Antiandrogenic Effects of Novel Androgen Synthesis Inhibitors on Hormone-dependent Prostate Cancer¹

Prostatic carcinoma is the most commonly diagnosed malignancy in men in the United States and is second only to lung cancer in cancer-related deaths (1, 2). Androgens play an important role in controlling the growth of the normal prostate gland, and in promoting BPH⁴and prostatic carcinoma (3). The two most important androgens in prostate cancer etiology are testosterone and DHT. Testosterone is synthesized primarily in the testes and also, to a lesser extent, in the adrenals. Testosterone is further converted to the more potent androgen DHT by the enzyme 5α-reductase, which is localized primarily in the prostate (4). Although testosterone and DHT both stimulate the growth of normal and malignant prostate tissue, DHT is believed to be the more important androgen(5, 6).

Androgen ablation therapy has been shown to produce the most beneficial responses in patients with hormone-responsive prostatic tumors. Orchidectomy (castration, either surgical or medical with a luteinizing hormone-releasing hormone analogue) remains the standard treatment option for most patients. However, both methods, which result in reduced androgen production by the testes, fail to alter androgen production by the adrenal glands. Studies in the United States and Europe have reported that although androgen ablation therapy alone is an effective treatment, a combination therapy of orchidectomy with antiandrogens, to inhibit the action of adrenal androgens,significantly prolongs the survival of prostate cancer patients(7, 8, 9). Given that efforts to block the production or effects of adrenal androgens result in worthwhile therapeutic gains,this laboratory has been designing and evaluating novel compounds that inhibit androgen production from all sites in the body. The compounds developed are steroidal antagonists of the steroidogenic enzymes C_17,20-lyase and 5α-reductase. C_17,20-lyase catalyzes both the 17α-hydroxylation and the cleavage of the C_17,20-side chain during the conversion of the 21-carbon steroids pregnenolone and progesterone to the 19-carbon androgens dehydroepiandrosterone and androstenedione, respectively(10). The enzyme has an identical amino acid sequence in both testicular and adrenal tissue (11), indicating that inhibitors of this enzyme would be equally effective at both sites. Two isoforms of 5α-reductase occur in the body (type I and type II), and the type II enzyme is the predominant form in the human prostate. Inhibitors of this isoenzyme would essentially prevent prostatic accumulation of the potent androgen DHT.

A number of C_17,20-lyase-inhibitors have been described. However, most are not specific, and only the imidazole antifungal agent ketoconazole has been used clinically to reduce testosterone levels in patients with advanced prostate cancer(12, 13, 14). The major drawback with ketoconazole is that it is not very potent or specific. It is only a moderate inhibitor of C_17,20-lyase, inhibits cortisol production, and has a number of significant side effects. Nevertheless, recent studies have reported that ketoconazole was effective in reducing PSA levels in 55% (15) and 62.5% (16) of patients who had progressed after antiandrogen (flutamide) withdrawal. These results indicate that compounds more specific and selective than ketoconazole may be more effective for the treatment of prostate cancer. Several inhibitors of 5α-reductase have also been described and finasteride,which is a more potent inhibitor of the type II than the type I isoenzyme (17), has been approved for the treatment of BPH(18). Although effective at reducing DHT levels in patients with prostate cancer, finasteride also increases the bioavailable levels of testosterone (19), which can stimulate tumor growth (20). Therefore, compounds designed to inhibit both C_17,20-lyase and 5α-reductase would be expected to be more clinically beneficial to prostate cancer patients than compounds which inhibit only one of the enzymes.

The growth effects of testosterone and DHT on prostate cancer cells are mediated by the androgens binding to their cognitive nuclear receptors,which in turn bind to specific response elements in the promoter regions of androgen-regulated genes. The products of these genes modulate cellular proliferation (21). Antiandrogens such as flutamide have also been used clinically for the treatment of prostate cancer. However, the results were disappointing. Some patients tended to improve after flutamide was withdrawn after relapse(22, 23). Nonetheless, as described above, clinical trials which combined the therapies of orchidectomy with the antiandrogen flutamide reported a significantly longer survival period than orchidectomy alone (7, 8, 9). This implies that a treatment regimen involving total androgen ablation in combination with an antiandrogen may be a more effective treatment option for patients with androgen-dependent prostate tumors.

Previously we have reported the synthesis and testing of several steroidal inhibitors of C_17,20-lyase and 5α-reductase (24, 25, 26, 27, 28). These compounds were shown to be effective inhibitors of human testicular C_17,20-lyase and prostatic 5α-reductase in vitro (See Table 1). In the present study we report that some of these compounds are also very potent antiandrogens, and that this property contributes,at least in part, to their growth-inhibitory effects on androgen-dependent LNCaP prostate cancer cells in vitro and in vivo. LNCaP cells are the most frequently studied AR-positive prostate cancer cell line that can be readily grown in tissue culture (29, 30, 31). LNCaP cells are not only androgen-responsive but also androgen-dependent. They respond to androgens with increased cellular proliferation and elevated expression of PSA (32). Although LNCaP cells have a mutated AR(33, 34), they have been used extensively for research on the causes, treatment, and prevention of prostate cancer(35). We have used these cells in culture and as tumors in nude mice to compare the efficacy of our compounds with the known inhibitors of C_17,20-lyase and 5α-reductase,ketoconazole, and finasteride, respectively. The antiandrogenic properties of the compounds were compared with flutamide.

