The Bcl-2 Homology Domain 3 Mimetic Gossypol Induces Both Beclin 1-dependent and Beclin 1-independent Cytoprotective Autophagy in Cancer Cells

Abstract

Gossypol, a natural Bcl-2 homology domain 3 mimetic compound isolated from cottonseeds, is currently being evaluated in clinical trials. Here, we provide evidence that gossypol induces autophagy followed by apoptotic cell death in both the MCF-7 human breast adenocarcinoma and HeLa cell lines. We first show that knockdown of the Bcl-2 homology domain 3-only protein Beclin 1 reduces gossypol-induced autophagy in MCF-7 cells, but not in HeLa cells. Gossypol inhibits the interaction between Beclin 1 and Bcl-2 (B-cell leukemia/lymphoma 2), antagonizes the inhibition of autophagy by Bcl-2, and hence stimulates autophagy. We then show that knockdown of Vps34 reduces gossypol-induced autophagy in both cell lines, and consistent with this, the phosphatidylinositol 3-phosphate-binding protein WIPI-1 is recruited to autophagosomal membranes. Further, Atg5 knockdown also reduces gossypol-mediated autophagy. We conclude that gossypol induces autophagy in both a canonical and a noncanonical manner. Notably, we found that gossypol-mediated apoptotic cell death was potentiated by treatment with the autophagy inhibitor wortmannin or with small interfering RNA against essential autophagy genes (Vps34, Beclin 1, and Atg5). Our findings support the notion that gossypol-induced autophagy is cytoprotective and not part of the cell death process induced by this compound.

Introduction

The recent development of small molecule antagonists of prosurvival Bcl-2 family proteins looks promising for the future treatment of cancer. These novel compounds, also known as BH3³ mimetics, bind to prosurvival Bcl-2 proteins and neutralize them. Gossypol is a natural polyphenol compound isolated from cottonseeds. Various levels of anticancer activity of gossypol have been observed against many different cancer cell lines both in vitro and in vivo (1–3). Evidence is accumulating that gossypol and its derivatives act as BH3 mimetics, killing multiple tumor cell lines, at least in part by activating the Bcl-2-regulated apoptotic pathway (4–7). Gossypol suppresses both telomerase activity and NF-κB activity in human leukemia cells (8, 9). However, the precise mechanisms responsible for the inhibition of cell growth and the stimulation of cell death induced by gossypol are poorly understood. We demonstrate that gossypol potently induces caspase-dependent apoptosis in the absence of Bak and Bax (Bcl-2-associated X protein) by converting Bcl-2 from an inhibitor to an activator of apoptosis (4).

Autophagy has only recently become an important area in cancer research and is emerging as a key regulator of death pathways (10–13). Autophagy is a genetically programmed, evolutionarily conserved process that degrades long-lived cellular proteins and organelles. Autophagy is induced by the formation of the initial membrane nucleation that requires a kinase complex including Beclin 1, a BH3-only protein, and hVps34 to generate the phospholipid PI3P. WIPI-1 (WD repeat domain phosphoinositide-interacting protein 1) binds to PI3P and was found to localize to isolation membranes (14, 15). The isolation membrane chooses its cargo (for example a mitochondrion) and elongates until its edges fuse to form a double-membraned structure known as the autophagosome. Two ubiquitin-like conjugation systems forming LC3-phosphatidylethanolamine (LC3-II) and Atg5-Atg12, respectively, are necessary for the elongation of the isolation membrane (16, 17). The autophagosome matures by fusing with endosomes and lysosomes, finally forming the autolysosome where the degradation of the cargo occurs (11, 18). The mammalian target of rapamycin (mTOR) is considered to be the central regulator of autophagy, because inhibition of mTOR by rapamycin induces autophagy (19).

Under conditions of metabolic stress, autophagy is induced to provide the nutrients and energy required for cell viability (20, 21). Besides its function in cell survival, autophagy has recently emerged as a key regulator of death pathways (11, 19, 22, 23). It is now known that autophagy can be involved in the execution of cell death via caspase-dependent and caspase-independent mechanisms (reviewed in Refs. 19 and 24). Autophagy plays an important role in determining the response of tumor cells to therapy and to change environmental stimuli (25, 26). Autophagy is frequently activated in tumor cells that have been exposed to chemotherapy (tamoxifen, rapamycin, arsenic trioxide, temozolomide, histone deacetylase inhibitors vitamin D analogs, and etoposide (27) or radiation (28)). Moreover, numerous studies have shown that apoptosis and autophagy share some common signaling pathways and are mutually regulated (29–31). For example, Bcl-2, which has been described as an apoptosis guard, also appears to be important in autophagy. Recent observations suggest that the BH3 domain of Beclin 1 interacts with Bcl-2 family proteins, and disruption of their interaction between Bcl-2 and Beclin 1 can stimulate autophagy (29, 32, 33). However, the contributions of autophagy and apoptosis to cell death in cancer induced by various cytotoxic agents remain to be determined. Furthermore, autophagy is known to protect some cancer cells against anticancer treatment by blocking the apoptotic pathway (“protective” autophagy). Growing evidence supports the existence of two distinct forms of autophagy in cancer cells, which we term “canonical” and “noncanonical” autophagy (34). In contrast to classical or canonical autophagy, noncanonical autophagy is a process that does not require the entire set of Atg (autophagy-related) proteins to form the autophagosome. Nishida et al. (35) reported a form of autophagy that is independent of Atg5/Atg7. We and others (36–38) have reported autophagy that is independent of both Beclin 1 and Vps34.

