmi1 regulates cell senescence and proliferation via transcriptional silencing of tumour suppressor genes, such as p16 INK4a , p19 ARF , and p21 Cip1 , and results in tumourigenesis. [6][7][8] In addition, Bmi1 is known to inhibit other tumour suppressor genes, including PTEN, 9 BCL2L11, 10,11 and WWOX. 12 A recent study has revealed that Bmi1 regulates memory CD4 T cell survival and function through the direct repression of Noxa gene. 13 Teshima et al 11 report that Bmi1 directly regulates pro-apoptotic genes such as BCL2L11/Bim and PMAIP1/Noxa, leading to enhance anti-apoptotic potential of mantle cell lymphoma.
It has been reported that Bmi1 functions as an oncogene. Aberrant overexpression of Bmi1 has been reported in multiple tumour types, including breast cancer, 14 colon carcinoma, 15 melanoma, 16 and hepatocellular carcinoma. 17 Although upregulation of Bmi1 in human NSCLC has been reported, 18 the role of Bmi1 in the pathogenesis of NSCLC and its exact target genes have not been extensively studied.
Accumulation of evidence has demonstrated that Bmi1 plays a critical role in cancer cell invasion, metastasis, and chemoresistance. 19 Overexpression of Bmi1 correlates with cancer development, progression, and therapy failure. 17,20,21 Moreover, experimental decrease of Bmi1 protein levels results in cancer cells apoptosis and/ or senescence, increasing susceptibility to cytotoxic agents and radiation therapy. 22,23 These data suggest that reducing Bmi1 protein level may have a beneficial effect in multiple types of cancer including NSCLC.
Noxa, a pro-apoptotic BH3-only member of the Bcl-2 family of proteins, 24 albeit showing weak pro-apoptotic potential on its own, appears to be crucial in fine-tuning cell death decisions by targeting the pro-survival molecule Mcl-1 for proteasomal degradation. 25 Noxa expression is traditionally known to be modulated by p53-dependent mechanisms. 24,26 Many p53-independent mechanisms of Noxa upregulation have been identified. For instance, the transcription factors c-Myc, 27 HIF-1α, 28 CREB, 29 and E2F1 30 have been described to mediate p53-independent transcription of Noxa.

Furthermore, recent studies have shown that Bmi1 suppresses
Noxa expression in memory CD4 T cells and mantle cell lymphoma. 11,13 However, the mechanisms underlying Noxa induction and the functional significance of Noxa in NSCLC have not been studied.
Deguelin is a natural rotenoid extracted from several plants, including Derris trifoliata Lour (Leguminosae), Mundulea sericea (Leguminosae). It has shown great potential as a cancer chemopreventive and therapeutic agent for various types of cancer, including lung and breast cancers. 31 Deguelin has been reported to induce cell apoptosis through inhibiting many signalling pathways, such as PI3K/Akt/ HK2, 32,33 IKK/IκBα/NF-κB, 34 and AMPK/mTOR/survivin. 35 Additionally, the anti-cancer effect has been associated with many other mechanisms, including inhibition of tumour cell propagation and malignant transformation through p27/cyclinE/pRb/E2F1 or Aurora B for cell cycle control, [36][37][38][39] HIF-1α/VEGF and HGF/c-Met for anti-angiogenic, 40,41 and GSK-3β/β-catenin for anti-metastasis. 42 These findings suggest that deguelin functions as an anti-tumourigenic agent targeting apoptosis, cell cycle arrest and anti-angiogenesis for cancer therapeutic intervention. Thus, the mechanism by which deguelin induces apoptosis in human cancers including NSCLC need to be fully revealed.
In this study, we investigated the underlying mechanism of deguelin-induced apoptosis in NSCLC cell lines.

| Lentiviral infection and transient transfection
The generation of gene stable knockdown cell lines was performed as described previously. 44

| Protein preparation and Western blotting
Protein preparation and Western blotting were performed according to the method previously described. 44 The mitochondrial fraction was prepared using the Mitochondria Isolation Kit (cat#89874; Thermo Scientific, Rockford, IL, USA) according to the instructions provided. Protein concentration was determined using the BCA Assay Reagent (cat#23228; Pierce, Rockford, IL, USA). Western blotting was performed as previously described. 44

| Cell proliferation assays
Cells were seeded at a density of 2 × 10 3 cells per well in 96-well plates in 100 μL of RPMI 1640 medium containing 10% of FBS without or with different concentrations of deguelin and incubated in a 37°C, 5% of CO 2 incubator. After culturing for 24, 48, 72, or 96 hours, 10 μL of the WST-1 reagent (#11644807001; Roche, Mannheim, Germany) were added to each well and cells were incubated for 2 hours at 37°C. The absorbance of the cellular reduction of WST-1 to formazan was measured at 450 nm as previously described. 44 Three independent experiments were performed in triplicate. tures were maintained at 37°C in a 5% of CO 2 incubator for 2 or 3 weeks and colonies were counted under a microscope as previously described. 44

| Immunohistochemistry
Tumour tissues obtained from euthanized xenografted mice were embedded in paraffin and subjected to immunohistochemical staining LI ET AL.

