*2.5. Tangeretin Treatment Modestly Reduces the CD44*+/*CD24*<sup>−</sup> *Population Size*

MDA-MB-231 and MCF-7 cells were treated with tangeretin, and we analyzed the BCSC marker-expressing CD44+/CD24<sup>−</sup> subpopulation. Tangeretin modestly reduced the CD44+/CD24<sup>−</sup> cell population size from 84.1% to 56.8% in MDA-MB-231 cells and from 2.2% to 0.9% in MCF-7 cells (Figure 6). This result shows that tangeretin reduces the frequency of a BCSC trait.

**Figure 6.** The effect of tangeretin on the CD44high/CD24low cell proportion. The CD44high/CD24low subpopulation within an MDA-MB-231 and MCF-7 cell population treated with tangeretin (80 μM) or DMSO for 24 h was analyzed by flow cytometry. For flow cytometry analysis, 2 <sup>×</sup> 10<sup>4</sup> cells were acquired. The gating was based on the binding of an antibody without tangeretin (red cross). The data from triplicate experiments are represented as the mean ± SD; \*\* *p* < 0.01.

#### *2.6. Tangeretin Inhibits the Stat3 Signaling Pathway and Reduces the Sox2 Level in Mammospheres*

To investigate the cellular mechanism by which tangeretin inhibits mammosphere formation, we assessed the expression levels of Stat3 and pStat3 in mammospheres. Our results showed that tangeretin decreased the total protein levels of Stat3 and pStat3 in BCSCs (Figure 7A). The protein level of phospho-Stat3 was significantly reduced in the cytosol and nucleus of mammosphere cells. The Stat3 protein level was also decreased, as shown in Figure 7B. Additionally, we investigated Stat3 probe DNA binding to tangeretin-treated nuclear extracts by EMSA. We examined Stat3 probe DNA binding to mammosphere nuclear proteins using a Stat probe. Tangeretin reduced Stat3 DNA binding (Figure 7C, # 3). The specificity of Stat3 binding was determined using a self-competitor (100×) (Figure 7C, # 4) or a mutated Stat oligo (100×) (Figure 7C, # 5). The band indicated by arrows is Stat3 and the specific DNA complex. To examine the effect on Stat3 on mammosphere formation, we performed mammosphere formation using siRNA of Stat3. Our data showed that Stat3 reduction decreased mammosphere formation (Figure 7D). To analyze the cellular function of tangeretin, after tangeretin treatment, we checked the Sox2 level because Stat3 dimer activated Sox2 gene [18]. Our data showed that tangeretin reduced the level of transcripts and protein of Sox2 (Figure 7E). Sox2 plays a role in CSC progression [19]. Our data showed that the Stat3/Sox2 signal is important for mammosphere formation. Our data showed that tangeretin, which disregulates the Stat3/Sox2 signaling pathway, was essential for inhibiting the proliferation of BCSCs (Figure 8).

**Figure 7.** Tangeretin regulates Stat3 signaling and Sox2 regulation. (**A**) The total levels of Stat3 and pStat3 were measured in mammospheres composed of MDA-MB-231 cells and MCF-7 cells after treatment with tangeretin (0 or 60 μM) for 48 h using western blotting. Total lysates were subjected to immunoblot analysis with anti-Stat3 and anti-pStat3 antibodies. β-actin was used as an internal control. (**B**) The levels of Stat3 and pStat3 in the cytosolic and nuclear protein fractions were measured in mammospheres composed of MDA-MB-231 cells after treatment with tangeretin for 48 h using western blotting. Nuclear and cytosolic proteins were run on SDS-PAGE gels, followed by immunoblotting with anti-Stat3, anti-pStat3, anti-β-actin and anti-Lamin B antibodies. (**C**) EMSA was used to analyze mammosphere nuclear proteins after treatment with tangeretin. Nuclear extracts were reacted with a Stat3 probe and subjected to 6% native PAGE. Lane 1: Stat3 probe; lane 2: nuclear extracts with the Stat3 probe; lane 3: tangeretin-treated nuclear proteins with the Stat3 probe; lane 4: nuclear proteins incubated with a self-competitor oligo (100×); and lane 5: nuclear extracts incubated with a mutated-stat3 probe (100×). The arrow indicates the DNA/Stat3 complex in the mammosphere nuclear lysates. (**D**) Effect of Stat3 on mammosphere formation using siRNA of Stat3. Stat3-downregulated MDA-MB-231 cells were cultured for seven days using mammosphere media. Images were obtained by microscrope at 100× magnification. (**E**) The transcriptional level of the Sox2 gene in MDA-MB-231 was determined in tangeretin-treated mammospheres. A Sox2-specific primer was used for real-time RT-qPCR. Western blot analysis of mammosphere under tangeretin. β-actin served as an internal control. The data are presented as the mean ± SD of three independent experiments. \*\* *p* < 0.01 versus the DMSO-treated control group.

