Follicle-stimulating Hormone Regulates Pro-apoptotic Protein Bcl-2-interacting Mediator of Cell Death-Extra Long (BimEL)-induced Porcine Granulosa Cell Apoptosis*

Background: In the mammalian ovary, 99% of follicles are removed through follicular atresia caused by granulosa cell apoptosis. Results: BimEL can induce porcine granulosa cell apoptosis, and its expression is regulated by FSH via the PI3K/Akt/FoxO3a pathway. Conclusion: The pro-apoptotic protein BimEL is involved in porcine granulosa cell apoptosis. Significance: This provides novel insights into the molecular mechanisms underlying follicular atresia. The pro-apoptotic protein Bim (B-cell lymphoma-2 (Bcl-2)-interacting modulator of cell death) has recently been identified and shown to promote cell death in response to several stimuli. In this report, we investigated the role of Bim in porcine follicular atresia. Initially, Bim cDNA was cloned and characterized from porcine ovarian tissue. Porcine Bim had three alternative splicing variants (Bim-extra long, Bim-long, and Bim-short), all containing the consensus Bcl-2 homology 3 domain. We then found the Bim-extra long (BimEL) protein, the most abundant isoform of Bim, was strongly expressed and co-localized with apoptotic (TUNEL-positive) granulosa cells from porcine atretic follicles. Furthermore, overexpression of BimEL triggered apoptosis in granulosa cells. In primary granulosa cell cultures under basal conditions, we observed that BimEL expression was dampened by treatment with follicle-stimulating hormone (FSH). The role of the PI3K/Akt pathway in the regulation of repression was clarified by the use of the PI3K inhibitor, LY294002, and by transfection with Akt siRNA. Forkhead Box Protein O3a (FoxO3a), a well defined transcriptional activator of Bim, was phosphorylated at Ser-253 and inactivated after FSH stimulation. Also, FSH abolished FoxO3a nuclear accumulation in response to LY294002. Finally, chromatin immunoprecipitation assays demonstrated that FoxO3a directly bound and activated the bim promoter. Taken together, we conclude that BimEL induces porcine granulosa cell apoptosis during follicular atresia, and its expression is regulated by FSH via the PI3K/Akt/FoxO3a pathway.

The pro-apoptotic protein Bim (B-cell lymphoma-2 (Bcl-2)-interacting modulator of cell death) has recently been identified and shown to promote cell death in response to several stimuli. In this report, we investigated the role of Bim in porcine follicular atresia. Initially, Bim cDNA was cloned and characterized from porcine ovarian tissue. Porcine Bim had three alternative splicing variants (Bim-extra long, Bim-long, and Bim-short), all containing the consensus Bcl-2 homology 3 domain. We then found the Bim-extra long (Bim EL ) protein, the most abundant isoform of Bim, was strongly expressed and co-localized with apoptotic (TUNEL-positive) granulosa cells from porcine atretic follicles. Furthermore, overexpression of Bim EL triggered apoptosis in granulosa cells. In primary granulosa cell cultures under basal conditions, we observed that Bim EL expression was dampened by treatment with follicle-stimulating hormone (FSH). The role of the PI3K/Akt pathway in the regulation of repression was clarified by the use of the PI3K inhibitor, LY294002, and by transfection with Akt siRNA. Forkhead Box Protein O3a (FoxO3a), a well defined transcriptional activator of Bim, was phosphorylated at Ser-253 and inactivated after FSH stimulation. Also, FSH abolished FoxO3a nuclear accumulation in response to LY294002. Finally, chromatin immunoprecipitation assays demonstrated that FoxO3a directly bound and activated the bim promoter. Taken together, we conclude that Bim EL induces porcine granulosa cell apoptosis during follicular atresia, and its expression is regulated by FSH via the PI3K/Akt/ FoxO3a pathway.
During oogenesis, Ͼ99% of developing follicles undergoes atresia (1)(2)(3)(4). It is widely accepted that the viability of pri-mordial and primary follicles is determined mainly by survival factors derived from the oocyte, whereas the relative expression level of tumor suppressors, apoptotic proteins, and survival factors in granulosa cells determines whether an ovarian follicle will grow or undergo atresia in the late preantral stage of the follicles (5). FSH functions as a critical survival signal for antral follicles by blocking apoptosis in granulosa cells (6); however, molecules acting as signals to suppress granulosa cell apoptosis and promote growth during this particular stage of follicular development remain largely undetermined.
The B cell lymphoma 2 (Bcl-2) family proteins are considered major regulators of apoptosis (7). These proteins share homology in at least one of four Bcl-2 homology (BH) 2 domains and constitute three subfamilies, including Bcl-2 anti-apoptotic proteins (e.g. Bcl-2 and Bcl-xl), Bax pro-apoptotic proteins (e.g. Bax and Bak), and BH3-only proteins (e.g. Bim, Bmf, and Puma) (8). According to data generated from different genetic models, the balance between the proand anti-apoptotic Bcl-2 members presumptively determines the fate of a follicle (9 -12). Therefore, the "Bcl-2/Bax balance" is proposed to be responsible for the regulation of granulosa cell apoptosis and survival (13). However, there is limited understanding of the involvement of BH3-only proteins, such as Bim, in this process.
