Steroid 5α-Reductase Isozymes I and II in Recurrent Prostate Cancer

Androgen target cells in peripheral tissues use testosterone to activate androgen receptor, which interacts with androgen response elements in DNA to regulate gene transcription. Alternatively, intracellular testosterone acts as a prohormone that is converted to dihydrotestosterone, a more potent androgen receptor ligand. In the prostate, an intracrine pathway (1) uses the enzyme steroid 5α-reductase (EC 1.3.99.5) to metabolize testosterone to dihydrotestosterone. The adrenal androgen androstenedione is also converted to 5α-reduced androstanedione by steroid 5α-reductase in the prostate (2) and androstenedione has been implicated as a source of dihydrotestosterone in prostate tissue after castration (3).

Steroid 5α-reductase is a membrane-associated, NADPH-dependent enzyme that catalyzes the irreversible stereospecific reduction of C¹⁹ 3-keto-Δ^4-5 steroid to 5α-reduced metabolites. Steroid 5α-reductase isozymes I (S5αRI) and II (S5αRII) in humans are composed of 260 and 254 amino acids, respectively, with 47% sequence identity and distinct biochemical properties (4). S5αRI exhibits a broad neutral to basic optimum pH range, whereas S5αRII displays a narrow acid pH optimum for maximum activity. Both isozymes contain an NH₂-terminal steroid binding domain and a COOH-terminal NADPH binding domain. S5αRI has higher turnover and decreased substrate affinity, whereas S5αRII has lower turnover and increased substrate affinity. The apparent dissociation constant for NADPH cofactor is similar for both isozymes.

Prostate cancer is the second leading cause of death and the most common non–skin cancer in American men. Current estimates predict 232,090 new cases of prostate cancer in American men and 30,350 deaths from prostate cancer in 2004 (5). Men who fail curative therapy or present with advanced disease usually receive androgen deprivation therapy that causes regression of androgen-dependent prostate cancer through programmed cell death (6). However, androgen deprivation therapy is palliative in advanced prostate cancer (7) because prostate cancer recurs and almost always causes death. A molecular role for androgen receptor in the transition from androgen-stimulated prostate cancer to recurrent prostate cancer is supported by androgen receptor expression in recurrent prostate cancer (8–10) and expression of androgen-regulated genes (10–12). Potential mechanisms include androgen receptor amplification (13), androgen receptor gene mutations leading to transactivation by low levels of dihydrotestosterone (14) or promiscuous ligand binding (15) and posttranslational modification of androgen receptor through peptide growth factor signaling (16, 17). In addition to these mechanisms, our investigations have recently found tissue levels of testosterone and dihydrotestosterone in recurrent prostate cancer that are sufficient for activation of even wild-type androgen receptor (10). However, lack of efficacy of finasteride, a S5αRII-specific inhibitor, suggests that S5αRII does not contribute significantly to dihydrotestosterone production (18).

We investigated the expression levels, subcellular location and in vitro enzymatic activity of S5αR isozymes I and II in recurrent prostate cancer compared with androgen-stimulated benign prostate and androgen-stimulated prostate cancer. S5αRI and II expression were characterized using tissue microarray immunohistochemistry and Western analysis, and enzymatic activity was assessed using pH-optimized assays for each isozyme.

Patient and tissue samples. All prostate specimens were acquired in compliance with the guidelines of the University of North Carolina at Chapel Hill Lineberger Comprehensive Cancer Center Clinical Protocol Review Committee and Institutional Review Board and the federal Health Insurance Portability and Accountability Act protected health information regulations. Immediately after surgical removal, tissue specimens were formalin-fixed and paraffin-embedded or placed in cryovials, snap-frozen in liquid nitrogen, and cryopreserved until further processing. All histologic diagnoses were confirmed by examination of frozen and corresponding formalin-fixed, paraffin-embedded tissue specimens. Prostate specimens from a total of 105 men were used for tissue microarray immunohistochemistry (n = 45), immunoblot analysis (n = 24), and enzyme assays (n = 36).