Gossypol is the first naturally occurring compound to have been shown to inhibit Bcl-2, Bcl-X_L, and Mcl-1 and to induce apoptosis. Nevertheless, gossypol-induced autophagy and the cross-talk between apoptosis and autophagy in cancer cells have not so far been explored. In this study, we report that gossypol induces both Beclin-dependent and Beclin-independent autophagy in cancer cells. Notably, we demonstrate that the suppression of autophagy using either pharmacological inhibitors or RNA interference with essential autophagy genes enhances cell death induced by gossypol.

EXPERIMENTAL PROCEDURES

Antibodies and Reagents

Antibodies directed against total and phosphorylated p70S6K (polypeptide 1 ribosomal protein S6 kinase-1) and 4E-BP1 (eukaryotic initiation factor 4E-binding protein 1), against caspase-3 (clone 8 G10) and against Vps34 (3811) were purchased from Cell Signaling. Anti-PARP antibody (C2-10) was from Alexis Biochemicals.

Anti-actin antibody, anti-Atg5, and rabbit polyclonal anti-LC3B antibody were supplied by Sigma. TRITC-conjugated goat anti-mouse IgG was purchased from Jackson. Anti-Bcl-2 (C-2, sc-7382) and horseradish peroxidase monoclonal anti-Bcl-2 antibody (sc-509, from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-cytochrome c for immunofluorescence staining (6H2.B4) and anti-Beclin 1 antibodies were from BD Biosciences PharMingen (San Jose, CA) or from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA; sc-10086).

Earle's balanced salt solution (EBSS) was from HyClone, and rapamycin was from Calbiochem. Gossypol and bafilomycin A1 were from Sigma. Lipofectamine 2000 and fetal bovine serum were from Invitrogen. z-VAD-fmk is from Biovision. The ECLTM Western blotting detection kit and donkey anti-rabbit antibody were purchased from Amersham Biosciences. Goat anti-mouse and swine anti-goat antibodies were obtained from Bio-Rad and Caltag (Burlingame, CA), and 4′,6-diamidino-2-phenylindole and Mito-Tracker Red CMXRos were purchased from Molecular Probes, Inc. (Eugene, OR), respectively.

Cell Culture

The human cervical cancer cell line HeLa and transfected cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin at 37 °C under 5% CO₂. To establish LC3/GFP-stable transfectants, HeLa cells were transfected with pEGFP-C3-LC3 plasmid DNA, and positive clones overexpressing LC3 were selected with 1 mg/ml G418. Hela/GFP-LC3 were kindly provided by T. Yoshimori (Osaka University, Osaka, Japan). The cancer cell line MCF-7 and MCF-7/GFP-LC3 cells were kindly provided by M. Jäättelä (Institute of Cancer Biology, Copenhagen, Denmark) and cultured in RPMI supplemented with 2 mm l-glutamine and 10% fetal bovine serum. Mouse embryonic fibroblast wild type and Atg5^−/− were kindly given by N. Mizushima (Tokyo Medical and Dental University, Tokyo, Japan) and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum.

GFP-LC3 Assay

The assay was performed in MCF-7 or HeLa cells stably transfected with rat GFP-LC3. Prior to analysis, the cells were treated as described in the text. Autophagy was then measured by light microscopic or confocal counting of cells with GFP-LC3 puncta. A minimum of 50–100 cells/sample was counted in triplicate samples/condition/experiment.

GFP-WIPI-1 Assay

The assay was performed in HeLa cells transiently transfected with GFP-WIPI-1 (14, 15). After 36 h the cells were treated for 12 h with 10 μm gossypol, EBSS, or with culture medium. The cells were fixed in 3.7% paraformaldehyde and analyzed by fluorescent microscopy. From three independent sets of experiments, a total of 300–400 cells/treatment were counted. From this the percentages of GFP-WIPI-1 puncta-positive cells were deduced, and representative confocal images were acquired using a LSM 510 (Zeiss).

Western Blot Analysis

Western blot analysis was done as described previously (4). All of the experiments were repeated at least three times. Representative autoradiographs are shown. The antibody dilutions were as follows: anti-LC3, 1:5,000; anti-AKT-P(Ser⁴⁷³), 1:2000; anti-AKT, 1:1000; anti-Bcl-2-P(Ser⁷⁰), 1:1000; anti-Bcl-2 horseradish peroxidase, 1:8000; anti-Beclin, 1 1:2500; anti-actin, 1:20,000, anti-caspase-3, 1:1000; anti-PARP, 1:10,000; anti-Vps34, 1:1000; and anti-Atg5, 1:1000.

Electron Microscopy

The cells were fixed for 1 h at 4 °C in 1.6% glutaraldehyde in 0.1 m Sörensen phosphate buffer (pH 7.3), washed, fixed again in aqueous 2% osmium tetroxide, then dehydrated in ethanol, and embedded in Epon. Ultrathin sections stained with uranyl acetate and lead citrate were then processed for electron microscopy with a Zeiss EM 902 transmission electron microscope at 80 kV.