The protein Bim (also known as Bcl-2-related ovarian death gene, Bod) was originally identified as a Bcl-2-interacting protein by screening a bacteriophage cDNA expression library constructed from a mouse thymic lymphoma (14). Bim promotes apoptosis by binding with high affinity to anti-apoptotic Bcl-2 family members, thereby inducing mitochondrial release of cytochrome c, which subsequently activates caspase-9 and caspase-3 and death effector molecules (15). More than 12 additional Bim mRNA splicing variants have been identified in a wide range of human and rodent tissues (16 -19), but most tissues express one pre-dominant isoform of Bim as determined by Western blotting analysis, termed Bim EL . Studies on the transcriptional regulation of Bim have mainly focused on FoxO3a, a member of the Forkhead family of transcriptional regulators that directly stimulates Bim transcription (20). Bim proteins were also identified in granulosa cells of primordial, primary, secondary, and mature murine follicles (21); however, their cellular functions and relative contribution to granulosa cell apoptosis remain largely unknown.
It prompted us to investigate the putative role of Bim in porcine granulosa cell apoptosis. Porcine Bim homologues were cloned as the first step toward elucidating differences in Bim EL expression between atretic and healthy follicles. An overexpression model was created to determine the pro-apoptotic effect of Bim EL in transfected granulosa cells. We then demonstrated that Bim EL was down-regulated after FSH treatment in primary cultured granulosa cells. Finally, the mechanism of FSH-regulated Bim EL expression was determined.

EXPERIMENTAL PROCEDURES
Unless otherwise specified, all chemicals used in this study were purchased from Sigma.
Classification of Follicles and Recovery of Granulosa Cells-Porcine ovaries were collected at a local abattoir and transported to the laboratory in a vacuum flask (30 -35°C) containing sterile physiological saline within 2-3 h of collection. In the laboratory, ovaries were washed twice with sterile physiological saline (37°C) containing 100 IU/liter penicillin and 50 mg/liter streptomycin. Follicles were classified as healthy or atretic, according to previously established morphological criteria (22). Briefly, healthy follicles had vascularized theca interna and clear amber follicular fluid with no debris. Follicles lacking any of these criteria were classified as atretic. The slightly atretic and atretic follicles had gray theca interna and flocculent follicular fluid in varying degrees.
Granulosa cells from follicles of different health status were isolated by puncturing follicles with a 25-gauge hypodermic needle and gently expelling the cells into a culture dish containing M199 medium. The cumulus-oocyte complex and ovarian tissue were discarded under a stereo microscope. Granulosa cells were then harvested by centrifuging the media (500 ϫ g for 10 min). Pooled cells were washed three times in PBS for immunoblotting.
Granulosa Cell Culture Conditions and Treatment-For experiments investigating FSH-mediated effects, primary cultures of granulosa cells from healthy follicles (2-5 mm in diameter) were selected, largely due to the fact that the follicles in this phase are highly responsive to FSH in pig (23). Briefly, viable granulosa cells (assessed by trypan blue exclusion) were cultured in 6-well plates (5 ϫ 10 6 /well) with 2 ml of DMEM/Ham's F-12 (Invitrogen) and supplemented with 100 units/ml penicillin and 50 mg/ml streptomycin without (control) or with 0.01, 0.1, and 1 IU/ml porcine FSH (Sioux Biochemical, Inc., Sioux Center, IA). Cells were incubated in a humidified 5% CO 2 atmosphere incubator at 37°C for the indicated time periods. When examining FoxO3a phosphorylation, LY294002 (20 M) was added 30 min before FSH treatment. Granulosa cells from each treatment were collected from the wells and snap-frozen for subsequent analysis by immunoblotting or caspase-3 activity assay.
Caspase-3 Activity Assay-The caspase-3 activity was used to estimate apoptotic response and determined using caspase-3 activity assay kits (Applygen, Beijing, China). After treatment, granulosa cells were pelleted by centrifugation (1000 ϫ g, 10 min). The cell pellet was lysed in 80 l of lysis buffer for 20 min on ice. After the lysate was centrifuged at 13,000 ϫ g for 15 min, the supernatant was collected, and the protein concentration was determined with BCA protein assay kit (CW BioTech, Beijing, China). Aliquots (20 g of protein) of the supernatant in triplicate were transferred to the 96-well plate and incubated in the dark with 100 l of buffer containing caspase substrates at 37°C for 4 h. Absorbance at 405 nm was measured using a microplate reader (Bio-Rad).
RNA Isolation-Total cellular RNA (tcRNA) from various tissues (porcine lung, liver, and ovary and murine liver and ovary) was extracted using an RNAprep pure tissue kit (TIAN-GEN Biotech, Beijing, China) according to the manufacturer's instructions. Total RNA content was determined by spectrophotometry (260 nm).
Rapid Amplification of cDNA 3Ј Ends-The tcRNA (500 ng) from porcine ovarian tissue was used to convert mRNAs into cDNAs, using a 3Ј-full rapid amplification of cDNA 3Ј end core set version 2.0 kit (Takara Biotech, Dalian, China), which included avian myeloblastosis virus reverse transcriptase and a universal oligo(dT) containing-adapter primer. The PCR was conducted in accordance with manufacturer's instructions. The Takara LA Taq polymerase was used in rapid amplification of cDNA 3Ј end-PCR. An aliquot of the first PCR (outer PCR) was used for a subsequent nested PCR (inner PCR). Amplified cDNA was gel-purified and subcloned into the pBST-II vector (TIANGEN Biotech, Beijing, China) and sequenced (SUN Biotech, Beijing, China).
