Activin and Follicle-Stimulating Hormone Signaling Are Required for Long-Term Culture of Functionally Differentiated Primary Granulosa Cells from the Chicken Ovary¹

Abstract

Follicle-stimulating hormone, activin A, and transforming growth factor (TGF) α are important regulators of chicken granulosa cell (cGC) function. Hence, we aimed to test whether these growth factors are useful for establishing a suitable in vitro cell culture model system of primary cGC. Although cGC are easily isolated from distinct follicular stages, a long-term cGC culture system for in vitro studies has been unavailable. Here, we report a novel, long-term cell culture system that allows for cGC proliferation in vitro while maintaining the epithelial phenotype that granulosa cells exhibit in vivo. The cGC rapidly lose their epithelial morphology and acquire a mesenchymal or fibroblastoid phenotype when cultured in the absence of activin A. This process is strongly enhanced by TGFα, a well-known granulosa cell mitogen. However, FSH stimulates cGC proliferation without enhancing morphological changes and dedifferentiation. Interestingly, a combination of both activin A and FSH stimulates cGC proliferation and supports maintenance of differentiated epithelial morphology. Furthermore, activin A and FSH synergistically induce granulosa cell-specific differentiation markers such as inhibin α and chicken zona pellucida protein C, suggesting that cultured cGC resemble functionally differentiated granulosa cells. Our data demonstrate that activin signaling is necessary to sustain a morphologically differentiated phenotype of cGC in vitro. The results also suggest a pivotal importance of activin signaling for granulosa cell function in vivo.

Introduction

The ovary of the domestic chicken (Gallus gallus) constitutes an ideal model system for studies of ovarian biology and oocyte development. Several large follicles are present in the ovary at any given time, representing the follicular hierarchy in which F1 is the most mature follicle that ovulates before F2 and so forth. Chicken granulosa cells (cGC) are easily isolated from such hierarchical follicles without significant contamination by other cell types [1]. The architecture of follicles from the center to the periphery consists of a single yolk-filled oocytic germ cell, the acellular perivitelline layer (PVL), the granulosa cells, a basal membrane, and theca cells. The PVL surrounds the oocyte and consists of fibrous network of proteins that also contains the sperm receptors essential for fertilization [2]. Hence, the PVL is the avian equivalent of the mammalian zona pellucida. A major component of the PVL is chicken zona pellucida protein C (chZPC), the orthologue of mouse zona pellucida 3 (ZP3). Mouse ZP3 is only expressed by the oocyte [3], whereas chZPC is exclusively produced and secreted from granulosa cells [4].

Granulosa cells are epithelial cells of endodermal origin [5], forming a monolayer of closely packed cuboidal cells sandwiched between the PVL and the basal membrane. Granulosa cells are typical steroidogenic endocrine cells and, in avian species, produce mainly progesterone, which is further metabolized by juxtaposed theca cells to androgens and estrogens [6]. Granulosa cells are in closest proximity to the germ cell and are thought to play an important role in oocyte development and early embryogenesis. For example, granulosa cells may produce maternal factors destined for the developing embryo [7]. Moreover, cGC secrete high-density particles containing apolipoprotein A-I toward the oocyte [8].

Two important regulators of cGC proliferation are transforming growth factor (TGF) α and FSH. The TGFα signals through a mitogen-activated protein kinase (MAPK) pathway via the epidermal growth factor (EGF) receptor (EGF-R), whose activation stimulates cGC proliferation [9]. The FSH is produced in the anterior pituitary gland and stimulates proliferation and differentiation of ovarian granulosa cells as well as testicular Sertoli cells in vivo. These are the only somatic cell types expressing the cognate FSH receptor (FSH-R) [10], a typical seven-transmembrane G-protein-coupled receptor. The FSH-R activates adenylyl cyclase, and elevation of intracellular cAMP levels causes protein kinase A activation. Release of FSH from the pituitary gland is tightly controlled by the activin/inhibin growth factors, with activin A activating FSH release [11] and inhibin blocking FSH release [12]. Hence, these growth factors generate a feedback loop to control FSH biosynthesis, thereby regulating FSH availability in reproductive tissues.

