**The Antiproliferative Activity of Oxypeucedanin via Induction of G2**/**M Phase Cell Cycle Arrest and p53-Dependent MDM2**/**p21 Expression in Human Hepatoma Cells**

### **So Hyun Park, Ji-Young Hong, Hyen Joo Park and Sang Kook Lee \***

College of Pharmacy, Natural Products Research Institute, Seoul National University, Seoul 08826, Korea; hanirela@snu.ac.kr (S.H.P.); jyhong7876@snu.ac.kr (J.-Y.H.); phj00@snu.ac.kr (H.J.P.)

**\*** Correspondence: sklee61@snu.ac.kr; Tel.: +82-2-880-2475

Received: 5 December 2019; Accepted: 21 January 2020; Published: 23 January 2020

**Abstract:** Oxypeucedanin (OPD), a furocoumarin compound from *Angelica dahurica* (Umbelliferae), exhibits potential antiproliferative activities in human cancer cells. However, the underlying molecular mechanisms of OPD as an anticancer agent in human hepatocellular cancer cells have not been fully elucidated. Therefore, the present study investigated the antiproliferative effect of OPD in SK-Hep-1 human hepatoma cells. OPD effectively inhibited the growth of SK-Hep-1 cells. Flow cytometric analysis revealed that OPD was able to induce G2/M phase cell cycle arrest in cells. The G2/M phase cell cycle arrest by OPD was associated with the downregulation of the checkpoint proteins cyclin B1, cyclin E, cdc2, and cdc25c, and the up-regulation of p-chk1 (Ser345) expression. The growth-inhibitory activity of OPD against hepatoma cells was found to be p53-dependent. The p53-expressing cells (SK-Hep-1 and HepG2) were sensitive, but p53-null cells (Hep3B) were insensitive to the antiproliferative activity of OPD. OPD also activated the expression of p53, and thus leading to the induction of MDM2 and p21, which indicates that the antiproliferative activity of OPD is in part correlated with the modulation of p53 in cancer cells. In addition, the combination of OPD with gemcitabine showed synergistic growth-inhibitory activity in SK-Hep-1 cells. These findings suggest that the anti-proliferative activity of OPD may be highly associated with the induction of G2/M phase cell cycle arrest and upregulation of the p53/MDM2/p21 axis in SK-HEP-1 hepatoma cells.

**Keywords:** oxypeucedanin; *Angelica dahurica*; antiproliferation; G2/M phase cell cycle arrest; p53; SK-Hep-1; hepatoma cells

### **1. Introduction**

Hepatocellular carcinoma (HCC) is the sixth most frequently diagnosed cancer and the second most common cause of death from cancer [1]. Approximately 30 to 40% of patients with HCC diagnosed at early stages are potentially effectively treated by surgical therapies such as hepatectomy or liver transplantation [2]. However, the disease diagnosed at an advanced stage or re-progression stage after treatment has poor prognosis due to the fundamental liver disease and lack of other treatment options [2–4]. In accordance, any systemic therapy has not effectively contributed to survival for patients with advanced HCC [5]. Therefore, it is necessary to procure potential chemotherapeutic agents for the treatment of advanced HCC cells.

*Angelica dahurica* (Umbelliferae) is an indigenous plant mainly distributed in Korea, China, and Russia. The root of *Angelica dahurica* has been used for the control of hysteria, bleeding, menstrual disorder, neuralgia and pain as a traditional medicine in Korea. Previous phytochemical studies revealed that the plant is a rich source of furanocoumarins, including oxypeucedanin [6]. Oxypeucedanin (OPD) (Figure 1), a coumarin-type major constituent of the root of *Angelica dahurica*, has been reported to have several biological activities, including anti-mutagenic, uterus contraction, blood pressure increase, and cytotoxic and antiproliferative effects against cancer cells [7–9].

**Figure 1.** Chemical structure of oxypeucedanin (OPD).

When the DNA damage stimulus comes to the cells, the response of the cellular system is complex and finely controlled. The cellular response involves the functions of gene products that recognize DNA damage signals and activate processes such as the inhibition of proliferation, the stimulation of repair mechanisms, or the induction of apoptosis [10–12]. Generally, the cellular response to DNA damage and the disturbance to replication involve the activation of checkpoints, following signal transduction pathways for regulation of cell cycle progression and cell division [10,11]. Deficiency in these checkpoint responses can induce cell death, genomic instability, or predisposition to cancer [12–14]. In addition, cell cycle progression is finely regulated by cell cycle regulatory proteins and checkpoint proteins depending on the specific cell cycle phase. In particular, the G2/M phase of the cell cycle is governed by the expression of key G2/M transition regulatory proteins and the ATR/Chk1 signaling pathways. p53, a tumor suppressor, is also considered to be involved in the transcriptional regulation of a large number of growth-arrest and apoptosis-related genes [15].

Although several biological activities of OPD have been previously reported, the precise molecular mechanism of OPD related to its antiproliferative activity against human liver cancer cells has not been fully elucidated. In the present study, the growth-inhibitory activity and the underlying mechanisms of action of OPD were investigated in SK-Hep-1 human hepatoma cells.

### **2. Results**

### *2.1. Antiproliferative E*ff*ects of Oxypeucedanin (OPD) in SK-Hep-1 Human Hepatoma Cells*

To primarily evaluate whether OPD shows potential growth-inhibitory effects on human cancer cells, the anti-proliferative activity of OPD was determined in a panel of human cancer cell lines. As summarized in Table 1, OPD inhibited the growth of human cancer cells. Among the cell lines tested, SK-Hep-1 human hepatoma cells were the most sensitive to OPD. Therefore, further study on the mechanism of action of OPD in the regulation of cell proliferation was conducted in SK-Hep-1 cells. In addition, since OPD was shown to have the anti-proliferative activity against SK-Hep-1 cells, the isolated OPD analogs from the roots of *Angelica dahurica* were also evaluated for their antiproliferative activity in SK-Hep-1 cells. Among the test compounds, OPD was the most active growth inhibitor against SK-Hep-1 cells (Table 2).


**Table 1.** Anti-proliferative effects of furanocoumarins from *Angelica dahurica* on various human cancer cells.

<sup>a</sup> Cancer cell line: MDA-MB-231 (breast), T47D (breast), SK-Hep-1 (liver), A549 (lung), SNU638 (stomach), <sup>b</sup> IC50 value: μM.


**Table 2.** Anti-proliferative effects of oxypeucedanin analogs on SK-Hep-1 cells.

Since the anti-proliferative activity of OPD was found in SK-Hep-1 human hepatoma cells for 72 h, the growth-inhibitory activity of OPD was also investigated for 24 or 48 h in SK-Hep-1 cells. As shown in Figure 2, OPD exhibited growth-inhibitory activity against SK-Hep-1 human hepatoma cells in a concentration- and time-dependent manner.

**Figure 2.** Anti-proliferative effects of OPD in SK-Hep-1 human hepatoma cells (left) and MRC5 human normal cell line (right). SK-Hep-1 cells were cultured in a 96-well plate and treated with the indicated concentrations of OPD for 24–72 h. MRC5 cells were cultured in a 96-well plate and treated with the indicated concentrations of OPD for 72 h. The cell proliferative activity was determined using the SRB assay. The % cell proliferation value was calculated by the mean absorbance of samples/absorbance of the vehicle-treated control as described in the Materials and Methods. Data are presented as the means ± S.E. (*n* = 3).

The IC50 value of OPD with a 72 h treatment was 32.4 μM. In addition, the growth-inhibitory activity of OPD was also determined in a normal cell line. OPD was unable to affect the growth rate of MRC5 normal human lung fibroblast cells (IC50 >100 μM). These data suggest that OPD may be able to selectively inhibit the proliferation of human hepatoma cancer cells compared to normal cells. Under the same experimental conditions, the IC50 value of etoposide, a positive control, was 0.3 μM.

### *2.2. E*ff*ects of OPD on the Cell Cycle Distribution of SK-Hep-1 Cells*

To further elucidate the anti-proliferative mechanisms of OPD in SK-Hep-1 cells, the cells were treated with the indicated concentrations of OPD for 24 h, and flow cytometry analysis was performed with PI staining. As shown in Figure 3A, OPD enhanced the accumulation of the G2/M phase peak from 22.66% (control) to 35.90% (75 μM). These data suggest that the antiproliferative activity of OPD in SK-Hep-1 cells is in part associated with the induction of G2/M phase cell cycle arrest. To further investigate whether the G2/M phase cell cycle arrest by OPD is correlated with the regulation of the checkpoint proteins, the expression of the G2/M cell cycle regulatory proteins was determined by western blot analysis. Since OPD did not show significant cytotoxicity at the test concentration up to 100 μM for 24 h (Figure 2), the cells were treated with OPD (50, 75, or 100 μM) for 24 h, and then the checkpoint protein expression related to G2/M phase cell cycle regulation was measured in SK-Hep-1 cells. As shown in Figure 3B, the expression levels of Chk1, p-cdc25c (Ser198), cdc25c, cyclin B1, cdc2, and p-cdc2 (Thr161) were downregulated, but the levels of p-Chk1 (Ser345) were upregulated by OPD treatment. Chk1 (checkpoint kinase 1) is a multifunctional protein kinase that coordinates the response to specific types of DNA damage [16]. Cdc25 is a protein phosphatase responsible for dephosphorylating and activating cdc2, a pivotal step in directing the cells toward mitosis [17]. When DNA damage ocurrs, the Chk1 phosphorylates cdc25c, which then leads to translocation of cdc25c from the cytoplasm to the nucleus, where cdc25c can interact with cdc2/cyclin B during mitosis [18,19]. Moreover, the activity of the cdc2-cyclin B1 complex is dependent on the phosphorylation/dephosphorylation status of cdc2 [11,13,20]. The entry of eukaryotic cells into mitosis is regulated by cdc2 activation, including the binding of cdc2 to cyclin B1 and its phosphorylation at the Thr161 residue. In this study, we found that cdc25c was inactivated by phosphor-Chk1 with OPD treatment, and the activation of the cdc2-cyclin B1 complex was also suppressed by OPD in a concentration-dependent manner, indicating the induction of G2/M phase cell cycle arrest by OPD. These findings suggest that the activation of Chk1 and sequential regulation of signal transduction pathways by OPD may be due to the induction of G2/M phase cell cycle arrest by OPD in SK-Hep-1 cells.

**Figure 3.** *Cont.*

**Figure 3.** Effects of OPD on the regulation of cell cycle distribution in SK-Hep-1 cells. (**A**) SK-Hep-1 cells were treated with various concentrations of OPD for 24 h. Both adherent and floating cells were collected, fixed with 70% cold ethanol overnight, and then incubated with RNase A and PI for 30 min. The cell cycle distribution was analyzed by flow cytometry. (**B**) The cells were treated with the indicated concentrations of OPD for 24 h, and the expression of cell cycle regulatory proteins was determined by western blot analysis. β-Actin was used as an internal control.

### *2.3. E*ff*ects of OPD on the Regulation of p53-associated Signaling Pathways in SK-Hep-1 Cells*

To further validate the association of p53-mediated signaling molecules in the anti-proliferative activity of OPD in SK-Hep-1 cells, the cells was treated with OPD for 24 h, and then the expression of p53-associated molecules were determined by western blot analysis. OPD significantly enhanced the protein expression of p53, p-p53(Ser15), MDM2, p21, and GADD45α in the cells (Figure 4A). Previous studies have reported that the phosphorylation of p53 at Ser15 or Ser20 upregulates p53 stability by disrupting the interaction between p53 and MDM2 [21,22]. Because OPD is able to induce the stability of p53, the accumulation of p53 by OPD may subsequently enhance its downstream target genes, including MDM2, p21, and GADD45α, in the cells. In addition, the expression of p53-associated proteins was also monitored for up to 48 h with treatment with 50 μM OPD and was found to be up-regulated after 4 h of treatment in SK-Hep-1 cells (Figure 4B). Further study was designed to confirm whether the cellular localization of p53 is affected by treatment of cells with OPD. Generally, in the nucleus, MDM2-mediated ubiquitination led to the transportation of p53 into the cytoplasm or its degradation by the 26S proteasome [21,22]. As an important transcriptional factor, the stability and nuclear localization of p53 are considered essential for its tumor suppressor activity [23]. Therefore, we determined the levels of p53 and MDM2 protein expression in both the cytosol and nucleus fraction following treatment with OPD in SK-Hep-1 cells. As shown in Figure 4C, the levels of p53 protein expression were upregulated and localized in the nucleus upon OPD treatment for 24 h in a concentration-dependent manner. The levels of MDM2 protein expression were also shown to be upregulated in the nucleus by OPD treatment in the cells. To further confirm the effects of OPD on the expression of p53-associated proteins, OPD treatment of cells was conducted in the absence or presence of caffeine. Because caffeine is known as an ATR/Chk1 kinase inhibitor [24,25], the effect of caffeine on the enhanced p53-associated protein expression by OPD was investigated in the cells. As expected, treatment with OPD (50 μM) for 24 h led to a significant upregulation of p53, p21, and MDM2 expression in the absence of caffeine, but the pretreatment with caffeine for 2 h alleviated the enhancement of p53-associated protein expression by OPD in SK-Hep-1 cells (Figure 4D). Although the expression of p53-associated proteins was downregulated by the pretreatment with caffeine, OPD also exhibited a slight upregulation of p53 and p21 expression in the presence of caffeine (Figure 4D). These data suggest that the induction of G2/M phase cell cycle arrest by OPD seems to be associated with not only the ATR/Chk1 signaling pathway but also the regulation of p53-mediated pathways. Further study examined the maintenance of p53 stability by OPD with the cotreatment of cycloheximide (CHX), a protein synthesis inhibitor [26], in SK-Hep-1 cells. As shown in Figure 4E, OPD significantly suppressed the degradation of p53 in the presence of cycloheximide, suggesting that the half-life of p53 degradation was sustained by OPD treatment compared to vehicle-treated control cells. However, the rate of MDM2 degradation was not greatly affected by the down-regulation of p53 turnover with OPD treatment of cells.