Co-immunoprecipitation Assays

5 × 10⁶ cells were collected and lysed in CHAPS lysis buffer (20 mm Tris (pH 7.4), 137 mm NaCl, 2 mm EDTA, 10% glycerol, and 2% CHAPS) for 3 h at 4 °C. The cleared lysates were subjected to immunoprecipitation with anti-Beclin 1 antibody and protein G-Sepharose. The immunoprecipitates were analyzed by Western blotting using antibodies against Bcl-2 and Beclin 1.

RNA Interference

A human Beclin 1 shRNA hairpin (target sequence, 5′-CTCAGGAGAGGAGCCATTT-3′) and Atg5 shRNA hairpin (target sequence, 5′-GAAGTTTGTCCTTCTGCTA-3′) were cloned into the BamH1 and HindIII sites of pSilencer 2.1-U6 Hygro. Two sequences for human Vps34 siRNA (5′-⁶⁷⁵GTGTGATGATAAGGAATAT⁶⁹³-3′ and 5′-¹³⁵GTTCTCAGGACTATATCAA¹⁵³-3′ were used. Control scrambled and shRNA plasmids were transfected into HeLa cells. The RNA interference efficiency was verified by Western blotting.

Cell Death Analysis

The cells were plated in a six-well plate at a density of 5 × 10⁵ cells/well and incubated overnight. The medium was then replaced with either fresh medium (control medium) or medium supplemented with 10 μm of gossypol for various times. Apoptotic events were evaluated by annexin V labeling, using the annexin V-fluorescein isothiocyanate/propidium iodide assay kit (BD Biosciences) according to the standard protocol. Apoptotic cell death was also determined by quantification of apoptotic nuclei (fragmentation and condensation of nuclei) following Hoechst 33258 staining. A total of 300–500 nuclei were counted for each sample.

Fluorescence Microscopy

Apoptosis was induced by incubating with gossypol for the times indicated in culture medium. Nuclear condensation was assessed by staining with Hochest 33342. The cells were grown to 70% confluence on a coverslip, and the nuclear material was stained with 1 μg/ml Hochest 33342 at 37 °C for 15 min. The cells were washed twice with phosphate-buffered saline and then fixed with freshly prepared 3.7% formaldehyde at 37 °C for another 15 min. The cell images were obtained using a LSM 510 Zeiss confocal microscope.

Protein Degradation Assay

Protein degradation was measured as previously described (39). Briefly, the cells were radiolabeled for 24 h with 0.05 mCi/ml of l-[U-¹⁴C]valine. At the end of the labeling period cells were rinsed three times with phosphate-buffered saline and then incubated for 6 h with 10 mm valine full medium either in the presence or in the absence of 10 μm gossypol or EBSS.

Caspase-Glo® 3/7 Assay

The assay (Promega) provides a proluminescent caspase-3/7 substrate, which contains the tetrapeptide sequence DEVD. Following caspase activation, this substrate is cleaved to release aminoluciferin, a substrate of luciferase used in the production of light. Luminescence is proportional to the amount of caspase activity.

Caspase-Glo® 3/7 assay was performed according to the manufacturer's protocol. Briefly, the cells were plated in white-walled multiwell plates suitable for cell culture and compatible with the luminometer at a density of 2 × 10⁴ cells/well and incubated overnight. The medium was then replaced with either fresh medium (control medium) or medium supplemented with 10 μm of gossypol for 24 h. When required 20 μm z-VAD-fmk was added to the medium. Caspase-Glo® 3/7 reagent was prepared according to the standard protocol and added in equal volumes to the well. Luminescence was measured by luminometer.

Statistical Analysis

Statistical analysis of the differences between the groups was performed using Student's t test. p < 0.05 was considered statistically significant.

RESULTS

Gossypol Activates Autophagic Flux in Both MCF-7 and HeLa Cells

Both MCF-7 and HeLa cell lines are sensitive to gossypol, express detectable levels of Bcl-2, and trigger autophagy in response to cancer treatment (40, 41). We first tested the ability of gossypol to induce autophagy in these two cell lines. We therefore looked for changes in the cellular distribution of the autophagy marker LC3. In response to autophagy, the cytoplasmic protein LC3 was conjugated to phosphatidylethanolamine to yield LC3-II, a form that localizes to the presautophagosomal and autophagosomal membranes. Gossypol induced time-dependent accumulation of the LC3-II form in both MCF-7 and HeLa cells (Fig. 1A). The cellular localization of LC3 can also be evaluated by following stably transfected cells using the fluorescent autophagy marker GFP-LC3. As illustrated in Fig. 1B, gossypol caused an accumulation of a punctuate fluorescent pattern, indicating the redistribution of LC3 to autophagic structures, whereas untreated cells displayed diffuse staining, confirming once again that gossypol activates the autophagic process in these cells. Transmission electron microscopy experiments were also used to confirm the presence of bona fide autophagic structures in the cells. Gossypol-treated cells had more typical autophagic structures/cell than untreated cells (Fig. 1C). As illustrated in Fig. 1C and in supplemental Fig. S1, gossypol also induced mitochondrial fragmentation and swelling and the loss of cristae. This is consistent with data recently reported by Vogler et al. (42) suggesting that gossypol induces mitochondrial swelling in chronic lymphocytic leukemia cells.

FIGURE 1.

