Withaferin A Induces Oxidative Stress-Mediated Apoptosis and DNA Damage in Oral Cancer Cells

Withaferin A (WFA) is one of the most active steroidal lactones with reactive oxygen species (ROS) modulating effects against several types of cancer. ROS regulation involves selective killing. However, the anticancer and selective killing effects of WFA against oral cancer cells remain unclear. We evaluated whether the killing ability of WFA is selective, and we explored its mechanism against oral cancer cells. An MTS tetrazolium cell proliferation assay confirmed that WFA selectively killed two oral cancer cells (Ca9-22 and CAL 27) rather than normal oral cells (HGF-1). WFA also induced apoptosis of Ca9-22 cells, which was measured by flow cytometry for subG1 percentage, annexin V expression, and pan-caspase activity, as well as western blotting for caspases 1, 8, and 9 activations. Flow cytometry analysis shows that WFA-treated Ca9-22 oral cancer cells induced G2/M cell cycle arrest, ROS production, mitochondrial membrane depolarization, and phosphorylated histone H2A.X (γH2AX)-based DNA damage. Moreover, pretreating Ca9-22 cells with N-acetylcysteine (NAC) rescued WFA-induced selective killing, apoptosis, G2/M arrest, oxidative stress, and DNA damage. We conclude that WFA induced oxidative stress-mediated selective killing of oral cancer cells.


INTRODUCTION
Most anticancer drugs effectively kill cancer cells; however, they also non-selectively kill normal cells, which limits their therapeutic value. It is currently believed that deregulated cell proliferation and deregulated apoptosis contributes to carcinogenesis (Reed, 1999). When the balance between proliferation and apoptosis is interrupted during tumor development, cell proliferation is deregulated (Scully et al., 2000;Evan and Vousden, 2001). Therapeutics that target oral squamous cell carcinoma (OSCC) cell proliferation and apoptosis regulators can enable these cancer cells to evade the regulatory system (Evan and Vousden, 2001).
Reactive oxygen species (ROS) are primarily generated in the mitochondria, and they cause a loss of mitochondrial membrane potential (MMP) (Li et al., 2006;Oh and Lim, 2006). ROS is an important inducer for the early stages of apoptosis (Samhan-Arias et al., 2004) and DNA damage (Barzilai and Yamamoto, 2004;Chen et al., 2016). Several drugs that modulate ROS have been reported (Nicco et al., 2005;Trachootham et al., 2006;Wu and Hua, 2007;Widodo et al., 2010) to regulate apoptosis for selective killing. Thus, ROS might mediate the selective activation of apoptosis for selective killing in cancer chemotherapy (Pollack et al., 2001;Daniel et al., 2003;Real et al., 2004).
Recently, several natural products have also been reported to induce apoptosis involving ROS (Ding et al., 2009;Chiu et al., 2013;Lee et al., 2013;Vyas and Singh, 2014). For example, the antiproliferation effect of cancer cells for the natural product from Withania somnifera (W. somnifera) is commonly reported to steroidal lactones (withanolides) (Vyas and Singh, 2014). Of the various withanolides derived from the root or leaf of W. somnifera, withaferin A (WFA) appears to be the most active against cancer (Vyas and Singh, 2014). The molecular anticancer activities of WFA have been reported through its antioxidant, anti-inflammatory and metabolic activities (Vanden Berghe et al., 2012;Vyas and Singh, 2014). WFA also induces ROS production and mitochondriamediated caspase activation in apoptosis of HL-60 myeloid leukemia cells (Malik et al., 2007). Thus, WFA has potential for modulating apoptosis and oxidative stress. Moreover, WFA has been widely used in several types of cancer (Malik et al., 2007;Uma Devi et al., 2008;Woo et al., 2014;Lee et al., 2015), but its use in oral cancer cells has not been sufficiently investigated.
We therefore hypothesized that WFA killed oral cancer cells by regulating oxidative stress-mediated apoptosis. We tested this hypothesis by evaluating cell viability, cell cycle changes, annexin V, caspases, ROS, MMP, and DNA damage. Moreover, the role of oxidative stress in WFA-induced cell killing of oral cancer cells was examined by the addition of an antioxidant N-acetylcysteine (NAC). Therefore, this work sheds light on exploring the roles of oxidative stress-mediated WFA-induced cell killing mechanism of oral cancer cells.

