Gambogic acid inhibits the catalytic activity of human topoisomerase IIα by binding to its ATPase domain

Topoisomerases are essential for DNA metabolism, where they act to adjust the topology of DNA during transcription, replication, recombination, repair, and mitosis (1). Topoisomerases create transient nicks (Topo I) or breaks (Topo II) in the double-stranded DNA polymer, allowing DNA to be converted between topological isomers. Nuclear Topo I and Topo II have been identified as clinically important targets for cancer chemotherapy, and their inhibitors are central components in many therapeutic regimens (2, 3). These topoisomerase-targeting agents can be divided into two categories according to their mechanisms of action: topoisomerase poisons and catalytic inhibitors. Topoisomerase poisons are able to stabilize the reversible covalent topoisomerase-DNA complex termed the cleavage complex, whereas catalytic inhibitors, most of which target Topo II, act on the other steps in the catalytic cycle without trapping the covalent complex. Although topoisomerase poisons are among the frequently used regimens in the clinical treatment of human malignancies, there are still some limitations such as dose-limiting toxicities and drug resistance leading to treatment failure after initial effective therapy. Moreover, drugs from different chemical families, which share a common cellular target, generally exhibit different spectra of anticancer activity. Consequently, there is an increasing interest focusing on the development of new kinds of anticancer drugs targeting human topoisomerases.

Recently, Topo II catalytic inhibitors attract some interest for their multiple activities, including as antineoplastic agents, cardioprotectors, or modulators, to enhance the efficacy of other anticancer agents (4). Thus far, several structurally diversified compounds, including the anthracycline derivative aclarubicin, the thiobarbituric acid derivative merbarone, the coumarin drugs novobiocin, the epipodophyllotoxin analogue F 11782, fostriecin, chloroquine, and the bisdioxopiperazine derivatives, have been identified to be catalytic Topo II inhibitors (4). An interesting feature of most catalytic topoisomerase inhibitors is their capacity to modulate the cytotoxic effects of Topo II poisons (5–7).

Gambogic acid (GA) is a natural product isolated from the gamboge resin of Garcinia hanburyi tree in Southeast Asia. The resin is used as a traditional Chinese medicine exerting effects of detoxification, hemostasis, and parasiticide. Early investigations have identified GA as a potent apoptosis inducer (8), and subsequent studies have confirmed that both activation of caspases and mitochondrial pathway are involved in GA-induced apoptosis (9). Results from in vivo studies indicate that GA has significant antitumor activity (10). Chinese Food and Drug Administration has recently approved a phase II clinical trial of GA injection as an antitumor candidate. Although activation of the apoptotic pathway, down-regulation of telomerase (11, 12), and reduction of CDK7 kinase activity (13) partly explain the potent anticancer activity of GA, the molecular mechanism of its nonselective cytotoxicity remains poorly understood.

In the present study, we used biochemistry, cellular, and surface plasmon resonance experiments demonstrating that GA exerts inhibitory action on human Topo IIα by binding to the ATPase domain, thereby prevents DNA cleavage and inhibits ATP hydrolysis. These results distinguish GA from other topoisomerase-targeting agents, which promises GA a potential antitumor candidate for clinical development.

GA was isolated from gamboge resin of Garcinia hanburyi with the purity above 95% as determined by high-performance liquid chromatography. Etoposide (VP16), Adriamycin, amsacrine, camptothecin, aclarubicin, and App(NH)p were ordered from Sigma. Purified human topoisomerases I and IIα, kinetoplast DNA (kDNA), and plasmid pBR322 were obtained from TopoGEN. [γ-³³P]ATP (3,000 Ci/mmol stock) was purchased from Amersham Biosciences. All compounds were dissolved in DMSO or normal saline as stock solutions, which were kept in aliquots at −20°C.

