Important Roles of the AKR1C2 and SRD5A1 Enzymes in Progesterone

Chemico-Biological Interactions xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Chemico-Biological Interactions

journal homepage: www.elsevier.com/locate/chembioint

Important roles of the AKR1C2 and SRD5A1 enzymes in progesterone metabolism in endometrial cancer model cell lines ⇑ Maša Sinreih a, Maja Anko a, Sven Zukunft b, Jerzy Adamski b,c,d, Tea Lanišnik Rizˇner a, a Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia b Institute of Experimental Genetics, Genome Analysis Centre, Helmholtz Zentrum München, München, Germany c Lehrstuhl für Experimentelle Genetik, Technische Universität München, 85356 Freising-Weihenstephan, Germany d German Centre for Diabetes Research, 85764 Neuherberg, Germany article info abstract

Article history: Endometrial cancer is the most frequently diagnosed gynecological malignancy. It is associated with Available online xxxx prolonged exposure to estrogens that is unopposed by progesterone, whereby enhanced metabolism of progesterone may decrease its protective effects, as it can deprive progesterone receptors of their active Keywords: ligand. Furthermore, the 5a-pregnane metabolites formed can stimulate proliferation and may thus 3-Keto/20-keto-reductases contribute to carcinogenesis. The aims of our study were to: (1) identify and quantify progesterone 5a-Reductases metabolites formed in the HEC-1A and Ishikawa model cell lines of endometrial cancer; and (2) pinpoint Pre-receptor metabolism the enzymes involved in progesterone metabolism, and delineate their roles. Progesterone metabolism 5a-Pregnanes studies combined with liquid chromatography–tandem mass spectrometry enabled identification and 4-Pregnenes quantification of the metabolites formed in these cells. Further quantitative PCR analysis and small-interfering-RNA-mediated gene silencing identified individual progesterone metabolizing enzymes and their relevant roles. In Ishikawa and HEC-1A cells, progesterone was metabolized mainly to 20a-hydroxy- pregn-4-ene-3-one, 20a-hydroxy-5a-pregnane-3-one, and 5a-pregnane-3a/b,20a-diol. The major difference between these cell lines was rate of progesterone metabolism, which was faster in HEC-1A cells. In the Ishikawa and HEC-1A cells, expression of AKR1C2 was 110-fold and 6800-fold greater, respectively, than expression of AKR1C1, which suggests that 20-ketosteroid reduction of 5a-pregnanes and 4-pregnenes is catalyzed mainly by AKR1C2. AKR1C1/AKR1C2 gene silencing showed decreased progesterone metabolism in both cell lines, thus further supporting the significant role of AKR1C2. SRD5A1 was also expressed in these cells, and its silencing confirmed that 5a-reduction is catalyzed by 5a-reductase type 1. Silencing of SRD5A1 also had the most pronounced effects, with decreased rate of progesterone metabolism, and consequently higher concentrations of unmetabolized progesterone. Our data confirm that in model cell lines of endometrial cancer, AKR1C2 and SRD5A1 have crucial roles in progesterone metabolism, and may represent novel targets for treatment. Ó 2014 Elsevier Ireland Ltd. All rights reserved.

1. Introduction The majority of cases are diagnosed in post-menopausal women, while 20% of EC patients are pre-menopausal and 5% are younger Endometrial cancer (EC) is the most frequently diagnosed than 40 years of age [4]. EC patients are usually treated surgically, gynecological malignancy [1]. Worldwide, it is the sixth most com- but with patients who wish to preserve fertility, alternative mon cancer in women, with 320,000 new cases diagnosed in 2012 treatments with progestins are used [5,6]. [2]. Both incidence and mortality from EC appear to be rising [3]. From a histological and molecular pathology perspective, the majority of EC cases can be divided into two groups. About 80%

Abbreviations: AKR1C1/2/3, aldo–keto reductase 1C1/1C2/1C3; EC, endometrial of diagnosed cases are endometrioid type (type 1). These tumors cancer; HSD17B2, 17b-hydroxysteroid dehydrogenase type 2; SRD5A1/2, 5a- are well differentiated and estrogen-dependent, and they have reductase types 1/2. good prognosis. Most type 1 cases originate from endometrial ⇑ Corresponding author at: Institute of Biochemistry, Faculty of Medicine, hyperplasia, and they occur in premenopausal and postmeno- University of Ljubljana, Vrazov trg 2, 1000 Ljubljana, Slovenia. Tel.: +386 1 pausal patients and are less aggressive. Type 2 tumors are 5437657; fax: +386 1 5437641. E-mail address: [email protected] (T.L. Rizˇner). diagnosed mainly in postmenopausal patients and are considered http://dx.doi.org/10.1016/j.cbi.2014.11.012 0009-2797/Ó 2014 Elsevier Ireland Ltd. All rights reserved.

