Comparative regulation of inhibin, activin and human chorionic gonadotropin production by placental trophoblast cells in culture

Placenta(1994), 15, 803-818

Comparative Regulation of Inhibin, Activin and Human Chorionic Gonadotropin Production by Placental Trophoblast Cells In Culture j. KEELAN a, Y. S O N G & J. T. FRANCE Research Centrefor ReproductiveMedicine, Department of Obstetrics and Gynaecology, National Women's Hospital, Claude Rd, Epsom, Auckland, New Zealand a To whom correspondenceshould be addressed Paper accepted8. 6.1994

SUMMARY

In the present study, we investigated the roles of cyclic adenosine monophosphate (cAMP), intracellular calcium, glucocorticoids, protein kinase-C and gonadotrophin-releasing hormone (GnRH) in regulating human chorionic gonadotrophin (hCG), inhibin and activin production in cultured human term placental trophoblast cells. Inhibin and hCG were measured in conditioned media by radioimmunoassay, while putative forms of inhibin and activin were characterized by western blotting using affinity-purified antisera directed against the inhibin o~- and flAsubunits. Inhibin and hCG secretion were stimulated by dexamethasone (0.2 tZM), GnRH (5-25tZM), calcium ionophore A23187 (0.2-1 lXM), phorbol-12-myristate-13-acetate (22 nM) and epinephrine (1 ~M), with increasing response over successive 24-h treatment periods. Two molecules M r - 3 0 and 32 kDa appeared to be the predominant dimeric forms of inhibin secreted by the cells, while 26 kDa activin was present in excess over inhibin. Large amounts of 40-44 kDa protein were detected by the or-directed antisera only, which may be a form of the inhibin asubunit precursor protein. Secretion of activin was responsive to phorbol estermediated stimulation but not to the presence of GnRH or elevated cAMP concentrations. The divergence in maternal serum inhibin and hCG concentrations during late pregnancy remains unexplained by thesefindings.

INTRODUCTION Inhibin, originally described as a gonadal non-steroidal hormone regulating follicle stimulating hormone (FSH) production by the pituitary (deJong, 1988; Ying, 1989), is now known to be produced by a variety of different tissues including the human placenta (Bandivdekar et al, 1981; McLachlan et al, 1986). Inhibin or- and 13A-subunit messenger RNAs have been 0143-4004/94/080803 + 16 $08.00/0

9 1994W. B. Saunders CompanyLtd

804

Placenta (1994), VoL 15

detected in the human placental villous trophoblast in situ (Petraglia et al, 1992), while immunohistochemicalstudies have localized the inhibin subunits (Petraglia et al, 1991, 1992) and the 13A homodimer activin-A (Rabinovici et al, 1992) to placental cytotrophoblast and syncytiotrophoblast cells. The elevated levels ofimmunoreactive (Abe et al, 1990; Tabei et al, 1991) and bioactive (Qu et al, 1991; Qu and Thomas, 1992) inhibin present in the maternal circulation and cord blood during preganancy are believed to arise from the placenta. However, the bioactive/immunoactive potency of circulating inhibin in pregnancy is much lower than in the non-pregnant state (Qu et al, 199 I), raising questions as to the structure and relative proportions ofinhibin/activin molecules secreted by the placenta. The cellular mechanisms regulating the placental synthesis of inhibin and activin are incompletely understood. In cultured trophoblast cells from first trimester placentae, Rabinovici et al (1992) failed to detect any significant alteration to activin immunohistochemical staining following exposure of the cells to increased cyclic adenosine monophosphate (cAMP) concentrations, dexamethasone, gonadotrophin-releasing hormone (GnRH), transforming growth factor-J3 (TGF-[3), interleukin-1 (IL-1), inhibin-A or activin-A. On the other hand, Petraglia et al (1987) have presented evidence supporting a role for cAMP in mediating the hormonal regulation ofinhibin o~-subunit production in term placental cell cultures, observations recently confirmed by Qu, Brulet and Thomas (1992). It has also been proposed that a short closed-loop feedback system might operate between inhibin, GnRH and human chorionic gonadotrophin (hCG), where hCG stimulates inhibin production which in turn, possibly via suppression of local GnRH production, inhibits hCG secretion (Petraglia et al, 1987, 1989). Concentrations of hCG in maternal serum during pregnancy remain constant throughout the third trimester, while those of inhibin increase towards term, suggesting that different factors regulate the production of these hormones at this time. Circulating levels of activin in maternal serum during pregnancy have not yet been reported. Numerous in vitro studies have identified hCG synthesis to be responsive to modulation by a wide range of factors. These include cAMP (Strauss et al, 1992), dexamethasone (Ringler, Kallen and Strauss, 1989), progesterone and oestradiol (Iwashita et al, 1989), intracellular calcium (BElisle et al, 1989), GnRH (Iwashita, Evans and Catt, 1986), protein kinase-C (Iwashita et al, 1992), TGF-13 (Matsuzaki et al, 1992), tumour necrosis factor-o~ (Ohashi et al, 1992; Ying et al, 1992), and epidermal growth factor (EGF) (Morrish et al, 1987). In an endeavour to identi~ agents that may differentially regulate placental inhibin and hCG production, we have investigated the effects of some known modulators of hCG production on placental inhibin oL- and [3A-subunit production by term placental trophoblast cells in culture. Furthermore, using immunoblotting techniques, we have studied the subunit composition of inhibin molecules secreted by cultured human trophoblastic cells under basal and stimulated conditions.

MATERIALS AND METHODS

Reagents Hank's balanced salt solution (HBSS), Ham's F12 medium (H), Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), BME vitamin solution, sodium selenite, insulin and EGF were purchased from Gibco/BRL Life Technologies Ltd, Auckland, NZ. Forskolin, dexamethasone, GnRH, calcium ionophore A23187, epinephrine, phorbol 12-myristate-13-acetate (PMA), bovine serum albumin (BSA), ovalbumin

