Testosterone accumulation in prostate cancer cells is enhanced by facilitated diffusion

Abstract

Background:

Testosterone is a driver of prostate cancer (PC) growth via ligand-mediated activation of the androgen receptor (AR). Tumors that have escaped systemic androgen deprivation, castration resistant prostate cancers (CRPC), have measurable intratumoral levels of testosterone, suggesting that a resistance mechanism still depends on androgen-simulated growth. However, AR activation requires an optimal intracellular concentration of androgens, a situation challenged by low circulating testosterone concentrations. Notably, PC cells may optimize their androgen levels by regulating the expression of steroid metabolism enzymes that convert androgen precursors into androgens. Here we propose that testosterone entry into the cell could be another control point.

Methods:

To determine whether testosterone enters cells via a transporter, we performed in vitro ³H-testosterone uptake assays in androgen-dependent LNCaP and androgen and AR-independent PC3 cells. To determine if the uptake mechanism depended on a concentration gradient, we modified UGT2B17 levels in LNCaP cells and measured androgen levels by LLE-MS (liquid-liquid extraction mass spectrometry). We also analyzed CRPC metastases for expression of AKR1C3 to determine whether this enzyme that converts adrenal androgens to testosterone was present in the tumor stroma (microenvironment) in addition to its expression in the tumor epithelium.

Results:

Testosterone uptake followed a concentration gradient but unlike in passive diffusion, was saturable and temperature-dependent, thus suggesting facilitated transport. Suppression of UGT2B17 to abrogate a testosterone gradient reduced testosterone transport while overexpression of the enzyme enhanced it. The facilitated transport suggests a paracrine route of testosterone uptake for maintaining optimal intracellular levels. We found that AKR1C3 was expressed in the tumor microenvironment of CRPC metastases in addition to epithelial cells and the pattern of relative abundance of the enzyme in epithelium versus stroma varied substantially between the metastatic sites.

Conclusions:

Our findings suggest that in addition to testosterone transport and metabolism by tumor epithelium, testosterone could also be produced by components of the tumor microenvironment. Facilitated testosterone uptake by tumor cells supports a cell non-autonomous mechanism for testosterone signaling in CRPC.

INTRODUCTION

The androgenic hormones testosterone and the testosterone metabolite dihydrotestosterone (DHT) are of clinical importance as activators of the androgen receptor (AR). Ligand-bound AR engages a gene expression program that promotes the survival and proliferation of prostate carcinoma (PC).^{1, 2} Therefore, testosterone suppression by surgical or pharmacological methods, commonly termed androgen deprivation therapy (ADT), is a standard approach for treating advanced PC. While most patients initially respond to ADT, disease progression to castration resistant prostate cancer (CRPC) is nearly universal. Notably, the majority of CRPCs retain AR activity through a variety of mechanisms that include the maintenance of high intratumoral androgen levels despite undetectable or low plasma concentrations. Prior studies indicate that these high intracellular testosterone or DHT levels are derived by the metabolism of circulating adrenal androgen precursors, androstenedione (AED) and dehydroepiandrosterone sulfate (DHEAS), into testosterone, and subsequently into DHT, or by synthesizing testosterone from progesterone or cholesterol precursors.^3–6 Thus, new AR pathway therapies have been developed to inhibit testosterone/DHT signaling either by blocking the DNA binding domain of AR or by inhibiting testosterone biosynthesis.⁷

An alternative mechanism contributing to intracellular testosterone accumulation involves the regulated trans-cellular transport of testosterone or metabolic precursors. Several models have been proposed for testosterone interaction with the cell. The first model holds that due to the small size, neutral charge, and lipophilic character of the testosterone molecule, unbound testosterone enters cells by passive diffusion down a concentration gradient.^{8, 9} In the second model, protein bound testosterone is transported by endocytosis. In this model steroid hormone binding globulin (SHBG) bound testosterone forms a complex with the endocytic receptor megalin in the clathrin-coated pits in the cell membrane. This complex is then internalized into endocytotic vesicles which then release the hormone into the cytoplasm.^{10, 11} A third model comes from the observation in rats and rams where testosterone enters seminiferous tubules and the rete testis fluid much more rapidly than its structurally similar metabolites, DHT and 3α-androstanediol (3α -diol, 5α-androstane-3α,17β-diol), suggesting the possibility of a specific carrier. In these cases, testosterone accumulation is saturable and driven by a downward concentration gradient, in line with carrier-mediated passive transport, i.e. facilitated diffusion.^12–15 A fourth model involves energy-dependent, i.e. active transport for testosterone, and is supported by studies of epididymal epithelium in the rat.¹⁶ In addition, recent work indicates that OATP1B3, an anionic low affinity transporter for e.g. estrone-3-sulfate (E1S) and DHEAS, could also transport testosterone at 20uM concentrations.¹⁷