**Figure 4.** *Cont.*

**Figure 4.** Effects of OPD on the regulation of p53-associated signaling pathways. (**A**) SK-Hep-1 cells were treated with the indicated concentrations of OPD for 24 h. The expression level of proteins was analyzed by western blot. β-Actin was used as an internal control. (**B**) SK-Hep-1 cells were treated with 50 μM OPD for various time points up to 48 h. The expression level of proteins was analyzed by western blot. β-Actin was used as an internal control. (**C**) SK-Hep-1 cells were treated with various concentrations of OPD for 24 h. After cells were harvested, the proteins were separated into cytosolic and nuclear fractions as described in the Materials and Methods. The protein expression levels were analyzed by western blot. β-Actin was used as an internal control. (**D**) SK-Hep-1 cells were pretreated for 2 h with 1 mM caffeine, and then cultured with 50 μM OPD for an additional 24 h. The expression level of proteins was analyzed by western blot. β-Actin was used as an internal control. (**E**) SK-Hep-1 cells were treated with 50 μM OPD for 24 h and then treated in the absence or presence of 20 μg/mL cycloheximide. The cells were harvested at 0, 20, 40, and 60 min. The expression level of proteins was analyzed by western blot, and the band density was quantified using the NIH ImageJ software (Bethesda, MD).

### *2.4. E*ff*ects of OPD on Cell Proliferation Depending on the p53 Status in Hepatoma Cells*

Since p53 seems to be highly associated with the anti-proliferative activity of OPD in SK-Hep-1 cells, the effects of p53 status on the growth-inhibitory activity of OPD were evaluated in human hepatoma cells with different p53 statuses. The cells were treated with various concentrations of OPD for 72 h, and then cell proliferation was determined by the SRB assay. As shown in Figure 5A, wild-type p53 cell lines, such as SK-Hep-1 and HepG2, were more sensitive than the p53-null Hep3B cell line, indicating that p53 plays an important role in the anti-proliferative activity of OPD in human hepatoma cells. To further confirm the involvement of p53 in the antiproliferative activity of OPD, p53 siRNA was transfected into wild-type p53 SK-Hep-1 cells, and the effect of p53 siRNA on the cell proliferation was determined in SK-Hep-1 cells. When the p53 siRNA transfected-SK-Hep-1 cells were treated with OPD (50 μM) for 48 and 72 h, the antiproliferative activity of OPD was decreased compared to that of the control cells (no transfection or scrambled siRNA transfection), suggesting that the anti-proliferative activity of OPD is partly associated with wild-type p53 expression in hepatoma cells (Figure 5B). In addition, the OPD treatment exhibited upregulation of p53 and MDM2 in the

control cells. Although p53 expression was not shown in p53 siRNA-treated SK-Hep-1 cells, OPD slightly up-regulated the expression of MDM2 in p53 siRNA-treated SK-Hep-1 cells (Figure 5C).

**Figure 5.** Effects of p53 expression on the cell proliferation of hepatoma cells by OPD. (**A**) SK-Hep-1, HepG2, and Hep3B cells were cultured in a 96-well plate and treated with the indicated concentrations of OPD for 72 h. The cell proliferative effect was determined using the SRB assay. Data are presented as the means ± S.E. (*n* = 3). (**B**) SK-Hep-1 cells were transfected with 5 nM p53 siRNA or scrambled siRNA for 24 h, and then OPD treatment (50 μM) was given for 48 or 72 h for the measurement of cell proliferation. Data are presented as the means ± S.E. (*n* = 3). (**C**) The effects of p53 siRNA on the expression level of p53 and MDM2 were analyzed by western blot in control cells and p53 siRNA-transfected SK-Hep-1 cells.

### *2.5. E*ff*ects of OPD in Combination with Gemcitabine on the Cell Proliferation of SK-Hep-1 Cells*

To further investigate the effect of the combination of OPD with an anticancer agent on the antiproliferative activity of hepatoma cells, we selected gemcitabine, a pyrimidine-based DNA synthesis inhibitor, as a candidate compound based on the use of patients with liver cancer in the clinic. As shown in Figure 6A, cotreatment with OPD and gemcitabine exhibited a synergistic effect (CI < 1.0) on the antiproliferative activity of SK-Hep-1 cells. The combination of the higher concentrations of OPD and gemcitabine showed a relatively stronger synergistic activity. In accordance, when OPD (20 μM) was treated with various concentrations of gemcitabine (0.1–0.4 μM), the combination with the higher concentration of gemcitabine (0.4 μM) was shown to have a stronger synergistic activity (Figure 6).


**Figure 6.** Combination effect of OPD with gemcitabine on the cell proliferation of SK-Hep-1 cells. (**A**) The cells were treated with the indicated concentrations of OPD and gemcitabine for 72 h, and then cell proliferation was measured by the SRB assay. (**B**) The cells were treated with OPD (20 μM) and various concentrations of gemcitabine (0.1–0.4 μM) for 72 h, and then cell proliferation was measured by the SRB assay. Data are presented as the means ± S.E. (*n* = 3). Based on the cell proliferation data.

Natural products have been used for traditional medicines and also play an important role in drug discovery and development programs. In particular, anticancer drugs are mainly developed based on natural product-originated small molecules. In our continuous efforts to procure small molecules from natural sources in the discovery of antitumor agents, oxypeucedanin (OPD), a furanocoumarin isolated from the root of *Angelica dahurica,* was considered a potential candidate. Previous biological activities of OPD include the effective intervention of sunitinib-induced nephrotoxicity, the hepatoprotective activity of tacrine-induced cytotoxicity in liver cells, and the cytotoxicity against cancer cells such as gastric cancer, prostate cancer, and melanoma cells [27,28]. Although the antiproliferative activities of OPD have been reported in cancer cells, the precise molecular mechanism of OPD in the anticancer activity of human liver cancers remains to be elucidated. This study demonstrates that OPD regulates the induction of G2/M phase cell cycle arrest and p53-dependent MDM2/p21 signaling pathways in human hepatoma SK-Hep-1 cells.

It is known that the cell cycle is a finely controlled sequence of events in the growth and proliferation of eukaryotic cells. Cell cycle progression occurs in an ordered manner that is monitored by cell cycle checkpoints. Among the checkpoints, the DNA damage-induced G2/M checkpoint guarantees the fidelity of genomic stability. However, cancer cells have abnormally activated in cell division as a result of diverse factors, including uncontrolled checkpoint protein expression. These defects in the checkpoints lead to genomic instability, cell death, or carcinogenesis. Therefore, the regulation of the G2/M checkpoint is often applied as a potential therapeutic target to evaluate the efficacy of natural product-derived antitumor agents in cancer cells. In the present study, we found that OPD effectively inhibited the growth of SK-Hep-1 human hepatoma cells. Cell cycle distribution analysis also revealed that the antiproliferative activity of OPD is in part associated with the induction of G2/M cell cycle arrest in cells. This result was consistent with the previous report of G2/M phase arrest of OPD in prostate cancer cells [27]. However, a previous study of the antiproliferative activity of OPD on the prostate cancer cells did not investigate in detail the regulatory biomarkers involved in G2/M phase cell cycle arrest in cancer cells. The present study revealed that the induction of G2/M phase cell cycle arrest by OPD was correlated with the regulation of Chk1-mediated G2/M checkpoint proteins, including cdc25c/cdc2 and cdc2/cyclin B1 complex pathways, in hepatoma cells. Chk1 is considered to play an important role in the G2/M checkpoint via the ATM-RAD3-related (ATR) pathway. Chk1 also regulates the activity of its shared downstream substrate, cell division cycle 25c (cdc25c). OPD effectively modulates the Chk1-cdc25c activation pathway axis, leading to the induction of G2/M cell cycle arrest in SK-Hep-1 cells. We also found that OPD effectively suppressed the activity of cyclin B1 and its partner cell division cycle 2 (cdc2) expression, which in turn evokes the prevention of entry into the mitosis (M) phase cell cycle transition in hepatoma cells.

Tumor suppressor p53 is mutated, deleted, or rearranged in more than half of all human tumors, and thus, p53 is considered an important target in anticancer drug development [29,30]. p53 is also considered an important factor in the control of the cell cycle at the G1/S and/or G2/M transition through diverse mechanisms [31]. In this study, we found that OPD significantly up-regulated the expression of p53 by the accumulation of p53 and increased its stability in SK-Hep-1 cells. The upregulation of p53 levels by OPD led to the activation of downstream target expression such as MDM2, p21 and GADD45α in cells. The association of p53 in G2/M phase cell cycle signaling was further confirmed by the significant suppression of p53, p21 and MDM2 expression by treatment with caffeine, an ATR/Chk1 inhibitor, in SK-Hep-1 cells. However, OPD did not fully recover the levels of p53 and its downstream target protein expression in the presence of caffeine, indicating that OPD may cause the activation of p53 independently of the ATR/Chk1 pathway. The involvement of p53 in the anti-proliferative activity of OPD in SK-Hep-1 cells was also confirmed using cell types with different p53 statuses and deletion methods. OPD was found to be more effective in p53-wild-type expression cells, and p53 siRNA-transfected cells were less sensitive to the OPD in the antiproliferative activity on liver cancer cells.

The function of p53 is mainly governed by the stability and nuclear localization of p53 in cells. We found that OPD effectively inhibits the degradation of p53 and thus sustains the stability of p53 in cancer cells. In addition, OPD was found to induce the nuclear localization of p53, which may activate the transcriptional activity of its downstream target genes associated with cell cycle regulation in cancer cells.

Gemcitabine (2 ,2 -difluorodeoxycytidine), a pyrimidine-based antimetabolite, is currently used to the treatment of pancreatic cancer as well as other cancers [32,33]. Gemcitabine inhibits DNA replication by activating the S-phase checkpoint. A study also showed that gemcitabine activates the ATR/Chk1 pathway [34]. In the present study, we found that OPD in combination with gemcitabine exhibits synergistic anti-proliferative activity in SK-Hep-1 cells. Therefore, OPD might be a promising lead compound in combination with cancer chemotherapeutic agents in advanced liver cancer treatment. In addition, in terms of clinical relevance of furanocoumarins, the bioavailability of oxypeucedanin in plasma samples may be important parameters. In the previous study, Chen et al. [35] reported that the pharmacokinetic parameters of OPD were the Tmax (12 h), T1/2 (2.4 h), and oral bioavailability (F, 10.1%) in rat models by oral administration of OPD. These data suggest that the oral bioavailability of

OPD is a relatively low and thus needed to be further study for improvement of the bioavailability either structural modifications or appropriate formulation of administration of OPD.

### **3. Materials and Methods**

### *3.1. Chemicals*

Dulbecco's modified Eagle's medium (DMEM) and fetal bovine serum (FBS) were purchased from HyClone Laboratories (Logan, UT, USA). Antibiotics-antimycotics solution, Lipofectamine®RNAiMAX, negative control siRNA, and Opti-MEM®Reduced Serum Medium were purchased from Invitrogen (Grand Island, NY, USA). p53-specific siRNA was purchased from Bioneer (Daejeon, Korea). The nuclear extract kit was purchased from Active Motif (Carlsbad, CA, USA). Bovine serum albumin (BSA), dimethyl sulfoxide (DMSO), trichloroacetic acid (TCA), sulforhodamine B (SRB), propidium iodide (PI), ribonuclease A (RNase A), and other agents were purchased from Sigma-Aldrich (St. Louis, MO, CA). Antibodies against p-chk1 (Ser345) (#2348), chk1 (#2360), p-cdc25c (Ser198) (#9529), cdc25c (#4688), p-cdc2 (Tyr15) (#9111), p-cdc2 (Thr161) (#9114), p-p53 (Ser15) (#9284), p21 (#3733), and α/β tubulin (#2148) were purchased from Cell Signaling (Danvers, MA, USA). Antibodies against cyclin B1 (#752), cdc-2 (#54), β-actin, p53 (#126), MDM2 (#965), GADD45α (#797), lamin B1 (#20682) were obtained from Santa Cruz Biotechnology (Dallas, TX, USA)

OPD (Figure 1) and its analogs, isolated from the root of *Angelica dahurica*, were provided by Dr. Jin-Woong Kim (College of Pharmacy, Seoul National University, Seoul, Korea). *Angelica dahurica* (9 kg, dry weight) was purchased from Kyungdong Market Herbal Medicine in Seoul, and ultrasonically extracted three times for 120 min with 100% MeOH. The extract was filtered and concentrated under reduced pressure to obtain MeOH extract (910 g), which was suspended in distilled water to obtain a CHCl3 fraction (229 g). OPD and its analogs were obtained by various column chromatography. Each component was analyzed by 1H-NMR, 13C-NMR, and FABMS, and then identified compared with the literature values of corresponding compounds. OPD and its analogs, dissolved in 100% DMSO.

