FOXO1 Negates the Cooperative Action of FOXL2C134W and SMAD3 in CYP19 Expression in HGrC1 Cells by Sequestering SMAD3.

Adult granulosa cell tumor (aGCT) is a rare type of ovarian cancer characterized by estrogen excess. Interestingly, only the single somatic mutation FOXL2 C134W was found across virtually all aGCTs. We previously reported that FOXL2C134W stimulates CYP19 transcription synergistically with SMAD3, leading to elevated estradiol synthesis in a human granulosa cell line (HGrC1). This finding suggested a key role for FOXL2C134W in causing the typical estrogen overload in patients with aGCTs. We have now investigated the effect of FOXO1, a tumor suppressor, on CYP19 activation by FOXL2C134W in the presence of SMAD3. Intriguingly, FOXO1 antagonized the positive, synergistic effect of FOXL2C134W and SMAD3 on CYP19 transcription. Similar to FOXL2C134W, FOXO1 binds SMAD3 but not the proximal FOXL2C134W binding site (-199 bp) of the CYP19 promoter identified in our earlier studies. The results of a competitive binding assay suggested a possible underlying mechanism in which FOXO1 sequesters SMAD3 away from FOXL2C134W, thereby negating the cooperative action of FOXL2C134W and SMAD3 in inducing CYP19 expression. To our knowledge, this study is the first to demonstrate the ability of FOXO1 to restore an altered CYP19 expression by FOXL2C134W and SMAD3 and provides insight as to why FOXO1 deficiency promotes GCT development in mice.

The forkhead box (FOX) family of transcription factors consists of 19 subfamilies clustered by sequence conservation, from FOXA to FOXS [1]. FOX proteins are encoded by >170 distinct genes identified in 14 species, including 50 in humans, and characterized by an evolutionarily conserved DNA-binding domain known as the FOX domain (also known as the winged-helix domain). Most FOX proteins bind to DNA as monomers [2]. They regulate a diverse range of biological and physiological activities, including developmental and homeostatic processes [3]. In the field of ovarian biology, three of the FOX family members, FOXL2, FOXO1, and FOXO3, have been extensively studied and are known to be associated with gonadal development, sterility, and tumorigenesis [4].

Studies in humans, mice, and goats demonstrated that FOXL2 is expressed in selected cell types, such as mesenchyme of developing eyelids, embryonic and adult pituitary gonadotrope and thyrotroph cells, the endometrium, and fetal and adult granulosa cells (GCs) in the ovary [5-12]. Two Foxl2 germline null mouse models developed in different laboratories displayed perinatal mortality of 50% to 95%, associated with craniofacial malformations and impaired development of the ovary and eyelid. Alternatively, the surviving female mice were infertile due to a block in follicle development at the primary stage associated with a failure of GCs to complete the squamous to cuboidal transition, suggesting that FOXL2 is essential for early folliculogenesis [11, 13]. In contrast, somatic and inducible FOXL2 knockdown experiments in adult female mice have demonstrated a second critical role, which is the maintenance of ovarian structure and function. This is supported by the fact that FOXL2 knockdown in adult female mice leads to the de-repression of Sox9 and the ovary-to-testis sex reprogramming associated with the transdifferentiation of ovarian granulosa/theca cells into Sertoli/Leydig-like cells, the upregulation of testicular genes, the downregulation of ovarian genes, such as CYP19, and the loss of oocytes [14]. This finding suggests that the Foxl2 gene needs to be activated throughout life to suppress male-specific gene expression in the gonad.

More than 120 germline mutations in the FOXL2 gene have been found to be the cause for blepharophimosis/ptosis/epicanthus inversus syndrome (BPES), an autosomal-dominant disease characterized by dysplasia of the eyelids associated with (BPES type I) or without (BPES type II) premature ovarian failure [5, 15]. In contrast, a single somatic missense mutation, FOXL2, which is never observed in BPES, is found in virtually all adult GC tumors (aGCTs)—a rare, but unpredictable, type of ovarian cancer (2% to 5%). These studies included >400 tumors from patients of diverse ethnicities [16, 17]. Clinically, aGCTs are the most common sex cord stromal cell tumors of the ovary and are characterized by increased estradiol levels. Surgery for the complete tumor resection is the basis for the treatment of primary and relapsed tumor. Chemotherapy treatment is administered only for advanced or nonresectable tumors. Most of the aGCTs are diagnosed at an early stage with a favorable prognosis. However, survival upon recurrence is modest in this pathology [18, 19]. The uniqueness of the FOXL2C134W mutation is a compelling finding in a tumor with an otherwise modestly impacted genome, which lacks the consistent induction of oncogenes or inactivation of tumor suppressor genes found among other tumor types. In fact, immunohistochemical evaluation of different oncogenes and tumor suppressors (e.g., MYC, CDKN1A, ERBB2, and TP53) failed to detect a correlation with patient outcome [20]. Earlier studies have suggested that FOXL2C134W is potentially involved in the onset of aGCTs [15, 21–26], yet there is no clear mechanism to account for the causal relationship.