Keelan et al: Regulation of lnhibin, Aaivin and hCG Production

805

(Ova), phenylmethylsulphonyl fluoride (PMSF), and antibiotics were obtained from Sigma Chemical Co., St Louis, MO, USA. Transferrin, dibutyryl cAMP (Bt2cAMP), dispase (grade II), and DNAse-I were supplied by Boehringer Mannheim NZ Ltd, Auckland, NZ. Pharmacia (Uppsala, Sweden) was the supplier of Percoll, density marker beads, protein-A Sepharose and Sepharose 6B. Recombinant human (rh-) inhibin-A was donated by Genetech Inc., San Francisco, CA, USA, while rh-activin-A was a gift from Dr Y. Eto, Ajinomoto Co. Inc., Kawasaki, Japan. FSH from human pituitaries, was supplied by Dr A. Renwick, University of Auckland. Prior to use, the FSH preparation was dialysed in isotonic saline to remove potentially cytotoxic factors. The immunological potency of the dialysed material was determined by immunochemiluminometric assay (Amerlite: Kodak Clinical Diagnostics, Sidney, Australia) standardized against the second FSH IRP 78/549. Antisera (polyclonal) to pan-cytokeratin ~ 1anti-chymotrypsin were purchased from BioGenex Laboratories, San Ramon, CA, USA. Trophoblast cell culture Human term placentae were obtained after normal vaginal delivery with informed consent of the mother in accordance with the established guidelines of the Auckland Area Health Board Ethical Committee. Placental trophoblasts were isolated by a modification of the method of Kliman et al (1986). About 60 g ofviUous tissue was dissected free of connective tissue, washed with saline, then cut into 1-2 mm pieces. The tissue was then digested in a solution of 625 mg dispase-II in 250ml of HBSS at 37~ for 45 min with gentle stirring. After this time, 10mg DNAse-I in 10ml HBSS was added and digestion allowed to proceed for a further 15 rain. Cellular clumps were allowed to settle for 1 min, then the supernatant solution containing liberated cells in suspension was decanted off and centrifuged for 10min at 1000g. The pelleted cells were reconstituted in 25 ml H / D M E M + media, which is a 1:1 mixture of Ham's F12:Dulbecco's modified Eagle's medium containing 5 mg/l insulin, 10 mg/l transferrin, 10 p.g/1 EGF, 0.2 nmol/l sodium selenite, 1 per cent BME vitamin solution, 100 mg/l streptomycin sulphate and 100 000 U/1 penicillin (Ralph, Lee and Thorburn, 1989) supplemented with 10 per cent FBS. The cells were dispersed by aspirating through a pasteur pipette, passed through two layers of cheesecloth to remove large clumps and repelleted by centrifugation as before. They were resuspended in HBSS, and centrifuged at 1000 g for 20 min over a continuous Percoll gradient (45 per cent Percoll in HBSS, centrifuged for 30min at 37000g). Cells in the re#on corresponding to a density of 1.049-1.062 g/l, as judged by the position of the density marker beads run in a parallel Percoll gradient tube, were harvested with a pasteur pipette (Kliman et al, 1986). These cells, which were predominantly single or small clumps of trophoblasts, were recovered by centrifugation and resuspended in plating medium ( H / D M E M + supplemented with 10 per cent FBS) at a concentration of 2 x 106 viable cells/ml. Viability was checked using trypan blue exclusion. The cell suspension was plated out into 24-well Falcon multidishcs (Becton Dickinson Labware, Lincoln Park, NJ) at 1 x 106 viable cells per well in 0.5 ml media and cultured at 37~ in a humid atmosphere of 5 per cent CO2/95 per cent air. The next day the plating media was replaced with H / D M E M + containing 1 per cent FBS. For immunohistochemical studies (n = 3), cells were plated on coverslips placed in the culture medium and assessed after 24- and 96-h culture according to established procedures (Douglas and King, 1989). Experiments were conducted over days 5-10 of culture. The cells were exposed to test substances over three consecutive periods of either 24 or 48 h. Once removed from the cells, 20 p.1 of a preservative/protease-inhibitor solution (0.5 mg/ml PMSF, 5 mg/ml

806

Placenta (1994), Vol. 15

sodium azide, 187.5 mg/ml EDTA, dissolved in 50 per cent methanol/50 per cent 25 mM phosphate-buffered saline (PBS) pH 7) was added per ml of media and the samples were stored in polypropylene tubes at -20~ prior to assay. Hormones and secretagogues were prepared as concentrated stock solutions and added in small amounts to the culture wells to achieve the desired final concentration. Stock solutions of forskolin, dexamethasone and epinephrine were prepared in 50-66 per cent ethanol. A23187 was dissolved in dimethylsulphoxide (DMSO) or ethanol. Other test substances were dissolved in either saline or media. The final concentration of ethanol or DMSO in the media never exceeded 1 per cent. Three replicate culture wells were used for each test dose, and each experiment was repeated at least three times. The results from different treatment groups and their respective controls (wells treated with solvent only) were tested for statistical differences by unpaired two-tailed Student's t-test or analysis of variance when the results of multiple experiments were being compared. A P value < 0.05 was considered to be significant. Results are given as mean _+ standard error of the mean (s.e.) unless otherwise stated. Hormone immunoassays hCG determinations were performed using the Amerlite hCG-60 assay (Kodak). The assay has a between-batch coefficient of variation (CV) of 6.3 per cent at 195 IU/I (n = 23). Reagents and assay methodologies for the inhibin radioimmunoassay (RIA) were obtained from Dr G. Bialy (NICHHD, Bethesda, MD, USA), and were produced by Dr D. M. Robertson, Monash University, Melbourne, Australia. Inhibin iodination and purification were carried out as recommended, except rh-inhibin (Genentech; lot 13140-90/R) was radioiodinated rather than the purified bovine inhibin supplied. Antibody-bound inhibin tracer was precipitated with 1:200 goat anti-rabbit IgG (Sigma) in 6 per cent (w/v) polyethylene glycol 6000 containing 0.1 per cent (v/v) triton X-100. The sensitivity of the inhibin RIA, using the recommended human inhibin reference standard rhINH-R-90/1, was approximately 100pg/ml (equivalent to 5.1 U/ml calibrated against WHO international reference preparation 86/90). The rhINH-R-90/1 preparation was eightfold less potent in the RIA than the rh-inhibin from Genentech. The within-assay precision was 5.9 per cent over the range 200-5000 pg/ml (n = 200), while between-assay precision was 11 and 14 per cent at 437pg/ml and 3035pg/ml, respectively (n = 14). The assay is known to cross-react with free OL-subunit molecules (Robertson et al, 1989), but shows minimal reactivity to activin or TGF-~3 (Robertson et al, 1988). Antisera used in the immunoblotfing experiments . Antibodies were raised in rabbits against synthetic peptides corresponding to the t~1-26 sequence of porcine inhibin and the [3A81-107 sequence common to all known species. The peptides were synthesized on a model 430 Applied Biosystems automated peptide synthesizer, using the BOC-chemistry profiles. BSA and Ova conjugates of both peptides were prepared and used for inoculation. The anti-oL-subunit antisera was purified by affinity chromatography using the a l - 2 6 peptide coupled to Sepharose 6B. The recovery of anti-peptide activity varied between 25-50 per cent. Attempts to purify the [3-subunit directed antisera by a similar approach resulted in poor recoveries. Therefore this antisera was purified by incubation with BSA- and Ova-Sepharose 6B to remove antibodies specific to the two carrier proteins. The IgG from the serum was then recovered by protein-A chromatography. This procedure greatly reduced the high non-specific background that was otherwise observed with the unpurified antisera. The anti-a-subunit antisera recog-

