Golden Berry-Derived 4β-hydroxywithanolide E for Selectively Killing Oral Cancer Cells by Generating ROS, DNA Damage, and Apoptotic Pathways

Background Most chemotherapeutic drugs for killing cancer cells are highly cytotoxic in normal cells, which limits their clinical applications. Therefore, a continuing challenge is identifying a drug that is hypersensitive to cancer cells but has minimal deleterious effects on healthy cells. The aims of this study were to evaluate the potential of 4β-hydroxywithanolide (4βHWE) for selectively killing cancer cells and to elucidate its related mechanisms. Methodology and Principal Findings Changes in survival, oxidative stress, DNA damage, and apoptosis signaling were compared between 4βHWE-treated oral cancer (Ca9-22) and normal fibroblast (HGF-1) cells. At 24 h and 48 h, the numbers of Ca9-22 cells were substantially decreased, but the numbers of HGF-1 cells were only slightly decreased. Additionally, the IC50 values for 4βHWE in the Ca9-22 cells were 3.6 and 1.9 µg/ml at 24 and 48 h, respectively. Time-dependent abnormal increases in ROS and dose-responsive mitochondrial depolarization can be exploited by using 4βHWE in chemotherapies for selectively killing cancer cells. Dose-dependent DNA damage measured by comet-nuclear extract assay and flow cytometry-based γ-H2AX/propidium iodide (PI) analysis showed relatively severer damage in the Ca9-22 cells. At both low and high concentrations, 4βHWE preferably perturbed the cell cycle in Ca9-22 cells by increasing the subG1 population and arrest of G1 or G2/M. Selective induction of apoptosis in Ca9-22 cells was further confirmed by Annexin V/PI assay, by preferential expression of phosphorylated ataxia-telangiectasia- and Rad3-related protein (p-ATR), and by cleavage of caspase 9, caspase 3, and poly ADP-ribose polymerase (PARP). Conclusions/Significance Together, the findings of this study, particularly the improved understanding of the selective killing mechanisms of 4βHWE, can be used to improve efficiency in killing oral cancer cells during chemoprevention and therapy.


Introduction
Oral cancer is the sixth most common cancer worldwide [1]. Its high morbidity and mortality are partly due to its relatively poor chemotherapy outcomes [2]. Because of their high cytotoxicity in normal cells, the various anti-oral cancer drugs developed so far have limited therapeutic applications. Therefore, a continuing challenge is to develop an anti-oral cancer therapy that is safer and more effective, particularly in terms of selective killing efficiency.
Accumulating evidence in recent studies indicates that selective activation of apoptosis improves the effectiveness of cancer chemotherapy [10][11][12][13][14][15]. For improved modulation of the apoptotic potential of cancer cells, one promising line of research is the use of withanolides that selectively kill tumor cells but have low toxicity in healthy cells [13]. However, the potential use of apoptosisinducing withanolides such as 4bHWE [4] and withaferin A [7][8][9] for selective killing, especially in oral cancer cells, is rarely discussed.
Therefore, this study examined the potential effectiveness and related mechanisms of P. peruviana-derived 4bHWE used for selectively killing oral cancer cells.

Assessment of intracellular reactive oxygen species (ROS)
Intracellular ROS was measured using 29,79-dichlorodihydrofluorescein diacetate (DCFH-DA) from Sigma Chemical Co. (St. Louis, MO) as previously described [17]. After 4bHWE treatment, cells were washed with PBS and then incubated with 10 mM H2DCF-DA in PBS for 30 min at 37uC in darkness. Cells were then harvested and washed with PBS. After centrifugation, cells were resuspended in PBS and analyzed with a FACSCalibur flow cytometer (Becton-Dickinson, Mansfield, MA, USA) with Win-MDI software (http://facs.scripps.edu/software.html) at excitation and emission settings of 480 and 525 nm, respectively.

Assessment of mitochondrial membrane potential
Mitochondrial membrane potential (DY m ; MitoMP) was measured using a MitoProbe TM DiOC 2 (3) assay kit (Invitrogen, San Diego, CA, USA) as described in [16]. The 4bHWE-treated cells were suspended in 1 ml of warm PBS at approximately 1610 6 cells/ml, loaded with 5 ml of 10 mM DiOC 2 (3), and incubated at 37uC in 5% CO 2 for 20-30 min. After harvest, cells were washed, resuspended in PBS, and analyzed immediately using a flow cytometer with Win-MDI software at excitation and emission settings of 488 and 525 nm, respectively.

