Gastric Parietal Cells - Semantic Scholar

0013-7227/02/$15.00/0 Printed in U.S.A.

Endocrinology 143(8):3162–3170 Copyright © 2002 by The Endocrine Society

Gastric Parietal Cells: Potent Endocrine Role in Secreting Estrogen as a Possible Regulator of GastroHepatic Axis TAKASHI UEYAMA, NOBUYUKI SHIRASAWA, MITSUTERU NUMAZAWA, KEIKO YAMADA, MOMOKO SHELANGOUSKI, TAKAO ITO, AND YOSHIHIRO TSURUO Department of Anatomy and Cell Biology (T.U., N.S., T.I., Y.T.), Wakayama Medical University, Wakayama 641-8509, Japan; and Tohoku Pharmaceutical University (M.N., K.Y., M.S.), Aobaku, Sendai 981-8558, Japan Estrogen, if it is produced in the gastrointestinal tract, may overflow into the systemic circulation in the case of increased portal-systemic shunting. This idea is in accord with a significant step-up in serum estradiol (E2) concentration in the portal vein of rats, compared with that in the artery. Gene expression of aromatase, estrogen synthetase, was demonstrated by RT-PCR in the gastric mucosa of male and female adult rats, equivalent to that in the ovary. Aromatase activity and production of E2 in the gastric mucosa were demonstrated by 3H2O assay and gas chromatography-mass spectrometry, and they were inhibited by aromatase inhibitor, 4-hydroxyandrostenedione. Conversion of 14C-androstenedione to 14C-E2 through 14C-testosterone in cultured gastric mu-

E

STROGEN HAS VARIOUS physiological functions, such as growth and differentiation, via not only endocrine but autocrine or paracrine fashion (1). Though the ovary is a major source of systemic estrogen in premenopausal nonpregnant females, estrogen is produced also in extraovarian tissues, including skin (2), testis (3), brain (4), adipose tissues (5), osteoblasts (6), vascular smooth muscle cells (7), and T cells in the spleen (8). These extraovarian tissues, however, probably do not produce a sufficient quantity of estrogen to affect the circulating level. Patients with liver cirrhosis often show an increase of serum estradiol (E2)/testosterone ratio, which is supposedly caused by an abnormal estrogen metabolism in the liver or elsewhere, and this is believed to be responsible for manifestations of spider angioma, loss of pubic hair, testicular atrophy, and gynecomastia (9). As is often the case with this condition, portal hypertension can cause an increase of portal-systemic shunting via anastomosis between the portal vein and the vena cava, whereby products in the gastrointestinal organs may overflow into the systemic circulation, without metabolism and degradation in the liver. In accord with this notion, previous studies showed that experimental portal-systemic shunting in adult male rats resulted in a marked increase of systemic E2 concentration (10, 11). We therefore formulated the following hypothesis: under normal physiological conditions, a large quantity of estrogen equivalent to ovary is Abbreviations: DAB, 3,3⬘-Diaminobenzidine tetrahydrochloride; E2, estradiol; ER, estrogen receptor; GC-MS, gas chromatography-mass spectrometry; NADPH, nicotinamide adenine dinucleotide phosphate; 35 S-dATP, 35S-deoxy-ATP; TLC, thin-layer chromatography.

cosa was also demonstrated. Parietal cells exhibited strong signals for aromatase mRNA and immunoreactive protein by in situ hybridization histochemistry and immunohistochemistry. Estrogen receptor ␣ mRNA and immunoreactive protein were demonstrated in hepatocytes by RT-PCR, in situ hybridization histochemistry, and immunohistochemistry. Total gastrectomy reduced portal venous E2 concentration, without changing systemic E2 concentration, together with downregulation of estrogen receptor ␣ mRNA level in the liver. These findings indicate that gastric parietal cells play a potent endocrine role in secreting estrogen that may function as a regulator of the gastro-hepatic axis. (Endocrinology 143: 3162–3170, 2002)

secreted from an unidentified region of the gastrointestinal organs into the portal vein and most of it is then trapped in the liver, but in the case of increased portal-systemic shunting, it overflows into the systemic circulation, resulting in the elevation of systemic E2 concentration, to cause estrogenexcess signs associated with liver cirrhosis. Until now, however, little attention has been given to the possibility that estrogen may be produced in the gastrointestinal tract; and so, it seems of great value to know at which site estrogen is produced. The hypothesis prompted us to investigate whether a substantial amount of estrogen is secreted into the portal vein and then to identify the exact organ and cellular localization of aromatase, estrogen synthetase (12), in the gastrointestinal tract of rats. Materials and Methods Measurement of serum estrogen concentration Adult male Wistar rats (16-wk-old) were purchased from Kiwa Lab. (Wakayama, Japan). Under anesthesia with sodium pentobarbital (40 mg/kg), 2 ml of the blood was collected from the portal vein close to the liver, and the left ventricle (arterial blood), and 1 ml of each serum was obtained. After extraction of the water-insoluble fraction with diethyl ether, the dried-up pellet was reconstituted in 200 ␮l standard serum (zero concentration). Extraction efficiency of E2 from serum was 92.6%. E2 content was measured with E2 RIA kit (Diagnostic Products, Los Angeles, CA) according to the manufacturer’s protocol. Cross-reactivities to estrone and estriol were 1.1% and 0.32%, respectively. No crossreactivities to other steroids were observed. The minimal limit of detection was 2 pg E2 in 1 ml serum.

Estimation of mRNA levels by RT-PCR Total RNAs were prepared by ISOGEN (Nippon Gene, Tokyo, Japan) from esophagus, stomach, small intestine, colon, spleen, liver, and ovary