Human gastric adenocarcinoma SGC-7901, hepatocellular carcinoma BEL-7402 and SMMC-7721, and the ovarian carcinoma HO-8910 cell lines were obtained from the cell bank of Chinese Academy of Sciences. Human promyelocytic leukemia HL-60, chronic myelogenous leukemia K562, lymphoblastic leukemia MOLT-4, lung adenocarcinoma A549 and NCI-H23, hepatocellular carcinoma HepG2, colorectal adenocarcinoma HT-29, oral epidermoid carcinoma KB, cervical carcinoma HeLa, mouse embryo fibroblast NIH-3T3, and human umbilical vein endothelial cells were purchased from the American Type Culture Collection. Human gastric adenocarcinoma MKN-28 and MKN-45; colorectal carcinoma HCT-116 and HCT-15; breast carcinoma MCF-7, MDA-MB-435, and MDA-MB-468; and ovary adenocarcinoma SK-OV-3 cell lines were from the Japanese Foundation of Cancer Research. DOX-selected multidrug-resistant (MDR) cell sublines K562/A02 and MCF-7/Adriamycin were ordered from the Institute of Hematology, Chinese Academy of Medical Sciences. Vincristine-selected MDR subline KB/vincristine was obtained from Zhongshan University of Medical Sciences. All these cell lines were maintained strictly according to the supplier's instructions and established procedures.

Cytotoxicity of GA was assessed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay or sulforhodamine B assay as described previously (14).

Relaxation assays were carried out in a final volume of 20 μL containing Topo I reaction buffer [10 mmol/L Tris-HCl (pH 7.5), 100 mmol/L NaCl, 1 mmol/L phenylmethylsulfonyl fluoride, and 1 mmol/L mercaptoethanol], 0.25 μg supercoiled pBR322 and 1 unit of Topo I (4 units were required for cleavage assays and antagonism assays; 1 unit of Topo I can relax 0.25 μg of supercoiled DNA in 30 min at 37°C). For antagonism assays, GA was incubated with Topo I reaction buffer in the presence of 4 units of Topo I at 4°C for 10 min before 10 μmol/L camptothecin was added. Reactions were used at 37°C for 15 min and terminated by the addition of 2 μL of 10% SDS. Reaction products were separated on a 1% agarose gel for catalytic assays or on a 1% agarose gel containing 0.5 μg/mL ethidium bromide for cleavage assays and antagonism assays.

Relaxation or decatenation assays were carried out in a final volume of 20 μL containing Topo II reaction buffer [50 mmol/L Tris-HCl (pH 8.0), 120 mmol/L KCl, 10 mmol/L MgCl₂, 1 mmol/L ATP, 0.5 mmol/L DTT], 0.2 μg kDNA or 0.25 μg supercoiled pBR322, and 1 unit of Topo IIα (1 unit of Topo II can decatenate 0.2 μg of kDNA in 30 min at 37°C). The ATP concentration in competitive inhibition assays on kDNA decatenation was 2 or 5 mmol/L.

DNA cleavage reactions contained 6 units of Topo IIα and 0.25 μg supercoiled pBR322 in a total volume of 20 μL reaction buffer (15). Complete cleavage reactions were carried out in the presence of 1 mmol/L ATP. DNA cleavage/religation equilibrium before enzyme-mediated strand passage was established in the absence of ATP. DNA cleavage/religation equilibrium after enzyme-mediated strand passage was established in the presence of 1 mmol/L App(NH)p. Cleavage intermediates were trapped with 2 μL of 4% SDS and 2 μL of 250 mmol/L NaEDTA (pH 8.0). Proteinase K was added (2 μL of 0.8 mg/mL), and reaction mixtures were incubated at 45°C for 30 min to digest the Topo IIα. Samples were mixed with loading buffer, heated at 70°C for 2 min, and subjected to electrophoresis on a 1% agarose gel.

Topo II–mediated DNA religations were carried out as described previously (15). CaCl₂ replaced MgCl₂ to promote high levels of enzyme-mediated DNA cleavage (16, 17). Initial DNA cleavage/religation equilibria were established at 37°C for 6 min. For prestrand passage religation, the Topo II-DNA cleavage complex was trapped by the addition of 0.8 μL of 250 mmol/L NaEDTA. To prevent recleavage of the DNA, 0.6 μL of 5 mol/L NaCl was added, and samples were placed on ice to slow reaction rates. Religation was initiated by the addition of 1.8 μL cold MgCl₂ (8.5 mmol/L final). For poststrand passage religation, samples were shifted from 37°C to 55°C to initiate the religation. Tested compounds were added to the reaction mixtures just before the initiation of DNA religation, and the reactions were terminated by the addition of 2 μL 10% SDS. The samples were processed as described above.

