Autophagy and Apoptosis in Hepatocellular Carcinoma Induced by EF25-(GSH)2: A Novel Curcumin Analog

Curcumin, a spice component as well as a traditional Asian medicine, has been reported to inhibit proliferation of a variety of cancer cells but is limited in application due to its low potency and bioavailability. Here, we have assessed the therapeutic effects of a novel and water soluble curcumin analog, 3,5-bis(2-hydroxybenzylidene)tetrahydro-4H-pyran-4-one glutathione conjugate [EF25-(GSH)2], on hepatoma cells. Using the MTT and colony formation assays, we determined that EF25-(GSH)2 drastically inhibits the proliferation of hepatoma cell line HepG2 with minimal cytotoxicity for the immortalized human hepatic cell line HL-7702. Significantly, EF25-(GSH)2 suppressed growth of HepG2 xenografts in mice with no observed toxicity to the animals. Mechanistic investigation revealed that EF25-(GSH)2 induces autophagy by means of a biphasic mechanism. Low concentrations (<5 µmol/L) induced autophagy with reversible and moderate cytoplasmic vacuolization, while high concentrations (>10 µmol/L) triggered an arrested autophagy process with irreversible and extensive cytoplasmic vacuolization. Prolonged treatment with EF25-(GSH)2 induced cell death through both an apoptosis-dependent and a non-apoptotic mechanism. Chloroquine, a late stage inhibitor of autophagy which promoted cytoplasmic vacuolization, led to significantly enhanced apoptosis and cytotoxicity when combined with EF25-(GSH)2. Taken together, these data imply a fail-safe mechanism regulated by autophagy in the action of EF25-(GSH)2, suggesting the therapeutic potential of the novel curcumin analog against hepatocellular carcinoma (HCC), while offering a novel and effective combination strategy with chloroquine for the treatment of patients with HCC.


Introduction
Hepatocellular carcinoma (HCC) is the fifth most common cancer and the third leading cause of cancer death worldwide [1]. Certain regions in Asia and Africa are disproportionally affected, while China alone accounted for half of the new liver cancer cases occurring worldwide during 2008 [2]. Chemotherapy plays a crucial role in the treatment of HCC especially at advanced stages when curative therapies like resection and liver transplantation are inapplicable [3,4]. However, since most widely used chemotherapeutic drugs show severe side effects, development of novel and safe agents is mandatory.
Curcumin, a natural compound isolated from the commonly used spice turmeric, has been shown to inhibit cell proliferation in various types of cancer cells in vitro and in vivo [5]. Numerous curcumin derivatives and analogues have been developed in recent years in order to enhance anti-tumor efficacy and overcome limitations such as poor aqueous solubility, relative low bioavailability and intense yellow staining [6,7]. The compounds 3,5-bis(2flurobenzylidene)piperidin-4-one (EF24) and 3,5-bis(pyridin-2ylmethylene)piperidin-4-one (EF31), synthetic structural analogues of curcumin [8], exhibit improved anticancer activity and a safety profile similar to curcumin [8][9][10][11]. Synthetic manipulation of these agents generates EF24-(GSH) 2 and EF31-(GSH) 2 , double glutathione conjugates with no less anticancer capability compared to EF24 and EF31. However the conjugates exhibit superior stability in solution, water solubility and lack of color [8]. The structurally related compound 3,5-bis(2-hydroxybenzylidene)tetrahydro-4Hpyran-4-one (EF25) and its double glutathione conjugate EF25-(GSH) 2 , are under investigation and reported here for the first time.
Although much of the research into the anti-cancer mechanisms of curcumin has focused on its ability to induce apoptosis, curcumin has also been found to induce other types of cell death including autophagic cell death, mitosis catastrophe and paraptosis [5,[12][13][14]. By testing different cell lines, it has been found that the mode of cell death induced by curcumin varies among different cell lines, and the mechanisms of different cellular responses remains a mystery [14].
Here we show that in vitro EF25-(GSH) 2 exhibits preferential toxicity to malignant liver cancer cells compared with immortalized human hepatic cells. In parallel, in vivo EF25-(GSH) 2 significantly suppresses the growth of hepatocellular carcinoma (HepG2) xenografts and is relatively nontoxic to mice. Further investigation into the mechanism of action reveals that EF25-(GSH) 2 induces a mixed mode of cell death in hepatoma cells in which autophagy, cell cycle arrest, cytoplasmic vacuolization, caspase-dependent and caspase-independent apoptosis all take place.

