GA, N-acetyl-cysteine (NAC), staurosporine (STS), hydrogen peroxide (H₂O₂), N-phenylmaleimide (NPM), 3 methyladenine (3MA), JC-1 and anti-β-actin (A5411) were purchased from Sigma Co. (St. Louis, MO, USA). Anti-AIF (113306), anti-BAX (109683), anti-BCL2 (100064), anti-PARP1 (100573), anti-porin (114187), anti-histone (122148) and z-VAD-FMK were purchased from GeneTex (ICON-GeneTex Inc., Taipei, Taiwan); anti-caspase-8 (ab138485), caspase-9 (ab115161), anti-caspase-3 (ab90437), anti-LC3 (ab48394), anti-beclin 1 (ab62557) and anti-P62 (ab155686) were purchased from Abcam (Cambridge, UK).

MDA-MB-231 human breast carcinoma cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) (Sigma Co.) supplemented with 10% fetal bovine serum (FBS) (v/v), 100 U/ml penicillin/streptomycin/amphotericin (Sigma Co.) and 0.1 M sodium bicarbonate. Cells were maintained in a humidified incubator at 37°C with 5% CO₂ and passage by 0.25% trypsin-EDTA every 2–3 days.

Cell viability was analyzed using Cell Counting Kit-8 (CCK-8) that detects the metabolic activity of cells. At the end of the various treatments, 10 µl of the CCK-8 reagent was added to each well (at 1×10⁴ cells/well), and the cells were then incubated at 37°C for 4 h. Absorbance was recorded using an ELISA microplate reader at 450 nm.

Cell morphology was studied by Hoechst 33258 and propidium iodide (PI) double staining for live-cell images. After treatment, cells (at 1×10⁵ cells/well) were stained with Hoechst 33258 and PI solution at room temperature for 10 min. Cells were washed twice with phosphate-buffered saline (PBS) for analysis by an inverted fluorescence microscope (Carl Zeiss Axiovert M200; Carl Zeiss, Jena, Germany).

Cell survival was assessed by labeling cells with PI. Briefly, the cells (at 1×10⁵ cells/well) were washed twice with PBS, re-suspended in a binding buffer and stained with 1 µg/ml of a PI solution (Sigma Co.) for 15 min at room temperature in the dark. After incubation, the cells were analyzed using flow cytometry.

Intracellular production of ROS, namely hydrogen peroxide was measured using DCFH-DA (Enzo Life Sciences, Inc., Farmingdale, NY, USA). Cells (at 1×10⁵ cells/well) from each experimental condition were stained with DCFH-DA (2 µM) at 37°C for 30 min. ROS production of cells was evaluated by flow cytometry (FACSCalibur; Bioscience, Tokyo, Japan). Values were expressed relative to the fluorescence signal of the control.

Cells (at 1×10⁵ cells/well) from each experimental condition were stained with JC-1 (10 µg/ml) reagent and MitoSOX Red (Molecular Probes Inc., Eugene, OR, USA) at 37°C for 30 min for detection of the mitochondrial membrane potential and mitochondrial ROS, respectively. Cells were washed twice with PBS, trypsinized, and collected, and the pellet was re-suspended in PBS for analysis using flow cytometry.

The immunoassay was performed as previously described (28).

Caspase-8, −9 and −3 activities were measured by the colorimetric activity assay kits (Chemicon International, Temecula, CA, USA). The assay is based on cleavage of the chromogenic substrates, DEVD-pNA and LEHD-pNA, by caspase-3 and −9, respectively. Cells (at 1×10⁴ cells/well) were lysed with the chilled lysis buffer on ice for 10 min and centrifuged for 5 min at 10,000 × g. Supernatant of each treatment was transferred to a fresh tube and equal protein concentration was adjusted to that of the positive control before incubating with the caspase inhibitor in a 96-well microplate for 10 min. Caspase substrate solution with specific peptide substrate was then added into the supernatant and incubated for 2 h at 37°C before measuring by ELISA reader at 405 nm.