HeLa cells grown at 30% to 50% confluence were incubated in MEM containing 10% fetal bovine serum (Life Technologies). Small interfering RNAs (siRNA) for Topo IIα, Topo IIβ, Topo I, or nonspecific control siRNA were transfected into the cells by Oligofectamine (Invitrogen) according to the manufacturer's protocols. The siRNAs oligonucleotides for Topo IIα (sense 5′-GGUAUUCCUGUUGUUGAAC-3′), Topo IIβ (sense 5′-AGAAGAGUCAGAAACUGCC-3′), Topo I (sense 5′-CUUGACAGCCAAGGUAUUC-3′), and negative control (sense 5′-UUCUCCGAACGUGUCACGU-3′) were synthesized by GenePharma, and the final concentration in each well was 100 nmol/L.

HeLa cells transfected with indicated siRNAs were exposed to GA for 3 h. They were collected and fixed overnight in 70% ice-cold ethanol. Before analysis, cells were washed once in PBS, resuspended in 1 mL DNA-staining solution (20 μg/mL RNase, 20 μg/mL propidium iodide in PBS), and incubated for 30 min in the dark. At least 1.0 × 10⁴ cells for each sample were examined by a FACSCalibur analyzer (Becton Dickinson) with Cellquest. DNA contents were analyzed with Modfit LT 3.0 (Verity Software House).

The tested tumor cells were treated as indicated and were lysed in 1× SDS gel loading buffer [50 mmol/L Tris-HCl (pH 6.8), 100 mmol/L DTT, 2% SDS, 0.1% bromphenol blue, 10% glycerol]. Cell lysates were boiled for 5 to 10 min. Samples were separated on polyacrylamide gels and then electrotransferred to nitrocellulose membranes. Membranes were incubated with the indicated primary antibodies against Topo IIα, Topo IIβ, Topo I, phosphorylation of c-Jun (p-c-Jun, Ser⁶³), and β-actin (Santa Cruz Biotechnology), as well as phosphorylated H2AX (γ-H2AX, Ser¹³⁹; Cell Signaling). Membranes were then incubated with horseradish peroxidase–conjugated secondary antibodies, and were visualized by exposure to X-ray film using SuperSignal West Pico Chemiluminescent Substrate (Pierce, Inc.) according to the manufacturer's instructions.

HL-60 cells were pretreated with either GA or aclarubicin for 20 min, and then exposed to 20 μmol/L VP16 or 200 nmol/L camptothecin for 1 h. DNA double-strand break (DSB) were evaluated using neutral single cell gel electrophoresis assay as described previously (18), and observed with a fluorescence microscope (Olympus, BX51). DNA damage was quantified in at least 50 randomly selected cells per slide as the Olive Tail Moment variable (arbitrary units, defined as the value of the percentage of DNA in the tail multiplied by the tail length) analyzed by Komet 5.5 software (Kinetic Imaging, Ltd.).

Clonogenic assays were done as described earlier (19). Briefly, treated or untreated cells were collected by trypsinization, resuspended in ice-cold medium, and seeded at 500 cells per well in six-well plates. Cells were stained with crystal violet 2 weeks later. Individual colonies containing >50 cells were counted.

Interactions between GA and DNA were assessed by two independent assays. First, unwinding assays were done to determine whether GA could intercalate into plasmid DNA (20). Second, ethidium bromide displacement fluorescence assays were used according to published procedures (21). Briefly, test compounds were incubated with 2 μmol/L ethidium bromide in the presence of 20 μmol/L calf thymus DNA (Sigma) in the reaction buffer containing 2 mmol/L acetate (pH 5.0), 9.3 mmol/L NaCl, and 0.1 mmol/L NaEDTA for 10 min at room temperature. Fluorescence was read with a NOVOstar fluorometer (BMG Labtech) with the excitation wavelength of 546 nm and the emission wavelength of 595 nm.

The effect of GA on Topo II·DNA binding was monitored by a fluorescence polarization assay (22). A 40-mer double-stranded oligonucleotide with fluorescein incorporated at the 5′ terminus of the top strand was synthesized by Sangon. The sequences of the top and bottom strands, respectively, were 5′-TGAAATCTAACAATG↓CGCTCATCGTCATCCTCGGCACCGT-3′ and 5′-ACGGTGCCGAGGATGACGATG↓AGCGCATTGTTAGATTTCA-3′. The arrows represent the sites of Topo II–mediated DNA cleavage. Binding assays contained 5 nmol/L fluoresceinated DNA oligonucleotide and 50 nmol/L human Topo IIα in 20 μL of reaction buffer [50 mmol/L Tris-HCl (pH 8.0), 120 mmol/L KCl, 10 mmol/L MgCl₂, 0.5 mmol/L DTT]. Fluorescence polarization was measured with EnVision 2101 plate reader (Perkin-Elmer) using a 480-nm excitation filter and a 535-nm emission filter.