Here we present additional support for findings that challenge the orthodoxy of passive, non-specific diffusion as the only entry mechanism of testosterone into PC cells. We found at the prostate cell level the same phenomenon as described in seminiferous tubules and rete testis,^12–15 in that testosterone followed a concentration gradient but was saturable (thus specific) and temperature dependent. While this is consistent with a facilitated transport mediated by a specific carrier, in all cells evaluated we still observed a residual testosterone transport that could not be competed, implicating a background of passive, non-specific diffusion. Furthermore, we discuss possible sources of intratumoral testosterone by showing that in PC3 cells AED is converted into testosterone and secreted into the media. Interestingly, we also found that aldo-keto reductase family 1-member C3 (AKR1C3), an enzyme that converts AED into testosterone, is expressed not exclusively in the PC cells but in tumor microenvironment (stromal) cells in a subset of CRPC metastases. In this case testosterone could be delivered to tumor cells via paracrine secretion. Taken together our data suggest that in addition to the primary mechanism of passive diffusion intracellular testosterone concentration can be augmented in PC cells by facilitated diffusion.

MATERIALS AND METHODS

Cells and reagents.

LNCaP, 22Rv1, VCaP, PC3, DU145, RWPE-1, Caco2, and HepG2 cells were obtained from American Type Culture Collection (ATCC, Rockville, Maryland, USA). Immortalized PSC27 (PSC27i) prostate myofibroblast cells were generated by Dr. Beatrice Knudsen. BPH1, a benign hyperplastic prostatic epithelial cell line immortalized by SV40-LT antigen, was a generous gift from Dr. Simon Hayward. All cells were used within 20 passages after receipt (Supplemental Materials and Methods). Testosterone (T), Estradiol (E2), progesterone (P), DHT, DHEAS and, E1S were purchased from Sigma (St. Louis, MO). Enzalutamide was obtained from Medivation (San Francisco, CA).

Isotype labelled steroids were purchased as follows: Androstene-3,17-dione-2,3,4-¹³C₃ (Cerilliant, Round Rock, TX), Testosterone-2,3,4-¹³C₃ (Cambridge Isotope laboratories, Tewksbury, MA), 5α-dihydrotestosterone-2,3,4-¹³C₃ (Cambridge Isotope laboratories, Tewksbury, MA), 5α-Androstane-3,17-dione-¹³C₃ (Isosciences, Ambler, PA), and Dehydroepiandrosterone-¹³C₃ (Isosciences, Ambler, PA). β-Glucuronidase was from Helix Pomatia, Type H-5 (Sigma, St. Louis, MO).

Transgenic cell lines.

The generation of PC3-AR overexpressing cell line has been reported.¹⁸ AR expression was verified by immunoblotting using AR- antibodies (Supplemental Materials and Methods). pGIPZ-shUGT2B17 and non-silencing lentiviral control pTRIPZ-tet-on-shNSC were purchased from Open Biosystems (Pittsburgh, PA); shRNA targeting sequences are noted in the Supplemental Materials and Methods. shUGT2B17 was subcloned into inducible pTRIPZ lentiviral vector (Dharmacon, Lafayette, CO) according to standard protocols. The subsequent clones were validated by sequencing. To generate lentivirus 293FT cells were first transduced with pTRIPZ-tet-on-shUGT2B17 construct together with lentiviral plasmids psPAX2 and pMD2.G (gift from Dr. Yu Ding) in Opti-MEM (Invitrogen, Carlsbad, CA) with Lipofectamine 2000 (Invitrogen, Carlsbad, CA). To generate UGT2B17-deficient cell line, LNCaPs were transduced with lentiviral particles containing pTRIPZ-tet-on-shUGT2B17 (hereafter referred as i-shUGT2B17) in the presence of 6μg/ml polybrene (Santa Cruz Biotechnology, Dallas, TX). Puromycin (Invitrogen, Carlsbad, CA) selection was kept for at least one month before use.

To make UGT2B17 over-expression cell lines, 293FT cells were transduced with the pLenti6.3/TO/V5-DEST-tet-on-UGT2B17-ORF (i-ORF) construct following the same protocol as above and the supernatants were applied to pre-established LNCaP transactivator (TR) cells. Both geneticin and blasticidine were applied for positive clone selection. 1 μg/ml of doxycycline (Invitrogen, Carlsbad, CA) was introduced to the cells transduced with i-shUGT2B17 or i-ORF construct at least 3 days before each experiment to ensure that gene expression was fully induced. The transduction efficiency of knockdown and overexpression was determined by qRT-PCR and western blot.

Steroid measurements.

To measure steroid concentration in cell pellets, i-shUGT2B17 expressing LNCaP cells were plated in 10cm plates. Once the cells reached the confluency of 85%−90%, complete culture medium was replaced with serum free medium for 24 hours before collection of cell pellets. i-ORF expressing LNCaP cells were starved for 24 hours followed by 5nM of DHT treatment for 12 hours before cell pellet collection. The cells were then trypsinized and triple washed with cold 1x PBS. The testosterone levels in cell pellets were measured using liquid chromatography-tandem mass spectrometry (LC-MS-MS) as described previously.^{5, 19}

To measure testosterone concentrations in media, cells were incubated for 48 h at 37°C in serum-free, phenol red-free medium containing 1μM ¹³C₃-androstenedione. Cells and media were collected separately and frozen at −70°C for LC–MS–MS analysis (Supplemental Materials and Methods). Limit of detection (LOD) for the LLE–MS assay was 0.00098nM.

