CHM-1, a novel synthetic quinolone with potent and selective antimitotic antitumor activity against human hepatocellular carcinoma in vitro and in vivo

Hepatocellular carcinoma ranks as one of the most common cancers in the world (1). In Taiwan, where the incidence of hepatocellular carcinoma is among the highest in the world, chronic infection with hepatitis B virus appears to account for at least 70% of hepatocellular carcinoma (2). The incidence of hepatocellular carcinoma is dramatically increasing in the United States and other developed countries most likely due to the increasing prevalence of hepatitis C (3). Unfortunately, most patients with hepatocellular carcinomas are not curable because extensive resection is not possible, there is massive liver dysfunction caused by cirrhosis, or the disease is not identified at an early stage. Systemic chemotherapy and chemoembolization are the remaining treatment options in advanced hepatocellular carcinoma (4). However, hepatocellular carcinoma is well known for its expression of the multidrug resistance gene and its poor response to currently available chemotherapeutic agents (5, 6). Therefore, innovations to explore effective chemotherapeutic agents for hepatocellular carcinoma are urgently needed.

Antimitotic agents interact in the tubulin/microtubule system and constitute a major class of anticancer drugs (7). Because microtubules play crucial roles in the regulation of the mitotic apparatus, disrupting microtubules can induce cell cycle arrest in M phase, consequentially triggering of signals for apoptosis. Known antimitotic agents generally fall into three distinct groups (8). Microtubule-stabilizing agents, which bind to fully formed microtubules and prevent the depolymerization of tubulin subunits, are presented by paclitaxel and the taxanes, the epothilones, and discodermolide. The other two groups function by binding to the tubulin dimer and inhibiting their polymerization into microtubules. The Vinca alkaloids, typified by vinblastine, vincristine, and vinorelbine, are one of these groups. Colchicine and colchicine-site binders define the third group of antimitotic agents. Both the taxanes and the Vinca alkaloids are widely used to treat various kinds of human cancers, whereas colchicine-site binders are being vigorously pursued as potential new chemotherapeutic agents (9).

Quinolone derivatives are initially discovered as bacterial DNA gyrase inhibitors and thus developed as antibacterial agents with low side effects (10). Our previous work showed that 2-phenyl-4-quinolones exhibited promising cytotoxicity in a broad spectrum of human cancer cell lines and inhibited platelet aggregation (11–16). We also showed the 2-phenyl-4-quinolone series inhibited tubulin polymerization by binding to tubulin at the colchicine-binding site and was exerted as antimitotic agents (11–15). Recently, some 2-phenylpyrroloquinolin-4-ones were synthesized from the 2-phenyl-4-quinolone and have been shown to inhibit the growth of hepatocellular carcinoma both in vitro and in vivo (17). Notably, 2′-fluoro-6,7-methylenedioxy-2-phenyl-4-quinolone (CHM-1) has been identified from 6,7-substituted 2-phenyl-4-quinolone derivatives in our laboratory as an anti-invasive agent in hepatocellular carcinoma cells (18). We found that CHM-1 inhibited cell invasion via the down-regulation of MMP-9 in SK-Hep-1 cells. In this study, we investigated the antitumor activity of CHM-1 against hepatocellular carcinoma in both in vitro and in vivo assays and further elucidated the mechanism of CHM-1-induced effects in hepatocellular carcinoma cells.

CHM-1 was obtained from the Graduate Institute of Pharmaceutical Chemistry, School of Medicine, China Medical University. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), propidium iodide, anti-β-tubulin monoclonal antibody, anti-endonuclease G polyclonal antibody, FITC-conjugated secondary antibody, and other chemical agents were bought from Sigma Chemical. Antibody to anti-OxPhos complex IV and Mitotracker Red CMXRos were obtained from Molecular Probes. Antibodies to cyclin A, cyclin B1, Cdc2, and nucleolin were purchased from Santa Cruz Biotechnology. Antibodies to stathmin, phospho-Cdc2 (Tyr¹⁵), phospho-Cdc2 (Thr¹⁶¹), caspase-8, and caspase-9 were purchased from Cell Signaling Technologies. Antibody to caspase-3 was purchased from Imgenex. Antibodies to glyceraldehyde-3-phosphate dehydrogenase (GAPDH), cIAP1, cIAP2, XIAP, and survivin were purchased from Abcam. Antibodies to MAP4, caspase-6, and caspase-7 were purchased from BD Biosciences PharMingen. Antibody to MPM2 was purchased from Upstate Biotechnology.