Gossypol triggers autophagy. A, immunoblot analysis of LC3-I and LC3-II levels in MCF-7 (left panel) and HeLa (right panel) cell lines treated with 10 μm gossypol for the times indicated. B, top panels, MCF-7/GFP-LC3 and HeLa/GFP-LC3 cell lines were treated with 10 μm gossypol for the times indicated, fixed, and then visualized by fluorescent microscopy, Bars, 5 μm. Bottom panels, the number of GFP-LC3 dots was scored on ∼50–100 cells. LC3 levels correlate with the numbers of autophagosomes in the cells at a snapshot in time. The data are presented as the means ± S.D. from three independent experiments and analyzed using Student's t test (*, p < 0.05; **, p < 0.001). C, representative electron micrographs of HeLa cells (top panel) and MCF-7 cells (bottom panel) treated or untreated with 10 μm gossypol for 18 h. Arrowheads, presence of autophagosomes. N, nucleus; M, mitochondria; ER, endoplasmic reticulum. Bars, 1 and 0.5 μm in frames and enlargements, respectively. D, autophagic flux. The effect of gossypol on autophagosome degradation can be inferred by comparing untreated samples with those treated with an inhibitor of lysosomal proteolysis such as bafilomycin A1 (Baf). Immunoblot analysis of LC3-I and LC3-II levels in MCF-7 (left panel) and HeLa (right panel) cell lines treated with bafilomycin (final concentration, 100 nm) for 2 h. E, gossypol promotes long-lived protein degradation in cells cultured in full medium. MCF-7 and HeLa cells were radiolabeled for 24 h with 0.05 mCi/ml of l-[U-¹⁴C]valine. At the end of the labeling period, the cells were rinsed three times with phosphate-buffered saline. The cells were then incubated in full medium ± 10 μm gossypol or in EBSS with 10 mm valine for 6 h. The data are presented as the means ± S.D. from three independent experiments and analyzed using Student's t test (*, p < 0.05).

Because the cellular level of LC3-II alone may not accurately reflect autophagic activity, the autophagic flux into the lysosomal compartment was investigated by analyzing LC3-II in cells treated with bafilomycin A1, a specific inhibitor of vacuolar-type H+-ATPase (see Fig. 1D and supplemental Fig. S2). The accumulation of LC3-II determined by immunodetection (Fig. 1D), and the accumulation of GFP-LC3 dots (supplemental Fig. S2) were enhanced in the presence of bafilomycin A1, supporting the notion that gossypol causes the activation of autophagy by promoting the synthesis of autophagosome and increasing autophagic flux. As a positive control, we used EBSS (starvation medium), a classical stimulus used to induce the build-up of autophagosomes and autophagic flux.

The stimulatory effect of gossypol on autophagy was confirmed by analyzing the rate of degradation of long-lived protein (39). As illustrated in Fig. 1E, gossypol treatment increased the degradation of long-lived proteins in both MCF-7 and HeLa cells cultured in complete medium.

Beclin 1 Is Required for the Stimulation of Autophagy Triggered by Gossypol in MCF-7 but Not in HeLa Cells

Autophagy is tightly regulated by the activity of the Beclin 1 complex in initiating the formation of autophagosomes (43). We first compared the increase in the abundance of LC3-II caused by gossypol in MCF-7 cells treated with gossypol following the shRNA-mediated silencing of Beclin 1. As expected, the increase in the abundance of LC3-II caused by gossypol (Fig. 2A, left panel) was reduced by the down-regulation of Beclin 1 expression (Fig. 2A, right panel). We next compared the accumulation of number of GFP-LC3 puncta/cell in MCF-7 cells treated with gossypol following the shRNA-mediated silencing of Beclin 1. Accordingly, GFP-LC3 assays revealed a marked decrease in the number of GFP-LC3 dots when Beclin 1 expression was depleted, indicating the requirement of Beclin 1 for the stimulation of autophagy by gossypol in MCF-7 cells (Fig. 2C). We next compared the increase in the abundance of LC3-II caused by gossypol in HeLa cells following the shRNA-mediated silencing of Beclin 1 (Fig. 2B). Surprisingly, and in contrast to the effects of starvation and of rapamycin treatment, the increase in the abundance of LC3-II caused by gossypol (Fig. 2B, left panel) was not reduced by Beclin-shRNA (Fig. 2B, right panel). Accordingly, GFP-LC3 assays revealed the same number of GFP-LC3 puncta/cell when Beclin 1 expression was depleted, indicating that Beclin 1 is not involved in the stimulation of autophagy by gossypol in HeLa cells (Fig. 2D).

FIGURE 2.

Gossypol-induced autophagy is dependent on Beclin 1 in MCF-7 cells but not in HeLa cells. Immunoblot analysis of LC3-I and LC3-II levels of MCF-7 (A) and HeLa (C) cells transfected with control (left) and with Beclin 1 RNAi (right) treated with 10 μm gossypol for the indicated times. Beclin 1 immunoblotting was used as a transfection control. Actin immunoblotting was used as a loading control, and EBSS or rapamycin treatment was used as an autophagy control. MCF-7 (B) and HeLa (D) GFPLC3 cells were transfected with control or with Beclin 1 RNAi and then treated with 10 μm gossypol for the indicated time, followed by fixation, and visualized by fluorescent microscopy. Bars, 5 μm. The number of GFP-LC3 dots was scored on ∼50–100 cells (**, p < 0.001). E, gossypol induces Beclin 1·Bcl-2 complex dissociation in MCF-7, but not in HeLa cells. Top row, MCF7 (left panel) or HeLa (right panels) cells were cultured for 24 or 6 h, respectively, either in complete medium (control) or in medium supplemented with 10 μm gossypol. EBSS treatment (4 h) or ABT-373 were used as an autophagy control. Endogenous Beclin 1 or Bcl-2 was immunoprecipitated using goat polyclonal anti-Beclin 1 antibody (1:40 dilution) or horseradish peroxidase monoclonal anti-Bcl-2 antibody, respectively. The immunoprecipitated proteins were subjected to immunoblotting using a horseradish peroxidase monoclonal anti-Bcl-2 or anti-Beclin 1 antibodies. Bottom, lysates were immunoblotted with the antibodies indicated. The Western blots are representative of three independent experiments.

