Activation of the Androgen Receptor by Intratumoral Bioconversion of Androstanediol to Dihydrotestosterone in Prostate Cancer

Prostate cancer development and growth depend on the androgen receptor (AR), a ligand-dependent transcription factor required for normal male reproductive function. AR binds testosterone and dihydrotestosterone (DHT) with high affinity to mediate androgen-dependent gene transcription (1). AR is expressed during all stages of prostate cancer progression, and increased AR transcriptional activity is a hallmark of the disease. Inhibition of prostate cancer cell growth by small inhibitory RNAs that target AR provides further evidence for obligatory AR function in prostate cancer development and progression (2).

Prostate cancer growth is stimulated initially by circulating testicular androgens. After treatment by medical or surgical castration, prostate cancers adapt to the androgen-deprived environment to maximize AR function through mechanisms facilitated by the genetic instability of cancer cells. Mechanisms for increased AR transactivation during prostate cancer progression to castration-recurrent growth include AR gene amplification (3), somatic AR gene mutations that provide a gain-of-function by decreasing AR ligand specificity (4, 5), and increased AR interactions with coregulators whose levels also increase during prostate cancer progression (6, 7). Prostate cancer tissue production of androgen develops during androgen deprivation therapy (8–10), and increased mitogen signaling and AR phosphorylation influences AR transcriptional activity (6, 11).

The present study investigated whether AR activation by the conversion of 5α-androstane-3α,17β-diol (androstanediol) to DHT contributes to prostate cancer growth when circulating testicular androgen levels are low. Cell-specific bioactivation of androstanediol to DHT and catabolism to androsterone were investigated as mechanisms for increased AR transcriptional activity that mediates castration-recurrent growth in men undergoing androgen deprivation therapy to treat advanced prostate cancer. Levels of bioavailable DHT reflect activities of several metabolic enzymes. Intracellular DHT derives primarily from the irreversible conversion of testosterone by 5α-reductase. Androstenedione and dehydroepiandrosterone sulfate are major circulating adrenal androgens converted to androstanediol and DHT in human prostate (12). Androstanediol is an inactive DHT metabolite that can be reversibly oxidized to DHT.

Herein, we demonstrate AR transcriptional activity in prostate cancer cell lines in the presence of androstanediol is related directly to mRNA and protein levels of 17β-hydroxysteroid dehydrogenase 6 (17β-HSD6) to a greater extent than the retinol dehydrogenase 5 (RDH5), or dehydrogenase/reductase short-chain dehydrogenase/reductase family member 9 (DHRS9), enzymes that convert androstanediol to DHT. Results were corroborated using mass spectrometry to measure DHT and androsterone levels in cells after incubation with androstanediol, and after the administration of androstanediol dipropionate to athymic mice bearing the CWR22R castration-recurrent prostate cancer (CR-CaP) xenograft. Androstanediol was converted to DHT in the CWR22R xenograft tumor, where mRNA levels were measured for 17β-HSD6, DHRS9, and RDH5. The results support the notion that androstanediol metabolism to DHT coupled with 5α-reductase activity contributes to optimal utilization of androgen precursors and catabolites for AR transactivation during prostate cancer development and progression.

Patient specimens of androgen-stimulated benign prostate (AS-BP) and androgen-stimulated and CR-CaP correspond to samples analyzed for AR and melanoma antigen-A11 mRNA expression (7). Procedures using mice were performed in accordance with the National Institutes of Health and Roswell Park Cancer Institute Institutional Animal Care and Use Committee and Institutional Biosafety Committee. Serially transplanted androgen-dependent CWR22 human prostate cancer xenografts were propagated in athymic nu/nu mice (13) and excised before and at different times after castration for RNA analysis using quantitative real-time (RT) PCR (7). To demonstrate intratumoral conversion of androstanediol to DHT, male athymic nude mice 4–5 weeks old were purchased from Harlan Sprague Dawley, Inc. and housed individually in the Division of Laboratory Animal Research Facility, Roswell Park Cancer Institute. One day after castration, mice were inoculated subcutaneously on one flank with 10⁶ CWR22R castration-recurrent xenograft cells suspended in Matrigel (1:1 mixture, BD Biosciences). When CWR22R tumors measured 0.5 cm³, seven mice were injected subcutaneously at the tumor site with 1 mg androstanediol dipropionate (Steraloids) in 0.1 mL sesame oil, and five mice received vehicle alone. Tumors were excised 2 days later, cut into 0.1 cm³ pieces, frozen in liquid N₂ and stored at −80°C until analysis using mass spectrometry.