Ethics Statement
All procedures involving mice were approved by Anhui Medical University Animal Care Committee, which follows the protocol outlined in The Guide for the Care and Use of Laboratory Animals published by the USA National Institute of Health (NIH publication No. 85-23, revised 1996). The details of animal welfare and steps taken to ameliorate suffering were in accordance with the recommendations in The Guide for the Care and Use of Laboratory Animals, and all efforts were made to minimize suffering.

Synthesis of EF25 and EF25-(GSH) 2
EF25 was prepared as previously reported where it was originally named ''compound 11'' [15], while EF25-(GSH) 2 was obtained by a procedure identical to that for EF24-(GSH) 2 [8]. It should be noted that EF25 combined with glutathione much more slowly by comparison with EF24. EF25 (64.0 mg, 0.2 mmol, 1.0 eq.) in CH 3 CN (0.2 ml) was added dropwise to a solution of GSH (123.0 mg, 0.4 mmol, 2.0 eq.) in water at room temperature. The reaction mixture was refluxed for 2 hr until the disappearance of both the yellow color and EF25 by LC/MS. Evaporation of the solvent delivered the product as a white powder in quantitative yield. HR-ESI-MS (m/z): [M+H] + calcd for C 39 H 51 O 16 N 6 S 2 923.28053, found, 923.28121 ( = 00068 amu) (Fig. S1).
The 1 H NMR spectrum of EF25-(GSH) 2 in DMSO-d6 and D 2 O (buffer pH7) are complex due to the presence of diastereoisomers resulting from GSH conjugation at the two C = C bonds of EF25. The 1 H NMR spectrum of the unconjugated EF25 in DMSO-d6 exhibits a sharp singlet at 7.89 ppm assigned to the olefinic(C = )C-H proton and sharp aromatic signals at 6.8-7.3 ppm. The intensity of the olefinic signal decreases for the conjugated EF25-(GSH) x , and the sharp aromatic signals observed for unconjugated EF25 are broadened for EF25-(GSH) x . These observations indicate a mixture of the mono-and bis-conjugates EF25-(GSH) and EF25-(GSH) 2 , respectively, and possibly rapid exchange between them (Fig. S2). The comparison of the 1 H NMR spectra of EF25 in DMSO-d6 and EF25-(GSH) 2 in D 2 O (pH7) illustrates the absence of observable quantities of unconjugated EF25 (Fig. S3). Thus, in these solvents, the equilibrium lies primarily on the side of the conjugates, although in biological tissues it is shifted to the unconjugated form as the hydrophobic EF25 interacts with its target proteins.

Cell viability assay
Cells (8610 3 per well) were seeded onto 96-well plates in supplemented DMEM and incubated overnight. Then the cells were treated in triplicate for the indicated time with increasing doses of EF25-(GSH) 2 in 10% FBS containing DMEM or RPMI 1640 without antibiotic. Treated cells were then incubated in the presence of 0.5 mg/mL 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide (MTT) for 4 h. The formazan crystals were dissolved in DMSO and monitored at an absorbance of 490 nm. Absorbance values were normalized to those obtained for the untreated cells to determine percentage survival. All experiments were repeated at least three times. IC 50 values (50% inhibition concentration) were then calculated using the Statistical Package for the Social Sciences (SPSS, Inc.).

Colony formation assay
Twenty-four-well plates were seeded with 500 viable cells in complete medium and incubated overnight. The cells were then treated in triplicate with EF25-(GSH) 2 in 10% FBS containing DMEM without antibiotic for 24 h. The compound-containing medium was then removed, and the cells were washed with PBS twice and incubated in complete medium for another two weeks. Medium was replaced once at the end of the first week. The cell colonies formed were fixed in 10% formalin for 10 min and visualized by staining with Giemsa [22]. 7. DNA content analysis 5610 6 HepG2 cells were seeded into six-well plates and incubated overnight. The cells were treated with EF25-(GSH) 2 and then collected by trypsinization and fixed in precooled 70% ethanol overnight. Cells were then stained with 50 mg/mL propidium iodide (PI) in the presence of 100 mg/mL RNase A. DNA content was analyzed by FACSCalibur (Becton Dickinson), and data were analyzed by the Flowjo software. The percentage of cells in sub-G 1 -G 0 was used to represent the apoptosis rate.

