Follicle-stimulating hormone/cAMP regulation of aromatase gene expression requires β-catenin

β-Catenin Enhances FSH Regulation of Key Steroidogenic Enzyme mRNAs.

We initially evaluated the effect of β-catenin on induction of mRNAs encoding CYP19A1 and rate-limiting components of steroid synthesis, steroidogenic acute regulatory protein (STAR), and cytochrome P450 side-chain-cleavage enzyme (CYP11A1). Preantral rat GCs were cultured in the absence of added steroids to establish a model where FSH induction of some target genes is modest (34), permitting optimal detection of contributions from β-catenin. GCs were transduced with an adenovirus encoding a β-catenin mutant lacking amino-terminal residues involved in its targeted degradation (Δ90), while retaining interactions with DNA-binding proteins (35). An adenovirus containing EGFP (36) provided a control for estimating the effect of Δ90 β-catenin in untreated and FSH-stimulated cells. Semiquantitative RT-PCR analysis indicated that, when compared with EGFP controls, both 24-h FSH treatment and Δ90 β-catenin alone could induce all three steroidogenic enzyme mRNAs, albeit to extents that varied between experiments (Fig. 1A and B). In contrast, Δ90 β-catenin consistently and significantly enhanced FSH induction of Cyp19a1 and Cyp11a1 mRNAs (P < 0.05), but not Star (Fig. 1 A and B). These data provide evidence that β-catenin can selectively modulate FSH regulation of two key steroidogenic enzymes in GCs, CYP19A1 and CYP11A1. Because CYP19A1 showed a marked sensitivity to β-catenin, subsequent analyses focused on the gene encoding this protein.

Residues 235–238 in the SF1 Ligand-Binding Domain Mediate a Functional Interaction with β-Catenin.

The ability of β-catenin to augment Cyp19a1 expression could be because of interactions with SF1, because the nuclear receptor has been shown to bind the cofactor via residues 235–238 in the proximal activation domain (Fig. 2A) (30). To test this hypothesis, we used recombinant adenoviruses to coexpress Δ90 β-catenin with either wild-type SF1, or SF1 with its β-catenin interaction domain mutated [SF1(235-4A)] in primary GCs, and measured the accumulation of Cyp19a1 mRNA.

Compared with EGFP controls (Fig. 2B, lane 1), wild-type SF1 increased Cyp19a1 mRNA expression in the absence of FSH (P < 0.05, Fig. 2B, lane 3). Coexpression of Δ90 β-catenin significantly increased the stimulatory effect of SF1 alone (P < 0.05, Fig. 2B, lane 4). In contrast, SF1(235-4A) did not induce Cyp19a1 mRNA, despite levels of expression similar to those of its wild-type counterpart (Fig. 2B, lane 5). The Δ90 β-catenin mutant was unable to rescue gene expression by SF1(235-4A) (Fig. 2B, lane 6). Similar results were observed when lower titers of wild-type and mutant SF1 were used (data not shown). These data suggest that the trans-activation of Cyp19a1 by SF1 requires both an intact β-catenin binding site and a functional interaction with β-catenin.

The Functional Interaction Between FSH and SF1 Requires an Intact β-Catenin-Interaction Domain.

We next determined whether the β-catenin interaction domain of SF1 was required to facilitate the FSH signal in rat GCs. In the absence of added steroids, both FSH and SF1 alone produced small increases in Cyp19a1 mRNA when compared with untreated EGFP controls (Fig. 3, lanes 1–3). The stimulatory effect of FSH was dramatically enhanced by the presence of SF1 (P < 0.05, Fig. 3, lane 4), but not SF1(235-4A) (Fig. 3, lane 6). Thus the β-catenin-interaction domain of SF1 is also essential for the synergistic actions of FSH and SF1 on Cyp19a1.

FSH/cAMP Regulation of the CYP19A1 Type II Requires β-Catenin.

Modulation of Cyp19a1 mRNA levels by β-catenin could reflect either enhanced transcription or posttranscriptional stabilization of the message. Transient expression assays and RNA interference were used to explore the effect of β-catenin on activity of a cAMP-responsive fragment of the CYP19A1 PII in primary GCs, as well as a human granulosa tumor cell line (KGN). Like GCs, KGN cells express endogenous SF1 and increase steroidogenesis in response to cAMP signaling (data not shown and ref. 37).