| Chromatin-immunoprecipitation assay
Chromatin-immunoprecipitation (ChIP) assays were performed as previously described. 44 Briefly, the deguelin-treated NSCLC cells  (Table S1) were used to amplify the Noxa promoter regions present in the immunoprecipitated DNA.

| In vivo tumour growth assay
All animal procedures were approved by the Institutional Animal Care and Use Committee of the Second Xiangya Hospital, Central South University, China. Xenograft tumours were established by s.c.
Mice were weighed and tumours measured by caliper every 2 days.
Tumour volume was calculated from measurements of 2 diameters of the individual tumour according to the following formula: tumour volume (mm 3 ) = (length × width × width/2).

| Statistical analysis
Statistical analysis was performed with SPSS 16.0 (SPSS Inc, Chicago, IL, USA). Results expressed as mean ± SD were analysed using the Student's t test. Differences were considered significant when P < 0.05.  Figure 1A). We next sought to examine the expression of Bmi1 in human NSCLC and matched normal adjacent tissue specimens. In matched normal adjacent samples, Bmi1 was expressed at a relatively low level. On the contrary, Bmi1 was significantly higher in the group of tumour samples ( Figure 1B). Importantly, Noxa was highly expressed in normal adjacent tissues but F I G U R E 1 Expressions of Bmi1 and Noxa in human non-small cell lung cancer. A, Western blot analysis was performed to examine Bmi1 and Noxa expressions in several NSCLC cell lines and normal MRC5 lung cells. β-actin was used as a loading control. B, Bmi1 and Noxa protein levels in six representative NSCLC cases was assessed by Western blot analysis. β-actin was used as a loading control. N, normal adjacent tissue; T, tumour (left panel). Western blotting determined Bmi1 and Noxa protein levels in the malignant and the corresponding normal adjacent tissues of 22 NSCLC patients (right panel). The intensity was evaluated using Image J (NIH) computer software. ***P < 0.001, significant difference between groups as indicated. C, knockdown of Bmi1 attenuated NCI-H23, NCI-H1299, and NCI-H460 anchoragedependent cell growth. WST-1 assays were performed as described in Materials and Methods. Data represent mean ± SD from three independent experiments. *P < 0.05, **P < 0.01, significant difference compared with the shGFP control cells. D, Knockdown of Bmi1 attenuated NCI-H23, NCI-H1299, and NCI-H460 anchorage-independent cell growth. Soft agar assays were performed as described in Materials and Methods. Data represent mean ± SD from two independent experiments. ***P < 0.001, significant difference compared with the shGFP control cells dramatically decreased in tumour tissues ( Figure 1B). Furthermore, knockdown of Bmi1 ( Figure S1) inhibited both anchorage-dependent ( Figure 1C) and -independent cell growth ( Figure 1D)

| Deguelin inhibits anchorage-dependent and
-independent growth of human NSCLC cells Deguelin has shown potential chemopreventive and chemotherapeutical activities against various human cancers. In the present study, we first examined the effect of deguelin on the anchorage-dependent growth of NSCLC cell lines, including NCI-H23, NCI-H1299, and NCI-H460 cells. We found that deguelin significantly inhibited the growth of all tested cell lines in a dose-dependent manner (Figure 2A). Notably, we observed that H460 cells were far more sensitive to deguelin and all died at 2 μM concentration (data not shown).
Therefore, the maximum concentration of deguelin for H460 exposure in the present study was 1 μM. Meanwhile, the inhibitory effect of deguelin on anchorage-independent growth was examined in a soft agar medium. The result showed that deguelin significantly suppressed the colony formation of NSCLC cells in soft agar ( Figure 2B).
The number of colonies formed was reduced in a dose-dependent manner. In addition, the majority of the colonies in the deguelin-treated group were smaller than those in the vehicle-treated control ( Figure 2B). The results indicate that deguelin suppresses both anchorage-dependent and -independent growth of NSCLC cells.