The human hepatocellular carcinoma cell lines Hep3B and HepG2 were obtained from the American Type Culture Collection and cultured in DMEM. The human hepatocellular carcinoma cell line HA22T was obtained from Culture Collection and Research Center and cultured in DMEM. MRC5 (American Type Culture Collection) was cultured in MEM. AML12 (American Type Culture Collection) was cultured in DMEM/F-12. All cells were maintained in humidified air containing 5% CO₂ at 37°C.

Cells were inoculated in 96-well plates at a density of 10⁴ per well, and the percentage of cell survival was assessed using MTT colorimetric assay.

Terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling immunofluorescence staining (Promega) and Cell Death ELISA^PLUS kit (Roche Molecular Biochemicals) were used to determine apoptosis.

The cell cycle distribution was determined using FACScan flow cytometric analysis by DNA staining with propidium iodide.

After treatment, cells were fixed with methanol, blocked with 3% bovine serum albumin, stained with anti-β-tubulin or apoptosis-inducing factor (AIF) monoclonal antibody, and then FITC-conjugated antimouse IgG antibody. Mitotracker Red CMXRos (100 nmol/L) was added to the culture medium at 37°C for 30 min to stain mitochondria before fixation. Nuclear staining was done with 4,6-diamidino-2-phenylindole. Cells were imaged with Leica TCS SP2 Spectral Confocal System.

The CytoDYNAMIX ScreenTM3 kits was purchased from Cytoskeleton and used for the detection of polymerization of tubulin/microtubule.

The cell lysate in which the cytosolic and cytoskeletal fractions containing soluble (s) and polymerized (p) tubulin, respectively, was separated by centrifugation, resolved by electrophoresis through SDS-PAGE, and immunoblotted with an antibody against β-tubulin.

Nuclear and mitochondrial fractionation were done as described previously (19). Total protein was determined and equal amounts of protein were separated by 10-15% SDS-PAGE and immunoblotted with specific primary antibodies. The signal detected using an enhanced chemiluminescence detection kit (Amersham). For immunoprecipitation studies, total protein was precipitated with anti-Cdc2 antibody and mixed overnight at 4°C. The following day, protein A/G^PLUS-agarose beads were added and incubated with gentle rocking at 4°C. Then, immunoblotting was done using anti–cyclin B1 antibody and anti-Cdc2 antibody.

A Cdk1/Cdc2 Kinase Assay Kit (Upstate Biotechnology) was used to determine the activity of Cdc2 kinase according to the manufacturer's instructions.

The commercial assay kit (BioVision) was used to determine caspase activity.

Stealth RNA interference targeting AIF mRNA was obtained from Invitrogen and used according to the manufacturer's instructions. The AIF small interfering RNA (siRNA) sequence used was 5′-GGCCAGGGUACUGAUUGUAUCUGAA-3′, and the control siRNA, with no homology to any human mRNA, was 5′-GGCGGUGCAGUUAGUAUCUUACGAA-3′. For transfection, HA22T cells were seeded in 24-well plates or 10-cm dishes at 50% confluency with siRNA duplexes (200 nmol/L) using LipofectAMINE 2000 (Invitrogen). Following siRNA treatment, cells were treated with indicated agents for 24 h and then analyzed using Western blot analysis.

CHM-1 was dissolved completely in a vehicle mixture of DMSO/Cremophor EL/saline. As a comparison, the positive control doxorubicin was diluted with sterile saline before injection. Male severe combined immunodeficient mice (4-6 weeks of age), were purchased from the Laboratory Animal Center of the Medical College of National Taiwan University. HA22T cells (6 × 10⁶) were injected s.c. into the flank of each animal. When tumors reached an approximate volume of 50 mm³, mice were selected and distributed for drug studies. Tumor volumes were determined by measuring the length (l) and the width (w) and the volumes were calculated as V = lw² / 2. The mice were sacrificed and the tumors were excised after 1 month of treatment with CHM-1 and doxorubicin by i.p. injection twice weekly.

HA22T cells were injected i.p. at 10⁷ per mouse into male severe combined immunodeficient mice on day 0; all drugs were delivered by i.p. against i.p. tumors on day 1. CHM-1 and doxorubicin were administrated twice weekly until doxorubicin-treated mice were all dying. The mice were observed daily and any deaths were recorded. Antitumor activity was assessed as the ratio of median survival time in the treatment group (T) to median survival time in the control group (C), and the results are shown as T/C: lifespan T/C (%) = [(median survival time of drug-treated group) / (median survival time of control group)] × 100.