Assessment of DNA damage by comet-NE assay
The nuclear extract (NE) of HGF-1 cells was used to perform comet-NE assay according to a previously described protocol [4,20] with slight modification. Briefly, cell suspensions were mixed with equal volumes of 1.2% low-melting-point agarose, immediately loaded onto 1.2% regular agarose pre-coated slides, and then cooled with ice until solidification. The third layer with an equal volume of 1.2% low-melting agarose gel was then loaded onto the solidified second gel and again cooled with ice. After cell lysis treatment at 4uC for 2 h, the slides were processed to NE digestion with a coverslip and incubated at 37uC for 2 h in a humidified space. Denaturation of the slides in 0.3 N NaOH and 1 mM EDTA for 20 min was followed by electrophoresis. After washing, the slides were transferred to 0.4 M Tris-HCl (pH 7.5), and 40 ml propidium iodide (PI, 50 mg/ml; Sigma, St Louis, MO, USA) was added for fluorescence microscopy observation (TE2000-U; Nikon, Tokyo, Japan). In the comet assay, a freeware program (http://tritekcorp.com) was used to measure DNA damage in terms of percentage of tail DNA [21].

Assessment of cell cycle distribution and sub-G1 population
The PI staining was performed as described previously [20]. Briefly, cells were treated with vehicle (DMSO only) or 0.5, 1, 2, 5, and 10 mg/ml 4bHWE for 24 and 48 h. After harvest, cells were fixed overnight with 70% ethanol. After centrifugation, the cell pellets were incubated with 10 mg/ml PI and 10 mg/ml RNase A in PBS for 15 min at room temperature in darkness. The samples were then analyzed with a FACSCalibur flow cytometer and Win-MDI software.

Assessment of apoptosis
Apoptosis was measured by annexin/PI double staining (Pharmingen, San Diego, CA, USA) as previously described [22]. Briefly, cells were treated with vehicle or with 4bHWE at doses of 1, 2, and 5 mg/ml for 24 h. The cells were then incubated with 10 mg/ml of annexin V-fluorescein isothiocyanate and 5 mg/ ml of PI and analyzed with a FACSCalibur flow cytometer (Becton-Dickinson) with Win-MDI software.

Statistical analysis
All data were presented as means 6 SD. Group differences in cell viability and cell cycle were assessed by using JMPH 9 software to perform one-way ANOVA with Tukey HSD Post Hoc Test. Levels not connected by the same lower-case letter indicated significant differences. Other data were analyzed by Student t-test.

Assessment of growth inhibition
The MTS assay of cell viability after 24 h and 48 h treatment with 4bHWE (0, 1, 2, 5 and 10 mg/ml) showed that, for each experimental concentration of 4bHWE, proliferation of Ca9-22 oral cancer cells was significantly lower than that of HGF-1 normal cells (one-way ANOVA) ( Figure 1). In HGF-1 cells, treatment with 10 mg/ml 4bHWE slightly decreased cell viability to 89.0261.17% and to 65.4262.26% after 24 h and 48 h, respectively. In contrast, the viability of similarly treated 4bHWE-

Assessment of ROS
Some withanolides such as withaferin A reportedly induce cell death in melanoma cells by generating ROS [8]. Therefore, this study next compared the regulating effects of ROS on the proliferation of Ca9-22 and HGF-1 cells. Figures 2A and 2B show the ROS fluorescence intensities in terms of percentages of Ca9-22 and HGF-1 cells positive for DCFD-A, which were counted after treatment with 3.6 mg/ml 4bHWE for varying time intervals. Figure 2C shows the mild induction of ROS observed in HGF-1 cells compared to the relatively dramatic time-dependent induction of ROS in Ca9-22 cells. For each experimental concentration of 4bHWE, the percentage of DCFD-A positive cells was significantly higher in the Ca9-22 oral cancer cells than in the normal HGF-1 cells (P,0.0001).

Assessment of mitochondrial membrane potentials (mitoMP)
Next, a DiOC 2 (3) assay was performed to examine the effects of 4bHWE-induced ROS induction on mitoMP. Figures 3A and 3B show the mitoMP fluorescence intensities in terms of percentages of DiOC 2 (3)-positive Ca9-22 and HGF-1 cells after 24 h treatment with 0, 1, 2, and 5 mg/ml 4bHWE. Figure 3C shows that mitoMP moderately decreased in HGF-1 cells but substantially decreased in Ca9-22 cells in a dose-dependent manner. For each experimental concentration of 4bHWE, the percentage of cells positive for DiOC 2 (3) was significantly lower in the Ca9-22 cells than in the HGF-1 cells (P,0.0005).

