Melatonin and (−)-Epigallocatechin-3-Gallate: Partners in Fighting Cancer

We have demonstrated previously that melatonin attenuates hepatotoxicity triggered by high doses of (−)-epigallocatechin-3-gallate (EGCG) in mice. The current work investigated the influence of melatonin on the oncostatic activity of EGCG in two cancer cell lines, wherein melatonin induced an opposite response of p21. In human tongue cancer TCA8113 cells, melatonin-induced p21 and EGCG-mediated formation of quinoproteins were positively associated with the oncostatic effects of melatonin and EGCG. Melatonin-stimulated an increase in p21 which was correlated with a pronounced nuclear translocation of thioredoxin 1 and thioredoxin reductase 1, both of which are known to induce p21 via promoting p53 trans-activation. Melatonin did not influence the EGCG-mediated increase of quinoprotein formation nor did EGCG impair melatonin-induced p21 up-regulation. Co-treatment with both agents enhanced the cell-killing effect as well as the inhibitory activities against cell migration and colony formation. It is known that p21 also plays a powerful anti-apoptotic role in some cancer cells and confers these cells with a survival advantage, making it a target for therapeutic suppression. In human hepatocellular carcinoma HepG2 cells, melatonin suppressed p21 along with the induction of pro-survival proteins, PI3K and COX-2. However, EGCG prevented against melatonin-induced PI3K and COX-2, and melatonin probably sensitized HepG2 cells to EGCG cytotoxicity via down-regulating p21, Moreover, COX-2 and HO-1 were significantly reduced only by the co-treatment, and melatonin aided EGCG to achieve an increased inhibition on Bcl2 and NFκB. These events occurring in the co-treatment collectively resulted in an enhanced cytotoxicity. In addition, the co-treatment also enhanced the inhibitory activities against cell migration and colony formation. Overall, the results gathered from these two cancer cell lines with a divergent p21 response to melatonin show that the various oncostatic activities of melatonin and EGCG together are more robust than each agent alone, suggesting that they may be useful partners in fighting cancer.


Cell Lines and Cell Culture
Human tongue squamous cell line TCA8113 was provided by the Key Laboratory of Oral Biomedicine of Shanghai Jiaotong University (Shanghai, China). Human hepatocellular carcinoma (HCC) cell line HepG2 was obtained from the Stem Cell Bank of Chinese Academy of Sciences (Shanghai, China). Both cell lines were cultured in RPMI 1640 medium containing 10% (v/v) fetal bovine serum, 100 U/mL penicillin and 100 µg/mL streptomycin. The medium for HepG2 cells additionally contained 2 mM L-glutamine. The cells were maintained at 37 • C under 5% CO 2 and 95% air.

MTT Assay
TCA8113 or HepG2 cells were seeded in 96-well plates at a density of 50,000 or 100,000 cells/well, respectively, 24 h prior to each experiment. The cells were exposed to melatonin or/and EGCG for 48 h. The medium was then removed and 200 µL RPMI-1640 medium containing 0.5 mg/mL MTT was added to each well. After a 4-h incubation, the medium was replaced with 150 µL dimethylsulfoxide. Optical density was measured by a microplate reader (Molecular Devices SpectraMax 190, Sunnyvale, CA, USA) at 490 nm. Data are presented as percentages of viable cells versus control.

Cell Migration
Cell migration was evaluated by scratch assay. TCA8113 cells or HepG2 cells were seeded on 6-well plates at a density of 1 × 10 6 /well. After 24 h incubation, sterile pipette tips were used to scratch cell layers. Following a brief wash with phosphate buffer solution (PBS), the cells were treated with melatonin or/and EGCG for 48 h. The cells were photographed, and the images were used to measure the width of the wound. Relative migration ratio of the scratched cells was calculated according to the formula: (width at 0 h − width at 48 h)/width at 0 h.