HepG2 cell tumor xenograft in mice
Five-week-old male athymic mice were obtained from Beijing Vital River Laboratory Animal Co., Ltd. Animals were given ad libitum access to water and standard mouse chow. 5610 6 HepG2 cells were injected subcutaneous into the left flank and allowed to form a xenograft. Treatment was initiated when the tumor reached a group mean of 100 mm 3 . EF25-(GSH) 2 was dissolved in PBS and administrated i.p. daily at a dose of 1.5 mg/kg body weight for 30 days [23,24]. The control group was given the same  2 showed similar toxicity towards six human tumor cells (BEL-7402, HCT116, HepG2, A549, SMMC-7721 and Hela) (a) and the toxicity of curcumin was much lower than that of EF25-(GSH) 2 (b). c, cells were incubated with increasing doses of indicated compounds for 24-, 48-, and 72-h periods and analyzed by MTT assay. The IC 50 of each agent at each time period was calculated and compared using SPSS. The IC 50 of EF25-(GSH) 2 is much lower than that of curcumin and essentially equivalent to that of cisplatin. d, the cytotoxicity of EF25-(GSH) 2 to HL-7702 cells was much lower than that of cisplatin and similar to curcumin after 48-hour incubation as determined by MTT assay (*, p,0.01, **, p,0.001). (C) Cells were incubated with 0.5 mmol/L of the indicated compound for 24 h and subsequently allowed to grow into colonies (2 weeks). EF25-(GSH) 2 totally inhibited colony formation leading to clean plates, while curcumin and cisplatin did not. Results are representative of three independent experiments. doi:10.1371/journal.pone.0107876.g001 volume of PBS only. Tumor volume was calculated using the formula V = a 2 6b/2, where a and b represent the shorter and longer diameters of the tumor, respectively. At the end of the treatment, the mice were sacrificed under etherization, and the tumors were weighed.

Transmission electron microscopy
Treated cells were collected by trypsinization, washed twice with PBS, and then fixed with ice-cold 3% glutaraldehyde in 0.1 mol/L cacodylate buffer at 4uC overnight. Cells were then postfixed in osmium tetroxide and embedded in Polybed resin. Ultrathin sections were double stained with uranyl acetate and lead citrate and examined with a JEM-2100 electron microscope.

4, 6-diamidino-2-phenylindole (DAPI) staining
At the end of the EF25-(GSH) 2 treatment, DAPI was added to the medium at a final concentration of 1 mg/mL. After staining with DAPI for 15 min, cell morphology was examined by laser confocal microscopy. Uniformly stained nuclei with clear margins were regarded as normal, while condensed or fragmented nuclei with strengthened fluorescence were considered apoptotic.

Transient transfection
HepG2 cells were seeded in six-well plates and incubated overnight. mCherry-GFP -LC3 was transfected using Lipofectamine 2000 according to the manufacturer's instructions. The transfection mixture was replaced with 10% FBS containing DMEM without antibiotic 6 hours after transfection and incubated for another 24 hours. Cells were then treated with EF25-(GSH) 2 for the indicated time. Green (GFP) and red (mCherry) fluorescence was observed under a laser confocal microscope.

Western blot assay
Cell lysates were subjected to SDS-PAGE and blotted onto polyvinylidene difluoride (PVDF) membranes. The membranes were then incubated with the each primary antibody and appropriate secondary antibody. The immunoblots were visualized by a chemiluminescence HRP substrate.

Statistical analysis
All values are expressed as the mean6SE. Data were analyzed using two-tailed student's t test. P#0.05 was considered statistically significant. Statistical analysis was performed using SPSS.

