Potent Antiandrogen and Androgen Receptor Activities of an Angelica gigas–Containing Herbal Formulation: Identification of Decursin as a Novel and Active Compound with Implications for Prevention and Treatment of Prostate Cancer

Prostate cancer is second only to lung cancer as the cause of cancer deaths in American men, responsible for an estimated 30,000 deaths per year (1). Cytotoxic chemotherapy and radiotherapy provide no survival benefit to patients with hormone-refractory prostate cancer, which usually arises after hormonal ablation therapies (2–4). Presently, only some taxane drugs have shown limited efficacy for advanced prostate cancer (5). All of these treatments cause significant side effects. Nontoxic alternatives are needed to decrease the risk and burden of prostate cancer.

Chemoprevention has become recognized as a plausible and cost-effective approach to reducing cancer morbidity and mortality by inhibiting precancerous events before the occurrence of clinical disease (6, 7). Androgen and androgen receptor (AR)–mediated signaling are crucial for the development and function of the normal prostate (2) as well as for prostate cancer (2, 3). The importance of androgen in prostate cancer is supported by the observations that prostate cancer rarely occurs in eunuchs or in men with deficiency in 5α-reductases, the enzymes that convert testosterone to its active metabolite 5α-dihydrotestosterone (2, 3). Targeting androgen and AR signaling represents a rational strategy for the chemoprevention of prostate cancer (6, 7). As a proof of concept, a clinical trial with finasteride (Proscar), which inhibits 5α-reductase II within the prostate gland, had shown a significant reduction of total prostate cancer incidence (8). However, prostate cancer that developed in subjects in the intervention group seemed to be more advanced in tumor stages than those from the placebo group, raising doubt about the overall survival benefit of this single-target approach. Novel agents that target multiple aspects of androgen and AR signaling will be more desirable.

Oriental herbal medicine has been used since ancient times to treat malignancies. Systematic characterization of active phytochemicals in medicinal herbs and their mechanisms of action are important for providing the rationale for their efficacy and for transforming herbal practices into evidence-based medicine. In collaborative work aimed at developing safe and efficacious Oriental herbal formulas for prostate cancer chemoprevention, our laboratories focused on herbs and their phytochemicals that target androgen and AR signaling as well as cell cycle arrest and apoptosis. With respect to biomarkers of androgen and AR signaling, prostate-specific antigen (PSA) is a gene tightly regulated by androgen in normal prostate and some prostate cancer cells (9). A member of the kallikrein family (KLN3), PSA is a serine protease with highly prostate-specific expression and is elevated in blood circulation of patients with prostate cancer. Circulating PSA is widely used clinically as a marker for prostate cancer screening and is particularly useful as an indicator of prostate cancer response to therapy and recurrence (9). The LNCaP human prostate cancer cells are perhaps the best-studied in vitro model for androgen and AR signaling in prostate cancer. They possess a high-affinity mutant AR and produce high levels of PSA, which is extremely responsive to androgen stimulation (10, 11). In addition to PSA expression, another known outcome of androgen deprivation or blockage of AR signaling in these cells is an induction of neurite-like projections that have been termed neuroendocrine differentiation (12–15). Androgen signaling represses these morphologic manifestations and the associated molecular markers, such as neuron-specific enolase (NSE; ref. 15). Therefore, we chose this cell line as the primary cell targets for screening novel antiandrogen and AR agents using PSA and neuroendocrine differentiation as key functional biomarkers.

We report here strong antiandrogen and AR activities of the ethanol extract of an anticancer formula (KMKKT) containing Korean Angelica gigas Nakai (Umbelliferae family; AGN) root and nine other herbs in LNCaP cells. We describe, for the first time, the identification of decursin from AGN as a novel antiandrogen and AR agent. Decursin was first isolated from Angelica decursiva (Fr. et Sav.) in Japan in 1966 and later from AGN (16, 17). Decursin has been reported to induce cytotoxic activity of leukemia cell lines in vitro (17, 18) and to be active against sarcoma growth in an animal model (19). For prostate cancer cells, a recent article has shown G₁ cell cycle arrest and apoptosis induced by decursin in LNCaP, DU145, and PC-3 cell lines (20). The antiandrogen and AR activities described in the present article, together with the reported antiproliferative and apoptotic activities of decursin in prostate cancer cells, have significant implications for the chemoprevention and treatment of prostate cancer with decursin and decursin-containing herbs and formulas.

