Hinokitiol up-regulates miR-494-3p to suppress BMI1 expression and inhibits self-renewal of breast cancer stem/progenitor cells

Hinokitiol (β-thujaplicin) is a tropolone-related compound that has anti-microbe, anti-inflammation, and anti-tumor effects. Cancer stem/progenitor cells (CSCs) are a subpopulation of cancer cells with tumor initiation, chemoresistant, and metastatic properties and have been considered the important therapeutic target in future cancer therapy. Previous studies reported that hinokitiol exhibits an anti-cancer activity against murine tumor cells through the induction of autophagy. The current research revealed that hinokitiol suppressed the self-renewal capabilities of human breast CSCs (BCSCs) and inhibited the expression of BMI1 at protein level without suppressing its mRNA. Treatment of hinokitiol in mammospheres induced the expression of miR-494-3p and inhibition of miR-494-3p expression in BCSCs. This treatment abolished the suppressive effects of hinokitiol in mammosphere formation and BMI1 expression. BMI1 is a target of miR-494-3p by luciferase-based 3′UTR reporter assay. Overexpression of miR-494-3p in BCSCs caused the down-regulation of BMI1 protein, inhibition of mammosphere forming capability, and suppression of their tumorigenicity. Moreover, miR-494-3p expression was significantly and inversely correlated with patient survival in two independent public database sets. Furthermore, treatment of hinokitiol in vivo suppressed the growth of xenograft human breast tumors as well as the expression of BMI1 and ALDH1A1 in xenograft tumors. In conclusion, these data suggest that hinokitiol targets BCSCs through the miR-494-3p-mediated down-modulation of BMI1 expression.


Research Paper
cells with surface markers of CD24-CD44+ [6] or high intracellular aldehyde dehydrogenase (ALDH) activity [7]. CSC activity could also be determined by tumorsphere cultivation, a non-adherent culture condition that enables the enrichment of CSC population from cancer cell lines or primary cancer cells [8][9][10]. The polycomb complex protein BMI1 regulates the self-renewal capability of normal and malignant mammary stem cells [11]. Overexpression of BMI1 in normal mammary epithelial cells increased mammosphere formation [11]. Cooverexpression of Bmi1 and activated H-Ras (RasG12V) in MCF10A cells resulted in the formation of poorly differentiated carcinomas with epithelial-mesenchymal transition features in severe combined immunodeficient mice [12]. Recently, a small molecule inhibitor of BMI1, PTC-209, has been identified to cause a decrease in BCSCs through the up-regulation of tumor suppressor microRNAs (miRNAs), such as miR-200 and miR-141 [13].
MiRNAs belong to a family of small non-coding RNAs with 19-24 nucleotides in length. This group functions as gene regulators to suppress gene expression by binding to the 3′-untranslated region to cause translation inhibition or mRNA degradation of their target genes [14]. The functions of miR-494-3p in cancer are controversial. miR-494-3p has been reported to demonstrate oncogenic effects by modulating NOTCH1 and PTEN/PI3K/AKT signaling in non-small cell lung cancer [15], accelerating cell proliferation in liver cancer, and down-regulation or mutation in colorectal cancer [16]. On the other hand, miR-494-3p could inhibit gastrointestinal stromal tumor cell proliferation by targeting KIT [17] or suppressing invasion of prostate cancer cells through down-regulation of CXCR4 [18]. Recently, miR-494-3p was demonstrated to target PAK1 in breast cancer cells, leading to the suppression of cell invasion [19]. However, the role of miR-494-3p in regulation of self-renewal of BCSCs remains unclear.
Hinokitiol is a natural monoterpenoid originally extracted from Taiwanese hinoki and has antiinflammatory and anti-microbial abilities [20,21]. Furthermore, hinokitiol induces apoptosis in cancer cells through a caspase 3-dependent pathway or through cellcycle arrest [22][23][24]. Previous studies demonstrated that hinokitiol caused cell death in murine breast and colorectal cancer cells through the induction of autophagy [25]. These studies indicated that hinokitiol could serve as a novel anti-cancer compound. The CSC targeting effect of this compound needs further investigation.
The present study reveals that hinokitiol inhibited the self-renewal of BCSCs and down-regulated BMI1 protein expression without affecting its mRNA level. The expression of miR-494-3p in mammospheres was induced by hinokitiol. Inhibition of miR-494-3p abolished the inhibitory effect of hinokitiol in targeting BMI1. Overexpression of miR-494-3p in human breast cancer cells suppressed tumor growth of BCSCs in vivo. The expression of miR-494-3p was significantly and inversely correlated with breast cancer patient survival in two independent public database sets. Finally, hinokitiol suppressed tumor growth in a xenograft breast cancer model. The upregulation of miR-494-3p as well as downregulation of BMI1 was observed in hinokitiol-treated xenograft tumors. In conclusion, hinokitiol could target BCSCs in vitro and in vivo through upregulation of miR-494-3p, inhibiting BMI1 expression.