EF25-(GSH) 2 inhibited proliferation of tumor cells
The structures of curcumin analogues, EF25 and EF25-(GSH) 2 , examined in this study are presented in Figure 1A. We first determined the effect of EF25, EF25-(GSH) 2 and curcumin on cell proliferation of HepG2 cells. Cisplatin, a widely used chemotherapeutic drug, was also examined under the same conditions. EF25-(GSH) 2 , which is far more effective than curcumin, showed similar cytotoxicity to cell lines derived from three types of hepatomas (HepG2, SMMC-7721 and BEL-7402) and three other carcinomas (HCT116, A549 and Hela) (Fig. 1B a, b). EF25 and EF25-(GSH) 2 exhibit similar cytotoxicity in a dose-and time-dependent manner, indicating that GSH association does not change the cell- kill capacity of the EF25 conjugate. The mechanism of this phenomenon was elucidated in our previous investigation showing that the conjugate is reversible [8]. Thus, the active agent in both cases would appear to be EF25. That the latter showed slightly better activity than EF25-(GSH) 2 at 24 h, and that the difference between them diminished as time prolonged to 72 h (Fig. 1B c), is mostly likely due to differential cell penetration regulated in part by the equilibrium shift from conjugate to free EF25. In effect, EF25-(GSH) 2 serves as a pro-drug capable of releasing the active agent EF25 by reversal of the well-known Michael reaction in cells [8]. The cytotxicity of EF25-(GSH) 2 to HepG2 cells is much greater than that of curcumin as characterized by its much lower IC 50 value (7.2 mmol/L at 48 h), which is close to that of cisplatin (9.1 mmol/L at 48 h) (Fig. 1B c). In order to examine if EF25-(GSH) 2 can preferentially kill malignant cells, cytotoxicities of the conjugate and cisplatin against immortalized human hepatic cell line HL-7702 were examined at 48 h post-treatment by the MTT assay. The results show that the cytotoxicity of EF25-(GSH) 2 to HL-7702 cells was much lower than that of cisplatin (Fig. 1B d).
To examine the long-term effect of EF25-(GSH) 2 treatment, the colony formation assay was performed. A quantity of 0.5 mmol/L EF25-(GSH) 2 totally inhibited colony formation, while 0.5 mmol/ L curcumin had nearly no effect. An 0.5 mmol/L cisplatin treatment also caused a significant reduction in colony number, but was much less efficiently by comparison with EF25-(GSH) 2 (Fig. 1C).

EF25-(GSH) 2 suppresses HepG2 xenograft growth
To assess the antitumor potential of EF25-(GSH) 2 in vivo, HepG2 xenograft-bearing mice were given EF25-(GSH) 2 i.p. daily for 30 days (1.5 mg/kg body weight, PBS as vehicle, n = 6). For comparison, cisplatin was dissolved in PBS and given i.p. every other day at a dose of 0.5 mg/kg body weight (n = 4). Compared with vehicle treatment (n = 6), EF25-(GSH) 2 treatment significantly suppressed the growth of tumor volume which was much more efficient than cisplatin treatment ( Fig. 2A, B). Notably, while the EF25-(GSH) 2 -treated group maintained normal weight gain, the cisplatin-treated animals suffered a remarkable weight loss throughout the treatment (Fig. 2C). The tumor weights of EF25-(GSH) 2 -treated mice were also significantly lower than that of the control group (Fig. 2D). There was no apparent change in liver, kidney and spleen weight in the EF25-(GSH) 2 -treated group, while  the weight of these organs dropped dramatically in the cisplatintreated group (data not shown).

The morphological appearance of EF25-(GSH) 2 -treated HepG2 cells
The morphological changes in EF25-(GSH) 2 -treated cells were observed under a light microscope to observe apparent vacuolization in the cytoplasm. The number of cells that suffered cytoplasmic vacuolization and its extent varied when treated with different concentrations of EF25-(GSH) 2 , but all reached a maximum at about 16 hours post-treatment (Fig. 3A). When treated with 5 mmol/L EF25-(GSH) 2 , the cells experienced moderate vacuolization which regained normal morphology after about 8 hours (Fig. 3C). In contrast, most cells exposed to 10 mmol/L EF25-(GSH) 2 showed extensive and irreversible vacuolization in the cytoplasm. At 20 mmol/L, EF25-(GSH) 2 not only induced massive cytoplasmic vacuolization but also caused apoptotic membrane blebbing (Fig. 3A).
In the cytoplasm of HepG2 cells treated with 10 or 20 mmol/L EF25-(GSH) 2 for 16 hours, large vacuoles of varying size were content-free and single membrane bounded, while the small vacuoles resemble autophagic vacuoles (Fig. 3D, E).