Previous studies have shown that the anti-apoptotic protein Bcl-2 down-regulates autophagy by binding to Beclin 1 (33, 44, 45). Dissociation of the Beclin 1·Bcl-2 complex stimulates autophagy induced by starvation or by BH3 mimetic molecules. In a previous study we demonstrated that gossypol is a potent BH3 mimetic molecule (4). However, the role of gossypol in the regulation of the Beclin 1·Bcl-2 complex remained unknown. We used MCF-7 and HeLa cells, which express detectable endogenous levels of both Beclin 1 and Bcl-2, to investigate the effect of gossypol on the dissociation of Beclin 1 and Bcl-2. Co-immunoprecipitation experiments showed that when gossypol induced autophagy, dissociation of the Beclin 1·Bcl-2 complex was observed in MCF-7 cells, but not in HeLa cells, as in starvation (Fig. 2E). Regarding the noncanonical data in HeLa cells, we carried out a flip/flop co-immunoprecipitation experiment (IP·anti-Bcl2 and IB· anti-Beclin 1) for this cell line. Previous study reported that ABT-737 (another pharmacological BH3 mimetic) competitively inhibits the interaction between Beclin 1 and Bcl-2/Bcl-X_L, antagonizes autophagy inhibition by Bcl-2/Bcl-X_L and hence stimulates autophagy in HeLa cells (33). Our data show that in contrast to ABT-737 or EBSS, when gossypol was used, dissociation of the Beclin 1·Bcl-2 complex was not observed in HeLa cells (Fig. 2E).

Previous studies have demonstrated that c-Jun N-terminal kinase (JNK) phosphorylates Bcl-2, triggering its release from Beclin 1, following starvation or ceramide-induced autophagy (45, 46). However, the role of gossypol in the phosphorylation pathways of Bcl-2 remains unknown. We used MCF-7 and HeLa cells to investigate the effect of gossypol on Bcl-2 phosphorylation. As illustrated in supplemental Fig. S3, gossypol did not induce a time-dependent increase in the phosphorylated Bcl-2 levels in either MCF-7 or HeLa cells. A slight decrease in the phosphorylated Bcl-2 levels was even observed. From this finding we concluded that Bcl-2 phosphorylation is not involved in the dissociation of the Bcl-2· Beclin 1 complex or in the subsequent gossypol-induced autophagy.

Gossypol-induced Autophagy Is Dependent on Vps34 in Both MCF-7 and HeLa Cells

The Vps34 class III phosphatidylinositol 3-kinase (hVps34) complex plays an essential role in autophagosome formation (47). Treatment of both MCF-7 and HeLa cells with wortmannin, a phosphatidylinositol 3-kinase inhibitor, inhibited LC3 II formation (Fig. 3A) and GFP-LC3 recruitment to the autophagosome in both cell lines (Fig. 3B), although the effect was less severe in HeLa than in MCF-7 cells. This finding indicates that wortmannin exerted an autophagy-inhibiting effect.

FIGURE 3.

Gossypol-induced autophagy is dependent on Vps34 in both cells lines. A, immunoblot analysis of LC3-I and LC3-II levels in MCF-7 (left panel) and HeLa (right panel) cell lines treated with 10 μm gossypol (Go) alone or in combination with 100 nm of wortmannin (Wo) for the times indicated. B, top row, cells were cultured for 6 h with or without 10 μm gossypol and then incubated for 5 h with or without 10 μm gossypol alone or in combination with 100 nm wortmannin. The cells were then fixed and visualized by fluorescent microscopy (bars, 10 μm). Bottom, the number of GFP-LC3 dots was scored on ∼50–100 cells. The values represent the means ± S.E. of two independent experiments (**, p < 0.001; ***, p < 0.0001). C, immunoblot analysis of LC3-I and LC3-II levels in MCF-7 (left panel) and HeLa (right bottom panel) cells transfected with scramble and with Vps34 RNAi treated with 10 μm gossypol for the indicated times. D, gossypol treatment stimulated GFP-WIPI-1 puncta formation. GFP-WIPI-1 was transiently expressed in HeLa cells and treatments using 10 μm gossypol, EBSS, or control culture medium were carried out for 12 h. Representative confocal images were taken (top; bars, 20 μm), and the percentage of GFP-WIPI-1 puncta-positive cells from three independent sets of experiments counting a total of 300–400 cells/treatment was presented (bottom). The values represent the means ± S.D. (**, p < 0.01; ***, p < 0.001).

