Identification of the Major Oxidative 3α-Hydroxysteroid Dehydrogenase in Human Prostate That Converts 5α-Androstane-3α,17β-diol to 5α-Dihydrotestosterone: A Potential Therapeutic Target for Androgen-Dependent Disease

Abstract

Androgen-dependent prostate diseases initially require 5α-dihydrotestosterone (DHT) for growth. The DHT product 5α-androstane-3α,17β-diol (3α-diol), is inactive at the androgen receptor (AR), but induces prostate growth, suggesting that an oxidative 3α-hydroxysteroid dehydrogenase (HSD) exists. Candidate enzymes that posses 3α-HSD activity are type 3 3α-HSD (AKR1C2), 11-cis retinol dehydrogenase (RODH 5), L-3-hydroxyacyl coenzyme A dehydrogenase , RODH like 3α-HSD (RL-HSD), novel type of human microsomal 3α-HSD, and retinol dehydrogenase 4 (RODH 4). In mammalian transfection studies all enzymes except AKR1C2 oxidized 3α-diol back to DHT where RODH 5, RODH 4, and RL-HSD were the most efficient. AKR1C2 catalyzed the reduction of DHT to 3α-diol, suggesting that its role is to eliminate DHT. Steady-state kinetic parameters indicated that RODH 4 and RL-HSD were high-affinity, low-capacity enzymes whereas RODH 5 was a low-affinity, high-capacity enzyme. AR-dependent reporter gene assays showed that RL-HSD, RODH 5, and RODH 4 shifted the dose-response curve for 3α-diol a 100-fold, yielding EC₅₀ values of 2.5 × 10⁻⁹m, 1.5 × 10⁻⁹m, and 1.0 × 10⁻⁹m, respectively, when compared with the empty vector (EC₅₀ = 1.9 × 10⁻⁷m). Real-time RT-PCR indicated that L-3-hydroxyacyl coenzyme A dehydrogenase and RL-HSD were expressed more than 15-fold higher compared with the other candidate oxidative enzymes in human prostate and that RL-HSD and AR were colocalized in primary prostate stromal cells. The data show that the major oxidative 3α-HSD in normal human prostate is RL-HSD and may be a new therapeutic target for treating prostate diseases.

ANDROGENS ARE ESSENTIAL for the development and regulation of male sexual characteristics (1–6). Androgens exert their action by binding to the androgen receptor (AR), resulting in the trans-activation of androgen-responsive genes (7, 8). Consequently, androgen action is highly regulated, and its dysregulation can result in androgen-dependent prostate diseases, such as benign prostatic hyperplasia (BPH) and prostate adenocarcinoma (CaP). BPH affects approximately 50% of men by age 50, and its incidence increases with age (2, 4). CaP is the second leading cause of cancer-related deaths in men with approximately 184,000 new cases a year and approximately 32,000 related deaths a year (1, 9). Androgens are essential for the development of the two diseases, as prepubescent castrated male beagles never develop BPH or CaP (10–12), and androgen ablation can be a beneficial therapy in the treatment of these diseases.

5α-Dihydrotestosterone (DHT) is the most potent androgen and is responsible for the growth, development, and maintenance of the normal secretory function of the prostate (1, 2, 6, 13, 14). Within the prostate, DHT is formed from the irreversible reduction of testosterone by type 2 5α-reductase (13, 15, 16). Studies in rat (17–19), dog (11, 20–23), and marsupials (24, 25) show that DHT may also be formed from the inactive androgen 5α-androstane-3α,17β-diol (3α-diol) by an unknown oxidative 3α-hydroxysteroid dehydrogenase(s) (HSD) (Fig. 1).

Fig. 1.

Androgen Metabolism in Human Prostate The formation of DHT from either the reduction of testosterone by type 2 5α-reductase or by the oxidation of 3α-diol (back reaction) by an oxidative 3α-HSD are shown. One source of circulating 3α-diol is the hepatic metabolism of DHT by AKR1C4. Inhibition of type 2 5α-reductase with finasteride and the inhibition of the oxidative 3α-HSD is also indicated as therapeutic targets for prostate diseases. RED, Reductase; OX, oxidase.

Regulating the levels of DHT by hormonal ablative therapy is achieved either by surgical castration to remove the testis and hence remove its precursor testosterone or by directly blocking the intracrine formation of DHT within the prostate by targeting type 2 5α-reductase with the mechanism-based inactivator finasteride. Surgical castration reduces the prostate size by more than 80% and reduces prostate DHT levels by 90% (26, 27). However, this approach has undesired side effects that are associated with global changes in androgen levels, e.g. osteoporosis (28). Selective targeting of the type 2 5α-reductase by finasteride reduces both the volume and size of the prostate by approximately 25% and decreases prostate DHT levels by 80% with fewer side effects than castration (29–31). It is noteworthy that neither approach completely attenuates DHT levels in the prostate, suggesting that other sources of this potent androgen exist.

Other sources of DHT include reduction of testosterone by type 1 5α-reductase, which may be up-regulated in the diseased prostate (32), and oxidation of 3α-diol, which can potently stimulate the growth of prostate across species (19, 20, 23, 25). Because 3α-diol has a low affinity for the AR [dissociation constant (K_d) = 10⁻⁶m[;rsqb];, it was concluded that 3α-diol is converted back to DHT by an unidentified oxidative 3α-HSD. Furthermore, administration of 3α-diol, but not its epimer 5α-androstane-3β,17β-diol (3β-diol), resulted in the induction of prostate growth in castrated beagles (20, 23). Administration of [³H]3α-diol in human males suggests a human oxidative 3α-HSD exists because approximately 65% and 50% of the radioactivity was found to be [³H]DHT in the plasma and prostate, respectively (33, 34). These data suggest that the back reaction could be an important source of DHT in humans and a potential therapeutic target for treating prostate diseases.

In humans the candidate oxidative 3α-HSDs are all members of the short-chain dehydrogenase/reductase (SDR) family and include the following: 11-cis retinol dehydrogenase (RODH 5) (35, 36), L-3-hydroxyacyl coenzyme A dehydrogenase/type 10 17β-HSD [endoplasmic reticulum amyloid β-peptide binding protein (ERAB) (37, 38)], RODH like 3α-HSD also known as human oxidative 3α-HSD (RL-HSD) (39), novel type of human microsomal 3α-HSD (NT 3α-HSD) (40), and retinol dehydrogenase 4 (RODH 4) (41, 42) (Table 1). Recently, aldo-keto reductase (AKR) 1C2 was shown to reduce DHT to 3α-diol but was unable to oxidize 3α-diol back to DHT in transfected cells (9, 43). The ability of the other candidates to oxidize 3α-diol to produce sufficient DHT to activate AR-dependent gene transcription has not been compared.

Table 1.

