Regulation of Aromatase Expression in Human Ovarian Surface ...

Download PDF
47 downloads 39 Views 385KB Size Report
Comment
can be detected in human ovarian surface epithelial tumors. ... surface epithelial, sex-cord stromal, and germ cells. ... One type of sex-cord stromal cell in the.
0021-972X/00/$03.00/0 The Journal of Clinical Endocrinology & Metabolism Copyright © 2000 by The Endocrine Society

Vol. 85, No. 12 Printed in U.S.A.

Regulation of Aromatase Expression in Human Ovarian Surface Epithelial Cells* TOMOHARU OKUBO, SAMUEL C. MOK,

AND

SHIUAN CHEN

Division of Immunology (T.O., S.C.), Beckman Research Institute of the City of Hope, Duarte, California 91010; and Laboratory of Gynecology Oncology, Department of Obstetrics, Gynecology and Reproductive Biology (S.C.M.), Brigham and Women’s Hospital, Harvard Medical School. Boston, Massachusetts 02115 ABSTRACT Ovarian cancer originates mainly from surface epithelial cells, which are potential targets of estrogen action. Using immunohistochemistry and RT-PCR analysis, aromatase (estrogen synthetase) can be detected in human ovarian surface epithelial tumors. In this study, we functionally characterized the aromatase expressed in a primary cell culture, normal human ovarian surface epithelial (HOSE) 17. The apparent Km and Vmax values were determined to be 5.8 ⫾ 0.5 nM, and 0.3 ⫾ 0.0 pmol/mg䡠h, respectively. The aromatase activity in HOSE 17 cells can be induced effectively by phorbol esters and forskolin, suggesting that estrogen biosynthesis in HOSE 17 cells is mainly regulated through protein kinase C- and protein kinase A-mediated mechanisms. Exon I-specific RT-PCR revealed that phor-

O

VARIAN CANCER, THE fifth most common cancer in women in United States (1), remains a difficult cancer to treat in that the mortality rate has not improved in the past few decades (2). The ovary consists of three types of cells: surface epithelial, sex-cord stromal, and germ cells. The vast majority of ovarian cancers originate from the surface epithelial cells. Several epidemiological studies suggest that local production of estrogen is associated with the development of ovarian cancer. It has been found that women taking oral contraceptives have a lower risk of developing ovarian cancer (3), possibly as a result of a negative feedback mechanism that suppresses the local production of estrogen through gonadotropins. On the other hand, women who take antiestrogen to stimulate ovulation (usually clomiphene) have an increased risk factor for ovarian cancer (4), possibly as a result of a positive feedback mechanism that increases the local production of estrogen. It is also well known that women who have a history of breast cancer have a much higher risk of developing ovarian cancer and women who have a history of ovarian cancer have a higher risk of developing breast cancer (5). Estrogen is known to play an important role in promoting breast tumor growth. These findings strongly indicate that localized estrogen formation may be important in the development of ovarian cancer. Recently, Lau et al. (6) have reported the coexpression of estrogen receptor ␣ (ER␣) and ER␤ messenger RNA (mRNA)

Received May 30, 2000. Revision received July 31, 2000. Rerevision received August 24, 2000. Accepted September 2, 2000. Address correspondence and requests for reprints to: Dr. Shiuan Chen, Division of Immunology, Beckman Research Institute of the City of Hope, Duarte, California 91010. E-mail: [email protected]. * Supported by the NIH Grants CA44735, ES08258, and CA65767.

bol esters predominantly up-regulated promoter II. Whereas forskolin treatment increased exon I.3A-containing messenger RNA, the aromatase activity remained low in the cells treated with this agent. In vitro transcription/translation analysis using plasmids containing T7 promoter and the human snail gene (SnaH) as a reporter capped with different untranslated exon Is revealed that exon PII-containing transcripts were translated more effectively than exon I.3-containing transcripts. These findings explain why aromatase activity is higher in cells with the PII-containing transcripts than is cells with the I.3-containing transcripts. Our results indicate that aromatase is functionally expressed in human ovarian surface epithelial cells and its expression is regulated at both the transcriptional and translational levels. (J Clin Endocrinol Metab 85: 4889 – 4899, 2000)

in several normal human ovarian surface epithelial (HOSE) cell lines (including HOSE 17) and the disruption of ER␣ mRNA expression in most ovarian cancer cells. However, a definitive correlation of estrogen formation and the development of ovarian cancer has not yet been established. Aromatase (cytochrome P450 aromatase or P450arom) is the enzyme that synthesizes estrogen. As part of maintenance of normal physiological function, this enzyme is expressed in several normal human tissues, such as ovary, placenta, testis, skin, adipose, bone, and brain (7–10). Aromatase is expressed pathologically in some tumor tissues, such as the great majority of human breast (11), some of ovarian (12) and endometrial cancers (13). One type of sex-cord stromal cell in the ovary, granulosa cell, is the major source of estrogen biosynthesis in premenopausal women. Whereas aromatase expression in ovarian surface epithelial tumors has been demonstrated by several methods (12, 14, 15), the catalytic properties and the control of the expression of aromatase in ovarian surface epithelial cells are not known. We recently detected and measured aromatase activity in a primary HOSE cell line that is derived from the ovary of a 28-yr-old woman. As the first step toward developing an understanding of the role of aromatase in HOSE cells, we functionally characterized the enzyme expressed in this primary cell culture. We have found that phorbol 12-myristate 13-acetate (PMA), a phorbol ester, is an important inducing agent for the expression of aromatase in this cell. Furthermore, by RT-PCR, we have evaluated the exon I/promoter usage associated with aromatase expression in HOSE cells. We report the detection of the differences in the translatability of transcripts with different untranslated exons I us-

4889

4890

OKUBO ET AL.

ing an in vitro transcription/translation system. These results are discussed. Materials and Methods Materials PMA, forskolin, and dexamethasone (DEX) were obtained from Sigma (St. Louis, MO). Bisindolylmaleimide I hydrochloride was purchased from Calbiochem (La Jolla, CA). Dimethylsulfoxide (DMSO) was from Mallinckrodt, Inc. (Paris, KY). [1␤-3H(N)]-androst-4-ene-3,17dione was purchased from NEN Life Science Products (Boston, MA). [35S] methionine was from Amersham Life Science, Inc. (Arlington Heights, IL). The mammalian cell-expression vector pSG5 was obtained from Stratagene (La Jolla, CA). Oligonucleotide primers were synthesized in the DNA/RNA synthesis laboratory at the City of Hope.