KMKKT. This herbal formula was originally designed to promote vital energy, to remove blood stasis, and to decrease inflammation for treating lung cancer. The 10 ingredients and their proportion (w/w) were as follows: Benincasa hispida, China, 17.2%; Bletilla striata, China, 8.6%; Tulipa edulis, Korea, 8.6%; Panax ginseng, Korea, 8.6%; Phaseolus angularis, Korea, 17.2%; Zanthoxylum piperitum, Korea, 6.9%; Patrinia villosa, China, 8.6%; Astragalus membranaceus, Korea, 8.6%; Asini gelatinum, Korea, 8.6%; and AGN, Korea, 6.9%. The herbal ingredients were obtained from Oriental Medical Hospital of Kyunghee University (Seoul, Korea) and kindly authenticated by Dr. Nam-In Baek (Department of Oriental Herbal Materials, Kyunghee University). In studies to be reported elsewhere, the ethanol extract of this formula given by i.p. injection every other day decreased the growth of human PC-3 prostate cancer tumor xenograft in athymic nude mice by 68% with a dose of 100 mg/kg and decreased the growth of transplantable mouse Lewis lung carcinoma in syngeneic mice by 86%, without any adverse effect on body weight of the mice.³

Extraction of KMKKT and constituent herbs. The ethanol extract was prepared as follows: The dried and pulverized medicinal herbs were mixed together and each 350-g batch was soaked with ethanol (2 liters × 3 changes) at room temperature for 7 days. The extract was filtered through filter paper (pore size, 3 μm), evaporated (rotary evaporator, model NE-1, Japan), and lyophilized (freeze dryer, Lioalfa-6, Telstar, Terrassa, Spain) to produce 12.9 g of powder (yield, 3.7%). The individual herbs were extracted by the same procedure for each 200-g batch with 1 liter ethanol × 3 changes. The yield (w/w) was as follows: B. hispida, 1.76%; B. striata, 5.8%; T. edulis, 1.2%; P. ginseng, 2.15%; P. angularis, 1.69%; Z. piperitum, 7.15%; P. villosa, 10.6%; A. membranaceus, 4.12%; A. gelatinum, 0.05%; and AGN, 6.48%.

Fractionation of AGN extract. AGN powder (1.2 kg) was obtained from Kyung-Dong Pharmaceutical Co. (Seoul, Korea) and was extracted thrice with ethanol. After filtration, concentration, and freeze drying as described above, the ethanol extract (253.6 g) was obtained (yield, ∼21.1%). The ethanol extract was reconstituted in distilled water and partitioned with an equal volume of n-hexane. The resulting aqueous phase was partitioned in succession with an equal volume of methylene chloride, ethyl acetate, and butanol, respectively. The n-hexane, methylene chloride, ethyl acetate, and butanol fractions and the residual liquid were concentrated and lyophilized. The yields for the various fractions as percentage of input AGN ethanol extract were as follows: n-hexane fraction (2.94 g, 1.16%), methylene chloride fraction (35.2 g, 13.8%), ethyl acetate fraction (16.0 g, 6.3%), butanol fraction (12.2 g, 4.8%), and residue (17.0 g, 6.7%).

Purified and synthetic decursin. Decursin was extracted and purified by methods reported in ref. 19. The purity was determined to be ∼98.6%. A crystalline sample (100%) of synthetic decursin (18) was kindly provided by Prof. Hogyu Han (Department of Chemistry, Korea University, Seoul, Korea).

Assay for decursin. Decursin standard and extract/fraction samples were spotted on Silica Gel 60 thin-layer plates (20 × 20 cm; Merck, Darmstadt, Germany). After development in hexane/ethyl acetate (1:2 v/v), 10% sulfuric acid was sprayed on the TLC plates to char carbon-containing compounds. A high-performance liquid chromatography (HPLC) method was also developed to measure decursin contents in herbal extracts (Supplementary Fig. S1).

Cell culture and treatments. LNCaP cells were obtained from the American Type Culture Collection (Manassas, VA) and grown in RPMI 1640 supplemented with 10% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA) and 2 mmol/L l-glutamine, 10 mmol/L HEPES, 1 mmol/L sodium pyruvate, and 45 g/L glucose without antibiotics. When cells were 50% to 60% confluent (usually 48 hours after plating), the medium was changed and treatment with ethanol extracts of KMKKT, its constituent herbs, or purified decursin was started. To standardize the condition of agent exposure, cells were cultured in a volume-to-surface area ratio of 0.2 mL/cm² (15 mL for a T75 flask and 5 mL for a T25 flask).

In experiments where androgen stimulation was required, cells were seeded in phenol red–free medium containing 5% charcoal-stripped serum (CCS; Atlanta Biologicals) to decrease background signaling. Androgen stimulation was provided by a nonmetabolizable analogue mibolerone, which was a kind gift from Dr. Charles Young (Mayo Clinic, Rochester, MN). Bicalutamide, also known as Casodex, was purchased from LKT Labs (St. Paul, MN) and used as a positive antiandrogen agent for comparison with decursin in key experiments.

