Pseudolaric Acid B, a Novel Microtubule-Destabilizing Agent That Circumvents Multidrug Resistance Phenotype and Exhibits Antitumor Activity In vivo

Natural products have played pivotal roles in the drug discovery and development process. This is particularly evident in the field of cancer therapeutics, where >50% of the approved drugs introduced from 1981 to 2002 were of natural origin (1). Within this category, many exhibit adequate efficacy without the need for structural modifications (e.g., paclitaxel), whereas some require synthetic modifications (etoposide) for optimal function. It has been noted that natural products may embody more “privileged structures” than purely synthetic chemical libraries (2), and they are a rich resource of new chemical motifs. Therefore, the natural product–based drug discovery program remains an important avenue toward the development of small-molecule therapeutics for cancer as well as other diseases (3).

Among the very large number of proteins being targeted for anticancer drug development, microtubules have been suggested as one of best validated targets known today (4). It is because microtubules, which are polymers of α-tubulin and β-tubulin heterodimers, are indispensable to the mitotic process. Because tubulin is one of the most highly conserved proteins in evolution (5), it is not surprising that many antimitotic agents, identified through natural product screening, target the microtubules (6). Moreover, compounds with diverse chemical structures interact with tubulin. Well-known examples include the vinblastine and the paclitaxel classes of compounds. Although some of these microtubule-targeting agents, such as paclitaxel, have proven to be very effective in a range of malignancies, not all tumors are equally susceptible to these agents (7). Furthermore, neurotoxicity and acquired drug resistance are common problems associated with the use of tubulin inhibitors (7). In addition, recent observations indicate that microtubule-targeting agents exhibit antiangiogenic or antivascular properties (8) and that a remarkable synergism is observed when different microtubule-targeting agents have been used concurrently (9). This prompted efforts to optimize and combine known drugs for better overall activity as well as the development of novel microtubule-targeting agents.

The tree Pseudolarix kaempferi Gorden (Pinaceae) is indigenous to central China. The bark of this tree has been processed into the medicinal herb “tujinpi,” which is used in Chinese folk medicine to treat dermatologic fungal infections. Subsequent phytochemical studies led to the isolation and identification of the diterpenoids pseudolaric acid A (PAA; Fig. 1A) and pseudolaric acid B (PAB; Fig. 1B) as the major biologically active components (10, 11). Besides antifungal activities, studies on PAA and PAB revealed that they also display antifertility effects (12). Subsequently, it was discovered that PAA and PAB also exhibited considerable cytotoxicity (IC₅₀ ∼3 and 1 μmol/L, respectively) toward several cancer cell lines, including those of lung, colon, breast, brain, and renal origins (13). Preliminary minor modifications of the functional groups resulted in a dramatic reduction in cytotoxicity (13), suggesting that a rather specific structural configuration of this compound is required to exert its biological effects. At this point, neither the in vivo efficacy of this class of compounds against tumors nor its biological target was known. This impeded further development of these compounds into effective cancer therapeutics. In the present study, we provide evidence to show that PAB is an antimitotic compound that exerts its cytotoxic effect through targeting and destabilizing microtubules. Furthermore, we show that PAB circumvents the multidrug resistance (MDR) mechanism activated by P-glycoprotein (P-gp) overexpression in cells. Lastly, we show that PAB is active against liver tumors in the murine xenograft model, highlighting its potential use in the treatment of hepatocarcinoma that are resistant to other existing microtubule-targeting drugs.