To initially test whether cAMP stimulation of CYP19-luc required β-catenin, a pool of four siRNA duplexes was used to knock down endogenous β-catenin protein (CTNNB1) in primary GCs (Fig. 4A). A similar pool of nontargeting siRNA duplexes controlled for non-sequence-specific effects. FSH or forskolin (24 h) significantly induced CYP19-luc in primary GCs cotransfected with control siRNA compared with vehicle-treated groups (Fig. 4A). In contrast, cotransfection with Ctnnb1-specific siRNA attenuated basal activity of the promoter and compromised the efficacy of FSH and forskolin (P < 0.05, Fig. 4A).

Similar results were observed in KGN cells by using two different approaches to reduce endogenous β-catenin. In addition to RNAi, we overexpressed Axin1, which encodes a rate-limiting component of a cytoplasmic protein complex that degrades β-catenin (38–40) and negatively regulates β-catenin-dependent transcription (39, 41). Because KGN cells have low levels of FSH receptors (37), we used forskolin to activate adenylyl cyclase and elevate cAMP. Cotransfection with either CTNNB1-specific siRNA or an Axin1 expression vector significantly inhibited forskolin stimulation of CYP19-luc (P < 0.05, Fig. 4 B and C).

We also tested whether increasing β-catenin levels affected cAMP responsiveness of CYP19A1 PII. Adenoviral transduction of Δ90 β-catenin in KGN cells synergistically enhanced forskolin stimulation of endogenous CYP19A1 mRNA (data not shown), mimicking results obtained in rat GCs treated with FSH (Fig. 1). Transiently transfected Δ90 β-catenin was insufficient to activate CYP19-luc in KGN cells (Fig. 4D). However, forskolin stimulation of CYP19-luc was significantly enhanced by Δ90 β-catenin (P < 0.05, Fig. 4D). Collectively, these data demonstrate that the actions of β-catenin on CYP19A1 in GCs are dependent on hormone-induced cAMP cascades.

SF1 Requires β-Catenin for Basal and cAMP Regulation of the Type II CYP19A1 Promoter.

Because the CYP19A1 PII region contains both cAMP- and SF1-response elements, we determined whether β-catenin was required for SF1-dependent activation of this promoter. Transient Δ90 β-catenin expression in KGN cells potentiated SF1 stimulation of CYP19-luc (P < 0.05, Fig. 5A). This result is consistent with observed effects on the endogenous gene in primary GCs (Fig. 2B) and emphasizes β-catenin as a transcriptional partner of SF1.

Conversely, depleting endogenous β-catenin by CTNNB1-specific RNAi significantly inhibited the basal trans-activity of SF1 (P < 0.05, Fig. 5B), and compromised the enhanced stimulation of CYP19-luc by forskolin and SF1 together (P < 0.05, Fig. 5B). Similar results were observed with Axin1 (data not shown). These data are in accordance with effects of SF1(235-4A) on the endogenous Cyp19a1 gene (Figs. 2 and 3), and they suggest that functional interactions with β-catenin link the cell surface gonadotropin/cAMP cascade to SF1.

SF1 and β-Catenin Localize to the Endogenous CYP19A1 PII.

Finally, ChIP analysis of KGN cells with a β-catenin antibody revealed modest but consistent association of the cofactor with CYP19A1 PII chromatin (Fig. 6, lane 4). Increased association of SF1 and β-catenin to this regulatory region was detected after 1-h forskolin stimulation in two of three experiments (Fig. 6, lanes 7 and 8). Antibodies to Gα_q/11 failed to immunoprecipitate CYP19A1 PII under any condition (Fig. 6, lanes 2 and 6), and none of the antibodies precipitated chromatin distal to the CYP19A1 gene (data not shown). Thus, β-catenin is present in transcription complexes assembled on the endogenous CYP19A1 PII, and its association, along with that of SF1, is likely regulated by cAMP-dependent mechanisms.

Plasmids and siRNA.