Cell culture The HOSE 17 cell line was derived from surface scraping of the normal ovary of a 28-yr-old woman. The cells can be maintained in vitro for approximately four to five passages and have been characterized (16, 17). HOSE 17 was cultured in growth medium containing a 1:1 mixture of MCDB 105 and M199 (Sigma) supplemented with 10% (vol/vol) heat-inactivated FCS (Gemini, Calabasas, CA) and 1⫻ Fungi-Bact Solution (Irvine Scientific, Santa Ana, CA), including 100 U penicillin G/mL, 100 ␮g streptomycin sulfate/mL, and 0.25 ␮g Amphotericin/mL.

Aromatase assay Aromatase activity was determined by a modification of the tritiated water method of Thompson and Siiteri (18), using [1␤-3H]androst-4ene-3,17-dione as substrate (specific activity, 28.5 Ci/mmol). In the “Incell” aromatase assay, the cells were washed twice with PBS, and 1 mL serum-free medium containing 100 nm [1␤-3H] androstenedione as well as 500 nm progesterone (used to suppress the endogenous 5␣ reductase that also consumes the androgen substrate) was added to each well. After a 5-h incubation at 37 C, the reaction mixture was removed and extracted with an equal volume of chloroform. The mixture was then centrifuged at 100 ⫻ g for 10 min, and the aqueous upper layer was mixed with charcoal-dextran to remove any trace amount of unreacted substrate. In the second extraction, the sample was vortexed and subsequently centrifuged at 15,000 ⫻ g for 5 min. Supernatant aliquot was counted in a liquid scintillation counter. Aromatase activity was calculated as pmol/mg protein/h. Protein concentrations were determined by the Bradford method (19). In dose-response experiments, cells were plated on Corning 6-well plates (Corning, Inc., Corning, NY) in the medium. When ⬃60 –75% confluent, they were refed with the medium containing 0.25% (vol/vol) DMSO vehicle (control), PMA at 0.01–50 nm, forskolin at 0.5–100 ␮m, or DEX at 0.1–200 nm, and then incubated for 24 h. At the end of the preincubation, the cells were assayed for aromatase activity.

Reaction intermediate analysis The assay was performed using [1␤-3H] androstenedione (specific activity, 20.0 Ci/mmol) as a substrate. The reaction media were extracted with an equal volume of chloroform. A 500-␮L aliquot of the chloroform phase containing steroid intermediates was withdrawn, and the solvent chloroform was removed by centrifugation under vacuum. The residue was dissolved in 100 ␮L acetonitrile and a 50-␮L aliquot mixed with 10 ␮L internal standards (the concentration of each internal control was 200 ␮m). The reaction intermediates were separated by reverse phase highperformance liquid chromatography on a C18 column (218TP54; VYDAC, Hesperia, CA), using a solvent system of acetonitrile:water (25:75, vol/vol) at a flow rate of 1 mL/min. Two-milliliter fractions were collected manually, and 500-␮L aliquots were counted in 3 mL ScintiSafe 30% (Fisher Scientific, Pittsburgh, PA). The retention times of 19hydroxy-4-androstene-3,17-dione(19 ol A), 4-androsten-19-al-3,17dione(19 al A), 19-nor-4-androstene-3,17-dione(19 nor A), and androstenedione were detected based on the internal standards’ absorbance at 214 nm and eluated at 9.1 min, 20.8 min, 44.0 min, and 66.8 min,

JCE & M • 2000 Vol. 85 • No. 12

respectively. The radioactivity associated with each peak was used to calculate the amount of each steroid. The level of the product estrone was estimated from the amount of tritiated water formed.

RNA isolation, RT-PCR, and Southern blot analysis For total cellular RNA isolation, HOSE 17 cells were incubated for 24 h in growth medium containing DMSO (0.25%) or each compound alone or in combination. The cells were then harvested from a 175-cm2 tissue culture flask with a scraper. RNA was isolated from the cultured cells using a procedure described in unit 7.12 of Molecular Cloning: A Laboratory Manual, Second Edition (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY). After extraction, RNAs were quantified by measuring their absorbance at 260 and 280 nm. Exon I primer-specific RT-PCR was shown to be a useful tool for examining the alternative usage of aromatase gene exon Is/promoters in aromatase-expressing tissues (e.g. Refs. 20 and 21). We synthesized seven oligonucleotides including 1a, 1c, 1b, 15, N69, 1d, and 2a with the sequences indicated in the studies by Harada (20), Toda et al. (22), and Shozu et al. (23). These primers have sequences derived from exons I.1, I.3, I.4, I.5, I.6, PII, and II, respectively. A reverse primer (2d) with a sequence derived from exon II and situated downstream from 2d was also synthesized and used. For RT-PCR, 2 ␮g total RNA was reverse-transcribed by Superscript II (Life Technologies, Inc., Gaithersburg, MD) using random primer (Life Technologies, Inc.) in a concentration of 50 ng/␮L. The 2.5 ␮L complementary DNA (cDNA) was subjected to PCR amplification in a 25-␮L reaction containing 20 mm TrisHCl (pH 8.3), 50 mm KCl, 15 mm MgCl2, 0.2 mm dNTPs, 12.5 pmol of each primer, and 5 U of AmpliTaq (PerkinElmer Corp., Norwalk, CT). PCR was performed for 25 cycles for semiquantitative analysis using the following temperature profile: 51 C, 1 min (primer annealing); 72 C, 2 min (primer extension); and 94 C, 1 min (denaturation). An additional extension cycle was performed for 5 min at 72 C before cooling the reaction mixture to 4 C. Because all aromatase mRNA contain exon II regardless of which exon I is present, PCR for each exon I was performed with a unique exon I primer and exon II reverse primer, 2d. As a control, we performed PCR using primers 2a and 2d to amplify the exon II region. Furthermore, we performed PCR using a set of human ␤-actin-specific primers, ␤-actin1 (sense 5⬘AGGAGCACCCCGTGCTGCTGA-3⬘) and ␤-actin2 (antisense 5⬘CTAGAAGCATTTGCGGTGGAC-3⬘), to amplify the human ␤-actin gene, which served as an internal control to normalize aromatase mRNA expression under each condition. The amplified products were subjected to electrophoresis on 1.5% agarose gel and blotted on Zetaprobe membranes (Bio-Rad Laboratories, Inc. Hercules, CA). The membranes were hybridized with exon IIspecific probe that corresponds to the middle of exon II (5⬘-ATGGTTTTGGAAATGCTGAA-3⬘). The oligonucleotide was labeled with [␥-32P]ATP by T4 kinase and purified with a STE select-D G-25 spin column (5 Prime-3 Prime, Inc., Boulder, CO). The conditions of hybridization were according to the Bio-Rad Laboratories, Inc. instruction manual. For quantification, the membranes were exposed to a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA). Precaution was taken to make sure that our RNA preparation did not contain genomic DNA. The RNA preparations were treated with RNasefree DNase. In addition, a control, such as PCR analysis with RNA without treating with RT, was performed to assure ourselves that the PCR products were derived from aromatase mRNA. PCR was performed at nonsaturating conditions for 25 cycles. This allowed us to compare the relative usage of different exons I in RNA from the same sample. It has been shown that with 25 cycles, product accumulation was exponential (21). In addition, the quantity of PCR products generated with 25 cycles of PCR increases in an aromatase mRNA concentrationdependent manner (21) and reconfirmed in this study (Fig. 1).