Chemicals and antibodies. DMSO, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide salt, and etoposide were purchased from the Sigma Chemical Co., St. Louis, MO. Nocodazole, colchicine, paclitaxel, and vinblastine were purchased from Calbiochem, San Diego, CA. Protease inhibitor cocktail tablets were purchased from Roche (Indianapolis, IN). Anti-p21 and anti–cyclin B1 antibodies were purchased from Santa Cruz Biotechnology Inc., (Santa Cruz, CA). Anti-Chk1 (Ser³⁴⁵), anti-Chk2 (Thr⁶⁸), anti–caspase-3, anti-Chk1, and anti-Bid antibodies were purchased from Cell Signaling Technologies Inc. (Beverly, MA). Polyclonal anti-bcl-2 and anti–cytochrome c antibodies were purchased from Upstate Group (Serologicals Corp., Charlottesville, VA). Monoclonal anti–α-tubulin and β-actin antibodies were purchased from the Sigma Chemical.

Animals. Male BALB/c-nu/nu mice, originating from Charles River Laboratories, Inc., (Wilmington, VA) were obtained from the Laboratory Animal Unit, The University of Hong Kong (Hong Kong Special Administrative Region, China). All of the experiments were carried out in accordance with the guidelines of the Committee on the Use of Live Animals in Teaching and Research, The University of Hong Kong.

Purification of pseudolaric acid A and pseudolaric acid B from Pseudolarix kaempferi. PAA and PAB were isolated from the root bark of P. kaempferi as reported previously (14). Each compound was purified by repeated flash column chromatography on silica gel using hexane-ethyl acetate and dichloromethane-acetone mixtures and the purity was checked to be >98% by high-performance liquid chromatography using an Alltima C18 5 μm column and a mobile phase of methanol/water on an Agilent 1100 Series high-performance liquid chromatography.

Cell culture. All cells were obtained from the American Type Culture Collection (Rockville, MD) unless otherwise stated. The human breast cancer cell line MCF-7, liver cancer cell line Hep3B, colon cancer cell line HCT-8, mouse lung cancer LLC-1, and human breast carcinoma cell lines MDA435/LCC6 and MDA435/LCC6MDR (kindly provided by Dr. Robert Clarke, Georgetown University, Washington, DC) were maintained in DMEM. The human liver cancer cell line HepG2, human cervix cancer cell line HeLa, and human lung fibroblast CCD-19Lu were maintained in MEM. The human liver cancer cell lines QGY-7703 and QGY-TR50 (ref. 15; generously provided by Prof. Yong Xie, Hong Kong University of Science and Technology, Hong Kong Special Administrative Region, China), human nasopharyngeal cancer cell line CNE1, and human prostate cancer cell line LNCaP (generously provided by Prof. Y.C. Wong, The University of Hong Kong) were maintained in RPMI 1640. The human breast cancer MCF-10A cell line was maintained in DMEM/F-12, and the human colon cancer cell line SW620 was maintained in L15 medium without NaHCO₃. All media were supplemented with 10% fetal bovine serum and antibiotics. Cells were cultured at 37°C in the presence of 5% CO₂. The human colon cancer SW620 was cultured in a 37°C incubator.

Cytotoxicity assays. The test compounds were dissolved in DMSO at final concentrations of 100 mmol/L and stored at −20°C. Cytotoxicity was assessed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay as described (16). Cells were seeded on 96-well plates (2 × 10³-10 × 10³ cells per well). After overnight preincubation, the cells were exposed to different concentrations of PAB (0.039-100 μmol/L) or other test compounds for 6 days. Subsequently, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide reagent (10 μL) was added to each well and incubated at 37°C for 4 hours followed by the addition of 100 μL solubilization buffer (10% SDS in 0.01 mol/L HCl) and overnight incubation. A_{585 nm} was determined from each well on the next day. The percentage of cell viability was calculated using the following formula: Cell viability (%) = A_treated / A_control × 100. Data were obtained from three independent experiments.

Cell cycle analysis. After drug treatment, cells were harvested at indicated times, washed twice with PBS, and resuspended in 1 mL of 70% ice-cold ethanol for 30 minutes at 4°C. The cells were centrifuged and the pellets were resuspended in 100 μL PBS containing RNase A and propidium iodide (Beckman Coulter, Inc., Fullerton, CA). The DNA contents and cell cycle distributions were analyzed using an Epics XL-MCL flow cytometer (Beckman Coulter).