3162

Ueyama et al. • Gastric Parietal Cells Secrete Estrogen

of 12-wk-old male and female Wistar rats. Expression of aromatase or estrogen receptor (ER) genes was determined by RT-PCR. Total RNA (1 ␮g) was converted into cDNA by reverse transcription using poly(dN)6 and poly(dT)12–18 primers (Amersham Pharmacia Biotech, Buckinghamshire, UK) and Moloney murine leukemia virus reverse transciptase (Life Technologies, Inc., Rockville, MD) in a total reaction vol of 50 ␮l. Primers were made using the following sequences based on the nucleotide sequences in the rat. Aromatase: nucleotides 1522–2059 of rat aromatase cytochrome P450 mRNA (13), 5⬘-TGACACCATGTCCGTCACTCT-3⬘ (forward), 5⬘-AATGGGGCTGTCCTCATCTA-3⬘ (reverse); ER␣: nucleotides of 1510 –1870 of rat ER␣ mRNA (14), 5⬘-TGGCTACGTCAAGTCGATTCC-3⬘ (forward), 5⬘-AGACGATGAGCATCCAGCAT-3⬘ (reverse); ER␤: nucleotides of 1051–1500 of rat ER␤1 mRNA (15) and 635-1138 of rat ER␤2 mRNA (16), 5⬘-AGCTACTGCTGAGCACCTTGA-3⬘(forward), 5⬘-TGAGGAGGATCATGGCCTTCA-3⬘(reverse), or nucleotides (17) of 454 –715 of rat ER␤1 mRNA and 38 –299 of rat ER␤2 mRNA, 5⬘-TTCCCGGCAGCACCAGTAACC-3⬘(forward), 5⬘-TCCCTCTTTGCGTTTGGACTA-3⬘(reverse). As an internal control, we also estimated the expression of rat ␤-actin mRNA (18) using the following primer set: 5⬘-TTGTAACCAACTGGGACGATATGG-3⬘ (forward), 5⬘GATCTTGATCTTCATGGTGCTAGG-3⬘ (reverse). The amplification protocol consisted of 25 cycles of denaturation for 30 sec at 95 C, annealing for 30 sec at 55 C, and extension for 1 min at 72 C. The PCR products were electrophoresed on 2% agarose gel stained with ethidium bromide and visualized by UV illumination. Specifically amplified products were quantified by densitometric scanning using NIH image software. ER␣ mRNA levels were normalized by ␤-actin mRNA.

In situ hybridization histochemistry Male Wistar rats (12-wk-old, n ⫽ 3) were deeply anesthetized with sodium pentobarbital (40 mg/kg) and perfused transcardially with saline followed by 4% paraformaldehyde in 0.1 m phosphate buffer, pH 7.4. The specimens were postfixed in the same fixative for 2 h and cryoprotected in PBS containing 30% sucrose. Frozen sections of 10 ␮m in thickness were cut in a cryostat, thaw-mounted on silane-coated slides, and stored at ⫺80 C until use. The probe sequences of aromatase and ER␣ were as follows: aromatase probe 1 (40-mer), complementary to nucleotides 741–780, and aromatase probe 2 (40-mer) 1331–1370 of rat aromatase mRNA (13); ER␣ probe 1 (40-mer), complementary to nucleotides 501–540, and ER␣ probe 2 (40-mer) 951–990 of rat ER␣ mRNA (14). Computer-assisted homology search revealed no identical sequences in any genes in the GenBank database. The probes were labeled with 35S-deoxy-ATP (35S-dATP) using terminal deoxynucleotidyl-transferase (Takara, Ootsu, Japan). The specific activity of each probe was 5–10 ⫻ 108 cpm/␮g. Excess (100⫻) amounts of nonlabeled probes completely eliminated the hybridization signals for the respective mRNAs, indicating that these signals were specific. Tissue sections were treated with 0.2 n HCl and digested with 1 ␮g/ml proteinase K at 37 C for 15 min. After postfixation with 4% paraformaldehyde in 0.1 m phosphate buffer (pH 7.4) for 5 min, the sections were immersed in 2 mg/ml glycine in PBS for 20 min. These sections were rinsed in PBS and then dehydrated through a graded ethanol series (70 –100%). Sections were hybridized overnight at 37 C in 100 ␮l buffer containing 4⫻ standard saline citrate, 50% formamide, 0.12 m phosphate buffer, 1⫻ Denhardt’s solution, 0.2% sodium dodecylsulfate, 250 ␮g/ml yeast transfer RNA, 10% dextran sulfate, and 0.1 m dithiothreitol with 5 ⫻ 106 cpm labeled probe per slide. After hybridization, sections were washed four times for 20 min at 55 C in 1⫻ standard saline citrate, immersed briefly in distilled water and dehydrated with a graded ethanol series, and then dried. The slides were coated with Ilford k-5 emulsion (Ilford, Knutsfold, UK) diluted 1:2 with water, for autoradiography, and then exposed for 4 –16 wk at 4 C. Slides were developed in D-19 (Eastman Kodak Co., Rochester, NY), and sections were counter-stained with hematoxylin for morphological examination. All slides were processed simultaneously for the same probes.

Immunohistochemistry Sections were incubated with 3% H2O2 in distilled water for 20 min to quench the endogenous peroxidase activity. After rinsing twice with PBS, they were incubated with the primary antiserum against human placental aromatase (rabbit, R-8-1), originally generated by Osawa and

Endocrinology, August 2002, 143(8):3162–3170 3163

colleagues (19, 20), diluted 1:1,000; or with the primary antiserum against rat ER (rabbit, no. 409), originally generated by Hayashi and colleagues (21), diluted 1:10,000 with 0.1 m PBS containing 5% normal goat serum and 0.3% Triton X-100 for 48 h at 4 C. The specificity and characterization of the antisera against human placental aromatase and rat ER were as described (20 –23). Briefly, the antibody against human placental aromatase suppresses human aromatase activity, with an IC50 value of 0.6 ␮l/ml incubation mixture, and it monospecifically reacts with aromatase cytochrome P450 in the Western blotting (20). The antibody against rat ER can react with the receptor, regardless of whether it is occupied by E2 (21). Omission of the primary or secondary antibody, or incubation with preimmune rabbit serum instead of the primary antiserum, completely eliminated the immunoreactivity as described (22, 23). After washing in PBS, they were incubated with the secondary antibody (biotinylated goat antirabbit IgG; Vector Laboratories, Inc., Burlingame, CA) diluted 1:200 in PBS, for 1 h at 37 C. After rinsing twice with PBS, they were incubated with avidin-biotin-HRP complex (ABC Elite kit; Vector Laboratories, Inc.) for 1 h. After washing in 0.05 m Tris-HCl buffer (pH 7.6), they were incubated in 0.05 m Tris-HCl buffer (pH 7.6) containing 0.02% 3,3⬘-diaminobenzidine tetrahydrochloride (DAB) and 0.005% H2O2, for 2–5 min. For electron microscopic examination, some of the sections were fixed in 1% OsO4 after DAB reaction, then embedded in Epon. Ultrathin sections of gastric mucosa showing strong positive immunostaining were examined with an electron microscope. For double-fluorescence immunohistochemistry, sections were incubated simultaneously with the primary antiserum against aromatase (rabbit, R-8-1), diluted 1:1,000, and monoclonal antibody against gastric proton pump (␤-subunit) (mouse; Affinity BioReagents, Inc. Golden, CO) diluted 1:6,000 with 0.1 m PBS containing 5% normal goat serum and 0.3% Triton X-100, for 48 h at 4 C. After rinsing twice with PBS, sections were incubated with the secondary antibody (biotinylated goat antirabbit IgG; Vector Laboratories, Inc.), diluted 1:200 in PBS, for 1 h at 37 C. Finally, they were incubated in 1:100 dilution of Texas-Red avidin D (Vector Laboratories, Inc.), simultaneously with 1:100 dilution of fluorescein isothiocyanate-conjugated goat antimouse IgG antibody (Jackson ImmunoResearch Laboratories, Inc., West Groove, PA) in 0.1 m PBS containing 5% normal goat serum and 0.3% Triton X-100, for 1 h at 37 C. They were rinsed twice with PBS and coverslipped with an antifade solution (VECTASHIELD; Vector Laboratories, Inc.).