Expression and reporter vectors were generously donated as follows: murine Δ90 β-catenin in the pUHD10–3 vector by W. J. Nelson (Stanford University School of Medicine, Palo Alto, CA) (52); murine Axin1 in a pCS2+MT vector by F. Costantini (Columbia University, New York, NY) (53); murine Sf1 and Sf1(235-4A) in the pCMX vector by K. Morohashi (University of Tsukuba, Tsukuba, Japan) (30); and human CYP19A1-517 promoter fragment in pGL3-basic by S. Bulun (Northwestern University, Chicago, IL) (54). Full-length murine Sf1 in pcDNA3 was described in ref. 26. Silencing RNA SMARTpools, CTNNB1 and Ctnnb1, and siCONTROL nontargeting siRNA were purchased from Dharmacon (Lafayette, CO).

Recombinant Adenovirus Construction.

Adenovirus vectors encoding SF1, SF1(235-4A), and Δ90 β-catenin were constructed by using the AdMax system (Microbix Biosystems, Toronto, ON, Canada). Full-length Sf1 and Sf1(235-4A) were excised from pcDNA3 and pCMX, respectively, and subcloned into the EcoRI site of the pDC316(io) adenovirus shuttle vector. Δ90 β-catenin was excised from pUHD10–3 with SacII and XbaI, blunt ends were generated, and the gene was subcloned into the SmaI site of pDC316. Constructs were confirmed by restriction digestion and sequencing. Recombinant adenoviruses were generated by cotransfecting HEK293 cells with adenovirus shuttle vectors and pBHGloxΔE,3Cre plasmid (circularized adenovirus genome) as described (55). Sample absorbance at 260 nm was used to estimate viral content by using the relationship 10¹² virus particles per ml (vpm)/A₂₆₀ unit (56). Titers used were determined empirically by Western blot and functional assays.

Cell Culture.

Female Sprague–Dawley rats (20 days old), purchased from Charles River Laboratories (Hollister, CA), were maintained in accordance with Washington State University Institutional Animal Care and Use Committee guidelines. At 23 days, rats were injected daily with 0.15 mg of 17β-estradiol s.c. for 3 days. Ovaries were harvested and GCs were isolated as described (46). Cells were plated (2–3 × 10⁶ per 60-mm dish) in DMEM/F12 medium (Invitrogen, Carlsbad, CA) supplemented with 10% (vol/vol) FBS and 1% (vol/vol) penicillin/streptomycin (Invitrogen, Grand Island, NY) for 4 h at 37°C in 5% CO₂ before adenovirus transduction. KGN cells (RIKEN, Kobe, Japan) were cultured in DMEM supplemented with 10% FBS/1% penicillin/streptomycin at 37°C in 5% CO₂.

Adenovirus Transduction and Hormone Stimulation.

Primary GCs were exposed to recombinant adenoviruses diluted in serum-free medium supplemented with 1% penicillin/streptomycin for 13 h, and the medium was then replaced with fresh DMEM/F12/penicillin/streptomycin. At 24 h after transduction, cells were incubated in the presence or absence of 100 ng/ml purified human FSH (National Hormone and Peptide Program, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health) under serum-free conditions (24 h), then harvested for total RNA and protein isolation.

RT-PCR.

Total RNA was isolated and DNase-treated to remove genomic DNA contamination by using RNeasy Kits (Qiagen, Valencia, CA). First-strand cDNA was synthesized (1 μg total RNA) by using oligo(dT) primers and SuperScript II Reverse Transcriptase (Invitrogen, Carlsbad, CA). Primer sequences for Cyp19a1 (NM_017085) and Cyp11a1 (NM_017286) were published previously (46). Additional sequences include the following: Gapdh [NM_017008; (F) 5′-ATGGTGAAGGTCGGTGTGAACG-3′, (R) 5′-GTTGTCATGGATGACCTTGGCC-3′], and Star [NM_031558; (F) 5′-GCCAGCAGGAGAATGGAG-3′, (R) 5′-AGCCAGCTCATGGGTGAT-3′]. Preliminary studies identified the linear signal range from 15 to 40 cycles. PCR was performed on 20 ng of total cDNA, and PCR products were resolved by agarose gel electrophoresis and ethidium bromide staining. Quantity One 4.5.0 1-D Analysis software (Bio-Rad, Hercules, CA) was used for image capture and densitometry. Gapdh signal was used to normalize for cDNA content variability between samples.

Western Blots.

