0013-7227/90/1276-2804$02.00/0 Endocrinology Copyright © 1990 by The Endocrine Society
Vol. 127, No. 6 Printed in U.S.A.
Transforming Growth Factor-/? 1 Inhibits Ovarian Androgen Production: Gene Expression, Cellular Localization, Mechanisms(s), and Site(s) of Action* ELEUTERIO R. HERNANDEZ, ARYE HURWITZ, DONNA W. PAYNE, A. M. DHARMARAJAN, ANTHONY F. PURCHIO, AND ELI Y. ADASHI Division of Reproductive Endocrinology, Department of Obstetrics and Gynecology, University of Maryland School of Medicine (E.R.H., A.H., D. W.P., E. Y.A.), Baltimore, Maryland 21201; the Department of Gynecology and Obstetrics, Johns Hopkins University School of Medicine (A.M.D.), Baltimore, Maryland 21205; and Oncogen (A.F.P.), Seattle, Washington 98121
ABSTRACT. It is the aim of this study to establish ovarian transforming growth factor-/Jl (TGF/31) gene expression, to reevaluate its cellular localization, and to explore potential interactions of this regulatory peptide on ovarian androgen biosynthesis. Northern analysis of whole ovarian polyadenylated RNA revealed a single 2.5-kilobase transcript corresponding to the TGF01 precursor. Immunohistochemical staining localized the protein to the thecal-interstitial (interfollicular) compartment. To explore potential autocrine effects of TGF/31, use was made of whole ovarian dispersates from immature rats the differentiation of which was monitored by the acquisition of androgen biosynthetic capacity. The accumulation of androsterone, the major androgenic steroid detectable in this culture system, increased 5.4-fold over baseline in response to treatment with hCG (1 ng/ml). This effect was further optimized (2- to 4fold) by supplementation with insulin (1 fig/ml) and insulin-like growth factor-I (50 ng/ml). In the absence of these optimizing supplements, TGF/S1 (10 ng/ml) was without effect on basal androsterone accumulation, producing distinct, albeit relatively limited (25%), inhibition of hCG hormonal action. In contrast, supplement-mediated optimization of ovarian androgen biosyn-
T
RANSFORMING growth factor-/31 (TGF01), a homodimeric polypeptide comprising two identical 112-amino acid chains, is now well recognized as a polyfuctional regulatory molecule (1). Originally identified by its ability to elicit an acute reversible phenotypic transformation of normal mammalian cells, TGF/31 has now been shown to exert numerous regulatory actions in a wide variety of both normal and neoplastic cells (1). At
Received September 24,1989. Address all correspondence and requests for reprints to: Dr. Eleuterio R. Hernandez, Division of Reproductive Endocrinology, Department of Obstetrics and Gynecology, University of Maryland School of Medicine, Room 11-007, Bressler Building, 655 West Baltimore Street, Baltimore, Maryland 21201. * This work was supported in part by a grant from the Frank C. Bressler Research Fund (to E.R.H.), the Lalor Foundation and the Rockefeller Foundation (to A.M.D.), NIH Research Grant HD-19998, and USPHS Research Career Development Award HD-00697 (to E.Y.A.).