Candidate Oxidative 3α-HSDs Assigned Names for this Study with Other Names with GenBank Accession Numbers and Formal Gene Names

Name Assigned in this Study	100% Nucleotide Identity with GenBank Identification No.	Formal Gene Name
AKR1C2	Aldo-keto reductase family 1, member C2 (BC063574)	AKR1C2
	Dihydrodiol dehydrogenase 2 (AB021654)
	Bile acid binding protein (NM_205845)
	Type 3 3α-HSD (NM_001354)
ERAB	17β-HSD type 10 (AF035555)	HADH2
	Short chain L-3-hydroxyacyl-CoA dehydrogenase (NM_004493)	HADSC
	Endoplasmic reticulum amyloid β-peptide-binding protein (U96132)
RL-HSD	Human oxidative 3α-HSD (U89281)	HSD17B6
	Homo sapiens 3-hydroxysteroid epimerase (AF223225)
	RODH-like 3α-HSD
RODH 5	Human 11-cis retinol dehydrogenase (U43559)	RDH5
	Retinol dehydrogenase 5 (NM_002905)
NT 3α-HSD	Homo sapiens 3α-HSD (AF343729)	DHRS9
RODH 4	Homo sapiens sterol/retinol dehydrogenase (AF057034)	RODH-4
	Homo sapiens retinol dehydrogenase 16 (NM_003708)

We report the identification of the major oxidative 3α-HSD in human prostate as RL-HSD based on activity assays, transfection studies, trans-activation of the AR, and expression levels in human prostate. Moreover, RL-HSD was shown to be colocalized with the AR in primary prostate stromal cells, suggesting that it can regulate androgen signaling in this cell type.

RESULTS

Experimental System to Monitor 3α-Diol Oxidation by 3α-HSDs in Transfected Cells

The cDNAs that code for five candidate SDRs previously shown to convert 3α-diol to DHT in transient transfection studies were obtained by RT-PCR from human liver RNA. The PCR products were subsequently cloned into either pcDNA3 or into pcDNA3-LacZ to yield bicistronic constructs. Each construct was transfected into either COS-1 cells or PC-3 cells, and the ability of the transfected cells to convert 3α-diol to DHT was measured. COS-1 cells (44) were selected as a null environment for androgen metabolism, whereas PC-3 cells (45) were chosen because they are a prototypic androgen-metabolizing prostate cell line. The oxidative activity of the 3α-HSDs to convert 3α-diol to DHT was determined using a double transfection (pcDNA3–3α-HSD and pCMV-β-galactosidase) or a single transfection of a bicistronic construct (pcDNA3–3α-HSD-LacZ).

At the commencement of these studies we determined that the steroid formed by the oxidation of 3α-diol by the 3α-HSDs was, in fact, DHT. In transfection studies the product formed comigrated with an authenticated DHT standard by thin-layer chromatography (TLC) (Fig. 2). The reaction was replicated with unlabeled 3α-diol, and the resulting material was isolated and identified by liquid chromatography/mass spectrometry (LC/MS). Commercially available DHT was also examined by LC/MS, and the reaction product and the standard gave a single chromatographic peak with a retention time of 15.42 min. The mass spectrum for both species was identical and gave a dominant molecular ion [M⁺NH₄-H₂O]⁺ at m/z 290.44 (Fig. 2) where the predicted molecular ion m/z = 291.44 [MH⁺].

Fig. 2.

Representative 3α-Diol Metabolism in Transfected Cells and Validation of DHT as the Product A, Androgen standards separated by TLC run three times using methylene chloride/ethyl ether [11:1 (vol/vol). B, Representative analysis of metabolic profiles using the automatic TLC-linear analyzer. Standard [³H]3α-diol (1 ); COS-1 cells transiently transfected with RODH 5 using 5 μm [³H]-3α-diol, 0.25 h (2 ) and 2 h (3 ); standard [¹⁴C]DHT (4 ). C, LC/MS/MS tandem mass spectrometry of DHT (m/z = 290.44, [M+NH₄-H₂O]+, predicted MH+ m/z = 291.44) and the unknown product predicted to be DHT (m/z = 290.44, [M+NH₄-H₂O]+, predicted MH+ m/z = 291.44)]. a, Androstanedione; b, DHT, c, androsterone, d, 3β-diol; e, 3α-diol; f, polar metabolites.

Representative conversion of 5 μm [³H]3α-diol to DHT by COS-1 cells transiently transfected with RODH 5 is also shown over time (0.25 h and 2 h) (Fig. 2). Because 3α-diol has two functional substituents (a 3α-hydroxy and a 17β-hydroxy group), oxidation of the 3α-position was anticipated to yield DHT. Over time, the initial product was converted to a second product that comigrated with 5α-androstane-3,17-dione (Adione). Oxidation of DHT by an endogenous oxidative 17β-HSD activity would be responsible for this second product.

Oxidation of 3α-Diol to DHT in Cells Transiently Transfected with SDRs

The metabolism of 3α-diol in COS-1- and PC-3-transfected cells was compared using both low and high substrate concentrations as previously reported (35, 43). The resulting activities were normalized either to β-galactosidase by a double-transfection procedure, whereby both the pcDNA3–3α-HSD and pCMV-β-galactosidase were cotransfected, or by a single transfection of a bicistronic construct, whereby both the 3α-HSD and β-galactosidase were expressed under the control of the same cytomegalovirus promoter. This approach was taken because antibodies were unattainable for all the enzymes used in the study. The β-galactosidase values were used to normalize the activity of the enzyme as a means to better control for transfection efficiency. The variation of the β-galactosidase activity for the double transfection was as much as 2.5-fold; however the variation of β-galactosidase activity using the bicistronic construct was only 1.25-fold. Consequently, the bicistronic construct gave a more reliable representation of transfection efficiency, but at the cost of expression level. Both the expression of the 3α-HSD (as indicated by real-time PCR) and the β-galactosidase (as indicated by activity measurements) were decreased when the bicistronic construct was transfected. This is probably due to the processing of the long mRNA and its subsequent translation. Despite these differences, similar patterns for DHT formation were seen in COS-1 and PC-3 cells irrespective of the transfection protocol.

Transiently transfected cells were analyzed for the formation of DHT using a low concentration (0.1 μm) of 3α-diol, and the activity was normalized to β-galactosidase (Fig. 3). The formation of DHT was highest for RODH 5, RODH 4, and RL-HSD for both the double-transfection and the single-transfection procedures. Due to the high endogenous 17β-HSD activity of the cells, DHT formed from 0.1 μm 3α-diol by the transfected 3α-HSDs was quickly converted to Adione. Consequently, a more representative picture of the 3α-HSD oxidase activity is obtained by combining the DHT and Adione formed as a function of time (Fig. 3). The metabolic profiles indicated that ERAB and NT 3α-HSD were poor oxidases in comparison with the other candidate enzymes because they were only able to produce trace amounts of DHT when a high concentration (5 μm) of 3α-diol was incubated for longer times (data not shown).

Fig. 3.

