Ovarian granulosa cells utilize scavenger receptor SR-BI to evade cellular cholesterol homeostatic control for steroid synthesis.

Cellular cholesterol is known to be under homeostatic control in nonsteroidogenic cells, and this intrigued us to understand how such control works in steroidogenic cells that additionally use cholesterol for steroid hormone synthesis. We employed primary culture of rat ovarian granulosa cells to study how steroidogenic cells adapt to acquire sufficient cholesterol to meet the demand of active steroidogenesis under the stimulation of gonadotropin follicle-stimulating hormone (FSH) and cytokine transforming growth factor (TGF)β1. We found that TGFβ1 potentiated FSH to upregulate scavenger receptor class B member I (SR-BI) and LDL receptor (LDLR), both functional in uptaking cholesterol as hHDL(3) and hLDL supplementation enhanced progesterone production, and the effect of each lipoprotein was completely or partially blocked by SR-BI selective inhibitor BLT-1. Uptaken cholesterol could also be stored in lipid droplets. Importantly, LDLR and SR-BI responded to sterol with different sensitivity. Giving cells lipoproteins or 25-hydroxycholesterol downregulated Ldlr but not Scarb1; Scarb1 was ultimately downregulated by excessive sterol accumulation under 25-hydroxycholesterol and aminoglutethimide (inhibitor of steroidogenesis) cotreatment. Furthermore, transcription factors sterol regulatory element-binding protein (SREBP)-2 and liver receptor homolog (LRH)-1 crucially mediated Ldlr and Scarb1 differential response to sterol challenge. This study reveals that ovarian granulosa cells retain the cholesterol homeostatic control machinery like nonsteroidogenic cells, although during active steroidogenesis, they utilize SR-BI to evade such feedback control.

Pituitary-secreted FSH is the master hormone that promotes ovarian follicle development and granulosa cell differentiation before ovulation ( 21 ), and ovarian cytokine transforming growth factor (TGF ␤ 1) plays an important role in facilitating FSH action ( 22,23 ) including enhancement of progesterone production through increased expression of steroidogenic proteins StAR, P450scc, and 3 ␤ -HSD (24)(25)(26)(27). We were intrigued by how cellular cholesterol is controlled in steroidogenic cells when facing a great demand of cholesterol for active steroidogenic activity. This study, therefore, was initiated to explore how cholesterol entry mediators LDLR and SR-BI are regulated in response to FSH and TGF ␤ 1 stimulation of steroidogenesis in ovarian granulosa cells. Primary culture of granulosa cells from gonadotropin-primed immature rats was used, as both SR-BI and LDLR pathways are active in in vitro culture condition ( 8,13 ). There were two specifi c aims. The fi rst was to investigate FSH and TGF ␤ 1 regulation of the expression of SR-BI and LDLR and their subcellular localization as well as functional involvement in steroidogenesis. The second aim was to examine sterol responsiveness of the two receptors and the potential involvement of two key transcription factors SREBP-2 and LRH-1.

Animals
Immature Sprague Dawley-derived rats (23-25 days of age) were obtained from the Animal Center at National Yang-Ming University (Taipei, Taiwan). Rats were maintained under controlled temperature (22-25°C) and light conditions (12 h light/12 h dark). Food (Lab Diet from PMI Feeds Inc., St. Louis, MO) and water were available ad libitum. This study was conducted in accordance with the United States National Research Council's Guide for the (SR-BI), which selectively uptakes cholesteryl ester (CE) without internalizing the whole lipoprotein particle ( 4,5 ). Lipoprotein utilization in steroidogenic cells could be different among species. For rodents, HDL is the major circulating lipoprotein and SR-BI serves as the principal lipoprotein receptor in steroidogenic tissues ( 4,(6)(7)(8), whereas for humans, LDL is the major cholesterol-carrying lipoprotein and the role of LDLR and SR-BI is not entirely clear ( 9 ). Earlier limited studies suggest that LDL and LDLR participate in human steroidogenesis, as adrenocorticotropic hormone (ACTH)-increased cortisol level was lower in LDLR-defi cient homozygous familial hypercholesterolemia patients and LDL-defi cient hypobetalipoproteinemia patients compared with normal subjects ( 10,11 ). Additionally, a recent study indicates SR-BI also takes part in human adrenal steroidogenesis, as human subjects with a missense mutation of SR-BI had reduced corticotropinstimulated adrenal corticosteroid level, and cells with such mutation had defective uptake of HDL-CE ( 12 ).
In steroidogenic cells, SR-BI and LDLR are regulated by tropic hormones. Gonadotropins and ACTH increased the expression of SR-BI and LDLR in murine granulosa cells and adrenocortical cells, respectively ( 8,13,14 ). Treatment with an ovulation-inducing stimulus of folliclestimulating hormone (FSH) followed by human chorionic gonadotropin (hCG) also increased the expression of LDLR and SR-BI in rhesus macaque granulosa cells ( 15 ). Also, luteinized granulosa cells retrieved from women receiving gonadotropin-based ovulatory regimen had increased cholesterol uptake from LDL and HDL to support steroidogenesis ( 16 ). Previous studies revealed that in nonsteroidogenic cells, the cellular cholesterol level is under surveillance by the homeostatic control mechanism, with LDLR being a key regulatory target ( 17 ). When cells are at a state of sterol insuffi ciency, the endoplasmic reticulum-bound transcription factors sterol regulatory element-binding proteins (SREBP) are fi rst transported to the Golgi apparatus through the assistance of a sterolsensing molecule, SREBP cleavage-activating protein (SCAP). Then under the action of Golgi proteases, the N-terminal transcription factor domain of SREBP is released and translocates into the nucleus, and thereby activates transcription of target genes. There are two SREBP genes encoding three proteins, the splice variants SREBP-1a and SREBP-1c, and SREBP-2. SREBP-2 preferentially controls cholesterogenic genes, including LDLR ( Ldlr ) and HMG-CoA reductase ( Hmgcr ), SREBP-1c controls lipogenic genes, and SREBP-1a activates both categories. Also, it has been reported that SR-BI is potentially regulated by SREBP-1a transactivation through sterol response element (SRE)-like sites in its promoter ( 18 ). Notably, SR-BI is a known target of liver receptor homolog-1 (LRH-1), a nuclear receptor of the NR5A subfamily that is critically involved in the regulation of cholesterol metabolism, including reverse cholesterol transport, bile acid synthesis, and ovarian steroidogenesis ( 19 ). Whether and how the SREBP-mediated cholesterol feedback control mechanism and LRH-1 are involved in the regulation of LDLR and SR-BI in steroidogenic cells is not clear ( 20 ). viewed under a fl uorescence microscope (Olympus BX50, Tokyo, Japan) equipped with a mercury arc lamp, and photographed using SPOT image capture system (Diagnostic Instruments Inc., Sterling Heights, MI). Only the fl uorescence emission intensity, exposure time, and signal output gain value could affect the outcome signal intensity. To preclude overexposed images, the system was set to automatically capture the sample image that displayed the most intense fl uorescence signal, that is, FSH+TGF ␤ 1-treated cells. Then the exposure time and signal output gain value of this image was used to capture all of the subsequent samples in a single experiment. In this way, the fl uorescent images could refl ect the actual emission signal for comparison among samples.