### *3.2. Cell Culture*

SK-Hep-1, HepG2, and Hep3B cells were purchased from the American Type Culture Collection (Manassas, VA, USA). Cells were cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS) and antibiotics-antimycotics (PSF; 100 units/mL penicillin G sodium, 100 μg/mL streptomycin, and 250 ng/mL amphotericin B). Cells were incubated in a humidified atmosphere containing 5% CO2 at 37 ◦C.

### *3.3. Cell Proliferation Assay*

Cell proliferation was measured by the sulforhodamine B (SRB) assay [36]. Cells were seeded in 96-well plates (3 <sup>×</sup> 104 cells/mL), incubated for 24 h, and either fixed (for zero day controls) or treated with various concentrations of test compounds (total volume of 200 μL/well) for 24, 48, and 72 h. After treatment, the cells were fixed with 50% TCA solution and dried at room temperature. Fixed cells were stained in 0.4% SRB in 1% acetic acid, and unbound dye was washed with 1% acetic acid. Stained cells were dried and dissolved in 10 mM Tris (pH 10.0). The absorbance was measured at 515 nm, and cell proliferation was determined as follows: cell proliferation (%) = (average absorbance compound – average absorbance zero day) / (average absorbance control – average absorbance zero day) × 100. IC50 values were calculated by non-linear regression analysis using the TableCurve 2D v5.01 software (Systat Software Inc., San Jose, CA, USA).

### *3.4. Cell Cycle Analysis*

SK-Hep-1 cells were plated at a density of 1 <sup>×</sup> 10<sup>6</sup> cells per 100-mm culture dish and incubated for 24 h. Fresh media containing various concentrations of test sample were added to the culture dishes. Following a 24 h incubation, the cells were harvested (trypsinization and centrifugation) and fixed with 70% ethanol overnight at 4 ◦C. Fixed cells were washed with PBS and incubated with a staining solution containing RNase A (50 μg/mL) and propidium iodide (50 μg/mL) in PBS for 30 min at room temperature. The cellular DNA content was analyzed with a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA, USA). At least 20,000 cells were used for each analysis, and the distribution of cells in each phase of the cell cycle was displayed as histograms.

### *3.5. Western Blot Analysis*

SK-Hep-1 human hepatoma cells were exposed to various concentrations of OPD for the indicated times. After incubation, the cells were lysed, and the protein concentrations were determined by the bicinchoninic acid method [37]. Each protein was subjected to 6–15% SDS-PAGE. Proteins were transferred onto PVDF membranes (Millipore, Bedford, MA, USA) by electroblotting, and membranes were blocked for 1 h with blocking buffer [5% bovine serum albumin (BSA) in tris-buffered saline-0.1% Tween 20 (TBST)] at room temperature [38]. Membranes were then incubated with indicated antibodies (mouse anti-β-actin, diluted 1:10,000; other antibodies, diluted 1:500–1:1000 in 5% BSA/TBST) overnight at 4 ◦C and washed three times for 10 min with TBST. After washing, membranes were incubated with corresponding secondary antibodies diluted 1:2000 in TBST for 2 h at room temperature, washed three times for 10 min with TBST, and visualized with an enhanced chemiluminescence (ECL) detection kit (LabFrontier, Suwon, Korea) using an LAS-4000 Imager (Fuji Film Corp., Tokyo, Japan).

### *3.6. RNA Interference*

RNA interference of p53 was performed using siRNA duplexes purchased from Bioneer (Daejeon, Korea). The coding strand for p53 siRNA was as follows: sense CACUACAACUACA UGUGUA and antisense UACACAUGUAGUUGUAGUG. For transfection, reverse transfection was conducted using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's recommendations. Compound treatments occurred 24 h after transfection. Cells were harvested after 24 h and examined by western blotting.

### *3.7. Combination Assay*

Determination of the effect of combination therapy was performed using the SRB assay. On day 1, 3000 cells/well in a volume of 100 μL were plated in 96-well plates. On day 2, gemcitabine (100, 200, or 400 nM) and OPD (20, 30, or 40 μM) were each added in a volume of 50 μL, in all combinations. After 72 h, the cells were fixed with 50% TCA solution for 30 min at 4 ◦C, rinsed 5 times with water, and air-dried. Fixed cells were colored with 80 μL of 0.4% sulforhodamine B in 0.1% acetic acid) rinsed with 0.1% acetic acid, and air dried. Sulforhodamine was redissolved in 200 μL/well of 10 mM Tris, pH 10, and the absorbance was measured at 515 nm. After calculating the percent of inhibition by OPD and gemcitabine, the combination index (CI) was estimated to evaluate the synergistic effect of OPD and gemcitabine:

$$\text{CI} = \frac{(D)\_1}{(D\_m)\_1 [f\_a/(1-f\_a)]^{\frac{1}{m\_1}}} + \frac{(D)\_2}{(D\_m)\_2 [f\_a/(1-f\_a)]^{\frac{1}{m\_2}}}$$

where D = dose; Dm = median-effect dose; m = kinetic order; fa = fraction affected.

The combination index (CI) is a quantitative measure based on the mass-action law of the degree of drug interaction in terms of synergism and antagonism (Table 3) for a given endpoint of the effect measurement.


**Table 3.** Description and symbols of synergism or antagonism in drug combination studies analyzed with the combination index method.

### *3.8. Statistical Analysis*

All experiments were repeated at least three times. Data are presented as the means ± standard error (SE) for the indicated number of independently performed experiments and analyzed using Student's *t*-test. Values of *p* < 0.05 were considered statistically significant.

### **4. Conclusions**

In summary, the present study demonstrates the antiproliferative activities of OPD on SK-Hep-1 human hepatoma cells. The inhibition of proliferation of cancer cells was in part associated with cell cycle arrest at the G2/M phase and the tumor suppressor p53-mediated signaling pahway (Figure 7). These findings suggest that OPD is a promising new chemotherapeutic candidate for the management of human hepatoma cell treatment.

**Figure 7.** Schematic representation of the mechanisms of action of OPD against SK-Hep-1 human hepatoma cells.

**Author Contributions:** Conceptualization, S.K.L.; Methodology, S.H.P. and J.-Y.H.; Validation, S.H.P.; Formal Analysis, S.H.P.; Investigation, S.H.P.; Resources, S.K.L.; Data Curation, S.H.P.; Writing-Original Draft Preparation, S.H.P.; Writing-Review & Editing, J.-Y.H., H.J.P. and S.K.L.; Visualization, S.H.P.; Supervision, S.K.L.; Project Administration, S.K.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by a National Research Foundation of Korea (NRF) Grant funded by the Korean Government (NRF-2016M3A9B6903499) and the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2019 R1I1A1A01060558).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Ca**ff**eic Acid Attenuates Multi-Drug Resistance in Cancer Cells by Inhibiting E**ffl**ux Function of Human P-Glycoprotein**

### **Yu-Ning Teng 1, Charles C.N. Wang 2, Wei-Chieh Liao 3, Yu-Hsuan Lan 3,\* and Chin-Chuan Hung 3,4,\***


Academic Editor: Kyoko Nakagawa-Goto Received: 14 November 2019; Accepted: 6 January 2020; Published: 7 January 2020

**Abstract:** Multidrug resistance (MDR) is a complicated ever-changing problem in cancer treatment, and P-glycoprotein (P-gp), a drug efflux pump, is regarded as the major cause. In the way of developing P-gp inhibitors, natural products such as phenolic acids have gotten a lot of attention recently. The aim of the present study was to investigate the modulating effects and mechanisms of caffeic acid on human P-gp, as well as the attenuating ability on cancer MDR. Calcein-AM, rhodamine123, and doxorubicin were used to analyze the interaction between caffeic acid and P-gp, and the ATPase activity of P-gp was evaluated as well. Resistance reversing effects were revealed by SRB and cell cycle assay. The results indicated that caffeic acid uncompetitively inhibited rhodamine123 efflux and competitively inhibited doxorubicin efflux. In terms of P-gp ATPase activity, caffeic acid exhibited stimulation in both basal and verapamil-stimulated activity. The combination of chemo drugs and caffeic acid resulted in decreased IC50 in *ABCB1*/Flp-InTM-293 and KB/VIN, indicating that the resistance was reversed. Results of molecular docking suggested that caffeic acid bound to P-gp through GLU74 and TRY117 residues. The present study demonstrated that caffeic acid is a promising candidate for P-gp inhibition and cancer MDR attenuation.

**Keywords:** caffeic acid; cancer multidrug resistance; P-glycoprotein; phenolic acid

### **1. Introduction**

As cancer is one of the leading causes of death worldwide, cancer treatment is always on the top of listed hot research topics [1]. With advanced scientific researches and abundant medical resources in recent decades, diverse options have been developed to conquer cancer and related diseases. Nevertheless, the multi-drug resistance (MDR) of cancer treatment is still an ever-changing problem and more in-depth studies have been conducted to unveil the complicated characteristics of cancer treatment. Cancer MDR manifests cross resistance to several structurally and mechanically different chemo-agents and could be contributed to the following reasons [2]. The change in tumor microenvironment [3–5], decreased drug uptake [6], adapted cell apoptotic pathways [7–9], drug inactivation through metabolism [10,11], the influence of epigenetic regulation [12,13], mutation of drug target site [14], and the increased drug efflux [15] have been reported to play important roles in

causing cancer MDR. Among the above mechanisms, the increased drug efflux by ATP-binding cassette (ABC) transporters has been regarded as the most influential cause. ABC transporter superfamily consists of several subfamilies, and P-glycoprotein (P-gp) is one of the most comprehensively studied proteins [16]. P-gp is encoded by human *ABCB1* gene and can recognize various clinically used drugs, including antidepressants, HIV protease inhibitors, immunosuppressive agents, and chemotherapeutic drugs [17,18]. The diverse structures recognizing and effluxing the ability of P-gp, result in insufficient chemo-drug concentration inside cancer cells, therefore, causing cancer MDR.

There have been a series of P-gp inhibitor developments along the cancer MDR reversing agents discovering history, and the improvements have been based on previous failure experiences [19]. First generation P-gp inhibitors are potent but toxic as the required dose is high; examples of this are quinidine and verapamil [20]. Second generation inhibitors have exhibited better effects with lower IC50, but the involvement of these inhibitors in CYP450 interaction has impeded their further application [21,22]. Third generation inhibitors, including tariquidar and zosuquidar, have demonstrated prominent MDR reversal effects. However, they have still faced failure in clinical studies [23,24]. Therefore, severe toxicity and interaction of the above chemical reagents have turned the research direction toward natural resources, aiming at discovering low toxic and potent structures from plants, fungi, or marine organisms.

Among various natural resources, phytochemicals such as flavonoids and phenolic acids get much attention due to their multiple pharmacological effects, including antioxidant and antitumor activity [25,26]. Several phytochemicals, such as cyanidin, catechin, quercetin, caffeic acid, and ellagic acid, have been related to the down-regulation of human LDL oxidation [27]. Ellagic acid and ursolic acid have been reported to exhibit preventive and therapeutic effects against breast cancer cells [28]. Caffeic acid (Figure 1), a phenolic acid that widely exists in vegetables, fruits, and tea extracts, is well-known as a natural antioxidant [29]. Besides, caffeic acid has also been identified to have anti-inflammatory, antibacterial, and antiviral effects [30,31]. With regards to cancer treatment, caffeic acid and its derivative, caffeic acid phenethyl ester (CAPE), exhibit some therapeutic effects toward lung cancer and breast cancer cells, as well as breast cancer pre-clinical models [32–34]. CAPE has been well studied in previous researches, including its *MDR1* gene down-regulating effects in MCF-7 and MDA-MB-231 breast cancer cells [34] and P-gp inhibitory effects in HeLa resistant cancer subline and human intestinal LS174T cell line [35,36]. Nevertheless, the P-gp inhibitory and MDR modulating information of the caffeic acid was insufficient and warrants further detailed investigation.

**Figure 1.** The chemical structure of caffeic acid.

Therefore, in the present study, comprehensive researches of caffeic acid were conducted. The interaction of caffeic acid with human P-gp, as well as the inhibitory effects and mechanisms were assessed in P-gp over-expressing cell line *ABCB1*/Flp-InTM-293. The cancer MDR reversing ability of caffeic acid was then evaluated in both *ABCB1*/Flp-InTM-293 and KB/VIN MDR cancer cell lines. The present study demonstrated that caffeic acid is a promising candidate for P-gp inhibition and cancer MDR attenuation.