The wild-type LNCaP human prostate cancer cell line and CV-1 green monkey kidney cells were obtained from the American Type Culture Collection (Rockville, MD). PC-3 prostate cancer cells and PC-3 cells stably transfected with the human wild-type androgen receptor(designated PC-3AR) were kindly provided by Dr. Marco Marcelli (Baylor College of Medicine, Houston, TX; Refs. 36  and 37). RPMI 1640 medium, penicillin/streptomycin solution, trypsin/EDTA solution,G418, Opti-MEM, Lipofectamine, and Lipofectamine-Plus were from Life Technologies, Inc. (Grand Island, NY). Phenol red-free IMEM and phenol red-free trypsin/EDTA were from Biofluids Inc. (Rockville, MD). FBS was supplied by Summit Biotechnology, Inc. (Fort Collins, CO), and the PSA ELISA kit was from DSL, Inc. (Webster, TX). Hydroxypropyl cellulose was from Sigma Chemical Co. (St. Louis, MO). The luciferase reporter construct pMAMneoLUC was purchased from Clontech(Palo Alto, CA). The luciferase assay kit was from Promega (Wisconsin,MI). Wild-type human AR-expressing vector pCMV5-hAR and the LNCaP AR-expressing vector pCMV5-LNCaPAR were generous gifts from Dr. Elizabeth Wilson (University of North Carolina, Chapel Hill,NC). Matrigel was kindly supplied by Dr. Hynda Kleinman (NIH,Bethesda, MD).

Testosterone, DHT, flutamide, ketoconazole, and the antiprogestin triamcinolone acetonitride were purchased from Sigma Chemical Co. (St. Louis, MO). Finasteride was a gift from Merck Research Laboratories(Rahway, NJ). L-2, L-10, L-35, L-36, L-37, and L-39 were synthesized in our laboratory according to the procedures described by Ling et al. (27). [³H]R1881(specific activity 70–87 Ci/mMol) was obtained from DuPont NEN(Boston, MA). Testosterone RIA kits were purchased from DSL, Inc.

LNCaP, PC-3, and CV-1 cells were grown in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin solution. To determine the effect of steroids and novel compounds on cell proliferation, hormone-dependent LNCaP and hormone-independent PC-3 cells were transferred into steroid-free medium 3 days prior to the start of experiments. Steroid-free medium consisted of phenol red-free IMEM supplemented with 5% dextran-coated, charcoal-treated serum, and 1% penicillin/streptomycin solution. Serum was depleted of steroids as described previously (38). Growth studies were then performed by plating cells (1.5 × 10⁴) in 24-place multiwell dishes (Corning, Inc.,Corning, NY). After a 24-h attachment period, the medium was aspirated and replaced with steroid-free medium containing vehicle or the indicated concentrations of androgens and novel compounds. This medium was changed every 3 days and the cells were counted 9 days later using a Coulter Counter model Z-1 (Coulter Electronics, Hialeah, FL). Cell numbers were expressed as a percentage of vehicle-treated cells.

Wild-type LNCaP cells and CV-1 cells were transfected with the pMAMneoLUC plasmid as we have described previously(39). Briefly, 2 × 10⁵ cells in a 35-mm² dish were exposed to 3 ml of Opti-MEM containing Lipofectamine (30 μl) and 6 μg of the pMAMneoLUC plasmid for 5 h at 37°C in a 5% CO₂ incubator. The medium was then changed to routine culture medium for 72 h. Cells were then grown in medium supplemented with 750 μg/ml G418. The surviving colonies were picked and grown in selective media. Stable selectants were tested twice per month for luciferase activity, as described in the later section. After 3 months (∼23 passages), the transfectants with the highest luciferase activity were selected and designated LNCaP-LUC and CV-1LUC,respectively. These cell lines and the PC-3AR cell line were routinely cultured in RPMI 1640 medium supplemented with 10% FBS, 1%penicillin/streptomycin solution, and 750 μg/ml G418.

Competitive binding studies with the synthetic androgen R1881 were performed essentially as described by Wong et al.(40) and Yarbrough et al. (41). Wells in 24-place multiwell dishes were coated with poly-l-lysine (0.05 mg/ml) for 30 min and dried. To determine the kinetics of R1881-binding to the LNCaP AR and the wild-type AR, LNCaP cells and PC-3AR cells (2–3 × 10⁵) were plated in steroid-free medium and allowed to attach. The following day, the medium was replaced with serum-free, steroid-free IMEM supplemented with 0.1% BSA and containing [³H]R1881 (0.01–10 nm) in the presence or absence of a 200-fold excess of cold R1881 to determine nonspecific binding and 1μ m triamcinolone acetonitride to saturate progesterone and glucocorticoid receptors. Following a 2-h incubation period at 37°C, cells were washed twice with ice-cold DPBS and solubilized in DPBS containing 0.5% SDS and 20% glycerol. Extracts were removed and the cell-associated radioactivity counted in a scintillation counter. The data were analyzed using RADLIG 40 software(Biosoft, Ferguson, MO), and K_d and B_max determined by Scatchard plot transformation. When the concentration of R1881 required to almost saturate AR in both cell lines was established, the ability of the test compounds (5.0 μm) to displace[³H]R1881 (5.0 nm) from the receptors was determined as described above.