Measurement of aromatase activity by 3H2O assay method Gastric mucosa and ovaries were obtained from 12-wk-old male or female Wistar rats. After deep anesthesia with 40 mg/kg sodium pentobarbital, the stomach was removed. The cavity was washed in cold saline, and the mucosal layer was scraped off with a blade and quickly frozen with liquid N2 and stored at – 80 C until assayed. The aromatization rate was determined by measuring the amount of tritiated water released from the labeled substrate into the incubation medium during aromatization (24). Briefly, tissues (40 –150 mg) were incubated, at 37 C for 60 min, with 4 ␮m [1␤-3H]androstenedione (specific radioactivity, 27.5 Ci/mmol; NEN Life Science Products, Boston, MA) in 50 ␮l of 50% methanol, 15 mg nicotinamide adenine dinucleotide phosphate (NADPH) in 3 ml of 67 mm phosphate buffer (pH 7.4) containing 250 mm sucrose and 1 mm EDTA. The reaction was stopped with 4 ml CHCl3, and the radioactivity of 1 ml aliquot of water layer was measured. After the cancellation of background radioactivity (NADPH free reaction), the aromatase activity was expressed as picomoles/100 mg tissue/60 min. In the inhibition experiment, 4-hydroxyandrostenedione (Formestane) (25), synthesized according to the method previously reported (26), was included in the incubation medium at a concentration of 10 or 20 ␮m, and the aromatase activity was determined as described.

Aromatization studies by gas chromatography-mass spectrometry (GC-MS) Production of E2 in the gastric mucosa was further confirmed by GC-MS (27, 28). Gastric mucosa (about 100 mg) from adult male Wistar rats was incubated at 37 C for 60 min with 4 ␮m androstenedione in 50 ␮l of 50% methanol, 15 mg NADPH in 3 ml of 67 mm phosphate buffer (pH 7.4) containing 250 mm sucrose and 1 mm EDTA. At the end of incubation, 200 ng [2, 4, 16, 16, 17␣-2H5]E2, which was prepared from

3164

Endocrinology, August 2002, 143(8):3162–3170

the corresponding nonlabeled estrone with the known method (29), was added as the internal standard to each incubation mixture and extracted twice with 7 ml ethyl acetate. The extract was reduced with 1 mg NaBH4 and then placed in a Sep-Pak C18 cartridge (Waters Corp., Milford, MA). The steroid fraction eluted with 80% methanol was placed in Sephadex LH 20 column (Amersham Biosciences Corp., Piscataway, NJ) in which each aromatized product was obtained in 1–14 ml fraction (benzene/ methanol, 95:5). The aromatized product was converted to the bistrimethylsilyl ether. In short, the aromatized product obtained from Sephadex LH 20 column was dissolved in dry pyridine (30 ␮l). Bistrimethylsilyltrifluoroacetamide (30 ␮l) was added separately to the solution, and the mixture was heated at 60 C for 30 min, and then the solvent was removed under a stream of N2 gas. The residue was dissolved in anhydrous hexane (25 ␮l), and then 2-␮l portions of the solution were subjected to analysis by GC-MS. The recovery rate of E2 was in the range of 65–70%. A MAT SSQ GC-MS instrument ( Finnigan, San Jose, CA) was used. Gas chromatographic conditions were as follows: column, 30 m ⫻ 0.250 mm internal diameter fused silica DB5 (J & W Scientific, Folsom, CA); column temperature, from 50 C to 250 C at 25 C/min and at 10 C/min to 280 C; carrier gas, helium at a flow rate of 1 ml/min. Retention time of the E2 silyl ether was 14.51 min. MS conditions were as follows: ionization energy, 70 eV; ion source temperature, 150 C. The quantitative analysis of the aromatized product was performed with a selected ionmonitoring method; the molecular ion [M⫹, m/z (mass/charge) ⫽ 416.3] was the base peak ion for E2, and then the amount of the E2 produced was obtained by the relative abundance of the molecular ion of the product to that of the internal standard.

Production of E2 in cultured gastric mucosa Gastric mucosa was obtained from 12-wk-old male Wistar rats as described above. Tissues (about 200 mg) were incubated in 0.5 ml DMEM (Invitrogen Corp., Carlsbad, CA) containing 10% charcoalabsorbed fetal calf serum (Equitech-Bio Inc., Ingram, TX), 100 U/ml penicillin, and 100 ␮g/ml streptomycin at 37 C under 95% air-5% CO2. One nanomole of [4-14C]androstenedione (specific radioactivity, 53.6 mCi/mmol; Perkin-Elmer Life Science Products, Boston, MA) was added to each dish. After 5 h, the reaction was stopped by putting on ice and adding excess amounts (30 nmol in each) of nonlabeled steroids: androstenedione, estrone, E2, dehydroepiandrosterone, and dihydrotestosterone. The media and tissues were collected separately. The media were extracted twice with 4 ml of ether. The tissues were homogenized by sonication in 0.5 ml of 50 mm Tris-maleate buffer, containing 250 mm sucrose (pH 7.5), and extracted twice with 4 ml diethylether. These extracts were dried up with N2 gas and redissolved with 20 ␮l methanol. Extraction efficiency of radioactivities was 90 –93%. Thin-layer chromatography (TLC) was performed on silica gel 60 F254 TLC plates (Merck KGaA, Darmstadt, Germany) developed once in the solvent system chloroform/ethyl acetate (3:1, vol/vol). The reference fronts of reference compounds were in increasing order: testosterone, 0.322; E2, 0.370; dehydroepiandrosterone, 0.459; dihydrotestosterone, 0.507; androstenedione, 0.589; estrone 0.685. The radioactive areas corresponding to 14C-metabolites were visualized and estimated using the Bioimaging analyzer BAS 2500 (Fuji Photo Film Co., Ltd., Tokyo, Japan). To observe the binding activity of the products after the incubation of [4-14C]androstenedione with gastric mucosa, the labeled androstenedione and its derivatives were extracted, as described above, without the addition of nonlabeled steroids. They were subjected to TLC; and the individual radioactive areas corresponding to testosterone, E2, androstenedione, and estrone were scraped off and extracted with 30 ␮l methanol. Each extract was diluted in 500 ␮l 10 mm Tris-HCl buffer (pH 7.4) containing 0.9% NaCl, and 0.35% BSA, and the aliquots (20 ␮l) were added to 160 ␮l of the antibody solution against E2 (7010 –7050; Biogenesis, Poole, UK). After incubation for 3 h at 37 C, 200 ␮l 5% activated charcoal solution, with 1% dextran in the same buffer as above, was added and centrifuged at 10,000 ⫻ g for 3 min at 4 C. The supernatants were dissolved in ACS II scintillation cocktail (Amersham Biosciences Corp.), and the radioactivity was counted by liquid scintillation counter LSC-5100 (Aloka, Tokyo, Japan).


