Mothers Against Decapentaplegic-Related Protein 2 Expression in Avian Granulosa Cells Is Up-Regulated by Transforming Growth Factor β during Ovarian Follicular Development.

Although mothers against dpp (MAD) and its related proteins (MADR) are believed to be important components of the cell signaling pathway for the transforming growth factor β (TGFβ) superfamily, the presence and regulation of these signaling molecules in ovarian cells by TGFβ is not known. In the present studies, we have examined the presence of MADR2 and MADR1, two members of the MADR family, in hen granulosa cells at different stages of follicular development. The influence of TGFβ in vitro on their expression was assessed, particularly in the context of TGFβ-induced down-regulation of cytosolic phospholipase A2 (cPLA2), a key enzyme in the biosynthesis of eicosanoids. We have demonstrated for the first time the presence of MADR2 and MADR1 in hen granulosa cells at different stages of follicular development. The expression of MADR2, but not of MADR1, was up-regulated by TGFβ in vitro in a concentration- and time-dependent manner. Granulosa cell MADR2 expression was maximal during early stages of follicul...

T HE TRANSFORMING growth factor ␤ (TGF␤) superfamily is one of the largest groups of polypeptide growth factors and has distinct but diverse growth and differentiative functions in many physiological systems. Although little is known concerning the mechanism(s) of action of these regulators, recent studies indicate that they act via binding to serine/threonine kinase receptors (1). More recently, the novel gene family MAD (mothers against dpp) and its related proteins (MADR) have been identified in a variety of species as important components of the signal transduction pathway involving serine/threonine kinase receptor signaling, including that of TGF␤ (2)(3)(4)(5)(6)(7)(8)(9). The binding of TGF␤ to its receptor results in the formation of a heteromeric receptor complex that is activated via phosphorylation (1,10). One of these signaling proteins, MADR2, is then phosphorylated by the type I receptor, which leads to nuclear accumulation of MADR2 (11). Similarly, activin, another member of the TGF␤ superfamily, is also believed to act through an MADR2-mediated signaling pathway (2,12). It has been suggested that activin receptor activation results in the binding of MADR2, but not MADR1, to FAST-1 (a winged-helix DNA-binding protein) to form a site-specific transcriptional regulatory complex that has an important role at the promoter region of the activin-responsive gene Mix.2 (12).
Ovarian follicular development is the culmination of proliferation and differentiation of granulosa and theca cells, the nature and extent of which are dependent on the actions and interactions of gonadotropins and intraovarian regulators (13). TGF␤, a secretory product of hen granulosa and theca cells throughout follicular development (14), seems to act antagonistically with TGF␣ in the regulation of granulosa cell function (15)(16)(17)(18). Previous studies from our laboratory have demonstrated that PGs play an important role in the mitogenic response of granulosa cells to TGF␣ and that the production of PGs by granulosa cells is modulated by TGF␤ in a follicular stage-dependent manner (19). Although cyclooxygenase II seems to be a site of regulation by TGF␤ (20), more recent evidence points to the fact that this growth factor also suppresses the expression of cytosolic phospholipase A2 (cPLA 2 ), a key enzyme in the biosynthesis of PGs in granulosa cells (21). The decrease in granulosa cell cPLA 2 mes-senger RNA (mRNA) abundance in the presence of TGF␤ seemed to be the consequence of suppressed gene transcription and, interestingly, was more pronounced in cells at early stages of follicular development, when they were less differentiated and proliferatively more active (21). The reason(s) for this developmental dependency of TGF␤ action is not clear, nor is it known if the follicular stage-specific regulation of MADR2 expression by this growth factor is a determinant of the relative responsiveness of this signaling pathway to TGF␤. To date, the presence and regulation of MADR2 in the ovary by TGF␤ have not been examined.
The overall objective of the present studies was to demonstrate the presence of MADR2 in the hen granulosa cells and to study the regulation of its expression by TGF␤ in vitro during ovarian follicular development, particularly with reference to its possible association with the follicular stagedependent control of granulosa cell cPLA 2 gene transcription by the growth factor.