For AR transcription assays, cells were cultured as described in refs. 1, 7. CWR-R1 (2 × 10⁵ cells/well of 12 well plates), RWPE-2 (1.5 × 10⁵), LAPC-4 (2 × 10⁵), LNCaP (3.5 × 10⁵), LNCaP-C4-2 (2 × 10⁵), and DU145 cells (1.5 × 10⁵) were transfected using Effectene (Qiagen) with 0.01 μg pCMV-AR and 0.1 μg (CWR-R1 and LAPC-4) or 0.25 μg (RWPE-2, LNCaP, LNCaP-C4-2, and DU145) prostate-specific antigen enhancer luciferase reporter (PSA-Enh-Luc). PWR-1E (2 × 10⁵), PC-3 (1.5 × 10⁵), and HeLa cells (5 × 10⁴) were transfected in 12 well plates using FuGENE 6 (Roche Applied Science) with 0.01 μg pCMV-AR and 0.25 μg PSA-Enh-Luc. CV1 cells (4.2 × 10⁵ cells/6 cm dish) were transfected using calcium phosphate with 0.1 μg pCMV-AR and 5 μg PSA-Enh-Luc. Cells were transferred to serum-free phenol red-free medium 24 hours after transfection, and incubated for 24 hours with and without testosterone, DHT, and androstanediol (Sigma). Cells were washed with phosphate buffered saline and harvested in 0.25 mL (or 0.5 mL for CV1 cells) lysis buffer containing 1% Triton X-100, 2 mmol/L EDTA, and 25 mmol/L Tris phosphate, pH 7.8. After rocking for 30 minutes at room temperature, 0.1 mL aliquots were assayed for luciferase activity using an automated Lumistar Galaxy (BMG) luminometer.

Immunoblots were performed by plating cells in 10-cm dishes with serum-containing medium to achieve 40% to 60% confluence the next day. After 48 hours, cells were scraped into cold phosphate buffered saline, extracted in 0.1–0.2 mL lysis buffer containing 1% Triton X-100, 1% deoxycholate, 0.1% SDS, 0.15 mol/L NaCl, 2 mmol/L EDTA, 2 mmol/L sodium vanadate, 0.05 mol/L sodium fluoride, 50 mmol/L Tris–HCl, pH 8.0 and 0.1 μM DHT, and 150 μg protein/lane analyzed on 12% acrylamide gels containing SDS. Nitrocellulose transfer blots were probed using AR52 (10 μg/mL) and AR32 (0.3 μg/mL) antibodies, and HSD17B6 rabbit polyclonal antibody (Abcam ab62221, 1.25 μg/mL). To probe β-actin using a mouse antibody (Abcam, 1:1000 dilution), blots stored at 4°C were stripped at 55°C for 25 minutes in buffer containing 2% SDS, 92 mmol/L β-mercaptoethanol and 62.5 mmol/L Tris–HCl, pH 6.7. The blot was washed twice for 10 minutes with 0.15 mol/L NaCl, 0.05% Tween 20, and 10 mmol/L Tris–HCl, pH 7.5, and blocked for 1 hour in the same buffer containing 5% nonfat dry milk prior to antibody addition. Immunoreactive bands were visualized by chemiluminescence (SuperSignal Western Dura Extended Duration substrate, Pierce Biotechnology, Inc.).

LAPC-4 cell growth assays were carried out in triplicate by plating 4 × 10⁴ cells/well of 24-well plates in RPMI-1640 medium containing 10% charcoal stripped fetal bovine serum (Atlanta Biologicals), 20 mmol/L l-glutamine, penicillin, and streptomycin. The next day and 48 hours later, 0.1 mL serum-free medium was added per well to a final concentration of 0.1 nmol/L DHT and 10 nmol/L androstanediol. Twenty-four hours after hormone addition (day 1) and at 24-hour intervals, media were aspirated, 0.2 mL fresh serum-free medium and 20 μL WST-8 cell counting reagent (Dojindo Molecular Technologies) were added per well. Cells were incubated for 2.5 hours at 37°C and optical density determined at 485 nm. Statistical analysis was performed using the Pearson Product Moment Correlation.

