Selective Loss of AKR1C1 and AKR1C2 in Breast Cancer and Their Potential Effect on Progesterone Signaling

[CANCER RESEARCH 64, 7610–7617, October 15, 2004] Selective Loss of AKR1C1 and AKR1C2 in Breast Cancer and Their Potential Effect on Progesterone Signaling

Qing Ji,1 Chisa Aoyama,5 Yih-Dar Nien,2 Paul I. Liu,5 Peter K. Chen,5 Lilly Chang,3 Frank Z. Stanczyk,3,4 and Andrew Stolz1 Departments of 1Medicine, 2Surgery, 3Obstetrics and Gynecology, and 4Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California; and 5Department of Pathology, Olive View-UCLA Medical Center, University of California Los Angeles, Sylmar, California

ABSTRACT Initiative trial receiving estrogen and a progestin had increased incidence of breast cancer compared with those receiving estrogen alone. These Progesterone plays an essential role in breast development and cancer findings strongly suggest that exogenous progestin may be a causative formation. The local metabolism of progesterone may limit its interactions factor for breast tumor development (1, 2, 7, 8). with the progesterone receptor (PR) and thereby act as a prereceptor regulator. Selective loss of AKR1C1, which encodes a 20␣-hydroxysteroid The local regulation of hormone synthesis and catabolism are key dehydrogenase [20␣-HSD (EC 1.1.1.149)], and AKR1C2, which encodes a modifiers for the development and growth of hormone-dependent 3␣-hydroxysteroid dehydrogenase [3␣-HSD (EC 1.1.1.52)], was found in tumors, including breast cancer (9–11). For example, induction of 24 paired breast cancer samples as compared with paired normal tissues aromatase in breast cancer results in the in situ synthesis of estrogen, from the same individuals. In contrast, AKR1C3, which shares 84% promoting tumor growth, and aromatase inhibitors effectively prevent sequence identity, and 5␣-reductase type I (SRD5A1) were minimally breast cancer recurrence (10, 12). Thus, estrogen is an effective target affected. Breast cancer cell lines T-47D and MCF-7 also expressed re- for chemoprevention. Similarly, progesterone and/or its metabolites duced AKR1C1, whereas the breast epithelial cell line MCF-10A ex- may provide novel targets for chemoprevention of breast cancer or pressed AKR1C1 at levels comparable with those of normal breast tissues. treatment strategies for breast cancer because they act as either growth Immunohistochemical staining confirmed loss of AKR1C1 expression in breast tumors. AKR1C3 and AKR1C1 were localized on the same myo- promoters or cell cycle inhibitors in breast cancer cells (4, 6). Little is epithelial and luminal epithelial cell layers. Suppression of ARK1C1 and known about the pathways responsible for catabolism of progesterone AKR1C2 by selective small interfering RNAs inhibited production of in the human breast, but in rat placenta, robust induction of 20␣- 20␣-dihydroprogesterone and was associated with increased progesterone hydroxysteroid dehydrogenase [20␣-HSD (EC 1.1.1.149)] expression in MCF-10A cells. Suppression of AKR1C1 alone or with AKR1C2 in immediately precedes reduction in local progesterone levels at partu- T-47D cells led to decreased growth in the presence of progesterone. rition (13, 14). In humans, a similar increase in placental 20␣-HSD Overexpression of AKR1C1 and, to a lesser extent, AKR1C2 (but not activity occurs, but not in other pathways, implying a common mech- AKR1C3) decreased progesterone-dependent PR activation of a mouse anism for the local elimination of progesterone (15, 16). mammary tumor virus promoter in both prostate (PC-3) and breast ␣ (T-47D) cancer cell lines. We speculate that loss of AKR1C1 and AKR1C2 The 20 -HSDs are members of a newly emerging family of NADPH- in breast cancer results in decreased progesterone catabolism, which, in dependent cytosolic oxidoreductases, known as the aldo-keto reductase combination with increased PR expression, may augment progesterone super gene family (17, 18). AKR1C1 is one of four highly related human signaling by its nuclear receptors. hydroxysteroid dehydrogenases that share Ͼ84% sequence identity but have distinctive substrate specificities and tissue distributions. Three of INTRODUCTION these AKR1C family members are expressed in the human breast and catalyze the following reactions: (a) AKR1C1 reduces the carbonyl group In women, breast cancer is the most common noncutaneous malig- at the 20 position, and their products are weaker activators of the tran- nancy, and lifetime exposure to ovarian hormones is a well-recognized scriptional activity of PR (19); (b) AKR1C2 encodes for a 3␣-hydroxy- risk factor for breast cancer development (1, 2). The actions of both steroid dehydrogenase [3␣-HSD (EC 1.1.1.52)] that reduces the carbonyl estrogen and progesterone are required for normal growth and maturation group at the 3 position of progesterone metabolites that are also weak PR of breast tissues, and progesterone is required for terminal duct formation activators (20); and (c) AKR1C3, also referred to as 17␤-HSD type V, required for lactation (3). Estrogen can also regulate expression of the has minimal 20␣-HSD or 3␣-HSD activity for progesterone (21). Re- progesterone receptor (PR), linking the action of both these two hor- markably, AKR1C2 differs by only 7 of 323 amino acids from AKR1C1 mones and suggesting a complex interplay between estrogen and regu- (20, 22), and all three genes are in close proximity on chromosome lation of progesterone-dependent genes. Inconsistent results in breast 10p14, suggesting evolution by gene duplication (18, 23). cancer cell lines and animal studies have made it difficult to assess a role To assess whether AKR1C may act as a prereceptor regulator of of progesterone in either development or promotion of breast cancer hormone activity by adjusting the availability of hormones to act with (4–6). However, women receiving progestin are at greater risk for their cognate receptors, we recently demonstrated significant reduction of development of increased mammographic breast density, suggesting AKR1C2, which catalyzes reduction of dihydrotestosterone to the weak greater cellular proliferation (1, 7), and women in the Women’s Health androgen 3␣-androstanediol, in prostate cancer as compared with unaffected tissue (24). We hypothesized that selective reduction of this key Received 5/11/04; revised 7/12/04; accepted 8/11/04. androgen-catabolizing gene would result in maintenance of intracellular Grant support: University of Southern California/Norris Comprehensive Cancer Center grant 5P30 CA14089–29 from the National Cancer Institute; Robert E. and Mary levels of dihydrotestosterone in prostate tumors and thereby provide a R. Wright Foundation; Margaret E. Early Medical Research Foundation; and University selective growth advantage for these malignant cells. Because progester- of Southern California Research Center for Liver Disease DK98-016. This work was presented in part at the 85th ENDO Annual Meeting in Philadelphia, Pennsylvania on June one is likely to modify growth of breast cancer cells (4, 5), we now report 21, 2003. the relative expression of three AKR1C family members in paired breast The costs of publication of this article were defrayed in part by the payment of page cancer and normal tissue samples using our recently developed real-time charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. polymerase chain reaction (PCR) methodology (24). Effects of small Requests for reprints: Andrew Stolz, Hoffman Medical Research, Room 101A, 2011 interfering RNA (siRNA) suppression of AKR1C1 alone or with AKR1C2 Zonal Avenue, Los Angeles, CA 90033. Phone: 323-442-2699; Fax: 323-442-5425; E-mail: [email protected]. on progesterone metabolism and cellular proliferation were also deter- ©2004 American Association for Cancer Research. mined in MCF-10A and T-47D cell lines, indicating that AKR1C1 and 7610