Luciferase reporter assay.

Dual-Luciferase® Reporter Assay System (Promega, Madison, WI) was used to measure AR promoter activities. LNCaP cells were plated in a 24-well plate, 5 × 10⁴/well, transfected with an ARE luciferase reporter construct using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). Promoter activities were expressed relative luciferase activity (firefly luciferase light units normalized for Renilla luciferase values).

Western analysis.

Whole cell lysates were prepared with RIPA buffer containing 10% phosSTOP phosphatase inhibitor cocktail (Roche, Indianapolis, IN) and 1x protease inhibitor cocktail (Cell Signaling, Danvers, MA). Protein concentrations were quantified with the BCA Protein Assay kit (Pierce, Rockford, IL). 40 μg or 10 μg of total protein was separated by SDS–PAGE respectively. Gels were transferred onto 0.2 μm nitrocellulose membranes and the blots were probed using anti-UGT2B17, anti-V5, anti-AR antibodies (see Supplemental Materials and Methods). Anti-GAPDH antibody (GeneTex, Irvine, CA) was used as a loading control. Band intensity was analyzed by Image J (NIH, Bethesda, MD).

In vitro ³H-Testosterone uptake assay.

Measurement of testosterone accumulation in cells was conducted on confluent monolayers of various cell lines seeded on poly-L-lysine coated 24-well plates. Cells were first pretreated with uptake buffer (135 mM NaCl, 5 mM KCl, 2.5mM CaCl₂, 0.6 mM MgSO₄, 6 mM D-glucose, and 10 mM HEPES, pH 7.4) for 10–20 minutes and then incubated for 10 sec to 60 min in uptake buffer supplemented with 1mCi/ml [1,2,6,7–3H(N)] - testosterone (70 Ci/mmol, PerkinElmer Life and Analytical Sciences, Waltham, MA), which was used at the concentrations of 1–50nM. After incubation cells were then washed 3X in ice cold PBS on ice, solubilized in 0.2% SDS, mixed with scintillation fluid, Ecoscint (National Diagnostics, Atlanta Georgia, USA), and radioactivity (cpm) measured via liquid scintillation counting (LS 6000IC scintillation counter, Beckman Coulter, Brea, California). Each experiment was conducted in duplicates or triplicates and the counts (cpm) were converted into molarity (M) and normalized to total protein concentration (mg). Salt and pH modified uptake buffers were made according to Ogihara et al.²⁰ (Supplemental Materials and Methods). The specificity of OATP-mediated uptake was shown by the use of a specific transport inhibitor bromosulphophthalein (BSP, vWR, Radnor, PA) at a 100 μM final concentration, which was added into both the preincubation buffer and the transport buffer.²¹ Enzalutamide was used at a 5–10μM concentration 15 min before experiment to block the AR binding domain. Cellular protein content was determined by using the BCA Protein Assay Reagent (Thermo Fisher Scientific, Rockford, IL) with bovine serum albumin as the standard.

RNA Extraction and Quantitative real-time PCR.

Total RNA was extracted from cells using an RNeasy kit (Qiagen, Carlsbad, CA) and reverse-transcribed into cDNA with High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions. cDNA quantifications were done in triplicate with SYBR Green PCR master mix (Applied Biosystems, Foster City, CA). Water was used as negative control. Gene expression was normalized to housekeeping gene expression of RPL13A. The primer sequences for SDHA, RPL13A, UGT2B15 and UGT2B17 are accessible in the Supplemental Materials and Methods.

Prostate tissue microarray (TMA) and immunohistochemistry (IHC).

The CRPC TMA was created as previously described.²² For IHC staining details see Supplemental Materials and Methods. IHC staining intensity was scored in a blinded fashion by two pathologists, L.D.T. and R.G.C. The expression of AKR1C3 was assessed for tumor cells and on adjacent tumor-associated cells, i.e stromal cells, bone marrow cells of bone marrow and hepatocytes of liver metastases. Protein expression levels were based on compositional immunohistochemical stain intensity, and was the sum of multiplicands of percentage of cells staining at each of three levels of stain intensity – none, faint, intense – yielding a score between 0 and 200 in each cell type. Lesions with scores <20 were categorized as negative.

Data and statistical analysis.

Biological replicates of the experiments, each performed in duplicate or triplicate (as technical replicates) for each condition, were treated as independent data to assign p-values (n.s.= non-significant, *P<0.05, **P<0.005, ***P<0.0005)). Representative results are shown and error bars represent the standard deviation. When three or more groups were compared, one-way ANOVA was used, followed by Turkey HSD post-hoc test (SPSS Statistics 22, IBM). When two groups were compared, an unpaired 2-tailed Student’s t test was used (Microsoft Office Excel 2010) to analyze data. For the significance of the difference in the dot density graph analysis of the metastatic samples we used the Mann-Whitney U-test. Transport assay kinetic parameters were determined by using non-linear regression analysis with Prism 5.01 and the curves were fitted to the Michaelis-Menten equation v=v_max*(S)/(Km+(s)).