To investigate the role of Vps34 directly, we knocked down the expression of the Vps34. In accordance with Fig. 3A, the increase in the abundance of LC3 II caused by gossypol (Fig. 3C) was reduced by Vps34-sh RNA in both cell lines. From these findings we concluded that gossypol induces Vps34-dependent autophagy in HeLa cells. We next tested the ability of gossypol to induce WIPI-1 puncta formation in HeLa cells (Fig. 3D). WIPI-1 acts downstream of Vps34, specifically binds PI3P, and accumulates at the initiation site for autophagosome formation (14, 15). Gossypol treatment stimulated the formation of GFP-WIPI-1 puncta. From these findings we conclude that gossypol induces activation of Vps34, the generation of PI3P, and the binding of PI3P by WIPI-1 and leads to the induction of autophagy.

Gossypol-induced Autophagy Is Dependent on Atg5 in HeLa Cells

Previous studies have shown that the formation of autophagosomes as a result of Beclin 1-independent autophagy is dependent on the Atg12-Atg5 conjugation system. To determine the function of Atg5 in gossypol-induced autophagy, we compared the accumulation of autophagosomes before and after shRNA-mediated silencing of Atg5 in HeLa cells (Fig. 4). As expected, the increase in the abundance of LC3-II caused by gossypol (Fig. 4A, left panel) was reduced by Atg5-shRNA (Fig. 4A, right panel). Accordingly, GFP-LC3 assays revealed a marked decrease in the number of GFP-LC3 dots when Atg5 expression was depleted, indicating the requirement of Atg5 for stimulation of autophagy by gossypol in HeLa cells (Fig. 4B). Similar results have been obtained in MCF-7 (data not shown).

FIGURE 4.

Atg5 is critical for gossypol-induced autophagy. A, immunoblot analysis of LC3-I and LC3-II levels in HeLa cells transfected with control (left panel) or with Atg5 RNAi (right panel) and then treated with 10 μm gossypol for the times indicated. Atg5 immunoblotting was used as a transfection control, and actin immunoblotting was used as a loading control. B, left panel, HeLa cells were transfected with both Beclin 1 RNAi and GFP/LC3 vectors, then fixed, and visualized by fluorescent microscopy. Right panel, the number of GFP-LC3 dots was scored on ∼50/100 cells (**, indicate p < 0.001).

Gossypol-induced Autophagy in MCF-7 and HeLa Cells Is Associated with the mTOR Pathway

Within the complex regulation of autophagy, the protein kinase mTOR is part of an important signaling pathway (48). mTOR in complex 1 (mTORC1) controls the initiation of autophagy upstream of the ULK1 (uncoordinated-51-like kinase 1, the mammalian ortholog of Atg1) complex (47). Activation of mTORC1 down-regulates autophagy and is responsible for the phosphorylation of 4E-BP1 and the 70-kDa p70S6K, two proteins involved in protein synthesis. We monitored the activity of mTOR by immunodetecting the levels of phosphorylation of its substrates, 4E-BP1 and pS6K at Thr^37/46 and Thr³⁸⁹, respectively (supplemental Fig. S4). Medium containing serum was used as a positive control of the phosphorylation level of 4E-BP1 and p70S6K (supplemental Fig. S4, control). The level of phosphorylation of cells cultured in serum-free medium for 4 h was used as a control of the decrease in the phosphorylation level of 4E-BP1 and p70S6K (supplemental Fig. S4, EBSS). Gossypol treatment of MCF-7 and HeLa cells caused a time-dependent decrease in the phosphorylation levels of p70S6K and 4E-BP1 (supplemental Fig. S4, Gossypol). These findings show that gossypol-induced autophagy is correlated with mTORC1 inhibition.

Autophagy is tightly regulated by proteins upstream of mTOR including Akt/PKB (19). Whereas the phosphorylation level of AKT is inhibited by starvation in MCF-7 and HeLa cells, gossypol treatment does not decrease Akt phosphorylation. These findings show that gossypol induces autophagy downstream of Akt in the mTOR signaling pathway.

Inhibition of Autophagy Potentiates Gossypol-mediated Apoptotic Cell Death

The interconnection between gossypol-induced autophagy and apoptosis was first investigated using a chemical inhibitor of autophagy and by shRNA-dependent suppression of the expression of essential autophagy genes.

Wortmannin alone did not induce significant apoptosis in HeLa and MCF-7 cells. However, co-treatment with both gossypol and wortmannin potentiated gossypol-mediated apoptotic cell death in both cell lines (Fig. 5A).

FIGURE 5.

Inhibition of autophagy by wortmannin potentiates gossypol-induced cell death. A, cells were treated with 10 μm gossypol (Go1, 6 h for HeLa and 12 h for MCF-7; Go2, 16 h for HeLa and 24 h for MCF-7) alone or in combination with 100 nm of Wo (5 h). Cell death was measured by annexin V staining. The values represent the means ± S.E. of three independent experiments. B, immunoblot analysis of caspase-3 level in MCF-7 (left panel) and HeLa (right panel) cell lines. MCF-7 was treated for 24 h with 10 or 20 μm gossypol (Go) alone or in combination with 100 nm of wortmannin (Wo) for 5 h. HeLa cells were treated with 10 μm gossypol alone or in combination with 100 nm of wortmannin for the times indicated. C, caspase-3/7 activity was performed by Caspase-Glo® 3/7 assay in MCF-7 (left panel) and HeLa (right panel) cell lines. MCF-7 was treated for 24 h with 10 or 20 μm gossypol alone or in combination with 20 μm of z-VAD-fmk for 24 h or 100 nm of Wo for 5 h. HeLa cells were treated for 24 h with 20 μm of z-VAD-fmk for 24 h, or 100 nm of Wo for 5 h. D, immunoblot analysis of PARP level. The cells are treated as in B or in the presence or absence of z-VAD-fmk 20 μm of for 24 h.