Steroidogenic response to lipoprotein supplementation
Human HDL 3 (hHDL 3 ) and human LDL (hLDL) were prepared as previously described ( 30 ). In brief, blood samples from healthy donors were centrifuged at 2,000 g for 30 min to obtain sera, which were then supplemented with 3 mM EDTA-disodium and 1 mM phenylmethylsulfonyl fl uoride to avoid proteolytic degradation. Lipoproteins were isolated from the sera through sequential fl oatation ultracentrifugation using KBr solution (hHDL 3 at 1.125 g/cm 3 < < 1.21 g/cm 3 , and hLDL at 1.019 g/cm 3 < < 1.063 g/cm 3 ). The isolated lipoprotein fractions were dialyzed against PBS using a centrifugal fi lter unit with a mw cutoff of 10 kDa (Millipore Corp., Billerica, MA), and then sterilized by passing through a 0.22 m syringe fi lter. Protein concentration of the lipoprotein preparations was determined by the Bradford method, and apolipoprotein constitution was confi rmed by gel electrophoresis after delipidation with ether and ethanol extraction as previously described ( 31,32 ).
To assay for lipoprotein-supplemented steroidogenesis, granulosa cells (approximately 5.5 × 10 5 ) were cultured in matrigelcoated 24-well plates, treated with FSH (10 ng/ml) and/or TGF ␤ 1 (0.5 ng/ml) for 24 h, and then given hHDL 3 or hLDL for an additional 3 h. At the end of culturing, progesterone content in the conditioned medium was determined by enzyme immunoassay described below. To understand the potential role of SR-BI in the process, a selective inhibitor BLT-1 was given 1 h prior to lipoprotein administration. Also we determined the progesterone content in lipoprotein preparations; the estimated progesterone content in our highest treatment dose of hHDL 3 and hLDL was approximately 20 ‫ف‬ 40% of that detected in the conditioned media of unstimulated control cells, and it was less than 1% in FSH+TGF ␤ 1-treated group. The progesterone production of granulosa cells was calculated by subtracting the progesterone content in supplemented lipoprotein from that in the conditioned medium sample and then normalizing to the cell number determined by crystal violet assay as previously described ( 33 ).

Enzyme immunoassay
Progesterone levels in the conditioned media were measured using enzyme immunoassay as previously described ( 26,27 ). Progesterone antisera and progesterone-conjugated horseradish peroxidase were produced and verifi ed in the laboratory of Dr. Leang-Shin Wu (National Taiwan University, Taipei, Taiwan). In brief, progesterone standards and conditioned medium samples were incubated in the presence of progesterone-conjugated horseradish peroxidase in anti-progesterone antibody-coated 96well plates for 2 h at room temperature. This was then rinsed off, and peroxidase substrate 2,20-azino-bis(3-ethylbenzthiazoline-6sulfonic acid) diammonium (Sigma Chemical Co.) was added and incubated for 2 h. The absorbance of reaction products was measured at 410 nm using a multimode microplate reader (Infinite 200 PRO, Männerdorf, Switzerland).
Care and Use of Laboratory Animals and institutional guidelines, and it was approved by the Institutional Animal Care and Use Committee of National Yang-Ming University.