Immunohistochemical studies were done on formalin-fixed, paraffin-embedded tumor tissues as described previously (20). The primary antibodies AIF (Cell Signaling Technologies) and MPM2 (Dako) were used in this study. A standard labeled streptavidin-biotin technique (Dako) was used to detect the reaction products.

Data are presented as the mean ± SE for the indicated number of separate experiment. Statistical analyses of data were done with one-way ANOVA followed by Student's t test, and P values less than 0.05 were considered significant.

We first determined the effect of CHM-1 on the growth of hepatocellular carcinoma cell lines using the MTT assay. CHM-1 induced significant concentration-dependent growth inhibition in HA22T, Hep3B, and HepG2 cells, with the most potent effects observed in HA22T cells (IC₅₀, 0.75 ± 0.11 μmol/L; Fig. 1B, left). Importantly, CHM-1 was less toxic to normal cell types, including human fibroblasts (MRC5) and mouse hepatocytes (AML12; Fig. 1B, right). These data indicate that CHM-1 induces selective cytotoxicity in human hepatocellular carcinoma cells but less in normal cells. CHM-1 caused the accumulation of G₂-M phase with concomitant losses in G₀-G₁ phase, with a maximum effect observed at 24 h (Fig. 1C). The accumulation of cells with G₂-M DNA content was followed by an increase in hypodiploid cells at the later time points (48 h) as indicated as apoptotic cells. The terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling assay and quantitative assessment of oligonucleosomal DNA fragmentation were then used to identify apoptotic cell death. CHM-1 induced cell apoptosis with positive terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling reactions in HA22T cells (Fig. 1D, left). After 24- and 48-h treatment, CHM-1 caused apoptotic cell death in a concentration-dependent manner (Fig. 1D, right). These results indicate that CHM-1 induced G₂-M arrest of the cell cycle followed by apoptosis.

As shown in Fig. 2A, the microtubule network exhibited normal arrangement and organization in HA22T cells in the absence of drug treatment. CHM-1 treatment caused changes in microtubule distribution starting with a 4-h exposure, and treatment with CHM-1 for 24 h resulted in findings similar to those of colchicine-induced microtubule changes. Colchicine and vincristine caused cellular microtubule depolymerization with short microtubule fragments scattered throughout cytoplasm. Paclitaxel resulted in microtubule polymerization with an increase in the density of cellular microtubules. In in vitro microtubule assembly assay, CHM-1 inhibited tubulin polymerization in a concentration-dependent manner (Fig. 2B). For comparison, tubulin polymerization was significantly promoted by paclitaxel and completely inhibited by vincristine. The effect of CHM-1 on microtubule assembly was compared with that of colchicine, paclitaxel, and vincristine in in vivo assays. After cells were treated with various antimitotic agents for 24 h, CHM-1, colchicine, and vincristine caused inhibition of microtubule assembly (Fig. 2C). In contrast, paclitaxel induced dramatic promotion of tubulin polymerization. The percentage of polymerized tubulin levels was 40.5 ± 2.0% for control samples and 20.8 ± 6.2%, 22.2 ± 5.8%, 57.1 ± 1.4%, and 36.2 ± 1.2% for treatment with CHM-1, colchicine, paclitaxel, and vincristine, respectively. Stathmin is a major microtubule-destabilizing protein that promotes microtubule depolymerization, and MAP4 is a microtubule-associated protein that stabilizes microtubules (21). In the present study, CHM-1 (1-10 μmol/L) did not affect the expression of stathmin but induced a concentration-dependent inhibition of MAP4 expression (Fig. 2D).

As shown in Fig. 3A, CHM-1 did not change the levels of cyclin A, but the expression of cyclin B1 was sustained at high levels in CHM-1-treated cells. There was no difference in total Cdc2 and Tyr¹⁵-phosphorylated Cdc2 expression after CHM-1 treatment. Nonetheless, phosphorylation at the Thr¹⁶¹ site of Cdc2 was increased after CHM-1 treatment at the indicated times. Furthermore, CHM-1 significantly increased the binding of cyclin B1 to Cdc2 in HA22T cells (Fig. 3B). The Cdc2 kinase activity was low in control cells and was substantially increased in cells treated with CHM-1, similar to that of paclitaxel (Fig. 3C, left). Moreover, we found that treating the cells with CHM-1 together with roscovitine, a specific cyclin-dependent kinase inhibitor, significantly abolished CHM-1-induced Cdc2 activity and growth inhibition (Fig. 3C, right). Next, we examined the status of phosphorylated polypeptides found only in mitotic cells using MPM2 antibody. After 24 h of treatment, CHM-1 induced a concentration-dependent elevation in MPM2 phosphoepitopes, and CHM-1-induced effects were prevented by cotreatment with roscovitine (Fig. 3D).