Cell growth and cell death/apoptosis assays. The effect of herbal extracts on cell number was estimated by either quantitating cell protein yield by Lowry method or by the mitochondrial metabolism-based 2,3-bis[2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanilide, inner salt or 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS) colorimetric methods in 96-well plate format (21). Cell cycle analyses were carried out with propidium iodide staining according to Krishan's protocol (22) and flow cytometry using a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA). Cell death was detected by either an ELISA kit purchased from Roche Diagnostics Corp. (Indianapolis, IN; ref. 23) or by an immunoblot analysis of the caspase-mediated cleavage of poly(ADP-ribose) polymerase (PARP) as described previously (24).

ELISA for PSA protein in conditioned medium and cellular extract. An assay kit from United Biotech, Inc. (Mountain View, CA) was used for measurement of PSA in both conditioned medium (secreted) and cell lysate (cellular) as described previously (25).

Reverse transcription-PCR for mRNA of PSA, AR, NSE, and glyceraldehyde-3-phosphate dehydrogenase. After treatment for the desired duration, LNCaP cells were used for RNA extraction by RNeasy kit (Qiagen, Valencia, CA). Reverse transcription was done with 3 to 4 μg total RNA and oligo(dT) primers using SuperScript II RT (Life Technologies, Foster City, CA). PCR was carried out using Qiagen HotStarTaq Master Mix kit under optimized conditions for detecting differences in transcript abundance. Oligonucleotide primers were synthesized by Sigma-Genosys (The Woodlands, TX) as follows: (a) PSA gene (710 bp), 22 cycles, forward 5′-GATGACTCCAGCCACGACCT-3′ and reverse 5′-CACAGACACCCCATCCTATC-3′, annealing temperature 57°C; (b) AR gene (590 bp), 32 cycles, forward 5′-ATGGAAGTGCAGTTAGGG-3′ and reverse 5′-CAGGATGTCTTTAAGGTCAGC-3′, annealing temperature 57°C; (c) NSE gene (662 bp; Genbank X51956), 40 cycles, forward 5′-GTTCTGAACGTCTGGCTAAATAC-3′ and reverse 5′-CATTGAGTTATGGGGAAATGA-3′, annealing temperature 60°C; and (d) housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (230 bp), 25 cycles, forward 5′-TCAAGAAGGTGGTGAAGCAG-3′ and reverse 5′-CTTACTCCTTGGAGGCCATG-3′, annealing temperature 57°C.

Immunoblotting of PSA, AR, α-tubulin, and cleaved and total PARP. Cell lysate preparation and immunoblotting were as described previously (24, 25). Antibody for PSA was purchased from DAKO (Glostrup, Denmark), and an antibody for AR was purchased from BD PharMingen (San Diego, CA).

In experiments assessing effects of decursin or herbal extracts on AR nuclear translocation, the cells were harvested and separated into nuclear and cytosolic fractions using NE-PER Nuclear and Cytoplasmic Extraction Reagents kit (Pierce, Rockford, IL) and the specificity of separation was confirmed by immunoblotting for total PARP (antibody from Cell Signaling Technologies, Beverly, MA) and α-tubulin (antibody from Santa Cruz Laboratories, Santa Cruz, CA) as the respective nuclear and cytosol markers.

PSA promoter-luciferase assay. LNCaP cells were cotransfected with a luciferase reporter plasmid driven by a 6-kb PSA promoter and a cytomegalovirus (CMV)-β-galactosidase (β-gal) vector (generous gifts of Dr. Charles Young) and incubated in serum-free, phenol red–free medium for 24 hours to reduce AR signaling as described previously (25). The cells were treated with different levels of decursin plus androgen (0.1 nmol/L mibolerone) for 24 hours in 5% CSS medium. Cell extracts were prepared for luciferase and β-gal assays. Luciferase activities were normalized to β-gal activities to correct for differences in transfection efficiency.

Neuroendocrine differentiation. LNCaP cells were seeded into T25 flasks to ∼20% confluence with phenol red–free medium containing 5% CSS unless noted otherwise in some experiments in which 5% whole serum was used. Treatment with KMKKT, AGN, or decursin was initiated at seeding time in most experiments. Androgen stimulation with mibolerone was started 2 days after seeding. The overall extent of neuroendocrine differentiation for each flask was estimated by examining at least five random fields under ×100 magnification. Then, representative morphologic changes were recorded with a CCD camera.

Replications and statistical evaluations. Each result was replicated in at least two independent experiments, and many were done multiple times. When appropriate and necessary, ANOVAs and two-tailed t tests were used to test statistical significance of differences from untreated control.

