Shikonin induces mitochondria-mediated apoptosis and enhances chemotherapeutic sensitivity of gastric cancer through reactive oxygen species

The prognosis of gastric cancer remains poor due to clinical drug resistance. Novel drugs are urgently needed. Shikonin (SHK), a natural naphthoquinone, has been reported to trigger cell death and overcome drug resistance in anti-tumour therapy. In this study, we investigated the effectiveness and molecular mechanisms of SHK in treatment with gastric cancer. In vitro, SHK suppresses proliferation and triggers cell death of gastric cancer cells but leads minor damage to gastric epithelial cells. SHK induces the generation of intracellular reactive oxygen species (ROS), depolarizes the mitochondrial membrane potential (MMP) and ultimately triggers mitochondria-mediated apoptosis. We confirmed that SHK induces apoptosis of gastric cancer cells not only in a caspase-dependent manner which releases Cytochrome C and triggers the caspase cascade, but also in a caspase-independent manner which mediates the nuclear translocation of apoptosis-inducing factor and Endonuclease G. Furthermore, we demonstrated that SHK enhanced the chemotherapeutic sensitivity of 5-fluorouracil and oxaliplatin in vitro and in vivo. Taken together, our data show that SHK may be a novel therapeutic agent in the clinical treatment of gastric cancer.

SHK induces apoptosis in the human gastric cancer cells HGC-27 through mitochondria-mediated pathway 20 and causes cell cycle arrest in human gastric cancer (AGS) by early growth response 1 (Egr1)-mediated p21 gene expression 21 . However, studies about the effects of SHK on gastric cancer are insufficient and its precise anti-carcinogenic mechanisms against gastric cancer warrant further studies before it can be applied into the clinical setting.
Apoptosis, a highly regulated and controlled process, controls embryonic development by inducing the separation of fingers and toes, and removing infected cells to maintain internal environment stability 22,23 . Excessive apoptosis causes autoimmune diseases and insufficient apoptosis results in uncontrolled cell proliferation in tumourigenesis 24,25 . Apoptosis can be initiated through two pathways: the extrinsic cell death pathway and the intrinsic cell death pathway 26,27 . Both pathways induce cell death by activating caspase cascade 28 . Many chemotherapy drugs achieve anti-tumour effects by inducing apoptosis but drug resistance often occurs by caspase escape 29 . Recently, researchers discovered that cells can die and display apoptosis morphology without caspase activation via the translocation of apoptosis inducing factor (AIF) and Endonuclease G (Endo G) from the mitochondria to nucleus 30,31 . This sheds new light on overcoming the drug resistance for the tumours which are not sensitive to caspase dependent apoptosis. Therefore, inducing active cell apoptosis shows immense value in tumour therapy.
Reactive oxygen species (ROS) are oxygenic reactive chemical compounds formed as natural by-products of cellular metabolism 32 . ROS are two-edged swords in maintaining cellular homeostasis, which not only induce antimicrobial defence but also damage cell component 33,34 . The accumulations of intracellular ROS are the mechanism of most chemo-therapeutic and radio-therapeutic agents killing cancer cells 35 . SHK induces the generation of ROS and induces apoptosis in colon and hepatocellular cancer 14,36 . Among gastric cancer research, SHK time-dependently induced necrosis or apoptosis in gastric cancer cells via generation of ROS 37 .
In this study, we detected the anti-tumour effects of SHK on gastric cancer in vitro and in vivo. The results showed that SHK induces the generation of intracellular ROS, depolarizes the mitochondrial membrane potential (MMP), releases Cytochrome C from mitochondria and ultimately induces apoptosis via the caspase cascade. On the other hand, SHK induces caspase-independent apoptosis by the nuclear translocation of AIF and Endo G, which is first reported in our research. In addition, SHK enhances chemotherapeutic sensitivity of gastric cancer cells in vitro and in vivo. These results indicate that SHK is a novel inducer of apoptosis in gastric cancer and a promising candidate as a chemo-sensitizer for drug resistant gastric cancer patients.