As shown in Fig. 4A, CHM-1 did not induce the activation of effector caspases (caspase-3, caspase-6, and caspase-7) or initiator caspases (caspase-8 and caspase-9) in HA22T cells. After treatment for 12 and 24 h, the levels of cIAP1, cIAP2, XIAP, and survivin were not altered in CHM-1-treated cells (Fig. 4B). CHM-1 also did not increase the activity of caspase-3, caspase-8, or caspase-9 after a 24-h application (Fig. 4C). Staurosporine was used as a positive control for the cleavage and activation of caspases. Moreover, z-VAD-fmk, a general inhibitor of caspases, did not inhibit CHM-1-induced growth inhibition in HA22T cells (Fig. 4D).

Subcellular fraction showed that treatment of cells with CHM-1 dramatically redistributed AIF, but not endonuclease G, from the mitochondria to the cytosol, subsequent to the nucleus (Fig. 5A). Analysis by confocal microscopy revealed that AIF was translocated into the nucleus and caused nuclear condensation after 24-h treatment with CHM-1 (Fig. 5B). As shown in Fig. 5C, both CHM-1 and roscovitine induced AIF release in the cytosolic fraction, and cotreatment with roscovitine did not prevent CHM-1-induced AIF translocation. We next showed that the AIF protein level was greatly reduced at 24 h after treatment with AIF siRNA but not after treatment with control siRNA (Fig. 5D, top). Moreover, siRNA targeting of the AIF substantially attenuated CHM-1-induced AIF translocation to the cytosol (Fig. 5D, bottom). Thus, this RNA interference further confirms that the mitochondrial-nuclear translocation of AIF is indeed the target of CHM-1 treatment.

To investigate the in vivo antitumor activity of CHM-1, we carried out xenografts with HA22T cells in athymic severe combined immunodeficient mice. As shown in Fig. 6A, CHM-1 induced a dose-dependent inhibition of HA22T tumor growth (10 mg/kg CHM-1, 128 ± 26 mm³, P < 0.01 versus vehicle; vehicle-treated group = 491 ± 83 mm³ on day 70). In line with the tumor growth suppression results, CHM-1 prolonged the lifespan of the HA22T-bearing mice (Fig. 6B). CHM-1 showed significant antitumor activity, with a T/C value of 167% (P < 0.01). In comparison, doxorubicin conferred no antitumor benefit and the doxorubicin-treated mice all died on day 53 possibly due to the toxicity of the drug. Moreover, in these two animal models, body weights were not significantly affected by CHM-1. In support of the antitumor mechanism in CHM-1-treated cells, we next examined the immunohistochemical AIF and MPM2 protein expression using tumor specimens from control and CHM-1-treated animals. The distribution of AIF was almost detected in the cytosol in the vehicle-treated group, but the translocation of AIF and accumulation of MPM2 in nuclei of apoptotic cells increased significantly in tissue treated with CHM-1 (Fig. 6C).

In this study, CHM-1 exhibited promising antitumor actions against human hepatocellular carcinoma cells. CHM-1 induced a significant concentration-dependent growth inhibition and apoptosis in hepatocellular carcinoma cells. CHM-1 produced a dose-dependent tumor regression and prolonged the lifespan of mice carrying liver cancer xenografts, indicating its in vivo efficacy. We also compared the influence of CHM-1 with that of doxorubicin in hepatocellular carcinoma and normal cells. The range of IC₅₀ values between hepatocellular carcinoma cells and normal cells were broader for CHM-1 than for doxorubicin. In the in vivo model, the antitumor activity of CHM-1 was found to be equal and even superior to doxorubicin; importantly, body weight was not significantly affected by CHM-1 compared with doxorubicin. Therefore, we suggest that CHM-1 could be a more efficacious and safe anticancer agent for treatment of human hepatocellular carcinomas.

In previous studies, we showed 2-phenyl-4-quinolone derivatives, including CHM-1, interacted with tubulin at colchicine-binding site and inhibited tubulin polymerization with IC₅₀ values equivalent to those obtained with podophyllotoxin and combretastatin A-4 (11–15). Consistent with these biochemical effects, in intact cells, CHM-1 disrupted the intracellular microtubule network and caused inhibition of the microtubule assembly. Numerous proteins and kinases that interact with microtubules and/or free tubulin dimmer also have the potential to regulate both catastrophe and rescue rates (22). The overexpression/activation of stathmin and/or the down-regulation/inactivation of MAP4 should increase the dynamicity and decrease the stability of microtubules. In our study, the expression of MAP4 was inhibited by CHM-1 in HA22T cells. This suggests that the down-regulation of MAP4 may be involved in CHM-1-induced inhibition of tubulin polymerization.