| Deguelin inhibits the growth of NSCLC cells in a xenograft mouse model
To determine the inhibitory effect of deguelin in vivo, we explored NCI-H1299 and A549 athymic nude xenograft mouse models. Data showed that deguelin significantly inhibited tumour growth in both H1299 ( Figure 3A and C, Figure S2A) and A549 ( Figure 3B and D, Figure S2B) xenograft models. In vehicle-treated group, the average tumour volume of H1299 ( Figure 3A) and A549 ( Figure 3B) Figure 3E). Meanwhile, no obvious toxicity was observed as evaluating the change of body weight of tumour-bearing mice between the vehicle-and the deguelin-treated groups ( Figure S2C and D). In order to further evaluate the in vivo toxicity of deguelin, the blood analysis was conducted. Data showed that deguelin treatment has no obvious effect on WBC and RBC count ( Figure S2E). Moreover, the expression of Hb, ALT, AST, and BUN were consistent in vehicle-and deguelin-treated groups (Figure S2E), which indicated that deguelin had no significant toxicity to vital organ functions and no hematologic toxicities. H&E staining also showed there is no detectable toxicity in normal tissue (heart, liver, spleen, lung, and kidney) in deguelin-treated animals ( Figure S2F).
The results imply that deguelin is a well-tolerated compound at the dose of 3 mg/kg, and the inhibitory effect of deguelin on the xenograft tumour growth may partly depend on the suppression of Bmi1 and the induction of Noxa expression.

| Suppression of Bmi1 level by deguelin accompanies increased Noxa expression and apoptosis in NSCLC cells
Deguelin has been found to cause apoptosis in several cancer cell lines. 46 Our data showed that deguelin- F I G U R E 2 Deguelin inhibits anchorage-dependent and -independent growth of NSCLC cells. A, Deguelin suppresses anchorage-dependent growth of NSCLC cells. NCI-H23, NCI-H1299, and NCI-H460 NSCLC cells were treated with DMSO or the indicated concentrations of deguelin in medium containing 10% of FBS and growth was measured at the indicated times using the WST-1 assay. Data represent mean ± SD from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, significant difference compared with the DMSO control cells. B, Deguelin suppresses anchorage-independent growth of NSCLC cells. Soft agar assay was performed as described in Materials and Methods. The cultures were incubated for 14 days and then colonies were counted. Data represent mean ± SD from two independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, significant difference compared with the DMSO control cells F I G U R E 4 Effect of deguelin on the expression of apoptotic-related proteins in NSCLC cells. A, NCI-H23, NCI-H1299, and NCI-H460 NSCLC cells were treated with DMSO or the indicated concentrations of deguelin in medium containing 10% of FBS for 48 hours. After treatment, attached and floating cells were harvested. Expression of the indicated proteins was analysed by Western blotting with specific antibodies. β-actin was used as a loading control. B, H1299-shGFP, and H1299-shNoxa stable cells were treated with deguelin as indicated, whole cell extract was analysed by Western blotting with specific antibodies. C, H1299-shGFP, and H1299-shBmi1 stable cells were treated with deguelin as indicated, whole cell extract was analysed by Western blotting with specific antibodies. D and E, H1299 cells were treated with a panel of inhibitors as indicated, Western blotting was performed to detect apoptosis. F, H1299, and A549 cells were transfected with the Bmi1 plasmid, whole cell extract, cytosolic fractions, and mitochondrial fractions were subjected to Western blotting analysis with specific primary antibodies treatment in NCI-H23, NCI-H1299 and NCI-H460 cells ( Figure 5A).

Bmi1-mediated Noxa upregulation in NSCLC cells
Since pharmacological inhibition of Bmi1 by deguelin was accompanied with Noxa upregulation in NSCLC cells ( Figure 4A, left), we speculated that suppression of Bmi1 by RNA interference might also increase Noxa level. Knockdown of Bmi1 ( Figure 5B) led to increases in both Noxa protein level ( Figure 5B) and luciferase expression driven by the Noxa promoter ( Figure 5C), indicating that Bmi1 regulated the induction of Noxa gene in these NSCLC cells. To further determine whether Bmi1 directly activates the Noxa promoter, pGL3-Noxa-N1 reporter was co-transfected with the Bmi1 expression plasmid to assess the contribution of Bmi1 to the Noxa promoter activity. Results demonstrated that overexpression of Bmi1 ( Figure 5D) significantly decreased the Noxa promoter activity in NCI-H23, NCI-H1299, and NCI-H460 cells ( Figure 5E). Collectively, these data suggest that deguelin partially induces apoptosis through downregulation of Bmi1 and induction of Noxa in NSCLC cells.