Keelan et al: Regulation of lnhibin, Activin and hCG Production

807

nized rh-inhibin and reduced ~-subunit molecules on western blotting, while the anti-13Asubunit antisera detected intact inhibin-A, activin-A and reduced [3A-subunit. Antisera neutralized with solid-phase peptide antigen were used to confirm the specificity of the immunological response. Electrophoresis and western blotting

SDS-PAGE on 12-14 per cent acrylamide gels and western blotting onto nitrocellulose membrane (BioRad Laboratories Ltd, Auckland, NZ) was carried out using a Mini Protean II/Trans-blot system (BioRad) according to the supplier's instructions. Trophoblast-conditioned media (TCM) samples (5-11 ml) were purified and concentrated before electrophoresis by passage through 0.4ml Matrex Red-A gel (Amicon Corp., Lexington, MA, USA), a dye-affinity matrix which binds inhibin and rleated proteins (Jansen et al, 1981). Bound material was eluted with 1 ml 4 M urea/1.2 M KC1, then diluted 1:4 with 4 M urea to reduce the salt concentration and reconcentrated by ultrafiltration on 10 000 Mr cut-off Microsep cartridges (Filtron Technology Corp., Northborough, MA, USA) pre-blocked with 10 per cent glycerol. In order to standardize the loading of MRA-extracted proteins for electrophoresis in experiments comparing the effects of different treatments or duration of culture, lz5I-labelled inhibin was added to the TCM samples before purification to assess individual recoveries. The volume of the concentrates was adjusted with 4 M urea to obtain an equal number of counts/~l. After electrophoresis and transfer, the nitrocellulose membranes were blocked with 5 per cent (w/v) non-fat dried milk (BLOTTO) in PBS pH 7.2. Affinity purified anti-porcine ~1-26 antisera (1:500) or protein-A-purified anti[3A81-107 IgG (1:480), diluted in PBS containing 0.5 per cent (v/v) Tween-20 and 5 per cent BLOTTO, were used to probe the blots. Following a second incubation step with anti-rabbit IgG-HRPO conjugate (Tago Inc., Burlinghame, CA, USA), or streptavidinHRPO complex (Amersham) to allow detection of the biotinylated molecular weight markers (BioRad), immunoreactive protein bands were visualized on X-ray film using an enhanced chemiluminescence method (ECL, Amersham Australia Pty Ltd, Sydney, Australia).

RESULTS P r o d u c t i o n of immunoreactive-inhibin (ir-inhibin) and hCG during the course of cell culture

The placental trophoblast purification method used in this study is based on the widely used method of Kliman et al, 1986. The cells obtained by this method: (i) had uniform morphology under light microscopy; (ii) exhibited morphological changes (i.e. aggregation and syncytia formation) in culture similar to those that have been described for cytotrophoblasts (Ringler and Strauss, 1990); and (iii) were capable of secreting hCG in vitro. In culture preparations from three placentae, immunohistochemical staining of the cells 24 h after plating, using a commercial antisera specific for pan cytokeratin, revealed that > 95 per cent of cells were of epithelial origin, staining positive. In contrast, < 5 per cent of cells stained with an ~l-anti-chymotrypsin antibody, a marker for cells of myeloid lineage. Based on the above evidence we have assumed that our cultures consist principally of trophoblast cells. The level of production of ir-inhibin and hCG varied markedly between different placental cultures, though the pattern of secretion through the period of culture was

808

Placenta (1994), Vol. 15

consistent. Typically, secretion of both hormones was low until day 3 - 4 of culture then rose to maximum levels between days 6-10 [Figure l(a) and Co)], declining thereafter. The increase in h C G and ir-inhibin production was preceded by the appearance of multinucleated syncytiotrophoblast-like cells, which became the dominant cellular morphology by day 2 - 3 of culture. 500

5

400

~

**o0~176176

4 2

..~....."""'

300

r

3

r2 200

2~

100

1.~

1

3

5 7 9 Day of culture Figure 1. Production of immunoreactive-inhibin(-'"O-'") and human chorionic gonadotrophin (--O--) in culture. The mean + SEM24-h productionrates of fiveseparateplacentalculturesare shown. Levelsofir-inhibin on day 1 werecorrectedfor exogenousinhibinpresent in the media due to the high fetalcalfserum (10 per cent).

T h e actions o f cAMP on ir-inhibin and h C G p r o d u c t i o n Production of hCG (Strauss et al, 1992) and ir-inhibin (Petraglia et al, 1987; Qu, Brulet and Thomas, 1992) by cultured placental cells have previously been shown to be regulated by intracellular cAMP. T o confirm these observations.and establish the validity of our culture system, we first investigated the effects of Bt2cAMP and forskolin on h C G and inhibin secretion. Maximal stimulation of secretion of the two hormones was achieved with Bt2cAMP concentrations between 50-250 V,M and with forskolin concentrations between 0.33-6.66 O~M,varying between different culture preparations (results not shown). Significant effects were recorded during the first 48 h of treatment, becoming greater over further subsequent treatment periods. A similar pattern of response was seen with epinephrine (1 p~M), a physiological stimulator of cAMP production (Figure 2). Since the actions of F S H in ovarian tissue are known to be mediated primarily by elevations in intracellular cAMP concentrations, and include the stimulation of ovarian and testicular inhibin production, we examined the response of our trophoblast cultures to FSH. In four separate experiments we found no effect of up to 109 IU/1 F S H on either basal, G n R H - or foskolin-stimulated inhibin production. Regulation of ir-inhibin and hCG production by GnRH and calcium ionophore A23187 Evidence has been presented that G n R H is involved in the regulation of placental inhibin production (Petraglia et al, 1987, 1989). T o confirm this hypothesis, we challenged the trophoblast cultures with a range of G n R H concentrations (1-25 ~M). Although some cell culture preparations responded poorly to G n R H with only minor increases in inhibin and h C G secretion, six out of 10 preparations gave a significant and dose-dependent response.