Reverse Transcription-PCR (RT-PCR)-In each sample, tcRNA (500 ng) was used for cDNA synthesis. In the first step, secondary structures were removed, and an oligo(dT) 18 primer was annealed to the mRNA by heating the sample at 70°C for 5 min. The reaction was then quenched rapidly on ice, and 5ϫ RT buffer, 10 mM dNTPs, 2 mM dithiothreitol, 20 units of RNase inhibitor (TaKaRa Biotech, Dalian, China), and 200 units of Moloney murine leukemia virus reverse transcriptase (TIAN-GEN Biotech, Beijing, China) were added (total reaction mixture volume, 25 l). This mixture was heated at 42°C for 60 min and then at 70°C for 15 min to terminate the reaction. Subsequent PCR analysis was carried out with 1 l of the resulting cDNA, and products were separated by electrophoresis on a 1.5% agarose gel. For analysis of Bim isoform expression profiles, bands were purified and sequenced.
Multiple Sequence Alignments-Multiple sequence alignments were done with the ClustalW algorithm in the BCM Search Launcher.
Construction of Fluorescent Protein Expression Vectors and Transfection-Bim EL and FoxO3a construct was generated by subcloning PCR-amplified full-length porcine Bim EL and FoxO3a cDNA into pEGFP-N1 (Clontech). The reading frame of Bim EL and FoxO3a cDNA was connected with enhanced GFP to obtain a recombinant pEGFP-N1-Bim EL and pEGFP-N1-FoxO3a plasmid. The plasmid was propagated using DH5␣ (TIANGEN Biotech, Beijing, China) as the host strain, and purified with an EndoFree Plasmid maxi kit (Qiagen Inc., Hilden, Germany), and sequenced to confirm the open reading frame.
Transient transfection of porcine granulosa cells with plasmids was performed using Lipofectamine TM 2000 (Invitrogen), according to the manufacturer's instructions. Expression of the fusion construct was evaluated by fluorescence microscopy after transfection. Meanwhile, at 2 and 8 h after transfection, cells were stained Hoechst 33342 to evaluate nuclear morphology. At the end of the experiments, floating cells were collected from the medium by centrifugation and combined to attached cells harvested by trypsin/EDTA treatment. Cells were then washed in PBS and used for Western immunoblot.
Western Immunoblot-Cells were lysed in Laemmli sample buffer (Bio-Rad). Equal amounts of proteins (15 g/lane) were separated by SDS-PAGE (7.5% acrylamide running gel) and transferred to a nitrocellulose membrane (BioTraceNT, Pall Corp.). Nonspecific binding to the membrane was blocked with 5% nonfat milk in TBS-T (10 mM Tris (pH 7.5), 150 mM NaCl, and 0.1% Tween 20) for 1 h at room temperature. Membranes were then incubated with primary antibodies (4°C, overnight), washed in Tris-buffered saline containing 0.1% Tween, and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature. Antibodies against Bim (C34C5), FoxO3a (75D8), and phospho-FoxO3a were all purchased from Cell Signaling Technology (Beverly, MA). Anti-GFP, anti-actin, and anti-histone 3.1 antibodies were obtained from Abmart Biotechnology (Shanghai, China). HRP-conjugated anti-rabbit antibody was obtained from Zhongshan Biotechnology (Beijing, China). The protein bands were visualized by enhanced chemiluminescence detection reagents (Applygen Technologies Inc., Beijing, China) and X-Omat BT film (Eastman Kodak Co.), according to the manufacturer's instructions. The films were digitized, and densitometry analysis was performed with ImageJ 1.44p software (National Institutes of Health). The relative intensity of the bands was quantified and normalized to the respective loading control. Nuclear and cytoplasmic protein extracts were prepared using the ProteoJET TM cytoplasmic and nuclear protein extraction kit K0311 (Fermentas, Canada).
Histology-Porcine ovaries were fixed overnight in 4% phosphate-buffered formaldehyde at 4°C for 7 days and then embedded in paraffin. Randomly selected sections (5 m each) were mounted on poly-L-lysine-coated slides (Zhongshan Biotechnology, Beijing, China) for subsequent immunohistochemical and TUNEL analyses. Prior to Bim localization, sections were deparaffinized by washing with xylene and then rehydrated in decreasing concentrations of ethanol (100 to 50% ethanol) and washed in PBS. To block endogenous peroxidase activity, sections were incubated for 10 min in 0.5% hydrogen peroxide. After washing with PBS, sections were permeabilized with 0.1% Triton X-100 in PBS for 10 min. Nonspecific binding was blocked by incubation with 10% inactivated goat serum. A monoclonal rabbit Bim (C34C5) antibody (1:200; Cell Signaling Technology, Beverly, MA) was used to detect Bim EL protein expression. To eliminate nonspecific binding, nonimmune goat serum was added to the primary antibody. Sections were incubated overnight at 4°C. The next day, after washing with PBS, the HRP-conjugated anti-rabbit antibody (1:100, Zhongshan Biotechnology, Beijing, China) was applied for 1 h at room temperature. The binding of primary antibody was visualized using diaminobenzidine for 3-5 min. Sections were subsequently washed with twice-distilled water. Finally, sections were rinsed with water, dehydrated in ethanol and toluene, and mounted in Neutral Balsam. Bim labeling was examined using a Leica microscope, and images were recorded (Leica DC 200 digital camera; Leica, Wetzlar, Germany). As a negative control, normal rabbit serum was applied to some sections in lieu of specific antiserum.