SHK suppresses proliferation and induces death of gastric cancer cells but does small impacts for gastric epithelial cells.
The structure of SHK is presented in Fig. 1a. Immortalized gastric epithelial cell (GES-1) and gastric cancer cells (BGC-823 and SGC-7901) were treated with different concentrations of SHK. Shapes of BGC-823 and SGC-7901 cells became distorted, round and fragmented, but these effects were not significant in GES-1 cells (Fig. 1b). Cell viability detections showed that SHK suppresses cell proliferation in dose-dependent (Fig. 1c) and time-dependent (Fig. 1d,e,f) manners. The IC50 concentrations of SHK were 9.63 ± 0.43 μ M (SGC-7901), 13.35 ± 0.57 μ M (BGC-823) and more than 20 μ M (GES-1) at 24 h. Moreover, we found that increasing concentrations of SHK caused S-phase cell cycle arrest in SGC-7901 cells (Fig. S1). Cell death was quantified by Annexin V and PI staining. SHK significantly induced death of BGC-823 and SGC-7901 gastric cancer cells, but the effects were mildly for GES-1 cells (Fig. S2). Compared with BGC-823 cells, SGC-7901 Values are presented as the mean ± SD of eight independent experiments for MTT and three independent experiments for flow cytometry. *P < 0.05, **P < 0.01. cells were more susceptible to SHK and were chosen as the main research object. In addition, 10 μ M SHK was chosen as a normal experimental concentration, which was close to IC50, 5 and 20 μ M SHK were regard as low and high experimental concentrations, respectively. These data indicate that SHK is a suppressor of cell growth and inducer of cell death in gastric cancer cells.

SHK induces caspase-dependent apoptosis in SGC-7901 gastric cancer cells.
To detect the manner of death induced by SHK, we observed chromatin condensation by Hoechst staining in SGC-7901 cells (Fig. 2a). This means that SHK induces apoptosis in SGC-7901 gastric cancer cells. Furthermore, to determine whether the apoptosis induced by SHK was caspase-dependent, the pan-caspase inhibitor ZVAD-FMK was coadministered and SHK-induced cell death was significantly inhibited (Fig. 2b). The elevated lactate dehydrogenase (LDH) induced by SHK also decreased when ZVAD-FMK was added (Fig. 2c). SHK-induced splicing events that occurred in PARP, Caspase 8, 9 and 3 were also blocked by ZVAD-FMK (Fig. 2d). These data proved that the apoptosis induced by SHK is caspase-dependent. Meanwhile, Caspase 3 inhibitor ZDEVD-FMK and Caspase 9 inhibitor ZLEHD-FMK also strongly inhibited SHK-induced apoptosis of SGC-7901 gastric cancer cells (Fig. S3). Previous report 37 also found that SHK induced necrosis at 6 h and apoptosis at 24 h of AGS gastric cancer cells, which was time-dependent. We treated SHK for shorter time to detect whether SHK could induce necrosis in SGC-7901 gastric cancer cells. However, we did not find necrotic cells at 6 h and 12 h in SGC-7901 gastric cancer cells (Fig. S4), which may be due to the variability in cell types. Taken together, SHK activates caspase cascade and induces caspase-dependent apoptosis in gastric cancer cells.

SHK induces caspase-independent apoptosis by nuclear translocation of AIF and Endo G.
As is reported, AIF and Endo G play critical roles in caspase-independent apoptosis 30,31 . These proteins translocate to the nucleus and induce DNA degradation. Western blot showed the translocation of AIF and Endo G from the cytoplasm to nucleus with treatment of SHK (Fig. 3a), and this result was reaffirmed by immuno-fluorescence staining analyses (Fig. 3b). RNA interference of AIF and Endo G was utilized (Fig. 3c). AIF and Endo G knockdown significantly inhibited SHK-induced cell death (Fig. 3c) and the elevated LDH release (Fig. 3d). Meanwhile ZVAD-FMK had synergistic effects with AIF and Endo G knockdown (Fig. 3c). We also detected the DNA fragments and found that AIF and Endo G knockdown inhibited DNA laddering (Fig. S5). These data suggested that in addition to the caspase-dependent pathway, SHK also triggers caspase-independent apoptosis that is dependent on nuclear translocation of AIF and Endo G.