Please cite this article in press as: M. Sinreih et al., Important roles of the AKR1C2 and SRD5A1 enzymes in progesterone metabolism in endometrial cancer model cell lines, Chemico-Biological Interactions (2014), http://dx.doi.org/10.1016/j.cbi.2014.11.012 2 M. Sinreih et al. / Chemico-Biological Interactions xxx (2014) xxx–xxx as estrogen independent. These are more aggressive forms of this 2. Materials and methods cancer that arise spontaneously. Histologically, type 2 tumors are serous, clear-cell, and adenosquamous adenocarcinomas [3]. 2.1. Reagents and chemicals The risk factors for the development of EC include nulliparity, early age at menarche, late age at menopause, obesity, diabetes, All chemicals and solvents were of the highest purity available. tamoxifen treatment, and long-term use of estrogen-only hormone The commercial standards of testosterone, 3a-hydroxy-pregn-4- replacement therapy. Many of these risk factors support the ene-20-one, 3b-hydroxy-pregn-4-ene-20-one, 20a-hydroxy- hypothesis that chronic exposure to estrogens that is unopposed pregn-4-ene-3-one ((20S)-20-hydroxy-pregn-4-ene-3-one), 20b- by progesterone or progestins can lead to increased mitotic activity hydroxy-pregn-4-ene-3-one ((20R)-20-hydroxy-pregn-4-ene-3- of the endometrial cells, and increased DNA replication errors and one), 5a-pregnane-3,20-dione, 3a-hydroxy-5a-pregnane-20-one, somatic mutations, which can result in a malignant phenotype 3b-hydroxy-5a-pregnane-20-one, 20a-hydroxy-5a-pregnane-3- [3,7]. one ((20S)-20-hydroxy-5a-pregnane-3-one), 20b-hydroxy-5a- Progesterone has protective effects on the endometrium, as it pregnane-3-one ((20R)-20-hydroxy-5a-pregnane-3-one), 5a-preg- has anti-proliferative actions and can stimulate differentiation nane-3a,20a-diol ((20S)-5a-pregnane-3a,20-diol), 5a-pregnane- [8]. In pre-menopausal women, progesterone is synthesized 3b,20a-diol ((20S)-5a-pregnane-3b,20-diol), and 5a-pregnane- mainly in the corpus luteum, and its plasma concentrations change 3a,20b-diol ((20R)-5a-pregnane-3a,20-diol) were from Steraloids according to the menstrual cycle. After menopause, the concentra- (Newport, RI, USA). Progesterone, formic acid, Eagle’s minimal tion of progesterone in the blood decreases as the ovaries cease to essential medium, McCOY 5A, fetal bovine serum, and trypsin– function. However, progesterone can still be formed in the adrenal EDTA solution were from Sigma–Aldrich Chemie GmbH (Deisenho- gland and in some other peripheral tissues, including cancerous fen, Germany). OptiMEM was from Life Technologies (Carlsbad, CA, endometrium [9–11]. Progesterone that is formed locally de novo USA). from cholesterol can act in the same cell (intracrine action) or on Solvents were obtained from Carlo Erba Reagents (Rodano, neighboring cells (paracrine action) via progesterone receptors Italy), Carl Roth GmbH (Karlsruhe, Germany), Sigma–Aldrich Che- [9], or it can be further metabolized. mie GmbH (Deisenhofen, Germany), Thermo Fisher Scientific (Geel, The metabolism of progesterone in the endometrium proceeds Belgium) and Alfa Aesar (Heysham, UK). via the actions of 20-ketosteroid reductases (to form 20a- The following steroids were synthesized by the group of Dr. hydroxy-pregn-4-ene-3-one), 3-ketosteroid reductases (to form Gobec from the Faculty of Pharmacy, University of Ljubljana 3a/b-hydroxy-pregn-4-ene-20-one) and 5a-reductases (to form (Ljubljana, Slovenia) (unpublished data): pregn-4-ene-3a,20a-diol 5a-pregnane-3,20-dione). The reduction of the 20-keto group of ((20S)-pregn-4-ene-3a,20-diol), pregn-4-ene-3a,20b-diol ((20R)- progesterone is catalyzed by the aldo–keto reductases AKR1C1– pregn-4-ene-3a,20-diol), pregn-4-ene-3b,20a-diol ((20S)-pregn- AKR1C3, which act as 3-keto, 17-keto and 20-ketosteroid 4-ene-3b,20-diol), and pregn-4-ene-3b,20b-diol ((20R)-pregn-4- reductases, with varying catalytic efficiencies [12–15]. The pro- ene-3b,20-diol). The reduction of progesterone was performed as gesterone metabolite 20a-hydroxy-pregn-4-ene-3-one thus described earlier by Wiebe et al., with some modifications [21]. formed can be oxidized back to progesterone by the 20a- hydroxysteroid dehydrogenase activity of 17b-hydroxysteroid 2.2. Cell culture dehydrogenase type 2 (HSD17B2) [16]. The AKR1C enzymes do not reduce the 3-keto group of 4-pregnenes [17], and thus the The Ishikawa cell line was obtained from The European Collec- conversion of progesterone to 3a/b-hydroxy-pregn-4-ene-20-one tion of Cell Cultures (ECACC) and the HEC-1A cell line was obtained is catalyzed by as-yet-unidentified ketosteroid reductases. The from American Type Culture Collection (ATCC). The cell lines were formation of 5a-pregnane-3,20-dione from progesterone is cata- cultured according to the ECACC and ATCC recommendations. The lyzed by 5a-reductases types 1 and 2 (SRD5A1, SRD5A2). 5a- Ishikawa cells were cultured in Eagle’s minimal essential medium Pregnane-3,20-dione can be further metabolized to mono- and containing 5% fetal bovine serum, and the HEC-1A cells were cul- di-hydroxy-5a-pregnanes by the actions of the AKR1C1–AKR1C3 tured in McCOY 5A medium containing 10% fetal bovine serum. enzymes (Fig. 1) [15,18]. Both of these cell lines were grown without any antibiotics, in a Enhanced metabolism of progesterone may decrease its protec- humidified atmosphere of 5% CO and 95% air, at 37 °C. The cells tive effects, as it can deprive the progesterone receptors of their 2 were passaged at a 1:6 dilution. Cell lines in the 5th to 15th pas- active ligands. Furthermore, the progesterone metabolites formed sage were used in this study. may activate other cellular pathways. The 5a-pregnanes have been shown to stimulate cell proliferation and detachment in breast cancer cell lines, while the 4-pregnenes have the opposite effects 2.3. Progesterone metabolism studies [19]. Moreover, in breast cancer cell lines and tissue, the ratio between 5a-pregnanes and 4-pregnenes is changed in favor of For identification and quantification of progesterone metabo- 5a-pregnanes [19]. Recently, it was also shown that 5a-preg- lites, Ishikawa and HEC-1A cells were plated into 6-well plates at nane-3,20-dione and 3a-hydroxy-pregn-4-ene-20-one can regu- a density of 106 cells/well. To avoid potential steroid-mimicking late induction and growth of human breast cancer cells in a effects, the standard media were replaced after 24 h with serum- mouse model [20]. free and phenol-red-free media, and progesterone was added to a The aims of the present study were to: (1) identify and quantify final concentration of 50 nM for Ishikawa cells, and 50 and the progesterone metabolites formed in the HEC-1A and Ishikawa 5000 nM for HEC-1A cells (stock in dimethyl sulfoxide), with a final model cell lines of endometrial cancer; and (2) pinpoint the dimethyl sulfoxide concentration of 0.25%. The cells were incu- enzymes involved in progesterone metabolism, and delineate their bated with progesterone for 1, 2, 4, 8 or 24 h. The medium from roles in progesterone metabolism. Our data show differences in two wells of 6-well plates (4 mL) was combined, testosterone progesterone metabolism in these model cell lines of endometrial (100 ng/mL final concentration) and ethyl acetate (3 2 mL) were cancer, and they confirm the important roles of the AKR1C2 and added, with the mixture vortexed for 3 min. After centrifugation SRD5A1 enzymes in progesterone metabolism. In this manner, (5 min, 1000g), the upper ethyl acetate phase was collected, the the present study contributes to a better understanding of proges- solvent evaporated off, and the steroids dissolved in a mixture of terone metabolism in endometrial cancer. acetonitrile/ water/ formic acid (50/50/0.5; v/v).

Fig. 1. Progesterone metabolism in peripheral tissues. Reduction of the 20-keto group of progesterone is catalyzed by the aldo–keto reductases AKR1C1–AKR1C3. The progesterone metabolite 20a-hydroxy-pregn-4-ene-3-one can be oxidized back to progesterone by the 20a-hydroxysteroid dehydrogenase activity of 17b-hydroxysteroid dehydrogenase type 2 (HSD17B2). The formation of 5a-pregnane-3,20-dione from progesterone is catalyzed by 5a-reductases types 1 and 2 (SRD5A1, SRD5A2). 5a-Pregnane- 3,20-dione can be further metabolized to mono- and di-hydroxy 5a-pregnanes, through the actions of the AKR1C1–AKR1C3 enzymes. Metabolites 3a/b-hydroxy-pregn-4- ene-20-one are formed from progesterone by jet unidentiﬁed 3-ketosteroid reductases.

2.4. SiRNA-mediated gene silencing Hilden, Germany), according to the manufacturer instructions. Due to the small amount of cell material used for siRNA-mediated For transient transfection with small-interfering (si)RNAs, gene silencing, the total RNA was isolated with NucleoSpin RNA 120,000 Ishikawa and HEC-1A cells/well were seeded into 6-well isolation kit (Macherey–Nagel; Düren, Germany), according to plates, for 30–40% confluency after 24 h. The cells were transfected manufacturer protocol. The total RNA was transcribed to cDNA with 60 pmol pre-designed Silencer siRNAs (Ambion; Austin, TX, with superscript VILO cDNA kit (Invitrogen; Carlsbad, CA, USA). USA) (Table 1) using x-treme-GENE siRNA transfection reagent Quantification was achieved using the LightCyclerÒ 480 Real-Time (8 lL) (Roche; Penzberg, Germany), according to the manufacturer PCR system (Roche; Basel, Switzerland), using TaqManÒ Universal protocol. As a negative control, Silencer Negative Control #1 PCR Master Mix and the universal thermocycling parameters rec- (Ambion, 4390843) was used (60 pmol), and as a positive control, ommended by Applied Biosystems. The expression levels were Silencer Select GAPDH (Ambion, 4390849) was used (60 pmol). determined using the exon-spanning hydrolysis probes (labeled Transfection was performed in OptiMEM reduced-serum medium, with the dye 50 6-carboxyfluorescein [FAM]) that are commercially which was replaced with standard medium 6 h after transfection. available as Assay on Demand (Applied Biosystems; Foster City, CA, Then, 48 h after transfection, the cells were treated with progester- USA), using optimized primer and probe concentrations (Table 3). one. Eight hours and 24 h after addition of progesterone, the ste- The primers and fluorescent TaqManÒ minor groove binder roids were extracted from the media, as described above. To (MGB) probes for the specific amplification of AKR1C1–AKR1C3 determine the efficiency of transfection, total RNA and proteins were designed in our laboratory (Table 4) [22]. The RT-PCR samples were isolated 48 h and 72 h after transfection (Table 2). were run in triplicates in 384-well plates (Roche; Basel, Switzer- land), in a reaction volume of 5.0 lL. 2.5. RNA isolation and quantitative real-time PCR For gene expression analyses, the normalization factor for each sample was calculated based on the geometric mean of both of the For the gene expression analyses, total RNA was isolated from reference genes (HPRT1, POLR2A). The PCR amplification efficiency three biological replicas of each cell line using Quiazol (Qiagene; was determined from the slope of the log-linear portion of the

Table 1 The predesigned Silencer siRNAs used in this study.