RNA was extracted using TRIzol Reagent (Invitrogen). First strand complementary DNA (cDNA) was prepared using SuperScript II reverse transcriptase (Invitrogen). PCR primers and fluorogenic probe for the constitutive housekeeping gene, peptidylprolyl isomerase A (PPIA; cyclophilin A) were described (14). Primers and probe for human RDH5 (Hs00161263-m1, Applied Biosystems) yield a 126-base pair amplicon spanning exon 3 and 4 junction at assay location nucleotide 681 (NM-002905.2). The 17β-HSD6 primers and probe (Hs00366258-m1, Applied Biosytems) yield a 84-base pair amplicon spanning exon 3 and 4 junction at assay location 679 nucleotides (NM-003725.2). Human DHRS9 primers and probe (Hs00608375-m1, Applied Biosytems) result in a 66 base pair amplicon spanning exon 3 and 4 junction at assay location nucleotide 872 (NM-199204.1). PCRs (20 μL) contained cDNA prepared from 0.04 μg total RNA for 17β-HSD6 and RDH5 mRNA, and 0.4 μg total RNA for DHRS9 and peptidylprolyl isomerase A. cDNA was combined with 4 μL LightCycler TaqMan Master Mix (Roche) and 0.5 μL 20× TaqMan Mix (Applied Biosystems). Thermal cycler reactions were carried out in triplicate and repeated twice using a Roche LightCycler at 95°C for 10 minutes, followed by 55 cycles at 95°C for 15 seconds, 60°C for 25 seconds, and 72°C for 1 second. mRNA copy number was calculated based on CT value of 21.85 for 4 × 10⁵ copies with an amplification efficiency of 2, normalized by total RNA in the reaction.

Liquid chromatography tandem mass spectrometry analysis of DHT, androstanediol and androsterone was performed as described (9). Cells were incubated for 48 hours at 37°C with 100 nmol/L androstanediol in serum-free, phenol red-free medium. Cells and media were collected together in triplicate and data pooled from two experiments. Deuterated 5α-androstan-17β-ol-3-one-16,16,17-d₃ (DHT-d₃; CDN Isotopes; 12 ng) was added as internal standard and samples were extracted twice with 1.5 mL 9:1 chloroform:acetone. Organic layers were combined, evaporated under vacuum, and concentrated using solid phase extraction carbon-18 columns (Varian). DHT, androstanediol and androsterone were measured using an Agilent 1100 capillary liquid chromatography system coupled to an Applied Biosystems–MDS Sciex API-3000 triple quadrupole mass spectrometer (MDS Sciex, Concord). Positive ions were formed via dopant-assisted atmospheric pressure photoionization (Applied Biosystems). A Phenomenex Luna C18 column (3 μm, 150 × 0.5 mm²) and a gradient profile using mobile phase A (2 mmol/L ammonium formate) and mobile phase B (2 mmol/L ammonium formate in methanol) at 175 μL/min (65% to 80% at 2.25 minutes and 95% at 13 minutes) were used at 60°C. Androgen parent-product ion pairs monitored (mass to charge ratio, m/z) were 305.2–255.2 for DHT, 273.2–255.2 for androstanediol, 291.2–273.2 for androsterone, and 308.2–258.2 for the DHT-d₃ internal standard. DHT, androstanediol, and androsterone standards were from Sigma-Aldrich. DHT-d₃ deuterated twice at carbon 16 and once at carbon 17 was used to quantitate androgen. Steroid concentrations were calculated based on picomole per gram assuming 1000 g/L. The 10 pg limit of detection for DHT had signal to noise ratio >3, and the 50 pg limit of quantitation of DHT had signal to noise ratio >10.

AR transactivation in the presence of testosterone, DHT or androstanediol was compared with benign and cancer-derived cell lines that included monkey kidney CV1 cells, human cervical carcinoma HeLa cells, benign human prostate-derived PWR-1E cells, and PC-3, LAPC-4, and CWR-R1 human prostate cancer cells (Fig. 1). A concentration-dependent increase in AR transcriptional activity was demonstrated in response to testosterone and DHT in all cell types subsequent to transfection of pCMV-AR and the PSA-Enh-Luc reporter gene. The minimal transactivation of the PSA enhancer after transfection with pCMV5 empty vector showed that androgen-dependent gene activation resulted from the expressed wild-type AR, with maximal luciferase activity between 0.1 and 10 nmol/L testosterone or DHT. CWR-R1 and LAPC-4 cells appeared more responsive to low levels of testosterone and DHT (Fig. 1F).