Apoptosis was analyzed by TUNEL (in situ cell death detection kit, Roche Applied Science) according to the manufacturer's protocol. Briefly, after deparaffinization and rehydration, sections were incubated with proteinase-K (20 g/ml) for 15 min at room temperature, quenched in 3% H 2 O 2 in PBS for 10 min (to block endogenous peroxidase), incubated in a humidified chamber with equilibration buffer for 5 min, and finally incubated with terminal deoxynucleotidyltransferase for 1 h at 37°C. Negative control slides were incubated as described above, without the final addition of terminal deoxynucleotidyltransferase. After TUNEL reactions were complete, slides were washed in PBS, sealed under coverslips with nail varnish, and examined under Leica fluorescence microscopy.
Chromatin Immunoprecipitation Assays (ChIP)-Porcine granulosa cells were cultured in 15-cm dishes and transfected using GFP-FoxO3a plasmid as indicated above. To transfer fusion protein into the nucleus, 4-h post-transfection cells were incubated in the presence of LY294002 (20 M) for 2 h. Then the transiently transfected cells were processed for ChIP assay according to manufacturer's directions in the chromatin immunoprecipitation assay kit (Upstate). Briefly, cells were cross-linked in 1% formaldehyde solution for 10 min at room temperature. Cells were lysed in 500 l of SDS buffer and sonicated to generate 200 -1000 bp of DNA fragments. The sheared chromatin concentration was measured; about 10% of the chromatin was kept as an input. The rest of the chromatin was incubated with a GFP antibody (Abmart Biotechnology, Shanghai, China) for immunoprecipitation. Upon DNA isolation, amplification of bim promoter DNA was assessed by PCR using the following primers: forward (Ϫ218/Ϫ199), 5Ј-GATGGGTGCGCACTGAGCCA-3Ј; reverse (Ϫ106/Ϫ87), 5Ј-GCGTGCAGGCTCTGACAGGTA-3Ј. Analysis of the PCR products was performed on a standard 2% (w/v) agarose gel, by electrophoresis in Tris acetate/EDTA buffer.
Statistical Analyses-Data of the amounts of Bim EL , FoxO3a proteins, and caspase-3 activity were analyzed using a general linear model and one-way post hoc test using the MIXED procedure models of SAS (version 9.1, SAS Institute Inc., Cary, NC). Differences were considered statistically significant when p Ͻ 0.05.

RESULTS
Complementary DNA Sequence and Structure of Procine Bim-Based on a partial cDNA of porcine Bim EL (GenBank TM accession number. AJ606302), a 552-bp open reading frame (ORF) of Bim EL cDNA (GenBank TM accession number FJ825619) was obtained from porcine ovarian tissue. The porcine Bim EL putative mature protein contains 183 amino acid residues and has amino acid sequence homology of 87.3 and 84.2% with its human and mouse counterparts, respectively (GenBank TM accession numbers AAC39593 and AAC40029) (Fig. 1A). BLAST results indicated that murine and porcine Bim EL have similar characteristics. The porcine Bim EL has five exons (124, 168, 90, 110, and 60 bp), contains a conserved BH3 domain (LEDIGD; Fig. 1A, boxed), lacks the BH1, BH2, and BH4 domains present in channel-forming BCL-2 family proteins, and possesses a dynein-binding domain (DKSTQT; Fig. 1A, boxed). Unlike the murine Bim EL , porcine Bim EL did not have a COOH-terminal hydrophobic region (Fig.  1A, overlined). One antibody against both murine and porcine Bim EL protein was used for Western blot analysis, and it recognized a 25-kDa protein from porcine ovary extracts and a 27-kDa protein from murine ovary tissue (Fig. 1B). The difference in size may be explained by the fact that porcine Bim EL lacked the hydrophobic region.
The expression profiles of Bim mRNA in porcine lung, liver, and ovary were determined using RT-PCR with a primer pair that spanned the entire Bim coding sequence. Positive controls were tcRNA isolated from murine liver and ovary. As shown in Fig. 1C, three distinct Bim mRNA products were detected in the porcine lung, liver, and ovary. Nucleotide sequence analysis indicated that the two smaller products were splicing variants from porcine Bim EL (Fig. 1D). These isoforms had a similar splicing mode with that of the mouse Bim EL and were subsequently named porcine Bim L and porcine Bim S (GenBank TM accession numbers FJ825620 and FJ825621), respectively.
Bim EL Expression Correlates with Atretic Follicles and Apoptotic Granulosa Cells-To identify the hypothesis that Bim protein was present and coincident with follicular atresia in vivo, we applied double labeling with TUNEL staining and immunohistochemical staining using antibody against Bim EL FIGURE 1. Complementary DNA sequence and structure of porcine Bim. A, deduced amino acid sequence for porcine Bim EL aligned to its human and murine counterparts. Identical amino acids are indicated by asterisks, and amino acids with conserved similarities are indicated by dots or colons. The BH3 and dynein-binding domains are boxed. The COOH-terminal hydrophobic region is overlined. The amino acid numbers are indicated on the right. B, antibody that recognizes both murine and porcine Bim EL proteins was used to conduct Western blotting on samples from murine and porcine ovary tissue. C, identification of multiple RT-PCR products of porcine Bim. The coding region of Bim cDNA was amplified by RT-PCR using total RNA from porcine lung, liver, and ovary and murine liver and ovary, using animal-specific primers. The PCR products were subjected to agarose gel electrophoresis and photographed. The segment size is marked on both sides. D, genomic organization of porcine Bim isoforms.