Foxo1 knockout mice showed embryonic lethality after embryonic day 10.5 [27, 28] owing to defects in vascularization and blood vessel formation, which precluded early studies of FOXO1 function in the ovary in vivo. More recently, the selective depletion of the Foxo1 gene in mouse GCs revealed normal ovarian follicle morphology with subfertility, but there was no observation of tumorigenesis [29]. Similarly, mice with GC-specific depletion of the Foxo3 gene also exhibited normal ovarian follicle morphology with no GC tumors (GCTs). However, large GCTs were developed in ∼20% of the Foxo1 and Foxo3 double-knockout mice [29]. Moreover, the loss of Foxo1, Foxo3, and the tumor suppressor Pten in mouse GCs increased levels of estradiol and SMAD3 activation, and developed GCTs in 65% of female mice [30], ∼10-fold higher than observed following targeted depletion of Pten alone [31]. Clinically, FOXO proteins are involved in various diseases, including cancer, diabetes mellitus, and Alzheimer’s disease [32]. It is noteworthy that FOXO1 is targeted and inactivated by FHL2 and by the PI3K–AKT pathway; FHL2 is known as a transcriptional cosuppressor of FOXO1 and is implicated in aGCT aggression, whereas an elevated PI3K–AKT pathway is a hallmark of cancer [33]. Collectively, FOXO1 is considered a tumor suppressor that limits cell proliferation and induces apoptosis [34, 35].

We recently investigated the role of FOXL2C134W mutation in GC function with a hypothesis that FOXL2C134W regulates the expression of aGCT markers differently from FOXL2wt. Specifically, we investigated FOXL2 regulation of inhibin B and CYP19 using a recently established human GC line, human GC line (HGrC1) [36]. The results showed that neither FOXL2wt nor FOXL2C134W regulates Inhbb expression levels when cells were treated with or without activin A, or in the presence or absence of SMAD3 overexpression. However, FOXL2C134W, but not FOXL2wt, was able to induce CYP19 expression selectively in the presence of activin A or overexpressed SMAD3. Moreover, FOXL2C134W interactions with SMAD3 and with the FOX binding element (FBE) located at −199 bp upstream of the ATG initiation codon of human CYP19 were more sustainable than FOXL2wt.

To extend our studies on the role of FOXL2C134W in regulating CYP19 expression via enhanced recruitment of SMAD3 to a proximal FBE, we have now assessed whether another forkhead protein, FOXO1, is involved in the mechanism because it was demonstrated to partner with SMAD3 in the regulation of activin induction of Fshb in immortalized murine LβT2 gonadotrope cells [37].

In this study, we have shown the actions and interactions of three factors potentially involved in the regulation of an aGCT marker, CYP19, namely SMAD3, FOXL2, and FOXO1 [26, 38, 39], following the hypothesis that FOXO1 can oppose FOXL2C134W actions. We have focused on whether and how FOXO1 differentially regulates FOXL2C134W and FOXL2wt activities on CYP19 expression and whether it occurs in association with SMAD3. It is remarkable that the finding of a new possible target, FOXO1, that modulates the function of FOXL2C134W has implications for both aGCT establishment and potential future modulation of the tumor. Certainly, new agents able to abolish the typical aGCT CYP19 upregulation appear to be important in the field of hormone-dependent cancer.

1. Materials and Methods

A. Plasmids and Reagents

Antibodies against FOXO1 (catalog no. 9454, RRID:AB_823503) [40], SMAD2/3 (catalog no. 3102, RRID:AB_10698742) [41], and β-actin (catalog no. 4967, RRID:AB_330288) [42] were purchased from Cell Signaling Technology (Danvers, MA); the antibody against FOXL2 (catalog no. IMG-3228, RRID:AB_613022) [43] was from Imgenex (San Diego, CA); the antibody against c-Myc tag (catalog no. 551101, RRID:AB_394046) [44] was from BD Pharmingen (Franklin Lakes, NJ); and the antibody against FLAG tag (catalog no. F1804, RRID:AB_262044) [45] was from Sigma-Aldrich (St. Louis, MO). Activin A was previously made in our laboratory and reported [46]. HGrC1 cells were established as previously described [36]. Expression plasmids encoding N-terminally Flag-tagged hFOXO1wt (catalog no. 13507), N-terminally Flag-tagged hFOXO13A (catalog no. 13508), and N-terminally Flag-tagged hFOXO13A-DBD (catalog no. 10696) were purchased from Addgene (Cambridge, MA). FOXO13A has Ala residue substitutions at Thr24, Ser256, and Ser319 and is thus constitutively active whereas FOXO13A-DBD lacks the DNA binding site of FOXO13A. N-terminally Flag-tagged human FOXL2wt (FOXL2wt) was provided by Dr. Louise Bilezikjian [47], and its mutant (FOXL2C134W) was made in our laboratory as previously reported [48, 49]. The expression plasmids encoding human SMAD3 with or without N-terminal Myc tag were prepared from those provided by Drs. Kohei Miyazono [50, 51] and Louise Bilezikjian [47]. Luciferase reporter plasmids of human CYP19 promoter in different lengths were made in our laboratory using PCR-based methods. Biotinylated oligonucleotide probes were purchased from Integrated DNA Technologies (San Diego, CA).

B. HGrC1 Cell Culture

HGrC1 cell properties and derivation were described earlier [52]. Notably, these cells bear two normal alleles of the FOXL2 gene [52]. Cells were cultured in a standard incubator at 37°C in DMEM-F12 medium (catalog no. 12400-024, Thermo Fisher Scientific, Waltham, MA) complemented with antibiotics (penicillin and streptomycin) and 10% fetal bovine serum (catalog no. F6178, Sigma-Aldrich). HGrC1 cells at passage 14 were used for all of the experiments in this study. For some experiments, HGrC1 cells were treated with 100 ng/mL activin A for 24 hours in serum-free DMEM-F12 medium.