Colony Formation
TCA8113 or HepG2 cells were seeded in 6-cm cell culture dishes at a density of 500 or 40,000 cells/dish, respectively. After 24 h, the cells were treated with melatonin or/and EGCG. After a 14-day incubation, the cells were washed with PBS, fixed with the methanol for 15 min, and stained with 0.1% crystal violet for 20 min for counting colony number.

Western Blot
Cells cultured in 10-cm dishes were washed thrice with PBS and then lysed in Radio-Immune Precipitation Assay (RIPA) lysis buffer containing 1% PMSF on ice for 30 min. Following centrifugation (15,000× g, 10 min, 4 • C), the supernatants were incubated at 95 • C for 10 min in 5×SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis) loading buffer. After protein separation through electrophoresis on a 15% SDS-PAGE gel, proteins in the gel were transferred to a Polyvinylidene difluoride (PVDF) membrane. The membrane was blocked using 5% nonfat dried milk in TBS-T (10 mM Tris-HCl, pH7.8, 150 mM NaCl and 0.05% Tween-20 (v/v)) for at least 2 h at room temperature. After a short wash with the TBS-T, the membrane was incubated with a primary antibody diluted in the TBS-T by 1000-to 5000-fold for at least 3 h at room temperature. The membrane was washed thrice with the TBS-T and then was incubated with a secondary antibody diluted in the TBS-T by 2500-to 5000-fold for 1 h at room temperature. Following three washings with the TBS-T, protein signals were detected using ChemiDoc XRS + detection system (ECL, Bio-Rad, Hercules, CA, USA). The Quantity One ® Image Analyzer software program (Bio-Rad) was used for quantitative densitometric analysis.

Preparation of Nuclear Fraction
Cell nuclear fraction was prepared according to the protocol described in our previous study [51]. With minor modifications. In brief, cells were washed thrice with cold PBS containing 1 mM EDTA-Na 2 and lysed for 20 min at 4 • C in cold isotonic buffer (pH 7.0) containing 250 mM sucrose, 20 mM HEPES, 0.15 mM EDTA, 0.015 mM EGTA, 1 mM dithiothreitol, 10 mM KCl, 1 mM phenylmethylsulfonyl fluoride, 20 mM NaF, 1 mM sodium pyrophosphate, 1 mM Na 3 VO 4 , 1% (v/v) Nonidet P-40 and 1 µg/mL protease inhibitor cocktail. The homogenates were centrifuged at 2000× g for 10 min at 4 • C. The supernatants were discarded and the precipitates were washed once with the above isotonic buffer but without containing the protease inhibitor cocktail. The nuclear pellets were sonicated in RIPA buffer containing 1% PMSF on ice for 5 min. The homogenates were centrifuged at 15,000× g for 15 min and the supernatants were collected as nuclear fraction. Proteins in the nuclear fraction were assessed using Western blot as described above. Validation of the purity of the cellular fractions is provided as Supplementary Figure S1.

Quinoprotein Measurement
Cell quinoprotein levels were measured essentially according to our previous protocol used for animal tissues [52]. Briefly, cells cultured in 10-cm dishes were washed thrice with PBS and then lysed in RIPA buffer containing 1% PMSF on ice for 30 min. After a centrifugation (15,000× g, 10 min, 4 • C), the supernatants were collected, and the protein concentrations were measured by BCA protein assay kit (Beyotime Biotechnology, Shanghai, China). The supernatants containing at least 0.5 mg protein were then mixed with 10% trichloroacetic acid at a volume ratio of 1/1 to precipitate soluble protein.
After centrifuging at 1000× g for 10 min, the precipitate was washed with ethanol and then centrifuged at 4000× g for 5 min. The precipitate was further treated with chloroform-methanol (2:1, v/v) and centrifuged at 5000× g for 10 min. The resultant delipidated protein precipitate was suspended with 2 M potassium glycinate (pH 10.0) at 37 • C for 1 h and the supernatant was obtained by centrifuging at 15,000× g for 15 min. Finally, 4 µL NBT stock solution (12 mM) was added to 200 µL supernatant. The mixture was incubated at 37 • C in the dark, and absorbance at 530 nm was measured using the aforementioned microplate reader. Quinoprotein levels were expressed as OD 530 nm /mg protein.