Protein extraction and Western blotting. After drug treatment, adherent and floating cells were lysed with lysis buffer [50 mmol/L Tris-HCl (pH 7.5), 250 mmol/L NaCl, 5 mmol/L EDTA, 50 mmol/L NaF, 1 mmol/L DTT, 1% Triton X-100, 1 mmol/L sodium orthovanadate, 100 μg/mL phenylmethylsulfonyl fluoride, protease inhibitor cocktail]. Protein concentrations were determined using the Bio-Rad protein assay (Bio-Rad Laboratories, Inc., Hercules, CA). Protein samples (50 μg each) were subjected to electrophoresis on SDS polyacrylamide gels and transferred to Hybond enhanced chemiluminescence nitrocellulose membranes (Amersham Biosciences, Piscataway, NJ), and the membranes were blocked with 5% dry milk protein. Membranes were then incubated with the indicated primary antibody. The binding of the antibody was visualized by peroxidase-coupled secondary anti-mouse antibody using enhanced chemiluminescence. To detect the release of cytochrome c, cells were treated as described with a few modifications (17). In brief, cells were harvested by centrifugation at 1,000 × g at 4°C for 5 minutes. Cell pellets were washed once with ice-cold PBS and resuspended in 5 volumes of buffer A [20 mmol/L HEPES-KOH (pH 7.5), 10 mmol/L KCl, 1.5 mmol/L MgCl₂, 1 mmol/L sodium EDTA, 1 mmol/L sodium EGTA, 1 mmol/L DTT, 0.1 mmol/L phenylmethylsulfonyl fluoride, protease inhibitor cocktail] containing 250 mmol/L sucrose. After incubation on ice for 15 minutes, cells were homogenized with a 25-gauge needle for 20 strokes, and homogenates were centrifuged at 1,000 × g at 4°C for 5 minutes. The supernatant was further centrifuged at 60,000 × g at 4°C for 1 hour, and the supernatant was used for Western blotting.

Indirect immunofluorescence microscopy. HeLa cells or human umbilical vascular endothelial cells (passages 2-5) were grown on glass coverslips in six-well plates (35-mm diameter). After exposure to drugs, cells were washed once with PBS and then fixed with 4% paraformaldehyde in PBS for 15 minutes. The fixed cells were washed thrice with PBS and then permeabilized with methanol for 2 minutes. The methanol-fixed cells were rehydrated with PBS and then stained with 4′,6-diamidino-2-phenylindole for 10 minutes to visualize the DNA. To stain microtubules, cells were then incubated with α-tubulin (1:2,000) in 3% bovine serum albumin for 2 hours. Excess antibody was removed by multiple washes with TBS [50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl]. TRITC-conjugated goat anti-rabbit IgG (1:500) antibody was used as the secondary antibody. After incubation for 1 hour, the cells were washed again with TBS thrice. Coverslips were mounted onto microscope slides with FluorSave reagent (Calbiochem) and examined under the Olympus IX71 microscope (Melville, NY). Images were captured with CCD digital camera RT COLOR, Spot (Diagnostic Instruments, Inc., Melville, NY).

Tubulin polymerization assay.In vitro tubulin polymerization was monitored using the Tubulin Polymerization Assay kit (Cytoskeleton, Inc., Denver, CO) according to the manufacturer's instructions. In brief, 10 μL of the drug in G-PEM buffer [80 mmol/L PIPES, 2 mmol/L MgCl₂, 0.5 mmol/L EGTA, 1 mmol/L GTP (pH 6.9), 5% (v/v) glycerol] were added to a 96-well plate (0.33 cm³ per well) and incubated at 37°C for 10 minutes. Then, 100 μL of 3.0 mg/mL reconstituted tubulin in G-PEM buffer at 4°C were pipetted into each designated well containing the drug of interest. The plates were incubated in a plate reader at 37°C and the absorbance was recorded immediately. The effects on polymerization of the test compounds were quantified by measuring the absorbance at 340 nm (A₃₄₀) for 60 minutes using a Spectramax OD Reader and SOFTmax PRO 2.2.1 software (Molecular Devices, Sunnyvale, CA) using the kinetic mode (one reading per minute).