Assessment of DNA damage by comet-NE assay
In the comet-NE assay ( Figure 4A), the ''tailing'' effects observed in the Ca9-22 cells were largest at high 4bHWE concentrations. In contrast, none of the experimental 4bHWE concentrations induced an observable tailing effect in HGF-1 cells. Figure 4B shows that the HGF-1 cells showed no visible increase in% tail DNA whereas Ca9-22 cells showed dramatically increased% tail DNA in a dose-response manner. For each experimental concentration of 4bHWE, the DNA damage in terms of% DNA in the tails of cells treated with 4bHWE was significantly more severe in HGF-1 cells than in Ca9-22 cells (P,0.0001).  Figure 5B shows that, after treatment with 4bHWE concentrations lower than 2 mg/ml, the HGF-1 cells maintained low levels of c-H2AX expression whereas Ca9-22 cells showed dramatic dose-dependent increases in c-H2AX expression. At higher concentrations of 4bHWE, however, the fold change in the% of c-H2AX-positive cells was significantly larger in Ca9-22 cells than in HGF-1 cells (P,0.0005). Figure 6 shows that, after 24 h treatment with 4bHWE, the percentage change in sub-G1 populations in HGF-1 cells was not statistically significant at concentrations lower than 2 mg/ml. The percentage change in sub-G1 slightly increased to 2.75% when the concentration reached 5 mg/ml. In contrast, the percentage change in sub-G1 in Ca9-22 cells treated with 4bHWE for 24 h began showing significant changes at concentrations as low as 2 mg/ml and eventually reached 30.59%. After 48 h treatment, the percentage of sub-G1 populations in HGF-1 cells moderately increased to 25.03% and 29.42% after treatments with 2 and 5 mg/ml 4bHWE, respectively. In contrast, the percentage of sub-G1 in Ca9-22 cells treated with 4bHWE for 48 h showed a statistically significant change (42.76%) after treatment with 1 mg/ ml and a much larger change (78.89%) after treatment with 5 mg/ ml.

Assessment of cell cycle distribution
At 24 h, analyses of G1 and G2/M populations in HGF-1 cells showed a non-significant change in the G1 population and a basal level of change in the G2/M pollution. In contrast, after 24 h 4bHWE treatment, the Ca9-22 cells showed significant changes in G1 arrest (57.49% and 40.75% at 0.5 and 1 mg/ml, respectively) and in G2/M arrest (50.44% and 62.08% at 1 and 2 mg/ml, respectively). After 48 h treatment with 5 mg/ml bHWE, the G1 population in the HGF-1 cells slightly decreased to 56.27% whereas the G2/M population slightly increased. In contrast, the Ca9-22 cells showed substantial (79.23%) G1 arrest after 48 h treatment with a 4bHWE concentration of only 0.5 mg/ml. In the G2/M population, treatment with 1 mg/ml 4bHWE resulted in moderate (27.34%) G1 arrest coupled with dramatic subG1 accumulations as described above.

Assessment of apoptosis
To examine the involvement of apoptosis in 4bHWE-induced sub-G1 accumulation, flow cytometry-based annexin V/PI double staining was performed. Figure 7A shows the c-H2AX/PI staining profiles of HGF-1 and Ca9-22 cells treated with varying 4bHWE concentrations. The double positive areas of c-H2AX and PI intensities are commonly defined as late apoptosis. Analyses of late apoptosis ( Figure 7B) showed a mild dose-dependent increase in HGF-1 cells but a dramatic dose-dependent increase in Ca9-22 cells. For each experimental concentration of 4bHWE, the percentage of cells that underwent late apoptosis after 4bHWE treatment was significantly higher in Ca9-22 cells than in HGF-1 cells (P,0.0001).