(our following Western blot experiment showed that the Bim protein recognized by the antibody in our immunohistochemistry assay was Bim EL ). Atretic follicles were evident by the presence of TUNEL-positive granulosa cells, uneven granulosa cell layers, and exfoliating granulosa cells that could occasionally be detected within the antral cavity ( Fig. 2A, panel a1). An adjacent section of the same follicle clearly showed that Bim EL immunostaining was concentrated in both granulosa and theca cells, especially in the inner surface of the granulosa layer ( Fig.  2A, panel a2). In addition, moderate TUNEL staining and Bim EL immunostaining were shown in slightly atretic follicles ( Fig. 2A, panels b1 and b2). However, healthy follicles that had a well organized and intact granulosa cell layer showed negative Bim EL immunostaining and TUNEL staining ( Fig. 2A, panels c1  and c2).
To further verify the results from immunohistochemistry, the relative abundance of the Bim EL protein in isolated granulosa cells was quantified by Western blot. Granulosa cells were isolated from freshly procured porcine follicles and classified as healthy, slightly atretic, or atretic follicles according to morphological criteria (see "Experimental Procedures"). As shown in Fig. 2B, granulosa cell Bim EL protein expression was indistinguishable in the healthy follicles and increased as atresia progressed. The expression of Bim L and Bim S , the two other major isoforms of Bim, was not detected in our present Western blot experimental system even with a longer exposure time (data not shown).
Overexpression of Porcine Bim EL Induces Granulosa Cells Apoptosis in Vitro-In light of the positive correlation observed between Bim EL immunoreactivity and granulosa cell apoptosis, further experiments were done to investigate the proapoptotic potential of Bim EL protein in granulosa cells in vitro. Cultured granulosa cells were transiently transfected with an empty EGFP vector or pEGFP-N1-Bim EL as a control. Eight hours after transfection, green fluorescence for all constructs was detected in ϳ10% of transfected cells (data not shown). At 2 h post-transfection, pEGFP-N1-Bim EL -transfected cells exhibited green fluorescence, which was found exclusively in the cytosol (Fig. 3A, panel a). At 8 h post-transfection, fluorescence microscopic examination showed that the majority of granulosa cells overexpressing Bim EL exhibited cytomorphological alterations typical for apoptosis, including nuclear fragmentation, cell rounding and shrinkage, plasma membrane blebbing, and apoptotic body formation (Fig. 3A, panel b) (24). In contrast, cells expressing only GFP (control) displayed normal appearing morphology and remained healthy and flat, with the green fluorescence signal located both in the nucleus and cytosol at all times (Fig. 3A, panel c). Western blot analysis of the transfected cells showed that the exogenously expressed Bim EL protein levels were much higher than that seen with endogenous Bim EL expression (Fig. 3B). Quantitative analysis of apoptotic cells was also assessed by caspase-3 activity assay. Overexpression of EGFP-Bim EL caused a 7-fold increase in caspase-3 activity (Fig. 3C), indicating that Bim EL can induce porcine granulosa cells apoptosis in vitro.
Bim EL Expression in Granulosa Cell Is Suppressed by FSH in Vitro-FSH stimulation is the most critical for the survival of antral follicles during follicle selection and development. FSH withdrawal induces granulosa cell apoptosis and follicular atresia (6). To determine whether Bim EL expression was associated with FSH, cultured granulosa cells were deprived of FSH and serum for 24 h to induce apoptosis. By immunoblot analysis, high amounts of Bim EL proteins were observed within 24 h in cells cultured in FSH and serum-free medium. A decrease in Bim EL protein was evident after 24 h of FSH stimulation (Fig.  4A). However, the dose-dependent manner was not observed. The 0.01 IU/ml dose of FSH already yielded maximal protection and thus was the concentration used in subsequent experiments. Next, the time course of repression of Bim EL protein by FSH was studied. The primary granulosa cells were cultured in FSH and serum-free medium analyzed as unstimulated (0-h time point) or were analyzed following stimulation with FSH (0.01 IU/ml) or not for 24 or 48 h. In unstimulated cells, the levels of Bim EL protein were indistinguishable (Fig. 4C) and consistent with the results from follicles (Fig. 2B). There was a dramatic increase in the amount of Bim EL protein at 24 h (5.1fold) and 48 h (6.7-fold) following serum and FSH-free incubation. When granulosa cells were cultured in medium contained FSH (0.01 IU/ml), the Bim EL protein levels increased only 1.7and 2.9-fold (24 and 48 h, respectively) than in control (Fig. 4C). Thus, FSH could counteract the induction of Bim EL after serum deprivation of granulosa cells isolated from FSH-dependent healthy follicles. Also, Bim EL protein seemed to migrate with several additional bands, indicating that Bim EL could be phos-phorylated by FSH-signaling pathways. In accordance with this, as analyzed by caspase-3 activity assay, there was a direct correlation between Bim EL expression levels and the extent of FSH-regulated apoptosis (Fig. 4, B and D).

PI3K/Akt/FoxO3a Pathway Is Involved in FSH Regulation of Bim EL Expression-The intracellular signaling molecules that mediate FSH-induced granulosa cell survival remain unclear.
Accumulating evidence suggests that the PI3K pathway can be activated by FSH signals (25) and subsequently provide a strong survival signal to granulosa cells by partially regulating the activities of the Bcl-2 family of proteins (26). To explore the possibility that FSH activation of PI3K plays a role in repressing Bim expression, in vitro cultured primary granulosa cells were pretreated with LY294002, a specific pharmacological inhibitor for PI3K, followed by FSH stimulation in serum-free medium. The results suggested that the repression effect of FSH on Bim EL expression was abolished by PI3K inhibitor at concentrations known to efficiently block PI3K activity (Fig. 5A).