RESULTS

Assessment of testosterone uptake in neoplastic prostate epithelial cells

We performed competition assays for testosterone accumulation in a panel of AR (+) and AR (−) PC cell lines to provide evidence for a specific transport mechanism for testosterone entry into cells. Tritiated testosterone (³H-T, 10 nM) was competed for 10 minutes with 100x excess (1μM) of unlabeled ‘cold’ T (C-T), and intracellular ³H-T accumulation was measured. For baseline (uncompeted) uptake, we found that VCaP cells, which have amplified AR, accumulated more ³H-T than any other AR-expressing PC cell line, including LNCaP, 22Rv1 and LAPC4. Cells with no AR (PC3, DU145) (Supplemental Figure S1) accumulated ³H-T in a similar manner as the LAPC4 cells that express wild type (wt) AR (Figure 1A and 1B). When ³H-T was competed with 100x excess of C-T, we observed competition, but not all ³H-T could be prevented from entering cells. In all cell lines a similar amount of residual ³H-T uptake that could not be competed with excess C-T was observed (Figure 1A and 1B). Similar levels of ³H-T signal were found (without competition) in benign prostate epithelial cells (RWPE1 cells), immortalized benign prostate fibroblasts (PSC27i), and benign prostatic hyperplasia epithelial cells (BPH1), which all lack or express very low levels of AR, as well as in two non-prostate cell lines, the colon cancer CaCo2 and HepG2 hepatocellular carcinoma lines (Figure 1C).²³ These results suggest that while there is a T-specific, saturable import mechanism, 20–50% of T (depending on the cell line) can enter cells in a non-specific (unsaturable manner). Moreover, while the absence or presence of normal levels of AR did not affect the basal uptake, high levels of AR, providing a larger sink for a chemical gradient, enhanced basal uptake.

Figure 1. Testosterone uptake into various cell lines.

(A, B) ³H-T accumulation in PCa cell lines with distinct AR status: LNCaP and 22Rv1, mutated AR; LaPC4, wild type AR; PC3 and DU145, no AR; and VCaP, amplified AR. 10 nM of ³H-T was applied to cells for 10 min with or w/o 100x excess of cold T. Both AR(+) and AR(−) cells accumulated ³H-T which could be competed with cold T. (C) 10nM ³H-T accumulation after 10 min in normal prostate epithelial cells (RWPE1), prostate hyperplasia cells (BPH1), immortalized normal prostate stromal cells (PSC27i), hepatocellular carcinoma cells (HepG2) and colon carcinoma cells (Caco2). Counts were determined from cell lysates and normalized to weight (mg). Results are presented as mean ± S.D. for two separate experiments. (n.s.= non-significant, *P<0.05, **P<0.005, ***P<0.0005)

Dose response and kinetics of testosterone uptake in prostate cancer cells

We next focused on two cell lines, LNCaP and PC3, as examples of AR(+) and AR(−) cells, respectively, and extended the measurements across time and testosterone concentrations in order to determine kinetics and dose-dependence.

Kinetics of testosterone uptake in LNCaP and PC3 cells.

Accumulation of testosterone was linear over time for the first 3 minutes in both cell types, approximately at an initial uptake rate of 390 ± 175 pmol_*mg⁻¹_*min⁻¹ for PC3 cells and 972 ± 234 pmol_*mg⁻¹_*min⁻¹ for LNCaP cells (Figure 2). However, testosterone uptake exhibited a second phase, in which the uptake rate was 16 ± 1.6 pmol_*mg⁻¹_*min⁻¹ for PC3 cells and 54 ± 13 pmol_*mg⁻¹_*min⁻¹ for LNCaP cells (Figure 2 and Supplemental Figure S2A). At 20 minutes after the addition of testosterone, accumulation reached a plateau in both cell lines. The repeatedly seen early peak at 5 min may indicate a biphasic process caused by two distinct mechanisms (Supplemental Figure S2A).

Figure 2. Characteristics of ³H-T uptake.

(A, B) ³H-T uptake into LNCaP and PC3 cells as function of time. ³H-T accumulation in cells was determined at the times indicated. (C, D) Competition of ³H-T uptake with cold T in LNCaP and PC3 cells. Uptake of 10 nM ³H-T was determined after 20 minutes for increasing concentrations of cold testosterone (cold T). (E) Temperature-dependence of 10nM ³H-T uptake into LNCaP and PC3 cells on ice and 37^oC. (F) Effect of cross-linking fixative on 10nM ³H-T uptake into LNCaP cells for 3 and 60 minutes. (G) Comparison of accumulation following exposure to 5 nM and 10 nM ³H-T for 3 minutes in PC3 cells expressing AR [AR(+)] and in control cells [AR(−)] which were pretreated 15 minutes with 10 μM enzalutamide (ENZA). A presentative experiment from three separate experiments is presented. (n.s.= non-significant, *P<0.05, **P<0.005, ***P<0.0005)

Dose response of testosterone uptake in LNCaP and PC3 cells.