EF25-(GSH) 2 induced autophagy in HepG2 cells
The ultrastructural details of HepG2 cells treated with 20 mmol/L EF25-(GSH) 2 for 16 hours were further examined by transmission electron microscopy. Typical multimembrane autophagic vesicles engulfing cytoplasmic components and organelles were identified in the cytoplasm (Fig. 4).
To further confirm whether EF25-(GSH) 2 triggered autophagy in HepG2 cells, we examined the expression of the two forms of microtubule-associated protein 1 light chain 3 (LC3). In the process of autophagy, LC3-I residing in the cytosol is modified to LC3-II, which binds to the autophagosome membrane. Thus, the degree of LC3-I to LC3-II conversion correlates to the extent of autophagosome formation [25]. EF25-(GSH) 2 treatment obviously increased the expression level of both LC3-I and LC3-II as early as 12 hours post-treatment, but the bands corresponding to LC3-I were weakened and there was no obvious augmentation in the LC3-II expression when EF25-(GSH) 2 treatment was prolonged or the dosage was increased, indicating that the lack of conversion of LC3-I to LC3-II may due to incomplete autophagy (Fig. 5A).
The increase in LC3-II expression can be associated with either an enhanced formation of autophagosome or an impaired autophagic degradation [26]. Chloroquine (CQ) is a lysosomal trophic agent that raises the lysosomal pH and, hence, blocks autophagy at the late stages [27]. Accordingly, CQ was used to test if EF25-(GSH) 2 can induce complete autophagic flux [28,29]. In cells treated with 5 mmol/L EF25-(GSH) 2 , the LC3-II showed progressive accumulation in the presence of CQ at 24 h and 48 h. However, at 10 and 20 mmol/L, EF25-(GSH) 2 -treated samples with and without CQ were indistinguishable with respect to LC3-II expression (Fig. 5A). This data indicates that autophagy flux was achieved at 5 mmol/L EF25-(GSH) 2 but was blocked at 10 and 20 mmol/L.
In addition, we examined the localization of autophagosomespecific protein LC3B in HepG2 cells treated with EF25-(GSH) 2 for 24 hours using Cherry-GFP-LC3B plasmid. When autophagy is induced, exogenous LC3 distributes to the membrane of autophagosomes and shows characteristic green (GFP) or red (mCherry) dots. Because GFP is acid-labile, only mCherry red fluorescence can be seen in autophagolysosmes, while the neutral structures display both green and red fluorescence [30]. In untreated cells, mCherry-GFP-LC3B showed a homogeneous distribution, whereas the EF25-(GSH) 2 -treated cells showed fluorescent dots. At 5 mmol/L, the cells exhibit mostly only red dots, suggestive of autophagic degradation. Meanwhile, at 10 mmol/L, cells expressed doubletagged fusion proteins indicating that autophagic degradation was blocked (Fig. 5B). The data coincide well with the immunoblot analysis of LC3B in the presence of CQ.

EF25-(GSH) 2 induced G 2 /M cell cycle arrest and apoptosis in HepG2 cells
Curcumin and its analogs have consistently been reported to induce apoptosis [31,32]. To determine whether EF25-(GSH) 2 acts similarly in HepG2 cells, the DNA content of permeabilized PIstained cells was examined by flow cytometry at 24 h and 48 h posttreatment. The cell cycle analysis showed obvious G 2 /M cell cycle arrest at 24 h, and the percentage of cells in sub-G 1 -G 0 was greatly augmented at 48 h in a concentration-dependent manner (Fig. 6A).
DAPI staining of the nuclei also indicated that EF25-(GSH) 2treated cells underwent apoptosis, the extent of which was concentration dependent. Untreated HepG2 cells showed uniformly stained nuclei, while nuclei of EF25-(GSH) 2 -treated cells were condensed or fragmented with strengthened fluorescence (Fig. 6B).
These findings were further confirmed by analysis of the expression level of cleaved caspase-8 and caspase-3, both of which were augmented at 24 h post-treatment and maintained a high level up to 48 h at concentrations of 10 mmol/L and 20 mmol/L, whereas caspase activation was undetectable at 5 mmol/L (Fig. 6C).

Wortmannin advanced EF25-(GSH) 2 induced cell death in HepG2 cells in the early period
Autophagy modulation is a double edged sword in cancer treatment, possibly due to various cellular settings [33]. To test whether autophagy contributed to or hampered EF25-(GSH) 2 promoted HepG2 cell death, an inhibitor of autophagic sequestration (wortmannin (Wm)) was used to block autophagy at the early stages [26]. In the presence of 100 nmol/L Wm, the expression levels of both LC3B I and II types were largely reduced, indicating that Wm was effective in inhibiting EF25-(GSH) 2induced autophagy formation (Fig. 5C). Wm at 100 nmol/L was only slightly toxic to HepG2 cells but clearly promoted the EF25-(GSH) 2 -indued death process in the first 24 hours as evidenced by earlier cell shrinkage, rounding up (data not shown) and a 7-12% fall in cell viability examined by the MTT assay. However, as time progressed, the MTT assay at 48 h showed a slight increase rather than a further decrease of cell viability in Wm-pretreated cells. This indicates that Wm treatment advanced cell death only in the early period but had no obvious effect on the ultimate cytotoxicity of EF25-(GSH) 2 (Fig. 7A).