Cell culture and treatment
Isolation of ovarian granulosa cells from gonadotropin-primed immature rats was performed as previously described ( 25 ). In brief, immature rats were injected once subcutaneously with 15 IU of eCG to stimulate the development of multiple follicles to antral follicle stage. Forty-eight hours later, ovarian granulosa cells of mid-to large-sized antral follicles were isolated and seeded into EHS matrigel-coated culture wares in growth medium (DMEM/F-12, 1:1, 2 g/ml bovine insulin, 0.1% fatty acidfree BSA, 100 U/ml penicillin, and 100 g/ml streptomycin), and were allowed to attach for 20 h at 37°C, 5% CO 2 -95% air. Cultured cells were then incubated in incubation medium (DMEM/F12, 1:1, 0.1% lactalbumin hydrolysate, 100 U/ml penicillin, and 100 g/ml streptomycin) for an additional 20 h before the beginning of treatment. Cultured granulosa cells at this point and through additional 48 h of cultivation period exhibited a fl attened and fi broblastic morphology, while treatment with FSH rapidly stimulated cell rounding that faded after 24 h; on the other hand, combined treatment with FSH and TGF ␤ 1 induced cell rounding phenomenon that lasted to 48 h, and cells were able to revert to fi broblastic shape 24 h after changing to fresh culture medium ( 28 ). Biochemical data also showed that TGF ␤ 1 potentiated FSH action to stimulate granulosa cell progesterone production, a differentiation maker that remained low in the control group during 48 h culture period (25)(26)(27)(28)(29). Together, these observations indicated that this cell model does not undergo spontaneous luteinization and is ideal for studying regulation of granulosa cell differentiation.

Immunoblotting analysis
Granulosa cells (approximately 6 × 10 6 ) were cultured in matrigel-coated 60 mm culture dishes and treated as described in the fi gure legends. At the end of culturing, cells were extracted with lysis buffer (50 mM Tris-HCl, pH 7.4, containing 150 mM NaCl, 1 mM EDTA-disodium, 1% NP-40 alternative, 0.25% sodium-deoxycholate, 1 mM phenylmethylsulfonyl fl uoride, 1 g/ml each of aprotinin, leupeptin, and pepstatin, 5 mM sodium orthovanadate, and 5 mM sodium fl uoride) and sonicated using an ultrasonicator (Missonix Inc., Farmingdale, NY). Cell lysates were analyzed for SR-BI, LDLR, SREBP-2, and LRH-1 with ␤ -actin as an internal control. SDS-PAGE analysis was performed with equal protein amount for each sample. After electrophoresis and electroblotting, blots were blocked with 5% fat-free milk and then incubated with primary antibodies. Specifi c signals were visualized using HRP-conjugated secondary antibodies and enhanced chemiluminescence substrate (PerkinElmer Inc., Waltham, MA). Signals on X-ray fi lms were quantifi ed using a two-dimensional laser scanning densitometer (Molecular Dynamics, Sunnyvale, CA).

Immunofl uorescence analysis
Granulosa cells (approximately 5 × 10 plus TGF ␤ 1 for 24 h. At the end of culturing, cells were fi xed in 1% formaldehyde for 10 min at room temperature, and the crosslinking reaction was stopped by adding glycine (fi nal 125 mM) and incubating for 5 min. Cells were then rinsed with cold PBS and extracted for chromatin in 200 l lysis buffer [50 mM Tris-HCl, pH 8.0, containing 10 mM EDTA-disodium, 1% SDS, and a cocktail of protease and phosphatase inhibitors (1 mM phenylmethylsulfonyl fl uoride, 1 g/ml each of aprotinin, leupeptin, and pepstatin, 5 mM sodium orthovanadate, and 5 mM sodium fl uoride)]. This was then sonicated using an ultrasonicator to obtain DNA fragments with sizes in the range of approximately 400-1000 bp. Then the samples were frozen, thawed on ice, and centrifuged at 15,000 g for 15 min at 4°C to precipitate SDS. The supernatant was diluted 5-fold with dilution buffer (16.7 mM Tris-HCl, pH 8.0, containing 167 mM NaCl, 1.2 mM EDTA-disodium, 1.1% Triton X-100, 0.01% SDS, and cocktail of protease and phosphatase inhibitors) and precleared by incubating with 50 l of protein G agarose blocked with 0.1% BSA and 50 g/ml salmon sperm DNA for 2 h at 4°C. This was centrifuged, and the supernatant was incubated with antibodies overnight at 4°C, followed by another 50 l of blocked protein G agarose for 6 h. To remove nonspecifi c binding, the sample was washed sequentially with low-salt wash buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTAdisodium, 1% Triton X-100, 0.1% SDS), high-salt wash buffer (20 mM Tris-HCl, pH 7.4, 500 mM NaCl, 2 mM EDTA-disodium, 1% Triton X-100, 0.1% SDS), LiCl buffer (10 mM Tris-HCl, pH 8.0, 250 mM LiCl, 1 mM EDTA-disodium, 1% NP-40, 1% sodium deoxycholate), and TE buffer (10 mM Tris-HCl, 1 mM EDTA-disodium, pH 8). The sample was then eluted with 400 l of elution buffer (0.1 M NaHCO 3 , 1% SDS). The eluate was then incubated at 65°C in the presence of 0.2 M NaCl for 4 h to reverse crosslinks, and then incubated with 100 g/ml proteinase K at 50°C for 1 h to remove associated proteins. After phenol/chloroform extraction and ethanol precipitation, the DNA sample was dissolved in TE buffer and analyzed by PCR and agarose gel electrophoresis. The primer pairs used are listed in Table 1 .

Statistical analysis
Quantitative data are presented as the mean (± SE) and were analyzed by ANOVA and Duncan's multiple-range test at a signifi cance level of 0.05 using the general linear model of the SAS program (SAS Institute Inc., Cary, NC). Differences between two treatment groups were analyzed using the Student t -test at a signifi cance level of 0.05.

RESULTS
Pituitary gonadotropin FSH acting together with cytokine TGF ␤ 1 facilitates ovarian granulosa cell differentiation marked by an enhancement of progesterone synthesis (24)(25)(26)(27)29 ). We were interested in understanding how FSH and TGF ␤ 1 regulate cholesterol availability for active steroidogenesis and how the machinery of cellular cholesterol homeostatic control works in ovarian granulosa cells.