AKR1C2 may act as prereceptor regulators of PR activity. The signifi- body, and color was developed using AEC Substrate Pack (NeoMarkers, cance of these findings has been extended to assess whether changes in Fremont, CA), according to the manufacturer’s protocols. Sections were coun- expression of AKR1C family members could modify progesterone- terstained with Meyer’s hematoxylin (NeoMarkers) and mounted in Glycergel dependent transcriptional activity of the mouse mammary tumor virus (Dako). Both preimmune sera and peptide-preabsorbed antisera were used as (MMTV) promoter by PR-B. controls for IHC staining specificity. Suppression of AKR1C1 and/or AKR1C2 Expression by Small Interfer- ing RNAs. Four siRNAs were designed by Qiagen, and the resultant sequence MATERIALS AND METHODS was compared with the human genome to assess their cross-reactivity with other AKR1C family members. Ultimately, the following ARK1C1 sequences Chemicals and Supplies. All chemicals were of molecular biology grade from M86609 were used and are listed with their percentage similarity to the or higher and were purchased from Sigma (St. Louis, MO), unless otherwise other AKR1C family members expressed in human breast tissue: Seq-A, stated. Molecular biology reagents were purchased from Promega (Madison, AAGCTTTAGAGGCCACCAAAT (85% sequence identity to both AKR1C2 WI), Roche Molecular Biochemicals (Indianapolis, IN), and Life Technolo- and AKR1C3); Seq-B, AACTGCTGGATTTCTGCAAGT (100% sequence gies, Inc. (Gaithersburg, MD). Cell culture supplies were purchased from identity to AKR1C2 and 90% sequence identity to AKR1C3); Seq-C, AAAGC- Invitrogen (Carlsbad, CA). CAGGTGAGGAAGTGAT (100% sequence identity to AKR1C2 and 81% Breast Tissues and RNA Extraction. Human breast samples were pro- sequence identity to AKR1C3); and Seq-D, AAGGTCACTGAAAAATCT- cessed according to our previously described method (24). Twenty-four paired TCA (100% sequence identity with AKR1C2 and 71% sequence identity with breast tumors and their paired surrounding unaffected tissues were selected AKR1C3). Transfection of siRNA (Qiagen) was performed according to the from the University of Southern California/Norris Comprehensive Cancer manufacturer’s protocol, and quantitative real-time PCR was used to monitor Center (Los Angeles, CA) or Olive View-University of California Los Angeles suppression of AKR1C family members. Medical Center (Sylmar, CA) after institutional review board approval. Sam- MCF-10A cells (8 ϫ 104) were seeded onto 6-well plates and grown for 24 ples were fresh frozen in liquid nitrogen, and sections (5 ␮m) were subse- hours to approximately 50% confluence. Small interfering RNAs (4.5 ␮g) were quently reviewed by pathologists for pathological diagnosis along with immu- diluted in 100 ␮L of suspension buffer, vortexed, and mixed with 15 ␮Lof nohistochemistry staining for estrogen receptor (ER), PR, Ki-67, and Her2/neu RNAiFect. Cells were incubated with the samples for 5 to 10 minutes at room status. Only paired samples in which breast tumors contained Ͼ90% tumor temperature, the medium was exchanged with 300 ␮L of growth medium, and cells and normal tissue lacking any tumor cells were used. Clinical information then cells were allowed to recover for 24 hours. Nonsilencing fluorescein-labeled and surgical pathology reports were available without any patient identifiers. siRNA (Qiagen) was used as the control for siRNA transfections. The frozen tissues were homogenized with tissue pulverizers (Spectrum Lab- Progesterone Metabolism. MCF-10A cells (8 ϫ 104) were seeded in 6-well oratories, Rancho Dominguez, CA), and total RNA was prepared using the tissue culture plates and cultured 24 hours before siRNA transfections. Twenty- RNeasy Mini Kit (Qiagen, Valencia, CA) as described previously (24). four hours after siRNA treatments, 0.2 mCi of [3H]progesterone (Perkin-Elmer Quantitative Real-Time Polymerase Chain Reaction Assay. Real-time Life Science, Boston, MA) was added to the media, and cells were then grown for PCR for AKR1C1, AKR1C2, AKR1C3, and SRD5A1 was performed as an additional 16 hours. Medium and cells were harvested separately, and metab- described previously (24). SRD5A1 (forward primer, 5Ј-ATCCTCCTG- olites were separated by thin-layer chromatography. Aliquots (0.5 mL) of lysed GCCATGTTCC-3Ј; reverse primer, 5Ј-TCGCATCAGAAACGGGTAA- cells and supernatants were extracted twice with ethyl acetate:hexane (3:2) and AT-3Ј; probe, fluorescein amidite-5ЈCGTCCACTACGGGCATCGGTGC- evaporated under nitrogen at 40°C to remove the solvent, and 10 ␮g of proges- 3Ј-black hole quencher) was designed using Primer Express version 2.0 terone and 20␣-dihydroprogesterone were added as carriers. Extracts were applied software (Applied Biosystems, Foster City, CA). Random primed cDNA to a thin-layer chromatography chromatography plate (Silica gel 60F254 on libraries were prepared for TaqMan quantitative PCR by using the Omni- 20 ϫ 20-cm aluminum sheet; EM Science, Gibbstown, NJ) and run for 1.5 hours script Kit (Qiagen) with random hexamers (Applied Biosystems), and using chloroform:diethyl ether (10:3) as solvent mixture. Locations of progester- relative expression was calculated as described previously (24). one and 20␣-dihydroprogesterone were identified by ultraviolet light, the regions Cell Culture. MCF-7, MCF-10A, T-47D, and PC-3 cell lines were all were scraped, and radioactivity was determined by liquid scintigraphy. purchased from American Type Culture Collection (Rockville, MD). Cell lines T-47D Proliferation Studies. T-47D cells were grown in RPMI 1640 with- were cultured in RPMI 1640 with 10% fetal bovine serum (FBS), except for out phenol red with 4% charcoal dextran-treated FBS (HyClone, Logan, UT) and MCF-10A, which was grown in Clonetics Mammary Epithelial Growth Me- treated with different concentrations of progesterone [0.1% in ethanol (v/v)] or dium (MEGM Bullet Kit, CC-3051, serum-free) supplemented with 100 with 0.1% (v/v) ethanol as controls. Cell proliferation assays were performed as ng/mL cholera toxin. Total RNA was isolated as described above for quanti- described previously (27, 28) with the following modifications: Twenty-four hours tative real-time PCR assays. Stable PC-3 prostate cancer cell lines expressing before initiation of growth studies, cells were treated with siRNA as described either AKR1C1, AKR1C2, or AKR1C3 were developed as described previ- above. A total of 1.5 ϫ 104 cells were then seeded into 12-well tissue culture ously (24) and used to assess activation of MMTV luciferase reporter assays. dishes from a common cell culture and grown for an additional 24 hours. Proges- Development and Characterization of AKR1C Polyclonal Antibodies. terone was then added, and half of the media was replaced on a daily basis. The Polyclonal antibodies from rabbits for individual AKR1C family members were number of cells in quadruplet wells was then determined daily with the 3-(4,5- developed using peptides and recombinant protein as immunogens. An AKR1C3- dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay using Bio-Rad Mi- specific antiserum was developed as described previously (25) using peptide croplate Reader Model 3550 (Bio-Rad, Hercules, CA) after a standard curve sequence N-GLDRNLHYFNSDSFASHPNYPYS to generate antiserum ␣7548. confirmed a linear relationship between cell number and absorbance. Rabbits injected with the peptide N-FNHRLLEMIL(C) were used to generate PR-B–Dependent Transactivation Assay. Permanently transfected PC-3 antiserum ␣6621, which recognizes both AKR1C1 and AKR1C3, but not cell lines with varying stable expression levels of AKR1C1, AKR1C2, or AKR1C2. Bacterial-expressed protein was used to immunize rabbits, and the AKR1C3 were used for progesterone-dependent transactivation experiments resultant antisera, ␣1850, recognizes all family members (24). Specificity of with cotransfection of human PR-B. These PC-3 cell lines were individually antisera was assessed by Western blots with lysates of PC-3 cell lines that stably transiently transfected with 2 ␮g of pMMTV-Luc (kindly provided by Dr. express individual AKR1C family members. Western blots were performed as Gerhard A. Coetzee, Keck School of Medicine at University of Southern described previously (24) using a 1:2,000 dilution for all antisera. California, Los Angeles, CA), 0.2 ␮g of pCMV-hPR-B expression plasmid Immunohistochemical Staining of Breast Tissues. Immunohistochemical (kindly provided by Dr. Michael R. Stallcup, Keck School of Medicine at (IHC) staining was performed according to the previously described method University of Southern California), and 5 ng of Renilla luciferase plasmid (26) with the following modifications. Five-micrometer sections of paraffin- pRL-SV40 (Promega) to control for transfection efficiency. Cells were trans- embedded breast tissues were cut and pretreated for 5 minutes with serum-free fected with 15 ␮L of SuperFect (Qiagen) in 600 ␮L of RPMI 1640 and allowed protein blocking solution (Dako, Carpinteria, CA) and then incubated with the to recover for 36 hours in RPMI 1640 with 4% charcoal dextran-treated FBS. primary antihuman AKR1C polyclonal antisera (␣6621, ␣7548, or ␣1850) at Cells were then washed with PBS and treated for 16 hours with 100 pmol/L 1:1,000 to 1:2,000 dilution overnight in a humidified chamber at 4°C. Slides progesterone in 4% charcoal dextran-treated fetal calf serum. Luciferase ac- were then incubated with antirabbit horseradish peroxidase secondary anti- tivity was measured using the Luminoskan Ascent instrument (Labsystems, 7611