3α-Diol Metabolism in PC-3 and COS-1 Cells A, Normalized formation of DHT in PC-3 cells following the double-transfection protocol (pcDNA3–3α-HSD plus pCMV-β-galactosidase) using 0.1 μm 3α-diol. B, Normalized formation of DHT and Adione in PC-3 cells using the double-transfection protocol using 0.1 μm 3α-diol. C, Normalized formation of DHT in COS-1 cells by a single transfection of the bicistronic construct (pcDNA3–3α-HSD-Lac Z) using 0.1 μm 3α-diol. D, Normalized formation of both DHT and Adione in COS-1 cells by a single-transfection protocol using 0.1 μm 3α-diol. Normalized activity is expressed as percent of total DHT or DHT + Adione formed divided by milliunits of β-galactosidase with error bars representing the highest and lowest values with the average of three independent experiments (see Materials and Methods). β-Gal, β-Galactosidase; NT, no transfection; NT 3α-HSD, novel type 3α-HSD.

RL-HSD was also found to act as an epimerase converting 3α-diol first to DHT and then reducing DHT back to 3β-diol, as previously reported (46). However, this activity was observed only when high substrate concentrations (5 μm 3α-diol or 5 μm DHT) were incubated and was not observed when a low substrate concentration (0.1 μm 3α-diol) was used (data not shown). The metabolism profiles indicated that three candidates (RODH 5, RODH 4, and RL-HSD) could be responsible for the conversion of physiological concentrations of 3α-diol back to DHT in the human prostate. Of all the enzymes studied, only AKR1C2 was unable to convert 3α-diol to DHT in vivo at all concentrations tested, and this confirmed previous findings (9, 43).

Reduction of DHT to 3α-Diol in Cells Transiently Transfected with SDRs

The steroid specificity and preferred directionality of the 3α-HSDs was investigated by also determining their ability to reduce 5 μm DHT in COS-1 and PC-3 cells. Normalized formation of 3α-diol by transiently transfected COS-1 and PC-3 cells is shown in Fig. 4. Our results indicated that AKR1C2 acted as a robust reductase as it catalyzed the reduction of DHT to 3α-diol. Under these conditions the 3α-HSDs were unable to reduce DHT to 3α-diol; however, substantially lower 3α-diol was present in the RODH 5-, NT 3α-HSD-, RODH 4-, and RL-HSD-transfected cells as compared with the no-transfected or the pcDNA3 (empty vector)-transfected controls. These differences can be explained by other activities of these enzymes, which are revealed when the high endogenous 17β-HSD activity is suppressed. It has been noted previously that high steroid concentrations will suppress the 17β-HSD activity in these recipient cells (43). For example, at 5 μm DHT, transiently transfected NT 3α-HSD and RODH 4 exhibited oxidative 17β-HSD activity whereby DHT and the 3α-diol formed were now converted to Adione and androsterone (3α-hydroxy-5α-androstan-17-one), respectively. On the other hand, transiently transfected RL-HSD exhibits reductive 3β-HSD activity at the higher substrate concentrations and catalyzed the conversion of DHT to 3β-diol. Although these additional activities were observed (17β-HSD oxidation and 3β-HSD reduction), they are minor in comparison with their 3α-HSD oxidative activity, because they only occur over a much longer time frame. In contrast, RODH 5 is 3α-HSD specific because no other detectible activities were observed. Thus, the decreased level of 3α-diol observed with RODH 5 was solely due to the ability of this enzyme to oxidize the endogenously formed 3α-diol back to DHT (Fig. 4).

Fig. 4.

DHT Metabolism in COS-1 and PC-3 Cells Normalized formation of 3α-diol in COS-1 (A) and PC-3 (B) cells following a double-transfection (pcDNA3 and pCMV-β-galactosidase) protocol using 5 μm DHT. Normalized activity is expressed as percent of total 3α-diol formed divided by milliunits of β-galactosidase with error bars representing the highest and lowest values with the average of three independent experiments. β-Gal, β-Galactosidase; NT, no transfection; NT 3α-HSD, novel type 3α-HSD.

Steady-State Kinetic Analysis for the Conversion of 3α-Diol to DHT

Based on the transfection studies, RODH 5, RL-HSD, and RODH 4 were all capable of converting 3α-diol to DHT. To identify the most efficient enzyme, the steady-state kinetic parameters (V_maxapp, K_m, and V_maxapp/K_m) were determined for the transfected enzymes using isolated membrane fractions. It was found that RODH 4 and RL-HSD have similar kinetic constants and were identified as being high-affinity (submicromolar K_m values), low-capacity enzymes (low V_maxapp values); however, RODH 5 was found to be a low-affinity (micromolar K_m value) and high-capacity enzyme (high V_maxapp value) (Table 2). The steady-state kinetic analysis indicated that RODH 4 and RL-HSD had much higher utilization ratios (7- to 16-fold greater) than RODH 5. Our steady-state kinetic parameters for RODH 4, RL-HSD, and RODH 5 were similar to those previously reported (36, 39, 42).

Table 2.

Determination of the Steady-State Kinetic Parameters for 3α-Diol Oxidation Catalyzed by 3α-HSDs Using Isolated Membrane Fractions from Transiently Transfected COS-1 Cells

Oxidative 3α-HSD	Km (μm)a	Vmaxapp (nmol/min/mg protein)a	Vmaxapp/Km (nmol/min/mg protein)/(μm)a
RL-HSD	0.4 ± 0.04	5.9 ± 0.26	14.8
RODH 5	18.5 ± 4.1	39.6 ± 3.3	2.1
RODH 4	0.27 ± 0.03	8.74 ± 0.19	32.4

^a

Kinetic parameters were determined from initial velocity measurements determined in triplicate, and the data were pooled from two experiments.

Trans-Activation of the AR by 3α-Diol in Cells Transfected with SDRs

The ability of the oxidative 3α-HSDs to convert 3α-diol into sufficient amounts of DHT to trans-activate the AR was assessed using a reporter gene assay. The chloramphenicol acetyl transferase (CAT) reporter gene assay was performed using the appropriate controls to ensure that the response was mediated through the trans-activation of the AR. The specificity of the CAT assay was determined by using a nontransfected control, a dimethylsulfoxide (DMSO) control, a no-steroid control, an AR minus control, a p-tk-CAT [minus the androgen response element (ARE) tandem repeat] control, and a pbasic-CAT (minus the tk promoter) control, and the results are shown in Fig. 5.

Fig. 5.

Trans-Activation of the AR by 3α-Diol in the Presence of Oxidative 3α-HSDs A, The specificity of the CAT assay was validated by comparing the no-transfected control with the positive control [pCMV-AR, p-(ARE)₂-tk-CAT, pCMV-β-galactosidase and pcDNA3] and with all other possible combinations. All the systems tested included a no-steroid and steroid (DHT; 1.0 × 10⁻⁷m) treatment. B, Activation of the (ARE)₂-tk-CAT reporter gene by the AR in the presence of cotransfected HSDs vs. the concentration of 3α-diol was varied (10⁻¹² to 10⁻⁶m). Activation of the reporter gene is the average of three independent experiments with the activity expressed as percent of CAT activity (fold activation of the CAT divided by the maximal fold activation multiplied by 100%). C, Inhibition of the CAT response mediated by oxidative 3α-HSDs and 3α-diol using flutamide. The concentration of 3α-diol used was the EC₅₀ value in each case. The black triangle represents increasing flutamide concentrations (0.01 μm, 0.1 μm, 0.3 μm, 1 μm, 3 μm, and 10 μm; not shown, 0.001 μm flutamide) with the first value representing no-flutamide treatment as indicated by a zero and the highest concentration (10 μm) shown. Inhibition of the reporter gene is the average of three independent experiments with the activity expressed as percent of CAT activity (fold activation of the CAT divided by the maximal fold activation multiplied by 100%). NT HSD, Novel-type 3α-HSD.