thesis revealed TGF/31 to be a highly potent inhibitor (>80%) of hCG hormonal action. This reversible TGF/J1 action proved time and dose dependent, with a minimal time requirement of 72 h and a median inhibitory dose of 2.6 ng/ml. TGF/31 action was not due to diminution in the viable cell mass or altered cAMP generation and, therefore, most likely involved a site(s) of action distal to or independent of cAMP generation. Cellular radiolabeling studies of TGF/31 -treated ovarian cells disclosed the accumulation of steroid intermediates proximal to the 17ahydroxylation step, suggesting TGF/81-mediated blockade at the level of the steroidogenic enzyme 17a-hydroxylase/17-20-lyase. Taken together, these observations are in keeping with the view that TGF/31, possibly of thecal-interstitial origin, may not only play a positive paracrine role at the level of the adjacent granulosa cell (as previously reported), but may also constitute one of several autocrine signals concerned with the regulation of ovarian androgen economy. As such, these findings reaffirm the polyfunctional nature of TGF/31 action, as manifested by its diametrically opposed effects in different ovarian compartments. (Endocrinology 127: 2804-2811,1990)
the level of the ovary, TGF/31 has been shown to profoundly alter the proliferation (2-4) and differentiation (5-14) of rat granulosa cells. In addition, an increasing body of evidence now suggests that the ovarian thecalinterstitial (2, 3, 15) as well as granulosa (16) cells may be sites of TGF/31 production. The potential relevance of TGF/31 to ovarian physiology was highlighted by the recent discovery that ovarian inhibins are members of a growing gene family of structurally related regulatory peptides, of which TGF01 is a prominent representative (17). Moreover, activin-A (/3A/ /3A) and inhibin (a//3A) have been shown to exert opposing effects on gonadotropin-driven thecal-interstitial cell androgen biosynthesis (18). The effect of TGF/31 on ovarian androgen biosynthesis has not been explored. The present study was designed to establish ovarian TGF/31 gene expression, to reevaluate its cellular local-
2804
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 13 November 2015. at 03:13 For personal use only. No other uses without permission. . All rights reserved.
TGF/31 INHIBITS OVARIAN ANDROGEN PRODUCTION
2805
ization, and to explore the possibility that TGF/31 may also be involved in the regulation of ovarian androgen biosynthesis.
acetate was added to extract the product. The [3H]pregnenolone product was repurified on HPLC (90% pure) and coeluted with standard unlabeled pregnenolone.
Materials and Methods
Preparation of whole ovarian dispersates and highly enriched thecal-interstitial cells
Animals Mature (>60 days old) and immature (21-23 days old) Sprague-Dawley female rats from Johnson Laboratories, Inc. (Bridgeview, IL), were killed within 1-2 days after delivery. Reagents and hormones McCoy's 5a medium (modified, without serum), penicillin (10,000 IU/ml), streptomycin (10,000 Mg/ml solution), L-glutamine (29.2 mg/ml), trypan blue stain (0.4%, wt/vol), BSA, and DNase (bovine pancreas; 2,100 U/mg) were obtained from Gibco (Grand Island, NY). Collagenase (Clostridium histolyticum; CLS type I; 144 U/mg) was obtained from Worthington Biochemical Corp. (Freehold, NJ). Highly purified insulin-like growth factor-I (IGF-I), revealing no detectable impurities on polyacrylamide gel electrophoresis, was prepared from Conn fraction IV of human plasma and generously provided by Dr. Judson J. Van Wyk, University of North Carolina (Chapel Hill, NC). Bovine insulin was obtained from Sigma Chemical Co. (St. Louis, MO). Highly purified platelet-derived TGF/31 of human origin was generously provided by Dr. Michael B. Sporn, Laboratory of Chemoprevention, NCI, NIH (Bethesda, MD). Highly purified TGF/31 precursor and TGF/32 of human origin were generously provided by Dr. David Twardzik, Oncogen (Seattle, WA). Highly purified inhibin (a//3A) of porcine follicular fluid origin was generously provided by Drs. Wylie W. Vale and Joan M. Vaughan, The Salk Institute (San Diego, CA). Human high density lipoprotein (HDL) was generously provided by Dr. Michael Henson, University of Maryland (Baltimore, MD). Highly purified activin-A (/3A//3A) of porcine origin was generously provided by Dr. Nicholas C. Ling, The Salk Institute. Highly