Analysis of mRNA Expression by Real-Time PCR
Total RNA was extracted by using TRIzol reagent (Takara Biotechnology, Kusatsu, Shiga, Japan) according to the manufacturer's protocol. The preparation of the cDNA was carried out following the instructions described in the kit manual (RT-for-PCR kit, Takara Biotechnology). Real-Time Polymerase Chain Reaction (RT-PCR) was performed with a CFX System (Bio-Rad) according to the manufacturer's instructions (Takara Biotechnology). The mRNA level was normalized with β-actin as an internal reference. The following primers were used for PCR reactions.

Statistical Analysis
Data are presented as means ± SEM. The significance differences between groups were examined by student's t test, one-way ANOVA post hoc Dunnett test, or two-way ANOVA post hoc Bonferroni test, as appropriate. All statistical analyses were performed using GraphPad software (Prism version 5, San Diego, CA, USA). A p value of < 0.05 was considered statistically significant. We designed experiments and treatment groups based on the results of our previous experiments and did not do formal a priori power analysis.

Melatonin-Induced p21 Correlates with the Increase of Nuclear Trx1 and TrxR1 in TCA8113 Cells
Confluent TCA8113 cells were treated with melatonin. Melatonin dose-dependently decreased viable cells at 48 h ( Figure 1A) and increased p21 at 24 h ( Figure 1B); this suggests that p21 is positively associated with melatonin-induced oncostatic activity. To elucidate how melatonin up-regulates p21, we detected the alteration of p53 mRNA. A dose of 4 mM melatonin was used because it substantially reduced the number of viable cells and dramatically increased p21 ( Figure 1A,B). Melatonin transiently increased p53 mRNA within 24 h ( Figure 1C). It is known that cisplatin stimulates translocation of Trx1 from the cytosol into the nucleus, leading to augmentation of p21 trans-activation via Trx1-dependent redox regulation of p53 [53,54]. We thus evaluated the influence of melatonin on nuclear Trx1 and its reductase (TrxR1). At 24 h, nuclear Trx1 and TrxR1 levels were dramatically increased by 4 mM melatonin ( Figure 1D). To the best of our knowledge, this is the first finding that melatonin can stimulate nuclear translocation of Trx1 or TrxR1. Because this concept was established using high concentration of melatonin which initiates pronounced cytotoxic response, we subsequently investigated the influence of a lower concentration of melatonin (that caused marginal cytotoxicity) on p21 and nuclear Trx1/TrxR1. For this purpose, 1.5 mM melatonin was employed according to the dose response shown in Figure 1A. The low concentration of melatonin also significantly increased p21 ( Figure 1E) and nuclear Trx1 and TrxR1 ( Figure 1F) at 24 h. Cytoplasmic TrxR1 remained basically unchanged but cytoplasmic Trx1 significantly reduced following the melatonin treatment (Supplementary Figure S2). Overall, melatonin, at typical oncostatic concentrations, readily promotes nuclear translocation of both Trx1 and TrxR1, leading to avid up-regulation of p21 in TCA8113 cells.

EGCG Increases Quinoprotein Levels in TCA8113 Cells
Auto-oxidation of EGCG produces ROS and EGCG quinone, which can covalently bind to the free thiol group of cysteine residues in proteins, resulting in the formation of quinoproteins and consequent alterations of protein functions [30,55]. Intracellular quinoprotein levels of cancer cells subjected to EGCG treatments have never been investigated. Thus, herein we evaluated the influence of EGCG on quinoprotein levels in TCA8113 cells. The derivative product of quinoproteins and NBT can be kinetically assessed at 530 nm [52]. With this approach, we found that EGCG dosedependently increased quinoprotein levels following a 24-h incubation ( Figure 2A). Based on the area under curve, EGCG at 400 μM raised quinoprotein levels 12.2-fold, whereas EGCG at 200 μM elevated quinoprotein levels 6.3-fold, relative to control. It has been demonstrated that the amounts of intracellular EGCG are no longer measured after 3 h in EGCG-treated cancer cells [56]. In this regard, quinoprotein containing the molecular skeleton of EGCG stands out as a biomarker signature because of the straightforward dose response relationship at apparently protracted time point. EGCG caused marginal cytotoxicity at 200 μM and reduced the number of viable cells by 65% at 400 μM ( Figure 2B). Quinoprotein levels are apparently associated with EGCG-triggered oncostatic effect. Compared to the control, * p < 0.05, ** p < 0.01 and *** p < 0.001.