C. Transient Transfections

HGrC1 cells were cultured into either 12-well tissue culture plates (0.5 × 106 cells per well) or six-well tissue culture plates (1 × 106 cells per well). Medium was replaced the day after with serum-free DMEM-F12 (complemented with antibiotics), and transfection was performed using the above-described plasmids for 4 hours with Lipofectamine LTX with Plus or Lipofectamine 3000 reagents (catalog no. 15338100 or no. L3000008, respectively, Thermo Fisher Scientific) following the manufacturer’s procedures. After 4 hours, medium was replaced with new serum-free DMEM-F12 and transfected cells were cultured at 37°C for 24 hours with or without activin A. At this point cells were lysed for further analyses.

D. RNA Extractions and Quantitative RT-PCR

After 4 hours of transfection and additional culture for 24 hours in serum-free DMEM-F12, HGrC1 cells were lysed using TRIzol reagent (catalog no. 15596026, Thermo Fisher Scientific) and RNA was extracted with a Direct-zol RNA MiniPrep kit (catalog no. R2052, Zymo Research, Irvine, CA) following the manufacturer’s protocol. A high-capacity cDNA reverse transcription kit (catalog no. 4368814, Thermo Fisher Scientific) was used to reverse transcribe 1 μg of RNA. mRNA expression was quantified by qRT-PCR amplification of cDNA using SYBR Green PCR master mix (catalog no. 4309155, Thermo Fisher Scientific) and the ABI 7300 real-time PCR system thermal cycler from Applied Biosystems (Foster City, CA). The primer sequences used to amplify CYP19 were: forward, 5′-GGAATTATGAGGGCACATCC-3′, reverse, 5′-GTTGTAGTAGTTGCAGGCAC-3′; for L19: forward, 5′-GGGATTTGCATTCAGAGATCAG-3′, reverse, 5′-GGAAGGGCATCTCGTAAG-3′. Target gene expression was normalized on L19 expression.

E. Western Blot Analysis

After 4 hours of transfection and additional culture for 24 hours in serum-free DMEM-F12, HGrC1 cells were lysed in lysis buffer (RIPA buffer, catalog no. 89901, Thermo Fisher Scientific), phosphatase inhibitor cocktail (catalog no. 78420, Thermo Fisher Scientific), and protease inhibitor cocktail (catalog no. P8340, Sigma-Aldrich), and a Pierce bicinchoninic acid protein assay kit was used for total protein quantification (catalog no. 23227, Thermo Fisher Scientific). NuPAGE lithium dodecyl sulfate sample buffer 4× (catalog no. NP0007, Thermo Fisher Scientific) and β-mercaptoethanol (catalog no. 6010, Calbiochem, Billerica, MA) were added to cell lysates, and samples were denaturized at 95°C for 5 minutes. Protein separation occurred on 12% SDS-PAGE gels and with subsequent transfer to nitrocellulose membranes. Membranes were then incubated for 1 hour with blocking solution (BSA, catalog no. A30075-100, Research Products International, Mount Prospect, IL), and with primary antibodies overnight at 4°C. Membranes were washed three times, then incubated for 1 hour with horseradish peroxidase–conjugated secondary antibodies, further washed three more times, and incubated with SuperSignal West Femto maximum sensitivity substrate to detect chemiluminescence (catalog no. 34095, Thermo Fisher Scientific). Alternatively, primary and secondary antibodies incubations were performed using solutions 1 and 2, respectively, of Signal Enhancer HIKARI (catalog no. NU00101 and no. NU00102, Nacalai USA, San Diego, CA) to enhance protein detection. β-Actin was used as a loading control.

F. Luciferase Reporter Assay

After 4 hours of transfection and additional culture for 24 hours in serum-free DMEM-F12, HGrC1 cells (0.5 × 106 cells per sample) were lysed in a solution composed of 100 mM potassium phosphate (pH 7.8) and 0.2% Triton X-100. A luciferase reporter assay was performed with a Nano-Glo® Dual-Luciferase® reporter assay system (catalog no. N1610, Promega) following the manufacturer’s instructions. The FLUOstar Omega microplate reader (BMG Labtech, Ortenberg, Germany) was used to measure the relative firefly luciferase activity, which was normalized by β-galactosidase activity. We used the same plasmid constructs as reported in our earlier studies [52].