Gene silenced Sense Antisense Catalog number* AKR1C1, AKR1C2 CAGUUGACUUCAGAGGAGAtt UCUCCUCUGAAGUCAACUGga S3989 AKR1C3 AGAAAUCUCCACUAUUUUAtt UAAAAUAGUGGAGAUUUCUgt S16447 SRD5A1 GCAUGUUGAUAAACAUCCAtt UGGAUGUUUAUCAACAUGCcc S13415

* Ambion, Austin, TX, USA.

Table 2 Securityguard cartridges (C18; 4 3.0 mm; Phenomenex; Aschaff- Efficiency of siRNA-mediated gene silencing. enburg, Germany). Samples of 5 lLor20lL were injected via a PAL Silenced gene Ishikawa HEC-1A HTC-xt autosampler (CTC Analytics; Zwingen, Switzerland). The After 48 h After 72 h After 48 h After 72 h mobile phase consisted of (A) 5% acetonitrile, 0.5% formic acid in water, and (B) 0.5% formic acid in acetonitrile, and the flow rate AKR1C1 88.7 ± 5.6 91.8 ± 5.1 84.4 ± 0.3 82.3 ± 10.5 AKR1C2 93.2 ± 0.8 89.9 ± 5.9 84.6 ± 0.4 84.8 ± 3.6 was 690 lL/min. The column temperature was 25 °C. The gradient AKR1C3 89.5 ± 2.3 90.3 ± 7.8 79.5 ± 0.6 76.7 ± 6.0 elution was: initial condition, 0.0–10.0 min, 60% A; 10.0–25.0 min, SRD5A1 92.7 ± 0.6 92.5 ± 3.5 89.4 ± 3.2 84.3 ± 4.2 60–35% A; 25.0–30.0 min, 35% A; 30.0–30.1 min, 35–60% A; 30.1– 40.0 min equilibration at 60% A. The HPLC was coupled to a Mean ± SEM. QTRAPÒ 5500 system (AB Sciex Deutschland GmbH; Darmstadt, Germany), with all controlled by the Analyst 1.6 software (AB Sciex Deutschland GmbH; Darmstadt, Germany). Tandem mass spec- Table 3 trometry analysis was performed in positive ion mode and under TaqManÒ PCR assays on demand for the genes investigated in this study. constant electrospray ionization conditions. The concentration of * Gene Assay ID Gene name each steroid was calculated on the basis of standard curves. PPIA Hs99999904_m1 Peptidylprolyl isomerase A (cyclophilin A) Calibration curves were constructed from eight concentrations, HPRT1 Hs99999909_m1 Hypoxanthine phosphoribosyltransferase 1 covering the range from 2 ng/mL to 1000 ng/mL. Testosterone (Lesch-Nyhan syndrome) was added to each sample at a final concentration of 100 ng/mL, GAPDH Hs99999905_m1 Glyceraldehyde-3-phosphate dehydrogenase POLR2A Hs00172187_m1 Polymerase (RNA) II (DNAdirected) polypeptide and used as internal standard. The retention times and monitoring A, 220 kDa transitions for testosterone, progesterone and 16 of its metabolites SRD5A1 Hs00602694_Mh Steroid-5-alpha-reductase polypeptide 1 used in the assay are provided in Table S1. SRD5A2 Hs00165843_m1 Steroid-5-alpha-reductase polypeptide 2

* Applied Biosystem, Foster City, CA, USA. 2.7. Western blot

Protein aliquots of 30 lg were separated by SDS PAGE on 12% Table 4 Tris–glycine gels. The proteins were transferred from the gels to Sequences of the primers and probes for the specific amplification of AKR1C1, AKR1C2 PVDF membranes (Millipore, USA) and blocked with 5% non-fat and AKR1C3. milk for 2 h, to avoid non-specific binding. Afterwards, the Gene Primer/ probe Sequence membranes were incubated with the primary mouse polyclonal anti-AKR1C3 antibodies (1:5000; a6229; Sigma–Aldrich Chemie AKR1C1 Forward primer 50-TGCAGAGGTTCCTAAAAGTAAAGCTTTA-30 Reverse primer 50-GGAAAATGAATAAGGTAGAGGTCAACATAA-30 GmbH; Deisenhofen, Germany) in TTBS with 1% non-fat milk Fluorescent MGB-NFQ 50-TTCAATTGCCAATTTGGT-30 FAM powder for 2 h. Next, the membranes were incubated with the sec- probe ondary antibodies (1:10,000; peroxidase-conjugated goat anti- AKR1C2 Forward primer 50-CCTAAAAGTAAAGCTCTAGAGGCCGT-30 mouse IgG + IgM (H + L), Jackson ImmunoResearch Laboratories Reverse primer 50-ACATGTTTACAATA ATGAGGAGCAGGT-30 Inc., West Grove, PA, USA,) for 2 h at room temperature. Protein 0 0 Fluorescent MGB-NFQ 5 -AAGCCGGGTTCCACCA-3 FAM levels were examined in three replicates of HEC-1A cells transfec- probe ted with AKR1C3 specific siRNA and three replicates of HEC-1A cells 0 0 AKR1C3 Forward primer 5 -GTTGCCTATAGTGCTCTGGGATCT-3 transfected with scrambled siRNA, as a negative control. GAPDH Reverse primer 50-GGACTGGGTC CTCCAAGAGG-30 Fluorescent MGB-NFQ 50-CACCCATCGTTTGTCTC-30 FAM was used as the normalization control. Membranes were incubated probe overnight with the primary mouse polyclonal anti-GAPDH antibodies (1:3000; G8795; Sigma–Aldrich Chemie GmbH; Deis- [22]. enhofen, Gerrmany) in TTBS with 1% non-fat milk powder. The next day the membranes were incubated with the secondary antibodies (1:10,000; peroxidase-conjugated goat anti-mouse Ig- calibration curve for each gene investigated, and this was G + IgM (H + L), Jackson ImmunoResearch Laboratories Inc., West accounted for in the further calculations. Gene expression for each Grove, PA, USA) for 2 h at room temperature. The SuperWest Pico Cq sample was calculated from the crossing-point value (Cq) as E , (Thermo Fisher Scientific; Rockford, IL, USA) was used for the 10 divided by the normalization factor and multiplied by 10 . detection of the bound antibodies according to the manufacturer The siRNA-mediated gene silencing data were analyzed using instructions. Quantification of Western blots was carried out with the comparative DDCT method, by calculation of the difference the Multi Gauge software (Fujifilm software, Fujifilm, Tokyo, Japan) between the threshold cycle DCT values of the negative control using a Fujifilm LAS4000 image reader (Fujifilm, Tokyo, Japan). siRNA-treated sample, versus the DCT of the targeted siRNA-treated sample. Fold changes were calculated as a function of DDCT: 3. Results and discussion Fold change ¼ 2DDCT 3.1. The AKR1C1–AKR1C3 and SRD5A1 genes are expressed in the The gene expression normalization factor for each sample was Ishikawa and HEC-1A model cell lines calculated based on the geometric mean of two reference genes: HPRT1 and POLR2A. First, we examined the expression of genes encoding the progesterone-metabolizing enzymes in the Ishikawa and HEC-1A 2.6. Liquid chromatograph–tandem mass spectrometry epithelial cell lines. The Ishikawa cell line was established from the endometrial adenocarcinoma of a 39-year-old woman [23]. The chromatographic separation of the steroids was performed This cell line expresses estrogen and progesterone receptors and using an Agilent Infinity 1260 HPLC system with a Kinetex 2.6 lM is one of the most widely used human endometrial-derived cell XB-C18 column (150 4.6 mm; Phenomenex; Aschaffenburg, culture models [24]. The HEC-1A cell line was established from Germany), equipped with a Securityguard guard column and the endometrial adenocarcinoma of a 71-year old patient, and it