In vitro transcription/translation A 795-bp SnaH cDNA (GenBank accession no. AF125377) was generated by PCR and subcloned into pSG5 at the BamHI site at the 5⬘ end and BglII site at the 3⬘ end in the sense direction with T7 promoter as a reporter gene. Untranslated regions (UTRs) of pII, I.3A, I.3B, I.6H, and I.6L (117, 344, 243, 965, and 167 bp in length, respectively), followed by the 38-bp UTR sequence just upstream of aromatase gene start site (i.e.

AROMATASE EXPRESSION IN HOSE CELLS

FIG. 1. Accumulation of RT-PCR product in an aromatase mRNA concentration- dependent manner. HOSE 17 cell RNA, at different amounts, was subjected to RT-PCR (25 cycles) using primers 2a and 2d. PCR products were separated on agarose gel and transferred to Zetaprobe membranes. The PCR products were hybridized with a 32 P-labeled probe, which contains a sequence derived from exon II and analyzed with a Molecular Dynamics, Inc. M425S PhosphorImager. UTR in the exon II of the human aromatase gene), were generated by PCR and were inserted just upstream of SnaH cDNA in the vector at the BamHI site (see Fig. 8). The correct orientations and sequences of those fragments in the vector were confirmed by both restriction digestion and direct DNA sequencing. Human snail protein was synthesized in vitro using the TNT-coupled reticulocyte lysate system (Promega Corp., Madison, WI) with T7-RNA polymerase, according to the manufacturer’s instructions, using an optimal plasmid concentration of 1 ␮m and an incubation time of 40 min at 30 C. These conditions were selected so that the yield of the translation product, as calculated by [35S] methionine incorporation, would be on the linear part of the translation curve (data not shown). The expressed protein was separated on 15% SDS-PAGE. After the gels were dried, the expressed protein was visualized by autoradiography on Kodak BioMax film (Eastman Kodak Co., Rochester, NY) and quantified using a Phosphorimager, as described above.

DNA sequencing PCR products were subcloned into pSG5 or pCR2.1 vector (TA-cloning kit, Invitrogen, Carlsbad, CA) and sequenced using the dideoxy sequencing method (sequencing kit, version 2.0; Amersham Pharmacia Biotech, Cleveland, OH), according to the manufacturer’s instruction.

Statistical analysis The results were analyzed by Student’s t test. Results were considered significant when the P value was less than 0.05.

Results Kinetic analysis of aromatase in HOSE 17 cells

Aromatase activity in HOSE 17 cells was measured by an “In-cell” 3H-water release method. The Km and Vmax values were determined to be 5.8 ⫾ 0.5 nm and 0.3 ⫾ 0.0 pmol/mg䡠h, respectively (Fig. 2). The Km value was approximately one sixth of that of Chinese hamster ovary (CHO) cell expressing human placenta aromatase (24) (Table 1). Aromatase activity in these cells was further confirmed by reaction product analysis (see below).

4891

Because the kinetic parameters of aromatase activity in HOSE 17 cells are significantly different from those in human placenta (18, 24), it was thought that the aromatase enzyme in HOSE 17 cells might be different from that in placenta. Aromatase isoforms have been identified in goldfish (25, 26) and pig (27, 28). To investigate whether another aromatase form exists in HOSE 17 cells, we sequenced two parts of aromatase cDNA in HOSE 17 (56% of total coding sequence). Total RNA was extracted, and RT-PCR was performed using two sets of primers for the human placenta aromatase cDNA, exon II-Iv: II top (sense 5⬘-GCGGAACACAAGATGGTTTTGG-3⬘), V top (antisense 5⬘-GTCGACAGGCCGGGGCCTGACAGAG-3⬘), and for aromatase exon VI-VIII: V end (sense 5⬘-TAGATGAGGATCCCTTTGGACG-3⬘), IX top (antisense 5⬘-GGTACCATCTTTATGTCTCTCTCAC-3⬘). These primers were derived from regions that are conserved, but with subtle differences, among human and porcine aromatase isoforms. The two PCR products were subcloned into the pCR2.1 vector and sequenced as described in Materials and Methods. We found that the sequences of the RT-PCR products match completely with those of human placenta aromatase, indicating that HOST 17 cells express an identical aromatase protein as that in placenta. Only one aromatase protein in human has been identified so far. Although it is unlikely, we cannot totally eliminate the possibility that an aromatase isoform that is significantly different from placenta aromatase is present but could not be amplified with the primers used here. Reaction intermediate/product analysis has revealed that significantly more 19-al androstenedione (19-al A), by a comparison with the level of estrone, is produced by aromatase in HOSE 17 cells than human placenta aromatase (Table 2). Estrone (114.7 pmol) and 0.20 pmol 19-al A (less than 1/500 amount of the formed estrone) were generated by the CHO cell-expressing human placenta aromatase when the assay was performed with a 1-h incubation in the presence of 50 nm androstenedione, similar to that measured with human placental microsomes (20). On the other hand, 1.36 pmol estrone and 0.74 pmol 19-al A (approximately, a half of estrone formation) were generated by HOSE 17 cell-expressed aromatase when the assay was performed with a 5-h incubation in the presence of 200 nm androstenedione. The production of excess 19-al A or noneffective conversion of androgen to estrogen in HOSE 17 cells may be due to a disproportionate action of aromatase and cytochrome NADPH-P450 reductase. It has been previously shown that metabolic ratio is a sensitive indicator of aromatase-cytochrome P450 reductase interactions in the microsomal environment (29). The results of this reaction intermediate analysis further confirm that aromatase is expressed in HOSE 17 cells. Aromatase activity in HOSE 17 cells treated with PMA, forskolin, and DEX

A complex mechanism is involved in the control of human aromatase expression. Seven untranslated exon Is in the human aromatase gene have been reported. It is thought that aromatase expression in different tissues is driven by the promoters situated upstream from these exon Is, providing tissue-specific control of aromatase expression. Aromatase

4892

OKUBO ET AL.

JCE & M • 2000 Vol. 85 • No. 12

FIG. 2. Kinetic analysis of aromatase activity in HOSE 17 cells using the “Incell method.” The assay was performed in triplicate, and the results are shown with the SE.