Activins and inhibins are members of the TGFβ family of growth factors, acting as homo- or heterodimers of three different subunits: inhibin β_A, inhibin β_B, and inhibin α. The αβ heterodimers, αβ_A and αβ_B, are called inhibin A and inhibin B, respectively; the ββ homodimers, β_Aβ_A and β_Bβ_B, are called activin A and activin B, respectively; and the β_Aβ_B heterodimer is called activin AB [13]. Like all members of the TGFβ family, activins signal through specific receptors, all of which share intrinsic serine-threonine kinase activities [14]. Activin A also plays important roles in hematopoiesis [15]; embryogenesis, during which activin A is a potent inducer of mesoderm [16]; and axial structure formation [17]. Granulosa cells are a major source for activin A [18–20]. Moreover, a role for activin A in granulosa cell physiology through autocrine or paracrine signaling is becoming increasingly established [21]. However, knowledge of molecular events triggered by activin A within the ovary remains limited. In the present manuscript, we report the first, to our knowledge, long-term in vitro culture system for primary untransformed epithelial-like granulosa cells. We show that cultured granulosa cells require activin A to maintain a differentiated epithelial phenotype in vitro, suggesting a pivotal role for activin A in cGC physiology in vivo. This model system enables studies of cGC differentiation and morphology as well as of signal transduction events specific for this highly specialized follicular cell type. Finally, this system will be instrumental in answering key questions concerning follicular architecture, function, growth, and maturation.

Materials and Methods

Animals and Isolation of cGC for Cell Culture

Brown Derco laying hens were obtained from Heindl Co. (Vienna, Austria) and maintained on layer's mesh with water and feed provided ad libitum under a photoperiod of 16L:8D. All animal experimentation was in accordance with the regulations of the animal ethics committee of the University of Vienna. Hens were killed by decapitation, and granulosa sheets were isolated from large preovulatory follicles (POFs) as described previously [1] using minor modifications. Briefly, follicles were punctured with sterile forceps, and the yolk was carefully removed until granulosa sheets appeared. The sheets, which consist of the PVL and adhering granulosa cells, were washed in cold, sterile PBS to remove residual yolk and digested with 1 mg/ml of collagenase type IV (Sigma, St. Louis, MO) in PBS for 15 min at 37°C until cells were completely separated. The cGC were washed twice by resuspension in cold PBS, followed by a 4-min centrifugation at 250 × g at 4°C. Cell viability of cGC exceeded 95% after their isolation. Cells were plated at densities of 1 × 10⁵ cells/cm² in Dulbecco modified Eagle medium (DMEM; catalog no. 41965; GibcoBRL Life Technologies, Paisley, U.K.) supplemented with penicillin (50 U/ml), streptomycin (50 μg/ml), and 5% fetal calf serum (FCS; GibcoBRL Life Technologies). Cells were incubated under standard conditions (37°C, 5% CO₂). Unless otherwise indicated, the following growth factor concentrations were used: 50 ng/ml (1.4 nM) of human pituitary FSH (Calbiochem, San Diego, CA), 25 ng/ml (1.8 nM) of human recombinant activin A, and 10 ng/ml (1.6 nM) of human recombinant TGFα (R&D Systems, Minneapolis, MN). All experiments were performed in the presence of 5% FCS.

Monitoring of Cell Proliferation

Cell numbers were measured using the CellTiter96 AQ_ueous Non-Radioactive Cell Proliferation Assay (Promega, Madison, WI) according to the manufacturer's protocol. Briefly, a freely diffusible substrate was reduced by nonspecific cellular enzymes, thereby producing a colored formazan analogue. The amount of formazan was measured by spectrophotometry, and absorbance was directly proportional to the cell number.

Long-Term Proliferation Studies

The cGC isolated from F4 and F5 follicles were pooled and plated at half-confluence in DMEM containing the indicated amounts of FSH and activin A. After reaching confluence, cells were trypsinized and split into two aliquots. One aliquot was counted in a CASY 1 TTC cell counter (Schaerfe System, Reutlingen, Germany); the other aliquot was split 1:3 and again grown to confluence. This procedure was repeated until cell growth was no longer detectable. After 15, 18, and 24 days in culture, total protein extracts were prepared as described below.