A tissue microarray was constructed that contains a total of 68 cores from a total of 45 men. Androgen-stimulated benign prostate and prostate cancer cores were obtained from the transition zone of formalin-fixed, paraffin-embedded radical prostatectomy specimens from 23 men with clinically localized prostate cancer. The patients had not received radiation or hormonal therapy prior to surgery. Mean age was 57 years (range 46-73) and Gleason sums ranged from 5 to 8. Recurrent prostate cancer cores were obtained from formalin-fixed, paraffin-embedded transurethral prostatectomy specimens from 22 men who had increasing serum prostate-specific antigen levels and urinary retention from local recurrence of prostate cancer after surgical or medical androgen deprivation therapy. Mean age was 72 years (range 57-86) and Gleason sums ranged from 8 to 10.

Steroid 5α-reductase enzyme assays were performed using snap-frozen operative specimens from 36 men, the samples were then stored in liquid nitrogen until further use. Twelve samples of androgen-stimulated benign prostate, 12 samples of androgen-stimulated prostate cancer, and 12 samples of recurrent prostate cancer were sources of steroid enzymes. The patients with androgen-stimulated benign prostate had a mean age of 63 years (range 50-75). Patients with androgen-stimulated prostate cancer had a mean age of 54 years (range 39-76) and Gleason sums were 6 or 7. Patients with recurrent prostate cancer had a mean age of 71 years (range 60-86) and Gleason sums ranged from 8 to 10.

An immunoblot was constructed from frozen prostate tissue specimen lysates from 24 different men (19) that yielded eight samples each of androgen-stimulated benign prostate, androgen-stimulated prostate cancer, and recurrent prostate cancer. The androgen-stimulated benign prostate patients had a mean age of 65 years (range 58-71). Androgen-stimulated prostate cancer patients had a mean age of 59 years (range 41-71) and Gleason sums were 6 or 7. Recurrent prostate cancer patients had a mean age of 71 years (range 60-86) and Gleason sums ranged from 8 to 10.

Microarray construction. A high-density tissue microarray was constructed using formalin-fixed, paraffin-embedded human prostate specimens as previously described (20). The tissue microarrays were constructed with matched pairs of androgen-stimulated benign prostate and prostate cancer from radical prostatectomy specimens from 23 men and recurrent prostate cancer from transurethral resection specimens from 22 men using the Beecher Instruments (Silver Spring, MD) manual tissue arrayer. Benign prostate obtained by transurethral resection, colon cancer, and mouse liver cores were included as internal controls for standardization.

Six-micrometer sections were cut from donor paraffin blocks and stained with H&E using routine methods. A pathologist (S.J. Maygarden) evaluated the tissue sections and identified benign prostate and prostate cancer. The regions of interest were sampled by removing a 1.5 mm tissue core. These cores were implanted into a recipient paraffin block to create an array containing a total of 84 tissue cores that includes 22 cores of recurrent prostate cancer, 23 cores of androgen-stimulated benign prostate, 23 cores of androgen-stimulated prostate cancer, and 16 duplicate and control tissues.

Immunohistochemistry. Optimal conditions were defined for immunodetection of 5α-reductase isozymes (21). Tissue microarray immunohistochemistry was done using the tape-transfer method of tissue mounting (Instrumedics; Hoboken, NJ) and the EnVision + Peroxidase System (Dako Cytomation; Carpinteria, CA). Six-micrometer sections of formalin-fixed, paraffin-embedded specimens were deparaffinized and rehydrated using Hemo-De and graded alcohols. Antigens were retrieved by incubation in citrate buffer (pH 6.0; Biocare Medical, Walnut Creek, CA) for 2 minutes at 120°C and 22 psi. Endogenous peroxidases were blocked using 0.03% H₂O₂ for 5 minutes at 37°C. Specimens were incubated for 30 minutes at 37°C with anti-human S5αRI (12 μg/mL) and S5αRII (6 μg/mL) IgG specific for NH₂-terminal epitopes (21). Tissues were incubated with polymer-conjugated anti-rabbit IgG (Dako Cytomation) followed by visualization using diaminobenzidine tetrahydrochloride (Vector Labs, Burlingame, CA). Counterstaining was done using hematoxylin (Gill's formula, Vector Labs). Specificity was confirmed when S5αRI and S5αRII polyclonal antibody binding to antigen was prevented by preincubation with S5αRI or S5αRII epitope peptide (22).