Effects of KMKKT on cell growth and apoptosis. As shown in Fig. 1A, (left), KMKKT treatment for 48 hours in complete medium containing 10% whole serum inhibited LNCaP cell growth in a concentration-dependent manner in the range of 20 to 100 μg/mL. At ≥200 μg/mL and longer exposure (72 hours), KMKKT induced apoptotic DNA fragmentation (Fig. 1A,, right) and increased PARP cleavage (data not shown). The growth suppression effect was accompanied by G₁ arrest (Table 1). Therefore, KMKKT caused G₁ arrest at low concentration levels and apoptosis at ≥200 μg/mL concentrations.

KMKKT exerts a sustained inhibition of PSA expression. We next determined the effect of nonapoptotic concentrations of KMKKT on PSA expression and secretion after 48-hour exposure. As shown in Fig. 1B, (left), KMKKT decreased cellular PSA protein level and PSA secreted into the conditioned medium in a concentration-dependent manner. The IC₅₀ was estimated to be ∼7 μg/mL based on three independent experiments. In time course experiments, exposure to 50 μg/mL KMKKT for 6 hours did not decrease cellular PSA level and by 12 hours decreased cellular PSA by half (Fig. 1B,, right). By 72 hours, the cellular PSA expression was almost completely blocked (Fig. 1B). The secreted PSA followed the same kinetics of suppression as the cellular PSA.

To determine whether the effect of KMKKT occurred at the transcription level, we examined the steady-state level of the PSA mRNA over a course of exposure of 48 to 96 hours (Fig. 1C,, left) and in the acute duration of 9 to 24 hours (Fig. 1C,, right). We included methylseleninic acid as a positive treatment that we had shown earlier to decrease PSA mRNA and AR mRNA (25) to validate the reverse transcription-PCR (RT-PCR) detection methodology (Fig. 1C,, left). The RT-PCR assay detected a significant reduction of PSA mRNA abundance at 48 and 96 hours of KMKKT treatment (Fig. 1C), in excellent agreement with the ELISA data in Fig. 1B. KMKKT treatment decreased the PSA mRNA level starting at 9 hours (Fig. 1C,, right) and in a concentration-dependent manner when examined at 24 hours of exposure (Fig. 1C , right).

The AR mRNA was not decreased during KMKKT treatment from 9 to 96 hours (Fig. 1C). Because AR function is necessary for PSA transcription, we analyzed by immunoblot whether KMKKT decreased AR protein abundance and found that treatment for 24 hours significantly decreased AR protein level (Fig. 1D). These results indicate that the suppressing action of KMKKT on PSA protein level was largely mediated by decreasing PSA mRNA transcript abundance, which was accompanied by decreased AR protein level without changing the level of AR mRNA.

KMKKT inhibits androgen-stimulated AR nuclear translocation. To determine whether KMKKT affects AR nuclear translocation, we grew LNCaP cells in phenol red–free medium supplemented with 5% CSS for 2 days to decrease basal signaling. The cells were pretreated with increasing concentrations of KMKKT extract for 1 hour and were stimulated with mibolerone (1 nmol/L) for 2 hours in the continued presence of KMKKT extract. Nuclear and cytosolic fractions were prepared for immunoblot analyses. As shown in Fig. 1E, PARP and α-tubulin detection confirmed the specificity of nuclear and cytosol preparations. Mibolerone stimulation converted a predominantly cytosolic distribution pattern for AR under androgen-deprived state (Fig. 1E , lanes 1 and 2) to one that was mostly nuclear localized (lanes 3 and 4). KMKKT extract decreased nuclear AR level in a concentration-dependent manner (compare lanes 5 and 7 with lane 3). These results indicate that KMKKT treatment rapidly inhibited androgen-stimulated AR nuclear translocation in addition to the suppression of AR protein abundance.

KMKKT induces expression of NSE and morphologic features of neuroendocrine differentiation. It has been known that deprivation of androgen or a blockage of AR signaling by bicalutamide leads to neuroendocrine differentiation in LNCaP cells and to the induction of molecular markers, such as NSE (12, 15). We therefore examined the mRNA level of NSE and found this neuroendocrine differentiation marker gene significantly increased by as little as 20 μg/mL KMKKT at 24 hours of exposure in the complete medium (Fig. 1C,, right, lanes14-16 versus lane 13) and the NSE induction was sustained through 96 hours (Fig. 1C , left, lane 6 versus lane 5, lane 8 versus lane 7). The NSE expression data provided a molecular clue for the likely induction of neuroendocrine differentiation by KMKKT under the appropriate cell seeding conditions.