Cells were grown in steroid-free IMEM for 3 days and plated(4 × 10⁴ cells/well) in 24-well plates in phenol red-free IMEM supplemented with 10% charcoal-stripped serum and no antibiotics. After a 24-h incubation period, the cells were washed twice with DPBS and each well was incubated with 250 μl of phenol red-free IMEM containing 2 μl of PLUS-reagent, 4 μl of Lipofectamine, and 4 μg of the pMAMneoLUC plasmid for PC-3AR cells and 0.5 μg of the pCMV5-hAR or pCMV5-LNCaPAR plasmids for CV-1LUC cells. After a 5-h incubation period, 250 μl of routine medium were added to each well and the cells were incubated for an additional 24 h. The resultant cells, designated PC-3AR/LUC,CV-1LUC/hAR, or CV-1LUC/LNCaPAR, were assayed for luciferase activity as described in the following section.

LNCaP-LUC cells were transferred to steroid-free medium 3 days before the start of the experiment and plated at 1 × 10⁵ cells/well in steroid-free medium. PC-3AR/LUC cells,CV-1LUC/hAR cells, and CV-1LUC/LNCaPAR cells were transiently transfected as described above. After a 24-h incubation period in steroid-free medium, each well was treated with ethanol vehicle or the selected steroids and novel compounds (5 μm) in triplicate. After a 24-h treatment period, the cells were washed twice with ice-cold DPBS and assayed using the Luciferase kit according to the manufacturer’s protocol. Briefly, the cells were lysed with 200μl of luciferase lysing buffer, collected in a microcentrifuge tube and pelleted by centrifugation. Supernatants (100 μl aliquots) were transferred to the corresponding wells of white 96-well plates(Polyfiltronics, Inc., Boston, MA). Luciferin (50 μl) was added to each well, and the light produced during the luciferase reaction was measured in a Victor 1420 Multilabel counter (Wallac,Inc., Gaithersburg, MD). The effects of the steroids and novel compounds on DHT-induced luciferase transcription were determined using the same protocol.

Male SCID mice and athymic nude mice 4–6 weeks of age were purchased from the National Cancer Institute (Frederick, MD). Animals were housed in a pathogen-free environment under controlled conditions of light and humidity and received food and water ad libitum.

LNCaP tumors were grown s.c. in male SCID mice essentially as described by Sato et al. (42) with modifications based on the breast cancer model described by Yue et al. (43, 44). LNCaP cells were grown in routine culture medium (RPMI 1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin)until 80% confluent. Cells were scraped into DPBS, collected by centrifugation, and resuspended in Matrigel (10 mg/ml) at 3 × 10⁷ cells/ml. Each mouse received s.c. injections at one site on each flank with 100 μl of cell suspension. Tumors were measured weekly with calipers, and tumor volumes were calculated by the formula 0.5236 × r₁² × r₂(r₁ < r₂).

Hormone-dependent PC-82 tumor xenografts were kindly provided by Dr. John Isaacs (John Hopkins School of Medicine, Baltimore, MD). PC-82 tumors were minced into fine pieces, washed with DPBS, filtered, and resuspended in Matrigel at a concentration of 100 mg/ml The mice were inoculated s.c. at one site on each flank with 100 μl of tumor suspension using a 18-gauge needle. Tumors were allowed to grow (∼3 months) and then were measured weekly. Tumor volumes were then calculated weekly according to the formula 0.5236 × r₁² × r₂ (r₁ < r₂). When tumors reached a measurable size, the mice were randomized into treatment groups (five animals/group) and treated as described in the following section.

For LNCaP tumor experiments, treatments began 4–5 weeks after cell inoculation when measurable tumor volume was 500 mm³ (For PC-82 xenografts, this was ∼3 months). For each experiment, groups of six mice (five mice for PC-82 xenografts) with comparable total tumor volumes were either castrated or treated with the novel compounds at 50 mg/kg/day. Mice were castrated under methoxyfluorane anesthesia via the abdominal approach. Compounds and reference drugs were prepared at 10 mg/ml in a 0.3%solution of hydroxypropyl cellulose in saline, and mice received s.c. injections daily. Control and castrated mice were treated with vehicle only. Treatments lasted for 28 days, after which time the animals were sacrificed by decapitation and the blood was collected. Tumors were excised, weighed, and stored in liquid nitrogen for additional analysis.

For measurement of serum testosterone levels, 50 μl of mouse serum were assayed according to the instructions provided with the ¹²⁵I-testosterone RIA kit supplied by DSL, Inc. Radioactivity was counted using a Packard Cobra II gamma counter. For measurement of tumor testosterone levels, whole tumors were homogenized in phosphate buffer (pH 7.4; 0.1 m). The homogenates were then centrifuged at 2000 × g for 20 min to remove debris. Fifty-μl aliquots of the tissue supernatant were used to determine the tumor testosterone concentration, as described above.