ATPase assays were done as described by Osheroff et al. (23). ATP hydrolysis reactions contained 1 unit of Topo IIα, 0.25 μg supercoiled pBR322, and 1 mmol/L [γ-³³P]ATP in a total volume of 20 μL reaction buffer. Reactions were initiated by the addition of Topo IIα and incubated at 37°C. Samples (2 μL) were removed at intervals up to 25 min, and spotted on polyethyleneimine-impregnated thin layer cellulose chromatography plates (Merck). Plates were developed by chromatography in freshly made 400 mmol/L NH₄HCO₃ and analyzed using Typhoon 9410 (Amersham) with the Image-Pro Plus software (Media Cybernetics).

Human Topo IIα ATPase domain was overexpressed and purified as described previously (24) with some modifications. The surface plasmon resonance assay was done using a Biacore 3000 instrument (Biacore AB). Human Topo IIα ATPase was immobilized onto CM5 sensor chips after activation of the surface carboxyl groups, and a response of 6,000 to 7,000 resonance units was reached. Serial dilutions of compounds were injected to over the sensor chip, and sensorgrams were recorded. Equilibrium constants (K_D) for evaluating protein-ligand binding affinity were determined using the steady-state affinity fitting analysis.

The molecular docking was done according to the procedures described by Hu et al. (25). The crystal structure of the ATPase domain of human Topo IIα in complex with 5′-adenylyl-β,γ-imidodiphosphate (ADPNP) was retrieved from the Brookhaven Protein Data Bank (PDB entry 1ZXM; ref. 26). The docking program AutoDock 3.03 was used to dock ligands to the human Topo IIα ATPase domain.

The cytotoxicity of GA was evaluated using a panel of human tumor cell lines and two immortalized cell lines (NIH-3T3 and human umbilical vascular endothelial cell). GA displayed potent cytotoxicity without distinct selectivity among the cells lines tested (Fig. 1B). The average IC₅₀ of GA against the 21 cell lines was 1.23 μmol/L. We next investigated the cytotoxicity of GA and reference compounds (Adriamycin, vincristine, and VP16) in three MDR-expressing cell lines, K562/A02, MCF-7/Adriamycin, and KB/vincristine (Fig. 1C). For each pair of these cell lines, the resistance factor, defined as the ratio of the IC₅₀ value of resistant cell to that of the parental cell, was calculated. GA showed almost equivalent cytotoxicity against the MDR sublines comparing to the parental cells, with the resistance factor values of 1.24, 3.24, and 1.17 (Fig. 1D), which were much lower than those of reference drugs. These results show that GA displays potent antiproliferative activity against a wide panel of human tumors with significant anti-MDR activity.

GA Inhibits the Catalytic Activity of Topo I. First, we examined the effect of GA on the Topo I–mediated relaxation of supercoiled pBR322. The results indicated that GA caused a dose-dependent inhibition of Topo I catalytic activity with an approximate IC₅₀ of 50 μmol/L (Fig. 2A). A DNA cleavage assay was next applied to detect whether GA induced Topo I–DNA cleavage complexes. As shown in Fig. 2B, GA did not cause nicked DNA (FII), which represents drug-induced Topo I cleavage complexes even up to 50 μmol/L. To further confirm that GA acts as a catalytic inhibitor of Topo I, an antagonism assay was done. As shown in Fig. 2C, 12.5 μmol/L GA pretreatment abated the camptothecin-induced cleavage complexes and 25 μmol/L GA completely abrogated the effect of camptothecin, suggesting that inhibition of the catalytic activity of Topo I by GA resulted in a failure to induce DNA cleavage by camptothecin.