Quantitative morphometry. Tissue sections from 36 prostate specimens were cut at 6 μm, deparaffinized, and stained with H&E. An average of 12 images per patient specimen were collected and analyzed using Image Pro Plus 4.5 (Media Cybernetics, Inc. Silver Spring, MD). Each visible lumen was circled and the enclosed area calculated. Epithelium was circled and the area calculated by subtracting the area of the lumen. The area of the stroma was determined by subtracting the area of the lumen and epithelium from total image size. Percent epithelium was the area of the epithelium divided by the sum of the area of the epithelium and the area of the stroma. Percent stroma was the area of stroma divided by the sum of the area of epithelium and the area of stoma (23).

Image acquisition analysis. One or two random images were acquired at 400× magnification from each core on the tissue microarray using Leica DMRA2 microscope (Leica Microsystems, Inc., Bannockburn, IL) with a Ludl stage controller (Ludl Electronic Products, Ltd., Hawthorne, NY) and a Hamamatsu 3 Chip CCD camera with controller (Hamamatsu, Bridgewater, NJ) interfaced with a Flashpoint three-dimensional image grabber card (Integral Technologies, Indianapolis, IN) in a Pentium IV-based PC. Image Pro Plus 4.5 (Media Cybernetics) software was used to capture and store the images. The images have 24-bit color depth and 640 × 480 pixel resolution and were stored as TIFF image files. An image album was created using Adobe Photoshop 7.0 (Adobe Systems, Inc., San Jose, CA) with the acquired images for visual scoring.

Tissue microarray sections were visually scored by experienced observers (M.A. Titus and O.H. Ford) blinded to type of antibody and the androgen status of the patients from whom tissue was procured. In addition, O.H. Ford was blinded to the nature of the experiment. Epithelial nuclear and cytoplasmic S5αRI and S5αRII immunostaining were semiquantitatively assessed on a scale ranging from 0 (no expression) to 3 (strong immunostaining) in each of 100 nuclei or cells, respectively, to yield a visual score ranging from 0 to 300 for each feature for each specimen (24).

Statistical analysis. Data were described by the mean visual scores of S5αRI and S5αRII immunostaining. Statistical analyses were done using Statgraphics Plus 4.1 (Manugistics, Inc., Rockville, MD). Student's t test was used to compare S5α-RI and II nuclear immunostaining in androgen-stimulated benign prostate, androgen-stimulated prostate cancer, and recurrent prostate cancer, followed by ANOVA (Tukey HSD multiple comparison test). Differences were considered significant at P < 0.05.

Steroid 5α-reductase in vitro assays. Tissue preparation and enzyme incubations protocols by Moore and Wilson (25) were used with modifications. All steps were carried out at 4°C. Fifty milligrams of prostate tissue were used to allow sufficient protein concentrations for multiple assays. The total protein concentration of each of 12 samples of androgen-stimulated benign prostate, 12 samples of androgen-stimulated prostate cancer, and 12 samples of recurrent prostate cancer was converted to epithelial protein concentration by dividing total protein concentration by percent epithelium. S5αRI and S5αRII specific activities were expressed in picomoles per milligram of epithelial protein per minute for each specimen.

Prostate tissue was pulverized in liquid nitrogen using mortar and pestle and homogenized in 1 mL ice-cold homogenization buffer [10 mmol/L Tris-HCl (pH 7.4), 0.25 mol/L sucrose, 1 mmol/L EDTA, 1 mmol/L DTT, 1 mmol/L fresh phenylmethylsulfonyl chloride, 1.3× Halt protease inhibitor (Pierce, Rockford, IL), 3 mmol/L NADPH (Roche Applied Science Indianapolis, IN), and 0.5 nmol/L androstenedione (Sigma-Aldrich, St. Louis, MO)] using a PowerGen 700 for three 10-second bursts. Final homogenates were centrifuged at 800 × g for 5 minutes to remove connective tissue. The resulting cell-free supernatants were stored in 250 μL aliquots at −80°C. Protein concentrations were measured using the procedure of Lowry et al. (26).