RIPK1 promotes survival of SGC-7901 cells in SHK-induced apoptosis.
Necroptosis is a programmed form of necrosis and SHK has been reported as a necroptosis inducer in previous studies 16,18,38 . Necroptosis is typically inhibited by the blockade of RIPK1. Although no necroptosis (PI+ , Annexin V+ ) was found, previous reports 18,19 indicated that there were necrotic cells in the late apoptosis stage of SHK-induced cell death. In addition, some reports 39,40 support that inhibition of RIPK1 enhances SHK-induced apoptosis. To detect whether there are necrotic components in the late apoptosis stage or RIPK1 could regulate apoptosis in SHK-induced gastric cancer cell death, the RIPK1 inhibitor Nec-1 was used. Interestingly, Nec-1 did not inhibit but increased SHK-induced cell death (Fig. 4a). This increase of cell death was inhibited by ZVAD-FMK (Fig. 4a). These data suggest that SHK does not induce necroptosis and Nec-1 promotes SHK-induced apoptosis in SGC-7901 cells. Western blot showed that RIPK1 protein levels were decreased after treatment with SHK ( Fig. 4b). We constructed over-expression plasmids and down-expression shRNA lentiviral particles of RIPK1. RIPK1 over-expression prevents SHK-induced apoptosis (Fig. 4c), whereas RIPK1 knockdown contributes to SHK-induced apoptosis (Fig. 4d). In addition, ZVAD-FMK had synergistic protective effects with RIPK1 over-expression (Fig. 4c), but RIPK1 knockdown robbed the protection of ZVAD-FMK in SHK-induced cell death (Fig. 4d). In order to detect whether RIPK1 knockdown could change cell death modality, we increased the concentration of ZVAD from 20 μ M to 50 μ M (Fig. S6). The high concentration of ZVAD had better protective effect, which indicated that the SHK-induced death modality in RIPK1 knockdown cells was apoptosis. Moreover, LDH levels ( Fig. S7) confirmed the same results with Fig. 4c,d. In conclusion, SHK does not induce necroptosis in gastric cancer cells, and the expression of RIPK1 has a protective role in SHK-induced apoptosis.
Intracellular ROS play an upstream role in SHK-induced apoptosis. The generation of ROS exhibit a close correlation with tumour cell death 36,41 . To identify roles of intracellular ROS in SHK-induced gastric cancer cell death, ROS scavengers N-acetyl-L-cysteine (NAC) and L-glutathione (GSH) were used. NAC and GSH significantly inhibited SHK-induced death (Fig. 5a). ROS were measured by DCF probes and exhibited dose-dependent and time-dependent manners in response to SHK treatment (Fig. 5b). Intracellular GSH and GSSG (oxidized forms of GSH) levels reflect redox state of cells, and the GSH/GSSG ratio was significantly reduced response to SHK treatment (Fig. 5c). In addition, the antioxidants NAC and GSH completely inhibited the generation of ROS (Fig. 5d). However, ZVAD-FMK, AIF, Endo G and RIPK1 played less or no roles in the regulation of ROS (Fig. S8). These data indicate that ROS play an upstream role of caspase, AIF, Endo G and RIPK1 in SHK-induced apoptosis. ROS typically promote mitochondrial dysfunction and mitochondria-mediated apoptosis [42][43][44] . We consider that ROS play an upstream role and mediate mitochondria-related apoptosis in gastric cancer cells. SHK depolarizes MMP and leads to mitochondria-mediated apoptosis through ROS. Accumulation of intracellular ROS typically depolarizes mitochondrial membrane potential (MMP) 45 . Thus, we detected the MMP of SGC-7901 gastric cancer cells after treatment of SHK. SHK depolarized MMP in a dose-dependent manner (Fig. 6a), and this effect could be blocked by ROS scavengers NAC and GSH (Fig. 6b). We also detected release of mitochondrial Cytochrome C (Fig. 6c); cleavage of caspase 8, 9, 3 and PARP (Fig. 6d); and nuclear translocation of AIF and Endonuclease G (Fig. 6e) by Western blot. However, all of these events could be partially or completely inhibited by the ROS scavengers NAC and GSH (Fig. 6c-e). The release of Cytochrome C, caspase cascade and the nuclear translocation of AIF and Endo G are the main forms of mitochondria-mediated apoptosis. These data indicate that the generation of ROS, depolarization of MMP and mitochondria-mediated apoptosis are components of a programmed event in SGC-7901 cancer cells treated with SHK.