Analysis of the cell death of each condition was performed using an Alexa Fluor Annexin V/Dead Cell Apoptosis kit (Molecular Probes Inc.) according to the manufacturer's protocol. Briefly, cells (at 1×10⁵ cells/well) were harvested after treatment. Cells were then washed twice with cold binding buffer, resuspended in binding buffer and stained with 5 µl each of Annexin V-FITC and PI solution for 15 min at room temperature in the dark. After incubation, 1 ml binding buffer was added, and cells were analyzed by flow cytometry.

The intensity of bands in western blot analyses or fluorescence was quantified using ImageJ software (NIH) and Zeiss Zen software. The intensity value for each protein was normalized against the intensity of the loading control for that same sample. The values after normalizing to the loading control in the control groups were set as 1.0. Data presented are the mean ± standard error of the mean (SEM) from at least 3 independent experiments and were analyzed using a Students t-test with two-tailed distribution between groups as indicated in the graphs. All calculations were performed using GraphPad Prism 6.01 version.

The effect of GA on cell survival is shown in Fig. 1A. The cytotoxicity of the GA was measured by CCK-8 assay. As shown in Fig. 1A, GA exerted a significant cytotoxic effect at concentrations 2.5–20 mM at 24 h and the IC₅₀ value was found to be 20 µM. These results indicated that the GA treatment significantly reduced cancer cell viability in a dose-dependent manner. To further evaluate the cytotoxic effect of GA-induced breast cancer cell death, cells were stained with DAPI after exposure to GA for 24 h. Significant change in blebbing, cell shrinkage, DNA condensation was observed in the MDA-MB-231 breast cancer cells; an increase in fluorescence intensity was observed by 2.5 D reconstruction image after treatment with GA (Fig. 1B). As shown in Fig. 1C, GA markedly increased the percentage of PI-positive cells by live-cell fluorescent. In addition, cell cycle analyses showed an increase in the subG1 phase population at 24 h in the GA treatment groups. The increased subG1 population of GA-treated breast cells at 24 h coincided with that of the cell viability assay (Fig. 1D).

To investigate the effect of GA on intracellular ROS generation, MDA-MB-231 cells were treated with GA at different concentrations and stained with fluorescent probe DCFH-DA and MitoSOX, which detect hydroperoxide and mitochondrial ROS, respectively. Cells following GA treatment showed higher green and red fluorescence compared with that in the normal control group (Fig. 2A and B). A flow cytometric analysis was performed to quantify the effect of GA-induced intracellular and mitochondrial ROS generation, and the results are presented in Fig. 2A and B. GA induced a significant shift in green and red fluorescence, whereas following GA, enhanced total ROS and mitochondrial ROS production were well apparent. To investigate whether ROS were involved in the GA-induced cell death, we introduced NAC, an inhibitor that blocks ROS generation, before GA treatment. The results showed that NAC alone had no effect on the mitochondrial ROS. However, NAC treatment blocked the GA-induced mitochondrial ROS in MDA-MB-231 cells (Fig. 2C). Moreover, there was increased in cell viability after treatment with GA and NAC (Fig. 2D).

Annexin V/PI double-staining was used to detect the PS out-flipping phenomenon, one of the clearest characteristics of apoptotic cell death. To examine whether inhibition of GA-induced cell proliferation was associated with apoptosis, cells were detected after 24 h of GA exposure by flow cytometry. GA reduced the number of viable MDA-MB-231 breast cancer cells from 87 to 60% in a dose-dependent manner. However, we found that the cell population shifted directly to PI⁺ quadrants without going through the PI⁻/Annexin V⁺ transition. This observation suggests that cell death induced by GA differs from apoptosis in MDA-MB-231 cells, apparently relying mainly on necrosis or autophagy (Fig. 3A). Mitochondrial integrity or the loss of Δψm has been linked to initiation of apoptosis and multiple cell death, including autophagy and necrosis. To determine Δψm loss, breast cancer cells were stained with JC-1 when harvested and subjected to flow cytometric analysis. JC-1 concentrates in the mitochondria as red fluorescent aggregates at high membrane potentials and is converted to green monomers when Δψm is lost. Red fluorescence in MDA-MB-231 breast cancer cells after GA treatment was reduced from 99 to 64%, indicating mitochondrial membrane potential (MMP) depolarization in breast cancer cells (Fig. 3B). To determine whether GA-induced apoptosis, characterized mainly by activation of caspase-3-associated proteins leading to cleavage of PARP1, protein expression was measured by western blotting (Fig. 3C). Along with decreased levels of PARP1 and a significantly increased BAX/BCL2 ratio was observed after GA treatment (Fig. 3D). No changes in cleaved caspase-3, −9 and −8 expression and activity were found in MDA-MB-231 breast cancer cells (Fig. 3C and E). To corroborate that GA-induced cell death was independent of caspase, the cells were pre-treated with 1 mM pan-caspase inhibitor z-VAD-FMK for 2 h. No significant recovery in cell viability was observed after inhibitor pre-treatment in the MDA-MB-231 cells, indicating that GA induced cell death in a caspase-independent manner (Fig. 3F).