Protein Preparation, SDS-PAGE, and Immunoblotting

Adherent granulosa cells were lysed directly in the culture dishes by addition of 30 μl/cm² of SDS-PAGE sample buffer (5% [v/v] glycerol, 1% [w/v] SDS, 2.5% [v/v] β-mercaptoethanol, 30 mM Tris/HCl [pH 6.8], and 0.025% [w/v] bromphenol blue). After boiling for 5 min, lysates were stored frozen at −20°C. Protein extracts were separated on 10% [w/v] polyacrylamide gels and transferred to Protran nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany). Membranes were stained with Ponceau S to verify protein transfer and equal loading of protein and then blocked with 5% (w/v) fat-free milk powder in PBS + 0.1% (v/v) Tween for 1 h at room temperature. Antibody incubation was done at 4°C overnight using rabbit anti-chZPC antibodies [4] at a dilution of 1:5000. After incubation with secondary goat-anti-rabbit immunoglobulin G-horseradish peroxidase (Oncogene, Boston, MA) at a dilution of 1:10 000, bands were visualized using enhanced chemiluminescence Western blotting detection reagent (Amersham Pharmacia Biotech, Buckinghamshire, U.K. ) as suggested by the manufacturer.

Immunofluorescence

Large POFs from adult hens were embedded in freezing agent (Microm, Walldorf, Germany) and shock-frozen on dry ice immediately. Cryostat sections (thickness, 20 μm) were prepared using an HM 500 OM cryomicrotome (Microm) and transferred to Superfrost-Plus slides (Menzel, Braunschweig, Germany). Alternatively, cGC were isolated from POFs and spread on a microscope slide. After fixation in acetone:methanol (1:1) at −20°C for 15 min, sections or sheets were rehydrated for 15 min in PBS at 37°C. Cultured cGC were grown on collagen-coated culture slides (Becton Dickinson, Bedford, MA) and fixed as above. Mouse anti-pan-cadherin antibodies (Sigma) were used at a dilution of 1:500 in 1% fat-free milk powder/PBS; polyclonal rabbit anti-ZO-1 antibodies (Zymed, San Francisco, CA) were used at a dilution of 1:200. The secondary antibodies Alexa Fluor 594 goat-anti-mouse and Alexa Fluor 488 goat-anti-rabbit (Molecular Probes, Eugene, OR) were used at dilutions of 1:500 in 1% fat-free milk powder/PBS for 1 h at 37°C. After several PBS washes, slides were mounted in fluorescence mounting medium (DAKO, Carpenteria, CA) and inspected using an Axiovert 135 microscope (Zeiss, Jena, Germany).

Isolation of Total RNA and Northern Blot Analysis

Pooled F4 and F5 cells were cultured in DMEM containing activin A and FSH. Cells were split 1:2 and incubated for 48 h in DMEM containing either activin A, TGFα, FSH, or the combinations TGFα/activin A or FSH/activin A at the concentrations indicated above. Total RNA was isolated using Tri-Reagent (Molecular Research Center, Cincinnati, OH) according to the manufacturer's recommendations. Approximately 20 μg of total RNA were incubated with one volume of denaturing solution (2 M glyoxal, 5 mM sodium phosphate buffer [pH 6.8], 2.5mM EDTA, and 66% dimethyl sulfoxide) for 1 h at 50°C to resolve RNA secondary structures. Next, a one-sixth volume of 6× loading solution (0.25% [w/v] bromphenol blue, 0.25% [w/v] xylenecyanol, and 30% glycerol) was added, and samples were loaded onto 1.5% [w/v] agarose gels. The RNA gels were run in 10 mM sodium phosphate buffer (pH 6.8) at 10 V/cm. Separated RNAs were blotted to positively charged Hybond-N⁺ nylon membranes (Amersham Pharmacia Biotech) in 20× SSC (3 M NaCl and 0.3 M sodium citrate, pH 7.0) using capillary transfer. The RNA was ultraviolet-cross-linked using an energy output of 120000 μJ (Stratalinker; Stratagene, La Jolla, CA). The RNA on membranes was visualized by methylene blue staining (0.04% [w/v] methylene blue in 0.5 M sodium acetate [pH 4.5]). After destaining with H₂O, membranes were prehybridized for 4 h at 65°C in 0.5 M sodium phosphate buffer [pH 6.8] containing 7% SDS.