Franklin, MA) and the Dual Luciferase Activity Kit (Promega). The ratio of blocked by preincubation with excess peptide, thereby confirming the firefly to Renilla luciferase activity was normalized to total protein, and specificity of IHC staining. luciferase activity of control PC-3 cells was arbitrarily defined as 100 Ϯ SD. As illustrated in Fig. 1B, IHC staining with both ␣6621 and ␣7548 revealed predominant staining on the myoepithelial and luminal epi- RESULTS thelial cells. Because ␣6621 cross-reacts with both AKR1C1 and ARK1C3, selective localization for AKR1C1 was determined in Relative Expression Profile of AKR1C Family Members and paired BT6 and BN6 tissue samples that lacked AKR1C3 expression. SRD5A1 in Paired Breast Tissue Samples. The relative expression In Fig. 1C, no IHC staining with ␣7548 was observed, confirming the profiles of AKR1C1, AKR1C2, AKR1C3, and SRD5A1 were determined lack of expression of AKR1C3. IHC staining with ␣6621, which in 24 paired breast cancer and normal tissue samples using RNase P as could only correspond to AKR1C1, was localized on the myoepithe- the internal control. RNase P expression was equivalent in tumor and lial and luminal epithelial cells. Endothelial cells were also stained normal tissue (data not shown). Table 1 lists the relative changes in gene with this antiserum (data not shown). Decreased IHC staining was expression for AKR1C family members and SRD5A1 in individual pairs observed on the corresponding tumor sample, in close agreement with of tissue samples, along with clinical features and PR, ER, Her2/neu, and the 48-fold decrease in relative expression of AKR1C1. Thus, Ki-67 status. A substantial (defined as Ͼ5-fold) decrease in AKR1C1 AKR1C1 is normally expressed in the same location as AKR1C3, expression was found in 13 of 24 cases. AKR1C2 expression was absent with staining observed on a majority of myoepithelial and luminal in tumors or reduced in 6 of these 13 cases. AKR1C3 expression was only epithelial cells, the latter being the same site of PR expression (3, 29). significantly reduced in one case without a significant decrease in ␣ ␣ AKR1C1. SRD5A1 expression was significantly decreased in only four IHC staining with 6621 and 7548 was then performed on all cases and increased (Ͼ5-fold) in three cases, and no apparent relationship available samples to determine the relationship between relative gene existed between reduced expression of SRD5A1 and that of AKR1C1.No expression profile and protein levels. Fig. 1D illustrates two examples of consistent pattern was found in the expression profile for AKR1Cs or paired tissue samples in which AKR1C1 mRNA expression was reduced SRD5A1 and PR, ER, Her2/neu, or Ki-67 status. Thus, reduced gene or absent in tumor samples associated with minimal changes in AKR1C3. ␣ expression of ARK1C1 appears to be unrelated to PR or ER status. Decreased staining of 6621 in both cases corresponds with AKR1C1 ␣ Development of AKR1C-Specific Antisera and Immunohisto- expression levels, whereas 7548 staining for AKR1C3 was unchanged chemical Staining in Breast Tissue Samples. Rabbits received injec- in accordance with the real-time PCR data. These cases suggest that tion with synthesized peptides to develop antisera that recognize specific decreased ␣6621 staining closely parallels AKR1C1 mRNA expression family members, despite the high sequence homology of family mem- patterns and confirm the selectivity of ␣7548 IHC staining for AKR1C3. bers. Lysates harvested from permanently transfected PC-3 cells express- Fig. 1E demonstrates additional cases in which ␣6621 IHC staining ing AKR1C1, AKR1C3, or AKR1C2 (24) were used to assess specificity matches AKR1C1 expression but not ARK1C3 expression. Finally, Fig. of rabbit antisera by Western blot. As illustrated in Fig. 1A, ␣1850 1F confirms that ␣7548 staining corresponds with relative changes in recognized all transfected AKR1C family members expressed in PC-3 AKR1C3 expression. Pathologists, who were unaware of relative expres- cells. We confirmed that ␣7548, designed according to Pelleiter et al. sion patterns, then independently compared the IHC staining intensity of (25), selectively recognized only ARK1C3, whereas ␣6621 recognized ␣6621 and ␣7548 for all available cases. Table 2 demonstrates that both AKR1C1 and AKR1C3, but not AKR1C2. As shown in Fig. 1B, approximately two thirds of the ␣7548 and ␣6621 IHC staining corre- both ␣6621 and ␣7548, but not their preimmune sera, stained normal sponded with relative expression of AKR1C1 and AKR1C3. Taken to- myoepithelial and luminal epithelial cells in the alveoli, which was gether, these findings confirm that decreased expression of AKR1C1