To further confirm the involvement of PI3K pathway in regulating Bim EL expression, siRNA targeting Akt, a downstream substrate of PI3K, was used to transfect granulosa cells. By using two validated Akt siRNA target sequences, we were able to knock down Akt expression in granulosa cells by 75%. Nonsilencing siCTR had no effect on Akt expression. Akt knockdown using siRNA also significantly increased Bim EL expression (Fig. 5B). These data clearly show that Akt protein is required for Bim EL expression in granulosa cells.
It was recently reported that the Forkhead transcription factor, FoxO3a, the major downstream effector of PI3K/Akt signaling pathway, can bind to the bim promoter region and con-trol Bim expression at the transcriptional level (27,28). FoxO3a activity and nucleus translocalization is inhibited by PI3K-mediated phosphorylation at three sites (Thr-32, Ser-253, and Ser-315) (29). We hypothesized that in porcine granulosa cells, FSH activates PI3K/Akt, which in turn triggers phosphorylation of FoxO3a and ultimately prevents FoxO3a from inducing Bim expression. Time course experiments demonstrated that FoxO3a (Ser-253) phosphorylation increased dramatically  The graph demonstrates the results of the densitometric analysis of Bim EL protein levels normalized against loading controls (arbitrary units, no FSH and LY294002 ϭ 1). B, effect of Akt siRNA on Bim EL expression. Granulosa cells were transfected with Akt siRNA (siAkt) or siCTR for 36 h and then were harvested and analyzed by immunoblotting. C, Western blot analysis was completed to determine whether FSH affected FoxO3a phosphorylation status. Granulosa cells were preincubated for 30 min with 20 M LY294002 or not and then were harvested at specific times after FSH treatment (15-90 min). Phosphorylated (p) Foxo3a and total Foxo3a expression was assessed by immunoblotting. The graph demonstrates the ratio of the densitometric analysis of phosphorylated forms of FoxO3a proteins normalized against its total forms (arbitrary units, 0 h ϭ 1, respectively). D, FSH regulates FoxO3a protein subcellular localization. Granulosa cells plated on glass bottom 35-mm dishes were 80% confluent when transfected with pEGFP-N1 (panel a, control) or pEGFP-N1-FoxO3a (panels b and c) constructs incubated with or without FSH. At 6 h post-transfection, cells were then stained with Hoechst 33342 to visualize nuclei, and the subcellular localization of FoxO3a was monitored by GFP fluorescence. E, Western blot (WB) analysis of fusion protein expression in granulosa cells transfected with a full-length Bim EL -GFP construct or mock vector. Transfected granulosa cells lysates were subjected to SDS-PAGE and immunoblotting using the anti-GFP antibody. Molecular masses of the protein bands are shown on the left side. The graph demonstrates the results of the densitometric analysis of Bim EL protein levels normalized against loading controls (arbitrary units, GFP ϭ 1). F, FoxO3a is up-regulated in granulosa cells from atretic follicles. Isolated granulosa cell lysates, obtained from follicle of different health status (H, healthy follicle; SA, slightly atretic follicle; A, atretic follicle), were subjected to SDS-PAGE and immunoblotting using anti-FoxO3a antibodies. The graph demonstrates the results of the densitometric analysis of FoxO3a protein levels normalized against loading controls (arbitrary units, H ϭ 1). G, immunoblots showing the expression of FoxO3a in cytosolic and nuclear extracts of granulosa cells from the healthy and atretic follicles. Actin and histone 3.1 were used as fractionation controls, respectively. The graph demonstrates the results of the densitometric analysis of FoxO3a protein levels normalized against loading controls (arbitrary units, H ϭ 1). All values represent means Ϯ S.E. of three experiments. *, p Ͻ 0.05; **, p Ͻ 0.01. beginning at 15 min after FSH stimulation. The ratio of p-FoxO3a/total FoxO3a (as measured by densitometry) increased more than 2-fold at 15 and 30 min and 3-fold at 45 and 90 min compared with 0 min. However, this induction was consistently blocked by concomitant LY294002 (20 M) treatment (Fig. 5C).
Having demonstrated FSH stimulation of FoxO3a phosphorylation, we next sought to determine whether FSH regulated the intracellular localization of FoxO3a protein by transfecting cultured granulosa cells with plasmids expressing GFP or FoxO3a-GFP fusion proteins followed by LY294002 (20 M) treatment. Fluorescence microscope observations showed cells have diffuse fluorescence in both the nucleus and cytoplasm when granulosa cells were transfected with the control GFPonly vector as expected (Fig. 5D, panel a). When granulosa cells were transfected with a FoxO3a-GFP plasmid but left FSHuntreated, the fusion protein was largely in the nucleus (Fig. 5D,  panel b). In contrast, treatment of these cells with FSH (0.01 IU/ml) led to relocalization of fusion protein to the cytoplasm, although a portion was still present in the nucleus (Fig. 5D,  panel c). Western blot analysis showed that Bim EL protein in the FoxO3a-GFP plasmid-transfected cells was about 1.83-fold increased as compared with control cells (Fig. 5E). This result indicated that LY294002 activated FoxO3a protein and sequestered it in the nucleus and thereby induced Bim EL expression, which can be reversed by FSH treatment.