TABLE 1 Kinetic properties of aromatase in HOSE 17 cells and CHO cell-expressing human placenta aromatase

HOSE 17 CHO cell-expressing human placenta aromatase

Km (nM)

Vmax(pmol/mg䡠h)

5.8 ⫾ 0.5 31 ⫾ 6

0.3 ⫾ 0.0 137 ⫾ 27

CHO, Chinese hamster ovary.

expression in human tissue is regulated by several hormones and chemicals, such as cAMP, glucocorticoids, and phorbol esters. Different promoters respond differently to these agents. To evaluate the effects of these agents on aromatase activity in HOSE 17 cells, the cells were incubated for 24 h in media containing 10% serum and six different concentrations of PMA (a phorbol ester), forskolin (an inducing agent for cAMP synthesis), or DEX (a glucocorticoid). Dose-response studies of the induction of aromatase by these agents were performed (Fig. 3). PMA stimulated aromatase activity maximally at a concentration of 4 nm, and the activity decreased at higher concentrations. Shozu et al. (30) also found a biphasic response of PMA on aromatase expressed in THP-1 cells and suggested that the decrease in aromatase activity might be attributed to a desensitization phenomenon. Phorbol ester inhibits protein kinase C activity at the higher concentration (31). The maximal activity with 4 nm PMA was ⬃130-fold higher than that detected in the untreated cells. Both forskolin and DEX also stimulated aromatase activity. However, the

inductive effects were significantly weaker than that of PMA. These results indicate that, among the three agents, PMA is the strongest inducing agent of aromatase activity or expression in HOSE 17 cells. Combined effects of PMA, forskolin, and DEX on aromatase activity in HOSE 17 cells

As can be seen in Fig. 4, forskolin significantly facilitated the stimulatory action of PMA on aromatase activity. The inductive effect of PMA plus forskolin on aromatase activity in HOSE 17 cells was synergistic (i.e. the activity induced by PMA plus forskolin was significantly higher than the sum of individual activities induced by PMA and forskolin). In contrast, DEX did not facilitate the stimulatory action of PMA on aromatase activity. The effect of PMA plus DEX (as well as forskolin plus DEX) seems to be additive. These findings with HOSE 17 are different from the results seen with THP-1 cells, in that forskolin inhibits the stimulatory action of PMA and DEX facilitates the action (30). Interestingly, the activity in response to PMA plus forskolin plus DEX was lower than that in response to PMA plus forskolin and almost the same as that in response to PMA plus DEX. This result indicates that DEX suppresses the stimulatory action of PMA plus forskolin. To further evaluate the stimulatory action of forskolin on the increase of aromatase activity by PMA, the cells were incubated in media containing six different concentrations of forskolin and 1.3 nm PMA for 24 h. As can be seen in Fig. 5, the aromatase activity increased in a dose-dependent man-

AROMATASE EXPRESSION IN HOSE CELLS

ner with forskolin in the range of 0 –33 ␮m. At 1.3 nm PMA and 13 ␮m forskolin, the activity was significantly higher than an additive action of individual treatments of 1.3 nm PMA and 13 ␮m forskolin. However, the aromatase activity was not affected at a low concentration of forskolin (i.e. 0.7 ␮m).

TABLE 2. Reaction intermediate analysis of aromatase in HOSE 17 cells and CHO cell-expressing human placenta aromatase

19 ol A 19 al A Estrone

HOSE 17 200 nM[␤-3H] androstenedione 5-h incubation

CHO cell-expressing human placenta aromatase 50 nM [1␤-3H] androstenedione 1-h incubation

0.097 pmol 0.74 pmol 1.36 pmol

0.0506 pmol 0.202 pmol 114.7 pmol

FIG. 3. Dose-response studies of aromatase activity in HOSE 17 cells in response to PMA, forskolin, or DEX. The data are shown as mean ⫾ SE of folds of the untreated control. The assays were performed in triplicate. *, P ⬍ 0.05 compared with each control.

4893

Combined effects of PMA, forskolin, and DEX on aromatase mRNA level in HOSE 17 cells

To understand the mechanism of the stimulatory action of PMA plus forskolin on aromatase activity, the cells were incubated under the same conditions as described previously, and then the total cytoplasmic RNA was extracted. We have examined the level of aromatase mRNA by semiquantitative RT-PCR with the procedure described in Materials and Methods, using a set of primers (2a and 2d) that are derived from exon II of the human aromatase gene. We also amplified ␤-actin mRNA as an internal control to normalize the level of aromatase gene expression. It has been found that, with few exceptions, the relative aromatase mRNA levels (Fig. 6A) correlate with the relative activity levels (Fig. 4). We have performed three independent sets of experiments in which HOSE 17 cells were treated with different agents, and we have done three independent RT-PCR Southern experi-

4894

OKUBO ET AL.

FIG. 4. Combination effects of PMA, forskolin, and DEX on the aromatase activity in HOSE 17 cells. Each column represents the mean ⫾ SE of values obtained from triplicate assays. *, P ⬍ 0.05.

ments to evaluate aromatase mRNA levels in each treatment. The results from a representative set of experiments are shown in Fig. 6A. We found that the mRNA level in response to forskolin was very high. This pattern was not consistent with that in the activity measurement. The aromatase activity in forskolin-treated cells was enhanced only moderately (see Fig. 4). Although the increase fold of the aromatase activity and that of aromatase mRNA for PMA-treated cells were comparable, there was a significant difference between the increase fold of the aromatase activity and mRNA for forskolin-treated cells (P ⬍ 0.05). We have also performed exon I-specific RT-PCR to examine exon I/promoter usage under each treatment condition (Fig. 6, B and C). Promoter II-driven transcripts were dominant in cells treated with PMA alone or PMA plus forskolin. Promoter II-driven transcripts were always dominant when PMA was included in the treatment. On the other hand, in response to forskolin, promoter I.3-driven transcripts (I.3A mainly) were dominant, followed by those driven by promoters II (i.e. pII) and I.6 (i.e. I.6H1). In the presence of DEX, promoter I.4-driven transcripts were stimulated and also, to a small extent, promoter I.1 was used. It should be noted that the major portion of the RNA messages in response to forskolin were unspliced forms, I.3A and I.6 H1, that include the upstream region from exon II (molecular nature of these exon Is see Fig. 8). By comparing results shown in Figs. 4 and 6B, it is thought that relative aromatase activity in HOSE 17 cells correlates with the relative level of promoter II-driven transcript better than total mRNA level. It should be noted that theoretically RT-PCR using primers 1d and 2d can amplify not only promoter PII-driven transcripts but also unspliced promoter I.3- and I.6-driven transcripts (i.e. I.3A and I.6H1, respectively). It is also possible that RT-PCR using primers 1c and 2d can amplify promoter I.3-driven as well as promoter I.6-driven transcripts. There-