GA Inhibits the Catalytic Activity of Human Topo IIα. Effect of GA on the catalytic activity of human Topo IIα was evaluated using the enzyme-mediated negatively supercoiled pBR322 relaxation. As seen in Fig. 2D, GA inhibited pBR322 relaxation in a dose-dependent manner, and most of pBR322 remained in supercoiled state in the presence of 0.6 μmol/L GA. This inhibitory effect is much more potent than that of VP16. No existence of drug-induced Topo II cleavage complexes was observed in a Topo II–mediated DNA cleavage assay, even at the concentration of that completely restrained Topo II activity (Fig. 2E), indicating the effect of GA on Topo II was different from Topo II poisons such as VP16 or Adriamycin. To further quantify the inhibitory effect of GA on Topo IIα, Topo II–mediated kDNA decatenation was carried out. The results showed that GA was almost equipotent with aclarubicin, a dual topoisomerase catalytic inhibitor, with IC₅₀ values of 6.3 ± 0.2 and 6.9 ± 0.4 μmol/L, respectively (Fig. 2F). Taken together, these results show that GA acts as a dual catalytic inhibitor of Topo I and Topo IIα, and it seems that its activity on Topo IIα is more significant than that on Topo I.

Topo IIα RNA Interference Decreased the Sensitivity of HeLa Cells to GA. We found that GA exerts inhibitory effects on Topo I and Topo IIα in a cell-free system. To determine whether the effects exist in cells, and to find the correlation between topoisomerase activities and GA action, RNA interference (RNAi) was done to down-regulate the expression of Topo I, Topo IIα, or Topo IIβ in HeLa cells, respectively. Down-regulation of endogenous topoisomerases after transfection with siRNA oligonucleotides for 24 h was detected by Western blot analysis (data not shown and Fig. 3C). We observed a significant reduction of cells undergoing apoptosis in Topo IIα siRNA groups treated with 10 μmol/L GA for 3 h (Fig. 3A). However, down-regulation of Topo IIβ and Topo I had little effect on GA-induced apoptosis (data not shown). To better evaluate the effect of Topo IIα–targeted siRNA on drug sensitivity, a clonogenic assay was carried out. As shown in Fig. 3B, cells transfected with siRNA against Topo IIα gained resistance to GA with a survival rate of 53%, which was much higher than those of cells transfected with siRNAs for Topo I, Topo IIβ, or mock control upon GA treatment. Taken together, these results imply that Topo IIα is a preferential cellular target of GA.

Topo IIα RNAi Abated GA-Mediated Phosphorylation of c-Jun. The phosphorylation and activation of the transcription factor c-Jun plays a central role in apoptosis (27). Our previous experiments showed that treatment of HeLa cells with 10 μmol/L GA for 3 h dramatically increased phosphorylation of p38 mitogen-activated protein kinase and extracellular signal-regulated kinase, eventually leading to c-Jun phosphorylation (Fig. 3C).⁴

Topo II poisons are able to stabilize the cleavage complex, thus leading to the accumulation of cytotoxic DNA breaks. Topo II catalytic inhibitors, such as aclarubicin which prevents topoisomerase binding to DNA (28), merbarone which inhibits DNA cleavage, and bisdioxopiperazines which trap of the enzyme in the closed-clamp form (29), are well known to circumvent the cytotoxicity of Topo II poisons. These prompted us to investigate whether GA might act as a catalytic inhibitor of Topo II in cells, considering that GA was unable to trap the cleavage complex (Fig. 2E) and did not induce DNA damage in neutral comet assays (Fig. 4A).⁴

GA Antagonized VP16-Induced DNA Damage, whereas It Did Not Affect Camptothecin-Induced DNA Damage. DNA DSBs induced by VP16 was examined using the neutral comet assay. As shown in Fig. 4A, obvious comet tails were detected in HL-60 cells after exposure to 20 μmol/L VP16 for 1 h, whereas pretreatment with aclarubicin or GA for 20 min dramatically decreased VP16-triggered DNA DSBs. Semiquantitative analysis with Komet 5.5 software indicated that 200 nmol/L GA was almost equipotent with 100 nmol/L aclarubicin to antagonize VP16-induced DNA damage (Fig. 4B). DNA DSBs are always accompanied by the elevation of γ-H2AX (30). We detected whether GA also abrogated VP16-induced γ-H2AX by Western blot. As shown in Fig. 4E, γ-H2AX level was significantly enhanced in HL-60 cells after exposure to 20 μmol/L VP16 for 1 h. As expected, pretreatment with 100 nmol/L aclarubicin or 200 nmol/L GA for 20 min antagonized the induction of γ-H2AX by VP16 in HL-60 cells (Fig. 4E). Similar results were obtained in HeLa cells exposed to 40 μmol/L VP16 for 1 h after pretreatment with 200 nmol/L aclarubicin or 500 nmol/L GA for 20 min (Fig. 4E).