Materials and Methods Materials
Culture media reagents were purchased from Gibco/Bethesda Research Laboratories (Burlington, ON, Canada). Collagenase-1A and trypsin inhibitor (type II) were obtained from Sigma Chemical Co. (St. Louis, MO). The RNeasy Kit was purchased from Qiagen (Hildon, Germany). The Random Primed DNA Labeling Kit was from Boehringer Mannheim (Mannheim, Germany); ␣-32 P deoxycytidine triphosphate and the enhanced chemiluminescence Western blotting detection kit were purchased from Amersham (Oakville, ON, Canada). -probe blotting membrane, trans-blot supported nitrocellulose membrane, and the Bio-Rad protein assay kit were from Bio-Rad Laboratories (Hercules, CA). X-ray films were from Eastman Kodak Company (Rochester, NY). Recombinant human TGF␤ was purchased from Collaborative Research (Bedford, MA). The 2.8-kb hen cPLA 2 complementary DNA (cDNA) probe was a gift from Dr. Lih-Ling Lin (Genetics Institute, Cambridge, MA). The 1.4-kb human MADR2 and 1.2-kb human MADR1 probes have been described previously (3,22).

Isolation and culture of granulosa cells
White Leghorn hens in their first year of lay were obtained from a local egg farm and were individually caged in a windowless, air-conditioned room with a 14-h light, 10-h dark cycle. The birds had free access to feed and water. The time of ovulation was predicted from the time of the previous oviposition, on the basis that the former occurs 15-75 min from the latter. Approximately 10 -14 h before the expected time of ovulation, hens were killed by cervical dislocation, and the ovaries were excised and placed in ice-cold Medium 199 supplemented with HEPES (25 mm, pH 7.4). Follicles from two to three hens were grouped together according to the stage of development, as follows: the largest (F1), third largest (F3), and fifth and sixth largest (F5-6) preovulatory follicles and large white follicles (LWF). Granulosa cells from each group of follicles were isolated, as described previously (Asem et al., 1984) and dissociated by incubation (15 min at 37 C) in 2 ml Medium 199-HEPES containing collagenase (540 U) and trypsin inhibitor (0.2 mg/follicle). Dispersed granulosa cells (2 ϫ 10 5 ) were incubated in 16-mm tissue culture wells (Falcon, Becton Dickinson Labware, Lincoln Park, NJ) in 0.5 ml MEM, without phenol red but supplemented with l-glutamine (0.29 mg/ml), nonessential amino acids (0.1 mm), penicillin (100 U/ml), and streptomycin (100 g/ml), at 39 C under an atmosphere of 5% CO 2 and 95% air. After an incubation period of 3 h to allow cell plating, the medium was changed to one containing different test agents, and cells were cultured for up to 24 h. At the end of the culture period, the media were removed, and protein and RNA extracts were prepared. Cell viability, determined by the trypan blue dye exclusion test in both experimental groups (F1 and F5-6), was greater than 90% before and after culture.

Solubilized cell extracts and immunoblot analysis
Because MADR proteins translocate from the cytosol to the nucleus upon receptor activation (3,23), whole-cell lysates, rather than cytosolic fractions from granulosa cells, were used for Western blot analysis. Total cell protein extracts were prepared as follow: granulosa cells were sonicated (8 sec/cycle, 3 cycles) on ice in 10 mm HEPES buffer (pH7.4) containing 1 mm EGTA and 2 mm phenylmethylsulfonyl fluoride. The sonicates were stored at Ϫ20 C until electrophoretic analyses were performed. Protein concentration was determined using the Bio-Rad Protein Assay Method.
Equal amounts of proteins (60 -100 g per lane, depending on specific experiment) present in cell extracts were resolved by one-dimensional SDS-PAGE and electrophoretically transferred to nitrocellulose membrane. Membranes were blocked with 5% nonfat milk and subsequently incubated with polyclonal human MADR1 or MADR2 antibody (or preimmune sera for negative controls) diluted in 10 mm Tris-buffered saline (pH7.5) containing 5% milk. An enhanced chemiluminescence kit was used to visualize immunopositive proteins. For preparation of anti-MADR1 and MADR2 antibodies, the nonconserved domain (3) of each protein was expressed as a glutathione-S-transferase fusion protein in pGEX4T1 (Pharmacia). Rabbits were immunized with 1 mg of bacterially expressed protein, purified by glutathione-sepharose chromatography, using standard protocols (24).