To allow cells to progress from the G₂ to the M phase, the Cdc2/cyclin B1 complex is active (23), meaning that Cdc2 has undergone an activating phosphorylation (on Thr¹⁶¹) by CAK, that inhibitory phosphorylation (on Thr¹⁴ and Tyr¹⁵) has been removed by active Cdc25C phosphatase, that Cdc2 associates with cyclin B1, and that complex translocates to the nucleus (24). In this study, CHM-1 arrested the growth of cancer cells at the G₂-M phase and then induced apoptotic cell death. CHM-1 significantly increased the protein level of cyclin B1, Thr¹⁶¹-phosphorylated Cdc2, but had no effect on the expression of Tyr¹⁵-phosphorylated Cdc2 and total Cdc2. The kinase assay revealed that CHM-1 activated Cdc2 kinase activity, and the formation of Cdc2/cyclin B1 complex was increased by CHM-1. The pattern of CHM-1-induced activation of cyclin B1 and Cdc2 kinase activity is consistent with the effects of the antimitotic agents (25, 26). In addition, the CHM-1-induced activation of Cdc2 kinase activity, the elevation of MPM2 phosphoepitopes, and the inhibition of cell growth were all attenuated by roscovitine. Taken together, we show that in addition to directly disrupting microtubules CHM-1 induces mitotic arrest by increasing the activation of Cdc2/cyclin B1 complex activity.

Caspases exist as inactive zymogens in normal and surviving cells and perform proteolytic processing on activation during apoptosis. Members of the inhibitor of apoptosis protein family counteract both the activation and the activity of caspases (27). In the present study, neither initiator caspases nor effector caspases were activated by CHM-1. Furthermore, the general caspase inhibitor z-VAD-fmk could not prevent CHM-1-induced cell death. Therefore, CHM-1 appears to induce apoptosis via a caspase-independent mechanism in human hepatocellular carcinoma cells. Recent studies suggest that caspase activation is not the sole pathway to induce apoptotic cell death (28). Mitochondria can release factors involved in caspase-independent cell death, including AIF and endonuclease G. AIF is ubiquitously expressed both in normal tissues and in a variety of cancer cell lines and is found to be localized in mitochondria-rich areas (29). Herein, our data showed that CHM-1 induced AIF translocation and nuclear condensation in HA22T cells. CHM-1-induced AIF translocation was not attenuated by roscovitine, and roscovitine alone would trigger AIF release in cytosol. This roscovitine-induced event was also observed in U937 cells (30). Moreover, siRNA knockdown of the AIF protein in HA22T cells effectively attenuated CHM-1-induced AIF translocation. In immunohistochemical analyses, we found that CHM-1 significantly increased AIF translocation in nuclei in HA22T tumor tissues. These findings suggest that AIF-dependent pathway plays a pivotal role in CHM-1-induced apoptosis in human hepatocellular carcinoma cells.

Caspase-independent cell death pathways are important safeguard mechanisms to protect the organism against unwanted and potential harmful cells when caspase-mediated routes fail but can also be triggered in response to cytotoxic agents or other death stimuli (31). The major challenge in treating cancer is that many tumor cells carry mutations in key apoptotic genes, such as p53, Bcl-2 family proteins, or those affecting caspase signaling, thus making treatment with traditional chemotherapeutic agents ineffective (32). The triggering of caspase-independent processes, therefore, has become an alternative approach to eradicate tumor cells (33, 34). Accordingly, the capability of inducing cell death even in the absence of the caspase cascade may be a key feature of CHM-1 and explains its efficiency to trigger cell death in a variety of drug-resistant cell types including, in particular, tumor cells with defects in caspase activation.

In this study, we show that CHM-1 induces mitotic arrest and apoptosis by binding to tubulin and inhibiting tubulin polymerization. CHM-1 causes mitotic arrest, at least partly, by modulating Cdc2/cyclin B1 complex activity and then apoptosis executed by a caspase-independent pathway. Additional studies on the pharmacokinetic properties of CHM-1, including absorption, distribution, metabolism, excretion, and toxicity, are needed to proceed in the future. We suggest that CHM-1 deserves further investigation and has the potential to replace doxorubicin, which is used in the systemic chemotherapy or transarterial chemoembolization for patients with hepatocellular carcinoma.