[³H]Vinblastine and [³H]colchicine tubulin-binding assay. Colchicine and vinblastine tubulin-binding assays were done using the centrifugal gel filtration method (18) with minor modifications. Aliquots (50 μL) of 0.5 mg/mL purified tubulin were incubated separately with 40 nmol/L [³H]colchicine (1 mCi/mL; specific activity, 76.5 Ci/mmol; Perkin-Elmer, Melbourne, Australia) and 80 nmol/L [³H]vinblastine (0.25 mCi/mL; specific activity, 10.8 Ci/mmol; Amersham Biosciences) in the presence or absence of the test compound at the indicated concentrations for 10 minutes at room temperature. After the incubation, 50 μL of each sample were loaded onto a 1 mL spin column filled with Sephadex G-50 (Amersham Biosciences) and centrifuged at 2,000 × g for 1 minute to obtain the tubulin fraction. After the addition of the scintillation fluid, the radioactivity of [³H]ligand-bound tubulin was measured using a LS 6500 multipurpose scintillation counter (Beckman Coulter). The percentage of [³H]ligand binding was expressed as relative binding compared with control (tubulin + [³H]ligand only).

Assay of in vivo antitumor activity of pseudolaric acid B. The antitumor effect of PAB was examined in the murine xenograft model using the taxol-resistant QGY-TR50 human liver cancer cell line. A PAB solution was prepared by dissolving PAB in an aqueous solution of 6% polyethylene glycol (Sigma Chemical), 3% ethanol, and 1% Tween 80 (Sigma Chemical). A dose tolerance test was conducted to determine the appropriate dose ranges for PAB. Male, 5- to 7-week-old BALB/c-nu/nu mice were injected with 2 × 10⁶ tumor cells s.c. in the mid-dorsal region. Tumors were allowed to grow without treatment for 5 days. When the tumors became palpable (3-4 mm in diameter), PAB (at 10, 15, and 25 mg/kg/d in 0.1 mL) was given i.p. daily for 22 days. For the control group, mice received identical volumes of the aqueous solution without PAB. Animal weight and tumor dimensions were measured twice weekly with calipers, and tumor volumes were estimated using two-dimensional measurements of length and width and calculated with the formula [l × (w)²] × 0.52 (l is length and w is width) according to described procedures (19). Each treatment group consisted of at least 10 mice.

Pseudolaric acid B induces G₂-M cell cycle arrest and apoptosis of tumor cells. PAA and PAB from P. kaempferi were purified to apparent homogeneity by silica gel column chromatography. The purities of PAA and PAB were checked by high-performance liquid chromatography. Because it has been shown that PAB is comparatively more active than PAA in inhibiting the growth of various cancer cells (13), we focused on evaluating the cytotoxicity of PAB to a panel of cancer cell lines. Similar to previous reports, we found that PAB displays significant cytotoxicity (mean IC₅₀, 0.9 μmol/L) against tumor cells of different origins, including liver, lung, colon, breast, nasopharyngeal, cervix, and prostate (Fig. 1C). Apparently, the p53 status of these cells does not affect their sensitivity to this drug, because the IC₅₀ of cancer cells with wild-type p53 expression (e.g., HepG2 and MCF-7) was similar to those of cells in which the p53 gene was mutated or deleted (e.g., Hep3B and SW620). We found that PAB was ineffective against normal fibroblasts (CCD-19Lu) derived from the lung, as increasing concentrations of this compound (up to 100 μmol/L) failed to inhibit their proliferation (>50%; Fig. 1C). We further examined if PAB is selective for actively dividing cells. To this end, NIH-3T3 cells were stably transfected with the versican V1 isoform (20). Expression of the versican V1 isoform (NIH-3T3/V26-1) promotes cell proliferation and inhibits cell apoptosis markedly, in contrast to the control vector-transfected NIH-3T3 fibroblasts (21). Comparing the vector-transfected control (NIH-3T3/Vector Ctrl) and NIH-3T3/V26-1 cells, the latter were 5-fold more sensitive to PAB cytotoxicity (2.5 versus 0.47 μmol/L; Fig. 1C). Taken together, these data suggest that PAB is selectively inhibiting the growth of actively proliferating cancer cells, and its effect on slow growing cells derived from normal tissue is minimal.