Table 1 Clinical features and changes in relative expression of AKR1C family members and SRD5A1 mRNAs in paired samples of breast tumors versus unaffected tissues Sample Age AKR1C1 AKR1C2 AKR1C3 SRD5A1 code Diagnosis (y) ER PR Ki-67 Her2 (mean Ϯ SD) (mean Ϯ SD) (mean Ϯ SD) (mean Ϯ SD) BTN1 Inv duct 64 ϩϩ15% ϩ 12.8 Ϯ 1.1 ND 6.5 Ϯ 0.3 84.5 Ϯ 6.5 BTN4 Inv duct 52 ϪϩNA ϩϪ3.7 Ϯ 0.4 1.9 Ϯ 0.1 1.1 Ϯ 0.1 30.3 Ϯ 2.2 BTN5 Inv lob 68 Ϫϩ 5% Ϫ 5.0 Ϯ 0.5 ND ND Ϫ24.5 Ϯ 2.1 BTN6 Inv lob 53 ϩϪ40% ϪϪ48.3 Ϯ 3.7 ND ND 1.9 Ϯ 0.2 BTN7 Inv duct 35 ϩϩ10% ϩ/ϪϪ14.7 Ϯ 1.3 ND ND 1.7 Ϯ 0.1 BTN8 Inv duct 46 ϪϪ90% ϪϪ342 Ϯ 25.8 Ϫ3.6 Ϯ 0.3 Ϫ4.9 Ϯ 0.3 Ϫ4.6 Ϯ 0.3 BTN9 Inv duct 33 ϪϪ20% ϩϪ285 Ϯ 16.9 Ϫ3.5 Ϯ 0.5 Ϫ1.5 Ϯ 0.1 Ϫ4.6 Ϯ 0.2 BTN10 Inv duct 50 ϩϩ 5% ϩ/Ϫ 1.4 Ϯ 0.1 ND Ϫ1.3 Ϯ 0.1 1.5 Ϯ 0.1 BTN13 Inv duct 56 ϩϩ20% NA Ϫ67.1 Ϯ 5.8 Ϫ2.4 Ϯ 0.2 Ϫ3.7 Ϯ 0.2 Ϫ6.6 Ϯ 0.4 BTN15 Muc CA 49 ϩϩ10% ϪϪ2.1 Ϯ 0.1 ND ND 1.7 Ϯ 0.1 BTN17 ADH 58 NA NA NA NA Ϫ3.6 Ϯ 0.2 1.7 Ϯ 0.1 1.4 Ϯ 0.1 Ϫ237 Ϯ 29.6 BTN19 Inv lob 58 ϩϩ10% ϪϪ330 Ϯ 25.4 Ϫ81.4 Ϯ 7.8 Ϫ3.6 Ϯ 0.2 Ϫ2.1 Ϯ 0.1 CHTN1 Inv duct 28 NA NA NA NA 6.4 Ϯ 0.2 2.4 Ϯ 0.2 Ϫ3.6 Ϯ 0.3 4.8 Ϯ 0.2 CHTN2 Inv duct 37 NA NA NA NA Ϫ42.6 Ϯ 3.7 3.6 Ϯ 0.3 2.9 Ϯ 0.3 6.6 Ϯ 0.4 BTN20 DCIS 50 ϪϪ15% ϩ 7.7 Ϯ 0.5 12.7 Ϯ 0.9 Ϫ1.5 Ϯ 0.1 1.1 Ϯ 0.1 BTN21 Inv duct 65 ϩ/Ϫϩ/Ϫ 20% ϩϪ63.3 Ϯ 4.4 Ϫ15.3 Ϯ 1.1 1.2 Ϯ 0.1 Ϫ1.2 Ϯ 0.1 BTN22 Inv duct 40 ϩϩ20% ϪϪ1198 Ϯ 83.8 ND Ϫ34.5 Ϯ 2.4 2.4 Ϯ 0.2 BTN23 Inv lob 47 ϩϩ15% ϪϪ10.9 Ϯ 0.8 Ϫ11.5 Ϯ 0.8 Ϫ1.2 Ϯ 0.1 Ϫ1.5 Ϯ 0.1 BTN24 Inv lob 72 ϩϪ15% ϪϪ199.2 Ϯ 14.1 Ϫ288 Ϯ 20.2 Ϫ6.5 Ϯ 0.5 Ϫ1.7 Ϯ 0.1 BTN26 Inv duct 37 ϩϩ10% ϩϪ2506 Ϯ 175.4 Ϫ24.6 Ϯ 1.7 Ϫ1.6 Ϯ 0.1 Ϫ2.9 Ϯ 0.2 BTN28 Inv duct 42 ϪϪ10% ϩ ND in cancer ND in cancer Ϫ3.6- Ϯ 0.3 Ϫ4.6 Ϯ 0.2 BTN30 Inv duct 62 ϩϩ25% ϩϪ1.6 Ϯ 0.1 2.8 Ϯ 0.2 Ϫ5.7 Ϯ 0.4 Ϫ9.2 Ϯ 0.6 BTN31 Inv duct 58 ϩϩ10% ϩ 24.2 Ϯ 1.7 8.1 Ϯ 0.6 15.5 Ϯ 1.1 Ϫ2.3 Ϯ 0.2 BTN32 Inv duct 51 ϩϩ20% Ϫ 5.5 Ϯ 0.4 2.9 Ϯ 0.2 1.2 Ϯ 0.1 1.6 Ϯ 0.1 NOTE. Numbers listed in the table represent fold changes in mRNA expression in tumor samples compared with paired normal samples relative to RNase P expression. A positive number refers to increased mRNA expression in tumors, whereas a negative number represents decreased expression in tumor as compared with unaffected tissue. Abbreviations: Inv duct, invasive ductal carcinoma; ND, not detectable in both normal and cancer tissues; NA, not available; Inv lob, invasive lobular carcinoma; Muc CA, mucinous carcinoma; ADH, atypical ductal hyperplasia; DCIS, ductal carcinoma in situ; ND in cancer, not detectable in cancer tissues. 7612

Fig. 1. IHC staining of ␣6621 and ␣7548 in paired breast cancer and normal tissue samples. A and B, specificity of ␣6621 and ␣7548 used for IHC localization of AKR1C family members in breast tissues. A. Specificity of antipeptide antibodies ␣6621 and ␣7548 compared with ␣1850, gen- erated against the entire AKR1C1, was determined by Western blot using lysates (30 ␮g) from PC-3 cells stably expressing AKR1C1, AKR1C2, or AKR1C3 compared with nontransfected PC-3 cells. B. Specificity and localization of IHC staining with ␣6621 and ␣7548 was determined by respectively staining normal breast tissue from samples BN13 and BN9 with preimmune, immune, and immune sera preabsorbed with corresponding peptides. Absence of staining with preimmune sera and by preabsorption of sera with peptide confirms the specificity of IHC staining for each antiserum. CϪF, comparison between IHC staining of ␣6621 and ␣7548 in paired breast tissues and mRNA levels. C. Tissue localization of AKR1C1 was determined in paired samples BN6 and BT6 because neither tumor nor normal tissue expressed AKR1C3. ␣7548 lacked IHC staining, confirming the absence of AKR1C3. In normal tissue, ␣6621 staining is seen on myoepithelial and luminal epithelial cells, which is lost in the corresponding tumor sample. Real-time PCR data for each of the IHC samples are listed to compare IHC staining with relative changes in mRNA expression. Rela- tive changes in AKR1C1 or AKR1C3 expression in tumor as compared with unaffected tissue are in- cluded (Ϫ, reduced tumor expression; ND, not detectable). D, decreased ␣6621 staining in two pairs of samples in which AKR1C3 staining and gene expression are minimally affected. Staining with ␣6621 paralleled changes in gene expression for AKR1C1, whereas ␣7548 staining remained relatively unchanged. E, IHC staining with ␣6621 in two paired tissue samples that express varying amounts of AKR1C1 and AKR1C3. F, IHC staining of ␣7548 in paired tissue samples that express varying amounts of AKR1C3.

mRNA identified in the tumor samples closely corresponds with loss of Reduced Relative Expression of AKR1C1 in Human Breast protein expression. Due to the lack of AKR1C2-specific antisera, we Cancer Cell Lines. Gene and protein levels of AKR1Cs were deter- suspect (but cannot confirm) that decreased AKR1C2 mRNA will be mined in the established human breast cancer cell lines MCF-7 and associated with loss of protein expression. T-47D as well as in human breast epithelial cell line MCF-10A (30).