SHK enhances chemotherapeutic sensitivity of 5-fluorouracil and oxaliplatin. 5-Fluorouracil
(5-Fu) and oxaliplatin (Oxa) are commonly-used chemotherapy drugs for gastric cancer. Drug resistance of cancer is a major cause of chemotherapy failure. We found that a low dose (5 μ M) of SHK exhibited significantly synergistic effects with 5-Fu and Oxa (Fig. 7a). However, SHK (5 μ M), 5-Fu and Oxa alone did not induce significant apoptosis, and the combination of SHK (5 μ M) with 5-Fu or Oxa increased cell apoptosis dramatically (Fig. 7b). We also found that the combination of low dose SHK with 5-Fu or Oxa increases intracellular ROS (Fig. 7c) and depolarizes MMP (Fig. 7d). These data suggest that even at a low dose, SHK enhances the chemotherapeutic sensitivity of gastric chemotherapy drugs. This effect may be dependent on the accumulation of intracellular ROS and depolarization of MMP.

SHK induces apoptosis of xenografts and enhances anticancer activities of chemotherapy drugs in vivo.
To further evaluate the anti-tumour effects of SHK in gastric cancer, we established immuno-deficient nude mice xenografts bearing SGC-7901 gastric cancer cells. The maximum diameter and weight of xenografts were measured. SHK inhibited the size and weight of xenografts ( Fig. 8a and b). We detected malondialdehyde (MDA) content of xenografts, which represents lipid peroxidation, and xenografts treated with SHK exhibited ROS accumulation in vivo (Fig. 8c). To evaluate whether SHK could induce apoptosis in xenografts, a TUNEL assay was performed. As is shown in Fig. 8d, TUNEL-positive cells were significantly increased with increasing concentrations of SHK. In addition, the combination of SHK with 5-Fu or Oxa increased tumour restriction rates in mice (Fig. 8e), indicating that SHK also enhanced the sensitivity to chemotherapy drugs in vivo. Consistent with the results in vitro, the results suggest that SHK induces cell death and enhance anti-tumour effects of chemotherapy drugs in vivo experiments. Pathways of SHK-induced apoptotic cell death in gastric cancer cells are shown in Fig. 8f.