Keelan et al: Regulation of lnhibin, Activin and hCG Produaion

809

1000 r800

600 =

8 400 200

NS

*

II~

24

48 72 Hours of t r e a t m e n t Figure 2. Response of immunoreactive-inhibin ([]) and human chorionic gonadotrophin (11) production to 1 ~M epinephrine over days 5 - 7 of culture. Media was removed every 24 h from day 4 and replaced with fresh media containing the secretogogue or vehicle (50 per cent ethanol in saline). * P < 0.05; ** P < 0.01; *** P < 0.001; NS, not significant. Similar results were obtained in two other experimental cultures.

Maximum stimulation of the secretion of both hormones occurred at GnRH concentrations of 25 tXM, the highest dose tested. Cells repeatedly exposed to GnRH produced progressively increasing amounts of ir-inhibin and hCG (Figure 3). 10000

=

1000

100 24

48 72 Hours of t r e a t m e n t Figure3. Effect of 25 ~M gonadotrophin-releasing hormone (GnRH) on immunoreactive-inhibin ([2) and human chorionic gonadotrophin (ll) production over day 5-7 of culture. Media was removed every 24h from day 4 and replaced with fresh media containing GnRH or saline vehicle (control). Note the log scale on the y-axis ( per cent of control). * P < 0.05; ** P < 0.01; *** P < 0.001; NS, not significant. The results shown are representative of three similar experiments.

The effect of changes in intracellular calcium on inhibin and hCG secretion by the trophoblast cultures was investigated by exposing the cells to the calcium ionophore A23187 over three consecutive days in culture. Again the responses were variable, with A23187 having toxic effects in the majority of culture experiments. However, four out of 11 culture preparations responded with an increase in hCG and ir-inhibin production at doses of A23187 ranging from 0.2-1.0 p~M; higher doses resulted in reduced hormone production rates which we attribute to the toxicity of the agent. The stimulation of hCG and inhibin production elicited by the calcium ionophore increased in magnitude over successive days of treatment (Figure 4). In three separate trophoblast preparations that responded

Placenta (1994), Vol. 15

810

to the 1 WM A23187 dose, hCG concentrations showed a mean increase of 759 _ 449 per cent after 72 h exposure, while ir-inhibin levels increased to 154 _ 10.2 per cent relative to controls. In one of these preparations, there was a 16-fold increase in hCG concentrations in the media, much greater than the 1.3-fold rise observed in ir-inhibin concentrations.

***

2500

]

2O00

8

15oo 1000 500

Ns i 0

~ 48 72 H o u r s of t r e a t m e n t Figure 4. Effects of calcium ionophore A23187 (1 ~M) on immunoreactive-inhibin ([3) and human chorionic gonadotrophin (ll) production over days 5 - 7 of culture. Media was removed every 24 h from day 4 and replaced with fresh media containing the ionophore or dimethyl sulphoxide vehicle (control). * P < 0.05; ** P < 0.01; 9** P < 0.001; NS, not significant. The results shown are representative of three similar experiments. 24

L

Stimulation of ir-inhibin and hCG production by dexamethasone The glucocorticoid agonist dexamethasone stimulated h C G production in all experiments (n = 8), and in six out of eight placental preparations also stimulated inhibin production. In a dose-response study, the results of which are shown in Figure 5, hCG secretion was significantly increased by dexamethasone concentrations over the range 0.04-1 ~M after 48-72 h treatment. Inhibin production in the cultures, on the other hand, was considerably less sensitive to dexamethasone stimulation, showing a significant response (P < 0.05) to 0.2 p~M dexamethasone only after the third day of treatment, with no further increase at the higher dose.

The role of protein kinase-C in regulating inhibin and h C G secretion The protein kinase-C activator, PMA, had sometimes variable effects in different placental cultures, but in five of nine cell preparations at the dose employed (22 nM) it markedly increased inhibin secretion (mean + s.c. per cent increase relative to control: 226 + 37). The secretion of hCG was unaffected by the PMA treatment in six of the culture experiments but increased in the other three (mean + s.e. per cent increase: 178 + 25). The level of response to PMA did not show any consistent changes over successive treatment periods. We were unable to demonstrate any consistent and significant effects of PMA on either G n R H - or A23187-stimulated ir-inhibin and h C G production in our cultures (results not shown).

Western blotting data The ir-inhibin measured in the culture media in the above experiments was determined by RIA with an antiserum possessing specificity for the ot-subunit (Robertson et al, 1989). It

Keelan et al: Regulation of lnhibin, Activin and hCG Production 250

**

- (a)

2000

200

811 (b)

***

1500

~150

8

~I000 I00 500

50

control 0

i

i

i

10 40 200 Dexamethasone(nM)

i

I

1000

[] i

0

I

i

I

I

i

I

i0 40 200 Dexaraethasone (nM)

I

I

i000

Figure 5. Dexamethasone dose-response study. The response of (a) immunoreactive-inhibin and (b) human chorionic gonadotrophin production was studied over day 5 - 7 of culture. Media was removed every 24h from day 4 and replaced with fresh media containing dexamethasone or vehicle (50 per cent ethanol in saline). * P < 0.05; 9 * P < 0.01; *** P < 0.001; NS, not significant. The results shown are representative of three similar experiments. ([2), 72h; (O), 48h, (9 24h.