FSH and TGF ␤ 1 regulation of SR-BI and LDLR in steroidogenic granulosa cells
We fi rst investigated the temporal profi le of FSH and TGF ␤ 1 regulation of SR-BI and LDLR expression in rat ovarian granulosa cells. Immunoblotting analysis showed

Fluorescent staining of cellular neutral lipids
Granulosa cells (approximately 5 × 10 5 ) grown on cover slips were treated with FSH and/or TGF ␤ 1 for 24 h followed by the addition of hHDL 3 (500 g/ml) or hLDL (100 g/ml) in the absence or presence of aminoglutethimide (0.5 mM; an inhibitor of P450scc) for an additional 24 h. At the end of culturing, cells were fi xed with 4% paraformaldehyde for 15 min, and then stained with 2 g/ml BODIPY 493/503 (Molecular Probes/Life Technologies Inc., Carlsbad, CA) for 10 min. Cells on cover slips were then mounted on glass slides and photographed as described above.

RT-PCR analysis
Granulosa cells (approximately 2 × 10 6 ) were cultured in matrigel-coated 35 mm culture dishes and treated as described in the fi gure legends. At the end of culturing, cells were extracted with 1 ml of Trizol reagent per 35 mm dish. After adding 200 l of chloroform, the mixture was centrifuged at 12,000 g for 15 min for phase separation. The upper aqueous phase was retained, mixed with isopropanol at 1:1 ratio, allowed to stand at room temperature for 30 min, and then centrifuged at 12,000 g for 10 min to precipitate RNA. The pellet was washed with 75% ethanol and then dissolved in ddH 2 O containing 0.1% DEPC. RNA concentration of each sample was estimated using ND-1000 spectrophotometer (Thermo Scientifi c Inc., Wilmington, DE). RNA sample was then reverse-transcribed to cDNA using MMLV reverse-transcriptase according to the manufacturer's protocol. Taq DNA polymerase was used to amplify cDNA for Scarb1 , Ldlr , and Hmgcr , with Actb used as an internal control. The forward and reverse primer pairs used are listed in Table 1 . The PCR cycle number used for each gene was within the linear range of amplifi cation. PCR was performed using Aztec PC-816 thermal cycler (Astec Co., Fukuoka, Japan), and the products were analyzed by 1.5% agarose gel electrophoresis with gel containing HealthView nucleic acid stain (Genomics Co., Taipei, Taiwan) and visualized with a UV transilluminator (Quantum ST-1000, Eberhardzell, Germany).

Chromatin immunoprecipitation analysis
Chromatin immunoprecipitation (ChIP) analysis was performed as previously described ( 28 ) with modifi cations ( 34 ). Granulosa cells (approximately 1.8 × 10 7 ) were cultured in matrigel-coated 100 mm culture dishes and treated with FSH immunostaining activity of SR-BI and LDLR at 24 h posttreatment was more intense in FSH+TGF ␤ 1-treated cells compared with the control and FSH-treated cells. To further understand the contribution of SR-BI and LDLR in cholesterol uptake in steroidogenic granulosa cells, we next assayed for functional activity of the two receptors.