In contrast to testosterone and DHT, there were major cell-type differences in AR transactivation with androstanediol. AR transactivation in the presence of 10 nmol/L androstanediol was insignificant in CV1 cells, but evident in HeLa cells at 10 nmol/L androstanediol (Fig. 1A and B). In PC-3, LAPC-4, CWR-R1, and PWR-1E cells, androstanediol was nearly equipotent with testosterone and DHT (Fig. 1C–F). AR binds androstanediol with low affinity (15), suggesting AR transactivation with androstanediol resulted from cell-specific conversion to DHT.

To investigate whether cell-specific AR transactivation in the presence of androstanediol resulted from oxidative conversion to DHT, different cell types were incubated with 100 nmol/L androstanediol and androgen metabolites measured using mass spectrometry (Fig. 2A). Cell incubations were carried out for 24 hours to parallel the AR transactivation assays (Fig. 1) and for 48 hours to assess metabolite stability. Low to undetectable androstanediol was measured in all cell types after 24 and 48 hours (Fig. 2A). In contrast, DHT was highest in PWR-1E cells, at moderate levels in the LAPC-4, CWR-R1, and PC-3 cells, but nearly undetectable in HeLa and CV1 cells. Testosterone was undetectable (LAPC-4, PC-3, and HeLa cells) or 5-fold less than DHT (CWR-R1 cells; not shown). Highest levels of androsterone were measured in HeLa and CV1 cells, with lower levels of androsterone in prostate-derived cells in an inverse relationship with DHT. For LNCaP and LNCaP-C4–2 cells, mass spectrometry indicated essentially undetectable levels of androstanediol, DHT, androsterone, testosterone, and 5α-androstenedione after 24 and 48 hours incubations with 100 nmol/L androstanediol, consistent with the inability of androstanediol to activate AR in LNCaP cells (not shown), and with the high glucuronosyltransferase activity in LNCaP cells, an enzyme that irreversibly converts androstanediol to the sugar conjugate for excretion (16).

An additional indicator of androstanediol conversion to an active androgen was provided by the ability of 10 nmol/L androstanediol or 0.1 nmol/L DHT to increase the growth of LAPC-4 cells (P < 0.001) that contain a wild-type AR (Fig. 2B). Together the results indicate cell-specific AR transcriptional activity and prostate cancer cell growth in the presence of androstanediol results from the oxidative metabolism of androstanediol to DHT.

Several enzymes are involved in the oxidative metabolism of androstanediol to DHT, and in the reductive conversion of androstanediol to androsterone (Fig. 3). Androstanediol is converted to DHT by oxidative 3α-HSD activity of 17β-HSD6 (17), RDH5 (18), and DHRS9 (19; Table 1). Androstanediol is metabolized to androsterone by 17β-HSD6 and 11, and androsterone is glucuronidated for excretion (20–22).

Quantitative RT-PCR analysis of RNA from different cell types indicated higher levels of 17β-HSD6 mRNA in prostate-derived cells than DHRS9 or RDH5 mRNA (Fig. 4A), which correlated directly with AR transactivation and DHT levels after incubation with androstanediol (Figs. 1 and 2A). 17β-HSD6 mRNA levels were lowest in CV1 cells, where AR was transcriptionally inactive with androstanediol. HeLa cell 17β-HSD6 mRNA levels were ∼7-fold greater, consistent with the increase in AR activity with 10 nmol/L androstanediol. Highest levels of 17β-HSD6 mRNA were in PWR-1E and RWPE-2 benign prostate cells, and in LNCaP, LNCaP-C4–2, CWR-R1, and DU145 prostate cancer cells, where AR transactivation with androstanediol was similar to testosterone or DHT. In contrast, DHRS9 and RDH5 mRNA levels were lower than 17β-HSD6 mRNA (Fig. 4A) and did not correlate with DHT levels or AR transactivation with androstanediol. RDH5 mRNA levels were ∼4-fold higher in CWR-R1 cells than other cell types. However, this increase was not associated with a shift in dose response of AR activity with androstanediol relative to other prostate cancer cells (Fig. 1).