also is known to have high cross-reactivity with the inhibin pro-otC protein. To better characterize the 'inhibin' proteins secreted by the cells, T C M was semi-purified, concentrated and then subjected to gel electrophoresis and immunoblotting using antibodies specific for the inhibin ~t- and 13A-subunits. Unstimulated T C M samples from days 1-9 of culture, blotted and probed with anti-or subunit antibody [Figure 6(a)], contained an immunoreactive band with an estimated Mr of 40-42 kDa which was abundant during the first 5 days of culture. The molecular weight of this protein is similar to that predicted ( - 4 0 - 4 5 kDa) for the immature inhibin ot-subunit precursor otNotC (Sugino et al, 1992). A much fainter band of 31 kDa was evident in the day 5 T C M sample in this blot which migrated on the gel identically to the inhibin standard. This protein was also detected by the 13A-directed antibody in samples from day 4-9 of culture, coincident with peak levels of ir-inhibin as determined by RIA, and thus is likely to be dimeric inhibin. Media from the same experiment was found to contain a - 2 6 k D a protein reactive with the anti-i3A antibody [Figure 6(b)]. The migration position of this protein on the gel was identical to that of the activin standard (26 kDa). There were no clear fluctuations in concentrations of the putative 'activin' protein in the media throughout the culture period. Higher molecular weight proteins (> 50 kDa) of unknown composition detected in this blot were artifacts due to interference from proteins present in the FCS. Conditioned media from another culture experiment, electrophoresed under reducing conditions, contained a 44kDa protein band recognized by the anti od-26 antibody [Figure 7(a)] while a band at 14 kDa interacted with the anti 13A 81-107 antibody [Figure 7(b)]. The levels of the 44 kDa protein in the media increased when the cells were treated with forskolin (1.66 p~M)and with PMA (22 nM) [Figure 7(a)], consistent with the ir-inhibin RIA measurements made on the samples prior to electrophoresis. No protein bands were detectable with a Mr around that expected of the mature ot-subunit (21 kDa). Media concentrations of the 14kDa protein detected by the anti-[3A subunit antibody were increased by PMA treatment but not by forskolin [Figure 7(b)]. The electrophoretic migration of this 14 kDa protein is consistent with that of the mature I3A inhibin subunit after reduction. In a subsequent experiment, electrophoresis and immunoblotting of non-reduced T C M

Placenta (1994), VoL 15

812 (a)

(b) Day of culture 3 5 7

Mr

9

71k

71k

58k

58k

40k

Day of culture 3 5 7

Mr

40k

~

32k

32k

26k

26k ~ a n t i - a 1-26

~

~

~

~

anti-~ 81-107

Figure 6. Western blotting analysis of the time-course of inhibin ct- and [3A-immunoreactivity in trophoblastconditioned media (TCM) collected through 9 days of culture. After day 1 of culture, media from one 24-well plate was removed, pooled and replaced every 48 h thereafter. After dye affinity-extraction and concentration TCM samples were electrophoresed under non-reducing conditions, transferred to nitrocellulose membrane, then probed with affinity-purified (a) anti-~xl-26 antisera or Co) anti-13A81-107 IgG. Media from day 1 contained 10 per cent fetal calf serum (FCS), whereas 1 per cent FCS was used in the remainder of the experiment. The bands visible in the blots with Mr >50 kDa were also present in fresh media, and hence do not represent immunoreactive proteins secreted by the trophoblast cells. Levels of immunoreactive inhibin in the media prior to extraction were 0.4, 1.18, 4.04, 2.12 and 4.86 ng/ml on days 1, 3, 5, 7 and 9, respectively.

samples from cultures challenged with forskolin, GnRH or PMA again revealed the presence of 13A-subunit-immunoreactive proteins of molecular weights 26, 30 and 3233 kDa, consistent with the presence of mature dimeric activin and inhibin (Figure 8). The appearance of two 'inhibin' bands, molecular weight 30 and 32 kDa, are likely to be the result of different degrees of a-subunit glycosylation as observed in rh-inhibin preparations (Tierney et al, 1990). While levels of 'inhibin' in the blot were clearly increased by treatment with forskolin, neither PMA nor GnRH had discernible effects on levels of 'inhibin' immunoreactivity, although increased ir-inhibin concentrations were measured in the media by the RIA. In contrast, levels of 26 kDa 'activin' were markedly increased in response to PMA, but showed no detectable response to forskolin or GnRH. Reprobing the same blot with a-subunit-directed antisera (not shown) again revealed the presence of the - 4 2 kDa protein that was undetectable with the anti-13A antibody. There was no evidence in any experiment that the trophoblast cell cultures produced pro-aC.

DISCUSSION The cell isolation and culture techniques that have been used in this study are based on methods that have been well-characterized in the literature as producing relatively pure placental trophoblast cells (Kliman et al, 1986; Ringler and Strauss, 1990). We used dispase digestion rather than trypsin to liberate the cells, as in our hands this yielded cells with greater hormone production potential. Supplements, including insulin and EGF, were

813

Keelan et al: Regulation of Inhibin, Activin and hCG Produaion

(a)

(b) Inh

P

F

C

Inh

P

F

C

Mr

-

67

-44

-21

-

14

Figure 7. Westernblotting analysis of pooled trophoblast-conditionedmedia samples, from day 7-8 of culture.

Samples were dye affinity-extracted,electrophoresedunder reducing conditions, and probed with the (a) anti-c~ and 13A-subunit antisera. Inh, recombinant human-inhibin (rh-inhibin) (30 ng); P, PMA-treated (22nM); F, forskolin-treated(2 p~M);C, control (from unstimulated cultures). Bovineserum albumin was present in the rhinhibin standard preparation.The high molecularweightbands visiblein the blots were present in fresh, unused media and do not representimmunoreactiveproteinssecretedby the trophoblastcells. Levelsofimmunoreactiveinhibin in the samples before extraction (determined by radioimmunoassay)were 85, 224 and 179pg/ml in control, forskotin-and PMA-treated cultures, respectively. included in the culture media to promote basal production of the hormones. Although it is unlikely, we cannot exclude the possibility that these supplements had synergystic or obligatory effects on the responses to the various secretogogues observed in this study. Similar to the experience of other investigators (Yagel et al, 1989; Kato and Braunstein, 1990; Morrish, Bhardwaj and Paras, 1991), we experienced wide variations in hormone production rates and responses to certain stimuli. The results presented in this report were derived from cultures that had relatively high concentrations of hormones in the conditioned media (hCG > 5 0 IU/1/24 h on day 5-6). Results from these cultures were taken to be the most representative of a valid in vitro response. Hormone production from cultures with lower h C G secretion rates, however, responded to the various stimuli with similar trends. The production of h C G and ir-inhibin during the course of culture showed, in general, similar characteristics. Although syncytia formation began from day 2 - 3 of culture, hCG and ir-inhibin production did not begin to increase significandy until day 4-5. The low levels of hCG and ir-inhibin released by the cells on day 1 of culture, and the coincident rise in the production rates of both hormones subsequently, suggests that they are both products of the syncytiotrophoblast. Similar rates of inhibin production have been recently described by Qu, Ying and Thomas (1992). Western blotting analysis of T C M with the a-directed antisera revealed the presence of relatively large amounts of a monomeric 4 0 - 4 4 kDa immunoreactive protein which may be a form of the o~-subunit precursor molecule (i.e. ~xNoLC). However, our preliminary evi-