Because GA also inhibits topoisomerase I catalytic activity, and abated the camptothecin-induced cleavage complexes in vitro, we then detected whether this effect exists in cells using the neutral comet assay. As shown in Fig. 4C, treatment with 200 nmol/L camptothecin for 1h caused potential DNA damage of HL-60 cells, and obvious “comet tails” were observed. However, pretreatment with 200 nmol/L GA was incapable of attenuating cytotoxicity of camptothecin. This result reveals a lack of inhibition of Topo I activity of GA in cells in pharmacologic concentration range.

GA Attenuated the Antiproliferative Activity of VP16. Because DNA DSBs are believed to be responsible for the antiproliferative activity of Topo II poisons (31), we investigated whether GA also abrogated the antiproliferative activity of VP16 using clonogenic survival assays. As shown in Fig. 4F, VP16 treatment significantly reduced the survival clones to 9% of that in control groups. Pretreatment with 200 nmol/L aclarubicin or 500 nmol/L GA dramatically abrogated the activity with the survival rates of 39% and 30%, respectively.

The catalytic cycle of Topo II can be divided into six discrete steps (32). To determine at which step the process was influenced by GA, various steps in the catalytic cycle were probed.

Interactions between GA and DNA. Because chemical compounds that intercalate into DNA or bind to the minor groove of DNA have dramatic effects on the activity of Topo II, such as amsacrine, Adriamycin, ellipticine, saintopin, and TAS-103 (33), we investigated whether GA was the case. First, an unwinding assay was done using linearized pBR322 and T4 DNA ligase. In this assay, DNA intercalators unwind the linearized DNA, which results in a change in the twist of the duplex helix. Circularization of drug-bound DNA by T4 DNA ligase freezes the linking number of the unwound DNA. Upon drug removal, the twist changes back to normal whereas the linking number remains constant, which causes the introduction of negative superhelicity into the DNA. As shown in Fig. 5A, an obvious inhibition of the activity of the T4 ligase by 20 μmol/L Adriamycin was observed, which was due to strong intercalation ability (34). However, both VP16 and GA had no influence on the DNA topoisomers, even at a concentration of 100 μmol/L. Second, the ability of GA to displace ethidium bromide from DNA was determined by a fluorescence assay. The DNA-bound form of ethidium bromide has a significantly stronger fluorescence emission than free ethidium bromide. Thus, displacement of ethidium bromide from DNA can be monitored by a decrease in fluorescence signal (35). As seen in Fig. 5B, 100 μmol/L of GA was incapable of displacing ethidium bromide. In contrast, the DNA intercalative amsacrine readily dislodged the bound fluorophore. Taken together, these results indicate GA is a nonintercalative agent.

GA Does Not Impair Topo II·DNA Binding. The first step in the Topo II catalytic cycle is the noncovalent interaction of the protein with DNA. The influence of GA on this interaction was examined using a fluorescence polarization assay. The free fluoresceinated DNA substrate is likely to rotate more rapidly and therefore to have a lower fluorescence polarization than does the Topo II·DNA complex; thus, examining the fluorescence polarization can be used to determine the effect of a compound on Topo II·DNA binding. A double-stranded 40-mer oligonucleotide that contained a single cleavage site was used in the present study. The results indicated that GA had no obvious effect on the fluorescence polarization of Topo II·DNA complex even at a high concentration of 100 μmol/L (Fig. 5C). In contrast, aclarubicin dramatically decreased the fluorescence polarization of the sample, which had been shown to prevent the binding of Topo II to DNA due to strong intercalation ability (36).

Effects of GA on Prestrand and Poststrand Passage DNA Cleavage/Religation Equilibria. In the presence of divalent cation, Topo II establishes a DNA cleavage/religation equilibrium. This equilibrium is called prestrand passage to distinguish it from the cleavage/religation that is established after the strand passage event. Because strand passage requires ATP binding, the enzyme prestrand and poststrand passage DNA cleavage/religation equilibria can be separated experimentally. In our experiments, prestrand passage equilibria were established in the absence of a nucleotide triphosphate and poststrand passage equilibria were established in the presence of 1 mmol/L App(NH)p. As seen in Fig. 5D and E, significant decreases in relaxed and linear DNA were observed by addition of GA in prestrand and poststrand passage Topo II–mediated cleavage assays. It seemed that inhibitory potency of GA against poststrand equilibrium was a little stronger than that against prestrand equilibrium. Increases in relaxed and linear DNA were detected in VP16-treated samples. There were no obvious changes in relaxed and linear DNA in the presence of GA in prestrand and poststrand passage Topo II–mediated DNA religation assays, whereas dramatic increases in relaxed and linear DNA were observed in the presence of VP16 (data not shown). These results showed that GA potently inhibited the Topo II–mediated DNA cleavage and had no effects on religation. This mechanism of action is different from that of VP16, which traps cleavage complex mainly through enhancing the enzyme-mediated DNA cleavage and inhibiting religation.