Discussion
Cancer is the leading killer of human being with an estimated 14.1 million new cases and 8.2 million cancer deaths per year worldwide 3 . Searching for effective anti-tumour medicines has been a focus of researchers. Plants generally contain a variety of anti-tumour components, and many plant extracts have been used in the clinic to treat cancer, such as vincristine, paclitaxel, topside and topotecan. SHK is a naphthoquinone extracted from Chinese herbal medicine plant and is considered to be safe due to its long history use in Chinese traditional medicine. Recent studies confirmed its anti-tumour effects in osteosarcoma 46 , leukaemia 39 , colon cancer 14 , breast cancer 16 and glioma 19 . In this study, we found that SHK suppresses proliferation and induces death in SGC-7901 and BGC-823 gastric cancer cells in dose-and time-dependent manners. In contrast, this agent has minimal effects on immortalized gastric epithelial GES-1 cells. Kim et al. demonstrated that SHK induces cell cycle arrest in human AGS gastric cancer cells 21 and we also found that SHK caused S-phase cell cycle arrest in SGC-7901 and BGC-823 gastric cancer cells. Previous study also supported that SHK inhibits the cell viability, adhesion, invasion and migration of the human gastric cancer cell line MGC-803 47 . These data demonstrated the anti-gastric cancer effects of SHK.
ROS are double-edged swords for tumour. On the one hand, ROS cause severe DNA damages and gene mutations and active inflammation, proliferation, invasion, metastasis and angiogenesis of tumour cells 48 . On the other hand, once the levels of ROS reach the threshold, tumour growth is suppressed, cell structures break down and the cancer cells die. This is the mechanism by which most chemo-therapeutic and radio-therapeutic agents kill cancer cells 35 . In our study, we found that SHK induces the generation of intracellular ROS in time-and dose-dependent manners, depolarizes the MMP and ultimately induces mitochondria-mediated apoptosis. GSH is an important antioxidant in cellular metabolism and capable of preventing damages caused by ROS. Once oxidized, GSH is converted to its oxidized form GSSG. The ratio of GSH/GSSG is a measure of cellular oxidative stress 49 . We found that the GSH/GSSG ratio was significantly decreased when treated with SHK. In addition, as ROS scavengers, GSH and NAC inhibited SHK-induced cell death. However, the pan-caspase inhibitor ZVAD-FMK and knockdown of AIF and Endo G do not affect intracellular ROS, indicating that ROS play an upstream role in SHK-induced apoptosis.
The accumulation of ROS and depolarization of MMP may cause mitochondrial swelling and increase permeability of the mitochondrial membrane, which induces intrinsic apoptosis (also called mitochondria-mediated apoptosis) 28 . On the one hand, Cytochrome C leaks out from the perforated mitochondria and binds to apoptotic protease activating factor-1, forming apoptosomes. Then, apoptosomes activate caspase 9 and initiate the chain reaction of caspase cascade. These processes are defined as caspase dependent-apoptosis 50 . On the other hand, AIF and Endo G are released from perforated mitochondria, translocated into the nucleus and induce chromosomal DNA fragmentations. These processes are defined as caspase-independent apoptosis 51 . Our data showed that SHK induces apoptosis of gastric cancer cells via both the caspase-dependent and -independent pathways. SHK induces apoptotic cell morphology, release of Cytochrome C from mitochondria and cleavage of caspase 3, -9 and PARP, which is blocked by ZVAD-FMK. In addition, AIF and Endo G are translocated from the cytoplasm to the nucleus, which was affirmed by Western blots and immunofluorescence assays, and knockdown of AIF and Endo G inhibit SHK-induced apoptosis. These data confirm that SHK induces caspase-independent apoptosis in gastric cancer cells. Many cancer cells are not sensitive to drugs due to caspase escape and caspase-independent apoptosis is a new strategy to overcome drug resistance. In addition, we also observed the activation of caspase 8, indicating that extrinsic apoptosis (death receptor signaling pathway) may be involved in SHK induced-apoptosis. This finding must to be further confirmed. RIPK1 is an enzyme that plays an important role in the NF-κ B pathway, apoptosis and necroptosis. In TNFR1-elicited signal pathways, RIPK1 is a constituent component of complex I, which includes RIPK1, TRADD, cIAPs, TRAF2 and TRAF5. In cIAP-mediated ubiquitylation, RIPK1 initiates NF-κ B pathways, which promote cell survival. RIPK1, RIPK3, caspase 8, TRADD and FADD form complex II. Caspase 8 cleaves RIPK1 and RIPK3 and triggers the caspase cascade to induce apoptosis. When caspase 8 is blocked, RIPK1 and RIPK3 become phosphorylated and trigger necroptosis 52 . Recent studies reported that RIPK1 mediates necroptosis induced by SHK in osteosarcoma, breast cancer and glioma 16,18,19 . According to previous report of Lee et al., SHK induced necrosis at 6 h and apoptosis at 24 h of gastric cancer cell AGS, which was time-dependent 37 . In our study, we did not find necrotic cells at 6 h and 12 h in SGC-7901 gastric cancer cells. Interestingly, we demonstrated that RIPK1 inhibitor Nec-1 does not rescue but increases SHK-induced cell death. RIPK1 Over-expression prevents SHK-induced cell death in gastric cancer cells, whereas RIPK1 knockdown contributes to it. These data suggest that SHK may not trigger RIPK1-dependent necroptosis in gastric cancer cells and RIPK1 plays a prosurvival role in SHK-induced apoptosis.
The ultimate aim of developing novel anti-tumour drugs is clinical application and developing chemo-sensitive strategies and agents is of great importance 8 . Multi-drug combinations are a novel approach to overcome drug resistance. We evaluated the effect of SHK combined with 5-Fu or Oxa and found that low dose SHK enhances anti-tumour effects. The molecular mechanisms of this synergistic effect may be involve the accumulation of ROS and depolarization of MMP. Furthermore, data also suggest that SHK suppresses tumour growth in xenograft animal models. SHK increases lipid peroxidation and induces apoptosis in vivo, which is consistent with the results in vitro. In addition, SHK enhances the anti-tumour activities of 5-Fu and Oxa in vivo.
In conclusion, SHK is an intracellular ROS inducer and we demonstrated its ability to suppress cell proliferation. SHK induces apoptosis in gastric cancer cells via caspase-dependent and -independent manners, which has never been reported in previous studies. In addition, we also demonstrate that SHK enhances chemotherapeutic sensitivity in gastric cancer both in vitro and in vivo. Therefore, this study suggests that SHK may be a novel therapeutic agent in the clinical treatment of gastric cancer.