Activation of the CAT reporter gene occurred only in the presence of androgen when the AR and the (ARE)₂-tk-CAT constructs were cotransfected, indicating that the CAT assay was specific to monitor trans-activation of the AR. Furthermore, the activation of CAT by DHT, testosterone, and 3α-diol was inhibited with the AR antagonist flutamide. Consequently, the CAT assay we developed could be used to determine the ability of the oxidative 3α-HSDs to modulate gene transcription by trans-activation of the AR using 3α-diol as the substrate. Activation of the (ARE)₂-tk-CAT construct by the oxidative 3α-HSDs using 3α-diol is shown in Fig. 5 and in Table 3. RODH 4, RODH 5, RL-HSD, and NT 3α-HSD were all able to mediate the trans-activation of the reporter gene construct at low levels of 3α-diol yielding EC₅₀ values of 1.0 × 10⁻⁹m, 1.5 × 10⁻⁹m, 2.5 × 10⁻⁹m, and 5.5 × 10⁻⁸m, respectively. In contrast, 3α-diol gave an EC₅₀ value for trans-activation of the AR equal to 1.9 × 10⁻⁷m in pcDNA3 (empty vector)-transfected cells. Thus RODH4, RODH 5, and RL-HSD each increased the potency of 3α-diol for the AR by greater than 100-fold. From the metabolism data it is not surprising that ERAB (EC₅₀ = 2.1 × 10⁻⁷m) and AKR1C2 (EC₅₀ = 1.9 × 10⁻⁷m) were unable to trans-activate the reporter gene construct with 3α-diol. Using the EC₅₀ values for 3α-diol activation of the reporter gene, the CAT reaction was subsequently inhibited with the AR antagonist flutamide for pcDNA3, RL-HSD, RODH 5, RODH 4 (Fig. 5), and NT-3α-HSD (data not shown). The inhibition of the CAT response by flutamide produced identical IC₅₀ values for all the constructs tested, suggesting that the response was indeed AR dependent and not construct dependent (pcDNA3 using 3α-diol, IC₅₀ = 3.5 × 10⁻⁷m; pcDNA3 using DHT, IC₅₀ = 3.5 x 10⁻⁷m; pcDNA3 using testosterone, IC₅₀ = 4.6 × 10⁻⁷m; RL-HSD, IC₅₀ = 3.6 × 10⁻⁷m; RODH 5, IC₅₀ = 4.5 × 10⁻⁷m; RODH 4, IC₅₀ = 2.8 × 10⁻⁷m; NT 3α-HSD, IC₅₀ = 3.4 ×10⁻⁷m).

Table 3.

EC₅₀ Values for Trans-Activation of the AR Using a CAT Reporter Gene Assay

Steroid	Construct	EC50 Value (m)a
3α-diol	pcDNA3	1.9 × 10−7
	AKR1C2	1.9 × 10−7
	ERAB	2.1 × 10−7
	RL-HSD	2.6 × 10−9
	RODH 5	1.5 × 10−9
	NT 3α-HSD	5.5 × 10−8
	RODH 4	1.0 × 10−9
T	pcDNA3	2.5 × 10−9
DHT	pcDNA3	6.8 × 10−10

^a

Activation of the (ARE)₂-tk-CAT reporter gene by the oxidative 3α-HSDs in COS-1 cells are the average of three independent experiments. T, Testosterone.

The activation of the (ARE)₂-tk-CAT reporter by 3α-diol using transfected RODH 5, RL-HSD, and RODH 4 was compared with that observed with DHT, testosterone, and 3α-diol (Table 3). The ability of RODH 5, RL-HSD, and RODH 4 to activate the CAT reporter using 3α-diol gave a dose-response curve equivalent to that observed with testosterone (EC₅₀ = 2.5 × 10⁻⁹m), but less than the response observed with DHT (EC₅₀ = 6.8 × 10⁻¹⁰m). These differences can be explained because at low concentrations (submicromolar), both 3α-diol and DHT are quickly metabolized to the inactive androgens, androsterone and androstanedione, respectively, by a very high oxidative 17β-HSD activity. Taken together, RODH 5, RL-HSD, and RODH 4 are able to alter gene transcription by oxidizing low levels of 3α-diol to DHT when compared with the pcDNA3 control.

Expression Levels of the SDRs in Human Prostate and Cell Type Using Real-Time RT-PCR

The metabolism studies, kinetic parameters, and trans-activation experiments indicate that three enzymes (RODH 5, RODH 4, and RL-HSD) may be responsible for the formation of DHT from 3α-diol in the human prostate. Consequently, a real-time PCR method was developed to investigate the expression levels of the 3α-HSD oxidases in normal human prostate. The development and specificity of the real-time PCR assay included 1) the identification of primers that amplified the desired gene, which was validated by sequencing the PCR product; 2) placement of the primers to cross over exon-intron boundaries to prevent the nonspecific amplification of genomic DNA; and 3) melting curves were analyzed at the end of each run to ensure specificity between reactions. This RT-PCR methods were linear (> r = 0.995) over a dynamic range (10⁹) as determined by plotting the log₁₀ fluorescence intensity vs. the number of cycles and could be used to determine variable expression levels of the SDRs within the prostate. The expression levels of the SDRs were determined using total RNA pooled from 32 Caucasian human prostates and normalized to the high-abundance housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and the low abundance PBGD, and similar patterns were observed. The real-time PCR data indicated that all the candidate oxidative 3α-HSDs and the AR were expressed within whole prostate, but to varying levels (Fig. 6). The expression patterns revealed that AR was the highest expressed. Of the oxidative 3α-HSDs, ERAB and RL-HSD were expressed more than 15-fold higher in comparison with the other candidates and would therefore be the most likely candidates to be the major oxidative 3α-HSD in human prostate. However, from the metabolism studies, ERAB was unable to oxidize low levels of 3α-diol and was unable to alter gene transcription using the CAT assay. Consequently, the major oxidative 3α-HSD in normal human prostate is identified as RL-HSD.

Fig. 6.