3.2. Ishikawa and HEC-1A cells differ in their progesterone metabolism pregnane-3a,20a-diol, and 5a-pregnane-3b,20a-diol to a lesser In Ishikawa cells, the metabolism of 50 nM progesterone pro- extent. Progesterone may also be metabolized to low levels of 4- ceeded mainly to 20a-hydroxy-pregn-4-ene-3-one in the first 4 h pregnenes: 3a-hydroxy-pregn-4-ene-20-one, pregn-4-ene- and 8 h (47.6 ng/mL after 4 h and 65.0 ng/mL after 8 h). In addition 3a,20a-diol and 3b-hydroxy-pregn-4-ene-20-one. Dihydroxy to 20a-hydroxy-pregn-4-ene-3-one also 5a-pregnane-3,20-dione metabolites of progesterone are then further metabolized forming (35.4 ng/mL) and 3a-hydroxy-pregn-4-ene-20-one (20.8 ng/mL) sulfates (data not shown). were formed at higher concentrations after 4 h and 5a-pregnane- The metabolism of progesterone in HEC-1A cells proceeded rap- 3,20-dione (23.7 ng/mL), 20a-hydroxy-5a-pregnane-3-one idly, and therefore the higher concentration of 5 lM progesterone (22.0 ng/mL) and 3a-hydroxy-pregn-4-ene-20-one (21.8 ng/mL), was first used for the metabolism studies. At all-time points cho- 5a-pregnane-3a,20a-diol (14.0 ng/mL), 3b-hydroxy-5a-pregnane- sen, 20a-hydroxy-pregn-4-ene-3-one was the most abundant 20-one (12.4 ng/mL), 3a-hydroxy-5a-pregnane-20-one (9.6 ng/ metabolite detected, with concentrations of 669.7 ng/mL, mL) and 5a-pregnane-3b,20a-diol (9.1 ng/mL) after 8 h. After 1336.7 ng/mL and 1519.7 ng/mL after 4, 8 and 24 h, respectively. 24 h, most of progesterone was metabolized, with three main After the 24-h incubation, most of the progesterone was metabo- metabolites detected: 5a-pregnane-3a,20a-diol (63.7 ng/mL), lized, and we also detected pregn-4-ene-3a,20a-diol (431.0 ng/ 20a-hydroxy-pregn-4-ene-3-one (22.5 ng/mL) and 5a-pregnane- mL), 5a-pregnane-3a,20a-diol (232.7 ng/mL), 5a-pregnane- 3b,20a-diol (19.5 ng/mL). Lower amounts of other metabolites 3b,20a-diol (214.0 ng/mL), 20a-hydroxy-5a-pregnane-3-one were also detected (Table 5 and Fig. 3). (176.7 ng/mL) (Table 6 and Fig. 4). Other metabolites were also Based on these data, we can propose the following pathway of detected at lower concentrations. progesterone metabolism in Ishikawa cells: first, progesterone is The proposed pathway of 5 lM progesterone metabolism in reduced at either 20-keto group or C5 position to form 20a- HEC-1A cells includes first the formation of 20a-hydroxy-pregn- hydroxy-pregn-4-ene-3-one and 5a-pregnane-3,20-dione, 4-ene-3-one and 5a-pregnane-3,20-dione, followed by reduction respectively. Next, 20a-hydroxy-5a-pregnane-3-one is formed; at 20-keto position. The 20a-hydroxy-5a-pregnane-3-one formed this is then metabolized at the 3-keto group to form mainly 5a- in this manner is then reduced to 5a-pregnane-3a,20a-diol and 5a-pregnane-3b,20a-diol. After 8 and 24 h, high levels of 4-pregnenes are also formed: pregn-4-ene-3a,20a-diol, pregn-4-ene- 3b,20a-diol, 3b-hydroxy-pregn-4-ene-20-one and 3a-hydroxy- pregn-4-ene-20-one. To better understand the first steps of progesterone metabolism in HEC-1A cells we also studied metabolism of 50 nM progesterone after 1, 2 and 4 h (Table 7 and Fig. 5). Results here were more similar to metabolic profiles observed in Ishikawa cell line. After 1 h, progesterone was metabolized mainly to 5a-pregnane-3,20-dione (62.2 ng/mL) and 20a-hydroxy-pregn-4-ene-3-one (40.6 ng/mL). After 2 h of incubation the major metabolites detected were 20a- hydroxy-pregn-4-ene-3-one (59.0 ng/mL), 5a-pregnane-3,20- dione (52.5 ng/mL), and 20a-hydroxy-5a-pregnane-3-one (49.6 ng/mL). After 4-h metabolism the 5a-pregnane metabolites prevailed with 5a-pregnane-3a,20a-diol (90.4 ng/mL), 20a- hydroxy-5a-pregnane-3-one (84.9 ng/mL), 20a-hydroxy-pregn-4- ene-3-one (61.2 ng/mL), 5a-pregnane-3b,20a-diol (40.2 ng/mL) and pregn-4-ene-3a,20a-diol (39.0 ng/mL). Other metabolites were also detected at lower concentration. Based on these data, we can propose the following pathway of Fig. 2. Expression of genes encoding progesterone metabolizing enzymes in the 50 nM progesterone metabolism in HEC-1A cells: first, 20-keto Ishikawa and HEC-1A cell lines. The mRNA levels for gene expression were normalized to HPRT1 and POLR2A. Data are means + SD, from three biological group of progesterone is reduced to form 20a-hydroxy-pregn-4- replicates. ene-3-one and progesterone is reduced at C5 to form 5a-preg-

Fig. 3. Metabolites of 50 nM progesterone formed after 4, 8 and 24 h in Ishikawa cells. In Ishikawa cells, the metabolism of 50 nM progesterone proceeded mainly to 20a- hydroxy-pregn-4-ene-3-one in the ﬁrst 4 h and 8 h. In addition to 20a-hydroxy-pregn-4-ene-3-one also 5a-pregnane-3,20-dione and 3a-hydroxy-pregn-4-ene-20-one were formed at higher concentrations after 4 h and 5a-pregnane-3,20-dione, 20a-hydroxy-5a-pregnane-3-one and 3a-hydroxy-pregn-4-ene-20-one, 5a-pregnane-3a,20a-diol, 3b-hydroxy-5a-pregnane-20-one, 3a-hydroxy-5a-pregnane-20-one and 5a-pregnane-3b,20a-diol after 8 h. After 24 h, most of progesterone was metabolized, with three main metabolites detected: 5a-pregnane-3a,20a-diol, 20a-hydroxy-pregn-4-ene-3-one and 5a-pregnane-3b,20a-diol. Lower amounts of other metabolites were formed. Data are mean ± SEM of three biological replicas.

Table 6 Metabolites of 5 lM progesterone formed in the HEC-1A cell line.

Concentration of individual steroid* (ng/mL) After 4 h* After 8 h* After 24 h* Progesterone 3953.3 ± 37.1 3063.3 ± 242.5 437.7 ± 165.9 20a-Hydroxy-pregn-4-ene-3-one 669.7 ± 73.5 1336.7 ± 97 1519.7 ± 345.2 20b-Hydroxy-pregn-4-ene-3-one 52.5 ± 1.8 48.4 ± 2.4 12.1 ± 2.4 3a-Hydroxy-pregn-4-ene-20-one 114.0 ± 6.5 169.0 ± 8 52.2 ± 12.2 3b-Hydroxy-pregn-4-ene-20-one 39.5 ± 0.9 48.9 ± 1.6 58.1 ± 7.2 Pregn-4-ene-3a,20a-diol 23.8 ± 2.9 88.2 ± 14.0 431.0 ± 37.3 Pregn-4-ene-3b,20a-diol 7.2 ± 1.6 24.1 ± 2.7 108.6 ± 4.4 Pregn-4-ene-3a,20b-diol 42.1 ± 2.6 29.0 ± 1.6 7.2 ± 1.3 5a-Pregnane-3,20-dione 97.3 ± 36.2 49.7 ± 6.6 52.8 ± 21 20a-Hydroxy-5a-pregnane-3-one 25.9 ± 3.3 75.8 ± 7.6 176.7 ± 6.5 3a-Hydroxy-5a-pregnane-20-one 14.3 ± 1.1 14.8 ± 0.4 4.4 ± 0.3 3b-Hydroxy-5a-pregnane-20-one 44.5 ± 3.5 29.7 ± 1.5 7.3 ± 1.4 5a-Pregnane-3a,20a-diol 5.9 ± 1.1 28.7 ± 5.8 232.7 ± 52.3 5a-Pregnane-3b,20a-diol 10.9 ± 2.1 46.4 ± 8 214.0 ± 35.2