To investigate the effect of PAB on cell cycle control, exponentially growing HepG2 cells were treated with 10 μmol/L PAB (a concentration equivalent to IC₇₀) for 0 to 36 hours followed by analysis using flow cytometry. As shown in Fig. 2A, treatment of HepG2 cells with PAB caused an evident reduction of cells in the G₀-G₁ phase and a concomitant accumulation of cells arrested in the G₂-M phase of the cell cycle. Treatment of other cell types originating from breast (MCF-7), colon (SW620), liver (Hep3B and QGY-TR50), and cervical (HeLa) cancers with PAB all led to a time-dependent G₂-M block in the cell cycle (Fig. 2B).

To investigate whether PAB treatment leads to cell apoptosis, we treated hepatocarcinoma cells QGY-TR50 with PAB and examined the activation of the apoptotic markers. Cells treated with 10 μmol/L PAB resulted in a time-dependent increase in caspase-3 and Bid cleavage, bcl-2 phosphorylation, and cytochrome c release, respectively (Fig. 2C), suggesting that PAB induces apoptotic cell death.

Pseudolaric acid B is a microtubule-targeting agent that induces mitotic arrest in tumor cells. A variety of anticancer agents induce cell cycle arrest at the G₂-M phase, leading to apoptosis. Examples include the topoisomerase II inhibitors, such as etoposide and doxorubicin, which are potent inducers of DNA strand break (22). Etoposide treatment activates the damage-responsive kinase ataxia telangiectasia mutated and the ataxia telangiectasia mutated and rad-3 related protein, which in turn phosphorylates the serine kinase Chk1 and Chk2, leading to the deactivation of Cdk1-cyclin B1 complex and causing cells to be arrested at the G₂-M boundary of the cell cycle (23). Microtubule-targeting agents, such as paclitaxel and nocodazole, which disrupt mitotic spindle dynamics (7), also induce cell cycle arrest at the mitotic phase. To better understand the mechanism of PAB-induced G₂-M arrest, HepG2 cells were treated with PAB, etoposide and nocodazole, respectively, and the expression of cell cycle–related proteins was analyzed. As expected, treatment of cells with etoposide resulted in the overexpression of total Chk1 and phosphorylated Chk1. In addition, the level of phosphorylated Chk2 was also elevated. Furthermore, p21, an inhibitor of the Cdk1-cyclin B1 complex (24), was also significantly amplified after treatment with etoposide (Fig. 3). On the contrary, treatment with nocodazole or PAB did not lead to a significant increase in any of the G₂ checkpoint proteins. However, cyclin B1, a marker for mitotic arrest induction, was up-regulated in the cells treated with PAB or nocodazole (Fig. 3). Taken together, these data show that PAB treatment does not activate the G₂ checkpoint and suggest that PAB-treated cells are probably arrested at the mitotic phase of the cell cycle.