Relative Expression of Oxidative 3α-HSDs in Human Prostate and Cell Type as Determined by Real-Time RT-PCR Normalized to PBDG A, The expression of the oxidative HSDs in normal human prostate. Total RNA (1 μg) from 32 pooled human prostates was reverse-transcribed to cDNA, and 12.5 ng of cDNA was added to each real-time PCR experiment that was performed in triplicate with the mean shown. Data are normalized to the housekeeping gene PBGD and are represented as expressed femtograms of each protein per ng of total cDNA. B, The expression of the oxidative HSDs in primary prostate epithelial and stromal cells show a cell type-specific pattern. Total RNA (1 μg) from primary prostate epithelial (n = 14) and stromal (n = 15) cells was reverse-transcribed to cDNA, and 50 ng of cDNA was added to each real-time PCR experiment that was performed in triplicate with the median shown. NT HSD, Novel type 3α-HSD.

Although RL-HSD was expressed in prostate, its abundance may be suppressed by examining the whole gland. Subsequently, the expression levels of the oxidative 3α-HSDs were analyzed in cultured human prostate primary epithelial and stromal cells to determine their cellular localization. Furthermore, because AR is prominently expressed in the stromal cells of the prostate (47), it was important to determine whether the SDRs are colocalized with the AR so that they can regulate the androgen signal. Total RNA was isolated from normal prostate primary epithelial (n = 14) and primary stromal cells (n = 15), and 1 μg was reverse transcribed. The results indicated that the enzymes displayed a cell type-specific distribution (Fig. 6). ERAB was most highly expressed in both cell types in comparison with the other oxidative candidates with an approximate 3-fold preference for the epithelial cells, but it is a weak oxidase in comparison with the other candidate enzymes. RL-HSD was expressed in the stromal cells with an approximate 20-fold preference. The AR was expressed in both epithelial and stromal cells with an approximate 10-fold preference for stromal cells as previously shown (47). The stromal colocalization of RL-HSD and the AR indicated that RL-HSD is positioned to regulate the trans-activation of the AR. This is important because changes in the expression levels of RL-HSD could lead to an increase in activation of androgen-sensitive genes, due to the stromal colocalization of the AR. RODH 5 was expressed equally in epithelial and stromal cells, but its expression level was 5-fold lower in the stromal cells as compared with RL-HSD. NT 3α-HSD was expressed only in the epithelial cells; however, it poorly metabolizes 3α-diol and only weakly trans-activated the AR with 3α-diol. RODH 4 was very lowly expressed in both epithelial and stromal cells, with a slight preference for the epithelial cells.

DISCUSSION

For more than 30 yr, animal experiments (rat, beagles, and marsupials) have indicated that 3α-diol will promote growth of the prostate, yet it has negligible affinity for the AR. It was assumed that the growth produced was accomplished by the oxidation of 3α-diol to DHT; however, the identity of the oxidative 3α-HSD has remained elusive. In the literature five SDRs have been implicated by individual investigators as being involved in this reaction, leading to confusion as to its identity. In this study we have compared all candidate oxidative enzymes in a single study and identified RL-HSD as being the major oxidative 3α-HSD in human prostate.

All the oxidative 3α-HSD candidate enzymes were able to oxidize 3α-diol to DHT except for AKR1C2. The 3α-diol metabolic profiles showed that three enzymes (RODH 5, RODH 4, and RL-HSD) could be responsible for the back reaction. In contrast, ERAB and NT 3α-HSD were only able to convert 3α-diol to DHT at a high concentration (5 μm) and over an extended time course. These findings are supported by a steady-state kinetic analysis of the transfected enzymes. Thus, RL-HSD and RODH 4 had the highest utilization ratios (V_maxapp/K_m). The metabolism studies also indicated that RL-HSD was able to act as an epimerase by converting 3α-diol to DHT and then back to 3β-diol, but only at higher substrate concentrations (5 μm 3α-diol or 5 μm DHT). Irrespective of this epimerase activity, RL-HSD was able to regulate gene transcription when low concentrations of 3α-diol were incubated.

The ability of the oxidative 3α-HSDs to reduce DHT to 3α-diol was also investigated. Our data confirmed that AKR1C2 was the only enzyme that could reduce DHT to 3α-diol (9, 43). By contrast, in the RODH 5-, RL-HSD-, RODH 4-, and NT 3α-HSD-transfected cells the amount of 3α-diol formed was significantly lower; this is due to the minor 17β-HSD activity of NT 3α-HSD and RODH 4, which converts DHT to Adione, the minor reductive 3β-HSD activity of RL-HSD that converts DHT to 3β-diol, and the ability of RODH 5 to oxidize 3α-diol formed endogenously back to DHT.

We also investigated the ability of the oxidative 3α-HSDs to convert sufficient 3α-diol to DHT to trans-activate the AR. Using a reporter gene assay we found that RODH 5, RL-HSD, and RODH 4 were able to shift the dose-response curve of 3α-diol to the left by 2 orders of magnitude as compared with the pcDNA3 control. This increased activation was blocked with flutamide, indicating that this reaction was AR dependent. The significant finding is that the oxidative 3α-HSDs altered gene transcription at the prereceptor level by changing the concentration of active androgen available to the AR.

The mRNA expression from normal human prostate indicated that all the oxidative 3α-HSD candidates were expressed; however, RL-HSD was expressed more than 15-fold as compared with the other two remaining candidates (RODH 5 and RODH 4). Consequently, the major oxidase was identified as RL-HSD in normal human prostate. Furthermore, RL-HSD was shown to be colocalized with the AR in primary prostate stromal cells; thus RL-HSD could regulate androgen-sensitive genes by oxidizing physiological concentrations of 3α-diol back to DHT in this cell type.

Our data show that the major enzyme responsible for the elimination of DHT in the prostate is likely AKR1C2. Because the major oxidative 3α-HSD responsible for DHT formation from 3α-diol is RL-HSD, we have identified the HSD pair that modulates ligand access to the AR in the human prostate (Fig. 7).

Fig. 7.

Regulation of the AR by 3α-HSDs within the Prostate The reduction of DHT by AKR1C2 and the oxidation of 3α-diol by RL-3α-HSD regulate ligand access to the AR. AKR1C2 is highly expressed in human prostate cancer epithelial cells (43 ).

RL-HSD was identified as the major oxidative 3α-HSD in normal human prostate, but Northern analysis indicates that RL-HSD is not ubiquitously expressed (48). Because RODH 5, RL-HSD, or RODH 4 can form DHT in transfected cells, the expression levels of these enzymes need to be examined in other androgen-dependent diseases, like acne or alopecia. Excess DHT is implicated in acne and alopecia, and RODH 4 and RODH 5 may be important in regulating DHT levels in the epidermis and sebaceous glands. For example, Jurukovski et al. (41) showed that RODH 4 was expressed in the epidermis and Karlsson et al. (49) showed that it was potently inhibited by 13-cis-retinoic acid, a clinically relevant treatment for acne. The authors suggested that the inhibition of acne by 13-cis-retinoic acid could be due to the inhibition of RODH 4 and a local decrease of DHT where DHT increases the differentiation and growth of sebaceous glands. New approaches in the treatment of androgen-dependent prostate diseases, acne, and/or alopecia may be to inhibit RL-HSD, RODH 4, or RODH 5 by decreasing the local production of DHT in these target tissues.