814

Placenta H994), Vol. 15

C

F

P

G

Inh 4

Mr 116k

967k 9

50k

9

32k 30k 26k

9 21k

Figure 8. Western blot of trophoblast-conditioned media samples electrophoresed under non-reducing conditions and probed with the anti-13A antisera. In this experiment, cells were cultured at 2.6 • 106 cells/ml to achieve high concentrations of inhibin accumulation in the media. C, Control (unstimulated cultures); F, forskolin-treated (20 p~M);P, PMA-treated (22 WM);G, gonadotrophin-releasing hormone (GnRH)-treated (25 p~M);lnh, recombinant human-inhibin (rh-inhibin) standard (35 ng). Levels of ir-inhibin (determined by radioimmunoassay) in the samples prior to extraction were 11.6, 55.5, 32.7 and 25.5 ng/ml in control, forskolin-, PMA- and GnRH-treated cultures, respectively. The - 5 0 kDa band seen in the inhibin standard lane is due to the presence of the higher molecular weight form ofinhibin as a minor component of the rh-inhibin preparation.

dence from analysis of T C M proteins eluted from the electrophoresis gel that suggests that this molecule possesses little immunoreactivity with the. antisera used in the inhibin RIA. The decline in levels of this protein in the media during the course of culture suggests that it may be stored within the trophoblast cells and released during the first few days of culture. The detection of this protein in T C M specifically by the a-subunit directed antisera suggests that it is a form of the ~x-subunit precursor protein containing the inhibin od-26 amino acid sequence. The lack of reactivity of this protein in the inhibin RIA explains why the observed pattern ofir-inhibin accumulation in T C M through culture does not show elevated levels in media from days 1-3. Our results confirm previous findings (Petraglia et at, 1987; Qu Brulet and Thomas, 1992) that ir-inhibin secretion by placental trophoblasts in culture is stimulated by agents that potentiate cAMP production or action. Immunoblotting studies demonstrated increased concentrations of dimeric inhibin in the media following treatment of the cells with forskolin, while 13A/activin levels were apparently unaffected. Increased 13A mRNA levels in response to elevated cAMP concentrations has been previously observed in the human placenta (Tanimoto et al, 1992), fetal adrenal (Voutilainen, Erama and Ritvos, 1991), and rat granulosa cells (LaPolt et al, 1990). However, a cAMP-dependent response element has yet to be located on the J3A-subunit gene and it has been suggested that these observations may be due to cAMP-induced alterations in I3A mRNA stability (Tanimoto et al, 1992). It is possible that the elevated levels of ot-subunit might have combined with existing static levels of 13A-subunit to form inhibin at the expense of activin production. While we were unable to detect a marked decline in the levels of activin immunoreactivity in this sample,

Keelan et al: Regulation of lnhibin, Activin and hCG Production

815

the blotting method is not sufficiently quantitative to reliably detect comparatively small changes in antigen concentration. Our failure to demonstrate an effect of FSH (which stimulates cAMP production in target tissues) on placental inhibin production, even at levels as high as those seen in the sera of postmenopausal women, is in contrast with recently published observations (Qu, Brulet and Thomas, 1992). However, these authors used supraphysiological levels of FSH (30 ng/ml, - 2 0 0 mlU/ml) to achieve a small but significant increase in ir-inhibin release. Since circulating FSH levels are low ( - 3 mlU/ml) during pregnancy, the physiological significance of this response is doubtful. A putative glucocorticoid response element has been identified on the rat inhibin ~subunit gene (Albiston et al, 1990). In our trophoblast cultures dexamethasone stimulated placental ir-inhibin secretion, though it was a much more potent stimulator of hCG production. The observed hCG response to dexamethazone was similar to that of Ringler, Kallen and Strauss (1989), who reported near maximal stimulation of hCG release at dexamethasone concentrations of 1 IXM. Our results also confirm the postulated role of GnRH in regulating placental inhibin production (Petraglia et al, 1987, 1989). The less potent stimulation of ir-inhibin production by GnRH and A23187, like that observed with cAMP modulators and dexamethasone, suggests that the potential dynamic range for placental inhibin production may be intrinsically lower than that of hCG. Our results do not allow us to assess the extent to which the stimulatory actions of GnRH on placental inhibin production are mediated by an initial rise in hCG concentrations. Protein kinase-C activation has been reported to lead to increased hCG production in JEG-3 cells (Ilekis and Benveniste, 1985) and in untransformed trophoblasts (Iwashita et al, 1992). In our experiments, although the effects ofphorbol esters on hCG and ir-inhibin production were inconsistent, a marked stimulatory effect was measured with respect to both hormones in a number of placental cultures (inhibin, six out of nine; hCG, three out of nine). Increased production of inhibin following treatment with phorbol esters has not been previously reported in other tissues to our knowledge. In our immunoblotting experiments, PMA clearly stimulated the production of the - 4 0 kDa putative inhibin ~x-subunit precursor protein, in agreement with inhibin RIA results. Levels of hCG were less frequently and acutely responsive to PMA-induced stimulation. This is in accordance with the data of Bdisle et al (1989) who found hCG production to be uncoupled to phospholipid hydrolysis and protein kinase-C activation. On the other hand, Iwashita and colleagues (1992) more recently reported the stimulation of basal (but not GnRH-stimulated) hCG release in first trimester placental tissue following protein kinase-C activation. Our observations could reflect a greater sensitivity of inhibin secretion to protein kinase-C stimulation than that of hCG. Possibly a PMA-induced increase in 13Aproduction leads to greater dimeric inhibin formation. Alternatively, other regulatory signals might be produced in response to PMA and these may modulate the production of the two hormones in different ways. Immunohistochemical studies of first trimester placental trophoblasts (Rabinovici et al, 1992), and analysis of placental extracts by RIA (Shintani et al, 1989) suggest that the placenta might be a source of activin. The present immunoblotting data is the first reported evidence that 'activin' is produced in culture by term placental trophoblast cells, although bioassay data is required for confirmation of the identity of the 26 kDa protein seen in our blots. The relatively constant rate of 'activin' production throughout culture is consistent with the 13A gene being expressed in a constitutive fashion. Production of 'activin' in our cultures was highly responsive to protein kinase-C stimulation. The stimulation of 13Asubunit synthesis by phorbol esters has been previously reported in other tissues (Scher et