purified hCG (CR-121; 13,450 IU/mg) was generously provided by Dr. R. E. Canfield through the Center for Population Research, NICHHD, NIH (Bethesda, MD). [7-N-3H]Pregnenolone (15 Ci/mmol) and [9,ll-N-3H]androsterone (42 Ci/mmol) were obtained from New England Nuclear (Boston, MA). [3H]Pregnenolone (3a-hydroxy-5apregnan-20-one), not commercially available, was biologically and enzymatically synthesized by us as follows. Granulosa cells from immature intact rats were cultured for 72 h in the presence of [3H]pregnenolone (5 X 106cpm/3 x 105 cells). Labeled steroids in the medium were then extracted, pooled, and subjected to HPLC (see below). Fractions containing the major metabolite, 5a-pregnane-3a,20a-diol (5a-pregnanediol), were collected, evaporated to dryness, and enzymatically oxidized (66% conversion) to pregnenolone using a bacterial 17/3-hydroxysteroid dehydrogenase (HSD) containing minor 20aHSD activity (19). In addition to the enzyme, the reaction mixture (1 ml) contained 100 mM 2-[iV-cyclohexylamino]ethanesulfonic acid (pH 9), 10% methanol, 0.2 mM NAD, and 0.05% BSA. The reaction proceeded for 60 min at room temperature before ethyl
Whole ovarian dispersates from immature rats were prepared as previously described (20). Briefly, ovaries were dissected free of their bursae, divided into four to six pieces, washed (twice) with 10 ml McCoy's 5a medium, and subjected to enzymatic dispersion for 45 min at 37 C using 0.4% (wt/vol) collagenase, 0.001% (wt/vol) DNase, and 0.1% (wt/vol) BSA. In the course of this incubation period, the ovaries were dissociated into a cell suspension by repeated pipetting every 30 min with a graded series of micropipettes (id, 0.5-1.0 mm). At the end of the incubation period, the cells were collected by centrifugation at 250 x g for 5 min, washed (three times) with McCoy's 5a medium, and then resuspended into a known volume of the same. The composition of whole ovarian dispersates at the time of plating was previously discussed (21), granulosa cells constituing 80% and thecal-interstitial cells 10-20% of the whole. Purification of ovarian thecal-interstitial cells was carried out by density gradient centrifugation, as previously described (22). Cell viability, as assessed by trypan blue dye exclusion, was consistently greater than 85%. Tissue culture procedures Whole ovarian dispersates (3 x 105 viable cells/culture) and highly enriched thecal-interstitial cells (5 X 104 viable cells/ culture) were inoculated onto tissue culture dishes (35 x 10 mm; Falcon Plastics, Oxnard, CA) containing 1 ml McCoy's 5a medium (modified without serum) supplemented with L-glutamine (2 mM), penicillin (100 U/ml), and streptomycin sulfate (100 fig/ml). Cell cultures were maintained for 96 h (unless indicated otherwise) at 37 C under a water-saturated atmosphere of 5% CO2and 95% air. The various experimental agents (diluted in sterile culture medium) were applied in 50-^1 aliquots. HPLC Ether- or ethyl acetate-extracted medium steroids were fractionated on a Lichrosorb Diol column (EM Reagents, Cherry Hill, NJ) calibrated with 35 steroids, as previously described (23). Radiolabeled metabolites were detected in-line with a radioactivity detector (Flo-One Beta, Radiomatic Instruments, Tampa, FL). RIA Androsterone (3a-hydroxy-5«-androstane-17-one), the main androgenic steroid identified in culture medium by HPLC, was quantified by RIA as previously described (23). The antiserum (AO-1) used was raised against androsterone- 16-BSA, generously provided by Dr. D. T. Armstrong, University of Western Ontario (London, Canada). cAMP RIA kit was purchased from Biochemical Technologies, Inc. (Stoughton, MA).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 13 November 2015. at 03:13 For personal use only. No other uses without permission. . All rights reserved.
TGF/31 INHIBITS OVARIAN ANDROGEN PRODUCTION
2806
Endo • 1990 Vol 127 • No 6
-28s
Northern blot analysis Polyadenylated [poly(A)+] RNA was isolated by the guanidinium hydrochloride procedure, fractionated on a 1% agaroseformaldehyde gel, transferred to a nylon membrane (Hybond, Amersham, Arlington Heights, IL), and hybridized to a 32Plabeled simian TGF/31 cDNA probe, as previously described (24). This probe codes for the entire TGFjSl precursor polypeptide (391 amino acids), the mature TGF/31 portion of which consists of the 112 carboxy-terminal residues (24).