Surgery Total gastrectomy (10-wk-old male Wistar rats, n ⫽ 6) was carried out by resecting the stomach, followed by anastomosing the duodenum and the esophagus end-to-end. Sham operation (n ⫽ 6) consisted of manipulation of the viscera. The mortality of this operation was 8%. Three weeks after surgery, blood samples were collected from the portal vein and the left ventricle, and tissue samples were taken from the liver. Portal-systemic shunting was performed using a two-stage procedure (10). The first operation was the sc transposition of the spleen, using 4-wk-old male Wistar rats. Five weeks later, the portal vein was ligated near the hilum of the liver (n ⫽ 6). Control (n ⫽ 6) was submitted to the first stage and to sham laparotomy at the time of the second stage. Four weeks after the second stage, blood samples from the left ventricle and tissue samples from the liver were taken. The mortality of these operations was 50%.

Statistical analysis Values are presented as the mean ⫾ sem. Statistical analysis was performed by either one-way ANOVA followed by Fisher’s protected least-significant-difference test or Student’s t test, as appropriate, using StatView software (Abacus Concepts, Berkeley, CA).

Results Step-up of serum E2 concentration in the portal vein

We compared the serum E2 concentration in the portal vein with that in arterial blood in adult male rats, using RIA. There was a significant step-up in serum E2 concentration (portal vein, 22.4 ⫾ 2.8 pg/ml; artery, 10.0 ⫾ 1.4 pg/ml; P ⬍ 0.01, n ⫽ 6). These data suggest that E2 produced in the gastrointestinal tract or other organs, including spleen, drains into the portal vein. Expression of aromatase in gastric parietal cells

To know the site of estrogen production in the gastrointestinal tract, we surveyed the mRNA encoding the rat aromatase, estrogen synthetase (12), using RT-PCR. Gastric mucosa obtained from adult rats of both sexes contained a substantial amount of aromatase mRNA, equivalent to that in the ovary (Fig. 1A). Weak expression of aromatase mRNA in the spleen was confirmed by increasing the number of PCR cycles (35 cycles) (data not shown). No aromatase mRNA was detected in other gastrointestinal organs, such as the liver, esophagus, small intestine, and colon. Strong signals for aromatase mRNA and immunoreactive aromatase protein were demonstrated in gastric parietal cells using in situ hybridization histochemistry (Fig. 1, B and C) and immunohistochemistry (Fig. 1, D and E), respectively. Parietal cells are characterized by a gastric proton pump (H⫹/K⫹ATPase), which produces a large quantity of gastric hydrochloric acid. Double fluorescence immunohistochemistry clearly indicated that immunoreactivity for the proton pump was present around the intracellular canaliculi, whereas immunoreactivity for aromatase was mainly located near the basal side (Fig. 1E). Immunoelectron microscopy demonstrated that immunoreactive aromatase protein was located on the rough endoplasmic reticulum (Fig. 1F). These data clearly indicate the expression of aromatase in gastric parietal cells.



Enzyme activity of aromatase and production of E2 in the gastric mucosa

We demonstrated aromatase activity in the gastric mucosa by the 3H2O assay method. Enzyme activities were equal among male and female gastric mucosa and the ovary (Fig. 2A). The ovarian aromatase activity changes, depending on the state of follicular development (30). Because we did not specify the stage of estrous cycle, we presented our data of the ovarian aromatase activity as a mean level, which was proved to be similar to gastric aromatase activity. These activities were inhibited by a suicide substrate of aromatase, 4-hydroxyandrostenedione, Formestane (25), in a dosedependent manner (Fig. 2B). These data show that the gastric mucosa can produce a large quantity of estrogen, equivalent to that in the ovary. Production of estrogen was further confirmed by GC-MS. The aromatization product was postulated to be a mixture of estrone and its 17␤-reduced compound, E2. Thus, the aromatization product obtained under the conditions used for the 3H2O assay method was reduced to E2 on treatment with NaBH4. The E2 was purified and analyzed as the bistrimethylsilyl ether by GC-MS (EI mode). Mass spectrum and retention time of the product were identical with those of authentic E2 (Fig. 2, C and D), indicating that the product was E2. Incubation with an adequate amount of substrate and the cofactor resulted in the formation of a significant amount of E2 (Table 1). The E2 production was significantly inhibited by 20 ␮m Formestane, as seen in the radiometric assay. Even in the incubation without the substrate (androstenedione) and the cofactor (NADPH), a low level of E2 was detected, suggesting the endogenous content of E2 in the gastric mucosa. Because we did not show the proportion of E2 to estrone in GC-MS analysis, we further confirmed that the gastric mucosa produced and secreted E2 as the major product in physiological conditions. After incubation in culture me-

FIG. 1. Expression and localization of aromatase in gastric parietal cells. A, Total RNAs (1 ␮g), prepared from esophagus, stomach, small intestine, colon, liver, spleen, and ovary of 12-wk-old male and female Wistar rats, were converted into cDNA by reverse transcriptase. The amplification protocol consisted of 25 cycles of denaturation for 30 sec at 95 C, annealing for 30 sec at 55 C, and extension for 1 min at 72 C. RT-PCR shows that aromatase mRNA is expressed in the gastric mucosa of male and female, as well as in ovary. B and C, Results of in situ hybridization histochemistry shows the expression of aromatase mRNA in the parietal cells. B, Darkfield photomicrograph; M, mucosa; ME, muscularis externa. C, Brightfield photomicrograph counter-stained with hematoxylin. The stomachs of male Wistar rats (12-wk-old), perfused with 4% paraformaldehyde, were sectioned in a cryostat (10 ␮m in thickness). The probes specific to rat aromatase

mRNA were labeled with 35S-dATP (specific activity, 5–10 ⫻ 108 cpm/␮g), and in situ hybridization histochemistry was performed. The slides were coated with Ilford k-5 emulsion for autoradiography and then exposed for 4 wk at 4 C. D, immunohistochemistry for aromatase (cytoplasmic staining, brown). Sections were reacted with the primary antiserum against human placental aromatase generated by Osawa et al., diluted 1:1000 for 48 h at 4 C. They were incubated with biotinylated goat antirabbit IgG, diluted 1:200 for 1 h at 37 C, followed by avidin-biotin-HRP complex for 1 h. Finally, they were visualized by DAB reaction. E, Double immunofluorescence for aromatase (red) and the proton pump (green). The sections were reacted simultaneously with the primary antiserum against aromatase (rabbit), diluted 1:1000, and monoclonal antibody against gastric proton pump (␤-subunit) (mouse), diluted 1:6000, for 48 h at 4 C. They were incubated with biotinylated goat antirabbit IgG, diluted 1:200 for 1 h at 37 C. Finally, they were incubated in 1:100 dilution of Texas-Red avidin D, simultaneously with 1:100 dilution of fluorescein isothiocyanate-conjugated goat antimouse IgG antibody, for 1 h at 37 C. F, Ultrathin sections of gastric mucosa showing strong positive immunostaining were examined with an electron microscope. The immunoelectron microscopy shows the localization of aromatase in the rough endoplasmic reticulum (arrows). N, Nucleus; m, mitochondria; icc, intracellular canaliculi; sER, smooth endoplasmic reticulum; L, lysosome. Scale bars, 200 ␮m (B), 30 ␮m (C and D), 15 ␮m (E), and 500 nm (F).