G. Coimmunoprecipitation and Oligonucleotide Binding Assays

After 4 hours of transfection and additional culture for 24 hours in serum-free DMEM-F12, HGrC1 cells (2.5 × 106 cells per sample for coimmunoprecipitation assay; 5 × 106 cells per sample for oligonucleotide binding assay) were lysed in lysis solution [25 mM Tris-HCl (pH 7.4), 140 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40, 0.5% Triton X-100, 10% glycerol, 2 mM NaF, freshly added 1 mM Na4P2O7, protease inhibitor cocktail (catalog no. P8340, Sigma-Aldrich) and Benzonase (catalog no. E1014, Sigma-Aldrich)]. Cell extracts were then sonicated for 3 to 4 seconds twice, spun down at 4°C for 10 minutes at 13,000 rpm, and supernatants were collected. For the oligonucleotide binding assay, the EpiQuik™ general protein–DNA binding assay kit (catalog no. P-2004-96, EpiGentek, Farmingdale, NY) was used. For coimmunoprecipitation assays, supernatants were first precleared for 1 hour with mouse IgG agarose (catalog no. A0919) and immunoprecipitated using anti-FLAG M2 affinity gel (catalog no. A2220, Sigma-Aldrich) overnight at 4°C. The precipitants were washed three times in lysis buffer and eluted with FLAG peptide (catalog no. F3290, Sigma-Aldrich). After the elution step, immunoblot analysis of the samples was performed with c-myc antibody to detect the immunoprecipitated proteins tagged with cMyc. Secondary antibodies, goat anti-mouse IgG antibody (IRDye 800CW, LI-COR Biosciences, Lincoln, NE, RRID AB_10793856) [53], and goat anti-rabbit IgG antibody (IRDye 800CW, LI-COR Biosciences, RRID AB_10796098) [54], which are conjugated to IR fluorescent dyes excitable at 800 nm, were used to visualize the desired antigens using an 800-nm laser scanner (Odyssey, LI-COR Biosciences).

H. Statistical Analysis

All statistical analyses were performed using GraphPad Prism software. Data analysis was also performed with JMP Pro 12 software (SAS Institute, Cary, NC) using Box–Cox transformation and one-way ANOVA analysis with a least-squares means Tukey honestly significant difference test or a least-squares means Student t test. Significance was set at P < 0.05, and data are shown as mean ± SEM. For experiments where variance was not similar between the groups being compared, a parametric unpaired t test with a Welch correction or nonparametric Kolmogorov–Smirnov test was used.

2. Results

A. FOXO1wt Suppresses CYP19 mRNA Expression Induced by FOXL2C134W and SMAD3 in HGrC1 Cells

HGrC1 cells were transfected with SMAD3 and FOXL2wt or FOXL2C134W in the presence or absence of FOXO1wt, or the matched empty vectors as controls. Cell lysates were collected for immunoblot analysis. With this protocol, we have previously demonstrated expression levels of endogenous and exogenous SMAD3, FOXL2wt, and FOXL2C134W [52]. Figure 1A shows expression levels of endogenous (empty vector) and exogenous FOXO1 (upper panel) and β-actin (lower panel). Total RNA was collected in parallel to evaluate CYP19 mRNA expression by qRT-PCR. Previously [52], we showed that SMAD3 or FOXL2wt alone or in combination have indistinguishable effects on CYP19 mRNA expression from empty vector controls, whereas the combination of SMAD3 and FOXL2C134W markedly induces CYP19 mRNA expression. The current study confirms the induction of CYP19 mRNA expression by SMAD3 with FOXL2C134W (17-fold, Fig. 1B). Strikingly, a dramatic reduction (4.5-fold) of CYP19 mRNA expression was observed in the presence of FOXO1wt. These results suggest that FOXO1 has a selective inhibitory role on CYP19 mRNA expression induced by SMAD3 and FOXL2C134W in HGrC1 cells.

Figure 1.

Effect of FOXO1 on SMAD3- and FOXL2C134W-induced CYP19 mRNA expression in HGrC1 cells. HGrC1 cells were transfected with 500 ng of SMAD3, Flag-FOXL2wt, Flag-FOXL2C134W, and Flag-FOXO1wt plasmids as shown, or their empty vectors to equalize the amount of plasmid transfected per well. After 4 h, serum-free medium was replaced and cells were cultured for 24 h, followed by the collection of cell lysates and total RNA. (A) Cell lysates were subjected to western blot analysis to show the expression levels of Flag-FOXO1wt proteins. The expression levels of FOXL2 and SMAD3 in these cells were reported earlier [52]. Antibodies against FOXO1 and β-actin (for loading control) were used. Representative results from duplicate experiments are shown. (B) qRT-PCR was performed to measure CYP19 mRNA levels. Representative results (n = 3) from triplicate experiments are presented. Data are shown as mean ± SEM. Different letters denote significant difference among experimental conditions (P < 0.05).

B. FOXO13A Abolishes SMAD3 and FOXL2C134W-Induced CYP19 mRNA Expression

To approach the mechanism by which FOXO1 inhibits SMAD3- and FOXL2C134W-induced CYP19 mRNA expression, we next tested the effect of the FOXO13A mutant that is confined to the nucleus and is thereby constitutively active [55]. As an additional control, we included the FOXO13A-DBD mutant in this study, which lacks the capacity to bind SMAD3 as well as DNA [37].

HGrC1 cells were transfected with SMAD3 with different combinations of FOXL2wt, FOXL2C134W, FOXO13A, and FOXO13A-DBD plasmids, or matched empty vectors as a control for total DNA transfected. Figure 2A shows the relative expression levels of endogenous (empty vector) and exogenous mutant FOXO1 proteins (upper panel) and β-actin (lower panel). Notably, FOXO13A showed a more slowly migrating band compared with endogenous FOXO1 due to the Flag tag. In FOXO13A-DBD, the Flag tag is offset by the deletion of 13 amino acids (amino acid positions from 208 to 220) from the DNA binding domain. Total RNA was again collected in parallel to evaluate CYP19 mRNA expression by qRT-PCR (Fig. 2B). Appreciably, while addition of FOXO13A or FOXO13A-DBD to SMAD3 and FOXL2wt did not alter the CYP19 mRNA expression, FOXO13A abolished the SMAD3 and FOXL2C134W induction (13-fold) of CYP19 expression down to control levels, indicating that the nuclear localization of FOXO1 is essential to suppress the cooperative induction of CYP19 by FOXL2C134W and SMAD3. FOXO13A-DBD did not exhibit any effects on SMAD3 and FOXL2C134W-induced CYP19 mRNA expression, possibly due to its incapacity to bind SMAD3 and/or DNA.