* Mean ± SEM; ND, not detected. nane-3,20-dione. Next, 20a-hydroxy-5a-pregnane-3-one is to metabolize 100-fold higher concentrations of progesterone with formed from 5a-pregnane-3,20-dione; this is then metabolized at the equal cell count and in the same time frame, thus metabolism the 3-keto group to form mainly 5a-pregnane-3a,20a-diol, and was evaluated at two different concentrations. When metabolism 5a-pregnane-3b,20a-diol to a lesser extent. Progesterone may also of 5 lM progesterone was studied, 4-pregnenes, especially 20a- be metabolized to low levels of 4-pregnenes: 3a-hydroxy-pregn-4- hydroxy-pregn-4-ene-3-one prevailed, while with lower, 50 nM ene-20-one, and via 20a-hydroxy-pregn-4-ene-3-one to pregn-4- concentrations of progesterone, there were more 5a-pregnane ene-3a,20a-diol and pregn-4-ene-3b,20a-diol. metabolites. The ratio between 5a-pregnane and 4-pregnene The major difference between Ishikawa and HEC-1A cell lines metabolites was 2.0 after 24-h. Similarly, also in Ishikawa cells was the rate of progesterone metabolism. HEC-1A cells were able 50 nM progesterone was metabolized mainly to 5a-pregnane

Fig. 4. Metabolites of 5 lM progesterone formed after 4, 8 and 24 h in HEC-1A cells. In HEC-1A cells, the metabolism of 5 lM progesterone proceeded rapidly. At all-time points chosen, 20a-hydroxy-pregn-4-ene-3-one was the most abundant metabolite detected. After the 24-h incubation, most of the progesterone was metabolized, and we also detected pregn-4-ene-3a,20a-diol, 5a-pregnane-3a,20a-diol, 5a-pregnane-3b,20a-diol, 20a-hydroxy-5a-pregnane-3-one. Other metabolites were also detected at lower concentrations. The data are mean ± SEM of three biological replicas.

Table 7 When Collins et al. incubated a cancerous tissue homogenate with Metabolites of 50 nM progesterone formed in the HEC-1A cell line. progesterone, the major metabolite detected was 5a-pregnane- Concentration of individual steroid* 3,20-dione as well, except that here the metabolism of progester- (ng/mL) one was more rapid, with higher levels of 5a-pregnane-3,20-dione After 1 h* After 2 h* After 4 h* formed, compared to normal proliferative endometrium [29],as also observed in HEC-1A cell line. On the basis of these data, we Progesterone 675.7 ± 23.6 396 ± 25.2 191.3 ± 30.2 20a-Hydroxy-pregn-4-ene-3-one 40.6 ± 4.2 59 ± 2.9 61.2 ± 6.3 hypothesize that 5a-metabolites may have an effect on cell prolif- 20b-Hydroxy-pregn-4-ene-3-one ND ND 5.4 ± 0.2 eration, while 4-pregnenes may stimulate differentiation, similar 3a-Hydroxy-pregn-4-ene-20-one 10.4 ± 2.4 13.6 ± 1.6 15.8 ± 0.9 to that suggested for breast cancer [20,30]. The biological effects 3b-Hydroxy-pregn-4-ene-20-one 5.4 ± 0.3 4.7 ± 0.1 4.8 ± 0.3 of these two groups of metabolites, 5a-pregnanes and 4-pregn- Pregn-4-ene-3a,20a-diol 28.3 ± 5.5 34.9 ± 1.3 39 ± 1.5 Pregn-4-ene-3a,20b-diol 20.8 ± 1.2 24.3 ± 1.0 17.1 ± 2.0 enes, have been studied extensively in breast cancer, which has Pregn-4-ene-3b,20a-diol 4.6 ± 1.2 3.7 ± 0.4 3.9 ± 0.6 shown that 5a-pregnanes stimulate proliferation and detachment, 5a-Pregnane-3,20-dione 62.2 ± 5.8 52.5 ± 3.9 34.1 ± 5.0 while 4-pregnenes counteract these effects [20,30]. So far, the 20a-Hydroxy-5a-pregnane-3-one 19.7 ± 2.3 49.6 ± 2.9 84.9 ± 4.2 possible actions of these progesterone metabolites on EC cell pro- 3a-Hydroxy-5a-pregnane-20-one 12.5 ± 0.3 19.5 ± 1.4 22.5 ± 1.3 liferation, migration and invasiveness have not been studied. 3b-Hydroxy-5a-pregnane-20-one 20.6 ± 1.3 23.3 ± 1.4 16.7 ± 2.0 5a-Pregnane-3a,20a-diol 4.3 ± 0.6 24.2 ± 3.5 90.4 ± 8.3 5a-Pregnane-3b,20a-diol 3.1 ± 0.2 13.8 ± 1.3 40.2 ± 2.1

* Mean ± SEM; ND, not detected. 3.3. Gene silencing reveals important roles for AKR1C2 and SRD5A1 in progesterone metabolism in Ishikawa and HEC-1A cells

To identify the enzymes involved in progesterone metabolism, metabolites. The ratio between 5a-pregnane and 4-pregnene we performed siRNA-mediated gene silencing. The AKR1C1/ metabolites formed in 24 h was 1.7. This is in line with lM KM val- AKR1C2, AKR1C3 and SRD5A1 genes were successfully silenced ues for the reduction of progesterone and 5a-pregnanes by AKR1C using Ambion siRNAs (Table 1), with efﬁciencies from 77% to 93% enzymes [15], and low lM KM value for the reduction of progester- (Table 2). Due to the high homology of their nucleotide sequences, one by SRD5A enzymes [27]. AKR1C1 and AKR1C2 were silenced concurrently using siRNAs that It was previously shown by Collins et al. that normal secretory recognize both of these genes. As there were very low mRNA levels endometrium metabolizes progesterone mainly to 20a-hydroxy- of SRD5A2, this gene was not silenced. Scrambled siRNA and siRNA pregn-4-ene-3-one, while in normal proliferative endometrium, against GAPDH were used as the negative and positive controls, the major metabolite detected was 5a-pregnane-3,20-dione [28]. respectively. Successful silencing of one of these genes, AKR1C3,

Fig. 5. Metabolites of 50 nM progesterone formed after 1, 2 and 4 h in HEC-1A cells. After 1 h, progesterone was metabolized mainly to 20a-hydroxy-pregn-4-ene-3-one and 5a-pregnane-3,20-dione. After 2-h incubation major metabolites detected were 5a-pregnane-3,20-dione, 20a-hydroxy-pregn-4-ene-3-one and 20a-hydroxy-5a-pregnane- 3-one. After 24-h incubation major metabolites detected were 5a-pregnane-3a,20a-diol, 5a-pregnane-3,20-dione, 20a-hydroxy-pregn-4-ene-3-one, 5a-pregnane-3b,20a- diol and pregn-4-ene-3a,20a-diol. Other metabolites were also detected at lower concentration. The data are mean ± SEM of three biological replicas. at the protein level was also conﬁrmed in HEC-1A cells, showing 24 h), 5a-Pregnane-3,20-dione (72% after 24 h), 20a-hydroxy-5a- 66% decrease in protein levels 72 h after siRNA transfection (Fig. 6). pregnane-3-one (21% after 24 h), 3a-hydroxy-5a-pregnane-20- The metabolism of progesterone was then studied 48 h after one (53% after 8 h and 100% after 24 h), 5a-pregnane-3a,20a-diol siRNA-mediated gene silencing. The progesterone metabolites that (64% after 8 h, and 55% after 24 h) and 5a-pregnane-3b,20a-diol were formed after 8 h and 24 h were extracted and analyzed using (41% after 24 h). The decreases in 20a-hydroxy-pregn-4-ene-3- liquid chromatography–tandem mass spectrometry. The differ- one and 20a-hydroxy-5a-pregnane-3-one levels were less ences in the metabolism observed between cells transfected with pronounced compared to when AKR1C1/C2 were silenced (30% the negative control and cells where individual genes of progester- and 21% versus 58% and 41%, respectively, after 24 h). Similarly, one metabolism were silenced are discussed below (Tables 8–11). as observed for the silencing of AKR1C1/AKR1C2, the total concen- When AKR1C1 and AKR1C2 were silenced in Ishikawa cells, trations of the progesterone metabolites decreased. However, as higher levels of progesterone (123% after 24 h) were detected, lower levels of progesterone were also detected, there might be compared to the negative controls (Tables 8 and 9). We also alternative metabolic pathways that are activated when AKR1C3 detected lower levels of several metabolites, and especially 20a- is silenced. These might form metabolites that are not detected hydroxy-pregn-4-ene-3-one (58% after 8 h, and 58% after 24 h), by our liquid chromatography–tandem mass spectrometry 20a-hydroxy-5a-pregnane-3-one (30% after 8 h, and 41% after analysis. 24 h), 3a-hydroxy-5a-pregnane-20-one (100% after 8 h), 5a-preg- The rate of progesterone metabolism drastically decreased nane-3a,20a-diol (91% after 8 h, and 90% after 24 h) and 5a-preg- when SRD5A1 was silenced. As expected, there were higher levels nan-3b,20a-diol (47% after 8 h, and 52% after 24 h), which supports of both progesterone (239% after 24 h) and 20a-hydroxy-pregn- the role of AKR1C1/AKR1C2 in the reduction of steroidal 20-keto 4-ene-3-one (339% after 24 h), while concentrations of 5a-preg- and 3-keto groups, and suggests that these enzymes can form nane-3,20-dione, mono- and di-hydroxy 5a-pregnanes decreased 20a-hydroxy metabolites of 5a-pregnanes (Fig. 7). Moreover, also by up to 92% compared to the negative control. This supports the 3a-hydroxy-pregn-4-ene-20-one and pregn-4-en-3a,20a-diol important role of SRD5A1 in the formation of 5a-pregnanes, and were not detected after AKR1C1 and AKR1C2 were silenced. The it conﬁrms that SRD5A1 silencing reroutes progesterone metabolic total concentrations of all of the progesterone metabolites detected pathway in the direction of 4-pregnenes (up to 316% higher con- decreased 1.2-fold after 8 h, and 1.9-fold after 24 h, both compared centrations after 24 h). to the negative control. Similar data to Ishikawa cells were also seen for HEC-1A cells. When AKR1C3 was silenced in the Ishikawa cells, lower levels of When AKR1C1 and AKR1C2 were silenced, higher levels of proges- progesterone were detected (14% after 24 h), with lower levels of terone were detected (110%), when compared to the negative its metabolites, and especially 20a-hydroxy-pregn-4-ene-3-one control after 24 h (Tables 10 and 11). There were lower levels of (30% after 24 h), 3a-hydroxy-pregn-4-ene-20-one (61% after 20a-hydroxy-pregn-4-ene-3-one (45% after 8 h, and 43% after