Hinokitiol's ability to inhibit the self-renewal capabilities of BCSCs
The cytotoxic effects of hinokitiol were examined in two human breast cancer cell lines, AS-B145 and BT-474. As shown in Figure 1, the half-maximal inhibitory concentration (IC 50 ) values of hinokitiol in AS-B145 ( Figure 1A) and BT-474 ( Figure 1B) cells were 266.9 ± 42.6 μM and 46.5 ± 8.0 μM, respectively. The potential anti-CSC effects of hinokitiol was further evaluated in the range of concentrations without causing massive cell death at 0-10 μM with a mammosphere cultivation assay. This assay determines the self-renewal capability of BCSCs [8,26]. Hinokitiol significantly inhibited primary and secondary mammosphere formation at 10 μM, a concentration below IC 50 value, in both AS-B145 ( Figure  1C) and BT-474 ( Figure 1D). These results indicate that hinokitiol displays an anti-self-renewal activity of BCSCs in vitro.
The mediation of miR-494-3p on the suppressive effects of hinokitiol to the self-renewal capability and BMI1 expression in BCSCs BMI1 positively regulates the self-renewal capability of BCSCs [11,12]. The effect of hinokitiol to the BMI1 expression of mammospheres in AS-B145 and BT-474 breast cancer cells was examined. As shown in Figure 2, hinokitiol significantly inhibited BMI1 expression in both AS-B145-and BT-474-derived mammospheres at a concentration of 10 μM ( Figure  2A). BMI1 was further overexpressed in BT-474 cells in order to reduce the therapeutic effects of hinokitiol in targeting BCSCs. The ALDEFLUOR assay findings show that hinokitiol treatment obviously decreased ALDH+ BCSCs within BT-474 mammospheres (from 69.5% to 20.7%), but the inhibition was less efficient in BMI1overexpressing cells (the ALDH+ cells remained 40.6%) ( Figure 2B). However, hinokitiol did not suppress the mRNA expression of BMI1 in AS-B145-or BT-474derived mammospheres ( Figure 2C). This result suggests that the inhibitory effects of hinitiol to BMI1 expression may be mediated by a specific miRNA. Previous experiments demonstrated that BMI1 was a target of miR-494-3p in oral squamous carcinoma cells [27]. The expression of miR-494-3p in mammospheres was then detected after hinokitiol treatment. Results showed that hinikitiol induced miR-494-3p in mammospheres derived from AS-B145 cells ( Figure 3A). When AS-B145 or BT-474 cells were transfected with miR-494-3p inhibitor, the inhibitory effects of hinokitiol to BMI1 expression ( Figure 3B) or mammosphere formation ( Figure 3C) was abolished. The results of the luciferase reporter assay suggest that transfection of miR-494-3p mimic into 293T human embryonic kidney cells or BT-474 breast cancer cells significantly reduced the activity of luciferase fused with wildtype 3′-UTR of BMI1, but not to the luciferase fused with mutated 3′-UTR by deletion of miR-494-3p binding sites ( Figure 4A). Furthermore, overexpression of miR-494-3p in AS-B145 and BT-474 mammospheres suppressed BMI1 expression ( Figure 4B). These results indicate that the inhibitory effect of hinokitiol to the selfrenewal capability of BCSCs is mediated by miR-494-3pinduced BMI1 down-regulation.