Testosterone uptake into LNCaP and PC3 cells was measured at pH 7.4 in a confluent monolayer of cells that had been cultured in normal medium for 24–49 h. ³H-T uptake was competed with increasing concentrations of C-T in LNCaP and PC3 cells. Uptake of 10 nM ³H-T was determined at 30 min, i.e. when the plateau had been reached. Increasing the dose of C-T revealed an asymptotic approach to the above-observed residual non-competable accumulation of ³H-T (at 10 min), suggesting that this non-competable fraction of uptake was a constitutive feature, consistent with passive non-specific diffusion through the cell membrane, i.e. not mediated by a saturable testosterone-specific transporter (Figure 2C,D).

Transport on ice.

To assess temperature dependence, which would indicate a complex process involving an active (energy consuming) transporter, we compared uptake of 10nM ³H-T by LNCaP and PC3 cells on ice and at 37^oC. Uptake was less efficient on ice in both LNCaP and PC3 cells: 60% and 30%, respectively (Figure 2E). Thus, the reduced rate on ice was compatible with the Van’t Hoff rule for slowing chemical kinetics at lower temperatures.²⁴

Transport in crosslinked conditions.

To further distinguish passive diffusion from active membrane transport, we fixed LNCaP- cells with a crosslinker, 1% paraformaldehyde, for 10 min. We observed that in the initial phase, there was a 75% decrease in the uptake of testosterone in paraformaldehyde-treated compared to untreated cells. In the saturated phase at 60 min the uptake in paraformaldehyde-treated cells was reduced by 23% (Figure 2F). This partial reduction of uptake following crosslinking is consistent with a contribution of passive diffusion of testosterone into the cell especially in the initial phase.

Role of AR in accumulation.

The target of testosterone within the cell is AR, which upon binding of testosterone triggers downstream biological responses. Often in cellular signaling such responses would involve a feedback loop, which could modulate testosterone uptake indicating that uptake is a specific biological process that can be regulated. To determine if presence of AR in the cell affects intracellular testosterone accumulation, we blocked the ligand binding sites of AR by treatment with 10 μM of the AR ligand binding site antagonist enzalutamide for 15 min before testosterone treatment. In the presence of enzalutamide, testosterone accumulation in AR overexpressing PC3 cells (AR+) was reduced to levels similar to those in the parental mock transduced PC3 cells AR(−) (Figure 2G). This and the above (Figure 1A and 1B) findings suggest that the AR protein may play a role in acting as a sink to allow for accumulation of testosterone within the cell, consistent with both a gradient-dependent entry and an internal reservoir.

Assessment of organic anion-transporting polypeptides (OATPs) in prostate cancer cell testosterone transport

Recent studies have suggested that OATP1B3, an anionic low affinity solute transporter for compounds such as E1S and DHEAS, could also transport testosterone.^{17, 25, 26} Therefore, we established OATP1B3 overexpressing LNCaP lines and examined testosterone entry into the cells.²⁷ Following administration of 20 nM ³H-T no difference in testosterone accumulation in OATP1B3 overexpressing LNCaP cells compared to naïve LNCaP cells was observed (Figure 3A). Moreover, rifampicin,²⁸ which is known to inhibit OATP1B3-mediated uptake, failed to inhibit ³H-T entry into naïve LNCaP or into OATP1B3 overexpressing LNCaP cells (Figure 3A). Because both LNCaP and PC3 cells appear to express several OATPs, including OATP1B1, −1B3, −4A1 and 5A1 as determined by qRT-PCR, we treated cells with sulfobromophthalein (BSP), a general inhibitor for organic anion transporters^{29, 30} but found that BSP did not inhibit uptake of testosterone at 10 nM into PC cells with endogenous OATP expression (Figure 3B and C).

Figure 3. Role of OATP inhibitors, salts and pH on ³H-T uptake.

(A) ³H-T accumulation in LNCaP cells exposed to 20 nM in the presence of 100x and 200x of cold T in comparison to rifampicin, an OATP1B1 and OATP1B3 inhibitor. (B) Inhibition of 10 nM ³H-T uptake to LNCaP and PC3 cells by 1 μM bromosulphophthalein (BSP). (C) RT-PCR analysis of SLCO transcript levels in LNCaP and PC3 cells presented as Ct-value chart. SDHA is a house keeping gene. (D) Effect of sodium (Na+) and chloride (Cl-) on 10nM ³H-T uptake in LNCaP cells. The 40% decrease in ³H-T uptake in sodium-free Krebs-Ringer phosphate buffer (pH 7.4) suggests that the transporter is sodium-dependent. (E, F) pH had no effect on ³H-T transport in LNCaP or PC3 cells. Abbreviations: Rif, rifampicin; Buffer, the uptake buffer as described in materials and methods; no Na⁺ and no Cl^-, uptake buffers without sodium and chloride ions, respectively. A presentative experiment from three separate experiments is presented. (n.s.= non-significant, *P<0.05, **P<0.005, ***P<0.0005)

Changes in salt concentrations and pH have been shown to affect membrane mediated transport. However, we found that in LNCaP cells, testosterone uptake was only partially sodium dependent, but not dependent on chloride (Figure 3D) and that pH had no effect on T accumulation (Figure 3E and 3F). Together these findings suggest that OATPs do not play a major role in the uptake of testosterone by PC cells at low extracellular concentrations.