To provide in vivo evidence that the FoxO3a protein was involved in follicular atresia, the relative abundance of the FoxO3a protein in isolated granulosa cells from different healthy follicles was quantified. We showed that the amount of the FoxO3a protein was markedly increased in granulosa cells as atresia progressed (4.58-and 7.26-fold higher in lightly atretic and atretic follicles than in healthy follicles, respectively) (Fig. 5F). In addition, the relative cytoplasmic and nuclear distributions of FoxO3a protein were also evaluated in granulosa cells from healthy and atretic follicles. As shown in Fig. 5G, FoxO3a protein in the cytoplasmic fraction from atretic granulosa cells was about 3.08-fold increased as compared with that from healthy granulosa cells, and this increase reached about 8.22-fold in nucleus fraction. The results support our hypothesis that up-regulation and nuclear translocation of the FoxO3a protein is a necessary prerequisite for the activation of proapoptotic genes, such as bim, in response to follicular atresia.
FoxO3a Binds to bim Promoter to Activate Transcription-To determine whether FoxO3a activates Bim EL expression at the promoter level and to identify the transcriptional element involved, we set out to clone the porcine bim promoter. Essafi et al. (28) showed that the activity of the mouse bim promoter is completely retained in the 500-bp region upstream of the transcriptional start site. Based on the information, we cloned by PCR the porcine bim promoter (442 bp). Sequence analysis of the porcine and mouse promoter regions showed 68% identity and contained a common putative FHRE (Forkhead-responsive element) proximal to the transcription start site (Fig. 6A), which can be involved in the induction of Bim expression.
We next performed ChIP assays to determine FoxO3a occupancy of the bim promoter. The chromatins in GFP-FoxO3a plasmid-transfected granulosa cells were sheared into 200 -1000-bp fragments (Fig. 6B). As demonstrated in Fig. 6C, the anti-GFP antibody, but not the control antibody (IgG), precipitated the region of the bim promoter containing the FoxO site in GFP-FoxO3a plasmid-transfected granulosa cells. These results strongly suggest that FoxO3a is recruited directly to the FHRE of the bim promoter to activate its transcription. Taken together, these results suggest that the PI3K/Akt/FoxO3a pathway is involved in FSH regulation of Bim EL expression.

DISCUSSION
Follicular atresia is a complicated process that requires intrinsic signaling molecules as well as extrinsic regulators such as FSH (31). Some of these factors control follicular atresia by regulating the fate of granulosa cells, specifically granulosa cell apoptosis (5). The Bcl-2 subfamily of proteins represents one of the intrinsic factors that play essential roles in promoting granulosa cell survival, whereas Bax subfamily protein serves as proapoptotic signals to induce granulosa cell death (32). It is suggested that the Bcl-2 and Bax protein subfamilies control the balance of "life and death" in granulosa cells. In this study, we examined whether Bim, a BH3-only protein member, had a dynamic expression pattern in atretic follicles and ultimately mediated FSH-regulated granulosa cells apoptosis. To our knowledge, this is the first evidence demonstrating the role of Bim in female reproduction and more specifically in follicular atresia.
It is important to note that, based on the characterization done in this study, porcine Bim does not contain an identifiable COOH-terminal hydrophobic tail. This tail is present in numerous members of the Bcl-2 family (e.g. Bak, Bcl-2, and Bcl-xl) and is used to localize the protein in the membranes of mitochondria, endoplasmic reticulum, or the nuclear envelope (33). In contrast, several pro-apoptotic Bcl-2 family proteins, e.g. Bax, Bad, and Bid, are inactive cytosolic proteins under normal conditions and translocated to the mitochondria via this tail during apoptosis (34). It was recently reported that Bim␣1, -␣2 (19), and -␥ (17), which also lack the COOH-terminal hydrophobic region, were located in the mitochondria, whereas Bim␤1-␤4 isoforms (19), which lack both the BH3 domain and the hydrophobic regions, were located only in the cytosol. Perhaps the BH3 domain itself is sufficient for mediating membrane localization, and the COOH-terminal hydrophobic domain is not essential for their apoptotic capability. This model is supported by our overexpression data.
Similar to other members of the Bcl-2 family (35,36), Bim isoforms show tissue-specific expression patterns (Fig. 1C). There were distinct temporal and spatial roles for specific Bim isoforms during apoptosis (37). However, only the Bim EL protein was detected in apoptotic granulosa cells by Western blot in our experiment. Based on our observation that porcine Bim EL expression levels in granulosa cells varied with the health status of follicles, we speculated that Bim EL might serve to initiate atresia of porcine follicles. Although Bim S was considered the most potent compound for promoting apoptosis in human lymphoma cells (38), and its mRNA was detected by RT-PCR in ovary tissue (Fig. 1C), the expression of its protein in porcine granulosa cells was not detected in this study. Therefore, mech-anisms governing the tissue specificity of Bim isoform expression require further study.
As a pro-apoptotic protein, Bim was highly expressed in atretic follicles. Bim EL immunostaining was barely detectable in stroma cells and concentrated in both granulosa and theca cells, especially in the inner surface of the granulosa layer. Apoptosis in atretic follicles in most mammals was confined to granulosa cells, so we focused on the Bim function in granulosa cells in this study. However, the Bim expression patterns and function in theca cells are also worth further investigation. Our results indicated that there was a clear correlation between follicular atresia and increased Bim EL protein levels. To our knowledge, this work is the first to show the Bim protein function in the follicular atresia in any species.
It is well established that FSH is a critical survival factor for granulosa cells during follicular development (39). Sufficient exposure of antral follicles to FSH is essential for the follicles to escape from atresia and to reach preovulatory stage (4). To explore whether Bim EL expression is regulated by FSH, an in vitro granulosa cell culture system was utilized. Our results demonstrated that FSH significantly suppressed granulosa cell apoptosis. This result was consistent with other reports (41). Interestingly, granulosa cell apoptosis attenuated by FSH was accompanied by a decrease in Bim EL accumulation. This is also the first indication that Bim EL accumulation is induced after FSH withdrawal and contributes to FSH withdrawal-induced apoptosis in granulosa cells.