miR-494-3p as an oncosuppressor miRNA in breast cancer
With transfection of miR-494-3p mimic, the overexpression of miR-494-3p significantly suppressed the primary and secondary mammosphere formations of AS-B145 ( Figure 5A) and BT-474 cells ( Figure 5B). FACS was performed to analyze the intracellular miRNA expression utilizing Smartflare fluorescent beads [28]. In this analysis, BT-474 cells were further sorted into two populations of miR-494-3p low and miR-494-3p high groups with Smartflare miR-494-3p fluorescent probe ( Figure 5C). Results show that the expression of BMI1 protein was higher in miR-494-3p low BT-474 cells than those of miR-494-3p high counterparts ( Figure 5C). With mammosphere assay, the miR-494-3p high BT-474 cells displayed a poor CSC activity that formed less number of mammospheres, as well as decreased in mammosphere size, than the miR-494-3p low counterparts ( Figure 5D). The miR-494-3p overexpression was introduced by the lentiviral delivery of miR-494-3p precursor into the BT-474 cells and performed xenograftment assay in vivo in NOD/SCID immunocompetent mice; the tumor growth of BT-474 cells with miR-494-3p overexpression was significantly slower than control tumors ( Figure 5E, p= 0.0114). Immunohistochemistry analysis demonstrated a reduction of the positive stain of nuclear BMI1 in tumors derived from miR-494-3p overexpressed BT-474 cells ( Figure 5F). Furthermore, the down-regulation of BMI1 protein expression in xenograft tumors derived from miR-494-3p overexpressed BT-474 cells was confirmed by Western blot ( Figure 5F). Kaplan-Meier survival analysis was performed in breast cancer patients from GSE37405 dataset (overall survival among ER+ breast cancer patients) by MIRUMIR ( Figure 6A) and the Cancer Genome Atlas (TCGA) dataset (metastasis-free survival among invasive breast carcinoma patients) by PROGmiR V2 online tools ( Figure 6B). The results showed that the lower expression of miR-494 had a significant poor survival time ( Figure 6, p= 0.00113 for GSE37405 and p=0.0125 for TCGA dataset). Moreover, the results indicate that miR-494-3p functions as an oncosuppressor in breast cancer. To examine the therapeutic potential of hinokitiol in vivo to breast cancer, the BT-474 mammosphere cells were injected into the mammary fat pads of NOD/SCID and treated with 40 mg/kg hinokitiol intraperitoneally when the tumor volume reached to 100 mm 3 . Compared with ethanol-treated group, the tumor growth of hinokitioltreated group was significantly reduced ( Figure 7A, p=0.014). By qRT-PCR analysis, the expression levels of miR-494-3p in hinokitiol-treated tumors were significantly increased when compared to ethanol-treated group ( Figure  7B, p=0.0002). The BMI1 protein expression level in hinokitiol-treated tumor samples was reduced as compared with the EtOH-treated tumor in Western blot analysis ( Figure 7C). By immunohistochemistry analysis, the expression level of ALDH1A1, one of the BCSC markers [7], and nuclear BMI1 was decreased in hinokitioltreated tumors ( Figure 7D). These data suggest that the therapeutic effect of hinokitiol in vivo in the suppression of breast tumor growth is associated with the upregulation of miR-494-3p, leading to the down-regulation of BMI1 expression.