EGCG Increases Quinoprotein Levels in TCA8113 Cells
Auto-oxidation of EGCG produces ROS and EGCG quinone, which can covalently bind to the free thiol group of cysteine residues in proteins, resulting in the formation of quinoproteins and consequent alterations of protein functions [30,55]. Intracellular quinoprotein levels of cancer cells subjected to EGCG treatments have never been investigated. Thus, herein we evaluated the influence of EGCG on quinoprotein levels in TCA8113 cells. The derivative product of quinoproteins and NBT can be kinetically assessed at 530 nm [52]. With this approach, we found that EGCG dose-dependently increased quinoprotein levels following a 24-h incubation (Figure 2A). Based on the area under curve, EGCG at 400 µM raised quinoprotein levels 12.2-fold, whereas EGCG at 200 µM elevated quinoprotein levels 6.3-fold, relative to control. It has been demonstrated that the amounts of intracellular EGCG are no longer measured after 3 h in EGCG-treated cancer cells [56]. In this regard, quinoprotein containing the molecular skeleton of EGCG stands out as a biomarker signature because of the straightforward dose response relationship at apparently protracted time point. EGCG caused marginal cytotoxicity at 200 µM and reduced the number of viable cells by 65% at 400 µM ( Figure 2B). Quinoprotein levels are apparently associated with EGCG-triggered oncostatic effect.

Concurrent Increases of p21 and Quinoprotein in TCA8113 Cells Treated with Melatonin and EGCG
We examined the combined effects of melatonin and EGCG on p21 and quinoprotein. We found that EGCG did not influence the p21 protein up-regulation induced by melatonin and melatonin did not affect the dramatic increase of quinoproteins caused by EGCG at either 12 h ( Figure 3A,B) or 24 h ( Figure 3C,D). Because the above experiments showed that p21 and quinoprotein levels were positively associated with melatonin-and EGCG-induced oncostatic effects, respectively, the concurrent increase of p21 and quinoprotein by the co-treatment suggests that the two agents could generate enhanced oncostatic actions. This concept was reinforced by the following experiments using TCA8113 cells. The co-treatment was more efficient than the single agent in 1) increasing DNA damage as reflected by γ-H2AX ( Figure 3E), 2) inducing apoptotic response as suggested by cleaved caspase 3 ( Figure 3E

Concurrent Increases of p21 and Quinoprotein in TCA8113 Cells Treated with Melatonin and EGCG
We examined the combined effects of melatonin and EGCG on p21 and quinoprotein. We found that EGCG did not influence the p21 protein up-regulation induced by melatonin and melatonin did not affect the dramatic increase of quinoproteins caused by EGCG at either 12 h ( Figure 3A,B) or 24 h ( Figure 3C,D). Because the above experiments showed that p21 and quinoprotein levels were positively associated with melatonin-and EGCG-induced oncostatic effects, respectively, the concurrent increase of p21 and quinoprotein by the co-treatment suggests that the two agents could generate enhanced oncostatic actions. This concept was reinforced by the following experiments using TCA8113 cells. The co-treatment was more efficient than the single agent in (1) increasing DNA damage as reflected by γ-H2AX ( Figure 3E), (2) inducing apoptotic response as suggested by cleaved caspase 3 ( Figure 3E