### **2. Results**

### *2.1. Ca*ff*eic Acid Is Non-Cytotoxic toward Experimental Cell Lines and Is Not a Substrate of P-gp*

Before conducting further experiments, the cytotoxicity of caffeic acid was examined in HeLaS3, KB/VIN, Flp-InTM-293, and *ABCB1*/Flp-InTM-293 cell lines to select a rational concentration range. Caffeic acid exhibited higher than 80% cell viability in all tested cell lines under the treatment of 100 μg/mL for 72 h. Hence, the following assays were conducted with caffeic acid of not more than 100 μg/mL.

The first characteristic of caffeic acid on P-gp was demonstrated through MDR1 shift assay, which revealed whether a compound is a substrate of P-gp. P-gp's substrates activate a conformational change detected by the structure-sensitive UIC2 antibody. As Figure 2 showed, the fluorescent peaks of caffeic acid 20 and 25 μg/mL did not shift to the right as the positive control vinblastine did, indicating that the conformation of P-gp was not influenced by caffeic acid. Therefore, caffeic acid is not P-gp's substrate.

**Figure 2.** The result of MDR1 shift assay. The conformation of P-gp was not influenced under the treatment of 20 and 25 μg/mL caffeic acid. Vinblastine (a standard substrate of P-gp) was used as a positive control.

### *2.2. The Inhibitory E*ff*ects, Mechanisms and Binding Modes of Ca*ff*eic Acid on Human P-gp Function*

The inhibitory effect of caffeic acid on P-gp function was screened with calcein-AM accumulation assay. Calcein-AM is a non-florescent substance and P-gp's substrate. It will be transformed to fluorescent calcein (not a P-gp substrate) by cell esterase. Therefore, under the treatment of P-gp's inhibitor, the intracellular calcein fluorescence is higher than the normal condition. The results of caffeic acid are revealed in Figure 3a. When *ABCB1*/Flp-InTM-293 cell line was treated with caffeic acid in amounts of 5, 10, and 20 μg/mL, the intracellular calcein fluorescence was increased in a concentration-dependent manner. Hence, the efflux function of P-gp could be inhibited by caffeic acid.

Caffeic acid's inhibitory effects and mechanisms were further demonstrated via the other two substrates of P-gp, rhodamine123 and doxorubicin. As Figure 3b showed, the efflux of fluorescent substrate rhodamine123 was inhibited by caffeic acid 10 and 20 μg/mL treatment and followed Michaelis-Menten kinetics. The Lineweaver-Burk plot (Figure 3c) indicated that caffeic acid inhibited rhodamine123 efflux in an uncompetitive pattern, both Vmax and Km decreased when the *ABCB1*/Flp-InTM-293 cell line was treated with increased caffeic acid concentrations (Table 1). Same as rhodamine123, the efflux of the chemotherapeutic drug doxorubicin was also inhibited by caffeic acid dose-dependently (Figure 3d). However, the inhibitory mechanism of caffeic acid on doxorubicin was competitive inhibition, different from rodamine123 (Figure 3e). When the *ABCB1*/Flp-InTM-293 cell line was treated with an increased concentration of caffeic acid, the Km (affinity) increased accordingly and the Vmax remained constant (Table 1).

**Figure 3.** *Cont*.

**Figure 3.** The effects of caffeic acid on human P-gp efflux function. (**a**) Intracellular calcein fluorescence with or without caffeic acid pretreatment in *ABCB1*/Flp-InTM-293 cell line (over-expressing human P-gp). (**b**) Michaelis-Menten kinetics of rhodamine123 efflux with or without caffeic acid pretreatment in *ABCB1*/Flp-InTM-293. (**c**) Lineweaver-Burk plot analyses of caffeic acid on the transport of rhodamine123 in human P-gp. (**d**) Michaelis-Menten kinetics of doxorubicin efflux with or without caffeic acid pretreatment in *ABCB1*/Flp-InTM-293. (**e**) Lineweaver-Burk plot analyses of caffeic acid on the transport of doxorubicin in human P-gp. \* denotes *p* < 0.05 compared with the control group. Data were presented as mean ± SE of at least three experiments, each in triplicate. (**f**) Molecular docking analysis of caffeic acid (PubChem CID: 689043) docked posed of compounds in the P-gp (PDB entry 6QEX) binding pocket of 3D structure.


**Table 1.** The effects of caffeic acid on the transport of rhodamine123 and doxorubicin by human P-gp.

\* *p* < 0.05 as compared to the rhodamine123 or doxorubicin transport without caffeic acid.

Ki 14.13 ± 0.005

+ Caffeic acid, 20 μg/mL 107.52 ± 0.001 426.16 ± 0.00 \*

In order to investigate the supposed binding pattern and possible interaction between the ligand of caffeic acid and pocket of P-gp, the ligand of caffeic acid was virtually docked to the crystal structures of the ligand-binding domain of P-gp using the docking program CDOCKER. The virtual binding result is shown in Figure 3f. The docking results showed that caffeic acid had the the best binding energies active site of P-gp with a -CDOCKER energy score of 20.1292, and binding energy was 44.4058 Kcal/mol. The binding model clearly indicated that the caffeic acid bound to P-gp with residues GLU74 and TRY117. Our docking results further demonstrated the binding behavior between P-gp and caffeic acid, providing insight into the design of novel P-gp modulators.

The interaction between caffeic acid and ATP binding site of P-gp was carried out with Pgp-GloTM Assay System. As Figure 4a shows, when the P-gp membrane was treated with caffeic acid with amounts of 1, 10, and 50 μg/mL, the basal P-gp ATPase activity was inhibited. On the other hand, the ATPase activity was stimulated under the treatment of 100 μg/mL caffeic acid. Besides, when combining caffeic acid with 200 μM verapamil, the elevated ATPase activity produced by verapamil was further stimulated and especially high under 100 μg/mL concentration (Figure 4b).

**Figure 4.** The P-gp ATPase modulating effects of caffeic acid. (**a**) Caffeic acid stimulated ATPase activity dose-dependently (10–100 μg/mL). (**b**) The verapamil-stimulated ATPase activity was further stimulated by caffeic acid. Data were analyzed in terms of the change of luminescence (ΔRLU). Data were presented as mean ± SE of at least three experiments, each in triplicate.

(a)

### *2.3. The Influences of Ca*ff*eic Acid on Human P-gp Expression*

The modulatory ability of caffeic acid on *ABCB1* gene expression was performed in both *ABCB1*/Flp-InTM-293 and KB/VIN cell lines. In *ABCB1* overexpressing cell line *ABCB1*/Flp-InTM-293, caffeic acid after 72 h treatment slightly down-regulated the expression of P-gp (Figure 5a). Nevertheless, the same treatment for MDR cancer cell line KB/VIN exhibited the opposite phenomenon. Caffeic acid elevated *ABCB1* gene expression under 72 h treatment (Figure 5b).

**Figure 5.** The modulating effects of caffeic acid on human P-gp expression. (**a**) The *ABCB1* expression in *ABCB1*/Flp-InTM-293 was slightly down-regulated after treating the cells with 40 μg/mL caffeic acid for 72 h. (**b**) The *ABCB1* expression in KB/VIN was slightly up-regulated after treating the cells with 40 μg/mL caffeic acid for 72 h. (**c**) The intracellular ROS production under the treatment of caffeic acid with or without doxorubicin in HeLaS3. (**d**) The intracellular ROS production under the treatment of caffeic acid with or without doxorubicin in KB/VIN. Data were presented as mean ± SE of at least three experiments, each in triplicate. \* denotes *p* < 0.05 compared with the control group. \*\* denotes *p* < 0.05 compared with the doxorubicin 1 μM group in Figure 5c and doxorubicin 10 μM group in Figure 5d.

To study whether the regulation of *ABCB1* gene expression was related to the intracellular reactive oxygen species status, the intracellular total ROS activity assay was performed. As Figure 5c,d indicates, caffeic acid significantly decreased ROS production in HeLaS3 cell line and slightly decreased ROS production in KB/VIN cell line. When both cell lines were treated with chemotherapeutic drug doxorubicin and caffeic acid, the ROS production exhibited no difference compared to caffeic acid treatment alone. However, the doxorubicin-induced oxidative challenge was significantly reversed by caffeic acid in amounts of 10 μg/mL and 100 μg/mL in both HeLaS3 and KB/VIN cell lines.

### *2.4. The Modulating E*ff*ects of Ca*ff*eic Acid on Cancer Multi-Drug Resistance*

The MDR reversal ability of caffeic acid was examined in both P-gp over-expressing cell line *ABCB1*/Flp-InTM-293 and MDR cancer cell line KB/VIN. As Table 2 shows, 30 μg/mL caffeic acid reversed vincristine, paclitaxel, and doxorubicin resistance by 3.90, 4.96, and 15.11-fold, respectively. The IC50 of doxorubicin decreased from 9023.61 nM to 569.90 nM with the treatment of caffeic acid 30 μg/mL in *ABCB1*/Flp-InTM-293. The MDR reversal phenomenon was further approved and analyzed by cell cycle. Compared to paclitaxel alone treatment, the addition of caffeic acid 20 and 25 μg/mL significantly increased subG1 population (from 11.3% to 24.1% and 33.0%), indicating that the cell underwent obvious apoptosis under combinatorial treatment (Figure 6a and Table 3).


**Table 2.** The reversal effects of caffeic acid on chemotherapeutic drug resistance in P-gp over-expressing cell line *ABCB1*/Flp-InTM293.

\* *p* < 0.05 as compared to the chemotherapeutic drug treatment (vincristine, paclitaxel, or doxorubicin) without caffeic acid. The reversal fold (RF) was calculated by dividing the individual IC50 of chemotherapeutic drugs by the IC50 of chemotherapeutic drugs in the presence of caffeic acid.

**Table 3.** The percentage of each cell cycle phase under various treatments in *ABCB1*/Flp-InTM-293 cell line and KB/VIN cell line.


**Figure 6.** *Cont*.

**Figure 6.** The cytotoxicity-enhancing effects of combinatorial treatment with caffeic acid. (**a**) The cell cycle distribution of 72 h treatment in *ABCB1*/Flp-InTM-293 cell line. (**b**) The cell viability of KB/VIN under the treatment of chemotherapeutic drugs with or without caffeic acid. Data were presented as mean ± SE of at least three experiments, each in triplicate. \* *p* < 0.05 as compared to the chemotherapeutic drug treatment (doxorubicin, paclitaxel, or vincristine) without caffeic acid. (**c**) The cell cycle distribution of 72 h treatment in KB/VIN cell line.

In MDR cancer cell line KB/VIN, with 100 μg/mL caffeic acid treatment, the cytotoxicity of 100 nM doxorubicin, paclitaxel, and vincristine significantly increased. The cell viability decreased from nearly 100% to 67.91%, 61.18%, and 59.50% for doxorubicin, paclitaxel, and vincristine, respectively. In addition, the cytotoxic-enhancing ability of caffeic acid on chemotherapeutic drugs was in a dose-dependent manner (Figure 6b). However, the further cell cycle analyses showed that the combination of caffeic acid and paclitaxel did not prominently increase the apoptosis of KB/VIN cells, revealing distinct cell effects among *ABCB1*/Flp-InTM-293 and KB/VIN (Figure 6c and Table 3).

### **3. Discussion**

Caffeic acid, a dietary non-flavonoid phenolic compound, has been a popular candidate among several research fields. The present study has demonstrated its usability in cancer MDR. Caffeic acid can attenuate this severe resistant problem by inhibiting the efflux function of human P-gp. Through diverse modulating mechanisms, caffeic acid helps resistant cancer cells retain chemotherapeutic drugs inside their cells, promoting further apoptosis and cell death.

Through investigating the history of P-gp inhibitor development, the ideal characteristics of potential candidates have been revealed. The inhibitor itself is not a substrate of P-gp, but is one of the favorable properties [19]. Our present research performed an experiment and the results indicated that caffeic acid was not P-gp's substrate. In this way, more caffeic acid could stay inside the cells to help P-gp inhibition, resulting in a higher intracellular chemotherapeutic drugs concentration.

The inhibitory effects of caffeic acid on P-gp efflux function were demonstrated on three P-gp fluorescent substrates, calcein-AM, rhodamine123, and doxorubicin. The different binding modes of each substrate revealed the inhibitory mechanisms of caffeic acid on P-gp drug binding sites. A previous investigation found that doxorubicin was a R-site substrate while rhodamine123 exhibited both M and R sites binding affinity [37–39]. Our efflux assay results indicated that caffeic acid showed uncompetitive inhibition on rhodamine123 transport and competitive inhibition on doxorubicin transport. Therefore, caffeic acid might compete the R drug binding site with doxorubicin, resulting in efflux inhibition. In terms of rhodamine123 inhibition, caffeic acid exhibited an allosteric modulation on M site, indirectly prohibiting the pump out behavior of P-gp.

