Gambogic acid is an antagonist of antiapoptotic Bcl-2 family proteins

Gambogic acid (GA) is a medicinal compound derived from the gamboges resin of the tree Garcinia hanburyi. GA has documented cytotoxic activity against tumor cell lines in culture, with concentrations required for killing 50% of cells (LD₅₀) of ∼1 μmol/L (1, 2). This natural product also displays antitumor activity in preclinical mouse models involving human tumor xenografts (3–5). In contrast, GA is reportedly well tolerated in mice and rats (2, 4, 6), suggesting that a therapeutic window might be identified at which tumor but not normal cells are killed. It would therefore be interesting to know the cytotoxic mechanism of GA.

The mechanism by which GA kills tumor cell lines involves apoptosis, a cell death processing involving caspase family proteases. In fact, GA was identified as an active compound in a cell-based high-throughput screening assay that measured caspase activation (1). Among the regulators of apoptosis are Bcl-2 family proteins. Humans have six genes encoding distinct antiapoptotic Bcl-2 family proteins: Bcl-2, Bcl-X_L, Mcl-1, Bfl-1, Bcl-W, and Bcl-B (7, 8). These proteins typically localize to intracellular membranes, especially mitochondrial membranes, where they have been shown to block the release of apoptogenic proteins such as cytochrome c, SMAC endonuclease G, and AIF (9–11). Several antiapoptotic Bcl-2 family proteins are known to become pathologically overexpressed in human cancers, conferring apoptosis-resistant phenotypes (12–17).

The antiapoptotic proteins are neutralized endogenously by proteins containing an α-helical interaction motif, known as BH3 (7, 18–20). Synthetic BH3 peptides bind antiapoptotic Bcl-2 family proteins with nanomolar affinities, promoting apoptosis (21, 22). Nonpeptidyl compounds have been identified that compete with BH3 peptides for binding to antiapoptotic Bcl-2 family proteins, mimicking BH3 peptides and creating interest in development of these molecules as potential cancer therapeutics (23, 24).

We show here that GA has the ability to compete with BH3 peptides for binding to several antiapoptotic Bcl-2 family proteins in vitro. GA also neutralizes the function of these proteins with respect to release of SMAC from isolated mitochondria. The cytotoxic activity of GA, however, appears to include Bcl-2 family-independent mechanisms, suggesting that suppression of Bcl-2 and related proteins represents only one of the cytotoxic mechanisms of this natural product.

The MicroSource Spectrum Collection Library (Discovery Systems) is a ∼2,000-compound collection of mostly pure natural products and their derivatives. Compounds were supplied as 10 mmol/L stocks of Me₂SO, stored at -20°C, and thawed immediately before analysis. Aliquots were dissolved to a final concentration of 10 μmol/L for fluorescence polarization assays (FPA). GA was purchased from Calbiochem. All the GA analogues were purchased from MicroSource.

Glutathione S-transferase (GST)-fusion proteins containing Bcl-X_L, Bcl-2, Bcl-W, Bcl-B, Bfl-1, and Mcl-1 lacking their COOH-terminal transmembrane domains (∼last 20 amino acids; “ΔTM”) were expressed from pGEX4T-1 plasmid in XL-1 Blue cells (Stratagene) as described previously (24). Briefly, cells were grown in 2 L Luria-Bertani with 50 μg/mL ampicillin at 37°C to an A_{600 nm} of 1.0. IPTG (0.5 mmol/L) was added, and the cultures were incubated at 25°C for 6 h. Cells were then recovered in 20 mmol/L phosphate buffer (pH 7.4), 150 mmol/L NaCl, 1 mmol/L DTT, 1 mmol/L EDTA, and 1 mmol/L phenylmethylsulfonyl fluoride followed by sonication. Cellular debris was sedimented by centrifugation at 27,500 × g for 20 min, and the resulting supernatants were incubated with 10 mL glutathione-Sepharose (Pharmacia) at 4°C for 2 h. The resin was washed three times with 20 mmol/L phosphate buffer (pH 7.4), 150 mmol/L NaCl, and 1 mmol/L DTT, and 10 mmol/L reduced glutathione dissolved in 50 mmol/L Tris-HCl (pH 8.0) was used to elute the GST-fusion proteins.

Other recombinant proteins used here, including His₆-Bid and His₆-caspase-8, were expressed and purified using methods similar to prior publications (25).