DISCUSSION
Hinokitiol displays an anti-cancer activity by the induction of cell-cycle arrest [23], apoptosis [24], DNA damage [29], or autophagic cell death [25]. This study reports that the anti-CSC activity of hinokitiol was partially mediated by miR-494-3p. The inhibition of miR-494-3p expression partially abolished the anti-BCSC effects of hinokitiol ( Figure 3). This experiment shows that the anti-cancer activity of hinokitiol is mediated by the induction of an oncosuppressor miRNA. Hibino et al. have demonstrated that inhibition of the enhancer of zeste homolog 2 (EZH2) by 3-deazaneplanocin A induced tumor suppressor miRNAs in liver cancer cells [30]. A similar phenomenon was also found in a recent multiple myeloma study [31]. In hinokitiol-treated xenograft tumors derived from BT-474 mammospheres, the downregulation of EZH2 was observed when compared with EtOH-treated group (Supplementary Figure 1). This finding suggests that hinokitiol may induce miR-494-3p expression through the down-regulation of EZH2, but the underlying molecular mechanism requires further investigation. Previous experiments reveal that hinokitiol could induce proteasomal degradation of epidermal growth factor receptor (EGFR), which led to the suppression of vasculogenic mimicry activity of BCSCs [32]. The EGFR signaling has been demonstrated to contribute to the enhancement of BCSC self-renewal induced by versican [33] or by tumor-associated macrophages [34]. Downregulation of EGFR expression by hinokitiol may also involve in the anti-BCSC effect but remains to be further investigated. The relationship between the expression of miR-494-3p and EGFR in breast cancer cells will be further investigated.
The function of miR-494-3p in carcinogenesis is controversial. Several reports suggested that miR-494-3p is an oncogenic miRNA. This miRNA shortens the disease-free survival time in lung cancer patients with high expression levels [15] or the inhibition of PTEN by miR-494-3p, leading to the activation of Akt in nasal natural killer cell lymphoma [35]. By contrast, other reports indicated that miR-494-3p could be a tumor suppressor miRNA. The direct targeting of Sox9 by miR-494-3p led to the suppression of cell migration, invasion, as well as tumor growth of chondrosarcoma cells [36]. The dual role of a single miRNA in carcinogenesis would not be surprising. miR-375 functioned as both an oncomiR and tumor suppressor miRNA in prostate cancer, depending on the stage of tumor progression and hormone status [37]. Furthermore, the effect of miRNAs to carcinogenesis mainly depends on the cellular context in different tumors [38]. This study presents the tumor-suppressive function of miR-494-3p in breast cancer by inhibiting BCSC selfrenewal and directly targeting BMI1 (Figure 4 and 5). These findings are consistent with the previous studies, which demonstrated that miR-494-3p induced cellular senescence in oral squamous carcinoma cells through the downregulation of BMI1 [27]. These results also show the tumor suppression effects of miR-494-3p. In addition, miR-494-3p was negatively correlated with the overall survival of ER+ breast cancer patients (GSE37405) ( Figure 6A) or metastasis-free survival of breast invasive carcinoma patients (TCGA data) ( Figure 6B). These data were consistent with a recent study from Zhan et al., showing that the expression of miR-494 was significantly reduced in tumors when compared with adjacent nonneoplastic breast tissues [19]. Altogether, these data strongly suggest that miR-494-3p is an oncosuppressor miRNA in breast cancer.
The current research shows that irradiation could induce epithelial-mesenchymal transition in breast cancer cells by the upregulation of BMI1 [39]. Knockdown of BMI1 abolished irradiation-induced cell migration in breast cancer cells [39]. Results regarding the inhibitory effect of hinokitiol in BMI1 expression through the induction of miR-494-3p suggest that hinokitiol has the potential to develop into a sensitization agent in breast cancer radiotherapy. On the other hand, utilization of    between miR-494 expression and overall survival in a GSE37405 dataset of ER+ breast cancer patients (A) or metastasis free survival in TCGA invasive breast cancer dataset (B) was analyzed by MIRUMIR or PROGmiR V2 website. www.impactjournals.com/oncotarget miR-494-3p oligos as an enhancer is also recommended to facilitate the therapeutic efficiency of hinokitiol in the future. In conclusion, these data demonstrated that hinokitiol could suppress the self-renewal and tumorigenicity of BCSCs through miR-494-3p-mediated BMI1 inhibition. These findings suggest that hinokitiol has a potential to be developed as a chemoprevention agent in breast cancer.