JCE & M • 2000 Vol. 85 • No. 12

FIG. 5. Synergistic action of forskolin on PMA-stimulated aromatase activity in HOSE 17 cells. The cells were incubated with various dose of forskolin in the presence of 1.3 nM PMA. Each bar represents the mean ⫾ SE obtained from triplicate assays. *, P ⬍ 0.05.

fore, we evaluated the usage of aromatase gene promoter II by subtracting the intensity of the I.3A signal from that of PII in Southern blots. To evaluate the usage of promoter I.3, we subtracted the intensity of the I.6 H1 signal from that of I.3A. Effects of the 5⬘ UTRs on the translation of aromatase mRNA

The results that forskolin-treated cells have a high level of both I.3A- and I.6H1-containing message, but a low level of aromatase activity (shown in Figs. 4 and 6B) suggest that the translational efficiency of aromatase transcript could be affected by the alternatively spliced 5⬘ UTRs. To test this hypothesis, pSG5 expression constructs containing each alternatively spliced UTR sequence inserted immediately upstream from the sequence encoding human snail protein were prepared for use in an in vitro translation assay (Fig. 7). The reason to choose this protein instead of aromatase as the reporter protein is that it is smaller than aromatase and expressed as a stable and soluble form,, whereas aromatase is a membrane-bound protein. The coupled reticulocyte lysate system allowed transcription using a T7 promoter and then translation of human snail protein from the AUG start site. Considering the fact that the promoter regions in all pSG5 constructs were not modified, the transcriptional efficiency of the cDNA constructs should be theoretically equal when in vitro transcription/translation reactions are started with equimolar amounts of the cDNA constructs. This has been confirmed by Chopra et al. (32). We used the construct without any 5⬘ UTR as a positive control and the construct without snail cDNA as a negative control. The translation products from all of six cDNA constructs were separated by SDS PAGE. The products were quantified with a PhosphorImager, and the results expressed as a percentage of the positive control (Fig. 7). The level of the product obtained from translation of the

AROMATASE EXPRESSION IN HOSE CELLS

4895

FIG. 6. A, RT-PCR analysis of exon II of aromatase gene in HOSE 17 cells under different treatments using semiquantitative RT-PCR. The cells were incubated in media containing the stimulants (as indicated in Fig. 6B) for 24 h, and then total RNA was isolated. A 2-␮g aliquot of total RNA was subjected to RT-PCR-Southern analysis, as described in Materials and Methods. The ␤-actin transcript was amplified by the same method as an internal control. B, Effects of PMA, forskolin, and DEX on exon I/promoter usage of aromatase gene in HOSE 17 cells. The cells were incubated in media containing the stimulants for 24 h, and then total RNA was isolated. A 2-␮g aliquot of total RNA was subjected to exon I-specific RTPCR-Southern analysis, as described in Materials and Methods. Each bar represents folds of the untreated control, and the relative exon I/promoter usage in each treatment is also shown. C, Autoradiograms of RT-PCR products generated with exon I-specific RT-PCRSouthern analysis. HOSE 17 cells were treated with 1.3 nM PMA (left), 17 ␮M forskolin (middle), and 1.3 nM PMA plus 17 ␮M forskolin (right). Lane 1, Exon I.1; lane 2, exon I.3 (I.3A and I.3B); lane 3, exon I.4; lane 4, exon I.5; lane 5, exon I.6 (i.e. I.6L); lane 6, pII; lane 7, exon II. The sizes of the detected PCR products for exons I.3A, I.3B, I.6L, pII, and II are 333bp, 232bp, 1037bp, 234bp, and 169bp, respectively.

construct with pII was the highest of all, followed by the positive control, then with I.6L, I.3B, I.3A, and I.6H1. Except for the positive control, this is in the same order as that of the lengths of each 5⬘ UTR, starting with the shortest. The translational efficiency of the constructs containing I.3A and I.6H1 were ⬃6.7% and 0.8% of the construct containing pII. To rule out the possibility that splicing may happen in this system, we performed RT-PCR on the in vitro transcription/translation reaction mixtures with a set of exon I-specific primers after treatment with DNase I. The spliced transcripts, I.6L or I.3B, were not detected in the reaction. This result supports the hypothesis that promoter II-driven transcript is translated more efficiently than the unspliced forms of transcript driven by promoter I.3 and I.6, including I.3A and I.6H1. Discussion

Ovarian surface epithelial tumors account for 60% of all ovarian neoplasms and 80 –90% of primary ovarian malig-

nancies (33, 34). A number of epidemiological studies suggest that endocrine factors play an important role in the development of ovarian surface epithelial malignancies (3, 4). Although high levels of gonadotropins in women in early postmenopausal have been postulated to play a role, there has been no conclusive evidence regarding a correlation between local estrogen level and the development of ovarian surface epithelial malignancies. In addition, although studies on aromatase expression in ovarian surface epithelial tumors have been published (12, 14, 15), there have been no report as to whether aromatase is expressed in normal HOSE cells. In this study, by directly measuring the aromatase activity, we have demonstrated that normal HOSE cells are capable of synthesizing estrogen. The levels of the activity are similar to those measured by Noguchi et al. (14) in ovarian epithelial tumors. Thompson et al. (35) first reported the aromatizaton of testosterone by epithelial tumor cells cultured from patients with ovarian carcinoma. It was concluded that at least

4896

OKUBO ET AL.

FIG. 7. Effects of the 5⬘ untranslated exon Is of the aromatase gene on the expression of the human Snail protein using an in vitro transcription/translation system. Equimolar amounts of the constructs were transcribed/translated in a reticulocyte lysate system. An autoradiogram of [35S] methionine-labeled protein, fractionated by PAGE, is shown (top). The molecular size of the protein product is 30 kDa and was confirmed by coelectrophoresis with molecular size markers. The results are also shown as a column graph in which the control (pSG5 SnaH) is taken as 100%.