3166



FIG. 2. Aromatase activity and production of E2 in the gastric mucosa. A, Comparison of enzyme activity among male and female gastric mucosa, and ovary (n ⫽ 4), estimated by 3H2O assay. Tissues (40 –150 mg) obtained from 12-wk-old male or female Wistar rats were incubated at 37 C for 60 min with 4 ␮M [1␤-3H]androstenedione containing 15 mg NADPH in 3 ml 67 mM phosphate buffer, pH 7.4. After the cancellation of background radioactivity (NADPH free reaction), the aromatase activity was expressed as picomoles/100 mg tissue/60 min. Data are shown as the mean ⫾ SEM. There were no significant differences among the three groups. B, Inhibition of aromatase activity in gastric mucosa by a suicide substrate of aromatase, Formestane (n ⫽ 4), was also estimated by 3H2O assay. **, P ⬍ 0.001 vs. control. C and D, Mass spectrum of bistrimethylsilyl ether of authentic E2 (C) and bistrimethylsilyl ether of the product (D). Gastric mucosa (about 100 mg) from adult male Wistar rats was incubated at 37 C for 60 min with 4 ␮M androstenedione containing 15 mg NADPH in 3 ml 67 mM phosphate buffer, pH 7.4. The metabolites were extracted and reduced with 1 mg NaBH4 and then purified in a Sep-Pak C18 cartridge and subsequently in Sephadex LH 20 column. The aromatized product converted to the bistrimethylsilyl ether was subjected to analysis by GC-MS. Mass spectrum and retention time of the product were identical with those of authentic E2, indicating that the product was E2. E, TLC of [4-14C]androstenedione and its metabolites in the cultured gastric mucosa and cultured medium compared with preincubation sample (Control). Gastric mucosa (about 200 mg) from 12-wk-old male Wistar rats was cultured in 0.5 ml DMEM containing 1 nmol [4-14C]androstenedione (specific radioactivity, 53.6 mCi/ mmol) and 10% charcoal-absorbed fetal calf serum at 37 C, under 95% air-5% CO2, for 5 h. The metabolites were extracted and separated by TLC on silica gel 60 F254 TLC plates in the solvent system chloroform/ ethyl acetate (3:1, vol/vol). The radioactive areas corresponding to 14C-metabolites were visualized and estimated using the Bioimaging analyzer BAS 2500. T, E2, AD, and E1 are the spots corresponding to testosterone, E2, androstenedione, and estrone, respectively; 1 and 2 are unknown. Arrowhead, Origin of TLC. In the tissue culture system, gastric mucosa produced testosterone and E2 from androstenedione.

TABLE 1. E2 content, estimated by GC-MS analysis Androstenedione

NADPH

4 ␮M 4 ␮M

15 mg 15 mg

Formestane

20 ␮M

E2 analyzed (pmol/60 min 䡠 100 mg tissue)

12.8 ⫾ 0.9 (n ⫽ 4) 67.5 ⫾ 3.5 (n ⫽ 8)a 28.2 ⫾ 3.5 (n ⫽ 4)b

Data are shown as the mean ⫾ SEM. Incubation with enough substrate and the cofactor resulted in the formation of a significant amount of E2. The E2 production was significantly inhibited by 20 ␮M Formestane. Even in the incubation without the substrate and the cofactor, a low level of E2 was detected, suggesting the endogenous content of E2 in gastric mucosa. a P ⬍ 0.0001 vs. control. b P ⬍ 0.0001 vs. Formestane-free. 14

dium containing [ C]androstenedione, a considerable quantity of [14C]testosterone and [14C]E2 were detected in both the mucosal tissue and the culture medium (Fig. 2E). The prod-

uct from the spot corresponding to [14C]E2 bound to the antibody against E2, and the other products corresponding to testosterone, androstenedione, and estrone did not bind to the antibody (data not shown). The specific binding of the products to the antibody was reciprocally inhibited by authentic E2 at the concentration from 0.1–1000 nmol/reaction. The Scatchard analysis revealed that the product was identical with E2. Signals for [14C]estrone were weaker than that of [14C]E2 (8% of [14C]E2). These data suggest that gastric mucosa produces and secretes E2 rather than estrone in physiological conditions. Expression of ER␣ in liver and stomach

We investigated the ER subtypes in the gastrointestinal tract by RT-PCR (Fig. 3A), in situ hybridization histochemistry (Fig. 3, B and C), and immunohistochemistry (Fig. 3D).



ER␣ mRNA was demonstrated in the liver, gastric muscular layer, and ovary, whereas ER␤ mRNAs (ER␤1 and ER␤2) were observed only in the ovary. ER␤ mRNAs were not observed in the gastric mucosa by the same primer set that was reported previously (17). In the liver, hepatocytes showed strong signals for ER␣ mRNA and immunoreactive ER␣ protein, equivalent to those in the ovary, as previously reported (31). These data indicate that hepatocytes have the capacity to respond to estrogen derived from gastric parietal cells. Smooth muscle cells in the gastric wall also showed weak signals (Fig. 3, A and C). Effect of total gastrectomy or portal-systemic shunting

We estimated serum E2 concentration and hepatic ER␣ mRNA levels in response to total gastrectomy or portalsystemic shunting. Gastrectomy was performed using male rats at 10 wk of age, and the samples were taken after 3 wk. Gastrectomy resulted in a disappearance of the significant step-up between portal venous and arterial E2 concentration in the sham-operated rats (Fig. 4A), indicating that the stomach, but not the spleen, is the main source of portal venous E2. Interestingly, there was no significant difference in arterial E2 concentration between sham operation and gastrectomy (Fig. 4A), suggesting that, in the normal physiological state, most of the E2 produced in the stomach is trapped in the liver. In accord with the change of portal venous E2 concentration, hepatic ER␣ mRNA levels were down-regulated by gastrectomy (Fig. 4B). Portal-systemic shunting was produced by ligation of the portal vein (10). Portal-systemic shunting resulted in a marked increase of circulating E2 concentration (Fig. 4C), in agreement with previous observations (9, 10), but the hepatic ER␣ mRNA levels were not changed (Fig. 4D). Discussion