GA Inhibits Topo IIα–Catalyzed ATP Hydrolysis. In the processes of Topo II reaction, ATP hydrolysis is responsible for the conformational changes leading to DNA transport and final enzyme turnover. A number of Topo II–targeted agents, including novobiocin, genistein, etc., were found to inhibit enzyme-catalyzed ATP hydrolysis (37). Here, we studied the effect of GA on the ATPase activity of human Topo IIα. As seen in Fig. 5F, 10 μmol/L GA substantially inhibited Topo IIα–catalyzed ATP hydrolysis, and the inhibitory rate reached up to ∼80% after 25-min incubation.

GA Binds to the Human Topo IIα ATPase Domain. Because GA inhibited human Topo IIα–mediated ATP hydrolysis, we next determined whether this effect of GA was associated with its direct binding to human Topo IIα ATPase domain using the surface plasmon resonance assay. Recombinant human Topo IIα ATPase domain purified from yeast was immobilized on CM-5 sensor chips. Real-time sensograms was recorded. General fitting and steady-state binding affinities were calculated using BIAeval 3.1 software. Results indicated that GA revealed a high binding affinity for ATPase domain, with the equilibrium dissociation constant (K_D) of 3.23 × 10⁻⁶ ± 0.23 × 10⁻⁶ mol/L, which was much smaller than that of VP16 (4.16 × 10⁻⁵ ± 0.46 × 10⁻⁵ mol/L; Fig. 6A).

GA Potentially Binds to the ATP Pocket of Human Topo IIα. To identify the potential binding site of GA to the ATPase domain, docking simulation was done. To obtain the appropriate docking variables, ATP and ADPNP were initially docked with the ATPase domain of human Topo IIα. The result indicated that ADPNP lay in the ATP binding sites (Fig. 6B, left), suggesting that the variables for docking are reliable. Further studies showed that GA might bind at the ATP site of human Topo IIα ATPase domain (Fig. 6B, right). As indicated in the interaction model between GA and human Topo IIα ATPase domain, GA may form hydrogen bonds with the residues of the enzyme: The carboxylic acid group of GA forms bonds with residue Ser¹⁴⁸; the other oxygens in GA may form several hydrogen bonds with Arg⁹⁸, Ile¹⁴¹, Lys¹⁶⁸, and Thr¹⁴⁷, respectively. Remarkably, the long side chain of GA overlaps with the phosphate and ribose moieties of ADPNP in the ATP binding sites (Fig. 6B). All these observations indicate that GA may compete with ATP for binding to the ATP binding sites of human Topo IIα.

GA Blocked kDNA Decatenation in an ATP-Dependent Manner. To test whether GA could compete with ATP for binding to human Topo IIα, we did a competitive inhibition assay on kDNA decatenation catalyzed by human Topo IIα at different concentrations of ATP. As shown in Fig. 6C, increased concentration of ATP resulted in weaker inhibitory activity of GA on Topo IIα. This result strongly suggests that GA functions as an ATP competitor.

In the past few years, GA has been shown to have potent antitumor activity, and progress has been made in the mechanism exploration (9, 11–13). One notable result is that transferrin receptor has been identified as target of GA (9). However, we found that GA displayed nonselective cytotoxicity against the 21 cell lines tested (Fig. 1B) and, in particular, GA exhibited almost the equipotent cytotoxicity against T47D (high transferrin receptor cell surface expression) and HMEC (negative transferrin receptor cell surface expression).⁴ These results suggested that transferrin receptor might not be the dominant target of GA in the cell lines we tested. Here, we found that Topo IIα is a cellular target for GA to exert its antiproliferative activity, which will help understand the nature of GA and facilitate designing chemotherapy regimens in clinical application as well.