Potential functional role of SR-BI and LDLR in steroidogenic granulosa cells
The cholesterol uptake activity of SR-BI and LDLR in ovarian granulosa cells was investigated by examining FSH and TGF ␤ 1 stimulation of progesterone production under hHDL 3 or hLDL supplementation with the combined use of BLT-1 (experimental designs shown in Figs. 2A and 3A ). BLT-1 is a small molecule inhibitor that blocks the selective CE uptake activity of SR-BI ( 35,36 ). hHDL 3 is rich in apoA-I and devoid of apoB and apoE, and it was reported to have a high affi nity for SR-BI but not LDLR ( 5,37 ). hLDL, on the other hand, is a ligand for LDLR and SR-BI ( 5 ), and thus, cells could obtain hLDL-derived cholesterol from both pathways. that at 6 and 12 h posttreatment, FSH±TGF ␤ 1 similarly increased SR-BI and LDLR protein levels, with a greater effect on SR-BI ( Fig. 1A ). While TGF ␤ 1 treatment alone had no effect (data not shown), it augmented FSH action and retained high levels of SR-BI and LDLR at 24 and 48 h posttreatment ( Fig. 1A ). Interestingly, granulosa cells treated with FSH for 48 h still had an SR-BI protein level signifi cantly higher than the control, whereas the LDLR level was lower than the control ( Fig. 1A ). Next, we employed immunofl uorescence analysis to observe subcellular localization of SR-BI and LDLR. After 24 h of treatment, control and TGF ␤ 1-treated cells showed weak immunostaining activity of SR-BI present at the nucleus and cytoplasm, and FSH treatment increased such staining activity in both compartments ( Fig. 1B ). Treatment with FSH+TGF ␤ 1 further increased SR-BI immunostaining activity, particularly at the plasma membrane protrusion processes and perinuclear region ( Fig. 1B ). On the other hand, LDLR, regardless of treatment, was predominantly distributed at cytoplasm in punctate patterns ( Fig. 1B ). Consistent with the result of immunoblotting analysis, the with FSH (10 ng/ml) and/or TGF ␤ 1 (0.5 ng/ml) for various time periods. Cell lysates were prepared and analyzed by immunoblotting for SR-BI and LDLR with ␤ -actin used as an internal control. The relative density ratio was calculated using 12 h or 48 h FSH+TGF ␤ 1-treated value as one. Each bar represents the mean (± SE) relative density (n = 3 ‫ف‬ 4). Different lowercase letters indicate signifi cant differences among groups in the same treatment period ( P < 0.05). (B) Cultured cells were treated with FSH and/or TGF ␤ 1 for 24 h, and then analyzed by immunofl uorescence for SR-BI and LDLR with normal rabbit IgG used as a specifi city control. Bar = 20 m. or presence of TGF ␤ 1 for 24 h, and then supplemented with lipoproteins for an additional 24 h ( Fig. 4A ). To observe cholesterol accumulation without the complication of differential steroidogenic activity among treatment groups (control, FSH, and FSH+TGF ␤ 1), aminoglutethimide (AMG; an inhibitor of P450scc, the fi rst enzyme in steroidogenesis) was given to cells together with lipoproteins to block steroidogenesis. In the absence of AMG and with lipoprotein supplementation, FSH+TGF ␤ 1 treatment but not FSH alone increased neutral lipid content in cells with diffuse fl uorescent staining prominently seen in the cytoplasm, and a few lipid droplets were also present ( Fig. 4B, C ). Giving cells hHDL 3 increased the number of neutral lipid-containing lipid droplets in FSH±TGF ␤ 1-treated cells, and cells supplemented with hHDL 3 plus AMG exhibited more prominent increases of lipid droplets in all groups (control, FSH, FSH+TGF ␤ 1) ( Fig. 4B ), suggesting increased uptake and storage of hHDL 3 -derived cholesterol. On the other hand, cells given hLDL also displayed increased neutral lipid-staining intensity in all groups (control, FSH, FSH+TGF ␤ 1), and concomitant treatment with hLDL and AMG further enhanced such staining intensity, except in the FSH-treated group ( Fig. 4C ).
These results together suggest that in ovarian granulosa cells, FSH and TGF ␤ 1-upregulated SR-BI is functional for hHDL 3 -cholesterol uptake and that hLDL-cholesterol could be internalized through both LDLR and SR-BI pathways; furthermore, cholesterol from both routes could be used for steroid hormone synthesis and stored in lipid droplets. Relative ratio of progesterone production was calculated using the value of FSH+TGF ␤ 1-treated value as one. Each bar represents the mean (± SE) relative progesterone production (n = 6). Different lowercase letters indicate signifi cant differences among groups without BLT-1 treatment in the respective control, FSH, or FSH+TGF ␤ 1 treatment set ( P < 0.05). Asterisk indicates a signifi cant difference compared with the respective control without BLT-1 treatment ( P < 0.05). Granulosa cells were fi rst treated with FSH and TGF ␤ 1 for 23 h to upregulate SR-BI level (control < FSH < FSH+TGF ␤ 1; Fig. 1A ) and then treated with or without BLT-1 for an additional hour prior to treatment with hHDL 3 for another 3 h ( Fig. 2A ). Without exogenous lipoprotein supply, TGF ␤ 1 augmented FSH stimulation of progesterone synthesis during the 27 h treatment period ( Fig. 2B ). Giving cells hHDL 3 further increased FSH and TGF ␤ 1-stimulated progesterone production in a dosedependent manner; this increase was nearly abolished by BLT-1 treatment ( Fig. 2B ), suggesting that granulosa cells could effectively uptake hHDL 3 -derived cholesterol through SR-BI for steroidogenesis. We then tested the effect of BLT-1 in hLDL-supplied cells ( Fig. 3A ). Administration of hLDL dose-dependently increased progesterone production in FSH and TGF ␤ 1-treated cells ( Fig. 3B ). Noticeably, BLT-1 only partially suppressed hLDL enhancement of FSH+TGF ␤ 1-stimulated progesterone synthesis, whereas BLT-1 had no signifi cant effect on the FSHtreated group ( Fig. 3B ). This fi nding implies that granulosa cells could uptake hLDL-derived cholesterol through LDLR and SR-BI routes to meet the high demand of active steroidogenesis.
Uptaken cholesterol could also be stored in the lipid droplet. We next investigated whether FSH and TGF ␤ 1 modulate lipid droplets in granulosa cells using a fl uorescent dye BODIPY 493/503 that stains for neutral lipids. Granulosa cells were treated with FSH in the absence displayed a similar effect to that of hHDL 3 , except that the control 48 h group (i.e., 24 h post-hLDL supplementation) also had reduced mRNA levels of Ldlr and Hmgcr ( Fig. 6B ). These results suggest that the expression of Scarb1 , unlike Ldlr and Hmgcr , is relatively insensitive to cholesterol feedback control and, therefore, may be preferentially used in ovarian granulosa cells to uptake and accumulate cholesterol during active steroidogenesis.
Next, we employed a cell-permeable cholesterol analog, 25-hydroxycholesterol (25-OHC) to bypass the different cellular uptake mechanisms for cholesterol, and 25-OHC was also reported to inhibit SREBP transcriptional activity by blocking SCAP-SREBP complex exit from the endoplasmic reticulum ( 39 ). Additionally, AMG was used to prevent the diversion of 25-OHC to steroid synthesis and thus could further increase cellular cholesterol accumulation in the presence of 25-OHC. RT-PCR analysis reveals that FSH±TGF ␤ 1 similarly increased the