Effects of steroids on testosterone uptake

We next studied the specificity of testosterone uptake by competition with other steroids. We observed that in the presence of other neutral steroids, such as progesterone, estradiol and DHT, accumulation of ³H-T in both LNCaP and PC3 cells was significantly reduced, respectively, by 58% and 40% when competed with progesterone, by 65% and 41% when competed with estradiol, and by 69% and 41% when competed with DHT. By contrast, sulfated steroids E1S and DHEAS did not compete with testosterone (Table 1). In summary, these findings suggest that at least one component of testosterone uptake by PC cells is mediated by facilitated transport embodied by a passive transporter that exhibits some substrate specificity.

Table 1.

Inhibitory effects of various compounds on 10nM ³H-testosterone uptake into LNCaP and PC3 cells

Compound	Concentration	LNCaP	PC3
	(fold)	Uptake (% of control)
	Hormone
Progesterone	100x	58.13±1.87*	39.80±1.06*
	10x	58.50±1.51*	41.65±1.88*
Estradiol	100x	64.97±4.72*	40.97±1.38*
	10x	79.69±2.29**	47.50±1.04*
Estrone-3-sulfate	100x	97.9±7.0 n.s.	95.4±4.8 n.s.
DHEAS	100x	91.55±6.0 n.s.	102.90±3.83 n.s.
DHT	100x	68.87±4.79*	40.89±0.89*

For each condition, results are displayed as mean values and standard deviation (n= two biological replicates, treated as independent data to assign p-values.) P values are from comparison of treated samples to untreated (n.s.= non-significant,

^*P<0.0005,

^**P<0.005,

^***P<0.05)

The influence of testosterone metabolism on testosterone uptake

Cellular testosterone is subject to irreversible glucuronidation,^{31, 32} a process which converts steroids into a more water-soluble metabolite, thereby abolishing their affinity for steroid receptors and routing them for elimination.^{33, 34} In the prostate, two UDP-glucuronosyltransferases, UGT2B15 and UGT2B17, are responsible for the glucuronidation of testosterone and DHT as well as for metabolizing DHT into 3α-diol-G (Figure 4I).³⁵ In accordance with a recent report that the expression of these genes requires AR,³⁶ we found that only LNCaP but not PC3 cells express both UGT2B15 and UGT2B17 (Supplemental Figure S3A).

Figure 4. Modifications of UGT2B17 expression changes intracellular androgen levels, AR expression and its activity creating an androgen gradient, which alters testosterone uptake into cells.

(A) Testosterone levels were increased in UGT2B17 depleted (i-shUGT2B17) LNCaP cells. (B) DHT treatment increased AR expression in UGT2B17 knockdown (i-shUGT2B17) cells in comparison to control cells as shown by Western blotting (C), and quantified by densitometry analysis. (D) By contrast, T levels were decreased in UGT2B17 over-expressing (i-ORF) LNCaP cells. (E) DHT treatment in UGT2B17 knockdown cells increased AR activity, while R1881 treatment had no effect. (F) By contrast, DHT treatment in UGT2B17 overexpressing cells decreased AR activity, while R1881 had no effect. Results are representative of two independent experiments. (G) A schematic presentation of testosterone inactivation (red arrows) together with the converting enzymes that are involved. (H) ³H-T uptake (5nM for 3 minutes) by UGT2B17 wild type (dox-) LNCaP cells grown in FBS, in serum free media containing either 5 nM or 500 nM testosterone was decreased when compared to UGT2B17-deficient (dox+) LNCaP cells. (I) ³H-T uptake (10 nM) into UGT2B17-overexpressing LNCaP cells (dox+) was increased in comparison to control cells (dox-). Results are mean values ± S.D. for two separate experiments. (n.s.= non-significant, *P<0.05, **P<0.005, ***P<0.0005)