Serum PSA levels were determined using a PSA ELISA kit supplied by DSL,Inc. Briefly, 2.5 μl of serum diluted 1:10 in DPBS was mixed with assay buffer to a final volume of 75 μl and added to duplicate wells in the 96-well plate that had been coated with an anti-PSA antibody. Following a 1-h assay and extensive washing of the plate, the wells were treated for 30 min with a second anti-PSA antibody labeled with horseradish peroxidase. After washing, the wells were treated with the tetramethylbenzidine substrate for 10 min, and the absorbance was read at 450 nm with a Dynatech MRX plate reader.

One-way ANOVA on SigmaStat for Windows version 1.0 was used to compare the different treatment groups at the 95% confidence level. A P of <0.05 was considered to be statistically significant.

The structures of the compounds are shown in Fig. 1, and their in vitro activities against C_17,20-lyase and 5α-reductase are provided in Table 1. For growth studies, androgen-responsive LNCaP cells and androgen-independent PC-3 cells were deprived of steroids for 3 days before the experiments. After plating and attachment, cells were treated with the compounds at concentrations of 0.1μ m, 1.0 μm, and 5.0μ m for 9 days (Fig. 2,A and 2,B). In agreement with our previous findings (45, 46), DHT and testosterone (both at 1 nm)were potent stimulators of LNCaP cell growth in vitro, and increased cell number by 5.9-fold and 4.4-fold, respectively (Fig. 2,A). Neither testosterone nor DHT had any effect on the growth of hormone independent PC-3 prostate cancer cells (Fig. 2,B). At each of the concentrations tested, the antiandrogen flutamide also increased the growth rate of LNCaP cells, whereas ketoconazole had no effect on cellular proliferation. Finasteride (1μ m and 5 μm) inhibited growth by 60% compared with vehicle-treated cells. L-37 was the most potent inhibitor of cell growth and did so in a dose-dependent manner. Names of the “L” compounds are shown in Fig. 1. L-37 inhibited cell growth by 34% at 0.1 μm, 54%at 1.0 μm, and 67% at 5μ m. L-35 inhibited cell growth by 40% at both 1-μm and 5-μmconcentrations. L-10 increased cell number by approximately 2-fold at each of the concentrations tested. L-2 was found to be the most potent stimulator of cell growth and 0.1 μm, 1.0μ m, and 5.0 μm L-2 increased cell number by 4.6-fold, 4.4-fold, and 2.9-fold,respectively. L-36 showed a biphasic effect on cell growth. The lower concentrations (0.1 μm and 1.0μ m) increased the growth rate of LNCaP cells by 2.3-fold and 2-fold, respectively. However, at a 5-μm concentration, L-36 inhibited cell growth by 50%. L-39 was mildly growth-stimulatory to LNCaP cells at the two lower concentrations, but at the higher concentration of 5.0μ m, L-39 had little effect on the growth rate. When the compounds were tested against androgen-independent PC-3 cells over the same 9-day treatment period, the only compounds that modulated growth were L-37 and L-39, which at the 5.0-μmconcentration inhibited the growth of PC-3 cells by approximately 30%(Fig. 2 B).

To determine whether the mechanism of action of the compounds was mediated by the AR, the ability of the compounds to compete with DHT and modulate cell proliferation was investigated (Fig. 3 A). Flutamide and ketoconazole had no effect on DHT-induced LNCaP cell proliferation. Surprisingly, the 5α-reductase inhibitor finasteride reversed the growth stimulatory effect of DHT on LNCaP cells in a dose-dependent manner. L-2 and L-10 had no effect on DHT-induced LNCaP cell proliferation at concentrations of 0.1μ m and 1.0 μm. At a concentration of 5.0 μm, both compounds slightly reduced proliferation to levels similar to those when the compounds were tested in the absence of DHT. No synergistic or additive effects were observed when LNCaP cells were cotreated with the growth stimulatory compounds L-2 or L-10 and DHT. L-36 had no effect on DHT-induced LNCaP cell proliferation at concentrations of 0.1μ m and 1.0 μm. However at 5.0 μm concentration, L-36 inhibited cell proliferation by >90%. L-37 was shown to be the most potent inhibitor of DHT-induced LNCaP cell proliferation. L-37 inhibited DHTinduced growth by 73%, 88%, and 95% at concentrations of 0.1μ m, 1.0 μm, and 5.0 μm, respectively. L-35 and L-39 were also effective at reversing the growth effects of DHT and did so in a dose-dependent manner. At the 5.0 μmconcentration, L-35 and L-39 inhibited DHT-induced LNCaP cell proliferation by 66% and 79%, respectively. These results suggest that the compounds may be acting to block the action of DHT in stimulating cell proliferation.

The ability of the novel compounds to compete with testosterone for the AR and modulate cell growth in vitro was also investigated(Fig. 3 B). When cells were cotreated with testosterone (1 nm) and the reference drugs or novel compounds for 9 days, the results were similar to those obtained when cell growth was stimulated by DHT. L-37, L-35, and L-39 again proved to be the most effective compounds at inhibiting testosterone-induced growth of LNCaP cells in vitro. Each of these compounds inhibited growth in a dose-dependent manner.