Concurrent Increases of p21 and Quinoprotein in TCA8113 Cells Treated with Melatonin and EGCG
We examined the combined effects of melatonin and EGCG on p21 and quinoprotein. We found that EGCG did not influence the p21 protein up-regulation induced by melatonin and melatonin did not affect the dramatic increase of quinoproteins caused by EGCG at either 12 h ( Figure 3A,B) or 24 h ( Figure 3C,D). Because the above experiments showed that p21 and quinoprotein levels were positively associated with melatonin-and EGCG-induced oncostatic effects, respectively, the concurrent increase of p21 and quinoprotein by the co-treatment suggests that the two agents could generate enhanced oncostatic actions. This concept was reinforced by the following experiments using TCA8113 cells. The co-treatment was more efficient than the single agent in 1) increasing DNA damage as reflected by γ-H2AX ( Figure 3E), 2) inducing apoptotic response as suggested by cleaved caspase 3 ( Figure 3E

Melatonin Reduces p21 in HepG2 Cells and Sensitizes the Cells to EGCG
We have demonstrated previously that melatonin reduces EGCG-triggered hepatotoxicity in mice [50]. Thus, we determined the influence of co-treatment of melatonin and EGCG on human HCC HepG2 cells. When HepG2 cells approached confluence, the cells were exposed to 1.5 mM melatonin for 24 h. The melatonin significantly suppressed p21 probably owing to melatonin-induced low expression of p53 ( Figure 6A). In a 24-h incubation, 3 mM melatonin also substantially suppressed p21, whereas 400 μM EGCG appeared to increase p21; however, the melatonin significantly attenuated p21 elevation induced by the EGCG (Figure 6B). Many studies have shown that attenuation of p21 in certain types of cancer cells could sensitize them to chemotherapeutic agents [57][58][59][60][61]. In glioblastoma U-87 MG cells, taxol also appeared to induce p21, similar to the action of EGCG in HepG2 cells. A p21 antisense oligonucleotide, which could dramatically decrease the steady state levels of p21, reduced p21 induction by taxol and significantly enhanced taxol-induced apoptosis [62]. Thus, we inferred that melatonin may enhance the oncostatic effects of EGCG in HepG2 cells. Five lines of evidence demonstrated that the combination of melatonin and EGCG could achieve an enhanced oncostatic effects compared to each agent alone. 1) Either 400 μM EGCG or 3 mM melatonin significantly reduced the number of viable cells measured by the MTT assay, but the co-treatment was more effective than the single agent ( Figure 6C). 2) Neither 400 μM EGCG nor 3 mM melatonin significantly caused DNA damage as indicated by γ-H2AX; only the co-treatment effectively damaged DNA ( Figure 6D). 3) Neither 400 μM EGCG nor 3 mM melatonin significantly affected nuclear H3 levels; only the co-treatment dramatically reduced nuclear H3 levels ( Figure 6E), suggesting that the nucleus was severely damaged by the co-treatment. 4) Either 400 μM EGCG or 3 mM melatonin significantly reduce cell migration evaluated by wound healing assay, but the cotreatment was more potent than each single agent ( Figure 7A). 5) Either 20 μM EGCG or 1 mM melatonin significantly inhibited colony formation, but the co-treatment was more robust than each single agent ( Figure 7B).