We then investigated whether inhibiting autophagy affected caspase-3 processing and activity. Caspase-Glo® 3/7 assay was used to directly measure caspase activity in both cell lines. MCF-7 cells co-treated with gossypol and wortmannin did not exhibit significant cleavage of the proapoptotic caspase-3 (Fig. 5B, left panel) or caspase activity (Fig. 5C, left panel), whereas treatment with gossypol alone or with a combination of gossypol and wortmannin had induced caspase-3 processing in HeLa cells at 13 h (Fig. 5B, right panel) and had enhanced caspase-3/7 activity in HeLa cells (Fig. 5C, right panel). This activity is abolished in the presence of z-VAD-fmk. From these findings we conclude that caspase-3 processing and activity were involved in HeLa cells but not in MCF-7 cells.

Proteolytic cleavage of PARP is a hallmark of apoptosis. The co-treatment of gossypol with wortmannin potentiated gossypol-mediated PARP cleavage in both cell lines (Fig. 5D). Collectively, these findings suggest that wortmannin inhibited gossypol-mediated autophagy and that it also increased the sensitivity of both cell lines toward the cytotoxic action of gossypol, with the induction of apoptosis.

Because wortmannin could enhance gossypol-induced cell death by mechanisms independent of its effects on autophagy, we knocked down the expression of the essential autophagy genes Atg5, Vps34, and Becn in MCF-7 and HeLa cells using specific shRNAs and analyzed the cross-talk between apoptosis and autophagy. As described above, we found that the down-regulation of Atg5 expression blocked the gossypol-induced accumulation of the autophagic LC3-II isoform (Fig. 4) and enhanced the cell death-inducing effect of gossypol (Fig. 6). Quantification of apoptotic cells showed that after exposure to gossypol, the apoptosis rate in Atg5 RNAi-transfected cells was nearly twice that in the untransfected cells. The role of Beclin 1-dependent autophagy in gossypol-mediated cell death was studied by knocking down Beclin 1 expression using RNAi. The expression of Beclin 1 was markedly inhibited in MCF-7 and HeLa cells transfected with Beclin 1 RNAi (Fig. 2, A and C). Notably, the down-regulation of Beclin 1 expression enhanced the cell death-inducing effect of gossypol (Fig. 5, A and B) in MCF-7 but not in HeLa cells. Finally, we found that the down-regulation of Vps34 expression blocked the gossypol-induced accumulation of the autophagic LC3-II isoform (Fig. 3C) and enhanced the apoptotic cell death-inducing effect of gossypol (Fig. 6B). Taken together, these findings suggest that both canonical and noncanonical autophagy protects tumor cells against gossypol-induced cell death. These observations support the notion that the autophagy mediated by gossypol is protective in nature.

FIGURE 6.

Inhibition of autophagy by Vps34 and Atg5 RNAi potentiates gossypol-induced cell death in MCF-7 and HeLa cells. A, cells were transfected with either Atg5 RNAi or Beclin 1 RNAi and then treated with 10 μm gossypol for the times indicated, followed by fixing, then stained with Hoechst 33342, and visualized by confocal microscopy. Bars, 10 μm. The arrows indicate the condensed, fragmented, brightly stained nuclei, which are the hallmark of apoptosis. B, cells were transfected with Atg5 RNAi, Beclin 1 RNAi, or Vps34 RNAi and then treated with 10 μm gossypol for the times indicated, and then the percentage cell death was determined by annexin V/PI staining followed by flow cytometry analysis.

DISCUSSION

Finding new therapeutic strategies to kill tumor cells is a major challenge facing cancer research. Currently, gossypol is being evaluated in phase I and II clinical trials for use as a single agent in B-cell malignancies and prostate cancer and in combination with other antitumor agents in a variety of hematologic, lymphoid, and solid tumor malignancies (49).

In this study we provide evidence that gossypol induces autophagy followed by apoptotic cell death in both MCF-7 and HeLa tumor cell lines. We found that Beclin 1 is critical to trigger the stimulation of autophagy by gossypol in MCF-7 cells but not in HeLa cells. We next identified the Beclin 1·Bcl-2 complex involved in autophagosome formation as a new target for gossypol. Dissociation of this complex is required in MCF-7 cells, but not in HeLa cells, to trigger gossypol-induced autophagy. We conclude that gossypol induces autophagy in both canonical and noncanonical manners. Notably, we demonstrate that inhibiting autophagy enhances the apoptosis induced by gossypol in tumor cells, thus indicating for the first time that gossypol-induced autophagy is cytoprotective rather than being part of the cell death process induced by this compound.