Radiolabeling of cDNA Probes and Hybridization

Approximately 25-ng aliquots of cDNA probes were radiolabeled using a random-prime labeling system (Megaprime cDNA Labeling Kit; Amersham Pharmacia Biotech) in the presence of 50 μCi [α-³²P]dCTP (Hartmann Analytic, Braunschweig, Germany) as suggested by the manufacturer. Unincorporated nucleotides were removed by gel filtration chromatography using Sephadex G-50 columns (NICK Columns; Amersham Pharmacia Biotech). Radiolabeled cDNA probes were denatured at 95°C for 5 min and added directly to the prehybridization solution. Hybridization was carried out overnight at 65°C. After washing in wash buffer (40 mM sodium phosphate buffer [pH 6.8] and 1% SDS), membranes were wrapped in Saran wrap and exposed to phosphor-imaging screens for up to 3 days. Bands were scanned in the optical scanner Storm 840 (Molecular Dynamics, Sunnyvale, CA). Northern blots were also exposed to x-ray films to visualize RNA bands by autoradiography.

Statistical Analysis

All experiments were repeated at least three times. Northern blots and immunoblots show results of representative experiments. Colorimetric cell proliferation assays were done in quadruplicates, with values expressed as the mean ± SD. Eight entirely independent proliferation experiments were carried out, which showed remarkable reproducibility. One-way ANOVA followed by the Tukey honestly significant difference post-hoc test was performed. All discussed differences proved to be statistically significant (P < 0.05 or P < 0.01).

Results

Granulosa Cell Proliferation In Vitro

Studies on granulosa cell biology are hampered by the fact that all currently available in vitro cell culture systems do not allow long-term propagation of primary granulosa cells. This prompted our efforts to determine the serum and growth factor requirements for culturing primary cGC, an epithelial cell type from the avian follicle. For cell attachment and viability, FCS proved to be essential. We also examined the effects of various growth factors on cGC proliferation and differentiation. Thus, TGFα, EGF, FSH, and activin A were tested for their effect on cGC proliferation, because they are important regulators of ovarian function. Compared to cGC grown in 5% serum without additional growth factors (untreated control cells), TGFα, FSH, and EGF stimulated cell proliferation to different extents in a dose-dependent manner, with TGFα being the most effective mitogen (P < 0.05). Low doses of 1.6 nM TGFα already led to a maximal response. After 48 h, TGFα-treated cultures contained twice as many cells as untreated control cultures incubated for the same time period. Although EGF recognizes the same receptor as TGFα, it was much less efficient in stimulating cGC proliferation (P < 0.05), because only EGF concentrations greater than 4 nM markedly stimulated cGC growth. The FSH promoted cGC proliferation to a lesser extent than TGFα, although it triggered maximal response at comparable concentrations (Fig. 1A).

Fig. 1.

A) TGFα, FSH, and EGF stimulate cGC proliferation to different extents. Identical numbers of cGC were plated in DMEM containing increasing amounts of the indicated growth factors in the presence of 5% FCS. After 48 h, cell number was assessed by MTS assay as described in Materials and Methods. Values obtained for TGFα differ significantly from values obtained for EGF and FSH (P < 0.01) at the three highest concentrations used. B) Activin A inhibits granulosa cell proliferation. Identical numbers of cGC were plated at 25% confluence in DMEM containing the indicated growth factors. After 2, 24, and 48 h, MTS (3-[4,5-dimethylthiazol-2-yl]-5-[3-carboxymethoxyphenyl]-2-[4-sulfophenyl]-2H-tetrazolium); assays were performed as described in Materials and Methods. Values measured after 2 h were considered to be the plating number and used as reference. All values are expressed as the mean ± SD (n = 4).