The ability of the novel compounds to modulate the growth of LNCaP cells and to inhibit DHT and testosterone-induced cell proliferation implied that they may be interacting with the LNCaP AR. To determine whether this was specific for the mutated AR expressed in LNCaP cells,kinetic studies were performed with the synthetic androgen methyltrienolone [³H]R1881 binding to the mutated AR expressed in LNCaP cells and the wild-type AR expressed in PC-3AR cells. The data were linearized by Scatchard transformation(data not shown) and the K_d and B_max for both types of AR were determined. LNCaP cells express a single class of high affinity binding sites with K_d = 0.78 ± 0.01 nm and B_max = 2.60 × 10⁵ ± 2.27 × 10⁴ receptors/cell. PC-3AR cells also express a single class of high-affinity binding sites with K_d = 0.20 ± 0.01 nm and B_max = 4.8 × 10⁴ ± 6.5 × 10³ receptors/cell. The ability of the compounds to compete with 5 nm[³H]R1881 for binding to both types of AR was determined (Fig. 4). Ketoconazole was ineffective at preventing[³H]R1881 from binding to the AR in either cell line. The antiandrogen flutamide prevented 52% of the[³H]R1881 from binding to the LNCaP AR, and 51% of the labeled androgen from binding to the wild-type AR in PC-3AR cells. Interestingly, L-35, which was one of the most effective compounds at inhibiting the growth of LNCaP cells in vitro,was the least effective compound at preventing[³H]R1881from binding to either type of AR. In LNCaP cell cultures, L-35 prevented 34% of the[³H]R1881 from binding to the AR, and in PC-3AR cells, the inhibition was 31%. Each of the other 5 novel compounds were very effective at preventing [³H]R1881 from binding to the cellular AR. The pattern was similar regardless of whether they were tested against the mutated LNCaP AR or the wild-type AR. Moreover, they prevented [³H]R1881from binding to the ARs with higher efficiencies than flutamide. In the presence of L-2, L-10, L-36, L-37, and L-39, the amount of[³H]R1881 bound to the LNCaP cells was 30%,5.2%, 0.7%, 15.2%, and 3.3%, respectively. In PC-3AR cells, the amount of [³H]R1881 bound in the presence of L-2, L-10, L-36, L-37, and L-39 was 13.5%, 2.6%, 5.1%, 11.7%, and 2.5%, respectively.

To determine whether the compounds are acting as AR agonists or antagonists, luciferase activity assays were performed using LNCaP-LUC and PC-3AR/LUC cells. DHT (5.0 nm) significantly(P < 0 0.001) stimulated luciferase production mediated by the LNCaP AR in LNCaP-LUC cells (Fig. 5,A). Ketoconazole (5.0 μm) and flutamide (5.0 μm) also activated transcription of the luciferase gene in LNCaP-LUC cells by approximately 2-fold each. Finasteride had no effect on luciferase activity. L-2, L-10, and L-36(all at 5 μm) stimulated luciferase production in LNCaP-LUC cells by 6-fold, 4.5-fold, and 8-fold, respectively,confirming that each of these compounds are agonists of the mutated LNCaP AR. L-39 also, surprisingly, stimulated luciferase activity in LNCaP-LUC cells by 2.5-fold, which suggested that it also has some agonistic properties. L-35 and L-37 (5 μm)significantly (P < 0.001) lowered luciferase transcriptional activity in LNCaP-LUC cells to barely undetectable levels, indicating that they are potent antagonists of the LNCaP AR. These studies were also performed using PC-3AR/LUC cells, which express wild-type AR (Fig. 5 B). In these cells, DHT (5 nm) significantly stimulated(P < 0.01) transcription of the luciferase gene by 3.3-fold. This was considerably less than with LNCaP-LUC cells,and was attributed to the cells reacting negatively to the double transfection of foreign genes. Neither ketoconazole nor flutamide had any appreciable effects on luciferase activity levels mediated by the wild-type AR, suggesting that they are only agonistic with the mutated AR. Finasteride (5 μm) did not modulate wild-type AR-mediated luciferase activity levels. When the novel compounds were tested, L-2, L-10, and L-36 (all at 5μ m) significantly stimulated(P < 0.05) luciferase production in PC-3AR/LUC cells by 2.1-fold, 2.8-fold, and 1.7-fold, respectively,confirming that each of these compounds are also agonists of the wild-type AR. L-35 and L-37 (5 μm)significantly lowered (P < 0.05) luciferase transcriptional activity in PC-3AR/LUC cells, again indicating that these compounds are antagonists of both the wild-type and mutant AR. In contrast to the results obtained with LNCaP-LUC cells, L-39 inhibited wild-type AR-mediated luciferase transcriptional activity in PC-3AR/LUC cells by 35%, suggesting that this compound is an antagonist of the wild-type AR and a weak agonist of the mutated LNCaP AR. The results indicate that L-2, L-10, and L-36 function as agonists of both receptor types, L-35 and L-37 are antagonists of both receptor types, and L-39 has mixed agonist/antagonistic activity depending on receptor type.