RNA isolation and Northern analysis
Total RNA from granulosa cells was isolated by using an RNeasy Kit (Qiagen) and quantified spectrophotometrically at 260 nm. RNA samples (6 -12 g) were denatured at 55 C for 15 min in 45% formamide and 5.4% formaldehyde and electrophoresed at room temperature in 1% formaldehyde-agarose gel. Gels were stained with ethidium bromide to confirm equal loading of RNA samples (by comparing staining of 28S and 18S ribosomal bands). After transfer to a membrane, blots were UV-cross-linked and hybridized to 32 P-labeled hen cPLA 2 , human MADR1, and MADR2 cDNA probes. The probes were labeled using a Random Primed [ ␣-32 P] DNA Labeling Kit. Blots were washed twice (5 min/wash) at room temperature, 4 times (15 min/wash) at 65 C in 0.3 ϫ saline sodium citrate, 0.5% SDS, and subsequently exposed to x-ray film at Ϫ80 C. Densitometric analysis of mRNA and 28S ribosomal ribonucleic acid (rRNA) band was performed using an Image Analysis System from Bio-Rad Laboratories. Data were normalized by the respective 28S RNA and expressed as a percentage of the control (defined as 100%).

Statistical analysis
Follicles from two or three birds were used for each experiment. Results were expressed as the mean Ϯ sem of three to five experiments. Statistical analysis were carried out by ANOVA, and significant differences between treatment groups were determined by the Tukey test. Where required, data were transformed logarithmically, before statistical analysis, to remove heterogeneity of variance, as determined by Barlett's test. Statistical significance was inferred at P Ͻ 0.05.

In vivo expression of MADR2 in granulosa cells at different stages of follicular development
To determine whether MADR2 is expressed in granulosa cells in vivo and whether its cellular content is dependent on the stage of follicular development, granulosa cell layers from F1, F3, and F5-6 follicles (developing follicles; cells ranging from highly differentiated to proliferatively active, respectively) and LWF (follicles awaiting selection and recruitment into the developmental pool; undifferentiated cells) were isolated, and proteins were extracted for Western analysis. As shown in Fig. 1, MADR2 appeared as a 60-kDa protein in granulosa cells at all stages of follicular development examined. Expression of this protein was minimal in granulosa cells from LWF, highest in F5-6 granulosa cells, and decreased with follicular development (Fig. 1).

Regulation of MADR2 expression by TGF␤
MADR2 is coupled to activin and TGF␤ signaling systems (3,25) and has been suggested to function as a transcriptional coactivator in target cell gene transcription (4). Thus, we examined whether MADR2 expression in granulosa cells was regulated by TGF␤ in vitro at the transcriptional or translational level. MADR2 protein content in F1 granulosa cells was significantly increased by TGF␤ in a concentrationdependent manner (P Ͻ 0.05), reaching a level 300% above the control at 20 ng/ml (P Ͻ 0.05; Fig. 2), whereas the abundance of MADR2 transcript (ϳ3.6 kb) was increased six-fold by the growth factor (P Ͻ 0.05; Fig. 3). MADR1 (ϳ55 kDa), an MADR specifically phosphorylated in bone morphogenetic protein (BMP) signaling pathways (3), also was present in granulosa cells. It has a transcript size of 3.6 kb but was not significantly affected by the presence of TGF␤ at concentrations as high as 20 ng/ml (P Ͼ 0.05; see Figs. 3 and 5). In addition, the presence of MADR1 and MADR2 in the hen granulosa cell seems authentic, because no detectable signals were evident in immunoblots when the anti-MADR1 and anti-MADR2 antibodies were replaced with their respective preimmune antisera (data not shown). Interestingly, the basal MADR2 transcript level was highest early in the culture period (0.5 h) but significantly decreased with the duration of culture (Fig. 4). Addition of TGF␤ (20 ng/ml) to the cell cultures had no significant influence on granulosa cell MADR2 mRNA abundance in both F1 and F5-6 until the last time point examined (6 h), when the transcript levels were significantly elevated by 250% (Fig. 4a) and 500% (Fig. 4b), respectively. Thus, granulosa cell MADR2 is selectively regulated by TGF␤ at both the transcriptional and translational levels.