Differential sterol responsiveness of SR-BI and LDLR in steroidogenic granulosa cells
We further explored how lipoprotein supplementation modulates the expression of SR-BI gene ( Scarb1 ) and two key targets in the cholesterol homeostatic control process, LDLR ( Ldlr ) and HMG-CoA reductase ( Hmgcr ). Granulosa cells were stimulated with FSH and TGF ␤ 1 for 24 h, and then given hHDL 3 or hLDL for an additional 3 or 24 h ( Figs. 5A and 6A ). RNA samples were then isolated and analyzed by RT-PCR. It is known in nonsteroidogenic cells that sterol excess leads to downregulation of Ldlr and Hmgcr ( 38 ). We showed that such cellular cholesterol control mechanism functions in steroidogenic granulosa cells as well. Administration of hHDL 3 for 3 or 24 h reduced the mRNA levels of Ldlr and Hmgcr in FSH-and FSH+TGF ␤ 1treated groups ( Fig. 5B ). Interestingly, hHDL 3 had little effect on Scarb1 expression in all groups (control, FSH, and FSH+TGF ␤ 1) ( Fig. 5B ). Administration of hLDL , Ldlr , and Hmgcr with Actb used as an internal control. Relative density ratios were calculated using FSH+TGF ␤ 1-treated value as one at each treatment time period. Each bar represents the mean (± SE) relative density (n = 3). Different lowercase letters indicate signifi cant differences among groups in the absence of lipoprotein supplementation ( P < 0.05). Asterisk indicates a signifi cant difference compared with respective control without lipoproteins ( P < 0.05). SREBP-2 and LRH-1. SREBP-2 is the SREBP isoform that preferentially regulates cholesterogenic genes, including Ldlr ( 38 ). Also, a previous study reported that Scarb1 promoter contains a proximal SRE and a nuclear receptor NR5A response element (NRE) ( 18 ). Additionally, LRH-1 has been known to transactivate Scarb1 ( 40 ), and is specifi cally expressed in the granulosa cell layer in ovary ( 41 ).
We fi rst determined FSH and TGF ␤ 1 regulation of SREBP-2 proteolytic activation using immunoblotting to analyze the full-length and the cleaved mature forms of SREBP-2, respectively designated as SREBP-2(f) and SREBP-2(m). FSH signifi cantly increased the SREBP-2(m) level at 6 and 12 h posttreatment, and TGF ␤ 1 further enhanced FSH effect at 24 and 48 h posttreatment ( Fig. 8A ). Cells given FSH plus TGF ␤ 1 also had an increased level of SREBP-2(f) at 48 h posttreatment ( Fig. 8A ). Additionally, FSH also increased LRH-1 protein level at 6 and 12 h posttreatment, with the potentiating effect of TGF ␤ 1 lasting 12-48 h posttreatment ( Fig. 8B ). To confi rm SREBP-2 and LRH-1 direct involvement in the expression of Ldlr and Scarb1 , ChIP analysis was used. Cells treated with FSH plus TGF ␤ 1 for 24 h had increased SREBP-2 binding to both mRNA levels of Scarb1 , Ldlr , and Hmgcr at 4 h posttreatment, and TGF ␤ 1 further enhanced FSH action at 24 h posttreatment ( Fig. 7 ). Administration of 25-OHC downregulated the basal and FSH±TGF ␤ 1-increased mRNA levels of Ldlr and Hmgcr at 4 and 24 h posttreatment, whereas 25-OHC had no effect on Scarb1 mRNA level ( Fig. 7A ). Coadministration of 25-OHC and AMG for 4 h had an effect similar to that of 25-OHC alone, and 25-OHC plus AMG administration for 24 h further suppressed FSH±TGF ␤ 1-increased mRNA levels of Ldlr and Hmgcr , and interestingly, this also attenuated Scarb1 expression ( Fig. 7B ).
These results together clearly demonstrate that Ldlr and Hmgcr , and Scarb1 genes display differential response to sterol challenge in steroidogenic granulosa cells, and that SREBP may be involved in the process.