Figure 2.

Effects of FOXO13A and FOXO13A-DBD on SMAD3- and FOXL2C134W-induced CYP19 mRNA expression in HGrC1 cells. HGrC1 cells were transfected with or without 500 ng of SMAD3, Flag-FOXL2wt, Flag-FOXL2C134W, Flag-FOXO13A, and Flag-FOXO13A-DBD plasmids, alone or in combination. Empty expression vectors for FOXO13A and for FOXO13A-DBD were added as necessary to equalize the amount of plasmid transfected per well. After 4 h, serum-free medium was replaced and cells were cultured for 24 h, followed by the collection of cell lysates and total RNA. (A) Cell lysates were subjected to western blot analysis to show the expression levels of FOXO13A and for FOXO13A-DBD proteins using antibodies raised against FOXO1 or β-actin (for loading control). Representative results from duplicate experiments are shown. (B) qRT-PCR was performed to measure CYP19 mRNA levels. Representative results (n = 3) from triplicate experiments are presentedData are shown as mean ± SEM. Different letters denote significant difference among experimental conditions (P < 0.05).

C. FOXO13A Significantly Reduces CYP19 Promoter Activity Induced by SMAD3 and FOXL2C134W

Using the Universal PBM Resource for Oligonucleotide Binding Evaluation database [56] to identify potential binding sites for FOX and SMAD proteins, five FBEs (FBE1 to FBE5) and one SMAD binding element (SBE) were found within 1 kb upstream of the translation start site in human CYP19 promoter [52] (Fig. 3A). A luciferase reporter plasmid was used with a sequence of 823 bp including the five potential FBE and one potential SBE sites, designated −823/CYP19-Luc as reported previously [52] (Fig. 3B). The relative luciferase activity of −823/CYP19-Luc in HGrC1 cells transfected with combinations of SMAD3, FOXL2C134W, and FOXO13A plasmids (or matched empty vectors) is shown in Fig. 3C. We confirmed that the expression of FOXL2C134W with SMAD3 increased CYP19 promoter activity, replicating past results that the combination (but not SMAD3 alone) could activate CYP19 promoter activity [52]. However, the coexpression of FOXO13A completely depressed −823/CYP19-Luc activity to control levels. These data together suggest that induction of CYP19 expression by FOXL2C134W and SMAD signaling is opposed by FOXO1 at a transcriptional level.

Figure 3.

Human CYP19 promoter activity regulated by FOXL2 and FOXO1. (A) DNA sequences for potential binding sites of FOX (FBE1 to FBE5) and SMAD (SBE) proteins within 1 kb of the proximal region of human CYP19 promoter. (A) Human CYP19 promoter (1 kb) was explored by the Universal PBM Resource for Oligonucleotide Binding Evaluation database that includes data generated by universal protein binding microarray technology on the in vitro DNA binding specificities of proteins. The coding sequence of hCYP19 is underlined. (B) Luciferase reporter of human CYP19 promoter (−823/CYP19-Luc). Potential FOX (FBE1 to FBE5) and SMAD (SBE) binding sites are shown. HGrC1 cells were transfected with the −823/CYP19-Luc reporter and combinations of SMAD3, Flag-FOXL2C134W, and Flag-FOXO13A plasmids. Empty expression vectors were added as necessary to equalize the amount of DNA transfected per well. After 4 h, serum-free medium was replaced and cells were cultured for 24 h, followed by collection of cell lysates. (C) Relative luciferase activities are shown. Representative results (n = 3) from triplicate experiments are presented. Data are shown as mean ± SEM and are analyzed using the two-tailed Mann−Whitney test. ***P < 0.001 for difference among experimental conditions.

D. FBE5 (−199) and SBE (−163) Are Involved in FOXO13A Inhibition of CYP19 Promoter Activation by SMAD3 and FOXL2C134W

To resolve the cis elements responsible for FOXO1 action on CYP19 promoter, the −823/CYP19-Luc reporter was truncated to obtain four shorter constructs (Fig. 4). Their activities were evaluated in HGrC1 cells transfected with FOXL2C134W and SMAD3 with or without FOXO13A plasmids. All constructs except the most proximal (−56/CYP19-Luc) showed a significant inhibition of FOXL2C134W-induced CYP19 promoter activation by FOXO13A. These data imply that either the FBE5 (−199), the SBE (−163) [52], or both elements are accountable for FOXO1 regulation of FOXL2C134W activity (in the presence of an activated SMAD3 signaling) on the CYP19 promoter in HGrC1 cells.

Figure 4.

Luciferase activities of truncated human CYP19 promoters. (Upper panels) HGrC1 cells were transfected with an indicated CYP19-luciferase reporter (A, B, C or D) and combinations of SMAD3, Flag-FOXL2C134W and Flag-FOXO13A plasmids. After 4 h, serum free medium was replaced, and cells were cultured for 24 h, followed by collection of cell lysates. (Lower panels) Ratio of relative luciferase activities on the effect of FOXL2C134W or FOXO13A are shown. Representative results (n = 3) from triplicate experiments are presented. Data are shown as mean ± SEM and are analyzed using the two-tailed Mann-Whitney test. The asterisk denotes a significant difference among experimental conditions (***P < 0.001). n.s., not significant.