FIG. 3. Expression and localization of ER␣ in the liver and stomach. A, Total RNAs (1 ␮g), prepared from stomach, small intestine, liver, and ovary of 12-wk-old male and female Wistar rats, were converted into cDNA by reverse transcription. The amplification protocol consisted of 25 cycles denaturation for 30 sec at 95 C, annealing for 30 sec at 55 C, and extension for 1 min at 72 C. RT-PCR shows that ER␣ mRNA is strongly expressed in the liver and ovary and weakly in the stomach (gastric muscular layer) and small intestine, whereas ER␤1 (upper middle) and ER␤2 (lower middle) are expressed only in the ovary. Bright (B) and dark (C) field photomicrographs of in situ hybridization histochemistry show the expression of ER␣ mRNA in hepatocytes (B) and the gastric muscular layer (C). The stomach and the liver of male Wistar rats (12-wk-old), perfused with 4% paraformaldehyde, were sectioned in a cryostat (10 ␮m in thickness). The probes specific to rat ER␣ mRNA were labeled with 35S-dATP (specific activity, 5–10 ⫻ 108 cpm/␮g), and in situ hybridization histochemistry

The present study has shown that gastric parietal cells produce and secrete a substantial amount of E2 into the portal vein, using both biochemical and morphological methods. We showed a significant step-up of E2 concentration in the portal vein, compared with that in the artery, which was not seen after gastrectomy, and also an increase of arterial E2 concentration after portal-systemic shunting. This indicates that the stomach, but not other organs, is the main origin of portal venous E2. The strong expression of mRNA encoding aromatase, estrogen synthetase, was confirmed by RT-PCR in the gastric mucosa. A low amount of aromatase mRNA was also detected in the spleen, as reported (8); however, production of E2 seems to be limited. The gastric mucosa of both male and female adult rats showed a substantial activity

was performed. The slides were coated with Ilford k-5 emulsion for autoradiography and then exposed for 4 –16 wk at 4 C. D, Immunolocalization of ER␣ (brown) in the nuclei of hepatocytes. Sections were reacted with the primary antiserum against rat ER (rabbit) generated by Hayashi and colleagues (21), diluted 1:10,000 for 48 h at 4 C. They were incubated with biotinylated goat antirabbit IgG, diluted 1:200 for 1 h at 37 C followed by avidin-biotin-HRP complex for 1 h. Finally, they were visualized by DAB reaction. Scale bars, 15 ␮m (B and D) and 200 ␮m (C).

3168



FIG. 4. Alteration of serum E2 and hepatic ER␣ levels, in response to surgical treatments. A, Comparison of E2 concentration among sham-operated portal venous and arterial samples (n ⫽ 6), and gastrectomized portal venous and arterial samples (n ⫽ 5). **, P ⬍ 0.01 vs. the other three groups. Total gastrectomy was carried out by resecting the stomach, followed by anastomosing the duodenum and the esophagus end-to-end. Sham operation consisted of manipulation of the viscera. E2 content in the serum was measured with E2 RIA. Gastrectomy resulted in a disappearance of the significant step-up between portal venous and arterial E2 concentration in the sham-operated rats, indicating that the stomach is the main source of portal venous E2. There was no significant difference in arterial E2 concentration between sham operation and gastrectomy, suggesting that, in the normal physiological state, most of the E2 produced in the stomach is trapped in the liver. B, Comparison of ER␣ mRNA levels in the liver, between sham operation and gastrectomy, by semiquantitative RT-PCR. **, P ⬍ 0.01 vs. Sham. Specifically amplified products were quantified by densitometric scanning using NIH image software. ER␣ mRNA levels were normalized by ␤-actin mRNA. Hepatic ER␣ mRNA levels were down-regulated by gastrectomy. C, Comparison of arterial E2 concentration between sham operation (n ⫽ 3) and portal-systemic shunting (n ⫽ 3). *, P ⬍ 0.05 vs. Sham. Portal-systemic shunting was performed using a two-stage procedure. Controls were submitted to the first stage and to sham laparotomy at the time of the second stage. Portal-systemic shunting resulted in a marked increase of circulating E2 concentration, in agreement with previous observations. D, Comparison of ER␣ mRNA levels in the liver, between sham operation and portal-systemic shunting, by semiquantitative RT-PCR (no significant differences). Data are shown as the mean ⫾ SEM.

of aromatase. Enzyme activity and production of E2 were clearly suppressed by an aromatase inhibitor. In the physiological tissue culture system, gastric mucosa produced testosterone and E2 from androstenedione. These data indicate the presence of aromatase in the gastric mucosa. Because gastric mucosa contains a mixed population of cells, we further investigated the localization of aromatase mRNA and its protein, using in situ hybridization histochemistry and immunohistochemistry. These histochemical data indicate that the parietal cells in gastric mucosa have aromatase mRNA and its protein. Taken together, we confirmed the presence of biologically active aromatase in gastric parietal cells. We further showed that the signals for ER␣ mRNA and protein were located in hepatocytes (by RT-PCR, in situ hybridization histochemistry, and immunohistochemistry) and mRNA levels for hepatic ER␣ were regulated by E2 concentration in the blood flowing into the liver. Arterial E2 concentration did not show any significant decrease after gas-

trectomy but a marked increase after portal-systemic shunting. We therefore suggest that, in the normal physiological state, hepatocytes sequester most of the E2 secreted from gastric parietal cells. These findings of the present study led us to propose an expanded role of the gastric parietal cells beyond the wellknown exocrine function of secreting hydrochloric acid and intrinsic factor into the gastric juice. We provided evidence, for the first time, that the gastric parietal cells serve an endocrine function, whereby estrogen is synthesized and secreted into the portal vein, the consequences of which may be of clinical significance in the case of portal-systemic shunting. Generally, production of estrogen is known to be dependent on the availability of circulating C19 precursors in other tissues, such as adipose tissue, bone, and brain. In the ovary, the granulosa cells in preovulatory follicles are the main sites of synthesizing E2 from testosterone, which is provided by the theca interna and the interstitial gland (32).