The ability of the compounds to modulate DHT-induced luciferase activity was determined in both cell types. In each of the experiments the cells were cotreated with DHT (5 nm) and the reference drug or novel compound (5 μm). DHT (5.0 nm)significantly (P < 0 0.001) stimulated luciferase production mediated by the LNCaP AR in LNCaP-LUC cells(Fig. 6,A). Ketoconazole had no effect on DHT-induced luciferase activity. In this cell line, flutamide and finasteride significantly lowered DHT-induced transcriptional activity suggesting that they compete for the LNCaP AR and prevent DHT from binding. All of the novel compounds significantly inhibited DHT-induced LNCaP AR-mediated luciferase transcription. The order of potency of the compounds was L-37 > L-39 > L-2 > L-35 > L-10 > L-36, and DHT-induced luciferase activity was inhibited by 99.1%, 86.8%, 73.8%, 72.5%,69.8%, and 42.1%, respectively. No synergistic or additive effects were observed on luciferase activity levels when LNCaP-LUC cells were cotreated with the agonistic compounds, L-2, L-10, or L-36 in combination with DHT. These compounds compete with DHT for binding to the mutated LNCaP AR and mediate transcription to the same level as in the absence of DHT. When the compounds were tested against the wild-type receptor in PC-3AR/LUC cells, a similar pattern was obtained(Fig. 6 B). Ketoconazole and finasteride had no effect on DHT-induced transcriptional activity and flutamide significantly inhibited (P < 0.01) luciferase activity induced by DHT. As was observed with the mutated LNCaP AR, all of the novel compounds were effective at inhibiting DHT-induced transcriptional activity mediated by the wild-type AR. The order of potency of the compounds was L-37 > L-39 > L-35 > L-2 ≥ L-36 ≥ L-10, which inhibited luciferase activity levels by 89.1%, 82.2%, 69.8%, 62.9%, 59.5%, and 58.9%,respectively.

To ensure that the results obtained with these cell lines reflected the ability of the compounds to bind to the AR and not another steroid hormone receptor, these assays were performed in steroid receptor-negative CV-1LUC cells that were transiently transfected with either the wild-type AR or the mutated LNCaP AR constructs. The results obtained were in agreement with the results described above, indicating that the effects of the compounds were specific for the AR (data not shown).

In the first experiment, the effects of L-2, L-35, and L-36 on tumor growth were determined, and orchidectomy and ketoconazole were used as the reference treatments. The mice were grouped 28 days after cell inoculation when measurable tumor volumes were approximately 500 mm³ (Fig. 7,A). After 21 days of treatment tumor volumes in the mice treated with L-2, L-36, and L-35 were similar to the castration group. However, by 28 days tumor volume in the control mice increased 4-fold over the 28 days of treatment, and tumor volume in the castrated mice increased by only 2-fold (52% reduction). None of the treatments,including ketoconazole, had any appreciable effect on the growth of the tumors after 28 days of treatment (Fig. 7,A). The reductions in tumor volumes in the mice treated with ketoconazole, L-2, L-35, and L-36 were 23%, 28%, 8%, and 12% respectively. These results were also reflected in the weights of the tumors. The only treatment that resulted in significantly (P < 0.01) smaller tumors was orchidectomy (Fig. 7,B). Tumor weight in the castrated mice was reduced by 62%. Mice that had been castrated also had significantly (P < 0.02) lower serum levels of PSA when compared with vehicle-treated animals (Table 2). In this experiment, serum and tumor levels of testosterone were significantly reduced in the mice that were castrated. Although, serum testosterone levels in the mice treated with L-2, L-35, and L-36 were significantly lower compared with the vehicle-treated mice, tumor testosterone levels were unaffected.

In the second experiment, the effects of L-10, L-37, and L-39 on tumor growth were determined and orchidectomy, finasteride, and ketoconazole were used as the reference treatments (Fig. 8,A). The mice were grouped 35 days after cell inoculation when measurable tumor volume was approximately 500 mm³(Fig. 8,A). Total tumor volume in the control mice increased 4.3-fold over the 28 days of treatment and tumor volume in the castrated mice increased by only 2.6-fold (38% reduction compared with control). Tumor volumes increased 3.2-fold in the mice treated with finasteride and, as in the previous experiment, ketoconazole had no appreciable effect on tumor growth. Tumor volumes in the mice treated with L-10 increased by 3.3-fold (22% reduction compared with control)over the 28 days of treatment. This compound was as effective as finasteride in this model. In contrast to the growth inhibition observed with L-37 in vitro (Fig. 2,3,4,5,6), tumor volumes in the mice treated with this compound increased 5.2-fold. In the mice treated with L-39, tumor volume increased by 2.4-fold, which was a 44%reduction versus control mice. In this experiment, L-39 was as effective as castration at inhibiting tumor growth. Tumor weights in the castrated and L-39-treated mice were significantly(P < 0.05) lower than those in the control mice (Fig. 8,B). Compared with vehicle-treated mice, tumors in the castrated and L-39-treated mice weighed 75% and 72% less,respectively. As shown in Table 3, serum PSA levels were significantly reduced in the mice that were castrated or treated with finasteride or L-39 (all P < 0.02), and L-10 (P < 0.05). Serum testosterone levels were significantly reduced in the groups treated with castration (P < 0.001), L-37 (P < 0.02), and L-39(P < 0.05). Tumor testosterone was significantly reduced (P < 0.001) in the mice treated by castration, and of the novel compounds tested, only L-37 significantly reduced (P < 0.02) tumor levels of testosterone.