Table 2 Immunohistochemical staining of ␣6621 and ␣7548 in paired breast samples compared with AKR1C1 and AKR1C3 gene expression profile. Sample Changes in AKR1C1 in Changes in AKR1C3 in code tumor vs. normal tissue (fold) ␣6621 in normal tissue ␣6621 in cancer tissue tumor vs. normal tissue ␣7548 in normal tissue ␣7548 in cancer tissue (BTN1 13 1؉ 2ϩ 72ϩ, myoepithelial cell 2ϩ, focally (10% BTN5 5 2ϩϩ/Ϫ to 1ϩ ND NA NA BTN6 Ϫ48 3ϩ, myoepi Ͼ luminal 1ϩ ND Negative Negative BTN13 Ϫ67 2ϩ, myoepi Ͼ luminal 1ϩϪ42ϩ, myoepiϾluminal Negative BTN17 Ϫ42ϩ, myoepi Ͼ luminal 1ϩ 12ϩ, myoepiϾluminal 1ϩ. myoepi BTN20 8 2ϩ 2ϩϪ1NANA BTN21 Ϫ63 2ϩϪto ϩ 11ϩ 1ϩ BTN22 Ϫ1198 2ϩ, myoepi Ͼ luminal 2ϩ, diffuse cytoplasm Ϫ35 Negative Negative BTN23 Ϫ11 2ϩ 2ϩϪ12ϩ Negative BTN24 Ϫ199 2ϩ, myoepi, luminal squamous 1ϩϪ71ϩ, myoepiϾluminal, Negative epithelial cell and smooth muscle smooth muscle BTN28 ND in cancer 3ϩ, myoepi Ͼ luminal ϩ/ϪϪ41ϩ, rare myoepithelium ϩ/Ϫ to 1ϩ, diffusely NOTE. The intensity of the staining was graded as follows on a scale of 1 to 3: negative, lack of staining; 1ϩ, weak staining; 2ϩ, moderate staining; and 3ϩ, strong staining. Distribution of the staining was reported as diffuse (staining present in a majority of the glands) or focal (staining on only some of the glands). Abbreviations: ND, not detectable; NA, not available; myoepi, myoepithelial cell; luminal, luminal epithelial cell. 7613

As shown in Fig. 2A, significantly less AKR1C1 was found in MCF-7 Progesterone Metabolism Catalyzed by AKR1C1 and AKR1C2. and T-47D cell lines as compared with MCF-10A, whereas AKR1C3 To assess the role of AKR1C1 and AKR1C2 on progesterone metab- was equivalently expressed in all cell lines. ␣7548 Western blot data olism, siRNAs were developed. Pilot results showed that Seq-A in Fig. 2B confirmed that AKR1C3 was equally expressed in all cell selectively suppressed AKR1C1, whereas Seq-D suppressed both lines, whereas use of ␣6621 showed significantly less AKR1C1 AKR1C1 and AKR1C2 without affecting expression of AKR1C3,as expression in MCF-7 and T-47D cell lines as compared with MCF- illustrated in Fig. 2C. The ability of these two siRNAs to alter 10A, which paralleled the AKR1C1 gene expression profile. progesterone metabolism was then monitored in MCF-10A cells,

Fig. 2. Suppression of AKR1C1 and AKR1C1/AKR1C2 in T-47D cells enhanced progesterone-mediated inhibition of cellular growth. A and B, reduced expression of AKR1C1 in established breast cancer cell lines. Expression levels of AKR1C1 and AKR1C3 in the breast cancer cell lines MCF-7 and T-47D and breast epithelial cell line MCF-10A were determined at both the mRNA and protein levels by real-time PCR and Western blot. A. Relative expression of AKR1C1 and AKR1C3 was determined using real-time PCR in triplicate compared with relative levels of MCF-7 cells (arbitrarily defined as 1 Ϯ SD). Relative expression is compared with a normal breast tissue. Substantially lower AKR1C1 levels were found in MCF-7 and T-47D breast cancer cell lines as compared with breast epithelial cell line MCF-10A and a normal breast tissue. No substantial changes in AKR1C3 expression were noted in all three cell lines. B. Protein levels in breast cell lines were determined by Western blot with ␣7548 and ␣6621. Relative equivalent amounts of AKR1C3 were found in all three cell lines, whereas decreased expression of AKR1C1 was found in MCF-7 and T-47D compared with MCF-10A. CϪE. Suppression of AKR1C1 alone or with AKR1C2 reduces formation of 20␣-dihydroprogesterone from progesterone in MCF-10A. C. AKR1C1-specific siRNA (Seq-A) substantially suppressed AKR1C1 expression but not AKR1C2 and AKR1C3 in MCF-10A cells compared with nonspecific siRNA (Ns)-treated MCF- 10A cells as a control. The other siRNA (Seq-D) was able to significantly suppress both AKR1C1 and AKR1C2 as compared with nonspecific siRNA-treated MCF-10A cells. AKR1C3 expression was relatively unaltered by either siRNA. D. Production of 20␣-dihydroprogesterone from progesterone in MCF-10A cells treated with AKR1C1-specific siRNA (Seq-A) or AKR1C1/AKR1C2- specific siRNA (Seq-D) was able to substantially reduce 20␣-dihydroprogesterone in both media and cell lysates. Note that the majority of the metabolite was distributed in the media. E. Decrease in 20␣-dihydroprogesterone is associated with increased progesterone in siRNA-treated cells. Suppression of both AKR1C1 and AKR1C2 leads to more unmetabolized progesterone in both cell lysates and media. FϪI. Suppression of AKR1C1 and AKR1C1/AKR1C2 in T-47D cells enhanced the suppression of cellular proliferation by progesterone. F. AKR1C1-specific siRNA (Seq-A) selectively suppressed AKR1C1 with minimal reduction of AKR1C2, whereas siRNA (Seq-D) suppressed both AKR1C1 and AKR1C2 in T-47D cells. AKR1C3 expression was not affected with either treatment. G. Cellular proliferation of T-47D treated with siRNA (Ns, nonspecific siRNA) demonstrated no significant changes in cellular growth with suppressed AKR1C1 alone or with suppressed AKR1C2 expression when progesterone was not presented. H and I. Growth of T-47D cells with siRNA treatments in the presence of 10Ϫ10 (H)or10Ϫ9 mol/L (I) progesterone demonstrated a substantial decrease in cell numbers, with loss of AKR1C1 alone or in combination with AKR1C2. Concentrations of 10Ϫ10 and 10Ϫ9 mol/L progesterone were selected based on our pilot study of the effects of progesterone on T-47D growth inhibition. It confirmed that 10Ϫ11 and 10Ϫ8 mol/L progesterone had effects on T-47D growth inhibition similar to published studies (refs. 41, 42; data not shown). 7614