E. FBE5 Is Required for FOXO1 to Inhibit FOXL2C134W on CYP19 Transcription

Two further constructs were used from −823/CYP19-Luc reporter, namely an SBE mutant (Fig. 5A) and an FBE5 mutant (Fig. 5B) [52]. For the new constructs, six bases were mutated within the sequence of each binding element and the reporters were then tested for their activities. Interestingly, the SBE mutant (Fig. 5A) maintained significantly different activities between with or without FOXO13A, whereas the FBE5 mutant (Fig. 5B) failed to show inhibition by FOXO13A, indicating that FBE5 is sufficient to for FOXO13A-mediated reduction of CYP19 promoter activity induced by FOXL2C134W. These data suggested that FOXO13A can inhibit FOXL2 C134W binding to CYP19 promoter, but also opened the question of whether this activity was due to a competitive action of FOXO13A for FOXL2 C134W binding to the FBE5 site or to SMAD3.

Figure 5.

Contribution of FBE5 or SBE sites to FOXL2C134W and FOXO13A activation of CYP19 transcription. SBE (A) and FBE5 (B) sites were mutated at six internal bases as indicated within the −823/CYP19-luciferase reporter, and HGrC1 cells were transfected with these constructs with combinations of SMAD3, Flag-FOXL2C134W, and Flag-FOXO13A plasmids. After 4 h, serum-free medium was replaced and cells were cultured for 24 h, followed by collection of cell lysates. Relative luciferase activities are shown. Representative results (n = 3) from triplicate experiments are presented. Data are shown as mean ± SEM. Data are shown as mean ± SEM and are analyzed using the two-tailed Mann–Whitney test. **P < 0.01 for difference among experimental conditions. n.s., not significant.

F. FOXO1 Associates With SMAD3 in HGrC1 Cells

Despite the lack of contribution of the SBE in the role of FOXO1 attenuation of FOXL2C134W activity, our earlier [52] and current studies implicate SMAD3 as a critical FOXL2C134W partner for CYP19 expression. In fact, SMAD3 forms a more stable protein complex with FOXL2C134W than does FOXL2wt in HGrC1 cells [52]. To investigate whether FOXO1 interacts with SMAD3, cells were transfected with combinations of SMAD3, FOXL2C134W, and FOXO13A plasmids, and coimmunoprecipitation assays were performed. Figure 6 shows the input levels of SMAD3, FOXL2C134W, and FOXO13A (top three panels) and the levels of SMAD3 coprecipitated with FOXL2C134W or FOXO13A (bottom panel). Interestingly, FOXO13A, similar to FOXL2C134W, associates with SMAD3, consistent with the putative role of SMAD3 in regulating FOXO1 activity [37].

Figure 6.

Protein–protein interactions between SMAD3 and FOXL2C134W or FOXO13A. HGrC1 cells were transfected with combinations of Myc-SMAD3, Flag-FOXL2C134W, and Flag-FOXO13A plasmids. Empty expression vectors were added as needed to equalize the amount of DNA transfected per well. After 4 h, serum-free medium was replaced and cells were cultured for 24 h, followed by collection of cell lysates. (A) INPUT levels of SMAD3, FOXL2, and FOXO1 proteins before immunoprecipitation were detected with anti-Myc, anti-FOXL2, and anti-FOXO1 antibodies respectively (top three panels). Protein lysates were immunoprecipitated with anti-Flag agarose beads, and SMAD3 proteins were detected with anti-Myc antibody (bottom panel). (B) Relative signal intensities of SMAD3 protein coprecipitated with FOXL2 and FOXO1 were measured. Representative results from triplicate experiments are presented. Data are shown as mean ± SEM.

G. FOXO13A Does Not Bind to FBE5 in HGrC1 Cells

We previously reported that FOXL2C134W in the presence of SMAD3 binds to the FBE5 site in the proximal region of CYP19 promoter [52]. To investigate whether FOXO1 competes with FOXL2 to bind to the FBE5 site, we performed oligonucleotide binding assays using probes containing the FBE5 and SBE sites (Fig. 7A). Cells were transfected with combinations of SMAD3, FOXL2C134W, and FOXO13A plasmids and treated with activin A for the final 24 hours, then lysed and subjected to a binding assay. In this case, activin A treatment was used to establish a uniform activation of SMAD3 activity in HGrC1 cells as previously demonstrated [52]. Figure 7B confirms this previous observation; specifically, that FOXL2C134W binds to the probe selectively in the presence of Smad3 [52]. In contrast, FOXO13A did not bind to the same probe. Importantly, FOXL2C134W did not bind to the probe when cotransfected with FOXO13A. These findings suggest that FOXO13A acts to prevent FOXL2C134W binding to the FBE5 site.

Figure 7.