Melatonin Reduces p21 in HepG2 Cells and Sensitizes the Cells to EGCG
We have demonstrated previously that melatonin reduces EGCG-triggered hepatotoxicity in mice [50]. Thus, we determined the influence of co-treatment of melatonin and EGCG on human HCC HepG2 cells. When HepG2 cells approached confluence, the cells were exposed to 1.5 mM melatonin for 24 h. The melatonin significantly suppressed p21 probably owing to melatonin-induced low expression of p53 ( Figure 6A). In a 24-h incubation, 3 mM melatonin also substantially suppressed p21, whereas 400 µM EGCG appeared to increase p21; however, the melatonin significantly attenuated p21 elevation induced by the EGCG (Figure 6B). Many studies have shown that attenuation of p21 in certain types of cancer cells could sensitize them to chemotherapeutic agents [57][58][59][60][61]. In glioblastoma U-87 MG cells, taxol also appeared to induce p21, similar to the action of EGCG in HepG2 cells. A p21 antisense oligonucleotide, which could dramatically decrease the steady state levels of p21, reduced p21 induction by taxol and significantly enhanced taxol-induced apoptosis [62]. Thus, we inferred that melatonin may enhance the oncostatic effects of EGCG in HepG2 cells. Five lines of evidence demonstrated that the combination of melatonin and EGCG could achieve an enhanced oncostatic effects compared to each agent alone. (1) Either 400 µM EGCG or 3 mM melatonin significantly reduced the number of viable cells measured by the MTT assay, but the co-treatment was more effective than the single agent ( Figure 6C). (2) Neither 400 µM EGCG nor 3 mM melatonin significantly caused DNA damage as indicated by γ-H2AX; only the co-treatment effectively damaged DNA ( Figure 6D).
(3) Neither 400 µM EGCG nor 3 mM melatonin significantly affected nuclear H3 levels; only the co-treatment dramatically reduced nuclear H3 levels ( Figure 6E), suggesting that the nucleus was severely damaged by the co-treatment. (4) Either 400 µM EGCG or 3 mM melatonin significantly reduce cell migration evaluated by wound healing assay, but the co-treatment was more potent than each single agent ( Figure 7A). (5) Either 20 µM EGCG or 1 mM melatonin significantly inhibited colony formation, but the co-treatment was more robust than each single agent ( Figure 7B

EGCG Reverses Melatonin-Induced Pro-Survival Responses and Melatonin Does Not Impair Death-Promoting Action of EGCG in HepG2 Cells
In the aforementioned five experiments, except colony formation, the other studies were carried out using 3 mM melatonin and 400 μM EGCG. We further used these doses to examine the reciprocal interaction of melatonin and EGCG in HepG2 cells following a 24-h incubation. Melatonin treatment up-regulated pro-survival PI3K and COX-2 proteins ( Figure 8A) probably as an adaptive response to pronounced reduction of p21 and moderate cytotoxixity ( Figure 6B,C); these adaptive responses were thoroughly suppressed in the co-treatment of melatonin and EGCG ( Figure 8A). As expected, the EGCG treatment dramatically increased quinoprotein levels, and the melatonin treatment did not significantly affect EGCG-elevated quinoprotein levels ( Figure 8B). The EGCG treatment suppressed apoptosis inhibitor survivin, anti-apoptotic Bcl2, pro-survival transcriptional factor NFκB ( Figure  8C). However, the co-treatment with melatonin enhanced the effects of EGCG ( Figure 8C). Moreover, while the EGCG treatment seemed to suppress cytoprotective enzyme HO-1 (heme oxygenase-1), melatonin aided to significantly decrease HO-1 ( Figure 8C).

EGCG Reverses Melatonin-Induced Pro-Survival Responses and Melatonin Does Not Impair Death-Promoting Action of EGCG in HepG2 Cells
In the aforementioned five experiments, except colony formation, the other studies were carried out using 3 mM melatonin and 400 µM EGCG. We further used these doses to examine the reciprocal interaction of melatonin and EGCG in HepG2 cells following a 24-h incubation. Melatonin treatment up-regulated pro-survival PI3K and COX-2 proteins ( Figure 8A) probably as an adaptive response to pronounced reduction of p21 and moderate cytotoxixity ( Figure 6B,C); these adaptive responses were thoroughly suppressed in the co-treatment of melatonin and EGCG ( Figure 8A). As expected, the EGCG treatment dramatically increased quinoprotein levels, and the melatonin treatment did not significantly affect EGCG-elevated quinoprotein levels ( Figure 8B). The EGCG treatment suppressed apoptosis inhibitor survivin, anti-apoptotic Bcl2, pro-survival transcriptional factor NFκB ( Figure 8C). However, the co-treatment with melatonin enhanced the effects of EGCG ( Figure 8C). Moreover, while the EGCG treatment seemed to suppress cytoprotective enzyme HO-1 (heme oxygenase-1), melatonin aided to significantly decrease HO-1 ( Figure 8C).