UGT2B17 converts testosterone and DHT to their glucuronide derivatives at a 10- and 6-fold higher rate than UGT2B15.^37–40 Consequently, we genetically modified the expression of UGT2B17 to augment a testosterone gradient by reducing intracellular concentrations of free testosterone to demonstrate the role of facilitated diffusion in testosterone uptake. We generated LNCaP cell lines in which either UGT2B17 over-expression (i-ORF) or shRNA mediated knockdown (i-shUGT2B17) was inducible by doxycycline (dox) (Supplemental Figure S3B–3E). We first performed functional studies to ensure that the genetically manipulated cell lines behaved as expected. Accordingly, dox treatment of i-shUGT2B17 LNCaP cells reduced UGT2B17 levels and resulted in a 5.7–13.7-fold increase in testosterone levels (from 0.013–0.031 pg/mg to 0.178 pg/mg) compared to non-induced control cells (Figure 4A). In this system, we also confirmed the previously reported commensurate effect of UGT2B17 knockdown on AR expression in the presence of DHT: following UGT2B17 knockdown with dox in LNCaP cells, DHT treatment led to increased AR expression as shown by immunoblotting, compared to parental LNCaP cells (Figure 4B and 4C). By contrast, over-expression of UGT2B17 in LNCaP cells decreased intracellular testosterone levels by approximately 67% (Figure 4D). As expected, modification of UGT2B17 expression also affected AR activity: knockdown of UGT2B17 stimulated AR activity while the over-expression of UGT2B17 depressed AR activity in the presence of DHT (Figure 4E and 4F). Consistent with previous findings^{32, 41} these modifications of UGT2B17 expression had no effect on AR activity when stimulated by R1881, a non-metabolized androgen (Figure 4E and 4F).

The above findings suggest that manipulation of the expression of UGT2B17 alters free intracellular testosterone, and hence the transmembrane gradient, allowing us to demonstrate facilitated transport. Partial knockdown of UGT2B17 by shRNA produced increased levels of intracellular testosterone, and we measured approximately 2-fold less intracellular testosterone accumulation after exposure to exogenous testosterone in UGT2B17 deficient cells (Figure 4G). By contrast, in UGT2B17overexpressing cells, with or without AR blockade with ENZA, there was a slight increase in testosterone accumulation (Figure 4H) suggesting that changing the level of intracellular testosterone can affect the uptake of external testosterone, consistent with facilitated diffusion.

AKR1C3 in human CRPC metastasis and conversion of adrenal androgen AED into testosterone in vitro and in vivo

The in vitro testosterone uptake experiments suggest that the PC cell lines LNCaP and PC3, established from PC metastases,^{42, 43} utilize a specific transport mechanism for importing extracellular testosterone. One source of testosterone in ADT treated patients is the adrenal gland that secretes AED and DHEA(S) into the circulation. AED and DHEA can then be converted into testosterone by AKR1C3 (Figure 5A).^44–46 Therefore, we next examined whether LNCaP and PC3 cells can synthesize testosterone from AED. We found that AR+ LNCaP cells glucuronidated essentially all the newly synthesized secreted testosterone (Figure 5B). Interestingly, in AR-null PC3 cells that do not express either UGT2B15 or-17, both free and glucuronidated testosterone and DHT were identified in the media, suggesting that another UGT, potentially UGT2B7, could be involved in glucuronidation in this cell type (Figure 5C).⁴⁷ These results suggest that PC cells can secrete active free testosterone into the extracellular microenvironment which may vary depending on intracellular glucuronidation activity as reported by others.^{48, 49} This observation opens the possibility of paracrine exchange of testosterone.

Figure 5. AED conversion to testosterone.

(A) A schematic of AED-metabolic pathway. (B, C) In LNCaP cells (AR+) ¹³C₃-AED is converted into testosterone and other androgens; however testosterone is glucuronidated before secretion into media. Interestingly, in PC3 cells (AR-) both free and glucuronidated testosterone is secreted into media. (D) Distribution of number of TMA cores with respect to positive (+) and negative (−) stain for AKR13C in either epithelial (E) or stromal (S) compartment, defining these four groups E+/S+, E+/S-, E-/S+, E-/S-. (E) Pie chart showing prevalence of these groups for various metastatic sites. (F) The dot density graph shows that the H-score for AR immunohistochemical staining (y-axis, calculated as described in methods) is higher in the E+/S+ compared to E+/S- and E-/S+ groups. (G) Selected CRPC metastatic tissue sections showing immunohistochemical staining for AKR1C3 and AR in various metastatic sites. Red arrows point to AKR1C3-positive hepatocytes in liver and to endothelial cells and leukocytes in bone. Scale bar, 100 μm. (n.s.= non-significant, *P<0.05). Abbreviations: AED, androstenedione; T, testosterone; DHT, dihydrotestosterone; 5α-Adione, androstenedione; LN, lymph node.

Since the conversion of AED to testosterone is catalyzed by the enzyme AKR1C3 we determined whether metastatic CRPC tumors have the potential to produce testosterone from AED by assessing AKR1C3 expression in CRPC metastases.^{22, 50} We used IHC to assess AKR1C3 protein expression in 160 tumor cores from 42 CRPC patients organized in tissue microarrays. Overall 157 of 160 cores stained positive for AKR1C3. In the majority of the samples AKR1C3 was expressed in both the tumor epithelial cells (E) and stromal cells (S) (E+/S+, 101/161). However, all possible patterns of expression with respect to epithelium or stroma were observed in subsets of tumors: E+/S-, 30/161; E-/S+, 27/161; and E-/S-, 3/161 (Figure 5D). Interestingly, AKR1C3 staining was overwhelmingly prominent in tumor epithelial cells in lymph nodes and soft tissues and less so in stromal cells in these metastatic sites, while in bone and liver metastases stromal AKR1C3 expression contributed substantially to the staining patterns. Intriguingly in liver metastases AKR1C3 expression was pronounced in the hepatocytes surrounding the tumor cells (Figure 5E an 5G).