A direct association between Beclin 1 and Vps34 is essential to promote canonical autophagy (50). Haplo-insufficiency of Beclin 1 expression levels is sufficient to inhibit autophagy both in vivo and in vitro (51, 52). As expected, down-regulation of Beclin 1 expression by RNA interference effectively reduced gossypol-, rapamycin-, and starvation-induced autophagy in MCF-7 or HeLa cells, respectively (Fig. 2). However, in HeLa cells, despite effective knockdown of Beclin 1 protein expression throughout the duration of gossypol treatment, the autophagy elicited was unaffected by Beclin 1 RNAi treatment. In addition, the PI3P-binding protein WIPI-1 was stimulated to localize at autophagosomal membranes upon gossypol treatment in HeLa cells. Taken together with the findings of the co-immunoprecipitation assay, these results implicate the existence of Beclin 1-independent and Vps34-dependent mechanisms of gossypol-induced autophagy in HeLa cells.

Beclin 1 is a BH3-only protein that binds Bcl-2 anti-apoptotic family members. Inducers of autophagy can stimulate the dissociation of Beclin 1 from its inhibitors, either by activating BH3-only proteins (such as Bad) or by post-translational modifications of Bcl-2 (such as phosphorylation) that may reduce its affinity for Beclin 1 and BH3-only proteins. In this study, we demonstrate that gossypol-induced dissociation of the Beclin 1·Bcl-2 complex does not require the phosphorylation of Bcl-2. Gossypol acts as a BH3 mimetic, binding to the BH3 pocket of Bcl-2 or Bcl-xl and disrupting Bcl-xl-Bim and Bcl-xl-Bax complexes (53). In a recent study involving reconstituted proteoliposomes, it has been shown that gossypol competes with Bax/Bid BH3 ligands for binding to the Mcl-1 hydrophobic grove, which explains how gossypol restores Bax permeabilizing function in the presence of Mcl-1 (7). This suggests that gossypol may trigger canonical autophagy by binding to the BH3 pocket of Bcl-2 to disrupt the Beclin 1·Bcl-2 complex and hence stimulates autophagy in MCF-7 cells. Nevertheless, further investigations are needed to elucidate the molecular mechanism by which the noncanonical autophagy induced by gossypol is regulated in HeLa cells. In contrast, ABT-737, another pharmacological BH3 mimetic, inhibits the interaction between Beclin 1 and Bcl-2/Bcl-X(L) and thus stimulates canonical autophagy in HeLa cells (33), indicating that the noncanonical autophagy stimulated by gossypol is not cell type-dependent. However, the ubiquitin-like conjugation systems Atg12-Atg5 and LC3-PE are both required for autophagosome formation in both canonical and noncanonical autophagy. Furthermore, both the noncanonical and canonical autophagy induced by gossypol are characterized by the inhibition of mTOR signaling, which probably occurs upstream of the Beclin 1·Bcl-2 complex (47).

A recent study has shown that gossypol induces apoptosis in chronic lymphocytic leukemia through mitochondrial outer membrane permeabilization with increased production of reactive oxygen species, which in turn is associated with a loss of intracellular ATP, activation of Bax, and the release of cytochrome c and apoptosis-inducing factor, thus causing apoptosis (3). However, antioxidants did not abrogate gossypol-induced cell death, indicating that reactive oxygen species generation appears to be an effect rather the cause of gossypol-induced cell death. In this study we provide evidence that gossypol induces mitochondrial swelling (Fig. 1C and supplemental Fig. S1), which is consistent with data reported in chronic lymphocytic leukemia cells (42). Mitochondrial swelling can result from the induction of a permeability transition pore (MPTP). The induction of MPTs is known to play an important role in controlling the fate of the cell and whether it adopts autophagy or apoptosis. We therefore hypothesize that during the early stages of gossypol treatment, the degree of MPTP induction is relatively low, and thus self-defensive autophagy occurs to ensure the turnover of damaged mitochondria. However, following prolonged gossypol treatment or an increase in the dosage, the increased MPTP reaches a “threshold” level above which extensive autophagy occurs and is followed by apoptosis. The defense provided by autophagy then diminishes, and the extent of apoptosis increases (compare Figs. 1B and 6). Autophagy therefore seems to delay the onset of apoptosis. This notion is further supported by our findings that the inhibition of autophagy by wortmannin, or more specifically by Vps34, Atg5, or Beclin 1 RNAi, increased cell sensitivity to gossypol-mediated apoptosis (Figs. 5 and 6). Zhu et al. (36) were the first to demonstrate the existence of Beclin 1-independent autophagy elicited by the neurotoxin 1-methyl-4-phenylpyridinium in pathological situations. Moreover, pathological stimulation of Beclin 1-independent autophagy is associated with neuronal mitochondrial cell death. Recently we reported that noncanonical autophagy induced by resveratrol, a polyphenol found in grapes and other fruits and vegetables, can act as a caspase-independent cell death mechanism in breast cancer cells (38). In this study we report for the first time that noncanonical autophagy is Beclin 1-independent but Vps34-dependent. We also report that this noncanonical autophagy is associated with the protection of cancer cells against anticancer treatment by blockade of the apoptotic pathway.

In conclusion, our study shows that the naturally occurring BH3 mimetic gossypol induces autophagy and apoptosis in cancer cells. Our findings show that gossypol-induced autophagy is cytoprotective rather than being part of the cell death process induced by this compound.

Addendum

During review of this paper, Smith et al. (54) reported a Beclin 1-independent but Vps34-dependent form of autophagy in ovarian cancer cells in response to arsenic trioxide.