2.5 kb
-18s
Immunohistochemistry The rabbit TGF/31 polyclonal antibody [anti-LC-(l-30)] was generously provided by Kathleen C. Flanders, Laboratory of Chemoprevention, NCI, NIH. Sections (5 /xm) of Bouin-fixed dehydrated paraffin-embedded ovaries were deparaffinized, blocked, and incubated with 0.1 mg/ml anti-TGF/3 (1:60 dilution) at room temperature overnight. Sections were washed extensively and then incubated with biotinylated goat antirabbit immunoglobulin G (IgG). Antibody localization was achieved using a rabbit IgG ABC-Vectastain kit (Burlingame, CA) with 3,3'-diaminobenzidine (Sigma Chemical Co.) as the substrate. Slides were counterstained in Gill's hematoxylin (Sigma Chemical Co.). Control sections were stained in parallel with normal rabbit IgG or buffer alone. Neither control exhibited detectable staining.
M FIG. 1. Ovarian TGF/31 gene expression in immature (I) and mature (M) intact rats. Ten micrograms of poly(A)+ from the ovaries of immature and mature rats were employed to generate the Northern blot, as described in Materials and Methods. 18S and 28S = Ribosomal RNA.
Data analysis Data are presented as the mean ± SE of three or more separate determinations. RIA data analysis, dose-response curve fitting, and determination of the median inhibitory (EC50) dose were carried out on an IBM AT PC using CURVE FIT software based on the four-parameter logistic equation designed to fit the results to a sigmoidal function curve. Statistical significance was determined by Student's t test or analysis of variance, as indicated.
Results Ovarian TGFfil gene expression and cellular localization To determine if the ovary is a site of TGF/31 gene expression, poly(A)+ RNA extracted from the ovaries of immature and mature rats was analyzed (Fig. 1). Northern blot analysis revealed a single major transcript with an apparent size of 2.5 kilobase corresponding to the TGF/31 precursor (24). This transcript appears identical to that observed in a variety of species including the rat (1) and establishes the ovary as a site of TGF/31 gene expression. Given that immunoreactive TGF/31 has been localized to both the thecal-interstitial (2, 3) and granulosa (16) cells, we have also evaluated its cellular localization in situ in the immature ovary by immunohistochemical staining. As shown (Fig. 2), the TGF/31 protein, as in the mice (15), was exclusively localized to the thecal-interstitial (interfollicular) compartment. No staining was
FIG. 2. Ovarian TGF/81: cellular localization. Sections were processed as described in Materials and Methods. Interstitial staining was absent in control sections, in which antibody has been replaced with nonimmune rabbit serum or buffer (not shown).
detected when employing nonimmune rabbit serum or buffer (not shown). TGFfil inhibits gonadotropin-supported androsterone accumulation by cultured whole ovarian dispersates To examine the role of TGF/31 in thecal-interstitial cell differentiation, whole ovarian dispersates were cultured in the absence or presence of hCG (1 ng/ml), with or without increasing concentrations (0.01-10 ng/ml) of TGF/31 (Fig. 3). Basal androsterone accumulation was relatively low (15 ± 2ng/ml) and remained unaffected after treatment with TGF/31 (10 ng/ml) by itself. As expected, treatment with hCG substantially increased androsterone accumulation (5.8-fold). However, concurrent treatment with increasing concentrations of TGF01
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 13 November 2015. at 03:13 For personal use only. No other uses without permission. . All rights reserved.
TGF/31 INHIBITS OVARIAN ANDROGEN PRODUCTION
2807
400 .,
Z
o
80
H 60
200 . 3 O U
z o x w H
40
UJ
z o
(T UJ h
en
20
O X
o
Q
(E
Z
Q
Z
Vol. 127, No. 6 Printed in U.S.A.