Current views are increasingly recognized that a large number of compounds that target Topo II have now been identified to be capable of simultaneously targeting Topo I, and vice versa. These are exemplified by aclarubicin, TAS 103, and tafluposide (F 11782; ref. 35). However, GA stands in contrast to this case. In the present study, we found that GA could dramatically block Topo IIα-catalyzed supercoiled DNA relaxation and kDNA decatenation, but to a comparatively less extent inhibit Topo I–mediated supercoiled DNA relaxation. Reduction of GA-induced apoptosis and abrogation of cellular response were only observed in cells with down-regulated Topo IIα but not in cells with knockdown of Topo IIβ or Topo I. Moreover, GA was reported to cause G₂-M arrest (10, 13), which is consistent with the notion that most Topo II inhibitors arrest cells in S or G₂-M phases. All these indicated GA preferentially functioned as a Topo IIα inhibitor in cells. However, the mechanism behind the selectivity that GA specially targets to human Topo IIα remains unknown. Given human Topo IIα is the major important target of anticancer therapy (38) due to its high abundance in rapidly proliferating cells and vital roles in mitotic processes, it is thus encouraged to predict that favorable inhibition of GA on Topo IIα may promise it to be more selective against rapidly proliferated tumor cells in its clinical settings.

Two classes of Topo II inhibitors have been characterized: topoisomerase poisons, which stabilize DNA cleavage complex and induce DNA damage (39), and the catalytic inhibitors, which interfere with different stages of the catalytic cycle (40). Our present study established GA as a catalytic inhibitor of Topo II in cell-free system and in cells based on the following evidences. First, GA failed to trap topoisomerase-DNA complexes in DNA cleavage assays. Second, GA was unable to stimulate DNA damage or induce γ-H2AX in cells. Finally, GA was able to abrogate DNA damage and cytotoxicity caused by well-characterized Topo II poison VP16 by using HL-60 and HeLa cells.

In the Topo II catalytic cycle, the enzyme binds to double-stranded DNA, introduces a transient DNA DSB, and passes the unbroken strand through the break. Topo II then religates the transient break and dissociates from DNA. The Topo II poison etoposide stabilizes the DNA-Topo II complex by preventing religation, and produces DNA DSBs. Thus, compounds that act at or before the formation of the transient DNA DSB are likely to attenuate the actions of Topo II poisons, such as aclarubicin, which intercalates into DNA (36), and merbarone, which acts by blocking DNA cleavage (22). In the present study, the fact that GA blocked enzyme-mediated DNA cleavage without intercalating into DNA or impairing enzyme-DNA interaction, substantiated that GA acts distinctly from aclarubicin, but similarly with merbarone.

Bypassing the above events, GA was particularly noted for its inhibitory action on enzyme-catalyzed ATP hydrolysis, a key step involved in course of the Topo II reaction. Thus far, the well-characterized drugs that impede ATP hydrolysis are coumarins such as novobiocin and chlorobiocin, and bisdioxopiperazines including ICRF-187, ICRF-193, and their analogues. Earlier works indicated that novobiocin and ICRF-193 were able to directly interact with the NH₂-terminal ATPase domain (24, 41, 42), suggesting a potential modulation of this domain on drug sensitivity. Data presented here further show that GA also exhibits high affinity for the ATPase domain of human Topo IIα, which may be responsible for its inhibition of enzyme-catalyzed ATP hydrolysis. Molecular docking predicted some potential binding sites for GA within ATPase domain, revealing a possibility that GA might compete with ATP for binding to human Topo IIα. We further showed that GA blocked DNA decatenation in an ATP-dependent manner, which strongly suggested the competition between GA and ATP of binding to human Topo IIα. However, we cannot rule out the possible involvement of other domains in the effect of GA on Topo II activities, because GA also inhibits the enzyme-mediated DNA cleavage. Thus far, the binding sites of most nonintercalative Topo II–targeted inhibitors and their precise mechanisms remain unclear. It is generally accepted that distinct regions of Topo II cooperatively modulate drug sensitivity. For instance, two potential binding sites for etoposide within human Topo IIα have been identified, one in the catalytic core and one in the ATPase domain (43). Point mutation assays indicate that different protein domains are involved in mediating the effect of bisdioxopiperazine compounds (44).

In summary, we provide evidence for the first time that GA should be classified as a catalytic inhibitor of human Topo IIα by binding to ATPase domain. This classification provides rationale for further clinical evaluation of GA, and favors it a promising anticancer candidate when combined with its well-defined antitumor activities and its distinct anti-MDR action.