Higher levels of 17β-HSD6 mRNA in prostate-derived cell lines, where AR activity was greatest in response to androstanediol, were also associated with higher levels of 17β-HSD6 protein (Fig. 4B). The ∼30 kDa 17β-HSD6 protein was detected in all prostate-derived cell lines, but not in CV1 or HeLa cells which had correspondingly lower levels of 17β-HSD6 mRNA and AR was less active with androstanediol. AR protein levels were higher in LAPC-4, CWR-R1, and LNCaP prostate cancer cells than in PC-3, PWR-1E, HeLa, or CV1 cells (Fig. 4B). Longer exposure of the transblot revealed low levels of 17β-HSD6 in HeLa cells and AR in PWR-1E cells (not shown). However, 17β-HSD6 siRNA introduced into cells did not decrease AR transactivation in response to androstanediol due to insufficient knockdown and the contributions of other enzymes that metabolize androstanediol to DHT.

The results suggest that 17β-HSD6 contributes to androstanediol conversion to DHT in benign and malignant prostate, and bioactivation of androstanediol to DHT accounts for AR transactivation in the presence of androstanediol.

To investigate whether enzymes involved in DHT biosynthesis in prostate cancer tissue contribute to prostate cancer progression during androgen deprivation therapy, mRNA levels for 17β-HSD6, DHRS9, and RDH5 were determined in the CWR22 xenograft at different times after castration. The CWR22 human prostate cancer xenograft undergoes remission after castration and regrowth ∼120 days after castration and thus mimics the clinical response to androgen deprivation (23, 24). 17β-HSD6, DHRS9, and RDH5 mRNA levels varied in the CWR22 xenograft depending upon time after castration (Fig. 5A). 17β-HSD6 mRNA levels decreased ∼5-fold within 2 days after castration and remained low throughout castration-recurrent tumor growth. DHRS9 and RDH5 mRNA levels increased transiently after castration, but were unchanged in the castration-recurrent xenograft CWR22 tumor relative to the tumor prior to castration.

The decline in 17β-HSD6 mRNA after castration in the CWR22 xenograft was investigated further using AS-BP and androgen-stimulated and CR-CaP tissue. Similar to results with the CWR22 prostate cancer xenograft, 17β-HSD6 mRNA levels were highest in benign prostate, more variable in androgen-stimulated prostate cancer (AS-CaP), and declined ∼10-fold in most specimens of CR-CaP (Fig. 5B).

To obtain further evidence for the in vivo intratumoral metabolism of androstanediol to DHT, castrated CWR22R-bearing athymic mice were injected subcutaneously in oil at the tumor site with 1 mg androstanediol dipropionate. Intratumoral DHT measured 48 hours later using tandem mass spectrometry (Table 2) was 28-fold higher (28.1 ± 6.2 nmol/L DHT) after administering androstanediol dipropionate than control tumor-bearing mice injected with vehicle (1.0 ± 0.7 nmol/L DHT). Androsterone levels increased 4.3-fold after injection of androstanediol dipropionate. Androstanediol was not detected in control or treated animals, consistent with its metabolism to DHT and androsterone. The results demonstrate intraprostatic tumoral metabolism of androstanediol to DHT may contribute to AR transactivation and prostate cancer progression.

The metabolic processes that contribute to CR-CaP appear to be diverse, as reflected by an apparent wide variation in mechanisms and tissue androgen levels. On the other hand, AR continues to be a focus for new drug development to achieve sustained inhibition or eradication of prostate cancer growth during prolonged androgen deprivation therapy. AR activity depends on high affinity binding of testosterone or DHT, whose steady state levels are regulated by a family of reductive and oxidative enzymes (Table 1). The major circulating androgen testosterone synthesized in the testis from 4-androsten-3,17-dione (androstenedione) by reductive 17β-HSD (25; Fig. 3, Table 1) is irreversibly converted to DHT by type 2 steroid 5α-reductase (20) that also converts androstenedione, a major adrenal androgen, to 5α-androstane-3,17-dione. Androstanedione is reversibly converted to DHT by 17β-HSD, or indirectly through androsterone and androstanediol (26). DHT is reversibly inactivated by conversion to androstanediol by reductive 3α-HSD activity (21, 27) of type 3 3α-HSD aldo-keto reductase 1C2 (AKR1C2) that functions as both 3- and 17-ketosteroid reductase (19, 28–30). The 17β-HSD activity of aldo-keto reductase 1C3 (AKR1C3) converts androstanediol to androsterone, and androstenedione to testosterone (28, 31). Increased levels of AKR1C ketosteroid reductases were associated with CR-CaP (32). Androstanediol is metabolized by oxidative 17β-HSD activity of 17β-HSD6 and 17β-HSD11 to androsterone, which is glucuronidated and excreted (20–22).