Fig. 6. Reduced AKR1C3 protein levels in HEC-1A 72 h after transfection. (A) Representative membrane with AKR1C3 and GAPDH staining is shown. M, markers; ssiRNA, protein fraction of HEC-1A cells after transfection with scrambled siRNA; siAKR1C3, protein fraction of HEC-1A cells after transfection with AKR1C3 specific siRNA. (B) Scatter dot plot of Western blotting with the specific anti-AKR1C3 antibodies performed on three samples of HEC-1A cells after siRNA-mediated gene silencing of AKR1C3 compared to three samples of negative controls. AKR1C3 protein levels were quantified, normalized to GAPDH levels, and analyzed with unpaired t test.

Table 8 Concentrations of the individual progesterone metabolites formed in Ishikawa cells after siRNA-mediated gene silencing 8 h after progesterone treatment.

Concentrations of individual steroid (ng/mL), according to gene silenced* AKR1C1/C2 AKR1C3 SRD5A1 Negative control Progesterone 624.0 ± 23.0 564.5 ± 34.5 729.0 ± 14.0 571.5 ± 25.5 20a-Hydroxy-pregn-4-ene-3-one 69.1 ± 6.2 156.5 ± 5.5 275.0 ± 24.0 165.5 ± 16.5 20b-Hydroxy-pregn-4-ene-3-one ND ND ND ND 3a-Hydroxy-pregn-4-ene-20-one ND 14.5 ± 0.6 16.1 ± 1.4 11.6 ± 1.2 3b-Hydroxy-pregn-4-ene-20-one 9.5 ± 0.2 9.5 ± 0.1 9.7 ± 0.6 9.3 ± 0 Pregn-4-ene-3a,20a-diol ND ND 4.8 ± 1.3 ND 5a-Pregnane-3,20-dione 16.4 ± 1.1 18.7 ± 1.5 5.4 ± 0.1 22.6 ± 3.4 20a-Hydroxy-5a-pregnane-3-one 13.6 ± 4.3 32.5 ± 5.8 9.2 ± 0.6 25.4 ± 7.0 3a-Hydroxy-5a-pregnane-20-one ND 7.7 ± 1.2 ND 16.6 ± 2.8 3b-Hydroxy-5a-pregnane-20-one 15.9 ± 3.9 12.5 ± 0.4 ND 6.1 ± 3.2 5a-Pregnane-3a,20a-diol 3.2 ± 1.9 20.6 ± 2.0 71.2 ± 58.8 56.7 ± 1.4 5a-Pregnane-3b,20a-diol 2.7 ± 1.2 9.7 ± 2.8 21.5 ± 18.6 7.1 ± 2.7

* Mean ± SEM; ND, not detected.

Table 9 Concentrations of the individual progesterone metabolites formed in Ishikawa cells after siRNA-mediated gene silencing 24 h after progesterone treatment.

Concentrations of individual steroid (ng/mL), according to gene silenced* AKR1C1/C2 AKR1C3 SRD5A1 Negative control Progesterone 139.0 ± 4.0 97.9 ± 23.1 270.5 ± 65.5 113.2 ± 63.8 20a-Hydroxy-pregn-4-ene-3-one 71.3 ± 11.6 120.3 ± 25.8 579.0 ± 38.0 171.0 ± 45.0 20b-Hydroxy-pregn-4-ene-3-one ND ND ND ND 3a-Hydroxy-pregn-4-ene-20-one ND 3.3 ± 0.0 14.1 ± 2 8.5 ± 2.7 3b-Hydroxy-pregn-4-ene-20-one 12.4 ± 1.9 9.2 ± 0.1 12.3 ± 0.2 9.1 ± 0.1 Pregn-4-ene-3a,20a-diol ND 8.2 ± 0.4 38.5 ± 1.4 14.9 ± 1.1 5a-Pregnane-3,20-dione 4.7 ± 3.2 2.3 ± 0.1 4.2 ± 2.9 8.2 ± 1.2 20a-Hydroxy-5a-pregnane-3-one 18.7 ± 3.9 30.2 ± 5.1 17.5 ± 1.4 38.4 ± 6.4 3a-Hydroxy-5a-pregnane-20-one 13.2 ± 8.1 ND 1.1 ± 0.2 13.9 ± 8.7 3b-Hydroxy-5a-pregnane-20-one ND ND ND ND 5a-Pregnane-3a,20a-diol 12.1 ± 3.6 72.8 ± 0.8 99.3 ± 61.7 162.0 ± 44.0 5a-Pregnane-3b,20a-diol 25.4 ± 8.5 17.7 ± 1.2 10.3 ± 2.3 30.0 ± 2.5

* Mean ± SEM; ND, not detected.

24 h), pregn-4-ene-3a,20a-diol (53% after 24 h), 20a-hydroxy-5a- decreases in the levels of these metabolites were not as consider- pregnane-3-one (60% after 8 h and 49% after 24 h), 5a-pregnan- able as when AKR1C1 and AKR1C2 were silenced. As decreased or 3a,20a-diol (57% after 24 h) and 5a-pregnane-3b,20a-diol (47% unchanged levels of progesterone were detected after 8 h and after 8 h, and 38% after 24 h). This again supports the role of 24 h, respectively, and as the total rate of metabolism also slightly AKR1C1/AKR1C2 in the reduction of steroidal 20-keto and 3-keto increased (1.1-fold after 8 h and 24 h), this again suggests that groups. there might be some alternative metabolic pathway activated after When AKR1C3 was silenced in HEC-1A cells, there were lower AKR1C3 is silenced. levels of 20a-hydroxy-pregn-4-ene-3-one (29% after 8 h and 28% As for Ishikawa cells, slower rates of metabolism were observed after 24 h), 20a-hydroxy-5a-pregnane-3-one (20% after 8 h, and when SRD5A1 was silenced in HEC-1A cells. Higher levels of 25% after 24 h), 5a-pregnane-3a,20a-diol (25% after 24 h) and progesterone (115%) were seen after 24 h, and higher levels of 5a-pregnane-3b,20a-diol (20% after 8 h and 15% after 24 h). These 20a-hydroxy-pregn-4-ene-3-one after 8 h and 24 h (106% and

Table 10 Concentrations of the individual progesterone metabolites formed in HEC-1A cells after siRNA-mediated gene silencing 8 h after progesterone treatment.