Placenta (1994), Vol. 15

816

al, 1990; Voutilainen, Erama and Ritvos, 1991). The present western blotting data indicates that activin production rates in culture exceed those of inhibin (even taking into account that activin has two potential epitopes for the f3A-directed antibody while inhibin has only one). Since activin would reduce the apparent inhibin concentrations measured by bioassay, this may be a possible explanation for the low bioactive/immunoactive inhibin ratio reported in human maternal sera and cord blood (Qu and Thomas, 1992). Our data show that hCG and inhibin production are both stimulated by GnRH, glucocorticoids, cAMP, intracellular calcium and protein kinase-C, though, with the exception of protein kinase-C, inhibin secretion is less responsive to these signals. In contrast, activin production appeared to be responsive only to protein kinase-C stimulation. The divergence in maternal serum concentration of hCG and inhibin during pregnancy suggests that other as yet unidentified factors have important independent effects on their synthesis and secretion in vivo. One possibility is that expression of receptors within the placenta for a number of regulatory agents may vary with gestational age. Alternatively, as suggested recently by Cole et al (1992), an increase in the metabolic clearance of hCG may explain the absence of a rise in maternal serum hCG concentrations in the last trimester during pregnancy. ACKNOWLEDGEMENTS This project was supported by grants from the New Zealand Foundation for the Newborn and the Health Research Council of New Zealand. We thank Genentech for their gift of rh-inhibin, and Dr Y. Eto for the rhactivin. The assistance of the nursing staff at National Women's Hospital, Auckland, with the collection of placentae is also gratefully acknowledged.

REFERENCES Abe, Y., Hasegawab Y., Miyamoto, K., Yamagnchi, M., Andoh, A., Ibuki, Y. & Igarishi, M. (1992) Concentrations of immunoreactive inhibin during normal pregnancy in women. Journal of Clinical Endocrinology and Metabolism, 71, 133-137. Albiston, A. L., Lock, P., Burger, H. G. & Krozowski Z. S. (1990) Cloning and characterization of the rat alpha-inhibin gene. Molecular and CellularEndocrinology, 68, 121-128. Bandivdekar, A. H., Vijayalakshmi, S., Jaswaney, V. L. & Sheth A. R. (1981) Inhibin in human placenta. Indian Journal of ExperimentalBiology, 19, 744-745. BElisle, S., Petit, A., Bellbarba, D., Eseher, E., Lehoux, J. G. & Gallo-Payet, N. (1989) Ca 2+, but not membrane lipid hydrolysis, mediates human chorionic gonadotrophin production by LHRH in human term placenta. Journal of Clinical Endocrinology and Metabolism, 69, 117-122. Cole, L. A., Kardana, A., Sang-Yoon, P. & Braustein G. L. (1992) The deactivitation ofhCG by nicking and dissociation. Journal of Clinical Endocrinology and Metabolism, 76, 704- 710. de Jong, F. D. (1988) Inhibin. PhysiologicalReviews, 68, 555-607. Douglas, G. C. & King, B. F. (1989) Isolation of pure villous cytotrophoblasts from term placenta using immunomagnetic microspheres. Journal oflmmunologicalMethods, 119, 259-268. Ilekis, J. & Benveniste, R. (1985) Effects of epidermal growth factor, phorbol myristate actetate, and arachidonic acid on choriogonadotrophin secretion by cultured human choriocarcinoma cells. Endocrinology, 116, 24002409. Iwashita, M., Evans, M. I. & Catt, K.J. (1986) Characterization ofa gonadotrophin-releasing hormone receptor site in term placenta and chorionic villi.Journal of Clinical Endocrinology and Metabolism, 62, 127-132. Iwashita, M., Watanabe M., Adachi, T., Ohiro A., Shinozaki, Y., Takeda, Y. & Sakamoto, S. (1989) Effect of gonadal steroids on gonadotrophin-releasing hormone-stimulated chorionic gonadotrophin release by trophoblast cells. Placenta, 10, 103-112. Iwashitab M., Watanabe, M., Setoyama, T., Mimuro, T., Nakayamah S., Adaehi, T., Takeda, Y. & Sakamoto, S. (1992) Effects of diacylglycerol and gonadotrophin-releasing hormone on human chorionic gonadotrophin release by cultured trophoblast cells. Placenta, 13, 213-221. Jansen, E. H., Steenbergen, J., de Jong, F. H. & van der i o l e n , H.J. (1981) The use of affinity matrices in the purification of inhihin from bovine follicular fluid. Molecular and CellularEndocrinology, 21, 109-117.