In some prostate cancers, gain-of-function AR mutations may account for androstanediol activation of AR and increased cell growth (4, 33). However, most prostate cancers retain wild-type AR that is transcriptionally inactive when bound to androstanediol (15). Alternative mechanisms for prostate cancer tissue DHT production during androgen deprivation therapy are supported by the lower levels of DHT in prostate after treatment with the type 2 5α-reductase inhibitor, finasteride (34), although some evidence suggests an increase in 5α-reductase type 1 activity (35). Treatment with the dual type 1 and type 2 5α-reductase inhibitor dutasteride decreased prostate tissue DHT from 11.5 to 0.8 nmol/L, with a compensatory increase from 0.2 to 11.8 nmol/L testosterone (36).

The biological significance of the back conversion of androstanediol to DHT is supported by androstanediol-induced prostate growth in beagle dogs (37), conversion of androstanediol to DHT in humans (38), and androstanediol-induced masculinization of the tammar wallaby (39, 40). Conversion of androstanediol to DHT has been attributed to several oxidative 3α-HSD enzymes. These enzymes include retinol dehydrogenase 4 (RoDH4) and RDH5 (18), 17β-HSD6 (18, 19, 41) and 10 (54), and DHRS9 (19, 42; Table 1). These hydroxysteroid dehydrogenases oxidize or reduce the 3 or 17 ketone group of androgens, with a preference for oxidization of the 3 ketone that contributes to androstanediol oxidation to androsterone. 17β-HSD6 functions primarily as a 3α-hydroxysteroid oxidoreductase that converts androstanediol to DHT, and lacks stereospecificity in the reductive direction (17). 17β-HSD6 is expressed predominantly in stromal cells at higher levels than RoDH4 or RDH5, and is considered a predominant enzyme in androstanediol conversion to DHT in human prostate (19, 29). A mitochondrial 17β-HSD10 converts androstanediol to DHT in prostate tissue (Table 1) (43). Androstanediol also is acted upon by glucuronosyltransferase for rapid excretion of androstanediol glucuronide as observed in LNCaP cells (16).

Our studies have shown that benign prostate and prostate cancer cells metabolize androstanediol to DHT. Administration of androstanediol dipropionate at the CWR22R tumor site caused a 28-fold increase in DHT. Androstanediol also promoted the androgen-dependent growth of LAPC-4 cells. Since androstanediol was nearly equipotent with DHT in AR transcriptional activation in most of the prostate cell lines, the results indicate that conversion of androstanediol to DHT was sufficient to activate AR.

Higher levels of DHT and lower levels of androsterone correlated in most cell lines with greater AR transactivation in the presence of androstanediol. Quantitative mRNA measurements and enzyme protein levels suggested 17β-HSD6 is a predominant multifunctional enzyme in the oxidation and reduction of 3 and 17 keto groups, and levels correlated directly with the extent of androstanediol back-conversion to DHT and AR transactivation in the presence of androstanediol.

The ability of benign and malignant prostate cells to convert androstanediol to DHT provides a mechanism for increased AR transactivation in response to adrenal-derived androgen precursors (39). De novo synthesis of DHT from progesterone is an alternative pathway involved in environmental androgen production (44) and in CR-CaP (10). During androgen deprivation therapy, androstenedione is a major circulating adrenal androgen converted to DHT through testosterone. Androstanediol is not a major adrenal androgen. However, 5α-reductase type 1 is expressed in rat adrenal and levels increased after androgen withdrawal (45). This raises the possibility that 5α-reductase type 1 can convert adrenal 17α-hydroxyprogesterone to 17α-hydroxy-dihydroprogesterone, an androstanediol precursor (46). Adrenal androstanediol also may derive from the metabolism of androstenedione to androstanedione by 5α-reductase type 1, the conversion of androstanedione to androsterone, and androsterone to androstanediol by AKR1C3 (32). Increased availability of adrenal androstanediol in patients undergoing androgen deprivation therapy could provide the substrate necessary for conversion of androstanediol to DHT by 17β-HSD6.