**Figure 8.** The proposed model for CSC death induced by tangeretin is shown.

#### **3. Discussion**

It has been postulated that high consumption of fruits can prevent more than 20% of all cancer cases [20]. This preventive effect is predominantly mediated by phytochemicals interacting with specific target proteins that play important roles in cancer [21–23]. Citrus, one of the most important food sources of phytochemicals with health benefits, has many biological properties and controls key pathways involved in pathologies such as cancer [24–26]. The combination CD44+/CD24<sup>−</sup> has emerged as the most important marker for BCSC isolation, and the CD24 population of MDA-MB-231 is low [27].

First, we purified a CSC inhibitor from citrus. Assay-based fractionation and several chromatographic methods isolated one compound from a citrus powder, tangeretin. Tangeretin is the major flavonoid of citrus. It also has antioxidant, anti-inflammatory and anticancer properties [17]. Tangeretin modestly induces apoptosis in bladder cancer cells through mitochondrial dysfunction [28]. Tangeretin and nobiletin induces G1 cell cycle arrest but not apoptosis in breast and colon cancers [29]. Nobiletin inhibits CD36-dependent tumor angiogenesis, migration, invasion and sphere formation through the CD36/Stat3/Nf-Kb signaling axis [30,31]. Quercetin suppresses breast CSCs through its inhibition of the PI3K/Akt/mTOR signaling pathway [32]. Despite numerous studies, there are no studies on tangeretin-induced antiproliferative and anti-CSC effects. Our results showed that tangeretin suppresses the proliferation of BCSCs. Tangeretin inhibited mammosphere formation in breast cancer cells (Figure 3) and modestly induced apoptosis in CSCs (Figure 4).

Our data showed the reduction of CD44+/CD24<sup>−</sup> subpopulation, mammosphere formation, colony formation and tumor formation. It is known that tangeretin did not inflict damaging effects sufficient to result in a reduced capacity to survive and proliferate. However, inhibition of the growth of breast cancers without inducing cancer cell death may be advantageous in treating breast tumors. Breast cancer cells resumed growth similar to untreated control within a day of tangeretin removal [29]. The Stat3 protein is essential for the maintenance of BCSCs [4]. The acetylated derivative of tangeretin (5-AcTMF) had anticancer effects on human glioblastoma multiforme cells through blockade of Stat3 signaling [33]. Extracellular IL-8 protein is a factor for BCSCs formation [13]. We investigated Stat3 signaling under tangeretin treatment. Tangeretin suppressed the total protein levels of Stat3 and pStat3. The nuclear protein levels of Stat3 and pStat3 were also decreased by tangeretin. We assessed Sox2 transcript levels in BCSCs under tangeretin treatment and confirmed that the Sox2 mRNA level was decreased in the treated samples. In addition, the protein level of Sox2 was decreased in treated cells compared with control cells (Figure 7). Finally, tangeretin had an inhibitory effect on tumor growth in a breast cancer xenograft model. The tangeretin-treated group had a smaller tumor size and lower tumor weight than the control group.

In our study, tangeretin inhibited BCSC formation and targeted BCSCs by inhibiting the Stat3/Sox2 signaling pathway. Tangeretin is a possible therapeutic agent for breast cancer and BCSCs.

#### **4. Materials and Methods**

#### *4.1. Reagents*

Silica gel 60 and TLC plates were purchased from MERK (Darmstadt, Germany) and Sephadex LH-20 was obtained from Pharmacia (Uppsala, Sweden). High-pressure liquid chromatography was performed with a Shimadzu application system (Shimadzu, Kyoto, Japan). Cell viability was measured using the EZ-Cytox Cell Viability Assay Kit (DoGenBio, Seoul, Korea). Tangeretin was obtained from ChemFaces Co. (Hubei, China).

#### *4.2. Plant Materials*

Citrus peel was collected from Jeju Island, South Korea. The citrus peel was freeze-dried, and the dried citrus was ground. The samples (No. 2017\_030) were deposited in the Department of Biomaterials, Jeju National University, JeJu-Si, South Korea.