Oligonucleotide binding assay. (A) Biotinylated oligonucleotide probes were used. FBE5 and SBE sites are boxed. (B) HGrC1 cells were transfected with Flag-FOXL2C134W or Flag-FOXO13A plus Myc-SMAD3 plasmids. Empty expression vectors were added as needed to equalize the amount of DNA transfected per well. After 4 h, serum-free medium was replaced and cells were cultured for 24 h in the presence of 100 ng/mL activin A, followed by collection of cell lysates. Protein lysates were incubated with the probe precoupled to streptavidin-conjugated agarose plates following procedures described in vendor protocol (see Materials and Methods), and FOXL2 and FOXO1 proteins were detected with an anti-Flag antibody. Relative signal intensities of FOXL2 and FOXO1 protein bound to streptavidin-conjugated probe were measured. Representative results from triplicate experiments are presented. Data are shown as mean ± SEM. **P < 0.01 for difference among experimental conditions. n.s., not significant.

H. FOXO13A Sequesters SMAD3 From the FOXL2C134W/SMAD3 Complex

To further understand FOXO13A action in the presence of SMAD3/FOXL2C134W, activin A–treated cells expressing combinations of SMAD3, FOXL2C134W, and FOXO13A plasmids were subjected to immunoprecipitation and Western blot analyses. Figure 8A shows that although the input levels of FOXO13A and FOXL2C134W are comparable (top two panels), SMAD3 selectively coprecipitates with FOXO13A (bottom panel). These findings suggest a model in which SMAD3 preferentially associates with FOXO13A relative to FOXL2C134W; therefore, FOXO13A opposes FOXL2C134W due to sequestration of SMAD3, a critical partner essential for its regulation of CYP19 transcription.

Figure 8.

Competitive interactions between SMAD3 and FOXL2C134W or FOXO13A. HGrC1 cells were transfected with combinations of Flag-SMAD3, Myc-FOXL2C134W, and Myc-FOXO13A plasmids. Empty expression vectors were added as needed to equalize the amount of DNA transfected per well. After 4 h, serum-free medium was replaced and cells were cultured for 24 h in the presence of 100 ng/mL activin A, followed by collection of cell lysates. (A) Protein lysates were immunoprecipitated with anti-Flag agarose beads, and FOXO1 and FOXL2 proteins were detected with anti-Myc antibody. Input levels of SMAD3, FOXL2, and FOXO1 proteins before immunoprecipitation were detected with anti-Flag and anti-Myc antibodies, respectively. (B) Relative signal intensities of FOXL2 and FOXO1 protein coprecipitated with SMAD3 were measured. Representative results from triplicate experiments are presented. Data are shown as mean ± SEM.

3. Discussion

Ovarian cancer is the most lethal of gynecological malignances with a 5-year survival rate of 20% to 30% for patients with advanced disease [57]. GCTs account for ∼5% of ovarian cancers, with an overall frequency of 1 per 100,000 women [58]. They can be divided into two subtypes, the rarer juvenile type and the more common adult type, based not only according to the differences of the age of the patients at their onset, but also by histology, nuclear morphology, and potential recurrence. Although juvenile GCTs can occur during infancy or puberty, they usually do not bear any FOXL2 mutation, and in fact, loss of FOXL2 expression is observed in more aggressive forms [59, 60]. In contrast, aGCTs, the most common type of GCTs, occur during perimenopausal and postmenopausal age, and almost all patients with aGCTs are heterozygous for the FOXL2C134W mutation [16, 61]. Although most aGCTs are diagnosed at an early stage that can be treated with surgery, there is a high rate of recurrence, particularly among patients diagnosed at later stages [62], and 70% of patients with recurrence die as a result of GCTs. Traditional chemotherapy is relatively ineffective in women with advanced-stage or recurrent GCTs, which promotes the need for more effective therapeutic options for these patients. Additionally, as recurrent or metastatic tumors can manifest many years after removal of the primary tumor, with intervals up to 10 to 20 years not uncommon [63], there is a need for a diagnostic marker to allow accurate assessment of recurrence risk during clinical follow-up of patients.

The FOXO transcription factors play crucial cellular processes by orchestrating programs of gene expression, including (i) initiation of apoptosis, (ii) promotion of cell cycle arrest, (iii) reactive oxygen species detoxification, (iv) repair of damaged DNA, and (v) facilitation of glucose metabolism and energy homeostasis [64]. In particular, their roles in the cell cycle arrest and the initiation of apoptosis support the role for FOXO factors as tumor suppressors. Our current finding in the present study may suggest that FOXO1 exerts its tumor-suppressive activity by limiting FOXL2C134W-induced activity in CYP19 expression and consequently estradiol synthesis in GCs. FOXO1 is the most highly expressed FOXO family member in insulin-responsive tissues, where insulin signaling through the PI3K–AKT pathway plays a key regulation of FOXO1 activities [65]. Indeed, FOXO1 activities are regulated through phosphorylation at Thr24, Ser256, and Ser319 in response to the PI3K–AKT signaling, resulting in translocation of the factors outside the nucleus and their consequent inactivation [66]. Following nuclear exclusion, FOXO1 maintains its inactive state and is subject to ubiquitination and subsequent proteasomal degradation. The activation of PI3K–AKT signaling is common across many tumor types [67]. The inactivation of FOXO1 in response to PI3K represents one mechanism by which tumor cells avoid cell cycle arrest and apoptosis [68].

In the present study, we describe a novel role for the transcription factor FOXO1 in the inhibition of CYP19 expression induced by the synergistic action of mutant FOXL2C134W and its cofactor SMAD3. However, FOXO1 showed no effect on FOXL2wt action, indicating that FOXO1 has an inhibitory role selectively on FOXL2C134W-induced CYP19 expression in the presence of SMAD3.