It was recently reported that Forkhead transcription factor FoxO3a has emerged as a major regulator of ovarian function. For instance, gene silencing of FoxO3a in mice causes accelerated follicular initiation, resulting in premature follicular depletion and ovarian failure (42). Matsuda et al. (43) also reported that FoxO3a was expressed in porcine follicles and induced granulosa cell apoptosis. However, the question regarding which protein regulated by FoxO3a could contribute to the above process is unanswered. In this study, we showed that Bim was one candidate that can be up-regulated by FoxO3a overexpression and that is modestly induced in granulosa cells by atresia. However, other FoxO3a-regulated proteins could also be involved in this apoptosis. One method to detect whether other FoxO3a-regulated proteins might participate in this apoptosis would be to assess whether FoxO3a overexpression can induce apoptosis in bim-deficient granulosa cells.
FoxO3a was also expressed and hormonally regulated in the mouse ovary. Both of its transcription and phosphorylation states were determined by FSH in cultured granulosa cells (44). In this study, we found that FSH could rapidly phosphorylate FoxO3a at Ser-253 and ultimately inhibit Bim EL expression. In granulosa cells, both phosphorylation of FoxO3a and repression of Bim EL expression were diminished after treatment with the PI3K inhibitor, LY294002, suggesting that the PI3K signaling pathway was involved in the regulation of the Bim protein. Indeed, previous studies had shown that LY294002 increases FIGURE 6. FoxO3a binds to bim promoter to activate transcription. A, nucleotide sequence of the 442-bp fragment of the porcine bim promoter aligned to the corresponding mouse sequence. Identical bases are indicated by asterisks, and amino acids with conserved similarities are indicated by dots or colons. The arrow indicates the first nucleotide of the transcription start site. The putative Forkhead-responsive element is boxed. B, granulosa cells in Lysis Buffer at a cell concentration of 1 ϫ 10 7 per ml sheared with six sets of 10-s pulses (40 watts) on wet ice using a high intensity ultrasonic sonicator (Cole Parmer). Sheared chromatin was then electrophoresed through a 2% agarose gel and stained with ethidium bromide. The picture shows that the majority of the DNA has been sheared to a length between 200 and 1000 bp. C, chromatin fragments were subjected to immunoprecipitation with antibodies to IgG (nonspecific) or GFP as indicated. Upon DNA isolation, amplification of the bim promoter DNA was assessed by PCR and gel electrophoresis. Input DNA was used as positive control. A representative graph of three independent experiments is shown.
Bim protein levels in Ba/F3 cells (45) or sympathetic neurons (46). It is now clear that the inhibition of PI3K activity by LY294002 allows FoxO3a to translocate into the nucleus, where they are known to be direct activators of bim gene expression (46). Our results indicated that FoxO3a might also regulate Bim expression in granulosa cells using FoxO3a overexpression and ChIP model. Because of this, we speculated that FoxO3a could be translocated to the nucleus to promote Bim expression when follicular atresia originates under either physiological or pathological conditions.
In our studies on the effects of FSH on Bim EL expression, we found that FSH not only decreased the levels of Bim EL expression but also shifted the molecular weight of the Bim EL protein. This finding is in agreement with the results of several previous studies in which changes in the electrophoretic mobility of Bim EL were attributed to phosphorylation and dephosphorylation. Möller et al. (47) showed that a stem cell factor-induced band shift of Bim EL in mouse mast cells was due to phosphorylation and indicated that the larger band represented phosphorylated Bim EL after treatment with alkaline phosphatase. Abayasiriwardana et al. (48) also reported that this shift was abolished by alkaline phosphatase-mediated dephosphorylation but was restored if the phosphatase was inhibited by sodium orthovanadate in malignant mesothelioma cells. Moreover, recent studies have demonstrated that this phosphorylation could regulate the apoptotic function of Bim EL . In Ba/F3 cells, Ser-87 of Bim EL was found to be an important phosphorylation site targeted by Akt to attenuate the pro-apoptotic function of Bim EL (49). Similarly, in a human B lymphoma cell line, phosphorylation of Bim EL by Erk1/2 on Ser-69 selectively led to its proteasomal degradation, thereby promoting cell survival (30). However, the JNK-induced phosphorylation of Bim EL at Ser-65 promoted the apoptotic effect of Bim EL in primary cerebellar granule neurons (40). These results suggested that different kinase pathways might ultimately result in different patterns of Bim EL phosphorylation that determine whether phosphorylated Bim EL facilitates or antagonizes apoptosis. In this study, our results showed that BimEL phosphorylation status decreased during apoptosis of the porcine granulosa cells, which indicated that phosphorylation of BimEL might protect the porcine granulosa cells from apoptosis. However, the detailed mechanism of Bim EL phosphorylation in granulosa cells is as yet unknown and will form the basis of further investigation.
A hypothetical model of Bim protein regulation by FSH in granulosa cells is outlined in Fig. 7. Although there are some questions that still needed to be resolved, in this study, we showed that the level of Bim EL expression impacted porcine follicular atresia, and FSH could prevent granulosa cells apoptosis by down-regulating Bim EL expression via the PI3K/Akt/FoxO3a pathway. This work provides new insight into how the Bcl-2 protein family, especially Bim, functions in follicular atresia.