MATERIALS AND METHODS

Chemicals

The radioactive steroids were [4-¹⁴C]5α-DHT (57.3 mCi/μmol, PerkinElmer LLC, Norwalk, CT) and [9,11-³H(N)]5α-androstane-3α,17β-diol (40.0 Ci/μmol, PerkinElmer LLC). The following nonradioactive steroids used in the study were all purchased from Steraloids (Wilton, NH): 3α-diol, 3β-diol, androsterone, Adione, and DHT

Construction of the pcDNA3 Constructs and the Bicistronic Constructs

The enzymes investigated were all cloned from human liver total RNA (BD Bioscience, Palo Alto, CA). Total RNA (1μg) was reverse transcribed using GeneAmp RNA PCR Kit (Applied Biosystems, Foster City, CA) and the cDNA was amplified for each of the genes of interest using primers as previously reported [AKR1C2 (43, 51), ERAB (38), RL-HSD (50), RODH 5 (35), NT 3α-HSD (40), and RODH 4 (42)]. Each cDNA was successfully subcloned into the mammalian expression vector pcDNA3 using restriction sites previously reported (for AKR1C2, KpnI and ApaI (43); for ERAB, BamHI and EcoRI (38); for RL-HSD, EcoRI (50); for RODH, 5 BamHI and EcoRI (35); for NT 3α-HSD, EcoRI (40); and for RODH 4, EcoRI (42)], and its identity was validated by dideoxy sequencing. The bicistronic construct, pIRES (internal ribosome entry sequence)-Lac Z, was purchased from BCCM/LMBP (Ghent University, Belgium), and subsequently ligated to pcDNA3 to form a pcDNA3-pIRES-LacZ construct (pcDNA3-LacZ) using EcoRI and XhoI. The construct was validated by dideoxy sequencing, and after transfection the expression of β-galactosidase activity was determined using the β-galactosidase enzyme assay system (Promega Corp., Madison, WI). The 3α-HSD enzymes were subcloned from pcDNA3 into the bicistronic construct (pcDNA-3α-HSD-LacZ) using the same restriction enzymes that produced the respective pcDNA3–3α-HSD construct, and the final constructs were sequenced. The constructs used in the reporter gene assays were p-(ARE)₂-tk-CAT, p-tk-CAT, and the pbasic-CAT (52); the pCMV-β-galactosidase was purchased from CLONTECH Laboratories, Inc. (Palo Alto, CA).

Cell Culture

The cell lines (COS-1 and PC-3) were purchased and maintained according to the protocols provided by the American Type Culture Collection (Manassas, VA). COS-1 cells were maintained in DMEM (Invitrogen Corp., Grand Island, NY), 10% fetal bovine serum (FBS), 1% penicillin/streptomycin, and 2% l-glutamine. COS-1 cells were plated in six-well dishes at a density of 2.5 × 10⁵ cells and were transfected using FuGENE6 (Roche Diagnostics, Indianapolis, IN) with 0.7 μg pcDNA3–3α-HSD, 0.7 μg pcDNA3, and 0.4 μg pCMV-β-galactosidase. The plating condition for COS-1 cells was held constant for transient transfection of the bicistronic construct; however, 1.0 μg of plasmid (pcDNA3–3α-HSD-LacZ) was transfected. Approximately 3 h before metabolism studies, the medium was changed to DMEM (minus phenol red) (Invitrogen Corp.), 1% charcoal/dextran-treated FBS (CDT-FBS) (HyClone Laboratories, Inc., Logan, UT), 1% penicillin/streptomycin, and 2% l-glutamine for COS-1 cells. PC-3 cells were maintained in FK-12 (Invitrogen Corp.), 10% FBS, 1% penicillin/streptomycin, and 1% l-glutamine. PC-3 cells were plated in six-well plates at a density of 5.0 × 10⁵ cells and were transfected using FuGENE6 with 1.0 μg of construct and 0.2 μg of β-galactosidase. The plating condition for COS-1 cells was held constant for transient transfection of the bicistronic construct; however, 2.0 μg of plasmid was transfected. Approximately 3 h before metabolism studies, the medium was changed to RPMI (minus phenol red) (Invitrogen Corp.), 1% CDT-FBS, 1% penicillin/streptomycin, and 1% l-glutamine for PC-3 cells.

Metabolism Studies Using Transiently Transfected COS-1 and PC-3 Cells

The steroid stock concentrations used were determined by weight and titrated by enzymatic conversion as previously published (53). To study the metabolism of 0.1 μm and 5 μm 3α-diol or 5 μm DHT in transiently transfected COS-1 and PC-3 cells, a mixture of radioactive and non-radioactive steroid was used containing 2,200,000 cpm of [³H]5α-androstane-3α,17β-diol or 40,000 cpm of [¹⁴C]dihydrotestosterone. The organic soluble steroids were dried down under nitrogen, redissolved in DMSO (Fisher Scientific, Pittsburgh, PA), and added to the cells to give a final concentration of 0.25% DMSO, which had no effect on cell viability. Aliquots (500 μl) of the culture media were removed over time and extracted twice using 1 ml of water-saturated ethyl acetate (>96% recovery). The ethyl acetate was evaporated to complete dryness using a Sorvall Speed Vacuum and redissolved in ethyl acetate-methanol-chloroform (1:0.5:0.5) (Fisher Scientific) containing reference steroids as previously reported (53). The dissolved steroids were plated on Whatman LK6D Silica TLC plates (Fisher Scientific), prerun twice, and developed three times using methylene chloride-diethyl ether [11:1 (vol/vol)]. The TLC plates were analyzed with an automatic TLC-linear analyzer (Bioscan Imaging Scanner System 200-IBM with an AutoChanger 3000; Bioscan, Inc., Washington, DC). The computer-aided software quantitatively determined the radio signals emitted from the TLC plates and calculated the percentages of each metabolic product vs. the total radioactivity. The positions of radioactive steroid signals on the TLC plates were verified by staining reference standards and quantified by scintillation counting as described previously (53).

Validation that DHT Is the Product of the Enzymatic Oxidation of 3α-Diol

To validate that DHT was formed from 3α-diol by the 3α-HSD oxidases, 5 μm unlabeled 3α-diol was added to RODH 5-transiently transfected COS-1 cells as described earlier. The reaction was terminated by removal of the medium after 4 h, a time by which more than 90% of 3α-diol would be converted to DHT. The medium (15 ml) was extracted three times with 50 ml of ethyl acetate, concentrated by rotary evaporation, and separated by TLC. Standards and a radioactive control were run on the sides of the plate to determine the location of DHT. The product was isolated, dissolved in ethyl acetate, and separated over a silica column. DHT was eluted using 10 ml of chloroform and ethyl acetate (1:1) and condensed to 1 ml by rotary evaporation. The resulting liquid was transferred into a small vial, dried under nitrogen, and dissolved in 200 μl of acetonitrile. The product (predicted to be DHT) was analyzed by LC/tandem mass spectrometry using a Thermo Finnigan LCQ ion trap mass spectrometer (Thermo Finnigan, San Jose, CA) equipped with an electrospray ionization source in positive ion mode. Operating conditions were as follows: capillary temperature at 250 C. Nitrogen was used as the sheath (60 psi) and auxiliary (4 arbitrary units) gas to assist with nebulization. Full scanning analyses were performed in the range of m/z 100 ∼ 400.