FPAs were done as described previously using various Bcl-2 family proteins and FITC-conjugated Bid BH3 peptide (24, 26). Briefly, Bcl-2 proteins were incubated with 5 nmol/L FITC-Ahx-EDIIRNIARHLAQVGDSMDR in the dark. Fluorescence polarization was measured using an Analyst AD Assay Detection System (LJL Biosystem) in PBS (pH 7.4). IC₅₀ determinations were done using GraphPad Prism software.

Methods for competitive peptide displacement assays were similar to previous publications (24). Briefly, 100 nmol/L GST-Bcl-2 proteins were incubated with the compounds at various concentrations for 5 min at room temperature in PBS. Then, 5 nmol/L FITC-Bid BH3 peptide was added and fluorescence polarization was measured after 10 min. IC₅₀ determinations were generated by fitting the experimental data using a sigmoidal dose-response nonlinear regression model with GraphPad Prism software.

For time-resolved fluorescence resonance energy transfer (TR-FRET) assays, GST-Bcl-X_L and anti-GST-terbium (Invitrogen) were mixed together with the FITC-Bad BH3 peptide in PBS containing 0.005% Tween 20 in 96-well plates in a total volume of 20 μL/well. After incubation at room temperature for 30 min, 2 μL GA-containing solutions were added to the reaction mixtures containing 10 nmol/L Bcl-X_L, 10 nmol/L FITC-Bad BH3 peptide, and 2 nmol/L anti-GST-terbium for 30 min at room temperature. TR-FRET signals were measured with a SpectraMax M5 plate reader (Molecular Devices) using the following settings: excitation at 330 nm, emission for FITC signal at 490 nm, and emission for terbium signal at 520 nm.

HeLa cells were pelleted by centrifugation and then washed once in HM buffer [10 mmol/L HEPES (pH 7.4), 250 mmol/L mannitol, 10 mmol/L KCl, 5 mmol/L MgCl₂, and 1 mmol/L EGTA] containing 1 mmol/L phenylmethylsulfonyl fluoride and a mixture of protease inhibitors (Roche Molecular Biochemicals). The cell pellet was then homogenized in HM buffer by 50 strokes of a Dounce homogenizer using a B-type pestle. The homogenate was centrifuged twice at 600 × g for 5 min to remove nuclei and debris. The resulting supernatant was centrifuged at 10,000 × g for 10 min, and the resulting mitochondria-containing pellet was washed twice with the HM buffer.

For mitochondrial protein release assays, 10 μL mitochondria (50 μg) were added into a final volume of 50 μL HM buffer containing GA, tBid, or tBid preincubated with GA or Bcl-2 family proteins at 30°C for 15 min. The reactions were further incubated at 30°C for 40 to 60 min, and mitochondria were pelleted by centrifugation and the supernatants were collected, boiled in Laemmli sample buffer, and analyzed by SDS-PAGE/immunoblotting using anti-SMAC antibody (25).

Peptides were synthesized using Fmoc solid-phase synthesis on an ACT 350 multiple peptide synthesizer. FITC-conjugated Bid were synthesized on Fmoc-alanine Wang resin to give peptides with COOH-terminal carboxyl groups. FITC was linked to the Ahx. The crude peptides were purified with a Gilson high-performance liquid chromatography instrument and analyzed by matrix-assisted laser desorption/ionization time-of-flight mass analysis with an Applied Biosystems Voyager System 6264.

HeLa, HL-60, Jurkat, PPC1, and mouse embryonic fibroblast cells were maintained in DMEM (Irvine Scientific) supplemented with 10% fetal bovine serum, 1 mmol/L l-glutamine, and antibiotics. Apoptosis was assessed using staining with Annexin V-FITC and propidium iodide (PI) followed by flow cytometry analysis using FL-1 and FL-3 channels of a flow cytometer (Becton Dickinson; FACSort). Annexin V–positive/PI-negative cells were considered apoptotic.

For caspase assays, cell lysates were prepared and normalized for protein content, and 10 μg aliquots of cell lysates were incubated with 100 μmol/L DEVD-AFC, measuring enzyme activity by the release of AFC fluorescence. Data are reported as relative fluorescence units of product produced per minute per microgram of total protein.

We devised a high-throughput screening in which binding of a FITC-conjugated BH3 peptide to recombinant purified Bfl-1 protein is measured by FPA. A library of ∼2,000 natural products was screened for suppression of BH3 peptide binding by >50% using this FPA, resulting in ∼30 hits, among which was GA.