Primary GCs were scraped into PBS, pelleted (400 × g, 4 min), and lysed in radioimmunoprecipitation assay buffer [0.2 mM NaVO₃, Complete Mini Protease Inhibitor tablets (Roche Diagnostics, Indianapolis, IN) added fresh]. Protein concentrations were estimated by using Coomassie Plus Protein Assays (Pierce, Rockford, IL). Samples (15 μg) were separated by SDS/4–20% PAGE Tris·HCl gels (Bio-Rad) as described by Laemmli (57) and transferred to PVDF membrane (Bio-Rad) in Towbin buffer. Primary antibodies against β-catenin (BD Transduction Laboratories, Lexington, KY), SF1 (Upstate Biotechnology, Lake Placid, NY), and Akt (Cell Signaling Technology, Beverly, MA) were incubated in 5% milk/0.05% Tween 20/Tris-buffered (pH 8.0) saline blocking solution overnight at 4°C. The SF1 antibody cross-reacts with LRH1 (58), but the antigens are distinguished by molecular weight. After HRP-conjugated secondary antibody incubations (Pierce), antigen–antibody complexes were detected by chemiluminescence (Immobilon; Millipore, Billerica, MA).

Transient Cotransfection and Luciferase Assays.

KGN cells (6 × 10⁴) were plated in complete medium to achieve 70–80% confluence, and primary GCs (3 × 10⁵) were incubated in DMEM/F12/10% FBS for 3–4 h before transfection. Transient transfections using Lipofectamine (KGN) or Lipofectamine 2000 (GC) (Invitrogen, Carlsbad, CA) followed manufacturer’s recommendations. Total DNA transfected was equalized with empty pcDNA3 vector where indicated. All groups were cotransfected with pHRG-TK Renilla luciferase reporter vector (Promega, Madison, WI) to normalize for transfection efficiency differences. Cells were treated for 24 h with vehicle (0.25% DMSO), FSH (100 ng/ml), or forskolin (10 μM) (Sigma–Aldrich, St. Louis MO) in serum-free medium 24 h after transfection for transient expression assays, and 48 h after transfection for RNAi studies. Testosterone propionate (10 nM) was added in GC treatments. Cells were harvested for luciferase assays by using the Dual-Luciferase Reporter Assay System (Promega).

ChIP Assays.

ChIPs were performed as detailed by Weck and Mayo (58). Briefly, KGN cells (1.5 × 10⁶ per 60-mm dish) were plated overnight. Cells were treated with vehicle or 10 μM forskolin for 1 h (37°C, 5% CO₂), and cross-linked with 1% formaldehyde for 15 min at room temperature. Cells were lysed (250 μl final volume), and sonicated with a Fisher Sonic Dismembrator Model 100 set at 1 (six times, 15 s each). One tenth of the sample was removed for input and one tenth was removed to evaluate sonication efficiency. Remaining samples were divided into three groups, diluted in 1.5 ml of immunoprecipitation buffer, precleared with salmon sperm DNA/Protein A agarose (Upstate Biotechnology) (45 min, 4°C), and immunoprecipitated overnight at 4°C with 1.0 μg of each antibody [Gα_q/11 and β-catenin (Santa Cruz Biotechnology, Santa Cruz, CA), and SF1 (Upstate Biotechnology)]. Eluted protein/DNA complexes were digested with proteinase K, and DNA was purified with the Qiaquick PCR purification kit (Qiagen, Valencia, CA). PCR was performed on purified DNA by using primers upstream [(F) 5′-GCATCAGGGAACACTGGTTT-3′; (R) 5′-GGTACCTGGACCACAGCACT-3′] or within PII of the human CYP19A1 gene [(F) 5′-TTTCCACACTACCGTTGGCCG-3′; (R) 5′-GGCAATCTTCTTCCCTTGAAGC-3′]. Products were separated on agarose/ethidium bromide gels and visualized as indicated above.

Statistical Analyses.

All experiments were repeated at least three times on independent days. Transient transfection experiments were performed in triplicate wells each day unless otherwise noted. All experiments were analyzed by using general linear model procedures of SAS (SAS Institute, Cary, NC), for a completely randomized design. Least-square means are reported. Statistical significance was set at P < 0.05, indicated by asterisks in figures.