a portion of ovarian adenocarcinoma possess sufficient aromatase activity to convert ovarian stromal androgen to estrogen. Two other groups have demonstrated the presence of aromatase in ovarian surface epithelial neoplasms by aromatase assay, immunocytochemistry and RT-PCR analysis. Kitawaki et al. (12) detected aromatase activity and reported that aromatase was immunohistochemically localized in the cytoplasm of neoplastic cells in both benign and malignant ovarian epithelial tumors. However, Kaga et al. (15) demonstrated aromatase immunoreactivity in the stromal cells adjacent to the carcinoma and also at the site of flank invasion in ovarian cancer. The reasons for the difference between the two immunohistochemical studies are not currently understood. Differences in immunostaining of aromatase in breast tumors were also observed and thought to be due to the fact that different antibodies were used. Immunocytochemical analysis from our laboratory first identified the presence of aromatase in breast cancer epithelial and stromal cells (36), whereas others had reported the presence of aromatase only in the stromal tissue (37, 38). Our findings have been recently confirmed by independent in situ hybridization studies and cell proliferation assays showing that aromatase is expressed in breast cancer epithelial cells (39). The HOSE 17 cell line is a primary cell culture derived from the normal ovary of a 28-yr-old woman, and, therefore, it may more closely resemble normal cells than the established

JCE & M • 2000 Vol. 85 • No. 12

ovarian cancer cell lines. To determine the regulatory mechanism of aromatase activity in HOSE 17 cells, we have found that phorbol esters such as PMA are important inducing agents, and forskolin can facilitate the inductive action of phorbol esters. Although the basal aromatase activity in HOSE 17 cells is similar to that measured in ovarian epithelial tumors, the activity can increase greatly in the presence of PMA and forskolin. Phorbol esters such as PMA are potent tumor promoters that exert biological effects by the activation of protein kinase C (40). PMA has been shown to be a potent stimulator of aromatase activity in placenta (41) as well as in MCF-7 cells (42). Forskolin activates adenylate cyclase, resulting in the induction of the formation of cAMP. The activators of the protein kinase A pathway stimulates aromatase activity in granulosa cells and adipose stromal cells (43, 44), whereas they inhibit it in THP-1 cells, even in the presence of PMA (30). Aromatase activity was also stimulated synergistically by PMA plus forskolin in breast cancer cell lines, MCF-7 cells (42), but not in THP-1 cells. DEX is a synthetic glucocorticoid that stimulates aromatase expression in adipose stromal cells, but not in granulosa cells (43). Aromatase activity in HOSE 17 cells could be stimulated by DEX alone, and the synergistic action by PMA plus forskolin was inhibited by DEX. Therefore, aromatase activity in HOSE 17 cells is regulated differently from that reported in other tissues. It should be pointed out that the results generated with the HOSE 17 cell line may be specific for this cell line. We plan to extend our studies to other HOSE cell lines and will compare the results generated with other cell lines to the HOSE 17 cell line. We have recently performed RTPCR analysis on RNAs isolated from 10 HOSE cell lines and found that aromatase is expressed in all of these cell lines (our unpublished results). These cell lines contain mainly promoter II-, promoter I.3-, and promoter I.1-driven transcripts. The basal aromatase activity in these 10 cell lines is lower than that in the HOSE 17 cell line. The regulatory mechanism of the expression of aromatase in these cell lines has not yet been determined. Previous studies from our and other laboratories have revealed that, in human, different aromatase gene promoters respond to PMA, forskolin, and DEX differently. Promoters I.1, I.3 and II, and I.4 mainly respond to phorbol esters, forskolin, and DEX, respectively. We observed that switching of exon I/promoter usage occurred in HOSE 17 cells in response to these chemicals. In the nonstimulated conditions, the major utilization was an unspliced form of transcript (I.3A) (Fig. 6B). Under forskolin stimulation, promoter I.3driven unspliced transcripts (i.e. I.3A) were dominant, followed by those driven by promoter II and I.6. A cAMPresponse element upstream from promoter I.3 was recently identified in our laboratory (45). However, it remains unclear why the splicing at the downstream region of exon I.3 does not occur. In the presence of PMA, promoter II-driven transcripts were the major transcripts, not promoter I.1-driven transcripts, and were augmented by forskolin. These results suggest that a PMA-response element is present near promoter II. The molecular mechanism of the stimulatory action of forskolin on PMA induction is not yet determined. DEX induced promoter I.1- and I.4-driven transcripts to a small extent and inhibited promoter II- and I.3-driven transcripts

AROMATASE EXPRESSION IN HOSE CELLS

4897

FIG. 8. Diagram of the structures of the cDNA constructs containing the SnaH coding region capped with the 5⬘ UTRs of the human aromatase gene. These sequences were inserted into the pSG5 expression vector. The top lane represents the structure of the human aromatase gene. The length of each region is shown in bp number. The potential upstream ORFs associated with each construct are presented by dark gray boxes placed in three rows, indicating that they are originated from the three reading frames. There is no ORF that overlaps with the coding sequence of SnaH.

that were stimulated by PMA and forskolin, respectively. It should be noted that DEX inhibits cell growth in breast cancer cell lines (46) in contrast to the effects of both PMA and forskolin. In fact, DEX inhibited the cell proliferation in HOSE 17 cells (data not shown). In our study, we observed two PCR products with different sizes (i.e. I.3A/I.3B, I.6 H1/L) when we used primers derived from exon I.3 (i.e. 1c) or exon I.6 (i.e. N69) and a reverse primer from exon II (i.e. 2d). The presence of exon I.3A and I.3B has been already reported (21). The sequence of I.3A has been shown to be an unspliced variant of I.3B, and we confirmed both I.3A and I.6H1 by direct DNA sequencing of the PCR product that both of them are unspliced transcripts (results not shown). We have ruled out the possibility that the unspliced forms result from genomic DNA contamination because no PCR products were formed in the absence of RT. Furthermore, if these unspliced forms were due to genomic DNA contamination, we would expect that I.3Aand I.6H1-containing messages in each sample would be present at similar levels in each treatment. Although we detected only two different PCR products of promoter I.6driven transcripts (I.6H1/I.6l), Shozu et al. (23) detected five products of different sizes. They concluded that these alternatively spliced variants of promoter I.6-driven transcripts appeared to occur in a higher proportion of aromatase gene

transcripts in malignant cell lines than in primary culture cells. Considering the findings that forskolin treatment increased the level of both exon I.3A- and I.6H1-containing transcripts and moderately enhanced aromatase activity, we propose that RNA messages containing these exon Is (i.e. I.3A and I.6H1) are not translated effectively. Our hypothesis is supported by the results from in vitro transcrption/translation studies (see Fig. 7). Alternative usage of exon Is affects translational efficiency. Initiation of translation in eukaryotes can be influenced by some aspects of mRNA structure [such as the m7G cap (47)], the context surrounding the AUG codon, the position of AUG codon, leader length, and secondary structure (48). We have pointed out that the translation efficiency is in the same order as that of the lengths of each 5⬘ UTR, starting with the shortest exon I (i.e. PII), except the positive control. Furthermore, a scanning mechanism for initiation may explain our results partially. Our study used SnaH gene as a reporter that does not have the consensus sequence for initiation of translation surrounding the AUG codon. We have searched open reading frames (ORFs) in each aromatase 5⬘ UTR. There are 1, 3, 3, 4, and 13 ORFs in PII, I.6L, I.3B, I.3A, and I.6H1, respectively, and there is no ORF that overlaps with the coding sequence (Fig. 8). These ORFs could reduce the initiation efficiency of translation at