Transforming Growth Factor-/? 1 Inhibits Ovarian Androgen Production: Gene Expression, Cellular Localization, Mechanisms(s), and Site(s) of Action* ELEUTERIO R. HERNANDEZ, ARYE HURWITZ, DONNA W. PAYNE, A. M. DHARMARAJAN, ANTHONY F. PURCHIO, AND ELI Y. ADASHI Division of Reproductive Endocrinology, Department of Obstetrics and Gynecology, University of Maryland School of Medicine (E.R.H., A.H., D. W.P., E. Y.A.), Baltimore, Maryland 21201; the Department of Gynecology and Obstetrics, Johns Hopkins University School of Medicine (A.M.D.), Baltimore, Maryland 21205; and Oncogen (A.F.P.), Seattle, Washington 98121
ABSTRACT. It is the aim of this study to establish ovarian transforming growth factor-/Jl (TGF/31) gene expression, to reevaluate its cellular localization, and to explore potential interactions of this regulatory peptide on ovarian androgen biosynthesis. Northern analysis of whole ovarian polyadenylated RNA revealed a single 2.5-kilobase transcript corresponding to the TGF01 precursor. Immunohistochemical staining localized the protein to the thecal-interstitial (interfollicular) compartment. To explore potential autocrine effects of TGF/31, use was made of whole ovarian dispersates from immature rats the differentiation of which was monitored by the acquisition of androgen biosynthetic capacity. The accumulation of androsterone, the major androgenic steroid detectable in this culture system, increased 5.4-fold over baseline in response to treatment with hCG (1 ng/ml). This effect was further optimized (2- to 4fold) by supplementation with insulin (1 fig/ml) and insulin-like growth factor-I (50 ng/ml). In the absence of these optimizing supplements, TGF/S1 (10 ng/ml) was without effect on basal androsterone accumulation, producing distinct, albeit relatively limited (25%), inhibition of hCG hormonal action. In contrast, supplement-mediated optimization of ovarian androgen biosyn-
T
RANSFORMING growth factor-/31 (TGF01), a homodimeric polypeptide comprising two identical 112-amino acid chains, is now well recognized as a polyfuctional regulatory molecule (1). Originally identified by its ability to elicit an acute reversible phenotypic transformation of normal mammalian cells, TGF/31 has now been shown to exert numerous regulatory actions in a wide variety of both normal and neoplastic cells (1). At
Received September 24,1989. Address all correspondence and requests for reprints to: Dr. Eleuterio R. Hernandez, Division of Reproductive Endocrinology, Department of Obstetrics and Gynecology, University of Maryland School of Medicine, Room 11-007, Bressler Building, 655 West Baltimore Street, Baltimore, Maryland 21201. * This work was supported in part by a grant from the Frank C. Bressler Research Fund (to E.R.H.), the Lalor Foundation and the Rockefeller Foundation (to A.M.D.), NIH Research Grant HD-19998, and USPHS Research Career Development Award HD-00697 (to E.Y.A.).