To detect the morphologic features of neuroendocrine differentiation, we treated sparsely plated LNCaP cells in 5% whole serum–supplemented medium with a single exposure of KMKKT. By 7 days after KMKKT treatment, ∼70% cells displayed “neurite” morphology (Fig. 2A,, b), whereas only ∼10% of the control cells showed such a feature (Fig. 2A,, a). As a positive control (15), LNCaP cells in phenol red–free medium supplemented with 5% CSS formed neurites in ∼60% of the cells (Fig. 2A,, c), which was effectively reversed by mibolerone (Fig. 2A,, d). Under the experimental conditions described in Fig. 2A  (a and b), a single treatment with KMKKT induced and sustained neuroendocrine differentiation without cell loss for 4 weeks when the experiment was terminated (data not shown).

To further establish the efficacy of KMKKT to specifically block androgen actions in LNCaP cells in phenol red–free medium with CSS, we titrated the mibolerone concentration equivalent of KMKKT using neuroendocrine differentiation and PSA expression as biomarkers (Fig. 2B). In the absence of KMKKT treatment, as little as 0.1 nmol/L mibolerone significantly increased LNCaP cell proliferation as evident from photographs (Fig. 2B,, top, DMSO) and protein yield (Fig. 2C,, control) and also increased PSA secretion (Fig. 2D,, control). Maximal growth and PSA responses were obtained with 0.2 nmol/L mibolerone. In the concentration range of 0.1 to 1 nmol/L, mibolerone significantly decreased the proportion of cells with neurite projections (Fig. 2B,, top). In the presence of 50 μg/mL KMKKT, ≥0.5 nmol/L mibolerone was needed to reverse neuroendocrine differentiation (Fig. 2B,, middle) and to partially restore cell proliferation (Fig. 2C). With 100 μg/mL KMKKT, ≥1 nmol/L concentration of mibolerone was required to override neuroendocrine differentiation (Fig. 2B,, bottom) and to partially restore cell proliferation (Fig. 2C). Both concentrations of KMKKT greatly suppressed the androgen-stimulated PSA secretion into the medium (Fig. 2D). Together, these data supported strong and lasting inhibitory activities on androgen-stimulated cell proliferation and PSA expression and on the androgen suppression of neuroendocrine differentiation by KMKKT.

A. gigas extract contains antiandrogen agents. To identify the active herb(s), we prepared the ethanol extract of each of the 10 herbs and evaluated their effects individually on PSA expression. Mixing all 10 extracts in the proportion used for the formulation of KMKKT reconstituted the PSA-suppressing efficacy (Fig. 3A,, column 3, Mix). When each extract was tested at 20 μg/mL, AGN extract completely suppressed cellular PSA (Fig. 3A,, column 17), whereas A. gelatinum came second potent among the 10 herbs (Fig. 3A,, column 12). The IC₅₀ for AGN extract was estimated to be ∼1 μg/mL (Fig. 3A) and that for A. gelatinum was ∼15 μg/mL (data not shown). Due to its low extraction yield of ∼0.05% by ethanol, the actual contribution of A. gelatinum to the antiandrogen and AR activities of KMKKT could be much less than this differential in IC₅₀ values. Herbs 1-8 (Fig. 3A , columns 4-11) had negligible effect (within 30% of control) at the concentration of 20 μg/mL.

Decursin is a novel antiandrogen and AR compound. Next, we fractionated AGN extract to identify the active compound(s). Because decursin is a known major component of Korean AGN with cytotoxic activity in leukemia cells (17, 18), we focused on this phytochemical (structure shown in Fig. 3D) as a likely candidate. As shown in Fig. 3B, TLC analyses showed the presence of a small amount of decursin in KMKKT (lane 2). HPLC analysis estimated this content to be ∼1.7% (Supplementary Fig. S1). The decursin content was enriched in the methylene chloride (lane 5) and ethyl acetate (lane 6) fractions in comparison with the starting AGN extract (lane 3), whereas its content was decreased in the butanol fraction (lane 7). The n-hexane, methylene chloride, and ethyl acetate fractions were more potent than the AGN extract for suppressing cellular PSA, whereas the butanol fraction was less efficacious (Fig. 3C). Similarly, these fractions showed the same order of potency in suppressing LNCaP cell viability/growth after 24-hour exposure (Fig. 3C). The decursin content in general correlated and predicted the potency of each solvent fraction in suppressing PSA and cell growth (Fig. 3C).