We then examined the role of the TGFβ-like activin A, which is produced mainly in ovarian tissues (Fig. 1B). Notably, activin A efficiently blocked cGC proliferation if applied alone. However, both TGFα and FSH counteracted this antiproliferative role of activin A, allowing for cell proliferation of activin A-treated cells, albeit at attenuated efficacy. To mimic FSH action, we also added the cAMP analogue 8-Br-cAMP to the culture medium. Whereas 10 μM 8-Br-cAMP failed to overcome activin A-induced proliferation arrest, 100 μM 8-Br-cAMP permitted cell proliferation of activin A-treated cells to an extent comparable to that of untreated cells, indicating that FSH signals mainly via cAMP. Taken together, these data demonstrate that activin A inhibits cGC proliferation in vitro, whereas this antiproliferative effect can be overcome by either TGFα, FSH, or 8-Br-cAMP (Fig. 1B).

Granulosa Cell Morphology In Vivo and In Vitro

The cGC are believed to exhibit an epithelial phenotype in developing follicles in vivo. To verify this epithelial phenotype, we employed immunofluorescence staining of cryosections from POFs or isolated granulosa sheets using antibodies against epithelial marker proteins such as cadherins (Fig. 2). In contrast to the juxtaposed theca cells, cadherin staining in cGC illustrates the typical epithelial phenotype of cGC in vivo (Fig. 2A). Cadherins localize to the plasma membrane at cell-cell contacts and form belt-like adherens junctions that are characteristic for epithelial cells (Fig. 2B). Similarly, we have used other epithelial marker proteins, such as β-catenin and ZO-1, to demonstrate the epithelial nature of cGC in vivo (data not shown).

Fig. 2.

Epithelial phenotype of cGC in vivo. A) Twenty-micrometer cryosections of POFs were immunostained with anti-pan-cadherin antibodies and visualized by immunofluorescence microscopy. Granulosa cells (GC) show a cobblestone-like staining pattern, indicating an epithelial phenotype, with extensive belt-like adherens junctions. By contrast, theca cells (TC) exhibit a mesenchymal phenotype characterized by an irregular distribution of cadherins. OO, Oocyte. B) Granulosa cell sheets containing parts of the perivitelline layer attached to the cGC monolayer were prepared from POFs and immunostained with anti-pan-cadherin antibodies. Nuclear DNA was stained with DAPI (4,6-diamidino-2-phenylindole). The insertion shows a control without primary antibodies

However, cGC rapidly lost this epithelial morphology within 48 h on cultivation in vitro. Cells started to spread, losing the characteristic cobblestone-like growth pattern and, at the same time, acquired a mesenchymal or fibroblastoid morphology. As depicted in Figure 3, the localization of cadherin and ZO-1 remained normal despite these morphological changes. Addition of TGFα to the culture medium enhanced cell scattering, which otherwise occurred spontaneously, indicating a dedifferentiating effect of the mitogen TGFα. Moreover, cadherins and ZO-1 became redistributed, leading to a loss of correct plasma membrane localization, although immunoblots revealed that expression levels of these proteins remained constant (data not shown). Interestingly, activin A prevented cell scattering, and cells acquired an epithelial morphology with intact cell-cell contacts. The cGC formed clusters despite low cell numbers because of the activin A-induced proliferation arrest, and junctional proteins localized predominantly to the plasma membrane. Cells treated with a combination of TGFα and activin A proliferated and were able to form confluent monolayers, showing an epithelial cobblestone-like morphology, with a predominantly plasma membrane localization of junctional proteins. Treatment with FSH induced proliferation but failed to prevent cell spreading and scattering; localization of junctional proteins resembled the situation in untreated control cells. Finally, cells treated with a combination of FSH and activin A grew to high densities and still displayed an epithelial morphology, and cell scattering was efficiently prevented. Cadherin and ZO-1 showed strong membrane localization, because neighboring cells were tightly connected by adherens junction-like structures. Hence, considering cell proliferation and morphology, a combination of TGFα/activin A or FSH/activin A allowed for culturing cGC in vitro while maintaining the epithelial morphology seen in vivo. Importantly, activin A was required for the maintenance of an epithelial morphology, whereas FSH or TGFα were necessary to overcome the activin A-induced proliferation arrest.

Fig. 3.