To verify whether PAB induces mitotic arrest, we examined the effect of PAB on the organization of microtubules, as most drugs that cause mitotic arrest are microtubule targeting (25). To this end, we investigated the effect of PAB on microtubule networks by immunofluorescence staining of α-tubulin. As shown in Fig. 4, treatment of HeLa cells (Fig. 4A-H) and human umbilical vascular endothelial cells (Fig. 4I-P) with 2 μmol/L PAB resulted in a time-dependent disruption of the microtubule fibers of cells in the interphase of the cell cycle (compare Fig. 4A and B-D and Fig. 4I and J-L). This disruption is accompanied by cellular deformation, consistent with the role of microtubules in the maintenance of cell shape. Compared with the untreated control (Fig. 4E), a substantial increase in the proportion of cells was found in the prometaphase after 6 and 12 hours of PAB treatment (Fig. 4F and G). Whereas untreated cells undergoing mitosis have mitotic spindles, which could be easily identified (Fig. 4E and M), multipolar spindles with weak staining were found in the PAB-treated cells (Fig. 4F and G and Fig. 4N and O). Moreover, after 6 and 12 hours of PAB treatment, no cells in the metaphase were found (Fig. 4F and G). Although there was a reduction in the proportion of cells in prometaphase 24 hours after treatment (Fig. 4H), there was a significant increase in multinucleated cells (Fig. 4D and L) as well as a notable reduction in attached cells (data not shown). Taken together, these data suggest that PAB is a microtubule-disrupting agent. Like other known microtubule-disrupting compounds, such as colchicine or vinblastine, PAB interferes with the functions of the mitotic spindle apparatus and arrests cell cycle progression at the mitotic phase.

Pseudolaric acid B inhibits polymerization of purified tubulin in vitro. To determine whether PAB interacts with microtubules directly, we examined microtubule polymerization in vitro. Purified tubulin, in the absence of microtubule-associated protein, was allowed to polymerize in the presence or absence of PAB. As shown in Fig. 5A, in the presence of the microtubule-stabilizing agent, paclitaxel, a dramatic increase in the rate and extent of tubulin polymerization was observed compared with the control, whereas the addition of colchicine led to a pronounced reduction in microtubule assembly (IC₅₀, 1.4 ± 0.1 μmol/L, >90% inhibition at 10 μmol/L). Similarly, tubulin polymerization was also dose-dependently repressed by the addition of PAB (IC₅₀, 10.9 ± 1.8 μmol/L).

It has been reported that PAB exhibits a higher cytotoxicity than PAA across a panel of cancer cell lines (13). These two compounds differ only by a substituent on the cycloheptene hemisphere (Fig. 1, -methyl group in PAA and -carbomethoxy group in PAB). To investigate the structural-activity relationship, we compared the ability of PAA and PAB to inhibit tubulin polymerization. As shown in Fig. 5B, a structural change from a carbomethoxy group in PAB to a methyl group in PAA rendered PAA 50% less effective in repressing tubulin polymerization. The reduced capability of PAA in microtubule destabilization correlates with its decreased antiproliferative activity toward HepG2 and MCF-7 cells (Fig. 5C).

Pseudolaric acid B competes with colchicine for tubulin binding. The above findings suggest that PAB interacts with tubulin directly, leading to microtubule destabilization. Because a variety of known microtubule-disrupting agents interact at either the Vinca domain (26) or the colchicine site (27) of tubulin, we examined if PAB exerts its activity through binding to either of these two sites. To investigate if PAB binds to the colchicine site, a [³H]colchicine competition-binding assay was conducted. As expected, unlabeled colchicine (100 μmol/L) inhibited the binding of [³H]colchicine to tubulin (Fig. 5D). On the other hand, increasing concentrations of PAB gradually displaced [³H]colchicine from binding to tubulin (IC₅₀, 58.3 ± 5.7 μmol/L; Fig. 5D). In a parallel study using a [³H]vinblastine competition-binding assay, unlabeled vinblastine (100 μmol/L) inhibited the binding of [³H]vinblastine to tubulin, but increasing concentrations of PAB did not displace [³H]vinblastine (Fig. 5D). Taken together, these competition experiments suggest that PAB interacts with tubulin through the colchicine site and not at the Vinca domain.