Knockdown of Atg5 and Beclin-1 expression did not rescue EF25-(GSH) 2 -treated HepG2 cells
In order to avoid the non-specific effect of Wm, we knocked down the cellular expression of two autophagy essential genes, Atg5 and Beclin-1, separately, using specific small hairpin RNAs (shRNA) delivered by the lentiviral expression system. The cells were transduced with lentivirus expressing the shRNA targeting LacZ, Atg5 or Becllin 1, and were selected with puromycin. Puromycin-selected cells were then treated with 10 mmol/L EF25-(GSH) 2 for 24 h. Atg5-and Beclin-1-knockdown was evident by reduced expression level of Atg5 and Beclin-1, respectively. Furthermore, both Atg5-and Beclin-1-knockdown resulted in the attenuated expression level of LC3II visualized with immunoblotting (Fig. 8A). The MTT assay showed no obvious distinction in cell viability between LacZ-knockdown and Atg5/Beclin-1knockdown HepG2 cells, which produced similar results with Wm (Fig. 8B). Furthermore, Atg5/Beclin-1-knockdown did not prevent the extensive cytoplasmic vacuolization induced by EF25-(GSH) 2 , suggesting that this phenomenon is not directly induced by autophagic degradation (Fig. 8C).

CQ promoted cytoplasmic vacuolization, apoptosis and cell death induced by EF25-(GSH) 2 in HepG2 cells
It has been previously reported that inhibition of autophagy at different stages of the process can lead to distinct results [34,35]. In our study, inhibition of autophagy at an early stage by Wm did not significantly alter either the extent of cytoplasmic vacuolization or the final cell viability at 48 h. However, the late stage inhibitor CQ not only advanced the cell death process but also significantly enhanced cytoplasmic vacuolization, apoptosis and cytotoxicity induced by EF25-(GSH) 2 . This combination effect of CQ was most dramatic by treatment of EF25-(GSH) 2 at 5 mmol/L.
When exposed to 5 mmol/L EF25-(GSH) 2 alone, only a small portion of cells was moderately vacuolated, but in the presence of CQ the majority of cells underwent extensive vacuolization to an extent similar to that caused by 10 mmol/L EF25-(GSH) 2 (Fig. 7B). This observation indicates that EF25-(GSH) 2 -induced autophagy exhibits a cytoprotective role at lower concentration.
The MTT assay showed that CQ enhanced the effectiveness of EF25-(GSH) 2 within a 24-h period, continued to 48 h and proved especially clear-cut at the concentration of 5 mmol/L where cell viability dramatically dropped from 52.4% to 14.5% after 48 h treatment (Fig. 7A).
We also found that combining CQ with EF25-(GSH) 2 greatly augmented the apoptosis rate evidenced by a large increase in the percentage of cells in the sub-G 1 -G 0 stage in a concentrationdependent manner (Fig. 6A).
In the presence of CQ, the expression level of activated caspase-3 increased at concentrations of 5 mmol/L and 10 mmol/L, but unexpectedly decreased at 20 mmol/L. Similarly, the expression level of cleaved caspase-8 was clearly increased by CQ treatment except at a concentration of 20 mmol/L EF25-(GSH) 2 at 48 h (Fig. 6C).