Growth factors exert pronounced effects on cGC morphology and localization of junctional proteins. Identical numbers of cGC were plated in DMEM to approximately 50% confluence and incubated for 48 h in the presence of the indicated growth factors. Phase-contrast view (middle), cadherin immunostaining (red, left), and ZO-1 immunostaining (green, right) are shown

Marker Protein Expression in cGC

We then investigated the effects of medium compositions on cGC marker protein expression. In this respect, cGC treated with FSH/activin A resembled the in vivo phenotype much more closely than cGC treated with TGFα/activin A. We used Northern blot analysis and immunoblotting to analyze expression levels of three distinct cGC marker proteins: chZPC, FSH-R, and inhibin α. The chZPC provides an excellent marker for differentiated cGC because of its granulosa cell-specific expression [4]. Untreated cGC expressed low amounts of chZPC protein in vitro (Fig. 4). Importantly, TGFα completely abolished chZPC expression irrespective of the absence or presence of activin A. This confirms the dedifferentiating function of the mitogen TGFα. By contrast, FSH and activin A treatment moderately elevated chZPC protein levels. Surprisingly, the combined action of activin A and FSH increased chZPC expression in a more than additive manner, leading to very high protein levels comparable to those of chZPC observed in vivo (Fig. 4).

Fig. 4.

chZPC expression is repressed by TGFα and induced by FSH/activin A. Granulosa cells from F4 follicles were incubated for 48 h in DMEM containing the indicated growth factors. Protein extracts were analyzed by immunoblotting using polyclonal anti-chZPC antibodies. The first lane contains an extract from the same number of freshly isolated F4 granulosa cells, illustrating the in vivo expression levels of chZPC

Inhibin α is considered to be another specific marker protein for cGC [22]. Northern blot analysis showed that cultured cGC expressed inhibin α, and steady-state levels of inhibin α mRNA were increased by activin A. The TGFα did not influence inhibin α expression or up-regulation. Whereas FSH alone had no effect, a combination of FSH and activin A led to a pronounced increase of inhibin α mRNA levels (Fig. 5). These data further support the notion of a functional synergism between FSH and activin A. Finally, expression of the FSH-R is a pivotal requirement for FSH responsiveness of cultured cGC cells [23], which in turn is essential for granulosa cell function. The TGFα promoted down-regulation of FSH-R, again confirming the dedifferentiating action of this mitogen. By contrast, activin A induced FSH-R mRNA in our cGC culture system (Fig. 5). Although different splice variants of the FSH-R exist in different species [10], to our knowledge only a single 4.3-kilobase (kb) mRNA species has been described in chicken tissues so far [24]. However, we discovered two major FSH-R mRNA species of 4.3 and 3.9 kb in cultured cGC. Both FSH-R mRNA species showed an activin A-dependent regulation pattern. Hence, activin A most likely acts at the level of transcription and/or mRNA stability.

Fig. 5.

Regulation of inhibin-α and the FSH-R mRNAs. Identical numbers of granulosa cells from F4 follicles were plated in DMEM and incubated for 24 h in the presence of the indicated growth factors. Total RNA was isolated and subjected to Northern blot analysis. Methylene blue staining of 28S rRNA served to verify equal RNA loading

Long-Term Proliferation of cGC in Optimized Culture Medium

Our data indicated that combining FSH and activin A represents an ideal growth factor cocktail to propagate primary cGC in vitro. The cGC were both proliferating and exhibiting an epithelial morphology closely resembling the in vivo situation. The FSH responsiveness was sustained, and important marker proteins were highly expressed. By contrast, a combination of TGFα and activin A failed to maintain marker protein expression, suggesting that the mitogenic action of TGFα is linked to cGC dedifferentiation. To evaluate the actual long-term growth potential of primary cGC in vitro, cells were propagated in the optimized medium for prolonged time periods. Culture was performed in DMEM containing 50 ng/ml of human FSH and 25 ng/ml of human recombinant activin A. These conditions provide the minimal growth factor concentrations necessary for maintenance of an epithelial phenotype on the one hand and maximal stimulation of proliferation on the other. Cells proliferated evenly over a time period of more than 3 wk, thereby increasing cell number by a factor of 160, which is equivalent to at least seven generations (Fig. 6A). By contrast, untreated control cells lost the epithelial phenotype within 48 h, poorly survived trypsinization, and died within 1 wk (data not shown). During the exponential growth period up to Day 21, FSH/activin A-treated cells exhibited a generation time of approximately 70 h, maintaining both their epithelial phenotype (Fig. 6B) and expression of the cGC marker chZPC (Fig. 6C). Taken together, our results establish the first, to our knowledge, culture system for untransformed, primary ovarian granulosa cells. This culture system allows studies of granulosa cell differentiation and morphology as well as of signal transduction events specific for this highly specialized cell type, and it will be invaluable for answering key questions concerning folliculogenesis as well as oocyte growth and development.