Pseudolaric acid B is cytotoxic toward P-glycoprotein–overexpressing and multidrug-resistant tumor cell lines. One major obstacle in cancer therapy is dealing with acquired drug resistance of cancer cells. The MDR phenotype is commonly mediated by the overexpression of a 170-kDa P-gp efflux pump encoded by the mdr1 gene. This accounts for the acquired resistance to many tubulin-binding agents (28, 29). To study whether PAB is a substrate of the P-gp efflux pump, we tested the cytotoxicity of PAB toward a breast cancer cell line that had been transduced with a retroviral vector that directs the constitutive overexpression of mdr1 gene (MDA435/LCC6MDR1; ref. 30). Both doxorubicin and PAB were cytotoxic to the parental cell line MDA435/LCC6 with IC₅₀ of 0.2 and 0.47 μmol/L, respectively (Fig. 6A). However, whereas there was a 17-fold reduction in doxorubicin-mediated cytotoxicity (IC₅₀, 3.4 μmol/L) toward the MDA435/LCC6MDR1 cells, there was only a modest reduction in the cytotoxicity of PAB (IC₅₀, 1.1 μmol/L; Fig. 6A and B). We further evaluated the cytotoxicity of PAB toward the QGY-TR50 cell line, which is a paclitaxel-selected, P-gp-overexpressing, and MDR liver cancer cell line derived from the parental cell line QGY-7703 (15). Although taxol is 250-fold less effective toward QGY-TR50 compared with the parental cell line (data not shown), PAB exhibits similar cytotoxicity toward both cell lines (Fig. 6C). Collectively, these data suggest that PAB circumvents drug resistance induced by P-gp overexpression.

Efficacy of pseudolaric acid B in xenograft experiments in vivo. The efficacy of PAB against human tumors was assessed in nude mice. Our in vitro cytotoxicity assays showed that tumor cells of liver origin were most sensitive to PAB treatment. The MDR liver cancer cell line QGY-TR50 was chosen for testing the efficacy of PAB in vivo because, unlike HepG2 and Hep3B cells, it forms tumors readily when implanted in nude mice. Moreover, the MDR phenotype rendered QGY-TR50 a more rigorous model for evaluating the efficacy PAB in vivo. As shown in Fig. 6D, PAB treatment resulted in a dose-dependent suppression of tumor growth. Tumor growth inhibition was most evident in mice treated with PAB at 25 mg/kg/d, where ∼50% reduction in tumor size were observed compared with mice treated with the vehicle. The dosages tested were well tolerated by the animals, with no signs of toxicity or body weight loss. Moreover, we did not observe any animal lethality during the experiment.

The root bark of P. kaempferi has been used as antifungal remedy in traditional Chinese medicine for a long time, and more recently, its major constituents have been isolated and characterized (10, 11). PAB is the major antifungal constituent (10). Although structurally similar acids, including PAA, -C1, and -C2, have been isolated and identified from the extracts of the same plant, PAB showed the highest potency toward cancer cell lines (13). The present study investigated the mode of action accounting for the antitumor activity of PAB. We provided evidence showing that PAB exerts its cytotoxicity by directly interacting with tubulin, disrupting the formation of mitotic spindles and microtubules, leading to apoptotic cell death. Because microtubule-targeting agents, such as griseofulvin and spongistatins also exhibit antifungal activity (31–33), it is very likely that the microtubule-targeting property of PAB also accounts for its antifungal activity. Although recently it has been suggested that PAB could function as a peroxisome proliferator-activated receptor agonist (34), we believe that tubulin is the major target of PAB, because the amount required for peroxisome proliferator-activated receptor activation is 10- to 100-fold higher than that required for disrupting microtubules.