The effects of L-39 on the growth of hormone-dependent PC-82 tumor xenografts were determined because these tumors express wild-type AR. Castration, flutamide, and finasteride were used as the reference treatments. The mice were grouped approximately 21 weeks after tumor inoculation, when measurable tumor volume was approximately 500 mm³ (Fig. 9,A). Tumor volume in the control mice increased 2.2-fold over the 28 days of treatment, and tumors in the castrated mice actually regressed by 13% over the duration of the experiment. Compared with the control mice, finasteride slowed tumor growth but not to the same extent as the antiandrogen flutamide. Tumor volume in the mice treated with flutamide increased by only 1.1-fold over the 28 days of treatment. L-39 was almost as effective as flutamide at slowing the growth rate of the PC-82 tumors. Tumor volume in the mice treated with L-39 increased by 1.3-fold during the course of the experiment. These results were also reflected in the weights of the tumors. Compared with the control group, castration, flutamide, and L-39 all significantly lowered (P < 0.02) PC-82 tumor weight (Fig. 9,B) and serum levels of PSA (Table 4). Castration also significantly lowered (P < 0.01) serum and tumor levels of testosterone. Flutamide had no effect on the serum levels of testosterone. Serum and tumor testosterone levels were significantly lower (P < 0.01)in the mice treat with L-39.

In the present study, we evaluated the ability of novel inhibitors of either C_17,20-lyase and/or 5α-reductase to inhibit the growth of hormone-dependent prostatic cancer cells in vitro and in vivo. We had reported previously that several of these novel compounds showed very good biological activity by lowering androgen levels in a normal male rat model, suggesting that they may be effective treatment options for hormone-dependent prostatic carcinoma (47, 48). In this report we describe that some of our novel androgen synthesis inhibitors also interact with the AR expressed in prostate cancer cells, and that the antiandrogenic properties of some of the compounds contribute, at least in part, to their ability to inhibit the growth of hormone dependent prostate cancers.

As an initial approach, the effects of the compounds on the growth of hormone-dependent LNCaP cells were determined. Of the six novel compounds tested, three stimulated LNCaP cell growth (L-2, L-10, and low concentrations of L-36), two inhibited cell proliferation (L-35 and L-37), and one had no effect (L-39). The mitogenic effects of L-2 and L-10 were surprising because both of these compounds are pregnane derivatives and would not expected to be recognized as ligands by the LNCaP AR. However, the LNCaP AR contains a mutation in the ligand-binding domain (threonine to alanine at residue 877) that is responsible for it recognizing other steroidal compounds as either agonists or antagonists (33, 34). It has recently been reported that residue 877 contacts the ligand directly, and the mutation alters the stereochemistry of the binding pocket(49). The mutation broadens the specificity of ligand recognition and the LNCaP AR recognizes estrogens, progestins, and antiandrogens such as flutamide as androgens. The cells respond to these compounds with increased proliferation (34, 50, 51, 52). Consistent with previously published reports from this laboratory,flutamide was found to be growth stimulatory to LNCaP cells and finasteride was growth inhibitory (Fig. 2; Refs. 45, and 46). Our results show that the effects of flutamide and finasteride are specific for the LNCaP AR and not for the wild-type AR. This is in contrast to the results obtained with our novel compounds,which interacted with both AR types and maintained the same agonistic/antagonistic properties regardless of receptor type. This indicates that some of our antiandrogenic compounds may be useful for the treatment of patients with tumors expressing either wild-type or mutated AR, or for patients with amplified AR expression. Androgen responsive cells are believed to respond to androgen ablation therapy by acquiring androgen-independence and adapting to growth in the absence mitogenic androgens. However, recent studies have suggested that in a subset of tumors, cells respond to reduced androgen levels by amplifying AR gene expression (53, 54). Therefore,AR-mediated signaling is likely to be important in the advanced-stage disease. Compounds such as L-39, which is a potent inhibitor of C_17,20-lyase and 5α-reductase and exhibits antiandrogenic properties, may be a suitable treatment option for such patients.

Prostate cancer cells with mutated AR respond to flutamide and hydroxyflutamide, the active metabolite of flutamide, with growth stimulation rather than growth inhibition. Our studies, in vitro and in vivo, used flutamide, because we have shown previously that LNCaP cells respond equally well to both compounds (45, 46). Findings of growth stimulation by flutamide led to the development of bicalutamide (Casodex), a second-generation, nonsteroidal antiandrogen (56, 57). Bicalutamide has a 4-fold-higher affinity for the AR than flutamide and is recognized by the LNCaP AR as an antiandrogen. Clinically,bicalutamide is providing better responses and is better tolerated than flutamide (57). Experiments ongoing in this laboratory are presently comparing the antitumor effects of bicalutamide to some of our more potent novel compounds. Although the LNCaP prostate cancer cell line is the most characterized androgen-dependent model of prostate cancer, the mutation in the AR is problematic for investigators researching the effects of antiandrogens. Furthermore,LNCaP cells also respond to androgens in a biphasic concentration-dependent response, with high concentrations of androgens inhibiting cell proliferation (58). We observed this with L-36, which was a potent stimulator of cell proliferation at the lower concentrations and a potent inhibitor of cell growth at the 5-μm concentration. Therefore, whereas the competitive binding studies with the synthetic androgen[³H]R1881 indicate that the compounds are interacting with both the mutated LNCaP AR and the wild-type AR,they cannot discriminate whether the compounds are acting as agonists or antagonists. We used transcriptional activation assays with luciferase activity regulated by an androgen responsive promoter to overcome these difficulties. LNCaP-LUC cells were selected in this laboratory and responded to DHT and testosterone in a manner similar to the parent cells, indicating that no clonal variation had occurred during the selection process. The studies confirmed that the growth stimulatory compounds (L-2, L-10, and L-36) were functioning as pure AR antagonists whereas the growth inhibitory compounds (L-35, L-37, and L-39) were AR antagonists. It is interesting to note that the pregnane derivatives L-2 and L-10 were agonists of both receptor types. This suggests that the side chain modification at carbon 17 may play an important role in mediating these effects. L-35, L-37, and L-39 were also determined to be antagonists of both the wild-type AR and the LNCaP AR, indicating that they are more comparable with the second-generation antiandrogen bicalutamide than with flutamide.