Concentrations of individual steroid (ng/mL), according to gene silenced* AKR1C1/C2 AKR1C3 SRD5A1 Negative control Progesterone 5130.0 ± 230.0 4990.0 ± 20.0 5100.0 ± 40.0 5155.0 ± 55.0 20a-Hydroxy-pregn-4-ene-3-one 402.0 ± 7.0 517.5 ± 3.5 774.5 ± 1.5 728.0 ± 55.0 20b-Hydroxy-pregn-4-ene-3-one 96.6 ± 5.4 84.6 ± 0.7 73.9 ± 0.4 76.1 ± 2.4 3a-Hydroxy-pregn-4-ene-20-one 68.9 ± 1.3 82.7 ± 3.8 77.3 ± 0.5 81.4 ± 1.5 3b-Hydroxy-pregn-4-ene-20-one 87.8 ± 4.3 88.1 ± 2.1 86.3 ± 1.2 88.9 ± 1.2 Pregn-4-ene-3a,20a-diol ND 7.7 ± 0.6 11.5 ± 0.4 11.2 ± 0.5 Pregn-4-ene-3a,20b-diol ND ND ND ND 5a-Pregnane-3,20-dione 76.8 ± 3.5 87.1 ± 3.8 17.5 ± 0.1 80 ± 3.6 20a-Hydroxy-5a-pregnane-3-one 8.2 ± 0.7 16.4 ± 0.5 4.2 ± 0.1 20.4 ± 1.6 3a-Hydroxy-5a-pregnane-20-one 10.3 ± 0.9 15.9 ± 0.1 ND 13.3 ± 0.2 3b-Hydroxy-5a-pregnane-20-one 48.1 ± 2.1 45.1 ± 0.3 7.3 ± 1.3 40.7 ± 1.7 5a-Pregnane-3a,20a-diol ND ND ND ND 5a-Pregnane-3b,20a-diol 5.0 ± 0.3 7.6 ± 0 2.3 ± 0.5 9.4 ± 0.6

* Mean ± SEM; ND, not detected.

Table 11 Concentrations of the individual progesterone metabolites formed in HEC-1A cells after siRNA-mediated gene silencing 24 h after progesterone treatment.

Concentrations of individual steroid (ng/mL), according to gene silenced* AKR1C1/C2 AKR1C3 SRD5A1 Negative control Progesterone 4650.0 ± 70.0 4345.0 ± 235.0 4855.0 ± 35.0 4210.0 ± 60.0 20a-Hydroxy-pregn-4-ene-3-one 1245.0 ± 5.0 1570.0 ± 50.0 2790.0 ± 0 2180.0 ± 70.0 20b-Hydroxy-pregn-4-ene-3-one 85.9 ± 1.3 77.3 ± 2.5 64.6 ± 0.7 68.9 ± 0.1 3a-Hydroxy-pregn-4-ene-20-one 107.0 ± 2.0 119.0 ± 2.0 119.5 ± 0.5 115.5 ± 0.5 3b-Hydroxy-pregn-4-ene-20-one 167.5 ± 3.5 164.5 ± 8.5 172.0 ± 2 152.5 ± 1.5 Pregn-4-ene-3a,20a-diol 49.9 ± 0 80.8 ± 2 130.5 ± 1.5 106.0 ± 4.0 Pregn-4-ene-3a,20b-diol 8.2 ± 0.6 8.6 ± 1 10.5 ± 0.6 9.8 ± 0.3 5a-Pregnane-3,20-dione 44.0 ± 0.7 42.5 ± 2.2 8.4 ± 0.5 44.8 ± 1.5 20a-Hydroxy-5a-pregnane-3-one 46.3 ± 0.9 68.2 ± 2.6 16.5 ± 0.3 90.9 ± 2.1 3a-Hydroxy-5a-pregnane-20-one 7.6 ± 2.2 6.9 ± 0.9 4.0 ± 0.9 7.9 ± 0.1 3b-Hydroxy-5a-pregnane-20-one 28.0 ± 1.3 27.2 ± 0.7 5.2 ± 0 24.7 ± 0 5a-Pregnane-3a,20a-diol 17.6 ± 0.8 30.4 ± 0.1 7.7 ± 0.4 40.6 ± 0.8 5a-Pregnane-3b,20a-diol 44.7 ± 0.3 61.5 ± 0.7 14.0 ± 1.6 72.6 ± 2.4

* mean ± SEM; ND, not detetected.

Fig. 7. Proposed progesterone metabolism pathway in the Ishikawa and HEC-1A model cell lines. The main pathway of progesterone metabolism in Ishikawa cells is given by the violet arrows (or arrows with open arrowheads), and in HEC-1A cells by the pink arrows (or arrows with solid arrowheads). Secondary pathways are shown with dashed arrows. First, either 20-keto group of progesterone is reduced to form 20a-hydroxy-pregn-4-ene-3-one, or progesterone is reduced at C5 to form 5a-pregnane-3,20-dione. Next, 20a-hydroxy-5a-pregnane-3-one is formed; this is then metabolized at the 3-keto group to form mainly 5a-pregnane-3a,20a-diol, and 5a-pregnane-3b,20a-diol to a lesser extent. Progesterone may also be metabolized to low levels of 4-pregnenes.

128%, respectively), while the concentrations of 5a-pregnanes higher catalytic efﬁciency for 20-keto reduction of progesterone were decreased (80% after 8 h, and 69% after 24 h). These data than AKR1C2 [13,14], our data here suggest that in Ishikawa and again support an important role for SRD5A1 in progesterone HEC-1A cells, 20-ketosteroid reduction of progesterone and 5a- metabolism. pregnanes are catalyzed mainly by AKR1C2 (Fig. 7), while AKR1C1 As seen from the gene expression proﬁles, AKR1C2 mRNA levels and AKR1C3 have secondary roles. All three AKR1C isoforms were 110-fold greater than those for AKR1C1 in Ishikawa cells, and catalyze 3-keto reduction of 5a-pregnane-3,20-dione to form 3a- 6,800-fold greater in HEC-1A cells. Although AKR1C1 has a 10-fold hydroxy-5a-pregnane-20-one, but do not catalyze the formation