Assessment of apoptotic signaling
Apoptotic signaling was examined by Western blotting assays of Ca9-22 and HGF-1 cells treated with varying concentrations of 4bHWE (0, 1, 2, and 5 mg/ml). Figure 8 shows that increasing concentrations of 4bHWE induced stepwise increases in ATR phosphorylation associated with cellular DNA damage in the Ca9-22 cells. Cleavage of caspase 9, caspase 3 and PARP was also
Recently, withanolide derived from Ashwagandha leaf extract [32] has shown a potential role in selectively killing breast cancer MCF7 cells at a concentration of 24 mg/ml [33]. The current  study similarly showed that the IC 50 values of Ca9-22 oral cancer cells were 3.6 mg/ml (7.16 mM) and 1.9 mg/ml (3.78 mM) after 24 h and 48 h treatment with 4bHWE, respectively. We hypothesize that the selective killing effect of 4bHWE may be comparable or even more potent in other oral cancer cell lines. In HGF-1 cells, however, IC 50 values were undetectable by MTS assay at concentrations lower than 10 mg/ml. In other studies of 4bHWE treatments for 24 h, an IC 50 value of 6 mM was reported in an MTT assay of breast cancer MDA-MB-231 cells [6] and an IC 50 value of 1.41 mM was reported in a trypan blue assay of lung cancer H1299 cells [4]. These data indicate that the drug sensitivity of 4bHWE varies according to cancer cell type. Moreover, the current study is the first to confirm that 4bHWE kills oral cancer cells preferentially to normal oral cells.
Furthermore, the protective effects of withanolides are found in normal fibroblast cells. For example, withanone derived from the Ashwagandha leaf protects normal human fibroblasts against methoxyacetic acid-induced senescence-like growth arrest in terms of senescence-associated b-galactosidase staining [34]. Similarly, the current study found that the morphology of HGF-1 cells treated with 1, 2, and 5 mg/ml 4bHWE was similar to that of untreated fibroblast HGF-1 cells (data not shown). Further studies of growth arrest phenotypes are needed to determine whether 4bHWE is safe for normal cells.
Although accumulating evidence agrees that withanolides induce ROS-mediated apoptosis, their potential use for selectively killing cancer cells is relatively unclear. For example, withaferin A is known to induce ROS-mediated apoptosis in breast cancer [7], melanoma [8], and leukemia [9,31]. Although cancer cells are expected to have higher oxidative stress compared to normal cells [35], normal cells may tolerate an exogenous oxidative stress level sufficient to prevent overload resulting in cell death. In contrast, cancer cells exposed to high oxidative stress cannot tolerate exogenous ROS-modulating agents, and cell death increases as the threshold level is exceeded [36]. This concept may partly explain the selective killing effects of 4bHWE in oral cancer cells observed here and those of withanone in breast cancer cells as reported in   [33], i.e., both ROS induction and the reduction of mitochondrial membrane potential are higher in cancer cells compared to normal cells. These results suggest that therapies for selectively killing cancer cells should be targeted at modulating redox status in both cancer cells and normal cells [36]. However, the roles of caspases and caspase inhibitors [37] in 4bHWE-induced selective apoptosis need further study.
The ROS-mediated effects of withanolides may be modulated by antioxidants. For example, N-acetylcysteine can reportedly rescue human melanoma cells from withaferin A-induced, ROSmediated apoptosis [8]. Accordingly, the role of ROS in the selective killing of 4bHWE may be further clarified by studying ROS modulators.
The ROS are known to induce DNA damage and checkpoint responses [38]. For example, the comet-NE and c-H2AX assays in this study revealed that 4bHWE induced DNA damage and G1 or G2/M cell cycle arrest, both of which have been observed earlier in other withanolides. For example, tubocapsanolide A reportedly inhibits proliferation of lung cancer A549 cells via G1 arrest [39]. However, 4bHWE [4] and withanone [33] are known to induce DNA damage and then arrest at G2/M in lung cancer H1299 cells and in breast cancer MCF-7 cells, respectively.
Selective DNA damage (i.e., DSB) can be monitored by H2AX, which is phosphorylated in an ATR-dependent manner [40]. The H2AX also helps to stabilize the genome [41] and is essential for caspase-activated DNA fragmentation [42]. As expected, 4bHWE treatment induced higher expressions of ATR and caspase signaling proteins in Ca9-22 cells compared to HGF-1 cells.
After  (Figure 4), respectively. These results further confirmed that 4bHWE treatment selectively induces ROS and DNA damage in Ca9-22 cells in preference to HGF-1 cells.
In conclusion, the results of this study confirm that 4bHWE treatment selectively induces ROS, mitochondrial depolarization, and DNA damage, which in turn induces selective apoptosis signaling, which ultimately results in the selective killing of oral cancer cells (Figure 9). Accordingly, this study showed, for the first time, that 4bHWE treatment selectively kills oral cancer cells in preference to normal oral cells. Further study of the target candidates and signaling mechanisms reported here may also provide a sufficiently improved understanding of the selective killing mechanisms of 4bHWE to enable its effective use in treating oral cancer with minimal adverse effects.