Keelan et al: Regulation of lnhibin, Activin and hCG Production

817

Kato, Y. & Braunstein, G. D. (1990) Purified first and third trimester placental trophoblasts differ in vitro hormone secretion. Journal of Clinical Endocrinology and Metabolism, 70, 1187-1192. Kliman, H.J., Nestler, J. E., Sermasi, E., Sanger, J. M. & Straus III, J. F. (1986) Purification, characterization, and in vitro differentiation of cytotrophoblasts from human term placentae. Endocrinology, 118, 15771582. LaPoit, P. S., Piquette, G. N., Soto, D., Sineich, C. & Hsueh A. J. (1990) Regulation of inhibin subunit messenger ribonucleic acid levels by gnnadotrophins, growth factors, and gonadotrophin-releasing hormone in cultured rat granulosa cells. Endocrinology, 127, 823-831. Matsuzaki, N., Ying, L., Masushiro, K., Toushun,J., Shimoya, K., Taniguchi, T., Fumita, S. & Osarnu, T. (1992) Trophoblast-derived transforming growth factor-131 suppresses cytokine-induced, but not gonadotrophin-releasing hormone-induced, release of human chorionic gonadotrophin by normal human trophoblasts. Journal of Clinical Endocrinology and Metabolism, 74, 211-216. McLachlan, R. I., Healy, D. L., Robertson, D. M., Burger H. G. & de Kretzer, D. M. (1986) The human placenta: A novel source of inhibin. Biochemicaland Biophysical Research Communications, 140, 485-490. Morrish, D. W., Bhardwaj, D. & Paras, M. T. (1991) Transforming growth factor-131 inhibits differentiation and human chorionic gonadotrophin and human placental lactogen secretion. Endocrinology, 129, 22-26. Morrish, D. W., Bhardwaj, D., Dabbach, L. K., Marusyk, H. & Siy, O. (1987) Epidermal growth factor induces differentiation and secretion of human chorionic gonadotrophin and placental lactogen in normal human placenta. Journal of Clinical Endocrinology and Metabolism, 65, 1282-1290. Ohashi, K., Saji, F., Kato, M., Wakimoto, A. & Tanizawa, O. (1992) Tumour necrosis factor-cx inhibits human chorionic gonadotrophin secretion. Journal of Clinical Endocrinologyand Metabolism, 74, 130-134. Petraglia, F., Vaughan, J. & Vale, W. (1989) Inhibin and activin modulate the release of gnnadotrophinreleasing hormone, human chorionic gnnadotrophin and progesterone from cultured human placental cells. Proceedingsof the National Academy of Science USA, 86, 5114-5117. Petraglia, F., Sawchenko, P., Lim, A. T., Rivier, J. & Vale, W. (1987) IJocalization, secretion, and action of inhibin in human placenta. Science, 237, 187-189. Petraglia, F., Garuti, G. C., Calza, L., Roberts, V., Giardino, L., Genazzani, A. R., Vale, W. & Meunier, H. (1991) Inhibin subunits in human placenta: localization and messenger ribonucleic acid levels during pregnancy. AmericanJournal of Obstetrics and Gynecology, 165, 750-758. Petraglia, F., Woodruff, T. K., Botticelli, G., Botticelli, A., Genazzani, R., Mayo, K. E. & Vale, W. (1992) Gonadotropin-releasing hormone, inhibin and activin in human placenta. Journal of Clinical Endocrinology and Metabolism, 74, 1184-1188. Qu, J. & Thomas, K. (1992) Changes in bioactive and immunoactive inhibin levels around human labour.Journal of Clinical Endocrinology and Metabolism, 74, 1290-1295. Qu, J., Brulet, C. & Thomas, K. (1992) Effects of epidermal growth factor in inhibin secretion in human placental cell culture. Endocrinology, 131, 2173-2181. Qu, J., Ying, S.-Y. & Thomas, K. (1992) Inhibin production and secretion in human placental cells cultured in vitro. Obstetricsand Gynecology,79, 705-712. Qu, J. P., Vankrieken, L., Brulet, C. & Thomas, K. (1991) Circulating bioactive inhibin levels during human pregnancy. Journal of Clinical Endocrinology and Metabolism, 72, 862-866. Rabinovici, J., Goldsmith, P. C., Librach, C. L. & Jaffe, R. B. (1992) Localization and regulation of the activin-A dimer in human placental cells. Journal of Clinical Endocrinology and Metabolism, 75, 571-576. Ralph, M. M., Lee, C. S. & Thorburn, G. D. (1989) Identification and characterization of monolayer cultures of sheep trophoblast cells maintained in bicameral culture chambers. Biology of Reproduction, 41,481-489. Ringler, G. E. & Strauss III, J. F. (1990) In vitro systems for the study of human placental endocrine function. EndocrineReviews, 11, 105-123. Ringler, G. E., Kallen, C. B. & Strauss III, J. F. (1989) Regulation of human trophoblast function by glucorticoids: dexamethasone promotes increased secretion of chorionic gonadotrophin. Endocrinology, 124, 1625-1631. Robertson, D. M., Hayward, S., Irby, D., Jacobsen, J., Clarke, L., McLachlan, R. I. & de Kretser, D. M. (1988) Radioimmunossay of rat serum inhibin: changes after PMSG stimulation and gonadectomy. Molecular and CellularEndocrinology, 58, 1-8. Robertson, D. M., Giaeometti, M., Foulds, L. M., Lahnstein, J., Goss, N. H. Hearn, M. T. & de Kretser, D. M. (1989) Isolation of inhibin alpha-subunit precursor proteins from bovine follicular fluid. Endocrinology, 125, 2141-2149. Seher, W., Eto, Y., Ejima, D., Den, T. & Svet-Moldavsky, I. A. (1990) Phorbol ester-treated human acute myeloid leukemia cells secrete G-CSF, GM-CSF and erythroid differentiation factor into serum-free media in primary culture. Biochimica BiophysicaActa, 1055, 278-286. Shintani, Y., Takada, Y., Yarnasaki, R. & Saito, S. (1991) Radioimmunoassay for activin A/EDF. Method and measurement of immunoreactive activin A/EDF levels in various biological materials. Journal oflmmunological Methods, 137, 267-274. Strauss III, J. F., Kido, S., Sayegh, R., Sakuragi, N. & Gafvels, M. E. (1992) The cAMP signalling system in human trophoblast function. Placenta, 13, 389-403. Sugino, K., Nakamura, T., Takio, K., Miyamoto, K., Hasegawa, Y., Igarashi, M., Titani, K. & Sugino, H.

818

Placenta (1994), Vol. 15

(1992) Purification and characterization of high molecular weight forms of inhibin from follicular fluid. Endocrinology, 130, 789-796. Tabei, T., Ochiai, K., Terashima, Y. & Takanashi, N. (1991) Serum levels of inhibin in maternal and umbilicalblood during pregnancy.American Journal of Obstetrics and Gynecology, 164, 896-900. Tanimoto, K., Tamura, K., Ueno, N., Usuki, S., Murakami, K. & Fukamizu, A. (1992) Regulationofactivin beta A mRNA level by cAMP. Biochemicaland Biophysical Research Communications, 182, 773-778. Tierney, M. L., Goss, N. H., Tomkins, S. M., Kerr, D. B., Pitt, O. E., Forage, R. G., Robertson, D. M., Hearn, M. T. & de Kretzer, D. M. (1990) Physicochemicaland biologicalcharacterizationof recombinant human inhibin. Endocrinology, 126, 2368-2370. Voufilainen, R., Erarna, M. & Ritvos, O. (1991) Hormonallyregulated inhibin gene expressionin human fetal and adult adrenals.Journal of Clinical Endocrinology and Metabolism, 73, 1026- ! 030. Yagel, S., Casper, M. D., Powell, W., Perhar, R. S. & Lala, P. K. (1989) Characterization of pure first trimester cytotrophoblastcells in long-term culture: growth pattern, markers and hormone production.AmericanJournal of Obstetrics and Gynecology, 160, 938-945. Ying, S. Y. (1989) Inhibins, activins,and follistatins.Journal of Steroid Biochemistry, 33, 705-713. Ying, L., Matsuzaki, N., Masushiro, K., Kameda, T., Taniguchi, T., Saji, F., Yone, K. & Tanizawa, O. (1992) Trophoblast-derived tumour necrosis factor-cd induces release of human chorionic gonadotrophin using interleukin~ (IL-6) and IL-6-receptor-dependent system in the normal human trophoblasts.Journal of Clinical Endocrinology and Metabolism, 74, 184-191.