In contrast, gastric parietal cells are capable of producing their own testosterone and androstenedione (33). Accordingly, gastric parietal cells are unique, in that they secrete E2 and can produce its precursor, testosterone. In fact, we demonstrated that cultured gastric mucosa converted androstenedione to testosterone, and subsequently to E2. Le Goascogne et al. (33) reported that cholesterol side-chain cleavage enzyme (P450scc) and 3␤-hydroxysteroid dehydrogenase were undetectable in the gastric mucosa. We also confirmed their findings by RT-PCR (unpublished observation). Thus, gastric parietal cells may use circulating progesterone derived from adrenal cortex and gonads as the first precursor for successive steps, or they can also use circulating testosterone preferentially as the substrate of aromatase. Detailed analysis of steroid biosynthesis, including estimation of portal venous levels of testosterone, progesterone, and other measurable steroids, is in progress in our group. Gastric parietal cells are also endowed with a large number of mitochondria and an elaborate tubulovesicular system, similar to a smooth-surfaced endoplasmic reticulum. These structural characteristics seem to be common in adrenal cortical cells and testicular interstitial cells. A second novel finding is that the gastric estrogen is not a simple sex steroid specific to female but a steroid common to both sexes, given that aromatase activity and mRNA in the gastric mucosa of both sexes were similar quantitatively to the levels found in the ovary. A functional role for gastric estrogen has still to be elucidated. As one possibility, gastric estrogen may act as a local regulator of the gastro-hepatic axis, because ER␣ mRNA and immunoreactive protein were expressed in hepatocytes, and mRNA levels of hepatic ER␣ were regulated by E2 concen-

FIG. 5. Schematic diagram showing the possible involvement of E2, secreted from gastric parietal cells, in the regulation of the gastro-hepatic axis. In the normal physiological state (left), E2 produced in the stomach is trapped by hepatic ER␣. Gastrectomy (upper right) results in reduction of portal E2 concentration and down-regulation of hepatic ER␣ levels. After portal-systemic shunting (lower right), gastric E2 overflows into the systemic circulation via collateral blood vessels, resulting in the elevation of systemic E2 concentration. Pink solid circles, E2; brown arcs, ER␣; arrows, the direction of blood flow.


tration in the blood flowing into the liver. The estrogen/ER system is thought to be involved in the process of liver regeneration (34), and also in the up-regulation of apolipoprotein E (35) and A-1 (36), thereby modulating the plasma lipid profile. Estrogen might affect gastric motility (37), because a low level of ER␣ mRNA was expressed in the muscularis externa of stomach. Gastric estrogen might also act directly on parietal cells in an autocrine or intracrine fashion, because the protein and mRNA expression of ER␣ and ER␤ was reported in gastric parietal cells (17, 38), but we could not confirm this finding. Or possibly, gastric estrogen might be involved in bone formation (39), considering that gastrectomy induces osteopenia in both humans and rats (40), and that estrogen is one of the factors known to affect bone metabolism. However, it should be noted that gastrectomy did not lower the levels of E2 in the arterial blood; thus, it is unlikely to be directly involved in the mechanism of bone loss. Last, feminization, which is often seen in male patients with increased portal-systemic shunting, is likely caused by the overflow of gastric estrogen into the systemic circulation, although the existence of aromatase in the human gastrointestinal tract has not been reported. In preliminary studies, we have observed the existence of aromatase mRNA and enzyme activity in the stomach, obtained from human samples, the detailed analysis of which is in progress in our group. The scheme in Fig. 5 summarizes the present results that illustrate the possible involvement of endocrine E2 secreted from gastric parietal cells in the regulation of gastro-hepatic axis. In the normal physiological state (left), gastric parietal cells produce and secrete a considerable amount of E2 into the portal vein, and the parietal cell-derived E2 is trapped by

3170


ER␣ of hepatocytes, resulting in only a slight overflow into the vena cava. Arterial E2 concentration, therefore, is not dependent on the E2 delivered from the stomach in the normal physiological state. In response to gastrectomy (upper right), portal venous E2 concentration is reduced to the circulating level, and the hepatic ER␣ is down-regulated; thus, arterial E2 concentration remains unchanged after gastrectomy. After portal-systemic shunting (lower right), a large amount of E2 produced in the stomach drains directly into the vena cava via collateral flow, resulting in an abnormal increase of arterial E2 concentration, which may be a cause of feminization in humans. Acknowledgments We thank Mr. Tasuo Sugawata and Mr. Noboru Iwasaki (Kawasaki, Kanagawa, Japan) for assistance with measuring E2 concentration, and Dr. Edith D. Hendley (Burlington, VT) and Dr. Arthur D. Loewy (St. Louis, MO) for helpful comments and careful reading of the manuscript. Received December 18, 2001. Accepted April 22, 2002. Address all correspondence and requests for reprints to: Dr. Takashi Ueyama, Department of Anatomy and Cell Biology, Wakayama Medical University, Kimiidera 811-1, Wakayama 641-8509, Japan. E–mail: [email protected]. This work was supported, in part, by Grants-in-Aid from Japan Society for the Promotion of Science, Wakayama Medical Award for Young Researchers, and Medical Research Grants from Wakayama Foundation for the Promotion of Medicine.

References 1. Simpson E, Rubin G, Clyne C, Robertson K, O’Donnell L, Davis S, Jones M 1999 Local estrogen biosynthesis in males and females. Endocr Relat Cancer 6:131–137 2. Leshin M, Baron J, George FW, Wilson JD 1981 Increased estrogen formation and aromatase activity in fibroblasts cultured from the skin of chickens with the Henny feathering trait. J Biol Chem 256:4341– 4344 3. Brodie A, Inkster S 1993 Aromatase in the human testis. J Steroid Biochem Mol Biol 44:549 –555 4. Lephart ED 1996 A review of brain aromatase cytochrome P450. Brain Res Rev 22:1–26 5. Ackerman GE, Smith ME, Mendelson CR, MacDonald PC, Simpson ER 1981 Aromatization of androstenedione by human adipose tissue stromal cells in monolayer culture. J Clin Endocrinol Metab 53:412– 417 6. Purohit A, Flanagan AM, Reed MJ 1992 Estrogen synthesis by osteoblast cell lines. Endocrinology 131:2027–2029 7. Harada N, Sasano H, Murakami H, Ohkuma T, Nagura H, Takagi Y 1999 Localized expression of aromatase in human vascular tissues. Circ Res 84: 1285–1291 8. Samy TS, Kno¨ferl MW, Zheng R, Schwacha MG, Bland KI, Chaudry IH 2001 Divergent immune responses in male and female mice after trauma-hemorrhage: dimorphic alterations in T lymphocyte steroidogenic enzyme activities. Endocrinology 142:3519 –3529 9. Braunstein GD 1999 Aromatase and gynecomastia. Endocr Relat Cancer 6:315–324 10. Farrell GC, Koltai A, Murray M 1988 Source of raised serum estrogens in male rats with portal bypass. J Clin Invest 81:221–228 11. Stauber RE, Rosenblum E, Eagon PK, Gavaler JS, Van Thiel DH 1991 The effect of portal-systemic shunting on hepatic sex hormone receptors in male rats. Gastroenterology 100:168 –174 12. Simpson ER, Mahendroo MS, Means GD, Kilgore MW, Hinshelwood MM, Graham-Lorence S, Amarneh B, Ito Y, Fisher CR, Michael MD, Mendelson CR, Bulun SE 1994 Aromatase cytochrome P450, the enzyme responsible for estrogen biosynthesis. Endocr Rev 15:342–355 13. Hickey GJ, Krasnow JS, Beattie WG, Richards JS 1990 Aromatase cytochrome P450 in rat ovarian granulosa cells before and after luteinization: adenosine 3⬘,5⬘-monophosphate-dependent and independent regulation. Cloning and sequencing of rat aromatase cDNA and 5⬘ genomic DNA. Mol Endocrinol 4:3–12 14. Koike S, Sakai M, Muramatsu M 1987 Molecular cloning and characterization of rat estrogen receptor cDNA. Nucleic Acids Res 15:2499 –2513 15. Kuiper GG, Enmark E, Pelto-Huikko M, Nilsson S, Gustafsson JA 1996


16.