The buffer for all incubations contained 150 mmol/L Tris-citrate, 0.5 mmol/L DTT, 0.5 mmol/L EDTA, 0.5 mmol/L fresh phenylmethylsulfonyl chloride, 0.25 mg/mL bovine serum albumin, 3 mmol/L NADPH, and 50 nmol/L [³H]androstenedione (New England Nuclear, Boston, MA). Incubations were done in a total volume of 200 μL at pH 7.5 for S5αRI and pH 5.5 for S5αRII. Enzymatic reaction was initiated by the addition of 50 μL supernatant that was mixed gently for 5 seconds and shaken for 30 minutes in a 37°C water bath. The incubations were quenched by adding of 1 mL ice-cold chloroform/methanol (8:2, vol/vol) that was vortexed vigorously and placed on ice after centrifugation to separate organic and water phases. Homogenization buffer at 0.9% ethanol was added to control incubations in place of cell-free supernatant. All incubations were done in duplicate under protein and time linearity.

The conversion of androstenedione to 5α-reduced metabolites androstanedione and dihydrotestosterone was measured using TLC (25, 27). An aliquot of dissolved organic extraction containing tritiated steroids and internal standards (0.25 mg/mL dihydrotestosterone, androstenedione, androstanedione, testosterone; Sigma-Aldrich) were separated using silica-coated flexible plates and chloroform/methanol (98:2, vol/vol) mobile phase. The developed plates were visualized using anisaldehyde spray. Zones corresponding to stained reference steroids were transferred into vials containing 5 mL of liquid scintillation cocktail and assayed for tritium. Control incubations lacking enzyme were analyzed as described above.

The relative amount of each radioactive steroid was calculated as percent total radioactivity (³H) recovered from the TLC lane. Blank values were subtracted from tissue metabolism rates.

Immunoblot analysis. Protein lysates from frozen patient samples were isolated as described previously (19). Briefly, 100 mg tissue pieces were pulverized under liquid nitrogen and mixed with 1 mL of radioimmunoprecipitation buffer containing protease inhibitors [0.15 mol/L NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 5 mmol/L EDTA, 50 mmol/L Tris (pH 7.4), 0.5 mmol/L phenylmethylsulfonyl fluoride, 10 μmol/L pepstatin, 4 μmol/L aprotinin, 80 mg/mL leupeptin, 0.2 mmol/L sodium vanadate, and 5 mmol/L benzamidine] and 1 μmol/L dihydrotestosterone. After homogenization and incubation for 15 minutes on ice, lysates were centrifuged at 12,000 × g twice for 15 minutes. Supernatant proteins (100 μg) were electrophoresed on 10% polyacrylamide gels containing SDS and electroblotted to Immobilon-P membranes (Millipore Corp., Bedford, MA). Immunoblot analysis was done using S5αRI and S5αRI antibodies (1.0 μg/mL) and horseradish peroxidase conjugated rabbit IgG. Specific signals were detected using enhanced chemiluminescence (SuperSignal West Dura Extended Duration Substrate, Pierce).

Nuclear S5αRI immunostaining was moderate and similar in androgen-stimulated benign prostate, androgen-stimulated prostate cancer, and recurrent prostate cancer (Table 1; Fig. 1). Epithelial nuclei exhibited lower S5αRII than S5αRI immunostaining in androgen-stimulated benign prostate, androgen-stimulated prostate cancer, and recurrent prostate cancer (P < 0.0001). The difference between S5αRI and S5αRII nuclear immunostaining varied marginally across the three tissue types (ANOVA, P = 0.055). Nuclear S5αRII immunostaining in recurrent prostate cancer was lower than androgen-stimulated benign prostate and androgen-stimulated prostate cancer (P < 0.0001). Cytoplasmic S5αRI immunostaining was lower in recurrent prostate cancer than androgen-stimulated benign prostate or androgen-stimulated prostate cancer (P < 0.00001). Recurrent prostate cancer cytoplasmic S5αRII had decreased immunostaining versus androgen-stimulated benign prostate or androgen-stimulated prostate cancer (P < 0.0001). S5αRI and S5αRII immunostaining was not detectable in androgen-stimulated benign prostate, androgen-stimulated prostate cancer, and recurrent prostate cancer stroma.