Indeed, exposure of LNCaP cells to purified decursin (purity, ∼98.6%) for 48 hours in complete medium decreased cellular and secreted PSA with IC₅₀ of ∼0.4 μg/mL (1.3 μmol/L; Fig. 4A). Synthetic crystalline decursin (18) generously provided by Prof. Hogyu Han produced the same potency and pattern of inhibition (data not shown). Exposure for 24 hours to decursin (1.6, 3.3, and 6.6 μg/mL) and AGN extract (5, 10, and 20 μg/mL) recapitulated in a concentration-dependent manner the effects of KMKKT on PSA mRNA and protein abundance (Fig. 4B). Like KMKKT, decursin or AGN extract did not affect the AR mRNA level (Fig. 4B) but decreased AR protein abundance (Fig. 4B). In addition, the AGN extract and decursin induced potent G₁ arrest in a concentration-dependent manner as did KMKKT (Table 1). Furthermore, treatment with decursin or AGN extract blocked the androgen-stimulated translocation of AR to the nucleus as did KMKKT (Fig. 4C , compare lanes 5, 7, and 9 with lane 3). These results suggested that decursin in neat form or in herbal extracts inhibited AR translocation into the nucleus and decreased its protein abundance, preventing AR from activating PSA mRNA transcription and protein expression.

To verify this prediction of transcriptional inhibition by decursin, we evaluated the PSA promoter-luciferase activity in a transient transfection assay as described previously (25). As shown in Fig. 4D, decursin treatment for 24 hours inhibited androgen-stimulated PSA promoter transcription in a concentration-dependent manner, with IC₅₀ of ∼1.6 μg/mL (5 μmol/L).

Furthermore, decursin and AGN extract recapitulated the effect of KMKKT in the neuroendocrine differentiation assay (Fig. 5, photographed 9 days after seeding). In the absence of androgen stimulation (a-f), decursin (e and f), AGN extract (c and d), and KMKKT extract (b) decreased the basal PSA secretion by 4-fold (from 12 to 3 ng/mL). In the presence of androgen (g-l), decursin (k and l) and AGN extract (i and j) inhibited the androgen-stimulated reversal of neuroendocrine differentiation and androgen-stimulated cell proliferation as well as inhibited PSA secretion in a concentration-dependent manner, achieving a complete block on these variables with 3.3 μg/mL (10 μmol/L) of decursin and 10 μg/mL AGN extract, respectively. The effect of a single treatment lasted through at least 17 days when the experiment was terminated (data not shown), indicating the long-lasting androgen suppression action of decursin and AGN extract. These data unequivocally supported decursin as an antiandrogen and AR compound in AGN and KMKKT extracts.

PSA suppression persists after removal of decursin or herbal extracts. To further confirm the long-lasting action of decursin on androgen and AR signaling, we exposed LNCaP cells for 3 days in complete medium with either decursin, AGN, or KMKKT extract and then carefully removed the conditioned medium and washed the cells twice with serum-free medium. The cells were then fed fresh complete medium for 1, 2, or 3 days. At each time point, the conditioned medium was harvested and the cells were lysed for PSA ELISA measurement to determine the kinetics of recovery. As shown in Fig. 6A, (left), the decursin-, KMKKT-, and AGN-exposed cells showed minimal recovery of cellular PSA level (normalized on protein content) at 24 hours after removal. By 3 days after removal, a partial recovery of <40% and 30% was detected for cells exposed to decursin and the two extracts, respectively. Immunoblot analysis of cellular PSA in decursin-treated cells in a separate experiment confirmed this result (Fig. 6B). Similar patterns of persistent suppression of PSA secreted into the medium by decursin and KMKKT or AGN extract were observed (Fig. 6A , right, and B).

To determine whether the suppression of PSA expression after the removal of decursin was due to a persistent decrease in AR abundance, we immunoblotted AR and PSA in the cell lysate (Fig. 6B). Decursin-treated cells maintained a profound suppression of cellular PSA throughout 72 hours, whereas the AR protein rebounded to control level within 24 hours of removal of decursin. These results suggested that decursin exerted a sustained inhibitory action on androgen and AR signaling even after AR protein abundance had fully recovered.