Determination of cell viability and cell death.
MTT assay (Roche, Mannheim, Germany) was used to detect cell viability following the manufacturer's instructions. Cells were seeded in 96-well plates, cultured overnight and treated with drugs. MTT (5 mg/ml, Sigma) was added for 4 hour-incubation, removed supernatant and solubilized MTT using DMSO (Sigma, USA). The absorbance was measured by a 96-well plate reader (Bio-Tec Instrument) at 490 nm. Hoechst 33342 staining (Sigma-Aldrich, St. Louis, MO, USA) was used to analyze the nuclear morphology. Cells were incubated with Hoechst 33342 (5 μ g/ml, Sigma) and PI (10 μ g/ml, Sigma) at 37 °C, then washed in PBS and visualized with a fluorescence microscope (Nikon, Japan). Annexin V and PI staining kit (Beyotime Biotechnologies, Jiangsu, China) was applied to quantify the cell death. After treatment, cells were washed with PBS, resuspended in 200 μ l binding buffer, incubated with 5 μ l Annexin V-FITC and 10 μ l PI for 20 min at room temperature in the dark. The stained cells were quantified by flow cytometer (BD bioscience, USA).
Cell cycle analysis. Cells were seeded in 6-well plates and treated with shikonin for 24 h, then collected, washed with cold PBS and fixed with 70% ethanol overnight at 4 °C. Cells were incubated with PI (50 μ g/ml, Sigma) and RNase (100 μ g/ml, Sigma) for 1 h, then analysed by flow cytometer (BD bioscience, USA). LDH assay. LDH assay (Beyotime Biotechnologies, Jiangsu, China) was used to detect the cytotoxicity according to the manufacturer's protocol. Briefly, after treatment the media supernatants were transferred to a 96-well plate, added 50 μ l LDH detection reagent, incubated for 1 hour at 37 °C, added chromogenic agent, finally the absorbance was measured by a 96-well plate reader (Bio-Tec Instrument).

Isolation of nuclei and mitochondria.
A nuclear/cytosol fractionation kit (Beyotime Biotechnologies, Jiangsu, China) was used to isolate the nuclei according to the manufacturer's protocol. Briefly, cells were collected, resuspended with cytosol extraction reagent by vigorous vortex mixing for 30 min and centrifuged. The supernatant was collected as cytosolic fraction and the pellet was collected as nuclear fraction.
A mitochondria isolation kit (Beyotime Biotechnologies, Jiangsu, China) was used to isolate the mitochondria according to the manufacturer's protocol. Briefly, cells were collected, resuspended with extraction buffer, incubate on ice, homogenized the cells using a homogenizer for 30 strokes and centrifuged the homogenate at 600 g Scientific RepoRts | 6:38267 | DOI: 10.1038/srep38267 for 10 min at 4 °C. The supernatant was carefully transferred to a fresh tube and centrifuged at 11,000 g for 10 min at 4 °C. The supernatant was collected as cytosolic fraction and the pellet was collected as mitochondria fraction.
Immunofluorescence staining analyses. After treatment, cells were fixed with 4% paraformaldehyde, permeabilized with 0.5% triton X-100 for 15 min, blocked with PBS buffer containing 5% bovine serum albumin and washed with PBS three times. The cells were incubated with primary antibodies overnight at 4 °C and followed by 2 hour-incubation with FITC or Cy3 conjugated secondary antibodies. The nuclear was stained with DAPI (Sigma, USA) and images of immunostained cells were captured with a confocal microscope (Nikon, Japan) using the × 60/1.35 NA oil objective.
RNA interference, plasmid transfection and lentivirus infection. The siRNA against Endo G (target sequence of 5′-CCAUGGACGACACGUUCUA-3′), AIF (target sequence of 5′-GCAGUGGCAAGUUACUUAU-3′) and control siRNA (target sequence of 5′-UUCUCCGAACGUGUCACGU-3′ ) were synthesized by GenePharma Technologies (Shanghai, China). The RIPK1 plasmids were synthesized by RT-PCR and identified by gene sequencing. The RIPK1 shRNA sequence (GCAAAGACCTTACGAGAATTT) was synthesized by Genwiz Technologies (Suzhou, China) and constructed with pLKO.1 lentiviral vector plasmids. Procedures of RNA interference, plasmid transfection and lentivirus infection are previously described 53 . DNA laddering. DNA ladder was extracted from cells after treatment with Qiagen DNA Extraction Kit (Germany) by the manufacturer's recommendations and then visualized in an agarose gel by ethidium bromide staining.