Discussion
The present study using the assays of cytotoxicity, cell migration and colony formation reveals that the oncostatic activity of the co-treatment of melatonin and EGCG is always more robust than each single agent in two cancer cell lines with divergent p21 response to melatonin.
Early studies indicated that melatonin at a concentration as low as 1 nM significantly increased p21; thus, melatonin-induced increase of p21 is considered to be an important cancer-inhibitory mechanism [63,64]. The present study showed that melatonin at typical oncostatic doses markedly increased p21 in TCA8113 cells. The p21 up-regulation was correlated with Trx1/TrxR1 shuttling into the nucleus. Trx, TrxR and NADPH comprise a Trx system responsible for thiol-disulfide exchange reactions [65]. TrxR and Trx work together to reduce oxidized cysteine residues within the DNAbinding domain of several transcription factors including p53, maintaining maximal DNA-binding activity [66]. Transfection of Trx enhances p53-dependent expression of p21 [53]. Cisplatin-triggered nuclear translocation of Trx1 stimulates p53-dependent p21 trans-activation [53,54]. In estradiol-17βtreated cells, an alternative splicing variant of TrxR1 was found in the nucleus [67]. However, none of these studies revealed a concurrent nuclear translocation of Trx1 and TrxR1. We found that melatonin at oncostatic concentrations potently increased both nuclear Trx1 and nuclear TrxR1 levels in TCA8113 cells.