thesis revealed TGF/31 to be a highly potent inhibitor (>80%) of hCG hormonal action. This reversible TGF/J1 action proved time and dose dependent, with a minimal time requirement of 72 h and a median inhibitory dose of 2.6 ng/ml. TGF/31 action was not due to diminution in the viable cell mass or altered cAMP generation and, therefore, most likely involved a site(s) of action distal to or independent of cAMP generation. Cellular radiolabeling studies of TGF/31 -treated ovarian cells disclosed the accumulation of steroid intermediates proximal to the 17ahydroxylation step, suggesting TGF/81-mediated blockade at the level of the steroidogenic enzyme 17a-hydroxylase/17-20-lyase. Taken together, these observations are in keeping with the view that TGF/31, possibly of thecal-interstitial origin, may not only play a positive paracrine role at the level of the adjacent granulosa cell (as previously reported), but may also constitute one of several autocrine signals concerned with the regulation of ovarian androgen economy. As such, these findings reaffirm the polyfunctional nature of TGF/31 action, as manifested by its diametrically opposed effects in different ovarian compartments. (Endocrinology 127: 2804-2811,1990)
the level of the ovary, TGF/31 has been shown to profoundly alter the proliferation (2-4) and differentiation (5-14) of rat granulosa cells. In addition, an increasing body of evidence now suggests that the ovarian thecalinterstitial (2, 3, 15) as well as granulosa (16) cells may be sites of TGF/31 production. The potential relevance of TGF/31 to ovarian physiology was highlighted by the recent discovery that ovarian inhibins are members of a growing gene family of structurally related regulatory peptides, of which TGF01 is a prominent representative (17). Moreover, activin-A (/3A/ /3A) and inhibin (a//3A) have been shown to exert opposing effects on gonadotropin-driven thecal-interstitial cell androgen biosynthesis (18). The effect of TGF/31 on ovarian androgen biosynthesis has not been explored. The present study was designed to establish ovarian TGF/31 gene expression, to reevaluate its cellular local-
2804
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 13 November 2015. at 03:13 For personal use only. No other uses without permission. . All rights reserved.
TGF/31 INHIBITS OVARIAN ANDROGEN PRODUCTION
2805
ization, and to explore the possibility that TGF/31 may also be involved in the regulation of ovarian androgen biosynthesis.
acetate was added to extract the product. The [3H]pregnenolone product was repurified on HPLC (90% pure) and coeluted with standard unlabeled pregnenolone.
Materials and Methods
Preparation of whole ovarian dispersates and highly enriched thecal-interstitial cells
Animals Mature (>60 days old) and immature (21-23 days old) Sprague-Dawley female rats from Johnson Laboratories, Inc. (Bridgeview, IL), were killed within 1-2 days after delivery. Reagents and hormones McCoy's 5a medium (modified, without serum), penicillin (10,000 IU/ml), streptomycin (10,000 Mg/ml solution), L-glutamine (29.2 mg/ml), trypan blue stain (0.4%, wt/vol), BSA, and DNase (bovine pancreas; 2,100 U/mg) were obtained from Gibco (Grand Island, NY). Collagenase (Clostridium histolyticum; CLS type I; 144 U/mg) was obtained from Worthington Biochemical Corp. (Freehold, NJ). Highly purified insulin-like growth factor-I (IGF-I), revealing no detectable impurities on polyacrylamide gel electrophoresis, was prepared from Conn fraction IV of human plasma and generously provided by Dr. Judson J. Van Wyk, University of North Carolina (Chapel Hill, NC). Bovine insulin was obtained from Sigma Chemical Co. (St. Louis, MO). Highly purified platelet-derived TGF/31 of human origin was generously provided by Dr. Michael B. Sporn, Laboratory of Chemoprevention, NCI, NIH (Bethesda, MD). Highly purified TGF/31 precursor and TGF/32 of human origin were generously provided by Dr. David Twardzik, Oncogen (Seattle, WA). Highly purified inhibin (a//3A) of porcine follicular fluid origin was generously provided by Drs. Wylie W. Vale and Joan M. Vaughan, The Salk Institute (San Diego, CA). Human high density lipoprotein (HDL) was generously provided by Dr. Michael Henson, University of Maryland (Baltimore, MD). Highly purified activin-A (/3A//3A) of porcine origin was generously provided by Dr. Nicholas C. Ling, The Salk Institute. Highly