Follicular stage dependence of MADR2 response to TGF␤
The regulation of granulosa cell MADR2 expression by TGF␤ was dependent on the stage of follicular development. Though the growth factor significantly increased MADR2 mRNA abundance at 6 h (Fig. 4) and protein content at 12 h (Fig. 5) in granulosa cells in both F1 and F5-6 follicles (P Ͻ 0.05), MADR2 expression in the presence of TGF␤ (20 ng/ml) seemed to be more pronounced in the less differentiated cells (F5-6; Figs. 4 and 5). This is supported by the outcome of statistical analysis by ANOVA, which shows a significant interaction between follicular stage and TGF␤ effects (P Ͻ 0.05), presumably brought about by the greater stimulatory action of the growth factor at the F5-6 stage (Fig. 4). Similarly, MADR2 protein content in F5-6 granulosa cells in the presence of TGF␤ (20 ng/ml) was two times higher than in the F1 cells (Fig. 5). In contrast, TGF␤ had no significant influence on granulosa cell MADR1 protein content, irrespective of the stage of follicular development (P Ͼ 0.05; Fig.  5).

Reciprocal expression of MADR2 and cPLA 2 in granulosa cells cultured in the absence and presence of TGF␤
To investigate whether the TGF␤-induced changes in cPLA 2 expression in hen granulosa cells possibly could be associated with the up-regulation of MADR2, experiments were performed to compare concentration-dependent and temporal responses to TGF␤. In the absence of TGF␤, cPLA 2 mRNA abundance increased, whereas MADR2 mRNA levels decreased, with the duration of culture. The earliest significant changes were observed at the same time point (6 h; Fig.  4). Addition of TGF␤ to the cultures increased MADR2 mRNA abundance and decreased that of cPLA 2 , after 6 h of culture, when compared with their respective controls (Fig.  4). The divergent effects of TGF␤ on the expression of MADR2 and cPLA 2 both were concentration-dependent, with significant responses evident at the same concentration of the growth factor (20 ng/ml; Fig. 3). Moreover, both MADR2 and cPLA 2 responses to TGF␤ were follicular stage-dependent and were greater in the granulosa cells from the early stage of follicular development (Figs. 4 and  5). Thus, a reciprocal relationship seems to exist between cPLA 2 and MADR2 expression in granulosa cells that are responsive to TGF␤. Furthermore, this relationship is follicular stage-dependent.