Chromatography for LC/MS experiments using a gradient system was performed using a Waters Alliance 2690 HPLC system (Waters Corp., Milford, MA). An YMC C18 octadecylsilyl silicon-AQ column (250 × 2.0 mm inner diameter, 3 μm; Waters, Milford, MA) was used with a flow rate of 0.15 ml/min. Solvent A was 5 mm ammonium acetate in water with 0.1% formic acid, and solvent B was 5 mm ammonium acetate in acetonitrile with 0.1% formic acid. The linear gradient was as follows: 50% B at 0 min, 50% B at 2 min, 90% B at 15 min, 90% B at 17 min, 50% B at 18 min, and 50% B at 25 min.

Formation of DHT Normalized to β-Galactosidase

Cells (1.0 × 10⁶) were lysed using reporter lysis buffer (Promega Corp., Madison, WI) at the 2-h time point, and the formation of DHT across the entire time course was normalized to the coexpressed β-galactosidase activity. The level of β-galactosidase was detected spectrophotometrically following the manufacturer’s protocol (Promega). In the protocol, 2× reaction β-galacosidase enzyme solution was made using 200 mm sodium phosphate (pH 7.3) (Fisher Scientific), 2 mm MgCl₂, 100 mm β-mercaptoethanol, and 4.4 mm ο-nitrophenyl-β-d-galactopyranoside (Sigma-Aldrich Corp, St. Louis, MO) as specified by Promega. The reaction was terminated using 1 m sodium carbonate, and the formation of the ο-nitro-phenolate anion at 420 nm was measured using the molar extinction coefficient (E = 6900 m⁻¹ cm⁻¹). Percent conversion of steroid substrate was then normalized to milliunits of β-galactosidase.

Isolation of the Oxidative 3α-HSDs from Transiently Transfected COS-1 Cells

COS-1 cells were plated at 2.5 × 10⁵ cells and transfected using 2 μg of cDNA (pcDNA3–3α-HSD). After 20 h the cells were washed twice with PBS and harvested using a Tris-HCl-sucrose buffer (pH 7.4) containing 50 mm Tris-HCl, 250 mm sucrose, 1 mm EDTA, and 1 mm β-mercaptoenthanol. The cells were sonicated using 4× 10 bursts four times, and the resulting supernatant was centrifuged at 800 × g for 10 min at 4 C to remove the cellular debris. The supernatant was removed and centrifuged at 100,000 × g for 1 h at 4 C to isolate the cytosolic and membrane fractions. After the high-spin centrifugation the supernatant was removed, and glycerol was added to the supernatant to a final concentration of 30% and stored as the cytosolic fraction. The membrane pellet was washed in the Tris-HCl-sucrose buffer, resonicated, resuspended, and centrifuged at 100,000 × g for an additional hour at 4 C. The cytosolic fraction and this second soluble fraction had no enzymatic activity with steroid substrate. The membrane pellet was resuspended in the Tris-HCl-sucrose buffer using a homogenizer, and glycerol was added to a final concentration of 30% and stored at −80 C freezer. The total protein for the cytosolic and membrane fractions was determined for each sample using the method of Bradford (54).

Steady-State Kinetic Analysis of the Expressed 3α-HSDs

The steady-state kinetic parameters for the 3α-HSDs were determined using the isolated cytosolic and membrane fractions from transiently transfected COS-1 cells. The kinetic analysis (final volume of 200 μl) was conducted at 37 C in the Tris-HCl-sucrose buffer, pH 7.4 (final concentrations were 40 mm Tris-HCl, 200 mm sucrose, 0.8 mm EDTA, and 0.8 mm β-mercaptoenthanol), 10 mm MgCl₂, 4% methanol, and a combination of [³H]3α-diol and unlabeled steroid to obtain the final substrate concentration. Isolated cytosolic and membrane fractions were added, and the reactions were initiated by the addition of the oxidized cofactor nicotinamide adenine dinucleotide to a final concentration equal to 1 mm as previously determined (36, 39, 42). Reactions were terminated using 1 ml of water-saturated ethyl acetate, extracted, dried, and plated for TLC analysis as described earlier. This discontinuous assay measures the formation of DHT from 3α-diol vs. time. To obtain progress curves, time points were collected to determine the initial velocity for each concentration tested by linear regression. Plots of velocity vs. substrate concentration were hyperbolic and could be iteratively fit to the Michaelis-Menten equation [v = (V_max * S)/(K_m + S)] to yield values for V_maxapp, V_maxapp/K_m and K_m, and their associated ses for the oxidation of 3α-diol.

CAT Reporter Gene Assays

COS-1 cells were plated into six-well plates at a density of 3.5 × 10⁻⁵ cells using DMEM (− phenol red), 1% CDT-FBS, 1% penicillin/streptomycin, and 2% l-glutamine. Twenty hours after plating, the cells were transfected with FuGENE6 using 0.2 μg of pCMV-AR, 0.4 μg of (ARE)₂-tk-CAT, 0.1 μg of pCMV-β-galactosidase, and 0.2 μg of pcDNA3 or the pcDNA3–3α-HSD of choice. Twenty-four hours after the transfection, steroids were added to the individual wells. Steroid concentrations included DHT (1 × 10⁻¹² to 1 × 10⁻⁶m), testosterone (1 × 10⁻¹² to 1 × 10⁻⁶m), and 3α-diol (1 × 10⁻¹² to 1 × 10⁻⁶m). The specificity of the CAT assay was determined by using the nontransfected control, the minus AR control (i.e. p-(ARE)₂-tk-CAT, pCMV-β-galactosidase, and pcDNA3); the tk-CAT promoter control (i.e. pCMV-AR, p-tk-CAT, pCMV-β-galactosidase, and pcDNA3); the pbasic-CAT basic or no promoter control (i.e. pCMV-AR, pbasic-CAT, pCMV-β-galactosidase, and pcDNA3); and the DMSO control (i.e. pCMV-AR, p-(ARE)₂-tk-CAT, pCMV-β-galactosidase, and pcDNA3).