In addition to drug binding sites, the interaction between caffeic acid and ATP binding sites of P-gp was also studied. According to the tested compound's behavior toward P-gp ATPase regulation, substances could be categorized into three classes: dual regulators, stimulators, and inhibitors [40,41]. Dual regulators stimulate both basal and verapamil-stimulated ATPase activity at a lower dose, but inhibit the activity at a higher dose, such as paclitaxel and vinblastine. Stimulators like valinomycin and bisantrene increase ATPase activity dose-dependently while inhibitors decrease both basal and verapamil-stimulated ATPase activity, such as rapamycin and cyclosporine A. The results of ATPase assay in the present study indicated that caffeic acid exhibited stimulatory activity from 10 μg/mL to 100 μg/mL in a dose-dependent manner. Therefore, caffeic acid was an ATPase stimulator. Besides, the results of verapamil-stimulated ATPase activity further revealed the binding behavior of caffeic acid on ATPase binding sites. Caffeic acid increased verapamil-stimulated ATPase activity regardless of the dose, implying its binding site on ATPase was different from verapamil. This allosteric stimulation advanced the consumption of ATP, indirectly inhibiting P-gp efflux function.

Whether the promising P-gp inhibitory effects of caffeic acid were helpful in reversing cancer MDR was than studied in our following experiments. In *ABCB1*/Flp-InTM-293 P-gp over-expressing cell line, caffeic acid significantly decreased the required doses of chemo-agents, including vincristine, paclitaxel, and doxorubicin. Under the treatment of 30 μg/mL caffeic acid, the IC50 of paclitaxel largely decreased from 604.09 nM to 121.55 nM. This advanced cytotoxicity was related to the increased apoptotic effects revealed by cell cycle assay results. With caffeic acid as a combinatory agent, the percentage of subG1 apoptotic population induced by paclitaxel significantly increased in a dose-dependent manner. The above results were consistent with previous researches, which revealed that caffeic acid could sensitize ovarian carcinoma cells and lung cancer cells to cisplatin and paclitaxel, respectively [33,42]. Caffeic acid exhibited chemo-sensitizing effects in the combination group by cell cycle arresting in G2/M (caffeic acid 20 μg/mL with paclitaxel) and G1 (caffeic acid 25 μg/mL with paclitaxel). These effects were not only due to the modulation of P-gp, other cellular targets and multiple mechanistic possibilities may be involved and need further investigation. The MDR reversing ability of caffeic acid was also investigated in MDR cancer cell line KB/VIN. The results exhibited a trend on increasing the cytotoxicity of chemo-agents. With the treatment of 100 μg/mL caffeic acid, the cell viability decreased from nearly 100% to 67.91%, 61.18%, and 59.50% for doxorubicin, paclitaxel, and vincristine, respectively. However, compared to the promising results in *ABCB1*/Flp-InTM-293 cell line, the MDR modulating effects of caffeic acid in KB/VIN seemed to be less potent and did not show increased apoptosis in the cell cycle analyses, exhibiting cell type-dependent effects. This phenomenon could be explained by the regulation of caffeic acid on P-gp expression. As Figure 5a,b shows, caffeic acid slightly decreased *ABCB1* gene expression in *ABCB1*/Flp-InTM-293 but increased the expression level in KB/VIN cell line. This up-regulating trend in KB/VIN diminished the functional inhibitory potency of caffeic acid, resulting in weaker MDR reversing effects. Previous research has revealed that the oxidative stress might have a role in the regulation of P-gp expression [43]. As caffeic acid exhibited significant ROS-related anti-oxidant effects, the influence of caffeic acid on ROS production in KB/VIN cell line was performed. The results showed that caffeic acid significantly decreased ROS

production in HeLaS3 cell line and slightly decreased ROS production in KB/VIN cell line. However, the doxorubicin-induced oxidative challenge was significantly reversed by caffeic acid in amounts of 10 μg/mL and 100 μg/mL in both HeLaS3 and KB/VIN cell lines. Therefore, the relationship between the reactive oxygen species levels and the up-regulation of *ABCB1* gene in KB/VIN might be related to the insufficient ROS regulation of caffeic acid. The above results indicated that the MDR reversal effects of caffeic acid might be cell line-dependent and warrant further detailed investigation.

The present study provided in-depth and comprehensive researches on the relationship between caffeic acid and human P-gp, and demonstrated the ability of caffeic acid on sensitizing MDR cancer cells toward chemotherapeutic drugs treatment. In order for caffeic acid to find a role in clinical application, some attempts could be applied to this phenolic prototype agent, including structural modification and pharmaceutical design.

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

### *4.1. Chemicals and Reagents*

Acetic acid, β-Mercaptoethanol (β-ME), caffeic acid, dimethyl sulfoxide (DMSO), ethanol (Absolute; analytical grade), paclitaxel, rhodamine 123, sulforhodamine B (SRB), trichloroacetic acid (TCA), tris base, (±)-verapamil, and vincristine were obtained from Sigma-Aldrich Co. (St Louis, MO, USA). Calcein-AM was from AAT Bioquest (Sunnyvale, CA, USA), and doxorubicin was from US Biological (Woburn, MA, USA). Dulbecco's Modified Eagle Medium, RPMI 1640 medium, fetal bovine serum (FBS), phosphate buffered saline (PBS; pH 7.2), Trypsin-EDTA, and hygromycin B were purchased from Invitrogen (Carlsbad, CA, USA). Zeocin was from InvivoGen (San Diego, CA, USA).

### *4.2. Cell Lines*

Human cervical epithelioid carcinoma HeLaS3 was purchased from Bioresource Collection and Research Center (Hsinchu, Taiwan), and the multi-drug resistant human cervical cancer cell line KB/VIN was kindly provided by Dr. Kuo-Hsiung Lee (University of North Carolina, Chapel Hill, NC, USA) and maintained with vincristine regularly. The human P-gp stable expression cells (*ABCB1*/Flp-InTM-293) and parental cell line Flp-InTM-293 were constructed as previously described [44]. All cells were cultured in DMEM or RPMI-1640 containing 10% FBS at 37 ◦C in a humidified atmosphere of 5% CO2.

### *4.3. Cytotoxicity Determination Assay (SRB Assay)*

The method has been described in our previous research [45]. Briefly, after 72 h of treatment of a series of concentrations of chemotherapeutic drugs with or without caffeic acid, 50% trichloroacetic acid (TCA) was added to fix cells for 30 min, and then the cells were washed with water and air-dried. After that, cells were stained with 0.04% sulforhodamine B (SRB) for 30 min, and then the unbound dye was removed by washing cells with 1% acetic acid and air-dried. The bound stain was solubilized in 10 mM Tris Base and the absorbance was measured using a BioTek Synergy HT Multi-Mode Microplate Reader (Winooski, VT, USA) at 515 nm.

### *4.4. MDR1 Shift Assay*

The method has been described in our previous research [46]. The conformation change of P-gp after the addition of caffeic acid was examined by using a MDR1 Shift Assay kit (EMD Millipore Corp., Billerica, MA, USA) according to the manufacturer's protocol. UIC2 shift was shown in the presence of a P-gp substrate such as vinblastine. A total of 5 <sup>×</sup> 105–1 <sup>×</sup> 106 cells were prepared per reaction and resuspended with warm UIC2 binding buffer. Cells were incubated at 37 ◦C for 10 min and then treated with DMSO or vinblastine or test compounds. Cells were incubated at 37 ◦C for 30 min and then treated with IgG2a (negative control antibody) or UIC2 working solution (P-gp conformational sensitive antibody). Cells were incubated at 37 ◦C for 15 min and then washed with iced UIC2 binding buffer twice. A secondary antibody, goat anti-mouse IgG ALEXA 488, was added at 4 ◦C for 15 min, and then iced UIC2 binding buffer was added. The fluorescence was measured by FACS analysis (BD FACSCantoTM II System, South City-I, Haryana, India).

### *4.5. Intracellular Calcein Accumulation Assay*

The method has been described in our previous research [46]. For the screening of an inhibitory effect on human P-gp efflux function, intracellular calcein accumulation assay was performed. Briefly, 1 <sup>×</sup> 10<sup>5</sup> cells/well were seeded in 96-well black plates for 24 h. Before the assay, cells were washed and pre-incubated with warm Hanks' balanced salt solution (HBSS) for 30 min and subsequently with caffeic acid for 30 min. After pre-incubation and three times washing with PBS, the calcein-AM was added (substrate of P-gp), and the calcein fluorescence generated within the cells was detected by BioTek Synergy HT Multi-Mode Microplate Reader using an excitation wavelength of 485 nm and emission wavelength of 528 nm at 37 ◦C temperature every 3 min for 30 min. Each experiment was performed at least three times, each in triplicate on different days.

### *4.6. Rhodamine123 and Doxorubicin E*ffl*ux Assay*

The method has been described in our previous research [46]. 1 <sup>×</sup> 105 cells/well were placed on 96-well plates and incubated overnight. Before the efflux assay, cells were washed and pre-incubated with warm HBSS for 30 min, and subsequently with caffeic acid for 30 min. Then, the cells were treated with rhodamine123 for 30 min or doxorubicin for 3 h at 37 ◦C. After being washed with warm PBS, cells were allowed to efflux fluorescent rhodamin123 and doxorubicin for 10 min and 2 h, respectively. Supernatant samples (100 μL) were transferred to 96-well black plates. The fluorescence of rhodamine123 and doxorubicin was measured using a BioTek Synergy HT Multi-Mode Microplate Reader (excitation/emission: 485/528 nm for rhodamine123, 485/590 nm for doxorubicin). Each experiment was performed at least three times, each in triplicate on different days. Kinetic parameters were estimated by nonlinear regression using Scientist v2.01 (MicroMath Scientific Software, Salt Lake City, UT, USA) according to the following equation:

$$\mathbf{V} = \frac{\mathbf{V\_{max}} \times \mathbf{C}}{\mathbf{K\_m} + \mathbf{C}}$$

where V denoted the efflux rate; Vmax, the maximal efflux rate; Km, the Michaelis-Menten constant; and C is the substrate concentration.

### *4.7. P-gp ATPase Activity Assay*

The method has been described in our previous research [46]. For the evaluation of P-gp ATPase activity of caffeic acid, Pgp-GloTM Assay System from Promega (Madison, WI, USA) was used. In a 96-well untreated white plate, 25 μg of recombinant human P-gp membranes were incubated with Pgp-GloTM Assay Buffer (untreated control), 200 μM verapamil (positive control for drug induced P-gp ATPase activity), 100 μM sodium orthovanadate (selective inhibitor for P-gp ATPase activity), or a series of concentrations of caffeic acid. The reaction was initiated by adding 5 mM MgATP and incubated for 40 min at 37 ◦C, followed by stopping the reaction with 50 μL ATPase Detection Reagent for 20 min at room temperature. Luminescence was measured using a BioTek Synergy HT Multi-Mode Microplate Reader, and data were presented as Change in Luminescence (ΔRLU).

### *4.8. Real-Time Quantitative RT-PCR*

The method has been described in our previous research [46]. *ABCB1* mRNA expression levels were quantified by real-time RT-PCR. Total RNA was extracted from HeLaS3, KB/VIN, Flp-InTM-293, and *ABCB1*/Flp-InTM-293 cells using Qiagen RNeasy kit (Valencia, CA, USA). Taqman Assay-On-DemandTM reagents of primers and probes for *ABCB1* (Hs00184500\_m1) and *GAPDH* (Hs02758991\_g1) genes were provided by Applied Biosystem (Foster City, CA, USA). The relative *ABCB1* mRNA expression levels were normalized to the amount of *GAPDH* in the same cDNA and evaluated by StepOnePlusTM Real-Time PCR System (Applied Biosystems®).

### *4.9. Intracellular Total ROS Activity Assay*

The influence of caffeic acid on intracellular reactive oxygen species (ROS) was evaluated with Cell Meter™ Fluorimetric Intracellular Total ROS Activity Assay Kit (Catalog number: 22900) purchased from AAT Bioquest (Sunnyvale, CA, USA). Briefly, 4 <sup>×</sup> 104 cells/well were seeded in 96-well black plates for 24 h. Then the cells were stained with AmpliteTM ROS Green working solution for 1 h; after that, caffeic acid with or without chemotherapeutic drugs were added to induce ROS production at room temperature for at least 15 min. The fluorescence was measured using a BioTek Synergy HT Multi-Mode Microplate Reader at 490/525 nm (same as FITC filter).

### *4.10. Cell Cycle Analysis*

The method has been described in our previous research [45]. Cells were plated to 6-well plates with serum-free medium for starvation. Twenty-four hours later, cells were treated with chemotherapeutic drugs with or without caffeic acid for 72 h. After that, cells were harvested and washed in cold phosphate-buffered saline (PBS), followed by fixing in ice-cold 70% ethanol for at least 24 h. Then, cells were incubated with 50 μg/mL PI at 4 ◦C for 24 h in the dark. Cells were then analyzed by FACS analysis (BD FACSCantoTM II System with excitation laser 488 nm, measuring at emission 575 nm for PI).