Comparison with the effects of bicalutamide. To further probe the modes of action of decursin on AR signaling, we compared its effects in the presence and absence of androgen stimulation with those of bicalutamide, which is a clinically indicated androgen-binding antagonist (26). LNCaP cells were cultured in 5% CSS medium for 48 hours and exposed to DMSO or indicated concentrations of decursin or bicalutamide for 1 hour before the addition of mibolerone for 24 hours. As shown in Fig. 6C, decursin decreased PSA mRNA and protein as well as AR protein abundance in the absence (lanes/columns 2 and 3 versus lane/column 1) and presence (lanes/columns 7 and 8 versus lane/column 6) of mibolerone without affecting the AR mRNA level. Bicalutamide did not significantly change AR mRNA level either (lanes 4 and 5 versus lane 1 in the absence of mibolerone and lanes 9 and 10 versus lane 6 in the presence of mibolerone). In contrast to decursin, bicalutamide increased the basal levels of PSA mRNA and cellular and secreted PSA as well as AR abundance in the absence of androgen (lanes/columns 4 and 5 versus lane/column 1), acting as a partial AR-binding “agonist.” As expected of its androgen-binding blocker action, bicalutamide decreased PSA mRNA, protein, and secretion in the presence of mibolerone in a concentration-dependent manner (lanes/columns 9 and 10 versus lane/column 6). With respect to LNCaP survival, decursin was more potent than bicalutamide at a same molar concentration of exposure (40 μmol/L) to induce apoptosis as indicated by the cleaved PARP (lane 3 versus lane 5, lane 8 versus lane 10). These results supported distinct novel mechanisms by which decursin inhibited androgen and AR signaling in comparison with bicalutamide.

The work communicated here represents a beginning of collaborative efforts to identify novel phytochemicals that target androgen and AR signaling from Oriental medicinal herbs. Identification of the active compound(s) is essential to understanding the mechanisms of action and for developing clinically useful agents for the prevention and treatment of prostate cancer and other androgen-dependent diseases, such as benign prostatic hyperplasia (BPH; ref. 27) and male baldness (28). Starting from the ethanol extract of the KMKKT formula, we discovered potent antiandrogen and AR signaling activities, including the suppression of PSA expression at both protein and mRNA levels (Fig. 1B and C) and the induction of neuroendocrine differentiation (Fig. 2) using the androgen-dependent LNCaP model. We found that AGN was primarily responsible for the antiandrogen and AR activities (Fig. 3A). Solvent fractionation and TLC data (Fig. 3B) confirmed decursin as a major phytochemical of AGN extract (16, 17) and supported a direct correlation of decursin content with the potency of the various solvent fractions to inhibit PSA and cell growth (Fig. 3C). Treatment with purified decursin recapitulated all the effects of KMKKT and AGN that were examined, including decreasing PSA protein abundance and secretion (Fig. 4A and B) and PSA mRNA level (Fig. 4B), decreasing AR protein level without affecting its mRNA abundance (Fig. 4B), inhibiting androgen-stimulated AR translocation to the nucleus (Fig. 4C) and androgen-stimulated PSA promoter transcription (Fig. 4D), suppressing androgen-stimulated cell proliferation and overriding the ability of androgen to block neuroendocrine differentiation (Fig. 5), and inducing a profound G₁ arrest (Table 1). These results unequivocally supported decursin as a potent antiandrogen and AR compound in AGN and KMKKT extracts but did not exclude the possibility of additional active antiandrogen and AR compounds. For example, we have found recently that decursinol angelate, an isomer of decursin, possesses a similar potency for inhibiting PSA expression. Efforts are being continued to identify the structural determinants of the antiandrogen and AR activities of decursin analogues and derivatives.

Our data showed that the antiandrogen and AR activities were not a consequence of apoptosis induced by decursin or by KMKKT extract. In our experiments, decursin induced apoptosis (Fig. 6C; Supplementary Fig. S2) at exposure concentrations of ≥40 μmol/L (13.2 μg/mL). This is consistent with a recent report of apoptosis induction in LNCaP cells by ≥50 μmol/L (16.5 μg/mL) decursin (20). KMKKT induced apoptosis at ≥200 μg/mL concentrations (Fig. 1A). These apoptotic levels of decursin or KMKKT were much higher than the IC₅₀ for PSA suppression, 0.4 μg/mL decursin (Fig. 4A) and 7 μg/mL KMKKT (Fig. 1B), respectively. The antiandrogen and AR activities were observed in concentration ranges that caused G₁ arrest (Table 1) and were likely a primary cause for the cell cycle arrest effect. This was further supported by the ability of increasing mibolerone doses to partially restore cell growth in the presence of KMKKT (Fig. 2B and C). Because decursin has been reported to induce G₁ arrest not only in LNCaP cells but also in androgen-independent DU145 and PC-3 cells (20), decursin could therefore exert antiproliferative effects through both androgen/AR–dependent and androgen/AR–independent mechanisms, the details of which remain to be elucidated. The antiandrogen and AR activities could be predicted to make decursin more active against androgen-dependent prostate cancer cells than androgen-independent cells for growth inhibition and apoptosis. Consistent with this, we found (Supplementary Fig. S2) that PC-3 cells were less sensitive than LNCaP cells for decursin-induced apoptosis, in agreement with a recent report (20).