#### *4.3. Extraction and Isolation of an Inhibitor*

Citrus powder was extracted with 100% methanol. The bioassay-based isolation protocol is summarized in Figure 1A. The methanol extracts were mixed with water, and the methanol was evaporated. The water-suspended samples were extracted with equal volumes of ethyl acetate. The ethyl acetate-concentrated part was loaded onto a silica gel column (3 × 35 cm) and fractionated with solvent (chloroform-methanol, 20:1) (Figure S1). The three parts were divided and assayed by evaluating mammosphere formation. The #2 part potentially inhibited mammosphere formation. The #2 part was loaded onto a Sephadex LH-20 open column (2.5 × 30 cm) and eluted in three fractions (Figure S2). The three parts were obtained and assayed by evaluating mammosphere formation. Part # 3 showed inhibition of mammosphere formation. Part #3 was isolated using preparatory TLC (glass plate; 20 × 20 cm) and developed in a TLC glass chamber. Individual bands were separated from the silica gel plate. Each fraction was assayed by evaluating mammosphere formation (Figure S3). The #1 fraction was loaded onto a Shimadzu HPLC instrument (Shimadzu, Tokyo, Japan). HPLC used an ODS 10 × 250 mm C18 column (flow rate; 3 mL/min). The mobile phase was water and acetonitrile. For elution, the acetonitrile proportion was initially set at 20%, increased to 80% at 20 min and finally increased to 100% at 40 min (Figure S4).

#### *4.4. Structural Analysis of the Purified Sample*

The chemical structures of the isolated compounds were determined by mass and nuclear magnetic resonance (NMR) measurements. The molecular weight was established as 372 by ESI-mass spectrometry, which showed a quasimolecular ion peak at *m*/*z* 373.3 [M + H]<sup>+</sup> in the positive mode (Figure S9). The 1H NMR spectrum in CDCl3 exhibited signals due to four aromatic methine protons at δ 7.86 (2H) and 7.01 (2H), which are attributable to a 1,4-disubstituted benzene, one aromatic singlet methine at δ 6.59, and five methoxy groups at δ 4.09, 4.01, 3.94, 3.93 and 3.87. In the 13C NMR spectrum, there were 20 carbon peaks, including a carbonyl carbon at δ 177.3; nine sp<sup>2</sup> quaternary carbons at δ 162.3, 161.2, 151.4, 148.4, 147.7, 144.0, 138.0, 123.8 and 114.8; five sp<sup>2</sup> methine carbons at δ 127.7 (×2), 114.5 (×2) and 106.7; and five methoxy carbons at δ 62.2, 62.0, 61.8, 61.6 and 55.5 (Figure S5). All proton-bearing carbons were assigned by the HMQC spectrum, and the 1H-1H COSY spectrum established a partial structure, 1,4-disubstituted benzene (Figures S6, S7 and S9). The chemical structure was determined to be from the HMBC spectrum, which exhibited long-range correlations from the methine proton at δ 6.59 to the carbons at δ 177.3, 161.2, 123.8 and 114.8, and from the methine protons at δ 7.86 to the carbon at δ 161.2, implying that this compound was a polymethoxylated flavone (Figures S8 and S9). Finally, long-range correlations from the five methoxy proton peaks to the oxygenated sp2 quaternary carbons established the structure of this compound as that of tangeretin (Figure 2).

### *4.5. Culture of Breast Cancer Cells and Mammospheres*

Two breast cancer cell lines, MCF-7 and MDA-MB-231, were purchased from the American Type Culture Collection (Rockville, MD, USA) and maintained in DMEM supplemented with 10% fetal bovine serum (FBS; HyClone Fisher Scientific, CA, USA) and 1% penicillin/streptomycin (Gibco, Thermo Fisher Scientific, CA, USA). Cancer cells (5 <sup>×</sup> <sup>10</sup><sup>4</sup> or 1 <sup>×</sup> <sup>10</sup>4) were incubated in an ultralow-attachment 6-well plate with MammoCultTM culture medium (STEMCELL Technologies, Vancouver, BC, Canada). All cells were cultured in a humidified 5% CO2 incubator at 37 ◦C for 7 days. The formation of mammospheres was assessed by the NICE program [34]. Mammosphere formation was examined by examining mammosphere formation efficiency (MFE) (%) [35].