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2004 American Association for Cancer Research. DECREASED AKR1C1 AND AKR1C2 IN BREAST CANCER which express levels of AKR1C1 comparable with those found in normal tissue. In Fig. 2D and E, suppression of AKR1C1 in MCF-10A cells significantly reduced the conversion of progesterone to 20␣- dihydroprogesterone in media and cell lysates. Suppression of AKR1C1 and AKR1C2 leads to additional and significant inhibition of 20␣-dihydroprogesterone formation. This reduction in 20␣-dihydroprogesterone correlated with a significant increase in progesterone in both media and cell lysates. Suppression of AKR1C1 Alone or with ARK1C2 Affects Proges- terone-Mediated Inhibition of Cellular Growth of T-47D. To assess potential loss of AKR1C1 and AKR1C2 on progesterone-dependent growth, siRNA suppression of AKR1C1 and AKR1C1/AKR1C2 in T-47D cells was determined because they endogenously express PRs. In Fig. 2F, suppression of AKR1C1 alone or with AKR1C2 is illustrated. The same cells were then used for proliferation assays with progesterone treatments. As shown in Fig. 2G, no significant changes were found in those cell lines treated with siRNAs in the absence of progesterone In Fig. 2H and I, progressive inhibition of cellular proliferation was observed with suppression of both AKR1C1 and AKR1C1/AKR1C2 in the presence of 10Ϫ9 or 10Ϫ10 mol/L progesterone, suggesting that reduced metabolism of progesterone in T-47D is responsible for progesterone-mediated inhibition of cellular growth. Inhibition of Progesterone-Dependent Transactivation by AKR1C1 and AKR1C2. To determine whether progesterone catabolism can alter PR-B ligand-dependent activity, the activity of a MMTV luciferase reporter gene was evaluated in the prostate cancer cell line PC-3 stably expressing AKR1C1, AKR1C2, or AKR1C3. These cells, along with PC-3 cells stably transfected with vector alone, were transiently transfected with pCMV-hPR-B, a luciferase reporter driven by the MMTV promoter, as well as a Renilla luciferase to control for transfection efficiency. In the absence of progesterone or PR-B, no activity was detectable (data not shown). In Fig. 3A, two different expression levels of AKR1C1 and AKR1C2, but not AKR1C3, in PC-3 cells were able to inhibit progesterone-dependent activation of the MMTV promoter, indicating that AKR1C1 and AKR1C2 metabolize progesterone to weaker activators of PR-B (17). Despite low levels of AKR1C1 expression, AKR1C1 completely inhibited luciferase activity, confirming that 20␣- Fig. 3. Effect of AKR1C1, AKR1C2, and AKR1C3 on PR-B–dependent transactiva- HSD can effectively block PR-B activation. Despite adequate levels of tion of MMTV promoter. A. Prostate cancer PC-3 cell lines stably expressing different amounts of AKR1C1, AKR1C2, and AKR1C3 as determined by Western blot with ␣1850, AKR1C3, AKR1C3 failed to inhibit progesterone-dependent activation, along with vector-transfected PC-3 cell line as a negative control, were transiently confirming that progesterone is a poor substrate for AKR1C3. These transfected with pMMTV-Luc (2 ␮g), pCMV-hPR-B expression plasmid (0.2 ␮g), and findings were extended to T-47D, which endogenously expresses PRs. Renilla luciferase plasmid pRL-SV40 (5 ng) to control for transfection efficiency. Pro- gesterone (100 pmol/L) was used to treat transfected cells for 16 hours before luciferase As shown in Fig. 3B, different amounts of AKR1C1 plasmid transfected assays. Ratio of firefly to Renilla luciferase activities was normalized to total protein to into T-47D cells together with MMTV luciferase reporter plasmid sig- compare relative promoter activity from different cell lines. Luciferase activity of control PC-3 cells was arbitrarily defined as 100 Ϯ SD in three independent experiments nificantly inhibited luciferase activity. performed in triplicate. Relative luciferase activities of modified PC-3 cell lines expressing different levels of AKR1C family members are compared with those of control PC-3 cells. Statistical significance was determined by comparing changes in luciferase activity DISCUSSION of the PC-3 cells expressing AKR1C family members versus PC-3 control cells using Student’s t test. B. T-47D was used to confirm the findings presented in A by transient Catabolism of steroid hormones can modulate availability of critical transfection with pMMTV-Luc (1 ␮g) and two concentrations of pSVL-AKR1C1 using ligands for their cognate transcription factors and thereby function as 100 pmol/L progesterone for 16 hours before luciferase assays. Luciferase activity of control T-47D cells was arbitrarily defined as 100 Ϯ SD in two independent experiments effective prereceptor regulators of gene expression. This type of regula- performed in triplicate. Relative luciferase activity of the T-47D cell line transfected with tion is best characterized for the mineralocorticoid receptor, in which different amounts AKR1C1 plasmid was compared with control cells. Student’s t test ␤ compared changes in luciferase activities between each T-47D cell line transfected with selective expression of 11 -hydroxysteroid dehydrogenase type II in AKR1C1 and the T-47D control. aldosterone target tissue prevents inappropriate activation of the mineralocorticoid receptor by cortisol (31). A similar paradigm is also devel- oping for the prereceptor regulation of progesterone activity by AKR1C1 may reach up to 300 to 700 nmol/L by the end of the third trimester (34). (17, 31). For example, as the human placenta comes to term, increased The function of AKR1C1 in the kidney may thus imitate that of 11␤- 20␣-HSD (but not 3␣-HSD or 17␤-hydroxysteroid dehydrogenase) ac- hydroxysteroid dehydrogenase type II to prevent inappropriate occupa- tivity is observed, implicating it as the major catabolic pathway for tion of ligand-dependent transcription factor. progesterone elimination (15, 16). Besides the placenta, studies in the We initially assessed the relative expression of ARK1C1 and two human kidney further support this concept because expression of highly related family members using a gene-specific real-time PCR. AKR1C1 protects the mineralocorticoid receptor from binding with pro- Although mRNA levels of the AKR1C family members had been quan- gesterone, an effective aldosterone antagonist (32, 33). This may be tified by semiquantitative reverse transcription-PCR and RNase protec- especially important during pregnancy, when serum progesterone levels tion methods (35, 36), real-time PCR provides a reliable and highly 7615