Our results reveal at least two possible mechanisms through which decursin may exert the antiandrogen and AR activities: inhibiting androgen-stimulated AR nuclear translocation and decreasing AR protein abundance. With respect to the first mechanism, it remains to be determined whether decursin competes with androgen for binding to AR to prevent it from translocating to the nucleus to activate transcription of AR-responsive genes, such as PSA (Fig. 4D). A comparison with bicalutamide has revealed significant differences between their actions, particularly under androgen-free conditions (Fig. 6C). That is, bicalutamide increased (instead of inhibited) basal AR protein and PSA abundance and secretion in contrast to decursin, which suppressed these variables. In addition, decursin was more potent than a same molar concentration of bicalutamide for induction of apoptosis (Fig. 6C). These results support decursin as a novel and potent antiandrogen and AR agent with several aspects of mechanisms of action distinct from bicalutamide. With respect to the second mechanism, because AR protein is down-regulated by decursin, AGN, and KMKKT extracts, but the AR mRNA level was not affected (Figs. 1B, 4B, and 6C), we are currently investigating whether AR protein synthesis and/or degradation are targeted by decursin. Furthermore, decursin has been reported as an activator of protein kinase C (PKC) in test tube assays and to inhibit leukemia differentiation induced by some tumor-promoting PKC activators (17, 18). Because PKC activators, such as 12-O-tetradecanoylphorbol-13-acetate and phorbol-12-myristate-13-acetate, have been reported to suppress PSA (29, 30) and induce apoptosis in LNCaP cells (30, 31), whether the antiandrogen and AR activities of decursin in LNCaP cells are mediated by PKC pathway should be examined in the future.

One salient feature of the antiandrogen and AR activities of decursin or AGN and KMKKT extracts is the long-lasting action. The PSA-suppressing effect of a single treatment by KMKKT was initially manifested at 9 hours of exposure with the lowering of PSA mRNA (Fig. 1C) and then of cellular PSA protein at 12 hours of exposure (Fig. 1B) and persisted for 96 hours without any indication of recovery (Fig. 1B and C). Similarly and more strikingly, we observed that neuroendocrine differentiation persisted for 2 to 4 weeks with a single treatment with decursin or the herbal extracts (Figs. 2 and 5). The long-lasting action was further evident when the treatments were removed with only minor partial recovery of cellular PSA protein 3 days afterward (Fig. 6A and B). The lasting action after decursin removal did not result from a continually suppressed AR protein abundance because it rebounded to control level 24 hours after decursin was removed (Fig. 6B). We are currently investigating whether this persistent suppression of PSA might be due to a sustained blockage of AR nuclear translocation or an epigenetic inactivation of the transcribability of the PSA promoter. If translatable pharmacologically to in vivo, the long-lasting action of decursin would have a significant advantage in terms of the dosing frequency to maintain an effective suppression of androgen and AR activity.

The identification of decursin as a novel antiandrogen and AR signaling compound at low concentration levels (<10 μmol/L) provides specific mechanistic rationales for its potential use in the chemoprevention of androgen-dependent prostate cancer and for the treatment of BPH (27) and male baldness (28). This discovery and the recent report (20) of induction by decursin of G₁ arrest in DU145 and PC-3 AR-negative and androgen-independent prostate cancer cells at 25 μmol/L (8.2 μg/mL) and apoptosis at 100 μmol/L (32.8 μg/mL) suggest that decursin could also be useful for the chemoprevention or treatment of hormone-refractory prostate cancer. Regarding the latter, recent studies have shown that even in hormone-refractory prostate cancer AR is mostly present in wild-type status and that ligand-independent AR signaling is important for the growth and survival of these prostate cancer cells (2, 32, 33). The ability for decursin to suppress AR abundance and signaling in the absence of androgen (Fig. 6C) could therefore be important for the management and prevention of hormone-refractory prostate cancer in combination with hormone ablation therapies. Animal models of prostate cancer chemoprevention and therapy are clearly warranted to establish the efficacy of decursin and herbal extracts before any translational research into human populations is contemplated.

Grant support: The Hormel Foundation, Eagles' Telethon, National Cancer Institute grant CA95642 (J. Lu), and Korean Biogreen21 Project grant 1000520030030000-1, Korean Ministry of Health and Welfare grant 01-PJ9-PG1-01CO05-0004, and Interdisciplinary Research and SRC Program of Korea Science and Engineering Foundation (S-H. Kim).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Nam-In Baek for authenticating the herbs used in this study, Prof. Hogyu Han for the crystalline decursin sample, Dr. Charles Young for the generous gifts of PSA promoter, CMV-lacZ plasmids, and mibolerone, Dr. Hongbo Hu for help with apoptosis ELISA assay, Todd Schuster for flow cytometry analyses, and Andria Hanson for secretarial support.