We also compared the above groups defined by AKR1C3 expression patterns with respect to their expression of AR and PSA as measured by the H-score. We did not observe a statistically significant difference in the expression levels of AR between these groups. Interestingly, AR expression levels in the (E+/S+) group displayed a broad distribution among the tissue cores, ranging from tissues with H-score of zero to samples with H-score of 200. By contrast in the other groups, (E+/S-) and (E-/S+), a bimodal distribution was detected, with samples expressing either high or low levels of AR (Figure 5F). By contrast, the same analysis for PSA suggested that its expression levels were on average higher in the (E+/S+) group in comparison to the (E+/S-) group of samples (Supplemental Figure S4).

DISCUSSION

The experimental results provide evidence for a specific carrier-meditated, passive (facilitated) entry of testosterone into PC cells. We found that testosterone accumulation reaches a plateau after 15–20 min. However we consistently observed a peak intracellular abundance during the early rapid rise, causing a departure from the monotonic kinetics of intracellular accumulation (see Figure 2A and Supplemental Figure S2). This biphasic increase of cellular testosterone suggest that an additional, efficient transient uptake mechanism is involved. While less studied, neutral unconjugated steroid hormones, e.g. estrogen, progesterone, cortisol, and thyroid hormone, have also been shown to enter cells by transporter-mediated uptake.^{20, 51–55} In PC, the cell membrane protein OATP1B3 has been shown to transport testosterone into cells;^{17, 26} however concentrations were1000x higher than those used here to utilize this transporter, which is in line with the concentrations needed for the transport of conjugated steroids, such as DHEAS and E1S, by OATPs.²⁵ Although the molecular identity of the carrier studied here remains elusive, our results provide a starting point for further characterization. Such a transporter could provide a new class of target for suppressing intracellular testosterone levels in PC cells, further helping to address the problem of resistance to ADT.

PC undergoes an adaptive response to ADT by increasing intratumoral androgen synthesis. This synthesis depends on de novo cholesterol conversion or the utilization of circulating adrenal androgen precursors. One of the enzymes needed to convert DHEA and AED into testosterone is AKR1C3,^{56, 57} whose expression in CRPC is upregulated.⁴⁶ In addition to prominent AKR1C3 expression in epithelial cells, as previously published,²² we found that AKR1C3 was also expressed in the tumor microenvironment/stroma (endothelium, leukocytes and fibroblasts) as well as hepatocytes adjacent to liver metastases. Interestingly, depending on the metastatic tissue site, the expression pattern of AKR1C3 enzyme with respect to epithelium and stroma, varied substantially. In bone metastases the most prominent AKR1C3 staining was found in the bone marrow stromal cells, consistent with in vitro studies. By contrast in lymph node and soft tissue metastases tumor epithelium was the main site of AKR1C3 expression. In liver metastases, AKR1C3 staining was most prominent in hepatocytes. Collectively, these data suggest that the conversion of AED into testosterone may take place in stromal cells^56–60 which then deliver testosterone in a paracrine fashion to adjacent carcinoma cells. It is worth noting that another source for paracrine testosterone are platelets, which express the required steroidogenic enzymes at increased levels in CRPC.^{61, 62} All of these non-cell autonomous (or paracrine) mechanisms would benefit from facilitated transport of testosterone into the cells.

DHEA and AED can be converted into testosterone and/or to DHT by sequential use of the enzymes AKR1C3, HSD3B2, and SRD5A1.⁶³ However, one needs to consider two metabolic pathways for synthetizing T and/or DHT in CRPC: the ‘front door’ and/or the ‘back door’ pathways. Both have been suggested to be involved in the development of CRPC.^{64, 65} In the classical front door mechanism testosterone either acts as a ligand for AR or is converted into the high affinity AR ligand DHT. Conversely, in the backdoor mechanism DHT is mainly synthetized from 5α-dione, thus independent of testosterone as a precursor (Figure 5A).^{64, 66, 67} Pathway utilization depends on the order in which these three enzymes are activated. The order of activation in turn depends on precursor concentrations, enzyme expression levels and affinity of the given enzyme for its substrate. Whether the expression site of the enzyme, e.g. AKR1C3, in epithelium vs stroma plays a role in pathway activation remains to be explored.

CONCLUSIONS:

Our results suggest that PC cells may utilize cell autonomous and cell non-autonomous, paracrine modes of testosterone delivery to the cancer cells. The latter could benefit from the facilitated import and transport of testosterone. The putative transporter, whose identity has yet to be determined, may offer a novel therapeutic target. The cells responsible for the external testosterone synthesis include juxtaposed tumor cells as well as stromal cell constituents of the tumor microenvironment and may vary depending on the metastatic site.