Discussion
The present study using the assays of cytotoxicity, cell migration and colony formation reveals that the oncostatic activity of the co-treatment of melatonin and EGCG is always more robust than each single agent in two cancer cell lines with divergent p21 response to melatonin.
Early studies indicated that melatonin at a concentration as low as 1 nM significantly increased p21; thus, melatonin-induced increase of p21 is considered to be an important cancer-inhibitory mechanism [63,64]. The present study showed that melatonin at typical oncostatic doses markedly increased p21 in TCA8113 cells. The p21 up-regulation was correlated with Trx1/TrxR1 shuttling into the nucleus. Trx, TrxR and NADPH comprise a Trx system responsible for thiol-disulfide exchange reactions [65]. TrxR and Trx work together to reduce oxidized cysteine residues within the DNA-binding domain of several transcription factors including p53, maintaining maximal DNA-binding activity [66]. Transfection of Trx enhances p53-dependent expression of p21 [53]. Cisplatin-triggered nuclear translocation of Trx1 stimulates p53-dependent p21 trans-activation [53,54]. In estradiol-17β-treated cells, an alternative splicing variant of TrxR1 was found in the nucleus [67]. However, none of these studies revealed a concurrent nuclear translocation of Trx1 and TrxR1. We found that melatonin at oncostatic concentrations potently increased both nuclear Trx1 and nuclear TrxR1 levels in TCA8113 cells. EGCG readily undergoes auto-oxidation to produce ROS and EGCG quinone, resulting in oxidative stress and formation of quinoproteins [30,55]. Such a prooxidant action of EGCG is viewed as a crucial anti-cancer mechanism [27,28,55,68]. Although the cancer-inhibitory effect of EGCG has been extensively studied [27,28,68], never has quinoprotein been examined in cancer cells subjected to EGCG treatment. The covalent binding of highly reactive EGCG quinone and protein containing free cysteine residue(s) leads to stable accumulation of quinoproteins, suggesting that quinoprotein is a biomarker signature of EGCG and could mirror EGCG-produced oxidative stress. The current study showed that p21 and quinoprotein levels were positively associated with melatonin-and EGCG-induced cell death, respectively. Cancer cell death caused by melatonin or EGCG is believed to be a consequence of oxidative stress [4,69,70]. Thus, p21 and quinoprotein could respectively reflect melatonin-and EGCG-triggered oxidative stress in TCA8113 cells. The concomitant increase of p21 and quinoprotein levels after the co-treatment indicated that two distinct anti-cancer routes resulted from each single agent were integrated and suggest that the two agents together could generate enhanced anti-cancer activity.
Although the up-regulation of p21 by melatonin is a putative anti-cancer mechanism, the down-regulation of p21 by melatonin at oncostatic concentrations has been shown in rat glioma cells and human gastric cancer cells [71,72]. However, the biological importance of such an opposite effect of melatonin on p21 is poorly understood. It is well known that p21 has a strong inhibitory effect on cell cycle; however, it also plays a powerful anti-apoptotic role in some cancer cells and confers these cells with a survival advantage, making it an appropriate target for suppression by pharmacological agents [57,73]. Inhibition of p21 by antisense oligodeoxynucleotide causes apoptosis in human breast cancer cells [74,75]. Compelling evidence demonstrates that attenuation of p21 in cancer cells make a battery of DNA-damaging chemotherapeutic agents more effective [57][58][59][60][61]. HCC is a highly malignant tumor and is refractory to most chemotherapeutic agents [76]. Sorafenib is a first-line chemotherapeutic drug for patients with advanced HCC [77]. It has been documented that sorafenib can markedly reduce p21 levels in tumors or cancer cells including HepG2 cells [78,79], suggesting that p21 suppression plays a role in for HCC therapy. The present study showed that melatonin dramatically suppressed p21 in HepG2 cells, which could be ascribed to melatonin-induced reduction of p53 ( Figure 6A). We further found that melatonin-induced reduction of p53 in HepG2 cells was associated with melatonin-triggered increase of MDM2 (Supplementary Figure S3). Thus, melatonin-induced reduction of p53 possibly involved p53 degradation by proteasome via MDM2-mediated ubiquitination of p53. Currently there is ample evidence indicating that melatonin is a highly appealing candidate for modulating HCC to achieve a successful chemotherapy [80][81][82][83][84][85]. In this setting, the down-regulation of p21 by melatonin as evidenced by the present study may play a pivotal role. Thus, the pursuit using melatonin as an adjunct agent for HCC therapy may be promising. Two earlier studies indicated that melatonin up-regulated p21 in HepG2 cells [86,87], whereas the present study found the opposite effect. The discrepancy may result from different cell density and incubation time. We carried out the experiments when HepG2 reached confluence and investigated the influence of a 24-h incubation of melatonin on p21. The results that melatonin increased p21 [86,87] were obtained at much higher cell density after a 48-h incubation.
The present study showed that while melatonin down-regulated p21 and caused moderate cytotoxicity, pro-survival PI3K and COX-2 proteins were up-regulated in HepG2 cells. Perhaps such a drug resistant response attenuated the cytotoxic effect of melatonin. EGCG reduced the expression of survivin, Bcl2 and NFκB. Nonetheless, oncostatic activity of EGCG by itself was inadequate. When the drugs were combined, (1) melatonin probably sensitized HepG2 cells to EGCG toxicity via suppressing p21; (2) EGCG inhibited melatonin-induced pro-survival responses; (3) COX-2 and HO-1 were significantly suppressed only by the co-treatment; and (4) melatonin aided EGCG to achieve an increased inhibition of Bcl2 (from 30% to 50%) and NFκB (from 50% to 80%). These molecular pathways together resulted in enhanced cytotoxicity in the co-treatment.

Conclusions
EGCG has shown anti-cancer effects in many studies. However, high-dose EGCG is known to cause hepatotoxicity. Our previous study has demonstrated that melatonin reduces hepatotoxicity triggered by high-dose EGCG [50], making intake of EGCG at high levels more tolerable. The present study reveals that co-treatment of melatonin and EGCG in cancer cells is more potent than each single agent in reducing the number of viable cells, inhibiting cell migration and suppressing colony formation. Taking both the safety and anti-cancer effect of EGCG into account, melatonin plays dual roles in reducing hepatotoxicity risk of EGCG and increasing the anti-cancer potential of EGCG; thus, melatonin and EGCG appear to be ideal partners in fighting cancer.

Conflicts of Interest:
The authors have no conflict of interest to declare.