### *4.11. Molecular Docking*

Molecular docking helps us in predicting the intermolecular framework formed between a protein and a small molecule and suggests the binding modes responsible for inhibition of the protein. In this study, the existing structure of P-gp (PDB entry 6QEX) was used as a template for docking caffeic acid (PubChem CID: 689043) putative ligands using Discovery Studio 4.5. After removing all crystallized H2O molecules from the former construction, hydrogen was added into the CDOCKER module. CDOCKER is a powerful CHARMm-based docking method that has been used to generate highly accurate docked poses. In this refinement application, the ligands were conceded to tilt around the rigid receptor [47].

### *4.12. Statistical Analysis*

Statistical differences were evaluated by ANOVA followed by post hoc analysis (Tukey's test) or Student's t-test. The statistical significance was set at *p* value < 0.05.

**Author Contributions:** Conceptualization, Y.-N.T. and C.-C.H.; Data curation, Y.-N.T.; Formal analysis, Y.-N.T., C.C.N.W. and W.-C.L.; Funding acquisition, Y.-N.T. and C.-C.H.; Investigation, Y.-N.T., C.C.N.W., W.-C.L. and Y.-H.L.; Methodology, Y.-N.T., C.C.N.W., W.-C.L. and C.-C.H.; Project administration, C.-C.H.; Resources, Y.-H.L. and C.-C.H.; Software, Y.-N.T., C.C.N.W. and W.-C.L.; Supervision, Y.-H.L. and C.-C.H.; Validation, C.-C.H.; Visualization, C.-C.H.; Writing—original draft, Y.-N.T., C.C.N.W. and W.-C.L.; Writing—review & editing, Y.-H.L. and C.-C.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by China Medical University (CMU108-MF-69), I-Shou University (ISU 108-S-02), and Ministry of Science and Technology (MOST 108-2320-B-039-042 and MOST 108-2320-B-214-009).

**Acknowledgments:** Flow cytometry analyses were performed at the Medical Research Core Facilities Center, Office of Research & Development at China Medical University, Taichung, Taiwan, R.O.C.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

1. Silva, R.; Vilas-Boas, V.; Carmo, H.; Dinis-Oliveira, R.J.; Carvalho, F.; de Lourdes Bastos, M.; Remiao, F. Modulation of P-glycoprotein efflux pump: Induction and activation as a therapeutic strategy. *Pharmacol. Ther.* **2015**, *149*, 1–123. [CrossRef] [PubMed]


**Sample Availability:** Samples of the compounds are not available from the authors.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Pterostilbene Suppresses both Cancer Cells and Cancer Stem-Like Cells in Cervical Cancer with Superior Bioavailability to Resveratrol**

### **Hee Jeong Shin 1, Jang Mi Han 1, Ye Seul Choi <sup>1</sup> and Hye Jin Jung 1,2,3,\***


Academic Editor: Kyoko Nakagawa-Goto

Received: 23 November 2019; Accepted: 4 January 2020; Published: 6 January 2020

**Abstract:** Increasing studies have reported that cancer stem cells (CSCs) play critical roles in therapeutic resistance, recurrence, and metastasis of tumors, including cervical cancer. Pterostilbene, a dimethylated derivative of resveratrol, is a plant polyphenol compound with potential chemopreventive activity. However, the therapeutic effect of pterostilbene against cervical CSCs remains unclear. In this study, we compared the anticancer effects of resveratrol and pterostilbene using both HeLa cervical cancer adherent and stem-like cells. Pterostilbene more effectively inhibited the growth and clonogenic survival, as well as metastatic ability of HeLa adherent cells than those of resveratrol. Moreover, the superior inhibitory effects of pterostilbene compared to resveratrol were associated with the enhanced activation of multiple mechanisms, including cell cycle arrest at S and G2/M phases, induction of ROS-mediated caspase-dependent apoptosis, and inhibition of matrix metalloproteinase (MMP)-2/-9 expression. Notably, pterostilbene exhibited a greater inhibitory effect on the tumorsphere-forming and migration abilities of HeLa cancer stem-like cells compared to resveratrol. This greater effect was achieved through more potent inhibition of the expression levels of stemness markers, such as CD133, Oct4, Sox2, and Nanog, as well as signal transducer and activator of transcription 3 signaling. These results suggest that pterostilbene might be a potential anticancer agent targeting both cancer cells and cancer stem-like cells of cervical cancer via the superior bioavailability to resveratrol.

**Keywords:** cancer stem cell; cervical cancer; pterostilbene; resveratrol

### **1. Introduction**

Cervical cancer is one of the most common types of female malignant tumor, with worldwide incidence of more than 500,000 cases and mortality rate of 9% per year [1]. High-risk human papillomavirus (HPV) types such as HPV-16 and -18 are known to cause cervical cancer through the overexpression of viral oncoproteins E6 and E7 [2]. Although the worldwide death rate from cervical cancer has declined due to the current treatment modalities, including HPV vaccines, surgery, radiation therapy, and chemotherapy, the cancer recurrence, metastasis, and the adverse drug effects remain major problems [3]. Therefore, safer and more effective therapeutic options are needed to improve the treatment of cervical cancer.

Accumulating evidence has demonstrated that cancer stem cells (CSCs), a small subpopulation of tumor cells with self-renewal and multi-lineage differentiation capacities, crucially drive the development, metastasis, relapse, and chemo/radio-resistance of cervical cancer [4,5]. In addition, HPV oncoprotein E6 has been found to be involved in self-renewal and maintenance of stemness in cervical CSCs by upregulating Hes1, a downstream gene of Notch1 [6]. HPV16 E7 also upregulates the expression of stemness-related genes such as Oct3/4, Sox2, Nanog, and fibroblast growth factor 4 to maintain the self-renewal capacity of cervical CSCs [7]. Accordingly, targeting the cervical CSCs is a promising therapeutic strategy for the high-risk HPV-positive cervical cancer.

Various scientific studies have suggested the potential of natural active compounds isolated from plants or herbs for prevention and treatment of cancer [8,9]. Stilbenes are a class of polyphenolic compounds and naturally found in various dietary sources, such as grapes, blueberries, red wine, peanuts, and some medicinal plants [10]. Recently, stilbenes such as resveratrol (3,4 ,5-trihydroxy-trans-stilbene) and its dimethylated analog, pterostilbene (trans-3,5-dimethoxy-4 -hydroxystilbene), have received considerable attention due to their potent antioxidant, anti-inflammatory, antidiabetic, and anticarcinogenic properties (Figure 1) [11,12]. Resveratrol and pterostilbene have been considered as excellent anticancer agents because of their low toxicity and abilities to regulate multiple molecular signaling pathways involved in cancer progression [13,14]. However, resveratrol has a low bioavailability that may lower its biological efficacy, while pterostilbene is more lipophilic, and thus, it exhibits better bioavailability [15]. Pterostilbene shows stronger antiproliferative and apoptotic effects than those shown by resveratrol in the human colon and cervical cancer cells [16,17]. However, the therapeutic effect and anticancer mechanism of pterostilbene against cervical CSCs compared to resveratrol have not been studied.

Here, anticancer effects of resveratrol and pterostilbene were compared using both HeLa cervical cancer adherent and stem-like cells. The abilities of the two compounds to suppress growth, migration, and stemness of HeLa cells were evaluated and the underlying molecular mechanisms were further explored. The results revealed that pterostilbene more effectively inhibited the stem-like properties of HeLa cells than resveratrol through stronger downregulation of specific CSC markers and signal transducer and activator of transcription 3 (STAT3) signaling. This is the first study to demonstrate the potential inhibitory activity of pterostilbene against cervical cancer cell stemness.

**Figure 1.** Chemical structures of resveratrol and pterostilbene.

### **2. Results**

### *2.1. Pterostilbene Inhibited the Growth of Cervical Cancer Cells with Higher Potency Compared to Resveratrol*

First, we compared the inhibitory effects of resveratrol and pterostilbene on the growth of HeLa, CaSki, and SiHa cervical cancer adherent cells using the 3-(4,5-dimethylthiazol-2-yl)-2,5 diphenyltetrazolium bromide (MTT) assay at various concentrations (0–200 μM). Resveratrol and pterostilbene suppressed the growth of HeLa, CaSki, and SiHa cells in a concentration-dependent manner (Figure 2A). The results showed that pterostilbene (IC50 = 32.67 μM for HeLa; 14.83 μM for CaSki; 34.17 μM for SiHa) exhibited stronger growth inhibitory effect than resveratrol (IC50 = 108.7 μM for HeLa; 44.45 μM for CaSki; 91.15 μM for SiHa). Next, we evaluated the effects of pterostilbene and resveratrol on the colony formation of HeLa, CaSki, and SiHa adherent cells. The colony forming ability of the cells was more effectively inhibited by pterostilbene than resveratrol (Figure 2B). These

data demonstrate that pterostilbene is more potent in suppressing the growth and clonogenicity of cervical cancer adherent cells compared with resveratrol.

**Figure 2.** Growth inhibitory effects of resveratrol and pterostilbene on HeLa, CaSki, and SiHa cells. (**A**) The effects of resveratrol and pterostilbene on the growth of HeLa, CaSki, and SiHa adherent cells. The cells were treated with increasing concentrations of the two compounds (0–200 μM) for 72 h, and cell growth was measured by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. (**B**) The effects of resveratrol and pterostilbene on the colony forming ability of HeLa, CaSki, and SiHa adherent cells. The cells were incubated in the absence or presence of the two compounds (10 and 20 μM) for seven days. The cell colonies were detected by crystal violet staining. \* *p* < 0.05 versus the control.

### *2.2. Pterostilbene Exhibited Stronger Migration Inhibitory E*ff*ect than Resveratrol in Cervical Cancer Cells*

To compare the effects of resveratrol and pterostilbene on the metastatic ability of cervical cancer cells, we examined whether the two compounds inhibit the migration and invasion of HeLa adherent cells. A monolayer wound healing assay was performed to evaluate their effects on cell migration. Pterostilbene more markedly decreased the migration of HeLa cells at both 24 and 48 h after treatment when compared to resveratrol (Figure 3A). The effects of the two compounds on cell invasion were assessed using a Matrigel-coated Transwell chamber system. Both resveratrol and pterostilbene

resulted in a significant reduction in the invasiveness of HeLa cells (Figure 3B). In particular, the invasion inhibitory effect of pterostilbene was more potent than that of resveratrol.

**Figure 3.** Effects of resveratrol and pterostilbene on the metastatic ability of HeLa cells. (**A**) The effects of resveratrol and pterostilbene on the migration of HeLa adherent cells. The migratory potential of HeLa cells was analyzed using a wound healing assay. The cells were incubated in the absence or presence of the two compounds (20 μM) for 48 h. The cells that migrated into the gap were counted using an optical microscope. Dotted white lines indicate the edge of the gap at 0 h. (**B**) The effects of resveratrol and pterostilbene on the invasion of HeLa adherent cells. The invasiveness of HeLa cells was analyzed using Matrigel-coated polycarbonate filters. The cells were incubated in the absence or presence of the two compounds (10 and 20 μM) for 48 h. The cells penetrating the filters were stained and counted using an optical microscope. \* *p* < 0.05 versus the control.

### *2.3. Comparison of the Cell Cycle Arrest and Apoptosis-Inducing E*ff*ects of Resveratrol and Pterostilbene in Cervical Cancer Cells*

To determine whether the growth inhibitory effects of resveratrol and pterostilbene on HeLa adherent cells were caused by cell cycle arrest, the effects of the two compounds on the cellular cell cycle distribution were quantified using flow cytometry analysis. Both resveratrol and pterostilbene induced cell cycle arrest at the S and G2/M phases along with a decrease in G0/G1 phase duration when compared with the control cells (Figure 4A). Notably, pterostilbene was more potent than resveratrol in blocking cell cycle progression. The induction of tumor suppressor protein p53 and its downstream target p21 can trigger cell cycle arrest by inhibiting the activity of cyclin-dependent kinase (CDK)–cyclin complexes [18]. Therefore, the effects of resveratrol and pterostilbene on the expression of these cell cycle regulators were assessed. Results revealed that the cell cycle arrest at the S and G2/M phases of HeLa adherent cells by resveratrol and pterostilbene was associated with the promotion of p53 and p21 expression and subsequent downregulation of cyclin E1 and cyclin B1 that are active in the S and G2 phases, respectively (Figure 5B). Furthermore, pterostilbene not only more significantly increased the expression levels of p53 and p21, but also decreased those of cyclin E1 and cyclin B1 compared to resveratrol.