Z-VAD-FMK prolonged vacuolization and G2/M cell cycle arrest partially rescues HepG2 cells from EF25-(GSH)2 toxicity
To determine whether caspase activation plays a crucial role in EF25-(GSH) 2 -induced cytotoxicity, the pan caspase inhibitor Z-VAD-FMK was employed. In the presence of this compound at 48 h post-treatment, cell cycle analysis showed a clear decrease of cells in the sub-G 1 -G 0 stage especially at concentrations of 10 (from 37.9% to 16.7%) and 20 mmol/L (from 63.3% to 17.2%) (Fig. 6A). These data indicate that apoptosis induced by EF25-(GSH) 2 is primarily caspase-and concentration-dependent and partially caspase-independent to the extent of about 17%.
However, compared to the 21.2% (10 mmol/L) and 46.1% (20 mmol/L) decrease of apoptotic cells in the presence of Z-VAD-FMK at 48 h, only a 6.3% (10 mmol/L) and 19.3% (20 mmol/L) rise in cell viability was observed when pretreated with Z-VAD-FMK as examined by the MTT assay at 48 h (Fig. 7A), indicating the operation of non-apoptotic cell death. The latter was accompanied by prolonged cytoplasmic vacuolization and G 2 /M cell cycle arrest.
The extent of cytoplasmic vacuolization was not significantly enhanced in the presence of Z-VAD-FMK (Fig. 7B), but the period of vacuolization was prolonged beyond the 48 h treatment. Cells exposed to 10 mmol/L EF25-(GSH) 2 alone avoided vacuolization and begin to shrink at 48 h. In contrast, cells exposed to cotreatment of 10 mmol/L EF25-(GSH) 2 and 30 mmol/L Z-VAD-FMK exhibited extensive vacuolization (Fig. 7C).
Cell cycle analysis showed obvious G 2 /M cell cycle arrest in the presence of Z-VAD-FMK at 48 h post-treatment, which is similar to what was observed at 24 h post-treatment with EF25-(GSH) 2 alone, indicating that Z-VAD-FMK prolonged the status of G 2 /M cell cycle arrest induced by the EF25 conjugate (Fig. 6A).