Cell Cultures and Chemicals
Ca9-22 and CAL 27 oral cancer cell lines and HGF-1 normal human gingival fibroblast cell lines were kept in Dulbecco's Modified Eagle's Medium (DMEM) (Gibco, Grand Island, NY, USA) plus 10% fetal bovine serum (FBS) at 37 • C in a humidified atmosphere containing 5% CO 2 , as previously described (Chang Y. T. et al., 2016). WFA was purchased from Selleckchem.com (Houston, TX, USA) and dissolved in dimethyl sulfoxide (DMSO) for experiments. An antioxidant or free radical scavenger NAC (Sigma-Aldrich; St. Louis, MO, USA) was pretreated before WFA treatment to diminish cellular ROS and confirm the role of oxidative stress in WFA treatment.

Cell Cycle Distribution
DNA content was detected using 7-aminoactinomycin D (7AAD), a DNA dye, (Vignon et al., 2013). After fixation in 70% ethanol, cells were treated with 1 µg/ml 7AAD for 30 min at 37 • C. Finally, the cells were resuspended in phosphate-buffered saline (PBS), and signals were examined using the FL3 channel of the Accuri C6 flow cytometer, and software (BD Biosciences, Franklin Lakes, New Jersey, USA). The G2/M percentage was calculated as the G2/M population among the whole cell cycle phases (G1, S, and G2/M).

Pan-caspase Assay to Measure Apoptosis
Apoptosis was measured using a generic caspase activity assay kit (Fluorometric-Green: ab112130; Abcam, Cambridge, UK)  for detecting the activity of caspases-1, 3-9. After the cells had been treated with WFA or NAC/WFA, they were treated with 2 µl of 500X TF2-VAD-FMK per 2 ml of medium for 2 h in a cell cultureincubator. After the cells had been washed with PBS, they were resuspended in 0.5 ml of assay buffer, and signals were immediately detected using the FL1 channel of the Accuri C6.

ROS Production Assay
2 ′ ,7 ′ -Dichlorodihydrofluorescein diacetate (H 2 DCF-DA) was used for ROS detection, as previously described (Shih et al., 2014). After the cells were treated with WFA or NAC/WFA, they were treated with 100 nM of H 2 DCF-DA in PBS for 30 min in a cell culture incubator. After the cells were harvested and washed, they were resuspended in PBS and their signals were immediately analyzed using the FL1 channel of the Accuri C6.

Mitochondrial Membrane Potential (MMP) Assay
MMP was detected using an assay kit (MitoProbe TM 3,3 ′diethyloxacarbocyanine iodide (DiOC 2 (3)) (Invitrogen, San Diego, CA, USA)) as previously described (Yen et al., 2012). After the cells were treated with WFA or NAC/WFA, they were washed with PBS and treated with 10 of 10 µM DiOC 2 (3) per 1 ml of medium/well in a 6-well plate in a cell culture incubator for 20-30 min. After the cells were harvested and washed, they were resuspended in PBS and their signals were immediately analyzed using the FL1 channel of the Accuri C6.
Phosphorylated Histone H2A.X (γH2AX) Assay γH2AX, DNA double-strand break biomarker, was detected using flow cytometry as previously described (Chen et al., 2016). After the cells were treated with WFA or NAC/WFA, they were fixed with 70% ethanol. After they had been washed, they were treated with 2 µg/ml of phospho-Histone 2A.X (Ser139) Antibody (H2AX) (sc-101696; Santa Cruz Biotechnology, Santa Cruz, CA, USA) at 4 • C for 1 h. After the cells were washed, they were treated with a secondary antibody (Jackson Laboratory, Bar Harbor, ME, USA) at room temperature for 30 min. Finally, the cells were treated in 20 µg/ml of PI, and their signals were detected using the FL1/FL2 channels of the Accuri C.

Statistical Analysis
All data are presented as means ± SD. Each analysis was performed in three separate experiments at different times (n = 3). All data were analyzed using Student paired t-test of Sigmaplot 10.0 (Scientific Data Analysis and Graphing Software, Systat Software Inc., Chicago, IL, USA).