Having established the in vitro activities of the novel compounds, the in vivo effects were determined using LNCaP prostatic cancer cells grown as tumor xenografts in male SCID mice. In a preliminary experiment (data not shown), the growth of LNCaP prostate cancer xenografts in male SCID mice was determined to be hormone-dependent, and tumor weights were significantly lower in the animals that were castrated or treated with the 5α-reductase-inhibitor finasteride. Flutamide was found to stimulate the growth of LNCaP tumor xenografts, but not to the same extent as DHT. Additionally, the maximum time of treatment was determined to be 28 days. After this time, tumors in the castrated group appeared to have acquired hormone-independence based on a determination of weekly serum PSA levels (data not shown).

In agreement with the results obtained in vitro, the androgenic compounds L-2, L-36, and L-10 had no appreciable effect on tumor growth measurements. L-10 was found to be as effective as finasteride at inhibiting tumor growth, and L-2 and L-36 significantly lowered serum levels of testosterone but not tumor testosterone levels. Therefore, it is likely that although the compounds are effective inhibitors of androgen synthesis, this is overshadowed by their androgenic properties, which results in LNCaP tumor growth. The inability of L-35 and L-37 to inhibit LNCaP tumor growth in vivo was especially disappointing. Both of these compounds were the most effective at inhibiting LNCaP cell growth in vitro,and they were both determined to be the most potent antiandrogens. We have recently determined that L-37 is an inhibitor of the liver cytochrome P450 enzymes 3α-hydroxysteroid-oxidoreductase and 3β-hydroxysteroid-oxidoreductase, which catalyze the metabolism of DHT to polar metabolites that are excreted in the urine (data not shown). Although we were unable to reliably measure serum levels of DHT in the serum of the animals because of cross-reactivity with testosterone and the novel compounds, the results (not shown) did suggest that there were elevated serum DHT levels in the animals treated with L-37. These elevated DHT levels may explain why L-37 had no effect on tumor growth in vivo. It should also be noted that L-37 does not inhibit the growth of PC-82 prostate cancer xenografts grown in athymic nude mice in vivo (data not shown). Currently, we do not know the reasons for the lack of growth inhibition observed with L-35. The inability of L-35 to inhibit LNCaP tumor growth may be attributed to its weak affinity for the AR. Alternatively, it may be attributable to the compound being converted to an inactive metabolite, poor uptake of the compound in the animals,or rapid clearance from the animals.

The most effective compound at inhibiting LNCaP tumor growth in vivo was L-39, which paralleled castration. Moreover, L-39 was also very effective at inhibiting the growth of hormone-dependent PC-82 prostate tumor xenografts in athymic nude mice. PC-82 prostate cancer cells expresses wild-type AR, and the cells cannot be grown in vitro. In these experiments, PC-82 tumors grew to a measurable size in approximately 12 weeks. The in vitro experiments had determined that L-39 was effective at inhibiting LNCaP cell growth in the presence of exogenous steroids, and that L-39 is extremely effective at preventing high affinity androgens from binding to the AR. However, when compared with L-35 and L-37, L-39 was not the most potent antiandrogen. We have recently reported on the pharmacokinetic profile of L-39 in normal (non-tumor-bearing) male mice(59). L-39 (50 mg/kg s.c.) was shown to be cleared from the blood of the animals within 6 h of injection, and a nonpolar metabolite was detected in the serum. The metabolite has not yet been characterized, and it is not yet known whether it is contributing to the antitumor activities of L-39. It is also important to note that the multifunctional activities L-39 are most likely to explain the antitumor effects of this compound. L-39 is an inhibitor of androgen synthesis, and it is a potent antiandrogen when tested against both the mutated LNCaP AR and the wild-type AR. Although additional studies are required to determine the full potential of this compound, L-39 holds considerable promise as a treatment option for hormone dependent prostate cancer.

We are grateful to Dr. John Isaacs (The John Hopkins School of Medicine, Baltimore, MD) for providing the doner PC-82 tumor xenografts, to Dr. Marco Marcelli (Baylor College of Medicine, Houston,TX) for providing PC-3 cells and PC-3 cells stably transfected with the human wild-type androgen receptor, to Dr. Elizabeth Wilson (University of North Carolina, Chapel Hill, NC) for providing the wild-type human AR-expressing vector and the LNCaP AR-expressing vector, to Dr. Hynda Kleinman (NIH, Bethesda, MD) for providing the Matrigel, and to Merck Research Laboratories (Rahway, NJ) for providing the finasteride.