| Deguelin inhibits the directly binding of Bmi1 to the Noxa gene locus
To demonstrate the effect of deguelin on the interaction of Bmi1 with the Noxa promoter in vivo, we performed ChIP assays. NSCLC cells were treated with DMSO or deguelin for 48 hours, chromatin was immunoprecipitated and the binding of Bmi1 to Noxa specific genomic regions was analyzed by PCR using primers designed around the Noxa promoter region ( Figure 6A, Figure S4, Table S1).
ChIP assays with NCI-H23, NCI-H1299, and NCI-H460 cells showed that Bmi1 interacted with the Noxa gene locus ( Figure 6B-D). Notably, accumulation of Bmi1 was always observed within the CpG islands ( Figure 6A and Figure S4A) of Noxa gene in all three NSCLC cell lines ( Figure 6B-D, #2 and #3). Also, deguelin significantly disrupted the interaction of Bmi1 with the Noxa locus ( Figure 6B-D).
These results suggest that deguelin directly targets Bmi1 to inhibit its binding to the Noxa promoter and relieves Bmi1-mediated Noxa repression to increase Noxa expression, leading to deguelin-induced apoptosis in NSCLC cells.

| DISCUSSION
Although deguelin has been reported to induce apoptosis in various cancer cells, the molecular details of this regulation remain to be fully disclosed. In this report, we showed that BH3-only protein Noxa is up-regulated during deguelin-induced apoptosis in a panel of NSCLC cell lines, which is independent of p53. Moreover, the upregulation of Noxa is consistently accompanied by deguelin-reduced Bmi1, an important component of the PRC1. We further confirmed that deguelin attenuates the binding of Bmi1 to the Noxa promoter and removes Bmi1-caused repression, thus resulting in Noxa induction. This study has revealed a novel mechanism by which deguelin activates the apoptotic machinery in NSCLC cells.
Noxa prefers to localize in mitochondria 24  Although Noxa was originally described as a p53-regulated gene, it can also be modulated in a p53-independent manner. 25 To determine whether the p53 status affected Noxa induction after deguelin treatment, p53 mutant NCI-H23, p53-null NCI-H1299, and p53 wildtype NCI-H460 were selected for further experiments. Noxa protein increased by deguelin in p53 mutant NCI-H23 cells, p53-null NCI-H1299 cells or p53 wild-type NCI-H460 cells was not significantly affected ( Figure 4A, left). These results suggested that induction of Noxa by deguelin is irrespective of the p53 status in these NSCLC cell lines.
Our results indicate that deguelin increases Noxa and decreases Deguelin is reportedly an Hsp90 inhibitor. It directly binds to the ATP-binding pocket of Hsp90, interferes with Hsp90 chaperone function, decreases the expression of many Hsp90 client proteins, and induces apoptosis in cancer cells, which reduces tumour growth. 53 Although deguelin has been mentioned as an Akt inhibitor, 32 (Figure 6). However, whether binding of Bmi1 to the Noxa promoter depends on the methylation status of the promoter and affects Noxa expression or not, needs to be further investigated.
Our results demonstrated that treatment with various concentrations of deguelin had no effect on trimethylation of H3K9 and H3K27 ( Figure S3), which are linked to gene repression. 55 Moreover, ChIP results indicated that deguelin did not inhibit the binding of H3K27me3 to the Noxa promoter ( Figure S4B and C). H3K27 methylation was reportedly served as a binding site for the recruitment of PRC1 complex. 13,56 In addition, the level of H3K27 trimethylation was significantly decreased at the Noxa gene locus but was not affected at the Ink4a gene locus, another target gene of Bmi1, in the Bmi1 −/− cells. 13 Based on these results, deguelin did not inhibit the level of H3K27me3 as well as the binding of H3K27me3 to Noxa promoter. Thus, we speculated that deguelin did not dissociate Bmi1 from all sites on chromatin but directly targeted Bmi1 to inhibit its binding to the Noxa promoter, and relieved Bmilmediated Noxa repression, causing increased expression of Noxa.
However, there is still a possibility that deguelin decreases the binding activity of Bmi1 on the promoter of other potential target genes under distinct conditions.
Our results indicated that H460 cells were far more sensitive to deguelin. Chemosensitivity is a multi-factored phenotype, which could be influenced by various molecular determinants (i.e., DNA, RNA, or protein). In other words, gene polymorphisms, gene expression alterations, protein expression, and modification differences among individuals and within individual cancers can influence drug sensitivity. 57 The genetic or epigenetic changes in genes that regulate apoptosis, DNA repair and senescence affects their intrinsic sensitivity to chemotherapy, which contributes to their intrinsic quality of chemosensitivity, when challenged with chemotherapeutic agents. 58

CONFLI CT OF INTEREST
The authors have declared no conflicts of interest.