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2004 American Association for Cancer Research. DECREASED AKR1C1 AND AKR1C2 IN BREAST CANCER reproducible method to monitor large changes in relative gene expres- 47, 48). In the recent Women Health Initiative Study, increased sion. On average, we noticed that approximately half of the 24 cases incidence of total and invasive breast cancer was observed in the demonstrated a substantial loss of AKR1C1 expression, defined as a estrogen ϩ progestin group as compared with the placebo group (7). Ͼ5-fold reduction in tumor as compared with paired unaffected tissue. A The risks of invasive lobular carcinoma and invasive ductal carcinoma similar reduction in AKR1C2 expression was also found in half of these were also significantly increased among women who used combined cases, whereas AKR1C3 expression was relatively unchanged. Reduction estrogen ϩ progestin therapy (with continuous or sequential progestin in AKR1C1 expression was observed in a previous microarray study of use), but not among those who solely used estrogen therapy (49), human breast cancer, and loss of 20␣-HSD activity occurs in 7,12- strongly implicating progesterone in combination with estrogen as dimethylbenz(a)anthracene-induced breast tumors in rats (37, 38). Others stimulating cellular proliferation. reported a similar reduction in AKR1C1 and AKR1C2 gene expression by To explain how loss of AKR1C1 and AKR1C2 expression could reverse transcription-PCR in nine breast cancer tumors as compared with modify progesterone-dependent PR-B activation, a progesterone-depen- unaffected surrounding tissue from the same individuals (39). They also dent MMTV reporter activity was monitored in two different cell lines. noted a significant increase in SRD5A1 expression in tumors as compared Prostate cancer PC-3 cells stably expressing either AKR1C1 or AKR1C2, with normal breast tissues in contrast to our findings, which may be due but not AKR1C3, were able to reduce progesterone-dependent transac- to differences in the tissue sample pool sizes, stage of tumor samples, and tivation of the MMTV reporter construct with cotransfected PR-B. accuracy of the RNA quantitative techniques. We confirmed that de- AKR1C1 was able to completely inhibit progesterone-dependent activity, creased protein expression of AKR1C1 was associated with the previ- despite its low level of expression. Increasing concentrations of proges- ously reported reduction in AKR1C1 gene expression in the MCF-7 and terone were able to overcome this kind of inhibition (data not shown), T-47D breast cancer cell lines in contrast to the breast epithelial cell line indicating that metabolism, not some nonspecific process, is responsible MCF-10A (35). This pattern differed from that of AKR1C3, which was for reduced PR-B activation. Studies in the T-47D cell line confirmed that equally expressed in all cell lines. Furthermore, reduced AKR1C1 and AKR1C1 was able to inhibit progesterone-dependent PR-B activation in AKR1C2 expression in MCF-10A cells by siRNA suppression leads to a breast-derived cell line. In addition to 20␣-HSD, the 3␣-HSD of decreased production of 20␣-hydroxyl or 3␣-hydroxyl progesterone me- AKR1C2 can also inhibit progesterone-dependent activation of PR-B, tabolites. Thus, decreased expression of AKR1C1 and AKR1C2 observed although not as effectively as AKR1C1. in human samples is mimicked in certain breast cancer cell lines, which In normal breast tissues, AKR1C1 is localized predominately in the will prove useful for future studies on the AKR1C gene family. myoepithelial cells and in the luminal epithelial cells, which also express Cellular localization of specific AKR1C members is difficult to PR (3, 50). We speculate that expression of AKR1C1 and possibly determine due to their high sequence similarity. Immunohistochem- AKR1C2 in normal tissue may regulate progesterone-dependent gene istry with AKR1C3-specific antiserum ␣7548 prominently stained expression by limiting progesterone’s interactions with nuclear PRs. This myoepithelial and luminal epithelial cells as observed previously (25), observation resembles our recent finding of selective reduction in and AKR1C1 was localized to the same epithelial layers in samples AKR1C2 expression, but not ARK1C3, in prostate cancer samples (24), that lacked ARK1C3 expression (25). Although ␣6621 was able to suggesting that specific yet highly related AKR1C family members can recognize both AKR1C1 and AKR1C3 using the Western blot tech- act as prereceptor regulators for ligand-dependent transcription factors. nique, ␣6621 IHC staining closely matched AKR1C1 gene expression, Thus, AKR1C1 may limit progesterone-dependent signaling of nuclear and ␣6621 IHC staining could be used to evaluate relative AKR1C1 PRs and thereby indirectly regulate gene expression. This is compatible expression in pathological samples. with the role of AKR1C1 to limit the inhibitory interactions of proges- The MCF-10A cell line was used to assess the physiologic function terone with mineralocorticoid receptor. of AKR1C1 because it expresses AKR1C1 at levels comparable with We hypothesize that loss of key catabolic enzymes in breast cancer those found in normal breast tissues. Selective reduction of AKR1C1 tissues in combination with increased expression of PR may enhance by siRNA suppression in MCF-10A leads to a decrease in conversion progesterone-dependent gene expression. In addition to increased proges- of progesterone to 20␣-dihydroprogesterone, present in both cell terone levels and differences in progesterone metabolites as a result of lysate and media, which was further reduced with loss of AKR1C2. loss of AKR1C1 or AKR1C2, changes in relative expression of PR-A Substantial decrease in 20␣-dihydroprogesterone production was associated with a corresponding increase in progesterone levels in MCF-10A cells. Progesterone-mediated inhibition of growth in T-47D cells was also observed with suppression of AKR1C1, which was further enhanced in the absence of AKR1C2 expression. The loss of these catabolic enzymes in breast cancer presumably leads to increased progesterone levels, which, in combination with increased PR, could enhance progesterone-responsive gene expression. Confounding findings have been reported in the literature on the proliferative effects of progesterone in established human breast cell lines, with a majority of studies reporting growth-inhibitory effects (6, 28, 40–43). However, progesterone was found to cause cells to enter one mitotic cycle with induction of genes associated with cell growth (44, 45). Although progestins are proliferative when administered in a transient or sequential manner, sustained treatment results in growth arrest (46). Our data demonstrated that suppression of AKR1C1 or AKR1C1 and AKR1C2 can significantly inhibit cellular proliferation by progesterone as compared with control cell lines. The role of progesterone as an antiproliferative agent in cell lines Fig. 4. Potential role of AKR1C1 as prereceptor regulator of progesterone-dependent gene expression. Potential role of AKR1C1 in regulating availability of progesterone (F) to interact does not mimic its function in vivo. A majority of studies in animals with nuclear PRs by shunting a certain percentage of progesterone to metabolites (‚) that are and humans find that progestin treatment promotes growth in vivo (40, weaker transcriptional activators and thereby reduce PR signaling by progesterone. 7616