Please cite this article in press as: M. Sinreih et al., Important roles of the AKR1C2 and SRD5A1 enzymes in progesterone metabolism in endometrial cancer model cell lines, Chemico-Biological Interactions (2014), http://dx.doi.org/10.1016/j.cbi.2014.11.012 M. Sinreih et al. / Chemico-Biological Interactions xxx (2014) xxx–xxx 11 of 3b-hydroxy-5a-pregnane-20-one. All three of these isoforms are References also involved in the formation of the dihydroxy 5a-pregnanes: 5a- pregnane-3a,20a-diol and 5a-pregnane-3b,20a-diol. [1] A.J. Ryan, B. Susil, T.W. Jobling, M.K. Oehler, Endometrial cancer, Cell Tissue Res. 322 (2005) 53–61. In both of these cell lines, siRNA-mediated silencing of SRD5A1 [2] J. Ferlay, E. Steliarova-Foucher, J. Lortet-Tieulent, S. Rosso, J.W. Coebergh, H. had the most drastic effects on progesterone metabolism. When Comber, D. Forman, F. Bray, Cancer incidence and mortality patterns in Europe: silenced, the rate of metabolism decreased, and thus higher estimates for 40 countries in 2012, Eur. J. Cancer 49 (2013) 1374–1403. progesterone concentrations were detected. As progesterone coun- [3] Y. Sonoda, R.R. Barakat, Screening and the prevention of gynecologic cancer: endometrial cancer, Best Pract. Res. Clin. Obstet. Gynaecol. 20 (2006) 363–377. teracts the mitogenic effects of estrogens and directs cells toward [4] A.E. Schindler, Progestogen deficiency and endometrial cancer risk, Maturitas differentiation, higher progesterone levels can protect against 62 (2009) 334–337. development and progression of EC. Based on our experimental [5] T. Kaku, H. Yoshikawa, H. Tsuda, A. Sakamoto, M. Fukunaga, Y. Kuwabara, M. Hataeg, S. Kodama, K. Kuzuya, S. Sato, T. Nishimura, M. Hiura, H. Nakano, T. data, we suggest that SRD5A1 represents a potential target in the Iwasaka, K. Miyazaki, T. Kamura, Conservative therapy for adenocarcinoma treatment of EC. Higher levels of progesterone were also detected and atypical endometrial hyperplasia of the endometrium in young women: when the AKR1C1/2 genes were silenced; therefore AKR1C2, which central pathologic review and treatment outcome, Cancer Lett. 167 (2001) 39– 48. is highly expressed in endometrial cancer, also shows potential as a [6] K. Niwa, K. Tagami, Z. Lian, K. Onogi, H. Mori, T. Tamaya, Outcome of fertility- new target. However, the precise role of AKR1C1 and AKR1C2 preserving treatment in young women with endometrial carcinomas, BJOG enzymes should be studied further by pharmacological approach. 112 (2005) 317–320. [7] T.J. Key, M.C. Pike, The dose–effect relationship between ‘unopposed’ oestrogens and endometrial mitotic rate: its central role in explaining and 4. Conclusions predicting endometrial cancer risk, Br. J. Cancer 57 (1988) 205–212. [8] J.J. Kim, T. Kurita, S.E. Bulun, Progesterone action in endometrial cancer, endometriosis, uterine fibroids, and breast cancer, Endocr. Rev. 34 (2013) 130– In the present study, we identified the progesterone 162. metabolites in Ishikawa and HEC-1A endometrial cancer cell lines. [9] T. Sugawara, E. Nomura, S. Fujimoto, Expression of enzyme associated with In Ishikawa cells, 50 nM progesterone was metabolized mainly to steroid hormone synthesis and local production of steroid hormone in endometrial carcinoma cells, J. Endocrinol. 180 (2004) 135–144. 20a-hydroxy-pregn-4-ene-3-one, 5a-pregnane-3,20-dione 20a- [10] S.A. Missmer, A.H. Eliassen, R.L. Barbieri, S.E. Hankinson, Endogenous estrogen, hydroxy-5a-pregnane-3-one, and 5a-pregnane-3a/b,20a-diol. androgen, and progesterone concentrations and breast cancer risk among The metabolism of 50 nM and 5 lM progesterone in HEC-1A cells postmenopausal women, J. Natl. Cancer Inst. 96 (2004) 1856–1865. [11] M. Sinreih, N. Hevir, T.L. Rizner, Altered expression of genes involved in proceeded rapidly, there were more 5a-pregnane metabolites progesterone biosynthesis, metabolism and action in endometrial cancer, formed at lower progesterone concentration and more 4-pregnene Chem. Biol. Interact. (2012). metabolites at higher progesterone concentration. The major dif- [12] T.M. Penning, Y. Jin, S. Steckelbroeck, T. Lanisnik Rizner, M. Lewis, Structure– function of human 3 alpha-hydroxysteroid dehydrogenases: genes and ference between these two cell lines is the rate of progesterone proteins, Mol. Cell. Endocrinol. 215 (2004) 63–72. metabolism and not the metabolic profile, as both cell lines metab- [13] N. Beranic, P. Brozic, B. Brus, I. Sosic, S. Gobec, T. Lanisnik Rizner, Expression of olized 50 nM progesterone mainly to 5a-pregnane metabolite. Our human aldo-keto reductase 1C2 in cell lines of peritoneal endometriosis: potential implications in metabolism of progesterone and dydrogesterone and gene expression study together with the siRNA-mediated gene inhibition by progestins, J. Steroid Biochem. Mol. Biol. 130 (2012) 16–25. silencing indicate that in these Ishikawa and HEC-1A cells, 20- [14] N. Beranic, S. Gobec, T.L. Rizner, Progestins as inhibitors of the human 20- ketosteroid reduction is catalyzed mainly by AKR1C2. AKR1C1/ ketosteroid reductases, AKR1C1 and AKR1C3, Chem. Biol. Interact. 191 (2011) AKR1C2 gene silencing resulted in increased levels of unmetabo- 227–233. [15] T.L. Rizner, T.M. Penning, Role of aldo-keto reductase family 1 (AKR1) enzymes lized progesterone, which further supports the important roles of in human steroid metabolism, Steroids 79 (2014) 49–63. these enzymes. Our experimental data also show that in both of [16] L. Wu, M. Einstein, W.M. Geissler, H.K. Chan, K.O. Elliston, S. Andersson, these cell lines, 5a-reduction is catalyzed by SRD5A1. The silencing Expression cloning and characterization of human 17 beta-hydroxysteroid dehydrogenase type 2, a microsomal enzyme possessing 20 alpha- of SRD5A1 had the most pronounced effect on progesterone hydroxysteroid dehydrogenase activity, J. Biol. Chem. 268 (1993) 12964– metabolism, with decreased rate of metabolism and concurrently 12969. elevated levels of unmetabolized progesterone. Based on these [17] T.M. Penning, M.E. Burczynski, J.M. Jez, C.F. Hung, H.K. Lin, H. Ma, M. Moore, N. Palackal, K. Ratnam, Human 3alpha-hydroxysteroid dehydrogenase isoforms data, we suggest that SRD5A1 and AKR1C2 represent potential (AKR1C1–AKR1C4) of the aldo–keto reductase superfamily: functional targets for treatment of EC. plasticity and tissue distribution reveals roles in the inactivation and formation of male and female sex hormones, Biochem. J. 351 (2000) 67–77. [18] Y. Higaki, N. Usami, S. Shintani, S. Ishikura, O. El-Kabbani, A. Hara, Selective Conflict of Interest and potent inhibitors of human 20alpha-hydroxysteroid dehydrogenase (AKR1C1) that metabolizes neurosteroids derived from progesterone, Chem. The authors declare that there are no conflicts of interest. Biol. Interact. 143–144 (2003) 503–513. [19] J.P. Wiebe, Progesterone metabolites in breast cancer, Endocr. Relat. Cancer 13 (2006) 717–738. Transparency Document [20] J.P. Wiebe, G. Zhang, I. Welch, H.A. Cadieux-Pitre, Progesterone metabolites regulate induction, growth, and suppression of estrogen- and progesterone receptor-negative human breast cell tumors, Breast Cancer Res. 15 (2013) R38. The Transparency document associated with this article can be [21] J.P. Wiebe, V. Dave, J.B. Stothers, Synthesis and characteristics of allylic 4- found in the online version. pregnene-3,20-diols found in gonadal and breast tissues, Steroids 47 (1986) 249–259. [22] T.L. Rizner, T. Smuc, R. Rupreht, J. Sinkovec, T.M. Penning, AKR1C1 and AKR1C3 may determine progesterone and estrogen ratios in endometrial cancer, Mol. Cell. Endocrinol. 248 (2006) 126–135. Acknowledgements [23] M. Nishida, K. Kasahara, M. Kaneko, H. Iwasaki, K. Hayashi, Establishment of a new human endometrial adenocarcinoma cell line, Ishikawa cells, containing estrogen and progesterone receptors, Nihon Sanka Fujinka Gakkai Zasshi 37 This study was supported by a Young Researcher Grant to M.S. (1985) 1103–1111. and a J3-5510 Grant to T.L.R., both from the Slovenian Research [24] G. Vollmer, Endometrial cancer: experimental models useful for studies on Agency. We thank Dr. Chris Berrie for manuscript editing. molecular aspects of endometrial cancer and carcinogenesis, Endocr. Relat. Cancer 10 (2003) 23–42. [25] H. Kurarmoto, M. Hamano, M. Imai, HEC-1 cells, Hum. Cell 15 (2002) 81–95. Appendix A. Supplementary data [26] E. Castro-Rivera, S. Safe, Estrogen- and antiestrogen-responsiveness of HEC1A endometrial adenocarcinoma cells in culture, J. Steroid Biochem. Mol. Biol. 64 (1998) 287–295. Supplementary data associated with this article can be found, in [27] D.W. Russell, J.D. Wilson, Steroid 5 alpha-reductase: two genes/two enzymes, the online version, at http://dx.doi.org/10.1016/j.cbi.2014.11.012. Annu. Rev. Biochem. 63 (1994) 25–61.

[28] J.A. Collins, D.M. Jewkes, Progesterone metabolism by proliferative and [30] J.P. Wiebe, M.J. Lewis, V. Cialacu, K.J. Pawlak, G. Zhang, The role of progesterone secretory human endometrium, Am. J. Obstet. Gynecol. 118 (1974) 179–185. metabolites in breast cancer: potential for new diagnostics and therapeutics, J. [29] J.A. Collins, D.M. Jewkes, Progesterone metabolism in adenocarcinoma of the Steroid Biochem. Mol. Biol. 93 (2005) 201–208. endometrium, Am. J. Obstet. Gynecol. 120 (1974) 779–784.