To further elucidate the mechanisms underlying the anticancer effects of the two compounds in cervical cancer cells, cellular apoptosis was quantitatively measured using flow cytometry analysis following annexinV-FITC/propidium iodide (PI) double staining. Annexin V is a marker of early apoptosis and PI is a marker of late apoptosis and necrosis. The total amount of early and late apoptotic cells was markedly increased after resveratrol and pterostilbene treatment in comparison with the control (Figure 4B). Moreover, the apoptosis-inducing effect of pterostilbene was stronger than that of resveratrol in HeLa adherent cells (from 4.67% to 30.58% and 50.46% by resveratrol and pterostilbene, respectively). The elevation of intracellular reactive oxygen species (ROS) plays an important role in mediating apoptotic processes [19]. Thus, to determine whether ROS are involved in the regulation of resveratrol- and pterostilbene-induced apoptosis, the levels of intracellular ROS in HeLa adherent cells were measured using the fluorescent 2 ,7 -dichlorofluorescein diacetate (DCFH-DA) product. Pterostilbene more prominently elevated the production of ROS in comparison with resveratrol at the indicated doses (Figure 5A). In addition, the cell apoptosis induced by resveratrol and pterostilbene was involved in the activation of caspase-3 and caspase-9, as well as the downregulation of antiapoptotic proteins such as Bcl-2 and Bcl-XL (Figure 5B) [20]. These data also showed that pterostilbene was more effective in regulating the expression of these apoptosis-related proteins than resveratrol.

Matrix metalloproteinases (MMPs) play a critical role in the degradation of the extracellular matrix (ECM) during cancer metastasis [21]. To further define the mechanism through which resveratrol and pterostilbene reduce cervical cancer cell migration and invasion, the protein expression levels of MMP-2 and MMP-9 in HeLa adherent cells were investigated. Pterostilbene more strongly suppressed the expression of MMP-2 and MMP-9 compared with resveratrol (Figure 5B). These findings suggest that pterostilbene may possess enhanced activity in inhibiting the metastasis of cervical cancer cells than resveratrol, through more effective downregulation of MMP-2 and MMP-9 expression.

Therefore, the superior growth and migration inhibitory effects of pterostilbene compared to resveratrol in HeLa adherent cells were mediated through the enhanced activation of multiple mechanisms, including cell cycle arrest at S and G2/M phases, induction of ROS-mediated caspase-dependent apoptosis, and inhibition of MMP-2/-9 expression.


**Figure 4.** Effects of resveratrol and pterostilbene on the cell cycle and apoptotic cell death of HeLa cells. (**A**) The cell cycle distribution of HeLa adherent cells was evaluated by flow cytometry after the treatment of the two compounds (40 μM) for 48 h. (**B**) HeLa adherent cells were treated with resveratrol and pterostilbene (40 μM) for 48 h. Apoptotic cells were determined by flow cytometry analysis following annexin V-FITC and propidium iodide (PI) dual labeling.

**Figure 5.** Identification of molecular mechanisms underlying the growth and migration inhibitory effects of resveratrol and pterostilbene in HeLa cells. (**A**) The effects of resveratrol and pterostilbene on reactive oxygen species (ROS) generation in HeLa adherent cells. The cells were treated with resveratrol and pterostilbene (20 and 40 μM) for 48 h. Intracellular ROS levels were detected with 2 ,7 -dichlorofluorescein diacetate (DCFH-DA). (**B**) The effects of resveratrol and pterostilbene on the expression of cleaved caspase-3, cleaved caspase-9, Bcl-2, Bcl-XL, p21, p53, cyclin E1, cyclin B1, MMP-2, and MMP-9 in HeLa adherent cells. The cells were treated with the two compounds (20 and 40 μM) for 48 h, and the protein levels were detected by Western blot analysis using specific antibodies. The levels of β-actin were used as an internal control. Arrowheads indicate true bands for the molecular markers. \* *p* < 0.05 versus the control.

### *2.4. Potent Inhibitory Activity of Pterostilbene against the Growth and Migration of Cervical CSCs*

CSCs, which play critical roles in therapeutic resistance, recurrence, and metastasis of tumors, have been identified in various solid tumors and hematological cancers including cervical cancer [22,23]. Therefore, anticancer agents with the potential to eliminate cervical CSCs may provide novel therapeutic opportunities for more effective treatment of cervical carcinoma.

To assess the effects of resveratrol and pterostilbene against stem-like properties of cervical cancer cells, the CSC population from HeLa cells was enriched in serum-free suspended spheroid culture condition [24,25]. First, we examined whether the two compounds affect the clonogenic growth as tumorspheres of cancer stem-like cells derived from HeLa cells. The tumorsphere forming ability of HeLa cancer stem-like cells was significantly suppressed by treatment with resveratrol and pterostilbene (Figure 6A). They decreased both the number and size of HeLa cancer stem-like cells. Particularly, pterostilbene was much stronger in inhibiting the tumorsphere formation of HeLa cancer stem-like cells compared with resveratrol.

**Figure 6.** Effects of resveratrol and pterostilbene on the tumorsphere-forming ability and migration of cervical cancer stem cells (CSCs). (**A**) HeLa cancer stem-like cells were treated with the two compounds (10 and 20 μM) and incubated with the CSC culture media for eight days. The number of formed tumorspheres in each well was counted under a microscope. (**B**) HeLa cancer stem-like cells were seeded into laminin-coated culture plate and incubated with the CSC culture media in the absence or presence of resveratrol and pterostilbene (10 and 20 μM) for 24 h. The cells that migrated into the gap were counted under an optical microscope. White lines indicate the edge of the gap at 0 h. \* *p* < 0.05 versus the control.

Next, we investigated the effects of resveratrol and pterostilbene on the migration of HeLa cancer stem-like cells. The wound healing assay showed that the two compounds reduced the migration of HeLa cancer stem-like cells when compared to the control conditions (Figure 6B). In addition, the migration inhibition activity of pterostilbene was much more potent than that of resveratrol. These findings underscore the superior therapeutic potential of pterostilbene to eliminate cervical CSCs.

### *2.5. Pterostilbene Exhibited Better Capacity for Inducing Cell Cycle Arrest and Apoptosis of Cervical CSCs Compared to Resveratrol*

To further elucidate the inhibitory effects of resveratrol and pterostilbene on the growth of cervical CSCs, the cell cycle progression and cellular apoptosis of HeLa cancer stem-like cells were measured by flow cytometry analysis. As shown in Figure 7A, both resveratrol and pterostilbene induced S phase arrest (increase in the proportion of arrested cells from 19.87% to 23.83% and 36.93% by treatment with resveratrol and pterostilbene, respectively) when compared with the control cells. Moreover, pterostilbene more strongly induced cell cycle arrest than resveratrol in HeLa cancer stem-like cells. Our data also showed that the number of early and late apoptotic cells was markedly increased after resveratrol and pterostilbene treatment in comparison with the control (Figure 7B). The apoptosis promoting effect of pterostilbene was more potent compared to resveratrol in HeLa cancer stem-like cells (increase in the proportion of total apoptotic cells, from 15.72% to 58.65% and 83.46% by resveratrol and pterostilbene, respectively). These results indicate that pterostilbene has a better capacity for inducing cell cycle arrest and apoptosis of cervical CSCs than resveratrol, thereby causing a stronger inhibition in the tumorsphere-forming ability of cervical CSCs in comparison with resveratrol.

**Figure 7.** Effects of resveratrol and pterostilbene on the cell cycle and apoptotic cell death of cervical CSCs. (**A**) The cell cycle progression and (**B**) cellular apoptosis of HeLa cancer stem-like cells were measured by flow cytometry analysis after the treatment of resveratrol and pterostilbene (40 μM) for 48 h.

### *2.6. E*ff*ect of Pterostilbene on the Expression of Stemness Markers in Cervical CSCs*

To explore the mechanism by which resveratrol and pterostilbene inhibit the growth and migration of cervical CSCs, their effects on the expression of transcription factors, Sox2, Oct4, and Nanog, were investigated. These transcription factors have been reported to induce stem-like properties, in HeLa cancer stem-like cells [26–28]. The two compounds effectively reduced the expression levels of the key stemness-related transcription factors as well as CD133, a cell surface marker for CSCs, suggesting that the inhibitory effects of resveratrol and pterostilbene against cervical CSCs may be associated with the downregulation of these stemness regulators (Figure 8A). Pterostilbene treatment more noticeably decreased the expression levels of stemness markers compared with resveratrol treatment.

**Figure 8.** Effects of resveratrol and pterostilbene on stemness markers and signal transducer and activator of transcription 3 (STAT3) signaling in cervical CSCs. (**A**,**B**) HeLa cancer stem-like cells were treated with resveratrol and pterostilbene (10 and 20 μM) for 48 h, and the protein levels were detected by Western blot analysis using specific antibodies. The levels of β-actin were used as an internal control. Arrowheads indicate true bands for the molecular markers. \* *p* < 0.05 versus the control.

The STAT3 pathway is involved in the maintenance of cervical CSCs by regulating the expression of stem cell-related transcription factors [29,30]. To further understand the molecular mechanism underlying the anticancer effects of resveratrol and pterostilbene against cervical CSCs, we investigated whether the two compounds affect STAT3 signaling. Our results confirmed that the compounds significantly decreased the expression levels of phosphorylated STAT3, without inhibiting the total protein levels of STAT3 in HeLa cancer stem-like cells (Figure 8B). Furthermore, pterostilbene more profoundly inhibited the phosphorylation of STAT3 than resveratrol. These results demonstrate that pterostilbene is more effective in suppressing the stem-like properties of cervical cancer cells than resveratrol through stronger downregulation of specific CSC markers and STAT3 signaling.

#### **3. Discussion**

The health beneficial effects of stilbene, a class of natural polyphenolic compounds, have been extensively studied in the past several decades [10,11]. Resveratrol and pterostilbene, the most widely known stilbenes, have gained increasing attention due to their roles in the potential prevention of major non-infectious chronic diseases such as cancer, cardiovascular disease, diabetes, and neurological degeneration [12–14]. Although both resveratrol and pterostilbene possess the therapeutic activities to inhibit various mechanisms for these human diseases, the bioavailability of pterostilbene with two methoxy groups is higher than resveratrol with two hydroxyl groups [12]. According to several studies, resveratrol and pterostilbene exhibit bioavailability at approximately 20% and 80% in vivo, respectively [31,32]. Such structural differences between the two compounds are expected to make pterostilbene more easily absorbed by oral ingestion through the promotion of lipophilicity and membrane permeability, compared with resveratrol.

The superior anticancer effects of pterostilbene have been reported in various tumors including lung, colon, breast, and cervical cancers [13]. Pterostilbene effectively suppressed cancer progression and metastasis by regulating apoptosis-dependent and apoptosis-independent signaling pathways. Accumulating evidence has shown that the anticancer effects of pterostilbene against cervical cancer are associated with the induction of apoptosis by activating the endoplasmic reticulum (ER)/nuclear factor erythroid 2-related factor 2 (Nrf2) pathway and downregulating the HPV oncoprotein E6 that causes the degradation of tumor suppressor protein p53 [17,33].

In the current study, we thoroughly investigated the cellular mechanisms responsible for the improved anticancer effects of pterostilbene compared to resveratrol in cervical cancer. Our results revealed that the superior growth and migration inhibitory effects of pterostilbene than resveratrol in HeLa cervical cancer cells could be attributed to the following reasons: the enhanced activation of multiple mechanisms, including cell cycle arrest at S and G2/M phases through the reduction of cyclin E1 and cyclin B1 expression following the induction of p53 and its downstream target p21; apoptosis through the activation of caspase-3 and caspase-9 mediated by ROS, as well as the downregulation of antiapoptotic proteins such as Bcl-2 and Bcl-XL; and the inhibition of MMP-2 and MMP-9 expression.

In addition, this is the first study to investigate the suppressive activities of pterostilbene and resveratrol against cervical cancer stemness. The critical role of CSCs in cancer progression and metastasis has already been identified and validated in many studies [4,5,22,23]. Notably, CSCs display resistance to many types of therapies, which results in cancer recurrence. Thus, it is important to suppress the self-renewal, proliferation, and metastasis abilities of CSCs for more reliable cancer treatment. In several cancers, the therapeutic effects of pterostilbene to eradicate CSCs have been confirmed. Pterostilbene inhibits the tumorsphere formation, migration, and stemness-related gene expression of CD133+ CSCs by downregulating multifaceted oncoprotein (MUC1), NF-κB, and Wnt/β-catenin-dependent signaling pathways in lung, breast, brain, and liver tumors [34–38]. However, the inhibitory effect of pterostilbene against cervical CSCs has not been studied previously.

In this study, pterostilbene significantly suppressed both the tumorsphere-forming ability and migration of HeLa cancer stem-like cells. Particularly, the therapeutic potential of pterostilbene to suppress cervical CSCs was markedly stronger than resveratrol. A tumorsphere is a solid, spherical structure developed from the proliferation of the cancer stem/progenitor cells. In these results, pterostilbene showed a better capacity for inducing cell cycle arrest and apoptosis of cervical CSCs in comparison with resveratrol. Therefore, the cell cycle arrest and apoptosis promoting effect of pterostilbene compared to resveratrol led to a stronger inhibition in the tumorsphere formation of HeLa cancer stem-like cells.

The stemness supporting transcription factors, such as Sox2, Klf4, c-Myc, Oct4, and Nanog, are upregulated in various types of CSCs [22,23,26–28]. Our results demonstrated that pterostilbene effectively decreased the expression levels of the major stemness transcription factors, including Sox2, Oct4, and Nanog, as well as a cell surface marker for CSCs, CD133, suggesting that the inhibitory effect of pterostilbene against cervical CSCs may be associated with the downregulation of these stemness regulators. In addition, pterostilbene more markedly suppressed the expression of stemness markers than resveratrol.