Immunoblot analysis confirmed immunohistochemical results. In recurrent prostate cancer, the 23 kDa S5αRI was expressed at high levels in six specimens and intermediate levels in two specimens (Fig. 2). S5αRII expression was below detectable levels in all eight recurrent prostate cancer specimens even after extended exposure. Expression levels of S5αRI and S5αRII were similar across all eight androgen-stimulated benign prostate specimens. The eight androgen-stimulated prostate cancer specimens revealed variable levels of expression of S5αRI and S5αRII. All three tissue types, androgen-stimulated benign prostate, androgen-stimulated prostate cancer, and recurrent prostate cancer, showed greater expression of S5αRI than S5αRII.

The proportion of the 5α-reducing capability determined under optimal conditions varied in the three tissue types. Epithelium comprised a median of 25% in androgen-stimulated benign prostate, 52.7% in androgen-stimulated prostate cancer, and 94.8% in recurrent prostate cancer (Table 2). In both androgen-stimulated benign prostate and androgen-stimulated prostate cancer, S5αRII activity exceeded S5αRI activity (Table 3). However, in recurrent prostate cancer, S5αRI activity was 3.7-fold greater than S5αRII activity.

S5αRI and S5αRII immunostaining was detected in nuclear and cytoplasmic compartments of androgen-stimulated benign prostate, androgen-stimulated prostate cancer, and recurrent prostate cancer epithelial cells but no immunostaining was detected in stroma. In recurrent prostate cancer, S5αRI localized predominantly to the nucleus and S5αRII localized predominantly to the cytoplasm. The subcellular nuclear localization of S5αRI and cytoplasmic localization of S5αRII in recurrent prostate cancer epithelium shown in this study has been reported previously in normal prostate tissue (28). Bonkhoff et al. (29) characterized S5αRI and S5αRII immunoreactivity in benign prostate tissue and recurrent prostate cancer. S5αRI and S5αRII showed increased nuclear and cytoplasmic immunostaining in recurrent prostate cancer, respectively, compared with benign prostate (29). In this study, S5αRI nuclear immunostaining remained similar between androgen-stimulated benign prostate and recurrent prostate cancer and cytoplasmic S5αRII was significantly lower in recurrent prostate cancer. The ratio of S5αRI/S5αRII immunostaining changed from approximately 5:1 in androgen-stimulated benign prostate and 3:1 in androgen-stimulated prostate cancer to 12:1 in recurrent prostate cancer. The COOH-terminal specific S5αRI and S5αRII antibodies and extended antibody exposure time used in the previous study may explain the differences in the two studies. Our immunohistochemistry data was supported further by Western blots and enzyme assays.

5α-Reductase activity in human hyperplastic and malignant tissues has been reported previously (30). More recently, individual isozyme activities have been shown in specimens of benign hyperplasia (31) and prostate cancer (32). Our results for androgen-stimulated benign prostate and androgen-stimulated prostate cancer agree with S5αRI and S5αRII activities reported previously by Soderstrom et al. (32). To our knowledge, individual S5αRI and S5αRII isozyme activities have not been reported in recurrent prostate cancer. S5αRI activity is 3.7-fold that of S5αRII in recurrent prostate cancer. The increased S5αRI enzyme activity is consistent with higher levels of S5αRI enzyme found by immunohistochemistry and immunoblot. The residual S5αRII enzyme activity in recurrent prostate cancer tissue may be attributed to S5αRI activity (33) which persists at optimal pH in this in vitro system.

The switch toward S5αRI expression in recurrent prostate cancer may be in response to angiogenesis (34) and its role in epithelial microenvironment regulation. Neovascularization mediated by vascular endothelial growth factor, a universal characteristic of solid tumors, has been linked to aggressiveness of prostate cancer (35). Androgens regulate vascular endothelial growth factor production in prostate cancer (36). Although angiogenesis increases tumor vascularization, portions of the tumor and even cells adjacent to neovessels may be hypoxic (37). A hypoxic microenvironment stabilizes hypoxia-inducible factor 1α and changes tumor metabolism by increasing expression of hypoxia-inducible factor 1α target genes. One such protein, carbonic anhydrase 9 (38), is a tumor-associated transmembrane enzyme that may influence microenvironmental pH. Additionally, the promoter of carbonic anhydrase 9 has been reported to be sensitive to increased cell density (39) that is characteristic of recurrent prostate cancer. Wykoff et al. (38) proposed carbonic anhydrase 9 expression provides a mechanism for maintaining extracellular acidosis and intracellular alkalosis (40) which promotes tumor growth. The intracellular basic pH would serve to optimize S5αRI activity in recurrent prostate cancer. The S5αRI dihydrotestosterone production in turn stimulates tumor growth and hypoxia-inducible factor 1α expression, which promotes tumor growth and basic intracellular pH. The hypoxic tissue microenvironment favors S5αRI conversion of testosterone to dihydrotestosterone and stabilizes hypoxia-inducible factor 1α transcription factor which in turn increases expression of its target genes.