The activity of GA against the six human antiapoptotic Bcl-2 family proteins was contrasted using FPA we have established previously (24). GA displaced to various extents FITC-BH3 peptide binding to all six proteins, with apparent IC₅₀ values of 1.47 μmol/L for Bcl-X_L, 1.21 μmol/L for Bcl-2, 2.02 μmol/L for Bcl-W, 0.66 μmol/L for Bcl-B, 1.06 μmol/L for Bfl-1, and 0.79 μmol/L for Mcl-1. Thus, Mcl-1 and Bcl-B showed the greatest sensitivity (apparent IC₅₀, <1 μmol/L) and Bcl-2 showed the least sensitivity (Fig. 1). Due to solubility, we were unable to increase the GA concentration to achieve complete BH3 peptide displacement for some Bcl-2 family proteins in these assays. Incubating GA with FITC-BH3 peptide in the absence of Bcl-2 family proteins only marginally affected baseline (background) fluorescence polarization (∼Δ20 mP at concentrations >20 μmol/L), excluding a direct effect of GA on the peptide probe as an explanation for the results (Fig. 1). In contrast to Bcl-2 family proteins, GA did not inhibit in FPA using rhodamine-SMAC peptide binding to BIR3 of XIAP (data not shown) or FITC-ATP binding to Hsp70 (data not shown), thus showing the specificity for antiapoptotic Bcl-2 family proteins.

To confirm by an independent method the ability of GA to compete with BH3 peptides for binding to Bcl-2 family proteins, we devised a TR-FRET assay for Bcl-X_L, in which FITC-Bad BH3 peptide bound to recombinant purified GST-Bcl-X_L is excited by emissions from terbium-conjugated anti-GST antibody. FITC-Bad BH3 peptide bound to GST-Bcl-X_L (but not GST; data not shown) in a concentration and saturable manner, with apparent K_d values of 2 nmol/L (Fig. 2A). GA inhibited FITC-Bad BH3 peptide binding to GST-Bcl-X_L, as measured by TR-FRET, with apparent IC₅₀ values of 1.5 μmol/L (Fig. 2B), thus yielding similar results as those obtained using FPA.

An active, NH₂-terminally truncated form of the proapoptotic BH3-containing protein Bid (tBid) induces release of apoptogenic proteins such as SMAC from isolated mitochondria in vitro (25, 27, 28). Using this assay, we showed that five of the six human antiapoptotic Bcl-2 family proteins negate tBid-induced release of SMAC from isolated mitochondria (Fig. 3). For these five proteins (Bcl-2, Bcl-X_L, Bfl-1, Bcl-W, and Mcl-1), adding GA restored tBid-induced SMAC release in a concentration-dependent manner (Fig. 3). Complete restoration was typically achieved with 5 μmol/L GA, representing ∼10:1 molar excess of GA relative to antiapoptotic Bcl-2 family proteins. GA also enhanced tBid-induced release of SMAC from isolated mitochondria when recombinant Bcl-2 family proteins were not added (Supplementary Data),¹

GA has cytotoxic activity against various tumor cell lines in culture and induces apoptosis (1, 2). We confirmed the apoptosis-inducing activity of GA using Jurkat T-cell acute lymphoblastic leukemia and HL-60 acute promyelomonocytic leukemia cell lines using Annexin V/PI staining, counting Annexin V–positive/PI-negative cells as apoptotic (Fig. 4A). The concentration of GA required to induce apoptosis of ∼50% of the cells within 20 to 24 h was ∼0.2 and ∼0.5 μmol/L for Jurkat and HL-60, respectively. GA also induced clearly detectable proteolytic processing of procaspase-3, a marker of apoptosis, at concentrations >2 μmol/L, as determined by immunoblotting using lysates from prostate cancer cell PPC1 treated with GA (Fig. 4B). To assess whether GA affects expression of Bcl-2 family proteins, we did immunoblotting experiments in which HL-60 leukemia cells were treated with caspase inhibitor benozyl-valinyl-alanyl-asparatyl-fluoromethylketone to prevent apoptosis and thus avoid secondary declines in protein expression. Levels of Bcl-2 and Bcl-X_L were not affected by treatment with GA (Supplementary Fig. S2).¹

We tested the cytotoxic activity of GA against HL-60 cells that had been stably transfected with neo-control or Bcl-2 expression plasmids (29). Overexpression of Bcl-2 reduced the sensitivity of HL-60 cells to GA, shifting the dose-response curve to the right, such that higher concentrations of GA were required to kill the cells (Fig. 5A). Treating HL-60 neo-control and Bcl-2-transfected cells with an upstream activator of the mitochondrial pathway for apoptosis, staurosporine, confirmed that Bcl-2-overexpressing HL-60 cells have a block to apoptosis (Fig. 5B), which GA overcomes. Experiments in which staurosporine and GA were combined showed that GA overcomes Bcl-2-mediated resistance to staurosporine (Supplementary Fig. S3).¹