Fig. 6.

Long-term culture of F4 granulosa cells. A) Cells proliferate exponentially throughout the culture period of up to 25 days. B) Epithelial cobblestone-like morphology of cGC in long-term culture. The picture was taken after 20 days in culture. C) Immunoblot showing protein levels of chZPC in long-term cultures of cGC

Discussion

Studies concerning the role of cGC in folliculogenesis require appropriate in vitro cell culture systems. The present work establishes a long-term cell culture system for nontransformed, primary cGC, which is useful to study growth factor signaling in the ovary. The culture of primary cGC has been hampered by weak proliferation rates, loss of morphology, and loss of responsiveness toward FSH [23]. To be useful as a model system, cultured cGC have to meet different requirements, including proliferation and growth as well as responsiveness to FSH. Importantly, the epithelial cobblestone-like morphology as well as expression of marker proteins such as chZPC, a specific granulosa marker in the chicken [4], should be maintained in vitro. Steroidogenesis is frequently used as marker for functionally differentiated granulosa cells, especially in the establishment of stably transfected cell lines [25]. However, steroidogenesis in birds serves only as marker for terminally differentiated, weakly proliferating granulosa cells [26, 27]. By contrast, chZPC expression is a feature of both terminally differentiated and proliferating granulosa cells.

Untreated cGC proliferate moderately and only weakly express chZPC in vitro. Furthermore, such cells fail to sustain a proper epithelial morphology and acquire a rather mesenchymal phenotype that is characterized by flat, spread cells without pronounced cell-cell contacts. Notably, granulosa cell-conditioned medium rapidly induces cell spreading and scattering in cultures of freshly dispersed cGC (data not shown). Hence, these spontaneously occurring morphological changes are perhaps caused by the action of some unknown factor acting in an autocrine fashion. Two possible candidate factors triggering spontaneous loss of the epithelial phenotype are EGF and TGFα. The latter acts as a strong mitogen that also induces pronounced dedifferentiation events, namely enhanced cell spreading and scattering, as well as acquisition of a mesenchymal phenotype reminiscent of the morphology of untreated cGC in culture. Both TGFα and EGF, both of which are ligands of EGF-R, activate MAPK pathways, and both factors promote cGC proliferation. However, human recombinant TGFα stimulates proliferation much more efficiently than human recombinant EGF. This can be explained by the fact that the chicken EGF-R exhibits a 100-fold higher affinity for human TGFα than for human EGF [28, 29]. Whether chicken EGF or chicken TGFα is the main effector in vivo remains unknown. Besides induction of a mesenchymal phenotype, TGFα-mediated suppression of chZPC expression is another major drawback of this growth factor regarding its use in cGC culture and, again, points to dedifferentiation. Moreover, steady-state levels of FSH-R mRNA are decreased in the presence of TGFα, perhaps causing reduced FSH responsiveness. This observation is consistent with results obtained in the rat ovary [30]. Taken together, TGFα stimulates cell proliferation most efficiently but fails to sustain a differentiated phenotype. It redistributes cadherins and ZO-1, whose correct localization is heavily disturbed, induces a mesenchymal morphology that lacks appropriate marker protein expression, and is likely to decrease FSH responsiveness. Therefore, TGFα alone is of limited use for cGC cell culture. Similarly, FSH stimulates granulosa cell proliferation, albeit to a lesser extent than TGFα. Unlike TGFα, FSH neither triggers cell spreading nor represses expression of chZPC; however, it fails to suppress spontaneously occurring morphological changes. Altogether, FSH promotes proliferation and sustains the in vivo phenotype much better than TGFα, yet FSH fails to sustain an epithelial phenotype of cGC in vitro.