4898

OKUBO ET AL.

the proper start site (49). In addition, the rate of translation may be affected by the formation of a stable secondary structure in 5⬘ UTRs. We used the SEQWEB version 1.1 (Genetics Computer Groups, Inc., Madison, WI) program to predict free energy as the stability of the secondary structure in the 5⬘ UTR of each exon I-containing transcript. The analysis revealed that the secondary structure of PII containing transcript is most unstable (⌬G ⫽ ⫺25.9 kcal/mol), followed by I.6L (⌬G ⫽ ⫺40.1 kcal/mol), I.3B (⌬G ⫽ ⫺50.3 kcal/mol), I.3A (⌬G ⫽ ⫺77.5 kcal/mol), and I.6H1 (⌬G ⫽ ⫺227.1 kcal/ mol). This analysis also supports our findings on the relative translation efficiency of different exon I-containing transcripts. Whereas the results from the in vitro transcription/ translation studies cannot adequately explain the results as to why exon PII usage did not correlate well with the aromatase activity in samples treated with forskolin only or treated with PMA plus DEX (cf. Figs. 4 and 7), the results do support the conclusion that aromatase expression can also be regulated at the translational level. In summary, our results indicate that functional aromatase is expressed in normal HOSE cells and the aromatase activity is induced in a cooperative fashion by PMA and forskolin, suggesting transcriptional regulation through the PKC and PKA pathways. Furthermore, our in vitro transcription/ translation studies reveal that aromatase expression can also be modulated at the translational level. Although several epidemiological studies support the hypothesis that local production of estrogen is associated with the development of ovarian cancer, the exact mechanism is not yet known. We have observed that aromatase gene promoter/exon I switching occurs by comparing promoter/exon I usage in ovarian surface epithelial neoplasms and that in normal HOSE cell lines (our unpublished results), suggesting that aromatase expression in ovarian cancer is regulated differently from that in normal HOSE cells. Furthermore, Lau et al. (6) have reported that ER␣ and ER␤ mRNA are present in normal HOSE cells and the transcription of ER␣ is disrupted in most ovarian cancer cells. These results indicate that the regulation of the synthesis and the molecular action of estrogen in normal HOSE and cancer tissue are different. Further investigations are needed to evaluate the function of ERs and the affect of in situ estrogen formation on the proliferation and transformation of ovarian surface epithelial cells. Acknowledgments We gratefully acknowledge the generous support of Prof. Hideo Honjo (Department of Obstetrics/Gynecology, Kyoto Prefectural University of Medicine, Kyoto, Japan) for Tomoharu Okubo to study in Dr. Shiuan Chen’s laboratory, the kind technical support of Dr. Yeh-Chih Kao, and the help of Ms. Karen Feintuch in the preparation of this manuscript.

References 1. Landis SH, Murray T, Bolden S, Wingo PA. 1998 Cancer statistics, 1998. CA Cancer J Clin. 48:6 –29. 2. Herbst AL, Berek JS. 1993 Impact of contraception on gynecologic cancers. Am J Obstet Gynecol. 168:1980 –1985. 3. Rossing MA, Daling JR, Weiss NS, Moore DE, Self SG. 1994 Ovarian tumors in a cohort of infertile women. N Engl J Med. 331:771–776. 4. Greene MH, Clark JW, Blayney DW. 1984 The epidemiology of ovarian cancer. Semin Oncol. 11:209 –226. 5. Sutcliffe S, Pharoah PD, Easton DF, Ponder BA. 2000 Ovarian and breast

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

22. 23. 24. 25. 26. 27.

28. 29. 30. 31.

32. 33.