The CAT assay monitors the transfer of n-butyryl from n-butyryl coenzyme A (Sigma-Aldrich Corp.) to d-threo-[dichloroacetyl-1,2–14C] chloramphenicol (60 mCi/mmol, PerkinElmer LLC) whereby the products are separated by TLC. The CAT reporter gene assay was performed using the method described by Promega. Briefly, after 20 h of incubation with steroid, the cells were washed twice with PBS, and 200 μl of reporter lysis buffer was added to each well and incubated at 37 C for 15 min. The cells were scraped, collected, and frozen at −80 C until further processing. The cells were thawed, lysed by vortexing for 15 sec, and centrifuged, and the resulting supernatant was diluted with 800 μl of reaction lysis buffer. The β-galactosidase assay was performed as described earlier, and the samples were normalized to the no-steroid-treated cells using the β-galactosidase activity. The lysates were heated at 60 C for 10 min to inactivate native cell deacetylases, as described by Promega, and the CAT assay was subsequently performed at 37 C for 45 min. The reactions were terminated using 1 ml of water-saturated ethyl acetate, extracted, dried, redissolved with ethyl acetate, and plated for TLC analysis. The TLC plates were prerun twice and developed using methylene chloride/acetone [92.5:7.5 (vol/vol)]. The TLC plates were analyzed using the TLC-linear analyzer as described previously. The percentage of CAT activity was calculated and compared with the no-steroid control to determine fold induction. The EC₅₀ values were obtained by plotting the percent of CAT activity (fold-activation/maximum activation * 100) vs. the Log₁₀ concentration of steroid (M) using Grafit 5 [y = (Range)/(1 + exp(slope factor * ln(abs(S/EC₅₀)))) + Background]. Inhibition of CAT activity was monitored using the AR antagonist flutamide (Sigma-Aldrich Corp.) in these experiments. Flutamide (0.001 μm–10 μm) and the test steroid (EC₅₀ concentration) were combined, dried down under nitrogen, dissolved in DMSO, and added to the cells as a cocktail.

Real-Time RT-PCR

Real-time RT-PCR determined the relative mRNA expression levels of the oxidative 3α-HSDs in human prostate tissue. The development of the real-time PCR assay included the identification of specific primers to amplify only the desired gene and placement of the primers to cross over an exon-intron sequence to prevent the nonspecific amplification of genomic DNA. Primer specificity was determined by separating the PCR product on a 3% gel and by sequencing to ensure only the amplification of the desired gene. Real-time PCR was performed using a DNA Engine Opticon2 Continuous Fluorescence Detector (MJ Research, Inc., Waltham, MA), and each plate contained nine standards in duplicate and four no-template controls. At the end of the PCR reaction, melting curves were performed to ensure amplification of the desired gene. The RT-PCR method was linear (> r = 0.995) over a dynamic range (10⁹) as determined by plotting the log₁₀ fluorescence intensity vs. the amount of plasmid. Total RNA pooled from 32 human Caucasian male prostates was purchased from BD Bioscience (Palo Alto, CA) and 1 μg of total RNA was reverse transcribed using the GeneAmp RNA PCR Kit. The primer sequences for amplification of ERAB, AR, and the housekeeping genes GAPDH and PBGD were obtained from previous publications (55–57). The primers for RL-HSD are: forward, 5′-dGCT TTC TTT GTA GGA GGC TAC TGT G-3′; reverse, 5′-dTCC TTA ATA TGC TTG GGG GCT TCT-3′, giving a 187-bp product. Primers for RODH 5 are: forward, 5′-dGAG GCC TTC TCT GAC AGC CTG AG-3′; reverse, 5′-dCCA TAG TGG GCC TGT GTG GCA-3′, giving a 169-bp product. NT 3α-HSD primers are: forward, 5′-dCCA AGT TGG GGA GAA AGG TCT-3′; reverse, 5′-dCAC TGG AGA CAT TAA TAA CTC TCC-3′, giving a 197-bp product. Primers for RODH 4 are: forward, 5′-dGAC CGG TCC AGT CCA GAG GTC-3′; reverse, 5′-dTAG CGA GTA CGG GGG TGG CAG-3′, giving a 160-bp product. The conditions for the real-time PCR using SYBR Green (QIAGEN, Inc., Valencia, CA) were as follows: 95 C for 15 min followed by 40 cycles of 94 C for 15 sec, X°C for 30 sec, and 72 C for 30 sec (where X = 58 C for GAPDH, PBGD, and AR; 60 C for RL-HSD, NT 3α-HSD, and RODH 4; and 63 C for ERAB and RODH 5). Full-length standards (2,500,000 fg − 0.025 fg) were generated for ERAB, RL-HSD, RODH 5, NT 3α-HSD, and RODH 4 from their appropriate cDNA plasmids (pcDNA2-ERAB, pcDNA3-RL-HSD, pcDNA3-RODH 5, pcDNA3-NT 3α-HSD, and pcDNA3-RODH 4). PCR product standards (2,500,000 fg − 0.25 fg) were generated for AR, GAPDH, and PBGD by isolating the desired PCR product after a PCR reaction using total human liver RNA. The product was then isolated by gel purification and used as standards with correction factors for GADPH (3.30), PBGD (7.48), and AR (11.79) due to the difference in molecular weight between full-length and PCR product standards.

RT-PCR in Cultured Prostate Primary Epithelial and Stromal Cells

The cultured primary prostate epithelial and stromal cells were maintained as previously reported (58, 59). Total RNA (1 μg) from primary prostate epithelial (n = 14) and stromal (n = 15) cells was reverse transcribed using GeneAmp RNA PCR Kit, and 50 ng of cDNA was subsequently used to determine the expression of the SDRs using the real-time PCR method described above.

Acknowledgments

This work was supported by National Institutes of Health (NIH) Grant R01 CA-90744 (to T.M.P.), and Department of Defense (Army) Grant PC040420 (to D.M.P.). D.R.B. was supported, in part, by NIH Training Grant 1R25-CA-101871-D1.

Abbreviations

 
• Adione

  5α-Androstane-3,17-dione;

 
• AKR

  aldo-keto reductase;

 
• AKR1C2

  type 3 3α-HSD;

 
• androsterone

  3α-hydroxy-5α-androstan-17-one;

 
• AR

  androgen receptor;

 
• ARE

  androgen response element;

 
• BPH

  benign prostatic hyperplasia;

 
• CaP

  prostate adenocarcinoma;

 
• CAT

  chloramphenicol acetyl transferase;

 
• CDT-FBS

  charcoal/dextran treated FBS;

 
• DHT

  5α-dihydrotestosterone;

 
• 3α-diol

  5α-androstane-3α,17β-diol;

 
• 3β-diol

  5α-androstane-3β,17β-diol;

 
• ERAB

  endoplasmic reticulum amyloid β-peptide binding protein/L-3-hydroxyacyl coenzyme A dehydrogenase/type 10 17β-HSD;

 
• GAPDH

  glyceraldehyde-3-phosphate dehydrogenase;

 
• HSD

  hydroxysteroid dehydrogenase;

 
• IRES

  internal ribosome entry sequence;

 
• LC/MS

  liquid chromatograpy/mass spectrometry;

 
• NT 3α-HSD

  novel type of human microsomal 3α-HSD;

 
• PBGD

  porphobilinogen deaminase;

 
• RL-HSD

  RODH-like 3α-HSD/human oxidative 3α-HSD;

 
• RODH

  retinol dehydrogenase;

 
• RODH 4

  retinol dehydrogenase 4;

 
• RODH 5

  11-cis retinol dehydrogenase;

 
• SDR

  short-chain dehydrogenases and reductases;

 
• TLC

  thin-layer chromatography.