To further investigate the cellular mechanism of GA-induced cell death, the protein levels of AIF were detected. AIF is a FAD-dependent oxidoreductase that has a vital role in oxidative phosphorylation and is independent of the caspase pathway. We investigated whether the translocation of AIF occurred in the GA-induced cell death of MDA-MB-231 cells. As shown in Fig. 4A and B, addition of GA resulted in a significant increase in overlap coefficient between nuclear AIF. AIF translocation occurred and subsequently appeared in the nuclear fraction of MDA-MB-231 cells treated with GA for 24 h. We also confirmed and quantified the cellular distribution of AIF. Immunoblotting revealed that AIF mostly resided in the mitochondria of the cancer cells (Fig. 4C). To investigate whether the AIF pathway is involved in the GA-induced cell death, we introduced NPM, an inhibitor that blocks AIF activation, before GA treatment. The results showed that NPM combined with GA had a significant increased on the expression of AIF and PARP1 (Fig. 4D). After the treatment of NPM, JC-1 gathered in the mitochondrial matrix and produced red fluorescence (Fig. 4E). As shown in the results of flow cytometry, GA caused a dramatic reduction in red fluorescence which indicated a loss of mitochondrial membrane potential (ΔΨm) and damage of mitochondria in the MDA-MB-231 cells (Fig. 4F). The ratio of red and green fluorescence represented the level of depolarization in mitochondria: NPM protected MDA-MB-231 cells against the damage caused by GA treatment (Fig. 4G). Moreover, there was increased cell viability after treatment with NPM, indicating that AIF may be the major factor involved in GA-induced cell death (Fig. 4H). Furthermore, NAC treatment also blocked the GA-induced AIF translocation to the nucleus, suggesting that ROS may be the regulator involved in this process (Fig. 4I).

Autophagic protein LC-3 is a component of double-membrane vesicle formation into autophagosomes. To explore whether autophagy was involved in GA-induced cell death, the fluorescence expression was utilized. A large number of bubbles were observed in the GA-exposed cells after 24 h (Fig. 5A). Moreover, a number of autophagosomes or lysosomes containing segregated were observed in the GA-treated cells. Conversion from cytosolic LC3-I to LC3-II, the membrane-binding form, was analyzed by western blotting. The protein expression of LC3-II was significantly increased in the MDA-MB-231 breast cancer cells and p62 was downregulated, which is one of the selective substrates for autophagy. These results suggested that AIF and LC3-II may be involved in GA-induced autophagy (Fig. 5B and C). Autophagy is a major process of organelle and protein degradation. An increasing number of studies have shown that autophagy may be a new target for anticancer therapy. To further confirm that autophagy was involved in GA-induced cell death, 2 mM 3 methylamphetamine (3MA), an autophagy blocker that inhibits PI3K class III protein, was added. After addition of 3MA, the increase in LC3 was significantly inhibited compared with the control cells (Fig. 5D). Considering all of the above results, we conclude that GA-induced cell death in MDA-MB-231 cells was mainly due to autophagy. Moreover, there was increased in cell viability after treatment with 3MA, indicating that autophagy may be the major factor involved in GA-induced cell death (Fig. 5E). We further determined the involvement of ROS in GA-induced autophagy, and protein expression of associated signaling molecules (Fig. 5F). NAC reversed the effects of GA on protein expression of LC3 and P62. However, treatment with AIF inhibitor NPM showed no significant effects (Fig. 5G). These results suggested that the anticancer effects of GA on breast cancer cells were mediated by ROS-dependent mitochondrial dysfunction with the involvement of AIF and LC3 signaling pathway.