The TGFβ-like activin A is secreted by granulosa cells and exerts potent effects by regulating FSH availability. Because cGC express activin receptors and secrete activins, autocrine and/or paracrine signaling events are, perhaps, crucial for cGC function. We demonstrate that cultured cGC indeed respond to exogenous activin A and that activin signaling is active within this cell type. Furthermore, activin A seems to be a pivotal differentiation factor for cGC, which not only regulates FSH availability but also cooperates with FSH to sustain cGC functions in vitro. Activin A arrests cell proliferation, and cells are not able to form confluent monolayers in culture. Activin A-treated cells cluster in islands and do not scatter, despite the availability of sufficient space. Within these clusters, cells exhibit an epithelial phenotype, displaying a cobblestone-like growth pattern. Inhibition of cell proliferation is a well-known property of the prototypic member of the growth factor family TGFβ, which is known to signal through the same subset of Smad proteins as activin A. Activin A itself has been shown to arrest cell proliferation of endothelial cells [31], hepatocytes [32], and breast cancer cells [33].

The TGFα stimulates cGC proliferation with concomitant induction of a mesenchymal phenotype. Membrane localization of the junctional proteins cadherin and ZO-1 disappears, leading to a loss of normal cell-cell junctions. However, activin A attenuates TGFα-induced effects, such as proliferation and dedifferentiation, being essential for the maintenance of an epithelial morphology. In turn, TGFα abolishes chZPC expression and represses activin A-dependent chZPC induction. Hence, activin A and TGFα seem to feed antagonizing signaling events. Notably, TGFα signals mainly through ras activation, which can lead to atypical Smad protein phosphorylation patterns, thereby suppressing TGFβ signaling in mammary and lung epithelial cells [34]. This finding provides a possible explanation for impaired activin A action in the presence of TGFα.

The cGC further respond to activin A by induction of inhibin α mRNA and, importantly, of FSH-R mRNA, which is also true for mammals [35, 36], thereby ensuring continued FSH responsiveness. In contrast to previously published results [24], we detected two distinct FSH-R transcripts, despite the fact that the same cDNA probe was used for Northern hybridization. Differential splicing of the FSH-R mRNA is not unusual and, to our knowledge, has been reported for all mammalian species examined so far [10]. In summary, a combination of FSH and activin A in medium containing 5% serum proved to be optimal for proliferation of functionally differentiated cGC. We cannot exclude a dependence of cGC proliferation on additional, as-yet-unknown, serum-borne growth factors or hormones, but our data clearly demonstrate a strict requirement for both FSH and activin A. First, activin A hardly slows down FSH-induced proliferation. Second, the cells exhibit a cobblestone-like phenotype with correct localization of junctional proteins. Third, cells remain FSH responsive, and fourth, chZPC is massively induced to in vivo expression levels. These features are maintained for up to 4 wk in culture. These results demonstrate synergistic effects of FSH and activin A, which were not only observed in the induction of chZPC protein but also that of inhibin α mRNA, which is massively induced by this growth factor combination. A synergistic action of FSH and activin A on gene expression has been reported for genes associated with proliferation, including cyclin D2 and proliferating cell nuclear antigen [37]. A possible mechanism for this synergistic induction remains to be uncovered.

For reasons outlined above, the ovary of the domestic chicken (G. gallus) provides an ideal system to investigate granulosa cell biology. Our future work concerning the role of granulosa cells in folliculogenesis can rely on a novel in vitro cell culture system of nontransformed cGC. We also demonstrate that cGC strongly depend on direct, probably autocrine activin A signaling, so as to regulate cell proliferation and to develop an epithelial, polarized phenotype, a prerequisite for granulosa cell function and follicular architecture. Finally, our results provide evidence that activin A and FSH trigger cooperating signaling pathways, which synergize in inducing chZPC and inhibin α expression. Importantly, our data suggest that well-established FSH actions are modulated by activin A signaling, expanding the role of activin A from a mere regulator of FSH biosynthesis to an FSH cooperator and intraovarian modulator of FSH action.

Acknowledgments

We would like to thank Wolfgang Schneider for providing the anti-chZPC antibodies, Alan Johnson for the FSH-R cDNA probe, and Julia Hilscher for technical assistance.

Author notes

¹

Supported through grant SFB-604 from Austrian Science Foundation (FWF).