Potential role of SREBP and LRH-1 in differential sterol responsiveness of SR-BI and LDLR in steroidogenic granulosa cells
To further identify the underlying mechanisms for differential sterol effect on Ldlr and Scarb1 in ovarian granulosa cells, we focused on two key transcription factors, Fig. 7. Effect of 25-OHC and AMG on FSH and TGF ␤ 1-induced expression of cholesterogenic genes in steroidogenic granulosa cells. Cultured cells were pretreated for 1 h with (A) 5 g/ml 25-OHC or (B) 0.5 mM AMG plus 25-OHC, and then treated with FSH and/or TGF ␤ 1 for additional 4 or 24 h. RNA extracts were analyzed by RT-PCR, and relative density ratios of Scarb1 , Ldlr , and Hmgcr were calculated as described in Fig. 5 . Each bar represents the mean (± SE) relative density (n = 4). Different lowercase letters indicate signifi cant differences among groups in the absence of 25-OHC±AMG ( P < 0.05). Asterisk indicates a significant difference compared with the respective control without 25-OHC±AMG ( P < 0.05). ND indicates density too weak to be accurately quantifi ed. no obvious LRH-1 binding to NRE1 and NRE2 site in Ldlr promoter was detected ( Fig. 8D ).
We then further investigated the effect of AMG and 25-OHC on SREBP-2 proteolytic activation and LRH-1 protein level in correlation with the changes of LDLR and SR-BI protein levels. Administration with AMG and/or 25-OHC dramatically suppressed the FSH-and FSH+TGF ␤ 1-upregulated protein levels of SREBP-2(m) and LDLR, with 25-OHC exhibiting a stronger effect ( Fig. 9 ). Neither AMG nor 25-OHC signifi cantly affected LRH-1 level at 6 h after FSH treatment ( Fig. 9 ). While AMG alone had no effect, 25-OHC significantly decreased the FSH+TGF ␤ 1-upregulated LRH-1 at 24 h posttreatment, and the inhibitory effect was greater under 25-OHC and AMG cotreatment; interestingly, a similar effect Scarb1 and Ldlr promoter regions, with a greater effect on Ldlr ( Fig. 8C ). While Scarb1 promoter was reported to have an NRE site that could respond to LRH-1 ( 18,40 ), whether LRH-1 regulates Ldlr expression is unclear. To identify potential LRH-1 binding sites in Ldlr promoter, we analyzed the promoter sequence and found two putative LRH-1 binding sites with only one mismatch from the consensus YCAAGGYCR (where Y = T/C; R = G/A) ( 42 ) within the range of 2,000 bp upstream of Ldlr gene ( Ϫ 173 ‫ف‬ Ϫ 181 and Ϫ 831 ‫ف‬ Ϫ 839, here designated Ldlr -NRE1 and Ldlr -NRE2). Therefore, we constructed two primer pairs to encompass each individual site ( Table 1 ). ChIP analysis showed that cells treated with FSH plus TGF ␤ 1 for 24 h had increased LRH-1 binding to Scarb1 promoter, whereas Cell lysates were analyzed by immunoblotting for SREBP-2 with ␤ -actin used as an internal control. The full-length and proteolytic mature forms of SREBP-2 are respectively designated as SREBP-2(f) and SREBP-2(m). (B) Cells were treated with FSH and/or TGF ␤ 1, and cell lysates analyzed by immunoblotting for LRH-1. The relative density ratio was calculated using 12 h or 48 h FSH+TGF ␤ 1-treated value as one. Each bar represents the mean (± SE) relative density (n = 4). Different lowercase letters indicate signifi cant differences among groups in the same treatment period ( P < 0.05). (C, D) Cells were treated with FSH+TGF ␤ 1 for 24 h, and cell lysates were analyzed by ChIP assay using SREBP-2 or LRH-1 antibody with isotypic IgG serving as a negative control. After normalization to input, relative density ratio was calculated with control value as one. Each bar represents the mean (± SE) relative density (n = 3). Asterisk indicates a signifi cant difference compared with the control ( P < 0.05). upregulation of SR-BI and LDLR in ovarian granulosa cells to meet the high demand of cholesterol for active steroidogenic activity. First, treatment with FSH increased protein levels of SR-BI and LDLR, and TGF ␤ 1 could sustain such an increase for at least 48 h posttreatment ( Fig. 1 ). Second, FSH and TGF ␤ 1-upregulated SR-BI and LDLR played active roles as lipoprotein receptors in granulosa cells, as supplying granulosa cells with hHDL 3 or hLDL further increased FSH+TGF ␤ 1-stimulated progesterone production and cholesterol accumulation in lipid droplets ( Figs. 2-4 ). hHDL 3 was reported to have a high affi nity for SR-BI but not LDLR, whereas hLDL is a ligand for LDLR and SR-BI ( 5,37 ). Treatment with BLT-1, a selective inhibitor of SR-BI, almost completely blocked the hHDL 3 supplementation-enhanced FSH+TGF ␤ 1-stimulated progesterone production, but it only partially reduced the effect of hLDL ( Figs. 2 and 3 ), suggesting that hHDL 3 is largely uptaken through SR-BI, whereas hLDL could be uptaken through both LDLR and SR-BI to enhance progesterone synthesis. Moreover, we found that lipoprotein-derived cholesterol accumulation (refl ected by BODIPY 493/503-stained neutral lipids) was apparent and related to the receptor level and that this phenomenon was enhanced when the was observed on SR-BI protein level ( Fig. 9 ). Altogether, these results indicate SREBP-2 and LRH-1 crucially mediate differential sterol responsiveness of LDLR and SR-BI.