purified hCG (CR-121; 13,450 IU/mg) was generously provided by Dr. R. E. Canfield through the Center for Population Research, NICHHD, NIH (Bethesda, MD). [7-N-3H]Pregnenolone (15 Ci/mmol) and [9,ll-N-3H]androsterone (42 Ci/mmol) were obtained from New England Nuclear (Boston, MA). [3H]Pregnenolone (3a-hydroxy-5apregnan-20-one), not commercially available, was biologically and enzymatically synthesized by us as follows. Granulosa cells from immature intact rats were cultured for 72 h in the presence of [3H]pregnenolone (5 X 106cpm/3 x 105 cells). Labeled steroids in the medium were then extracted, pooled, and subjected to HPLC (see below). Fractions containing the major metabolite, 5a-pregnane-3a,20a-diol (5a-pregnanediol), were collected, evaporated to dryness, and enzymatically oxidized (66% conversion) to pregnenolone using a bacterial 17/3-hydroxysteroid dehydrogenase (HSD) containing minor 20aHSD activity (19). In addition to the enzyme, the reaction mixture (1 ml) contained 100 mM 2-[iV-cyclohexylamino]ethanesulfonic acid (pH 9), 10% methanol, 0.2 mM NAD, and 0.05% BSA. The reaction proceeded for 60 min at room temperature before ethyl
Whole ovarian dispersates from immature rats were prepared as previously described (20). Briefly, ovaries were dissected free of their bursae, divided into four to six pieces, washed (twice) with 10 ml McCoy's 5a medium, and subjected to enzymatic dispersion for 45 min at 37 C using 0.4% (wt/vol) collagenase, 0.001% (wt/vol) DNase, and 0.1% (wt/vol) BSA. In the course of this incubation period, the ovaries were dissociated into a cell suspension by repeated pipetting every 30 min with a graded series of micropipettes (id, 0.5-1.0 mm). At the end of the incubation period, the cells were collected by centrifugation at 250 x g for 5 min, washed (three times) with McCoy's 5a medium, and then resuspended into a known volume of the same. The composition of whole ovarian dispersates at the time of plating was previously discussed (21), granulosa cells constituing 80% and thecal-interstitial cells 10-20% of the whole. Purification of ovarian thecal-interstitial cells was carried out by density gradient centrifugation, as previously described (22). Cell viability, as assessed by trypan blue dye exclusion, was consistently greater than 85%. Tissue culture procedures Whole ovarian dispersates (3 x 105 viable cells/culture) and highly enriched thecal-interstitial cells (5 X 104 viable cells/ culture) were inoculated onto tissue culture dishes (35 x 10 mm; Falcon Plastics, Oxnard, CA) containing 1 ml McCoy's 5a medium (modified without serum) supplemented with L-glutamine (2 mM), penicillin (100 U/ml), and streptomycin sulfate (100 fig/ml). Cell cultures were maintained for 96 h (unless indicated otherwise) at 37 C under a water-saturated atmosphere of 5% CO2and 95% air. The various experimental agents (diluted in sterile culture medium) were applied in 50-^1 aliquots. HPLC Ether- or ethyl acetate-extracted medium steroids were fractionated on a Lichrosorb Diol column (EM Reagents, Cherry Hill, NJ) calibrated with 35 steroids, as previously described (23). Radiolabeled metabolites were detected in-line with a radioactivity detector (Flo-One Beta, Radiomatic Instruments, Tampa, FL). RIA Androsterone (3a-hydroxy-5«-androstane-17-one), the main androgenic steroid identified in culture medium by HPLC, was quantified by RIA as previously described (23). The antiserum (AO-1) used was raised against androsterone- 16-BSA, generously provided by Dr. D. T. Armstrong, University of Western Ontario (London, Canada). cAMP RIA kit was purchased from Biochemical Technologies, Inc. (Stoughton, MA).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 13 November 2015. at 03:13 For personal use only. No other uses without permission. . All rights reserved.
TGF/31 INHIBITS OVARIAN ANDROGEN PRODUCTION
2806
Endo • 1990 Vol 127 • No 6
-28s
Northern blot analysis Polyadenylated [poly(A)+] RNA was isolated by the guanidinium hydrochloride procedure, fractionated on a 1% agaroseformaldehyde gel, transferred to a nylon membrane (Hybond, Amersham, Arlington Heights, IL), and hybridized to a 32Plabeled simian TGF/31 cDNA probe, as previously described (24). This probe codes for the entire TGFjSl precursor polypeptide (391 amino acids), the mature TGF/31 portion of which consists of the 112 carboxy-terminal residues (24).