Cell viability assay
Cells were treated with a sequential concentration of hinokitiol (0-1000 μM) and cultured for 48 h. Cell survival was assessed using WST-1 reagent by measuring absorbance at 440nm wavelength. The half maximal inhibitory concentration (IC50) value was calculated by GraFit software (version 7, Erithacus Software Ltd., Surrey, UK)

Immunoblot analysis
The total protein concentration in each sample was determined with the bicinchoninic acid (BCA) protein assay (Pierce Biotechnology, Rockford, IL, USA). Proteins were separated by SDS-PAGE, transferred onto Polyvinylidene Fluoride (PVDF) membranes (Pall Corporation, Port Washington, NY, USA), and probed with rabbit polycolnal anti-BMI1 (Novus Biologicals, Littleton, CO, USA) or rabbit polyclonal anti-GAPDH (GeneTex International Corporation, Hsinchu City, Taiwan). Horseradish peroxidase-conjugated goat anti-rabbit IgG polyclonal antibody (GeneTex) was used as the secondary antibody followed by incubation with T-Pro LumiFast Chemiluminescent Substrate (JF Ji-Feng Biotechnology, New Taipei City, Taiwan) and the signals were visualized and captured with a FUSION Solo S Imaging system (Vilber Lourmat, MArne-la-Valée, France). The signals were quantified with ImageJ software (version 1.51i, National Institute of Mental Health, Bethesda, MA, USA). For quantification of BMI1 protein in xenograft tumors by western bot, 10 μm of paraffin embedded tumors were sliced and proteins were extracted according to the protocol of Guo et al. [40].

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)
qRT-PCR was performed as previously described [27]. Briefly, total RNA was extracted and purified by Quick-RNA™ MiniPrep Plus (Zymo Research Corp, Irvine, CA, USA). 1 μg total RNA was used for complementary DNA (cDNA) conversion by RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). For miR-494-3p detection, a specific RT primer for miR-494-3p (Guangzhou RiboBio Co., Ltd., Guangzhou, China) was used for cDNA conversion and qPCR was then performed by KAPA SYBR ® FAST qPCR Kit (Kapa Biosystems, Inc., Wilmington, MA, USA) and StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific) with specific qPCR primer pair (Guangzhou RiboBio Co., Ltd.) under a condition as described in [27]. RNU6B was used as internal control for analyzing miR-494-3p expression. The primers sequences for detection of BMI1 and MRPL19 were used as described in [27].

Transfection
The transfection of plasmid DNA, miRNA mimic or inhibitor was performed by TurboFect Transfection Reagent (Thermo Fisher Scientific) with the manufacture's protocol. Briefly, 100 nM miRNA mimic or inhibitor (purchased from Guangzhou RiboBio Co., Ltd.) or 1 μg of pCMV14-3X flag or pCMV-BMI1-flag was complexed with transfection reagent as a ratio of 1 μg nuclei acid: 2 μl reagent at room temperature for 15 minutes and then added into wells of 6-well-plates with cell attachment at 60% confluency. Cells were then harvested at 48h post transfection for further experiments.

Constructs and luciferase-based reporter assay
Human BMI1 gene was amplified by PCR from cDNA of BT-474 cells and cloned into pCMV14-3X flag vector with following primers: KpnI-BMI1-F (5'-CgCggTAccATgCATCgAACAACgAgAATC-3') and BMI1ns-BamHI-R (5'-AgCggATCCACCAgAAgAAgTT gCTgATgAC-3'). The firefly luciferase reporter plasmid with full length BMI1 3'-UTR was purchased from OriGene Technologies, Inc. (Rockville, MD, USA) and a reporter plasmid of mutant BMI1 3'-UTR with a deletion of putative miR-494-3p binding region (deletion of nucleotide positions from 762 to 768 in 3'-UTR of BMI1) was constructed with site-directed mutagenesis kit as described in [27]. For reporter assay, the firefly luciferase reporter plasmid was mixed with a renilla luciferase plasmid as a ratio of 50:1 and transfected together with miR-494-3p mimic or negative control mimic into 293T or BT-474 cells. The luciferase activity was detected at 48h post transfection as described in [27].