RESULTS
The Viability of Oral Cancer Cells and Normal Oral Cells Treated with WFA Was Significantly Affected in WFA-Treated Cells with NAC Pretreatment MTS assays showed that the relative cell viability (%) of the Ca9-22 and CAL 27 oral cancer cells were significantly lower than control after 24 h WFA treatments of 0.5, 1, 2, and 3 µM in a dose-dependent manner ( Figure 1A). In contrast, HGF-1 normal oral cells treated with WFA showed no reduction in viability.
The involvement of oxidative stress in drug treatment is usually validated by pretreating cells with an antioxidant like NAC (Chan et al., 2006;Shieh et al., 2014;Hung et al., 2015;Lien et al., 2017). Cells treated with NAC-only [NAC pretreatment (2 mM)/WFA posttreatment (0 µM)] differed non-significantly from untreated controls (no NAC pretreatment and no WFA posttreatment in all three types of cells ( Figure 1B). Moreover, WFA-induced antiproliferation was significantly inhibited in two types of WFA-treated oral cancer cells with NAC pretreatment (NAC/WFA) (p < 0.05-0.001).
To further validate the low cytotoxicity of WFA-treated HGF-1 normal oral cells, the levels of WFA-induced apoptosis in HGF-1 cells were evaluated using the pan-caspase assay. The flow cytometric pan-caspase patterns of WFA-treated HGF-1 cells are shown in Figure 1C. Generic caspase activities in WFAtreated HGF-1 cells slightly increased at 1-3 µM WFA about 60% compared to the control (50%) (p < 0.001) (Figure 1D), suggesting that WFA only induced minor signs of apoptosis (only 10% induction) with low cytotoxicity to HGF-1 normal oral cells compared to the control.

Annexin V/PI-Induced Apoptosis of CA9-22 Oral Cancer Cells Treated with WFA Was Inhibited in WFA-Treated Cells with NAC Pretreatment
The flow cytometric annexin V/PI patterns of Ca9-22 oral cancer cells treated with WFA are shown in Figure 3A. The annexin V positive (+) expression (%) for WFA-treated Ca9-22 cells was higher than the control in a dose-dependent manner ( Figure 3B).
The involvement of caspases in the change of pan-caspase activity was validated using western blotting. In Figures 4E,F, the expression of apoptosis signaling proteins, such as cleaved PARP and cleaved-caspases 3, 8, and 9, was higher in WFA-treated Ca9-22 cells; although some of them declined at a concentration of 3 µM. In contrast, these caspase signaling proteins of the WFAtreated Ca9-22 cells were inhibited in WFA-treated Ca9-22 cells with NAC pretreatment (NAC/WFA).

DISCUSSION
WFA has anticancer (Uma Devi et al., 2008;Woo et al., 2014;Lee et al., 2015) and ROS production (Lee et al., 2009;Li et al., 2017;Liu et al., 2017) effects in several types of cancer. However, the selective killing effect of WFA was rarely investigated as yet, especially in oral cancer cells. The aim of this study is to examine the selective killing effects of WFA against oral cancer cells and to explore the role of oxidative stress in WFA-treated oral cancer cells. The cell killing, apoptosis, cell cycle, DNA damage, and oxidative stress effects of WFA in oral cancer cells are discussed as follows.

Comparison of IC 50 Using WFA in Different Types of Cancer Cells
Different cancer cell types are sensitive to WFA. For example, the IC 50 values of WFA were 2 µM for myeloid leukemia (HL-60) cells (24 h; MTT assay) (Malik et al., 2007), about 2 µM for pancreatic cancer cells (Panc-1 and MIAPaCa-2) (48 h; MTS assay) (Li et al., 2015), and 0.2-1.2 µM for cervical cancer cells (C33a, CaSki, HeLa, and SiHa) (24 h; MTT assay) (Munagala    al., 2011). However, the selective killing effect of WFA was not thoroughly investigated in these studies. We also found that the IC 50 -values of WFA in Ca9-22 and CAL 27 oral cancer cells were 3 and 2 µM (24 h; MTS assay), respectively. In contrast, WFA did not harm HGF-1 normal oral cells. Accordingly, WFA induces selective killing against oral cancer cells without adverse effects for the viability of normal oral cells. Therefore, we provide here the first report that documents WFA's selective killing effects against oral cancer cells.
However, HGF-1 normal human gingival cells are a type of fibroblast cell lines and oral cancer cells arise from the epithelium. It was reported that some epithelial and fibroblast cells had different responses to various signaling molecules. For example, epithelial cells undergo growth arrest in response to TGF-β, whereas fibroblasts undergo morphological changes, and proliferate (Wilkes et al., 2003). Therefore, the possibility of selective killing of WFA against oral cancer cells warrants further detailed investigation. In a follow-up study we will compare oral cancer cells with normal epithelial cells using a similar experimental design.