2.5 kb
-18s
Immunohistochemistry The rabbit TGF/31 polyclonal antibody [anti-LC-(l-30)] was generously provided by Kathleen C. Flanders, Laboratory of Chemoprevention, NCI, NIH. Sections (5 /xm) of Bouin-fixed dehydrated paraffin-embedded ovaries were deparaffinized, blocked, and incubated with 0.1 mg/ml anti-TGF/3 (1:60 dilution) at room temperature overnight. Sections were washed extensively and then incubated with biotinylated goat antirabbit immunoglobulin G (IgG). Antibody localization was achieved using a rabbit IgG ABC-Vectastain kit (Burlingame, CA) with 3,3'-diaminobenzidine (Sigma Chemical Co.) as the substrate. Slides were counterstained in Gill's hematoxylin (Sigma Chemical Co.). Control sections were stained in parallel with normal rabbit IgG or buffer alone. Neither control exhibited detectable staining.
M FIG. 1. Ovarian TGF/31 gene expression in immature (I) and mature (M) intact rats. Ten micrograms of poly(A)+ from the ovaries of immature and mature rats were employed to generate the Northern blot, as described in Materials and Methods. 18S and 28S = Ribosomal RNA.
Data analysis Data are presented as the mean ± SE of three or more separate determinations. RIA data analysis, dose-response curve fitting, and determination of the median inhibitory (EC50) dose were carried out on an IBM AT PC using CURVE FIT software based on the four-parameter logistic equation designed to fit the results to a sigmoidal function curve. Statistical significance was determined by Student's t test or analysis of variance, as indicated.
Results Ovarian TGFfil gene expression and cellular localization To determine if the ovary is a site of TGF/31 gene expression, poly(A)+ RNA extracted from the ovaries of immature and mature rats was analyzed (Fig. 1). Northern blot analysis revealed a single major transcript with an apparent size of 2.5 kilobase corresponding to the TGF/31 precursor (24). This transcript appears identical to that observed in a variety of species including the rat (1) and establishes the ovary as a site of TGF/31 gene expression. Given that immunoreactive TGF/31 has been localized to both the thecal-interstitial (2, 3) and granulosa (16) cells, we have also evaluated its cellular localization in situ in the immature ovary by immunohistochemical staining. As shown (Fig. 2), the TGF/31 protein, as in the mice (15), was exclusively localized to the thecal-interstitial (interfollicular) compartment. No staining was
FIG. 2. Ovarian TGF/81: cellular localization. Sections were processed as described in Materials and Methods. Interstitial staining was absent in control sections, in which antibody has been replaced with nonimmune rabbit serum or buffer (not shown).
detected when employing nonimmune rabbit serum or buffer (not shown). TGFfil inhibits gonadotropin-supported androsterone accumulation by cultured whole ovarian dispersates To examine the role of TGF/31 in thecal-interstitial cell differentiation, whole ovarian dispersates were cultured in the absence or presence of hCG (1 ng/ml), with or without increasing concentrations (0.01-10 ng/ml) of TGF/31 (Fig. 3). Basal androsterone accumulation was relatively low (15 ± 2ng/ml) and remained unaffected after treatment with TGF/31 (10 ng/ml) by itself. As expected, treatment with hCG substantially increased androsterone accumulation (5.8-fold). However, concurrent treatment with increasing concentrations of TGF01
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 13 November 2015. at 03:13 For personal use only. No other uses without permission. . All rights reserved.
TGF/31 INHIBITS OVARIAN ANDROGEN PRODUCTION
2807
400 .,
Z
o
80
H 60
200 . 3 O U
z o x w H
40
UJ
z o
(T UJ h
en
20
O X
o
Q
(E
Z
Q
Z