17. 18. 19.

20. 21.

22. 23. 24. 25. 26. 27. 28. 29. 30.

31. 32. 33.

34.

35.

36. 37. 38. 39. 40.

Cloning of a novel receptor expressed in rat prostate and ovary. Proc Natl Acad Sci USA 93:5925–5930 Maruyama K, Endoh H, Sasaki-Iwaoka H, Kanou H, Shimaya E, Hashimoto S, Kato S, Kawashima H 1998 A novel isoform of rat estrogen receptor ␤ with 18 amino acid insertion in the ligand binding domain as a putative dominant negative regular of estrogen action. Biochem Biophys Res Commun 246: 142–147 Campbell-Thompson ML 1997 Estrogen receptor ␣ and ␤ expression in upper gastrointestinal tract with regulation of trefoil factor family 2 mRNA levels in ovariectomized rats. Biochem Biophys Res Commun 240:478 – 483 Nudel U, Zakut R, Shani M, Neuman S, Levy Z, Yaffe D 1983 The nucleotide sequence of the rat cytoplasmic ␤-actin gene. Nucleic Acids Res 11:1759 –1771 Osawa Y, Yoshida N, Fronckowiak M, Kitawaki J 1987 Immunoaffinity purification of aromatase cytochrome P-450 from human placental microsomes, metabolic switching from aromatization to 1␤ and 2␤-monohydroxylation, and recognition of aromatase isozymes. Steroids 50:11–28 Kitawaki J, Yoshida N, Osawa Y 1989 An enzyme-linked immunosorbent assay for quantitation of aromatase cytochrome P-450. Endocrinology 124: 1417–1423 Okamura H, Yamamoto K, Hayashi S, Kuroiwa A, Muramatsu M 1992 A polyclonal antibody to the rat oestrogen receptor expressed in Escherichia coli: characterization and application to immunohistochemistry. J Endocrinol 135: 333–341 Tsuruo Y, Ishimura K, Fujita H, Osawa Y 1994 Immunocytochemical localization of aromatase-containing neurons in the rat brain during pre- and postnatal development. Cell Tissue Res 278:29 –39 Tsuruo Y, Ishimura K, Hayashi S, Osawa Y 1996 Immunohistochemical localization of estrogen receptors within aromatase-immunoreactive neurons in the fetal and neonatal rat brain. Anat Embryol 193:115–121 Siiteri PK, Thompson EA 1975 Studies of human placental aromatase. J Steroid Biochem 6:317–322 Wiseman LR, Goa KL 1996 Formestane. A review of its pharmacological properties and clinical efficacy in the treatment of postmenopausal breast cancer. Drugs Aging 9:292–306 Numazawa M, Ogata M 1986 Methoxylation of enolizable steroidal 4-ene-3ones using hypervalent iodine. J Chem Soc Chem Commun 14:1092–1093 Numazawa M, Yoshimura A, Oshibe M 1998 Enzymatic aromatization of 6-alkyl-substituted androgens, potent competitive and mechanism-based inhibitors of aromatase. Biochem J 329:151–156 Numazawa M, Yoshimura A 1999 Biological aromatization of ␦4, 6- and ␦1, 4, 6- androgens and their 6-alkyl analogs, potent inhibitors of aromatase. J Steroid Biochem Mol Biol 70:189 –196 To¨ke´s L, Throop LJ 1972 Introduction of deuterium into steroid system. In: Freid J, Edwards JA, eds. Organic reactions in steroid chemistry. Vol 1. New York: Van Nonstrand Reinhold; 145–221 Lephart ED, Doody KJ, McPhaul MJ, Simpson ER 1992 Inverse relationship between ovarian aromatase cytochrome P450 and 5␣-reductase enzyme activities and mRNA levels during the estrous cycle in the rat. J Steroid Biochem Mol Biol 42:439 – 447 Yamashita S, Korach KS 1989 A modified immunohistochemical procedure for the detection of estrogen receptor in mouse tissues. Histochemistry 90: 325–330 Ishimura K, Yoshinaga-Hirabayashi T, Tsuri H, Fujita H, Osawa Y 1989 Further immunocytochemical study on the localization of aromatase in the ovary of rats and mice. Histochemistry 90:413– 416 Le Goascogne C, Sanane`s N, Eychenne B, Goue´zou M, Baulieu EE, Robel P 1995 Androgen biosynthesis in the stomach: expression of cytochrome P450 17␣-hydroxylase/17, 20-lyase messenger ribonucleic acid and protein, and metabolism of pregnenolone and progesterone by parietal cells of the rat gastric mucosa. Endocrinology 136:1744 –1752 Francavilla A, Polimeno L, DiLeo A, Barone M, Ove P, Coetzee M, Eagon P, Makowka L, Ambrosino G, Mazzaferro V, Starzl TE 1989 The effect of estrogen and tamoxifen on hepatocyte proliferation in vivo and in vitro. Hepatology 9:614 – 620 Srivastava RA, Srivastava N, Averna M, Lin RC, Korach KS, Lubahn DB, Schonfeld G 1997 Estrogen up-regulates apolipoprotein E (ApoE) gene expression by increasing ApoE mRNA in the translating pool via the estrogen receptor ␣-mediated pathway. J Biol Chem 272:33360 –33366 Lamon-Fava S, Ordovas JM, Schaefer EJ 1999 Estrogen increases apolipoprotein (apo) A-I secretion in hep G2 cells by modulating transcription of the apo A-I gene promoter. Arterioscler Thromb Vasc Biol 19:2960 –2965 Bond EF, Heitkemper MM, Bailey SL 1998 Estrogen suppresses gastric motility response to thyrotropin-releasing hormone and stress in awake rats. Res Nurs Health 21:221–228 Campbell-Thompson M, Reyher KK, Wilkinson LB 2001 Immunolocalization of estrogen receptor ␣ and ␤ in gastric epithelium and enteric neurons. J Endocrinol 171:65–73 Compston JE 2001 Sex steroids and bone. Physiol Rev 81:419 – 447 Lehto-Axtelius D, Stenstro¨m M, Johnell O 1998 Osteopenia after gastrectomy, fundectomy or antrectomy: an experimental study in the rat. Regul Pept 78:41–50