Our experimental approach to address the underlying molecular mechanism showed that FOXO1 competes with FOXL2C134W in binding the essential cofactor SMAD3. Data showed that SMAD3 sequestration by FOXO13A appeared to be the main molecular event upstream of the FOXO1 reversal of the CYP19 upregulation by the FOXL2C134W/SMAD3 complex. This finding is in good accord with reports by others that FOXO1 partners with SMAD3 in other physiological and pathological conditions [69, 70]. For example, in hepatocytes, TGF-β1/Smad3 promotes gluconeogenesis via the PP2A–AMPK–FOXO1 pathways during fasting/feeding as well as during pathologic conditions such as insulin resistance and diabetes [71]. Similarly, SMAD3/4 and FOXO1 have essential functions in oligodendrocyte differentiation [72], and it was suggested in 2004 that the FOXO1/SMAD3 complex may control neuroepithelial and glioblastoma cell proliferation in brain tumors [73].

Consistent with these existing views of FOXO1 functions, FOXO13A in the current study exhibited more potent activity than did FOXO1wt. Interestingly, the deletion of the DNA binding domain in FOXO13A (FOXO13A-DBD) compromised the inhibition of FOXL2C134W by FOXO1, despite a lack of binding by FOXO1 to FBE5. Indeed, the DNA binding domain of FOXO1 is also essential for SMAD3 binding [37], further supporting a model in which FOXO1 directly competes with FOXL2C134W for binding to SMAD3. FOXL2C134W action at the FBE5 site on CYP19 promoter occurs with SMAD3 as a cofactor [52]. This is not broadly true of all SMADs, as SMAD2 [52] and other SMADs, including SMAD1, SMAD5, and SMAD8 (data not shown), fail to cooperate with FOXL2C134W to regulate CYP19 transcription in HGrC1 cells. Given that FOXO1 does not bind to FBE5, but exhibits more stable interaction with SMAD3 than does FOXL2C134W, we propose a model in which FOXO1 prevents DNA binding of FOXL2C134W via a competition for SMAD3, and not for the same DNA binding site. We suggest a possible mechanism by which the FOXO1 sequestration of SMAD3 inhibits FOXL2C134W/SMAD3-induced CYP19 expression in HGrC1 cells as depicted in Fig. 9. An inhibitory role for FOXO1 on CYP19 regulation was previously postulated by Park et al. [74] using rat primary GCs by showing that a synergistic induction of FSH and activin A on CYP19 expression was supported by endogenous FOXO1 phosphorylation (thus inactivation) via the PI3K–AKT signaling activated by FSH whereas overexpression of FOXO13A suppresses the synergistic induction.

Figure 9.

Possible molecular mechanism for the regulation of CYP19 transcription by FOXO1 and FOXL2C134W. (A) and (B) Findings of our previous study where FOXL2C134W and SMAD3 synergistically stimulate CYP19 transcription by binding to FBE5 in CYP19 promoter more strongly than FOXL2wt and SMAD3. (C) Data collected in this study show the ability of FOXO1 to counteract the synergistic activity of FOXL2C134W and SMAD3 in the stimulation of CYP19 transcription. FOXO1 and FOXL2C134W both bind SMAD3, but FOXO1 does it more strongly than FOXL2C134W. Therefore, FOXO1 negates the ability of FOXL2C134W to bind SMAD3 and thus to enhance a steady-state level of CYP19 transcription.

In summary, we have used HGrC1 cells to investigate the role of FOXO1 in the regulation of CYP19 expression in opposition to FOXL2C134W mutant reported earlier [52]. Our findings are that (i) FOXO1wt decreases CYP19 mRNA expression cooperatively induced by FOXL2C134W and SMAD3, and the constitutively active mutant FOXO13A is more potent but the FOXO13A-DBD mutant has no effect because it does not bind SMAD3 [37], (ii) the above-observed regulation occurs at the transcriptional level, (iii) the presence of FOXO13A prevents the FOXL2C134W/SMAD3 complex binding to the FBE5 site in the proximal region of CYP19 promoter, and (iv) FOXO13A interacts more strongly with SMAD3 but not with FBE5, suggesting that its inhibitory action is possibly related to a sequestration of SMAD3 by FOXO1. Based on our data and in accordance with the literature, these molecular events take place in the nucleus because FOXO13A cannot be phosphorylated in the mutated residues and this prevents its translocation in the cytoplasm. Also, FOXL2C134W and SMAD3 are located in the nucleus as described in the literature [37, 75–77].

To our knowledge, this study is the first to demonstrate the ability of FOXO1 in restoring an altered CYP19 transcriptional activity by FOXL2C134W mutant. It is therefore conceivable that FOXO1 acts as a tumor suppressor by preventing FOXL2C134W activity to enhance CYP19 expression, leading to estrogen access in women with aGCTs. Because the activity of FOXO1 is tightly regulated by posttranslational modifications, including phosphorylation, acetylation, and ubiquitylation [78], therapeutically targeting such modification steps could lead to the discovery and development of efficacious agents against aGCTs. In conclusion, this work addressed to one of the main features of aGCTs: the FOXL2C134W-driven upregulation of aromatase expression and the subsequently increased estradiol levels in these patients. Our findings of the new possible target, FOXO1, that modulates the activity of FOXL2C134W appear to be important in the field of ovarian cancer because it has implications for both aGCT establishment and potential future modulation of the tumor.