Microtubules are highly dynamic polymers that switch between phases of extension and shortening at individual microtubule ends (35). This process, known as microtubule dynamics, is particularly important for mitosis when the rapid reorganization of microtubules is required for the alignment of chromosomes during metaphase as well as for subsequent chromosome separation (36, 37). Given the indispensable role of microtubules during mitosis, microtubule-targeting agents have been shown to be effective against various cancers (7). We have shown presently that PAB displaces colchicine from binding to tubulin, suggesting that PAB may compete with colchicine for the same binding site. However, because PAB displaced colchicine only at high concentrations, this may reflect the lower affinity of PAB than colchicine for this binding site. Alternatively, given that PAB and colchicine have rather different chemical structures, it is also possible that PAB binds to a distinct site on tubulin, which induces conformational changes to the colchicine site and thus weakens colchicine binding. Further experiments to elucidate the exact binding site of PAB on tubulin may provide novel insights on the mechanisms of tubulin polymerization.

We have shown that PAB is unique in several aspects that render it an attractive candidate for further development into an anticancer drug. Firstly, PAB has broad cytotoxicity toward cancer cells of different origins, and the cytotoxicity is independent from the p53 status (Fig. 1C). Secondly, PAB could find potential application in the treatment of cancers that have developed acquired resistance to microtubule-targeting agents. The treatment of cancers with tubulin-binding agents, including taxanes and vincristine, is often associated with the eventual development of drug resistance, which lowers their efficacies significantly. It is well established that the overexpression of the P-gp efflux pump is one of major mechanisms leading to the MDR phenotype (38). Therefore, substantial efforts have been devoted to the development of P-gp inhibitors and the identification of new microtubule-targeting drugs that will not be removed by P-gp (39, 40). In these studies, we have shown that the overexpression of P-gp did not render the cells more resistant to PAB and that MDR cells seem to be equally sensitive to PAB. Finally, PAB is well tolerated by the animals, because there was no sign of toxicity or weight loss at efficacious dosages in our xenograft tumor model. At present, taxanes have been used to treat a variety of cancers, such as ovarian, breast, and non–small cell lung, whereas vinblastine and vincristine have been used in leukemia and breast cancer. However, these tubulin-targeting agents are ineffective against hepatocarcinoma. Further studies to explore the potential use of PAB in the treatment of hepatocarcinoma, for which there is no effective therapeutic at present, are warranted.

Besides being cytotoxic agents, some microtubule-targeting molecules are also potent antiangiogenic or antivascular agents. For example, the estradiol derivative 2-methoxyestradiol (41, 42) and the alkylating agent phenyl-3-(2-chloroethyl)ureas (43) are able to block angiogenesis. On the other hand, combretastatin derivatives, such as ZD6126, are antivascular agents that can induce rapid collapse of the tumor vasculatures, resulting in a reduction of blood flow to the tumor (44). Interestingly, while this article is in preparation, Li et al. showed that PAB exhibited antiangiogenic properties (45). Because it has been shown that microtubule-disrupting agents can inhibit angiogenesis through reducing the availability of hypoxia-inducible factor 1α (41), our findings also provide a molecular mechanism to account for the antiangiogenic properties of PAB.

Our structure-activity relationship studies indicated that the cytotoxicity of PAB is modified by the substituent at the cycloheptene hemisphere (Figs. 1 and 5B and C), providing a basis for subsequent structural modifications. At present, PAA and PAB can only be obtained from the bark of P. kaempferi. These natural compounds do not allow extensive modifications because of structural and functional group constraints. In addition, due to the rarity of this tree, it is difficult to obtain these compounds in large amounts. Recently, we have successfully achieved the total synthesis of both PAA and PAB.⁶

We thank G.W. Qin and S.K. Leung for assistance with the natural product isolation, Dr. Maurice Ho for help in vinblastine and colchicine tubulin-binding assay, Prof. Y.C. Wong for providing the LNCaP cells, Prof. Y. Xie for providing the QGY-7703 and QGY-TR50 cells, and Drs. D.Y. Jin and X.P. Lu for critical comments of the article.