Fluorescence-activated cell sorting (FACS)
ALDH+ BCSCs within BT-474 mammospheres were detected by ALDEFLUOR assay (StemCell Technologies, Inc., Vancouver, BC, Canada) according to our previous report [43]. The fluorescence signals were analyzed by COULTER TM Epics XL flow cytometry (Beckman Coulter, Inc. Brea, CA, USA). In order to sort the differential miR-494-3p expressing cells, the Smartflare beads was used (Merck Millipore, Temecula, CA, USA). BT-474 cells were grew in a 10 cm dish for 80% confluency with 8 ml culture medium, added 8 μl miR-494-3p Smartflare-Cy5 beads and then incubated at 37°C for 16 hours. After incubation, cells were harvested with trypsin/EDTA and suspended in DMEM/F12 supplemented with 5% FBS and performed cell sorting with FACSAria cell sorter (BD Biosciences).

Human breast cancer xenograftment model
All the animal studies were operated following a protocol approved by Institutional Animal Care & Utilization Committee of Chung Shan Medical University. For analysis of the expression of miR-494-3p to the tumorigenicity of BT-474 cells, cells were transduced with lentivirus carrying negative control or miR-494-3p precursor which were purchased from BioSettia (San Diego, CA, USA) for 3 days and injected into mammary fads of NOD/SCID mice (purchased from National Laboratory Animal Center, Taipei, Taiwan) as 1×10 6 cells/50 μl Matrigel/site. For examine the in vivo therapeutic effect of hinokitiol, BT-474 cells were firstly cultured into secondary mammospheres and dissociated by HyQTase treatment. The dissociated secondary mammosphere cells were suspended in 2.5 mg/ml Matrigel (BD Biosciences) and injected into mammary fat pads as 2×10 4 cells/50 μl/site. Hinokitiol treatment was performed intraperitoneally when tumors reached 100 mm 3 . Tumor volume was calculated as d 2 ×D×π/6 where d and D were the shortest and longest diameter in mm, respectively [44].

Immunohistochemistric analysis
Xenografted tumors were harvested, fixed with 3.7% formaldehyde and embedded into paraffin. 5 μm sections were sliced and the expression of BMI1 or ALDH1A1 was detected by polyclonal rabbit anti-BMI1 antibody (Novus Biologicals, LLC) or polyclonal rabbit anti-ALDH1A1 antibody (GeneTex Inc.) followed by a standard avidinbiotin-peroxidase complex method. 3,3'-Diaminobenzidine (DAKO, Carpinteria, CA) was then used to detect the antibody binding. The images of sections were scanned by TissueFAXS Plus (TissueGnostics GmbH, Vienna, Austria).

Analysis of the association between miR-494-3p expression and overall survival rate in breast cancer patients
The association between miR-494-3p expression and overall survival rate of breast cancer patients was analyzed using public breast cancer datasets of GSE37405 and TCGA by online analysis tools. The dataset of GSE37405 was analyzed by MIRUMIR website (http:// www.chemoprofiling.org/cgi-bin/GEO/MIRUMIR/ web_run_MIRUMIR.V1.pl) whereas TCGA dataset was analyzed by PROGmiR V2 website (http://xvm145. jefferson.edu/progmir/).

Statistical analysis
One-way analysis of variance (one-way ANOVA) was used to identify differences between experimental groups and the control group. A P value less than 0.05 was considered to be statistically significant.

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.

GRANT SUPPORT
This work is supported by Ministry of Science and Technology in Taiwan (grant No. MOST 103-2314-B-040-015-MY3).