JCE & M • 2000 Vol. 85 • No. 12

cancer risks to women in families with two or more cases of ovarian cancer. Int J Cancer. 87:110 –117. Lau KM, Mok SC, Ho SM. 1999 Expressionof human estrogen receptor-a and -␤, progesterone receptor, and androgen receptor mRNA in normal and malignant ovarian epithelial cells. Proc Natl Acad Sci USA. 96:5722–5727. Means GD, Kilgore MW, Mahendroo MS, Mendelson CR, Simpson ER. 1991 Tissue-specific promoters regulate aromatase cytochrome P450 gene expression in human ovary and fetal tissues. Mol Endocrinol. 5:2005–2013. Tsai-Morris CH, Aquilana DR, Dufau ML. 1985 Cellular localization of rat testicular aromatase activity during development. Endocrinology. 116:38 – 46. Grodin JM, Siiteri PK, MacDonald PC. 1973 Source of estrogen production in postmenopausal women. J Clin Endocrinol Metab. 36:207–214. Naftolin F, Ryan KJ, Davies IJ, et al. 1975 The formation of estrogens by central neuroendocrine tissues. Recent Prog Horm Res. 31:295–319. Reed MJ, Owen AM, Lai LC, et al. 1989 In situ oestrone synthesis in normal breast tissue and breast tumor tissue: effect of treatment with 4-hydroxyandrostenedione. Int J Cancer. 44:233–237. Kitawaki J, Noguchi T, Yamamoto T, et al. 1996 Immunohistochemical localisation of aromatase and its correlation with progesterone receptors in ovarian epithelial tumors. Anticancer Res. 16:91–98. Yamaki J, Yamamoto T, Okada H. 1985 Aromatization of androstenedione by normal and neoplastic endometrium of the uterus. J Steroid Biochem. 22:63– 66. Noguchi T, Kitawaki J, Tamura T, et al. 1993 Relationship between aromatase activity and steroid receptor levels in ovarian tumors from postmenopausal women. J Steroid Biochem Mol Biol. 44:657– 660. Kaga K, Sasano H, Harada N, Ozaki M, Sato S, Yajima A. 1996 Aromatase in human common epithelial ovarian neoplasms. Am J Pathol. 149:45–51. Tsao S-W, Mok SC, Fey EG, et al. 1995 Characterization of human ovarian surface epithelial cells immortalized by human papiloma viral oncogenes (HPV-E6E7 ORFs). Exp Cell Res. 218:499 –507. Auersperg N, Edelson MI, Mok SC, Johnson SW, Hamilton TC. 1998 The biology of ovarian cancer. Semin Oncol. 25:281–304. Thompson Jr EA, Siiteri PK. 1974 Utilization of oxygen and reduced nicotinamide adenine dinucleotide phosphate by human placental microsomes during aromatization of androstenedione. J Biol Chem. 249:5364 –5372. Bradford MM. 1976 A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem. 72:248 –254. Harada N. 1993 A unique aromatase (P-450arom) mRNA formed by alternative use of tissue-specific exons I in human skin fibroblasts. Biochem Biophys Res Commun. 189:1001–1007. Zhou C, Zhou D, Esteban J, et al. 1996 Aromatase gene expression and its exon I usage in human breast tumors. Detection of aromatase messenger RNA by reverse transcription-polymerase chain reaction. J Steroid Biochem Mol Biol. 59:163–171. Toda K, Simpson ER, Mendelson CR, Shizuta Y, Kilgoren MW. 1994 Expression of the gene encoding aromatase cytochrome P450 (CYP19) in fetal tissues. Mol Edocrinol. 8:210 –217. Shozu M, Zhao Y, Bulun SE, Simpson ER. 1998 Multiple splicing events involved in regulation of human aromatase expression by a novel promoter, I.6. Endocrinology. 139:1610 –1617. Kao Y-C, Cam LL, Laughton CA, Zhou D, Chen S. 1996 Binding charasteristics of seven inhibitors of human aromatase: a site-directed mutagenesis study. Cancer Res. 56:3451–3460. Gelinas D, Pitoc GA, Callard GV. 1998 Isolation of a gold fish brain cytochrome P450 aromatase cDNA: mRNA expression during the seasonal cycle and after steroid treatment. Mol Cell Endocrinol. 138:81–93. Tchoudakova A, Callard GV. 1998 Identification of multiple CYP19 genes encoding different cytochrome P450 aromatase isozymes in brain and ovary. Endocrinology. 139:2179 –2189. Choi I, Simmen RCM, Simmen FA. 1996 Molecular cloning of cytochrome p450 aromatase comprementary deoxyribonucleic acid from periimplantation porcine and equine blastocysts identifies multiple novel 5⬘-untranlated exons expressed in embryos, endometrium, and placenta. Endocrinology. 137:1457–1467. Choi I, Collante WR, Simmen RCM, Simmen FA. 1997 A developmental switch in expression from blastocyst to endometrial/placental-type cytochrome :450 aromatase genes in the pig and horse. Biol Reprod 56:688 – 696. Grogan J, Shou M, Zhou D, Chen S, Korzekwa KR. 1993 Use of aromatase (CYP19) matabolite ratios to characterize electron transfer from NADPHcytochrome p450 reductase. Biochemistry. 32:12007–12012. Shozu M, Zhao Y, Simpson ER. 1997 Estrogen biosynthesis in THP-1 cells is regulated by promoter switching of the aromatase (CYP19) gene. Endocrinology. 138:5125–5135. Huges R, Timmermans P, Schrey MP. 1996 Regulation of arachidonic acid metabolism, aromatase activity and growth in human breast cancer cells by interleukin-1␤ and phorbol ester: dissociation of a mediatory role for prostaglandin E2 in the autocrine control of cell function. Int J Cancer. 67:684 – 689. Chopra R, Kendall G, Gale RE, Thomas NSB, Linch DC. 1996 Expression of two alternative spliced forms of the 5⬘ untranslated region of the GM-CSF receptor chain mRNA. Exp Hematol. 24:755–762. Koonings PP, Campbell K, Mishell Jr DR, Grimes DA. 1989 Relative fre-

AROMATASE EXPRESSION IN HOSE CELLS

34.

35. 36. 37. 38. 39.

40.

quency of primary ovarian neoplasm: a 10-year review. Obstet Gynecol. 74:921–926. Katsube Y, Berg JW, Silverberg SG. 1982 Epidemiologic pathology of ovarian tumors: a histopathologic review of primary ovarian neoplasms diagnosed in the Denver Standard Metropolitan Statistical Area, 1 July-31 December 1969 and 1 July-31 December 1979. Int J Gynecol Pathol. 1:3–16. Thompson MA, Adelson MD, Kaufman LM, Marshall LD, Coble DA. 1988 Aromatization of testosterone by epithelial tumor cells cultured from patients with ovarian carcinoma. Cancer Res. 48:6491– 6497. Esteban JM, Warsi Z, Haniu M, Hall PF, Shivery JE, Chen S. 1992 Detection of intratumoral aromatase in breast carcinomas, an immunohistochemical study with clinicopathogic correlation. J Am Pathol. 140:337–343. Sasano H, Nagura H, Harada N, Goukon Y, Kimura M. 1994 Immunolocalization of aromatase and other steroidogenic enzymes in human breast disorders. Hum Pathol. 25:530 –535. Santen RJ, Martel J, Hoagland M, et al. 1994 Stromal spindle cells contain aromatase in human breast tumors. J Clin Endocrinol Metab. 79:627– 632. Lu Q, Nakamura J, Savinov A, et al. 1996 Expression of aromatase protein and messenger ribonucleic acid in tumor epithelial cells and evidence of functional significance of locally produced estrogen in human breast cancers. Endocrinology. 137:3061–3077. Nishizuka Y. 1984 The role of protein kinase C in cell surface signal transduction and tumor promotion. Nature. 308:693– 698.

4899

41. Toda K, Terashima M, Kawamoto T, et al. 1990 Structural and functional characterization of human aromatase P-450 gene. Eur J Biochem. 193:559 –565. 42. Ryde CM, Nicholls JE, Dowsett M. 1992 Steroid and growth factor modulation of aromatase activity in MCF7 and T47D breast carcinoma cell lines. Cancer Res. 52:1411–1415. 43. Steinkampf MP, Mendelson CR, Simpson ER. 1987 Regulation by folliclestimulating hormone of the synthesis of aromatase cytochrome P-450 in human granulosa cells. Mol Endocrinol. 1:465– 471. 44. Mendelson CR, Cleland WH, Smith ME, Simpson ER. 1982 Regulation of aromatase activity of stromal cells derived from human adipose tissue. Endocrinology. 111:1077–1114. 45. Zhou D, Chen S. 1999 Identification and characterization of a cAMP-responsive element in the region upstream from promoter I.3 of the human aromatase. Arch Biochem Biophys. 371:179 –190. 46. Lippmen M, Bolan G, Huff K. 1976 The effects of glucocorticoids and progesterone on hormone responsive human breast cancer in long term tissue culture. Cancer Res. 36:4602– 4609. 47. Shatkin AJ. 1976 Capping of eucaryotic mRNAs. Cell. 9:645– 653. 48. Kozak M. 1991 Structural features in eukaryotic mRNAs that modulate the initiation of translation. J Biol Chem. 266:19867–19870. 49. Mueller PP, Hinnebusch AG. 1986 Multiple upstream AUG codons mediate translational control of GCN4. Cell. 45:201–207.