STAT3 is a transcription factor that is activated in many cancer types and can regulate pathways involving cell proliferation, cell survival, angiogenesis, and tumorigenesis [39]. Recent studies have revealed that STAT3 is an important regulator for self-renewal and survival of CSCs in various tumors [40]. In cervical carcinoma, STAT3 upregulates the stem-like characteristics of cervical cancer cells by increasing the expression of the stemness supporting markers such as Sox2, Oct4, and Nanog [29,30]. Therefore, targeting STAT3 signaling may be a promising approach to combat the survival of cervical CSCs. In the present study, pterostilbene resulted in a reduction in the expression levels of phosphorylated STAT3, without affecting the total protein levels of STAT3 in HeLa cancer stem-like cells. Furthermore, pterostilbene was more effective in suppressing the phosphorylation of STAT3 than resveratrol.

Our results collectively demonstrate that pterostilbene can suppress the stem-like properties of cervical cancer cells by downregulating specific CSC markers and STAT3 signaling. Thus, pterostilbene might serve as a potential anticancer agent for more effectively eradicating cervical cancer by targeting both cancer cells and CSCs, with superior bioavailability compared to resveratrol. However, the precise mechanism underlying how pterostilbene modulates the phosphorylation of STAT3 remains unclear. It is well known that STAT3 is phosphorylated by various protein kinases, including epidermal growth factor receptor (EGFR), Janus kinases (JAK), and Src family kinases (SFKs) [41]. Accumulating evidence has revealed that pterostilbene and resveratrol can induce cell cycle arrest and apoptosis by inhibiting the upstream kinase activities of STAT3 signaling in several cancers such as breast, pancreatic, prostate, and bone tumors [42–45]. Further studies to understand the mechanism of action of pterostilbene against cervical CSCs will help the discovery of the upstream cellular mediators of STAT3 activity regulated by pterostilbene. Moreover, further in vivo experiments using tumor xenograft animal models will be required to better verify the therapeutic potential of pterostilbene for cervical cancer compared to resveratrol.

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

### *4.1. Materials*

Resveratrol and pterostilbene were purchased from Sigma-Aldrich (Saint Louis, MO, USA) and dissolved in dimethyl sulfoxide (DMSO) at a concentration of 100 mM to prepare a stock solution. Matrigel and gelatin were obtained from BD Biosciences (San Jose, CA, USA). Laminin and the Transwell chamber system were obtained from Koma Biotech (Seoul, Korea) and Corning Costar (Acton, MA, USA), respectively. Antibodies against p21, p53, MMP-2, MMP-9, Bcl-2, Bcl-XL, cleaved caspase-3, cleaved caspase-9, cyclin E1, cyclin B1, STAT3, phospho-STAT3, Sox2, Oct4, Nanog, β-actin, rabbit IgG, and mouse IgG were purchased from Cell Signaling Technology (Danvers, MA, USA). Anti-CD133 antibody was obtained from MiltenyiBiotec GmbH (BergischGladbach, Germany).

### *4.2. Cell Culture*

Human cervical cancer HeLa, CaSki, and SiHa cell lines were obtained from the Korean Cell Line Bank (KCLB). The cells were cultured in Dulbecco's modified Eagle medium (DMEM; Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin–streptomycin–amphotericin B (Lonza, Walkersville, MD, USA) and then maintained at 37 ◦C in a 5% CO2 humidified incubator.

### *4.3. Cell Growth Assay*

HeLa, CaSki, and SiHa cells (3 <sup>×</sup> 103 cells/well) were seeded in a 96-well culture plate and then treated with various concentrations of resveratrol and pterostilbene (0–200 μM) for 72 h. Cell growth was measured with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) colorimetric assay (Sigma-Aldrich). The absorbance of each well was determined at a wavelength of 540 nm using a microplate reader (Thermo Fisher Scientific, Vantaa, Finland). The IC50 values from obtained data were analyzed using the curve-fitting program GraphPad Prism 5 (GraphPad Software, La Jolla, CA, USA).

### *4.4. Colony Formation Assay*

To evaluate the colony forming inhibitory effects of resveratrol and pterostilbene, HeLa, CaSki, and SiHa cells (2.5 <sup>×</sup> 102 cells/well) were seeded in a six-well culture plate. After 24 h incubation, the cells were treated with resveratrol and pterostilbene (10 and 20 μM) and incubated for 7 days until colonies were formed. Following this, the colonies were fixed with 4% formaldehyde and stained with 0.5% crystal violet solution for 10 min and washed with double-distilled water. The number of colonies on each well was counted and the percentage of compound-treated colonies relative to DMSO-treated control colonies was calculated.

### *4.5. Wound Healing Assay*

HeLa cells (2.5 <sup>×</sup> <sup>10</sup><sup>5</sup> cells/well) were seeded in a 24-well culture plate. The confluent monolayer cells were scratched using a tip and each well was washed with PBS to remove non-adherent cells. The cells were treated with resveratrol and pterostilbene (20 μM) and then incubated for up to 48 h. The perimeter of the central cell-free zone was observed under an optical microscope (Olympus, Center Valley, PA, USA).

### *4.6. Invasion Assay*

The invasiveness of cells was examined using Transwell chamber inserts with a pore size of 8.0 μm. The lower surface of the polycarbonate filter was coated with 10 μL of gelatin (1 mg/mL), while the upper surface was coated with 10 <sup>μ</sup>L of Matrigel (3 mg/mL). HeLa cells (5 <sup>×</sup> <sup>10</sup><sup>4</sup> cells/well) were seeded in the upper chamber of the filter; resveratrol and pterostilbene (10 and 20 μM) were added to the lower chamber filled with medium. The chamber was incubated at 37 ◦C for 48 h, and then the cells were fixed with 70% methanol and stained with hematoxylin and eosin (H and E). The total number of cells that invaded the lower chamber of the filter was counted using an optical microscope (Olympus).

### *4.7. Cell Cycle Analysis*

The cell cycle distribution was analyzed using the Muse Cell Cycle Assay Kit (Millipore, Hayward, CA, USA) according to the manufacturer's instructions. To this end, HeLa cells (5 <sup>×</sup> <sup>10</sup><sup>5</sup> cells/dish) were seeded in a 60-mm culture dish and treated with resveratrol and pterostilbene for 48 h. The cells were collected by centrifugation and washed using PBS and fixed in ice cold 70% ethanol at −20 ◦C for more than 3 h. The cells were then incubated with the Muse Cell Cycle reagent for 30 min at room temperature in the dark. The cell cycle analysis was carried out using the Muse Cell Analyzer (Millipore).

### *4.8. Apoptosis Analysis*

The apoptotic cell distribution was determined using the Muse Annexin V and Dead Cell Kit (Millipore) according to the manufacturer's instructions. Briefly, after treatment with resveratrol and pterostilbene, HeLa cells were collected and diluted with PBS containing 1% bovine serum albumin (BSA) as a dilution buffer to a concentration of 5 <sup>×</sup> <sup>10</sup><sup>5</sup> cells/mL. The single cell suspension was mixed with the Muse Annexin V/Dead Cell reagent at a 1:1 ratio and incubated in the dark for 20 min at room temperature. The cells were then analyzed using the Muse Cell Analyzer (Millipore).

### *4.9. Reactive Oxygen Species (ROS) Measurement*

Intracellular ROS levels were measured using a ROS-sensitive fluorescence indicator, 2 ,7 -dichlorofluorescein diacetate (DCFH-DA; Sigma-Aldrich). HeLa cells (1 <sup>×</sup> 105 cells/well) were seeded in a 96-black well culture plate and treated with resveratrol and pterostilbene (20 and 40 μM) for 48 h. The cells were then incubated with 10 μM of DCFH-DA for 20 min and washed with PBS. The fluorescence intensity of DCF was detected using a multimode microplate reader (Biotek, Inc., Winooski, VT, USA) at the excitation and emission wavelengths of 495 and 529 nm, respectively. The fluorescent images were also acquired using an Optinity KI-2000F fluorescence microscope (Korea Lab Tech, Seong Nam, Korea).

### *4.10. Western Blot Analysis*

Cells were lysed using RIPA buffer (Sigma-Aldrich) supplemented with a protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) on ice. Protein concentrations of the extracts were determined using a BCA Protein Assay kit (Pierce; Thermo Fisher Scientific, Inc., Waltham, MA, USA). Equal amounts of cell lysate were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and the separated proteins were transferred to polyvinylidene difluoride (PVDF) membranes (EMD Millipore) using standard electroblotting procedures. The blots were blocked in Tris-buffered saline with Tween-20 (TBST) containing 5% skim milk at room temperature for 1 h and immunolabeled with primary antibodies against p21, p53, MMP-2, MMP-9, Bcl-2, Bcl-XL, cleaved caspase-3, cleaved caspase-9, cyclin E1, cyclin B1, STAT3, phospho-STAT3, Sox2, Oct4, Nanog, CD133 (dilution 1:2000), and β-actin (dilution 1:10000) overnight at 4 ◦C. After washing with TBST three times, the membranes were incubated with horseradish peroxidase-conjugated anti-rabbit or anti-mouse (dilution 1:3000) secondary antibody for 1 h at room temperature. Immunolabeling was detected with an enhanced chemiluminescence (ECL) kit (Bio-Rad Laboratories, Inc., Hercules, CA, USA) according to the manufacturer's instructions. The band density was analyzed using ImageJ software (version 1.5; NIH).

### *4.11. CSC Culture*

CSCs were cultured using a non-adhesive culture method [24,25]. To propagate cervical cancer stem-like cells, HeLa cells grown in the serum-based media were cultured in Dulbecco's modified Eagle medium/nutrient mixture F-12 (DMEM/F12; Gibco) containing 1× B-27 serum-free supplement (Gibco), 5 μg/mL heparin (Sigma-Aldrich), 2 mM L-glutamine (Gibco), 20 ng/mL epidermal growth factor (EGF; Gibco), 20 ng/mL basic fibroblast growth factor (bFGF; Gibco), and 1% penicillin/streptomycin (Gibco). Tumorspheres grown in the serum-free media were subcultured every 7 days by dissociating with Accutase (Millipore) and maintained at 37 ◦C in a 5% CO2 humidified incubator.

### *4.12. CSC Tumorsphere Formation Assay*

HeLa cancer stem-like cells were seeded in a 96-well culture plate at a density of 500 cells/well using the serum-free media with EGF and bFGF. After 8 days of resveratrol and pterostilbene treatment (10 and 20 μM), the number of tumorspheres formed in each well was counted under an optical microscope (Olympus).

### *4.13. CSC Migration Assay*

For the CSC migration assay, the ibidi culture inserts (IBIDI GmbH, Martinsried, Germany) were placed in a laminin-coated 24-well culture plate. HeLa cancer stem-like cells were prepared at a density of 5 <sup>×</sup> 10<sup>5</sup> cells/mL, of which 70 <sup>μ</sup>L was transferred to each chamber. After cell attachment for 24 h, the culture inserts were removed, and the attached cells were incubated with the serum-free media

containing EGF and bFGF, in the absence or presence of resveratrol and pterostilbene (10 and 20 μM) for 24 h. The perimeter of the central cell-free zone was confirmed under an optical microscope (Olympus).

### *4.14. Statistical Analysis*

The data were presented as the mean ± standard error (SE) of three independent experiments. Student's *t*-test was used to determine statistical significance between the control and the test groups. A *p*-value of <0.05 was considered to indicate a statistically significant difference.

### **5. Conclusions**

The present study focused on the therapeutic effect and mechanism underlying the anticancer effects of pterostilbene against both cancer cells and cancer stem-like cells in cervical cancer compared to resveratrol. Our results demonstrated that the superior inhibitory effects of pterostilbene compared to resveratrol were associated with the enhanced activation of multiple mechanisms, including cell cycle arrest at S and G2/M phases through the induction of p53 and p21 and subsequent reduction of cyclin E1 and cyclin B1; apoptosis through the activation of caspase-3 and -9 mediated by ROS and the downregulation of Bcl-2 and Bcl-XL antiapoptotic proteins; and the inhibition of MMP-2 and -9 expression. Notably, pterostilbene exhibited a greater inhibitory effect against HeLa cancer stem-like cells than resveratrol through more potent inhibition of the expression levels of stemness markers, such as CD133, Oct4, Sox2, and Nanog, as well as STAT3 signaling. Based on these findings, we conclude that pterostilbene is a better potential candidate than resveratrol for more effectively treating cervical carcinoma.

**Author Contributions:** H.J.J. conceived and designed the experiments; H.J.S., J.M.H. and Y.S.C. performed the experiments and analyzed the data; H.J.S. and H.J.J. wrote the paper. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2016R1D1A1B03932956) and the NRF grant funded by the Ministry of Science and ICT (No. 2019R1A2C1009033). This work was also supported by the Brain Korea 21 Plus Project, Republic of Korea.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


**Sample Availability:** Samples are available from the first or corresponding author.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