DISCUSSION
Ovarian steroid hormones are essential for female reproductive competence and general well-being, and it is important to understand the control of steroidogenesis, including cellular uptake of the steroid precursor, cholesterol. Although it has been well demonstrated that cellular cholesterol is under homeostatic control in nonsteroidogenic cells ( 17,43 ), how it is regulated in steroidogenic cells remains poorly understood. Using the primary cell culture model system, this study reveals that ovarian granulosa cells retain the cholesterol homeostatic control machinery like nonsteroidogenic cells, although during active steroidogenesis, they utilize SR-BI to evade such control.
Previous studies indicated that lipoprotein receptors SR-BI and LDLR could be upregulated by FSH in ovarian granulosa cells ( 8,13 ). Here, we further provide experimental evidence supporting that TGF ␤ 1 potentiates FSH Fig. 9. Effect of 25-OHC and AMG on FSH and TGF ␤ 1 upregulation of lipoprotein receptors and key transcription factors SREBP-2, LRH-1, SR-BI, and LDLR in steroidogenic granulosa cells. Cells were pretreated with 0.5 mM AMG and/or 5 g/ml 25-OHC for 1 h, and then treated with FSH and/or TGF ␤ 1 for 6 h or 24 h. Then 10 M MG-132 was added 2 h prior to the end of culturing. Cell lysates were analyzed by immunoblotting for LDLR, SR-BI, SREBP-2, and LRH-1 with ␤ -actin used as an internal control. The relative density ratio was calculated using 6 h FSH-treated or 24 h FSH+TGF ␤ 1-treated value as one. Each bar represents the mean (± SE) relative density (n = 4). Different lowercase letters indicate signifi cant differences among groups in the same treatment period ( P < 0.05). ND indicates density too weak to be accurately quantifi ed. study identifi es that, in steroidogenic granulosa cells, LDLR is also subjected to cholesterol homeostatic control mechanism similar to nonsteroidogenic cells; interestingly, SR-BI responds to sterol challenge differently from LDLR. This is supported by the following lines of evidence. First, supplying granulosa cells lipoproteins downregulated the FSH and TGF ␤ 1-increased expression of Ldlr and Hmgcr but not Scarb1 ( Figs. 5 and 6 ). Also, the cell-permeable cholesterol analog 25-OHC exerted a similar effect as lipoproteins ( Fig. 7A ). Interestingly, we found that granulosa cells given 25-OHC together with AMG (an inhibitor preventing cholesterol conversion to steroids) could also reduce the expression of Scarb1 at later time and with lesser extent compared with Ldlr and Hmgcr ( Fig. 7B ). These results imply that cellular LDLR and SR-BI display different sensitivity to sterol challenge. Second, FSH stimulated SREBP-2 proteolytic maturation with an increased level of SREBP-2(m), and TGF ␤ 1 further enhanced FSH action for a longer duration ( Fig. 8A ). Additionally, FSH plus TGF ␤ 1 steroidogenic process was blocked by the P450scc inhibitor AMG ( Fig. 4 ). In cells given hHDL 3 ±AMG, the staining intensity was related to SR-BI level at 48 h posttreatment (control < FSH < FSH+TGF ␤ 1; Figs. 1A and 4 ). In cells given hLDL±AMG, the staining intensity was related to the cellular level of LDLR (FSH < control < FSH+TGF ␤ 1; Figs. 1A and 4 ). Interestingly, we also noticed that, at 24 h posttreatment, FSH-treated cells had an LDLR protein level similar to the control group while having an SR-BI level greater than the control; however, hLDL supplementation-enhanced progesterone production was not affected by BLT-1 pretreatment ( Figs. 1A and 3 ). Although both receptors were reported capable of uptaking cholesterol from LDL ( 5 ), our study suggests that hLDL-cholesterol may be more effi ciently uptaken through LDLR than SR-BI.
In nonsteroidogenic cells, cellular cholesterol has been clearly demonstrated to be under negative feedback regulation involving transcription factor SREBPs, with Ldlr and Hmgcr being two canonical SREBP targets ( 17 ); whether SREBP regulates SR-BI is unclear. This ultimately reduce the SR-BI level ( Fig. 10D ), and this might protect cells from cholesterol overload. In all, this study reveals that the cellular cholesterol homeostatic control mechanism is conserved in steroidogenic granulosa cells and that cells evade such control through SR-BI in order to acquire suffi cient cholesterol infl ux for steroid hormone production.
increased SREBP-2 binding to an Ldlr promoter fragment containing SRE with moderate binding to Scarb1 promoter ( Fig. 8C ). Furthermore, AMG and/or 25-OHC attenuated FSH±TGF ␤ 1-increased SREBP-2(m), and this is parallel to the change of LDLR ( Fig. 9 ). These results suggest that FSH and TGF ␤ 1 upregulation of SREBP-2(m) is likely due to cholesterol shortage consequent to active steroid synthesis and that SREBP-2 predominantly modulates the expression of Ldlr and, less so, of Scarb1 . Importantly, we identifi ed LRH-1 as a crucial transcription factor that mediates SR-BI response to sterol challenge and that LRH-1 is regulated by FSH and TGF ␤ 1. FSH increased LRH-1 protein level, and TGF ␤ 1 potentiated FSH action ( Fig. 8B ). Additionally, FSH plus TGF ␤ 1 increased LRH-1 binding to Scarb1 promoter region containing a putative NRE site, whereas no signifi cant binding was detected on Ldlr ( Fig. 8D ). Moreover, 25-OHC in the absence or presence of AMG reduced the FSH plus TGF ␤ 1-increased LRH-1 at 24 h posttreatment, and this is parallel to the change of SR-BI; whereas no effect was detected at 6 h posttreatment ( Fig. 9 ). This fi nding indicates that cells, when facing high extent of cholesterol accumulation for long duration, may turn off cholesterol uptake pathways other than LDLR through mechanisms including inhibition of LRH-1 transactivation of Scarb1 . Granulosa cells given 25-OHC had SREBP-2(m) reduced faster and perhaps at a lower sterol threshold than LRH-1 ( Fig. 9 ), and although SREBP-2 could bind to Scarb1 promoter ( Fig. 8C ), it may not be the key player of SR-BI response to sterol challenge. However, in a situation of sterol defi ciency, SREBP-2(m) may act with LRH-1 to upregulate Scarb1 .
A working model is proposed accordingly ( Fig. 10 ) to illustrate how steroidogenic granulosa cells adapt to obtain suffi cient cholesterol for active steroidogenesis with the conservation of a cholesterol negative feedback control mechanism involving SREBP seen in nonsteroidogenic cells. Under basal condition without tropic hormone stimulation, the LDLR level is monitored by a cellular cholesterol homeostatic control mechanism: SREBP-2-driven expression of LDLR would be downregulated when excess cholesterol is uptaken ( Fig. 10A, B ). Under FSH and TGF ␤ 1 marked stimulation of progesterone production (upregulation of StAR, P450scc, and 3 ␤ -HSD), two mechanistic pathways can be activated but are regulated differentially to meet the great demand of cholesterol in granulosa cells ( Fig. 10C, D ). During active steroidogenesis, cholesterol conversion into progesterone would cause a decrease in intracellular cholesterol, resulting in proteolytic activation of SREBP-2 that subsequently transactivates Ldlr ; moreover, FSH+TGF ␤ 1 upregulates LRH-1, which acts with SREBP-2 to augment Scarb1 expression ( Fig. 10C ). Cellular cholesterol accumulation rapidly represses SREBP activation and LDLR expression, making FSH+TGF ␤ 1-increased SR-BI the major receptor responsible for bulk entry of cholesterol from HDL and LDL to support steroidogenic need ( Fig. 10D ). Further increase of the cholesterol level would