Discussion
Cell signaling by TGF␤ is mediated by two types of transmembrane serine/threonine kinase receptors, types I and II (1,11,23). The type II receptor seems to be constitutively active and, upon ligand binding, forms a heteromeric complex with type I receptor, which activates the type I receptor by phosphorylation (10). Recently, MAD and MADR have been identified, in a variety of species, as important components of the signal transduction pathways of the TGF␤ superfamily (11,23). It has been demonstrated that MADR2 is rapidly phosphorylated by activation of TGF␤ signaling pathway (22) and redistributes from the cytoplasm to the nucleus, presumably for transcriptional activation of specific genes (26). In the present studies, we have demonstrated, for the first time, the presence of MADR2 protein (60 KDa) and transcript (3.6 Kb) in granulosa cells and, more importantly, its expression is up-regulated by TGF␤ at both the transcriptional and translational levels. This is evident by a time-and concentration-dependent increase in granulosa cell MADR2 mRNA abundance and protein content in the presence of TGF␤. Although significant TGF␤-induced MADR2 transcript level was noted after 6 h of exposure to the growth factor, an earlier response to TGF␤ could not be ruled out. Nonetheless, these findings suggest that, in addition to being phosphorylated and activated by TGF␤ receptor activation as has been shown in other systems (22), MADR2 also is up-regulated by TGF␤ and, thus, presumably acts as positive feedback to enhance its own actions. Interestingly, as shown in the time course study, MADR2 transcript levels, in granulosa cells maintained in growth factor-free conditions, decrease with the duration of culture and can be reversed by the addition of exogenous TGF␤ (Fig. 4, A and B), suggesting that the growth factor may be necessary for the maintenance of physiologically important cellular levels of MADR2 in vivo.
Previous studies from our laboratory have demonstrated that TGF␤ suppresses granulosa cell cPLA 2 and cyclooxygenase II mRNA abundance and PG production and increases urokinase plasminogen activator activity in vitro. These responses were more pronounced at early stages of follicular development (19,20,27). In the present study, we have demonstrated that MADR2 is up-regulated by TGF␤ in vitro, with maximum response observed in granulosa cells from F5-6 follicles. Whereas TGF␤ is expressed in hen granulosa and theca cells throughout follicular development (14), our present in vivo studies indicate that MADR2 abundance was minimal in granulosa cells from LWF, highest in F5-6 granulosa cells, and decreased with follicular maturation. These findings raise the interesting possibility that the follicular stage-specific high abundance of MADR2 may be an important determinant for the relative responsiveness of the signaling pathway to the growth factor and that homologous up-regulation of MADR2 may be physiologically important. The physiological significance for the low MADR2 abundance in LWF granulosa cells observed in the present studies is not clear. It has been shown that TGF␤ is capable of inducing apoptosis in other systems (28). It is possible that the suppression of MADR2 expression in vivo at this follicular stage may be one of the mechanisms in place to prevent these cells from undergoing cell death until the fate of these follicles (i.e. development vs. atresia) has been determined.
MADR2 is believed to be a potential transcriptional regulator, based on its ability to accumulate in the nucleus in response to TGF␤ and the observation that MADR1 can act as transcriptional activator (4). Furthermore, MAD3, another member of MAD family, seems to be able to potentiate a transcriptional response to TGF␤ after its overexpression (29). However, the physiological target gene for MADR2 in the granulosa cell is not known, although it is tempting to speculate that the down-regulation of cPLA 2 by TGF␤ may be related to the follicular stage-dependent expression of MADR2. This notion is consistent with our earlier observation that the suppression of granulosa cell cPLA 2 level by the growth factor resides at the level of gene transcription and is not a result of decrease in mRNA stability (Li et al., unpublished data). In addition, it is of interest to note, from the present studies, that alterations in MADR2 expression seem tightly coupled to reciprocal changes in cPLA 2 mRNA abundance, as evidenced by increased basal MADR2 mRNA and protein levels, whereas cPLA 2 transcript abundance decreased with follicular development. Furthermore, basal MADR2 expression decreased, whereas that of cPLA 2 increased, with the duration of culture. Although the precise mechanism for the increase in cPLA 2 mRNA and protein content is not known, one possibility is that granulosa cell cPLA 2 expression is suppressed by endogenous inhibitory factors (e.g. TGF␤) and their signaling machinery to maintain appropriate levels in vivo. When these inhibitory factors were removed by culturing cells in serum-free medium, cPLA 2 expression spontaneously increased. This notion is consistent with our present observation that the increase in basal cPLA 2 expression, with increased duration of culture, is accompanied by reciprocal decreases in mRNA abundance and protein content of MADR2. Finally, the reciprocal MADR2 and cPLA 2 response to TGF␤ followed a similar time course and concentration-and follicular stage-dependency. Further experiments, including MADR2 knock-out with antisense, are required to determine directly whether and what precise relationship exists between MADR2 signaling and cPLA 2 expression in the granulosa cell.
It has been demonstrated that MADR1 shares considerable sequence homology with MADR2 but is not involved in the TGF␤ signaling pathway (3). Although significant levels of MADR1 mRNA and protein were detected in granulosa cells in the present studies, they were not significantly affected by the presence of TGF␤, indicating that the response in MADR2 expression to the growth factor was specific. It is known also that MADR4 (DPC4), another MADR protein, is involved in TGF␤ signaling (30). Whether MADR4 also is upregulated by TGF␤ and can act as a transcriptional coregulator in the regulation of gene expression remains to be determined.
In summary, we have demonstrated, for the first time, the presence of MADR1 and MADR2 in hen granulosa cells. The expression of MADR2, but not of MADR1, was up-regulated by TGF␤ in vitro in a concentration-and time-dependent manner. Granulosa cell MADR2 expression was maximal during early stages of follicular development, when the cells are proliferatively most active and the cPLA 2 system most responsive to TGF␤. These findings are consistent with the hypothesis that MADR2 expression is autoregulated and that this regulation may be an important determinant in the follicular stage-specific responsiveness of the cells to TGF␤.