WFA Induces Apoptosis in Cancer Cells
In the current study, WFA induced apoptosis in Ca9-22 oral cancer cells associated with subG1 accumulation and caspase activation. This is consistent with the findings that WFA induced apoptosis in CaSki cervical cancer cells (Munagala et al., 2011) and WFA induced apoptosis in MDA-MB-231 and MCF-7 breast cancer cells (Hahm and Singh, 2013). In the current study (Figure 4E), the cleaved caspases (3, 8, and 9) and cleaved PARP showed high cleavage (high apoptosis) in 1 and 2 µM of WFA, but showed less cleavage (less apoptosis) in 3 µM of WFA. Similarly, other drugs have related findings. For example, etoposide induced more cleavage of PARP at 2 and 4 days, but declined at day 6 in human non-small cell lung cancer cells (H1437) (Chiu et al., 2005). Ganciclovir induced more cleavage of PARP, caspases 3/9, and cytochrome c at 1 or 2 days, but declined at 3 days in human non-small cell lung cancer cells (CL-1) (Chiu et al., 2002). It is possible that higher dose treatment or longer exposure may lead to more cell death and fail to proportionally cleave apoptotic proteins. Furthermore, caspase-1 may mediate apoptosis (Bergsbaken et al., 2009;Miao et al., 2011;Sollberger et al., 2014) and it warrants further investigation for the role of caspase-1 in WFA induced apoptosis.
In addition to the induction of subG1 accumulation, WFA induces G2/M arrest and a mitotic catastrophe in prostate cancer cells in terms of cell cycle analysis and western blotting for several G2/M arresting proteins (Roy et al., 2013). WFA also induces G2/M arrest in gastric cancer cells (Kim et al., 2017), osteosarcoma cells (Lv and Wang, 2015), and breast cancer cells (Zhang et al., 2011). Similar to the current study, WFA induced G2/M arrest in Ca9-22 cells at lower concentrations (1 and 2 µM) but did not present at higher concentration (3 µM) ( Figure 2B) where subG1 population (apoptosis) was gradually increased.

WFA Induces DNA Damage in Cancer Cells
Drug-induced ROS generation is associated with DNA damage Chiu et al., 2013;Chen et al., 2016). Since WFA is known to induce ROS production (Lee et al., 2009;Li et al., 2017;Liu et al., 2017) (Figure 5), the DNA damage effect warrants for investigation. In the current study, we found that WFA treatment induced γH2AX expression in Ca9-22 oral cancer cells, which is consistent with the findings that WFAinduced γH2AX expression in MCF7 breast cancer cells, but not in normal human TIG-3 fibroblasts (Widodo et al., 2010). This DNA damaging ability was shown to be associated with apoptosis in several other cancer cell studies (Norbury and Zhivotovsky, 2004;Roos and Kaina, 2006).
WFA-Induced Cell Killing, Apoptosis, Cell Cycle Change, and DNA Damage Are Medidated by Oxidative Stress in Cancer Cells NAC is a common free radical scavenger. NAC pretreatment is reported to effectively diminish cellular ROS and to confirm the role of oxidative stress in drug treatment. Recently, several WFA-based cancer therapies were reported to be rescued by NAC pretreatment. For example, WFA induced apoptosis in human melanoma cells was accomplished by generating ROS (Mayola et al., 2011), and NAC pretreatment rescued ROS-induced damage. NAC also inhibited WFA-induced ROS production and caspase activation of human HL-60 myeloid leukemia cells (Malik et al., 2007).
Similarly, we found that NAC pretreatment rescued WFAinduced apoptosis in Ca9-22 oral cancer cells by subG1 accumulation, annexin V/PI, pan-caspases, and caspase signaling in western blotting. In our study, low concentration of WFA-induced G2/M arrest in Ca9-22 cells was rescued by NAC pretreatment. NAC pretreatment rescued WFA-induced oxidative stress in Ca9-22 cells by ROS and MMP analysis. NAC pretreatment also rescued γH2AX-based DNA damage in Ca9-22 cells as shown by the flow cytometry assay. This is the first evidence that NAC pretreatment protects WFA-treated oral cancer cells against oxidative stress.

CONCLUSION
WFA selectively killed oral cancer cells with less toxic effects to normal oral cells. WFA also induced apoptosis, oxidative stress, and DNA damage, which ultimately inhibited the proliferation of oral cancer cells.
National Sun Yat-Sen University-KMU Joint Research Project; KMUH105-5R61 from the Kaohsiung Medical University Hospital; and MOHW106-TDU-B-212-144007 from the Taiwan Ministry of Health and Welfare surcharge on tobacco products.