Discussion
Effective and less toxic alternative chemotherapeutic agents against HCC are needed to address the emerging problem of drug resistance and severe side effects [2]. Widely used drugs like cisplatin exhibit no selectivity for malignant cells, while natural compounds like curcumin, which possess a good safety profile, exert inadequate effectiveness [7].
By contrast, our in vitro and in vivo data show the novel compound EF25-(GSH) 2 to exert preferential toxicity toward HCC cells, offering potential as a promising anti-HCC therapeutic agent. In addition, the double GSH conjugation successfully solves the problems of instability and water insolubility which limits the usage of curcumin and its analogues [8].
Basic and clinical studies have clearly established the importance of apoptosis in therapeutic tumor-cell death, but many notable studies have confirmed that other forms of cell death are crucial for effective cancer therapy, and that apoptosis is not the comprehensive answer especially when dealing with the whole tumor instead of isolated tumor cells [36]. We found that the action of EF25-(GSH) 2 is complex in terms of which death pathways are involved. In EF25-(GSH) 2 treated HepG2 cells, autophagy and apoptosis were detected and extensive cytoplasmic vacuolization was observed. These events do not occur independently, but are closely connected.
The role of autophagy in cancer therapy is complex and depends on the specific cellular setting and treatment scenario. Under some circumstances, autophagy rescues cells under stress conditions and, in this sense, may suppress apoptosis and/or other When treated with EF25-(GSH) 2 at concentrations of 5 mmol/L or lower, cells experienced full-scale autophagy that displayed moderate cytoplasmic vacuolization, ultimate recovery and partial rescue of cells from the resulting stress. In contrast, the protective autophagy was blocked in cells treated with EF25-(GSH) 2 at concentrations of 10 mmol/L or higher which led to massive cytoplasmic vacuolization. The latter cells arrested in the G 2 /M phase succumbed to both caspase-dependent and caspase-independent cell death. EF25-(GSH) 2 treatment alone led mainly to caspase-dependent apoptotic cell death, but also to a significant proportion of caspase-independent apoptosis. The action of EF25-(GSH) 2 could be modulated by CQ (green) and Z-VAD-FMK (blue). Co-treatment of EF25-(GSH) 2 with CQ promoted autophagy blockage and cytoplasmic vacuolization, which then enhanced apoptosis for both caspase-dependent and caspase-independent mechanisms. Co-treatment of EF25-(GSH) 2 with Z-VAD-FMK inhibited caspase activation and subsequently blocked the caspase-dependent apoptotic death route. Thus, cells were trapped by cytoplasmic vacuolization and G 2 /M cell cycle arrest, which eventually led to non-apoptotic cell death. doi:10.1371/journal.pone.0107876.g009 types of cell death. In other scenarios, irreversible self-destruction caused by massive autophagy leads to cell demise [33]. To investigate the exact role of autophagy in chemotherapy, autophagy inhibitors at different stages have been previously employed. Interestingly, the blockade of autophagy at an early or late stage has been reported by some groups to cause different effects. For example, the late stage inhibition by Bafilomycin A1 was found to enhance apoptosis and cell death, whereas inhibition of autophagy at early stages using 3-MA failed to do so [34,35].
Our autophagy inhibitor data using Wm and CQ also show different effects. Inhibition of autophagy at early stages by Wm advanced the cell death process during early phases of EF25-(GSH) 2 treatment, but altered the final toxicity insignificantly. In contrast, CQ greatly enhanced cytoplasmic vacuolization, apoptosis and cell death. These data suggest that autophagy does not directly execute cell death through extensive digestion of cellular cytoplasm, but exhibits a cytoprotective role and functions as a fail-safe response to the stressful condition induced by EF25-(GSH) 2 . In spite of this, protective autophagic degradation is only operative at low concentrations and is blocked by the action of the compound itself at higher and more cytotoxic concentrations.
However, we found that blocked autophagy contributes to cell death induced by EF25-(GSH) 2 . In EF25-(GSH) 2 -treated HepG2 cells, autophagy degradation blockage is accompanied by extensive cytoplasmic vacuolization. The latter phenomenon was found in tumor cells under various chemotherapeutic treatments. Although the cells present with a common morphology, various mechanisms were proposed [14,[37][38][39]. Hence, we conclude that accumulation of autophagosomes instead of autophagic degradation promotes the formation of extensive cytoplasmic vacuolization and subsequent cell death. In some effective cancer therapies, impaired autophagy has been observed [40], which may cause metabolic dysfunction and make cells more susceptible to other types of cell death. Notably, preclinical investigations combining the autophagy late stage inhibitor hydroxychloroquine (HCQ) with various chemotherapies has already entered clinical trials [41].
With EF25-(GSH) 2 alone, the number of cells within the G 2 /M stage was found to be augmented at a 24 h post-treatment and then largely diminished at 48 h when the number of apoptotic cells greatly soared. Non-apoptotic cell death occurs when the process from G 2 /M stage arrest to caspase-dependent apoptosis is blocked by Z-VAD-FMK, implying that cells arrested at G 2 /M have already reached a ''point of no return'' in the lethal process and that caspase activation may not be necessary. Notably, this result suggests that for cells failed to undergo apoptosis, EF25-(GSH) 2 induced cell death through non-apoptotic mechanisms, although less effectively without the participation of caspase activation.
As expected, activation of both caspase-3 and caspase-8 was enhanced under co-treatment of CQ and 5 mmol/L or 10 mmol/L EF25-(GSH) 2 . However, caspase activation was undetectable with co-treatment of CQ and 20 mmol/L EF25-(GSH) 2 , whereupon the apoptosis rate soared, indicating that the apoptosis in this scenario is mainly caspase-independent. Cell cycle analysis in the presence of the pan-caspase inhibitor Z-VAD-FMK suggests that EF25-(GSH) 2 alone causes mainly caspase-dependent apoptosis, but also partially caspase-independent apoptosis.
To sum up, we propose an anti-hepatoma mechanistic model for EF25-(GSH) 2 in Figure 9. When treated with EF25-(GSH) 2 at a concentration of no more than 5 mmol/L, cells experience successful autophagic degradation. In this case, moderate cytoplasmic vacuolization takes place followed by subsequent recovery, which partially rescues cells from a stressed condition. However, EF25-(GSH) 2 at a concentration of 10 mmol/L or higher leads to impaired autophagy during which the autophagic degradation step is blocked and followed by massive cytoplasmic vacuolization. At this point, cells undergo both caspase-dependent and caspase-independent apoptosis. EF25-(GSH) 2 treatment alone leads mainly to caspase-dependent apoptosis accompanied by partial caspase-independent apoptosis. Co-treatment with CQ stimulates autophagosome accumulation and cytoplasmic vacuolization, which then promotes both caspase-dependent and caspase-independent apoptosis. Z-VAD-FMK inhibits caspase activation and subsequently blocks the path to apoptotic death. In this case, the status of vacuolization and G 2 /M cell cycle arrest is prolonged and eventually leads to non-apoptotic cell death.
In conclusion, our results show that the novel curcumin analog EF25-(GSH) 2 has promising potential as a low toxicity chemotherapeutic agent for HCC. Similar to curcumin, the anti-tumor action of EF25-(GSH) 2 involved in autophagic, apoptotic and nonapoptotic mechanisms would broaden its application. The combination of EF25-(GSH) 2 with chloroquine is suggested to provide a safer and more effective treatment for HCC.