Antiapoptotic Bcl-2 family proteins suppress mitochondria-initiated cell death by inhibiting proapoptotic Bcl-2 members Bax and Bak (10, 20, 30). Consequently, when cells are genetically engineered to lack Bax and Bak, then Bcl-2 and related cytoprotective proteins no longer display an antiapoptotic phenotype (31). Cells doubly deficient in Bax and Bak therefore provide a context for assessing the mechanism of GA-induced toxicity. GA killed transformed mouse embryonic fibroblasts generated from bax^+/+bak^+/+ embryos in a concentration-dependent manner (Fig. 5C). However, GA also killed bax^−/−bak^−/− mouse embryonic fibroblasts, requiring only ∼2-fold higher concentrations of compound to reach the LD₅₀. Treating these mouse embryonic fibroblasts with staurosporine confirmed that Bak/Bak double knockout cells have a profound block to cell death (Fig. 5D), which GA overcomes. We conclude therefore that GA induces cytotoxicity through mechanisms that are only partly dependent on Bcl-2 family proteins.

We examined the activity of analogues of GA with respect to displacement of Bid BH3 peptide from Bfl-1 in vitro (Fig. 6A) and cytotoxic activity (Fig. 6B). In the FPA, dihydro-GA and acetyl-iso-GA displaced BH3 peptide from Bfl-1 with potencies only 3- to 6-fold less than GA (Fig. 6A). In contrast, tetrahydro-GA and garcinolic acid had greatly reduced activity in BH3 displacement assays. In agreement with the BH3 displacement data, dihydro-GA and acetyl-iso-GA exhibited cytotoxic activity against cultured leukemia cells, whereas tetrahydro-GA showed markedly reduced activity (∼50-fold less potent) and garcinolic acid was nontoxic (Fig. 6B). Thus, the cytotoxic activity of GA analogues correlates roughly with their ability to compete with BH3 peptides for binding to an antiapoptotic Bcl-2 family member.

We show here that the cytotoxic natural product GA competes for BH3 peptide binding sites on several antiapoptotic members of the Bcl-2 family and neutralizes the ability of these proteins to suppress release of apoptogenic proteins from isolated mitochondria. Structure-function analysis of GA using analogues suggested a general correlation between BH3 competition and cytotoxicity activity, but experiments with bax/bak double knockout cells suggest GA-induced cytotoxicity is only partially dependent on Bcl-2 family proteins.

GA has been reported to affect other molecular events relevant to cytotoxicity, including inducing increases in Bax protein levels, reducing Bcl-2 protein levels, suppressing transferrin receptor internalization, inhibiting the catalytic activity of human topoisomerase IIα, and modulating the nuclear factor-κB signaling pathway (2, 32–34). GA also reportedly induces G₂-M-phase arrest of dividing cells (35), suggesting that this natural product hits targets involved in cell cycle. Thus, like many natural products, GA may have several targets in mammals, among which are antiapoptotic members of the Bcl-2 family.

The in vivo activity of GA and its therapeutic index has been explored previously in rodents (2, 6). Using a prostate cancer xenograft model, for example, it was recently reported that GA effectively inhibited tumor angiogenesis and suppressed tumor growth with few side effects, where GA was delivered by s.c. injection daily for 15 days (4). Also, using a mouse model of glioma, i.v. injection of GA daily for 14 days was reported to significantly reduce tumor volumes with little evidence of side effects (5). Antitumor activity was attributed to induction of apoptosis and to suppression of angiogenesis.

It is interesting to speculate why plants might produce compounds that neutralize Bcl-2 family proteins. Insects, nematodes, and other animal species that eat plants contain evolutionarily conserved antiapoptotic Bcl-2 family proteins (reviewed in ref. 36). Consequently, plants might produce antagonists of these compounds as a mechanism to defend themselves against animal species. In this regard, several natural products from diverse plant species have been shown to competitively displace BH3 peptides and neutralize the activity of antiapoptotic Bcl-2 family proteins, including epigallocatechin gallate from green tea, theaflavins from black tea, and gossypol from cotton seeds (37, 38). These compounds are structurally diverse, but all display <1 μmol/L activity against a variety of human antiapoptotic Bcl-2 family proteins. It remains to be determined whether their affinity is greater for Bcl-2 family members from insects or other lower organisms of the animal kingdom.

No potential conflicts of interest were disclosed.

We thank Tessa Siegfried and Melanie Hanaii for article preparation.