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2004 American Association for Cancer Research. DECREASED AKR1C1 AND AKR1C2 IN BREAST CANCER and PR-B in breast cancer could also alter the spectrum of progesterone- 23. Lou H, Hammond L, Sharma V, et al. Genomic organization and chromosomal regulated genes (51). Wiebe et al. (28, 35) have also reported a differ- localization of a novel human hepatic dihydrodiol dehydrogenase with high affinity bile acid binding. J Biol Chem 1994;269:8416–22. ential effect of specific progesterone metabolites as either growth pro- 24. Ji Q, Chang L, VanDenBerg D, et al. Selective reduction of AKR1C2 in prostate moters or inhibitors of the cell cycle in the MCF-7 cell line, suggesting a cancer and its role in DHT metabolism. Prostate 2003;54:275–89. 25. Pelletier G, Luu-The V, Tetu B, et al. Immunocytochemical localization of type 5 specific role for progesterone metabolites. The appearance of specific 17beta-hydroxysteroid dehydrogenase in human reproductive tissues. J Histochem progesterone metabolite binding sites on MCF-7 membranes further Cytochem 1999;47:731–8. supports a role for progesterone metabolites as signaling molecules (52). 26. Han YP, Nien YD, Garner WL. Recombinant human platelet-derived growth factor and transforming growth factor-beta mediated contraction of human dermal fibroblast Thus, metabolites of progesterone may also have unsuspected effects on populated lattices is inhibited by Rho/GTPase inhibitor but does not require phos- cell growth. Fig. 4 illustrates the potential role of AKR1C1 and AKR1C2 phatidylinositol-3Ј kinase. Wound Repair Regen 2002;10:169–76. functioning as a prereceptor regulator of progesterone interactions with 27. Syed V, Ulinski G, Mok SC, et al. M. Expression of gonadotropin receptor and growth responses to key reproductive hormones in normal and malignant human PRs. Loss of AKR1C1 in combination with increased PR in selective ovarian surface epithelial cells. Cancer Res 2001;61:6768–76. breast cancers could therefore provide therapeutic opportunities to block 28. Wiebe JP, Muzia D, Hu J, et al. The 4-pregnene and 5alpha-pregnane progesterone metabolites formed in nontumorous and tumorous breast tissue have opposite effects ligand-dependent activation of PRs by using antiprogestational agents. on breast cell proliferation and adhesion. Cancer Res 2000;60:936–43. 29. Anderson E. The role of oestrogen and progesterone receptors in human mammary development and tumorigenesis. Breast Cancer Res 2002;4:197–201. ACKNOWLEDGMENTS 30. Soule HD, Maloney TM, Wolman SR, et al. Isolation and characterization of a spontaneously immortalized human breast epithelial cell line, MCF-10. Cancer Res We thank Dr. Mike Stallcup and Dr. Michael Press (Keck School of 1990;50:6075–86. 31. Tomlinson JW, Stewart PM. Cortisol metabolism and the role of 11beta-hydroxy- Medicine at University of Southern California) for their thoughtful discussions steroid dehydrogenase. Best Pract Res Clin Endocrinol Metab 2001;15:61–78. and review of the manuscript. 32. Bumke-Vogt C, Bahr V, Diederich S, et al. Expression of the progesterone receptor and progesterone-metabolising enzymes in the female and male human kidney. J Endocrinol 2002;175:349–64. REFERENCES 33. Quinkler M, Johanssen S, Bumke-Vogt C, et al. Enzyme-mediated protection of the mineralocorticoid receptor against progesterone in the human kidney. Mol Cell 1. Ross RK, Paganini-Hill A, Wan PC, et al. Effect of hormone replacement therapy on Endocrinol 2001;171:21–4. breast cancer risk: estrogen versus estrogen plus progestin. J Natl Cancer Inst 34. Johansson ED, Jonasson LE. Progesterone levels in amniotic fluid and plasma from (Bethesda) 2000;92:328–32. women. I. Levels during normal pregnancy. Acta Obstet Gynecol Scand 1971;50: 2. Pike MC, Spicer DV, Dahmoush L, et al. Estrogens, progestogens, normal breast cell 339–43. proliferation, and breast cancer risk. Epidemiol Rev 1993;15:17–35. 35. Wiebe JP, Lewis MJ. Activity and expression of progesterone metabolizing 5alpha- 3. Soyal S, Ismail PM, Li J, et al. Progesterone’s role in mammary gland development reductase, 20alpha-hydroxysteroid oxidoreductase and 3alpha(beta)-hydroxysteroid and tumorigenesis as disclosed by experimental mouse genetics. Breast Cancer Res oxidoreductases in tumorigenic (MCF-7, MDA-MB-231, T-47D) and nontumorigenic 2002;4:191–6. (MCF-10A) human breast cancer cells. BMC Cancer 2003;3:9. 4. Lange CA, Richer JK, Horwitz KB. Hypothesis: progesterone primes breast cancer 36. Burczynski ME, Lin HK, Penning TM. Isoform-specific induction of a human cells for cross-talk with proliferative or antiproliferative signals. Mol Endocrinol aldo-keto reductase by polycyclic aromatic hydrocarbons (PAHs), electrophiles, and 1999;13:829–36. oxidative stress: implications for the alternative pathway of PAH activation catalyzed 5. Pasqualini JR, Paris J, Sitruk-Ware R, et al. Progestins and breast cancer. J Steroid by human dihydrodiol dehydrogenase. Cancer Res 1999;59:607–14. Biochem Mol Biol 1998;65:225–35. 37. Perou CM, Jeffrey SS, Van de Rijn M, et al. Distinctive gene expression patterns in 6. Clarke CL, Sutherland RL. Progestin regulation of cellular proliferation. Endocr Rev human mammary epithelial cells and breast cancers. Proc Natl Acad Sci USA 1990;11:266–301. 1999;96:9212–7. 7. Chlebowski RT, Hendrix SL, Langer RD, et al. Influence of estrogen plus progestin 38. Mori M, Tominaga T, Tamaoki BI. Steroid metabolism in normal mammary gland on breast cancer and mammography in healthy postmenopausal women: the Women’s and in the dimethylbenzanthracene-induced mammary tumor of rats. Endocrinology Health Initiative Randomized Trial. JAMA 2003;289:3243–53. 1978;102:1387–97. 8. Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus 39. Lewis MJ, Wiebe JP, Heathcote JG. Expression of progesterone metabolizing enzyme progestin in healthy postmenopausal women: principal results from the Women’s genes (AKR1C1, AKR1C2, AKR1C3, SRD5A1, SRD5A2) is altered in human breast Health Initiative Randomized Controlled Trial. JAMA 2002;288:321–33. carcinoma. BMC Cancer 2004;4:27. 9. Zhu BT, Conney AH. Functional role of estrogen metabolism in target cells: review 40. Horwitz KB. The molecular biology of RU486. Is there a role for antiprogestins in the and perspectives. Carcinogenesis (Lond) 1998;19:1–27. treatment of breast cancer? Endocr Rev 1992;13:146–63. 10. Harada N. Aromatase and intracrinology of estrogen in hormone-dependent tumors. 41. Schoonen WG, Joosten JW, Kloosterboer HJ. Effects of two classes of progestagens, Oncology (Basel) 1999;57(Suppl 2):7–16. pregnane and 19-nortestosterone derivatives, on cell growth of human breast tumor 11. Sasano H, Suzuki T, Harada N. From endocrinology to intracrinology. Endocr Pathol cells. II. T47D cell lines. Steroid Biochem Mol Biol 1995;55:439–44. 1998;9:9–20. 42. Schoonen WG, Joosten JW, Kloosterboer HJ. Effects of two classes of progestagens, 12. Chen S, Zhou D, Okubo T, et al. Prevention and treatment of breast cancer by pregnane and 19-nortestosterone derivatives, on cell growth of human breast tumor suppressing aromatase activity and expression. Ann N Y Acad Sci 2002;963:229–38. cells. I. MCF-7 cell lines. Steroid Biochem Mol Biol 1995;55:423–37. 13. Stocco CO, Zhong L, Sugimoto Y, et al. Prostaglandin F2 alpha-induced expression 43. Lin VC, Ng EH, Aw SE, Tan MG, et al. Progestins inhibit the growth of MDA-MB- of 20alpha-hydroxysteroid dehydrogenase involves the transcription factor NUR77. 231 cells transfected with progesterone receptor complementary DNA. Clin Cancer J Biol Chem 2000;275:37202–11. Res 1999;5:395–403. 14. Wiest WG, Kidwell WR, Balogh K. Progesterone catabolism in the rat ovary: a 44. Musgrove EA, Lee CS, Sutherland RL. Progestins both stimulate and inhibit breast regulatory mechanism for progestational potency during pregnancy. Endocrinology cancer cell cycle progression while increasing expression of transforming growth 1968;82:844–59. factor alpha, epidermal growth factor receptor, c-fos, and c-myc genes. Mol Cell Biol 15. Milewich L, Gant NF, Schwarz BE, et al. Initiation of human parturition. IX. 1991;11:5032–43. Progesterone metabolism by placentas of early and late human gestation. Obstet 45. Hissom JR, Moore MR. Progestin effects on growth in the human breast cancer cell Gynecol 1978;51:278–80. line T-47D: possible therapeutic implications. Biochem Biophys Res Commun 1987; 16. Diaz-Zagoya JC, Wiest WG, Arias F. Metabolism of progesterone by placentas from 145:706–11. several mammalian species in vitro. Am J Obstet Gynecol 1979;135:809–13. 46. Eden J. Progestins and breast cancer. Am J Obstet Gynecol 2003;188:1123–31. 17. Penning TM. Molecular endocrinology of hydroxysteroid dehydrogenases. Endocr 47. Longacre TA, Bartow SA. A correlative morphologic study of human breast and Rev 1997;18:281–305. endometrium in the menstrual cycle. Am J Surg Pathol 1986;10:382–93. 18. Vergnes L, Phan J, Stolz A, et al. A cluster of eight hydroxysteroid dehydrogenase 48. Graham JD, Clarke CL. Physiological action of progesterone in target tissues. Endocr genes belonging to the aldo-keto reductase supergene family on mouse chromosome Rev 1997;18:502–19. 13. J Lipid Res 2003;44:503–11. 49. Li CI, Malone KE, Porter PL, et al. Relationship between long durations and different 19. Wilks JW, Spilman CH, Campbell JA. Steroid binding specificity of the hamster regimens of hormone therapy and risk of breast cancer. JAMA 2003;289:3254–63. uterine progesterone receptor. Steroids 1980;35:697–706. 50. Conneely OM, Mulac-Jericevic B, DeMayo F, et al. Reproductive functions of 20. Hara A, Matsuura K, Tamada Y, et al. Relationship of human liver dihydrodiol progesterone receptors. Recent Prog Horm Res 2002;57:339–55. dehydrogenases to hepatic bile-acid-binding protein and an oxidoreductase of human 51. Richer JK, Jacobsen BM, Manning NG, et al. Differential gene regulation by the two colon cells. Biochem J 1996;313:373–6. progesterone receptor isoforms in human breast cancer cells. J Biol Chem 2002;277: 21. Higaki Y, Usami N, Shintani S, et al. Selective and potent inhibitors of human 5209–18. 20alpha-hydroxysteroid dehydrogenase (AKR1C1) that metabolizes neurosteroids 52. Weiler PJ, Wiebe JP. Plasma membrane receptors for the cancer-regulating derived from progesterone. Chem Biol Interact 2003;143–144:503–13. progesterone metabolites, 5alpha-pregnane-3,20-dione and 3alpha-hydroxy-4- 22. Zhang Y, Dufort I, Rheault P, et al. Characterization of a human 20alpha-hydroxy- pregnen-20-one in MCF- 7 breast cancer cells. Biochem Biophys Res Commun steroid dehydrogenase. J Mol Endocrinol 2001;25:221–8. 2000;272:731–7.

7617

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2004 American Association for Cancer Research. Selective Loss of AKR1C1 and AKR1C2 in Breast Cancer and Their Potential Effect on Progesterone Signaling

Qing Ji, Chisa Aoyama, Yih-Dar Nien, et al.

Cancer Res 2004;64:7610-7617.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/64/20/7610

Cited articles This article cites 48 articles, 12 of which you can access for free at: http://cancerres.aacrjournals.org/content/64/20/7610.full#ref-list-1

Citing articles This article has been cited by 11 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/64/20/7610.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/64/20/7610. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.