Enzyme activity depends ultimately on mRNA and protein levels. Measurements of 17β-HSD6 mRNA and protein levels in prostate cancer cell lines correlated well with AR transactivation in response to androstanediol, and with the conversion of androstanediol to DHT based on mass spectrometry measurements. However, hydroxysteroid dehydrogenase activities also depend on the levels of oxidized and reduced NAD/H and NADP/H (43).

Our studies suggest that 17β-HSD6 is one of several enzymes involved in the peripheral conversion of androstanediol to DHT in prostate. 17β-HSD6 mRNA levels declined 5- to 10-fold in the castration-recurrent CWR22 and clinical specimens compared with AS-CaP, and did not rebound with castration-recurrent tumor growth. The decline in 17β-HSD6 mRNA in prostate cancer tissue after castration is consistent with the ∼90% decline in DHT to levels that remain sufficient to activate AR (8, 9). DHT levels may also decrease because of less 5α-reduction (35). In contrast, DHRS9 and RDH5 mRNA levels increased transiently after castration in the CWR22 tumor, but were similar in the androgen-stimulated and castration-recurrent CWR22 xenografts. The results contrast with a recent report that suggested a slight increase in RDH5 mRNA levels in the LNCaP xenograft upon progression to castration-recurrent growth (10).

Unlike DHT, testosterone persists in the normal range in CR-CaP tissue (8, 9) from local tissue production and from adrenal androgen conversion, such as from dehydroepiandrosterone in stromal cells (47). Testosterone may serve a greater role in AR transactivation in CR-CaP growth through mechanisms that include increased levels of AR coactivators (6, 7) and, in uncommon cases, AR somatic mutants, such as AR-H874Y, that increase testosterone-dependent AR transcriptional activity to levels similar to DHT (1, 48). However, our studies suggest that testosterone was not a major metabolite of androstanediol metabolism to DHT.

The importance of local active androgen production in prostate cancer growth during androgen deprivation therapy has gained renewed attention. Prostate tissue DHT persists after castration at levels approximately 10% of normal, with higher levels of testosterone (8–10, 49–51). Steroid 5α-reductase levels increase in AS-BP and prostate cancer (52), and a shift from steroid 5α-reductase type 2 toward type 1 contributes to the conversion of testosterone to DHT in prostate cancer (35). The prevalence of prostate cancer progression during androgen deprivation therapy by medical or surgical castration suggests mechanisms independent of AR may contribute to prostate cancer growth. However, inhibition of prostate cancer growth by reducing AR levels indicates that AR is a critical mediator of castration-recurrent growth (2). In agreement with these findings, a phase I clinical trial using abiraterone acetate, a selective steroid 17α-hydroxylase (CPY17) inhibitor, reduced PSA levels (53), which suggests that ligand-activated AR contributes to CR-CaP growth. We have shown that prostate-derived cells and tissues express steroid metabolic enzymes important in DHT synthesis that maintain AR-dependent gene transcription to a greater extent than in cells from other organs. AR may be transcriptionally active in CR-CaP through local bioactivation of androstanediol to DHT by 17β-HSD6. The acquired capacity of prostate cancer cells to produce testosterone and DHT from androgen precursors and catabolites establishes an environment for AR stimulation of recurrent growth during androgen deprivation therapy.

No potential conflicts of interest are disclosed.

National Cancer Institute Center grant P01-CA77739 (J.L. Mohler and E.M. Wilson) and Cancer Center Support Grant to Roswell Park Cancer Institute CA16156 (J.L. Mohler), University of North Carolina at Chapel Hill Lineberger Cancer Center CA34026 (J.L. Mohler), and US Public Health Service grants HD16910 from the National Institute of Child Health and Human Development (E.M. Wilson) and by the Intramural Program of the NIH, National Institute of Environmental Health Sciences (Z01 ES5050167; K.B. Tomer).

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.