The kinetic variables of S5αRI are distinct from S5αRII (41). S5αRI requires higher steroid substrate concentration compared with S5αRII to achieve half maximal rate of dihydrotestosterone production. In androgen-stimulated prostate, both isozymes contribute to dihydrotestosterone production, but loss of S5αRII activity decreases the amount of dihydrotestosterone formed in recurrent prostate cancer. Secondly, testosterone levels should be elevated because the S5αRI dissociation constant is increased requiring higher testosterone levels to drive S5αRI conversion of testosterone to dihydrotestosterone. Therefore, recurrent prostate cancer testosterone levels should be elevated because of minimal S5αRII expression and increased S5αRI substrate dissociation constant. Furthermore, recurrent prostate cancer dihydrotestosterone levels should decrease due to loss of both S5αRII and S5αRI expression in recurrent prostate cancer. These factors support the equivalent testosterone levels detected in androgen-stimulated benign prostate and recurrent prostate cancer specimens and decreased dihydrotestosterone levels in recurrent prostate cancer compared with androgen-stimulated benign prostate (10).

The relatively high testosterone level found in recurrent prostate cancer could also result in local estradiol biosynthesis by cytochrome P450 aromatase (EC 1.14.14.1, Fig. 3). Aromatase protein and activity has been shown in prostate cancer (42) and adipose tissue in men during aging (43). Estradiol biosynthesis may increase peripherally in adipose tissue or locally in recurrent prostate cancer as a consequence of blocking S5αRI and S5αRII formation of dihydrotestosterone. Prostate cancer patients receiving LY320236, a competitive inhibitor of 5α-reductase I and II, showed increased circulating estradiol levels with simultaneous reduction of serum PSA (44). Raloxifene, a selective estrogen receptor modulator, has been reported to increase apoptosis in recurrent prostate cancer cell lines DU145, PC3 (45) and the synthetic estrogen, diethylstilbestrol, decreased PSA levels in men with recurrent prostate cancer in a phase II clinical trials (46). Additionally, cytochrome P450 hydroxylation of estradiol to A-ring catechols can mediate cellular damage by redox cycling (47) and catechol phase II metabolite, 2-methoxyestradiol, has potent apoptotic and antiangiogenic effects (48).

We characterized S5αRI and S5αRII expression in recurrent prostate cancer using immunohistochemistry and immunoblot analysis and enzyme activity using in vitro assays. These data suggest that inhibition of both S5αRI and S5αRII may offer effective treatment for recurrent prostate cancer. Although finasteride, a selective S5αRII mechanism–based inhibitor and flutamide, an antiandrogen, have been shown to be an effective therapy for prostate cancer (49), finasteride is ineffective in recurrent prostate cancer (18). Dutasteride has been used safely to treat men with lower urinary tract symptoms (50) and LY320236, an inhibitor of S5αRI and II, decreased serum PSA levels in a limited study of nine castrated men with prostate cancer (44). Dutasteride, a dual S5αRI and II inhibitor, may decrease tissue dihydrotestosterone levels below those necessary for androgen receptor activation. Future characterization of androgen receptor–regulated gene expression after dutasteride treatment in prostate cancer cell lines or increased survival rates in clinical studies will support inhibition of both isozymes of S5αR. Dutasteride should be investigated to determine if it can extend the duration of remission induced by androgen deprivation therapy or induce re-remission of recurrent prostate cancer.

We thank Dr. Catherine B. Lazier for 5α-reductase I and II antibodies and Dr. Frank S. French for comments and critical review of the manuscript.