Measurement of intracellular ROS level and mitochondrial membrane potential. A Reactive
Oxygen Species Assay Kit (Beyotime Technologies, Jiangsu, China) was used by the manufacturer's recommendations. After treatment, the cells were collected and incubated with DCFH-DA at 37 °C in the dark. DCF fluorescence intensity was measured by flow cytometry (BD bioscience, USA). And a Mitochondrial Membrane Potential Assay Kit (Beyotime Technologies, Jiangsu, China) according to the manufacturer's recommendations. After treatment, cells were incubated with JC-1 for 30 min at 37 °C in the dark. Then cells were visualized with a fluorescence microscope (Nikon, Japan) or quantified by FACS Calibur flow cytometer (BD bioscience, USA).

Measurement of intracellular GSH/GSSG. A GSH and GSSG Assay Kit (Beyotime Technologies,
Jiangsu, China) was used according to the manufacturer's recommendations. After treatment, cell lysates were obtained by three freeze-thaw cycles in liquid nitrogen. The mixture was centrifuged and the supernatant was collected for GSH and GSSG measurement.

Western blot. Whole cells and subcellular components were suspended in Laemmli Buffer (Bio-Rad
Laboratory, USA). Protein concentrations were measured using the BCA Protein Assay Kit (Pierce, USA). Equal amounts of protein were separated by SDS-PAGE and transferred to PVDF membrane (Bio-Rad Laboratory, USA). The membrane was blocked with 5% skim milk, and incubated with primary antibodies overnight at 4 °C and secondary antibodies for 2 hours at room temperature.

Establishment of xenografts.
Five-week-old BALB/c nude mice were purchased from Vital River (Beijing, China). Procedures involving animals and their care were conducted in conformity with NIH guidelines and was approved by Animal Care and Use Committee of Chinese PLA General Hospital. SGC-7901 cells were harvested, and a mixture of 5 × 106 cells, PBS and matrigel (Corning, NY, USA) was injected subcutaneously in the right oxter. After the xenografts were established, the tumour-bearing models were administered SHK via gastric infusion combined with or without 5-Fu or Ox via intraperitoneal injection. The diameter of xenografts was measured by calipers every three days. One month later, the mice were sacrificed, and tumours were weighed. The tumour restriction rate was calculated as follows: (1-Tumour Weight of Test Sample/Tumour Weight of Control) × 100%.
MDA and TUNEL assays. MDA Assay Kit (Beyotime Technologies, Jiangsu, China) was used to detect the malondialdehyde (MDA) content of xenografts according to the manufacturer's protocols. Tumour were homogenized by sonication extraction and normalized by BCA Protein Assay Kit (Pierce, IL, USA). A TUNEL Assay Kit (Roche, IN, USA) was used to detect the apoptosis of xenografts according to the manufacturer's protocols. Tumours were fixed by formalin and embedded by paraffin. DNA strand breaks were identified by TUNEL probes.